Professional Documents
Culture Documents
An Introduction PDF
An Introduction PDF
CHEMICAL ENGINEERING
KINETICS & REACTOR DESIGN
CHARLES G. HILL, JR.
The University of Wisconsin
Bibliography: p.
Includes indexes.
1. Chemical reaction, Rate of. 2. Chemical
reactors—Design and construction. I. Title.
QD502.H54 660.2'83 77-8280
ISBN 0-471-39609-5
Printed in the United States of America
20 19 18
Board of Advisors, Engineering
One feature that distinguishes the education of the chemical engineer from that of
other engineers is an exposure to the basic concepts of chemical reaction kinetics
and chemical reactor design. This textbook provides a judicious introductory level
overview of these subjects. Emphasis is placed on the aspects of chemical kinetics
and material and energy balances that form the foundation for the practice of reactor
design.
The text is designed as a teaching instrument. It can be used to introduce the novice
to chemical kinetics and reactor design and to guide him until he understands the
fundamentals well enough to read both articles in the literature and more advanced
texts with understanding. Because the chemical engineer who practices reactor
design must have more than a nodding acquaintance with the chemical aspects of
reaction kinetics, a significant portion of this textbook is devoted to this subject.
The modern chemical process industry, which has played a significant role in the
development of our technology-based society, has evolved because the engineer has
been able to commercialize the laboratory discoveries of the scientist. To carry out
the necessary scale-up procedures safely and economically, the reactor designer must
have a sound knowledge of the chemistry involved. Modern introductory courses in
physical chemistry usually do not provide the breadth or the in-depth treatment of
reaction kinetics that is required by the chemical engineer who is faced with a reactor
design problem. More advanced courses in kinetics that are taught by physical
chemists naturally reflect the research interests of the individuals involved; they do
not stress the transmittal of that information which is most useful to individuals
engaged in the practice of reactor design. Seldom is significant attention paid to the
subject of heterogeneous catalysis and to the key role that catalytic processes play
in the industrial world.
Chapters 3 to 7 treat the aspects of chemical kinetics that are important to the
education of a well-read chemical engineer. To stress further the chemical problems
involved and to provide links to the real world, I have attempted where possible
to use actual chemical reactions and kinetic parameters in the many illustrative
examples and problems. However, to retain as much generality as possible, the
presentations of basic concepts and the derivations of fundamental equations are
couched in terms of the anonymous chemical species A, B, C, U, V, etc. Where it is
appropriate, the specific chemical reactions used in the illustrations are reformulated
in these terms to indicate the manner in which the generalized relations are employed.
Chapters 8 to 13 provide an introduction to chemical reactor design. We start
with the concept of idealized reactors with specified mixing characteristics operating
isothermally and then introduce complications such as the use of combinations of
reactors, implications of multiple reactions, temperature and energy effects, residence
time effects, and heat and mass transfer limitations that ari often involved when
heterogeneous catalysts are employed. Emphasis is placed on the fact that chemical
reactor design represents a straightforward application of the bread and butter tools
of the chemical engineer—the material balance and the energy balance. The
vii
viii Preface
fundamental design equations in the second half of the text are algebraic descendents
of the generalized material balance equation '
Rate of _ Rate of Rate of Rate of disappearance {p ..
input ~~ output' accumulation by reaction
In the case of nonisothermal systems one must write equations of this form both for
energy and for the chemical species of interest, and then solve the resultant equations
simultaneously to characterize the effluent composition and the thermal effects as-
sociated with operation of the reactor. Although the material and energy balance
equations are not coupled when no temperature changes occur in the reactor, the
design engineer still must solve the energy balance equation to ensure that sufficient
capacity for energy transfer is provided so that the reactor will indeed operate
isothermally. The text stresses that the design process merely involves an extension
of concepts learned previously. The application of these concepts in the design
process involves equations that differ somewhat in mathematical form from the
algebraic equations normally encountered in the introductory material and energy
balance course, but the underlying principles are unchanged. The illustrations in-
volved in the reactor design portion of the text are again based where possible on real
chemical examples and actual kinetic data. The illustrative problems in Chapter 13
indicate the facility with which the basic concepts may be rephrased or applied in
computer language, but this material is presented only after the student has been
thoroughly exposed to the concepts involved and has learned to use them in attacking
reactor design problems. I believe that the subject of computer-aided design should
be deferred to graduate courses in reactor design and to more advanced texts.
The notes that form the basis for the bulk of this textbook have been used for
several years in the undergraduate course in chemical kinetics and reactor design at
the University of Wisconsin. In this course, emphasis is placed on Chapters 3 to 6
and 8 to 12, omitting detailed class discussions of many of the mathematical deriva-
tions. My colleagues and I stress the necessity for developing a "seat of the pants"
feeling for the phenomena involved as well as an ability to analyze quantitative
problems in terms of design framework developed in the text.
The material on catalysis and heterogeneous reactions in Chapters 6, \%, and 13
is a useful framework for an intermediate level graduate course in catalysis and
chemical reactor design. In the latter course emphasis is placed on developing the
student's ability to analyze critically actual kinetic data obtained from the literature
in order to acquaint him with many of the traps into which the unwary may fall.
Some of the problems in Chapter 12 and the illustrative case studies in Chapter 1'3
have evolved from this course.
Most of the illustrative examples and problems in the text are based on actual
data from the kinetics literature. However, in many cases, rate constants, heats of
reaction, activation energies, and other parameters have been converted to SI units
from various other systems. To be able to utilize the vast literature of kinetics for
reactor design purposes, one must develop a facility for making appropriate trans-
formations of parameters from one system of urtits to another. Consequently, I have
chosen not to employ SI units exclusively in this text.
Preface ix
Like other authors of textbooks for undergraduates, I owe major debts to the
instructors who first introduced me to this subject matter and to the authors and
researchers whose publications have contributed to my understanding of the subject.
As a student, I benefited from instruction by R. C. Reid, C. N. Satterfield, and
I. Amdur and from exposure to the texts of Walas, Frost and Pearson, and Benson.
Some of the material in Chapter 6 has been adapted with permission from the course
notes of Professor C. N. Satterfield of MIT, whose direct and indirect influence
on my thinking is further evident in some of the data interpretation problems in
Chapters 6 and 12. As an instructor I have found the texts by Levenspiel and Smith
to be particularly useful at the undergraduate level; the books by Denbigh, Laidler,
Hinshelwood, Aris, and Kramers and Westerterp have also helped to shape my
views of chemical kinetics and reactor design. I have tried to use the best ideas of
these individuals and the approaches that I have found particularly useful in the
classroom in the synthesis of this textbook. A major attraction of this subject is that
there are many alternative ways of viewing the subject. Without an exposure to
several viewpoints, one cannot begin to grasp the subject in its entirety. Only after
such exposure, bombardment by the probing questions of one's students, and much
contemplation can one begin to synthesize an individual philosophy of kinetics. To
the humanist it may seem a misnomer to talk in terms of a philosophical approach
to kinetics, but to the individuals who have taken kinetics courses at different schools
or even in different departments and to the individuals who have read widely in the
kinetics literature, it is evident that several such approaches do exist and that
specialists in the area do have individual philosophies that characterize their ap-
proach to the subject.
The stimulating environment provided by the students and staff of the Chemical
Engineering Department at the University of Wisconsin has provided much of the
necessary encouragement and motivation for writing this textbook. The Department
has long been a fertile environment for research and textbook writing in the area of
chemical kinetics and reactor design. The text by O. A. Hougen and K. M. Watson
represents a classic pioneering effort to establish a rational approach to the subject
from the viewpoint of the chemical engineer. Through the years these individuals
and several members of our current staff have contributed significantly to the evolu-
tion of the subject. I am indebted to my colleagues, W. E. Stewart, S. H. Langer,
C. C. Watson, R. A. Grieger, S. L. Cooper, and T. W. Chapman, who have used
earlier versions of this textbook as class notes or commented thereon, to my benefit.
All errors are, of course, my own responsibility.
I am grateful to the graduate students who have served as my teaching assistants
and who have brought to my attention various ambiguities in the text or problem
statements. These include J. F. Welch, A. Yu, R. Krug, E. Guertin, A. Kozinski,
G. Estes, J. Coca, R. Safford, R. Harrison, J. Yurchak, G. Schrader, A. Parker,
T. Kumar, and A. Spence. I also thank the students on whom I have tried out my
ideas. Their response to the subject matter has provided much of the motivation for
this textbook.
Since drafts of this text were used as course notes, the secretarial staff of the
department, which includes D. Peterson, C. Sherven, M. Sullivan, and M. Carr,
Preface
deserves my warmest thanks for typing this material. I am also very appreciative
of my wife's efforts in typing the final draft of this manuscript and in correcting the
galley proofs. Vivian Kehane, Jacqueline Lachmann, and Peter Klein of Wiley were
particularly helpful in transforming my manuscript into this text.
My wife and children have at times been neglected during the preparation of this
textbook; for their cooperation and inspiration I am particularly grateful.
C. G. H., Jr.
Contents
Preface vii
1 Stoichiometric Coefficients and Reaction
Progress Variables 1
2 Thermodynamics of Chemical Reactions 5
3 Basic Concepts in Chemical Kinetics—Determination
of the Reaction Rate Expression 24
4 Basic Concepts in Chemical Kinetics—Molecular
Interpretations of Kinetic Phenomena 76
5 Chemical Systems Involving Multiple Reactions 127
6 Elements of Heterogeneous Catalysis 167
7 Liquid Phase Reactions 215
8 Basic Concepts in Reactor Design and Ideal
Reactor Models 245
9 Selectivity and Optimization Considerations in the
Design of Isothermal Reactors 317
10 Temperature and Energy Effects in Chemical Reactors 349
11 Deviations from Ideal Flow Conditions 388
12 Reactor Design for Heterogeneous Catalytic Reactions 425
13 Illustrative Problems in Reactor Design 540
Appendix A Thermochemical Data 570
Appendix B Fugacity Coefficient Chart 574
Appendix C Nomenclature 575
process is moving, it places a maximum limit on monoxide oxidation reaction in our notation as
the product yield that may be obtained. Chem-
ical engineers must also use thermodynamics to 0 = 2CO 2 - 2CO - O 2
determine heat transfer requirements for pro- instead of in the more conventional form, which
posed reactor configurations. has the reactants on the left side and the products
on the right side.
1.1 BASIC STOICHIOMETRIC CONCEPTS
2CO + O 2 - 2CO 2
where b, c, s, and t are the stoichiometric co- Alternatively, the reaction could be written as
efficients of the species B, C, S, and T, respec-
tively. We define generalized stoichiometric o = co 2 - co - io 2
coefficients vt for the above reaction by rewriting with
it in the following manner.
vCo = - = 1
0 = vBB -f vcC +
' vsS + vTT
(1.1.2) The choice is a matter of personal convenience.
where The essential point is that the ratios of the
vB = -b Vc = S
stoichiometric coefficients are unique for a
given reaction {i.e., v co /v O2 = ( —2/—1) =
vc = —c VT = t
[ —1/( —1/2)] = 2}. Since the reaction stoich-
The generalized stoichiometric coefficients are iometry can be expressed in various ways, one
defined as positive quantities for the products must always write down a stoichiometric equa-
of the reaction and as negative quantities for tion for the reaction under study during the
the reactants. The coefficients of species that are initial stages of the analysis and base subsequent
neither produced nor consumed by the indicated calculations on this reference equation. If a
reaction are taken to be zero. Equation 1.1.2 consistent set of stoichiometric coefficients is
has been written in inverted form with the zero used throughout the calculations, the results
first to emphasize the use of this sign convention, can be readily understood and utilized by other
even though this inversion is rarely used in workers in the field.
practice.
One may further generalize equation 1.1.2 1.1.2 Reaction Progress Variables
by rewriting it as
In order to measure the progress of a reaction
0 = X vtAi (1.1.3) it is necessary to define a parameter, which is a
i
measure of the degree of conversion of the
where the sum is taken over all components Ai reactants. We will find it convenient to use the
present in the system. concept of the extent or degree of advancement
There are, of course, many equivalent ways of of reaction. This concept has its origins in the
writing the stoichiometric equation for a reac- thermodynamic literature, dating back to the
tion. For example, one could write the carbon work of de Donder (1).
1.1 Basic Stoichiometric Concepts
Consider a closed system (i.e., one in which a variable that is simply related to the extent of
there is no exchange of matter between the reaction. The fraction conversion of a reactant
system and its surroundings) where a single At in a closed system in which only a single
chemical reaction may occur according to reaction is occurring is given by
equation 1.1.3. Initially there are ni0 moles of
constituent At present in the system. At some r = ni - nt = 1 - ^ (1.1.7)
later time there are nt moles of species At present. "iO n
i0
At this time the molar extent of reaction is
defined as The variable / depends on the particular species
chosen as a reference substance. In general, the
n•iO
~ (1.1.4) initial mole numbers of the reactants do not
constitute simple stoichiometric ratios, and the
This equation is valid for all species Ah a number of moles of product that may be formed
fact that is a consequence of the law of definite is limited by the amount of one of the reactants
proportions. The molar extent of reaction £ present in the system. If the extent of reaction is
is a time-dependent extensive variable that is not limited by thermodynamic equilibrium
measured in moles. It is a useful measure of the constraints, this limiting reagent is the one that
progress of the reaction because it is not tied determines the maximum possible value of the
to any particular species At. Changes in the extent of reaction (£max). We should refer our
mole numbers of two species j and k can be fractional conversions to this stoichiometrically
related to one another by eliminating £ between limiting reactant if / is to lie between zero and
two expressions that may be derived from unity. Consequently, the treatment used in
equation 1.1.4. subsequent chapters will define fractional con-
versions in terms of the limiting reactant.
nJ0) (1.1.5) One can relate the extent of reaction to the
fraction conversion by solving equations 1.1.4
and 1.1.7 for the number of moles of the limiting
If more than one chemical reaction is possible, reagent nUm and equating the resultant ex-
an extent may be defined for each reaction. If pressions.
£k is the extent of the kth reaction, and vki the
stoichiometric coefficient of species i in reaction
/c, the total change in the number of moles of
species At because of R reactions is given by or
equilibrium differs measurably from £max, we tions, one then arrives at a result that is an
will classify the reaction involved as reversible. extremely good approximation to the correct
From a thermodynamic point of view, all answer.
reactions are reversible. However, when one is
analyzing a reacting system, it is often conve- LITERATURE CITATION
nient to neglect the reverse reaction in order to j . D e Donder, Th., Lemons de Thermodynamique et de
simplify the analysis. For "irreversible" reac- Chemie-Physique, Paris, Gauthier-Villus, 1920.
2 Thermodynamics of
Chemical Reactions
2.0 INTRODUCTION various species in the reaction mixture. If r
reactions may occur in the system and equilib-
The science of chemical kinetics is concerned
rium is established with respect to each of these
primarily with chemical changes and the energy
reactions, it is required that
and mass fluxes associated therewith. Thermo-
dynamics, on the other hand, is concerned with = 0 k = 1, 2,. . . , r (2.0.2)
equilibrium systems . . . systems that are under-
going no net change with time. This chapter
These equations are equivalent to a requirement
will remind the student of the key thermo-
that the Gibbs free energy change for each
dynamic principles with which he should be
reaction (AG) be zero at equilibrium.
familiar. Emphasis is placed on calculations of
equilibrium extents of reaction and enthalpy AG = Y^t^i = 0 at equilibrium (2.0.3)
changes accompanying chemical reactions.
Of primary consideration in any discussion
2.1 CHEMICAL POTENTIALS AND
of chemical reaction equilibria is the constraints
STANDARD STATES
on the system in question. If calculations of
equilibrium compositions are to be in accord The activity at of species i is related to its
with experimental observations, one must in- chemical potential by
clude in his or her analysis all reactions that
occur at appreciable rates relative to the time . = tf + RT En a{ (2.1.1)
frame involved. Such calculations are useful in where
that the equilibrium conversion provides a R is the gas constant
standard against which the actual performance T is the absolute temperature
of a reactor may be compared. For example, if
the equilibrium yield of a given reactant system fii° is the standard chemical potential of species
is 75%, and the observed yield from a given i in a reference state where its activity is taken
reactor is only 30%, it is obviously possible to as unity
obtain major improvements in the process The choice of the standard state is largely
yield. On the other hand, if the process yield arbitrary and is based primarily on experimental
were close to 75%, the potential improvement convenience and reproducibility. The tempera-
in the yield is minimal and additional efforts ture of the standard state is the same as that of
aimed at improving the yield may not be the system under investigation. In some cases,
warranted. Without a knowledge of the equili- the standard state may represent a hypothetical
brium yield, one might be tempted to look for condition that cannot be achieved experi-
catalysts giving higher yields when, in fact, the mentally, but that is susceptible to calculations
present catalyst provides a sufficiently rapid giving reproducible results. Although different
approach to equilibrium. standard states may be chosen for various
species, throughout any set of calculations it is
The basic criterion for the establishment of
important that the standard state of a component
chemical reaction equilibrium is that
be kept the same so as to minimize possibilities
I yifk = 0 (2.0.1) for error.
i
Certain choices of standard states have found
where the fit are the chemical potentials of the such widespread use that they have achieved
Thermodynamics of Chemical Reactions
the status of recognized conventions. In parti- changes in the enthalpy and Gibbs free energy
cular, those listed in Table 2.1 are used in cal- of the system. The enthalpy change on reaction
culations dealing with chemical reaction equili- provides information that is necessary for any
bria. In all cases the temperature is the same as engineering analysis of the system in terms of
that of the reaction mixture. the first law of thermodynamics. It is also useful
in determining the effect of temperature on the
Table 2.1 equilibrium constant of the reaction and thus
Standard States for Chemical Potential Calculations on the reaction yield. The Gibbs free energy is
(for Use in Studies of Chemical Reaction Equilibria) useful in determining whether or not chemical
equilibrium exists in the system being studied
State of and in determining how changes in process
aggregation Standard state variables can influence the yield of the reaction.
Gas
In chemical kinetics there are two types of
Pure gas at unit fugacity (for an ideal
gas the fugacity is unity at 1 atm
processes for which one calculates changes in
pressure; this is a valid approx- these energy functions.
imation for most real gases). 1. A chemical process whereby reactants, each
Liquid Pure liquid in the most stable form at in its standard state, are converted into
1 atm
Solid
products, each in its standard state, under
Pure solid in the most stable form at
1 atm. conditions such that the initial temperature
of the reactants is equal to the final tem-
perature of the products.
Once the standard states for the various 2. An actual chemical process as it might occur
species have been established, one can proceed under either equilibrium or nonequilibrium
to calculate a number of standard energy conditions in a chemical reactor.
changes for processes involving a change from
One must be very careful not to confuse
reactants, all in their respective standard states,
actual energy effects with those that are asso-
to products, all in their respective standard
ciated with the process whose initial and final
states. For example, the Gibbs free energy
states are the standard states of the reactants
change for this process is
and products respectively.
AG° = (2.1.2) In order to have a consistent basis for
comparing different reactions and to permit
where the superscript zero on AG emphasizes the tabulation of thermochemical data for var-
the fact that this is a process involving standard ious reaction systems, it is convenient to define
states for both the final and initial conditions enthalpy and Gibbs free energy changes for
of the system. In a similar manner one can standard reaction conditions. These conditions
determine standard enthalpy (AH0) and stan- involve the use of stoichiometric amounts of
dard entropy changes (AS0) for this process. the various reactants (each in its standard state
at some temperature T). The reaction proceeds
by some unspecified path to end up with com-
2.2 ENERGY EFFECTS ASSOCIATED
plete conversion of reactants to the various
WITH CHEMICAL REACTIONS
products (each in its standard state at the same
Since chemical reactions involve the formation, temperature T).
destruction, or rearrangement of chemical The enthalpy and Gibbs free energy changes
bonds, they are invariably accompanied by for a standard reaction are denoted by the
2.2 Energy Effects Associated with Chemical Reactions
symbols AH0 and AG°, where the superscript called the enthalpy (or heat) of formation of the
zero is used to signify that a "standard" reaction compound and is denoted by the symbol AH°f.
is involved. Use of these symbols is restricted Thus,
to the case where the extent of reaction is 1 mole
Ai/? veralI = AiJ°walert/, + A//°propvlene - AiJ°propanoI(/,
for the reaction as written. The remaining
discussion in this chapter refers to this basis. (2.2.3)
and
Because G and H are state functions, changes
in these quantities are independent of whether AGo°verall = AG°walert/) + AG°propyIenc - AG°propanol(/l
the reaction takes place in one or in several (2.2.4)
steps. Consequently, it is possible to tabulate
data for relatively few reactions and use this where AG° refers to the standard Gibbs free
data in the calculation of AG° and AH0 for other energy of formation.
reactions. In particular, one tabulates data for This example illustrates the principle that
the standard reactions that involve the forma- values of AG° and AH0 may be calculated from
tion of a compound from its elements. One may values of the enthalpies and Gibbs free energies
then consider a reaction involving several of formation of the products and reactants. In
compounds as being an appropriate algebraic more general form,
sum of a number of elementary reactions, each AH0 = YJvi AH°fJ (2.2.5)
of which involves the formation of one com- i
For the overall reaction, the determination of the standard Gibbs free
energy and enthalpy of formation of its com-
AH0 = A//? + AH°2 + AH°3 (2.2.1) pounds. If AH0 is negative, the reaction is said
AG° = AG? + AG°2 + AG^ (2.2.2) to be exothermic; if AH0 is positive, the reaction
is said to be endothermic.
However, each of the individual reactions It is not necessary to tabulate values of AG°
involves the formation of a compound from or AH0 for all conceivable reactions. It is
its elements or the decomposition of a com- sufficient to tabulate values of these parameters
pound into those elements. The standard en- only for the reactions that involve the formation
thalpy change of a reaction that involves the of a compound from its elements. The problem
formation of a compound from its elements is of data compilation is further simplified by the
Thermodynamics of Chemical Reactions
fact that it is unnecessary to record AG°f and related to AG°/T and that equation 2.2.8 is
AH°f at all temperatures, because of the rela- useful in determining how this parameter varies
tions that exist between these quantities and with temperature. If one desires to obtain an
other thermodynamic properties of the reactants expression for AG° itself as a function of tem-
and products. The convention that is commonly perature, equation 2.2.7 may be integrated to
accepted in engineering practice today is to give AH0 as a function of temperature. This
report values of standard enthalpies of formation relation may then be used with equation 2.2.8
and Gibbs free energies of formation at 25 °C to arrive at the desired relation.
(298.16 °K) or at 0 °K. The problem of calculat- The effect of pressure on AG° and AH0
ing a value for AG° or AH0 at temperature T thus depends on the choice of standard states em-
reduces to one of determining values of AGJ ployed. When the standard state of each com-
and AH° at 25 °C or 0 °K and then adjusting ponent of the reaction system is taken at 1 atm
the value obtained to take into account the pressure, whether the species in question is a
effects of temperature on the property in ques- gas, liquid, or solid, the values of AG° and AH0
tion. The appropriate techniques for carrying refer to a process that starts and ends at 1 atm.
out these adjustments are indicated below. For this choice of standard states, the values of
The effect of temperature on AH0 is given by AG° and AH0 are independent of the system
pressure at which the reaction is actually carried
out. It is important to note in this connection
that we are calculating the enthalpy change for
(2.2.7) a hypothetical process, not for the actual process
as it occurs in nature. This choice of standard
where C°pi is the constant pressure heat capacity states at 1 atm pressure is the convention that is
of species i in its standard state. customarily adopted in the analysis of chemical
In many cases the magnitude of the last term reaction equilibria.
on the right side of equation 2.2.7 is very small For cases where the standard state pressure
compared to AH%9%A6. However, if one is to be for the various species is chosen as that of the
able to evaluate properly the standard heat of system under investigation, changes in this
reaction at some temperature other than variable will alter the values of AG° and AH0.
298.16 °K, one must know the constant pressure In such cases thermodynamic analysis indicates
heat capacities of the reactants and the products that
as functions of temperature as well as the heat
of reaction at 298.16 °K. Data of this type and
techniques for estimating these properties are
contained in the references in Section 2.3. (2.2.9)
The most useful expression for describing
where Vt is the molal volume of component i in
the variation of standard Gibbs free energy
its standard state and where each integral is
changes with temperature is:
evaluated for the species in question along an
isothermal path. The term in brackets represents
AG°
the variation of the enthalpy of a component
T AH0 with pressure at constant temperature (dH/dP)T.
(2.2.8)
dT It should be emphasized that the choice of
standard states implied by equation 2.2.9 is not
In Section 2.5 we will see that the equilibrium that which is conventionally used in the analysis
constant for a chemical reaction is simply of chemically reacting systems. Furthermore,
2.4 The Equilibrium Constant and its Relation to A(7°
in the vast majority of cases the summation term The following references contain techniques for
on the right side of the equation is very small estimating thermochemical data.
compared to the magnitude of AH°,dim and,
1. R. C. Reid and T. K. Sherwood, The Properties of Gases
indeed, is usually considerably smaller than the
and Liquids, Second Edition, McGraw-Hill, New York,
uncertainty in this term. 1966.
The Gibbs free energy analog of equation 2. S. W. Benson, Thermochemical Kinetics, Wiley, New
2.2.9 is York, 1968.
3. G. J. Janz, Estimation of Thermodynamic Properties of
AGg = AG? atm + (2.2.10) Organic Compounds, Academic Press, New York, 1958.
where the integral is again evaluated along an 2.4 THE EQUILIBRIUM CONSTANT AND
isothermal path. For cases where the species
ITS RELATION TO AG°
involved is a condensed phase, Vt will be a very
small quantity and the contribution of this The basic criterion for equilibrium with respect
species to the summation will be quite small to a given chemical reaction is that the Gibbs
unless the system pressure is extremely high. free energy change associated with the progress
For ideal gases, the integral may be evaluated of the reaction be zero.
directly as RT In P. For nonideal gases the
integral is equal to RT In / ? , where / ? is the AG = = 0 (2.4.1)
fugacity of pure species i at pressure P.
The standard Gibbs free energy change for a
reaction refers to the process wherein the
2.3 SOURCES OF THERMOCHEMICAL reaction proceeds isothermally, starting with
DATA stoichiometric quantities of reactants each in its
Thermochemical data for several common spe- standard state of unit activity and ending with
cies are contained in Appendix A. Other useful products each at unit activity. In general it is
standard references are listed below. nonzero and given by
1. F. D. Rossini, et al., Selected Values of Physical and AG° = £ v^ 0 (2.4.2)
Thermodynamic Properties of Hydrocarbons and Related i
Compounds, Carnegie Press, Pittsburgh, 1953; also loose-
leaf supplements. Data compiled by Research Project 44 Subtraction of equation 2.4.2 from equation
of the American Petroleum Institute. 2.4.1 gives
2. F. D. Rossini, et al., "Selected Values of Chemical
Thermodynamic Properties," National Bureau of Stan- AG - AG° = X MfjLt - /i?) (2.4.3)
dards, Circular 500 and Supplements, 1952. i
3. E. W. Washburn (Editor), International Critical Tables, This equation may be rewritten in terms of the
McGraw-Hill, New York, 1926.
activities of the various species by making use of
4. T. Hilsenrath, et al., "Thermal Properties of Gases,"
equation 2.1.1.
National Bureau of Standards Circular 564, 1955.
5. D. R. Stull and G. C. Sinke, "Thermodynamic Properties
of the Elements," Adv. Chem. Ser., 18, 1956. AG - AG° = RT X vf In at = RT In (]J a]*
6. Landolt-Bornstein Tabellen, Sechste Auflage, Band II,
Teil 4, Springer-Verlag, Berlin, 1961.
(2.4.4)
7. Janaf Thermochemical Tables, D. R. Stull, Project Direc- where the \\ symbol denotes a product over i
tor, PB 168370, Clearinghouse for Federal Scientific and i
Technical Information, 1965. species of the term that follows.
10 Thermodynamics of Chemical Reactions
volumes of the reactants and products. However, If all of the species are gases, combination of
this choice of standard states is extremely rare equations 2.6.1, 2.6.2, and 2.6.4 gives
in engineering practice.
(S) (2.6.1)
equilibrium constant Ka, the term Kf/P is
affected by changes in the system pressure as
well as by changes in temperature.
In a system that involves gaseous components, The product of Ky and ps + t~b~c is assigned
one normally chooses as the standard state the the symbol KP.
pure component gases, each at unit fugacity
(essentially 1 atm). The activity of a gaseous KP = s+t_6_c _
species B is then given by
(2.6.6)
+t-b-c
K K (2.6.9)
c
cj \nB + nc + nnTT
ris -r -f
This equation is extremely useful for cal- 264.2 atm. Assume that the only significant
culating the equilibrium composition of the chemical reaction is
reaction mixture. The mole numbers of the
various species at equilibrium may be related to H2O(g) + C 2 H 4 (0) ^ C 2 H 5 OH(#)
their values at time zero using the extent of Use only the following data and the fugacity
reaction. When these values are substituted into coefficient chart.
equation 2.6.9, one has a single equation in a
single unknown, the equilibrium extent of re-
action. This technique is utilized in Illustration Compound Tc (°K) Pc (atm)
2.1. If more than one independent reaction is
occurring in a given system, one needs as many H2O(g) 647.3 218.2
equations of the form of equation 2.6.9 as there C2H4(g) 283.1 50.5
are independent reactions. These equations are C 2 H 5 OH(#) 516.3 63.0
then written in terms of the various extents of
reaction to obtain a set of independent equa-
tions equal to the number of unknowns. Such a
system is considered in Illustration 2.2.
Compound AGj 298 16 (kcal) A//? 298 . 16 (kcal)
If one assumes that AH0 is independent of From the stoichiometry of the reaction it is
temperature, this equation may be integrated to possible to determine the mole numbers of the
give various species in terms of the extent of reaction
and their initial mole numbers.
l
= ni
2
100.0 - (0.280) 1
;.74 x 10~ 3 = •
/ 60.0 - A / 25.0 - (0.190)(0.885) (264.2)
\ 100.0 - U 1100.0 -
or
£(100.0 -
(60.0 - 0(25.0 _
This equation is quadratic in £. The solution is rium constant Ka. From equation 2.5.2 it follows
£ = 10.8. On the basis of 100 moles of starting that the equilibrium conversion is decreased as
material, the equilibrium composition is as the temperature increases for exothermic reac-
follows. tions. The equilibrium yield increases with in-
creasing temperature for .endothermic reactions.
Temperature changes also affect the value of
Mole numbers Mole percentages
Kf/p. The changes in this term, however, are
N2 15.0 16.8 generally very small compared to those in Ka.
H2O 49.2
C2H4 14.2 15.9 2.7.2 Effect of Total Pressure
C 2 H 5 OH 10.8 12.1
Total 89.2 100.0 The equilibrium constant Ka is independent of
pressure for those cases where the standard states
are taken as the pure components at 1 atm. This
2.7 THE EFFECT OF REACTION
case is the one used as the basis for deriving equa-
CONDITIONS ON EQUILIBRIUM YIELDS
tion 2.6.9. TJie effect of pressure changes then
Equation 2.6.9 is an extremely useful relation for appears in the terms Kf/P and _p + t+-••-*-*•-^
determining the effects of changes in process The influence of pressure on KffP is quite small.
parameters on the equilibrium yield of a given However, for cases where there is no change in
product in a system in which only a single gas the total number of gaseous moles during the
phase reaction is important. It may be rewritten reaction, this is the only term by which pressure
as changes affect the equilibrium yield. For these
s+t-b-c
K
a nc ns (2.7.1)
nbBncc K f/p
Any change that increases the right side of cases the value of Kf/P should be calculated
equation 2.7.1 will increase the ratio of products from the fugacity coefficient charts for the sys-
to reactants in the equilibrium mixture and thus tem and conditions of interest in order to deter-
correspond to increased conversions. mine the effect of pressure on the equilibrium
2.8 Heterogeneous Reactions 15
yield. For those cases where the reaction pro- agent above that which would be obtained with
duces a change in the total number of gaseous stoichiometric ratios of the reactants.
species in the system, the term that has the
largest effect on the equilibrium yield of products 2.8 HETEROGENEOUS REACTIONS
is Ps + t+ b-c--\ Thus, if a reaction produces a
decrease in the total number of gaseous com- The fundamental fact on which the analysis of
ponents, the equilibrium yield is increased by an heterogeneous reactions is based is that when a
increase in pressure. If the total number of gas- component is present as a pure liquid or as a pure
eous moles is increased by reaction, the equilib- solid, its activity may be taken as unity, provided
rium yield decreases as the pressure increases. the pressure on the system does not differ very
much from the chosen standard state pressure.
2.7.3 Effect of Addition of Inert Gases At very high pressures, the effect of pressure on
solid or liquid activity may be determined using
The only term in equation 2.7.1 that is influenced the Poynting correction factor.
by the addition of inert gases is nl. Thus, for
reactions in which there is no change in the total p V dP
(2.8.1)
number of gaseous moles, addition of inerts has RT
no effect on the equilibrium yield. For cases
where there is a change, the effect produced by where V is the molal volume of the condensed
addition of inert gases is in the same direction as phase. The activity ratio is essentially unity at
that which would be produced by a pressure moderate pressures.
decrease. If we now return to our generalized reaction
2.4.5 and add to our gaseous components B, C,
S, and T a pure liquid or solid reactant D and a
2.7.4 Effect of Addition of Catalysts
pure liquid or solid product U with stoichio-
The equilibrium constant and equilibrium yield metric coefficients d and u, respectively, the re-
are independent of whether or not a catalyst is action may be written as
present. If the catalyst does not remove any of The equilibrium constant for this reaction is
the passive restraints that have been placed on
the system by opening up the possibility of addi- (2.8.3)
tional reactions, the equilibrium yield will not 4«C«B
be affected by the presence of this material.
When the standard states for the solid and
liquid species correspond to the pure species at
2.7.5 Effect of Excess Reactants
1 atm pressure or at a low equilibrium vapor
If nonstoichiometric amounts of reactants are pressure of the condensed phase, the activities of
present in the initial system, the presence of the pure species at equilibrium are taken as unity
excess reactants tends to increase the equilib- at all moderate pressures. Consequently, the gas
rium fractional conversion of the limiting re- phase composition at equilibrium will not be
16 Thermodynamics of Chemical Reactions
affected by the amount of solid or liquid present. present at equilibrium. All of these reactions
At very high pressures equation 2.8.1 must be must be considered in calculating the equilib-
used to calculate these activities. When solid or rium composition.
liquid solutions are present, the activities of the In order to arrive at a consistent set of relation-
components of these solutions are no longer ships from which complex reaction equilibria
unity even at moderate pressures. In this case may be determined, one must develop the same
one needs data on the activity coefficients of the number of independent equations as there are
various species and the solution composition in unknowns. The following treatment indicates
order to determine the equilibrium composition one method of arriving at a set of chemical reac-
of the system. tions that are independent. It has been adopted
from the text by Aris (1).*
If R reactions occur simultaneously within a
2.9 EQUILIBRIUM TREATMENT OF
system composed of S species, then one has R
SIMULTANEOUS REACTIONS
stoichiometric equations of the form
The treatment of chemical reaction equilibria
outlined above can be generalized to cover the
situation where multiple reactions occur simul- £ v ^ . = 0 fc= 1 , 2 , . . . , / * (2.9.1)
i= 1
taneously. In theory one can take all conceivable
reactions into account in computing the com- where vki is the stoichiometric coefficient of
position of a gas mixture at equilibrium. How- species i in reaction k.
ever, because of kinetic limitations on the rate of Since the same reaction may be written with
approach to equilibrium of certain reactions, different stoichiometric coefficients, the impor-
one can treat many systems as if equilibrium is tance of the coefficients lies in the fact that the
achieved in some reactions, but not in others. In ratios of the coefficients of two species must be
many cases reactions that are thermodynami- identical no matter how the reaction is written.
cally possible do not, in fact, occur at appreciable Thus the stoichiometric coefficients of a reaction
rates. are given up to a constant multiplier X. The
In practice, additional simplifications occur equation
because at equilibrium many of the possible re-
actions occur either to a negligible extent, or £ foktA, = 0 (2.9.2)
i
else proceed substantially to completion. One
criterion for determining if either of these condi- has the same meaning as equation 2.9.1, pro-
tions prevails is to examine the magnitude of vided that X is nonzero. If three or more reactions
the equilibrium constant in question. If it is many can be written for a given system, one must test
orders of magnitude greater than unity, the re- to see if any is a multiple of one of the others and
action may be said to go to completion. If it is if any is a linear combination of two or more
orders of magnitude less than unity, the reaction others. We will use a set of elementary reactions
may be assumed to go to a negligible extent. In representing a mechanism for the H 2 — Br2
either event, the number of chemical species that reaction as a vehicle for indicating how one
must be considered is reduced and the analysis is may determine which of a set of reactions are
thereby simplified. After the simplifications are independent.
made, there may still remain a group of reactions
whose equilibrium constants are neither very * Rutherford Aris, Introduction to the Analysis of Chemical
small nor very large, indicating that appreciable Reactors, copyright 1965, pp. 10-13. Adapted by permis-
amounts of both products and reactants are sion of Prentice-Hall, Inc., Englewood Cliffs, NJ.
2.9 Equilibrium Treatment of Simultaneous Reactions 17
pressure is 25 atm. The following reactions are The values of KP for reactions A and B are
to be considered. given by
2CO + 2H 2 ^± CH 4 + CO 2 (A)
p2 p2 ,,2 ,,2 p2
CO + 3H 2 - CH 4 + H 2 O (B) r
COrH2
CO 2 + H 2 ^ H 2 O + CO (C)
•r.tf p p3
When the equilibrium constants for reactions CO H
A and B are expressed in terms of the partial The mole fractions of the various species may
pressures of the various species (in atmospheres), be expressed in terms of £A and £B so that the
the equilibrium constants for these reactions
above expressions for KP become
have the following values.
ZA
KPA = 0.046 KPB = 0.034
4 - 2^ - 2 4 - 2^ -
Determine the equilibrium composition of the
mixture.
Solution
2.10 SUPPLEMENTARY READING Tabulate the effects of the changes below on:
REFERENCES (a) Reaction velocity.
(b) Equilibrium constant.
The following texts contain adequate discussions
(c) Equilibrium degree of conversion.
of the thermodynamics of chemical reactions;
(d) Actual degree of conversion obtained in a
they can be recommended without implying
very short time interval at the start of the
judgment on others that are not mentioned.
reaction.
1. K. G. Denbigh, The Principles of Chemical Equilibrium, Variables:
Third Edition, Cambridge University Press, Cambridge,
(1) Increase in temperature.
1971.
(2) Increase of pressure.
2. O. A. Hougen, K. M. Watson, and R. A. Ragatz, Chemical
Process Principles, Part I, Material and Energy Balances
(3) Substitution of a less active catalyst.
and Part II, Thermodynamics, Second Edition, Wiley, (4) Dilution with argon at constant total
New York, 1954, 1959. pressure.
3. I. Prigogine and R. Defay, Chemical Thermodynamics, (5) Increase in the Reynolds number of the fluid
translated by D. H. Everett, Wiley, New York, 1954. in the reactor.
4. F. T. Wall, Chemical Thermodynamics, Second Edition, Appropriate tabular entries are: (increases), (de-
W. H. Freeman, San Francisco, 1965. creases), (no effect), (indeterminate).
5. K. S. Pitzer and L. Brewer, revision of Thermodynamics,
by G. N. Lewis and M. Randall, McGraw-Hill, New 2. Consider the ammonia synthesis reaction
York, 1961. being carried out at 450 °C and 101.3 MPa.
6. M. Modell and R. C. Reid, Thermodynamics and Its
Applications, Prentice-Hall, Englewood Cliffs, N. J., 1974. IN2 + | H 2 - NH 3
The feed stream consists of 60 mole percent
LITERATURE CITATIONS hydrogen, 20% nitrogen, and 20% argon. Calcu-
1. Aris, R., Introduction to the Analysis of Chemical Reactors, late the composition of the exit gases, assuming
Prentice-Hall, Englewood Cliffs, N J , copyright © 1965. equilibrium is achieved in the reactor. Make
Used with permission. sure that you take deviations from the ideal gas
2. Strickland-Constable, R. F., "Chemical Thermodynam- law into account. The equilibrium constant ex-
ics," in Volume 8 of Chemical Engineering Practice, pressed in terms of activities relative to standard
edited by H. W. Cremer and S. B. Watkins, Butterworths,
London, 1965. Used with permission.
states at 1 atm may be assumed to be equal to
8.75 x 10" 3 . Thefugacity of pure H 2 at 450 °C
PROBLEMS and 101.3 MPa may be assumed to be equal to
136.8 MPa.
1. In the presence of an appropriate catalyst, ac-
etone reacts with hydrogen to give isopropanol. 3. A new process for the manufacture of acety-
lene has been proposed. The process will involve
CH3COCH3 + H 2 ^ CH3CHOHCH3 the dehydrogenation of ethane over a suitable
catalyst (yet to be found). Pure ethane will be
Standard enthalpies and Gibbs free energies of
fed to a reactor and a mixture of acetylene,
formation at 25 °C and 101.3 kPa are as follows.
hydrogen, and unreacted ethane will be with-
drawn. The reactor will operate at 101.3 kPa
AH°f (kJ/mole) AG°f (kJ/mole) total pressure and at some as yet unspecified
temperature T.
Acetone (g) -216.83 -152.61 The reaction
Isopropanol (g) -261.30 -159.94
C 2 H 6 -> C 2 H 2 2H 2
20 Thermodynamics of Chemical Reactions
may be assumed to be the only reaction occur- is obtained, what is the composition of the
ring in the proposed reactor. The following data effluent mixture? The reactor may be as-
on the heats of formation of ethane and acety- sumed to operate isothermally at 101.3 kPa
lene are available. total pressure. Neglect any other reactions
that may occur.
AG/298 AHf2Q8
(kJ/mole) (kJ/mole)
4. As a thermodynamicist working at the Lower
Slobbovian Research Institute, you have been
Ethane -32.908 -84.724 asked to determine the standard Gibbs free en-
Acetylene 209.340 226.899 ergy of formation and the standard enthalpy of
formation of the compounds ds-butene-2 and
£raws-butene-2. Your boss has informed you that
The standard states of these materials are taken the standard enthalpy of formation of butene-1 is
as the pure components at 298 °K and a pressure 1.172 kJ/mole while the standard Gibbs free
of 101.3 kPa. The following data on the absolute energy of formation is 72.10 k J/mole where the
entropies of the hydrocarbons at 298 °K are standard state is taken as the pure component
available. at 25 °C and 101.3 kPa.
Your associate, Kem Injuneer, has been
testing a new catalyst for selective butene isom-
Entropy
erization reactions. He says that the only re-
(J/mole • K)
actions that occur to any appreciable extent over
Ethane 229.65 this material are:
Acetylene 200.95
butene-1 ^ ds-butene-2
ds-butene-2 ^ trarcs-butene-2
(a) What is the standard Gibbs free energy
change at 25 °C for the reaction as written? He has reported the following sets of data from
(b) What is the standard enthalpy change at his system as being appropriate for equilibrium
25 °C for the reaction as written? conditions.
(c) What is the absolute entropy of gaseous
hydrogen at 25 °C and a pressure of 101.3 Run I
kPa? Use only the above data in evaluating Reactor pressure 53.33 kPa
this quantity. Reactor temperature 25 °C
(d) What is the equilibrium constant for the Gas composition (mole percent):
reaction at 25 °C? butene-1 3.0
(e) If the standard enthalpy change may be as- cis-butene-2 21.9
sumed to be essentially independent of tem- trans-butene-2 75.1
perature, what is the equilibrium constant
for the reaction at 827 °C? Run II
Reactor pressure 101.3 kPa
(f) If we supply pure ethane at 827 °C and a
Reactor temperature 127 °C
pressure of 101.3 kPa to the reactor de-
Gas composition (mole percent):
scribed above and if equilibrium with respect butene-1 8.1
to the reaction cis-butene-2 28.8
trans-butQnQ-2 63.1
C 22 H 6 C2H2 2H 2
Problems 21
flowing it over a powerful catalyst at 1000 °K 11. Methanol may be synthesized from hydro-
and at various pressures. Over what range of gen and carbon monoxide in the presence of an
pressures may carbon deposit on the catalyst? appropriate catalyst.
Over what range of pressures will it not deposit?
CO + 2H 2 ^± CH 3 OH
Note. Assume that only the reactions given in
the table occur, and that equilibrium is attained. If one has a feed stream containing these gases in
stoichiometric proportions (H 2 /CO = 2) at 200
X atm and 275 °C, determine the effluent composi-
tion from the reactor:
3.16 (a) If it operates isothermally and equilibrium
C + 2H 2 O ^ CO 2 + 2H 2 5.01 is achieved.
CO 2 + C ^ 2CO 2.00 (b) If it operates adiabatically and equilibrium
CO + H 2 O ^ C O 2 + H 2 1.58 is achieved. (Also determine the effluent
temperature.)
For an initial ratio of CO 2 to H 2 of unity, what Data:
must the pressure be if exactly 30% of the carbon (1) AH0 = -17,530 - 18.19T + 0.0141T2 for
present in the feed can precipitate as solid AH0 in calories per gram mole and T in
carbon? degrees Kelvin.
The equilibrium constants are based on a (2) Molal heat capacities at constant pressure
standard state of unit fugacity for the gaseous H2 Cp = 6.65 + 0.00070 T
species and on a standard state corresponding CO Cp = 6.89 + O.OOO38 T
to the pure solid for carbon.
CH3OH Cp = 2.0 + 0.03T
10. Butadiene can be produced by the dehydro- for Cp in calories per gram mole-degree
genation of butene over an appropriate catalyst. Kelvin.
C4H8 (3) Equilibrium constant expressed in terms of
H2 C4H6
fugacities
In order to moderate the thermal effects asso-
ciated with the reaction, large quantities of 3835
Kf = - 9.150 l o g 1 0 T
steam are usually injected with the butene feed.
Steam/butene ratios of 10 to 20 are typical of the
+ 3.08 x 10" 3 T + 13.20
conditions employed in many industrial reac-
tors. If equilibrium is achieved within the reactor, (4) Note that in part (b) a trial and error solu-
determine the effluent composition correspond- tion is required. (Hint. The effluent tempera-
ing to the conditions enumerated below. ture will be close to 700 CK.)
12.* In a laboratory investigation a high-pres- From these experimental data and the physi-
sure gas reaction A ^ 2B is being studied in a cal properties listed below:
flow reactor at 200 °C and 10.13 MPa. At the end (a) Calculate the composition of the gas stream
of the reactor the gases are in chemical equilib- before the sampling valve.
l
rium, and their composition is desired. (b) Are the gases at, near, or far from chemical
Unfortunately, to make any analytical mea- equilibrium after the sampling valve? (Show
surements on this system, it is necessary to bleed definite proof for your answer.)
off a small side stream through a low-pressure Data and Allowable Assumptions:
conductivity cell operating at 1 atm. It is found (1) Heat of the reaction, AH = 29.31 kJ/mole of
that when the side stream passes through the A reacting, independent' of temperature.
sampling valve, the temperature drops to 100 °C (2) Heat capacities: 58.62 J/mole-°K for A
arid the conductivity cell gives compositions of and 29.31 J/mole-°K for B, independent of
yA = 0.5 and yB = 0.5 (mole fractions). temperature.
(3) The gas mixture is ideal at all pressures,
temperatures, and compositions.
: (4) The flow through the sampling valve is a true
Adapted from Michael Modell and Robert C. Reid,
Thermodynamics and Its Applications, copyright © 1974. Joule-Thompson expansion.
Reprinted by permission of Prentice-Hall, Inc. Englewood (5) Assume no heat loss in the sampling line or
Cliffs, N. J. across sampling valve.
Concepts in Chemical
3 Kinetics—Determination of the
Reaction Rate Expression
3.0 INTRODUCTION metric system of units, (e.g., mole/liter • sec or
molecules/cm3-sec).
This chapter defines a number of terms that are
Changes in the mole numbers nt of the various
used by the chemical kineticist and treats some
species involved in a reaction are related to the
of the methods employed in the analysis of
extent of reaction by equation 1.1.4
laboratory data to determine empirical rate ex-
pressions for the systems under investigation. P =
Hi
^ (3.0.2)
It is convenient to approach the concept of
reaction rate by considering a closed, isothermal, Differentiation of this equation with respect to
constant pressure homogeneous system of uni- time and substitution in equation 3.0.1 gives
form composition in which a single chemical
reaction is taking place. In such a system the rate
of the chemical reaction (r) is defined as:
If one defines the rate of increase of the moles
r (3.0.1) of species / as
V dt
r,= - V~dt (3.0.4)
where
V = system volume then
£ = extent of reaction r
i = (3.0.5)
t = time :itivp for nrr»Hn<M c anr\
Several facts about this definition should be negative for reactants, and since the reaction rate
noted. r is intrinsically positive, the various rt will have
the same sign as the corresponding vf and dnjdt
1. The rate is defined as an intensive variable. will have the appropriate sign (i.e., positive for
Note that the reciprocal of system volume is products and negative for reactants).
outside the derivative term. This considera- In the analysis of engineering systems, one
tion is important in treating variable volume frequently encounters systems whose properties
systems. vary from point to point within the system. Just
2. The definition is independent of any parti- as it is possible to define local temperatures,
cular reactant or product species. pressures, concentrations, etc., it is possible to
3. Since the reaction rate almost invariably generalize equations 3.0.1 and 3.0.4 to define
changes with time, it is necessary to use the local reaction rates.
time derivative to express tfye instantaneous In constant volume systems it is convenient
rate of reaction. to employ the extent per unit volume £*
Many sets of units may be used to measure
(3.0.6)
reaction rates. Since the extent of reaction is b
V
expressed in terms of moles, the reaction rate has
and to denote the rate in such cases as
the units of moles transformed per unit time per
unit volume. The majority of the data reported (3.0.7)
in the literature is expressed in some form of the V dt dt
24
3.0 Introduction 25
the bulk fluid properties. The conversion rate rium, the rate expressions can generally be
takes into account the effects of both chemical written in the form
and physical rate processes. The intrinsic rate
r = /c0(Q) (3.0.13)
may be thought of as the conversion rate that
would exist if all physical rate processes occurred where </>(Q) is a function that depends on the
at infinitely fast rates. concentrations (Q) of the various species
Chapter 12 treats situations where both present in the system (reactants, products,
physical and chemical rate processes influence catalysts, and inhibitors). This function (j){Ci)
the conversion rate; the present chapter is con- may also depend on the temperature. The
cerned only with those situations where physical coefficient k is called the reaction rate con-
rate processes are unimportant. This approach stant. It does not depend on the composition
permits us to focus our concern on the variables of the system and is consequently indepen-
that influence intrinsic chemical reaction rates dent of time in an isothermal system.
(i.e., temperature, pressure, composition, and the 5. The rate constant k generally varies with the
presence or absence of catalysts in the system). absolute temperature T of the system accord-
In reaction rate studies one's goal is a pheno- ing to the law proposed by Arrhenius.
menological description of a system in terms of a
k = Ae'E/RT (3.0.14)
limited number of empirical constants. Such
where
descriptions permit one to predict the time-
E is the apparent activation energy of the
dependent behavior of similar systems. In these
reaction
studies the usual procedure is to try to isolate the
R is the gas constant
effects of the different variables and to investigate
A is the preexponential factor, sometimes
each independently. For example, one encloses
called the frequency factor, which is as-
the reacting system in a thermostat in order to
sumed to be a temperature independent
maintain it at a constant temperature.
quantity
Several generalizations can be made about the
6. Very often the function </>(C;) in equation
variables that influence reaction rates. Those
3.0.13 is temperature independent and, to a
that follow are in large measure adapted from
high degree of approximation, can be written
Boudart's text (1).
as
1. The rate of a chemical reaction depends on
the temperature, pressure, and composition
where the product f| is taken over all com-
of the system under investigation.
ponents of the system. The exponents /?£ are
2. Certain species that do not appear in the
the orders of the reaction with respect to each
stoichiometric equation for the reaction
of the i species present in the system. The
under study can markedly affect the. reaction
algebraic sum of the exponents is called the
rate, even when they are present in only trace
total order or overall order of the reaction.
amounts. These materials are known as
7. If one considers a system in which both for-
catalysts or inhibitors, depending on whether
ward and reverse reactions are important, the
they increase or decrease the reaction rate.
net rate of reaction can generally be expressed
3. At a constant temperature, the rate of reaction as the difference between the rate in the for-
generally decreases monotonically with time ward direction f and that in the opposite
or extent of reaction. direction f.
4. If one considers reactions that occur in
systems that are far removed from equilib- r = r — r (3.0.16)
3.0 Introduction 27
rate is proportional to the product of the con- For Class II second-order rate expressions of
centrations of two different species. the form of equation 3.1.10, the rate can be
2 expressed in terms of the extent of reaction per
Class I: rr —
— kC
K^A (3.1.9)
unit volume as
Class II: r = kC4CR (3.1.10)
For Class I second-order rate expressions, rv = ^ = k(CA0 + v A£*)(C +
B0
combining equations 3.0.7, 3.1.2, and 3.1.9 gives
= C = Ci
i0 - vAkCAOt
Since two of the terms within the brackets in
equation 3.1.17 are the residual reactant con-
If vA is equal to — 1, equation 3.1.13 becomes centrations at time t or extent £*, it is often
useful to rewrite this equation as
1 1 mis)v
CA " CA0 " C \ IC \~1
7^ 7^ = ( C ^ o V B - CB0VA)kt
In testing experimental data to see if it fits this
type of rate expression, one plots 1/CA versus t.
If the data fall on a straight line, the rate expres- or
sion is of the form of equation 3.1.9, and the
slope and intercept of the line are equal to
— vAk and 1/CAO, respectively.
Many second-order reactions follow Class I (3.1.20)V
rate expressions. Among these are the gas-phase These equations are convenient for use in de-
thermal decomposition of hydrogen iodide termining if experimental rate data follow Class
(2HI -• H 2 + I 2 ), dimerization of cyclopen- II second-order kinetics in that they predict a
tadiene (2C 5 H 6 -> C 1 0 H 1 2 ), and the gas phase linear relationship between ln(CB/CA) and time.
thermal decomposition of nitrogen dioxide The y intercept is ln(CB0/CA0) and the slope is
(2NO 2 -* 2NO + O 2 ). {CAOvB - CBOvA)k.
30 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
Class II second-order rate expressions are one Class II Third-Order Rate Expression:
of the most common forms one encounters in the
laboratory. They include the gas phase reaction 1 1 1 '
of molecular hydrogen and iodine (H 2 -f I 2 -• ^AO r r
V
A
2HI), the reactions of free radicals with mole-
cules (e.g., H + Br2 -> HBr -f Br), and the hy-
= (CAOvB - CBOvA)kt (3.1.26)V
drolysis of organic esters in nonaqueous media.
In this case one obtains an expression for <!;*
3.1.1.3 Third-Order Reactions in Constant Vol- that cannot be manipulated to yield a simple
ume Systems. Third-order reactions can be algebraic form. However, if the concentration
classified into three primary types, according to of one species is known as a function of time, the
the general definition. concentrations of all other species may be deter-
Class I: (3.1.21) mined from the definition of the extent of reac-
tion per unit volume; that is,
Class II: r = kC\CB (3.1.22)
Class III: r = kCACBCQ (3.1.23) Ci - Ci0
(3.1.27)V
V;
If one uses reactants in precisely stoichio-
metric concentrations, the Class II and Class III Hence,
rate expressions will reduce to the mathematical
C
form of the Class I rate function. Since the mathe- J = CJo + ~(Ci - Cl0) (3.1.28)V
matical principles employed in deriving the re-
lation between the extent of reaction or the Class III Third-Order Rate Expression:
Q
- vBCQ0) In
Cr.
concentrations of the various species and time The last equation has extremely limited
are similar to those used in Sections 3.1.1.1 and utility and is presented more as a subject of
3.1.1.2, we will list only the most useful results. academic interest than of practicality.
Gas phase third-order reactions are rarely
Class I Third-Order Rate Expression:
encountered in engineering practice. Perhaps
1 1 the best-known examples of third-order reac-
2 = -2vAkt (3.1.24)V tions are atomic recombination reactions in the
C,40
presence of a third body in the gas phase and
and
the reactions of nitric oxide with chlorine and
CAO oxygen: (2NO + Cl 2 -* 2NOC1; 2NO + O 2 -»
- 1 (3.1.25)V
2NO 2 ).
3.1 Mathematical Characterization of Simple Reaction Systems 31
3.1.1.4 Fractional and Other Order Reactions in Unfortunately, when one combines this rela-
Constant Volume Systems. In chemical kinet- tion with the rate functions for various reaction
ics, one frequently encounters reactions whose orders, the situation is entirely different from
orders are not integers. Consider a reaction that which prevails in the constant volume case.
involving only a single reactant A whose rate One cannot develop explicit closed form expres-
expression is of the form sions for the extent of reaction as a function of
time for all the cases treated in Section 3.1.1.
= kCA = k(C (3.1.30)V Since the only common case for which one can
dt A0 develop such a solution is the first-order reac-
tion, we will start by considering this case. Since
Systems composed of stoichiometric propor-
tions of reactants also have rate expressions that 140
will often degenerate to the above form. C - UA
- (3.1.33)
CA
Except for the case where n is unity, equation ~V~
3.1.30 can be integrated to give the first-order rate expression becomes
1 1 = (1 n)Vjikt (3.1.34)
VAQ " c^ " V dt
Solution of this equation subject to the
or condition that £ = 0 at t = 0 gives
n-i rn-i ~ (1 n)\Akt
±A (3.1.35)
(3.1.32)V
W
Among the reactions whose orders are not or
integers are the pyrolysis of acetaldehyde (3.1.36)
(n = 3/2), and the formation of phosgene from
CO and Cl 2 [r = fc(Cl2)3/2(CO)]. For reaction orders other than unity, one
must treat the various reactions on an individual
3.1.2 Mathematical Characterization of basis and be able to express the total volume of
Simple Variable Volume Reaction Systems the mixture as a function of the composition,
temperature, and pressure of the system. In
From the viewpoint of an engineer who must
many cases of interest in kinetics, one is inter-
design commercial reactors to carry out gaseous
ested in reactions that occur at constant pressure
reactions involving changes in the total number
and temperature. In such situations one can say
of moles present in the system, it is important to
that
recognize that such reactions are usually ac-
companied by changes in the specific volume of V = V(nl9n29-")= V(nj) (3.1.37)
the system under study. These considerations
are particularly important in the design of If the functional form of this relation is known
continuous flow reactors. For these systems one (e.g., if one is dealing with a gaseous system that
must employ the basic definition of the reaction behaves ideally), this relationship can be com-
rate given by equation 3.0.1. bined with equation 3.0.1 and the appropriate
rate function to obtain a differential equation,
(3.0.1) which can then be integrated numerically or in
V dt explicit form. If we consider a generalized rate
32 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
Thus
RT
The various mole numbers can be expressed l>total(at/i = 1) - Hto = 0)]
in terms of the extent of reaction and the initial SA =
mole numbers. If the functional form of V is RT
known (e.g., ideal gas behavior), one obtains a
differential equation that can be integrated even = 1) - - 0)
if it is necessary to resort to numerical methods
«total(at/i = 0)
to do so.
In the analysis of experimental rate data taken (3.1.43)
in variable volume systems it is possible to If one starts with a mixture consisting of 2
develop another expression for the rate of moles of hydrogen, 4 moles of ethylene, and 3
formation of species i that is more convenient moles of inert gases, the following mole table
to use than equation 3.1.39. This alternative may be established.
approach has been popularized by Levenspiel
(2). It involves the use of fraction conversion
rather than concentration as a primary variable Moles fA= 1
and is applicable only to systems in which the
volume varies linearly with the fraction con- H2 2 0
version. C2H4 4 2
C2H6 0 2
V = K0(l + SAfA) (3.1.40) Inerts 3 3
The fraction conversion is based on the limiting Total 9 7
reagent, in this case assumed to be species A. The
parameter 5A is the fraction change in the volume In this case the hydrogen is the limiting re-
of a closed reacting system between zero con- agent and equation 3.1.43 gives
version and complete conversion. As such it may
take on both positive and negative values. Hence 7-9 2
A
~ ~~9 ~~ ~ 9
If 2 moles of ethane were present in the origi-
nal mixture used in the last example, the value of
This parameter takes into account the pres- 3A becomes —(2/11). The reader should verify
ence of inerts, the use of nonstoichiometric quan- this point.
tities of reactants, and the presence of one or If one is dealing with a gaseous system in
more of the reaction products in the original which deviations from ideality are negligible, one
system. To illustrate this point, let us consider as may take variations in the absolute temperature
an example the isothermal gas phase hydrogena- and the absolute pressure into account by a
tion of ethylene (C 2 H 4 + H 2 -• C 2 H 6 ) taking slight modification of equations 3.1.40 and
3.2 Experimental Aspects of Kinetic Studies 33
3.1.41. In this case, which one need not be concerned with consecu-
tive reactions. The assumption is also valid for
V= F0( ^1 ^ r
(3.1.44) many nonisothermal condensed phase systems.
For nonisothermal or variable pressure gaseous
with systems a modification of the form of equation
3.1.44 is more appropriate for use.
To develop expressions for the reaction rate
in variable volume systems, one need only return
(3.1.45) to the fundamental definition of the reaction
where the temperature T and the pressure P rate (3.0.1) and combine this relation with equa-
correspond to a given fraction conversion fA. tions 3.1.40 and 3.1.48.
To determine the concentration of the limiting n
A0 "JA
reagent at a given fraction conversion one need
sAfA v4 dt
only note that
nA = nA0(l - fA) (3.1.46)
CA0
(3.1.51)
dt
and that at constant temperature and pressure,
For variable volume systems equation 3.1.51 is
^4 = nA0{\ - fA) " fA) much simpler in form and easier to use than
= CA equation 3.0.9.
V V0(l + 5AfA) SAfA)
(3.1.47) 1 dC, Ci_d_V_
= (3-0.9)
v^v~dT
The concentrations of the other species pres- V; dt
ent in the reaction mixture may be found by The 6A concept will prove to be particularly use-
using the extent concept. From equation 1.1.9, ful in the design of tubular reactors for gas phase
reactions.
-nAOfA (3.1.48)
3.2 EXPERIMENTAL ASPECTS
Thus OF KINETIC STUDIES
viewpoint of the chemical engineer, who is in- reaction. Even though the reaction under investi-
terested in the design of commercial scale re- gation may be reported in textbooks, isolation
actors, a knowledge of the reaction mechanism and identification of the products should be
is useful, but not essential. Such individuals are carried out with the products of the reaction run
more concerned with the problem of determin- under the conditions of the rate measurement.
ing reaction rate expressions for use in design Errors exist in the literature, and one may have
calculations. Since this function must be deter- components present in his or her system that
mined experimentally, we now turn our atten- may promote side reactions and/or inhibit the
tion to some of the points to be considered in reaction that one thinks is being studied.
obtaining kinetic data.
3.2.1.2 Are Side Reactions Important? What is
3.2.1 Preliminary Questions to be Answered the Stoichiometry of the Reaction? When a
in Experimental Kinetics Studies mixture of various species is present in a reaction
vessel, one often has to worry about the possibil-
Since the proper design of chemical reactors ity that several reactions, and not just a single
requires a knowledge of the reaction rate expres- reaction, may occur. If one is trying to study one
sion for the system under consideration, it is particular reaction, side reactions complicate
essential that the chemical engineer thoroughly chemical analysis of the reaction mixture and
understand the methods by which these empiri- mathematical analysis of the raw data. The stoi-
cal functions are determined. One should be chiometry of the reaction involved and the rela-
extremely careful in experimental work to ob- tive importance of the side reactions must be
tain a valid rate expression by taking appro- determined by qualitative and quantitative anal-
priate precautions in the laboratory and in the lysis of the products of the reaction at various
interpretation of the data. Since many kinetic times. If one is to observe the growth and decay
data in the literature are in error in one impor- of intermediate products in series reactions,
tant respect or another and since many experi- measurements must be made on the reaction
enced chemists and chemical engineers have system before the reaction goes to completion.
erred in carrying out kinetics experiments, it is
crucial for the beginning student to be aware of 3.2.1.3 Are the Conditions of the Experiment
some of the pitfalls. The sections that follow Properly Specified? All kineticists should re-
treat several aspects of kinetic experimentation cognize the importance in experimental work of
that must be considered if one is to obtain using pure materials, of controlling and mea-
meaningful data. The discussion is patterned suring temperatures accurately, and of recording
after Bunnett's (3) treatment in Technique of times properly. Nonetheless, it is very common
Organic Chemistry. for individuals to "save time" by using reactants
or solvents that are not properly purified.
3.2.1.1 Is the Reaction Under Study Properly The first tendency of the chemist involved in
Identified? It may seem ridiculous to ask the experimental kinetics work might well be to say
experimenter whether the reaction whose ki- that one should purify the reagents and solvents
netics are being studied is actually the reaction as much as possible. On the other hand, the first
he or she thinks it is. Nonetheless, the literature tendency of a practicing engineer would be to use
contains many examples of the incorrect identi- a minimum of purification procedures, particu-
fication of reactions. One cannot always rea- larly if the ultimate goal is to design a commer-
son that the products of a reaction are the cial scale reactor that will use a relatively impure
corresponding analogs of a well-characterized feedstock. In practice, what one does is to accept
3.2 Experimental Aspects of Kinetic Studies 35
reagents and solvents at a certain degree of initiating, or for quenching a reaction is likely to
purification as being conditionally satisfactory introduce larger uncertainties in the rate mea-
subject to the reservation that preliminary ex- surement than is inherent in the performance of
periments might indicate the need for further a good timer. One generally determines the value
purification. These preliminary experiments may of the reaction rate constant by plotting some
also indicate that the chemical engineer will function of the instantaneous concentrations
have to allow for purification of the feedstock in versus time and determining the slope of the
the design of the commercial scale facility.- straight line that results. Since this slope is deter-
The reason for stressing the importance of mined by differences between the time points and
working with relatively pure reagents and sol- not by their absolute values, it is not essential in
vents is that the rates of many reactions are most cases to know these absolute values of
extremely sensitive to the presence of trace im- time. However, a word of caution is appropriate
purities in the reaction system. If there is reason when one is concerned with the analysis of con-
to suspect the presence of these effects, a series of secutive reactions. In such cases a knowledge of
systematic experirrlents may be carried out to the true zero of time is required. (See Section 5.3.)
explore the question by seeing how the reaction
rate is affected by the intentional addition of
3.2.1.4 Does the Analytical Method Properly
impurities. In many cases, lack of reproducibility
Represent the Extent of Reaction ? One should
between experiments may be an indication that
trace impurity effects are present. test the analytical procedure by analyzing mix-
tures of known compositions that are similar to
Temperature control for most reactions in
those expected for samples from the reaction
solution can be provided by immersion of the
vessel. Even if this test gives an excellent check,
reactor in a liquid constant temperature bath
one must still face the possibility that the pri-
that can be controlled to ±0.1 °C or better,
mary reaction under investigation may be ac-
depending on the desired degree of control. For
companied by unexpected side reactions that
gas phase reactions, particularly those that occur
consume reactants or produce products in such
at high temperatures, the reactor can be im-
a way as to distort the meaning of the analytical
mersed in a fluidized sand bath. The fact that
results. One general approach that provides a
these thermostatted baths can be held at a con-
measure of protection against being misled by
stant temperature for prolonged periods of time
analytical information is to follow the progress
may lead to such a sense of complacency on the
of the reaction by two or more different and
part of the experimenter that he or she fails to
independent analytical methods. Where possi-
record the temperature at periodic intervals. One
ble, each of these methods should involve analy-
of the laws of experimental work ("If anything
sis for a different species.
can go wrong, it will") dictates that the investi-
gator adhere to a strict timetable for recording
the system temperature. When the kinetic data of 3.2.1.5 Is the Experiment Properly Planned to
different laboratories disagree, the failure to cali- Provide Significant Data ? One cannot directly
brate temperatures properly is often at fault. measure a reaction rate. One is restricted to
Consequently, it is wise to calibrate periodically measurements of the concentrations of various
the device used to record the bath temperature. species or of a physical property of the system as
Although some precautions must be observed a function of time. Thus one must plan the ex-
in timing a reaction, the direct measurement of periments so as to obtain significant differences
time seldom affects the accuracy of a rate deter- in the quantities that are observed in the labora-
mination. The time required for sampling, for tory. In a properly planned experiment the
36 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
kineticist chooses experimental techniques that Such numbers are usually adequate to allow
provide differences between successive samples curvature to be distinguished from linearity, even
that are many times greater than the expected when one needs to reject one or two points
error in the measurement. This procedure per- because of scatter of the data.
mits the kineticist to derive a meaningful rate
constant from the original measurements. 3.2.1.6 Are the Data Reproducible ? If experi-
One decision that the kineticist must make in mental results are to be accepted as meaningful
planning experiments is the number of observa- by the scientific community, they must be capa-
tions to be made during a run and the times at ble of being reproduced by investigators in other
which these observations are to be made. One laboratories as well as by the original investiga-
usually tries to work up the data in a form such tor. An individual should assure this reproduc-
that a plot of some function of the concentra- tion by replicating the experiment.
tions in the reaction vessel versus time gives a There are several sources of irreproducibility
straight line. The reaction rate constant is then in kinetics experimentation, but two of the most
derived from the slope of this line. The question common are individual error and unsuspected
then becomes, "How many points does it take to contamination of the materials or reaction ves-
establish the linearity and slope of the line?" Two sel used in the experiments. An individual may
points are adequate to determine a straight line, use the wrong reagent, record an instrument
but they are not sufficient to establish linearity of reading improperly, make a manipulative error
data. Any number of curved lines could be drawn in the use of the apparatus, or plot a point in-
through two points. Various functional forms correctly on a graph. Any of these mistakes can
4>{Ci) corresponding to different reaction orders lead to an erroneous rate constant. The prob-
could be used as the ordinate in these plots, and ability of an individual's repeating the same
one could not distinguish between them on the error in two successive independent experiments
basis of two experimental points. The straight is small. Consequently, every effort should be
line joining these points will not give any indica- made to make sure that the runs are truly inde-
tion of whether or not other experimental points pendent, by starting with fresh samples, weighing
would lie on the same straight line. Moreover, these out individually, etc. Since trace impurity
this approach to determining a rate constant effects also have a tendency to be time-variable,
cannot reveal any of the experimental defects or it is wise to check for reproducibility, not only
complications that are often called to one's at- between runs over short time spans, but also
tention by curved or scattered plots. between runs performed weeks or months apart.
If one has three or four completely accurate It is commendable experimental procedure to
points, an adequate test of the linearity of the repeat each run in duplicate and to be satisfied
assumed concentration function could be made. if the two results agree, but this is expensive in
However, one cannot be sure whether departures terms of the labor costs involved. Moreover,
from linearity are the result of scatter of the data, repetition of each run is not always necessary.
of a slowly curving function, or of some combina- For example, if one is studying the effect on the
tion of these circumstances. When one further reaction rate of a variable such as temperature or
recognizes that one or two points in a set of data reactant concentration, a series of experiments
may be in error as a result of an experimental in which the parameter under investigation is
mishap and that these points may deviate sub- systematically varied may be planned. If a plot
stantially from the line established by the rest of of the results versus this parameter yields a
the data, one comes to the conclusion that it is smooth curve, one generally assumes that the
desirable to have eight to ten data points per run. reproducibility of the data is satisfactory.
3.2 Experimental Aspects of Kinetic Studies 37
In addition to worrying about the reproduc- 3.2.2.1 General Reactor Types. In planning ex-
ibility of the data, one should also take care to periments the chemist is most apt to think in
insure that the calculations are reproducible by terms of closed systems and has traditionally
an independent calculation by a colleague. The used some form of batch reactor. The chemical
literature contains numerous values of mis- engineer, on the other hand, is accustomed to
calculated rate constants. One has the usual dealing with open systems and the eventual
problem of conversion of units and the problem design of large-scale continuous processes. Con-
of conversion from natural logarithms to com- sequently, the chemical engineer will often
mon logarithms. Thus, even though one's raw choose some form of flow reactor for laboratory
data might be highly accurate, the calculated scale investigations. The discussion in this chap-
results may be off by a large multiple. If a ter is restricted to batch reactors, but the
colleague uses the raw data, determines the rate relations developed in Chapter 8 may be used to
constant for a particular experimental run with analyze kinetic data from flow reactors.
reference only to first principles, and obtains
the same rate constant as the original investi- 3.2.2.2 Methods of Following the Course of a
gator, one has a good indication that a system- Reaction. A general direct method of measur-
atic error in calculations has not been made. ing the rate of a reaction does not exist. One can
only determine the amount of one or more
3.2.2 Experimental Techniques and product or reactant species present at a certain
Apparatus time in the system under observation. If the
In order to determine reaction rate constants composition of the system is known at any one
and reaction orders, it is necessary to determine time, then it is sufficient to know the amount of
reactant or product concentrations at known any one species involved in the reaction as a
times and to control the environmental condi- function of time in order to be able to establish
tions (temperature, homogeneity, pH, etc.) dur- the complete system composition at any other
ing the course of the reaction. The experimental time. This statement is true of any system whose
techniques that have been used in kinetics studies reaction can be characterized by a single re-
to accomplish these measurements are many and action progress variable (£ or fA). In practice it
varied, and an extensive treatment of these is always wise where possible to analyze occa-
techniques is far beyond the intended scope of sionally for one or more other species in order
this textbook. It is nonetheless instructive to to provide a check for unexpected errors, losses
consider some experimental techniques that are of material, or the presence of side reactions.
in general use. More detailed treatments of the In more complex cases when several reactions
subject are found in the following books. are occurring simultaneously in the system
under observation, calculations of the composi-
1. R. B. Anderson, Experimental Methods in Catalytic
Research, Academic Press, New York, 1968.
tion of the system as a function of time will
require the knowledge of a number of indepen-
2. H. W. Melville, and B. G. Gowenlock, Experimental
Methods in Gas Reactions, Macmillan, New York, 1964. dent composition variables equal to the number
3. S. L. Freiss, E. S. Lewis, and A. Weissberger, Technique of of independent chemical equations used to
Organic Chemistry, Volume VIII, "Investigation of Rates characterize the reactions involved.
and Mechanisms of Reactions," Parts One and Two, In principle one can use any of the many tools
Second Edition, Wiley-Interscience, New York, 1961.
and methods of the analytical chemist in carry-
4. A. Weissberger, Techniques of Chemistry, Volume VI,
"Investigation of Rates and Mechanisms of Reactions," ing out these determinations. In practice analy-
Parts I and II, Third Edition, Wiley-Interscience, New tical methods are chosen on the basis of their
York, 1974. specificity, accuracy, ease of use, and rapidity of
38 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
measurement. For our discussion these methods It is useful at this point to consider briefly the
may be classified into two groups: physical and relative merits of chemical and physical methods
chemical. Regardless of the method chosen, for monitoring the course of a reaction. A
however, it must meet the following criteria. chemical method involves removal of a portion
of the reacting system, inhibition of the reaction
1. It must not disturb the system under investi-
occurring within the sample, and subsequent
gation by affecting the kinetic processes
analysis of the system composition. This anal-
occurring therein.
ysis may be carried out using conventional wet
2. The measurement should be representative
chemistry techniques, a spectroscopic method, or
of the system at the time it is made, or at the
any one of a variety of other analytical tech-
time the sample analyzed was taken.
niques. The essence of the chemical method,
3. The method must provide a true measure of
however, lies in the fact that it requires removal
the extent of reaction.
of a sample from the reacting system. Chemical
In the choice of an analytical technique for methods give absolute values of the concentra-
following the course of a reaction, it is important tions of the various species present in the re-
to recognize that no aspect of the measurement action mixture. However, it should be stressed
should affect the kinetic processes occurring in that even the most precise of these methods has
the system. For example, a solution conductivity an accuracy that is no better than the accuracy
method that uses platinized electrodes should of the standard analytical procedure that was
not be used in the study of a reaction that is used to calibrate the method. Another disad-
catalyzed by platinum black. vantage inherent in the use of chemical methods
Perhaps the most obvious method of studying is that the sampling procedure involved does not
kinetic systems is to periodically withdraw provide a continuous record of the reaction pro-
samples from the system and to subject them to gress. It is impossible to study very fast reactions
chemical analysis. When the sample is with- using these methods.
drawn, however, one is immediately faced with Physical methods involve the measurement of
a problem. The reaction will proceed just as a physical property of the system as a whole
well in the test sample as it will in the original while the reaction proceeds. The measurements
reaction medium. Since the analysis will require are usually made in the reaction vessel so that
a certain amount of time, regardless of the the necessity for sampling with the possibility
technique used, it is evident that if one is to of attendant errors is eliminated. With physical
obtain a true measurement of the system com- methods it is usually possible to obtain an
position at the time the sample was taken, the essentially continuous record of the values of the
reaction must somehow be quenched or inhi- property being measured. This can then be
bited at the moment the sample is taken. The transformed into a continuous record of reactant
quenching process may involve sudden cooling and product concentrations. It is usually easier
to stop the reaction, or it may consist of elimina- to accumulate much more data on a given re-
tion of one of the reactants. In the latter case, action system with such methods than is possible
the concentration of a reactant may be reduced with chemical methods. There are certain
rapidly by precipitation or by fast quantitative limitations on physical methods, however. There
reaction with another material that is added to must be substantial differences in the contribu-
the sample mixture. This material may then be tions of the reactants and products to the value
back-titrated. For example, reactions between of the particular physical property used as a
iodine and various reducing agents can be measure of the reaction progress. Thus one
quenched by addition of a suitably buffered would not use pressure measurements to follow
arsenite solution. the course of a gaseous reaction that does not
3.2 Experimental Aspects of Kinetic Studies 39
involve a change in the total number of moles in ful calibration curves relating the property being
the gaseous phase. measured to the extent of reaction or species
It is always wise to calibrate physical methods concentrations.
of analysis using mixtures of known composition
under conditions that approximate as closely
3.2.2.3 Physical Methods that have been Used
as practicable those prevailing in the reaction
to Monitor Reaction Kinetics. In this section
system. This procedure is recommended because
some physical property measurements of general
side reactions can introduce large errors and
utility are discussed. One of the oldest and most
because some unforeseen complication may
useful techniques used in kinetics studies in-
invalidate the results obtained with the tech-
volves the measurement of the total pressure
nique. Ivor example, in spectrophotometric
in an isothermal constant volume reactor. This
studies of reaction kinetics, the absorbance that
technique is primarily used to follow the course
one measures can be grossly distorted by the
of homogeneous gas phase reactions that involve
presence of small amounts of highly colored
a change in the total number of gaseous mole-
absorbing impurities or by-products. For this
cules present in the reaction vessel (e.g., the
reason, when one uses indirect physical methods
hydrogenation of propylene).
in kinetic studies, it is essential to verify the
stoichiometry of the reaction to ensure that the H i f^ LJ C^ TU (1 T 1 \
products of the reaction and their relative mole 2 + <-3"6 —• <-3ri8 p.Z.lj
numbers are known with certainty. For the same Pressure measurements can be accomplished by
reason it is recommended that more than one any one of a number of different types of
physical method of analysis be used in detailed manometric devices without disturbing the sys-
kinetic studies. tem being observed. Another type of reaction
In principle, any physical property that varies system that can be monitored by pressure mea-
during the course of the reaction can be used to surements is one in which one of the products
follow the course of the reaction. In practice one can be quantitatively removed by a solid or
chooses methods that use physical properties liquid reagent that does not otherwise affect
that are simple exact functions of the system the reaction. For example, acids formed by
composition. The most useful relationship is that reactions in the gas phase can be removed by
the property is an additive function of the con- absorption in hydroxide solutions.
tributions of the different species and that each From a knowledge of the reaction stoichio-
of these contributions is a linear function of the metry and measurements of the total pressure as
concentration of the species involved. This a function of time, it is possible to determine the
physical situation implies that there will be a extent of reaction and the partial pressures or
linear dependence of the property on the extent concentrations of the various reactant and
of reaction. As examples of physical properties product species at the time at which the measure-
that obey this relationship, one may cite elec- ment is made. Illustration 3.3 indicates how
trical conductivity of dilute solutions, optical pressure measurements can be used to determine
density, the total pressure of gaseous systems a reaction rate function.
under nearly ideal conditions, and rotation of The various forms of spectroscopy find wide-
polarized light. In sufficiently dilute solutions, spread application in kinetic studies. They are
other physical properties behave in this manner usually well suited for application to in situ stud-
to a fairly good degree of approximation. More ies of the characteristics of the reaction mixture.
complex relationships than the linear one can be The absorption by a reacting system of electro-
utilized but, in such cases, it is all the more magnetic radiation (light, microwaves, radio-fre-
imperative that the experimentalist prepare care- quency waves, etc.) is a highly specific property
40 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
of the system composition and the dimensions tion of a reactant or product as a function of
of the reaction vessel. Both the magnitude of the time, how does one proceed to determine the
absorption coefficient and the wavelengths at reaction rate expression?).
which absorption maxima occur are character- The determination of the reaction rate expres-
istic of the absorbing compound and, to a lesser sion involves a two-step procedure. First, the
extent, its physical state. By appropriate choice concentration dependence is determined at a
of the wavelength of the incident radiation, one fixed temperature. Then the temperature depen-
is able to take advantage of the remarkable dence of the reaction rate constant is evaluated
specificity of the various forms of spectroscopy to give a complete reaction rate expression. The
to monitor the progress of the reaction. Among form of this temperature dependence is given by
the various forms of spectroscopy that can be equation 3.0.14, so our present problem reduces
used in in situ kinetic studies are visible, ultra- to that of determining the form of the concentra-
violet, infrared, microwave, nuclear magnetic tion dependence and the value of the rate con-
resonance, and electron spin resonance spectro- stant at the temperature of the experiment.
scopy. For treatments of the limitations and Unfortunately, there is no completely general
uses of spectroscopic techniques, consult trea- method of determining the reaction rate expres-
tises on analytical chemistry. sion or even of determining the order of a re-
In addition to spectrophotometric or spectro- action. Usually one employs a trial-and-error
scopic measurements, there are a number of procedure based on intelligent guesses and past
other optical measurements that can be used to experience with similar systems. Very often the
monitor the course of various reactions. Among stoichiometry of the reaction and a knowledge
the optical properties that can be used for these of whether the reaction is "reversible" or "ir-
studies are optical rotation, refractive indices, reversible" will suggest the form of the rate equa-
fluorescence, and colorimetry. tion to try first. If this guess is incorrect, the
There are several electrical measurements that investigator may then try certain forms that are
may be used for analysis of solutions under in suggested by assumptions about the mechanism
situ conditions. Among the properties that may of the reaction. Each reaction presents a unique
be measured are dielectric constants, electrical problem, and success in fitting a reaction rate
conductivity or resistivity, and the redox poten- expression to the experimental data depends on
tial of solutions. These properties are easily the ingenuity of the individual investigator.
measured with instrumentation that is readily The discussion below is largely confined to
adapted to automatic recording operation. How- irreversible reactions with simple rate expres-
ever, most of these techniques should be used sions of the form of equation 3.0.17.
only after careful calibration and do not give
r = kCpAACiBCpcc (3.0.17)
better than 1% accuracy without unusual care in
the experimental work. but the methods developed are more generally
applicable. We have chosen this course to keep
the discussion as simple as possible and to pre-
3.3 TECHNIQUES FOR THE INTER-
sent the material in a manner that avoids the
PRETATION OF REACTION RATE DATA
introduction of more complex rate expressions.
In Section 3.1 the mathematical expressions that Most of the methods presented below are appli-
result from integration of various reaction rate cable, regardless of the mathematical form of the
functions were discussed in some detail. Our rate expression, and they may be readily ex-
present problem is the converse of that con- tended to cover the rate expressions that will be
sidered earlier (i.e., given data on the concentra- encountered in subsequent chapters.
3.3 Techniques for the Interpretation of Reaction Rate Data 41
The techniques used to determine reaction from a knowledge of the concentration of one of
rate functions may be broken down into three the reactants or products at some time r, the
general categories. initial composition of the system, and the reac-
tion stoichiometry (from which dA may be deter-
1. Integral methods based on integration of the
mined), it is possible to use equation 3.1.47 or
reaction rate expression. In these approaches
equation 3.1.50 to determine the fraction con-
one analyzes the data by plotting some func-
version of the limiting reagent at this time.
tion of the reactant concentrations versus
Equation 3.1.51 may then be used to determine
time.
the reaction rate at this conversion level.
2. Differential methods based on differentiation
of experimental concentration versus time ^ A 1
data in order to obtain the actual rate of reac- r = (3.1.51)
1 + dt
tion. In these approaches one analyzes the
data by postulating various functional rela- Thus, for both variable and constant volume
tions between the rate of reaction and the systems, one can manipulate concentration ver-
concentrations of the various species in the sus time data to obtain values of the reaction
reaction mixture and tests these hypotheses rate as a function of time or as a function of the
using appropriate plots. concentrations of the various species present in
3. Methods based on simplification of the re- the reaction mixture. The task then becomes one
action rate expression. In these approaches of fitting this data to a reaction rate expression
one uses a vast excess of one or more of the of the form of equation 3.0.13.
reactants or stoichiometric ratios of the re-
actants in order to permit a partial evaluation r = k<KCt) (3.0.13)
of the form of the rate expression. They may Since data are almost invariably taken under
be used in conjunction with either a differ- isothermal conditions to eliminate the tempera-
ential or integral analysis of the experimental ture dependence of reaction rate constants, one
data. is primarily concerned with determining the con-
Each of these general categories is discussed in centration dependence of the rate expression
the subsections that follow. [0(Q)] and the rate constant at the temperature
in question. We will now consider two differ-
ential methods that can be used in data analysis.
3.3.1 Differential Methods for the Treatment
of Reaction Rate Data
3.3 J.I Differentiation of Data Obtained in the
In experimental kinetics studies one measures Course of a Single Experimental Run. If one
(directly, or indirectly) the concentration of one has experimental results in the form of concen-
or more of the reactant and/or product species tration versus time data, the following general
as a function of time. If these concentrations are differential procedure may be used to determine
plotted against time, smooth curves should be 4>(Ci) and k at the temperature in question.
obtained. For constant volume systems the re-
action rate may be obtained by graphical or 1. Set forth a hypothesis as to the form of the
numerical differentiation of the data. For vari- concentration dependent portion of the rate
able volume systems, additional numerical ma- function, 0(Q).
nipulations are necessary, but the process of 2. From the experimental concentration-time
determining the reaction rate still involves differ- data, determine the reaction rate at various
entiation of some form of the data. For example, times.
42 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
3. At the selected times, prepare a table listing chiometric ratios of the reactants. One must also
the reaction rate and the concentrations of assume that the rate expression is of the form of
the various species present in the reaction equation 3.3.1 with the pt for all species other
mixture. Calculate 0(C/) at each of these than reactants taken as zero.
points. From the use of a stoichiometric feed ratio,
4. Prepare a plot of the reaction rate versus
4>(Ct). If the plot is linear and passes through "AO 'lB0 nCo
(3.3.3)
the origin, the form of 0 is consistent with the
experimental data. Consequently, insofar as From the definition of extent of reaction,
the accuracy of the data is concerned, the
n
form of the rate expression is satisfactory for i = ^_nAo. ^= _B^n_Bo. a n d ^n_c_n_c*
subsequent use in design calculations. The V A
Y
A
V A
A
V D V R
D O
Vr-
r
W
CC
slope of this straight line is the reaction rate (3.3.4)
constant. If the plot is not linear or does not
From equations 3.3.3 and 3.3.4, one can see
pass through the origin, one must return to
that at all times, with a stoichiometric feed ratio,
step 1 and hypothesize a new form for the
concentration dependence of the reaction
rate. 5 d = nJL = *L (3.3.5)
Differential procedures are illustrated sche- some sort of graphical procedure to analyze the
matically in Figure 3.1. The first diagram in- data. From such plots one may see that certain
dicates how the rate may be determined from points are seriously in error and should not be
concentration versus time data in a constant weighted heavily in the determination of the
volume system; the second schematic illustrates reaction rate function. The consistency and pre-
the method just described. The third diagram cision of the data can also be evaluated by eye by
indicates the application of our general differ- observing its deviation from a smooth curve
ential method to this system. (ideally its deviation from a straight line).
It is always preferable to use as much of the The following example illustrates the use of
data as possible to determine the reaction rate the differential method for the analysis of kinetic
function. This often implies that one should use data. It also exemplifies some of the problems
Log concentration
(b)
Figure 3.1
Schematic representation of the application of the differential method
for data analysis.
44 Basic Concepts in Chemical Kinetics;—Determination of the Reaction Rate Expression
Bromine
Time, t concentration - ^ X 103
C C15
(min) (moles/liter) At (moles/liter) (moles/liter)15
0 0.3335
2.25 16.44 0.3150 0.1762
0.2965
4.50 13.56 0.2812 0.1491
0.2660
6.33 11.48 0.2555 0.1291
0.2450
8.00 11.68 0.2353 0.1141
0.2255
10.25 9.11 0.2153 0.0999
0.2050
12.00 0.1910 8.00 0.1980 0.0881
13.50 0.1794 7.73 0.1852 0.0797
15.60 0.1632 7.71 0.1713 0.0709
17.85 0.1500 5.87 0.1566 0.0620
19.60 0.1429 4.06 0.1465 0.0561
27.00 •0.1160 3.64 0.1295 0.0466
30.00 0.1053 3.23 0.1107 0.0369
38.00 0.0830 2.79 0.0942 0.0289
41.00 0.0767 2.10 0.0799 0.0226
45.00 0.0705 1.55 0.0736 0.0200
47.00 0.0678 1.35 0.0692 0.0182
57.00 0.0553 1.25 0.0615 0.0153
63.00 0.0482 1.18 0.0518 0.0118
100
Figure 3.2
Log-log plot of reaction rate versus bromine concentration for
illustration 3.1.
46 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
constant volume systems or by measuring the be simple integers or half-integers. When this is
initial slope of a fraction conversion versus time the case, one should note that the values of ft
curve and using equation 3.1.51 to calculate the that fit the data may not have mechanistic
initial reaction rate. Numerical calculations significance. Instead, they merely provide ex-
could also be used. In either event, for each run ponents for a reasonable mathematical approx-
one obtains a value of the initial reaction rate imation to the true rate expression, which itself
r 0 at a given mixture composition. If the reac- may be a much more complex mathematical
tion rate expression is of the form function.
One advantage of the initial rate method is
r0 = = k (3.3.9) that complex rate functions that may be ex-
tremely difficult to integrate can be handled in
one may determine the order of the reaction a convenient manner. Moreover, if one uses
with respect to an individual component (e.g., initial reaction rates, the reverse reactions can
A) by making a rate measurement at two be neglected and attention can be focused solely
different initial concentrations of this species on the reaction rate function for the forward
while holding all other concentrations constant reaction. More complex rate functions may be
between the two runs. If we denote the two tested by the choice of appropriate coordinates
observed rates by r 0 and r 0 and the corre- for plotting the initial rate data. For example, a
sponding initial concentrations by CA0 and reaction rate function of the form
C'A0, equation 3.3.9 may be written for these
runs as kCA
r
o = (3.3.13)
r0 = (fcCfeC£§ • • -)CA-0 (3.3.10) k'CA
and may be tested by plotting the reciprocal of the
(3.3.11) initial rate versus the reciprocal of the initial
Division of equation 3.3.10 by equation 3.3.11 concentration of species A. Since
and solution of the resulting expression for f}A 1 1 k'
gives (3.3.14)
kC AO
this plot should be linear with an intercept equal
(3.3.12) to k'/k and a slope equal to 1/fc.
log
c
3.3.2 Integral Methods for the Treatment
By varying the initial concentration of each
of Reaction Rate Data
component of the reaction mixture in turn, it is
possible to determine the order of the reaction When integral methods are used in data analysis,
with respect to each species. After this has been measured concentrations are used in tests of
established, equation 3.3.9 may be used to proposed mathematical formulations of the
determine the reaction rate constant. reaction rate function. One guesses a form of
If one has initial rate data available at several the reaction rate expression on the basis of the
different concentrations of species A and at the reaction stoichiometry and assumptions con-
same initial concentrations of all other species, cerning its mechanism. The assumed expression
a more accurate value of f}A may be determined is then integrated to give a relation between the
from a log-log plot of equation 3.3.10. In many composition of the reaction mixture and time.
cases orders determined in this fashion will not A number of such relations were developed in
48 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
Figure 3.4
(3.3.15) Test of reaction rate data by the integral
method of analysis.
2. Separate variables and integrate.
linear in time with a slope equal to the rate con- taken at 80%, 90%, or higher conversions of the
stant k. However, it is more convenient to limiting reagent.
rearrange this equation as Perhaps the most discouraging type of de-
viation from linearity is random scatter of the
CBO
In [ ^ ] = In (CAOVB - CBOvA)kt data points. Such results indicate that something
cAO, is seriously wrong with the experiment. The
(3.3.20) method of analysis may be at fault or the reac-
and plot the left side versus time. One still tion may not be following the expected stoichio-
requires the data plot to be linear, but the metry. Side reactions may be interfering with the
restriction that the line pass through the origin analytical procedures used to follow the progress
is modified so that the Y intercept becomes of the reaction, or they may render the mathe-
ln(CB0/CA0). matical analysis employed invalid. When such
The graphical approach to the analysis of plots are obtained, it is wise to reevaluate the
kinetic data has several advantages. The best entire experimental procedure and the method
straight line through the experimental points used to evaluate the data before carrying out
can often be found by inspection (i.e., by moving additional experiments in the laboratory.
a transparent straight edge until it appears to Another form of deviation from linearity that
fit the data with a minimum of deviation). Since is often encountered in plots of \l/(Ct) versus
the errors in the concentration measurements time is curvature of the data, as shown in
are usually much greater than those in the time Figure 3.5. If the assumed order is greater than
measurements, in estimating the best straight the true order of the reaction, upward curvature
line through a series of points one should think will be observed. If the assumed order is less
of the residuals as parallel to the concentration than the true order, downward curvature will
axis (i.e., vertical), not as perpendicular to the result.
line. The graphical method readily shows trends
and deviations from linear behavior. Scattered
points that are obviously in error can be
recognized easily and eliminated from further
consideration. In those rare cases where the
accuracy of the data exceeds that obtained in
plotting on a reasonable scale, one should use
the numerical methods described in Section
3.3.2.2.
Since the integral method described above is
based on the premise that some rate function
exists that will lead to a value of i/KQ) that is
linear in time, deviations from linearity (or A = Positive curvature
curvature) indicate that further evaluation or B = Negative curvature
interpretation of the rate data is necessary.
Many mathematical functions are roughly linear
Time, t >-
over sufficiently small ranges of variables. In
order to provide a challenging test of the Figure 3.5
linearity of the data, one should perform at Observation of curvature in tests of reaction
least one experimental run in which data are rate data.
50 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
Solution RT
+ 20 - no]
7T - 7T0
Assume that the reaction follows first-order
kinetics. The integral form of the reaction rate ^00 - ^0 RT 2n0
expression is given by equation 3.1.8. (3n0 - H0) —
Cf or
= -kt (A)
n — n0 \ ( n
2
(D)
n0 3TT 0 - TTO \ TT0
At the temperature and pressures of this exper-
iment, ideal gas behavior may be assumed. Thus
The mole fraction ether at time t is given by
r - RT
(B)
RT
n0 - n0
where n is the total pressure in the system. y* = (E)
n0
Combination of equations A and B gives
n0
YE7J
= -kt (Q Combination of equations D and E gives
n0
Time, t 0 t oo
This last equation is one against which the
Mole data can be tested to see if the assumption of
number first-order kinetics is appropriate. The data are
CH3OCH3 n0 n0 - £ 0 worked up below.
CH 4 0 £ ^ = n0
H2 0 £ {00 = *o
Time, t (sec) Pressure (mm Hg) 3TT0 - n
CO 0 = «0
0 420 840
Total
57 584 676
moles no n0 + 2^ 3«0
85 662 598
114 743 517
Pressure no 7T IT. = 3TT0
145 815 445
182 891 369
219 954 306
261 1013 247
Now, from this table and the ideal gas law, 299 1054 206
it is evident that
3.3 Techniques for the Interpretation of Reaction Rate Data 53
(3.3.21)
W i= i
3.3.2.2 Integral Methods for the Analysis of
Kinetic Data—Numerical Procedures. While where kOti is the value of the rate constant com-
the graphical procedures discussed in the pre- puted using the concentration value determined
vious section are perhaps the most practical and at time tt. For a Class I, second-order reaction
useful of the simple methods for determining (vA = —1) taking place in a constant volume
rate constants, a number of simple numerical system, the several values of the rate constant
procedures exist for accomplishing this task. may be determined from equation 3.1.15.
54 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
adapted to various graphical techniques. Illus- Many computer libraries contain programs that
tration 3.4 indicates how this approach is used perform the necessary statistical calculations
to determine a reaction rate constant. and relieve the engineer of this burden. For
Orre may also use the methods of the statis- discussions of the use of weighted least squares
tician to determine average rate constants (e.g., methods for the analysis of kinetic data, see
the standard unweighted least squares pro- Margerison's review (8) on the treatment of ex-
cedure). perimental data and the treatments of Kittrell
The unweighted least squares analysis is based et al. (9), and Peterson (10).
on the assumption that the best value of the rate
constant k is the one t,hat minimizes the sum of
3.3.2.3 Integral Methods for the Analysis of
the squares of the residuals. In the general case
Kinetic Data—Fractional Life Methods. The
one should regard the zero time point as an
adjustable constant in order to avoid undue time necessary for a given fraction of a limiting
weighting of the initial point. An analysis of this reagent to react will depend on the initial con-
type gives the following expressions for first- centrations of the reactants in a manner that is
and second-order rate constants determined by the rate expression for the re-
action. This fact is the basis for the development
First order: of the fractional life method (in particular the
half-life method) for the analysis' of kinetic data.
T Zfo *n CAi) - [£> t -)(I In CAi)/ri] The half-life, or half-period, of a reaction is the
\2 i ~\ v =r ~2
l) /"] - 2Ji time necessary for one half of the original re-
Second order: actant to disappear. In constant volume systems
it is also the time necessary for the concentra-
tion of the limiting reagent to decline to one
4
CM C half of its original value.
k = (3.3.26)
The fractional life approach is most useful as
a means of obtaining a preliminary estimate of
the reaction order. It is not recommended for
These expressions apply to constant volume the accurate determination of rate constants.
systems in which vA = — 1. The sums are taken Moreover, it cannot be used for systems that do
from i = 1 to i = n. Although the use of these not obey nth order rate expressions.
equations is somewhat laborious for hand cal- If one combines the definition of the reaction
culations, they are easily handled by even the rate in variable volume systems with a general
simplest types of computers. nth-order rate expression, he finds that the time
While the unweighted least squares method of necessary to achieve a specified fraction con-
data analysis is commonly used for the deter- version is given by
mination of reaction rate constants, it does not
yield the best possible value for k. There are two dfA (3.3.27)
principal reasons for this failure. Jo VkC\ -vAVkCnA
1. All points are assigned equal weights. or, using equations 3.1.40 and 3.1.47,
2. Functions of the measured concentrations
must be used in the equations defining the
residuals.
Statisticians have developed general calcula- This relation indicates that for all values of
tion procedures that avoid these difficulties. n, k, vA, and 5A, the fractional life is inversely
56 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
Table 3.1
Fractional Life Relations for Constant Volume Systems
Reaction
order rc: 0 1 2 3
V"1
2 ;
f f A0 //(I - / ) (2/ - / ) l(1l - -r,
J ) -•
l f
f
vAk k 2(1 - f)2vAkCA0
2""' -- 1
tll2/tl/4- 2.000 2.409 3.000 3.857
(4/3)"~l _ j
tfl £n(l -
for n = 1 - ft) - = In
tf2 - f2
(3.3.31a)V
(3.3.35)
The value of this ratio is characteristic of the Similar equations are valid at times t2, t3, etc.
reaction order. Table 3.1 contains a tabulation In all cases, however^ the right side of these
of partial reaction times for various rate expres- equations will be a constant, since the time
sions of the form r = kCA as well as a tabulation increment A is a constant. Thus, at time th
of some useful ratios of reaction times. By using
ratios of the partial reaction times based on ex- - Q = vAktt + constant (3.3.36)
perimental data, one is able to obtain a quick
A plot of tn(£ — £') versus t will therefore give
estimate of the reaction order with minimum
a straight line with a slope equal to vAk for first-
effort. Once this estimate is in hand one may
order kinetics.
proceed to use a more exact method of deter-
This technique is readily adaptable for use
mining the reaction rate parameters.
with the generalized additive physical approach
discussed in Section 3.3.3.2. It is applicable to
3.3.2.4 Integral Methods for the Analysis of systems that give apparent first-order rate con-
Kinetic Data—Guggenheim's Method for First- stants. These include not only simple first-order
Order Reactions. Guggenheim (11) has devel- irreversible reactions but also irreversible first-
oped a special method that is useful in obtaining order reactions in parallel and reversible re-
the rate constant for a first-order reaction when actions that are first-order in both the forward
an accurate value of the initial reactant concen- and reverse directions. The technique provides
tration is not available. It requires a series of an example of the advantages that can be ob-
readings of the parameter being used to follow tained by careful planning of kinetics experi-
the progress of the reaction at times tl9t2,t3, etc. ments instead of allowing the experimental
and at times t1 + A, t2 + A, t3 + A, etc. The design to be dictated entirely by laboratory
time increment A should be two or three times convention and experimental convenience.
the half life of the reaction. If we denote the Guggenheim's technique has been extended
extent of reaction at times tx and rx + A by ft to other order reactions (12, 13), but the final
and ft, respectively, equation 3.1.36 indicates expressions are somewhat cumbersome.
58 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
CH 3
If one notes that vA = — 1 and that the
The dilatometer readings (h) listed below are quantity In^h^ - /i0)/^oo] i s itself a constant
given in arbitrary units and are arranged in
pairs taken at a fixed interval of 2 hr. They are £n(/i — h') — —kt + a new constant (B)
related to the extent of reaction by Note that k can be determined from the slope
g = *- K of a plot of the left side of equation B versus time
and that this plot can be prepared without a
1. Use Guggenheim's method to determine the knowledge of the dilatometer readings at times
pseudo first-order rate constant. zero and infinity. The data are worked up below
2. Given the additional fact that the reading at and plotted in Figure 3.8.
"infinite" time is 12.16, use a numerical
averaging procedure to determine the re- Time, t (min) h — ti
action rate constant.
0 5.34
Data 4.56
10
Initial concentrations (moles/liter): 20 4.00
30 3.51
HC1O4 = 0.3974; C 4 H 8 = 0.00483. 40 3.06
50 2.69
Dilatometer Dilatometer 60 2.38
Time, t reading reading 70 2.07
(min) at time t at time t + 2 hr
/.u I I tI I I I CAi - h
4. (from 1 and 3)
6.0 - -
(
5.0 - Thus the average value of the reaction rate
constant can be calculated numerically from
4.0 ^ < -
1 - ft;
k =
(n + 1)A ^ o - ft;
3.0
where n is the number of pairs of data points.
Numerical calculations indicate that
2.0 -
N k =
(7
1
(12.768) = 0.0133 min"
used to determine k' and fiA. It should be noted t may be expressed in terms of the extent of
that this technique should always be used in reaction per unit volume £*.
conjunction with other methods of determining
the rate expression, because it forces the data to t = Ci0 (3.3.43)
fit equation 3.3.37 when, in fact, the rate expres- The experimental variable X may now be written
sion may be more complex. If one observes that in the following manner.
the exponents f}t are nonintegers that vary con-
tinuously with temperature or pressure, this is an X = XM 4- XA + XB + Xz (3.3.44)
indication that the rate expression is more com-
plex than equation 3.3.37. where XM includes the contribution of the solvent
or medium in which the reaction is carried out
as well as the contributions of any inert species
3.33.2 Use of Physical Property Measurements that are present. It also includes any contribu-
as a Measure of the Extent of Reaction. The tions arising from the reaction vessel itself. Com-
most useful physical properties for use in kinetics bination of equations 3.3.41 and 3.3.44 gives
studies are those that are an additive function of
the contributions of the various constituents, X = XM + gA + hACA + gB + hBCB + gz + hzCz
the contribution of each species being a linear (3.3.45)
function of its concentration. Total pressure, ab-
sorbance, optical rotation, and the electrical con- If the initial property value is denoted by Xo,
ductivity of dilute solutions are all properties Xo = XM + gA + hACA0 + gB + hBCB0 +gz + hzCZ0
of this type. This section indicates how such (3.3.46)
physical property measurements may be simply
related to the extent of reaction per unit volume. The change in the value of the property X
Consider some physical property X that results between time zero and time t is given by:
from the contributions of the various species X - Xo = hA(CA - CA0) + hB(CB - CB0)
present in the reaction mixture
+ hz(Cz - Czo) (3.3.47)
A = (3.3.40) or
X - Xo = hAvAZ* + hBvB£* + hzvz£* = £* X M i
where X is the experimentally measured quantity i
the concentrations of the various species at time Equation 3.3.50 is an extremely useful relation
3.3 Techniques for the Interpretation of Reaction Rate Data 61
O O
I
C—Cl CH.OH C—OCH 3 H Cl
B-+C
When more than two data points are avail- If one assumes that the errors in the four
able, the graphical method is much better to use quantities tu f2, CA1, and CA2 are independent
than common averaging techniques. It gives one and that n is a known constant, the expected
a visual picture of the fit of the data to 3.3.55. If random error in the rate constant Ak may be
one has several data points and estimates of the expressed as follows.
uncertainty in each point, a weighted least
squares fit of the data would be appropriate. A/c \2 dink
T d In tly
3.3.5 Precision of Rate Measurements
d In k\2
for Simple Irreversible Reactions
d £n tj
In order to obtain a feeling for the major sources
of uncertainty and error in the calculation of dink
reaction rate constants, it is useful to consider d lnCA CA1
the nature of the errors inherent in the measure- dink &cA2
ment of these parameters. dlnCA cA2
3.3.5.1 Precision of Reaction Rate Constants. (3.3.59)
A reaction rate constant can be calculated from or
the integrated form of a kinetic expression if one AkV At,
has data on the state of the system at two or more
different times. This statement assumes that ~k t2 ~ h
sufficient measurements have been made to (n -
establish the functional form of the reaction rate
expression. Once the equation for the reaction cA1
rate constant has been determined, standard /AC A2
techniques for error analysis may be used to 'A2
evaluate the expected error in the reaction rate (3.3.60)
constant.
In the general case of a dependent variable For the case of n = 1, the corresponding
y = /(x l5 x2, x 3 , . . . , xw), which is a known equation is
function of the m independent variables xl9 x2, /A/c
x3,...,xm, the expected relative error in y due
to the relative errors in x l5 x2, x 3 ,. . . , xw is
IT
given by
-I
Equation 3.1.32 applies to a constant volume
system that follows nth-order kinetics. If we
take vA = — 1 it can be rewritten as
(3.3.58)
(3.3.61)
k =
n - l)(t2 -
To illustrate the use of equation 3.3.60 for
where CA2 and CA1 are the concentrations of A estimating the relative uncertainty in the reac-
present at times t2 and t l9 respectively, and where tion rate constant, consider an example where
n # 1. n = 2, (t2 — r j = 100 sec, the uncertainty in
64 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
each measurement (A^ and At2) is 1 sec, CA2 = centration measurements. Benson (17) has pre-
0.9 CA1 (10% reaction), and the relative un- sented a useful table that summarizes the
certainty in each concentration measurement is converse of the problem we have just considered
0.01 (1%). Substitution of these values into (i.e., what precision in analytical capability is
equation 3.3.60 gives required to measure k to a given degree of
accuracy). It is reproduced here as Table 3.2.
Afc 1 1
Benson states that the analytical error is about
T 100 100 the same for orders of reaction from zero to
0.9 four and is primarily dependent on the extent
(0.01)2 of reaction occurring between the two points
1 - 0.9
chosen. Inspection of this table illustrates the
dilemma the kineticist faces in planning exper-
(0.01)2
iments. In order to measure k to within 1%
accuracy he or she must be able to analyze for
= 1O"4 + 1O~4
his or her reactants with a precision of approx-
+ 81 x 1O~4 + 100 x 10~ 4 imately 0.1% at a conversion level of approx-
= 183 x 10" 4 imately 15%. If higher precision in k is desired,
much higher analytical precision must be
Thus, the relative uncertainty in the rate con-
obtained and higher degrees of conversion must
stant is
be used.
Afc = (183 x 10" 4 ) 1/2 = ±13.7 x 10" The situation is significantly changed if it is
possible to monitor the appearance of products
instead of the disappearance of reactants. The
= ±0.137
mathematics of the situation lead one to the
The major sources of error are the concen- general principle that whenever possible it is
tration measurements and attempts to improve advantageous to monitor the products during
the accuracy of the rate constant should focus the initial stages and the reactants during the
on trying to improve the accuracy of the con- final stages of the reaction.
Table 3.2
Errors in Calculated Rate Constants Caused by Analytical Errors
1 5 10 20 30 40 50
Analytical
precision (%) Error in k (%)
3.3.5.2 Precision of Activation Energy Mea- At room temperature the relative uncertainty
surements. The activation energy of a reaction in measuring E over 10 °C temperature intervals
can be determined from a knowledge of the is generally about ±5% for gas phase reactions
reaction rate constants at two different tempera- and about ± 3% for liquid phase reactions.
tures. The Arrhenius relation may be written in
the following form.
LITERATURE CITATIONS
17. Benson, S. W., The Foundations of Chemical Kinetics, Determine the reaction order and rate constant
McGraw-Hill, New York, copyright © 1960. Used with for the reaction by both differential and integral
permission.
methods of analysis. For orders other than one,
C o will be needed. If so, incorporate this term
PROBLEMS into the rate constant.
1. Moelwyn-Hughes (Physical Chemistry, page
1109, Pergamon Press, New York, 1957) has 3. The data given below are typical of the poly-
tabulated the following values of the rate con- merization of vinyl phenylbutyrate in dioxane
stant for the reaction solution in a batch reactor using benzoyl per-
oxide as an initiator. The reaction was carried
N 2 O 5 -> N 2 O 4 + \O2 out isothermally at 60 °C using an initial mono-
mer concentration of 73 kg/m 3 . From the fol-
Temperature lowing data determine the order of the reaction
T (°K) k( sec - 1 ) and the reaction rate constant. Note that there is
an induction period at the start of the reaction so
288.1 1.04 x 1(T 5 that you may find it useful to use a lower limit
298.1 3.38 x 10" 5 other than zero in your integration over time.
313.1 2.47 X 10" 4 The reaction order may be assumed to be an
323.1 7.59 X 10" 4
integer.
338.1 4.87 X 10" 3
If the rate constant is of the form of equation Time, t (ksec) Monomer concentration (kg/m3)
3.0.14, determine the parameters A and E.
7.2 40.6
10.8 23.2
2. The decomposition of hexaphenylethane to
14.4 13.3
triphenylmethyl radicals in liquid chloroform 18.0 7.4
has been studied at 0 °C. 21.6 4.16
25.2 2.32
(C 6 H 5 ) 3 CC(C 6 H 5 ) 3 -> 2(C 6 H 5 ) 3 C 28.8 1.30
32.4 0.74
The following results were obtained.
Time, t (sec) C/Co for hexaphenylethane 4. Huang and Dauerman \Jnd. Eng. Chem. Pro-
duct Research and Development, 8 (227), 1969]
0 1.000 have studied the acetylation of benzyl chloride in
17.4 0.941 dilute solution at 102 °C.
35.4 0.883
54.0 0.824
NaAc + C 6 H 5 CH 2 C1 ->
174.0 0.530
209 0.471 C 6 H 5 CH 2 Ac Na + + Cl
313 0.324
367 0.265
Using equimolal concentrations of sodium ace-
434 0.206
584 0.118
tate and benzyl chloride (0.757 kmole/m 3 ), they
759 0.059 reported the following data on the fraction benzyl
chloride remaining unconverted versus time.
Problems 67
7. The following initial rate data [-d(B 2 H 6 )/ What does the fact that the final observed
dt] were reported for the gas-phase reaction of pressure is 124.1 kPa imply?
diborane and acetone at 114 °C.
B2H6 + 4Me2CO -> 2(Me2CHO)2BH
9. The reaction rate constant for the gas phase
decomposition of ethylene oxide is 0.0212 min~ *
Initial pressure (torr) at 450 °C.
TnitiPil ratp s\v IKJ
J.1I1L1CX1 l u l C
10^
Run B2H 6 Me 2 CO (torr/sec)
CH2 - CH2(g) CO{g)
in the complex concentration during the time Determine the order of the reaction and the re-
interval studied? action rate constant. The order is an integer.
(b) What is the order of the reaction with respect
to Br2 concentration? What is the product 13. When one makes use of the assumption that
of the true rate constant and the term cor- the volume of a system is linear in the fraction
responding to the dependence of the rate conversion
on complex concentration (pseudo rate V = V0(l + df)
constant)? the parameter 3 is based on the limiting reagent.
11. Svirbley and Roth [J. Am. Chem. Soc, 75 Calculate 3 for the following gaseous reactions
(3106), 1953] have reported the following data and the feed concentrations given. / refers to the
for the reaction between HCN and C 3 H 7 CHO concentration of inerts in the feed stream.
in aqueous solution at 25 °C. (a) C 2 H 5 OH -> H 2 O + C 2 H 4
Initial concentration ratios
HCN C 3 H 7 CHO C 2 H 5 OH:H 2 O:C 2 H 4 :/ =o= 1:1:1:2
Time, t (sec) (moles/m3) (moles/m3) (b) CH 3 CHO + H 2 -> CH 3 CH 2 OH
Initial concentration ratios
166.8 99.0 56.6 C H 3 C H O : H 2 : C H 3 C H 2 O H : / ^ 1:2:0:1
319.8 90.6 48.2 (c) CH 3 CHO + H 2 -• CH 3 CH 2 OH
490.2 83.0 40.6 Initial concentration ratios
913.8 70.6 28.2
C H 3 C H O : H 2 : C H 3 C H 2 O H : / - 2:1:0:0
1188. 65.3 22.9
oo 42.4 0 (d) C 2 H 6 + Cl2 -> HC1 + C2H5C1
Initial concentration ratios
C 2 H 6 :C1 2 :HC1:C 2 H 5 C1:/ =0= 1:2:1:0:3
Determine the order of the reaction and the
(e) C H 2 = C H — C H = C H 2 + 2H2 -> Q H j
reaction rate constant.
Initial concentration ratios
12. Baciocchi, et al. [J. Am. Chem. Soc, 87 v^±l 2 —v^rl—\^rl—^rl 2 . r l 2 . \^ 4 rl j
(3953), 1965] have investigated the kinetics of 2:1:1:0
the chlorination of several hexa-substituted ben- 2N 2 O 5 -> 2N 2 O 4 + O 2
zenes. They have reported the following data for Initial concentration ratios
the chlorination of dichlorotetramethylbenzene N 2 0 5 : / : N 2 0 4 : 0 2 o 1:0:0:0
in acetic acid solution at 30 °C.
Initial concentrations: 14. The reaction of cyclohexanol and acetic acid
C 6 (CH 3 ) 4 C1 2 = 34.7 moles/m 3 in dioxane solution as catalyzed by sulfuric acid
was studied by McCracken and Dickson \_Ind.
Cl 2 = 19.17 moles/m 3 Eng. Chem. Proc. Des. and Dev., 6 (286), 1967].
The esterification reaction can be represented by
Time, t (ksec) Fraction conversion the following stoichiometric equation.
0 0
48.42 0.2133 A + B : C + W
85.14 0.3225 acetic cyclo- cyclo- water
135.3 0.4426 acid hexanol hexyl
171.3 0.5195 acetate
222.9 0.5955
257.4
The reaction was carried out in a well-stirred
0.6365
batch reactor at 40 °C. Under these conditions,
70 Basic Concepts in Chemical Kinetics—Determination of the Reaction Rate Expression
the esterification reaction can be considered as (b) If acetic acid exists primarily as a dimer
irreversible at conversions less than 70%. (C2H4O2)2 in solution at 40 °C, postulate a
set of mechanistic equations that will be con-
sistent with this rate expression.
The following data were obtained using identical 15. The saponification of the ester phenylace-
sulfuric acid concentrations in both runs. tate by sodium phenolate has the following
stoichiometry.
Run 1 : Co = Ceo = 2.5 kmoles/m 3
H2O + CH3COOC6H5 + Na +
CB (kmoles/m3 Time, t (ksec) CH 3 COO"Na 2C6H5OH
2.070 7.2 The reaction takes place in aqueous solution.
1.980 9.0 Equimolal concentrations of the ester and the
1.915 10.8
phenolate are used. These concentrations are
1.860 12.6
1.800 14.4
equal to 30 moles/m3. By the time the sam-
1.736 16.2 ples are brought to thermal equilibrium in the
1.692 18.0 reactor and efforts made to obtain data on ester
1.635 19.8 concentrations as a function of time, some sa-
1.593 21.6 ponification has occurred. At this time the con-
1.520 25.2 centration of ester remaining is 26.29 moles/m3,
1.460 28.8 and the concentration of phenol present in the
reactant mixture is 7.42 moles/m3. The rate ex-
Run 2: CA0 = 1 kmole/m 3 pression for the reaction is believed to be of the
Cflo = 8 kmoles/m 3 form
tester ^phenolate
CA (kmoles/m3) Time, t (ksec) Rate =
cphenol
0.885 1.8
0.847 2.7 Determine the reaction rate constant from the
0.769 4.5 following data.
0.671 7.2
0.625 9.0 Time, t (ksec) Ester concentration (moles/m3)
0.544 12.6
0.500 15.3 0 26.29
0.463 18.0 0.72 22.00
2.16 18.30
4.32 15.29
(a) Determine the order of the reaction with 8.64 12.15
14.34 9.99
respect to each reactant and the rate constant
20.64 8.50
for the forward reaction under the conditions 27.24 7.30
of the two runs. The rate constant should
differ between runs, but the conditions are
such that the order will not differ. The indi- Be very careful to note the appropriate stoi-
vidual and overall orders are integers. chiometric coefficients for use in your analysis.
Problems 71
16. Consider a gas phase dehydrogenation re- 18. Daniels (Chemical Kinetics, page 19, Cornell
action that occurs at a constant temperature of University Press, 1938) has carried out a series of
1000 °K. investigations of the N 2 O 5 decomposition re-
action. The following data are typical of the time
RCH2OH(#) -> RCHO(#) + H2{g) for 50% decomposition of a given initial charge
At this temperature the reaction rate constant is of N 2 O 5 in a constant volume reactor at various
equal to 0.082 liters/mole-min. If the reaction temperatures. It may be assumed that the initial
takes place in a constant pressure reactor, starting concentration of N 2 O 5 is the same in all cases
with pure gaseous alcohol at a pressure of 2 and that each run is carried out isothermally.
atm, find the time necessary to reach 20%
decomposition: Temperature, T (°C) Half-life (sec)
(a) In the sense that the number of moles of
alcohol in the reactor decreases by 20%. 300 3.9 x 10~5
(b) In the sense that the concentration of alcohol 200 3.9 x 10~3
150 8.8 x 10" 2
in the reactor decreases by 20%. 100 4.6
Ideal gas laws may be assumed in both cases. 50 780
in the gas phase at 45 °C. The reaction takes Determine the apparent order of reaction and
place in a constant volume system. the apparent rate constant in suitable units.
Hints
(1) The resistance is inversely proportional to
Time, t (min) PN2o5 (mm of Hg)
the conductance of the solution. The latter is
0 an additive property of the contributions of
10 247 the various species in the solution. The re-
20 185 action order is either 0, 1, or 2. Note that the
30 140 concentrations of nitroethane and pyridine
40 105 are much greater than that of the iodine.
50 78 (2) Show that the extent of reaction at time t
60 58 relative to the extent at time infinity is given
70 44 by
80 33
90 24 =
100 18 \~R R~\ R
dU2O is the chemical shift that would be ob- have been corrected to allow for heat losses from
served in the absence of unprotonated the system (i.e., adiabatic operation of the batch
species reactor may be assumed). The reaction was
5H + O is the chemical shift that would be ob- carried out in diethyl ether solution using initial
served if all species were protonated concentrations of C 2 H 5 MgBr and pinacolin of
403 and 10 moles/m 3 , respectively. The stoi-
X is the fraction of the water molecules that
H2O chiometry of the reaction may be represented as
are not protonated
XH + O is the fraction of the water molecules that A + B -+ C
are protonated Changes in the potentiometer reading may be
It may be assumed that all the protons formed by assumed to be proportional to the temperature
reaction immediately react to form protonated change of the reaction mixture. Determine the
water molecules. The following chemical shifts order of the reaction with respect to pinacolin
were observed as a function of time. and the apparent reaction rate constant.
[Data taken from article by Horman and Unfortunately, the potentiometer reading at
Strauss, J. Chem. Education, 46 (114), 1969.] If time zero is uncertain.
reaction (A) may be considered as an irreversible
reaction, what is the order of this reaction with 24. Pannetier and Souchay* have reported the
respect to BHB? The initial concentration of data below as an example of the application
BHB is 360 moles/m3. The water is present in of dilatometry to kinetics studies.
sufficient excess that its concentration may be The hydration of ethylene oxide
considered to be essentially constant. What is the
reaction rate constant? CH 2 OH
CH 2 —CH 2 H2O
CH.OH
23. Tuulmets {Kinetics and Catalysis, 5 (59),
1964] has studied the kinetics of the reaction of
was studied in aqueous solution at 20 °C using
ethyl magnesium bromide with pinacolin. He
an initial oxide concentration of 120 moles/m 3
used a calorimetric technique to monitor the
and a catalyst concentration of 7.5 moles/m3.
progress of the reaction. The overall temperature
increase of the reaction mixture was less than a * Adapted from Chemical Kinetics by G. Pannetier and
degree. Mixture temperatures were determined P. Souchay, p. 68, copyright © 1964. Used with permission
with a sensitive potentiometer. The data below of Elsevier Scientific Publishing Company.
74 Basic Concepts in Chemical Kinetics—Determination pf the Reaction Rate Expression
4 Kinetics—Molecular Interpretations
of Kinetic Phenomena
4.0 INTRODUCTION available experimental data, representing a sug-
gestion as to the sequence of molecular events
This chapter treats the descriptions of the molec- that occur in proceeding from reactants to pro-
ular events that lead to the kinetic phenomena ducts. It does not necessarily represent the actual
that one observes in the laboratory. These events events that occur during the reaction process,
are referred to as the mechanism of the reaction. but it must be consistent with the available ex-
The chapter begins with definitions of the vari- perimental facts. Often there will be more than
ous terms that are basic to the concept of re- one mechanism that is consistent with these
action mechanisms, indicates how elementary facts. The problem of designing an experiment
events may be combined to yield a description that will eliminate one or more of the competing
that is consistent with observed macroscopic mechanisms is a challenging problem for the
phenomena, _and discusses some of the tech- kineticist. In some cases it is an impossible task.
niques that may be used to elucidate the mecha- Implicit in the use of the term "sequence of
nism of a reaction. Finally, two basic molecular molecular events" is the idea that the chemical
theories of chemical kinetics are discussed—the transformation that one observes in the labora-
kinetic theory of gases and the transition state tory is not the result of a single molecular process
theory. The determination of a reaction mecha- but is the end result of a number of such pro-
nism is a much more complex problem than that cesses. If one considers reactions such as
of obtaining an accurate rate expression, and
the well-educated chemical engineer should have C 7 H 16 + 11O2 -> 7CO2 + 8H2O
a knowledge of and an appreciation for some of
(4.0.1)
the techniques used in such studies.
+
There are at least two levels of sophistication 5SOJ + 2IO3- + 2H -+ 5SO^ + I2 + H2O
at which one may approach the problem of pro- (4.0.2)
viding a molecular description of the phenomena
that occur during the course of a chemical re- it should be evident from a purely statistical
action. At the first level the sequence of molec- viewpoint that there is virtually zero probability
ular events is described in terms of the number that the number of molecular species involved
and type of molecules and molecular fragments in these reactions would simultaneously be in
that come together and react in the various steps. spatial and electronic configurations such that
The second level of description contains all of more than a few chemical bonds could be broken
the elements of the first but goes beyond it to and/or made in a single step. These equations
treat the geometric and electronic configurations merely describe the stoichiometry that is ob-
of the various species during the different stages served in the laboratory. Nonetheless, they do
of the reaction sequence. For this textbook, the reflect the conversion of reactant molecules into
first level of description is adequate. The second product molecules. It is quite plausible to assume
level is more appropriate for study in courses of that the events that occur on a molecular level
physical organic chemistry or advanced physical are encounters at which atomic rearrangements
chemistry. occur and that the observed reaction may be
The mechanism of a reaction is a hypothetical interpreted as the sum total of the changes that
construct. It is a provisional statement based on occur during a number of such encounters. This
76
4.0 Introduction 77
assumption is the foundation on which all stud- stoichiometric chemical equation remains valid
ies of reaction mechanisms are based. The major when multiplied by an arbitrary factor, a mecha-
justifications for its use lie in the tremendous nistic equation loses its meaning when multiplied
success it has had in providing a molecular inter- by an arbitrary factor. Whereas stoichiometric
pretation of kinetic phenomena and the fact that coefficients and reaction orders may be integers
it has been possible to observe experimentally or nonintegers, the molecularity of a reaction is
some of the intermediates postulated in se- always an integer. The following examples indi-
quences of elementary reactions. cate the types of rate expression associated with
Each elemental process contributing to the various molecularities.
overall mechanism is itself an irreducible chemi-
cal reaction. The elementary reactions may also Unimolecular: A-+B + C r = kxCA
be referred to as simple reactions, mechanistic (4.0.3)
reactions, or reaction steps. The superposition
in time of elementary reactions leads to the ex- Bimolecular: M + N ^ P r = k2CMCN
perimentally observed reaction. Since each step (4.0.4)
in the mechanism is itself a chemical reaction
and is written as such, the equation representing Termolecular: R + 2S ^ T + U r = k3CRCj
an elemental process looks the same as an equa- (4.0.5)
tion that represents the stoichiometry of a chemi-
cal reaction. The mechanistic equation, however, Note that for the last reaction the stoichiometric
has an entirely different meaning than the stoi- coefficients require that
chiometric equation. It is only from the context rs = -2k3CRCJ (4.0.6)
of their use that the reader knows which type
of equation is being used, and the beginning Each elementary reaction must fulfill the fol-
kineticist must be sure to distinguish between lowing criteria.
their meanings in his or her thinking.
1. On a molecular scale the reaction occurs
The number of chemical species involved in a
exactly as written.
single elementary reaction is referred to as the
2. The simple reaction should break or form as
molecularity of that reaction. Molecularity is a
few bonds as possible. Normally only one
theoretical concept, whereas stoichiometry and
bond is made and/or broken. Occasionally
order are empirical concepts. A simple reaction
two bonds are broken and two new ones made
is referred to as uni-, bi-, or termolecular if one,
in a four-center reaction. Only very rarely do
two, or three species, respectively, participate as
more complex processes occur.
reactants. The majority of known elementary
3. The simple reaction must be feasible with
steps are bimolecular, with the balance being
respect to bond energies, atomic geometry,
unimolecular and termolecular.
and allowed electron shifts.
Since an elementary reaction occurs on a
molecular level exactly as it is written, its rate In all but the simplest cases, the mechanism of
expression can be determined by inspection. A a reaction consists of a number of steps, some of
unimolecular reaction is ^ first-order process, which involve reacting species that do not ap-
bimolecular reactions are second-order, and ter- pear in the overall stoichiometric equation for
molecular processes are third-order. However, the reaction. Some of these intermediate species
the converse statement is not true. Second-order are stable molecules that can be isolated in the
rate expressions are not necessarily the result of laboratory. Others are highly reactive species
an elementary bimolecular reaction. While a that can be observed only by using sophisticated
78 Molecular Interpretations of Kinetic Phenomena
experimental techniques. These reaction inter- orderly manner, the thought processes leading
mediates are sometimes referred to as active to these results involve elements of experience,
centers. intuition, luck, knowledge, and guesswork.
Among the various intermediate species that An individual who has started work on a
may participate in a reaction sequence are stable kinetics research project will usually have some
molecules, ions, free atoms, free radicals, car- preconceived ideas about the mechanism of the
banions, carbonium ions, molecular and ionic reaction being investigated and will usually
complexes, and tautomeric or excited forms of know the major products of the reaction before
stable molecules. If the intermediate is, indeed, a starting any kinetics experiments. As a result of
stable substance, then its presence can be de- a literature search, the kineticist often has a
tected by any of the standard techniques of knowledge of experimental rate expressions and
chemical analysis, provided that the interme- proposed mechanisms for similar reactions. As
diate can be isolated (i.e., prevented from partici- the kineticist proceeds with the experimental
pation in the processes that would normally work and determines the complete product
follow its formation). If isolation is impossible, distribution, the order of the reaction, and- the
then the techniques available for the study of effects of temperature, pressure, solvent, etc., on
stable intermediates are the same as those for the reaction rate, his or her ideas relative to the
the study of highly reactive species. For a de- mechanism will evolve into a coherent and
tailed discussion of appropriate experimental logical picture. Along the path, however, the
techniques, consult the references listed in Sec- preliminary ideas may be shuffled, added to,
tion 3.2.2 or the review by Wayne (1). discarded, broadened, or refurbished before
they can be manipulated into a logical sequence
4.1 REACTION MECHANISMS of events that is consistent with all of the
available facts. Each mechanism study evolves
At first glance the problem of "finding the
in its own fashion as the investigator designs
mechanism" of a reaction may appear to the
experiments to test provisional mechanisms that
beginning student as an exercise in modern
have been postulated on the basis of earlier
alchemy. Nonetheless, there are certain basic
experiments. These experiments constitute a
principles underlying the reasoning that inter-
more or less systematic attempt to make mecha-
venes between experimental work in the labora-
nistic order out of what may appear to the
tory and the postulation of a set of molecular
uninitiated observer to be a random collection
events that give rise to the chemical reaction
of experimental facts. By the time the experi-
being investigated. This section is a discussion
mental research is completed and the results are
of the nature of the problem, the means by
ready for publication, the investigator will have
which one derives a rate expression from a
resolved his or her ideas about the mechanism
proposed sequence of reactions, and some tech-
into a fully coherent picture that can be pre-
niques that are useful in the elucidation of
sented in a systematic and rational manner.
reaction mechanisms.
Since this end result is all that is published for
the judgment of the scientific community, the
4.1.1 The Nature of the Problem
uninitiated observer obtains the false impression
The postulation of a reaction ^mechanism is that the research work itself was actually carried
the result of inductive rather than deductive out in this systematic and rational fashion. In
thinking. Even though the kinetics researcher the actual program there may have been many
may present the ideas and experiments that false starts and erroneous ideas that had to be
lead to a proposed mechanism in a logical overcome before arriving at the final result. To
4.1 Reaction Mechanisms 79
a fellow kineticist, however, the investigator's of its constituent elementary reactions. The
systematic and logical rendering of the experi- fundamental principle from which one starts is
mental facts is an efficient means of communi- that the rate of an elementary reaction is pro-
cating the results of the research. portional to the frequency of collisions indicated
The problem of determining the mechanism by the stoichiometric-mechanistic equation for
of a chemical reaction is one of the most in- the reaction (i.e., to the product of the concentra-
teresting and challenging intellectual and experi- tions indicated by the molecularity of the re-
mental problems that a chemical engineer or action). In addition, one usually bases the
chemist is likely to encounter. Even for relatively analysis on one or more of the following
simple mechanisms, it is occasionally difficult to simplifications in order to make the mathematics
determine from the assumptions about the amenable to closed form solution.
elementary reactions just what the exact form First, one often finds it convenient to neglect
of the predicted dependence of the rate on re- reverse reactions (i.e., the reaction is considered
actant concentrations should be. In many cases to be "irreversible"). This assumption will be
the algebra becomes intractable and, without valid during the initial stages of any reaction,
an a priori knowledge of the magnitudes of since the number of product molecules available
various terms, it may be difficult or even im- to serve as reactants for the reverse reaction will
possible for the theoretician to make simplifica- be small during this period. If the equilibrium
tions that will permit these predictions to be constant for the reaction is very large, this
made. Many chemical reactions are believed to assumption will also be true for intermediate
have mechanisms involving a large number of and later stages of the reaction.
elementary steps [e.g., the mechanism of re- Second, it may be convenient to assume that
action between H 2 and O 2 contains at least 26 one elementary reaction in the sequence occurs
steps (2)], and it becomes an extremely formida- at a much slower rate than any of the others.
ble task to manipulate the corresponding mathe- The overall rate of conversion of reactants to
matical relations to predict the dependence of products may be correctly calculated on the
the rate on the concentrations of stable mole- assumption that this step governs the entire
cules. How much more formidable is the process. The concept of a rate limiting step is
problem that our investigator must tackle—that discussed in more detail below.
of deriving such a mechanism consistent with Third, it is often useful to assume that the
all available experimental data! concentration of one or more of the intermediate
Since the problem of deriving a rate expression species is not changing very rapidly with time
from a postulated set of elementary reactions is (i.e., that one has a quasistationary state situa-
simpler than that of determining the mechanism tion). This approximation is also known as the
of a reaction, and since experimental rate ex- Bodenstein steady-state approximation for inter-
pressions provide one of the most useful tests mediates. It implies that the rates of production
of reaction mechanisms, we will now consider and consumption of intermediate species are
this problem. nearly equal. This approximation is particularly
good when the intermediates are highly reactive.
Fourth, one often finds it convenient to as-
4.1.2 Basic Assumptions Involved in the
sume that one or more reactions is in a quasi-
Derivation of a Rate Expression from a
equilibrium condition (i.e., the forward and
Proposed Reaction Mechanism
reverse rates of this reaction are much greater
The mechanism of a chemical reaction is a than those of the other reactions in the sequence).
microscopic description of the reaction in terms The net effect is that these other reactions do not
80 Molecular Interpretations of Kinetic Phenomena
produce large perturbations of the first reaction If one of the times (say ^ L S ) in the denomina-
from equilibrium. tor of equation 4.1.3 is very much larger than
In the sequence of elementary reactions any of the others, then this equation becomes
making up the overall reaction, there often is
one step that is very much slower than all the 1
— r
RLS (4.1.5)
subsequent steps leading to reaction products. L
RLS
In these cases the rate of product formation
where the subscript RLS refers to the rate
may depend on the rates of all the steps pre-
limiting step.
ceding the last slow step, but will not depend on
If some of the processes prior to the rate
the rates of any of the subsequent more rapid
limiting step are characterized by times com-
steps. This last slow step has been termed the
parable to that of the rate determining step,
rate controlling, rate limiting, or rate determining
equation 4.1.4 may be rewritten as
step by various authors.
In a mechanism consisting of several consecu- 1
tive elementary reaction steps, fragments of the r = (4.1.6)
1 1 1
initial reactant pass through a number of inter-
mediate stages, finally ending up as the products
of the reaction. The total time necessary to where terms corresponding to events beyond
produce a molecule of product is simply the the last slow step in the process have been
sum of the discrete times necessary to pass dropped. Kinetic events preceding the last slow
through each individual stage of the overall step can influence the overall reaction rate, but
reaction. Where a reversible reaction step is subsequent steps cannot.
involved, the net time for that step is necessarily
increased by the "feedback" of intermediate
4.1.3 Preliminary Criteria for Testing a
accompanying that step. The mean reaction
Proposed Reaction Mechanism—
time (rmean) is thus
Stoichiometry and Derivation of a
Lean = h + t2 + £3 + ' ' ' + tn (4.1.1) Rate Expression for the Mechanism
where the tt represent effective times necessary There are two crucial criteria with which a
on the average to accomplish each step. proposed mechanism must be consistent.
The overall reaction rate r may be defined as
1. The net effect of the elementary reactions
the reciprocal of this mean reaction time.
must be the overall reaction and must explain
1 the formation of all observed products.
r == (4.1.2) 2. It must yield a rate expression that is con-
Thus sistent with the kinetics observed experi-
mentally.
r = (4.1.3) In the case of relatively simple reaction
+ t2 + U + + mechanisms, the net or overall effect of the
The rates of each individual step are given by elementary reactions can be determined by
analogs of equation 4.1.2. Thus adding them together. For example, the stoi-
chiometric equation for the decomposition of
1
r = (4.1.4) nitrogen pentoxide is
1 1 1 1
— + — + — + ••• H 4NO, (4.1.7)
4.1 Reaction Mechanisms 81
Mechanism A [N 2 O 2 ] = y [ N O ] 2 = X e [ N O ] 2 (K)
2NO + H 2 ^ N 2 + H 2 O 2 (slow) (C)
where Ke is an equilibrium constant for reaction
H 2 O 2 + H 2 ^ 2H2O (fast) (D) E.
82 Molecular Interpretations of Kinetic Phenomena
Substitution of this expression for the N 2 O 2 is a classic reaction in kinetics because it was
concentration in equation I gives the first gas phase, first-order reaction to be
reported (3).
[NO] 2 [H 2 ] = The reaction has been studied by a number
of investigators, and the following mechanism
(L) is the one that appears to give the best agreement
These equations are consistent with the ob- with the available experimental facts.
served kinetics if we identify k as ^.
(B)
k = (M)
+ NO 2 -*NO + O 2 (slow) (C)
Thus mechanism B, which consists solely of
bimolecular and unimolecular steps, is also (D)
consistent with the information that we have In reaction C the N O 2 itself does not react
been given. This mechanism is somewhat simpler but plays the role of a collision partner that may
than the first in that it does not requite a ter- effect the decomposition of the N O 3 molecule.
molecular step. This illustration points out that The N O 2 and N O 3 molecules may react via the
the fact that a mechanism gives rise to the two paths indicated by the rate constants k2
experimentally observed rate expression is by and k3. The first of these reactions is believed
no means an indication that the mechanism is to have a very small activation energy; the
a unique solution to the problem being studied. second reaction is endothermic and conse-
We may disqualify a mechanism from further quently will have an appreciable activation
consideration on the grounds that it is incon- energy. On the basis of this reasoning, Ogg (4)
sistent with the observed kinetics, but consis- postulated that /c3 is much less than k2 and that
tency merely implies that we continue our reaction C is the rate controlling step in the
search for other mechanisms that are consistent decomposition. Reaction D, which we have
and attempt to use some of the techniques dis- included, differs from the final step postulated
cussed in Section 4.1.5 to discriminate between by Ogg.
the consistent mechanisms. It is also entirely
Derive an expression for the overall reaction
possible that more than one mechanism may be
rate based on the above mechanism.
applicable to a single overall reaction and that
parallel paths for the reaction exist. Indeed,
Solution
many catalysts are believed to function by open-
ing up alternative routes for a reaction. In the The rate of the overall reaction is given by
case of parallel reaction paths each mechanism either of the following expressions.
proceeds independently, but the vast majority
of the reaction will occur via the fastest path. 4o 2 ]
dt dt
The second is more useful for our purposes,
ILLUSTRATION 4.2 USE OF STEADY-STATE
since oxygen is produced in only one step of
AND PSEUDO EQUILIBRIUM
the mechanism and that step is the rate con-
APPROXIMATIONS FOR INTERMEDIATE
trolling step. Thus
CONCENTRATIONS
The thermal decomposition of nitrogen pent- r = rRLS = /c 3 [NO 3 ][NO 2 ] (F)
oxide to oxygen and nitrogen dioxide Equation F contains the concentration of an
2N 2 O 5 -> O 2 + 4NO 2 (A) intermediate (NO 3 ) that does not appear in
4.1 Reaction Mechanisms 83
the overall stoichiometric equation. This term step. From the equilibrium condition,
may be eliminated by means of the Bodenstein
steady-state approximation.
<*[NO3]
- /c 2 [NO 2 ][NO 3 ] - /c 3 [NO 3 ][NO 2 ] - /c 4 [NO][NO 3 ] = 0 (G)
dt
4NO]
= * 3 [NO 2 ][NO 3 ] - £ 4 [NO][NO 3 ] = 0 (I)
dt
working principles are known to experienced Guideline 3. All of the elementary reactions
kineticists, they are able to make implicit involved in a mechanistic sequence must be
chemical judgments so automatically that they feasible with respect to bond energies. Compared
often cannot see the problem from the student's to the average density of molecular energies in
viewpoint. These principles are seldom stated a reacting mixture, the energies required to
explicitly in the literature. Edwards et al. (5) break interatomic bonds in a molecule are quite
have assembled perhaps the most extensive large. Because of this, elementary reactions will
collection of guidelines. Those that are presented normally involve relatively simple acts in which
below permit a kineticist to judge the reason- few bonds are involved. Complicated rearrange-
ableness of a proposed mechanism. Often they ments consisting of a series of concerted motions
will permit the elimination of a proposed must be viewed with skepticism in the absence
sequence of reactions as being unreasonable in of strong supporting evidence.
the light of available experimental data. Bond energy considerations also lead to the
Guideline 1. The most fundamental basis for conclusion that highly endothermic elementary
mechanistic speculation is a complete analysis of reactions will be slow processes because of the
the reaction products. It is important to obtain a large activation energies normally associated
complete quantitative and qualitative analysis with these reactions.
for all products of the reaction. Inasmuch as Guideline 4. A number of elementary reactions
many chemical reactions give a complex array sufficient to provide a complete path for the
of products, the relative proportions of which formation of all observed products must be
change as the time of reaction increases, it is employed.
very useful to carry out a complete analysis for
the shortest possible reaction time and for Guideline 5. All of the intermediates produced
successively longer periods. In this manner one by the elementary reactions must be consumed
can differentiate between primary products by other elementary reactions so that there will
formed directly from the reactants and sec- be no net production of intermediate species.
ondary products formed by subsequent reaction
Guideline 6. The great majority of known ele-
of the primary products. Such analyses can give
mentary steps are bimolecular, the remainder
valuable clues as to the identity of reaction
intermediates. being unimolecular or termolecular. Any reac-
tion where the stoichiometric coefficients of the
Guideline 2. The atomic and electronic structure reactants add up to four or more must involve
of the reactants and products may provide a multiplicity of steps. The ammonia synthesis
important clues as to the nature of possible reaction is known to occur by a number of
intermediate species. The degree of atomic and steps rather than as
electronic rearrangement that takes place will
often indicate which portions of the reactant N , + 3H, 2NH, (4.1.11)
molecules participate in the reaction act and Guideline 1. A postulated mechanism for a reac-
which would be involved in elementary reac- tion in the forward direction must also hold for
tions leading to the formation of reaction the reverse reaction. This guideline is a conse-
intermediates. The structural arrangement of quence of the principle of microscopic reversi-
atoms in the molecules that react must corre- bility. (See Section 4.1.5.4.) Three corollaries of
spond at the instant of reaction to interatomic this guideline should also be kept in mind when
distances appropriate for the formation of new postulating a reaction mechanism. First, the
species. rate limiting step for the reverse reaction must
4.1 Reaction Mechanisms 85
be the same as that for the forward reaction. Guideline 9. When the overall order of a reac-
Second, the reverse reaction cannot have a tion is greater than three, the mechanism probably
molecularity greater than three, just as the has one or more equilibria and intermediates
forward reaction is so limited. Consequently, prior to the rate determining step.
the ammonia decomposition reaction
Guideline 10. Inverse orders arise from rapid
2NH 3 -> N 2 + 3H 2 (4.1.12) equilibria prior to the rate determining step. In
order to illustrate this point, consider the
cannot occur as a simple bimolecular process. oxidation of arsenious acid in an aqueous
Third, if the reaction rate expression for the solution.
forward reaction consists of two or more
independent terms corresponding to parallel H 3 AsO 3 + I3 + H 2 O = H 3 AsO 4 31" + 2H +
reaction paths, there will be the same number (4.1.13)
of independent terms in the rate expression for
the reverse reaction. At equilibrium not only is
The experimentally observed rate expression is
the total rate of the forward reaction equal to
the total rate of the reverse reaction, but the /c[H 3 AsO 3 ][I 3 ]
forward rate by each path is equal to the r= (4.1.14)
reverse rate for that particular path.
Guideline 8. Transitory intermediates (highly re- and two proposed mechanisms are based on the
active species) do not react preferentially with following sequence of elementary reactions.
k'
H 2 O I + + H 2 AsO 3 " -• H 2 A s O 3 I + H 2 O (Slow step for mechanism I)
(Equilibrium constant K3 for mechanism II) (4.1.17)
H2AsO3+ + H 2 O ^ ± H 3 A s O 4 + H+ (4.1.19)
one another to the exclusion of their reaction For the first mechanism the overall reaction
with stable species. The concentration of these rate may be taken to be that of the rate limiting
intermediates is usually so low that the number step.
of encounters an intermediate may undergo r = /c / [H 2 OI + ][H 2 AsO 3 ] (4.1.20)
with other intermediates will be very small
compared to the number of encounters it will With the use of the equilibrium steps 4.1.15
undergo with stable molecules. and 4.1.16, this expression may be converted to
86 Molecular Interpretations of Kinetic Phenomena
a form consistent with the experimental rate Guideline 13. / / the order of a reaction with re-
expression spect to one or more species increases as the con-
centration of that species increases, it is an
= kK1K2 -12
(4.1.21) indication that the reaction may be proceeding
[I"] by two or more parallel paths. Liebhafsky and
since the solvent concentration is just another Mohammed (6) have reported that for the re-
constant. action between hydrogen peroxide and iodide
For the second mechanism, equation 4.1.18 ion, the order with respect to hydrogen ion in-
is the rate controlling step, and the arrows in creases from zero to one as the pH is decreased.
parentheses indicate the assumptions made
r = /c 2 [H 2 O 2 ][H + ][r]
regarding the reversibility of reactions 4.1.17
and 4.1.18. The overall reaction rate is now (4.1.24)
r = /c"[H2AsO3I] (4.1.22) Such expressions are typical of many acid and
base catalyzed reactions. (See Section 7.3.1.)
With the use of the equilibrium expressions for
the first three reactions, equation 4.1.22 may be Guideline 14. If there is a decrease in the order of
converted to the following form, a reaction with respect to a particular substance
as the concentration of that species increases, the
[H3AsO3][l3-] dominant form of that species in solution may be
(4.1.23)
[H+][r]2 undergoing a change brought about by the change
which is also consistent with the experimental in concentration. A decrease in reaction order
rate expression. with respect to hydrogen ion concentration with
In both cases the negative reaction orders increasing acidity has frequently been observed
arise from equilibria that are established prior for reactions involving weak acids.
to the rate controlling step. A final rate expres- The guidelines enumerated above are by no
sion depends only on equilibria that are estab- means complete, but they do provide a starting
lished by elementary reactions prior to the rate point for the beginning student. Further details
determining step. Subsequent equilibria (e.g. are available in publications by Edwards et al.
4.1.19) do not influence its form. (5) and King (7).
Guideline 11. Whenever a rate law contains non-
integer orders, there are intermediates present in 4.1.5 Additional Methods and Principles
the reaction sequence. When one observes a frac- used in Investigations of Reaction
tional order in an empirical rate expression for a Mechanisms
homogeneous reaction, it is often an indication
Although reaction rate expressions and reaction
that an important part of the mechanism is the
stoichiometry are the experimental data most
splitting of a molecule into free radicals or ions.
often used as a basis for the postulation of re-
Guideline 12. If the magnitude of the stoichio- action mechanisms, there are many other experi-
metric coefficient of a reactant exceeds the order mental techniques that can contribute to the
of the reaction with respect to that species, there elucidation of these molecular processes. The
are one or more intermediates and reactions after conscientious investigator of reaction mecha-
the rate determining step. Before applying this nisms will draw on a wide variety of experi-
rule, one must write the stoichiometric equation mental and theoretical methods in his or her
for the reaction in a form such that all coefficients research program in an attempt to obtain infor-
are integers. (See Illustrations 4.1 and 4.2.) mation about the elementary reactions taking
4.1 Reaction Mechanisms 87
place in the system of interest. Some useful and Gowenlock (9), Anderson (10), Friess, Lewis,
techniques are described below. The information and Weissberger (11), Lewis (12), and Hammes
that they supply may provide a basis for choice (13).
between two or more alternative mechanisms, or
it may provide additional circumstantial evi-
4.1.5.2 Isotopic Substitution Techniques. If one
dence that a proposed mechanism is-, indeed,
correct. can discover where the various portions of
reactant molecules end up in the product mole-
cules, a great deal of insight into the mechanism
4.1.5.1 Studies of Reaction Intermediates. In of the reaction can be gained. By isotopically
the past two decades the most fruitful progress in labeling a functional group or atom in a mole-
experimental studies of reaction mechanisms has cule and examining the distribution (or lack
come about through the development of instru- thereof) of the tagged species after reaction, it is
mental techniques for detecting and identifying possible to determine which bonds are broken
the trace amounts of active intermediates pro- and formed. In studies with radioactive and sta-
duced in complex reaction systems. Suchistudies ble isotopes it is generally assumed that the
place a mechanism on a much firmer basis than kinetic behavior of a labeled atom is essentially
exists in those cases where the identity of inter- the same as that of an unlabeled atom. The im-
mediates can only be surmised. portant exceptions are studies involving deu-
A stable species that is suspected to ^ct as an terium and tritium.
intermediate in a complex chemical reaction can There are three basic types of mechanistic
often be added to a reaction mixture and its information that are derived from isotopic
effects observed. If the appropriate products are experiments.
formed at a rate no less than that of the "un-
1. A knowledge of which bonds are broken and
interrupted" or "original" reaction, this is strong
thus give rise to intramolecular or inter-
evidence that the reaction proceeds through the
molecular rearrangements.
intermediate that has been isolated. This evi-
2. A knowledge of whether the reaction pro-
dence is, however, not unequivocal. Similarly,
ceeds in an isolated molecule or requires the
the failure of other presumed intermediates to
participation of more than one molecule.
give the correct products under appropriate re-
action conditions will cause a kineticist to, revise 3. A knowledge of which products are the pre-
his or her ideas concerning the mechanism of cursors of others.
the reaction being investigated. For example, As an example, consider the thermal and
i photochemical decomposition of acetaldehyde
the Clemmensen reduction of carbonyl (C=O)
i '
groups to methylene (CH2) groups, using zinc CH3CHO -> CH4 + CO (4.1.25)
amalgam and hydrochloric acid, cannot proceed One proposed mechanism involved an in-
i tramolecular rearrangement, while a second
through the corresponding carbinol (HCOH), involved a free radical chain mechanism com-
since the carbinols themselves are not generally posed of the following sequence of elementary
reduced by the same combination of reactants reactions:
(8). CH 3 CHO^CH 3 +CHO- (4.1.26)
For detailed treatments of the experimental
methods used in studies of reaction interme- - CHjCHO-^CH4 + CH 3 O (4.1.27)
diates, consult the works of Wayne (1), Melville CH 3 CO-^CH 3 +CO (4.1.28)
88 Molecular Interpretations of Kinetic Phenomena
where reactions 4.1.27 and 4.1.28 occur many, an equal flow in the opposite direction. This
many times for each time that reaction 4.1.26 implies that the reaction path established as most
occurs. If one uses a mixture of CH 3 CHO and probable for the forward direction must also be
CD 3 CDO to study this reaction, the first mecha- the most probable path for the reverse reaction.
nism predicts that only CD 4 and CH 4 will be This consequence is also known as the principle
formed, while the second predicts that a mixture of detailed balancing of chemical reactions. Its
of CD 4 , CH 4 , CH 3 D, and CD 3 H will be formed. relationship to the principle of microscopic re-
The fact that the statistically expected distribu- versibility has been discussed by Denbigh (18).
tion of the latter methane species was observed If we consider a substance that can exist in three
experimentally (14-16) is taken as evidence that intraconvertible isomeric forms, A, B, and C
the major path for this reaction is the chain (e.g., frans-butene-2, ds-butene-2, and 1-butene),
reaction sequence. there is more than one independent reaction that
occurs at equilibrium. The conditions for ther-
4.1.5.3 Stereochemical Methods. Many or- modynamic equilibrium would be satisfied if
ganic reactions involve reactants that can exist there were a steady unidirectional flow at the
in stereoisomeric forms. If an optically active molecular level around the cycle
reactant yields a product that is capable*of ex-
hibiting optical activity, but that does not exhibit
such activity, it is usually assumed that the re-
action must involve an intermediate that is not
(4.1.29)
optically active (e.g., a carbonium ion interme-
diate). If it can be determined that the configura-
tion of an optically active product differs from
or is similar to that of the optically active re-
actant, one gains information about whether
there is an inversion of configuration. Retention such that the concentration of each species re-
of configuration might imply a two-step process, mains constant. However, this flow would not be
the first step turning the molecule "inside out," in accord with the principle of microscopic re-
and the second step turning it "outside in" again. versibility. If this principle were not applicable,
It could also mean that none of the bonds of the then the concentrations of the various species
optically active carbon atom are broken in the would show oscillations if one started with a
reaction. nonequilibrium system and allowed it to ap-
proach equilibrium. Several attempts have been
4.1.5.4 The' Principle of Microscopic Reversi- made to observe oscillatory phenomena of this
bility. The principle of microscopic rever- type, but they have not led to definitive results.
sibility is based on statistical mechanical One concludes that each of the reactions should
arguments and was first formulated by Tolman be balanced individually.
(17) in 1924.
In a system at equilibrium, any molecular process
and the reverse of that process occur on the aver-
age at the same rate.
(4.1.30)
The most "significant consequence of the prin-
ciple for kineticists is that if in a system'' at
equilibrium there is a flow of reacting molecules
along a particular reaction path, there must be
4.1 Reaction Mechanisms 89
The requirement of detailed balancing has where the reaction is again written in the
implications with regard to the relationships exothermic direction, the activation energy is
that must exist between reaction rate constants 28% of the sum of the AB and CD bond
and equilibrium constants. It requires that at strengths. For the reverse endothermic re-
equilibrium each reaction must be balanced in action, the activation energy is the sum of the
the forward and reverse directions; that is, standard energy change for the reaction and
fcj[A] = fc_![B] (4.1.31) the activation energy for the exothermic
reaction.
fc2[B] = fc_2[C] (4.1.32)
/c3[C] = fc_3[A] (4.1.33) These rules provide only very crude estimates
or of activation energies and are not sufficiently
sensitive to show differences in a series of related
^-=P^ = K1 (4.1.34) reactions.
Semenov (20) has suggested that for exother-
C mic abstration and addition reactions of atoms
= K, (4.1.35)
J k- and small radicals, the following relation is
A useful.
= K? (4.1.36) E = 11.5 - 0.25^ (4.1.37)
k..
The principle of detailed balancing provides where q is the heat evolved in the reaction ex-
an automatic check on the self-consistency of pressed in kilocalories and E is also expressed in
postulated reaction mechanisms when equilib- kilocalories. For the endothermic case,
rium can be approached from both sides. E = 11.5 + 0.15q (4.1.38)
4.1.5.5 Activation Energy Considerations. Ac- Other theoretical approaches to the problem
tivation energy considerations can provide a of predicting reaction activation energies exist
basis for eliminating certain elementary reac- (21-23). For our purposes, however, it is suffi-
tions from a sequence of reactions. Unfor- cient to recognize that "ball-park" estimates are
tunately, the necessary activation energy data is the best one can expect. Such estimates are often
seldom available, and one must estimate these adequate for purposes of differentiating between
parameters by empirical rules and generaliza- alternative mechanisms on the basis of a compar-
tions that are of doubtful reliability. ison of predicted and actual activation energies.
Activation energies for bimolecular reactions
have been correlated with bond energy data by 4.1.5.6 Families of Reactions. Research in
the use of the Hirschfelder rules (19). chemical kinetics can be influenced in several
ways by similarities in reactivity among com-
1. For a simple displacement reaction involving pounds having similar chemical structures. Con-
atoms or radicals, such as A + BC -• AB + sequently, one of the first steps in a kinetics
C, where the reaction is written in the exo- investigation should be a careful study of the
thermic direction, the activation energy is published literature on related reactions. While
5.5% of the dissociation energy of the bond one may occasionally be led astray by tentative
which is broken. For the reverse endothermic conclusions drawn from a survey of the litera-
reaction, the standard energy change of re- ture, the possibility of achieving large savings in
action must be added to this quantity in order time and laboratory effort is so much more likely
to obtain the activation energy. that it should be considered a must. Such a
2. For an exchange reaction such as survey furnishes a large bank of tested ideas and
AB + CD -+ AC + BD hypotheses about elementary reactions on which
90 Molecular Interpretations of Kinetic Phenomena
the investigator may draw for inspiration and ciple is named for the fourteenth-century philos-
stimulation. They provide a framework against opher William Ockham (William of Occam),
which one can orient and test ideas about the who was famous for his hardheaded approach to
mechanism of a specific reaction. problem solving. He believed in shaving away
The fact that compounds having similar chem- all extraneous details—hence the term Occam's
ical structures often react in similar ways implies Razor. In essence his axiom states that when
that they follow corresponding mechanisms in there are several possible solutions to a problem,
proceeding from reactants to products. One the right one is probably the most obvious.
must have a very sound basis in experimental Therefore one should not postulate or accept
fact to be able to defend successfully a proposed mechanisms that are more complex than neces-
reaction mechanism that contains elementary sary to explain all observed experimental facts.
reactions that represent a major departure from
those accepted by the scientific community as
4.2 CHAIN REACTIONS
being a proper mechanistic interpretation of re-
lated reactions. On occasions such departures The reaction mechanisms treated thus far have
are necessary, and they themselves may provide involved the conversion of reactants to products
the key to an improved understanding of the by a sequence of elementary reactions proceed-
common family of related reactions. ing in simple stepwise fashion. Once these se-
quences have been completed, all of the reaction
4.1.5.7 Occam''s Razor—A Rule of Simplicity. intermediates haye disappeared and only stable
Another principle occasionally used as a basis product molecules remain. These types of re-
for choice between two alternative mechanisms action are classified as open sequence reactions
is the rule of simplicity that is derived from a because they always proceed in the stagewise
more general philosophical principle known as fashion with no closed reaction cycles wherein a
Occam's Razor. The kineticist resorts to this rule product of one elementary reaction is fed back to
when all of the experimental and theoretical react with another species in an earlier elemen-
information that has been brought to bear on the tary reaction. (Reversible reactions are, however,
problem of determining the mechanism of a re- permitted in open sequence mechanisms.) This
action leaves two (or more) mechanisms, both section deals with closed sequence or chain re-
consistent with the facts. In this case one assutoes action mechanisms in which one of the reaction
that nature prefers the simple to the complex and intermediates is regenerated during one stage of
that the reaction follows the simplest path. In the reaction and is fed back to an earlier stage
many cases a serious question may exist as to to react with other species so that a closed loop
which mechanism is simpler but, »in others, the or cy<fle results. The intermediate species are
investigator may decide that one mechanism is thus periodically renewed by reaction, and the
more likely because it involves only bimolecular final products are the result of a; cyclic repetition
steps and no termolecular processes, or because of the intervening processes. Because chain re-
it involves a smaller change in chemical structure actions play such an important role in systems of
than an alternative mechanism. Complicated practical industrial significance, it is important
atomic rearrangements that occur Hn a jingle thAt the chemical engineer understand their
step are regarded with suspicion, and simpler basic nature. Failure to comprehend the impli-
stepwise processes are generally considered more cations of chain reaction processes can lead to
probable. hazardous explosions.
Occam's Razor states that "multiplicity ought To illustrate the nature of chain reactions, we
not to be posited without necessity." This prin- will consider a classic example—the formation
4.2 Chain Reactions 91
of hydrogen bromide from molecular hydrogen actions that do not involve the formation of
and bromine in a homogeneous gas phase species capable of maintaining the chain; for
reaction. example,
H 2 + Br2 ^± 2HBr (4.2.1) Br- + Br- + M -> Br2 + M (4.2.7)
This stoichiometric equation does not give any where M represents a third body or molecule
indication of the complex nature of the mecha- capable of soaking up and carrying away the
nism by which this reaction proceeds. energy released by the formation of a Br-Br
In certain highly energetic collisions with any bond.
molecule M in the system, a bromine molecule Bond energy considerations indicate that the
may be dissociated in a homolytic split of the initiation reaction (4.2.2) should be quite slow
bond joining two bromine atoms. because its activation energy must be quite high
(at least equal to the bond dissociation energy).
Br2 + M -> Br- + Br- + M (4.2.2) If one were dealing with an open sequence reac-
tion mechanism, such a step would imply that
The collision must be sufficiently energetic
the overall reaction rate would also be low be-
that enough energy is available to break the
cause in these cases the overall reaction becomes
chemical bond linking the two bromine atoms.
approximately equal to that of the rate limiting
This type of reaction is called an initiation reac-
step. In the case of a chain reaction, on the other
tion because it generates a species that can serve
hand, the overall reaction rate is usually much
as a chain carrier or active center in the following
faster because the propagation steps occur many
sequence of elementary reactions.
times for each time that an initiation step occurs.
Br- + H 2 -+ HBr + H- (4.2.3) The chain length of a reaction is defined as
the average number of cycles in which a chain
H- + Br2 -+ HBr + Br- (4.2.4) carrying species participates from the time that
^ + H- (4.2.5) it is first formed until the time that it is destroyed
in a termination reaction. Depending on the
H- + Br2 -• HBr + Br- (4.2.6) relative rates of the propagation and termina-
tion steps, a chain may involve only a few links
etc. or it may be extremely long. For example, a
chain length of 106 has been determined for the
Equations 4.2.3 and 4.2.4 are the elementary gas phase chain reaction between hydrogen and
reactions responsible for product formation. chlorine. Chain length represents a statistical
Each involves the formation of a chain carrying concept; some cycles will terminate before and
species (H- for 4.2.3 and Br- for 4.2.4) that propa- others after the average number of propagation
gates the reaction. Addition of these two rela- steps. This fact leads to a distribution of molecu-
tions gives the stoichiometric equation for the lar weights in the polymers formed by chain re-
reaction. These two relations constitute a single actions. In terms of quantities that can be
closed sequence in the cycle of events making up determined in the laboratory, we find it con-
the chain reaction. They are referred to as propa- venient to define the chain length by the follow-
gation reactions because they generate product ing ratio.
species that maintain the continuity of the chain.
_ rate of disappearance of reactants
The chain carriers need not all react with
reactant molecules in chain propagation reac- rate of chain carrier formation by initiation
tions. Some will disappear in termination re- (4.2.8)
92 Molecular Interpretations of Kinetic Phenomena
In the case of the hydrogen-bromine reaction, 4.2.1 The Reaction between Hydrogen and
each of the elementary propagation reactions Bromine H2 -h Br2 ^ 2HBr
led to the formation of a single chain carrier.
In 1906 Bodenstein and Lind (24) investigated
This type of reaction is said to be a straight or
the gas phase homogeneous reaction between
linear chain reaction. Some mechanisms involve
molecular bromine and molecular hydrogen at
elementary propagation reactions in which more
pressures in the neighborhood of 1 atm. They
than a single chain carrier is formed by the
fitted their experimental data with a rate expres-
reaction. This type of reaction is known as a
sion of the form
branching reaction. Examples of such reactions
are 4HBr] /c[H2][Br2] 1/2
(4.2.10)
and • OH + O
*
(4.2.9)
where k! is a constant independent of tempera-
Both of these reactions involve the production ture and k follows the normal Arrhenius form
of two active centers where there was only one for a reaction rate constant with an activation
before. When reactions of this type occur to a energy of 40.2 kilocalories per mole.
significant extent, the total number of active In 1919 Christiansen (25), Herzfeld (26), and
centers present in the system can increase very Polanyi (27) all suggested the same mechanism
rapidly, since a multiplication effect sets in as the for this reaction. The key factor leading to their
chains propagate. The growth of chain carriers in success was recognition that hydrogen atoms
a branched chain reaction is pictured below. and bromine atoms could alternately serve as
chain carriers and thus propagate the reaction.
By using a steady-state approximation for the
concentrations of these species, these individuals
were able to derive rate expressions that were
consistent with that observed experimentally.
Their original mechanism consists of the fol-
lowing elementary reactions.
Initiation:
Br9 2Br- (4.2.11)
Propagation:
When chain termination processes are opera- k2
Br- - H- -f - HBr (4.2.12)
tive such that they destroy chain carriers at a
rate that makes the average net production of H- -+ B r 2 ^ Br- -f HBr (4.2.13)
chain carriers unity, one has a situation corre- k4
H- + HBr -> H 2 -f- Br- (4.2.14)
sponding to a nuclear power reactor generating
energy at steady-state conditions. The effect of Termination:
the destruction processes is to cause the chain
2 B , ^ Br. (4.2.15)
branching process to degenerate to a straight
chain process. On the other hand, as the sketch Other reactions need not be considered on
indicates, there can be a multiplication of chain the basis of the arguments presented below. Re-
carriers. In physical terms this can lead to the action 4.2.14 is the reverse of reaction 4.2.12 and
chemical analog of a nuclear explosion. (See is responsible for the inhibition of the reaction
Section 4.2.5.) by HBr. Steps 4.2.12 to 4.2.14 all produce one
4.2 Chain Reactions 93
chain carrier to replace the chain carrier de- This relation expresses the fact that under
stroyed by the reaction. The sharp-eyed observer steady-state conditions, the rate of the initiation
may have noted that the initiation and termina- reaction is equal to the rate of the termination
tion reactions proposed by these individuals reaction, and the steady-state bromine atom
differ slightly from those discussed previously. concentration is equal to that which would arise
The reasons for this are dealt with later. from the equilibrium Br2 ^± 2Br; that is,
From these equations, the overall rate oTfor- l/2
mation of HBr is given by [Br]ss = [^ [Br2] (4.2.20)
As it stands, this expression is awkward and which, in turn, may be solved for the steady-state
inconvenient to test because it contains the concentration of atomic hydrogen.
concentrations of bromine and hydrogen atoms, 1/2
parameters that are not easily measured. In [H2]
principle these quantities could be eliminated by
solving this equation simultaneously with the [H],, = /c 3 [Br 2 ] + /c4[HBr]
(4.2.22)
of the major breakthroughs in the evolution of others based on studies of various atomic recom-
the theory of chemical reaction mechanisms. bination reactions imply that the "correct"
Since the mechanism was first published, modi- initiation and termination reactions are not
fications in the initiation and termination steps reactions 4.2.11 and 4.2.15 but reactions 4.2.25
have been required as additional experimental and 4.2.26 (28).
facts have been established by subsequent in- The next step in analysis of the mechanism is
vestigators. It is enlightening to consider these to indicate why we have limited the number of
modifications and the rationale underlying the elementary reactions in the mechanism to five.
omission of certain elementary reactions from To one uninitiated in the task of dealing with
the mechanism. reaction mechanisms, it is difficult to see why
Because the steady-state assumption leads to elementary reactions such as
the equilibrium relation for the bromine atom H 2 + Br2 -> 2HBr (4.2.27)
concentration (4.2.20), it does not matter what
mechanism is assumed to be responsible for H2 + M • 2H + M (4.2.28)
establishing this equilibrium. Alternative ele- HBr + M • H + Br + M (4.2.29)
mentary reactions for the initiation and termina- Br + HBr • Br 2 + H (4.2.30)
tion processes, which give rise to the same
equilibrium relationship, would also be consis- H + Br + M • HBr + M (4.2.31)
tent with the observed rate expression for HBr H + H + M • H2 + M (4.2.32)
formation. For example, the following reactions were rejected on an a priori basis and what con-
give rise to the same equilibrium: siderations led to their rejection. We will see
Initiation: Br2 + M -i 2Br + M (4.2.25) that they may be eliminated from consideration
through the use of some of the concepts of
Termination: M + 2Br 4 Br2 + M (4.2.26) probability and the bond energy requirements
where M is a third body. It represents any avail- outlined previously. If any of these reactions
able molecule that is capable of supplying to were important, the rate expression for the
bromine molecules the energy necessary for dis- revised mechanism would have a different math-
sociation of the Br-Br bond and of carrying ematical form than the empirical rate expression.
away enough energy in the termolecular re- Equation 4.2.27 may be rejected even if it
action so that the Br-Br bond does not immedi- provides a parallel path competitive with the
ately dissociate. mechanism described above because its pre-
If equations 4.2.25 and 4.2.26 are substituted sence would require a second-order term in the
for equations 4.2.11 and 4.2.15, respectively, in experimental reaction rate expression.
the mechanism described above, the effect is to The initiation reaction is the dissociation of
replace kx by fc'^M] and k5 by fc'5[M] every- molecular bromine rather than the dissocia-
where that they appear. Since these quantities tion of molecular hydrogen because the energy
appear as a ratio in the final rate expression, the necessary to dissociate the latter is much greater
third body concentration will drop out and than that required to dissociate the former. This
(/Ci/fcs) becomes identical with (k\/k'5). The ne- fact is evident from a consideration of the
cessity for the use of the third body concentra- standard enthalpies of reaction.
tion thus is not obvious in kinetic studies of the Br2 + M -> 2Br + M AH, = 46.1 kcal/g-mole
thermal reaction. However, from studies of
(4.2.33)
photochemical reaction between hydrogen and
bromine, there is strong evidence that the ter- H 2 + M -> 2H + M AH n = 103.4 kcal/g-mole
mination reaction is termolecular. This fact and (4.2.34)
4.2 Chain Reactions 95
Since the activation energy of an endothermic If the preexponential factors of these rate con-
elementary reaction cannot be less than AH, the stants are comparable in magnitude, further
reaction with the significantly lower activation simplification is possible.
energy will occur much more frequently, other
9-(E2-E3)/RT (4.2.39)
factors being equal. Similar arguments permit
one to eliminate the HBr dissociation reaction [Br]
from consideration. If the estimation techniques discussed in
The Br + HBr reaction will be unimportant Section 4.1.5.5 are employed, one finds that E3
in competition with the Br + H 2 reaction be- is equal to 2.5 kcal. From 4.2.36, E2 is equal to
cause energy considerations again dictate that 17.6 kcal. At 540 °K the ratio of atomic con-
the former will have a much higher activation centrations is thus of the order of 10"6. Conse-
energy than the latter. quently, the reaction between atomic hydrogen
and atomic bromine is only one millionth as
Br + HBr-+Br2 + H AHIV = 41.2 kcal/g-mole likely as a reaction between two bromine atoms
(4.2.35) and that between two hydrogen atoms is only
Br + H 2 ^ H B r + H EA = 17.6 kcal/g-mole 1/1012 as likely. Although the atomic hydrogen
(4.2.36) concentration is extremely low under reaction
conditions, it has been estimated to be four
Both reactions are endothermic, but the inter- orders of magnitude greater than that which
action of bromine atoms with HBr is much more would be in equilibrium with molecular hydro-
so than the interaction with molecular hydrogen. gen. The excess hydrogen atom concentration
Consequently, the former reaction will occur over that which would exist at equilibrium is
much less frequently than the latter. "paid for" by the energy liberated by the reacting
Equations 4.2.31 and 4.2.32 represent radical mixture. This feature is a general characteristic
recombination reactions that might be con- of chain reactions. Superequilibrium steady-
sidered as alternative termination reactions. state concentrations of highly reactive inter-
Since these reactions and reaction 4.2.26 are rad- mediates can be produced as a result of the
ical recombination reactions, they will all have energy released by reaction.
very low activation energies. The relative rates The reaction between hydrogen and bro-
of these processes will then be governed by the mine illustrates many of the general features of
collision frequencies and thus the concentrations straight chain reactions. This reaction system
of the reacting species. The relative concentra- has been studied by many investigators, and
tions of hydrogen and bromine atoms can be the body of accumulated experimental data
determined from the steady-state form of equa- represents as nearly a consistent and complete
tion 4.2.17. collection of data as is available for any reaction
in the literature.
(4.2.37)
[Br] 4.2.2 Chain Reaction Mechanisms—
If we consider this ratio during the initial General Comments
stages of the reaction when the HBr concentra-
tion is extremely low and when the concentra- The essential characteristic of a chain reaction
tions of hydrogen and bromine are nearly equal, mechanism is the existence of a closed cycle of
this expression reduces to reactions in which unstable or highly reactive
intermediates react in propagation steps with
[H] k2 stable reactant molecules or other intermediates
(4.2.38)
[Br] ~ T3 and are regenerated by the sequence of reactions
96 Molecular Interpretations of Kinetic Phenomena
that follows. The mathematical methods and presence of such mechanisms may be recognized.
approximations used to express the overall rate Not all chain reactions will exhibit all of the
of reaction in terms of the individual rate con- characteristics enumerated below, but the pre-
stants and the concentrations of stable species sence of several of them should be a strong
are merely extensions of those discussed earlier. indication to the investigator that a chain re-
The elementary reactions comprising the action is probably taking place. The absence of
chain reaction mechanism are generally classi- any of these criteria should not be regarded as
fied as initiation, propagation, or termination particularly significant.
reactions. In the initiation reaction an active Some pertinent criteria for recognizing the
center or chain carrier is formed. Often these are presence of a chain reaction are:
atoms or free radicals, but ionic species or other
intermediates can also serve as chain carriers. 1. An induction period is present.
In the propagation steps the chain carriers inter- 2. Extremely large increases or decreases in the
act with the reactant molecules to form product reaction rate occur when relatively small
molecules and regenerate themselves so that the amounts of other substances are added to the
chain may continue. The termination steps reaction mixture. These materials may act
consist of the various methods by which the either as initiators or inhibitors of the reac-
chain can be broken. tion, depending on whether they participate
The key assumption that permits one to pro- in the reaction as initiators or terminators of
ceed from mechanistic equations to an overall the chain.
rate expression is that the steady-state approxi- 3. Small changes in pressure, temperature, or
mation for intermediates is valid. One is also composition can markedly affect the overall
usually required to make the long-chain approxi- reaction rate or cause an explosion.
mation (i.e., the by-products resulting from the 4. In gas phase reactions an increase in the sur-
initiation and termination reactions are assumed face to volume ratio of the reaction vessel may
to represent only a very small fraction of the reduce the reaction rate, while addition of
total products). The reaction then will remain "inert" gases may increase the reaction rate.
stoichiometrically simple to within an approxi- (See Section 4.2.5.)
mation that improves with increasing chain 5. Abnormally high quantum yields may occur
length. It is also sometimes necessary to make in photochemical reactions. Einstein's law of
the additional assumption that at steady state photochemical equivalence is the principle
the rate of the initiation process must be equal to that light is absorbed by molecules in discrete
the rate at which the various chain breaking amounts as an individual molecular process
processes are occurring. Otherwise, the concen- (i.e., one molecule absorbs one photon at a
tration of chain carriers would depend explicitly time). From optical measurements it is pos-
on time. Often this equality is obvious from the sible to determine quantitatively the number
algebraic equations resulting from the use of the of photons absorbed in the course of a reac-
steady-state approximation. When there are two tion and, from analyses of the product mix-
or more alternative termination reactions, the ture, it is possible to determine the number of
algebraic relations may be so complex that the molecules that have reacted. The quantum
equality is not apparent. yield is defined as the ratio of the number of
In any attempt to elucidate the mechanism of molecules reacting to the number of photons
a chemical reaction, it is very important to deter- absorbed. If this quantity exceeds unity, it
mine at an early stage whether or not a chain provides unambiguous evidence for the exis-
reaction is occurring. Consequently, we now tence of secondary processes and thus indi-
note some of the characteristics by which the cates the presence of unstable intermediates.
4.2 Chain Reactions 97
Thus
//c 1 [Cl 2 ][O 3 ]
[CIO 3 -L = (E)
' 2k,
Substitution of this expression into equation B gives
/c 1 [Cl 2 ]+/c 3
2k,
[C1O 2 ] SS = - (G)
Substitution of the steady-state concentrations of the two intermediates into the equation for the
rate of disappearance of ozone gives
Ik.
= -/c1[Cl2][O3]-/c2 («)
dt
or
4O3]
dt
(I)
The first term in equation I refers to ozone of ozone, a mechanism using this equation as
consumption via the initiation step. This term a termination step would give a rate expression
is negligible compared to the second. Thus, of the same form as that described above.
been recognized previously because earlier chain presented in Section 4.3.1.3, this initiation re-
reaction mechanisms had led to rate expressions action will shift to a bimolecular process at low
that were more complicated than those normally pressures.
associated with organic pyrolysis reactions. This particular mechanism assumes that Rx
The basic premises on which Rice-Herzfeld and Ri are different radicals and that the latter
mechanisms are based are as follows. do not participate in the propagation reactions.
Initiation: In the more general case the radical R\ can
1. Free radicals are formed by scission of the participate in propagation reactions analogous
weakest bond in the molecule. to reactions (2) and (3). These propagation steps
Propagation: consist of a bimolecular hydrogen abstraction
2. One or both of the radicals formed in the ini- reaction followed by a unimolecular decomposi-
tiation step abstracts a hydrogen atom from tion reaction.
the parent compound to form a small satu- The three possible termination reactions have
rated molecule and a new free radical. been written as combination reactions when, in
fact, disproportionation reactions may also oc-
3. The new free radical stabilizes itself by split-
cur. Under a given set of experimental con-
ting out a simple molecule such as an olefin
ditions, only one of the three chain breaking
or CO.
steps (4a), (4b), or (4c) can be expected to be
RCH2 - CH2- -> R- + CH2 = CH2 (4.2.40) important.
With this general sequence of reactions as our
Termination:
proposed mechanism, we are now prepared to
4. The chain is broken by a combination or
use the corresponding rate expressions and our
disproportionation reaction between two
standard assumptions to show that rate expres-
radicals.
sions with reaction orders of 1/2, 1, and 3/2 can
These basic premises go a long way in cor- be derived. The form of the rate expression that
relating and tying together the extraordinary results is an indication of the nature of the chain
complexity of many pyrolysis reactions. In breaking process.
terms of mechanistic equations, they may be A first-order rate expression results when the
written as termination process involves the dissimilar radi-
cals, Rx and R2. That is, reaction (4b) is the
Initiation:
chain breaking step.
(1) Mk-X R 1 + R 1 (4.2.41) Based on the set of mechanistic equations set
Propagation: forth above, the rate of disappearance of reac-
tant M is given by:
(2) Ri + M k-i • R 2 + R1H (4.2.42)
rf[M]
(3) R2 ^ R, -I- F (4.2.43) - /c 2 [Rj[M] (4.2.47)
Termination: For long chain reactions the amount of reac-
(4a) R, + R, p2 (4.2.44) tant undergoing decomposition via the ini-
tiation reaction is small compared to that
(4b) R, + R 2 ; p3 (4.2.45) decomposed by the propagation reactions (i.e.,
(4c) R2 + R 2 ' p4 (4.2.46) ^ [ R i ] » ^I- Thus, to a good approximation,
The initiation step has been written as a
= -k 2 [R!][M] (4.2.48)
unimolecular reaction. In terms of the theory dt
100 Molecular Interpretations of Kinetic Phenomena
fc3[R2] - (4.2.49)
dt
- /c 3 [R 2 ] - (4.2.50)
dt
Adding these two equations and rearranging, and it is apparent that the overall activation en-
we find that ergy of the reaction is given by
- E4 b
(4.2.51) E = (4.2.56)
2/c4b[Ri]
The average chain length i f is given by the
Substitution of this relation into equation following ratio.
4.2.50 gives
rate of disappearance of reactants
_, (4.2,2) rate of initiation process
k3 V / 2 (4.2.58)
Substitution of this relation into equation
4.2.48 gives an expression that is independent of the initial
reactant concentration.
[M] (4.2.54) The following mechanism has been proposed
2/c,4 b
for the decomposition of ethane into ethylene
which is a first-order rate expression. If each of and hydrogen. The overall rate expression is
the individual rate constants is written in the first order in ethane.
Initiation:
c2 H 6 ^2CH3 (4.2.59)
Chain transfer: CH 3 HH C 2 H 6 ^ C H 4 -f C 2 H 5 (4.2.60)
Propagation: c2 H5 - C 2 H 4 + H (4.2.61)
Termination:
H -HC2 H 6
r H 2+
H -h C 2 H 5 ^ C 2 H 6
C2H5 (4.2.62)
(4.2.63)
4.2 Chain Reactions 101
4C 2 H 6 ] 1
•[C 2 H 6 ] (4.2.64)
dt
2k, 2k,
which is a first-order rate expression. Physical Substitution of this result into equation
arguments indicate that kl/k3 must be very small 4.2.67 leads to a relation for [ R j .
compared to unity {k1/k3 « 1). Thus equation 1/2
4.2.64 becomes: (4.2.70)
2kd [M] 1/2
[C 2 H 6 ] (4.2.65) Combination of equations 4.2.66 and 4.2.70
dt
gives
A reaction rate expression that is proportional
to the square root of the reactant concentration
results when the dominant termination step is dt
reaction (4c), that is, the termination reaction (4.2.71)
occurs between two of the radicals that are in-
volved in the unimolecular propagation step. Since the first term on the right side of this
The generalized Rice-Herzfeld mechanism con- equation merely represents twice the rate of
tained in equations 4.2.41 to 4.2.46 may be em- initiation, it is negligible compared to the
ployed to derive an overall rate expression for second term. Hence
this case. d[W] 1/2
As before, the time rate of change of reactant = -fc, A [M] 1/2 (4.2.72)
concentration is given by dt 2k4c
and one has half-order kinetics.
-/c2[Rj[M] (4.2.66) In order for the overall rate expression to be
dt
3/2 order in reactant for a first-order initiation
The steady-state approximation may be used process, the chain terminating step must involve
for each of the chain carrying species. a second-order reaction between two of the
radicals responsible for the second-order prop-
dt agation reactions. In terms of our generalized
(4.2.67) Rice-Herzfeld mechanistic equations, this means
that reaction (4a) is the dominant chain breaking
= *2[Ri][M] - /c3[R2] - 2/c4c[R2]2 « 0 process. One may proceed as above to show that
dt
(4.2.68) the mechanism leads to a 3/2 order rate expres-
sion.
Adding these two equations and rearranging, For Rice-Herzfeld mechanisms the mathe-
one finds that: matical form of the overall rate expression is
strongly influenced by the manner in which the
R2 l ,42.69,
" 2k4c J chains are broken. It can also be shown that
102 Molecular Interpretations of Kinetic Phenomena
changing the initiation step from first order to additive that slows down a reaction is called an
second order also increases the overall order of inhibitor.
the reaction by 1/2 (30). These results are easily In the case of chain reactions a mere trace of
obtained from the equations derived previously inhibitor can reduce reaction rates by orders of
by substituting k\ [M] for kl everywhere that magnitude. Such inhibitors break the chain,
the latter term appears. Since k1 always appears perhaps as a result of a reaction in which a
to the 1/2 power in the final rate expressions, relatively nonreactive free radical is formed.
the exponent on M in the existing rate expres- Another manner in which an inhibitor may act
sion must be added to 0.5 in order to obtain the is by combining with a catalyst and rendering
overall order of the reaction corresponding to it inoperative.
the bimolecular initiation step. In like manner, Other additives can promote chain reactions
shifts from bimolecular to termolecular termi- by acting as initiators of the chain. These
nation reactions will decrease the overall order compounds are materials that can be more
of the reaction by 1/2. readily dissociated than the primary reactants.
Although the above Rice-Herzfeld mecha- For example, benzoyl peroxide has often been
nisms lead to simple overall rate expressions, used to initiate the polymerization of olefinic
do not get the impression that this is always the monomers because of the ease with which it
case. More detailed discussions of these types dissociates.
of reactions may be found in textbooks (31-34) In many cases complex chain reactions are
and in the original literature. characterized by an induction period (i.e., a
period at the beginning of the reaction during
4.2.4 Inhibitors, Initiators, and which the rate is significantly less than that
Induction Periods which subsequently prevails). This period is
often caused by the presence of small amounts
Since the complex decomposition reactions of inhibitors. After the inhibitors are consumed,
considered in the previous section are propa- the reaction may then proceed at a much faster
gated by a series of elementary steps, it might be rate.
expected that they and other chain reaction
processes will show an unusual sensitivity to 4.2.5 Branched Chain Reactions
any substance or physical condition that inter- and Explosions
feres with the propagation of the chain. It is
therefore not surprising to find that the rates of Chain reactions can lead to thermal explosions
such reactions can be markedly reduced either when the energy liberated by the reaction
by the presence of trace quantities of certain cannot be transferred to the surroundings at a
chemical substances or by changes in the sufficiently fast rate. An "explosion" may also
physical condition of the surface of the vessel in occur when chain branching processes cause a
which the reaction is being studied. The latter rapid increase in the number of chains being
changes include both variations in the surface/ propagated. This section treats the branched
volume ratio of the reactor brought about by chain reactions that can lead to nonthermal
adding spun glass or some other high surface explosions and the physical phenomena that
area material to the reactor and variations in are responsible for both branched chain and
the nature of the surface itself through the use thermal explosions.
of Teflon, syrupy phosphoric acid, or other A chain branching reaction is one that leads
coatings. The reduction in the rate may occa- to an increase in the number of chain carriers
sionally be referred to as negative catalysis, but present in a reacting system (e.g., 4.2.9). When
it is more commonly known as inhibition. An such reactions occur to a significant extent, the
4.2 Chain Reactions 103
total number of radicals present in a reacting propagation and termination steps are written
mixture can increase rapidly so that steady-state as first-order processes, but this is not essential
conditions are no longer maintained. The re- to the argument that follows.
action proceeds at a very high velocity and, The rate expressions have been written in
since the associated release of chemical energy generalized fashion with the terms fp, fb9 fst, and
is also extremely rapid, the viewer sees this fgt containing the reaction rate constants, sto-
phenomenon as an explosion. ichiometric coefficients, and concentrations of
The generalized mechanism by which the various stable species present in the reaction
branched chain reactions proceed provides a mixture. If one also wished to consider bi-
basis for a semiquantitative understanding of molecular radical processes, these could also be
explosions resulting from chain branching.* lumped into the / parameters.
d[R]
Initiation: r; = (4.2.73)
dt
Propagation:
Termination:
.d[R]
Chain breaking at R -• ? (4.2.76)
solid surface dt
4R]
Chain breaking in R -• ? (4.2.77)
dt
gas phase
where R represents the various active centers or In terms of the above rate expressions, the
free radical chain carriers involved in the re- Bodenstein steady-state approximation for free
action, and the detailed chemistry of the various radical intermediates gives:
steps is unspecified. The number of radicals
formed in the chain branching step per radical = 0 (4.2.78)
that reacts is assigned the symbol a. All of the
Note that this equation does not contain a
:
Adapted from Chemical Kinetics by K. J. Laidler. Copy- term corresponding to the straight chain step in
right © 1965. Used with permission of McGraw-Hill which the primary products are formed. This
Book Company. step does not have any net effect on the total
104 Molecular Interpretations of Kinetic Phenomena
(4.2.79)
ft + fgt ~ /*(« - 1)
The overall rate of reaction is therefore Steady
reaction
(4.2.80) Explosion
fst + fgt ~
Since the branching parameter a is greater
than unity (usually it is 2), it is conceivable that
under certain circumstances the denominator of
the overall rate expression could become zero.
In principle this would lead to an infinite re-
action rate (i.e., an explosion). In reality it
becomes very large rather than infinite, since
the steady-state approximation will break down Total pressure
when the radical concentration becomes quite
large. Nonetheless, we will consider the condi- Figure 4.1
tion that fb((x — 1) is equal to (fst -f fgt) to be The rate of the hydrogen-oxygen reaction as a
a valid criterion for an explosion limit. function of the total pressure. (Adapted from
Chemical Kinetics by K. J. Laidier. Copyright
Because there are two positive terms in the
© 1965. Used with permission of McGraw-Hill
denominator of equation 4.2.85 (either of which
Book Company.)
may be associated with the dominant termina-
tion process), this equation leads to two explo-
sion limits. At very low pressures the mean free semiquantitative arguments can be used to
path of the molecules in the reactor is quite explain the phenomenon known as the second
long, and the radical termination processes explosion limit. At these pressures the dominant
occur primarily on the surfaces of the reaction processes by which chains are terminated will
vessel. Under these conditions gas phase colli- occur in the gas phase. The increased pressure
sions leading to chain breaking are relatively hinders the diffusion of radicals to the vessel sur-
infrequent events, and fst » fgt. Steady-state faces and provides a number density of gas phase
reaction conditions can prevail under these radicals that is sufficient for radical recombina-
conditions if fst > fb(a — 1). tion and disproportionation processes.
As the pressure in the reaction vessel increases, If the pressure in the reaction vessel is now
the mean free path of the gaseous molecules will decreased, the rate at which chains are broken
decrease and the ease with which radicals can will also decrease. Eventually the pressure will
reach the surfaces of the vessel will diminish. reach a point at which the rates of the termina-
Surface termination processes will thus occur tion processes will become equal to the rate at
less frequently; fst will decline and may do so which the radical concentration is increasing
to the extent that fst + fgt becomes equal to because of the chain branching process; that is
fb(oc — 1). At this point an explosion will occur.
This point corresponds to the first explosion
limit shown in Figure 4.1. If we now jump to At this point an explosion will occur cor-
some higher pressure at which steady-state re- responding to the second explosion limit in
action conditions can again prevail, similar Figure 4.1.
4.3 Molecular Theories of Chemical Reaction Kinetics 105
There is a third explosion limit indicated in the sequence of molecular events that occur
Figure 4.1 at still higher pressures. This limit is during the course of a chemical reaction. Re-
a thermal limit. At these pressures the reaction action mechanisms should not be regarded as
rate becomes so fast that conditions can no experimental facts. They are plausible explana-
longer remain isothermal. At these pressures the tions of experimental data that are consistent
energy liberated by the exothermic chain reac- with the data, but that are subject to revision
tion cannot be transferred to the surroundings as new data are obtained. Even if one proposed
at a sufficiently fast rate, so the reaction mixture mechanism gives agreement with all available
heats up. This increases the rate of the process experimental facts, this is no sign that the
and the rate at which energy is liberated so one mechanism is unique and that other mechanisms
has a snowballing effect until an explosion could not give such agreement.
occurs.
It should be evident from this discussion that 4.3 MOLECULAR THEORIES OF
the first explosion limit will be quite sensitive CHEMICAL REACTION KINETICS
to the nature of the surface of the reaction vessel
A "complete" theory of reaction kinetics would
and its area. If the surface is coated with a
provide a basis for calculating the rate of an
material that inhibits the surface chain termina-
elementary reaction from a knowledge of the
tion process, the first explosion limit will be
properties of the reacting molecules and their
lowered. Inert foreign gases can also have the
concentrations. In terms of the present state of
effect of lowering the first explosion limit, since
our theoretical knowledge, a "complete" theory
they can hinder diffusion to the surface. If some-
can be regarded as a goal that is far, far down
thing like spun glass or large amounts of fine
the road. Existing theories are extremely un-
wire are inserted, one can effect an increase in the
satisfactory. Nevertheless, chemical engineers
first explosion limit by changing the surface/
must be cognizant of the primary features of
volume ratio of the system.
these theories.
4.2.6 Supplementary References 4.3.1 Simple Collision Theory
Our discussion of chain reaction processes is Before a chemical reaction can occur, energy
necessarily incomplete. For more detailed treat- must be available and localized in such a way
ments consult the following references. that it is possible to break and make certain
1. C. H. Bamford and C. F. H. Tipper (Editors), Compre- chemical bonds in the reactant molecules. More-
hensive Chemical Kinetics, Volume II, The Theory of over, the participants in the reaction must be
Kinetics, Elsevier, New York, 1969. in a spatial configuration such that it is possible
2. G. J. Minkoffand C. F. H. Tipper, Chemistry of Combus- for the atomic and electronic rearrangements to
tion Reactions, Butterworths, London, 1962. occur. The most common means by which such
3. N. N. Semenov, Some Problems in Chemical Kinetics and redistributions of energy and changes in geome-
Reactivity, Volumes I and II, Princeton University Press,
tric configurations can occur are molecular
Princeton, N.J., 1958, 1959.
collision processes. In this sense all theories of
reaction are collision theories. However, we will
4.2.7 Cautionary Note on
use this term in a more limited sense. We restrict
Reaction Mechanisms
its use to the theory that links chemical kinetics
It is appropriate to conclude our discussion of to the kinetic theory of gases by the use of the
reaction mechanisms on a note of caution. A theoretical expression for bimolecular collision
mechanism is merely a logical hypothesis as to frequencies.
106 Molecular Interpretations of Kinetic Phenomena
4.3.1.1 Rates of Bimolecular Reactions. For other books dealing with kinetic theory (35-39).
purposes of chemical reaction kinetics the pri- Different authors may write equations 4.3.1 and
mary result of the kinetic theory of gases is that 4.3.4 in slightly different forms because of differ-
the total number of collisions between A mole- ent assumptions involved in their derivations
cules and B molecules per unit volume per unit and the degree of mathematical rigor that they
time is given by ZAB. choose to use. They may precede these expres-
sions by some numerical factor that will not
Z
AB = n'An'Bno2AB (4.3.1) differ appreciably from unity. Such differences
are insignificant for purposes of chemical kine-
where tics. From a practical viewpoint the uncertainties
involved in predicting reaction rates do not
n'A and n'B are the number densities of molecules
justify using anything more complex than the
A and B, respectively
simple hard sphere model of the kinetic theory.
T is the absolute temperature
Bimolecular processes are the primary vehicle
kB is the Boltzmann constant
by which chemical change occurs. The frequency
oAB is the arithmetic average of the hard
with which these encounters occur is given by
sphere diameters of molecules A
equations 4.3.1 and 4.3.4. However, only an
and B
extremely small number of the collisions actually
OA + On lead to reaction; for predictive purposes one
(4.3.2)
needs to know what fraction of the collisions
are effective in that they lead to reaction.
and \xAB is the reduced mass of the system ex- Since chemical reactions involve the making
pressed in terms of the molecular masses mA and breaking of chemical bonds with their as-
and mB as sociated energy effects and geometric require-
mAmB ments, it is not unreasonable to assume that
VAB = (4.3.3)
mA mR these factors play an important role in deter-
mining the probability that a bimolecular colli-
If A and B are identical, then mA = mB = ra, sion will lead to chemical reaction. In addition
\i = m/2, GA = oB = oAB = a, and n'A = n'B = to these factors there are restrictions on bi-
n'. To avoid double counting of collisions, an molecular combination or association reactions
additional factor of 1/2 must also be introduced and quantum mechanical requirements that can
into equation 4.3.1 to obtain an expression for influence this probability.
the number of collisions between identical mole- For complex organic molecules, geometric
cules per unit volume per unit time. considerations alone lead one to the conclusion
that only a small fraction of bimolecular colli-
ZAA = (n')2~a sions can lead to reaction. One can represent
the fraction of the collisions that have the proper
geometric orientation for reaction by a steric
= 2(n')V (4.3.4) factor (Ps). Except for the very simplest reactions,
this factor will be considerably less than unity.
For a typical gas at standard conditions, ZAA On the basis of simple collision theory, it is not
and ZAB are both of the order of 10 28 collisions possible to make numerical estimates of Ps9
per cubic centimeter per second. although it may occasionally be possible to
Derivations of the above expressions are con- make use of one's experience with similar re-
tained in most physical chemistry texts and actions to determine whether Ps for a given
4.3 Molecular Theories of Chemical Reaction Kinetics 107
reaction will be large or small. This failure to provides the surprisingly simple result that the
be able to predict values of Ps is one of the fraction of the collisions that will involve at least
major weak points of the collision theory. It an energy sc directed along the line of centers is
results from the inability of the theory to con- given by e~£c/kBT or e~Ec/RT. If the steric factor
sider the individual geometric shapes of various defined previously is introduced, the rate of re-
reactant molecules. Since Ps must be determined action between unlike molecules becomes
empirically, it may be considerably in error
because it will contain all pf the errors asso- e — KnAnB
rAB = Psn'An'B™2AB\
ciated with the assumption, of hard sphere
molecules as well as any quamtmi mechanical (4.3.7)
restrictions on the reaction in question.
Of all the factors that determine the effective- Comparison of this equation with the Ar-
ness of a collision, the energy requirement is by rhenius form of the reaction rate constant reveals
far the most important. Reaction cannot occur a slight difference in the temperature dependence*
unless sufficient energy is provided and localized of the rate constant, and this fact must be ex-
so that the appropriate bonds can be broken plained if one is to have faith in the consistency of
and new bonds formed. It is reasonable to the collision theory. Taking the derivative of the
suppose that there will be a threshold energy natural logarithm of the rate constant in equa-
requirement below which these processes can- tion 4.3.7 with respect to temperature, one finds
not occur. It is usually assumed that a collision that
is effective for reaction purposes only if the rl Pn h 1 V
+ (4.3.8)
relative kinetic energy along the line of col- dT IT RT2
lision centers is greater than or equal to this
threshold value. Kinetic energy associated with while the Arrhenius form is given by equation
motion perpendicular to this line corresponds 3.3.54.
to a sideswipe that leads to changes in rotational dlnk EA
levels but cannot be expected to be effective in (4.3.9)
dT RT2
promoting reaction.
The principles of kinetic theory may be used These last two equations can be reconciled
to arrive at an expression for the number of only if the experimental activation energy is tem-
collisions whose relative kinetic energy along perature dependent in such a way as to accom-
the line of centers is greater than sc. The result modate the variation in collision frequency with
is the following expression for the number of temperature; that is,
such collisions per unit volume per unit time. RT
EA = Ec (4.3.10)
Qh
n
— AnBn(JAB
Usually the last term on the right side of equation
4.3.10 is only a very small fraction of the total.
If this equation is now written in terms of energy
For example, if the observed activation energy
per mole instead of per molecule, it becomes
(EA) is 10 kcal at 500°K, the RT/2 term is only
i 1
/ 2 0.5 kcal or 5%. Such discrepancies are usually
,-Ec/RT
(4.3.6) within the scatter of the data.
Illustration 4.4 indicates how collision fre-
an expression that becomes identical with equa- quency calculations are used to obtain a reaction
tion 4.3.1 if we set Ec equal to zero. Equation 4.3.5 rate expression.
108 Molecular Interpretations of Kinetic Phenomena
comings of the theory lie in its failure to relate One may estimate the relative frequency of
the steric factor and the activation energy to bimolecular and termolecular collisions using
molecular parameters from which a priori pre- equations 4.3.1 and 4.3.11.
dictions can be made.
1 1
nAnBnc (4.3.11)
where the symbols correspond to those used actions of small atoms and radicals (e.g., H +
previously and / is a somewhat indefinite para- H + M -> H 2 + M).
meter that should have a value typical of the
length of a chemical bond (i.e., t = 1 A). If one
4.3.1.3 Vnimolecular Reactions. Experimen-
assumes that the same considerations that gov-
tal evidence for unimolecular processes posed
ern the efficiency of bimolecular collisions are
a dilemma for early kineticists. They observed
relevant to termolecular processes, then the re-
that several species that appeared to be stable at
action rate of these processes is given by
low temperatures underwent first-order decom-
r = ZABCP«e -E/RT (4.3.12) position or isomerization reactions at higher
110 Molecular Interpretations of Kinetic Phenomena
For the case where 5 = 2 this expression energy values near the top of this minimal bar-
reduces to the simple exponential form of Ar- rier is referred to as the activated complex. The
rhenius. For values of S greater than 2, it yields region in hyperspace corresponding to this con-
a much larger probability of reaction than one figuration is referred to as the transition state.
would obtain from the normal Arrhenius form. Progress along the path corresponding to the
The enhancement may be several orders of series of positions occupied by the system as it
magnitude. For example, when S = 10 and moves from reactants to the activated complex
E/RT = 30, the ratio of the probability factor and on to the final product is measured in terms
predicted by Hinshelwood's approach to that of distance along the reaction coordinate. If this
predicted by the conventional Arrhenius method quantity is used as a measure of the progress of
is (30)4/4! = 3.375 x 104. The drawback of the reaction, it is possible to convert the multi-
approach is that one cannot accurately predict dimensional potential energy diagram into a
S a priori. When one obtains an apparent steric two-dimensional free energy diagram, one form
factor in excess of unity, this approach can often of which is shown in Figure 4.2. Such figures
be used in interpretation of the data. emphasize the commonly used analogy in whfch
the reaction and its energy requirements are
compared to the movement of a sphere across
4.3.2 Transition State Theory
hilly terrain in the presence of a gravitational
The transition state theory provides a useful field.
framework for correlating kinetic data and for In terms of the transition state theory the rate
codifying useful generalizations about the dy- of reaction is controlled by the rate of passage
namic behavior of chemical systems. This theory through the region at the top of the energy
is also known as the "activated complex theory," barrier. This passage may correspond either to
the "theory of absolute reaction rates," and movement over the top of the barrier or, in some
"Eyring's theory." This section introduces chem- cases, to quantum mechanical tunneling through
ical engineers to the terminology, the basic the barrier. Several approaches to formulating
aspects, and the limitations of the theory. the reaction rate in terms of such movements
The transition state theory differs from col- have been described, but we will employ the one
lision theory in that it takes into account the utilized by Eyring and his co-workers (52-54).
internal structure of reactant molecules. It de- The basic hypothesis on which Eyring's formu-
scribes a reaction in terms of movements on a lation is based is that an equilibrium exists
multidimensional surface in which the potential between certain activated complex species and
energy of the system is depicted as a function of the reactants. This equilibrium determines the
the relative positions of the nuclei constituting number of activated complexes corresponding
the participants in the reaction. The progress of to specified reactant concentrations, regardless
the reaction is characterized in terms of move- of whether or not an overall chemical equilib-
ment along the surface. Although one may en- rium exists between reactant and product spe-
vision an infinite number of paths linking the cies. This equilibrium hypothesis also permits
points corresponding to the original reactants formulation of the rate expression either in terms
and the final products, certain of these paths will of the characteristic partition functions of sta-
require significantly smaller gross energy inputs tistical mechanics or in terms of pseudo ther-
to surmount the energy barrier for the reaction. modynamic parameters.
The minimum energy requirement is related to The task of expressing the energy of a system
the activation energy for the reaction. The comprised of a number of nuclei and electrons
atomic configuration corresponding to potential as a function of their spatial configuration is
4.3 Molecular Theories of Chemical Reaction Kinetics 113
Activated
complex
Energy barrier
for forward reaction
Difference in ground
state energies
Reaction coordinate
Figure 4.2
Schematic representation of reaction progress in terms of the
reaction coordinate.
obviously a complex problem. In fact, it is one to depict the potential energy surface in three
that is not amenable to exact solution for even dimensions, as shown in Figure 4.3.
the simplest possible chemical reaction. How- Certain qualitative and semiquantitative as-
ever, theoretical physicists have attacked this pects of potential energy surfaces are conve-
problem with diligence during the past several niently illustrated using this figure. The vertical
decades, and a number of useful approximate elevation represents the potential energy of the
methods have been developed. In principle they system as a function of the two internuclear
permit one to generate multidimensional poten- distances rxy and ryz. Such surfaces may be gen-
tial energy surfaces linking reactants and erated using the techniques of the theoretical
products. physicist. They correspond to two valleys linked
Consider a reaction involving three atoms. by a saddle-shaped pass or col. One side of each
valley corresponds to a very steep hill, said
X + YZ -> XY + Z (4.3.23)
terrain arising from the strong dependence of the
In general three position variables will be needed repulsive component of interatomic forces on
to specify the potential energy of the reaction separation distance. The opposite sides of the
system. These may be the X-Y, Y-Z, and X-Z valley rise less steeply and eventually lead to a
internuclear distances or two internuclear dis- plateau at large values of both internuclear
tances and the included angle. Even in this separations. The valleys do not join except by
relatively simple case, four dimensions would be way of a pass or saddle point near K and L.
required for generation of the potential energy If a plane is passed perpendicular to the rxy
surface. However, if we restrict our attention to axis at a sufficiently large value of this parameter,
linear configurations of these atoms, it is possible the cross section of the potential energy surface
114 Molecular Interpretations of Kinetic Phenomena
'xy
© © ©
Figure 4.3
Schematic representation of the potential energy of a system comprised
of three atoms in a linear configuration as a function of the internuclear
separation distances. The dashed line represents the reaction X + YZ —*
XY+ Z.
one obtains is essentially the potential energy the potential energy curve characteristic of the
curve governing the formation of the YZ mole- formation of XY from its constituent atoms.
cule whose constituent atoms exert both attrac- Point M on the figure corresponds to a Z atom
tive and repulsive forces. At smaller values of and an X Y molecule. At point N both rxy and ryz
rxy, atom X interacts with atoms Y and Z to are large, so atoms X, 7, and Z are well sepa-
perturb the shape of the potential function so rated and no molecules are present. At points
that the depth of the potential well is less. Point K and L both internuclear distances are small,
J is located on the floor of one of the valleys and as in a collision between an X atom and a YZ
has a large value of rxy and a very small value for molecule (point K) or between a Z atom and an
ryz. It corresponds to an X atom far removed XY molecule (point L). The region in the neigh-
from a YZ molecule. borhood of these points is referred to as the
Similarly, it can be shown that a cross section transition state, and the configuration assumed
perpendicular to the ryz axis at large ryz gives by the atoms in this state is referred to as the
4.3 Molecular Theories of Chemical Reaction Kinetics 115
activated complex. We shall denote this complex figure, the system can obviously achieve reaction
by the symbol XYZ%. with significantly lower energy requirements. In
The reaction X + YZ -> XY + Z will corre- physical terms the process of forming the X-Y
spond to motion from point J in one valley to bond continuously contributes to the energy
point M in the second valley. For this reaction to requirement for breaking the original Y—Z
take place with a minimum amount of energy bond.
the system will travel along the floor of the first For reactions involving more than three at-
valley, over the col, and down into the second oms the number of "dimensions required to
valley. This path is indicated by a dashed line. depict the potential energy surface exceeds hu-
Energy considerations dictate that the majority man capacity for visualizing the surface. Thus
of the reaction systems will follow this path. The it may be more convenient to consider such
elevation of the saddle point above the floor of reactions as taking place between various moi-
the first valley is thus related to the activation eties that play the same role as the atoms X, Y
energy for the reaction. and Z in the discussion above.
Another common representation of the pro- The semiempirical nature of the methods used
gress of the system as it moves from reactants to to construct multidimensional potential energy
products is that shown in Figure 4.2. Here the surfaces makes the quantitative validity of the
potential energy of the system is plotted versus results questionable. Hence the present state of
distance along the dashed line shown in Figure the theoretical calculation of activation energies
4.3. This variable is referred to as the reaction is unsatisfactory.
coordinate. In both figures the energy differences The development of reaction potential energy
between the transition state and the reactant and surfaces was an essential precursor to the formu-
product states are related to the activation lation of the transition state theory. The first cal-
energies for the forward and reverse reactions, culation of reaction rates in terms of a specific
respectively. Energy differences between the re- potential energy surface is attributed to Pelzer
actant and product states are related to the and Wigner (55), who studied the reaction be-
normal thermodynamic energy functions char- tween atomic and molecular hydrogen. Eyring
acteristic of the reaction. For exothermic reac- and his co-workers (52-54) later developed and
tions the potential energy of the products lies extended the basic concepts, leading to codifica-
below that of the reactants, and for endothermic tion of the theory and thereby forging the close
reactions the relative elevations are reversed. link between his name and the theory. At about
Thus, Figure 4.2 indicates an endothermic re- the same time Evans and Polanyi (56-58) pre-
action. sented a somewhat similar formulation of the
It is important to recognize that the mecha- problem. These and other analyses of the prob-
nism indicated by the figures is quite different lem lead to very similar conclusions. The chem-
from one corresponding to an initial complete ical engineer should be familiar with these
dissociation of the YZ molecule followed by conclusions, even if he or she does not choose to
combination of the X and Y atoms. This latter become acquainted with the analytical details.
mechanism corresponds to movement from one In Eyring's formulation of the problem he
valley at J up to the plateau at N and back assumes that an "equilibrium" exists between
down to the second valley at M. The activation the activated complex species and the reactant
energy for the latter mechanism would be equal molecules. This "equilibrium" is said to exist at
to the energy required to dissociate the YZ all times, regardless of whether or not a true
molecule. By moving along the valley as in the chemical equilibrium has been established be-
mechanism indicated by the dashed line in the tween the reactants and products. Although the
116 Molecular Interpretations of Kinetic Phenomena
basic assumption is not unreasonable, neither complex, h is Planck's constant, and (XYZX) is
does it lend itself to simple verification. It does, the total concentration of activated complexes.
however, lead to results that are consistent with Combination of equations 4.3.24 to 4.3.26 gives
those obtained by procedures that do not require
one to assume equilibrium between reactants r = -(XYZX) (4.3.27)
and the activated complex. h
The reaction rate is then taken to be the Thus the effective frequency with which acti-
product of the frequency at which activated vated complexes are transformed into reaction
complexes cross the energy barrier and their con- products is kBT/h. At a temperature of 300 °K,
centration at the top of the barrier. If (Cx)' repre- this group has a value of 6 x 10 12 sec" 1 , which
sents the concentration lying within a region of is comparable in magnitude to normal molecular
length Sx at the top of the barrier (see Figure 4.2) vibration frequencies.
and if vx is the mean speed at which molecules If one describes the equilibrium between re-
move from left to right across the barrier, the actants and the activated complex by
rate is given by
(XYZ%)
x
{C )'v x Kx = (4.3.28)
(4.3.24) (X)(YZ)
r = 5X then
These activated complexes differ from ordi- (4.3.29)
r = h
nary molecules in that in addition to the three
normal translational degrees of freedom, they which is a second-order form.
have a fourth degree of translational freedom The problem of predicting a rate constant thus
corresponding to movement along the reaction reduces to one of evaluating Kx. There are two
coordinate. This degree of freedom replaces one basic approaches that are used in attacking this
vibrational degree of freedom that would other- problem: one is based on statistical mechanics
wise be observed. and the other on thermodynamics. From statis-
The motion of activated complexes within the tical mechanics it is known that for a reaction of
transition state may be analyzed in terms of the type X + YZ^XYZ1 the equilibrium
classical or quantum mechanics. In terms of constant is given by
classical physics, motion along the reaction
7~Eo/RT
coordinate may be analyzed in terms of a one- Kx = iXYZ (4.3.30)
dimensional velocity distribution function. In QxQ YZ
terms of quantum mechanics, motion along the where the g's are molecular partition functions
reaction coordinate within the limits of the tran- for the various species and Eo is the energy
sition state corresponds to the traditional quan- increase accompanying the reaction at absolute
tum mechanical problem involving a particle in zero when 1 mole of activated complex is formed.
a box. Combination of equations 4.3.29 and 4.3.30
Such analyses lead to the conclusions that indicates that the second-order rate constant is
11/2 given by
8* = % (4.3.25)
2nm \
and
This relation indicates that the rate constant
(Cx)f = (XYZx)j(2nmxkBT)1/2 (4.3.26) can be determined from a knowledge of the parti-
tion functions of the activated complex and the
where mx is the effective mass of the activated reactant species. For stable molecules or atoms
4.3 Molecular Theories of Chemical Reaction Kinetics 117
the partition functions can be calculated from An equilibrium constant is simply related to
experimental data that do not require kinetic a standard Gibbs free energy change, as indi-
measurements. However, they do require that cated by equation 2.4.7.
molecular constants such as the vibrational fre-
quencies and moments of inertia be evaluated AGX = -RT (4.3.33)
from spectroscopic data. Evaluation of the parti- Hence, the reaction rate constant can be
tion function for the activated complex {QXYZ%) written as
presents a more difficult problem, since the
required moments of inertia and vibrational
(4.3.34)
frequencies cannot be determined experimen-
tally. In principle theoretical calculations would
permit one to determine moments of inertia Since the factor (kBT/h) is independent of the
from the various internuclear distances and nature of the reaction, this approach to the
vibrational frequencies from the curvature of transition state theory argues that the free
the potential energy surface in directions normal energy of activation (AGJ) determines the re-
to the reaction coordinate. In practice one would action rate at a given temperature.
seldom (if ever) have available a sufficiently The free energy of activation can also be
accurate potential energy surface for the reac- expressed in terms of an entropy and an enthalpy
tion whose rate constant is to be determined. of activation in conventional thermodynamic
Moreover, QXYZ$ is a partition function from fashion.
which the degree of freedom corresponding to AG+ = AHX - TAS* (4.3.35)
motion along the reaction coordinate has been
removed. The contribution of this motion has These quantities provide another manner of
already been taken into account in the analysis expressing the rate constant in thermodynamic
leading to equation 4.3.27. Readers who wish to terms.
acquaint themselves with the manner in which
partition functions are calculated are encour-
aged to consult standard texts in physical chem-
If k is expressed in liters per mole per second,
istry and statistical mechanics.
the standard state for the free energy and entropy
Although equation 4.3.29 refers to a second-
of activation is 1 mole/liter. If the units of k
order reaction between an atom X and a mole-
are cubic centimeters per molecule per second,
cule YZ, the theory is readily generalized to
the corresponding standard state concentration
other reaction stoichiometries. An expression
is 1 molecule/cm3. The magnitudes of AG* and
characterizing the equilibrium between reac-
AS* reflect changes in the standard state, so it
tants and the activated complex is used to
is not useful to say that a particular reaction is
eliminate the latter from equation 4.3.27, and
characterized by specific numerical values of
the desired result is obtained.
these parameters unless the standard states
Instead of formulating the reaction rate ex-
associated with them are clearly identified.
pression in terms of molecular partition func-
These standard states are automatically deter-
tions, it is often convenient to employ an
mined by the units chosen to describe the re-
approach utilizing pseudo thermodynamic func-
actant concentrations in the phenomenological
tions. From equation 4.3.29, the second-order
rate expressions.
rate constant is given by
In terms of the collision theory a bimolecular
reaction rate is written as
k = ^ K ' (4.3.32)
r = P<ZA ,-E/RT (4.3.37)
118 Molecular Interpretations of Kinetic Phenomena
where Ps is the steric factor, ZAB is the collision degree of order (large entropy decrease) neces-
frequency, and E is the activation energy of the sary to obtain the proper molecular configura-
reaction. In terms of the thermodynamic formu- tion for complex formation.
lation of the transition state theory a bimolecular The thermodynamic formulation of the transi-
reaction rate is given by tion state theory is useful in considerations of
reactions in solution when one is examining a
r = eAStlRe-AHt/RT{A)(B) (4.3.38) particular class of reactions and wants to extra-
polate kinetic data obtained for one reactant
Comparison of these expressions indicates that system to a second system in which the same
the activation entropy is related to the steric function groups are thought to participate (see
factor for the reaction. One may interpret the Section 7.4). For further discussion of the predic-
steric factor in terms of the degree of order of tive applications of this approach and its limita-
molecular configurations required to bring tions, consult the books by Benson (59) and
about the reaction, and this viewpoint is gener- Laidler (60). Laidler's kinetics text (61) and the
ally regarded as more satisfactory from an in- classic by Glasstone, Laidler, and Eyring (54)
tellectual viewpoint than is that which regards contain additional useful background material.
Ps as an a posteriori correction factor neces- Although the collision and transition state
sary to obtain agreement between theory and theories represent two important methods of
experiment. attacking the theoretical calculation of reaction
Experimental values of AS* are readily cal- rates, they are not the only approaches available.
culated from measured values of reaction rate Alternative methods include theories based on
constants and activation energies. These experi- nonequilibrium statistical mechanics, stochastic
mental values provide significant insight into the theories, and Monte Carlo simulations of chemi-
nature of the transition state and the structure cal dynamics. Consult the texts by Johnson (62),
of the activated complex. Loosely bound com- Laidler (60), and Benson (59) and the review by
plexes have higher entropies than tightly bound Wayne (63) for a further introduction to the
ones. A positive entropy of activation implies theoretical aspects of reaction kinetics.
that the entropy of the complex is greater than The various theories can provide useful in-
that of the reactants. More often there is a de- sight into the way in which reactions occur, but
crease in entropy associated with formation of we must again emphasize that they must be
the activated complex. For bimolecular reactions regarded as inadequate substitutes for experi-
the complex is formed by association of two mental rate measurements. Experimental work
molecules, and there is a loss of three trans- to determine an accurate reaction rate expres-
lational and at least one rotational degree of sion is an essential prerequisite to the reactor
freedom. Hence AS* is usually negative and, in design process.
some cases, is not too different from the overall
entropy change conventionally associated with LITERATURE CITATIONS
the reaction. In these cases for reactions of the
type A + B -• AB, it indicates that the activated 1. Wayne, R. P., "The Detection and Estimation of Inter-
mediates" in Comprehensive Chemical Kinetics, Volume
complex closely resembles the product molecule I, The Practice of Kinetics, edited by C. H. Bamford and
in its structure. For a long time such reactions C. F. H. Tipper, Elsevier, New York, 1969.
were regarded as abnormal because they had 2. Gardiner, W. C , Jr., Rates and Mechanisms of Chemical
unusually low steric factors. However, the tran- Reactions, W. A. Benjamin, New York, 1969.
sition state theory clearly indicates that these 3. Daniels, F., and Johnston, E. H., J. Am. Chem. Soc, 43
low steric factors are merely a result of the large (53), 1921.
Literature Citations 119
4. Ogg, R. A., J. Chem. Phys., 15 (337), 1947; 18 (572), 22. Semenov, N. N., op. cit., Chapter I.
1950. 23. Laidler, K. J., Chemical Kinetics, Chapter 3, McGraw-
5. Edwards, J. O., Greene, E. F., and Ross, J., J. Chem. Hill, New York, 1965.
Education, 45 (381), 1968. Adapted with permission. 24. Bodenstein, M., and Lind, S. C , Z. Physik. Chem., 57
6. Liebhafsky, H. A., and Mohammed, A., J. Am. Chem. (168), 1907.
Soc, 55 (3977), 1933. 25. Christiansen, J. A., Kg I. Danske Videnskab. Selskab.,
7. King, E. L., in Catalysis, edited by P. H. Emmett, Mat.-Fys. Medd., 1 (14), 1919.
Volume 2, Part 2, pp. 337-456, Reinhold, New York, 26. Herzfeld, K. F., Ann. Physik, 59 (635), 1919.
1955.
27. Polanyi, M., Z. Elektrochem., 26 (50), 1920.
8. Gould, E. S., Mechanism and Structure in Organic
Chemistry, p. 136, Holt, Rinehart and Winston, New 28. Jost, W., and Jung, G., Z. Physik. Chem., B3 (83), 1929.
York, 1959. 29. Rice, F. O., and Herzfeld, K. F., J. Am. Chem. Soc, 56
9. Melville, H. W., and Gowenlock, B. G., Experimental (284), 1934.
Methods in Gas Reactions, Macmillan, London, 1963. 30. Goldfinger, P., Letort, M., and Niclause, M., Contribu-
10. Anderson, R. B., Experimental Methods in Catalytic tion a I'etude de la structure moleculaire, Victor Henri
Research, Academic} Press, New York, 1970. Commemorative Volume, p. 283, Desoer, Liege, 1948.
11. Friess, S. L., Lewis, E. S., and Weissberger, A., Editors, 31. Laidler, op. cit., Chapters 7 and 8.
Investigation of Rates and Mechanisms of Reactions, 32. Frost, A. A., and Pearson, R. G., Kinetics and Mecha-
Volume VIII, Parts I and II of Technique of Organic nism, Second Edition, Chapter 10, Wiley, New York,
Chemistry, Second Edition, Interscience, New York, 1961.
1961, 1963. 33. Benson, op. cit., Chapter XIII.
12. Lewis, E. S. (Editor), Investigation of Rates and Mecha- 34. Dainton, F. S., Chain Reactions, Methuen and Co.
nisms of Reactions, Third Edition, Volume VI, Part I of London, 1956.
Techniques of Chemistry, Wiley, New York, 1974.
35. Kennard, E. H., Kinetic Theory of Gases, McGraw-Hill,
13. Hammes, G. G. (Editor), Investigation of Rates and New York, 1938.
Mechanisms of Reactions, Third Edition, Volume VI,
36. Present, R. D., Kinetic Theory of Gases, McGraw-Hill,
Part II of Techniques of Chemistry, Wiley, New York,
New York, 1958.
1974.
37. Guggenheim, E. A., Elements of the Kinetic Theory of
14. Morris, J. C , / . Am. Chem. Soc, 63 (2535), 1941; 66
Gases, Pergamon Press, New York, 1960.
(584), 1944.
38. Golden, S., Elements of the Kinetic Theory of Gases,
15. Zemany, P. D., and Burton, M., J. Phys. and Colloid
Addison Wesley, Reading, Mass., 1964.
Chem., 55 (949), 1951.
39. Kauzmann, W., Kinetic Theory of Gases, W. A. Ben-
16. Wall, L. A., and Moore, W. J., J. Phys. and Colloid
jamin, New York, 1966. .
Chem., 55 (965), 1951; / . Am. Chem. Soc, 73 (2840),
1951. 40. Moelwyn-Hughes, E. A., Physical Chemistry, p. 1109,
Pergamon Press, New York, 1957.
17. Tolman, R. C , Phys. Rev., 23 (699), 1924; The Principles
of Statistical Mechanics, p. 163, Clarendon Press, 41. Tolman, R. C , cited in Frost and Pearson, op. cit.,
Oxford, 1938. pp. 68-69.
18. Denbigh, K. G., The Thermodynamics of the Steady 42. Lindemann, F. A., Trans. Faraday Soc, 77(598), 1922.
State, pp. 31-38, Methuen and Co., London, 1958. 43. Benson, op. pit., Chapter 11.
Adapted with permission. 44. Laidler, op. cit., pp. 143-175.
19. Hirschfelder, J., J. Chem. Phys., 9 (645), 1941. 45. Wayne, op. cit., pp. 256-289.
20. Semenov, N. N., Some Problems in Chemical Kinetics 46. Bunker, D. L., Theory of Elementary Gas Reaction
and Reactivity, Volume I, pp. 29 and 64, translated by • Rates, pp. 48-74, Pergamon Press, Oxford, 1960.
M. Boudart, Princeton University Press, Princeton,
N.J., 1958. 47. Johnston, H. S., Gas Phase Reaction Rate Theory, pp.
263-297, Ronald Press, New York, 1966.
21. Benson, S. W., Thermo chemical Kinetics, Methods for
the Estimation of Thermochemical Data-and Rate Param- 48. Kondratiev, V. N., Chemical Kinetics of Gas Reactions,
eters, Second Edition, Wiley, New York, 1976. pp. 2'83-321, Pergamon Press, Oxford, 1964.
120 Molecular Interpretations of Kinetic Phenomena
49. Gardiner, op. cit., pp. 111-135. (b) Since these investigators carried out their
50. Slater, N. B., Theory of Unimolecular Reactions, Cornell study in a stirred flow reactor, they were able
University Press, Ithaca, N. Y., 1959. to measure the reaction rate directly as a
51. Hinshelwood, C. N., The Kinetics of Chemical Change, function of ethyl nitrate concentration. They
p. 81, Clarendon Press, Oxford, 1940.
have reported the following data.
52. Eyring, H., J. Chem. Phys., 3 (107), 1935.
53. Wynne-Jones, W. F. K., and Eyring, H., / . Chem. Phys., Rate Ethyl nitrate concentration
3(492), 1935.
(moles/m3 • ksec) (moles/m3)
54. Glasstone, S., Laidler, K. J., and Eyring, H., The Theory
of Rate Processes, McGraw-Hill, New York, 1941. 13.4 0.0975
55. Pelzer, H., and Wigner, E., Z. Physik. Chem., B15 (445), 12.2 0.0759
1932. 12.1 0.0713
56. Evans, M. G., and Polanyi, M., Trans. Faraday Soc, 31 23.0 0.2714
(875), 1935. 20.9 0.2346
57. Evans, and Polanyi, op. cit., 33 (448), 1937.
58. Polanyi, M., / . Chem. Soc, 1937, 629. Is the proposed mechanism consistent with
59. Benson, S. W., Thermo chemical Kinetics, Wiley, New this data? What are the order of the reaction
York, 1968. and the rate constant that they observed
60. Laidler, K. J., Theories of Chemical Reaction Rates, experimentally?
McGraw-Hill, New York, 1969.
61. Laidler, op. cit., Chapter 3, 1965. 2. Under appropriate conditions the decom-
62. Johnston, H. S., Gas Phase Reaction Rate Theory, position of ethyl bromide may be explained by
Ronald Press, New York, 1966.
the following mechanism.
63. Wayne, R. P., in Comprehensive Chemical Kinetics,
Volume 2, The Theory of Kinetics, edited by C. H. Bam- Initiation:
ford and C. F. H. Tipper, Elsevier, Amsterdam, 1969.
C 2 H 5 Br ^ C 2 H 5 - + Br- £fl§1
PROBLEMS Propagation:
1. Houser and Lee [J. Phys. Chem., 71 (3422), Br- + C 2 H 5 Br ^ HBr + C 2 H 4 Br- Eati
1967] have studied the pyrolysis of ethyl nitrate
using a stirred flow reactor. They nave proposed C 2 H 4 Br- '-I C 2 H 4 + Br- £flf3
the following mechanism for the reaction.
Termination:
Initiation:
Br- + C 2 H 4 Br- ^ C 2 H 4 Br 2 EaA
C 2 H 5 ONO 2 ^ C 2 H 5 O + N O 2
(a) Derive an expression for the rate of disap-
Propagation:
pearance of ethyl bromide.
C 2 H 5 O ^ CH 3 - + CH 2 O (b) What is the apparent order of the reaction?
(c) What is the apparent activation energy for
CH 3 - + C 2 H 5 ONO 2 ^ CH 3 NO 2 + C 2 H 5 O
the reaction in terms of the activation en-
Termination: ergies of the individual steps?
Note that the amount of ethyl bromide that
2 C 2 H 5 O k-X CH 3 CHO + C 2 H 5 OH
disappears by the initiation reaction is small
(a) What rate expression is consistent with this compared to that which disappears via reac-
mechanism? tion 2.
Problems 121
3. Khan and Martell [J. Am. Chem. Soc, 91 What is the resultant expression for the overall
(4668), 17, 1969] have reported the results of a rate of consumption of oxygen in terms of the
kinetic study of the uranyl ion catalyzed oxida- individual rate constants and the concentrations
tion of ascorbic acid. The stoichiometric equa- of stable species? Experimentally, the reaction
tion for this reaction mav be represented as kinetics follow the expression:
H2A + O 2 ^ A + H2O2 d(O2)
= /c(CH 3 CHO) 3/2 (Co + + + )
The rate expression for this reaction is of the dt
form. Note that the overall stoichiometry requires
d(H2A) [d(O2)/dt] = [d(CH3CHO)/dt~]. What assump-
= £[H 2 A][H + ][O 2 ][UO 2 ++ ] tions are necessary to make the postulated
dt
mechanism agree with the observed kinetics?
Is the following mechanism consistent with the
experimentally observed rate expression?
+
H2A ^ H A ~ + H fast, equilibrium constant
+ +
HA" + UO^" ^± UO2HA fast, equilibrium constant K2
UO2HA+ + H+ ^UO 2 H 2 A + + fast, equilibrium constant K3
UO 2 H 2 A ++ + H + ^UO 2 H 3 A + + + fast, equilibrium constant KA
UO2H3A+ + + + O2 ^UO 2 H 3 A(O 2 ) + + + fast, equilibrium constant
UO2H3A(O2)+ + + ^ UO 2 H 3 A(O 2 ) + + + slow, rate constant k6
UO 2 H 3 A(O 2 ) + + + ^UO 2 + + + H 2 O 2 - 1- A + H + fast, rate constant k-,
4. Havel [Sb. Ved. Praci, Vysoka Chem. Tech- 5. The following mechanism has been proposed
nol, Pardubice 1 (83), 1965] has suggested the for the oxidation of nitric oxide.
following chain reaction mechanism for the
NO + O2 ^ NO 3 equilibrium constant Kc
Co + + + catalyzed oxidation of aldehyde to
peracetic acid. NO 3 + N O ^ N O 3 - N O
Initiation:
NO 3 + NO ^ 2NO2
NO + NO3-NO k-X 2NO2 + NO
Propagation:
k2
O2 + NO 3 NO k-X 2NO + 2O2
CH3CO-H-O2^CH3CO3- Derive a rate expression that is consistent with
CH 3 CO 3 +CH 3 CHO^CH 3 CO 3 this mechanism. Treacy and Daniels [J. Am.
Chem. Soc, 11 (2033), 1955] have determined
Termination: that the orders of the reaction with respect to
k4
2CH3CO3-^inactive products oxygen and nitric oxide are one and two, respec-
tively, at high pressures and less than one and
3- + CH3CO-^inactive products
greater than two at low pressures. Is the pro-
2CH3CO-4inactive products posed mechanism consistent with this data?
122 Molecular Interpretations of Kinetic Phenomena
6. It is thought that the fluorination of both cis- actor operating at 60 °C starting with an
and trans-perfluorobutene-2 may proceed by the initial tetraborane pressure of 14.13 kPa.
mechanism shown below.
Initiation: Tetraborane concentration
Time, t (sec) (moles/m3)
«1 o '
0 4.83
C 2 F 4 = C 2 F 4 -+• F 2 —> C 2 F 4 — C 2 F 4 + F* 1800 3.68
3600 3.38
Propagation: 5400 2.91
7200 2.46
F 9000 2.18
C 2 F 4 = C 2 F 4 + F-k4 C 2 F 4 - C 2 F 4
F What is the experimentally observed order of
k
C 2 F 4 - C 2 F 4 + F 2 Z C 2 F 5 - C 2 F 5 + F- the reaction? Under what circumstances is the
proposed mechanism consistent with the experi-
Termination: mentally observed rate expression? The order is
equal to n(l/2) where n = 0, 1, 2, 3, or 4.
F
phosgene (COC12) from carbon monoxide and what change occurs in the order of the
chlorine and the decomposition of phosgene reaction? What does the overall activation
into these species. energy become?
(d) The chain length is given by the ratio of the
Mechanism I: rate of chain propagation (reaction 3) to
Cl 2 ^ 2C1 equilibrium constant K1 that for termination (reaction 5). What is the
Cl + CO ^± COC1 equilibrium constant K2 chain length in this case?
12. Experimental studies of the catalytic de- approach. He proposed the following mech-
composition of ozone in the presence of nitrogen anism.
pentoxide follow a rate expression of the form
Ni(CO) 4 ^ Ni(CO) 3 + CO
r = /c(O 3 ) 2/3 (N 2 O 5 ) 2/3 k2i
an initial pressure of 39.46 kPa, determine the (b) The average number of HC1 molecules
collision frequency. If the activation energy for formed per entering chlorine atom.
the process is 243 kJ, what is the apparent (c) The probability that a collision between a
second-order reaction rate constant? The exper- chlorine atom and a hydrogen molecule
imental rate constant is 0.0404 (m3/mole-ksec). leads to reaction.
On the basis of this approach, what is the steric
factor? Comment. It may be assumed that each reaction of the
type
Cl + H 2 -> HC1 + H
17. Rodebush and Klingelhoefer [J. Am. Chem.
Soc, 55 (130), 1933] have studied the reaction is immediately followed by the much faster
of atomic chlorine with molecular hydrogen. reaction:
Chlorine atoms were formed by partial disso-
H + CU -* HC1 + Cl
ciation of molecular chlorine in an electrodeless
discharge. A stream of this gas was then mixed Data
with a hydrogen stream and passed through a
thermostatted reaction vessel. At the far end of <JH2 = 2.39 A a cl = 2.97 A
the vessel the reaction was effectively quenched
by using a piece of silver foil to catalyze the Hydrogen flow rate 6.3 cm3 (STP)/min
recombination of chlorine atoms. The products Chlorine flow rate (as Cl2) 9.1 cm3 (STP)/min
of the reaction were determined by freezing out Fraction Cl 2 dissociated 11°/
1 1
/o
Volume of reaction vessel
the Cl 2 and HC1 in liquid air traps and titrating
Pressure in vessel 10 cm3
samples with standard thiosulfate and alkali 0.340 mm Hg
Temperature in vessel
respectively. On the basis of the data and as- Length of run 0°C
sumptions listed below, determine: Thiosulfate titre of products 10 min
36.5cm3of0.2N
(a) The average number of collisions that a solution
Alkali titre of products
chlorine atom undergoes with hydrogen 9.1 cm3of0.1N
molecules in the reaction vessel. solution
5 Chemical Systems Involving
Multiple Reactions
5.0 INTRODUCTION
r - $ % - , - (M.1,
The chemical composition of many systems can
be expressed in terms of a single reaction pro-
In this section we discuss the mathematical
gress variable. However, a chemical engineer
forms of the integrated rate expression for a few
must often consider systems that cannot be
adequately described in terms of a single extent simple combinations of the component rate ex-
of reaction. This chapter is concerned with the pressions. The discussion is limited to reactions
development of the mathematical relationships that occur isothermally in constant density
that govern the behavior of such systems. It systems, because this simplifies the mathematics
treats reversible reactions, parallel reactions, and and permits one to focus on the basic principles
series reactions, first in terms of the mathemati- involved. We will again place a "V" to the right
cal relations that govern the behavior of such of certain equation numbers to emphasize that
systems and then in terms of the techniques that such equations are not general but are restricted
may be used to relate the kinetic parameters of to constant volume batch reactors. The use of
the system to the phenomena observed in the the extent per unit volume in a constant volume
laboratory. system (£*) will also serve to emphasize this
restriction. For constant volume systems,
127
128 Chemical Systems Involving Multiple Reactions
-(k,
t = --, (5.1.5)V
the analysis of kinetic data where Ao, Bo, Ko, and So refer to the initial
concentrations of the various species. As in the
KCA - CB
= ln\ previous section, one may separate variables
—C and integrate to obtain an expression for £* as
B0
a function of time. However, the resultant
CAe expression is of extremely limited utility for
= In purposes of analyzing kinetic data.
- cBO A somewhat more useful approach is to con-
sider the relations that should exist between the
extent of reaction per unit volume at time t and
In the analysis of kinetic data from reactions the extent at equilibrium. At equilibrium equa-
believed to be first-order in both directions, the tion 5.1.19 becomes
equation that is most suitable for use depends
on the pertinent equilibrium data available. - ^ - K(Ro
0
Equations 5.1.17 and 5.1.11 are perhaps the most
useful, but others may be more appropriate for
use in some cases. The integrated forms permit This equation may be rewritten in terms of the
one to determine the sum (k1 + fc-J, while equi- equilibrium constant by recognizing that
librium data permit one to determine the equilib-
rium constant K = ki/k^1. Such information K = ^ (5.1.22)V
is sufficient to determine fcx and k^v
0 = K(A0 - g)(B0 - ^) - (Ro + e ) ( 5 0 + ^
5.1.1.2 Opposing Second Order Reactions (5.1.23)V
A + B^R + 5. ^* is the root of the quadratic expression that
kr corresponds to physical reality (i.e., it does not
From a mathematical standpoint the various give rise to a reactant or product concentration
second-order reversible reactions are quite sim- that is negative). The equilibrium concentrations
ilar, so we will consider only the most general of all species may be evaluated using stoichio-
case—a mixed second-order reaction in which metric relations of the form
the initial system contains both reactant and
product species. Cie = (5.1.24)V
r = kfCACB - krCRCs (5.1.18) At any time the concentrations of the various
In a constant volume system the rate may be species are given by relations of the form
written in terms of the extent per unit volume as Q = Ci0 + v££* (5.1.25)Y
= {kfA0B0 - krR0S0)
dt ^
- [kf(A0 + Bo) + kr(R0 + S o )]£*
+ (kf - kr)£*2 (5.1.20)Y (5.1.27)V
130 Chemical Systems Involving Multiple Reactions
- £*
= dt
{(kf - kr)(Z* ~ a) - [kf(Ae + Be) + kr(Re +
Integration gives
~{(kf - kM* ~ 4*) - \kf{Ae + Be) + kr{Re
t = In
kf(Ae + Be) + kr(Re + Se) - £*){(kf - K)g + kf(Ae Se)} J
(5.1.32)V
If one again employs the relation between the equilibrium constant and the reaction rate constant
(5.1.22), equation 5.1.32 can be written as
5.7.1.3 Second-Order Reaction Opposed by First- 5.1.2.1 General Techniques for the Interpreta-
kf tion of Reaction Rate Data for Reversible
Order Reaction A + B ^ R. For a rate ex- Reactions. The determination of the mathe-
kr
pression of the form matical form of a reaction rate expression is
generally a two-step procedure. One first deter-
r = kfCACB - krCR (5.1.36)
mines the dependence of the rate on the con-
the procedures employed in Sections 5.1.1.1 and centrations of the various reactant and product
5.1.1.2 lead to the following relation for the time species at a fixed temperature and then evaluates
dependence of £*. the temperature dependence of the various rate
B) - En (5.1.37)V
- In
where
K(A0 + Bo) + 1 + y/K2(A0 - B0)2 + 2K(A0 + Bo + 2J?T
(5.1.38)V
2K
From thermodynamic data it is possible to constants appearing in the rate expression. For
evaluate all quantities pertaining to equilibrium reversible reactions, at least two rate constants
in equation 5.1.37, reducing it to the form will be involved and the Arrhenius relation
must be used to analyze data on the temperature
[kf(Ae + Be) + kr~\t dependence of each rate constant individually
in order to determine the influence of tempera-
ture changes on the overall reaction rate.
Approaches to the determination of the
where Ml and M 2 are numerical constants. This concentration-dependent terms in expressions
equation is of the same form as 5.1.34 and may for reversible reactions are often based on a
be handled in the manner outlined previously. simplification of the expression to limiting cases.
By starting with a mixture containing reactants
alone and terminating the study while the reac-
5.1.2 Determination of Reaction Rate
tion system is still very far from equilibrium, one
Expressions for Reversible Reactions
may use an initial rate study to determine the
The problem of determining the mathematical concentration dependence of the forward reac-
form of the rate expression for a chemical reac- tion. In similar fashion one may start with
tion is one that involves a combination of careful mixtures containing only the reaction products
experimental work and sound judgment in the and use the initial rates of the reverse reaction
analysis of the data obtained thereby. In many to determine the concentration dependence of
cases the analytical techniques discussed in this part of the rate expression. Additional sim-
Section 3.3 are directly applicable to studies of plifications in these initial rate studies may arise
reversible reactions. In other cases only minor from the use of stoichiometric ratios of reactants
modifications are necessary. and/or products. At other times the use of a vast
132 Chemical Systems Involving Multiple Reactions
Each of the reactant concentrations is given Expansion and the use of equation 5.1.30 gives
by Alt* _
A = B = A,o —- <•> = kfA2e - krR2e -
*' R
' < 5 - 8 *2 4.98 (B) (F)
K
kr AeBe (2.60) -4)(-&)- (2KAe + 4Re)
Hence Now
kf = 4.98/c, (Q
2KAe + 4Re = 2(4.98)(2.60) + 4(5.80)
Equations of the form of 5.1.26 may now be = 49.10 kmoles/m 3
written for each species.
Substitution of numerical values in equation F
B = A = Ae-(Z*-g) R = Re + 2(Z*-&) (D) gives
Substitution of these relations into equation A
gives 49.10/cr ) \2.90
(4.98 - 4)(g* - 2.90) - 49.10
In
L((4.98 - 4)(-2.90) - 49.10
134 Chemical Systems Involving Multiple Reactions
If we denote the deviation from equilibrium the equilibrium constant for the reaction one
conditions by can determine the ratio of these rate constants
(k1/k- J . Such information is adequate to deter-
A£* = f* - & (5.1.45)V
mine the individual rate constants k1 and k-v
then
For certain types of reversible reactions it is
= Ae- A£*, B = Be (5.1.46)Y necessary to use only slight perturbations from
equilibrium so that the differential equations
and resulting from the analysis are amenable to in-
tegration. The theoretical treatment in these
(5.1.47)V
dt dt cases varies slightly from that presented above
and indeed varies slightly with the orders of the
Combining equations 5.1.43, 5.1.46, and 5.1.47
forward and reverse reactions. Consider the
gives
reaction represented by the following second-
order mechanistic equations.
A + B ^ C +D (5.1.53)
= M.-fc-A-(fei+fc-i)A^* The reaction is assumed to occur in a constant
(5.1.48)V volume system. A relaxation analysis of the type
employed for the first-order reaction leads to
Combination of equations 5.1.42 and 5.1.48 the following analog of 5.1.49.
gives
-\_kl(Ae + Be) + fc_i(Cc + De)1 A£*
- ( / q + fc_i)A^* (5.1.49)V dt
(5.1.54)V
Integration subject to the condition that A£* =
(A£*)o at t = 0 gives If the displacement from equilibrium is small,
the term involving (A£*)2 is small compared to
= (fcx + /c_t)r (5.1.50)V the term involving A£*. This condition implies
that the terms (Ae + Be) and (Ce + De) are both
very much greater in magnitude than A£*. Under
The relaxation time (£*) for the chemical re-
these conditions equation 5.1.54 becomes
action is denned as the time corresponding to
Be)
In = 1 (5.1.51)V dt
(5.1.55)V
The relaxation time is thus the time at which which, after integration, becomes
the distance from equilibrium has been reduced
to the fraction l/e of its initial value. From €n Be)
equation 5.1.50, it is evident that
(5.1.56)V
1
t* = (5.1.52)V
From the basic definition of the relaxation
time it is evident that
Consequently, if one can experimentally deter-
mine the relaxation time of a system, the sum 1
t* = (5.1.57)V
of the rate constants (ki + /c_ x) is known. From Be) De)
136 Chemical Systems Involving Multiple Reactions
As before, the constants kx and k^1 can be Consider an arbitrary reaction, which we may
separated by making use of the fact that /ct//c_ x write in the form
is the equilibrium constant. Alternatively, the
rate constants can be separated by measuring a A + bB ^± rR + sS (5.1.58)
t* at various values of (Ae + Be) and (Ce + De). Suppose that under conditions where the
Chemical engineers should be aware of the concentrations of species JR and S are very small,
existence of relaxation techniques for studies of experimental evidence indicates that the initial
very fast reactions. However, since relaxation reaction rate is of the form
time measurements call for sophisticated ex-
perimental equipment and techniques, they are rf = k[Af[By[R]p[SY (5.1.59)
seldom made outside of basic research labora-
tories. where the various orders may be positive, nega-
tive, or zero. We would like to know what
thermodynamics has to say about permissible
5.1.3 Thermodynamic Consistency of forms of the concentration dependence of the
Rate Expressions reverse reaction. It is logical to expect that the
form of the reverse reaction rate will also be
For reversible reactions one normally assumes
a product of powers of concentrations, for
that the observed rate can be expressed as a
difference of two terms, one pertaining to the example,
forward reaction and the other to the reverse rr = k'[AY[BY[RY[SY (5.1.60)
reaction. Thermodynamics does not require that
the rate expression be restricted to two terms or Our problem is that of determining the allowable
that one associate individual terms with intrinsic reaction orders in the last equation.
rates for forward and reverse reactions. This The net rate of reaction is given by
section is devoted to a discussion of the limita-
tions that thermodynamics places on reaction
rate expressions. The analysis is based on the (5.1.61)
idea that at equilibrium the net rate of reaction Hence, at equilibrium, one requires that
becomes zero, a concept that dates back to the
historic studies of Guldberg and Waage (2) on
the law of mass action. We will consider only = lAeY
cases where the net rate expression consists of
two terms, one for the forward direction and one where the subscript es indicate any set of con-
for the reverse direction. Cases where the net centrations at which equilibrium exists. The rate
rate expression consists of a summation of constants k and k! and thus their ratio are in-
several terms are usually viewed as corres- dependent of the system composition and
ponding to reactions with two or more parallel depend only on temperature. In an ideal solution
paths linking reactants and products. One may thermodynamics indicates that the equilibrium
associate a pair of terms with each parallel path constant expressed in terms of concentrations
and use the technique outlined below to deter- (Kc) is also a function only of temperature. Con-
mine the thermodynamic restrictions on the sequently the ratiofc/fc'must be a function of Kc.
form of the concentration dependence within
each pair. This type of analysis is based on the (5.1.63)
principle of detailed balancing discussed in
Section 4.1.5.4. Now, for equation 5.1.58, the equilibrium con-
5.1 Reversible Reactions 137
and if (a' - a), (/?' - /?), (pf - p\ and (a' - a) Denbigh and Prince (4) have studied the kinetics
are each the same multiple of the corresponding of the absorption of gaseous N O 2 in aqueous
stoichiometric coefficient (3), nitric acid solutions. The following stoichio-
metric relation governs the process.
This type of rate expression provides a good On the basis of their experimental studies and
correlation of the experimental data for dilute a search of the literature, Denbigh and Prince
aqueous solutions. indicated that the correct choice of n is 1/2. They
If it is assumed that in more concentrated found that the data could be fitted by an ex-
solutions the rate of the forward reaction con- pression of the form
tinues to follow this rate expression, what forms
of the reverse rate are thermodynamically con- r = - C[N 2 O 4 ] 1/4 [NO] 1/2
sistent in concentrated acid solution? Equili- where the "constant" C depends on the acid
brium is to be established with respect to concentration. (In very concentrated solutions
equation A when written in the N 2 O 4 form. It one must worry about the extent of dissociation
may be assumed that the dependence on NO 2 of HNO 3 , s o some difficulties arise in the HNO 3
and N 2 O 4 concentrations may be lumped to- term.)
gether by equation C.
as well as cases where one reactant may not be 5.2.1.1 Simple Parallel Reactions. The sim-
common to both reactions (competitive parallel plest types of parallel reactions involve the
reactions). irreversible transformation of a single reactant
+B into two or more product species through
reaction paths that have the same dependence
(5.2.2)
on reactant concentrations. The introduction
of more than a single reactant species, of
In this section we will consider the kinetic reversibility, and of parallel paths that differ in
implications of both general classes of parallel their reaction orders can complicate the analysis
reactions. considerably. However, under certain condi-
tions, it is still possible to derive useful mathe-
5.2.1 Mathematical Characterization matical relations to characterize the behavior
of Parallel Reactions of these systems. A variety of interesting cases
are described in the following subsections.
When dealing with parallel reactions, it is
necessary to describe the time-dependent be- 5.2.1.1.1 Irreversible First-Order Parallel Reac-
havior of the system in terms of a number of tions. Consider the irreversible decomposition
reaction progress variables equal to the number of a reactant A into two sets of products by
of independent reactions involved. When one first-order reactions.
tries to integrate the rate expressions to deter- (5.2.3)
mine the time dependence of the system com- A
position, one finds that in many cases the algebra (5.2.4)
becomes unmanageable and that closed form
For the first reaction,
solutions cannot be obtained. On the other
hand, it is often possible to obtain simple rela-
tions between the concentrations of the various dt if (5.2.5)V
species, thereby permitting one to determine the
fractions of the original reactants transformed and, for the second,
by each of the reactions. The technique of
eliminating time as a dependent variable is an SI (5.2.6)V
dt
invaluable asset in the determination of these
relations. It enables one to determine relative where £? and £ | are the extents per unit volume
values of reaction rate constants while at the of reactions 1 and 2, respectively. These two
same time circumventing the necessity for ob- differential equations are coupled and must be
taining a complete solution to the differential solved simultaneously. In the present case,
equations that describe the reaction kinetics. We division of the first by the second gives
will use this technique repeatedly in the sections
that follow. We will start by describing the
dtJ
mathematical behavior of simple parallel re-
actions and then proceed to a discussion of
=dM = r (5-2.7)v
competitive parallel reactions. Throughout the
discussion we will restrict our analysis to con-
Integration gives
stant volume systems in order to simplify the
mathematics and thereby focus our attention on
the fundamental principles involved. (5.2.8)V
140 Chemical Systems Involving Multiple Reactions
Combination of equations 5.2.6 and 5.2.8 gives reaction is immeasurably fast, it is still possible
to determine relative values of the rate constants
by measuring the increments in species concen-
trations and using equation 5.2.15.
Separation of variables and integration leads 5.2.1.1.2 Reversible First-Order Parallel Reac-
to tions. This section extends the analysis devel-
k2 oped in the last section to the case where the
*] (5.2.10)V reactions are reversible. Consider the case
where the forward and reverse reactions are
Combination of 5.2.8 and 5.2.10 gives all first-order, as indicated by the following
mechanistic equations.
R (5.2.16)
(5.2.13)V
(5.2.19)V
= Ao
(5.2.14)V Similarly,
result is
(5.2.39)V
mx — m2
In order to evaluate A£f and hence 5(0 one need only substitute the relation for , into equation
5.2.31 and solve the differential equation. The result is:
(Se - S 0 ) * - (m 2 + fc2
S = Se +
m2 —
(5.2.40)V
The time dependence of the various species do not fit the categories discussed in the last
concentrations will depend on the relative mag- two subsections. Of particular interest are the
nitudes of the four rate constants. In some cases reactions between different chemical species to
the curves will involve a simple exponential give two or more different products (e.g. the
rise to an asymptote, as is the case for irrevers- formation of otho-, meta-, and para- derivatives
ible reactions. In other cases the possibility of of an aromatic compound. This section is de-
overshoot (exists, as indicated in Figure 5.2. voted to a discussion of the mathematical
Whether or not this phenomenon will occur relations that govern such reactions. Consider
depends on the relative magnitudes of the rate the following two stoichiometric equations as
constants and the initial conditions. However, representative of this class of reactions.
the fact that both roots of equation 5.2.34 must
be real requires that there be only one maximum
in the curve for R(t) or S(t). A + B -• R + •
dt
5.2.1. J.3 Higher-Order Irreversible, Simple Par-
allel Reactions. Many simple parallel reactions (5.2.41)V
5.2 Parallel Reactions 143
and
R - Ro
(5.2.45)V
s-s0
The fractional yield of product R is given by
R-Ro
(5.2.46)V
Ao- A
Similarly,
O —
(5.2.47)V
Ao - A + k2
Time, t
Thus the fractional yields of the different prod-
Figure 5.2 ucts are independent of time and of the orders
Concentration versus time curves for reversible of the reaction. If the increments in the products
parallel reactions indicating the possibility of a can be determined (in many cases Ro and So
maximum in the concentration of one product will be zero), then it is possible to evaluate the
species. ratio kl/k2. Moreover, for this case,
AA
r (5.2.48)V
A +B = k2{A)m2{Bf2 dt
dt
and the methods developed in Chapter 3 may
(5.2.42)V
be used to evaluate m1? nl and (k{ + fc2). The
The form of the solutions that one obtains for last result together with a knowledge of the
reactions of this type is dictated by the relative ratio k1/k2 permits one to determine individual
values of the four reaction orders. For some values of these parameters.
choices of these orders it is not possible to
obtain simple closed form solutions to these Case II. The Orders with Respect to Each of the
rate equations unless one places additional re- Constituents Are Different, but the Total Order
strictions on the initial composition of the (m + n) of Each Reaction Is the Same.
reaction mixture. The three possible- cases are In this case elimination of time as an indepen-
discussed below. dent variable gives
Case I. The Orders with Respect to Each of the
Reactants Are Equal (m2 = mx and n2 = nt).
In this case elimination of time as an indepen-
dent variable gives or
d£% _kx # j|tmi _
(5.2.43)V \Do si Qi)
k2
which leads to (5.2.50)V
differential equation. In this particular instance In this case separation of variables and inte-
gration leads to
(
]M = Y (5.2.51)V k,
1 + — Ao
and the equations involved in the solution
become identical with those of Case I. 1+ l -(« + «)]
In the more general case, the product dis-
F2
tribution is strongly influenced by the manner (5.2.55)V
in which the reaction is carried out—in partic- In order to evaluate the relative values of the
ular the manner of contacting and the type of reaction rate constants one need only plot
reactor used. (£* + £*) versus ^f and take the slope at the or-
igin. From equation 5.2.54 this slope is equal to
Case III. The Total Order is Not the Same for 1 + (kJk2)A0 when [m1 + nl - (m2 + n2)~\ = 1.
Both Reactions. From the slope one can determine k1/k2. This
In this case mx + n1 differs from m2 + n2 and ratio and the relation between £f and £f given
there are a variety of possible forms that the by equation 5.2.55 may be used with either of
rate expression may take. We will consider only the original rate expressions 5.2.41 and 5.2.42 to
some of the more interesting forms. In this case obtain individual values of kx and k2 for specified
elimination of time as an independent variable values of m1 + n1 and m2 + n2.
leads to the same general result as in the previous In this subsection we have treated a variety
case (equation 5.2.50). As before, in order to of higher-order simple parallel reactions. Only
obtain a closed form solution to this equation, by the proper choice of initial conditions is it
it is convenient to restrict our consideration to possible to obtain closed form solutions for
a system in which Ao = Bo. In this specific case some of the types of reaction rate expressions
equation 5.2.50 becomes one is likely to encounter in engineering prac-
tice. Consequently, in efforts to determine the
kinetic parameters characteristic of such sys-
h; = ir(Ao-£* ~ £*) (5.2.52)V
tems, one should carefully choose the experi-
mental conditions so as to ensure that potential
The relationship that must exist between £? and simplifications will actually occur. These simpli-
£f is most readily obtained by noting that fications may arise from the use of stoichiometric
ratios of reactants or from the degeneration of
(5.2.53)V reaction orders arising from the use of a vast
excess of one reactant. Such planning is particu-
or, using equation 5.2.52, larly important in the early stages of the re-
search when one has minimum knowledge of the
system under study.
a<;2
5.2.1.2 Competitive Parallel Reactions. Com-
(5.2.54)V petitive parallel reactions are those in which two
A closed form solution to this equation exists or more reactant species compete for yet another
for all values of \_(m1 + nx) — (ra2 + n 2 )]. How- reactant; for example,
ever, the resultant function will depend on this C
difference. One case that occurs often is that in (5.2.56)
which the difference is unity. D
5.2 Parallel Reactions 145
If the kinetic parameters for the upper reaction In the case where n1 and n2 are both unity,
are denoted by the subscript 1 and those for equation 5.2.59 becomes
the lower reaction by the subscript 2, the appro-
"si
priate rate expressions for constant volume (5.2.60)V
systems may be written as d£* k2 (B'o - fj)
Separation of variables and integration gives
dt
(5.2.61)V
(5.2.57)V or
and (5.2.62)V
= k2(A)m2(B')"2
dt Thus, from simultaneous measurements of the
unreacted fractions of the two competing species,
= k2(A0 - £* - one may readily determine the ratio of reaction
(5.2.58)V rate constants.
The general relationship between £? and £f In order to indicate the types of complications
may be determined by eliminating time as an involved in proceeding from this point to equa-
independent variable in the usual fashion. tions that indicate the time-dependent behavior
of the various species concentrations, it is in-
d£X k1 p^mi-m2 (^o ~ ^l)" 1 structive to consider the following rate expres-
x 2
dE>* k2 ° (Bf0 — £f)"2 sions.
(5.2.59)V dtF*
rectly. Reactions of this type are the only ones Substitution of this result in equation 5.2.63
that we will consider in more detail. gives
dt -»,{..-«-*[.-(^)~}d, (5.2.66)V
146 Chemical Systems Involving Multiple Reactions
In general this equation must be solved using stants that are sufficiently accurate for purposes
numerical methods. Once £? has been related to of reactor design.
time in such fashion, equation 5.2.65 may be The first point to be established in any experi-
used to evaluate £f as a function of time. Basic mental study is that one is dealing with parallel
stoichiometric principles may then be used to reactions and not with reactions between the
determine the corresponding concentrations of products and the original reactants or with one
the various product and reactant species. another. One then uses data on the product dis-
When m1 = m2 and n1 and n2 are both two, tribution to determine relative values of the rate
equation 5.2.59 can be integrated to give constants, employing the relations developed in
Section 5.2.1. For simple parallel reactions one
1 1 1 1 then uses either the differential or integral meth-
ods developed in Section 3.3 in analysis of the
(5.2.67)V data.
There are few short-cut methods for analyzing
or simple parallel systems. One useful technique,
( however, is to use stoichiometric ratios of reac-
(5.2.68)V
k2\B' tants so that the ratio of the time derivatives of
the extents of reaction simplifies where possible.
When m1 equals m2 and n1 and n2 are one For higher-order irreversible simple parallel
and two, respectively, integration of equation reactions represented by equations 5.2.41 and
5.2.59 gives 5.2.42, the degenerate form of the ratio of
reaction rates becomes
1 1
In (5.2.69)V h
1
w2) (5 2 70)V
k2
Equations 5.2.62, 5.2.68, and 5.2.69 indicate
that simultaneous measurements of the concen- For these conditions the product distribution
trations of the competitive species permit one is time independent when the overall reaction
to determine the relative values of the two rate orders are identical. However, when (rax + nx) >
constants using plots appropriate to the rate (m2 4- n2\ the product distribution changes as
expressions in question. time proceeds with the ratio of the product
formed by reaction 1 to that formed by reaction
2, declining as the reaction proceeds. Conversely,
5.2.2 Techniques for the Interpretation
if (m2 + n2) > {m1 + wj, this ratio will in-
of Kinetic Data in the Presence of
crease. Experiments of this type can provide
Parallel Reactions
extremely useful clues as to the form of the
In general an analysis of a system in which non- reaction rate expression and as to the type of
competitive parallel reactions are taking place experiments that should be performed next.
is considerably more difficult than analyses of Another technique that is often useful in studies
the type discussed in Chapter 3. In dealing with of these systems is the use of a large excess of one
parallel reactions one must deal with the prob- reactant so as to cause a degeneration of the
lems of determining reaction orders and rate concentration dependent term to a simpler form.
constants for each of the individual reactions. In the case of competitive parallel reactions
The chemical engineer must be careful both in one has the option of studying each reaction
planning the experiment and in analyzing the independently by varying the composition of
data so as to obtain values of the kinetic con- the initial reaction mixture. If a chemical species
5.2 Parallel Reactions 147
is absent, it obviously cannot react. What is value that permits one to convert the relative rate
more interesting from the viewpoint of the kine- constants into absolute values for each
ticist is the possibility of using reactions of this parameter.
type to gain information about complex reaction Illustrations 5.3 and 5.4 indicate how one
systems. If one reaction is well characterized utilizes the concepts developed'in this section
on the basis of previous work and if another in the determination of kinetic parameters for
reaction is very rapid and not amenable to competitive parallel reactions.
investigation by conventional techniques, com-
parative rate studies can be very useful. In such ILLUSTRATION 5.3 DETERMINATION OF
cases it is convenient to "starve" the system with RELATIVE RATE CONSTANTS FOR
respect to the species for which other reactants COMPETITIVE PARALLEL SECOND-
compete by providing less than a stoichiometric ORDER REACTIONS (FIRST-ORDER IN
amount of this species for any of the reactions EACH SPECIES)
involved. One lets the competitive parallel When a mixture of benzene (B) and benzyl
reactions go to completion and analyzes the chloride (C) are reacted with nitric acid (A) in
resultant product mixture. Since the equations acetic anhydride solution, the products are
resulting from the elimination of time as an in- nitrobenzene (NB) and nitrobenzyl chloride
dependent variable are applicable to the final (NC).
product mixture, compositions of such mixtures
may be used to determine relative values of the A + B -> NB + H 2 O (A)
rate constants involved.
A + C + H2O r 2 = k2(A)(C) (B)
Relative values of the rate constants are useful
in themselves; by measuring such values for the The data below are typical of this reaction at
reactions of a series of compounds with the same 25 °C when both B and C have initial concen-
reactant, one is able to determine the rank order trations of 1 kmole/m 3 (5). The data were
of reactivity within the series. Such determina- recorded after very long times so all the nitric
tions are useful in the development of correla- acid present initially was consumed. What is the
tions of the effects of substituent groups on the value of k2/ki<?
Nitrobenzene Nitrobenzyl
Initial concentration chloride concentration
HNO 3 after complete after complete
Run (kmoles/m3) reaction (kmoles/m3) reaction (kmoles/m3)
CH—CH CH=CH
X X CH 2 CH 2
CH 2 CH 2
CH 2 —CH
CH=CH2 CH=CH 2
CH—CH CH=CH
X X CH2 CH2
CH 2 CH 2
CH2=CH2 CH 2 —CH 2
i
CHO CHO
(A + B - • D) r2 = k2CACB (B)
5.2 Parallel Reactions 149
0 649
The product distribution thus changes as the c t- -
reaction proceeds.
^AO ^AOy
Rearrangement gives
(F)
2k,
Equation E relates ft and ft, so ft may be
d(CA0 - ft) k7
evaluated at various values of ft using the value
If we let x = CA0 — ft and r = k1/k2. of r determined above.
150 Chemical Systems Involving Multiple Reactions
v2 ( 0 . 1 2 3 ) - 1
(0.123)
1 - 1 - 1 -
CA0 2(0.123) - 1 cAO
-0.754
= 0.163 ( 1 - 1 - —
cAO. c,10.
Using the values of ^yCA0 determined from this product species will be present during the normal
relation, it is possible to integrate equation F course of the reaction. The general scheme for
numerically. The result is these reactions may be represented as
k2CAOt = 2.37 [A . • •] _+ [ B • • •] ^ [ C • • •] (5.3.1)
The initial reactant concentration may be where the quantities in brackets may denote
determined from the ideal gas law. more than one molecular species. Among the
many general types of reactions that fall into
this category are those leading to mono-, di-,
RT (0.082)(603) tri-, etc. substituted products, sequential partial
= 1.01 x 10" 2 moles/liter oxidation reactions, multiple cracking reactions,
and polymerization reactions.
Thus,
This section discusses the kinetic implications
2.37 of series reactions. We will be concerned only
^2 -
(1.01 x 10"2)(40) with those cases where the progress of the various
stages of the overall transformation is not in-
= 5.86 liters/mole-minute fluenced by either parallel or reverse reactions.
Consequently, The discussion will again be limited to constant
volume systems.
fci = (jA k2 = (0.123)(5.86)
5.3.1 Mathematical Characterization of
= 0.72 liters/(mole-minute) Series Reactions
These values of the reaction rate constants 5.3.1.1 Consecutive First-Order Reactions. The
differ from those cited by Pannetier and Souchay simplest case of series reactions is that in which
(6) because these individuals erroneously treated every reaction in the sequence obeys first-order
the two reactions as if they were of the simple kinetics. It may be represented in terms of the
parallel type instead of as if there were a com- following sequence of mechanistic equations.
petition between the acrolein and butadiene k.2 kj,
molecules for other butadiene molecules. A B -> C -> D (5.3.2)
The classic example of "reactions" of this type
is a sequence of radioactive decay processes that
5.3 SERIES OR CONSECUTIVE result in nuclear transformations. The differential
REACTIONS—IRREVERSIBLE equations that govern kinetic systems of this
The term "series reactions" refers to those type are most readily solved by working in terms
reactions in which one or more of the products of concentration derivatives. For the first
formed initially undergoes a subsequent reac- reaction,
A A
tion to give yet another product. Significant
— = -M (5.3.3)V
amounts of both the intermediate and the final
5.3 Series or Consecutive Reactions—Irreversible 151
•2 0.6 -
0.2
2 3 4 5 6 7 8
Dimensionless time, k^t
Figure 5.3
Time dependence of species concentrations for the consecutive reactions
A S B ^> C ^> D for k2 = 0.5 Jr, and k3 = 0.9 k,.
time and setting the derivative equal to zero. One may clearly extend the technique to
The result is include as many reactions as desired. The irre-
versibility of the reactions permits one to solve
the rate expressions one at a time in recursive
fashion. If the first reaction alone is other than
(5.3.7)V first-order, one may still proceed to solve the
system of equations in this fashion once the
The conditions under which the maximum can initial equation has been solved to determine
exist may be determined by examination of the A(t). However, if any reaction other than the
initial slope. When it is positive, a maximum will first is not first-order, one must generally resort
occur; if it is negative, there will be no maximum. to numerical methods to obtain a solution.
From equation 5.3.5, When one is dealing with a finite series of
dB reactions, it is possible to use stoichiometric
— klA0 k2B0 (5.3.8)V principles to determine the concentration of the
It
final species. For example, if only four species
Hence there will be a maximum when kxA0 > (A, B, C, D) are involved in the sequence of
k2B0. When k2B0 > fc^o, the curve for B starts equation 5.3.2, then
out with a negative slope and decreases mono-
tonically to zero at infinite time, as shown in
Figure 5.4. (5.3.1O)V
In order to arrive at an equation for the time
dependence of the concentration of species C, The concentration of species D a t a particular
one may proceed in a similar fashion. The result time may then be determined using this relation
is and the equations for species A, B, and C derived
above.
C p
k2e~k2t k2e~k*
C = k3 - K2 (5.3.9)V
y-k3t
- kj (k3 - k2)(k3 -
Obviously the curve depicting the time- 5.3.1.2 Consecutive Reactions that are other than
dependent behavior of the concentration of First-Order. For consecutive reactions that are
species C can take on even more forms than that not first-order, closed form analytical solutions
for species B. The shape of the curve is dependent do not generally exist. This situation is a con-
on the initial concentrations of the various sequence of the nonlinearity of the set of dif-
species and the three reaction rate constants. ferential equations involving the time derivatives
Figure 5.3 depicts the time-dependent behavior of the various species concentrations. A few two-
for the specific case where only species A is member sequences have been analyzed. Un-
present initially. fortunately, the few cases that have been
5.3 Series or Consecutive Reactions—Irreversible 153
analyzed are seldom encountered in industrial ble for the destruction of the intermediate. The
practice. For the most part one must resort to process involves taking differences in the slopes
approximate methods or to initial conditions of two concentration versus time curves (or a
that cause a degeneration of the reaction rate simple combination thereof). Since these dif-
expression to a more simple form (e.g., using a ferences may be very imprecise, this method is
stoichiometric ratio of reactants or a large often inappropriate for use.
excess of one species). In some cases one is able to start with the
One general technique that is often useful in stable intermediate and determine the rate at
efforts to analyze the behavior of these systems which it disappears using conventional methods.
is the elimination of time as an independent For first-order and pseudo first-order reac-
variable. tions of the series type several methods exist for
determining ratios of rate constants. We will
consider a quick estimation technique and then
5.3.2 Techniques for the Interpretation
describe a more accurate method for handling
of Kinetic Data in the Presence of
systems whose kinetics are represented by
Series Reactions
equation 5.3.2.
The first point that must be established in an The first approach is based on equation 5.3.7
experimental study is that one is indeed dealing and makes use of the fact that the time at which
with a series combination of reactions instead of the maximum concentration of the intermediate
with some other complex reaction scheme. One B is reached is a function only of the two rate
technique that is particularly useful in efforts of constants and initial concentrations. For the
this type is the introduction of a species that is case where no B is present initially, equations
thought to be a stable intermediate in the reac- 5.3.7 and 5.3.6 can be written as
tion sequence. Subsequent changes in the dy-
namic behavior of the reaction system (or lack
thereof) can provide useful information about
the character of the reactions involved. and
If one monitors the rate of disappearance of
the original reactant species the general differ- Bw
(5.3.12)V
ential and integral approaches outlined in Ao K - 1
Section 3.3 may be used to determine the rate
expression for the initial reaction in the sequence. where
Once this expression is known one of several
other methods for determining either absolute (5.3.13)
or relative values of the rate constants for sub-
sequent reactions may be used. Combination of these relations and further
If the data are sufficiently accurate, one may algebraic manipulation gives
use a general differential approach in which the
expression for the net rate of formation of a K/il K)
stable intermediate is postulated and tested = K - (5.3.14)V
against experimental data. The difference be-
tween the rate of formation by the initial reaction This equation indicates that the maximum
for which the kinetics are known and the actual becomes more pronounced as k2/k1 becomes
net rate of formation is tested against the pro- smaller. If k1 is known and the maximum con-
posed rate expression for the reaction responsi- centration of intermediate is measured, k2 may
154 Chemical Systems Involving Multiple Reactions
Table 5.1
Series First-Order Reactions. Time-Percentage Reaction Relations for Various
Relative Rate Constants
Adapted from Kinetics and Mechanism 2nd edition, by A. A. Frost and R. G. Pearson.
Copyright ©1961. Reprinted by permission of John Wiley and Sons, Inc.
5.4 Complex Reactions 155
CN CONH ? COOK
fast
R >R
KOH
R + NH,
\ H2O s
CN CN
CN
COOK COOK
COOK
k2 fast
R R R NH,
H2O KOH
CONH7 COOK
The kinetics observed are typical of two first- cussed previously. There are a great number of
order reactions in series. possible combinations of parallel, series, and
The following data represent the progress of reversible reactions that we have not considered,
the reaction as reported in Swain's (7) analysis but that may occur in nature in chemical pro-
of Kaufler's work. cessing operations. Benson (10) has described
this situation in an apt fashion: "Kinetic systems
Percent reaction (5O<5*) Time, t (hr) when investigated in detail display an anarchistic
tendency to become unique laws unto them-
15 0.367 selves." In this section we turn our attention to
35 1.067
these more complex systems.
70 4.200
It is extremely difficult to generalize with
regard to systems of complex reactions. Often it
Use the time ratio method to determine is useful to attempt to simplify the kinetics by
kjk2. using experimental techniques which cause
a degeneration of the reaction order by using a
Solution large excess of one or more reactants or using
From the data indicated it is possible to arrive stoichiometric ratios of reactants. In many cases,
at three time ratios and their logarithms. We however, even these techniques will not effect a
will evaluate K using each of these ratios and simplification in the reaction kinetics. Then one
average the results. From the ratios and Table must often be content with qualitative or semi-
5.1: quantitative descriptions of the system behavior.
log(f7oA15) = 1-0586 KX = 0.194 This text has not treated all the mathematical
descriptions of reacting systems that have ap-
log(r7oA35) = 0.5951 Ku = 0.193
peared in the literature. Indeed, such coverage
Ais) = 0-4635 KIU = 0.196 goes far beyond the scope and spirit of this text.
The average value of fe2/fe1 is thus 0.194. For material of this type, consult the kinetics
literature.
156 Chemical Systems Involving Multiple Reactions
5.4.2 Competitive Consecutive There are two other limiting forms of these
Second-Order Reactions equations that are also of interest. If kx » fc2, the
first step is very rapid compared to the second,
Competitive consecutive reactions are com-
so that it is essentially complete before the latter
binations of parallel and series reactions that
starts. The reaction may then be treated as a
include processes such as multiple halogenation
simple irreversible second-order reaction with
and nitration reactions. For example, when a
the second step being rate limiting. On the other
nitrating mixture of HNO 3 and H 2 SO 4 acts on
hand, if k2 » kl9 the first step controls the
an aromatic compound like benzene, NO 2
reaction so the kinetics observed are those for a
groups substitute for hydrogen atoms in the
single second-order process. However, the anal-
ring to form mono-, di-, and tri-substituted nitro
ysis must take into account the fact that in this
compounds.
case 2 moles of species A will react for each mole
Since many reactions of this type involve a
of B that is consumed.
series of second-order processes, it is instructive
For the more general case of arbitrary rate
to consider how one analyzes systems of this
constants, the analysis is more complex. Various
sort in order to determine the kinetic parameters
approximate techniques that are applicable to
that are necessary for reactor design purposes.
the analysis of reactions 5.4.1 and 5.4.2 have
We will follow a procedure described previously
been described in the literature, and Frost and
by Frost and Pearson (11). Consider the follow-
Pearson's text (11) treats some of these. One
ing mechanistic equations.
useful general approach to this problem is that
of Frost and Schwemer (12-13). It may be re-
(5.4.1)
garded as an extension of the time-ratio method
discussed in Section 5.3.2. The analysis is pre-
C + A^D + Z' (5.4.2)
dicated on a specific choice of initial reactant
These reactions may be regarded as parallel concentrations. One uses equivalent amounts of
with respect to species A and series with respect reactants A and B (Ao = 2B0) instead of equi-
to species B, C, and D. Only species A and B are molal quantities.
present at time zero. A material balance on species A indicates that
The equations for the time derivatives of
Ao - A = C + ID (5.4.6)V
certain species concentrations are given by
while a balance on species B gives
dA
~k XBA - k2CA (5.4.3)V
dt Bo = B + C + D (5.4.7)V
With this substitution equation 5.4.3 becomes In this case equation 5.4.17 becomes
dA fiK
= (2/c2 - kx k2A2 B R l - 2K)$ I Ac
dt dx* Bo L 2(1 - K) ' 2(1 - K)J Bo
Elimination of time as an independent vari- (5.4.19)V
able between equations 5.4.4 and 5.4.11 gives
Recognition that Ao/Bo = 2 leads to
dfi
Frost and Schwemer have developed a time- kv This value and the value of K then may be
ratio technique based on equations 5.4.21 and combined to evaluate k2. The following illustra-
5.4.16 in order to facilitate the calculation of tion indicates the application of this technique.
second-order rate constants for the class of re-
actions under consideration. Data for A/Ao ver-
sus T* at various values of K are presented in ILLUSTRATION 5.6 DETERMINATION OF
Table 5.2, and time ratios are given in Table 5.3. REACTION RATE CONSTANTS FOR
The latter values may be used to determine K COMPETITIVE CONSECUTIVE SECOND-
by using various time ratios from a single kinetic ORDER REACTIONS
run if one recognizes that (if/rf) = (tjt2). Once
K has been determined, Table 5.2 may be used to Burkus and Eckert (14) have studied the kinetics
determine the T* values at a given A/Ao and K. of the triethylamine catalyzed reaction of 2,6-
Equation 5.4.18 may then be used to determine tolylene diisocyanate with 1-butanol in toluene
Table 5.2
i* as a Function of K and AIA0 (11)
A/Ao = 0.8 A/Ao = 0.7 A/Ao = 0.6 A/Ao = 0.5 A/Ao = 0.4
20% 30% 40% JU/ O 60%
reaction reaction reaction reaction reaction
Table 5.3
Time Ratios as a Function of K (11) (f60//20 is the ratio of the time for
60% A reacting to the time for 20% reaction, or where A/Ao = 0.4
and 0.8, respectively)
CH 3
r ??
+ C4H9OH -> O C N _ ^ Y _ N — C — O C 4 H 9
5 + A -• C (A)
CH 3 H O
I OH CH3 H O
:-OC4H9 fc2 || | • ' "
+ C 4 H 9 OH -> C 4 H 9 O — C ^ N N—C—OC 4 H 9
C + A (B)
0.360 20 3 282
0.657 30 9 117
= - « = 6.16
1.116 40
1.866 50
3.282 60 = 2-840 1/K = 6.09
tso/ho =
160 Chemical Systems Involving Multiple Reactions
LITERATURE CITATIONS
PROBLEMS
1. Hellin, M., and Jungers, J. C , Bull. Soc. Chim., 1957
(386). 1. One of the classic examples of a reversible
2. Guldberg, C. M., and Waage, P., Etudes sur les affinites reaction that is first-order in both directions is
chimiques, Br0gger and Christie, Christiania, Oslo, the conversion of y-hydroxybutyric acid into its
1867. lactone in aqueous solution.
3. Denbigh, K. G., The Principles of Chemical Equilibrium, In aqueous solution the water concentration
Second Edition, p. 445, Cambridge University Press, may be considered constant, so the reverse re-
1966.
action follows pseudo first-order kinetics. The
4. Denbigh, K. G., and Prince, A. J., J. Chem. Soc, 1947 data below on this reaction have been taken
(790).
from Emanuel and Knorre. Use them to deter-
5. Emanuel', N. M., and Knorre, D. G., Chemical Kinet- mine the values of both first-order rate constants.
Problems 161
Give two possible rate laws for the reverse (Note. The absorbance could be measured only
reaction. to the nearest 0.5 unit.)
5. In an unpublished study, Grieger and Hansel 6. Ziman et al. [Kinetika i Kataliz, 9 (117), 1968]
have used absorbance measurements to monitor have studied the kinetics of the catalytic oxida-
the progress of the reaction forming an enzyme- tive dehydrogenation of various butene isomers
substrate complex of imidazole (S) and metmyo- to form 1,3-butadiene. Over a Bi-Mo catalyst
globin (E) at high pressures. Formation of the the following reactions are important.
enzyme-substrate complex (ES) may be repre-
sented by the following equation 1-butene ^ trans-2-butene
k2
E + S ^± ES 1-butene ^ ds-2-butene
k4
where the forward reaction is first-order in
each component (second-order overall) and the trans-2-butene ^± ds-2-butene
ke
reverse reaction is first-order. Since the equilib-
rium is rapidly established, a relaxation ap- 1-butene ^± 1,3-butadiene + H 2
proach was used to investigate the reaction
rrans-2-butene ^± 1,3-butadiene + H 2
kinetics. The data below are typical of the re- kg
action at 103.42 MPa and 19.5 °C. What is the ds-2-butene ^ 1,3-butadiene + H 2
relaxation time characteristic of these condi-
At 396 °C these authors have reported the fol-
tions? What additional experiments must be
lowing values of the reaction rate constants in
performed in order to determine the individual
units of inverse seconds.
reaction rate constants?
N(a) + H(a) ^ NH(a) (3) The allo-ocimene may undergo subsequent re-
k action to form a- and /?-pyronene and a dimer
of aZZo-ocimene.
NH(a) + H(a) ^ NH 2 (a) (4) Fuguitt and Hawkins have reported the data
k-4
below as typical of these reactions. The reported
data have been modified somewhat for the pur-
NH 2 (a) + H(a) ^ NH 2 (5) poses of this problem. By summing the amounts
of dimer, a- and /?-pyronene, and aZZo-ocimene
The symbol (a) denotes an adsorbed species.
present in the final mixture, one may determine
If all steps are at equilibrium and if the second
the total amount of aZZo-ocimene formed in the
step is believed to be rate controlling, what rela-
initial reaction. These investigators have postu-
tion must exist between the overall equilibrium
lated that all three of the parallel reactions are
constant and the observed rate constants? The
first-order. Is the data consistent with this hy-
rate of the forward reaction is to be taken as
pothesis? If so, what are the values of the three
/c2CH2 where fc2 is the rate constant observed for
first-order rate constants?
the forward reaction. Start by determining the
appropriate form of the rate constant observed
Data
for the reverse reaction in terms of the kt values
used above. Reaction of a-Pinene at 204.5 °C
Reprinted with permission from J. Am. Chem. Soc, 69 (319), 1947. Copyright by the American Chemical
Society.
8. When optically active a-pinene is heated 9. Schmid and Heinola [J. Am. Chem. Soc,
above 200 °C in the liquid phase, it becomes 90 (131), 1968] have studied the liquid phase
optically inactive. Initially, the reaction pro- reactions of 1-phenylpropyne (A) and 2,4-di-
cesses may be written as nitrobenzenesulfenyl chloride (B) in chloroform
solution at 51° C. The primary products of the
d-pinene -• dZ-limonene reaction are: l-phenyl-£raHS-l-chloro-2-(2, 4 di-
k2
nitrophenylthio) propene (C) and 1-phenyl-
d-pinene -> aZZoocimene tr<ms-2-chloro-l-(2, 4-dinitrophenylthio) pro-
pene (D). The reaction-stoichiometry may be
d-pinene -» dl pinene
164 Chemical Systems Involving Multiple Reactions
k2
B - > £>
11. In carrying out an enzyme assay it may be
convenient to introduce an auxiliary enzyme to
Using equimolal concentrations of reactants A the system to effect the removal of a product
and B, the authors obtained the following data. produced by the first enzymatic reaction.
McClure [Biochemistry, 8 (2782), 1969] has
described the kinetics of certain of these coupled
Time, t (ksec) A (kmoles/m3) enzyme assays. The simplest coupled enzyme
assay system may be represented as
0.0 0.181 k2
16.2 0.141 B
21.6 primary enzyme auxiliary enzyme c
0.131
32.4 0.119 If the first reaction is regarded as zero-order
37.8 0.111 irreversible (i.e., the enzyme is saturated with
86.4 0.0683 substrate), and the second reaction is first-order
99.0 0.0644 in the product JB, determine the time-dependent
108.0 0.0603
behavior of the concentration of species B if no
B is present initially. How long does it take to
reach 98% of the steady-state value if kx =
It may be assumed that the rate expressions for 0.833 mole/m3-ksec and k2 = 0.767 sec" 1 ?
reactions 1 and 2 are identical except for the What is this steady-state value?
values of the rate constants. If the reaction order
is an integer, determine the overall order of the
reaction and the rate constants k1 and k2. 12. Debande and Huybrechts [Int. J. Chem.
Kinetics, 6 (545), 1974] have studied the gas
phase Diels-Alder additions of propylene (P)
10. Regna and Caldwell [J. Am. Chem. Soc, 66 to cyclohexa-1, 3-diene (Chd) to give the exo-
(246), 1944] have studied the kinetics of the acid- and endo-isomers of 5-methylbicyclo (2,2,2)
catalyzed transformation of 2-ketopolyhydroxy oct-2-ene (MBO). The initial reaction rates were
acids into ascorbic acids and other products.
both found to be of the mixed second-order
k2 type with
K
^endo and r
exo — ^ e
where K is a 2-ketohydroxy acid, A is the corre-
sponding ascorbic acid analog, F is the furfural
formed by the decomposition of the ascorbic and
acid, and P is the polymerized furfural. At log/c e n d o = - 5 6 9 8 / T + 5.74
59.9 °C, the reaction of 2-keto-i?-gluconic acid log/cexo = -6577'/T + 6.66
follows pseudo first-order kinetics with ki =
2.53 10 - 3 mm - 1 and k, = 4.91 x 10" where T is expressed in degrees Kelvin and k in
1 cubic meters per kilomole per second.
min . If one starts with a solution that is 0.05
molar in both 2-keto-JS?-gluconic acid and its The following biradical mechanism has been
ascorbic acid derivative, what is the maximum proposed for this reaction.
Problems 165
(endo-MBO)
6 Q
(exo-MBO)
(a) Does this mechanism give rise to a rate where both reactions are second-order. From
expression that is consistent with the initial the data below determine the two rate constants
rate data? if the reaction of 127.3 mg of p-chlorophenyl-
(b) If the reverse Diels-Alder reactions are faster silane and 198.2 mg of benzyl alcohol in the
than the isomerization reaction between the presence of metallic copper liberates hydrogen
exo- and endo-forms, what does the rate gas as follows.
expression reduce to?
(c) If the endo-form is the desired product, do
you recommend operation at the high or Time, t (sec) 62 116 198 396 878
low end of the temperature range 512 to cm 3 H 2 8.12 12.18 16.24 20.30 24.36
638 °K? Determine the product distribution
for 30% conversion of cyclohexadiene at
The initial silane concentration is equivalent to
your specified temperature.
0.5 kmole/m 3 .
13. The reaction of p-chlorophenylsilane with
benzyl alcohol may be characterized as follows. 14. Eldib and Albright [Ind. Eng. Chem., 49
k (825), 1957] have indicated that the main reac-
RSiH 3 + R'OH -X RSiH 2 OR' + H 2
tions in the catalytic hydrogenation of cotton-
RSiH 2 OR / + R'OH ^ RSiH(OR) 2 + H 2 seed oil are
If each of these reactions is regarded as ir- kx = 0.0133 min" 1 /c4 = 0.0024 m m " 1
reversible pseudo first-order, derive equations ^ = 00108 min-i ^ = 00080 min-i
for the time dependence of each species in terms . _1
3 min
of the initial concentrations and the appropriate ~~ '
rate constants. If O ne were to start with pure linoleic acid,
For run 3, the following parameter values sketch the time dependent behavior of all species
were estimated. in terms of normalized concentrations {CJCA0).
6 Elements of
Heterogeneous Catalysis
6.0 INTRODUCTION a source of scientific theory, many of the con-
cepts that we would now interpret as catalytic
Since the chemical, petrochemical, and petro- in nature played a key role in scientific "theory."
leum industries rely heavily on catalytic pro- The "Philosopher's Stone" was a hypothetical
cessing operations, chemical engineers must be substance that could turn base metals into gold
cognizant of the fundamental and applied as- and could promote good health and long life.
pects of catalysis. In 1964 it was estimated that The chemical action of the "Philosopher's
18% of the total wholesale value of manufactured Stone" required that a small quantity of this
goods produced in the United States directly substance could bring about a large change, an
involved catalytic technology (1). Some com- idea inherent in the modern view of catalysis.
mercially significant catalytic processes are listed The aura of mysticism that is associated with
in Table 6.1. Chapter 6 discusses the chemical catalytic phenomena reaches back to the days
aspects of heterogeneous catalytic phenomena. of the alchemist. Only in recent years has
Chapter 12 is devoted to the engineering aspects catalysis been regarded more as a science than
of catalytic phenomena which are involved in an art. Indeed, much art still remains, and our
the design of commercial scale reactors. imperfect understanding of catalysts has long
When philosophical and metaphysical ideas been reflected in the manufacturing processes
enjoyed precedence over experimental facts as employed to produce commercial catalysts.
Table 6.1
Typical Industrial Catalytic Reactions (2, 3)
In this list, the first five processes are among the most important commercial catalytic processes in terms of total
processing capacity.
167
168 Elements of Heterogeneous Catalysis
In 1835 Berzelius coined the term catalysis process. The ratio of metal to oxygen in an oxide
to describe the influence of certain substances on catalyst will frequently change with changes in
the nature of diverse reactions, the substances temperature and in the composition of the fluids
themselves apparently being unchanged by the with which it is in contact. Pure metal catalysts
reaction. He imbued these materials with a will often change in crystal structure or surface
catalytic force capable of awakening the poten- roughness on use. Many commercial catalysts
tial for chemical reaction between species that are gradually deactivated by poisoning reactions
would normally be nonreactive at a given that accompany the main reaction process. The
temperature. In more modern terms the fol- catalysts used to initiate free radical polymeriza-
lowing definition is appropriate. tion reactions end up as an integral part of the
polymer formed by the reaction and are thus
A catalyst is a substance that affects the rate or consumed. In neither case does a stoichiometric
the direction of a chemical reaction, but is not whole number relationship exist between the
appreciably consumed in the process. amount of the catalyst consumed and the
amounts of reactants converted to products.
There are three important aspects of the The number of moles of reactants converted is
definition. First, a catalyst may increase or orders of magnitude greater than the number of
decrease the reaction rate. Second, a catalyst moles of catalyst consumed.
may influence the direction or selectivity of a For reversible reactions the principle of
reaction. Third, the amount of catalyst con- microscopic reversibility (Section 4.1.5.4) indi-
sumed by the reaction is negligible compared cates that a material that accelerates the forward
to the consumption of reactants. reaction will also catalyze the reverse reaction.
Catalysts that decrease reaction rates are In several cases where the catalytic reaction has
usually referred to as inhibitors. They usually been studied from both sides of the equilibrium
act by interfering with the free radical processes position, the observed rate expressions are con-
involved in chain reactions, and the mechanism sistent with this statement.
differs from that involved in accelerating a Catalytic reactions are often classified as
reaction. The most familiar example of the use homogeneous or heterogeneous. True homo-
of inhibitors is the addition of tetraethyl lead to geneous catalysis takes place when the catalyst
gasoline to improve its antiknock properties. and the reactants are both present in the same
A catalyst cannot change the ultimate equili- fluid phase. Acid and base catalyzed reactions
brium point set by thermodynamics, but it can in aqueous solution are reactions of this type.
affect the rate at which this point is approached. Reactions that occur between a gas and a liquid
However, it can facilitate approach to equili- or between two liquid phases are also generally
brium with respect to a desired reaction while considered to fall in this category. Although the
not influencing the rates of other less desirable mass transfer characteristics in these cases differ
reactions. In optimizing yields of desired prod- from those where the reaction takes place in a
ucts, chemical engineers are very concerned with single phase, the reactants are mobile in both
the selectivity or specificity of a catalyst. For phases. The characteristics of these reactions
commercial applications, selectivity is often are much more similar to those of reactions
more important than activity per se. occurring in a single fluid phase than they are
The idea that a catalyst remains unaltered by to those of reactions occurring at a fluid-solid
the reaction it catalyzes is naive and misleading. interface. The term heterogeneous catalysis is
Physical and chemical changes can and do occur generally restricted to catalytic phenomena in-
either during or as a result of the catalytic volving a solid catalyst and reactants in a gas
6.1 Adsorption Phenomena 169
or liquid phase. These phenomena are some- involving permanent dipole, induced dipole,
times referred to as contact catalysis. and quadrupole interactions. It involves van der
Catalysts may be employed in solid, gaseous, Waals or secondary valence forces. It is akin to
or liquid form. However, the overwhelming condensation. Chemisorption, on the other
majority of industrial catalytic processes involve hand, involves a chemical interaction with
solid catalysts and gaseous reactants. In many attendant transfer of electrons between the
cases it is possible to develop useful new catalysts adsorbent and the adsorbing species (adsorbate).
on the basis of analogies drawn between the The adsorbed species are held to the surface by
behavior of certain chemical systems in the valence forces that are the same as those that
presence of heterogeneous or homogeneous hold atoms together in a molecule.
catalysts. For example, reactions that take place In practice one can usually distinguish be-
in the presence of strong' acids in liquid solution tween chemisorption and physical adsorption
are often catalyzed by acidic solid catalysts such on the basis of experimental evidence. However,
as silica-alumina. In some cases inorganic salts it is frequently necessary to consider several
or metal complexes that have catalytic activity criteria in making the classification.
in aqueous solution will exhibit activity for the The most important criterion for differen-
same reaction when deposited on a solid sup- tiating between physical and chemical adsorp-
port. For example, an aqueous solution of cupric tion is the magnitude of the enthalpy change
chloride and palladium chloride can be used as accompanying the adsorption process. The
a catalyst for oxidizing ethylene to acetaldehyde energy evolved when physical adsorption occurs
in the Wacker process (2). These same materials is usually similar to the heat of liquefaction of the
can be deposited on a solid support to obtain a gas (i.e., approximately 2 to 6 kcal/mole). Like
heterogeneous catalyst for this reaction, al- condensation, the process is exothermic. For
though this is not done in the current commer- physical adsorption the heat of adsorption often
cial process. lies between the heat of vaporization and the
heat of sublimation. The enthalpy change
accompanying chemisorption is significantly
6.1 ADSORPTION PHENOMENA
greater than that for physical adsorption and,
This section treats chemical and physical ad- in some cases, may exceed 100 kcal/mole. More
sorption phenomena in order to provide neces- often it will lie between 10 and 50 kcal/mole.
sary background material for the discussion of In practice chemisorption is regarded as an
catalytic reaction mechanisms in Section 6.3.1. exothermic process and, in the past, the view
Adsorption is the preferential concentration of was frequently expressed that endothermic ad-
a species at the interface between two phases. sorption cannot occur. However, if a molecule
Adsorption on solid surfaces is a very complex dissociates on chemisorption and the dissocia-
process and one that is not well understood. The tion energy of the molecule is greater than the
surfaces of most heterogeneous catalysts are not energy of formation of the bonds with the
uniform. Variations in energy, crystal structure, surface, then the process can be endothermic.
and chemical composition will occur as one Since the Gibbs free energy change accom-
moves about on the catalyst surface. In spite of panying any spontaneous process must be
this it is generally possible to divide all adsorp- negative, the only cases in which endothermic
tion phenomena involving solid surfaces into adsorption can be observed are those for which
two main classes: physical adsorption and there will be a large positive entropy change.
chemical adsorption (or chemisorption). Physi- The dissociative chemisorption of hydrogen on
cal adsorption arises from intermolecular forces glass is endothermic and is an example of this
170 Elements of Heterogeneous Catalysis
relatively rare phenomenon. More often dis- room temperature is often used as a quick test
sociative chemisorption is exothermic, like other for determining the nature of the adsorption
chemisorption processes. process.
A second criterion that is frequently used to A third empirical criterion is based on the
differentiate empirically between chemical and effect of temperature on the amount adsorbed.
physical adsorption is the rate at which the For physical adsorption the amount of gas ad-
process occurs and, in particular, the tempera- sorbed always decreases monotonically as the
ture dependence of the rate. For physical ad- temperature is increased. Significant amounts of
sorption of a gas on a solid surface, equilibrium physical adsorption should not occur at tem-
is usually attained very quickly and is readily peratures in excess of the normal boiling point
reversible. The activation energy of the process at the operating pressure. Appreciable chem-
is usually less than a kilocalorie. Physical ad- isorption can occur at temperatures above the
sorption usually takes place so fast that the boiling point and even above the critical tem-
observed rate is limited by the rate at which perature of the material. Because chemisorption
molecules can be transported to the surface in- can be an activated process that takes place at
stead of by the adsorption process itself. Chem- a slow rate, it may be difficult to determine the
isorption may occur at rates comparable to amount of chemisorption corresponding to true
those of physical adsorption, or it may occur at equilibrium. Moreover, the process may not be
much slower rates, depending on the tempera- reversible. It is also possible for two or more
tures involved. Most types of chemisorption are types of chemisorption or for chemical and
characterized by a finite activation energy and physical adsorption to occur simultaneously on
thus proceed at appreciable rates only above the same surface. These facts make it difficult to
certain minimum temperatures. This type of generalize with regard to the effect of tempera-
phenomenon is referred to as activated chem- ture on the amount adsorbed. Different behavior
isorption. However, in some systems, chemi- will be observed for different adsorbent-
sorption occurs very rapidly even at very low adsorbate systems.
temperatures and has an activation energy near The experimental procedure for measuring the
zero. For example, chemisorption of hydrogen extent of adsorption generally involves admit-
and oxygen occurs on many clean metal surfaces ting a gas to a chamber that contains the sample
at liquid nitrogen temperatures. These cases are and waiting for a fixed period of time, during
described as nonactivated chemisorption. It is which "equilibrium" is supposedly achieved.
often observed that for a given adsorbent and Figure 6.1 indicates one common type of beha-
adsorbate, the initial chemisorption is non- vior that can be observed if physical adsorption
activated, while subsequent chemisorption is and one type of activated chemisorption are
activated. present. At low temperatures, physical adsorp-
Another aspect of rate measurements that is tion processes dominate and the amount of
useful in discriminating between the two types chemisorption is insignificant. The amount of
of adsorption involves studies of the rate of material physically adsorbed decreases as the
desorption. The activation energy for desorption temperature increases. Within the time scale of
from a physically adsorbed state is seldom more the normal adsorption experiment, the amount
than a few kilocalories per mole, whereas that of material adsorbed via the chemisorption
for desorption from a chemisorbed state is process will increase as the temperature in-
usually in excess of 20 kcal/mole. Consequently, creases in the low temperature range. Conse-
the ease with which desorption occurs on quently its contribution to the total will become
warming from liquid nitrogen temperature to more and more significant as the temperature
6.1 Adsorption Phenomena 171
general criteria discussed above, taken collec- engineer, who is interested in the design of
tively, provide a suitable basis for discrimina- heterogeneous catalytic reactors: chemisorption
tion. Table 6.2 summarizes these criteria. because it is an essential precursor of the reac-
One of the basic tenets of virtually all mecha- tion, and physical adsorption because it provides
nistic approaches to heterogeneous catalysis is a means of characterizing heterogeneous cata-
that chemisorption of one or more reactant lysts in terms of specific surface areas and the
species is an essential step in the reaction. Be- distribution of pore sizes within a porous
cause of the large changes in activation energies catalyst.
that are observed when shifting from a homo-
geneous reaction to one that is heterogeneously 6.2 ADSORPTION ISOTHERMS
catalyzed, energetic considerations indicate that Plots of an amount of material adsorbed versus
chemisorption must be involved. The forces pressure at a fixed temperature are known as
involved in physical adsorption are so small adsorption isotherms. They are generally classi-
relative to those involved in chemical bonding fied in the five main categories described by
that it is hard to imagine that they can distort Brunauer and his co-workers (4). In Figure 6.2
the electron clouds around a molecule to such adsorbate partial pressures (P) are normalized
an extent that they have an appreciable effect by dividing by the saturation pressure at the
on its reactivity. Many heterogeneous catalytic temperature in question (Po). Type I is referred
reactions take place at temperatures above the to as Langmuir-type adsorption and is charac-
critical temperatures of the reactants involved terized by a monotonic approach to a limiting
and, presumably, physical adsorption cannot amount of adsorption, which presumably cor-
occur under these conditions. Chemisorption responds to formation of a monolayer. This
has, however, been observed in many such cases. type of behavior is that expected for chemisorp-
This section has treated the qualitative fea- tion.
tures of chemisorption and physical adsorption No other isotherms imply that one can reach
processes. Both are important to the chemical a saturation limit corresponding to completion
Table 6.2
Comparison of Chemisorption and Physical Adsorption
are referred to as adsorption sites. Each site a monolayer can cause an extremely dis-
can accommodate only a single adsorbed proportionate reduction in catalytic activity.
species. 2. Heats of chemisorption usually decrease
2. The energy of an adsorbed species is the markedly as adsorption proceeds and the
same anywhere on the surface and is in- surface becomes covered.
dependent of the presence or absence of
In using the Langmuir adsorption isotherm
nearby adsorbed molecules. This assumption
as a basis for correlating heterogeneous catalytic
implies that the forces between adjacent ad-
rate data, the fact that the heat of chemisorption
sorbed molecules are so small as to be
varies with extent of adsorption is less significant
negligible and that the probability of adsorp-
than would appear at first glance. Molecules
tion onto an empty site is independent of
adsorbed on high-energy sites may be bound so
whether or not an adjacent site is occupied.
strongly that they are unable to serve as inter-
This assumption usually implies that the
mediates in the catalytic reaction. On the low-
surface is completely uniform in an energetic
energy sites the energy change on adsorption
sense. If one prefers to use the concept of a
may be insufficient to open up the catalytic
nonuniform surface with a limited number of
path for the reaction. It is possible that only
active centers that are the only points at
those sites of intermediate energy are effective
which chemisorption occurs, this is permis-
in catalysis.
sible if it is assumed that all these active
centers have the same activity for adsorption 6.2.1.1 Derivation of the Langmuir Equation
and that the rest of the surface has none. Adsorption of a Single Species. The kinetic ap-
3. The maximum amount of adsorption that is proach to deriving a mathematical expression
possible is that which corresponds to a for the Langmuir isotherm assumes that the rate
monolayer. of adsorption on the surface is proportional to
4. Adsorption is localized and occurs by colli- the product of the partial pressure of the ad-
sion of gas phase molecules with vacant sites. sorbate in the gas phase and the fraction of the
5. The desorption rate depends only on the surface that is bare. (Adsorption may occur only
amount of material on the surface. when a gas phase molecule strikes an uncovered
site.) If the fraction of the surface covered by an
For physical adsorption processes the third adsorbed gas A is denoted by 6A, the fraction
assumption is the poorest of these assump- that is bare will be 1 — 9A if no other species
tions. For the case of chemical adsorption the are adsorbed. If the partial pressure of A in the
worst of these assumptions is the second. There gas phase is PA, the rate of adsorption is given by
is a significant amount of experimental evidence
' adsorption A(\ VA) (6.2.1)
that is contradictory to this assumption. Taylor
(7) was the first to emphasize that adsorption where k may be regarded as a "pseudo rate
sites may vary in energy. He noted that atoms constant" for the adsorption process.
at peaks on the surface and along crystal edges The rate of desorption depends only on the
will be in high-energy states and will be the number of molecules that are adsorbed. Thus
points at which adsorption first occurs. Other
evidence for the lack of surface uniformity in- Tdesorption (6.2.2)
cludes experimental data indicating that: where k' may be regarded as a pseudo rate
constant for the desorption process.
1. Adsorption of a catalyst poison to an extent At equilibrium the rates of adsorption and
that represents only a very small fraction of desorption are equal.
6.2 Adsorption Isotherms 175
If these equations are solved for the fractions KAPA + KBPB KCPC
occupied by each species, (6.2.17)
Solving for ^6h where we have assumed that only one species
is adsorbed.
The desorption process must involve a reac-
(6.2.14)
+ tion between two adsorbed atoms to regenerate
6.2 Adsorption Isotherms 177
the gas phase molecules. Consequently, it may mett, and Teller (8). It takes multilayer adsorp-
be regarded as a second-order reaction between tion into account and is the basis of standard
two surface species: methods for determining specific surface areas
of heterogeneous catalysts. The extended form
v = k'AG2A (6.2.19)
' desorption ^* ^ of the BET equation (4) can be used to derive
At equilibrium all five types of isotherms as special cases.
The BET approach is essentially an extension
- 0 A ) 2 = k'A92A (6.2.20) of the Langmuir approach. Van der Waals
forces are regarded as the dominant forces, and
or
the adsorption of all layers is regarded as
(6.2.21) physical, not chemical. One sets the rates of
i-eA
adsorption and desorption equal to one another,
Solving for 9A, as in the Langmuir case; in addition, one re-
quires that the rates of adsorption and desorp-
(6.2.22) tion be identical for each and every molecular
1 + layer. That is, the rate of condensation on the
At low pressures where the surface is sparsely bare surface is equal to the rate of evaporation
covered, the fraction of the sites occupied by of molecules in the first layer. The rate of evapo-
fragments of species A will be proportional to ration from the second layer is equal to the rate
the square root of the gas phase partial pressure. of condensation on top of the first layer, etc.
If several species can adsorb but only species One then sums over the layers to determine the
A dissociates, it is easily shown, using the type total amount of adsorbed material. The deriva-
of analysis employed in Section 6.2.1.2, that tion also assumes that the heat of adsorption of
each layer other than the first is equal to the
heat of condensation of the bulk adsorbate ma-
KCPC terial (i.e., van der Waals forces of the adsorbent
are transmitted only to the first layer). If it is
(6.2.23)
assumed that a very large or effectively infinite
and that
number of layers can be adsorbed, the following
result is arrived at after a number of relatively
KBPB + KCPC elementary mathematical operations
(6.2.24)
v = (6.2.25)
Inasmuch as the Langmuir equation does not where c is a constant exponentially related to
allow for nonuniform surfaces, interactions the heats of adsorption of the first layer and the
between neighboring adsorbed species, or mul- heat of liquefaction, and P o is the saturation
tilayer adsorption, a variety of theoretical ap- pressure.
proaches that attempt to take one or more of This equation can be rearranged to give a
these factors into account have been pursued somewhat more familiar form.
by different investigators. The best-known al-
1
ternative is the BET isotherm, which derives its + x (6.2.26)
v(l - x) vmc
name from the initials of the three individuals
responsible for its formulation, Brunauer, Em- where x is the normalized pressure (P/Po)-
178 Elements of Heterogeneous Catalysis
If one plots the left side of this equation versus the steps that occur on a molecular scale in the
x, a straight line should result, with following manner.
1 1. Mass transfer of reactants from the main
(6.2.27) body of the fluid to the gross exterior surface
slope + intercept
of the catalyst particle.
Equation 6.2.9 may then be used in determining 2. Molecular diffusion and/or Knudsen flow of
the specific surface area of an adsorbent. Alter- reactants from the exterior surface of the
native arrangements of equation 6.2.25 are catalyst particle into the interior pore
possible, but they have the inherent disadvan- structure.
tage of placing undue emphasis on the low- 3. Chemisorption of at least one of the reac-
pressure data points that are most susceptible tants on the catalyst surface.
to error. 4. Reaction on the surface. (This may involve
If one restricts the number of layers of ad- several steps.)
sorbate that may be stacked up, as would be 5. Desorption of (chemically) adsorbed species
the case in the very narrow capillaries of a from the surface of the catalyst.
porous catalyst, the BET analysis must be 6. Transfer of products from the interior cata-
modified to allow for this. If n is the number of lyst pores to the gross external surface of the
permissible layers, it can be shown that the catalyst by ordinary molecular diffusion
adsorption isotherm becomes: and/or Knudsen diffusion.
vmcx[\ - (n 7. Mass transfer of products from the exterior
==
v - (c - - cxn+1] surface of the particle into the bulk of the
fluid.
(6.2.28)
Several of these steps are shown in schematic
When n = 1, this equation reduces to the
fashion in Figure 6.3. Of course, if the catalyst
Langmuir form. Since x is always less than
is nonporous, steps 2 and 6 are absent. Steps 1,
unity, it does not take very large values of n to
2, 6, and 7 are obviously physical processes,
approach the limiting form indicated by equa-
while steps 3 to 5 are basically chemical in
tion 6.2.25. By appropriate choice of the param-
character. The rates of the various steps depend
eters in equation 6.2.28, it is possible to generate
on a number of factors in addition to the con-
the five types of curves shown in Figure 6.2.
centration profiles of the reactant and product
Many individuals have developed more
species.
elegant theoretical treatments of adsorption
Steps 1 and 7 are highly dependent on the
processes since Brunauer, Emmett, and Teller
fluid flow characteristics of the system. The
published their classic paper. Nonetheless,
mass velocity of the fluid stream, the particle
the BET and Langmuir isotherms are the
size, and the diffusional characteristics of the
most significant ones for chemical engineering
various molecular species are the pertinent
applications.
parameters on which the rates of these steps
depend. These steps limit the observed rate only
6.3 REACTION RATE EXPRESSIONS FOR
when the catalytic reaction is very rapid and
HETEROGENEOUS CATALYTIC REACTIONS
the mass transfer is slow. Anything that tends
When a heterogeneous catalytic reaction oc- to increase mass transfer coefficients will en-
curs, several physical and chemical processes hance the rates of these processes. Since the
must take place in proper sequence. Hougen rates of these steps are only slightly influenced
and Watson (9) and others have broken down by temperature, the influence of these processes
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 179
Fluid
boundary
layer
Fluid /
boundary /
l a y e r /
Figure 6.3
Schematic representation of heterogeneous catalytic reaction on a porous
catalyst.
on the overall conversion rate will vary as the in which one measures the conversion achieved
temperature changes. Their influence is often as a function of the ratio of the weight of cata-
negligible at low temperatures, but may be quite lyst (W) to the molal feed flow rate (F). One
significant at higher temperatures. then repeats the experiments at a different linear
There are two commonly used techniques for velocity through the reactor (e.g., by changing
determining the range of process variables where the length-to-diameter ratio of the catalyst bed
these steps influence conversion rates. The first in a tubular reactor). The data are then plotted
approach involves carrying out a series of runs as shown in Figures 6.4a and 6.4b. If the two
180 Elements of Heterogeneous Catalysis
curves do not coincide as shown in Figure 6Ab, with caution; Chambers and Boudart (10) have
mass transfer resistances are significant, at least shown that it often lacks sensitivity under con-
in the low-velocity series. If the curves coincide ditions commonly employed in laboratory scale
as shown in Figure 6.4a, mass transfer effects studies. Heat and mass transfer coefficients are
probably do not influence the observed con- not strongly dependent on fluid velocity for
version rate. This diagnostic test must be applied Reynolds numbers below 50. (The characteristic
length dimension is taken as the pellet diameter.)
Consequently, at low Reynolds numbers, the
conversion will not change significantly with
fluid velocity at constant W/F, even though
heat or mass transfer limitations may be present.
The scatter in the experimental data will be
Curves for both series coincide
sufficient to mask these effects for many reac-
W/F tions. Thus, when the experimental diagnostic
test indicates that mass transfer limitations are
probably not present, one's conclusions should
be verified by calculations of the sensitivity of
the test to variations in fluid velocity. The
Low velocity
series methods developed in Sections 12.4 and 12.5
may be used for this purpose.
W/F •
(b)
The second approach involves simultaneous
variation of the weight of catalyst and the molal
flow rate so as to maintain W/F constant. One
then plots the conversion achieved versus linear
velocity, as shown in Figures 6.4c and 6Ad. If
the results are as indicated in Figure 6.4d, mass
Linear velocity at constant W/F transfer limitations exist in the low-velocity re-
(c) gime. If the conversion is independent of ve-
locity, there probably are no mass transfer
limitations on the conversion rate. However,
this test is also subject to the sensitivity limita-
tions noted above.
Linear velocity at constant W/F
When fluid velocities are high relative to the
(d)
solid, mass transfer is rapid. However, in stag-
nant regions or in batch reactors where no
Figure 6.4 provision is made for agitation, one may en-
Tests for external mass transfer limitations on counter cases where mass transfer limits the
conversion rates, (a) External mass transfer prob- observed reaction rate. We should also note
ably does not limit conversion rate, (b) Mass that in industrial practice pressure drop con-
transfer limitations are present, (c) External mass
straints may make it impractical to employ the
transfer probably does not limit conversion rate.
exceedingly high velocities necessary to over-
(d) Mass transfer limitations are present at low
velocities. Adapted from Chemical Engineers' come the mass transfer resistance associated
Handbook, Fourth Edition, edited by R. H. Perry, with highly active catalysts.
C. H. Chilton, and S. D. Kirkpatrick. Copyright c For porous catalysts the external gross surface
1969. Used with permission of McGraw-Hill Book area of a particle often represents only an in-
Company.) significant fraction of the total surface area that
I a strong rate of c
182 Elements of Heterogeneous Catalysis
6.3.1 Rate Expressions for obtained. The latter two individuals were re-
Heterogeneous Catalytic Reactions Limited sponsible for analyzing a number of possible
by the Rates of Chemical Processes reaction mechanisms in these terms (11, 12) and
for popularizing this approach to analyzing
In the treatment of rate expressions for het-
catalytic reaction rate data. For simplicity the
erogeneous catalytic reactions the definition of
term "Hougen-Watson models" will be used
local reaction rates in terms of interfacial areas
throughout the remainder of this text.
(3.0.10) is appropriate.
In addition to the assumptions implicit in
the use of the Langmuir isotherm the following
r (6.3.1)
~ sdt assumption is applicable to all Hougen-Watson
where S is the surface area of the solid catalyst. models: the reaction involves at least one species
If an attempt is made to fit heterogeneous chemisorbed on the catalyst surface. If reaction
catalytic reaction rate data to a rate expression takes place between two adsorbed species, they
of the form must be adsorbed on neighboring sites in order
for reaction to occur. The probability of reaction
r" = (6.3.2) between adsorbed A and adsorbed B is assumed
to be proportional to the product of the fractions
it is often found that the reaction orders are not
of the sites occupied by each species (0A6B).
integers, that the orders depend on temperature
Similar considerations apply to termolecular
and perhaps on concentration, and that the
reactions occurring on the surface.
observed activation energy may depend on
In addition to the aforementioned assump-
temperature. Such expressions are of rather
tions, one normally elects one of two mutually
limited use for design purposes and can be used
incompatible assumptions as a basis for his or
with confidence only under conditions that
her analysis. These two limiting cases are as
closely resemble those under which the experi-
follows.
mental data were recorded. The basic reason
this sort of approach often gives rise to such Category I. Adsorption equilibrium is maintained
difficulties is that it assumes that the driving at all times, and the overall rate of reaction is
force for reaction is the concentrations of the governed by the rate of chemical reaction on the
species present in the fluid phase. A more logical surface. The expressions developed for 9t in
approach is to use the surface concentrations of Section 6.2 can be used in this case.
the adsorbed species in an expression analogous
Category II. The rate of chemical reaction on
to equation 6.3.2; that is,
the surface is so rapid that adsorption equilibrium
r" = (6.3.3) is not achieved, but a steady-state condition is
reached in which the amount of adsorbed material
where 0t is the fraction of the surface covered
remains constant at some value less than the
by species i. If the various 9{ can be related to
equilibrium value. This value is presumed to be
bulk fluid phase concentrations, one has an
that corresponding to equilibrium for the surface
appropriate mathematical form to use in testing
reaction at the appropriate fractional coverages
a proposed reaction rate expression. The basis
of the other species involved in the surface reac-
for most analyses of this type is the Langmuir
tion. The rate of adsorption or desorption of one
isotherm. When expressions for the various Qt
species is presumed to be much slower than that
in terms of the appropriate partial pressures
of any other species. This step is then the rate
involved in the Langmuir isotherm are substi-
limiting step in the overall reaction.
tuted in equation 6.3.3, a Langmuir-Hinshel-
wood or Hougen-Watson rate expression is We will treat Category I models first.
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 183
6.3.1.1 Hougen-Watson Models for the Case of the sum in the denominator. If the product R is
Equilibrium Adsorption. This section treats strongly adsorbed, equation 6.3.6 provides a
Hougen-Watson mathematical models for cases basis for explaining inhibition of the reaction
where the rate limiting step is the chemical by that material. For example, if the dominant
reaction rate on the surface. In all cases it is term in the denominator is KRPR, the appro-
assumed that equilibrium is established with priate approximation for the rate expression is
respect to adsorption of all species.
kKAPA
Case I. Irreversible Reaction (Unimolecular) r = (6.3.7)
where E is the activation energy of the surface For the case of bimolecular reaction between
reaction and AHt is the enthalpy change accom- two adsorbed A molecules, similar arguments
panying chemisorption of species i. In the last lead to the following rate expression.
case, the relative magnitudes of the heats of
adsorption of A and R will determine whether r = (6.3.12)
the apparent activation energy will be greater KAPA + KRPR + KSPS)2
than or less than that of the surface reaction.
Case III. Irreversible Bimolecular Reactions Be-
Case II. Irreversible Bimolecular Reaction Be- tween Species Sorbed on Different Types of Sites
tween Adsorbed Species on the Same Type of With certain types of catalysts it is easy to
Site postulate that more than one type of chemisorp-
Consider the following irreversible bimolecular tion site may exist on the solid surface. For
reaction, which takes place between two species example, in the case of metal oxide catalysts,
adsorbed on a catalyst surface. one might speculate that certain species could
chemisorb by interaction with metal atoms at
A + B -+ R + S (6.3.9) the surface, while other species could interact
In this case the reaction rate is assumed to be with surface oxygen atoms. Consider the possi-
proportional to the product of the fractions of bility that species A adsorbed on one type of
the sites occupied by species A and B. site will react with species B adsorbed on a
second type of site according to the following
r" = k0A0B (6.3.10) reaction.
In the general case where all species may be
adsorbed on»the same type of surface site equa- 1 site) + ^(Type 2 site) (6.3.13)
tions 6.2.16 and 6.2.17 may be used for 0A and
For this mechanism the rate expression is given
6B. Thus
by
kKAKBPAPB r" = k0A1dB2 (6.3.14)
KAPA KBPB KRPR KSPS)2 where 0Ai is tfre fraction of the sites of Type 1
(6.3.11) occupied by species A and 6B2 is the fraction of
There are severaLiimiting forms of this rate the sites of Type 2 occupied by spef ies B. If the
expression, depending on which species are Langmuir adsorption isotherm equation is writ-
strongly or weakly adsorbed. ten for each type of site, if only species A is
If all partial pressures except that of one re- adsorbed on sites of Type 1, and if only species
actant, say species A, are held constant and the B is adsorbed on sites of Typte 2,
partial pressure of species A is varied, trje rate
will go through a maximum. The rate increases r" = k (6.3.15)
KB2PB
initially with increasing partial pressure of re-
actant A as the fraction of sites occupied by Rate expressions of this type have been pro-
species A increases. However, as this fraction posed to correlate the data for the reaction
increases, the fraction occupied by species B between hydrogen and carbon dioxide on
declines as B molecules are displaced by A tungsten.
molecules. Eventually one reaches a point where ^ Since adsorption occurs independently on
the decline ip the value of 0B more than offsets each type of site and displacement of one species
the increase in 6A and the product 6A0B goes by another does not occur, this type of reaction
through a maximum. will not exhibit a maximum of the type noted in
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 185
Case II. Instead, the rate will increase with in- maximum as the concentration of a reactant is
creasing pressure of one reactant, eventually increased at constant values of other species
approaching an asymptotic limit corresponding concentrations.
to saturation of the type of site in question. If the surface is nearly covered (0A « 1) the
reaction will be first-order in the gas phase re-
Case IV. Irreversible Bimolecular Reactions Be- actant and zero-order in the adsorbed reactant.
tween an Adsorbed Species and a Fluid Phase On the other hand, if the surface is sparsely
Molecule covered (8A « KAPA) the reaction will be first-
order in each species or second-order overall.
The first three examples consider cases where Since adsorption is virtually always exothermic,
only adsorbed species participate in the reaction. the first condition will correspond to low
The possibility also exists that reaction may temperature and the second condition to high
occur between a gaseous molecule and an ad- temperatures. This mechanism thus offers a
sorbed species. This mechanism is referred to as ready explanation of a transition from first-
a Rideal mechanism. If species A refers to the to second-order reaction with increasing tem-
adsorbed species and species B to a gas phase perature.
molecule, one form of this bimolecular reaction
may be written as
Case V. Reversible Reactions Between Adsorbed
A + B{g) ^R + S (6.3.16) Species (Change in Number of Moles on Reaction)
with a rate expression given by Consider the reaction
r" = k6APB (6.3.17) + S (6.3.20)
If we consider the general case where all species where all species are adsorbed on surface sites.
may be adsorbed, 9A is given by equation 6.2.16, It is assumed that an adsorbed A molecule re-
and the rate becomes acts with a vacant site to form an intermediate
that then dissociates to form adsorbed R and S
r = species. This implies that the number of sites
KAPA KBPB KRPR KSPS involved in a reaction obeys a conservation
(6.3.18) principle.
The rate of this surface reaction can be
If species B had been adsorbed rather than
written as
species A, the corresponding rate equation
would be r" = kx0A9v - k-tOzOs (6.3.21)
, , k K B P A P B where 0v is the fraction of the sites that are
r KAPA KRPR KSPS vacant. This quantity is given by
(6.3.19) 0v = 1 - (0A + 6R + 9S) (6.3.22)
Both equations have the same mathematical or
K
form. In these case's there will not be a maximum A?A + KRPR + KSPS
0,, = 1 -
in the rate as the pressure of one species varies. KAPA + KRPR + KSPS
Instead, the rate will approach an asymptotic 1
value at high pressures. Consequently this type (6.3.23)
of mechanism can be excluded from subsequent KAPA + KRPR + KSPS
consideration if it can be shown that the ex- Substitution of this result into equation 6.3.21
perimentally observed rate passes through a together with the appropriate expressions for
186 Elements of Heterogeneous Catalysis
the fractional coverages of other species gives Driving Force Term: In all rate expressions
the driving force must become zero when thermo-
k1KAPA-k_iKRKsPRPs
r = dynamic equilibrium is established. The "equi-
+ KAPA + KRPR +• KSPS)2 librium" constant K appearing in equation
(6.3.24) 6.3.29 can be regarded as the appropriate ratio
or of partial pressures for the overall reaction.
KRKSPRPS
K = (6.3.31,)
KrK4
r =
Kx -r KAPA + KRPR + KSPS) In the general case the value of K appearing in
(6.3.25) the driving force term is the product of the
where Kr is the equilibrium constant for the equilibrium constant for the surface reaction
surface reaction. Kr and the product of the adsorption equilib-
rium constants for the reactants divided by the
Kr -
K (6.3.26)
product of the adsorption equilibrium constants
for the reaction products.
If we let Kinetic Term: The designation kinetic term
k == k1KA (6.3.27) is something of a misnomer in that it contains
both rate constants and adsorption equilibrium
and
constants. For the cases where surface reaction
K= (6.3.28) controls the overall conversion rate it is the
product of the surface reaction rate constant
equation 6.3.25 can be rewritten as for the forward reaction and the adsorption
equilibrium constants for the reactant surface
kd ~ A
K )
species participating in the reaction. When
V p ^2
adsorption or desorption of a reactant or
"(1 + KAPA + KRPR + K s^s) product species is the rate limiting step, it will
(6.3).29) involve other factors.
Adsorption Term: When all reactants, prod-
ucts, and inerts are adsorbed without dissocia-
The cases discussed above represent only a
tion such that adsorption equilibrium prevails,
small fraction of the surface reaction mecha-
the adsorption term is given by
nisms which might be considered. Yang and
Hougen (12) have considered several additional (1 + KAPA + KBPB + KRPR + KSPS + KtP,r
surface reaction* mechanisms and have devel-
where Kr and P 7 are the adsorption equilibrium
oped tables from which rate expressions for
constant and the gas phase partial pressure of
these mechanisms may be determined. They
inerts, respectively. When dissociation of a
approached' this problem by writing the rate species occurs, the term KtPi must be replaced
expression in the following form. by ^KiPi as long as adsorption equilibrium
(kinetic term) x (driving force) holds for all species. When adsorption equilib-
Rate = rium is not maintained, other modifications in
(adsorption term)"
the adsorption term are necessary.
(6.3.30)
The exponent n on the adsorption term is
Let us examine each of the terms in this formu- equal to the number of surface sites participating
lation of the rate expression. in the reaction, whether they hold adsorbed
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 187
reactants or participate as vacant sites. If more The rates of adsorption and desorption of
than one site is involved, we assume that the species B are equal. Hence,
sites must be adjacent for the surface reaction
— K-1BUB (6.3.35)
to occur.
where 6V is the fraction of the sites that are
6.3 J.2 Hougen-Watson Models for Cases where vacant, and klB and k_ 1B are the rate constants
Adsorption and Desorption Processes are the for adsorption and desorption of species B.
Rate Limiting Steps. When surface reaction Solving for 0B,
processes are very rapid, the overall conversion
rate may be limited by the rate at which ad- 0B = ^ ^ = KBPB6V (6.3.36)
sorption of reactants or desorption of products
takes place. Usually only one of the many species A similar analysis is applicable to other
in a reaction mixture will not be in adsorptive species that attain adsorptive equilibrium. Thus,
equilibrium. This generalization will be taken
as a basis for developing the expressions for OR = KRPROV and 6S = KSPSOV
overall conversion rates that apply when ad- (6.3.37)
sorption or desorption processes are rate lim- The sum of the fractions of sites occupied by
iting. In this treatment we will assume that the various species and the fraction vacant must
chemical reaction equilibrium exists between be unity.
various adsorbed species on the catalyst surface,
even though reaction equilibrium will not pre- I = 0A + 0B + 0R + 0s + 0v (6.3.38)
vail in the fluid phase. Combination of equations 6.3.34 and 6.3.36
Consider the reversible surface reaction to 6.3.38 gives
1
A + B R (6.3.32) 6V =
k-i
This Hougen-Watson model for the case tive equilibrium; that is,
where the adsorption of a single species is rate
limiting is of the same mathematical form as
eA = KApAev eB = KBPBOV es = KSPSOV
equation 6.3.30. In this case the kinetic term is (6.3.47)
merely the rate constant for the rate controlling The reaction on the surface is also assumed to
adsorption process. The driving force term will be at equilibrium.
depend on the stoichiometry of the surface
reaction, which is presumed to be at equilibrium. Kr9A6B
(6.3.48)
The adsorption term must also be modified in
these cases, the change occurring in the element
in the summation corresponding to the species The sum of the various types of fractional
that is not at adsorptive equilibrium. This term coverage is unity.
will also depend on the nature of the surface l=0A + 0B + 0R + 0s + 0v (6.3.49)
reaction involved.
Extension of the analysis developed above to Combination of equations 6.3.47 to 6.3.49 and
cases where the stoichiometry of the surface rearrangement gives
reaction differs from that considered is relatively
simple and straightforward. An interesting case a- 1
is that where the overall conversion rate is . , KrKAKBPAPB
limited by adsorption of a species that disso-
ciates on adsorption. Consider a reaction whose (6.3.50)
stoichiometry can be represented by
The conversion rate is equal to the net rate of
A2 — R + S (6.3.43) desorption of species R. Hence,
where dissociation of the A2 molecule occurs
on adsorption. The mechanism of the reaction r" = k-1ROR-klRPROv=k-1R(0R-KRPR6v)
on the catalyst surface is represented by
(6.3.51)
2A R + S (6.3.44) Combination of equations 6.3.47, 6.3.48,
6.3.50, and 6.3.51 gives
In this case the above procedure can be used
to show that KrKAKBPAPB
fc-i
KRKSPRPS
P — r =-
r
A^ KAKr KrKAKBPAPB
r" =-
K
RKSPRPS
(6.3.52)
(6.3.45) The analyses developed in this section are
readily extended to reactions with different
Now consider the case where the controlling
stoichiometries. Regardless of whether an ad-
step is the rate of desorption of a product species
sorption or a desorption process is rate limiting,
R for a reversible surface reaction of the form
the resulting rate expressions may be written in
A + B^ R + S (6.3.46) the typical Hougen-Watson fashion represented
by equation 6.3.30. A comprehensive summary
All other species are presumed to be in adsorp- of such relations has been developed by Yang
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 189
and Hougen (12). Illustration 6.1 indicates the The following mechanistic equations for
development of a Hougen-Watson model in an chemisorption and reaction are appropriate.
attempt to fit rate data for a catalytic reaction.
P + a (A)
ILLUSTRATION 6.1 DEVELOPMENT OF A
HOUGEN-WATSON RATE EXPRESSION H, + 2a (B)
FOR A HETEROGENEOUS CATALYTIC Ku^ = adsorption equilibrium constant for
REACTION hydrogen
Oldenberg and Rase (13) have studied the Pff + 2H a - ka + 2a (C)
catalytic vapor phase hydrogenation of pro-
pionaldehyde over a commercially supported K eq = equilibrium constant for surface reaction
nickel catalyst. Their data indicate that the A, - A + a (D)
mathematical form of the reaction rate at very
low conversions and 150 °C can be expressed 1/XA = reciprocal of adsorption equilibrium
quite well in the following manner. constant for alcohol
The rate of the overall reaction is equal to
/cPp
that of the rate controlling step.
K = or (G)
Solution
It is convenient to assign the following symbols From equation D,
before proceeding to a discussion of the mecha-
nism proper.
or v (H)
P = propionaldehyde
A = propanol The expressions for 0A and 0H may now be
a — surface site substituted into equation G to give
PCT = adsorbed propionaldehyde—surface cov- KAPAOV
erage = 0P (I)
Ha — atomically adsorbed hydrogen—surface
coverage = 6H The sum of the fractional coverages must be
unity. Hence,
Aa = adsorbed alcohol—surface coverage =
190 Elements of Heterogeneous Catalysis
Substitution of equations F, H, and I into 3 may be applied to data from catalytic reac-
equation J and rearrangement gives tions, (i.e. one postulates a rate expression and
then uses an integral or differential method to
1
determine if the data fits the rate expression).
Because of the complexity of Hougen-Watson
rate models, it is often convenient to use an
Substitution of this result into equation E initial rate approach to analyze the data. The
gives remainder of this section is (jievoted to a dis-
cussion of the use of this approach to determine
r ,, = k1PPP-(k_1PKAPJKeqKH2PH2) reaction rate expressions.
Studies of the influence of total pressure on
the initial reaction rate for pure reactants present
in stoichiometric proportions provide a means
For an initial rate study, PA « 0, and the last of discriminating between various classes of
equation becomes Hoi^gen-Watson models. Isolation of a class of
„ k1PPP probable models by means of plots of initial
reaction rate versus total pressure, feed com-
position, and temperature constitutes the first
If the second term in the denominator is step)in developing a Hougen-Watson rate model.
large compared to unity, Hougen (14) has considered the influence of
total pressure for unimolecular and bimolecular
surface reactions; the analysis that follows is
adopted from his monograph.
In initial rate studies no products need be
iJnder these conditions the Hougen-Watson present in the feed, and the terms in the rate
model is consistent with the experimentally expression involving the partial pressures of
observed rate expression. these species may be omitted under appropriate
experimental conditions. The use of stoichiome-
tric ratios of reactants may also cause a simpli-
6.3.2 Interpretation of Experimental Data
fication of the rate expression. If one considers
The problem of determining which, if any, of a reversible bimolecular surface reaction be-
the Hougen-Watson models developed in the tween species A and B,
previous section fit- experimental data for a
A + + S (6.3.53)
heterogeneous catalytic reaction is generally
more complex than the corresponding problem an analysis of the type employed in Section
for a homogeneous phase reaction. Usually 6.3.1.1 indicates that if the surface reaction is
many possible controlling steps must be exam- controlling, the conversion rate is given by:
ined. Physical limitations on the overall reaction
rate must first be eliminated by operating in a i KAKBPAPB -
turbulent system and using sufficiently small
r =
catalyst pellets. The multitude of Hougen- + KAPA + KBPB + KRPR + KSPS)2
Watson rate expressions for the cases where (6.3.54)
adsorption, desorption, and reaction on the
surface are rate limiting is then considered. The If an equimolar feed of species A and B is used,
trial-and-error procedures developed in Chapter the dependence of the initial rate on total pres-
6.3 Reaction Rate Expressions for Heterogeneous Catalytic Reactions 191
mathematical formulation (which is not neces- formulation and preparation technique will
sarily unique), tempts one to try to correlate produce a material that is suitable for industrial
experimental data by each of a variety of such applications. The magnitude of the surface area
mathematical forms and then to "conclude" that alone is insufficient to predict the efficacy with
the one which best fits the data is the "true" which a given mass of catalyst promotes a given
kinetic mechanism. Considerably more judg- reaction. For fast reactions, the greater the
ment than this is required and, although a given amount of catalyst surface open to easy access
mathematical expression may provide a good by reactant molecules, the faster the rate at
fit of the data, it does not necessarily imply that which reactants can be converted to products.
the presumed mechanism is indeed the "correct" If we are to be able to develop useful predictive
one. For rate expressions of the form of equation models of catalyst behavior, we must know
6.3.58, there are five adjustable constants; when something about the interior pore structure of
reversible reactions, the possibility of adsorption the catalyst and the facility with which reactant
of inerts, or reactions in parallel or series are and product species can be transported within
considered, still more constants are introduced. catalyst pellets.
With the flexibility given by these constants it This section discusses the techniques used to
is not hard to get a good fit of data to a mathe- characterize the physical properties of solid
matical expression. In the words of one of my catalysts. In industrial practice, the chemical
colleagues, "With six constants you can fit a engineer who anticipates the use of these cata-
charging rhinoceros." lysts in developing new or improved processes
Furthermore, one often finds that "best" fits must effectively combine theoretical models,
of data may give rise to negative adsorption physical measurements, and empirical informa-
equilibrium constants. This result is clearly im- tion on the behavior of catalysts manufactured
possible on the basis of physical arguments. in similar ways in order to be able to predict
Nonetheless, reaction rate models of this type how these materials will behave. The complex
may be entirely suitable for design purposes if models are beyond the scope of this text, but
they are not extrapolated out of the range of the principles involved are readily illustrated by
the experimental data on which they are based. the simplest model. This model requires the
We should also point out that the adsorption specific surface area, the void volume per gram,
equilibrium constants appearing in the Hougen- and the gross geometric properties of the cata-
Watson models cannot be determined from lyst pellet as input.
adsorption equilibrium constants obtained from The specific surface area is usually deter-
nonreacting systems if one expects the mathe- mined by the BET technique discussed in
matical expression to yield accurate predictions Section 6.2.2. For the most reliable BET mea-
of the reaction rate. One explanation of this fact surements the adsorbate gas molecules should
is that probably only a small fraction of the be small, approximately spherical, inert (to
catalyst sites are effective in promoting the avoid chemisorption), and easy to handle at the
reaction. temperature in question. For economy, nitrogen
is the most common choice with measurements
6.4 PHYSICAL CHARACTERIZATION OF
usually made at 77 °K, the normal boiling point
HETEROGENEOUS CATALYSTS
of liquid nitrogen. Krypton is another material
The ease with which gaseous reactant molecules that is frequently employed.
achieve access to the interior surfaces of a The remainder of this section is devoted to a
porous substance is often of crucial significance discussion of the experimental techniques used
in determining whether or not a given catalyst to determine the other physical properties of
6.4 Physical Characterization of Heterogeneous Catalysts 193
catalysts that are of primary interest for reactor determine the volume occupied by the helium
design purposes, the void volume and the pore (VHe). This volume is equal to the sum of the
size distribution. volume exterior to the pellets proper and the
void volume within the pellets (Kvoid). The helium
6.4.1 Determination of Catalyst is then pumped out, and the chamber is filled
Void Volumes with mercury at atmospheric pressure. Since the
mercury will not penetrate the pores of most
The simplest method of determining the void
catalysts at atmospheric pressure, the mercury
volume or the pore volume of a given catalyst
will occupy only the volume exterior to the
sample is to measure the increase in weight that
pellets proper {VHg). Hence:
occurs when the pores are filled with a liquid of
known density. Water, carbon tetrachloride, and ^void - Kie " (6.4.2)
various hydrocarbons have been used success-
fully. The procedure involves boiling a known and the void volume per gram (Vg) is
weight of dry catalyst pellets in the liquid for K
Hg
20 to 30 min to displace the air in the pores. (6.4.3)
After the boiling fluid is replaced with cool w
liquid, the pellets are placed on an absorbent This method determines the pore volume cor-
cloth and rolled to remove the excess liquid. responding to pore radii below 75,000 A. By
They are then weighed. The difference: between varying the pressure on the system, it is possible
the wet and dry weights (Wwet — Wdry) divided to force the mercury into some of the pores and
by the density of the imbibed liquid (pL) gives determine the void volumes corresponding to
the void volume. The void volume per gram of different pore radii. We will pursue this point
catalyst (Vg) is obtained by in the next section.
The porosity of the catalyst pellet (sp) is
w —w defined as the void fraction.
Vn = rK
wet rK
d (6.4.1)
_ void volume of catalyst particle
Jp (6.4.4)
This technique will determine the total volume total volume of catalyst particle
of the pores with radii between approximately
10 and 1500 A. It is limited in accuracy by the For a particle of mass mP,
fact that it is difficult to dry the external surface mPVn
of the particles without removing liquid from (6.4.5)
the large pores. Some liquid also tends to be mPVg mP
Pskeletal
retained around the points of contact between
particles. These two sources of error offset one where p ske i etal is the true density of the bulk
another. Any air retained within the pores after solid and where the second term in the denom-
boiling will lead to erroneous results. inator is the volume occupied by the solid
A much more accurate method of determining proper. Many commercial catalysts have po-
the pore volume of a catalyst sample is the rosities in the neighborhood of 0.5, indicating
helium-mercury method. One places a known that the gross particle volume is about evenly
weight of catalyst (W) in a chamber of known split between void space and solid material.
volume. After the chamber has been evacuated, Significantly higher porosities are not encoun-
a known quantity of helium is admitted. From tered in commercial catalysts because of the
the gas laws and measurements of the tempera- problems of achieving sufficient mechanical
ture and pressure, one may then proceed to strength at these porosity levels.
194 Elements of Heterogeneous Catalysis
The total pore volume can also be determined while the surface area per particle is given by
from adsorption measurements if one knows
the volume of vapor adsorbed under saturation mPSg = np(2nrL) (B)
conditions. For high surface area catalysts the where we have neglected the area at the end of
amount of material adsorbed on particle ex- the cylinder, since L » 7 for virtually all systems
teriors will be negligible compared to that of interest. Division of equation A by equation
condensed in the pores. Hence the liquid phase B and rearrangement gives
volume equivalent to the amount of gas ad-
sorbed is equal to the pore volume. The liquid 2Va
density is assumed to be that corresponding to r = (Q
the saturation conditions in question. This
technique is less accurate than that described
For a monodisperse system this result is in
previously.
good agreement with the values obtained from
Illustration 6.2 indicates how void volume
pore size distribution measurements, but it can
and surface area measurements can be combined
be significantly in error if one is dealing with a
in order to evaluate the parameters involved in
bimodal pore size distribution (see Section
the simplest model of catalyst pore structure.
6.4.2).
To evaluate the average pore length, it is
ILLUSTRATION 6.2 EVALUATION OF necessary to recognize that the porosity sp will
PARAMETERS IN A MATHEMATICAL represent not only the volumetric void fraction
MODEL OF POROUS CATALYSTS but also at any cross section the fraction of the
area occupied by the pore openings. If the
To develop analytical models for processes average open area associated with each pore is
employing porous catalysts it is necessary to assumed to be n~r2, the definition of the porosity
make certain assumptions about the geometry indicates that
of the catalyst pores. A variety of assumptions
are possible, and Thomas and Thomas (15) mPVa
(D)
have discussed some of these. The simplest
model assumes that the pores are cylindrical
where Vp and Sx are the gross geometric volume
and are not interconnected. Develop expres-
and geometric surface area, respectively.
sions for the average pore radius (7), the average
Substitution of equation A into equation D
pore length (L), and the number of pores per
gives
particle (np) in terms of parameters that can be
measured in the laboratory [i.e., the apparent nJnT2)L np7ir2
particle dimensions, the void volume per gram
(Vg), and the surface area per gram (Sg)~\. which can be rearranged to give
Solution L-V»
We will start by developing an expression for
For a spherical catalyst pellet of radius R,
the average pore radius?. If we denote the mass
of an individual catalyst particle by mP, simple R
geometric considerations indicate that the void L = 2
volume per particle is given by
4nR T
The number of pores per particle may be
mPV = nJnr2L) (A) obtained by solving equation D for np and
6.4 Physical Characterization of Heterogeneous Catalysts 195
smaller pore sizes, such results are suspect would expect to be most significant from a
because the molecular diameter of the adsorbate catalytic viewpoint. Some other industrial cat-
molecules is comparable in magnitude to the alysts are deliberately fabricated with a macro-
pore radius in this regime. Moreover, the anal- pore-micropore structure in order to minimize
ysis assumes that the properties of the condensed or eliminate pore diffusion limitations on the
phase in the capillaries are the same as those of reaction rate. The macropores serve as express-
bulk liquids, yet the concepts of surface tension ways that facilitate reactant transport to the
and a hemispherical surface become hazy or internal surface area of the catalyst.
unrealistic below radii of 10 A, and perhaps
6.5 CATALYST PREPARATION,
even at larger radii. In the regions where the
FABRICATION, AND ACTIVATION
mercury penetration approach and the desorp-
tion method overlap, they agree reasonably well. The task of developing a suitable catalyst for
However, in many cases of practical significance, commercial applications involves many consid-
it is necessary to employ both techniques to erations, ranging from obvious factors like
cover the entire range of pore sizes (10 to catalyst activity and selectivity to variables like
10,000 A) that may be present. This is particu- the catalyst shape and the composition of the
larly true of catalysts like pelleted alumina in binder used in a pelletizing process. This section
which one has a bimodal pore size distribution is devoted to a discussion of these considerations
of the type discussed below. and of the techniques involved in manufacturing
The catalysts employed in commercial fixed industrial catalysts.
bed reactors often exhibit a bimodal pore size Figure 6.6 illustrates some of the requirements
distribution, sometimes referred to as bidisperse to be kept in mind when developing a catalyst
or macro-micro distributions. In these systems for industrial applications. The art of catalyst
the bulk of the catalytic reaction occurs in pores formulation involves reconciling conflicting de-
with radii below 200 A, since this is where the mands that may be imposed by such require-
bulk of the surface resides. However, the tran- ments. The keystone of the arch is a material
sport of reactants to these small pores occurs that has high intrinsic activity and selectivity
primarily through macropores with radii rang- for promoting the reaction of interest. This
ing from 200 to 10,000 A. These catalyst pellets material may be used in bulk form or, more
are prepared by compacting fine porous pow- commonly, it will be supported on a carrier
ders. The micropore region arises from the pore material that may or may not have catalytic
structure within each of the particles of the activity of its own. The activity per unit volume
original powder, while the macropores are is of practical significance because process eco-
formed by the passageways around the com- nomics are often strongly dependent on the cost
pacted particles. As the pelleting pressure is of packed reactor space. Consequently, the cata-
increased, the macropores become successively lyst should have a high specific surface area,
reduced in size, but the micropores remain un- and it should have a pore structure such that
affected unless the crushing strength is exceeded. reactants can gain easy access to the inner sur-
Most commercial forms of alumina catalysts faces of the catalyst pellet. The first of these
exhibit this type of bimodal pore size distribu- requirements follows directly from the fact that
tion. For example, Rothfeld (17) has observed the reaction takes place at the fluid-solid inter-
two peaks, one centered at a radius of 6250 A face, while the requirement for an appropriate
and the other at 60 A. The micropores accounted pore structure follows from a desire to avoid
for about 99% of the surface area and 65% of diffusional limitations on chemical conversion
the void volume. It is these pores that one rates. (See Chapter 12.)
6.5 Catalyst Preparation, Fabrication, and Activation 197
Intrinsic
x • ^
Resistance to \
\ activity
and / Appropriate^^
/\ V
poisoning \
and
\ fouling
\
\
selectivity / pore size / > v
/ distribution/ N.
/ /Proper f l u i d \
/
/>v
Stable
structure \ J
f
^^
/ flow
/characteristics^/ \
>v
\ ^ -
/ Reproducible N. ^v X Low \
/ in its ^ \y^ cost \
yt^ behavior
/
Figure 6.6
Requirements for industrial catalysts.
The lifetime of a catalyst is another factor ponents into a structure that is not greatly
that plays a key role in determining process weakened or changed by sintering during use.
economics. This lifetime is the period during Very seldom does an industrial catalyst con-
which the catalyst produces the desired product sist of a single chemical compound or metallic
in yields in excess of or equal to a designated element. Most often a catalyst formulation
value. The life of a catalyst may terminate be- consists of a multitude of components, each of
cause of unacceptable changes in activity and/or which performs an essential task in the creation
selectivity, because of physical attrition, or of a commercially viable catalyst. Figure 6.7,
because of unacceptable changes in mechanical
properties arising from thermal cycling or other
factors in the process environment. Resistance
to poisoning and fouling as well as to other
catalyst deactivation processes is an attribute to
be preferred in commercial catalysts.
Mechanical and geometric properties also
play a strong role in determining if a given
formulation will give rise to a commercially
viable catalyst. Catalyst pellets must be suitably
shaped so that the reactant fluid can pass
through a packed bed without excessive pressure
drop or maldistribution of the flow. The finished
catalyst must also be mechanically robust to
avoid attrition during handling and loading
and to avoid crushing under sustained loads
during fixed bed operation. The requirements Figure 6.7
of high strength and a stable structure depend Components of industrial catalysts. (Ownership of
on a firm welding together of the catalyst com- this diagram is by ICI)
198 Elements of Heterogeneous Catalysis
adapted from the Catalyst Handbook (18), in- of quality, and cost of manufacture. The two
dicates that the components commonly found most commonly used techniques for catalyst
in industrial catalysts can be classified as cata- preparation are the only ones we will consider.
lytic species, catalyst supports, and catalyst
1. Precipitation or gel formation of the active
stabilizers. The support may be a ceramic matrix
component or components through the inter-
bound together by a hydraulic cement. The
action of aqueous solutions of two or more
support gives body and strength to the catalyst
chemical compounds.
granule and is frequently used to decrease the
2. Impregnation of a carrier using a solution
concentration of the usually more expensive
containing a compound of the desired cata-
catalyst proper. The catalytic agents consist of
lyst component.
the catalyst proper together with any materials
added to promote or modify the active surface. Precipitation is frequently used in the prep-
Stabilizers may be present in the form of gross aration of single- and multiple-component cata-
crystals readily seen under the scanning electron lysts. If possible, the catalyst precursor is
microscope, or they may be present in highly precipitated as a compound that will undergo
dispersed form. They prevent sintering and some chemical change before it is converted into
loss of active surface area through diffusional the substance that actually functions as the
merging of contiguous small crystals and crys- catalyst. Dehydration of a hydrated precipitate
tallites. Catalysts in which the active constituent may be sufficiently drastic to accomplish this
is a metal with a relatively low melting point purpose. The nature of the impurities occluded
normally are stabilized through the use of or adsorbed on the precipitate must be con-
refractory materials such as alumina, chromia, sidered in making up the starting solutions.
and magnesia, which act as spacers between Experience has indicated that the use of dilute
the readily sinterable metal crystals. Since com- solutions, nitrates or sulfates of the metal
mercial catalysts are complex multicomponent elements, and ammonia or ammonium salts as
mixtures, it is appropriate to describe in broad the precipitating agents generally facilitates the
terms some of the procedures used in the prep- removal of impurities. In these cases subsequent
aration of these materials. heat treatment often suffices to drive off the
impurities in volatile form. Significant variations
6.5.1 Catalyst Preparation in catalytic activity have been observed for
Natural products and common industrial chem- precipitates of the same material formed under
icals in massive form are seldom useful as different conditions. When preparing multicom-
catalysts because they have low specific surface ponent catalysts by coprecipitation techniques,
areas, may contain various amounts of im- the process variables must be manipulated to
purities that have deleterious effects on catalyst produce a uniform product. By adding a solu-
performance, do not usually have the exact tion of two metal salts to an excess of the
chemical composition desired, or are too ex- precipitating counter ion (carbonate, sulfide,
pensive to use in bulk form. The preparation of hydroxide, etc.) simultaneous precipitation of
an industrial catalyst generally involves a series both species is obtained. If the addition is
of operations designed to overcome such prob- carried out in the reverse order there is a pref-
lems. Many catalysts can be produced by several erential precipitation of one metal salt relative
routes. The actual choice of technique for the to the other because of differences in solubility
manufacture of a given catalyst is based on products.
ease of preparation, homogeneity of the final One of the simplest, most commonly used
catalyst, stability of the catalyst, reproducibility techniques for preparing a catalyst involves dis-
6.5 Catalyst Preparation, Fabrication, and Activation 199
persing an active component (or components) 200-mesh screen. Catalysts employed with gas-
on a support material. Normally one impreg- eous feeds in fixed bed reactors are most
nates the carrier material with a solution of a frequently used in the form of pellets, granules,
soluble precursor of the catalyst and then or extrusions, whereas those used in fluidized
converts this precursor to the desired product beds are usually ground to pass through a 60-
by oxidation, reduction, thermal decomposi- mesh screen. With cylindrical pellets the ratio
tion, or some other suitable step. Where appro- of length to diameter is usually kept below 3:1,
priate, it is preferable to use a granular support so that the pellets will not pack parallel to one
instead of a powder, since it eliminates the another and thus lead to excessive pressure
necessity for pelleting or extrusion in order to drop or excessive bypassing, channeling, or
obtain the final product. maldistribution of the flow. The ratio of the
In many cases drying operations are critical diameter of the reaction vessel to a characteristic
to the production of successful commercial pellet dimension is also significant, since it is
catalysts. Close control of the drying process almost impossible to achieve uniform packing
is necessary to achieve the proper distribution if this ratio is less than 5:1.
of the catalyst precursor within the pore struc- More detailed treatments of catalyst manu-
ture of the support. Drying also influences the facturing processes are available in the literature
physical characteristics of the finished catalyst (19-22).
and the ease with which subsequent pelleting or
extrusion processes may be carried out. 6.5.2 Catalyst Supports, Promoters,
In general all materials used to catalyze and Inhibitors
reactions of a vapor phase feed are calcined or It is very seldom that a commercial catalyst
activated at temperatures above 400 to 500 °F. consists of only a single chemical compound or
This heat treatment may accomplish one or element. Often the active constituent is sup-
more of the following tasks. ported on a carrier material that may or may
1. Reduce the moisture content to a level con- not possess catalytic activity of its own. En-
sistent with the temperature at which the hanced catalytic activity, selectivity, or stability
catalyst will be employed. may also be achieved by the addition of other
2. Decompose salts containing the catalyst pre- materials referred to as promoters or inhibitors.
cursors such as metal nitrates, formates, Early workers viewed carriers or catalyst
oxalates, or acetates. (Oxides are the usual supports as inert substances that provided a
products of the decomposition process.) means of spreading out an expensive material
3. Form metal catalysts by reduction of metal like platinum or else improved the mechanical
oxides with hydrogen. strength of an inherently weak material. The
4. Increase the strength of the finished pellet primary factors in the early selection of catalyst
or extrudate. supports were their physical properties and their
5. Influence the initial activity and stability of cheapness; hence pumice, ground brick, char-
the catalyst. coal, coke, and similar substances were used.
No attention was paid to the possible influence
The form in which the catalyst is employed of the support on catalyst behavior; differences
is strongly dependent on the reaction involved, in behavior were attributed to variations in the
the scale of the process, the specific nature of the distribution of the catalyst itself.
catalyst, and the type of reactor. Catalysts used However, it is now recognized that the cata-
for reactions carried out with a liquid feed are lyst carrier does more than provide a physical
usually ground so as to pass through a 100 to framework on which the catalyst is supported.
200 Elements of Heterogeneous Catalysis
In many cases there is an interaction between 8. Inert to attack from components of the feed-
the carrier and the active component of the stream and product stream and any com-
catalyst so that the character of the active sur- ponents present in any regeneration streams
face will change. For example, the electronic to which it may be exposed.
character of the supported catalyst may be 9. Noncatalytic with regard to any undesirable
influenced by the transfer of electrons across side reactions.
the catalyst-carrier interface. In some cases the
carrier itself has a catalytic activity for the The average pore size and the pore size dis-
primary reaction, an intermediate reaction, or tribution should be such that physical limita-
a subsequent reaction, and a "dual-function" tions are not placed on the conversion of
catalyst is thereby obtained. Materials of this reactants to products. The particle size of the
type are widely employed in reforming processes. carrier must also be suitable for the purpose
There are other cases where the interaction of intended (i.e., small for fluidized bed reactors and
the catalyst and support are much more subtle significantly larger for fixed bed applications).
and difficult to label. For example, the crystal The types of materials used commercially as
size and structure of supported metal catalysts catalyst supports can be categorized as having
as well as the manner in which the metal is high or low specific surface areas. The low sur-
dispersed can be influenced by the nature of the face area (up to 1 m2/g) materials are generally
support material. ceramic materials like silica or alumina. They
One should distinguish between true catalyst are often formed by high-temperature fusion
supports and diluents. A catalyst support (or (2000 °C) in electric furnaces. The product of
carrier) is a material on which a thin layer of this operation is crushed, sieved, and fabricated
catalyst is deposited; a diluent is an inert into various shapes (spheres, granules, rings,
material thoroughly mixed with the catalyst to cylindrical annuli, or irregular shapes). Partial
enhance the binding properties of a powdered sintering is then accomplished by firing at
catalyst or to assist in pelleting or extrusion temperatures in the neighborhood of 1400 °C.
fabrication procedures. The resultant products are stable mechanically
An ideal carrier is one that is: and are easily impregnated with a variety of
catalyst precursor solutions. The final product
1. Inexpensive. is quite porous with pore sizes typically in the
2. Available in large quantities of uniform range of 20 to 100 microns. Low surface area
composition. carriers are used where the need for a high
3. Sufficiently porous to permit dispersion of activity per unit mass of finished catalyst is less
catalyst on its interior surfaces. important than other factors like catalyst cost
4. Free of any components that may poison the or selectivity.
catalyst. The high specific surface area supports (10 to
5. Stable under operating and regeneration 100 m2/g or more) are natural or man-made
conditions. materials that normally are handled as fine
6. Strong enough to resist any thermal or me- powders. When processed into the finished cat-
chanical shocks that it is likely to suffer as alyst pellet, these materials often give rise to
well as any disruptive action arising from pore size distributions of the macro-micro type
fouling material that may be deposited in its mentioned previously. The micropores exist
pores. within the powder itself, and the macropores
7. Resistant to attrition. are created between the fine particles when they
6.6 Poisoning and Deactivation of Catalysts 201
are compressed together in a pellet press. Dia- step in the reaction. In this sense it may act
tomaceous earth and pumice (or cellular lava) as one component of a dual function catalyst.
are naturally occurring low-cost materials that 3. The promoter may slow down, or otherwise
are representative of this class of catalyst sup- influence crystal formation and growth, or
port. Among the synthetic carriers that can be produce lattice defects. These effects may
created by modern technology are those derived lead either to a higher activity per unit area
from clays, bauxite, activated carbon, and or to a higher specific surface area.
xerogels of silica gel and alumina gel.
Promoters may influence selectivity by poi-
Although a carrier is often the major con-
soning undesired reactions or by increasing the
stituent of a catalyst pellet, one often finds that
rates of desired intermediate reactions so as to
there are materials that are added in small
increase the yield of the desired product. If they
amounts during catalyst preparation in order
act in the first sense, they are sometimes referred
to impart improved characteristics to the fin-
to as inhibitors. An example of this type of action
ished catalyst. These materials are referred to as
involves the addition of halogen compounds to
promoters. They may lead to better activity,
the catalyst used for oxidizing ethylene to
selectivity, or stability. The manner in which
ethylene oxide (silver supported on alumina).
promoters act is not well understood, although
The halogens prevent complete oxidation of the
a number of plausible explanations have been
ethylene to carbon dioxide and water, thus per-
set forth. They remain one of the reasons for
mitting the use of this catalyst for industrial
the "black magic" aura of catalysis.
purposes.
In 1920 Pease and Taylor (23) proposed a
Another example of catalyst promotion is the
definition of promoter action that is still appro-
use of additives to inhibit loss of active surface
priate today. Promotion occurs in those "cases
area during operation. Since this effect enhances
in which a mixture of two or more substances
catalyst stability, such additives are often called
is capable of producing a greater catalytic effect
stabilizers. These promoters may inhibit sin-
than can be accounted for on the supposition
tering of active sites on the surface or growth of
that each substance in the mixture acts inde-
microcrystalline regions. They may form solid
pendently and in proportion to the amount
solutions with higher melting points than the
present." Hence "better" catalyst performance
active agents alone, thus permitting a reaction
implies an improvement beyond that expected
to be carried out at a higher temperature than
from a simple additive rule. The phenomenon
would otherwise be possible. Alumina is used in
is often characterized by a steep rise in activity
the manufacture of iron-based ammonia syn-
to a sharp maximum at a low concentration of
thesis catalysts for this purpose.
promoter followed by a decline to a lower or
In commercial catalysts it is often useful to
zero level of activity, depending on whether or
employ a multiplicity of promoters, each of
not the promoter alone is capable of acting as a
which serves a particular function or acts in
catalyst.
conjunction with other constituents to produce
Some hypotheses that have been offered in an a more desirable catalyst.
effort to explain promotion effects are:
6.6 POISONING AND DEACTIVATION
1. The promoter may change the electronic
OF CATALYSTS
structure of the solid in such a way that the
activity per unit area is increased. Since the earliest days of heterogeneous cataly-
2. The promoter may catalyze an intermediate sis, decreases in activity during use have been
202 Elements of Heterogeneous Catalysis
observed. The rates at which catalyst deactiva- activity. This process is referred to as reactiva-
tion processes take place may be fast or slow. In tion. If the preferential adsorption of the poison
some cases the decline in activity is so rapid that cannot be readily reversed, a more severe chemi-
the catalyst ceases to function effectively after a cal treatment or complete replacement of the
few minutes or hours of exposure to a reactant catalyst may be necessary.
feedstream. On the other hand, there are cases In addition to fouling and poisoning, there is
where the deactivation processes occur so slowly a third catalyst deactivation process that is com-
that the catalyst may function effectively for monly encountered in industrial practice. This is
months or years. In the design of commercial the phenomenon referred to as aging. It involves
catalytic processes, one obviously must take a loss in specific activity due to a loss in catalyst
these factors into account so as to allow for surface area arising from crystal growth or
periodic replacement or regeneration of his sintering processes. This process becomes more
heterogeneous catalyst. rapid as the temperature increases. It may also
If the deactivation is rapid and caused by the be increased by the presence of certain com-
decomposition or degradation of reactants or ponents of the feedstream or product stream or
products on the catalyst surface, the process is of trace constituents of the catalyst. In some
termed fouling. In this case a deposit is formed cases a flux or glaze capable of blocking catalyst
on the surface or in the pores that physically pores may be produced.
blocks a portion of the catalyst and prevents it Various schemes have been proposed for
from catalyzing the reaction. Such fouling is classifying poisons, but the one that is perhaps
particularly rapid with silica-alumina cracking the most convenient for chemical engineers in-
catalysts and in this application, the design of a terested in reactor design is the classification in
unit for catalyst regeneration is as important as terms of the manner by which the poison affects
the design of the reactor proper. An example of a chemical activity. In these terms one can dis-
cracking reaction leading to fouling might be tinguish between four general but not sharply
C 1 0 H 2 2 -+ C 5 H 1 2 + C 4 H 1 0 + C (solid on cat- differentiated classes.
alyst surface). The carbon deposit may be burned
off with oxygen and/or steam to regenerate the 1. Intrinsic Activity Poisons. These poisons de-
catalyst so that the active surface is no longer crease the activity of the catalyst for the
covered. primary chemical reaction by virtue of their
If the activity of the catalyst is slowly modified direct electronic or chemical influence on the
by chemisorption of materials that are not easily catalyst surface or active sites. The mecha-
removed, the deactivation process is termed nism appears to be one that involves coverage
poisoning. It is usually caused by preferential of the active sites by poison molecules, re-
adsorption of small quantities of impurities moving the possibility that these sites can
(poisons) present in the feedstream. Adsorption subsequently adsorb reactant species. Com-
of extremely small amounts of the poison (a mon examples of this type of poisoning are the
small fraction of a monolayer) is often sufficient actions of compounds of elements of the
to cause very large losses in catalytic activity. groups Vb and VIb (N, P, As, Sb, O, S, Se, Te)
The bonds linking the catalyst and poison are on metallic catalysts.
often abnormally strong and highly specific. 2. Selectivity Poisons. These poisons decrease
Consequently, the process is often irreversible. the selectivity of the catalyst for the main re-
If the process is reversible, a change in the tem- action. In many cases impurities in the feed-
perature or the composition of the gas to which stream will adsorb on the catalyst surface and
it is exposed may be sufficient to restore catalyst then act as catalysts for undesirable side re-
Literature Citations 203
actions. The classic examples of poisons of deactivation processes, consult the article by
this class are heavy metals such as Ni, Cu, V, Butt (26).
and Fe, which are present in petroleum stocks
in the form of organometallic compounds LITERATURE CITATIONS
such as porphyrins. When these feedstocks
1. Haensel, V., Industrial and Engineering Chemistry, 57
are subjected to catalytic cracking, these (6), p. 18, 1965.
organometallics decompose and deposit on
2. Thomas, C. L., Catalytic Processes and Proven Cata-
and within the catalyst. The metals then act as lysts, Academic Press, New York, 1970.
dehydrogenation catalysts. The product dis- 3. Thomson, S. J., and Webb, G., Heterogeneous Cata-
tribution is markedly affected with lower lysis, Oliver and Boyd, Edinburgh, 1968.
yields of gasoline and higher yields of light gas, 4. Brunauer, S., Deming, L. S., Deming, W. E., and Teller,
coke, and hydrogen. Very small poison con- E., J. Am. Chem. Soc, 62 (1723), 1940.
centrations suffice to produce large changes 5. Langmuir, I., J. Am. Chem. Soc, 38 (221), 1916.
in selectivity. In cases where a large number 6. Langmuir, I., J. Am. Chem. Soc, 40 (1361), 1918.
of parallel and series reactions are involved, 7. Taylor, H. S., Proc Roy. Soc, A108 (105), 1925.
the adsorbed poison is thought to change the
8. Brunauer, S., Emmett, P. H., and Teller, E., J. Am.
activation energy barrier for some of the com- Chem. Soc, 60 (309), 1938.
petitive intermediate steps, resulting in in- 9. Hougen, O. A., and Watson, K. M., Chemical Process
creased rates of undesirable side reactions Principles, Part 3, Kinetics and Catalysis, pp. 906-907,
relative to that of the primary reaction. Wiley, New York, copyright © 1947. Used with per-
3. Stability Poisons. These poisons decrease the mission.
structural stability of the catalytic agent or of 10. Chambers, R. P., and Boudart, M., J. Catalysis, 6 (141),
the carrier by facilitating recrystallizations 1966.
and other structural rearrangements. Steam 11. Hougen, O. A., and Watson, K. M., Ind. Eng. Chem.,
acts as this type of a poison for silica-alumina 35(529), 1943.
gel catalysts. It acts not so much by reducing 12. Yang, K., and Hougen, O. A., Chem. Eng. Prog., 46
the intrinsic activity per unit surface area, but (3), p. 149, 1950.
by reducing the active area, thus decreasing 13. Oldenburg, C. C , and Rase, H. F., AIChE J., 3 (462),
1957.
the activity per unit weight of catalyst. The
temperature at which the reactor operates has 14. Hougen, O. A., "Reaction Kinetics in Chemical Engi-
neering," Chem. Eng. Prog. Monograph Series, No. 1,
a marked effect on stability poisoning. Sinter- 47, Chapter 5, copyright © 1951. Adapted with
ing, localized melting, and recrystallization permission.
occur much more rapidly at high tempera- 15. Thomas, J. M., and Thomas, W. J., Introduction to the
tures than at low temperatures. Principles of Heterogeneous Catalysis, pp. 206-211,
4. Diffusion Poisons. This phenomenon is Academic Press, New York, 1967.
closely akin to catalyst fouling. Blockage of 16. Cranston, R. W., and Inkley, R A., Adv. Catalysis, 9
pore mouths prevents full use of the interior (143), 1957.
surface area of the pellets Entrained dust par- 17. Rothfeld, L. B., AIChE J., 9 (19). 1963.
ticles or materials that can react on the cat- 18. Catalyst Handbook, Springer-Verlag, New York, 1970.
alyst to yield a solid residue give rise to this 19. Ciapetta, F. G., and Plank, C. J., "Catalyst Prepara-
type of poisoning. tion," Chapter 7 in Catalysis, Volume I, edited by
P. H. Emmett, Reinhold, New York, 1964.
20. Ciapetta, F. G., Helm, C. D., and Baral, L. L., "Com-
Many reviews listing specific poisons for spe-
mercial Preparation of Industrial Catalysts,"'Chapter 2
cific catalysts are available in the literature (23- in Catalysis in Practice, edited by C. H. Collier, Rein-
25). For a more up-to-date review of catalyst hold, New York, 1957.
204 Elements of Heterogeneous Catalysis
Figure 6P.1
PROBLEMS
Typical adsorption rate curve for the adsorption of
1. The data below were recorded at constant hydrogen on zinc oxide at 1 atm pressure. The
temperature for the adsorption of nitrogen on curve to the left of A applies to 0 °C; at A the tem-
silica gel at — 196 °C. Prepare both a Langmuir perature is raised to 111°, at B to 154°, at C to
184, and at D to 218°. [Reprinted with permission
and a BET plot using coordinates that would be
from J. Am. Chem. Soc, 69 (1306), 1947. Copyright
expected to give rise to linear plots. What specific by the American Chemical Society.]
surface areas are predicted in each case? The
area covered per molecule of adsorbed nitrogen
is 0.162 nm2. The normal boiling point of liquid
nitrogen is — 196 °C. 3. Figure 6P.2 contains the Raman spectra of
Comment on your results. pyridine adsorbed on silica gel at three tempera-
tures as reported by Schrader and Hill [Rev. Sci.
lustrum, 46 (1335), 1975]. Four bands are ap-
Pressure Volume adsorbed per gram of sample parent at 991, 1006, 1032, and 1069 cm" K If the
(kPa) (cm3 at STP) band intensities are proportional to surface con-
centrations and if each band is associated with
0.8 6.1 one vibrational degree of freedom of an adsorbed
3.3 12.7
species, what is your interpretation of these data?
18.7 17.0
30.7 19.7
38.0 21.5 4. Restelli and Coull [AIChE J , 12 (292), 1966]
42.7 23.0 have studied the transmethylation reaction of
57.3 27.7 dimethylamine in a differential flow reactor
67.3 33.5 using montmorillonite as a catalyst. They
measured initial reaction rates under isothermal
conditions for this heterogeneous catalytic pro-
2. Taylor and Liang [J. Am. Chem. Soc, 69 cess. Steady-state operating data were recorded.
(1306), 1947] have studied the extent of adsorp- The initial reaction is represented by the fol-
tion of hydrogen on zinc oxide at a constant gas lowing stoichiometry.
pressure and at successively increasing tempera-
tures. A typical curve of the amount adsorbed as 2(CH 3 ) 2 NH^(CH 3 ) 3 N + CH 3 NH 2
a function of time is shown in Figure 6P.1.
Interpret these data. The feedstream consisted of essentially pure di-
Problems 205
10.0 1 1
1 1 ' 1
P
D~ 0.50 atm C -
> 8.0 -
6.0-
292.0°C
-cr ° °-
^ - * —
1 4.0-
278.8° c _
:* 2.0 -
-
, | . 1 . 1 1 . 1
100 200 300
Mass velocity, G (g/hr—cm2)
Figure 6P.3
Initial rate versus mass velocity. [Reprinted with
permission from AlChE., 12 (292), 1966.]
0.0033 20
0.0085 50
0.0140 90
0.0236 140
Time, t (sec) Total pressure (kPa) (a) In many previous studies of the catalytic
oxidation reaction it has been found that the
0 26.7 reaction is first-order with respect to carbon
10 30.4 monoxide.
60 34.1 r = -kPco
120 36.3
240 38.5 Determine if the data obtained by Brittan,
360 40.0 Bliss, and Walker are consistent with this
720 42.7 type of rate expression.
(b) On the basis of their experimental data,
It has been postulated that the rate expression Walker, Bliss, and Brittan proposed the fol-
for the reaction is of the form lowing mechanism for this reaction.
CO + Oa ^ O J C O ) slow
CO2? Explain qualitatively, in terms of the nickel catalyst. Instead of using a more complex
mathematical model, and in mechanistic Hougen-Watson model for the kinetic equation,
terms. they used an equation of the following form to
correlate the data.
11. An addition reaction of the type
A + B -+ C r0 =
where
is being carried out over a relatively inactive
metal catalyst at 400 °K. At this temperature r0 = initial reaction rate, g-moles
the rate appears to be first-order in the gas propionaldehyde/ksec-g unreduced
phase concentrations of both A and B. catalyst
r = kPAPB Pa, Ph = partial pressures of propionaldehyde,
hydrogen, kPa.
In studies of the temperature dependence of
the reaction, the rate goes through a maximum m,n = exponents in rate equation
(at fixed PA and PB) at a temperature near For their runs at approximately 150 °C, the
400 °K. How do you interpret these facts in unreduced catalyst weight was 0.7961 g and the
terms of the Hougen-Watson framework? following data were obtained.
Partial pressures
(kPa) Space velocity
2-moles nronionaldehvde/ Percent
propionaldehyde Hydrogen ksec-g unreduced catalyst conversion
12. Oldenburg and Rase [AIChE J , 3 (462), Determine the values of fc, m, and n in the
1957] studied the catalytic vapor phase hydro- simplified rate equation. Assume that the cata-
genation of propionaldehyde by making low lyst activity change is negligible over the course
conversion runs on a commercial supported of the experiments.
210 Elements of Heterogeneous Catalysis
14. The kinetics of the irreversible reaction number of gram moles of A converted per
A -» B + C has been studied in a laboratory minute per gram of catalyst.
scale recycle reactor. A nonporous solid catalyst
is being used to obtain an appreciable reaction (a) What is your interpretation of these data
rate. The data reported by your technician are in terms of the Hougen-Watson models?
shown in Figure 6P.6. Each of the sets of data (b) How would you replot these data in order
was recorded holding the partial pressures of to determine the rate constants and ad-
the other species constant. The homogeneous sorption equilibrium constants involved in
gas phase reaction may be neglected. The the model you propose in part (a)? These
reaction rate is expressed in terms of the plots are to be straight lines.
212 Elements of Heterogeneous Catalysis
80
Figure 6P.7
Effect of particle size and linear velocity on conversion.
[Reprinted with permission from the contribution of D. A.
Dowden and G. W. Bridger to Advances in Catalysis, 9 (669).
Copyright © 1957, Academic Press.]
Problems 213
0.2 -
0.1 -
0.05 -
\
0.02
0.01 I I I \
2.0 2.2 2.4 2.6 2.8
1000
10 20 30 40 50 60 70 80 90 100 T,°K
Temperature, °C
Figure 6P.10
Figure 6P.8 Effect of temperature on decomposition rate at a
Equilibrium decomposition of nickel tetracarbonyl. Reynolds number of about 15 and a pressure of
[Reprinted with permission from AlChE J., 13 (86), 20 torr. [Reprinted with permission from AlChE J.,
1967.] 13 (86), 1967.]
214 Elements of Heterogeneous Catalysis
5.0
I I 1 1 I
T = 225 C C^^°
2.0 -
^ ^
c o
.2 1 1.0 . — • — — • • —
~ - — •
II = 175°C
S 0.5
1
0.2 1 1 | I I
10 20 50 100
Reynolds number
Figure 6P.12
Effect of flow rate on nickel deposition at 50 torr
total pressure. [Reprinted with permission from
AlChEJ., 13 (86), 1967.]
215
216 Liquid Phase Reactions
Electrostatic effects other than ionization are actant molecules. These encounters will differ
abo important. Interactions between reacting appreciably from gas phase collisions in that
ions depend on the local electrical environ- they will occur in close proximity to solvent
ment of the ions and thus reflect the influence molecules. Indeed, in liquids any individual
of the dielectric constant of the solvent and the molecule will always be interacting with several
presence of other ions and various solutes that surrounding molecules at the same time, and the
may be present. In dilute solutions the influence notion of a bimolecular collision becomes rather
of ionic strength on reaction rates is felt in the arbitrary. Nontheless, a number of approaches
primary and secondary salt effects (see below). to formulating expressions for collision fre-
Another important physical effect arising quencies in the liquid phase have appeared in
from the presence of solvent molecules is the the chemical literature through the years. The
efficiency of energy transfer. In condensed phases simplest of these approaches presumes that the
the rapid energy transfer in the abundant colli- gas phase collision frequency expression is
sions between reactant and solvent molecules directly applicable to the calculation of liquid
maintains vibrational thermal equilibrium at all phase collision frequencies. The rationale for
times. Other solvent effects on reaction rates and this approach is that for several second-order
mechanisms include effects arising from the gas phase reactions that are also second-order
acidity of the medium; effects of selective solva- in various solvents, the rate constants and pre-
tion by one component in a mixed solvent or of exponential factors are pretty much the same
different degrees of solvation by different in the gas phase and in various solvents. For
solvents of the reactants, reaction interme- further discussion of the collision theory ap-
diates (including activated complexes), and re- proach (o reactions in liquids, consult the
action products; effects-arising from hydrogen monograph by North (3).
bonding; effects of changing from a protic sol- There is another aspect of collisions in liquid
vent (one that may be regarded as a hydrogen solution that is of particular interest with regard
bond donor, such as H 2 O, NH 3 , and alcohols) to chemical reactions. Collisions in solution are
to a dipolar aprotic solvent (not a hydrogen often repeated, so that multiple collisions of the
bond donor, such as acetone, SO 2 , or nitroben- same two molecules occur. Consider the mole-
zene); and cage effects. The various solvent cules labeled A and B in Figure 7.1. Each
effects may, of course, be superimposed or over- molecule is surrounded by several neighboring
lap one another such that they are inextricably molecules. In view of the short-range order
linked. The magnitude of these solvent effects typical of liquids, the neighboring molecules
may range from insignificance to several powers will all be located at approximately the same
of ten. For example, at 25 °C the reaction distance from the molecule in question. This
CH3I + c r ^CH 3 CI + r distance will be somewhat larger than a typical
occurs more than 106 times as fast in dimethyl- hard sphere molecular diameter, but consider-
acetamide as in acetone (1). On the other hand, ably less than twice this diameter. Consequently,
the decomposition of nitrogen pentoxide in geometric and molecular force considerations
several different solvents is characterized by rate indicate that the passage of any specified refer-
constants that lie within a factor of two of the ence molecule between any two neighbors will
gas phase rate constant (2). be restricted by the repulsive portion of the
Since the forces giving rise to the formation intramolecular potential. To escape from the
of chemical bonds are very short-range forces, "cage" formed by the surrounding molecules,
reactions in liquid solutions will require some the reference molecule must surmount the
sort of encounter or "collision" between re- energy barrier presented by the repulsive forces.
7.1 Electrostatic Effects in Liquid Solution 217
and Owen (9)]. The Debye-HUckel theory is the Debye-HUckel limiting law is applicable.
useless for quantitative calculations in most of The slope of such lines is nearly equal to the
the reaction systems encountered in industrial product of the ionic charges of the reactants
practice because such systems normally employ
concentrated solutions. However, it may be There are three general classes of ionic reac-
used together with the transition state theory to tions to which these equations may be applied.
predict the qualitative influence of ionic strength
1. Reactions between ions of the same sign
on reaction rate constants.
{zAzB is positive) for which the rate constant
For a bimolecular reaction between species A
increases with increasing ionic strength.
and B, the analysis gives an equation of the form
2. Reactions between ions of opposite sign
(zAzB is negative) for which the rate constant
k = k( (7.1.1) decreases with increasing ionic strength.
TAB
3. Reactions between an uncharged species and
where k0 is the rate constant in infinitely dilute an ion that this equation predicts to be in-
solution and yA, yB, and yAB are the activity co- dependent of ionic strength (zAzB = 0)
efficients of species A, B, and the activated
complex, respectively. Several tests of the validity of the last equation
If limiting forms of the Debye-HUckel expres- have been made through the years. Figure 7.2,
sion for activity coefficients are used, this taken from Laidler (10), indicates the variation
equation becomes of log(/c//c0) with ionic strength for a variety of
reactions. All factors considered, the agreement
of theory and experiment are quite remarkable
for a great many reactions. Note that the bulk
In k = In k0 + 2zAzBS[~\x
2(kBTD)3'2 of the data are reported in a range of ionic
(7.1.2) strengths that would normally be expected to be
where outside the range of validity of the Debye-
zA and zB represent the number (and sign) of HUckel limiting law (below yfjl = 0 . 1 for 1:1
charges on the ions A and B electrolytes and below yfji = 0.03 for higher
q is the charge on an electron valence ions). Equation 7.1.2 appears to be
No is Avogadro's number capable of extrapolation to much higher ionic
kB is the Boltzmann constant strengths for the reactions indicated. Also note
T is the absolute temperature that the magnitude of the effect can be quite
D is the dielectric constant of the medium significant, particularly for reactions that involve
\i is the ionic strength of the solution ions with multiple charges. Even in the case of
reactions where zAzB = ± 1 , the use of "inert"
H = iCLQzf) (7.1.3) salts to increase the ionic strength can cause the
where C, is the molar concentration of species i. rate to increase or decrease by as much as 50%.
In dilute aqueous solution at 25 °C Equation 7.1.2 characterizes what is known
as the primary salt effect (i.e. the influence of
log k = log k0 + 2zAzB^Ji(0.509) (7.1.4) ionic strength on the reaction rate through the
This equation is known as the Br0nsted- activity coefficients of the reactants and the
Bjerrum equation. It may be-derived in several activated complex). Much early work on ionic
ways. It predicts that a plot of log k versus the reactions is relatively useless because this effect
square root of the ionic strength should be was not understood. Now it is common practice
linear over the range of ionic strengths where in studies'of ionic reactions to add a considerable
7.2 Pressure Effects on Reactions in Liquid Solution 219
another species in the solution or by a decom- Early studies of the catalytic hydrolysis of
position process. esters indicated that in many cases, for strong
The relative ease with which proton transfer acids, the observed rate constants were inde-
is accomplished is responsible for the impor- pendent of the anion, and it became generally
tance of the generalized acid-base concept in accepted that the active catalyst was a hydrogen
solution chemistry. The Br0nsted concept of ion. In other reactions, it became necessary to
acidity is most useful in this respect. Br0nsted consider the effects of the hydroxide ion con-
defined an acid as a species that tends to give up centration and also the rate of the uncatalyzed
a proton and a base as a species that tends to reaction. The result was a three-term expression
accept a proton. In this sense any proton for the apparent rate constant:
transfer process having the general form
Apparent = K +
HX + Y -> HY + X (7.3.1) (7.3.2)
may be regarded as an acid-base reaction where the first term corresponds to the un-
between a Br^nsted acid HX and a Br0nsted catalyzed reaction, the second to catalysis by
base Y to form a conjugate base X and a hydrogen ions, and the third to catalysis by
conjugate acid HY. The various species involved hydroxide ions. The three rate constants ap-
in these reactions may be ions or neutral pearing in this expression vary with temperature
molecules. and with the nature of the reaction involved.
The unique properties of the proton have been Depending on the reaction conditions and
attributed by some authors to the fact that it has the reaction involved, one or two of these terms
no electronic or geometric structure. The ab- may be negligible compared to the other(s).
sence of any electron shell implies that it will When only the hydronium ion is effective in
have a radius that is about 105 times smaller catalyzing the reaction, the process is referred
than any other cation and that there will be no to as specific acid catalysis. When only hydroxide
repulsive interactions between electron clouds ions are effective, the process is classified as
as a proton approaches another reactant species. specific base catalysis. When both terms are
The lack of any geometric or electronic structure significant, the catalysis is characterized as
also implies that there will not be any steric specific acid and specific base catalysis.
limitations with regard to orientation of the Although the concepts of specific acid and
proton. However, it still must attack the other specific base catalysis were useful in the analysis
reactant molecule at the appropriate site. of some early kinetic data, it soon became
The small size of the proton relative to its apparent that any species that could effect a
charge makes the proton very effective in proton transfer with the substrate could exert a
polarizing the molecules in its immediate vi- catalytic influence on the reaction rate. Conse-
cinity and consequently leads to a very high quently, it became desirable to employ the more
degree of solvation in a polar solvent. In general Br0nsted-Lowry definition of acids and
aqueous solutions, the primary solvation process bases and to write the reaction rate constant as
involves the formation of a covalent bond with * = ^o + /cH+ (H 3 O + ) + fco
the oxygen atom of a water molecule to form a
hydronium ion H 3 O + . Secondary solvation + X kHXi(HXx) + X kXj(Xj)
1
j
of this species then occurs by additional water
(7.3.3)
molecules. Whenever we use the term hydrogen
ion in the future, we are referring to the H 3 O + where HXt and X} represent all other acid and
species. base species present in the solution apart from
222 Liquid Phase Reactions
H 3 O + and OH . Reactions that are dependent beyond the scope of this text. Consult the books
on the concentrations of HXt and Xj are by Bell (22, 23), Frost and Pearson (20), Laidler
categorized as involving general acid and general (16), Bender (17), and Hine (31) for a further
base catalysis. Table 7.1, adapted from Ashmore introduction to this topic.
(30), indicates a number of catalytic reactions Since the rates of acid and base catalyzed
of the specific and general acid-base types in reactions are sensitive to variations in the
order to provide some orientation as to the types solution pH, it is instructive to consider the
of reactions in the various categories. A thor- types of behavior that can be observed in
ough discussion of these reactions is obviously aqueous solution in the laboratory. The disso-
Table 7.1
Examples of Acid and Base Catalysis in Aqueous Solution (30)
Type of
catalysis Brief title of reaction Equation of reaction
Reprinted with permission from Catalysis and Inhibition of Chemical Reactions by P. G. Ashmore. Copyright
1963, Butterworths.
7.3 Homogeneous Catalysis in Liquid Solution 223
k = k (7.3.5)
^apparent 0
-pH -pH
Alternatively,
k — k 4- (7.3.6)
^apparent ^0 '
-pH -pH
In many cases one or two of the three terms
on the right side of these equations is negligible
compared to the others. In a 0.1N strong acid,
for example, the second term is /cH+ x 1CT1,
while the last term is /cOH- x 10~ 13 , since -pH
sociation equilibrium, it is evident that ionic strength of the solutions in which they
take place. There are two types of salt effects
(H + ) = Kaf^- = Kaq (7.3.12) that are significant in acid-base catalysis. The
first of these is the primary salt effect, discussed
where Ka is the dissociation constant for the in Section 7.1. This effect is significant for reac-
acid. The hydrogen ion concentration will not tions in solution involving ionic species. It
vary if the ratio (HA)/(A~) is maintained con- operates by influencing the activity coefficients
stant. Thus the first three terms on the right side of the reactants and the activated complex. The
of equation 7.3.11 may be represented by a con- second effect is referred to as the secondary salt
stant fcx under these conditions. The equation effect. It operates by changing the actual con-
may then be rewritten as centrations of the catalytically active ions. It
may cause either an increase or a decrease in the
Apparent = K + (HA)(^HA + kA-/q) (7.3.13) rate constant, and it may be either larger or
By plotting the measured rate constant versus smaller than a primary salt effect. In some
the undis'sociated acid concentration, one ob- instances both effects occur simultaneously, and
tains for this type of catalysis a straight line with a cancellation of effects can make the reaction
intercept ki and slope OL1 = (/cHA + kA-/q). If rate appear to be independent of ionic strength.
the procedure is repeated for other ratios, The secondary salt effect is important when
enough information is obtained to permit eval- the catalytically active ions are produced by
uation of kHA and kA-. The hydrogen and hy- the dissociation of a weak electrolyte. In solu-
droxide ion concentrations corresponding to a tions of weak acids and weak bases, added salts,
given ratio q may be determined from equation even if they do not exert a common ion effect,
7.3.12 and the dissociation constant for water. can influence hydrogen and hydroxide ion con-
It should be evident that the calculations and centrations through their influence on activity
the necessary experimental program are simpli- coefficients.
fied if one or more of the contributions is If one considers a reaction catalyzed by
negligible under appropriate conditions. In a hydrogen ions formed by the dissociation of a
relatively simple system in which either fcHA or weak acid HA, the hydrogen ion concentration
kA- is zero, the existence of general acid or is governed by the following relation:
general base catalysis may be deduced from the
pH dependence of the reaction. Ka = (7.3.14)
A reaction catalyzed by undissociated acid 7HA (HA)
will have the dependence of log k on pH shown where Ka is the equilibrium constant for the
in Figure 13g. Specific acid and specific base reaction and the y/s are the activity coefficients
catalysis are presumed to be absent. If specific of the various species. Taking logarithms and
and general acid and base catalysis are both rearranging gives
operative, one is able to obtain a variety of
interesting log k versus pH curves, depending
log(H + ) = log Ka + log C^) + log
on the relative contributions of the different
\A )
terms in various pH ranges. Curves i and j of
(7.3.15)
Figure 7.3 are simple examples of these types.
Since specific acid and specific base catalysis The concentration of hydrogen ions depends
and generalized acid and generalized base cata- on the ionic strength through the ratio of
lysis by cations and anions all involve ionic activity coefficients appearing in the dissociation
species, these processes are influenced by the equilibrium expression. Any change in ionic
226 Liquid Phase Reactions
strength affects the y terms and thus the hydro- ciently large that any variations caused by the
gen ion concentration. Consequently, if a reac- progress of the reaction will be negligible. This
tion is catalyzed by hydrogen (or hydroxide approach is similar to the method of excess
ions), the rate becomes dependent on ionic concentrations discussed in Chapter 3. It is
strength through this secondary salt effect. particularly useful in attempts to determine the
For the range of concentrations where the rate expression for reactions that would involve
Debye-Huckel theory is applicable it is possible significant changes in ionic strength if carried
to place the theory of the secondary salt effect out in the absence of extraneous ionic species.
on a more quantitative basis by using Debye- Unfortunately, the rate constants determined in
Huckel relations for the various activity coeffi- this fashion may be quite different from those
cients appearing in equation 7.3.15 or its more in highly dilute solutions. Nonetheless it is good
general analog. For an acid of charge z and its laboratory practice to add small quantities of
conjugate base of charge (z — 1), the analogous electrolytes to those reaction systems believed
equation is to involve ionic species in order to determine
the possible presence of ionic strength effects.
HAg Some judgment must be used in the selection
log(H + ) - \ogKa + log
of the added ions in order to choose species that
are noncatalytic in themselves and thus influence
+ log - (7.3.16) reaction rates only through the charges resulting
from electrolytic dissociation.
If the Debye-HUckel limiting law is used to
evaluate the various activity coefficients in 7.3.2 Catalysis by Enzymes
aqueous solution at 25 °C, the last equation
becomes Enzymes are protein molecules that possess
exceptional catalytic properties. They are essen-
HAZ) tial to plant and animal life processes. Enzymes
log(H + ) = \ogKa + log
are remarkable catalysts in at least three re-
spects : activity, specificity, and versatility.
z - 1) (7.3.17)
The high activity of enzymes becomes ap-
In the most common case z = 0, and the parent when the rates of enzyme catalyzed
secondary salt effect implies that the hydrogen reactions are compared to those of the corre-
ion concentration will increase with increasing sponding nonenzymatic reaction or to the same
ionic strength. However, the direction of the reaction catalyzed by an inorganic species. Rate
effect is determined by the sign of the quantity enhancements on the order of 108 to 10 11 are
(z - 1). not unusual in the presence of enzymes. Enzyme
The existence of the primary and secondary efficiencies are often measured in terms of turn-
salt effects indicates the importance of main- over numbers. This number is defined as the
taining control over ionic strength in kinetics number of molecules that are caused to react
studies. One may choose to keep the ionic in 1 min by one molecule of catalyst. For many
strength low so as to minimize its effects, or common reactions the turnover number is in
one may make a series of measurements at excess of 103 and, in some cases, it may exceed
various ionic strengths in order to permit extra- 106. High enzyme turnover numbers are largely
polation to the limit of infinitely dilute solution. the result of greatly reduced activation energies
Another useful alternative is to maintain the for the enzymatic reaction relative to other
ionic strength constant at a value that is suffi- modes of effecting the reactions in question.
7.3 Homogeneous Catalysis in Liquid Solution 227
Enzymes are often considered to function by the substrate, thereby producing a reaction
general acid-base catalysis or by covalent cata- intermediate in the sequence of steps leading to
lysis, but these considerations alone cannot product formation. The detailed configuration
account for the high efficiency of enzymes. of the enzyme molecule—including the con-
Proximity and orientation effects may be par- formation of the protein in folds or coils as well
tially responsible for the discrepancy, but even as the chemical structure near the active site—
the inclusion of these effects does not resolve is quite important, and it is said that the geo-
the disparity between observed and theoretically metric configuration is such that only molecules
predicted rates. These and other aspects of the with certain structural properties can fit. This
theories of enzyme catalysis are treated in the is the famous "lock and key" hypothesis for
monographs by Jencks (33) and Bender (34). enzyme activity that dates back to the work of
Enzyme specificities are categorized in terms Fischer in 1894. More sophisticated models that
of the manner in which enzymes interact with purport to explain enzyme specificity have been
various substrates. Some enzymes will cause proposed through the years, but the basic con-
only a single substrate to react. This type of cept that specificity results from steric or geo-
specificity is known as absolute specificity. An metric considerations remains unchanged.
example is urease, which catalyzes only the The third remarkable aspect of enzyme cata-
hydrolysis of urea. Other enzymes will react lysis is the versatility of these species. They
only with substrates having certain functional catalyze an extremely wide variety of reactions—
groups in certain positions relative to the bond oxidation, reduction, polymerization, dehydra-
to be attacked. This situation is called group tion, dehydrogenation, etc. Their versatility is a
specificity. An example of enzymes of this type reflection of the range and complexity of the
is pepsin, which requires an aromatic group to chemical reactions necessary to sustain life in
be present in a certain position relative to a plants and animals.
peptide linkage in order to effect its hydrolysis.
Reaction specificity is the least specific type of 7.3.2.1 Rate Expressions for Enzyme Catalyzed
enzyme catalysis. It requires only that a certain Single-Substrate Reactions. The vast majority
type of bond be present in the substrate. En- of the reactions catalyzed by enzymes are
zymes such as the lipases will catalyze the believed to involve a series of bimolecular or
hydrolysis of any organic ester. Many enzymes unimolecular steps. The simplest type of en-
exhibit stereochemical specificity in that they zymatic reaction involves only a single reactant
catalyze the reactions of one stereochemical or substrate. The substrate forms an unstable
form but not the other. Proteolytic enzymes, complex with the enzyme, which subsequently
for example, catalyze only the hydrolysis of undergoes decomposition to release the product
peptides composed of amino acids in the L species or to regenerate the substrate.
configuration. Reaction rate expressions for enzymatic reac-
Enzyme specificity is often explained in terms tions are usually derived by making the Bo-
of the geometric configuration of the active site denstein steady-state approximation for the
of the enzyme. The active site includes the side intermediate enzyme-substrate complexes. This
chains and peptide bonds that either come into is an appropriate assumption when the substrate
direct contact with the substrate or perform concentration greatly exceeds that of the enzyme
some direct function during catalysis. Each site (the usual laboratory situation) or when there is
is polyfunctional in that certain parts of it may both a continuous supply of reactant and a
hold the substrate in a position where the other continuous removal of products (the usual
parts cause changes in the chemical bonding of cellular situation).
228 Liquid Phase Reactions
The "classic" mechanism of enzymatic cata- Combining equations 7.3.21 and 7.3.23 gives
lysis can be written as
V = (7.3.24)
E -f - S^ES (7.3. 18)
k2
Since most kinetic studies of this type involve
F<\ —• F -4- P (7.3. 19) initial rate experiments, it is usually necessary
where to rederive this expression in terms of the initial
concentrations and the initial rate. From ma-
E represents the enzyme terial balance considerations
S represents the substrate
(£0) = (E) + (ES) (So) = (S) + (ES)
ES represents the enzyme-substrate complex
(7.3.25)
P represents the product of the reaction
For the conditions commonly encountered
The stoichiometry of the reaction may be in the laboratory, So » Eo. Since ES cannot
written as exceed Eo, this implies that So « S.
S -> P (7.3.20) Solving equation 7.3.25 for E and substituting
In a sense this mechanism is akin to Linde- this result into equation 7.3.22 gives
d(£S) = /
o) - (ES)](S0) - k2(ES) - k3(ES) ^ 0 (7.3.26)
dt
d(ES)
= kx(E)(S) - k2(ES) - k3(ES) « 0 This equation predicts that the initial rate
dt
will be proportional to the initial enzyme con-
(7.3.22) centration if the initial substrate concentration
or is held constant. If the initial enzyme concen-
k,(E)(S) tration is held constant, the initial rate will be
(ES) = (7.3.23)
k2 + k3 proportional to the substrate concentration at
7.3 Homogeneous Catalysis in Liquid Solution 229
low substrate concentrations, and substantially Equation 7.3.31 gives rise to what is known as
independent of substrate concentration at high an Eadie or Hofstee plot, while equation 7.3.32
substrate levels. The maximum reaction rate is gives rise to a Hanes plot. The Eadie plot has
equal to k3E0, and this product is often assigned the advantage of spreading the points out more
the symbol Vmax. The group (k2 + k3)/k1 is evenly and of determining K and Fmax separately.
often assigned the symbol K and is known as The three types of plots are shown schematically
the Michaelis constant. Equation 7.3.28 can be in Figure 7.4. The Lineweaver-Burk and Eadie
written in terms of these parameters as plots are the ones used most frequently in data
analysis.
(7.3.29) Although the Michaelis-Menten equation is
applicable to a wide variety of enzyme catalyzed
reactions, it is not appropriate for reversible re-
The Michaelis constant is numerically equal to
actions and multiple-substrate reactions. How-
the value of the initial substrate concentration
ever, the generalized steady-state analysis re-
that gives an initial velocity that is half that
mains applicable. Consider the case of reversible
of the maximum.
decomposition of the enzyme-substrate complex
Although equation 7.3.28 and, in particular,
into a product molecule and enzyme with
equation 7.3.29 are known as Michaelis-Menten
mechanistic equations.
rate expressions, these individuals used a some-
what different approach to arrive at this E + S^±ES^±E +P (7.3.33)
mathematical form for an enzymatic rate ex-
pression (35). In this case the net rate of reaction is given by
In attempting to determine if a given set of d(P)
experimental data is of the same mathematical V = = k3(ES) - k4(E)(P) (7.3.34)
dt
form as equation 7.3.29, there are three routes
The steady-state approximation for the inter-
that permit the graphical determination of the
mediate complex is
parameters Vmax and K. The most frequently
used plot is known as a Lineweaver-Burk or
reciprocal plot. It is based on rearrangement of 0 « ^ ~ = fci(£)(S) - k2(ES)
equation 7.3.29 into the following form.
- k3(ES) + kt(E)(P)
1 1 K (7.3.35)
(7.3.30)
If the conservation equation for total enzyme
If the data fit this model, a plot of 1/VO versus concentration (7.3.25) is employed, the last
1/(SO) should be linear with a slope K/Vmax and equation becomes
intercept 1/Knax- It 1 S analogous to that used in fci[(£0) - ES](S) - k2(ES) - k3(ES)
determining the constants in the Langmuir
equation for adsorption on solid surfaces. Other + /c4[(E0) - (ES)~](P) ^ 0 (7.3.36)
forms that may be used to prepare linear plots or
are UE0)(P)
(7 137)
UP) -
V — V — K (7.3.31)
y /v
0 ~~ 'max (So) Combining equations 7.3.25, 7.3.34, and 7.3.37
and gives
(So) (So) ^ k - k2kA(E0){P)
(7.3.32)
"o "max "max k2 /c4(P)
Reaction velocity, Vo
o
ine
CD
m 3
Q. Q-
CD Sf
X CD
s *h
O 3
7? N
•o o
I °-
D) - i
3 CD
CD Q>
(0 O
7.3 Homogeneous Catalysis in Liquid Solution 231
If one defines the following enzyme reaction A comparison of equations 7.3.43 and 7.3.38
parameters, shows that they are of the same mathematical
form. Both can be written in terms of four
(7.3.39) measurable kinetic constants in the manner of
equation 7.3.40. Only the relationship between
K.= the kinetic constants and the individual rate
constants differs. Thus, no distinction can be
the four kinetic constants (Vx max, F 4 m a x , X s , and made between the two mechanisms using steady-
Kp) can be determined from initial rate studies state rate studies. In general, the introduction
of the forward and reverse reactions. Equation of unimolecular steps involving only isomeriza-
7.3.38 can be rephrased in terms of these param- tion between unstable intermediate complexes
eters as does not change the form of the rate expression.
(7.3.40)
KsKp + Kp(S) + KJLP)
7.3.2.2 Inhibition Effects in Enzyme Catalyzed
At equilibrium the net reaction velocity must Reactions. Enzyme catalyzed reactions are
be zero. In terms of the enzymatic kinetic con- often retarded or inhibited by the presence of
stants, equation 7.3.40 then indicates that species that do not participate in the reaction in
question as well as by the products of the re-
(7.3.41) action. In some cases the reactants themselves
(S) can act as inhibitors. Inhibition usually results
where Keq is the thermodynamic equilibrium from the formation of various enzyme-inhibitor
constant for the overall reaction. Equation 7.3.41 complexes, a situation that decreases the amount
is known as the Haldane relation. It indicates of enzyme available for the normal reaction
that the enzymatic kinetic parameters are not sequence. The study of inhibition is important
all independent, but are constrained by the in the investigation of enzyme action. By deter-
thermodynamics of the overall reaction. mining what compounds behave as inhibitors
In the above example the enzyme combined and what type of kinetic patterns are followed,
with either the substrate or the product to form it may be possible to draw important conclu-
the same complex. This assumption is not realis- sions about the mechanism of an enzyme's
tic, but it still leads to the correct form of the action or the nature of its active site.
rate expression. Inhibitors may act reversibly or irreversibly,
The following mechanism for a reaction of but this classification is not particularly useful.
identical stoichiometry introduces a second It may even be misleading, because it suggests
complex into the sequence of elementary re- that reversible and irreversible inhibitors act in
actions. different ways when, in fact, both act by com-
bining with the enzyme to give inactive com-
E + S^ES^±EP^E + P (7.3.42)
k2 k4 ke plexes, but with quite different "dissociation
where ES and EP are enzyme-substrate and constants." The irreversible inhibitors give
enzyme-product complexes, respectively. complexes that have very small dissociation
For this mechanism, the steady-state rate constants; the reversible inhibitors have signifi-
equation is: cantly higher dissociation constants.
•V = (7.3.43)
(k2k5 + k2k4 + k3ks) ks)(S) + k6(k2
232 Liquid Phase Reactions
A much more useful classification of inhibitors These rules can be used in conjunction with
can be made on the basis of the mechanisms by experimental inhibition studies to assess the
which they act. Competitive inhibitors combine plausibility of possible enzymatic mechanisms.
with the enzyme at the same site as the substrate
does, thus blocking the first step in the sequence.
7.3.2.3 The Influence of Environmental Factors
Noncompetitive inhibitors combine with the
on Enzyme Kinetics. Because enzymes are pro-
enzyme at some other site to give a complex
teins, they are unusually sensitive to changes in
that can still combine with the substrate, but
their environment. This is true not only with
the resultant ternary complex is unreactive. Un-
regard to variations in inhibitor concentrations,
competitive inhibition results when the inhibitor
but also with respect to variations in pH and
and substrate combine with enzyme forms as in
temperature. Most enzymes are efficient cata-
the following mechanism.
lysts only within relatively narrow ranges of
E + P (7.3.44) pH and temperature.
When the rate of an enzyme catalyzed re-
EA + / ^ EAI no reaction (7.3.45) action is studied as a function of temperature,
k5 it is found that the rate passes through a maxi-
These three classes of inhibition can be dis- mum. The existence of an optimum temperature
tinguished by virtue of the effect of variations can be explained by considering the effect of
in inhibitor concentration on the slopes and temperature on the catalytic reaction itself and
intercepts of reciprocal plots. For competitive on the enzyme denaturation reaction. In the
inhibition only the slope varies. For uncompeti- low temperature range (around room tempera-
tive inhibition only the intercept varies, while ture) there is little denaturation, and increasing
for noncompetitive inhibition both the slope and the temperature increases the rate of the catalytic
the intercept vary. reaction in the usual manner. As the tempera-
If more than one substrate participates in an ture rises, deactivation arising from protein
enzymatic reaction, the kinetic effects of an denaturation becomes more and more impor-
inhibitor can be quite complex. In this case, tant, so the observed overall rate eventually will
rules formulated by Cleland (36) are useful in begin to fall off. At temperatures in excess of
gaining a qualitative picture of the inhibition 50 to 60 °C, most enzymes are completely
patterns to be expected of a given mechanism. denatured, and the observed rates are essentially
1. If the inhibitor combines with an enzyme zero.
form different from one with which the Enzyme activity generally passes through a
variable substrate combines, the vertical maximum as the pH of the system in question
intercept of the corresponding reciprocal is varied. However, the optimum pH varies with
plot will be affected. substrate concentration and temperature. Pro-
2. If the inhibitor combines with an enzyme vided that the pH is not changed too far from
form that is the same as or is connected by the optimum value corresponding to the maxi-
a series of reversible steps to the same form mum rate, the changes of rate with pH are
with which the variable substrate combines, reversible and reproducible. However, if the
the slope of the corresponding reciprocal plot solutions are made too acid or too alkaline, the
is affected. activity of the enzyme may be irreversibly
3. These effects can occur separately, in which destroyed. Irreversible deactivation is usually
case the inhibition is either competitive or attributed to denaturation of the proteinaceous
uncompetitive; or they can occur jointly, enzyme. The range of pH in which reversible
resulting in noncompetitive inhibition. behavior is observed is generally small and this
7.4 Correlation Methods for Kinetic Date—Linear Free Energy Relations 233
behavior is almost certainly due to changes in itself. The development and improvement in
the amounts and activities of the various ionic these quantitative correlations for reaction rate
forms of the enzyme, the substrate, and the and equilibrium constant data has been one of
enzyme complex. The maximum in the activity the most striking developments in the evolution
of the enzyme is a reflection of the maximum in of physical organic chemistry. In the area of
the concentration of the catalytically active kinetics these relationships constitute a suitable
species. framework for the extrapolation and interpola-
tion of rate data, and they can also provide a
7.4 CORRELATION METHODS FOR useful insight into the events that take place on
KINETIC DATA—LINEAR FREE reaction at the molecular level.
ENERGY RELATIONS It is instructive to consider the rationale
underlying the various linear free energy corre-
A primary objective of the practicing kineticist
lations and to indicate in qualitative fashion
is to be able to relate the rate of a reaction to
how substituents may influence reaction rates.
the structure of the reactants. Although the
The relation between an equilibrium constant
collision and transition state theories provide
and the standard free energy change accom-
useful frameworks for the discussion of the
panying a reaction is given by
microscopic events involved in chemical re-
actions, they do not in any meaningful way AG° = -RTlnKa (7.4.1)
permit one to predict reaction rates on an a
priori basis for even relatively simple reactions. Since the transition state formulation of a
However, chemical engineers should be aware reaction rate expression treats the activated
that methods exist for correlating empirical rate complex as being in equilibrium with the
data for homologous reactions. These correla- reactants, the resultant expression for the reac-
tions attempt to describe quantitatively the tion rate constant depends similarly on the free
influence of variations in chemical structure on energy difference between reactants and the
rate constants for a series of reactions involving activated complex. In this case equation 4.3.34
the same functional groups. This section briefly can be rewritten as
describes the most useful correlation method in kBT
order to indicate to beginning kineticists an AGX = -RT\ Ink - In
empirical approach which has given good results
(7.4.2)
in the past. The useful correlation methods are
based on what are referred to as linear free where the second logarithmic term is constant
energy relations. These relations presume that at a fixed temperature. Thus, estimates of
when a selected molecule undergoes reactions reaction rate and equilibrium constants may be
with two different homologous compounds, the regarded as equivalent to estimates of free
activation energy changes associated with the energy differences between different species.
rate processes will be influenced in a similar If one examines a series of reactions, differences
fashion by the changes in structure. In essence in £n k or In K are then simply related to
one treats the molecule as if its structure can be differences in the associated free energy changes.
arbitrarily broken up into a reaction center X Consider two molecular processes involving
and a nonreacting residue with structural ele- the reaction of structurally similar reactants
ments that can influence the rate of reaction at with a common reagent; for example,
X. Similar treatments are used in the correlation
of free energy changes associated with reaction Y1NXl + A ^ Y,NX2 + B (7.4.3)
equilibria and hence of the equilibrium constant Y2NXX + A k-X Y2NX2 + B (7.4.4)
234 Liquid Phase Reactions
where Y1 and Y2 are monovalent substituent energy of our representative Y1NX1 mole-
atoms or groups; Xx and X2 are monovalent cule will include terms due to the individual
atoms or groups that may be regarded as the yl5 JV, and Xx groups and terms arising from
reactant and product group, respectively; N is the interactions of these groups. For a represen-
the core of the molecule that links the substituent tative molecule the standard free energy under
groups to the reactant and product groups; these conditions may be written as
and A and B are other reactant and product
GN GYlN
molecules. A specific example of the above
pair of generalized reactions is the following. (7.4.8)
(7.4.13)
It is generally thought that the interaction lady true for those cases where N is a relatively
energies of groups that are not directly bonded rigid group, such as an aromatic ring. In this
to one another result from the following effects. case the various X and Y groups will be the
same distance apart in the species involved in
1. Inductive or polar effects. These effects in-
reactions 7.4.3 and 7.4.4. Furthermore, if the X
volve electron displacements that are trans-
and Y groups are sufficiently far apart that there
mitted along a chain of atoms without any
are no direct steric interactions and no direct
reorganization of the formal chemical bonds
resonance interactions between the X and Y
in the molecule. For example, the introduc-
groups, only the polar interactions contribute
tion of a methyl group in a pyridine ring
significantly to the GYXN terms. This situation
involves a displacement of electrons to the
occurs in many meta and para substituted ben-
nitrogen atom from the methyl group. This
zene derivatives. In this case GYXN can be as-
effect falls off rapidly with separation dis-
sumed to be proportional to the product of
tance.
parameters for the substituent group Y and the
2. Resonance or electromeric effects. Certain
reactant group X. These parameters are referred
molecular structures are characterized by the
to as polar substituent constants GYN and GXN.
possibility of having two or more compatible
The proportionality constant depends on the
electronic structures and the molecules exist
solvent, the temperature, and the nature of the
in a resonance state intermediate between the
core group N linking the substituents. It pro-
several extremes. These effects are particu-
vides a measure of how effectively the influence
larly characteristic of aromatic structures and
of one group is transmitted to the other. This
other molecules containing conjugate double
proportionality constant may be written as
bonds.
( — TNRT In 10), so that the interaction term
3. Steric effects. These effects arise as a con-
can be written as
sequence of the molecular geometry of the
species involved in the reaction. They include GYXN = -(TNRT In 10)(JYN(TXN (7.4.14)
236 Liquid Phase Reactions
Combining equations 7.4.13 and 7.4.14 then the second class permits one to obtain estimates
gives of the rate constants for reactions involving the
In I — I — (7.4.15)
\K2/
1 1 1 1 1 I
-4.0 -
m-\° N.
-4.2 p-ClV
op-F
- H
°\. p-Me
p-Eta3 p-MeO
-4.4 p-i-Pro \ —
p-t-Bu \
1 1 1 1 1 1
-3.6 -4.0 -4.4
log K' for benzoic acids
Figure 7.5
Log-log plot of ionization constants of benzoic and phenyl-
acetic acids in water at 25°. (From Physical Organic Chemistry
by J. S. Hine. Copyright © 1962. Used with permission of
McGraw-Hill Book Company.)
25 CC). The value of o may be determined from of a and p values are available (37, 41-45). Note
its definition in terms of the ionization constant that one requires different o values for meta and
if the appropriate benzoic acid derivative has para substituents, but that only one p value is
been measured. Such a values may then be used required. Typical a values range from about — 1
to determine p values for other reactions, and to +2, while p values range from —6 to +4.
these p values in turn lead to the possibility of Hence the Hammett equation may be used to
determining new er values. Equation 7.4.24 im- correlate data covering several orders of magni-
plies that a plot of log KA or log kA versus oA tude in the rate and equilibrium constant values.
should be linear for a given series of reactions Illustration 7.1 involves the use of Table 7.2 to
involving the same reactive groups. Extensive estimate a reaction rate constant.
experimental evidence attests to this relation. The substituent constants have also been as-
Table 7.2 contains some values that have been sociated with the ability of the substituent group
reported in the literature. Extensive tabulations to alter the charge density at the reaction site.
Table 7.2
Hammett Substituent Constants (46)
a 0
The groups with positive o values are regarded and those for another must exist if the Hammett
as electron withdrawers, while those with nega- equation is obeyed.
tive cr's refer to electron donor substituents. Re-
actions involving a transition state with a highly ILLUSTRATION 7.1 USE OF THE HAMMETT
electron deficient center would be expected to EQUATION FOR THE DETERMINATION OF
be sensitive to the stabilizing effect of substi- A REACTION RATE CONSTANT
tuents able to donate charge to the center. The
Kindler [Ann., 450(\\ 1926] has studied the
opposite situation exists for reactions involving
alkaline hydrolysis of the ethyl esters of a
a reaction center with an excess electron density.
number of substituted benzoic acids. The m-
In this case the stabilizing effect occurs in the
nitro compound was found to have a rate
presence of substituents that act to withdraw
constant 63.5 times as fast as the unsubstituted
charge from the center. Reactions with positive
compound. What relative rate constant is
p values are accelerated by electron withdrawal
predicted for the reaction of p-methoxybenzoate
from the ring (positive cr), while those with
by the Hammett equation? The value based on
negative p values are retarded by electron
experimental results is 0.214.
withdrawal.
In its original form the Hammett equation
Solution
was appropriate for use with para and meta
substituted compounds where the reaction site From Table 7.2 the o value for the m-nitro
is separated from the aromatic group by a non- group is 0.710. Substitution of this value and
conjugating side chain. Although there have the ratio of reaction rate constants into equation
been several extensions and modifications that 7.4.19 gives
permit the use of the Hammett equation beyond
these limitations, it is not appropriate for use log 63.5 = p(0.710)
with ortho substituted compounds, since steric or
effects are likely to be significant with such p = 2.54
species. The results obtained using free radical for the alkaline hydrolysis of ethyl benzoates.
reactions are often poor, and the correlation is For p-methoxy substitution, Table 7.2 indicates
more appropriate for use with ionic reactions. that a = -0.268. In this case equation 7.4.19
For a detailed discussion of the Hammett equa- becomes
tion and its extensions, consult the texts by
Hammett (37), Amis and Hinton (12), and - ) = 2.54(- -0.268) = -0.681
Johnson (47).
Before terminating our discussion of the
or
Hammett equation, we should note that the
existence of linear correlations of the type indi-
cated by equation 7.4.20 implies a linear free
K) - ° 209
energy relationship. The rate or equilibrium This value compares quite favorably with the
constants can be eliminated from this equation experimental results.
using equation 7.4.1; that is,
7.4.2 Other Correlations
- C (7.4.25)
2303RT 2.303RT
In order to correlate rate constant data for
Thus a linear relationship between the free aliphatic and ortho substituted aromatic com-
energies for one homologous series of reactions pounds, one must allow not only for the polar
240 Liquid Phase Reactions
effects correlated by the Hammett equation but 14. Eckert, C. A., Ann. Rev. Physical Chem., 23 (239), 1972.
also for resonance and steric effects. Taft (48-49) 15. Bell, R. P., "Acid-Base Catalysis," p. 3, Oxford Uni-
has shown that it is possible to extend the range versity Press, Oxford, copyright © 1941. Reprinted
with permission.
of linear free energy relations significantly by
assuming that the polar, steric, and resonance 16. Laidler, K. J., op. cit., 1965.
effects may be treated independently. Other 17. Bender, M. L., Mechanisms of Homogeneous Catalysis
from Protons to Proteins, Wiley Interscience, New York,
useful extensions of the Hammett equation in-
1971.
clude those of Swain and Scott (50-51), Edwards
18. Jencks, W. P., Catalysis in Chemistry and Enzymology,
(52-53), Grunwald and Winstein (54), and McGraw-Hill, New York, 1969.
Hansson (55). In order to obtain substituent
19. Ashmore, P. G., Catalysis and Inhibition of Chemical
values for the parameters appearing in the Reactions, Butterworths, London, 1963. ,
various equations, consult the text by Wells (41), 20. Frost, A. A., and Pearson, R. G., Kinetics and Mecha-
the original references, or texts in physical nism, Second Edition, Wiley, New York, 1961.
organic chemistry. The practicing design engi- 21. Moelwyn-Hughes, E. A., The Chemical Statics and
neer should be aware of these and other cor- Kinetics of Solutions, Academic Press, New York, 1971.
relations in order to minimize the experimental 22. Bell, R. P., "Acid-Base Catalysis," Oxford University
work necessary to generate required kinetic Press, Oxford, 1941.
parameters. 23. Bell, R. P., The Proton in Chemistry, Methuen, London,
1959.
LITERATURE CITATIONS 24. Laidler, K. J., The Chemical Kinetics of Enzyme Action,
1. Parker, A. J., Adv. Phys. Org. Chem., 5 (192), 1967. Clarendon Press, Oxford, 1958.
2. Laidler, K. J., Reaction Kinetics, Volume 2, Reactions 25. Plowman, K. M., Enzyme Kinetics, McGraw-Hill, New
in Solution, p. 2, Pergamon Press, Oxford, 1963. York, 1972.
3. North, A. M., The Collision Theory of Chemical Re- 26. Gould, R. F., Editor, "Homogeneous Catalysis: Indus-
actions in Liquids, Methuen, London, 1964. trial Applications and Implications," ACS Adv. in
4. Rabinowitch, E., and Wood, W. C , Trans. Faraday Chem., 70, Washington, 1968.
Soc.,32(\m), 1936. 27. Schrauzer, G. N., Transition Metals in Homogeneous
5. Franck, J., and Rabinowitch, E., Trans. Faraday Soc, Catalysis, Marcel Dekker, New York, 1971.
30(120), 1934. 28. Jones, M. M., Ligand Reactivity and Catalysis, Aca-
6. Debye, P., and Hiickel, E., Phys. Z., 24 (305), 1923. demic Press, New York, 1968.
7. Debye, P., and Hiickel, E., ibid., 25 (145), 1924. 29. Basolo, F., and Pearson, R. G., "Mechanisms of In-
organic Reactions," Wiley, New York, 1960.
8. Debye, P., and Hiickel, E., Trans. Faraday Soc, 23 (334),
1927. 30. Ashmore, P. G., op. cit., pp. 30-31.
9. Harned, H. S., and Owen, B. B., The Physical Chemistry 31. Hine, J. S., Physical Organic Chemistry, Second Edition,
of Electrolytic Solutions, Third Edition, Reinhold McGraw-Hill, 1962.
Publishing, New York, 1958. 32. Skrabal, A., Z. Elektrochem., 33 (322), 1927.
10. Laidler, K. J., Chemical Kinetics, p. 221, McGraw-Hill, 33. Jencks, W. P., Catalysis in Chemistry and Enzymology,
New York, 1965. McGraw-Hill, New York, 1969.
11. Amis, E. S., Solvent Effects on Reaction Rates and 34. Bender, M. L., Mechanisms of Homogeneous Catalysis
Mechanisms, Academic Press, New York, 1966. from Protons to Proteins, Wiley Interscience, New York,
12. Amis, E. S., and Hinton, J. F., Solvent Effects on 1971.
Chemical Phenomena, Academic Press, New York, 1973. 35. Michaelis, L., and Menten, M. L., Biochem. Z., 49
13. Clark, D. and Wayne, R. P., in Comprehensive Chemical (333), 1913.
Kinetics, Volume 2, The Theory of Kinetics, pp. 302-376, 36. Cleland, W. W., Biochim. Biophys. Ada, (57(188), 1963.
Elsevier, Amsterdam, 1969. Adapted with permission.
Problems 241
37. Hammett, L. P., Physical Organic Chemistry, Second The influence of ionic strength on the two rate
Edition, McGraw-Hill, New York, 1970. constants was noted to be as follows.
38. Hammett, L. P., Chem. Rev., 77(125), 1935.
39. Hammett, L. P., Trans. Faraday Soc, 34 (156), 1938.
40. Hine, J. S., op. cit., p. 85. jj, (kmoles/m 3 ) (m 3 /kmoles-sec) (m 6 /kmoles 2 -sec)
41. Wells, P. R., Linear Free Energy Relationships, Aca-
demic Press, New York, 1968. 0.000 0.658 19.0
42. Gordon, A. J., and Ford, R. A., The Chemist's Com- 0.0207 0.663 15.0
panion, pp. 145-149, Wiley, New York, 1972. 0.0525 0.670 12.2
0.0925 0.679 11.3
43. Jaffe, H. H., Chem. Rev., 53 (191), 1953.
0.1575 0.694 9.7
44. McDaniel, D. H., and Brown, H. C , J. Org. Chem., 23
0.2025 0.705 9.2
(420), 1958.
45. Leffler, J. E., and Grunwald, E., Rates and Equilibria of
Organic Reactions, Wiley, New York, 1963. Are these data consistent with the results
46. Hine, J. S., op. cit., p. 87. predicted by equation 7.1.4?
47. Johnson, C. D., The Hammett Equation, Cambridge
University Press, London, 1973.
2. The reaction of the nitrourethane ion with
48. Taft, R. W., Jr., J. Am. Chem. Soc, 74 (2729, 3120),
1952; 75(4231), 1953.
hydroxide ions can be written as
49. Taft, R. W., Jr., in Steric Effects in Organic Chemistry, NO 2 =NCOOC 2 H 5 " + OH" ->
M. S. Newman, Editor, Chapter 13, Wiley, New York,
1956.
N2O + C2H5OH
50. Swain, C. G., and Scott, C. B., J. Am. Chem. Soc, 75 Near room temperature the reaction is essen-
(141), 1953. tially irreversible and second-order. In the
51. Swain, C. G., Mosely, R. B., and Brown, D. E., J. Am. limit of zero ionic strength at 293 °K, k =
Chem. Soc, 77(3731), 1955. 2.12m3/kmole • ksec. Determine the initial reac-
52. Edwards, J. O., J. Am. Chem. Soc, 76 (1540), 1954. tion rate in a solution that is 0.05 kmole/m3 each
53. Edwards, J. O., J. Am. Chem. Soc, 78 (1819), 1956. in potassium nitrourethanate, NH4OH, and
54. Grunwald, E., and Winstein, S., J. Am. Chem. Soc, 70 KC1. The ionization constant for NH 4 OH at
(846), 1948. 293 °K is 1.7 x 10"5 kmoles/m3. The effect of
55. Hansson, J., Svensk, Kern. Tidskr., 66 (351), 1954. the slight ionization of NH 4 OH on ionic
strength and the variation of activity coefficients
PROBLEMS with temperature between 293 and 298 °K may
be neglected.
1. The oxidation of iodide ions by hydrogen
peroxide 3. The following reaction takes place in aqueous
solution.
H 2 O 2 + 2H + + 21 2H2O + I2
[Co(NH3)5Br] + + + OH" -»
has been studied by Bell et al. \_J. Phys. Chem., [Co(NH3)5OH] + + + Br-
55 (874), 1951]. At 25 °C the reaction appears to it may be regarded as bimolecular and irre-
proceed by two parallel paths so that the ob-
versible. Determine the ratio of the reaction rate
served rate expression is of the form
in a system initially containing 10 moles/m3
r = k1(H2O2)(r) + /c2(H2O2)(I-)(H+) [Co(NH3)5Br](NO3)2,100 moles/m3 NH4OH,
242 Liquid Phase Reactions
14.7 9.298
Since the spontaneous reaction term dominates
4000 11.13 in the pH range 4 to 6, studies of the reaction
8000 13.05 in this range are particularly suitable for mea-
12000 15.28 suring the small catalytic effects of weak acids
16000 17.94 and weak bases. If one employs the more gen-
eral expression for k involving contributions
from the undissociated acid and the anion re-
What is the activation volume for this sulting from dissociation, determine the coeffi-
reaction? cients of these terms from the data below.
5. Bronsted and Guggenheim [J. Am. Chem.
Soc, 49 (2554), 1927] have studied the mutaro- Hydrogen
tation of glucose as catalyzed by acids and bases. Sodium Propionic Ion
The reaction takes place slowly in pure water, propionate acid concentration /cxlO 3
is weakly catalyzed by hydrogen ions, and is normality normality (kmoles/m3) (min" 1 )
strongly catalyzed by hydroxide ions. When 0.010 5.65
0.010
strong acids and bases are employed as cata- 0.040 0.020 6.53
lysts, the apparent first-order rate constants can 0.050 0.050 6.81
be written as 0.075 0.075 7.57
0.100 0.100 8.21
0.125 0.125 9.15
where kOH/kH+ is of the order of several thou- 0.150 0.150 9.85
sands. In the pH range 4 to 6, the contribution 0.040 0.020 0.00001 6.53
of the spontaneous reaction dominates, since 0.040 0.060 0.00002 6.60
both catalytic terms are negligible in this region. 0.040 0.110 0.00006 6.76
In dilute solutions the primary salt effect is not 0.040 0.160 1.00010 6.77
applicable to solutions of strong acids and bases.
On the basis of the data below determine the 6. Bell and Baughan \_J Chem. Soc. 1937, 1947]
parameters k0 and kH+ at 18 °C: have investigated the generalized acid-base
Problems 243
catalysis of the depolymerization of dimeric 7. The kinetic data below were reported for an
dihydroxyacetone. In terms of the general enzyme catalyzed reaction of the type E + S ^
formulation of the first-order rate constant ES ^± E + P. Since the data pertain to initial
reaction rates, the reverse reaction may be
fcOH (OH-) neglected. Use a graphical method to determine
+ /cA(A) + kB(B) the Michaelis constant and Vmax for this system
where k0 is the rate constant for the water at the enzyme concentration employed.
catalyzed reaction and kx is the catalytic
constant for species x. This equation can also be Initial substrate Initial rate
written as concentration (M) (jumoles/liter-min)
k = k" + OL(B)
where 2 x 10" 3 150
2 x 10" 4 149.8
k" = k0 2 x 10~5 120
and 1.5 x 10" 5 112.5
1.25 x 10~6 30.0
glucoamylase and an immobilized version there- In both cases the initial substrate concen-
of. Their goal was to obtain the necessary data tration was 0.14 kmole/m 3 . In cases A and B
to be able to design a commercial facility for the Michaelis constants (Km) are reported to be
dextrose production. Their studies were carried 1.15 x 10" 3 and 1.5 x 10" 3 kmoles/m 3 , re-
out in a batch reactor at 60 °C. Compare the spectively. The rate constant fc3 is equal to
data below with that predicted from a Michaelis- 1.25 /mioles/m3-sec-unit in both cases.
Menten model with a rate expression of the If the data are not consistent, provide
form r = fc3(£0)(S)/[Km + (S)] plausible explanations for the discrepancy.
(a) Conventional enzyme (Eo = 11,600 units) 10. Gould has reported the following acidity
constants (in H 2 O at 25 °C) for some substituted
Yield benzene seleninic acids ArSe O 2 H.
Time, t (sec) (mg dextrose produced/cm3)
From Mechanism and
Substituent K x 105
0 12.0 Structure in Organic
900 40.0 Chemistry by E. S.
None 1.6
1800 76.5 Gould. Copyright ©
p-MeO— 0.89
3600 120.0 1959. Used with per-
m-MeO— 2.2
5400 151.2
3.5
mission of McGraw-
m-Cl
7200 155.7 Hill Book Company.
m-NO 2 8.5
9000 164.9
p-Br 3.2
p-C6H5O- 1.3
(b) Immobilized enzyme data (Eo = 46,400
units)
(a) Show that the equilibria for the reaction
Yield
series
Time, t (sec) (mg dextrose produced/cm3) ArSeO 2 H + H 2 O ^ ArSeO2~ + H 3 O +
0 18.4 are governed by the Hammett equation and
1,800 200.0 calculate the p value for this series. The
3,600 260.0 dissociation constant for benzoic acid is
5,400 262.0 6.3 x 10" 5 .
7,200 278.0 (b) Calculate the dissociation constant for p-
9,000 310.0 nitrobenzeneseleninic acid and compare it
13,500 316.0
with the observed value of 1.0 x 10~ 4 .
18,900 320.0
24,900 320.0
(c) What does the above data indicate is the
o value for the p-C 6 H 5 Q—substituent?
8 Basic Concepts in Reactor Design
and Ideal Reactor Models
8.0 INTRODUCTION 3. Is a catalyst necessary or desirable? If a
The chemical reactor must be regarded as the catalyst is employed, what are the ramifi-
very heart of a chemical process. It is the piece cations with respect to product distribution,
of equipment in which conversion of feedstock operating conditions, most desirable type
to desired product takes place and is thus the of reactor, process economics, and other
single irreplaceable component of the process. pertinent questions raised below?
Several different factors must be considered in 4. What operating conditions (temperature,
selecting the physical configuration and mode pressure, degree of agitation, etc.) are re-
of operation to be used to accomplish a specified quired for most economic operation?
task. This chapter discusses the fundamental 5. Is it necessary or desirable to add inerts or
principles that the chemical engineer engaged other materials to the feedstock to enhance
in the practice of reactor design utilizes in yields of desired products, to moderate
making these selections. thermal effects, or to prolong the useful life
of any catalysts that may be employed?
8.0.1 The Nature of the Reactor 6. Should the process be continuous or inter-
Design Problem mittent? Would batch or semibatch opera-
tion be advantageous?
The chemical engineer is required to choose
7. What type of reactor best meets the process
the reactor configuration and mode of operation
requirements? Are there advantages asso-
that yields the greatest profit consistent with
ciated with the use of a combination of
market constraints for the raw material and
reactor types, or with multiple reactors in
product costs, capital and operating costs,
parallel or series?
safety considerations, pollution control require-
ments, and esthetic constraints that may be 8. What size and shape reactor(s) should be
imposed by management, society, or labor used?
unions. Usually there are many combinations 9. How are the energy transfer requirements
of operating conditions and reactor size and/or for the process best accomplished? Should
type that will meet the requirements imposed one operate isothermally, adiabatically, or
by nature in terms of the reaction rate expres- in accord with an alternative temperature
sion involved and those imposed by manage- protocol?
ment in terms of the required production 10. Is single pass operation best, or is recycle
capacity. The engineer is thus faced with the needed to achieve the desired degree of
task of maintaining a careful balance between conversion of the raw feedstock?
analytical reasoning expressed in quantitative 11. What facilities are required for catalyst
terms and sound engineering judgment. In an supply, activation, and regeneration?
attempt to maintain this balance, some or all of 12. What are the reactor effluent composition
the following questions must be answered. and conditions? Are any chemical separa-
tion steps or physical operations required
1. What is the composition of the feedstock in order to bring the effluent to a point
and under what conditions is it available? where it is satisfactory for the desired end
Are any purification procedures necessary? use?
2. What is the scale of the process? What 13. Are there any special materials requirements
capacity is required? imposed by the process conditions? Are the
245
246 Basic Concepts in Reactor Design and Ideal Reactor Models
process fluids corrosive? Are extremely high uses essentially the reverse of the design
temperatures or pressures required? procedures developed later in this chapter
to determine the effective reaction rate from
The remainder of this text attempts to estab-
the pilot plant data. It may be difficult to
lish a rational framework within which many
separate the intrinsic chemical effects from
of these questions can be attacked. We will see
any heat and mass transfer effects in the
that there is often considerable freedom of
analysis of data of this type.
choice available in terms of the type of reactor
3. Operating data from commercial scale reac-
and reaction conditions that will accomplish a
tors. If one's company has access to actual
given task. The development of an optimum
operating data on another commercial in-
processing scheme or even of an optimum
stallation of the same type as that contem-
reactor configuration and mode of operation
plated, it provides the closest approximation
requires a number of complex calculations that
to the conditions likely to be encountered in
often involve iterative numerical calculations.
industrial practice. Such access may result
Consequently machine computation is used
from licensing arrangements or from previous
extensively in industrial situations to simplify
experience within the company. Unfortunate-
the optimization task. Nonetheless, we have
ly, such data are often incomplete or in-
deliberately chosen to present the concepts used
accurate, and the problems of backing the
in reactor design calculations in a framework
intrinsic chemical kinetics out of the mass
that insofar as possible permits analytical solu-
of data may be insurmountable. In such
tions in order to divorce the basic concepts
systems physical limitations on rates of heat
from the mass of detail associated with machine
and mass transfer may disguise the true ki-
computation.
netics to a significant degree.
The first stage of a logical design procedure
involves the determination of a reaction rate
In the design of an industrial scale reactor for
expression that is appropriate for the range of
a new process, or an old one that employs a
conditions to be investigated in the design
new catalyst, it is common practice to carry out
analysis. One requires a knowledge of the
both bench and pilot plant studies before
dependence of the rate on composition, tem-
finalizing the design of the commercial scale
perature, fluid velocity, the characteristic di-
reactor. The bench scale studies yield the best
mensions of any heterogeneous phases present,
information about the intrinsic chemical kinetics
and any other process variables that may be
and the associated rate expression. However,
significant. There are several potential sources
when taken alone, they force the chemical
of the experimental data that are essential for
engineer to rely on standard empirical corre-
proper reactor design.
lations and prediction methods in order to
1. Bench scale experiments. The reactors used determine the possible influence of heat and
in these experiments are usually designed to mass transfer processes on the rates that will be
operate at constant temperature, under con- observed in industrial scale equipment. The
ditions that minimize heat and mass transfer pilot scale studies can provide a test of the
limitations on reaction rates. This facilitates applicability of the correlations and an indica-
an accurate evaluation of the intrinsic chem- tion of potential limitations that physical pro-
ical effects. cesses may place on conversion rates. These
2. Pilot plant studies. The reactors used in these pilot plant studies can provide extremely useful
studies are significantly larger than those in information on the temperature distribution in
bench scale laboratory experiments. One the reactor and on contacting patterns when
8.0 Introduction 247
more than a single phase reactant-catalyst operated in semibatch fashion. In this mode the
system is employed. tank is partially filled with reactant(s), and
additional reactants are added progressively
8.0.2 Reactor Types until the desired end composition is achieved.
Alternatively, one may charge the reactants all
In terms of the physical configurations en- at once and continuously remove products as
countered, there are basically only two types of they are formed. In the continuous flow mode
reactors: the tank and the tube. of operation the stirred tank reactor is conti-
The ideal tank reactor is one in which stirring nuously supplied with feed; at the same time an
is so efficient that the contents are always uni- equal volume of reactor contents is discharged
form in composition and temperature through- in order to maintain a constant level in the
out. The simple tank reactor may be operated tank. The composition of the effluent stream is
in a variety of modes: batch, semibatch, and identical with that of the fluid remaining in the
continuous flow. These modes are illustrated tank.
schematically in Figure 8.1. In the simple batch The ideal tubular reactor is one in which
reactor the fluid elements will all have the same elements of the homogeneous fluid reactant
composition, but the composition will be time move through a tube as plugs moving parallel to
dependent. The stirred tank reactor may also be the tube axis. This flow pattern is referred to as
plug flow or piston flow. The velocity profile at materials handling costs involved in filling,
a given cross section is flat and it is assumed emptying, and cleaning of these reactors. While
that there is no axial diffusion or back-mixing batch reactors are being filled, emptied, or
of fluid elements. cleaned, and while the reactor contents are
Batch reactors are often used for liquid phase being heated to the reaction temperature or
reactions, particularly when the required pro- cooled to a point suitable for discharge, batch
duction is small. They are seldom employed on reactors are not producing reaction products.
a commercial scale for gas-phase reactions The sum of the nonproductive periods may
because the quantity of product that can be often be comparable in length to the time
produced in reasonably sized reactors is small. necessary to carry out the reaction. In deter-
Batch reactors are well suited for producing mining long-term production capacities for
small quantities of material or for producing batch reactors, these dead times must be taken
several different products from one piece of into account.
equipment. Consequently they find extensive Continuous flow reactors are almost in-
use in the pharmaceutical and dyestuff industries variably preferred to batch reactors when the
and in the production of certain specialty processing capacity required is large. Although
chemicals where suchflexibilityis desired. When the capital investment requirements will be
rapid fouling is encountered or contamination higher, the operating costs per unit of product
of fermentation cultures is to be avoided, batch will be lower for continuous operation than for
operation is preferable to continuous processing batch reaction. The advantages of continuous
because it facilitates the necessary cleaning and operation are that it:
sanitation procedures.
When the specified production capacities are 1. Facilitates good quality control for the
low, processes based on batch reactors will product through the provision of greater
usually have lower capital investment require- constancy in reaction conditions.
ments than processes calling for continuous 2. Facilitates automatic process control.
operation, so batch reactors are often preferred 3. Minimizes the labor costs per unit of product.
for new and untried processes during the
initial stages of development. As production Often the decision to select a batch or con-
requirements increase in response to market tinuous processing mode involves a determi-
demands, it may become more economic to nation of the relative contributions of capital
shift to continuous processing but, even in these and operating expenses to total process costs
cases, there are many industrial situations where for the proposed level of capacity. As Denbigh
batch operation is preferable. This is particularly (1) points out, what is best for a highly indus-
true when the operating expenses associated trialized country with high labor costs is not
with the reactor are a minor fraction of total necessarily best for a lesser developed country.
product cost. At low production capacities, In many cases selectivity considerations deter-
construction and instrumentation requirements mine the processing mode, particularly when
for batch reactors are usually cheaper than for the reaction under study is accompanied by
continuous process equipment. Moreover, it is undesirable side reactions. The yield of the
generally easier to start up, shut down, and desired product may differ considerably between
control a batch reactor than a comparable batch and continuous operation and between
capacity continuous flow reactor. the two primary types of continuous processes.
The disadvantages associated with the use of When the yield is lower for a continuous
a batch reactor include the high labor and process, this factor may be so important in the
8.0 Introduction 249
overall process economics as to require the use fluid is noncorrosive and readily and safely
of a batch reactor. pumped, it may be preferable to employ an
At this point we wish to turn to a brief dis- external heat exchanger and circulation pump.
cussion of the types of batch and flow reactors Good temperature control can be achieved
used in industrial practice for carrying out using an external reflux condenser for cases
homogeneous fluid phase reactions. Treatment where appreciable vapor is given off. The selec-
of heterogeneous catalytic reactors is deferred tion of either internal or external heat transfer
to Chapter 12. equipment is governed by the required area, the
susceptibility of the heat transfer surface to
8.0.2.1 Batch Reactors (Stirred Tanks). Batch fouling, the temperature and pressure require-
reactors are usually cylindrical tanks and the ments imposed by the heat transfer medium,
orientation of such tanks is usually vertical. and the potential adverse effects that might
Cylindrical vessels are employed because they occur if the process fluid and the heat transfer
are easier to fabricate and clean than other medium come in direct contact through leakage.
geometries and because the construction costs For high-pressure operation, safety considera-
for high-pressure units are considerably less tions are extremely important and care must be
than for alternative configurations. For simple taken to ensure proper mechanical design. Clo-
stirred vertical batch reactors, the depth of sures must be designed to withstand the same
liquid is usually comparable to the diameter of maximum pressure as the rest of the autoclave.
the reactor. For greater liquid height to diameter Various authors have treated the problems in-
ratios more complex agitation equipment is volved in designing medium- and high-pressure
necessary. Agitation can be supplied by stirrer batch reactors (2-4).
blades of various shapes or by forced circulation
with an external or built-in pump. Where more 8.0.2.2 Continuous Flow Reactors—Stirred
gas-liquid interfacial area is required for evap- Tanks. The continuous flow stirred tank re-
oration or gas absorption, or where it is neces- actor is used extensively in chemical process
sary to minimize the hydrostatic head (e.g., to industries. Both single tanks and batteries of
minimize the boiling point rise), horizontal tanks connected in series are used. In many
reactors will be used. The latter orientation may respects the mechanical and heat transfer aspects
also be preferable when the reactor contents are of these reactors closely resemble the stirred
quite viscous or take the form of a slurry. Batch tank batch reactors treated in the previous sub-
reactors may be fabricated from ordinary or section. However, in the present case, one must
stainless steel, but there are often advantages to also provide for continuous addition of re-
using glass or polymer coatings on interior actants and continuous withdrawal of the
surfaces in order to minimize corrosion or product stream.
sanitation problems. It is possible to employ either multiple in-
Because of the large energy effects that often dividual tanks in series or units containing
accompany chemical reaction, it is usually neces- multiple stages within a single shell (see Figure
sary to provide for heat transfer to or from the 8.2). Multiple tanks are more expensive, but
reactor contents. Heating or cooling may be provide more flexibility in use, since they are
accomplished using jacketed walls, internal coils, more readily altered if process requirements
or internal tubes filled with a heat transfer fluid change. In order to minimize pump require-
that is circulated through an external heat ex- ments and maintenance, one often chooses to
changer. Energy may also be supplied by elec- allow for gravity flow between stages. When
trical heating or direct firing. If the process the reactants are of limited miscibility, but differ
250 Basic Concepts in Reactor Design and Ideal Reactor Models
Reactants
Products
L
w L -. Products
L Reactants Products
Reactants 3L
7 /
(a) (b) (c)
Figure 8.2
Types of staged reactors, (a) Reactor battery, (b) Vertically staged, (c) Compartmented. (Adapted
from Reaction Kinetics for Chemical Engineers by S. M. Walas. Copyright © 1959. Used with permission
of McGraw-Hill Book Company.)
in density, the vertical staged shell lends itself sure in tanks constructed of inexpensive mate-
to countercurrent operation. This approach is rials such as mild steel.
useful when dealing with reversible reactions In addition to lower construction costs, a
between immiscible fluids. continuous flow stirred tank reactor possesses
For purposes of calculation, each stage in a other advantages relative to a tubular flow
multiple-stage unit is treated as an individual reactor, such as the facilitation of temperature
reactor. The process stream flows from one control. Efficient stirring of the reactor contents
reactor to the next, so that there is essentially insures uniform temperature and the elimination
a step change of composition between successive of local hot spots. The large heat capacity of the
reactors. The step change is a direct consequence reactor contents also acts as a heat sink to
of efficient mixing. Unless the fluid phase is moderate temperature excursions when changes
highly viscous, it is not difficult to approach occur in process conditions. The physical con-
perfect mixing in industrial scale equipment. All figuration of cylindrical tanks provides a large
that is required is that the time necessary to heat transfer area on the external surface of the
distribute an entering element of fluid uniformly tank and permits augmentation of this area
throughout the tank be very small compared to through the use of submerged coils within the
the average residence time in the tank. tank. However, the rate of heat transfer per unit
Because of the dilution that results from the volume of reaction mixture is generally lower in
mixing of entering fluid elements with the re- a conventional stirred tank reactor than in
actor contents, the average reaction rate in a conventional tubular reactors because of the
stirred tank reactor will usually be less than it lower ratio of heat transfer surface area to
would be in a tubular reactor of equal volume volume in the tank reactor. Consequently,
and temperature supplied with an identical feed tubular reactors are preferred for fast reactions
stream. Consequently, in order to achieve the when energy transfer requirements are very
same production capacity and conversion level, large.
a continuous flow stirred tank reactor or even Ease of access to the interior surface of stirred
a battery of several stirred tank reactors must tanks is an additional advantage of this type of
be much larger than a tubular reactor. In many reactor. This consideration is particularly signi-
cases, however, the greater volume requirement ficant in polymerization reactors, where one
is a relatively unimportant economic factor, needs to worry about periodic cleaning of
particularly when one operates at ambient pres- internal surfaces.
8.0 Introduction 251
Selectivity considerations may also dictate The single-jacketed tube reactor is the sim-
the use of stirred tank reactors. They are pre- plest type of tubular reactor to conceptualize
ferred if undesirable side reactions predominate and to fabricate. It may be used only when the
at high reactant concentrations, and they are heat transfer requirements are minimal because
also useful when one desires to "skip" certain of the low surface area to volume ratio charac-
concentration or temperature ranges where by- teristic of these reactors.
product formation may be excessive. When the shell-and-tube configuration is
Stirred tank reactors are employed when it utilized, the reaction may take place on either
is necessary to handle gas bubbles, solids, or a the tube side or the shell side. The shell-and-tube
second liquid suspended in a continuous liquid tubular reactor has a much greater area for
phase. One often finds that the rates of such heat transfer per unit of effective reactor volume
reactions are strongly dependent on the degree than the single-jacketed tube. Consequently, it
of dispersion of the second phase, which in turn may be used for reactions where the energy
depends on the level of agitation. transfer requirements are large. On occasion
Large stirred tank reactors are generally not the reaction zone may be packed with granular
suited for use at high pressures because of solids to promote increased turbulence or better
mechanical strength limitations. They are used contacting of heterogeneous fluid phases or to
mainly for liquid phase reaction systems at low act as a thermal sink to facilitate control of the
or medium pressures when appreciable residence reactor. In many cases energy economies can be
times are required. achieved using countercurrent flow of a hot
product stream to preheat an incoming reactant
8.0.2.3 Continuous Flow Reactors—Tubular stream to the temperature where the reaction
Reactors. The tubular reactor is so named occurs at an appreciable rate. Two commercial
because the physical configuration of the re- scale processes that employ this technique are
actor is normally such that the reaction takes the synthesis of ammonia from its elements and
place within a tube or length of pipe. The the oxidation of sulfur dioxide to sulfur trioxide.
idealized model of this type of reactor assumes Tubular furnaces are used only when it is
that an entering fluid element moves through the necessary to carry out endothermic reactions at
reactor as a plug of material that completely fairly high temperatures on very large quantities
fills the reactor cross section. Thus the terms of feedstock. Thermal reforming reactions and
piston flow or plug flow reactor (PFR) are often other reactions used to increase the yield of
employed to describe the idealized model. The gasoline from petroleum-based feedstocks are
contents of a given elemental plug are presumed commercial scale processes that employ this
to be uniform in temperature and composition. type of reactor. A tubular furnace is basically
This model may be used to treat both the case a combustion chamber with reactor tubes
where the tube is packed with a solid catalyst mounted on its walls and ceiling. Tube dimen-
(see Section 12.1) and the case where the fluid sions are typically 3 to 6 in. in diameter with
phase alone is present. lengths ranging from 20 to 40 ft. As many as
The majority of tubular reactors may be several hundred tubes may be used, with either
classified in terms of three major categories: series or parallel connections possible depending
1. Single-jacketed tubes. on the required residence time.
2. Shell-and-tube heat exchangers. Because there is no back-mixing of fluid ele-
3. Tube furnaces, in which the tubes are exposed ments along the direction of flow in a tubular
to thermal radiation and heat transfer from reactor, there is a continuous gradient in reactant
combustion gases. concentration in this direction. One does not
252 Basic Concepts in Reactor Design and Ideal Reactor Models
encounter the step changes characteristic of mul- of the reactants is fed. This point is demonstrated
tiple stirred tank reactors. Consequently, for the quite dramatically in Illustration 10.7. This mode
same feed composition and reaction tempera- of operation is also desirable when high reactant
ture, the average reaction rate will generally be concentrations favor the formation of undesir-
significantly higher in a plug flow reactor than it able side products or when one of the reactants
would be in a single stirred tank or a battery of is a gas of limited solubility.
stirred tanks with a total volume equal to that of Another mode of semibatch operation in-
the tubular reactor. The more efficient utilization volves the use of a purge stream to remove con-
of reactor volume is an advantage of the tubular tinuously one or more of the products of a
reactor that permits one to use it in processes reversible reaction. For example, water may be
which demand very large capacity. Because vari- removed in esterification reactions by the use of
ations in temperature and composition may a purge stream or by distillation of the reacting
occur in the axial direction in tubular reactors, mixture. Continuous removal of product(s) in-
these systems may be somewhat more difficult to creases the net reaction rate by slowing down
control than continuous flow stirred tank re- the reverse reaction.
actors. However, the problems are usually not Semibatch or semiflow processes are among
insurmountable, and one can normally obtain the most difficult to analyze from the viewpoint
steady-state operating conditions which give rise of reactor design because one must deal with an
to uniform product quality. open system under nonsteady-state conditions.
Other advantages of the tubular reactor rela- Hence the differential equations governing en-
tive to stirred tanks include suitability for use at ergy and mass conservation are more complex
higher pressures and temperatures, and the fact than they would be for the same reaction carried
that severe energy transfer constraints may be out batchwise or in a continuous flow reactor
readily surmounted using this configuration. The operating at steady state.
tubular reactor is usually employed for liquid
phase reactions when relatively short residence 8.0.3 Fundamental Concepts Used in
times are needed to effect the desired chemical Chemical Reactor Design
transformation. It is the reactor of choice for The bread and butter tools of the practicing
continuous gas phase operations. chemical engineer are the material balance and
the energy balance. In many respects chemical
8.0.2.4 Semibatch or Semiflow Reactors. Semi- reactor design can be regarded as a straight-
batch or semiflow operations usually take place forward application of these fundamental prin-
in a single stirred tank using equipment ex- ciples. This section indicates in general terms
tremely similar to that described for batch opera- how these principles are applied to the various
tions. Figure 8.1 indicates some of the many types of idealized reactor models.
modes in which semibatch reactors may be
operated. 8.0.3.1 Material and Energy Balances in the
One common mode of operation involves Design oj Industrial Reactors. The analysis of
loading some of the reactants into a stirred tank chemical reactors in terms of material and en-
as a single charge and then feeding in the re- ergy balances differs from the analysis of other
maining material gradually. This mode of opera- process equipment in that one must take into
tion is advantageous when large heat effects account the rate at which molecular species are
accompany the reaction. Exothermic reactions converted from one chemical form to another
may be slowed down and temperature control and the rate at which energy is transformed by
maintained by regulating the rate at which one the process. When combined with material and
8.0 Introduction 253
energy balances on the reactor, the reaction rate tinuous flow reactors operating at steady state,
expression provides a means of determining the the accumulation term is omitted. However, for
production rate and the composition of the pro- the analysis of continuous flow reactors under
ducts as functions of time. Both steady-state and transient conditions and for semibatch reactors
time varying situations may be analyzed using it may be necessary to retain all four terms. For
the same fundamental relations. Differences in ideal well-stirred reactors the composition and
the analyses result from the retention of different temperature are uniform throughout the reactor
terms in the basic balance equations. and all volume elements are identical. Hence the
A material balance on a reactant species of material balance may be written over the entire
interest for an element of volume AF can be reactor in these cases. For tubular flow reactors
written as: the composition is not independent of position
For completeness, the terms corresponding to 8.0.3.2 Vocabulary of Terms Used in Reactor
the entry of material to the volume element and Design. There are several terms that will be
exit therefrom must contain, in addition to the used extensively throughout the remainder of
ordinary enthalpy of the material, its kinetic and this text that deserve definition or comment. The
potential energy. However, for virtually all cases concepts involved include steady-state and tran-
of interest in chemical reactor design, only the sient operation, heterogeneous and homoge-
enthalpy term is significant. Since only changes neous reaction systems, adiabatic and isothermal
in internal energy or enthalpy can be evaluated, operation, mean residence time, contacting and
the datum conditions for the first, third, and holding time, and space time and space velocity.
fourth terms must be identical in order to achieve Each of these concepts will be discussed in
an internally consistent equation. Although heat turn.
interactions in chemical reactors are significant, Large-scale industrial reactions are almost in-
shaft work effects are usually negligible. The variably carried out on a continuous basis with
chemical reaction rate does not appear explicitly reactants entering at one end of the reactor net-
in equation 8.0.3, but its effects are implicit in all work and products leaving at the other. Usually
terms except the second. The first, third, and such systems are designed for steady-state opera-
fourth terms reflect differences in temperature tion but, even during the design of such systems,
and/or in composition of the entering and leav- adequate care must be made to provide for the
ing streams. The energy effects associated with transient condition that will invariably be in-
composition changes are a direct reflection of curred during start-up and shut-down periods.
the enthalpy change associated with the reac- By the term steady-state operation we imply that
tion. conditions at any point in the reactor are time-
There are a variety of limiting forms of equa- invariant. Changes in composition occur in the
tion 8.0.3 that are appropriate for use with differ- spatial dimension instead of in a time dimension.
ent types of reactors and different modes of It should be emphasized that operation at
operation. For stirred tanks the reactor contents steady state does not imply equilibrium. The last
are uniform in temperature and composition term is restricted to isolated systems that un-
throughout, and it is possible to write the energy dergo no net change with time. Insofar as the
balance over the entire reactor. In the case of a analysis of continuous flow reactors is con-
batch reactor, only the first two terms need be cerned, the major thrust of this and succeeding
retained. For continuous flow systems operating chapters involves steady-state operation. The
at steady state, the accumulation term disap- basic principles described in the previous sub-
pears. For adiabatic operation in the absence of section remain valid for the analysis of transient
shaft work effects the energy transfer term is systems. However, in most transient cases it is
omitted. For the case of semibatch operation it necessary to resort to numerical solutions in
may be necessary to retain all four terms. For order to predict the response of a continuous
tubular flow reactors neither the composition flow reactor network to changes in operating
nor the temperature need be independent of conditions. Batch reactors are inherently un-
position, and the energy balance must be written steady-state systems, even when the reactor con-
on a differential element of reactor volume. The tents are uniform throughout.
resultant differential equation must then be Both homogeneous and heterogeneous reac-
solved in conjunction with the differential equa- tion systems are frequently encountered in com-
tion describing the material balance on the differ- mercial practice. The term homogeneous reaction
ential element. system is restricted in this text to fluid systems in
8.0 Introduction 255
which the system properties vary continuously Although the concept of "mean residence
from point to point within the reactor. The term time" is easily visualized in terms of the average
embraces both catalytic and noncatalytic reac- time necessary to cover the distance between
tions, but it requires that any catalysts be uni- reactor inlet and outlet, it is not the most funda-
formly dispersed throughout the fluid phase. The mental characteristic time parameter for pur-
term heterogeneous reaction system refers to poses of reactor design. A more useful concept is
systems in which there are two or more phases that of the reactor space time. For continuous
involved in the reaction process, either as re- flow reactors the space time (T) is defined as the
actants or as catalysts. ratio of the reactor volume (VR) to a charac-
Adiabatic operation implies that there is no teristic volumetric flow rate of fluid (Y°).
heat interaction between the reactor contents
and their surroundings. Isothermal operation r = (8.0.4)
implies that the feed stream, the reactor contents,
and the effluent stream are equal in temperature The reactor volume is taken as the volume of the
and have a uniform temperature throughout. reactor physically occupied by the reacting fluids.
The present chapter is devoted to the analysis of It does not include the volume occupied by agita-
such systems. Adiabatic and other forms of non- tion devices, heat exchange equipment, or head-
isothermal systems are treated in Chapter 10. room above liquids. One may arbitrarily select
The terms holding time, contact time, and resi- the temperature, pressure, and even the state of
dence time are often used in discussions of the aggregation (gas or liquid) at which the volumet-
performance of chemical reactors. As employed ric flow rate to the reactor will be measured. For
by reactor designers, these terms are essentially design calculations it is usually convenient to
interchangeable. They refer to the length of time choose the reference conditions as those that
that an element of process fluid spends in the prevail at the the inlet to the reactor. However, it
reactor in question. For a batch reactor it is is easy to convert to any other basis if the
usually assumed that no reaction occurs while pressure-volume-temperature behavior of the
the reaction vessel is being filled or emptied or system is known. Since the reference volumetric
while its contents are brought up to an ignition flow rate is arbitrary, care must be taken to
temperature. The holding time is thus the time specify precisely the reference conditions in
necessary to "cook" the contents to the point order to allow for proper interpretation of the
where the desired degree of conversion is resultant space time. Unless an explicit state-
achieved. The terms "contact time" and "mean ment is made to the contrary, we will choose our
residence time" are used primarily in discussions reference state as that prevailing at the reactor
of continuous flow processes. They represent the inlet and emphasize this choice by the use of the
average length of time that it takes a fluid ele- subscript zero. Henceforth,
ment to travel from the reactor inlet to the re-
actor outlet. For plug flow reactors, all fluid
elements will have the same residence time. T = (8.0.5)
However, for stirred tank reactors or other re-
actors in which mixing effects are significant, where % is the volumetric flow rate at the inlet
there will be a spread of residence times for temperature and pressure and a fraction con-
the different fluid elements. This has important version of zero.
implications for the conversions that will be Using this convention, a space time of 30 min
achieved in such reactors. means that every 30 min one reactor volume of
256 Basic Concepts in Reactor Design and Ideal Reactor Models
feed (measured at inlet conditions) is processed after the nature of the changes occurring within
by the reactor. One reactor volume enters and the reactor is known.
one reactor volume leaves, but this statement The reciprocal of the space time is known as
does not imply that we have simply displaced the the space velocity (S).
original charge from the reactor. Some or all of
the original contents may leave, and some of the 5 = 1= (8.0.6)
fresh charge may leave as well. In the latter T
circumstance the reactor contents become a mix-
ture of the original contents and new material. Like the definition of the space time, the defini-
The space time is not necessarily equal to the tion of the space velocity involves the volumetric
average residence time of an element of fluid in flow rate of the reactant stream measured at
the reactor. Variations in the number of moles some reference condition. A space velocity of
on reaction as well as variations in temperature 10 hr" 1 implies that every hour, 10 reactor
and pressure can cause the volumetric flow rate volumes of feed can be processed.
at arbitrary points in the reactor to differ appre- When dealing with reactions where a liquid
ciably from that corresponding to inlet condi- feed must be vaporized prior to being fed to the
tions. Consequently, even though the reference reactor proper, one must state very clearly
conditions may be taken as those prevailing at whether the space velocity is based on the vol-
the reactor inlet, the space time need not be equal umetric flow rate of the feed as a liquid or as a
to the mean residence time of the fluid. The two gas. Unless an explicit statement to the contrary
quantities are equal only if all of the following is made, the term space velocity in this text will
conditions are met. refer to the ratio of the volumetric flow rate
evaluated at reactor inlet conditions td the reactor
1. Pressure and temperature are constant volume.
throughout the reactor. The term space velocity has somewhat differ-
2. The density of the reaction mixture is inde- ent connotations when dealing with heteroge-
pendent of the extent of reaction. For gas neous catalytic reactors. In this case it denotes
the ratio of the mass flow rate of feed to the mass
phase reactions this requirement implies that
of catalyst used (W).
there can be no change in the number of
moles on reaction. In terms of equation 3.1.45,
we require that 3 = 0. WHSV = (8.0.7)
3. The reference volumetric flow rate is eval-
uated at reactor inlet conditions. where p is the mass density of the feed and
WHSV is termed the weight hourly space velocity.
When the space time and the mean residence Sometimes the term volumetric hourly space
time differ, it is the space time that should be velocity (VHSV) is used to denote the ratio of
regarded as the independent process variable the volumetric flow rate of a gaseous feed to the
that is directly related to the constraints imposed weight of the catalyst bed.
on the system. We will see in Sections 8.2 and 8.3
that it is convenient to express the fundamental
VHSV = (8.08)
design relations for continuous flow reactors in w
terms of this parameter. We will also see that for
these reactors the mean residence time cannot be On other occasions the volume of catalyst in-
considered as an independent variable, but that stead of the mass of catalyst may be used in the
it is a parameter that can be determined only denominator. The units associated with a parti-
8.1 Design Analysis for the Batch Reactor 257
The accumulation term is just the time deriva- There are a number of limiting forms of equa-
tive of the number of moles of reactant A con- tion 8.1.5 that should be mentioned briefly. If
tained within the reactor (dNJdt). This term also h =0andfAi = 0,
may be written in terms of either the extent of
reaction (£) or the fraction conversion of the dfA
t = N AO
Jo ( _ A)VR (8.1.6)
limiting reagent (fA). {A is presumed to be the (~r
258 Basic Concepts in Reactor Design and Ideal Reactor Models
If, in addition, the reactor volume (fluid density) The use of equations like this was treated in
is constant, equation 8.1.6 becomes great detail in Chapter 3. There our primary
dfAA _ _ rcAA dCAA objective was the determination of the mathe-
[SA
. _ r matical form of the reaction rate expression
° (~rA)~ Jc*o(-rA from data on the extent of reaction or fraction
(8.1.7) conversion versus time. At present our objective
For reactions where the fluid volume varies is just the reverse: to determine the time nec-
linearly with the fraction conversion, as indi- essary to achieve a given degree of conversion
cated by equation 3.1.40, equation 8.1.6 becomes using our knowledge of the mathematical form
of the reaction rate expression and the reaction
t = N
dfA conditions. Although it is often more convenient
'Jo (Z^r SAfA) to work in terms of extent of reaction when
dfA analyzing rate data, it is usually more conve-
= CA (8.1.8) nient to work in terms of fraction conversion in
SAfA)
analyzing design problems. The two concepts
where VR0 is the volume occupied by the reacting are simply related, and the chemical engineer
fluid at zero fraction conversion. This equation should learn to work in terms of either with
would be appropriate for use when low-pressure equal facility.
gas phase reactions involving a change in the The degree of difficulty associated with
number of moles on reaction take place in a evaluating the integral in any of the batch
batch reactor at constant pressure. However, gas reactor design equations (equations 8.1.5 to
phase reactions are rarely carried out batchwise 8.1.10) depends on the composition and tem-
on a commercial scale because the quantity of perature dependence of the reaction rate expres-
product that can be produced in a reasonably sion. For nonisothermal systems an energy
sized reactor is so small. The chief use of batch balance must be employed to relate the system
reactors for gas phase reactions is to obtain the temperature (and through it the reaction rate)
data necessary for the design of continuous flow to the fraction conversion. For isothermal
reactors. systems it is not necessary to utilize an energy
For simple nth-order kinetics where volumet- balance to determine the holding time necessary
ric expansion effects may be significant, equation to achieve a given fraction conversion. One
8.1.8 becomes merely substitutes rate constants evaluated at
t = C AO r
Jo
dfA the temperature in question directly into the
rate expression. However, an energy balance
must be used to determine the heat transfer
requirements necessary to maintain isothermal
conditions. The reader should recognize that
=r
Jo
(8.1.9) even if a simple closed form relation between
time and fraction conversion cannot be ob-
Equations 8.1.4 to 8.1.8 may also be written tained, it is still possible to evaluate the integral
in terms of the extent of reaction (£) or the graphically or numerically using standard meth-
extent per unit volume (£*). In terms of £ the ods to assure convergence. The latter approach
most general design relation (equation 8.1.5) is invariably necessary when departures from
becomes isothermal conditions are large.
(8.1.10) The following illustrations indicate how the
basic design relations developed above are used
8.1 Design Analysis for the Batch Reactor 259
to answer the two questions with which the Since the reaction takes place at constant
reactor designer is most often faced. volume, the pertinent design equation is equa-
tion 8.1.7.
1. What is the time required for converting a
quantity of material to the desired level under dfB
t = CBO
specified reaction conditions? - fB)(CC0 - fBCB0)
2. What reactor volume is required to achieve
a given production rate? Since the reaction occurs under isothermal
conditions, the rate constant may be taken
outside the integral sign. Integration and sim-
ILLUSTRATION 8.1 DETERMINATION OF
plification then gives
HOLDING TIME REQUIREMENTS FOR THE
FORMATION OF A DIELS-ALDER ADDUCT
Wassermann (6) has studied the Diels-Alder In
reaction of benzoquinone (B) and cyclopen-
t =
tadiene (C) at 25 °C. k(C{CO CBO)
.CH.
xo s
CH CH CH
CH 9
CH CH CH
CO CH'
adduct r = kCBCc
silicate and sodium aluminate. Kerr (7) and determine the time necessary to achieve 98%
Liu (8) have studied an alternative method of conversion of the substrate to zeolite in a
synthesis from amorphous sodium alumino- well-stirred batch reactor. The conditions to
silicate substrate and aqueous sodium hydroxide be considered are isothermal operation at
solution. The reaction (essentially a crystalliza- 100 °C and a hydroxide concentration of
tion or recrystallization process) can be viewed 1.5 kmoles/m3.
as 2. Assuming the reaction conditions noted in
Liu's description of the kinetics of the zeolite part 1, determine the reactor size and total
formation process can be formulated in terms weight of charge necessary to produce zeolite
of the following equation. A at an average rate of 2000 kg/day. Only
dCA _ /c2[OH ] XCAQZ ,A^ one reactor is to be used, and it will be
{ necessary to shut down for 1.8 ksec between
~~df ~ k3Cz + (k3 + \)CA ' batches for removal of product, cleaning,
where and start-up. The zeolite to be recycled to
[ O H ] = hydroxide ion concentration the reactor will come from the 2000 kg
(kmoles/m3) produced daily.
Cz = concentration of zeolite crystals Solution
(kg/m3)
Since the reaction takes place in slurry form,
CA = concentration of amorphous
the reactor volume may be regarded as constant,
substrate (kg/m3)
and equation 8.1.7 is appropriate for use.
and ku k29 and k3 are kinetic constants. Since
CSA dfA
the total weight of solids in a batch reactor h= (B)
^AO
Jo T~r
(~rA)
must satisfy an overall material balance, we
require that the quantity (CA + Cz) be a con- where the subscript A refers to the amorphous
stant equal to the total weight of the charge substrate.
divided by the effective reactor volume. Note The principles of stoichiometry may be used
that the reaction is catalyzed by the presence to write the reaction rate in terms of the fraction
of product in that the zeolite concentration conversion. The desired conversion level is
appears in the numerator of the rate expression. expressed in terms of the initial substrate level.
Consequently, it is desirable to include some Thus CA0 = 24 kg/m 3 . At any time the instan-
zeolite in the feed to the reactor to enhance the taneous concentration of substrate can be
reaction rate. Reactions of this type are labeled written as
as "autocatalytic." They are treated in more CA = CA0(l - fA) (C)
detail in Section 9.4. while the corresponding concentration of zeolite
At 100 CC the following values of the rate is given by
constants are appropriate for use. Cz = C z o + fACM (D)
fct = 2.36 k2 = 0.625 ksec" 1 k3 = 0.36 The instantaneous reaction rate can be expressed
1. If one utilizes a slurry containing 1 kg/m 3 of in terms of the fraction conversion by combining
zeolite and 24 kg/m 3 of amorphous substrate, equations A, C, and D.
8.2 Design of Tubular Reactors 261
Substitution of numerical values into this relation The weight of amorphous solids per batch
gives must equal the product of the reactor volume
_ (0.625)(1.5)2-36(24)(l - fA)(l + 24/J and the solids concentration. Thus
'A 0.36(1 + 24fA) + (1.36)(24)(1 - fA) 126 kg
= 5.25 m 3 _ ^ 1387 gal
_ (39.05K1 - fA)(l + 24/,) 24 kg/nr
33.0 - 24fA (E) Note that in this case the batch reactor is
Combining equations E and B gives operating in the production mode only two
thirds of the time. This situation is not extra-
(33.0 - 24fA) dfA ordinary and allowances must be made for
t = 24
39.05(1 - / , ) ( ! + 24/,) downtime in order to meet design requirements
properly.
This integral may be broken up into terms that
can be evaluated using standard tables.
When there are significant heating and cooling
t = 20.28 Jo 4
p -i — dfA periods associated with the utilization of a
_ L)(\ + 24/J batch reactor, one must take these circumstances
into account in design calculations. If only one
- 14.75 Jop fAdfA
reaction is involved, the holding time computed
(1 - + 24/.) using the methods of this section will be a
or conservative estimate of the length of time the
20.28 1 + 24/A reactor contents should be held at the specified
temperature. However, if undesirable side reac-
1- / J tions take place at temperatures below the
14.75 operating temperature, the magnitude of the
i«n(l +24/J adverse effects must be considered. If they are
significant it may be necessary to resort to
The time required to achieve any desired rapid quenching techniques or to separate pre-
degree of conversion may be calculated from heating of coreactants to the desired tem-
this expression. For fA = 0.98: t = 3.54 ksec = perature.
0.982 hr.
For part 2, for 98% conversion, each batch 8.2 DESIGN OF TUBULAR REACTORS
will require a holding time of 3.54 ksec and a
downtime of 1.80 ksec for emptying, cleaning, Tubular reactors are normally used in the
and start-up. The total time consumed in chemical industry for extremely large-scale
processing one batch is thus 5.34 ksec. In order processes. When filled with solid catalyst parti-
to produce 2000 kg of zeolite A per day we cles, such reactors are referred to as fixed or
require that each batch contain an amount of packed bed reactors. This section treats general
amorphous solids equal to design relationships for tubular reactors in
which isothermal homogeneous reactions take in the real world, we should keep in mind that
place. Nonisothermal tubular reactors are the model can only reflect the idealizations
treated in Section 10.4 and packed bed reactors built into it and that deviations from ideal
in Section 12.7. behavior can fall into three categories.
8.2.1 The Plug Flow Reactor (PFR)—Basic 1. There will be velocity gradients in the radial
Assumptions and Design Equations direction so all fluid elements will not have
the same residence time in the reactor.
The simplest model of the behavior of tubular
Under turbulent flow conditions in reactors
reactors is the plug flow model. The essential
with large length to diameter ratios, any
features of this idealized model require that
disparities between observed values and
there be no longitudinal mixing of fluid elements
model predictions arising from this factor
as they move through the reactor and that all
should be small. For short reactors and/or
fluid elements take the same length of time to
laminar flow conditions the disparities can
move from the reactor inlet to the outlet. The
be appreciable. Some of the techniques used
model may also be labeled the slug flow or
in the analysis of isothermal tubular reactors
piston flow model in that it may be convenient
that deviate from plug flow are treated in
to picture the reaction as taking place within
Chapter 11.
differentially thin slugs of fluid that fill the
2. There will be an interchange of material
entire cross section of the tube and that are
between fluid elements at different axial
separated from one another by hypothetical
positions by virtue of ordinary molecular
pistons that prevent axial mixing. These plugs
diffusion and eddy diffusion processes arising
of material move as units through the reactor,
from turbulence and/or the influence of any
and this assumption is conveniently expressed
packing in the bed. Convective mixing
in terms of a requirement that the velocity
arising from thermal gradients in the reac-
profile be flat as one traverses the tube diameter.
tor may also contribute to the exchange of
Each plug of fluid is assumed to be uniform in
matter between different fluid elements.
temperature, composition, and pressure, which
3. There may be radial temperature gradients
is equivalent to assuming that radial mixing is
in the reactor that arise from the interaction
infinitely rapid. However, there may well be
between the energy released by reaction,
variations in composition, temperature, pres-
heat transfer through the walls of the tube,
sure, and fluid velocity as one moves in the
and convective transport of energy. This
longitudinal direction. With respect to these
factor is the greatest potential source of
variations, however, the model requires that
disparities between the predictions of the
mass transport via diffusion or turbulent mixing
model and what is observed for real systems.
processes be negligible and that the plugs of
The deviations are most significant in non-
material not interact with one another except
isothermal packed bed reactors.
for transmission of the hydrodynamic forces
giving rise to the fluid motion. Some of the The tubular flow reactor is a convenient
aforementioned requirements may be removed means of approaching the performance charac-
in more complex mathematical models of teristics of a batch reactor on a continuous
tubular reactors. In the present chapter, how- basis, since the distance-pressure-temperature
ever, we will limit our discussion to the simplest history of the various plugs as they flow through
possible model—the plug flow model. None- the reactor corresponds to the time-pressure-
theless, when we attempt to compare the results temperature protocol that is used in a batch
predicted by the model with what we observe reactor. Although this analogy is often useful,
8.2 Design of Tubular Reactors 263
Rate of disappearance
/ Rate of flow \ /Rate of flow of \
of reactant by chemical
of reactant into = reactant out of | (8.2.1)
reactions within the
\ volume element / \ volume element /
volume element
264 Basic Concepts in Reactor Design and Ideal Reactor Models
AO
a given segment of the reactor must be taken into The following set of mechanistic equations is
account in arriving at the reactor volume re- consistent with their experimental results:
quired to accomplish a given task. No new
principles are involved; one merely breaks the A1C13 + M -> A1C13M (fast) (A)
reactor up into a series of segments that can be
assumed to operate at an appropriate average B + A1C13-M k-i OAICI3 (slow) (B)
pressure. The pressure drop across a segment is C-A1C13 -• C + AICI3 (fast) (C)
calculated from the Bernoulli equation using
appropriate empirical relations where necessary where reaction B is the rate limiting step.
to estimate the friction factor, equivalent lengths (A1C13-M) and (C-A1C13) represent complexes
for u-bends or other fittings, and losses asso- formed between the dissolved A1C13 and the
ciated with changes in tube diameter. The basic species in question. The concentration of the
design equation is then applied to each segment first reactive complex is essentially constant as
using the average of the inlet and outlet pressures long as enough methyl acrylate remains in solu-
in the rate expression. One marches through the tion to regenerate (A1C13*M) efficiently. The con-
reactor assuming conversion increments that centration of methyl acrylate in excess of A1C13
are sufficiently small that calculations allowing will not affect the rate of reaction during the
for the pressure drop lead to convergence of the early stages of reaction but, as reaction proceeds
integral. The case study in Section 13.1 indicates and the methyl acrylate concentration drops
one possible approach to this problem. below the initial A1C13 concentration, the
Illustrations 8.3 and 8.4 indicate the applica- amount of complex present is limited by the
tion of the above analysis to isothermal tubular amount of methyl acrylate remaining.
reactors with negligible pressure drop. These investigators report that the second-
order rate constant for reaction B is equal to
ILLUSTRATION 8.3 DETERMINATION OF 1.15 x 10" 3 m3/mole-ksec at 20 °C. Determine
REQUIRED PLUG FLOW REACTOR the volume of plug flow reactor that would be
VOLUME UNDER ISOTHERMAL necessary to achieve 40% conversion of the input
OPERATING CONDITIONS—CONSTANT butadiene assuming isothermal operating condi-
DENSITY CASE tions and a liquid feed rate of 0.500 m 3 /ksec. The
feed composition is as follows.
Inukai and Kojima (9) have studied the alumi-
num chloride catalyzed diene condensation of Butadiene {B) 96.5 moles/m3
butadiene and methyl acrylate in benzene solu- Methyl acrylate (M) 184 moles/m3
tion. The stoichiometry for this Diels-Alder AICI3 6.63 moles/m3
reaction is
CH 2 O
|| o
CH C—OCH 3 CH, C—OCH 3
|
CH V CH CH HC
CH 2 CH 2 CH CH 2
\\ /
CH 2
butadiene (B) methyl acrylate (M) -+ adduct (C)
266 Basic Concepts in Reactor Design and Ideal Reactor Models
since the volume change accompanying liquid Below 565 °C the pyrolysis reaction is essen-
phase reactions is negligible. Equation 8.2.9 may tially first order with a rate constant given by
be used as the basic design relationship.
k = 7.8 x l O 9 * " 1 9 2 2 0 ' 7 ^ " 1
dfB
- rB° JO 7Z rB) where T is expressed in degrees Kelvin.
If one desires to design a pilot scale tubular
- C
= CBO r-
Jo ,
dfB
0
reactor to operate isothermally at 500 °C, what
length of 6-in. pipe will be required to convert
Since the quantities CA1Cb 0 and k are constant 90% of the raw feedstock to methyl acrylate?
for the conditions cited, equation E may be The feedstock enters at 5 atm at a flow rate of
integrated to give 500 lb/hr. Ideal gas behavior may be assumed.
A 6-in. pipe has an area of 0.0388 ft2 available
for flow. Pressure drop across the reactor may be
1-/* neglected.
T =
VR = xr0 = 67.0(0.500) = 33.5 m 3 For the case where the feed is pure methyl
acetoxypropionate,
The space time and reactor volume required
to accomplish the specified conversion in a plug
flow reactor are sufficiently high that they make
8.2 Design of Tubular Reactors 267
Since there is a change in the number of moles of From the definition of the space time,
gaseous species on reaction, the concentration
VR = = (0.193)(29.9) = 5.78 ft3
corresponding to a given fraction conversion is
From geometric considerations the required
1 -h (B) length of 6-in. pipe is equal to (5.78 ft3)/(3.88 x
- Wo
10~ 2 ft 2 )or 149 ft.
Combining equations A and B gives
Illustrations 8.3 and 8.4 indicate how the plug
(\ + fA) dfA flow design equations may be applied to homo-
T =
- fA) geneous fluid phase reactions. We now wish to
consider the form of the plug flow reactor design
Since the reaction takes place isothermally at
equation for heterogeneous catalytic reactions.
500 °C, the reaction rate constant may be moved
For reactions of this type the reaction rate must
outside the integral sign. At this temperature it is
be expressed per unit weight or per unit area of
equal to 0.124 sec" l . Thus
catalyst. Since the two are related through the
specific surface area of the material, we will
develop relations only in terms of the former.
Consider the segment of a tubular reactor shown
Evaluation of the integrals gives in Figure 8.3 in which we now presume that the
differential volume element dVR contains an
0.124i = - - fA) - fA - €n(l - fA) amount of catalyst dW. The rate of disappear-
- fA) - fA ance of reactant A within the differential volume
F o r / , =0.90, element is then equal to { — rAm) dW. A material
balance for steady-state operating conditions
T = [ - 2 £n(l - 0.9) - 0.900]/0.124 = 29.9 sec based on equation 8.2.1 gives
If we had erred and not included the effect of Input Output Disappearance by reaction
volumetric expansion on reaction, we would
FA =FA + dFA + (-rAm)dW
have calculated a space time of 18.6 sec, which
would have been off by 38%, and this error would (8.2.11)
propagate through the remainder of the calcula- Introduction of the fraction conversion through
tions below. equation 8.2.5 leads to the following analog of
If the gas behaves ideally the volumetric flow equation 8.2.7.
rate at the reactor inlet is given by the product
of the molal flow rate [(500/146) lb moles/hr] dfA (8.2.12)
and the molal volume at the pressure and tem- {-rA
perature in question. The latter may be calcu-
In this case the reaction rate will depend not
lated by correcting the standard molal volume
only on the system temperature and pressure but
(359 ft3/lb mole) for variations in temperature
also on the properties of the catalyst. It should be
and pressure between the reactor inlet and stan-
noted that the reaction rate term must include
dard conditions. Hence
the effects of external and intraparticle heat and
mass transfer limitations on the rate. Chapter 12
treats these subjects and indicates how equation
8.2.12 can be used in the analysis of packed bed
= 696 ft3/hr o= ft3/sec reactors.
268 Basic Concepts in Reactor Design and Ideal Reactor Models
8.2.2 Residence Times in Plug space time is the more meaningful for reactor
Flow Reactors design purposes. A knowledge of mean residence
time 1 does not permit us to determine the re-
For plug flow reactors all fluid elements take the
quired reactor volume, but a knowledge of the
same length of time to travel from the reactor
space time i readily leads to this quantity. The
inlet to the reactor outlet. This time is the mean
space time is an independent variable directly
residence time 7.
related to system parameters, whereas the mean
Consider the general case of a reaction ac-
residence time is a dependent variable found by
companied by a volumetric expansion or con-
integration of equation 8.2.17 or by tracer
traction. The time necessary for a plug to travel
studies (see Chapter 11).
from inlet to outlet of a tubular reactor is given
by ILLUSTRATION 8.5 DETERMINATION OF
MEAN RESIDENCE TIME IN A PLUG FLOW
t = (8.2.13)
REACTOR UNDER ISOTHERMAL
OPERATING CONDITIONS—VARIABLE
where V is the volumetric flow rate. This param-
DENSITY CASE
eter can be written in terms of the inlet reference
volumetric flow rate (%), the fraction conver- Consider the plug flow reactor used for the
sion, and the volumetric expansion parameter. pyrolysis of methyl acetoxypropionate in Illus-
TT = ^ ( 1 + dAfA) (8.2.14) tration 8.4.
The increment in reactor volume can be written C 6 H 10 O 4 C2H4O2 C4H6O2
in terms of the fraction conversion using equation A B C
8.2.6. For isothermal operation at 500 °C and 5 atrn,
17
>dfA it was shown that the space time required to
dVR = (8.2.15)
(~rA) achieve 90% conversion was 29.9 sec. Compare
this value with the mean residence time of the
Combining equations 8.2.13 to 8.2.15 gives
material in the plug flow reactor.
dfA
1= (8.2.16) Solution
For first-order kinetics in a system where SA is
which can be rewritten in terms of the inlet
nonzero, the reaction rate expression is of the
reference concentration of the limiting reagent as
form
dfA (1
1 = CAO (8.2.17) _ -
r kc -M
JfA in
+dAfA)
This equation is the basic relation for the mean In the present case 8A = 1. Hence the equation
residence time in a plug flow reactor with arbi- for the mean residence time (equation 8.2.17)
trary reaction kinetics. Note that this expression becomes
differs from that for the space time (equation (*0.9
8.2.9) by the inclusion of the term (1 + SAfA) and _
that this term appears inside the integral sign.
t — Jo kC AO - fA) SAfA)
The two quantities become identical only when
dA is zero (i.e., the fluid density is constant). The Simplification gives
differences between the two characteristic times 0.9
ro.9 dfA -£n(l -fA)
may be quite substantial, as we will see in
Illustration 8.5. Of the two quantities, the reactor r 0
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 269
dfA
, total - F P2 r J
P ]
~\ (8.2.19)
- A Jj-A~- A)J
From the principles of calculus, the quantity in Continuous flow stirred tank reactors are
brackets can be rewritten as a single integral. normally just what the name implies—tanks into
Hence which reactants flow and from which a product
stream is removed on a continuous basis.
CFSTR, CSTR, C-star, and back-mix reactor
Thus j plug flow reactors in series with a total are only a few of the names applied to the
volume VRtioial give the same conversion as a idealized stirred tank flow reactor. We will use
single reactor of volume T^>total. the letters CSTR as a shorthand notation in this
When plug flow reactors are connected in textbook. The virtues of a stirred tank reactor
parallel, the most efficient utilization of the total lie in its simplicity of construction and the rela-
reactor volume occurs when mixing of streams tive ease with which it may be controlled. These
of differing compositions does not occur. Con- reactors are used primarily for carrying out
sequently the feed rates to different parallel legs liquid phase reactions in the organic chemicals
270 Basic Concepts in Reactor Design and Ideal Reactor Models
industry, particularly for systems that are cha- molecules have short residence times, there is a
racterized by relatively slow reaction rates. If it rapid response at the reactor outlet to changes
is imperative that a gas phase reaction be carried in the reactor feed stream. This characteristic
out under efficient mixing conditions similar to facilitates automatic control of the reactor.
those found in a stirred tank reactor, one may Because the mixing process makes the prop-
employ a tubular reactor containing a recycle erties of the entire reactor contents equal to
loop. At sufficiently high recycle rates, such those of the effluent stream, there will be a step
systems approximate CSTR behavior. This sec- change in fluid composition and temperature at
tion is concerned with the development of design the point where the feedstream enters the reactor.
equations that are appropriate for use with the In the present chapter we will restrict the dis-
idealized stirred tank model. cussion to cases where there is no temperature
change on entrance to the reactor unless other-
8.3.1 Individual Stirred Tank Reactors wise noted. For these isothermal CSTR's the
8.3.1.1 Basic Assumptions and Design Relation- drop in reactant concentration as the fluid
ships. The most important feature of the enters the reactor implies that in the vast
CSTR is its mixing characteristics. The ideal- majority of cases the volume average reaction
ized model of reactor performance presumes rate will be low by comparison to that in an
that the reactor contents are perfectly mixed so isothermal plug flow reactor. Consequently,
that the system properties are uniform through- when both operate at the same temperature, a
out. The effluent composition and temperature CSTR will have to be significantly larger than
are thus identical with those of the reactor con- a PFR in order to effect the same composition
tents. This feature greatly simplifies the analysis change in a given amount of fluid. (Exceptions
of stirred tank reactors vis a vis tubular reactors occur in the case of autocatalytic reactions.)
for both isothermal and nonisothermal opera- However, by using a battery of stirred tanks in
tion. It is not difficult to obtain a good approxi- series, differences in total volume requirements
mation to CSTR behavior, provided that the can be significantly reduced. Moreover, be-
fluid phase is not too viscous. The approxima- cause of the simplicity of their construction,
tion to CSTR behavior is valid if the time stirred tank reactors normally cost much less
necessary to disperse an entering element of per unit of volume than tubular reactors. Thus,
fluid (e.g., a shot of dye or radioactive tracer) in many cases, it is more economic to employ a
uniformly throughout the tank is very much Targe stirred tank reactor or a battery of such
shorter than the average residence time in the reactors than it is to use a tubular reactor. Size
tank. comparisons of these two types can be treated
Unlike the situation in a plug flow reactor, the quantitatively, and we will return to this subject
various fluid elements mix with one another in in Sections 8.3.1.3 and 8.3.2.3. First, however, we
a CSTR. In the limit of perfect mixing, a tracer must develop the basic design equations.
molecule that enters at the reactor inlet has Consider the schematic representation of a
equal probability of being anywhere in the continuous flow stirred tank reactor shown in
vessel after an infinitesimally small time incre- Figure 8.5. The starting point for the develop-
ment. Thus all fluid elements in the reactor have ment of the fundamental design equation is again
equal probability of leaving in the next time a generalized material balance on a reactant
increment. Consequently there will be a broad species. For the steady-state case the accumula-
distribution of residence times for various tracer tion term in equation 8.0.1 is zero. Furthermore,
molecules. The character of the distribution is since conditions are uniform throughout the
discussed in Section 11.1. Because some of the reactor volume, the material balance may be
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 271
F
A in = l n l e t molal flow rate
[A in = Fraction conversion
of limiting reagent
at inlet
CA in = Inlet concentration
of limiting reagent
Figure 8.5
Schematic representation of CSTR indicating process variables.
In terms of the symbols indicated in Figure 8.5, The last equation may be rewritten in terms
of the fraction conversion as
^ m = FAout + (-rAF)VR (8.3.2)
where we again emphasize that the appropriate ^ o ( l - L m ) = ^ o ( l - L o u t ) + (~rAF)VR
volume is that physically occupied by the re- (8.3.3)
acting fluid. The quantity (— rAF) is the rate of
disappearance of reactant A evaluated at reactor where FA0 is again the molal flow rate corre-
outlet conditions. sponding to zero conversion. Rearrangement
272 Basic Concepts in Reactor Design and Ideal Reactor Models
In this situation the mean residence time and the For most of the commonly encountered types
reactor space time become identical. of reaction rate expressions the rate decreases
monotonically with increasing fraction con-
8.3.1.3 Relative Size Requirements for an In- version. The fact that a step change in composi-
dividual Continuous Stirred Tank Reactor and a tion occurs as a fluid enters an ideal CSTR
Plug Flow Reactor. In the development of the
final reactor design for a proposed production
requirement, the chemical engineer must con-
sider a variety of reactor types and modes of
operation. Several factors must be considered
in the development of the final design, some of Path followed in
which may be only peripherally related to the plug flow reactor
implies that for these cases the reaction rate will the area under a plot of the reciprocal reaction
be much smaller in this type of reactor than it rate ( — l/rA) versus fraction conversion was
would be in a tubular reactor being used to equal to the ratio VR/FA0 for a plug flow reactor.
accomplish the same composition change. Con- In the case of a CSTR equation 8.3.4 indicates
sequently, the CSTR must be significantly larger that on a similar plot, the quantity VR/FA0 is
than the PFR. These considerations are evident equal to the rectangular area shown in Figure
from a brief consideration of Figure 8.6a. In a 8.7. Thus, the ratio of the rectangular area to the
plug flow reactor one moves from right to left area under the curve is equal to the ratio of
along the rate curve as he or she proceeds from reactor volumes if identical molal flow rates and
inlet to outlet. In a CSTR, on the other hand, conversion increments are employed. For any
the reaction rate is constant throughout the rate expression that decreases monotonically
reactor and equal to that prevailing at the outlet. with increasing fraction conversion, the CSTR
Except in the case of autocatalytic reactions (see will always require a larger reactor volume than
Section 9.4), this rate corresponds to the lowest the corresponding PFR used to accomplish the
point on the PFR rate curve and implies that a same task.
larger volume will be required to accomplish
the same composition change between inlet and
outlet streams. Reciprocal,
rate curve
It is possible to reduce the disparity in reactor
volumes by using several tanks in series so that
one obtains stepwise changes in composition as
one proceeds from tank to tank. This situation is
depicted in Figure 8.6b for the case of three
CSTR's in series. In this case it is readily ap-
parent that each of the CSTR's would have a
larger volume than that of a PFR necessary to
effect the same composition change, so that the
total volume of the three CSTR's would exceed
that of a single PFR used to effect the same
overall change in composition. (We have shown
earlier that the three PFR's in series would be
equivalent to a single PFR with a volume equal
to the sum of those of the individual PFR's.) As
one increases the number of CSTR's in series, J A out
general class of reactions that follow simple nth- 3. The higher the order of the reaction, the
order kinetics with n normally lying between greater the ratio of sizes at a fixed conversion
zero and three. Following Levenspiel's analysis level. For zero-order reactions reactor size is
(11), we will presume that the inlet fraction con- independent of reactor type.
version is zero and that the volumetric expansion 4. Variations in fluid density on reaction can
parameter (3A) has some arbitrary value. Thus have significant effects on the size ratio, but
the effects are secondary when compared to
- fA)" the variations in reaction order. For positive
— r4 = (8.3.11)
SAfA)n values of the expansion parameter dA, the
volume ratio is increased, for negative values
For these conditions the general design equation
of SA, the volume ratio decreases. However,
for a plug flow reactor (8.2.7) becomes
the fact that in practice CSTR's are used only
vn 'Mi +sAfA)ndfA for liquid phase reactions makes this point
(8.3.12) academic.
,40/PFR
while that for a single CSTR (equation 8.3.4) The larger volume requirement of the CSTR
becomes does not necessarily imply extra capital costs,
especially for reactions that occur at ambient
+ sA/Ay pressure. However, the fact that the required
(8.3.13)
/csTR kC"A0 \ l - f A CSTR volume increases rapidly at high con-
version levels leads to some interesting optimiza-
Dividing equation 8.3.13 by equation 8.3.12 gives
tion problems in reactor design. The chemical
engineer must find the point at which he obtains
I
AO /CSTR 1-/4
an economic trade-off between the high fraction
conversion obtained in a large reactor versus the
dAfAy dfA
VRC%\
/PFR
p - fAr
low conversions in a small reactor. In the first
situation the equipment costs will be high and
Jo
(8.3.14) the product separation costs and raw material
costs low; in the second case equipment costs
Levenspiel (11) has evaluated the right side of will be low and the other costs high. The optim-
equation 8.3.14 for various values of n and SA. ization problem may be further complicated by
His results are presented in graphical form in allowing the number of CSTR's employed to
Figure 8.8. For identical feed concentrations vary, as we will see in Section 8.3.2.3.
(CA0) and molal flow rates (FA0) the ordinate of
In Illustrations 8.3 and 8.6 we considered the
the figure indicates the volume ratio required
reactor size requirements for the Diels-Alder
for a specified conversion level.
reaction between 1,4-butadiene and methyl
There are several useful generalizations which acrylate. For the conditions cited the reaction
may be gleaned from a thorough study of may be considered as a pseudo first-order reac-
Figure 8.8. They include the following. tion with 5A = 0. At a fraction conversion of
0.40 the required PFR volume was 33.5 m 3 , while
1. If the reaction order is positive, the CSTR is the required CSTR volume was 43.7 m 3 . The
larger than the PFR at all conversion levels. ratio of these volumes is 1.30. From Figure 8.8
2. The higher the fraction conversion involved, the ratio is seen to be identical with this value.
the greater the disparity between the sizes of Thus this figure or equation 8.3.14 can be used
the CSTR and the PFR. in solving a number of problems involving the
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 277
100
~2-
Io
0.01
Figure 8.8
Comparison of performance of single CSTR and plug flow reactor for the
nth-order reactions
The ordinate becomes the volume ratio KCSTR/VPFR or the space time ratio
T CSTR /T PFR if the same quantities of identical feed are used. (Adapted from
Chemical Reaction Engineering, Second Edition, by O. Levenspiel. Copyright
© 1972. Reprinted by permission of John Wiley and Sons, Inc.)
to the nonsteady-state analysis of reactors, we For liquid phase reactions where SA = 0, the
wish to consider briefly the relations that govern following expression is appropriate.
the transient behavior of an individual CSTR
operating under isothermal conditions. In some Kn
ln ~ ^ t (8-3.19)
cases it is possible to obtain easily analytical dt
solutions describing the approach to the steady
Combining equations 8.3.18 and 8.3.19 gives
state because of the uniform composition of the
reactor contents. This situation is in distinct dCAF ,„
contrast to that prevailing in a plug flow reactor ^ (8.3.20)
dt
where one must invariably resort to numerical
solutions of the transient material balance rela- There are several interesting forms of equation
tions in order to describe the approach to steady- 8.3.20 that correspond to various limiting con-
state conditions. ditions. For example, if both VR and %n are
For nonsteady-state operating conditions the time invariant, we have the situation corre-
generalized material balance on reactant A is as sponding to a shift from one steady-state
follows. operating condition to a second, and the quantity
T r^ . . Disappearance
rsr . .
InputA = Output + u . K
-f Accumulation
by reaction
F = F A- (— r \V (8.3.15)
A in -* A out ' V AF/ R dt
where the reaction rate is evaluated at the con- VRl%n is just the reactor space time T. Hence,
ditions prevailing at the reactor outlet and NA for this case,
is the total number of moles of species A in the C — C
dC AF
reactor at time t. NA may change as a result of (8.3.21)
changes in the volume occupied by the reacting dt
fluid or as a result of changes in the composition It is readily apparent that equation 8.3.21 re-
of the fluid. Since duces to the basic design equation (equation
8.3.7) when steady-state conditions prevail.
= CAFVR (8.3.16)
Under the presumptions that CA in undergoes a
then step change at time zero and that the system is
isothermal, equation 8.3.21 has been solved for
various reaction rate expressions. In the case of
first-order reactions, solutions are available for
If the molal flow rates are written as the pro- both multiple identical CSTR's in series and in-
duct of a concentration and a volumetric flow dividual CSTR's (12). In the case of a first-order
rate, and if equation 8.3.17 is combined with irreversible reaction in a single CSTR, equation
equation 8.3.15, 8.3.21 becomes
dC AF cA in -kCAF (8.3.22)
dt T T
dC4
+ (8.3.18) This equation is an ordinary linear differential
dt dt
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 279
T-l ^
Figure 8.9
Schematic representation of reciprocal rate curve for cascade of three
arbitrary size CSTR's.
K , T,
Figure 8.10
Notation for a cascade of N CSTR's in series.
280
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 281
recognizes that the use of such cascades is systems. In this section we presume a knowledge
almost invariably restricted to liquid systems of the reaction rate expression as a function of
and that in such systems density changes the concentration of reactant A, at least in
caused by reaction or thermal effects are usually graphical terms. That is, we presume
quite small, then additional relations or simpli-
fications can be developed from these starting (8.3.30)
equations. In particular this situation implies
that at steady state the volumetric flow rate where g{CA) denotes some arbitrary function.
between stages is substantially constant. It also For present purposes we assume that the CSTR's
implies that for each reactor, Jt = xb and that all operate at the same temperature and that
the following relation between concentration g{CA) is known at this temperature.
and fraction conversion is appropriate: Equation 8.3.29 may be written in the form
(8.3.28) c —
CAi = CAO(1 - ft) W(i- 1) (8.3.31)
where CAi is the concentration of reactant A
leaving the ith reactor. We will make the as- \n graphical terms this equation indicates that
sumption of constant density throughout the a plot of( — rAi) versus CAi is a straight line with
remainder of Section 8.3.2. We should also recall a slope — 1/ij that cuts the abscissa at C^-^.
that the development of equation 8.3.26 pre- An Analysis of the reactor design problem
sumed steady-state operation and thus this involves the simultaneous solution of equation
assumption is also made throughout Sections 8.3.30 and several equations of the form of
8.3.2.1 to 8.3.2.3. equation 8.3.31 (one for each reactor). These
In view of equation 8.3.28 and these assump- equations are the basis for the solution of the
tions, equation 8.3.27 can be rewritten as two types of problems with which the reactor
designer is most often faced in the analysis of
batteries of ideal CSTR's.
c
Ao
Concentration
Figure 8.11
Plots used in the graphical analysis of cascades of ideal CSTR's.
Now, for the first reactor, equation 8.3.31 through the point where the ordinate is zero
becomes and the abscissa is CA0. However, only the point
CAO CAI of intersection of the straight line with the curve
-r, = ~ (8.3.32) representing equation 8.3.30 has physical mean-
ing. It represents the conditions f hat must
For a specified inlet concentration this equation prevail in the first reactor. Hence the solution
indicates that a plot of —i\ versus CA1 is a to the first part of our problem consists of
straight line with slope (—1/iJ that passes drawing straight line EF through point E with
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 283
slope ( —1/ij). CAl is the point of intersection bination of CSTR's that is best suited to
of this straight line with curve I, as shown in achieving a specified conversion level. We will
Figure 8.11. begin by considering the case of two arbitrary
Now that CA1 is known, it is evident that a size ideal CSTR's in series operating under
similar process can be used to find CA2 because, isothermal conditions and then briefly treat the
in this case, equation 8.3.31 becomes problem of using multiple identical CSTR's in
series. Consider the two cascade configurations
- r 2 = °A1 ~ CA2
(8.3.33) shown in Figure 8.12, taken from Levenspiel (13).
For the first reactor, equation 8.3.26 becomes
Thus we may construct straight line GH in
Figure 8.11, by drawing a line of slope ( — 1/T2) Ji Jo
(8.3.34)
through the point with an ordinate of zero and AO
an abscissa of CAl. The intersection_of this line while, for the second, it becomes
with curve I gives us the effluent concentration
from the second reactor. This same procedure VR2 f2 ~ h
can be repeated for any other reactors that may FAO (-rA2)
be part of the cascade. The straight line JK was The relations indicated by these two equations
constructed in this fashion for the present case. are shown graphically in Figure 8.12 for two
Let us turn now to the second of the problems alternative configurations. In both cases the
mentioned earlier—determination of the corn- cascade operates between the same initial and
/=o
Rectangular area
measures size of
first unit
Figure 8.12
Graphical representation of variables for two CSTR's. (Adapted from Chemical Reaction
Engineering, Second Edition, by O. Levenspiel. Copyright © 1972. Reprinted by permission
of John Wiley and Sons, Inc.)
284 Basic Concepts in Reactor Design and Ideal Reactor Models
120
1 I I I I
- Solution for single CSTR
-Solution for cascade of three CSTR's
~ 100 -
*E 80
3
60
£ 40
20
Figure 8.14
Graphical solution for cascade of CSTR's.
kmole/m 3 , our initial choice of slope was correct. quite generally applicable tp the steady-state
If not, one must choose a new point J and repeat analysis of CSTR cascades, it is not highly
the procedure until such agreement is obtained. accurate in numerical terms, particularly when
Figure 8.14 indicates the construction for this the generation of the curve representing the
case. The slopes of the straight lines in this reaction rate expression involves graphical or
figure are equal to (18.2 - 0) x 10" 3/(0.034 _ 0.08) numerical differentiation of rate data. Algebraic
or —0.396 ksec" 1 . Thus the volume of each rhethods are capable of greater accuracy than
individual CSTR is equal to 0.278/0.396 or graphical ones if the functional form and con-
0.70 m 3 . The combined volume of all three stants involved in the reaction rate expression
CSTR's is thus 2.1 m 3 , which is a volume re- are knpwn. In this subsection we will again
duction of more than a factor of three when consider a cascade of CSTR's where the reactor
compared to the single CSTR case. In Section volumes are not necessarily equal Jto one another.
8.3.2.3 we will see that such large volume re- We will again use the nomenclature shown in
ductions are typical of the use of cascades of Figure 8.10 and start from equation 8.3.29,
CSTR's. rearranged as follows.
CAi - (8.3.36)
8.3.2.2 Algebraic Approach tq the Analysis of
Batteries of Stirred Tank Reactors Operating at Starting with the first reactor and using algebraic
Steady State. Although the graphical ap- iteration, we proceed to analyze the cascade for
proach presented in the previous subsection is particular forms of the reaction rate expression.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 287
CM =
CAO
(1 + + /C2T2) • • • (1 + kNTN) when we have discarded the negative root be-
(8.3.41) cause reactant concentrations cannot be
negative. Thus
These relations are valid regardless of whether
the reactors all operate at the same temperature
or at different temperatures. In the case where (8.3.47)
the cascade is isothermal and all reactors have
the same size, - 1 + yfl
C2 =
CAO
(8.3.42)
(1 + kxf -1+ /l+4/c 2 T 2
where T is the space time for an individual 2k?2<<2
T
CSTR. (8.3.48)
One may use the same general approach when
the reaction kinetics are other than first-order. and so on. Although a general expression for
However, except in the case of zero-order CAN would be rather complex algebraically,
kinetics, it is not possible to obtain simple closed involving N nested square roots, the solution for
form expressions for CAN, particularly if unequal the isothermal case is readily obtained graph-
reactor volumes are used. However, the numer- ically by the procedure outlined in the previous
ical calculations for other reaction orders are subsection. A number of other graphical solu-
not difficult to make for the relatively small tions have been discussed in the reactor design
number of stages likely to be encountered in literature by various individuals (15-18).
industrial practice. The results for zero-order Illustrations 8.8 and 8.9 indicate how the
kinetics may be determined from equation techniques developed in this and the previous
288 Basic Concepts in Reactor Design and Ideal Reactor Models
section may be used in the design analysis of This value is considerably larger than that
cascades of stirred tank reactors. calculated in Illustration 8.7 for a nonstoichio-
metric feed ratio, thus indicating the potential
ILLUSTRATION 8.8 DETERMINATION OF desirability of using an excess of one reagent
REACTOR SIZE REQUIREMENTS FOR A when it appears to be a positive power in the
CASCADE OF CSTR's—ALGEBRAIC rate expression. In any economic analysis of a
APPROACH process, however, the costs of separation and
recovery or disposal of the excess reagent must
Consider the Diels—Alder reaction between be taken into account.
benzoquinone (B) and cyclopentadiene (C), dis-
cussed in Illustrations 8.1 and 8.7. Case II—Two Identical CSTR's in Series
B + C -> adduct In this case it will be necessary to determine the
If one employs a feed containing equimolal con- concentration in the effluent from the first
centrations of reactants, the reaction rate ex- reactor in order to determine the required reactor
pression can be written as size. One way of proceeding is to write the design
equation for each CSTR.
r = kCcCB = kCB
CBO(/BI -0)
Determine the reactor size requirements for (A)
/Bl)2
cascades composed of one, two, and three iden-
tical CSTR's. Use an algebraic approach and CB0(/B2 "
assume isothermal operation at 25 °C where the (B)
/cCg O (l —
reaction rate constant is equal to 9.92 m 3 /
kmole-ksec. Reactant concentrations in the feed For identical reactors equations A and B may
are equal to 0.08 kmole/m 3 . The liquid feed rate be combined to give
is equal to 0.278 m3/ksec. The desired degree of
JB1 JB2 JB
conversion is equal to 87.5%.
(1 - (1 " hl
Solution For fB2 = 0.875,
For the specified degree of conversion the
effluent concentration of benzoquinone must be
(1 - 0.875)2(/*i) = (0.875 - - fB1)2
equal to (1 - O.875)(O.O8) = 0.01 kmole/m 3 . This equation may be solved by trial and error
Case I—Single CSTR or by graphical means to determine the com-
position of the effluent from the first reactor.
In this case equation 8.3.45 can be solved for the Thus fB1 = 0.7251 and CBl = 0.08(1 - 0.7251) =
reactor space time directly. 0.02199 kmole/m 3 . Either equation A or equa-
Q40 ~ CAI tion B may now be used to determine the re-
T =
kC2 quired reactor size. Hence
or (0.278)(0.08)(0.7251)
0.08 - 0.01 VR1 = = 3.36 m 3
9.92(0.08)2(l - 0.7251):
y
R\
= 70.56 ksec
9.92(0.01)2
The total volume of the two reactors is 6.72 m 3 ,
Thus
which is considerably less than half that required
VR1 = 70.56(0.278) = 19.6 m 3 if only a single CSTR is employed.
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 289
Case III—Three Identical CSTR's in Series This equation may be solved numerically by trial
and error, recognizing thalfBl must lie between
In this case there are two intermediate unspec-
0 and 0.875. The appropriate value of fB1 is
ified reactant concentrations instead of just the
0.6285. Equation F may now be used to deter-
single intermediate concentration encountered
mine that fB2 = 0.8038. With a knowledge of
in Case II. At least one of these concentrations
these conversions, equation C, D, or E may be
must be determined if one is to be able to ap-
used to determine the required reactor volume.
propriately size the reactors. In principle one
Thus
may follow the procedure used in Case II where
the design equations for each CSTR are written 0.278(0.08)(0.6285)
VR1 = ~ = 1.60 m 3
and the reactor space times then equated. This 9.92(0.08)2(l - 0.6285)
procedure gives three equations and three un-
knowns (VR1, fBU and fB2). Thus, for the first The total volume of the cascade is then 3 (1.60)
reactor, or 4.8 m 3 , which is again a significant reduction
in the total volume requirement but not nearly
/-,. as great as that brought about in going from one
kC2B0(l - fBl)2 to two CSTR's in series.
For the second, Obviously this approach is not easily extended
to cascades containing more than three reactors
2- fBl) and, in those cases, an alternative trial and error
(D)
- fB2)2 procedure is preferable. One chooses a reactor
volume and then determines the overall fraction
and, for the third, conversion that would be obtained in a cascade
of N reactors. When one's choice of individual
VR3 ^ C BO (0.875 - fB2)
(E) reactor size meets the specified overall degree of
i/- kC2 (\ — 0 RIS}2
conversion, the choice may be regarded as the
Combining equations C and E gives desired solution. This latter approach is readily
amenable to iterative programming techniques
0.875 - fB2 fB1 using a digital computer.
(0.125)2 (1 - far)2
or
(0.125)2/B1 ILLUSTRATION 8.9 DETERMINATION OF
/ „ = 0.875 - (F) OPTIMUM REACTOR SIZES FOR A
A
v ~~ /BI)
CASCADE OF TWO CSTR's
Combining equations C and D gives
Consider the Diels-Alder reaction between
JBI _ JB2 ~ JBI benzoquinone (B) and cyclopentadiene (C) dis-
(G) cussed earlier in Illustrations 8.1, 8.7, and 8.8.
(1 - fBl)2 " (1 - fBl)2
Now fB2 may be eliminated from the last B + C -> adduct
equation by using equation F. We wish to determine the effect of using a
2 cascade of two CSTR's that differ in size on the
(0.125) / B1
0.875 - -ft Bl volume requirements for the reactor network.
fB (1 - fBl)2 In Illustration 8.8 we saw that for reactors of
2
(1 - fm)2 (0-125) /fll equal size the total volume requirement was
1 - 0.875 +
(1 - fBi)2 6.72 m3. If the same feed composition and flow
290 Basic Concepts in Reactor Design and Ideal Reactor Models
V -4- VV
F Rn
gu rI fm
jB1 '0.875 -
Rl + R2 ~— kC~2 7j 7~x2 8.3.2.3 Size Comparisons Between Cascades of
(0.125)2
Ideal Continuous Stirred Tank Reactors and Plug
(Q Flow Reactors. In this section the size require-
It is this sum that we desire to minimize. The ments for CSTR cascades containing different
easiest approach to finding this minimum is to numbers of identical reactors are compared with
plot the quantity in brackets versus fB1. The that for a plug flow reactor used to effect the
minimum in this quantity then gives the mini- same change in composition.
mum total volume, and the value of fB1 asso- One may define a space time for an entire
ciated with the minimum may be used in cascade (ic) in terms of the ratio of the sum of
equations A and B to determine the optimum the component reactor volumes to the inlet
distribution of the total volume between the volumetric flow rate. Hence
two reactors. The minimum occurs when fB1 =
0.702.
Now vRl
T, = = I T,- (8.3.49)
FBo = CBOro = (0.08)(0.278)
- 0.02224 kmole/ksec If all component reactors have the same volume
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 291
N(N - 1) r o
N - 1
AN 2! CAAN J cascade
N(N - 1)(JV - 2) //cr CA0
£n
C I
^ A out/ plug flow
(8.3.53) (8.3.59)
and if ekXc is expanded in series form, If the effluents from the two streams are to be
identical and if equimolal feed rates and com-
positions are employed, the ratio of space times
2! 3! becomes equal to the ratio of total volume re-
(8.3.54) quirements. Thus, for constant density systems
where CAout = CAN = CAO(1 - fA),
a term by term comparison in the limit as N
approaches infinity indicates that
^ Total, cascade
r..
CA
kT
e < as N -> oo (8.3.55) -fA)
(8.3.60)
or
Figure 8.15, reproduced from Levenspiel (21),
as N -> oo (8.3.56) is in essence a plot of this ratio versus the frac-
tion conversion for various values of N, the
This relation is identical with that which number of identical CSTR's employed. The
would be obtained from equation 8.2.10 for a larger the value of N, the smaller the discrepancy
plug flow reactor with first-order kinetics. in reactor volume requirements between the
292 Basic Concepts in Reactor Design and Ideal Reactor Models
30
First—order
reaction
* ' * 0
>
\ ,
10
1
^ •
. > <
AT ^^*****"
^ ^
<
——
— ^
— 1
—— - --I -
~N = ^ "o s
1.0
0.01 0.1
= C /C
A A0
Figure 8.15
Comparison of performance of a series of N equal-size CSTR reactors with
a plug flow reactor for the first-order reaction
R, = 0
For the same processing rate of identical feed the ordinate measures the
volume ratio VWCSTR -S/^PFR or the space time ratio T WCSTRS /T PFR directly. (Adapted
from Chemical Reaction Engineering, Second Edition, by O. Levenspiel.
Copyright © 1972. Reprinted by permission of John Wiley and Sons, Inc.)
CSTR cascade and a PFR reactor. Note how with the greatest change occurring on addition
rapidly PFR behavior is approached as N in- of a second tank.
creases. Levenspiel has also included lines of Although the major thrust of the material
constant kz on this figure, and these lines may be presented in this subsection has concerned the
useful in solving certain types of design prob- relative size requirements for CSTR cascades
lems, as we will see in Illustration 8.10. and plug flow reactors, the practicing chemical
Levenspiel (22) has prepared a similar plot for engineer will be more concerned with the relative
second-order reactions (Figure 8.16). It is based economics of the two alternative reactor net-
on a generalization of equation 8.3.46 for N work configurations. In this regard it is worth
identical reactors in series and the integral form repeating that the additional capital costs asso-
of the plug flow design equation for second- ciated with the larger volumes of CSTR's are
order kinetics. We again see that increasing the relatively small, particularly when the units are
number of reactors in the cascade causes the designed for operation at atmospheric pressure.
total volume discrepancy between the cascade Consequently, a plot of the total costs associated
and a plug flow reactor to diminish rapidly, with the use of N reactors versus the number N
8.3 The Continuous Flow Stirred Tank Reactor (CSTR) 293
100
X Second—order
xX C\ •
reaction
\
V
\
\
/
/ i 4>
10
/ / >
\s
y X
y y
7? . f / X <*>
r /
yt /
^^
^.
"^
— i«
—-
•»»
T-
i
/ x^
\
• • — I
Figure 8.16
Comparison of performance of a series of N equal-size CSTR reactors with a
plug flow reactor for elementary second-order reactions
2A -> products
A + B -> products, CA0 = Cfl0
with negligible expansion. For the same processing rate of identical feed the
ordinate gives the volume ratio 1^NCSTRS/^PFR or the space time ratio TNCSTRS^PFR
directly. (Adapted from Chemical Reaction Engineering, Second Edition, by O.
Levenspiel. Copyright 1972. Reprinted by permission of John Wiley and Sons,
Inc.
will usually look something like that shown in the cascade. Consequently, the reactor designer
Figure 8.17. One obtains an economic trade-off must consider cascades containing different
between the costs associated with the high vol- numbers of reactors in the search for an eco-
ume requirements when very few reactors are nomic optimum.
employed and the additional engineering, fabri- The following illustration indicates how Fig-
cation, installation, and maintenance costs ures 8.15 and 8.16 are used in handling simple
incurred by using a larger number of reactors in reactor design calculations.
294 Basic Concepts in Reactor Design and Ideal Reactor Models
the volume ratios for the two reactors. Hence, at E. Increase in Conversion Arising from the Use
the same fB, we see that VCSTR/VPFR = 20. Thus of a Cascade of Two CSTR's at a Specified Feed
K:STR = 20(2.20) = 44 m 3 . Rate
If the feed rate is maintained constant while the
C. Conversion in a CSTR of the Same Size as number of reactors is doubled, the overall space
the Tubular Reactor time for the cascade will double. In the present
For a CSTR equal in volume to the tubular case kCB0TN = 2 = 2(350) = 700. From Figure
8.16, at this value of the dimensionless group
reactor, one moves along a line of constant
and N = 2, it is seen that fB = 0.99.
kCBOr in Figure 8.16 in order to determine the
conversions accomplished in cascades composed
of different numbers of reactors but with the
same overall space time. The intersection of the 8.3.2.4 Analysis of CSTR Cascades under Non-
linefcCfloi = 19.6 and the curve for N = 1 gives steady-State Conditions. In Section 8.3.1.4 the
fs = 0-80. equations relevant to the analysis of the transient
behavior of an individual CSTR were developed
D. Increase in Processing Rate Arising from the and discussed. It is relatively simple to extend the
Use of a Cascade of Two CSTR's at a Specified most general of these relations to the case of mul-
Degree of Conversion tiple CSTR's in series. For example, equations
The values of the group kCBOz that correspond 8.3.15 to 8.3.21 may all be applied to any indi-
to 95% conversion and one or two CSTR's in vidual reactor in the cascade of stirred tank re-
series may be determined from Figure 8.16. They actors, and these relations may be used to
are approximately 350 and 70, respectively. analyze the cascade in stepwise fashion. The
Thus, difference in the analysis for the cascade, how-
ever, arises from the fact that more of the terms
*r = 2 =2 70 in the basic relations are likely to be time variant
- 0.20 when applied to reactors beyond the first. For
(kCB0T)N=i 350
example, even though the feed to the first re-
Since actor may be time invariant during a period of
nonsteady-state behavior in the cascade, the
feed to the second reactor will vary with time as
and
the first reactor strives to reach its steady-state
condition. Similar considerations apply further
= 2-
TN=1 downstream. However, since there is no effect of
variations downstream on the performance of
then upstream CSTR's, one may start at the reactor
where the disturbance is introduced and work
= 10
o.2O downstream from that point. In our generalized
notation, equation 8.3.20 becomes
Consequently, the processing rate for the cas-
cade will be an order of magnitude greater than -Ai
that for a single CSTR. Note that operation of (8.3.61)
dt
the two reactors in parallel would have merely
doubled the processing capacity. Hence there is where the reaction rate is evaluated at conditions
a very strong case for operating with the units in prevailing in reactor i and where we have pre-
a series configuration. sumed both a constant density system and a
296 Basic Concepts in Reactor Design and Ideal Reactor Models
constant reactor volume. In this case both CAi the limit of zero recycle the system approaches
[and thus (— rAi)~\ and CA{i-1} are generally time plug flow behavior, and at the limit where only
variant quantities. an infinitesimal proportion of net product
For first-order irreversible reactions and iden- stream is produced, the system will approach
tical space times it is possible to obtain closed CSTR behavior. To develop characteristic de-
form solutions to differential equations of the sign equations that describe the performance of
form of 8.3.61. In other cases it is usually neces- recycle reactors, we will follow Levenspiel's
sary to solve the corresponding difference equa- treatment (23). The recycle ratio R is defined as
tions numerically.
/Volume of fluid \
If we were to extend the analysis developed in
returned to the I
equations 8.3.22 to 8.3.25 to the case of just two
\ reactor entrance /
CSTR's in series, the equation we would have to n j
(8.3.63)
solve to determine the composition of the /Volume of fluid leaving\
effluent from the second reactor would be the [ the system in net
following form of equation 8.3.61. product stream
\
_ i r cAin + C* - c4 Q A2 (8.3.62)
(it T2 \_kzl + kx, + 1
= FAO H -fA2)
Figure 8.18
Schematic representation of recycle reactor indicating symbols used in design analysis.
Parallel combinations are governed by the are concerned. Total feed capacity then increases
general principle enunciated in Section 8.2.3. in proportion to the flow rates that can be
handled by the various parallel legs. The com-
For most efficient utilization of the available
ponent reactors of each of the parallel legs can
reactor volume, all parallel streams that meet
then be considered as a series combination of
must have the same composition.
reactors and can be optimized in terms of the
Under these circumstances each parallel leg general principles enunciated below.
may be considered to be operating indepen- Consider the series combination of PFR and
dently insofar as the space time requirements CSTR's shown in Figure 8.19. In terms of the
necessary to effect a given composition change fundamental design equations for these idealized
Figure 8.19a
Series combination of PFR and CSTR's.
Figure 8.19b
Graphical representation of equations 8.4.1 to 8.4.3.
8.5 Comparison of Isothermal Stirred Tank and Plug Flow Reactors 299
reactor types (8.2.7 and 8.3.4), it can be said the arrangement of units depends on the actual
that shape of curve, the conversion level desired, and
Ik.= (*'" Jk. the units available.
(8.4.1)
FAO J'*° (~rA) (See Section 9 4 for an illustration of this type.)
A 2 - LA\ (8.4.2)
-rA 8.5 SUMMARY OF FUNDAMENTAL
and DESIGN RELATIONS—COMPARISON OF
~~ Xl 2 ISOTHERMAL STIRRED TANK AND
(8.4.3) PLUG FLOW REACTORS
AO — rA3
where the reaction rates in the last two equations Table 8.1 summarizes the fundamental design
are evaluated at the effluent conditions for each relationships for the various types of ideal
reactor and where FA0 is the molal flow rate reactors in terms of equations for reactor space
of reactant A that would correspond to zero times and mean residence times. The equations
conversion. The above relations are depicted in are given in terms of both the general rate
graphical form in Figure 8.19b. If the mathe- expression and Jith-order kinetics.
matical form of the reaction rate expression is If the various expressions are compared, it is
known, these equations can be used to specify evident that for constant density situations
completely the necessary design parameters.
^batch, constant pressure ^batch, constant volume
For optimum utilization of a given set of
= ~tpFR = TPFR (8.5.1)
ideal reactors operating as ari isothermal reac-
tor network, an examination of the l/{ — rA) Moreover, for negligible pressure drop through
versus CA curve is a good way to find the best a plug flow reactor,
arrangement of units. The following general
==
rules have been enunciated by Levenspiel (24). ^batch, constant pressure ^PFR (o.J.Z)
1. For a reaction whose rate-concentration curve As we stressed earlier, the reactor space time
rises monotonically (any nth-order reaction, is the independent variable at the control of the
n > 0) the reactors should be connected in reactor designer. This parameter is more mean-
series. They should be ordered so as to keep the ingful than the mean residence time in the
concentration of reactant as high as possible if reactor.
the rate-concentration curve is concave (n > 1), If we wish to make size comparisons of batch
and as low as possible if the curve is convex and continuous processing equipment in terms
(n < 1). of space time requirements, we must recognize
that there are nonproductive periods associated
For all reaction orders greater than unity, the with filling, heating, cooling, draining, cleaning,
appropriate order is: plug flow, small CSTR, etc., and that the long-term space time require-
large CSTR. In the case of reaction orders less ment would be given by
than unity, the reverse order should be em-
L
batch = * + ts (8.5.3)
ployed. For a first-order reaction the conversion
will be independent of the arrangement of the where
various reactors.
ts is the average nonproductive time per
2. For reactions where the rate-concentration batch processed
curve passes through a maximum or minimum t is given by the equations in Table 8.1
300 Basic Concepts in Reactor Design and Ideal Reactor Models
Table 8.1
Summary of Design Equations Given that V = Vo(1 + SA fA)
I. Batch
A. Holding time
1. Constant volume
(~rA) ~ JA>
2. Constant pressure
dfA
•J,7 + kCTa1 - fA)n
B. Equivalent space time x = t + t.
1
B. Space time
dfA = r'1 + ^
i-rA) /cQo'J
fcC".n' J (1a-f
- Ar
III. CSTR
Adapted from Chemical Reaction Engineering, First Edition, by O. Levenspiel. Copyright © 1962. Reprinted by
permission of John Wiley and Sons, Inc.
a
Indicated design relations not recommended for use in reactor analysis.
and temperature of the fluid are uniform (— rAF) is the reaction rate per unit volume ex-
throughout the tank. pressed in terms of the mass of reactant
For semibatch operation, the term "fraction A disappearing per unit time and eval-
conversion" is somewhat ambiguous for many uated at the conditions prevailing within
of the cases of interest. If reactant is present the reactor
initially in the reactor and is added or removed in subscripts 0 and F indicate input and
feed and effluent streams, the question arises as effluent conditions, respectively
to the proper basis for the definition of / . In If the initial condition of the reactor contents
such cases it is best to work either in terms of the is known and if the feedstream conditions are
weight fraction of a particular component pre- specified, it is possible to solve equation 8.6.1 to
sent in the fluid of interest or in terms of con- determine the effluent composition as a function
centrations when constant density systems are of time. The solution may require the use of
under consideration. In terms of the symbols material balance relations for other species or
shown in Figure 8.20 the fundamental material a total material balance. This is particularly true
balance relation becomes: of variable volume situations where the following
Wo#mo = vvF0mF + {-rAF)VR + T(wFpmVR)
overall material balance equation is often useful.
at
(8.6.1) <pm0 = O wF H (Pm^R) (8.6.2)
where
w is the weight fraction reactant By working in terms of total mass, the reaction
Om is the mass flow rate term disappears because the principle of con-
servation of mass must be satisfied.
pm is the mass density of the fluid in the There are a number of specific cases of equa-
reactor tion 8.6.1 that are of potential interest for com-
mercial applications. We wish to consider one
mode of semibatch operation using Illustra-
tion 8.11 in order to indicate the general
$ m f = Outlet mass flow rate principles involved in the analysis of these
wF = Weight fraction of reactamt systems.
A in effluent stream
Kladko (25) has presented a very interesting
case study of a reactor design problem involving
an exothermic isomerization reaction. Although
the reaction in question was well behaved in
laboratory scale apparatus, it behaved quite
differently when first run on a commercial scale
in a batch reactor. The system ran out of control
with the temperature increasing so rapidly that
the batch erupted violently through a safety
VR = Reactor volume valve and vented out over the building area. The
pm = Mass density of fluid
fact that the strong exotherm and its concomitant
effect on the reaction rate could have been pre-
4>m0 = Inlet mass flow rate dicted a priori on the basis of energy balance
w0 = Weight fraction of reactant A in feed stream calculations indicates the necessity of consider-
Figure 8.20
ing thermal effects in reactor design calculations.
Schematic representation of semibatch reactor These effects are the subject of Chapter 10p
indicating process variables. be particularly careful to take them into account
302 Basic Concepts in Reactor Design and Ideal Reactor Models
when moving from bench or pilot scale reactor material sump. The sump not only acts as a
systems to commercial scale equipment. Be- thermal sink for the cold incoming reactant, but
cause of the proprietary nature of the product also dilutes it, thereby reducing the reaction rate
many details of the reaction were omitted. How- and the rate of energy release by reaction.
ever, by making certain assumptions or engi- If the temperature of the reactor contents is
neering estimates regarding heat capacities and maintained constant at 163 °C, determine the
molecular weights, it is possible to generate the total amounts of species A and B in the reactor
necessary input data to permit evaluation of as functions of time when it is loaded according
several alternative reactor designs. Some alter- to the schedule shown below.
natives are considered as illustrative examples
in Chapter 10, and the next illustration indicates Time, t (h) Feed rate of A (lb/hr)
the type of analysis appropriate to isothermal
semibatch operation. 0-3 175
3-6 225
ILLUSTRATION 8.11 ISOMERIZATION IN A 6-7 275
SEMIBATCH REApTOR 7-8 325
Reagent A undergoes an essentially irreversible 8-11 400
isomerization reaction that obeys first-order 11-12 325
12-13 275
kinetics.
13-14 225
A-+B 14-15 175
Both A and B are liquids at room temperature 15-16 100
and both have extremely high boiling points. 16-17 50
A 1000-gallon, glass-lined kettle is available 17 + 0
for carrying out the reaction. The kettle may be
maintained at essentially isothermal conditions As we will see in Illustration 10.7, this type of
by a heat transfer fluid that circulates through a filling schedule is necessary to avoid dramatic
jacket on its external surface. The heat transfer exotherms that would result from sudden ter-
fluid may be cooled or heated as required by mination of the feed and to ensure that the heat
, circulation through appropriate heat exchangers. transfer capability of the system is not exceeded.
Since the reaction is exothermic, Kladko and his
co-workers (25) wished to consider the possi- Data and permissable assumptions
bility of using cold reactant feed to provide a
heat sink for some of the energy liberated by 1. The reactor contents are perfectly mixed.
reaction. By controlling The rate of addition of 2. The rate expression is first-order in species A.
feed, they could also obtain a measure of control 3. At 163 °C the reaction rate constant is
over the rate of energy release by reaction. Hence 0.8 hr" 1 .
a semibatch mode of operation appeared to be
Solution
an attractive alternative. Since cold< incoming
reactant would crack the hot glass liner, they A material balance involving the amount of
considered the possibility of starting with 1500 lb species A contained within the reactor can be
of product B in order to provide a thermal and written as
kCAV'R
dt
8.6 Semibatch or Semiflow Reactors 303
where nA is the instantaneous number of moles where <$>AQ is the mass rate of flow of species A
of species A contained within the reactor and into the reactor.
V'R is the instantaneous volume occupied by the Equation B may be solved in piecewise fashion
liquid solution. This equation is similar to to determine the mass of species A present in
equation 8.6.1, but it lacks the term correspond- the reactor as a function of time. The solution
ing to the effluent stream and it has been written can be written as
in molal units.
Now, at any time, m
A(t) dmA
r -" A j. ®A() - km,
dmA
kmA (B) - In = t - tt
dt
Table 8.1.1
Material Balance Analysis for Illustration 8.11
Fraction of input A
Time, t Total that remains
(hr) mA (lb) mB (lb) mass (lb) unconverted
PROBLEMS
vA0 at — mA
1. Baciocchi et al. [J. Am. Chem. Soc, 87 (3957),
Values of the amounts of A and B present in the
1965] have studied the chlorination of dichloro-
reactor at various times are given in Table 8.1.1.
tetramethylbenzene in acetic acid at 30° C. The
reaction of interest has the following stoichio-
LITERATURE CITATIONS metry.
1. Denbigh, K. G., Chemical Reactor Theory, p. 3, Cam- C 6 Me 4 Cl 2 + Cl 2 -> HC1 + C 6 Me 3 (CH 2 Cl)Cl 2
bridge University Press, Cambridge, 1965.
The data below are typical of those recorded in
2. Gooch, D. B., Ind. Eng. Chem., 35 (927), 1943.
a well-stirred batch reactor. Initial concentra-
3. Clark, E. L., Golber, P. L., Whitehouse, A. M., and
tions were as follows.
Storch, H. H., Ind. Eng. Chem., 39 (1555), 1947.
Cl 2 = 19.2 moles/m 3
4. Moss, F. D., Ind. Eng. Chem., 45 (2133), 1953.
5. Walas, S. M., Reaction Kinetics for Chemical Engineers,
C 6 Me 4 Cl 2 - 34.7 moles/m 3
p. 79, McGraw-Hill, New York, 1959.
6. Wassermann, A., J. Chem. Soc, 1936 (1028). Time, t (ksec) Fraction Cl 2 reacted
7. Kerr, G. T., J. Phys. Chem., 70 (1047), 1966.
0 0
8. Liu, S., Chem. Eng. 5c/., 24 (57), 1969.
48.4 0.2133
9. Inukai, T., and Kojima, T., J. Org. Chem., 32 (872), 85.1 0.3225
1967. 135.3 0.4426
10. Ratchford, W. P., and Fisher, C. H., Ind. Eng. Chem., 37 171.3 0.5195
(382), 1945. 222.9 0.5955
11. Levenspiel, O., Chemical Reaction Engineering, pp. 257.4 0.6365
125 127, Second Edition, Wiley, New York, copyright
© 1972. Used with permission.
12. Mason, D. R., and Piret, E. L., Ind. Eng. Chem., 43
(a) What is the order of the reaction and the
(1210), 1951. reaction rate constant?
13. Levenspiel, O., op. cit, pp. 139-143. (b) Determine the plug flow reactor volume
14. Szepe, S., and Levenspiel, O., Ind. Eng. Chem. Process
necessary to achieve 90% conversion of the
Design Develop., 3 (214), 1964. input chlorine using an input volumetric
15. Eldridge, J. M., and Piret, E. L., Chem. Eng. Prog., 46 flow rate of 0.15 m 3 /ksec and the same
(290), 1950. initial concentrations as used in the batch
16. Jones, R. W., Chem. Eng. Prog., 47 (46), 1951. experiments.
17. Weber, A.P., Chem, Eng. Prog., 49 (26), 1953.
18. Jenney, T. M., Chem. Eng., 62 (12), p. 198, 1955. 2. The following data were recorded using a
19. Crooks, W. M., British Chem. Eng., 2 (710), 1966. laboratory scale continuous stirred tank reactor
Problems 305
in which the stoichiometrically observed reac- (a) If the reactor is a CSTR with a volume of
tion was 600 ft3 and 60% of the B is converted to C,
2A -> R + S determine
(1) The space time for the reactor.
Moles A fed/ksec Moles A leaving/ksec (2) The average holding time in the reactor.
(3) The outlet flow rate from the reactor
8.0 0.8 (cubic feet per minute)
36 7.2 (b) If the reactor is a plug flow reactor with a
192 76.8 volume of 600 ft3 and 90% of the B is
converted to C, determine
The effective reactor volume was 1000 cm3. (1) The space velocity in the reactor.
The initial concentration of A was the same in (2) The effluent volumetric flow rate.
all runs and equal to the solubility limit of (3) The rate constant k.
species A in water. No R or S is present in the (4) The average holding time in the reactor.
feed. You have been asked to scale up this
reactor to produce significantly larger quantities 4. A chemical reaction is being studied in a
of R. It has been suggested that you use a laboratory scale steady-state flow system. The re-
tubular flow reactor with an inside diameter of actor is a well-stirred 1000 cm 3 flask containing
2 cm and that your feed be a saturated solution an aqueous solution. The reactor contents
of species A. No R or S is present in the feed. (1000 cm 3 of solution) are uniform throughout.
(a) What is the order of the reaction? The stoichiometric equation and data are given
(b) If you are to process 0.15 mole/sec of input below. What is the expression for the rate of
A to 90% conversion, what length of tubular this reaction? Determine the reaction order and
reactor do you require? The feed concen- the activation energy.
tration of A will be the same as in part (a).
diacetal + H 2 O -» Aldehyde + 2 Alcohol
3. Two gaseous streams are available for use
in carrying out a chemical reaction. The first Concentration
contains pure A and is produced at a rate of of Aldehyde
400 ft3/min. The second contains 50% B (re- Feed Rate Temperature, T in Effluent
mainder is an inert material) and has a flow rate Run (cm3/sec) (°Q (kmoles/m3)
of 200 ft3/min. These streams are mixed instan-
1 0.5 10 0.75
taneously and fed to a flow reactor. Both
2 3.0 10 0.5
streams are at the same temperature (86 °C) and 3 3.0 25 0.75
pressure (1 atm) and these quantities remain
unchanged during the instantaneous mixing
process. The gases behave ideally. A and B Feed concentration of diacetal is 1 kmole/m 3 .
react to form an addition product Note. Data are hypothetical. The reaction may
be assumed to be zero-order in aldehyde and
A + B -> C alcohol, and apparently zero-order in water.
with
rc = kCACB
5. Roper \Chem. Eng. Sci., 2 (27), 1953] has
The reaction is carried out isothermally in two studied the reaction of chlorine {A) with 2-
flow reactors. Both reactors operate at a con- ethylhexene-1 (B) in carbon tetrachloride solu-
stant total pressure of 1 atm. tion. Solutions of these materials were prepared
306 Basic Concepts in Reactor Design and Ideal Reactor Models
and brought together in a mixing chamber at the (4) n = 1/2 two plug flow reactors
inlet to a tubular flow reactor. The following in parallel
data were reported at 20 °C. The initial concen- (5) n = 1 two plug flow reactors
trations refer to values calculated on the basis of in parallel
perfect mixing. (6) n = 3 two plug flow reactors
in parallel
cA Assume that the exit composition from
(moles/m3) the last reactor remains unchanged in
Run (moles/m3) (moles/m3) at exit T (sec) all cases and that none of the A has been
converted to products prior to entering
1 91 209 23 0.600 the reactor. Do not assume 5 = 0.
2 91 209 32 0.376 Obtain a general solution. In series
3 91 209 45 0.284 operation the small reactor precedes the
4 110 211 34 0.525
large reactor in the sequence.
5 110 211 46 0.324
6 110 211 59 0.232 (b) Repeat part (a) for the case of continuous
stirred tank reactors. Assume that there
is no change in the number of moles on
reaction.
It has been suggested that the rate expression
for the reaction is of the form 7. Walter [J. Chem. Eng. Data 5 (468), 1960]
r = kCACB has studied the kinetics of ethylene chloride
pyrolysis over a pumice catalyst.
Graphically determine if this expression is
consistent with the above data. If so, what is the C1CH2—CH2C1 -* HC1 + CH 2 =CHC1
reaction rate constant? If not, what do you
recommend? The reactor consists of a cylindrical tube 59 cm
long packed with pumice stone. The catalyst
6. (a) The rate of a chemical reaction is given charge was constant in all of the runs below
by and equal to W. The reactor void volume was
approximately 100 cm3. The reaction is believed
r = kC" where to be first-order in ethylene chloride under the
V dt
conditions of this study and in all cases the feed-
If 90% of the reactant A is converted to stock was pure reactant. Both the reactants and
products in a reactor, and if one obtains the products are gases at the conditions involved.
a second reactor that is one half the size Ideal gases may be assumed. The following data
of the first, determine the increase in were reported at 600°C and one atmosphere.
feed capacity that results from the
following types of operation.
(1) n = 1/2 two plug flow reactors Input flow rate
in series (moles/ksec) Fraction conversion
(2) n = 1 two plug flow reactors 0.86
0.550
in series 0.544 0.85
(3) n = 3 two plug flow reactors 0.344 0.94
in series
Problems 307
(a) What is the reaction rate constant at 600 GC 9. The catalytic dehydrochlorination of tetra-
in sec" 1 ? chloroethane has been studied by Shvets,
(b) If the reaction rate constant at 500 °C is Lebedev, and Aver'yanov [Kinetics and Cata-
0.141 sec" 1 , what is the activation energy lysis, 10 (28), 1969].
for the reaction?
C 2 H 2 C1 4 -> C 2 HC1 3 HC1
8. Young and Hammett [J. Am. Chem. Soc,
The reaction is first-order with respect to tetra-
72 (286), 1950] have studied the alkaline bromi-
chloroethane with a rate constant
nation of acetone in a stirred flow reactor
(VR = 118 cm3). The stoichiometric equation k = lO 1 2 *?- 2
for the bromination is usually considered to be:
when T is expressed in degrees Kelvin. During
(CH 3 ) 2 CO 3BrO" - the reaction, small amounts of other products
CHBr 3 CH 3 CO 2 2OH are produced by side reactions. The chlorine
produced by the reaction can serve as a catalyst
and the reaction rate is believed to be propor- poison at concentrations above 150 ppm on a
tional to the concentrations of acetone and mole basis. These investigators have reported
hydroxyl ions and independent of the concen- the following values for the mole ratio of
tration of hypobromite ion. Determine if the chlorine to HC1.
data below are consistent with this rate expres-
sion. If they are, what is the value of the reaction Temperature, T( C C) Ratio
rate constant?
408 1.7 X 10~ 4
440 3.2 x 10" 4
455 4.0 X 10" 4
Inlet hydroxide Inlet acetone Inlet hypobromite Effluent acetone Effluent hydroxide
concentration concentration concentration concentration concentration Total flow
Run (moles/m3) (moles/m3) (moles/m3) (moles/m3) (moles/m3) rate (cm3/ksec)
It has been suggested that a pilot plant operation have studied the reaction of ethylene and
to determine the feasibility of developing this chlorine in liquid ethylene dichloride solution
process be carried out in a tubular flow reactor in a CSTR. The stoichiometry of the reaction is
with a volume of 0.15 m 3 . It is suggested that
the reactor operate at 450 °C and 1 atm with a C2H4 + CI2 —> C2H4CI2
feed flow rate of 41.7 moles of pure tetra-
chloroethane per kilosecond. Will the catalyst Equimolar flow rates of ethylene and chlorine
be susceptible to poisoning under these op- were used in the following experiment, which
erating conditions? was carried out at 32 °C.
11. Balasubramanian, Rihani, and Doraiswamy Feed ratio H 2 /C 3 H 8 = 2.0 in all cases
[Ind. Eng. Chem. Fundamentals, 5 (184), 1966] Reactor pressure = 7.0 MPa in all cases
Problems 309
Space Fraction
required for a 50% increase in total pressure is
Temperature, T (°C) time (sec) propane converted 197 sec. The initial pressure is 1 atm. The
reaction is known to be second-order in ace-
538 0 0 taldehyde. You have been asked to determine
42 0.018 the volume of a plug flow reactor necessary to
98 0.037 process 120 liters/min of pure acetaldehyde to
171 0.110 an 80% conversion level. The feed pressure is 1
593 40 0.260 atm. The reaction may be assumed to be
81 0.427 essentially irreversible. The pressure drop along
147 0.635 the length of the plug flow reactor is negligible.
For this problem the stoichiometry of the main 15.* Ratchford and Fisher found that methyl-
reaction may be considered to be of the form acetoxypropionate decomposes on heating to
form acetic acid and methyl acrylate:
H, C,H. CH 4 C2H6
CH3COOCH(CH3)COOCH3 ^
13. Consider the following homogeneous gas CH3COOH + CH 2 =CHCOOCH 3
phase reaction.
The pyrolysis closely approximates a first-order
A + B -• C + D
irreversible reaction with a rate constant given
The reaction is essentially "irreversible," and its by:
rate in a constant volume batch reactor is
given by ic,=7.8xl0*exp^-38'20°N
16. Buckles and McGrew [J. Am. Chem. Soc. where E is the ester produced by the reaction.
88 (15), 1966] have studied the dimerization of The reaction occurs at a constant temperature
phenyl isocyanate in liquid solution in the in an acetone solution of these species. The
presence of a catalyst. extent of reaction is limited by equilibrium
conditions. Neglect the volume change on the
O reaction.
I! The following two data points have been
c reported
2C 6 H 5 NCO ^± C 6 H 5 N NC 6 H 5
kb ^ /
Space velocity Fraction conversion of acid
c
33 ksec" 1 0.50
phenyl o 67.65 ksec" 1 0.40
isocyanate 1,3 diphenyl 2,4-uretidinedione
(monomer) (dimer) What is the equilibrium degree of conversion?
The forward reaction is third-order (second- 18. Kem Engineer has been asked to scale up
order with respect to monomer and first-order an existing process to obtain an increased pro-
with respect to catalyst). The reverse reaction is duction capacity for compound B. At present
second-order overall (first-order with respect to the process is carried out in two CSTR's in
both catalyst concentration and dimer). The series. The reaction involved has the following
reaction is catalyzed by tributylphosphine at a stoichiometry.
concentration of 0.05 moles/liter.
The following data relative to the reaction at 2A - • B + C (liquid phase reaction)
25 C are available: Unfortunately, the data from which the rate
constants were originally determined have been
Ke = 0.178 lost. Kem believes that he can determine a rate
kf = 1.15 x 10~ 3 Iiters2/mole2-sec constant from measurements on the plant's
EAf = 1.12 kcal/mole present system for manufacturing compound B.
EAb = 11.6 kcal/mole Data
Volume of first CSTR 30 gal
If a monomer solution at a concentration of Volume of second CSTR 40 gal
1 mole/liter is fed to a CSTR at 0 °C, determine Feed to first CSTR pure A
the space time necessary to achieve a conversion Fraction conversion of A in first
corresponding to 90% of the equilibrium value. CSTR 0.60
If the reactor volume is 100 liters, what is the Overall fraction conversion of A
corresponding volumetric flow rate? by the two CSTR's in series 0.80
Volumetric feed rate to first
17. An alcohol A and an acid B are fed to a CSTR 900 gal/hr
CSTR in equimolal proportions. The mechanis- The reaction is known to be second-order in A.
tic and stoichiometric equation for the reaction (a) Is it possible for Kem to determine a rate
is constant for the reaction from these data?
A + + H2O If not, what additional data does he need?
Problems 311
has been studied in a tubular flow reactor (inside reactor each minute. The fraction of the inlet A
diameter = 0.1 m) filled with silica-alumina converted to R in the first reactor is such that
catalyst by Hanson and Engel [AIChE /., 13 the effluent from the first reactor leaves at 85 °F.
(260), 1967]. The reaction occurred at 900 °F The temperature of the stream leaving the
and 1 atm with pure ethylbenzene feed. The second reactor is 100 °F.
density of the catalyst was 2.75 g/cm 3 and the The concentration of A in the entrance stream
void fraction was 0.57 (volume of voids/total is 1.5 lb moles/ft3. The following table provides
volume). The catalyst was found to degenerate some information about the temperature depen-
with time, and the first-order rate constant dence of the rate constants k1 and fc_ j .
yaried accordingly. The following data were
obtained. T( F) k{ (ft3/lb mole-min) /<_ l (ft3/lb-mole-min)
all of the water present. The gases leaving the is homogeneous. The reverse reaction may
cooler pass to a long pipe followed by a series of be neglected.
absorption towers. In these towers further (d) Because of the large percentage of inerts in
oxidation of NO to NO 2 takes place. Of the NO thcfeed stream, the volume change on reac-
present in the gases leaving the cooler, 90% is tion may be neglected. This implies that our
now being oxidized to NO 2 in the oxidizing answer will only be a first approximation to
chamber and absorption towers. the true solution.
It has been suggested that the production (e) It may be assumed that absorption of the
capacity of the plant can be increased by intro- NO 2 by H 2 O in the absorption towers does
ducing additional air into the gases leaving the not affect the oxidation rate.
cooler. Bucky Badger suggests that the decrease (f) The plant capacity is directly proportional
in residence time will be more than compensated to the quantity of NO 2 produced per unit
for by the increased rate of reaction. He proposes time.
that the plant throughput may be increased, (g) The reaction rate at 30 °C is given by:
keeping the conversion of the NO fixed at 90%.
(a) Is it possible to increase the production "CJVO _ iri r
NO
capacity as suggested? ITt ~~ °2
(b) If so, what should be the number of moles of where concentrations are expressed in kilo-
air fed per mole of gas leaving the cooler in moles per cubic meter and where
order to maximize the production capacity?
What is the percent increase in capacity k = 8.0 x 103 m6/kmole2-sec
under these conditions?
(h) Composition of air:
Data and Suggestions N2 79%
The composition of the gases leaving the cooler O2 21%
is:
N2 83.0% (i) Note that the addition of air increases the O 2
concentration, but also adds N 2 as an inert
O2 8.8%
diluent. Thus there will be some optimum
NO 8.2% air feed rate.
(a) Assume that this composition will be un- (j) A useful approach is to set up an expression
for the desired result in terms of the moles of
changed by any change in the plant produc-
air fed per 100 moles of gas leaving the
tion capacity.
cooler.
(b) For simplification, assume the gas flow ap-
proximates plug flow at a constant tempera-
ture of 30 °C. 31. An addition reaction of the form A + B -•
(c) The oxidation reaction E takes place in the liquid phase in a combina-
tion of two plug flow reactors in series as shown
2NO + O 2 2NO 2 below.
50 Ib moles/hr - A
- > •
A, I% E ^
•^ V ~ 10 ft 3 VR ~ 10 ft 3
B
50 Ib moles/hr -
Reactor I Reactor II
316 Basic Concepts in Reactor Design and Ideal Reactor Models
The concentrations of species A and B at the reactors be greater than, less than, or equal
inlet to reactor I are both equal to 1.5 lb moles/ to twice the fraction conversion obtained in
ft3. The rate expression for this reaction is given part (a)? Why!
by (c) Would operation of the reactors in parallel
r = kCACB give greater, lower, or the same overall con-
version of A as that obtained in the present
If isothermal operation at 25 °C is assumed, the mode of operation?
value of the rate constant k is 2.0 ft3/lb mole-hr. Consider two conditions:
(a) What fraction of the inlet A is converted to (1) Fifty percent of the A-B mixture is fed to
E in the first reactor? each reactor.
(b) Will the fraction of the inlet A that is con- (2) Sixty percent of the A-B mixture is fed to
verted to E in the series combination of one reactor and 40% to the other.
Selectivity and Optimization
317
318 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
reactor, but the raw material and separation It follows that one obtains maximum selecti-
costs necessary to produce a given amount of vity by choosing reaction conditions such that
desired product are unacceptably high. the ratio (rv/rw) always has its highest value. This
The bulk of this chapter is devoted to a dis- ratio is often referred to as the instantaneous
cussion of optimization with regard to selectivity selectivity.
considerations. In the sections that follow we For a specific system at a given temperature,
will take 5 = 0 in order to concentrate on the nature dictates the values of fcl9 fc2, a1? and a 2 .
primary effects and to simplify the discussion. The only factor that the engineer is at liberty to
Consequently, in this chapter, the terms space adjust and control is CA. It may be maintained
time, mean residence time, and holding time may at a high level by using a batch or plug flow
be used interchangeably. reactor, by operating at low conversions, by in-
creasing the pressure in gas phase systems, and
9.1 PARALLEL REACTIONS by avoiding the use of inert diluents in the feed.
Low concentrations of a reactant are achieved
The possibility of a species reacting by parallel
using a CSTR, operating at high conversions,
paths to yield geometric isomers or entirely
lowering the pressure in gaseous systems, and
different products is often responsible for low
adding inerts to the feed stream. Note that in
yields of a desired product. If circumstances are
this case the desire for selectivity works at cross
such that the orders of the desired and unwanted
purposes to the desire for a small reactor size so
reactions are different with respect to one or
that a good design with respect to one constraint
more species, it is possible to promote the desired
may be poor with respect to the other. In this
reaction by an appropriate choice of reactor
situation a detailed economic analysis is neces-
type and reaction conditions.
sary to optimize the design.
Following the treatment of Levenspiel (1) we
Let us now consider the three possible rank-
shall consider a set of parallel reactions in which
ings of the reaction orders in order to determine
only a single reactant species has any influence
when the concentration of species A should be
on the corresponding reaction rate expressions.
kept at high or low values.
A -> V (desired or valuable product)
Case I. at i > a 2
(9.1.1) In this case the order of the desired reaction is
higher than that of the unwanted reaction so the
A -> W (undesired or relatively worthless
product) exponent on the concentration is positive. The
instantaneous selectivity is promoted by employ-
(9.1.2)
ing high concentrations of reactant. Conse-
The corresponding rate expressions are: quently, batch or plug flow reactors are most
appropriate from a selectivity viewpoint.
<V = (9.1.3) Case II. oe2 > «!
In this situation the order of the unwanted reac-
dCjr tion is greater than that of the desired reaction,
(9.1.4)
dt so the selectivity is enhanced by using low con-
Elimination of time as a variable between centrations of reactant. A CSTR is appropriate
equations 9.1.3 and 9.1.4 gives from a selectivity standpoint. Unfortunately,
this situation is the one in which the selectivity
Ty dCy liX (aj _ a2 ) considerations work against the desire for a
(9.1.5)
rw dCw k2 small reactor size.
9.1 Parallel Reactions 319
Desired strategy
Desired strategy
B B
~1 1/ 1
JA • » /
-4- T
Figure 9.2
Contacting patterns for various combinations of high and low concentration of reactants in
continuous flow operations. (Adapted from Chemical Reaction Engineering, Second Edition,
by O. Levenspiel. Copyright © 1972 Reprinted by permission of John Wiley and Sons, Inc.)
situations. The most desirable contacting pattern and unity. The overall fractional yield is the
and mode of operation can be determined only average of the instantaneous fractional yield
by considering several alternative processing integrated over the reactor. The proper averaging
modes and the possibility of recycle of reactants, technique depends on the type of reactor em-
either prior to or subsequent to separation of ployed. CSTR's do not present any problems
the products. with respect to the averaging process, because
In order to determine the product distribution the fluid composition is constant throughout
quantitatively, it is necessary to combine material the volume of the reactor. In this case the in-
balance and reaction rate expressions for a given stantaneous and overall yields are identical:
reactor type and contacting pattern. On the
other hand, if the reactor size is desired, alter- (9.1.9)
native design equations reflecting the material
balances must be employed. For these purposes where the rate expressions are evaluated at the
it is appropriate to work in terms of the fractional reactor effluent composition.
yield. This is the ratio of the amount of a product For a plug flow or a batch reactor where the
formed to the amount of reactant consumed. reactant concentration varies with position or
The instantaneous fractional yield of a product V with time, the overall yield can be determined by
(denoted by the symbol y) is defined noting that
dCv cVF- c v o
y = —- (9.1.7) JCA0 Vj
vv dCA
Substitution into the definition for the overall
while the overall fractional yield Y is defined as yield gives
CVF — Cvc
(9.1.8 rA1 dC
CAF - CAO JCAO
CAF ~ ^A(.
Appropriate stoichiometric coefficients are FA°ydCA (9.1.11)
employed to ensure that 3; and Y lie between zero cA0-c AF JC AF
9.1 Parallel Reactions 321
It is possible to extend this treatment to the taneous yield in equations like equation 9.1.11
case of multiple CSTR's operating in series by and then evaluating the integral directly. Un-
adapting the procedure outlined by Denbigh and fortunately, there are many significant industrial
Turner (2). Let (ACA)U (ACA)2, and (ACA)h rep- reactions, particularly heterogeneous catalytic
resent the changes in the concentration of reactions utilizing a complex feed stpck, for
species A that take place in tanks one, two, and which the formal rate expressions have not been
i, respectively. determined. In these cases the concept of instan-
taneous yield can sometimes be quite useful.
(9.1.12) This situation occurs when the instantaneous
yield depends on only a single composition
where (CA)t is the concentration of reactant A in
variable. In such cases y can often be measured
tank i. The associated changes in the concentra-
with far less effort than would be needed to
tions of the valuable product V are given by
determine the formal rate expressions for the
various competing reactions. Plots of the instan-
(9.1.13) taneous yield versus composition may be deter-
mined by carrying out a series of steady-state
where yt is the yield characteristic of the steady- experiments in a CSTR. All runs are carried out
state concentrations prevailing in the fth tank. at the same temperature and catalyst conditions.
This relation follows directly from equation The shape of the curve determines which type
9.1.7 as applied to a finite process taking place of reactor configuration gives rise to the optimum
at constant y. The overall change in the concen- product distribution.
tration of species V is obtained by summing Figure 9.3 contains typical instantaneous yield
over the pertinent number of tanks (n). versus reactant concentration plots and the
shaded areas indicate the composition changes
ACV = CVn - C vo = - £ U-(ACJJ (9.1.14) of the desired product that are effected by vari-
V
A i = l ous reactor types. From the definition of the
overall yield,
Combination of this result with the definition
of the overall yield given by equation 9.1.8 gives
CVF ~ Cvo = ~ Y(CA0 - CAF) (9.1.16)
(a)
(c)
Figure 9.3
Instantaneous yield versus reactant concentration curves and their
relation to overall product concentration changes, (a) plug flow or
batch reactor, (b) CSTR. (c) three CSTR's in series. (Adapted from
Chemical Reaction Engineering, Second Edition, by O. Levenspiel.
Copyright © 1972. Reprinted by permission of John Wiley and Sons,
Inc.)
so the pertinent area in this case is the rectangle plug flow or batch reactor. If the instantaneous
shown in Figure 9.3b. For a staged cascade of yield decreases continuously with increasing
stirred tank reactors a similar analysis indicates reactant concentration, the optimum product
that the pertinent area is that given by the sum distribution is achieved using a continuous
of the rectangles corresponding to the individual stirred tank reactor. When maxima or minima
tanks. are observed in the instantaneous yield curve,
The shape of the instantaneous yield curve the desired selectivity is enhanced by the use of
determines the optimum reactor configuration a combination of plug flow and stirred tank
and flow pattern for a particular reaction net- reactors. However, considerations other than
work. For cases where the instantaneous yield selectivity may influence the final choice of
increases continuously with increasing reactant reactor type, especially when the instantaneous
concentration, the optimum reactor configura- yield curve is relatively flat over the reactant
tion from a product selectivity viewpoint is a concentration range of interest.
9.1 Parallel Reactions 323
complete conversion, the yields at a given con- at which the concentration of a particular inter-
version level can be evaluated from equations D mediate passes through a maximum. If interested
and F if the values of the reaction rate constants in designing a continuous flow process for pro-
and the initial concentration are known. In all ducing this species, the chemical engineer must
cases, however, the yield from the CSTR will make appropriate allowance for the flow con-
exceed that from the PFR. ditions that will prevail within the reactor. That
This illustration has provided us with a con- disparities in reactor configurations can bring
crete example that indicates in quantitative form about wide variations in desired product yields
the validity of the general rule of thumb that we for series reactions is evident from the considera-
have stated for analyzing parallel reactions. High tions in Illustrations 9.2 and 9.3.
concentrations favor the higher-order reaction,
and low concentrations favor the lower-order ILLUSTRATION 9.2 QUANTITATIVE
reaction. DEVELOPMENT OF SERIES REACTION
RELATIONSHIPS FOR BATCH AND PLUG
9.2 CONSECUTIVE (SERIES) REACTIONS FLOW REACTORS
The time corresponding to maximum yield reactor from the standpoint of maximizing pro-
of V is obtained by differentiating equation B duction of the intermediate. What will be the
with respect to time and setting the derivative effluent concentration of V for this optimum
equal to zero. operating condition? It may be assumed that
species V and W are not present in the
feedstream.
Solution
dC VF kxx)k2 k2x)k
— Ki ( _
dx + +
0.9
or
optimum (G) • 0.8
0.7
Substitution of this result into equation E gives
an expression for the maximum possible effluent 0.6
concentration of the desired species. -4 -3 -2 -1
w
(K
) iog, o ^
n
CvFmax Figure 9.4
cA0 r Ai
1 4-
Comparison of maximum yields for series reactions
in stirred tank and plug flow reactors.
1
(H) strongly influence the overall process economics.
+ T^
Figure 9.4 indicates the range of values of k2/k1
for which yield considerations are significant.
Successive substitution reactions usually have
The ratio of the maximum yield achievable in rate constants that lie near the minimum in the
a CSTR to that which can be obtained in a relative yield curve.
batch reactor is given by the ratio of equation
H to equation E of Illustration 9.2. Hence,
As we noted earlier, there may be heat transfer
y 1 considerations or other factors that dictate the
J
max,CSTR h \k2Hk2-ki)
i
max,PFR 1+ use of a CSTR even when the yield consider-
'rJllii ations are unfavorable. In such cases the yield
9.2 Consecutive (Series) Reactions 327
of the desired product may be upped significant- the plug flow or batch reactor requires a lower
ly by using a battery of CSTR's in series. If space or holding time than a CSTR to achieve
desired, one has the additional flexibility granted the maximum concentration of intermediate.
by using tanks with different volumes or tanks The more this ratio departs from unity, the
operating at different temperatures in the pro- greater the difference in space times. This fact
duction line. The relative capacities of the becomes evident on substitution of numerical
various tanks may then be chosen to optimize values into equations C and G of Illustrations
product yield. This problem has been considered 9.2 and 9.3, respectively, or when plots of
previously by Denbigh (4). He has shown that if CV/CAO versus kxz are prepared for various
both reactions are first-order and if isothermal ratios of k2/k1. [See, for example, Levenspiel
conditions prevail, the capacities of all tanks (5).] In general, for series reactions, the maxi-
should be the same. However, if the degradation mum possible yields of intermediates are ob-
reaction is of higher order than that producing tained when fluids of different compositions and
the desired product, the capacities of the tanks at different stages of conversion are not allowed
should become smaller from the first tank to mix.
onward. Conversely, if the degradation reaction Second, it is possible to plot the data in time-
is of lower order, then the capacities of the independent form in order to obtain curves that
tanks should increase from the first tank onward. are useful in the determination of k2/kx in
There are a few other points worthy of note kinetic studies. The experimental points are
that become evident on closer inspection of matched with one of the family of curves on the
the equations developed in Illustrations 9.2 and graph corresponding to the type of reactor
9.3. First, except for the case where k2/k1 = 1, employed in the investigation. Figure 9.5 is an
A = 1- (CA/CA0)
Figure 9.5
Dimensionless representation of product distribution from series reactions
A ^ v ^ w (stirred tank reactor).
328 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
example of this type of plot for a continuous the initial A converted. Levenspiel (6) has
stirred tank reactor. prepared such plots, and Figure 9.6 is repro-
For cases where it is possible to readily duced from his textbook. This figure presents
recover unused reactant A from the product the fractional yield of intermediate V as a
mixture for recycle to the reactor inlet, the use function of the ratio of rate constants {k2/k1)
of the definition of yield employed in Illustra- and the fraction A converted. The curves
tions 9.2 and 9.3 is not appropriate. In this indicate that the fractional yield of the inter-
situation, a more appropriate definition is mediate species is always higher in plug flow
than when extensive back-mixing occurs, re-
Cy Cy
y = (9.2.1) gardless of the conversion level. Moreover, the
-AO Cy CW figure has important implications with regard
The equations derived earlier for the effluent to the conversion level of species A for which
concentrations in the PFR and CSTR cases may one should design. If the reaction under consid-
be substituted into equation 9.2.1 to obtain eration has a value of k2/kl that is greater than
numerical values of the fractional yield of the unity, the fractional yield of V drops very
intermediate V as a function of the fraction of rapidly, even at low values of the conversion
t 0.4
0.2
Figure 9.6
Comparison of the fractional yields of V in mixed and plug flow
reactors for the consecutive first-order reactions. A ^ V -^ W.
(Adapted from Chemical Reaction Engineering, Second Edition,
by O. Levenspiel. Copyright © 1972. Reprinted by permission of
John Wiley and Sons, Inc.)
9.3 Series-Parallel Reactions 329
of species A. Thus, in order to avoid excessive concentration will shift the location of the
production of the undesirable product W, it is maximum intermediate concentration, and this
necessary to design for a low conversion of A can be used to optimize the product distribution.
per pass and recycle this species after separation
from the product mixture. In such cases the 9.3 SERIES-PARALLEL REACTIONS
load on the separation equipment will be high,
and large quantities of material will have to be Reaction networks that consist of a multiplicity
processed. Consequently, the costs of the sep- of reactions in which steps in series and steps in
aration process will strongly influence the over- parallel are both present are often referred to as
all process economics. As the ratio k2/kl series-parallel reactions. These systems often
decreases below unity, the fractional yield of are characterized by rate expressions that place
intermediate increases at a given conversion conflicting demands on the type of contacting
level. It is evident that at low values of k2/k1, desired, so that it is often impossible to obtain
this yield becomes relatively insensitive to a unique answer to a design problem. In such
fraction conversion until the conversion level cases the practicing chemical engineer must
begins to exceed 80 to 90%. In this situation exercise creativity and judgment in the choice
one should design for relatively high conversion of contacting pattern and reactor type. To
levels. illustrate the type of conflicts involved, one may
Although the bulk of the discussion in this consider the following combination of mecha-
section and the illustrative examples have been nistic equations:
restricted to successive first-order reactions,
concentration versus time curves can be de- A-*V (9.3.1)
veloped for other cases, including those where A + A W2 (9.3.2)
the consecutive reactions differ in order from
where
one another. For the batch and plug flow cases,
the development requires the simultaneous V is the desired product
solution of the pertinent differential equations. Wx and W2 are undesirable products
In the case of stirred tank reactors an analogous
set of simultaneous algebraic equations are To optimize the production of V, one would be
obtained that will generally be nonlinear. In persuaded by the presence of the consecutive
such cases closed-form solutions are generally reactions to use a plug flow or batch reactor.
not available, and computers must be used to On the other hand, since the parallel reaction
obtain numerical solutions. Fortunately, the leading to W2 is of higher order in reactant A
concentration-time curves are similar in shape than that leading to V, a stirred tank reactor is
to those for first-order reactions, and the rules called for. These conflicting considerations
of thumb enunciated previously for that case imply that there will not be a unique solution
may be regarded as general for all irreversible as to the type of reactor and operating conditions
reactions in series. Little can be said about the to be employed. The answer will depend on the
product distribution curves for reactions other feed concentrations available and the values of
than first order, because they depend on the the pertinent rate constants. To illustrate the
initial reactant concentration. One may con- general principles involved in tackling these
clude, as in the case of parallel reactions, that series-parallel design problems, we will find it
high concentrations of reactant favor the higher- instructive to consider two common classes of
order reaction and low concentrations favor the reactions of this type: multiple substitution
lower-order reaction. Variations in the feed reactions and polymerization reactions.
330 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
of original A, the concentration of V will reach see that when all the A that will ever be added is
zero, and we will have a solution containing only present in the mixing vessel, as in modes 2 and 3,
W. Figure 9.8 is a schematic representation of appreciable amounts of intermediate V are
the changes which occur. formed. When it is added incrementally, as in
In mode 3, the reaction rate is slow compared mode 1, no appreciable V is formed. Such-be-
to the mixing process, and one has, in effect, a havior is characteristic of reactions in series, as
well-stirred batch reactor. Initially A and B com- noted in Section 9.2. As far as A, V, and W are
bine to form V, which can then compete with concerned, their behavior is analogous to
other A molecules for the B molecules that re-
main. During the early stages the V molecules w (9.3.5)
are at a numerical disadvantage, and more V will From a comparison of the qualitative results
be produced than is consumed. As its concentra- of modes 2 and 3, we see that the concentration
tion rises, V competes more successfully, even- level of B does not have a major influence on the
tually passing through a maximum and declining product distribution and the reaction sequences
thereafter. The general behavior of this mode of involved. (It will, however, influence the overall
mixing in terms of product distribution is quite conversion rate.) This behavior is characteristic
similar to that of mode 2, which was depicted in of parallel reactions of the same order with
Figure 9.8. respect to a particular species. From the view-
point of species B, reactions 9.3.3 and 9.3.4 may
be regarded as
B: (9.3.6)
W
Levenspiel (8) has discussed these same ex-
amples and has proposed the following very
useful rule.
Series-parallel reactions can be analyzed in terms
of their constituent series reactions and parallel
Time or moles B added differentially reactions in that optimum contacting for favor-
able product distribution is the same as for the
Figure 9.8 constituent reactions.
Schematic representation of product distribution
versus time for parallel-series combination of For the reaction set 9.3.3 and 9.3.4 where V
reactions—Case II: parallel component added is the desired product, the rule indicates that a
slowly. (Adapted from Chemical Reaction Engi- mixture containing A, which has reacted, should
neering, Second Edition, by O. Levenspiel. Copy- not be back-mixed with fresh A, while B may, be
right © 1972. Reprinted by permission of John added in any fashion which is convenient. One
Wiley and Sons, Inc.) must apply the maxim with discretion, however,
and where possible should work out the mathe-
The product distributions shown in Figures matics appropriate to the reaction set involved
9.7 and 9.8 are obviously quite different, so we in order to obtain as much insight as possible
again see that the manner of contacting can into the factors that will influence the product
have significant effects on product yields. We distribution.
332 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
We are now prepared to develop quantitative which gives Cw. A material balance on species
relations for series-parallel reactions of the B gives
multiple substitution type considered above. QBO ~~ CB = (Cv — Cvo) + 2(CW — Cwo)
Case I. Quantitative Treatment for Plug Flow (9.3.12)
or Batch Reactor
and this provides a means of evaluating CB.
The first step in the determination of the product
Case II. Quantitative Treatment of Stirred
distribution for reactions 9.3.3 and 9.3.4 is an
Tank Reactor
evaluation of the instantaneous yield.
vA dCv When reactions 9.3.3 and 9.3.4 take place in a
V— — single continuous stirred tank reactor, the route
vvdCA
to a quantitative relation describing the prod-
dCy uct distribution involves writing the design
dt — k22CCvC
vCB
equations for species V and A.
~dCA
T = (9.3.13)
dt - r4 - rv
(9.3.7) or
Hence, for a plug flow or batch reactor, -AO
Cy
cAO =—V- (F 1 ) - ^ +^(^ L ) for k2 / k1 (9.3.9)
and
Cy_
(9.3.10)
cAO cAO.
One of the last two equations gives the relation- which is the difference equation analog of
ship between Cv and CA at any time in a plug equation 9.3.8. Equation 9.3.15 may be solved
flbw or batch reactor. The stoichiometry of the for Cv in terms of CA to give
system provides the additional information
necessary to describe completely the system + 1 -
composition. Cy
cA cAO (9.3.16)
A material balance on species A indicates that
- 1
Q 40 = CA Cv + Cw
cA
Equations 9.3.11 and 9.3.12 are also applicable
(9.3.11) to a CSTR reactor, since they represent overall
9.3 Series-Parallel Reactions 333
material balances. They provide the additional consumption of B is independent of the manner
relations necessary to determine the complete in which B is added. It depends only on the total
effluent composition. consumption. If one recognizes that the 45-
It is possible to represent the product distri- degree line would represent complete conver-
butions for both the batch and CSTR cases in sion to the desired product, it is evident that the
the form of time independent plots, as shown in highest fractional yields are obtained at low
Figures 9.9 and 9.10. The plots are prepared using fraction conversions of species A, regardless of
equations 9.3.9 (or 9.3.10), 9.3.11, 9.3.12, and whether k2/k1 is large or small. However, the
9.3.16. As the reaction proceeds, B is consumed, larger the value of k2/ku the faster the fractional
and one moves from left to right along the curve yield of V drops off with increasing conversion
representing the appropriate value of k2/kl. The of A. Thus, if it is possible to remove small
dashed lines of slope 2 on the figures indicate amounts of V cheaply from large volumes of the
the amount of B consumed to reach a particular reaction mixture, the optimum reactor configu-
point on the curve being followed. This value of ration and mode of operation would involve the
V + B —--+ W
Batch or plug flow
0.1
<\/---'\i
0 0.1
v
0.2
V. V.
0.30.4 0.5 0.6 0.7 0.8 0.9 1.0
Figure 9.9
Product distribution in batch or plug flow reactors for the indicated reac-
tions (Adapted from Chemical Reaction Engineering, Second Edition, by
O. Levenspiel. Copyright © 1972. Reprinted by permission of John Wiley
and Sons, Inc.)
334 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Figure 9.10
Product distribution in a continuous stirred tank reactor for the indicated
reactions. (Adapted from Chemical Reaction Engineering, Second
Edition, by O. Levenspiel. Copyright © 1972. Reprinted by permission
of John Wiley and Sons, Inc.)
use of a plug flow reactor with low conversions will still be along the same k2/kl line. However,
of A per pass coupled with a separator to remove the starting point will no longer be the point
the product V and to recycle unconverted reac- Cv = 0, CA = CA0, but the intersection of the
tants. The exact conversion level to be employed curve corresponding to the value ofk2/k1 charac-
will depend on an economic analysis of the teristic of the system and a straight line passing
combined reactor-separator system. through the point Cv = CA = 0, with a slope
As Levenspiel points out in his discussion of ( — CV0/CA0). The effect of the presence of inter-
this same reaction network, it is relatively easy mediate V in the feed is to reduce the net
to extend the use of figures like Figure 9.9 (PFR) fractional yield of species V.
to cases in which intermediates may be present Illustration 9.4 indicates how the concepts we
in the feed to the plug flow reactor either by have developed may be used in attempting to
virtue of their presence in a recycle stream or in develop a rational reactor design for carrying
the raw feedstream. The progress of the reaction out multiple substitution reactions.
9.3 Series-Parallel Reactions 335
ILLUSTRATION 9.4 DESIGN above, the product mixtures will contain sub-
CONSIDERATIONS FOR A SPECIFIED stantial amounts of unreacted A and B. It is
PRODUCT DISTRIBUTION thus appropriate to consider the questions of
separation and recycle for each species in turn.
Statement
Reactant A
For a set of parallel-series reactions represented HA has significant economic value then it should
b
y be separated from the reactor effluent stream and
A+B^V n ^kxCACB recycled for subsequent use. Since the conver-
sion level is higher in the plug flow reactor, the
V + £ -i W r2 = k2CvCB recycle rate will be much smaller and the de-
it is desired to operate under conditions such mands on the separation equipment for reclaim-
that the relative yield of V is 75%, based on the ing species A will also be somewhat smaller.
amount of A that reacts. If the reaction rate Even when species A is of relatively little eco-
constants are numerically equal, indicate briefly nomic value, there may be circumstances when
the constraints within which the idealized re- the costs associated with meeting the pollution
actor types must operate. Consider conversion control requirements for the process effluent
levels, mole ratios of the feed, and the possibility will dictate separation and recycle of this reac-
of separation and recycle of reactant or product tant as the most economic alternative.
species. It may be presumed that species A is
the limiting reagent. Reactant B
The product distribution is insensitive to the
Solution concentration of reactant B. If B is cheap and
does not offer a potential pollution problem
Figures 9.9 and 9.10 are useful for determin- downstream, its concentration may be kept at
ing the conversion level at which each type any convenient level. If B is costly or must be
of reactor should be operated. The locus of removed for other reasons, one has the options of
points along which the relative overall yield operating with low B concentrations at high
V is 75% is given by the straight line linking conversions in a relatively large reactor to
the point fA = 0, CV/CAO = 0 to the point produce a product containing very little B, or
fA = 1,CV/CAO = 0.75. Points above this straight of operating at higher B concentrations in a
line will correspond to yields in excess of the smaller reactor with separation and recycle of
minimum specified value of 75%. The point unused B. The specified product distribution
where this straight line intersects the yield curve requires that the mole ratio of V to W be 3:1.
for the specified value of k2/ki (1.0) indicates the To produce 1 mole of W and three of K, one
highest conversion level at which the specified must consume 4 moles of A and 5 moles of B.
product distribution can be achieved. When such The feed ratio employed in an actual situation
a straight line is superimposed on Figure 9.9, it may differ appreciably from 1.25 to enhance the
is found that for a batch or plug flow reactor, the reaction rate or to allow for discarding some A
maximum possible conversion of A that is con- and£.
sistent with the constraint on the product
distribution is fA = 0.42. For a CSTR the cor- Products V and W
responding maximum conversion level may be Separation and recycle of species V or W has
found using Figure 9.10. In this case/^ = 0.25. no merit. Neither product enhances the reac-
At the limiting conversion levels determined tion rate. The low conversion of A that we are
336 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
forced to employ is a consequence of the fact that molecule and as a parallel reaction from the
otherwise there would be insufficient V in the viewpoint of the monomer molecules being
product mixture. consumed. Of course, if polymer molecules can
react with one another,
residence time and consequently will reach a For cases where the growth period is the same
molecular weight greater than the average value. as the residence time in the reactor, as in poly-
Whether the first or the second factor domi- condensation processes, the residence time dis-
nates depends on the type of polymerization tribution is the dominant factor influencing the
process involved. If the period during which molecular weight distribution. In this case one
the polymer molecule is growing is short com- obtains a broader molecular weight distribution
pared to the residence time of the molecule in from a CSTR than from a batch reactor. Figure
the reactor, the first factor dominates. This sit- 9.12 [also taken from Denbigh (11)] indicates
uation holds for many free radical and ionic the type of behavior expected for systems of this
polymerization processes where the reaction type.
intermediates are extremely short-lived. Figure
9.11, taken from Denbigh (10), indicates the 1 1 1 1
types of behavior expected for systems of this
type.
is
1 1
P£ I Batch
1 11
1 1 1 1 1
V
11
c _ J\ I Batch reactor _ \
o 0.8
3nz:ati
1 1
1 \ flow reactors
0.7 —I \ —
i i Continuous
11
III
CD 0.6
0)
\ 1
S5 ~7 10 20 30 40
_C
0.5 Degree of polymerization, r
> 0.4
4 / \\ni
1 —
Figure 9.12
Molecular weight distribution function for the case
o
Q. where the length of the growth stage is long
~co
0.3
In \W compared to the reactor residence time or for the
tot
o.i Hi/ \ \ \ —
i
0 f 1 1 ^ H ^ 1 1
100 200 300 400 500
9.4 REACTOR DESIGN FOR
Degree of polymerization, r
AUTOCATALYTIC REACTIONS
Figure 9.11
Molecular weight distribution function for the case 9.4.1 Basic Concepts
where the length of the growth stage is short
compared to the residence time in reactor. (Re- There are many reactions in which the prod-
printed with permission from Chemical Reactor ucts formed can themselves act as catalysts for
Theory, by K. G. Denbigh and J. C. R. Turner. the reaction. Thus there will be a range of com-
Copyright (c) 1971 by Cambridge University Press.) positions for which the reaction rate accelerates
338 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
as the reaction proceeds. This phenomenon is Under normal circumstances, when a material
known as autocatalysis, and reactions of this reacts, its initial rate of disappearance is high, but
type pose some interesting problems in the selec- the rate decreases progressively as the reactant
tion of an optimum reactor configuration. is consumed. In an autocatalytic reaction, on
An autocatalytic reaction is one in which the the other hand, the initial rate is relatively slow,
reaction rate is proportional to a product con- because little or no product is present. The rate
centration raised to a positive exponent. Some increases to a maximum as products are formed
of the first articles in the literature of chemical and then decreases to a low value as reactants
kinetics deal with reactions of this type. For are consumed or equilibrium is achieved. If
example, in 1857, Baeyer (12) reported that the there are no product species present in the initial
reaction of bromine with lactose was autocata- reaction mixture, autocatalytic reactions exhibit
lytic. The hydrolyses of several esters also fit the type of behavior shown in Figure 9.13a. If
into the autocatalytic category, since the acids the product species that is catalytic is present in
formed by reaction give rise to hydrogen ions the original reaction mixture, the type of be-
that serve as catalysts for subsequent reaction. havior that the system will exhibit is shown in
Among the most significant autocatalytic reac- Figure 9.13b.
tions are the fermentation reactions that involve It should be noted that we have not let the
the action of a microorganism on an organic initial rate become zero in Figure 9.13 even when
feedstock. no catalyst is present, because the possibility
Rate Rate
Time, t max
Reactant concentration
Rate Rate
Time, t
Reactant concentration
(b)
Figure 9.13
Characteristics of autocatalytic reaction, (a) No products present in
initial reaction mixture, (b) Some product species present in initial
mixture.
9.4 Reactor Design for Autocatalytic Reactions 339
CH 3 CH 3 XH3
CH—O CH—N- CO 2
I I (A)
CH 2 C, CH2 C OH
\ / O \ /N
NH NH
VR. No special constraint is to be placed on the The right side of this expression is identical with
volumetric feed rate or the initial concentration the rate of production of species B and C. Hence
of species A. the maximum production rate for a fixed reactor
volume occurs when the reactor contents have
Solution a composition that maximizes the specific reac-
tion rate. Now, in terms of the fraction
The observed rate of reaction is identical with
conversion,
the rate of formation of species C
(B) = CMfA (L)
r = kM? + k3(AB)2 (F) or,
this species. In the hydrolysis of an ester formed The maximum rate may be evaluated by com-
from a weak acid, an order approaching one bining equation 9.4.4 or 9.4.5 with equation
half may be observed. 9.4.1.
The key characteristic of all autocatalytic rate
expressions is that plots of the rate versus time 9.4.2 Reactor Design Considerations
are of the general form shown in Figure 9.13.
If converted into plots of fraction conversion Plug flow reactors generally give greater pro-
versus time, these forms give rise to a charac- duction capacities than CSTR's of equal volume
teristic S shape. These plots first rise, showing but, in the case of autocatalytic reactions, this
autoacceleration as the rate increases, then pass generalization is not valid. For this class of
through an inflection point as the rate reaches a reactions back-mixing of reacted material with
maximum, and finally taper off so that the frac- fresh feed is often beneficial in optimizing the
tion conversion approaches unity or its equili- overall reactor design. This mixing can be
brium value as the time approaches infinity. achieved by employing a continuous stirred tank
In the more general case for liquid phase reactor or by recycling the unseparated product
reactions or other cases where S = 0, the auto- mixture. With the CSTR it is possible to operate
catalytic term in the reaction rate expression all the time at the highest reaction rate (i.e., at the
can be written as points in Figure 9.13 labeled C max ). One could
thus require a lower volume than would be re-
= k(C (9.4.1) quired to achieve the same conversion level in
A0
where A and P are reactant and product species, a plug flow reactor, which would operate at an
respectively. The extent of reaction per unit average rate less than the optimum. This mode
volume corresponding to the maximum auto- of operation would require an effluent com-
catalytic rate may be determined by setting the position corresponding to Cmax. If lower con-
derivative equal to zero. versions are desired, the CSTR will still require
dr A
•\CP0 + vPnPp + PPVp(CA0 + vAt*YA(CP0 + = 0
(9.4.2)
determine the relative costs of operating in the you maximize the formation of V. In what
separation and recycle mode vis a vis the costs manner would you carry out this experi-
of .utilizing the second stage plug flow reactor ment; that is, how would you mix the re-
and the attendant separation costs necessary to actants and at what rate?
obtain the final product. (b) If you desire to produce V in a flow reactor,
what type of reactor and operating condi-
LITERATURE CITATIONS tions do you recommend?
(c) If the activation energies for the rate con-
1. Levenspiel, O., Chemical Reaction Engineering, Second
Edition, pp. 166-167, Wiley, New York, copyright ©
stants kl9 k2, and k3 are 30, 25 and 25 kJ/
1972. Used with permission. mole, respectively, what additional state-
2. Denbigh, K. G., and Turner, J. C. R., Chemical Reactor ment can you make regarding recommended
Theory, pp. 123-124, Cambridge University Press, operating conditions?
Cambridge, 1971.
3. Trambouze, P. J., and Piret, E. L., AIChE J., 5 (384), 2. Consider the following chlorination of ben-
1959. zene reaction system of elementary reactions.
4. Denbigh, K. G., Chem. Eng. Sci., 77(25), 1961.
C\2 + C 6 H 6 ^ C 6 H 5 C 1 + HC1
5. Levenspiel, O., op. cit., p. 180.
6. Levenspiel, O., op. cit., p. 181. Cl 2 C 6 H 5 C1 ^ C 6 H 4 C1 2 HC1
7. Levenspiel, O., op. cit., pp. 186-192. Assume that chlorination of dichlorobenzene
8. Levenspiel, O., op. cit., p. 189. does not occur to an appreciable extent. Mono-
9. Levenspiel, O., op. cit., pp. 191-192. chlorobenzene is the desired product.
10. Denbigh, K. G., and Turner, J. C. R., op. cit., p. 115. (a) Obtain an equation that relates the concen-
11. Denbigh, K. G., and Turner, J. C. R., op. cit., p. 114. tration of monochlorobenzene to the con-
12. Baeyer, A., Ann. Chem. Pharm., 103 (178), 1857. centration of benzene, assuming that only
13. Walles, W. E., and Platt, A. E., Ind. Eng. Chem., 59 benzene and chlorine are present initially:
(6), p. 41, 1967. (1) ForaCSTR.
(2) For a batch reactor.
PROBLEMS (b) Utilizing this information, for each of the
above cases (1 and 2) in order to minimize
1. Your plant has two liquid streams available the rate of production of dichlorobenzene
containing solutes that are not profitably mar- relative to the rate of production of mono-
ketable at the present time. The first contains chlorobenzene, would you want a high or
solute A, the second solute B. low conversion of benzene? Prove this math-
Two possible reactions may occur between ematically for each of the following cases.
these species. The stoichiometric equations and
rate expressions are given below. kx > k2
k, =k2
A + B -• V (valuable product)
k, < k2
rv = k,CACBC\)2 + k2CACB
A + B —» U (undesired byproduct) (c) For an industrial situation, what other con-
sideration is particularly important in addi-
rv = k3CACB
tion to minimizing the rate of production of
(a) You have two beakers containing samples of undesired products relative to that of the
the two streams and desire to carry out a desired product? Answer this question very
small-scale laboratory experiment in which briefly in a qualitative manner.
Problems 343
3. Species A and B react as follows. the whey. It is believed that if very high tempera-
tures are used for short times, this result can be
A + 2B k-X 2R + S (-rA)i = KCAC\ obtained without degrading the protein of the
whey, which, besides ruining the food value,
A + B^T + U (-rA)2 = k2CACB
gives it a bad taste.
An equimolar mixture of A and B is introduced A PFR has been designed in which steam is
into an infinitely long plug flow reactor (CA0 = injected into the whey as it enters the reactor.
CB0 = 1.0 kmole/m 3 ). The concentration of A After it passes through the PFR, the temperature
in the effluent is 0.1 kmole/m 3 . No B is present in is dropped rapidly to 10 °C, where the rate of
the effluent stream. What conclusions can you degradation is negligible. A holding time of 7 sec
draw concerning the relative magnitudes of k1 is needed to obtain 99.99% spore kill at a tem-
and/c 2 ? perature of 127 CC. For the whey to meet health
and flavor tests, a maximum of 10% degradation
4. The reactions indicated below take place of protein is to be allowed. As a first approxima-
in the liquid phase with the indicated rate tion in your analysis, the reaction is to be treated
expressions. as isothermal at 127 °C.
Another research group has previously con-
ducted tests on the denaturation kinetics and
(ii) 2A -+ C r = k2CA2 k2 = 2 m3/kmole-ksec has reported the data below.
The feed stream consists of A dissolved in a Will the proposed process meet the standards
solvent S such that the initial concentration of A set for protein degradation?
is 2 kmoles/m 3 . If 50% of the A fed to a flow
reactor undergoes reaction, determine the effl- Data
uent composition: Whey contains 1.5 kg/m 3 of lactalbumin and
(a) IfaCSTRisused. 3.0 kg/m 3 of lactoglobulin. At 85 C the follow-
(b) If a PFR is used. ing data are available on the kinetics of protein
degradation, using initial concentrations that
5. Xylene isomerization reactions can be accom- are the same as those in whey.
plished by contacting a hot gas stream with a
solid catalyst. Under these conditions the isom-
Lactalbumin Lactoglobulin
erization reactions may be regarded as reversible
and first-order. Unfortunately, the catalyst also Percent Percent
catalyzes disproportionation reactions. These Time, t residual Time, t residual
reactions may be regarded as essentially second- (sec) protein (sec) protein
order and irreversible. If one desires to achieve
an equilibrium mixture of isomers with minimal 0 100 0 100
material losses due to disproportionation, what 120 82.5 20 83
do you recommend concerning the mode in 250 72.5 35 77
which one should operate a continuous flow 360 65.0 80 67.5
725 47.5 100 58.0
reactor?
850 42.0 125 54.5
6. A process for ultra-high temperature steriliza- 960 39.0 140 52.5
1075 35.0 160 48.0
tion of whey has been proposed. The purpose of 1200 30 180 44.5
the process is to destroy 99.99% of the spores in
344 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
Over the temperature range of interest, the acti- (2) If the reaction at 2500 °K may be con-
vation energy for the lactalbumin degradation sidered as occurring at constant volume,
reaction is 25.1 kJ/mole while that for the lacto- determine the times corresponding to
globulin degradation reaction is 121 kJ/mole. the two maxima. Note that the time in the
Note that there are problems associated plasma jet will be exceedingly small,
with the determination of time zero for these (c) Roughly what fraction of the methane is
reactions. converted to acetylene at 2500 °K if the
Assume that the orders of both reactions are contact time in a plasma jet is that corre-
integers, although not necessarily the same sponding to the maximum in acetylene
integer. concentration?
7. Gulyaev and Polak [Kinetics and Catalysis, 8. Consider the sequential first-order reactions
fci k2
6 (352), 1965] have studied the kinetics of the A -• V -> W where V is the desired product.
thermal decomposition of methane with a view These liquid phase reactions are to be carried
toward developing a method for the commer- out in a cascade of two equal volume CSTR's in
cial production of acetylene in a plasma jet. The series. If the reactors are to be sized so as to
following differential equations represent the maximize the concentration of species V in the
time dependence of the concentrations of the effluent from the second reactor, determine the
major species of interest. reactor volumes necessary to process 500 gal/hr
d(CR4 of feed containing 6 moles/gal of species A. No
= -MCH4) V or W is present in the feed. What fraction of the
dt
A ends up as VI The rate constants k1 and k2
are both equal to 0.5 hr" 1 .
^ = -/c 3 (C 2 H 4 )
dt 9. A sequential isomerization reaction is be-
d(C2H2) lieved to be characterized by the following mech-
= -/c 4 (C 2 H 2 ) + fc3(C2H4) anistic equations.
dt
A -+ B -> C
(a) Integrate these differential equations to de-
termine the time dependent behavior of each These reactions are being studied in a series
species concentration. Assume that the only combination of two CSTR's. Each of these re-
species present initially is methane and that actors has a volume of 2 liters. Liquid is being
its initial concentration is (CH 4 ) 0 . fed to the first reactor at a rate of 1 liter/min.
(b) (1) Determine the times corresponding to This liquid contains only pure A and the solvent.
maximum concentrations of C 2 H 4 and The inlet concentration of A is 2 g moles/liter.
C 2 H 2 . It may be noted that Under the conditions of operation, k1 = 0.5
min" 1 and k2 = 0.25 min" 1 . Both reactors op-
k3 » fc4 erate at the same temperature.
and that (a) What is the composition of the effluent from
the first reactor? Include all species of
k, = 4.5 x lO 13 *- interest.
k3 = 2.57 x l O V (b) Repeat part (a) for the second reactor.
fe4 = 1.7 x lO 6 *" 1 10. Kostyuk, Belen'ku, and Tezhneva [Kinetics
where T is expressed in degrees Kelvin. and Catalysis, 10 (53), 1969] have studied the
Problems 345
OH
+ CH3COCH3
(acetone)
CH 3 —C—CH 3
O—OH
(p-hydroxycumyl hydroperoxide)
and
OH OH
CH 3 COCH 3
CH 3 —C—CH 3 OH
I
O—OH (hydroquinone)
Both reactions are first-order in catalyst con- 11. Vaidyanathan and Doraiswamy [Chem.
centration and in the reactant involved. If one Eng. ScL, 23 (537), 1966] have studied the cat-
incorporates the catalyst concentration into the alytic partial oxidation of benzene in a composi-
rate constants k1 and k2, the following values tion range where the reactions of interest all
are typical of the results reported by these in- follow pseudo first-order kinetics. The pertinent
vestigators at 40 °C. stoichiometric equations are
^ = 0.95 C 6 H 6 + |O 2 2H 2 O
C 4 H 2 O 3 + 2CO2
kt = 3.6 x 10 sec
C 4 H 2 O 3 4- 3O2 4CO 2 + H 2 O
C6H6 6CO2 + 3H 2 O
Both reactions take place in liquid solution. If
one desires to carry out these reactions in a con-
tinuous stirred tank reactor, determine the space The corresponding rate expressions are
time corresponding to a maximum yield of
p-hydroxycumyl hydroperoxide. If the initial
p-diisopropylbenzene dihydroperoxide concen-
tration is 10 moles/m3, what are the concentra-
tions of the various species in the effluent? r3 = k3PB
346 Selectivity and Optimization Considerations in the Design of Isothermal Reactors
taken as the rate of disappearance of monomer (a) Derive equations for the effluent concentra-
tion of each species in terms of the rate
constants kx and klh the reactor volume, the
initial volumetric flow rate, and the initial
show that the average rate of polymerization concentrations of species A and B.
is given by (b) What reactor volume will give the highest
concentration of species C in the effluent
= NokrkaP2(2kd + krP) stream? What will this concentration be?
(krP + kd)(kaP + kd)
The breadth of the molecular weight distribu- 15. An autocatalytic reaction is to be carried out
tion may be measured by the ratio MJMn. Show in aqueous solution in two identical continuous
that this ratio is given by stirred tank reactors operating in series. The
reaction stoichiometry is
S
- - + \ + A A -> B
w y y
M. while the reaction rate expression is
2+ -
r = kCACB
where If the system is to operate isothermally at 50 °C
where the reaction rate constant is equal to
y=
LP 0.9 m3/kmole-ksec, determine the reactor vol-
ume necessary to achieve an overall fraction
By considering the limits of 1/y as zero and
conversion of 0.80. Species A is to be fed at a
infinity corresponding to infinitely rapid desorp-
rate of 0.3 mole/sec and an initial A concentra-
tion or slow reaction and very slow desorption
tion of 2 kmoles/m3.
or rapid reaction, respectively, show that
16. An autocatalytic reaction represented by the
mechanistic equations
which is a rather narrow molecular weight A + R + R
fc-i
distribution. is being carried out in two identical stirred tanks
operating in series. The reaction is reversible and
14. The following reactions take place in the exothermic. The reaction stoichiometry is
liquid phase in a continuous stirred tank reactor
operating at steady state. A -• R
rate constants kx and /c_x in units of cubic feet The reaction rate expression for this stoichio-
per pound mole per minute. metric equation is
r = kCACB
Temperature
/<- The feed to the CSTR contains a concentration
of A of 2.0 kmoles/m3, while neither species B
70 0.55 ? or C is present in this stream. Both reactors
85 2.90 0.078 operate isothermally at the same temperature.
100 ? 0.625 Steady-state operation may be assumed.
The CSTR and the PFR each have a volume
If the reactor volume is 50 ft3, determine the of 0.1 m 3 . The rate at which A is fed to the first
composition of the stream leaving each reactor. reactor is 150 kmoles/ksec. Note that this is an
autocatalytic reaction in which a product acts
as a catalyst for subsequent reaction.
17. Consider the following combination of
(a) If the fraction conversion leaving the CSTR
reactors. is 0.25, what is the reaction rate constant?
(b) What is the effluent composition from the
second reactor?
(c) It has been suggested that the rate of produc-
tion of species C by this network can be
increased by operating the first reactor at a
CSTR PFR fraction conversion of 0.45. To what value
must the input feed rate of A be changed
under these operating conditions? Will the
It has been suggested that the following liquid rate of production of species C by this com-
phase reaction be carried out in this reactor bination of reactors be increased under these
network. circumstances? The composition of the feed-
A -+ B + C stream is to remain unchanged.
1 f\
\J
Temperature and Energy Effects
in Chemical Reactors
10.0 INTRODUCTION maintain the temperature within at least mod-
The energy changes associated with chemical erate limits. Adiabatic operation is possible only
reactions play an extremely important role in when the concomitant temperatures do not
the design of commercial scale reactors. Even rise so high that the rate becomes excessive or
in those cases where one desires to use a mode drop so low that the rate becomes unpractically
of operation approximating isothermal be- low. This mode of operation is favored when
havior, energy balance considerations are the enthalpy change accompanying the reaction
important in determining the heat transfer re- is small in magnitude, when it is possible to
quirements of the process. In laboratory scale adjust the initial temperature of the reactant
experiments it is relatively easy to maintain mixture to a level where subsequent changes
substantially isothermal conditions because of will not take the mixture out of the workable
the large surface to volume ratio of small equip- range, or when a solvent or other inert materials
ment and because the economic considerations can be added to the reactant mixture to mod-
of the heat transfer process are unimportant. erate temperature changes via sensible or latent
In industrial scale equipment, both physical heat effects. In some cases it is advantageous
limitations on the rate of heat transfer and the to divide the reactor into several stages, each
economics of this process can become quite of which operates adiabatically but between
important, particularly when large enthalpy which heat exchangers are used to heat or cool
changes accompany the reaction. the reactant mixture. In commercial scale equip-
Several factors govern the temperature range ment it is often impractical to try to maintain
in which one may choose to operate a com- truly isothermal conditions, but one can often
mercial reactor. The dependence of the reaction achieve adequate control at a satisfactory level
rate expression on temperature and the position of the reaction rate by exchanging heat with
of chemical equilibrium are two key factors external heat reservoirs.
influencing the choice of temperature level. The For nonisothermal reactors the key questions
temperature dependence of the main reaction that the reactor designer must answer are:
and of important side reactions will govern the (1) How can one relate the temperature of the
selectivity of the conversion. Other properties reacting system to the degree of conversion
of the reaction mixture may also influence the that has been accomplished? and (2) How does
choice of operating temperature through sec- this temperature influence the subsequent per-
ondary considerations. For example, the dew formance of the system? In responding to these
point of a gaseous mixture, the bubble point of questions the chemical engineer must use two
a liquid mixture, the temperature at which a basic tools—the material balance and the en-
liquid separates into two immiscible phases, ergy balance. The bulk of this chapter deals
and the temperature dependence of corrosion with these topics. Some stability and selectivity
reactions all can limit the temperature range considerations are also treated.
within which one might choose to operate. Cat-
alyst activity, selectivity, and deactivation must 10.1 THE ENERGY BALANCE AS APPLIED
also be considered in selecting the operating TO CHEMICAL REACTORS
temperature for catalytic processes. The principle that energy must be conserved
In order to achieve adequate control during has been expressed by innumerable textbook
the operation of a reactor, it is necessary to authors in a number of mathematical forms,
349
350 Temperature and Energy Effects in Chemical Reactors
depending on the particular orientation and work interaction can take place only at those
interests of the individual concerned. One form portions of the system boundary that are im-
that is especially useful for chemical engineering permeable to the flow of matter. For further
applications is the following. discussion of general forms of the energy bal-
vl + 4- — +4- — ZZ (10.1.1)
dt -o-*.+ .i. t + e r
input
min -
output
+
2gc gc °
streams streams
where
-— is the kinetic energy per unit mass
inlet
=E mn (10.1.3)
streams
— Z is the gravitational potential energy per
In most circumstances of interest to the de-
y<-' unit mass
signer of chemical reactors, the kinetic and po-
m is the mass flow rate into or out of the tential energy effects are negligible as is the
system shaft work term. Under these circumstances,
subscripts in and out refer to the conditions 6= I /i in m in (10.1.4)
that prevail at the various points of entry inlet
streams
and exit, respectively.
Recall that the term shaft work refers to me- For the vast majority of situations of interest
chanical forms of energy that are interchanged in the design of industrial scale chemical reac-
between the system and the surroundings by tors, equation 10.1.4 provides an adequate ap-
means of a shaft that protrudes from the equip- proximation to the true situation for continuous
ment and either rotates or reciprocates. A shaft flow reactors operating at steady state. Since
10.2 The Ideal Well-Stirred Batch Reactor 351
the right side of this expression represents the The vast majority of the reactions carried out
difference between the total enthalpy that leaves in industrial scale batch reactors involve reac-
the system in time dt and the total enthalpy tants in condensed phases. Since the specific
that enters during the same time, one often volumes of both liquids and solids are very
sees the statement that small, the difference between internal energy
and enthalpy for these materials is usually
q « AH (10.1.5) negligible. Thus one often sees the statement
where that for batch reactions taking place at constant
volume:
q is the amount of heat transferred per unit
Q « AH (10.1.8)
mass of entering material
AH is the difference between the enthalpies per where the A now represents the difference be-
unit mass of the exit and entrance streams tween the final and the initial values of the
enthalpy of the system. For both batch and
It should be emphasized that the enthalpy flow reactors, the enthalpy changes that occur
change includes not only sensible heat effects during the course of the reaction are then im-
but also a heat-of-reaction term and in some portant in determining the heat transfer require-
cases pressure effects. If there are multiple inlet ments of the system under consideration. The
and outlet streams, appropriate averaging tech- methods used in determining the enthalpy
niques must be used to employ this equation. changes that accompany chemical reactions
For batch reactors the appropriate form of were discussed in detail in Section 2.2.
the first law to use is that for closed systems. The rate at which heat is transferred to a
system can be expressed in terms of an overall
(10.1.6) heat transfer coefficient U, the area through
where which the heat exchange occurs and on which
U is based, and the difference between the
A£ total is the change in the total energy of the temperature of the heat source (or sink) Tm,
system and that of the reactor contents T.
Q is the heat transferred from the surround-
Q = UA(Tm - T) (10.1.9)
ings to the system
W is the work done by the system on the This expression may be combined with equa-
surroundings tions 10.1.4 and 10.1.8 in order to analyze the
different situations that may arise in operating
If one again takes note of the fact that work the various types of ideal reactors. These anal-
effects and kinetic and potential energy effects yses are the subject of Sections 10.2 to 10.4.
are usually negligible in chemical reactors,
equation 10.1.6 simplifies to
10.2 THE IDEAL WELL-STIRRED
(10.1.7) BATCH REACTOR
where
The key assumption on which the design anal-
AE is the change in the internal energy ysis of a batch reactor is based is that the degree
of the system of agitation is sufficient to ensure that the com-
subscript V is used to indicate that the volume position and temperature of the contents are
is assumed to be substantially con- uniform throughout the reaction vessel. Under
stant these conditions one may write the material
352 Temperature and Energy Effects in Chemical Reactors
and energy balances on the entire contents of to liquid phase reactions where the restrictions
the reaction vessel. of constant volume and constant pressure go
If one considers a batch reactor in which the together and to gas phase reactions occurring
chemistry is characterized by a single extent of at constant pressure. We should again emphasize
reaction, the material balance analysis presented that the enthalpy change appearing in equation
in Section 8.1 indicates that the holding time 10.2.4 contains both sensible heat terms and
necessary to change the fraction conversion heat of reaction terms. Consequently, one can
from fA1 to fA2 is given by interpret equation 10.2.4 as implying that
l
a, r 0 Cpi dT (10.2.13)
(10.2.10)
If the partial molar heat capacities are sub-
where nA0 is the number of moles of the limiting stantially constant over the temperature range
reagent charged to the batch reactor. of interest, this equation may be solved to deter-
On a differential basis, mine the relationship between the temperature
and the fraction conversion.
- T) = nA0 —
dt
1
The total processing time per batch is thus essen-
fA T (from B) k (from C) *U ~ /A) *U " /A)
tially equal to 5 hr. While operating 7000 hr one
will be able to process 7000/5 or 1400 batches. 0.00 436 0.8 0.8 1.250
Each batch must contain 2,000,000/(0.97)(1400), 0.05 444.3 1.49 1.42 0.706
or 1473 lb. Since the liquid has a density 0.10 452.6 2.73 2.46 0.407
of 0.9 g/cm3, the required volume will be 0.15 460.9 4.87 4.14 0.242
1473/(0.9)(62.5), or 26.2 ft3. This volume is 0.20 469.2 8.51 6.81 0.147
equivalent to 196 gal. 0.25 477.5 14.60 10.95 0.091
The maximum heat flux that will have to be 0.30 485.8 24.6 17.2 0.058
0.35 494.1 40.7 26.5 0.037
maintained in order to operate isothermally will
0.40 502.4 66.3 39.8 0.025
be that generated at the start of the reaction. 0.45 510.7 106.1 58.4 0.017
Q = AHRrVR = AHRkCA0VR 0.50 519.0 167.5 83.8 0.012
0.55 527.3 260.6 117.3 0.009
or 0.60 535.6 400.0 160 0.006
Q = AHRknA0 0.65 543.9 605.6 212 0.005
0.70 552.2 905.9 272 0.004
In consistent units 0.75 560.5 1339 335 0.003
0.80 568.8 1957 391 0.003
AH = - 0.85 577.1 2828 424 0.002
g 252 cal lb 0.90 585.4 4045 405 0.002
BTU 0.95 593.7 5729 286 0.003
= -149.5 0.97 597.0 6567 197 0.005
tT
356 Temperature and Energy Effects in Chemical Reactors
The time necessary to accomplish this exo- The material balance on a single CSTR op-
thermic reaction under adiabatic operating con- erating at steady state may be represented by:
ditions is only an extremely small fraction of
= F
iout ~ (10.3.1)
that necessary for isothermal operation. In fact,
the times necessary to fill and drain the reactor where the rate expression is evaluated at the
and to heat it to a temperature where the rate effluent composition and temperature. If only
becomes appreciable will be greater than that one reaction is taking place one may rewrite this
necessary to accomplish the reaction. Thus, equation in terms of the familiar design equation
Thus, Solution
Solution (D)
The solution to this problem requires a trial
and error iterative procedure. Since both the This equation can be solved for T using a trial
reactor volume and the initial volumetric flow and error procedure. This gives T = 410 °K.
rate are known, the space time per reactor may Thus
be calculated and we may focus our attention
initially on the first reactor.
The volumetric flow rate is equal to
Division by FA0 and combination of the These equations must be solved simultaneously
integrals gives using a knowledge of the temperature depen-
dence of reaction rate expression.
(fA2 - IAI) ^ ^ = L ,dT Note that in the case of the plug flow reactor,
•> 1 1
V A
there may be a variation in the temperature of
or the reactor contents from point to point within
the reactor. This condition is in contrast to the
two ideal reactor cases discussed earlier, where
the temperature of the contents of a single
reactor was assumed to be constant throughout.
. 0,05 The presence of temperature gradients in the
same direction as the flow is by no means con-
trary to the assumption of plug flow. In this
Combining equations B, E, and F gives chapter we will confine our analysis to situations
in which both the temperature and the com-
0.5(T2 - 410)
0.705 position are uniform across the cross section
83 of the reactor. The complications arising from
0.705 + 25(2.61 x the existence of radial temperature gradients
(G) and variations in the composition with radial
1 + 25(2.61 x
position are treated implicitly in conjunction
A trial and error solution gives T2 = 458.5 K with the discussion of the two dimensional
(approximately 186 °C) and fA2 = 0.997. model of fixed bed reactors in Section 12.7. For
With this arrangement one would be able to homogeneous systems the effects are usually
produce in excess of 2 million lb of B annually small.
if side reactions do not occur at this higher If one considers the rate at which heat is
temperature and if the vapor pressure of the being supplied to a differential length of a
solution still lies in a range that does not create tubular reactor, geometric considerations imply
problems. that
This section treats the material and energy where D is the inside diameter of the tubular
balance equations for a plug flow reactor. For reactor.
steady-state operation the energy balance anal- A material balance on this differential reactor
ysis leading to equation 10.1.4 is appropriate. volume yields the following result.
T = (10.4.2)
and the energy balance on this differential
362 Temperature and Energy Effects in Chemical Reactors
T utofelement T ntoelem m
F ° ' C dT
at To dfA (10.4.6)
where the first summation involves the molal tion is a simple example of a Diels-Alder
flow rates of the various species leaving the reaction.
volume element and the second summation
CH 2 CH 2
involves the molal flow rates entering the volume • / / \
The summation involves the effluent molal flow ethylene at 450 °C and 1 atm is fed to a reactor,
rates. This equation and equation 10.4.2 must be determine the space times required to convert
solved simultaneously in order to determine the 10% of the butadiene to cyclohexene for iso-
tubular reactor size and to determine the manner thermal and for adiabatic modes of operation.
in which the heat transfer requirements are to
be met. For either isothermal or adiabatic
Data
operation one of the three terms in equation
10.4.7 will drop out, and the analysis will be Wasserman (6) has reported the following data
much simpler than in the general case. In the for this reaction.
illustrations which follow two examples are
treated in detail to indicate the types of situa- k = io 7 -V27.5oo/jtr Hters/mole-sec
tions that one may encounter in practice and AHR = - 30,000 cal/g mole
to indicate in more detail the nature of the
design calculations. The reverse reaction may be neglected.
The following values of gas phase heat ca-
ILLUSTRATION 10.4 DETERMINATION OF pacities may be assumed to be constant over the
THE VOLUME REQUIREMENTS FOR temperature range of interest.
ADIABATIC OPERATION OF A TUBULAR
REACTOR WITH EXOTHERMIC REACTION CP,C4H6 = 36.8 cal/g mole-°K
Butadiene will react with ethylene in the gas CP,C2H4 = 20.2 cal/g mole-°K
phase at temperatures above 500 °C. This reac- Cp,c6Hl0 = 59.5 cal/g mole-°K
10.4 Temperature and Energy Considerations in Tubular Reactors 363
-r(fj' 0/ ^ , 0 ( l - fAf
(A)
Cyclohexene
Ffi^T - T o ) = FAOfA59.5(T - T o )
The initial reactant concentration can be de- or
termined from the ideal gas law
^CpidT = (57.0F^0 + 2.5fAFAO)(T - T o )
T ^—
kCA
o.i
JA-2 en(l - fA) -
1 "/Jo
kC AO
T = o = 80.19 sec
0.156(8.43 x 10" 3 )
In the case of adiabatic operation the tern- where T is expressed in degrees Kelvin. At a
perature must be related to the fraction con- fraction conversion of 0.1 the gas temperature
version so that equation A can be integrated. will have risen to 775 °K.
364 Temperature and Energy Effects in Chemical Reactors
where T is given by equation C. inside area of the pipe may be taken as 200
Numerical evaluation of this integral gives kcal/(hr-m2-°K). Volumetric expansion effects
T = 47.11 sec. are negligible.
This space time is about 59% of that required Solution
for isothermal operation. This number may be
somewhat low because the reaction is exo- For a plug flow reactor the appropriate design
thermic and the rate of the reverse reaction may equation for the reaction at hand is:
be appreciable at the highest temperatures in- fO-97 dfA _ ™.97 dfA
volved in our calculation. — r>~ JO
(A)
Hl-fA)
We now wish to examine the case where we From equation C of Illustration 10.1,
allow for heat exchange with a substantially k = 2.61 x l O 1 ^ - 1 4 ' 5 7 0 ^ ^ - 1 (B)
constant temperature heat sink (e.g., an evapo- The temperature may be related to the frac-
rating or condensing fluid or a material flowing tion conversion by means of an energy balance
at a velocity such that its temperature change such as equation 10.4.7.
over the reactor length is quite small compared In the present case
to the driving force for heat transfer).
U = 200,000 cal/m 2 -hr°K
ILLUSTRATION 10.5 DETERMINATION OF Tm = 433 °K
THE VOLUME REQUIREMENTS FOR
D = 4(2.54 x 10" 2 ) = 0.1016 m
OPERATION OF A TUBULAR REACTOR
UNDER NONISOTHERMAL CONDITIONS FA0 = 133,700 g/hr (from Illustration 10.2)
WITH HEAT EXCHANGE Cp = (0.5) cal/g-°K
Consider the reaction used as the basis for To = 436 °K
Illustrations 10.1 to 10.3. Determine the volume
required to produce 2 million lb of B annually
in a plug flow reactor operating under the condi-
tions described below. The reactor is to be Division of equation C by FA0 and insertion of
operated 7000 hr annually with 97% conversion the reaction rate expression and numerical val-
of the A fed to the reactor. The feed enters at ues gives
163 C. The internal pipe diameter is 4 in. and
the piping is arranged so that the effective reactor >fA (2 x 10 5 )(433 - T)4 dfA
volume can be immersed in a heat sink main- Jo (0.1016)/cQ0 (1 - fA)
tained at a constant temperature of 160 °C. The
(0.5) dT - S3fA (D)
overall heat transfer coefficient based on the
10.5 Autothermal Operation of Reactors 365
The initial reactant concentration is equal to This volume corresponds to a reactor length
0.9 g/cm3 or 9 x 105 g/m 3 in units that are of 69.6 ft.
consistent with the other parameters we are The temperature of the reactor contents rises
using. from 436 °K at the inlet to 443.20 °K at a point
Substitution of this value into equation D and corresponding to fA = 0.22 and then drops to a
rearrangement gives temperature of 433.19 °K at the reactor outlet.
Equations A, B, and E provide sufficient infor- The maximum temperature is significantly less
mation to determine the necessary reactor vol- than that which would be obtained by adiabatic
ume. However, they do require an iterative trial operation (597 °K). This fact may be important
and error solution. in design considerations when side reactions
A simple approach to solving this set of equa- have significant reaction rates at high tempera-
tions can be represented as follows. tures. The required reactor volume will be
greater than that for isothermal operation be-
1. Consider increments of reactor-volume corre- cause, under the specified operating conditions,
sponding to increments of fraction conver- once a fraction conversion of 0.63 is obtained,
sion fA1,fA2, • • • ,fAn. the temperature of the reactant mixture drops
2. Guess a temperature corresponding to fA1. below the feed temperature. In practice one
3. Calculate values of the integrand in equation might wish to dispense with all or part of the
E corresponding to conversions zero andfAl. cooling capacity in the later portions of the re-
4. Use the trapezoidal rule to approximate the actor. This technique would have the effect of
integral. reducing the required reactor volume. An alter-
5. Calculate the temperature corresponding to native iteration procedure for handling tubular
fAl and iterate until the calculated and as- reactor analyses involving a heat transfer term
sumed values agree. is employed in the case study in Section 13.1.
6. Repeat the procedure for successive incre-
ments in the fraction conversion.
7. Take the functional relationship between T 10.5 AUTOTHERMAL OPERATION
and fA determined above and use it in con- OF REACTORS
junction with equation A to determine the In the design of processes involving exothermic
required reactor volume. reactions it is generally desirable to use the en-
ergy liberated by reaction at some other point
When this procedure is followed, one finds
in the process or to make it available for use
that the space time corresponding to the op-
elsewhere in the industrial plant. For example,
erating conditions outlined above is equal to
it can be used to preheat the feed components,
4.642 hr. The required reactor volume is then
particularly when the reaction takes place at
equal to
high temperatures and the feed components are
VR = xf0 = 4.642(39.3) - 182.4 gal supplied at much lower temperatures. The term
"autothermal operation" is applied to modes of
or processing in which exothermic reactions are
VR = 24.3 ft3 carried out such that the energy released by
366 Temperature and Energy Effects in Chemical Reactors
reaction is fed back to the incoming reactant ments for the various tanks may be of opposite
stream. Use of this term implies that the reactor sign. Some tanks will require a net input of
system is to a large extent self-supporting in thermal energy, while others will need to be
terms of its thermal energy requirements. Con- cooled. It is often useful in such situations to
sequently, it is possible to run at the desired consider the possibility of adiabatic operation
operating temperature without using external of one or more of the tanks in series, remember-
heat sources to preheat the feed. An autothermal ing the constraints that one desires to place on
reaction is akin to the autocatalytic reactions the temperatures of the process streams. An-
discussed in Section 9.4 in the sense that a pro- other means of achieving autothermal operation
duct of the reaction (in this case thermal energy) is to use a network consisting of a stirred tank
serves to enhance the reaction rate. reactor followed by a tubular reactor. This case
Since operation in an autothermal mode im- is considered in Illustration 10.6.
plies a feedback of energy to preheat the feed,
ILLUSTRATION 10.6 AUTOTHERMAL
provision must be made for "ignition" of the
OPERATION OF A REACTOR NETWORK
reactor in order to attain steady-state operation.
CONSISTING OF A STIRRED TANK
The ordinary gas burner and many other rapid
REACTOR FOLLOWED BY A PLUG
combustion reactions are examples of autother-
FLOW REACTOR
mal reactions in which the reactants are pre-
heated to the reaction temperature by thermal Consider the reaction studied in Illustration
conduction and radiation. (Back diffusion of 10.1. Autothermal operation is to be achieved
free radicals also plays an important role in using a CSTR with an effective volume of 1000 gal
many combustion processes.) followed by a PFR of undetermined volume.
There are several techniques by which the Pure species A enters at a rate of 40.0 gal/hr and
feedback of energy may be accomplished. at a temperature of 20 °C. The overall fraction
conversion is to be 0.97. This flow rate and con-
1. In a semibatch reactor, a cold feed may be version level will suffice to meet the annual pro-
heated by mixing with the reactor contents. duction requirement of 2 million lb of B. Both
This technique is discussed in Illustration the CSTR and the PFR are to be operated
10.7 later in this section. adiabatically. What PFR volume will be re-
2. In reactors operating under continuous flow quired, and what will be the temperature of the
conditions there may be heat exchange be- effluent stream?
tween the effluent stream or the reactor con-
tents and the feed stream. This exchange may Solution
occur in the reactor proper, in heat ex- The analysis of the performance of the CSTR is
changers between various portions of a re- identical with that of the first CSTR considered
actor network, or by various fluid mixing in Illustration 10.3.
processes. In a stirred tank reactor the fresh The space time for the CSTR is
feed is rapidly mixed with the reactor con- 1000
tents to promote energy transfer. Energy - 25 hr
4O0
feedback in tubular reactors can be accom-
plished by recycling a fraction of the reactor The fraction conversion at the CSTR exit for
effluent for mixing with the fresh feed. this first-order reaction is given by
kr 25k
In Illustration 10.2 we saw that when one uses
25k
a series of stirred tanks for carrying out an exo-
thermic reaction under isothermal conditions The rate constant is given by
there may be occasions when the heat require- k = 2.61 x i o 1 4 e - 1 4 (B)
10.5 Autothermal Operation of Reactors 367
To C p d T
T = 410 + 166(fA - 0.705) (F)
At the exit from the PFR, the temperature will
or, assuming a constant heat capacity, be 454 °K or 181 °C
Equations B, D, and F may now be solved
P(T0Ut - To) 0.5(Tout - 293)
Xn = (Q simultaneously to determine the required space
AHR 83
at To time in the plug flow reactor.
Equations A to C must now be solved simulta- For the present situation r = 3.72 hr so the
neously using a trial and error iterative proce- required PFR volume is 3.72(40.0) = 148.8 gal
dure. One finds that fA1 = 0.705 and T = or 19.9 ft3.
410 °K. By operating in this mode, the necessity for
These properties are those of the stream en- using heat exchangers is eliminated or mini-
tering the plug flow reactor. The design equation mized. If one regards the exit temperature of
for this reactor is 181 C as excessive, a heat exchanger could be
used between the CSTR and the PFR, or ar-
T = °- 97 dfA ^ 00.97 dfA rangements could be made for internal cooling
~~ JO 705 within the PFR.
- fA fe(l - L )
(D)
where k is again given by equation B. In Illustration 10.7 we will consider how to
An energy balance on the PFR operating at meet the vast majority of the heat transfer re-
steady state is given by equation 10.4.6. For quirements for operation of a semibatch reactor
adiabatic operation this equation becomes by a semiautothermal mode of processing. By
^ l e a v i n g PFR ^ e n t e r i n g PF R
0= - AHu at To (E)
Since the heat capacity of the reaction mixture intermittent addition of a cold reactant stream
is given as independent of temperature and com- to the hot contents of a well-stirred semibatch
position, equation E simplifies to reactor, it is possible to maintain the system
temperature within prescribed limits.
0 =.-- ILLUSTRATION 10.7 QUASIAUTOTHERMAL
OPERATION OF A SEMIBATCH REACTOR
' ^p\ l
leaving PFR i
entering PFR/ USING ADDITION OF COLD FEED
The relationship between the temperature and At the time Kladko's article (5) was prepared,
the fraction conversion at a particular point in market conditions dictated that it would not
the plug flow reactor can be obtained by setting be financially remunerative to develop a con-
IAI = /A and TleavingPFR equal to T. Thus, tinuous process for producing B by the reaction
discussed in Illustration 10.1 and other illustra-
AH, tions in this chapter. Consequently they con-
T = T
A
ig PFR "
entenn:
-fA sidered the possibility of using a semibatch
368 Temperature and Energy Effects in Chemical Reactors
reactor with continuous addition of cold feed However, for condensed phases, the difference
to maintain the temperature of the reactor between internal energy and enthalpy is usually
contents within prescribed limits. The basic negligible;
problem was considered earlier in Illustration
Q « Hsys - H sys , 0 (A)
8.11. However, in addition to the material
balance aspects of the design, we now wish to The enthalpy difference is given by
consider the heat transfer requirements for
TT I
operation in accordance with the filling schedule sys
'° J2O°C
rnB ~ mB0)CdT
outlined earlier. For the conditions indicated in m
Illustration 8.11 (isothermal, 163 GC), determine ~ B,o)
the direction and magnitude of the heat transfer where C is the average heat capacity per unit
requirements. mass and where we have assumed that the
initial material sump (mB0) is at 163 °C.
Equations A and B can be combined with
Solution
appropriate numerical values to give
The material balance equations have been solved
Q = (mA + mB - mBO)(0.5)(163 - 20)
earlier with the result that
- 83(/% - mB0)
-k(t-ti) = (mA + mB - mBO)71.5 - S3(mB - mB0)
= rnAie
= 7l.5mA - 11.5(mB - mB0)
where
where
mA is the mass of A remaining in the reactor Q is expressed in calories
at time t m is expressed in grams
mAi is the mass of A in the reactor at time tt If m is expressed in pounds and Q in BTU,
<X>A0 is the mass feed rate of A appropriate to
Q = 12SJmA - 20J(mB - mB0)
the time interval between t and tt
The numerical values characterizing mA and
An energy balance on the reactor can be derived mB at various times are obtained as indicated in
from equation 10.1.1 by omission of appropriate Illustration 8.11. They may be used to determine
terms. the net heat required between time zero and
dE, time t. The pertinent data are presented in
? = Q + hinmin
dt Table 10.1.2. It is evident that heat must be
supplied to the reactor during the first several
Multiplying by dt and integrating between time
hours, but the reactor reaches a point beyond
0 and time t gives
which cooling must be supplied.
Instantaneous heat transfer requirements may
be determined by recognizing that the thermal
If the temperature of the inlet stream is taken energy input must be used either to effect a
as the datum temperature for enthalpy and sensible heat change or the reaction. If q is the
internal energy calculations, heat transfer rate,
Since CA = mA/V'R and the reaction obeys In more conventional engineering units,
first-order kinetics,
q = 128.70)^0 - U9.5mA
q = ®AO [' CdT + kmA AHR where
q is now expressed in BTU's per hour
Substitution of numerical values gives
<$)A0 is expressed in pounds per hour
q = $40(0.5X163 - 20) + 0.8m^(-83) mA is expressed in pounds
Calculated values of this quantity are also
where presented in Table 10.1.2.
If one were to operate this semibatch reactor
q is expressed in calories per hour
under a filling schedule, which for the first 11 hr
Q>AQ is expressed in grams per hour is identical to that considered previously, and
mA is expressed in grams then proceed to feed A at the maximum rate
of 400 lb/hr for an additional 2.875 hr, the same
Table 10.1.2 total amount of A would have been introduced
Material and Energy Balance Analysis for to the reactor. However, in this case, the heat
Illustration 10.7 transfer requirements would change drastically.
There would be a strong exotherm beginning
Total at the moment the cold feed is stopped. The
Time, t mass x 10~ 3
Jtal q x 10~ 3 results for this case are presented in Table
(hr) mA (lb) (lb) (BTU) (BTU/hr)fl 10.1.3.
0 0 1500 0 22.52 Table 10. 1.3
1 120 1675 14.31 8.18 Results of Analysis for Abrupt Termination of Feed
2 175 1850 18.90 1.61 at 13.875 hr
1 QQ 1 8 86 S18
Lyy lo.ou J.1O
4 244 2250 20.93 -0.20 Total
5 265 2475 19.41 -2.71 Time, t mass Qtotal X
10 3 q x 10" 3
6 274 2700 16.10 2.65 (hr) mA (lb) (lb) (BTU) (BTU/hr)a
-7 1 £ C\Q A ^A
/ ly ID 16.Uo 4.J4
3 \2.
8 364 3300 17.12 7.98 11 488 4500 10.81 -6.84
9 439 3700 20.05 -0.98 12 494 4900 3.42 -7.55
10 473 4100 16.85 -5.04 13 498 5300 -4.26 -8.03
11 488 4500 10.81 -16.49 13.875 499 5650 -11.35 -59.63
12 443 4825 -2.64 -17.55 14.00 452 5650 -18.38 -54.01
13 388 5100 -16.55 -17.41 14.50 303 5650 -40.64 -36.21
14 329 5325 -30.02 -16.79 15.00 203 5650 -55.58 -24.26
15 268 5500 -42.76 -19.16 15.50 136 5650 -65.59 -16.25
16 189 5600 -56.63 -16.15 16.00 91 5650 -72.31 -10.87
17 119 5650 -68.13 -14.22 17 41 5650 -79.78 -4.90
18 54 5650 -77.84 -6.45 18 18 5650 -83.22 -2.15
19 24 5650 -82.32 -2.87 19 8 5650 -84.71 -0.96
20 11 5650 84.26 1.31
a
Tnctonto
1 T"\£*/-\iic n 1r»n1 a torJ o ft(±r
3
Instantaneous heat transfer rates are calculated changes are made in the mass flow rates at the times
after changes are made in the mass flow rates at the indicated.
times indicated.
370 Temperature and Energy Effects in Chemical Reactors
Consequently, if one desires to minimize the feed rate and feed composition, there are several
possibility that this reaction will run away as a curves that may result, depending on the nature
result of exceeding the capacity of the cooling of the reaction or reactions involved. Figure 10.1
network, it would be advisable to operate in a shows the nature of the curve for the case of a
mode such that the feed rate is gradually simple first-order irreversible reaction. At low
diminished instead of abruptly terminated at a temperatures the reaction rate is negligible, so
high feed flow rate level. very little energy is released. At high tempera-
tures the reaction rate constant is so large that
10.6 STABLE OPERATING CONDITIONS
very little unreacted material remains in the
IN STIRRED TANK REACTORS
effluent stream. Consequently, still higher values
of T cannot increase Qg significantly, and the
In Illustrations 10.2 and 10.3 the temperatures curve must approach an asymptote. At inter-
and degrees of conversion leaving stirred tank mediate temperatures one has a competition
reactors were determined by simultaneous solu- between the increase in reaction rate arising
tion of rate, material balance, and energy from the effect of increased temperature and
balance equations. In this section we will see the decrease in the rate arising from depletion
that there may be circumstances when there of the reactants. The equation of the curve is
will be more than one set of operating condi- easily obtained from first principles.
tions that will satisfy these equations. In such
cases questions arise as to the stability of the Qg = rVR AH (10.6.1)
operating points corresponding to these solu-
tions. For the elementary cases that are treated where |A//| is the magnitude of the heat of
in this text, the stability can be readily deter- reaction for the exothermic reaction.
mined on the basis of simple physical arguments.
There are more complex cases in the general
area of reactor stability, however, that require
sophisticated mathematical analysis. Many of
these cases have been treated in the extensive
body of literature that exists on this subject.
Perlmutter (7) has summarized much of this
material in textbook form.
When a reactor is operating at steady state,
the rate of energy release by chemical reaction
must be equal to the sum of the rates of energy
loss by convective flow and heat transfer to the
surroundings. This statement was expressed in
algebraic form in equations 10.3.4 and 10.4.6
for the CSTR and PFR, respectively. It will
serve as the physical basis that we will use to
examine the stability of various operating
Temperature, T
points.
Let Qg represent the rate at which thermal Figure 10.1
energy is released by an exothermic chemical Rate of energy release by reaction versus tem-
reaction in a CSTR. If Qg is plotted versus the perature for an irreversible exothermic reaction
temperature of the reactor contents for a fixed carried out in a CSTR.
10.6 Stable Operating Conditions in Stirred Tank Reactors 371
region where thermal energy would be released operating conditions, as do points a and c.
faster than it could be dissipated. This would Point b, however, is unstable. Also note that a
cause the system to increase further in tempera- relatively small slope for the heat loss curve will
ture. The temperature deviation would continue give rise to a high reaction temperature and low
to be amplified until the reactor arrived at yield. If the cooling capacity is increased one
point c, the upper stable operating point. A moves from line 4' to line 3' and achieves a
negative temperature fluctuation would also be higher degree of conversion for this equilibrium
amplified, the system losing energy faster than reaction. The highest yield evidently corres-
it is generated by the reaction. In this case the ponds to using an energy loss curve that inter-
temperature drops until the lower stable oper- sects the energy release curve near or at its
ating point is attained. This situation corres- peak. However, because of the shape of the
ponds to extinguishing the reaction. Because of curve it may be very difficult from a process
the nature of its response to deviations in tem- control viewpoint to operate at this condition.
perature, point b is referred to as an unstable Increases in the cooling rate can increase the
operating point. slope of the energy loss curve, possibly leading
To operate at a point like c on curve 2 in to extinction of the reaction. This situation can
Figure 10.3, one must provide a means of by- be visualized by rotating line 2' to the left about
passing or circumventing the lower stable state a its intersection on the T axis.
and the unstable state b. This may be accom- The problem of ignition and extinction of
plished by temporarily preheating the feed, op- reactions is basic to that of controlling the
erating at a reduced flow rate, or reducing the process. It is interesting to consider this problem
capacity of the cooling system. These procedures in terms of the variables used in the earlier
will permit one to move to the right of point b discussion of stability. When multiple steady-
and "ignite" the reaction. state solutions exist, the transitions between the
These stability considerations are not limited various stable operating points are essentially
to first-order irreversible reactions. Figure 10.4 discontinuous, and hysteresis effects can be ob-
depicts the Qg and Qr curves for a reversible served in these situations.
exothermic reaction. The intersections of the Consider the energy release and energy loss
Qg curve and lines 3' and 4' represent stable curves shown in Figure 10.5. These curves cor-
respond to three different operating conditions
where one would be varying either the tempera-
ture of the feed stream, To, or the temperature
of the medium to which heat is being transferred,
Tm. The intercepts on the temperature axis cor-
responding to the three cases are Tr, T" and
T". We will assume that all other operating
variables are held constant. When the T inter-
cept is V, the reactor may operate in a stable
mode at point A. However, if the cooling rate
s is increased (by decreasing Tm) or if the tem-
cc
perature of the feed stream drops, curve Q'r will
Temperature, T
move to the left and the reaction will be extin-
Figure 10.4 guished. Consequently, point A represents a
Energy release and energy loss curves for revers- critical extinction point. For combustion re-
ible exothermic reaction in a CSTR. actions the temperature at this point may be
374 Temperature and Energy Effects in Chemical Reactors
reaction rate has the highest possible value. For temperature profile is one where the temperature
endothermic reactions this implies the maximum falls with increasing conversion. Severe heat
possible feed temperature. For exothermic re- transfer requirements may be needed to make
actions this means that the exit temperature the actual temperature profile approach the
should lie below that corresponding to equili- desired ideal. Two ways in which this may be
brium at the desired fraction conversion and achieved ar£ indicated in Figure 10.7. The case
above that correspondiilg to the maximum rate of internal heat exchange between the reaction
at this conversion. A trial and error procedure mixture and the feed stream has been discussed
can locate an exit temperature within this range by Van Heerden (8) and Kramers and Wester-
that corresponds to the minimum reactor vol- terp (9). The other alternative is to use staged
lume. For a CSTR the reactor effluent tempera- operations with interstage cooling between
ture should be that which gives a maximum adiabatic sections. This case has also been
rate at the desired conversion level. The inlet treated by Kramers and Westerterp (9). For
temperature may be adjusted to give this endothermic reactions, multiple stages with re-
effluent condition when the reactor is operated heating between stages are commonly used to
adiabatically. keep the temperature from dropping too far.
For an exothermic reaction, adiabatic opera- Illustration 10.8 indicates how one determines
tion gives an increase in temperature with the optimum temperature at which a single
increasing conversion. However, the optimum CSTR should be operated.
Heat out Heat out Heat out
i
Very 1 A A A
Hot Warm
hot <yJlMJ
-Cold feed
Heating of feed
Distance Distance
(a) (b)
Figure 10.7
Techniques for approaching optimum temperature profiles for exothermic reaction, (a) Adiabatic operation
of reactors with interstage cooling, {b) Countercurrent heat exchange. (Adapted from Chemical Reaction
Engineering, Second Edition, by O. Levenspiel. Copyright ® 1972. Reprinted by permission of John Wiley
and Sons, Inc.)
10.7 Selection of Optimum Reactor Temperature Profiles . . . Thermodynamic and Selectivity Considerations 377
k2 = A2e~E2/RT
RT2 Jc
RT2
Determine the minimum reactor volume that or
will be required to obtain a fraction conversion <! E 2 ( fc
fc if the feed is pure C and if the input volumetric k2 E , \\ - fc
flow rate is %,. What will be the temperature
Substituting the Arrhenius form of the reac-
of the effluent stream?
tion rate constants,
Solution E2( fc
E , 11 - fe
For this CSTR the appropriate design equation
is One can then solve for the optimum tempera-
ture:
v Crci dfc c f
T =
VR coj o Jc _CCOJC ( A ) Ji M OJE2A2( fc
-rt CF — r.
CF T\ R R
with
or
— rCF — k2CBF (B)
T=
F r o m stoichiometric considerations Rln E2A2 fc RlnP
l - fc
CCF = C c o ( l - fc) (Q
At this temperature
CBF = CCOJC (D)
__ ^ O-Ei/(£2-£i)
Combining equations A to D gives
and
fc fc k2 = AJ
r0 M I - /,) - hf *i - fc{kt + k2) Thus the minimum reactor volume is given by
In order to minimize the required reactor
volume one may set the temperature derivative VR =
of VR equal to zero. 'Xi - fc) -
378 Temperature and Energy Effects in Chemical Reactors
10.7.2 The Influence of Selectivity High temperatures favor the reaction with the
Considerations on the Choice of higher activation energy and lower temperatures
Reactor Operating Temperatures favor that with the lower activation energy.
When multiple reactions arfc possible, certain of For the parallel reactions in equation 10.7.1
the products have greater economic value than one may use this general rule to select the follpw-
others, and one must select the type of reactor ing operating conditions as optimum from a
and the operating conditions so as to optimize selectivity viewpoint when the reactor operates
the product distribution and yield. In this sub- isothermally.
section we examine how the temperature can
If E1 > E2, use a high temperature.
be manipulated with these ends in mind. In our
If E2 ^> Eu use a low temperature.
treatment we will ignore the effect of concentra-
tion levels on the product distribution by as- On the other hand, if it is possible to use a
suming that the concentration dependence of temperature progression scheme and if one de-
the rate expressions for the competing reactions sires to obtain the maximum amount of the
is the same in all cases. The concentration effects desired product per unit time per unit reactor
were treated in detail in Chapter 9. volume, somewhat different considerations are
Consider the following simple parallel applicable. If Ex > E2, one should use a high
reactions. temperature throughout, but if E2 > Eu the
temperature should increase with time in a batch
D (desired product)
reactor or with distance from the reactor inlet in
a plug flow reactor. It is best to use a low tem-
A + B (10.7.1) perature initially in order to favor conversion
to the desired product. In the final stages of the
(undesired product) reaction a higher temperature is more desirable
in order to raise the reaction rate, which has
It is evident that fallen off because of depletion of reactants. Even
dD though this temperature increases the produc-
(10.7.2) tion of the undesirable product, more of the
dU
desired product is formed than would otherwise
If reaction 1 is to be enhanced and reaction 2 be the case. Thus one obtains a maximum
depressed, the ratio of the rate constants must production capacity for the desired product.
be made as large as possible. This ratio may be Now consider the basic series reaction scheme
written in the Arrhenius form as k2
D -4U (10.7.4)
-Ei/RT
where species D is the desired product. In this
case the desired product distribution is favored
(10.7.3)
when the ratio of the rate constants (k1/k2) is
It changes with temperature in a manner that mad£ very large Therefore?, for operation at a
depends on whether E1 is greater than or less single temperature, one should use a high tem-
than E2. When the temperature increases, the perature if El > E2 and a low temperature if
ratio increases if Ex > E2 and decreases if E2 > E2 > Ex. On the other hand, if a temperature
Ex. Consequently, the reaction having the larger progression is to be employed, the relative mag-
activation energy is the one that is most sensitive nitudes of the activation energies determine
to variations in temperature. The following gen- whether the initial temperature should be greater
eral rule is appropriate to competing reactions. than or less than the final temperature. For
10.7 Selection of Optimum Reactor Temperature Profiles .. . Thermodynamic and Selectivity Considerations 379
example, if E2 > El9 the temperature should space time or reactor holding time should be
start out high to accelerate the first reaction and used. Levenspiel (12) gives the following example
thus obtain a large output from a given reactor of a situation of this type:
volume. However, the temperature should be
progressively reduced as species D accumulates R T (desired)
in order to take advantage of the fact that the
undesirable side reaction D -+ U slows down U W
faster with decreasing temperature than does the
(10.7.7)
useful reaction. The problem of selecting an where
optimum temperature schedule has been treated Ei >E2
by Bilous and Amundson (10).
For series-parallel combinations of reactions £4 > £3
a number of possible situations can arise. Two E5 > E6
general guidelines are useful to keep in mind (11). The optimum temperature progression to use
First, if one has two undesirable products being in this sequence is a high temperature to start
formed in parallel with a desired species or with followed by lower temperatures where the con-
an intermediate product that can subsequently centration of species R is high and then in-
react to form the desired species creasing temperatures when the concentration
of species S becomes appreciable. Levenspiel (13)
has summarized the results of several analyses
(10.7.5) of the optimum temperature level and progres-
sion to be used for several general reaction
schemes.
Although a determination of the operating
there are three significant orderings of the activa- temperature that produces the most favorable
tion energies. The type of temperature to use for product distribution is important in working up
each of these cases is given below: a reactor design for a reaction scheme involving
1. Ei > E2 and E± > £ 3 ; use high temperature. multiple reactions, other considerations may be
2. Ei < E2 and £ x < E3; use low temperature. equally important from a design viewpoint.
3. Ei > E2 and Ex < E3 or £ x < E2 and Et > Optimization of the reactor economics requires
E3; use intermediate temperature. a favorable product distribution, high conver-
sions, and low capital costs. If the most favorable
In case 3 an intermediate temperature will give product distribution is obtained at low tempera-
the most favorable product distribution. If Ex > tures, very little of any product will be formed.
E2andEi < E3, one can show that the optimum In this case it will be more economical to use an
temperature is given by intermediate temperature, because the adverse
effects of a change in the product distribution
E3 -E2 will be more than offset by the increase in re-
T = (10.7.6)
action rate and conversion. This will have the
Rln
!i - E2J A2 effect of increasing the production capacity of a
given reactor. In the case where the most favor-
A second generalization is applicable to steps able product distribution is favored by operation
occurring in series. If an early step requires a at high temperatures, the highest permissible
high temperature and a subsequent step is temperature should be used, because the reac-
favored by a low temperature, then a tempera- tion rates will be high and the requisite reactor
ture sequence that decreases with increasing size small.
380 Temperature and Energy Effects in Chemical Reactors
For the proposed design the reactor volumes 7. The hydrogenation of cottonseed oil pro-
are both 0.8 m 3 , the input volumetric flow rate ceeds by the following irreversible steps.
is 1.10 m 3 /ksec, and the overall fraction con-
version of the initial B is to be 0.80. Since the
reaction is carried out in dilute liquid solution,
the effective heat capacity of the liquid mixture
is substantially unaffected by the reaction and
may be regarded as a constant that is equal to
3.47 J/cm3-°K. If the initial concentration of
species B is 5.6 kmole/m 3 and the feed stream
enters at 70 °C, determine the required heat A: Linoleic acid C: Iso-oleic acid
transfer area for each reactor. B: ds-oleic acid D: Stearic acid
Problems 383
Relative activation energies have been estimated will operate isothermally, but each will operate
by comparison with similar reactions: at a different temperature. You have been asked
to design a reactor capable of processing feed at
Reaction EA 1.7 m3/sec.
The following additional data are available.
1 Low
2 High Specific heats Inerts
A B C
3 Low (J/mole -°K) 62.8 75.4 125.6 75.4
4 High Species concentrations A B C Inerts
5 High (kmole/m3)
Feed 3.0 3.0 0.0 32.0
In a batch reactor at 100 °C operated for a short Effluent from 0.3 5.7 2.7 32.0
time, 90 mole percent A remains unreacted, 6.2% second reactor
forms B, and 3.8% forms C.
(a) It is desired to produce compound C. You Feed temperature = 330 °K
are determining the optimum temperature
— - 70,000 J/mole for the reaction as
for operating a CSTR. Assuming all reac-
written (at 330 °K)
tions are first-order, explain the fact that two
optimum temperatures are found. Reactor operating temperatures: First — 330 °K,
Second - 3 5 8 °K
Activation energy = 108.4 kJ/mole
At33O°K,/c - 33Oksec- 1
(a) What is the volume of each CSTR?
(b) How much thermal energy must be removed
in each reactor if the system is to operate
in the fashion described above? Be very care-
ful to write the energy balance for the second
reactor in the appropriate form.
Temperature, T
9. The liquid phase reaction A -• B is to be
carried out in a plug flow tubular reactor at a
(b) If B is desired, is it better to operate at low constant pressure of 202.6 kPa. The feed is
or high temperature? Why? 600 kmoles/ksec of pure A with an inlet tem-
(c) Assume that reactions 4 and 5 have the same perature of 200 °C. Pure A has a specific volume
value of EA. Which is larger, fc4 or k5 ? Why? of 0.056 m 3 /kmole. The heat of reaction at
200 °C is —15 kJ/mole. The molar specific heats
8. The decomposition reaction A —• B + C oc- of A and B are both 42 J/mole-°K. The reaction
curs in the liquid phase. It has been suggested rate constant in this range is given by the
that your company produce C from a stream expression
containing equimolar concentrations of A and fc = 110 + 0.8(7 - 200)
B by using two continuous stirred tank reactors
where
in series. Both reactors have the same volume.
The reaction is first-order with respect to A and T is measured in degrees Celsius
zero-order with respect to B and C. Each reactor k has the units ksec~*
384 Temperature and Energy Effects in Chemical Reactors
The reactor would be run adiabatically, but reactor each minute. The temperature of the
the maximum reaction temperature allowable is pure A stream is 70 °F. The fraction of the A
400 °C, since above this temperature undesirable converted to R in the CSTR is 0.4, and the
by-products are formed. Calculate the minimum temperature of the effluent from the CSTR is
reactor volume required to obtain 80% con- 100 °F. The mole density of pure liquid A is
version of A. What must the heat transfer rate 1.5 lb moles/ft3. The following table gives the
be in the cooling section of the reactor? temperature dependence of the rate constants
kx and k-v
- Adiabatic - -Cooling-
i 400°C maximum
/ r — ^ 200° C 400° C D = 5 cm
Q Pure A I I (80% conversion)
70 0.55 0.010
75 0.93 0.020
80 1.67 0.040
85 2.90 0.078
90 5.0 0.156
95 8.65 0.312
100 15.0 0.625
where E1 > E2 and E1 < E3. What tempera-
ture will give the maximum yield of species D?
Heat capacity of pure A = 25 Btu/lb mole-°R
11. An autocatalytic reaction of the type Heat capacity of pure R = 30 Btu/lb mole-°R
A + R ^± R + R AHR = -10,000 Btu/lb mole R formed for the
fc-i
isothermal reaction at 70 °F
is being carried out in the combination of re-
actors shown below. (a) Determine the CSTR volume.
(b) Determine the heat transfer rate (and direc-
tion) in the first reactor.
Pure A
(c) The plug flow reactor is to be operated iso-
thermally at 100 °F. The PFR effluent con-
version of the A fed to the CSTR is to be
90%. Determine the heat transfer require-
Ideal ments for the PFR and the necessary reactor
plug flow reactor volume.
The reaction is reversible and exothermic. The 12. A tubular reactor is being used to investigate
reaction stoichiometry is A -> R. One hundred an endothermic isomerization reaction that may
pound moles of pure liquid A are fed to the be represented schematically as A -> B.
Problems 385
The following data are applicable. The standard enthalpy change for this reaction
AHR = 46,500 J/mole A reacted at 30 °C at 200 °C is —126 kJ/mole for the reaction as
written. Each CSTR has a volume of 0.2 m 3 . The
CP,A = Cp,B = 100 J/mole-°K
feed rate is 66 cm3/sec. The following data on
Reactor diameter = 50 cm average heat capacities are available.
Pressure = 203 kPa (may be considered con-
stant over the length of the reactor)
Species Heat capacity (J/mole -CK)
Feed: Pure A at a rate of 12.6 moles/sec
A 54.4
Initial temperature = 320 °C B 71.1
If electrical heating is applied along the length C 96.3
of the reactor to provide energy at a constant D 29.2
rate of 575 J/sec-cm of reactor length and this S 41.9
input is evenly distributed along the length of
reactor, determine the temperature profile in the
reactor. Heat losses from the reactor are negli- At 500 °C the second-order rate constant is
gible. The experimental data on the concentra- 1.7 m3/kmole-ksec. Determine:
tion profile are as follows. (a) The composition of the effluent from the
first reactor.
Distance from inlet (m) Fraction A converted (b) If the heat transfer rate in the first reactor
is —1.05 kJ/sec, what is the temperature of
0 0 the effluent from the first reactor?
0.85 0.1 (c) What is the reaction rate constant at this
1.74 0.2 temperature?
3.05 0.3
4.27 0.4
6.10 0.5 14. An irreversible liquid phase dimerization
reaction of the type
13. The reaction A + B -> C + D takes place
2M -* D
in the two ideal CSTR's operating in series
shown below. The reaction is first-order with follows second-order kinetics. You have been
respect to both A and B. Temperatures and asked to estimate the fraction of the monomer
concentrations of the species of interest are as that can be converted to dimer in an existing
indicated. S denotes the solvent. continuous stirred tank reactor facility designed
Feed 1 1 /
CA0 = 0.6 kmole/m3 | /
7~
1
CBO = 1-1 _/
CC0 = 0.0
Cm = 0.0 ^ Tfinal = 500° C
x — ?
Cso = 20 c= ? CAF = 0.1 kmole/m3
7in = 200° C CBF = 0.6
C CF = 0.5
CDF = 0.5
CSF = 20
386 Temperature and Energy Effects in Chemical Reactors
for adiabatic operation. The reactor volume is (f) In order to maximize production of B when
0.4 m 3 and the available input volumetric flow x = 480 sec, what inlet temperature should
rate is 1.3 m 3 /ksec of pure monomer. Additional be specified?
data are given below.
Inlet temperature = 312 °K 16. A dissociation reaction of the type A -•
Standard enthalpy change on reaction at B + C is being studied in a pilot plant reactor
300 °K - - 4 2 kJ/mole D formed having a volume of 0.5 m 3 . The reaction involves
Liquid heat capacity = 2.0 J/cm 3 o K ideal gases with the following heat capacities.
Monomer feed concentration = 16 kmoles/m 3 CpA= 160J/mole°K
Thermal expansion effects may be neglected. CpJS= 120J/mole-°K
k = 2.70 x 10 9 ^- 1 2 ' 1 8 5 / T rn 3 /mole-ksecforTin CpC= 120J/mole-°K
degrees Kelvin
Pure A is charged to the reactor at 400 kPa and
330 °K. The reaction is first-order in species A.
15. Consider a CSTR that is used to carry out a
The variation of the reaction rate constant withv
reversible isomerization reaction of the type
temperature is given below.
CH CH CH
I C H 2 ||
CH CH CH
Two feed streams each containing one of the benzene = 1-75 J/g-°K
reactants at a concentration of 0.2 kmoles/m 3 of
PBenzene = 880 kg/m 3
benzene and each at 25 °C are to be rapidly
mixed and fed to a tubular reactor. The reverse reaction and side reactions may
If the flow rate of each feed stream is 0.139 be neglected.
m 3 /ksec and if 50% conversion is to be achieved, Note that on mixing the reactant concentra-
determine the reactor volumes required for iso- tions will drop to half those prevailing before
thermal operation and for adiabatic operation. mixing.
11 Deviations from Ideal Flow Conditions
11.0 INTRODUCTION nonideal flow systems in terms of data that are
easily obtained experimentally. From measure-
The flow patterns in real reactors do not con-
ments on the feed and effluent streams, one may
form exactly to those postulated for ideal plug
develop parameters that can be used to charac-
flow and continuous stirred tank reactors. None-
terize the magnitude of system nonidealities and
theless, the conversions achieved in many real
to serve as input to more complex models of
reactors so closely approximate those predicted
reactor behavior than ideal CSTR's and PFR's.
on the basis of the idealized models that the
We will begin by indicating how such measure-
design equations for these reactors can be used
ments can be used to determine the residence
with negligible error. In other cases significant
time distribution function for a reactor, and then
differences are noted between observed and
treat three different mathematical models that
predicted results. These differences may arise
permit estimation of the conversion that will be
from a number of sources—from channeling of
attained. Throughout this chapter we will re-
fluid as it moves through the reaction vessel,
strict our discussion to systems in which a single
from longitudinal mixing caused by vortices and
reaction takes place in a homogeneous isother-
turbulence, from the presence of stagnant re-
mal reactor. Volume changes on reaction are
gions within the reactor, from bypassing or
also assumed to be negligible (3 ~ 0). These
short-circuiting of portions of a packed reactor
restrictions will permit us to focus our attention
bed, from the failure of impellers or other mix-
on the nonideal flow conditions.
ing devices to provide perfect mixing, etc. In
this chapter we hope to establish a rational basis
for examining quantitatively and qualitatively 11.1 THE RESIDENCE TIME
the effect of departures from idealized flow DISTRIBUTION FUNCTION dF(t)
behavior on the performance of a reactor. Read
this chapter with a view toward developing a Except for the case of an ideal plug flow reactor,
"seat of the pants" feeling for the magnitude of different fluid elements will take different lengths
deviations from ideal flow conditions in various of time to flow through a chemical reactor. In
types of reactors so that you will know when order to be able to predict the behavior of a
these effects can be neglected and when they given piece of equipment as a chemical reactor,
must be treated by the techniques developed in one must be able to determine how long differ-
this chapter. ent fluid elements remain in the reactor. One
In principle, if the temperatures, velocities, does this by measuring the response of the
flow patterns, and local rates of mixing of every effluent stream to changes in the concentration
element of fluid in a reactor were known, and of inert species in the feed stream—the so-called
if the differential material and energy balances stimulus-response technique. In this section we
could be integrated over the reactor volume, one will discuss the analytical form in which the
could obtain an exact solution for the composi- distribution of residence times is cast, derive
tion of the effluent stream and thus the degree relationships of this type for various reactor
of conversion that takes place in the reactor. models, and illustrate how experimental data
However, most of this information is lacking are treated in order to determine the distribution
for the reactors used in laboratory or commer- function.
cial practice. Consequently, it has been necessary The mathematical relations expressing the
to develop approximate methods for treating different amounts of time that fluid elements
388
11.1 The Residence Time Distribution Function dF{i) 389
catalysts present in the reactor, by settling out The interpretation of the F(t) curve as the
or being filtered out as it moves through the probability that a fluid element entering the
reactor, or by chemical reaction in the case of reactor at time zero has left by time t may be
nonradioactive tracers). used to indicate how the curve may be generated
There are three general stimulus techniques from experimental data. Let us consider the case
commonly used in theoretical and experimental where at time zero one makes a step change in
analyses of reactor networks in order to charac- the weight fraction tracer in the feed stream
terize their dynamic behavior. from WQ to w'o . A generalized stimulus-response
curve for this system is shown in Figure 11.2.
1. A step function change in which the input
concentration is changed from one steady-
Stimulus
state level to another.
2. A pulse input in which a relatively small
amount of tracer is injected into the feed
stream in the shortest possible time. w0 fw
t E
3. A sinusoidal input. The frequency of the 1
sinusoidal variation is changed and the /Response
steady-state response of the effluent at differ-
ent input frequencies is determined, thus w0
1
generating a frequency-response diagram for
the system.
Time, t
The time variations of the effluent tracer con-
Figure 11.2
centration in response to step and pulse inputs Generalized response of an arbitrary reactor to a
and the frequency-response diagram all contain step change in input tracer concentration.
essentially the same information. In principle,
any one can be mathematically transformed into At time t the fraction of the effluent fluid
the other two. However, since it is easier experi- characterized by an age less than t (and thus
mentally to effect a change in input tracer con- with a composition WQ ) is just equal to F(t).
centration that approximates a step change or At the same time, the fraction of the effluent
an impulse function, and since the measure- characterized by an age greater than t (with the
ments associated with sinusoidal variations are original inlet composition WQ) is equal to
much more time consuming and require special 1 — F(t). The time dependence of the weight
equipment, the latter are used much less often fraction tracer in the effluent (wE) is then given
in simple reactor studies. Even in the first two by
cases, one can obtain good experimental results
only if the average residence time in the system w£(£) = wZF(t) + wo [1 " F{t)] (11.1-3)
is relatively long. Thus
Kramers and Alberda (6) have discussed the
manner in which sinusoidal variations are F(f) = (11.1.4)
analyzed, but we will discuss only the first two
types of stimuli. They are sufficient for the and we see that in general F(t) would give the
analysis of the majority of situations that will relative response of the system to a step function
be encountered by the chemical engineer en- input.
gaged in the practice of designing chemical Now consider the case where the input to the
reactors. system is a pulse of tracer. The amount of tracer
11.1 The Residence Time Distribution Function dF(t) 391
leaving in a time increment dt is wE<bm dt where Determine the average residence time of the
<J>m is the mass flow rate. For constant density fluid and the F(t) curve for this system.
systems mass and volume fractions are identical
and the basic definition of F(i) indicates that Solution
I C,
392 Deviations from Ideal Flow Conditions
Table 11.1.1
Data Workup for Illustration 11.1
ZQ(g/m3)
Time, t (sec) Q ( g ;/m )
3
0 F(t) tCE (g • sec/m3) t2CE (g • sec2/m3)
0 0.0 0.0 0 0 0
120 6.5 6.5 0.13 780 93,600
240 12.5 19.0 0.38 3000 720,000
360 12.5 31.5 0.63 4500 1,620,000
480 10.0 41.5 0.83 4800 2,304,000
600 5.0 46.5 0.93 3000 1,800,000
720 2.5 49.0 0.98 1800 1,296,000
840 1.0 50.0 1.00 840 705,600
960 0.0 50.0 1.00 0 0
1080 0.0 50.0 1.00 0 0
icE = 50.0
0
YtCE = 18,720
n
Table 11.I.I contains a workup of the data The plug flow reactor has a flat velocity profile
in terms of the above analysis. In the more and no longitudinal mixing. These idealizations
general case one should be sure to use appro- imply that all fluid elements leaving the reactor
priate averaging techniques or graphical inte- have the same age (I). The F(t) function for this
gration to determine both F(t) and 7. When system must then be
there is an abundance of data, plot it, draw a F(t) = 0 for 0 < t <1
smooth curve, and integrate graphically instead (1LL7)
of using the strictly numerical procedure em- F(t) = 1.0 for t > l
ployed above. The responses of this system to ideal step and
On the basis of the tabular entries pulse inputs are shown in Figure 11.3. Because
the flow patterns in real tubular reactors will
18
- > 720 always involve some axial mixing and boundary
layer flow near the walls of the vessels, they will
distort the response curves for the ideal plug
11.1.2 F(t) Curves for Ideal Flow Patterns flow reactor. Consequently, the responses of a
real tubular reactor to these inputs may look
For a few highly idealized systems, the residence like those shown in Figure 11.3.
time distribution function can be determined a The next case to be considered is the ideal
priori without the need for experimental work. continuous stirred tank reactor. The key to the
These systems include our two idealized flow derivation of the F(i) curve for this type of re-
reactors—the plug flow reactor and the con- actor is the realization that the assumption of
tinuous stirred tank reactor—and the tubular perfect mixing implies that upon entry in the
laminar flow reactor. The F(t) and response reactor an element of volume can instanta-
curves for each of these three types of well- neously appear in any portion of the reactor.
characterized flow patterns will be developed Therefore its past or its future history cannot be
in turn. derived from its position. Furthermore, the prob-
11.1 The Residence Time Distribution Function dF{t) 393
0 0 t 0 T
Time, t Time, t Time, t
Input signal Response of plug flow reactor Response of real tubular reactor
W
E
0
Time, t Time, t Time, t
Figure 11.3
Response of ideal plug flow reactor and real tubular reactor to step and impulse inputs.
The relative response of a single CSTR to an These fractions may lead to an irregular re-
ideal pulse input may be obtained by taking the sponse curve. This effect cannot be described
time derivative of equation 11.1.13. quantitatively for a general system, since it will
dF{t) e~t/T depend on the relative orientations of the inlet
(11.1.14) and outlet with respect to the impellers and any
dt
baffles present in the system. Examples of
From this equation it is evident that there is possible response curves are shown in Figure
a wide distribution of residence times in a 11.5. The effluent response will also be strongly
stirred tank reactor. dependent on the mixing time. The approx-
The responses of a single ideal stirred tank imation of an actual stirred tank reactor by an
reactor to ideal step and pulse inputs are shown ideal CSTR improves as the ratio of the mean
in Figure 11.4. Note that any change in the residence time to the mixing time increases.
reactor inlet stream shows up immediately at For most design purposes a ratio greater than
the reactor outlet in these systems. This fact is 10 gives a very good approximation [7].
used to advantage in the design of automatic The final idealized flow situation that we will
control systems for stirred tank reactors. consider is laminar flow in a tubular reactor in
The performance of the stirred tank flow the absence of either radial or longitudinal
reactors that are encountered in industrial diffusion. The velocity profile in such a reactor
practice may differ significantly from the ideal is given by
case discussed above. The feed stream entering
the stirred region will not be instantaneously u =u (11.1.15)
dispersed and mixed with the entire contents of °
the vessel. The mixing process will require a for 0 < r < R
finite amount of time in order to produce a
microscopically homogeneous solution, so there where
will be a time lag between the change in the u0 is the centerline velocity
input stream and the change in the character- r is the distance from the centerline
istics of the effluent stream. Moreover, a fraction of the pipe
of the feed stream may pass through the reactor
outlet without undergoing complete mixing. R is the inside radius of the pipe
wE
> 0.632 (W+O -
J I
0 T 0 T
Time, t Time, t
Response to step change Response to pulse input
Figure 11.4
Response of ideal continuous stirred tank reactor to step and pulse
inputs.
11.1 The Residence Time Distribution Function dF(t) 395
w0
0 I 0 t
Time, t Time, t
Response to pulse input
Response to step change
Figure 11.5
Response of real stirred tank reactor to step and pulse inputs.
The average velocity with which a' fluid It leaves at a time tmin given by
element moves is given by
_ L _T
(11.1.20)
" = y (11.1.16) u0 2
Thus
Since u varies with r, the residence times of the
various fluid elements will also vary with r. The F(t) = 0 for t<- (11.1.21)
time that it will take a fluid element to traverse
the reactor is given by The fraction of the volumetric flow rate that
takes place in the region bounded by r = 0 and
r = r will be equal to F(t) where t is equal to
"»!'-(« the time necessary for a fluid element to traverse
the reactor length at a given r (equation 11.1.17).
where L is the length of the reactor. Thus
Combining equations 11.1.17 and 11.1.18 gives Combining equations 11.1.15 and 11.1.23 and
simplifying gives
r dr
F(t) = (11.1.24)
The fluid at the centerline is moving the
Jo r dr
fastest so this material will be the first to leave.
396 Deviations from Ideal Flow Conditions
1.0
0.8
0.6
Fit)
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Dimensionless time, t/t
Figure 11.6
F(t) curves for reactors with idealized flow patterns.
11.1 The Residence Time Distribution Function dF(t) 397
It is convenient to classify deviations from that the fluid velocity and reactant concentra-
ideal flow conditions into two categories. tion are constant across the tube diameter. The
magnitude of the dispersion is assumed to be
1. Different fluid elements may move through
independent of position within the vessel, so
the reactor at different velocities. The ele-
there will be no stagnant regions and no by-
ments remain segregated1 as they move
passing or short-circuiting of fluid in the model
through the reactor.
reactor. By changing the magnitude of the
2. Fluid elements with different ages may mix
dispersion parameter, one may vary the per-
on a microscopic scale. However, the mixing
formance of the reactor from that of plug flow
does not occur to the extent that it does in
{Q)L = 0) to that of a single continuous stirred
an ideal CSTR.
tank reactor {Q)L = oo).
These two types of deviations occur simul- The axial dispersion parameter &L is a term
taneously in actual reactors, but the mathe- that accounts for mixing both by molecular
matical models we will develop assume that the diffusion processes and by turbulent eddies and
residence time distribution function may be vortices. Since these two types of phenomena
attributed to one or the other of these flow are to be characterized by a single parameter,
situations. The first class of nonideal flow con- and since we force the model to fit the form of
ditions leads to the segregated flow model of Fick's law of diffusion, Q)L should be regarded
reactor performance. This model may be used as an effective dispersion coefficient having the
with the residence time distribution function to units of an ordinary molecular diffusivity
predict accurately conversion levels for first- (Iength2/time). However, it is significantly
order reactions that occur isothermally. (See greater in magnitude because of turbulence
Section 11.2.1.) The second class may be modeled effects.
in several ways, depending on the additional The response of the axial dispersion model
assumptions one is willing to make concerning to step or pulse tracer inputs can be determined
the nature of the mixing processes. Once these by writing a material balance over a short
assumptions are made the parameters of the tubular segment and then solving the resultant
mathematical model can be determined from differential equations. A transient material bal-
the F(i) curve. The remainder of this section is ance on a cylindrical element of length AZ gives
devoted to a discussion of the interpretation of where the input and output terms allow for
response data in terms of two of these mixing both dispersive and convective transport.
models. Dividing by the cross-sectional area, At, and
AZ, and taking the limit as the last two param-
11.1.3.1 The Axial Dispersion Model. The
eters approach zero gives
axial dispersion model is often used to describe
the behavior of tubular reactors. This model 8C
characterizes mass transport in the axial direc- (11.1.29)
tion in terms of an effective or apparent longi-
tudinal diffusivity Q)h that is superimposed on The initial and boundary conditions that
the plug flow velocity. The model also assumes apply to this equation depend on whether one
398 Deviations from Ideal Flow Conditions
1.0 c
I / j T ^ ^ ^ ~^t = °' 0 2 5
0.8 — 0.001—-
i
-A
0.4 —
0.2 —
0.5
n 1.0
Dimensionless time, t/T*
1
1.5 2.0
Figure 11.7
Residence time distribution curves for dispersion model.
is dealing with a pulse or step input and the equation 11.1.29 are:
characteristics of the system at the tracer in-
jection and monitoring stations. At each of f Co for Z > 0 at t = 0
C = (11.1.30)
these points the tubular reactor is characterized $ for Z < 0 at t = 0
as "closed" or "open," depending on whether
or not plug flow into or out of the test section Co at Z = - oo for t > 0
C = (11.1.31)
is assumed. A closed boundary is one at which Co at Z = + oo for t > 0
there is plug flow outside of the test section; an
open boundary is one at which the same dis- where C o and CQ are the tracer concentrations
persion parameter characterizes the flow condi- before and after the step change, respectively.
tions within and adjacent to the test section. In this case a closed form solution is possible
There are then four different possible sets of with
boundary conditions on equation 11.1.29, de- C - Co
pending on whether a completely open or (11.1.32)
completely closed vessel, a closed-open vessel,
or an open-closed vessel is assumed. Different where the term in parentheses is the argument
solutions will be obtained for different boundary of the error function.*
conditions. Fortunately, for small values of the The relative response at the end of a tubular
dispersion parameter, the differences between reactor of length L is identical with the F(t)
the various solutions will be small.
If we now consider a step change in tracer * The error function is tabulated in most handbooks of
concentration in the feed to an "open" tube that mathematical tables. It is useful to note that erf( — x) =
can be regarded as extending to infinity in both — erf(x) and that
directions from the injection point, the appro- 2 rx y
erf x = — e dy
priate initial and boundary conditions on JnJo
11.1 The Residence Time Distribution Function dF(t) 399
- erf - /— -t (11.1.35)
-[l-(t/7*
m1
(11.1.37)
The response in this case is shown in Figure It should be noted that an overall material
11.7 to be the expected step-function response. balance requires that all the tracer pass the
At the other extreme, a value of SJJUL equal monitoring station.
to infinity corresponds to an ideal stirred tank
reactor. CL(t)rodt = mT (11.1.38)
In the case where one injects a perfect pulse
of tracer Levenspiel and Smith (8) have shown Combining equations 11.1.36 to 11.1.38 gives
CL(t) 1
(11.1.39)
0 CL(t)d(t/V (t/1*)
uL
that the solution to equation 11.1.29 is If the right side of this equation is plotted
versus dimensionless time for various values of
MjL
c = the group Q)JuL (the reciprocal Peclet number),
the types of curves shown in Figure 11.8 are
where obtained. The skewness of the curve increases
with SiJuL and, for small values of this param-
mT is the amount of tracer injected in eter, the shape approaches that of a normal
consistent units error curve. In physical terms this implies
L and VR are the length and volume of the test that when @L/uL is small, the shape of the
section, respectively axial concentration profile does not change
400 Deviations from Ideal Flow Conditions
2.0
I I I
1.8 —
1.6 —
^-[9L/uL) = 0.002 (small dispersion)
1.4 —
1.2 —
0.8 —
**>
I
0.6 —
Ni [@L/uL) = 0.2
\\^*~ (large dispersion)
0.4 —
\ X. {2>r/uL)= 2.0
\ ^ v / " ^ (large dispersion)
0.2
0.0 V I X^ i i
0.0 0.5 1.0 1.5 2.0 2.5
Dimensionless time, t/1*
Figure 11.8
Effluent response curves for perfect impulse injection of tracer (axial
dispersion model). (Adapted from Chemical Reaction Engineering,
Second Edition, by O. Levenspiel. Copyright © 1972. Reprinted by
permission of John Wiley and Sons, Inc.)
appreciably in the time interval required for the time at which the maximum tracer concentra-
fluid to pass the monitoring station. However, tion is observed at the monitoring station. The
when @L/uL is of order 0.01 or greater, the shape result is
changes significantly in this time interval.
Differentiation of the last equation with -l-r) (H.1.40)
respect to (t/1*) can be used to determine the T
uL
11.1 The Residence Time Distribution Function dF(i) 401
This approximation is valid to within 5% at this where the subscripts 1 and 2 refer to the up-
limit. Since the axial dispersion term itself may stream and downstream locations, respectively.
be viewed as a perturbation or correction term This equation reflects the fact that variances
for real tubular reactors, errors of this magni- are additive for flow through independent ves-
tude in Q)L lead to relatively minor errors in the sels or regions. This property implies that the
conversion predicted by the model. variance of the residence time distribution can
In principle any of the equations 11.1.40, be determined for any region if the variances
11.1.44, 11.1.45, 11.1.47, or 11.1.48 could be used of the inlet and effluent streams are known. In
to determine the dispersion parameter. How- similar fashion mean residence times are addi-
ever, both equations 11.1.40 and 11.1.44 require tive; thus, for the test section,
that one accurately determine a small difference
^test M (11.1.50)
in large numbers to evaluate Q}LjuL. Hence
equations 11.1.45 and 11.1.47 are preferred for The aforementioned investigators (10-12)
evaluation of SfJuL for open and closed vessels, have derived equations relating the measured
respectively. For small @L/uL, equation 11.1.48 mean residence times and variances to the
is appropriate. Peclet number or dispersion parameter for the
The discussion thus far presumes that a test section. For the case where the conditions
perfect pulse or step input is employed when, at both monitoring probes correspond to a
in fact, such inputs can only be approximated. doubly infinite pipe, it can be shown that
In the case of pulse inputs one is also faced
with two conflicting constraints in attempting Acx2 = (11.1.51)
uL
to generate a perfect pulse or delta function.
Since a finite amount of tracer cannot be in- If the measuring points are chosen far enough
jected in zero time, as much material should be away from the ends of the test vessel so that end
injected in as short a time as possible in order effects are negligible, this expression may be
to approximate a delta function. However, the used with confidence. Bischoff and Levenspiel
injection process should not disturb the system (12) have presented design charts that permit
significantly. This latter requirement implies one to locate monitoring stations so as to avoid
that the tracer should be injected very slowly. end effects. For example, in a packed bed re-
Fortunately, these difficulties may be circum- actor with a tube to pellet diameter ratio of 15,
vented by using the imperfect pulse method de- where the packed bed is followed by an open
scribed by Aris (10), Bischoff (11), and Bischoff tube, at least 8 pellet diameters are required
and Levenspiel (12). The technique involves between the measurement point and the open
monitoring the concentration at two points in tube if errors below 1% are to be obtained.
the test section instead of at a single point. In addition to the aforementioned slope and
It does not matter where the injection point is variance methods for estimating the dispersion
located, provided that it is upstream of the two parameter, it is possible to use transfer functions
monitoring stations. Any type of imperfect in the analysis of residence time distribution
pulse input may be employed. The variances of curves. This approach reduces the error in the
the concentration-time curves at the two moni- variance approach that arises from the "tails"
toring stations are determined and their dif- of the concentration versus time curves. These
ference is taken. tails contribute significantly to the variance and
can be responsible for significant errors in the
= o\ - o\ (11.1.49) determination of £^L.
11.1 The Residence Time Distribution Function dF(t) 403
uLj d(Z/L)2 ' -©f"° used when dealing with flow through packed
beds.
In addition to the three methods described
where
above, nonlinear regression methods or other
L is the distance between monitoring stations
transform approaches may be used to determine
L/u is again defined as 7* the dispersion parameter. For a more complete
For a section of a continuous system where treatment of the use of transform methods,
end effects are negligible several investiga- consult the articles by Hopkins et al. (15) and
tors (e.g., 13-14) have shown that the transform Ostergaard and Michelsen (14).
of this equation with respect to time leads to Illustration 11.2 indicates the use of the slope
the following expression. and variance methods for evaluating 3)JuL.
1/2
uL
= exp - 1 + ILLUSTRATION 11.2 DETERMINATION OF
uTu REACTOR DISPERSION PARAMETER FROM
1/2 EXPERIMENTAL RESIDENCE TIME DATA
this data to determine the average residence time to a pulse, the variance in t is given by
and the F(t) curve. If the pulse is assumed to
]
be perfect, what value of(@JuL) gives a reason- dt
able fit of the experimental data? Use the slope dt
and variance methods to evaluate this param- dF(t)
dt
eter. Jo dt
dF(t)
Solution — - dt - I2
Jo V- dt
First consider the slope method for determining
2JuL. From a plot of the F(t) curve the value If steady-state operation, invariant mass flow,
of dF(t)/dt at t = 1 = 374.4 sec or 0.3744 ksec and evenly spaced time increments are assumed,
is equal to 2.17 ksec" 1 . From equation 11.1.42, as in Illustration 11.1, this equation may be
written as
VdF(t) t =00
l = 0.3744(2.17)
I <2cE -i 2
2 t= 0
Thus I cE
= 0.121
uL From the data in Table 11.I.I it is evident that
A more refined value may be obtained using Since the total average residence time in the
equation 11.1.44 to determine 7*. actual reactor (7) is given by
7 374.4 T/
"^CSTR
r = = 321.5 t = (11.1.60)
1 + 2(0.0822)
1
equation 11.1.59 may be rewritten as
As a second approximation,
f + f" = ^ p (11.1.61)
= 0.1481
uL (321.5)2(2) If we are determining the response of the
series of stirred tank reactors to a step change
or in inlet tracer concentration from 0 to CQ at
— = 0.104 time zero, the initial condition for this differ-
uL
ential equation is
Subsequent iterations lead to the conclusion
that 9JuL = 0.113. This result differs from Cn = 0 for t=0 and n>0
that obtained by the slope approach by 7%. The solution to this equation may be obtained
Note that if one has a value for VR/% as in the using the integrating factor approach to give
normal situation, the iterative procedure is un-
necessary. However, in this case equation 11.1.44 Cn = e~nt/r (11.1.62)
should not be used to determine @L/uL, because
significant errors may be involved in accurately The integral may be evaluated for each stage
determining a small difference in large numbers. of the reactor network in turn. For the first
stage, Cn^1 = Co, so that
or
e2tlT - 1) dt =
where t is the mean residence time for the two reactor cascade.
One may proceed stepwise in this fashion to develop a general recursion formula for the
concentration leaving reactor) in an n reactor cascade.
,-ntll nt nt nt nt:
(11.1.68)
The response curve for the network as a whole is obtained by setting ./equal to n in equation 11.1.68.
Note t h a t ! is given by equation 11.1.60.
1 nt
(11.1.69)
Note that in this case the right side of equation 11.1.68 is zero for t = 0 and unity for t = oo. Figure
11.9 contains several F(t) curves for various values of n. As n increases, the spread in residence time
decreases. In the limit, as n approaches infinity the F(t) curve approaches that for an ideal plug flow
reactor. If the residence time distribution function given by 11.1.69 is differentiated, one obtains an
0.75 -
0.25 —
Figure 11.9
Residence time distribution curves for the n-CSTR model.
11.1 The Residence Time Distribution Function dF(t) 407
equation for the slope of the curve from An alternative to the slope approach to
which n may be determined by comparison determining the appropriate value of n for use
with the experimental F{t) curve. in model calculations is based on a determina-
tion of the variance of the response of the actual
=w
W "^[ 1 + T + ^[T]a + "'-+ ( ^ [ T r
ntlT (11.1.70)
(n - 2)!
tion. The closest approximation to this condition S: No mixing of the contents of the elements
that one can encounter in engineering practice as they move through the reactor (complete
is a laminar flow reactor in which molecular segregation).
diffusion in both longitudinal and radial direc- M: Complete mixing of the contents immedi-
tions is negligible. Another real-life situation in ately on entrance followed by flow through
which a segregated flow pattern is largely the reactor in the completely mixed state.
responsible for the spread in residence times is
If we assume a rate expression of the form
one in which there is some short-circuiting or
bypassing of portions of the reactor volume as r = kCn (11.2.1)
the fluid moves through the reactor. The by-
the total conversion rates under the two lim-
passed regions may be considered as "dead"
iting circumstances outlined above are given by
spaces that contribute to the^ total reactor
volume but that do not contribute to the overall Rs = Vk(C1 + C\) (11.2.2)
conversion rate in proportion to their volume. and
Since they are not effectively purged by the
prevailing flow pattern, they often contain fluid (11.2.3)
in which the reaction has gone substantially to
completion. It should be evident that this For a differential reactor and the same resi-
situation should be avoided in the proper dence time the ratio of these rates will be equal
design of chemical reactors. In general, for to the ratio of the conversion levels attained.
reaction orders greater than unity, the segre-
gated flow model places an upper limit on the C,
(11.2.4)
conversion that will be obtained in an actual fu 2 -
reactor. For a first-order reaction occurring
isothermally, the model can be used to predict
Since
accurately the conversion that will be attained
C\ + C\ C2
in a real reactor whose F(t) curve is known. (11.2.5)
The details of the analysis are discussed in
Section 11.2.1. unless n = 1 or n = 0, these are the only cases
In general, the larger the range of the distri- for which the two extreme situations will con-
bution of residence times, the greater the verge to the same limit. In these cases the fraction
discrepancy between the conversion levels pre- conversion will be the same whether or not fluid
dicted on the basis of the segregated flow model elements having different reactant concentra-
and those predicted by the various mixing tions remain segregated or are mixed prior to or
models. For narrow distribution functions the during the reaction.
conversions predicted by both models will be For reaction orders greater than unity the
in good agreement with one another. segregated flow model will predict a higher con-
To illustrate the nature of the limits that the version level than the various mixing models.
segregated flow and mixing models place on the For reaction orders less than unity the mixing
expected conversion level, it is useful to examine models will predict higher conversion levels than
what happens to two elements of fluid that have the segregated flow model. The magnitude of the
the same volume V9 but that contain different differences for representative cases are shown in
reactant concentrations C1 and C 2 . We may Illustrations 11.5 to 11.7. The following illustra-
imagine two extreme limits on the amount of tion shows that for the first-order case, reactor
mixing that may occur. combinations involving different mixing patterns
410 Deviations from Ideal Flow Conditions
but the same residence time distribution function Solving for fA2
will give rise to the same predicted conversion.
ILLUSTRATION 11.4 COMPARISON OF fA2 = -kTc
CONVERSION LEVELS ATTAINED IN TWO
or
DIFFERENT REACTOR COMBINATIONS _ 1 + /CTC - £T k t *
HAVING THE SAME RESIDENCE TIME ^A1 ~ 1 + /CTC
DISTRIBUTION CURVE—FIRST-ORDER
REACTION Consider the alternative arrangement
The F(t) curve for a system consisting of a plug
flow reactor followed by a continuous stirred
tank reactor is identical to that of a system in
which the CSTR precedes the PFR. Show that
the overall fraction conversions obtained in these
two combinations are identical for the case of an where fAl and/^ 2 are the conversions at the out-
irreversible first-order reaction. Assume isother- let of the CSTR and the PFR, respectively.
mal operation. From the design equation for the CSTR,
Solution
T =
IA l
Let TP and TC represent the space times of the - fxi) HI - fAAl)
plug flow reactor and the continuous stirred
or
tank reactor respectively. Consider the following
reactor combination (A)
where fAi andfA2 are the conversions at the out- Integration gives
let of the first and second reactors, respectively.
From the design equation for a PFR,
dfA or
T = CA0 [Ul (B)
p AV j 0 - fA)
Combining equations (A) and (B) and rearrang-
Integration gives ing gives
kzp = - £ n ( l -fA1) e~kZp _ 1 + /CTC - e~kTp
or =
" /CT C /CT C
f — 1 p-kip
This relation is identical to that obtained for
From the design equation for a CSTR, the first arrangement. This illustration provides
an example of the general principle that for ir-
reversible first-order reactions carried out iso-
T, =
- fA2) - fA2) thermally, all reactor combinations having the
11.2 Conversion Levels in Nonideal Flow Reactors 411
same residence time distribution function give ILLUSTRATION 11.5 USE OF THE
rise to the same overall conversion. SEGREGATED FLOW MODEL TO
You should repeat the analysis for a second- DETERMINE THE CONVERSION LEVEL
order reaction to verify that for other reaction OBTAINED IN A NON-IDEAL FLOW
orders the overall conversion will depend on REACTOR
whether the PFR precedes the CSTR or vice Use the F(t) curve generated in Illustration 11.1
versa. to determine the fraction conversion that will be
achieved in the reactor if it is used to carry out
a first-order reaction with a rate constant equal
11.2.1 The Segregated Flow Model
to 3.33 x 10" 3 sec" 1 . Base the calculations on
The basic premise of the segregated flow model the segregated flow model.
is that the various fluid elements move through
the reactor at different speeds without mixing Solution
with one another. Consequently, each little fluid
Equation 11.2.7 is the key equation to the solu-
element will behave as if it were a batch reactor
tion of this problem.
operating at constant pressure. The conversions
attained within the various fluid elements will
</A> = j™°o° fA(t) dF(t) (A)
be equal to those in batch reactors with holding
times equal to the residence times of the different
For a first-order reaction,
fluid elements. The average conversion level in
the effluent is then given by
where the summation extends over all possible so that equation A becomes
residence times. In terms of an integral
)= 1.0 <fA> = l-
</.> = j ; fA(t)dF(t) (11.2.7) ^ dt
An alternative interpretation of the segregated Substituting equation 11.1.6 into this relation
flow model that leads to precisely the same result gives
involves replacing the actual reactor by a num-
dt
ber of plug flow reactors in parallel so that the f= O
(B)
combination will give rise to the F(t) curve ob-
served experimentally. Under these conditions
the degree of conversion in a fraction of the If we assume a constant density system and a
effluent stream dF(i) is equal to that which would constant mass flow rate and replace the integrals
occur in an ideal PFR with a residence time t. by finite sums, we have
When the different streams are recombined at
the outlet, the average conversion level is again £ (e-ktwE At)
given by equations 11.2.6 and 11.2.7. In order to
indicate how one makes use of these relations,
wE At
consider Illustration 11.5.
412 Deviations from Ideal Flow Conditions
If we now replace the mass fractions by con- obtained above with those that will be obtained
centrations and note that our data are recorded in Illustrations 11.6 and 11.7.
at evenly spaced time increments,
For systems with a wide distribution of resi-
I dence times, the degree of conversion can be
<fA> = 1 - calculated accurately only in the case of a first-
order reaction. However, if it is possible to
estimate in some rough fashion the extents to
The data are worked up below. which mixing and segregation effects contribute
to the observed residence time distribution, one
e-kt
can bracket the actual performance of the reactor
Time, t (sec) Q CEe~kl with the solutions obtained from the completely
0 0.0 1.00 0.0
segregated flow model and those obtained from
120 6.5 0.670 4.36 models consisting of various combinations of
240 12.5 0.449 5.61 ideal reactors. It is an extremely rare situation in
360 12.5 0.301 3.76 which one knows the flow pattern in the reactor
480 10.0 0.202 2.02 sufficiently well that an exact calculation of the
600 5.0 0.135 0.68 conversion can be performed. Instead, one must
720 2.5 0.091 0.23 be satisfied with bracketing the solution in terms
840 1.0 0.061 0.06 of models based on segregated flow and various
960 0.0 — 0.0 mixing effects. Two commonly used mixing
1080 0.0 — 0.0 models are described in the next two sections.
YfiE = 50.0 = 16.72
11.2.2 The Longitudinal Dispersion Model
Thus in the Presence of Chemical Reaction
— A + (-rA)nR2 AZ (11.2.8)
'dZ+ ) Output Disappearance
Input by reaction
11.2 Conversion Levels in Nonideal Flow Reactors 413
where we have taken into account the fact that conversions and large S!L/uL significantly larger
material enters and leaves the volume element reactors are required than would be predicted
by bulk flow and by longitudinal dispersion. If by a plug flow analysis. However, at large SiJuL,
we divide by AZ and take the limit as AZ ap- . the dispersion model may well be inappropriate
proaches zero, equation 11.2.8 becomes for use. Consequently, only the lower segment
of Figure 11.10 is of general utility for design
d2CA calculations.
2 .—±+rA = 0 (11.2.9)
Iz Numerical solutions to equation 11.2.9 have
been obtained for reaction orders other than
Note the similarities and differences between unity. Figure 11.11 summarizes the results ob-
this equation and the time-dependent equation tained by Levenspiel and Bischoff (18) for
used to evaluate the dispersion parameter second-order kinetics. Like the chart for first-
(equation 11.1.29). order kinetics, it is most appropriate for use
For a first-order irreversible reaction equation when the dimensionless dispersion group is
11.2.9 becomes small. Fan and Bailie (19) have solved the equa-
d2CA tions for quarter-order, half-order, second-order,
- kCA = 0 (11.2.10) and third-order kinetics. Others have used per-
dZ2 dZ
turbation methods to arrive at analogous results
Wehner and Wilhelm (17) have obtained an for the dispersion model (e.g. 20,21).
analytical solution to this equation for the case There are several closed form approximate
where 8A = 0. The solution is valid both when solutions to both the general and first-order
one has plug flow and when one has dispersion forms of the dispersion equations (11.2.9 and
in the regions adjacent to the test section. 11.2.10). For example, Levenspiel and Bischoff
CA
- 1 f ~ - (11.2.11)
jA
cM~ ~d
where (2, 3, 4) indicate that for small dispersion num-
bers (SiJuL) equation 11.2.11 can be rewritten
in the following form if one expands the ex-
ponentials and drops higher-order terms.
Levenspiel and Bischoff (18) have compared
this solution with that for the plug flow case. CAO
plug
= 1 - fA = e~kx = e~kLlu (11.2.13) = -kx + ( / C T ) 2 - ^ (11.2.14)
uL
They determined the ratio of the dispersion or, using equation 11.2.13 for the same reactor
reactor volume to the plug flow reactor volume space time and the same reactor volume,
necessary to accomplish the same degree of con-
version for several values of the dimensionless In fC^cdispersionN| = (kr)2 ^ (11.2.15)
dispersion parameter 2JuL. Figure 11.10 sum- V ^piug / uL
marizes their results. It is evident that for high This equation indicates that the conversion in
414 Deviations from Ideal Flow Conditions
i i 11 1 I M I II
First order
0.0625
y-j^l = 0 (plug flow)
1/ I II I I II
0.01 0.1
= C /C
A A0
Figure 11.10
Comparison of real and plug flow reactors for the first-order reaction A ->
products, assuming negligible expansion (SA = 0). (Adapted from Chemical
Reaction Engineering, Second Edition, by O. Levenspiel. Copyright © 1972.
Reprinted by permission of John Wiley and Sons, Inc.)
the dispersion reactor will always be less than where rA exit and rA0 are the rates at the exit a n d
that in the plug flow reactor (C^ dispersion > C/lplug). entrance of a plug flow reactor with the same
For the case where one fixes the effluent com- space time as the dispersion (real) reactor.
position instead of the reactor size, equations The influence of dispersion on the yield of an
11.2.13 and 11.2.14 can be manipulated to show intermediate produced in a series reaction has
that, for small also been studied. When Q)JuL is less than 0.05,
Tichacek's results (22) indicate that the frac-
-^dispersion ^dispersion 4 °^L tional decrease in the maximum a m o u n t of
— — — 1 ——
^ plugflow *pp lugflow intermediate formed relative to plug flow con-
(for same conversion) (11.2.16) ditions is approximated by Q}L/uL itself. Results
obtained at higher dispersion numbers are
Levenspiel and Bischoff (3) have used a deri- given in the original article.
vation by Pasquon and Dente (20) to obtain an Douglas and Bischoff (23) have considered the
expression that gives an approximate solution influence of volumetric expansion effects on the
to 11.2.9 for small Q)LjuL and arbitrary kinetics. yields obtained with dispersion.
CA dispersion
A plug S)' r -" ta (7f (11.2.17)
11.2 Conversion Levels in Nonideal Flow Reactors 415
100
\
0 = < 000
Siooc)
rN
/ " \
/ X 10
s,,
.2 it
N
S
\ \200 (L ines of const ant volij m e
or holdin 3 time)
/ s
10
\ >J 00
/ /
/
"^
s^ ' \
*
/ •
/^ '—«,
/
• * ^ , / /
\
X V,
\ 20
/
> >3
/
V>
/
/
\ "
7
/ 4 > ^
2
—7 » —
=
\
/
/ 0.06 25 / N
= 0 (plug flov /) 7 •
—— ss
0.01 0.1
Figure 11.11
Comparison of real and plug flow reactors for the second-order reactions
A +B products, CA0 = CB
2A products
Illustration 11.6 indicates how the longitudinal 11.1. Assume that the value of the first-order
dispersion model may be used to predict reactor rate constant is 3.33 x 10" 3 sec" 1 .
performance.
Solution
ILLUSTRATION 11.6 USE OF THE Equation 11.2.11 is applicable in this case. Using
DISPERSION MODEL TO DETERMINE THE the results obtained in Illustrations 11.1 and 11.2,
CONVERSION LEVEL OBTAINED IN A
NONIDEAL FLOW REACTOR p = [1 + 4(3.33 x l(T3)(0.113)(374.4)]1/2
= 1.25
Use the dispersion parameter determined in
Illustration 11.2 to predict the conversion that where we have used the value of @L/uL deter-
will be attained in the reactor of Illustration mined by the variance approach.
416 Deviations from Ideal Flow Conditions
Thus where
fA = 1 - 0.327 = 0.673
T = (11.2.20)
This result compares to a value of 0.666 pre-
dicted on the basis of the segregated flow model. The final effluent concentration is related to
Excellent agreement should be obtained for the the inlet concentration in the following manner.
first-order case if the dispersion parameter gives
a good fit of the experimental F(t) curve. Agree- r^
Cn
(11-2.21)
ment for reaction orders other than unity will
not be nearly as good.
where/, is the fraction conversion leaving tank n.
11.2.3 Determination of Conversion
The use of the reactor cascade model to esti-
Levels Based on the Cascade of Stirred
mate the conversion level attained in a first-
Tank Reactors Model
order reaction is discussed in Illustration 11.7.
In Section 11.1.3.2 we considered a model of
reactor performance in which the actual reactor ILLUSTRATION 11.7 USE OF THE CASCADE
is simulated by a cascade of equal-sized con- OF STIRRED TANK REACTORS MODEL TO
tinuous stirred tank reactors operating in series. PREDICT REACTOR PERFORMANCE
We indicated how the residence time distribution
Use the model based on a cascade of stirred tank
function can be used to determine the number of
tanks that best model the tracer measurement reactors to predict the conversion that will be
data. Once this parameter has been determined, attained in the reactor of Illustration 11.1. As-
the techniques discussed in Section 8.3.2 can be sume that the value of the first-order rate con-
used to determine the effluent conversion level. stant is 3.33 10" 3 sec" 1 .
If the case of an irreversible first-order reac-
tion is considered, a material balance on the nth Solution
CSTR gives In Illustration 11.1 we found! or T to be 374.4
sec, while in Illustration 11.3 we determined that
the value of n must lie between 4 and 5. From
equation 11.2.21, with n = 4,
Input Output Disappearance
by reaction 1
where / = 1 " 3.33 x 10"3(374.4)
VR is the volume of the actual reactor
VR/n is the volume of the CSTR used in the model = 1 - 0.337 = 0.662
Equations of this form were treated in Section
8.3.2.2. For a constant density system, For n = 5, / = 0.671, and for n = 4.59, / =
0.668.
Cn = (11.2.19) All these values are close to those predicted
by the segregated flow and dispersion models.
Literature Citations 417
11.3 GENERAL COMMENTS AND the reactor size may have to be increased to a
RULES OF THUMB value appreciably greater than that required by
In the previous section we indicated how various plug flow conditions. The additional volume
mathematical models may be used to simulate requirements may be determined using the mod-
the performance of a reactor in which the flow els outlined previously.
patterns do not fit the ideal CSTR or PFR condi- Note that even though various flow models
tions. The models treated represent only a small will often predict conversion levels that are
fraction of the large number that have been within a few percent of one another, one must be
proposed by various authors. However, they are extremely careful in the overall design calcula-
among the simplest and most widely used mod- tions, particularly if the conversion level is high.
els, and they permit one to bracket the expected If an ideal plug flow reactor gives a conversion
performance of an isothermal reactor. However, of 98% and proper accounting for nonideal flow
small variations in temperature can lead to much conditions by various models gives a range of
more significant changes in the reactor perfor- conversion levels from 95 to 97%, the magnitude
mance than do reasonably large deviations inflow of subsequent separation problems will vary
patterns from idealized conditions. Because the widely depending on the model chosen. For
rate constant depends exponentially on tempera- example, if product specifications call for a purity
ture, uncertainties in this parameter can lead to of 99% or better, the required separations would
design uncertainties that will make any quantita- be quite different. In the ideal case one would
tive analysis of performance in terms of the have to reduce the impurity level from 2% to 1%;
residence time distribution function little more in the nonideal flow model it would have to be
than an academic exercise. Nonetheless, there reduced from 5% to 1%. The costs of the equip-
are many situations where such analyses are ment necessary to accomplish this task would
useful. vary greatly, depending on the input impurity
Denbigh (24) has provided a set of generaliza- level. Of course, it is only fair to point out that
tions that are useful in deciding when the con- any other sources of error in the conversion
versions attained in tubular reactors will deviate estimate (e.g., temperature variations) would
significantly from those predicted on the basis of have the same effect.
the plug flow model. For laminar flow, molecular The dispersion and stirred tank models of
diffusion in the longitudinal direction will not reactor behavior are in essence single parameter
appreciably affect reactor performance if the models. The literature contains an abundance of
reactor length is much greater than its diameter. more complex multiparameter models. For an
Molecular diffusion in the radial direction tends introduction to such models, consult the review
to destroy any concentration gradients that have article by Levenspiel and Bischoff (3) and the
been established and thus serves to offset devia- texts by these individuals (2, 4). The texts also
tions resulting from the velocity profile. In other contain discussions of the means by which resi-
words, it works in favor of the idealized assump- dence time distribution curves may be used
tions instead of against them. For turbulent to diagnose the presence of flow maldistribu-
flow, eddy diffusion is the dominant mode of tion and stagnant region effects in operating
dispersion, and it can significantly affect reactor equipment.
performance. A crude generalization that is valid
for simple reactions is that if the Reynolds num- LITERATURE CITATIONS
ber is greater than 104 and if the L/D ratio of the 1. Levenspiel, O., Chemical Reaction Engineering, First
reactor is at least 50, deviations from plug flow Edition, pp. 244-53, Wiley, New York, 1962.
because of longitudinal dispersion may be ne- 2. Levenspiel, O., Chemical Reaction Engineering, Second
glected. However, if these criteria are not met, Edition, pp. 253-314, Wiley, New York, 1972.
418 Deviations from Ideal Flow Conditions
3. Levenspiel, O., and Bischoff, K. B., Adv. Chem. Eng., kg/m3 of NaCl, which does not appear to affect
4(95), 1963. the reactor performance or to be at all involved
4. Himmelblau, D. M., and Bischoff, K. B., Process Anal- in the reaction under study. Your assistant re-
ysis and Simulation, pp. 59-86, Wiley, New York, 1968.
commends that you carry out a residence time
5. Danckwerts, P. V., Chem. Eng. Sci., 2 (1), 1953. distribution study by making a step change in
6. Kramers, H., and Alberda, G., Chem. Eng. Sci., 2 (173), the inlet NaCl concentration and then observing
1953.
the effluent concentration of NaCl as a function
7. Kramers, H., and Westerterp, K. R., Elements of of time. The data below were reported for this
Chemical Reactor Design and Operation, Academic
Press, New York, 1963.
experiment. Use the data given to prepare an
F(t) curve and to determine the average residence
8. Levenspiel, O., and Smith, W. K., Chem. Eng. Sci., 6
(227), 1957. time in the reactor network.
9. Levenspiel, O., and Bischoff, K. B., op. cit, p. 112.
10. Aris, R., Chem. Eng. Sci., 9 (266), 1959. Effluent
11. Bischoff, K. B., Chem. Eng. Sci., 72(69), 1960. concentration
12. Bischoff, K. B., and Levenspiel, O., Chem. Eng. Sci., Time (kg/m3)
77(245), 1962.
13. Mixon, F. O., Whitaker, D. R., and Orcutt, J. C , 8:00 p.m. 1.0
AIChEJ., 13(21), 1967. 8:01 1.0
8:02 1.0
14. Qstergaard, K., and Michelsen, M. L., Can. J. Chem.
Eng., 47 (107), 1969. 8:03 1.0
15. Hopkins, M. J., Sheppard, A. J., and Eisenklam, P., 8:04 1.0
Chem. Eng. Sci., 24 (1131), 1969. 8:05 1.0
16. Buffham, B. A., and Gibilaro, L. G., AIChE J., 14 8:06 1.0
(805), 1968. 8:07 1.2
17. Wehner, J. F., and Wilhelm, R. H., Chem. Eng. Sci., 6
8:08 1.4
(89), 1959.
8:09 1.6
8:10 1.8
18. Levenspiel, O., and Bischoff, K. B., Ind. Eng. Chem., 51
(1431), 1959.
8:11 2.0
8:12 2.2
19. Fan, L. T., and Bailie, R. C , Chem. Eng. Sci., 13 (63),
8:13 2.4
1960.
8:14 2.6
20. Pasquon, I., and Dente, M., J. Catalysis, 1 (508), 1962. 8:15 2.8
21. Burghardt, A., and Zaleski, T., Chem. Eng. Sci., 23 8:16 3.0
(575), 1968. 8:17 3.0
22. Tichacek, L. J., AIChEJ., 9 (394), 1963. 8:18 3.0
23. Douglas, J. M., and Bischoff, K. B., Ind. Eng. Chem., 8:19 3.0
Process Design Develop., 3 (130), 1964. 8:20 3.0
24. Denbigh, K. G., Chemical Reactor Theory, pp. 50-51, 9:00 3.0
Cambridge University Press, London, 1965.
the performance of the unit approximated that 3. The residence time distribution for a con-
of an ideal continuous stirred tank reactor, these tinuous stirred tank reactor may be represented
investigators carried out a series of tracer experi- in terms of the F(t) curve as
ments at different agitator speeds and different F(t) = 1 - e'tlx or 1 - e~tlT
volumetric flow rates. The response of a non-
where the mean residence time 1 and the space
reacting system to a pulse input of helium in-
time T are identical when there is no volume
jected into a steadily flowing air stream was used
change on reaction. Show that the fraction con-
to characterize the reactor. The data are nor-
version in a CSTR calculated for a first-order
malized by assuming that the concentration of
reaction on the basis of the design equations will
the effluent at time zero is equal to the total mass
be identical with that predicted by the segregated
of tracer injected divided by the effective volume
flow model and the above F(t) curve. If k = 3.0
of the reactor. On the basis of the two sets of data
ksec" 1 , the reactor volume is 1.2 m 3 , and the
presented below, does the reactor appear to be a
feed rate is 4.0 m 3 /ksec, what will this conversion
good approximation to an ideal CSTR? Be sure
be?
to use all the data in your analysis.
4. The following data have been reported as a
result of an effort to determine the distribution
Run 1 of residence times in a packed bed reactor. Use
Volumetric flow rate: 6.25 cmVsec these data to generate an F(t) curve and to deter-
Agitator speed: 1290 rpm mine the average residence time in the reactor.
A pulse of tracer is fed to the reactor at time
^effluent/*-0
where C o is the zero.
normalizing concentration tjx
Time, t Effluent tracer
0.85 0.13 (sec) level (g/m3)
0.72 0.30
0.51 0.65 0 0.0
0.37 0.95 48 0.0
0.275 1.25 96 0.0
0.215 1.55 144 0.1
0.16 1.85 192 5.0
240 10.0
288 8.0
336 4.0
Run II 384 0.0
Volumetric flow rate: 10.6 cmVsec 432 0.0
Agitator speed: 630 rpm
If one desires to utilize this reactor to carry
c effluent /c 0 t/x
out a first-order isomerization reaction of the
0.80 0.20 type A -• R, and if the rate constant for the
0.65 0.40 reaction is 7.5 ksec" 1 , determine the average
0.49 0.70 conversion that one expects in this reactor. Com-
0.34 1.10 pare this value with those one would obtain in an
0.26 1.25 ideal PFR and in an ideal CSTR having the
0.19 1.63 same average residence time as our actual
0.14 2.05 reactor.
420 Deviations from Ideal Flow Conditions
5. The following results were reported from the after the change. It may be assumed that the
operation of a flow reactor in which the reaction total mass flow rate (Om) remains substantially
fci
constant.
A -> B was taking place. The reaction is first-
order, irreversible with k1 = 0.0433 sec" l. Pure 8. Levenspiel and Smith [Chem. Eng. Sci., 6
A enters the reactor. The exit stream consists of (227), 1957] have reported the data below for a
\0%A and 90% B. residence time experiment involving a length of
2.85 cm diameter pyrex tubing. A volume of
Time, / (sec) KMnO 4 solution that would fill 2.54 cm of the
tube was rapidly injected into a water stream
24 0.00 with a linear velocity of 35.7 cm/sec. A photo-
30 0.01 electric cell 2.74 m downstream from the injec-
45 0.15 tion point is used to monitor the local KMnO 4
54 0.40 concentration. Use slope, variance, and maxi-
60 0.50
mum concentration approaches to determine
66 0.60
the dispersion parameter. What is the mean
90 0.80
180 0.95 residence time of the fluid?
360 1.00
KMnO 4 concentration
Time, t (sec) (arbitrary units)
(a) Determine t, the average residence time for
this system. 0 0
(b) If the same conversion is to be obtained in an 2 11
ideal plug flow reactor, what is the corre- 4 53
sponding value of T? 6 64
(c) What is the value of 7 for a CSTR for the 8 58
same conversion? 10 48
(d) Comment on your results for parts (a), (b), 12 39
and (c). 14 29
16 22
18 16
6. The F(t) curve for a system consisting of a 11
20
plug flow reactor followed by a continuous 22 9
stirred tank reactor is identical to that of a 24 7
system in which the CSTR precedes the PFR. 26 5
Show that the overall fraction conversions ob- 28 4
tained in these two combinations are different 30 2
when the reactions are other than first-order. 32 2
Derive appropriate expressions for the case of 34 2
second-order irreversible reactions and indicate 36 1
how the reactors should be ordered so as to 38 1
40 1
maximize the conversion achieved.
42 1
7. Derive the F(t) curve for a CSTR by con-
sidering its response to a step change in the input 9. Bucky Badger has been investigating the resi-
tracer concentration. Let WQ and H?Q represent dence time characteristics of a reactor that he
the weight fraction tracer in the feed before and plans to use for pilot plant work. He suggests
Problems 421
that the F(t) curve can be approximated by the 11. In an effort to determine the cause of low
following relation. yields from a reactor network, your technicians
F(t) = 0 0 < t < 0.4 have carried out some tracer studies in which
F{t) = 1 - g-i-25<i-o.4) > 0.4
4.000 kg of an inert material are quickly injected
t
at the feed port. The tracer levels leaving the
where all times are measured in kiloseconds. reactor at various times after injection were as
(a) Calculate the average residence time in the follows.
reactor.
(b) If Bucky uses this reactor to carry out a first- Time, t (sec) Tracer concentration (kg/m3)
order reaction of the type A -> P, calculate
the fraction conversion expected for iso- 12 1.960
24 1.930
thermal operation with a reaction rate con-
120 1.642
stant equal to 0.8 ksec~ *. Use the segregated
240 1.344
flow model. 600 0.736
(c) Calculate the overall conversion expected 1200 0.268
for this reaction in a reactor network con- 2400 0.034
sisting of a PFR with a space time of 0.4 ksec 3600 0.004
and a CSTR with a space time of 0.8 ksec.
Consider the case where the PFR is first as At any time the reactor contains 2 m 3 of fluid.
well as the case where it is second. The feed and effluent rates remain constant at
3.3 m3/ksec. Does the response of the system
10. Use the F(t) curve for two identical CSTR's
approximate that of any simple ideal reactor?
in series and the segregated flow model to predict
What conversion level is expected if the reaction
the conversion achieved for a first-order reaction
has a first-order rate constant of 15 sec" ll
with k = 0.4 ksec" *. The space time for an in-
dividual reactor is 0.9 ksec. Check your results 12. Hopkins and co-workers [Chem. Eng. Set,
using an analysis for two CSTR's in series. 24 (1131), 1969] have indicated that the data in
140
o
\
120
1 :1
/ %I
100
Output curve commences
\ < 62.5 sec after input
80
X
h/ <1
S 60
o Input
\ ^ Output
40
I x-
c
20
L
4X x X
c
2 4 6 8 10 12 14 16 18
y
20
Chart length, in. 1 in. = 12.5 sec
Figure 11P.1
Tracer response curve. Reprinted from M. J. Hopkins, A. J.
Sheppard and P. Eisenklam, Chem. Eng. Sci., 24 (1131), 1969.
Used with permission of Pergamon Press, Ltd.
422 Deviations from Ideal Flow Conditions
two reactors will be the same as in part (a) (a) Assuming ideal pulse injection determine
What ratio of reactor volumes gives a (1) The F(t) curve for overall residence times
good representation of the F(t) curve? for both cases.
What conversion is expected on the basis (2) The F(t) curve for the contact time
of this model? Comment on the suitability distribution.
of the model. (3) The effective dispersion parameter for
(f) Prepare a table summarizing the fraction the inert gas case (<3JuL).
conversions predicted by the various (b) For a first-order irreversible isomerization
models. reaction (A -> R)withk = 0.45 sec" 1 , deter-
In parts (c), (d), and (e) draw the F(t) curves pre- mine the conversions predicted using the
dicted on the basis of the various models so that following reactor models.
they may be compared with the experimental (1) Plug flow with T =1 for reactive gas
data. experiment and with T = average con-
tact time.
15. When catalytic reactions are carried out in a
(2) CSTR with T = 1 for reactive gas experi-
fluidized bed reactor, one may observe conver-
ment and with T = average contact time.
sions lower than those predicted by either PFR
(3) Segregated flow with reactive gas resi-
or CSTR models. Such observations are often
dence time curve.
attributed to gas bypassing by bubbles that do
(4) Segregated flow with reactive gas contact
not efficiently contact the catalyst. Iwasaki et al.
time curve.
[Kagaku Kogaku,4 (170), 1966] have described
It is assumed that the reactive gas A has
a method for measuring the distribution of times
adsorption characteristics similar to those
that reactive gases are in contact with the solid.
of the gas used in the pulse experiment.
The technique is based on pulse injections of two
gases, one of which is adsorbed by the solid and
the second of which is not adsorbed. Typical
16. The techniques that we have developed in
data are given in the table below.
this chapter for the analysis of residence time
distribution functions can be used in analyzing
Time, t (sec) Concentration (arbitrary units) flow conditions in a stream or river where one
wishes to determine the dispersion of pollutants
Inert gas Adsorbed gas
from a given source. The following data are
0.0 0.0 0.0 taken from a USGS survey of the South Platte
0.8 0.0 0.0 River [R. E. Glover, "Dispersion of Dissolved
1.6 0.0 0.0 or Suspended Materials in Flowing Streams,"
2.4 0.01 0.01 Geological Survey Professional Paper 433-B
3.2 5.0 5.0 (1964)].
4.0 10.0 7.5 Average flow rate: 554 ft3/sec
4.8 8.0 8.5 Length of reach: 19,900 ft
5.6 4.0 8.0 Naturally occurring concentration of K + in the
6.4 0.0 6.5 stream: 8.2 mg/liter
7.2 0.0 4.0
At time zero, 1000 lb of K 2 CO 3 are dumped
8.0 0.0 2.0
into the upstream end of the reach. Periodically
8.8 0.0 1.0
9.6 0.0 0.0 samples were taken at the downstream end of
the reach and analyzed for K + ions with the
424 Deviations from Ideal Flow Conditions
1. Reactors in which the solid catalyst particles The use of multiple catalyst sections usually
remain in a fixed position relative to one arises because of the need to maintain adequate
another (fixed bed, trickle bed, and moving temperature control within the system. Other
bed reactors). constraints leading to the use of multiple beds
2. Reactors in which the particles are suspended include those of pressure drop or adequate fluid
in a fluid and are constantly moving about distribution. In addition to the shell and tube
(fluidized bed and slurry reactors). configuration, some of the possibilities for heat
transfer to or from fixed bed reactors include
In the sections that follow, each of these the use of internal heat exchangers, annular
reactor classes is discussed in more detail, with cooling spaces or cooling thimbles, and cir-
particular emphasis on fixed and fluidized bed culation of a portion of the reacting gases
reactors. through an external heat exchanger.
425
426 Reactor Design for Heterogeneous Catalytic Reactions
The packing itself may consist of spherical, types. One of its prime attributes is its simplicity,
cylindrical, or randomly shaped pellets, wire with the attendant consequences of low costs
screens or gauzes, crushed particles, or a variety for construction, operation, and maintenance
of other physical configurations. The particles relative to moving bed or fluidized bed opera-
usually are 0.25 to 1.0 cm in diameter. The tion. It requires a minimum of auxiliary equip-
structure of the catalyst pellets is such that the ment, and is particularly appropriate for use in
internal surface area far exceeds the superficial small commercial units when investments of
(external) surface area, so that the contact area large sums for control, catalyst handling, and
is, in principle, independent of pellet size. To supporting facilities would be economically
make effective use of the internal surface area, prohibitive. Another major advantage of this
one must use a pellet size that minimizes mode of operation is implicit in the use of the
diffusional resistance within the catalyst pellet term "fixed bed reactor"; i.e., there are no
but that also gives rise to an appropriate pressure problems in separating the catalyst from the
drop across the catalyst bed. Some considera- reactor effluent stream. (In many fluidized bed
tions which are important in the handling and systems, catalyst recovery can be quite trouble-
use of catalysts for fixed bed operation in some and require substantial equipment costs.)
industrial situations are discussed in the Catalyst Another important attribute of fixed bed re-
Handbook (1). actors is the wide variation in space times at
The most commonly used direction of re- which they can be operated. This flexibility is
actant flow is downward through the bed. This extremely important in situations where one is
approach gives a stable bed that will not fluidize, likely to encounter wide variations in the
dance, or lift out of the reactor. This approach quantity or quality of the feedstock to be pro-
minimizes catalyst attrition and potential en- cessed. For extremely high temperature or high
trainment of catalyst fines. When processing pressure reactions employing solid catalysts,
conditions are such that the reactor is subjected economic considerations usually dictate that the
to wide variations in feed flow rates or when process becomes commercially viable only when
the feed is a dense fluid, it is imperative that the a fixed bed reactor is employed.
flow direction be downward. The attendant There are some cases where the disadvantages
disadvantages of downward flow are the ten- of fixed bed operation prevent the use of such
dency of the bed to compress itself and the reactors. Heat transfer to or from a large fixed
gravitation of catalyst fines (resulting from bed of catalyst often represents a significant
attrition) down through the bed. Both phe- problem. In some cases it is possible to circum-
nomena may lead to increased pressure drop vent this potential difficulty by appropriate
and channeling or maldistribution of the flow. variations in the physical configuration of the
Upflow has the advantage of lifting catalyst catalyst bed (e.g., using a shell and tube arrange-
fines or fragmented particles from the bed ment). There are also a variety of operating
thereby avoiding channeling and blockage of techniques that can be used to facilitate control
the bed. However, this mode of operation is over the bed temperature. These include the use
disadvantageous because it may lead to un- of inert diluents in the feed stream to moderate
stable beds at high flow rates. It leads to dancing the temperature changes, and the use of the
in a pulsating flow that causes catalyst abrasion "cold shot" or gas bypass technique in which a
and, in unusual circumstances, may lead to fresh cold reactant stream is mixed with a hot
fluidization. stream that has undergone partial reaction.
A fixed bed reactor has many unique and Recycle of a product stream and partial tem-
valuable advantages relative to other reactor porary poisoning of the catalyst are two other
12.1 Commercially Significant Types of Heterogeneous Catalytic Reactors 427
expedients that can be used to help regulate the rapid, costs associated with the catalyst re-
temperature in fixed bed reactors. generation or replacement step may render the
The problem of heat transfer is a difficult one entire process unattractive from a commercial
because the rate of energy release or consump- standpoint. The effective catalyst life necessary
tion along the length of the reactor is not uni- to render the process economic for fixed bed
form. The major share of the reaction takes operation depends on the details of the process
place near the reactor inlet. In an exothermic under study, but if the lifetime is not at least
reaction, the rate will be relatively large near several months, the costs of the shutdowns will
the inlet because of high reactant concentration^. normally be exorbitant. In situ regeneration
If the rate of energy release is greater than the offers a possible way around this difficulty, but
rate at which heat can be transferred to a coolant if continuous operation is to be maintained, the
fluid, the reacting fluid will continue to heat up, use of in situ regeneration requires two or more
leading to an even faster reaction rate. This reactors in parallel with concomitant higher
phenomenon continues as the mixture moves capital costs. The most successful fixed bed
down the tube, until the decline in reactant operations are those where the catalyst activity
concentration has a larger effect on the rate than does not decline markedly over long time
the temperature increase. The net result is that periods. A technique that occasionally can be
for exothermic reactions one often observes a used to prolong the time between regenerations
maximum in a plot of temperature versus reactor and shutdowns is the use of catalyst beds that
length. are longer than are required to accomplish the
The maintenance of uniform flow distribution desired degree of conversion. If the required bed
in fixed bed reactors is frequently a problem. depth is relatively short, doubling or tripling
Maldistribution leads to an excessive spread in the depth of the catalyst packing will greatly
the distribution of residence times with adverse prolong the time that the unit can remain on
effects on the reactor performance, particularly stream. When the unit is first brought on stream,
when consecutive reactions are involved. It may the desired conversion will be accomplished in
aggravate problems of hot-spot formation and the first portion of the bed. As the catalyst
lead to regions of the reactor where undesired activity declines, the portion of the bed in which
reactions predominate. Disintegration or attri- the bulk of the reaction is accomplished will
tion of the catalyst may lead to or may aggravate move down the bed until, finally, insufficient
flow distribution problems. catalyst activity remains to accomplish the
Another disadvantage of fixed bed reactors is required degree of conversion. Unfortunately,
associated with the fact that the minimum pellet this approach can be used only with certain
size that can be used is restricted by the per- reactions, the most common example being the
missible pressure drop through the bed. Thus ammonia synthesis reaction.
if the reaction is potentially subject to diffusional
limitations within the catalyst pore structure, 12.1.1.2 Trickle Bed Reactors (2). A trickle bed
it may not be possible to fully utilize all the reactor utilizes a fixed bed over which liquid
catalyst area (see Section 12.3). The smaller the flows without filling the void spaces between
pellet, the more efficiently the internal area is particles. The liquid usually flows downward
used, but the greater the pressure drop. under the influence of gravity, while the gas
One of the major disadvantages of fixed bed flows upward or downward through the void
operation is that catalyst regeneration or re- spaces amid the catalyst pellets and the liquid
placement is relatively difficult to accomplish. holdup. Generally cocurrent downward flow of
If the catalyst deactivation rate is sufficiently liquid and gas is preferred because it facilitates
428 Reactor Design for Heterogeneous Catalytic Reactions
a uniform distribution of the liquid across the of these topics outlined in Sections 12.3 and 12.4
catalyst bed and permits the employment of are in general applicable to trickle bed reactors,
higher liquid flow rates before encountering analyses specific to such reactors have been
flooding constraints. reviewed by Satterfield (2).
The primary uses of trickle bed reactors are
for hydrodesulfurization, hydrocracking, and 12.1.1.3 Moving Bed Reactors. In moving bed
hydrotreating of various high-boiling petroleum reactors, a fluid phase passes upward through
fractions. The direct and capital costs are signi- a packed bed of catalyst pellets. Solid is fed to
ficantly less for trickle bed operation than for the top of the bed, moves downward under the
an equivalent hydrodesulfurization unit oper- influence of gravity in a manner approximating
ating entirely in the vapor phase. The use of plug flow, and is removed from the bottom. The
this reactor type makes it possible to process catalyst pellets are then transferred to the top
feedstocks with such high boiling points that of the reactor in external equipment by pneuma-
straight vapor phase operation would lead to tic or mechanical means in continuous fashion.
excessive undesirable side reactions. During the period when the catalyst pellets
In spite of the fact that trickle bed reactors are not in the reactor proper, they may be re-
often approach plug flow behavior, the use of a generated or reconditioned in an auxiliary
liquid phase in the feed introduces several facility.
complications in the design analysis. The nature It is necessary to design special control valves
and extent of the liquid distribution within the to provide proper solids flow and to maintain
catalyst bed vary drastically with changes in the close control over the solids level within the
liquid and vapor flow rates, the properties of the reactor. Moreover, care must be taken in the
reaction mixture (especially its viscosity and design of these reactors to prevent bypassing of
wetting characteristics), and the design of the the bed by the fluid reactant stream, and to
reactor (especially the liquid distribution sys- ensure good distribution of the solids at all
tem). As the liquid distribution changes, there are levels. These potential difficulties indicate why
concomitant changes in the contacting efficien- these types of reactors are less frequently used
cies between the liquid, vapor, and catalyst. than fixed or fluidized bed reactors. When
Usually, gaseous reactants must first be ab- catalyst decay is slow, so that fixed bed opera-
sorbed and transported across a thin liquid film tion is satisfactory with infrequent shutdown for
to the exterior of the catalyst pellet. They must catalyst regeneration, moving bed operation
then be subsequently transported through liquid does not offer a viable alternative. In catalytic
filled pores to sites in the catalyst interior. By cracking where catalyst deactivation is so rapid
using a low ratio of liquid to catalyst in the that fixed beds must be regenerated after only
reactor, one is able to minimize the extent of a few minutes of time on stream, the disadvan-
the homogeneous reaction. However, a balance tages of these reactors are obvious. Either
must be struck for highly exothermic reactions, moving bed operation or fluidized bed opera-
since the energy released by reaction may be tion is imperative in such cases. The primary
sufficient to volatilize the liquid in substantial disadvantage of moving bed operation vis a vis
portions of the bed. In such cases, portions of fluidized bed operation is in its reduced capa-
the bed will not be wetted, and there will be bility to facilitate heat transfer in either the
poor contacting between liquid and catalyst. reactor proper or the regeneration unit. This
One often finds that either external or intra- reduced capability makes it difficult to achieve
particle mass transfer effects are significant in the control of the catalyst temperature necessary
reactors of this type. Although the treatments to prevent high temperature deactivation pro-
12.1 Commercially Significant Types of Heterogeneous Catalytic Reactors 429
cesses. In instances where catalyst deactivation Typically the catalyst particles used in fluid-
is not rapid but is still appreciable and can be ized bed operations have dimensions in the
circumvented by intermittent regeneration, range of 10 to 300 microns. For optimum
moving bed operation may offer advantages fluidization, it is important to employ the proper
over using several fixed bed reactors in parallel. particle size distribution. Beds of large uni-
The economics are dependent on the deactiva- formly sized solids often fluidize poorly with
tion rate and other process details, and no bumping, slugging, and spouting, which can
general rule of thumb is appropriate. cause serious structural damage in large beds.
The catalyst in moving bed operations usually In such cases the quality of the fluidization
has an important role as a heat carrying medium. can frequently be drastically improved by adding
The energy released by carbon burnoff in small amounts of fines. Fine particles with a
exothermic regeneration reactions can be used wide size distribution will remain fluidized over
to supply some of the energy requirements of a wide range of gas flow rates, permitting
endothermic cracking reactions with attendant flexible operation with relatively large beds.
overall energy savings. Commercial scale fluidized bed reactors may
be extremely large pieces of equipment. Diam-
12.1.2 Heterogeneous Catalytic Reactors eters of 10 to 30 ft are not unusual for catalytic
in which there is Significant Motion of the cracking units. Typically, height to diameter
Catalyst Particles Relative to One Another ratios of 2:1 or greater are employed. Large
systems are required in part to accommodate
12.1.2.1 Fluidized Bed Reactors. A fluidized heat transfer equipment, cyclone separators,
bed reactor is one in which relatively small and other internal equipment.
particles of catalyst are suspended by the Several advantages are associated with the
upward motion of the reacting fluid. In virtually use of fluidized bed reactors. A remarkably uni-
all industrial applications the fluid is a gas which form temperature can be maintained throughout
flows upward through the solid particles at a the catalyst bed. This property is a consequence
rate which is sufficient to lift them from a of the high degree of turbulence within the bed,
supporting grid, but which is not so large as to the high heat capacity of the solid catalyst
carry them out of the reactor or to prevent them comprising the bed relative to the gas contained
from falling back into the fluidized phase above therein, and the extremely high interfacial area
its free surface. The particles are in constant for heat transfer between the solid and the gas
motion within a relatively confined region of phase. These factors facilitate control over the
space, and extensive mixing occurs in both the temperature of the reactor and its contents.
radial and longitudinal directions of the bed. This in turn enhances the selectivity that can
Fluidized bed reactors were first employed be achieved and permits very large-scale
on a large scale for the catalytic cracking of operations.
petroleum fractions, but in recent years they The primary advantage of fluidized bed
have been employed for an increasingly large reactors, however, is that they permit conti-
variety of reactions, both catalytic and non- nuous, automatically controlled operations
catalytic. The catalytic reactions include the using reactant-catalyst systems that require
partial oxidation of naphthalene to phthalic an- catalyst regeneration at very frequent intervals.
hydride and the formation of acrylonitrile from Fluidized bed operation permits one to easily
propylene, ammonia, and air. The noncatalytic add or remove the catalyst from the reactor or
applications include the roasting of ores and the regenerator. Regeneration can be accom-
f
he fluorination of uranium oxide. plished by any convenient procedure, but the
430 Reactor Design for Heterogeneous Catalytic Reactions
use of fluidized bed regeneration permits con- In spite of the drawbacks enumerated above,
tinuous operation and is usually most economi- fluidized bed reactors have a number of com-
cal. Furthermore, the circulation of solids pelling advantages, as we have noted previously.
between two fluidized beds makes it possible By proper design it is possible to overcome
to transport large quantities of energy between their deficiencies so that their advantages pre-
the reactor and the regenerator. This feature is dominate. This book does not discuss in detail
particularly useful in catalytic cracking reac- the manner in which this problem can be solved,
tions where the exothermic regeneration reac- although the design considerations outlined in
tion can be used thereby to supply some of the subsequent sections of this chapter are quite
energy requirements for the endothermic crack- pertinent. For detailed treatments of fluidized
ing reaction. bed reactor design, consult the excellent refer-
Still another advantage of fluidized bed ence works by Kunii and Levenspiel (3) and by
operation is that it leads to more efficient Davidson and Harrison (4).
contacting of gas and solid than many com-
petitive reactor designs. Because the catalyst
72.7.2.2 Slurry Reactors. Slurry reactors are
particles employed in fluidized beds have very
commonly used in situations where it is neces-
small dimensions, one is much less likely to
sary to contact a liquid reactant or a solution
encounter mass transfer limitations on reaction
containing the reactant with a solid catalyst. To
rates in these systems than in fixed bed systems.
facilitate mass transfer and effective catalyst
Disadvantages are also associated with flu-
utilization, the catalyst is usually suspended in
idized bed reactors. They cannot be used with
powdered or in granular form. This type of
catalyst solids that will not flow freely or that
reactor has been used where one of the reactants
have a tendency to agglomerate. Attrition of
is normally a gas at the reaction conditions and
the solids also causes some loss of material as
the second reactant is a liquid, e.g., in the hydro-
fines, which are blown out of the reactor.
genation of various oils. The reactant gas is
Extensive solids collection systems including
bubbled through the liquid, dissolves, and then
cyclone separators and electrostatic precipi-
diffuses to the catalyst surface. Obviously mass
tation must often be provided to minimize
transfer limitations can be quite significant in
catalyst losses and contamination of the en-
those instances where three phases (the solid
vironment. Another disadvantage of fluidized
catalyst, and the liquid and gaseous reactants)
bed operation is that it leads to a larger pressure
are present and necessary to proceed rapidly
drop than fixed bed operation with concomitant
from reactants to products.
higher operating costs. Erosion of pipes and
Satterfield (5) has discussed several advan-
reactor internals via abrasion by the particles
tages of slurry reactors relative to other modes
can occur. In general, operating and mainte-
of operation. They include the following.
nance costs will be relatively high for this mode
of operation compared with similar scale 1. A well-agitated slurry may be kept at a
operations with other reactor types. Fluidized uniform temperature throughout, eliminating
bed operations also have the disadvantage that "hot" spots that have adverse effects on
the fluid flow deviates markedly from plug catalyst selectivity.
flow, and the bypassing of solids by bubbles 2. The high heat capacity associated with the
can lead to inefficient contacting. This problem large mass of liquid facilitates control of the
is particularly significant when dealing with reactor and provides a safety factor for
systems in which high conversions are desired. exothermic reactions that might lead to
It can be circumvented to some extent by using thermal explosions or other "runaway"
multiple beds in series or internal staging. events.
12.2 Mass Transport Processes within Porous Catalysts 431
3. Since liquid phase heat transfer coefficients column with various stages placed above one
are large, heat recovery is practical with another in a single shell. Since a single slurry
these systems. reactor has a residence time distribution ap-
4. The small particles used in slurry reactors proximating a CSTR, the last mode of construc-
may make it possible to obtain much higher tion gives an easy means of obtaining stagewise
rates of reaction per unit weight of catalyst behavior and more efficient utilization of the
than would be achieved with the larger pellets reactor volume.
that would be required in trickle bed reactors. For more detailed treatments of slurry re-
This situation occurs when the trickle bed actors, see the texts of Satterfield (5) and Smith
pellets are characterized by low effectiveness (6).
factors (see Section 12.3).
5. Continuous regeneration of the catalyst can
12.2 MASS TRANSPORT PROCESSES
be obtained by continuously removing a
WITHIN POROUS CATALYSTS
fraction of the slurry from which the catalyst
is then separated, regenerated and returned Heterogeneous catalysis involves an interaction
to the reactor. between a solid substrate and reactant molecules
6. Since fine catalyst particles are desired, the in a fluid phase. This interaction occurs at the
costs associated with the pelleting process interface between the two phases, and efficient
are avoided, and it becomes possible to use utilization of the catalyst requires that the sur-
catalysts that are difficult or impossible to face area per unit weight of catalyst be as large
pelletize. as possible. For commercial applications, this
requirement implies that the catalyst should be
A major deterrent to the adoption of con- porous with a large internal surface area. Typical
tinuous slurry reactors is the fact that published industrial catalysts have specific surface areas
data are often inadequate for design purposes. ranging from 1 to 1000 m 2 /g. With the exception
Solubilization and mass transfer processes may of the use of wire gauzes for high temperature
influence observed conversion rates and these oxidation processes (e.g., NH 3 or SO 2 oxida-
factors may introduce design uncertainties. One tion), nonporous catalysts have not been used
also has the problems of developing mechanical extensively in commercial applications. For
designs that will not plug up, and of selecting porous catalysts the superficial external surface
carrier liquids in which the reactants are soluble of the catalyst will normally represent only an
yet which remain stable at elevated temperatures insignificant fraction of the total surface area.
in contact with reactants, products, and the A spherical catalyst pellet with a diameter of
catalyst. A further disadvantage of the slurry 0.5 cm weighing 0.1 g would have a geometric
reactor is that the ratio of liquid to catalyst is surface area of only 7.8 x 10" 5 m 2 . If it had
much greater than in a trickle bed reactor. a modest specific surface area of 10 m 2 /g, the
Hence, the relative rates of undesirable homo- internal surface area would be 1 m2. Hence,
geneous liquid phase reactions will be greater even a thin outer shell of the catalyst, which is
in the slurry reactor, with a potential adverse only 0.05 cm deep, would have a surface area
effect on the process selectivity. associated with it that would be orders of
Slurry reactors may take on several physical magnitude greater than the external surface
forms: they may be simple stirred autoclaves; area. Consequently, to obtain high specific
they may be simple vessels fitted with an external activities, it is necessary to use highly porous
pump to recirculate the liquid and suspended catalysts in industrial situations.
solids through an external heat exchanger; or It requires only an extremely limited knowl-
they may resemble a bubble-tray rectifying edge of fluid flow processes to recognize that the
432 Reactor Design for Heterogeneous Catalytic Reactions
pressure drop through an ordinary catalyst bed Unfortunately, it is not possible to obtain
is not sufficient to force any perceptible amount effective diffusivities merely by correcting bulk
of fluid through the very small pores of catalyst phase diffusivities for the reduction in cross-
particles that are required to obtain surface sectional area due to the solid phase. Two
areas of 1 to 1000 m 2 /g. If reactants are to come primary factors render this simple approach
in contact with the interior surface of catalyst invalid:
particles, they must do so by diffusion. If one
1. The geometry of the pore structure makes it
has a fast reaction and relatively long catalyst
impossible to determine accurately the effec-
pores, reactant molecules will be converted to
tive length of the diffusion path. Interconnec-
products before they have time to diffuse very
tions within the pore structure, the tortuous
far into the pore structure. In this case, the
character of individual pores, and variations
interior of the particle will be filled primarily
in cross-sectional area along the pore length
with product molecules, and only the outer
all contribute to the difficulty of the task.
peripheral layer of the catalyst pellet will be
2. One or more of several different mechanisms
fully effective in promoting reaction. Section 12.3
may be responsible for the mass transfer
discusses the problem of determining what
process. These include ordinary bulk diffu-
fraction of the total catalyst surface area is
sion, Knudsen diffusion, surface diffusion,
used effectively in promoting chemical reaction
and bulk flow. For the majority of the
and the implications of this situation for chemi-
catalysts and conditions used in industrial
cal reactor design. In the most general case, the
practice, the only significant mechanisms are
problem involves a complex interaction between
bulk diffusion and Knudsen diffusion. The
chemical reaction, diffusive mass transfer, and
relative importance of these two processes
heat transfer processes.
depends on the relative values of the mean
One must understand the physical mecha-
free path and the pore dimensions.
nisms by which mass transfer takes place in
catalyst pores to comprehend the development Ordinary or bulk diffusion is primarily respon-
of mathematical models that can be used in sible for molecular transport when the mean free
engineering design calculations to estimate what path of a molecule is small compared with the
fraction of the catalyst surface is effective in diameter of the pore. At 1 atm the mean free
promoting reaction. There are several factors path of typical gaseous species is of the order
that complicate efforts to analyze mass transfer of 10~ 5 cm or 103 A. In pores larger than
within such systems. They include the facts that 10" 4 cm the mean free path is much smaller
(1) the pore geometry is extremely complex, and than the pore dimension, and collisions with
not subject to realistic modeling in terms of a other gas phase molecules will occur much more
small number of parameters, and that (2) differ- often than collisions with the pore walls. Under
ent molecular phenomena are responsible for these circumstances the effective diffusivity will
the mass transfer. Consequently, it is often useful be independent of the pore diameter and, within
to characterize the mass transfer process in a given catalyst pore, ordinary bulk diffusion
terms of an ''effective diffusivity" i.e., a transport coefficients may be used in Fick's first law to
coefficient that pertains to a porous material in evaluate the rate of mass transfer and the con-
which the calculations are based on total area centration profile in the pore. In industrial
(void plus solid) normal to the direction of practice there are three general classes of reac-
transport. For example, in a spherical catalyst tion conditions for which the bulk value of the
pellet, the appropriate area to use in charac- diffusion coefficient is appropriate. For all
terizing diffusion in the radial direction is 4nr2. catalysts these include liquid phase reactions
12.2 Mass Transport Processes within Porous Catalysts 433
and very high pressure reactions where the T is the absolute temperature
fluid density approaches the critical density of M is the molecular weight
the material. The third class is made up of low-
AC is the concentration difference over
pressure, gas phase reactions that take place on
catalysts with very large pores, say greater than the length of the pore
5000 A in diameter. When bulk diffusion is AX is the pore length
operative, a number of correlations may be
Since n72 is the cross-sectional area for flow,
used to estimate the ordinary molecular dif-
and the term under the radical is the average
fusivity in the absence of experimental data.
molecular velocity (U), equation 12.2.1 can be
Reid and Sherwood (7) have summarized them
written in the form of Fick's first law as
in convenient form.
Knudsen diffusion will be the dominant mecha-
nism of mass transfer whenever the mean free (12.2.2)
TT? AX
path between collisions is large compared with
the pore diameter. This situation prevails when where the Knudsen diffusivity DK is defined as
the gas density is low or when the pore dimen-
sions are very small. It is not observed with
DK = -1
liquids. The molecules hitting the walls are
momentarily adsorbed and then are given off
in random directions (diffusely reflected). After The Knudsen diffusivity is thus directly pro-
a collision with the pore wall, the molecule will portional to the pore radius (7).
usually fly to another spot on the wall before Equation 12.2.3 is often written in cgs units as
having a collision with a second gas phase
molecule. Many collisions with the walls will DK = 9.7 x 10 3 r J— (12.2.4)
take place for each collision between gas phase
molecules. The molecule moves down the
where
catalyst pore by a series of random flights,
interrupted by collisions with the pore walls. DK is expressed in square centimeters per second
The gas flux is reduced by the wall "resistance." 7 is expressed in centimeters
This causes a delay because of the finite time
T is expressed in degrees Kelvin
the molecules are adsorbed and because, after
a collision, the molecule is just as apt to reverse The symbols refer to a single component. Since
its direction as it is to continue along its original molecular collisions are rare events in Knudsen
path. If the equation for the gas flow in a straight flow, flow and diffusion are synonymous and
circular pore for the case of Knudsen flow is each component of a mixture behaves as if it
analyzed according to the principles of the alone were present. Numerical values of the
kinetic theory of gases, it can be shown that the Knudsen diffusivity for molecules of ordinary
molar flow rate (F) can be written as weight at ordinary temperatures range from
0.01 cm2/sec for pores with a radius of 10 A up to
F = (n72)~7 (12.2.1) about 10 cm2/sec for pores with 10,000 A radii.
By comparing the relative magnitude of the
where mean free path (/.) and the pore diameter (27), it
is possible to determine whether bulk diffusion
r is the pore radius or Knudsen diffusion may be regarded as negli-
R is the gas constant gible. Using the principles of the kinetic theory
434 Reactor Design for Heterogeneous Catalytic Reactions
of gases, it can be shown that this ratio is equal equation 12.2.6 reduces to
to the ratio of the bulk diffusivity (DAB) to the
1
Knudsen diffusivity. (12.2.8)
l/DK + l/DA
DAB Equimolar counterdiffusion takes place in cata-
(12.2.5)
27 DK lyst pores when a reaction with a stoichiometry
of the form A -+ B occurs under steady-state
Obviously, there will be a range of pressures conditions.
or molecular concentrations over which the In the case of nonequimolal counterdiffusion,
transition from ordinary molecular diffusion to equation 12.2.6 suffers from the serious dis-
Knudsen diffusion takes place. Within this re- advantage that the combined diffusivity is a
gion both processes contribute to the mass trans- function of the gas composition in the pore.
port, and it is appropriate to utilize a combined This functional dependence carries over to the
diffusivity (i^c). For species A the correct form effective diffusivity in porous catalysts (see be-
for the combined diffusivity is the following. low), and makes it difficult to integrate the com-
bined diffusion and transport equations. As
1
(12.2.6) Smith (12) points out, the variation of 2C with
VDK - OLYA)/DAB
composition ( l y is not usually strong, and it has
been an almost universal practice to use a com-
where
position independent form of Q)c (12.2.8) in as-
YA is the mole fraction of species A in the gas sessing the importance of intrapellet diffusion.
phase In fact, the concept of a single effective diffusivity
loses its engineering utility if the dependence on
DAB is the bulk diffusivity in a binary gaseous
composition must be retained.
mixture of A and B,
Whether Knudsen or bulk diffusion domi-
a is given by nates the mass transport process depends on the
relative magnitudes of the two terms in the de-
a = 1+ ^ (12.2.7) nominator of equation 12.2.6. The ratio of the
two diffusivity parameters is obviously impor-
where NA and NB are the molar fluxes of A tant in establishing these magnitudes. In this
and B relative to a fixed coordinate system. regard, it is worth noting that DK is proportional
The combined diffusivity is, of course, defined to the pore diameter and independent of pressure
as the ratio of the molar flux to the concentration whereas DAB is independent of pore size and
gradient, irrespective of the mechanism of trans- inversely proportional to the pressure. Conse-
port. The above equation was derived by sepa- quently, the higher the pressure and the larger
rate groups working independently (8-10). It the pore, the more likely it is that ordinary bulk
is important to recognize that the molar fluxes diffusion will dominate.
(Nt) are defined with respect to a fixed catalyst Surface diffusion is yet another mechanism
pellet rather than to a plane of no net transport. that is often invoked to explain mass transport
Only when there is equimolar counterdiffusion, in porous catalysts. An adsorbed species may be
do the two types of flux definitions become transported either by desorption into the gas
equivalent. For a more detailed discussion of phase or by migration to an adjacent site on the
this point, the interested readers should consult surface. It is this latter phenomenon that is
Bird, Stewart, and Lightfoot (11). When there is referred to as surface diffusion. This phenome-
equimolal counterdiffusion NB = —NA and non is poorly understood and the rate of mass
12.2 Mass Transport Processes within Porous Catalysts 435
transfer by this process cannot be predicted with be identical with the volume fraction solids
any degree of accuracy. The interested student [(1 — Ep) where ep is the porosity of the pellet].
should refer to the discussion of Satterfield (13) Similarly, the fraction of the cross section area
and review articles on this subject (14-16). through which mass transfer can take place is
Bulk or forced flow of the Hagan-Poiseuille identical with the void fraction sp. Consequently,
type does not in general contribute significantly our first step in obtaining an effective diffusivity
to the mass transport process in porous catalysts. will be to multiply our combined diffusivity by
For fast reactions where there is a change in the ep, since the area available for mass transport
number of moles on reaction, significant pres- has been reduced by this factor from the gross
sure differentials can arise between the interior area of a cross section normal to the flow. The
and the exterior of the catalyst pellets. This second step requires recognition that the length
phenomenon occurs because there is insufficient of the diffusion path that the molecules must
driving force for effective mass transfer by forced traverse in real pores will be greater than that
flow. Molecular diffusion occurs much more of a straight line linking the origin and termina-
rapidly than forced flow in most porous catalysts. tion of the diffusion path. In addition, the
Our discussion of the various types of diffu- avenues through which diffusion proceeds are
sion has presumed that they take place in a well- irregular in shape and vary in cross section.
characterized pore structure, that is, straight Since the constrictions offer resistances to diffu-
cylindrical pores. However, the catalysts used in sion that are not offset by enlargements, both of
industry have extremely complex structures with these factors cause the flux to be less than it
interconnecting pores, tortuous pores, and wide would be in a straight cylindrical pore of the
variations in pore diameter as one moves along same length and mean radius. Consequently, if
the length of the pore. Consequently, we need to we introduce a length factor L' and a shape factor
convert the combined diffusivities that are ap- Sf (both of which are greater than unity) to allow
propriate for Knudsen and/or bulk diffusion for these effects, the relation between the effective
into effective diffusivities, which are required for diffusivity and the combined diffusivity may be
use in systems of unspecified pore geometry. written as
These effective diffusivities can be calculated if
we presume some sort of a realistic model for the ®eff = 1^7 = ™ (12.2.9)
geometry of the pore structure. The test of the
model is, of course, whether or not it predicts where the tortuosity factor T' is used to replace
results that are in accord with what is observed the factor L'S", since it is virtually impossible to
in the laboratory or the plant. The preferred separate these factors for most real porous sys-
method of obtaining an effective diffusivity for tems. For accurate work this parameter, or ^ e f f
design calculations involves measurement of itself, must be determined experimentally. There
this property. It should be noted, however, that have been several attempts to relate U and/or S'
the values obtained should be interpreted with to other readily measured properties of catalyst
care. We briefly discuss both the calculational pellets, but they are of limited applicability. The
and the experimental routes to effective diffusivi- combined factor T' has been measured for a
ties in the paragraphs below. variety of commercial catalysts. Tortuosity fac-
If the pores of a catalyst pellet are randomly tors from slightly more than one to greater than
oriented, geometric considerations require that ten have been reported (17). Values of 3 to 7 are
if one takes an arbitrary cross section of the easily reconciled with physical reality by mul-
porous mass, the fraction of the area occupied tiplying together physically reasonable values
by the solid material will be a constant that will of L and S'.
436 Reactor Design for Heterogeneous Catalytic Reactions
In the above discussion, we have presumed rate of efflux of this component into a second
that the tortuosity factor T' is characteristic of component. Although somewhat simpler experi-
the pore structure but not of the diffusing mole- mentally, this efflux method has the disadvan-
cules. However, when the size of the diffusing tage that adsorption-desorption effects on the
molecule begins to approach the dimensions surface can conceivably complicate the results
of the pore, one expects the solid to exert a for high surface area materials. Since it is not a
retarding influence on the flux and this effect steady-state experiment, the equations for non-
may also be incorporated in the tortuosity steady-state diffusion must be used in inter-
factor. This situation is likely to be significant preting data. For gases and typical catalyst
in dealing with catalysis by zeolites (molecular pellets, the time frame of the experiment is so
sieves). short that it is difficult to obtain accurate diffu-
Because a priori estimates of effective diffu- sivity values. Consequently, the technique is
sivities are not adequate for accurate design largely restricted to liquids or to very finely
calculations and recourse must be made to ex- porous solids, which give rise to very low effec-
perimental methods to evaluate these para- tive diffusivities (e.g., zeolites).
meters, a few comments on the techniques used Barrer (19) has developed another widely used
in these studies are in order. Since catalytic nonsteady-state technique for measuring effec-
reactors are normally operated at steady state tive diffusivities in porous catalysts. In this
and nearly constant pressure, effective diffu- approach, an apparatus configuration similar to
sivities are often measured under these con- the steady-state apparatus is used. One side of
straints. The most commonly used apparatus the pellet is first evacuated and then the increase
for measuring counterdiffusion rates was first in the downstream pressure is recorded as a
described by Wicke and Kallenbach (18). It function of time, the upstream pressure being
involves sealing off the sides of a cylindrical held constant. The pressure drop across the
pellet and flowing two different pure gases over pellet during the experiment is also held rela-
the two flat faces. The rate of steady-state diffu- tively constant. There is a time lag before a
sion through the pellet is determined by analyz- steady-state flux develops, and effective diffusion
ing each of the two pure streams for contamina- coefficients can be determined from either the
tion by the other component. One difficulty with transient or steady-state data. For the transient
this approach is that since it measures diffusion analysis, one must allow for accumulation or
through the pellet, it ignores dead-end pores, depletion of material by adsorption if this occurs.
which may contribute to reaction and which are The following illustration indicates how the
included in a measured pore-size distribution. It concepts we have developed thus far in this
can be used over a wide range of pressures but chapter can be used in determining effective
cannot be readily adapted to temperatures much diffusivities for use in the analyses we will de-
above ambient. The principles of kinetic theory velop in subsequent sections.
can be used to calculate effective diffusivities for
both species from measurements on only one of
ILLUSTRATION 12.1 ESTIMATION OF
the two gas fluxes. In practice, however, mea-
COMBINED DIFFUSIVITY FOR CUMENE
surements of both fluxes are desirable to facili-
IN A CRACKING CATALYST
tate the detection of leaks in the apparatus and
of the presence of surface diffusion. The cumene (isopropyl benzene) cracking re-
A second approach to measuring effective action is often used as a model reaction for
diffusivities involves filling the pore structure determining the relative activities of cracking
with one component and then measuring the catalysts.
12.2 Mass Transport Processes within Porous Catalysts 437
CH, CH 3
CH
This reaction takes place on silica-alumina cat- In terms of the indicated property values,
alysts in the temperature range from 300 to
600 °C. It is a clean reaction with negligible pro-
duction of by-products.
For the purposes of this illustrative example,
Hence
we wish to calculate the combined and effective
diffusivities of cumene in a mixture of benzene
and cumene at 1 atm total pressure and 510 °C
within the pores of a typical TCC (Thermofor
Catalytic Cracking) catalyst bead. For our pre- = 2.61 nm = 26.1 A
sent purposes, the approximation to the com- and
bined diffusivity given by equation 12.2.8 will be
sufficient because we will see that the Knudsen DK = 9.7 x 103(2.61 x 10"
diffusion term is the dominant factor in deter-
mining the combined diffusivity. = 6.46 x 10" 3 cm2/sec
Additional information. The following property
Substitution of the numerical values for the bulk
values are associated with the TCC beads:
and Knudsen diffusion coefficients into equation
12.2.8 gives
sp = 0.51
1 1 1
equivalent particle diameter = 0.43 cm = 154.8 + 6.67
(6.46 x 10" 3 ) 0.150
density of an individual particle pp = 1.14 g/cm3
A narrow pore-size distribution and a tort- or
uosity factor of three may be assumed. Using Q)c = 6.19 x 10" 3 cm2/sec
the methods suggested by Reid and Sherwood
(7), the ordinary molecular diffusivity DAB is From the magnitudes of the diffusion coefficients,
found to be 0.150 cm2/sec. it is evident that under the conditions cited the
Solution
majority of the mass transport will occur by
Knudsen diffusion. Equation 12.2.9 and the
The Knudsen diffusion coefficient may be eval- tabulated values of the porosity and tortuosity
uated from equation 12.2.4 if the catalyst prop- may be used to determine the effective diffusivity.
erty values are used to estimate the average pore
(6.19 x 10"3)(0.51)
radius. From equation C of Illustration 6.2,
_ 2K
= 1.05 x 10" 3 cm2/sec
438 Reactor Design for Heterogeneous Catalytic Reactions
12.3 DIFFUSION AND REACTION IN achieving the necessary mass transfer rate.
POROUS CATALYSTS Assuming that the catalyst particle is isothermal,
the average reaction rate under these conditions
When diffusion and reaction occur simulta- will, in general, be less than that which would
neously within a porous solid structure, con- prevail if there were no mass transfer limitations.
centration gradients of reactant and product In this case, one does not measure true intrinsic
species are established. If the various diffusional kinetics. In this section we examine the general
processes discussed in the previous section are problem of determining what fraction of the
rapid compared with the chemical reaction rate, catalyst surface area is effective when reaction
the entire accessible internal surface of the and intraparticle mass transfer effects interact.
catalyst will be effective in promoting reaction Our objective is to develop expressions for the
because the reactant molecules will spread rate of reaction averaged over the entire catalyst
essentially uniformly throughout the pore struc- particle, but expressed in terms of the tempera-
ture, before they have time to react. Here only ture and concentrations prevailing at its exterior
a small concentration gradient will exist between surface.
the exterior and interior of the particle, and Quantitative analytical treatments of the
there will be diffusive fluxes of reactant mole- effects of mass transfer and reaction within a
cules in and product molecules out that suffice porous structure were apparently first carried
to balance the reaction rate within the particle. out by Thiele (20) in the United States, Dam-
In this case, we measure true intrinsic kinetics. kohler (21) in Germany, and Zeldovitch (22) in
On the other hand, if the catalyst is very active, Russia, all working independently and reporting
many reactant molecules will have been con- their results between 1937 and 1939. Since these
verted to products before they have had time to early publications, a number of different re-
diffuse very far into the pore structure. There search groups have extended and further de-
will be steep concentration gradients of both veloped the analysis. Of particular note are the
reactant and product species near the periphery efforts of Wheeler (23-24), Weisz (25-28), Wicke
of the particle. Within the central core of the (29-32), and Aris (33-36). In recent years, several
particle, the reactant concentrations will be very individuals have also extended the treatment to
low and their gradients will be small. In the core include enzyrties immobilized in porous media
the product concentrations will be high. Vir- or within permselective membranes. The im-
tually the entire reaction will take place within portant consequence of these analyses is the
a thin shell on the periphery of the particle. In development of a technique that can be used to
this case the internal surface area is not used analyze quantitatively the factors that determine
effectively. Still, under steady-state operating the effectiveness with which the surface area of
conditions, diffusion of reactants in and pro- a porous catalyst is used. For this purpose we
ducts out must balance the rate of reaction define an effectiveness factor rj for a catalyst
within the pore. Here, however, it is large peri- particle as
pheral concentration gradients rather than high Hence the effectiveness factor is the ratio of the
diffusivities that are mainly responsible for actual reaction rate to that which would be
12.3 Diffusion and Reaction in Porous Catalysts 439
If a material balance is written over the dif- so that equation 12.3.10 may be rewritten as
ferential element of pore length shown in Figure
12.1, one finds that at steady state: d2C
= h2TC (12.3.12)
d(x/L):
Input = output + disappearance by reaction
The solution to this equation may be written in
1
Try I - 3)c —— = wr ' - c the form
dx L x+Ax
+ (2n7 Ax)(k';C) (12.3.6) c 0 cos:
(12.3.13)
cosh
where we have used the symbol Q)c to emphasize
the fact that the diffusive flux may represent a which describes the concentration profile along
combination of Knudsen and ordinary mole- the length of the pore. While this equation is not
cular diffusion and where we have assumed that our ultimate goal, a brief digression to examine a
the stoichiometric coefficient of the reactant for possible physical interpretation of the square of
which we are writing the material balance is — 1. the Thiele modulus may shed some light on the
Dividing through by Ax, taking the limit as discussion that follows.
Ax goes to zero, and rearranging gives Equation 12.3.11 may be rewritten as
2k'[L2 (dC
(d
(12.3.11) c I 7 I nr2 (12.3.15)
\ax/x=o
12.3 Diffusion and Reaction in Porous Catalysts 441
From equation 12.3.13, the concentration The approximation is quite good for many
gradient at the pore mouth is found to be engineering purposes, particularly when one
x
dx j L Q>
cosh hT
< sinh hT d )\\( hA
hT Co sinh hr
(12.3.16)
L cosh hT
According to this model, the reaction rate that recognizes the disparity between the physical
would be observed within the catalyst pore is reality represented by a real porous catalyst and
given by a combination of equations 12.3.15 the assumptions implied by the model.
and 12.3.16. The effectiveness factor for a single pore is
identical with that for the particle as a whole.
rpore = ^ ^ ^ t a n h / , r (12.3.17)
Thus the reaction rate per unit mass of catalyst
can be written as
If there were no limitations placed on the
reaction rate by intraparticle diffusion (i.e., if r
mass — (12.3.22)
the reactant concentration were C o throughout
the pore), the reaction rate would be given by
r ideal = 2n7Lk';C0 (12.3.18)
From the definition of the effectiveness factor
and equations 12.3.17 and 12.3.18,
nr — = — tanh hT
rj-=
1
ideal 2nrLk'[C0
hT tanh h7 (12.3.19)
In the laboratory, the measured rate constant they must have a means of evaluating the effec-
for the first-order reaction would have to equal tiveness factor for the catalyst particles they
the product r\k'{ if this constant were expressed plan to use in their reactor. There are two basic
per unit area of catalyst. This relationship then routes to the evaluation of rj, and each is dis-
gives us an alternative interpretation of the cussed in turn.
effectiveness factor in terms of our model The first method of determining rj is based on
k" equation 12.3.23. It requires experimental data
1 measured (12.3.23) on the rate constants (or the reaction rate at
n= k"
similar external concentrations) as a function of
'M intrinsic
It is the ratio of the measured rate constant to the gross particle size of the catalyst. Experi-
the intrinsic chemical rate constant. ments are carried out using smaller and smaller
The Thiele modulus may be written in terms particles obtained by grinding or crushing the
of measurable quantities using equations 12.3.2 large particles that are intended for use in the
and 12.3.3: commercial scale reactor. When a size regime
IW; where the rate constant is independent of
hT = L
particle size is reached, we are measuring the
true or intrinsic rate constant, and this parameter
(12.3.24) gives us the denominator of equation 12.3.23.
The numerator is, of course, the rate constant
For hT > 2 the observed reaction rate can be measured for the particle size one intends to use
approximated by combining equations 12.3.21, in the reactor. This method requires rate mea-
12.3.22, and 12.3.24: surements on at least two different size catalysts,
and preferably more.
A second approach to determiningy\ requires
rate measurements on only a single size of
catalyst but it requires that we know Vp, Sx, Vg,
(12.3.25) Sg and that we know or can estimate Q)c. From
a single rate experiment, we can obtain k'[ or its
There are two points worth noting relative to equivalent k'[Sg. Then there are alternative
equation 12.3.25. First, the ratio of the external means of solving equations 12.3.11, 12.3.20, and
surface area Sx to the gross geometric volume of 12.3.23 for rj in terms of the input parameters.
a catalyst particle is inversely proportional to a A simple trial and error technique consists of
characteristic dimension of the particle. For the following steps.
geometrically similar particles, this fact implies
that the observed rate per unit mass of catalyst 1. Assume a value for rj.
• will be inversely proportional to the particle 2. Calculate hT from equation 12.3.20 or read
size. Second, the observed rate will be propor- the value of this parameter from the curve in
tional to the square root of the true rate constant. Figure 12,2.
Consequently, over the temperature range where 3. Calculate the intrinsic rate constant from the
intraparticle mass transfer strongly influences input parameters, using equation 12.3.24.
the reaction rate, the apparent activation energy 4. Compare the measured rate constant k![ with
for the reaction will be approximately one half the product of the assumed Y\ and the intrinsic
the true activation energy for the surface reaction. rate constant determined in step 3. If the two
If reactor designers desire to make use of the are equal, the assumption in step 1 was
analysis developed thus far in this subsection, correct; if they are unequal, one assumes a new
12.3 Diffusion and Reaction in Porous Catalysts 443
value of r\ and iterates until consistent results relatively clean reaction that can be used as a
are obtained. measure of the relative activities of different
cracking catalysts. Some of the data on this re-
A variation on the second approach involves
action have been obtained using integral flow
a somewhat different manner of combining the
reactors. For this type of reactor, the variation
equations used in the trial and error procedure.
of conversion with reactor space time is rela-
From equation 12.3.11,
tively insensitive to variations in the form of the
(12.3.26) reaction rate expression at degrees of conversion
that are far from equilibrium. Consequently,
data on these systems are often reported in
and, from equations 12.3.23 and 12.3.20,
terms of a simple first-order reaction rate con-
i,, ^ 1 measured stant, assuming that the volume change accom-
tanh hT panying the reaction is negligible. An expression
that is often used to determine this constant is
Combining equations 12.3.26 and 12.3.27 gives
W dfA
K measured = M t a n h hT) (12.3.28) " Jo f - r . J ~ Jo0 kC, 0 (l - fA
-«n(l -/A
which can be written in terms of experimentally kC AO
measurable variables using equations 12.3.2 and
12.3.3: where W is the weight of catalyst employed and
the reaction rate constant is expressed per unit
= hT(tHTih hT) (12-3.29) weight of catalyst. (The basic equation was de-
rived earlier as equation 8.2.12.) If the data of
This equation can be solved for hT, and equation Corrigan and his co-workers (37) are expressed
12.3.20 may then be used to determine the in these terms, at 510 °C the apparent rate con-
effectiveness factor. stant is approximately 0.716 cm3/sec-g catalyst)
The equations for effectiveness factors that we as determined from Figure 6 of the reference
have developed in this subsection are strictly cited. If the catalyst employed in this study has
applicable only to reactions that are first-order the properties enumerated below, determine the
in the fluid phase concentration of a reactant effectiveness factor for the catalyst. The value of
whose stoichiometric coefficient is unity. They the combined diffusivity from Illustration 12.1
further require that no change in the number of may be used in your analysis.
moles take place on reaction and that the pellet
Data on Catalyst Properties
be isothermal. The following illustration indi- (Silica-alumina)
cates how this idealized cylindrical pore model
is used to obtain catalyst effectiveness factors. Equivalent diameter 0.43 cm
Particle density 1.14 g/cm3
ILLUSTRATION 12.2 DETERMINATION OF Specific surface area 342 m2/g
CATALYST EFFECTIVENESS FACTOR FOR Porosity 0.51
THE CUMENE CRACKING REACTION Void volume per gram 0.447 cm3/g
FROM MEASUREMENT OF AN APPARENT
RATE CONSTANT Solution
The cumene cracking reaction has been studied Since we are given the apparent rate constant
by a number of investigators because it is a rather than the true rate constant, a trial and
444 Reactor Design for Heterogeneous Catalytic Reactions
error solution will be required. Either of the failure to allow for the change in the number of
approaches described above may be used, but moles on reaction, and the possibility that the
we will employ equation 12.3.29. Since our ap- pellets cannot be maintained in an isothermal
parent rate constant is expressed per unit weight condition for this endothermic reaction (see Sec-
of catalyst rather than per unit surface area, the tion 12.3.2). The discrepancy indicates the ne-
first step involves multiplying both sides of this cessity for good experimental data as input to
equation by 5 reactor design calculations for heterogeneous
catalytic reactions. Model predictions will not
K measured^ = M ^ n l l hT reflect the real world if the assumptions on which
they are based are in error. For best results, one
where the left side is equal to 0.716 cm3/(sec-g should use experimental data taken under condi-
catalyst). The equivalent radius of the catalyst tions approximating the proposed industrial
particles is 0.215 cm. Hence application.
Vp = I = |7r(0.215)3 = 0.0416 cm 3
12.3.1.2 The Effectiveness Factor for a Straight
and Cylindrical Pore: Second- and Zero-Order
Reactions. This section indicates the predic-
Sx = 4nR2 = 4TT(0.215) 2 = 0.581 cm2
tions of the straight cylindrical pore model for
Thus isothermal reactions that are zero- and second-
(0.447)(0.581)2(6.19 x 10 - 3 V
0.716 = /iT(tanh hT)
_ (0.0416)2
reaction (h2) is defined as The reaction rate within the pore is equal to
the rate of diffusion into the pore as indicated
h7 = I (12.3.32) by equation 12.3.15. Thus
d(C/C0)
h22(C/C0)2
d(x/L)
(12.3.34) This relation is an exact consequence of the
differential equation 12.3.30, but it suffers from
Integration gives
the disadvantage that it contains an unknown
rd(c/co)j parameter, C L , the reactant concentration at the
L d(x/L) J hl(C/C0)3 end of the pore. However, for even moderately
2 3 l fast reactions we can say that CL is significantly
less than C o and, because of the nature of the
(12.3.35)
function under the radical sign, it is a very good
where Cx is a constant of integration. approximation to say that
At x/L = 1 the first derivative must be zero,
1 0.8165
and C/CQ must be CJC0. Hence r\ ^ for h? > 2.5
d(C/C0)
d(x/L) •m-m
1
L\Co/ \ c o /(12.3.36) Curve B of Figure 12.3 [adopted from Wheeler
(12.3.42)
1.0
i i form of the appropriate differential equation is
d2(C/C0) 2L2K
2 = hi (12.3.43)
0.8 —
— \ \ \ c d(x/L) 13)CC o
| -
aeh-
\\ V \ V >
— Zero order
subject to the boundary conditions of equations
12.3.8 and 12.3.9. The parameter /c'o' is the zero-
order rate constant based on unit surface area
of the catalyst and h0 is the Thiele modulus for
I 0.4 the zero-order reaction.
— First o r d e r - ^ S ^
The solution to equation 12.3.43 can be
I Second order —^ written as
0.2 —
c
Single pore model ^ - \ \ ^ \ I (12-3.44)
I l i
1 2 3 4 5 The reactant concentration C will be greater
Thiele modulus, ht than zero throughout the entire length of the
pore provided that h0 < yfl. In this case the
Figure 12.3 effectiveness factor will be unity because the
Plots of effectiveness factors versus corresponding reaction rate is independent of concentration.
Thiele moduli for zero-, first-, and second-order For values of h0 > >/2, equation 12.3.44 would
kinetics based on straight cylindrical pore model.
call for negative values of the reactant concen-
For large h~ values of r\ are as follows:
tration at large values of x/L, a situation that
l2k'0'L2 is clearly impossible. Hence the boundary con-
Zero-order r\ =
ditions on equation 12.3.43 must be changed
so that both the reactant concentration and
First-order r\ = — its gradient become zero at a point in the pore
h that we label with a coordinate xc. In this
/2/3 situation, the concentration profile becomes
Second-order t) =
C = 0 for x (12.3.45)
[From A. Wheeler, Adv. Catalysis, 3 (249), 1951.
Copyright (c) 1951. Used with permission of Aca-
demic Press.] Co L^l 2 \Lj2,
Notice that in the region of fast chemical re- (12.3.46)
action, the effectiveness factor becomes inversely
where xc can be determined from the point at
proportional to the modulus h2. Since h2 is
proportional to the square root of the external which dC/dx = 0; that is,
surface concentration, these two fundamental
relations require that for second-order kinetics, for h0 (12.3.47)
the fraction of the catalyst surface that is effective
will increase as one moves downstream in an Thus, for h0 > y/2, the effectiveness factor is
isothermal packed bed reactor. given by
As Figure 12.3 indicates, it is also possible to
obtain an analytical solution in terms of the 2nrLk0 L n0
straight cylindrical pore model for the case of
a zero-order reaction. Here the dimensionless This function is plotted as curve C in Figure 12.3.
12.3 Diffusion and Reaction in Porous Catalysts 447
The differential equation governing simulta- new variable z defined such that C = z/r. This
neous chemical reaction and diffusion then substitution converts equation 12.3.57 to
becomes
(12.3.58)
d
*.2o? dC
\ _ k';PpSgCr2 = 0 (12.3.52) dr2
This equation is of the same form as equation
If the effective diffusivity is assumed to be a 12.3.10 and the solution can be expressed in
constant, equation 12.3.52 can be written as terms of exponential functions as
>sr/R
z = Cr = (12.3.59)
dr2 r dr ®pff
where the constants of integration Cx and C 2
The boundary conditions for this differential can be evaluated from the boundary conditions.
equation require that: At r = 0, the fact that dC/dr = 0 implies that
1. At r = R = C0 (12.3.54) C, = -C2 (12.3.60)
Equation 12.3.59 can thus be rewritten in
2. At r = 0 (12.3.55) terms of hyperbolic functions as
recognizing that the overall rate of reaction where we have used the defining equation for
within the spherical pellet must be equal to the cf)s (12.3.56). For large values of 0 S , the hyperbolic
rate of mass transfer into the pellet at steady tangent approaches unity and the effectiveness
state. factor approaches 3/0 s . Figure 12.6 is a plot of
Rate = 4nR2@eH (~ ) (12.3.63) the function described by equation 12.3.68. It
\drJr = R is appropriate for an isothermal first-order
where we have eliminated the negative sign on irreversible reaction in a sphere. At low values
the derivative, because we are interested in the of </>s, the effectiveness factor approaches unity.
flux into the pellet. Illustration 12.3 indicates the use of the
The concentration gradient at the exterior effective diffusivity approach for estimating
surface can be obtained by differentiating catalyst effectiveness factors when this param-
equation 12.3.62 with respect to r eter is determined experimentally or may be
estimated.
cosh (t>s -
dC \ K
(12.3.64)
sinh(0s) sinh(</>s)
1.0
\ or
fcltruA = 4.081
\
\ The measured value of k'{Sg is 0.716 cm3/(sec-#
, 0.8 catalyst) and the ratio of this value to k'[ true S 3
\
\
should be equal to our assumed value for the
\ effectiveness factor, if our assumption was cor-
rect. The actual ratio is 0.175, which is at
0.6 \
\ variance with the assumed value. Hence we pick
\ a new value of rj and repeat the procedure until
agreement is obtained. This iterative approach
\
\
produces an effectiveness factor of 0.238, which
4
I °- \ s
\
corresponds to a <j)s value of 11.5. This result
differs from the experimental value (0.17) and
that calculated by the cylindrical pore model
0.2
(0.61). In the above calculations, an experimental
value of ®eff was not available and this cir-
cumstance is largely responsible for the dis-
crepancy. If the combined diffusivity determined
0.0 in Illustration 12.1 is converted to an effective
15 18
diffusivity using equation 12.2.9, the value used
above corresponds to a tortuosity factor of 2.6.
If we had employed Q)c from Illustration 12.1
Figure 12.6 and a tortuosity factor of unity to calculate
Effectiveness factor plot for spherical catalyst ^ e f f , we would have determined that rj = 0.65,
particles based on effective diffusivities (first-order which is consistent with the value obtained from
reaction).
the straight cylindrical pore model in Illustra-
tion 12.2.
Assume that the effective diffusivity for cumene
in these particles is equal to 1.2 x 10~ 3 cm2/sec,
which is a typical value for silica-alumina TCC
123.14 The Effectiveness Factor Analysis in
beads (39).
Terms of Effective Diffusivities: Extension to
Solution
Reactions Other than First-Order and Various
Catalyst Geometries. The analysis developed
The trial and error procedure outlined in in Section 12.3.1.3 may be extended in relatively
Section 12.3.1.1 is quite general if appropriate simple straightforward fashion to other integer-
modification is made for the use of 0S values order rate expressions and to other catalyst
and effective diffusivities. Thus we will start by geometries such as flat plates and cylinders.
assuming that Y\ = 0.20. From Figure 12.6, we Some of the key results from such extensions
find that 4>s = 13.4. are treated briefly below.
Substitution of numerical values into the For a zero-order chemical reaction, an analy-
defining equation for 0S (12.3.56) gives sis similar to that employed in Section 12.3.1.2
indicates that
1.2 x 10 - 3 r\ = 1 for <£so < (12.3.69)
12.3 Diffusion and Reaction in Porous Catalysts 451
the case of gases contacting the inside of a diffusion and chemical reaction in porous
porous annulus. Refer to Satterfield (41) and catalysts of different geometries. For discussion
Aris (35) for additional discussion along these of the asymptotic solution approach to catalyst
lines and the pertinent references. However, we effectiveness factors, see the texts of Aris (36,
should point out that Luss and Amundson (42) 43) and Petersen (44).
have demonstrated that for simple first-order
isothermal reactions, the spherical particle has
12.3.1.5 Isothermal Effectiveness Factors—Mis-
the lowest effectiveness factor for all possible
cellaneous Considerations.
shapes having the same volume. This demon-
stration is largely due to the fact that the sphere 12.3.1.5.1 Implications of Nonequimolal Coun-
has the lowest possible external surface/volume ter-diffusion for Effectiveness Factors. When
ratio. One is tempted to conclude that catalyst there is a change in the number of gas phase
particles should be fabricated in shapes other species on reaction, there has to be at steady state
than spheres to minimize intraparticle diffu- a net molar flux either into or out of a porous
sional limitations on the reaction rate. However, catalyst. When bulk diffusion is responsible for
in practice, such other considerations as the all or even a reasonably significant fraction of
resultant pressure drop across a packed bed, the the total mass transfer, the pressure gradient
flow distribution through the bed, and the ease through the pellet will usually be negligible, but
of manufacture weigh heavily in determining the the increase in the number of moles on reaction
particular geometric shape in which the catalyst makes it more difficult for reactants to diffuse
will be employed. into the catalyst, thereby decreasing the effective-
For spherical pellets, for semi-infinite slabs, ness factor. A decrease in the number of moles
and for infinite cylinders at sufficiently large facilitates diffusion of reactants into the pellet
values of the Thiele modulus (/>, the effectiveness relative to the equimolal counterdiffusion case,
factor becomes inversely proportional to the thereby increasing the catalyst effectiveness
modulus with the constant of proportionality factor. While we have previously neglected
depending on the gross catalyst geometry and consideration of these phenomena, we now wish
the order of the reaction. Under these conditions, to briefly consider the influence of changes in
the reactant concentrations rapidly approach the number of gas phase species during reaction
zero as one penetrates into the catalyst, and it on catalyst effectiveness factors. Treatments of
is only the outer periphery of the catalyst that is this subject date back to the classic paper of
effective in promoting reaction. This region is Thiele (20), but the most comprehensive general
very thin when 0 is large and, in many respects, treatment is that of Weekman and Goring (45),
it may be regarded as a semi-infinite flat plate who analyzed isothermal zero-, first-, and
with a thickness Vp/Sx, irrespective of the true second-order reactions in porous spherical
catalyst geometry. This limiting condition gives catalysts at constant total pressure. They de-
rise to the so-called "asymptotic forms" of the scribed the response of the effectiveness factor
relation between the effectiveness factor and the to changes in the number of moles on reaction
Thiele modulus, in which the inverse propor- in terms of a volume change modulus 0, and
tionality noted above is observed. This approach the conventional Thiele modulus <fis. The volume
is extremely appealing from the viewpoint of change modulus is defined as
the mathematician, because it emphasizes the
0 = (£yt)Y (12.3.75)
mathematical similarity of the differential
equations and their corresponding boundary where the sum of the stoichiometric coefficients
conditions when one analyzes simultaneous is the increase in the number of moles on re-
12.3 Diffusion and Reaction in Porous Catalysts 453
action and Y is the mole fraction reactant at mole fraction at the external surface of the
the external surface of the catalyst. Their results catalyst, one expects the influence of 9 to be
were summarized in the form of plots of the most significant near the inlet to a plug flow
ratio of the effectiveness factor for the volume reactor vis a vis its effects at other points along
change case (nf) to that for no volume change the reactor length. Moreover, the effect should
(rj) versus the volume change modulus 9 at be small in cases where the reactants are highly
various values of the Thiele modulus <ps. diluted by the presence of inerts. This approach
Figure 12.8 is a reproduction of their curves for has also been extended to nonisothermal systems
first-order reactions. The more significant the (46).
departures of 9 and (/>s from zero, the greater
the departure of the ratio (rj'/rj) from unity. The 12.3.1.5.2 Implications of the Effectiveness Fac-
departure of (rj'/rj) from unity also becomes more tor Concept for Kinetic Parameters Measured
significant as the order of the reaction increases, in the Laboratory. It is useful at this point to
but the effect is relatively small compared to discuss the effects of intraparticle diffusion on
the effects of variations in 9 or variations in <fis the kinetic parameters that are observed ex-
at low values of 4>s. Since the volume change perimentally. Unless we are aware that intra-
modulus is directly proportional to the reactant particle diffusion may obscure or disguise the
1.5 1— L^10
1.4
1.3 \\
£0
\
\ \
v>V
1.2
1.1
1.0
— ——
— 0
0.9
~o~
i —• —
— I—
— — .
.
—
1
—
.
-
—
-
= = » — m
• -
-0.5
-1.0
0.8 ' —• ^ .
'
— —
- " . -2.0
0.7 §5 ^
=3.0
^6.0
0.6-
0.5
0.4
-1 1
Volume change modulus,
Figure 12.8
Effectiveness factor ratios for first-order kinetics on spherical catalyst pellets.
true intrinsic chemical kinetics, we may draw raised to the nth power, it can be said that
incorrect conclusions regarding the order and
temperature dependence of the intrinsic chemi- rr -Cn cr ° — r<" +] (12.3.78)
cal reaction. If one is interested in the funda-
mental nature of the chemical reaction, it is Thus a zero-order reaction appears to be 1/2
important to employ experimental conditions order and a second-order reaction appears to
that eliminate the influence of both external be 3/2 order when dealing with a fast reaction
and intraparticle mass transfer processes on taking place in porous catalyst pellets. First-
the observed rate. From the viewpoint of the order reactions do not appear to undergo a shift
chemical engineer engaged in the practice of in reaction order in going from high to low
reactor design, it is even more important to effectiveness factors. These statements presume
recognize that in dealing with pelletized hetero- that the combined diffusivity lies in the Knudsen
geneous catalysts, the apparent kinetics may range, so that this parameter is pressure in-
differ appreciably from the true intrinsic kinetics. dependent.
If we consider a reaction with intrinsic kinetics If the dominant mode of transport within the
of simple nth order form that takes place within catalyst pores is ordinary molecular diffusion,
the pores of a catalyst pellet, the observed rate the analysis becomes somewhat more complex.
of reaction per unit mass of catalyst may be The ordinary molecular diffusivity is inversely
written as proportional to the pressure so that in this case
(12.3.76) h or (f""M 1 / 2 • P 1 / 2 (\11 79^1
where C o is the reactant concentration in the If the reactant is by far the dominant species,
gas phase. If we adopt the straight cylindrical equation 12.3.79 can be written as
pore model and write a material balance over
a differential element of pore length, we find (C0RT)1/2 or (12.3.80)
that for nth-order kinetics, the analog of
equations 12.3.10, 12.3.31, and 12.3.43 becomes and, in the limit of low effectiveness factors, the
rate will be of the form
d2(C/C0) 2/c;T2 C
cn
d(x/L)2 (12.3.81)
°
so that here one would again observe discre-
KCoJ pancies between the observed and true orders
(12.3.77) of the reaction.
In a similar fashion, it is easily shown that
where the square of the generalized Thiele the apparent activation energy of the reaction
modulus (the quantity in brackets) is propor- may differ appreciably from the intrinsic activa-
tional to the external concentration of reactant tion energy of the chemical reaction. The ap-
raised to the power (n — 1). parent rate constant is equal to the product of
In the limit of low effectiveness factors where n the effectiveness factor and the true rate constant
becomes inversely proportional to the Thiele and, in the limit of low effectiveness factors, it
modulus, the apparent order of the reaction may can be said that
differ from the true order. In this case, since the
h
rate is proportional to the product of the effec- k or true
(12.3.82)
tiveness factor and the external concentration ^apparent ^~ i
12.3 Diffusion and Reaction in Porous Catalysts 455
In terms of the generalized Thiele modulus of higher temperatures one is much more likely
equation 12.3.77, the last equation becomes to observe such limitations. In the regime where
catalyst effectiveness factors become small one
k cr will observe an activation energy that is given
^apparent "-
by equation 12.3.85. A typical Arrhenius plot
for the range of temperatures where this change
takes place will resemble the one shown in
where we have retained only the temperature
Figure 12.9.
dependent terms from the Thiele modulus.
If the combined diffusivity is written as
®c = Ae~Ed^ion/RT (12.3.84) T
At high temperatures the slope
(where A is a temperature independent quantity), will be approximately 1/2 the _
slope observed at low
the Arrhenius relation can be used with equa- temperature (77 « 1)
tions 12.3.83 and 12.3.84 to show that
diffusion = — (12.3.86)
The high temperature activation
If Knudsen diffusion dominates, — energy is approximately half that
observed at low temperatures
RT
^diffusion = ~y (12.3.87) I I I
Reciprocal temperature, 1/T
Under normal circumstances the true activation
Figure 12.9
energy term in equation 12.3.85 will far exceed
Schematic representation of shift in activation
the diffusional activation term calculated from energy when intraparticle mass transfer effects
either equation 12.3.86 or equation 12.3.87 and, become significant.
to a good approximation, it may be said that
in the limit of low effectiveness factors
12.3.1.53 Effectiveness Factors for Hougen-
-^apparent ^ ~ (IZJ.OOl Watson Rate Expressions. The discussion thus
far and the vast majority of the literature dealing
At low temperatures diffusion will be rapid with effectiveness factors for porous catalysts
compared to chemical reaction and diffusional are based on the assumption of an integer-
limitations on the reaction rate will not be power reaction rate expression (i.e., zero-, first-,
observed. In this temperature regime, one will or second-order kinetics). In Chapter 6, how-
observe the intrinsic activation energy of the ever, we stressed the fact that heterogeneous
reaction. However, since chemical reaction rates catalytic reactions are more often characterized
increase much more rapidly with increasing by more complex rate expressions of the
temperature than do diffusional processes, at Hougen-Watson type. Over a narrow range of
456 Reactor Design for Heterogeneous Catalytic Reactions
may exist within the particle itself or within the The heat transfer path through the solid phase
boundary layer separating the particle from the offers considerable thermal resistance for many
bulk fluid. The presence of such gradients can porous materials, particularly if the pellet is
have significant consequences for reactor design formed by tableting of microporous particles.
calculations. This section treats the influence of Such pellets may be regarded as an assembly of
intraparticle temperature gradients in terms of particles that contact one another at only a
their influence on catalyst effectiveness factors. relatively small number of points that act as
This influence is most significant in cases where regions of high thermal resistance.
highly exothermic reactions are carried out on An approximate relationship that is some-
catalyst pellets that have low effective thermal times useful in predicting effective thermal
conductivities. If catalyst effectiveness factors conductivities is the geometric average value
are defined in terms of the ratio of the observed discussed by Woodside and Messmer (53).
rate to a rate evaluated assuming that the reac-
tant concentration and the temperature at the (12.3.94)
pore mouths prevail throughout the pellet, it - ^
is possible to observe values of this parameter where kf and ks are the thermal conductivities
in excess of unity. When the catalyst interior is of the bulk fluid and solid, respectively, and sp
hotter than the peripheral regions, the increase is the porosity of the pellet.
in temperature as one moves inward may more In spite of the difficulties of predicting fceff
than offset the decline in reactant concentrations on an a priori basis, it is still possible to choose
so that one obtains faster local rates of reaction a value that is reasonably correct because the
in the interior than in the peripheral regions. possible range of values is roughly only 0.1
to 0.4 BTU/(hr-ft-°F) or 1.6 to 6.4 x 10~ 3
J/(sec-cm-°C). Typically, the thermal conduc-
12.3.2.1 Effective Thermal Conductivities of Po-
tivity of reactant gases at room temperature
rous Catalysts. The effective thermal conduc-
will range from 8 to 24 x 10" 5 J/(sec-cm-°C),
tivity of a porous catalyst plays a key role in
and these values are an order of magnitude less
determining whether or not appreciable tem-
than those of most porous catalysts under
perature gradients will exist within a given cata-
vacuum. Consequently, in this case the bulk of
lyst pellet. By the term "effective thermal
the energy transport will occur through the
conductivity", we imply that it is a parameter
solid phase. On the other hand, for liquids, the
characteristic of the porous solid structure that
thermal conductivity will be an order of magni-
is based on the gross geometric area of the
tude greater and, when the catalyst pores are
pellet perpendicular to the direction of heat
filled with liquid, both phases will make signifi-
transfer. For example, if one considers the
cant contributions to the heat transfer process.
radial heat flux in a spherical pellet one can
Butt (54) has developed a useful model for the
say that
thermal conductivity of porous catalysts that
dT has shown good agreement with experimental
qr = -4nr2kleff (12.3.93)
dr data.
where qr is the rate of heat transfer in the radial
direction and fceff is the effective thermal 12.3.2.2 Effectiveness Factors for Nonisother-
conductivity. mal Catalyst Pellets. Here we indicate how
The effective thermal conductivities of com- previous effectiveness factor analyses may be
mon commercial porous catalysts are quite low extended to situations where the pellet is not
and fall within a surprisingly narrow range. isothermal. Consider the case of a spherical
458 Reactor Design for Heterogeneous Catalytic Reactions
pellet within which a catalytic reaction is taking ment that the rate of mass transport equal the
place. If we examine an infinitesimally thin reaction rate.
spherical shell with internal radius r similar to The maximum temperature difference be-
that shown in Figure 12.4 and write a steady- tween the center of the pellet and the external
state energy balance over the interior core of the surface (Tc — T o ) occurs when the reactant
pellet, it is obvious that the heat flow outward by concentration vanishes at r = 0.
conduction across the sphere of radius r must
be equal to the energy released by reaction (T c - T 0 ) max = -
within the central core. The latter quantity is
just the product of the reaction rate and (minus)
the enthalpy change accompanying the reaction Evaluation of the maximum temperature dif-
(AH). Hence ference provides a useful criterion for deter-
mining if departures from isothermal behavior
are significant. Substitution of the following
-47rr2fceff — = -AHM (12.3.95) property values into the above relation leads
dr
to a temperature difference of 200 °C between
where & is the reaction rate in the interior core. the center of the pellet and the exterior surface.
Since the reaction rate within the core must
also be equal to the rate of transfer of reactant AH = - 80,000 J/mole
into the core by diffusion, the previous equation ®eff = 10" 1 cm2/sec
may also be written as
Co = 4 x 10" 5 moles/cm3 (1 atm)
dC Kti = 16 x 10~ 4 J/(cm-sec-°C) (alumina)
-4nr2kcn = ~AH
~dr ~dr
Obviously, if the reactant concentration does
(12.3.96) not go to zero at the center, the temperature
difference will be less but, in many cases, it will
Negative signs are required on the right sides
still be quite large. Large values of this tempera-
of the last two equations so that for exothermic
ture difference can lead to effectiveness factors
reactions, the temperature will be hotter in the
for exothermic reactions that are considerably
core than at the periphery. Integration of the
in excess of unity.
last equation between radius r and the gross
In Section 12.3.1.3 it was shown that for a
pellet radius R gives
first-order reaction of the type A -> B, a material
*eff(r - T o ) = Aif®eff(C - Co) (12.3.97) balance on the infinitesimally thin spherical shell
of Figure 12.4 lea^s to the following differential
where T o and C o are the temperature and reac- equation.
tant concentration at the external surface of the
pellet. Rearrangement gives d2C 2 dC
(12.3.100)
At steady state, an energy balance on this same used throughout this chapter. It is important to
volume element gives remember that here the effectiveness factor is
defined as the ratio of the actual rate to the one
Output — input = energy released by reaction
that would occur if all portions of the interior
(by conduction)
of the pellet were exposed to reactant at the same
A T
-47rr 2 £ eff — ) = -4nr2 Ar{k'{CSgPp)AH (12.3.102)
MY Jr + Ar
Rearranging and taking the limit as Ar -> 0 concentration and the same temperature as exist
gives at the outside surface of the pellet.
Regardless of the approach used to generate
d (^ dT\ M2 KjCS9pp AH Q3)
values of the effectiveness factor, this parameter
dr = r can be expressed in terms of three dimensionless
or groups.
d2T 2dT k';CSgppAH
(12.3.104) I A Thiele type modulus.
dr r dr /c eff
reactions on spherical catalyst pellets in terms in the nonisothermal shell of catalyst at its
of four figures in which rj is plotted versus 4>s. periphery.
Each figure represents a fixed Arrhenius number The shapes of the curves in Figure 12.10 that
and contains a series of curves for values of the correspond to highly exothermic reactions are
heat generation parameter ranging from —0.8 such that a single value of <ps may give rise to
to +0.8. Figure 12.10 is adapted from their as many as three values of rj at fixed values of
work and represents an Arrhenius number that y and /?. These three values correspond to three
corresponds to a set of variables that may different sets of circumstances in which the rate
typically be encountered in industrial practice of energy release by reaction equals the rate of
(i.e., E = 28,000 cal/g mole; T = 700 °K). Their heat removal. In many respects the situation is
original article also contains curves for Arrhe- similar to the problem of stability for a CSTR
nius numbers of 10, 30, and 40. treated in Section 10.6. The intermediate value
Inspection of Figure 12.10 indicates that for of r\ for given values of <ps, y and fi corresponds
exothermic reactions (/? > 0), the effectiveness to unstable conditions. The high and low values
factor can exceed unity by a considerable of r\ represent stable conditions and both can
amount. This situation occurs when circum- be realized in practice. Which condition will
stances are such that the increase in rate caused prevail in the real world depends on the direc-
by the increase in temperature as one moves tion from which the steady-state condition is
toward the center of the pellet more than offsets approached. The highest value of rj corresponds
the drop in reactant concentration accom- to a steep temperature profile in the pellet, and
panying the movement. The overall rate of physical transfer processes limit the overall re-
reaction is therefore greater than it would be action rate. The lowest value of rj will be close
if the same temperature and reactant concentra- to unity and, in this case, the temperature
tion prevailed throughout the catalyst pellet. gradients within the pellet will be quite small. In
While values of r\ that exceed unity imply the last instance the overall rate is controlled by
efficient catalyst utilization, there may be con- the rate of the chemical reaction on the catalyst
comitant disadvantages. The large temperature surface.
increase at the center of the pellet may hasten We would be remiss in our obligations if we
catalyst deactivation processes in the interior did not point out that the regions of multiple
core, or it may lead to adverse effects on catalyst solutions are seldom encountered in industrial
selectivity if the activation energies of reactions practice, because of the large values of fi and y
leading to undesired products are greater than required to enter this regime. The conditions
that of the reaction of interest. under which a unique steady state will occur
The numerical solutions indicate that under have been described in a number of publica-
conditions where r\ » 1, there is a large tem- tions, and the interested student should consult
perature gradient at the periphery of the pellet, the literature for additional details. It should
but the gradient flattens out considerably as one also be stressed that it is possible to obtain
approaches the center of the pellet. For fast effectiveness factors greatly exceeding unity at
exothermic reactions (large 0S, /? > 0), the relatively low values of the Thiele modulus. An
interior of the pellet will be at a relatively uni- analysis that presumed isothermal operation
form high temperature, but there will be a fairly would indicate that the effectiveness factor
sharp decrease in temperature as one nears the would be close to unity at the low moduli
exterior surface. In this regime rj becomes in- involved. Consequently, failure to allow for
versely proportional to cj)s, as in the isothermal temperature gradients within the catalyst pellet
case, and the bulk of the reaction takes place could lead to major errors.
500 1000
Figure 12.10
Effectiveness factor chart for first-order reaction in spherical pellets for
y = 20. [From P. B. Weisz and J. S. Hicks, Chemical Engineering Science,
17 (265), 1962. Copyright c 1962. Reprinted with permission of Pergamon
Press.]
461
462 Reactor Design for Heterogeneous Catalytic Reactions
Under isothermal conditions, we have seen pellets, made by compressing these particles in
that the apparent activation energy of the reac- a steel mold. They also measured the tempera-
tion is approximately one half the intrinsic ture difference between the center of the pellet
value when rj is sufficiently low. When r\ exceeds and the external surface.
unity, an opposite effect occurs (i.e., the apparent The reactor feed mixture was "prepared so as
activation energy will exceed the true activation to contain less than 17% ethylene (remainder
energy). hydrogen) so that the change in total moles
Consideration of Figure 12.10 indicates that within the catalyst pore structure would be
for endothermic reactions (/? < 0), nonisother- small. This reduced the variation in total pres-
mal conditions do not have as significant an sure and its effect on the reaction rate, so as to
effect on rj as they do for exothermic cases. For permit comparison of experiment results with
endothermic reactions, rj is always less than theoretical predictions [e.g., those of Weisz and
unity and the temperature within the catalyst Hicks (61)]. Since the numerical solutions to the
interior is less than that at the exterior surface. nonisothermal catalyst problem also presumed
Both concentration and temperature decrease first-order kinetics, they determined the Thiele
as one moves radially inward. A good approxi- modulus by forcing the observed rate to fit this
mation for the effectiveness factor can be ob- form even though they recognized that a
tained by considering only the temperature Hougen-Watson type rate expression would
gradient through the pellet and neglecting con- have been more appropriate. Hence their Thiele
centration gradients, even when they are appre- modulus was defined as
ciable. This approach has been described by
Maymo, Cunningham, and Smith (62). (A)
The following illustration indicates how ex-
perimental and calculated values of catalyst where ^ m was the reaction rate per unit weight
effectiveness factors may be determined for a of catalyst for very small catalyst particles, and
specific case. C o was the reactant concentration at the external
surface of the pellet.
Using this definition of the Thiele modulus,
ILLUSTRATION 12.4 EFFECTIVENESS
the reaction rate measurements for finely divided
FACTOR DETERMINATION FOR A
catalyst particles noted below, and the addi-
NONISOTHERMAL CATALYST P E L L E T -
tional property values cited below, determine
EXOTHERMIC REACTION
the effectiveness factor for 0.5 in. spherical
Cunningham et al (63) have studied the rate of catalyst pellets fabricated from these particles.
hydrogenation of ethylene at 1 atm on a copper- Comment on the reasons for the discrepancy
magnesium oxide catalyst. They used flow between the calculated value of rj and the ratio
reactors to study the reaction kinetics over both of the observed rate for 0.5 in. pellets to that for
finely divided catalyst particles and spherical fine particles.
12.3.3 The Influence of Catalyst Poisoning This situation is termed "pore-mouth" poisoning.
Processes on Catalyst Effectiveness Factors As poisoning proceeds the inactive shell thickens
and, under extreme conditions, the rate of the
In the design of commercial scale heterogeneous catalytic reaction may become limited by the
catalytic reactors, the activity of the catalyst will rate of diffusion past the poisoned pore mouths.
almost invariably change with time. We now The apparent activation energy of the reaction
wish to focus our attention on the implications under these extreme conditions will be typical of
of poisoning reactions for efficient use of cata- the temperature dependence of diffusion coeffi-
lyst surface areas. Since reactant molecules must cients. If the catalyst and reaction conditions in
interact with unpoisoned catalyst sites before question are characterized by a low effectiveness
reaction can occur, the poisoning process may factor, one may find that poisoning only a small
have two effects on the reaction rate one fraction of the surface gives rise to a dispropor-
observes. tionate drop in activity. In a sense one observes
1. It always decreases the total number of cata- a form of selective poisoning.
lytic sites or the fraction of the total surface The two limiting cases for the distribution of
area that has the capability of promoting deactivated catalyst sites are representative of
reaction. some of the situations that can be encountered
2. It may increase the average distance a re- in industrial practice. The formation of coke
actant molecule must diffuse through the pore deposits on some relatively inactive cracking
structure before undergoing reaction. catalysts would be expected to occur uniformly
throughout the catalyst pore structure. In other
The poisoning reaction must be viewed like situations the coke may deposit as a peripheral
any other chemical interaction between a gas shell that thickens with time on-stream. Poi-
phase reactant and the solid surface. The manner soning by trace constituents of the feed stream
in which the poisoned sites are distributed often falls in the pore-mouth category.
throughout the pore structure is a direct conse- By analyzing both of the limiting situations
quence of the relative values of the rate at which described above, Wheeler (64-65) was able to
the poison diffuses into the pore and the rate at show that the interaction of the poisoning pro-
which the poisoning reaction takes place. The cess with the influence of intraparticle diffusion
problem of determining this distribution is the on the rates of the primary and poisoning reac-
same as the more general problem we have been tions can lead to a variety of interesting relations
analyzing in the last several subsections: "What between observed catalytic activity and the frac-
do the concentration profiles of reactant and tion of surface poisoned. Each of the limiting
product species look like for a given catalyst cases is analyzed below in terms of the mathe-
particle?" If the poisoning reaction is slow rela- matical model set forth by Wheeler. In both cases
tive to diffusion of the poison, the poisonous the reaction is presumed to proceed isothermally
reactants will be distributed uniformly through- in a manner that is first-order in the gas phase
out the porous catalyst and we will have homo- concentration of reactant. For convenience, the
geneous poisoning of the catalyst. This situation straight cylindrical pore model of the catalyst is
obviously corresponds to that in which the effec- used, but the qualitative results may be re-
tiveness factor for the poisoning reaction is garded as essentially independent of the model
essentially unity. On the other hand if the poi- employed.
soning reaction is rapid relative to diffusion of
the poison, the other periphery of the catalyst 1233.1 Uniform Distribution of Activity Loss.
particle will be completely poisoned before the If the species giving rise to the poisoning reaction
interior core suffers significant activity losses. must make several collisions with the catalyst
12.3 Diffusion and Reaction in Porous Catalysts 465
x = 0 x = L
where hT is the Thiele modulus for the unpoi- In order to demonstrate the selective effect of
soned reaction. pore-mouth poisoning, it is instructive to con-
The unknown concentration Cc can be elimi- sider the two limiting cases of reaction conditions
nated between equations 12.3.117 and 12.3.120 corresponding to large and small values of the
to give Thiele modulus for the poisoned reaction. For
the case of active catalysts with small pores, the
arguments of all the hyperbolic tangent terms in
C = equation 12.3.124 will become unity and
2k
2k"
1 + tanh [(1 - a)/iT]aL j - ^
1
[for M l - a) > 2] (12.3.125)
ochT
1 + (xhT tanh [(1 - a)/zT]
This equation indicates that a small amount of
(12.3.121) poisoned surface can lead to a sharp decline in
apparent activity. For example, if only 10% of
The reaction rate within the pore may now be the catalyst surface has been deactivated in the
obtained by combining equations 12.3.120 and case where the Thiele modulus for the unpoi-
12.3.121. soned reaction is 40, !F = 0.200 so that the
468 Reactor Design for Heterogeneous Catalytic Reactions
apparent activity has dropped by some 80%. In penetrating. In instances where hT(\ — a) is
this situation, both the primary reaction and the large (say greater than 6) and a is appreciable
poisoning reaction try to take place on the same (say between 0.2 and 0.8), we find that the de-
surface near the pore mouth. As this region is activated portion of the pore represents by far
poisoned, the reactants are forced to diffuse the major portion of the depth to which reactants
further into the pore before encountering a cata- are capable of penetrating in appreciable con-
lytically active surface. This has a concomitant centrations. In this case the overall reaction rate
strong adverse effect on the observed reaction becomes limited by the rate of diffusion through
rate. Curve D in Figure 12.11 is representative the poisoned region, and the activation energy
of this type of behavior. will reflect the temperature dependence of the
Now consider the other extreme condition combined diffusivity, i.e., it will be a kilocalorie
where diffusion is rapid relative to chemical or so. Thus as the temperature increases, there
reaction [i.e., hT(l — a) is small]. In this situa- is a transition from the activation energy of the
tion the effectiveness factor will approach unity intrinsic chemical reaction to an activation
for both the poisoned and unpoisoned reactions, energy approaching zero.
and we must retain the hyperbolic tangent terms
in equation 12.3.124 to properly evaluate # \ 12.3.4 The Influence of Intraparticle Mass
Curve C in Figure 12.11 is calculated for a value Transfer Limitations on Catalyst Selectivity
of/?T = 5. It is apparent that in this instance the
The present section deals with the influence of
activity decline is not nearly as sharp at low
intraparticle mass transfer processes on catalyst
values of a as it was at the other extreme, but it is
selectivity following the basic pattern established
obviously more than a linear effect. The reason
by Wheeler (64, 65) in his classic papers.
for this result is that the regions of the catalyst
pore exposed to the highest reactant concentra-
12.3.4.1 Independent Parallel Reactions of Dif-
tions do not contribute proportionately to the
ferent Species on the Same Catalyst. One often
overall reaction rate because they have suffered
requires a catalyst that promotes the reactions
a disproportionate loss of activity when pore-
of one component of a feedstock but does not
mouth poisoning takes place.
promote the reactions of other constituents of
For situations where the reaction is very slow
the mixture. For example, one might desire to
relative to diffusion, the effectiveness factor for
dehydrogenate six-membered rings, but not five-
the poisoned catalyst will be unity, and the ap-
membered rings. This type of selectivity behav-
parent activation energy of the reaction will be
ior may be represented by mechanistic equations
the true activation energy for the intrinsic
of the form
chemical reaction. As the temperature increases,
however, the reaction rate increases much faster A k-X U + V (12.3.126)
than the diffusion rate and one may enter a
regime where hT(l — a) is larger than 2, so the X k-X W + Z (12.3.127)
apparent activation energy will drop to that
given by equation 12.3.85 (approximately half The corresponding rate equations for constant
the value for the intrinsic reaction). As the volume systems can be written as
temperature increases further, the Thiele mod-
ulus \_hT(l — a)] continues to increase with a _ ^ = -ktf^A) (12.3.128)
concomitant decrease in the effectiveness with
which the catalyst surface area is used and in the d(X)
= -k2rj2{X) (12.3.129)
depth to which the reactants are capable of dt
12.3 Diffusion and Reaction in Porous Catalysts 469
where the effective rate constants are k1r]1 and rate constants (e.g., when the larger molecule is
k2rj2, respectively. the more reactive), then the selectivity ratio will
The selectivity of the catalyst performance can decline as the effectiveness factors decrease.
be described by the ratio of equations 12.3.128 However, if the more reactive species also has a
and 12.3.129. higher effective diffusivity, it is possible to obtain
an enhanced selectivity for the porous catalyst.
d(A) =
(12.3.130) This situation may readily occur in dealing with
d(X) k2ri2(X) molecular sieve catalysts in which certain reac-
tants can be essentially excluded from the catalyt-
For an isothermal system separation of variables
ically active regions by virtue of their molecular
and integration gives
size or shape.
favor the lower-order reaction. For the case cited Figure 12.13 contains a plot of the yield of the
above, the yield of V relative to W will increase. intermediate V as a function of the fraction A
When the desired reaction is of lower order than reacted for a value of k1/k2 equal to 5. In this
the undesired reaction, catalyst properties that case, we see that the maximum yield of V based
lead to low effectiveness factors give a higher on the initial concentration of A is equal to
selectivity than those that lead to an effectiveness 66.9%.
factor of unity. If the desired product is formed When consecutive reactions take place within
by the higher-order reaction, we should mini- a porous catalyst, the concentrations of A and
mize the concentration gradient in the pellet V within the pellet will be significantly different
and thus should use catalysts with effectiveness from those prevailing at the external surface.
factors approaching unity. The intermediate V molecules formed within the
pore structure have a high probability of reacting
123A3 Consecutive Reactions Where an Inter- further before they can diffuse out of the pore.
mediate Is the Desired Product. Consecutive
reactions in which an intermediate species (V) is
the desired product are often represented as a
series of pseudo first-order reactions
Ak-Xv%W (12.3.136)
Good yields of the desired product can be ob-
tained only when k1 > k2. Among the indus-
trially significant reactions of this type are those
involving partial oxidation and multiple sub-
stitution. Reactions represented by equation
12.3.136 were discussed in detail in Section 9.2,
and that material is also pertinent here.
For a catalyst-reactant system in which the
effectiveness factors for the first and second
reactions both approach unity, the analysis pre-
sented in Sections 9.2 and 9.3 is appropriate. If
yields are based on the initial concentration of
reactant A and if no V or W is present in the Figure 12.13
original feed, equations 9.3.9 and 9.3.10 can be Yield of intermediate versus fraction conversion
rearranged to express the yield of V as a function for k"lk'z = 5. Upper curve: r\ = 1 ; lower curve:
of the fraction A reacted: x\ < 0.3.
(1 - - 1] for k2 * /M (12.3.137)
and
or
d\Cv/CAg) =
_
d(x/L)2 CA cosh
K cosh
CAg
(1 - x/L) J|) cosh [Ml -
1 +
CVg h \
- 1 cosh ( hT
1 cosh hT
*V /
(12.3.146)
472 Reactor Design for Heterogeneous Catalytic Reactions
The rate of production of V within a single pore must be equal to the net rate of diffusion out of the
pore.
(12.3.147)
dx
Evaluation of the derivative and substitution of this result into equation 12.3.147 gives
K
Vg_
Ag
hT tanh hT (12.3.148)
- 1
Kii
The relative rate at which the external gas phase concentrations change is given by the ratio of the rate
at which V is produced within a pore (12.3.148) to the rate at which A is consumed within a pore
(12.3.17):
C Vg
tanh
cA - 1
dC Vg (121149)
dCA
- 1
tanh hT w
This equation gives the differential yield of V for a porous catalyst at a point in a reactor. For equal
combined diffusivities and the case where hT approaches zero (no diffusional limitations on the reaction
rate), this equation^ reduces to equation 9.3.8, since the ratio of the hyperbolic tangent terms becomes
•s/kf2@A/k'i@v. As hT increases from about 0.3 to about 2.0, the selectivity of the catalyst falls off con-
tinuously. The selectivity remains essentially constant when both hyperbolic tangent terms approach
unity. This situation corresponds t6 low effectiveness factors and, in tnis case, equation 12.3.149
becomes
dCy
(12.3.150)
~~dCA
- 1 - 1
k'ii
12.3 Diffusion and Reaction in Porous Catalysts 473
or
dCVg
(12.3.151)
dC A(l
Simplification gives
dCVa
(12.3.152)
dC
Ag
This equation may be solved using an integrating factor approach to determine the absolute yield of
V as a function of the fraction A reacted in an isothermal reactor. For the case where no V or W
are present in the original gas stream, integration gives:
(12.3.153)
1 1 -
where CA0 is the concentration of reactant at the entrance to the catalytic reactor and CAg is the local
gas phase concentration. This equation gives the yield of V based on the initial reactant concentration.
As a function of the fraction conversion of A
k'2
[(1 - / - fA)
(12.3.154)
-11 -
The lower curve in Figure 12.13 is a plot of the yield of V versus the fraction conversion of species
A for @)v = 2A, k'{/kf2 = 5, and low effectiveness factors for the first reaction. In general, for the low
effectiveness factor regime and comparable combined diffusivities, the maximum yield of an inter-
mediate in a series of consecutive reactions will only be abouOhalf that achieved when the effectiveness
factor approaches unity (64). At effectiveness factors below 0.3, the observed Selectivity is nearly
independent of the Thiele modulus and the effectiveness factor. The major decline in selectivity takes
pla'ce as the effectiveness factor drops from unity to 0.3. If one is interested in the mtermediate produced
by series reactjans, and the reactions in question are being carried out on catalyst pellets characterized
by an effectiveness factor for the initial reaction that lies between 0.3 and 1, it is possible to increase the
yield of this intermediate by going to smaller catalyst pellets orvby altering the pore structure so as to
increase the effective diffusivity. However, if the effectiveness factor lies well below 0.3, it is necessary
to go to very large reductions in pellet size or a much more open pore structure to bring about a
significant improvement in selectivity.
Wheeler's classic analysis (64, 65) has also been extended to nonisothermal situations (67-70).
474 Reactor Design for Heterogeneous Catalytic Reactions
Generally, under either isothermal or noniso- this transport process and its implications for
thermal conditions, intrapartiole diffusional lim- chemical reactor design.
itations are undesirable because they reduce the In trying to analyze the flow fields within a
selectivity below that which can be achieved in heterogeneous catalytic reactor, we encounter
their absence. The exception to this generaliza- an extremely complex problem, regardless of
tion is a set of encjothermic reactions that take whether we consider a fixed bed, a trickle bed, a
place in nonisothermal pellets where the second fluidized bed, or any other commonly used in-
reaction has an activation energy that is greater dustrial configuration. We oan, in principle, set
than that of the first. up differential equations for the conservation of
The problem treated in Section 13.2 indicates' mass, momentum, and energy that take into
how the concepts discussed above can be applied account the chemical reaction at the fluid-solid
to a practical problem. interface. Appropriate equations are needed
both for processes taking place external to the
catalyst particle and for those occurring within
12.4 MASS TRANSFER BETWEEN THE
the particle. The sets of equations are coupled
BULK FLUID AND EXTERNAL SURFACES
through the conditions at the exterior surface of
OF SOLID CATALYSTS
the particle, and here one must match the local
When a solid acts as a catalyst for a reaction, fluxes of energy and chemical species as well as
reactant molecules are converted into product the local temperatures and compositions. Al-
molecules at the fluid-solid interface. To use the though these equations are readily written, no
catalyst efficiently, we must ensure that fresh completely general method of solution exists.
reactant molecules are supplied and product Sometimes it is possible to obtain analytical
molecules removed continuously. Otherwise, solutions by making appropriate assumptions
chemical equilibrium would be established in the so as to simplify the problem. While the as-
fluid adjacent to the surface, and the desired sumptions are in many cases inconsistent with
reaction would proceed no further. Ordinarily, physical reality as expressed by the situations
supply and removal of the species in question normally encountered in industrial practice, the
depend on two physical rate processes in series. predictions of conversion may well be within the
These processes involve mass transfer between accuracy of the kinetic information. Some of
the bulk fluid and the external surface of the these analytical approaches are described in the
catalyst and transport from the external surface text by Petersen (71) and interested students
to the internal surfaces of the solid. The concept should consult this source. There are several
of effectiveness factors developed in Section 12.3 intrinsic problems and/or limitations with such
permits one to average the reaction rate over the analyses, particularly when one attempts to em-
pore structure to obtain an expression for the ploy conventional film theory. We shall find it to
rate in terms of the reactant concentrations and be more convenient to utilize average mass trans-
temperatures prevailing at the exterior surface fer coefficients in our analyses and to make use
of the catalyst. In some instances, the external of the semiempirical correlations of these param-
surface concentrations do not differ appreciably eters that have been developed over the years.
from those prevailing in the bulk fluid. In other
cases, a significant concentration difference
12.4.1 External Mass Transfer in Packed
arises as a consequence of physical limitations on
the rate at which reactant molecules can be Bed Reactors
transported from the bulk fluid to the exterior The velocity patterns within a fixed bed reactor
surface of the catalyst particle. Here, we discuss reflect the interactions between fluid elements
12.4 Mass Transfer between the Bulk Fluid and External Surfaces of Solid Catalysts 475
flowing over different particles, variations in the In gas phase systems, it is convenient to define a
available cross section for flow, the intrinsic mass transfer coefficient based on a partial pres-
physical properties of the fluid, and the average sure driving force {kG) as
rate at which the fluid is supplied. The problems
J
of analyzing the various physical transport pro- *G=,O '" (12A2)
cesses in a fixed bed reactor are thus extremely i,ES
complex. The conventional engineering ap- where PiB and PiES are the partial pressures of
proach to analyzing sudh systems involves the species i in theiWk and at the surface, respec-
definition of average heat and mass transfer tively. If the gas behaves ideally, then
coefficients for the bed. Such coefficients are
presumed to apply to the entire external surface kc = kGRJ (12.4.3)
of a given catalyst particle, even though experi- where Rg is the gas constant.
mental studies have shown that this situation To explore further the usages of these and
does not correspond to physical reality. This other mass transfer coefficients, consult stan-
assumption is sometimes stated by saying th&t dard references on mass transfer (72, 73) or
the surface k uniformly accessible. It implies that transport phenomena (74).
the equations describing the transport processes Dimensional analysis of the variables charac-
can be treated as unidimensional. teristic of mass transfer under flow conditions
The errors that result from the use of aver- suggests that the following dimensionless groups
age transport coefficients are not particularly are appropriate for correlating mass transfer
serious. The correlations that are normally data.
employed to predict these parameters are them- DnG
selves determined from experimental data on Reynolds number = NRe = (12.4.4)
packed beds. Therefore, the applications of the
correlations and the data on which they are based
correspond to similar physical configurations. Schmidt number = NSc = — (12.4.5)
pQ)
Depending on the driving force we choose to
employ in our analysis, there are several defini- kj)n
Sherwood number = N*h = (12.4.6)
tions of mass transfer coefficients that may be
considered appropriate for use. If we consider an where
arbitrary interface between a fluid and the exter-
nal surface of a catalyst particle, we might choose Dp is the equivalent diameter of the catalyst
to define a mass transfer coefficient based on a particle
concentration driving force (kc) as G is the mass velocity based on the total (super-
ficial) cross-sectional area of the reactor
(12.4.1) ja is the fluid viscosity
i,ES
p is the fluid density
where 2) is the molecular diffusivity of the species being
Jt is the molar flux of species i to- transferred in the system of interest
ward the surface relative to the The Sherwood number is also known as the
molar average velocity Nusselt number for mass transfer. Notice that
CiB and CiES are the concentrations of species i the diameter of the catalyst pellet is used in the
in the bulk and at the surface, Reynolds and Sherwood numbers as the charac-
respectively. teristic length dimension of the system. For flow
476 Reactor Design for Heterogeneous Catalytic Reactions
through packed beds, the transition between Some of the more useful relations are summa-
laminar and turbulent flow occurs at a Reynolds rized in Table 12.1.
number of approximately 40. At high Reynolds numbers the correlations
The equivalent particle diameter appearing in for liquids and gases are quite similar, but axial
these dimensionless groups is the diameter of a mixing becomes increasingly significant in gases
sphere having the same external surface area as at low Reynolds numbers. The values ofjD pre-
the particle in question. Thus for a cylinder of dicted by equations 12.4.10 and 12.4.11 differ at
length Lc and radius rc, the equivalent particle most by 15% for 55 < Re < 1500, but for low
diameter is given by Reyno'kis numbers jD for liquids will be less than
jD for gases. These equations apply to packed
D V beds in which a single fluid fills the voids be-
- ^ I = 2nrcLc + 2nr? (12.4.7) tween particles. The problem of predicting ap-
propriate mass transfer coefficients in trickle bed
or reactors is much more complex, and it is beyond
Dp = y/2rcLc + 2r2 (12.4.8)
the scope of this text. For an introduction to the
The most convenient mathematical form for problems involved, consult Satterfield's mono-
correlating mass transfer data is in terms of the graph (80). For other catalyst geometries (e.g.,
well-known Chilton-Colburn (75, 76) jD factor. woven screens) or flow conditions (e.g., pulsatile
flow), jD factors are available and may be ob-
(12.4.9) tained from the current literature or standard
handbooks. In addition, it is possible to estimate
mass transfer coefficients using data on heat
The functional dependence of jD on Reynolds
transfer coefficients, since the jD factor is equal
number has been the subject of study by many
to an analogous factor (jH) defined for the corre-
investigators [e.g., Thodos and his co-workers
lation of heat transfer data (see Section 12.5).
(77, 78), and Wilson and Geankoplis (79)]. A
variety of equations have been proposed as
12.4.2 External Mass Transfer in
convenient representations of the experimental
Fluidized Bed Reactors
data. Many of these correlations also employ the
bed porosity (sB) as an additional correlating Since the catalysts employed in fluidized bed
parameter. This porosity is the ratio of the void reactors have characteristic dimensions in the
volume between pellets to the total bed volume. 10 to 300 micron range, the external surface
Table 12.1
Correlations for Mass Transfer Factors (jD) in Packed Beds
0.357
Gas Petrovic and Thodos (78) 3 < NRe < 2000 £
BJD ^r0.359 12.4.10
iV
Re
1.09
0.001^ < NRe < 55 8
BJD N2,3
12.4.12
12.4 Mass Transfer between the Bulk Fluid and External Surfaces of Solid Catalysts 477
Table 12.2
Correlations for Mass and Heat Transfer Factors in Fluidized Beds
area per unit weight of catalyst is significantly in fluidized beds, see the text of Kunii and
greater than that employed in fixed bed reactors. Levenspiel (84) and the review by Beek (85).
This fact ensures that even in relatively low
density fluidized beds, overall mass transfer rates
12.4.3 Implications of External Mass
will be high compared to those in fixed beds.
Transfer Processes for Reactor
The large extent of turbulent mixing within such
Design Calculations
systems also serves to enhance the mass transfer
coefficient, but it is the high external surface to The only instances in which external mass trans-
volume ratio that is most significant in ensuring fer processes can influence observed conversion
rapid mass transfer. rates are those in which the intrinsic rate of the
Table 12.2 summarizes the more useful cor- chemical reaction is so rapid that an appreciable
relations for jD and jH that have been proposed concentration gradient is established between
by various investigators. The indicated cor- the external surface of the catalyst and the bulk
relations are quite consistent with one another fluid. The rate at which mass transfer to the
in the regions where they overlap. Although the external catalyst surface takes place is greater
mass transfer coefficients determined from these than the rate of molecular diffusion for a given
correlations are not particularly large, they concentration or partial pressure driving force,
lead to high mass transfer rates when they are since turbulent mixing or eddy diffusion pro-
multiplied by the external surface area of the bed. cesses will supplement ordinary molecular diffu-
For more detailed treatments of mass transfer sion. Consequently, for porous catalysts one
478 Reactor Design for Heterogeneous Catalytic Rea
does not encounter external mass transfer limita- Under steady-state operating conditions, the
tions, except in those circumstances where intra- observed reaction rate must be exactly counter-
particle diffusional limitations are also present. balanced by the rate at which reactants are
In the presence of intraparticle mass transfer supplied to the exterior surface of the particle.
limitations, the rate per particle is expressed in On a unit mass basis, the latter rate can be
terms of the species concentrations prevailing written as
at the exterior of the catalyst. However, when
(12.4.21)
external mass transfer limitations are also
present, these concentrations will differ from where
those prevailing in the bulk. Since bulk con-
PAB and PAtES are the reactant concentrations
centrations are what one measures in the
in the bulk fluid and at the
laboratory, exterior surface concentrations must
external surface of the catalyst,
be eliminated to express the observed conversion
respectively,
rate in terms of measurable concentrations. In
the paragraphs that follow, the manner in kG is the mass transfer coefficient
which one eliminates surface concentrations in appropriate units
is indicated in some detail for a specific case. am is the external surface area per
Consider an irreversible reaction of the form unit mass of catalyst.
B (12.4.18) In this equation the entire exterior surface of
that takes place on a solid catalyst surface. For the catalyst is assumed to be uniformly acces-
the moment, we shall presume that the intrinsic sible. Because equimolar counterdiffusion takes
chemical reaction rate follows an arbitrary place for stoichiometry of the form of equation
kinetic expression with a functional form 12.4.18, there is no net molar transport normal
to the surface. Hence there is no convective
rm = O(species concentrations, temperature) transport contribution to equation 12.4.21. Let
(12.4.19) us now consider two limiting conditions for
steady-state operation. First, suppose that the
where the rate is expressed per unit mass of
intrinsic reaction as modified by intraparticle
catalyst. If we take intraparticle mass transfer
diffusion effects is extremely rapid. In this case
limitations into account, the previous relation
PAES will approach zero, and equation 12.4.21
may be written as
indicates that the observed rate per unit mass
rm = n ^'(species concentrations, temperature) of catalyst becomes
(12.4.20) rm = kGamPAB (12.4.22)
where the new function O' must be expressed in The reaction will then appear to follow first-
terms of the temperature and species concentra- order kinetics, regardless of the functional form
tions prevailing at the exterior surface of the of the intrinsic rate expression and of the effec-
catalyst pellet. For reactions that obey rate tiveness factor. This first-order dependence is
expressions that differ from simple first-order characteristic of reactions that are mass transfer
kinetics, rj itself depends on the species concen- limited. The term "diffusion controlled" is often
trations, and the apparent order of the reaction applied to reactions that occur under these
observed in the laboratory may differ from the conditions.
intrinsic order, as we have noted previously in Now consider the other extreme where the
Section 12.3.1.5.2. external mass transfer rate is sufficiently rapid
12.4 Mass Transfer between the Bulk Fluid and External Surfaces of Solid Catalysts 479
to ensure that PAtES approaches PAB. In this The exterior surface concentration is then
case, the concentration dependence of the reac- given by
tion is given by equation 12.4.20. The apparent
orders of the reaction will be given by the
A,ES
functional form of r\<&'. One may or may not kGam + rjk['rueSg
observe first-order kinetics. The apparent order
will depend on the interaction of the intrinsic (12.4.27)
surface reaction and intraparticle diffusion
1+ kaam
effects.
To illustrate the masking effects that arise
With respect to the bulk concentration, the
from intraparticle and external mass transfer
observed rate per unit weight of catalyst then
effects, consider a surface reaction whose intrin-
becomes
sic kinetics are second-order in species A. For
this rate expression, equation 12.4.20 can be
written as V —
2
rm = *lkintrinsicP A (12.4.23) 1+ kGam
In the limit of low effectiveness factors where
,B
r\ is proportional to (l//i), equation 12.3.78 indi- (12.4.28)
cates that the observed rate becomes
kGam
^observed 1 A (12.4.24)
when the dominant mode of transport is The apparent rate constant per unit mass that
Knudsen diffusion. one would observe in the laboratory would
therefore obey the following relation.
Equations 12.4.22 and 12.4.24 indicate that
the observed reaction order will differ from the 1 1 1
intrinsic reaction order in the presence of intra- (12.4.29)
particle and/or external mass transfer limita-
tions. To avoid drawing erroneous conclusions When kGam is very large compared to rjk[frueSg,
about intrinsic reaction kinetics, we must be the exterior surface concentration is not appre-
careful to either eliminate these limitations by ciably different from the bulk fluid concentra-
proper choice of experimental conditions or to tion, and the apparent rate constant is equal to
properly take them into account in our data rfk[frueSg, which reflects the interaction of the
analysis. intrinsic chemical reaction and intraparticle
Let us now consider how the external surface diffusion processes. When kGam is very small in
concentrations can be eliminated when our comparison with r]k['rueSg, the exterior surface
reaction follows simple irreversible first-order concentration will be significantly less than the
kinetics. In this instance equation 12.4.20 be- bulk fluid value, and the apparent rate constant
comes becomes equal to kGam.
rm = VKUAJPA.ES (12.4.25) The above discussion indicates an approach
At steady state, this rate is balanced by the mass that may be used in deriving an expression for
transfer rate (equation 12.4.21). the reaction rate in terms of the physical and
chemical parameters of the system. However,
for most practical catalyst systems, it will not
(12.4.26) be possible to arrive at closed-form expressions
480 Reactor Design for Heterogeneous Catalytic Reactions
for the reaction rate per unit mass of catalyst. O2 = 0.900 cm2/sec
Consequently, this approach is of extremely 2
SO2 = 0.602 cm /sec
limited utility for reactor design purposes. The
SO3 = 0.515 cm 2 /sec
most common approach to the analysis of 2
external mass transfer limitations in hetero- SO2 = 0.598 cm /sec
Thus
5.39 x 10~4®ilf
I 0.594
and
- yso2(Q) + ySo3 -
^so2-so2 SO2-SO3
or
rSO2
,
f/
so 2 -o 2
0.0652
1 -
- 0.564 cm2/sec
0.0652
0.1963 -
0.7385 0.06521
0.598 0.602 + 0.303
482 Reactor Design for Heterogeneous Catalytic Reactions
The Reynolds number in consistent units is convenient to write the last equation as
then
N _£A_'"~
DpG _ (O.389)(3.33 x 1O~2)
v
Re = 41
~T~ " 3.2 x 10" 4
From the ideal gas law, the catalyst surface. Hence one must establish
a concentration difference for nitrogen, so that
yp total
the backward diffusion of nitrogen exactly
Q = counterbalances the convective transport to the
RT
catalyst surface. We should also indicate that
Thus
V.ES = -(S2.06K731)
VI,ES 790/760
= -9.93 x 1 0 - 3 ( N S c ) 2 / 3 ( v ; + ^
Substitution of the previously calculated the mole fractions at the external surface sum to
Schmidt numbers and composition values, and qnly 0.9987. In large measure this discrepancy
an appropriate average value for yt permits one arises because we have used feed composition
to determine the difference in mole fractions. values in our calculations rather than local
In the present instance, we may use yiB values average bulk fluid compositions.
484 Reactor Design for Heterogeneous Catalytic Reactions
We would be remiss if we did not indicate much greater than the decrease encountered
that a significant temperature difference also when intraparticle diffusion processes limit the
exists between the bulk fluid and the external catalytic reaction rate.
surface. This AT has a far greater effect on the
observed rate than does the SO 2 concentration 12.5 HEAT TRANSFER BETWEEN THE
difference. Illustration 12.6 indicates how the BULK FLUID AND EXTERNAL SURFACES
temperature difference may be calculated. OF SOLID CATALYSTS
In view of the fact that our results are reason-
ably sensitive to the estimate of the bed poro- The energy effects that accompany chemical
sity used in the analysis, these results are not reactions can lead to significant temperature
bad. If one had employed a value of 0.3 or 0.5 differences between a bulk fluid and the external
rather than 0.4 for sB, jD would change signif- surface of a catalyst on which reaction is taking
icantly and this would have a major influence place. At steady state, the rate of energy release
on the calculated concentration (or mole frac- (or consumption) by reaction must equal the
tion) differences. Unfortunately, bed porosity rate at which heat is transferred to the fluid.
data were not noted in the article cited. In an The temperature gradient necessary to sustain
experimental program being conducted as an the heat transfer may be appreciable, even in
aspect of a reactor design, this parameter could situations where concentration differences be-
easily be determined. tween the bulk fluid and the external catalyst
surface are negligibly small. Consequently, these
temperature differences can act to obscure the
Yoshida, Ramaswami, and Hougen (88) have reaction kinetics in many more situations than
developed a nomograph that in essence is based can the mass transfer effects previously dis-
on the procedure employed in the last example. cussed. In laboratory studies aimed at a deter-
Although it employs a jD correlation that pre- mination of the rate expression for the intrinsic
dates those given in Table 12.1 and thus is less chemical reaction, external gradients in tem-
accurate, it eliminates many of the detailed cal- perature and concentration can be made
culations involved in determining when signifi- negligible by operating under the following
cant external concentration differences exist. For constraints.
all practical purposes, the results obtained by
1. Using a reactant stream that is diluted with
using the nomograph are indistinguishable from
inerts to reduce the reaction rate so that the
those obtained using the procedure employed in
energy evolved per unit volume is greatly
Illustration 12.5.
reduced below that encountered in the ab-
Before terminating the discussion of external
sence of inerts.
mass transfer limitations on catalytic reaction
2. Employing high mass velocities to minimize
rates, we should note that in the regime where
resistances to heat and mass transfer.
external mass transfer processes limit the reac-
tion rate, the apparent activation energy of the In such studies one may also eliminate intra-
reaction will be quite different from the intrinsic particle gradients of temperature and composi-
activation energy of the catalytic reaction. In the tion by using very fine catalyst particles or by
limit of complete external mass transfer control, confining the catalytic species to the exterior
the apparent activation energy of the reaction surface of a nonporous or impervious pellet.
becomes equal to that of the mass transfer coeffi- Unfortunately, the conditions that are optimum
cient, typically a kilocalorie or so per gram mole. for the elucidation of the intrinsic chemical
This decrease in activation energy is obviously kinetics are often inappropriate for use in
12.5 Heat Transfer between the Bulk Fluid and External Surfaces of Solid Catalysts 485
commercial scale reactors, and the design bers where both transfer coefficients are small,
engineer must be able to take into account both it is quite possible for substantial temperature
exterior and intraparticle gradients in concen- differences to exist between the bulk fluid and
tration and temperature. We discuss here the the external surface, even when the corres-
methods used to evaluate exterior temperature ponding concentration differences are small and
gradients in both fixed and fluidized bed re- when intraparticle concentration and tempera-
actors. ture gradients are negligible (rj ^ 1). The major
Heat transfer between a fluid and a catalyst resistance to heat transfer is the laminar film
particle occurs primarily through the same adjacent to the solid surface rather than the
combination of molecular and convective pro- pellet itself.
cesses that are responsible for mass transfer in Experimental data on heat transfer in fixed
such systems. At sufficiently high temperatures, and fluidized bed reactors are correlated in
one must also allow for radiation contributions terms of a j factor for heat transfer.
to the energy transfer processes within the
reactor. Radiation effects are not considered in (12.5..)
the correlations discussed below. They are not
significant at temperatures below 400 °C for where
packed bed reactors comprised of pellets with
[i is the fluid viscosity
characteristic dimensions of less than 1/4 in. (89).
At higher temperatures, they may be included Cp is the constant pressure heat capacity per
in an energy balance on a segment of the reactor unit mass of fluid
if appropriate view factors and other geometric K is the thermal conductivity of the fluid
considerations are specified. For present pur- h is the heat transfer coefficient between the
poses, however, their inclusion would greatly catalyst particle and the bulk fluid
complicate the analysis that follows, without
JVPr is the Prandtl number
shedding additional light on the fundamental
principles involved. The functional dependence of jH on Reynolds
The approach we use in analyzing the heat number has been the subject of study by many
exchange between a solid catalyst particle and investigators, and a variety of equations have
the surrounding fluid to determine if a significant been proposed for correlation of the available
temperature difference exists is in many respects data for fixed bed (77, 88) and fluidized bed re-
quite similar to that employed in the analysis actors (81-83). Boundary layer theory indicates
of mass transfer in Section 12.4. Both analyses that the Chilton-Colburn analogy, jH = j D ,
are based largely on the Chilton-Colburn represents an asymptotic solution for forced con-
analogies for the corresponding transport pro- vection in three-dimensional flows in any ge-
cesses. However, our earlier comment that ometry, provided that the Peclet number is large
external mass transfer limitations are less (90). [The Peclet number for mass transfer is the
significant than intraparticle limitations cannot product of the Reynolds and Schmidt numbers
be generalized to the heat transfer process. At (DpG/2p\ while the Peclet number for heat
high Reynolds numbers, the heat transfer co- transfer is the product of the Reynolds and
efficients, like the mass transfer coefficients, will Prandtl numbers (DPGCP/K).~] The relation^ =
be large and the corresponding temperature jH agrees well with experimental data for many
and concentration differences between the bulk flow geometries. Although the literature correla-
fluid and the exterior surface of the catalyst tions frequently indicate that the ratio jH/jD is
will be small. However, at low Reynolds num- slightly greater than unity for flow through
486 Reactor Design for Heterogeneous Catalytic Reactions
packed beds (e.g., 79), the deviations are usually rate at which thermal energy is exchanged be-
small. They may be explained by the fact that tween the fluid and the solid catalyst.
measured rates of vaporization or sublimation
r m (-AH) - ham(TES - TB) (12.5.4)
cancel out of the expression for the ratio jH/jD
when one attempts to calculate these parameters Noting that rAm = vArm, equations 12.5.3 and
from conventional experimental data. The devia- 12.5.4 may be combined to give
tion from unity then merely reflects an average of
,B - CA,ES)am(-AH)
values determined from a large number of = haJTES - TB)
readings with a particular humidity chart (91).
For most purposes, the correlations for jD
presented in Tables 12.1 and 12.2 also suffice for (12.5.5)
estimating jH. There is, however, one additional
correlation for fluidized beds that is worth or
noting. On the basis of data for the fluidization kCtA(AH){CAtB -
TES ~TB-
of 20 to 40 mesh silica and alumina gel particles
in air at Reynolds number values (DpG/fi) ranging h [ v A - yA 2 . v.
from 9 to 55, Kettenring et al. (92) suggest that
(12.5.6)
If the expressions for h and kcA, derived from
hDn equations 12.5.1 and 12.4.9, are substituted into
= O.O135(NRe)1 (12.5.2)
the last equation, we find that
(CAtB -
ES (12.5.7)
v
Sc,
1 -
Notice, however, that none of these correlations The temperature difference is thus directly
are appropriate for use when radiation makes a proportional to the heat of reaction per mole
significant contribution to the heat transfer of diffusing species and to the difference in con-
process. centration between the bulk fluid and the ex-
At steady state, the rate at which reactants are terior surface of the solid. If we recognize that
supplied to the external surface of the catalyst jD ~ jH from the Chilton-Colburn relation, and
by mass transfer must be equal to the rate at that the ratio of Prandtl and Schmidt numbers
which they are consumed by the catalytic reac- is close to unity for many simple gas mixtures,
tion. Per unit mass of catalyst, the rate of dis- the previous relation may be approximated as
appearance of species A is then given by
\ (CAtB - CAtES)
TES -
r
A,m ~
(12531
(for gases) (12.5.8)
where the term in brackets is the drift factor. The Thus if the method developed in Section 12.4
rate of reaction multiplied by the energy release is used to evaluate the concentration difference,
per unit extent of reaction must also equal the this equation may be used to determine an
12.5 Heat Transfer between the Bulk Fluid and External Surfaces of Solid Catalysts 487
approximate value for the temperature differ- for heat transfer coefficients. Hence
ence. (In many cases the drift factor term in
brackets is essentially unity.) TEss ~TB =
I
^ (12.5.11)
E rlClm
In the limit where the external surface con-
centration becomes very small compared to the which may also be written in terms of the jH
bulk fluid concentration, we obtain the maxi- factor as
mum temperature difference.
*• FS •/ R — (12.5.12)
CpGjHNfr2/3am
(for gases) Yoshida, Ramaswami, and Hougen (88) have
1 also developed nomographs for the solution of
- yA I (vj/vA)
.7=1 (12.5.9) equation 12.5.12. Illustration 12.6 indicates an
analytical approach by which the temperature
In many cases of interest, the maximum tem- difference can be evaluated using the correla-
perature difference will be over 100 °C, and tions given in Tables 12.1 and 12.2.
values of this magnitude have been observed
experimentally for several catalytic reactions. ILLUSTRATION 12.6 HEAT TRANSFER IN
Combination of equations 12.5.8 and 12.5.9 A FIXED BED REACTOR—TEMPERATURE
gives GRADIENTS BETWEEN THE BULK FLUID
AND THE EXTERNAL CATALYST SURFACE
T ES T — I 1 _ cA max (12.5.10)
In Illustration 12.5, we considered the problem
cA of estimating the concentration differences that
Because of the large magnitude of AT max , it is exist between the bulk fluid and a catalyst used
quite possible that significant temperature dif- for the oxidation of sulfur dioxide. If the reported
ferences will exist, even when the ratio CA ES/CA B temperature is that of the bulk fluid, determine
is close to unity. For example, if the exterior the external surface temperature corresponding
surface concentration is 95% of the bulk fluid to the conditions cited. Additional useful data
level and the adiabatic temperature rise for the are:
Heat of reaction evaluated for the temperature range of interest = —96.95 kJ/mole
Heat capacity (Cp) of the reaction mixture = 1.07 J/g-°K
Thermal conductivity of the reaction mixture = 3.70 x 10~ 4 J/(cm-sec-°K)
tions, we normally try to operate under con- global rate that is useful for design purposes, one
ditions that minimize the effects of heat and mass must combine the intrinsic chemical reaction
transfer on conversion rates so as to facilitate rate,expression with expressions for heat and
the extraction of the rate expression for the mass transfer rates corresponding to industrial
intrinsic chemical reaction from the observed operating conditions. As a general rule, the
reaction rate (the global rate). We use very fine global rate reduces to the intrinsic expression
particles and high fluid velocities to minimize evaluated at bulk fluid properties when one of
these limitations so that differences in composi- the chemical steps at the catalyst surface is slow.
tion and temperature between the bulk fluid However, if economic considerations require
and the fluid within the pore structure of the the use of fluid velocities or catalyst pellet sizes
catalyst are insignificant. In the absence of such that differ from those employed in the laboratory,
differences, the global reaction rate corresponds the global rate may be quite different from the
to the intrinsic reaction rate, evaluated at the intrinsic rate at bulk fluid conditions. In such
bulk fluid composition and temperature. When cases the reaction may be mass or heat transfer
such differences are significant, the global rate limited, or controlled by pore diffusion pro-
may be appreciably higher or appreciably lower cesses. Figure 12.14 indicates some of the reac-
than the intrinsic rate corresponding to bulk tant concentration profiles that may result from
fluid properties. control by various steps in the sequence enu-
Laboratory reactors and industrial scale merated earlier. In commercial scale fixed bed
equipment are seldom operated under similar reactors, external resistances to mass transfer are
flow and heat transfer conditions. To obtain a usually small compared to the intraparticle
Catalyst pore
Figure 12.14
Schematic representation of reactant concentration profiles in various
global rate regimes. I: External mass transfer limits rate. II: Pore dif-
fusion limits rate. Ill: Both mass transfer effects are present. IV: Mass
transfer has no influence on rate.
490 Reactor Design for Heterogeneous Catalytic Reactions
For isothermal systems, it is occasionally pos- ployed much more widely in the chemical indus-
sible to eliminate the external surface concentra- try than any other basic reactor type. This
tions between equations 12.6.1 and 12.6.2 and section indicates in some detail the procedures
arrive at a global rate expression involving only involved in the design of a tubular fixed bed
bulk fluid compositions (e.g., equation 12.4.28 reactor. This task involves the use of global rate
was derived in this manrler). In general, however, information obtained through the techniques
closed form solutions cannot be achieved and developed earlier to predict the composition of
an iterative trial and error procedure must be the effluent from a reactor for a given set of
employed to determine the; global rate. One pos- design parameters.
sible approach is summarized below. The desigri 4>roblem can be approached at
1. Use the j factor correlations of Section 12.4 to various levels of sophistication using different
determine appropriate heat and mass transfer mathematical models of the packed bed. In cases
coefficients. of industrial interest, it is not possible to obtain
2. Assume a value for TES. closed form analytical solutions for any but the
3. Determine CES from 12.6.4. simplest of models under isothermal operating
4. From the effectiveness factor relation, deter- conditions. However, numerical procedures can
mine ri(TES, CES). be employed to predict effluent compositions on
5. Calculate the reaction rate per unit mass using the basis of the various models. In the subsec-
equation 12.6.2. (This implies a knowledge of tions that follow, we shall consider first the
the intrinsic rate expression.) fundamental equations that must be obeyed by
6. Determine TES from equation 12.6.3 using all packed bed reactors under various energy
the reaction rate determined in step 5. transfer constraints, and then discuss some of the
7. Compare the initial guess of TES with the simplest models of reactor behavior. These dis-
calculated value and iterate until agreement cussions are limited to pseudo steady-state op-
is reached. erating conditions (i.e., the catalyst activity is
8. Substitution of the final values of TES and/or presumed to be essentially constant for times
CES into any of the various equations for rm that are long compared to the fluid residence
gives the global rate. time in the reactor).
Models of fixed bed'reactors can be catego-
This procedure obviously requires machine rized in a couple of ways. One basis lies in the
computation capability if it is to employed in number of spatial coordinates employed in the
reactor design calculations. Fortunately, there equations used to describe the model. One-
are many reactions for which the global rate dimensional models take into account variations
reduces to the intrinsic rate, which avoids the in composition and temperature along the length
necessity for calculations of this type. On the of the reactor, while two-dimensional models
other hand, several high tonnage processes (e.g., also allow for variations in these properties in
SO 2 oxidation) are influenced by heat and mass the radial direction. A second basis for cate-
transfer effects and one must be fully cognizant gorizing reactor models lies in the manner in
of their implications for design purposes. which one envisions the reaction as being dis-
tributed throughout the catalyst bed. In this
12.7 DESIGN OF FIXED BED REACTORS
sense, the models are viewed as either pseudo
Fixed or packed bed reactors have many advan- homogeneous or heterogeneous. For the pseudo
tages relative to other types of heterogeneous homogeneous models, one envisions the reac-
catalytic reactors and, consequently, are em- tions as taking place throughout the reactor
492 Reactor Design for Heterogeneous Catalytic Reactions
volume, and not as localized at the catalyst sur- in words. Unfortunately, this solution is not so
face. The rate expressions for use with these easily expressed in the quantitative terms the
models are obtained by taking the product of engineer needs to prepare equipment specifica-
the global reaction rate per unit mass of catalyst tions and cost estimates. In principle, the solution
and the bulk density (pB) of the catalyst (i.e., the involves the use of a heterogeneous reaction rate
total mass of catalyst divided by the total volume expression in a set of material balance equations
of the reactor). for various chemical species and a correspond-
ing conservation equation for energy. The resul-
= rmpB (12.7.1)
tant differential equations and the attendant
The units of rv are moles converted/(volume- boundary conditions are then evaluated over the
time), and rv is identical with the rates employed catalyst bed to determine the conditions at the
in homogeneous reactor design. Consequently, reactor outlet. Before discussing these equations
the design equations developed earlier for homo- in detail, however, we consider briefly a few
geneous reactors can be employed in these terms subjects that are relevant to the general problem
to obtain estimates of fixed bed reactor per- of fixed bed reactor design. These include pres-
formance. Two-dimensional, pseudo homoge- sure drop, dispersion, and heat transfer in fixed
neous models can also be developed to allow for beds.
radial dispersion of mass and energy.
Heterogeneous models of fixed bed reactors 12.7.1.1 Pressure Drop in Fixed Bed Reactors.
explicitly account for the presence of the solid In order to simplify a reactor design analysis, we
catalyst by writing material and energy balance often assume that the pressure drop across a
equations for both the solid phase and the fluid reactor is negligible and has little influence on
phase. Several types of heterogeneous models reactor size requirements. However, there are
can be employed, depending on the complica- some situations in reactor design where it is
tions one is willing to introduce. The basic important to consider the implications of pres-
heterogeneous model considers only transport sure drop, particularly when dealing with the
by plug flow, but differentiates between bulk flow of compressible fluids through a fixed or
fluid properties and those prevailing at the exter- fluidized bed reactor. Because of its significance
nal surface of the catalyst pellet. Models may ip. a wide variety of chemical engineering opera-
also be developed that allow for intraparticle tions, flow through packed beds has been studied
gradients and for radial variations in system by many investigators through the years. The
properties. The necessity for using a heteroge- most useful approaches to predicting pressure
neous model can be circumvented through the drop through packed beds are those based on
use of global rate expressions in terms of rv in empirical correlations for the friction factor
the corresponding pseudo homogeneous models (/ M ), defined in the following manner.
described in Section 12.7.2. I have chosen to
- &L 4/ M
utilize this approach for the level of this text. (12.7.2)
[See Froment (94, 95) for critical reviews of
several heterogeneous models.] where
L is the length of the packed column
12.7.1 General Considerations p is the fluid density
Prediction of the performance of a fixed bed v0 is the superficial velocity (the average linear
reactor under specified operating conditions is velocity the fluid would have in the absence
a problem whose solution is readily expressed of packing)
12.7 Design of Fixed Bed Reactors 493
D* is the equivalent particle diameter defined as outlet densities. This approximation is useful in
computer-aided design problems.
D*=-
12.7.1.2 Dispersion in Packed Bed Reactors.
where In Chapter 11, we indicated that deviations from
plug flow behavior could be quantified in terms
av is the area per unit volume of an individual of a dispersion parameter that lumped together
particle the effects of molecular diffusion and eddy dif-
& represents the combined effects of static fusivity. A similar dispersion parameter is useji
pressure (P) and gravitational force and is to characterize transport in the radial direction,
defined as and these two parameters can be used to describe
radial and axial transport of matter in packed
g> = p + pgz (12.7.3)
bed reactors. In packed beds, the dispersion
where z is the distance upward (as opposed to results not only from ordinary molecular dif-
gravity) from a chosen reference plane. fusion and the turbulence that exists in the
Subscripts 0 and L refer to inlet and outlet &> absence of packing, but also from lateral de-
values, respectively. flections and mixing arising from the presence
Several correlating equations for the friction of the catalyst pellets. These effects are the domi-
factor have been proposed for both the laminar nant contributors to radial transport at the
and turbulent flow regimes, and plots of fM (or Reynolds numbers normally employed in com-
functions thereof) versus Reynolds number are mercial reactors.
frequently presented in standard fluid flow or If there is no correlation between the direc-
chemical engineering handbooks (e.g., 96, 97). tions in which a given fluid molecule is deflected
Perhaps the most useful of the correlations is at successive layers of pellets, the radial distance
that represented by the Ergun equation (98) through which the molecule is transported as it
D* 150(1 - eB)
2 + 1.75 (12.7.4)
G
where each term in parentheses is a dimension- moves longitudinally through the bed can be
less group. At high Reynolds numbers, the first described using Einstein's random walk ap-
term on the right drops out, and we obtain the proach. Each deflection will be comparable in
Burke-Plummer equation for the turbulent flow magnitude to the characteristic dimensions of
regime. Similarly, at low Reynolds numbers, the the pellet. If a large number of molecules are
second term drops out, and the Ergun equation considered, the effect is that of radial diffusion
reduces to the Blake-Kozeny relation for the characterized by an apparent diffusion coeffi-
laminar flow regime. cient in the radial direction that is given by (99):
Notice that the superficial mass velocity G is
constant throughout the bed, but that p will vary uzDp
(12.7.5)
for compressible fluids. When the pressure drop 11.2
is small compared with the absolute pressure,
equation 12.7.4 may be used for gases by em- where uz is the superficial velocity in the axial
ploying the arithmetic average of the inlet and direction. For gases, the magnitude of 2)R is
494 Reactor Design for Heterogeneous Catalytic Reactions
usually much greater than the molecular diffu- mated that the longitudinal Peclet number
sion coefficient. This disparity is even greater for should have a value of approximately 2. Experi-
liquids. The dimensionless group uzDp/@R is mental results indicate that for gases N Pe z is
known as the Peclet number for radial transport. equal to two for modified Reynolds numbers
(DpGjjX) above 10. For liquids, the longitudinal
iVp c , EE ^ (12.7.6) Peclet number is less, particularly at low Rey-
nolds numbers.
At high Reynolds numbers where molecular It should be emphasized that uz is the apparent
diffusion effects are negligible, experimental evi- superficial velocity in the longitudinal direction
dence confirms the general validity of equation (i.e., it is based on total area, voids plus solid).
12.7.5. Figure 12.15. indicates how the Peclet Similarly, in packed beds both S)L and 2)R are
number for radial mixing varies with the fluid normally based on total cross section.
Reynolds number. Above a Reynolds number of The extent to which longitudinal dispersion
40, the radial Peclet number is approximately 10. must be taken into account depends on the ratio
The Peclet number for axial dispersion is de- of the reactor length to the size of the particles.
fined in a manner similar to the radial parameter If the ratio is 100 or more, as is the usual case in
commercial reactors, the effect of longitudinal
dispersion is negligible in comparison with mass
Nft,z = ^ (12.7.7) transport due to bulk flow. Only in very short
reactors at high conversions and low velocities
where 3)L is the longitudinal dispersion param- is the effect significant. In such cases, however,
eter. Using a mathematical model that regards dispersion models of the reactor would not be
the interstices in the packing as a series of mixing realistic, and channeling and nonuniform flow
chambers, Aris and Amundson (101) have esti- distribution effects would be significant.
20 I I I I 1 i
' 1'
• • Berna -d and Wi helm
Figure 12.15
Correlation of average Peclet number with Reynolds number and the ratio
of particle to tube diameter (DPIDT). {From Chemical Engineering Kinetics
by J. M. Smith. Copyright c 1970. Used with permission of McGraw-Hill
Book Company.]
12.7 Design of Fixed Bed Reactors 495
12.7.13 Heat Transfer in Fixed Bed Reactors. ratios of the coefficient in the packed bed to that
In reactor design, it is important to be able to in the empty tube. Typical ratios range from 5 to
describe in quantitative terms the heat transfer 7.8, depending on the ratio of the pellet diameter
processes taking place within the reactor. In the to the tube diameter. Jakob (102) has proposed
design of homogeneous reactors, one normally the following correlation for wall heat transfer
assumes that radial mixing is sufficiently rapid coefficients as a generalization of an earlier
that all the resistance to energy transfer is con- correlation by Colburn (103):
centrated at the wall, and that the heat transfer 0.83
can be described by a single heat transfer T _ r*pOA
NPr (12.7.9)
coefficient. However, in fixed bed reactors, the
catalyst pellets inhibit radial mixing of the fluid, where
and significant gradients in temperature can
exist in both the radial and axial directions. DT is the tube diameter in feet
Here, we treat heat transfer between the con- Dp is the particle diameter in feet
taining wall of the reactor and its contents and / * is a coefficient given by Figure 12.16 with Dp
heat transfer within the packed bed. The latter in inches and DT in feet
process is described in terms of an effective
Kg is the thermal conductivity of the fluid in
thermal conductivity that lumps together contri-
BTU's per hour-foot-degree Fahrenheit
butions from a number of heat transfer mech-
anisms. Both this parameter and the wall hw is the wall heat transfer coefficient in BTU's
heat transfer coefficient can be estimated on the per hour-square foot-degree Fahrenheit
basis of correlations that summarize extensive \x is the fluid viscosity in pounds mass per hour
experimentation in this area. per foot
12.7.1.3.1 Heat Transfer to the Containing Wall G is the superficial mass velocity in pounds
Heat transfer between the container wall and the mass per hour per square foot
reactor contents enters into the design analysis This correlation indicates that the ratio of the
as a boundary condition on the differential or fixed bed coefficient to the empty tube coefficient
difference equation describing energy conserva- goes through a maximum, as Dp/Dr is varied.
tion. If the heat flux through the reactor wall is
0.24
designated as qw, the heat transfer coefficient at
the wall is denned as
K, = (12.7.8)
T T
1 1
W ~ B
where
Tw is the wall surface temperature
TB is the mean temperature of the bulk fluid in 0.3
the volume element at the axial position in
question
Figure 12.16
Normally, such heat transfer coefficients for Coefficient f* for use in equation 12.7.9. (From
packed beds are significantly greater than those Heat Transfer by M. Jakob. Copyright © 1957.
for empty tubes at the same gas flow rate. Early Reprinted by permission of John Wiley and Sons,
reports of such data were usually reported as Inc.)
496 Reactor Design for Heterogeneous Catalytic Reactions
Other useful correlations of heat transfer co- homogeneous body through which heat is trans-
efficients in packed beds have been proposed. ferred by conduction. Consider the volume
Among these is a simple relation of Calderbank element shown in Figure 12.17. It consists of a
and Pogorski (104) cylindrical shell of radius R, thickness AR, and
length Az, which is concentric with the tube axis.
(12.7,«) If an energy balance is written for the volume
element, the following input terms are necessary:
where the symbols have their usual significance. Input of energy by convective transport in the
Leva (105) has also proposed two correlations axial direction
for heating:
2nR(AR)puzCJT - T0)\zAt
hw = 0.813 °- 90
DT
= 0.125-
2 = 0
for 0.35 < - ^ < 0.6 (12.7.12)
DT
and one for cooling.
0 10
-
Input of thermal energy by effective radial con- and let Cp refer to the heat capacity of the gas
duction at temperature T.
Similar terms exist for output of thermal
At energy by these mechanisms. In addition to the
input and output terms, we also need terms for
Input of thermal energy by effective longitudinal the generation of thermal energy by chemical
conduction reaction
rv(-AH)2nR(AR)(Az)At
K*2nR{AR)
- * - * ? • At
8z
and the accumulation of energy within the
where To is the datum temperature for enthalpy volume element
dT
Input 2nR(AR)puzCp(T - To% At - K12UR{AZ) At - K*2nR(AR) At
Tz
+ generation + rv( - AH)2nR(AR)(Az)(At)
dT
= output = 2nR(AR)puzCp(T - T0)\z + Az At - K*2TLR(AZ) — At
cR R + AR
AR At
oz
+ accumulation
(12.7.14)
498 Reactor Design for Heterogeneous Catalytic Reactions
A T - T0)\z-puzCpR(T ^ T0)\z.
gas + U ~ ^To ) 1 R R r (
At —
Az
\ C l\ IR + AR R
+ AR
• R (8T\
fdT\
- + Az
(12.7.15)
Az
and in the limit, as the various deltas approach zero,
For steady-state operation, the time deriva- range of effective thermal conductivities typical
tive vanishes and we have the form that is of of various Reynolds numbers. The data pre-
greatest interest for purposes of this text. sented are most appropriate for use at tempera-
12.7 Design of Fixed Bed Reactors 499
Z.UU
I I fluid, the effects of transport by turbulent motion
of fluid eddies, the effects of radiation transport
through the void spaces, and the effects of con-
duction through the solid. The last term lumps
1.50 — —
V
4 together a number of effects associated with
energy transfer from one pellet to adjacent
/ /
pellets by radiation, point contacts, and con-
1.00 — vective transport to and from the fluid. The
numerical constant in the radiation term implies
that British Engineering Units (with time in
V hours) must be used for the various parameters
0.50 — and, in particular, the bed temperature T must
be expressed in degrees Rankine.
V Most of the available data have been recorded
0 i I under conditions such that only the terms for
200 400 600 800 eddy transport and conduction through the
DpG/fi solid are significant. Equation 12.7.19 requires
Figure 12.18
that K* increase with particle diameter, mass
Heat transfer in packed beds. Effective thermal velocity, and the conductivity of the solid. It is
conductivity as a function of Reynolds number. consistent with data for low conductivity solids,
Curve 1 : Coberly and Marshall. Curve 2: Campbell but some discrepancies arise for very high
and Huntington. Curve 3: Calderbank and Pogorski. conductivity solids (108). At Reynolds numbers
Curve 4: Kwong and Smith. Curve 5: Kunii and greater than 40, the contribution of the molecu-
Smith. (From G. F. Froment, "Chemical Reaction lar conduction term is negligible.
Engineering," Adv. Chem. Ser., 109, 1970.) The parameter h', which combines several
effects, must be at least as large as the convective
tures below 300 °C where radiation heat transfer heat transfer coefficient (/ic), calculated from the
is negligible. jH factor correlations discussed in Section 12.5.
Argo and Smith (106, 107) have presented a It is given by an expression of the following
detailed discussion of heat transfer in packed form
beds and have proposed the following relation h! = hc + hR + hp (12.7.20)
for the effective thermal conductivity in packed where hR and hp represent the contributions of
beds: radiation between particles and conduction for
pellets in contact. These quantities can be
100z estimated as:
Solution
ILLUSTRATION 12.7 DETERMINATION OF
Equations 12.7.20 to 12.7.23 may be used to
THE EFFECTIVE THERMAL CONDUCTIVITY
evaluate the parameter h'. We begin by con-
OF A PACKED BED OF CATALYST PELLETS
verting the convective heat transfer coefficient
In Illustrations 12.5 and 12.6, some data on the hc, determined in Illustration 12.6, to British
catalytic oxidation of SO 2 were used to deter- Engineering Units.
mine composition and temperature differences hc = 8.79 x 10" 3 J/cm2-sec-°K
between the bulk fluid and the fluid at the pellet-
gas interface. Use the data and results of these =c= 15.56 BTU/ft2-hr-°R
illustrations and the new data given below to From equation 12.7.20 it is evident that W must
predict the effective thermal conductivity of the be at least as large as hc, and this fact provides
bed. a useful starting point for solving these tran-
scendental equations. We begin by assuming
Data and Results from Illustrations 12.5 and that h' = 35 BTU/(hr-ft2-°R). From equation
12.6 12.7.23,
T = 458 °C - 731 K = 1316 °R . o g K p = -1.76 +0.0129 »
P = 790 mm Hg
G= 245 lb/hr-ft2 = 3.33 x 10" 2 g/sec-cm2 or
KP = 0.0199
Kg = 3.70 x 10" 4 J/(cm-sec-°K)
From equation 12.7.22,
= 0.0214 BTU/(hr-ft°F)
_ 2(1.8) + 35(0.01276)
= 0.032 cp (0.0199)
p
~~ (0.01276X1.8)
Pp = 1.81 g/cm3
= 0.40 = 3.51 BTU/(hr-ft2-°R)
Kg = 0.0214 BTU/(hr-ft-°F)
DpCpG _ (1.276 x 10"2)(0.255)(245)
= 0.1935 BTU/(hr-ft-°F)
(10.3)(0.40)
4(0.65) (1316)3
D,(0.173) 4 (1.276 x 10~2)(0.173)
2 - a 100 2 - 0.65 (100)4
= 0.0969 BTU/(hr-ft-°F)
- eB)h'KsDp _ (1 - 0.40)(36.1)(1.8)(1.276 x 10~2)
= 0.1225 BTU/(hr-ft-°F)
2KS + h'Dp 2(1.8) + (36.1)(1.276 x 10~2)
(composition and temperature) and the reactor diameter, the adiabatic model may provide
volume. Usually, it is necessary to transform a good representation of actual operating
the coupled differential equations into differ- experience.
ence equations and to use numerical methods However, the most complex analysis is that
to solve the transformed equations. The con- in which heat transfer through the reactor walls
ventional approach to this problem involves is taken into account. This type of operation
the use of a stepwise procedure in which one must be employed when it is necessary to supply
starts at the reactor inlet and marches longitu- or remove energy from the system so as to
dinally through the reactor in appropriate moderate the temperature excursions that would
volume increments. One-dimensional or two- otherwise follow. It is frequently employed in
dimensional models may be employed de- industrial reactors and, to model such systems,
pending on the degree of sophistication desired, one must often resort to two-dimensional
the cost constraints on computer aided design, models of the reactor that allow the concentra-
and the level of approximation required. Either tion and temperature to vary in both the radial
model can be handled by modern computing and axial directions. In the analysis of such
machines, but the one-dimensional model is systems, we make incremental calculations
used most often in preliminary design calcula- across the diameter of a given longitudinal
tions because it provides a good approximation segment of the packed bed reactor, and then
to the desired result with limited consumption proceed to repeat the process for successive
of computer time, and it can be used to deter- longitudinal increments.
mine the effect of changes in design parameters The various energy transfer constraints enter
and operating conditions on effluent conditions into the analysis primarily as boundary condi-
(see Section 12.7.2.1). The two-dimensional tions on the difference equations, and we now
model is more complex, but provides essential turn to the generation of the differential equa-
information about radial temperature profiles tions on which the difference equations are
within the bed. Such information is particularly based. Since the equations for the one-dimen-
useful in evaluating the potential for catalyst sional model are readily obtained by omitting
deactivation or in dealing with systems in which or modifying terms in the expressions for the
selectivity considerations are significant (see two-dimensional model, we begin by deriving
Section 12.7.2.2). the material balance equations for the latter.
The computational effort required to carry For purposes of simplification, it is assumed
out the design analysis is determined mainly by that only one independent reaction occurs with-
the magnitude and spatial distribution of the in the system of interest. In cases where multiple
temperature variations that are taken into reactions are present, one merely adds an ap-
account. The maximum temperature difference propriate term for each additional independent
between the inlet and outlet of the reactor occurs reaction.
when the reactor operates adiabatically. In this Consider the element of reactor volume
case, heat transfer to the reactor wall is neglected shown in Figure 12.17. The cylindrical shell is
so there is no temperature variation in the radial concentric with the cylinder axis, about which
direction. However, the temperature does vary there is general symmetry. A material balance
in the axial direction, so the material and energy on the volume element requires consideration
balance equations are coupled through the de- of several terms. The amount of a particular
pendence of the reaction rate on temperature. species A that enters the volume element during
If the reactor is well insulated, and/or of large a time increment At consists of material entering
12.7 Design of Fixed Bed Reactors 503
by longitudinal bulk flow, the total area (void plus solid) normal to the
direction of transport. These parameters are
\_CAuz2nR(ARj]z(At)
those involved in the definition and discussion
material entering as a consequence of axial of the Peclet numbers presented in Section
dispersion, 12.7.1.2.
The amount of A that leaves the volume
element during the time increment At consists
1 7 > I (At) of three similar terms. The amount of A that
and material entering by radial dispersion, accumulates within the volume element during
time At is (ACA)2nR(AR)(Az\ and the amount
of A that is generated by chemical reaction with-
in the volume element is vArv2nR(AR)(Az)(At),
where rv is the reaction rate expressed in pseudo
where &>L and Q)R are effective dispersion param- homogeneous form [i.e., the number of moles
eters that lump together molecular diffusion transformed per unit time per unit of total
and eddy diffusion arising from turbulence and reactor volume (voids plus solid)].
packing effects. As used in the pseudo homo- The various terms may be combined in a
geneous models, both diffusivities are based on generalized material balance to give:
= output [C>z27cR(AK)]z + Az At - I ^ L At
- \sR2nR{Az) (^A] At
+ accumulation +(ACA)2nR(AR)(Az) (12.7.24)
Az
dC,
+ v4r,,R = R
AR Az Af
(12.7.25)
504 Reactor Design for Heterogeneous Catalytic Reactions
and, in the limit, as Az, AJR, and At, all approach zero
where we have presumed that Q)L and the tube diameter are constant over the length of the reactor, and
that 3)R is constant over the reactor diameter. It is necessary to retain uz within the partial derivative
to cover those cases where there is a change in fluid density on reaction. When there is an increase or
decrease in the number of moles on reaction, or a change in fluid temperature or pressure, the velocity
component in the longitudinal direction will vary along the length of the reactor. For gas phase
reactions, the effect can be quite significant.
The last equation can be rewritten as
C { dC
*
At steady state, the right side vanishes, and one obtains an equation that describes in pseudo homo-
geneous fashion the concentration profile in the reactor.
For isothermal systems this equation, together with an appropriate expression for rv, is sufficient to
predict the concentration profiles through the reactor. For nonisothermal systems, this equation is
coupled to an energy balance equation (e.g., the steady-state form of equation 12.7.16) by the depen-
dence of the reaction rate on temperature.
i ^ - ^ 1 ^ +^ =0 (12.7.30)
K
tj^-GYz {C T)
" ~ 4~ hJT ~ TJ
" rv(AH)
=° (12.7.31)
where Tw is the tube wall temperature.
12.7 Design of Fixed Bed Reactors 505
The boundary conditions that govern both exceeds 30, a plug flow model of the velocity
the one- and two-dimensional models are profile provides a good representation of the
usually stated in the following manner. actual profile across the central core of the
tubular reactor. Departures from the average
velocity are significant only in the thin annular
region near the tube walls. In single tube
commercial-scale reactors this criterion is usu-
(12.7.34) ally met [e.g., if the pellets have a characteristic
at R = 0 for all z dimension of 0.6 cm ( ~ l / 4 in.), the required
tube diameter is only 18 cm ( ^ 7 in.)]. However,
(12.7.35) in the multitube reactors required to meet the
dR
large heat transfer requirements of very rapid
= 0 at R = Ro for all z (12.7.36) highly exothermic reactions, the assumption of
dR a relatively flat velocity profile is not a good
In the formulation of the boundary condi- one. Such systems are difficult to model in
tions, it is presumed that there is no dispersion mathematical terms. In order to generate the
in the feed line and that the entering fluid is necessary data for the design of such reactors,
uniform in temperature and composition. In the practicing engineer frequently has to resort
addition to the above boundary conditions, it is to pilot scale reactors that consist of one or
also necessary to formulate appropriate equa- more tubes of the same diameter and length as
tions to express the energy transfer constraints those to be employed in the commercial-scale
imposed on the system (e.g., adiabatic, iso- multitube reactor.
thermal, or nonisothermal-nonadiabatic opera- We turn now from a discussion of general
tion). For the one-dimensional models, bound- principles to specific models of reactor behavior.
ary conditions 12.7.34 and 12.7.35 hold for all
R, and not just at R — 0. 12.7.2.1 The One-Dimensional Pseudo Homo-
When the velocity uz varies with radial posi- geneous Model of Fixed Bed Reactors. The
tion, equation 12.7.28 must be solved by a design of tubular fixed bed catalytic reactors
stepwise numerical procedure. Experimental has generally been based on a one-dimensional
evidence indicates that the axial velocity does model that assumes that species concentrations
indeed vary with radial position in fixed bed and fluid temperature vary only in the axial
reactors. The velocity profile is relatively flat in direction. Heat transfer between the reacting
the center of the tube. As one moves radially fluid and the reactor walls is considered by
outward, the velocity increases gradually until presuming that all of the resistance is contained
a maximum is reached at a point about one within a very thin boundary layer next to the
pellet diameter from the tube wall. It then falls wall and by using a heat transfer coefficient
rapidly, until it reaches zero at the wall. If the based on the temperature difference between
ratio of the tube diameter to the pellet diameter the fluid and the wall. Per unit area of the tube
506 Reactor Design for Heterogeneous Catalytic Reactions
wall, the heat flow rate from the reactor con- and that
tents to the wall is then hw(T — Tw). The cor-
relations for hw presented in Section 12.7.1*3 V = (12.7.41)
may be used to evaluate this parameter.
The one-dimensional model is advantageous equation 12.7.39 can be written as
because it provides a rapid means of (1) ob-
taining an estimate of the reactor size necessary
to achieve a given conversion, and (2) examining
the influence of several design variables on the Now dFA = —FA0 dfA, so the previous equa-
behavior of the reactor. Unfortunately, the tion can be written as
model provides no information concerning the
possibility of achieving an excessive temperature VR_= C djA
at the center of the tube that may be markedly (12.7.43)
FA0 * ~rAe
different from the mean temperature at the same
longitudinal position. Such temperatures may which is identical with the equation for an ideal
be unacceptable for reasons of reactor stability, plug-flow reactor, derived as equation 8.2.7. If
process selectivity, or catalyst deactivatioit. To both sides of this relation are multiplied by the
ensure that such excessive temperatures are not bulk density of the catalyst, we again generate
achieved along the reactor axis, one should equation 8.2.12.
determine if they are present -using a two-
dimensional model to analyze those sets of PBVR W _ CPHJ/A _ f dfA (12.7.44)
operating conditions that the one-dimensional AO AO
model indicates are promising. The two-dimen-
sional model is discussed in Section 12.7.2.2. The general energy balance for the one-
The equation describing trie steady-state ma- dimensional, packed bed reactor cannot be
terial balance for tubular packed bed reactors obtained by omitting the^ radial derivatives of
can be obtained from the more general relation the temperature in equation 12.7.29, because
(12.7.28) by omitting the terms corresponding these terms ultimately lead to heat transfer
toradial transport of matter. Hence the material through the walls of the reactor. They can be
balance relation becomes omitted only for adiabatic operation. In the
-general... case of a one-dimensional model, the
d2cA d(cAuz) rate of heat loss through the wall must be
_2 + v rv = 0 (12.7.38) considered. For a differential length of reactor
dz dz
(Az), this loss is given by
Further simplification of the material balance
hwnDT(T -TJ(Az)
equation occurs if the axial dispersion term is
where
neglected. In this case,
DT is the tube diameter
d(CAut) T w is the local wall temperature
- v/,, = 0 (12.7.39)
dz
hw is a wall heat transfer coefficient which can
be evaluated from the empirical correlations
If one recognizes that
presented in Section 12'.7.1.3.1.
nD
r An energy balance over the differential length
(12.7.40)
of reactor for steady-state operating conditions
12.7 Design of Fixed Bed Reactors 507
C (T - T 0 ) z +
P y rv(-AH) Az +
4 / ^p
d
~ - Tw) = rv(-AH) (12.7.46)
dz
This equation can also be rewritten in terms of the superficial mass velocity G, which does not vary
along the length of the reactor.
8
IK* ( j^hw(T - TJ = rv(-AH) (12.7.47)
Tz\KL\dz dz
As indicated earlier, the axial conduction are coupled through the dependence of the
term is almost always negligible compared to reaction rate on temperature. However, for
the convective enthalpy transport term. There- isothermal operation, the equations are not
fore, equation 12.7.47 is usually simplified to coupled, and dT/dz becomes zero. For adiabatic
give operation, the heat losses through the wall are
negligible, and the energy balance becomes
4
hw(T - Tw) = rv(-AH)
~D~T
(12.7.48) G-(CpT) = rv(-AH) (adiabatic case)
(12.7.49)
Equations 12.7.48 and 12.7.39 provide the
simplest one-dimensional mathematical model Depending on the operational constraints,
of tubular fixed bed reactor behavior. They one of the two equations 12.7.48 or 12.7.49 or
neglect longitudinal dispersion of both matter the choice of isothermal behavior must be used,
and energy and, in essence, are completely together with the general material balance
equivalent to the plug flow model for homo- relation (equation 12.7.39), to determine the
geneous reactors that was examined in some composition and temperature profiles along
detail in Chapters 8 to 10. Various simplifi- the length of the reactor.
cations in these equations will occur for different In many instances, the pressure drop through
constraints on the energy transfer to or from the the reactor will be relatively small so that we
reactor. Normally, equations 12.7.48 and 12.7.39 may employ a mean value for the total pressure
508 Reactor Design for Heterogeneous Catalytic Reactions
in our calculations. However, the Ergun equa- energy and matter. The dispersion that repre-
tion presented in Section 12.7.1.1 may be used, sents the combined effects of molecular diffusion,
if necessary, to evaluate the pressure at various normal fluid turbulence, and the influence of
points in the bed. Solutions of the appropriate the packing is accounted for by superimposing
material and energy balance equations for this "effective" heat and mass transfer terms on the
simplified one-dimensional model can be ob- plug-flow equations. Because of the assumptions
tained by straightforward numerical procedures involved in the definitions of the effective
employing digital computation to solve the thermal conductivity and dispersion terms, they
corresponding difference equations. While the implicitly contain the effect of radial velocity
correspondence of these solutions to the behav- profiles within the tubes. Normally, one uses
ior of actual reactors must be viewed as only an values averaged over the tube diameter, as
approximation, this model provides a means of expressed by Peclet numbers for heat and mass
simulating the steady-state response of the transfer. However, the velocity profile is not
system under investigation to changes in various considered explicitly. For design purposes, we
process parameters at minimum cost in com- take the longitudinal Peclet number based on a
puter time and programming effort. The sim- pellet diameter, to be approximately 2. Effective
ulations enable us to answer questions such as: thermal conductivities may be estimated using
the method described in Section 12.7.1.3.2.
1. What tube length will be required to achieve
In many respects, the solutions to equations
a given conversion?
12.7.38 and 12.7.47 do not provide sufficient
2. What will be the corresponding effluent
additional information to warrant their use in
temperature?
design calculations. It has been clearly dem-
3. What does the longitudinal temperature
onstrated that for the fluid velocities used in
profile look like for given inlet temperatures
industrial practice, the influence of axial dis-
and/or wall temperature profiles? Are the
persion of both heat and mass on the conversion
hot spots excessive for reasons of selectivity,
achieved is negligible provided that the packing
catalyst deactivation, etc?
depth is in excess of 100 pellet diameters (109).
4. How do changes in the tube diameter in-
Such shallow beds are only employed as the
fluence the design calculations?
first stage of multibed adiabatic reactors. There is
5. How do changes in catalyst pellet size affect some question as to whether or not such short
system behavior? beds can be adequately described by an effective
6. How can one avoid excessive sensitivity of transport model. Thus for most preliminary de-
the performance to changes in process param- sign calculations, the simplified one-dimensional
eters? model discussed earlier is preferred. The dis-
The last question pertains to the problem of crepancies between model simulations and
parametric sensitivity. There is extensive liter- actual reactor behavior are not resolved by the
ature dealing with this topic, but it is beyond inclusion of longitudinal dispersion terms. Their
the scope of this book. effects are small compared to the influence of
A more general one-dimensional model of radial gradients in temperature and composi-
tubular, packed bed reactors is contained within tion. Consequently, for more accurate simula-
equations 12.7.38 and 12.7.47. These equations tions, we employ a two-dimensional model
include all of the elements of the simple model (Section 12.7.2.2).
discussed above and, in addition, account for Nevertheless, one feature of the one-dimen-
the longitudinal dispersion of both thermal sional model containing dispersion terms is of
12.7 Design of Fixed Bed Reactors 509
considerable interest. These terms increase the The following illustrations indicate the man-
order of the partial differential equations and, ner in which the one-dimensional model is
under certain conditions, lead to nonuniqueness employed in reactor design analyses.
of the steady-state profile through the reactor
(110). For certain ranges of operating conditions
and parameter values, three or more steady- ILLUSTRATION 12.8 PRODUCTION OF
state profiles can be obtained for the same feed SULFUR TRIOXIDE IN AN ADIABATIC
conditions. The two outlying steady-state pro- FIXED BED REACTOR
files will be stable (at least to small perturba-
tions), while the intermediate one will be Industrial scale production of sulfuric acid is
unstable. The profile generated as a solution to dependent on the oxidation of sulfur dioxide to
equations 12.7.38 and 12.7.47 will depend on the sulfur trioxide in fixed bed catalytic reactors.
initial guesses of T and C involved in the trial
and error solution. so2 so,
In physical terms, this assessment means Through the years, several catalyst formulations
that the steady state that would be achieved in have been employed, but one of the traditional
a real reactor will depend on the initial profile catalytic agents has been vanadium pentoxide.
in the reactor. For all situations where the Calderbank (114) has indicated that for a
initial values differ from the feed conditions, catalyst consisting of V2O5 supported on silica
we must solve the time dependent differential gel, the kinetic data are represented by a rate
equations describing the reactor to determine expression of the form
which profile will prevail. In order to determine
if such computations are necessary, one must
know whether or not multiple steady states are pl/2 (A)
r
possible. Several criteria that permit us to make so 2
this decision on an a priori basis have been that may be regarded as a degenerate form of
discussed by various authors (111-113). For typical Hougen-Watson kinetics. The rate con-
details of these analyses, consult the indicated stants are given by
references. For present purposes we are inter-
ested only in the general conclusions derived 31,000
£n fcx = 12.07 - (B)
from such studies. The range within which such RT
multiple steady states are possible is very and
narrow, and such conditions are rarely en- 53,600
In k7 = 22.75 - (Q
countered in either laboratory or industrial RT
scale reactors. They are most likely to be
where
realized for highly exothermic reactions carried
out in extremely short beds under adiabatic T is expressed in degrees Kelvin
operating conditions. The length of virtually
R is expressed in calories per mole-degree
all industrial scale reactors precludes the ne-
Kelvin
cessity of including longitudinal dispersion terms
in a one-dimensional model of reactor behavior kx is expressed in moles per second-gram of
and of worrying about their implications for catalyst-atmosphere 3/2
multiple steady states. The effects of radial k2 is expressed in moles per second-gram of
gradients are much more likely to be significant. catalyst-atmosphere
510 Reactor Design for Heterogeneous Catalytic Reactions
and
The partial pressures of the various species are numerically equal to their mole fractions since the
total pressure is one atmosphere. These mole fractions can be expressed in terms of a single reaction
progress variable-the degree of conversion-as indicated in the following mole table.
8.0/
o2 13.0 13.0 -
2
(13.0 - 4.0/)/(100 - 4.0/)
In terms of the fraction conversion, the catalytic reaction rate expression then becomes
The mass velocity G will be constant over the reactor length, and this quantity may be determined
from the specified production rate and the reactor dimensions. To produce 50 tons/day of 100%
H 2 SO 4 , the number of pound moles of SO2 that must be oxidized per second is:
For 99% conversion of the SO 2 , the inlet molal flow rate must be
1.18 x 1(T 2
= 0.149 lb moles/sec
(0.99)(0.08)
The average molecular weight of the inlet gas is (0.08)(64) + 0.13(32) + 0.79(28) = 31.4. The mass
velocity G is then (0.149)(31.4)/[TI(6) 2 /4] = 0.165 Ib/(ft2-sec).
We now wish to examine the heat capacity per unit mass to determine if it varies significantly with
conversion. At the inlet conditions, the molal heat capacity of the gaseous feed will be equal to ^(y f C pi ).
Hence at / = 0, with Cp in units of calories per gram degree Kelvin and T in degrees Kelvin.
If complete conversion were to take place ( / = 1), the heat capacity of the effluent mixture could
be evaluated with the aid of the mole table on page 511.
Equations I and J indicate that for the temperature range of interest (640 to 830 °K), the heat capacity
per unit mass is substantially independent of the conversion level. Furthermore, the temperature
dependent contribution to the heat capacity will not vary much over the temperature range involved.
Hence without introducing errors comparable to those inherent in the use of a one-dimensional model,
we may take the heat capacity as constant at 0.250 cal/g-°K or 0.250 BTU/(lb-°F).
Combining equations G and D indicates that
GCp~ = (-AH)CAOuofz
where we have used our assumption that Cp is a constant. Integration of this equation between the
reactor inlet and location z gives
0 08(
- ° 2 1 4 9 ) = 4.22 x 1 0 - l b moles/ft^-sec
7c(6)74
where the factor 1.8 is necessary to convert T in degrees Kelvin to consistent units.
12.7 Design of Fixed Bed Reactors 513
Hence the relation between the temperature and the fraction conversion at any point in the adiabatic
reactor is given by
T - To = (251.9 - 0.0204T)(/- / 0 )
or
T=T0 +251.9(/ - /Q)
l
1 + 0.0204(/ - /„) '
Equations G and H may be combined to give
CAOuo df fci^OKl - /)]1/2(13.0 - 4.0/) M8.O/M13.O - 4
-°/) 1/2
3/2 1/2 (N)
pB dz (100 - 4.0/) [8.0(1 - / ) ] ( 1 0 0 - 4.0/)
Equations M and O may now be solved numerically to determine the bed depth corresponding to a
given conversion.
If we rewrite equation O as
Az = j5(Afl (P)
where
(100 - 4/)
the coefficient /? may be evaluated at the beginning and the end of the conversion increment, and an
average value employed to determine Az.
At the inlet to the first reactor/ = 0 and To = 370 °C = 643.16 °K. Hence
(10Q)3/2
^o — (438 x io 1 0 )(13)e~ 3 1 ' 0 0 0 / ( 1 ' 9 8 7 ) ( 6 4 3 1 6 ) =
^'^
If we chose our first conversion increment as A/ = / i — f0 = 0.05, equation M indicates that
643.16 + 251.9(0.05)
Tl
- (0.0204X0.05)
514 Reactor Design for Heterogeneous Catalytic Reactions
Thus
[100 - 4(0.05)]
Pi = 10 1/2 - 40.2
4.38 x 10 (1 - 0.05) [13 - 4(0.05)] 1.90 x 1015(0.05)[13 - 4(0.05)]11/2
1/2 31 000/(1 987)(65509)
[100 - 4(o.05)] e ' - ~ (1 - 0.05) 1/2 e 5
and
= zx - 0 =
or
zl = 2.51 ft
Proceeding to the next increment for A/again equal to 0.05, we find f2 = 0.10 and
„. 643.16 + 251.9(0.10) = m m ^
0.0204(0.10)
Thus
[100 - 4(0.1)]
= 27.4
4.38 x 10 10 (l - 0.1)1/2[13 - 4(0.1)] _ 1.90 x 1015(0.1)[13 - 4(0.1)] 1/2
PlOO — 4 ( 0 1 ) ] 1 / 2 e
31 00
' °/<1-987)(666.99) "~ /j Q j_ e53'600/(1'987)(666"99)
Table 12.1.1
Temperature and Conversion Profiles for Packed Bed Reactor Network
Stage 1 Stage 2
Catalyst Catalyst
depth depth
SA Temperature, T(°K) (ft) IA Temperature, T(°K) (ft)
to be willing to incur the associated capital Before proceeding to the next illustration, it is
charges. instructive to see if our assumption of constant
The total weight of catalyst required is given total pressure was indeed appropriate. If one
by the product of the bulk density and the total assumes 1/4 in. spherical pellets, a nominal gas
reactor volume viscosity of 0.09 lb/hr-ft and a bed porosity of
0.4, the Reynolds number of the gas is given by
Weight of catalyst - 37.4 ^ - ( 8 . 4 8 + 28.76)
(0.165)(3600)
- 39,400 lb or 19.7 tons
In practice, one could oversize both stages (0.09)
and operate with reduced inlet temperatures that = 137.5
could subsequently be raised as catalyst deacti-
vation proceeds. From equation 12.7.4,
150(1 - 0.4)
+ 1.75 = 2.40
1 - £, 1315
516 Reactor Design for Heterogeneous Catalytic Reactions
37.24 (Q-165)2 lb m
3
0.4 1 3.6 x 10" 2 ft-sec 2
4 V12
lb,, lb f sec 2 ft2
= 3.05 x 104
ft-sec2 32.2 lbm-ft 144 in.2
= 6.57 psi
This pressure drop is a significant fraction of the suming that the heat transfer to the surroundings
total pressure. Consequently, the analysis should can be characterized by an overall heat transfer
be repeated, breaking up the reactor into seg- coefficient based on the temperature difference
ments that could be treated as having a constant between the reactor contents and ambient con-
average pressure. As an alternative, we could ditions (70 °F). When based on the inside area
go to larger diameter, shorter beds of catalyst to of the reactor tube, the heat transfer coefficient
reduce the pressure drop, while maintaining the has a numerical value of 1.2 BTU/(hr-ft2-°F) or
same conversion according to the one-dimen- 6.0 x 10~4(BTU/sec-ft2-°K). The second-stage
sional homogeneous model. For example, by inlet temperature may be taken as 410 °C
going to a nine foot diameter bed, the total bed instead of 370 °C, as in Illustration 12.8.
length could be reduced from 37.24 to 16.6 ft and
the pressure drop to 0.7 psi. Total catalyst re- Solution
quirements would be unchanged. Such trade-offs
The finite difference form of the material balance
indicate why large diameter beds with small
equation developed in Illustration 12.8 (equa-
length-to-diameter ratios are often employed in
tion P) is again applicable:
catalytic reactors. Section 13.1 indicates how
to treat the pressure variation along the length Az = j8A/ (A)
of a tubular reactor. where
transfer coefficient (^ll) and with a corresponding However, solution of the difference equations
change made in the temperature driving force. requires a knowledge of the temperature at the
end of the conversion increment. Consequently,
- T ambient ) = (-AH)rv a trial and error procedure is indicated. One
CZ
assumes a value for the temperature at the end
The right side of this equation can be rewritten of the increment, computes Az from equation A,
in terms of the fraction conversion, using equa- and checks the temperature assumption using
tions E and G of Illustration 12.8. equation B. Under normal circumstances, the
d_
G-ACJ) D,
~ ^ambient) = rmpB(-AH) = ( - AH)CA0U0 —
dz' CZ
655.09 + 643.16
-[-
AT = 251.9 - 0.0204
= 7.13 °K
(0.05) - 5.387 x - 294.27 j (2.51)
AT =[ 2 51.9-0.0204 (0.05)
= 6.81 °K
orTx = 649.97 GK, which is in better agreement 12.1.2). In the present case, thermodynamic con-
with the initial estimate of the temperature. To straints on the rate indicate that the first stage
achieve the desired degree of agreement, one may effluent should correspond to a conversion in the
iterate as often as necessary. vicinity of 0.87 and an effluent temperature near
For the second conversion increment (A/ = 824.4 °K. Beyond this point, the system is so
0.05, / = 0.10), we will assume T = 657 °K close to thermodynamic equilibrium that sub-
(AT - 7.03 °K). Here, stantial increments in the reactor length do not
produce noticeable increments in the conversion.
jS2 = 39.0 Here, one may employ a higher temperature
for the second-stage feed and still achieve the
and, from equation A, desired conversion level. The conversion and
temperature profiles as calculated from equa-
'48.5 + 39.0 tions A and B are given in Table 12.1.2.
Az = (0.05) = 2.19 ft
The total weight of catalyst required is equal to
657 + *
AT 2 = 251.9 - 0.0204 (0.05)
= 7.69 K
12.7 Design of Fixed Bed Reactors 519
Table 12.1.2
Conversion and Temperature Profiles for Packed Bed Reactor of Illustration 12.9
This value is considerably less than that ob- 72.7.2.2 The Two-Dimensional, Pseudo Homo-
tained for pure adiabatic operation (19.7 tons). geneous Model of Fixed Bed Reactors. Two-
The heat losses tend to partially remove thermo- dimensional models permit more realistic simu-
dynamic constraints on the reaction rate and lation of fixed bed reactor behavior than the
permit a closer approach to the optimum tem- one-dimensional models discussed previously.
perature profile corresponding to minimum Experimental measurements indicate that the
catalyst requirements. fluid temperature and composition are not uni-
A few words concerning the results of our form across a section of the tube normal to the
analyses in Illustrations 12.8 and 12.9 are in flow. The one-dimensional models discussed
order. Obviously, better estimates of the catalyst earlier neglect the radial resistance to heat and
requirements could be obtained by using smaller mass transfer and thus predict uniform tem-
conversion increments. We have not attempted perature and composition for each longitudinal
to fully optimize the reactor stages in terms of position. This assumption is obviously a vast
catalyst minimization. Furthermore, we have oversimplification when reactions with large
again neglected pressure drop in each stage. heat effects are considered. Whenever there is
Further calculations would remedy each of the extensive heat exchange between the packed bed
aforementioned shortcomings of the analysis. reactor and its surroundings, one requires at
They are readily accomplished with the aid of least a two-dimensional model to simulate the
machine computation. reactor performance. In such cases, the design
engineer requires a model that predicts the de-
520 Reactor Design for Heterogeneous Catalytic Reactions
tailed temperature and composition patterns in ding difference equations. As boundary condi-
the reactor to be able to avoid hot spots along the tions, one needs to know the temperature and
reactor axis when they would be detrimental for composition profile across the tube diameter at
reasons of selectivity, catalyst deactivation, etc. the reactor inlet. In addition, the solution must
A complete two-dimensional model would satisfy the requirements that
account for the radial distribution of velocity,
for the radial concentration and temperature 8CA
^ = 0 at R = 0 for all z (12.7.51)
gradients in terms of Peclet numbers that them- cR
selves varied with radial location, and for axial
dT
dispersion of heat and mass. With machine com- = 0 at R = Ofor all z (12.7.52)
putation, such considerations may be handled ~d~R
if the cost in computer time and programming (for reasons of symmetry) and that
effort can be withstood. However, the most sig-
nificant aspects of two-dimensional models can
be demonstrated by considering simple two- —~ = 0 at R = ^ = Ro (tube wall)
<m 2
dimensional models. Such models assume that
(12.7.53)
the mass velocity G and the radial Peclet num-
bers for heat and mass transfer are constant since the tube is not permeable to the fluid. The
across the tube diameter and that the effective boundary condition at the wall on the energy
thermal conductivity and effective dispersion conservation equation can be framed as a re-
contributions to longitudinal transport of energy quirement that the heat flux at the wall be of the
and mass are insignificant (compared to the con- following form:
vective transport terms). The appropriate mate-
rial and energy balance equations that provide Q= ~KR
a two-dimensional description of the reactor
performance at steady state have been derived (12.7.54)
earlier in discussing general aspects of heat trans-
where
fer and of mass conservation in packed beds.
They are equation 12.7.18 Tw is the inside wall surface temperature
. d(CpT) TRo is the temperature of the surrounding fluid
+ rv(AH) = 0 hw is a heat transfer coefficient
cz dR2
(12.7.18) This approach in essence assumes a temperature
discontinuity at the wall. Alternative methods
for energy conservation and a simplified form
of formulating this constraint have also been
of equation 12.7.28
proposed (94, 95, 115), together with empirical
1 dCAl d(CAu: methods of evaluating the heat transfer coeffi-
+ vArv = 0 cients introduced by each method.
3R2 R dR J cz
Various numerical procedures may be em-
(12.7.50)
ployed to solve the difference equations corre-
for mass conservation. sponding to equations 12.7.18 and 12.7.50 Many
These coupled partial differential equations sophisticated numerical procedures may be em-
can be solved for the temperature and composi- ployed, but they are more properly treated in
tion at any point in the catalyst bed by using textbooks dealing with numerical methods or
numerical procedures to solve the correspon- more advanced texts in chemical reactor design.
12.8 Design of Fluidized Bed Catalytic Reactors 521
For a general introduction to some of the tech- extent of this interchange between the gas bub-
niques employed to solve difference equations of bles and the remainder of the bed has obvious
the types encountered in chemical engineering, implications with respect to the degree of con-
consult the text by Lapidus (116). Detailed version that can be expected in a fluidized bed
numerical examples of one method of numerical reactor. The bubbles are essentially empty, al-
solution to the two-dimensional reactor problem though from time to time a quantity of particles
are contained in the texts of Smith (117) and may rain through the bubble from the roof. In
Jensen and Jeffreys (118). some cases, this movement may cause the bubble
to split. The lowest third of a sphere enclosing
the bubble contains a wake of particles that
12.8 DESIGN OF FLUIDIZED BED
travel upward with the bubble. Solid particles
CATALYTIC REACTORS
also move in streamlines around each rising bub-
The design of a commercial scale fluidized bed ble, so that a spout of material is also drawn up
reactor for a new process is in many respects one behind the bubble. The solids moving upward
of the most challenging technical assignments in the wake behind the bubble and in the spout
that a practicing chemical engineer can be as- below provide rapid mixing of solids from bot-
signed. Even a superficial study of an operating tom to top.
unit reveals its intrinsic complexity. It is obvious These factors combine to make it extremely
that such systems pose tremendous analytical difficult not only to develop predictive analytical
problems for someone trying to develop realistic models of the performance of such systems, but
mathematical models of the interactions be- also to scale up experimental data from labora-
tween the chemical reaction phenomena and the tory operations, particularly when the commer-
various physical transport processes that occur cial scale equipment may contain baffles, heat
in these reactors. Not only do we have to worry exchangers and other internal fixtures within
about the possibilities of intraparticle and exter- the bed. Some useful models have been devel-
nal mass and heat transfer limitations of the oped for fluidized beds, but the design of indus-
types discussed previously, but we must also be trial scale reactors is generally dominated by
concerned with the complexity of the fluid-solid empirical correlations and component designs
contacting process and the manner in which the that have proven useful in past generations of
gas and solid catalyst are distributed throughout equipment. This feature is particularly true of the
the reactor. equipment used for distribution of the incoming
Within a fluidized bed, low density regions or gas. This component has a marked effect on the
"gas bubbles" are formed at apparently unpre- solids recirculation pattern and on potential
dictable rates. These bubbles may subsequently channeling through the bed.
coalesce, split, or grow, perhaps even to the The velocity at which gas flows through the
extent that their size approaches the physical dense phase corresponds approximately to the
dimensions of the reactor. Indeed, in some cases velocity that produces incipient fluidization. The
it is necessary to provide for reactor internals to bubbles rise, however, at a rate that is nearly an
limit bubble size and the problems of slugging, order of magnitude greater than the minimum
poor contacting, and mechanical vibration, fluidization velocity. In effect, then, as a con-
which become serious when the bubbles become sequence of the movement of solids within the
too large. Experimental studies clearly demon- bed and the interchange of fluid between the
strate that there is significant movement of gas bubbles and the dense regions of the bed, there
from the high density regions through the bub- are wide disparities in the residence times of
bles and back to the high density regions. The various fluid elements within the reactor and in
522 Reactor Design for Heterogeneous Catalytic Reactions
the times that the fluid elements are effectively ficient to permit a prediction of degrees of con-
in contact with the catalyst particles. version, since there is not an equal distribution
From the standpoint of attempting to develop of catalyst between the bubble and dense phases.
mathematical models for the simulation of It is the presence of bubbles and their effect on
fluidized bed reactors, one must determine if the gas/solid contacting and mixing that lies at the
phenomena mentioned above and other aspects root of the problem and that causes the discrep-
of the behavior of fluidized beds can be described ancy between the residence time distribution
in terms of a relatively small number of para- (easily measured) and the contact time distribu-
meters. In particular, those aspects of the be- tion (not easily measured).
havior of fluidized beds that have a significant Because of the inadequacies of the aforemen-
influence on the conversion achieved must be tioned models, a number of papers in the 1950s
adequately characterized by the model if it is to and 1960s developed alternative mathematical
be useful for purposes of reactor design. Many descriptions of fluidized beds that explicitly di-
models, especially if they are sufficiently com- vided the reactor contents into two phases, a
plex and contain many "constants," can describe bubble phase and an emulsion or dense phase.
a given set of data equally well. However, if the The bubble or lean phase is presumed to be
models are to be useful for design purposes, it essentially free of solids so that little, if any,
should be possible to determine the parameters reaction occurs in this portion of the bed. Re-
appearing therein from simple laboratory experi- action takes place within the dense phase, where
ments or from correlations of experimental data. virtually all of the solid catalyst particles are
Indeed, the utility of the model is judged by the found. This phase may also be referred to as a
success obtained when extrapolating data from particulate phase, an interstitial phase, or an
one set of experimental reaction conditions emulsion phase by various authors. Figure 12.19
(usually bench or pilot scale experiments) to is a schematic representation of two phase mod-
predict the behavior of a large-scale unit. els of fluidized beds. Some models also define a
The physical situation in a fluidized bed reac- cloud phase as the region of space surrounding
tor is obviously too complicated to be modeled the bubble that acts as a source and a sink for gas
by an ideal plug flow reactor or an ideal stirred exchange with the bubble.
tank reactor although, under certain conditions, On the basis of different assumptions about
either of these ideal models may provide a fair the nature of the fluid and solid flow within each
representation of the behavior of a fluidized bed phase and between phases as well as about the
reactor. In other cases, the behavior of the system extent of mixing within each phase, it is possible
can be characterized as plug flow modified by to develop many different mathematical models
longitudinal dispersion, and the unidimensional of the two phase type. Pyle (119), Rowe (120),
pseudo homogeneous model (Section 12.7.2.1) and Grace (121) have critically reviewed models
can be employed to describe the fluidized bed of these types. Treatment of these models is
reactor. As an alternative, a cascade of CSTR's clearly beyond the scope of this text. In many
(Section 11.1.3.2) may be used to model the cases insufficient data exist to provide critical
fluidized bed reactor. Unfortunately, none of tests of model validity. This situation is especially
these models provides an adequate representa- true of large scale reactors that are the systems
tion of reaction behavior in fluidized beds, par- of greatest interest from industry's point of view.
ticularly when there is appreciable bubble The student should understand, however, that
formation within the bed. This situation arises there is an ongoing effort to develop mathe-
mainly because a knowledge of the residence matical models of fluidized bed reactors that
time distribution of the gas in the bed is insuf- will be useful for design purposes. Our current
Literature Citations 523
28. Weisz, P. B., and Prater, C. D., Adv. Catalysis, 6 (143), 56. Schilson, R. E., and Amundson, N. R., Chem. Eng.
1954. Sci., 13, pp. 226, 237, 1961.
29. Wicke, E., and Kallenbach, R., Kolloid Z., 97 (135), 57. Beek, J., AIChEJ., 7(337), 1961.
1941. 58. Tinkler, J. D., and Pigford, R. L., Chem. Eng. Sci.,
30. Wicke, E., and Brotz, W., Chem.-Ing.-Tech., 21 (219), 75(326), 1961.
1949. 59. Petersen, E. E., Chem. Eng. Sci., 17 (987), 1962.
31. Wicke, E, and Hedden, K., Z. Elektrochem., 57 (636), 60. Tinkler, J. D., and Metzner, A. B., Ind. Eng. Chem.,
1953. 53(663), 1961.
32. Wicke, E., Chem.-Ing.-Tech., 29 (305), 1957. 61. Weisz, P. B., and Hicks, J. S., Chem. Eng. Sci., 17
33. Aris, R., Chem. Eng. Sci., 6 (262), 1957. (265), 1962.
34. Aris, R., Ind. Eng. Chem., Fundamentals, 4 (227), 1965. 62. Maymo, J. A., Cunningham, R. E., and Smith, J. M.,
35. Aris, R., Ind. Eng. Chem., Fundamentals, 4 (487), Ind. Eng. Chem., Fundamentals, 5 (280), 1966.
1965. 63. Cunningham, R. A., Carberry, J. J., and Smith, J. M.,
36. Aris, R., The Mathematical Theory of Diffusion and AIChEJ., 77(636), 1965.
Reaction in Permeable Catalysts, Volumes I and II, 64. Wheeler, A., Adv. Catalysis, 3 (249), copyright © 1951.
Clarendon Press, Oxford, 1975. Used with permission of A. Wheeler and Academic
37. Corrigan, T. E., Garver, J. C , Rase, H. F., and Kirk, Press.
R. S., Chem. Eng. Prog., 49 (603), 1953.
65. Wheeler, A., Catalysis, 2 (105), edited by P. H.Emmett,
38. Wheeler, A., op. cit., p. 136, 1955. Reinhold, New York, 1955. Adapted from the contri-
39. Satterfield, C. N., op. cit., p. 152, 1970. bution of A. Wheeler to Catalysis, 2, edited by P. H.
Emmett. Copyright © 1955 by Litton Educational
40. Satterfield, C. N., op. cit., p. 134, 1970.
Publishing, Inc. Adaptation with permission of Van
41. Satterfield, C. N., op. cit., pp. 135-138, 1970. Nostrand Reinhold Company.
42. Luss, D., and Amundson, N. R., AIChEJ., 13 (759), 66. Smith, J. M., op. cit., pp. 457-462.
1967.
67. Carberry, J. J., Chem. Eng. Sci., 77(675), 1962.
43. Aris, R., Introduction to the Analysis of Chemical
68. Ostergaard, K., Ada, Chem. Scand., 15 (2037), 1961.
Reactors, Prentice-Hall, Englewood Cliffs, N.J., 1965.
69. Butt, J. B., Chem. Eng. Sci., 21 (275), 1966.
44. Petersen, E. E., Chemical Reaction Analysis, p. 64,
Prentice-Hall, Englewood Cliffs, N.J., 1965. 70. Hutchings, J., and Carberry, J. J., AIChE J., 12 (20),
45. Weekman, V. W., Jr., and Goring, R. L., J. Catalysis, 1966.
4(260), 1965. 71. Petersen, E. E., Chemical Reaction Analysis, pp. 129-
164, Prentice-Hall, Englewood Cliffs, N.J., 1965.
46. Weekman, V. W., Jr., J. Catalysis, 5 (44), 1966.
47. Chu, C , and Hougen, O. A., Chem. Eng. Sci., 77(167), 72. Treybal, R. E., Mass Transfer Operations, Second
1962. Edition, McGraw-Hill, New York, 1968.
48. Roberts, G. W., and Satterfield, C. N., Ind. Eng. 73. McCabe, W. L., and Smith, J. C , Unit Operations of
Chem., Fundamentals, 4 (288), 1965. Chemical Engineering, McGraw-Hill, New York, 1967.
49. Roberts, G. W., and Satterfield, C. N., Ind. Eng. 74. Bird, R. B., Stewart, W. E., and Lightfoot, E. N.,
Chem., Fundamentals, 5 (317), 1966. Transport Phenomena, Wiley, New York, 1960.
50. Knudsen, C. W., Roberts, G. W., and Satterfield, 75. Chilton, T. H., and Colburn, A. P., Ind. Eng. Chem., 26
C. N., Ind. Eng. Chem., Fundamentals, 5 (325), 1966. (1183), 1934.
51. Satterfield, C. N., op. cit., pp. 176-200, 1970. 76. Colburn, A. P., Trans. AIChE, 29 (174), 1933.
52. Kao, H. S-P., and Satterfield, C. N., Ind. Eng. Chem., 77. Acetis, J. and Thodos, G., Ind. Eng. Chem., 52 (1003),
Fundamentals, 7(664), 1968. 1960.
53. Woodside, W., and Messmer, J. H., J. Appl. Phys., 32 78. Petrovic, L. J., and Thodos, G., Ind. Eng. Chem.,
(1688), 1961. Fundamentals, 7(274), 1968.
54. Butt, J. B., AIChEJ., 11 (106), 1965. 79. Wilson, E. J., and Geankoplis, C. J., Ind. Eng. Chem.,
55. Damkohler, G., Z. Phys. Chem., A193 (16), 1943. Fundamentals, 5 (9), 1966.
Problems 525
80. Satterfield, C. N., Mass Transfer in Heterogeneous 104. Calderbank, P. H., and Pogorski, L. A., Trans. Inst.
Catalysis, pp. 79-97, MIT Press, Cambridge, 1970. Chem. Engrs. (London) 35 (195), 1957.
81. Chu, J. C, Kalil, J., and Wetteroth, W.A., Chem. Eng. 105. Leva, M., Ind. Eng. Chem., 42 (2498), 1950.
Prog.,49{\4\), 1953. 106. Argo, W. B., and Smith, J. M., Chem. Eng. Prog., 49
82. Riccetti, R. E., and Thodos, G., AIChE J., 7 (442), (443), 1953.
1961. 107. Smith, J. M., loc. cit., pp. 513-522.
83. Sen Gupta, A., and Thodos, G., Chem. Eng. Prog., 58 108. Beek, J., Adv. Chem. Eng., 3, pp. 229-230, 1962.
(7), p. 58, 1962.
109. Carberry, J. J., Wendel, M., AIChE J., 9 (132), 1963.
84. Kunii, D., and Levenspiel, O., loc. cit., Chapter 7.
110. Raymond, L. R., Amundson, N. R., Can. J. Chem.
85. Beek, W. J., "Mass Transfer in Fluidized Beds," in Eng.,42(\73), 1964.
Fluidization by J. F. Davidson and D. Harrison,
Academic Press, New York, 1971. 111. Luss, D., Amundson, N. R., Chem. Eng. Sci., 22 (253),
1967.
86. Olson, R. W., Schuler, R. W., and Smith, J. M., Chem.
Eng. Prog., 46 (614), 1950. 112. Luss, D., Chem. Eng. Sci., 23 (1249), 1968.
87. Reid, R. C , and Sherwood, T. K., The Properties of 113. Hlavacek, V., Hofmann, H., Chem. Eng. Sci., 25
Gases and Liquids, Second Edition, McGraw-Hill, pp. 173, 187, 1970.
New York, 1966. 114. Calderbank, P. H., Chem. Eng. Prog., 49 (585), 1953.
88. Yoshida, F., Ramaswami, D., and Hougen, O. A., 115. Froment, G. F., Ind. Eng. Chem., 59 (18), 1967.
AIChE J., 8 {5), 1962.
116. Lapidus, L., Digital Computation for Chemical Engi-
89. Argo, W. B., and Smith, J. M., Chem. Eng. Prog., neers, McGraw-Hill, New York, 1960.
49(1443), 1953. 117. Smith, J. M., op. cit., pp. 536-547.
90. Stewart, W. E., AIChE J., 9 (528), 1963.
118. Jenson, V. G., and Jeffreys, G. V., Mathematical
91. Sherwood, T. K., Trans. AIChE, 39 (583), 1943. Methods in Chemical Engineering, pp. 412-428, Aca-
92. Kettenring, K. N., Manderfield, E. L., and Smith, demic Press, New York, 1963.
J. M., Chem. Eng. Prog., 46 (139), 1950. 119. Pyle, D. L., in "Chemical Reaction Engineering,"
93. Smith, J. M., loc. cit., Chapters 10-12. ACS Adv. Chem. Series, 109, p. 106, 1972.
94. Froment, G. F., "Analysis and Design of Fixed Bed 120. Rowe, P. N., Fifth European Symposium on Chemical
Catalytic Reactors," in Chemical Reaction Engineering, Reaction Engineering, Elsevier, Amsterdam, 1972.
Adv. Chem. Series, 109 (1), American Chemical
121. Grace, J. R., AIChE Symposium Series, 67 (116),
Society, Washington, D.C., 1972.
p. 159, 1971.
95. Froment, G. F., Fifth European Symposium on
122. Kunii, D., and Levenspiel, O., Fluidization Engineer-
Chemical Reaction Engineering, Elsevier, Amsterdam,
ing, Wiley, New York, 1969.
1972.
123. Davidson, J. F., and Harrison, D., Fluidized Particles,
96. Bird, R. B., Stewart, W. E., and Lightfoot, E. N.,
Cambridge University Press, London, 1963.
Transport Phenomena, Wiley, New York, 1960.
97. Perry, R. H., and Chilton, C. H., Chemical Engineer's
Handbook, Fifth Edition, McGraw-Hill, New York, PROBLEMS
1973.
98. Ergun, S., Chem. Eng. Prog., 48 (93), 1952. 1. Thiophene (C 4 H 4 S) is representative of the
99. Wehner, J. F., and Wilhelm, R. H., Chem. Eng. Sci., organic sulfur compounds that are hydro-
6(89), 1956. genated in the commercial hydrtodesulfurization
100. Smith, J. M., loc. cit., p. 504. of petroleum naphtha. Estimate both the com-
101. Aris, R., and Amundson, N. R., AIChE J., 3 (280), bined and effective diffusivities for thiophene in
1957. hydrogen at 660 °K and 3.04 MPa in a catalyst
102. Jakob, M., Heat Transfer, p. 553, Wiley, New York, with a BET surface area of 168 m 2 /g, a porosity
1957. of 0.40, and an apparent pellet density of
103. Colburn, A. P., Ind. Eng. Chem., 23 (910), 1931. 1.40 g/cm3. A narrow pore sized distribution
526 Reactor Design for Heterogeneous Catalytic Reactions
The entire apparatus was immersed in a When the hydrogen pressure is 1 atm, and the
constant temperature bath at 25 °C. The up- temperature is 77 °K, the experimentally ob-
stream hydrogen pressure was maintained con- served (apparent) rate constant is 0.159 cm 3 /
stant at 77.00 cm Hg. The time variation of the sec-g catalyst. Determine the mean pore radius,
pressure in the constant volume container is the effective diffusivity of hydrogen, and the
given in the next column. catalyst effectiveness factor.
Problems 527
4. A catalyst for cracking cumene is available centrations that the reaction may be regarded
commercially in the form of 0.35 cm diameter as pseudo first-order in both directions.
pellets. These pellets have a specific surface area Determine the effectiveness factor for the ion
of 420 m 2 /g and a void volume of 0.42 cm 3 /g. If exchange resin at 85 °C, assuming that the
the apparent first-order rate constant for this reaction is reversible even though the authors
reaction is 1.49 cm3/sec-g catalyst at 412 °C, presumed the reaction to be irreversible in
determine the effectiveness factor of the catalyst. reporting their data. They note that at 100 °C
the equilibrium for the reaction corresponds to
5. Barnett et al. [AIChE J., 7 (211), 1961] have a conversion greater than 94%. If the equilibrium
studied the catalytic dehydrogenation of cyclo- constant for the reaction is expressed as the
hexane to benzene over a platinum-on-alumina ratio of the £-butanol concentration to the
catalyst. A 4 to 1 mole ratio of hydrogen to isobutylene concentration and corrected for the
cyclohexane was used to minimize carbon temperature change in going from 100 °C to
formation on the catalyst. Studies were made 85 °C, a value of 16.6 may be considered appro-
in an isothermal, continuous flow reactor. The priate for use.
results of one run on 0.32 cm diameter catalyst The ion exchange particles may be regarded
pellets are given below. as isothermal, and the effective diffusivity of
Temperature, 705 °K isobutylene within the particles may be taken
Pressure, 1.480 MPa as 2.0 x 10~ 5 cm2/sec. The resin particles may
H 2 feed rate, 8 moles/ksec be considered as spheres with radii equal to
Cyclohexane feed rate, 2 moles/ksec 0.0213 cm. The density of the swollen resin is
Conversion of cyclohexane, 15.5% assumed to be equal to 1.0 g/cm3.
Quantity of catalyst, 10.4 g Rate measurements on these particles indicate
Catalyst Properties: that at 85 °C the rate is equal to 1.11 x 10" 2
Pore volume, 0.48 cm 3 /g moles/ksec-g catalyst, when the conversion
Surface area, 240 m 2 /g achieved is only 3.9%. From thermodynamic
Pellet density, 1.332 g/cm3 equilibrium data contained within the article,
Pellet porosity, 0.59 cm 3 voids/cm3 it is estimated that under these conditions, the
isobutylene concentration in the reactor at the
If the effectiveness factor of the catalyst is exterior surface of the resin may be taken as
known to be 0.42, estimate the tortuosity factor (0.961)(1.72 x 10" 5 ) moles/cm3.
of the catalyst assuming that the reaction obeys
first-order kinetics and that Knudsen diffusion 7. A well-insulated pilot scale packed bed
is the dominant mode of molecular transport. reactor (5 m long by 5 cm diameter) is being
used to carry out an irreversible reaction of the
6. Gupta and Douglas [AIChE J., 13 (883), type
1967] have studied the catalytic hydration of M + N -+ R + S
isobutylene to f-butanol, using a cation ex-
All species are gases at the conditions of interest.
change resin catalyst in a stirred tank reactor.
The reaction is quite exothermic (AH =
—120 kJ/mole). Side reactions are unimportant.
(CH- 3 ) 2 C=CH 2 + H 2 O ±^ (CH 3 ) 3 COH There are two different catalyst manufacturers
whose products are being considered for use in
The water is present in such large excess a commercial scale reactor facility. Both pro-
relative to the isobutylene and t-butanol con- ducts are 3.2 mm diameter spherical pellets that
528 Reactor Design for Heterogeneous Catalytic Reactions
are believed to be essentially the same in to 475 C, both catalysts show little loss of
chemical composition. activity with time onstream and both catalysts
The pilot plant results indicate that the de- give essentially the same yield. (It is, however,
activation characteristics of the two catalysts significantly lower than the yield obtained
are somewhat different. The same gas pressure, with the 550 °C feed temperature.) The deactiva-
feed composition, and molal feed rate are tion behavior of both catalysts is very repro-
employed in all cases. If the inlet temperature ducible.
is 550 °C, the behavior indicated in Figure 12P.2 (a) How do you interpret these data?
is observed. If the feed temperature is reduced (b) What modifications in the manufacturing or
operating conditions do you suggest to
obtain improved performance from catalyst
100 A1
0.01
0.70 0.74 0.78 0.82 0.86 0.90
103/T, °r1
Figure 12P.3
Arrhenius plots for poisoned and unpoisoned catalysts. [From
AlChE J., 7 (211), 1961. Used with permission.]
Problems 529
2. Spherical pellets (1.27 cm diameter) fabricated Note. Engineering estimates of property values
by compressing unreduced granular particles have been used, where necessary, to fill out the
in a steel mold. data.
Pellets of three different densities were studied.
They were obtained by varying the quantity of 11. It is instructive to consider the relative
particles and the pressure used in the molding rates of mass transfer in fixed and fluidized bed
process. Pertinent physical property data for reactors. The rapid rate in the fluidized bed is
the various forms of catalyst used in these due not so much to the high mass transfer
studies are summarized below. coefficients involved, but to the very large
Apparent density
of particle or Equivalent K
Catalyst pellet (g/cm3) diameter (m2/g) (cm3/g) fi
total
Granular
particles 2.14 0.13 mm 90 0.18 0.38
Pellet A 1.16 1.28 cm 90 0.46 0.53
Pellet B 0.92 1.27 cm 90 0.73 0.77
Pellet C 0.72 1.27 cm 90 1.17 0.84
£
totai refers to the ratio of the void volume within and between pellets to the total
volume.
Reaction rate data were reported as a function exterior surface area per unit volume of reactor.
of temperature and are shown in Figure 12P.4. If one assigns the weight of catalyst per unit
Although the form of the intrinsic rate equation volume of reactor the symbol pB, the rate of
for ethylene hydrogenation for this specific mass transfer to the external surface per unit
catalyst is not known, one might anticipate an reactor volume is given by
equation of the form
?MT = kmaJCB - CES)pB = kmav{CB - CES)
(A)
r =
where we have assumed equimolar counter-
diffusion^ = 0) in order to simplify the analysis.
on the basis of other studies of the catalytic The product ampB represents the external surface
hydrogenation of ethylene. What is your inter- area of catalyst per unit volume of bed (av).
pretation of the data shown in Figure 12P.4? For spherical pellets of radius R, this area is
If it is possible to back up your arguments using given by
semiquantitative arguments, do so using rough
numbers. / / multiple interpretations are possible, OmPB =
state them. In the low temperature region the
intrinsic activation energy based on the particle where pp is the apparent density of a catalyst
data is 17.8 kcal/g mole, but at high temperatures pellet. The ratio of the catalyst density within
the apparent activation energy drops to 11.800 the entire bed to that of a single pellet is equal
kcal/g mole. to the ratio of the volume occupied by the
Problems 531
Particles Pellets
Vb = 0.13-0.15
6
04 ft B
<T
^^Particle curve o 0.08 D 0.723
o\ <> 0.11 -a- 0.714
°- 0.16 0s)
0 -o 0.16 <> [• 0.952
* 0.068 -O-J
in- 5 u
9 0.049 6 1190
9 1.160
—k -Pellet curves -
T
I •
in- 6
ri
1 rX_ I*
1 1
--?
*
PelU
Pellet B
\\ i-r
-Pelle
\\
\
\ ^
10"7
2.20 2.40 2.60 2.80 3.00 3.20 3.40 3.60
1000/7, °K~1
Figure 12P.4
Arrhenius plots for particles and pellets. {yb = ethylene mole fraction)
[From AlChE J., 11 (636), 1965. Used with permission.]
pellets themselves to the entire bed volume. typical property values to determine the relative
In terms of the external void fraction of the bed rates of mass transfer in these systems. Mass
(sB), the volume fraction occupied by the pellets velocities for fixed and fluidized beds may be
is equal to 1 — £B. Hence taken as 0.15 and 0.03 g/(cm2-sec), respectively.
Bed void fractions may be taken as 0.30 and
^ = 1- g
PP
0.80 for the fixed and fluidized beds, respective-
Hence ly. The corresponding catalyst sizes may be
taken as 0.5 cm and 0.0063 cm (250 mesh).
= _3 6(1 - eB) These numbers are chosen so as to favor fixed
(B)
bed mass transfer. The reacting fluid may be
regarded as a gas with a viscosity of 3.30 x
Combining equations A and B gives
10~4g/(cm-sec).
_ 6fcw(l - eB) (n
1
MT — ^ \^B ^ES) (Q 12.* The data on page 532 were obtained by
Karpenko in an unpublished study described by
Assuming the same concentration driving
force in both fixed and fluidized beds, use * Adapted with permission from C. N. Satterfield
532 Reactor Design for Heterogeneous Catalytic Reactions
Rylander in Catalytic Hydrogenation Over 13. Zajcew [J. Am. Oil Chem. Soc, 57(11), 1960]
Platinum Metals (p. 39, Academic Press, New has studied the hydrogenation of fatty oils for
York, 1967). Nitrobenzene in ethanol was shortening stock using a palladium catalyst. The
hydrogenated at room temperature and 1 atm experiments were carried out in a 1-gal hydro-
over various amounts of 5% Pd on carbon. genator provided with mechanical agitation, a
Four loading levels of catalyst were used. At gas dispersing system, and heating and cooling
each level, the reduction was carried out in two capability. Hydrogen gas is fed in at the reactor
different types of batch reactor. bottom and is rapidly consumed. Figure 12P.5
1. An equipoise shaker that gives very vigorous indicates the time necessary to achieve a speci-
agitation. fied conversion as a function of catalyst loading
2. A flask stirred by a rotating magnetic bar level. Note that semilog coordinates are em-
that gives relatively poor agitation. ployed. The table below reformulates these data
Each reduction proceeded at nearly constant in terms of the number of iodine units of
rate until the nitrobenzene was nearly exhausted. reduction per minute per 1% of catalyst.
Conditions: 0.02 18
Temperature — 185°C 0.01 40
Pressure - Atmospheric 0.005 72
H 2 Flow-- 4 liters/min
216 - 0.0025 130
Agitation - 620 rpm
•I! 144
122 A Catalyst - 0.5% Pd on carbon 0.00125
0.0005
215
354
II
112-110
100 ! "i! T
i
i i
i i
i
i
i
I
|
I
1
1 What is your interpretation of these data?
i i i i 1 1
14.* Price and Schiewetz \_Ind Eng. Chem.
!o o
; 49 (807), 1957] have studied the catalytic liquid
2
Pd in oil phase hydrogenation of cyclohexene in a lab-
Figure 12P.5
oratory scale semibatch reactor. A supported
Hydrogenation rate as a function of catalyst con- platinum catalyst was suspended in a cyclo-
centration. [Adapted from M. Zajcew, J. Am. Oil hexene solution of the reactant by mechanical
Chem. Soc, 37 (11), 1960. Used by permission of
American Oil Chemists Society.] * Adapted with permission from C. N. Satterfield
Problems 533
agitation of the solution. Hydrogen was bubbled At 25 °C the equilibrium constant for the
through the solution continuously. The reactor reaction
is described in their words in the next paragraph.
'The reactor consisted of a 1-liter three-
necked Morton flask. This flask has four equally
spaced perpendicular indentations about its
periphery and a concave bottom, which aid the is 2 x 108. Consequently, the reaction may be
action of the agitator. A four-bladed stirrer regarded as irreversible. These investigators
driven by a variable-speed motor extended into reported the data indicated on Figures 12P.6 to
the flask through a packing gland in the center 12P.11. Provide an interpretation of each figure
neck. Tubes were sealed into one side neck for individually and then an overall analysis of the
hydrogen delivery, sample withdrawal, and a factors that govern the rate of this reaction.
thermocouple well. An ice water condenser on Where multiple interpretations of an individual
the exit neck minimized solvent escape by figure are possible, indicate them. Notice that in
entrainment or vaporization. Hydrogen was Figure 12P.11, the apparent activation energies
dispersed into the reacting solution by a 20 mm determined from the slopes of the solid lines are
sintered-glass tube extending to the bottom of 4.80 and 12.8 kcal/g mole at 1300 and 2500 rpm,
the flask. The entire assembly was immersed in respectively.
a constant temperature bath maintained at
* This material and the figures on pages 533-536 are
±0.1 °C. However, because of the exothermic reprinted, with permission, from R. H. Price and R. B.
nature of the reaction, a temperature rise of 1 Schiewtz, Ind. Eng. Chem., 49 (807), 1957. Copyright © by
to 2 °C occurred in the reactor.* the American Chemical Society.
Figure 12P.6
Typical concentration-time relationship for experimental runs. Tem-
perature, 26°C; pressure, 746 mm of mercury; hydrogen flow, 30.7 x
10 5 mole/sec; catalyst weight, 0.975 g; stirrer speed, 1100 rpm;
slope, 13.7 x 1 0 5 mole/liter/sec.
534 Reactor Design for Heterogeneous Catalytic Reactions
25 1 1 1 1 1 1 I 1 1 1 1 1 . 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
'1
2 23 -
21 -
1
.
o 19
1
1
"'o
1 1 . 11
.
•S 1 7 -
1 1
15 -
-
13 . . 1 1 , . . 1 11 1 . 1 1 , , , 1 1 1 1 I 1 . . . . 1-
500 550 600 650 700 750 800 850
Absolute pressure (mm mercury)
Figure 12P.7
Effect of pressure on rate of reaction. Temperature, 25-26 °C; hydro-
gen flow, 27.6 x 10 5 mole/sec; catalyst weight, 0.976 g; stirrer
speed, 1500 rpm.
2 3 4 5 6 7 8
Weight of catalyst (g)
Figure 12P.8
Effect of catalyst weight on rate of reaction. Temperature, 25 °C;
pressure, 748 mm of mercury; hydrogen flow, 31.3 x 10 5 mole/sec;
stirrer speed, 1000 rpm.
30 - I ' I • I • I • I • I • I '
I i I i I i I i I i I i I i I i
4 6 8 10 12 14 16 18
Flow rate [(moles/sec) X 10 4 ]
Figure 12P.9
Effect of hydrogen flow rate on rate of reaction. Temperature, 2 5 -
26 °C; pressure, O = 534 mm and # = 741 mm of mercury; catalyst
weight, 0.974 g; stirrer speed, 1500 rpm.
45 1 1 1 1 1 1 1 1 1
1 i 1 1 ' 1 i •
r *—
40 -
< ^
30 -
/
s
IS 25 - -
o 20 -
o 15-
/
10 - /
5 -
• i 1 , 1 1 1 I 1 . i i i i 1 1 I
600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600
Stirrer speed (rpm)
Figure 12P.10
Effect of stirring on rate of reaction. Temperature, 25-27 °C; pressure, 746 mm
of mercury; hydrogen flow, O = 29.6 x 10~5 and • = 81.5 x 10~5 mole/sec;
catalyst weight, 0.976 g.
535
536 Reactor Design for Heterogeneous Catalytic Reactions
316 320 324 328 332 336 340 344 348 352 356
Figure 12P.11
Evaluation of apparent energies of activation. O = 1300 rpm and # = 2500
rpm.
15. Vanadium pentoxide may be used as a at 649 °F. The reaction rate expression pertinent
catalyst for the oxidation of sulfur dioxide to to these experimental conditions is believed to be
sulfur trioxide. of the form:
so2 so3 r =
^SO2^SO2)
In a series of laboratory scale experiments,
streams of oxygen and sulfur dioxide were fed The following data have been reported by
at different rates to a differential reactor con- Mathur and Thodos [Chem. Eng. ScL, 21 (1191),
taining 2.372 g of catalyst. The data below were 1966] and are sufficient for an initial rate
recorded under essentially isothermal conditions analysis of this heterogeneous catalytic process.
(a) Determine the constants k and KSOr the thermal effects accompanying the reaction.
(b) If pure SO 2 and O 2 are both fed to a reactor Consider inlet temperatures from 350 to 480 °F
at flow rates of 100 lb moles/hr, how many and N 2 /C 2 H 4 ratios from 0 to 5.0.
pounds of catalyst will be required for 20% Your analysis will be governed by the
conversion? The initial feed pressure is following constraints.
1 atm. Assume that the aforementioned rate (a) If the temperature at any point in the
expression is valid over the range of vari- reactor exceeds 550 °F, the conditions will be
ables concerned. The reactor may be as- inappropriate in that explosions may occur
sumed to operate isothermally at 649 °F. in this regime.
Set up the integral, but do not attempt to (b) If the temperature decreases as one moves
evaluate it. Make sure that all terms under down the reactor, the reaction must be
the integral are expressed in terms of a regarded as self-extinguishing.
single variable. (c) If the pressure drop exceeds 14 atm, the
calculations must be terminated.
16. Wan [Ind. Eng. Chem., 45 (234), 1953] has Determine the range of satisfactory perfor-
studied the partial oxidation of ethylene to mance and the resultant yields for various
ethylene oxide. reactor lengths for the following operating
specifications.
C2H4 C2H4O
1. Inlet gas pressure = 1 5 atm.
The catalyst consists of silver supported on 2. Inlet C 2 H 4 /O 2 ratio (moles) = 4 : 1 .
alumina and, while it is reasonably specific, 3. Superficial mass velocity = 8000 lb/hr-ft2.
appreciable amounts of CO 2 and H 2 O are also 4. Bulk density of catalyst = 83.5 lb/ft3.
formed. Over the range of interest, the yield of External heat and mass transfer effects are to
ethylene oxide is relatively constant so that for be neglected in your analysis, but you should
present purposes, we may regard the reaction estimate the potential magnitudes of these
stoichiometry as effects.
C 2 H 4 + 1.5O2 -+
Not many problems dealing with the various
0.6C 2 H 4 O + 0.8CO 2 + 0.8H 2 O
fixed bed reactor models are set forth above,
The rate of reaction may be expressed as because I feel that the best means of demon-
strating the features thereof is through case
r = 1.17 x lO6 studies. I find that the principles involved are
where the partial pressures are expressed in clarified by realistic examples involving the
atmospheres design of commercial scale reactors using
machine computation where appropriate. Two
T is expressed in degrees Kelvin brief examples that involve studies of specific
r is expressed in pound moles per pound of process variables are considered in Chapter 13.
catalyst per hour One could employ studies of the effects of other
If 1/8 in. catalyst pellets are packed in 1 in. ID process variables on the reactor performance as
tubes, which in turn are immersed in a liquid additional problems or else state the assignment
bath that maintains the tube walls at 300 °F, in more general terms, as in the problem below.
consider the effects of varying the feed tempera- Space limitations preclude setting forth many
ture and of diluting the feed with N 2 to moderate such problems, but the book Catalytic Processes
538 Reactor Design for Heterogeneous Catalytic Reactions
and Proven Catalysts, by C. L. Thomas, and the 2. Equilibrium yields of butadiene are rela-
descriptions of new and current catalyst tech- tively low. For example, the yield is 35% at
nology that appear frequently in Hydrocarbon 930 °F and 71% at 1110 CF when using a hydro-
Processing provide appropriate starting points carbon feed partial pressure of 0.1 atm.
for the acquisition of the necessary data for In recent years, considerable research effort
problem formulation. Old issues of Industrial has been invested in trying to develop an
and Engineering Chemistry and the more recent oxidative process to accomplish the desired
quarterlies evolving therefrom are also good transformation. Several catalyst formulations
sources of essential data on catalyst activity. with good selectivities for the desired reaction
The literature of various catalyst manufacturers have been developed, and you are asked to
provides information on the physical properties determine the desirability of using one of these
of general catalyst types when such information formulations in a fixed bed reactor configura-
is not included with the activity data. As tion. Several advantages are claimed for the
stated, the problems normally require literature oxidative processes:
searches, and the students usually find tt\at 1. The constraint of thermodynamic equilib-
information necessary to complete the design is rium for the butene dehydrogenation reaction
often missing and must be estimated using is effectively removed since hydrogen is con-
good engineering judgment. verted to water by oxidation. Equilibrium yields
then approach 100% over the complete tempera-
17. Oxidative Dehydrogenation of Butene to ture and partial pressure range of interest.
Butadiene 2. Selectivities are high, typically in excess of
UW Chemical Corporation 90%.
Madison, Wisconsin 3. Longer times on-stream are facilitated by
To: Student Team "Blue" the presence of oxygen in the feed. It inhibits
From: Big Red, Group Leader carbon buildup on the catalyst, thereby per-
Re: Butadiene production mitting one to operate for months without
A proposed expansion of the company's regeneration.
styrene-butadiene rubber production will re- 4. Input energy requirements for the process
quire an additional 10,000 tons/year of buta- are significantly reduced since the energy re-
diene as a raw material. leased by the exothermic oxidation reactions
For many years, butadiene has been manu- serves as a driving force for the endothermic
factured by dehydrogenating butene or butane dehydrogenation reaction.
over a catalyst at appropriate combinations of For preliminary discussions of the proposed
temperature and pressure. It is customary to expansion program, it is desirable to determine
dilute the butene feed with steam (10-20 moles the basic equipment requirements, although a
H 2 O/mole butene) to stabilize the temperature detailed economic evaluation is not essential.
during the endothermic reaction and to help An article by Sterrett and Mcllvried [Ind. Eng.
shift the equilibrium conversion in the desired Chem., Process Des. Develop., 13 (54), 1974]
direction by reducing the partial pressures of describes the use of a ferrite catalyst for the
hydrogen and butadiene. The current processes desired application. It contains kinetic data
suffer from two major disadvantages. for this catalyst and several references to other
1. Catalyst on-stream periods are short, since oxidative dehydrogenation catalysts. Select a
coke builds up rapidly on the catalyst. Hence promising catalyst material and prepare a
catalyst regeneration must take place at frequent reactor design proposal using a fixed bed
intervals. configuration. You are asked to consider factors
Problems 539
540
13.1 Pyrolysis of a Propane Feedstock 541
(hr-ft2 of external tube surface). You may also Rase (1) were particularly useful in arriving at
assume that the feed enters the reactor at 1100° F the numbers cited below.
and 60 psia. In order to maintain flow through
the downstream separation train the reactor Reaction Stoichiometry and Thermodynamic
effluent cannot be at a pressure lower than Data. The product distribution for a propane
25 psia. Propane conversion should be in the charge is shown in Figure 13.1. The reaction is
neighborhood of 85%. You are asked to prepare not clean even at low conversions. At higher
a preliminary reactor design for this system. Use conversions secondary reactions of the primary
the data tabulated below to determine length products become more and more significant.
requirements and general suitability for 2, 4, However, it is reasonable to assume that the
and 6 in. nominal pipe sizes (schedule 40). original reactants continue to decompose along
the path of the original mechanism, and first-
Data Summary. There is an extensive litera- order kinetics is observed to quite high conver-
ture dealing with hydrocarbon pyrolysis reac- sions. In the presence of established information
tions. The articles by Schutt (2) and Fair and on the product distribution as the conversion
-CH 4
+_, o
"x a
"8g 40 -
28
o
E
20 I—
10 —
20 40 50 60 70 80 90
Percent conversion of propane
Figure 13.1
Product distribution for propane pyrolysis. [From Schutt, Chemical
Engineering Progress, 50 (415), 1954. Used with permission.]
542 Illustrative Problems in Reactor Design
increases, the establishment of rate expressions pressed in degrees Kelvin and in calories per
for the secondary and tertiary reactions is not gram mole degree Kelvin.
essential from the standpoint of reactor design.
For present purposes we may consider the re- Viscosity Data. Our physical properties expert
action stoichiometry to be given by indicates that the following expression provides
a good representation of the mixture viscosity
C 3 H 8 -> 0.300C 3 H 6 + 0.065C2H6 0.6675C2H4
data over the temperature, pressure, and com-
+ 0.635CH4 + 0.300H2 (13.1.1) position range of interest.
The volumetric expansion parameter S may thus //(centipoise) =
be taken as 0.9675. The product distribution 10"4(2.37 + 0.708/-0.411/ 2 )T 2 / 3
will vary somewhat with temperature, but the
(13.1.8)
stoichiometry indicated above is sufficient for
preliminary design purposes. (We should also where
indicate that if one's primary goal is the produc-
/ is the fraction propane converted
tion of ethylene, the obvious thing to do is to
recycle the propylene and ethane and any un- T is expressed in degrees Kelvin
reacted propane after separation from the lighter [Note. Kinetic theory predicts that ft should be
components. In such cases the reactor feed proportional to T 1/2 , but the T 2 / 3 dependence
would consist of a mixture of propane, propyl- provides a better fit of the results calculated
ene, and ethane, and the design analysis that from methods outlined in Reid and Sherwood
we will present would have to be modified. For (3).]
our purposes, however, the use of a mixed feed
would involve significantly more computation 13.1.2 Reactor Design Analysis
without serving sufficient educational purpose.)
The following heat capacity data may be A preliminary design analysis for the proposed
employed for the temperature range of interest. reactor may be carried out in terms of the plug
(Temperatures must be expressed in degrees flow model for a nonisothermal system. If the
Kelvin.) At 1100 °F the standard heat of reac- flow conditions lie in the turbulent regime, the
tion is equal to 21.960 kcal/g mole. assumption of linear velocity profiles will be
reasonable and it will be appropriate to neglect
radial variations in velocity, temperature, and
Heat capacity
Compound Cp (cal/g mole-°K)
composition. These assumptions are in essence
the same as those employed in Section 10.4, and
C3H8 21.14 + 0.020567 (13.1.2) the discussion contained therein will be the
C3H6 17.88 + 0.016457 (13.1.3) basis for the development of our numerical
C2H6 13.34 + 0.015897 (13.1.4) results. However, we have added the additional
C2H4 12.29 + 0.010227 (13.1.5) complication of varying the pressure along the
CH 4 6.98 + 0.010127 (13.1.6) length of the reactor. This aspect of the problem
H2 6.42 + 0.000827 (13.1.7) is taken into account by breaking the reactor
up into a number of segments, each of which is
Kinetic Data. The pyrolysis reaction obeys considered to operate at a constant pressure.
first-order kinetics with a rate constant equal to In order to characterize the flow regime and
3.98 x l O 1 2 * - 5 9 ' 1 0 0 ^ 7 ^ - 1 , where T is ex- thereby justify the plug flow assumption, it is
13.1 Pyrolysis of a Propane Feedstock 543
appropriate to calculate the Reynolds number It is convenient to write the first-order reac-
tion rate expression as
DG
(13.1.9)
+SAfA TJ\PO
where D is the tube diameter.
where CA0 is the propane concentration at zero
Now
fraction conversion and the reactor inlet tem-
7000 8913 perature (T o ), and inlet pressure (P o ). (See
2 (13.1.10)
nD D2 equation 3.1.44.) Ideal gas behavior has been
assumed, but this is quite appropriate for these
nonpolar gases at the high temperatures and
where low pressures of interest.
Equations 13.1.13 and 13.1.14 may be com-
D is measured in feet
bined to give
G is measured in pounds mass per hour per
square foot VR (Vou. (1 +dAfA)TPodfA
(13.1.15)
% ' J/in k(l - fA)T0P
Thus
or
8913
(13.1.11) (1 + SAfA)TP0 dfA
(13.1.16)
j/in /c(l - fA)T0P
where ji must be expressed in pounds mass per In order to determine the required reactor
foot per hour. volume one must relate the temperature (and
At the reactor inlet (1100 °F or 866.49 °K) thus k) and the local pressure P to the fraction
conversion using an energy balance and con-
\i = 2.15 x 10" 2 cp =o= 5.21 x 10" 2 lbm/(ft-hr)
ventional fluid flow equations.
Thus For thermal cracking units, the majority of
the required energy input is supplied by radiant
8913 1.71 x 105 heat transfer in a furnace. In such cases one finds
(13.1.12)
5.21 x 10" 2 D D it convenient to express the rate at which thermal
energy is supplied to the reactor as the product
Since all of the tube diameters of interest are of an energy flux and the external surface area
less than 1 ft, it is evident that in all three cases of the tube. If we represent the flux by q and a
the flow will be highly turbulent and assumption segment of the reactor length by AL, then the
of plug flow conditions will be quite appropriate. total rate at which thermal energy is supplied
For steady-state operation of a plug flow (SQ) is given by
reactor the basic design equation (equation 8.2.9)
3Q = qnD0 AL (13.1.17)
can be written as
where Do is the outside diameter of the pipe.
o Ul dfA This expression may be substituted for the left
= C (13.1.13)
-rA side of equation 10.4.6 to give the following
representation of an energy balance on a differ-
where A refers to propane. ential element of reactor length.
544 Illustrative Problems in Reactor Design
1.0
Material k* ft
Drc wn tubing Smooth
Wrc>ught iron, stee I 0.00015
Gal vanized iron 0.0005
Cast iro n 0.0008
Cor lcret e 0.01-(1.00 1
0.1 \
^
s
s
k*/DT
).004 )o: c.00 1
0.01 \ / ^0.
£f^0.00 oeJ 0. 300^
55
— 9
_ _ ./ / /
sEli
! ^
Smooth -
1s — a
• • •• • m m
••» • m
0.00001 •••
a..
c .00005^ % 1 0002
0.001
102 10 3 10 4 105
I1 106
D.C)0 )1 I I
107
Reynolds number N R e
Figure 13.2
Friction-factor chart. (From Unit Operations of Chemical Engineering by W. L. McCabe and J. C. Smith.
Copyright c 1967. Used with permission of McGraw-Hill Book Company.)
Table 13.1
Potential Range of Viscosities and Friction Factors
=
/ 0 / = 0.8 / = 0 / = 0.8
Viscosity (cp) 2.15 x 10" 2 2.43 x 10" 2 2.51 x 10" 2 2.83 x 10" 2
N Re for 2-in. pipe 9.94 x 105 8.80 x 105 8.52 x 105 7.56 x 105
f"* for 2-in. pipe 0.0050 0.0050 0.0050 0.0051
N Re for 4-in. pipe 5.11 x 105 4.52 x 105 4.37 x 105 3.88 x 105
¥
for 4-in. pipe 0.0043 0.0044 0.0044 0.0045
rN for 6-in. pipe
Re 3.39 x 105 3.00 x 105 2.90 x 105 2.57 x 105
P* for 6-in. pipe 0.0042 0.0044 0.0044 0.0045
13.1 Pyrolysis of a Propane Feedstock 545
1 onl - fAin)
qnD0 AL = ^ J ^ ' " Cpi d ^ ' ™ Cpi
(13.1.18)
where the first summation involves the molal of the tabular entries indicates that the friction
flow rates of the various species leaving the factor does not vary much with either tempera-
volume element and the second involves the ture or fraction conversion and that an assump-
molal flow rates of the various species entering tion of constant / * for a given pipe size would
the volume element. The datum for the enthalpy be entirely appropriate. We will u s e / * = 0.0050
calculations has been taken as the reactor inlet for the 2-in. pipe and / * = 0.0044 for the 4-
temperature. and 6-in. pipes. Since the fluid density varies
The pressure drop through a tubular reactor along the length of the reactor as the pressure
of length AL may be expressed in terms of the and the temperature change, the local density
friction factor / * as must be used in equation 13.1.19 to evaluate the
2/* A L Q 2 ) pressure drop over the length segment.
AP
(13.1.19) Equations 13.1.16, 13.1.18, and 13.1.19 must
P be solved simultaneously in order to determine
where p is the mass density of the fluid and the the required reactor length for the various pipe
gc factor is introduced in order to make the sizes. We now wish to rewrite these equations in
units consistent. forms that are appropriate for solution using a
The friction factor may be calculated from a finite difference approach. In the present case
knowledge of the fluid Reynolds number using we will find it convenient to consider the reactor
the chart shown as Figure 13.2. Table 13.1 indi- as consisting of a series of elements of length
cates the friction factors that correspond to zero AL and employ these equations by incrementing
and 80% conversion for the three pipe sizes of the length. The following mole table may be
interest at two temperature levels. Examination prepared using stoichiometric principles.
Effluent
Species molal flow rate
Now, if we let
A
Lout = Lin + L (13.1.20)
the energy balance (equation 13.1.18) can be written as
_F ^^AfA (13.1.21)
vA
Since AHRaiTo = 21,960 cal/g mole and the stoichiometric coefficient vA = - 1 , the last equation
can be written as
AL
^ ° = fTom (21.14 + 0.02056T) dT - fAin fT°ut(0.35 + 0.00109T) dT
FAQ JTin jTi"
where one may recognize the term in brackets as the heat of reaction at the temperature of the fluid
leaving the length element. In this equation the units of T must be degrees Kelvin, and each term
must have the units of calories per gram mole.
Now
FA0 = 7
°^(Q954) = 7.208 x 104 g moles/hr
and
q = 104 x 252 = 2.52 x 106 cal/hr-ft2
Thus, if Do is measured in inches and AL in feet,
In order to provide a periodic check on the where P is measured in pounds per square inch
energy balance equation, one may use the form (absolute). Thus
of equation 13.1.18, which results from integra-
tion between the reactor inlet and a distance L (1 + 0.9615fA)TdfA
downstream where the fraction conversion is 0
1.040 x
J A out •
(13.1.30)
R a t TQJA out
= 1.562 x 1 0 2 ( 1 + 5 A/A)
D 2 ~ (13.1.33)
nD2L
4(144)(12.30) -st 3.98 x
(1 + 0.9675/4)7(14.7) dfA
(13.1.29)
548 Illustrative Problems in Reactor Design
The mass density p may be determined from Inspection of this equation indicates that the
the ideal gas law. pressure drop will be quite sensitive to the pipe
diameter employed. To minimize the pressure
P - drop, large tubes are desired, but the larger the
where pipe, the lower the heat transfer area per unit
M = 2>M,. (13.1.35) volume. The three specified pipe sizes have been
The various mole fractions may be determined chosen to illustrate the optimization problem.
from the ratio of individual effluent flow rates We are now prepared to carry out the calcula-
to the total effluent flow rate. (The necessary tions for a single length of each of these pipe
data was tabulated previously for use in analysis sizes. For hand calculations we have the neces-
of the heat capacity terms.) Thus sary relations in equations 13.1.24, 13.1.29, and
M =
.016)J
+ 0.6675/4(28.05) + 0.635/4(16.04) + 0.300/4(2.016)
(1 + 0.9675L
44.09 + 2.12 x l
(13.1.36)
1 + 0.9675/4
or within the accuracy of the data. 13.1.39. One manner in which we may proceed
44.09 for each size diameter is as follows.
M = (13.1.37)
1 + 0.9675/4 A. Choose a length increment AL.
Hence B. Assume a gas temperature at a distance L
P(44.09) from the reactor inlet and calculate an
9 = average gas temperature for the increment.
0.9675/4)(10.73)(1.8)T
C. Assume an average gas pressure for the
2.283P increment.
( 1 1 U 8 )
(1 + 0.9675fA)T D. Determine the increment in the fraction
Combination of equations 13.1.19, 13.1.33, conversion within the length element by
and 13.1.38 and conversion of AP to pounds using a finite difference form of equation
per square inch gives 13.1.30.
0.9675/*) 7 l 2
2.283P2/* AL 1.562 x 10
D2
AP =
+0.9675/4)T(32.2)(^)(144)
= 2.883 x (13.1.39)
13.1 Pyrolysis of a Propane Feedstock 549
First Increment.
A. Length = 20 ft.
B. Assume Tl = 885 °K 800
Tav = (866.49 + 885)/2 = 875.75 °K 100 200 300 400 500 600
Distance from inlet (ft)
C. Assume P av = 59.9 psia.
D. Use fA = 0 in equation 13.1.40 as a first Figure 13.3
estimate of an average value. Then, for a 4- Temperature and pressure profiles for pyrolysis
in. pipe with D = 4.026 in. reactor.
Table 13.2
Summary of Stepwise Calculations for 4-in. Pipe Pyrolysis Reactor
Temperature
Distance at inlet Average Average" Effluent
Increment from inlet to element AT temperature pressure APa fraction
number (ft) (°K) (°K) (°K) (psia) (psia) A/ x 103 conversion
This result is not consistent with that obtained above, and a second iteration will be in order.
Examination of equations 13.1.40 and 13.1.24 indicates that the assumed temperature will have to
be increased to bring the two calculated values of AfA1 closer together. Since AfAl = fAl — 0.000,
we may use the preliminary estimates of AfA to calculate average values of fA in equation 13.1.40
(Xi.av — 2 x 10~3). First, however, we must check the assumed pressure using equation 13.1.39
w i t h / * = 0.0044.
AP - ( 2 - 883 x
1Q-3)]875.75(2O) _
10)(0-0044)[l + 0.9675(2 x 1Q)]875.75(2O)
Thus, P av = 60 - — - = 59.82.
For the second trial we may use our estimates of fA1 and P av as input and increase our assumed
value of Tl to 885.24 K.
B. Tav = (866.49 + 885.24)/2 = 875.865
D. From equation 13.1.40,
_ 1.040 x IO 1 V 5 9 ' 1 0 0 / [ ( 1 - 9 8 7 ) ( 8 7 5 - 8 6 5 ) 1 [l - 2.0 x 10-3](59.82)(4.026)2(20) _3
'A1 =
[1 + 0.9675(2.0 x 10"3)]875.865 = 4A 4
° * 10
Second Increment.
A. Length = 20 ft.
B. Assume T2 = 901.81 °K
Tav = (885.24 + 901.81)/2 = 893.525 °K
C. Assume P av = 59.82 - 0.35 = 59.47 psia.
D. Use fA = 0.008 as a first estimate of the average fraction conversion for the second increment in
552 Illustrative Problems in Reactor Design
112 x 10~3
3
/^average = 4-10 X 10" + ~ = 7.46 X 10~ 3
which is quite consistent with our first estimate considering the insensitivity of AfA2 to small pertur-
bations in the average fraction conversion for the increment.
or
A/42 = 7.72 x 10 - 3
so our assumed temperature is consistent thus far.
F. From equation 13.1.39,
Table 13.3
Summary of Stepwise Calculations for 6-in. Pipe Pyrolysis Reactor
Temperature
Distance at inlet Average Average Effluent
Increment from inlet to element AT temperature pressure AP fraction
number (ft) (°K) (CK) (CK) (psia) psia A / x 103 conversion
Where two columns appear to be inconsistent, round-off errors are responsible for the numerical discrepancy.
these considerations, the 4-in. pipe size is to be For a discussion of modern methods of olefin
preferred. It represents an engineering com- manufacture by pyrolysis, consult the paper by
promise between the high surface area/volume Ennis and co-workers (5).
ratio characteristic of small pipe sizes and the
low pressure drop associated with large pipe
sizes. 13.2 PHTHALIC ANHYDRIDE PRODUCTION
The length of the reactor is comparable in VIA THE CATALYTIC OXIDATION OF
magnitude to that of units that have been em- NAPHTHALENE IN A FIXED BED REACTOR
ployed commercially, and it is smaller than
many. If desired, the other process parameters This problem treats some aspects of the design
could be changed, reducing the required length of a fixed bed reactor for a case where selectivity
somewhat (e.g., one could design the furnace to and safety considerations are very important.
give a somewhat higher heat flux, operate at a
different inlet pressure, or use parallel combina- 13.2.1 Charge to Design Engineer:
tions of reactors). The calculational technique
used in generating the numbers in Tables 13.2 To: Kem Inguneer
to 13.4 may be used to accomplish the necessary From: Barney Boss
calculations. The equations are readily adapt- Re: Design Specifications for Phthalic Anhydride
able to machine computation and the latter Production in a Fixed Bed Reactor
route is preferable to hand calculations in A proposed expansion of the corporation's vinyl
seeking an overall optimum design. Figure 13.4 plastics operation will require a commitment by
is a flow diagram for a program that could be the company to produce its own plasticizer.
used in machine computations to determine Our long range planning group has suggested
composition and temperature profiles from that 6 million pounds per year of new phthalic
equations 13.1.40, 13.1.24, and 13.1.39. For anhydride capacity would meet our internal
machine computation one could use signifi- needs and projected increases in demand from
cantly smaller length increments than those em- current customers.
ployed in the numerical calculations of this Either naphthalene or ortho-xylene is an
section. The accuracy of the results could acceptable starting material for partial oxida-
thereby be improved. Nonetheless, the numbers tion to phthalic anhydride, but current raw
presented in Tables 13.2 and 13.3 provide good materials costs favor the former as a starting
estimates of the required lengths of 4- and 6-in. material. Both fixed and fluidized bed processes
pipe. One must recognize that the pressure drop have been used on a commercial scale, but you
that would be associated with U-bends in the are to focus your attention on the former.
piping has not been taken into account. Greater Figure 13.5 is a schematic flow diagram of the
improvements in accuracy can be achieved by proposed process. Most research groups that
allowing for this factor in the hand calculations have studied the catalytic oxidation of naph-
than can be achieved by going to shorter-length thalene over vanadium pentoxide agree that the
segments in the machine computations. principal reactions are
naphthoquinone
maleic anhydride
naphthalene ^ (13.2.1)
CO 2 , H 2 O
phthalic anhydride''
Start
Have
, T,P
converged
satisfactorily
for this
increment
Exit
Finish
Figure 13.4
Generalized flow sheet for simulation of pyrolysis reactor by machine com-
putation.
555
556 Illustrative Problems in Reactor Design
Steam Steam
Heat transfer oil Water
Naphthalene Condensate Waste heat
Air boiler
filter
Overhead
phthalic (maleic, benzoic,
anhydride toluic acids)
Figure 13.5
Schematic diagram of phthalic anhydride production process. (Adapted from P. Ellwood, Chemical Engineer-
ing, June 2, 1969, pp. 81-82. Used with permission of Chemical Engineering.)
However, in order to design a reactor for immersed in a liquid heat transfer medium to
phthalic anhydride production, the reaction rate , give the desired results. A thorough analysis
constants for the temperature range of interest will require a study of the effects of tube diameter
are such that the following simplified set of on system performance but, for preliminary de-
reactions has been suggested as appropriate sign calculations, 2-in. diameter tubes are to be
(6, 7). employed. Cracking reactions begin to be sig-
naphthalene -• phthalic anhydride -> CO 2 , H 2 O nificant at about 673 °K, and the system should
not have any hot spots that exceed this tem-
(13.2.2) perature.^
The reactions are highly exothermic and very The necessity of avoiding potentially explo-
rapid. Consequently conventional practice in sive conditions requires that the mole percent
the design of fixed1 bed reactors for phthalic naphthalene in the feed be kept below 1%. It is
anhydride production has been based on the suggested that preliminary design calculations
use of multitube reactors to ensure good heat be based on a feed composition of 0.75 mole
transfer and good temperature control. These percent naphthalene (remainder air).
are required to ensure good selectivity. Often a Most commercial multitube reactors provide
thousand or more small diameter tubes may be the necessary cooling capacity through the use
13.2 Phthalic Anhydride Production Via the Catalytic Oxidation of Naphthalene in a Fixed Bed Reactor 557
of a molten salt that circulates around the throughout the reactor length at the feed tem-
reactor tubes and then to an external heat ex- perature. Determine the required reactor length
changer that can be used for the generation of and temperature and composition profiles for
high-pressure steam. It may be assumed that each case. Obviously there will be a trade-off
proper heat transfer equipment is available such between the number of tubes employed and the
that there is no substantial increment in the required reactor length. For present purposes
temperature of the heat transfer medium across assume that 1000 schedule 40, 2-in. stainless
the reactor. steel tubes will be used to handle 6 million lb/yr
It is known that high mass velocities are to of naphthalene feed. Use a one-dimensional
be employed within the reactor tubes to mini- model of the fixed bed reactor. The calculations
mize heat and mass transfer limitations on the should be terminated when one exceeds 99%
catalytic reaction rates. It is also known that conversion, when the pressure drops below
the effectiveness factors for the catalysts com- 1 atm, or when the temperature at any point in
monly employed often differ appreciably from the reactor exceeds 673 °K.
unity.
Your assignment is to use the data and as- Data Summary and Permissable Assumptions.
sumptions listed below to prepare a preliminary There are a number of groups that have at-
reactor design in which particular emphasis is tempted to simulate fixed bed performance for
placed on the variation of the maximum yield of this reaction system (7-9). Many of the data
phthalic anhydride with feed temperature. below have been taken from these sources.
The reactor inlet pressure may be taken as
25 psia. Consider feed temperatures of 630, 640, Reaction Stoichiometry and Thermodynamic
650, and 660° K. In each case assume that the Data. The stoichiometry of the reactions of
heat transfer medium maintains the tube walls interest may be represented as:
O
.C.
O + 7.5O2 -^-> 8CO 2 + 2H 2 O (13.2.4)
O
or
C8H4O3 7.5O2 8CO 2 + 2H 2 O (13.2.4)
558 Illustrative Problems in Reactor Design
The second of these reactions may not go to Under these conditions the mole fraction
completion, and the incomplete combustion oxygen will drop from 0.2084 to 0.1636, which
may yield maleic anhydride, benzoic acid, CO, represents a decrease of 21.5% or a variation
or other compounds as by-products. of ± 12.1%. At lower conversions the difference
There is considerable variation in the heat will be even less substantial.
of reaction data employed in different articles Because the feed is primarily air and because
in the literature that deals with this reaction. substantial amounts of N 2 and O 2 are present
Cited values differ by more than an order of in the effluent stream, we will assume that the
magnitude. If we utilize heat of combustion heat capacity of the reactant mixture may be
data for naphthalene and phthalic anhydride taken as that of air at 630° K (0.255 cal/g-°K).
and correct for the fact that water will be a gas The variations of the heat capacity with tem-
instead of a liquid at the conditions of interest, perature and pressure may be neglected.
we find that for the first reaction (equation The Prandtl number of the fluid may be
13.2.3) the standard enthalpy change will be taken as 0.79.
approximately — 429 kcal/g mole; for the second
reaction it will be approximately — 760 kcal/g Kinetic Data and Catalyst Data. The simpli-
mole. These values will be used as appropriate fied reactions indicated by equation 13.2.2 may
for the temperature range of interest. Any be assumed on the basis of arguments set forth
variation of these parameters with temperature by DeMaria et al. (8). For the catalyst employed
may be neglected. by this research group, pseudo first-order ki-
For the case where the inlet naphthalene netics are followed by each of the two reactions,
concentration is below the explosion limit and with
where one is successful in producing phthalic
anhydride in commercial yields, the mole frac- k, = 5.74 x I 0 1 3 6 ' - 3 8 ' 0 0 0 / K r s e c - 1 (13.2.5)
tion oxygen in the reactant gases remains rela-
tively constant. This point is readily seen by k2 = 2.19 (13.2.6)
the construction of a mole table for a feed
containing 0.75% naphthalene, of which 80% where T is expressed in degrees Kelvin and R
reacts to give phthalic anhydride and the re- in calories per gram mole-degree Kelvin. Both
mainder goes to CO 2 . reactions may be regarded as irreversible.
Naphthalene 0.75 0 0
Nitrogen 0.79(99.25) = 78.41 78.41 78.53
Phthalic anhydride 0 0.8(0.75) = 0.60 0.60
CO 2 0 2(0.8)(0.75) + 10(0.2)(0.75) = 2.70 2.70
H2O 0 2(0.8)(0.75) + 4(0.2)(0.75) = 1.80 1.80
Oxygen 0.21(99.25) = 20.84 20.84 - |(0.60) - 2.70 - 1(1.80) = 16.34 16.36
Total 100.0 99.85 99.85 99.99
13.2 Phthalk Anhydride Production Via the Catalytic Oxidation of Naphthalene in a Fixed Bed Reactor 559
The units on the rate constants reported by The overall heat transfer coefficient for ther-
DeMaria et al. indicate that they are based on mal energy exchange between the tube wall and
pseudo homogeneous rate expressions (i.e., the the reacting fluid may be taken as 1.0 x
product of a catalyst bulk density and a reaction 10~ 3 cal/cm2-sec-°K. The effective thermal con-
rate per unit mass of catalyst). It may be assumed ductivity of the catalyst pellets may be taken as
that these relations pertain to the intrinsic equal to 6.5 x 10~ 4 cal/(sec-cm-°C).
reaction kinetics in the absence of any heat or
mass transfer limitations.
13.2.2 Solution
The catalyst consists of V 2 O 5 supported on
silica gel with K 2 SO 4 and other promoters also The design of a fixed bed reactor for phthalic
present. The physical property values tabulated anhydride production in terms of a one-dimen-
below are typical of the low or intermediate sional model involves only the integration of a
surface area supports that one might expect to set of ordinary differential equations. In the
use in this application. DeMaria et al. (8) did general case for the consecutive reactions of
not report data of these types for their catalyst. interest, the reaction selectivity will depend on
several process variables. A thorough design
Pellet shape—spherical analysis would require an investigation of the
Pellet diameter = 0.318 cm effects of variations in inlet temperature, coolant
Vg = 0.26 cm 3 /g temperature, tube diameter, catalyst pellet size,
mass velocity through the tubes as influenced
Sg = 80 m 2 /g by the number of tubes, system pressure, and
Bulk density = 0.84 g/cm3 the feed composition on the variation of com-
Skeletal density = 2.2 g/cm 3 position and temperature with bed length. One
Bed porosity BB = 0.40 should also investigate the sensitivity of the
results to uncertainties in the wall heat transfer
Pellet porosity £p = 0.364 coefficient, reaction activation energies, etc. For
this highly exothermic reaction, an investigation
One need not worry about the problem of
of a two-dimensional model of the reactor
catalyst deactivation for this reaction. Many
would also be appropriate. We will restrict our
process descriptions quote lifetimes of 3 years
discussion to variations of the reactor inlet
or more.
temperature for the case where the coolant is
maintained at this temperature. Space limita-
Transport Properties. Because the feed is pri- tions preclude our study of the influence of other
marily air and because substantial amounts of variables, but the method discussed below can
N 2 and O 2 are present in the effluent stream, be used to determine the one-dimensional model
we will assume that the fluid viscosity is that predictions of reactor performance for varia-
of air for purposes of pressure drop calculations. tions in other process parameters.
For the temperature range of interest, the fluid One of the first things that should be done
viscosity may be taken as equal to 320 micro- in the analysis is to determine if pressure varia-
poise. The pressure range of interest does not tions along the length of a reasonable-size
extend to levels where variations of viscosity reactor will be significant for the specified
with pressure need be considered. The effective operating conditions. This will require a knowl-
diffusivities of naphthalene and phthalic anhy- edge of the superficial mass velocity through the
dride in the catalyst pellet may be evaluated tubes. This quantity may be calculated from
using the techniques developed in Section 12.2. the tube dimensions and the inlet flow rate and
560 Illustrative Problems in Reactor Design
m o l e s
6,000,000^ x 4 5 4 1 x -?L- x ** x^ _ x | _ _ = 0.6740^^
sec
Per tube the molal flow rate of naphthalene is The fluid density at inlet conditions may be
then 6.740 x 10 ~ 4 g moles/sec. The total inlet calculated from the ideal gas law
molal flow rate is equal to 0.6740/0.0075, or
PM
89.86 g moles/sec. The average molecular weight 9 = (13.2.9)
of the feed is given by
The total mass flow rate is then 29.58(89.86) = If we consider the case where T — 630 °K, then
2658 g/sec or 2.658 kg/sec. For 1000 tubes in 25
parallel the available cross section for flow is
given by 14.696/
X
I (29.58)
9 = - 9.73 x 10- 4 g/cm 3
(82.06)(630)
- 23.30 f t 2 ^ 2.165 m 2 (13.2.10)
Since G = up, the fluid velocity at inlet condi-
tions is then
The mass velocity is then 2.658/2.165 or
1.228 kg/m2-sec or 0.1228 g/cm2-sec. G 0.1228
The modified Reynolds number based on the u =— = r r = 126 cm/sec =c= 1.26 m/sec
p 9.73 x 10
catalyst pellet diameter of 1/8 in. or 0.318 cm (13.2.11)
is then
The pressure drop through the packed bed
reactor is governed by the Ergun equation dis-
M cussed earlier as equation 12.7.4.
The density may be eliminated using equation in the single pore model of a catalyst particle.
13.2.9 to give The Knudsen diffusivity is given by equation
12.2.4;
d& = G2RT(1 ~ eB) riSO(l -aB) 1
dL D*PM sB3 |_ (D*G/fi) ' J DK = 9.7 x 103r /— (13.2.17)
VM
(13.2.14) 2V
where r = —- for the cylindrical pore model.
Substitution of numerical values gives
where Thus
0* and P are expressed in dynes per square
centimeter r = = 6.5 x 1 0 " 7 c m ^ 6 5 A ^ 6 . 5 n m
80 x 10
L is expressed in centimeters (13.2.18)
T is expressed in degrees Kelvin. and
The inlet pressure of 25 psia corresponds to
1.724 x 106 dynes/cm2. Thus, when the inlet DK = 9.7 x 103(6.5 x
temperature is 630 °K, the pressure gradient at
the start of the bed is equal to (13.2.19)
9.73 x 1(T
3
( + 429,000)(1.41 x 10" )(0.0075)V 29.58
= 3.64 x 10" 4 (13.2.32)
(6.5 x l(T 4 )(630)
phthalic anhydride is the dominant product transfer coefficient between the catalyst surface
species and yields are in excess of 70%, the as- and the bulk fluid is given by
sumption of an isothermal pellet is a good one.
hP = j H (13.2.39)
Furthermore, /? will diminish as one moves
through the reactor, since C o will decrease as one or
proceeds downstream. In any event we will pre- hP = (0.159(0.255)(0.1228)(0.79)~2/3
sume throughout the remainder of our analysis = 5.83 x 10~ 3 cal/(cm 2 -sec-°K)
that the pellets are isothermal.
(13.2.40)
Since our calculations indicate that intra-
particle mass transfer limitations are significant, At steady state the rate of mass transfer must
we must now consider the possibility that tem- equal the reaction rate and, if one neglects the
perature and concentration differences will exist change in the number of moles on reaction, the
between the bulk fluid and the external surface drift factor may be taken as unity. Thus
of the catalyst. Appropriate mass and heat trans-
fer coefficients must therefore be determined. KACUES ~ Q> m = vtrm (13.2.41)
At the Reynolds number of interest, equation where am is the gross external surface area of the
12.4.10 indicates that catalyst pellets per unit mass. Now for spherical
pellets with a pellet porosity of 0.364 and a
0.357 0.357
JH = JD = = 0.159 skeletal density of 2.2.
0.4(122) 0 3 5 9
(13.2.35) nDl 6
axial dispersion term, a material balance on naphthalene gives the following expression.
or
where we have recognized that changes in system pressure and temperature influence the reactant
concentration through equation 3.1.44. We have, however, assumed that since the reactants are so
dilute, the volume expansion parameter 3 may be taken as approximately zero. Thus
(13.2.54)
N0
(13.2.55)
Thus
ACN = A fc (13.2.56)
Equation 12.3.149 relates the cnanges in phthalic anhydride and naphthalene concentrations in
the gas phase
ACPA y
PA Ir PA
(13.2.57)
ACN 3N V PA tanh hT
- 1 - 1
L\
where CPA and CN are the local concentrations of phthalic anhydride and naphthalene, respectively.
The diffusivity parameters take into account both knudsen diffusion and ordinary molecular diffusion.
The inlet naphthalene concentration can be determined from the ideal gas law
and equations 13.2.56 and 13.2.57 then permit one to determine the naphthalene and phthalic anhydride
concentrations at all points in the reactor by stepwise integration processes.
In order to use these equations, however, the local temperatures and pressures must be known.
Equation 13.2.16 can be used to evaluate the latter parameter, while an energy balance on the reacting
fluid gives the temperature. If the temperature difference between the bulk fluid and the external
surface of the catalyst pellets is neglected for the reasons noted earlier, the energy balance on the
reacting fluid can be written as:
(13.2.59)
where rlm and r2m are the rates of reactions 1 and 2 per unit mass of catalyst.
The rate of the first reaction is readily calculated from simple material balance considerations on
our differential length element
(13.2.62)
"Qv ~rlm r
lm
Thus,
(&CPA
(13.2.63)
Thus,
dCp
r l w AH, + r2mAH2 = rml \AH, + (l
dCN
(13.2.65)
13.2 Phthalic Anhydride Production Via the Catalytic Oxidation of Naphthalene in a Fixed Bed Reactor 567
LITERATURE CITATIONS 5. Ennis, B. P., Boyd, H. B., and Orriss, R., Chemtech, 5
(693), 1975.
1. Fair, J. R., and Rase, H. F., Chem. Eng. Prog., 50 r , ,
J J
415' ' 6. Westerterp, K. R., Chem. Eng. Sci., 77(423), 1962.
1 h
7. Carberry, J. J., and White, D., bid. Eng. Chem., 67(7),
2. Schutt, H. C , Chem. Eng. Prog., 43 (103), 1947. p 27, 1969.
3. Reid, R. C , and Sherwood, T. K., The Properties of 8 DeMaria, R., Longfield, J. E., and Butler, G., Ind.
Gases and Liquids, Second Edition, Chapter 9, McGraw- £nc, Chem 53(259) 1961
Hill, New York, 1966. 9 Froment> Q p ^ ^ ^ ^ ^ pp x%_^
A Thermochemical Data
The data in this appendix have been compiled College Station, Texas (Loose-leaf data sheets,
from a number of sources. Nearly all of the extant, 1977).
critical property data are taken from Appendix 2. Zwolinski, B. J., et al., "Selected Values of
A of The Properties of Gases and Liquids, Second Properties of Chemical Compounds," Ther-
Edition, by R. C. Reid and T. K. Sherwood, modynamics Research Center Data Project,
copyright © 1966, McGraw-Hill Book Com- Thermodynamics Research Center, Texas
pany. They are used with the permission of A & M University, College Station, Texas
McGraw-Hill Book Company. Most of the (Loose-Leaf data sheets, extant, 1977).
thermochemical data (AG°, A/f °, and S°) were
They are used with permission of the Thermo-
obtained from the following sources.
dynamics Research Center. Other sources used
1. Zwolinski, B. J., et al., "Selected Values of include the Handbook of Chemistry and Physics,
Properties of Hydrocarbons and Related 57th edition, edited by R. C. Weast, CRC Press,
Compounds," American Petroleum Institute Cleveland, 1974; and Lange's Handbook of
Research Project 44, Thermodynamics Re- Chemistry, 11th edition, edited by J. A. Dean,
search Center, Texas A & M University, McGraw-Hill Book Company, New York, 1973.
cO
° 2 5CC
Absolute
State of Molecular AG?, 2 5 . C A///,25 C entropy Tc Pc
Compound aggregation weight kcal/g-mole kcal/g-mole cal/g-mole-°K °K Atm
Elementary gases
Hydrogen Gas 2.016 0 0 31.211 33.3 12.80
Oxygen Gas 32.000 0 0 49.003 154.8 50.1
Nitrogen Gas 28.016 0 0 45.767 126.2 33.5
Fluorine Gas 38.00 0 0 48.6 144. 55.
Chlorine Gas 70.91 0 0 53.286 417. 76.1
Bromine Liquid 159.83 0 0 36.4 584. 102.
Iodine Crystal 253.82 0 0 27.9 785. 116.
Paraffins
Methane Gas 16.04 - 12.140 - 17.889 44.50 190.7 45.8
Ethane Gas 30.07 - 7.860 - 20.236 54.85 305.4 48.2
Propane Gas 44.09 - 5.614 - 24.820 64.51 369.9 42.0
^-Butane Gas 58.12 - 4.10 - 30.15 74.12 425.2 37.5
Isobutane Gas 58.12 - 5.00 - 32.15 70.42 408.1 36.0
rc-Pentane Gas 72.15 - 2.00 - 35.00 83.40 469.5 33.3
2-Methylbutane Gas 72.15 - 3.50 - 36.92 82.12 460.4 32.9
Neopentane Gas 72.15 - 3.64 - 39.67 73.23 433.8 31.6
ft-Hexane Gas 86.17 - 0.07 - 39.96 92.83 507.3 29.9
2-Methylpentane Gas 86.17 - 1.20 - 41.66 90.95 496.5 30.0
3-Methylpentane Gas 86.17 - 0.51 - 41.02 90.77 504.7 30.8
570
Appendix A Thermochemical Data 571
5
cO
°25C
Absolute
State of Molecular AG? i25 . c AH?, 25 c entropy T1
c
Pc
Compound aggregation weight kcal/g-mole kcal/g-mole cal/g-mole-cK °K Atm
Olefins
Ethylene Gas 28.05 16.282 12.496 52.45 283.1 50.5
Propene Gas 42.08 14.990 4.879 63.80 365.1 45.4
1-Butene Gas 56.10 17.09 - 0.03 73.04 419.6 39.7
ds-2-Butene Gas 56.10 15.74 - 1.67 71.90 434.6 40.5
trans-2-Butene Gas 56.10 15.05 - 2.67 70.86 428.6 41.5
Isobutylene Gas 56.10 13.88 - 4.04 70.17 419.7 39.5
1-Pentene Gas 70.13 18.96 - 5.00 82.65 464.8 39.9
ds-2-Pentene Gas 70.13 17.17 - 6.71 82.76 475.6 40.4
trans-2-Pentene Gas 70.13 16.76 - 7.59 81.36 475.6 40.4
3-Methyl-l-butene Gas 70.13 17.87 - 6.92 79.70 464.8 33.9
2-Methyl-2-butene Gas 70.13 14.26 - 10.17 80.92 470. 34.
1-Hexene Gas 84.16 20.94 - 9.96 91.93 504.0 31.1°
1-Octene Gas 112.21 24.96 - 19.82 110.55 578. 25.5*
Cyclopentene Gas 68.11 26.48 7.87 69.23 506.1 47.2°
Diolefins
Propadiene Gas 40.06 48.37 45.92 58.30 393.3 43.6fl
1,3-Butadiene Gas 54.09 36.43 26.75 66.62 425. 42.7
Acetylenes
Acetylene Gas 26.04 50.000 54.194 47.997 309.5 61.6
Propyne Gas 40.06 46.313 44.319 59.30 401. 52.8
Ethylacetylene Gas 54.09 48.30 39.48 69.51 463.7 37.5°
Dimethylacetylene Gas 54.09 44.32 34.97 67.71 488.7 37.5°
Cycloparaffins
Cyclopentane Gas 70.13 9.29 - 18.41 70.00 511.8 44.6
Methylcyclopentane Gas 84.16 8.55 - 25.50 81.24 532.7 37.4
Ethylcyclopentane Gas 98.18 10.66 - 30.37 90.42 569.5 33.5
Cyclohexane Gas 84.16 7.60 - 29.43 71.28 553.2 40.
Methylcyclohexane Gas 98.18 6.52 - 36.99 82.06 572.1 34.3
Aromatics
Benzene Gas 78.11 30.989 19.820 64.34 562.1 48.6
Toluene Gas 92.13 29.228 11.950 76.42 592.0 41.6
o-Xylene Gas 106.16 29.177 4.540 84.31 631.6 35.7
m-Xylene Gas 106.16 28.405 4.120 85.49 616.8 34.7
p-Xylene Gas 106.16 28.952 4.290 84.23 618.8 33.9
Ethylbenzene Gas 106.16 31.208 7.120 86.15 617.1 36.9
Alcohols
Methyl alcohol Gas 32.04 - 38.81 - 48.05 57.29 513.2 78.5
Liquid - 39.85 - 57.11 30.4
Ethyl alcohol Gas 46.07 - 40.13 - 56.03 67.54 516.3 63.0
Liquid - 41.65 - 66.20 38.53
5°
Absolute
State of Molecular AG°- 7 V C A / / / . 2 5-C entropy Tc Pc
Compound aggregation weight kcal/g-mole kcal/g-mole cal/g-mole-GK C
K Atm
572
^25 C
Absolute
State of Molecular AG°, 2 5 . C AH
f,25C entropy Tc Pc
Compound aggregation weight kcal/g-mole kcal/g-mole cal/g-mole-°K °K Atm
573
APPENDIX
Generalized Fugacity
B Coefficients of Pure Gases and
Liquids (Z c = 0.27)
Reduced pressure, Pr
0.1
(Reprinted from Chemical Process Principles Charts, Third Edition, by O. A. Hougen, K. M. Watson, and R. A.
Ragatz. Copyright 1964. Reprinted by permission of John Wiley and Sons, Inc.)
574
APPENDIX
C
Symbol
Nomenclature
Meaning ratio of reaction rate for poisoned
a activity catalyst to that for unpoisoned catalyst
a external (superficial) surface area G Gibbs free energy
A pre-exponential factor in rate constant G mass velocity
At representative chemical species h specific enthalpy
C concentration convective heat transfer coefficient
CAi concentration of rectant A at time tt
K
Thiele modulus for ith order reaction
or in effluent from reactor i
C* reactant concentration at time zero in
K Thiele modulus for poisoned catalyst
hT Thiele modulus
a CSTR operating under transient JD Chilton-Colburn factor defined by
conditions equation 12.4.9
Cp constant pressure heat capacity JH Chilton-Colburn factor defined by
D diameter equation 12.5.1
D dielectric constant Molar flux of species i relative to the
DAB bulk diffusivity molar average velocity
DK Knudsen diffusivity k reaction rate constant
3) dispersion or diffusivity parameter k thermal conductivity
@Am pseudo-binary diffusivity of species A kB Boltzmann constant
in a multicomponent gas mixture K mass transfer coefficient
<3)c combined diffusivity thermal conductivity of bulk fluid
E energy or activation energy mass transfer coefficient defined by
Ec relative kinetic energy directed along equation 12.4.2
the line of centers in a collision (on a rate constant for acid catalyzed
per mole basis) reaction
^Diffusion activation energy for diffusion defined mass transfer coefficient
by equation 12.3.84 rate constant in infinitely dilute solu-
Eo energy increase accompanying reac- tion or for uncatalyzed reaction
tion at absolute zero rate constant for base catalyzed
Eo initial enzyme concentration reaction
/ fraction conversion k pseudo rate constant for desorption
/? fugacity of pure species i at pressure P process
fi fugacity of species i as it exists in the K equilibrium constant
reaction mixture K Michaelis constant
fM friction factor equilibrium constant for reaction ex-
(f/P)i fugacity coefficient for species i pressed in terms of activities
F molal flow rate adsorption equilibrium constant
F'A hypothetical molal flow rate of species equilibrium constant expressed in
A corresponding to a stream in which terms of partial pressures
none of the A has reacted equilibrium constant for surface reac-
F(t) function used in characterization of tion expressed in terms of the frac-
fluid residence times (see Section 11.1) tional surface coverages 6t
575
576 Appendix C Nomenclature