this field has not suffered from a lack of APPLIED THERMODYNAMICS SYMPOSIUM

Certainly,
in either journals or textbooks. Dozens of
treatment
equations have been proposed in the past hundred years
and many reviews and chapters of books have been
written to compare their relative merits. Despite this
prolific outpouring on the subject, some facts and
approaches are either misunderstood or not fully appreci-
ated and these justify discussion.
In view of the large number of equations of state
which have been suggested, it is natural to search for the
motivation. There are two rather obvious reasons for
the widespread activity in this field over an extended
period of time. The first is the fact that the problem of
EQUATIONS
developing an equation of state is mathematically
fascinating and particularly tantalizing because it seems
so simple, at least at the start. Examination of the
data in regular tabular or graphical form can lead one to
OF
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

believe that it should not be difficult to obtain a satis-

factory algebraic representation. The second reason
concerns the exceptional power and utility of an equa- STATE
Downloaded via ECOPETROL SA on September 1, 2019 at 04:40:12 (UTC).

tion of state. When combined with appropriate thermo-

dynamic relations, a well behaved equation can predict JOSEPH J. MARTIN
 with high precision isothermal changes in heat capacity,
enthalpy, entropy and fugacity, vapor pressure, latent
heat of vaporization, activity coefficients, and vapor-
liquid equilibria in mixtures, not to mention the assist-
ance it offers in transport property correlations. Un-
fortunately, even though the useful applications of an
equation of state are so extensive and attractive, the
development of a high performance equation proves
to be so involved that to date no one has come close to
discovering a single relation which is truly good over a
Development of an equation of state is wide range of density. Because of the apparent sim-
plicity of the task, it is not surprising that many have
mathematically fascinating and particularly rushed in where more experienced workers have feared to
tread. This is not meant to frighten away interested
tantalizing because it apparently seems so but uninitiated workers; it is primarily to emphasize
the frontier challenge and opportunity for further re-
search.
simple, at least at the start, and the exceptional
Since the phrase “equation of state” has more than a
single connotation in thermodynamics, it is well to
power and utility answer the question,
emphasize that here it refers to the equilibrium relation,
in the absence of special force fields, among pressure,
"Why another article on equations of state?" volume, temperature, and composition of a substance
whether it be quite pure or in a uniform mixture. In
functional form this relation is f(P, V, T,x) 0, though
=

in this paper only pure chemical species will be treated

(x=
unity), and reference will be made simply to PVT
equations and PVT data. Equations of state may be
applied to gases, liquids, and solids, but application to
solids will not be considered.
Some persons are understandably interested only in
the final comparison of the predictions of an equation
of state and the experimental data. What is needed,
however, is thorough comprehension of the inherent
properties of certain algebraic expressions that are or
may be used in equations of state, and the characteristic
behavior of PVT data. Three previous papers (27,
23, 24) have covered in some detail the characteristics of
PVT data that should be kept in mind when seeking to
produce an equation to represent the data with high
precision. It is not the objective here to review all such
characteristics or to evaluate all of the equations that
have been proposed. Rather it is to look at the objec-
tives sought in the more recent equations and to seek
insight into how anequation of state is designed and what
must be accounted for if it is successfully to carry out its
job.
Two approaches to the development of an equation
of state may be followed. One is the theoretical ap-
proach based on either kinetic theory or statistical
mechanics involving intermolecular forces. This is exem-
plified by such equations as those of van der Waals (37),
Lennard-Jones and Devonshire (79), Hirschfelder, Bird,
and Spotz (73), and Flory, Orwall, and Vrij (9). The
other approach is empirical or at best only semitheo-
retical. This is exemplified by equations such as those of
Clausius (6), Bcrthelot (3), Dieterici (8), Wohl (38),
Keyes (77), the NBS (36), Beattie and Bridgeman (7),
Keyes, Smith, and Gerry (78), Benedict, Webb, and
Rubin (2), Rcdlich and Kwong (33), Martin and Hou
(23), Pings and Sage (30), Hirschfelder, Buehler, McGee,

VOL. 59 NO. 12 DECEMBER 1967 35

and Sutton (14), Strobridge (35), Costolnick and Thodos
(7), McCarty and Stewart (26), and Goodwin (10).
This paper focuses attention on the empirical or
semitheoretical approach since this is the one which
has had the greatest success in representing data with
high precision over a wide range of density. A com-
panion paper (20) is currently being prepared on the
determination of closed equations of state (as contrasted
to open ended series or “virial” equations as they are
called) through a statistical analysis of intermolecular
force potentials. It suffices to say, however, that such
theoretical equations as have been developed to date
are definitely more limited in range of application than
the empirical equations. This is not to infer that the
new developments of tomorrow will not come from
statistical mechanics; it is merely a realistic evaluation
of the field at this moment. For example, if a table of
thermodynamic properties were to be prepared today
for the gas and liquid phases of a pure substance for
which the PVT data were available, it would be based
on an empirical equation of state or even numerical
analysis rather than on an equation from kinetic theory
or statistical mechanics.
In the search for an equation of state, one must make
three decisions at the beginning. The first concerns
the amount and kind of data that will be required
to obtain the parameters in the equation, the second
concerns the range of density to be covered, and the
third concerns the precision with which the PVT data
are to be represented. For some applications, it is
desirable to have an equation which can be obtained
from a minimum of data—i.e., the critical temperature,
pressure, and volume, or the boiling point, or some molec-
ular parameter. In other instances an equation is
sought which will represent a large amount of experi-
mental PVT data within the precision of the experiment.
Density is paramount in all pressure-explicit equations.
If the same high order of precision is to be obtained at
all times, a relatively simple short equation will suffice
for low densities, whereas a long complicated equation
will be required if coverage is to include both low and
high densities. From our present knowledge high
precision equations have many arbitrary constants whose
number depends primarily upon the density range and
in a minor way upon the temperature range. For
example, as the situation exists today, an equation
representing the data precisely up to a fiftieth of the
critical density may need only two constants, while four
or five constants will be necessary to go to one half the
critical density. At least a half dozen or more con-
stants will be required to continue up to the critical
density, and twice that many will be needed if the goal
is one-and-a-half to two times the critical density. This
36 INDUSTRIAL AND ENGINEERING CHEMISTRY
may be a pessimistic view of the status of the field, but
it should be considered more of a challenge than a
limitation on what may be discovered in the future.
As examples of the objectives of various investigators,
Keyes (77) and Beattie and Bridgeman (7) designed
their equations to reproduce the data within the experi-
mental precision for densities up to a little over half the
critical density. Benedict, Webb, and Rubin (BWR)
(2) essentially modified the Beattie-Bridgeman equation
to extend the density range well past the critical and
still fit the data with high precision, while Bloomer and
Rao (4) modified the temperature coefficients of the
BWR equation to fit nitrogen data. A previous equa-
tion by the author (23) had essentially the same goals
of precision and density range as the Benedict, Webb,
and Rubin equation. Keyes, Smith, and Gerry (18)
carried their search for high precision to an extreme in
calculating thermodynamic properties of steam (16)
by developing an equation of state with nine constants
including one term with temperature to the forty-ninth
power; yet even with this complexity the equation
is valid only to a third of the critical density. Many
years ago Onnes (29) suggested that high precision
could be obtained by calculating pressure from a power
expansion in density with the coefficients being tempera-
ture functions. This is the virial equation and a number
of authors such as Hirschfelder, Bird, and Spotz (13),
Gyorog and Obert (72), and Joffe and Patel (15) have
devised theoretical and empirical procedures to obtain
the virial coefficients for low densities.
Michels (28) has long employed the isothermal virial
equation to fit precise experimental data to two-and-a-
half times the critical density using terms as high as the
ninth power of density. Hirschfelder, Buehler, McGee,
and Sutton and Costolnick and Thodos covered a wide
range of density and simultaneously maintained fair pre-
cision by dividing the PVT plateau of data into regions
and writing separate equations for the different regions.
Flory, Orwall, and Vrij and Lennard- Jones and Devon-
shire sought high precision in their equations but limited
the range of applicability to the liquid region where the
densities are always significantly greater than the critical
density. Van der Waals and others that followed him,
such as Clausius, Dieterici, Berthelot, Wohl, and Redlich
and Kwong, endeavored to find equations covering the
whole range of density from infinitely attenuated gas to
compressed liquid, but were willing to accept fairly
large deviations from the experimental results. It is
interesting to note, however, that Redlich in later
papers with other collaborators (31, 32) focused atten-
tion on higher precision representation and increased
the number of arbitrary constants in his equation from
2 to 44. With the improvement in experimental PVT
 Figure 7. Pressure-volume plot

techniques it isnatural that the trend in recent years

has been toward more complicated equations which can
fit the data within the experimental precision.

Design of Simple Equation of State

When one is developing an equation of state empiri- Figure 2. Pressure-temperature plot
cally, there are six graphs that it pays well to keep in
mind. These are Figure 1, the P-V plot; Figure 2, the
P-T plot; Figure 3, the generalized compressibility or z
plot; Figure 4, the generalized vapor pressure plot;
Figure 5, the generalized second virial coefficient plot;
and Figure 6, the generalized third virial coefficient
plot. These will be discussed as the occasion demands
their use. We shall start with the critical isotherm on
the P-V plot, just as van der Waals did almost 100
years ago, and seek to fit it with his equation :

RT a
~

(1)
V -
b V*

Other two-constant (exclusive of R) equations could be

used, but they offer no particular advantages. The
standard procedure given in a multitude of books will be
followed, but it is necessary to repeat it here to demon-
strate the thread of logic one must carry beyond the
normal end point. Because the critical point is a point
of horizontal slope and inflection, the usual first two
pressure-volume derivatives are set to zero:
-RTC 2 a

(ve -

by
+
iv"° (2)

and
2 RTC 6 a
(3) AUTHOR Joseph J. Martin is Professor of Chemical Engi-
“

(K -

b)3 v?
neering at the University of Michigan, Ann Arbor, Mich. The
Since Equation holds in general, it holds at the
1 author acknowledges the financial assistance and patience over a
critical point, so there are three equations available 5-year period of the National Science Foundation which made the
to solve for the two arbitrary constants, a and b. If study of many kinds of equations of state possible.

VOL 59 NO. 12 DECEMBER 1967 37

 +0.1
0.0

-0.5

1.0

-15

Figure 4. Generalized vapor pressure chart Figure 5. Generalized second virial coefficient

0.1

0.08

0.06

Eqn. (13)

0.02

Data

0.01
0

Figure 6. Generalized third virial coefficient

V, 1.29V, V

Figure 7. Critical isotherms of three van der Waals equations

38 INDUSTRIAL AND ENGINEERING CHEMISTRY

Pc, Ve, and Tc are known, any two equations can be this can be given in terms of both Pc and Tc in Equation
chosen for the solution. If Equations 1 and 2 are 6. The result is:
selected, the result is an equation for the critical isotherm
which passes through the critical point with zero slope,
but has a positive value of (d2P/dV2)Tc,vc for ordinary
W =

f (10)

values of Pc, Vc, and Tc, resulting in a minimum at the / S

,
p rp
critical point and a maximum at a slightly larger volume. b(Pc, Tc)
_
-

(11)
gP[
If Equations 1 and 3 are selected for the solution, the
critical isotherm passes through the critical point with a If the same pair of independent variables is employed
positive slope and exhibits a minimum at a slightly in calculating a and b, three different equations of state
smaller volume and a maximum at a slightly larger result:
volume than the critical. Although these two proce-
RT a(Vc, Tc)
dures of solution have some interesting properties, they P(V„ Tc) =
“

will not be pursued further here. F ~

b(Vc) V2

If Equations 2 and 3 are selected for the solution, the RT 9 RTcVc

critical isotherm has zero values of (t>PfdV)Tc,vt and (12)
\

8 V2
(d2P/5F2)rc,rc at Vc and Tc, but passes well above the
1

critical point for ordinary values of Pc, Vc, and Tc.

Now this is an exceptional case because there is no Pc RT a(Ve, Tc)
in Equations 2 and 3, so that solution gives a and b P(Pc, Vc) =
“

F ~

b(Vc) V2
in terms of only Vc and Tc, as follows:
RT 3 PcVc2
9 RTcVc (13)
(4) V2
1

RT a(Pc, Tc)
V, P(Pc, Tc) = '
* “
(5) F ~

b(Pc, Tc) V2
3
RT 27 R2T2
When Equations 4 and 5 are put into Equation 1, a 1

(14)
value of Pc at Vc and Tc is predicted to be: 1
E-H
I
64 PCV2
loo
3 RTC PcVc
Pc or 0.375 (6) These are all perfectly valid equations based on the
8 Vc RTc
van der Waals development. Guggenheim (77) recog-
Equation 6 furnishes some interesting possibilities. nized these several alternatives and stated, “Unfor-
One may forget that Vc and Tc were the independent tunately one meets the statement that van der Waals’
quantities in Equations 2 and 3, and assume in Equation constants a and b were computed from the critical data.
6 that Pc and Tc are independent and calculate Such a statement is at least ambiguous, if not non-
Vc. Alternately, Pc and Vc may be selected as in- sensical; for as van der Waals’ equation gives a false
dependent and Tc may be calculated. The selection value of PcVjRTc one can evidently obtain various
of any particular pair of independent quantities from inconsistent sets of values of a and b according as these
Pc, Vc, and Tc permits different expressions for a and b are computed from Pc, Vc, or from Pc, Tc, or from Vc,
by elimination of the dependent quantity through Tc.” It is probable that Guggenheim was thinking of
Equation 6. Thus, with the dependent quantity being the inconsistency that would arise if a and b were not
either Pc, Vc, or Tc, there are three possibilities for a, computed simultaneously from the same pair of inde-
which are, respectively, pendent quantities, but even for the consistent systems,
9 RTcVc Equations 12, 13, and 14, there are some striking differ-
a(Vc, Tc) (7) ences. The critical isotherms of the three are plotted
8
on Figure 7, along with the true curve which has been
27 R2T2 measured for argon. Equation 12 predicts Pc from the
a (Pc, Tc) (8)
64 Pc given Vc and Tc, and the value is high in the ratio of the
critical compressibility factors, zCtdw/zCA 0.375/0.291 =

a(Pc, Vc) = 3 PCV2 (9) = 1.29. Equation 13 does not give what seems to be a
There are only two possibilities for b, however, since in critical isotherm at the true Tc because it predicts a Tc
Equation 5 it is a function of only one quantity, Vc, and which is low by the ratio ZcJ^cdw 0.291/0.375 — =

VOL. 59 NO. 12 DECEMBER 1967 39

1/1.29. Thus, the line drawn is at an apparent reduced
temperature of 1.29. Equation 14 predicts Vc from the
given Pc and Tc, and the value is high in the ratio, 1.29,
in a manner similar to Equation 12.
result is as shown in Figure 8, and it is seen that the
transformation has been carried too far to the left.
If the transformation is enough to make the calculated
zc V* instead of 0.291, the curve lies as shown in
=

The principal use of Figure 7 is to determine if there
is a way that any of the three equations can be adjusted
to fit the data curve. Equation 12 can be eliminated
because a vertical adjustment in pressure will yield
Figure 9. One could refine this adjustment depending
upon where is it desired best to fit the data; in fact,
zc =
l/% works well for argon, but slightly smaller values
are better for substances whose critical compressibility

negative values of pressure at large volumes. Equation is somewhat below 0.291. In the case of argon, the
13 is eliminated because it does not have the right slope volume transformation is rounded off so that ze 0.335 =

at the critical volume and a transformation in tempera- instead of lj3 to make the numbers simple.
ture would yield incorrect limiting values at low or high The procedure to obtain the equation of state with
temperatures. Equation 14 does have the general numerical constants is to combine Equations 8, 11, and
appearance of the true curve except for a horizontal 15 to get:
displacement in volume. Thus, a linear transformation RT 27 R1 T2
in the volume coordinate is indicated as a good possi- (16)
RTC 64 PC(V + c)2
bility particularly because it has no effect on the two z—r + c

pressure-volume derivatives that must vanish at the

Because the dataon argon to be used for comparison are
critical point; it also offers no difficulties at the limit of
large volume or at the smallest volume where the equa- given in terms of reduced pressure, volume, and tem-
tion is to be used. This is accomplished by merely perature or:
adding a constant to the volume terms in Equation 1, or, /OP*, VR, Tr) 0 =
(17)
it is convenient to put Equation 16 into reduced form by
RT a
P =
(15) appropriate multiplications and divisions by Pc, Vc. and
V -
b + c (v + cj2 Tc, or:

P RTVcTc 27 R2T2V2
with a and b given by Equations 8 and 11. This
to be
idea is not new, as Clausius (<S) did the same thing Pr 64 P2V2{V + e)2
PcVcT,
shortly after van der Waals proposed his equation, but
the technique was not fully appreciated, possibly because (18)
Clausius offered a kinetic explanation that was only Since PR =
P/Pc, F* =
V/Vc, TIt =
T/Tc, and zc =

partially justified, rather than a mathematical explana- PcVjRTc,

tion based on a linear transformation of the independent
variable. Now if c is set equal to (zCvdw/zCt) Vc =
Tr
Vc Ph
—

0.29 Vc, the critical point of Equation 15 will fall pre- 0.125

cisely on the true critical point and the calculated zc

will be 0.291. Unfortunately when this is done, the

40 INDUSTRIAL AND ENGINEERING CHEMISTRY

Next, let c/Ve V25 zc 0.04/zc, which makes the
= =
Evaluating this at the critical point where VB =
1,
calculated zc 0.335, and the equation becomes:
= and with zc 0.291 for argon, we have,
=

tr 27
4.85
Pr 0.085N
(20) (23)

Zc /
64
7-7-7 This is obviously much lower than 6.0, so that it is
It is emphasized that this equation fits the critical point necessary to add something to the second term to make it
exactly with (dP/dV)Tc,vc (d2P/bV2)TciVc 0 only
— =
temperature-dependent. Again referring to Figure 2,
if zc 0.335; however,
=
as shown in Figure 9, Equation the age-old observation is made that all isometrics are
20 fits the actual critical point (where zc 0.291) very =
almost linear. As pointed out years ago (23), the
closely, but obviously with slightly incorrect values of critical isometric seems to be truly linear, the isometrics
Pc, (bP/bV)TCyVc and (b2P/bV2)TctVc, which were used to larger than the critical curve down slightly, those
develop it. This situation is the same as is evident between Vc and about Vc/1.8 curve up, and those whose
with the Redlich-Kwong equation (33). It is interesting volumes are less than Ve/1.8 are almost straight or curve
in connection with the latter that emphasis was laid on down a little. These facts hold if the temperature is
having b in the V b term equal to 0.26 Vc, supposedly
—
less than TB 2-2.5. For temperatures above TR
= =

on theoretical grounds, but when the application to 2-2.5, it appears that all isometrics curve down (23).
argon is made, b must equal 0.298 Vc or the results will be Since the curvature effects are small and since we are
way off. In short, it is ridiculous to say that b has been trying to develop a simple low precision equation, we
set theoretically. shall assume that all isometrics are linear. This means
Once the constants in Equation 20 have been set to that a linear temperature term is arbitrarily inserted
fit the critical isotherm, the next step is to determine the into Equation 20 as:
variation with temperature away from the isotherm. 21 (A + BTr)
Examination of Figure 2 shows that the critical volume Pr =
(24)
line is straight with a slope equal to the vapor pressure
0.085 {
64 z*(vB +
0.04\2
—

)
curve at the critical point, as mentioned previously (23). Zc

Also vapor pressures are correlated on a generalized plot Differentiating this with respect to temperature gives:
with the parameter being their slopes at the critical point
as shown in Figure 4. These curves can best be repre- 27 B
sented by the high precision vapor pressure equation: , 0.04V
64 z* ( VB + —

)
In P = A + ~

+ Cln T + DT +
(25)
E(F -

T) In At the critical point, this derivative should be about

T
(F -

T), (21)
6.0, so with VB 1 and zc 0.291:= =

thoroughly discussed by Martin, Kapoor, and Shinn 4.85 0.385 B =

6.0 or B = -0.3 (26)
-

(25). The utility of this plot cannot be overemphasized.

Its parameter is the key to the PVT behavior along the Since at TB = 1, A + BTR must be unity to fit the criti-
critical isometric, it shows how the complete vapor cal isotherm according to Equation 20, we can write:
pressure curve of any substance is determined from one
A -
0.3 (1) = 1 or A =
1.3 (27)
vapor pressure point and the critical point, and it is the
basis of one of the more precise generalized correlations Since the final equation is not too sensitive to A and B,
of latent heat of vaporization (22). we shall round these off to 4/3 and 1/3, respectively, and
In the case of argon the normal boiling point is 87.3° insert them into Equation 24 to obtain:
K. and the critical constants according to Shah and
Thodos (34) are Pc 48.34 atm., Vc 74.52 cc./gram 9(4 -

Tx)
= =
Pr =

0.085\ (28)
and K. At the
mole, Tc 150.88°
0.0207 and \/TB
=
boiling point then,
1.73. In Figure 4 this point /
64 ZJ VR + —Y
PB = =
zc , z„ /
lands practically on the curve with —M 6.0 (dPR/ = =

To test our handiwork in developing Equation 28, it is

dTB)CIitpt, Now if Equation 20
respect to temperature,
is differentiated with
compared with the data on argon which have been
assembled by Costolnick and Thodos (7). Reduced
pressures can be read from their graphs at round values
of pB( 1/Vr) and TR to a precision of about 1%. Table
I gives the comparison between the data points read

VOL. 59 NO. 12 DECEMBER 1957 41

 Table L Reduced Pressure for Argon
Tr 0.8 0.85 0.90 0.95 1.0 1.2 1.4 1.6 2.0 2.5
PR

0.1 0.232“ 0.250 0.268 0.286 0.305 0.378 0.452 0.523 0.672 0.855
0.232b 0.250 0.269 0.287 0.306 0.380 0.454 0.528 0.676 0.861
0.229c 0.248 0.267 0.286 0.305 0.380 0.454 0.528 0.674 0.855

0.2 0.462 0.500 0.538 0.700 0.850 1.00 1.31 1.70

0.462 0.502 0.541 0.700 0.858 1.017 1.334 1.731
+ 1.7 + 1.8 + 1.8
0.457 0.499 0.540 0.703 0.862 1.019 1.327 1.707
—

1.1** + 1.4 + 1.9 + 1.3

0.4 0.75 0.83 1.20 1.58 1.92 2.65 3.47

0.751 0.840 1.199 1.558 1.919 2.636 3.533
+ 1.2 -1.4 + 1.8
0.742 0.839 1.214 1.575 1.926 2.611 3.445
—
1.1 + 1.1 + 1.2 -1.5

0.6 0.970 1.59 2.20 2.78 3.93 5.33

0.969 1.572 2.174 2.776 3.980 5.486
-1.1 -1.2 + 1.3 +2.9
0.968 1.605 2.212 2.798 3.931 5.301

0.8 0.999 1.90 2.80 3.70 5.40 7.50

0.999 1.900 2.780 3.670 5.450 7.676
+ 2.3
0.999 1.949 2.847 3.712 5.373 7.373
+2.6 + 1.7 -1.7

1.0 1.00 2,20 3.40 4.63 7.03 10.0

1.004 2.232 3.459 4.687 7.143 1 0.21
+ 1.5 +1.7 + 1.6 + 2.1
1.003 2.32! 3.564 4.755 7.037 9.775
+ 5.5 + 4.8 + 1.6 -2.3

1.2 1.00 2.63 4.30 5.90 9.10 13.2

1.063 2.686 4.308 5.931 9.176 13.233
+ 6.3 + 2.1
1.060 2.81 1 4.457 6.033 9.047 12.655
+ 6.0 +6.9 + 3.7 + 2.3 -4.1

1.4 1.10 3.20 5.45 7.55 11.9 17.03

1.273 3.359 5.445 7.531 11.70 16.92
+ 15.7 + 5.0 -1.7
1.267 8.525 5.649 7.680 11.56 16.2!
+ 15.2 + 10.2 + 3.6 + 1.7 -2.9 -4.8

1.6 1.38 4.10 7.05 9.95 15.6 22.3

1.754 4.390 7.027 9.663 14.94 21.53
+27.1 +7.1 -2.9 -4.2 -3.5
1.748 4.61 1 7.305 9.883 14.82 20.72
+ 25.9 + 12.4 +3.5 -5.0 -7.1

1.8 1.20 2.15 6.0 9.70 13.4 20.7 29.4

1.846 2.671 5.973 9.275 12.58 19.18 27.43
+ 53.8 + 24.2 -4.4 -6.1 -7.4 -6.7
1.738 2.681 6.276 9.664 1 2.91 19.13 26.59
+ 45.6 + 24.7 + 4.6 -3.7 -7.6 -9.6

2.0 1.50 2.70 4.00 9.00 14.0 1 8.8 28.0 38.6

2.209 3.240 4.272 8.396 12.52 1 6.64 24.89 35.20
+47.3 + 20.0 + 6.8 -5.6 -10.6 -
1 1.5 -1 1.1 —
8.8
1.952 3.158 4.336 8.338 13.09 17.18 25.04 34.48
+ 30.1 + 16.9 + 8.4 -1.8 -6.5 8.6 -10.6 -10.7

2.2 2.70 4.50 6.10 7.80 14.0 20.0 26.0 37.5 50.5
3.059 4.354 5.649 6.944 12.12 17.30 22.48 32.85 45.80
+ 13.3 -3.2 -7.4 -11.0 -13.4 -13.5 -13.5 -12.4 -9.3
2.618 4.166 5.677 7.157 1 2.83 18.21 23.40 33.40 45.47
-3.0 -7.4 -6.9 -8.2 -8.4 -9.0 -10.0 -11.2 -10.0
"
Data recorded in first line throughout table. Equation 28, this paper, recorded in second line throughout table. c
Equation 29, Redlich-Kwong,
recorded in third line throughout table. J Deviation percentage recorded in this position whenever 1.0 or greater.

42 INDUSTRIAL AND ENGINEERING CHEMISTRY

from the graphs and Equation 28. Since the Redlich-
(30)
Kwong equation

For linear isometrics (b2P/bT2)v is always zero, so

Equation 28 predicts the constant volume heat capacity
(Fb_ 00^64) will not change as the volume is varied at a given tem-
perature. Equation 29, on the other hand, gives
0.42748
(29) negative values of (d2P/dT2)v at all points so it predicts
0.08664V that the constant volume heat capacity always decreases
tr°-wvr Vr +
) as the volume increases, or conversely that starting with
the ideal gas heat capacity, Cv*, this quantity increases
has been strongly supported by a number of investiga- continuously as the density increases. Now neither
tors including Shah and Thodos (34), the predictions of prediction is in accord with the facts. In the few cases
this equation are also tabulated for comparison. (Note where Cv has been measured over a wide range of densi-
that 0.08664/zc 0.08664/0.291
=
0.298
=
b/Vc, =
ties at the critical temperature or higher, it has been found
or b 0.298 Vc, as mentioned earlier.) The percentage
= that Cv increases to a maximum at the critical density
deviations between the data and the equations are only and then decreases. On the basis of Figure 2 that
given where the deviation is greater than 1%, because (c>2P/bT2)v is negative for 0 < p < pc, zero for p —

of inability to read the graphs more precisely. It
is readily noted that the predictions of the two equa-
tions are very similar. Equation 28 is somewhat better
up through pR 1.6, but the Redlich-Kwong equation
=
is a little better at the highest densities. It is to be
observed that the two equations are really quite similar.
In Equation 28 the denominator of the second term has
the symmetric quantity, (VR + 0.04/zc)2, while Equa-
tion 29 has the asymmetric quantity, (VR + 0)(VR +
0.08664/zc). When these quantities are multiplied out,
the results are almost the same. There are two principal
advantages to Equation 28, however. It can be ad-
justed at the time the linear transformation is made to
account for substances with different values of zc, and
the slope term, A + BTR, can be readily varied to fit M,
the parameter of the generalized vapor pressure plot.
Since the two equations predict very similar critical
isotherms, it means they could have been made almost
identical everywhere by either of two operations: In
Equation 24, TR>J‘ might have been introduced into the
pc,
positive for pc < p < 1.8 pc, zero for p =
1.8 pe and
negative again for p > 1.8 pc for temperatures two to
two-and-a-half times the critical temperature, it follows
that Cv for the gas should increase with density to a
maximum at the critical density, then decrease to a
minimum at 1.8 pc and increase again to the highest
densities. This should be the character possessed by a
good equation of state.
Isothermal variations of other thermodynamic proper-
ties, such as enthalpy, entropy, and free energy or fuga-
city, are not sensitive to the derivative (d2P/d T2) v, as
shown by the thermodynamic relations which give their
changes. Consequently, calculation of these properties
will be almost the same whether one uses A -f- BTR
or Tr~0-5 in the second term of Equations 28 and 29.
One can generalize that some properties change because
of higher order effects such as second derivatives. Heat
capacity is one of these, while enthalpy and entropy are
not, as they depend on first-order effects (first deriva-
tives).

denominator instead of A + BTR in the numerator. One further observation concerning equations such as
This was actually tried, and the predictions were sub- Equations 28 and 29 should be made. As good as
stantially the same as for the Redlich-Kwong equation. these simple equations are in their overall performance,
Alternatively, in the Redlich-Kwong equation A + they fail by far to fit the data within the experimental
BTR might have been introduced into the numerator precision. Even up to the critical density errors of the
instead of TR°-6 in the denominator. order of 5% or more can occur. This is considered com-
By putting TR°-S in the denominator, all isometrics pletely intolerable for investigations such as those of
curve downward. This
advantageous at about pR
is =
Beattie (7) or Michels (28) in fitting their high precision
0.6 and the higher pR It is disadvantageous any-
s. data where the errors are far under 1%. Despite the en-
where near the critical volume or pR 1 because of its
—
thusiasm of some workers in the field for such simple
linearity. In fact, for overall performance up through equations, one must conclude they cannot be seriously
pR —

1.6, the assumption of linear isometrics is better.
In the calculation of thermodynamic properties, an
interesting comparison can be made for the constant
volume heat capacity. The change of Cv with volume
is given by:
regarded as more than a qualitative prediction of PVT
behavior. This is not to imply that simple equations
do not have value. They are useful in predicting trends
and they can fit data quite precisely over limited ranges,
but they definitely are not the tools to correlate good

VOL. 59 NO. 12 DECEMBER 1967 43

 PVT data within the experimental precision. Some- changing conditions in the neighborhood of the critical
thing much more versatile in mathematical complexity point, (bP/bV)T and (b2P/bV2)r, vanish when given
or inherent characteristics is required. In this connec- as pressure with respect to volume while they go to
tion the author has experimented for almost 20 years infinity if taken as volume, a function of pressure, a
with numerous short equations with a variety of kinds pressure-explicit equation, is chosen. This is consistent
of terms such as V b, V + c, e~Hl /v, e~a/v/V2, and taV",
—
with van der Waals choice for a simple equation. The
but has never discovered any simple form better than procedure will be to determine the equation along the
Equation 28. Of course, this does not mean that such critical isotherm and then introduce appropriate tem-
an equation cannot be found some day, but it is not perature functions so that it will apply at other tempera-
readily obvious. tures.
The second decision concerns the nature of the volume
Design of High Precision Equation of State function, which we will assume at the start to be the
The papers already mentioned lay the groundwork virial equation of Onnes (29).
for the development to be discussed here. The first
decision to be made is whether to make the equation RT
'
MT) '
MT) ' ' ' '
MT) (31)
explicit in pressure, volume, or temperature. The latter V v2 V3 V”
would give an equation extremely awkward to use in
practice, so it is discarded. Because of the approximate The equation will now be fitted to a set of data along the
linearity of the constant volume lines on the P-T plot critical isotherm. Michels’ high precision data on
and because the derivatives that determine the rapidly argon (27) or carbon dioxide (28) might well be chosen

Table II. Critical Isotherm

Pressures and per cent deviations

v,
ft.yib. p!fC.i.a. A B C C F F G

0.43215 279.26 279.41 0,05 273.06 -2.22 279.28 0.01 280.87 0.58 279.27 0.01 280.18 0.33 279.29 0.01
0.34537 340.55 340.71 0.05 331.40 -2.68 340.55 0.00 342.69 0.63 340.54 0.00 341.73 0.34 340.56 0.00
0.28758 398.49 398.67 0.05 386.12 -3.10 398.49 0.00 401.12 0.66 398.48 0.00 399.89 0.35 398.50 0.00
0.24717 451.97 452.13 0.03 436.24 -3.48 451.94 -0.01 454.95 0.66 451.93 0.01 453.48 0.33 451.95 -0.01
0.21536 504.85 505.04 0.04 485.56 —
3.82 504.86 0.00 508.14 0.65 504.86 0.00 506.45 0.32 504.87 0.00
0.19100 554.09 554.25 0.03 531.24 —
4,12 554.10 0.00 557.52 0.62 554.10 0.00 555.64 0.28 554.10 0.00

0.17706 586.54 586.67 0.02 561.25 —

4.31 586.54 0.00 589.96 0.58 586.55 0.00 587.98 0.25 586.54 0.00
0.17195 599.35 599.45 0.02 573.07 -4.39 599.33 0.00 602.74 0.57 599.34 0.00 600.72 0.23 599.23 0.00
0.15607 642.58 642-60 0,00 612.94 —
4.61 642.53 -0.01 645.78 0.50 642.55 -0.01 643.66 0.17 642.52 -0.01
0.11717 774.97 774.83 -0,02 735.60 -5.08 775.02 0.01 776.60 0.21 775.03 o.o r 774.47 -0,06 774.96 0.00
0.088931 897.93 897.49 -0.05 852.81 -5.03 898.01 0.01 896.10 -0.20 898,00 0.01 894,54 -0,38 897.96 0.00
0.072230 976.50 975.84 -0.07 933.45 -4.41 976,51 0.00 971.51 -0.51 976.47 0.00 970.72 -0.59 976.54 0.00

0.060707 1025.7 1025.0 -0.07 990.78 -3.40 1025.6 -0.01 1019.1 -0.64 1025.5 -0.01 1019.0 -0.66 1025.7 0.00
0.052686 1051.7 1051.4 -0.03 1027.8 -2.27 1051.8 0.00 1045.8 -0.56 1051.8 0.01 1046.0 -0.55 1051.8 0.01
0.047541 1062.9 1062.7 -0.02 1048.0 -1.41 1062.9 0.00 1058.5 -0.41 1063.0 0.01 1058.7 -0.40 1062.8 -0.01
0.034226 1070.6 1070.6 -0.00 1 070.6 -0.00 1070.6 0.00 1 070.6 0.00 1070.6 0.00 1070.6 0.00 1 070.6 0.00
0.023008 11 94.2 969.10 —
18.85 1301.5 8.98 1 003.3 —
15.98 1217.1 1.92 871.90 -26.99 1209.1 1.25 2509.7 110.2
0.020951 1468.5 773.97 -47.30 1653.8 12.62 901.77 -38.59 1490.2 1.47 360,12 -75.48 1475,9 0.50 8273.0 463.3

0.019189 2093.6 357.07 -82.94 2341.1 1 1.82 744.59 -64.43 2097.6 0.19 —
1087.5 -151.95 2082.4 -0.53 30016.8 1333.0
0.017730 3262.0 -387.86 -111.89 3501.8 7.35 587.62 -81.99 3240.2 -0.67 -4551.6 -239.53 3240.5 -0,66 95451.4 2826.0
0.016474 5289.9 -1629.4 -1 30.80 5370.2 1.52 560.36 -89.41 5244.1 -0.87 -12277.7 -332.10 5283.1 -0.13 271602.0 5034.0
0.015367 8623.6 -3604.2 -141.79 8279.1 -4.00 941,75 -89.08 8582.7 -0.47 -28680.3 -432.58 8668.2 0.52 71 1 349.7 8149.0
0.014384 13845.8 -6609.6 —
147.74 12651.5 -8.63 2255.9 -83.71 13867.2 -0.15 -61825.8 -546.53 13922.3 0.55 1737758.8 12444.0
0.013585 20830.0 -10531.6 —
150.56 18323.3 -12.03 4987.4 -76.06 20967.6 0.66 -1 17977.3 -666.38 20730.2 -0.48 3704751.6 17685.0

A, 5-term virial using data points up to critical [Vc —

0.034226). F, 9-term virial using all data points.

fir 5-Term virial using all data points. G, 1 1-term virial using data points to critical.
C, 7-term virial using data points to critical. H, 1 1-term virial using first 18 data points.
D, 7-term virial using all data points. /, 1 1-term virial using first 20 data points.
E, 9-term virial using data points to critical. J, 1 1-term virial using all data points.

44 INDUSTRIAL AND ENGINEERING CHEMISTRY

 for this is representative of good modern experimental
studies. Since the latter go to slightly higher reduced
densities, CO2 will be selected. Now the technique
of using a number of derivatives of pressure with respect
to volume at the critical point equal to zero or small
numbers, up to say (d4/)/dI/’4) T might be used as sug-
gested earlier (23), if PVT data are not available. Since
there are ample data, only the first two derivatives will
be used plus the critical point itself.
To calculate the constants in Equation 31, the actual
data points arc fitted by regression analysis (least
squares), subject to the three conditions at the critical
point, namely P€, (dP/dV)T 0, and (d2P/dV2)r 0.=
These three conditions arc best brought into the least
squares algorithm by means of Lagrange multipliers.
It is instructive to vary n in Equation 31 over a wide
range, such as from 5 to 17, to learn just how complex
an equation will be needed. Note that Michels (28)
only went to n 9 because=electronic computers were
not available when that work was done.
The results are noteworthy, as shown in Table II.
When we fit the data up to the critical density, using only
those data points up to the critical, the seventh power
virial is better than either higher or lower powers.
If all data points are used up to 2.52 times the critical
density, increasing the number of terms gives an in-
creasingly better fit over the whole range, with the 17th
power virial being the best. In the latter Case, how-
ever, the fit up to the critical density is slightly inferior
to that of the seventh power equation. For any given
equation such as the 11th power, for example, increasing
the number of data points in the least squares algorithm
=
extends the good range of the equation to cover the
points used, but sacrifices some of the high precision
fitting at the lower densities. The implication of these
results is that only terms up to I77 are necessary for fitting
data up to the critical density, but that many more
terms are needed if higher densities are to be considered.
To go to 2.52 pc with good precision over the whole
range requires at least 17 volume terms, which will

ressures for CO,

5r several equations

H f J K L M N

279.15 -0.04 279.77 0.18 279.92 0.24 279.54 0.10 279.41 0.05 279.69 0.15 279.60 0.1 1

340.37 —
0.05 341.19 0.19 341.41 0,25 340.89 0.10 340.71 0.05 341.21 0.19 341.06 0.15
398.26 -0.06 399.25 0.19 399.53 0,26 398.90 0.10 398.68 0.05 399.44 0.24 399.24 0.19
451.68 -0.07 452.77 0.18 453,13 0.25 452.39 0.09 452.15 0.04 453.24 0.28 452.96 0.21
504.58 -0.06 505.72 0.17 506,14 0.25 505.32 0.09 505.07 0.04 506.57 0.34 506.22 0,27
553.81 -0.05 554.92 0.15 555,40 0.24 554.54 0,08 554.31 0.04 556.28 0.39 555.84 0.31

586.27 -0.05 587.30 0.13 587.81 0.22 586.95 0.07 586.73 0.03 589.09 0.44 588.58 0.35
599.07 -0.05 600.07 0.12 600.59 0.21 599,72 0.06 599.53 0.03 602.03 0.45 601.50 0.36
642.31 -0.04 643.11 0.08 643.67 0.17 642.83 0.04 642.69 0.02 645.81 0.50 645.19 0.40
775.13 0.02 774.60 -0.05 775.17 0.03 774.68 -0.04 774.95 0.00 780.56 0.72 779.59 0.60
898.77 0.09 895,80 -0.24 896.16 -0.20 896,47 -0.16 897.54 -0.04 906.47 0.95 905.16 0,81
977.74 0.13 972.82 -0.38 972.83 -0.38 973.86 -0.27 975.66 -0.09 986.84 1.06 985.43 0.91

1026.8 0.11 1021.3 -0.43 1021.0 -0.46 1022.4 -0.33 1024.5 -0.12 1 036.0 1.01 1034.7 0.88
1052.6 0.08 1047.9 -0.36 1 047.5 -0.40 1048.8 -0.28 1050.6 -0.10 1060.1 0.80 1059.3 0.72
1063.4 0.04 1060.1 -0.27 1059.7 -0.30 1060.6 -0.22 1062.0 -0.09 1068.4 0.51 1068.0 0.48
1 070.6 0.00 1070.6 0.00 1070.6 0.00 1 070.6 0.00 1 070.6 0.00 1 082,3 1.09 1079.9 0.86
1194.6 0.03 11 95.5 0.10 1 203.8 -0.80 1 195.8 0.13 1 1 86.87 -0.61 1791.8 50.0 1669.9 39.8
1468.4 -0.01 1464.6 -0.27 1471.3 0.19 1465.0 -0.24 1462.2 -0.43 2460.9 67.5 2210.8 50.5

1796.7 -14.2 2097,3 0.18 2085.6 -0.38 2091.6 -0.09 2114.4 -0.99 3557.9 70.0 3072.9 46.8
952.42 -70.8 3260.6 —
0.04 3254.5 -0.23 3270.6 0.26 3338.0 2.33 5210.9 59.7 4328.0 32,7
-5479.6 -203.6 5015.2 -5.19 5294.4 0.09 5291.5 0.03 5402.9 2.14 7712.6 45.7 6148.9 16.2
-30358.5 -452 6871.5 -20.32 8641.0 0.20 8600.3 -0.27 8663.9 0.47 1 1568.1 34.2 8809.3 2.15
-105944.3 -865 6723.2 -51.44 13848.5 0.02 13870.5 0.18 13560.5 -2.06 1 7658.7 27.5 12726.3 —
8.09
-27791.4 —
1434 89.23 -99.57 20814.3 -0.08 20821.8 -0.04 19901.3 -4.46 26543.0 27.3 17941.8 -
13.9

K, 17-term virial using all data points.

ir Equation 43.
M, Equation 29, RedlicK-Kwong.
N, Equation 28 with linear shift from xc —
0.375 to 0.330—i.e.r constants become 0.080 and 0.045 in place of 0.085 and 0.040, respectively.

VOL. 59 NO. 12 DECEMBER 1967 45

For purposes of generalization and extrapolation a short virial equation is

preferable to a longer one

require about three times as many temperature function the critical temperature is about —0.34 for all com-
constants. Even with modern electronic computers this pounds. For simplicity, assume this to be —y/%. Then
gets to be exceedingly cumbersome and time-consuming
BPC 1
in calculations; furthermore, it would not be useful if a _

(33)
RTC 3
vast array of data were not available. The virial
equations fitted to the critical isotherm can be differen- where B, the second virial coefficient, is defined by the
tiated to give the magnitude of the third and fourth equation
pressure-volume derivatives at the critical point. For PV B C
the seventh power virial fitted up to the critical density, A H---|---b
=
(34)
~RT y y*
(d3p/d V3) Tr is minutely positive, being less than 0.06%
of the largest term in the seven terms that are summed Comparison of Equations 32, 33, and 34 at the critical
to give the derivative. For the 17th power virial fitted temperature shows that
R?T*
up to 2.52 pc, (d3p/dV3)Tr is very slightly negative, being MTC) BRTC -~±
=
(35) = —

less than 0.3% of the largest term in its 17-term summa- 3 * C

tion. Although calculations of high order derivatives

are extremely sensitive, one may conclude from these
small values and their opposite signs in two carefully
fitted equations that (b3p/bV3)r vanishes at the critical
point.
For the seventh power virial, (i^P/dC*),* is slightly
Figure generalized correlation of the third virial
6 is a
coefficient. It is probably not so good as the more
theoretical correlation of Gyorog and Obert (12), but
it is simpler. From Figure 6 the generalized third virial
coefficient is 0.036 at Tc, and this is simplified to 1/30> so
_

negative, but is only 3% of the largest term in its summa-

CP* 1
=
(36)
tion. For the 17th power virial, this derivative is very R*T* 30
slightly positive, being just 0.5% of the largest term in Comparison of Equations 32, 34, and 36 at the critical
its summation. Again because of the small values and
temperature shows that
their differing signs, one concludes that within the
R3T3
accuracy of the calculations (d*P/dVi)T also vanishes MTe) CRTc =
(37)
30 ~P?
at the critical point. The suggestion made earlier (23)
that the third and fourth derivatives at the critical point With fi(Tc) and fz(Tc) given by Equations 35 and 37,
are either extremely small or vanish is substantiated by one may use the two van der Waals derivatives to solve
these results. The value of this conclusion will be forfi(Tc) and fi(Tc) as
limited to use in longer equations, for it is obvious that a R*T*V* R3T3VC
short equation such as van der Waals can only utilize f*(Te) =
-RTeV* + (38)
2 Pc 20 P*
two derivative conditions or the constants will be over-
specified. Setting four derivatives equal to zero at the 3 RTCV* 4 R*T*V3 R3T3V* , ,

h(Tc) =
+ (39)
critical point requires at least four volume terms in the 15 Pc 50 P*
equation, exclusive of that involving R. When these expressions are substituted into Equation 32,
Although a lengthy virial equation can be fitted to the critical compressibility is determined as:
data with high precision, it is preferable for purposes of
generalization and extrapolation to have a much shorter p y
= 0.237 (40)
equation. Knowing that a seventh power virial does a RTC
superb job up to the critical density, an expansion partly For CO2 the true value of z, is:
in powers of V and partly in other functions, such as
exponentials, is suggested. To obtain such a combina- ^^-Vo.03423
tion, consider again Equation 31 up to V5 applied to the
(l070.6
\ sq. in./\ —)
lb.,,,/
critical isotherm: RTr lb., -
ft.3
RTC
'
fj(Te)
'
MTe) ,
4"
MTC) ,
'
f&(Tc)
' '
0.2438
sq. in.
-

lb.,„ x) (547.54° R.)

y y2 j/3 y4 j/s 0.274 (41)
The constants in this equation will be calculated by Since this is greater than 0.237, a linear transformation
using the two van der Waals derivative conditions at the in volume is made by subtracting a quantity b from each
critical point, (bP/bV)T = (b*P/bV*)r 0, plus some
= volume term in Equation 32, rather than by adding as
generalized information from Figures 5 and 6. Figure 5 was done in Equation 15, so the curve is moved to the
shows that the generalized second virial coefficient at right. The result is:

46 INDUSTRIAL AND ENGINEERING CHEMISTRY

P =
RT' + M^) + MTc) +
order of \1 /Vc, so for some new substance this is the
v -
b (v -

by (v -

by value to try.
The test of Equation 44 is to fit it to a set of data.
MTC)
+
/»(r„) An algorithm is written for these/’s to be determined by
(42)
(v -

6)4 (v -

by least squares subject to three conditions at the critical

where point, Pc, C6P/bV)Tc =
0, and (5'2P/dVr)Tr =
0, just as
RTC was done with the virial Equation 31. Various values
A (0 .274 0.237) =
-

Pc of a and b, between 500 ± 100, and 0.0047 ± 0.0047,

respectively, should be tried. Clearly, b 0 is a
=

(0.2438) (547.54) possibility, for this is the virial equation and it should
(0.037) 0.0047 (43)
1070.6 always be tested. When this is done, the best fit of the
and Vc in Equations 38 and 39 is replaced with 0.237 CO2 critical isotherm occurs for a =
490 and b 0.004.
=

Several alternate procedures for deriving a critical

RTc/Pc, since the/’s must be the same before and after
the volume shift. The equation has been studied in this isotherm equation deserve mention. If there are in-
form and is equivalent to a seventh power virial equation sufficient data to employ least squares after the assump-
tion of a and b, the generalized second virial coefficient
up to the critical density. It is, however, the extension
to higher densities that is sought here. The value of b
and the vanishing of the third and fourth pressure-
is only an order of magnitude number, as the next move volume derivatives at the critical may be used along with
is to add terms to Equation 42 which will carry it well the three conditions given above to solve for the six
beyond the critical density. f(Tc)’s. Another technique is to drop the third deriva-
The kinds of terms to be added may well be exponen- tive condition and use the generalized third virial co-
tial as they must make a large contribution for small efficient. A quite different procedure starts by setting b
to zero in Equation 44 and using the second and third
changes in volume. Furthermore, exponentials repre-
sent complete series, so one exponential may be the virial coefficients and the four pressure-volume deriva-
tives to solve for the /(7))’s. The virial coefficients
equivalent of a series of powers of volume. Two kinds
of terms that have interesting properties have been have to be assumed a little large in anticipation of intro-
tried. They are \/eaV" and e~a!v"/Vm. The former ducing b because of the relations f% =
(B—

b)RT and
with n 1 was chosen because it is slightly simpler,
—
3
=
(C + b2 /— 2 bB)RT between Equations 34 and 44.
a series of terms like itself to be formed, This procedure predicts the critical volume (note it was
permits simple
and damps out rapidly at large volumes where one never needed in the calculations) to be too small so a
wants Equation 42 to give the total answer. Two terms linear transformation is made. This is best done by
in the series ft/eaV + f-j/e2 aV + /sA8 aV + give ...
making the second virial coefficient of the equation
good representation to densities as high as 2.3 pc. With agree with that predicted by the generalized correlation.
two exponential terms the equation for the critical At that point the critical volume will still be predicted a
isotherm becomes: little incorrectly, but inserting the correct critical volume
gives a critical pressure that agrees with the data within
RTC MTe) h(Tc) a few hundredths of a per cent because of the flatness
"** + +
V -
b (V -

b)2 (V -

b)3 of the critical isotherm at the critical point.

The results of the least square procedure are tabu-
MTe) MTc) MTc) MTc) lated in Table II, and they are encouraging. The
^ }
{V —

b)A ~(F —

by eaV e2aV
K

seven-term Equation 44 is greatly superior to the seven-

It is observed that a and b occur nonlinearly, so there is term virial Equation 30 when compared over the whole
no simple solution for them. We do know that b is of the range. It is as good as the 11-term virial up to 1.8 pc
order 0(0.0047). To determine a, pressures above the and better for higher densities. In fact, up to 1.8 pc,
critical density are calculated with Equation 42 and com- Equation 44 is as good as the 17-term virial, but not as
pared with the actual data by plotting A Raot Pcalcd = —

good above this point. In this range from 1.8 pc to 2.3

vs. V on semilogarithmic coordinates. The slope of this pc, the pressure rises so rapidly that deviations of several
graph as V approaches the critical from the low side is per cent are not considered unsatisfactory. These
—

a, because in this range only the/e term is needed, so

results indicate that only moderately long equations
that: can do a good job and that it may be advantageous to

MTc) try to develop closed equations from statistical mechanics

Pant ~
Picalcd + rather than to develop many virial coefficients up to the
17th or more. For the sake of completeness of Table II,
MTc) the pressures determined from Equations 28 and 29
or *
p ant, _
*
p o.alod = A _
“*

have been tabulated. Although neither is very good

above the critical density, the linear shift of Equation
or In A = In A —
aV (44) 28 is distinctly superior to the Redlich-Kwong Equation
For such graph for CO2, a is found to be about 500,
a 29.
which meansthe desired relation is Equation 44 with Equation 44 is only for the critical isotherm and is not
a =
0(500) and b 0(0.0047). In general, a is of the
=
complete until the temperature functions are decided

VOL. 5 9 NO. 12 DECEMBER 1967 47

 Table III. Comparison of Equation 48 and
Vol., Temp., Pressure, p.s.La. Vol., Temp., Pressure, p.s.i.a.
ft.3/lb. Exptl. Caled.“ Dev., % ft,3/lb. Exptl. Calcd.a Dev. %
0.62113 525.89 93.57 93.54 -0.03 0.33279 646.49 211.76 211.77 0.00
0.621 13 536.69 96.08 96.03 -0.05 0.33279 664.80 21 9.74 21 9.74 0.00
0.621 13 545.69 98.15 98.10 -0.06 0.33279 671.69 222.69 222.72 0.01
0.62113 563.69 102.24 102.19 -0.05 0.33279 692.39 231.60 231.61 0.01
0.62113 581.69 106.30 106.25 -0.04 0.33279 716.69 242.00 241.96 -0.02
0.62113 605.09 111.50 111.48 -0.02 0.33279 761.69 260.94 260.88 -0.03
0.62113 626.69 116.26 116.25 -0.01
0.62113 646.49 120.60 120.59 -0.01 0.27970 563.69 199.94 200.03 0.05
0.62113 664.80 124.61 124.58 -0.02 0.27970 581.69 210.21 210.23 0.01
0.62113 671.69 126.08 126.08 0.00 0.27970 605.09 223.23 223.23 0.00
0.62113 692.39 1 30.57 1 30.55 -0.01
0.27970 626.69 235.00 235.02 0.01
0.62113 716.69 135.83 135.78 -0.04
0.27970 646.49 245.66 245.67 0.01
0.62113 761.69 145.44 145.38 -0.04 0.27970 664.80 255.37 255.40 0.01
0.27970 671.69 259.02 259.04 0.01
0.48316 525.89 116.13 116.13 0.00 0.27970 692.39 269.87 269.88 0.00
0.48316 536.69 1 19.50 119.46 -0.03 0.27970 716.69 282.48 282.47 -0.01
0.48316 545.69 122.25 122.21 -0.04 0.27970 761.69 305.52 305.46 -0.02
0.48316 563.69 127.70 1 27.65 -0.04
0.48316 581.69 133.08 133.03 -0.04
0.24184 581.69 234.87 234.97 0.04
0.48316 605.09 139.97 139.94 -0.02
0.24184 605.09 250.44 250.46 0.01
0.48316 626.69 146.27 146.25 -0.01
0.24184 626.69 264.44 264.47 0.01
0.48316 646.49 151.97 151.97 0.00
0.24184 646.49 277.08 277.11 0.01
0.4831 6 664.80 157.20 157.22 0.01
0.24184 664.80 288.62 288.64 0.01
0.48316 671.69 159.17 159.18 0.01
0.24184 671.69 292.92 292.94 0.01
0.48316 692.39 1 65.07 165.06 -0.01
0.24184 692.39 305.76 305.77 0.00
0.48316 716.69 171.96 171.91 -0.03
0.24184 716.69 320.64 220.65 0.00
0.48316 / 61.69 184.55 1 84.49 -0.03
0.24184 761.69 347.84 347.78 -0.02

0.39571 536.69 141,05 141.07 0.01

0.39571 545.69 144.56 144.54 -0.01 0.21454 605.09 274.18 274.78 0.04
0.39571 563.69 151.45 151.40 -0.03 0.21454 626.69 290.42 290.47 0.02
0.39571 581.69 158.22 158.17 -0.03 0.21454 664.80 318.29 318.34 0.02
0.39571 605.09 166.87 166.84 -0.02 0.21454 671.69 323.25 323.30 0.01
0.39571 626.69 174.73 174.73 0.00 0.21454 692.39 338.01 338.06 0.02
0.39571 646.49 181.89 181.88 0.00 0.21454 716.69 355.13 355.17 0.01
0.39571 664.80 188.43 188.43 0.00 0.21454 761.69 386.44 386.31 -0.03
0.39571 671.69 190.88 190.87 0.00
0.39571 692.39 198.20 198.21 0.00 0.20633 263.17
581.69 263.48 0.12
0.39571 716.69 206.79 206.74 -0.02 0.20633 605.09 282.17 282.28 0.04
0.39571 761.69 222,43 222.37 -0.03 0.20633 626.69 299.19 299.25 0.02
0.20633 646.69 314.52 314.53 0.01
0.33279 545.69 1 66.08 166.14 0.03 0.20633 664.80 328.42 328.46 0.01
0.33279 563.69 174.56 174.55 -0.01 0.20633 671.69 333.59 333.65 0.02
0.33279 581.69 182.87 182.84 -0.02 0.20633 692.39 349.08 349.1 1 0.01
0.33279 605.09 193.45 193.43 -0.01 0.20633 716.69 367.00 367.02 0.01
0.33279 626.69 203.06 203.06 0.00 0.20633 761.69 399.69 399.62 -0.02

48 INDUSTRIAL AND ENGINEERING CHEMISTRY

Experimental Data for Chlorodifluoromethane
Vol„ Temp., Pressure, p.s.i.a. Vol., Temp., Pressure, p.s.i.a.
°
fl.3/lb. Exptl. Calcd." Dev. % ft.3/lb. R. Expfl. Calcd.0 Dev., %
0.17165 605.09 320.83 321.14 0.10 0.093570 646.49 526.75 526.99 0.05
0.17165 626.69 342.29 342.43 0.04 0.093570 664.80 564.36 564.25 -0.02
0.17165 646.49 361.50 361.55 0.01 0.093570 671.69 578.21 578.09 -0.02
0.17165 664.80 378.88 378.94 0.02 0.093570 692.39 619.31 619.15 -0.03
0.17165 671.69 385.38 385.42 0.01 0.093570 716.69 666.33 666.42 0.01
0.17165 692.39 404.62 404.70 0.02 0.093570 761.69 751.18 751.81 0.08
0.17165 716.69 426.91 426.99 0.02
0.083890 646.49 552.97 553.55 0.10
0.17165 761.69 467.53 467.47 -0.01
0.083890 664.80 596.62 596.40 -0.04
0.083890 671.69 612.59 612.31 -0.05
0.14254 605.09 360.36 360.10 0.20
0.083890 692.39 659.90 659.52 -0.06
0.14254 626.69 387.64 387.95 0.08
0.083890 71 6.69 713.90 713.86 -0.01
0.14254 646.49 411.95 412.04 0.02
0.083890 761.69 811.19 811.97 0.10
0.14254 664.80 433.86 433.91 0.01
0.14254 671.69 442.02 442.05 0.01 0.081554 646.49 559.43 559.94 0.09
0.14254 692.39 466.18 466.24 0.01 0.081554 664.80 604.69 604.39 -0.05
0.14254 716.69 494.02 494.17 0.03 0.081554 671.69 621.33 620.89 -0.07
0.14254 761.69 544.73 544.79 0.01 0.081554 692.39 670.35 669.84 -0.08
0.081554 716.69 726.50 726.18 -0.04
0.12202 626.69 425.52 426.07 0.13 0.081554 761.69 827.29 827.89 0.07
0.12202 646.49 455.22 455.37 0.03 0.075770 646.49 574.52 575.58 0.18
0.12202 664.80 481.93 481.95 0.00 0.075770 664.80 624.81 624.42 -0.06
0.12202 671.69 491.78 491.83 0.01 0.075770 671.69 643.08 642.55 -0.08
0.12202 692.39 521.12 521.17 0.01 0.075770 692.39 697.01 696.34 -0.10
0.12202 716.69 554.88 555.02 0.02 0.075770 716.69 758.53 758.25 -0.04
0.12202 761.69 616.03 616.26 0.04 0.075770 761.69 869.15 870.00 0.10

0.066950 646.49 596.40 597.72 0.22

0.12047 646.49 459.68 458.94 0.05
0.066950 664.80 655.61 654.92 -0.11
0.12047 664.SU 485.80 485.95 0.03
0.066950 671.69 677.26 676.16 -0.16
0.12047 671.69 495.83 495.99 0.03
0.066950 697.39 740.50 739.18 -0.18
0.12047 692.39 525.67 525.80 0.03
0.066950 716.69 812.83 811.73 -0.14
0.12047 716.69 559.90 560.19 0.05
0.066950 761.69 942.35 942.74 0.04
0.12047 761.69 621.92 622.41 0.08
0.054760 664.80 694.41 693.29 -0.16
0.10590 626.69 458.33 459.32 0.22 0.054760 671.69 722.63 720.66 -0.27
0.10590 646.49 494.18 494.42 0.05 0.054760 692.39 804.70 801.94 -0.34
0.10590 664.80 526.23 526.23 0.00 0.054760 716.69 898.17 895.67 -0.28
0.10590 671.69 538.05 538.05 0.00 0.054760 761.69 1065.12 1065.25 0.01
0.10590 692.39 573.11 573.13 0.00
0.044478 664.80 716.92 715.81 -0.15
0.10590 716.69 613.35 613.54 0.03
0.044478 671.69 753.69 751.11 -0.34
0.10590 761.69 686.25 686.60 0.05
0.044473 692.39 860.09 856.13 -0.46
0.044478 716.69 981.48 977.67 -0.39
0.099210 646.49 511.51 511.85 0.07
0.044478 761.69 11 99.99 1198.62 -0.11
0.099210 664.80 546.42 546.42 0.00
0.099210 671.69 559.34 559.27 -0.01 0.035992 664.80 723.72 723.37 -0.05
0.099210 692.39 597.39 597.37 0.00 0.035992 671.69 770.38 768.59 -0.23
0.099210 716.69 641.13 641.26 0.02 0.035992 692.39 906.48 903.78 -0.30
0.099210 761.69 720.05 720.56 0.07 (Continued on next page)

VOL. 59 NO. 12 DECEMBER 1967 49

 Vol., Tomp., Pressure, p.s.i.a. Vol., Temp., Pressure, p.s.i.a.
°
H.*/lb. Exptl. Calcd.a Dev., % ft. 3/lb. R. Exptl. Calcd.° Dev. %
0.035992 716.69 1064.54 1061.28 -0.31 0.029308 761.69 1546.22 1541.09 -0.33
0.035992 761.69 1353.43 1350.07 -0.25
0.023824 664.80 731.53 731.60 0.01
0.029308 664.80 724.18 724.44 0.04 0.023824 671.69 806.78 810.46 0.46
0.029308 671.69 780.03 782.26 0.29 0.023824 692.39 1043.07 1049.45 0.61
0.029308 692.39 952.74 956.26 0.37 0.023824 71 6.69 1330.71 1333.55 0.21
0.029308 716.69 1159.10 11 60.96 0.16 0.023824 761.69 1879.29 1868.03 -0.60
0
Equation 48 constants for chlorodifluoromethane:
R = 0.124098 Ct = -1.845051 X 10-4 C3 = 1.483763 b = 0.002
A2 =
-4.353547 Ae = 1.363387 X 10s A4 = 2.310142 X 10-3 k =
4.2/7,
fi5 =
2.407252 X 10~3 Bs = -1.672612 X 106 B4 = -3.605723 X 10"6 Te = 664.50° R.
C2 = -44.066868 Cs = 0 C4 = 0 Pc = 721.906 p.s.i.a.
A, = -0.017464 ^7 =
87 =
Ct = 0 A5 --3.724044 X lO^6 Vc = 0.030525 fl.3/lb
B3 = 7.62789 X 10-5 a =
548.2 Bs =
5.355465 X 10“8 °
R. -
°
F. + 459.69

upon and their constants evaluated. As discussed at e3

avterm but further research is required to obtain this.
some length (23), such functions might be: There are 18 constants in the temperature functions
of Equation 48, exclusive of k, which must be evaluated,
ft =
At + BtT + Cte-*T (45)
along with a and b that require the trial indicated for
Equation 44 along the critical isotherm. This, at
or ft =
At+ BtT + (46)
first, seems to be a rather formidable task, but it proves
to be relatively straightforward. When the critical
Cl Dt
or ft -
At + BtT + ^ (47) isotherm has been successfully fitted, a and b, and f2(Tc)
through f7(Tc) are known. This calls for a net of 18 —6
or 12 new facts or conditions. There are a number of
or evencomplicated. If Ct and Dt are taken as
more

zero, it is seen that linear isometrics would result, which ways to select these, but the following is a reasonably
would not be good in a high precision equation. Beattie simple way based on previous (23) discussions of the
and Bridgeman (7) used Equation 46 with n 2, and —
subject. All that is required is Pc, Vc, Tc, and one point
Rubin (2) followed suit. Claitor on the vapor pressure curve. If experimental PVT
Benedict, Webb, and
and Crawford (5) used Equation 47 with m 2 and = data are available, the predictions of the final equation
are compared with them and adjustments are made in
n =
4, and this was adopted by Bloomer and Rao (4).
Earlier studies (23) indicated Equation 45 as a good the several facts listed:
overall compromise, so it is used here with k ~ 5/ Tc. (1) From Figure 5 read the temperature at which
The resulting final form of equation is: B = 0. This is the Boyle temperature TB, a function of
Tc, as shown previously (23). It ranges from 2.1 Tc
RT A2+ B2T + CIe~*r for H20 to 3.3 Tc for H2, but is about 2.3 Tc for most
^ h
Vb
"

(V by -

substances.
Az + BtT + C3 e-kT A4 + B4T + C4 (2) From Figure 5, read the temperature at which
+ + ~

BPC/RT = ze 1. This is the T' temperature in the

—

(v by -

(v by -

earlier paper (23) and is very close to 0.8 Tc, as BPJRT

At + BbT + C6 e~*T As + BsT + C6 r** at TR < 1 is a function of zc.
+ + ~

(V by eav (3) From Figure 6, read off the third virial coefficient
-

at Tb, the Boyle temperature.

A, + B7T + C7
e2 aV
(48) (4) From Figure 6, read off the third virial coefficient
at T', the temperature of fact 2.
which is seen to be the same as that proposed earlier (23) (5) From Figure 4, read off the slope of the critical
with the addition of the eaV and e2 aV terms. The earlier isometric, (dPR/dTR)clit pt, which when multiplied by
equation was only good up to about 1.5 pc. The addi- PJ Tc gives (dP/dT) Vc.
tion of the exponential terms permits the equation to (6) Set (b2P/bT2)v = 0 at pc.
go to about 2.3 pc. This may not appear to be a big (7) Set (bP/bT2) r 0 at 1.8 pe. For extremely low
=

gain, but it actually is very significant because of the boiling substances this condition might have to be set
rapid rise of pressure with decrease in volume, as men- at 1,3 to 1.5 pc if high TRs are to be fitted.
tioned above. It would be highly desirable to have a (8) Set the slope of the 1.8 pe line as three-and-a-half
good equation to above 2.5 p„ such as by adding an times the slope of the pc line from fact 5. For low boiling

50 INDUSTRIAL AND ENGINEERING CHEMISTRY


Figure 10. Behavior of the Flory and Yang-Yendall equations
substances, this slope is to be set at the density of fact 7.
For example, if the 1.5 pc line is taken as linear, its slope
is about twice that of the pc line.
(9) Set the slope of the 2.0 pc line at T —* <*> as four
times the slope of the pc line. Alternatively, if a satu-
rated liquid point is known—i.e., P, V, and T—at about
2.0 pc, set this point instead of the slope.
(10) The high temperature slope (dP/'6T)v for some
isometric, V = aVe, is greater than the ideal gas slope,
R/V, by a ratio /S, or (dP/dT)v = jSR/V =
0R/aVe.
For most substances when a ~ 1.5, (8
=
a, so (dP/dT)v
=
R/Vc at 1.5 Ve.
(11) Set the generalized second derivative, b-Pg/bTg*
=
7)7/100 Pc at 1.4 pc and Tc. For low boiling sub-
stances this condition should be taken at pc < p < 1.3 pc.
(12) Set the generalized second derivative, hfPg/i>TB2
= —
7)7/50 Pc, at 2.1 pc and Tc.
Facts 1 and 2 above combined with the known ft(Tf)
=
A2 + B2TC -)- Cye~kTc perntit solving for Ay, B2,
and C2. Facts 3 and 4 plus the known fs(Tc) yield As,
Bs, and C». The four second-derivative facts, 6, 7, 11,
and 12, involve only the remaining C terms of which
there are just four, so C4, C5, C%, and C-, can easily be
calculated. The four slope facts, 5, 8, 9, and 10 involve
21’s and C’s, but since the C’s have already been deter-
mined, Ba, Bs, Bs, and By can be found. All that re-
mains are the four A terms. From the known B’s and
C’s and fi,(Tt), fs,(Tc), ftJ,Te), and fy(Tc), the values of
Ai, As, As, and Ay are obtained. Thus, all the con-
stants in the equation of state are found from simple
linear matrices of four equations and four unknowns.
The only time a more complex procedure occurs is in
using least squares to obtain the six /(Te) terms subject
to three conditions introduced by Lagrange multipliers,
and this introduces nine linear equations with nine
unknowns.
When all the constants have been determined and the
known data points calculated, the equation may be re-
fined by trying small changes in the input facts.
The curvature constant, k ~ 5/Tc, is the first to vary,
and it has a rather broad effect in all ranges except at
high temperature. Facts 1, 2, 3, and 4 have the greatest
impact at low and medium densities with 1 and 3 being
most important at high temperature and 2 and 4 being
most important at low temperature. Fact 5 has precise
control along pc and on either side of pc. Facts 7, 9,
10, and 11 have their greatest effects, as might be
expected, at the densities where they are applied.
With this knowledge of where the facts have their greatest
effect, it is not difficult to improve the equation in
specific regions. In the event that no actual PVT data
are available, the equation should be used to calculate z
and Pic at round values of Ts. The results should be
compared with the generalized compressibility chart,
Figure 3. This is an extremely important check—so
much so that probably it should be made even when
PVT data are available, as it often reveals erroneous
trends in the equation or the data itself.
As an example of this check, it is instructive to examine
the recently proposed Yang and Yendall (39) equation,
1
P =
(49)
V"
where P =
P/P*, V V/V*, and T
=
T/T*. The=
constants, P*, V*, and T* are the characteristic pressure
volume, and temperature of a given substance and are
determined either arbitrarily or from physical properties
such as the coefficient of compressibility. With m = 2/3
and n 2, the equation is that which Flory (9) developed
=
specifically for the liquid phase. Equation 49 may be
rewritten as
AT a
P =
(50)
V -
BVm V*
with A = P*V*/T*, B V*l~m, and a = P*V*n. =
For this to represent the gas phase and to reduce to the
ideal gas law at low pressure, Yang and Yendall set
A =
R. They then applied the equation to 11 different
substances using empirical procedures to obtain the four
constants. They show good fits for both gas and liquid
phases and so use the equation to calculate vapor pres-
sure and latent heat. This may appear to be the
answer to the need for a simple wide range equation of
state.
Unfortunately, the equation has a serious fault dis-
covered long ago when the author studied variations of
van der Waals equation on a strictly empirical basis.
By letting the constant b of Equation 1 be the variable
BVm where m < 1, and by letting n be less than 2 (Yang
and Yendall have m ~ 0.5 and n 1.8), the values of z
along the critical isotherm appear as in. Figure 10.
The anomalous behavior of z exceeding unity as the
pressure is reduced is not serious, practically, if one is
willing to accept deviations in calculated pressures of 1
or 2%; however, the behavior is quite unsatisfactory on
a theoretical basis, particularly when one notes that the
second virial coefficient is predicted to be infinity, since
VOL 59 NO. 12 DECEMBER 1947 51
the slope of the critical isotherm on the z-chart is infinity
at zero pressure. As mentioned earlier, this slope should
be very close to —0.34. Also integrations from zero
pressure, done when using ideal gas heat capacities or
when calculating fugacities, will be in error.
An alternative method of solving for the A, B, C
terms is by using least squares if sufficient data are
available. It should not be assumed, however, that such
a solution is automatically better than the trial proce-
dure using the 12 facts, because it tends to make the
predictions equally good in all regions, whereas one
usually seeks greater precision at lower densities. In a
trial procedure, one region can be made very good at the
expense of another. Of course, the least squares proce-
dure can do the same job by proper weighting of data in
certain regions, but this becomes as much of a trial as the
other procedure. Furthermore, the trial procedure is
quite informative in revealing the important facts and
terms in the equations as they affect different regions.
The least squares procedure reveals very little about the
relative importance of different constants. Also the
least squares procedure requires a solution of a 24 by 24
matrix (18 constants plus six Lagrange multipliers),
and round-off errors get to be significant, giving variable
results.
Note that the equation does not require vapor pressure
data except for the single point to establish the vapor
pressure curve on Figure 4. Even this can be eliminated
if critical isometric data are known. The final equation
can be used to predict the vapor pressure by a trial
procedure. A number of pressure points are calculated
along an isotherm for a range of volumes that bracket
the gas and liquid regions. High and low volumes
giving the same pressure are then used as limits in
VdP and when this integral vanishes, the correct
/ Vgas
V liq
pressure and gas and liquid volumes have been found
for the given temperature. This will only be good for
liquid densities as high as about 2.3 pc. As mentioned
earlier, it would be highly desirable to have an equation
that would be good to higher densities, so the calculated
vapor pressure range could be extended.
As an example of application of Equation 48, the
data of Michels (27) for chlorodifluoromethane have
been fitted. The equation is actually a simplified
version of Equation 48, as C4, C6, and fi{T) have been
taken as zero. The results are shown in Table III,
where it will be observed the predictions of the equation
are quite good. Thus, the equation can be used with
confidence to calculate all thermodynamic properties
that depend upon it.
Conclusions
A variety of objectives is possible in developing an
equation of state. An equation to satisfy one objective
may fail to satisfy another, and until a perfect equation
is discovered, no single equation satisfies all objectives.
At the present time there are simple equations, such as
Equation 28 or its equivalent, that cover a wide range
52 INDUSTRIAL AND ENGINEERING CHEMISTRY
of density with low precision, while there are complex
equations, such as Equation 48 or its equivalent, that
give high precision results over a fair range of density.
In the case of Equation 48, it is better than any other
equation because it goes to 2.3 pc. There is plenty of
room for improvement, however, for what is needed is
a high precision equation that is good
up to 2.5 or even
3.0 pc. It is hoped that this review of the field will
stimulate further work in this direction. Two general
routes are open to new investigators: One is to
carry
on with the empirical approach
employed here and the
other is to utilize statistical mechanics and the intermo-
lecular forces. The latter should consider both simi-
lar and dissimilar molecules, but it must be carried
well beyond the third virial coefficient of today’s calcula-
tions. In fact, because of the inability of short virial
equations to do an adequate job, it is recommended
that special attention be given to developing closed
equations from statistical mechanics as discussed in the
companion paper (20).
An additional field of research is concerned with
mixtures. Currently, combining rules for the constants
in Equation 48 are being studied. Because of the
generalized procedure for obtaining the constants for
the pure substances, there is greater consistency in the
mixture constants than in many other equations. We
expect to complete this study in the next year.
LITERATURE CITED
(1) Beattie, J. A., Bridgeman, O. C., Proc. Am. Acad. Arts Sci. 63, 229 (1928).
(2) Benedict, M., Webb, G. W., Rubin, L. C., J. Chem. Phys. 8, 334 (1940).
(3) Berthelot, D. J., J. Phys. 8, 263 (1899).
(4) Bloomer, O. T., Rao, K. N., III. Inst. Tech. Bull. 18, Chicago (1952).
(5) Claltor, L. C., Crawford, D. ft,, Trans. A.S.M.E. 71, 885 (1949).
(6) Clausius, R., Ann. Phys. Chem. IX, 337 (1880).
(7) Costolnick, J. J., Thodos, G., A.I.Ch.E. J. 9, 269 (1963).
(8) Dieterici, C., Ann. Phys. 69, 685 (1899).
(9) Flory, P. J., Orwall, R. A., Vrij, A., J. Am. Chem. Soc. 86, 3507 (1964).
(10) Goodwin, R. D., J. Res. Natl. Bur. Stds. 71A (3), 203 (1 967).
(11) Guggenheim, E. A., “Thermodynamics,” p. 138, Interscience, New York, 1950.
(12) Gyorog, D. A., Obert, E. F., A.I.Ch.E. J. 10, 621, 625 (1964).
(13) Hirschfelder, J. O., Bird, R. B., Spotz, E. L., Trans. A.S.M.E. 71, 921 (1949).
(14) Hirschfelder, J. O., Buehler, R. J., McGee, H. A., Sutton, J. R., Ind. Eng.
Chem. 50, 375 (1958).
(15) Jofie, J., Patel, G. R., Ind. Eng. Chem. Fundamentals 4, 374 (1965).
(16) Keenan, J. H., Keyes, F. G., “Thermodynamic Properties of Steam,” Wiley,
New York, 1936.
(17) Keyes, F. G., Proc. Nat. Acad. Sci. 3, 323 (1917).
(18) Keyes, F. G., Smith, L. B., Gerry, H. T., Proc. Am. Acad. Arts, Sci. 70, 319
(1936).
(19) Lennard-Jones, J. E., Devonshire, A. F., Proc. Roy. Soc. London A163, 53 (1937).
(20) Martin, J. J., paper presented at A.S.E.E. Chem. Engr. Div. Summer School,
East Lansing, Mich., June 1967.
(21) Martin, J. J., Chem. Engr. Progr. Symp.Ser. 59 (44), 120 (1963).
(22) Martin, J. J., Edwards, J. B., A.I.Ch.E. J. 11, 331 (1965).
(23) Martin, J. J., Hou, Y. C., Ibid., 1, 142 (1955).
(24) Martin, J. J., Kapoor, R. M., de Nevers, N. H., Ibid., 5, 159 (1959).
(25) Martin, J. J., Kapoor, R. M., Shinn, R. D., Proc. II, Congr. Eur. Fed. Chem.
Eng., Frankfurt, Germany, Dechema-Manographien 32, 48 (1958).
(26) McCarty, R. D., Stewart, R. B., “Advances in Thermophysical Properties at
Extreme Temperatures and Pressures,” A.S.M.E., New York, 1965.
(27) Michels, A., private communication, 1957.
(28) Michels, C., in cooperation with A. Michels, “Some Physical Properties of
Compressed Carbon Dioxide,” Univ. of Amsterdam, October 1937.
(29) Onnes, H. K., Commun. Phys. Lab., Leiden, Holland 71 (1901).
(30) Pings* C. J., Jr., Sage, B. H., Ind. Eng. Chem. 49, 1315 (1957).
(31) Redlich, O., Ackerman, F. J., Gunn, R. D., Jacobson, M., Lau, S., Ind.
Eng. Chem. Fundamentals 4, 369 (1965).
(32) Redlich, O., Dunlap, A. K., Chem. Eng. Progr. Symp. Ser. 50, 95 (1963),
(33) Redlich, O., Kwong, J. N. S., Chem. Rev. 44, 233 (1949).
(34) Shah, K. K., Thodos, G., Ind. Eng. Chem. Fundamentals 4, 369 (1965).
(35) Strobridge, T. R., Natl. Bur. Stds. Tech. Note 129 (1962).
(36) “Tables of Thermodynamic Properties of Ammonia,” Natl. Bur. Stds. Circ.
142 (1923).
(37) Van der Waals, J. D., doctoral dissertation, Leiden, Holland (1873).
(38) Wohl, A., Phys. Chem. 87, 1 (1914).
(39) Yang, C., Yendall, E. F., paper presented at 62nd Natl. A.I.Ch.E. Meeting,
Salt Lake City, May 1967.