ric
AAI, Inc. ~ i l m i n ~ ~~ o r t h ~ r o l i n ~ on, C
Larry L. Augsburger David E. Nichols University of ~ a r y l a n d ~ u r d u e University Baltimore, ~ a r y l a n d West Lafayette, Indiana Douwe D. Breimer Gorlaeus Laboratories Leiden, The l ether lands Trevor M. Jones The Association of the ~ritish Pharmaceutic~l Industry London, United Kingdom Hans E. Junginger Leiden/Amsterdam Center for Drug Research Leiden, The Netherlands Vincent H. L. Lee ~niversity of Southern California Los Angeles, California S t e ~ ~ e nS c h u l m ~ n G. University of Florida Gainesvilie, Florida Jerome P. Skelly Alexandria, Virginia
Felix Theeuwes Aka Cor~oration Palo Alto, C~lifornia
University of Sheffield Royal Hallamshire H o s ~ i t ~ i Sheffield, United K i n ~ d o m
Peter G. ~ e ~ ~ i n lnstitut de Recherche Jouvein~l Fresnes, France
tical Statistics: Practical a
evised and Expanded, edjted by J o s e ~ h Robjnson and ~ n c e nH. Lee R. t
rd A. Guarin~ 31 . Transdermal Controlled Systemic ~edications,edjted by Yje W. C ~ j e n 32. elivery Devices: Fundamentals and Ap~lications, edjted by ravee en
33. Pharmacokinetics; Regulatory
.
lndustria~ erspect~ves, ed~ted by ~ e t e G. Welljn~ Francis L. S. Tse r and ~ l ~ n i cDrug Trials and Tribulations, edited by ~ l l e n Cat0 ai E. ~ransdermalDrugDelivery: Deveiop~entalIssuesandResearchInitiatives, H. edjted by Jonathan a d ~ r a f t Ri~hard Guy ~ and Coatingsfor P ~ a r ~ a c e u t i c a l DosageForms, edjtedby
~ j l b S. ~ a n k eand Chrjsto~her7: ~hodes e r nufacturing and Pro~uctionTechnoloions, edjted by ~ a v j d Osborne and ~ n t o n W. H. A ~ a n n Stability: Principles and Practices, Jens 7. Carstensen istics: Practical and Clinical Applications, Second Edition, ed, Sanford ~ o l t o n radablePolymers as Drug De~ivery Systems, edjtedby ~ a asin in r ~
46. ~ r e c l i n i c a ~ Disposition: A LaboratoryHandbook, ~rancjsf. S. 7seand Drug J a ~ e J. Jaffe s LC in the Pharmaceutical Industry, e ~ j t e d ~ o d ~W. n by j ~~n~ and Stanley
r m a c e ~ t i ~ a l ~ i o e q u i v a l e n c e , by ~ e t e E. Well;n~, ~ranc;s edjted r f. and S ~ ~ j k a V.t ~ j n ~ h e n Pharmaceutica~ Dissolution Testing, ~ ~ e V. ~ a n a k a ~ s h Novel Drug ~elivery Systems:SecondEdition,Revisedand Expan~ed, Yie W. Chjen the Clinical Drug ~ e v e l o p m e n t Process, avid M. Cocchetto and ~ardj nufacturing Practices for ~harmaceuticals: A Plan dition, edited by Sjdney H. W;llj~ and J a ~ e s Prodrugs: Topical and Ocular Drug Delivery, edjted by ~ e n n P h ~ r m a c e u t i c a l i n h a ~ ~Aerosol Technolo~y, tion ed;ted by A~thony ~ ; c ~ e y J.
49. 50,
53. 54.
55, Radiopharmaceuticals: ~ h ~ m i s t rand y Pharmacology,
~ ~ n n
nd ~hrjster y s t r ~ m ~
Delivery, e ~ j t by~ j c h a eJ. ~athbone e ~ i
in Pharmaceutical Development, e~ited
fopment Increasing Process: Efficiency and CostPeter G, w1~lijng,Louis Las a, an^ U ~ e s hV. Effectiveness, ~ d i t e dby n~kar c r o ~ a ~ i c u l a tSystems for the Delivery of Proteins and Vaccines, e ~ d j ~ e y Srnadar ohe en and ~ o ~ a %ernstejn r d 78. Good ~ a n u f a c t u r i n gPr ces for Pharmaceuticals: A Plan Control,FourthEdition,visedand Expanded, ~ j ~ n e y N, R. ~ t o k e r 79. ~ q u e o ~Polymeric ~ o a t i n g s for s Pharmaceutical Dos Edition, Revised and Expanded, e ~ i t e d James w1 ~ c ~ j ~ j t y by Statistics:Practicaland ~ l i n ~ c Applicatjons,ThirdEdit~on, al
81
.
andb book ofPhar~aceutica~Granu~ationTechnoio~y, ~ ~ j t e d by
~ n o l o ~ ~ntibiotics: Second ofy Edition, Revised and by ~ j i i R. ~ t r~ h i j ~ o Ex~ande~, ~ j ~ e e
2.
rocess E n g ~ n e e r i n ~ ~ t ~ o nJ. n y
~epyrogenation, Second
This Page Intentionally Left Blank
MARCEL
DEKKER
fax: 41512618896
http://www. For more i n f o ~ a t i o nwrite to Special Sales/Professional Marketing at the headquarters address above. Switzerland tel: 41512618482. New York. Postfach 8 12.This book is printed on acidfree paper. recording.dekker.com The publisher offers discounts on this book when ordered in bulk quantities. . Inc. fax: 2126854540
Marcel Dekker AC Hutgasse 4. or by and any infor~ationstorage and retrieval system. CH4001 Basel. Current printing (last digit): 10987654321
CA
. including photocopying. N Y 10016 tel: 2126969000. electronic or mechanical. without permission in writing from the publisher. ~~rters L Dekker.
Neither this book nor any part may be reproduced or transmitted in any form or by any means. adison Avenue. micro~lming.
support. and love
.o my wife
with gratitude for her ~~nderstandi~g.
This Page Intentionally Left Blank
.
In so doing. The solids area of the pharmaceutical sciences has been explored more often in the last decade than in prior times. It focuses on the ~ r i ~ c i ~ Z e s science of pharof the maceutical sciences. type of solution to a problem is really patentable. and much misinterpretation occurs. Parts of this book address this aspect. not pr viously published. concerning the proper basic consideration in the approach to certain areas of pharmaceutical solid science. is due to the advent of sophisticated instrumentation and computer access. New (NDAs). and much of what is written in today’s literature is disregardful. in fact at times ignorant. for example. Also. However. as well as the more sophisticated segment ofpharmaceutical manufacturing personnel. in particular.This book is an outgrowth of my notes for a graduate course given at the ~niversity of Wisconsin for several decades. with particular emphasis on elucidating models for systems so that they can be of general use. The book should appeal to pharmaceutical scientists in industry. The book is written for those who are interested in the actual pro~esses the on microscopic level. The book also presents some entirely new aspects. not necessarily on details or particular examples. the ref~rences often not new. except when they are s~pportive material for the text. of the principles on which the instruments and programs are based. It should appeal to scientists in government agenci problem areas that might have bearing on. This. but rather give credit to the are scientists of yore who really were the innovators.
. It should have appeal to attorneys in patent law as well as patent examiners. it should be appealing to graduate students and to advanced undergraduate students who desire a place in the pharmaceutical solid sciences area. such advantages can lead to a certain mental laziness. because it elucidates whether a given.
This Page Intentionally Left Blank
.
Solid State Sta~ility olidState Stability lumes and Densities
223 267
28 1
. Amorphates
89
107
117
erms with Crystalline Solids e~o~si~era~ions iagrarns and Eutectics Particles and Surfaces
133 159 169 191 209
14. ~rystallization
7.Preface
ne ~omponent Systems operties of Solids 3. Solubility
v
1 13 27 51
61 6.
~ o a t i n g Tablets 28. ~ o m m i n ~ t i o n
323
335
21. Sustained Release by ~icroenca~sulation
Index
. Tablet Physics rinciples of Tablets sintegration and Dissolution
375
387 407 427 439
of 27.viii
17. Single Unit S ~ s t a i n e ~ Release Dosage
455 469 493
51 1
29. Cohesion
299 309
19.
et ~ r a ~ ~ ~ a t i o n
353
ard Shell Capsules 23.
nowadays. Karl c Fischer titration. the dissolution rate of the drug substance must first be known. this property (and many other) properties of the drug substance must be explored. Hence. Tools exist. Methods 1. therefore. singlecrystal Xraydiffraction. a function of the dissolution rate of the drug substance. solid dosage forms. in this case.
. for Ionic Compounds Symbols eferences
6
6
7
8
11
I1
The purpose of pharmaceutical research is to explore the causes of properties of dosage forms. Xray powder diffraction analysis. infrared (IR) spectroscopy. Such characterization of solidstate forms encompass microscopy. t h e r ~ o g r a v i ~ e t r ianalysis (TGA). so that the sources of derivative properties in the dosage form canbe adequately assigned. differential scanning calorimetry (DSC). but before an answer is attempted. Is the dissolution rate of a drug in a dosage form.7.5. for instance. Characteri~ation the of dosage form. Introduction to Polymorphism attice Energy. The properties of the dosage fo and a host of its qualities are a function of the neat drug.2
namic Functions 2
3
4 1. requires characterization of the drug substance and what its properties are. or is it influenced more by the excipients? Suchquestions cannot be answered a priori. that allow sharp definition of a solid. It is granted that such sourcing is never complete.
many also stem from the drug substance itself. a Z i q ~ needs simply an open threedimensional container.in this text. V is volume. ir’) be rubbery. n solid pharmaceuticalmarketed products. if its viscosity is higher than what it is at the glass transition temperature. and a i~ s ~ Z simply needs a two~imensiona~ i~ planar support.1)
. Their definitions are intuitive. 1997) and this willbe employedhere as the cutoff point for a solid.and (at times) solution calorimetry (Ledwige. both the drug substance and exciare present. a gas needs a threedimensional. A viscosity at 2’of 10l2 “ often used (Lu and Zografi. whereiris temperature.
There are three states of matter: (a) gases. of importance to discuss the ropert ties and testing approaches of the neat drug (the “drug. As shall be seenin later chapters. F is Helmholz free energy. a substance will mimic many of the qualities of a crystalline solid. it will. S is entropy. ction made in the present text now seems to be the accepted h research dealing with pharmaceutical products is directed ke them possible and also addresses the failures that might or do occur. hence it may be considered ve a viscosity. therefore. G is Cibbs’ energy. eference will be made to these methods in appropriate places in this book. however. and (c) solids. be conventional to even if it is amorphous. and N is enthalpy. It is. 1996). isnot specific in the terms of solids. however. In the rubbery state they are to be likened to (or actually are) supercooled melts or li~uids and.” at times. closed container to contain it. solids are either crystalline or amor~hous. The definition. amorphoussolids may and (above their glass temperature.” the “drug substance”) to assess the properties and difficulties asso~iatedwith the final product (the “dosage form” or the “drug product”). and in differe~tial form they are related as follows. they are glassy. P is pressure: (1. though many such failures stem from manufacturing and excipients. In the glassy state.
In this book. are liquids. as such. also refers to the drug product. but if defined in words. (b) liquids. (The word “drug. the followi~g terminology will be used for the four t h e ~ o d y n a ~ i c functions: E is free energy. and below this temperature.
at constant T? it is possible to indepen~entlydetermine AG and AH.L is chemical potential and n is number of ispecies transferred. (1. between phase 2 and 3. 10)then gives d{AG/T}/d~ ~dln[K]/dT = AH/T2 = (1. (1. temperature are variables and there are (n . in chemistry.8). for a chemical reaction with equilibrium constant (1. (1. An outcome of this is that (1*7) Often.
(1*6)
lpy change at constant pressure.13) constitute p .11) If ln[K] is known at several temperatures. (1.1) independent concentrations per
. systems are constant~ressuresystems. the chemical equilibrium criterion is that A F . it is in particular Eq.1).1) equations for each of the n compounds. and it is then possible to calculate A S from Another frequently employed relation is For instance. of a system isa measure of its oltzmann's law states that: (1. There are . It should finally bementioned that the entropy S.10) Inserting Eq. and so on the following holds.9) into Eq.3) that is affected.The chemical energy terms are not included in the foregoing. ecause there is equilibrium between phase 1 and 2.
dG ==: SdT
+ VdP +
(13
where I. It is particularly noted that dG = 0 during equilibrium. so the total number of equations is n(p . in a chemical situation. Note that Eq. and that. case. and A S may now be found from Eq. G is a convenient function in that d T and d P are zero at constant temperature and pressure. under these conditions. dG < 0.
It is of interest to estimate the number of phases that can be present under one particular energetic condition. not AG. s. but with these.12) mber of ways in which a system can be made up.1 equations. and that. for a voluntary process. be zero. (1. h o ~ e v ~situations arise that call for constantvolume considerations a r. Suppose an ensemble in equilibrium consists of rz c o ~ p ~ n e n tand p phases. ost often. A H for the reaction may be found.
1. because a given T dictates a certain P and vice versa. water (liquid). These are referred to in Table 1.14) is often difficult and it is st~essed that it applies only to an equili~riu~ sit~ation. and these may arrange themselves in different fashions (Fig. The angles. each of these occupies 120/(6 x
*
b
Example of crystal forms. As an example. The distance between these layers is denoted d below. so that by Eq. The term degree o f free do^ in this contextis exactly the oppositef its statistical~ e a n i n ~ s o (where itis the n ~ ~ o ~po~nts i ~ the s e r~ f u number o equations~. there are three phases.”lle of such a cornmolecular weight 180 and a true density 1. There are seven different crystal systems.21 as the difference between the number of equations and the number of unknowms. The use of Eq..e. and neither T nor P may be chamf ged without losing a phase. so that the number of variables is p(n .phase. is and the manner in which d is d e t e ~ i n e d as follows: To get an idea. first of all. Points in the (ideal) lattice are occupied by molecules or ions. b.14)
This means that there are d variables that may be changed without the system f “losing” a phase. there are two phases (liquid and gas). and w are shown in the left figures.there are different “layers” in different directions. (1. a beaker of water has one component. of
d =p(n1)+2n(pl)=np+2 f
+
(1. ice (solid). a. cause there are 6 x = 200 x cm3.as are the possible distances. and vapor.14) for the particular system and obtain [df . hence.5 = 120 cm3jmol.14) there is 1 degree of freedom (i. for order of magmole.1. Depending on which direction the lattice is viewed. it is prudent to actually do the derivation leading to Eq.1). or ions in the lattice may be visualized as a series of layers. u. and e. Increasing the t~mperature loses the solid phase (the ice melts) and lowerin it loses the liquid phase (the water freezes). f LL
A lattice is a periodic array. If the temperature is I increased a bit. The number of degrees of freedom is the n u m ~ e r variables minus the number of equations. ofthe magnitude of d consider a solid compound of The molar voll.1 ) 2. no phase will be lost. The situation is different at the freezing point. If. Such a point is called e a t r ~ l point.
.e. n = 1 and p = 2. Hence d = 0. molecules. However. (1. When in doubt.5 molecules in a pound would be 18011.. one variable [either T or P may be changed). v. (1. i.1 ositioning of atoms. as shown in Table 1. it is not possible to change both T and P at will. Here.
1 and 2. they will then reinforce one another.85 A
(1. by simple trigonometry. and ionic distances within a lattice. this distance must be a multiple of the wavelength h of the ray. to be equal to 2d sin[q Le. that when two Xrays are inphase.15)
(1. which relies on the fact. strike a surface at an angle of II . shown. hence.16)
where 1 is defined as 10"' cm3. 1. for them to be inphase. is usually referred to as 8.17)
A
Schematic for Bragg's law.Xrays are of this order of magnitude and are used for measurement of atomic. this distance must be equal to nh.85 x
cm3
cm3 = 5.. Ray 2 traverses a distance ABC (in bold in the figure) longer than ray 1. U. equaling thereflected angle. Two Xrays. This is performed according to rag's law. U. and the principle on which it is carried out is shown in Fig. 2d sin[U ] = nh (1. for attenuation to be monitored at the collector Q .Length of side Angle between Examples name distances axes a=b=c a=b$c aZb#c a # W c
u#v#w#9O0 u=t1=~#90" u = w = 90" w = 120"
Alternate System Cubic Regular Tetragonal Pyramidal 0rthorhombic AgN03 Rhombic Monoclinic Triclinic Trigonal Hexagonal Graphite ~hombohedral NaN03 NaCI Rutil ~~minobenzoic acid K2Cr07
a#b+ a=b=c a =b =c # d
nitude calculations~the arrangement is assumed to be cubic.
.2. The distance A. The incident angle. that is. molecular. the side length of the cube encasing the molecule would be given by
d3 = 200 x
or
d = 5.
54/(2 sin 3. Raman spectroscopy.5418tf)of diclofenac ~(2hydroxyethyl)pyrrolidin.19)
Schematic for derivation of the radius ratio rule. and showed that it exists in zwitterionic form in the solid state.. Hence. The carboxylic proton is present by the piperaz terminal nitrogen. Consider.18)
c i u ~ .e. employed they . The d i s ~ ~ (i. in which.Ledwidge et al. for instance. (1996).2. and molecules in the lattice.6
A
(1. reports an Xray diffraction pattern. The most common method is powder Xray d ~ ~ ~ ~ In t this method. Turel et al. A hypotenuse.e the dvalue) ~c corresponding to this would be
d = 1. and this is subjected to the type of detection shown in Fig.8") = 11. For instance. a c o ~ p o u n d consists of two ions. with radius r. (1997) usedXray crystallography to determine the crystal structure of ciprofloxacin hexahydrate.e and the smallest 20value where a peak occurs is 7.3. powder is packed into a cell. however. lanes present themselves in sufficient abundance to allow determination of the crystal lattice constants without determination of the position and direction of atoms. for instance. and thermal methods to determine that the water in the hexahydrate was present in complicated a network governed by hydrogen bonding.6". or ions in the lattice ~inglecrystal Xray crystallography allows determination of the position and direction of the ions. It is obvious from the figure at the right t~iangle. one smaller. and one larger. molecules.20)
+ (2R)2= (211 + 2r)2
(1. (using X rays with h = 1. with radius R. (2Ry or (1.
Inorganic ionic compounds consist of fairly spherical entities. and their packing is related to the relative radii of the two components of the systems. CA = 2fz + 2r.
. the situation in Fig. 1. 1. atoms. Adjunctly.
form 11) for the following reasons. because their orientation is different in thetwo different polymorphs (of the same crystal system). It is not possible. two polymorphs. and the concentration will level off at the the~odynamic e~uilibrium solubility.155 0. This concentration of drug in the solvent is reproduci~le.1550. 1. but the solution is not thermodynamically stable. The lattice constants.225 0.7331 1
Hexagonal Tetrahedral blende Octahedral Bodycentered cubic csc12 Facecentered cubic and also hexagonal
.4144.733 0. at most. at a given temperature. will have a higher vapor pressure. a. From a fictional point of view. The less stable form.2 emerge. 2. it might be tempting to think that there could. and so on.141R
(I 21)
Similar relations can be obtained for other arrangements (crystal systems). 4. There areno rules for the nomenclature I. this is the subject of the book ~ ~
t
Examples of the Radius Ratio Rule Applications Ratio
=
r/R
coordinatio~ number 2 3 4 6 8 12
Lattice
Example Carbon dioxide Boron nitride Zinc NaCX
00. in a practical sense. numbers simply signify the chronological order in which they were produced. 3. 6. the number is not limited to that. olymorphs will be su et to a special chapter (see Chapter 8) but at this point the following will be noted. The less stable form. Eventually precipitation of a more (the more) stable form will occur.414 0. at a given temperature. is different ~ Q Z forms ( ~ o r ~ ~ s )this p~enomenon denoted ~ o Z ~ ~ o r ~ ~ ecause there are seven crystal systems. however. and e. The molecules may be in different lattices.2250.
Whereas inorganic compounds often (if not most often) crystallize in one particular crystal system.V2) = 0. one (form I) will be (con~gurationally) more stable than the other (e. then might or would be different. to talk about the “most stable” polymorph.g.which has the positive root:
r = R(1 .. will have a higher ap~arent solubility. and the rules in Table 1. organic compounds have the capability of c~stallizing several in ~ ) and . beseven different po~ymorphsof one compound. 11. for a more stable polymorph may be discovered at a later time.
extensions to organic crystals are possible. in the situation shown in Fig. ere a more stable. ach interparticular distance (rY)is expressedas a number (pii) multiplied with ration (R) between two particles. Examplewise.4).
. as a function of atomic or molecular distance. are rationalized aron and Prutton. The molecules in solutions created by either a less or more stable polymorph are the same.22)
The value o f n is. ordinarily.r l
~ r ~ ~ l Kurt ~ o n n e g u t . water (Ice Nine) eventually causes the world’s oceans to freeze over. one ~ ~ r t i c ~ e . 1. The repulsive term depends on distance by some (the nth) power of the separation. thejth and the ith molecule wouldseparated by 2 “units” so that rij would equal In general this may be written as
rij = pijR
(123)
) isnowsummedover all interaction possibilities. The attractive force is. highermelting form of by e . 5. inversely proportionalto the seventh power of the interatomic or intermolecular distance. ( . from 10 to 13.” hen bonding occurs between two molecules.
The section to follow has been developed quite rigidly for inorganic ions. The potential energy u‘ of the interaction between two neighboring ions. may be written as
d =( ~ 1 (By) ) ~ ~
(I . a minimum will occur in the po~ential energy curve that exists betweenthem. the term ~ ~ r t j c Z e will often be used to signify “ion” or in some cases “molecule. 1.
”I
Potential energy curve. which then gives the pfor energy.5. 1965) by the existence of two opposing forces between the ms or molecules: an attractive force and a repulsive force. In the development. therefore. nergy curves. theoretically. This distance is known as the lattice constant Ro (Fig.
so that Eq. is obtained by obtaining the first ~erivative and equating it to zero (and at the same time ensuring that thesecond derivative will ~ 4 / = (7Au/R8) ~ R
+( ~ ~ b / R n " )
(12 8 )
For equilibriu~ to occur this must be zero.( 7 1 ~ ) ) [: =
(1.30)
40 [Aa/u:l . J I { l .[ ( 7 / ~ ) ( A a ) / RAla / R . ~ i t h of the Avogadro number N. (1.<"_"""" "" """">
r10. ( 1 2 7 ) to give
(1. Ro.32)
The energy per mole U can now be obtained by multi~lication 4 0 . = n or (7/n)(Aa)/R: = ~ b / R ~ This is now inserted into Eq.12
Schematic of interacting atoms or molecules. The following termsare introduced for thesake of conve~ience:
(1.( ~ b / R n )
(1.26) For a given crystal a and b are constants.24) becomes:
4 = ( A u / R 7 ).
(1.so that
.29)
or
7AaIR. so (7AaIR~) ( ~ ~ b / R ~ " )
+
(1.25)
and (1.31)
~
~
/
R
~
~
~
(1.24) where summationis over all i # j .27)
The distance at equilibrium.
40) (1. (1.27).39) to U/d d2
V 2 = (d2U/dR2)(dR/d V)2 (1
rom Eq.44) (1. For molecules (many organic molecules) the force terms are all positive. (1. leaving but two unknowns.
.37) (1. each second beingopposite in sign and giving rise to attraction).34) (1 (1. (1.
so that 1/k = Vd2U/dV2 olar volume with the nomenclature used here is
v = NR3
so that
dU/dV = ( d U / d ~ ) ( d R / d ~ ) he second derivative. (1.36) we have (dR/dV)2 = (3NR2)2== 9N4R4 q.33).36) to give ~)(9N4R~)(d2 U/dR2) = 9N3Ri(d2 / d R 2 ) U Equation (1. is most often acceptable to use only one or two terms m a ~ i n g it the s~mmation fairly simple.40) becomes
d2U/d V2 = (d2U/dR2)(9N4R4)
This is combined with Eq. The value of n may be obtained by lo~temperature compressibility measurements. (1.u =N
~ A ~ / . The definition for compressibility k. so that the first term vanishes.(7/n)}~ l R ~ ~
(1. the 7“ term in the definition for U vanishes so that
dU =: PdV.45)
which inserted in
I l k = 504N4(~a)[1 (7/n)](R~2) which allows calculation o f n. The actual value of a is obtained geometrically.33)
onv vent ion ally? U is equated with the enthalpy of sublimation. is
k
(l/V)dV/dP ‘35)
(1. hence. the exponent in the second term in Eq.41) (1.42)
t equilibrium^ dU/dR = 0. is
d2 U / d V 2 = (dU/d~)(d2R/dV2) (d2U/dR2)(dR/d~)2
+
(1. For ions. the terms alternate in sign (each secondion being negative.43) when differentiated twice gives
d2U / d R 2 = 56(NAa)[1 .thus exerting attraction. Ro values may be obtained by Xray analysis.36) (1. (1. because solids are considered constantvolume (rather than constantpressure) situations.The large negative value ofn.43)
(1.(7/n)](R~)
(1. A and y1 in Eq. reducing Eq.38)
t low temperature.39) .
Int J Pharm 15259. (1997). pp 70"79. 'Villa M. Wilcock DJ. Prutton CF (1965). (b) ionic radius of larger ion r = ionic radius of a smaller ion Ro = ~quilibrium distance between particles. molecules. ions. New York. P = pressure pij = factor forexpressing the distance between the ith andjth ion in units of = distance between the ith and jth ion = (a) distance between particles. Lu Q. Kittel (1962). = chemical potential u 8 = incident angle of an Xray h = wavelength
N n
=
=
enettlnetti 6. Ventura P (1990). Maron SM. molecules. Solid Phar~aceutics.A
constant in the energy versus distance equation adelung constant adelung constant B = constant in the energy versus distance equation d = distance between molecular layers E = energy
=
H
=
enthalpy ann's constant. ""Principles o f PhysicalChemistry. J Pharm Sci 86: 1374. London.4th ed. Italia A. ions. Draper SM. (b) integer in (c) exponent in potential energy versus distance equation. pp 728729. ions. Fronza 6 . Corrlgan 0 1 (1996). J Pharm Sci. Carstensen JT (198 1). Zografi G (1997). Academic Press.MechanicalProcesses andRate "Phenomena. (b) compressibility
Avogadro's number (a) number of particles. molecules S = entropy 1 = absolute temperature ( " U = crystal energy u' = potential energy between two ions Y = volume W = number of ways of building up a system . Introduction to Solid State Physics. 85:16. J Pharm Sci 79:470.
. pp 67. T. Pellegata R. Giordano F.
This Page Intentionally Left Blank
.
References
25
The text. 2. will deal first with properties of solids that are not.2. defects are to be expected. Several types of defects are depicted in Figs.
2. that a crystal contains nine molecules. 2. 2. From a statisticalmechanical point of view.3). a function of their subdivision.4. as mentioned earlier. If one of the internal molecules is moved to the surface. As they grow (a point that will be discussed later) planes may grow over one another on the surface (Fig.
Crystals are never perfect.5. 2. There is but one wayof arranging them. The Einstein Equation
13 16 18 19 22
2.g.1). Classic Heat Capacity 2.2 and 2. 2. primarily. Suppose (Fig.3. particle size). In essence they may be considered the properties of an infinitely large slab of the solid. They may also grow as a screw (a socalled screwdislocation).3. 2. and in this case there is a sort of pore that penetrates the crystal as the axis of the screw.1.6. Later chapters will deal with properties that are a function of the subdivision of the solid (e. The development of this concept in the following is based on the Schottky defect.. but it would also apply to vacancies of other natures. there will be { lo1}= 10 ways of doing
.7. shunting out areas of voids.2.
Schottky.
. and screw defects.
0 000 0 0 000
9 Posit’ions One Way
10 Positions 10 Ways
11
Posltlons
55 ways
Situation where one and two Schottky defects are created in a crystal with (originally) nine lattice sites.A
E 3
c
D
A nucleus or crystal (A) grows on the surface. Further growth and a site in a second layer are shown in (C) and in (D) the growth in “higher” sites grow over the lower sites creating a “hole.”
Schottky
Frenkel
Vacancy
Frenkel. and two sites are shown.
Itf two molecules were moved the surface. This isnow relative to n to give the equilibrium condition:
dF/dn = EL* kT ln([N .5
+
35.N Le.
. This rearranges to: is E:s= kT ln(n/[N . (2.n]/n)= 0
differentiated
(2.5) differentiated relative to n.8)
where the argum~nt Eq.n)ln(N .nln(n) = (2..
IC = nEs .
this.( N . then the number of ways would be to {112}= 11 x 10/2 = 55 ways. For a system of N the entropy would be
S = k ln[(N
+ n positions with n vacancies?
(.n]) or
n == NemEsk 1 / "
Normal range of vacancies is ofthe order of 0.TS
where Es is the energy associated with one vacancy.) 22
+ n)!/{N!n!}
Use is now made of Sterling's formula lnN! = N l n N .nln(n)}
+ n)!/{N!n!} N l n N .( N .5
Calcium replacing IC. so that in equilibrium considerations?it is the lmholtz free energy (rather than the Gibbs energy) that is applied.3
(2.MW 2 x 74. as a means of creating a vacancy.5
Cl
Cl K K 6 K C1 K C1 1 C K C1 K C1 K l K C1 K C1 K C1
K
C1
K
C1
C1 K C1 C1 Ca K C1 K C1 K C1 C1 K C1 K Cl K
K
C1
K
K
6 1
MW 75.1)
oltzmann constant.n) . olt~mann states that the entropy of a system S is proportional to the law logarithm of the number of' ways in which it can be made up:
S = kln(ways the system can be made up)
(2. as applied to this system ln[(N
so that
S = k{NlnN .5)
olid systems are usual1 onsidered constant volume systems.n ) l ~ ( N n) .001%.
(Data from Pick and Weber. do
There are several different definitions of ~ e n ~ i tThe ideal density ofa crystal can be y. the upper line the experimental line. Usual production sp~cifications not include criteria for mechanical properties. Hiestand (1997) states that “the ever present.e. and nd Grant (1987) have called attention to the fact that processing of solids causes lattice defects. ” yield value ofthe particles is The dependent on defects in the crystals.11) (2.4 and 2. If there are n calcium ions in a crystal with N positive ion sites. The lower line is the theoretical line.
n/N x 10000
Change in density of KC1 doped with calcium ions. so that the difference in density would be (2.givingrise to an increase in disorder. becausetheir a sizes are appro~imatelyequal.5
(2. H~ttenrauch and einer(1979 a.10) (2. where V is the molecular volume. should be (~~ of C1 being 35. and changes may occur in nearly all processing.5
The difference between these two numbers is
A W = 38n
The volume of the crystal is N x V .5) without vacancies.12)
+ n75.. with KCl. introducing a foreign molecule into ce of the compound in question). then each time a calcium ion (MW 40) is introduced.5
The weight W+ with vacancies would be
W+ = 2N74.5 + n35. a hole with one missing K 39) is created.)
. ancock and Zografi (1997) claim that this would give the particle a certain viscoelasticity.therefore. 1950.efects are often creating by doping (i. then the density is calculated as follows (Figs.5)
~~~
N“ = 2(N + n)74. For instance. is 38 per calcium ion. Grant and York (1986). plastic deformation profiles an explanation why lottolot problems are c o ~ m o n . calculated from knowledge of its lattice parameters and the molecular weight.b). (1984). ersey and Krycer (1981). the potassium ion ) may be replaced with relative ease bycalcium ion (MW 40). The loss in weight. 2. Longuemard et al.13)A ~
/ =N 38n/NV
~ttenrauch (1983). Moriata etal. If one considers a crystal with N ions of KCl.
which works on the same principle.6). Theformer is marginal at best if the solvent is selectedwith care. so that the crystallographic density wouldbe 3/28 = 3/1. This could be derived only by knowing the lattice para~eters. for the compound is C8N I 1016S. crystallographic densities are reported routinely in studies of the crystallographic details of a particular form of the compound. The pycnometer has a given volume V em3. for instance. Therefore. except the “liquid” is helium. and 8 and its molecular weight is 240. and is given by weight ( M 2 )of the ensemble is obtained ex~erimentally
The only unknown is p2. 2. The pycnometer is filledto a mark giving the density p l . At high dilution. ~ozwiakowski ai. the actual density would be less.( M / p 2 ) } em3.92 = 1. 7 .8 x em3. the lattice parameters of a orthorhombic crystal are 5. and (b) air entrapment.7). The volume it occupies is 5x 7x 8x = 2. In wet pycnometry. None of theseproblems exist inthe use of the helium pycnometer. and the weight of the contents W is determined. The net (Fig.g.If. water for a poorly watersoluble compound). so that the liquid now occupies{ V . These grams occupy M/p2 em3. The actual particle density is determined by either wet pycnometry or by helium pycnometry (Fig. which is the quantity sought. As an example. a liquid in which the solid is insoluble. Yielding A n Exact Volume
Liquid pycnometer.
Now M grams of solidare added. having the (unknown) density o f p2. then the mass of one molecule is 180/(6 x =3 x g. Ceolin (1997) has reported the volume of the triclinic unit cell of carba~azepine to be 2389 A3.
. isadvantages are (a) that the solid may be somewhat soluble in the pycnometer liquid. is selected (e.. ideal solutions are approached. (1996) reported on the solubility behavior of lamivudine et and in this process report on the lattice constants of the compound. ecause of lattice defects and vacancies. The following program in BA molecular weight em~loying precise atomic weights the
Ground Glass Stopper With Capillary Inserting Glass Stopper Allows Excess Liquld t o Escape. 2. of the solvent:
A.07 g/cm3. so that the volume contraction or expansion considerations are negligible. it is not to be expectedthat theparticle density isthe “true” density.Nowadays.
2 3
eat capacity plays a part in several pharrnaceutical considerations on a theoretical plane.17)
. Density = ”.y
.591

l.7 Densities as afunction. U4 =
INT “Mol Vol = ”. 1972.)
for aseriesofmonoclinic
4“ubstituted
1 X3 X4
=
= =
(8
*
12.01115)
11 * 1.0131e3x RA2 0. solubility.16)
is accompanied by an enthalpy. and the heat associated with this would be the heat of solution.0. 2 2 st.21 INT “Vol/grarn = ”. If a process goes from stage A to stage A+ (2.427
U4
=
Ul*~2*Y3/20 ecVol in AA3 ”.G r
120
130
140
150
160
170
Molecular Weight
. The process could be. X6
U1
=
10. for instance.9994 2
+ x 3 + X4 + x 5
”. This is often considered a constant.70668
+
5. (Data from Musa.00797 3 * 15.ofsubstituent benzoic acids. The heat capacity Cp of the solution is given by Cp = dAH/dT
(2.
For an ensemble of N harmonic oscillators.18) (2.08 63. but the problem with it is that it predicts constancy. and these are assumed to be harmonic oscillators.243 0.93 Heat capacity cal/"rnol 6. for a mole
Eavg 3RT =
So that. the socalled Einstein te~perature).54 114.90 6.43 5.42
.~ ~ Z o ~ ~
The foregoing holds in a classic sense.22)
Table 2.31 58. and they may be interchanged freely.29)
(~.056 0.156 0.20)
or. for a solid.82 24.. the heat capacities begin to drop. and it becomes important to obtain an idea of at which te~perature(a. with three degrees of freedom (the molecule may oscillate in three directions). It will be shown later that indium is used as a calibrator for diff~rential scanning calorimetry ( equation.and assuming that it is constant then implies that AH is temperatureindepen~ent. but they translate to organic molecules as well. he considerations to be outlined in the following are mostly based on work etals.109 Molecular weight 40. with systems for which one assumes constant enthalpies in a temperature range. the average energy of a system is kT' per degree of freedom. deviations may start to occur.85 6.
Heat Capacities at 25°C Compound Ca CU In Mg CO Heat capacity cal/"g 0.1 shows examples of this. There is but little difference between Cp and Cvfor solids.25 5. The equation is called the ~ e t i t . where k is the oltzmann constant and 7' is absolute temperature. In classic theory. the average energy is:
Eavg
=3
N
~
~
(2. the heat capacity should be
Cv = 6 cal/degmol
(2. it is possible to ~ o m m i t errors.092 0. The heat capacity is assumed to be associated with the energy E of the molecules in the lattice. The data in the table fairly well substantiates the Petitbut at lower temperatures.
for a harmonic oscillator is given by E = nhu = n(h/211.054 and u is the angular frequency. is used. v is frequency. The energy. and is an integer.30) (2..24) The energy of all the molecules is given by (2. (2. there will be various energy levels.ex) ence. E2 (with n = 2). ..) the is k's constant (1.29)] is a geometric series with factor ex.34)
. (2. the quantum mechanical concept that the energy takes on values only as integers of one another.+
then

a
(2.26) ~ntroducing Eq. In an ensemble of N molecules.3 1) (2. and so on. Eavg hu/{(exp(hu/~T) 1) = This should be applicable at all temperatures.20).28) If we use the notation
Y = (1 + ex + e2dV. (2. h/211. in Eq.
total number of molecules is givenby (2.) n h u = (2..e.~ ~ / R= hu/RT = X T (2. so that the sum is Y = 1/(1 . .29) (2.) so that. Le.In this type of development. (2.22) ck's constant (6.624 ergs). The fraction Vn) the molecules in energy of state n iven by the ~ o l t ~ m a n n distribution.19) .28) Eavg nhud In Y/dx = . but at high temperatures ( e x ~ ( ~ ~/ 1 x~ 1) (ho/rcT) + .33)
dY/dx = (e" + 2e" + .27)
we may write Eq.)(~211.1 = (hu/kT) k
+
(2. y1 is ntum number. El (with y1 = l). the average energy may now be calculated as (2.32) (2.24) as
(2.25) y introducing Eq. h (i.
4 TI0
0.) Debye later refined the model to include many frequencies and obtained an even better fit.so that Eavg ho/(ho/kT) ==
==I
kT
(2.0
iarnond heat capacities compared with the Einstein.
0.8
1.1)
so that the heat capacity becomes LdE/dT = Lk(c3p. provide^ a suitable choice is made of the fundamental oscillator frequency. (There should at least be three. so it is conventional to tie this in with the Einstein tem~erature by:
. ~ n t e ~ n a t i o n a l~ r i t i cTables uses this.
. and higher polynomial ap~roximations./{eXP(@E/T) . and it is more convenient to talk about temperatures than about f~equencies.35)
that is. one for each degree of freedom. Figure 2. The Einstein model gives profiles in reasonable agreement with experimental data.39) ~l For instance.3~) (2. The severe assumption in the Einstein model is that there is only one fundamental frequency.6
0. (2.37)
~ i t this terminology.lT)2{exp(c3p.38) where L is Avogadro’s number. o Both the terns h and kT are energy terms.when tabulating heat capacities and enthalpies as a function of te~perature. at temperatures higher than a given temperature c3pE (the socalled Einstein temperature). The important lesson to draw from this is that heat ca~acities ( ~ e ~ eon ~ i are n the c o m ~ o u ~ d ) times s u f ~ c i e n t ltewlperaturedependent and thatthis ~ h o ~ be at ~ ld taken into account. model with a 4of 1320 > .ho z== k@E
(2.0
0. ith experimental data.2
0. The most common appro~imation that is hH=Q+gT (2. it is possible nowto find (by iteration) a value of c3pE that makes the data fit the best. the energy equals the classic energy.8 is an example of this./T))/{exp(c3p~/T) (2.31) becomes h
Eavg
= kc3p. Eq.
but more easily attainable vapor pressures. that is. phase rule states that
cf = C . A similar set of equations for the solid ence.43)
n e~uilibrium two phases have the same chemical potential. It will be assumed that the equilibrium is between a solid and its vapor. the
Pg
= Ps
where the subscript g denotes gas and s denotes solid.0th liquids and solids have vapor pressures. For a onecomponent system this becomes df'=
E"+
(2.
{ d P ~ / d T= ~ } "8s
. ibbs..g..2) may be written (2. the following must hold: It is recalled that dG and that and where s and v are molar entropy and volume. although some pharmaceutical substances (nitroglycerin7ibuprofen) have significant vapor pressures at room temperature. The phase rule states that in the described situation there will be two degrees of freedom (e. For less precise.46)
is the G function per mole.
SdT
+ Vdp
SO
(2. is This loss rate is proportional to the vapor pressure in the pan. temperature and pressure) that may be changed. Eq. so that at equilibrium. A substance with known vapor pressure PC ( e g . P is number of phases.48) (2. A covered pan with a pinhole is a l l o ~ e d stay at a given temperature in the TGA.41)
where G is number of components.49) or
lu. and the vapor pressure of the solid (Py) is
The development to follow holds for any condensed phase of a onecomponent system. Vapor pressure of a solid is measured by means of a socalled which measures the escaping tendency of the gas through a capillary. thermal gravito metric analysis (TGA) is employed.P + 2 i
(2.45) that (2. benzoic acid) is run in parallel. Vapor pressures ofsolids may be quite ow.42) (2. and df is the degrees of hases. and the loss rate (dWx/d~) measured. (3.
680 2.51) and (2. An example of this is the vapor pressure of benzoic acid.914 2.1065 0. 2.934 0.002 2. its vapor pressure curve will follow the Clausius Clapeyron equation.54)
This integrates to lnb] = .~ ~ / /( ~ ~ ) 3
+
(2. the heat of vaporization).(2.832 2.52)
~ntroductionof Eqs. 2.767
0. (2.1539
3.51) where A H is the heat absorbed at constant temperature and pressure when X mol of substance passes from the solid to the gaseous state (Le. except that now the slope is AHvap (i.50) gives:
(2.52) into Eq.99 x ’7. It is noted that the heat of vaporization is
AH
==
1.9 and the logarithmic transformation is plotted in Fig.610
Source: West and Selby (1967)..230 0.7147 1.
Vapor Pressure of Benzoic Acid as a Function o f Tempe~ature
P
Temp (“C) 60 70 80 90 100 110
=
vapor pressure
lOOO/T K_l
1nVl 2. The direct data are plotted in Fig.4 ~cal/mol
It may also be noted that it is assumed that the enthalpy of vaporization is not temperaturedepend~nt.)
25
VI
=RT/P
(2.685 = 15.568 0.244 1.10. and if this is considered
(VI
.
.2085 0. (2.? it is the molar heat of egarding the volumes. the molar volume in the s d stateis negligi~le that in the gas phase.50)
(~.55)
whereis an integra~ion constant.v.3928 0.53)
(2.754 2. (from the source) it is not so in the temperature interval and shown.
If a substance is at a temperature suf~ciently high for it to be in a melted ~ondition.e.2592 2.336 0.
11) and melting points will assume a special significance when further discussion on polymorphism is presented. 6 8 5 1 ~R"2 .7 "2. but a few points and examples are appropriate to mention at this point. Car~amazepine (USP) is monoclinic. that is.
y
1
. Ceolin et ai.1
Data in Table 2.
olymor~hism the phenomenon of a chemical entity being able to exist in two is different crystal forms. The authors produced a crystal of dimensions 10 x 70 x 430 p m that they used for singlecrystal characterization of the polymorph.
. but single crystals are ~ i f ~ c u l tproduce ion to in this manner. It will be discussed in greater detail elsewhere in this text.
Vapor pressure curves (Fig. (2.12). 2.0
3. 2.20.000
0
1
2. They show the following topological p .8 2.6
2.55). T diagrams of carbamazepine. ~ u b l i ~ ~ t gives a triclinic polymorph.1.2 (see Fig. 2. but other polymorphic forms exist.826 . T diagram (Fig. have reported on p .50 60 70 80
9 0 100 110 120
Temp ("C)
Vapor pressure of benzoic acid as a function of temperature. (1997).7 .9
1000/T
3.9) treated according to Eq.
The heat of fusion AHmelt is the difference between the two.
. Int J P h a m 259. they found that 132 was the triple point. Huttenrauch R (1983). J P h a m Sci 86:987. 4) should read: B is the triple point between triclinic.
. Krycer I (1981). J Pharm Sci 86:1062. Grant DJW. and by monitor in^ the deposit and the temperature alongthe tube. liquid. Hersey JA. Einstein A (1907).)
They found the transition point by using a tube heated at the position of the solid. Int J Pharm Techno1 Prod Manuf 2(2):55. Huttenrauch R. Int J P h a m 30:161. Dzyabchen~oAV. Hiestand E (1997).. (Data from Ceolin et al. and vapor. Agafonov VN. P h a m Ind 45(4):435. Bachet B (1997). Ann Physik 22: 180. Keiner I (1979b). Keiner I (1979a). D is the triple point between triclinic. Toscanini S .51
Melting Point
v40
GO
80
100 120 140 160
Temp ("C)
Vapor pressure diagram of benzoic acid ( ~ e l t i n g point 122°C). 1977. York P (1986). Hutten~auchR. monoclinic and vapor. Powder Techno1 22289.F. Gardette M.
E
Vapor Phase
!32 190 Temperature ( " C )
The literature caption (the reference Fig.
The Ch~micai ubber Co. Longuemard P. PhD dissertation. J Pharrn Sci 87:193. Nakai Y. University of Wisconsin. Int J P h a m 17051. Jbilou My GuyotHerrnann A. Nakajima SI (1984). D 141. Spankcak CW (1996). Chern Pharrn (19’72). Cleveland. 143. Kukuoka E. Moriata M.. Madiso
try and Physics. Sisco JJ.
. p..Nguyen NT. Guyot JC (1998). OH. 48th ed.Jozwiakowsk~ MJ.
14. 3. 3.6.2. 3.9. 3. 3. 3. 3.11. ~ o ~ ~ ~ e ~ a t i o ~
~
42
42 44 44
and p 3.13. and the subject has ds been subdivide^ into the foregoing subtopics.
.4.10.7. Solub~lity
45
46
46 46 47
Solubility of c o ~ ~ o u nis of great importance in pharmaceutics. 3.1. ffect of Surfactants eat of Solution etermination: Effect of Temperature trolytes on Solubility ixed Solvent Systems lectric Constant on Solubility Parameters ultiple Solubility Peaks yclodextri~s rediction Equations for Solubility in
28
28 32 37 37 38 39 41
3. 3. 3.Solubility 3. 3.5.15.12.8. E ~ u i l i b r i u ~ 3.3.
There is a fair amount of misinterpretation of the “heats of solution” in literature. Add to that the fact that solids of higher energetics(metastable poly~orphs or amorphates) have higher apparent solubilities also confounds the issue.. 3. it assumes that a (stable) solid (the solute) is laced in contact with a li~uid (the o l ~ eand~the system is all owe^ to be agitate^ for a long s ~ ~ .henever the tern solubility is employed. in the case of benzodiazepam). the drugsubstance. the
When a substance (the solute) dissolves in a solvent there are certain changes that ome solutions are ideal solutions. 1933) that may be considered: e solid phase is a pure compound. while. by fact t ~ a the co~cent~ation solute has reached a co~stant t o f level. For the purposes of this book. and it will become a ~ ~ a r e that the effect of temperature on solubility is associated with an aspect of nt this thermal phenomenon. The two components form a solid solution in such a way that there is unlimited solubility in the solid phase. case 1 is ove~helmingly most cosituation. then the DSC thermogram will showtwo peaks. there are four types of equilibri~msituations ill. turned out to be a metastable polymorph. If the melts of compoundAand compound Bare immiscible. At times (e. Apparent equilibrium solubilities esta~lishedand were t h o u ~ h t be true equilibrium solubilities because the to figures were reproducible. the term e ~ u i l i b r i u ~ ly solubility is clouded to some degree with uncertainty. makes the experimental establishment of solubili~yof acompound dif~cult to achieve. and this had a lower solubility. Case 2 is at portant in differential scanning calorimetry work. there is limitedsolubility in the solid phase). one at each compound’s melting point. it is tacitly assumed that it is equilibrium solubility.
f these. and there is one liquid phase. and there is more than one liquid phase. In other words. Because it is never reallycertain that any drug substance produced is a c ~ ~ a lthe stable polymorph. 4. or by other ~ e a n allowed to reach a state o e ~ u i l i b r i ucharacteri~ed the s f ~. Such things as small is temperature ~uctuations. Inthis chapter to follow it is going to be assumed that solubility is exactly what the foregoing italicized de~nition purports it to be. This will be discussed in a later chapter. but comprehensive
. as first produced (in clinical trials). for property thatis of importance in the following isthe heat associated with the solution of a solid drug substance in a solvent (most often water). This de~nition by no means easy to establish in practice. otherwiseone broad peak will occur.. There aretwo solid solutions forming (Le.g. until one day the more stable form happened to be produced.and that solubility maybe a function of particle size. and in such solutions the volumes. and in this aspectitis fruitful to quote a very old. e solid phase is a pure compound.
and Ai is shown in Fig. in its simplest form. owever.in the followingwords. If heat is evolved when the solid is dissolved in an ( a l ~ o s st ~ t u r a t e d s o l ~ tthen. The equation corresponding to this may be derived in direct analogy with the [previously cited equations] containing nl + n2 moles but may also be obtained by introducing: A = (nl and
+ n2)Al
One. and the opposite. or H functions.
The relations between partial molarand inte~ralheats of solutions are ransted (1943a). 2 [re~onstructed as 3.directly translated (the word t ~ ~ ~ ~ o d uy ~ t ia n~ i ~ the tern A in the translatio~): f n n o used for
To visualize the connection one may utilize a graphic presentation in which one most advantageously utilizes the xconcentration scale and in place of A which applies to func~ion n1 + n2 molecules of mixture [utilizes] Ai. Az. but the final heat of solution. this text]. m much heat is evolved.reference (Taylor. the solubi~ io~ lity of the compound will decrease with increasingtemperature. the error would be flagrant. The initial heat of solution is positive.
At a further point in the text ansted (1943b) states that A the t h e ~ o d y namic function “can be the S. in the simple case. valid at constant temperature and pressure:
as well as the relation between the differential [thermodynamic functions]
The connection between AI. the curve will either rise or fall. If solubility ofa compound in a solvent isplotted versus the temperature then. hence.” The foregoing text talks to the difference between differential heats of solution and integral heats of solution. it may be that the total heat of solution is positive. if one deduced therefrom that the solubility of the compound decreased with the temperature.1 in Fig. obtains the following equation. is negative. examplewise the heat evolved per mole of sulfuric acid added to 1 mol of water. V. representing the dissolving of the last increment entering the solution at the saturation point. 193 1). is also true. the integral mixing [ ~ h e ~ o d y ~ a m i c ] for one mole of mixture.and hence a rise of temperature will result in the dissolving of another increment. 6. The heat of solution (IT)of a mixture of n1 moles of a compound A in n2 moles of a solvent
. to quote Taylor (1931):
nowledge that when water is poured upon solid p o ~ ~ s s i uhydroxide. E.
It is seen that the partial molar lution of water (A) and sulfuric acid ( are 6200 and 510 cal. (20) equals At ... . 3. The distance QP equals x times the slope at L. respectively.
where
and
ted the partial molar quantities of compone ubscript 2).A 2 . Hence onlypart of graph would apply. (20) equals A. his is shownfor a composition of 0. such as shown ig.
(subscri~t and compo1) of the sulfuric acid water
that the first column is the third column divide the second column. ). as shown in
+
. then the interce th the left axis gives HI = 3H/dnl and the intercept with the right axis gives p 2 = 3H/dn2 where n1 and n2 are the molesof water and acidin the particular amount of solution i.Enthalpy
N
M
0
x=o
ti20
x= 1
Solute
“The abscissa is x (Br~nsted. according to the Brransted Eq. Hence. x = ~ / ( n l na). is dissolved in water. value at D corresponds to H . tangent is drawn at any point of the curve.1 lists the heats of sol system.1933~): The thermodynamic function in the figure is alpy H . mu~tipliedby JD = (1 .x ) a ~ / a ~which. there is a limiting value for x. namely. then a graph.55 mol ofsulfuric acid added ater (i. (Le.. Table 3. The distance DJ is equal to the slope at L.A. x a H / ~ ~which according to the Brransted Eq.. a mole fraction of 0. the ordinate value ). If the se data are plotted.e. ). that correspond in^ to the solubility. = aHr/ax. at P equals A2 = a ~ ~ / a ~ .e.2 ensues.
30 .4
0.0
0.55 0.8 0.93 2.81 4.68 6.71 9. Heat of Solution as a Function of Compos~tion ole fraction acid ( X ) acid 1000 x heat evolved mole per of
A H per mole of solution
0 0.95 1.6 0.84 0.73 5.2
0.63 7.Water and Sulfuric Acid.87 406 3.58 1.
.
0.75 0.8
1.456 3.5 0.45 0.59 3. Marshall.95 1.4 0.oo
0 15.93 0.65 0.21 3.452 3.42 0.38 3.G
0.20 2.3 0.6 12. 1933.65 8.21 0.0
x
eats of solution of the sulfuric acid system: The abscissa is the mole fraction of sulfuric acid.45 0
0 1.2 2.56 2.6 1.64 2.97 1.1.35 0. 1909.2 0.37 3.43 0
Source: Data from Brmsted.1 0.9 0.94 10.7 0.85 0.
If less time is used. The concentrations are a function of time. then a conventional repr~sentation solubility versus temperature results. QU.1 exhibits a eutectic diagram such as shown in ig.3. 3.4a. for instance. There are many exceptions.3 the heats of solution are terminated by the solubility X .shown in Fig. 3. is a mpositional line where.G
0. so the solubilities are subtracted from 59 (see column 3 in logarithm takenof these numbers. is a solubility curve. The simple solution situationreferred to in Sec.
5:
X
Q
0
Q
c( 
I
V
0.
The subject of eutectic diagrams willbe taken up in a later chapter. These are plotted in Fig. there is an equilibrium between solid solute I and a solution of in water of composition x. It is noted that in Fig.Fig. at a given temperature.8
1.3. 3. 3. The socalled liquidous line in the right part of the eutectic. and if the axes are ~ i p p e as . and some of the situations of this will be discussed later in this chapter. but a short outline will be given at this point.1 but t e r ~ i n a t e ~ the solubility X . in essence.2
0.0
x
Heats of s o l u t i ~ ~ s depicted in Fig. as demonstrated Table 3. repreby senting the highest concentration. then the solubility may be obtained by iterative extrapolation. 83"C. ~ m m o n i u mnitrate solubility in water. 3. as shown in Fig. 3 This. It is seen that the data "seem" to level off at 59.4
0. ~ 3Ab. and the supernatant is assayed. T. of Solubility of solids are determined by placing an excess of solid in contact with the solvent in a hermetic containers (ampoule or closed testtube) and agitating it a in constant te~perature bath. and 126°C Maxima and minimain solubility/temperature curves also occur. It is conventional to use 7 2 h for e~~ilibration.
.5.0
0. statements may (incorrectly) imply that there is either an increase or a decrease in solubility of a compound with temperature. Samples are taken after certain in times (here multithen plotted as ples of 12 h).6. 3. exhibits breaks at 32"C.2.
872 1.ilit
Melting Pomt.639
0 30 45 52. X
H20
1
Solute. T
(a) Eutectic diagram of water and a solute.078 3.
60
M 0 0
L
. (b) The right side of the eutectic diagram from Fig.099
.. B.T Mole Fraction x
1
L
1
L
L
U
1 $0
Solid B+ Water
Q
Ice
+
Solid 8
4
a
Mole Fraction.4a plotted with reversed axes (i. Example of ~olubility Determination by Iterative ~ x t r a p o l a ~ i o ~ Time (h) Solub~lity(g/lOOO 59 g) 4.18 57.367 2.5 3 1.2
h[59 .
M
40
30
20
10
0
0
20
40
GO
80
Time (hours)
Data from Table 3.2 1
0 12 24 36 48 60
1.8
s
59 29 14 6. solubility as a functi~n of temperature).5 56 0.2.. B
Temperature. 3.e.
at a concentration level correspo~ding to an activity of a.e.998
. is given by
p 1
= po + RTlna
(3. (3..be the partial
(3.9) (3.6)
This when inserted in Eq.10)
. at a given tem~erature.4 treated by iteration. The conventional t r e a t ~ e n t of solubility as a function of temperature is to note that the chemical potential of a compound in solution. .
5
4
3 2
1
0
0
20 40 T m e (Hours)
60
Data from Fig. given by Eq. 1 mol of solid into a quantity of saturated solution and h" . lubility is best expressed as molality or as weightof solute per gram of .4403e2x R*2 0.6.h = (hs . The value of the iterant thatgives the best fit (the least sum of residual squares) is then assigned as the solubility. ~ ) =
h" . and obviouslyis the chemical potential when the is unity (i. 3. (3.ho~/T2}dT ~ d ( l n ~ a . (3.h. hen there is eq~ilibrium between a solid and a saturated solution activity (of the chemical potential of the solid p.ho)is the enthalpyassociated with tran~ferri~g . is commonly simply denoted h nes h. when a = 1 molal). po is a reference state.y
=
4 .6) then gives (after rearrangement)
{(h..4)
ere.4).. not per cubic centimeter of solution).5) relative to T now gives ividing through by T and di~erentiating { a ( ~ . as "the partial molar enthalpy of the component in the erefore. that is. equals that of the compound in solution. ~ / T ) / a T } ~ d{a(po/aT}~dT Rd(ln a.) = T It is recalled that
+
(3. 1 3 8 2 .e.
repeated with a figure different from 59.
by the heat of solutio^ not being temperatureindependent.S
(3.e.15)
d ln S/dT = A H / ( ~ ~ 2 ) ~ombining with Eq.15) then gives this Td In S/dT = { A / ( (3. cT2’ etc.12)
This is referred to as a Van’ with equilibrium constants9 written:
If y. is temperatureindependent9then the logarithm of the saturation on cent ration is linear in reciprocal absolute temperatur a plotting mode that is often use example of this is shown in Linearity of the Van’t ependent.he and solute at constant pressure. AC.. (3.16)
§olubility of ~ r t h ~ r h o m ~u~fanilamide Ethanol bi~ in Temp (“6) 47. (1984) hypothesizedthat if. (3.4 40.1 Solubility (g/ 1000 g) 28.15 2. i. rather than “the partial mol of solution of the solute. (3.where .34 3. 3 rant et al. is solubilities). is an integration constant. 11)
where S is the saturation concentration (in molality) and y. at the absolute zero of e t ~ a ~ tem~erature b is the change in the apparent partial molar heat capacity of t. e. If they are no off plot will not be linear.. is unnecessary.65 31 12 3. If this is assumed to be unity.36
Source: Data from ~ i l o s o ~ i c 1964. 3. More correctly Eq. we assume that it is a of linear function oftempe .2.15 1n[sl 3.19 3.) is temperatureindependent.82 2.. There isevidence that the introduction of terns containing hi powers of T. and (b) activity coefficient (y. is inde~endent temperature. = Y. An example of this is shown in Fig. h.3 29.
. = a + bT. is the activity co at saturation. properly.’9They interpret that “ a maybeconsidered to be the ~ y ~ ~ t ~valueof c AH.” If the curvature in Fig. as follows: AH.10) becomesthe well and often used equation ln[q = AH/(^^)}
+B
ff plot (although this latter. whichisitselfassumed to be independent of temperature.34 16.6 24. then Eq.1’7) (3.22 23. a.
(3.g.30 3. is the activity of the solute at saturation andis & given by
a.78 14.
This should produce a straight line with intercept B / R and slope
A and 13 may now be estimated from the slope and intercept of this line.y
=
12. T h i s ~ a c tin .5
1. 3.1998. .3
1000/T
3.e.17) versus 1/T. .000
1000/T
able 3. at constant temperature and pressure.3 plotted according to Eq.) i. et 1~96).
Equation (3. .18) corresponds to “the partial molar enthalpy o f the c o ~ p o n e n tin the . ~ r s t a p p r o x i ~ a ~ iin n s~ o n l i program. . itself.7
1.. (3. solution . (3. Consider the diagram in Fig.2 .1 the heat absorbed.5
Solubility of dl p[pseudoephedrine]. To obtain good estimates. solution. .8
1.1.” ( ~ e n b i g h1961). one must have a good estimate of A and B.
.17) integrates to Ins=
(3. and oa ~ e ~ ~ may be used a s . 8 9 6 1 ~R*2
.1. when 1 mole of the component dissolves in the . be emphasized that the enthalpy term in both (3. Suppos the depicted compound at a ternperature of TI had a solubility corresponding to and at a higher tem~erature a had
1. (Data from ~ u ~ i p e d d1996.18)
This may be fittedby nonlinear programs. . This approach has been employed byseveralrecent investigators (~udipeddi.380 .9
1. again. ~ozwiakows~i ai.6
1.4 3.2 3. It should.12).1
3. ~ a ~ itequite unde~standablewhy the Van’t ~o~ can not be expected to be s linear. . most graphing programswill calculate (d In Cs) which may then be multiplied by I“ and plotted by way of Eq.4
3. ii. butfor these to work.
the change in heat capacity also changes. Figure 3. for instance.10 shows the effect of sodium chloride concentration on the solubility of a bisnaphthalimide derivative.
The use of mixedsolvent systems isoften necessary in pharmaceuti~s when a drug is poorly soluble.992
25
10
3. 3. rather than in grams per 1000 g (g/lOOO g) of solvent
*
The solvent has a great i ~ ~ u e n on solubility and should always be speci~ed. hence. (1998) have reported on the solubility of aspirin in 38% alcohol.
solubility corresponding to L.78 .8 versus n
T". Cosolvents used are Ethanol Propylene glycol Glycerin olyoxyethylene glycols Ternary diagrams are used to visualize wheremaximum solubility occurs when more than one solvent is used (Fig. 3. an electrolyte). EC is the percentage of glycerol. To assign the change in heat capacity as an explanation to the nonlinearity is rational only in the sense that the composition changes with temperature. There are many examples of this. itself. the concentration of electrolytes may greatly affect the sol~bility of a compound. 3 6 2 ~R2 = 0.15
1000/T
'
I
I
I
Derivative curve (d I S ) of data in Fig. Longuemard et al. the length WB is the amount of ethanol and. they failed to obtain linearity according to the 'Van't Hoff.1 1). then the ~ ~ f e r e n t i a l e n t h ~ ~ i e s of solution^ w o u be ~ ~ a f ~ ~ c t i of nt e ~ ~ e r a t ~ r e . ce aqueous solutions. that this is particularly true for a compoun~ that is.9 3 . The length of PA is the percentage of water. linear. (It will be seen later. The lines in this presentation mode are parallel to the sides in the triangle.y
4 0 F"
331. although in this case the curvature may be because the ordinate is in grams per liter (g/L). but it is rather to be expected. it is not unexpected that the Van't Hoff plot is not hence. here. In a different presentation mode they
.
etc. A point inside the triangle. M
0.4. The figure to the left in Fig. such as would result. and such points can be connected to form isotherms and diagrams. the more of one cosolvent that is added.19)
where A and B are constants and E is the dielectric constant of the solvent. 20 mglg. An example of this is shown in Table 3.00
0.10
Effect of salt concentration on the solubility of a bisnaphthalimide. 3..
. ften.) for given compositions of the solvent.
requently the solubility is a fun~tionof the dielectric constant of the medium. in which the solubility of a c o m ~ o u n ~ is tabulated as a function of the dielectric constant of the medium (glycerin/water
Water
Glycerol
Ternary diagram. such as one given composition.12implies a maximum solubility. the larger the solubility is. whereas in the other diagram. the relation is that of the Jaffe equation: In[SJ = ( A / & ) l3 . 1996. then the solubility would be the same (10 mglg.
+
(3.05
INaCl].0
1
2
3 4
5
0. (Data from et al. If solubilities are determined for many solvent compositions.)
are cast perpendicularly to the axes.
the plotting as a function of their dielectric constant.912 0.
Effect of Dielectric Constant on Solubility of ~isnaphthalimide Dielectric constant 78.274 1. but they will often show maximumsolubility at a given dielectricconstant (Fig. with hydrophobic drugsthe solubility decreases withincreasconstant.9 52.833 0. It is particularly useful. and an example of this is shown in Table 3.14).086 0.02 1O O / ~ ~ 1. 1971.333
. to carry out solubilities in solvent pairs of different ratios tovary the dielectric constant.6 45.45 42.4 and Fig.5 74. 3. from a practical point of view.335 1. This is the socalled cha~eleonic effect (Sunwoo and Eisem. Shino~a 6 (1978) defines this as
(3 20)
s the heat of vaporization of the solvent.13.200 4.49 2.900 2.
olubility profiles visavis the solubility parameter of the solvent at times shows mult~plepeaks.15 gives an example both of plotting the solubility of a com~ound (caffeine) in solvent rnistures with different solubility parameters. V its molar volume.00 1. almost any other solvent pair willshow maximum stability atthat her than using dielectric constant as a measure. and the practical part of this is that once this is established. The opposite happens at times. Graphswill often be linear when plotted asin Fig.5 Solubility ( S .30 2.H 20
Hz0
Glyerol
Ethanol
Glycerol
Ethanol
Ternary diagrams of the two types of solubility. 3.42 1.13. the Hildebrand solubility ~arameter is often employed.09 0.517 1.9% 1nES3 0.693 0.9 65.
ost often. and 3. mg/mL) 2. 3.35 1 0.
D.)
0
10 20 30 Dielectric Constant
40
50
Effect of dielectric constant onthe solubility of phenobarbital in four systems: A.)
.. and the lower curve (with the lower abscissa) is the solubility versus the solubility parameter (ai).0006
.0 17
0.. 1961. 1996. 3 C.g1ycerin:ethano~. ropyleneglyco1:ethanol.2
0.
y
=
2.. 4 7 6 ~ R^2 = 0.. 1964. (Data from Martinet al.limide.0 12
'
1
I
0.999
0. € .022
l/(l)ielectric Constant)
ffect of dielectric constant on the solubility of bisnaphtha. (Data from aghavan et al.)
Dielectric Constant
100
20
40
60
80
"
5
10 155 2 20 Solubility Parameter
The solubility of caffeine in a solvent cansisti~gof dioxane and water at 25°C: Top curve (with top abscissa) is solubility versus dielectricconstant. watecethanol. propylene glyco1:water.8 6 . ata from Lordi et al.
Romero et al. drug A)? is called the substrate and the other ligand. a straight line should ensue with a slope of b. given by
(3..2~) The equilibri~m constant of this (the stability constant) is (3. one of the two components of the system (e. polar solutes in semipolar (or polar) solvents (Jouyba Acree.g. 1998). by in .26) ] in the equilibrium equation [see Eq.21)] gives which rearranges to (3. ~omplexation often applicable to solubility problems in pha~aceutics.28) ence by measuring the solubility as a function of the added ligand concentration).25)
and Inserting the expressions for [ (3. (3. 1994. (324)
so
(3. The measured solubility Sobs is the solubility S plus the amount complexed. it also exists for and the molecules appear to adjust their solubility to fit the tin et ai.ustamante et al. is A drug A (the substrate) will react with another compound I (the ligand) and form a 3 weak e ~ u i l i b r i u ~ . 1985). If.. ~ystems this type are often characterized by of nonspecific van der ~ a a l forces as well as strong specific interactions? so that the s Hildebrand solubility parameters no longer can explain the instance. An exampleis ascorbic acid/niacinamide (niacinamide ascorbate).
Drug substances maycomplexwithcomplexing agents. then (B) will be this a ~ o u nless the amount complexed. we denote the concentration of ligand calculated based on the t amount added..
(32’7)
. 1996). In general.21) K =[ ~ ~ l / [ ~ ~ [ ~ l The concentration of u n c o ~ p l e ~ substrate is the solubility when no ligand is ed present. [A] = S The ~uantities brackets are actual concentrations in the complexed system..
18). without ouessidjewe. and at 1 asslebach equation will predict that
y
100
1.3778
+
4 3 . Stella. for instance.” This affectsmany of the physicochemical properties of the complexed drug.. approximately
S=S.
he following ons side rations deal with the solubility of an acid as a functi but the inverse problem of an amine and its solubility as a function of follow the same lines. L6pez et al.000

80 60
‘r:
40
d
3
20
0
0 1 2 ~ ~ a c i n ~ mConcentration. in a softshell capsule. then the equation becomes. If the two compounds are placed. 1996.lope = B = K S / ( 1
+ KS)
(3.29) (3. which allows the complex to form. ~) id~ from Raghavan et al. 1996. 7 3 3 ~ RA2 1. The complexation equations have been e x a ~ i n e d extensively for 1:1 complexes iguchi and Connors. form inclusion compounds with many drugs. Loftson ng pharmacological properties ( rewster. ~omplexation is useful in several applications of solids.
A special note on cyclodextrins is in order. for example. So in general they are pre~ i time n reacted by placin~them in a mixer. and then drying the powder mass. M ide
3
Effect of a ligand ( n i a c i n a ~ i d on the solubility of a b i s n ~ p ~ t h a l i ~(Data..+K
as demonstrated in Fig. Mostlyan acid is less solublethan its salts. on the size of the ring). and Loftson and the ~ridriksdottir (1998) have shown the effect of watersoluble polymers and of a series of drugs on the solubility of pcyclodextrin in water (Figs.16. 1965. They take up either the entire molecule or some hydrophobic portion of it into their “cavity.17 and 3. ascorbic acid and niacinamide complexes . 3.. adding alcohol.)
. The problem may be considered in opposite sense.30)
hen S is small. Irie and and Uekama. 1996). and y depending p. These compounds (a. they will i n t e ~d ~ t (and the complex is yellow). 1991). 3. 1996. 198’7. Ahmed et al.
is lessthan what is calculated from the solubility product of the metal counterion (3. '
52
60
70
I65
mg Drug/g Cyclodextrin NoPol CMC PVP HPMC
The solubility of a hydrocor . let us assume that p moles of been added. (3. so that both [ 1 and [A"] is equal to p .32) until such t this point.)
or
where CHA is the solubility of the acid form.
/ycyclodextrin complex in aqueous solutions of from Loftson and Fridriksdottir.33). 1998.)
. CA is the concentration of the acid anion. 1997. (Data from Loftson and Fridriksdottir. PVP. then the amount in solution will increase by Eq. If the solubility of HA. denoted SHA. and th is ~uf~ciently that high the solubility is sim the concentration of A".70
1
E
HPMC Percent W/V
Effect of HPMC on pcyclodextrin solubility in solutions saturated in carbamazepine. (3.1% w/v HPMC. andply: is the pK of the acid.33) then the solubility S is given by
curve would have the appearance of a titration ould be given by Eq. A is increased by addition of an alkali . or 0.
As far as solubility is concerned. 1984. 1985. One of these is the RedlichKister (or. and a short note on the effect of micellar systems is of importance. Here Xm. and /lo. 1994. and . 1991. 1996) to establish a reliable predictor for an unexperienced solute solubility in binary mixtures of solvents with know neat properties. the phenomenon is one of partition. Williams and Amidon. Often the important aspect of solubility is indissolution testing. Ochsner et al. 1996.35)
If a bit more MOH is added. 1981. Acree. Hammad and Muller (1998). does not start taking place before the critical micelle concentration (CMC) is achieved... have the appearance shown in Fig. ~ j the e ~ c e ~ t that~the linear t io increase in solubility of. then often.” solubility in micellar systems is of importance. the drug. Solubilityplots. Micellar systems. 1980. If the acid is dibasic.B2 are leastsquares~tted constants in the equation:
~ o u y b a n ~ a r a m a l eand Hanaee (1997) have investigated this equation for a series ki of hydroxybenzoic acid esters. for instance. For instance. arzegar~alaliand Jouyban~haramaleki. the concentration of each ion would bep is the small amount. there is a p a r t i t ~ o ~ solute molecules of between solute in the aqueous phase and solutes in the micellar phase.K = p2
(3. for example. 1992.
~lthough perhaps not strictly applicable to “solids. in a certain way. 3.
There have been a series of attempts in literature (Martin et al. are similar to twophase systems with one big exception. ~alkowsky and Roseman. in micellarsystems there is an e ~ ~ i Z between j ~ i ~ ~ monomers in solution. therefore. the volume X b fractions of the two solvents A and El when no solute is present.36)
so that MA will precipitate out: a plateau would be reached.
. and the oil molecules in a droplet are fairly much the same independently oftime. the mole fraction of the solubility is related to fa and ji.19. the CNIElS/. Acree et al. and molecules of the amphiphile in the micelle. is smaller than that of HA and in this case the solubility versus pH curve would decrease above the second pH of the compound. 3 /11. so that in that aspect. the solubilit~ product of zZA.RK) equation. where ‘ and X denote the molefraction solubility in the neat solvents A and I of the solute. has reported on the solubility of clonazepam in mixed micelles. whereas in oil in water systems the oil droplets are continents unto themselves. But Cp+A}2>P2=K
+ A where A
(3. BarzegarJalali and Hanaee.
The con cent ratio^ profile. is governed by the following equation: dC/dt where
. The method of approaching this is as follows: the rate with which the compound goes into solution in V milliliters of liquid is given by:
d ~ / d = VdC/dt = kAS t
(3. ~ a l a v i o l l e al.
When a compound is poorly stable in aqueous solution.39) is mass not dissolved at time t.3
0..5%.40)
where firstorder kinetics are assumed. it is worthwhile selecting smaller value for the sampling interval.C)
(3.42) (3. 1987. ruelove et al. A is the surface area. (Data from H a ~ r n a d and Muller.1
&a
“ I
0
v)
0. For instance. 20 min. Then. and k is the intrinsic dissolution rate constant. 1984. et ogdanova et al.. 1980.1
0. q. because the longer the solubility experiment goes on. the a more drug substance will degrade.kl C = kl(. Z= ( k A S / k ~ ) 8 V
=I=
( k A S / V ) . for instance. and where k l is the decomposition rate constant.43)
This may be rewritten dC/(B .41) (3. S is the solubility.38)
or (3.G
SPC Mol Fraction
Solubility of chlonazeparn in soya phosphatidylcholine.00 0.4
0..5
0.)
Niacinamide is often used to increase solubility of drugs (Fawzi et al. therefore. 1994).2
0. 1998.0
0.C ) = kldt
. The rate with which the drug decomposes is given by:
dC/dt = klC
(3. (1998) have shown that melts made with niacinamide and indomethacin give rise to a maximum solubility at an indo~ethacinconcentration of about 7..8 . then the Nogami method may be particularly useful.Chen et al.
but other circu~stances the (temperature ~uctuations) could also account for this. T a ~ e n its fullest.’’ The equation.45)
where CT is the inte cia1 energy between solid and liquid.Co] (3. M is molecular weight. = activity of component A in saturated solution
=
3
=
. known as the stwaldFreundlich equation.C] =: ”k. the gas constant. and they are often used in on tests for poorly soluble drugs. relates the solubilities S1and S2 of particles of size rl and r2 by the following equation:
(3.. on crystal~ization. ligand or substrate 1 = ther~odynamic function of c o ~ p o n e n t 2 = Ther~odynamic function of co~pollen = t h e r ~ o d y n a ~function of ~ i x t u ~ e ic l/dnl = partial molal thermo~ynamicfunction of comp ~ A 2 / d ~partial t h e ~ o d y n a m i c = ~ function of component Jdx = slope of thermodyna~ic function versus c o ~ p o s i t i ox curve (~ ) a. s absolute te~perature. and this then gives both the value of S and kl (provided the surface area is known). such as the polysorbates.44)
The value of may be found by iteration.
stwald (1900) and ~reundlich (1922) postulated that the size of a particle affected Its ‘‘solubility. The derivation of and will be shown in Chapter 6. willsolubilize drug compounds when the surfactant is present in excess of its critical micelle concentration ( solubility will increase linearly with surfactant concentration. Jeannin et al. refuted ne reality of the equation is that. is difficult to prove e~perimenta~ly with prevalent values of and. the equation predicts that real equilibrium existsonly to between an infinitely large amount of liquid with a single. p is density. in~nitely large crystal.
urfactants. 1975).
constant in the extended Van’t off solubility equation concentration of A. here will be more description of this in the next chapter. but the equation has also been.t
+ ln[P . rfactants also aid in the wetting down of a solvent. smaller particles often disap~ear at expense of larger ones (ripenin.which integrates to 1np . in polydisperse suspensions. T p ) . ). or as the stwald ~ p e n i n g effect. real differences would be difficult to pts at this have been made ( ~ ~ o l and en 1972.
Guerrmi M. Naggi A.chemical potential of component A in saturated solution po = standard chemical potential of Component A in solution p = density CT = interfacial tension between solid and liquid /dnl
=
Acree WE Jr (1991). Acree WE Jr (1996). Int J P h a m 140:237. Int J Pharm 198:281. 216. J Pharrn Sei 56:847.
. M rl = size of a large particle r2 = size of a smaller particle S = solubility of component A SI = solubility of large particle S2 = solubility of smaller particle T = absolute temperature V = molar volume nl = number of moles of component y 2 = number of moles of component 1 x = mole fraction Xa. .P I . enthalpy associated with transferring I mol of solid into a (large quantity of ) saturated solution = heat of vaporization enthalpy of solid h.37) ivity coefficient of component A in saturated solution ebrand solubility parameter /A. Hanaee J (1994). (3. Acree WE Jr (1992). Kinkel AW. Aguiar AJ. New York.= =
constant in the e concentration
off solubility equation
1 = concentration o
h e = heat capaci
differencebetween component A in solution and solid
ions two solvents A and €3 when no solute is present slope of integral heat versus composition (n) curve olal enthalpy of component A olal enthalpy of component I3 enthalpy of solution. Krc J Jr. Thermochirn Acta 198:71. /32 = leastsquaresfitted constants in Eq. Int J Pharrn 77:247. .X b = ~olubility mol action of a compound in neat solvents in /3 = constant in the Van? ff equation for solubility BO. Academic Press. = enthalpy of component A in saturated solution R = gas constant K = complexation constant = (a) molecular weight or (b) counter ion 1 = concentration of metal counter ion. Ahmed SM. Bikeman JJ (1970). Thermochi~ Acta 178:152. Int J Pharrn 127:27. Samyn JC (1967). Physical Surfaces. JouybanGharamal~ki A (1996). Focher B (1991).
Fysisk Kemi. J Pharm Sci 86:147. M (1980). Needham T. Int J Pharm 169:55. Sisco JJ. pp 336338. . 336339. J Pharm Sci 69:487. ~hermochim arshall AL (1931). Gray DB. Chen A. 2. ransted JN (1909). ed. r0nsted JN (1933). Carst€nsen JT (1975). J Pharm Sci 85: 1017. 6. Lianos CME (1998). Int J P h a m l67:177. New York. "he PrinciplesofChemical Equilibrium. Van Nostrand. ir H (1998). pp 575602. Fysisk Kemi. Cohen S (1985). Liron Z. Georgieva S (1998). ed. Pharm Res 11:398. T Hill AE (1931). Int J Pharm 163:1. Martin A (1994). Mora JG. i A. p 155. i A. 536539. Nash R (1994). Int J Pharm 18:27. Pharm Sci 74: 132. Pharmaceuticaland Medical dextrins. p 139. Qchoa R. Copenhagen. artin A. Acree WE (1998). 996). CW Joz~iakowski (1972). alaviolle I. ~ p a ~ k c a k (1996). l rie T. Bernard MJB.openh hag en.~ogdanovaS. Adv Anal Chem Instr 4: 116. In: Taylor. School of Pharmacy. Van Nostrand. Diaz D. Reillo A. Acta 121:283. A. Nguyen NT. Zia W (1997). Hammad MA. oy K (1970). p 140. Munksgaard. Parta AN (1964). Pharm Dev Techno1 3:395. Medhizadeh M. J Pharrn Sci 53:463. Grant DJW. Cienc Ind Farm 6:325. Pharrn Dev Techno1 1231. A Treatise on Physical Chemistry. Chem tamante C. Amsterdam. Terol A. tamante C. Escalera JB (1994). nsted JN (1933~). Colloid and Capillary Chemistry. Uekama K (1997). Hanaee J (1997). p 257. DucheneD. J Pharm Sci 70: 1115. Rei110 A. University of Wisconsin. Hi~uchi (1958). ~ u n k s g a a r d . Chem arstensen JT (1977). Copenhagen. . p 141. Int J P h a m 163:115. Ann Pharm Fr 33:433. Masse J (1987). Brewster ME (1996). Int J Pharm 154:245. J P h a m Sci 87:193. Chern Pharm Bull 44: 1061. M. ~ e M a u r y Chauvet A. Karaivanova V. Ambrosio T. Lordi N. Zito S. Denbigh I (1961). Fairbrother JE (1984). HS. Guyot JC (1998). Squillante E. Chow AHL. Connors K A (1965). Escalera JB (1994). ~
. Munksgaard. Adjei A (1980). In: aylor HS. Wouessidjewe D (1996). New York. Sciarrone . Higuchi T. Phys Chem 68:700. Brsnsted JN (1933b). J P h a m Sci 69: 104. ~uyotHermann AM. Galan YCR (1987). New York. Adjei A (1981). J Pharm Sci 74:638. 6). Fysisk Kemi. Hussain MA (1996). Wu PL. Int J P h a m 17051. ed. Muller BW (1998). Cambridge UniversityPress. J P h a m Sci 61:281. MJ. Martin A. pp 19721 5 . FormulationandPreparation of Dos Elsevier/NorthHolland ~iomedical. Marcel Dekker. In: Durnitriu S. Freundlich H (1922). pp 336338. J Paint ~ e c h n o 42:76. A Treatise on Physical Chemistry. Loftsson T. Martin A. Paruta A . J Am P h a m Assoc Sci Ed 47:657. Int J P h a 159: 171. J Pharm Pharmacol 46172. New York. Qchoa R. Newburger J. Sidzhakova D. J P h a m Sci 85:1142. Dutton. C London.
s Yalk SH. Roseman TJ (1981). In: Yalkowsky SH. Int J F h a 19:17. Eisen H (1971). New York. A ~ keda M. ed. . Yamamoto K (1997). Chen N. Amidon CL (1984).
. Int J ~ i l l i a m NA. ~ishimuraK. pp 91134.Sunwoo C. ~ u s s a i n (1984). J Pharm Sci 60238. J Pharrn Sei 79:9. Techniques of Solu~ilization of . Dekker.
This Page Intentionally Left Blank
.
there are are vailable for particle size and particle distri~utionassessme~t ith a multitude of methods.5.7. 4.4.
52 52 53 53 5
4. 4. ome of the methods to be d e s c ~ i b e ~ old and tested.
58
58 59 59
In the previous chapters the subdivisions of the solid has been of no importance stwald~reundlich equation.9. but today.4.8.2. it becomes important to be nguishbetween the many definitions of particle size that 990).6. will
.In general.
4. The most im~ortant be discussed in the following.11. 4.1.
4. solids are multia solid sample is usually more than one particle). however. 4. 4. 4.
nd Cas Adsorption
55
lectronic Counters and Laser Counters eflectance Infrared (F
56 57
4.3.10.
s = r(v2l3)
(41)
where I“ is denoted the general shape factor. so the ratio is ~ / { n / 6 = ~ / ~ = 4. h) AB.1 ~ ~ l e
Calculate the shape factor for a sphere. but h is usually “hidden” because the particle lies on its short side. if the particle were either a sphere or a cube.84.1. This will be discussed further when shapes are discussed. ~ o ~ ~ d i s ~powder is one for which all the particles are the same size. common isometric shapes. that is. and a right cylinder (one for which its diameter equals its height). it could be either the small dimension (the height. b ( ~ ~or) one of the hich dimension is chosen is often a question of which measuring method is 0th b and E may be recorded.
ouu
0
Monodisperse Isometric Polydisperse Isometric
i
Polydisperse Nonlsometrlc
States of subdivision. A. the is nd2.
A ~ 4. s to the twothirds power of its volume v.2 shows two situations for which the type “diameter” must be defined.The three most prominent subdivisions of multiparticulates are illustrated in Fig. it will suffice to say that there are three. igure 4. long di~ension(the length. article is one of whichthe surface s may be expressed as the ratio of the surface.
. more than one definition becomes possible. In the parallelepiped. the width or breadth. a cube (with shape factor 6). erse erse powder is one. but once the shape of the particle deviates from that. a sphere. 1 ) AC. A. } 62/3n1/3
he “size” ofa particle would be easyto define. E ~ ~ 4. but in s this case one often records the diagonal. the . 4. here.1~ ~ ~ r
lume of a sphere is d3n/6 and the 2/3power of the volume is d2{n/6}2/3. the particles of which are not the same size. The same is true for the a ~ o r p h o u particle.
5)
+ (4
X
12. logically. but in general.An orthorho~bic crystal (a para~lelepiped)and an oddly shaped particle (e. ~edimentation methods (And
In the following.5)
+ (1
X
3~. briefly. be given by (2 X 1) + (3
X
5.1. a very dilute suspension of a sample (e. or treated.. as well as with distributions and surface areas. and placed on a hemocyto~eter slide. an a ~o r p h a te ) . schematic. in mineral oil) is made.g.. isometric particles will always be a ~ ~ r o x i m a t e d cubes.6 pm
This type of diameter is called the ~ ~ i t ~ ~ e~ t i i a ~ ~ and e denoted by ~e~~ ~ is ~ e ~
. and so on. is usually referred to as ~ i c ~ o ~ e ~ i t i c ~ . An average particle diameter would then.g.
In optical microscopy. because this by shape is closer to a that of a real particle. how theyare measured. The most common methods are ~icrosco~y Screen (sieve) analysis Electronic counting (Coulter..
The foregoing definitions refer to single particles. The results may present themselves (in a simpli~ed fashion) as shown in Table 4.e.5)/(~ 3
+ + 4 + 1) = 106/10 = 10. particle dimension will be denoted by a. multiparticulate).g. particles exist in a ~ o ~ ~(i. the measured. The easiest method of differentiating between the l a ~ ~ o ~ various types of particle sizes isto describe.. In this writing. The number of particles of a range in the field (e. then the number between 10 and 25. between zero and 10 pm) are counted and noted. The scientific field concerned with these matters.
Isometric particles are frequently approximated by spheres.
Withscanning optical microscopy (S ). the measured diameteris not necessarily representative of the larger lot from whichit t is collventional to measure in such a fashion that the total number of articles is about 300. diameter a2. used and triclinic phases of carbamazepine.38 1406.5 37. e ~n r one considers the volume V and the surface area of a sphere.25
.75 625
1 52734. (1997).5 5.5 499. and so on (Table 4. For isometric shapes this is inde
iven a particle population o f rzl particles of diameter a l .3)
so that for a nonspherical particle one may generalize that
as. for instance. diameter a.5 1
The problem with microscopicdiameters is that the sample size is verysmall.5 4 12. Ceolin et al.2).= ~ V / S
(4.75
4 12.5 78 12.. with. then the ratio of V to A has the dimension of a diameter
V / A = {(n/6)a3)/(na2) a/6 =
(4. it ispossible to increase the measured number considerably.4)
asv is denoted the surface volume mean diameter. but the sample size is still small.
uch larger magni~cations than achievable b . The type diameter obtained by pemeametry is called th surface ~ o l ~ ~ a ~ iea ~ e t eIf .Example of Microscopic Particle Count ~ u ~ b e ~ Particle size (a) p m 2 1 3 5.125 90. it is seen that
rz2
particles of
Example of Microscopic Particle Count Number ~ a r ~ i csize (4 pm le
Ea3 3
yla
2 1 2 2
3 37. hence.
ical microscopy are achieved by to distinguish between trigonal
In permeametr~ is actually the surface area that is measured and this method will it be treated in more detail at a later point.
The openin~s the screens are described by a U. would then give the surfacevolume diameter
a. As the wire has a width of its of cannot deduce the size of the opening by dividing 1 in.5.1.4. per gram of solid.5 410 GOTO 110 500 “”Diameter = ”. the volum
V = l/p
where p is the density. i sample).e. ultiplying this by 6 and dividing by A. shows common mesh sizes. very common manner of measuring particle size in industry ~nprocess sieve is analysis (Fig. A The surfacevolume mean diameter is fre~uently called the ~ ~ ~dia. = ~ ~ ( p A s )
Scanning electron microsco~y (SI3 ) may be used for small particle sizes and the procedures used are quite the same.5.DI 120 Q2 = Q2 + 1 140 VV2 = Nl*(D1*2) 150W3 = W3 + W1 160W4 = W4 + W2 170 D2 = W 3 ~ ~ ~ 180 IF Q2 = Q1 GOTO 500 190 GOTO 400 400 DATA 2.1. by the number. and the arithmetic mean diameter is.. which in indicates the n u m b ~ r strands per inch.3).. by T).3.5.so that
(i. A. to a.
Program for Converting a.Ql ~ ~ 110 READ N1. 100 I ~ ‘“Number of DataTSets = ”.r meter.. where V and S are the volume and surface area of the ~ o ~ ~ Z ~ tthe o ~ From the sums it can be calculated that
as. 4. esh Number. surface area measurements (to be discussed later). = 28. in similar fashion called t h e ~ ~ ~ ~ ~ i ~ ~ ethe e ~ ? t moment denoting the power of the n u ~ e r a t o r . for instance.I22
~~~~~
.74211 = 28.7 pm
r such conversions if they are often needed.S.12.
A. ng the specific surface area.37.
d2 . The electric conductivity over the aperture is then measured.063 0.59 0.
If a sieve analysis is conducted on a W gram sample.11)
w.250 0.S. the conductivity is reduced by an amount corresponding to the volume of
U. Every time a particle (with essentially negligible conductivity) passes through the aperture. is assumed to have an average diameter of d = (0. which is pumped and circulated through the aperture.A
B
Example of sieve analysis.210
80 100 120 200 230 270 325 400
Opening (mm) 0.053 0. This is a fourth~oment diameter and is denoted as the weight
mean diameter.21)/2 = 0. then the masses (weights) of the fractions collected on the various sieves are denoted w l .044 0.125 0. a sample that went through a 60mesh screen and was retained by a 70mesh screen.. conventionally? assumed to collect particles of diamters of d l .177 0. Mesh Openings Mesh 10 20 25 30 40 50 60 70 Opening (mm) Mesh 2 0. For example. .037
.69 0. w2. The “average” diameter of the entire sample may be expressed as
+
(4. Each sieve is.297 0.23 mm or 230 pm.*. of which the diameters are the average valuesof the diameters of the confiningscreens.. The electrode is placed in a dilute suspension of the drug substance.
er r In the ~ o ~ l t c o ~ ~ t ean?electrode with an aperture is employed.149 0.84 0.25 0.074 0.42 0. .
If the particles have a hydrodynamic radius of a. The new concentration C1 may now be used to calculate the fraction. At a giventime t. sodium metaphosphate is added as a deglomeration agent. the method is slow. The instrument is adjusted to a threshold value T l . larger than a. Particles from a population are sampled and added to water to a concentration no larger than 2%. directly.13) Andrks et al. a sample is taken through the stopcock. 4. whereas laser light scattering detected a trimodal distribution. a size simply not detected in optical (projected) light microscopy. the procedure is repeated. The possibility of two particles passing at the same time is taken care of by a coincidence factor. It gives. one (weak) mode about 1 pm.
. u and vb. An assay of this then gives the analytical concentration C ofsolids in the dispersion.. in the end. 2. mean (4. If the average of these is denoted ui. 3. IJsuaily.counting the number of particles of volume larger than V2. The sample size is large compared with other methods. with a density that is A p larger than the dispersion ~ e d i u m with viscosity r] will be given by
where g is gravitational acceleration.
113
(4. the alvern counter employs a laser beam that is interfered with by particles flowing in its path. and attime zero.so that only the number of particles ofa given volume V I . led to a monomo~al distribution. = [9~20/(2Ap~)~ havepassed the 20cm will . This threshold is then changed to a different threshold T2.and so on. It gives a defined radius (the hydrodynamic radius). obtained by microscopy. the oversize distribution. results may appear as a cumulative ~istribution function. then the steadystate velocity v of a particle. denoted the s u ~ f a c ~ d i a ~ e t e r . Particles.12)
In a similar fashion. so that one knows the number of particles rzi. but it has several advantages: 1. so that.liquid it replaces.4) depends on Stokes law. This leads to the concept of a crosssectional definition of a diameter.
The Andreasen apparatus (Fig. that are in a certain interval of volumes. The suspension is dispersed. It is possible to convert the cumulative distribution to a frequency function. (C f particles with a particle size larger than a. This introduces the concept of a v o Z u ~ e mearz ~i~~eter. is counted. then an aver~ge . y modern standards. (1998) have shown that comparing a set of data of particle size distributions of fenofibrate. diameter can be calculated from this.
The obtained results give good correlation between diameters observed and those observed from some (but not all) of the other methods mentioned.
larger than in the other “precision” methods. It is the ratio of th meter to the ~ e r i m e t ofrthe circle with the same area as the particle. therefore. and the quoted article does not do that. It was mentioned earlier that it is important to disting~ish between the various kinds of dia~eters. particles possessing a radius that is larger than a. =: [9q20/(2Apt)] willbebelow the mark B. will not be pipetted out. The e Z o ~ g ~ tfactor is the ratio of largest to smallest diam~ter (diagonal) is used as a measure of particle shape.S topcock
. that there are several ways of expressing iameters are often determined microscopically. These particles. leading to some specu~ation about why the diameter values from the FTZ diffuse method differed from a value obtained by SE
e will mention more about particle diameters related to their shapes as this text to say at this point. then the eywood factor will be close to unity. ~ which ~ Another method is the ~ e ~ factor. ctual geometric s~rface area (the area ~ a l ~ u l a from the geometry of the te~ . ~ close to circular.Liquid Level o f
Suspension
L
Andreasen apparatus: At a given time t. assuming to be it completely io~ smooth) is often calculated. ne presentation method for a ~ i a ~ e ~ e r . is the diameter which a ~ o u n t i n g the shape is the ~ r o J e c tse~~r f a for of a circlewith the same area as the particle.
. o o attempts to describe the circularity of the particle projection (microscopically).
which equals average ofconfining screen o~enings g = gravitational acceleratio~ i = ru~ning index t = time s = surface of a single particle S = surface area of a popluation (sample) ZI = (a) volume of a single particle. louquin P.
a = size. RCginault P. = specific surface area. Dzyabchenko AV. Lafaye A. area per gram h = height (small dimension) of a particle AB E = the long dimension of a particle b = breadth of a particle dum = ~ ~ j d Xi n i d ~ / ~ ~ i d ~ mean diameter in sieve analysis = / ~ X weight dj = average diameter insieveanalysis.
. Agafonov VN.Int J P h a m 147: 129. = (Xna2/Xn}I 2 . Toscanini S. Int J Pharm 171:lOl. Lafaye JM (1990). surface volume mean diameter d ~ / a. diameter a.Chebli and Cartilier (1998) have determined that samples of microcrystalline 101) and crosslinked cellulose (CLC) of particle diameters of about 50 pm and found their eywood factors to be 125 and 179. Cartilier L (1998). = hydrodynamic radius A. a. respectively. (b) Stokes velocity V = volume of a population (sample) W = X w i = weight of a sample for sieve analysis 17 = overall shape factor Ap = difference in densities of solid and liquid in a settling sus~ension q = viscosity
Andrks C. Bracconi P. Pourcelot Y (1 968). = X ~ Z ~ / Xarithmetic mean diameter ~Z. Bachet Chebli C. Rochat MH.. asu = ~ r ~End2. . Powder Techno1 60:205. Matsuda Y (1996)." = { X r ~ d ~ / X n } volume mean diameter "~. Cardette ME. J Pharm Sci 85:112.surface mean diameter ' a.
This Page Intentionally Left Blank
.
Porosity 5. Shape Factors by Way of Fractal Dimensions 5.2. and Teller) Isotherms istributions: The atchChoate Relations
62
62
63 65
66
67 67 69 74 75 75 78 79 79 81
85
85 87
~ulti~articulates often.5. 5.15. are notmonodisperse.10.12. 5. The expected distributions will be overvie~ed in
.6. they most contain a spectrum of particle sizes.3. Emmett.1. 5.14. 5.13. Distribution Types 5. 5.11. 5. 5. Permeametry (Carman5.16. FreundSich Isotherms 5.9.4.8.7. Symbols References olydisperse Particle Populations dsorption Isotherms er. Adsorption Isotherms 5. Surface Areas from article Size Distributions 5.5. as seen in the last chapter. 5.
efore discussing this it is worthwhile to have an overview of the concept and the definitions associated with “diameters” or “sizes.”
There are three commonly used dimensions employed in microscopy. erdan (1961) “Feret’s diameter is the mean leof the distance between opposite sidesof the apparent out1 f the particle parallel to direction and irrespective of the orientation of each particle There are different definitions for diameters. oisson. some of which havealready been ~ i s c u s s eThey are repeated in Table 5. 5. and binary. The normal frequency function. lognormal. and breadth. The latter two are not too significant in d i s t r i ~ ~ t i o ofs multin particulates.
. tions not yet touched on. the Martin’s and the Feret’s diameters are two others.
The most common distribution^ of numbers are normal. some of which also apply to crystalline samples. there are one several ways. as well as bimodal distributions are. For amorphous materials. The diameter of a circle with the same area as the (microscopically d e t e ~ i n e d ) crosssection of the particle is one.For crystalline mate~als.1 for convenience. in’s diameter is along a line that dissects the rticle in two equal areas. These are shown in Fig. and include some defini~. f ( a ) is given by
Martin
Area 1 = Area 2
~chematic showing the definitions of a artin and
a Feret diameter.1. length. but the former. uses height.
Diameter Definitionsa
Cna/Cn = a.
Often distributions are lognormal. this is defined by Eq.” bThis notation has been used in this text to distinguish between two diameters.4).
where d represents some of the aforementioned diameters. The probability of a particle having a diameter smaller than a* is then given by
he lognormal distribution is defined in ec. ence.: arithmetic mean diameter ~ n a ~ / = asu:a ~ ~ n surface volume mean diameter Cna4/Cna3 = awnz sum: weight mean diameter’ = = volume mean diameter’ = surface mean diameter
“The symbol a denotes size or “diameter. and cr is the standard deviation of the population.3.
where~denotes “function of. The “weight mean diameter” is. 5.Ink]} = . in that. the logarithms of which are between ln[a] and ln[a] aln[a]. The tributional form is then given by
resent at ion in dis
dlf(a)] is the number fraction of particles with diameters a. The normal distrib~tion function is the integral of this function. of i e The mean diameter is denoted dg (the g e o ~ e t ~~ c e ~ iaa ~~ ~ e r ) .” but to avoid confusion that convention will not be followed in this text. aavgis ~verage “diameter” (or size). at times. one plots the logarit~ms the sizes. The mast common distribution encountered in multiparticulate solids is the lognormal distribution. The Weibull istribution has the following form: ln{. where the symbol a is the differentia^ notation (used to distinguish it from the diameter notation of a).” a is “diameter” (or size) . referred to as “the volume mean diameter. (5.ln[a]
+C
( 5* 3)
where J is cumulative frequency. The number average of such a population is given by
+
. instead of plotting sizes (diameters) on the xaxis of a frequency plot or a distribution plot.
7)
that when a = 0.11)
The term under the exponent sign may be rewritten 0. however.u2]}du
(5.10)
Inserting these equations into Eq.15) may be r e ~ r i t t e ~ ln[a.(2/2/2) l n [ ~ ] } ~ ln2CY The substitutio~
+
(5.15)
where usehas been made of the gamma function in the evaluation of the integral.5 ln2a a.ln[ag1)/21/2 a] In
SO
(5.~ 2 ) d ~
= (ag/2/7r)(0. = [y/2 ln[al/ ln[a]y/2n]
agexp( 4'2~n[cr]u . The relations correlate the ea^ ~ i a m e t e ofs number d~~tributions those ~ with of weight distributions.16) has been repeated for convenience].6) now gives a. (5.The following substitution is now made:
u = (ln[a] . u = DO. (5.9) It is noted that
(5.14)
exp(0. (5. other will not. The re~aining equations are shown in Table 5.u2 = {u .16)
and is the first of the HatchChoate equations. but the mean diameter calculated on the assumption the of l o ~ n o ~ a l iwould have the value stated in the table.5 ln2a)e x p ( .2 [Eq. Eq. one of the distributions may be truly lognormal.] = 0.12)
is introduced.) ln2 =
J00
(5.
+ In
(5.5 a)2(2/~/2) exp(0.11) gives
00
(5. (5.5 ln2a + In a. a~cordin~ly. This may be rearranged to ln[a/agl= u[2'I2~ l a)] n (5 8)
or (5. ty
.5 y/2 ln[a]u. noting that
dm = du
and this inserted in Eq.
056 0.21) (5.2 In2a" ~n[a.1).] 2. A measuring device (e.I In2cro .l=1n[a.529 0. Count 14 32 43 44 45 40 18 9 6 Cumulative
(Pm)
215 24 1 267 293 3 19 345 37 1 397 423
f
0. a very small sample is taken from the population and a slide is prepared (usually a very dilute suspension in oil).]= 1n[ag1 1.The HatchChoate Relations" Relation ln[aJ = ln[a.. An example of this from microscopy is shown in Table 5.976
Normal 2value"
I .360 0.3.5 ln2 a" ln[a.] In2a ln[a.18) (5. The leastsquares equation is Z = 6.g. a set of similar data (40/50 mesh) are plotted in this fashion..115 1.] .550 1.1 or list of symbols at the end of the chapter.867 0.179 0.
+ + + +
(5. but it is more advantageous to plot the Zvalue (obtained from the ~ u m u l a t i v ~ frequency) as a function of particle size to see if the distribution is normal. The frequencies may be presented in histogram form.354 0.5 In2a ln[a.5 ln[as. The data seem to be normally distributed.550 1.] = ln[a.]= 1n[41. is In microscopy.980
aObtained from a normal error table.23) (5.939 0. The equation for this type of distributio~ Eq. a hemocytometer) allows the viewer to count the number of particles in certain particle ranges. 5.036 0. For diameter definitions consult Table 5.2.19) (5.183 0.159 0.590 0.031355b
bavg
Data Generated for 1 0 0 ~ 2 Mesh Salicylic Acid ~0 Range (Pm) 202 228 254 280 306 332 358 384 410 Frequency.072 0.5 In2a" ~n[a.
. distributions are often normal.171 0.905 0.24)
aSuperscript w implies distribution by weight.] = ln[ar] .20) (5.175 0.5 ln2a" Equation no.17) (5.
For narrow particle size.127 0.] + 0.] = ln[aF] . The Zvalues are found from normal error curve tables.5 In2a ln[ag] = ln[aF] 2.0.5 ln2 a ~n[a.8862
+ 0. (5.024
f
0.2.075 0. and lack of superscript implies distribution by number.056 0.22) (5. In Fig.16) (5.l== lnia.708 0.
1999.031355 = 32 pm (5. The mean (2= 0) is at ~. 5.88~2/0. In type I isotherms v ap oaches an e ~ u i l i b r i u ~ increasr of with ing pressure. the adsorbed volume ofgas v is plotted as a function of ~ r e s s ~P e the gas.)
where b is the breadth of the particle.031355220 pm = the standard deviation is 1/0. TI. Three of the conventionally accepted ty of (type I. norfor the isotherm to be dis~ussed next.
A~ount Absorbec
Type II
Type I I I
Type IV
Pressure
Four of the five conventional types of gas adsorption isotherms. n the isotherms.3.25) (5. whereas this is nottruefor either 11. and 111) are shown in Fig.24)
It wiltbe seen in the following that gas adsorption is employed extensi~elyin the me~surem~nt surface areas.300
Normalized presentation of size d i st ri b~t i o~s a 40/50 mesh cut of oxalic acid in ata from Dali and Carstensen.
.
oc asionally. At equilibrium the two rates will equal one rate another.8)P
(5.C)
And the relation between C' and m is often givenby equation.is the d e s o ~ t i o n constant.If a solid.8). which is proportional to the activity a of the gas.29)
where k is the adsorption rate constant and P is the pressure tion ofgaswill occur with a rate /3.3~~
where k. whichis pro~ortional to the am0 red by the gas. This is shown schematically in Fig. and proportionally to the fraction (1 . 5.
e+ = kB
k8 = k+(l . In logarithmic terms this becomes: The equation is empirical. It is assumed that 1. adso m a solution). 1918).4.27)
= qc'/"
where q and n are constants. and if this liquid before the addition contains C' material per milliliter. There is no interaction betweensites.26) the socalled Freun~lich (5. and the value of m does not approach an asymptotic value.
yM
(5. This equation willbededucedin the following for gas sorption. that is. The surface of the solid is smooth.
# + B
= k+(1 .then part of this will adsorb ontothe surface of the solid.8)P
(5. but the arguments hold equally well for solute adsorption ( e g . be filter material may also be adsorbed).
As ~entioned. surface areas (but not particle size distributions) may be o ~ t a i n e from gas adsorption or. equilibration. 2. Type isotherms are usually explained by means of Langmuir's equation (Langmuir9 1916. 3. and is assayed and now contains C m The adsorbed amount m is obtained as
m = V(C' . of volume Vs mL9is suspended ina liquid (of volume V mL) in which the solid is virtually insoluble. which is not already covered with gas. All the sites are identi~al. so that hen using the terminology for the equilibrium constant K
. the supernatant is separated by centrifugation (not filtration. by adsorption of solutes from liquids in which the solid isinsoluble. as molecules willadsorb onto the surface of the solid with a rate /3+.
32) b e c o ~ e s
(5.33) he amount of gasor solute M . given by
a = PIP0
(5.35) where a is the gas activity. Taking inverses gives (5.00 0.
=kJk.05
0. 1996b).32)
(5.34) where 4 is a proportionality constant.5 the asymptote is estimated at 0.
q. (5.Fraction Adsorbed
Adsorption Proporttonal
to
/
Fractton Not Covered and t o Vapor Pressure
Desorptfon Proportional t o Fraction Covered
Schematic o f Langmuir adsorption. which isadsorbed for each square centimeter or each gram of adsorbate would be ~ r o ~ ~ r t i oto a l surface fraction covered.0
Curve characteristic of a Langmuir isotherm.36)
In Figure 5.
. The data in Fig.28) to
0.5 are plotted according to Eq. if the crosssectional area of the gas or solute is e~nements of the asymptote calculation can be made statistically .4 0.0
0.2
0.8 Activity fP/Po)
1. (5.6 0.151 and the surface area of the solid can be estimated from this. 5. so that n the (5.
5. the area of the firstsorbed layer.4 plotted according to Eq. 5. therefore. It is assumed that the first layer may not be complete before the second and the third layer may form before both of the underlayers are complete situation depicted isat agiven pressure p . E is the energy required for the adsorption of the first~). and the rate constant is kl = bl e x ~ ( E / ~Here.38)
. ~ayznotbe obtaine~ p~ottiyz~ inve~ses by stby the s ~ ~ a$ . hence.y
. and use of nonlinear programs are advocated (Carstensen. isalso. holds at equilib~umfor the first alsop = blsl exp(E/
(5. in general.7 is.1. The volume adsorbed in Fig. The s u b s c ~ ‘‘I” here denotes the first layer that is sorbed. one part (s2) has two layers.~ . The surface area of the solid is A one part (so) is not covered. Inverse plots. Emmett. t also assumes no nearestneighbor interactions between sorbed molecules.7). the T modelassumes that multiple layers may for 5. and so on. It is also proporis ~t tional to the gas pressure p .000
+
14
12
8
“
0
2
4
1 /a
6
10
12
Data from Fig.
The Langmuir adsorption equationis based on theassumption that all surfaces are smooth. 7 6 9 2 3 ~R*2 . O O O O 0 . and 0). The molecules adsorbed take up a volume of vo/m2of layer. The assumption made in the derivation of the Lan~muir isotherm is that only one (mo ) layer is allowed. ~ e s 3
Type I1 isotherms are usually explainedby the derivationof Br~nauer.
so
+
SlVO
+ 2s2vo + ?s3vo +
* *

~
~
~
The rate with which adsor~tion occurs on the uncovered surface (to create the first layer of adsorption) is proportional to the uncovered area so. he following equation. It is assumed in the following that yz layers may form. and that all the particle sites are equally energetic. therefore. and the rate consta~t denoted a l . Aside from the other assumptio ’ the Langrnuir model. The d e s o ~ t i o n rate is ~roportional to sl. One part (sl) has one layer. (5. The rate. 1998).35).
a2slp = b2s2exp(EL/RT) T treatment now makes the following assumption: e value of EL is the enthalpy of liquefa~tion (condensation) of the gas on the surface.
x = PI~O
. (5. and the rate constant is tional to the surface area of the second bz exp(E2/RT). he equation for this situation is. This value applies to all s u b s e ~ layers.39) may now be written for the third layer. The ratio a j / b jis constant and simply denoted a/b for the first and higher layers (i. (5. The term
Y = P(al/W exP(E/RT) is now introduced into Eq.40).
The second layer is formed by molecules sorbing on an area of sl(m2) of the first layer. 5. so that the sorption rate is a2s1ptand the desorption rate is now proporlayer (s2). The next approximation made is:
6. is po = (a/b) e in exp(EL/RT) so that (5.43)
= YSO
The expression for the ~lausius~lapeyron ~ u a t i o ~this notation. in analogy with Eq. i # 1).40)
and so on.e..44) where po is the equilibrium pressure for a bulk mass of the adsorbent at the temperature T of the experiment.42) (5.39). ~ ~e~ An equation similar to Eq. (5.39)
(5.
a 3 s s = b3s3 exp(EL/RT)
(5.Schematic for the assumptional mechanism I n BET isotherms. which becomes
s 1
(5.41) (5.
. Eq.
+
3CSoX
3
+ .~ ~/~)}
~
)
/
~
~(5. = cx/[( 1 . (5.52)
The aim is to find the value of the volume v. the area of the solid A may be calculated.x + cx)]
Introducing Eq.x I&sox 2
= csovo Xix.55)
(5.~ = ~~ o c x ~ ~ sO ( ~+ ~ E+ x~ f ~ ~ ~ 2 ~ s o + = o l c i (5. 5.57)
.50) it follows that
qs. (5. because it may be converted to the number of molecules that are adsorbed and.54) The geometric series Ex' = x/( 1 .
(5.ncsoxn}
(5.s 2
= xs1
In general
si = sp"x
so that
One now denotes
c = y/x
so that
c = { [ ~ 1 / ~ 1 1 / ( exp{(E .7 the total volume v.54) will give
v/v.51) and (5.x). (5.x)2 so that introducin~ these terms into the summatio~s in Eq.47) and (5. knowing their effective cross section. .53).55) now gives
(5. (5.ns.49) ~
We may hence combine
s1 = csox
qs.
(5. of an adsorbed m o ~ o ~ a y e r .x)]ax} = x/(l .50)
The total area (see Fig. (5. It is also apparent that Eix' = x( aXxz/laJc]= x{a[x/(l . 5.} = vo{cs.54) is obtained. of adsorbant is
v = vo{sl + 2s2 + 3s3 + .44) into Eq. From the beginning and Eq.x)( 1 . (5.7) is given by
A = s o + s i + ~ ~ + ~ ~ ~ ~ +. (5.51) i
ig..
@/po)) versus @/po) should yield a ht line. and they account for type ption isotherms when c is not too large. 1978).2
0. and that this latter was c o n f i ~ e d means of lowangle Xray scattering. in the case of microcrystalline cellulose) ( arshall et al. . sible to calculate the number of molecules N in a monolayer.8
0 23. and to dissolution rates.
hen the second column is plotted versus the first a typical type I1 isotherm results (Fig. ern2.
~
~
a 5. for dry solids the value obtained with nitrogen is a reliable measure of the actual surface area. so that it is simply a matter of plottingp.l]/c (5.011 0. (5.021 0.18 98. (1959. 1961) later the solid measured would be 16 x N.. For surface area mea~urements.77
0 0. nitrogen is the most frequently used gas (krypton is also employed). 5.. and quotes that for toberrnorite the surface areas ranged from 20 to 90% of those obtained by water adsorption.81 32. moisture isotherms are often associated with ater bonding to “internal” s (e.
Nitrogen Adsorption onto a Solid Sample
0 0.g. 0 = 1 atm. 1974.041
.4 0. If a nitrogen isotherm is carried out at liquid nitrogen tempe~ature.57). then the slope/intercept ratio will be ~lope/intercept= [c . so that the area trogen has a projected adsorbed area of 16 A2 = 16 x runauer et al.ter
so that plotting the parameter @/po)/{v(l . and for large values of c.6
0. rather t h a n p / p ~ . the dry surface area may not be all that meaningful at times. The value of surface area measurements in pharmaceutics is its relation directly ilability (because often this increases with surface area of the drug). ollenbeck et al.52 49. account for If plotting is carried out according to Eq.4 for a 10g sample of a solid.57) are the BET equations. noted that nitrogen adsorption often gives low results. 1975.56) and (5.. thus.031 0. calculate the surface area and the specific surface area.8). The gas used is nitrogen at liquid nitrogen temperature.58)
v is the amount of gas adsorbed converted to standard temperature and presrom the isotherm it is possible to calculate the value of v (and c) and from v .1 p ~
~
e
iven the data in Table 5. by wever. Equations (5. Because this latter is associated with wetted surfaces.
and the slope is 0.57) and shown in Fig. Table 5.4 treated by Eq. (5.59)
that is.Ol/(l. It is seen that the intercept is 5 x (C . 5.l)/c = 5
X
104/0.05 = 0.60)
The slope is
(C . = 20 mL (STP)
(5.c) O.
The data in Table 5.05.8
1.1
P (atm)
Data from Table 5.~)/(v.61)
so that
v.6
0.99 = 1.62)
4.
.01
(5. (5. so that accord in^ to Eq.58)
5.05
(5.01
(5.00 0.2
'
I
I
1
1
0'4
0.9998e2x
Y
.
c = 110.Olv.9.000

0.4.0083e4
+
R 2 1.o
p(atm)
Data in.4 are now treated by way of Eq. (5.47).)
= 0.
63)
s adso~tion measures both external (real surface) area and internal (pore If the surface area A of a sample is divided by its mass ee ~ rA. At times the y o l u ~ e t ~spec$ic s u ~ ~ ar c ~ ~ .36 x lo2' molecules at 16 x 1@" cm2.
P2
Isotherms at three different t e ~ p e r a t ~ r eTI < s.36 x 102']f16 x
cm2 = 85 x IO4 cm2 = 85 m2
(5.64)
where @ is the fraction covered. (5. ic a ing the surface area of a sample by its real volume.
. If Eq. hence..65) where /? is a constant. (1995)havemodified a thermal activity monitor to allow measurements of eat of adsorption directly. strictly speaking.
dsorption is associated with an enthalpy of adsorption.4 = 0. udipeddi et al. as stated in the foregoing. he isosteric heat of adsorption is obtained in the manner shown in The isosteric diffe~ential heat of adsorption q (Jacobs and
q == ~ ~ 2 { ( ~ l n ~ ) / ~ ~ } ~
(5.893 x lW3 mol = 5.his corresponds to 20 x lO""22.
T2
<
T3. but may also be applied to the lowpressure region of a curve.
. The equation. In so doing.64) is integrated it becomes (5. the surface area of the solid sample is A
==
f5. was derived for a muir isotherm only. ~results. they can also construct the entire adsorption isotherm.
( distinguished from bed porosity.8 are treated at different levels of coverage). the G accounts for an intermediate state between the first. The heat of adsorption decreases as a function ofcoverage to a constant value (close to the heat of condensation of the adsorbate). It has tacitly been assumed that all sites are equally ener e manner in which this maybe investigated is through the isosteric heat of orption. If this is plotted versus temperature (ie. and all the others are unaffected by the solid. it becomes apparent at times. in is that of active sites. which will be covered in later chapters. below its critical temperature. if the data in Fig.” then the considerations alluded to in the foregoing hold true.) A liquid condensed in a pore of radius r will have a lower vapor pressure P” than that of the bulk liquid. and the relation is given by the Kelvin equation: ln[P”/Po]= e x p [ ( . first oyd (1942) and Jura and Harkins (1943). It is noted that anexternal vapor pressure has to exceed P” before condensation of the “adsorbent” can begin. has demonstrated this directly and states:
It should be noted that in real systems the heat ofinteraction of the adsorbatewith the solid surface isnot constant as assumed by the BET or its analogous model. Po. Guggenheim (1966). as the pressure is increased in the lowpressure range. 5. but most parti~ulate solids exhibit some degree of p~rticZe ~orosity.66)
where IV is the molar volume of the liquid. that there is a very distinct difference between the heat at low coverage and that at higher coverage)* In some isotherms this is direct1 demonstrable. might also be attributable to anotherassumption. the notion thatthe adsorbed layer is associated with one energy of adsorption. Capillary condensation will
. 5. rather severe ass~mptions the two models presented so far. for instance) will result as shown by 0 * in Fig..
A fair amount of adsorption work is performedby the heat of immersio~. dealing with moisture is di (1996).haveproposed and f o r ~ u l a t e da model. 11 scribed quantitatively by a dual distribution of the adsorptive energie~. ~ rou~ If all the pores have the same size and are evacuated fully.’~ The differencein heat of adsorption at different degrees of coverage. and the solid then exposed to vapor of a gas. namely. and tightly bound layer and the bulk layer. which is associated with higher pressures. a conventional isotherm (type I or 11.There are several. only by the heat of condensation. and this pressure is often referred to as the b r ~ ~ ~ t h pressure.
If a solid is “all solid. et (1961) explain that “the experimental isotherm shown in [Fig. This model will be discussed in further detail in Chapter 8.2 ~ ~ c o s ~ / ~ ~ r ) ]
(5. then.1 la. Ander oor (1968). ~ d s o r p t i o n isotherms of ar on cadmium been bromide reported Olivier by (1960) and are typically of type T (see V al. y is the interfacial tension between liquid and solid and 0 is the contact angle.
and the situation will be as shown in y hysteresis looping (C‘C) during the desorption. 5. mercury intrusion permeametry is the method of choice. and filling this to a given mark. than it really is. allows calculation of the apparent volume of the sample.66)] still applies. If there are large “pockets” in a solid of volumeV* (socalled inkwell porosity). there will be a different breakthrough for each size pore.. the very smallest of pores (e.Am t. however.13). may be obtained from knowled~e the true of ~ensity and the apparent density. ~ s u a l l yporosimeters have a maximum of 30. that gas permeametry will account for complete surfaces (i. (5. however. (Data by from Defay and Prigogine. For actual pore size distribution.01 pm) will not be accounted for. less than 0.000 psi. It is often the e ~ t e surface area that is of ~ ~ ~ Z most importance. 1966. then. it is possible to obtain the pore slze distribution.. The Kelvin equation [see Eq. the higher the pressure needed to obtain intrusion (Fig.)
commence at point A. but can go as high as . above 90”) so that an external pressure is required to intrude the merc~ry into the pore.12) The placing of a powder sample in a cuvette. I . It is seen. Adsorbed
L
0
P*
Pressure
(4
Isotherm for vapor adsorption an ideal (A) and a real (B) porous body. so that the smaller the pore radius r. 5. ence.e.. as E is given by p*
. and these are connected to the surface by smaller pores of radius r. of known volume. and this will continue (at the same pressurethe breakthrough pressure) until the pores are filled.g. The total‘ porosity. the distribution will fallaciously appear as V* larger at radius r. (Fig. the surfaces of the pores as well). so that the the apparent density p’ can be calculated. hen a distribution of pore sizes occur. and after this the adsorption (AB) will once again be conventional surface adsorption. the difference in between two readings will give the volume u intruded at a given pressure P. ore size distributions are usually elucidated byway of mercury intrusion etry. For this. By now increasin~ pressure of the mercury systematically and measuring the volume 660utside” the cuvette. the contact angle is about 135” (Le. Adsorbed
C
Amt. and insuch cases thearea as given by gas adsorption is irrelevant.
r(3)
r(3) Lower Pressure. E.
(5. will be smaller than E . ~ p r ~ s e n t r ing a certain pore radius rl. define the surface area A. as Ap = cwinr..68)
U (Volume)
U (Volume)
Schematic of mercury porosimetry trace. This unmeasured porosity represents small pores as well as defects (which are obviously not pores). It is apparent that the data will represent themselves as volumes ai.67) The porosity. and the difference E .
(~. The problem of inkwell pores has already been touched on.. and one may.
. Hg Intrusion o f Larger Pores
Schematic representing pores and defects and the principle of mercury intrusion: r(4) < r(3) > r(2) > r(1).E. measured at the highest pressure. therefore. can be converted to a radius that will represent the "average" of smaller pores. of the particle using volume fractions wi.
that is. therefore. and porosity E . ) [ ~ / (. = d:v
(5. maybe obtained by the consideration that the work W exerted by intruding a volume of du at a pressure of P is
= Pdu
(5. so that (5. then the velocity in the pores is larger by a factor of E . (5./mL of solid. A. and more correctly. also equal to the surface dA. if the viscosity of the liquid is q. times the interfa~ial tension.72)
where the inte~ral represents the (crosshatched) area under the curve in he total volume V of the sample is known from its weight and true density.is is obviously based on all the pores being cylinders (the socalled bundle of inders model) and more directly.76) If liquid approaches the bed with a velocity v*. ~ccordingto oisseuille’s law for a liquid flowing under a pressure head of AP. a surfacevolume mean diameter.73)
External surface areas may be obtained by gas permeametry.70) gives
~~~
= Pdu/y[cos8]
(5.however.69) and (5..
V*
= V’/E
(5. In a powder bed with speci~c surface area A.69)
It is. the socalled hydraulic radius r* is given by r* = ( 1 / 2 A .70) Equating Eqs. may formally be calculated as
V/6A. times the contact angle.
his may be rearranged to ~ ~= 8qvf/{xu4} / h (5. a volume of Y will pass through a capillary of radius r and length h in a time element t.E ) ] 1 (5.75) where v is the velocity of flow through the capillary.71)
rated this becomes
Ap = ~
l / ~ ~ C O S 0 ~ }
PJSllfl
(5.77)
oducin~this e~pression and the expression for the hydraulic radius into the seuille e~uation now gives
. dzl. of the pores.
ecause gas adsorption measures both externaZ and s of a solid. all quantities excep this latter can be calculated.E
) ~
(~.~to c t o r s e ~ particulate solids. the Fisher subsieve sizer). this will larger be be may attribute^ on an overall scale/surface gosi sit^.v. so that the bed volume is the hei allows calculation of E . which is employed in of powder that corresponds to 1 cm3 o solid (obtained by taki f material weighing l / p where p is the particle density of the soli crosssection of 1 cm2 is used.79)
and the weight mean diameter will be
Such calculations are often carried out forsieve analysis.
*
If a particle size distrib~tion~(a) a function of size (a) is known. that is diameter or projection diameter) to the particle volume v.g. ir (or another gasof known viscos through the bed at measured vel and the pressure diffe and efflux streams are measured.av
sf a ~articZe.
s = a.82)
Similarly a shape factor may be defined that converts the surface area s o a ~ a ~ t i c Z e f to its surface mean a. Such a surface area is denoted the geo~etric s~rface (not to be confused with area the geometric mean).
3 v = a. then it is possible as to calculate the surface area.
(5..
A com~only used shape factor converts the volume volumetric mean. 1937. ass~ming that the particles are spherical and smooth. and because the latter are not necessarily available for dissolution.h P / h = 32v’q(l .a 2
(5.”(see Table 5. and this converts a “size” a (however. possible to assign v o Z ~ ~ e t r i c ~ ~ h a ~a. as will be discussedlater.83)
.1). a.78)
This is the Car ny equation (Carman. If the external surface area A* is known.
= A*/A. e. to its size. For a weight mean diametertype calculation the area will be
A. permeametry is often a better means areas that have meaning in dissolution testing.. 193 ry (e. = ~ ~ w ~ a 2
(5..
It is.
77) gives (5. so that be relied on. A. A similar approach will show that (5. a.18)] from Table 1 into Eq. which may be converted to a volume E . Another approach is tedious accounting for N by counting methods.If a particle p~pulation fairly monodisperse. so the specific surface area A. surface of the sample A and its true volume V by the formula:
(5. are known so that this latter can be calculated. and assuming the t and (5. for each interval.86) ecause N . (5. is then 1 g of the cut will contain N particles.&
3 A. such as for a narrow mesh cut. p.
(5.87) where now all quantities except a. a. and if the particle density is p. However. that by introso the ap~ropr~ate Hatch~hoate relations [see Eqs.85)
It is a s s u ~ e d that the narrow mesh cut is lognormally distrib~ted. a. = 1 a ~
(5 34)
The surface area of the particle is s = asa2. y formally be calculated from microscopy. prerequisite for this is that the distribution be narrow.90)
. the height h is the same as the breadth. then for the 1 g sample
Nu = N ~ . o roften calculated by the It)is .The sum ~ a j a may be calculated.89)
or a particle the relation between a.. this will be covered and A shape factor (simply denoted the s ~ ~ ~ e ~ a c .. may be calculated.. the ass~mption made that the dimenis sion that is hidden in microscopy. The total number of partirepresents a weight. ' ecause the count consists of.88) from which a. = a..17) and (5.a 2 /a. (5. is this ure divided by the mass of the particle a.. listing the appropriate ni particles of ~ b ~ zes l e ~ ~ai h breadth bj. ifthe total number may be obtained experimentally.and I' is obviously given by:
(5.a. can be calculated. may be o b t a i ~ indire~tly} more l o ~ i ~ a l l y } f r o m e~ but d~ssolution data. and ag are measurable.
To do so it wouldbenecessary. however.91)]. The topologic dimension is still 1. An intuitive understanding of I) is de~onstrated Fig. (5.
~ c ~ e ~ a t i c principle of ~ e a s u r e ~ e oft fractal di~ensions~ showing n
. There are many applications of the concept. h. because ~nterest in it started with the work of Richardson (1961). but the Euclidean dimension is 2. by image analysis.first to describe what it is. to obtain a crosssectional representation of the particle and. 5. ~ i t h o udelving into the intricacies of this approach.15.91)
s a constant and 1) is referred to as the fractal dimension [and as demonEq. the space is to a great extent filled up by the line [it is not possible to entirely fill up space witha line]. and then to showhow it can beapplied to pha~aceutical problems. the dimension is 1.91) emergesas thenegative of the slope of a logarithmic L versus gplots are known as Richardson plots. for instance. coast line plots. because [Xniu:]2/3 Ze #
The introduction of fractal geometry as a mathematical tool is attributableto ~ a n d e l b r o t(1983). of a perimeter depends on. this were reduced to q then the periphery measured would be longer than if one had used a largerscaling length. who had the task of measurin~ the length of the coastline of In general it can be shown that the length I. For the wiggly lineon the bottom of the figure. or walking yardstick plots. and one could visualize this as having a dimension of 2. because of their origin in geographic and topological science. then the length obtained would depend on the “leng measuring stick. For the straightin line on top of the figure. the scaling length.I))logl[q]
+
(5.This.” the scaling length. and the intent here is. q. It is now possible to define the surface irregularity of a particle by the fractal dimension D [definedin Eq. This is the origin of the practical application of fractal dimensions. from this. (5. the general philosophy of t f. by the relation: log[L] = ”(1 . is not correct for values of the shape factor obtained by dividing the area of a s u ~ ~ with the twothirds power of its volume. For the line in the middle the fractal dimension could be visualized as being between 1 and 2. the length of a contour. if. such as depicted in ured. for instance.
5.16 is taken from the work by Ramadan and The slope is
H=SD
( 5 92)
the fractal i ~ c r e ~ ~ ~ is .D=1
D=ca. there are two distinct line segments.Z
Fractal dimensions of a contour. 5. The curves and concepts are approximate and are shown for definition purposes only.g. In the upper graph. assumedly.
0
1
2
3
4
log steplength
L as a function 1 (step length) for natural microspheres. 1990. S988). The point where a curve becomes a planefilling curve (e. 6 imply that there are two surface pop~lations (perimet~r lengths La and d the same group of in~estigators(Thibert et al. 0th the photomicrographs h spheres with “pimples”) published by the authors and the upper curves in ig. a Peano curve) is complex and beyond this writing. indicating two types of surface m o ~ h o l o g yWith lactose.)
. and Fig.l.a measure of the surface roughness.
obtain the fractal dimension. The curve is not which t linear (although the authors have treated it as such). fine structure of surface pores has a fractal . as exemplified in Again.. This indeed has been done. indicating that there are two selfsimilarity populations. in the pharmaceutical literature. ~imension different from that of the nonporous part of the surface. (Data from Ramadan and Tawashi. indeed. reported later such a behavior in the fractal analysis of lactose granules. This is done for graphic clarity.. The points in the upper graph are those in the lower graph + 0. the points are shown as bimodal. The lower curve shows all the data plotted in simple linear regression.
This method is better. then more of the roughness will manifest itself than if a larger molecule (see ere. (5. ible to probe surfaces in a more convenient manner^ that is. again. by gas re the property values are averages over the entire surface.18a). Ifa small adsorbent molecule is employed (seeFig.91) gives
Coverage of an irregular surface by differentsized adsorbant molecules. the crosssectional area of the molecule is a function its diameter h. then the sorbed area 6 is simp~y per mo1ecule. If the molecule has a circular crosssection ofarea b and by square arrange~ent. of a problem (although sample sizesare still smallin such work. 5.. 1988.4. h2
/3= h2
(~.)
rojection of crosssectional images can be misleading in that the observation can be a function of the orientation of the solid particle.88
0
1
2
3
4
log [Steplength]
The fractal character of lactose granulations.18. they are not simply one particle). 5.
. squared. and sampling (as in any other the type of microscopy) represents a problem.~3) (5'94)
or
h = (8)"2
mploying an approach similar to Eq. ractal approaches to surface sampling by gas analysis are based on the printlined inFig. hence. It is timeconsuming. the measured surface should be the larger the smaller t or adsorption. (Data from Thibert et al. the more sp~erical particle is.
) the dissolution reactive dimension (D.
Fractal plot of carbo? black: Amount o f adsorba tion of crosssectional area (A2) o f adsorbing molecule. Figure 5.16
n the foregoing example.) reported by Farin and Avnir (1987).)
. 199’7 a. the ap~licable equation is: log[A] == (D. Fini et al.100)
where A is the surface area obtained by gas (nitrogen) adsorption anddis the particle diameter.96)
Figure 5. where r is the number of molecules in a monolayer. (1996 a.
.3) log[d]
+ constant
(5. and reported on the fractal dimensions of and as the surface (D.b.] = D l n [ ( ~ > ” ~ Q = . (1983).c) reported that the fractal dimensions often depend on the e of crystalli~ation.( ~ / 2 )ln(P) + ]
+
(5. packing can be treated similarly.20 shows the BET (nitrogen) surface area of differentsized fractions of Aerosil (colloidal silica).0581
so that
L)
= 2. The more general case. where the molecular packing is other than square.to do nitrogen adsorption on various mesh fractions of the solid. more convenient (although still not practical from a quality control point of view). now
a =Jh 2
(5.. 1983. ~ p e r i ~ e n tsuch as those described. then the fractal equation becomes (5.~n[n.99) log[v] ( 4 n=
+ 1)
where v is the molar volume of the sorbed molecule.
(mmolis) in monolayer as a ata from et ala.95)
z . When this approach is used.19 is constructed from data published by Avnir et al. It is seen that the slope is L)/2 = 1. and it is s. the molecules are fairly spherical.b) studied the physical properties of salts of ursodeosycholic acid.These authors (Fini et al. and if an adsorflat on a surface. are still rather cumbersome.
(c)“diameter” (or size).4
y = 3. Hialeah.6
BET surface area.)
~lectronic counters and laser counters are the methods of choice in many presentday situations. = surface area of a sample divided by its real volume A. containing 0. and they employed jetmilled and bantammilled material and “spiked” it with larger particles to make the distribution lognormal by weight.8
1. (b) br bi = pre~xponential factor for adsorption of the ith layer C” = concentration before adsorption C = concentration after adsorption
D
d
= =
fractal dimension general size term for diameters
. adjusted to pH of 5.6 1
2.995
2. For instance. A (m2/g) as a function of particle diameter. = specific surface area Ap = surface area of the particle as A. (Data from Avnir et al. ~ = geometric surface area ~ A.005% of Tween 80. arithmetic mean diameter aavg= average“diameter” b = (a) rate constant of desorption (BET equation). (b) length of a particle. As electrolyte they use a 2% w ~ v solution of maleic acid.3996 1. = X:na/Cn. 1983. after which they saturate it with drug.. (d) rate constant of adsorption ( a. = volumetric specific surface area (per cm3 of solid) a = (a) activity (= PIPo).4.0722X R”2

0. Pfizer). = X Z W . d (nm) of various Aerosils.2. Their investigation dealt with a lognormal particle size distribution.
A = surface area of a sample A” = external surface area of porous solid (in a bed) A.0
1. FL) to measure the particle size distribution of an experimental drug (CP 118 954. Zhang and Johnson (1997)used a Coulter Counter (Coulter Electronics.
ln[j]} = . = (a) molecular volume (in elvin equation). (b) breakthrough pressure in intrusion porosimetry. the logarithms of which are between ln[a] and ln[a] E = enthalpy of adsorption of the first layer ( quation) E L = enthalpy of condensation (BET equation f = “function of” f(a> = (a) { T / [ O ( ~ X ) ” ~ } exp{(a .(a) pressure that is lower than the equilibrium pressure of a gas at a given temperature. (c) length of capillary. = (b) generally particle size distribution as a function of size (a) = slope of fractal dimension plot = (a) “length” of adsorbate molecule. . . ~ scaling factor. Po = equilibrium vapor pressure of a gas at a given temperature Pr(a < a”) = probability of a size a being smaller than a* = (a) Weibull distribution = In{. = length of perimeter (in fractal dimensions) c) = standard deviation for lognormal distribution r = amount of gas adsorbed on a solid w1 = (a) u .geometric mean diameter = exp( xrzi ln[di]/Gni] ~ / = surface volume mean pore size 6 ~ ~ dg = geometric mean diameter = exp f Eni ln[di]/ dwm = ~ ~ ~ = weight mean diameter a / x ~ ~ dh[f(a)] = number fraction of particles with diameters a. (b) length measuring stick (in fractal dimensions). ~ ) ~ / 2 a ~ } normal frequency function. (b) adsorbed amount (in Freundlich adsorption) N = number of particles in a population n = Freundlich exponent = gas pressure * .ln[a] C. (b) liquid volume Y” = inkwell pore volume ‘c/T/ = work (in mercury intrusion)
dg
=
d:
=
*
+
. (b) pore radius r* = hydraulic radius rl = pore radius of the ith pore s = surface area of a par ’ so = surface not covered s1 = surface covered with one adsorbent layer ( s. (c) factor in = isosteric ~ } (b) Langmuir equation.a . = rate constant of adsorption k. (b) constant (fractal dimension equation) q = (a) ~ ~ 2 { ( ~ l n ~ ) / ~ heat. (d) Freundlich prefactor R = gas constant r = (a) radius of capillary.= rate constant of desorption I. = surface covered with i layers of adsorbent T = absolute temperature t = time f. (d) height of particle i = running index = cumulative frequency in Weibull distribution = equilibrium constant in Langmuir equation k .(y/2/2)ln[rr] = integral substituent dummy variable.
Technomic publish in^. Trans Inst Chem Eng Lond 15: 150. J Am Chem Soc 68:686. Pitman Publishing. Drug Dev Ind Pharm 25:347. Carman PC (1938). Brunauer S. (b) volume intruded 2~/~ at a mercury pressure of P (a) molar volume of the sorbed molecule (in fractal dimension). Carstensen JT (1996bj. Lancaster. DallaValle JM (1943). p 39. J Am Ghem SOC 60:309. A / ( v ~ /(general) shape factor = ~) = ~x. Dali MV. Lancaster. J Phys Chem 97:3566. Microrneritics. ett PH. PA. (b) liquid velocity./[a:/~](general) shape factor for a particle = y = interfacial tension between adsorbate and solid 8 = the differential notation E = particle por~sity E. = weightof fr on of particles withsize a x = p(a/b)exp( Y =p(a/~) exP(~~/~~) Z = standard normal deviate a = j h 2 = scaling factor when adsorbate is not of square con~guration a. = a2/s = surface shape factor of a particle a. Carstensen JT (1999). pp 6373. Emmett PH. Modeling andData Treatmentinthe Pha~aceutic~l Sciences. = total porosity measured at the highest intrusion pressure S2 = A*/A.
.= rate with which gas molecules will desorb from a surface r = . ing. Pfeifer P (1983). (d) volume of adsorbent gas at a given
v' = velocity of flow v* = approach velocity of a liquid to a bed vo = volume of molecules per square meter of layer v = volume of a monomolecular layer of adsorbent gas ( . = rugocity 8 = (a) contact angle. = volumetric shape factor of a particle = a:/v #3 = square of scaling factor in three dimensions (h2 per molecule) B+ = rate with which gas molecules will adsorb onto a surface 6. . PA. J Am Chem Soc 62: 1723. Carman PC (1937). Modeling andData Treatment in the Pharmaceutical Sciences. (b) fractional coverage of a surface with adsorbed gas p' = particle apparent density p* true particle density q = viscosity a = standard deviation of a population ln(a) = standard deviation for lognormal distribution
Anderson RB (1946). (c) volume of a particle. J Soc Chern Ind 57:225. Teller E (1940).l n [ ~ z ~ ] ) / lna]. W = mass of a sample w. Teller E (1938). New York. p 28.u =
v
=
(a) substituent in integral = (ln[a] .
Royal Acad Sci Vienna Proc Class I 136:271. 1987). p 184. Fini A. Oxford. J Chem Phys 11:561. PhD dissertation. . Fini A. Jura G Harkins WD (1943). Thibert R. Holgado MA (1997b). Sixsmith D. Fazio 6 . Holgado MA (1997a). Int J Pharm 171:45. Tawashi R (1988). WH Freeman & Co. University of Wisconsin. J Phys Chem 91:5517. Rabasco AM (1996a). Rabasco AM (1996~). Butterworths. Langmuir I (1918). Hollenbeck RC. Olivier JP. Marshall I(. New York. Application of Statistical ~ a c h a n i c s Clarendon Press.
~~~b~~~~ h ~ 5 ~ t ~ ~ C
lot the data in Table 5. PhD dissertation. FernandezHervas MJ. p 319. Clarendon Press. Tawashi R (1990). Peck GE. Pudipedi M (1996). Hinchen JJ (1961). p 45. J Pharm Sci 86:1303. ( Olivier JP (1960). Drug Dev Ind Pharm 1:51. (1983). J Pharm Sci 77:724. Rabasco AM (1996b). Duddu SP. Fer~andezHervas MJ. Akbarieh M.deBoor JH (1968). FernandezHervas MJ. 2nd ed. FernandezHervas MJ. Washington.Beebe RA. Graham DP. Zhang Y. General Systems Yearbook 6:139. American Chemical Society. J Am Chem SOC 38:2221. Int J Pharm 154: 179. Carstensen JT (1995). StanleyWood N C (1972). J Pharm Pharmacol 24:138. FernandezHervas MJ. Small Particle Statistics. Ramadan MA. Holgado MA. eds. J Am Chem SOC 64:1195.
. TheDynamicalCharacterofAdsorption. J (1927). Richardson LF (1961). Sokoloski TD. The Fractal Geometry of Nature. Pudipeddi M. J Am Chem SOC 38:2221. Zisman WA. J Pharm Sci 8597 1. Calculate the appropriate mean and the standard deviation. London. J Pharm Sci 79:929. Eur J Pharm Fini A. StanleyWood N C (1974/1975). ir I(1916). Fazio 6 .Eur J Pharm Sci 4:231. Marshall I . Fazio G.3 and ascertain whether they are normal or lognormal. Kildsig DO (1978). ~ ~ g g e n h e iEA (1966). Oxford. J Pharm Sci 85:381. m Harkins WD. Sixsmith D. Fini A. DC. Herdan C (1960). In: Copeland LE. Solid Surfaces. Holgado MA. Rensselaer Polytechnic Institute. Johnson KC (1997). Ross S. olgado MA.Zettlemoyer AC. J Pharm Sci 67:1599. Boyd C E (1942).
2.6. a discussion of some of the fundamental factors affecting crystallization is presented. Crystal Growth
6. 6.5.6. because the shape of the particle may affect machinability (e.7.
6.
.12.13. The ~ q u i l i ~ r i u m
90 90 91 92 93 94 94 95 istributions After ~omogenous Nucleation 96 100 101 102 104 104 105
6. Temperature and Solubility rofiles During Thermal Recrystallization After Thermal ~ecrystallization 6.
omogeneous Nucleation
Zones 6.1.11.3. Nucleation 6.14. etastable Zones and Nucleation 6. Crystallization 6.and this. Particle Size ~istribution 6. 6. needleshaped particles may logjam when they flow through a hopper).9. The conditions which this i s carried out is of under importance in pha~aceutics. Yield and Metasta~le
6. most often..8. Nucleation and Critical Nucleus Size 6. eferences et~rogeneous Nucleation
The last step in drug substance manufacture is puri~cation.10. and themorphology and the shape may affectdissolution.4. consists of r~crystallization. Therefore.g.
That is. for instance. 8. 5% of an impurity solubility of 2 g/20 g of ethanol at 25°C. the material is produced. yvlt (1971)has shown that thewidth of the zone maybe a functionof to how high a temperaturethe solution has been heated and for how long. if. and these crystals will later transform.5 g of impurity) are dissolved in 40 nd then cooled to 25"C. then it will go into solution. stagewise. It is a common misunderstan~ingthat precipitationandrecrystallizatio~ occurs from saturated solutions. The width of this z metastable zone. to the most (or a more) stable crystal structure. thereby. and then cooled. in the foregoing example.
stwald (1899) formulated arule of stages: when a system firststarts c~stallizing. the drug isusually not pure when the overall is complete. whereas they actually occur from s ~ ~ e r s ~ ~
1s is illustrated in Fig. it will initially create the crystal structure that forms the smallest loss of free energy. 6. succinimide has a solubility of 1 g/20 g of ethanol and 1 g/4 g at 60°C. a solution at 300 rng/g is heated to 70°C (point past B). will not precipitate until point A (57°C). the heating was carried out at 90"C.1. If 300 mg of material is added to 1 g of solvent and heated to 70°C (point B). This will be dealt with futher in Chapter 8. For instance. for example. precip~tationwill not occur {in a reasonable length of time) until 57°C (A) is reached. In this case one speaks of r e ~ r e ~ ~ i t ~ ~ ~ o ~ . ecrystallizations may also be carried out by dissolving the substance in one solvent. precipitatin~out the pure chemical and (providing the solubilities allows it) keeping the impurities in solution. and adding another in which it is insoluble. it may contain.5 g of impurity will stay in solution.rug manufacture (synthesis).5 g of pure material will then precipitate out (1 g in solution) and the 0. then the zone would be
u10 20
30
40
50
60 70 80
Te~peratur~O C
If.
. If 10 g of crude material (containf pure chemical and 0.
diagonal).
In this chapter.2). AG is maximum when
d ~ G / d = p?r. and if it were kept there for 1 h. then AG from solid to solution must be zero.1)
=0
where p is chemical potential and cr interfacial tension. It is speculated that complete randomness of the solution is not ascertained until it has been kept at a temperature well above the solution temperature an given length of time.
. rather than for 5 min. argument that follows is. = 4a/p
This then is the critical nucleus size. alk about solubility. r
+ 12ar.
or
10
0
(3
10
20
0
2
4
6 8 1 0 1 2 Size (d)
Chemical potential and nucleus size. so to determine it per mole. it would then also be wider. the symbol r denotes dimension (radius. it must bedivided by the density p (to o in the chemical potential per gram) and be multiplied by the molecular weight ) to obtain the chemical potential per mole. then the nucleation will occur more readily. when
r. and that if complete rando~nessis not at hand.strictly speaking. because beyond size.2)
that is.wider. the growth of a n ~ c l e ~ s this ied by a negative AG (Fig 6. incorrect ( lassic form in ~arstensen (1980). p is potential per cubic centimerer (cm?).
(6. For a cubical nucleus of size r
AG = pr3
+ 6ar2
(6.
or would decrease. would be so long that it could not be carried out. One might invoke criteria such as that if one tested the con cent ratio^ every 24 h. but it is difficult to disprove. energy is simply assumed to be proportional to some “size. then
(6. 6. 1898). predicts that solubility is inversely proportional to the size of a particle. The size of that crystal will be such that the concentration of the supernatant will be given by the point on curve ABC which will give mass balance. where AGO refers to standard state. if the crystals grew and the equation were correct. indeed very possible. it predicts that if a multiparticulate system is placed in a liquid. were it true. then the crystals will grow (Ostwald ripening).” It is more logical to think that the real e uation would be one leading to a curve as shown in Fig. in that AG cannot be
quation (6.1 1) If. 3 may be ap~roximated a by (6. but there are problems withthe argument.until there is only one crystal left.8) ut this means that
his is known as the OstwaldFreundlich equation (Ostwald. It is possible. it is conventional to require 72 h for equilibrium to be attained.10) where (b is a constant and where (6.13)
. because the time it would take. so
If a system is taken from point A (with dimension r l ) to point sion rz). This is not reasonable. when r is large. In solubility work. that the basic equation [Eq. or more correctly. The equation. For instance there is no accounting for defect f o ~ a t i o nand the interfacial .].ut the lefthand sideof the equation is AGO+ RTln[C. (B/r) > ((b/r2).7) pre icts that the equilibrium state of a solid is an infinitely large crystal. seemingly.2)] is incomplete. (6.1 = B(l/r) (6. solubility had been reached when there was no “detectable” increase ut this is no guarantee that the concentrations over a 6month period would not increase.12)
so
(6. then the equation reduces to > WC.3.
14)
==
It is seen from this that if a solution is supersaturated to a degree of S then
C/C2(> l).15) gives (6.
odified modelchemical of
potential as a function of size.1) for a nucleus to spontaneously form in supercooled water vapor.17) ere. is given by
W = 6d2g
(6..16) Inserting Eq.
Time Required for Nucleation to Take Place Supersaturation ratio Time Infinite yr lo3 yr
0. S is the supersaturation ratio.
. the more easily a nucleus will form.&V
20
0
10
Length
ig.
I
The work required to create a nucleus of size d. (6. finite work is required to form a nucleus).1 s
101~ s
Source: Mullin (1961). ullin (1961) reports the following times (Table 6. 1n[q is 0 (i.15)
or (6.e. (6. and it shows when this is unity.14) into (6. The higher the supersaturation ratio is.
It is assumedin calculations that if the impurity is hi her than its solubility limit at the conditions of precipitation of the drug substance?then it is “removed. igure 6. and the explanation lies in an expansion of the metahomogeneous nucleation there is often a la time before c~stallization e foregoing example. and this often happens. the
Chow et al. the example of homogeneous nucleation is suspensions of amor~hous frusemide). and this maybe an important consideration? becauseonly relative small amounts of additive ( i ~ ~ u r i t y ) y affect a great number of properties of the crystallization and the crystals t firstsight it might be speculated that~cetoxyacetanilide increases the solu~ility acetaminophen. One effect is on the yield. he inclusion also affects the ability of the crystal to contain water (up to a se (as in the pr~viously cited case) the inclusion leads to lattice vaca~cies? “space” created presumably allows “room” for the water molecules. and in a suspensio~ amorpho of ort the following c~stallization profile
5
4
.4 is an exam e an asymptotic limit to the uptake (in this case.” in that owever inco~oration (doping) of solids by introducing guest ~olecules into ssible. The amorphous stateis much more energetic (more soluble) than states. the extent of s~lubility of increase (about 6% at the highest co~centration additive) does not explain the of dramatic decrease in yield.mpurities in the intermediate drug substance are usually removed by recrystallizaion. Chow reported that thoat seeding and the pretoxyace~anilide system did not start crystallizing in 2 h. ( 1 ~ 8 5 )st ed the effect of additives in the mother liquor on the outcomes of crystalliz n. but even though this is true.
6 2 ) where 6 2 is solubility.b~. was first tney (1897a.
llization. also used in early dissolution work. ordin ding to Noyes and / d t = k&4(C1 . erthoud and Valeton have shown that a reaction term must be i~cluded:
v)
2 $
0 0
50
100
150
Time (hours)
C ~ ~ s t a l ~ i ~ aofi o n 16% arnorphous furosemide (frusernide) suspension. Reference is m a film at the crystal surface exists. per
he decrease in concentration ( C ) follows a curve of the type
c . the at solution ori~inally contained 9 g of a c e t a ~ i n o ~ h e n 390 cm3 of water. an at the crystallization and dissolution rates were the same.cS (eo CJe4' = 
(6.18)
ut it should be noted that q is neither a growth nor a nucleation rate ons st ant. 1987. They also assumed that dissolution was the reverse of crystallization.The effect of ~acetoxyacetanili~e on yield of a c e t a ~ i n o p ~ e n 30°C.)
. ( t a from Doherty and York.
indeed. At a point .where C3 > Cs. the fraction rate of surface renewal.22) (1Im = (1lkd) + (1lkP) where kpis a "reaction on st ant'^ and kd is the diffusion constant. and Marc (1908).
.22) is that if the film were stagnant. who suggested that there were two processes. one a dislodging of molecules from the surface (the socalled reaction rate k. In the metastable form would approach a concentration of C = 200 mg/g if no con1 wever. A model overcoming this problem was proposed by Berthoud (1912) and Valeton (1924). The reason for the expression in Eq.24)
orph is allowed to dissolve. as shown in a previous higher solubilitynumber than a more stable form.
Crystal Surface
"".
(6. precipitation ens that in a time period (2040 time units in the figure). (6. In
~ ~= KA(C1 ~ / C3) t
(6. at a given point in time (10 time units). 1"
h
u
chernatic of concentration profile at a crystalsu~ersaturate~ solution interface. as shall be discussed shortly (6. found h to be zero at high velocities. igbie (1935) and Dankwerts (1951) suggesteda surface renewal theory where.23)
it stands to reason that the film thickness would have to depend on the speed. here. referred to in the foregoing). and the second being the dif'fusion as discussed in the foregoing.all the metastable material in steadystate dis~ontinues and concentration decreases toward the saturation con cent ratio^ C2 the of the more stable poly~orph.. then (
kd
= I)/h
(6. simply
kd = (Df)'/2
f is. This would implyan infinite growth rate at high liquid velocity.21)
where. the con~entration stays constant.
C)) equals the rate with which the stable form precipitates (kZ(C . but they introduce us to the manner in which crystallization events are translated into particle size distributi~ns the ensuing product. The crystal that was born in the first time interval will have the size:
a.
dpz/dt = 4
(6. n that have grown for T .The steadystate phenomenon happens because the rate at which a suspended metastable drug substance dissolves (kl(Cl . but the fact remains that where the concentration is fairly constant.26)
e is actually never quite accomplished.
Csteady
= (klC1 . Under such conditions. but serves is model well as an introduction into how distributions are arrived at The question that poses itself first is.2~)
0
20
40 60 Time
80
100
120
Conce~trationprecipitationtime curve. what is the rate of nucleation? We will see shortly that.C2).k2C2>/(kl .k 2 )
(6. (Data from Dofierty and York.csteady)] = ckZ(Csteady
.8) This situat~on an o~ersimplifie~ for general crystallizations. the nucleation and growth rates of the crystals are constant (Fig.)
. it is dependent on the degree ofsupersaturation. = k?
(~. that is.1 time units and so on. a. most often. 1987. in AB this is constant and it but will also be assumed that the growth of the size.~7)
Growth rates are also a function of supersatura~ion. nuclei are formed each time unit.CZ)]
(6. that the T time units are divided into intervals.2~)
that is. Csteady is given by
c"kI(c1 .28)
These are obviously severe assumptions for a general case. first. each of 1 time unit. but in the foregoing situation the degree of supersaturation (between A and wouldbe constant. follows Mac da/d~ k =
(6. and that pz . Then there will be pz nuclei that will have grown for T time units. 6. for the m o ~ e n tthat a crystallization event the range AB takes T time units. of Assume. Wewill assume.
1)3. then this
= pkn(T3 ( T . (6. (6. If the data points are taken off their Fig. tsuda (1998).t)4/T4] ut
(T.38)
so that
ln[f(> a*>]= 4 ln[a*/amax] (6.1)3 so that the total weight of the crystals will be
= pkfl(T3 ( T .37)
or by taking logarithms ln[f( > a*)] = 4 ln[(T . if simply judged from data in Table 6. (6.30)
+
+
.32)
If one considers the amount of material that has sources from times 0 to t.
+
(6.10 (plo he geometric mean is seen to be given by 373/ 1.. unlike what their article og. shown in As 6.33)
This material will have sizes at or above
a* = k(T . however. * ( T.The one born in the second time interval will have the size a1 = k(T .34)
T3dT = pkn(T .(Z3
+ 13)}
(6.97 that is.t ) 4 / T 4
(6. the sum of the series equals the integral:
T
T3dT = pknT4/4
(6.9 the distribution.I ) ~ + . could be either.t)4/4
(6.SO5 = 4.t)4/T4 = a*/aw. but Fig.2) shows it to be visually.2 to allow plotting according to either a normal or a lognormal distr~bution..2 and ated in Table 6.39) weight frequency of particles larger than a* versus the straight"1ine with slope 4.t13) .
(6.33) is equal to
Tt
(6. 6.40)
.3 1)
ecause the time unit is one. the distribution looks as shown in Table 6.35)
so that the weight fractionf(> a") of material with a particle size larger than a* is
f ( > a") = ( T .t )
q. 1 carried out).36) (6.
045 0.1.
100
80
44
s .298 5.5 9 15 20 25 45 55 70 80
2vaf ue
Mil
2.190 .996 3.
.7.880 2.2 treated as a lognormal d i s t ~ b ~ t i o n .840 0.689 3.1 3 5.Example of Lognormal Calculation Size Undersized
(4
20 30 40 50 60 70 80 90 125 150 200 220
("/I
0.1.880 .r3
20
0 0 
$
100 200 300 Particle Size ~ ~ i ~ r o n )
Graph of data from Otsuka and Natsuda (1996).912 4.401 3.600 .013 0.340 . 5 1 0 5 ~R"2 0.500 4..828 5.994

1
2
3
2
lnldl
4
5
6
The data in Table 6.382 4.248 4.2 1.670 0.4373 +
1 .094 4. .1.1.
Y 1
y'i
0
N
.394
2.845
Soztrce: Otsuka and Natsuda (1998).525 0.01 3 0.01 1 5.
6625 nowledge ofthe geometric meandiameter allows calculation with the Choate relation: ln[d.] .] = 1/1. is given by
ost often this is associated with an exponent l/n. larger for a multiparticulate.45)
where R is the gas constant and T is absolute temperature. Denoting time with the symbol (9 and the number of nuclei at any given time by N . Eq. a type of situation that will occur in thermal recrystallization) is given by
AG = RTln{C/S) = RT ln[q] (6. the rate may be expressed as dN/~(9.agM. s) is. J ..
he chemical energy AG associated with 1 mol transferring from a supersaturated solution of concentration C to a saturated solution (i.5 ln[a. therefore.0. symbolized with the letter q:
4 = C/S
(6. because (as demonstrated) the highermoment diameter is. The standard deviation is the inverse of the slope. C/S is denoted the supersaturation ratio and is.. in some form or manner. there must be a nucleus from which the c ~ s t a l ~grows. > becomes J = a(C' .S)'/" (6. not legitimate simply to compare agw with a.2) represents the energy of activation for nucleation.46)
ecause S represents the concentration in a solution in equilibrium with the solid state.42)
It is obvious that this figure is considerably higher than the value that the authors found from BET surface measurements.] = ln[ag.48)
. associated with the rate of nucleation. 144 pm =
(6. if C > S. ln[cr.. it is legitimate to state that the two methods give different results..41)
This corresponds fairly well with the "mean" arrived at by Otsuka and atsuda (1996). however.51 = 0.45) represents the energy of transfer of one mole from solution to the crystalline state.e. in the following.] from which (6. (6. The rate of cr~stallization therefore. For crystallization to occur. q. It is. so that the expression. by definition.43) (6. The nucleation rate. (6.
6.49) and (6. s that is.49)
An example of this when the ambient temperature is 23"C. Assume the solubilities at 40°C is50 and at 25°C is30.
T = To[l .
In thermal recrystallization. (6. Rather than arriving at the complex relations that would arise from this. either natural or induced. excess drug is dissolved in solvent at a higher tem~erature at which its solubility is more than its ambient solubility.49) would take the f o m :
40
20
0
2
4
Time
6
8
1 0 1 2
1 Cooling curve starting at 40°C toward ambient temperature of 23°C. an approximation approach has been taken as shown in the following example:
~ ~ 6. Rraw the temperature versus timeand the solubility versus time curves. Recrystalli~atio~ stopped at 25°C. then ln[q = ( . is shown in Fig. the starting temperature is 40"C. Combining Eqs.e.Frequently the value of y2 is 2 (i. If the solubility of a compound is assumed to follow a van't Hoff equation.A H / R T ) p
+
(6.1 1.50)
where p is a constant and AH is the heat of solution.1 a ~ ~ ~ e
Suppose a recrystallization takes place and the temperature is at 40°C at time 0 and at 25" at time 10. The question is: would the solubilities of the compound be as a function of cooling time? Heat transfer usually results in temperatures following a s i ~ a . is
. Assume the ambient (or cooling) temperature is 23°C.. 1/ n = 1/2). In the writing to follow the aim is to deduce what type of particle size distribution would result from thermal recrystallization..50) would then give the temperature as a function of time.m i n ufunction. A n ~ 6. and the harvesting temperature is 25"C.exp(k@)] (6. (6. and the temp~rature is then allowed to drop by cooling.1 e ~ ~ In the stated case. Eq.
23)(1 .014
This is now converted to solubility versus time.T = 23
+ (40 .51) (6.2*T) 1 = 3163.55)
C .exp(k@)
+ 17(1 . given the solubility of the compound to be recrystallized is at 40°C and 30 at 25°C then it is easilycalculated 50 that the solubility would be a function of temperature by way of ln[q = (3163. The solubility versus time curve is shown logarithmic fashion in Fig.S(23)I = 3.11.54)
+ 14. as the program in BASIC. 6.3 and 6.014 NEXT TI
.5/(T2 + 273. and the curve is a logarithmic in decay by way of Eq.
Program for ~ ~ ~ e rValues~ nWay of Eq.12.S(23) is the supersaturation and is given by
C .58) where A is the surface area.4). 6. (6.S(23) = 21.5/T) (6. The tem~erature versus time curve is shown in Fig. 7 = 25: ‘ 25 = 23 from which
k = 0.6 exp(0. 4 usually of value close to 2.exp(k@) +
(6.05 .51) a t by ~ For T1 = 0 to 10 T2 I= 23 + l~*EX~(O.1 1.exp(0. since at 4 = 10.23# (6. (4. 6.2
The curve T = 23 17(1 ..2@) is shown in Fig. most easily by programming (e. and A is a growth rate constant and g is an exponent.56)
as rom the example it is seenthat it is reasonable to assume that the s ~ ~ e ~ s a t ~ r a t i a function of time may be given by
h = (Co . Tables 6.S ) exp(k@) (6.55):
ln[C . gain assuming a van’t Hoff equation to hold then.15) S2 = SI + 14. ates of growth may frequently be expressed as (6.23@)
(6.0.g.57)
where Co is the concentration at the beginning temperature and S is the solublity at the ambient temperature.52)
or.
4 25.7 41.4 44. with side length r.62) (6.2 26.3 30.59)
so that
dm/dQ.2 34.61) now give
dm/dQ..~Q)
(6.2 32.3 30.000
3.3 28.g ~ # )
(6.6Q)and (6.6 30. 1 27.3 Time
0 1 2 3 4 5 6 7 8 9 10
Temperature 40 36.0
The mass of one single cubic particle.8 25
Solubility 49.9 34.1 38.61)
Eqs. = 3r2p(dr/d#) = ~ 6 r 2 ~ g e x p ( .5
Time
Concentr~tion ess saturation concentration at 23°C plotted in a s e m i l o ~ a r i t ~ i ~ 'L ig.
.1 31.0476 .0 . fashion versus time. = 3r2p(dr/dQ.) The area A of the surface of the cube is
A = 6r2
(6.6 29.3 36.63)
or
y
. is
m = r3p
(6. (6. 2 3 3 8 6 ~ R*2
=
1.3.5 33.Printout of Data Generated in Table 6.4 32.
65)
(6. 6. If r ~ ~ the growth is impaired in two directions.69)
s mentioned in the foregoing. may be such that one of the three situations is preferred. resemble lognormal distribution functions.13).The length of time given the crystallization is denoted t. This may eliminate the lag time and.
. per se. 6. It can either grow equally rapidly in all directions (situation i in Fig.66)
t
exp(~#/~)d# ah'/"{1 .64)
The size of the largest particle ro is obtained by inserting # = 0 in this equation:
ro = (2M/p)Ag[l . in which casea plate results. there is often a lag time before nucleation starts. 6.
nce a nucleus isformed at or beyond the critical size it will continue to grow.13). This. (6.67)
y the same argument. The drug substance. AG in Fig.13). (6. 6. the number of particles with particle size larger than r4 is denoted N .egk"]
he number of particles that are born between time Q1. in some ways.67)). and is given by integration of the integral in Eq.68)
Equation (6..67) from # to t:
N > r ~ @aA'/"{ek@/" ~ __
(6.e"kt1n1 =
(6.64) may be written [egk@] [pr4/(2MAg)] egkz = Inserting this in Eq. It is customary to seed a crystallization with seeds of the drug substance. (6. (Eq.
+
(6.68) now gives (6. There are some compounds that always crystallize out as needles. then a needle results (see situation iii in Fig.2). often.SI1/" = aACIn x p ( . for n = 1/2 and g = 2. The size ofa particle that to is born at time # is given by: exp(gk#)d4 = ( 2 M / p ) A ~ [ e '.e. (6. is tied in with the metastable zone.~ # / ~ ) e The total number of particles is obtained by integrating this from 0 to
z
(6. Carstensen (1980) and Rodriquez (1985) have shown that these functions. when the symbol r is substituted for r(#) is the cumulative distribution function.70) which. and dq5 is given by: J = dN/d@= a[(C(#).egkz] ~~~ (6. or the growth may be i ~ p a iin one direction (see ii in Fig.70) divided by Eq. reduces the energy ofactivation for the critical nucleus formation (i.
(Data from Chow et al. Int J Pbarm 24239258.iii
\
Creation of different crystal habits from a nucleus. New York.
. Chow PKK.. Chow A HL. Bikeman JJ (1970). Rodrugue~~ornedo (1985).14 shows the results from the presence of ~aceto~ytacetanilide on the dimensions (len~th~breadth ratio) of acetaminophen (Chow et al. (iii) two directional growth rates are lower than the third.
0
1
2
Additive g/L
ig. 1985). (ii) one directional growth rate is lower than the other two. p 215. ~ o d i ~ c a t i o nacetaminophen crystals: of influence of growth in aqueous solution containing ~aceto~ytacetanilide crystal properties. Int J Pharm 34:197205.) on
owever. J Chim Phys 10:624.Solid Pharmaceutics: Mechanical Properties and Rate Phenomena.. Figure 6. additives affect the dimensions in the crystallization of certain substances. Ind Eng Chem 43: 1460. Academic Press. Doherty C. 1985. Dankwerts PV (1951). New York. Carstensen JT (1980).
Berthold A (1912). (i) all directional growth rates are equal. pp 3032. J Pharm Sci 74: 1322. Zbognshan W. Academic Press. York P (1987). Physical Surfaces. N Carstensen JT. Grant DJW (1985).
Noyes AA. Noyes AA. J P h a m Sci 85:112. Matsuda Y (1996). Cleveland. Crystallization. Butterworths. Mullin JW (1961). Ostwald W (1897). p 106. Z Phys Chem 61:385. Chemical Rubber. Nyvlt J (1971). Marc R (1908). Valeton JJP (1923). Trans Am fnst Chem Eng 31:365. Industrial Crystallisatio~from Solutions.Z Phys Chem 47:689. London. Whitney WR (1897). OH. J Am Chem SOC19:930. Whitney WR (1 897). 2. 335. Phys Chem 22:289.
. Z Krist~llogr59:135.Higbie R (1935). Otsuka N.
. Use of Modulated ~ifferential 7.
7. 1993. In general.8.6. Pikal.5. et al.7.3. At times materials are produced in amorphous form by methods usually used for producing crystalline modi~cations.
ethods of Preparation
108 S9 O 109 110
7. ecrystallization from different solvents is not always successful. The theory of ‘Crolmer and Web ker and Doring (the states that crystals form as a function ter formation (of volume sat~ration as an end result. Amorphates 7. stability (Carstensen and Carstensen et al. the nucleation rate J is a function of interfacial tension S. Crystallization of Amorphates
t was previously mentioned that ~ ~ o r ~ ~ simplyedefinedas materials that are are ~ t s not crystalline.1.2. Ahmed 1992. Water Absorption “Isotherms” into Amorphates
110
112 112 I14 115 115
7. Polymers Symbols eferences
etermination of Amorphates
7. Chow and Grant (1988. 1998). 1989) have described that recrystallization of acetaminophen from a series of solvents gave rise to amorphous material and different crystal forms. 1978). although there have been some exceptions reported in the liter of this they have higher dissolution rates and apparent solubilities eezer.7..4. they are more energetic (less stable) than any cr stalline form. Class Transition Temperatures of Mixtures Scanning Calorimetry 7. 0 between solid and liquid by the following formula:
~~
.
produced by ~eZting sucrose. and in such a case. but does not apply well to supersaturation situations or melts. to create a substance in amorphous form when it is easily crystallized may be achieved. if this glass is suspended in 95% ethanol and stirred at length. is highly dependent on the viscosity of the melt.~ ~ lactose ~ g a high amorphate content. As an example of ~ ~ ~ u ~ . Turnbull and Fisher (1949) modified the equation to read:
~~~
f = e x ~ [ ~ . the problem in lies producing the crystalline substance.1 6 ~ ~ ~ u ~ ~ / { 3 R AG. Amo~icillin trihydrate becomes amorphous on ~ e ~ ~ ~ ~ ~ t i o n et al. in general. in the following manners: Sublimation Supercooling of melts ~eutralization an acid with a base (if the drug is an acid) or vice versa of Recrystallization from a variety of solvents ehydration of hydrates ophilization (e. particularly for more complex molecules.g./RT] l n ~ ) ~ ~ ] ~T~(
+
where AG. spraydried ~ i has . The opposite. AG. R is the gas constant. However. therefore. (1997) prepared three crystalline forms of ciprofloxacin HCl. furthermore. If it crystallizes during storage. hard candy. some means of nucleationmust be created. The equation holds well for vapors and solutions (Mullins and Leci.er
J = exp( . of amor~hous sugar and will be touched on later in the chapter. echanical interaction is often a means. by “kugeln”) raydrying
To name some examples. it is remembered from organic laboratory courses that students will produce adispersion ina test tube.. then it becomes cloudy and is considered defective candy. 1969). that certain substances that possess high viscosity at their melting point may be prone to become amorphous on melting and recooling
At times it is actually difficult to prepare an organic compound in crystalline form.1 6 ~ ~ 3 u 3 ~ / { 3 R 3 T 3 ( l n ~ ) 2 } ]
where N is Avogadro’s number. In general terms. Hildebrand and M~ller~oymann (1967) report on the produ ketoprofen by neutralization of ketoprofen with sodium hydroxide. a crystalline sodiumsalt will eventually occur. and Tis absolute temperature. and then scrape the side of the test tube with a spatula.
. It is obvious. is activation energy for motion of molecules across the matrix~luster interface. is amorphous. and water as solvents produce a hygroscopic glass. Tamann (1926) showed that for melts there is maximum in J at a particular temperature. they re pared the amorphous form by Z ~ o ~ ~ i ~ i ~yophilization sucrose produces z~ti~n.
If one melts a (stable) solid and recools it.Km2) l
(7. therefore. 7. 2 ) ~ ( m4. This requires nucleation. the viscosity of the melt is often of the magnitude of l0l2 Pa s (Lu and Zografi. Materials that are viscous about their melting point are. lower than the melting point. 1997).
Tg =
TgI
4.
.3)
where
’1
‘O
Rubbery
50
50 T(m)
150
250
Temperature O C
Molecular volume as a function of temperature of a solid prone to forming an amorphate.~ m 2 T .” For lyophilized materials that produce amorphous cakes.. 1 0 ’ ~a s) that will allow the cake to deteriorate. and below C it is referredto as a glass. An example of this is shown in Fig. in which the molecules are arranged in lattices (ordered arrays) and the orientation of each molecule is set. the “collapse temperature” is essentially the temperature at which the viscosity drops below a critical viscosity (e. At a temperature T..Solids that are not crystalline are denoted amorphous. there will be a physical in the amorphate.1. Just below the melting point. and this is the type of material referred to as amorphous. At a given high viscosity(attained at or lower than the melting point).. the melt will have the appearance of a solid. and molecular movements will be random in direction and magnitude (within the limits of the system). and nucleation propensity is a function of the viscosity of the liquid in which it occurs.
It is often of importa~ce to estimate the glass transition of an amorphate that has a ’ certain water (or solvent) content. prone to form supercooled solutions.g. If values of T (Tsl and Ts2)are known at two different water contents (ml and m2). 1952). and is referredto asthe “rubbery” state. At the glass transition temperature T. etween points A and 313 the properties of the amorphate is often similar to that of the melt. and this is the “cutoff point’’ between a “liquid” and a “solid. the molecules will have no specific orientation. then it should crystallize when the melting point is reached. then Tg at other water content may be estimated by using the CordonTaylor equation (Cordon and Taylor. as opposed to a crystalline material.
8
1. this line is an extension of the solution ik vapor pressure line (see AB in Fig. and one may consider the moist amorphate as a highly concentrated.2). As shown by Carstensen and ~ a n ~ c o(1989) for amorphous sugar. although amorphates in the rubbery state (just below the melting point of the crystalline form of the compound) are actually highly viscous li~uids.x) where x is the molefraction of solute. they will pick moisture in a fashion d up that is not like that of a BET isotherm (to be covered shortly).
A~orphates solids that are not cryst~lline. 1993). 7. (1998) investigated amorphous alactose and were able to measure the heat capacity at Tg separately from the endotherm.2
(14
Moisture isothermfor an amorphous solid. (1998) have described this techni~ue. and it may be considered a s o Z ~ t i o ~ . the water activity a. 7. 7.2).6
0.24 in Fig.4
0. rather the cornposition will change. then the isotherm is often quite linear if the amount of water absorbed is expressed as olef fraction (line DE in Fig. but the vapor pressure’willstay constant. the solution itselfwillbe saturated. Itis assumed at this point that the are term solid is selfevident.This is referred to in the following as MDSC. an amorphous form of the compound is produced and exposed to different relative humidities.2
0. which rather than using a linear cooling or heating ramp.I
Solutlon
+
Preclpitate
te
0. At a given point (x = 0. Hill et al.
f
6u.. a sinusoidal in temperature profileisused (Reading et al. (This concentration differs from compound to comyond this concentration.0
0. When exposed to h u ~ i atmospheres. will decrease linearly with (1 . (Data from Carstensen and VanScoik.Hill et al.
0
. In an ideal situation. In this type of graph rdinates are in a direction opposite that of a usual isotherm.
1988.)
. supersaturated “solution. and the ure will not change with further addition of compou~d.0
1. The moisture actually penetrates into the solid.”
Dilute Solutlon
.2) the solution becomes saturated.
(1996) found that [6fluoro2(2’fluorol. 7 as accurately as possible..5
0. y exposing a m o ~ h o u sucrose to various relative humidities. because here it is probably not an adsorption..? not of the BETtype) up to a relative humidity of 82%. When it is exposed to 75% water rapidly and changes into the hemihydrate. ancock and Zografi (1993) later used this principle in their investigation of whether solution theory could be applied to macromolecules. 1993.3 0.” Their data for polyvinylpyrrolidone (PVP) K30 are shown in Fig.haveshown moisture isotherms of cefadroxil. an absorption that is at play. the vapor pressure depression is larger (owing to the two or threefold number of ionic particles. over that of the molarity of the salt). If these moisture levels were expressed as molefraction of sucrose.2
0.. Both forms are quite watersoluble. 6 and Fig.lbipheny14y1)3met~yl]4quinolinec~rboxylic acid sodium salt (brequinar sodium) e an amorphous form or as a hemihydrate.0
Water Activity
Fit of vapor pressure data of aqueous solutions of PVP K30 at 30°C to the FloryHuggins equation.4
0. owing to crystalli~ation and because the crystalline phase only can handle surface adsorbed moisture (Le. The fraction to the right of point is the principle used for salt solutions to obtain constant relative humidity in de cators. With electrolytes. for amorphate..0
0. 7.e. Carstensen and VanScoik (1988) were the first to point out that for an amorphous substance. but rather.2
0. and the solubilities often high. 1993).) The points are taken frorn their Fig. as is the trace of the FloryHuggins equation.6
0.
.ecause of the random arrangement and the mobility of the molecules in an amorphate. then it should be possible to use basic solution theories to model the data.4
0. crystalline anhydrate. To quote “If one considers the absorption process to be completelyanalogous to the solution p~ocess. much less than the amorphous rubbery phase). as opposed to a crystalline modification. and hydrate.8
1. then the vapor pressures fell in line with vapor pressure curve of sucrose the solutions itself.0 0. Lehto and Laine (1998).1
0. amorphates are usually chemically less stable than crystalline modi~cations (Carstensen et al. at which point the isotherm suddenly drops. various moists ure leveIs were reached.6
0. u et al. The amorphate yields an isotherm that is constantly increasing (i. it is illogicalto use the traditional moisture isotherms. (Data from Hancock and Zografi. so that these are preferred for creating constant relative humidity in desiccators.3.
0.
the ~uantitativecontent of amor~hous c o ~ ~ o n emay be obtained by microcalorimetry. Cases often exist in which a drug substance or excipient is partially crystalline and partially amorphous. Densities ofamorphous materials (pa) are most often less than those (pa) of crystalline solids. for instance. B u c ~ t et ~ 1995.9 are not shown. (1980) as and by ~ ~ b s k a et aal. y icrocalorimetry has been a useful tool in the detection of minor contents of ous m a t e ~ acaused by.. but the adherence at activities above 0.. and Ryan (1986) optimized crystallinity of lyophilizates this. Hancock (1998) has cautioned that there are several shortcomings of this method. §tubberud and Forbes (1998) used microgravimetric method (CISORP Automated Sorption oni it or). enthalpies are subjected to correction to bring them to “the same temperature’’ by a method forwarded by offmann (1958). They f o u ~ d PVP to act as aninternal desiccant and delay the onset of crystallization. The Vrentas equation probably provides a better fit. (1977) d e t e ~ i n e ~ fraction the crystallinity in samplesofdigoxin powder. 7. 1998). (1995). Density measurements may also nt be used. (1983) have investiin gated the effect that crystallinity has on the enzymatic hydrolysisof the palmitate of chloramphenicol. Sebhatu l et al. 1994. so that the content f of amorphous component may be assessed from
The most common method of measuring the transformation of the etas stable amorphate to more stable crystalline forms is bywayof Xray if fraction.3 is that the data fit neither the Flor nor the Vrentas equation.way. can be d e t e ~ i n e dwhich is better than d e t e ~ i n a t i o n Xray diffraction. 1989).1
ter
The importantfeature in Fig. because melting and crystallization occur at different points (Hancock. However. Also..
. and contents of as little as 1% o al. . by Phillips (1997) described a means of estimation the content of amorphate in has pharm~ceutical powders by means of calorimetry. 1994. but many nonhygroscopic tablet excipients accelerate it. Amorphous materials. also shown by Imaizini etal. It is based on comparing the size of enthalpic changes in fusion and crystallization. sample to compare the test sample against.
Traditionally. Bernabei et 81. may occur in the same temperature range as melting (Ford and Timmins. Black and Lovering. milling (Briggner et al. data become slightly uncertain at such high h~midities. the fraction of a solid that is amorphous has been d e t e ~ i n e d by means of Xray diffraction. to study the crystallization of amorphous lactose. crystallographically pure. In such cases. 1996). Ahmedet al. events suchas transitions and desolvations. it is difficult to obtain a sufficiently. in any event. as shown by Carstensen and Morris (1989).. are less chemically stable than their crystalline counte~arts. for one. and Junginger (1977) ~ e t e r m i n e ~ the degree of phase t r a n s f o ~ a t i o n this manner.
the vapor pressure curve is in line with the vapor pressure curve of sucrose at less than saturation. 1996a. the sucrose will begin crystallizing. corre. ~icrocalorim~tric methods have also been used to study amorphous tocrystalline t r a ~ s ~ o ~ a t i o n s (Hansen et al. 1994). which is the inverse of the solubility and. it is logical to view this amor~hous state as a supersaturated solution. and the weight drops until all the sucrose has crystallized..6). 7. and at a certain given time. and the weight checked as a function of exposure time. 7. as mentioned. moist state behaves similar to a “solution” (Le. Hence. but rather.b. The first event that occurs is a contraction of the spheres in size.b). After freezedrying. The glass transition temperature is a function of moisture content. Saleki~erhardt et al. the plot has a positive slope. This is point A in Fig.
. apparently Tg decreases. The plot is plotted as the plateau level... and as such should follow a van? Hoff plot. presumably owing to a change from rubbery to glassy state.. themselves (AB) may.4. be considered solubility.
0 0
IO
20
30
40
Time (days)
Weight gain at 23°C and 33% RH. and lyophilizing them on petri dishes in a fashion such that no sphere touched another sphere. so that the weight drops. As seen in Fig. the petri dishes wereexposed to different relative humidities and temperatures. A plateau is then reached. They produced amorphous sugar spheres by pipetting sucrose solutions into liquid nitrogen (socalled kugeln). 7. The weight gain at a certain relative humidity.. a very concentrated.5.1992a. 1990) employed weight gain (Fig.4) to follow a probit function.4) as a means of studying the conversion of amorphous sucrose into crystalline sucrose. so that the transition is facilitated. such that the amorphous. would be part of an isotherm.. 1998. traditionally. 1991a. as they indeed do (Fig. The kinetics of transformation has been discussed by several authors. and as this increases.crystalline content as high as 10% may go undetected by this method (Ahmed et al. (1998)employ firstorder kineticsin the transformation of amorphous to crystalline griseof~lvin. supersaturated solution of sucrose in water). Carstensen and VanScoik (1989. Ahmed et al. Angberg et al. but these isotherms are not of the conventional type. The crystals cannot “hold” water in the same fashion that the amorphous phase can.b. Carstensen and VanScoik found the points after the drop in weight (see phase CD Fig. The levels. 7. 7. hence.
and one means of assessing the effectiveness ofa plasticizer is to record the glass transition temperature as a function of plasticizer content.4566~RA2= 0. The aspect of amorphicity is.4
0.2
I .3
3. The rubbery state confers elasticity to the polymer film. but on nucleation.4
1ooorr
Van't Hoff plot of plateau levels in moisture uptake by amorphates.8
0. (1991a.4
0.2
0.0.3
0. of great importance. and the concentration will decreaseto the level ofthe solubility of a crystalline form.6
0.5
0. Larsen et al.2 00
0. Transfo~ationsmay also betestedbyway of dissolution.o
'
rc\
e.97
.b) employedthe method to study the transformatio~ amorof phous lactose into the crystalline hydrate.6. Plasticizers are added to polymers to achieve this. here. precipitation will occur.
Angberg et al. This decomposes very rapidly into the anhydrate.
. also employing this method.6
Mole Fraction Sucrose
y=
1.965
3.4827+ 2.
olymers will be the subject of Chapter 26 and constitute a special case of pharmaceutical solids.1.2
3. so it is important that Tgbe as low as possible.I
0.1
3. showed thatamorphous acadesinecrystallizes bywayof a metastable hydrate. or approach it. As is true with metastable polymorphs. (1997). concentrations will first rise to a high level (the apparent solubility of the amorphate).
If it rises above room temperature. Nystrorn C. Castensson S (1992a). Castensson S (1991a). Castensson S (1992b). water is most often a good plasticizer. Rawlins DA (1996). soft gelatin capsules is controlled to within close limits. Nystrorn C. Angberg M. Nystrom C. Bucktor! C . Rawlins D A (1998). will
f
= fraction of amorphate in a batch of drug or excipient G. willbe brittle. impedes the compression process. and highfines ~roduction result. Int J Pharm 130:195. = Gibbs’ energy for transport of a mole from cluster to solution J = nucleation rate m1 = mass of amorphous component 1 in a mixture of amorphates m2 = mass of amorphous component 2 in a mixture of amorphates IV = Avogadro’s number PVP = polyvinylpyrollidone S = supersaturation T = absolute temperature Tg = glass transition temperature Tgl = glass transition temperature of component 1 in a mixture Tg2 = glass transition temperature of component 2 in a mixture K = weighted ratio constant in a mixture of amorphates pa = density of amorphous phase in a mixture of amorphous and crystalline phase pc = density of crystalline phase in a mixture of amorphous and crystalline phase p1 = density of component 1 in a mixture of amorphates p2 = density of component 2 in a mixture of amorphates v = cluster volume cr == interfacial tension
. Castensson S (1991b). and 90) have shown that the glass transition temperature of dependent on moisture content. Angberg M. The same holds true for wet granulations. then it becomes brittle and shatters during comminution.
. then a compressed mass is easily formed. Int J Pharm 167: 139. If the granulation is overdried. this is due to the elasticity of the bonding bridge of the binder that keeps the particles together. vinyl pyrollidone (PVP) is a frequently used binder. Angberg M. Int J Pharm 83:11. Angberg M. in turn. Compressibility of tablets made from granulations is a function of the moisture content. Buckton 6 . Int J Pharrn 8 1:153. Below a critical limit the capsules will become brittle. then the polymer willbein the glassy state at the time of grinding. and often. giving rise to nts of fines which. Nystrorn C. Int J Pharm 77269. Ahmed H. Int J Pharrn 73:209.For watersoluble polymers. Above a critical capsule will become too soft and deform in the bottle. If it deforms easily.
The States of Aggregation. MullerGoymann CC (1997). Ryan JA (1986). p 105. trans. Phillips EM (1997). Int J Pharm 105:125. Lisnyak YV. J Chem Phys 17:71. Nambu N. Hansen LD. Bernabei MT. Int J P h a m 135:3 1. Feely LC (1998).Becker R. Bucton G. J Pharm Sci 7:654. Oksanen. Weber A (1925). Int J Pharm 161:93. Laine E (1998). Chow AH. Hansen LD. Pyne MT. Laine E (1998). ~ullins JW. J Pharm Phamacol 29:634. Briggner LE. Int J P h a m 154: 103. Hill VL. Ann Physik 24:719. J Therm Anal 40:949. ~ u b s ~ a AV. Morris T (1993). Lu Q. Timmins P (1989). John Wiley & Sons. Taylor JS (1952). Iannucelli V. Darcy P. Mehl RF. Greenleaf D. Holbrook P (1995). Junginger H (1977). Ford JJ. Bystrom K. Dtsch Apoth Ztg 117:456. Wood RW (1996b). S Pharm Sci 82:657. Carstensen JT. Leci CL (1969). Wu LS. Morris T (1993). Craig DQM. J Appl Chem 2428. VesaPekka L. Int J Pharm 52:123. Forbes RT (1998). Blagoy YP (1995). Carstensen JT. Angberg M. J P h a m Sci 86:854. Int J Pharm 51:115. Zografi G (1990). Coppi G. Wood RW (1996a). Buckton G. CA. Black DB. Zografi G (1997). Lovering EG (1988). Hemming DJB. Pudipeddi M. Bergstrom RG. New York. Int J Pharm 137: 1. J Cryst Growth 5:75. Int J Pharm 149:267. Ahlneck C (1994). Nagai T (1980). J Chem Phys 29:1192. Franchini M. Pharm Res 79:1374. Int J P h a m 163:49. Makkhar AP. Darcy P (1994). Int J Pharm 104: 135. Int J Pharm 116:113. Stubberud L. Int J Pharm 160: 131. Int J Pharm 82:R710. Drug Dev Ind Pharm 21:1953. Beezer AE (1992). ya Hi~debrand GE. Grant DJW (1988). Drug Dev Ind Pharm 23:1063. Fisher JC (1949). Lehto E. Chow AH. Cravvford JW. Imaizini H. Van Norstrand. 3 Pharm Sci 86: 1374. Hancock BC (1998). Chem P h a m Bull 28:2565. Hansen LD (1997). Gordon M. Farm Ed Prat 38:391. Wood RW.
. New York. Hussain MA (1996). Pang J. Forni F. Keiser DR. Reading M. Sebhatu T. Cameroni R (1983). Z Physk Chem 119277. Buckton G. Elliot D and Hill VL (1993). Int J Pharm 163:145. Pharrn Dev Techno1 1:43. Mendiratta A (1997). ama an^ G (1926). Volmer M. Larsen MJ. Int J Pharm 163:198. Doring W (1935). offmann JD (1958). Pharmaceutical Thermal Analysis: Techniques and Applications. Turnbull D. Singh M. Drug Dev Ind Pharm 19:1811. Grant DJW (1989).
1.8. ~olubility Thermodynamic Functions 8.8.1 through 8.3. but also distinctly different degrees of biological absorption.9.10. depending on how they are recrystallized. Rates of Conversion in Symbols References
The pharmaceutical interest in polymorphs is attributable to the work by Aguiar et al.4. Methods and Detection and Monotropes 8. enantiotropes and monotropes. Inorganic (particularly ionic) solids usually are associated with one and only one crystal system.
. may occur in several different crystal modi~cations (polymorp~s). There are two types of polymorphism.6. ~nantiotropes The “Disappearing” 8.2. They are distinguished by their vapor pressure diagrams and differential scanning calorimetry (DSC) traces (see Figs.5. (1967). Organic solids. Mixtures of Polymorphs Rates of Polymorphs and ~seudopo~ymorphs 8.4). Stability of Metastable Polymorphs~
118 119 121
8. ell~nown to all is that sodium chloride is cubic. 8.
8. however. who demonstrated that different poly~orphic forms of chloramphenicol gave not only different dissolution rates. Pseudopolymorphism and 8.7. Polymorphs. Dissol~tion 8.
s of Conversion in Moist Storage
123 125 125 126 127 128 129 129 130
8.
t may almost always be assumed that more than one polymorph exists in a new drug substance.g. Two polymorphs (i and ii) were obtained.
FTIR Spectra of the TWOPolymorphs
Functional group NH C=O. under the mentioned ne way of distinguishing between different polymorphs is by difference in Xray patterns. infrared spectra (IR) show distinct difference 8.andethanol (Table 8. 199’7) maybe attempted.
ods for (attempts at) producing different crystal forms most often take the of recrystallization from different solvents.1). Chow and Grant (1988. ethyl acetate. One of the tasks of the p h a ~ a c e u t i c a l scientists is then to produce as many polymorphs as possible at the earliest time possible during the product development stages. was This done at room temperature (RT) or ice cooling (I) with (S) or without ( ~ O magnetic ) stirring. by deyilleiers et al. ethyl ether.1767 1887
Source: Giordano et al. however.
.. (1998). (1998) studied crystal forms o f piroxicam pivalate by recrystallization from toluene. ester Polymorph i 3253 1760 1682 Polymorph ii
3291. Schnitzer et al. thensub1imation (e. (1998). Often.
Pyroxicam Pivalate Polymorphs
~~
Solvent Toluene Ethyl ether Ethyl acetate Ethanol
RTS
1
RTWO
i i
1
Is
1
1
Iwo
i
1
i
1 1
1 1
i tii
1
I
I
Source: Giordano et al. 1989) have described that recrystallization of acetaminophen from a series of solvents gave rise to amorphous material and different crystal forms.. ecrystallization from different solvents is not always successful. If the compound is heatstable. 3350 1750. but this could not be duplicated~in the same solvents. (1998).2 shows the wave numbers (reciprocal centimeters) of certain bands in the transform infrared (FTIR) spectra of the two poly~orphs.
0
0
90
100
150
200
Temp "C
Vapor pressure diagram of an enantiotropic pair.
.
it is advisable to recool the mass and record the melting point on the down trace. The DSC trace of such a pair may take one of several forms. and the two lower traces are the heating of the metastable polymorph. This is exemplified in Fig. Traces of the metastable form may either show up this way. and then (second en~otherm) remelt. 8.3. The stable form will simply show up as a trace with one endothem (the melting point of the stable form). precipitate (exotherm) as the stable form I.0
20
40
6 0 80 100 120 1 4 0
Temperature "C
Possible DSC traces resulting from heating of the room temperaturestable form of an enantiotropic pair. It follows from thermodynamics that the change in Cibbs' energy by a path from metastable to stable form AG..
0
50
100
150
200
Temp "C
Graph of vapor pressures for a monotropic pair. Most often. If the compound is stable to melting. is given by
In the top trace.4. 8. or they may show up as the middle trace in Fig. the situation where one form (form 11) is metastable throughout the melting range). It is negative. however.as shown in the middle trace: melt. 8.
. which may either sim~lymelt (lower trace) or. decomposition of the solid and melt preclude conclusions from cooling curves.4 that is heated.
The other case is monotropism (i'e. it is the stable polymorph in Fig. so the form with the highest vapor pressure at a given temperature is the least stable (metastable) compound.
The rate of transformation is directly proportional to the amount of solid not yet nucleated (frequently denoted "random nucleation"): da/dt = k( 1 . The nucleation rate may also be proportional with time (ie.a) which integrates to (84
.. the longer the elapsed time. and the rate of transformation is given by da/dt = k (8*2) where a is fraction converted.lnf1 . There are no complicating factors. the roba ability is timeindependent. polymorph transforms to a less energetic crystal form. to the most (or a more) stable crystal structure. and these crystals will later transform. If the nucleation event is such that it can occur throughout the solid then there are three cases: 1.f O ' * ' ' r " ' l " " * ' 0 2 0 40 GO 80 100 120 140
Temperature O C
Some possible DSC traces of the heating of polymorphs that are monotropic.a]= kt 3. If a denotes the amount transformed at time t. the more likely is it that a site will transform.
da/dt = k*(l .a)t If. then some of the possibilities are elaborated in the following.
Ostwald (1 899) formulated a rule of stages: when system first starts crystallizing. then (8.his
10
r
. is the This ~olany~in~er equation:
a = kt
(8* 3)
2. stagewise. it a initially will create the crystal structure that forms the smallest loss of free energy.6)
. the term k* = 2k2 is introduced. dry. as an input function. so that a simply increases linearly with time. There are several types of ~ e c h a n i s m that can occur when a metasta~le.
dln[1
 a ] = 2k2t
(8.8)
which integrates to
 In[(1 a)] = k2t2
(89)
This is a form of the AvramiErofeev equation. An example ofthis is shown in Fig. 8.5. 4. Similarly, if the nucleation rate is proportional to t, but the nucleation can take place in three directions, then In[(l  a)]= k3t3 (8.10)
This, also, is a form of AvramiErofeev equation. f, in two dimensions (exemplified by a cylinder transforming in a radial direcly; Fig. 8.6), the nucleation starts at the surface and works its way in, then the fraction not decomposed is given by (1  a)= 7t(R q)2/7tR2 7t(R kt)2/7tR2= [I  (k/R)tI2 =
(8.1 1)
where q is the thickness of the transformed layer, and R is the radius of the cylinder. This may now be written:
E  (1  a)’/2] ( k / R 2 ) t 1 = (8.12)
Note that the rate constant ( k / R 2 )is larger, the smaller the particle ( R ) . It is easily shown that in threedimensional diffusion, this becomes [I  (1  4 l i 3 1 = ( k / ~ 3 ) t (8.13)
the same type dependence of particle size as in the cylinder example. mples of work in the pharmaceutical area are the publications of Umeda et dealing with transf~rmati~n aceta~olamidepoly~orphsand the of eniwa et al. (1985). Transformations are followed by disappearance (or appearance) of Xray peaks, and the data are then plotted by the various equations, and the bestfit isfound. In general, this is not a particularly good method (the data may plot wellby many different equations), but in transformation rates, it works quite well.
I
20
I
*
’
I
”
I
”
(
40
60
a0
too
Time (hrs)
Graph dealing with the conversion of (a) pure aform and (b) crystals containing 1% of the yform.Adherence to Eq. (8.9) is better than that to Eq. (8.10). (Data from ~ a n e n i w a al., 1985.) et
h
Schematic of cylindrical model for linear decomposition.
lagden et al. (1998a) reported on four different polymorphs of sulfathiazole. They (1998b)reported that ani ~ p u r i t y sulfathiazole synthesis, etha~idosulfathiain zole, in concentrations as small as 1 mol%, stabilizes two of the metastable forms, form IT and 111, of the drug. nd ways of stabilizing metastable polymorphs is of great indusym et al., 1996), because the metastable polymorph maygive d bioavailability. Often impu~ties induce twinning, which will may inhibit the t r a n s f o ~ a t i o n ,as with terephthalic acid(Davey et al., 1994). Sometimes additives may be used (e.g., polymers prevent the change of the centric to form of 3~ethamido4~pyrollidino nitrobenzene into the noncentric form) (Davey et al., 1997). If the “initial’? form of a new drug is a ~ o n o t r o p e ,and the stable form is unknown, then at one point in the development, seeds of the stable form may occur, and after this point it may be impossible to produce the metastable monotrope ernstein, 1995). I have had personal experience with such “diisappearing” polymorphs in that, in the early X960s, a batch of benzodiazepam was made that had a slightly (l°C) higher melting point than usual. The conventional wisdom was that thenew batch was purer (for it hada lower meltingpoint), but the truth was that it was a different polymorph, and after that batch had been made, it was impossible to recreate the “old” form. This, in turn, led to a large amount of duplication of clinical work, because the clinical results, up until that time, had been based on a metastable (more soluble), now unavailable polymorph.
It can beshown by ’s law that solubilities are (ap~roximately) linearly related to vapor pressures ly activities such as solubility are linearly related to fugacities). The graphs in Figs. 8.1 and 8.3 then become as shown in Figs. 8.7 and 8.8. enry’s law argument is applied to Eq. (8.l), then
r
0
0
20
40 Temp, O
60
C
80
.7 §olubilities (in mass of solute per mass of solvent) of an ena~tiotropi~ pair
where S denotes solubility, R the gas constant, and 7 absolute temperature. ' There are examples for which the solubilities are close overthe entire temperature range ( ~ a r s t e ~ s e n Fr~nchini, and 1984a,b) and, in such cases, may be difficult it to separate the two polymorphs in the final puri~cation (recrystallization or reprecipitation), and there are cases where companies have been forced to suggest specifications that stipulate a minimum and a maximum of one polymorph in relation to another. raphs, such as those in Figs. 8.7 and 8.8 are often presented in loginverse fom: ln[S] = (hH'~/~)(lOOO/T)B
+
(8.15)
where S is solubility, AHs is the heat of solution at saturation, R is the gas constant, 1 is absolute te~perature, /3 is a constant. It is recalled, however, AHs is not ' and tht necessarily te~peratureindependent (see Chapter Z), and if this is not true, then the rant equation (hdipeddi, 1998; J o z w i a ~ o ~et ~ i 1996) applies. s al., In S =  A / T
+ Bln[iT] +
(8.16)
'c
Temp, "C
Form II
§olubilities (in mass of solute per mass of solvent) of a ~ o n o ~ r o pair. pi~
eference is made to Figs. 8.7 and 8.8. The tendency of a metastable polymorph to convert to a more stable polymorph is a ~ n c t i o nof the difference in chemical energy. This, in turn [see Eq. (8.2)) is a function of their solubilities. If e curves in Figs. 8.7 and 8.8 are veryclose to one another (Carstensen and 1994a,b), inparticular, if the compound is very soluble, then the rate of t r a n s f o ~ a tion can be exceedingly slow, and the possibility of compounds crystallizing out in the two different crystal forms exists. The regulatory authorities, presumably, are interested in the morphicpurity of compounds because of the effect of polymorphism on bioavailability, and this, in turn, is tied to the solubility of polym~rphs.The metastable forms have higher ap~arent solubilities than the stable forms; hence, they are likely to have higher bioavailabilities. However, in a situation as just described, the solubilities can be sufficiently close that one form is as bioavailable as another, under similar conditions (particle size, moisture content).
One aspect of polymorphism is that the metastable form will have a higher “solubility” than the stable form. The word s o Z ~ ~ i Zhasybeen placed inquotation marks, ~t because theoretically a compound can only have one solubility. It has been seen, in Chapter 6, crystallization, that real equilibrium solubility happens only at infinite size of the particles, or at a secondary energy minimum. The point is that when the solubility is determined, an excess of solid is placed in contact with liquid that is stirred until“e~uilibrium”occurs. The facet of this solubility is that it is repeatable, so that for a metastable compound a reproducible number is arrived at, and this number is higher than the solubility of the more stable polymorph. The molecules insolution, however, are thesame, and the saturated solutionof the metastable polymerissimply asupersaturatedsolution of the compound. Seeding it or waiting for a sufficiently long time will result in precipitation of the more stable polymorph. One facet of polymorphism is, therefore, that solutions made from different polymo~hs contain the same compound. If a hydrateis dissolvedin water, then the solution will contain the same molecules as a solution made from the anhydrous material. Forthis reason, hydrates arecalled p s ~ ~ d o p o Z y ~ o ~ p prefix is derived The ~ s . from the fact that the solid composition differs (by water of crystallization^. The same argument holds for a solvate and solutions in the solvent in question. zuel(1991) and Golic et al. (1992) showed that norfloxacin forms different Sustar et al. (1993) showed that it formsat least two different crystalazuel (1991) and Golic et al. (1992) elucidated the crystal structure of norfloxacin, and Turel(l997) that cipro~oxacin of hydrates. The water is present in a complicated structure of hydrogen bonding. The manner in which the hydr made is as follows: The ciprofloxacin was dissolved in 1 1 molar ratio of Cs a : water, and an addition a few drops of 2 M sodium hydroxide would then of the solution. The crystals would grow in a couple of days ~hexahydrat ammonia isused to dissolve the ciprofloxacin, then, depending on the a m ~ o n i a co~centration, either a tetra or a hexahydra~eis formed.
There are often several (e.g., three different) polymorphic forms of an anhydrate, as well as solvates (Schnitzer et al., 1997) (e.g., there are three anhydrous, crystalline modi~cations prema~oxacin[forms I to 1 1 and two solvates [a hydrate and a of 11 met ha no late^). The DSC trace of form I is shown in Fig. 8.9. Note, that the upward peaks in this presentation are exotherms and that the events at appro~imately145" and 170°C are endothermic conversions, with a subsequent exotherm (indi~atinga change in morphism to a more stable form), and that the lone endotherm at about 205°C implies a single endothermic change (e.g,, melting) to a (physically) stable state (e.g., melt). They determined the enthalpy of solution of each form and found values of 33.2 kJ/mol for form 1 and 24.4 kJ/mol for f o m 111. The difference is 8.8 kJ/ mol, form I11 having the lower enthalphy solid phase. They take this difference to be ifference in molar entropy of the two forms:
AHI+III= 8.8 kJ/mol
(8.17)
The solubilities in ethyl acetate were sI = 3.23 mg/mL for form sIII 0.14 mg/mL for form 111. They then employed the (approximate) Cibbs' rela= tion. to calculate that
A G I  . + =~7.6 kJ/mol ~ ~
(8.19)
y employing the relation AGIbIII = AffI+III  TASI411 (820) and noting that both AH and AG were determined at T = 298 I it follows that (
A S = E7.6  (8.8)]/298 = 0.004 IkJfmol
= 4 J/mol
(8.21)
It follows, therefore, that the entropy term is rather insigni~cant comparison with in the enthalpy difference, and the authors conclude that forms I and I11 ~ o n s t i t a t ~ ~
6
8 100
120
140
160
180
200
220
Temperature, "C
DSC trace of form I of premafloxacin. Cooling and reheating produces only the endotherm at about 200°C. (Data from Schinzer et al., 1997.)
~ o ~ o t ~pair (i.e., the free energy of form 111 is lower than that of form I at all o ~ i c temperatures below the melting point). There are two dangers in the approach taken bythese authors (and many others, for that matter). The first is that Eq. (8.17) is based on bulk calor~metry; therefore, H is the integral heat of solution. To apply to saturated solutions, the e~thaipy term s ~ o u l d the ~ifferential be heat of solution at saturation (or nearsaturation) conditions (Pudipeddi, 1996). The other is that Eq. (8.18) is correct ~ ~ c t i v i t i e s , not c o ~ c e n t ~ a t i oare ,e ~ ~ ~ is true that in.dilute solutions activity coefficients ~s It o ~ e ~ are close to unity, but they need not be so in concentrated solutions, and the important question is the~ a g n i t u d of the ratio of activity coefficientsat Slrl and S I . If one e were, for example, 0.8and the other 1.2, it would not seem to be all that serious, but the ratio would be off by 33% (or 50% whichever way one looks at it). ecause S is calculate^ from the difference of twonumbers that aresuspect, it is always dangerous to ~r~~ co~cZ~sions from its magnitude, in particular, if it is small. There is an a~undanceof reports on poly~orphs today’s phar~aceutical in literature. For instance, Giordano et al, (1998) have reported on two polymers of piroxicam pivalate (PIP: . ~olymorph had the highermelting point ( I polymorph 1 the lower 36.5”). Their fusion enthalpies were found by 1 78.8 and 81.4 J/g. The “heat of fusion rule” (Ciordano etal., 1998; Yu, that if AHII  AHI is positive, then enantiotro~y what is expected. is Kakkar et al. (1997) have prepared three crystalline forms of cipro~oxacin e l , form I by cool evaporation from water, form 11 from coo1 eva~oration from 1:2 ~ a t e r / m e t h aboth~with a crystallization time of 50 h, and form 1 1 by cool ~~ , 1 evapo~ation from dimet~ylformamide. They furthermore prepared the amor~hous form by lyophilizati~n andby sprayd~ying. Some of the properties are shown in Table 8.3. Some trends are noteworthy, and often apply. The physically least stable is the amorphous and the most stable is form I judging from the solubilities. The least stable crystalline form has the lowest density.
.
ixtures of ~olymorphs may occur, but the extent to which this may occur be of can great importance and cause great difficulty in product development. As an example,
Cipro~o~acinHCl Forms Denslty (g/cm3) ~elting point (“C) 316.7a 0.796 0.980 1.042 3 12 3 1.5 0.5 No. of solvent moles per mole of drug Solubility at 37°C (mgicm3) 70 54 45 34
Sample
Amorp~ous Form 313.5 I Form I1 Form 316.3 I11
aSoftenmg point.
1
ter
Bergren et al. (1996) have described two forms of delaviridine mesylate (forms VI11 and form XI) both of which are anhydrous and nonhygroscopic. Sarver et al. (1998), however, crystallized delaviridine mesylate from acetonitrile at room temperature asforms I and 11. Both forms were very hygroscopic, and they subse~u~ntly recrystallized the compound from methanol under reflux. Acetone was added as acosolvent. This produced either form XI or form VIII, depen~ing on the amount of cosolvent used. In distinguishing between the many polymorphs, Sarver et al. employed,factor alinowski and Howery, 1981), in which large sets of data canbe segmenaller sets of orthogonal components. Of these, the first describes the e data, and followi~gcomponents deal with variance of less this means it is possible to distinguish between polymorphs in mixtures.
It has been shown in the previous sections that the less physically stable the polymorph or solvate, the higher is the solubility. Because dissolution rate, that is, the characteristic that is of importance biopharma~utically, directly related to soluis bility, there must be a connection between the two, This has, indeed, beenshown for phenylbutazone ( ~ a t s u n g aet al., 1976; Ibrahim et al., 1977; Muller, 1980; Tuladhar et al., 1983; ~ a t s u m o t e al., 1988; et Kaneniwa, 1988); for mefenamicacid (Aguiar and Zelmeer, 1969); for diflunisal (MartinezOhariz et al., 1994); for indomethacin (Kaneniwa et al., 1985); for tenoxicam (Nabulsi et al., 1992); and for oxyphe~butazone (Stoltz et al., 1988). Tros de Ilarduya et al.(1997) have described the dissolution behavior of pseudopolymorphs of sulindac. Table 8.4 shows the effect of solvates on dissolution rates. It would appear from the data in the table that the solvates are metastable in relation to the nonsolvate forms. ~ ~ y~ a ~ i ta affect dissolution rates as well (~arstensen,1973; ~ t may ~ Mitchel, 1980, 1981; Chow and Grant, 1989). The dissolution pattern of a metastable polymorph can be one of two types as shown in Fig. 8.10. The metastable solution may be fairly stable, so that the con
Dissolution Rates of Polymorphic Forms of Sulindac
I I1 (tabular habit) I1 (hexagonal habit) Acetonate ~h ~o r o f or mat e
Source: Tros de Ilarduya et al. (1997).
0.036 0.03 1 0.036 0.076 0.076
centrations with time will approach the metastable solubility (the middle curve); at one point in time, the stable modification may start precipitating out, and the concentrations will drop and eventually approac~ that of the stable modification. The latter phase is one manner in which r e c r ~ ~ ~ a Z Z i zcani be carried out a~ o~ strictly isothermally. The latter points of the precipitation in Fig. 8.10 (the mi curve) are frequently plotted semilogarit~mically versus time.
Good stability of a metastable compound can be achieved by (a) low temperature, (b) coarse crystals, and (c) dry storage. The moisture is the most serious contributor to conversion. oisture will condense onto the surface of the metastable form (11), and will then saturate the moisture layer to form a solution that supersaturated in (I). This is will eventually nucleate, and all of the I1 will convert to I. The conversion rate, therefore, is a function of the nucleation rate in “s0lution?” and it is a well known fact ( ~ u l l i n , 196l), that the n~cleationrate J is inversely proporti~nalto the viscosity of the solution, and also to the supersaturation ratio AS, by the following relation:
J = A exp(q / [ {T 3In AS)
+ (AGJRT)])
(8.22)
an equation discussed in Chapter 7 . For very soluble compounds, A S will be a very small number, and the tendency for one polymorph to change into another will be very small. F ~ r t ~ e r m o rif ,the solubility is high, then the AGm terms will not differ e much. An example of this is ranitidine (Carstensen and Franchini, 1994a,b).
AHaT= heat of solution at saturation, lsCal/mole = gas constant S = solubility, weight per weight of solvent
R
Metastable Pre~ipitation of
0I
0
5
10
Tine
iss solution of (squares) a metastabl~~ o l y m ~ r p h ; (circles) a metastable polymorph that on dissolution c o m m e n ~ s convert to a stable m o d i ~ c a t i ~ and (triangles) a to n, stable polymorph.
T
= =
absolute t ~ ~ ~ e ~ a t ~ ~ e constant in. the Van’t
Sustar I3, Bukovec N, Bukovec P (1993). J H e m Anal 40:475. Tros de Ilarduya MC, Martin C, Goni MM, Martinez~h~rriz MC (1997). Pharm 23: 1095. Tuladar MD, Carless JE, Summers MP (1983). J P h a m Pharmacol 35:208. devilliers MM, Wurster DE, van der Watt JG, Ketkar A (1998). Int J Pharm 163:219. Yu L (1995). J Pharm Sei 84:966.
This Page Intentionally Left Blank
.
Equilibria of Compounds Forming a Crystalline Anhydrate and Two Hydrate Forms 9. Condensation 9.12. bria of Co than Two Forming a Crystalline Anhydrate and Forms
148
150 151 153 156 156 157
9.1. Crystalline stalline.5.9.7. Equilibr~um
134 138 138 140 142
144
r Versus ~ o ~ p o s i ~ i o n 9.8.11.6.10.4. Critical T ~ m ~ e r a t uand Pressure re 9. Non~ydrateForming. Solvates Symbols eferences
. Substances that isture Adsorption or Absorption on or into Large.9. Critical Moisture Content Moisture Curves €or Salt Hydrates 9. E ~ u i l ~ ~ r iC o ~ p o u n d Forming a Crystalline Anhydrate and 146 9. WaterSoluble Substances 9. Presentatio~ Mode of ~ a t e Vapor ~ressure 145 Diagrams of a s 9.13. Moisture Equilibrium Curves of a Smooth Nature 9.
and the characterization “completely waterinsoluble” should be taken in this vein.1. These figures are shown in
. 4. owing to its large surface area.1 and Table 9. but do not form hydrates Amorphous substances Crystalline (anhydro~s) substances that f o m one hydrate Amorphous anhydrates that form one crystalline hydrate Crystalline anhydrates that form several crystalline hydrates
There is always somewater s~lubility associated with a compound. however poorly soluble. and definitions. 9. 7.very slightly and slightlysoluble) are not possible. 9. for the first in row Table 9. 9.g. 3. ubstances that are “completely” waterinsoluble (e. 9.hen a solid is placed in atmosphere. silica) ubstances that do not form stoichiometric hydrates. andthis can easily bedone. the concept involves pickup of moisture.2.7 mL. Table 9.2 x 273/1 = 43. 5. well as a kinetic property it also contains a As thermodynamic one.. There are seven distinct categories of solids.This substance.2. depending on the an substance) moisture from the atmosphere. but can ~ ~ s o r moisture by penetration (e. The e~uilibrium values are plotted as a function of relative humidity in Fig. at for example. In Fig.5 x mol.1.3.1 it is noted that there is an (approximately) linear rate at low time poin these are shown in Fig.g. The rate and extent to which this occurs is usually referred to as ~ygrosco~icity. 9. The volumeof this at 0°C and 1 atm would be IV = ~~~/~= 19. montmorillinite or cornstarch) Crystalline substances that are(moderately or very) watersoluble.. which will be treated in separate section in the following: 1. 6.. the weight of the sample will first rise fairly inetic phase) and the rate of this is referred to as the ~ o i s t ~~ e ~ t a ~ r In Fig.g. the number of is moles n = 35 x 103/18 = 19.2. and are seen to befaidy linear in lowlues. It is customary in isotherm work to convert these adsorbed amounts to the volume that would have been occupied 0°C and 1 atm. to n in Fig. it is also noted that the curves eventually 1 e~uilibrium level is a function of the relative humidity at which the e ~ ~ e r i m eis t n carried out. it will adsorb (or absorb. The levels are tabulated in the second column of Table 9. y its nature.5 x x 8. and that such pickup may be moderate at certain relative humidities and extensive at others indicates that classifications such as those used for solubility are not possible. as a desiccant in packaging of moisturesensitive drugs and drug prosilica gel is exposed an atmosphere of a given relativehumidity (RH). such as those used for solubility (e.2shows an exampleofthese equilibrium values at various relative humidities. but the large gamut of substances call for more detail ribe their hygroscopic classification. An example of this is silica gel. for instance). for instance) and very nonhygroscopic substances (sand. 2. At best one may talk about very hy~roscopicsubstance (choline salts.
9
128.8 20 45.0.1 20% RH 0 10 18 26 31 33 34 0
40% RH
60% RH 0 20.6 64.5 5 10 15 43.1 70.6 37.3 23.1
0 41.7 107.6
141* 1
.9 41.1 53.0 64.5 74.9 2 34.12100
+
3. 60.
Typical Adsorption Data as a Function of Time Time (days) 0 1 13.966

0
RH
Moisture uptake rates as a function of relative humidity for a “~ateri~soluble” compound. 40. and 80% R
y
.s
60%RH
4011 RH
20% RH
I
f
10 Time (days)
20
Moisture uptake curves for a sample of silica gel at 20.2 136.2550e2x R*2 0.
4. c is the cequation.
.2)
=: I
Treatment of the data in Table 9. STP)
RW/[ 100 . As mentioned in Chapter 5. Vmis the volume of a monolayer.8 72. 9.2 145.2 Equilibrium Values and BET Parameters from the Data in Table 9.5 89.2 plotted as adsorbed amount as a function of relative
humidity.. as mentioned.017 0.58)].1) is shown in Fig.9 180.9
the third column and are denoted Y .1)
(STP) adsorbed.022
0 43.4 versus t is seen (as mentioned in Chapter 5) that
20
40
GO
80
100
RW
Data from Table 9. (5.006 0.RH)] V(
0. Yjyt. so that the value of RH/~V{lOO RH)] (l/~~)[RH/lOO] (9. is the adsorbed volume (OOC. 1 atm) of water that constitutes just one layer on the entire surface of the solid sample. curves of this type are called BET isotherms [see Eqs.2
V (mL.4.l ) / ~ c Y ~ ) } ~ ~ H / l O O ~ 
+
(9. and the value of c is often large.6 59.01 1 0. 9. The Rvalue used is in units of cubic centimeters atmosphere (cm3atm).0 47. It can be shown that such data follow the BET equation: RH/[~{lOO RH}] = (1/{Ymc~((c .57) and (5. 9.1
H
0 20 40 60 80
Adsorbed (mg)
0 35.2 (last column).2 by Eq. and is plotted in Fig. RH/[ Y(100 been calculated in Table 9. (9. The data in the third column are shown in Fig.
number of molecules times the area of each molecule:
lo2'
X
(12.25 rn2/g.?3 12e2x R*2 = 1.2).1 = 0. such that I / Vm= slope = 0.0
0.2
0.5 X
= 125 m2
(9.2. This will be covered further in the section under pharmaceutical packaging.75 x lov4= 99
(9.5 x m2/molecule water is of usually employed ( ~ ~ d i p e d d1996). we may use Eq.3).027 that is. From the silica isotherm. then the specific surface area would be 31.0273/2. then. it is possibleto calculate the length of time the product is good.7475e4
+
2. = PVm/RT = 1 x 37/[82 x 2731 = 16.5 x lom4 mol
=6 X
X
16.
~lope/intercept c .1 =
so that
c . at which relative humidity this occurs. and dividing this figure by the moist~re penetration of the package.
vrn 1/0.0
RH/ 100
Data in Table 9.000
0.y
. if there is a critical moisture content beyondwhich the dosage form becomes unstable.so that in this example the entire surface area wouldbe the i.7)
A crosssectional area of 12.8)
If the weight of the sample were 4 g.6
0. one may then calculate how much moisture is taken up by the silica bag at this point. (9.2 treated by Eq.8
1. (9.3)
so
c = 100
Hence. If a bag of silica is placed in a bottle with a dosage form.4
0. it is possible to calculate from the isotherm of the dosage form.027 = 37 cm3 =
This can be converted to moles (n) and then to molecules (N):
n.
.5 x
= 1021molecules
(9.
microcrystallin givesrise to BET isotherms. (1980). nonprotic c ~ m p ~ u n (e. for instance.b).9. a solid is evacuated (Fig. saturated solution at exactly the critical relative h u ~ i ~ i t y . (b) active sites (BET model applies. ds sucrose). then a Other examples include. saturated solution of the compound.g. for example.1 and 9.2 and Fig. there is the sharp upswing at very high relative humidities. for example. and several electrolytes (e. a smooth equilibrium moisture curve will again. sodium chloride). When RM/[V(l . of the sphere that has a vapor pressure Pa. As me~tioned. below the critical relative humidity.. bulk.oisture isotherms are of great significance in ph moisture isotherms of polyvinylpyrrolidone (P xypropyl methylcellulose complex. and that each OH group in the MCC co~stitutes adsorption (absorption) site. organic electrolytes (e. (c) bulk. such as exemplified in the forego in^ are carried out on cornstarch. Cases in points and misoprostolhydro
For an organic compound. The purpose here is to derive a rational equation for the rate with which moisture is adsorbed onto a watersoluble solid. 9. riments. This phenomenon has been dealt with by Van c amp en et al.. such as starch. The surface area obtained m any fold larger than the area obtained by nitrogen ad (1972) and Zografi and Kontny (1986) have shown that water penetrates the solid. choline salts).g. an
Compounds of this nature are. organic. unsaturated solution.P)]is plotted versus RH.g. moisture will adsorb onto its surface by the same process as nitrogen in a
Modes of moisture adsorptlon: (a) evacuated solid.5a) and then placed in an atmoif which is lower than the vapor pressure P..3 result.5a. then (see Fig. 9. then results of the type shown in Tables 9.
aceutics.
. at RH values above the criticalrelative (d) humidity.
In the case of deliquescence. then the vapor pressure P.
w' = n / s
(9.6 it follows that. the solution formed is not saturated. a bulk layer o~moisture thermodynamically feasible. however. once the vapor pressure in the atmosphere Pa equals then the condition exists in which that of the saturated solution of the compound Ps. is If P.5c. l brium.
P. This quantity is very significant because. A consequence ofthis is that if a solid is placedin an atmosp~ere then con~ensationwill continue ad i n ~ n i t u m ~ the volumeof the vapor phaseis if infinite.11)
. continue until the concenthat o f tration of the now unsaturated solution is such that its vapor pressure ma~ches the atmosphere (Le..e.d).10)
If the solution is considered ideal. but will have a con cent ratio^ of n/w' (i. it is worthwhileto consider a more realistic situation forwhich these principles apply The following nomenclature will be used in the following:
*
6' = volume of vapor phase . It will. > Ps and the volume of the atmosphere is infinite. when the subject is the stability of drugs in exposure to moist atmoof spheres. ecause vaporphase volumes of infinity are not within the realm of the sible. 9. will be
P. then water will condense onto the solid..). and (b) where there is sufficient moisture in the vapor to dissolve all the drug and form an unsaturated solution (deli~~escence). this is the a ~ o u n t the sorbed bulk moisture layerand this dictate^ the expected stability rate of the drug s~bstance. until it is P.owever (seeFig. a phenomenon known as deliquescence. P. his will continue until all the solid has dissolved. because in this case nature will attempt to establish the equilibrium that exists at a concentration of solute of zero (pure water). = Po(1 .9)
After condensation the vapor pressure will be Ps.. after equilibriu~ has set in. with insufficient moisture for deliquescence.x ) = P(){w'/(n w')} +
(9. because nature will attempt to establish equili~ fRH. 9. ith the cited nomenclature and the situation depicted in Fig. the mole fraction in solution will be
x = {n/w')/{1 + (n/w')>= n/(n + w')
(9. and the s ution formed willbe saturated (because it is in equilibrium with a solid phase).
water vapor pressure of vapor phase before condensation water vapor pressure of vapor phase after condensation Po = water's vapor pressure at a temperature of T R = ideal gas constant 77 = absolute temperature w' = moles of water condensed s = solubility of compound in moles/mole of water n = number of moles of solid dissolved
= =
e now distinguish between two situations: (a) one where there is insufficient water present in the vapor phase to dissolveall the drug (and a socalled b ~ l moisture layer is formed). the amount of moisture condensed is such that the amount of solid dissolved is given by w's = n.
at time t (Carstensen.. A further assumption is that the amount of moisture adsorbed does not. (9.
Here. Such a sample would consist of N particles where: NTE= Npndo3/6 == 1
(9. it will be assumed that the foregoing situation (a) exists (i. the MUR curve). At time t. w grams dissolved per particle). as it is the number of moles of solid (all of which have dissolved). w (grams) has been adsorbed by a particular solid particle weighing w1 grams and of diameter do.V ~ V a~ u ~ a l e V Vapor Pressure P a
~
Vapor V o ~ V u ~ ~ Vapor Pressure A f t e r Cond. and can be solved. that the condensed water will dissolve solid). and it willbe assumed that the sorbed solution is saturated at all times. and the thickness h of the moisture layer.12)
Eqs.
.e. If there are W grams of water adsorbed by 1 g of solid (i..11) and (9. It is assumed in the following that 1 g of solid is studied and that the sample is ~onodisperse. n is known.
In the following. change the vapor pressure in the atmosphere surrounding the solid particles. 1986). and the diameter d of the solid itself will have decreased its original value. = PX n Moles of Mater
C~nd~ns~
Schematic for moisture adsorption on a solid in a closed container.e. at which the subscript denotes the condition before moisture adsorption. to a great extent.13)
The amount of solid present at time t is given by the original amount less the amount dissolved. Assume that. Px is related to w’by the ideal gas relation
(Pa P x ) ~ w f ~ ~ =/
(9. what sort of curve might be expected for the extent of moisture uptake withtime (the moisture uptake rate curve. The diameter from of the ensemble D is the sum of the diameter of the remaining solid.12) constitute two equations with two unknowns (Px and wf). a certain amount of moisture. some solid will have dissolved. then. The question is. moisture will have adsorbed.
and under ideal conditions it would be (9. . inthe ensemble particle.20)
e =p * ~ ( ~ l 6 )
quat ti on (9.I .X.16) where p* is the density of the adsorbed liquid.
(9.P.P. + FSW
where
I.18) may be written:
or
D = {[F+ (1 . by a mass transfer coefficient k .WS)/p]= QD3 . are the volume fractions of liquidand solid.15)
The volume of liquid adsorbed by one solid particle has a volume ofthe total particle minus the solid particle.ist
~ ( ~ 7 t / 6 )= N(m . Because it is assumed that it is always saturated. ) the difference between the water vapor pressure P in the atmosphere and the vapor pressure P.P
w = p*N(7t/6)D3. that is.Q/(l .23)
E = {I.26)
where Eq.[p*(l .FAY) w]/Q}1/3
The area a of the particle (solid plus liquid) is..W S ) / ( N ( p ~ / 6 )
(9. this gradient is a constant. it is timeindependent.22)
a = 7 t { [ ~ (1 .WS d3 where S is solubility in gram/gram. (9.)B[E
+ v2/3
(9. hence. Therefore:
(9.25)
The rateof condensation ( d w / d ~ is proportional tothe pressure gradient (Le. = P * l P
(9. so that we may write: dW/dt = ka(P. respectively.) = k(P. where (1.FAY)}2/3
(9.14)
d3 = (1 . This may be written:
. over a saturated solution).)Po
+ X.It follows from Eq.18) (9.17) that the amount of moisture adsorbed per gram can be expressed in terms of diameters as follows:
P* = (1 . (9.17) . (9. and po and p are the respective densities.x) and x.wS) = 1 .F A Y ) W J / Q }= / ~ i w ~ B[E where
+
]~/~
(9. It is also proportional to the surface area a.19) (9. . At a given atmospheric milieu.23) has been used for the last step.
will not form a “bulksorbed” layer).9. It is noted that there is a linear section of the curve (up to 6 days) as shown in Fig.(1980) studied the moisture pickup in e a vacuum system by usinga Cahn balance.) pe~
. 9.P. then it ght pick up moisture at a given rate.. and at 80% at an evenhigher rate.
here are humidities below which a solid will not adsorb (considerable amounts) of moisture (Le. and exposing the evacuated head space to relative humidities created by salt baths. The slope of this linear ) segment is the moisture uptake rate ( ~ ~ RThe . The rate with which it picks up moisture is determined by weighing the sample at given intervals.3.28)
3 6 = k(P .29) can be solved by iteration. The uptake rates can be simply obtained by weighing the sample after a given time (6 days). actual uptake rates (determined from the linear portions) are shown in Table 9. 9. An example of their results obtained by the latter method is is shown in Fig.dW/{[E where
+
= 3Gdt
(9.27) integrates to
where the initial conditions that W = 0 at t = 0 has been imposed.)B quation (9. at 50% RH at a higher rate. 1980. They also reported moisture uptake rates of choline chloride at room temperature and different relative humidities using a desiccator method. but in such a case it is assumed that the moisture uptake is still in the
100
80
60
40
20
0
0
300100
200
Hours
Data for choline chloride rnolsture adsorption at 100% V a n ~ a ~ et al.~ a n C a ~ p et nal.27) (9. As already covered to some extent.4.7. as demonstrated in Table 9..8. these are dictated by the solubility of the compound. ~akobsen al. (1997) et scussed this situation for three drug substances. ~ ~ u a t i o n (9. If the RH were 30%. 9. uppose a solid with a high solubility is placed in a room of a given shown in Fig. As an example of this.
3.
Moisture Uptake of a Highly ~ a t e r .~ o l u bCompound at 50% R le stored Days
2 6 38 36 €00 144 288
at 50% RH
Moisture pickup
(mg/g)
0.3
.5 1.0 4.25 3.4 3. of
0
100
200 300 Days at 50%RH
400
~oisture uptake data from Table 9.2 4.5 2.Mecha~ism moisture uptake.
and tbere are two types of curves that occur when equili~rium values are plotted against RH: salt pairs and continuous adsorption.45
It is known as the cri~ical o i s t u r e o n t e ~ and the critical relati~e ~ c t humidity.. As shown ig.10 and the deduced value of the critical relative humidity RH*. this.06264
+ O. hydrates. different amounts of water
RW
y = 0. A hydrate is ds defined as a ~ h e ~ i cc ol~ p o u n d a with a rational ratio of water to anhy~rous o ~ p o ~ n d c at d ~ f e r e n t tem~eratures.
function
of
RH.10. 9. for a nonhydrateforming compound.1 0.g. the mole fraction at saturation y may be calculated from
*=ly
(9. In some cases (e. for most salts intersect at much higher relative humidities. 9.
oisture
uptake rate as a with R2 = 0. It is visualized that the water molecules occupy definitive positions in the crystal lattice.9.999. especially ionic c o ~ ~ o ~ nform. the moisture level (the weight of the sample) will taper off and plateau at an equilibriu~value.
Leastsquares fit
is
.ter
Moisture Uptake Rate of ~ a t e r . is a rather small of and number.30)
The previous section dealt with the rate with which moisture is taken up. The former will be discussed first. Curves. and the reverse procedure may be used. such as the one shown in Fig. 9.~ o l uCompound ~le
25 50 80
0. ~ontmorillonite). at longer time periods. This equilibrium value is also a function of RH.25 0. Many co~pounds. namely. Because the value is related to the solubil~ty the compoun~. on a mole fraction scale. the saturated solution is often (nearly) ideal.O06374x. from plots such as that in Fig.
PH20
1
P
U
E
H
H20
cuso4
Mole Fra~tion CuSO4 >
Vapor pressure profile for CuS04 hydrates. they are presented with reference to original works. as for instance in magnesiumchloride. 9.
The vapor pressure of a salt hydrate as a function of “composition’’ will be referred to in the following simply as the vapor pressure pro~Ze the h y ~ r ~ tSome of the of e. concepts to bediscussed are quite ancient. This is an i n t e ~ e d i a t e case. the water molecules are part of the coordination shell of an ion. primarily because a u t o ~ a t e d literature searches usually do not go farther back than 1970. Occasionally. a dihydrate and a tetrahydrate. and hydrates usually have stepwise profiles when equilibrium vapor pressure is plotted versus composition. shows no inclination to be stepwise in nature).. The cases that will be discussed in the following are of the type for which an anhydrate can be produced by heating or by vacuum.1 1). and the crystal spacings between the layers of aluminum magnesium silicates increase in proportion. There is a great deal of rejnve~tjon in the field. 1943. but since t h e ~ o d y n a ~ i c s not a are function of the calendar time at which they were formulated. There are different conventions for presentation and the one proposed by rransted (1943) is as follows (Fig. but the curve is continuous (Le. Heating magnesium chloride tetrahydrate to 80“ 100°C will remove two of the molecules of water. among others. ut further heating results in the removal of 2 mol of hydrochloric acid. which existsas.)
. leaving magnesium hydroxide behind. The question is then whether this water sits “in a channel” (as appa~ently it does in montmo~llonite)or is bound in a different manner. The anhydrate can be produced by interaction between metallic magnesium and hydrochloric acid gas.may be adsorbed or absorbed. (Data from Brmnsted.
such as lactose ~ o n o ~ y d r aand p o t a s s i u ~ te tartrate ract with moisture to give a hydrate. the xhydrate will start formi~g.9. only one hydrate form exists. This diagram applies to one particular temperature and when the water activity (a = Hiloo) is low. and the solid phase is the pentah~drate. only the . (9. pure water’s vapor pressure.32) The equilibrium constant is given by K = P&. At a given value ofa . the compound (copper sulfate) was capable of existing in a crystalline anhydrate and several hydrate forms.12. salt is added to form unsaturated solution (u). and starting at point P.rmnsted (1943). considers water the left ordinate axis. then the interchange in
In the situation depicted in the previous sections. Frequently. say an xhydrate.* (9. suf~cient If copper sulfate is added to point .
. a lhowever. At point A the aqueous phase is saturated with copper sulfate. anhydrate exists. as opposed to convention nowadays.33)
The general situation is depicted in Fig.
the vapor pressure is equal to the vapor pressure of a saturated solution. hence. (9.13 (potassium tartrate dihydrate) that when A H is fairly temperatureinde~endent.35)
F is here water vapor pressure.P diagram for a compound that forms (only) a monohydrate. It is seen in Fig. x is the number of moles of hydrate water per mole of compo~nd. however. The situation.g. will be as shown in the C E part of the curve. so water will start ing. 25°C)
A
X+Y
X
X . From Eq. Increasing the vapor pressure to between al and a2 will not (but will cause some surface adsorption). A H is the enthalpy involved with one ~ o l e c u l e of water.2
Constant Temperature
(e. then loginverse t e ~ ~ p e r aplots are linear. At a2.33). q is a and constant.
and withinfinitevolume vapor phase the reaction according toEq. then dilute sol~tions form and will the profile.32)will .
. wher~asthe e~thalpyfor the saturated solution is for the reaction:
(9. 9. and (with an infinite vapor phase volume) adsorption will continue (line il all the salt has just dissolved to create one phase. t~~e The slopes of the vapor pressure over the salt hydrate and the saturated solution differ because the enthalpy from the slope o f the hydrate is for the reaction shown in Eq. 9. if the solutions are ideal.13 neither the salt pair nor saturated solutions have a vapor pressure of 760 (1 atm) until a temperature of 115°C is reached.5). (9.33) it follows that
which. e vapor pressure is further increased. as In Fi . i A H is te~peraturein~~pendent be integrated to f can ln[P] = .~ A H / ~ ~ ~ } (q l / ~ )
+
(9. (9.36) ~ a p o pressures over salt hydrate pairs also increase with te~perature r ( 9. is that the vapor pressure of the xhydrate will increase with temper~ture will the vapor pressure of a saturatedsolution of the c ~ m p o u n ~ .
For a monohydrate as depicted in Fig. This is denoted the critical t e ~ ~ e ~ and at ~ e a t ~ temperatures above this temperature only the anhydrate is thermodynamically stable.12 wouldapply at any temperature in this region.18 27.4
2.13.13) whenthe solubility increases with temperature its vapor pressure most often increases more rapidly for the hydrate than for the saturated solution.975 . 5 6 5 2 7 ~ R2 . 9.4 .4
3. 9.9742 .0.7HzO + 5HzOa 2.14.6
2.66 8.)
AT
In the situation shown in Fig.8
3.6
1000/T
3 Vapor pressure data and solubility data of KzC4H406.62120 + 4H20b 1. there will often be the case(see Fig. Therefore.4
0 15 20 25 30
Source: aPartington and Winterton (1930).6H20 + SrC12.05 SrC12.93 19. and (as seen in Fig. bBaxter and Lansing (1920). 19.y
y
.0
3. (Data from Krack. 8 7 3 0 ~ R2 . 9.14)wherethe xhydrate achieves the same vapor pressure as the saturated solution vapor pressure and ternperature.95 12.0 8. 1998. the hydrate stays as such in the entire temperature range of 101 15".1. and the general shape of the diagram in Fig. 9.the increase in vapor pressure over the salt hydrate will increase drastically once more than 1 mol of water is present per
Influence of Temperature on Vapor Pressure( m ~ H of Salt Pairs ~ ) ~ e ~ p e r a t u ("C) re NaHP04.12Hz0 + NaHP04.199 . The vapor pressure of both salt hydrate and saturated solution increases with temperature.000
0
h[sI
1nlPI
0
2.0 .2
3.
. 9.6.23 4.
because the section PA is simply a straight line. Above T3 the salt would have a higher vapor pressure than the saturated solution. (Data from Carstensen. and if a diagram as this ised out to 100°/~RH.38) + 1)
(9.39) H the system in equilibrium is infinite dilution (pure water). and this will continue until all is dissolved. It is the critica2 temperature for the hydrate. If a saturated solution consisted of 1 mol of salt and y mol of water. The data presentation in mole fraction (see Fig. then a sharply increasing curve should result at very high
.12. The shape of this curve is as shown in the curve in the right hand section (CDE) of Fig. 9. if it were ideal. 1986. The diagram in Fig. but this is thermodynamically untenable. z + 00.11) is simpler. At the temperature T3 the line for the salt pair has “caught up” with that o f the saturated solution. 9. then adding z mol of water to 1 mol of solid would give a mole fraction of
a = P/Po = 1 .37) At higher relative humiditi~s.water will condense on the solid to form an unsaturated solution.a)
as a ”+ 1. Moisture will then keep on condensing and converting the monohydrate to saturatedsolution. The line at point A has been drawn slightly to the left for graphic clarity. After that the vapor pressure will increase so that it is always in equilibrium with the concentration in the (now) unsaturated solution. and T3is simply the highest temperature (and a triple point) at which the monohydrate exists. would be (9.[l/(z
so that
+ l)] = z / ( z (9. If this is ideal. It occurs at 1 mol of water per mole of solid.39)
z = a/(l . 9. then vapor pressure of the saturated solution. It follows from Eq.14 is at a given tem~erature and shows a compound capable of forming a monohydrate at different temperatures. (9.)
mole of solid. if the solution is ideal.F
E
E
0
1
2
Mole W ~ t ~ r / ~ Solid ole
Single salt pair (monohydrate)vapor pressures as a function of temperatur~.
easeB
30 5 4 0 0
GO
70
Temperature.1s). in the second the mnhydrate vapor pressure reaches the vapor pressure of the saturated solution above the boiling point of water. Such a case has been reported by Chen and Grant (1998) for nedocrornil the sodium trihydrate/monohydrate system. 9. between mn and n0. 9. for m/O.15. is above the critical tem~erature . 9. he correct profile for this situation is depicted ~raphicallyas case 9. in Fig. if there is a tra~sition between two salt pairs there will be an intersection between. there are two possible situations. This is not feasible thermodyna~ically. 9. Note the nicks in the saturation curve at points E and L. 9. The other (hypothetical) case is where the mform's critical temperature ( N in case A in Fig.16. of the nform (case A.
Case A
\
n
0
7
10
20
Sat'd
Soln. an mn. A diagram similar to that of Fig. and schematically in Fig. mn and n0.so that the critical temperature for the mhydrate in this case would have to be the point at which it reaches the vapor pressure of saturated solution. 9. * C
Water vapor pressure diagrams of a compound forming two hydrates. and above the transitio the stable specieswillbe ( ed. an n0 pair (m n). and above the transition the stable specieswillbe
. (From ample. hypothetical case the mn vapor pressure curve crosses that of the saturated solution (denoted s) at T*. figure).19. when there is a transition between two salt pairs there will be an interS . T" ften. and the nick in tion vapor pressure curve(s) is absent. 1943). One is that the mform's critical temper~ture in case (IC is belowthat of the nform's critical temperature ( L in case B in Fig.and above this te~perature would (if it existed) give rise it to a situation in which the vapor pressure of the msalt would be higher than that of the saturated solution. In the M latter. Noteworthy are the nicks in the solubility curve at tem~eratures and N and at X and L.If there are two salt pairs. IS).16. 1 of that publication fails to show three plateaus. Often (but not shown in the .1isshown for cases A and Fig. In the first here is a conversion below the boiling point of water.
for example. It is seen in the table (and from Fig. not using molar concentration and content units). The moisture contents for the remaining hydrates are shown in Table 9. the dihydrate. and 12) aside from a crystalline anhydrate. hence. the moisture percentage is 100 x (36/178) = 20%.6.6. 9. the vapor pressure the of nick in the vaporpressure curve at T".17) that the relative h~midityof the at~osphereabove a mixture of anhydrous disodium hydrogen phosphate and the dihydrate is g 100 x (9/24) = 38% RH. There would be a increases. Any or of the anhydrous salt and the rate will give this relative h u ~ i d i t y . when its vapor pressure would equal that of the saturated solution (Fig. 9.e. accordingly.x diagram of a single hydrate (the monohydrate.
An example of this situation has already been described. (b) As temper~ture monohydrate will reach that of the dihydrate at a critical temperature. it existed. The numerical data ( ~ a ~ H ~ shown in Table 9.T = 25°C
T = T*
P
X
X
\
Case A
(a)
o/ 1 /2
012
Thereare two distinct situations hypothetically possible: (i) as temperature increases. the percentage of moisture in the dihydrate. is calculated as follows: disodium hydrogen phosphate has a molecular weightof 142. At T > T* the diagram would sunply bea P . would have a vapor pressure if above that of the dihydrat~.26(e)).x diagram will be that typical of a single hydrate (i.16(d)). In the usual presentation mode (i...16(b)). for the monohydrate (Fig. form three hydrates (2. The solid phase in equilibrium with saturated solution would be the monohydrate. The compound can are ~ ~ ) . the dihydrate has a molecular weight of 142 + 36 = 178. 9. the dihydrate vapor pressurewould reach its critical point T". which is not possible. 9.e. in Fig. 9. albeit it not indebil.16(c)). 7. T". Fig. the ~onohydrate. 9.11.Fig. Above T * . so at T > T" the P . It is noted that there is a nick in the vapor pressure curve. disodium hydrogen phosphate containing between 0 and 20% moisture will have
.
)
.5 salt)
Source: Maron and Prutton (1965). then the dodecahydrate isconverted to saturated solution.is obvious that the heats of solution of two hydrate forms e would be different. This is
M
U .~haracteristicsof Disodium Hydrogen Phosphate Moisture in solid (YO) Anhydrous Dihydrate Heptahydrate Dodecahydrat~ Saturated solution g (100 g ~ater/4. it will not pick up moisture (or will pick up only surface moisture). and this causes a different slope of the solubility curve. then it will start picking up substantial quantities of moisture until it has completely converted into the heptahydrate. Two further points need to be mentioned: (a) If disodium hydrogen phosphate is stored at an RHbetween 38 and 58%. In the solu~ilit~ there is a l w a ~ a nick in the curve at t h e ~ o i nwhere there is plots s t a c~iticaZt e ~ p e r a t u ~It . as shown in the table. of (Data from Brrirnsted. the heptahydrate contains 47% moisture.75
above it an atmosphere of 38% RH.1928. (b) If the relative humidity is raised to (slightly above) 92% RH. At higher RH values the equilibrium willbe dictated by the water vapor pressure over the now unsaturated solution.38 0.
pHz 0
Water activity (RH/lOO)
0 20 47 60
0.58 0. Once the relative humidity is raisedto (slightly above) 58%.92
Pair Pair Pair Pair
9 14 18 22
0. Similarly. and mixtures of diand heptahydrate give riseto water vapor pressures of 14 mmHg (58% RH). Similar plateaus exist for heptahydrate and dodecahydrate.r(
rrl
x
Dodecahydrate Dodecahydrateb Heptahydrate Dihydrate Anhydrate
Temp "C
7 Solubilty of Na2HP04 hydrates as a function temperature.
that solubility does not always increase with temperature. 9. The situation is shown for sodium carbonatedecahydrate and ~heptahydrate where both vapor pressure and solubilities are listed.and decahydrates. because
. AG is negative in going from the downcurve to the upcurve. and that these may be ofa ET nature. exhibit watervapor interactions that give rise to smooth (not stepwise) isotherms.18. It is seen here (for the solubility curve of the anhydrate at higher temperatures). There are situations for which A H is not constant with temperature. 1928. PP denotes vapor pressurecurves in centimeters mercury. This pheno~enon known as hysteresis. The is ordinates will be denoted simply as x and y in the following.19 will result. In general. however. It is seen that the transiti~n te~perature (the critica~ te~perature) is apparent (32°C) fro^ both types o curve. If such a substance is evacuated and allowed to adsorb moistsure up to a water activity closeto unity.Decahydrate bHeptahydrate PPDecahydrateI 1 PPHeptahydrate/
0
0
10
20
30
Solubility and vapor pressure data of sodium carbonate hepta. Obviously. (Data from Kracek. then a curve will result that is different from the adsorption curve (a): desorption curve (b). In this case the heat of solution is of a different sign so the solubility decreases with temperature. It is noted that yd is not anequilibrium condition. but in solubility plots it often is. Solubility data (grams of solid per 100 g of water). f
It was mentioned earlier that compounds. then a curve such as (a) in Fig. such as gelatin.)
exemplified for the disodium phosphate system in Fig. If the pressure is reduced again. It is necessary to express the solubility in moles of solute per mole of solvent. there is a correlation between water vapor pressure and solubility of hydrates. 9.
is not applicable to most realistic pharmaceutical conditions. therefore. 1980) is to approximate them h isotherms (Le.6
0.1. they have will different equilibrium There can.41)
Usually. be only one RH condition in the pore space of the form.8
1. 9.g.20). in pharma~utical engineering literature.. a tablet) is usuallymade to a given moisture content (e. r routine isotherms. not use the high end portion at all).42) The high RH tail of the curve is usually above 85% R and. that is.19.9.. In excipient study by ~angvekar (1974).0
0. so the result is that compound b will pick up moisture (move from
0)
0
'ablet
0)
0
10
20
30
Water Vapor Pressure (Torr)
Fr~undlichmoisture isotherm presentationof initial part of a BET isotherm..Because the drug and the exci ients have ~ifferentmoisture isotherms. (9.8g/100 g ofsolid.2
Water Activity (RHll 00)
ET adsorption and desorption moisture isotherm.
. the high relative humidity tail is difficult to obtain with precision. but it is applicable to one cted test (46C.4
Down Curve (b)
0. dosage form (e.g.Fig. when all the data are lumped together. 75%RW).0
1. and one approach (Carstensen.
everal common tablet excipients give riseto Langmuir isotherms.4
0.2 0. they follow an equation of the type:
1/y = ( A l p ) B
+
(9. however. the moisture equilibrium and curves are shown in a sense opposite that shown in Fig.
If we consult assume that the up curve isthat of drug (A) and the down curve that of excipient basis.44) and (9.compound a will lose moisture (moving from A to C). as seen from the figure. cause.45) The values of yz usually do not differ much (and the two isotherms. it is made at a particular moisture content of a fraction y basis) of q moisture (i. a total of m A q A grams of water). where would both. mass balance (assuming no loss of moisture) gives:
YC[~A
+ ~ B =I Y A ~ A Y B ~ B +
(9. The areas have not been taken into account.e. and A contains a fraction (on a there are m A grams of A on an ~nhydrous basis) ofq A moisture (i. This may be used to estimate the moisture movement ina solid dosage form after it is manufactured. therefore. then P is also known. and the isotherms apply to two samples of material (to account for the area. This can be verified byinspection of Fig. In the situation for which a known amount of A m~ is mixed with a known amount of B mB. and as the dosage form (e. plotting by ET would have to be done). fairly well. [ (9. 1980).e. Two moisture e~uilibrium rvesmay(in an abbreviated fashion) be repreFreundlich isotherms.. can be represented as differing only in the values ofthe e's). quantitatively..47)
and the amount of moisture lost can then be gauged from oisture loss in and for (9.48)
As y C is known. therefore. The dry weight of the dosage form. adhere to a Freundlich isotherm.19. is m A mB. tablet) is made. 9. it follows that A must giv (from point A to point C) and must take up moisture (from point
+
The moisture isotherms are of the type (9.g.yc)
(9.e. the relative humidity (the vapor pressure P) in the pore space must be one particular ure (P). There are mB grams o on an anhydrous basis. a total of mq = [mB+ mA]q grams of water)... The question is to estimate.46)
or:
YC
CYA~A YB~B)/[~A
+
+m ~ l
.43)]:
where q is a constant (Carstensen.49) A
= mAbA
(9.
. where (at what R ) the line DC will be. a total of mBqB grams of water). and contains a fraction (on a dry basis) of moisture (i.
O = water vapor pressure
.
What has been said inthe foregoing also applies the situation in which solvates to are formed. except is now the vapor pressure of the solvent. and the final relative humidity may be calculated).Ps)B/3 EST = enthalpy 15: = equilibrium constant k = mass transfer rate constant m = mass of one solid particle N = number of particles n = number of molecules (a) adsorbed. ethanol.50)
must equal the moisture gained by B: (9. that is of importance.Fs}]2/3 h = thickness of adsorbed moisture layer 1“ = P*/P G = k(P ..ter
If for instance the two compounds aremixed together. Mass balance about AC 9. (b) dissolved n. or other) occupy sites in much same the fashion as water occupies sites in hydrates. = number of molecules in a monolayer Po = water’s vapor pressure at a temperature of I ‘ Pa = water vapor pressure of vapor phase before condensation Ps = vapor pressure over a saturated solution PH.1”s}/Q]2/3 BET = Brunauer. so that P[= P C = PD]can be calculated(Le. Emmett. moisture added (as in a granulation). both moisture losses and gains. In this latter case. In these. the isotherms should be determined on samples that have been wettedand dried the same way the final mix has been wetted and dried (because the surface area changes).20 then gives that the moisture loss experienced by A
(9. solvent(methanol. it not of water. and what has been said about vapor pressures also applies in this case.51) All quantities are known. and Teller c = BET constant d = diameter of particle after condensation do = diameter of particle before condensation I) = diameter of particle plus condensed water E = [FQ/{l.
A
=
general symbol for a hydrateforming compound
a = area of solid plus condensed liquid
B = x[(l . then xc is known. and this is dried.
Maron SH. p 373. J Am Chem Soc 45:419. (b) symbol for number of moles of water in a hydrate x. In: Washburn EW. J P h a m Sci 76548. Int J Pharm 156:67. (b) volume of vapor phase Vm = volume (STP) of adsorbate in a monolayer w r = moles of water condensed w = weight of water adsorbed per particle W = weight of water dissolved per gram of solid x = (a) mole fraction. . e T* = critical temperature for a hydrate t = time q = general symbol for a constant V = (a) volume adsorbed. vol4. vol 3. P h a m Res 3: 187. ed. Int J Pharm 1825. (b) mole fraction at saturation.
.Px == water vapor pressure of vapor phase after condensation Q =p * ~ ~ ~ 6 R = gas constant RH = relative humidity s or S = solubility of a compound in rnoleslmole of water STP = standard pressure and temperature T = absolute t ~ ~ e r a t u rK . Buur A (1997). P h a m Dev Techno1 4:81. Copenhagen. Prutton CF (1965). Mc~rawHill. (c) moles of water per mole of salt at saturation p = density of liquid p = density of solid o p* = density of solid plus adsorbate
Baxter AB. McGrawHill. Zografi CyHancock P (1993). Kontny M (1986). Uchiyama M (1987). Nieman H. pp 181185. M a c ~ i l l a n New York. International Critical Tables. Medhizadeh N. 4th ed. Frokjaer S. Chen LR. Larsen C. Munksgaard. Grant DJW (1998). Carstensen JT (1980). Lansing CD (1920). ed. Bray MI. New York. New York. (1999). Fairbrother JE (1984). Vancampen L. Fysisk Kemi. Partington AB. Jakobsen DF. Principles of Physical Chemistry. Carstensen JT (1986). Zografi 6. Zografi G. International Critical Tables. In: Washburn EW. Brransted JN (1928). Int J P h a m 10:1263. Carstensen JT. Yoshioka S. p 23'7. Pharrn Dev Technol 4:487. = volume fraction of solid in solution z = moles of water per mole of solid for an unsaturated solution y = (a) amount adsorbed. p 253. Danjo IS. Pharm Technol 9 (Sept):4l. Grant DJW.Chow AHL. Kracek FC (1928). Bransted JN (1943). Int J Pharm 5:l. Winterton CD (1930). J Chem Soc 132:635.
This Page Intentionally Left Blank
.
he actual synthesis of drugs is not the subject of this text.2 10.6 10.5 10. but there are aspects at have a direct bearing on further development.3 10. and they are only important before the mar~eting a drug product). is the purification of the raw chemical that is of importan~e. what chemical form ~ ~ ~ Zfor instance) is the one that should be t ~ o r ~ pursued? What r e c r y s t ~ l Z i z ~ t~ o ~ e should be used? Will these decisions have an i ~ i ~ ~ in~uence p o l ~ ~ ~ ror ~~y g r o~~ c o ~ i c ~ ~ y ? on p i s
. therefore. there are a series of problems that are encountered and decisions that have to be made.10. a~ticularly. the first step in development and discovery. it First of all.8 ehydration Kinetics ~ymbols ferences 10.1 Salt Selection 10.4 10. of The syntheses of the drug is.7 10.9 Solvates Change Precipitation cation by Use of Mixed Solvent Technique
160 160 161 162 163 164 165 167 167 168 168
It goes without saying that the drug substance is the most important part of a pharmaceutical soliddosage form (except for placebos. and once a company decides to proceed with the development of a drug.
described by Sherman (1996a. patent 5. for instance. most often. for drugs that contain carboxylic acids or those with an amine group. but excessive solubility may be a drawback.0159 g would precipitate out.b) in U. and hygroscopic. 1990) where this type of approach is described. In general. 2. a sodium salt. Clavulanic acid. S. and the pH then lowered to below 4. amine hydrochlorides. then a change in pH may be used for purificatio~. Most salts of acids have higher solubility than the free acids. such as its m ~ c ~ i ~ ~ ~ i Z whetherit is easilymade into (i. It rnay also give rise to highly viscous. This
.1
81
These are problems thatboth the innovatorand the generics encounter. for example) and the reasons for using certain salts rather than the corresponding free bases or acids.
urification is the final step of drug substance synthesis. drugs that have ionizable groups are prepared as either sodium or potassium salts. lubility usually results in bioavailability that is better than (or at least equal to) that of a lesssoluble form. Examples of this are the sodium il. An example of a compound that might be purified in this fashion is naproxen. but excessive solubility causes higher hygroscopicity.0159 mg/cm3 and sodium naproxen has a solubility at 25°C of196. alts most often crystallize more easily. saturated solutions. example. as specific salts (sodium salts. i t ~ tablets or capsules) should not be impaired. The precipitation may be accomplished in several ways. hence. a free base or. igh solubility is usually desired. include the following: 1.0. which is a Blactamase inhibitor used in ~ u ~ e n t(SKB) is an oil. (The salt is also highly soluble.780 and in et al. for is ~ ~ ~ . Other aspects. a point that was discussed in the previous chapter).e. and in this manner may impair the rates of solubility. In such drugs a method of approaching formulation may be to employ the acid addition compound of the drug (the claimed substance) and neutralize it during wet granulation with sodium carbonateor sodium bicarbonate.
Drugs with ionizable functional groups are produced. The sodium salt of amphoteric compounds are quite soluble. The base or acid may be an oil. For ~ ~ ~ ~ o t e ~ i eom~ othere ~ the possibility of having either an addition salt. because it is of importance to place specifications on the physical state of the drug substance.The reaction is then brought to completion. If the compound is a protolytic substance in solution.. If. in but its potassium salt is well defined. 1990)... it has a low critical humidity. and the tablet made. an addition salt. where free naproxen is the primary species. 200 g of naproxen were added to approximately 1000 g of water at 25°C and made alkaline to dissolve it.such as a hydrochloride rnay be used. Thermal recrystallization and p~ecipitation are most common methods of achieving purity. ~ublimation the is resorted to at times. but is not commonly used.7 mg/cm3 in water at a pH of about 8 (Gu et al. because it has a solubility of 0. then 200 .573.
Adequate conditions (amount of water or other solvent.. Residual solvent is removed by drying by heat or by vacuum (or by both). In this manner it is possible to assess the severity of adsorption of different solvents and impurities. and if this impurity had a solubility in excess of 2 g/lOOO g of water at p 4. for instance. from containing 1% of impurity.8 g of purer drug substance. so that the drug substance had been made purer (Le. The attainment of zero percent impurity by any form of precipitation method is ideal. and 4000 g water then added. the ease of removal of solvent would depend on the equilibrium vapor pressure (at the drying temperature) of the solvent over the solid. 1 g/lOOO g of water at pH 4. as well as occlusion. Microwave drying would work only if the energy frequency was adjusted to the particular solvent. then 198 g of naproxen would precipitate and. 10. 1% of impurity. but is soluble in isopropanol to the extent of 3. If 3.ru
would be freed ofany impurity that had a solubility higher than the final conditions would dictate.9 mg/g of isopropanol. Residual solvent is a problem in precipitation purification. therefore. this would be free o the impurity. the amount adsorbed C". from solution is given by ln[C*] = In q
+ n ln[hf]
(10. total of 5000 g of solvent iscapable of dissolvingonly 0.. Adsorption will always occur. but also of the impurities. final pH) maybe arrived at to optimize the purification. Freundlich equation. Vacuum drying
. The purification process is.and most often Freundlich isotherms may be used to estimate gravity of impurity retention by adsorption. the solubility of the drug substance then drops to 0. it now contained only 0. One situation that may arise is that the drug substance forms a solvate.5% impurity). a function not only of the solubility profileof the drug substance. 198 g being naproxen and 1 g being impurity. If the impurity had a solubility of less than 2 g. theoretically. In the example in Fig.9 g of drug substance is dissolved in 1000 of isopropanol. If the naproxen used was not quite pure. but contained.1 g of a drug substance). rather than realistic. The purity obtained willbe a function of the levelof impurities in the compound before reprecipitation and of their solubilities.02 mg/g of (mixed) solvent (i. An example ofthe potential use ofsolubility in mixedsolvents for precipitation purification was published by Joszwiakowski et al. and in this case. Adsorption may be investigated researchwise. so that this water addition would allow 3.1 the drug is soluble in water only to the extent of < 0. then a precipitation by solvent change can be accomplished. (1996). for example.
IX If two solvents are miscible and the drug substance to be purified is soluble in one and poorly soluble in the other. then 199 g of solid would precipitate.1)
where q is a constant andhf is the amount in solution.e.02 mg/g of water.
to a great extent. and an impurity has the solubility ~haracteristi~s shown. adding toit 1000 g of solvent and heating it to 60°C will dissolveit all. is never quite true.e. 6. The best method the of ~‘freeing’’ solvent of this type is by way of comminution.5 g will precipitate out (under ideal.‘ assumethat a compoundis soluble to the extent shown in Table 10.Thismay happen when hard vacuumis e ~ p l o y e and in such a case. theoretically. or (b) make them sufficiently mobile allow escape of to the The third situation is that surface removal of solvent forms an impenetrable crust. 9. ~ s s u m that a particular batch of the drug s u b s ~ contains 2% of impurity e n~ (i.e. sidual solvent may also be the result of solvent ~ntrapped crystals as they in te (i. longer drying time.e.5
10
2
.0
20 40 60 80 Weight Percent Isopropanol
100
Example of ~re~ipitation puri~ca~iont e r / i ~ o p r o ~ a n o ~ . the solvent may occupy defect sites in crystal). 98% of drug substance). be “pure.2 g of (0.. because the milling may (a) expose the defect sites.5
25 60
0. an example..
Solubilities of a Drug Substance and an Impurity Solubility of drug (g/lOOO g of solvent) Solubility of i ~ p u r i t y (g/lOOO g of solvent)
0. not zero. been covered in Chap. are always finite. Taking 10 g of the batch (i. but some c ~ m ~ e n t this point at s As are of import~nce. The 0. By cooling it to 25”C. 9. therethis fore. equilibriu~ conditions).8 g of drug substance)..
This has. in ~ a
might be used to bring the pressure below that of the equilibrium vapor pressure of the solvent. g impurity will present a concentration less than sat~ration and the precipitated drug substance will.” ecause of adsorption and possible inclu~ion.1. trapping solvent on the inside.5 g~l000 of solvent). and limits on impurities. it may be corrected by using a lower vacuum and a ~.
0 0.A special case ofpurification is that of optical isomers. generically. the solubilities of (excess amounts of) a mixture of a certain amount of the dform and a certain (not necessarily the same) amount of the Eform would simply (approximately) be the sum of the solubilities of the two.)
. a compound.0
Mole Fraction ‘Iform ~olubilityphase diagram of the p s e u ~ o e ~ h e d ~system. that the solubility plot is exactly the upsidedown inverse of the melting point plot. however. however. Equimolar mixtures of chiral compounds (denoted dl) may.2 that the dlform is less soluble than either of the hiral purity” appears to be mandatory for new drugs. 10. omp pounds with one chiral center may occur as a dform. A means of ..2.to simply describean a~e equimolar composition of the two enantiomers without signifying whether it is a conglomerate or a racemic com~ound.2
0. 10. racemic compound is. is to derivatize the dZcompound with an optically a . or (in a racemic compounds) as a dEform. 1978).2. It is noted in Fig.6 0.Zform the situation is. A situation similar to that of pseudoephedrine exists for daxclamol hydrochloride (Liu and Hurwitz. ine
1995. there is no chemical interaction between dand Eforms) and. different. whencompared with a melting point phase diagram of a molecular compound to be treated shortly. (Rata from Pudipeddi. The expression ~ a c e ~ is often used. and may be conas a strong complex between the two components. exist as racemic compounds or conglomerates. presenceof a d. the dform derivative would have a different solubility that form derivative. and situati~ns makes separation by recrystallization impossible.8
1.e. (denoted enantiomers). as the word implies. in that case.4 0. dipeddi (1995)have reported on the isomersof pseudoe~he hase diagram of this system is shown in Fig. depending on the compound in question. and fractional recrystallization can now be carried out. from Fig. If the latter did not exist. an Eform. 10. then the mixture is a conglomerate (i. The resolved
4
0.
such as zeolites or bentonites. Some pore space drying may occur (e. uniform concentration of water or solvent is denoted co. The manner in which speci~cations set for moisture content is discussed in Chaps 14 and 15 are dealing with stability. more no evaporation is taking place). 1960) is given by (10. uniform concentration is denoted c. requirement for chiral purity seems to be only possible for “grandfather” drugs. it is actually unnecessary (e. This allows drying to be monitored for the end point. The integration requires that A is (fairly) constant. 1960). ~Ztochopherol). The same holds true for amorphous solids. and it adds to the cost of the drug. A high kvalue requires good heat and mass transfer such as in a fluid bed dryer or in spray drying.4)
. then the drying rate. it is filtered or centrifuged to a certain degree of dryness. Equation (10. dm/dt would be dictated by
q’ = dm/dt = kA(P0 . The expression for the average concentration c. costly.q’t
(10. In other cases. but a drying step is usually necessary.. If it is the drying of surface moisture. the solid particle will be considered spherical with a radius of ro. but their moisture is often part of the residual moisture allowed by specifications.enantiomer is then reformed by dederivitization. For crystalline compounds that are not hydrates.if agglomerates are formed). It also applies only to the phase where drying occurs. the pressure at the temperature of the airstream.2). often a necessity because of the toxicity of one enantiomer (the one not wanted). It is. contain internal water that dries by diffusion. and P is the partial vapor pressure of water or the solvent in the drying airstream.. Micropores may also be dried out down to a certain pore size. Drying of such drugs in general follows usual diffusion kinetics (Jost.
After purification of a drug substance by rec~stallization orreprecipitation.k is a masstransfer coefficient. but rather.g. Po is the vapor pressure of water or the solvent at the dew point of the airstream drying the solid.P )
(10.e. which can be expected in most drying conditions.g. and at time t the average concentration is denoted c.2) integrates to
m = mo . (10. and the final. Some dry solids. This is tedious. and yields are meager at times. drying is simply removal of surface moisture. at time t (Jost. however. The initial. because once drying is complete. For simplicity.. for a rise in temperature of the exit airstream indicates that water or solvent has been completely removed (i. t is time.3)
where q is given by Eq.2)
where m i s mass ofwater or othersolvent. A is the surface area of the solid.. P is no longer the pressure at the dew point.
10. Usually. the smaller the particle.9) becomes
1 . (10.)} = . for a cylinder it is 6).7. Overdrying is particularly directed toward drying of hydrates.
. and (c) overdrying.( m (10. oexp(t/z) / ~= ~
+
(10. and partly it occurs as ~ t ~ ~ water. It is noted that the a~proximation gives some zero time deviation from intercept with zero for Eq. Q may vary from 6 to 8.t / ~ } ln[6/n2] where (10. the water in hydrates is partly held in the coordination shell about the ions in the lattice. The e t ~ ~
~
Z
. m. it becomes four times smaller as u increases from 1 to 2).g.6)
1
The term (1/u2)exp[u2n2Dt/ri] decreases drastically with increasing value of 1/u2 (e.5)
so that
(10. and if it is assumed that the final amount of water or solvent. drying curves are separated into the three sections shown in Fig.g.8)
lo)
The left~hand side isthe fraction (or if multiplied by 100. (10. it is necessary to multiply by the volume V . The effect of ~eometry is such that the factor (A/n)romost often is close to unity. depending on the geometry of the solid. This shows that the drying process in a diffusional phase is loglinear in time. Add to that the effect that increasing Uvalues have on the term exp[v2n2Dt/ri] and it is seen that quite an adequate approximation would be
or ln{(co. m.
As mentioned in Chap.2) to mass of water or solvent.where u is a running index and where summation is withu going from 1 to infinity.9) and where. percentage) of moisture the left at time t. Eq..e. (b) drying by diffusion (the fallingrate phase).. in a particle. is a constant that is dependent on geometry (e. but may be expressed by considering that at t = 0. 9. (10. = 0.c)/(co . To convert Eq.3) (Pitlcin and Carstensen. as shall be discussed in the next section.{ .3: (a) drying of surface moisture (the linear drying phase). (10. rather than c~ncentration within a particle. 1973). and shows that it is the more rapid.
0 fails to give up the last two molecules of water when heated above rather gives off HCl.g.5. In the following. for instance (Zoglio et al. 10. 10. creating an anhydrous ionic network. 1 This further nucleates (see Fig. Xray examination of the dried compound has demonstrated that vacuum drying forms a skin possessing no crystalline structure. then. a process that is accelerated by presenceof water.5).
structural water is held much less tightly than the coordination water. rying with airstreams with a relative humidity equaling that of the equilibrium relative humidity of the salt hydrate will remedy that. It is unstable rearranges to a phase that has no welldefined crystallinity (seeFig. phase C). ampicillin. at times. it is the salt hydrate that is the desired form of the drug substance. whichis unstable (except for zeolites). or (b) it is desired to remove the (structural) water of hydration. but on the other hand. the “drying” of the hydrate is considered to be twofold. 1975) forms an. In va it is often advantageous not to employ a hard vacuum. crust during fluid bed drying under some conditions. It is apparent from the previous chapter that at any given temperature below the critical temperature T” of a salt hydrate. For instance.
. but onthe 0 nd. The drying then takes place through a continuous layer ) and a reduced surface (the cracks in C). depending on drying conditions. The finesse. is to stop the drying at the right point. and cephalosporins).Moisture Content A
onstant Rate Period
R@nlOV
able Moisture
D
d Molsture
0
The different drying phases. depending on purpose. This nucleation and crystallization gives cracks at right angles to the interface... as shown in the cited cases (Fig. amo~icillin. ying. adsorbed moisture may be deleterious to the compound as well as d ion of the hydrate. whereas drying in moister atmospheres does not iverise to this phenomenon. In the hard vacuum. drying is more rapid if lower humidities or higher temperatures are used. at times waterimpenetrable. ~acuum drying causes similar crusts to form (Garner. Either it is desirable to remove the surface moisture and not the water of hydration. Often (e. 1953)in the case of copper su~fate pentahydrate and magnesium tartrate dihydrate. moisture evaporates off of the surface. causes undesired effects. there is the possibility of removal of water (or solvent) of hydration.
It is not only water that may become part of the lattice of a compound. Pohlrnan et al. then being able to escape only through narrow cracks. 1981).
.
Taylor and York (1998) studied the dehydration of trehalose dihydrate. (1975).. one talks about solvates. the second one being trigonal (Lowes et ai. Drying is impaired by the moisture 3and first having to penetrate layer € . and such) may also occupy lattice sites. Solvents (ethanol. have shown that at least three polymorphs exist of carbamazepine. 198 1. anhydrous subphase (B) that then crystallizes to a phase (C) containing cracks. and in that case. the first being monoclinic (Reboul et al.0
1
2
3
4
5
(mm Hg)
6
Water Vapor Pressure
Drying rates of CuS045H20as a function of water vapor pressure. methanol. 1981).. for instance. The strucure of a dihydrate
A
B
c
Hard vacuum drying of CuSO4 (A) causes an amorphous. and found that none of the conventional equations would fit the dehyration data well.
Int J Pharm 167:215. J~zwiakowski MJ. Academic Press. Sax M. Pitkin C. Pohlman € . Acta Crystallogr B37:2242. S. Chang CH. Decamp WH (1981).11:709. depending on shape q = constant in the Freundlich equation q' = drying rate constant in the constant rate drying period ro = radius of a spherical particle being dried T* = critical temperature t = time u = running index z = ~ ~ / ( ~ 1unit that reduces drying time to reduced. Cristau B. University of Wisconsin. J Pharm Sci 67:636. S. University of Minnesota. 1983) have also been reported. Nguyen NT. Fromm GH. Minneapolis. Strickley RG.962. Chowhan ZT (1990). Dietz C (1986). Madison. J Pharm Sci 76:744. PhD dissertation.780. Lowes MMJ. Jost W (1960). J Pharm Sci 85193. Cryst Res Techno1 21:1463. Soyfer JC. Lotter AP. Reck G. p 46.573. 3rd printing. Astier JP (1981). Patent 5. P h a ~ a z i e 30H. Carstensen JT (1973). Reboul JP. m~lecular (b) weight y z = mass of water or other solvent in a solid being dried ) mo = orginal mass of water in a solid being dried n = exponent in the Freundlich equation Po = water (or solvent) vapor pressure at dew point P = water (or solvent) vapor pressure Q = coefficient in drying equation. Gases. Sherman BC (1960b). J Pharm Sci 62: 1215. Steng WH. WI. Liquids. U. 3 Culde C. Pharm a~ol ~gy Zoglio MA. Sisco JJ.. Jahn R. Cairfa MR. Spankcak CW (1996). Sherman BC (1960a). Acta Crystallogr B37: 1844. Terrence CF. Yo0 C (1983).eck and Dietz. p 77. New York.
. Taylor LS. U. Diffusion in Solids. 1986) and that of an acetonate (Terrence et al. J Pharm Sci 64:1869. Patent 5. 27:85. Himes VL. Chi LH. PhD dissertation. Mighell AD. Van Der Watt JG (1987).
A = surface area of the solid C* = amount in solution in equilibrium with a solid of concentration M c = average concentration of water (or solvent) in a particle being dried co = value of c at time zero .573. Carstensen JT (1975). Pudipeddi M (1995). Liu S. Hurwitz A (1978). York P (1998). no~~imensional = )) time
Duddu SP (1993). Pharm Res 7:379. Pfeiffer S (1975).c = value of c after drying is complete 1 = diffusion coefficient of water (or solvent) through a solid ) k = mass transfer coefficient M = (a) amount adsorbed (Freundlich e~uation). Gu L. MN.
The Separated Phase: Solid Solutions of the Second Kind
11.7.1.3.10.4. Melts 11.11.12.9. Miscible Melts
11.11.5. Freezing of Ideal Solutions and Ideal Solubility
170 171 172 174 175 176 176 178 179 182 183
11. Solid Solutions of the First Kind
11. Solid Solutions 11. Hydrous Amorphates 11. Lyophilization: Amorphous Cakes 118. Immiscible Melts 11. Coprecipitates
184
184 185
11.13.6. olecular Compounds 11.14.15. Partially Miscible Melts
11.16. Cogrinds
11.
186 186
187
3 87
issolution of Solid Dispersions Symbols References
. Eutectic Diagrams 11.2. Melting Point Depressions and Purity Assessment by the Van Laar Equation
11.
then a diagram such as shown in Fig.g.It is assumed that the separated phases are crystalline. If a plot is made of the mole fraction x. rd This may be rearranged to read In[xl = bg/[RT11
$
{kb/[RTI)
(1 1. If cooling is carried out fromcomposition S.o
v
Mole Fraction
w
Freezing point diagram of a solid and water
.]
(1 1. a may be substituted by x. because an upper temperature (e. and in similar fashion a solution of composition V will precipitate drug at point Q. It is noted (as opposed to conventional eutectic diagrams) the curve is not continued all the way to the right yaxis. J I L ~is then the s t a ~ ~ achemical energy of a solution o f a mole fraction of unity. the curve U Q is denoted the solubility curve of A.3)
00°C
I . Assume that a crystalline substance A is dissolved in water. The condition of equilibrium is that the chemical potentials of the solute in solid and dissolved form. and assume that the two do notform solid solutions.2)
since for an ideal liquid. p. The freezing point trace of water is the section NU and at the other side of point 77 . the mole fraction. a note on ideal solubility relations is in order.1 results.1)
but
kb
=
+ RT ln[a] = + RT In[. then there will be a separation of ice at a temperature of R. and p b are the same. at which one or the other solid phase (ice or drug) is in equilibrium with a solution.efore discussing melting point diagrams. the boiling point of water) is usually indicated (composition W ) . We may write
k =k b s
(11. 11.
1 to the left of U . it gives no x information about what the conta~inant and.0.5) gives rise to the Van Laar equation. and atthe righthand side ofpoint U it is the solubility equation of in the liquid A (e. and it is possible.. t eutecticshave intersections in "the middle.1. hence. it is possible to develop the entire melting point curve as follows: I" is calculated for several of xvalues by using Eq. For instance if AN equaled 8000 cal/mol and T0were 2OO"C.6) may be (1 written:
(11. at constant pressure and te
{ ln[x1/3 I " } ~= L / R T ~ 3
(11. achieve this is shown in Table 11. in its entirety. the separating phase is water. for instance A H = 8200 and To = 190 givesthe results in Table 11
A similar program may be written for the righthand side of the diagram (by s stituting (1 . with the terminology used.g.(1/T0)} or x =4 N / R { 4 ~ / ~ ~ } (1 1.water). not knowing its molecular is weight.7)
R is the gas constant and 4 N the heat of fusion. (1 1.
(11.6). (1 1.5)
For the left side ofthe freezing point diagram.
.x] * x= AH/R{(l/I") .Y l ) for Y l and rewriting the appropriate lines.~ ) / R I " 2
r partial molar heat of the solid compound in ideal solution and solid per mole. in so doing to construct the "eutectic curve'. It is noted that althoug~ may be determined in this fashion. I. the value of x cannot be translated into weight percent. E~uation 1. and ) n of water..6. It is common to employ differential scanning calorimetry (DSC) for this type of determination.x If the tes a mole fraction x and the melting point of the pure drug suben the melting point of the contaminated drug substance 2" given by: lnfl ..he temperature dependence of x may be written (and expanded by
( 1nfx113T)p= .6)
( (l/I") = (1/T0) .
Integration of Eq. both at a temperature of at absorbed. would be (1 .{ ( l / R ) ~ [ ~ ( ~ ~ / I " ) / 3 ~ } { ( l / ~ ) { f 3 ( ~ b / ~ ) / 3 I " l } 3
= {(H . For the ram in Fig." Inserting.( ~ / ~ ~ ) lnx) l ram gives x up to 0.8)
can beextended by changing the upper limit in
TO statement to 1. which allows assessment of the purity of a drug substance by obtaining its "melting point'' T . this equation is known as the freezing point equation. 11..
1 164.6 178.3 0.1 F2 = F1 + 273.depending on the rate of cooling.5 146. As the temperature decreases.r l Program for Eq. 11. In this figure the melting points of mixtures of piroxicam pivalate (PIRP) (polymor~h with piroxicam (PIR) areshown.Q1 INPUT “ELT. A schematic of a DSC trace is shown in Fig. The most common way to assess impurities in p r ~ f o r ~ u l a t i o nby the use of is DSC.2 0.8)
INPUT “HEAT O F FUSION = ”.15 PRINT Yl.12 mol I) fraction will start “freezing” (showing separation out of a solid phase) at 150°C.1 0. it follows that lowering the t ~ m ~ e r a t u r e 140°Cwillcause below supersaturation of both com
Results from Table 111 Using A H cal/mol and To= 190°C Mole fraction x Melting point 190 184.”C = ”. 1998) of a eutectic diagram is shown in Fig.4 0. there will be more PIRP precipitati~g out. The solid phase is PIRP and. the crystals formed are or can be fairly large (in a relative sense).6 STEP .7 172.TEMP. At point C the following occurs: if one considers the lines AC and DC solubility curves.2.Y7 NEXT Y 1
The eutectic point is rarelya rational fraction and is.15 F3 = lfF2 Y2 = 1. andthe liquid (melt) with which is in equilibriu~ is given by the corresponding composition of the liquidus line.F1 FOR Y1 = 0 TO .5 0. intersection the between two solubility curves..6
.9 156.991Q1 Y3 = LOC(1U1) Y4 = U2*Y3 Y5 = F3Y4 Y6 = l/Y5 Y7 = U6273. A melt at 170°C of 0. TI
A literature example (Giordano et al. in essence.8
=
8200
0 0. 11.3. (1 1 .
4
'
' i 0. C is denoted the eutectic composition. which precipitates out be the eutectic precipitation. and TG(or between TH and TJ).there will be two phases present. Because supersaturation will cause precipitation of small particles.3
2
1 140 120
160
180
200
220
Temperature ("C) Schematic of DSC trace of compound containing an impurity. and is not to be considered a "compound. Similar considerations apply for cooling along the line except hereit is PIR which constitutes the coarser part. area ACB consists of PIRP plus melt. In the former. the eutectic ~ i x t u r e precipitate out (soliwill dify) as a finely divided mixture of PIRV and PIR. 1998. solid PIRP and a liquid consisting of a mixture o f PIR and PIRP.o
'
Mole Fraction PIR
Binary phase diagrams piroxicampivalate (polymorph I) with piroxicam. and above line ACD only liquid (melt) exists. The liquidus line is essentially a line indicating the solubility of PIRP in molten PIR at the given temperature.)
. precipitation of both will occur. and the The area CDE is an area where PIR plus melt exists. The eutectic composition is not (necessarily) a rational ratio between the two compounds.6 '
*
0.and at temperatures between 7.
pounds. in addition to the coarser PIR already precipitated. Heating solid along composition xF or xH will cause an onset of melting at temperatures TF (or TH)..0
*i 0.(Data of from Giordano et al.." elow line BHE (the eutectic temperature) only solid phase exists.8
'
1 . therefore.
220
D
E
120' 0.2 '
0.
7IS. TGor TJ as the end of the endotherm. (Data from
. xE2 = 0. with a local maximum in enthalpy of fusion at about 33% ketoprofic acid.B.
In certain instances the binary melting point between two c o ~ p o u n d will have an s appearance as shown in Fig.] where x and y are simple for is a chemical compound.. in that case are A and [A. [A. One may think of the diagram consisting of t ~djacent9 simple eutectic diagrams. and the eutectic temperature is the onset of the endotherm. ebrand and Muller~o~mann (1997) have reported on mixed crystals of sodium salt. P
Mole Fraction Q
Compd. The diagram is established by DSC. for one and B and [AXBY] the other.x). knowing the value of AHI and AHII of the two polymorphs I and I1 of their at their melting points allows calculation of one leg of the eutectic . To demonstrate this experimentally the authors prepared sodium sa€t/acid ratio of 2: 1 and the crystals thusly formed exhibited only one melti~g peak at .26. 1997) calculated x E l = 0. and knowing A H of PIR at its melting point then allows calculation of the other leg. and 1997. i. 1995. The compounds. TE1 = 140.4.)
compound formation. the line CJD is the solubilitytemperature line of PI the latter. ~etoprofen (DSC) shows a melting endotherm with d a mixed crystal onset of 400 IS. TE1 AG to be proportional to T allowed them to calculate the transition tempe~ature to be 32°C and to establish the Cibbs energy diagram.Similarly. Xray and scanning electron microscopy (§EM) showed these crystals to be
Compd. and the complete “melting” temperature. If th e first peak should not have occurr oymann studied mixtures of the acid and sodium salt in various ratios and then und three peaks. 11.e.18. This would imply a molecular compound of two sodium salts to one acid. Q
inary melting diagram with molecular ~ilde~rand ~uller~oy~ann. Ciordano (1998) from this (according to Yu.. the solubility (in mole fraction) is (1 .
11. 10. showing the close relation between solubility and melting point diagrams. willbe discussed in which Sec.14. (1: The foregoing is distinguishedfrom saZid ~ i s ~ e r s i Q ~ s .
not.1.
. 11. It is also noted that Fig. 1:3). 1 2 . 2 3 .
Temperature Liquid
Liquld
Sol Id
So!id
a b c
M o l e Fraction B
"D
Case (iii)
Case ( i )
Case (ii)
Three situations in which solid solutions occur.4 isthe (horizontally flipped) mirror image of ig. It is noted that the point C should happen at a rational compositio~ 1.quite different from either sodium salt or acid.
Some reports in the pha~aceutical literature have dealt with systems of the type shown in Fig. and then the remaining solution (which under other conditions might crystallize as a eut ctic).6. These are formed by freezing aqueous solutions. ~arstensen Anik (1976) have reported on the proporand tional re~uirements must be met for a solid solution composition to occur. In practice this usually refers to lyophilize^ cakes.Temperature
Te
I
a
c b Fraction B
Binary melting point diagram for two c o ~ p o ~ n dA . and the lass transit~ontemperature can be arrived at from thermal analysis (see Fig. 1964a. indeed. 1965. ice will first freeze out. The collapse temperature in Fig.. 11. 1971). 1963.
.7 a er and Nail. In this i~stance point c is. An example ofthis is Fig. 11. Goldberg et al..” t here the ‘‘solid9’ is simplyvery viscoussolution. an eutectic point. et zi. Guillory et al. as shown in Fi
The S O Z ~in freezedrying is (when dried) referred to as a lyophilized cake. 1994). not of pure compoun~s and of A ms have been reported for p~armaceutical systems by (~ekiguchi and i~uchi al. 11. It is ~ mostly amor~hous. 1969. the
olids that are notcrystalline are denoted ~ ~An important category of this ~ ~ ~ is lyophilized cakes (for intravenous reconstitution).8 is a temperature dictated by mechanical properties. Just above the glass transition te~perature. will supercool and will become 66solid.
hese systems have not been fre~uently reported in the pharmaceutical literat have been reported in the metall~rgicalliterature). On such freezing (when part of the solid comes out as an amorphate). which form solid s solutions and an eutectic. 11. but the (finely subdivid~d) the solid parate out are compositions a and b. and B..S). The glass transition temperature can usually be arrived at from thermal analysis.
If.7 An example of a supercooled viscous solution.20
.)
.4 0
i
20
Solution
tsovtscosity
Curves
0
. 11. have viscosities of about lo6 Pa/s.10) and the solids conten is so%. the initial freezing te~perature 240 K ( is 11. then the composition would be at point C .) sucrose solutions. (Data from Her and Nail. because the diffusion coefficient 23 for evaporation of water. 1994.60
0
20
40
60
80
100
Solute Concentr~tion.10
0
IO
20
T e ~ ~ e r a t u*C re
Thermogram of aqueous solution of 10% PVP. The relative magnitudes of the endotherms for glass transition vxsavis melting is shown.9) consists of the evaporation of the crystalline ice. after the primary drying. for instance. and a glass of a water content in the range of 1215% results.40
. The generalsequence of eventsinfreezedryingisshown in Fig. If the tem~erature belowthe glass transition temperature. (Data from Her and Nail. is then this glass has a high viscosity and will dry slowly. so that the cake is left with “holes” in it. ut if sublimation were continuously carried out at this
300
200
100
50
40
30 20 . 11. 1994. between the Tcand Tgcurve. but below Tgthis figureis 1Ol2 Pals. YO . The primary drying (see Fig. will be high.9.
acKenzie. Suzuki and Franks. willmodify the positions of the two curves. (1997) have shown that the water in c i ~ r o ~ o ~ ais ipresent in a c n complicated hydrogenbonded network. and this is ofinterest whenever a binary or multinary system is melted. Levine and Slade. but this is primaril~ due to electrolytes (e. and sublimation would be very slow.. These aspects have been discussed in detail (Franks. therefore. 1993). then. 1977. is continuously increased. such that Tc increases.Schematic of freezedrying events.
utectics have been treated. These will crystallizeout andgive the cake structural strength. Turel et al. Some proteins have stabilities that depend on cooling rate. at point B.
. and the viscocity would become very high.. as well as the initial freezing rate.g. 1990.g. sodium chloride) and stabilizers (e. in t foregoing. but their presence. glycine) in the composition. there is a
300
250
200
0
20
40
60
80
100
Solute Conc
Limiting phases in a lyophilization event. such that the lyophilization temperature can stay within the bounds of the two curves. in an initial sense.
temperature. the glass transition would be passed. The temperature. s the heading implies. so that a slowcooling rate may provide a different (and sometimes worse) curve than when a fastcooling rate is employed.
X ) m L
If this is divided by the total mass (m.and the amount of liquid. and the two liquids will be immiscible (Le. = ( X L . The melting point of a mixt from the pure compound. 10) becomes:
(11.2
0. The systems are considered binary in the following chapter. q. (1 1. A an . otherthan serving asanintroductionto the andarenot of muc concepts to follow. but the experiment described would be a theoretical exercise. producing a in molten A. mixtures of inorganic electrolytes (sodium chloride) and organic materials would be of that ilk. to the temperature corresponding to point W ) .g.= {WX}rnL
1.then Eq. mL. and the liquid will become richer and richer in liquid compos~tion be X. and if their melts are totally immiscible. followed by a sharpmelting point for l3. + mL). if a mixture of A and a little (point V)..4 0. .and then cooled along the line CQ.0
A
0. It is assumed in the following that A has the lower melting point. Systems of this kind are rare 1 interest. form two phases).8
1.ctic
series of different systems that may arise.11 (which is repeated for convenience). If two substances. 11. v will solid m.. and the components are denoted A and l3.becausemost organic materials decompose at or before the temperatures at which inorganic electrolytes melt. are mixed. A DSC thermogram of such a mixture would simply show a sharpmelting point for A.9)translates to
xm.
*
The commonly referred to situation of eutectic diagrams is the one shown in Fig. on further heating. This “precipitate” may be fairly coarse. (1 1.10)
220
T2
0.more and more solid w (separated) out.0 B
Eutectic diagram.G Mole Fraction
0.9)
(1
If the composition at Cr is denoted x and the composition at point X is denoted xL.. As the cooling progresses (e. then heating a solidmixture the two will first result in A melting. Then. for example. is given by the socalled weight arm rule:
{VW}m. B will melt.
. then solid phase willseparate out when the temperature at C is reached.
the socalled Z i ~ ~ line represents the solubility curve of i d ~ ~ ET2 is the other liquidus line and represents the solubility of A in B). x(E) it would have the appearance in Fig. where the system is liquid. 11. divides the space intofour areas. and by Cibbs' phase rule the degrees of freedom n. 11. would be as shown in Fig. 11. The latter profile is exactly the same as for a pure compound. this~ ~ i ~ ~ i cannot occur.11. as shown in Fig. for instance). then the following dilemma occurs: Line TE. Along the line UCWQ Y in Fig. so that the situation is resolved innature by both A and €3 precipitating. consisting of coarse crystals of one of the components and a "eutectic mix" of finely subdivide^ crystals of A and B. (a) represents cooling of a noneutectic composition. the area below the line QE where the system is solid.12(b). 11. If the temperature were to drop below the eutectic temperature.11) wheref. is the mass fraction of solid and f L is the mass fraction of liquid.e. and the two triangular area consists of melt plusA or melt plus B. It follows from the type of situation that large crystals of either would not be possible ( ~ o u l d result in too large an increase above solubility of either compound).
. As this precipitation re lf re occurs..1 11. 1:3. The conventional eutectic diagram. 11. 11. If it were melt plusA plu of phasesp would be four (including vapor). The area above theline TET'2. would be given by
1
L
L
TOG
U
Melt
T"C
U
\ Melt
olid T
out
Time
1fx=O
Time
Temperature profile during cooling along line UCWQY in Fig.1
(11. withsolubleliquid phases.I 1 the t e ~ p e r a t ~profile.12(a) and at the eutectic composition. In an e ~ ~ i lsituation. x = x(E) in Fig. in l 3 and the solubility of B in A would be superseded. the eutectic ~ i x t ~will occur. x(E) wouldbe not be a convenient ratio (1:1. then the solubility of A. When the point E (the eutectic point. until the entire mass has frozen will the temperature drop again. so that what will happen at further cooling (i. assuming conre stant heat removal. removal ofheat) is that a very finely subdivided mixture of A and 13. 1:2. r e ~ o v a o heat will not result in a reduction in t e ~ ~ e r a t ~Not. whereas (b) is either one of the pure co~pounds (x = 0 or x = 1) or the eutectic composition [i..11. but for a eutectic.e. corresponding to the eutectic temperature and the eutectic composition) is reached.
Reference is made to Fig. is 2.13) (1
where AHA is the heat of fusion of A. This means that temperature cannot be changed if both A.the number of components. 11.13.AHtotal for the entire melting is the obtained (by comparison with i n d i u ~ traces) from the area. T is absolute temperature. and melt are present. then the curve may be co~structed (and the eutectic point may be calculated as the intercept between the two curves. 3.T2:
ln(l . a logarithmic term has to be added.( h H b / R T }
+ P2
(11. "C
Schematic of a DSC trace. and this is exactly the situation depicted by Fig. and heated in the calorimeter.xB) = {AHA/RT']
+ /31
1. The solubility equation for section TE would be
TE :
ln(1 . The lines TE and T2E are solubility curves. 3 1. as shown in Chap. Atotalunder the entire
15
10
. where it is assumed that the heating causes the first thermal response (the eutectic temperature) at 40°C and the last at 120"C. or the root of the two equations). Eutectic phase diagrams may be obtained by DSC." n 20
40
GO
80
100 1 2 0 1 4 0
Temperature.12)
where f. It is possible to carry out the trace with just one DSC determination. for which the eutectic temperature is about 40°C and the liquidus line is at 120°C.14)
If xA and "xBare known at two different temperatures. and one method for arriving at the diagram is the following: A finely ground mixture of A and and mixed well. in which case. at the cornposition in question. XB is the mole fraction of B. the liquidus line.
. where the solubility is expressed in mole fraction. and is a constant applying to €3 in the system.12. and this gives one point where the points C (liquidus te~perature) and Q (eutectic temperature) can be plotted. A H may not be temperatureindependent. if it is assumed that AH is temperatureinde~endent. R is the gas constant.n=fp+2=24+2=0
(11.t. xB is known from the composition. For the section ETi the same type equation applies:
E .xA)= .
shows the solubility of indomethacin as a function of its concentration in an nicotinam~deindomethaCinmelt. for example.4. The fraction meltedfL.e. the situation is different. willnot be pure rather. Figure 11. There have been occasional reports in literature pre~entingeutectic data as soZu~ilit~ data.. then at a tem erature corresponding to H . If a composition at H is allowed to cool. would appear similar to Fig. These authors studied melts of indomethacin and nicotinamide.g..)
. (1998). by comparing it with AHtotal. but rather it is a 3 solid that is a solid solutio^ of I in A (or A in on the righthand side of the lgure 11. the solid phase crystallizing out in the areas depicting solid plus melt are not the pure compound (e.as m~ntioned Sec. solid willprecipitate. so that by use of Eq. the solid would be containing an a m o u ~ of A t corresponding to N . ten portions and the reas computed for each Tvalue. therefore. Aw. This. however. (Data from Bogdanova et al.. If a comp were cooled from the melt. and the value of f L for each plotted versus temus erature to give the entire l i ~ u i ~ line. An example is the work by Bogdanova et al.12b if only the right (or entirely the left)of the molar con~position were considered]. AHw is obtained by the crosshatched area.16)
fs is 1 fw. and a diagram such as shown in Fig.
he trace may.
0
0
00
0
0
%
20
40 60 80 100 Indomethacin in Melt
~olubilityof indomethacin as a function of its concentration in the nicotina~ideindomet~acin melt. (1 1.15 serves demonstrate the definition a solid solution in the strictest to thermodynamic sense. The inter~retationof the data issimply that of a eutectic diagram.
In some systems. Note that the eutectic c o ~ ~ o s i tis o ~ (or only by accident) a rational fraci not in tion of moles of A and €3.trace. is now given by:
(1 1. 11.12a wouldlack the line segment CQ when a composition of the molar ratio is heated or cooled [i.14. 11. 1998. A on the left of the eutectic). be divided into. will contain an amount of A corresponding to the point M .For a molecular Compound.16)it is possible to calculate xL. 11.
forinstance. and would take long times to estabwould have to be chemically quite similar and. there is no eutectic at all. 1965. in that event. There cases were reported in the 1960s and 1970s (Sekiguchi and Sekiguchi et al.
The situation would require equilibrium. Carstensen9 1981). 1 and Niazi. Au and Ag) and..l6(i). There are inorganics that form solid solutions over the entire composition scale (e. a couple of words on miscibility of ~ i is in ~ order.
efore continuing. The two liquids are partially miscible between room temperature T) and the boiling point (Bp). KC SCN form solid solutions.1976...1964. The most c o ~ m o n is the situation depic in Fig.. ~ i ~ ~ miscible liquids may exhibit different temperature behaviors. In pharmaceutics9 there are no solidly documented cases of solid solutions. a f ~ a c t i of ~ ~ ~ o
B
x(Q)
(1)
B
A
(ii)
B
A
(lii)
B A
(1v)
B
Schematic of partial miscibility diagrams. If a composition of at temperature F . then the liquid mass will separate intotwo phases. 1l.
.ctie
Mole Fraction
Schematic of a situation leading to strictly solid solutions.g. but the strict criteria for solid solutions as described in the foregoing may be missing in most of these (Carstensen andAnik. Guillory et al. Goldberg et al. 1971). 1963.
1l. there is full miscibility at a certain temperature below the boiling point. 11. A situation a bit more complicated is that A crystallizes out but that. lowering the temperature will bring them into the case discussed in Secs.
Solid dispersions were originally suggested by Chiou and Riegelrnan (197 1) who dealt with the dispersion of drugs in a base of pol~ethyleneglycol (PEG). there will be a temperature range over which there is a single phase. There are also several possibilitiesfor the makeup of point Q (the eutectic).
AT
There are several possibilities for what will separate out from a molten mix as it is cooled and point C (see Fig.2. depending on the melting point of l3. cases: 1. cases(i) and (ii)maylead to "separate. and miscibility of A and I3 may be limited. although they may exhibit phase separation at certain temperature. If this occurs. In case (i) it is necessary to heat the mixture to a temperature above the melting point of higher melting constituent). the situation is as described in Sec. then. 11. as has been discussed under Sec. a ~ o r p h a ~ ethe situation may be one of two s. and even so. 11.l6(ii)) there is complete solubility above a given temperature below the boiling point.The weight arm rule applies. I3 remains amorphous. In cases (iii)and (iv) meltingat a temperature above the melting point of B will cause a single phase. At times (situation (iii)) there is a temperature range overwhich there isonly partial miscibility and at times (situation (iv)).and a fractionfG of composition xG. and the appearance of this mayresemble that of truly solid solution discussed in Secs. The final product. Here.9. y. The amorphates are mutually soluble.ll) is reached. at the eutectic temperature. the attainment of a solid state is the point at which the rubbery amorphate phase is sufficiently viscous. and such systems. This
. so that (11. is crystalline A dispersed in amorphous . portions of a phase diagram.17) At times (situation in Fig.composition x F . there will be two phases.8 and 11. elts are liquids.. f 2. a solid d i s ~ e ~ s i o ~ . conse~uently. accordingly. it is obviously a solid dispersion of in a m ~ ~ h oA s u that results. 11. but on further heating there will be a phase separation. third case is one in which both comp nds remain amorphous. 11. In cases (iii) or (iv). The amorphates are only partially miscible.10.6. is a solid sol~tion it o the second kind that results. If the composition to start with is to the left of the eutectic.5 and 11. For materials that are neatly crystalline. it is a second case of a solid dispersion that occurs. then all o f B is in a very jinely s~bdivided state.
Lloyd et al. DSC traces may detect this type of dissociation.. 1961). devilliers et al. (1998) have described coprecipitates of acetaminophen with polyvinylpyrrolidone (PVP) formed by (a) coprecipitation. folded once. if the microenvironmental p increased).g. (1997) and Craig and Newton (1991)have made paracetamol (aceta~inophen) and PEG 4000. without slowing down the release to any extent. Lipman and Summers (1980)and Horn and Dittert (1982). (1998)havesuggestedpoly(methylglyxy1ate) as a base for oral. whether these solid dispersions are (a) solid solutions in the strictest sense. likewise have discussed subject. or (b) by coprecipitating the drug with the polymer from water or a solvent. and by cooling.”
The solid dispersion process is carried out (a) by either comelting the drug and a meltable polymer. then the contents of the container (e. such as PEG. After reheating. (1998) employing DSC. The comelt process will be discussed the first.g. On melting a single endotherm occurs at 55”C.g. (b) finely subdivided. (b) recrystalli~ation.. This point is of interest because. tablets) should contain x mg us ~ i v ~ ~ u t e . x mg.g. e. would dictate that if the pivalate is the ‘4drug. if a label states piroxicam pivalate. or (c) amorphous drug in a polymer matrix. ~oprecipitation yielded a better dispersion of drug in the base than comelts. in terns of labeling. recrystal~zationthe exotherm occurs at 40°C and. The coprecipitates reported by Brachais et al.
. such as PEG. and strictly speaking the drug substance decreases (because there is less solid present as the salt. The process of solid dispersions is carried out by (a) either comelting the drug and a meltable polymer.was shown to give enhanced blood levels. Strict interpretation in terms of law. crystalline drug substance in a polymer matrix. Giordano et al. If water is the %ft” component. ases other than PEG have been suggested over the years [e.”then the free base is a “derivative drug. then the right half of the diagram constitutes the solubility~temperaturecurve of the “right” compound in water ( ~ a r s t e ~ s e n . It is a principle that has been successful commercially (GRISPEG). have discussed these aspects of the subject.. or (b) by coprecipitating the drug with the polymer from water or a solvent. determined eutectic diagrams of piroxicam pivalate and prioxicam and found them to be different for the two different polymorphs of piroxicam pivalate. (1998) lend themselves well to compression.(c) mechanical mixing. Decrease in cr~stallinity was observed only in cases for which both the PVP and the acetaminophen were soluble or partly soluble (ethanol and water). controlled drug delivery systems]. solid solution. In certain pharmaceu situations. solid dispersions containing 20% paracetamol by comelting in a DSC pan at 1h storage at 70°C. a doublet occurs at 52” and 55°C. and the formed amorphous phase was a glasslike.. contains a doublet. 1977. Eutectic diagrams differ if different polymorphs are used as the second component. Brachais et al. and have assessed the products by Xray diffraction and by solubility. Lipman and Summers (1980) and Hornand (1982). pivalate). The question is. Denbigh. a salt dissociates in the solid state (e. for example. The authors suggest that the dispersions contain recrystallized polymerin both an extended stable chain form and in a metastable form. and (d) freezedrying.
Portero et al. 1986a.. (b) finely subdivided.crystalline drug substance in a polymer matrix.b).6dihydro~hydroxysolventmethodproduced solid dispersions usi he dissolution equation is given by Idt = kA(S . 1981). (1998) studied cogrinds of furose~ide with crosspovidone (polyplas~one).. 1984. Upadrashta etal.. and gelatin (Kigasawa et ai.. most of them usingPEG as a dispersion vehicle.. chitin and j h et al. (1998) have reported on the improve~entin dissolution of () k ~(4cyanoanilino) 5. hin et al. compressed at 2 t/cm2.
The general purpose of solid dispersions is to improve dissolution rates. Figure 1117demonstrates this with.)
. This is a substanc~ in that is nature. or (c) amorphous drug in a polymer matrix. 1974. Usui et al. 1998).he question is whether these solid dispersions are (a) solid solutions in the strictest sense. insects. and shells 1977). Dissolution of medium: distilled water. 1978). 1996..... 1998)is . The literature is replete with examples. Chitosan is a good direct compression ingredient (Nagai et al.
It is a common practice to cogrind drugs with polymers.&(14~2amino2deoxy~glucose is and y ~ " d e a c e t ~ l a t the~polysaccharide chitin. such as ~cyclodextrin( ~ i t r e v e et al. tablet consisting 100 using USP paddle method at 37°C at 100 rpm from a 10 m m ~ i a m e t e r mg of ofloxacin.1997).. (Data from Okonogi et al.g..C) 1. Arias et al. 1992) whichenhances dissolution of many compounds (e. 1997.. a solid dispersion of ofloxacin (Okonogi et al. microcrystalline cellulose ( ~ a m a m o t o al. 1997). nifedipine. et t al.. Chitosan (Portero et al.18) (1
60
40
20
0
0
20
40
GO
80
Time ( m i d
7 Dissolution profilesof pure ofloxacin(circles) and in 1:4 solid dispersion in mannitol/urea made by the c o p r e ~ i p i ~ t i o n method. being the princi~al component of crustaceans.
S is solubility. The best plotting mode as shown in Fig. however. Increased dissolution rates. 1998. owever. There is often a limit on the temperature to which the melt may be heated. If the drug is compo then it is noted that forthe increased surface area and.. 11. (Data from rachais et al. It is also seen from the foregoing text. complex with the drug substance. (1998) have suggested poly(methyl~lyxy1ate) a base for solid as dispersion. and a disadvantage of the systems is actually that highlevel dosage forms do not 1 themselves to the approach. solid compound. For a m o ~ h a t e s and (with the ~stwald Freu~dlich equation) also for very finely subdivided crystalline solids. perhaps. This is an advantage. but their data do not lend themselves to this type of plotting or to squareroot in time plotting. the composition must be to the left of the eutectic.. (b) area a
=
Aw
1 = notation for a
amount melted deduced from area in a DSC peak molecular.999 R"2
09
0
2
4 6 Time (hr)
8
10
Solid dispersion produced by copre~ipitation. This is because a melt has to be produced in which soluble. at time t in a dissolution experiment bs9 degrees of freedom
y
=
. the solubility S is increased over coarse crystalline material. because it that the d~ugloading can be considerable.0. and A is area.
/= I
( ) component A.where M is the amount notdissolved. are also aided in that many me~tproducing substances. so that the limitation lies in producing a homogeneo~s melt at the higher temperature. most successful solid dispersions have a substantial amount of associated with them.22 1 1 4 ~ = 0.)
.k is the intrinsic dissolution rate constant.37 105 . and the complexes en have increased solubility. that the surface area of either may be greatly enhanced by forming a solid dispersion. achais et al. then the dissolution profile should be a cuberoot plot. solubility to happen.18 is by a loglinear decay with a slight initial burst.0.C is concentration at time t. If the dissolution occurred through erosion. such as P G.
Craig DQM. pp 105110. Wurster DE. Craig DQM (1990). Ketkar A (1998). John Wiley & Sons. Academic Press. Brachiais C€3. = fs = mass fraction of solid phase mL = f L = mass fraction of liquid phase n = number of compounds p = mumber of phases R = the gas constant S = solubility [r = absolute temperature To = melting point of pure compound t = time XB = mole fraction of B in solid solution XA = mole fraction of A in solid solution is the mole fraction of A in a mixture x = mole fraction xL = mole fraction in liquidus phase x. Riegelman S (1971). = ( H A ) M is amount not dissolved m. Anik S (1976). New York. J Pharm Sci 65:158. Int J P h a m 163:219. Georgieva S (1998). Sidzhakova D. = mole fraction in solidus phase B1 = a constant applying to B in the system B2 = a constant applying to A in the system
= =
Bogdanova S. Burger A. Van der Watt JG. The Principles of Chemical Equilibrium. P h a m Acta Helv 61:69. Int J Pharm 163:l. Chiou WL. Int J Pharm '76: 17. Microchim Acta 1979:259. Carstensen JT (1980). Huguet J. Ford JL (1986). d~e Denbigh K (1961). Carstensen JT (1977). Carstensen JT. Bunel C (1998). Mechanical Properties and Rate Phenomena:.
. Ramberger R (1979). New York.Solid Pharmaceutics.fL
fs
A ~ / = fraction melted ~ A ~ ~ ~ ~ 1 fw = fraction not melted H = partial molar heat of the solid compound in ideal solution h = enthalpy of the pure solid per mole A H = heat of fusion AHA = heat of fusion of A AHb = the heat of fusion of B ~ H = the crosshatched area (fraction melted) w AH^^^^* = heat of transition k = intrinsic dissolution rate constant I. pp 256264. CapelleHue ML. Chiou WL. Newton JM (1991). Niazi S (1971). Rrug Dev Ind Pharm 16:250. J Pharm Sci 60:1333 1281. Pharmaceutics of Solids and Solid Dosage Forms. 1569. Duclos R. J P h a m Sci 60: 1333. ~ a ~ b r i ~niversityPress. Vaugelade C. 1376. De~illiers MM. Int J P h a m 169:23. Karaivanova V.
Dittert W (1982). Puttipipatkhachorn S. p. Pever T. Yamamoto K (1998). VilaJato JL (1998). Cry0 Lett 9:21. J Pharm Sci 86:991. Remu~anLopez C. Usui F. Yu L (1995). Int J Pharm 156:175. Press. pp 2139. Upadrashta SM. H11:709. Okonogi S. Van Norstrand. Chem Pharm Bull. J P h a m Sci 87:333. Chang CH. 1123. Sekiguchi IC. Zanol M. Obi N. 6uillory JK. Quiros M (1997). Maeda K. 356.
. Porter0 A. Academic Y. Chitosan and Related Enzymes. Pohlman H. Ikeda M. CF. 27235. Reboul JP. Lee YB. Dev Biol Stand 3651. Lotter AP. Levine H. (1993). E. J Pharm Sci 58:301. Cryst Res Techno1 21:1463. Graig DQM. Sekiguchi IC. Lloyd GR. Acta Crystallogr B37: 1844. Turel I. Hildebrand GE. Pharmazie. Cairfa MR.Dietz G (1986). Huang S. l Horn D. New York. Gulde C. Int J P h a m 15259. Van Der Watt JG (1987). Giordano F. Chitin. Pergamon. J Chem SOCFaraday Trans. London. Franks. Oxford. ~ i g h e l AD. MacKenzie AP (1977). Sax M. Ventura P. Cristau B. J Pharm Sci 84:966. Fromm GH. Nishimura K. Gazzaniga A. Himes VL. Yonemochi E. and Smith A (1997). Choi JS (1998). Butterworths Scientific. Nagai T. Pfeiffer S (1975). Jahn R. Ueda Y (1964). Treatise on Physical Chemistry. J Pharm Sci 76:744. Orlando. pp 213216. 9366. Int J Pharm 170:247. Sa~ayanagi Nambu N (1984). Astier JP (1981). 11:1 108. Oh IJ. J Pharm Pharmacol 32:21P. Suzuki T. Int J P h a m 175: 17. Lach J (1969). Kanig JL (1965). Pharmacology. Summers IvIP (1980). Obi N (1961). Slade L (1988). Oguchi T. Soyfer JC. Bukovec P. Nakamori Y (1963). Katikaneni PR. Acta Crystallogr B37:2242. Moyano JR. J Pharm Sci 71:1021. Sekiguchi K. Lowes MMJ. J Pharm Sci 54:1145. 556. J P h a m Sci 86:854. 30. Yamamoto K (1997). Choi HIC. Obi N. Garner WE (1955). Chemistry of the Solid State. MullerGoymann CC (1997). 164. lnt J P h a m 175:75. Terrence. Decamp WH (1981). In: Taylor HS ed. Kusai A. Nuessle NO (1992). Cry0 Lett 11:93. Chem P h a m Bull.Franks F (1990). Goldberg AH. Gibaldi M. Drug Dev Ind P h a m 18: 1701. eck 6. Yo0 C (1983). FL. 12: 134. Ueda Y. Chem Pharm Bull. 89:3283. Frazer JCW (1931). Chitin. Shin SC. Carima L (1998). Muzarelli RAA (1977). Lippman EC.
This Page Intentionally Left Blank
.
4. The Nelson and Shah Equation 12. dm/dt = kA(S .C ) (12. The Wood’s Apparatus: Sink Plate iss solution 12.5. No~sink.
The equation developed by Noyes and Whitney in 1887.9. 12. o n s ~ aSurface Area
191 192 194 194 194 195 197 198 202 203 206 208
12. is given by rate. 12.1. Constancy of the Shape Factor
The rate with which a drug substance dissolves either from neat drug or from a dosage form is of great i ~ ~ o r t a because it often governs the biophar~aceutical n~e profiles of the drug.6. 12. when a substance (with solubility S ) dissolves from a planar surface of surface A. 12.12. issolution by ~alorimetry ffect of Variables ilm Theory and the Levich Equation ixsonCrowell (Cube ependence of the Shape Factor During Symbols References ~ nt 12.7.3.1)
. 12.8. The NoyesWhitney Equation 12. states that (Fig.2.10. then its ~issolu~ion dm/dt (where w1 is mass and t is time).1).
2)
s:
any criticisms have been voiced against Eq. a die (as shown in the upper left of the figure).
. With smaller amounts of drug available it suffices to make a small pellet and encasing it in wax and exposing only one face to a dissolution medium. the shaft is attached to a motor. Tablet
Platen
.
where k is the intrinsic dissolution rate constant (cm/s). (12. 12. and it will be assumedto be so in the following. This is then lowered into a dissolution container. ~xperimentationcan be carried out with constant surface as when using a Wood’s apparatus as shown in Fig.ter 1
Punch
Shaft
Powder. dissolves. If the dissolution takes place into a volume of dissolution liquid V. In all of these cases Eq.1.1 Wood’s apparatus.2.and a shaft screwed on to it (as shown in the upper right of the figure). and the die is rotated. Alternatively a fairly constant surface area can be assured by simply employing a large excess of powder so that only a small amount of the solid. is placed on a platen and filled with powder. (12.1) dm/dt = VdC/dt = kA(S .4) is shown in Fig. then the concentration will change with time by a modification of Eq.( C / S ) ]= (kA/V)t or
C = S[l .3)
(12. in the long run.C) (12.2). The powder is then compressed.2) may be integrated to give ln[l . 12. but in general it is correct. most often at 50 rpm. In this. The die is removed. and C is the on cent ration at time t.exp((kA/V)t)’J
(12. (12. (12.4)
A typical curve following Eq.
. should cause curvature in the dissolution plotted according to Eq. (12. 12. But even under such condition direct comparison is not exacting because the substances may have different yield values. 1984). This. and only for substances with very low elastic yield value. 12. essentially. 1998.. (Data from Okonogi et al.
. 1974. 12. they give different porosities. of dissolution.2). this manner it In is felt that the area will stay constant (see Fig. will this suffice to make a nonporous compact. using USP paddle method at 37°C at 100 rpm from a 10mmdiameter tablet containing 100 mg of OFX compressed at 2 t/cm2. because it provides some measure of comparison (e.g.
A
G
. The other problem is that often too little compression pressure is applied to prevent the compact from being porous.3C).3A) will stay constant during the dissolution and simply recede (see owever. 1977) (see Fig.)
The use ofa cornpressed disk is referred to asa Wood's apparatus ( ~ o o et al. it is more likely that it will become uneven during dissolution (Carstensen.40 r
v 
04 0
20 Time (min)
60
80
Dissolution of pure ciprofloxacin in distilled water. and probably a fairly inaccurate estimate. Plate dissolution is at best an estimate. d 1963) whenthe setup is as shown in the second drawing in Fig..4). 1997.3 Types of compact formations in plate dissolution. Often as little as 1ton/cm2pressure is applied. Grant et al. hence. t o w pressures lead to wrong kvalues and erroneous conclusions (Chen and Grant. The best method is to repeat an experiment several times using different compression pressures until a consistent value of k is obtained. c evert he less the method is useful. 12..1. The general idea is that the surface that is exposedto the liquid (Fig. between the kvalues for different salts of a drug substance).
then curvature results. the surface area. soliddosage forms (tablets or capsules) must eventually manufactured for the clinic (e. and the dissolution rate constant.g. dissolution of the particulate material can often be assessed (and compar d with that of other compounds). n the other hand.1 to 1 mg min’ cm2 are in a gray area.1) are the solubility.
If constant surface area dissolution is carried beyond the sink level. it is only a constant at a given
. by placing the powder in a ca~orimeter (Iba et al. and compounds with values of kA/ V of from 0..[ M / ~ 0 ] 1 = (2kS/pr)t ’3
(12. drug substance per se is subjected to a dissolution test in a Wood’s odet al. Figure 12.. 12. in clinical be phase 11). owing to the hydrodyna~ics being different inthe calorimeter than it would be in a dissolution apparatus.3) at 50 rpm.
1 . S is solubility.1. 1979) that if k is obtained under sink conditions over a p 37°C in a USP vessel by way of Eq. p is true density. Mo the initial amount subjected to dissolution. 1998). 1963). For ound selectivity it is frequently useful to express dissolution findings in terms of k (Le. although quite dependent on onditions. if the value is less than 0. in cm/s). it linearizes (Fig. It has be lman.a point to be discussed presently. as shall be discussed shortly. not absolute.ne reason that lowpressures are preferred by many investigators is that pressures may lead to pol~morphictransformations. then the drug is not prone to give dissolutionratelimited absorption problems. data must be plotted by a cuberoot equation. 1931). Although k is thought of as a constant. concern with iss solution is that of bioavailability. It is noted that i~itially (up to 10 min) the curve is fairly straight. l o ~ ~ In the critical time path for product development. It is a l ~ a a ~sv i s a b ~ e ~ to r e ~ o v e s ~ a l l a ~ o u oft solid from the back of the c o ~ ~ a and test it for a n ct ~ o r ~ h o (Cuillory. (Hixson and Crowell.
For a small amount of powder.. (12.4 demonstrates this (Usui et al. The method is simply comparative.5)
is mass not dissolved. (12. and r is the mean “radius” of the particle.. 1992). but then begins to curve. such problems can definitely be anticipated. then if the onstant ( k A / V ) is greater than 1 mg min’ cm . If plotted logarithmically.
The variables in Eq. It is possible from data of the fore oing type to calculate k (cmls). Thistestisuseful.. The surface area must be assessed microscopically (or by image a n a l y ~ e rand the ~.5). divided by the total area under the calorimetric curve. / ~ 0 7 here proportional is to the area under the (differential) calorimetric curve at time t. 1991) and measuring the heat evolved as a function of time.
y = 2.for an otherwise. There is also the problem with to adhered “dust. Its hydrodynamic variability will be discussed in a subsequent section. rarely discussed.3) assuming a s~lubility 10 pg/ of
mL. 12.1) recedes.9.6dihydro7hydroxy7~cyc~opental~~ pyrimidine) hydrochloride.should be fine. 1998.
. (12. several curves should be generated using different compression pressures. A plane surface allows dissolution of the solid into solution.3454 . is the pressure at which the compact is carry out the experiment in a duplicable fashion..4 treated according to Eq.
Equation (12. (Data from Usui et al.997
“
0
10 Time (min)
20
30
Data in Fig. Carstensen (1977) has pointed out that asthe surface (see Fig. particles can. “fall out. The pressurewhich the curves become duplicable is then the pressure that should be indicated. uthors most often indicate the pressure used. ~ssumed be the surface area.6. Relative to surface area.” which may give initial burst.4882e2x
RA2 = 0. but do not justify the choice. 12. the area may not be “smooth.1) was. for a while. The powder used. 12. explained in the following fashion: reference i s made to Fig.” and the cross section of the die. may not be so.” e hidden variable.Time (min)
Dissolution o f (f)4(4cyanoanilino5.)
temperature and under given hydrodynamic conditions.
14) becomes: (1/ ~ ) d ~ = D(dC/dy) / d ~ (12.
is assumed that there is a film. of thickness h. whereas at distance h. and that the layer adjacent to the surface is saturated.2) becomes
k =D/h
quat ti on (12. the concentration is that of the bulk solution (i. (12. then dC/dy
==:
(C .7) It is noted that the amount dissolved A4 equals the volume V of the dissolution mediumtimes the concentration.13) (12. so that Eq.Dissolution
Bulk Solution
Schematic of film model.C) = S hence. which is attached to this. the intrinsic dissolution rate constant from Eq.” This inserted in Eq. The ~tokesEinstein equation states that
(12. Fick’s law now gives:
J = D(dC/dy)
where D is the diffusion coefficient.
. Therefore. 12.10).2) is often written in the fashion of Eq. C).e.C) dC/dt = ( ~ / h ) ( A / ~ ) ( ~
that is.11)
(12.9) (12. (12. and J is the flux.14)
+
. y is distance perpendicular to the plate.1n[q] ln[~/(6na)]
(12. T is absolute temperature.7) becomes
dA4/d~ VdC/dt = ( ~ / h ) A (.S)/h
(12. if it isassumed that D is distanceindependent. (12. q is viscosity of the dissolution medium.19) gives
k =~ ~ ~ ( 6 h n ~ a )
or ln(k/[T) = .10)
so that (Eq. This latter is (1/A) dC/dt..12) where IC is the Boltzmann constant. (12. and a is the molecular “radius.8) (12.
y is distance from and pe~endicular tothe plate. it The initial and boundary conditions are
C.14)] was very close to the activation energy for water’s viscosity. y = O .
where B is a constant. x is distance along the xaxis. It is noted that viscosity decreases with increasing temperature so that for viscosity the activation energy. rather than the film theory as explained here.
In a stream passing over a plate. = S . and w is rotational speed of the plate.<L
(12.19)
x=o
. C is concentration. (12. there is good reason to believe that it is the Levich equation. the dissolution is dictated by the Ficksian equation
D32C/3y2. v is kinematic viscosity. It is noted that by conducting the experiment at different rotational speeds it is possible. then
vx(y)
V x @ > = BY (12. In spite of this evidence. that applies to plate dissolution.~arstensen Pate1 (1975) studied the dissolution of oxalic acid at different temand peratures and reasoned that In[k/Tl should have the temperature dependence of the viscosity: that is. and C is concentration.7 that the velocityincreases linearly with distance from the plate. D is diffusion coefficient.18) V. The first term is diffusional and the second one is convectional.16) where Q is a constant. The Levich equation states that (12.(y) is a linear velocity that may be converted to mass flow rate (Q cm3/s) by ~ u l t i p l y i n ~by the cross section of the channel. 12. Ea. 12.17)
Here (as denoted in Fig. Theyfound that the slope of the modified Arrhenius plot [see Hq.
. is negative.
. y = O o C.x=o’o(y<~ C .as is shown in the leftprofile in Fig.7 Schematic of dissolution from a plate by a stream. Vx is velocity in the x direction (the dependence on yvalue is denoted ). Ass~ming.=:o. O < x= O .79.Vx(y)3C/3x= 0
(12. by plotting the dissolution rate versus the square root of w to obtain the diffusion coefficient.
Sm is the solubility of griseofulvin in the micellar solution. and relation is probably more parabolic as shown in the righthand profile in
The foregoing text has concentrated on the basic mechanisms for dissolution from t~eparti~le s~~face into moving body ofwater.999 y = 0.6 1.["/N2B]''3bhL2'3
(12.B and H are the width and height of the channel and I. but that the position of the identical with the theoretical line.3 1.ao et al. will hold with the foregoing boundary and initial conditions. and their derivation.7 1. without the msubscript.0
Dissolution of griseofulvin into solutions of sodium dodecyl sulfate (20 mM) solution. (Data from Rao et al.8 that the relation is linear.5 1 .8
1.8
0. will be presented here. It is seen from Fig. is a composite diffusion coefficient.18130 + 0.8080~'3S. and indeed they find such a relation.30659~ RA2= 0. for diffusion into a simple solution. ne problem with the experiments is the assumption of linearity of Vx(y).20)
where D.6 1. 12. then it is denoted ~ o ~ y ~ i s p e r s e . v denotes the rate of flow.27599 + 0. with exception that the
y = 0.33721~ RA2 = 0.9
2..4
1.9
x
0
d)
0. in modified form. and b are the length and width of the compress.not. A similar e~uation. 1997.7
0.For dosage forms. the drug substance is usually present as a multiparticulate population. The dissolution of a particle and a ~ o n o d i s p e ~particle pop~lations se were first ixson and Crowell (1931). If the particles are all the same e The size then the population is m o n o ~ i s ~ e r sIf . 1975) of griseofulvin into a surfactant solution should exhibit the following flux. They assumed the particles to be spherical.)
.998
n Theory
0 Experimental
0. first topic at this point will be the dissolution of a monodisperse powder under sink conditions. J
J = 0.. (1997) concluded that the dissolution from a ShahNelson plate (Shah and Nelson. According to this the log of _R should be linear in the log of Q with a slope of 1/3.
If we assume the particle to be isotropic and dissolving under sink conditions. the surface shape factor.21)
lowercase s denotes the surface of the particle. a.id
ac
a cubical particle shape will be assumed.24)
.d3
(12. integrates to b = bo .23)
q.is defined by the equation
s = a. Similarly the volume v of the particle is given.. From the geometry it follows that
(12.29)
so that Eq. such as the width of the particle. by ( ~ a l l a ~ a l l1948) as e.25) This.d2
(12.21) becomes (1 3pb2dbldt = 6kb2S or (12.28)
It is noted that (12.26) may be rewritten:
(12.3 1) The rate constants K and K' are referred to as the parti~Ze cuberoot rate c~nstants. oyes~hitney equation becomes: drnldt = ksS rn = pb3 and
s = 6b2
(12. after initial conditions.33)
' The ove~aZZ shape factor I is given by the equation
.27)
where the cuberoot dissolution rate constant K is given by (12. but to repeat.26)
2.b = Kt
= (~kS1p)t
(12. and d is its "size" (i. a defined d~mension). b.32)
where s is the surface of the particle. Shape factors have been dealt with before.
21 = a. and the equation will be derived for a single particle of sidelength b.e.(2kSlp)t or b o
(12. (12. (12..
r l
??Values for Some Isometric Shapes
_____
Shape Sphere Cube Isometric cylindera
Shape factor
[6]2/3~1/3 6 f .36) for N particles is (12.35)
s =* ~ / ( * ~ / 3 )
The overall shape factor will be referred to as the ~articZes ~ actor in the text ~ ~ e immediately following. then s. (b) The other situation i s for oae particle.1 with the a~propriate I"values.34) (12. by combining these three equations it follows that
(12.36)
If N ~ a r ~ i c l eeach of mass m. volume v.54
"A cylinder for which the diameter equals the height. and examples of this are listed in Table 12.834 6 5. and density p are allowed to dissolve.37) (Np)dv/dt = NksS
or
€l !
Zero
€l !
tu 1
Zero
Time
Zero
Oissolu~ofl or Critical Time
2.9 The situation during dissolution of a monodisperse powder. (12. The shape factor for a sphere is smaller than that for any other particle shape.9). Note that all the particles disappear at the same time.~ h i t n eequation dictates that for N particles y dm/dt
=L=
kNsS
(12. The ~ o y e s .
. the mass m of a particle is the density p times the volume v and it follows that Eq. 12.
r =s / ( v ~ / ~ )
so that. Particle shapes for which the shape factor is independent of size are denoted isometric. and it is noted here that all the particles disappear at the same time.5 [4]2/3n'/3
Numerical value
4. In the following there will be two situatio~s that will be considered: (a) the dissolution of a monodisperse powder (Fig.
because incorrect conclusions may be drawn from employing an incorrect definition.38)
(12. (12.44)
Equation (12.[ v / v ~ ] ” ~ {(krS)/(p3v:I3)}t
It is noted that
vivo = N ~ / N p v o M / M o =
(12.40) gives
.44) in Eq.
= 1 .[M/MoI1l3= {(krs)/(p3(Mo/Np)’13)Jt
= {(krS)p213M~”3N113/3)tK ~ n t =
(1 2. (12.13
__
(12. The particles are isometric. the more familiar form of Eq. It is importa~t keep an to account of which dissolution rate constant isbeingdiscussed.40)
where Kvis the cuberoot dissolution rate constant based on volume for one particle. The various rate constants are summarized inthe symbol listat the end of the chapter.47) 2.49)
(1
Kmnis the cube~oot constant of N particles based on mass.43)
Since
v.39)
( 12.dv/dt = kFv2I3S/p ecause s = h213. It i s worth repeating that cuberoot dissolution in the strictest sense is derived from the following assumptions: 1.v‘’~= (kFS/3p)t = Kvt
(12.[M/Mo]’/3 {(~rs)/(p3v:13)}t
(12.
.43) may be rewritten 1 .[ M I ( N ~ =I( ~ w 3 p ) t ) k~~
or
M.l3 = (~01p)113 ~ =
~ ~ l ( ~ ~ ) ~ ~ 1 ~
(1 2.therefore dv/[v2’3 ] = (krS/p)dt This integrates (after initial conditions are imposed) to
v:I3 .40) emerges.46)
is the relative rate constant based on a population o N p~rticles. (12.45)
where
f& (
(krs)p213A4~13N‘13 0 13
(12. f Inserting Eq.42)
where A is the total mass of the undissolved powderat time t.48) (1
A4113
= [ ( ~ p ) l l / ~ ( k r s / 3 p=tKmnt )
where
ICmn = N1I3 krSpV2l3 13
2.41) 1 .
= (12. Hence. and Mo is the amount 4 before dissolution.
The initial steps o f the derivations follow fairly closely those of Pedersen and Brown (1975. however.who studied the dissolution from various faces of crystals of different morphology. h. it isassumed that the shapefactor is constant dissolution event(Le. owever. The former will be addressed first and after that a derivation will be presented to study the extent to which changes in shape factor may affect the cuberoot law. There is one (probably very common) situation that allows the equation to be used without adjustment for change in shape factor during the dissolution. Much dissolution work has been done with this assumption. 12.50) (12.rl
2. and kl. 3. Sink conditions prevail. then a parallelepiped with sides b. to study the shape factor as a function of dissolution time. given by
kh
= (h1b)k~ or khlh = kb/b
(12. If a crystal grows from a nucleus in the 6. The population is monodisperse. an isometric particle will dissolve according to this equation). The intrinsic dissolution rate constant k is assumed to be a constant for the substance. 1976). The interest in this lies in (a) whet he^ there are situations for which the shape factor stays constant.not only the effect of the shape factor thatis of importance. kh.
. crn/s) are numerically identical. h and I will dissolve by the equations
b
Schematic of ideal crystal. It is. if these constants are also dissolution rate constants. It is often postulated that dissolution rate constants (k. and (b) if it does not. in spite of the through workofPedersen and Brown(1976).10) with constants kb. the first assumption may be studied separately.
xsonCrowell equation. It is important to investigate whether the cuberoot equation also holds for situations in which the particle is not isometric. and that the shape factor calculations are the ones unique to our theme developed thereafter. but opposite in sign to crystallization rate constants. The dissolution is isotropic..5 1)
kt = (L/6)kb or k L / L = kblb
then. 4. and Ldirections (Fig. and derived equations that describe their dissolution behavior.
60) ai and Carstensen (1978) studied the effect for isotropically dissolving oxalic acid cylinders to establish the effect of the shape factor on the dissolution pattern.9 that most often the HixsonCrowell equation works well. 12. The extent to which this happens will be discussed in the following.(12. oni isotropic shapes were simulated by cylinders. under these conditions. and found that this model yields a better fit of dissolution data for hydrocortisone particles than a similar model using spheres as a model. This was then compared with the actual dissolution of a cylinder and good agreement wasobserved.52) (12. (12. when l? is independent of dimension) that this may be expected. It is only when a particle is isometric (i. hence. and the effect of dissolution on the theoretical shape factor determined. will stay constant during the dissolution event.e.54)
so
v2I3 = {L0b0h0}2f3{ (kb/bo)t]2 1 (12.. the shape factor.3 1) multiplied t h r o ~ g h p] by pdvldt = kirv2I3S
(12. (1993) used the concept proposed by Lai and Carstensen (1978) of imitating the shape of a crystal by a cylinder. the shape factor changes suf~ciently during a dissolution run that there are deviations from the cube root law.
It is for the reasons stated in Sec.
.58)
This is independentof time. the shape factor
s/u2I3
8(Lo
+ ho + b o ) / { [ L ~ b o h ~ ] ~ ’ ~ }
(12. Lu et al.57) 
That is. At times. It has been shownthat fora single isometric particle [see Eq.53) (12.
64) bh) hL ISince v is a composite function of L.65)
This is introduced into Eq.(K/ho)t} b = bo . (12. and height h as shown in Fig.
s = 2(Lb
+ hL + bh)
1)
(12.68) (12. 12. and ho are the initial length. These expressions can be integrated to obtain relations between the crystal dimensions and time of dissolution. bo.73)
. the rate of change of volume relative to time can be written as dvfdt = {dv/dL}{dL/dt} {dv/dt}{db/dt} {dv/dh}{dh/dt} = bh(dL/dt) Lh(db/dt) Lb(dh/dt)
+
+
+
+
+
(12. it follows that dL/dt where
I 
+ Lh{db/dt} + lb(dh/dt}
=T=
{2kS/p}(Lb
+ hL + bh)(12. and h.L/Lo) = u )
(12.Kt = ho(1 . a fact that is not correct and the rami~cations this will be discussed in the of following. (12. b.67) This differential equation is in the separate variable form.(K/L*)t] h = ho . then integration of Eq.10. (12. and (12.61). and height of the crystal. (1993) assume the shape factor to be constant during the dissolution event.(K/bo}t}
enoting: 1 .63)
~ombining Eqs. Hence.Kt = bo{1 .72)
(12.68 gives
L = Lo . Consider a parallelipiped of length L breadth b. 12. breadth. The area s and volume v of a particle with such a geometry are given by the follow in^ equations.69)
1°C = 2kS/p
This implies isotropic dissolution.(b/bo) = (1 .6
v = Lbh
The NoyesWhitney equation under sink conditions states that dm/dt
= ksS
dt
(12.66)
db/dt
I=I
dh/dt = K (12.70) (12. If Zo.63) to give bh{dL/dt] or (12.62).Kt = Lo(1 .( h / ~ o= 1 .71) (12.Lu et al.63) now gives pdvldt = 2kS(Lb (12.
density being constant. The slope of sucha plot gives the value of the intrinsic dissolution rate constant if the solubility. F can be obtained from dissolution data. A cube. h = 0.78 to 12. From Eq. it follows that
u = Kt/ho = Kt/bo = Kt/Lo
(12.uH1 . (12. namely.(ulbo)ho}{ .The criteria for choosing the “correct” root of the possible three is that it should be a real number between 0 and 1. The following equations ensue by inserting Eq.77)
Employing these i ~ e ~ t i t i Eqs.we consider the fraction of amount undissolved (I.(ulho)} 1 1
(12. at the point when the crystal completely dissolves. and a right circular cylinder are examples of isometric geomtries because their shape factors are independent of their dimensions.= ~ / ~ oI.
= ~ b h / ( ~ o b o= (Lo . (12.69) into Eq. and initial dimensions of the dissolving particle are known. Adequate linearity for a plot of u versus t has been demonstrated by Lai and Carstensen (1978) for cylindrical tablets of oxalic acid.75) (12. (12.where u is dimensionless.80) (12. Real particles are far from being isometric.~ h o ) / ( L o b o h ~ ) ~o) = { 1 .80) become: es L =I: Lo .76) (12.
I.(ulq)l{l .79)
Two shape ratios are now defined inthe following fashion which are indicative of the shape of the crystal.74) (12. which enables one to solve Eq.82) for u. Hence.8 1) (12.)can .uho)(bo .l]. be expressed as the ratio of instantaneous volume of the dissolving particle to its initial volume. it is clear that a linear relation between u and t exists. a sphere. The shape factor for one particle is defined as
r
= :
sv2f3
(12. density.83)
so for the parallelipiped model described in the foregoing
r = 2(Lb + bh + Lh)(lbh)2f3
(12.U l P }
This is a thirddegree equation in u.
P = Lolho
4 = bolho
(12. that is.74)
u ={2~~1(~hO)}t
(12.78)
To determine u. Tsotropicity and isonletricity are some of the basic assumptions in the derivation of the HixsonCrowell cuberoot law.uho
h = ho .82)
Rearranging the foregoing equation using these two ratios gives the following result
= (1 .uho)(ho .78). the domain of u is [O.(ulLo)ho}{ .uho
It should be noted that just before dissolution starts h = ho and.uho
b = bo . (12. (12.84)
.
83). lo oftimeofdissolution bo = 0.861 and reduced time [see Eq. signi~cantchangesin the shape factor wereencountered as this ratio increased above unity.
A
= surface area of a plane b = (a) breadth or (b) size of a dissolving particle bo = original (a) breadth or (b) size of a dissolving particle I) = diffusion coefficient d = general term for the size of a particle
y = . 12.Inserting Eqs.999
Time (min)
Fraction retained [see Eq. and (12. they observed that r changed si~nificantly after 50% dissolution. and h.75. They found that. This is exemplified in Figs. as a function of reduced time.)
.92) yields
(12. (Data from Dali.741 as a function of a ~ o t a s s i u ~ dichromate crystal of d i m e ~ s i o ~ s= 1.12. cm. gives
(12.85) earranging this equation to write it in terms of the shape ratios p and q.120 cm. 12. (12. Lai and Carstensen (1978) followed a similar approach for cylindrical tablets with different radius~hei~ht ratios. (12. (12.11 and 12.1. expectedly. 1997. 12. For a ratio of 2.86) of this expression it is possible to calculate the shape factor for the para1 geometry considered here.518 cm.841.4607e2x
RA2 = 0.0795e2 t 4. for an isometric tablet ( there was no change in the shape factor as the dissolution proceeded.85) into Eq.
518 cm. and ho cm as a function of dissolution time.ict
30
20
n
0
10
20
Time ( ~ i n u t ~ s )
Shape factor as a function of dissolution time of a potassium dichromate crystal of dimensions lo = 3.)
h = (a) thickness of film layer (b) height of a crystal J = flux K = ~ k S / p cuberoot dissolution rate constant for a cube = Kv = k r S / 3 p = cuberoot dissolution rate constant for a particle with a shape factor of r' KmB = N1/3krSp2/3/3= cuberoot dissolution rate constant for apo~ulation of N particles = (kI"S)N1/3p "2/3"1/3 /3 = relative cuberoot dissolution rate constant for a population of N particles k = intrinsic dissolution rate constant kb = intrinsic dissolution rate constant in bdirection kh = intrinsic dissolution rate constant in hdirection kL = intrinsic dissolution rate constant in Ldirection L = length of a crystal = mass of a po~ulation monodisperse particles of = original mass of a population of monodisperse particles m = mass not dissolved at time t for a single particle mo = original mass of a single particle = ~ o / ~ o = mass flow rate (cm3/s) 4 = bolho s = surface area of a single particle S = solu~ility t = time u = reduced time V = volume of liquid v ~ c ~ ) velocity Q (cm3/s) = linear y = direction ~erpendicular a dissolving surface to
. ( from Dali.120 cm. bo == 0. 1997.
J Pharm Sci 64: 151824. Pedersen PV. American Pharmaceutical Association. Englewood Cliffs. Frisella ME. Levich VC (1962). Physiochemical Hydrodynamics. Y a m a ~ o t o (1997). p 142. 23:689. The Academy of Pharmaceutical Sciences. ~onemochi Puttipipatkhachorn S. Carstensen JT (1996). Dali M (1997). eds. New York. Whitney W (1897). Int J Pharm 18:25. J Pharm Sci 86:1132. p 5. K 156: 175. ~allavalle JM (1948). P h a m Dev Techno1 4:487. Southard MZ (1997).r l
a. Grant DJW (1998). P h a m Res 13: 155. Larive CK. J P h a m Sci 64: 198 1. Pharm Res 10:308. Johnson KC (1993). Arakawa E Morris T. 2nd ed. Dali MV. Brown KF. NJ. Lu ATK. Noyes A. Lin M. Int J P h a m E. Dissolution testing in drug development and quality control. (1976). In: Leeson L. Grant DJW. American Pharmaceutical Association. Shah AC.
. Cuillory K (1992). Carstensen JT (1978). Rao MR. Task Force Committee.
=
surface shape factor of a particle shape factor density
av = volume shape factor of a particle /3 = constant in the flow conversion equation
lr
=
p =
s
Carstensen JT (1974). WI. Lai TYF. Personal Cornmunication. (1975). ~a s h ingt on. University of isc cons in. Iba I(. Printice Hall. Dissolution Technology. Carstensen JT (1991). Chen LR. PhD dissertation. Madison. Carstensen JT. Pitman Publishing. X. pp 871 16. p 3 1. In: Micromeritics. Oguchi T. Medhi~adeh Chow AHL Fairbrother JE (1984). Riegelrnan S (1979). Drug Dev Ind Pharm 17:77. J Am Chem SOC Olconogi S . M. Int J Pharm 1:33. Nelson KC. The Academy of Phamaceutlcal Sciences.
48) (i.l). Dissolution of Polydisperse Powders rticle Size ~istributions l is solution and ssolution After the Critical Time.” that is starting at size zero (with an infinitely small probability or fraction) and on the high side ending at an infinite large size ( a g a i ~ with an infinitely small probability or fraction). by Brooke (1973. then it is possible to solve dissolution patterns in closed form. ~ a r s t e n ~ e n ~ u s (1972) applied si~ulation a truncated l o ~ n o r ~disand a to al tribution of spherical particles to the principles of dissolution. lognormal powder. (12. t* Symbols Appendix eferences
209 210 217 219 220 22 1 22 1
If powders are polydisperse.13. 13. then if the population is “infinite. three additional assumptions were made: namely. This was confirmed experimentally by Carstensen and Patel (1975) who found that. (b) the expression for the distribution remained correct during the dissolution process.1. a more realistic powder population is one in which there is a finite. the plot of amount retained versus time would follow a cuberoot equation. approximately
. and (c) a numberbased lognormal distribution would also be l o g n o ~ a l after t r a n s f o ~ a t i o n the HatchChoate equation. 1974). and for such systems the type of dissolution pattern of the type in Eq.e.. As we will see in the following.1974) arrived at the conclusion that for a polydisperse. a cuberoot law adherence) will prevail. In the latter case. and this was followed up by a solution in closed form by Brooke (1973. smallest particle and a finite largest particle (Fig. but the slope would be inversely proportional to ln[a]. that (a) the smallest particle was zero size.
Coulter counter) of the particles from a random sample and suffer from the fact that (a) the volume is converted to an equivalent spherical radius. 1963.” ( oughton and Amidon. al. The importance of particle size distribution of powder substances for dissolun well documented inthe pharmaceutical literature (~arstensen and ke... exhibit skewed particle size distributions (Carstensen and Rodrigueznedo. “.. However.. havecautioned researchers about the misuseof the spherical approximation by stating that.. 1974. In today’s climate.Size Zero a(rnax)
cl
Zero
a(m1nf Time Zero
Zero C r i Ti ic a e t ml
t(3)
Dissolution Time
The situation in the dissolution of a polydisperse powder.
(13. 1991).g. 1975).1) where h. there are practical very problems thatare associated with particle size distribution and dissolution. Simply said. Carstensen and Patel. 1989 Pedersen and rown. Higuchi et al. It further has the dis~dvan
.. virtually all soliddosage form products are routinely subjected to dissolution testing. 1963. Size determination of needleshap~dparticles b techniques such as a oulter counter can lead to experimental difficulties such as 66coincidence’7 two of or more particles at the orifice leading to faulty results. Higuchi and Hiestand.direct comparisons between microscopederived mean particle size parameters and Coulter~ounter~derived data lead to erroneous conclusions if can no consideration isgiven to the shapes of particles.. 1975).
articulate solids resulting from unit processes such as crystallization. smaller particles dissolve at a faster rate than larger particles. 1974. is the geometric mean of the particle population and ln[o] is its standard deviation. milling. intz and Johnson.. precipitation. 1985. ~ l t ~ o u g h treatable from a theoretical point of view. and (b) that the sample size is always very small. most particle size distributions are derived from volumebased measurements (e. and the most common cause for product recalls is failure of a roduct to meet dissolution specifications (Cabana and article size limits are included in drug substance specifications. Steiner et. 1973. because particle size affects both dissolution characteristics and machinability (flow or compression) of the substance..
and the curve may be normalized. as shown in Fig. so that the area under the curve ( ~is unity. by microscopy. and p is particle density It is possible.Ferret diameter) is selected. The numbers in the ranges may be converted to fraction of particles. based on which the representativeness of the powder population has to be relied. 13. ughton and Amidon (1991) have suggested a microscopicbased image analysi utine procedure to have a check on the lottolot e particle size and shape characteristics of three lots of an investigaughton and Amidon. t o o ~ t a i meaningful inforn mation on particle size distribution of crystalline substances of nonspherical nature (Dali. Also. If a particle has an initial breadth bo.bo + dbo) would be given by
0.2.2)
where v is volume. the user is restricted to a relatively small sample s .002
180
230
280
330 430 380
Breadth (microns) Frequency distribution function o f a 401 drate. so the number fraction iVb of particles between the infinitesimally small interval (bo. dissolution in a USP dissolution apparatus.
+ 50 sieve fraction of oxalic aciddifiy
. and volumetric shape factor avothen the original mass. croscopy will yield information about the c ~ ~ a Z a dimensions of nonspherical . On the other hand. Invariably.tage that it does not directly address the main reason for its execution: namely. 1991). In this.008
0. and a certain number of particles examined. a particular dimension (length. with distinct advantage. m. the inherent tediousness and timeconsuming nature of this procedure limits its use. of that particle is given by
~
m = pv = p(avo)b~
(13. breadth. 1997). to determinetwoof the three dimensions and to plot these by a normalized fre~uency function f(bo). Powder dissolution of polydisperse samples can be used. and classified in size ranges. such methods rely on the redefined size and shape parameters.004
n
0.006
& 0. The major advantage of this method is that it can furnish size as well as shape information about nonisometric solids.
Edmundson and Lees. Schoonen et al. (13. and Dl are elaborated on in the following.. it would given by
(1 3. 1979). If the powder is allowed to dissolve then. Carstensen. Dali. An example is shown in Fig. here denote the largest and smallest dimension of the particles in the powder population and bo has reference to the fact that the particles will be placed in a iss solution medium at a given time of zero.4) applies to a polydisperse system ofparticles (for instance. It is at the critical time that the smallest particle disappears from the dissolution medium.B.Kt
(13. then the dimensions of the particle willdecreaselinearly with time (Carstensen and Musa. and anaverage valuefor the initial volume shape factor has been ascribed to that particular sieve fraction. a sieve fraction). If the powder is dissolved under sink conditions.the total number of particles in the system remains the same (Carstensen and Musa.8)
This term is obviously the original mass of the powder sample and p3 is the third moment of the probability ~ i s t r i ~ u t i o n f u ~ c (Bennett and Fran~lin.5)
Note that the initial distribution function can be used to calculate the mass undissolved until the critical time t*. and up to this point in time (t*).. 1975. 1972.
b = bo . 1996.3. 1972. 1965. The third coefficient in Eq. then M
==
A1 . tion 1961).4)
b .Denoting the initial mass of the whole powder population by Mo. 13.. 1997).n .8) is (13.9)
. B1.6)
If the cubed term is expanded. Carstensen and Patel. where the coefficients A I .t2 . Equation (13. at times t < t".. the mass undissolved will be (13.7)
C. and bm.t
+ C.
(1 3. t 3
(13.
.10)
where p2 is the second moment of the probability density function (Bennett and Franklin.0
0
100
200
300 Time (s)
400
500
600
Amount undissolved for the dissolution of a 40/ acid dihydrate in 0. The variance of the powder population is given by
s2 = lLL2 .t ~
0
c.12)
where use has been made of the fact that the probability density function as used is normalized? Le. The second coefficient in the expansion is
(13.7) may be divided through by A = Mo.13)
Equation (13.B.
+ 50 sieve fraction of oxalic
Where p l is the first moment of the probability density function and also the mean of the distribution ( ennett and Franklin.1 N HCI at 25°C and 50 rprn.11)
The coefficient to the last term is given by
(13.0
0
0
0.Pl
2
(13.t2 . 1961).14)
and the coefficients with subscript “2” are then the coefficients with subscript “1” divided by Mo.t3
(13. 1961).D.
.
(13. after which it takes the form:
~ 1
= 1 .
45
r^\
c
x
?4 
0.42 x 0. and 501 + 60. problem can be considered in the opposite direction (~arstensen and th the knowledge of K value for a compound (for instance oxalic acid dihydrate) under specified hydrodynamic conditions.2) through ( 1 3 4 . The results for volume shape factor obtainedby two methods are quite comparable.The coefficients of the terms in t.17)
In a typical research and development setting. Similarly.x2 and p3 can be evaluated.14) through (13. from Eqs. for that compound under specified h y ~ r o d y n a ~conditions can be determined from ic powder dissolution data and particle size analysis by microscopy ( can be done with Eqs. t2. The K value should be independent of particle size.4).15
0
100
200
300
400
500
Time (s)
Cuberoot law plot for the dissolution of a 40/ acid dihydrate showing the crltical time. the initial number of particles.17). the moments of the distribution function of a “dimensio~ sigof nificance” can be obtained. At the critical time. in the event that a new drug candidate is recognized by the drug discovery group.35
0
I
?I 
0. (13. (13.14) are given by the following equations: 1 2 = 3KP2/P3 3
c = 3K2P*/P3 2
D2 = K 3 / P 3
(13. the average initial volume shape factor for a polydisperse powder be can estimated ( issolution studies and particle size analysis on three sieve fractions of oxalic acid lhydrate: 301 40. there is a change in slope
+
+
0. (1 3.19) x lom4cm/s when dissolution was carried out in 0.16) (13.
+ 50 mesh fraction of oxalic
. the fraction undissolved as a function of time. Only the dissolution data up to the critical time are utilized in this manner (Fig. by knowing N .55
0. 401 50. I N P paddle apparatus at 25°C and 50 rpm. and the density of the solid. and t3 in Eq. yielded a K value of (1. From thedissolutio 2 is obtained and through the results from microscopy the moments . 13. then the dissolution rate constant K .
15) through (13. This was necessary to have assurance about the validity of the approach.47 0. the integrals that define the moments of distribution function are allowed to “float. the following question arises: Is it possible. 1931.5. The mean and §tandard deviation for a particular sieve fraction can be calculated using the following equations:
+
ean = p I
(13. to obtain the distribut~on parameters from powder is solution data. 1975).in the cuberoot law plot (Hixson and Crowell. the moments of the distribution function had to be known.4) through (13. The mean and standard deviation obtained are compared with those for the breadth of the particles from respective sieve fractions obtained from microscopy. For these sieve fractions the distribution parameters for the lengths and breadths of the particles were known. (13. with three p2. The moments of distribution.7) can be obtained. In all the three sieve fractions. ali (1998).3 shows the dissolution curve for a 401 50 mesh fraction o acid dihydrate. The cuberoot law plot for the same event is shown in Fig.
. the mean obtained directly from dissolution data is less than the mean from microscopy. equations used to solve for three unknowns.g. p l . (13.14). So it is of interest to determine which of the three dimensions of the particle is perceived by this approach.47
_ _ _
Source: Carstensen and Dale. oxalic acid dihydrate).56 0. and p3 can be evaluated from Eqs. In the process of working backward. From the leastsquarefit to the data the coefficients of terms in t as per Eqs.17). Carstensen and Patel. 13.. to exemplify these ideas.” In other words no restriction on the kind of dimension is imposed at this point. To obtain an estimate of a K value for for a compound (e. by not imposin~ any restrictions about the dimension of the
Distribution Parameters for Sieve Fractions of Oxalic Acid Dihydrate Determined from Dissolution Data and Comparison with Those Obtained from Microscopy Standard deviation from dissolution
( P 4
Sieve fraction 30/+40 40/+50 5O/t. (13. 1998. 13 which an estimate of the critical time was obtained.60
p(bo) from oi. The fraction undissolved data until the critical time can be leastsquare~tted to a thirddegree polynomial in time as dictated by Eq. Thus the restriction. of breadth being the de~ningdimension was imposed on the integrals before they could be evaluated numerically. The discussion pertaining to this aspect will be resumed subsequently. carried out dissolution of three sieve fractions of oxalic acid dihydrate.1 and Fig. In the wake of this observation. Figure 13. from microscopy microscopy (pm) (Pi4
Mean from dissolution
(Pm>
n = mean (holbo)
Predicted ~ ( ~ o )
(w4
410 299 240
85 55 32
84 70 33
0.18)
The results for the three sieve fractions are shown in Table 13.
000

Mass undissolved for the dissolution of a 5g 30/40mesh fraction of oxalic acid dihydrate.10 0.30 0.30 0.1.3.10 0.00
x
y 14
=
0.
.5459e2x
+
1 .
5.0046 .78888
+
1 .00 0.ter 1
Third Derivatives of Dissolution Curve in Fig. 4 6 7 2 ~ R2 0.998

12
10
8
G
4
2
3
4
5
6
’
7
8
9
Shape Factor From Dissolution
Correlation between microscopically and dissolutiondete~inedshape factors.6 Time
0 30 60 90 120 I50 180 210 240
d3M/dt3
0.8768e7x^3 R*2 1. 13.2648e4xA2.
That this is (approximately) so is shown in Fig.particles on the integrals comprising the moments of distribution. or an apparatus allowing slower dissolution is employed.00 1
0.5). 13. other solvents and apparatuses may be used.001
\
P
z
r. (13. It is seen from Eq. these two set of values in excellentagreement.1. thereby improving precision. As shown in Table 13.) that the third derivative should be independent of time.002
0.3 0.14. then this apparatus should be used. the model for dissolution must.7.7 Third derivative from data from a dissolution run of a 30/40 mesh cut of oxalic acid dihydrate. then longer time intervals priorto t* maybe used. Thus the mean of height ( ~ = Mho) for particles belonging to a ~ o particular sieve fraction can be predicted.5556e3
+
4.7619e6x Rn2 0.003
0. Ifhowever.002
0. If a solvent exerting less solubilizingpower on the substance is used. It is obvious that the longer the precritical time is.? 0.1and Fig.025

0. the value of (aqueous) K is sought under US. the better the assess~ent of the coefficients.004
0
100
200
Time
ig. the authors resorted to the volume shape factor data for these sieve fractions that were obtained microscopically(see Table 13. change. The ratio of mean height to the mean breadth can be calculated from the volume shape factors obtained frommicroscopy and dissolution. N/10 hydrochloric acid or sim~lated gastric fluids could be used as the dissolution media.
Y

1. but just for the purpose of determination of distribution parameters. It is natural to carry out the dissolution in water.
. by necessity. 13.000
'.? type dissolution apparatus conditions. 13. 13. Also the standard deviations of the are breadth of particles are comparable with those obtained from dissolution.6). These values can be compared with the means obtained directly from dissolution data (Fig.
After the smallest particle has dissolved. and water. that the smallest dimension (height) of the particles is recognized? To answer this query.
all the particles will have become smaller by an amount of [kt](Le.
. and.The smallest particle after a time of ey es t would be kt (point 6 )a particle just about todisappear (or just disappeared). after dissolution has taken place for a time of t.866
. therefore.0376e6~*2 R2 0.3933s3x .
Parabolic approximation of particle population.. Assuming a cubical particle. but the ~ u ~ b ofthe ~ a ~ t i c lwould still bef2N).fi where N is the total number of particles in the population.4. 13.
y
. The number of particles of this size is N times. the particles originally of sizea* would have a size of a*[kt]. After a given time t.0.Frequency
a(mm)
kt
Slze
a*
a(max)
Schematic of dissolution of a multipa~t~culate.8.
Consider a (normalized) distribution shown as an example in Fig.. The ~istributionshown is that of the powder before dissolution. the mass remaining M .29233 + 2.wherefi is the number denoted by the chord FG. The number of particles at this point would be [ k t ] f i . €or instance a particle of size a* has a frequency denoted fi (the length of the chord AB). is.
13. or (b) a mixed solvent maybe used.19)
larger than the critical time).~~e Z y~ o ~ei a Z t results. (1 3. (This is shown in o g n e in issolution curves beyond t" should.. be plottable in this ecause there is alwaysa nick in the curve at short times.2 are more likely to fit a seconddegree polynomial (Fig.so that the amount of mass dissolved in this experiment 1s also insuf~cientto change the surface area. taken aseither normal orlognormal. If
is inserted in Eq. 13. it is at best normal. which is far above the solubility of the compound. 13. and there are three principles thatare used to compensate for this: (a)the useof surfactants may sufficiently increase the solubility so that meaningful dissolution can be carried out.
Water or aqueous solvents as a dissolution medium have tacitly been assumed to be the case in the preceding writing. dealing with the
40
1
0
20
40
60
80
100
( 10^4)xCD (mol/l)
Dissolution rate constants (in M/s) of albendazole as a function of cyclodextrin concentration (CD). In particular the latter can be obtained easily by the point where the curve intersects with the xaxis. for populations. (Fig. for a sieve cut.2 that. 13.. it is possible to assign values to both amln and am.10)..
.19)then a ~ i ~ t ~ . 1998.eS7 previous sections is not possible in this instance because the lower limit isnot a given size.)
* The authorsemploy 1 mgt2. f ( a ) is usually. In fact the data shown ig. but ithas been seen in Fig.. or (c) the useofcomplexation may be employed. therefore.9). There are examples for which the solubility of a compound is sufficiently low that normal USP volumes (900 mL) are insufficient to dissolve all the solid.(13. A procedure similar to the one in the where t > t" (i. In work by Diaz et al.5 mL of water. (Data from Diaz et al.
increased to a total solubility of S.
A
=
surface area
a. it was found that the purely aqueous solubility So. since A and Y are constant.21)
where A is area. The authors tested the dissolution under sink conditions (the initial parts of their curves) and constant surface area* and found dissolution to be fairly linear in time. Y = volume of dissolution medium avo= volume shape factor K = ~olt~mann constant p l = mean (firstmoment) of a particle population p2 = secondmoment of a particle population . at a given ligand concentration of Ll. but as seen in the figure they are not. the slopes should be linear in ELt]. = amln size(length.u3 = thirdmoment of a particle population
.complexation of albendazole with cyclodextrins. Since S is linear in Lt. breadth. then the slopesoftheselines should be proportional to St. and Y is the dissolution volume. St is the solubility at the ligand concentration in question Ll. or width) of largest particle a. of the medium). = amax= size(length.. or width) of smallest particle =
A = surface area of the dissolving solid at time t b = width of a particle b.e. breadth. If k were independent of the ligand (i.. =geometric mean of a lognormal particle population 13 = width of channel for dissolution study C = concentration f ( b ) = normalized frequency function for the width of a particle h = height of a particle m = mass of an undissolved particle M = ~ultipart~culate undissolved mass Mo = initial multiparticulate mass before dissolution k = intrinsic dissolution rate constant (cmls) IC = linear (cuberoot) dissolution rate constant N = number or particles in a multiparticulate sample Q = a constant r = radius of particle R = the gas constant s = standard deviation of sizes in a particle population ln[s] = standard deviation of a lognormal particle population S = solubility SL = ligand solubility in the presence of substrate t = time (of dissolution) t* = critical time T = absolute temperature.
C =(kAS~t/Y)t
(13.
then the values of jo. The integral has a solution of the following type:
for reasons stated in the
( 1 3A.9.8)
(13A.13)
Barnett MI..3)
With knowledge of f3. 13. and j 2 maybe he amount remaining at time t > t* is the value of the integral: M/{pa.j.. Pharrn Techno1 6:49.10) (13A. rather than zero or a. and ao.5)
The lower integration limit is (kt).":
{(io
kt
+j l a +j2a2}(a3.ktgda
=
J... it is assumed that particle size distribution data exist and may be approximated by a parabola.p = particle density
In this section.) the values of jo.* text. Nystrorn C (1982).9)
(13A. as shown in Fig.~) and al(a. The maximum frequency occurs at f3 and is zero at the extremes ( 1 3A.
. the maximum and minimum diameters are denoted a.3a2(kt)+ 3a~kt)2 ( l ~ t ) ~ } d a 
(13A. The parabola is expressed in equation form as:
If the distribution is known. 12) (13A.jl and j2 are known from the following facts.11) ( 1 3A. ao(a.} = f " ' f ( a ) ( a .2) ( 1 3A. For convenience..
ixson A. Brooke D (1974). New York. New York. Pharm Forum 6:71. Pedersen PV. Rod~guezHorned0 N (1986). O’Neil R (1980). J Pharm Sci 74:1322. Micromeritics. J Pharm Sci 64: 1981. ~uizinga (1979). Lancaster. J Pharm Sci 64: 1770. 13:l 55. Brown KF (1976). Lianos CME (1998). Patel M. PA. (1965). Technomic. Pharm Res 95356. Dali MV (1998). oughton ME. (1997). PhD dissertation. Int J Pharm 51:9. de VriesNijboer T. Pharm Res. Steiner G. University of VIisconsinMadison. intz RJ. Modeling and Data Treatment in the Pharmaceutical Sciences. Patel M (1975). J Pharm Sci 61:223. Dali MV Carstensen JT (1996). J Pharm Pharmacol 17:193. Dali MV. Higuchi WI. Carstensen JT. J Pharm Sci 52:163. Carstensen JT (1974). Amidon GE (1991). J Pharm Sci 63:1395. Carstensen JT. Crowell J (1931). Rowe EL. J Pharrn Sci 68:163. Higuchi WI. Statistical Analysis in Chemistry and the Chemical Industry. Carstensen JT. lnd Eng Chem 23:923. Musa MN (1972).
. Carstensen JT (1966). FranklinNL (1961). 2nd ed. p 142. Pitman Publishing. Hiestand EN (1963). Hiestand EN (1963). John Wiley & Sons. Carstensen JT. J Pharm Sci 5257. J Pharm Sci 63:344. Brooke D (1973). ora JG. Drug Dev Ind Pharm 24:637. Pharm Dev Techno1 3(3):395. Johnson KS (1989).ennett CA. VI. Cabana BE. J Pharm Sci 62:795.
4.17. ~ e ~ p e r a t udependence of the solid to solidplusre gas reaction 14.14.9.5. Photolysis in the Solid State 14.
Transformations 14. Cases of Interaction of a Liquid with a Poorly Soluble Drug 14.9.2. The Solid to L i ~ u i d . ~nteract~ons
14. Surface Nucleation (ProutTomp~ins Model) 14.8.15.6. The Ng Equation 14. The solid to solid~lusgas reaction 14.11.16.5.14.1.13.
iffusion Controlled Interactions
14. Topochemical Reactions 14.
ando om ~ e c o m ~ o s i t i o n ~ AmorphatesSpontaneous in the Crystalline
~eactions
224 228 230 234 234 235 238 239 240 245 249 250 254 255 256 256 256 257 26 1 26 1 262 263
14.5. Choice of ~ o d e l
Involving a Liquid Phase 14.
eactions via the Gas Phase
. General Interactions in Dosage Forms 14. Pseud~polymorphic
14.12. Equilibria and Effects of Applied Pressure 14.~ a s 14. Nucleation F o l l o ~ e d Fast inetics (Poly~orphic Transfor~ations) 14.1.10.1 Tartaric acid and sodium bicarbonate 14.2.7.3. The AvramiErofeyev Equations by 14.p l ~ s .
It can be shown. This has also been demonstrated by I ~ a i ~ i et ial. in a situation such as with substituted benzoic acids. and the states are referred to as “glassy” below (the highest Tg in multiple glass transition te~peratures) and “rubbery’. Reactions in the crystalline state can be attributed to the presence of moisture or light. that reactions in the rubbery amorphous state are akin to solution kinetics.” and associate this with different types of lattice defects. They attribute this to (a) presence of an amorphous phase (if material has been milled or freezedried).. (1995). These usually possess a glass transition temperature Tg. but they are rare. 1992. but exceptions exist (Sukenik al. are less chemically stable than their crystalline counterparts. Pothisiri and Carstensen (1975) have shown that. 1958) for which the crystalline state is less solublethan the molecule in solution.
.1. rather than the rule.. then the amorphous form would be more stable than the crystalline (arrangement D). 1975. above Tg. arises.. fixedin position. or (b) that processing“increases the extent of disorder in the remaining crystal lattice. Arrangement I> can also be more adverse than a random orientation. There are examples that have been reported (Lemmon et al. O’Donnel and et Whittaker.The subject of solidstate stability is of great importance in pharmaceutics.e.. a function of moisture. Amorphous materials. as shall be seen in the Chap. This is the exception. If this occurs. The most interest and the largest body of work of amorphates is in the field of macromolecules. Suppose parts A and B of the molecule depicted in Fig. Stability patterns of solid dosage forms are partly a function of the stability of the drug substance in the dosage form. the situation. Stacey et al. (1980) n and by Gubskaya et al. For instance Shalaev et al. as is to be described (Carstensen and Morris. a compound is more stable in the crystalline state than in an amorphousstate. without the interference of water or light) (Carstensen.1 react. as shown in Fig.. In general. Only a few articles have appeared in the literature on the subject of chemical stability of amorphates. 15. arrangement C would give better stability than arrangement would be farther away from I3 in the former arrangement. in general. The parallel reactions (i. 1980. 14. but solids may also undergo deco~position or solidstate reactions in the “dry” state (i. 1996). and if that is true. 1959). to a great extent. Byrn et al. If the situation exists wherea group from one molecule reacts with another group in a neighbor. the amorphous versus the defect pathway) give rise to the same reaction products. but also.. 1993). the decomposition is between two groups in the same molecule. as shown by Carstensen and Morris (1989). in a crystalline state. (1997) have shown parallel reactions occurring in the solidstate methyl transfer of tetracycline methyl ester because fitting of the data gives good biexponential fits. 14.e. molecules are.
An attempt to explain this is made in the foll~wing. One purpose of the following writing is to seek an explanation for this pseudofirstorder (or indeed. Pika1 et al.. 7. implying different the distance between possibly interacting groups (A and B). and as seen a firstorder plot results.3 results. but kinetics cannot befollowedeasilybecauseof the slow reaction rate at room temperature. amorphous indomethacin was produced by melting a crystalline form of it to above melting (162°C) and recooling it to below 162°C. Of the few reports in literature dealing with the chemical stability of compounds in the amorphous state. 1979.S.1. then stable amorphates can be formed at room temperature.1 was a crystalline solid.165.oli
ilit
Afrangement C
Arrangement D
Different possiblearrangements of a molecule in solid state. a m o r ~ h cephalosporins (~feiffer o~ et al. Van Scoik Carstensen. If an Arrhenius plot is drawn of the data from 14. then Fig. and 185°C) and assayed from time to time. The pattern is strictly a firstorder one. The content of intact indomethacin was assessed by using the U. Amorphates made in this manner are morphologically stable down to 120°C so that their chemical stability can be monitored (If the temperatures are lowered rapidly. 14. then the potential energy between molecules would be inversely proportional to a power function of their
. with the fact that in the rubbery state. and 1990) and the material developed in the following all refers to anhydrous conditions. 14. The results obtained from the melt are shown in Fig.) At a range of temperatures below this temperature crystallization occurs too rapidly to permit assessment of amorphous stability. 14.2 and 14.. It is seenthat the Arrhenius plot of the amorphate continues into the Arrhenius plot of the melt. In the work by Carstensen and Morris (1993). 1990. The decomposition curves of amorphous indomethacin and a melt of indomethacin at different temperatures is shown in Figs. 1976.3.155. 1977) also adhere to a firstorder pattern. in a somewhat frozen (or much slowed) manner of that of the melt above the traditional melting point.175.
In the presence of moisture. Amorphous samples wereplaced at several constant temperature stations (145.2. truly firstorder) pattern. in some manner. Pharmacopeia ( U P ) method of analysis. If the substance in Fig. the molecules can arrange themselves in a random fashion. conversions from amorphous to crystalline modifications are promoted (Carstensen and Van Scoik. The explanation must lie. 150. Oberholtzer and Brenner.
then a group A in molecule a interacts with gr molecule b. (Data from Carstensen and Morris. 193 1).4
Arrhenius plot of indomethacin decomposition: Squares are amorphous and solid circles are melt.)
. it wouldbe akin to a LennardJones potential (LennardJones. triangles. The fraction of molecules that have this given energy El. 1993. necessary for reaction between A and l . 185°C (rate constant 0. and group would be assumed to be randomly distributed. squares.1)
1
2
c 
E
3
4
5
2.
A.)
distance (the lattice constant) (~arstensen and orris.13 h". The energy of the molecular pair will between the group A in one of the pair.145°C (k = 0.036 h').0
20
40
60
80
Time (hours)
Decomposition of amorphous indomethacin: Symbols.05 h'). 0.015 h'). and a certa molecular pairs would be at or above a critical energy Ei. Circles. if the decomposition is an intermolecular (rathan intramolecular) reaction. 155°C (k = 0. 1993. 175°C (rate constant 0. 1993~.3
2. is given 3 olt~mann distribution (Mo~lwynHu~hes.2
1000/T
2. phous state. 1961): (14. thatis.1
2. (Data from Carstensen and Morris. 165°C (rate constant 0.19 h").
if the difference between energy levels large compared with the is roundstate energy.8) (14. where the rate constant is given by kl = 4 ~ x ~ [ . ere have been proposals ( elwynHughes. by considering the energy differences small and integrating).4) Alternatively.( E ~ E o ) / ~ T I or its logarith~ic equivalent: ln[kl] = ln[q] .. Franks.. an Arrhenius equation where the activation energy is given by (14.
__.5) If.2 demonstrate the correctness of Eq.]/[eEolRT+ e(EotAE)lRT+ ***I .8). and Fig.3) that is.9)
The data in Fig. j / N . the expecta of a firstorder decomposition).I~E. 14. one may simply approximate the series in the numerator and denominator of these equations with their leading terns. first order. 1989) that the is best described in terns of the ~ i ~ l i a ~ s equation (Williams et al. by integrating Eq..Eo) ~ / ~ ~ that is. (14.1961 . (14.~}/(e~o/~~[~ e2AEIRT+ 11 . Another.10) (14. (14.6)) / ~ ~ l ~o = kl where kl (by definition in differential form) is a firstorder rate constant.lW= ~ e ~ P ~ ( E j(14.where N is the total number of moleculesand where the summation is overall energy levels. then....2) may be written: IV> I / N = IeEi/RT+ e4Ei+AE)IRT + . (14.3 demonstrates the correctness of (14. The fraction of molecules having energies in excess of E>then N .6) and imposing IV = No at time t = 0 ln[N/No] = kl t that is.E Ea = (Ei ./RT[~ +eAE/RT+e2AE/RT +.
(14... denoting this fraction q (l/XN)dN/dt = 4IIV>.g. that is.. This leads to the same result:
>z
/ N = eEi/RT/eEolRT
= e4EiEo>/RT
[exp(Eo/RT)] = exp[(Ej . 14. a fraction of the molecules (dXN/N) reaching Ei (or higher) react. is given by
k=o
k=o k=O
k==l
There are several ways of approaching these summations (e.7) (Le. discrete approach is to assume that the energy differenceAE between adjoining energy states is constant.EO)/RT] (14. 1955):
. in a time element dt.
982. if the heat of solution of amorphate and crystalline forms separately is known. 14.85
0. + C2 = 6: ln[k] = 45. 1939. 1939) equations:
(14.TS}/{ (1". Circles: Tg = loo". It would seem intuitive that if the Arrhenius equation fits. empiricallystate thata decomposition starts at the surface of the solid and works inward.97.771 In[k] + 0. (1999) described the crystallization of amorphous lactose above et the temperature of glass transition to follow the Johnson~ehlAvrami (Johnson and Mehl.T. C2 = 5 : { T . L TI
There are theories.48 .
0.1 17{T T g } { / C (2" Tg)} R = 0.47 { T . cefamandole nafate. It is far from certain that this equation would apply to chemical reactions.Tg)} R = 0.5). (1977) employed solution calorimetry to determine the amorphous content of cephalothin sodium. 1993.2 and 14.0289 R = 0.4 shows its application to the data in Fig. Because the amorphous forms are more energetic. Lo (1976) showed that ampicillin trihydrate dehydrated to amorphous ampicillin that had much poorer stability than the trihydrate.3 as a function of assumed of glass transition temperature.977. and cefamandole sodium. (Data from Carstensen and C Morris. 14. Several different values of C and Tg will give reasonable fits. akin to the foregoing.Tg} / ( C (2".54.12)
where x is amount decomposed. they have a higher heat of solution.3. Squares: TS = 120". k is a rate constant and t. C2 = 10: ln[k] = 25.5) or as threedimensional (as demonstrated in the sphere in Fig. 14.(14. Pika1 et al.)}] = 0. plotted by the inverse function of the WLF equation. but Fig. n is an integer between 1 and 4. and the percentage of amorphate may be obtained. Avrami.40 33. 14. as seen. that simply.)
+
+
.90
(TTg)/(C+TTg)
Possible dependencies of ln[k] in Figs.80
0. is a lag time. 14. Triangles: Tg= 80". This may be visualized as twodimensional (the cylinder in Fig. On storage the decomposition appears biphasic. ~chmitt al.11) where C2 and C1 are constants. then there would be values of C2 that would make the VVLF equation fit as well. cefazolin sodium.
17)
where N is the number of particles in the sample and p is the density of the solid. At time t.Only with excellent precision. 14. therefore. em. it is assumed that the decomposition "front" progresses in a linear fashion.ktI3 (14.18) that the rate constant (k/ao)is particlesize dependent.k2t)2
where H is the height of the cylinder.k2t)2/{H2Z(R)2] (1 .Examples of topochemical reactions.x is )
) = (1 . reaction orders in the solid state on purely stat~stical ~ o u n dand other infor~ation must be available before a mech~nisticmodel can be assigne~.
. the solid can be visualized as a cube originally with side a.16)
That is. ) would be given by o [a/ao13= [I . would be
HZ= H z ( R
. directional diffusion. (14. and with a fairly large number of assays and a sufficiently large decomposition. It is d ~ ~ c to dis~inguish e t ~ e e n u l ~ b ~ s.18)
It is noted from Eq. This is shown in Fig.. The original volume of the solid was Nai so that the fraction not decomposed (1 . In general it is not possible to distinguish between a reaction of the type described by Eq.h)2 = H z ( R . a layer h will have decomposed at time t ) and h
=I=
k2t
The amount remaining undecomposed at time t.x.
. will it be possible to distinguish between the two.(k/ao)t13 1 .[k2t/R])2
(14.6. there will be an amount undecomposed given by Npa3 = Np[ao .15)
For threedimensional.x = H2n(R .x = ~ p a ' / [ ~ p a= ] ~ (14. would be a = a. 1975). therefore. An example of this type of decomposition pattern is aspirin in an alkaline environment (Nelson et al..
For a cylinder ofradius R. Because the original volume is H ~ z ( R the )~ retained fraction (1 .e. the decomposition will work inward in a zeroorder fashion (i. (14. This is akin to physical phenomena such as crystal growth (the socalled McCabe law). so that after a given time the side length a.18) and a firstorder reaction.kt
(14.
Such a model would require a firstorder decay..ln[l . then.
k. 1975. although some instances have been reported in literature.1 . (Data from Nelson et al. then
I. (1974) have shown firstorder decomposition to occur for the dehydration of ampicillin trihydrate. that is. and Pothisiri and Carstensen (1974) have shown this to be true also for paminosalicylic acid.X] kt3 111
+ kl t .19)
where No is the original number of nuclei at the temperature to which the solid has been brought from a low temperature.[(klt)2/2!]+](kl t)3/3!J)
(1421) (14. and if this is followed by rapid twodimensional growth. and when there is ingestion of nucleation sites and the growth nuclei can overlap.. This rarely occurs. then in the decay period (t which is one common form of the Avrami equation. Shefter et al.1
2
3
Days
Aspirin decomposition in a solid. and k here are constants. alkaline environment. the decomposition in the solid phase of the decompositio~.x] = Q{ekt ..111 .. Shefter and Kmaclc(1967) studied the dehydration of theop~yllinehydrate and found it to follow a firstorder pattern.22)
f the lag time is denoted t.is first order.when the number of nuclei follow dN/dt = Noeklt (14. Jacobs and Tompkins (1955) have summarized the AvramiErofeyev equations as follows: When the nucleation is according to an exponential law. to be covered later. the AvramiErofeyev equation takes the form
.)
lecomposition is most often associated with active sites that start as nuclei. In Bawn kinetics.
> ti) this reduces to >
. In many situations the nuclei will grow and then overlap.
22)
__
Time.6
0 0. 1939) will be dealt with in the following in a somewhat simplified manner. 1994).22) to decomposition data. The program calculates a series of data according to Eq (14.1 0. and theophylline has been used as a model for several studies of this kind (Lin and yrn.080 0.The program in Table 14.396 0. Agbada and York. The previous sections have dealt with decompositions that occur randomly in a space or on a surface.211 0. (14.5 STEP . (14.2: 160 NEXT T
Decomposition Data According to Eqs. 21) and (14.22) (Fig. 14. The section to follow will deal with the situation in which decom~ositi~n tied to particular sites that are created as a function of is time.8). and that decomposition may occur in these areas and not in the areas that at time t still remain 6 6 n o n ~ u 
Program for Eqs.0’77 0. Suzuki et al.027 0.00950 0.7). 1989.22) 100 FOR T = 0 TO 1.Y 1.21) and the tabulation (see Table 14. In the hydrate water molecules form part of the matrix.001 0. 14. (14.2 0. in particular.399
0 0.327 0.21) (Fig.064 0.1 110 X1 = EXP(T) 120 X2 = (TA2)/2 130 X3 = (TA3)/3 140 V1 = X11 +T+X2X3 150Y2 = 1Y1 160 Y3 = LOG(Y2) 1702.2 demonstrates the danger in simply applying Eq. The AvramiErofeyev model used for this type of kinetics (Avrami.0955 0. to pseudopolymorphic transformations and dehydration kinetics of hydrates.2) gives the possibility of g~aphing according to q.509
0 0. and Eq.216 0. Dehydration kinetics of hydrates has had the attention of the pharmaceutical scientist for some time. 1979. kt
X
.512
.036 0.129 0. The model assumes that volumes within the solid at a given time t are activated.4 0.Y3.ln(x)
(k03

0 0.080 0. (14. = T A 3 150 PRINT T. This type of reaction has been assigned quite frequently in recent literature.21) and (14.3 0. (14..0362
0.5 0.1 and the printout in Table 14. (1982) has developed a generalized kinetic theory for isothermal reaction in solids.
ter 1
Time, kt
.7 Decomposition. data in Table 14.2 plotted according to Eq. (14.21). cleated.” This may occur in strings (onedimensional diffusion), areas (twodimensional diffusion), and volumes (threedimensionaldiffusion). The a p p r o ~ i ~ amathte ematical development follows the same path in the different cases, and only the latter will be derived. For simplicity it is assumed, in Fig. 14.9, that only the volume cornered by A is nucleated and the rest of the solid isnot. This could equally well have been scattered volumes ofa total volume equal to the condensed volume shown in Fig. 14.9, and the result, therefore, will be the same, except that, in the scattered case, the volumes may “grow together.” This is not considered in the model (but will be considered in the ~routTo~p~ins model). If the nucleation occurs zero order ineach. direction, then the sideof the nucleated cube, at time t, is kt, so that the number of nuclei, N 3 , at time t is
1.5
0.5
0.0
0
1
2
3
(kt)*3
Deco~position data Table 14.2 plotted according to Eq. (14.22). in
Schematic for approximate AvramiEreyefov model.
N3 = (k* t)3
(14.23) (14.24) (14.25)
If the nucleation occurs in a plane, then (twodimensional case)
N2 = (kt)2
and if it occurs along a line (a string), then (onedi~ensional case)
N , = (kt)
Figure 14.9 appliesto the threedi~ensionalcase, and the decomposition is assumed to be (a) inlinewith firstorder kinetics proportionalto the concentration of unreacted solid in the nucleated volume; (b) proportional to the number of nuclei and, hence, inview of Eq. (14.23) proportional to(kt)3; (c) not occurring at all in and the nonnucleated volume. This reasoning leads to
d( 1  x)/dt = q(l  x)(kt)3
(14.26a) (14.26'0)
in the threedimensional case, and in general to d(l
 x)/dt
= 4(1  X)(&)"
n being unity, two or three, depending on the dimension. Equation (14.26b) may be rewritten:
d ln[l  x] = qkntn which integrates to ln[l (14.27)
 x] = [qk"/(n
+ l)]{(t""f')) =  exp(&~)t(*"))
=
(14.28)
in line with the expected linearity in Fig. 14.8. Qr Taking logarithms of Eq. (14.28) now gives In( In{1  x = &I ) }
[gkn/(yl l)], here is a constant. (14.29)
+
+ (nln[t] + 1)
which is the conventional plotting mode, as employed by Dudu et al. (1995). These authors used micr~calori~etric methods and showed the dehydration of theop~ylline hydrate to be a twostep process obeying the equation
[ ln(1 = kt
y1
(14.30)
= 3. Hence, in their case, the process is a
which is a variant of Eq. (14.28), with thr~~dimensio~al, diffusional process.
r l
olymorphic transformation rates have lately become of importance; an example is a recent article by York et al. (1994), dealing with the dehydration kinetics of theophylline. The article by Ng (1972) is similarly instructive in the sense that it reviews all the equations that have been developed for polymorphic transformation kinetics. Usually the transformationkinetics are Sshaped curves, and before any model is imposed on the data, the following model should be considered. (This is comparable with the model proposed by Carstensen and Van Scoik, 1990): If the phenomenon that governs the t r a n s f o ~ a t i o n essentially the nucleation lag time, then the is curves may be considered as representing either a noma1 ora lognormal error curve and the mean would be the mean (or geometric mean) nucleation time. What this states is that each particle, in a sense, acts as itsown entity, that there is a nucleation time (with an error or a variance attached to it), and the particle will endure the nucleation time, and then decompose, individually, very rapidly. The reason for the lognormal relation is not difficult to rationalize. Solids are usually lognormally distributed. If the nucleation time is inversely p~oportional to size, then it, too, would be lognormally distributed. TOjudge whether such a relation pertains, the fraction decomposed is, therefore, converted to a cumulative Zvalue (by means of normal error table), and this a is plotted versus either t or ln[t], to yield a straight line: (14.31)
= 0 corresponds to the average nucleation time, tavg, that is
tavg
= exp(Q1/kl or
tavg
= Q2lk
(14.32)
Qvalues would correspond to the standard deviation of the nucleation time. ehydration, at times, results in a morphic transformation. For instance, Lo (1976) showed that the transformation of crystalline ampicillin trihydrate to amorphous penicillin was primarily firstorder and either was firstorder or followed a contracting cylinder model [(l  x)1/2being proportional to time].
f a solid is placedin a vacuum and exposed to temperatures at which it decomposes at a measurable rate, one of the following situations may arise:
I II III IV V VI
Solid + solid + solid Solid + solid + liquid Solid + liquid + liquid Solid "+ solid + gas Solid "+ liquid + gas Solid + gas + gas
ther schemes are theoretically possible, but notlikely.Of the foregoing, it is schemesIV and V that willbe treated insome detail in the following, because they are the ones most investigated in the pharmaceutical sciences. It will later be
shown that most pharmaceutical systems will not be of such a “purist” nature, but the experiences gathered from examining themwill throw light on several important, reallife situations.
Not all Sshaped curves will neatly fit topochemical or Avrami equations. The data in Table14.3 represent an Sshaped curve and were obtained by a reaction that produced a solid and a gas, and if plotted by Eq. (14.29) then Fig. 14.10 results. The plot may, at first glance, seem fairly linear, but the point is t t ~ ein~ ~ the deviations from the line are (+ ) (part AB), () in part that , (+) in part CD. Itis visually obvious, as well, that the curve is still Sshaped. again Such curves also fail to give an integer (2, 3, or 4) as dictated by the mo The solid + solid + gas type of reaction has been investigated by Tompkins (1944), used who potassium permanganate as a model substance. tical solids have been tested later [e.g., ~aminosalicy~ rnblum and Sciarrone (1964) and by Carstensen and typical example of such reaction is shown in Table 14.3and the readers may satisfy a themselves by plotting x versus t, that the plot is, indeed Sshaped. No solid has a smooth surface (Le., there are always surface imperfections). These could be“steps” in the surface or they could be crystal defects. These sitesare more energetic than the remaining sites. They are most likely to occur at surfaces, which, in any event, are populated with molecules that are unlike the molecules in the bulk of the crystal. For instance they have at least one less neighbor than bulk molecules. It is assumed that decomposition is more likely to occur at such “”activated” sites (Fig. 14.11). Once a molecule decomposesat anactivated site it changes its geometry; hence, the neighboring molecules are more likely to decompose. There will then be a chain or plane of activated molecules forming, with a probability of a (see secondfigure in Fig. 14.1 1). The a, of formation of activated molecules, N in number at time 1, is rate dlV/dt, and this is proportional to N , Initially this is then given by [dN/dt]o = a [ N
+ No]
(14.33)
Decomposition Data of 4.6 mol of a Solid Following the ~ r o u t  T o m ~ ~ i n s Model
Time ( t )
Gas (mrnol)/ 4.6 mmol solid decomposed
0 0.08 0.46 1.16 2.37 3.20 3.76 4.15
Mole fraction x
ln{xl(l  4
1
0 1 2 3 4 4.5 5 6
0.017 0.1 0.252 0.515 0.696 0.817 0.902
4.034 2.197 1.087 0.061 0.82’7 1.499 2.222
y =
 4.1317 t 2.7981~ R"2
=Z
0.995
Data in Table 14.3 treated according to Eq. (14.29).
It is obvious that after even a short period of time N becomes much larger than No, so that this latter can be dropped at times even remotely larger than zero. After a certain while (seelast inset in Fig. 14.1 planes will start tomerge, and l), hence there will be a termination probability / , so that at measurable times, Eq. 3 (1 4.33) becomes d N / d t = {a /3}N (14.33) 0th a and /3 are functions o f t (or what is equivalent, to the fraction decomposed x). It is reasonable to assume that
a =b
at
t = tl/2
(or x(14.34) = 0.5)
that is, at the time point at which onehalf of the substance has decomposed. Also,
/?=O
at
t=O
(orx=O)
(14.35)
for there can be no termination probability at time zero. One (not necessarily the correct) function which satisfies this condition is
/? = 2xa
hen this is inserted in Eq. (14.33) one obtains
d N / d t = a[1  2x]N
(14.36)
(14.37)
tion
Schematic of model leading to ProutTompkms kinetics: A and B are active surface sites. Propagation of A proceeds AC (third inset), as propagation at I starts. 3 Branc~ingthen occurs at C, and finally there is termination of one (or the other) of the branches.
7
The decomposition rate dx/dt is proportional to N; Le., dx/dt = kN or
N = (l/k){dx/dt}
(1438) (14.39) (14.40)
 2x]dx/dt
Equation (14.37) can now be written dN/dt = (a/k)[  2x1 dx/dt 1 Chain differentiation of dN/dt gives dN/dt = [dN/dx] [dx/dt] Introducing Eq. (14.39) into Eq. (14.40) gives dN/dt = [ d N / d ~[dxldt] (a/k)[l ] dN/dx = {a[l  2x]/k} which integrates to
N
==
(14.41) (14.42)
dx/dt is canceled out of the last part of this equation to give
(a/k)(x x2 )
(14.43) (14.44) (14.45)
Since, by (Eq. 14.38), N = (Ilk) {dxldt}, it follows from Eq. (14.43) that
(I / k ) dx/dt = ( ~ / k ) x ( X ) 1
which integrates to ln[x/(l  x)] = a(t  t1,2)
The equations have a zero time problem, because the equation is not defined for x = 0. This is a consequence of neglecting No. Similar paradoxes exist in the scientific literature. The Gibbs adsorption isotherm, for instance, is not defined for concentration, C = 0 (i.e., for a liquid without surfactant). In solidstate stability, it might be thought of in the vein, that as the material is being produced (Le., at time zero; e.g., through recrystallization)?it is already decomposing (however little). Data are plotted according to Eq. (14.35) in Fig. 14.12, and the linearity is good. There are several other aspects that may convince a scientist that this is the type of reaction at hand. First of all, Arrhenius plotting is good, and the activation energy is us~ally three to four times as high as in that o other reactions in the solid f
 6 1 " " " " * " " f 0 1 2 3 4 Time
5
6
7
Data from Table 14.3.
(and ~ i state.~The reason ifor this is that the ratedetermining parameter in Eq. ~ ~ ~ (14.45) is a (i.e., it is actually a propagation probability that is measured, not a rate constant in the usual sense). Whenever a compound “melts with decomposition,” then there is a good possibilitythat the melting range” depicts the interval in which the reaction occurs with a measurable rate, that it is too slow below this range, and too fast above the range, and in such a circumstance the activation energy is high, ompkins reaction may most likely be applicable. n the halflife is in order. There is frequently a substantial lag time (and other solid) type reactions. ecause many are carried out under vacuum (e.g., when breakseal tubes are used, or when manometers are glassblown directly unto the reaction vessel), and heat transmission, therefore, is poor, so that it will be a while before the solid itself actually attains the elevated temperature. An experimental remedy is to test the heat transmission by checking the length of time it takes for a stable solid substance with knownmelting point and heat capacity to melt at that temperature, and to do this with three substances (benzoic acid being one), a calibration curve. If it is then calculated that at a given test tem, it takes t minutes for the solid to attain the given ternpe e may be obtained by the integral mean value theorem. e subtracted from the time points used. s plotting, this does not apply, but it may be a source of The solid ”+ solidplusgas reaction embodies the dehydration of solid tes. Leung et al. (1998a,b)haveshown that aspartame 2.5 hydrate cyclizes outTompkins kinetics andthattherateconstants follow an ~ r r h e n i u s equation.
I t s h o u l ~ pointed out, thatthe solid to solidplusgasreaction may be so only over be a certain temperature range, or to a certain degree of decomposition. Figure 14.13 shows the eutectic diagram of a compound A with its solid decompositionproduct If the study is carried out at temperatures below the eutectic temperature T*, then the reaction will be solidto solidplusgas.If above the eutectic t~mperature, then the reaction will be solidto solid plus liquid plus gas. (If abovethe highest melting point, then it will be liquid kinetics.) The compounds reported in literature to be of the solid solidplusgas type are most often inorganic salts (e.g., p~tassium permanganate rout and Tompkins, 1944); silver permanganate (Coldstein and Flanag and some organic compounds, such as oxalic acid, ~aminosalicylic acid ( arrone, 1964; Pothisiri, 1975a,b), or indomethacin (~arstensen a Isen et al. (1997) showed cefaclor monohydrate to decompose (as judged by related substances) by firstorder kinetics. The rate constants could be plotted by ~rr~enius plotting and were consistent with ambient rate constants. The reaction scheme, whenamorphous material was present, was such that the rates were faster at time points and then becoming equal to those of the cry stall in^ m~dification. of us onclusion was that the initial phase was decom~osition a ~ o r ~ h ocontent parallel to conversion of amorphous to crystalline drug.
Liquid
E
Tompkins
Kinetics Compound A
omp pound B Decomposition Product
r Fraction of B Mole
Eutectic diagram of a compound and its decomposition product: At ternperatures higher than the melting point of B only liquid kinetics would be expected. At temperatures lower than the eutectic point only solid state kinetics (e.g., routTompkins kinetics) would be expected. In i ~ t e r ~ e d i atemperatures, socalled Bawn kinetics apply. te
At times the solidstate reaction cannot be completely specified, yet may be described in analytical terms. Tzannis and Prestrels~i (1999) described the effect of sucrose on the stability of trypsinogen, during spraydrying, by plotting denaturation t e ~ p e r a t u r e ~ a function ofmelting temperature and found a linear increase as between residual activity after spraydrying, and melting temperat~re.Adler and Lee (1999) have reported on the stability of lactate dehydrogenase in spraydried trehalose.
There are a m~ltitude “types’’ of Sshaped curves, and one, somewhat distorted, of shape is as shown in Fig. 14.14. Ng (1975) suggestedthe f o l l o ~ i global, em~irical n~ equation for this and other types of solidstate decomposition: dx/dt = kx”(l  X)’ (14.46)
If both y1 and p are unity, then the equation becomes the routTompkins equation. A set ofdata illustrating this is shownin Table 14.4. Theseare the data onwhich Fig. 14.14 is based. n the first two columns of the table the time required for decompositions of 0, 0.1,0.2, *, have been d e t e r m i ~ ~(Data treatment is actually easier if random times d. are used, with the associated fractions decomposed.) The average decompositi~nsat interval midpoints are then determined (columns 3 and 4), and the value of dx/dt is then calculated (as shown in the table footnote^. The Ng equation may be expressed in logarithmic form.
0
ln[dx/dt] = y1 ln[x] p ln[l  x]
+
+ ln[k]
10 r
Time
Sshaped curve following the Ng equation: data in Table 14.4.
If the data in Table 14.4 are transformed and l n [ d ~ / d is Imultiply regressed against ~ ln[x] and ln[l  XI, then values of yz = 2 and p = 3 are obtained.
any more compounds seem to decompose by this reaction scheme than by the solid to solidplusgas one. As mentioned in the caption to Fig. 14.13, this type of reaction
Example of Data Amenable to Treatment by the Ng Equation Fraction x decomposed Average time, t Average fraction, x decomposed dxldt (from curve)
Time
0
1.371 1.859 2.183 2,472 2.792 3.226 3.982 5.935 18.280
0
0.686 0.1 1.615 0.2 2.021 0.3 2.328 0.4 3.632 0.5 3.009 0.6 3.604 0.7 4.959 0.8 12.107 0.9
=
0.05
0.0729 O.204ga 0.3087 0.3450 0.3125 0.2304 0.1323 0.0512 0.08 10
0.15 0.25 0.35 0.45 0.55 0.65 0.75 0.85
"Obtained by: 0.1/(1.8591.371)
0.1/0.488
=
0.2048
kinetics is usually referred to as awn kinetics (Bawn, 1955). The sit~ation at time f is as shown in Fig. 14.15 and, as seen, there willbe a certain amount of liquid decomposition product. This amount corresponds to the amount of drug decomposed. However, the liquid decomposition product will dissolveparent compound to the extent, S (mole drug per mole decomposition product), to which it is soluble, so that the amount present in the solid state at time t is the original number of molesAo, minus the amount decomposed Aox, minus the amount dissolved, AoSx. The rate of decomposition would be the sum of the rates of decomposition in the solid state (assumed firstorder with rate constant k,, time" 1)and in the dissolved state (assumed first order with rate constant kl time"). The rate equation is hence dA/dt = k,[Ao(l  X )  AoxS] Noting that
A/Ao = (1  x )
 kl[Ao~S]
(14.48) (14.49) (14.50) (14.51) (14.52) (14.53)
it follows, by division through by A. that d( 1  x)/dt dx/dt where
B = [(kJk,)  l]S  1
k,[l  X  XS]  k l ~ S
or, noting that d(l  x = dx )
= I
k,[l  X  xS]
+ k l S ~ k,[l + Bx] =
quation (14.52) may be integrated, and yields ln[l
+ {Bx}] Bk,t =
Using B as an adjustable parameter, it is possible to find the value that makes the data profile through the origin, as dictated by Eq. (14.53), and also gives the best fit. Figure 14.16 and Table 14.5 show an example of data from decomposition of ~methylaminobenzoic acid. To plot this according to Eq. (14.53) it is necessaryto assume values ofB, plot the data, and assess the goodness offitbysome criterion. A different valueof B is then chosen, and this process repeated until a "best" value of B is arrived at. It is possible to show that in general the sumsof the squares of the deviations
Situation leading to Bawn kinetics.
85. All the correlation coefficients are good. which should be zero. this is also not a good criterion.40
20
0
200
Hours
Data from Table14.45 at this point. and2. At this time point. A n as different criterion is the correlation coefficient. the pressure is monitored as a function of time.45 = 1. t*. 1972. InEq. and x* = 0.using the correlation coefficientis nota good parameter. 14. Ao(l . The best criterion would be a criterion that dealt with curvature. but a simpler one. is simply to note the intercept. One can then in three or four tries arrive at a “best” value for B(= 0.x*)/.0. Eq. as stated.53) is valid from time zero to time t*.55) m0l/m01
Decol~~osition for ~ . the assumed value of the program run. it follows that
S = 0.*
(14.~ ) ~ /( 2)) of the points from the ensuing line is used a criterion.22 (14.1.7.: s (
= . At a given time point (which is quite reproducible).ats son statistics) are the best.5: ~ e c o m ~ o s i t i oof ~m~thylaminobenzoic n acid. (Data from Garstensen and Musa. (14. because it simply increases with increasin~ values of B up to avery high (unrealistic) value. Frequently. (14.85). present at time t*. the amount not decomposed. (14. It is best to do this by computer.54)
where x* is the mole fraction decomposed at time t*.x). ~urbin. In this. The number ofdata points are inserted. and a simple program in BASIC is shown in Table 14.6for three values of B (0. Fitting the data in this fashion is shown in Table 14. is just sufficient to dissolve the amount of liquid Aox.
x(y
s = (1 .~ e t h y l a m i ~ o b e n zAcid Data oic Time (h)
0 0 50 290 110 210 4 1 2
150 27.g. If t* = 350 (as in the example used here). and criteria for linearity (e.0).18).)
.. the last trace of solid will disappear (Fig. For data fitting to Eq.53).5
310
350 45
.9 8 20. Studies of this type are usually performed on a vacuum rack. also resulting in a very highintercept.53) the line must pass through the origin. and the sample can be observed. Therefore.55/0.
615 0.28 DATA 350.SPC(6).53) ln[ 1 Time (h)
B == 0.20 DATA 310.481 2.890 3.210 3.85
50 100 150 210 290 3 10 350
0.200 2.SPC(6).C.2 DATA 150.588 1. (14.y.N1 UT “Iteration Parameter.5) PRI~T “Correlation Coefficient = ”.095 0.( Z 5 * x l ) ) / ~ 2 Z6 PRINT RINT “Slope = ”.510
.45 Z2 = X2 .610 2.SPC (6).4 DATA 210.182 0.(Xl*Yl/N2) Z5 = 24/22 (Y l .1
+ Bx]
B=:!
B = 0.((X1 “2)/N2) z 3 = Y2 .710 4.Y IF N2 = N 1 goto 700 DATA 50.Z8 Z9 = (Z3 .Z9 =
¶>
=: I
.993 1.5 Treated by Eq. in 400 block” INPUT “Number of Data Points = ”.040 4. B = ”.334 0.((Z5*2)*Z2))/(N22~ PRINT “syxA2 ”.705
0.7 Program for Obtaining BestValuesby Manual Iteration
100 110 120 130 140 200 210 220 230 240 250 260 270 280 300 310 400 410 420 430 440 450 460 700 710 720 730 740 750 760 770 780 790 800 810 820 PRINT “Type in data as x.054 2.677
1.((YlA~)/N2) z 4 = z 1 .SPC(6).“LN(1+ BX) PRINT ‘‘ READ A.8 DATA 290.Z5 PRINT “Intercept = ”.099 1.l DATA 100.335 1.“X7’.C X = A Y = LOG(1 + B*C) X1 = X1 + x x 2 = x 2 + (XA2) Y1 = Y1 + Y Y2 = Y2 + (YA2) z 1 = 21 + (X*Y) N2 = N2+1 PRINT ~. Z6 27 = (Z4^2)/(Z3*22) Z8 = (27)^(0.099 1.B NT “T”.Data in Table 14.830 3.
] it follows that
k.x) = kl ( t . for ~methylaminobenzoic acid. The density of the liquid will actually change with time.t*) is (1 .012 h" kl is now calculated from Eq. therefore. The time is counted from t = t*.53).t*) )( *] or
(14.60)
ata of this type.03 found from the first part of the curve.59)
In[(l .x*) p.x*)/[Ao/p] (I . It is noted that when the total curve is plotted (Le.19.1
4
0
2
4
6
Storage (Years)
. mol.01 h".040 h' in quite good agreement with the value of 0.18 and 14. The Moles/cm3 density is denoted p and since there is a total number of A.x/ I . = slope/B = 0.58)
that is kl = 0. Unlike the ProutTomp~ins curve.85 = [(k1/0. (14.p] = k l t ) or
+ In[( 1 .22 . but it is assumed that both parent drug and decomposition product have appro~imatelythe same density.6 treated by Eq.
.85 = 0.x . 14. The first order rate constant obtained from this plot is kl = 0.11 1. (14.1
(14. one part relating to the phase where there is solid present. then an Sshaped curve results.
The slope in this case is 0.18 and 14.x p. and should decompose by firstorder kinetics. so that
*
In[( 1 . are presented in Figs.57) (14. Ao(l .x*) p]
a
(14.52)..012) . It isseen that the data are quite first order. the volume of liquid is Ao/p.01/0. 0.19 are combined). awn curve is a twophase curve. The initial molar concentration (at time t*)is. when Figs. the other to the part where all solid has dissolved.17 Data from Table 14. Since the slope is [L3k. and the = ) concentration at time ( t .03 h"
eyond t* the systemis a solution system.56)
(14. 14.
is in contact with another such solid . 19'72. 19'72. This is actually the value of the liquidus line on a eutectic diagram.)
.)
The values of x* obtained at t* will differ from temperature to temperature because the solubility is a function of temperature.5
1. A+B+C
00 .8
06 .. The melting point depression curve (Maron and Prutton.x = 0.18 treated according to Eq. (Data from Carstensen and Musa.62)
Figure 14.ilit
1.21 shows a situation where an ideally shaped solid A.20.55.61). 1958) is given by ln(1 x*) = (~ff/R)[(l/Tf) (1/T)] Such plots are quite linear.5
2. 14. as shown in Fig. that is. It is assumed that A in this situation. 1 . (Data from Carstensen and Musa. (14.
~ecomposition ~methylaminobenzoicacid after t* (350 h).
0.o
1. at which point of x = 0.o
0. 14. The contact area is assumed to be 1 cm2.0 350
375
400
425
450
Hours
Data in Fig.45 Le. (14.
. 3 7 ~ (R= 1.64)
ick's first law. y = 16.x*] as a function of lOOO/T: leastsquares equation.6 .65)
(14.73
2.)
As the reaction proceedsdecomposition product C will accumulatebetween A t a given time t .00). 1981. (Data from Carstensen and Kothari. hcm thick. necessitating diffusion of one of the reactants through the decomposition layer. A layer of B.68
1ooonr
ln[l . dB/dt is inversely proportional to h.19 . so that we may write pdh/dt = q/h (14. (Data from Carstensen and Musa.66)
Interaction betweentwo solids withdecompositionlayerseparating the two reacting species. hcm thick would contain
(14. 1972. for the reaction to density of I3 is denoted p. compound A must diffuse to the surface of hrough a layerof compound C.)
.
A system of this type is. some of them already alluded to. shows a literature example of such data.x ) ’ / ~= kt/ao }~ 2
(14. It is claimed that these data best fit a Jander equation (and such treatment is shown in Fig. it is obvious that the phase C in the Jander model (see Fig.(1 .x)) = kt (1n(1 .8. but at lower t e ~ ~ e r a t u r e s . again. It is seen that the rate constant is related to the particle size (i.(1 .21 are cubical.. but first of all the fit is not good. in two dimensions ~ i f f ~ s i o nthree dimensions (Jander equation) in
x2 = kt
(1 x)ln(l . forms of the Ng equation).. and the equation that gives the “best fit” is then assumed to be the mecha~ism.
~q u a tio n s Relating to Decomposition in the Solid State In@/( 1 .21) cannot possibly apply to a polymorphic transformation where the reaction is simply A + e ~ p h ~ s i hered t ~ u sorting out ~ e c h a ~by s~@tistical ~e t i s ~ ~ analysis can be fal~acious.x))”/”== kt 1 . 14. and a at time t.x)1/3)2= kt
. the autodecomposition of aspirin is higher than the diffusi coefficient (r~lated q). are listed in Table 14. If..x)l’n = kt Surface nucleation. There has been a tendency in recent literature to simply fit data to several (or all) o f these equations.g. Several such equations. as indicated in the lower line of Fig. Figure 14.71)
where x is fraction decomposed. ProutTompkins equation nDimensional nuclear growth (Avrami and Erofeyev) ~ . initially.~ i ~ e n s i onucleus growth nal nDimensional boundary reaction Diffusion in one dimension Diffusion.e. the aspirin. for instance.ilit
This may be integrated to
h2 = [24/p]t = k’t
(14.68)
A
h=aoa he amount retained at time t is (14.x ) + x = k t (1 . of side length a.70)
or
{ 1 .24). 1974).(1 .23. 14. ecently. and second.. the finer the particles the larger the rate constant).69)
(14.odium bicarbonate system. it has become customary to compare polymorphic and p s e u d o ~ ~ l y c transformation data with prevailing solidstate equations (e.IS)] (Nelson et al. and if
(14. and the reaction at higher temperatures then follows [see to .67)
or:
h = [kI t]1/2
k’ = 2g/p. 14. higher temperatures.
The amorphate showed a (slight) exotherm peaking at 81.6"C. Otsuka et al. and a sharp endotherm at 167.5"C.8"C and a sharp endotherm at 166. From this it would be reasonable to conclude that form I1 is stable at room temperature and t r a n s f o ~ s I at 116. (1999) have studied the dehydration and rehydration of diclofenac salt hydrate at ambient temperature.4"C.296
. 14.8'42. form I1 showed a slight endotherm at 116. and amorphate). 11.) I
Several modelistic investigations in this field have appeared in recent years. presumabl~owing to crystallization. (Figs. (1999) investigated three forms of glybuzole (I. 1999.72)
0.
(14.23 and 14. Neither form I nor 1 changed after storage at 40°C at 75% for 2 months.3"C. Fini et al. (Data from Otsuka et al.o
1.
.0
0.y = 52. 1956).24) and found all to have fairly 1 much the same solubility.994
l/(Tmax)
Kissinger plot of polymorph T of glybuzole.5
1. The authors estimated the polymorphic stability of form I bywayof the T issinger equation (Kissinger.723~ R"2
= 0.5
Time (hours)
Literature data dealing with two polymorphic transformations allegedly diffusional because it ahderes (somewhat) to a Jander model.. DSC for form I showed no peak other than a sharp them at 167. this latter form being stable at the higher to te~peratu~es.24.
e. in one of the cases to be discussed later (tartaric acid + sodium bicarbonate) this is true (in spite of the fact that the tablet can. particles are fused together (by either brittle fracture or by plastic deformation in tablets or tamping in capsules). such a model) cannot be conceived of in a polymorphic transformation. different Tmax values result. and R is the gas constant. and if the created contact area is between two different components of the tablet (one being the drug).g. Ea is the activation energy.” and such a layer (i..3
0. 15 is devoted to it.1
0.5
0. The topic discussed here will be of special cases in which water is not the interactant (or the main interactant). In fact.6 0. 6 5 2 4 . but this cannot be true.0. the Jander equation is based on an assumption of a layer of “reaction product.99 = 49.22 treated according to a Jandermodel. for all practical purposes.723 x 1.6
0. and in the case of glybuzole there were four such values.2 0.0 1.1
0. it is noted in the curve in Fig.4
0. It is highly likely that moisture plays a part in all of these. because what would be the “reaction product”?
It is tempting to think of a tablet as an agglomeration of individual particles. The following illustrative examples will be discussed: 1. 1993. Aspirin and phenylephrine 3. then there is the possibility ofinteraction. amorphous glybuzole). By their mere nature.
.20. independent of one another. 1999) employed the Jander equation to explain crystallization rates of compounds (e.0
0.This is a large topic in itself.2
Data from form B in. and Chap. If the experiment is conducted at different heating rates. The curve follows the leastsquares fit equation: J ( t ) = 0. Fig. Otsuka et ai. Tartaric acid and sodium bicarbonate 2. 14. However.4 0.
where 43 is the rate of heating.2 kCal/mol.194 + 0 ..24 that a lag time sometimes has to be invoked for the data to linearize. Tmax the temperature at the peak maximum in the is SC. The most common type ofinteraction in solid dosage forms is actually between water and drug. be anhydrous at the onset).00. Aspirin and lubricants In addition to the points made.8 1. (1991. 14.2
0. It can be seen from their graph that the activation energy is 24.
This is so sensitive that during manufacture extra precautions are taken to keep the relative humidity of the processing areas low. however.” Theproduct wasa (molecular compound type) interaction between niacinamide and ascorbate. Filibon. the product was quite “stable. If. When dissolved in water the individual components will regenerate. such that the solubility is maximum at a given indomethacin concentration. niacinamide and ascorbic n small scale. is a disadvantage. It contained among other vitamins. because if the reaction does occur in the solid state. the capsule “hardened up. and if so in what sense? 2. forming carbon dioxide. in solid dosageform formulation. It. The resulting powder was bright yellow. When the tablet is added to water. duringwhich the capsule contents were exposed to the moisture in the soft shell for longer times.
This is a common combination in effervescent tablets.The formation of molecular compounds as discussed in Chap. is of importance is . in which time lapses are short. attimes. (14. the author was in charge of scalethe up of a softshell capsule product.” The solidstate reaction has been investigated by Usui and Carstensen (1986) and Wright and Carstensen (1987).” in fact became a “bullet. the initial moisture is not low enough. Is moisture important. but the tubes in which effervescent products used to be sold were not tight. which produces the desired bubble effect. and (c) on “reconstitution” the bubbleeffectwillbereduced to the extent carbon dioxide was lost in storage. once marketed by American Cyanamid Company (Lederle). The solubility of the complex varies in a fashion. and the carbon dioxide could escape. Hence. t is necessary that this reaction does not take place before the time it reaches the consumer. ogdanova et al. the acid and the base will react. The same is true to a great extent in plastic bottle and in plastic blister packs. but the problem that the reaction (as shall be demonstrated later) is catalyzed by moisture. under the heading of ~ ~ t e c t i ca~type of solidstate interaction. and the problem was rectified by carrying out the reaction before blending the powders.73) or is it
. Whenthe reaction occurs in the solid state. For instance. there are two questions that present themselves:
1. and the aluminum foil has become apopular means of doing this. The niacinamide and ascorbic acid were simply mixed in a blender and “granulated” with ethanol. What is the stoichiometry? Is it that of Eq. 11. (1998) shown have a solidstate interaction between indomethacin and nicotinamide.
+ 2Naf + 2
To be strictly correct. and the internal pressure will cause the aluminum foil to “balloon. The evolution of carbon dioxide would normally build up pressure in a glass bottle. then (a) carbon dioxide will form in the container.
2(COOH)2+ 2NaHC03 + R2(COO). one must also pack the products in hermetic containers. other words. then the reaction will proceed. (b) the tablet willbecome softer. There have been occasional reports of solidstate interactions in the pharmaceutical literature. in. that thecontainer is not hermetic in this aspect.” but in largescale production. the lefthand side should be written in ionic form as well.
2(COOH)CO~~a HZ0
+
+ CO2
(14.25. The surface area is proportional to the twothirds power of the volume by the isometry factor r. 1985. The following nomenclature is used: there are M g of unreacted sodium bicarbonate at time t. (14.14.74)
Usui checked the weight loss of heated samples in hermetic containers?utili~ing different ratios of acid and base and established that the stoichiometry is that of Eq.76)
(1
It follows that (14. that is.)
. from Usui and Garstensen. It is noted that the decomposition rates are a function of applied pressure. that is.74). corresponding to that of the sodium bicarbonate itself. volume v and density p. and M0 initially. lost weight at a low rate. In other words in an open container. the moletomole interaction of tartaric acid and sodium bicarbonate. there was no interaction. There are N particles each of area a. e next studied the effect of compression on the decomposition of sodium bicarbonate. Characteristic curves are showninFig.77)
15 k
kP
20
40
60
80
100
Time (Hours)
Effect oftableting pressure on sodium bicarbonate decomposition at 70°C. e next studied the weight loss in open containers and demonstrated that the acid did not lose weight.and that the sodium bicarbonate and the mixture of sodium bicarbonate and tartaric acid. simply decarboxylation of the bicarbonate itself.
a = rv2I3 = ~p 213 m2J3
A = Nrp2~3m2~3
= I
N1~3rrp2~3M2~3
4. In the following it is assumed that the particles are isometric and that the reaction rate is proportional to the surface area of unreacted sodium bicarbonate.
and 15 kP.80)
This maybe integrated.earrangement of Eq.83) The data should. it is assumed that there is a very slight amount ( z mol) of water present in the tablet initially and that the reaction starts in a dissolved stage.)
. shows this to be so. The nomenclature used is: A' is the number of moles ofsodium bicarbonate left at time t . indeed. In a closed system there is a rapid interaction between the sodium bicarbonate and tartaric acid in compressed tablets.00). 0 0 3 ~ R = 1. and when initial conditions are imposed the following espression results:
( ~ / ~ ~ ) = (13 1 /
. That . therefore. (14. and M6 is the initial number of moles.8
where x is mole fraction decomposed. 14.27. and where
Eliminating N by inserting Eq. (14.X ) l l 3 = 1 . Even though the system is supposedly dry.
0
100
200
300
Hours at 55°C
Cuberoot plot of sodium bicarbonate decomposition at 55°C: leastsquares fit equations: 0 kP. and are in line with data reported by the Schefter et al.0. then. S is in I the solubility of the tartaric acid in water. y = 1 . The rate constants according to Eq. (1974).83) should be proportional to the specific surface area at time zero ( ~ ~ / M ~ )this is true is shown in Fig. 1985. y == 1 .77) into Eq. S is its sol~bility water and C is the concentration in the water present at time t. (14. The rate constants follow an Arrhenius plot. If this is true.82) gives (14. as water is produced in the reaction. 14. there will be an acceleration.0015~ R = loo).0 .26. plot by a cuberoot equation and Fig.kt
1)(14. (14. The data can be modeled in 4 the fashion shown in the following. (Data ( ( from Usui and Carstensen.78) gives d ~ / = k't ~ ~ ~ 3 (14.
2
0.6
S p ~ c ~ fSurface Area (sq.25
1. this figure is multiplied by volume of water present [i. 70°C. (14.447 ( R = 0.ln{X} = 1. 1985..84)J: dM'/dt where
==:
k
* ( ~ hM' + Z )
(14.M' + z)O.4
0. the expresthe sion in Eq.85)
where k2 is the secondorder rate constant.75
1' 0 0
1. y = 0. 1985.50
0
20
40
Hours
Decomposition of tartaric acid plus sodium bicarbonate tablets at 55°C (5 kP force): leastsquares fit.)
+
.25 versus specific surface areas: leastsquares fits.86)
0. To express this as number of moles decomposed.0291 * t (R= 0.e. and 55"C.788 3.98).3225 0.O18
The isa appearance rate of sodium bicarbonate in solution is given by dC/dt = k2SjS
(14.50
0.534 + 19. (Data from Usui and Carstensen. (Data (R from Usui and Carstensen.99). 188~ = 100).)
+
According to the reaction scheme the number of moles of water present at time t then is
' + z)mol = (Mh . 14.10
CY CY
rc
5
0
0. mlg) ic
7 Cuberoot constants from Fig. y = 1.
14. 1980).1
+ z = k*t + h[z] )
or. by Again it is noted that the reaction must be associated with some dissolution step in small amounts of water.88)
quation (14..89) then sim~lifies to ln[x] = k*t
+ ln(z/kfL]
(1 4. In the absence of (or at low levels of) moisture the reaction may not proceed. The produced tocopherol is much less stable. shows that the theories suggested by Wright (1983) are correct. in essence.For it to have meaning. The valueof z maybe estimated from the intercept and comes to about 0. the hydrolysis is accelerated hydroxyl ions. the mole fraction decomposed is (14.87) (14. The former is sensitive to high former to low pH. but both tochopheryl acet calcium pantothenate are cases in point. ~ Z (14.
In the strictest sense.28. n the case oftocopheryl acetate. and act as moisture scavengers.74.86) can now be recast in the following form:
” In(MA . (An exception to this is when a reaction is purposely carried out during a granulation.1 mgltablet. These are dodecahydrates. One way of retarding the reaction rate i s to preheat the bicarbonate to 95°C for a certain length of time CJVhite.In this manner an alkaline microenvironment is created. and makes it easier to compress.
. They hence stabilize the acid/base mixture in the solid state (if reasonable moisture barriers are provided): any ~ ~ amountZ of moisture created by a beginning reaction of the type of Eqs. Mohrle.k*
O.9
The water formed granulates the mixture. the hydrolysis and the presence ofwater are contraindicated. which is a reasonable figure. 1963.89) ecalling that z is a small number. and Eq. pH is not a term that is defined ina solid system. This is a particular instance where the useof alkaline excipients(e. for instance).g. there must be some water mediation. which ascertains the stability of the vitamin. hence. (14.73) or (14. the term z / M o is small. importantly the sodium carbonate formed can form double salts with the bicarbonate.OI8k2S1S
(14. This. employing x. This will react by the scheme
+ Na2C03+ H 2 0
1)
(14. It is also characteristic that often. It is seenthat the linearity is quite ood.will react with a mixture of the carbonate and bicarbonate to form a double salt hydrate. and yet maintain the reactivity in the solid state.90)
ata areplotted in this fashion in Fig. higher temperatures are not indicative of what will happen at room temperature. It is obviously of pharmaceutical importance in most situations to slow down the reaction in the solid state. hydroxyapatite) can be deleterious at higher temperatures. Calcium pantothenate is frequently admixed with magnesium oxide and granulated separately from the remaining i~gredients.
1997). Gu et al. They are. ice.ilit
If it is desired to control the of the microenvironment then citric. then stabili~ation resulted.92) so that different results may be obtained in DSC experiments depending on whether a crimped or open pan is used.7. ray et al.g. This essentially means that. (1990) report on drug excipient incompatibility studies of moexipril hydr loride. did not sufficiently stabilize the ut when the mixture was wet granulated. however.
Dehydration. and their and fumaric acids are the acids of pharmaceutical andl ling is far from ideal.5a][l.6. are not the a1 substances to handle in a tablet or capsule. and when s t o i e h i o ~ e ~ r i c a ~ o uf n t s o al~aZi were used. in several steps in dehydration of a hydrate. With an alkali. but they are abrasive. It might be argued that in such salt as a situation the sodium salt should be manufactured and used as such. The situation is referred to in the Federal ~ e ~ i s tasra ~ e r i v ~ t i v e e drug. (1999)have shown such a diagram for ~2(~)~toluenesulfonyl amino~3[[[5. and theymaybe su~~marized the denotes solid.e.. a lower hydrate or a crystalline anhydrate). may result in amorphous anhydrates. sodium carbonate. potassium clavulanate). hence. These are. too..g. in the solid state. sodium bicarbonate. (14. as mentioned before. the s o d i u ~ is sta~Ze opposed to the acid. §uryanarayanan. p s e ~ d o p o l ~ ~ o rt~ h incs f o ~ ~ a t i oThere are ~a ns. and mag~esiumand calcium oxides are common. D denotes drug. tartaric. properly speaking. and are not as corrosive as the acids mentioned. the acid addition salt).. (1997) employed have show that dehydration of theophylline mono~ydrate a twostep process. or in amphoteric substances. (enthalpy of dehydration = A
~ ~ )
(enthalpy of vaporization = AHv) *x that is. V denotes vapor. adding small amounts of water to a mixture of the drug with sodium bicarb e or sodium carbonate). but often the salt is very soluble and hygroscopic (e. and demonstrate that (even “wet”) adjustment of the microenvironmental (i. and I ..4~diazepin2yl]carbonyl]amino]propio Suihko et al. and they. all corrosive. r certain compounds it is necessary to control the “micr~environm~nt~~ in even more drastic fashion. is (enthalpy of transition
= AHT)
. It might be argued that it should be claimed as the active ingredient (equivalent to a certain amount of free acid.8tetrahydro4oxo5(pyrazolo~l. they are difficult to produce. but may also result in another crystalline phase (e.
(19’?4). It is possible. Carstensen (1995). so that if one broke a “discolored. and the model chosen is the one that fits the data “the best” (Sharp et ai. a solid tablet will decompose by photolytic decomposition only in the surface area. owever. be pressure induced. data are fitted to a series of equations. expressing the numbers as percent. Vitamin A beadlets equilibrate at about 75% of the original vitamin A potency. Lachman et al. There may not be a total loss.2 0
10
20 Time (sec)
30
40
Activity loss of catalase in the solid state inducedbypressure. (Carstensen. even at high pressure.)
(Data from
. and the same data treated by zeroorder kinetics give fairly much the same goodness of fit and activation energy.30. 14. valuable information may be gleanedin this fashion. 1977) Arrhenius plotting of a firstorder reaction. and tocopherol acetate. and this is shown in Fig. Wurster and Ternic. or may not. and by iteration it maybe found that.30).
Not much systematic work has been reported on photolysis of solids. most often. Kaminski et al. likewise. Often. (1997) have reported on the photoreactivity of mefloquine hydrochloride in the solid state.
Carstensen (1980) noted that topical reaction profiles were literally indistinguishable from firstorder decomposition profiles. These equilibria may.and at times. (1961) pointed out that. 1995.29 and 14. can achieve an equilibrium state in soliddosage forms. even at high pressures.” exposed tablet.There are times in which equilibrium sets up in the solid state. (1979) reported on a case where a combination of moisture and light caused an interaction between a dye and a drug (ethinyl estradiol) that permeated the entire tablet. More convincingly. Tonnesen et al. then the color would be unaffected in the interior. gives the best bias fit. Sharp et ala
0. 1996). 67% of activity left.3
9
0.4
d)
0
e 2
9
0. because the figures seem to taper off with increasing pressure. Wurster and Ternik (1995) havereported data that imply a pressureinduced activity loss in solidstate catalase (Figs. at times. 14. to invoke Arrhenius fitting to distinguish between reaction mechanisms as pointed out by Nelson et al..
” The original suggestions by Nelson et al. This can be thought of in many ways. There has to was be some moisture present to permit the hydrolysis of aspirin. sufficient to rule out the model as representing the decomposition mechanism.8 $ 2. Figure is not part of the reference publication. Ledwige and Corrigan (19969.29 treated by subtracting 67% from the percentage of zero pressure content. who fitted dehydration data to a series of oftused kinetic equation and applied the rate constants to the Arrhenius equation.
. (1974) and Carstensen (1980) were investigated by Taylor and York (1998).6 treated by the ProutTompkins equation.4139e3x
RA2 = 0. and Taylor and York (1998) have cautioned against that “lack of discrimination of the different best fitting models.
At times an active ingredient or a decomposition product in a solid dosage form is a liquid. It is seen that there is d~finite curvature in the plot.
Data from Table
14.4 treated by routTompkins kinetics. 14. and this may interact with other ingredients in the dosage form. They showed that the decomposition of ~henylephrine an acetylation. A typical example is the work by Troup and Mitchner (1964) dealing with aspirin and phenyhe authors showed that the decomposition of phenylephrine was linearly related to the formation of salicylic acid. 14.2 3.0 c 2. models can be ruZe~ The data in Fig. as did Carstensen (1980). If the salicylic acid is
0
620 400
80
Time (minut~s)
100 120
ig. 14. They. (Data from Wurster and Ternic.958
3.6 b 3. At times.31 is the data in Table out.l
100 300 200 Pressure (MPa)
Data from Fig.)
(1996).y = 3.4 $ 3.5279
n
.6 3 0
rr(
I. found that fits and activation energies from the different models remained fairly invariant.4. 1995. and plotting loglinearly against applied pressure.
it is assumed that there are two drugs.05 g/mL. so that the moisture does not play a part. (14. The concentration of D is S . but it is consumed .98). The reactions then are: (rate constant k ) (14. The rate at which I> disappears is the question to be solved. which is not probable.95)
+ 1/3R(OH). so that (in terms of moles) dC/dt = kA .g. whether the acetylation is achieved by acetic acid in vapor phase or in the liquid state or (more unli~ely) the whether it is a direct solidtosolid interaction. thatis
(14. If it i s an interaction in the liquid state. in which acetylation of codei~e aspirin1 in codeine combinations was demonstrated). then the acetic acid formed may react with the phenylephrine [R(OH)3]. 1966. then Jander kinetics should actually apply. But it is difficult to distinguish this and pseudofirstorder reactions. using and denoting
~'SMa =
(14.98) A = A. quantitatively in the overall reaction. for acetic acidat 25OC the density is1..100)
. so the equation for st first beestablished and solved.It is assumedthat the disappearance rate of A is pseudo~rstorder. by hydrolysis) to form a liquid decomposition product C . is not yet resolved.97)
+C
"+
(decomposition)
(rate constant k')
C is the species that is liquid. The rate of the latter step isgiven by a secondorder reaction term. the work by Jacobs et al.g.. There are other examples of the interaction of acetic acid with active ingredients (e.94) (14.96) (14. and let M denote the molarity of the liquid decomposition product (e. and the molecular weight of C is M . in other words 3) 3COOH
+ HZ0
"+
C6H4(0 1/3R(OC 1/3
(14. because its molecular weight is 60. A would be 1005160 == 4 16.k'SCA4 (14. C is created at a rate of kA.75). Let A be the number of moles of drug present at time t.93) (14.99) nserting Eq. In this case a saturated solution (S mol/mol) of is formed. Again. In more general terms.g. Ifit were the latter. C the number of moles of acetic acid.
3)
" +
+ 1/3R(OH)3 +
An alternate explanation would be that phenylephrine interacted directly with aspirin in an anhydrous solid state to transacetylate. and it is assumed that dissolution is fast.. exp(kt) The disappearance rate of D depends on how much C is present.formed by interaction of aspirin with traces of water.. A and decomposes (e. so that. The question is whether the acetic acid (which a sizable vapor pressure) interacts with has the phenylephrine as a gas with a solid reaction (to be covered shortly) or as a liquid with a solid reaction. then it probably occurs by the ~henylephrine dissolving in the acetic acid formed. again liberating water. The amount of C at time t is C.
the more liquid there will be to dissolve the furoic acid. An example of this is the intera~tion be OHR”) and substituted furoic acids (RCOO furoic acids decompose when heated by the decomposition product and carbon dioxide. At a
0
10
20
Time
Stability profile using A = 50. however. k = 0.0 = [kA. the more contact points.exp(k’SMt))
(14.aC Laplace transformation. The reaction accelerates because the larger the extent it has reacted.105)
) It follows that the decomposition rate of I is given by
by integrating Eq.k ) ~ [ { l / ( ~ . which will dissolve furoic acid to the extent of its solubility. A ~ different situation ariseswhen an insoluble component interacts with a drug in sa). There will be a reaction probability a. using Lnotation. and will spread over the microcrystalline cellulose. we arrive at the following equation: dC/dt = kAo exp(kt) .107)
Q is a liquid. In the presence of microcrystalline cellulose.k t ) } / ~ ) (1~. hence.where a is constant. There will a number of contact points N .2.
. the mixture f o m s carbon monoxide:
pi”
+ Q + Q’ + CO
(14. and k r S = 0. asso point. k = 0. (14. 14. at which interaction can take place (essentially the “wetted” f the microcrystalwitheach contact linecellulose).U/(S k))
(14. and a = 1.102)
+ k)]a
+ + 411
so by taking anti
C = [kAo/(krSM.k)]{exp(kt)
/dt = krSCM = aC
.32 usingA = 50. gives: SL.) (14.101) (14.104. we obtain
= [ k u A o / ( k r S ~k)][{[e~p(k’SM211/krSM) ( ~ e ~ ~ ( .1.2./(s or
= EkAo/(a .105) and multiplying by a.106) ]
An example of this is shown in Fig.
1 where g is a constant. Hence. that (14. It might be argued that the external surface of the microcrystalline cellulose would be insufficient to account for the total decomposition. creating one liquid decomposition molecule. 1972. since a and b are assumed constant).108)
where q = 6 a(S . y the decompositionat a contact point.L)/Lo
(14. If the initial number of contact points is No. whereas.r l
given time point there will be overcrowding. however.109)
+
N = No exp(qt)
Since.108) (whichcan be done. for instance. there is also a t e ~ i n a t i o n probability 6. will dislodge (dissolve) S molecules of furoic acid at the contact point. this is finite at time zero. water isotherms give surface areas 100 times as large (Hollenbeck.1 10) inserted inthis gives dx/dt = (1/Lo)giV Su~stituting (14. L being the number of intact alkoxyfuroic acid molecules dL/dt = gN 10) (14.1 11)
or dx/dt
I =
( l/Lo)dL/dt
(14. because dissolved molecules be next will to contact points that have already reacted. Zografiand Kontny. two types of sudace present in microcrystalline cellulose: nitrogen adsorption gives low surface areas (the external area).116)
Figure 14.. But unlike the ProutTompkins model. From the definition of L it follows that the mole fraction x ~ecomposed given by is
x = (Lo .1)]N = qiV
(14. 1986).1
This integrates to
where the term A = (Loq/gNo)has been introduced for convenience. at a given time t. it is assumed that the decomposition.1 15) isequivalent to Zn[l
+ Ax] = qt
(14. the rate of decomposition is proportional to the number of contact points. The factor arises from the fact that when a molecules react. then. then OS new contact points are created and one(the one at which the reaction took place) is lost. Equation (14.Marshall et al.
.109) into this gives Eq.112)
E~uation (14. then dN/dt = 16
+ a(S .1).33 shows data treated in this fashion. 1978. 13) (14. There are. It follows then from integrating Eq. (14.
and corne in intimate contact with other solids. If interaction as triplelayer tablets (or comprespotentials exist.)
There are cases for which there are liquids in a solid dosage form. An example is panthenol in a multivitamin tablet. (Data from Carstensen and Kothari.117) where Cf is the on cent ration at infinite time. This is a Lewis acid. The tern on the righthand side is actually the leading term of an infinite series. the liquid will stillooze into the layer containing its interactant.1 16). then separation tech ues.0
10
20
30
40
50
Hours
Furoic acid data treated according to Eq. is given by: (14.1 18)
(14. If the reaction with another drug ( ) is I)
D + B "+ deco~position
then the initial reaction rate is given by d{D}/dt kPB[D]A
(14. It can be shown (Jost. 1983. An example is ibuprofen (B).
Sometimes the vapor pressure of a drug is sufficiently highthat it may interact with other substances via the vapor phase. and may interact with Lewis bases. 1962) that the average concentration C of the liquid in the neighboring layer with which it is in contact. such sioncoated tablets) are resorted to re. such as triplelayer tablets.Usual measures. do not work in this case. for the interactant will be present in the gas phase. magnesium trisilicate is used. (14.119)
. Here it is customary to adsorb the liquid onto a solid carrier and for panthenol. At elevated ternperatures (and at room temperature under compression as well) the panthenol will ooze out of the carrier. but the process will be diffusion controlled.
initially ic & = mass of intact sample at time t ! M0 = initial mass of intact sample M‘ = number of moles of sodium bicarbonate left at time t = initial number of moles of sodium bicarbonate w1 = mass not reacted
ik?.) . A similar situation may be at work inthe aspirin incom~atibilities mentioned earlier. If. because (unlessthe drug is extremely finely subdivided) only a small fraction of the molecules are on the surface. since Eq. be a firstorder reaction.where {Dlis the surface density of Dmolecules (number of molecules/cm2)at time t and A is the surface area. = energy levels above Ei F = (a) a constant. the ibuprofen penetrates the crystal.t (14. the reaction will.120)
+ ln[Do]
where Do is the initial concentration.
. This will be true if only the surface of the solid interactant is affected. then Jander kinetics should prevail.S = iterant in the C = (a) generalsymbol for concentration..
A = number of moles of unreacted solid at time t Ao = original number of moles of unreacted solid a = side of a cube at a time t after reaction has started a = side of a cube before decomposition 0 13 = (kl/k. however. (14. (b) number of intact alkoxyfuroic acid molecules at time t Lo = number of intact a l ~ o x y f ~ r oacid molecules. therefore.1 .119) integrates to ln[D] = kAP. = (a) symbol for Laplace operator. (b) concentration in the water present in an effervescent tablet at time t CI = constant in the WLF equation Cz = constant in the WLF equation Cf = the concentration at infinite time D = diffusion coefficient DSC = differential scanning calorimetry E = energy Ea = activation energy E.”= enthalpy of vaporization h = thickness of a reacted layer k = general term for rate constant k l = firstorder rate constant k2 = rate constant for twodimensional diffusional decomposition k* = rate constant in effervescent interaction I. The extent of decomposition will be slight. As long as there is no penetration into the crystals. (b) pree~ponential constant in firstorder decay H = height of a cylinder A H = heat of fusion AHd = enthalpy of dehydration AHf = enthalpy of transition AH.
Lee G (1999). heat capacity or rate constant)of an amorphate at a temperature below or above its glass transition temperature. a constant in the Avrami treatment q = constant in the (a) Avrami treatment. two or three dimensions inapproximate Avrami model N. (b) mole fraction. very small amount of water present in an effervescent tablet a = propagation probability or rate / = termination probability or rate I I" = shape factor Cp = the rate of heating p = density of a solid
+
. (b) the Arrhenius equation. J Chern Phys 7:1103. Agbada CO. Banker GS..g. (b) property (e. = the property R of an amorphate at the glass transition temperature S = (a) solubility in water (for components of effervescent tablet). Int J Pharrn 106:33. I " ( if* = eutectic temperature Tf = melting point T. Peck GE (1982).fastreaction model. = glass transition temperature Tmax temperature at the peak maximum in a DSC = t = time t. = number of molecules with energy levels above El rz = (a) exponent in the Ng equation.N 2 . Avrami M (1939).
. (b) [qkn/(n l)].N = (a) number of nuclei. J Pharrn Sci 71:7.. Anderson NR. = lag time u = particle volume illiams. (b) number of particles in a sample No = initial number of intact molecules N1. (c) fraction decomposed. (c) radius of a cylinder R. (b) S = solubility (mol/mol)of a solid compound in its liquiddecomposition product SI = solubility of tartaric acid in water 1 = absolute temperature.ande elFerry x = (a) fraction. (c) Jander equation R = (a) ideal gas constant. (b) an integer between 1 and 4 (AvramiErofeyef equation) p = exponent in the Ng equation Q = constant in the expanded Avrami model Ql = (a) constant in the slownucleation. N3 = number of nuclei in one. York P (1994). (d) number of moles of water in a hydrate x* = the mole fraction inBawn kinetics where just enough material has decomposed to just dissolve the remainder of the parent compound Z = normal standard deviate z = original. J Pham Sci 88:199.
Z Anorg Chem 163: 1. J Pharm Sci 87509. Zografi G (1993). ~arstensen Pothisiri P (1975). Imaizini H. p 24. Chi LHH. Weinstein S. University of New York at Buffalo. 7:379. J Pharm Sci 64:37. 2ndreviseded. J Pharm Sci 55561. Olsen BA. Gubskaya AV. Solid State Chemistry of Drugs. Holgado MA (1999). Ofusa T. JMS Pure Appl Chem A29:l10. J Pharm Sci 82:657. Carstensen JT (1979). Academic Press. Fini A. p 45. Urbanyi T. Mehl R F (1939). Munson EJ. Lin CT. Musa MN (1972). Suryanarayanan R (1997). Process Biochem 24:38. 1 Dali MV (1995). Introduction to Solid State Physics. StanleyWood NG (1972). New York. Kothari R (1983). Oksanen CA. Erofeyev CR (1946). Carstensen JT. Franks F. FernandezHervas. Mol Cryst Liq Cryst 50:99. J Pharm Sci 50:141. MohrleR (1980). Drug Dev Ind Pharm 19: 18 1. McGuire JL. CR Acad Sci URSS 5251 1. New York. Padden BE. JT. 2nd ed. John Wiley & Sons. Snorek SV. Morris T 91993). Van Scoik K (1990). Windheuser J (1966). Fazio G. Sidzhakova D. eds. J Pharm Sci 63:755. PudipeddiM. Int J Pharm 114:247. J Chem Phys 9:177. p 31. J Pharm Sci 532393. O’Donnel JH. Padden BE. Strickley RG. PhD dissertation. J Res Nat Bur Stand 57:217. Han J. New York. Academic Press. Leung SS. Morris T (1993). Jahansouz H. Das NG. Int J P h a m 181:ll. Nelson E. Grant DJW (1998b). MoelwynHughes EA (1961). J Pharm Sci 68:368. Sixsmith D. Spera D. J Pharm Sci 72:1149. Academic Press. Carstensen JT. Franchini M. Grant DJW (1998a). Hollenbeck RG. Yoshihisa M (1999). 1112.rl
Avrami M (1940). Carstensen JT. Nagai T (1980). Lemmon RM. J Pharm Sci 67: 1599. Int J Pharm 163: 1. Carstensen JT. Kaminski EE. Swartz C. Proc Phys SOC(Lond) 43:461. Pharm Dev Techno1 2:303. Kittel C (1956). Ng WL (1975). Sokoloski TD (1995). Cooper J (1961). J Pharm Sci 68:863. Lisnyak W. Gu L.Physical Chemistry. Avrami M (1941). Kaufman MJ (1999). Lachman L. Chowhan ZT. Cohn RM. Jacobs A. Carstensen JT. AlvarezFuentes J. Pergamon Press. Drug Dev Ind Pharm 21:1953. Carstensen JT (1974). vol 1. Trans Am Inst Min Eng 132416. Parsons MA. Kelly TP. Birn SR (1979). A study of the solid state stability of ampicillin. Gordon PK. J Pharm Pharmacol 24:138.
. Aron E. Mazetti E. J Am Chem SOC 802730. Pharm Dev Techno1 4531. Kissinger HE (1956). Pharm Res. pp 5970. Lewellen PL (1997). Vance JJ (1966). New York. Oberholtzer ER. New York. (1958). Johnson WA. In: Lieberman HA. Leung SS. Diffusion. J Pharm Sci 61:273. Jost H (1962). Otsuka M. J Chem Phys 8:212. Pharm Res 7:1278. Eppich D. Peck GE. Karaivanova V. Jander W (1927). Whittaker AK (1992). Bray ML. Georgieva S (1998). (1989). Brenner GS (1979).Lachman L. Carstensen JT (1980)Solid Pharmaceutics: Mechanical PropertiesandRatePhenomena. Bogdanova S. Aust J Chem 28:1169. Weinstein S. Marcel Dekker. J Pharm Sci 87:501. Dilatusch A. Pharm Res 10:791. (1990). Chem Pharm Bull 28:2565. New York. Munson EJ. Xnt J Pharm 157:209. Blagoy YP (1995). Carstensen JT. Byrn SR (1982). Drug Dev Ind Pharm 25:197. Nambu N. Lo PKA (197’7). PharmaceuticalDosageForms: Tablets. Kildsig DO (1978). LennardJones JE (1931). Perry FM. Personal Communication Dudu SP. Marshall K.
J Am Chem SOC 97:5290. Int J Pharm 158:47. Bergman RC. 58:185. Suihko E.Otsuka M. WE. J Pharm Sci 74:1293. Pfeiffer RR. Pikal MJ. Personal communication. Ramalingam S (1974). Carstensen JT (1985). Roy ML. Van Scoik K. Engel GL. Madison. J Pharm Sci 84:190193. Kmack G (1967). Skrede G. Kontny M (1986). Lang JE. Stacey FW. Prestrelski SJ (1999). University of Wisconsin. Sukenik CN. Shimomura K. Pikal MJ. J Pharm Sci 88:351. Otsuka M. Lukes AL. Paronen P (1997). Teraoka R Matsuda Y (1991). J Am Chem Soc 77:3701. J Pharm Sci 53:375. Carstensen JT (1976). J Pharm Sci 88:291. Pudipeddi M (1 995). Landel RF. Bethesda. Usui F.105. MD. Onoe M. Williams ML. Ahlgren M. Master's dissertation. Antimicrob Agents Chemother 9:848. TQnnesen HH. Pothisiri P. Wurster DE. Pharm Res 3:187. Drug Dev Commun 1(1):29. Rickard EC. Pikal MJ. Drug Stability 1:249. Wright JL. Carstensen JT (1990). Troup A. Carstensen JT (1986). McKusick BC (1959). Gynther J. International Symposium on Product Biological FreezeDrying and Formulation. J Am Chem SOC81:987. Usui F (1984). Lukes AL. Schmitt EA. Dev Biol Stand 74:323340. Int J Pharm. Maloney AM (1990). Shalaev EY. Shefter E. Ternik RL (1995). LauPY. Zografi G. Zhang GGZ (1991). 75546. Coleman D (1976). Burn SR. (Karger. Lo A. Martinsen BK (1997). Zografi G (1997). Wood G (1975). Pharm Res 8:1066. Poso A. Law D. J Pharm Sci 66: 1312. Chem Pharm Bull 37:493. Matsuda Y (1993). J Pharm Sci 67:767. Shalaeva M. Gaines IS (1976). Shlyankevich A (1995). J Pharm Sci 56:1028. School of Pharmacy. Tzannis ST. Mitchner H (1964). Mandel NS.1792. Shefter E. Pharm Res 10:577. Sekiguchiki I (1987). Jang JE (1977). T White B (1963). Int J Pharm 152:75. Personal communication. Bonopace JA. Ferry JD (1955). J Pharm Sci 64: 1931. Suzuki E. US patent 3. Saucer JC.
. Ketolainen J. Basel. 199 1). J Pharm Sci.
This Page Intentionally Left Blank
.
10.6.3.
. 15. tablet and capsule formulations. in general.8.. Excess Water
268 268 268 269 27 1 273 274 275 276 277 278 278
Stability of drug substances in dosage forms is affected not only by their chemistry.1. Amorphates 15. Moisture Amounts at the Critical Moisture Content 15. 15. is a l he chapter to follow will deal with the nature of the interaction between water or water vapor with drug substances. 15.7.5. it is conventional ( ~ a r ~ t e net e ~ 1964) to study combinations both in the absence cirict tlw s al. presence o ~ ~ a t eThis is because. NonhydrateForming Drug Substances 15. 15.15.2.9. Compatibility studies are generally carried out with new drug substances in combination with common tablet or capsule i n ~ r e ~ i e nto ts ascertain that the excipients chosen are not detri~ental the integrity of the drug to (or of as little damage as possible). 15. When such programs are carried out.4. of a11 the types of substances one encounters in r. but also by their environ~ent. the most ~ e t r i ~ e ~ t water. Symbols References ound Water roenvironmental pH drateForming Drugs 15.
then moisture present on or in it will be of the types shown in Fig. If the vapor pressure is larger than the critical vapor pressure. 1990). such systems may be considered solutions and. What would be expected here is is that the surface moisture would interact withactivesites. 15. 15. kinetically. Morris.1. 15. It can be moisture that is adsorbed in an amount less than thatcorresponding to a monolayer (see Fig.
*
This ~ituation the one referred to in Fig. as such.the decomposition isnonhydrolytic.IC will be discussed.In this event (a) either the moisture acts solely in a catalytic sense (i.
If a substance does not form a hydrate. Oncethe critical vapor pressure for the compound (the water vapor pressure over a saturated solution) is reached (see Fig. and this bulk moisture layer will dissolve drug substance to the extent that a saturated solution is formed.la). as solution systems. 15. 15.ld). 15. orstarting to forma bilayer (see Fig. corresponding in concentration to the vapor pressure in question.la. and that the reaction would proceed from these active sites. This is not surprising in light of the previous findings thatamorphates are somewhatlikeliquids. should behave. 15. Carstensen and VanScoik (1990) showed that water vapor pressure over amorphous sucrose that contained water corresponded to a value that could be extrapolated from the vapor pressure curve of unsaturated solutions ofsucrose at the other endof the concentration ence. is formed (see Fig..
. moisture will condense on the solid in form of a bulk.lc). the situation leading to Fig. ora multilayer (not shown). lb).ter 1
s
Amorphous substances in the presence of water degrade according to firstorder kinetics (Pikal.e. Morewill be said about the situations in the following and. 1977. then water will adsorb until all the solid is dissolved and an unsaturated solution. or (b) the disappearance
<c*)
es of moisture adsorption on a crystalline solid. in particular.
Eleucovorin. ProutTompkins kinetics apply. circles.x)] = k(t .4). The hot spot theory is not new. as shall be seen shortly.1956.. 5% water with a pH 2. and usually referred to as an amorphate. and the rate constants follow an Arrhenius equation (Fig. In either event. and this is akin to the ProutTompkins model described in Chap.
This is the part of Fig. In fact Gluzman (1954. (The buffer forms hydrates. 15. Data plotted in this fashion is shown in Fig. give rise to firstorder kinetics. Intermediate moisture levels. This is simply an.2 represented by triangles. 15. low moisture content. 15.of moisture is replenished (however small coverage) in an amount determined by the the isotherm. the decomposition will proceed along strings or planes.3. in appearance being a solid. and the water contents and percentages added are not necessarilyavailable moisture).
0.1)
where x is fraction of drug activity retained. Literature data are most often insuf~cient todetermine if accounts for the profile.2 buffer in the solid state. t is time and t1/2is halflife. at very high moisture contents the situation in Fig.Leeson and Mattocks (1958). The sigmoid profiles shown in Fig. and the decomposition is simply solution kinetics(Le. intermediate moisture content. ICalcium leucovorin withmoisture andbuffers added. extension of the concept of active sites when the subject is adsorption.however. 14. This (as well as kinetics at very low moisture contents) have been explained in recent years by socalled hypotheses proposing a~orphouslike “hot spots” on the surface of a solid. 15. 15.. but with random molecular a~rangement.1958) and Gluzman and Arlozorov (1957) postulated that “part of a surface of a solid was actually in a liquid like state”in other words.2 shows the situations described in Fig.tiedin aspirin decomposition in this region to a Freundlichtype isotherm. At low water vapor pressure (square symbols).t I l 2 ) (15.
. Figure 15.1 as it applies to d . k is a rate constant (propagation constant). The development of such a model has been published by Attarchi (1984) and Carstensen and Attarchi (1988).is firstorder). 15.8
r
PseudoFirst Order
Exposure Time (hr)
d . triangles. squares.Id applies.2 adhere well to the ProutTompkins equation ln[x/(l .
paminosalicylicacids). 1975).5 shows the situation during which a water ~ o l e c u l e reacts with an activated drug molecule A*. and an e ~ ~ l a n a t i o n this is forwarded in the following.2
Arrhenlus plot of aspirin decomposition data in the presence o f limited amounts of moisture. (Data from Garstensen and Attarchi.2) is often first order. 1962). for Figure 15. The intermediate pattern (see triangles in Fig. 1988. the surface is lacking in onewater molecule.)
.0
400 200 Hours
8 06 0 0 0
Plot of aspirin decomposition data in the presence of low amounts of moisture.1
3.
0
c
3 .O
1OOO/T
4
2. then the logarithm of the rate constant at a given temperature Td. (Data from Carstensen and Attarchi. 1988. 15. for vitamin A esters at 55OC) (Cuillory iguchi.
(15.2)
This has been true in certain cases (e. which is then replenished from the atmosphere (Fig.g.. In this view the surface concentration of water molecules (however small) will remain constant (as long as the vapor phase does not significantly change). 15. of a series of analogous compounds in solid form should be inversely proportional to the inverse of the melting point.5).9
3. that is.. and substituted paminobenzoic acids (Carstensen an but in other cases (e. After reaction.g. it does not hold well ( and Carstensen.)
Cuillory and Higuchi (1962) hypothesized that if such a theory were correct.
many reallife examples give rise exactly this situation. and water condense until the e ~ u i l i b rpressure establishes. and dotted squares adsorbed moisture.6. explaining the i n t e ~ e d i a t e behavior a t i n t e ~ e d i a tmoisture levels. and the system will equilibrate. A*.3) is a pseudo~rstorderequation. then all of the drug may dissolve. If the ratio of headspace to solid is very large. i~~
. moisture will condense onto the
~ehaviorof a compound placed in a bottle with a water vapor pressure higher oisture will condense on the solid. owever. an anhydrous sample of solid may be transferred to a bottle where the atmospheric water vapor pressure is above the critical water vapor pressure. an A* molecule reacts with a water molecule and.k ~ ~ *= } ~ ~ ~ ~ ] = k'(A*}
(15. in line with the isotherm a new water molecule is adsorbed (second inset).? the vapor pressure that equals that of a saturated solution).3)
kinetic Equation (15. in which the vapor pressure over the solid is exactly the critical vapor pressure. Pcrit(Le. on the surface. and the pressure is higher than the critical. and this process will go on until the second situation occurs.Reaction of
Two Molecules
Ad~orption of One ~ o t e c u ~ e
Dotted squares represent activated A molecules. The bottle is capped.
ence the disappearance rate of A* is given by:
3(A*}/3t . seem unlikely that a drug substance in a closed system would experience exactly the critical vapor pressure. e
It might. In the first inset. As shown to 15. some of which will dissolve.
It is assumed that the decomposition is accounted for solely by the amount of material dissolved. C = S . holds this (Gerhardt 1989).5)
owever. and when a sufficient amount of water has condensed and formed saturated solution to lower the pressure above the “solid” to exactly the equilibrium pressure of the saturated solution (i. decomposed VdC/dt = dM/dt = kl VC 5.
.
M = Mo . and the drug a solubility of S . the larger is the volume V of the saturated solution formed from the solid.1(c)is akin to a suspension. and the smaller is the amount of actual solid left undissolved.b).VS)t
(15. and (c) (noting that the degradants are solutes) cause a decrease in the water vapor pressure with which the moisture layer is in contact so that.7 To quantitatively assess the decomposition.. Pothisiri and Carstensen. expressed as amount M . in this manner.7b)
in other words. andis assumedto remain saturated.8)
That this applies is shown in work by orris (1990) and Carstensen and Morris (1990a. Phenobarbital when it decomposes at 80°C in the presence of phos~hate ~uffer 6. Another example isthat reported by Morris (1990) and Morris and Carstensen (1990a.e. ~anahsouz al. 15. dC/dt = klC where C is concentration or. 1974. Gerhardt (1990) and Gerhardt and Carstensen (1989)have de~onstratedthat kinetic salt effects and saltingin of the drug into the moisture layer can explain the decomposition profilesexhibited by phenobarbital when moisture and buffers are present. et In this moisture region the moisture acts as a solution layer. given by
ko = kl VS
(1 5. one must know k l .b). will cause the vapor pressure to drop. but it has also been known to fail ( e g . 1975).7 is an example for which. the vapor pressure relation is not violated. so that dM/dt = kl VS (15. the critical water vapor pressure). sincethe solution is saturated. the decomposition is pseudozeroorder. and degradation compounds (a) increase or decrease the drug solubility. This.solid and dissolve solid. then e~uilibriumhas been reached. and S . 1990). (b) increase or decrease the kinetic parameter values of the drug. that is.( k .6) (1 (15. V . 15.7a)
that is.. with a slope equal to the pseudozeroorder rate constant ko. such as depicted in Fig. in the initial stages of deco~position. shown in Fig. If the saturated solution phase has a volume of V . and this should permit elucidation of the mechanism. in turn. The higher the atmospheric water vapor pressure P is. A situation. This often holds true (Pothisiri. then firstorder conditions prevail in the solution.
1995.9)
w* is often called kinetically u~available moisture or ~0~~~ water.e. The headspaceof the ampuls was knownand water amounts added were such that given water vapor pressures were obtained when the samplewas heated to 115°C.
The concept of bound water is one for which it is assumed that in a solid or a solid dosage form a certain amount of water will not affect the stability. In the previous sections.)
~arstensenand Pothisiri (1975) and right and Carstensen (1986) have done likewise. It is noted that the intercepts are nonzero. The decomposition is zeroorder. akin to a salt pair). to gauge the value of V . for quantitative assessment. Figure 15.b. The rate consta~ts pseudozeroorder and are plotted versus are ~oisture levels. instability as a fu~ction moisture content has been described. 1990. Samples were prepared by adding micro7 in amounts of water to solid indo~ethacln ampuls thatwere then sealed. the water activity saturated solution is of low magnitude).b). For calcium leucovorin (Nikfar et al. The problem here is.g. 1995) the amount of moisture that is ne and the critical relative humidity (CRH) is small (Le. ~arstensen Franchini. the more u~stablethe compo~~nd. Indomethacin decomposition at 11 5°C. [ V . This is true in many solidstate reactions. it is the situation at the critical moisture content that is by far the most severe. Franchiniand Carstensen. as opposed to the require
kt) = klS[V . and the larger the amount of water. 1990a. On the other hand it is high for poorly soluble drugs. 1990a.. is water of crystallization. or more simply a v ~ ~ ~~aob ~ s t u ~ e . at times.w"] is denoted kinetically available. and Morris and Carstensen.0.8 presents data from the work of Gerhardt (1989) and Gerhardt and ~arstense~ (1989). (Data from Morris. The bound moisture. il~
.6 r
/
90%RH 80% RH
70% FW
60% RH
0
Storage Time (Hours)
.. Thisdecomposition follows zeroorder kinetics at the onset. but of of these. Indeed the zeroorder rateconstant increases linearly with amount of water in the system. or very soluble drugs (e.. there are intermittent plateaus that correspond to a constantwater activity (~H/lOO) a series of water contents for (i. ranitidine.w"]
(1 5..
10). Gerhardt and Carstensen. graphed versus added moisture. The amount of time (t') required for the eutectic to form (for the mass to form a homogeneous liquid) is linear in water activity (a = ~ ~ / l O Othat is. 1989. but becomes the situation as theamount of parentdrug xamplesof this are the work by orris (1990) in which the indomethacin/ water system was studied in a closed system at 130°C (Fig. and q' are constants (Fig.q'a
(15.'Oo0
1
0 0
100
200
300
400
pL H20
Rate constants (pseudozeroorder) from plots such as shown in Fig. After a short time period a eutectic consisting of indomethacin. 15. 6
0
3
0
20
40
60
80
100
Storage Time (hrs)
D~composition indomethacin m the presence of moisture at 130°C. wherethe amount of water sufficesto bring all of the rug into solution) is denoted excess water. ).)
he situation in Fig. (Data from Cerhardt.
tl
=p
. and water is formed. deco~position products. This may not be applicable initially in drug dosage forms. 15. and from this point in time the decomposition is firstorder as expected for solution kinetics.ld (Le..7. Data from of
. 15.10)
where .9). 15. 1989.
30
20
10
0
50
60
70
ao
90
100
100 x Water Activity. citric he acid) or.. then the degradation consists of (a) dissolution up to where dissolution is complete. sod carbonates) in an attempt to make an adjustment of “the microenvironmental p In the area shown as Fig. It is seen from ikfar et al..b) that a displacement of 1. (Data from 1990. if it is acidsensitive. 2 ~ d.. (1987).he or she may employ bases (e‘g.9 plotted versus relative humidity. Gerhardt (1989) Gerhardtand Garstensen (1989) have demonstrated the existence of a “solid rofile” that parallels (but is not identical with) the traditional profiles of the drug in solution (Fig. after which (b) moi~tureond dens at ion continue until a concentration of the totally dissolved drug equals that of theofthe atmosphere. conversely.b) have reported similarly in relationship to ~ropanthelinebrooshioka and Uchiy 986a) showed that the mechanism changed at the critical rel~tive ~ ~ ~ ~e. one may. 1 5 . and Yoshioka and Garstensen (l990a. l990a. one assumes that the p value of a saturated buffe g r a ~ h i n ~ data from the moist solid. a. “buffer” a solid and dosage form. piece of evidence sorbed of moisture lay ut how does one definethe microenvironmen fully resolved. Carstensen et al.Nikfar (1990).g.)
Yoshioka and Uchiyama (1986a.b). 15. 15. The isp placed values are symbolize^ by squares in
.1 1). ut the sorbed solution could be of a p of t is displaced from that observed in solution. 1977).. ere is also the possibility of a kinetic salt effect.
If a formulator is aware that a compound is more stable in an acid than in a neutral or basic e~vironment or she may often formulate it with solid acids (e.If the solid is placed at constant relative humidity at values higher than the CRH. The shift in position of the kinetic le in solution from the valuesobtainedfrom lidstate decomposition 130. meaningfully. the t y where the water activity just equals ~ ( i point that of a solution saturated drug) (Carstensen.
(RH)
Lag times from Fig.4 p the rate consta~ts the solid state.
the “horizontal” lines are not really horizontal (e. The squares are points from solidstate decomposition shifted by 1. (Data from Nikfar. Point J is the point that corresponds with Fig. An ~ t y ~ i va~or/moisture c~2 curve for a salt pair is shown in Fig. the adsorption isotherm is what governs the type and amount of adsorption. 15. Thepoint is that.IC. The ordinate is moles of water per mole of solid.1. S
WS
x Moles
per Mole
1
0
1
O/*
F
Pressure/adsorptlon diagram for a salt pair. This corresponds tothe“horizontal” line at low moisture content in
Water
S Moles
per Mole
Moles
J L
aturated olution. 15. In the work published by Cerhardt (1990) it wouldbe necessary to force a 6pHunit shift to obtain coincidence.)
in Fig.
. The same holds true for the line HJ which is usually a horizontal depicting e salt pair. 15. 1992..11 and if such a shift is made.
In considering the situation in Fig.g. in contrast with the general description of salt airs. 15. and checked the stability of it as a function of moisture. one would truly obtain a line that woul slightly. then the data in solution would coincide with those in the solid state.Id. so there are still unexplained factors at work.Zcalcium leucovorin. 15. actually is part of an adsorption isoe m ) .12.4 pH units. and corresponds with Fig. the line range in which the anhydrate is stable.Solid Points Shifted
J
2
3
4
5
6
7
PH
1 pH rate profile of firstorder rate constants extracted from kinetics of decomposition of d. started with a crystalline anhydrate.
eyond this point the salt pair forms. An example of this is amoxicillin which. For amoxicillin the trih drate without adsorbed water is the maximum stability situation. Between H and J.13. After that. the rate constant increases s However. At this point. (This happens to be the maximum stability point. and rate constants that were constant at these various ~ o i s t u r e contents.) There would (might) be a small residual rate constant at 0% of moisture adsorbed. and on completion of this. After this unsaturated solutions form t i o (not the rate constant) increasesbecause the water and the ~ e c o ~ ~ o s irate ~ volume term increases. would have an increasing rate constant for RH values between 0 and about 30% (the e ~ u i l i b r i u ~ vapor pressure for the trihydrate). Nikfar (1990) has shown this in a situation in which calcium leucovorin in the presenceof sodium citrate and citric acid had a given R value for several moisture contents. as the RH is raised. the rate constant would be that for trihydrate without adsorbed rnoisture. and this has another stability profile. in general. and the trihydrate will start forming a saturated solution. increasing the rate constant from J to K. this increase is often not great. there will be given relative humidities for which one salt form reverts to another.
If hydratefor~ing buffers are present. the anhydrate formed by dehydration of the tri.givin to the line FG in 15. water additional water will simply convert the anhydrate to trihydrate. for the water is consumed by the buffer.(or higher) hydrate is amorphous. ET adsorption takes place. more water added to the systemwill not change the rate constant. and in this example the line FG has a steep positive slope because the amorphous anhydrate would be much more unstable than the crystalline form.
. and there . and in these regions. For hydrates. Between G and b form moisture is ~ b ~ o r to e ~ the trihydrate. experimentally.
Exaggerated schematic of rate constants as a fu~ction RH for a compound of forming a trihydrate. the line GH can go either up or down. if the anhydrate were crystalline to start with. Fig. and at point H this conversion is complete.and at J the critical RH is encountered. giving the smallest ko value. In the case of amoxicillinand other Blactarns. More and more bulk saturated solution will form. surface adsorption onthe trihydrate increases the rate constant.
Madison. University of Wisconsin. ed.. awn C (1955). PhD dissertation. ~ecomposition aspirin in the moist solid state. the pentahydrate in the case of streptomycins) is maintained. Cephalosporins are ofte s. however.
A* = concentration of active sites a = water activity (~H/lOO)
EmmettTeller isotherm n of drug in bulk layer water concentration on surface k = rate constant kl = firstorder rate constant
heat of fusion constant (in lag time equation) = relative humidity S = solubility T d = testing t e ~ ~ e r a t u r e . p 254. and at a given te~perature covered elsewhere) becomeszero.New York..g. and if formulated with Trona (a~proximate formula Na2C03. New York. therefore. Carstensen JT (1977). Chemlstry of the Solid State. ~ccelerated tests. Academic Press. must be carried out with control of integrity of the hydrate (e.2Na the antibiotic will remain in its pentahydrate form because the salt pfor e Trona encom asses that of the cephalosporin. WI. IS t = time tl12 = halflife t’ = lag time Y = volume of saturated bulk layer w* = mass or volume of “bound” water x = fraction decomposed B = constant (in lag time equation)
= =
Attarchi F (1984). Pharmaceutics of Solids and Solid Dosage Forms.ften hydrates are added to drug substances that are hydrates and are most stable as hydrate. IS Tm = melting point. John Wiley & Sons. This is the critical temperature for (as the hydrate. be a (unless moisture conditions completely d range of a salt pair. decreases with increasing temperature.
. The value ofko will. p 12. In: W Garner. the salt pair range of the Trona (Le. School of of Pharmacy.
Janahsouz H. Decom~ositlon dlcalcium leucovorin in the solid state. Mooney KG. of School of Pharxnacy. Jang JE (1977). Uch Zap Khar'kov Univ 54. Uchiyarna M (1987). Morris T. Carstensen JT (1975). Yoshioka S. 'University of Wisconsin. J Pharm Sci 7592. Nikfar F (1990). Mroso PV (1984). Yoshioka S. Carstensen JT (1986). Danjo K. Carstensen JT (1990b). Carstensen JT. PhD dissertation. J Pharrn Sci 75:546. Pharrn Res 7:S195. Hollenbeck RG. PhD dissertation. J Pharm Sci 77:318. Pothislri P. p 61. Carstensen JT. University of Wisconsin. University of Wisconsin. Z Fiz Khim 31:657. Madison. J Pharm Sci 71 :1096. WI. Nikfar F. Madison. Pika1 MJ. Kildsig DO (1978). Attarchi F (1988). J Pharm Sci 66:1312. Li Wan Po A (1993). J Pharm Sci 535235. Tornpkins FC (1944). WI. Marshall K. J P h a m Sci 67599. Mooney KG. J Pharm Sci 75:459. Prout EG. Pharm Res 7:S196. Carstensen JT (1990a). Van Scoik K (1990). WI. Gluzrnan M (1954). School of Pharmacy. J Pharm Sci 79:943. Nikfar F. StanleyWood NG (1972). School of of Pharmacy. Li Wan Po A. Carstensen JT (1989). Waugh W. Forbes SJ. Mattocks A (1958). Madison. Carstensen JT (1990a). Pharm Res 11:S238. Pothisiri P (1975). Int J Pharrn 18:287.Carstensen JT. Stella V (1990). Pothisir? P (1975). Uchiyama M (1986b). Irwin WJ (1982). Z Fiz Khim 32388. Peck GE. Mroso PV. J Pharm Sci 51 :100. Pharrn Res 6:S142. Pharm Res 7278. Arlozorov D (1957). Kornblum S. PhD dissertation. Int J Pharxn 83:87. Franchini M (1995). Decomposition of Phenobarbital in the Solid School of Pharmacy. Carstensen JT (1990b). Carstensen JT. Higuchi T (1962). Gerhardt A.
. PbD dissertation. Trans Faraday Sac 40:489. J Pharm Sci 76:548. L1 Wan Po A. J Pharrn Sci 64:1931. Carstensen JT. Guillory K. University of Wisconsin. Pharm Res 7:S127. Gluzman M (1958). orris T (1990). Sixsmith D. Pharm Res 7:S195. Yoshioka S. Carstensen JT. D e ~ o ~ p o s i t i o n indomethacin in the solid state. J Pharrn Pharrnacol 24:138. Carstensen JT (1990a). Carstensen JT (1994). ~ecoxnpositionof ~axninosalicyclicacid in the solid state. Yoshioka S. Gluzman M. Pharm Res 7:S195. Morris T. S. Gerhardt A (1990). Gluzman M (1956). Drug Dev Ind Pharrn 21:523. Franchini M. Tr Khim Fak NauchIssledovatel Inst Khim 14:197. 3 Leeson L. Madison. J Pharm Sci 64:7. Carstensen JT (1990b). Ku S. Lukes AL. J Am Pharxn Assoc Sci Ed 47:329. Tr Khim Fak NauchIssledovatel Inst 12333. WI. Sciarrone I (1964). Yoshio~a Uchiyama M (1986a). J Pharm Sci 79:799. Wright JL.
This Page Intentionally Left Blank
.
3. Stable onf figurations 16. Apparent Densities of Binary 16.16.4. 16.1)
.g.11. however.6.9.
ensity and Porosity Definitions etween Density and Porosity
282 282 283 285 286 288 288 290 29 1 29 1 295 296 297
16.8. ~ p ~ l i c a t i o n Real Powders 16. the volume V (of a drum. for instance) is partly occupied by solid particles. Wall Effect for Spheres to 16. Hence.2.” in that this term implies the weight of the substance per real volume. The Wall Effect 16. Characteristics of Mixtures of Spheres 16.1. then the ratio:
pf
==
M/V
(16. Closest Packing 16.When dealing with particulates. it has the dimension of mass per volume (g/cm3). if the mass M of powder in the drum is determined by weighing.7. and partly by void space. Hence.5. Compaction from ‘Vibration and Tapping Symbols eferences
A compact solid (e..10. Porosities of Mixtures of Several Particle Size Cuts 16. a crystal) has a fairly welldefined “density.
it will reduce in E‘. . This is referred to as the ~ o Z e c ~ l f f r density in the following.and an associated t a ~ ~ e d bed ~ o r o s i t ysiap. l k The importance of apparent density in pharmace~tical operations is obvious. then it will achieve a certain cascffdedupparent density p’.Whereas the molecular density is a given number.. and an associated bed~orosity and if it is then “tapped” in a reproducible fashion..” It is obviously smaller than the “true” density (e. therefore is somewhat smaller and is referred to as the erystulZograp~ic density. as mentioned in the introduction. piap. the number of capsules or tablets or the amount of bulk powder that willfit in a drum are examples for which the apparent density is perfor~ancecontrolling. l~ If a particulate powder is poured into a container. and the volume of the powder is Vl. The fraction of the volume that is occupied by solid is. then:
v. ranules are often produced in pharmaceutics./v
(16. and the two terms are usually almost equal. The amount of powder that will theoretically fit in a capsule or in a tablet die or punch volume is determined by this parameter.
.g. As mentioned in earlier chapters. and p‘ is referred to as appffrent density or ~ ~density. and the density p p is referred to as the p ~ r t i c density.
der is transferred to the cylinder.2)
where v. but these differences will mostly be small. is known as the solids f r ~ c t i o n .is a “density term. E ~ void space. purticZe porosity.
Schematic for definition of fractional volume and appar~nt density. volume and attain a somewhat higher tapped density. This is referred to as the . ofa crystal. real crystals have defects and flaws.
A perfect crystal would have a density determined by the crystallographic system and by the molecular weight of the substance. and these are aggregates of crystals and contain a certain fraction. depending on defect concentration. and the actual density. or the particles from which the powder is made up). =
v. the crystallographic density may vary ever so little.
The apparent density of the coarse powder is denoted p: (g/cm3) and its particle density isdenoted pc. The fraction of fines is
x = m/(p.O cm3 is p.2a). When w1 grams of €3 are added.
x )
Schematic of adding fines to 1 cm3 o f coarse material. (16.6)
6. 16.e..8)
since the volume has not increased. attention is directed toward the apparent densities ofmixtures of two powders. volume is 1 cm3.0 cm3. until the void space is filled The total up. and the porosity E is defined as the fraction of the volume that is void space. 1. obviously pc..3) yields the oftused equation this
E
= 1 = (p'/p)
(16. percolating powder (of apparent density pf) is added to the cylinder (Fig. that the small particles "fit'. then the density is
p=pc+m
(16. For simplicity (without losing generality) it is assumedthat the graduate has a volume of 1.
E=lvs
(1
This. hence. into the interstices of the coarse fraction). If fine.5)
Next. 16.7) (1 (16.. It is assumed that the fine powder can percoZffte(i.. is the void space in a volume of 1 cm3 with solidsfraction v. so the volume of the void space isE. that is.
WZ
= X&/(].
pl = vsp
or
v. the mass of powderin the 1.1 is considered be the coarse portion of to a mixture of a coarse and a fine powder. The weight of the contents is. The situation is that the material in Fig.The part that is not solid in the volume V is denoted the void space. then it will fitin the void space.
.4)
Intro~ucing into Eq. = p'/p
(16.m)
that is. efore the addition of the fine powder. The mass (weight) of this unit volume would be the apparent density.
14) .1 1)
When x > x . At the point of maximum density.10)~
1 /p’ = pc
+
YIZ
= (1(16. the weight of fine material is the volume it occupies {&J.“ C P f
(16.~)/pp~. of fine material. Hence. that so
1 / P L = N P p c .ut the mass (weight) of contents of the cylinder isequal to the apparent density.3). the reciprocal o f the apparent density of the mix (g~cm3).x)Ippc )
{(~ppc
p~)I(p~ppC)x}IIIPpc) +
(16. times its apparent density pi (Fig.p ~ ) / ( ~ ~ p ~ ~ left of the maximum apparent density To the ) . The volume w.
$.p~)/(piPp~)x}(l/Ppc)
+
(16. (1 . then the coarser particles are simply scattered in a bed o f the fine particles. so that
or 1IP’ = (1lPC){l . 16. (16.10)
This equation holds up to the point of maximum apparent density (mi~imum apparent volume). plus x/p).12)
The volume of 1 g of mix (cm3/g)is.x3 (16.the maximum apparent density of the mixture is
P. the
p’. however. [see Eq.where isthe particle density of the coarse fraction. ppC
Vmix
= ( x / ~ i + (1 .13)
This is a s t r a i ~ line with x = 0 intercept of (l/ppC)and intersecting the x = 1 ~t ordinate axis at (ppc.X)/&
S c h ~ ~ a t a ~r a n g ~ ~ ebeyond the point of maximum density. 1 g of mixture now contains a mass (weight) of 1 . of i r nt
. = L..x of coarser material occupied by this is and a mass (weight) of x.
009~ x values above theminimum and y = 1.
Whether tapping produces a closest packing or notis open to debate and is probably a function of the shape factor of the particles in question. according to ocke (1970/71) the lowest experimentally d e t e ~ i n e d fraction has been found to be 0. also known as facecentered cubic or hexagonal closepacked. that the traditional “onedimen. 16. 1978a). but the intercepts of the apparent volume plots fail to have the theoretical values.” a subject that will be discussed forthwith.26 1 . 1967.sional” tapping does not produce the closest packing. Hence.15) but.1 1 . Carstensen et al.sitie
so that here the reciprocal of the apparent density will be a straight line intersecting the x = 0 ordinate axis at l/p. experimentally (BenAimand LeGoff. but Berg et al. (1970/71) have reported this to be layers of hexagonal closepacked structure.2204 (16. Leastsquares fit lines are y = 0. a random shaking is necessary to obtain this. 1968. as shown in Fig 16.{7t/(3?72)} This is a rhombohedralordered packing. and the x = 1 ordinate axis at zero.
0
20
40
60
80
lOOD
% Fines
Linear relation of specific apparent volume as a function of fraction (or percent) for of fines.holds on either side of the maximum density.4. The relations have been borne out. The actual apparent density will be a function of finesfraction as shown in Fig. in the sense that linearity of apparent volumes with fines fraction.33 0. this would suggest a linear relation. Packing patterns in tapped density depictions are usually considered tetrahedral.5.0.
+
..01 lx for x values below the minimum. These authors have shown experimentally. ogers’ theorem (Rogers. the closest a packing of monodisperse spheres can attain is 1 . 1958).{31/2[~0~”(1/3) (7d/3)3} = 0. This is attributed to “wall effects.
d). or 0. The effect of hapes has been reported by Ridgway and Rupp (1969) and by Pitkin and cke (1970/71) found the relative mean spacing ( h / d ) to be independent of the reduced diameter ( d / D ) .4. as schematized in Fig. 16. the porosity I . but the important facet is essentially the finding by Rocke (1970/71) that the interlayer distance h is not affected by the reduced particle size. This author also showed that. for spheres.6.38 in such an arrangement. ne variable that governs pi is. if it is confined to a cylinder of diameter
. or real powders. f it is assumed. assuming layers to be arranged hexagonally.29.
~ / = 0. erg et al. and to have a value of 0. The dimenthe and arameter a = d / D (the reduced particle size) is often employed..of sphericity is overly simple. then the relative mean spacing h between layers is not affected by the reduced particle size.88. 16. cke (1970/71) showed that when monodisperse spheres are deposited in a cylindrical container. then. elly (1970/71)has reported on packing of a second layer placed a base layer.They treated data ~ublishedby y an inverse particle diameter plot. and (less intuitively) container diameter (2)) the mode ofcompaction. intuitively. mode of viewing the situation would be the following.88 d / D d
(16. but more realistic. (1969/70) derived a theory whereby the apparent density p'.0
25
50
75
100
% Fines
Apparent density data corresponding to the data in Fig. decreases to a limiting value. that the bed has an apparent density of pi at infinite bed diameter. ortional to l / d in a givenvessel. the assumption. a simpler.
~arstensen (1980) and Carstensen and Chan (1976) have reported on the porosity of conical heaps of spheres (encountered in repose angle measurements) and found a limiting porosity of 0.17)
ecause a pha~aceutical powder is not spherical (the closest to spherical being wet granulated granules). that is. and on has used probability theory to arrive at packing schemes. the particle size (diameter.
there be a wall effect at the wall and penetrating a certain e wall into the powder bed. of ass outer cylinder:(~/4)(L) g)2Hp. p1 = 0. and are From the model. the inner cylinder of diameter I) .Schematic of calculation of wall effect on overall apparent density of particles placed in a cylinder with height E? (cm).(L) . Wall effects are felt a distance of g/2 (cm) into the bed.8.g where the apparent density will be pi and the outer cylindrical shell of thickness / 2 .5. as an example.g)2}]p. who investigated apparent ortland cement and calcium sulfate hemihydrate.6.0.
(16.3.” eference ismade to Fig. They found that with
. and the graph ars this out.21) ta are generated using /3 = 1. (1969/70).18)
where is a constant. There are therefore two parts to the powder bed.7 is comparable with that found by rg et al. therefore. of the wall effect is a function of the diameter of the particle and is assumed to be of the form:
g = Bd
(16. 16.19) and similarly the mass (grams) of powder with density p i is ass of outer shell: [(n/4)H((D2) . The shape of the curve in Fig.8. p2 = 0. (16. This conforms with the model and Tarbuck (1966) except the wall effect is only assumed to ngth.depending on the material. g / 2 . d = 0. The weight (mass) of powder with density pi is. ecause the total volume V is (n/4)(D2H) follows that the apparent density is it (16.20)
The total mass (weight)M is the sum of these two terms. 16. The effective depth (thickness). where the apparent density will be g p i . the limiting apparent density (at infinite container diameter) should be pI = 0.
the maximum density configuration is the only stable one. depending on the method for tapping. such as the volumetric shape factor. vitaminA beadlets in vitamin formulations) the coarser particles will seem “to rise to the top. and to arrive at some sort of correlation between particle size distribution and stability of the powder. lending efficiencies will be a subject of a subsequent chapter.7
06 . but provided re no significant forces between the particles (that cohesion or electrostatic forces are not at play). For densities of bulk powders. Frequently. 16. it is quite reproducible.
c:
$ j
2
cz
05 .g. d = 0. but in accordance with the previous section is not necessarily the closest packing.
Real powders are not spheres.0.21) using B = 1.
. the packing radius (or the sphericity) is often referred to (Fig.5.8). using a third fraction of even finer particles that would fitinto the interstices of the binary mixture. is also fairly reproducible.0. the apparent density would increaseasymptotically to a l i ~ i t i n gvalue.” hat essentially happens is that the finer particles sift to the bottom.2it is obvious that if the fine powder is cascaded into the coarse powder. Hence.9’but eventually would settle down.
0.7 Graph plotted according to Eq. The tapped density. It could be possible to continue the argument.
increasing container diameter I).395 for cement and 0.2b). It is obvious that the porosity of a packed..4
0
2
4
6
8
1 0 1 2
~ o n ~ a i n Diameter er
.8. and they found that at these values. 16. it may temporarily “lie on topof it. (16. but surprisingly. This will be covered in a subsequent section. “real” powder sample may be less or more than that of a sphere. It is only when one has arrived at the maximum density that stability is imparted. E = 0. If one considers the situation in Fig.48 for calcium sulfate hemihydrate. p1 = 0. when one powder component is much larger than others (e. p2 = 0.
It is apparent that the cascaded density is not a stable configuration. 16.3. the bed would not be uniform (see Fig. and various means of describing the deviation from sphericity have been discussed in previous chapters.
then the packing of the mixture is based on the packing arrangement and on Table 16.24)
able
Arrangement
Bed Properties When One Small Particle Is Placed in Each Void Space
_ _ _ 
Porosity
0.22)
and the porosity may be expressed as
E = 0. so that the average coordination number for a ~ i x t u is ~ r given by
N = 1{ 12(21i2)Ff
+ 6Fr}/{2'/2Fr+ F'}
(16. If the fraction of the packing that is rhombohedra1 is denoted Fr. The coordination numbers are 6 and 12 for these packing modes..23)
The unit volume of a cubic packing is denoted unity. then the maximum diameter of the small sphere is related to that of the large sphere. and that of a rhombohedra1 packing is 2'12. the number of spheres in a unit volume are 1 and 2'/'. 16.1 are based on a cubic arrangement. hence..48Ff
(16. For theoretical considerations. is a mixture of cubic and rhombohedra1 packings.26 0. If packing radii are employed. Smith. then Fr
+ Ff = 1
(16.40
Diameter of the small sphere
Porosity of the mixture
Cubic ~hombohedral ~rthorhombic
Source: ~ a n e g o ~1936. on the average. et spherical approximations are often used.1 ( ~ a n e g o l d al. :
+ 0.48 0.26F.
. and refers simply to porosities. dIargee. If one places only one sphere in the voids between large spheres. It should be pointed out that the considerations in 16.. 1929. d. 1933) that the coordination number and the porosity.Loose and closer packing of irregular particles. With large steel spheres it has been shown (§mith et al.
The example in Fig. 1931). by the numbers shown in Table 16. respectively.2 is one for which a smaller particle is placed in the interstices of a larger particle. and that which is cubic is F f .1.
~ombining Eqs. and where the average value of N is 8.39.5 and then Eq. however.4E
6. 1967) it has been shown for a large range of porosities that
E
= 12.97)
It has been shown. and above this the packing density approaches 0. ( Tarbuck.414. 16. 1948). as shown in Fig. gure 16.5 .
t
nn
"N
0
5
10
15
21)
25
Distance&omwall (rrun)
Radial distribution of porosities of powder ina cylinder.25 and 0.25) 10.24) then gives
E
(0.23) and (16.0.9 shows random close packing where the solids (particle) fraction is 0.N(16.26) becomes.62. and the numbers used are between 8 and 9. (16.5.00043N2 (16. The coordination number will change if the packing changes positional~yin the bed.)
. when the particles are spherical.62. experimentally. that the packing density.62 ( cGeary.27)
(1
As mentioned when the particles are spherical. Experimentally (Ridgway and Tarbuck.is not constant throughout the powder mass.9. (16.
N = 22. ap~roximate~y. and rummer. and above this the packing density approaches 0. The porosity. 196Leva 1.414N . 1968.26)
ost encountered porosities are between 0.1 19iV
+ 0. the porosity increases as the ratio of
vessel to particle diameter increases to 10.
10 .6.072 .53)/'(0. where e 6 and 7 contacts between particles. increases as the ratio of vessel to particle diameter increases to 10.
Sphericity of particles as a measure of irregularity.Figure 16. the traces in Fig. ~ i g u r e 16. The abscissa in the figure is the coefficient of roughness divided by the particle density. it is ng to be expected that additives.5.6
07 .4 the curves will have shapes shown inthe figure.12. the and as this ratio increases from 0.8
0.995 .
mesh cuts are mixed.)
.10 employs the sphericity Se of the particles as a measure of ir . The data are quite scattered.1 to 0. It is seen in the figure that the following empirical relationship holds: ln[Se . and then decrease with increasing shape irregularity. Therefore. e 16.6
$ 3
02 03
0.0 for a perfect sphere.99) (~ For real powders.28) = 0.9 demonstrates the point that real powders may deviate from the spherical constructs described in the previous sections. 1950. such as lubricants (magnesium stearate) or glidants (talc). as defined by Wadell (1934) isthe ratio of the external surface are sphere of equivalent volume to the actual surface area of the particle.5
Pmsity
0.4
0. 16. 2 9 ~ & (16. (Data from Brown. will reduce the porosity in that they “fill out” the crevices (the roughness) of the surface.11 depicts the situation where the voids in the first component is 0.
1x)
0. but not their position. The size ratio between large and small particles will obviously have an effect. but the general trend of an increase in porosity with i ~ c r ~ a s i particle roughness is apparent. 16.8
09 . equal 1. This is of demonstrated in Fig.4 retain their shape.11 ==: 0.0. The shapes of particles ‘bed in haps 4 and 5.3 .
0. The rou~hness a particle also reflects the packing characteristics. then even tighter packings are possible.
039. each with one smaller sphere. and so on.2.)
. with as many level of spheres as possible) is 0. then the size of this smaller sphere can be calculated.
09
0.8
0.
Effect of size ratio of large to small particles on position. If the (now smaller)cavitiesbetween the spheres are nowfilledwithevensmaller (t~rtiary) spheres..e. 1937).)
upp pose the cavitybetween larger spheres in a given (e. 1961.4
Effects of particle roughness on packing c~ara~teristics. rhombohedral) arrangement (the sphere size arbitrarily set at size1. then the diameters of these may be calculated. (Data from Crosby. White and Walton.6 a 05
0.7
*10. 1929. ~haracteristi~s of orsfield packings are shown in Table 16.0) are filled. (Data from Furnas. This is denoted orsfield packing (Horsfield 1934. It can be shown that the limiting porosity (i.g.05
04 .
13 shows excellent linearity in all four cases.
. These ~alculatio~s shown in Table 16. nonseparable distribu~ u ~ tions.E ) =f3 = (1
v. If the fractions in Table 16. given by
vm = 1/(1 .207 0.&. but when they are filled.175
Following the development by ~hinohara (1990) the volumes of each of the sphere (Lee. It is instructive to examine the particle size distribution of the compositions with the minimum voids. secondary. a c ~ p s u l eare l o ~ n o r ~ a l l y fill ~ ist tribute^.2 for primary. then there is no potential for segregation. but it is one instance where closest packings that give ~ i n i porosity ( ~ n hence.g.(1
v = EVJ1 2
.190 0.3 gives the co~position when there are minimum voids. = (1 fl)&n2
where Vmis the unit particle volume.3 have the diameters indicated in Table 16. hig~est into a v o l u ~ e(e. secondary.31) (16.414 0. tertiary. It will later be seen. The fact that Horsfield packings give lognormal distributions does not exclude the possibility that other distributions may also give close.260 0.30) gives the total solids volume. nonsegregating.32)
v 3
= E2 Vm(l.158
Primary Secondary Tertiary Quaternary
1 .)/(l . Normal and Weibull distributions do not give good fits. then segregation may occur.2'7) to (16.0 0.primary. and so on) are given by
v1 = V.33)
vt= (1 .E)
=f1
(16.~haracteristi~s Worsfield Packings of Spheres
____
Ratio of size to primary sphere
Porosity of the mixture 0.E 2 )
~ u m m i n ~ (16. then the particle size can be calculated. that when voids are not completely filled.13.F 2 )
Table 16. V f as Eqs.225 0. The data in Table 16. Although there are only three points in each distribution. Fig.3 may be analyzed in light of the data in Table 16.4 and are shown graphically in are 16. and quaternary spheres.29)
E)
==h (1 f$)& =
(16.30)
(16.2.
(16. 16..
E
Data for Distribution of Horsfield Packing
= 0.84 1.0.176
0 0.5
d
11lfdJ
Y
70.1 24.6
Volume percent of fraction 1 70.08
0.7 86.8752 InEd].989).0
3
0.3 0. Z = 0. E = 0.39 2.08
0.55 1.995).0.414 0.6.5
Volume percent of fraction 4 1. squares.5 0.1 100
z
E
= 0.5.085 0.
Particle size distributions of Horsfield packings.1225 .3 96.063 . The symbols are small circles: = 0.53 2.4076 .O
0.63 1.6 = 0.5
00 . 2 = 0.5334 .9 3.9
"Fraction 1 is primary (coarsest).7 27.94201n[dJ.88 1.10 0.4 0.7 91.225 0.5 2.6 26.7 9.3 Cum
*A
61.7 61.o 0.3.0
1.8 98.4 16.50
46 73.6 90. (R= 0.998).1 100
0. large circles: E = 0.
3.0197ln[d]. 2 = 0.10 1.3 100
0.3 46.44 1.996). fraction 4 is finest.1.7 53.Composition When There Are ~ i n i m u m Voids When Four Componentsa Are Present Initial porosity in single component 0.4
E
E
= 0.0 t 4
05
0.8 13.6
Volume percent of fraction 3 6.0.9 6. 1931.91571n[d]. (R = 0. (R= 0.5 Cum E = 0. Source: Furnas. triangles: E = 0.0
Volume percent of fraction 2 21.3 Cum
E E
= 0.5
1 . Z = 0.4 Cum 2 Y*
E = 0.0 93. R = 0.1 100
= 0.3 9.276
.3 80.0 25
20
15
F
7 1.6 2 Y O Z
E
1.76
53.4.
p’ is the cascaded apparent density.e..14 show in particles that are 3 p m in size. and when powders are cohesiveand this. often occurs with fine powdersthen agglomerates may form. the powder will con solid at^ or compact.g. e.
. as we shall see in subsequent chapters. the apparent density after an large number of taps). but the basic process must be different. The curves.. 16. Each of the segments AB.14. from Fig.. (Data from Arakawa et al. All of the foregoing considerations have been based on the assumption of noncohesion. as shown in the figure. 16. The breaks in Fig. a grad~ated cylinder). This often ocxurs with very fine powders. 1966. and the break in the curves may be due to breakage of agglomerates. st is the amplitude.g. hen such a cylinder is tapped. are often parabolic.F
0
M
150
Number of Tap
Consolidation kinetics of white alundurn. then Compaction takes on different kinetics. 1956) that
It is noted.. the density that is obtained by pouring powder into. and ig. then it can be shown. but not in particles that hen a powder is vibrated. is plotted versus vibrational density 6 . 16.15 shows that the packing density goes through a maximum when packing density p.)
Theapparent densitiesreferred to in the foregoing are what is known as cascaded appare~tdensities(i.e. If piap is the tapped density(i. experimentally (Kuno. given by
where a is a constant. at points 6 ) in the curve for the 3pm powder. that there are “nicks” in the curves (e. and g i s gravitational acceleration. and pk is the apparent density after y1 taps. BC. and the line beyond C are quite linear..
0
Vibrational Intensity.1600e2xA2 R"2 = 0.50
1.0
3. = solids volume of particles in a bed vmlx = volume occupied by a mixture of a fine and coarse component v. = solids fraction x = weight fraction of fines in a didisperse powder bed
.5
4.5
3.5
2..932
~
0
Bulk mnsity
0.42388 f 0. (Data from Suzuki et al.0
2.)
See
a = constant in kinetic vibration equation D = (a) diameter of efflux tube. G
Packing density as a function of vibrational text for symbols. (b) gravitational acceleration H = height of cylindrical vessel ( ~ / = relative mean spacing in a closely packed bed ~ ) k = consolidation rate constant M = mass of powder m = mass of grams of fines in a didisperse powder bed N = average coordination number for a mixture n = number of taps Se = sphericity V = total volume of a bed V I = volume of a primary sphere Vz = volume of a secondary sphere Vn = volume of sphere in nth order void spaceVm = unit particle volume V.0
1. (b) diameter of cylindrical vessel d = particle diameter d / D = reduced diameter Fr = fraction in rhombohedral arrangement Ff = fractio~ powder bed in cubic arrangement of G = vibrational density g = (a) twice the width into a bed at which a wall effect is felt.y = 0.1 1 4 0 3 ~ 2. 1969.
BenAim R. Puisieu~ F. Powder Techno1 1:281. Ind Eng Chern 22:12061208.
E =
effect porosity S2 = vibrational amplitude p = particle density E.. McGeary RK (196 1). Academic Press. Chem Eng Sci 23:1147. p 214. M Manegold E. Carstensen JT (1977). LeGoff R (1967). Powder Techno1 2: 1. Zoglio MA (1978). = porosity between coarse fraction of a didisperse powder bed p' = apparent density. Crosby EJ (1961). In: Fayed ME.. Unit Operations. Furnas CG (1929). Rocke FA (1970/71). Chan PI. Garstensen JT (1990). Pharmaceutics of Solids. John Wiley & Sons. McDonald RL. J Roller PS (1930). pp 6385. = apparent density of a mixture of a didisperse powder bed at concenx trations of fines above.. Brown GG (1950).. Powder Technol 20:249. New York. Proc Fac Eng Keloh Leva M. Rupp R (1969). Trainor RJ Jr (1969/70). Berg TWO. eds. er Technol 4: 1801 86. Ridgway IC. Wiley. LeCoff R (1968). Ind Eng Ghem 40:415. Powder Technol 3: 183188. 966). New York. Horsfield HT (1934). Kelly EM (1970/71). KolloidZ 56:142. ~rurnmer (1948). housing a bed of powder p = cascaded apparent density ' p. 28 1. New York. Kagaku Kogaku 25: 124. R. Handbook of Powder Science and Technology. Garstensen JT (1980). Mehta A. H o f ~ a n n Solf K (1931). Bur Mines Bull 307:74. pp 9195. == apparent density after 72 taps
Arakawa M. Zairyo 15: 151. bulls density pc = particle density of coarse fraction of a didisperse powder bed p. J P h a m Pharrnacol 21:30S. Drug Dev Ind Pharm. J Pharm Sci 65: 12351239. Otten L.. pi = apparent density of a monodisperse powder in an infinitely wide bed pi = apparent density in layer next to wall of a cylinder. J SOC Ind 53:108. (1976). = fraction of fine fraction where maximum density occurs 3 = g / d = materialdependent factor connecting diameter of particle and wall . Carstensen JT. BenAim R. (1990). Rogers CA (1958).. Powder Technol Kuno H (1958).Solid Pharmaceutics: Mechanical Properties andRate Phenomena. Suito E (1966).x . 968). 44:5 13. J Pharm Pharrnacol 18:1683. = apparent density of fine fraction of a didisperse powder bed pmax= maximum apparent density of a didisperse powder bed ppc = particle density of coarse component pf = particle density of fine fraction pL. New York. Reinhold. Proc Lond Math SOC 3(8):609615. p 140. Okada T. Carstensen JT. J Am Cer SOC Pitkin C.
.
J Franklin Tnst 217549. Suzuki A. Physics 4:425. Walton SF (1937).:
. J Am Ceram SOC20:155. White HE. Phys Rev 343271. H. Foote PI>. I 0 Smith WO. ade ell H (1932). J Geol 4:310. Busang PF (1920). Powder Techno1 2:72. adell H (1934). Ta~ahashi Tanaka T (1969).Smith V (1933).
in the ejection of tablets from tablet dies.
The concept of friction is intuitively obvious. the weight divided by its cross section). 17.) It is also of importance in powder flow and. To move the object a tangential stress. z (tangential force divided by the same cross section). that ease of powder blending is. but its definition needs some elaboration. It will be seen in the following chapter. ith a setup as shown in Fig. the concepts of cohesion and friction need to be touched on. finally. then the downward stress would be the gravitational stress (0. The Concept of Friction and Frictional Coefficients 17. a function of the friction between particles.1. the load may be changed.
t=p0
(17. is necessary.1. The Concept of Cohesion
299
300
302 of Cohesion and Friction 303 306 306
efore progressing in this text. and the tangential force may be graphed as a function o f the
..g. If an object is placed on a support. Friction is of importance in several pharmaceutical aspects. to some degree. ~ a g n e s i u i ~ stearate) to powder mixes and granulations. that is.2. (This may partly be due to shape factors.17.1)
where p is the fricti~nal coefficient. The latter aspect necessitates the addition of lubricants (e. There is proportionality between CT and z.
or normal force. so that zero load cannot be applied as long as the experi~ent carried is out in a gravitational field such as that of the earth. thus.. blending. Even if this could be made weightless.r 17
f
N~rmal Stress
Schematic for the definition of frictional coefficient. the gravitational force wouldalwaysbe present.tance in blending and flow. All particles attract one another. and ejected it part way. ence. so is the concept that to blend particles they must be “pulled away from each other. It is also seen in Fig. it reveals nothing about the inter~articulatefriction. in the LaiCarstensen experiment. a discussion of the concept at this point is appropriate.
normal load. and cohesion and friction are intertwined in many aspects. that somehow a force must be applied to t an object to determine its frictional coefficient with a surface. that is. the tablet die. thesemust also be estimated by extra~olation.
Just as frictional coefficients are important in.g. For a noncohesive situation this will result in a straightline where the slope equals the frictional coefficient. when investigating the frictional coefficient between metal and a tablet surface. E is apparent from the foregoing text. 17. one bas to extrapolate to h powders there are also forces at play that cannot be eli~inated. compressed a tablet in a die on a hydraulic press. that can be applied is the weight of what “holds” the object in place.1 that the “smallest” load. The force q with which two particles attract one another is proportional to their mass m. and the force necessary to maintain speed could be measured. The surface of the tablets could then be dragged across a metal plate. Lai and Carstensen (1979). socalled cohesional forces and as will be seen. which is of such i ~ p o r . for instance.” This brings in the concept of cohesion.
q = /?’LE3
(17. This could then be repeated with different loads placed on the die. Although this may have meaning as far as tablet ejection is concerned (e. optimum amounts ofmagnesium stearate can be deter~inedin this fashion).12)
.
is the coordination number of the central particle with particles removed at distances ai from the central sphere (i.e. therefore. nearest neighbors.2.
. This particle has a coordination number of IV1. in the first shell.g. and they will cakeup when stored in drums.5)
where
/3 = p'"'
It has been shown by several investigators (Pilpel.8)
where N. Jordan. the number of particles in the ith she11 removed by ai from the central particle). and also has interactions with spheres farther away. the total force exerted on the central particle by all particles is (1 7. 17.7)
The force between the central sphere and one positioned at B would be
qj = / 3 ~ ~ ~y. It is also inversely 3 d proportional to the square of the distance betweenthem. here. As stated. the Each distance. This is why fine powders have a tendency to lump. there would be several Ni particles in the isphere.. For example. = ai= yid
(17.where d is the diameter of the particle and / ' is a constant.6)
(1
That is. The stress cr is the force per unit area. it follows that
/9 = /3"/d2
where /3" is a constant.9)
~c~ematic de~onstratin~ the concept of cohesion in a powder mass. the individual particle is attracted by all its neighbors. As shown in Fig. 1954) that cohesive force is proportional to the diameter of the particle.d)2/ = /3f~id/(y~) d3 (
(17. 1964. AB o distance is simply d. therefore
4 = Bd
(17.3)
7. 1937. andbecause the force acts across an area equal to the cross section of the particle it follows that
cr = b/d
7. of d.4)
(1 (17. the cohesive stress is the larger the smaller the particle. 1936. The considerations are made from thecentral (emphasized) spherical particle and its interaction (e. equal to 6. with a particle at position B). ~ s s u m i n g to be the same for both particles. Consider a central particle at A. Bradley.
(AB). yi.
and A gravitational force times cos [a] and the tangential force is the gravitational force times sin [a].1 1) mg sin[a]
hisis one equation with two unknowns.
mg = gpd31t/6
where p is the density. 17. Introducing this into Eq.12)
A
D
F
C
~ c h e ~ ashowing geometry in repose angles. 17. owder is placedin a hopper with the efflux tube blocked. The flow rate can be monitored in this fashion. ~bviously.3. and is called the repose angle (Fig. and the powder flowsout and forms a cone on the support below.1. when the distance ai is more than 3d.so that. and since the test is very easy carry out.. A particle on the slant is affected by two forces. gravitational force.1 1) gives
(i”7.10) where the summation is from i = 1 to infinity. assuming spherical particles of diameter d .3). It follows that. which is quite reproducible. The blocking is released. as shown in the second inset of Fig. AB. it follows that
+
h[C
+ mg cos[a]] =(17. and the powder forms an angle a with the support. but the number of particles will be muchgreater. AC and the the cohesive force. it still enjoys to popularity. putting this in the context of Fig.
n old technique that assesses cohesion and friction is that of the repose angle. the force contribution per particle will be considerably diminished. and the geometry involved yieldsa total force perpendicular to the slant plane of (AB AD). (17. ltd2/4)such that the stress is given by (17. ~ic
. 17. AB is the cohesive force. AC may be broken up by a force parallelogram. It is seen that the stress is inversely proportional to the size of the particle.he surface area over which all these forces act is the surface area of the central particle (Le. but since both are of importance in pharmaceutical applications.
with a plate corresponding to the cross section of the cylinders. as expected from Eq.14)
where $f and w are constants.7. as shown. This is not surprising. From the weight and the volume the porosity and the apparentdensity of the powder bed can be calculated. 1967. The types ~ e ~ loci~that may result are shown in Fig. and a horizontal force applied to the top ring.Carstensen and Chan (1976)haveshown the theoretical correlation between the particle size and the repose angle. if cohesionoccurs. The force required to move it (the socalled force at failure. i e If a powder is noncohesive. that n is not necessarily unity (Fig.985
~xperimental values of repose angleas a function of particle diameter. in a c o n ~ g u r a t i o ~ is stable).23162~ R A 2 = 0.e. because in Fig. sphere I will rest in the crevice between spheres T and 111.14) may be written: ln[a] = "PZ ln[d] (17. Nogami et ai. Often w is small.15)
+ in[+]
The result is. Powder is poured into them.
y
= :
4. 1961). 1955. A ( m a ~ i m ~ m ) isnow applied to the load powder (see Fig. however. 17. 1964.6.)
. The principle of this is shown in Fig. 1967) have shown that
a = ($f/d)
+
21
(17."1)542 .. Nelson. because the inte~rity the powder of bed fails) is recorded. 1964. in the lower curve ending in E. are the load and shear stress components at the degrees of consoli~ation used.6b).4). In general (Pilpel. owever.0. and at 30" or less. I that
Cohesion is most often measured by a socalled Jenike shear cell (Jenike.6c). 17.6a). They are placed. the connecting line between the centers of 1 and I1 will form an acute angle with the horizontal (i. 17. one on top of the other (see Fig. and the powder consolid~ted (see Fig. (17. Kananiwa et al. 17. so that Eq. 17. (Data from Pilpel. the line will be curved. 17. Theendpoints7 D and E. The apparatusconsists of two cylinders (rings).5 it is noted that at higherdiameter values the cohesive stress will become small. The limiting value of a as d + 00 is w and is often 30". to a certain degree... 17. 1965. Obviously. then a straight line results. Neuman. the consolidation has been lessthan in the upper curve.
Schematic showing the limiting repose angle of 30". 1957). therefore.:
Types of Jenike loci.
Shear Stress
a t Failure. powder bed. for that ~ o ~ at~ e degree o ~ c o ~ s o Z ~ ~ ~ t ~ be~noted that C is a function. 17. usually at an angle of 45" ( rks. fall on a straightline that intersects with the origin.
Features of the curve are the following: The endpoints to the right (D. as represented by the points G and 8 in Fig.
Top Ring
ons solid at ion
I
Load
The normal force is the loaded weights plus the weight of the powder and the ring. The intersect of the curve with the ordinate axis is the stress at zero 1 load.7. The weight ofthe upper ring and the powder in it constitute the ~ i n i load u ~ ~ that can be measured. r
"T
Applied Normal Load. and is the valueof the cohesive stress C (or force. of the state of the It should o . '
. dependi~g on that r whether force or stress is used as unit for the axes).
that this is not true.1 and Fig. An example of the foregoing concepts is represented by the data published by Kocova and Pilpel (1973) as shown in Table 17.16)
If the normal load at the endpoints (e. but rather the location where the method and the equation were developed). To this end. 1973)
YE
=1
+ 0. This may be measured in an apparatus similar to the shear cell.1 9)
They also found that the tensile strength is related to the maximum stress CT' by the relation T =h(pf/p)~
(17. 17.53d2'3
( 17.nT)/n' (n
YE
+
(17. it would be advantageous to have another point on the locus. The ~arrenSprings equation takes the form.
Jenike Locus Data Tensile strength.17)
is the shear index. 1973. a vertical upward force is applied. Therefore..8. then the equation may be expressed as
(z/cT')" = ~ n ( ~ f ) * .20)
where p' is the bulk density after compaction. but in place of loads applied to the powder bed.n
obtain C.
. T ( N m*> Normal stress. (Note that ~arrenSpringsdo not infer authors. and the force at which the powder bed fails is recorded. using the nomenclature of Fig.18)
The quoted authors found that
CR2T
(17. The curves follow the ~arrenSpringsequation. 1967. Hiestand and Peot (1974) have questioned the correctness of this and Carstensen and Geoffroy (1993) have shown. D and 9 is denoted CT'. and p is the particle density. 17.g. fairly long extrapolations are necessary. The curves do follow the ~arrenSprings equation but with some slight modification. ON ( N m*)
Shear at failure T(N m*)



174
226
330 630 930 1090 790 1090 1550
660 800 940 1070 1575 1920 2180
Source: Kocova and Pilpel. It frequently follows the relation (Farley and Valentin. Stanforth and Ashley.7. early researchers assumed that the intersect with the xaxis could be equated with the tensile stress of the powder bed.
= (0 T ) / T
+
(17. through iterationalfits of loci.
Units are in ~ ~ . (b) exponent in the correlation between tensile strength and apparent density of bed and particles N. Powder Technol 1:344. Powder Technol 15:129. 1973. p'.~ p r i n ~ s equation /!I. Chan PL (1976).
a = distance between two spherical particles b = constant relating force to diameter C = general term for cohesion d = diameter of a spherical particle g = ~ravitatioinalacceleration h = factor in the correlation between tensile strength and apparent density of bed and particles m = (a) mass of a particle. p" = constants relating interparticulate force to diameter of a particle a = repose angle yi = coefficient relating interparticle distance to particle diameter p = frictional coefficient @ = constant in repose angle versus diameter equation v = constant in repose angle versus diameter equation p = particle density p' = apparent density of bed CT = normal stress CT'maximum normal stress = z = tange~tial stress
radley RS (1936). Trans Faraday SOC 32:1088. (b) iterant in the ~ a r r e n . Carstensen JT. Valentin FHH (1967).~ / 1 0 (Data from Kocova and Pipel. = coordination number spheres of the ith shell YE = shear index q = force with which two particles attract one another T = (a) tensile strength of bed.) 00. Farley R.Normal Load/ 1000
The cohesive stress is indicated by the ordinate axis indicated at x = 0.
.
Int J Pharrn 1:33. Sugiwara M. Hiestand EN. AdvPharm Sci 2: 18 1. Ashley RC (1973). Neurnan B (1 967). Jordan D (1954). Powder Technol 8:33. Powder Technol 1: 199. Birks A (1957). Jenike AW (1961). Kocova S. J Pharrn Sci 63:605. Harnaker HC (1937). Pilpel N (1973). Aoki H (1967). Cartensen JT (1993). Williams JC. J Pharm Pharrnacol 16:705. Lai T YF. Powder Technol 76: 135. Nelson E (1955). Peot CB (1974). Kananiwa N. Kirnura S (1965). Ikekawa A. Powder Technol 7215. Stanforth PT.
. J Am Pharrn Assoc Sci Ed 44:435. Yakuzaigaku 25260. Nogarni H. Pilpel N (1964). Br J Appl Phys 5:S194. Physica 4:1058. Carstensen JT (1979). Chern Pharm Bull 15: 1441. Utah Eng Exp Stat Bull 108:l.Ceoffroy JM.
This Page Intentionally Left Blank
.
5.7.11. Types of Powder Flow 18.8.4. Wall Effects 18.6. Particle Enlargement
18.2.9.1. fractal di~ension) the particle of
.3. The roughness (rugosity. Powder Flow in Tableting 18. Dynamic Flow Rates
Symbols References
The prime interest in pharmaceutics relative to powder flow is that it affects tableting (and hardshell) operation in several ways. The shape of the particle 2. Effect of Efflux Tube Diameter 18. The ~easurement Repose Angles and Apparent Densities in of Flow Experiment s 18. The size of the particle 3.10. 18. Correlation Between Particle Diameter and Flow Rate 18. Correlation Between Repose Angle and Flow Rates 18. The flow rate of powders is affected by several properties:
1 . Flow of Po~ydisperse Powders and Powder Mixes
310 311 312 313 314 315 316 318 319 319 319 320 320 32 1
18. Effect of Moisture 18.18.12.
or other). There are some instances when it would be more appropriate to express the flow rate in units of cubic centimeters per second (cm3/s). the cohesion) 5. Conversely.g. If not otherwise specified in the heading. then the flow rates and repose angles may also differ. The moisture content
The firstsections in the followingwill deal with the flow of (a) first one fairly monodisperse component. The flow rate is then:
(18.in which casethe flow rate would be: (1 8. as is the mass (weight) M of the powder (Fig. and this may tie in with the problems. The chemical nature of the powder (e.2)
Time Recorder
Support (Table) or Plain
Recording
(a>
0)
Schematic for static powder flow ~ e a s u r e ~ e n t ~ . improper bulk density. If a series of batches of a drug product are made.1). easy to obtain. and the length of time 1 it takes for the hopper to empty is measured.
In its simplest form. 18. or the development of a product. It obvious that this type of is measurement also permits measurement of the repose angle. and one batch suddenly does not perform in a manner consistent with the others (poor tableting.4.1) This type measurement is often performed during the deve~opmental stages of a product and also. during scaleup and manufa~turing. it is a practical parameter. then (b) a polydisperse component or a mixture of two components. it is example (a) that is being discussed. powder flow is measured by placing powder in an appropriate funnel and blocking offthe exit tube. A timer isstarted at the time point the b l ~ ~ ~ i n g is released. and helps in the history of a product.. deviation from norm may bode problems in further processing.
. if it is used as an inprocess control. Although the shortcomings of repose angles has been men ti one^.
flow rate is the slope ofthe line..la). The radius of the cone r is then ( A / E ) " ~ the tangent of the repose angle a is and
a = h/r
(18. and when it becomes intact again the end time is recorded. the determination of the volume V is simply.
Repose angles are often measured in a rather oldfashioned way. The height h of the cone may be measured (best by an optical micrometer i. may bedetermined. The 18. conn~cted with the exit block.where p' is the apparent density of the powder. and that problems (e. a vertical micrometer with a telescope arrangement). ~onsequent~y. but refines the measurement.2. but the "regularity" of the flow that is of importance. Apparatuses have been constructed to facilitate the measurement of the beginning and the end of the flow. flow meters are equipped with a some ence. the recorded trace will have the shape shown in Fig.. The determina
Recorde Weight
(a)
(b)
Traces from a recording flow meter.e. 18. that is of i~portance. 18. and poor consistency during the flow operation maybe problematic. The conical heap may be caught on a piece of paper (Fig. because they may be different. The problem then is whether it is the apparent density of the powder in the hopper. and when the beam is interrupted the time records as zero. In 18. an electric beam past the exit of the efflux tube.g.
. The LewisHowe flow meter uses a timer. The radius of the cone maybe estimated by tracing the (almost circular) contour of the cone on the paper and d e t e ~ i n i n g area A (in the the simplest fashion by weighing it and knowing the weight of 1 cm2). At times it is not only the flow rate. In Fig. in t tablet in^) may be anticipated. An average flow rate may be estimated (either by least squares or drawing a line by eyea socalled leastsquares wrist).2a the powder is freeflowing. and from this the apparent density pN after flow. For pharma~eutical operations. This is optional.3)
aving the height and the base area. or that in the conical heap. flow of powderthrough the appropriate orifices must be of fairly uniform rate.2b the flow is highly irregular. but the i ~ p o r t a n feature is that the flow rate is not uniform.
there are two aspects of flow in this situation. powder is filled into a hopper (Figs. This is implied in the figu by having smaller amounts in the die at 1. 18.
In a tableting operation on a rotary machine. and a b necessary that the powder flow from the feed frame into the die cavities.3 and 18. one is the flow from the hopper into the feed frame. 18. then more at 132.5)
1 g. If the feed frame is a cm long. and if the die table has a radius of R. from whence it flows into the cavities formed by the lower punch and the tablet die. e. )
If the tablet being made has a compression weight of feed frame into hopper must be at least W = D / z = 1)2nwR/a
then the flow rate from
(18. and may be of value recording. one from the feed frame into the dies. v cm/s. is
z = af2nwR
(18. of a die is
u = 2naR
so that the time a die is in contact with the contents of the feed frame z. again for the assembly of a data base. this then flows into a feed frame. should a future batch of powder deviate from the n o m .
. and even more at 3. With good flow all these may be equal. These are two cascaded apparent densities that may (or may not) differ. it is necessary that the powder flow from the hopper into the feed frame.6)
The value of D is the volume of the die V times the appr~priate apparent density p’ so that
w = Vp’ZnwR/a
(18.3.4).4)
(18. and the rotational speed of the die table is w rotations per second (rps). The description is somewhat simplifi but demonstrates the principlesinvolved. then the linear speed. As shown in Fig.e.tion of the apparent density beforethe flow starts (i. ence.7)
Feed Frame
A
II
B2
B1
Lower Punch
. in the hopper of the funnel) p‘ may be determined by gradating the funnel.3 S i ~ p l i ~ e d schematic of powder flow in a tableting operation on a rotary tablet ~ a c h i ~Side view.
7). the dies will. and this force may be dissolved in direction PS (into the wall of the tube) and (toward the sphere B. Sphere A. Consider the situation shown in Fig. and this is one reason that flow rates are so important. Larhrib and Well (1988) have described the effect of the speed of com~ression tablets made from polyon ethylene glycol~icalciumphosphate mixtures. the more economic the operation.
. be full at point I33 (provided the flow from the hopper is adequate). Eq. It is economic to operate the machine at as high a speed as possible. given by Eq. in the next “row”). (18. so that the h i g ~ e r flow rate. (adjacent to the wall of the tube) is affected by gravity (PT).8) denotes the maximum. at which the desired tablet weight may be obtained.. Sphere I3 isaffected by the two
A
A
Schematic of flow patterns. critical speed. and the machine speed is set at such a level.
U I
Feed Frame
Q
Rotates at o Rotations/S~c Die Table
~ i m ~ l i sche~atic rotary tablet operation. ~ed of
The weight D delivered is ‘
D = wa/2nzc. Atspeeds below mcrIt.
The types of flow that may be encountered is discussed next. At speedsabove mcrlt the achievable fill weight willdrop inversely withthe speed. the Not much recent work has appeared on this subject. (18. coccit. ‘
(1 8. Top view.8)
For a given flow rate. therefore. 18.5.
If it is the former that governs the flow then the mass movesas a plug (plug flow. 18.5c. and all these forces add to a vertical force. and there will be no flow. and to the force normal to the tube. and the slower the flow. and if the interior moves “faster” than the particles close to the wall. hence.7 result (Carstensen andChan. At very small particle sizes the cohesional stress becomes more important. 18. as they move downwards. then there is particle movement in the cylinder of spheres (turbulent flow). 1976). Hence. and also by gravity. as shown in Fig. the flow rate is larger. consists of a layer next to the tube.6). 18. plots such as shown in Fig. for
~ r a v ~ t a t i o nStress al
Stacking
Schematic of blocked flow. 1956). the larger the contact area. and interior layer (Fig.neighbors in the row above it. When particles are toofine. It is obvious from the foregoing that the two main factors that affect flow are particle shape and size.. 18. because at one point wall effects start being of importance.
. they will not flow readily out of a hopper or into a die. the smaller the particle. however. then the surface area would be onefourth. the smaller the frictional coefficient is between the particles in the outer layer and the tube. and the interior a force in the downward direction. talc for instance (Strickland et al. ~xplanation this are attempted in the following sections.
This is the most important aspect of flow rates. The flow of poorly flowing powders may be improved by socalledglidants or run~ingpowders. so that the contact area would double. usually when the particle diameter approaches 1/20 of the efflux diameter. There is also an upper limit. Frequently.5d). therefore. If the diameter were halved. The former has a force toward the an wall. ecause of the cohesion associated with small particle size. for a sphere of diameter d. When flowis plotted versus particle diameter. and at a given particle diameter the cohesional stress in the lowest exposed layer will exceed the gravitational force on the column aboveit (Fig. The frictional stress is proportional to the contact area between the outer particles and the tube. other means of flow improvement must be found. The powder mass. For plug flow. increasing the particle size will improve flow. but thenumber of particles would be eight times as high. this would be related to nd2 per particle. The closer to spherical the better a particle (powder) flows. akin tolaminar flow).
1977.20
n
0
2 15
W
8
0)
c) . 1975. The force.. 1972)For the purpose of the discussion to follow. Similarly. the less extra force remains to cause flow. a small increase in the angle the tube exerts a~ainst the tal would cause flow.a]
(18. F at vertical position (angle being 90” with the horizontal). but if the angle is increased a bit. 1967.10) or: (18. at the contact angle (the tangential stress) to have the relation (in line with Fig.11)
. as seen. 18. just superseded when the angle of the tube is increased beyond the value of a. and this as demonstrated in the fashion shown in Fig. Kaneniwa et al.
F = Fa/ COS[9O0. 1967. Neuman. The flow of the powder is associated with the repose angle. Pilpel. being related to the force Fa.8 .The larger the contact angle. Cillard et al...
2
E
10
$
5
0
1000 Diameter (micron)
Correlation between particle diameter and flow rate. 18.8a a situation is shown in whichthe repose angle a. necessary for flow is. opper design (the angle of emux tubes. 18. the slant of the hopper cylinder) is essentially based on affording geometries that work as well as possible for as many types of powder as possible. reference ismade to Fig. flow wouldnot occur (it does not occur on the surface of the conical heap). ~uantitativelythis may be expressed as the “remaining” force. 1971. particles would flowdown from the side ofthe heap until just a is restored.)
. so that qualitatively it may be seen that a larger contact angle causes a slower flow rate. Fukuzawa et al.9)
een seen in Chap. 1965. 17 that the relation between repose angle a and particle diameter d is given by (18. ’ 7
In the past there has been a flurry of publications on this subject (Carr. (Data from Carstensen and Chan.8). If one filled a tube and placed it at an angle of a with the horizontal. is fairly sm is fairly large).
1 1) program To the in Table 18.9.1 l).6 Y5 = 2800/Y4 REM This is d PRINT X2.
With knowledge of K and Q. Y5. As demonstrated inSec.6” are plotted in Fig. “Flow Rate” 1. (18.9)”( 18.4. (18.15708/4) ngles from 30” to 60” Y1
=
SIN(x1)
is force. it is obvious that if a powder had particle sizes ranging from 100 to 2000 pm. 18.9 is similar to the one expected at lowdiameter values (see Fig. except that it does not decrease at very highdiameter values. This means that there is some upper limit on the fore
DataGenerating Profile for Eq. (1 8.9).9) to Eq.Y2 NEXT X1
. 18.6” PRINT “Angle”. It accounts for the increase in flow with increasing particle size.1 has been written. d may be calculated from a by means of and a quantity proportional to flow rate may be calculated as a function of a by way q.7). test the profile of such relations as Eqs. “Diameter”. CT It is seen that the curve in Fig.c
E
Schematic showing effect of repose angle on flow through a tube. Using~ value of K = a 280 and Q = 28.18. 18.0472 STEP (. but the behavior at high particle diameter is missing. (1 8. proportional to flow Y4 = X228. then the larger particles wouldblock an orifice which was 1000 p m wide. The results from the program using a value of K = 2800 and Q = 28.
and U = 0.
Program in Which Wall Effects Are Taken into Account [Eq.12)
where U is a constant.
going considerations.05 there will there be substantial wall effects.1 using a value of Q of 28. In general. Y5. characteristic of the powder. For a given tube diameter.1416 Y1 = SIN(x1) Y2 = 1jYl REM This is force. when d / D is larger than 0.1 by adding the steps Y6 = (1/Y2) . Q = 28. “”Diameter”. I INPUT ““Qvalue”. “Flow Rate” FOR X1 = S2. 18.0002 results in the data in Table 18. The printout using the values K = 2800. Many authors use the reduced diameter d / D . where D is the diameter of the efflux tube. 1/D will be a constant. U PRINT “Angle”. as the workable parameter.3 and the data are plotted in Fig. This feature may be incor~orated the p r o g r a ~ in in Table 18.04’72.Q INPUT “Uvalue”. This has been done in the program in Table 28. proportional to flow Y4 = X2Q Y5 = K/Y4 REM This i s diameter Y6 = Y2 .12)]
S INPUT “Kvalue = ”.U”d (18.10. (18.15708/4) REM this represents angles from 30” to 60” X2 = X1*180/3.(U”Y5) REM this is adjusted flow rate REM Y5 is d PRINT X2.36 TO 1.6” and a value of K = 280.Y6 NEXT XI
.6.0
5 0 100 1 5 0 200 2 5 0 Diameter (microns)
Data generated from the program in Table 18.2.STEP (.
It is noted that this translates into a dependence on orifice diameter of D2*5(i. a power of 2. a 30 32.6.4.67 1.e. arbitrary units 1.
It stands to reason that the larger the diameter of the efflux tube.25 24. d ( p m ) 2000 767 474 269 188 144 117 Flow rate. 18. d.
..
y comparing this curve with Fig.5 39 43.5).0002
Repose angle.32 1.5
(18.6 1.42 1.7 it is seen that it has the expected shape throughout the diameter range.3 Using the Values I(: = 2800.14)
where pp is particle diameter.24
.5 Diameter. the more rapid the flow. Q = 28.Diameter (microns)
Data from Table 18.4 = y D
+p
(18. The equation by Brown and Richards (1960) is of the following form:
4W/(nppg)0. This has been verified by Danish and alrrott (1971). and U = 0.3 Printout from Program in Table 18.72 1. is a constant that depends on the 8 particle diameter.13)
where D is the orifice of the efflux tube and .53 I . and a similar dependence was found by Jones and and c) who arrived at the following equation:
W=
15n)Wppg”2D2.5 48 52.
. is a “running powder’’ or a glidant. the coarse particles will “roll down” and separate out at the base of the conical heap. and spraydrying. as willbeseen in a later chapter. for instance.
euman (1967) found that moisture in solid samplesacts as an enhancer of flow. but for completion. They. so the two situations may be treated as one. thereby. with content uniformity. 1970 ~trick~and. The effect of moisture makes many flow experiments somewhat uncertain. is often a necessity. It was simply a matter of relative ~umidity the room.
Powder mixes are mostly polydisperse. that if a repose angle experiment is carried out. 1956).therefore. tablet operations. is improved withsmaller ut from the point ofview o f flow. roller compaction. the p r o b l e ~ of segregation occurs. in a powder. but suffice it say. technologists often talk about “fines’.” or powders that have a certain percentage of an ingredient or fraction (on a number percent basis) that is much coarser than the remainder. In general p~a~maceutical operations are geared at producing polydisperse powders of a fairly narrow particle range (e. in smallamounts the fines may act as a glidant ( anish and ~ a r r o t t . the powder ran fine..e. wet granulation. ~anipulation the drug recrystallization (rate of cooling. owever. a tablet where the crown potential of dislodging itself). For instance. where the flow blocked in a tableting operation. At one time I was called in as a consultant on a flow situation. of for instance^. In very few investigations has the relative in humidity in the environment been subject to scrutiny. tend to stick to surfaces of larger s and “keep them apart. the flow rates of a powder. it is mentioned here that it can be accomplished in four ways: slugging.g. weight percentagewise. Particle size enlargeme~t. when present in small amounts. This will be treated in some detail in chapters to follow. so that in such situations the initial transfer will result in a larger proportion
. it is associated with (although not necessarily the mechanical reason for) a defect known as a “capper” (i. it is obvious that (below the ~ a x i m a ) particle enlargement is of importance. and tableting was no problem. and should be considered in assessment of published data. granulations). This is a factor that should be included in such experiments. than the particle size of the largest fraction is often de~eterious.here. The presence of material that is much finer. in small amounts. may give somecontrol over the particle size distribution and. First.. dissolution is i m ~ r o v ~ d the larger specific surface area associated with the smaller article. and it will be the subject of several subsequent chapters. but with larger amounts slows down flow.
There are many reasons for keeping drug particles small. In addition to this. The same type of behavior will occur when a powder is discharged from a mixer into a drum. When the demonstration of this took place.
A bin below the die table would then catch the powder that flowed through the die.thereis aprofound difference. particularly in direct compression.
(a) area of the base of a cone in repose angledetermination. (b) constant in the Jones~ilpel equation a = (a) lengthof a feedframe. In this instance.
It is tempting to try to simply calculate the adequacy of a powderblend for a tableting operation by obtaining the flow rate by one of the mentioned methods.. furthermore. there is sufficient vibration in a tableting operation to make the statically found flow rate wrong and. as is the diameter of the die. there are machine attachments (forced feeders) at the bottom of a hopper. 16). With polydisperse powders where no size “predominates” (e. and these are used very often. If the coarse fraction is an active component (e. thelevelofpowder will change.g. but the vibration is. of maximum density as describedin Chap.of coarse particles at the wall of the drum into which the powder was discharged. The apparent density of the material flowing through the dieisless than even the cascaded apparent density. and because the test is nondestructive. so that compressionof powders on highspeedmachinesoccurswhere the material has a lowerdensity than the densities that may be determined by more static means in the laboratory. and the flow rate W wouldbe M / t . that will help the flow along. the mass (weight) M of powder could be deter~ined an ex~erimentally for determined time.(b)lengthofdie table inCarstensenLaughlin experiment D = (a) grams of powder in a die. it is possible to actually perform the flow rate experiment usingthe hopper and. They found the flow rate to be a function of die table velocity by: In[q = ln[u]
+
(18. then this may be a source of content uniformity problems. or the flow into the die. to have a flow meter using a die as the efflux tube. The diameters of the orifice of the hopper is known. (b) diameter of orifice of efflux tube D’ = D = grams of powder in a die I.. It should be noted that this latter is not all that important. The velocity would then be: u = a/t. t. For cohesive powders..15)
where q and k are constants. reposeangles are fairly reproducible and the conical heaps are fairly uniform.g. and (b) how much will flow into the tablet die. and then calculate (a) how much will flow out of the hopper per time unit. than in the center. For the hopper. In general the vibration helps. similarly. vitamin A beadlets). Carstensen and Laughlin (1979) constructed an experimental apparatus in which a rectangular die table (with one die) could be moved below a hopper with a rectangular opening of length a on the bottom. = force acting on powder in a tube Fa = force just sufficient to not allow a particle to slide down the slant of a cone h = height of a cone in repose angle determi~ation
A
=
.
Danish FQ.
.Solid Pharmaceutics. pp 9699. JT. Roland M (1972). Carstensen. Pharm Phamacol 18:429. Pilpel N (1971). Fukuzawa H. Pilpel N (1966a). Jarninet F. Lahrib H. Adv Pharm Sci 2: 181. Busse L. Jones T. Yakugaku Zashi 95:859. Fukoka E. Int J Pharm 160:197.iy = constant in the equation relating repose angle to particle diameter k = constant in the Carstensen~aughlinequation Ail = mass of m a t e ~ aflowed into or t ~ r o u g h die l a = constant in the equation relating repose angle to particle diameter q = constant in the Carstensen~aughlinequation R = radius of die table r = radius of a cone in repose angle determination rps = rotations per second t = time V = (a) volume of cone in repose angle determination. Pilpel N (1966~). New York. 184. J Pharm Belg 27:713. Powder Techno1 23:79. J Pharm Pharmacol 18:182s. Chem P h a m Bull 15: 1441. Adv Pharm Sci 3: 174. L. J Pharm Sci 66: 1235. Wells JI (1998). Ed 45:482. Carstensen JT. Academic Press. Higuchi T (1956). J Pharm Pharmacol 18:31. Aoki H (1967). Carstensen JT (1981). Pilpel N (1966b). Kimura S (1975). (b) volume of a die w = linear speed of a die W = flow rate (g/s) W' = volumetric flow rate a = repose angle = constant in the rown and Richards equation y = constant in the Brown and Richards equation w = rotational speed p = apparent density of material in a hopper ' p". Strickland WA Jr. = apparent density after flow p p = particle diameter z = contact time between powder and die
Carr R (1965). Ikekawa A. Neuman €3 (1967). Chem Eng Lond 72(C):163. Laughlin S (1979). Kaneniwa N. J Am Pharm Assoc Sci.MechanicalPropertiesandRate processes. Jones T. J Jones T. Parrott EL (1971). J Pharm Sci 60:550. Chan PL (1977). Gillard J.
This Page Intentionally Left Blank
.
it will in this region. Strain = &stress (19. return to its original shape when the stress is removed).1)
. The words may be used interchangeably. this book does not cover individual pieces of equipment. As mentioned in the introduction. the latter word implies the manner in which it is done. The introductory remarks to this chapter are simply that the intent is to outline the principles by which milling equipment works.5. however. The former word simply denotes that particles have been made smaller. a crystalline solid.. and that impact will cause crack propagation. ~illingInduced Particle Size ~istributions
330 332 333 334 334
illing~nduced Changes in Morphology Symbols eferences
The word c ~ ~ ~ i ~particle~sizei reduction.
The principle of grinding is that all particles have flaws. the word ~ i ~is ~ i denotes ~ ~ g most often used. when exposed to stress will first deform elastically (Le.323 324 327 330
19. In general.
applies in this region. 19. do not have the “correct.
. the particle will not. the density willalso remain unaltered. which roll at a predetermined speed. amorphicity may result. the yield point. and in certain types of milling (ball milling of long duration). The principle of a ball mill isshown in Fig. mortar and pestle are the means of grinding. The findings using preliminary procedures are often nonapplicable to the findings later on in the development of a drug. return to its original shape. At this stage only small amounts of drug substance are at hand. necessitating smallequipment that can be operated with a minimum of powder loss. the fracture point. For stress. To attain this they are milled. Powder and balls are charged into the cylinder (in upright position). raw materials. Usually. If the stress has not been applied for too long of a time. if the stress is released. In regions beyond the plastic limit.
Brittle Fracture
Plastic
Yield Point
4 1
Elastic Deformation
v1
$
E O
0
2
4
6
8
1 0 1 2
Strain (Arbitrary Unit)
Stressstrain diagram for a solid. At a given point. and milling is the first unit operation that is encountered in phar~aceutical pro~uction and development. a lid is placed on the open end and secured.’ particle size and surface area. the cylinder is then laid horizontally on a pair of rollers. if the stress is released. when largerscale means are used.
It has been seen in previous chapters. Types of millsthat will be discussed here are ball mills. and the solid will deform. and the work horses in scaleupeven moderate scaleupare the hammer mills. and fluid energy mills (micronizers). the crystal “breaks. This is denoted plastic ~ e ~ o ~ is obviousi that in . It ~ ~ t ~ ~ this region the crystal lattice is strained. In this region. applied for long time periods. Ball mills are usually used only in early stages. 19. that surface area and particle size d i s t r i ~ ~ t i o n are of importance in pharmaceutics. hammer mills. and the density could increase ever so slightly eyond a certain point (Fig. At the preformulation stage of product development.1). but even more important in compression. then the particle will return to its original shape. as received or synthesized.2. there could be some vacancy release. the elastic limit is exceeded.” These concepts are important in milling.
will decrease by ln[w. 1977. There is an optimum amount (weight) of balls to powder. however. omm mi nut ion isgenerallyconsidered a firstorder process (Austin et al. which is (in radians. to decrease in time t. but rarely in pha~aceutical production. If a millingis
. r. of material with the original diameter d‘. then the mass w. at time t. w). the mill should be only onethird full. This is the centrifugally limiting speed.2)
Ancient milling wascarried out with millstones. and (e) the speed of the mill. In general. Reid. 1965.3)
There are.. but in general. there are eight balls of radius of onefifthto onetenth. In general. the intensity of milling increases with speed. (b) the ratio of balls to powder.and in smaller scale is used this in tars a ~ ~ ~ e s t in e s laboratory. because they are only of very smallscaleinterest. but there is an upper limit. the morphological changes that may occur in a powder through milling are verypossiblein both andpestling and ballmilling. the mill is supplied with the optimum amount of balls. and when the ~entrifugal limit is reached.
w = (g/r)”*
(19. Not too much attention will be paid to the foregoing two types of mill in this text. However. Gardner and Robers. of a the radius of the cylinder. all millshave also been of interest in studies of milling kinetics. (1978) and Mehta et al. Gardnerand Austin.. all mills are frequently usedin laboratories. the balls will simply sit toward the wall of the mill and not move within the mill./w] = kt (19. The mill may be operated at different speeds.Principle of ball mill. 1976). the homogenizers that Z the depend on rotating cones and stators are based on this principle.. and the maximum and minimum amount of powder is recommended for a particular mill. At lower speeds. 1975. and then no omm minuting takes place. Austin. In liquid processing. Jindal and Austin. 1971/1972. (1977)have shown the following treatment to hold for pharmaceutical powders and granulations.
The degree ofcomminution depends on (a) the size ofthe balls. 1975. Carstensen et al. Austin (197111972) found that millingwill cause the mass of material that is of the original size. If w grams are milled. 1976). examples where this does not hold (Austin et al.
the energy input will be It has been discussed in earlier chapters that Arealvolume where
=
l?/d
(19.(l/d$’2} It is obvious. The energy input is proportional to time.( W l ) l
ond has suggested that it is square rootdep~ndent. depending on mill and equipment.11)
(19.13)
where q1 is a constant. and the original powder one of d. and a./d”]
(19.12)
(19. Introducing this into Eq. The weight mean diameter of the particles then is d” given by: where It follows that ick’slaw differs somewhat from ~ittinger’slaw.10)
where a. of the material that will have reduced in particle size to db. thatis..8) gives
E = QWd2) . there will be a certain amount wb.4)
where E is energy input and C is a constant depending on milling e q u i p ~ e n t and substance.carried out such that the milled material has a mean diameter of dl’.8)
where
=4 / 1c
(19.13) then gives
(19. E = q1t
(19. are surface and volume shape factors. assuming a surface energy of p2 erg/ cm’. that energy increases with increased surface. that is still of the t original diameter d. and a certain a m o ~ n w.which states that if the energy is used to create surface AA. At a given time t. in any event. then Kick’s law (Parrott. Employing Kick’s law in the following and combining Eqs.15)
. then. that is. E = ~ { ( 1 / ~ 2 ) 1 / 2.9)
r=
(19. 1970) is expressed as follows:
E = cln[d.. (19.(= d”). (19.4) and (19.
owder enters the feeding hopper.correlate with iterated values that impart linearity to the experimental data. Screens in the most common mills (where the energetics are intermediate at best) are interchangeable. The hammer is wedged on one side and if this side is forward in the rotation it is denoted “sharpedge forward. The powder will havea certain residence time in mill house. hence. from which it enters the mill house. (1978) showed that values of db. the feed rate must be smaller. The higher the speed. There is a relation between the size of the screen opening and the feeding rate. when sufficiently small. they may be metal screens with holes in them.4) that the particles that have leftthe mill house are always smallerthan the Screen
.and the particles will fracture under the impact of the hammers and. where hammers rotate. then their residence time in the mill house must be longer.
. the smaller the particle.Carstensen et al. They may be either metal wire on a metal frame. If they are large relative to the screen opening size. The former will be a function of the degree of reduction of the original particles. and if it is forward in the rotation. the particles will not leave the mill house perpendicularly.” This latter has a greater commuting effect than the former. or forthe larger sizes.
m e ~ From a practical point of view. The energy ofthe milling can be changed in several aspects:(a) the speed of the hammers may be changed. but rather at anangle. determined experimentally.3) are the most common. There are usually three speed settings. then one refers to it as “blunt edge forward. This effect ismore pronounced the higher the speed of the hammers. so that the particles are always smaller than the opening in the screen. Powder In
House
Powder Out
Hamm~r mill principle. will exit through the screen on the bottom of the mill house. the ~ ~ m mills (Fig. or (b) the direction of the hammers may be changed (blunt end or knife end forward). ecause ofthe whirl caused by the rotating hammer.” The other side of the hammer is simply straight. It should be noted ( 19. 19.
1~ = Noekt
3000
(19. then. then the mill cannot handle the load.ter 1
0
of the opening. and the mill house will fill up.and many mills are jacketed. Relative to the optimum rate of ~iZling. that is. an angle b).g. considerable conden~tion (Le.. but this may cause metal fatigue and cause breakage of hammers (which may then project through the mill house). moisture increase) may result.16)
*E!
c1
2000
lo00
'
a)
ii
c
3!
00. and the produced average particle size is shown in Fig.. the effective exit area for a particle is q cos[90 . some operations it is a practice to add dry ice directly to the feed.0
012 * 014 016 * 018
'
I:o
*
It2
Screen Opening.)
. any mills are equipped with controlledfeedin devices.
0
Manner in which a particle exists. When particles are milled to the 15 to 50pm range. the air must be very dry. 1993. Ifdone in this fashion. one viewis to consider that if No particles are introduced into the mill house per minute. 19. This is because. if a powder is fed too rapidly. the direction of exit is at an angle with the screen (e. hence.p] = q sin[b]. The optimum rate is the maxi mu^ rate that will permit milling without blockage of the mill house. m m
. Heat is evolvedduring the milling process.5. An example of the correlation between screen opening size. For one thing. at the optimum milling V rate. then the milling step is often referred to as micropulverizing*This is done in jacketed mills and (depending on the cooling liquid) is often referred to as cryogenic ~ i l l i n g In . the direction being at an angle to the direction
opening q. (Data from Carstensen. otherwise. i will survive fracture in an exponential manner. The mean particle size of a milled powder as a functionof the screen opening size. so that they may be cooled.5.
then the distributions will tend toward normal distributions. This point of view often suffices.If the particles are not toodifferent in size.81where /3 is the surface energy. but hard granules seem to mill into lognormally distributed stock. The bimodal distributions often occur when granules are produced by insufficient granulation (Le.e.) = ln(nAA)
(19. (1974) studied the effect of milling on the distribution of particle sizes in the milled material.21)
. which in turn gives rise to the energy E. expended. and k can be determined from the plot. z is then determined when it is determined what ~/~~ value is satisfactory. When granules are soft.the larger the number of holes a. This can then be carried out over several timeperiods. As mentioned.20) Equating Eqs. The efficiency of the milling operation would be givenby the ratio M / that at optimum efficiency. In between the particles willbe either bimodal or willfollow a Weibull function:
where f is cumulative fraction above a diameter of x. In a holed plate screen. each hole will cause a resistance. but throws no light on the particle size distribution that may be expected from a hammer mill operation.. E is also inversely proportional to the area throughwhich the milled powder is forced. then material will accumulate in the mill house. and this is assumed proportional to a d"' where the constant.20) gives:
A h[d] 3.18) If (95% of) the incoming powder is (above) a certain size Do.. so that the following equation would describe the energy consumption: (19. AA in surface area A [see Eq. and the material itself. h. Steiner et al. so that (19. the residence time z is (19. equipment.17) If the mill is operated at a higher input. and 8 and # are Weibull parameters. energy input is related to the increase. then the mass M will be proportional to N . but part is left still the way it was before the addition of the granulating liquid). / M . then deter~ining the amount that passes through a screen of this aperture after milling for z" minutes. in the process a part of the material has been granulated. (19. is a function of hole size. the less the resistance (i. the smaller the resistance).8) and (19. The larger the area.ln{az//3. 19.
a certain number of “original” particles ne r become agglomerated. The types of distribution curves obtained are. strong screens. and are jacketed for cooling. What. they do not contribute to the bonding in the tablet. makes solid bridging in tablet formation more difficult). the coarsest screen on the top.The finepowderisremoved centrifugally and iscaught in cyclones and airbags. or bimodal (Steiner et al. a too much harm donein havinga certain ~ ~ percentage of fines. the sieves are separated. anci In the second. and the amount of material on each sieved determined by weighing. This causes particles to hit one another and breakone another. Steendam and Lark (1968) report the use of cryogenic milling to grind granules of poly(D~lactic acid) of high molecular Adolfsen et al. hence. sieves are stacked on top of one another (a nest). Weibull. normal. The number of fines are a function of (a) the adequacy of the granulation procedure.. depending on the granulation process and the milling conditions. so that a strong turbulence is created in the milling chamber. In the former. The specific surface areas (for pharmacokinetic purposes) are minimally about 34 m”/g. cooling is necessary in milling. (19. ~ Z Z
.21) to hold well for nine pharmaceutical granulations. cause of their lack of binder. Air is introduced in two positions of a flat cylinder. The original material is usually premilled so that it has a particle size of 20/ 100 mesh. and the nest is then shaken in a prescribed and reproducible fashion. and micronizers are usually housed in separate explosionproof rooms. Even so. this method. There is always danger of dust explosions under such circumstances. The particle sizes attained are 120 pm. it is a case of small. The air at the two inlets is introduced at different pressures. (1974) found Eq. 1974). and ti n the usual means of obtaining particle size distributions is by way of sieve ~ n ~ Z ~ ~ i s . the particle sizes are usually large.r l
Steiner et al. the minimum particle size attainable with them is 2050 pm. lognormal.
If really large surface areas arerequired.) These are attrition mills. is of practical importance is the percentage of fines (particles smaller than 200 pm). and reappear in their native state in the final granulation. and too large percentages may cause tablets to split (cap) along failure lines made up of adjacent fine particles or pockets of fine particles. and (b) the time and intensity of milling. At times. and (c) affects fracture propagation in tablets. for too large a percentage of fines will cause problems when the granules are tableted. After shaking. Usually 100 g of granulation is placed on the top sieve. 21). (The most powerful hammer mills (micropulverizers)have very massive hammers. other means than plain hammer mills must be sought. gran~zate~particles. (b) increases the deformabi~ity asperites in the of surface. (1998) have shown that milling of sodium chloride (a) makes it more difficult for particle surfaces to rearrange (and.
For g ~ ~ ~ ~ Z a (toobe scovered in Chap.
2Na(l . where d' is a function of milling time t.. and for simplicity it is assumed that each impact will break a particle in half. because (e.a)2 6Na"I . regardless of what the value
+
Number of Particles that Remain After N Impacts Size Impacts
1
2
3
ma
N(1 . Carstensen et al. have shown.a)"l &2a(l .a) particles) will remain of size xol2. N(1 a)"
xo/.1 are obtained.. after ofie rotation.a)? 4Na(l . (rnJ3. Therefore. The total number of particles is now N(l a). N(l .a)ml Q24a2(1 etc. the fraction that has a size of x = xo/ZPis ~ ( ~ ) ( 2 ~ ( 1 a The fraction of size xO/Y is therefor^ (19. ere. that this relation is approximately semilo~arithmic. Assume that there are N particles of initial size xoin a sample to be milled. The total number of particles i s now
+
oceedin~ this fashion.
.g. p is a number between 0 and m. ln[P] = kt (19. that a fraction of (1 .. xo/* xo /
Total
N(1 . ln[2""'] = ln[2] ln[2"]). and as seen in the table. of the population.a > N2a
N(1 . (1978) and ehta et al.a ) of the a will N(1 .22)
y ~ar~tensen and ate1 (19'74) have shown. on a p r o ~ a ~ i l i tbasis.and that a fraction a of the 2Na particles with size xo/2 will be halved (ie.a ) 4Na2
N(1
yy1
N(1 . The particle diameters are loglinear.a ) of the particles will be left unchan~ed and 2a N of the particles will have a size of x0/2.effect of milling time may be expressed as the percentage P of material retained on a sieve size d'.a)I2 will remain ~nchanged]. in The number of particles after w1 impacts is N(1 + a)mand the possible sizes are from (x0/29 to xo.a) particles of original size [i.4 3 4Na(l ..e. that there will be 2Na2 particles of size xo/4).a)
~ 2
N(1
+a)
+ a)2
N(1
+ a)3
~
3
+
"The number Ql implies combinatonal
over I lie.. that if a monodisp~rs~ powder is fed at a constant rate into a hammer mill.23)
The distribution on the righthand side is normalized binomial. assumed that each impact (of the hammers in the mill) will fracture a given fraction a. (1977). N(l . the data in Table 19. then the resulting ground powder will be lognormal by number. and for large nzvalues this will approach a normal distribution.e.a ) of the 2Na particles [i. After two rotatio~s similar ar~ument show that a fraction (1 .
83 with a correlation coefficient of R = 0. t 1981).j 1996.02..)
. Figure 19. such as HPMC(Sugimoto et al. 1978).000 rpm hammer speed and a Jet mill (Jet0Mizer.6). Arias et reve al.
+
16
14
12
10
8
6
3
4
5
6
7
8
Pressure in Mill Chamber (bar)
Effect of applied pressure in a fluid energy mill on the a m o ~ n t amorphous of material produced. riggner et al. This increase in surface energy ispresumably and primarily due to conversion of crystalline to amorphous solid. 1994. NJ) equipped with a herringbone screen at 0.1976. Nakai et al. and using a 14.16x. Summit. 1986a. 1959).
Recent literature abounds with reports ofmillingincreasing surface energy and causing distortion of crystal lattices (in addition to the comminution) (Vamaguchi and Sakamoto. They found the particle distributions by weight to be lognormal with means of 6 and 18 pm. 1998). and manifests itself in an increase in dissolution rate. ~ . 1974. (Data from Briggner et al. 19.c y c l o d e x t r i ~ ( ~ i t et al.. amount of amorphous material produced.. (1998) studied cogrinds of furosemide with crosspovidone (polyplasdone. The intercept is not significantly different from zero. and found an increase in dissolution rate. microcrystalline cellulose ( ~ a m a ~ oet oal. Shin et al.... because simply grinding gives the same result as cogrinding.b). 1997).6 shows the effect of applied pressure in a fluid energy mill on the 2. Fluid Energy Alljet. therefore.99. will approach Zhan Johnson and (199’7) have used a Bantam mill (Bantam lverizing Machinery. and gelatin (Kigasawa et al. It is also a common practice to cogrind drugs with polymers. chitin and chitosan (Koh et al. The increase in dissolution rate is not due to the presence of crosspovidone. for large m values the distribution. Plumsteadville. The leastsquares fit isy = 0. PVP).ce... PA) at a 90 psi pressure and nitrogen gas to mill an e~perimental drug (CP 118 954).. (1994) employed isothermal microcalorimetry in the study of changes in crystallinity induced during the milling of powders (Fig. in.
as a function of milling time t p = number of impacts (between 0 and m) Q = coefficient in relation between energy and inverse diameter difference q1 = coefficient connecting energy of milling to time of milling Y = radius of a ball mill t = time of milling w = mass (weight) of material being milled w.T = milling rate constant = qI/C k = milling rate constant M = mass (weight) of particles after time t of milling Mo = mass of particles before milling N = number of particles (a) at time t. = weight of particles with diameter d. h = exponent in equation relating energy input to diameter of hole in screen and number n of holes o = rotations speed in radians z = residence time in mill
=
A C
=
. = mass (weight) of material being milled with the original diameter wg = weight of particles with diameter db x = particle size x = initial particle size before milling in a hammer mill 0 a./a. = volumetric shape factor a = fraction of particles impacted in one rotation in a hammer mill a2 = coefficient inequation relating energy input to diameter of hole in screen and number y1 of holes == Hooke’sslope = strain~stress 8 2 = surface energy 8 3 = Weibullcoefficient Q. (b) after m or p rotations in a hammer mill No = initial number of particles before milling m = number of impacts M = number of holes in a hammer mill screen P = percentage of material retained on a sieve size d’. = mean particle diameter before milling E = energy input into a milling process f = cumulative fraction of particles larger than x g = gravitational acceleration li. where dl.surface area of powder E / ln[d’/d”] = factor connecting energy input with particle size reduction during a milling process d’ = weight mean diameter of particle population after milling db = average diameter of the milled particles d. = ~ e i ~ u l l exponent I? = shape factor = d times area~volume= a. = surface shape factor a.
Bystrom K. Carstensen JT. Arita T. Sakamoto K (1959). Powder Technol 12:65. Clnem Pharm Bull 26:3419. Johnson KC (1997). Lancaster. cars tens el^ JT. J Pharm Sci 66: 1462. Puisieux F. Drug Dev Ind Pharrn 22: 1237. J Pharm Sci 63:1395. Bull Chem SOC Jpn 32: 1364. eds. Parrot EL (1986). Int J P h a m 160:187. The Theory and Industrial Pharmacy. New York. Cines JM (1997). Lachman L. Int J Pharm 153: 18 1. Int J Pharm 1:65. Konno T (1978). Carstensen JT (1993). riggner LE. Marcel Dekker. J Korean Pharm Sci 16:36. Powder Technol 14:35. Patel MR. In: Lieberman HA. 2. Pharmaceutical Principles of Solid Dosage Forms. L Mitrevej A. Rogers RS (1975). Pilpel N (1968). Zlnang Y. Carstensen JT (1977). flakugaku Zasshi 101:733. A. Adams K. Lerk CF (1998). Lea & Febiger. Lee YB. Nystrom C (1998). Lee YB (1986a). arwood CF. Nakano M. Shin SC. Arita T. Kooyama 0 (198 1). eds. Yarnamoto K. Kigasawa K.
~01. Austin LC. Gardner RP. Lieberman HA. Moyano JR. p 51. Koh IB. Buckton C . Nakal Y (1976). Austin LG (1975). Int J Pharm 154: 179. Trimarchi T. Mehta A. Nakano M. Tanaka M. Powder Technol 5: 1. Sinchaipanid N. Nakajima K.
Lantz RJ Jr (1981). ~arintournuwat (1996). Steiner C . Int J Pharm 175:33. J Pharm Sci 63:1494. Zoglio MA. Watabe K. Int J P h a m 105:125. J Pharmacokinet Yamamoto IS. Austin L C (1971/1972). Takayama Y. Arias MJ. Lee YB (1986b). J Pharm Sci 65:1484. Choi HK. mdal VK. Choi JS (1998). Shin SC. pp 2147. Carsensen JT (1974). Philadelphia. Weymont NP (1977). In: Lachman L. Junyaprasert V. Darcy P (1994). Int J Pharm 175:17. Arch Pharm Res 9:55. Nakai Y (1974). Nakai Y. Pha~aceutical Dosage Forms. PA. pp 77152. Terada K.Ado~fsson Caramella C. ardner RP. Pate1 MR (1974). Yamaguchi G. Powder Technol 12:247. Kanig JL. Technomic Publishing. Austin L C (1976). Powder Technol 17:109. Mehta A. Steendam R. Maruyama K. Oh IJ. Yamamoto K. Zoglio MA (1978).
. Koh IB. J Pharm Sci 57:478. Shin SC.
.9.1
*
Statistics of Ideal egregation of Noncohesive Powders ixing of Noncohesive Powders inetics of the Noncohesive
336 338 340 34 1 342 346
20.5. As such. it is desirable that the mixture be “uniform. and judging the “goodness of mix.8. or a powder.10. all go through a stage in which the product exists as a powder mixture.
20.4. and (b) the standard deviation.” the “completeness of mix.
20. 20. assaying these. ~ h e t h e r not the final dosage form is a to ry or tablet.” or “the de of mixing. The material to follow will concentrate on binary systems.
352
There are no solid dosage forms (except sachets) that are onecomponent systems. 20. 20.of the mixture. but the findings may also be extra~olated ~ ~ l t i n a systems.6.3.2.20.” and the means by which this is assessed i s by taking samples from various spots in the assembly.’ by way of comparing the results with (a) the theoretical mean x.
20.
20. a capsule.7. ffect of Particle Size rdered (Cohesive) Mixing Symbols eferences
347 347 348 35 1
352
20.
.
In general the number of ways IZ items may be removed from a total of N items is given by combinatorial N over IZ.1)
Examplewise 141}== 4 x 3 x 2 x 1/(3 x 2 x 1)( 1)= 4. 20. The factor “4” is the number of ways one particle can be taken from a set of four and is denoted combinatorial 4 over 1. as enumerated physically in following.E.
Perfectly blended spheres of equal size. a sample containing one drug particle could also be obtained as EDEE.. symbolically written here as 141>. for example. The probability ofpicking the first particle as a drug particle and the next three as excipients (Le. so that. all the samples would contain a fraction x of drug and have a zero standard deviation.n)!n!}
(20. To illustrate the theoretical effect of sampling ( E $ ) . However. When the combination containing two of each comes up.ter
Ideally..422). DEEE) would be (1/4) x (3/4) x (3/4) x (3/4) = 0. The total number of
This will simply be possibilities follow. First. however. and that the sample sizewas4.2) It is an oddity.105469 (i.105469. is referred to as an “ideal mixture” in the following. by sampling.0039 or 0.e. as shown in Fig.given by:
{&} = N ! / { ( N. The probability of picking a drug particle would be onequarter.1. and not explained here. and this. This number would be given by (42}= 4!/{2!2!} = 6 ways (20. the probability of picking four drug particles one after another would be (1/4) x (1/4) x (1/4) x (1/4) = 0. = 0.
Assume an ideal mixture of noncohesiveparticles. or EEED. so that the probability of obtaining one drug particle would be four times 0. hypothetical situation. suppose the sample were taken one particle at a time..3) accepted. a note on how degrees of blending are assessed.39%. or EE. how many combinations of 2 Es and 2 Ds are there).
. that
O! = 1
(20. the question is how many ways can this be done (Le. The mixture contains a fractional onequarter of drug (dark circles) and threequarters of excipient (light circles).
It is noted that the foregoing holds for an ideal mixture. and standard deviation s.2108 of the time 3 drug particles (43~(0.75)3= 0. In general the probability of obtaining n drug particles.1N]1’2
(20. is a o The mean x. this is approximately zero.0469 of the time = 4 drug particles (44}(0.3164 of the time 1 drug particle (41)(0.0039 of the time
(20..9)] are (20. of the distribution [see (20.75) 0.75)4 = 0.11) (20. from a sample of N particles containing x fraction of drug.4) (20.x)/(Nx)]1’2~~. As an example.x)Xp2
(20.75)~ 0. but some dosage forms (e. is the expression in (20. would be (20.X ) X
s = EN( 1 . and that accounts for all the possibilities). For large numbers of N it will approximate a normal distribution.00 as they should (one would obtain either no.6) is the s ~ a l l e spossi~le t a n ~ devi~tion t s a~~ that may be expected by samplin~ a ince N is usually large.7) (20. (20. two. (20.25)3 = 0.0 drug particles (40}(0. (20. variance s2. (20. may have a limited number of particles per dose.9/0.14)
So that
N = 250
.25)3(0.13). (20. rsd = loo[(1 . sustainedrelease pellets).5) (20.9) The distribution of Eq.8)
The n u ~ b e r add up to1.12) divided by Nx (the average number of particles in the mixture). that is.13)
the factor 100stemming from the fact that rsdvalues are usuallyexpressedin percent. a term employed often in blending science.9)is known as the binomial distribution. so that in such situations the relative standard deviation attributable toprobability ( E ~ ) may be rather large. or s three drug particles in a sample.12)
The standard deviation is the square root of Eq. and that the rsd in Eq. and the relative standard deviation (rsd).6) (20. one. that
6 = 100[0.10)
s2 = N( 1 .1l).10 fraction drug (and the remaining particles being blanks of the same size) and have it still meet USP requirements (rsd = 6%).4220 of the time = 2 drug particles (42}(0.25)(0. one might ask what the smallest number of particles that may be used in a sustained dosage form with x = 0.g. The answer to this is that the smallest number is given by Eq.25)2(0. The point here is that the assay one obtains from ra~dom samplin~ f~nctionf the sample size.
.E
F
G
Sampling device for model study of barrel rolling used by Pate1 (1975).
the inner cylinder is arrested. Insertion of the thief will often rupture the bond between snlall. the sample taken. Thieves with multiple holes also exist.
Sample Plug
Schematic of a plug thief.Open
Closed
Principle of thief sideport sampler. or cohesive powders (to becovered shortly). encasing a certain a ~ o u nof powder. The sample can then be harvested after the thief is co~pletely of the out mixture.
The sample size is dependent on the size of the cavity. er.” The principle of a thief fairly adequate for cohesive powders is shown in Fig. but have certain disadvantages. to be discussed later. At the appropriate (samplin~) location. and in certain thieves there is a large cavity. As this is cohesive it will not “fall out. an endthief maybeused. The sample size depends on the difference in position of the outside sleeve and the bottom of the inner cylinder at the point of sampling. that manner the same thief may be used for many sample sizes.
.4. adhered particle. will not be representative of the mixture. The thief is lowered with the outer cylinder flush with the bottom of the inner cylinder. 120. Hence. into which fits a similar volume plug with a smaller cavity. so that the powder will unmix at the point of insertion of the thief. dependent on ordered mixing (to be discussed in the following) the sideentry thief is not a good choice. for cohesive powders. and the outer sleeveislowered further. Cohesive powders form “plugs.” when the thief is t retracted.
. (a) is never noncohesive. being the larger the larger the difference between the two. this is shown in Fig. before discussing blending in detail. and that the standard deviation s of the population would approach an equilibrium .s i. They showed. complete mixing is never possible. it wouldbe disturbed by the mixer so that. To assess what factors affected segregation. ppie and coworkers also found that the rate constant k. by perturbing these completely uniform populations that the spheres would separate.. Because. Olsen and Rippie (1964). a sample size of four particles in square array would always be in a ratio of 3:l. and (c) ordered blending. The rate constant was also a function of the agitation intensity
~
Perfectly blended spheres of equal size. It is possible to perceive an arrangement of particles that is““completely blended” or “completely uniform.16) where k. was a function of the volumes v1 and v2 oftheballs. but for for larger particles it is insignificant. A microsecond after this geometry had been achieved. attributable to the mixing energy.7). For instance in the particles in Fig. 20.b) conducted experiments with balls. as visualized in the foregoing.e. sB. (20. not only from a sampling point ofview.. that from the point of view of noncohesive mixing. and ippie et al. would steel They arrange a “completelymixed”assemblyofsteelballsoftwotypes(e. k. that .6. in a cylinder. thereby. 20.” as long as the sample size islarger than the rational number that equals the ratio sum.two different diameters. On the other hand.5. type (c)is a subdivision of (b). We learn. The equilibrium level is a function of the intensity of the perturbation (the energy input of the mixer). there willalwaysbe a residual standard deviation. that it would be possible ina mixer to attain “perfect mixing’’ at a giventime point (Fig. (b) cohesive blending. some cohesion is alwaysat play. two different colors). (1964a. suppose. a zero s ~ a deviution is never attuinable in practice.There are several types of blending: (a) noncohesive blending. owever. In a manner of speaking. a short note on the opposite of blending (Le. 20.
. is a segregation constant. and then vibrate the cylinder.g. to repeat. segregation) is in order. but even under idealmixing conditions.
~ al
.0 High
A
u
Medium Low
0
10
Time (rnin)
Rippie’s segregation experiment. and will be discussed later. as well as the rate constants. were plottable by an “Arrheniustype equation” using llenergy (1 divided by amplitude of the vibration) as abscissa and lnfkJ as ordinate. 20.in that it will cause mixedness of a blend. (Data from Rippie.
Obviously a mix will becomerandomi~ed way of the mixing action.8. because the higher the energy input the h i ~ h e r be will the final standard deviatio~. The energy input. More modern blenders. does not affect the mixing. The faster the material “mixes” the more efficient is the mixer. The sampling procedure. but rather. are based on the same principle.b. essentially. and the final by mixedness will be a function of the following: The efficiency ofthe mixer in a positive sense.)
(the energy input). schematic ally^ in Fig.repre~Zes sents common blenders in existence for the last 50 years. 1964a. The types of blenders most often used in pha~~aceutical manufacturi~~ are shown. the result. however. The figure represents ~ ~ ~ ~ c iin fact it . and the equilibrium level.
A ~ y ~ o t h e ~3:lcperfect blend in a ribbon blender. This.
To achieve blending. 120.
Planetary Mixer
In the following. but the principles arrived at are equally applicable to multinary mixes. If two materials are placed in a mixer that is then started.BlenderDBarrel Ribbon
Types of blenders. so that the mixer may not be used to full capacity in any event. At times layers are alternated. but the essence is that the two components lie in separate layer. In any event. first necessary step is an expansion of the bed. The word “noncohesive” is a
Mode of noncohesive blending. This is one reason that one cannot “scaleup” capacities of mixers by simply determining the apparent density ( p ’ ) and then multiply the volume V ’ of the mixerby p‘ to calculate the mass (weight) of‘ particles that would fit init.
. The motor driving the ribbon or blades that cause the blending provide the energy E necessary to separate the particles and to make them move across one another. and must be allowed to pass one past the they other. most manufacturing situations call for a “round” number of tablets (e.9. because the larger the batch. 20. This is exemplified in Fig. they are then placed “on top of each other” (Fig. For the layers to assume a nondense configuration.9). the more economical the operations as concerns assay and labor costs. only binary mixtures will be considered.. 2 million). This in spite of industry’s zest for costcutting. the particles must be separated (a sort of ~uidization).g. a force must be applied that overcomes the cohesive force between particles.
of sampled N(l .O.
et us assume that mixing is to be carried out between two solidcomponents. (one of which could be a drug substance). and that there are two parts of ascertain that they were “completely mixed.g. If a thief is loweredat random at N points in the (1 ) mixer..x)x2
+ Nx(1 
= N(1 .x. The work expended deals more with moving layers of particles across other layers. but assume that it is a powder blend.x) times. then a fraction x of the time (e. ~ o ~ eis~ i o ~ the force c (small case) between two particles. but it is minimal.and will be treated first.x )
(20.x of B. onethird of the time in the foregoing example).x) of the time (e.~ Z ~
. in general.x)
(20. 20. and then A. Also. so that the variance is
s2 = Nx(1
.10 is the 2:l mixture described in the opening paragraph. and mixing such powders is referred to as rzoncohesive mixing. in which component I3 is placed in the mixer first.1. that is. So cohesion and friction are of importance in mixing. The mixture shown in Fig.” would be necessary it rent parts of the mixer and assay them. so even in noncohesive s blending thereis the work associated with separation of the particles. The average content is x. would differ from the mean by 1 .g. Sampling is never carried out at t = 0 and the figure is simply shown to demonstrate what the inrtial standard deviation ~ o be. an would differ from the mean by x (since their content is squares X would be
= N( 1 .convenient misnomer.17)
The number of degrees of freedom is N . wherethere is a fraction x of A and a fractio~ .x) = Nx( 1 .x)x(x
+ 1 . Nx times. the smaller the particle the larger the hen the particles are fairly large (more than 50 pm) then the cohesive stress is relatively small. and it is defined as proportional to the mass of the particles and inversely proportional to the s q u a r ~ of the distance between the two particles.x)/(N . For an ideal noncohesive mixing case the initial relative standard deviation willbegivenby the following argument. so samples co~ntingonly since their drug content would be 1. or. because there i cohesion at all times. the thief would sample A and a fraction (1 .18)
if N is large. the standard deviation would be
Example of sampling positions before blending. and this is where ~riction~Z forces have to be overcome. that is.1) a: x(1 . two~hi~ds the time in the foregoing example)..
s. (20.22) to ascertain that the particle size of the product (i.23)
or
s = .]
= kt
(20. (20.20)
If it is assumed that the rsd decreases to a limiting value and that this final rsd is governed by sampling error only. ~egulations. When . which is: rsd = lOOs/x
so that the “initial” rsd is:
rsdo = 100[N(1 . then it is dictated by a b i n o ~ i adistribution.24) exp(kt)
2
ut aside from this there is the standard deviation attributed by the energy input of the mixer. (20.)(20. A point should be made.e.25)
It suffices to say that the relative standard deviation will approach some finite number.24) simply becomes
s = so exp(kt)
(20.19) ( N . limits to a lower limit the number of particles that a dosage form may contain.)] = kt
(20. the more efficient is the t blending.s
+ (so . the number of particles in the dosage form) is larger than the minimum dictated by the equation. N is the number of samples taken. in which one the factors thatplays a role is the energy input in the mixer.s. For general dosage forms (immediatereleasecapsules and tablets). n is quite large. but for sustainedrelease pellets it may be rather small. rsd.22)
where y1 is the number of particles in question. The “allowed” standard deviation in blending sampling. From theory it can be shown that the final standard deviation of the binomial distrib~tion given by is
.x ) /(20. nowadays.e. therefore.so = [Nx(l .s.s
= {( 1 . employthe relative standard deviation.21)
t is of importance to know what the final condition is and how fast one arrives at it. so that the final standard deviation will be
2
sm
= Sbmornlal
2 + $energy
(20. ~alculations should.)/(so
.1)]1/2
ere. however. (20.. Equation (20.x ) / ( N . is 5%.
. the number of particles in the sample islarge).26)
and Eq.27)
This is what will be assumed in the following. which is the case even whenthe particles are moderately small (i. which is of sample size of13 times dose weight. Eq.22).1 ) ~ ] * / ~
(20.22).x ) / ( N . of I is the socalled blending constant. as a consequence. that l
so = 100[Nx(1 .x)/xn}]1/2
(20.23) becomes: ln[s/s. the complete blending equation is given by: 1n[(s . of the influence of Eq. always be made with Eq.s is small. The larger it is.1 ) p 2
(20.
1 plotted as shown in Eq. Fig. and the initial rsd. The blending rate constants k are a function of the fractions.13. x and (1 . by the foregoing. (20. An extremum in the initial variance [(see Eq. (20. The functions x(1 .
Standard Deviations of a 2:1 Powder Mix as a Function of Mixing Time Sample position
1 2 3 4 5 6 7 8 9 Avg
Time
=
0
Time = 1. consider the data in Table 20. are dealt with in the following.5 min
10 90 70 80 60 40 85 90 75 66.4
RSD
0 100 0 100 0 100 100 100 100 66.6 33.0 3. 20. The reason for this.As an example of mixing kinetics. that is.2x = 0.7 12.11 results.1. when x = 1/2.18)] occurs when 3[2]/3x = 1 .12 results. 20.6 50
0
10 Time
20
Data from Table 20. This is a 2: 1 powder mix.x and (1 .6 min
5 95 30 70 40 90 96 84 90 66. When these are plotted versus time.2
Time = 15 min
65 68 68 66 61 63 69 72 69 67.
. should be ~ o ( : 1) = 100(9( 1/3)(2/3)/8} = 50 2 (20. 20.28)
The table shows the individual assays at nine points in a mixer as a function of time.4
Time = 6.26). and when the logarithm of the standard deviation is plotted versus time.x ) / x ) are shown in Fig.x of ) the two components. and its qualitative consequences. Fig.
so that for a 100kg batch. This is not practical on an industrial scale.2
0. To these two parts of premix were added two parts of excipient to form premix 2. and it can beshown that.2). where strong mixing wasnot available. it was customary to take one part of drug substance and mix it with one part of excipient. It is usually carried out in barrel rollers. The concept of geome~ric mixing an old concept and is has its roots in the fact that in compounding. premixing isoften resorted to.4
X
0.992
10 Time
20
Plot of figures in Table 20. can be mixed most completely) and is the fastest to mix (and most convenient to handle i s the initial steps). ~onsequently.y
=
3.x)x. then the premix should be a total of (0.
10 r
x) /x
0. so that the ratio of the premix would be approximately twothirds the volume of the drum to 90% of the volume of the final mixer.
Traces of the functions x (1
. then the rate constant would be maximum at x = 0.8051 .8
10 . The principle of geometric mixing is belief that a 50:50 mix is the “easiest” the (i.6
0. and this preblend then mixed with the remaining 60 kg of excipient.x) and (1 .17640~ R A 2
= 0. the best amount of premix to employ would be given by (x)’/’.1 plotted as a function of Eq.
. For instance if x = 0.e..16.0.
or the reasons shown in Eq.5. under such circumstances. the 16 kg of drug substance would be mixed with 24 kg of excipient(to make a total of 40 kg). (20. the blending may cause a problem. so it is a general practice to make one premix (not a geometric number of premixes) and make it in a ratio that is convenient with available equipment. If this were true.22 it is obvious that for compositions in which the drug concentration is very low. 20. and so on until all was blended.16)’/’ = 0.4.
3. then mixing is exceedingly slow. This depends on the surface not roughness of both compounds to be of equal effect. it is seen that sufficient force or energy must reach all particles so that they may be separated. to be discussed shortly) that gave rise to the development of socalled highshear mixers.15 The powder that passes the choppers will have the cohesive stress overcome (i..154) and a fastmoving chopper (see Fig. with much less energetic mixing. the mixing process discussedhas concerned itself with the blending of mono~is~erse. impellers. If particles are of different sizes and particles of A can percolate in the ig. Fig. 2.e. then segregation will not occur.a significant influence of particle size on
ate1 (1978) and Carstensen and Patel (1977) have shown the following: 3. if nonexisting. (Data from Patel. and the blades A. the more stable the powder mix.
From what has been described about cohesion.)
. If particles are of equal size (Fig. but s e g r e ~ ~ t i o n also is fast. it should be noted that the degree of se~regationis a function of x and the smaller particle diameter(s). 20. i. There is. 20. and the closer the particle population is to such a state.14~) then blending is fast. 20.14. however. if not nonexistent. the particles will separate). binary mixtures. 20. If they are such that maxim~m density has been reached (the interstices are just filled).So far.b.e. 1978.14. 20. mixing is fast and complete.a)..If particles are of different sizes and particles of A do not fitin the (see Fig. and segregation very slow. does not percolate) or vice versa. In these there are slowmoving blades (paddles.
A and B same size
B smaller but cannot percolatepercolate can
B smaller but
Influence on particle size by mixing. elative to point 3. It is overcoming of the cohesion in powders (and this is particularly true about cohesive mixing.
by capillary interaction resulting from liquid bridges between drug and carrier (wet granulation). that is. situation 3 occurs as described in the foregoing. Some comments on it are in order before describing its origin and usefulness. the small particle may attach itself in a c h rather permanent fashion to the larger particle. and d2 that of the large particle. 20. and by solid bridging (Rumpf.29)
.2) and (17. ordered mixing or interactive mixtures was first introduced as a concept sey (1975).3)].8d:d.16. ecause. proportionality between cohesive force and particle diameter. Cohesion becomes more important as the particles become smaller [see Eq. 1982).2)
where dl is the diameter of the small particle. as has been mentioned in Chap. depending on the situation.3) are repeated here for convenience. the force on the the smaller particle is (20. but when the smaller particle becomes ~ ~ smaller than the large particle. Zimon. Reference is made to Fig. The cohesive force exerted upon the small particle is
C = . Equations (17. and this means slower mixingand. so that the last part of the lending should be carried out with the OH.Principle of a highshear blender.the energy input is high. There is.
will allow the now separated particles to mix. 1961. commonly. less complete mixing. The previous sections of the chapter have been concerned with “noncohesive” mixing. This gives riseto a desirable situation denoted ordered mixing.it will give riseto a high residual rsd. It is the latter that is. with the un~erstanding thatnoncohesiveness implies that the cohesion is small. The way in which drug adheres to a carrier is by electrical forces. referred to as ordered mixing. Whenthe particles of one component are small compared with the other. (17. whenthe choppers are on. Krupp. 1967.2) and (17. 17. The force is inversely proportiona~ to central distance squared.
(17. or interactive mixing. chop~ers
cally. and the force between particles is inversely proportional to the distance between their centers. What is needed is that the blending situation is such that all the particles will pass by the choppers.
unless a stress larger than that shown in Eq.31) is exerted on them. and this has been a reason for unreasonably high standard deviation values for certain powders of this nature and for such powders other thieves (e. care being taken to take the samples in the same spots that the spoon samples were taken. Sampling by spoon is nowdone in three spots in the halffull drum. then onequarter of the drum may be transferred to another drum (Fig.30)
< in cases where dl < d2.g.. then to the second layer. Samples are now taken at three designated spots on the surface. One such is the use of a sideentry thief for sampling such powders. The samples taken by spoon are nonperturbed. the stress is larger. so layers are notdisturbed. Thieves should always be vaZjdated before deemed appropriate for Sampling of a certain powder. The force works on a surface area equal to the cross section of the smaller particle. the small particles may attach themselves to the larger particles and remain there. and this is repeated in the subsequent threequarters~full drum. The thief to be tested is nowinserted into the powder in the appropriate places: to the bottom layer first. and then the third layer. The drum is then filled up. if the thief does not perturb the mixture during the sampling. This stress may occur in certain situations. or at best the standard deviations obtained should be the same. and carefully. so that the stress is
= ~ ’ d ~ d 2 / ( ~Bd~~d) ~ / ( ~ d 2 ~ =
(20. The results should be identical.4) are appropriate. The other situation is when certain “rugorous” excipients are used. because the spoon does not (or only very minimally) disturb the powder mass. using a large scoop. and transfer repeated with another onequarter of the content of the first drum. the plug type thief shown in Fig.
. York has reported on the adverse effect of pregelatini~edstarch on ordered mixes on spraydried lactose. (20. This can be done fairly easily.
so that the total force on the smaller particle is
&tal
= B dl d2 /(dl 3. so that the second is now halffull.3 1)
that is. The transfers must be carried out. 20. the smaller the small particle. When particles used are in the micron region. 20.17). If a small batch is made. and the large particles are fairly large.Formation of an ordered mix.d2>2a BJ&2 :
J 3 3
(20. or if a drum is collectedfrom a larger batch and barrelrolled to some degree ofuniformity.
in Table 20. 20. lactose.Sample by
Spoon
7 ~alidationofathief.16. and the surface of FloFast has. As depicted in Fig.5 m2/g to make satisfactory ordered mixes with spraydried lactose
Ordered Mixes and Surface Areas Surface area Excipient Dipac (Sucrose) Emdex Crystalline lactose available Surface area available
(> 2 elm)
(> 10 elm)
817 1985 4848
. Ni (198 1) showed glyberide would have be fineenough to have a surface that to area of 3. by comparison with other direct compression ingredients.whereas microcrystalline cellulose has a smoother surface and. and are shown. under scanning electron microscopy. fructose. or mannitol). by recrystallization under high pressure. and most efficient excipient for making ordered mixes is a grade of spraydried lactose known as FloFast.2 he most common. hence. The capability of “holding” fine (micronized)material will depend on the surface area that is accessible to the adherent fines. Granular dicalcium phosphate also possesses this attribute to somedegree. exactlysuch a moonlike appearance.. lends itself less to order mixes. has described techniques for making dendritic spherulites of sugars (e.
taniforth (1980).g. it is desirable that the larger particles have crevices into which the fines may fit.
then the repose angle becomes identical with that of spraydried lactose. it is to be expected that the content uniformity would be bette an in noncohesive blending. once the blend has been established. Supabhol and Stewart (1996) have shown interactive mixtures of ~ i c r ~ n i z ~ diazepam with the following direct compression ingredients: compactrol.9 42. for diazepam.3.g. and granules made by starchlactose~ovidone. 1990).9 40.or 18pm diameter. essentially. t is seen that when the drug is added to crystalline lactose.. and sodium starch glycolate that had been passed through an 80mesh screen.
Repose angle (") 41. 1979. Kata's data are shown in Table 20. spects of ordered mixing impinge on content uniformity of drug products. hang and Johnson (1997) prepared blends that contained 0. implying that the drug is imbedded in crevices in the surface of the excipient. when sampled in the official manner. the repose angle that there is no difference between this composition and that of pure drug. The strength of the interaction was established by using compression in a Wood's disk and rotation between 25 and 2000 RP . and there are reports literature on the effect of particle size on blend uniformity (e. and the drug substance was either of a 6. with prudent processing. The carrier was a mixture of microcrystalline cellulose. coarser powder. had an assay of 88130% the coarser powder was 97102%. dibasic calcium phosphate.1 39.Estimates from Repose Angles of Quality of Excipients to Form Ordered Composition
< Trimet~o%in 4 pm Crystalline lactose < 160 pin Spraydried lactose (SD) Crystalline lactose + trimethozin Spraydried lactose + trirnethozin
Source: Kata. and in rigorous blend occur.01 mg of drug per of carrier. In ordered blending.
C dz
= =
cohesional force on a smaller particle diameter of a larger particle
di = diameter of a smaller particle
. dicalcium phosphate (emcompress).and.5 48.0
Com~ent
about about
= =
a of trimethozin a of SD lactose
ata (1979) estimated the quality of excipients to form ordered mixes by determining repose angles a of their mixes. shows the effect of particle size on the force between the smaller and larger particles. ut when the drug is added to spraydried lactose. however. Yalkowslcy and lton. This. Yalkowslcy an lton (1990) point out that when it is simply keep the small particle o mechanical (van der Waal) force then attrition may affect uniformity. the content uniformit for such situations. 1and 15% concentrations.
In: KnepperWA. S. p 15. New York. Philadelphia. Faiman MD. Stewart RJ (1996). pp 93144
Lantz RJ. Marcel Dekker. patent 4. Stewart PJ (1993). Soebagyo SS. pp 32 1. Rumpf II (1961). U. pp 379414. Faiman MD (1964a).349. Adv Colloidal Interface Sci 1:11. Krupp H (1967). ~upabholR. patent 4. Powder Techno1 17:273. . PhD dissertaion. S.
. Staniforth JN (1980).. New York. eds. ~harmaceutical JB Dosage Forms. TheStrength of Granules and Agglomerates. (1979). Rippie EG. Adhesioin of Dust and Powder. x . Acta Pharm Technnol 25203. vol 2. University of isc cons in. U. pp 152.. N . Ni P F (1981).. New York. n) = probability of taking a sample of n drug particles from a totalof N particle from a sample containing a fraction x of drug.542 (issued 1982). Int J Pharrn 25227. Zimon AD (1982). S c h ~ a r t z (1981).163. = fraction of drug content in a blend or a population a = repose angle B' = constant relating particle size to cohesional force = sum of squares
Carstensen JT.916. Stewart PJ (1985). Rippie EG. J Pharm Sei 53: 1360. Soebagyo SS. ed. the number of ways of removing y1 drug particles from a total of N particles when sequence of selection is immaterial) n = number of particles in a sample r(x.f.~2 = residual variance of a blend caused by energy perturbation t = time x = fraction of drug . J Pharm Sci 56: 1523. when sequence of selection is immaterial s = standard deviation of a noncohesive blend at time t so = standard deviation of a powder before blending .e. Stewart PJ (1990). Int J Pharm 91:227.s = standard deviation of a noncohesive blend after infinite time ~. Soebagyo SS. J Pharm Sci 53: 147. 2nd ed..Interscience. The Theory and Practice of Industrial P h a ~ a e y Lea & Febiger... Patel MR (1977). Madison WI. Lachman L. Patel MR (1978). ~gglomeration. Rippie EG (1964). J Pharm Pharmacol48:1249. Int J Pharm 66:263. Kanig JL. In Lachman L. Consultants Bureau. In: Lieberman HA. Pramoda MK (1967b). eds. Olsen JL. Olsen JL.. Rippie E (1986). = force between two particles of unequal size k = blending rate constant ks = segregation constant N = number of samples taken in a powder before blend in^ { N E )= N combinatorial n (i. Lieberman HA.
7. in that it is more rapid from a large surface area. for instance. Earlier chapters have shown. and hydrophobic drugs will not wet down easily.9.10. in
. The larger the surface. Granule ~easurements Properties ~ndpoints 21. E~uipment 21.4.~ r a ~ ~ z a t i ~ g for powders further solids process in^. which essentially consists of “gluing” particles together. however. llets eterminatio~ (Sieve Test) issolution from ~etProcessed Granules Symbols Recommended Reading eferences
366 367 368 370 370 37 1
21.12. 21.2. ~xtragranular
354 354 356 358
358 359
362 364
21. ~pheronization
372
article sizes are often too small to allow good processing into tablets or capsules.5.6.
21. These are two reasons for ~ e t .11.21.3. One more reason is that the process. that flow can be affected. 21. The first step in dissolution.8. Granule Density and Porosity Porosity 21. also aids in forming bonds in tablets. ~ ~ a ~ u ~ a t i ~ n 21.1. iss solution of dosage forms is also affected in a positive way. the more dif~cult (the slower) will the wetting be. aterials and ~ e t h o d s hysics of the Process and 21. is wetting the surface. 21.
r a suitable length of time. the granulating agent is referred to as a ~ i ~ ~ e ~ . y be carried out in rotary processors by wet granulation. so that mixing can be achieved at high intensity in a small area of the mixer.
The principle of wet granulation is to add the binder by some means so that it will form bridges between the particles to form granules (Fig. As seen in Fig. that in wet granulation. are referred to texts such as those listed under ~ e c ~ ~ ~ ~ e e ~ n ~ ~ i eend ~ at the n ~ of this chapter.1).1 the general method is that he binder is dissolved in the water (or solvent).
. that this text does not e~phasize the actual machinery involved in solids manufacturing. e wet granules are dried by various means. ost often. and thatthe reader interested in the intricaciesof equipment used. et ~ranulation be carried out in kneaders. these can will include both an impeller and a chopper. the wet granule has formed.
Solid
Wet Granule Granulating Fluld
Overwet
Stages in granule formation. for instance in the introduction. 21.which case. The powders are mixed. for instance. ~ncreasing wettability 3. there main purposes in wet granulating: article enlargement 2. are three ence.
2. t granulation can also be accomplished by (a) extrusion or (b) pelletizing. nowadays.
solution of binder (the granulatin~ solution) is added. Adding binder to the particulate solid
It has been noted. and the “feeding” of all the material into and about the chopper can be assured by the impeller. fluid bed drying being the most common. 21.
after it is dried. This will be discussed further later. is the method of choice. Some c o ~ m o n binders are Acacia Ethyl cellulose (EC) ethylcellulose (MC) ydroxypropyl methylcellulose ( ydroxypropyl cellulose ( ~ P C )
Add binder
Solution
Dried Granules
Flow sheet for granule production. namely: 1. Adding somewhat too much water will cause an overwet granule (see Fig. Because. then the c~~ar liquid fillsup most of the void space betweenthe particles. The length of time the mixture is mixed he general method for doing this is shown in the flowsheet in Fig. If more water is added. or synthetic. Chowhan and Palagyi (19’78) and Chowhan (1980).2. a porous structure is desired. after drying. The granulating fluid is added to the powder. There are several variables that must be established during the formulation. natural.Alternatively. when tableted.3. pel1eti~ation (to covered shortly). Too many fines will also affect the flow rates and uniformity of the ensuing tablet. is added. 21. the drying temperature is of importance. but in such cases. and the structure is denoted a ~ r o ~ Z e t . as is the final moisture content. 21. A moisture content that is too low willcause the granules to become so brittle that they willbreak before the actual compression step. will become very hard. the granule. 21. the binder may be mixed withthe powder. or on agents. The amount of binder 2. the granulation is overwet. or water (whichever is applicable). At times this is desirable. “cappers” may form. 1958). which eventually forms what is denoted “wet granules” in Fig. may be used as binders. This corresponds to what is usually denoted a w e t ~ ~ ~ istructure (Newitt and Conway. for instance demonstrates the use of hydroxypropyl methylcelluPMC) as a granulating agent (binder). and solvent.
. be ny polymers. modified.1). and this. and this is denoted a wet c a ~ i Z Zstructure. For traditional granules. so that. usually in an amorphous state. in the formation of granules. When all the void space (and even more) is occupiedby granulat~r~ ing liquid. The volume of granulating liquid 3. is bound together by binder. and the granule should break or distort during compression. during the process.
Durrani et al. and granulating it to a total HPMC concentration of ak and Chalabala (1975). Krycer et al. ~ P has~ been C for wet granulation by several investigators (Nagy et al.. gelatin and pregelatinized starch.. (1983) made tablets of HPMC by dissolving it to 8% in water.. a total concentration of 0. however. in a ratio of 1:lO) to form a gel. used 1980.3 Various types of granules created by the type of granulation and the nulat la ti on
time. Table 21.e. The process for producing a wet mass is as shown in Fig.
olyvinylpyrrolidone (PVP) Starch TO this list it is possible to add. ethyl cellulose) are only solvent soluble. starch is only water soluble. Chowhan and Chow (1981) and Chowhan and Palagyi (1978) have wet granulated naproxen with HPMC. Chowhan (1980) has used HPMC in wet granulation of salicylic acid tablets. Shotton and Edwards (1974) used 4% methylcellulose as a binder in wet granulation of sulfadiazine tablets. the synthetic and cellulose derivative binders have been preferred. used 1000 g of met~ylcellulose(0. it is possibleto test the breaking strength of films made from it. for instance. Some of the binders mentioned (e.
To assess the “strength” of the particular granulating substance.g. 21. The table illustrates that the breaking strength is a function of moisture content. 1980). Methocel can be used in small percentages in wet granulation). for instance. Starch has to be added to boiling water (e. some both water and solvent soluble.. (1997) have described the use of and Carbopol934P for wet granulation.I illustrates this for a series of traditional binders. Gelatin bridges break more readily at higher moisture contents.1. Gudsoorkar and Khanna.g.17%) and ten times as much in a second setup (i. In recent years.5% in solution) in a total of 3000 (dry) g of (wet) granulation (Le...
.
2)
Granule “strength” may be tested in different manners (e.4 1..Brea~ing Strength of Various Binders inder Moisture in film
(%I
Gum arabic Gelatin ~ethylhydro~yethyl cellulose PVP (povidone) Starch
Source: Healey et al. and the work of adhesion.o 18
9. ~ a l a m a t a r i s and and Kortis (1997) tested the tensile strength of the granulations as a function of water consumption by methods published by Ashton et al. 1974.1
alamataris and Kortis (1997) studied granulations of lactose (subscript S in the following) and sulfadiazine (subscript L in the following) and employed a specially built rheometer to measure viscosity of the masses studied.v.4. and a con solid at in^ stress applied.cos[@]}
and the spreading coefficients.5 3.. (Rata from M~lamataris and Kortsis. Profiles co~structed from their data are shown in Fig. and the force required to sever it. The tensile strength at the particular packing fraction was obtained from the weight.4 12 7. Samples of gran~lation were placed in a cell. hLs given by
h = W@ 2y 
(21 1)
( I
(21. = y{ 1 1.)
.4
8. Wa which i s given by
r.2 3. namely. (1964) and Eaves and Jones (1972). 1997.8 13.
Breaking strength (J/cm2) 1. 1968). the crosssection. the surface tension of the liquids used. y representing either yL or ys.
9
d
20 0
I
I
I
10
30
I 40
Liquid Addition
(Percent)
Tensile strength of granulation as a function of water consumption.8 10. by the method published by ~ a r w o o d Pilpel.g. 21. the contact angle 0 between granulating liquid and powder. They measured wettability parameters.1
10.
It is the mass itself that causes the torque on the blades. As granulating liquid (or simply water) is added. 1987. Ritala and Virtanen. Hydro~ypropylcellulosewasused as binder. 1995). 21. hydroxyethyl cellulose. 1990 ancock et al. most often. 1971).
he mixer torque rheometer has been used to study wet granu~ations a series of by vestigators (Rowe and Sadeghnejad. there is a certain.5. The strength is a function of the “liquid saturation” (i tage of void space in the wet granules that is occupied by ganulation liquid).not an equilibrium process. Granule strength is also an (almost linear) function of granule size (Gold et al. 1973. (1988) and itala and Virtanen (1991). increases with moisture content (Ganderton and
When powder is placed inmixer and blender.. The critical liquid percentage depended on the system in question. Lindberger et al. The hydroxypropyl methylcellulose system was capable of extrusionspheronizing. 1989. 21. 1997.5 mL of water. Chatlapalali and have described torque versus time curves of diltiazem HCl~ellulose wet granulations. especiallywhensolubleexcipients (either drug or filler) are used. but not substantial.. 1984). Bier et al. 1984). Ritala et al. This is a point that is of importance in “granulation endpoints. However. lactose).. ~ranulationis. Parker et al.” to be discussed shortly. 1982. and the resistance is largely independe ris and Kortis. because the longer the granulation goes on. In other words. Usteri and Leuengberger(1989).g. using hydroxypropyl methylcellulose. and Wan and Prasad (1988)have also described the instrumentation of torque during the granulation process. poor liquid spreading.1992. in general. 1988.. It has been shown. substantial amounts of lactose may go into solution if the process is allowed to go on for too long (see Figs.. Lindberg et al. the liquid saturation is a function of granulation time. itala et al. (1979) originated the first systematic study of power consumption measurements in a granulation kneader to establish a ranulation endpoint. and microcrystalline cellulose as excipients. 1988. a resistance by the powder mass to the process. and granulations were carried out with isopropanol.1 and 21. experimentally (Capes. the more lactose will be dissolved. Landin et al. that when the crust is removed the granule attains its maximum tensilestrength at about20Y0 liquid saturation. 1991).The tensile strength of granules increase with increasing dryin (Capes.. and its solubility is 1 g in 2. A typical curve is shown in Fig. 1956). Wan and Prasad.. The forces that must be overcome are aals and frictional. and must h a l t e ~ be before reach
. The timeof kneading is also important.3). (1982). In all cases there were wetting. Although proposed earlier. The work exerted in crushing granules. Because lactose is often used.. and weak interaction between substrate and binder. this increases. HunterandGanderton. 1991. and it increases exponentially with the amount of binder content in the dry granules (Strickland et al. in the processing of granulations with soluble fillers (e. Usteri and Leuenberger.
the properties of density and porosity are interrelated. A schematic represen~ation pores in a granule is shown in Fig. r i io~. The capillary pressure. will allow signalsthat change as a function of kneading time. they will be treated compositely.
In granules. hence. Macropore
Ink
Bottle Pore
I
S~h~~atic representation of pores in a granule.. In recent years. P.
.6. After a certain time an “equilibrium granule” (which truly is not an “infinite time granule” will occur (~arstensen et al. of a liquid with interfactial tension y and contact angle [e] in a capillary of radius r is
P = 2zry cos[6~/(2~r2)2y cos[O]/r =
(21. 1976a. monitori~~g the torque on by of the mixing motor (or placing load cells on the mixing vessel).s5
“0
20 40 60 Liquid Addition (mL)
80
Typical torque versus liquid addition curve for a granulatio~ process. so that kneading first allows distribution of the water. the granulation endpoint was empirical. 21.3)
Micropore
.
ing an e ~ ~ i l is i~t ~ ~ t~For many years.b). ~ and it was d e t e ~ i n e d the operator. of size distributions may be deduced from hysteresis loops in adsorption isotherms. What happens in general is that the initial addition of liquid is too localized.
Above thisupper limit the pore the is essentiallypart of the surface rugosity. ~.6)
(21. inserted in Eq.) l
(Data
. then its volume V relates to its area A by
E
= 1 .where y is the interfacial tensionbetween adsorbate and substrate. if the true. The reason for this is the following: First. ~ercury intrusion porosimetry is usually usedfor measuring pore size ~ ~ s t r i ~ t i o ~This is applicable to pure solids as well as to granulations (Fig. V is its molar volume.048 pm denoted are micropores. (21.(P’/P)
V I A = h ~ ~ ~ / = r/2 r h ~ 2
(21.7). and Po is the bulk vapor pressure of the adsorbate. and 8 is the contact angle. The surface area can be calculated from the pore ~istribution graphically by integrating the penetration volume against the intrusion pressure. from ~ a r s h a l and Sixsmith.4) (2 where P is the vapor pressure of the adsorbate over the pore.04pm are mesopores. €974/75. can crystallographic density of the solid p is known. by measuring the apparent density p’ of the particle. 21.7) gives (21. Pores with radii of 0. assume that the pores are cylinders (the socalled bunch of cylinders model).5) ores with diameter above 8 pm are referred to as megapores (and the upper radius for those is usually in range of 2040 pm). R is the gas constant. T is absolute temperature. The total porosity E (regardless of distri~ution) be calculated.If a length h of cylinder is longcompared with its radius r.7 Approximate pore size distribution of microcrystalline cellulose (Avicei).7)
The term P V now becomes PV =Ph/2 Since (21
P = 2ypcos[B]/r
3)
this.9) = ypA cos[8] PV
. by a previously stated formula: (21. If such a liquidiscondensed into a capillary pore with radius r then the Kelvin equation states that ln[P/Po] = 2yV/rRT 1. and pores smaller than 0.
The rate constant obtained from this is an index of the granule hardness. Granule friability is often measured. (1978). 1991) applies (here repeated for convenience): P = 2y~0~[6]/r (21..e. in that as the work required to create an area of dA is PdI' = ypc0~[6]dA (21.10) or. in the sense that the amount left on the original retaining screen size decreases loglinearly in time.11) ypcos[6]A The relation may be deduced directly well. The rate at which fines wereproduced is app~oximately firstorder.
The particle density (i. then the ashb burn equation (Lowell and Shields. integrated from rl to
r2
Area under PY diagram = (21.48 N/m). the evacuated solid is exposed to a surrounding of mercury and a pressure P is applied. The bed density of populations of granules is.1 1).urn. Mercury porosimetry can be used to determine the distribution of pore sizes within a granule. as well.For microcrystalline cellulose this gives values of the order of 1. which was rotated for different lengths of time.12)
and integration of this.13).This is along the same lines as the findings of Carstensen and (1985). Baba and ~ u g i ~ o (1965) and to arsh (1961) have described methods for measuring this characteristic. dried granules of a certain mesh cut were in a ball mill. and found the granule density to include the pores that are closed (inkbottle pores) and open pores that have a radius less than 8 .urn. the intragranular porosity) is also obtained by mercury intrusion porosimetry. with radii larger than the value of r calculated from Eq. andby 6 the contact angle (usually 135140")of mercury with the solid. for any type ofpore shape. leads to Eq. directly. In the latter.or (21. (21.0m2/g. the interfacial tension between mercury and the solid (usually 0. Granule hardness has been measured by Harwood and Pilpel (1968) and by et al. Helium pycnometry (or wet pycnometry at times) can be used to determined the density of granules pg which includes inkbottle pores and pore space with pore radius of less than 8 . (21. ~ertommen al. Denoting by y.13)
The mercury porosimeter measures the total volume I' intruded at pressure. as discussed previously.
. In the latter case. and this volume represents the pore volume.(1998) haved e t e ~ i n e d ~ r ~ n density of pelletsmade by et the ~Ze spheronization (to be covered in the following).
8). Fair and h (1933)showed that the coefficientof variation of a diefill increased linearly the shape coefficient (i. 1971. that the packing of a powder (Le.” The extragranular porosity is affectedby the rate of addition of the granulation li~uid (Fig. and speed ofthe punch. but the larger this volume is.. These are subjects that will be discussed It is obvious from previous chapters. the longer the stroke of the punches will be. there is a certain (machine adjusted) volume between the die table and the lower punch at weight position.the extragranular porosity (or the dependent bulk density) is of importance. because one of the intents of wet ~ranulation indeed.67 and the logarithm of the porosity minus this figure is plotted versus rate of addition. Ridgway and Rupp (1969) and e t ~ ~ Carstensen (1990) has been quoted previously.g. In linewith the statements on the die fill..7102e2x
t
RA2 = 0. so that the number of bonds that are created would be the larger. This fact may be more applicable to direct compression (where there is no significant operational control of particle shape) than to wet granulates. This may varyfrom batch to batch. the higher the weight variation might be expected in a tableting operation). the powder consolidates in the die as the punches come down on it. the lower the porosity. The porosity approaches a limiting value of 0. /The work )by . 21.Furthermore.)
. in general. because to obtain a certain amount of powder in a tablet die.. the more “irregular” a granule shape. is greater the than that of the consolidation rate. particle “round.In highspeed machines.e.2. of the s ~ c ~ e ~ ~ c i~ ~ ~. to make the is. (Data from Davis and Gloor. The raw data for this graph are taken from the publica~ionby avis and Gloor (1971). and one of their findin the bulk density decreases linearly with increase in the shape coefficient.63066 . its bed or extragranular po~osity)is a function of the shape factors (e.931
t
I
J
120 100
80
Rate of Water Addition (glmin)
f 40
160
Effect of rate of water addition on the extragranular porosity of a wet granulation. Their data show that the porosity approaches a limiting
~
y =
0.
hence. 1980) that the moisture ried” granule is not.67. unless the granule is ause the surface moisture is a sort of lubricant.9. at acts as a lubricant. but that there is less on the surface than in the core. necessarily evenly ~istributed (Pitkin and Carstensen.
20
30
40
50
60
70
80
Temperature of Inlet Air
Effect of te~peratureon extragranular porosity. there willbe a closer packing with more surface moisture (the lower the drying temperature) and this in turn will give a smaller extragranular porosity. This is also demonstrated by the work of Armstrong and March (1976). 21.and that temperatures above 70°C are usually not used. and in the presentation here the logarithm of the porosity minus this figure is plotted versus rate of addition.)
. the smaller moisture dependency at higher diameters. That the specific surface area of the larger granule is smaller is the important factor. 21. then a linear plot ensues (Fig. 1976. The temperature of drying is also of importance.10.. If these coefficients are plotted versus diameter.9. in line with the trend shown in Fig. the larger the diameter of the particle. There is less sensitivity to moisture content. as seen in Fig. 1973).75). 21. 21. and this is shown in Fig. 1967) the more readily the powder will attain a closer configuration. and a parameter describing their “flatness” would be the coefficient to x2.10 are fairly well described by parabolas. because there is less “surface moisture” in the larger granules. and as seen arameter has an effect on the extragranular porosity. The frictional coefficient would be a function of surface moisture and the total friction would be proportional to this and inversely proportional to the area. other factors being equal (Fig. Hence. The curves inFig. 21.value of 0.10 simply gives gross. (Data from Davis and Gloor. may be assumed (as shown by Zoglio et al. 21. 0.
1971. overall moisture content of the granule). Often. the obvious that the porosity will approach a limiting value (in figure. the more the lubricant (Neuman. “External” water.11). manufacturing batch sheets do not include addition rates.
and the air velocity is then adjusted so that the particles become "airborne" (i.45
0. (Data from Armstrong and March. 21. The principle.4400e2x
RA2 = 0.0.2.40
10
20
30
40
of moisture
50
96 weightpervolume
~xtragranu~ar porosity as a function of moisture content of granules.
a )
E
x
2
iJ
20
40
60
80
100 120 140 1 6 0
Average Diameter (microns)
Flatness of distribution as a function of particle diameter.)
It is not the purpose in this text to delve on the intricacies of fluid bed drying. The powder to be granulated in is transferred to a basket with a mesh bottom. (Data from Armstrong and March. This is placed in the fluid bed dryer in such a fashion that air can be let in at the bottom of it.e.12. 1976. fluidized).. must be kept between the incipient ~ ~ i ~ i ~velocity o n the velocity that just
y = 3.)
. The air velocities ~ t i (Le..8595 . 1976. however.993
k
c . is briefly schematized Fig.
fluidized the powder).Fluid In
Schematic of Auid bed granulator. dnldt = bn (21.) The formation of the aggregates occurs by collision of either particles or aggregates with li~uid. 1977). gives rise to granules that have greater plastic deformability and less granule hardness. then the mass will simply wet down. and this is de~onstrated the following manner (Mehta et ai. The mixer granulations were harder than the fluid bed granulations. and with 70% drug (ethenzamide) have longer dissolution and disintegration times.. and if it is too slow. Arnaud et al. A delicate balance between spraying rate and air velocity and temperature must be maintained.. is proportional to y1. in comparison et with agitation granulation. he particle size distributions from fluidbedgranulated material are mostly lognormal. so that
. (For instance rz must be smaller than. where 1 < n < N . Dropsattach themselves to the powder and agglomerate it. and the ey1t~~iy1~ey1t veZocity(Le. If the spraying rate is too fast. (1999) have shown that fluidized bed granulation. The here. that is. or equal to. The granules are compact. and (d) roller compaction. The inlet fluid may be either a binder solution or water. the velocity that would simply carry the powder out the exit tube). ~ranulation liquid is then pumped into and sprayed onto the granulation. a granulating agent (PVP or pregelatinized starch) will bepart of the powder and will act as a binding agent when the water is added. b is a constant that depends on the collision and detach~ent mass ~. (1998) studied nitrofurantoin (20%) ~ranulations with 38% lactose and 37% cornstarch and compared four granulation procedures: (a) wet granulation in a Lodige mixer. The binders used by these authors were (a) lactosecornstarch and (b) HPC. the number of particles placedin the granulator.and the airstream dries it. etgranulated granules were harder than those made by dry process. In the latter. The manner in which a wet binder is incorporated into the massing is a most important factor. then the droplets may dry before reaching the solid particles.15) probabilities. ~ u n a d a al. in The granulation process is assumed to produce y1 particles. where N is an upper limit. (c) dry granulation. there is a proband ability of this happening that is proportional tothe number of particles or aggregates y1 present at time t. (b) granulation in a fluid bed granulator (Glatt).
There are several systems for pelletizing wet masses. this is relatively small compared with that exerted in pelletizing. Air is. 21. Shaping is often accomplished in a balling disk (see Fig.19)
] but. . 1968. 1996.. olm et al. 1998).z.
Wet Mass In
Cut Strings In
I
Perforated Plate
(a)
(b)
(a) Screw extruder. 1996.17)
where M is the average aggregate mass.
lthough some stress is applied in ordinary wet granulation.(21. (21. the most common being the screw extruder.1995. so that the wet strings (akin to spaghetti) are quite compact (non~orous).13b) and is often to as spheronization.n et al. A reduced time0 is now introdL~ced:
0“th
(21. 1997. (b) balling disk.16) where the proportionality constant is expressed as 114 for later notational convenience.z. The time ofgrowth t is different for each particular aggregate ( 1968) and is normally distributed about the mean growth time. 2 1. llets made by wet granulation followedby extrusion spheronization have cribed (Zimm et al. since 0 is normally distributed it follows that [ M/ Ma v gis normally distributed.. ~ntegration Eq. expelled.
. Johanssonet al. obviously..16) gives of
t =h
+ aj l n ~ ~ / ~ ~ v g ]
(21.
0 =4ln~”3vgl
(21. Vertrommen and Kinget. 1994.13a. This is akin to a meat grinder. 0 will be distributed normally with a mean of t .. Zimm et al.18)
Since t is distributed normally with a mean of z. a schematic of which is shown in Fig. and sufficient pressure is usually exerted so that some elastic and plastic deformation of the solid particles result. used microcrystalline cellulose and (10%) acetaminophen for their composition.
and air pockets form in the pellets.21)
It was not possible.Agglomeration liquids may be. ores are closed as the spheronization processes.. the specific surface area (by gas adsorption). and the pelletization procedure may be spheronization or extrusionspheronization. ~ertommen al. 2) after which the loglinear relation referred to in the
. each giving high statistical probability of fit. Carstensen. The sieving of particulate pha~aceuticalshas been reported (Fonner et al. 30/40 mesh) is selected. 1998).. butthat a tra~§ition occurred (to region no..for instance (as exemplified by Johannson et al. (1996) compared dissolution rates from spherical pellets by two odels described in the following. article size distributions may benormal or lognormal (Steiner et al. Carstensen (1977) found that the percensing through the sievewas a linear function of the logarithm of sieving avies (1990) found that the logarithm of the amount passing was linear in the lo~arithmoftime at lowtime points (region no.. ethanol/water 70 : 30. and resieved. The nomenclature used is: a = the radius of the pellet not extracted a.057 pm). 1974) whenthe granules are made in a kneader or. The mercury intrusion established the presence of macropores (0. If the fraction in one particular sieve fraction (e. Usually. imm et al. to distinguish between the two models.. (1998) have studied the structure of such products by deteret mining the true density by helium pycnometry. and appearance by scanning electron microscopy. statistic all^. = the radius of the pellet Q = mass of drug released per square centimeter (cm2) of pellet surface D = diffusion coefficient A I = grams of drug per cubic centimeter (cm3) of pellet S = solubility (g/cm3) t = time E = porosity (dimensionless) z = tortuosity The two models compared were: iguchi squareroot model. l977). tor speed and long spheronization times will reduce the pore volume. which is given by
= [ ( ~ & / ~ ) (A ~ 2 &S)St]1’2
(2120)
iguchi cube equation given by 1
+ ( u / u ~ ) ” ~3 ( ~ / ~=o6) D ~ ~ t / ( z A ~ ~ ) ~
(21. tby.g. 1). by fluid bed granulation. pore space (by mercuryintrusion). some of the material will pass through. 1977). asmentioned earlier (Mehta et al. 1958.
The most common method for determining granule sizes are by means of sieve tests. a giventime for a sievetestis and the weight obtained on the various screensis recorded.
which allows more rapid dissolution. M is the mass insidethe granule. The drug substance then forms a saturated solution of the liquid in the pore space. (1998) give rise to the dissolution profiles shown in Fig. and dissolution medium penetrates the pore space of the granule.o
1
2 3 4 In(Time in Seconds)
5
6
7
The percentage of material passing a sieve as a function of sieving time.)
. The et~ranulated products gave better qualities in other respects. The methods by which dissolution occurs from granules in general is dealt with in the following.
1. the better the drug should dissolve from it. (21. The manner in which a drug is released from a granule is the following: The contact angle is usually small (because the binder is hydrophilic).
It is intuitively obvious that the “looser” the granule. 21. One important aspect is the shape of the (two lower) curves. x is distance.15 and show the comparative dissolution rates.foregoing occurred (Fig. In this case the wet granulation imparts hydrophilicity to the composition. The trend is actually the opposite of what most often is encountered. Hence. 1990. The data reported by Arnaud et al. and D is the diffusion coefficient. at least with hydrophobic drugs. and this drops to the concentration in the bulk Cb over a distance of h (assuming there is a stagnant film of this thickness on the surface of the granule). so there is no wetting lag time.14).22) stems from the fact that the mass timegradient is of opposite sign from the concentration distance gradient. ~ s s y Fick’s law (l/A)dM/dt
==:
DdC/dx
(2 1. The concentration in the granule pore space is assumed to be the solubility. t is time. (Data from Davies. 21.22)
where A is the external surface area of the granule. ~quilibrium considered to occur when the tranis sition to region 2 occurs. and the drug ~ ~ ~ out intoethe bulk liquid.C is concentration. The minus sign in Eq. Prescribed times for screening should be made in such a fashion that region 2 has been reached.
F ] as a function of time.27) invoked.26)
which integrates to ln[S . where M o is the initial drug amount. is fraction of drug released. 21.25) into Eq.. (21. the value of hadbeen0. (21.15. then onemayhave plotted the lower curve in Fig.24) where V is the volume of the dissolution liquid.45 .28) results. 21. If this amount is denoted Fm.M .Slugging
0
20
Time
40
(mln.28) takes the form ln[Fm .25)
(21. 1998.C b )
(21.35of the total amount.o} ln[FO.29)
where Fm can be obtained by iteration.C.16 by plotting lnC0.9) some of the material has been “encased” in the granules to such an extent that it is no longer available. the concentration in the bulk liquid is (21. (21.then Eq. is the amount (mass) of drug released and Fo is the m a x i ~ u m ~ ~ u n t dissolvable in the dissolution medium.28)
been where the initial condition that Cb = 0 t = 0 has In[ VI to both sides Eq.)
(21.F ] = ( D / ~ z V } ~
a where I . It is noted that dM/dt = VdCldt Inserting Eqs. in Fig.23) The amountundissolved at time t is A4. where I. At times (as in the two lower curvesin Fig. By adding (21. (21.] = (D/hV}t
+ h(S}
+
(21.22) now gives
S VdCldt = { D / ~ } (.1
80 O0
F
0
ller Compaction. 21. ln(I.
. or knowledge of S (which here would be smaller thanFo/V). that is.If. (Data from Arnaud et al.Fj = { D / ~ V } t ln{Fm)
+
(21. so that amount dissolved is Mo .24) and (21.)
60
80
100
Dissolution curves from differently processed nitrofurantoin granules.
y =
.
. no adjustment is necessary.
In general. and Eq. (21. The porosity of extruded. It is most often used in the form ln[M/Mo] = K. Newton (1990) and Bains et al. the wet mix time. These may be fed into a slanted plate and rotated in such a fashion that they become spherical..29). = the radius of the pellet C = concentration Cb = concentration in bulk solution D = diffusion coefficient h = thickness of stagnant layer k = intrinsic dissolution rate constant (emis) Kg = granule dissolution constant ( m i d ) Mo = initial amount of drug
Al
.997
or
3
0
'
I
I
I
. and the water content are the principal variables leman et al. and a graph based on their data is shown in Fig.9 plotted according to Eq. it appears in strings that areeither cut or fall apart in cylindrical segments. the conditions of the feeding screw in the extruder.}
60
Data from Fig.
A = surface area = g of drug/cm3 of pellet a = the radius of the pellet not extracted a. (1997) studied these. the s ronizing time.3931e2x
RA2 = 0.44052 
3. spheronized granules is much lower than that of granules made by conventional means. the revolutions per minute (rpm) of the spheronizer. (21.
. Of these.17.0. 21. (21.30)
hen a wetted. solid mass isextruded.28) applies directly.t where Kg is the dissolution constant (in reciprocal time units). (1991) have investigated the effect of process variables.
20 40 Time (min. 21.
or h = spreading coefficient p = true. 2."
50
60
70
%Water
7 Yieldof20meshpellets after a spheronizing time of 12min as a function of mixing time and water content. vol. Active drug: theophylline with an nonionic binder. In: Lieberman HA.)
= amount of drug retained in the dosage form
P = (a) vapor pressure of adsorbate over a pore. 1997. Banker CS (1981). New York. eds. pp 185261. Marcel Dekker. vol. pp 245300. eutical Dosage Forms. Lachman L. Fonner DE (1990). (From Hileman et al. or (b) intrusion pressure
= bulk vapor pressure of the adsorbate = mass of drug released per square centimeter (cm2) of pellet surface I . 2. New York. Marcel Dekker. = amount of drug released into the dissolution medium Fm = amount of drug released at infinite time R = the gas constant r = pore radius S = solubility t = time T = absolute temperature V = molar volume Wa= work of adhesion as/av = shape coefficient y = interpdcial tension between adsorbate and substrate 4 = contact angle E = porosity hLs. ~harmaceuticalDosage Forms.. . Lachman I. Rosanske TW. crystallographic density of the solid p' = apparent density = bed density pg = granule density z = tortuosity
Anderson NR. eds.
. In: Lieberman HA.
Landin M. Parker MD. Pharm Dev Technol 2( 1):43. Hou XP. Trans Inst Chew Eng 36:422. Rowe RC (1992). Lai T. Otten L. J Pharm Pharmacol21:3OS. Adv Pharm Sci 2: 181. Annu Rep Shionogi Res Lab 15227. Flickner DW. Marshall IS. Carstensen JT (1985). Powder Technol 34:39. Hancock BC. Int J Pharm 64:207. p 41. Carstensen JT. Bonde M. Parker MD. J Pharm Sci 65:992. Hatch LP (1933). Shields JE. Van Norstrand Reinhold. S. a Nagay A. Zoglio MA. York P (1995). Rupp R (1969). Alderborn G (1998). Powder Technol 42:153. Chow YP (1981). Int J P h a m 38:229. Zoglio MA (1976). J Pharrn Pharrnacol 25S:71P. Drug Dev Ind Pharm 151. Int J Pharm 76:239. Sucker H (1979). ~ e h t A. eds. Rowe RC. Keresztes K. Carstensen JT (19'73). Fair GM. Otten L. and Porosity. Holm P. p 41. Huber HE. Hersey JA (1983). Brossard D. York P. Bier HP. Ritala M. Rains D. Rowe RC. Reenstjerna B (1982). Hou XP (1985). Krycer I. Parker MD. In: Fayed ME. Hungary] 35:168. Int J Pharm 80:179. Rohera BD (1998). Lindberg NO. J Pharm Sci 84:557. Sixsmith D (197411975). Chaumeil JC (1998). Drug Dev Ind Pharrn 16:l. Newltt DM. London. Chalabala M (1975). Newton JM (1991). In: Fayed ME. Parker MD (1991). Banker GS. Johansson B. Wigmore T (1996). Baba M. Int J Pharm 163:35. Chowhan ZT. Chatlapalali R. Int J Pharm 161: 179. Walters V (1974). Rowe RC (1992). (1991). J Pharm Pharmacol 26:41P. Drug Dev Ind Pharm 2457. Pitken C. Boutell LS. Acta Pharm Nord 3:229. York P. J Am Water Works Assoc 25: 155 Fonner DE. J Pharm Sci 62:1215. eds. Neurnann BS (1967). Kedvessy G (1980). Indian Drugs Pharm Ind. Int J Pharm 8:775. Kristensen HG (1988).
. Drug Dev Ind Pharm 14: 1041. Pharm Ind 41:375. Int J Pharm 25:207. New York. Marsh DM (1961). Int J Pharm 83:147. STP Pharma 6:396. ~ p a d r a s h t a Holm P. Hancock BC. Nicklasson F. eds. Chapman & Lowell S. Powder Surface Area Hall.ter
Arnaud P. Hunter BM. Pharm Univ Comenianae 28:35 [ Ridgway K. Van Norstrand Reinhold. Rak J. J Pharm Sci 67:1335. Leuenberger H. J Pharrn Sci 55576. Ritala M. Chowhan ZT (1980). IShanna S (1980). Row RC. Capes (1984). Carstensen JT (1978). Ganderton D (1973). Int J Pharm 72:243. ~ a l a m a t a r i s Kiortis S (1997). Swarbrick J (1966). Newton JM (1990). Gudsoorkar. PityeHbdy. Rowe RC (1991). Davies (1990). J Pharm Sci 69: 1. Carstensen JT (1990). SM. York P. Int J Pharm 69:233. 1. York P. Conway JM (1958). Schaefer T. J Pharm Sci 67:905. Pharm Technol Eur 8:22. Sadeghnejad GR (1987). Rowe RC (1990). J Pharm Sci 57:478. Palagyi L (1978). Chowhan ZT. New York. Hileman GA. Carstensen JT. 3rd ed. Healey JNC. York P. Selmecz B. Int J Pharm 154:9. J Sei Instrum 38:229. Leander L. Pitkin C. Rubinstein MH. J Pharm Sci 70: 1134. Pope DG. Virtanen S (1991). Handbook of Powder Science and Technology. Sugimoto K (1965). Handbook of Powder Science and Technology. Neau SH (1997). Harwood CF. Pilpel N (1968). Bull Pharm Technol Inst Univ Szeged [Szeged.
Edwards NJ (1974). Usteri M. Liew CV (1994). Kinget R (1998). Vertommen J. Sunada H. Pharm Dev Technol 1:37. Int J Phatm 161:225. Kokubo H. Drug Dev Ind Pharm 9:1417. Acta Pharm Technol 35:163. Wan LSC. Ind J Pharm 118213. Vertommen J. Drug Dev Ind Pharm 24225. Carstensen JT (1974). ASTM Spec Tech Pub1 235. J Pharm Pharmacol26: 107P. J Pharm Sci 63:1395. Hasegawa M. Carst~nsen (1983). Sakamoto H.
. Acta Pharm Technol 35:163. Carstensen JT (1976). Kawaguchi T (1998). Leuenberger H (1989). JT Zogiio M. Drug Dev Ind Pharm 23:39. Kinget R (1997). J Pharm Sci 65:1205. O’Connor RE (1996). Koehne G. Wan LSC. Tanino T. Prasad KPP (1988). Zimm KR. Whitby KT (1958).Shotton E. Huber HE. Rombaut P. Schwartz JB. Liew CV (1994). Rombaut P. Zoglio MA. Heng PWS. Tadashi M.Pate1 MR. Steiner 6 . Chan PL. Drug Dev Ind Pharm 20:2551. Wan LSC. Heng PWS. Fujita K.
This Page Intentionally Left Blank
.
. The time lapse between conception of a drug and its introduction into market place is of the importance.or wet granulationbased tablets (if those are possible with the drug substance). and the trials went ahead too rapidly to economically change the dosage form. Four years later. more foolproof than direct cornpression. results from decisions early in product development^ during whic original clinical trials were performed in capsules. at times.1. and even though hardshell capsules may be more expensive and cause other problems not encountered in other soliddosage forms. because of convenience. also.4. when the tablet. the development of the dosage form is easier and.6. 22.
*
.
TheTwoRingHardShellFillingMachine rinciples than the TwoRing Machine inciples of Fills and Volumes
376 379 380 38 1 382 383 384 385 385 386
22. and the tablets were never a success. ~ o m ~ a c t i o n uring Hardshell Filling
issolution and is integration of Hardshell Capsules
22. in some aspects.7.
~elliculation ~ustainedReleaseHardshell Capsules Symbols eferences ecommended Reading
dshell capsules are a dosage form that is resorted to when a drug substance rly compressible in the desireddosage strength and is moisturesensitive.22. the public was used to a capsule. An example of a drug substance that was introduced into the market place as a capsule because the development got too far ahead of itself is chlordiaydrochloride) capsules.
3: 0.
Common shapes and sizes of hardshell capsules.2 shows the basic principle on which the capsule separation operation is based. 22.2~. and two are shown in Fig.37.1 No2 _________. they are placed nst a plate (Fig. After one rotation the hopper is drawn back. it is frequently used in early product development (e. is drawn over the ring.3)where a movable hopper. Vacuum is applied.55 0.30.5. The rings may have one.13mL. 22.
. 22. and in the lower ring correspond in diameter to that of the body of the capsule (see Fig. first. 22. so that all the bodies are now in the bottom ring. and aligned so that the holes line up. to ~ ~ describe the basicprinciplesof ~ the original hardshell e ~ c ~ ~ ~ machine.4b) so that the bodies are forced into the tops. or three circles of holes. e discussing the required powder requirements for hardshell operations.75 0. phase I clinical batches). 22. The hopper has an auger (see Fig. The basics ofthe machine is a set of tworings that fit together.5) which
No.. two.4 No.21.0 No. Although it is not used much in actual production today. The volumes may depend on empty shellmanufacturer.25 0. so that the capsules can be ejected. 2: 0. The ring with the nowfilled bodies of the capsules is placed together with the top ring with the empty tops. 22. The bottom ring isnow transferred to a filling station (Fig. and the two rings are separated (see Fig. 1: 0.15 . placed in a hopper. Figure 22.4a). One is shown in Fig. and other sizes have beenreported in the literature: namely.2d).68. The mostly used shapes and sizes are shown in Fig.3 No.50. The once empty bodies of the capsules are now filled with powder. For larger batches it is too slow and laborintensive to retain the popularity it once had.The commonly usedhardshell capsules are made of gelatin. the pegboard pushes all the way into the rings. 0: 0.4c). 5: 0. 22.g. and the support plate is then removed (see Fig.4. and the powder fills into the body capsules. formed into shapes that allowfillingthemwith particulate matter.__Volume in mL 0. are fed down a socalled raceway witha rectifier bar that aligns the capsules so that they feed into the rings in the position shown in Fig. It is a convenient method for making small batches.1. 22. ~ontaining powder to be filled. A pegboard is placed in align~ent with them (see Fig. Capsules. 24. and all the tops are in the top ring. 22. 4: 0.2a and b).4
No. _ _ _ _ _ _ I _ 
0 3 0.5
" . it is . a) and then pushed in (see Fig. and that have holes which in the upper ring correspond in diameter to that of the top of the capsule. 22.
In the former case.
Principle of filling the empty capsules in the bottom ring. then powder fills in simply by gravity. aside from holding the halves together. and to avoid separation of the halves in shipping. Containing Bodies (Bottom) Tops (Top Ring.
may. By this procedure. if it is not engaged. a thin gelatin film is placed around the separation line between the halves and. as shall be discussed shortly. be engaged. because the auger might crush the sustaining film. For this. there is no internal pressure on the capsule.ib)
Ring Containing Tops
”+
ombined Side ngsView. The latter procedure is often used when sustainedrelease beadlets are filled. then it helps push powder into the shells. If it is engaged. or may not. there is a forced co~solidat~on the powder. this also tamperproofs the product.)
(c)
Top Ring Bottom Ring
Principle of a tworing hardshell machine. of there is none.
After Positioning Ring on Table Move Hopper into Position
After One Rotation Move Hopper Back
U
Capsules in Ring Empty
Capsules in Ring Full
ig. in the latter.
. the capsules may be banded.
.
There will be an effect of flow both in freeflow fill. If the 2 flow rate.Plate
Peg Board
Principle of ejection of capsules in a tworing machine. then in z s.1)
so that the contact time is
5
= a/(Q 2nR)
(22. in augerforced fill.2)
Auger
Holes in Lower Ring Side View
Filling process on a tworing machine. W z grams will flow into the capsule. and the shorter the rotational speed S rotations per second (rps). rate and The “dwell time” z (i.e. V The dwell time z is given by the fact that the linear speed ‘u of the hole under the hopper is
‘u= sz 2zR
(22.. of the die table. the length of time the body in the lower ring is in contact with the powder in the hopper) is the longer the length a of the hopper throat. forced or not. is F gls.
ighspeedfilling machines usuallydepend on a dosator principle. so that the fill weight can be adjusted by way of adjustment of the rps. which isoutlined in Fig. in spite of the b and c features of the hopper construction.6. then the amount that the flows into the die must be D or more. the fill weight variation experienced with multiplering filling is due to the different speeds of the holecircles. forced or not as situation may the be.R differs.
. This is compensated by the protrusion on the hopper. If fill dose is D g. 22. 22. the minimum flow rate.. For precision this should have the nature of a plug. The filling principle is then that the powder is stomped into the dosator by the downward movement of it.53. R(1) being larger than R(2) in Fig.e. 22. or simply gravity (or application of compressed air). ejects the powder. i. w e
W = ~ ( $27dR)la 2
(22. making the distance b in Fig.5 different from the distance c.) The dosator is then moved out of the bed. that is. the foregoing.4)
Note that L) is a function of $ . 2The machine can be run at different speeds.3)
For more than one row. to assure a certain free volume. (In some machines vacuum is employed the void sectionto obtain in the fill.where R is the radius of the circle of rings. in the limit
D = W T = ~ a / ( $ 27dR) 2
(22.
Leveler Blade
~ain
Hopper
1
Constant Level Hopper
Dosator principle. and downward movement of the pin. The empty capsules are separated and the body of the capsule is made to coincide with the dosator.giving different contact times. the distance a differs from inner to middle to outer row (for three rows of rings)and from inner to outerrow [for two rows of rings. The dosator is set at a given level.
the compressibility (and indirectly of the particle size distribution). the fill weight) is a function of the apparent density. (d) a dosator. recovery system powder. to (a) separate the cap andbody. depending on the machine used.The amount filled into the cavity of the dosator (hence. The principle of manufacturing the dosage form is. liesin the term f f ~ ~ r o ~ ~ i ~ because. pellets. Some models have presorters for empty capsules. The number after HK machines signifies sules that can be handled per minute with optimum operation ( e g . The constant level is important. the drug substance is mixed with mg of excipients. This principle applies to other capsulefilling machinery as well. as ~escribed earlier. and the mG2 being the most common. a sampling station for filled capsule.g. and tablets. somewhat inbetween. for flow then is impaired. because (see Fig. a capsuletransport portion. the Zanazi. the model G38 operating at 1000 capsules perminute. to some degree. There are several brands ofhighspeed hardshell caps~lefillingmachines. making a constant level of powder difficult to achieve. X of drug.5)
from which Y can be calculated. and (g) a closing and ejection station. iate If the f f ~ ~ r o ~ r apparent density is p’. 22.
. lactose and magnesium stearate) that is filled into the body of a hardshell capsule of a known volume. affliger Karg (HK) machine. within limits. It consists of (a) a hopper for empty capsules with rectifier. a rectifier. After the body is filled.
To make hard shell capsule products. and the volume of the capsule is V . then (22. It is obvious that the powder must be. The highest output of the Zanazi machines is 150. 2500 will handle 2500 capsules per minute). it is a ~ i ~ t of drug substance and excipient ~ r e (e. and an ejection station. because the stroke of the dosator. (c) a cleaning station. The HK machine has a storage hopper for empty capsules.OOO/h. The EZZ72 model handles powders. a bulk powder hopper (as shown in the foregoing). or even a larger density.6) the powder must flow in a controlled fashion from the main hopper into the constant level hopper. To obtain the correct fill in a capsuledosage form requiring. is selectedso that Y Y the mixture has the “correct” volume. (e) a powder hopper. and a checkfor weigher system. and (c) bring the cap and body section together again. 2 is a continuousmotion machine. On the other hand it cannot be too cohesive. by the leveler blade. The problem. it is made to coincide with the top. and the amount of powder it encounters governs both the fill weight and the precision of fill. however. a closing station. Control of this is accomplished. (b) “fill” the body. and an insertion mechanism causes joining of top and body. a (b) removal station for the capsule top. Flow rate of the powder is important. for example. (f) a top holder station. cohesive because type a of plug has to form. this could be the cascaded or the tapped dens~ty.
will affect the apparent density. 22. and shows that when a capsule is filled. There are several aspects to this.
~
If a solid is compressed below its elastic limit. Kawakati and Ludde (1970/71) have compiled a series
Strain Below the Elastic
n
Reversible Distortion
Release Pressure
Stress from Capsule Wall
(b)
. may be used to adjust it. then machine speed. (b) Linear stressstrain relation below the elastic limit (Hooke’s law).7a). then it will distort a bit. 16. apparent densities of mixtures depend on the state of subdivision of the two (or more) ingredients. which actually helps keeping the capsule halves together. the remedy is to adjust Y . if adjustments are made (e.g.6)
This integrates to
EXf1/(X
+ 1) = CP + Q
(22. if a trial of the filled capsule does not give the correct weight. The consolidating pressure P.. so that it becomes correct).7 demonstrates this.
.If the fillweightis appro~imatelycorrect. Kawakita and Taneya (1967)haveshown that the porosity I responds to pressure P in the following fashion: a&/aP= (22.1
In general.7)
where Q is an integration constant that can be derived from the apparent density at zero consolidation pressure. C ~ m ~ ~ actually resorted to. it will then remain within the elastic limit. to some degree. in all of the different types is c t i ~ ~ of capsule machines. and then return to its original shape once the stress is released (Fig. When a powder is “compressed” gently. Figure 22. as mentioned.7 (a) Shapes of a particle under stresses below the elastic limit. there is a residual stress. so that some measure of ~ ~ c sti ~ ment must be available to allow for fillweight variations that will occur from batch to batch ofraw material used. As has been seen in Chap.
For instance. Several authors (Mehta and Augsburger. preparation mode on release of active substance from hardshell capsules has been described (Whithey et al. 22. once the tablet has disintegrated... as compared with conventional disintegrants. for the data for 10kg pressure. Stewart et al. then: (22. in dosatorprinciple filling machines has been reported by ElShaboury et al. This may be recast as
]1’3
=I
+ Kt.. (1983) have described the effect of variables in formulation and process on release of drugs in capsule f o m . and uhammed et al. so that variables. Samyn and Jung. (1993). that cuberoot relations hold best. and K is the cuberoot dissolution rate constant. 1979).8) here Vois the volume ofthe powder at zero pressure. In fully automated filling machines the powder contents are compressed. 1970. and B2 are constants. an (1998) plot better in the semilogarithmic fashion (Fig. A systematic study of the use of effervescent salt. When their data are plotted as ln[M/Mo] versus time the plots are not nearly as good (as judged by correlation coefficients) as when a cuberoot relation is used (Fig. 22. however.b). 1972. and by Mo the initial amount.9 results.
Newton (1972). . divided by the valueof IC. when the co~pression pressure was increased. Khalil et al. It is not always. In the former case they found im~rovement (decrease in disinte~rationtime and t&).8). Cuberoot relations may be expected when the capsule contents. Newton et al. t. and V’ is the net volume of the particles. These authors studied effervescent salts as disintegrants as well as conventional disintegrants (Explotab. whereas plugs or agglomerates will behave in
. 1980.9) where t is time.10). One that has a bearing on compaction equations in Chaps. 1977. 23 and 24 is the Cooper equation: (22. such as force of compression can affect both dissolution and disintegration. 22.66 min.015894 = 10. Newton et al. presents itself as a particulate powder.1695/0. again. 1969.(Kt)
(22. microcrystalline cellulose). then Fig. Denoting by M the amount of drug notdissolved..of consolidation equations. When these are plotted versus disintegrat~ontimes from conventional disintegration tests. 1982. is lag time.10)
The disintegration time can be deduced from these curves: the intercept minus 1.. 1984) have investigated the use of disintegrants to improve disintegration and dissolution. V is the apparent volume at a consolidation pressure of P. (1971a. the disintegration time would be 0. tozolakis et al. The effect of additives and. 1981.
because moisture must penetrate the aggregates.2343 . In overlubricated powders. followed by (b) dispersion of the powder. followed by (c) dissolution of the drug from distinct drug particles.9
0.8
0.y = 1.1. The capsules depicted have no disintegrant or effervescent salt added. This type of dissolution curve should simply be a cuberoot law. disintegration of the shell may occur.
. but the powder mass will remain intact.
24 22
20
18
i.000 R A 2 = 0.0 0.3607e2x
R A 2 = 1. dissolve the drug. andpenetration of liquid into a (now hydrophobic) plug gives rise to a slower dissolutio~. which then can diffuse out. maybe with a slight lag time. o
20
30
40
50
Time {min)
Dissolution of fenoprofen capsules dosatorcompacted at different forces.
se~ilogarithmic fashion. This type of dissolution is likely to follow a square
In properly f o r ~ u l a t capsules the dissolution steps are (a) rapid dissolutio~ the e~ of shell.6 I.5894e2x y = 1.1.3035 .973 RA2 = 0.992
0
lokg
15 kg 25 kg
1.1695 .:
fri
16
14 12
10 10
30 t(i) From Disintegration
40
Disintegration values from dissolution plotted versus disintegration from conventional disintegration tests.5314e2x y = 1.7
0.1.
1998. and then. and th ested methods for increasing the pH.29471 . owing to their sustainedrel~ase nature. whereas a third (halfchange method) requires exchanging onehalf of the liquid with liquid ofa higher pH (Brossard and ~ouessidjewe. Most common dissolution media are (a) N/10 HCl and (b) water.9 . The most important aspect of these. The common apparatus usedis U. 1997) employ an acid medium at first. Kumar and Pandit.
.
L
Hardshell capsules are often used for coated beads and pellet sustainedrelease dosage forms. the socalled steppeddissolution can be accomplished in different manners (Marty et al. Pharmacopeia (USP) either method I (basket) or I1 (paddle). when introduced into the dissolution vessel.!
+ R"C : 0
"+
RNCR" + HZ0
(22. 1997).y = 0. 4 2 5 6 ~ ~ 2 R"2 = 1. (Data from Tang and Gan. whichhavehighersolubilities at higher pH.g..000 ~
'I
0
10
20
30
40
Time (min)
Dissolution of ciprofloxacin HC1 capsules.t y p e rate f reaction with the a m ~ n o groups o the gelatin: f NH. The is usuallyincreased from that of stomach (pH = 1. 1960. the rate will increase as time (pH) progresses. is their dissolution pattern. Marty et al. which encases the capsule and prevents the drug in it from dissolving. rossard (1976) have used a solid buffer addition. 1967). willform a film (a pellicle or pellicule). values. but some investigators (e. the capsule.to adjust
. Ritchel Orth. Munzel.11)
When this occurs. Increasing the pH as a function of time in this fashion.. These methods are rather impractical. 1990.)
Capsule powdersthatcontainsubstanceswithaldehyde or keto groups may e~perience~ i s s o l ~ t ~ o n decre~seon storage because o a ~ a i ~ l a r d . (1997). at various time periods increasing the pH Forcompounds such as ibuprofen.S.2) to that of intestine Two of the methods used to accomplish this involve a total exchange of medium.
Drug Dev Ind Pharrn 23:987. Newton JM (1983). Baylac C . Rhodes CT (1975). Ribon B.(1988) have suggested a method for which initially onehalf of the final volume is used. ElShaboury MH. p 71. de Fenin V.
a = throat length of hopper c = coefficient in pressure/fill equation D = fill weight K = HixsonCrowell cuberoot constant 1M = mass (weight) of drug not dissolved = initial mass of drug P = pressure on fill during filling R = radius of die table Q = constant in pressure/fill equation t = dissolution time t. Quincy C (1980). J Pharrn Sci 52:1145. Ausburger LL (1984). Botzolakis JE.Int J Pharm 7:327. Higuchi T (1963). and liquid addition of buffers then adjusts the pH upward. Cabr KE. Brazier JL. Aiache JM (1997). J Pharm Pharmacol 35:345. Taneya S (1967). Kawakita K. Plant Kogaku Sha. de Albuquerque M. ElCawad AHA. Munzel K (1960). = lag time Vb = the volume of the powder at zero pressure V = (a) apparent powder volume at a consolidation pressure of P. Gaudy D. J Pharm Sci 69:763. STP Pharma 5:750. Khan KA. Kawakita K. Ludde KH (1970/71). J Pharm Sci 64:166. Jacob M (1989). Arch Pham 293:766. Powder Technology. Pandig JK (1997). Augsburger LL (198 1). Small LE. Puech A. Botzolakis JE. Hashern FM (1993). STP Pharrna 6:728. Muhammed NAH. Kumar DS. Mehta AM.
. Tokyo. (b) volume of capsule V' = net volume of the particles w = linear speed of body of capsule in die W = flow rate (g/s) X = drug content ' = excipient content I E = porosity = constant in CooperEaton equation 82 = constant in CooperEaton equation t = dwell time f2 = rotations per second (rps) of the die table
Bannier A. Int J Pharm 12:341. Marty P. Augsburger LL (1982). Pharrn Ind 55:175. Khalil SA. Drug Dev Ind Pharrn 23:1135. Acta Pharrn Suec 9:563. Ali LM (1972). Sei Technol Pharm 5:353. J P h a m Pharmacol 36:77. Pinteur B. Brossard C. Brossard C (1976). Wouessidjewe I)(1990). Power Technol 4:61.
~
Hostetler V (1986). J Pharm Sci 59:169. Sallans F. J Pharm Pharmacol23:452. Patau P. J Pharm Pharmacol 32: 16'7. Tang V. Newton JM. Rodriguez F. Newton J N . Sablayrolles A. hith hey RJt Nainv~lle CA (1969). Pharm ~ e e ~ 107:485.Rouffiac R (1988). J Fharrn Pharrnacol 29:248. Lea & Febiger. The Theory and Practice of Industrial Pharmacy. 0). Tornblum JFV (1971b). J Pharma Belg 43241. In: Lachman L.
. Bader F (1980). . Liebeman HA.Tornblum JFV (1971a). J P h a Sci 58:1120. JM b l . Newtan J N (1979).~ e w t o n (1972). J Pharm Sci 56:773. J Pharm Pharmacol 3 1:1. eds. Razzo FN (1977). Drug Dev Ind Pharm 24549. J Pharm Pharmacol 23: 156% Ritchel WA. Orth H (1967). Combes 3. Philadelphia. Kanig JL. pp 374394. Can K (1998). Rowley 6. Rowley 6.
7. 23.
rinciples of SinglePunch Tablet Machines ultipleLayer and CompressionCoated Tablets Stress and Strain: The CooperEaton Equation onding in Tablets
388 389 390 39 1 393 394 396 398
Lubrication Energy Considerations in C o ~ ~ r e s s i o n Symbols eferences
400
40 1 404 405
05
As mentioned in the introduction.23.5. 23. 23. methods of instrumentation? and granulation techniques.4. 23.
.2. this text is not geared toward the actual machine and operational details of a solids operation. and only a cursory overview of tablet machines will be given here.8. 23.3. 23. This overview is necessary for the further discussion of properties of solids relative to compression. For further details the reader is referred to the recommended reference texts before the reference list at the end of the chapter for details on tablet machines. 23.1.6.9. 23.
e s s ence.1A). This also adjusts the ~ ~ r ~of~the tablet. that this becomes a problem as the dwell time becomes smaller. so that the fill weight of the tablet. It shall be seen. 23. about 60 tablets per minute. and the top punch comes down and o a tablet. U g. in general. in tablets made on a singlepunch machine. under rotary presses. and this can be adjusted.p ~ n & ~ ~ a & ~ i(oreeccentric press) is schematized in Fig. ecause most powders have flow rates in excess of 60 g/s there is generally no problem with the powder flowing into the die sufficiently rapidly (see Fig. The amount of powder that flows in is the volume ( Y cm3) times the cascaded apparent density (p' g/cm3) of the powder. epunch machines produce. The dimensions of the tablet are a function of the longest path this punch takes.4)
A
€3
C
D
Symbols:
Principle of singlepunch tablet machine.1) (23. 23. the hopper is in position over the empty die. ~ ~ i t It is obvious that the fill may be adjusted by the position of the lower punch in
pper has swung away.yis.then the amount of powder flowing into the die per minute is
w = NUg/s
(23. and powder flows from the hopper into the die. there is a functional relation between the thickness and the hardness. and the precision and accuracy of the fraction F . a function of both these factors. then the dose L) g. is given by:
u = Yp'
D = FVp'
(23. bordered below by the bottom punch. In tabZet n frame A. is given by
It is obvious that the accuracy and precisionof the dose are a function of the accuracy and precision of the fill weight W .The s i ~ g Z e .3)
W is the required flow rate for the powder.1. Co~tent ~ ~ o r hence. so that the ~ ~ i c ~of e s s ~ e tablet may be adjusted in this way. If r of tablets per second is denoted N..2)
If the fraction of the powder consisting of drug is I. at top speed. The ~weZZtime t s is given by
t = q/N
(23.
.
23. a r o ~ thed ~ periphery at 120" angles (Fig. the dwell time is
z =~
1 2 z ~ ~
(23. 23.4). highspeed machines are equipped with twoor three hoppers.where 4 is the fraction of the cycle (seeframes AD) the hopper stays in the position of frame A. a hopper feeds powder into a feed frame. ejected from the die by the extreme upward movement of the lower punch is removedfrom the die table at the end of the circle at the back of the hopper (the ff bar).
. and all ~anufacturing equipment is of the rotary type. 23. W is not (without further ~anipulation) changeable. because the machine operation is adjustable and.
s
Singlepunch machines are slow and are used mostly in product development and in initial clinical trial batches when raw material (drug substance) is in low quantity. rotary machines are resorted to.2). under which dies and lower punches receive the powder (Fig. Qnce outside the feed frame the upper punch descends and its downward movement. hence.lr = U / z = U 2 z ~ R I a
(23. One of the limiting quantities is the flow rate of the powder. produces the tablet. If the radius of the die table is denoted R. it follows that the required flow rate W is given by
I.6)
or. Some requirements are apparent from the general setup. in combination with the upward mo~ement the lower punch. for a given powder. then the linear speed of the dies is 2 z M . rotations per second) as possible. It is economically desirable to operate the machine at as high a speed (A rps.3). As opposed to a of singlepunch machine this is referred to as doublesided compression. and the cycle repeats. The tablet.5)
With a required fill weight of U . In a simple setup (Fig. the maximum allowable machine speed would be
Schematic of rotary machines. en larger quantities are required.
desirable pressure is applied. 23.4. three hoppers are placed at 60” anglesto one another. granulatio led into the die as it passes hopper I3 and Compressed a bit harder between hoppers and C. A is first 3 and dies in a feed frame between A and € . and three . and then finally C is filled into the die as it passes hopper C and the final.6) and (23.
Legend: A = Hopper. as shown in Fig. and C are placed in the appropriate hoppers. 1979) that flow rates are dynamic in a sense. and the flow rate from hopper to feed frame is L g/s then the lag time must be at least
z = NUIL
(23..7)
This is one of the flow rate requirements of the powder.for which the two compounds form very low temperature eutectics. Laughlin et al. 197’7). loosely compacted. One manner in which such a separation may be accomplished is by way of either a doublelayer tablet or a triplelayer tablet.5 then shows that the maximum machine speed would be: (23. It could be a case of physical chemical incompatibility. In the latter. 1979. If there are N stations to be filled under the feed frame.8)
Inserting this in Eq. B = Upper Punch.9) The smaller of the two Avalues in Eqs. the layout is. D = Tablet 0= Loose Powder = Denser Powder
. and a classic example of this or is methypry~on and caffeine (Carstensen. An example of chemical incompatibility is aspirin and dialminate.
=~
a
/
~
2
~
~
(23.A. 23.in that the powder in the feed frame differs from the static flow rate experienced whena powder flows through an orifice ofa stationary hopper. (23.8) is the limiting speed at which the machine can be operated and still produce weightquality product. but it should be mentioned at this point (Carstensen and Laughlin.
At times. Powder flow has been treated in Chap. 18. The tablet is then ejected at a knockout bar at the back of hopper A. C = Lower Punch.. it is necessary to “separate” two components of a soliddosage form.
but for incompatible drugs it may be a last resort.ressures and speeds are adjusted so that the layer separations are particularly if the layers havedifferent colors. and compressed (D). there may be a “spot” occurring. in the foregoing examples. it is then filled to the top with outer granulation (C).
.. In that case there will always be interaction in the contact area between. primarily because of the complicated nature of the construct. (1998) have described how core material properties affect the the compression and the properties of co~pressioncoatedtablets made with microcrystalline cellulose as the coating material. there may still be interaction in the ~ c ~ Z packed product. or whenspecialrelease effects are desired. Twolayer tablets are employed when the incompatibilities are less pronounced. Thomas et al. triplelayer tablets may be resorted to (Fig. In more pronounced situations of incompatibility.and here totally incompatible components can be separated by a neutral layer (see Fig. and the associated lower tablet machine speeds. if it is simply packed (in random arrangement) in a bottle. The process is slow (800 tablets per minute).4) or. bicoated or tricoated tablets are a solution. Tabl~tswit hi^tabletswithin tablets (tricoating) also is possible. The matter of flow has already been discussed.5). 23. Figure 23.. In twolayer tablet manufacture there are only two (or four) hoppers. there will be contact point between tablets. and a tablet is dropped into it (B). It is not used often. In that event. If the incompatibility is ~ ~ (e. The two incompatible compounds. would be ingranulations A and C. 23. and whenever layer A from one tablet touches layer C of another.~byway ~ of eutectic formation). the two layers.g.5 showsone end of a die table.5.
Hiestand states that the mechanical criteria for a successful tablet for~ulation are good flowability for powders and adequate strength without fracture for compacts. The principle of a bicoated tablet is shown in Fig. A die is first halffilled withthe outer granulation (A). 23. but further aspects of it will be dealt
00 0 00
0
OO
0
0
Tablet Die
0 0 0 0
0 0
0
0 0
O0
Schematic of triplelayer tablet turret.
” and sufficiently large. It is one of the two latter steps that is responsible for bond formation.e. Up toa certain pressure. Tableting is carried out by applying (compressional~ stress to a powder bed.7) breakage will occur.6)isexceeded (point I3 Fig. so that they will bond when they are compressed. 23. This is denoted the elastic limit or yield value. but it will not return to its original shape).It is also laborintensive and. The aspect of strength of materials and of compacts will also be discussed. if the pressure is lifted. the solidwillrelax. Once P(1) (see Fig. P(l). the deformation becomes irreversible (i. to create a “chemical” bond... is said to exhibit elasticity. t 1981).then it will “give” (ix.6 a cubical block is exposedto a pressure P.6). The solid. 23. and (c) fracture (Fig. it will becomethinner and wider).7). in the 1 9 6 0 ~ ~ was a sustained and successful attempt to there accomplish tableting of drug substances by simply mixing them with excipients and compressing them. and then removing the water by drying operations. 23. among others. involved in the tableting are (a) elastic deformation. 23. To this end the surface molecules must be d i s t a ~ by no more than molecular dimension distances. If.
with in the following.. so that they will flow well. and add binders. this is reversible (Le. The intent is. as exemplified in Fig. 23. further increases in pressure may then result in further deformation (see portion C in Fig. in this range of pressures. then the block will remain intact. has shown that the steps. It has been seen in previous chapters that efforts are made to make particles “round.7). by causing close proximity between the molecules in one particle to those in another. but will be distorted. (see point C in Fig.If the pressure isreleased at any point in this ion. eyond this point. This entails wetting the powder. if the pressure is released then the original form will be regained. and this is denoted brittle frac
. and as such is energyinefficient.A
I
B
Principle of bicoating (tablet within a tablet). (b) plastic deformation. 23. At a given pressure. Carstensen (1977.
7.
Strarn
Stress
Strain versus stress profile. which has been shown earlier and is repeated here for convenience. If. 23. graphically. ften. This is shown.P
Reversible
P(3)
Schematic of yield and fracture of a block. the
Up to a certain limit ofcom~ression pressure. and the bond formation that occurs in this region isinsuf~cient make a 4 4 g ~ ~ d 9 9 so the pressure has to be brought all to compact. however.
. the way up to P(3) for good bonding to occur. in Fig. the higher the co~pression pressure.7. tablets will becomethinner.
ture. (1994) have d e t e ~ i n e d Poisson’s ratio for microcrystalline cellulose. The slope of the line AB is denoted the Poisson’s ratio. This situation is denoted bonding by brittle fracture. Robers et al. u. the portion BC is small. 23. the thinner the tablet will become. then bonding is said to occur by plastic deformation. adequate bonding between particles occurs in the region C in Fig.
5 ton/cm2.1 are the experimental value.118 0.5 g/cm3.8showthickness data of a tablet as a function of compression pressure.11)
so that.1 are plotted in value calculated to be seen that 1/34 = 0. The fill weight is 900 mg.9/1. the smaller the porosity (E) and the socalled Athyeckel) equation states that
.61 0.4 1.032 0.35 0. that the slopes of suc where 4 denotes yield value.12) The thicknesses in Table 23.8.68 0. Cooper and Eaton (1962) improved on this by assuming that two processes were at play in the compression. so that if there were no porosity at all. The tablet has a crosssectional area of 1 cm2 and the true density of the contents is p = 1.66) = 0. The compression would first.
Example of Porosity as a Function of Applied Pressure ~o~pression pressure Thickness ( t o n s / c~2 ) Porosity 2000 2500 4000 5000 6000 7000 (8000)
(cm) 2.66. so that the porosity E can be calculated from (23.916 1.5 = 0. and the calculated porosities are shown in the third column of the table. It is strictly an arbitrary convention in rather than In[&].016 0
. of The ~egative the natural logarithm of the porosity is plotted versus applied pressure in Fig. then the thickness h of the tablet would be given. 23.4 0.all the porosity is gone (Le.1 and Fig.64 0.ln[Porosity]
0.62 0. through rupture of particles and their percolation.062 0. If latter were done.10)
where a and b are constants (Heckel. at a thickness of 0. by
h = 0. fill larger void spacesin the powder bed.442 4. so that the yield is 41/(3 x 0. Table 23.60.The thinner it is. eckel (1961) found. and the data linearize.067 2.In[&] aP + b =
(23.5 0.135
.78 1 3.6cm = 6mm
(23. nowing the thickness at any given compression pressure.137 2.
E
= 1 . then the data would the literature to plot . 23.{P’lP)
Linear~tyin the Athy~eckel equation is always somewhat lacking. experimentally. The data in Table 23. and once that process was over. The negative of its lo~arithm then is shown in the fourth column.In[&] still be linear.60 0.the tablet is simply solid). but with a positive slope. 1961). a similar procedure will give the apparent density p t ..
V ) / (V0 . a and k l are constants relating to the first fracture and percolation.10.
it would repeat. so that the ial values approach 1.10 6
o l .shed that for naproxen and PMC the equation gave a more rational fit than the ckel equation (Fig. V is volume at a pressure P . ~
Data In Table 23.0.whichis what it should be (V = V*). and the now smaller particles would rupture and fill the "smaller" void spaces. the exponents tend toward zero.10 has an asymptote at (Vo .1 plotted by the AthyHeckel equation.
( V0 .13)
where Vo is volume before compression.
+
5 4
0
1
2
3
4
5
6
7
CompressionPressure ( t ~ sem) ~ . Chowhan and Chow (1981b).
.1 plotted linearly.V*) = a exp(kl / P )
+ b exp(kz/P)
(23.11). 23.kP
6
7
8
Data in Table 23. An example of the trace produced by such an equation is shown in Fig. 23.5 = 1. ' " * ' " * ' " "
1 2 3 4 5 Compssion Pressure.5 0. and b and k2 are constants relating to the second step. en P tends toward infinity. hence. 23. This led to a compression profile that was best described by fractional volume compression. the curve in Fig.V ) / (V0 . and V" is the volume corresponding to true density.V*)= 0.0.
.
03 02 0.V)/(Vo. 23.
If a powder mass is confinedin a cylindrical space. at first sight. and a force is exerted on the top of it.Sexp(lO/P). how some of the force is “diverted” to the walls of the confining space. seem that the downward force would simply be propagated through the mass. Fig.81
0. and part o be “lost” to the die wall.0 0 2 4 6 8 Pressure 1 0 1 2
L
Trace of the equation (V. and CG and DG center of the sphere in the second layer. and the reactive force on the bottom of the space would be equal and opposite the force exerted on the top. osed into wall forces HJ and QK and the d o w n ~ a r d forces uring the compression there will be work exerted. P.1 0..)
the CooperEatonequation. ~ ~ ~ p i c ~ t i The diagram explains the socalled Uncleequatio~.Sexp(S/P)
+ O.V*)= O.6 05 p 0.85 0.(Data
from
. toward the wall.8. and two ispresented.001 x k& g. it might. 0.
1.86 0. one.82 0. 23. 1981. This force is propagated along the same center line to the two spheres in the bottom layer GQ and GH.83 0. the downward pressure. However. In the diagram the hypothetical situation of a “powder mass” of five spheres arranged as two.0
em
15
Example of data that would fit Chowhan and Chow. schematically.84 0.12 shows. This will be touched on further under Sec.4
f
$
r.0. The downward force is A and then decomposed in EC and DF.0
05
Pressure. that
0.
(1953) were the first to instrument a tablet machine. Applied forces are usually referred to by the symbol P . Schwartz and Weinstein. with a slope equal to the Poisson ratio u. from the surface by the relation: P . 23. The point P in Fig. then line (OP) to be regained. (1994) have reported on the ratio for microcrystalline cellulose. the first part of the compression cycle (part A) starts at the point when the punches experience a m~asurab~e from the powder. Perelman and Roman. that is. 1975. One ofthesewill predominate (Carstensen and
. that a more realistic representation is: Line segment OP: t = [ u / ( 1 .Leigh et al. 23. A fair amount of older literature on this subject deserves mentioning.13 represents the elastic limit.exp(kx) (23. Schwartz and oerner and McCabe.. but in dis~ussion the compression cycle in the following.14)
iguchi et al. 1967. die wall pressure (stress) will be denoted z. of the powder). so that the pressures on the upper and lower punches and on the die wall could be monitored during a compression cycle.13.as visualized in Fig.. then bonding will occur through plastic deformation or brittle fracture. 1971. and bonding removal of the load will cause the ori~inal will not occur. and applied pressure will be denoted a. When forces (or pressures) are applied above point P. Paris et al. = P.13. 23. namely.15)
Mars~all (1977) has shown.sic
I
Schematic showing die wall force. Robers et al.
decreases through the compression mass as a function of the distance x.
z = VCT
(23. If this ishigh. 1972. however. that of Long.u)]a (23. 1965. This will be shortly dealt with here.16)
As pharmaceutical examples of Poison’s ratio. 1960. If we refer to the upper portion of Fig. From this point (line 0 force wall stress (or pressure) z isfairly linear relative to the applied pressure (a.
“ I
P(lower) P(die wall)
C
D
S
0
Compression. 17. This isakin to the shear loci discussed in Chap.. die 1. and if the yield value isdenoted (3then the 1. It is noted that the cycle predicts a resi~ual wall stress of (3
iestand (1997) also points out thatdescribing the properties of compacts by linearity equations has limitations.18) At point R the die wall stress is greater than the applied stress by a value equal to the yield stress so the equation for line Segment Line
RS: z = cr
(23.. viscoelastic Mohr bodies.cycle. equation for the line segment PQ is:
Line S e g ~ e n PQ: z = cr . on then the solid behaves as a liquid. and onhomogeneous.
If the bonding is byplastic d e f o ~ ~ a t i (Le.(13 t where z is the stress on the die wall.19)
~arstensenandTour6 (1980) integrated these equationsto find the area ) within the compression cycle and found it to be linear with Tmax (Le. tosome degree applies to a powder bed as well as to a solid body.
. because compacts (in practice) are porous. the “applied compression pressure”).17)
+ (13
(23. obeys the equation (Parrott. 1990) (23. considering the body viscoelastic). ohr body is a construction showing z versus a and.
c The normal stress in the symmetrical situation equals 4 2
(23.NO in Fig. then failure occurs.and the normal c o ~ ~ owill be ~ t ~ e
C F
= (01
+q)/2
(23.22)
(23.16). only one Mohr circle then suffices. hence. t.
.
Line ~ e ~ OP:te [u/(l ~.21)
If total symmetry is med. ( 4 2 .23)
+ 212
and recalling that tangential stress equals friction times normal stress then gives
Q
(01
+03)/2
Shear locus and Mohr circles. cha~acterized that of by hr body.t / 2 The tangential stress in excess of C (see R . plane of shear.v)]o ~ = n
If the stress increases beyond C (see Fig.2/2) . equals 012 .14) is. then the segment OP isstill g (23. 23.14). 23. in this situation.
p)a/(l e n point (23.3
(a/2 .26) At one point (point ) the radial stress will exceedthe normal stress.Qrnax)l/(l . 23. the ““applied compression pressure”). hence.The residual die wall force.>)
(23. being the of residual force F(d). with the restriction that it goes through the point where CT = z. where
Tmax
= ~ r
E(1 .give information on whether bonding occurs primarily through plastic deformation or through brittle fracture..15 is the normal force and E is the tangential force. that is. and the descent will now be of a Poisson type.15.26)
This may be rearranged to express z as a function of cr. a certain ejection force E is necessary.25)
so that the equation for line segment
+ P ) ) + ([v(Q .2cl/(1+ J?L)
(23.
. divided by the area of the wall..212) .x.x>.24) The line segment QR is parallel to line segment OP and passes through the (qn.e. To eject the tablet. The equation for this line is Line Segment
RS: z = {[(I
+ p)/(l .p b r n a x . and is exemplified in Fig. so that the definition of frictional coefficients gives
E
Example of the definition of frictional coefficient.28)
Carstensen and Tour6 (1980) integrated these equations to find the area within the compression cycle and found it to be proportional to the square of zmax (i.[(2C)/(l + p)]
(23.p)]cr . which gives the equation for Line ~ ~ ~ m PQ: zt = [(l .c = p(cr/2
+ t/2)
+ p)]o . 23.1/2~/(1 p ) ] ) ( 2 ~ / ( .27)
It is noted that the residual die wall pressure is obtained by setting cr = 0. The d e f i ~ i t i o ~ frictional coefficient is recalled.
After completion of the cycle there remains a residual die wall pressure. F ( d ) in Fig. Such plots.p)} 1
(23. it is pressure:die esidual wall (23. %.
it is possible to monitor upper punch. the closer the ratio (23.E = pF(d)
(23.32) The energy lost (the elastic energy. If the thickness of the compressional mass is denoted h. will be obtained. the better the formula is lubricated.3 1)
y means of displacement gauges. We)after the pressure is released is given by
Punch Force. During compression a curve. Work equals force times distance.16 result..29)
where EA. and after pressure is released a curve.30) is to unity.is given by (23. Fgd and lower punch. forces and at the same time measure the depth of the upper punch intrusion (on a singlepunch machine). F l . is the frictional coefficient between the die wall and the tablet mass. There have been attempts in lit~rature to assess the frictional coefficient between a compressed powder mass and adjoining metal (Carstensen et al. (19’77) suggested that work on the lower punch divided by the energy input of the upper punch in the compression cycle would be a better index. 23. (23. such as AB. suchas OA. will be obtained. then c o ~ b i n i n g Eq. iguchi (1954) suggested that the ratio between upper Fu and lower punch pressure F1. 1980). ~ ~e~
. When such traces are obtained. the socalled Fratio. profiles such as shown in Fig. F
x=b
I
FO
C
X
0
X=O
D
Punch Displacement F o r c e ~ i s p l a c e plot. so that the compression work or energy Wc.30) with the Unckel equation then yields (23. Guyot et al. was an indication of the lubrication efficiency of the formulation. that is.
.0.(23. The deepest invasion of the upper punch is denoted x = q in Fig. The force displaceme~t profile isdenoted f ( x ) in Fig. and brittle fracture or plastic deformation of the particles will take place. F = ~ o b / ( .40 may be written:
(23. e (Data from Fessi et al.16. 23.0.36)
so that the total energy or work. the force equation would be F=F'/(bx) F' Fob O ~ x t b (23.16. although the slope differs from negative unity.)
. the maximum applied force F" (occurring at x = q) is
F" = F ~ b / ( . In Fig. 23.63039 .39) (23.4) b
Eq.X) b ala treated in the logarithmic form of this (Fessi et al. and Fuhrer (1965) r~entier (1974) suggested that f ( x ) was hyperbolic. 1981) are shown in Fig.37)
The first part of the compression event isa consolidation below the elastic limit (Fo) The energy consumption in this area is relative small. The fit is good.75357~ R = 0. 23.41)
y zz . 23. From this it follows that
= :
and.40) where F' is a characteristic constant. asymptoting at x = b. eyond this limit..17. Wtotal imparted on the tablet after the cycleis complete is given by
A
DO =
0
{f. 1981.(x) f2(x))dx
(23.38) (23. there will be a substantial amount of work needed for further inva n of the upper punch. 23. hence.16..995 2 '
2r
h(bx)in mm
7 ~ i p h e n h y d r a ~ i nhydrochloride tablets containing 50% polyvinyl polymer.
46)
so that Fo = exp(8.inKO]. It may be seen that the slope/intercept ratio gives ln[Fo] = l22/14.Fob ln[Fo] (23. Since work and intrusion distance are correlated by dW = Fdx it follows that the energy imparted on the tablet is
(I
Fob/(b.00+ 14.18 shows data by Fessi et al. 23.987
0' 9
I
I
I
12
10
11
wF+l. It and that the slope to intercept ratio should be .6 (23.43) 0
eferring to Fig. 1981. The mean yield pressure (MY ) has been reported for formulations.122. (24.(23..15 = 8.N
Ene~gy plotted against ln[F*]. (1981) treated by way of Eq. (23.40) and (23.x)dx = Fo~{ln{b/(b q)}] (23.43)
We = Fob ln[F*] .
y = . and Nystrorn (1985) determined that when microcr~stalline cellulose ( 100) is present with such compounds as lactose and acetaminophen (paraceta~ol). (23.)
. A formal way of doing this is to obtain done from the length of intrusion the porosity G of the fill at this force (which may be at the given value of Fo) and then assume that the force works over an area of Ar. this number would have to be divided by the area over which the force is in effect. f f Figure 23.41) into Eq.44).45) (~3.151~ R"2=0.42) where and Q are constants. is noted that this is a ~ e a n s o obtaining the elastic limit o the tablet powder.6 and introducing Eqs.44)
his predicts We (energy) whenplotted versus ln[F*]should give a linear trace. (Data from Fessi et al.6) = 5900N
To obtain the stress at the elastic limit.
Lahrib and Wells (1998) showed that at low PEG concentrations (mixed with dicalcium phos~hate). (b) punch depth at which porosity is zero C = cohesive stress E = ejection force F = fraction of fill that is drug F’ = force constant in Fessi equation F = force sensed by punch during compression cycle F* = ma~imum force during compression Fl = lower punch force Fu = upper punch force F* = maximum applied force occurring at x = q F ( d ) = residual die wall force f(x) = function describing punch force as a function of punch depth L) = dose h = thickness of a tablet (cm) k = rate constant in the Unckel equation k l .
A = crosssection of a tablet (em2) a = (a) length of feed frame. (b) slope of a Heckel plot b = (a) intercept of a Heckel plot. the MYP decreased linearly with polyethylene glycol (PEG) concentration. (b) R = F l / F u = Wiguchi R ratio rps = rotations per second U = fill or tablet weight Y = (a) die volume. (b) fractional volume of a powder at a given pressure P Vo = fractional volume of a powder at the point of closest packing Y* = volume at infinite pressure W = (a) required flow rate. = pressure at upper punch q = fraction of the cycle the hopper stays over the die = constant in the Fessi equation = see F* = (a) radius of die table.YP decreased in a linear fashion with MCC concentration. k2 = rate constants in the CooperEaton equation N = number of tablets per second P = pressure P . = pressure at a point x below the tablet reference plane P. (b) total energy imparted to a tablet during compression We = elastic energy lost after removal o f upper punch We = work imparted during compression Wf = total energy imparted on tablet after the cycle is complete x = distance in tablet mass from upper punch surface /3 = constant in the Fessi equation E = porosity h = rotational speed (rps) p = frictional coefficient
.
Academic Press. Perelman VE.Germany. Chowan ZT. Lea & Febiger. Handbook ofPowderScience Technology. Int J Pharm Technol Prod Manuf 6:27. 1001. Robers RJ. Chow YP (1981). Roman OV (1971). Technical University C a r l o . Mase GE (1970). McGrawHill. R o w RC. Schwartz JB. Oudernans GJ (1977). Eaton LE (1962). J Pharm Sci 56:888. Wells JI (1998). Puisieux F (1975). Carstensen JT (1979). and Carstensen JT (1984).SolidPharmaceutics:MechanicalProperties andRate Phenomena.Otten L. Continuum Mechanics. Pamentier W (1974). = yield value for viscoelastic solid p = particle density p' = apparent density crmax = maximum stress in a compression cycle (compression pressure) crl = minor stress component of normal stress in twodimensional Mohr circle a3 = major stress component of normal stress in twodimensional Mohr circle z = (a) dwell time. Higuchi T (1954). Fuhrer C (1965). Kanig JL. Lachman L. Parrot E (1990). Int J Pharm 105: 177. Van Norstrand Reinhold.~ i l h e ~zu ~rauhschweig. Int J Pharm 160:197. J Am Ceramic SOC4597. Schwartz EG. Trans Metallerg SOCAIME 221:671. Anderson NR (1986). New York. Wassink M. In: Fayed ME. New York. Fessi H. pp 236237. J Am Pharm Assoc Sci Ed 43:344. Dissertation. Lahrib H. In: Lachman L. eds. Philadelphia. Van Campen L. Int J Pharm Technol Prod Manuf 2:29.
Banker GS. Puisieux F. Leigh S. Duberg M. Lieberman HA. Powder Metallerg 6:73. Takinddin M. Laughlin SM. Carless JE. New York. Duchene D. Carstensen JT. Nystrom C (1985). Carstensen JT. Presented at the 2nd I nternational Conference on Compression. ~ o l ~ a AR (1969). Marcel Dekker. Int J Pharrn 3:32. Burt BW (1967). Marty JP.cr = normal stress Q. Duchene D.Powder Technol 18: 187. Long WM (1960). J Powder Metallerg 9:692. Sept 2 4 . Klein RJ. ~ . TourCe P (1990). p 170. Powder Technol 26:199. Proc 1972 Powder Metallerg Conf pp 225241. Carstensen JT (1981). Brighton.. J Pbarrn Sei 70:711. eds. Puisieux E. Laughlin SM. RankinPJ.eds. Laughlin SM (1979). Bannick J. Heckel RW (1961). i~a ~raunschweig. Arch ise en butt en we sen 18: 161. nd Strijbos S. Paris J.
. p 57. Dtsch Apoth Ztg 105:1150. Carstensen JT (1981). In: Lieberman HA. pp 293345. J Pharm Sci 70: 1005. Unckel H (1945). Powder Technol 23:79. pp 252269. England. Int J Powder Metallerg 5:79. McCabe WM (1972). pp 190. The Theory and Practice of Industrial Pharmacy. Cooper AR. Pharmaceutical Dosage Forms: T a ~ ~ evols2. New York. Koerner RM. (b) shear stress
Carstensen JT (1980). York P (1994).
This Page Intentionally Left Blank
.
pression Excipients crocrystalline cellulose 24. 1.8. 24.2.12.13.3.6. 24. ~ s p e r i ~ ~ 1 t i n g 24. 24. 24.7.15.5. Loading and Particle Size ons side rations ~ 24.3. ~e t Cranulated Tablets 24. Tensile Strength)
.9. 24. Direct Com~ression e c ~ a n i s m s te 24. Variables 24.10.17. C a ~ p i n ~ 24.1 1. Chitosan and xylitol t omp press ion Excipients
24. ~ a l t o ~ e x t r i n s 24.24.2. 24.14.16. Uniaxial Expansion 24.6.6. 24. Symbols eferences efects in Direct Cornpression oller omp pact ion ardness ( C r ~ s ~ i n g Strength.4.
irect Compression Conditions
408 409 41 1 41 1 413 413 414 414 414 414 15 15 416 416 417 18 419 420 42 1 22 424 425 426
24.
it might become more difficult to meet content uniformity standards.. and if it is possible to simply mix powders and compress them.e. then the process would become less laborintensive and more economical. aspirin) would have to be processed dry.. It is obvious that not all excipients or drugs would be directly compressible. wettability). But direct compression without these precompression operations is not possible for powders with very high yield values or poor flow characteristics. and excipients that are directly compressible are denoted “direct compression excipients” or C excipients. The chapter to follow will deal with some of the directly compressible excipients that may be used for this purpose.g. but the aspects of economics has made direct compression attractive.
Schematic for Process Selection DC excipient concentration Drug concentration High Low Drug flow rate Good Good or bad Bad Good Bad Drug compressibility Good Good or bad Good Bad Bad Method Direct compression Direct compression Slugging. and it is possible to simply place the powder in a die and compress it.
If the yield value of a powder is “low. tablets were primarily made by wet granulation.1 outlines the combination of properties that necessitate or allow the various processing options. the powder must also flow well for it to be directly compressible. that they have fairly lowyield values). and it will form a tablet (once the yield value has been reached). that for such a system to work it would be necessary that some degree of “autocompressibility” existed for the major portion of the ingredients (i. If the powder has a highyield value.” then it is often referred to as autocompressible. higher than practically achievable on a tablet machine. to s first add water to a mass of powder. It i illogical. then wet granulation is a means of achieving a compressible formula. Such an approach is denoted “direct compression” or simply DC.Historically. If the powder itself is a ~ t o c o ~ ~ r e s s ibut e . Some aspects might be lost (e. The attributes of flowability. However. That certain products (effervescent tablets. then it can be ~ Z does made flowable by wet granulation or. and then to remove it. Table 24. roller compaction Wet ~ranulation Wet granulation
Low igh
. as shall be seen later in this chapter. and wettability were considered best achievable by means of such a process. not flow well. compressibility. because the binder will have a sufficiently lowyield value that bonding can occur. by slugging or roller compaction. led the way to what is known as direct compression.
. 1980). also obvious what high means (e. it is possible to calculate the number of sm icles necessary to cover the large particle. equals onefourth of the surface area of the small particle (Fig. 1980)was arrived at on statistical grounds.g. even more importantly.
Saturation point
of small particles with ones. then that would constitute a weak area in the tablet which could give rise to capping and breaking.g. if one deals with a drug that is dispensed in microgram quantities) that the concentration is “l00w. knows that trialanderror is the only decidingfactor.1). This is particularly true in attaining adequate b~ending content u n ~ ~ ~ ~ i t and To achieve ordered mixing the maximum amount of material of the “small size” component (in this case the drug) is given by the following argument (refer to Table 24.” The 16%rule (Carstensen. This.1). The following nomenclature is now used: the number of small particles per large particle required for “full coverage” is n. A general rule of thumb (Carstensen. The surface area of the large particle is
AL
= nD2
(24. in the opposite case. Another consideration. and arrive at a critical weight ratio. the density of the small particle is ps. in turn. and for the large sphere the diameter is D and the density is p L . then direct compression is physically a possibility. As Kirchhoff once said: “Theory guideth. it is usually difficult to assess particle size distributions of small particles (of particle diameter d). er. even if the drug substance itselfis not autoco~pressible. These authors simply note that the projected area (the cross section) of the small particle is nd2/4. the argument being that if a certain number of drug particles would find themselves as neighbors (as calculated by probability statistics) in a compact. in some It situations. then the DC component should be sufficientto “cover” the drug substance and. whichis the crosssectional area of the small particle. is that if the drug content is 16% or less. experiment decideth.’~ is.. large
. (1982). Consider the surface of the large particle (AL). The question is where the cutoff point is.The following deals with mixtures of lowconcentrations of nonDC drugs with excipients. In some cases it is quite obvious (e. however.it can accommodate a certain number of small particles. the amount of drug should be enough to just cover the excipient. and it is easierto doas doneby Nystrom and Glazer (1985) and Nystrom et al. a sulfona~ide [usually in doses of 250500 mg]) would be a high concentration. 24. This is akin to ordered mixing.1)
114 Actual Area
Fig. Given the diameter of these. is that when the drug is present in higher concentration. The experienced formulator.
However.I/ at complete coverage is.nd3/6)
(24. direct compression can be dusty.When drug concentrations are low.7) now gives
W = 4RAs/AL
The advantages of direct compression isprimarily economic. particularly when the drug content is low.
. The attainment of adequate content uniformity can be difficult.2) and (24.and the surface area of the smaller particles is
As
rind2
(24. there are cases for which the opposite is true. (24. the hydrophobicity of the drug may be such that wetting is poor in the direct compression formulation. and punch wear is considerably higher than for wetgranulated products. but there are also disadvantages to direct compression. direct omp press ion is a distinct possibility for a drug candidate. (24.1) have been used in the last step. Furthermore.
so that the amount of material required to “fill up” the surface of the large particles is four times the surface area times the number of the small particles (i.10)
and introducing this into Eq.e. This may make a wet granulation a more desirable candidate for development because of dissolution and bioavailability considerations. n4AS).2)
The projected surface area (the cross section) is one quarter of this.. The number of small particles n that will accommodate it is
The mass wL of the large particle is:
W L = pLnD3/6
(24. or the large particle the specific surface area S2 is given by
and similarly for the small particles
ence.6)
The weight ratio I. the ratio R between the small and the large specific surface area is
= PsDIpLd
(24. therefore
where Eqs.5)
and the mass of small particles required to fill up the surface of the large particle is
~1
= 4{np.
at times. o al.
. for instance. The same ay hold for drugsubstances with low melting points (Skotnic~y. In interlocking (Fiihrer. van der Walls forces (Carstensen.It mightbe asked: “What physical characteristic of a substance makes it. 1953. In case (a) i. or periodic. or a mixture of it with other substances.’ allow the to surfaces to come in close contact. 1977). iguchi. electrostatic forces)..depe~ding on the molecular arrange~ent.. The point of asperite bonding was a favorite theory for a while ( iguchi.e. as with microcrystalline cellulose ( ~ y s t r et ~ 1993). hydrogen bonding. itis not a fusion in which there is total bonding (i. This type of bonding is the primary type for pharmaceutical materistrom et al. 1996. possible. Adolfsson et al.
B
Schematic of bonding by van der Waals forces. then fell in disrepute. As shown in Fig.. 1997).e. Suffice it to say that (a) adsorbed air must somehow be “removed. or (situawhen plane surfaces proximate within molecular distances.. it is a matter of placing surfaces together at molecular distances. ridging is a possibility as well. but obviously has some merit..2 this can happen (situation A) when asperites meet either other asperites or plane surfaces. whereas (situawhen twodiffe~ent substances “meet. large shape factors (irregular surfaces) and fractal dimensions (surface roughness) are the main contributors. and(b) even when they comein contact.. 1980). melting is the method by which bonding occurs. 1993). directly compressible?” There are several possibility for bonding mechanisms: (a) distance forces (van der Waals. one is not making o crystal out of two ut there are sporadic. It should also be mentioned that asperite melting is. 24. 1968).
With waxes. situations duri which two philic ome within molecular distances of one another d form a type of chemical bond.” they mayor may not bond. and has been reported (Olsson et al. 1968). (b) solid bridging. and (e) mechanical interlocking.
it is actually the eutectic temperature that is of importance.3. in general.12)
where VL is the molar volume of the melt. used thermodynamic and mechanical arguments. asperite melting would not be suspected ofcompounds having highmelting points. 24. this has been done by simply measuring. to show that ~ ~ stress e ~ (as the particle is during compression). in which particle of A touches a particle of a
. whereas it would normally would have a highermelting point if both liquid and solid were exposed to increased pressure. naphthalene had a loweredmelting point.
.. calorimetrically. Most solids would have a value of dT/dP = V. but the melt (liquid) is subjected to the atmospheric pressure in the void space. Under those conditions the solid is under a pressure P. For most substances VL > Vs so that the melting point increases with increasing pressure. There have been several reports in the literature in which an overall increase in temperature during tableting has been demonstrated. Again.Consider the situation in Fig. the pressure i s such that the melting point is also expected to increase. the point has never been made that the question is really whether the eutectic tem~erature increases with increasing pressure. it changes with pressure P and follows the Claperyron equation: dT/dP (VL .T/AH (24. If one simply considers the melting temperature T of a substance.Vs)T/AH (24. H is enthalpy of fusion. (Water and bismuth are exceptions to this statement). Vs is the molar volume of the solid.2”/atm. At times. but the local temperature rise at contact points may be much higher. Although this speaks against asperite bonding. Add to this the following argument: Stotnicky (1953) and Rankell and Higuchi (1968). ankell and Higuchi (1968) employed an expression arrived at by Carslaw and Jaeger (1959) and found that for sulfathiazole
Schematic of asperite bonding. So.13)
For instance. although the temperature rises during compression. Stotnicky (1953) demonstrated that under point pressure. so that. the temperature of the tablet mass as the tablets come off the machine. dT/dP will aZways be positive. and in that case dT/dP = V. A and B can form eutectics and the shaded area then may become a eutectic
mixture if the temperature at the particular surface point during compression is above the eutectic temperature.T/AH of 0.
. because it affects the nature of the surface. 1998) has an effect transcending the particle size i ~ ~ et effect.1. 10"5x2. there is a maximum load that the direct compression excipient can accommodate to make good tableting possible.lets
AT = 0. least squares fit: y = 0. 1998. Asperite melting is generallynot seriously considered as a mode of b o n d i ~ ~ in modern literature.
The most commonly used direct compression excipients are
praydried lactose Dicalcium phosphate anhydrous (Atab) Dicalcium phosphate dihydrate (Ditab) icrocrystalline cellulose (Avicel. Leastsquares fit: squares. The ~ a ~ t i c Z e of both drug substance and direct compression excipient are sizes of importance (see Fig. 24. the foregoing considerations apply to pure substances. As seen.1). it should never be ruled out as a possibility.046/f
wheref denotes the fraction of the total area that is in actual contact.4 shows the effect of milling of NaCl on the tensile strength of the resulting compacts. (Data from Adolphsson et al.3 1. lending feasibilityof asperite melting. MMC) re gelatinized starch
150
200
250
Compaction Pressure (MPa)
Effect of milling on tensile strength.)
+
+
.R2 = 0.55 10F2x.
There are a host of variables that may affect the tableting performance of a direct compression formulation. and the possibility of eutectic formation can make asperite melting possible in mixtures.995.5 104x2.2 . milled: y = 2. however improbable.3 102x . ~ i Z Z (~dolphsson al. when a drug substance is not autocompressible. Again. and circles. If this is IOe3 to 3 x then AT = 4515OoC..8 6. although there are reports from time to time of its occurring and. Figure 24. whereas it might not be feasible for the compound itself. unground NaC1.
(1998) found that this is not due to physiochemical changes (ie. 1994. and Papadi~itri
There are yet other C excipients that have been reported. Velasco (199'7) has reported the useof maltodextrins (Maltrin M510. in fact. Li and m as wet granulation excipients.is a useful filler considered. for have reported on the effect of machine speed on the performance of four pients. (199'7) have reported on the use of Xylitab 200" which is xylitol granulated with 2% sodium carboxymethyl cellulose. for instance the bulk density is low. 1995) have shown that there are five types of ~altodextrins reported on the effect of storage and and humidity on their direct compression quality. (1984) and ~ p a d r a s h t a al. altodextrins are glucose polymers that are watersoluble..
At times the mere mixing of directly compressible substance does not suffice to make a product that is satisfactory in all respects. 1996).. side from the actual flow and compression requirements. Joyce et al. Olssonet al. by many technologists the best of direct compression excipients. wetacid or enzymatic interaction with starch. For the purpose of flow improvement. If measuredby Malvern Mastersizer. This has been investigated in several cases. The mean aerodynamic diameter (obtained by means of an Aerosizer Mach 2) shifted from about 55 to 30pm.. the foregoing that have been reported in the literature. but Tobyn et al. Mollan and Celik (1993. it does not exhibit excellent flow. it has limitations (Bolhuis and Chowhan. Armstrong and Palfrey (1989). However.Tobyn et al.
There are direct compression excipients other than the ones listed in. There are definite advantages to silicifying MCC. Grain Processing Corporation. it is sensitiveto lubricants and lubricant level. no bulk chemical change and no morphological change were observed). (1998) have investigatedthe common practice of adding pyrogenic silicaas a glidant CC. direct compression excipients must also be able to perform under highspeedtableting conditions. Nagai et al. Flow rates were not t it isgenerally known that silicized MCC flows more readily than artaris et al. (1998) point out that during
. and its compression characteristics are somewhat dependent on moisture content. (1992) have found chitosan to be an excellent direct compression et excipient C excipients than the one mentioned in the foregoing exist and have been reported on. (1984) have reported on the plastoelasticity and tableting of mic~ocr~stalline cellulose (in Combination with acetaminophen (paracetamol]). it changed from 122 to 105pm. and the particle density (obtained by helium pycnometry) was unaltered.
i et al. thus binding them together. Adol~hsson al. then an increase in tensile strength can be brought about by adding a binder before compression. and the porosity of the powder bed isthereby reduced. Tablets of mixtures exhibiting tensile strengths higher than tablets made from the in~ividual components themselves may result. These are often ductile materials (e. however. 1997a. so that particletoparticle contact (without
. 1997. and the mecha~ism their bonding (Nystriim et al. for instance. is given by the relation (24. such as derivatives of cellulose and starch). Yu et al. Newton et al. and R.. ry binders usually deform plastically and bond to the drug (and other) particles during compression. polymers. it facilitates the approach of particles to one another. and here. f the strength of the tablet is less than desired.. 1985. Olsson et al. for instance. and this amount. Furthermore. 1977b) so that the history of the is of importance in such an application. in combination with dicalcium phosphate (Larhrib et al. particles are made to come in closer contact. The cooling rate from which PEG is made from a melt affects the morphology (Chath 1985. 1991. 1985. 1997a. 1989).
When two materials are blended.. This is the primary cause of bond formation. Larhrib et ah. ry binders and their properties have been studied. and Wells and Langridge (1981). In this case the mixtures give harder tablets than those made from the component excipients (Vromans and Lerk.b). 1982 of Nystriim. 1985. such as their fragment. (1998) have evaluated ~ Q Z y Q x y e t ~ y glycoZs (PEGS) of a range le~e of molecular weightsas dry binders.g. may also be affected by rugosity and particle shape (Lahrib and Wells. have studied the dicalcium phosphatemicrocrystalline cellulose system as a direct compression component. 1977). as shown earlier. These compounds have been used dry binders as in direct compression. is the actual surface area ratio between binder and carrier.. Nystrom and Glazer.compression.15) where A denotes weightspecific surface area.b). 1982.. Tensile strengths. The amount of dry binder must be e n o u g ~ cover all or a substantial part of the surfaces of the remain in^ to ingredients ( ~ y s t r o m al. (1999) have shown how the combination of Tablettose and microcrystalline cellulose as direct cornpression excipients and crosslinked sodium carbosymethyl cellulose(Acdisol) as disintegrant allows optimization ofdirectly compressed tablets. the binder is referred to as a dry binder.
e may be expected to have an effect on direct (and other types of) compresisture in small amounts canact as a lubricant. Larhrib and Wells. they often improve compressibility and reduce propensities for lamination and capping ofrugswithhighyieldvalues. resulting in a compact with a certain tensile strength. Nystrom and Glazer.. et et 1998). 1988. Craig and Newton.
milled. but at higher pressures the opposite was true.” It is true that the latter would have betterwetting properties. Nokhodchi et al. and the most often encountered ones are discussed in the following. This produces a paste that “glues” the drugandother excipient particles together. This may not be sufficient to warrant not using a direct compression approach. e. but it is possible to compare tablets made from direct compression components that contain pregelatinized starch.. but it may not always be that important. indeed. The processis one of (a) ~articZee ~ l a r ~ e ~(b)~ t . The wet mass is sized through a desirable size screen. ~ o a d i n has already been discussed.dried. Granules made by wet granulations were slightly more compressible than directly compressible mixtures when the compression pressures were low. e improvement of the r o ~ ~ ~of thes particles. 20.’~ then the tablet. is of importance.g. The aspects of this has been covered in Chap. Large amounts of moisture are undesirable when direct compression is carried out because a drug substance may be moisturesensitive (as.b) have reported on the effect of moisture content on the compression and energy aspects of ibuprofen compaction. as mentioned in Chap. which is made into a paste (e.. The most common defect in direct compression is content uniformity. in a 1 : 10 ratio) with water. the oldest binder employed is probably cornstarch. The manner in which the magnitude of the effect may be investigated is to carry out dissolution studies on both tablets and uncomp~essed powder. Historically. 21.
At times anhydrous compression is dictated by the stability of a drug substance. Onetenth of it is suspended in one part of cold water. then the finished tablet will have the properties of the direct compression ingredient.
his process was reviewed in Chap. This became less pronounced at highermoisture contents. or as in effervescenttablets when more than minimum amounts of water will cause reaction between acid and base on storage. 21. aspirin). then the wetting of the material. any drug substances are very hydrophobic and have a very highyield value. Defects will occur. when made. must not contain long “strings” of the drug substance. lubricated. If one considers the drug substance completely “in~ompressible. and the latter aids in compressibility of the blend. ~ e s ~ e former properties. An ideally formulated wetgranulated tablet will not contain the same ingredients as a directly compressed tablet. Cost considerations may outweigh small differences in dissolution rates. Suppose the tablet
. and (c) adding a ~ i ~ The two~ . and compressed.ter
interfacing air) is made easier. if only smallamounts of g drug substance are present. the presenceof drug maygiverise to capped tablets. aid in powder flow. It is difficult to compare directly compressible versus wetgranulated tablets. Chowhan and Chow (198 la) studied the effect of water on methylcellulose granules.but it is apparent that.g. and in such cases. (1995a. and added to nine parts of boiling water. prepared both dry and “wet. If the uncompressed powder exhibits a lag time and the wet processed one none (or a much smaller lag time).
2. 24.I
8
10
lOOx
% defects when n = 3
YOdefects when n = 5
5
15
% defects when y 1 = 10
% defects when n = 20
13 15
a
25 35 60 80
35 65
0. 2in. then there will be six points of contact. two rollers. The probability of having a row of three would be (6f)2. This is shown in Table 24. than the original powder.5). because they are broken up in the next step by coarse milling. If the presence of n particles in a row is sufficient to cause a defect. then n is given by this number. riction (or forcefeeding) now carries the particles down into the area EFG where they are compacted into a sheet. This has as its goal to make particles that are larger and. ~ e a r r a n g e m ~ n t particles makes the powder attain its tapped density p&. In slugging. of necessity. These tablets may not have good fill weight uniformity.
There are highdo~e drug substances that. are present in their formulas in a high concentration (e. The probability that two drug particles should be neighbors is 6f. and so on.
.1 1 10 40
1 15
n is the number of neighboring drug particles needed to produce a defect. These then flow well and can be compressed directly on a tablet machine to the correct dimensions and with adequate weight control. 24. rounder.. in the of space CDEF. are rotated. but flows poorly. It is possible. which can then be broken up and tableted directly. The powder is forced downward into the region CDEF by gravity and the weight of the powder above it (and in some equipment by force feed). so that the adequate slugging pressure (force divided by area) can be achieved. If the drug is of reasonable yield value (is compressible). aspirin). but this does not matter.rind
s
is considered as a bodycentered cubic array. with high pressure applied between their centers. diameter) are made using very heavyduty machines.5.
Ta Percentage of Defects Expected in Direct Compression for Different Numbers of na and x (see text)
("/.g. one hopes. to calculate the percentage of defects to be expected as a function off and y1. The powder is fed into the space ABCD in Fig. large tablets ( e g . which feeds out below the area GH. therefore. so that if there are n particles in a row. then one often resorts to slugging or roller compaction. In this process. A more convenient way of doing this is roller compaction (Fig. and the probability of a drug particle being one of these is equal to the fractionf of drug in the tablet. then the probability is (Sf)"'. and the powder is processed through them and exits as a compacted sheet. where it attains it cascaded apparent density p'.
a ~chleuniger Hardness en the force is divided by the area over which the force acts (the rectansectional area of the tablet.6). They proposed that compression exudes water from the granulation
. p. ~m The magnitude of the hardness change is related to the type and concentration of binder used in wetgranulation. when firstmade. The sheet itself usually will be thicker than its ex . may possess a certain hardness.
heangle a is denoted the grippin heangleof rolling.Schematic of a doubleroll compactor. tablet. often quite rapidly. it is referred to as the yield stress of the tablet.. A. 24. and some d e f o ~ a t i o n particles. or the angle of compaction. Chowhan and Palagyi (1978) stud tions of naproxen (e. e work reported in literature is p~eno~enological nature. there is a pressing force. Fig. but this may change with time. The angle where this final thickness is achieved is e angle of release. followed by pla of tle fracture occurs.g. and then level off a t a n e q u i l i ~ r ivalue.with hydroxypropyl methylcellulose [H granulating agents) with particular emphasis on the effect ofmoisture on “stability” of hardness. and an allin inclusive theory of roller compaction has not yet appeared (
ardness is measured by placing a tablet between two anvils and measuring the force recorded in kilopond) required to break it (e’g.. In the zone of compaction. because of elastic recovery of the compacted mass.
and attributed this to a reduction in the glass transition temperature (Tg). The aspects of tablet hardness and crushing strength for wetgranulated tablets is approximately the same.
into the void space. Stubberud and Forbes (1998) found that polyvinyl ~yrollidone (PVP) would delay the recrystallizatio~. ablets at different moisture levels increase in hardness on standing overnight. but the fact of the matter is that they are always somewhat curved. If the point occurs at a higher than achievable compression force. The ~ o j ~ t ~ r e induced hardness increases in tablets prepared from granulations containing different binders.7) is that the tablet will start laminating and capping at high pressures. of dissolution) does not affect dissolution. This.7 shows this type of plot. then linearity is fairly good. and this gives rise to a
.e. these are plotted in linear fashion (i.b) have studied the crushing strength versus lower punch work and re orted on tablet characteristics of tablets made by wet granulation to 8% w/w water. often referred to as a c o ~ ~~ r ~ ~ ZTypically (particularly in company literature showing the virtues o ~ e . Figure 24. Chowhan (1980) has used HPMC in salicylic acid tablets. 24. up to point A in Fig. 24. ess is a function of the magni de of the pressure that has been employed to make it. but they had no effect on the tablet disintegration time and in vitro drug dissolution. (1996) and Sebhatu et al.~ ~ i o ~ of a particular direct compression ingredient). with a total PMC concentration of 3% wiw).7). but that hydrophobic excipients would accelerate it. from a mechanical point of view. as that for directly rycer et al. (1983a.. increased bonding strength by way of solid bridge formation. Stubberud et al. This (as opposed to hardnessinduced decrease. eventually giving rise to crystallization. (1994) observed increased tabletcrushing strength during storage.
The reason for the parabolic nature of the plot (see point in Fig. and that this causes recrystallization of the drug or soluble excipients. in turn. induced by moisture. He linearly related to the amount of moisture lost under compression.Stationa~ Anvil Anvil Movlng
I
Breaks Cross Section 2 x Radius x Height (Thickness)
Schematic of a diametral hardness test. not attrib~table to moisture loss.
N'.. however. A hi~hcompression pressure would give rise to a large number N . several bonds.0900~ RA2 = 0.y = 3. This gives rise to decompressional stress on the tablet. of contact points (i.. the hardness will decrease).
. At a given point. the stress is released ~ ~ i a x i a Z Z ~ (i. Capped and laminated tablets are shown in Fig. the tablet expands in only one direction).e.
Laminated Tablet
Capped Tablet
Laminated and capped tablets. iestand has proposed indices that are directly measurablefor determining the propensity for capping. 24. the extra strength imparted by additional pressure is offset bythe additional stress inthe expansion.e. are created. It is almost linear. When materials are poorly compressible.
weaker tablet (Le. that is. and this number is proportional to the compression pressure.0100
+ 1. The reasons for the occurrence of cappers is that after the upper punch has reached its maximum pressure and starts retracting.9'79
10
9
0 0
5
A
0
2 4 6 8 1 Compression Force (tons)
0
.8. many bonds are formed).. so that doubling the pressure would cause a doubling in the number.
In compression.7 Compression profile: The section AB is what is most often shownin ~ublications. then curvature occurs at lower pressure values.
.6
%
Y'
0.19)
There are three different types of bonding: (a) Weak forces. The effect is shown in Fig. If Heckel plots are carried out fortwo different mesh cuts of a solid. (€3)plot of a coarser fraction of the same substance.16)
The return of the upper punch causes a number of bonds N d to be destroyed. 24.
Nd =qP 52
(24.
(24. and (c) the third is solid bridges. e.8
* .7.Nf = B P
(24. the initial compression will crush particles and. reduces the significance of solidbridge bonding. van der Waals forces. if the bonding is by brittle fracture. then. (b) The second type is mechanical interlocking (Fuhrer. 1977). i.g. and this number will often be proportional to P to a power (e. or adding a dry binder.2
0. P =
as depicted in Fig. and hydrogen bonding (distance forces).0
0
8 10 Compression Pressure
2
4
6
I2
AthyHeckel plots of a finemesh fraction of a substance (A) fracturing by brittle fracture.18)
The total number of surviving bonds after ejection will be N=NJNd=pP+$ and the hardness would be proportional to this number. AthyHeckel plots of a finemesh fraction (B) fracturing by plastic deformation and (C) the same plot of a coarser fraction of the same compound. to $). but increases the importance of weak distance forces..o
0. electrostatic forces. therefore. is a function of applied force by a parabolic (or other power) relation with a m a x i ~ u m at
I dN/dP = 0 = # +P. 24.17) (24. The hardness.. as the
1. It is noted that the tensile strength is higherfor the ground NaC1. but that the critical capping pressure is less.g.I c v1
0. Adolphson et al.e. (1998) have shown (for the case of sodium chloride) that milling of the particles.4.4
0.
magnesium stearate) may be incorporated without sizable influence on tablet hardness (e. The second method isto dooptimization while the experimentation is ongoing. 24. If. hence. ingredient). some combination of methods is poste ther than testing three levels in a complete factorial..5%) concentration of magnesium stearate. 243). If. that several functional ingredients are necessary: rug substance (unless the formula is a placebo)
Filler (to obtain the desired fill weight) lida ant (at times).
The area of statistical optimization of formulae is outside the scope of this book.g.g. It is conventional with some investigators to do screening first (i. will fail to form tablets at all above a certain. The first method is facilitated by knowing a (at least phenornonological) relation (equation) connecting responses (disintegration.... furthermore. however. and two methods are used for this.. then there would be five high(+) levels and five low () levels. then fairly large amounts of lubricants (e. as shall beseen. whereas a substance bonding by plastic deformation (e.9 as comparison between A and B). In practice. there are five independent variables x. compression pressure). hydroxyapatite). fix the ingredients that will be used in the formula and then optimize the amounts). a midlevel were desired. ith as little as six ingredients. however. This is obviously an excessive amount and. furthermore.pressure is increased.g. furthermore. so that the number of preparations made fora “complete” factorial study would be 35 (Le. what is the effect of different lots of each raw material?). complete factorials. This is a classic mathematical and search method.g. hardness) to variables s..e. then five zero levels would also be required. on the other hand. 1996). some mention of subject is in order. 1. If. but it is seen in the foregoing. dicalcium phosphate dihydrate). so that the number of combinations would be 25 = 32. A fairly large body of literature has the been written on the subject of optimization and pharmaceutical applications (Schwartz. as in the example. 1th an i n c o ~ ~ Z e factorial. then the coarse fraction (now C ) will become parallel with the fine fraction (now If bonding is by brittle fracture. the substance bonds primarily by plastic deformation.s. if one were to test high and low levels (and the filler the q. one might test more in an incomplete factorial.g.. they offer a great number of possibilities for interactions and.
. one must recognize that formulations are multicomponent systems. other variables are of importance (e. however. lead to excessive amountsof experiments. fairly low (e. the evolutionary operations (EVOP) and the simplex method. the behavioral difference between the two fractions will disappear (shown in Fig. In general it is wise to have as few components as possible.
whereas the other two dir next set of experiments is then started at I in three “directions. In this case. 1978) may be usedto further optimize the composition. and theywilleach optimize at different values of the variables.” and it is seen that 3 going to C gives a better (and the best result).. treated in exacting optimization procedures is the following:Several properties (responses) are tested for. Often dissolution decreases with increasing hardness.
. In Fig. statistics are actually abandoned. Other methods have been published. In this one “triangulates” a further experimental scheme (Fig. but acceptable for both.g.then the response.
2
Percent Lactose Example of attainment of optimum conditions in a tablet formulation. and c. then. the experimentation is started with a percenta e of lactose corA. These could be. then a EVOP method (Box et al. The question. values than 15 may be used. The next set of experiments now start at C. 24. is to decide which of the parameters are most important.10. and a decision ismade to accept a formula that is optimum for neither hardness nor dissolution. but all give results inferior to the composition at C. usually. percentage cappers. so that C is deemedto be a local maximum.. so that the parameters Cy) may be maximized.
Hardness
3
1
ig.20) Three experiments for each variable. for instance. What is not. hardness and dissolution.10). and this will (a) make precision better and (b) testing for interactions. hardness. (1999) have describeda procedure for which all factors havebeen combined in multiple regression plots to obtain ranges of variables giving the best tablet. Once one has arrived at the formula. then. gives an improvement. y (e. so that optimum dissolution might occur at hardnesses that are not acceptable. The amount of lactose that optimizes hardness may not optimize dissolution rates. and experiments are made in three directions.bj. i et al. 24. in this case a total of 15 experiments will give ‘‘first” values of ai. dissolution halflife) may be fitted to a polynomial of the form: (24.
g. hardness) a = the gripping angle in roller compaction p = (a) angle ofreleaseinroller compaction. (b) crosssectional area of tablet
AL = surface area of a large particle
14s = surface area of the smaller particles
AproJ projected surface area (the cross section) of a small particle = a = constant it the (a) Heckel. (b) dose d = diameter of small spheres adhering onto a larger sphere f = fraction of the total area of two particles that is in actual contact.. amount of lactose) y = response variable (e. A H = heat of fusion h = thickness of tablet i = number of contact points in compression V iVd = number of bonds broken during decompression Nf = number of bonds formed rz = the number for “full coverage” of a large particle by small particles q = heat transfer rate P = pressure P(l) = pressure at which deformation is reversible P(2) = pressure at which deformation is irreversible P(3) = pressures at which plastic deformation or brittle fracture occurs R. = actual surface area ratio between binder and carrier R = ratio between the specific surface areas of small and the large particles S2 = specific surface area of large particle SI = specific surface area of small particle T = melting point t = length of time of heating VL = molar volume of a melt Vs = molar volume of a solid W = w I / w L = (a) weight ratio at complete coverage of large particle by small V particles.A = (a) surface area.. (b) CooperEaton equation b = constant it the (a) Heckel. = composition variable of the ith component (e. (b) proportionality factor between pressure and number of bonds @ = factor to ? to describe number of bonds v = Poisson’s ratio p = particle density ps = density of small particles pL = density of a large sphere
. (b) CooperEaton equation c = specific heat D = (a) diameter of a large sphere. (b) F = flow rate wL = mass of a large particle w1 = mass of small particles required to fill up the surface of the large particle X.g.
Carslaw HS. Conduction of Heat in Solids. Palagyi L (1978). Papadimitriou E. Tobyn MJ. J Pharrn Sci 69: 1. Chitosan. Nagai T. Drug Dev Ind Pharm 19:2335. Pietsch WB (1970). Nystrorn C (1996). Staniforth JN. Pharm Res 11:1233. Larhrib H. Fiihrer C (1977). Edge S (1998). Cook DT. Mattson S. Efentakis M.
. Nambu N (1984). London. ~ubinstein MH. In: Alderborn G. Wells JI (1997). Handbook of Powder Science and Technology. Pilpel N (1984). Alderbor? B. Lahrib H. Krycer I. FL. Li LC. Powder Technol 46:67. MJ. Larhrib H. Pharm Res 10:13363. Nystrbm C. In: Fayed ME. Chow YP (1981). Rubinstein MH. Lahrib H. Duberg M. Academic Press. Sebhatu T. ~ubinstein MH (1997b). Van Norstrand Reinhold. Cirunay N. Chowhan ZT (1980). Academic Press. Int J P h a m 171:31. Stubberud L. ~ o l l a n Celik M (1995). Int J harm 23:255. University of London.. Duberg M. Int J Pharrn 134: 179. Drug Dev Ind Pharm 19:2143. Int J Pharm 143:233. pp 2139. p 170. Department of Pharmacy. Poukavoos N. J Pharm Pharrnacol 36516. Rankell AS. Wells JI. Jaeger JC (1959). Oxford University Press. Celik M (1994). Malamataris S. Materials for Direct Compression. Adolfsson A. Higuchi T (1968). Chitin. New York. New York. Int J Pharm 15351. J. Nystrom C (1998). Amstrong NA. Duberg M. Rubinstein MH (1997a). Int J Pharm 114:23. Nochodchi A. Olsson H. Arwidson HG. Drug Dev Ind Pharrn 23:245. Hersey JA (1983a). Bolhuis CK. Stubberud L. Drug Dev Ind P h a m 25:571. Glazer M (1985). Nystrorn C (1998). Chowhan ZT. Graffner C (1996). Bi YX. PlaizierVercammen A (1997). Sheen P. eds. Mazur J. Int J Pharm 169:183. Sawayanagi V. i Nystrom C. Int J Pharnz 163:145. Olsson H. Larsson A.Adolfsen Cararnalla C. Kim S (1989). J Pharrn Sci 57:574. Characterization of molten filled hard gelatin capsules. Celik M (1993). Peck GE (1990a). LaboPharm Prob Technol 25:759. Nystrom C (1986). eds. Sjogren J (1982). Buyot JC (1995a). Peck GE (1993). Int J Pharm 147:199. Chelsea College. Pharm Pharmacol 42:272. Int J Pharm 10:209. Newton JM. Carstensen JT (1980). New York. pp 276267. Wells JI. Stotnicky J (1953).
A. J Pharm Sci 70: 11 34. J Pharm Phamacol 41 :149. Drug Dev Pharrn 16: 1491. Holleborn CE (1977). PhD dissertation. Yonezawa Y. Nystrom C (1985). Czech J Phys 3:225. J Pham Phamacol 29:247. Pope DG. Drug Dev Ind Pharm 15:401. Choulis NH (1992). Chatham SM (1985). Elamin AA. Mollan MJ. Carehill PG (1993). pp 419501. McCarthy BP. Hersey JA (198313). Forbes RT (1998). Marcel Dekker. and Related Enzymes. Powder Technol 53:3. Danjo K (1999). Nystrorn C. Guyot JC (1995b). J Pharrn Sci 67:1335. Joyce J. Chowhan ZT. Int J Pharm 160: 187. Xnt J Pharrn 147:187. Int J Pharm 118:191. ~o c h o d c h A. Orlando. Mollan MJ. Nystrorn C. Chowham ZT (1996). p 75. Int J Pharrn 86: 131. Palfrey LP (1989). Li LC. Pope DG. Drug Dev Ind P h a m 20:3131. Peck GI3 (1990b). Int J P h a m Technol Prod Manuf 617. Powder Technol 34:39. Sunada H. Solid Fhamaceutics: Mechanical Properties and Rate Phenomena. Ahlneck C (1994). Krycer I. Int J Pharrn 120:13. Bin Baie S. Lahrib H. Otten L.
Lerk CF (1988).
. Jackson IM. Mondero C. Nuessle NO (1992). Int J Pharm Techno1 Prod Manuf 2:1. Velasco V. Lerk CF. Pharm Acta Hevl 61:150. Yu HCM. Van Kamp HV. Vromans H. Choong YL (1986). P h a m Acta Helv 6 1:22 1986.
Alderborn 6 . Jim~nezCastellanos R (1997). New Uork. Int J Pfiarm 46:183.Upadrashta SM. Int J P h a m 152:111. MunozRuiz A. DeBoer AH. Wan LSC. Marcel Dekker. Katikaneni PR. Langridge JR (1981). Drug Dev Ind P h a m 18: 1701. Drug Dev Ind Pharm 15:801. Elsabbagh HM (1989). Lieahuen L (1986). Rubinstein MH. Nystrom C (1996). Bolhuis CK. Materials for Direct Compression. Wells JI.
AvicelsR is listed (for instance in the
. There are several other are disintegrants for the f o r ~ u l a t o rto choose from. when made.5. disintegrants are added to tablet formulations (and at times to hardshell capsule formulas as well).1.c. must (in most cases) disintegrate to afford acceptable dissolution rates. Avicel P~lO1. There is often a direct correlation between disintegration times and dissolution rate constants (Carstensen et al. It is seen that ExplotabR and AcDiSolR. ~isintegrants work by swelling and causing a high degree of stress on the tablet. are superior to the three other disi~tegrants.the socalled superdisintegrants. and Avicel PI1102 have been compared by Chebli and Cartilier (1998). This will be covered in more detail in this chapter. To this end. Explotab. 1980a. CompressionCoated and ~ultipleLayer Tablets Symbols eferences ecom~ended Reading 432 435
436
437 438
isintegration and dissolution are crucial properties of tablets.
Tablets. and an extract of their results is reconstructed in Table 25. 1995). on Co~mon dis~nte~rants starch and modified starches.b.. Disintegrants also allow formation of channels that allow water (or other liquid) ~ e r ~ e a t i into the tablet. 1978.427 429
43 1
issolution of Drugs from Tablets 25. who also investigated crosslinked cellulose (CLC). AcDiSol.
and (b) they swell on contact with water. This means that the manner in which these disintegrants work is not by expansion. measuring is the volumes of sedimentation V .4 24. 1986)..~
The tablets contain 0. Avicel PH102 (both microcrystalline celluloses). This is performed by centrifuging suspensions of the disintegrant in both water and in paraffin and. 1986.
. by reducing the contact angle..8 75
Avicei PHlOl
___ __
Avicel PH102
> 1200
266 16
> 1080
39. 1989.5% magnesium stearate.1 460
Explotab 15 15. Tygon).2. Dicalcium phosphate dihydrate. Results for the CC samples and CLC (Chebli and Cartilier. 1998). Chebli and Cartilier (1998) have inv ated a crosslinked cellulose (CLC) as disintegrant and compared it with Avicel 01. and this slows down the disintegration. f . sically. o water uptake is also of importance (e. 1998) are listed in Table 25.. so that using these at a 2% level generally suffices to obtain maximum disintegration efficiency. but simply. This assembly is arranged vertically. but it has no solid planes on which to exert its force.. it is approximately equal to unity.
~ a ~ of ~ x c~ i eo tor. and AcDiSolR give approximately the same dison times at 2 and 30/0 levels. Some investigators check the sweZZi~g voZu~e.The ratio V w / V p denoted the s ~ e Z Z i ~ g capacity (Chebli and Cartilier.6 142
AcDiSol 7. A tablet is positioned in contact with the fritted disk. both ExplotabR. so that water will draw into the tablet. Poukavoos and Peck.g. and Vp. Van Kamp et al. The disintegrants work in that (a) they facilitate the penetration of liquid into the tablet (e. This is performed in an apparatus that consists of a glassfritted disk filter that connects to a 2mL pipette by way of tubing (e. then.ter
le Comparison of Disintegration Times of Direct Compression Tabletsa Containing Six Different Disintegrants at the 5% Level
Filler Erncompress Lactose 100mesh spraydried Lactose
a
CLCC25 15 16. in ~ ~ ~ n s o Lieberman et al.. AcdiSol.7
> 1200
__ _ _ _ _ _ ~ . but its disintegrating power is much less than the others. by aiding the filling of the void space with disintegrating liquid. This causes the disintegrant to swell. Chebli and Gartilier. 1989) as a disintegrant.. and Explotab. and for microcrystalline celluloses (Avicels).g. 1998). 1993. The reason spraydried lactose is more “difficult” to make disintegrate is that it is more soluble (owing to its amorphicity) than crystalline lactose dihydrate.g. The porosity of the tablet will affect the w~teruptakerates an and Choong. The uptake is then recorded as a function of time as the water level in the pipette changes.
Couvreur (1975) showed that unmodified cornstarch reduced the wetting angle.
Disintegration is a function of two factors: (a) first the disintegration medium (e. 1966. but (b) too low a pressure will allow the pores to be so large that the disinte~rant. when it swells. g is the gravitational acceleration.8 3. namely. If one considers a tablet to be porous and to possess one pore radius r (the average. and p is the density of the liquid.95 0.2. A first step in modeling would be to imagine a tablet simply d’ ating at constant volume into particles of the same size. and force the tablet apart. (25. 1950) will give rise to graphs of the type showii in Fig. an approximation).2)
where y is interfacial tension. where the yasymptote is assumed to be and 21 units). so that penetration rate is lowered. disintegration as a function of pressure (Berry and Ridout.. and the liquid must be as small as possible. and Q is a constant given by
= (2y cos[QI]/r) .isi
Ta
Excipient Avieel PHI01 Avicel PHI02 CLCC25
Rate of Water Uptake by Tablet as a Function of Time Swelling capacity
Vwater / wparaffin
lo3 x initial uptake rate of water (mL/s)
5.b..9 1. ~ a r s t e ~ s e n al. the wetting of the tablet..g p sin[a]
(25. One additional factor is at work. water) must penetrate the tablet.1975) L2 = { r ycos[qb]/2q}t (25.5
0. et 1978a. i.1. Hence. will not exert the desired pressure on the tablet.92
Source: Chebli and Cartilier.c). 1976. Q! is the angle between liquid and capillary wall. q is the viscosity of the disinteg~ation medium. 25. 25. Integration of Eq.95 0..2) gives (Couvr~ur. The effect of tableting pressure (Berry and Ridout.2) has been dropped. 1998. that is. The last part of this curve is often presented in sigmaminus functional form ( K e n ~ o n Swintosky. the surface must first “wet” SO the contact angle QI between the solid surface. 19’75) will be dL/dt = Q~2/8qL (25.e. and aided the flow of liquid into the pore space. then the rate of penetration of the liquid (Nogami et al. Couvreur. (25.g. This is not correct ( nd
*
. 1958) (Fig. ost tablets swell and then disintegrate (~arstensen. 1950) is such that (a) increased pressure decreases the pore size. and (b) then the disintegrant must swell.3)
where the last term in Eq.1)
where L is the length penetrated at time t.
~ t
(25.3. one may assume that the tablet consists of To particles. (1978a. 25.The weight (mass) remaining at time t therefore.1
.5452 .e.b. at time t .22
r
20
18
16
14
12 10
0 2 4 6 Tableting Pressure (Arbitrary
8 8
1 0 1 2
Units)
Schematic figure of the effect of tableting pressure on disintegration time.5)
if the tablet is nonswelling. and that these simply “fall off.. and the density of the granula) T = Toe@ (25. the weight of the tablet is proportional to the number of nondisintegrated particles T .000
4
2
8 10 Tableting Pressure
12
Logarithmic presentation of the end data in Fig.
hodes.4) where Tois the initial number of particles.0. be In this view. is given by
N =~ o e . but with the simpler model.” and that happens semilogarithmicallyin time (i. as shown in and 25. and q is a rate constant.c). That this is correct for tablets that swell minimally was demonstrated experimentally by Carstensen et al. the mathematics to follow is sufficiently simpli~ed to tested. because the resulting particles have definite distributions.69315~ R”2 = 1. 1975). A model by Kitamori and Shimamoto (1976) and considers the number of particles i that have dislodged at time t by V
y =
5.
1.Wett~ng
Penetration
Swelling+ Disintegration
More ~ i s i n t e g r a ~ e d Granules (May be Same Size or Different Size Dis~nt~~rated Granules
Schematic of ~ i s i n t e ~ r a t of a tablet. 25. stepwise.)
(25.1.4. is shown in Fig.c. (Data from Carstensen.72183 .
.
The previous section dealt with ~ i s i n t e ~ r a t i o n ~ the scheme by which disintegraand tion takes place. i~n
.0.R*2 Y = .6)
where tg is disintegration time and m is a constant.995
Time (min)
Weight of a di~integratingtablet as a function of time represented s e ~ ~ 1 o g a ~ t ~ mically.6 r
= 0. the tablet has to wet. 1978a.8761e2x 0. First.
. 1989.Then. Colombo et al. Then. 1995. and this force is proportional to its volume which.which decreases with a firstorder rate constant of k.” ence.C reaction (Carstensen. The weightof the dislodged granules W d The differential equations that this dictates are as follows: (25. then at time t the situation will be given t is divided into N intervals. The amount of particles IT released at a given time is given by
IT = IT.12)
If a tablet starts disintegrating at time zero. external surface area)..exp(kt))
(25.9) These equations areidentical with the equations foran A. the liquid has to penetrate.8) (25. The pressure exerted is proportional to the added weight. the tablet disintegrates Several authors (Timmermans et al. 1995). k is the release rate constant.10) The pressure P is proportional to W w . 3. and mo is the mass of the particle as it is dislodged from the tablet. 2. Caramella et al. Catellani et al. and k d . The weight of the dry (or dry partof the) tablet Ws. the number of particles for
. the disintegrant swells and the volume (mass) of the tablet increases 4. 1978. is proportional to its mass W . 1987) have described equipment with whichit is possible to measure the pressure the tablet exerts on confining plates.m = mo[1 . 1994. 1981. Curvefitting of the data will provide the values of k. and it will be assumed in the following that the granules are all of the same size and that the diff~sional release from the granule is given by
mo . whichis associated with an absorption rate constant kd 3. Faroongsamg and Peck. and the solution for the Eq. It is assumed that release occurs only from the granules. that the curve for P versus time will have a so maximum. 1984.
It is visualized that the tablet disintegrates by the foregoing equations.11)
where m is the mass of the individual granule at time t.8) of the equation system is (25. (25. inturn.. not from the tablet itself (which has a much smaller specific. Theweightof the adsorbedmoisture W. which would be W.exp(qt)) 1
(25. 1984.7) (25. The equation for the weight may be divided into 1.. at this time... and in the first interval has been &‘in existence” t min from its “birth.2.Then.(.
tN/N) time units.exp(qt).(t/N).
Time of birth Number of particles Amount dissolved
When these are summed. The amount of drug released from them is computed as follows: The amounts released during the first two and the last two intervals would be the number of particles released in each interval (tJ times the amount dissolved (mi) at time t after they had been allowed to dissolve for t. .13) where
x.14)
. The two last figures are t/N and zero. the total amount released is
(25.t(N . Hence. .5
Schematic for disintegration~issolution event. = n.
'
produced is To(l .
N3 N1
TO(1exp(qZt/N))born Mo( 1 exp(kZt/N))drug released
4
.l)/N) and (t ./N
(25. (t . t .TO( exp(qt))born 1 Mo( 1exp(kt))drug released
LLL
0 1 2
N2 Time N
1 ' 1 .
If either isi integration or dissolution are ratedetermining. note Fig.
. the socalled lagtime) sion of drug from the granules
In such a case.(1ld .exp[ktxj)dx
.exp[qtxjl)[l
1
. To convert this to an integral.{ +
1
1
exp[(q
+ k)tx]
.{ l / ( q 3.16) quat ti on (25.15) is a rather complicated equation.exp[ktx])dx
= tomolx .exp[qtx]. then the e~uations become ~ u more ~ a n a ~ e ~ h able. (25. of
.6 in which a graph of the sum is shown. the equation simply becomes:
dx x=o
x= 1
Sche~atlc the trace of Eq.{ 1 . the intervals become dx and the integral becomes
. 25.13).M = torno
J.{ 4% + w ) exp[(iq + 4 4 + { l f g )exP[qtxl
(25.{ I / ( q + w } exP[(q + W l
(f/q)t
+ (1/q)texp[qtx]
= tomot{J
+ ( 1 l t ) . As N goes towards infinity. the disintegration is rapid.1~ = torno
S.w
+( 1/ ( k t )exp[ktxl
} exp[(q
+ Ic)tx] + { l / q t }exp[qtx] +
I:. If.( l / k ) t (l/(~)texp[ktx])
= torno{ 1 + { f/(q
+ k ) ]t .and ranges from zero to unity. as ismost often true. and then the process becomes
J
= 1/(q
+ k ) .( I l k )
pid disintegration ( t = ti.15)
+ (1/(W exPl~~xl}
where (25.
7) (Zoglio and Carst In this approach a tablet within a tablet within a tablet is created. and this can be done.20) where A 4 1 is the original amount in the layer. is approximately constant (and equal the average between the surface area on the “outside” and “inside” of the layer. A .17) (25.~ ( 3 ) ( A 3 ) ~ (25.
Controlledrelease of drugs from dosage forms will discussed be in detail in owever. is best discussed at this point in the text (Fig. sustainedrelease by compression coating. ence. which are semipermeable both to the dissolving liquid.K(2)A(2)t = A42
and A4 = M3 .
(25. for instance. which intersects In time t = t. one such principle. This is now manipulated in such a fashion that + M3 is equal to the required dose. Similarly.. Mars. for the first layer) would be /dt = K(l)A(l) (25.17) and (25.A4oA4=0
O<t<t. Verhoeven et al. Another approach is to make the outer coat a restraining coat.18)
= ln[A4/A4~~ k(t . The release of drug from
+
~~
fl
Rate Constant K(I). and to the drug substance. an approxi~ate linear release.
. 1974).18) then a straightline ensues. and if the data are plott Eqs. Area A(1)
Approximate linear release rates from tricoated tablet.21). 4. 25.22) that is. using the same nomenclature
A4(25.t J
ost dissolution equations follow this path. layer is not equally thick.. the rate equation (e. and 8 h) are within the desired intervals.~ ( l ) A ( l ) t 4 (25.19) where K(1) is the dissolution rate constant and A(1) is the average layer of the the outside (first) coat.g. 1983. by using polymers. once it dissolves in the penetrated liquid (Conte et al. the surface area. t > t.. (25. for the other two layers. and the kc values and the layer weights are manipulated in such a fashion that the amount released at various points in time (1.. 1989. This integrates to A = A41 .
However. (1967) employed soluble porosity modifiers to the (otherwise waterinsoluble)compression coat so that. There are limitations to this (Le.K(3) = the rate constants for each of the layers in a
tricoated tablet L = length of penetration ashbu burn equation) M = amount not dissolved or disintegrated Mo = original amount w1 = exponent in Kitamori equation N = tablets per second or number of disintegrated particles or number of time intervals Q = constant ashbu burn equation) q. the effect of the compression pressure on the physical characteristics of the polymer).4 3 ) = surface areas of the three parts of a tricoated tablet g = gravitational acceleration k = dissolution rate constant kd = disintegration rate constal~t k . = water uptake rate constant k.1988) to explain the development of pore clusters and conducting channels that would span the compression coat. and this was used(Siegel. these would dissolve and create a porosity network allst strand and ~ k m a n 1983. s = (a) fraction of time hopper stays over die. ~tauffer (1985) applied percolation theory to the problem. Thombre et al. 1985. (c) disinte~rationrate constant r = pore radius R = radius of disintegrant particle.
A = A(1). T = number of particles in a tablet To= initial number of particles in a tablet t = time tg = disintegration time PVs = weight (mass) of unhydrated tablet Ww= mass of hydration part of a tablet Wd = mass of tablet disintegrated a = the angle between liquid and capillary wall y = interfacial tension # = contact angle
. (b) or exponent in disintegration equation. Shivanand and Sprockel (1998) have described a compressioncoated tablet. 1998). in some cases. its surface characteristics. did not behave exactly as predicted. onexposure to dissolution liquid. indeed.. A@). this is what Shivanand and Sprockel (1998) found.these is a function of such parameters as the amount of polymer. E((2). and particularly with sorbitol. It would be expected that the release would be linear after a lag time and.. for which the coating material was CAB and the porosity modifier was sodium chloride. 1989). Zenther et at.. and Fryklof et al. = disappearance rate constant of solid part of a tablet E( = E((l). and its compressibility (Shivanand and Prockel. the pore network.
New York. Khan KA. Ostholm GIV (1967). Controlled Release Drugs. Colombo P. Zoglio MA. Kitamori N. Drug Dev Ind Pharm 15:401. Philadelp~ia. In: M. LaboPharma Probl Technol 339: 11 5. ed. Colombo P (1989). Lievin V. Chebli C. Chem Pharm Bull 24:1789. De Boer AH. Drug Dev Pharm 7: 135. Kennon L. Tawashi R (1993). Shimamoto T (1976). Int J Pharm 171: 101. Pharm Acta 61:221986. J P h a m Sci 651837. Rork GS. Catellani PL. Siege1 RA (1988). Conte U. Kitamori N. Cartilier L (1998). J Pharm Sci 67: 1303.q
==
viscostity (of disintegration medium)
p == density
Berry H. GuyotHe~ann A . J Controlled Release10:205. Sheridan J (1980b). University Catholique de Louvain. Poukavoos N. Int J Pharm 51:63. Junginger HE (1989). Colombo P. J Pharm Sci 73:701. Columbo P. Lieahuen L (1986). Bellotti A. Peck GE (1993). Chowhan ZT (1980a).Danhof M. C. Dutch patent 7313696. Caram~lla Conte U. Faroongsamg D. Drug Dev Ind Pharm 13:2111. Blessel KW. Lnt J Pharm 167:83.
. Caramella C. La Manna A. Belgium. Tour6 P. New York. Wright JL. Chem Pharm Bull 14: 152. Himmelstein KJ (1989). Compositie met vertraagde afgifte. Predella P. Lerk CF. Geddo M. J Membr Sci 40:279. J Controlled Release 2:217. Lai TYF. Pharm Acta Helv 61:150. Caramella C. Choang YL (1986). Marcel Dekker. Conte U. Sheen P. Van Kamp HV. Sheridan J (1978). Sandell E. paper 79). Nogami H. Moes AJ (1995). La Manna A (1984). Mars P (1974). Carstensen JT. Sheridan J (1980). PressCoated Systems for Drug Release Control. p 87. Ridout CW (1950). Kim S (1989). Carstensen JT. Int J Pharm 5:157. J Pharm Sci 69:290. Ringard J (1981). patent 3317394. Kallstrand G. STP Pharma Sci 5: 110. Caramella C. Carstensen JT (1985). Gazzaniga A. Colombo P. Drug Dev Ind Pharm 6569. S. Drug Dev Ind Pharm 20:1777. Medicinal tablet and a method for its preparation. Sprock 988). Verhoeven J. Plenum. Rhodes CT (1975). LaManna A (1984). Wan LSC.Tayor & Francis. J Am Pharm Assoc Sci Ed 47:397. Caramella C. Shivanand P. Stauffer D (1985). Swintosky JV (1958). J Pharm Phamacol 2:619. Conte U. Kothari R. U . othari R. Fryklof LE. Abstracts. Timmermans J. Schutte SC. Int J Pharm Technol Prod Manuf 5:1. Uchida H (1966). IntontoPercolationTheory. p 25. Thombre AC. Zentner GM. La Manna A (1983). Bolhuis GK. Peck GE (1994). J Pharm Sci 72:772. Carstensen JT (1995). Cartilier L. Nagai T. Drug Stab~lityPrinciples and Practices. Peschier LJC. J Pharm Sci 67:1436. of VCH. Dissertation. Himmelstein KJ (1985). Docteur Sciences Pharmaceutiques. Zenther GM. Ekman B (1983). Colombo P. Carstensen JT. Carstensen JT (1976). Iga K (1978). Couvreur P (1975). Am Pharm Assoc Acad Pharm Sci Natl Meeting 21: (abst 6(2). Conte U. La Manna A (1987). Prasad VK. New York. STP Pharma Sci 3:213. Pharm Res 10:13363.
Modern ~ h a r ~ a c e u t i c s . Marcel Dekker. Lachman L. vols 1. 2. and 3. In: Banker GS. Lieberman HA. New York. eds.Rudnic EM. Kottke MK (1989).
. Pharmaceutical Dosage Forms: Tablets. Rhodes CT. Schwartz JB (1989). pp 348354. New York. 2nd arcel Dekker.
Cellulose and Cellulose 26.8.4.2.5.1 1. Intrinsic Viscosity olyethylene Glycols 26.12. ““Older” Polymers
444
C ) and Hydroxypropyl Methylcellulose (H
droxypropyl Cellulose (H 26. Cellulose Acetate
444 445 446 446 448 449 49 450 45 1 45 1
Polymers constitute a special group of excipients. and their primeuses following: 1.olecular Weights of Polymers 26. Diffusion Through Films 26. Ethyl Cellulose(EC) 26. Granulatin~ agents
are the
.14. Intrinsic Viscosity and
40 440 44 1 44 1 442 444 444
26.
Filmcoating materials 3. and T is absolute temperature (OK). at low concentrations. If the concentration is expressedas C* in grams per mole of solvent. The ideal gas law states that
PV = nRT
(26. (b) hydroxypropyl methylcellulose (Methocel. then the slope should have been T . n is number of moles. and (e) hydroxypropylcellulose The manner in which the molecular weight is determined is by preparing dilute solutions of it in an appropriatesolvent (for HPMC and HPC. in its most basic form. whereas in this chapter. If C had been in moles per mole of solvent. R is the gas constant.pter
2.
The important polymers in sustained release are (a) ethyl cellulose (Ethocel. P is pressure. easier means of obtaining it are important.1) becomes
rI = RTC
(26. Releasesustaining excipients The latter will be the subject of Chap. so that Eq.1).2)
where rI is the osmotic pressure. and the ratio between the slope found (by using e*) RT and allow calc~lation the molecular weight.
m)
L
Schematic for the definition of viscosity. HPMC). is visualized by an infinitely wide vessel containing the test liquid and providing a moving plate above a stationary bottom (Fig.1)
where V is volume.
The definition of viscosity. water) and measuring the osmotic pressure. To move the plate at a given speed v. and this force is inverselyproportional to the distance of the movable plate from the
Area A (sq.
. (26. EC). the properties of the polymers of interest will be discussed. in which the concentration C of solute. The ideal gas law also applies to very dilute solutions. 26. then. of ause d e t e ~ i n a t i o n molecular weight becomes a specification in lots of of polymer. is n / V . 29. the osmotic pressure will become linear with concentratio C*. requires a given force F .
that is. em2).
The notation for polyoxyethylene glycols is PEG. They vary in consistency from liquids to solids. Viscosity is often measured by monitoring the shear rate as a function of the shear stress. The molecular weights range from 190 to 20. = 1 and h = 1. if A. ~n. and for the Bingham body.1
Molecular Notation PEG 400 PEG 1000 PEG 4000 PEG 6000
Characteristics of PolyethyleneGlycols weight 380420 9501050 3004800 54006600 State at 25°C Liquid Solid Solid Solid
. there is a yield value. This is the viscosity divided by the Often the k i n e ~ a t i c density of the liquid. whichiscovered under direct compression) and derivatives ofcellulose
able26.000. then the liquid is denoted n e ~ t o ~ iIf not. viscosity is employed. first two cases. Polymers in concentrations of less than 1% v/v are usually newtonian. and they are usually referred to as polyethylene glycols.3)
The unit for this is stoke or centistoke. then it may be either thi~otropic. below which the liquid does not move under stress. d~latant. The notation PEG is followed by a number indicating an approximation of the molecular weight. The viscosity of water at room temperature is about 1 cp. olecular weights of polymers may also be obtained by gel permeation chromatography. The force required to move the plate is equal to the viscosity of the liquid.
Many of the useful polymers are either cellulose itself (microcrystalline cellulose [MCC].and the most common unit is the centipoise (cp). the shear rate is a power function of the shear stress. and if such a plot is linear.stationary bottom. the definition of intrinsic viscosity is (26.1. ~ i n g h a ~ In the or a body. Viscosity inthe cgssystem ismeasured in poise (P). ~seudoplastic. It is also directly proportional to the surface area of the movable plate (A.2) where qo is the viscosity of the pure solvent.
V = VlP
(26. The most common grades are shown in Table 26. For the pseudoplastic liquids. the viscosity changes when the liquid is shaken (or otherwise subjected to stress).
hw. is often obtained by way of the socalled Mark.92 0. 1982. cellulosic saccharide unit.7) where K is a function of the polymer and cr is an exponent that is a function of the shape of the molecule. hydrogen or methyl). 1959
Source: Rawe. as binding agents.55 0. which states that (26.. Powdered cellulose is marketed as Solkafloc or Elcema.g.2 is reconstructed from the cited publication. molecules with a degree of polymerization of ni. The derivatives’ primary use is as tablet (or capsule) excipients. (26.
~~~
Erl =
ouwink Constants for Derivatives of Cellulose. ’Cl (e.are used as filmcoating materials. and then reacting it with the appropriate chloride. 1958 Greminger and
0. methyl chloride): ’Cl + [ROR’]. A microcrystalline grade ( CC) is used as a direct compression excipient. the n.6 99.8 29.1963 Wirick.4) Placing the cellulosein alkaline solution causessome deterioration of the cellulose.8 1
1257 18126 320 414 414
Savage... but often used. is given by
M B= ~ h n i M ~ ~ ~ ~ n ~
+
(26.
.2
0
x MW range
Ref Neely. is not a single figure. so that there will be a host of cellulose sodium salts with different chain nce.1970 Brown.in alkali to form [RONa].4 11.
Cellulosic polymers are ([ROR’].. Theyare often obtained by placingcellulose [RO . to use the viscosityof a given concentration of polymer.5)
wheref.10 CC14
C6w6
lom3 g’ dL
316 2. and R’ the substituent. a y the same token.ouwink equation. for instance (as in the Dow Chemical Company’s bulletins on P M C ~ using a 2% . e.but co ‘ cular weight of one of the molecular segments of number molecular average weight M.g. is the number fraction of molecules having degree of polymerization of ni. and as excipients imparting sustained action to dosage forms. A less exacting. the weight m~lecular average weight is given by using the weight of fraction. in [ROR’]. Table 26. manner of correlation is rather than use the intrinsic viscosity. Used in SustainedRelease Formulations and in Film Coating Polymer Solvent MC HPC EC Water Ethanol 1.89 0.6) The value of M . where n is the degree of polymerization. Rowe (1982) has listed the values of K and cr for a series of polymers. nNaC1 (26.
aqueous solution and correlating this with the molecular weight. For instance. Interscience. and these two quantities relate to one another by Eq.5 could be inserted in the equation.313 8. 26. = 0.8: the table figures (the last two columns) are plotted in this fashion.
j3= l/a
(26.500 4.942
Molecular weight calculated from osmotic pressure as concentration approaches zero.p 504. Data from Dow Manuals.31382~.714 4. 26.12 x lo4 compared with 7. New of a y York. and the value of ln[MJ found to be y = 5. The designation method i s described in ASTM monographs D134772 and D236372.258 3.
Viscosity of 20/0 Solutions Viscosity 20/0 solution 10 40 100 400 1.605 5.3 lists the viscosity of grades of HPMC as a function of the number average molecular weight of the polymer M.689 4.
.000 8. M.l = Blnlrll that is.454 4.8)
(26.143 4. 1965.303 3.000 15. Fig. the
= lU*[q*]4
owe states that
This equation is very useful to the formulator since it can be used to predict molecular weight of samples of known nominal viscosity. are presented in the form ln[M. OOO] = 11. an M . .616 9. It is interesting to note that the molecular weight for the N100 sample (nominal viscosity 88 mPasTable 2) predicted using this equation was 8.787 4..700 4.8363 0.e.852 ln[MW/ 100O’J 2. Freque~tly nominal viscosity (q*) is used and is defined as: the
+
17* = ( q / q O ) 1 
(26.294 8.1 1)
In this case (Rowe 19821. John Wiley & Sons. p 504).10) arkHouwink equation takes the form (26. if S2 an C had a (2% aqueous) viscosity of lOO?OOO? then.44523.565 2. Source: E n c y c ~ o ~ e d ~ Polymer Science and T e c ~ ~ o l o g3..999. 10~ 13 20 26 41 63 86 110 120 140 1nhI 2. or Mtt = 232 (i.996 3. value of 232?000)could be estimated.9)
Table 26.000 19.987 9. if the value ln[100.991 7.1 x lo4 measured by gel permeation chromato~raphy .000 Number Avg Mol Wt.2 The leastsquares fit line is y = 1.
C H 2 60
R2
R1 C
1711
CH2
eo
R4
C CH2CO
C

eo
(26. It was mentioned in an earlier the chapter that HPMC maybeused as a granulating agent (e. by now.8). it is. largely abandoned in new formulations. as a particle size
. in the following
The abbreviated namesin parentheses in the heading willbeused often in the following.g.. They are known. hydros genated castor oil) have been used. as they have been in text precedingthis. but a d i f f e r ~ set of polymers are nowadays ~t most often considered in new product formulation.
The original sustainedrelease product patented by Smith Kline & French in 1950.12)
where the substituents determine the properties of the polymer. as ~ o v i ~ or ~ e is used as a tablet binder and o PVP as a rigid matrix in sustainedrelease preparations. as ~ ~ ~ r ~ g i t ~ . Other c o ~ p o u n d (e+.3 plotted according to Eq.F"3
0 0 0
2
4
6 12 8 In[Viscocity]
10
Data in Table 26. commercially. also known.
Polymethacrylates are substituted polymers of acrylic acid (anhydride) and have the general formula:
R1 R2
1
C . used shellac as a sustaining agent.
Polyvinylpyrrolidone. and these will be dealt with. (26.. Owing to its tendency to polymerize on storage.
able
Various Grades of Methyl and Hydroxymethyl Propylcellulose
Methylcellulose (Methocel). Schor (198 1) speculated that the slowerrelease rate arises from a decreased rate of swelling or a lower water solubility resulting from hydrogenbonding interaction between the carboxyl and the carbonyl groups that had been subjected to both hydrolysis and oxidation. In the invention as little as 0. The viscosities are from 15 to 100. Metolose SH is the tradename used by ShinEtsu.4.000 cp 15. but the most interesting use of the polymers is in creating sustainedrelease dosage forms. 1981 s. Ltd.000. The molecular weights were higher at the time than those used in the past and he used an amount of modified HPMC less than about onethird of the weight of the sustainedrelease dosage form. hydrolyzed HPMC (Methocel ESO) by exposing it to high humidity. S..01 ~ F M C USF 2208 (ISseries) . they are characterized by the viscosity of a ~ ~ ~ C 2% aqueous solution.000 cp 100.g.000 cp 18.000 to 190.
~
This is often used in combination with Ethocel. and sometimes acts as a plasticizer for the Ethocel. in a series of patents (Schor. USP 2905 (Fseries) Methocel. Ltd. The data in this table are reconstructed from data in Methocel ~ulletin. As mentioned.4. In all cases the carrier material was thoroughly intermixed with the medicament which was either powder or in solution.5% could be present. USP (Aseries) Methocel A4M Premium XD30345.000.g. Japan.000 cp
4. patent 3. F4M Premium
4. with a methoxyl content of 1624 wt%.790 employs up to 25% moisture and then obtains sustained release by control~ingthe degree of compression. and these viscosities correspond to number average molecular weights from 10. Although the actual mechanism is unknown. and there are various grades as shown in Table 26. Recent grades have molecular weights as high as 240. A iven trade name is an HPMC with a single viscosity (e. USP 2910 (Eseries) Methocel E4M Premium HPMC. There are manufacturers other than Dow Chemical Company (e. U. 1978.enlarger and as a binder). England).000 cp
. Formulating Su~tained Release P ~ ~ r ~ a c e u t iProducts cal it^ ~ ~1982.4. 4000 cp. HPMC and HPC are marketed by Dow Chemical Company. 100 cp.000. further improved the carrier base utilizing a grade HPMC with e following characteristics: Methocel K4N and K15M.00 (IS100M Premium) HPMC.000 cp. and in one case K100.870. Methocel K4M Premium Methocel K15M Premium XI)30018.000 cp 4.000 cp
4. 1979. Attempts have been made to m o ~ HPMC to tailormake it to certain sus~ y tainedrelease r e ~ u i r e ~ e n tSchor. 1982). and so on as shown in Table 26. M > 50.. Celanese. as shown in Table 26.
. In that situation.8 Yo 2.3 8
47. the hardness and the softening In point show a minimum when the ethoxyl content is 48%. There are various degrees of ethylation.54. 1991). the con cent ratio^ of
Etkoxy Content of Various Ethocel Grades Grade
~
~
G 44.or delayedaction systems (Rowe. Table 26. giving products with ent viscosities. because it accumulates in the solution on evaporation and forms a spongy.
cellulose (EG) is cellulose that is ethylated (i.546.5 Yo 2.53
Ethoxyl (E) E/Anydroglucose unit
Source: Hercules Co. and flexibility. of toluene/ethanol. this range.. C are often used in blends with EC in the formulation of microencapsulationbased sustained. If different types are compared.282.
45.5 3
. or caffeine) was reported.28
I (
N
Y
> 49% > 2. doublelayer film with a drugcontaining layer that incorporates the drug in HPC attached to a film containing EC with different percentages of either polyethylene glycol (PG) or HPC. forms ethyl ether bonds) at the roups in cellulose. Consider a microcapsule in which liquid has permeated into the interior and has become saturated with drug substance.Samuelov et al.e.2 12. barbital. but requires a 4 : 1 toluene/ethanol as a solvent when the ethoxyl content is 485 1%. the diffusion of water and oxygen through polymers is of importance. as is
In sustainedreleaseproducts. The extent moisture adsorption of ethyl cellulose goes down as the ethoxyl content goes up in the range of 435 1YO. (1979) have described a laminated. Ethocel is particularly useful in microencapsulation. by weight. This latter is determined in a solvent consisting of a 4 : 1 ratio. salicylic acid. 1966. it is important to keep water out of the solvent system. Alam and Eichel (1982) have described sustainedrelease pharmaceutical formulations of indop~ofen in with ethyl cellulose. using ethyl cellulose samples dried for 30 min at 100°C. 1980).545.5 shows the ethoxyl (E) contents and the amount of ethoxyl/glucosesunit (ElA) of various grades from Hercules Company. Ethyl cellulose is soluble in ethanol when the ethoxyl content is 4549%. porous film (Arwidsson and Johansson. Zeroorder release of the drug substances (tripelenamine. percentage elongation at rupture.422.9OYo 2. and the yardstick is the viscosity of a 5% solution in 4 : 1 toluene/ethanol then the following properties increase with increasing viscosity: tensile strength. As a film former.
and h is the film thickness.18)
hen there is no saturation (i.16) where I is the permeability and is given by 3 rr=QD (26. This inserted in Eq. at sink conditions.13) integrates to
=( ~ ~ S 2 ~ h ) t
(26. A is the surface area of the film.such as describedin the
h
Schematic of diffusion of a drug through a film. under sink ~ o n d i t i ~ n s the exit side and on saturation on the entry side of the ~ e ~ ~ r abe linear. when no solid drug phase is left on the entry en S I is not only smaller than saturation. and the co~cent~ationthe liquid on the exit side (to the “right”) is in assumed to be zero when sink conditions prevail.e.15) ient. The solution is saturated on the entry side (to the “left”). (26. SO that Fick’s first law will require that dM/dt = DAS2/h (26. !. Eq. and the slope .
B = (A ns1/ h )
(26.. but will also change with time. M is the amount released at time t. r the foregoing equations it is possible to calculate the value of the permen nstant from plots according to Eq.13) where S2 is the concentration of drug in the film on the donor side.16). The concentration at the exit side of the film. (26. Figure 26.drug in the liquid (on the “left) in Fig. 26.14)
The value of the concentration S2 of the drug in the polymer on the entry side is (26. will be zero.
.14) gives (26.?should be ne. (26.17)
A plot of M versus t should therefore.4 is constructed from data ublished by Donbrow and Friedman (1974).3 will be its solubility S1 in the liquid. D is the diffusion coefficient. onbrow and Friedman (1974) established that mass transfer through ethyl celluloseis controlled by a solubility diffusion process.
as the name implies. 1974. This method is. when used for films.)
foregoing. (1986) used a torsional braid balance to study the glass transition temperatures of mixtures. HPC when used as an additive. was not leachable. This is equivalent to a reduction in the thickness of the film. that the release profile of tripelennamine HCl through PEGethyl cellulose films followed a Higuchi law (Higuchi. although the film may erode more readily. The film is left with “holes” which will allow diffusion of drug from the interior of a microcapsule. Glass transition temperatures are usually determined by differential scanning calorimetry (DSC). then there is mutual insolubility between film former and additive. 1961). however. Plasticizers are usually used in applications to make the film more pliable and less likely to break. These authors mentioned that. In contact with water. If not. so that it lowers the temperature at which the plasticizer will be in a rubbery phase. apparently because the PEG dissolves in the aqueous intrusion phase and. this latter acts as a plasticizer. much more sensitive that DSC. who found. With inclusion of PEG ~000 (which acts as a plasticizer) there is an increase in permeability. containing either HPC or PEG as a coating material for potassium chloride has been quite successful. when PEG 4000 is used as an additive to the film. A plasticizer works by lowering the glass transition temperature T of ’ the polymer. Sakellariou et al. If the additive is insoluble. The leaching of PEG 4000 was confirmedby Samuelov and eoworkers (19’79). The use of Ethocel. increases porosity. The rubbery phase. a soluble additive will form channels in the film. Microcapsules prepared either by solvent method or by spraycoating can be made with filmsof very high tensilestrength
. and this is important in EC films. (Data from Donbrow and Friedman. If the glass transition is loweredby the addition of a compound.
Most of the mentioned polymers areamorphousto a large or small degree. thereby. then such holes will not occur.ter 2
”
0
1
2
3
4
5
6
Time (hr)
Transfer of caffeine through an ethyl cellulose film containing 50% PEG. in contrast to PEG. is less brittle than the glass phase. and dissolve in contact with the water of intrusion.
which will then not dissolve in the stomach. OH and one one CH2OH in four glucose units being left unsubstituted. the most often used porosity modifier was sodium chloride. because before that time.55. and found that although the vitro release wasprolonged to 4 h. Reasy et al. commercially.g. thein vivo in release from pain was prolonged by 8 h. CelZulose acetate phthaZate is used as an enteric coat. the appetite depressant ~ e ~ ~ a t e ~ o marketed in years past by Merrell Labs). with molecular weight about 220. te It is insoluble in acid. ela at in.Controlling the releasewas accomplished by adding watersoluble compounds. as Explotab and Primojel. sodium polymannuronateis used as a disintegrating agent and tablet binder.
Cellulose acetate p ~ t ~ a l awas. The isoelectric point is a function of the treatment. By using Ethocels ofhigh molecular weight. The principle is also applicable to beads.but also serves as a (mild) disintegrant in tablet formulations. There are numerous ~olysaccha~ides pharmaceutical interest.000. Hsiao and Chou (1989) had earlier described a similar process for controlled release of potassium chloride. ~ o t a b l y of of these is guar gum. by polymer deposition from cyclohexane by temperature change. a natural product derived from animal hide or bone. hiv van and et al. microcapsules could be p r o d ~ c ~ d lend themselves to ta~leting that ~ithout rupturing the film. Holliday et al. ~ar~o~oZ (carboxypolymethylene) is a carboxyvinyl polymer that has been used in various products in the past (e. but soluble at pH values in excess of 4. and obtained a product that was finer and had a longer sustained action than when a 10 cp grade of EC was used. The polymers are linear and have molecular weights from 15.000 to 250. for many years. for which thecoatingmaterial was cellulose acetate b~tyrate.(Chang and Rudnic. (1980). the best entericcoating compound. It is the sodium salt of the carboxymethyl ether of starch. This isof great importance.. microencapsulated sodium salicylate with ethyl cellulose 100 cp. tableting was considered impossible because cracks in the film would destroy the controlled. which is a galactomannan polysaccharide. but will dissolve in the intestinal tract. which can be either acid or alkaline. In this manner a coat may be placed about a tablet. odium s t ~ r c h glycolate is used as a tablet and capsule disintegrant. sustained effect of the individual microcapsules.
~lginic aci~so~ium alginate. 1991). Carbox~methylcellulose sodium ( CM C ) is used as a thickener in liquid formulation. The CH2OH and OH groups incellulose are substitutedwith either acetyl or phthal~l. (1998) have described compressioncoated tablets.
. whereby the outer layer would become porous. (1970) have described microcapsules of aspirin using ethyl cellulose. It is sold. is a mixture of fractions of amino acid groups that are tied together through peptide linkages. Gelatin is produced by hydrolysis.000.
and similar substituted acrylamides. 1996). This partially dissolves the polymer and is added to the alginate solution. se mechanism from poly(z>.. (1998) have described x a ~ t h a n gum.. 1994.. 1993. 1987. ~ e r v o o~1998a. 1996. As in et al. is a naturally Inulin it occurring polysaccharide found in many plants (Van Loo et al.b).. They demonstrated its use in indomethacin formulations. 1990. is linked fructose molecules. (1998) have described aliphatic polyesters on a base of lactideglycolide copolymers... chemically. possess a lower critical solution temperature. Lambert and Peck. 1995. as a potential use in oral sustaine~release matrices. Niwaet al. Jalil and Nixon. other cases. Fitzgerald and Corrigan. Coffin et al. Therefore. Talukar et al.b) described the use of t has in~Zin ~ydrogels. 1993). 1987).. 1986 at ong et al... 1996 a. 1986. The reaction product is stored at 22°C for 24 h until the reaction has completed.. As0 et al. A series of compounds have been incorporated into this type of matrix (Ike et al. such as poly(Nisopropylacrylamide). some polymers. su et al.3% w/w)is then added.. they swell reversibly at lo peratures and swell only slightly higher temperatures (Hoffman et al. a certain number of which have a glucose molecule at one chain end (~oberfroid. 1993.b). 1993.5 aqueous alginate solution is prepared. for instance. the release of drug from systems of this kind is a function of both drug diffusion and matrix erosion (Sanders et al.. Zhang et al. 1995). (1998) have described an azgi~ate~~dragit L30D system for sustained release. ervas et al. 1992. The microspheres that are formed are then filtered off and dried. and a solution of the drug added.
acetate ~ ~ t ~ a l a t e The oldest of these are shellac and cell~lose
. Sung et al. 1996a. 1987).Zactic is used as a matrixsustaining ingredient employing wet acid) granulation techniques in amounts of as low as 515% w/w poly(z>Llacticacid). Mauduit et al. The Eudragit is dissolved separately by useof a small amount of sodium hydroxide (neutralizing it to an extent of 30%)..lactide) ~oly(z>. itzgerald and Corrigan 1996. Calcium chloride (1. Stee~dam y.. Sat0 et al.r... 1988. ~ a ~ a k a w a 1990. ~ZuronicF127 is a polyoxyethylenepolyoxypropylene co forms an aqueous hydrogel above 32°C (Lenaerts et al.~oZy~~r.r. Asano et al. and Lerk (1998) have used it as a directcompression excipient in controlledrelease tablets. 1990). Hsu et al. There are some advantages to ~ncluding both hydrophilic and h y d r o ~ ~ o b i c moieties into backbone polymers (Serres et al. The beads are prepared in the following manner: 1. 1990. Chandrashekar and Upada..lactidecoglyor scribed (Hutchinson andFur. roperties such as glass transition temperature and molecular weights have been reported (~melczuk and ~ c ~ i n i t1992. 1996).
Eiehel HJ (1982). 1991. 1993).. ~ a t s u m o t o Uda H. hi ma no et al.Arwidsson H. Shirnamura € . has e t a t sensitive and has been used for microencapsulation and for film coating (~hinkuma al. It is insoluble in water. (b) slope of diffusion plot (= ( A I l S 1 / h ) q = viscosity of solution of polymer q0 = viscosity of pure solvent [q]= intrinsic viscosity fl = (a) osmotic pressure. Johansson (1991). (1998).
. Tanaka S. = number average molecular weight M ~= weight average molecular weight . patent 4.884.
A = diffusional area C = molality (moles solute per mole solvent) L) = diffusion coefficient EC = ethyl cellulose fi = number fraction of cellulosic derivative molecules with molecular weight Ml 5 " weight fraction of cellulosic derivative molecules with molecular '= weight M . U. S.exponent in intrinsic viscosity versus molecular weight. K. 3 Aikawa K. Int J Pharm 76:91. (b) permeability
Alam AS. fz = number of moles I" = pressure = e ideal gas constant = cellulose = su~stituted cellulose where n is the degree of polymerization OR'j = substituted cellulose unit = substituent in substituted cellulose S = drug on cent ration in the polymer directly inside the entry side z SI = drug con~entrationd ly on outside of entry sideofpolymer T = absolute temperature t = time V = volume = (a) l/a. = hydroxypropylcellulose C == hydroxypr~pylmethylcellulose h = film thickness llulose (often a type of HPMC) of cellulosic unit..316. I"o1yvi~yZalcohol isused primarily as a suspending or viscosityenhancing ingredient in liquid formulations. (b) mass diffused M . but is soluble in et gastric juice.~ o Z y v i ~ ~ Z a c e t a Z ~ i e t h y z a ~ i n o a ~ beene describedby Aikawa et al.
As0 Y. Okano T. Int J Pharm 11 3:39. patent 3. U. Kirit GC (1970). U. J Controlled Release 42:125. McGinity JW (1992).849. Bell SA. patent 4. Nixon JR (1990). Crit Rev Food Sci Nutri 33:103. 1070492. Schor JM (1981). J Controlled Release 40:293. Jalil R. J Controlled Release 25:43. iyazaki S. Ike 0. ashekar G.Asano M. Hsu YY. S. Kim YH. Donbrow M. Yoshioka S. patent 4. Bodmeier R.172. S. Pharm J Pharmacol48:669.418. H. Ikada Y (1992). J Biomater Sci Polym Ed 5:473. U. Gresser JD.226. Int J P h a m Techno1 Prod Manuf 3: 111. Gresser . Niwa T. FernandezHervas MJ. U. Nakamura T. Dong LC (1986).488. Schor JM. Kim SW (1996). Lambert WJ. J Controlled Release 13:279. Lyons CM. Dong LC. J P h a m Sci 68:325. Wagner RA (1975). DOWl n f o ~ a t i o n Sheet (1982). S. AS Dong LC. J Appl Chem 8:363. Holgado MA. White EFT (1985). Kunou N. H o f f ~ a n (1992). Dong LC. Furr ABJ (1990). Bukh N. [Table headed “Viscosities of Methylcellulose of Various Molecular Weights]. Couvreur P (1987). Hsu YY. patent 4. Brophy MR. J Microencapsul 753. DOW Handbook on Methocel Cellulose Ether Products. J Polym Sci A1:311. FernandezHervas MJ. Schor JM (1979). Triqueneaux C. patent 4. nPeppas L s NA (1989). Gaylord NG (1981). Biomaterials 13:230. Kim SW (1994). 192886682. Holiday WM. ~itzgerald JF. J Controlled Release 4:213. Kim W (1990). Roberfroid MB (19 . Bae YH. Berdick M. Chou T ( 9). Quarton M. J Controlled Release 9: 1 1 1. Mashimo T. Shimizu Y. Academic Press.743. Wada R. J Controlled Release 19:171. ang gad ha ram PRJ. J Controlled Release 13:21.357.369. Vert M (1993). Hoffman AS. McGinity JW (1987). Deasy PB. Trantolo DJ. U. Nigalaye A. Lenaerts V.Simons GA. S. Greminger GC. Savage AB (1959). J Pharm Sci 76:261. Pharrn Res 9:26. Afrassiabi AA.469. Yokouchi C. Int J Pharm 70:261. lnt J Pharm 163:23. Fell JT (1998). Wise DL (1996a). Rowe RC (1982). Chang RK. Wise DL (1996b). Int J Pharm 39:121. Neeley WB (1963). Friedman M (1974). RiegFalson F. Yoshida Kumakura M. Trantolo DJ. Mauduit J. Gangadharam PRJ. Hino T. J P h a m Pharmacol 32:15. Fint A. J Polym Sci B Polym Phys 28:923. P h a m Res 13: 196. Friedman M (1979). Schor JM (1982). Hutchinson FG. Serres A. Rowe RC. Terao T (1994). offman AS. Chang KT. J Controlled Release 3 1:33. del Cerro J (1995). Brown AM (1958). Peck KD (1995). Chem P h a m Bull 35: 1243. J Controlled Release 25:89. pp 565596. In: Whistler RL. J Controlled Release 28:143. METHOCEL. Stewart RR. patent 4. Industrial Gums~olysaccharides and Their Derivatives. Corrigan 01 (1996). Lyons CM. Ecanow B.863. Coffin MD. Bae YH. Hyon SH. British Patent. Y an Q. Rowe RC (1980). Rudnic EM (1991). Suzuki K (1989). No. Omelczuk MO. Rudin A. S. Controlled Release J 8:267. N (1996). Hoffman AS (1990). Yuasa H. S. Hsaio C. U. J a m Pharmacol 32: 116. J Appi Polym Sci 19:3361. Imai K. J Controlled Release 33:189.389. oore WR. Takada M (1987). Kawshima Y (1993). Vela MT. ed. Takeuchi H. Donbrow M. Int J Pharm 27:267. Samuelov Y. New York. Baudys M. J Pharm Sci 85:706. Sakellariou P. Yamanaka M.393.
. Fukuzaki M. Joy MM (1980). Van Q (1994). J Pharm Pharmacol27:633. PO ALW.
. Kawahara M. Hirose J. Augustijns P. Van Loo J. Ju TR. Yakuzaigaku 53:27 1. J Pharm Sci 79505. Yamakawa I. Shinkuma I). Topp EM. Vervoort L. Miyake Y (1990). Steendarn R. Zhang X. Vervoort L. Talukar MM. Int J Pharm 169:105. Rombaut P. Nixon PR. Goto S (1993). Crit Rev Food Sci Nutr 36525. Ito Y. Coussement P. Int J Pharm 172: 137. Higashi Y. Koyama L. Kobayashi T. Int J Pharm 14253. Augustijns P. YamanakaY. Pichora I). Siege1 RA. Amsden B. Lerk C F (1998). Pate1 MW (1998). Int J Pharm 172:127. Augustijns P.Miwa A. Mits G (1995). Firestone BA (1988). Kinget R (1998). Van den Mooter G. Harnaguchi T. Kaondo 0. Proc Symp Controlled Release Bioact Matter 15:164. Int J Pharm 175:33. Watanabe S. Verbeke N.Shimano K. Goosen MFA (1993). Waldman MH (1970). Van den Mooter G. Sung KC. Wirick MB. Kinget R (1998b).Mizuno N (1991). J Appl Polym Sci 14579. Wyss UP. Kinget R (1998a). TjandraMaga T. J Controlled Release 25:61. Gao P. Van den Mooter G. Yoshida T. De Leenheer L. Skoug JW. Int J Clin Pharmacol Ther Toxicol 29:303. Hoebregs H.
This Page Intentionally Left Blank
.
1. Aqueous Film Coating 27. 27.5.
roperties of Aqueous Films Symbols eferences ecommended Reading
27. rendering the encased drug substance more stable. SustainedRelease Coatings on Tablets
Coating tablets has several purposes.2.27. It. Solvent System 27.8. Plasticizers 27. Strength of Films and Effect of Storage 27. 27.
ransition ~emperatures rigin of Film Coating: ~ r g a n i c Solvent Coating
456 56 457 457 459
460
460 462 463 464 465 467 467 468
27. 27.7. li y 3.9. aids in the s ~ a ~ l o ~ a b i of ta tablet. 4. It facilitates making colored tablets with a m i n i ~ u ~coloring substance.
. in many instances. It is a barrier to moisture and (possibly) oxygen.
1. of 2.10.3.6.4. It masks taste.
plasticizer. opacifier. Once dry. in turn. A more controlled system is provided by fluidization and spray application (Wurster apparatuses). The process of tablet coating will not be covered here. 27. Solutions and syrups are added in just sufficient portions to make the tablets turn over.. but many popular products in today's marketplace that were once sugarcoated and are now filmcoated (e. andits use is much less common nowadays than a half a century ago. Operational variables may be minimized by spray rather than ladling systems. Operator skillisnecessary. It is traditionally done in coating pans. and spraying the liquid onto the tablets. when allcoating was carried out in this fashion. It is obvious fromthe figure that there is a series of steps necessary carry out to the operation.g. is lengthy. either in a coating pan or in a column. polymeric film. The reader is referred to the texts listed in the ~ e c ~ ~ ~ ~~ ~ e i e section at the end of the chapter. as shall be seen later in the chapter. when for necessary. may be water).
I
Build Up Coats
ing &t Coats
Schematic of sugarcoated tablet. much care is needed. OneADay vitamins).
Barrier Coat
Wax
/
* . is deposited on the surfaceof a tablet. one more application is added. The method of deposition consists of dissolving or dispersing the ingredients in a suitable solvent (which.
Sugar coating is the oldest of the coating methods. usually with a thickness of10100 bm. but even so. The film consists of polymer. exceptto the extent of a brief overview. and coloring agent. There are still several popular products on the market that are sugarcoated (ibuprofen products).1. and so on.
.in particular when pan coating isresorted to. The finished product has the overall composition shown in Fig.The manufacturing methods coating will be touched on only briefly. and the water is then evaporated off with hot air blasts. dried.
In film coating a thin. This process.
5 > 5. in a patent by Pita1 (1969). Some (e. 1985). osmosis. or drug diffusion. For engraved tablets. film must possess strength and elasticity. and covering over of the intagliation is also a problem.0 > 5.” As the solvent evaporates there will be a shrinkage.5
> 6. but disintegrate or dissolve inthe small intestine are often denoted enteric coatings. ~equirements anordinary film coat aredescribed. The dual characteristics are achieved by the combination in a coating composition of a filmforming water soluble solid.~
> 6.
Coatings that resist dissolution in the gastric fluids.
eq~irements a good film is that (a) the film be uniform. is Enteric films may be used simply as the barrier coat in a sugarcoated tablet. (b) that it be for without cracks. To withstand both of these vicissitudes. have the capability of reducing the stress buildup. ethyl acrylate ~olym~thacrylic acid Polymethacrylic acid..2. they contain carboxyl (or other acidic) groups. They are insoluble in acid and soluble in alkali.0 > 5.g. the film isnot aslikely to crack. a difference in thermal expansion and contraction of the substrate and the film. the coating on the tablet gives excellent protection against external media. and as such. (d) that it has no cracks or flaws in ap~earance (onionskin appearance).
7. because they reduce the cross section of intact film. Yet.0
. hence. furthermore. n~ethyl~ethacrylate Shellac Name in trade CAP HP50 Eudragit L30D Eudragit L Confectioner’s glaze pH range if soluble
. (c) that it adheres well to the tablet.1.(b) protective. 27. for for instance. insoluble fillers (opacifiers such as titanium dioxide) are detrimental. 1983.
A film coats tablets to form a membrane which is semipermeable to water thereby permitting the tablet to disintegrate in two or three minutes after administration. or (c) sustaining. talc). There is. A list of such ~olymers shown in Table 27.
TWO common types of defects in film coating are peeling and edge splitting. however. a water soluble polyglycol of high molecular weight. the grooves may not be ““filled in. The latter willbe dealt with in a special section of the chapter.1 Enteric Polymers
Polymer Cellulose acetate phthalate HPMC phthalate Polymethacrylic acid. The manner in which it acts (delays) is either by dissolution or erosion of the polymer. These are exemplified in Fig. For the most applications.atin
The purpose of the film is (a) esthetic (Rowe.
in doing so. as mentioned in the preface. a small amount of film is removed. where it is denoted edgesplitting. but contributors are the plasticizer and the concentration of plasticizer. and this problem has been addressed by Rowe (1982a.
~plitting can occur over the dome of the tablet or (more often) at the edges. up until a point where the embossing completely disappears from view. The foregoing types of defects may be problems with the film (plasticizer). who coated tablets with varying amounts of a 5%0 w/v solution of HPMC in water. Therefore. 1967). Rowe (1980) reported on the effect of the molecular weight of hydroxy~ropylmethylcellulose ( H ~ M C on the percentage of ) edgesplitting in filmcoated tablets. The thickness of the film coat affects embossobliteration. but may equally wellbe manufacturing problems.3 demonstrates the effect of film thickness ( ~ m on ) the percentage occurrence of defects. it follows that thin films will be pred~minated adhesive forces. The intrinsic adhesion. however. because such a pick causes dosedumping. which by means that the embossing willnot coatover. not the intent of this text.” The tablets will stick together for a short while and then separate and. but it should be men ti one^ in this connection that spraying at too high arate.
. Defect rates of this type are a function of the molecular weight of the polymer used. Discussionof manufacturing conditions is. at the constant and independent of thickness. The residual stresses in films caused byshrinkage this trend continues until a limiting value is reached ( Gardon. and having the tablet mass at too low a temperature cause defects known as “pickers.b).Shrinkage Stresses
Poor Adhesion
&Good Embossing Adhesion
bridged
Edge Splitting
Defect types in. This is particularly important in sustainedrelease films. filmcoated tablets. or byusing insufficient inlet air. The incidence of bridging (obliteration of the embossing) will be discussed further in the following. using glycerin (in an amount of 20% of the C) as plasticizer. Figure 27. he opposite is true forthick films.
)
Plasticizers. that is. 1980. 1959. diethyl succinate. 26.2.
. it is but seen from the table that the two criteria would lead to different conclusions. its suitability for tablet coating). Plasticizers include castor oil. this end some investigators use tensile strength and yield point as indicato~s. of plasticized films by differential scanning calorimetry " SC) and by thermal gra etric analysis (TCA) is complex. and the use of the ided torsion balance (I3 is a preferred method. ore to the point. depends on chain mobility.. the effect on the end use of the film.) values of polymers.b) (i. is the effect that plasticizers have on the glass transition (7. (Data from Rowe. have the capability of reducing the glass transition temperature of a film (Rowe 1982 a. as mentioned in Chap..The effect of various plasticizers on a methyl cellulose film is shown in Table 27. There is some sense in considering the mechanical effect that a plasticizer has (e.1
Film Thickness (microns)
The effect of the thickness of EC films on the occurrence of defects in film coating.g. and the purpose of a plasticizer is to improve the chain mobility. the value of Tg for a plasticized polymer may be estimated by knowledge
Properties of Plasticized ~et~ylcellulose Films (containing 30% Plasticizer) Tensile strength (MPa)
70 35 44 46
Plasticizer None PG PEG 6000 Glycerin
Yield point (MPa)
59 33 22 14
Ultimate elongation
("/. The measurement of Z. but the tensile strength does not. and polyethylene glycols (PEGS). for Cr.)
14 50 39 48
Source: Greminger and Savage.e. because the yield point decreases in the series. the polymer is more likely to be in the rubberya plasticstate). If such instrumentation is not available.
the torsional braid pendulum.
owe (1982a).4. The comparison of data is listed in Table 27. there is a maximum in intrinsic viscosity. Rowe and Forse. It so happens that this corresponds to the minimum in a series of properties (elongation at rupture. based on work by Sato (1980) and Chow (1976). a better method exists. considered the strain induced during storage of a filmcoated tablet. but a similar plot emerges when work at failure and elongation at rupture are plotted versus molecular weight. work done to produce failure. (1985) have compared Tg values obtained by DSC.5 is represented in loglog form. Table 27.5 for the tensile strength. for films. The plot in Fig. and 4p denotes volume fraction of polymer. namely.3 shows the effect of concentration and nature of plasticizer concentration on Tg values of hydroxymethyl propylcellulose (HPMC) film coatings. 27. 1979. 27. and tensile strength). TGA. The peaks for such transitions are weak and.of the Tg values for the individual components by using the formula by ueche (196 1): (27.4. In Fig. Sakellariou et al.I
97 91 87 73 26 24 22
0 10 30 10 30 10 30
177 141 75 140 70 153 103
Source: Entwistle and Rowe. and assumed that there is an isotropic linear strain E . 29. The functional effect of plasticizer concentration on Tg values of polymer films has been reported by Entwistle and Rowe (1979). and torsional braid pendulum (TBP). 1981.1) where the subscripts p and d denote “polymer” and “plasticizer. of
Effect of Plasticizers and Their Concentration on T Values of HPMC Films ’ and Effect on Bridging Concentration (W%) Bridging incidence
TgW)
Plasticizer None PG PG PEG 200 PEG 200 Glycerin Glycerin
(”/. This is demonstrated in Fig. simply for presentation convenience.” and where /?is the volumetric expansion coefficient (of the order of 5 x 104/0C).
Previous chapters have shown that the predominant method for determining glass transition temperature is differential scanning calorimetry.
.
) and
3. 3. 1979 Eiitwistle and Rowe. (Data from E n t ~ i s t ~ e Rowe.
v1
a . 1985
.6
3..
3 .. 1979. k. from Entwistle and Rowe. 1979 Okshrnafe and York. Porter and Ridgeway.)
Comparative Values for Tg by Different Methods
L7'g
Polymer CAP HP50 HPSS EC HPMC
by DSC 171 146 136 133 180
T by g
TGA 170 146 133 133 169174
Tg
by TBP 185
Literature Ref.
c
a .2
_( .. I
c
&
U
MW of PEG
Fi
Effect of molecular weight of plasticizer on intrinsic viscosity of HPMC.O
w
v)
. 1983
I29 177 155
Entwistle and Rowe.8
5
a . 1979.
00
3.4
f. (Data ig.o 4
5 6 7 8 In[Molecular Weight]
9
Tensile strength of HPMC films as a function of molecular weight of PEG.
Tables 27. hE1 and AE2 are the vaporization energies of the two components. This latter is close tensile films.4) where Vm is the volume of the final mixture. of original volume V . Crude as the method may seem.5)
the solubility parameter.U)}[AV/3V] (27. (27. In such a case. However..35. According to the HildebrandScott theory ( ildebrand and Scott. so that for the solvents used. because the concentration at which gelling takes place is governed by the solvent and (in a kinetic sense) the rate of evaporation. A H . it seems to give good qualitative results.3)
where u is the Poissons ratio. VI and V2 are the molar volumes of each component and $1 is the volume fraction of component “1. in Eq.2)
From this and the assumption that the strainsin the coating and substrate at corresponding points are identical (Stanley et al. absence oflumps). (27. and h V is the volume increase of a tablet. then the assessed solubility parameter is the midpoint between the solubility parameters of the two or three solvents. For aqueous filmcoating the solvent is essentially fixed.4) is taken to be a function of (61 . a word and two about this point may be in order. a 1% volume increase in storage of a tablet would result in P = 5. The heat of mixing. then that is assumed to be its solubility parameter. For cases where 2i1 = 62 there is complete miscibility.. Often directly compressed tablets expand to such an extent on moist storage angekar et al. If it is soluble in more than one.E
= AV/3V
(27.” It is customary to denote the term
[ ( A H . a polymer in a certain concentration isobserved in three solvents eshibiting “poor. it is possible to assess the com~atibility In between two polymers.5)] A H is the heat of evaporation. the enthalpy of mixing AH is given by (27. but notin the other two.” and “strong” hydrogen bonding. The solubility of the polymer in the three solvents is assessed (e. but assome filmcoatingis still carried out with solvents.62)2.R T ) / V}0’5= 6
(27. Rowe (1982) givesan example for a film with an E value of lo3 MPa and u = 0.” “moderate.owe arrived at the following formula for the stress in the film:
P = [E/(1 . in the solubility parameter definition [see Eq.. 1950). and if the polymer is soluble in one.
The solvent used is of importance in that it governs the shrinking stresses.1whereasa10% increase strength of most in volume would yield P = 51 MPa. complete solubility~ this manner.5 and 27. in particular owing to slow expansion caused by moisture uptake of the disintegrant.6 show examples of this. By this method. E is the Young’s modulus. on storage. 1972).g.
. 1980). by clarity. the calculation of 6 values can be urrell (1975) describes a method for calculating the valuesof the solubility parameter forpolymers. There are lists of vaporization heats of solvents published 973.
These.41 9. For instance.4 19.~) 1689 1627 1868 1.35 1.1997) report the useof 55. (1 996)have describedtechniques. and Felton et al.1 17. or PEG 6000.7 33. however. plasticized withtriethyl citrate..1997).6 17. 1967) have suggested constructing threedimensional solubility plots. c~uinity (1996. For the latter. when used for three ECgrades worked well with three solvents. Table 27. dibutyl sebacate. ansen. are difficult to apply to f o ~ u l a t i o ~ practices. Johnson and Zografi. tributyl citrate. 1986).6 shows that the tensile strength and the Young’s modulus decrease as the 8value of the solvent used for casting the film decreases.
Solvent Effects on Properties of EC NGrade Films &value. To evaluate the suitability of films.020. researchers have used the butt adhesion technique and the peel test.429.0 26.1
olvent with strong Hbonding 19. 1966.422. (MPR’.419. Eudragits have also found use.5. which determine the force that is needed to separate a polymer film from a substrate surface (Fung and Parrott. Felton and ~ c G u i n i t y (1996.
Film coatings were. in their early development applied using a solvent.5) do not show a consistent trend.57 Young’s modulus (MPa) 25. ~ ere. It is noted that the data for elongation (last column in Table 27.8 Tensile strength (MPa> 51. some authors (Cowley et al.2 15.423.8 51. To improve on the picture just presented.7 19423.2 59.ECgrade
Solvent having poor Hbonding 17.52 1. Wang et al.5 16. Fisher and 1976.0 Elongation
Solvent Chlorobenzene 2Nitropropane Benzene
(%I
. hydroxypropyl methylcellulose ( H P ~ or )hydro~ypropylcellulose ) with plasticizers have been polymers of choice.3
T10 N22 K200
The method. 1980. the Ngrade of EC has a &value of 19 Pa1/’.622. but environmental regulations are now suchthat virtually all coatings are made by watersoluble polymers.122. This value is within the intervals shown in the second line of Table 27. (1996). whereby the edge of a film on a tablet is just lifted.4
Solvent having moderate Hbonding 16.
and a plot. 27. 27. but not when it was waterinsoluble. and the plot is linear (Fig.
F
76) has studied the effect of the molecular weight of HPMC on the properas continuity of film.7).9 as being a linear function
" " 1 ' " ' 1 " "
Force of Adhesion
4
n
V
24
on
3
2
a
U
$
1
0
50 100 Deflection (microns)
150
Peel test to determine the force needed to separate a polymer film from a substrate surface.)
Young's modulus as a function of molecular weight of film polymer. The hydrophobicity of the surface influenced the adhesion when the plasticizerwas watersoluble.and attached to an instrumented platen that may be raised. of films made from Not unexpectedly. 27. (Data from Felton and McCinity. ~rineZ2 ~ r ~ n e (Fig.6. such as shown in Fig. 27. is obtained. the Young's ~ o ~ u Z u s increases with molecular weight. Other characteristics are functions of the molecular h ss weight of the film polymer.hardness. elasticity. The deflection is monitored as a function of force. 1997. and substratetofilm adhesion. rushing stre~gth shown in Fig. (Data from
.8) is a linear function of the is molecular weight.
G
v)
90
a
ij
80 70 8 1110 42
16
MW/1000
18
20
22
Crushing strength as a function of a polymer’s molecular weight. 1976. (Data from Rowe. 27. but the plot may be curved and asymptote.10).
It is obvious. that film coating should be capable of being used as sustainedreleasemediators.8
10
12
14
16
18 22 20
MW/1000
Brinell hardness as a function of the molecular weight from Rowe. from the foregoing discussion. Lindahl(1986) has described a sustainedreleasecoated tablet “comprising a dru~containing tablet and a coating or mem
.5 mm tablets. (Data
130
120 I10
100
IC
.)
of a film polymer. The ~ e ~ c e ~ t f f g e o f ~ ~ ~ defectst (as judged by mercury penetration) in particular batches of filmcoated Z e tablets is loglinearly decreasing with molecular weight (Fig.5 rnm tablets and triangles are 8. 1976.)
of molecular weight. Circles are 6.
Data from three tablet sizes are averaged.99
Time (hr)
Effect of plasticizer level HPMC) on release of drug from an EC (YO filmcoated tablet. (Data from
. castor oil and PG the plasticizer. wherein the coating. .t o
2"""""""
Percentage of tablets failing mercury penetration test for continuity of film.)
brane surround[ing] the same. of e (1985) shown has the effect of plasticizer (here.is waterinsoluble and insolublein ~astrointestinalfluids and consist[s]essentiallyof a terpolymer of polyvinylchloride.1 1 and 27. (Data from Rowe. Ethyl cellulose (EC) filmshavebeenused in this type of sustainedrelease 'ngs.
. i (1970) describes a similar principle in which CAP or nitrocellulose are the rs. The concentration of plasticizer in such instanc d be importance. 1976. ) in ethyl cellulosecoated tablets (Figs.12). 27. polyvinylacetate and polyvinylalcohol and a watersoluble porecreating substance. . . and acetone the solvent. .
Rowe RC (1979). J Pharrn Pharmcol 31:269. Entwistle CA.(AE2/V2)'. of mixing.
. (c) heat of ~)0. New York.Polymer Handbood. = molar volume of component 1 V2 = molar volume of component 2 PP = volumetric expansion coefficient of polymer P d = volumetric expansion coefficient of plasticizer 6 = solubility parameter q$p = volume fraction of polymer = volume fraction of component 1 u = Poisson's ratio
Burrell H (1975). 1985. 337. In: Bandrup J. 2nded.Wiley Interscience.5 (b) beat evaporation P = stress in film T = glass transition temperature ' 7= glass transition temperature of polymer " ' 7 '= glass transition temperature of plasticizer " .
10
20
30 4 0 65 0 0 MW/1000
70
Effect of MW of EC on the initial release from afilmcoated tablet with sustaining properties. Liu CA. eds. Chow 715. pp IV. Immergut EH.'~q$~q$~.)
E = Young's modulus EG = ethyl cellulose AEI = vaporization energies of component 1 AE2 = vapQrization energies of component 2 C = methylcellulose PG = propylene glycol PEG = polyoxyetbylene glycol methylcellulose PMC = hydro~ypropyl A H = (a) V ~ ~ ( A E ~ / V. J Polym Sci Polym Phys Ed 14: 131 1. Penwell RC (19'76). (Data from Rowe.
J Pharm Sci 75:529. S. Sat0 K (1980). Rowe RC. Rowe RC (1985). J Pharm Pharmacol28:886. New York. Rowe RC (1982a). Rowe RC (1982). Kanig JL. Trans Am SOC Mech Eng pp 697704. Theory and Practice of Pharmaceutical Technology. Philadelp~ia. McGinity JW (1997). Rose RC. Industrial GumsPolysa~charides and Their Derivatives. Zhang G. Rowe RC (1980). In: Lachman L. MarcelDekker. Johnson BA. 3 Pharm Pharmacol 35:341. Yeager JL (1986). Rowe RC (1983). In: Whistler RL. Pharm Ind 4:173. Malick AW. Felton LA. Int J Pharm 154: 167. Sheth PR (1972). J Pharm Pharmacol 32:584. J Pain Technol 39:104. Infeld MH. McGinity JW (1996). J Pharm Sci 69:439. Ridgway I (1 3). Fisher DG. pp 269324. S. York P (1985). Kelley FN. Greminger GC. Savage AB (1959). Felton LA. Malick AW. Gardon JL (1967).538.Felton LA.629. Lindahl AR. Shah NH. Zhang G. Meissner HP.214. patent 3. patent 3.
. Prog Org Coat 8:143. McGinity JW (1996). U. eds. Mehta SP. J Pharm Sci 61:939. McCinity JW (1996). New York. ed. Pharm Int Jan p 14.
Seitz JA. Pharm Res 2:19. Fung RM. Lea & Febiger. lnfeld MH. U. Bueche F (196 1). Rowe RC (1976). J Polymer Sci 50:549. Int J Pharm 27:267. Pharm Dev Technol 1:2 13. Pita1 G (1969). 3rd ed.476. Newton JM (1981). S Rowe RC (1976). filed 1964. vol 1. Sarli M. Pharm Dev Technol 1:381. J Pharm Pharmacol 35112. Porter SC. In: Treatise onAdhesion and Adhesives. Stanley P. White EFT (1985). S. Forse SF (1981). O~shmafe AO. Academic Press. Zografi C (1986). Polli GP (1970). Baldauf C H (1951). Rowe RC. Erlandson SAB (1986). Int J Pharm 127:203. J Pharm Pharmacol 33:557. U.588. Wang CC. Sa~ellariouP. pp 346373. Hansen CM (1967). Parrott EL (1980). patent 4.620. Int J Pharm Technol Prod Manuf 3:3. Shah NH. Lieberman HA. Pharm Acta Helv 51:330. Sangekar SA. pp 565596. J Pharm Pharmacol 33: 174.
Ideally. in principle. 6. then lowering its solubility will prolong the time it stays undissolved. the therapeutic levels (ThL in Fig. Then. Complexation and derivatization Erosion tablets Rigid matrices Swellable matrices Floatable tablets Osmotic pumps
If a drug appears to require frequent dosing. in this manner the drug will be absorbed more slowly. 5. It the former that will be discussed first. The rationale for such a dosage form is selfevident: taking one or two doses a day is preferable to a patient over taking two to four doses daily. eversince Smith Kline & French Laboratories marketed a sustainedrelease product in the early 1950s. 4. 2. acting in the sustained fashion. 28. and if the absorption is such. then the problem is often one of solubility. For a variety of reasons. be administered as rapidly disintegrating tablets or capsules. If the drug substance is consistently absorbed throughout the entire (or a substantial part) of the gastrointestinal tract. and this is the goal of most sustainedrelease formulation efforts. over a longer period of time. a linear.
Blood
Level
ThL
Time
Schematic of sustained release. Some of these consist of a singulardosage form. in vitro release pattern is preferable. and (b) prolonged action. 3.1) are reached over a longer time. when a derivative with lower solubility is used. there is the dual benefit of (a) reduced risk of reaching the toxic limit (TL). Such products can. others are mu~tiparticulate.
.The development of sustainedrelease products has been ongoing in the pharmaceutical industry. There are several principles that have been developed over the years. and these are the followis ing: 1.
and then passing into the small intestine. that is. controlled amounts of polyethylene glycol distearate were added. behave similar to erosion tablets (Christenson and Dale. as time in contact with dissolution medium progresses. 1968. The tablet is subjected to N/lO hydrochlori~ acid for 30 min1 h. because the releasedepends ontheproduct staying in the stomach for a certain length of time. and the dissolution medium is then changed to a pH 7 buffer.
Erosion tablets are tablets that do not disintegrate. 28.1) where a is the “diameter” of the tablet (assumed spherical). t is time. and more continuousrelease patterns are of advantage.s
Historically. Some grades of hydroxypropyl methylcell~lose (HPMC) form matrices that. Lapidus and Lordi. Ford et al.b. is also a disadvantage. in itself. Sterotex (hydrogenated vegetable oil) hasalso been used as a wax base. In vitro testing of such products aremost often carried out by a socalled half~hange method. Erosion tablets often follow a cuberoot equation.
. some of it (in the entericcoated core) will not release until the tablet has passed into the small intestine where the enteric coat may dissolve. integrity of the enteric coat being difficultto achieve in scaledup manufacture. This is a bimodal approach. 1987). and K is an erosion constant.2). The bimodal nature. The solid “sloughs off’ ata constant rate. in that some ofthe dosage form (in sugar coat of a tablet) will the release immediately. in part. Products of this type are difficult to make consistently.c. Carstensen and Valentine (1966) found this to hold true and used carnauba wax in which they imbedded the drug substance. This integrates to
Sugar Coat ~ontalnlng Half of Drug Dose
I
’
Enteric Coat
Principle of sustained release based on enteric coating. 1966. To control the rate of erosion.. the earliest attempts at sustainedrelease dosage forms by manipulation of the dosage form was in the form of entericcoated tablets (Fig. 1966. 1985a. but simply erode. The gastricemptyingtime is also a disadvantage. daldt = kl (28.
.2).a = a. Christenson and Dale(1966)find a linear erosion patternas described in Eq. in the case of certain eroding matrices. and 4 is a squareroot constant. except there is a lag time. . (28.t.5) hence. This is res sum ably the manner in which the sloughing off of the surface occurs. Inserting this in Eq.)'I2 = (28. and m0 is the original mass of the tablet.a = 4ct .)
. (28.( ~ / ~ 0 ) 1 ' 3 (4/ao) st .
Square Root of Time
Eroding front for HPMC matrix. HPMC. Eq.2) (28. (28.4)
or (28. to some degree.t.( y p z / ~ o ) ' l ~= (K/ao>t (28. 1998.8)
where t. its volume v is
v = (n/6)a3
(28. exhibits an eroding front which. ypz is mass not dissolved.(ala01 = (K/ao)t
and since the tablet is considered spherical.3.2) takes the form
a0 . because of initial wetting of the tablet surface.9)
Even with erosion tablets there will be some intrusion of liquid into the matrix.Kt
(28. (Data from Konrad et al.3)
1 .7) is followed in a wax matrix system. is a lag time.2). Even though wax tablets are not porous per se. (28. as shown in Fig.7) this now becomes 1 .7)
Equation (28.By dividing through by a0 and following a development akin to Eqs. but squareroot in time. If this is true.)'I2
(28. This may. 28.6)
where a0 is the original diameter of the tablet. there is always some residual porosity. alao = ~ v / v # ~== (ypz/ypzo)1/3 (28. is actually not linear. invalidate equations such as Eq.3) gives: 1 .8) and (28. (28.
Sn i
From a theoretical point of view (Carstensen, 1980), the intrusion of a liquid front into a (porous) matrix should follow the Washburn equation (Washb~rn,1921; Nogami et al., 1966; Couvreur, 1975): dL/dt =  ~ r 2 / ( 8 q = ) ~ q/L where (28.11) and where I, is the length of the intrusion at time t, r is the average radius of the pores, q is the viscosity of the liquid, and Q is a constant. If Eq. (28.9) holds, then the linear erosion would presumable be related to L; that is, L = P(ao  a) so that dL/dt = P(da/dt) = qP(ao  a) (28.13) (28.14) (28.12) (28.10)
so that by integration
 In[l  (alao)] = qt  S2
or
(28.15) where S2 is an initial condition constant related to the fact that wetting of the tablet surface requires a small, yet measurable, length of time. Other recession relations exist; for instance, Bamba et ai. (1979) have shown that in certain gums the erosion front itself follows a cuberoot law.
A matrix is a uniform mixture of drug, excipients, and (e.g.) polymer that is homogeneously fixed in a solid dosage form (Dow Methocel Bulletin, 1982). The basic principle of a rigid matrix was first developed Higuchi (1963),and by the aspects of it are depicted in Fig. 28.4. A twodimensional model is described here,as it illustrates the limitations and some of the misconceptions of the model in past literature. The drug substance, which has a solubility in the dissolution medium of Sg/cm3, is dispersed in the matrix, which is insoluble in the dissolution medium. The concentration of drug in the matrix is A g/cm3 of matrix. The matrix is porous, with a porosity o f E . Liquid will intrude from the bulk liquid, and in the model presented here will enter from the right in Fig. 28.4. The rate and extent of intrusion will follow Eq. (28.10), so that there will be a liquid front, as shown in the figure, which is x = L cm from the surface (where x = 0) at time t. The intruding liquid will dissolve drug substance, and at a given level of intrusion L, part of the matrix, between L and h, will still contain solid particles that are not yetcompletelydissolved,whereas in the volume to the right of h (Le., for 0 < x < h) all particles are dissolved. In the volume I, > x > h, the liquid will be saturated in drug substance, but when x < h, the concent~ationsteadily decreases
Solid Still
Present
No Solid P;esent
Liquid‘ Front
\
x=L
\ c=s
xh
x=o
x=o
x
Schematic of a rigid matrix in two dimensions with one side (to the right) exposed to liquid.
until it is zero at the interface with the bulk liquid. It is, as shown in the figure, assumed that Fick’s law applies, which will make the concentration gradient constant (i.e., the concentration will be linear in distance up until x = L). The volume of liquid in the volume 0 x < L is he and the average concentration is 0.5S, so that the amount of drug present in the volume at time t is O.5Seh. 3 The amount of material ( ,released at time t, will be the amount originally present in this volume (Ah), less what is present at time t; that is,
= Ah  0.5S~h
(28.16)
which in differential form is /dt ( A  0 . 5 s ~ dh/dt ) (28.17) Considering the amount in the liquid present at x = h to be SE,then the concentration gradient (see Fig. 28.4) is SE in the region 0 4 x < h. Fick’s law then gives dQ/dt = DSe/h where 21> is the diffusion coefficient. q. (28,16), (28.17), and (28.18) now gives dh/h = ( D S e / [ A O.SSe])dt (28.19) which may be integrated to 0.5h2 = ( D S e / [ A 0.5S&])t (28.18)
(28.20)
where t = 0 implies h = 0 so that initial conditions are met. Introducing Eq, 28.6 this becomes: 112 (28.21) = { 2DSe[A  0.5S~l) t 112
se
The domain of this equation is A > 0.5s~. this does not apply, then the equation If becomes (Fessi et al., 1982)
Q~ = a2Dt
(28.22)
where a is the area through which the diffusion can take place. There are two limiting concepts in this model, and this type of preparation: 1. The matrix mustbe““continuous.” 2. The pore space must be c continuous" (i.e., pores that are occluded will not work in the model). The limiting situations are then (a) when the drug content is very low and (b) when the matrix material content is very low. These twosituations will be dealt with at at a later point in this section. A couple of points are of importance. The porosity term in Eq. (28.21) is not the porosity, &table. of the original tablet, but rather, the porosity E in the volume 0 < x < h. This consists of the tablet porosity E,, plus the porosity ~ d created by the , complete dissolution of the drug substance in the volume. This latter is 3 (28.23) l d = A/pA cm where pA is the particle density of the drug substance; that is,
E
= E, 4” A / &
(28*24)
This is a means of controlling (increasing) the dissolution rate of the drug by adding soluble excipients (e.g., lactose). If they are present in a concentration o f Bg/cm3, then the porosity to be used in Eq. (28.21) would be
= E2
+ llflP‘4 + 4 9 s
(2~,25~
where pe is the particle density of the soluble excipient. Converselyif it is desired to decrease the dissolutio~ rate, then a larger ratio of matrix to drug substance (a decrease in A ) would be called for, The difference betweenEqs. (28.21) and (28.22) deals with the fact that below a certain porosity, the pore space is no longer continuous. This aspect is the subject of percolation theory.
ercolation theory is a mathematical tool, that allows prediction of the foregoing situations [Le., whether Eq. (28.21) or (28.22) applies]. It has been de vera1 authors (~euenberger al., 19 et 1995; Adrover et al., 1996; ervas et al., 1995, 1998; Towgen and or liquid intrusion (transport) to occur, it is necessary that “clusters” occur and that there are sites or bond percolation that take place (Stauffer and Aharoni, 1985). A continuous pathway of sites that “conduct” (percolation sites) must exist in the matrix. ~ h the~(tablet) porosity is very low, then the number of conducting n sites will be so low that a continuous pathway will not occur. The porosity at which this occurs is denoted the critical percolation threshold (E*). Above this there will be a part of the pores that are available for intrusion, and this is designated the accessible fraction ( E J , and the total porosity ( E ) is the sum of these two.
E, E,
= (E  E*)@
E
> E*
E*
(28.26) (28.27)
=0
when
E
a
ere, /3 is a constant of universal nature. In percolation theory, /3 = 0.34.4 for real systems. (For twodimensional systems, /3 = 0.14). In general, diffusion is defined byflux J (ie., amount dissolved by surface area).
J = Db(dC/dx)
)
where C is concentration, x is distance, 1) is diffusion coefficient, and Db is the bulk diffusion coefficient at steady state. In percolation theory, the concept of the dimensionless quantity D,given by
L)
=l)b/Da
(28.29)
is introduced. D, is the aqueous medium diffusion coefficient (Stauffer and Ahorony, 1985). These authors showed that
I) = L ) b / l ) ,
EA,(&
 E,)’
(28.30)
where EA, is a systemdependent constant, and (13 is a universal constant that has the value 2.0 for threedimensional matrices. The fraction of drug, denoted A (g/cm3), becomes the porosity Ed, in the exhausted part of the matrix, and Eq. (28.23) may be written A
&dP
E
(28.31) of the matrix mixture, hence, becomes (28.32)
The total porosity
E
= Ed
+
E,
where E, is the tableted porosity (i.e., the porosity before dissolution). The squareroot law, under these circumstances becomes
Q = { [DeES[2EdP E
q t ) 1’2
(28.33)
3 where 1, is the effective (traditional) diffusion coefficient of the drug substance in the medium within the pores. This term is a function of the tortuosity z and, according to the theory of percolation (Siegel, 1986),
Dh
= D,Ea/Z2
(28.34)
There is a substantial tortuosity effect at low drug loadings, but at higher drug loadings it becomesclose to unity, so that its effectmaybeneglected.Towgen glin (1998) have reported tortuosities at various loadings of hydrocortisone, nto polyethylenevinyl alcohol (EVAL) matrices and found that above a loadin of11+% the tortuosity becomes 3 or less. , suggests that, in a lattice, the sites can be either occupied by A mponent, and where the A threshold depends on (a) what type of lattice is created, and (b) at which concentration of I3 this substance is dominant in Caraballo et al. (1996) and Milan et al. (1998) have shown that there is a linear relation between the particle size in matrix tablets and the drug percolation threshold; in these studies, they used KC1 and caffeine as model drugs.
omb (1983) has reported work on percolation threshold in Eudragit matrices. The LeuenbergerBonnie equation may be written
p = C(E  E,)
(28.35)
where p is the slope of the Higuchi plot, E is total porosity (including porosity created by dissolution of drug andexcipients), and E, is the threshold porosity. Therefore, it should be possible to plot the slope of the Higuchi plots versus total porosity and determine the critical porosity by the xaxis intercept. That this is so is exemplified in Fig. 28.5. The figure shows that the threshold porosity is (intercept~slo~e)
E*
= 0.11181/1.461 = 0.076 (28.36)
As will be seen inthe following, most sustainedrelease products are based on the use of polymers of high molecular weight. It should be pointed out at the onset, that a polymer with merely a high molecular weight does not always qualify as a good sustainedrelease candidate. Even for some polymers that are generally known to have the desired characteristics this may not always be true. Kassem et al. (1978) have shown that many polymers, for instance, polyethylene glycols (PEGS), polyvinyl ~yrrolidone (PVP), hydroxypropyl methylcellulose (HPMC), and methylcellulose (MC), complex with drugs, such as spironolactone, and i these cases they enhance (speed up) ~ i s s o l ~ t i o n n rates. The general wisdom, however, isthat high molecular weight polymers will aid in the formation ofsustainedreleasedosage forms. For instance, Loftsson and ~ridriksdottir(1998) have shown that PVP, CMC, and HPMC complex with yand pcyclodextrin. In fact, when the task is to enhance dissolution rate, one commonpractice is to cogrind the drugs with polymers, such as HPMC (Mitrevej et al., 1996) chitin, and chitosan (Koh et al., 1986a,b), microcrystalline cellulose ( Nakai et al., 1978), or gelatin (Kigasawaet al., 1981). Shinet al. (1998) studied cogrinds of furosemide with crospovidone (polyplasdone,PVP). This i s the opposite ofsustainedreleasin
0.8 r
y =  0.1 1181
+
1.4610~ Rh2 = 0.980
0.0
0.1
0.2 0.3 0.4 Porosity
0.5
0.6
Aspects of a rigid matrix. (Data from Towgen and Binglin, 1998.)
substance, and is mentioned here, because some of the agents used in cogrinding, when used otherwise, work as sustaining agents (e.g., H P M ~ ) .
~ ~ d ~ oare defined “as networks of hydrophilic polymers which can absorb a ~ e l s si~nificant amount of water ( > 20% of their dry weight) without dissolving or loosing their structural integrity” (Vervoort et al., 1998a,b). Polymers of this type are albeit the swelling may be due to other causes, such as van der llites, hydrogen bonds, merephysical entanglement, or ionic polymers will, at least, swell somewhat in water, and the most common llable matrices is HPMC. There are a multitude of examples of ddu et al., 1993). An outline forusing of the possibility rations of the swellab~e (1982), have matrixdesc type been ~ o ~ ~ u l a t i n ~ Release s~ thaa ir ~ a cde ~ ~ i ~ a l with o d u c t s ~ ~ ~ ~~ There is, first, a protective gel layer formed, and then two mechanisms ensue (Fig. 28.6). The pseudogel allows additional liquid to penetrate into the tablet, and this extends the gel layer a further distance into the tablet. The outer gel layer then starts to hydrate more fully and may be dissolve in the dissolution media. A steady be reached, at which gel layer formation rate is balanced versus rate wever, the situation may be such that the sloughing off of gel is the controlli~g factor; then, the tablet simply becomes an erosion tablet. As shall be discussed further, some authors employ a power function for release (Solinis et al., 1998); that is, the amount not dissolved as a function of time will be given by (28.37) hen both processes happen at similar rates, then erosion and further wetting of the tablet will continue until the all of the tablet has been penetrated and wetted, and now erosion continues until all of the tablet has sloughed off or dissolved. The rate of diffusion is dependent on the molecular weight and the ““network,” se also affect the strength. he first work in relation to swellable dosage forms is attributable to Lapidus rdi (1966, 1968), Lapidus (1967). Lapidus and Lordi prepared granulations
/t”
Dry Interror
Hydrated “Skln”
Swell~ng,Formation of Outer “Skin” Diffusion
Erosion
Principle of how a hydrogel matrix functions.
the drug with the base materials and granulating with USP ethanol. For benzocaine, the solubility in USP dissolution media is very low, and the rate of release reduces to
) have given an exact solution to the swellable matri
with a ~ominalviscosityof15,000 c MC compositions) swell and do not disintegrate (Touitou a,b) used 25kN pressure on a 500mg 6 1 preparation of PMC (15000 cp). Tablets were made by direct comprestime plot was linear after 30% released and less th released, Solomon et al., (1979b) in another publication reported the use of in a study of sustained release (using KC1 as a tracer) for the 1. It may be directly compressed. 2. One obtains a gelled surface that is plane and uniform. 3. It is nonionic so that one avoids interaction with the tracer ( ionic. . It exists in a large range of viscosities.
It wasshownin Fig. 28.6 that the manner inwhich HP C matrices function involves penetration, swelling, diffusion, and erosion. Huber and Christenson that the erosion funct of HPMC matrices was linear in time owever,in other system amba et al., 1979), the decreasefollows a cuberoot law. The rate with which the gel sloughs off is, at times, referred to as ‘‘~isintegration,” Huber et al. (1966) used HPMC 4000 cp in their work on swellable matrices
y
= :
400.535  59.714~ R”2 = 0,992
0
2
3
4
5
6
Time (hr)
. Erosion function of HPMC matrices. (Data from Huber et al., 1966.) 7
and found disintegration time” to decrease with increasing gum concentration, but noted that the behavior would differ, in quantitative terns, with different drugs and useful for sustainedreleaseformulations were gums that would dy temperature, and they found sodium carboxymethyl cellulose ( ~ a C ~ and)~ C F to be such. For some they found constantrelease rates in ~ C certain time intervals When concentrations of HPMC become very low, especially in the poorly soluble drugs, such as naproxen, the intrusion matrix. system turns into an erosion system. Chiao and Kent (1983), for instance, used 4 9 % of H when even lower percentages are used, then the dissolution may follow a cuberoot law, rather than a squa~erootintime law, as exemplified by the figures in Table 28.1. The first two columns in Table 28.1 may be plotted to give the dissolution profile shown in Fig. 28.8. Note that the curve is smooth and that there is no ‘“lag” (i.e., the dissolution appears to begin right away). The third column in the table is the square root of the time points. If the amount dissolved is plotted as a function of the square root of time (Fig. 28.9), then a s t r u i ~ h ~ does not occur. !it is noted that this line cannot be explained away by the existence of a lag time, because the trace in Fig. 28.8 exhibits no lag time. It often happens that plots of this type are treated by the linear portion being extrapolated to zero percent release (in this case giving t1’2 = 1; i.e., t = l), and it is then concluded that there is a lag time. Lag times should be extracted only from linear plots. The fourth column in the 28.1 is the fraction not dissolved. This is obtained from subtracting the percentage dissolved from 100.00 and dividing this number by 100.00. The last column represents the cube root of the figures in the fourth column. If these figures (the cube roots of the fraction of drug not dissolved) are plotted as a function of time, then a neat straight~ine occurs (Fig. 28.10).
.I ~ i s s o ~ u t i o n Figures of a 3.7% HPMC as a Granulating Agent
Time (h) released
0.0 1.0 2.0 4.0 6.0 8.O 10.0 12.0 14.0 16.0 18.0
%
oderately Soluble, HighDosage Drug, Using
~

~
_
_
released
0.0 8.0 15.0 26.0 36.0 45.0 52.0 60.0 67.0 72.0 78.0
Squarerootoftime
0.0 1.o 1.414 2.0 2.449 2.828 3.162 3.464 3.742 4.0 4.243
Fraction not reIeased
1 .o 0.920 0.850 0.740 0.640 0.550 0.480 0.400 0.330 0.280 0.220
Cuberoot of fraction not released
~~
1.o 0.973 0.948 0.905 0.863 0.821 0.785 0.739 0.69 0.657 0.607
Ti me (hours)
Data from Table 28.1 where the amount of drug released as a functionof time is plotted, simply, versus time (column 2 versus column 1). Note that there is no time lag, the dissolution starts immediately.
1 00
80
60
40
20
0
0
1
2
3
4
5
Sq. Root of Time
Data from Table 28.1 plotted against square root of time.
y = 0.99398

2.1345e2x
RA2 = 0.999
Time (hours)
Graph constructed from data in Table 28.1 plotted by cube root.
proxen in which only4+9O/~ of lecular weight is 80135,000. These amounts of polymer are low; in fact, so low that it is difficultto imagine an actual matrix of polymer, and there is a minimum amount of polymer that will form a ~ontinuous phase, given by the amount of finedrug that can be %dhered’*to the solid drug. This problem was treated in an earlier chapter, and the ~ i n i m u m a ~ ~is u function of the surface areas, by the formula of Nystriim et al, (1982) a ~ t and Adolfsson et al. (1998):
=4 ~ A , / A ~
(~~.39)
is the smallest (critical) weight ratio of polymer to (high concentration) drug, A, is the actual surface area of the polymer particles, and AL is the actual surface area of the drug (whichhereis the component in high con cent ratio^). R is the surface area (m2/g) of polymer to the s ~ e c surface area of ~ c ratio of the s ~ e c i ~ c the drug. reased by increasing the surface area of the polymer ( e g , by this is doubly effective because it increases both the value of R. and that of A,.
Christenson and ale (1966) employed direct compression using onethird or more C. They showed that tablet hardness did not affect dissolution rates. The molecular weight of the polymer, however, was, and HPMC viscosity grades loo, 4000 and 15,000cp yielded tl12values of 1, 4, and 5 h.
( olomon et al. (1979a,b) used hydrosypropylmet~ylcellulose study of sustainedrelease (using KC1 as a tracer). It followed dependence with a lag time and showed the effect of the amou release (halftime) of KCl. The effect of concentration on the slope of the squarerootoftime plots is shown in Fig. 28.1 1. There is no consistent (or only negligible) effecton the lag time: 100 cp gives much faster release than a comparable formula with 1500 cp. The same was found by Huber and Christenson (1968), who found hardness to beof no i~portance.Aneffectofviscositywas demonstrated, and the valuesof tl12 for loo, 4.000 and 15,000cp grades for the particular formulations were, approximately, 1, 4, and 5 h. Lapidus and Lordi (1966,1968) and Lapidus (1967) also found that a squarerootintime relation held true, and thatit is the drug di~f~sivity, not the dissolution of polymer, and the water penetrability that were of importance in their system. ecause of the low solubility, the authors used the formula
/ P 2
= a[2DEASp2 = at3
(28.40)
y = 3.3121  3.2094e2x
R A 2 = 0.994
20
30
40
50
60
70
80
Percent HPMC
Release rates as a function of percentage of polymer. (Data from Solomon et al., 1979a.)
for the release, where G is the squareroot dissolution rate constant. If the tablet contains soluble diluent (in a volume fraction X ) and a poorly soluble drug is used, and the volume fraction of diluent it greatly larger than the other two contributors to porosity, then Porosity=&+A+X=ca X (28.41) Hence, G should be proportional toX 2 . Lapidus and Lordi found this to be approximately true (Fig. 28.12).
Attempts have been made to modify PMC to tailormake it to certain sustainedrelease requirements. Schor (1979, 1981, 1982) hydrolyzed H P (Methoce ~ ~ by exposing it to high humidity. It was mentioned that a previous patent (U.
x N
I
0
20
40 60 Lactose (%w/w)
80
ig.
1968.)
Effect of porosity on rate of drug dissolution. (Data from Lapidus and Lordi,
The carrier material was always thoroughly intermixed with the medicament. Schor et al. which is different from adding liquid water.43)
.
r
Some authors use the following presentation mode for dissolution of sustained release (and other) dissolution curves:
~1~~ = Kt. Lowey aand Stafford (1972) and Lowey(1979)use HPMC E50 (or HPC) humidified to 525% moisture admixed with 20% ethyl cellulose ( ~ 4 M (e.000.. . 1988. Methocel E50 and Metalose 60SH50. 1070492) in the absence of added water to form sustained release tablets. and a methoxy content of 1624 wt%.000 is effective. Although the actual mechanism isnot known.g. the tablets proved sustained release despite the failure to forrn soft rnucil a ~ i ~gel. In their invention. had the Higuchi equation held. mucilaginous gel barrier. for ) vitamin tablets). (1981) speculated that the slowerrelease rate arises from a decreased rate of swelling or a lower water solubility.790) employed up to 25% moisture and then obtained sustained release by controlling the degree of compression. of 93. behaves in an entirely different manner on contact with water.5% could be present.. The rate of solventuptake has been discussedby several authors (Peppas et al. Cao et al. 1996). K is here denoted the power dissolution rate constant.5. 1993. the water being added by a h~midificationprocess. resulting from hydrogenbonding interaction between the carboxyl and the carbonyl groups that had been subjected to both hydrolysis and oxidation. and n the dissolution index. MW > 50. When humidified and air dried in accordance with US Pat 3.171. which was either in a powdered or a solution form. . mucilaginous gel barrier on the surface of the tablet.~which is obtained when the higher molecular weight HPMCs are used. (1981) reports that a 4000cp grade of HPMC having an M . and forms little or no soft. The molecular weights were higher the time than at those used in the past and he used an amount of modified HPMC less than about onethird the weight of the sustainedr~leasedosage form. e.226.[but] have found that a similar tablet prepared from 50 cps HPMC having a mean molecular weight of 23. British patent 1.
Schor et al.000.3. 1980. Vergnaud. Ritger and Peppas (1987a.
by virtue of its ability to form a soft.g.870.
(28..b) use the following equation for the uptake of water in hydrogels:
4t14w = kt"
(28. He further improved the carrier base utilizing an HPMC grade with the following characteristics: Methocel K4M and K15M.849. Ranga Rao et al.691 discloses a product based on the aforementioned two polymers with undefined amounts of water.790 and 4. as little as 0..42)
where y1 would have been 0. and in one case K100.870. o s
HPMC maybeused in combination with ethyl cellulose(British patent.
Lippold et al. A precision hole is drilledinto the film. and in particular. with sorbitol. water). The mean dissolution time (MDT) (Mockel and Lippold. The electrolyte and drugwill dissolveand form a saturated solution of both. and its compressibility.. 1998). these would dissolve and create a porosity network co all strand and Ekman 1983. This approach has been mentioned in Chap. A core tablet containing the drug substance and an electrolyte (e. However.. 1988) was used explain the development of to pore clusters and conducting channels that would span the compression coat.where qt is the amount of solvent absorbed at times t and at infinite time.g. then this will create an osmotic pressure P.. N of 1989) for a maximum time of . which are semipermeable. at the onset.. sodium chloride) are coated with a film that is waterpermeable. and shells (Muzarelli. is given by way mean value integration of Eq. this latter will penetrate into the interior of the tablet (by diffusion and. the pore network. Stauffer and Aharony (1985) developed percolation theory to the problem. 1977.. 1993. Verhoevenet al.on exposure to the dissolution liquid. Fryklof et al. Thornbre et al.. being the principal component of crustaceans. 1983. The release of drug from these is a function of such parameters as the amount of polymer. The compression coat contains polymers. also through the hole). 1989).13..44) Sustained release by co~pressioncoating of tablets and porosity controlling the release has been suggested in the literature. Chitosan is /?( ~)~amino~deoxy~g~ucose obtained by Ndeacetylating the polysac1 and is charide chitin. insects. Zentner et al. There are limitations to the approach (e.)
The osmotic pump princi~Zeis demonstrated in Fig. and this (Siegel. did not behave exactlyas predicted. 28.2: (28. 1989). Chitosan has been used inhydrogel formulae (Porter0 et al. insomecases.given by
Waterper~eable Membrane
3 ~ c ~ e ~ f of tosmotic pump tablet. the effectof the compression pressure on the physical characteristics of the polymer). f ic
. If the solubility of the electrolyte is St mol/L. In contact with a dissolution liquid (e. but waterinsoluble. both to the dissolving liquid and to the drug substance (Conte et al. (1967) employedsoluble porositymodifiers to the (otherwise waterinsoluble) compression coat so that.g.g.. 25. (This is a substance that is abundant in nature. its surface characteristics. 1985. 28.
and this is the “release rate. ~yethylcellulose. where C K S is a steadystate co~centration of drug in the efflux liquid).
One of the p r o b l e ~ s encountered in sustained release is the gastric emptying time.g. will be forced out with a velocity given by
where
I. This efflux is proportional to
P. and h is film thickness. and it is oftea not ~ O s s ~ ~ Z e by
.” This a zeroorder release (Le. it is replenished by ( ~ u l k liquid ) diffusing into the interior and this.47)
and where r is the radius of the hole.and so on). from fasting versus nonfasting conditions.
The amount of drug leaving the tablet per second is then US. the most desirable form of release). the solution.where z is the ionic equivalence number of the electrolyte (e. in turn. The linearity will be lost once all the electrolyte or drugsubstance is exhausted. supposedly. hydroxypropylcellulose. In the steady state there is therefore... q is the viscosity of the liquid (in this case. but less than the electrolyte. but many of the release profiles are fairly close to one another in appearance. 2 for NaCL. containing S g/s of drug. and e approachin overcoming this is to produce a tablet thatwill “float” in the eth (19’78) described a composition consisting 080% of a dru substance 50/0 of either methylcellulose.1 exists between the liquid at the interior of the tablet and the bulk solution outside. I.
2. As the liquid is forced out through the hole. Efflux of v cm3/s.or sodium carboxymethylcellulose (or mixtures compositions resulted in products with buoyancy and were f o r ~ u ~ aasea~twot layer tablet with the composition such that itfloats in gastric juice. saturated in electrolyte and drug substance). prolongs the residence time in the stomach. 3 for CaC12. = ~ ~ ~ 4 / ( 8 ~ )
(28. issolution of v S (or vC. The drug in solution also contributes to this. so that a balance is established whereby the liquid influx equals the liquid efflux. therefore. dissolves electrolyte and drug. Influx of vcm3/s of dissolution medium. he principle is that such amounts of PMC and excipients are used and compression pressure so adjusted to correctly adjust the apparent density of the tablet. This. rs from patienttopatient.
The principles behind singledose sustainedrelease products have been discussed in the foregoing. These models give rise to certain profiles. A pressure differential of about P . Liquid.
95 and 0. is amount released ko is a zeroorder rate const t. (1993). Peppas (1985). and k is a rate constant. giving correlation coefficients between 0. It might be diff~rent. 1983. (1983). (b) slope of abbreviated a = initial diameter of an erosion tablet = concentration (g/cm3) of a diluent in a matrix C = ~on~entration
. to choose between an erosion and a diffusion m o ~ efor a wax for l tablet.b) given by (28. statistical scrutiny will fail to show differences in the models.53) was a good fit. A eibull equation might also fit. A40 is initial amount. and the rate constant for the erosion model is of inversely proportional to the size. instance. kiLIis a of iguchi rateconstant. All these equations have been discussed in the previous sections.it
~ e a E of the s h a ~ e the d ~ s s o ~ u t curve. authors then attempted fitting theprofiles by methods suggested by yer et al. what would be learned from that?
A = concentration of drug (g/cm3) in a matrix a = (a) diameter of an erosion tablet. but then. ood and Panchagnula (1998) investigated release profiles ofd i l t ~ a ~ efrom a m series of comm~rcial sustainedre leas^ preparations and attempted to fit them to the follo~ing equations:
= kot
(28.but aside from that there seemed to be no c o ~ m o n thread in the profile fitting.and kHc is a xsonCrowell rate constant. but because the diffusion model would be (fairly) indep~ndent tablet size.99 but it is simply a type of curve fitting. The authors found a linear relation between k l and kHC. given by
where F is the fraction released at time t. then many models should be tested. usually.. 1987a. and E ~ .itger and Peppas.49)
Q = kHt1l2
(28.51)
is a ~ o ~not t n released at time t . to deduct ~ h i of the ~ e ~ h a n iis at s s of i~n c ~ s~ ~ ~ Yet this is often being done.50) (28. ~dditional criteria can allow choosing one model over another. They also tested a orsenmeyer et al. It is logical to choose the former.53) where E is a constant and Qoo is the amount released at infinitetime. kl is a firstorder r e constant. The two latter suggest a general profile obtained from a spherical matrix. Equation (28. extra experiments might easily facilitate making a choice between the two. If so. and Karajgi et al.
(b) frac~). (b) volume of an erosion tablet x = distance z = number of ions into which an electrolyte dissociates = (a) erosion constant.) D. (b) thickness the of film in osmotic pump tablet J = flux: K = cuberoot dissolution rate constant KB = powder dissolution rate constant k = coefficient in the power function release rate equation kH = rate constant in Higuchi kHc = rate constant in HixsonCrowel equation ko = zeroorder rate constant kl = firstorder rate constant L = depth of liquid intrusion A = undissolved drug 4 Mo = intial amount of drug MDT = mean dissolution time m = mass of an erosion or matrix tablet mo = initial mass of an erosion or matrix tablet N = number of particles in a sample yt = exponent in the power function release rate equation P = osmotic pressure Q = (a) penetratioin equation constant. (b) exponent correlating percolation porosities. (b) radius of hole in osmotic pump S = solubility of drug (g/cm3) SI = solubility of electrolyte T = absolute temperature t = dissolution time t. = lag time I/ = volume of dissolution medium v = (a) velocity with whicha liquid exits from osmoticpump tablet. =f porosity accessible to intrusion E~ = p / A = porosity contributed by the dissolved drug E.ter
D = (a) diffusion coefficient. before dissolution
. = porosity from compression. = aqueous medium diffusion coefficient D = bulk diffusoin b
I . (b) dimensionless diffusion coefficient (&/D. tion released at time t h = (a) depth of a matrix in which all drug hasbeen dissolved. (b) amount of drug released per unit of surface area q = firstorder penetration constant qt = amount of solvent absorbed at times t qi = amount of solvent absorbed at infinite time I? = ideal gas constant r = (a) the average radius of pores. = (a)
= ~ ~ r 4 / ( 8factor in volume expression for osmotic pump. (c) slope of the Higuchi plot E == total porosity of the exhausted part of a matrix E* = percolatin threshold porosity E.
Grassi M (1996). Ostholm GIV (1967). Dow Information Sheet (1982). Percolation Structures and Processes. Ju TR. Edgar PJ (1987). Caramella C.317. Ford JL. Said S. J Controlled Release 9:lll. Yoshida M. Vyas SP (1993). Yamanaka H.143. Stemm NL. Pharm Phamacol 37:33. New York. eds. J Membr Sci 113:21. Marty JP. Zallen R. Caraballo I. Fini A. Ekman B (1983). McCaul F. Canadian patent 1 204 1671. Rabasco AM (1996).Mashimo T. Fessi H. Shehata E (1978). Kooyama 0 (1981). Puisieux F. Jamali F. Dale LB. Int J Pharm 23:327. New York. patent 3. J P h a m Sci 52:1145. Skoug JW. No 192886682.704. J Pharm Sci 71:749. Fouli AM. Christenson GL (1966). Adrover A. J Pharm Sci 55:974.In: Deutscher G.
. Kassem AA. Belgium. Int J Pharm 163:23. J Pharm Sci 72:772. MJ. Jain NK. Domb C (1983). Higuchi T (1963). FernandezHervas MJ.Imai K. patent 3. Gao P. Asano M. Hogan JE (1985a). S.q = viscosity of dissolution liquid p = coefficient in the relation between diffusion coefficients = integration constant in penetration equation = (a) squareroot in time constant. Puisieux F. PressCoated Systems for Drug Release Control. Yuasa . Dale LB (1966). Milan M. Int J Pharm 23:339. J P h a m Sci 12:732. J Pharm Scl 57:164. Sung KC (1996). Hogan JE. Giona M. Rubenstein MH. Grant DJW (1993). Christensen GL. Dissertation. Docteur Sciences Phamaceutiques. Academic Press. Sandell E.065. Kent JS (1983). Int J Pharm 160:187. Belgian Patent 623. P h a m Res 10:1648. Fryklof LE. del Cerro J (1995). Suzuki K (1989). Kallstrand G. Carstensen JT (1979). (b) exponent in the relation between diffusion coefficients pA = particle density of the drug substance in a matrix I pe = particle density of an escipient in a matrix 5 = tortuosity
Adolfsson Caramella C. Int J Pharm 3:87. Kumakura M. Rube~stein MH. Rubenstein MH. Karajgi J. La Manna A (1983). Vakilynejad M. British patent 1070492. Rubenstein MH. Kent J (1993). Couvreur P (1975). U. J Ford JL. Fukuzaki M. Ford JL. Watabe K. Tnt J Pharm 13:387. Christenson GL (1968). Carstensen JT.
A. Carstensen JT (1982). Huber HE. pp 1740. Huber HE. Holgado MA. Carstensen JT (1981). New York. Dow Handbook on Methocel Cellulose Ether Products [Table headed Viscosities of Methylcellulose of Various Molecular Weights]. p 87. J Drug Targ 1:1997.SolidPharmaceutics:MechanicalProperties and Rate Phenomena. 101:733. S. Duddu SP. p 220. Pleunum. Marty JP. Conte U. METHOCEL. Medicinal tablet and a method for its preparation. University Catholique de Louvain. Nixon PR. Bamba M. Ford JL.394. Fell JT (1998). Hsiao CH. Canadian patent 1 204 671. ~ a r u y a m a Tanaka M. Kigasawa K. American Institute of Physics. Hogan JE (1985~). Fernande~He~as Vela MT. Columbo P. Int J Pharm 113:39. U. Adler J. Valentine 'u' (1966). Yakugaku Zasshi K. Hogan JE (1985b). Bull Fac Pharrn Cairo Univ. Int J Pharm 40:223. Hsiao CH. Nystrom C (1998).
.
e n Zentner GM.Danhof M. Int J Pharm 172: 137. Pbys Rev 17273. Rombaut P. Van den Mooter C. Int J Pharm 172: 127. Kinget It (1998b). Augustijns P. Kinget R (1998a). Jungmger HE (1989). SchutteSC. Van den Mooter C. Washburn EH (1921)..Peschier LJC. Vervoort L. Vervoort I.
. J Controlled Release 10:205.. ~ i ~ ~ e l s tKJi (1985). Augustijns P.s
Verhoeven J. Rork CS. J Controlled Release 2:217.
This Page Intentionally Left Blank
.
9. Use of a Mixed Film and Multiple Films 29. d e i ~ end and for that purpose the reader is referred to the ~ e c o ~ ~ e ~~ e ~ d at the g of
. Coated Particles 29. Ethyl Cellulose Films 29.29.8.6. Sized Particles 29.11. Nonsink Conditions 29. and raw material selection.
494 495 496 498 499
50 1
503 503 503 504 504
is continuous Films
Films 29.2. The aim of the chapter is to cover ~ r j n c ~ and ~ Z e theory of products sustained by sizing and by coating.10. FilmThicknessCoated Granules 29. Multiple Osmotic Pump Principle 29.13. Coated Nonpareils 29. Other Films
505 507
508
Symbols ferences ecommended Reading
509 510
Attaining sustained release through particle size manipulation and coating are the subjects of the following. examples.1.12. ~ontinuous
29.4. Tableted Microcapsules 29.I. Application of Films 29. It is not the purpose to cover details about processes. The former will be treated first.5.3.
the dissolution profile given by Eq.1 andcolumn 2 in Table29. for a sphere is given by
K = 2kS/(rp) = Q / r
(29. The release patterns can be adjusted a bit by mixing the best fraction with small amounts of other fractions to “adjust” the profile. In that case the results of fraction released are those shownin Fig. notation will be used in the immediate following:
= mass not dissolved = initial mass before dissolution K = cuberoot dissolution rate constant t = time p = density k = intrinsic dissolution rate constant (cm/s) S = solubility (g/cm3)
r = radius of a spherical particle = 2kS/r t = time
As an example. naproxen). nitrofurantoin.O and r = 100pm. 7080Y0 in the our. Naproxen is another substance that gives a (alnlost) 12h in vivo release when as N C will prolong this to simply a ~ ~ i n i s t e r e d a particle. whichlargely is from submicron range to about Nevertheless. It is seen that the plain mesh cuts do notmeet the requirement for the first and third hour. It is often possible.1)
(29. by manipulation and mixing of mesh cuts to obtain a desired release profile. (29.1. whereas mixes do. Small amounts of 24 h.
.The second and third column show release rates for particles of sizes 7’5 and 50 pm.g.2) could allow calculation of some size r. The problem is that the valuesof r couldbe outside the pharmaceutically 500 pm acceptable range.. uppose one desired to obtain3 0 4 0 % released in the first hour.3) would be “sustained” to a set of specification one might require for sustained in vivo and in vitro release patterns. and tosome degree. (29. Some coverage of processes and raw materials willbegiven extent needed for covering the principles and theory. there are substances for which this is a practical solution (e.{ M / N o J ~ / Kt= where K . assumethat Q is 1.2)
which in Cartesian notation takes the form (29 3)
e
It is easy to visualize that Eq. andover 80% after 6 h.the chapters.
to the
It is known from previous chapters that particles will (often) dissolve by a Crowell cuberoot law: ~ 1 . at which the value of K and hence. 29.
896 0.334 0. which is most often waterinso~uble.891 0. 2.745 0.96fa3 0.919
0 0.717 0.964 0. the spray is halted intermittently.992 1 1
50pm
Ratio 1 : l : l
0 0.626 0.784 0.557 0.
Particle size manipulation is not a common way of producing sustained action. 4.657 0.350 0.992 1 1 1
.but waterpermeable.818 0.37 0.936 0.. (29.856
Ratio 1 : 4 : 6
0.784 0.908 0.0 and Different Size Particles
Time
100pum 0 0.790 0. This is continued until “the desired film thickness” is acquired.Time (hr) Ratios of 100 to 75 to 50 prn: circles 1 : 1 : 1.946 0. and drying by hot air is carried out.575 0.936 0. triangles I : 4 : 6.488 0.606 0.875 0.976
Ratio3:2:1 0 0. The most common method is by coating the particulate solid with a film. until a certain degree of wetness.271 0. Films may be applied in one o f four ways:
1.488 0. 3.
Release Rates According to Eq. the film solution is sprayed onto tumbling beads.3) with Q = 1. and so that the beads do not grow together.973
75pm 0 0.
Pan coating Fluid bed coating ~opreci~itat~on (CoasGervation) Evaporation
In method (l). This is not one continuous^^^ that is ~ r o d u ~ e ~ .329 0. squares 3 : 2 : 1.
(Many types of nomenclature e the exists. becomes a coatsieve principle.. waterpermeable film. or of drug and some excipient. The intrusion liquid will dissolve some of the drug and excipient to form a saturated solution. 28. the film isactually deposited by precipitation on the seeds or beads. a matrix.21. then a situation in Fig. and such films are. 29. and then up in a chamber and down again. and the film becomes continuous. but rather.but ruther a series o f ~ l o~ the same co~position. There may be some expansion of both the interior of the coated
Drug Diffuses Diffuses Drug Zero Order Firs
Saturated Solution Forqs. so that when a pellet (granule. the seeds or beads pass through a zoneofmist application. If the seeds or beads are completely dry when they reenter the spray zone. The drug release is akin to that of matrix tablets covered in Chap. the presenceof intrusion liquid and the subsequent pressure buildup may make the adhesion planes into pores. although only partly. 29. anchor together well. In a fluid bed (e. certain grades of ethyl cellulose).g. or only slowly soluble in water (e. but it is still. rather than a continuous film principle. is coated with the filmformer. completely continuous.g.
. certain grades of HPMC or HPC). just to be returned through the mist zone.in a Wurster apparatus). then the situation is the same as in the coating pan application. In method (3) which will be discussed in more detail shortly.. Each i n ~ i v i ~ ~ u Z drug particle is coated. ethod (4) is. bead) consisting of drug. principally. if the coating is carried and out correctly. but not with one ~ a ~ t i c u Z a rand ~ . but if they are still somewhat moist. not a film. a series of films. or can be.
TI There are films that are waterinsoluble ( e g . strictly speaking. is (or can be) drug ~ Z there particles in the surface of the film. The same may be argued for process (2). liquid will penetrate the coated bead. Expansion
9
Aqueous Llquld Penetrates
Film May also Expand
I
Solid Phase Present Inside Coated Bed
No Soiid Phase Left Inside
(a
(4
( b)
(c)
Schematic of a bed coated with a waterinsoluble.) At first [see situation (a) to (b) in Fig. f s The films.2will arise. so that the principle. The particle willbe denoted a coated bead or ~ i c r o s ~ ~in r e following. microcapsule being one. then the anchor between “layers” is actually continuous.
the concentration gradient is. then a schematic graph OAB will result (Fig..6)
(29. the last particle of drug will have dissolved. (This may be somewhat larger than the film thickness of the drycoated bead.7)
This will be treated further in Sec.~( A D / Vh)]t M [ ]
(29. Ifthe process is a ~ p r o ~ i m a t e d the three by zones (a) + (b). This is. the transition to first order.?
particle and the film itself.3 Idealized release profile of a situation such as shown in Fig. 29. Once a saturated solution is formed in the interior. 29. obviously. and (c)(d) after t*. where S is saturation.5)
(29. then
c =M/V
so
dM/dt = (AI)/Vh)M or. and t". However. occurs at B (6 h). g j occurs at A (i. a nonsteadystate period. and both that and the transition to poststeadystate will be gradual. 29. (c) lasting t* . and is a designated by the symbol t. the concentration gradient will be constant.t. transitions in (a)(b) involves some release. The following considerations assume that the film is c o n t i n ~ o This will be ~~. S l h .
12 Time (hr)
9.) At one point. and h is the thickness of the film at stages (b). and the rate will be given by dM/dt = ADC/h (29.4)
where M is drug mass not dissolved at time t. poststeadystate.2.3).e. t = t*. If the total volume of the interior of all the beads is denoted V .. and (d). and A is surface area. (c). at the interior of the bead. integrated: l n ~ ~ / = .12. under sink conditions at the exterior. proportional to the concentration C. I) is diffusion coefficient. discussed further at a later point in the chapter. hours. hours.
. lasting t. and from this point on.This period is denoted the Z g time in the following. at 1 h.
10) inserted in Eq.10)
Equation (29.8) then gives
. (29.The total weight of dry interior of the N coated beads is
so that
N =6 ~ / ( p ~ a 3 )
(29.
the length of steadystate. The larger h is 3. therefore. Solution. isforced outthrough the holes formed in the coating. is kept constant. again. Because the dry weight is given by W = N(g +f). as in Table 29. fills it. If the filmthicknessincreases. or the film must be sufficiently weak that the osmotic pressure formed on liquid intrusion will make the film yield and form channels through which the dissolved drug can escape). will disappear from the film. because blending of particles that do not percolate with place~os. then the number of particles N . then this component. therefore. The smaller D is . The film must be a twophase system to work in this manner (i. a reduction in q will make W smaller. (a) Water penetrates and swells the coating. and by Eq. with a rate of 6SDW/(hpa). on the other hand. $Jill w e i ~ h tis a ~ u s t e d s h o u l ~ p l a c e ~ o s a p p r o ~ i m a t e ~ y same size as the coated bead. increases.release rates go down.the value of t*. 3. if not impossible. or ethyl cellulose containing a watersoluble polymer (such as PEG). when the film is exposed to aqueous liquid. the w..8) the overall release rate. will decrease. the fill weight. the smaller N is) 2. In the former case the process that takes place when the microcapsule comes in contact with dissolution liquid (e. and the lag time t. 1. 2. to obtain a final release curve that is a linear combination of the individual release curves. saturated in drug substance.g. In the nonuse of the pareils there has been past practices in which the blank nonpareils were used as fillers. sugar.. However. resulting in devastating content uniformity problems. (29. e. will decrease.3) the the rate willbe
If. possible to blend fractions of coated beads of different film thickness.1withsized particles. f / C f + q) is increased. It is. There is a danger in this. Also. Desired release profiles. and (b) water dissolves the soluble plasticizer. and this increases the osmotic pressure P.g. The larger a is (or.one can be difficult. Water penetrates into the interior of the sphere. The smaller the solubilityis
(29.” so that the unit will act in a manner of an osmotic pump. dQ/dt..
. will increase.dQ/dt = ( ~ a 2 ) D S/ 6 h p ~ ~ 3= 6 S ( h p a ) W ( ) ~/D At a given fill weight and dose. weighted according to their fraction. may be obtained by manipulation of these ~ u ~ n t i twhen ade~uate i~s polymers are used. in a nonpareil seed).. be zeroorder. in the steadystate region (A to releasewill.
If an insoluble filmformer is mixed with a minor component that is soluble or leachable. leaving “holes. water) is 1. The most common coating is plasticized ethyl cellulose. i slower under the following conditions.11) in Fig. the minor component must be insoluble in the major filmformer. equivalently. 29.e. and becomes saturated with drug (and or excipient. but the drug concentration.
of volume release saturated liquid in the steadystate zone per hoZe will be
of
w1 = nPs4/vh
(29. Proportional to the solubility of the drug 2. of eight pore former = a H (29. The volume A.m~ltiplied by the concentration of drug. If the weight of the film is H .16)
E is constant for a given application.15)
where E =~ H / ( p ~ ~ h ) (29. and there are several ways of v i s ~ ~ Z i z i how they rz~ can be arranged in the film. the rate v. The total rate v is the rate per hole times the number of holes (i.. so the weight of the holes is ~4 ~ e i g of ~ h holes = h n p f ~ s 2= aH so that the number of holes E . so that
= d ~ / d= ESP~2/vh 4 a H S P 6 2 / ( v ~ 2 p ~ ) t
(29. A terminal phase results where the concentration. Proportional to the diameter of pores squared 3.19)
is a constant for a given application.18) where
G =4aH/(pf)
(29. The bead works as an osmotic pump (i. which is S during the steadystate period.. steadystate portion is 1.then the following holds.20)
It is seen that the release rate in the linear. Inversely proportional to the viscosity of the solution saturated in filler and drug 4.e. 5. of the rz particles is given by Volume of hole material = hrzxs2/4 The density of the hole material is p‘.17)
where q is viscosity.e. may be expressed as
YI
(29. C inside the sphere. decreases until it equals the concentration in the bulk liquid.14)
= 4 t ~ H / ( p ’ n 6=)E/a2 ~
(29. (29. using the expression for n derived earlier).13)
(29. Inversely proportional to the film thickness squared
.4. and a fraction of it is such a pore fomer a. The total rate of release (dM/dt) is the rate of volume release w.12) There are rz holes of diameter 6. This continues until all solid drug in the interior of the microcapsule has been dissolved. Coated beads are made by conventional means and the “weight” of the holes is.
is not necessarily a continuous film. the seeds or beads are simply loaded into the appropriate apparatus (pan. so that process reproducibility is of great importance. it should be recalled. (29. ~ranulations made and sized to a certain particle size are range (e. because the holes are notgoing to be cylinders. (29. when an initial amount has penetrated by diffusion.
. on the other hand. one for the polymer.23)
For monodisperse populations. in contact with water. These concepts lead to the following point of view. This may. 29. and would be good pore formers. reaction vessel). they may be tortuous. by the relation H =~ p f ~ n ~ 2 (29.10) into Eq.. repeated here for convenience. It is obvious that the amount of film H that is applied relates to the film thickness of a monodisperse population of beads of diameter a. Inserting Eq. and the largest a. Often overlooked is that the solubility of the “filler” contributes in two ways: first. In coating with contin~ous Jilms.that a film deposited by ~uidbedspr~ying method. It is an approximate picture anyway. the smallest diameter is a.The diameter of the pores squared is a processdependent term.
HIW =~prhl(pa)
(29. be the method of release of many continuousfilm products. (29. and channels may develop in the adhesion planes of the various layers of which the “continuous” film is made up. In some applications. They reduce the glass transition temperature by being soluble. (29. therefore. well anchored together. seems fairly insoluble in high molecular weight ethyl celluloses. in increasing the osmotic pressure P.20). Ethyl cellulose and HPC. but the overall effect is well described by Eq. Hole material is best insoluble in the polymer. However.21) then gives:
H = 6Wprhnn2/[p(na3)] 6 W p r h / ( p a ) =
(29. the osmotic pressure developed by dissolution of filler and drug substance causes a stress of the film.. The weight W of the beads is given by Eq. one for the pore former.9) W =~p(na3)/6.22)
For a desired film thickness h and bead particle size a the ratio of weights of filmformer to dry seeds or beads is.second. the poreforming model should not hold. where p is the density of the solid.fluid bed dryer.
TE
The effect of film thickness h has been mentioned on several occasions in the foregoing. and can be calculated from Eq. and in this case a thermogram (or better yet a torsion braid balance profile) shows two transition temperatures. Fig.4). The pore material is selected as a material that is insoZu~Zein the polymer. Polyethylene glycol of certain molecular weights. but rather. of certain grades are mutually soluble.21)
where p r is the density of the film.. and the film thickness will be (fairly) uniform. Again.
.g..21).. could be a series of films. in inc~easingthe viscosity. (29. actually. Plasticizers are usually not good pore materials becausethey are soluble in the polymer. and when using EC in such applications. the perco~ationthreshoZd must be exceeded.
(28.29) where
fi = % / N
is the number fraction of particles in the ith interval. N applied to a particle is assumed proportional to the surface area of the particle.24)]:
t N l = h i ( ~ 2ip~) a ~/ ? n i 2 a ~ = ~
(29.25)
The film thickness. The amount of film. so that the amount applied to each fraction would differ. this maybe written
(29.(4
(b)
(c)
Principle of thickness control by way of particle size distrib~tion~. hen N is large.27)
The release of material.24)
where /? is a proportionality constant and ni is the number of particles in that fraction.PIP’
(29. the amount adhering to them would either be a function of the actual surface area. The amount N applied to the ith fraction of particle size 1 aiwould. of the ith fraction would be givenby [Eq. from the ith fraction is given by
=a ~ a dqildt = ~ S ~ i ~ n i ~~ lS ~ i ~ p l ~
(29.26)
from which:
hi . qj. therefore. hi.~~)
.25) and (28.28)
so that the total release rate is given by (29.
ecause the seeds or beads are in a stream of coating spray. be
N = P n i ~2a 1
(29. If the ith fraction acquires a film thickness of hi then
H~ = n ~ h i ~ a ~ p f
(29.
(29. whether these mixtures provide the mechanical equivalent of an osmotic pump (byleaching out the minor watersoluble ingredient). the multi~1e osmotic pump princi~le was developed soon thereafter by use of mixed coats. or whether
. The continuous films ofsubstituted cellulosetype polymersare often produced by copreci~itation.
. deposited by melting and cooling. rod~ced copreby cipitation. A wax coating. 29.
The original concepts are of the type shown in Fig. However. One way to obtain a continuous “wall” is bytempe~ature effects.3 1)
@{min)
where^(^) is the particle size number distribution. It is often uncertain. It is noted that the integral is the secondmoment of the distribution function. the tem~erature droppe is and the polymer precipitates out on the suspended active substance. A solvent dissolves the polymer at higher temperature. p2 is equal to the variance of the distribution ( ennett and Franklin.2. 1961). and thatthis. will also be continuous. in general. but waxes are rarely used asJiZms.
~ontinuous films (such as depicted in Fig.2) are. 28.
such as ~icrocrystalline cellulose. of 9 parts ethylcellulose to a part hydroxypropylcellulose . . leach out the quinidine disposed on the nonpareil seed. whereas if the film is discontinuo~s. even with the very strong films provided by high molecular weight ethyl cellulose is voiced by Hsaio (1985) who states that
[Tlhe coated aspirin is mixed with a compression aid. siao (1985) teaches the use of a mixed coat of ethyl cellulose and hydroxypropylcellulose and the use of coprecipitation in its manufacture. and their pharmaceutical use is attributed to Hsiao (1987) who makes a claim “where the plurality of pellets are compressed into a tablet. . of a coating of quinidine over a nonpareil seed. often in combination with other polymers. and Klucel LF.Thesehave exceptional strength. orif the minor component is insoluble in the filmformer. less force is required to compress the mixture into tablets thereby minimizing disruption of the polymer film coating the aspirin crystals.e.By incorporation of a compression aid. . hence. patent 3. if the film is continuous. . plasut. . The inventor states: “the more watersoluble hydroxypropylcellulose in the outer coating provides ‘channels’ for the water to enter and. .ethycellulose to.
IC
Hermelin (1963) in the early 1960s implied the possibility of tableting coated beads. “Each pellet has a coating of quinidine over nonpareil seed. . The author uses type 10 (10 cp) ethyl cellulose. Enteric activity is provided the surface of individual particles.441 (Hermelin. ractical solutions to direct tableting of coated beads is tied in with the development of veryhigh molecular weight ethyl celluloses. then the release of drug would occur by erosion and diffusion. . They found that drug incorporation was less efficient in EC microsphere when the viscosity of the organic phase was
. the thus quinidine coated nonpareils are coated thereon with a coating. The patent describes the application of a coat of quinidine unto nonpareil seeds. The lowering of the main peak of the major component without the appearance of a peak for the minor component will indicate solubility.” In claim 1 they state: “pellets. . 1963) disclosesa tableted composition comprising cles of medicament covered several individual layers ofentericcoating mateby rial and included in a matrix of medicament and filler. . and the multiple osmotic principle may prevail.” The concern of breaking coated pellets during compression. such as hydroxypropyl ethylc cellulose ( H P ~ C and hydroxypropylcellulose (HPC). . These two situations may be distinguished by means of glass transition te~perature determination. . in a ratio of 7 parts to 3 parts.this latter is present as a plasticizer (i. advocating more that one coat in the dosage unit). then leachingmay occur.115. hydroxypropylcellulose. over a period of time. The quinidine coated nonpareils are then coated with a mixture of.” (i.
IL
Ethocel (EC) has been used as a membrane substance in sustainedrelease beads.e.. forms a solution with the major ingredient). ) Guyot andFawaz (1998) made microspheres of nifedipine using ethyl cellulose/ HPMC by means of solvent evaporation. . ..
Usually.. 1989. and Rowe (1986) has shown the effect of molecular weight on the properties of the ethyl cellulose film. The ethyl cellulose may be applied as a solution in (20% methylene chloride in methanol) or in suspension form ( e g .g. 1993). Suspensions are usually added in fluid bed equipment (e. plasticization of the ethyl cellulose is accomplishedby either HP HPC. for instance. Yuen et al.. good linear release is obtained with the coated pellets. the less is necessary. The flux through the film is. The release patterns are shown in
As mentioned in Fig. t.. If this were not so.2. 29. used HPMC ascoating material with Ethocel. The higher the molecular weight ofthe polymer. it was enhanced by decreasing the C/HPMC ratio or the EC/HPC ratio.. When suspensions are employed. Lippold et al. originally. The solubility of the drug substance is Smg/cm3 and M~ = SV for complete dissolution to be possible. or PEGS. Uniglatt. 1997.and the stronger the film. The subsequent release profile was treated in the previous sections as dictated by sink conditions (Le. On the other hand. but not all microspheres made in this manner withstand tableting (Palmieri and Wehrle. Belleville et al. then the dissolution profile would be dictated by the following considerations: The dissolution medium will have a volume of V cm3. The surface area of the beads is A cm2 and the thickness of the coating is h cm as in past sections. 1990. When solvent is added and plasticized with HPC. a film strong enough to withstand tableting pressures may result. under nonsink conditions given by
0
1
2
3
4
Time (hr)
Release patterns of coated pellets. a lag time.)
. the concentration in the bulk liquid surrounding the coated bead were zero). There will. and the authors concluded that the nifedipine was present as an amorphous phase. Surelease. be a total of Mo mg of drug substance in the coated beads. Germany)..increased... the phases (a) + (b) require a certain length of time. thy1 cellulose coating to attain sustained release has been treated extensively in recent literature (Porter. ethylcellulose pseudolatex). to be established. The microspheres exhibited no burst effect. (Data from Palrnieri and WehrlC. (1979). Yang et ai. The importance of somevariables are demonstrated: for ancihni and Vechio (1989) have demonstrated the effect of loading on release. 1997. hini et al. 1997).
C)
(29.34) and results in the following: dM/dt = [DA/(h)I{S .38) reduces to (29. (29.[ M . If q is small. then the Eq.M ) / V
This is inserted in the expression
(29.{ D A / h V } { [ S V Mol = +(q*
+M )
(29.32)
It is noted that C is the con~entration the dissolution medium and that it increases in as a function of time.38)
.til where the initial condition./dt = (AD/h)(S .Mo) = q* Inte~ration this yields (noting the initial condition dictated by ti of In{(4* + M)/(q* MO>)= B{t .37)
+
(29. that M = Mo when t = ti has been invoked.35)
+M }
where
and ( S V .33)
(29.M)/VI] CO
= . The concentration in the dissolution medium is given by
c = (Mo .
1998.5% t(1/2) =41.42)
The treatment.5%]
ln[lO%]
c . and one of these is chitosan.. has gained considerable interest in recent years as a polymer to sphere formulation (He et al. Hassan et al. and the slope of the linear approximation is not unity..8. but 0. lends credence to the modefin
There are other sustainedaction polymers in use. The inverse relation is fairly linear.100
tl
80
60
40
t(1yt) =420
O
12. 1994. The reason for this is that. They are plotted in Fig. in actuality7Eq. 1994.1999..
. 
0
In(7. ~ a l ~ i e r al. et i
In[ 12.d 3 )
(29. 1992). Acikgoz et al.5. Conte et al. 29.. 1996.. 6 in publication by Palmesi and ~ e h r l ~ (1 997).2. Wehr16.5 a 5%
2
4
6
8
Reduced Time
Fi
Release patterns of pellets under nonsink conditions. but undoubtedly somewhat curved.7 are shown in Table 29.40) should read
= ( N p / 6 ) ((d
+ h)3 . Chwala et al.5a/o]
I
I
I
0
1 00
200 Time (min)
300
Graph constructed from data in Fig. 1994.1997.)
(Data from Palrneri and
The slopes ofthe lines inFig. (29. 29. nevertheless..5%
20
0 0
t(l/2) = 240 A 10% t(lt2)= 135 i= 7.
9673 5 8.5 3.
A = surface area a = inside diameter of bead a(max) maximum particle size of a granulation = a(min) mi~imum = particle size of a granulation C = concentration at the interior of the bead D = diffusion coefficient E = constant (4aH/(p'n)or 4aH/(p') H = mass (weight) of film Hl = film weight on the ith fraction of coated granules h = film thickness K = cuberoot dissolution rate constant f' = grams of fill of the coated bead that is drug k = intrinsic dissolution rate constant (cm/s) M = mass not dissolved Mo = initial mass before dissolution
.982
y
I =
. Santhino et al.05 1 0.7
(50
Coating 3 1
Intercept 0.5 1.989 0.998 0.992 0.
A host of other compounds are also being investigatedand reported in current literature.994
10
8
x
0
6
4
2
8
.54644~ R"2
= :
0.09 0.0153
2.5381 10 7.0467 + 0. (1999) have developed methods for making casein microparticles.344
0. 29. For example.2 plotted as slope versus the inverse of coating amount. * (
10
1s
1 00/(9 Coat) 6
20
25
Data from Table 29.23
Slope x io3
R2
12.3.001 0. which have promise for targeting of drugs.LeastSquares Fit Rata from Fig.
O m a n M. patent 4. In: Florey KI. Guyot M. Belleville M. Drug Dev Ind P h a m 19:2021. p 71. S . Gallo JM (1992).399.115. Post A. J Pharm Biopharm 40:203. Kas HS. Bruni G.N = number of coated beads IZ = number of pores (holes) yli = the number of particles in the ith fraction P = osmotic pressure = (a) 2kS/r in HixsonCrowell equation. Pharm Res 9:190. Lechevin JC (1979). Orlando. Vecchio C (1993). patent 3. Int J Pharm 18753. He P. John Wiley.634. Modeling andData Treatment in the Pharmaceutical Sciences. Biancini R. Illum L (1998). p 32. Hassan EE.555. Hermelin V (1963). Franklin NJ (1961). U. Davis SS. Illum L (1999). New York.g. S . U. (b) total dose of drug dosage form (=JN). Maggi L. I1 Farm 44:645. Bennett CA. (b)factor in nonsink condition S = diameter of holes q = viscosity of saturated solution within the coated bead p = (a) weighted density average of the dry solids in the coated beads. U. ed. nonpareils) per coated bead q* = constant in nonsink conditions qi = the amount released from the ith fraction r = radius of a spherical particle S = solubility. In: Statistical Analysis in Chemistry and the Chemical Industry. Academic Press.993.. FL. Merie F. Johnson MCR (1994). U. Int J Pharm 108:233f. Int J Pharm 175:6l. Davis S S . S . g/cm3 t = time t* = critical time at which the last particle of drug will have dissolved Y = volume of the interior of all the dry. Chwala A. Newton JM. Carstensen JT (1996).155. NY.441. Hincla AA (1996). coated beads v = total volume of liquid released through pores vl = volume of liquid released per pore w = weight of fill (= q +f) = encapsulated dry weight [= N(4 +. Torre ML (1994).f)] a = fraction of film that is poreformer @ = (a) proportionality constant (granulation film thickness). Hsiao (1985). vol 13. Technomic Publishing. Int J Pharm 166:75. Giunchedi P. Taylor KMG. PA. He P. J Microencapsu 13: 141. S. amount released (e) 4 = grams of filler (e. Parish RC. Fawaz F (1998). Conte U. patent 4. Gazzamoga A. Analytical Profiles of Drug Substances. Zarembo JE (1984).
. Lancaster. Biancini R. patent 4. (b) general density term p’ = density of pore former 4 = a constant in nonsink condition = [ D A / (Y/3*)]
Acikgoz M. Hsiao (1987).587. pp 1281 80. Bavin PMG. Vecchio C (1989).
Int J P h a m 86:247.997.Collett JH (1999). patent 3. Pereira NL.623. Drug Dev Ind Pharm 15: 1495. GhebreSelassie J (1992). Int J P h a m 29:37. Wehrle . Palmeri BF. Int J P h a m 186: 191.
. Int J Pharm 54: 15. The Theory and Practice of Industrial Pharmacy. Lea & Febiger. Rork GS. Drug Dev Ind P h a m 20:2859. tr. Desmukh AA. Palmeri BF. Drug Dev Ind P h a m 19:855. Yang T. Kanig JL. Porter C (1990). de Freitas 0.
Bakan JA (1986). Santinho AJP. Powel TC (1971). Yuen KH.pp 412429. Rowe RC (1986). Drug Dev Ind Pharm 20:2859.Lippold H. International Patent Application WO 9410 1093.Weiss J. In: Lachman L. S. Sutter IS. Stamm A (1994). eds. 3rd ed. Haslarn JL (1994). Wehrle (1997). Lippold BC (1989). Philadelph~. Newton JM (1997). Liebeman HA. Van Savage 6.
74 Adsorption isotherms. applied to mixtures. 441442 Chitosan. 449 Amorphates. as solution systems. 109 108.67 hysteresis. 496498 Coating of tablets. 455467 enteric coating. 2 Amoxicillin amorphous anhydrate. Amorphous cakes. 302303 flow rates correlated to. 457 film coating. 302303 role in mixing. 299307 measurement of. 240245 BET isotherm. 176 hydrous. 456459 Cogrinds. 343 for two particles. 225227 Amorphous solid. 300
. 498499 Coating for sustained release. 75 La~gmuir. 225 decomposition of. 41 1413 thy~eckel equation. 449 Capping. 247 Ball mills. 282 of powder beds. 1121 18 determination of amorphate content. 324 in tableting. 414 CMC (carboxymethy1cellulose). 301303 and repose angles. 4 19420 CarmanKozeny equation. 230234. 281296 A~parent volumes of powders. 324327 Bawn model. 383 Buffers. 66 Adsorption models. 110 kinetics of decomposition of. 122123. 3 15 Apparent density of binary mixtures. 112 hydrous. 449 Cellulose derivatives. 354355 Binomial distributions. 303306 in powder beds. 2 02 17
AvramiErofeyev Equations. assumptions for. 277278 Buoyancy tablets. 283285 definitions. 108 trihydrate. 108 Andreasen apparatus. 7879 Cellulose. 58 Angle of repose. 75 Alginic acid. 1071 16 crystallization rates of. 336338 Blendi~g. 176 Amorphous indomethacin. 44 1442 Cellulose acetate derivatives. isosteric heat of. 421 Average particle size from dissolution I profiles. 449 Coated nonpareils. 224228. effect on solid state stability.6769 types IIT. 281296 Asperite melting. 6974 ~reundlich. 6674 BET. 6974 Binder. 393394. 268 methods of p r e ~ a r a t i o ~ .Adsorption. 486 CAP (cellulose acetate phthalate). 335352 Brittle fracture in milling. 186 Cohesion.
94 and equilibrium crystal size. 323334 (see also Milling) compaction of powder beds. 427429 rate of water uptake of. 126 Crystallization rates of amorphates. 94 reaction rate. 142144 Critical nucleus size. 427437 effect of tableting pressure on. determination of. 418419 Crystal growth. 282 determination of. 108
Dehydration kinetics. 197 nonsink. 104105 of polymorphs. 42 Defects in direct compression tablets. 6365
. 477 effect on dissolution. 234 Density. 409410 Disintegrants. 92 product yield. 245249 Diffusion through films. 416 Dehydration of amoxicillin trihydrate. 16 Diameter definitions. 58 Diffusion controlled solid reactions. 368370 Dissolution medium. 167 as a function of water vapor pressure. 415 particle size considerations. 102104 Crystal Systems. 148150 Critical time in Bawn model. 194 from particles and surfaces. 219 iss solution rates of polymorphs. 195 from hardshell capsules. 18 use in defect deter~ination. 416 dry binders. 415416 excipients for. 38 Diffuse reflectance IR. 96 resulting particle size distributions from. 25726 1 CooperEaton equation. 95 Crystal habit. 396398 Condensation kinetics. 62.5 Cyclodextrins. for producing continuous films. 420 Critical dissolution time from dissolution data. 203206 from tablets. 357 Contact points in solid state kinetics. Complex formation between drug and polymer. 428 Disintegration. 194 effect of temperature. effect on solubility. 408 defects in. 63 Dielectric constant. 96. 1121 14 Crystalline solid.100 and supersaturation. 90 thermal. 413 loading capacity. 17 of powder beds. importance in granulation. 503 Coulter counter. 196 Fick’s law in. 196 effect of viscosity. particle size distributions from. 4 42 1. 295296 Complexation. 191206 of polydisperse powders. 435 Compression cycles. 2 19220 Compressioncoated tablets. 430 models for. 186 Distributions. 140142 Contact angle. 432435 from wetprocessed granules. 4. 414 effect of moisture on. 185 Coprecipitation. 409 mechanisms. 56 Critical compression force. 200202 shape factors and. 2 determination of percent amorphate in. 167 Dehydration of theophylline. 62 lognormal. 196 film theory of. effect on dissolution rate. 429431 Dissolution.16 definitions. 446448 Direct compression tablets. 281296 substituent effect. 4 11 mixed excipients for. 242 Crushing strength of tablets.Comminution. 91 Critical temperature for hydrates. 427437 by calorimetry. 89108 cooling curves In. 217219 Critical moisture content. 101 effect of impurities. 382 HixsonCrowell equation. 128 Dissolution of solid dispersions. 394396 Coprecipitates. 95 Crystal growth rate. 112 Crystallization. 198 NelsonShah equation.
363 flurd bed granulation. 3 11 wall effects on. 1 modulated. 318 effect of moisture on. 164 of salt hydrates. 62 Film coating. 460 CordonTaylor equation. 319 measurement of 31 1 of powders. 359361 dissolution from. 1191 20 Energy cons~derations tableting. 449 Gibbs energy. 330 Fouriertransform diffuse reflectance IR. 109 GordonTaylor equation. 67 Friction.156166 Drug product. 471 Enthalpy. 249254 Einstein equation. 379 Drug. 192 1 Elastic limit.Dosage form. 54 Gas phase interactions in solid state kinetics. 170 Frenkel defect. 2 Dry binders. 471473 Eutectic diagrams. 1721 74 use of DSC in. 303306 Frictional coefficient. 3 Glass transition for plasticizers. 3 153 16 dynamic flow rates. 353370 effect of water addition rate. 457 Enteric coating sustained release. 3738 Enantiotropes. 362363 effect of moisture content. 358 Granule density. 460462 sustained release. 2 Gibbs’ phase rule. 1761 78 Freezing curves of ideal solutions. 299 measurement of. during hardshell filling. 362 endpoints. 457 of particulates for sustained release. 356357 porosity of. 367368
. 58 Fractal dimensions. 119 melting points and vapor pressure curves of. 401404 in Enteric coating. 358359 fluid bed. 459 Film coats aqueous. 109 Granulation. 445 Effervescent systems. 422423 Excipients for direct compression. 370 formation. 111 Flow rates correlation with repose angle. 180 Eutectics. 309321 regularity of. 381 Electrolytes effect on solubility. 364 effect of temperature on. 364 pelletizing. 462463 strength of. 256 Erosion tablets. 2 Freeze drying. 460 solvent systems for. 364366 Ferret’s diameter. 456459 defects. stability of. 2 Drug substance. 2 Dosators. 299300 in tableting. 356357 size determination. 320 effect efflux tube diameter on. 14 Freundlich Isotherm. 359 properties of. 165166 DSC. 2 Equilibrium. definition and measurement. 354 measurements of. wet. 1691 89 EVO~methodof optimization. 2. 413414 Extraganular porosity. 8 185 Free energy. 181 Eutectlc mixture. 118 Gas adsorption. 261 Gelatin. 486 FloryHiggins model. 316318 Fluid energy mills. 400 FTIR. use in polymorphic identification. 195
Floatable tablets. effect in solid state reactions. 366367 physics of. 110 EC (Ethyl cellulose). 495496 plasticizers for. 465467 Film theory of dissolution. 414 Drying. 463465 effect of storage on.
478482 effect of amount of polymer. 6 Lattice energy. 11 1 as solution systems. 210 Lubrication. 74 Isoviscosity curves. 375385 arriving at fill weights for. 357 types of. stability profiles. 276277 Hydrodynamic diameter. 379 effect of speed on fill weight. 391393 Horsfield packing. 359
. 445446 HPMC (hydroxypropyl methylcellulose). 472 HildebrandScott equation. 93 lag times in. 347 Langmuir isotherm. 421 Helmho~z energy. 327330 Hardness of tablets. 393394. 2832 Heckel equation. 482 effect of molecular weight of polymer. 198199 Ho~ogeneous nucleation. in tablet formation. 118 Isomers. 75 in moisture isotherms. dissolut~o~ from. 176178 Macropores. 384 tworing machine for making. 303 JohnsonMehlAvrami equation. 257261 Liquidus line. optical. 380381 compaction during. 52 Isosteric heat of adsorption. 323 applied to ~ i l l i 323324 ~ ~ . 810 Levich equation. 483 Hydrates. 343346 effect of particle size. 197 Liquid. 482 effect of drug loading. 177 Jander equation. 484
Hydrous amorphates. 6365 Heat capacity. 444445 modified. 134 Hysteresis in adsorption isotherms. 1821 classical. 20921 9 Lattice defects. 181 Lognormal distribution. 1531 56 Immiscible melts. 462 HixsonCrowell equation effect of particle shape. 176 PloryHiggins modeI. 6769 Largest particle size from dissolution profiles. 303 shear cell. 178179 Ink bottle pores. 228 Kinetics of decomposition of solids of benzoic acids. 163164 Isometric particle shape. 1516 Lattices. 111 Hygroscopicity. 482 erosion of. 228 crystallization of amorphous lactose and. 378379 pelliculation of. 356 Guar gum. 6365 Lognormal distributed powders. 2 Higuchi square root law. 382 dosator principle. in tablets. 1 use in polymorphic identification. 484485 use of mixtures of polymers. 18. 225 Kinetics of noncohesive mixing. 400401 Lyophilization. 381 disintegration of. 359 IR. 294 HPC (hydroxypro~yl cellulose. 209. 479482 release and equations for. 200 for monodisperse powder.I
[Granule] tensile strength. 13 energy of. 376379 HatchChoate relations.19 Einstein model. 57 Hydrogels. 224 nearest neighbor effect. 482 effect of diluents. 418419 Hardshell capsules. 192 1 eat of solution. 376 as sustained release dosage forms. 383 sizes. 175 using DSC to establish. 449 Hammer mills. 247 Jenike locus. 94 ooke's law. 382383 disso1ution from. 1 10112 Vrentas model. 2 Liquid interaction phases in solid state kinetics.
134 Molecular compounds. 65 NoyesWhitney equation. 1781 79 miscible. 125. 330332 opti~um feed rate. 52 Ng equation. 138 for multiple hydrates. 357 stability at the critical moisture content. I. 273274 effect on extragranular porosity. 174 Molecular weights and intrinsic viscosity. 343346 Nonsegregatmg mixtures. effect on solid state stability 254255. 356 Metastable zone. 62 Matrix tablets. 390. 100 homogeneous. 53.Maltodextrins.65 Milling. for large crystalline molecules. 323334 effect on particle size distribution. 93 Nucleation rates in Arnorphates 107108
. 161 Mixing. 269 stability effect of very low moisture levels. 267279 effect on tensile strength of granules. 136138 for crystalline solids. 347 efficiency. 171 Melting point diagrams. 328 Minimum particle size from dissolution profiles. 183 solid dispersions. 341 Modi~ed P M483 . 52 Monotropes. 1531 56 Moisture isotherms as BET isotherms. 129 effect on solidstate stability of. 361 for wet granules. 440441 Monodisperse powders. 2681272 Moisture exchange between excipients. 170 Melts immiscible. 493500 tableted.328 MC (methylcellulose). 359 Microscopy. 24. 414 MCC microcrystalline cellulose). 360 surface area from. 27 1273 stability effect of excess water. 191 Nucleation. ~ ~
Mohr bodies. 91. 330 Micropores. 414 Martin’s diameter. 145148. 127 Mixed solvents in purification. 325 Multilayer tablet machines. 398400 Moisture bound moisture. 1231 24. 505507 Normal distribution. 133158 for hydrates. 4445 Microcapsules. 274275 stability effect of intermediate moisture levels. 1691 89 of ideal solutions. 1191 20 Morphology Mortar and pestle. 347 modes of sampling m. 184 Mercury intrusion porosimetry. 338340 of noncohesive powders. 90 Micellar systems effect on solubility. 288290 Nonsink dissolution. 65 Zvalue. 275276 Micro~eritics. 1501 53 for nonhydrates. 475477 Mean particle size from dissolution profiles. 44243 of polymers. 1531 56 Moisture uptake rate. 7679 pore size distribution from. 504 Microenvironrnental pH. 2092 I9 Mixed polymorphs. 194 in microcapsules. 414. 438 Multiparticulates. 79. 138140 smooth. 119 melting points and vapor pressure curves of. kinetics of. 359361 Mesopores. Melting points. 473475 percolation theory pertaining to. 2324 of polymorphs. 335352 effect of partlcle size. 364 effect on stability of metastable polymorphs. 210217 as a function of screen aperture in rnillin~. 239240 Noncohesive mixing. 6188 Micronizers. 234. 1791 83 partially miscible. 25 eking point depressions.
92 Packing closest. effect on flow rate. 330332 surface area determination from. 54. 79 P effect in solid state kinetics. 58 effect on machinability.2425. 3 19 regularity of. 4244 Plasticizers. 58 surface mean. 2 102 17 after homogeneous crystallization. 319 Polymers. 118 mixed. 57 surface volume mean. 459460 Plastic deformation in milling. 393394
Polydisperse powders. 3 103 11 of polydisperse powders. 56 hydrodynamic. 7576 extragranular. 256 Prout~ompkins equation. 324 Poisson ratio. 3537 solubility and thermodynamic functions. 162 Purity assessment from melting point depression. 439450 molecular weight determinationy 440441 pH and temperature sensitive. dissolution patterns from. 118 pharmaceutical significance. 3 133 14 Precipitation by pHChange. 129 solubility of. 282 of multiparticulatey multidisperse mixtures 29 1294 Powder flow. 362 measurement. 319 Partly miscible melt PEGS (Polyethylene glycols). in tablet formation. 367 Pell~culat~on. 366367 Pellets. 55 volume mean. 449 Porosity from adsorption isotherms. 161 by pHchange. 444 Polymorphismy 7. 383 Percolation theory. 53 and solubility. 56 electronic counters for.117130 methods of detection. 1631 64 Ordered mixing. 311 static. 359361 Porosity of powder beds correlation with bed density. 160 by thermal recrystallization. 475477 Permeametry. 89 by mixed solvent technique. 441 Pelletizing. 58 arithmetic mean. 247. 234238. 255 Pseudopolymorp~ism.Onecomponent systems. 2092 19 flow rate of. 5161 by Andreasen apparatus. 448449 in film coating. 348350 Osmotic pumps. 117 Polymorphs dissolution rates of. 5 1. 3 10 in tableting. 52 Particle shape. 282 definltion. 346 Pressure effect in solid state reactions. 254255 effect on solubility. 57 related to shapes. 285286 Horsfield. 52 dissolution from. 89 Particle size. 309321 definitions. 485486 OstwaldFreundlich equation. 96100 log normal after milling. 160 Premixing. 46 Particle size distributions from dissolution. 294 Particle diameters. 121123 Polysaccharides. 114. 314 effect on mixing. 3 123 13 types of. 57 Particle dimensions. 126129 effect of moisture on transformat~on of.127 moist storage. 1 Optical isomers.125. 123125 methods of preparation. 5 16 1 effect in direct compression tablets. 125 Purification. 171
. 269 ~seudopolymorphic transformations. 4094 10 effect on flow rate. effect on. 79 Particle size enlargement. 450451 Polymethacrylates. 347 measurement. 1261 27 stability of.
234 by surface nucleation. drying of. 2749 determination of. 340 SEM. 144145 Salt selection. 3738 effect of particle size. 456 Surface areas from mercury porosimetry. 165166 equilibrium moisture content. 384 by hydrogels. 102104. 41. 44 temperature. 42 effect of dielectric constant. 32 effect of complexation. 57 Stress and strain in tablet formation. 57 Surface volume mean diameter. 391 Subsieve sizer. 46 Sustained release by chemical modification. 21 3265 diffusion controlled. 55 Sieve test. 798 1 during dissolution. effect on. 449 Solid. 1501 53 for hydrates. 13 Solid dispersions. 357 States of matter. 193 Slugging. 2 Statistics of ideal mixtures. 37 Spheronization. 370 Spreading coefficient. 183 of the first kind. 3 15 Reprecipitation. 167168 Solvents. 14 Segregation of noncohesive powders. 496498 EC films in. 145148
Radius ratio rule. 2 102 17 from fractal dimensions. 37 multiple peaks. 470 by coated particles. 471 equations for. 486487 by erosion. 184 dissolution of. 184 Solid state stability. 38 Solvates. 3 11 flow rates as a function of. 367368 Silica gel. 224228 Recrystallization. 339 validation of. 55 Surfactants. 6 Random decomposition in solids. 175176. 4546 prediction of. 389390 Salt hydrates drying of. 235238 Solidus line. 302303.162 Repose angle.7
PVP (polyvinyl pyrollidone). 134 Single punch tablet machines. 501502 by enteric coating. 4764477
. 79 Sieve analysis. 90 Residual standard deviation of powder mixes. 14 Screw defect.124. 160 Sampling thief. 3537. thermal. 388380 Sink conditions. 38 effect of electrolytes. effect on solubility. 471 film coats. 444 PX diagrams for compounds forming more than one hydrate. 504505 effect of filmthickness on. 79 volumetric mean shape factor. 245249
[Solid state stability] by nucleation followed by fast reaction. 234238 Solid to liquidplus gas reaction. 196 Stokes law. 238239 Solid to solidplusgas reaction. 54. effect on solubility. 202206 from dissolution profiles. 3237 Solubility parameters. 182 of the second kind. 465467 films used for. 4 17 Rotary tablet machines. 126127 of poorly stable substances. 4 17 Smallest particle size from dissolution profiles. 46 effect of pH. 54 Sugar coating. 3941 of optical isomers. 240245 temperature dependence of. 175 Solubility. 2 properties of. 78 from particle size distributions. 350 Schottky defect. 337 Roller compaction. 8 185 surface mean shape factor. 336338 StokesEinstein equation. 54 Shape factors. 186 Solid solutions. 4244 effect of solvents. 79 Surface mean diameter. 507508 hardshell capsules. 163164 of polymorphs. 2092 19 Sodium starch glycolate.
421422 brittle fracture mechanism. 162 particle size distribution from. 46 Validation of sampling thief. 38 1383 Tablet physics. 501502 by multiple osmotic pumps. 391 tensile strength of. 354 Wood’s Apparatus. 420421 wet g r a ~ ~ l a t e416 d. 354 particle enlargement. 171172 Vapor pressure of hydrates. 387404 Tapped density. 494 use of mixed and multiple films in. 353370. 503 Tablet coating. 5 Xylitol. 416 equipment. effect on granule porosity. 418419 TGA. 41 1413 bonding types in. 290 Water adsorption into amorphates. 350 Van Laar equation. 469483 by sized particles.X
[Sustained release] by microencapsulatin. 324 in tablets. 249254
Temperature of gra~ulation. yield value. of capsule fills. 383 capping of 418419 capping pressure. 121123 ‘Unstable compounds. 420 coating of. 388380 slugging for. 418419 uniaxial expansion of. 381 Xray Xray crystallography. 1 Thermal recrystallization. 40W01 multilayer. 493508 by multiple ~lmthickness. 438 optim~zation 422423 of. 145148 of solids. 420 direct compression. physical principles of. 363 Tensile Strength of Tablets. 417 rotary machines. 396398 critical compression force. 286288 for spheres. 4 Xray diffraction. 418 lubrication. 2 ~~c Topochemical reactions. 192 effect of variables. 1101 12 Wet granulation. 407420 roller compaction for.23 Vrentas model. 390. 414
. 499501 by nonpareils. 1. 455467 compression coated. 194195 Yield value in milling. 498499 percolation theory pertaining to. 316318 in powder beds. 383 Young’s modulus. 455467 Tablets asperite melting as bonding in. 389390 single punch machines. 22. 435 compression cycles. 295 Tartaric acidsodium bicarbonate system. 40841 1 energy consumptjon for. 417 stress and strain in. 102 T h e ~ m o d y n a functions. 354 flow sheet. 111 Wall effects effects on flow rates. 40 1 4 0 4 hardness test for. 475477 single unit dosage forms. determination of stability of. 228230 Transformation of polymorphs rates and models of.