`
`
`
`ENCYCLOPEDIA
`
`OF
`
`SEPARATION SCIENCE
`
`Editor-in—Chief
`
`IAN D. WILSON
`.
`.
`.
`.
`.
`.
`.
`.
`.
`.
`7
`Access for a limited period to an on-line VCI‘SlOn of the Encyclopedia of Separation Sc1ence is
`included in the purchase price of the print edition.
`
`'
`
`Managing Technical Editor
`
`EDWARD R. ADLARD
`
`Editors
`
`MICHAEL COOKE
`
`COLIN F. POOLE
`
`
`
`F'
`
`
`
`This on—line version has been uniquely and persistently identified by the Digital Object Identifier
`(D01)
`
`10.1006/rwss.2000
`By following the link
`
`http://dx.doi.org/IO.1006/rwss.2000
`
`from any Web Browser, buyers of the Encyclopedia of Separation Science
`will find instructions on how to register for access.
`
`If you have any problems with accessing the on—line version, e-mail:
`idealreferenceworks@harcourt.c0m
`
`
`
`.
`
`
`
`ACADEMIC PRESS
`
`
`A Harcourt Science and Technology Company
`
`San Diego San Francisco New York Boston
`London Sydney TOKYO
`
`Argentum EX1029
`Argentum EX1029
`Page 1
`
`Page 1
`
`

`

`This book is printed on acid—free paper.
`
`Copyright
`
`(‘ 2000 by ACADEMIC PRESS
`
`The following articles are US Government works in the
`public domain and not subject to copyright:
`
`III/FOOD TECHNOLOGY/Supercritical Fluid Chromatography
`III/FORENSIC SCIENCES/Liquid Chromatography
`
`III/INSECTICIDES/Gas Chromatography
`Crown Copyright 1999
`
`III/MECHANICAL TECHNIQUES: PARTICLE SIZE SEPARATION
`Copyright
`(‘ 1999 Minister of Natural Resources, Canada
`
`II/CENTRIFUGATION/Large—Scale Centrifugation
`Copyright C 2000 Minister of Public Works and Government Services, Canada
`
`III/AIR LIQUEFACTION: DISTILLATION
`Copyright _(‘ 2000 Air Products and Chemicals, Inc
`
`All Rights Reserved.
`No part of this publication may be reproduced or transmitted in any form or by any
`means, electronic or mechanical, including photocopying, recording, or any
`information storage and retrieval system, without permission in writing from the publisher.
`
`”/0
`
`A 3
`S
`
`
`
`Academic Press
`A Harcourt Science and Technology Company
`Harcourt Place, 32 Jamestown Road, London NW1 7BY, UK
`http://www.academicpress.com
`
`Academic Press
`
`A Harcourt Science and Technology Company
`525 B Street, Suite 1900, San Diego, California 92101-4495, USA
`http://www.academicpress.com
`
`ISBN 0—12—226770—2
`
`Library of Congress Catalog Number: 99-61531
`
`A catalogue record for this book is available from the British Library
`
`Access for a limited period to an on-line version of the Encyclopedia of Separation Science is
`included in the purchase price of the print edition.
`
` EDITORS
`
`
`——————————-———.____—____—___—_
`
`EDITOR-lN-CHIEF
`
`Ian D. Wilson
`AstraZeneca Pharmaceuticals Limited
`
`Mereside, Alderley Park
`Macclesfield
`
`Cheshire SK10 4TG, UK
`
`MANAGING TECHNICAL EDITOR
`
`Edward R. Adlard
`
`Formerly of Shell Research Limited
`Thornton Research Centre
`PO Box 1
`
`Chester CH1 38H, UK
`
`EDITORS
`
`Michael Cooke
`
`Royal Holloway, University of London
`Centre for Chemical Sciences
`
`Egham Hill, Egham
`Surrey TW20 OEX, UK
`
`Colin F. Poole
`
`Wayne State University
`Department of Chemistry
`Detroit
`
`MI 48202, USA
`
`
`
`
`This on-line version has been uniquely and persistently identified by the Digital Object Identifier
`(DOI)
`
`
`
`http://dx.doi.org/l 0.1 006/rwss.2000
`from any Web Browser, buyers of the Encyclopedia of Separation Science
`will find instructions on how to register for access.
`
`10.1006/rwss.2000
`
`By following the [in/e
`
`
`
`
`idealreferenceworks@harcourt.com
`
`If you have any problems with accessing the on—line version, e—mail:
`
`Typeset by Macmillan India Limited, Bangalore, India
`Printed and bound in Great Britain by The Bath Press, Bath, Somerset, UK
`000102030405BP987654321
`
`Page 2
`
`Page 2
`
`

`

`
`
`FOREWORD Vii
`
`M F
`
`oreword
`
`Separation science was first recognized as a distinct area of physical and analytical chemistry in the 19605. The
`term was first coined, I believe, by the late j. Calvin Giddings, Research Professor at the University of Utah.
`Calvin Giddings recognized that the same basic physical principles governed a wide range of separation
`techniques, and that much could be learnt by applying our understanding of one such technique to others. This
`was especially true for his first loves, chromatography and electrophoresis and latterly field flow fractionation.
`Of course there are many separation techniques other than chromatography, many with a history at least as
`long, or indeed longer, than that of chromatography: distillation, crystallization, centrifugation, extraction,
`flotation and particle separation, spring to mind. Other separation techniques have emerged more recently:
`affinity separations, membrane separations and mass spectrometry. Most people, a few years ago, would
`not have classed mass spectrometry as a separation technique at all. However, with modern ionization
`methods, which minimize fragmentation, mixtures of compounds can first of all be separated and then each
`component identified through fragmentation by secondary ion—molecule collisions and further mass spectro-
`metry. With the scale of mass spectrometry now matching that of microseparation methods such as capillary
`electrophoresis and capillary electrochromatography, combinations of orthogonal methods can now provide
`extremely powerful separation and identification platforms for characterizing complex mixtures.
`Basically, all separation techniques rely on thermodynamic differences between components to dis—
`criminate one component from another, while kinetic factors determine the speed at which separation can be
`achieved. This applies most obviously to distillation, chromatography and electrophoresis, but is also obvious
`in most of the other techniques; even particle size separation by sieving can be classified in this way. The
`thermodynamic aspect is, of course,
`trivial being represented by the different sizes of the particles, as
`indeed it is for the size exclusion chromatography of polymers. However, the kinetic aspects are far from
`trivial. Anyone who has tried to sieve particles will have asked the question: is it better to fill the sieve nearly to
`the top and sieve for a long time, or is it better to dribble the material slowly into the sieve and just remove the
`heavies from time to time? One might further ask: how does one devise a continuous sieving process where
`large particles emerge from one port of the equipment, and small ones emerge from the other port? And how
`does one optimize throughput and minimize unit cost?
`The publication of this Encyclopedia ofSeparation Science is a landmark for this area of science at the start
`of the third millennium. It will undoubtedly be of enormous value to practitioners of separation science
`looking for an overview and for guidance as to which method to select for a new problem, as well as to those
`who are at an early stage, simply dipping their toes into the waters, and trying to find out just what it is all
`about. Most impcrtant of all, by providing a comprehensive picture, it advances the whole field of separation
`science and stimul itCS further work on its development and application. The publishers, their editors and their
`authors are to be congratulated on a splendid effort.
`
`John H. Knox
`
`Edinburgh
`8 March 2000
`
`Page 3
`
`Vi
`
`EDITORIAL ADVISORY BOARD
`
`Editorial Advisory Board
`
`Richard W. Baker
`
`Membrane Technology & Research Inc (MTR)
`1360 Willow Road, Suite 103
`Menlo Park
`CA 94025, USA
`
`Kenneth L. Busch
`
`Kennesaw State University
`Office of Sponsored Programs
`1000 Chastain Road
`Kennesaw
`
`GA 30144, USA
`
`Howard A. Chase
`
`University of Cambridge
`Department of Chemical Engineering
`Pembroke Street
`
`Cambridge CB2 3RA, UK
`
`Jan Cilliers
`
`University of Science and Technology in Manchester
`Department of Chemical Engineering
`PO Box 880
`Manchester M60 10D, UK
`
`Alan Dyer
`University of Salford
`School of Sciences, Cockroft Building
`Salford M5 4WT, UK
`
`Heinz Engelhardt
`Universitat des Saarlandes
`
`Instrumentelle Analytik/Umweltanalytik
`Postfach 15 11 50
`
`66041 Saarbruecken, Germany
`
`William F. Furter
`
`Royal Military College of Canada
`Department of Chemical Engineering
`Kingston
`Ontario K7K 5L0, Canada
`
`Josef Janca
`Universite de La Rochelle
`
`Pole Sciences et Technologie
`Equipe de Physico—Chimie Macromoleculaire
`Avenue Michel Crepeau — 17042 La Rochelle, France
`
`Walt Jennings
`J&W Scientific incorporated
`91 Blue Ravine Road
`
`Folsom CA 95630, USA
`
`Kenneth Jones
`
`Affinity Chromatography Limited
`Freeport
`Ballsalla
`
`Isle of Man, UK
`
`Chris Lowe
`
`University of Cambridge
`Institute of Biotechnology
`Tennis Court Road
`
`Cambridge, UK
`
`David Perrett
`
`St Bartholomew's and the Royal London School
`of Medicine and Dentistry
`St Bartholomew’s Hospital, Department of Medicine
`West Smithfield
`London EC1A 7BE, UK
`
`Douglas E. Raynie
`The Procter and Gamble Company
`Miami Valley Laboratories
`PO Box 538707
`Cincinnati
`OH 45253-8707, USA
`
`Peter Schoenmakers
`
`Shell Research and Technology Centre Amsterdam
`(SRTCA) and University of Amsterdam
`Postbus 38000
`1030 BN Amsterdam, The Netherlands
`
`Darrell N. Taulbee
`
`University of Kentucky—Center for Applied Energy
`Research (UK—CAER)
`2540 Research Park Drive
`
`Lexington
`KY 40511, USA
`
`Gerda M. van Rosmalen
`
`Delft University of Technology
`Laboratory for Process Equipment
`Leeghwaterstraat 44
`2628 CA Delft, The Netherlands
`
`Edward Woodburn*
`
`University of Science and Technology in Manchester
`Department of Chemical Engineering
`PO Box 880
`Manchester M60 1QD, UK
`
`+deceased
`
`Page 3
`
`

`

`Crystallization
`
`W. Beckmann and U. Budde, Schering AG,
`Berlin, Germany
`Copyright ^ 2000 Academic Press
`
`Two decades ago, crystallization was called both an
`art and a science. However, the Reld is improving
`quickly. The crystallization of pharmaceuticals is still
`sometimes regarded an art and rather a mystery.
`However, crystallization processes are widely used
`throughout the production processes of the active
`ingredient of a drug product, and a lot of knowledge
`is nowadays available.
`For the crystallization of drug substances several
`aspects have to be considered, as the crystallization
`process is the last step in the chemical manufacture
`of pharmaceuticals. The crystallization determines
`a number of important properties of the drug sub-
`stance, namely the purity and residual solvent con-
`tent, the polymorphic form, crystal size and size
`distribution, and it affects downstream processes
`such as drying, ease of comminution and formulation
`of the Rnal drug product.
`The crystallization of all drug intermediates have
`the same goals and follows the same procedures as
`for other organic substances and thus will not be
`discussed here separately.
`
`General Considerations for the
`Development of the Crystallization
`Process
`In general, the demands on the crystallization of a
`drug substance differ according to the Rnal use,
`e.g. if the product is used in oral dosage forms, in
`ointments or in liquid formulations. However, for the
`sake of simplicity, it is assumed here that the crystal-
`lized drug substance is to be used in an oral dosage
`form.
`Figure 1 shows a typical crystallization process of
`a drug substance and the downstream processes up to
`the formulation of the drug product. The crystalliza-
`tion and the properties of the product have a great
`inSuence on all the following steps.
`
`Impurities
`
`Foreign and related compounds The requirements
`on the purity of a drug substance are strict; guidelines
`require a purity of typically '98%. Individual impu-
`rities with a known structure have to be below 0.5%
`and unidentiRed impurities have to be below 0.1 or
`
`III / PHARMACEUTICALS / Crystallization
`
`3729
`
`0.05%. In addition, the toxicological effect of all
`impurities must have been assessed in the Rrst toxico-
`logical tests, i.e. no new impurity is allowed that has
`not been present in the batch used for toxicological
`experiments.
`The puriRcation of a drug substance via crystalliza-
`tion cannot be predicted easily. While foreign impu-
`rities can mostly be easily reduced, related substances
`like impurities stemming from side reactions in the
`synthesis behave in an unpredictable way. In general,
`the puriRcation via crystallization will decrease with
`an increase in the yield, especially if the yields are
`'90}95%.
`
`Residual solvent content Beside obvious solvent
`properties such as a certain solubility for the drug
`substance and an appropriate puriRcation to yield
`ratio, the choice of the solvent for the crystallization
`of a drug substance is governed by the permissible
`limitations placed on the residual solvent content of
`the drug substance. All typical solvents have been
`classiRed according to their toxicity and tolerated
`daily uptakes of a solvent have been established, that
`are not to be exceeded by the drug product.
`Three classes of solvents are distinguished: (i) those
`that should be avoided; (ii) those that have a limit to
`their daily uptake; and (iii) those for which no limits
`have been set up so far. Examples are benzene and
`dichloroethane for class 1, methanol and dich-
`loromethane for class 2 and ethanol, ethyl acetate and
`acetone for class 3. In addition, good manufacturing
`practice (GMP) requires the manufacturer to limit the
`residual solvent to the lowest content possible.
`
`Figure 1 Typical crystallization and downstream processes up
`to formulation.
`
`Page 4
`
`

`

`3730
`
`III / PHARMACEUTICALS / Crystallization
`
`Table 1 Productivity, yield and development prerequisites for the separation of isomers via crystallization, enzymatic resolution and
`chromatography
`
`Parameter
`
`Crystallization
`
`Enzymaticresolution
`
`Chromatography
`
`Productivity
`Selectivity
`Prerequisite
`Development time
`
`High
`Varies
`High
`High
`
`Low
`High
`High
`High
`
`Medium
`Very high
`Low
`Low to medium
`
`Two types of limits for the residual solvent content
`of a drug are distinguished. Case 1 is a dosage-inde-
`pendent concentration limit and Case 2 is a limit for
`the total uptake through the drug product, that must
`not be exceeded by the solvent content of the drug
`substance and the excipients.
`The mechanisms of incorporation of solvent into
`the crystals can be described as follows:
`
`E The solvent is incorporated into the lattice at
`Rxed positions during the crystallization (solvate
`formation). In this case, the incorporation cannot
`be avoided directly. In some cases, a solvent of
`crystallization is removed or replaced by water.
`E The solvent is incorporated into the lattice as three-
`dimensional inclusions. The formation of inclu-
`sions is facilitated by the speed of crystallization,
`thus, the amount of residual solvent can be de-
`creased by lowering the rate of crystallization. For
`some systems, the tendency to form three-dimen-
`sional inclusions of solvent increases with the crys-
`tal size.
`
`If a problem with the residual solvent content of a
`drug arises, the clear remedy is a change of solvent.
`
`Separation of isomers An increasing number of
`pharmaceutical active ingredients are either isomers
`or enantiomers. Typically, different isomers of a
`chemical compound exhibit different biological or
`
`therapeutic activities, with one of the isomers being the
`carrier of the activity. In some cases, the second isomer
`can even have an adverse biological activity. In any
`case, the inactive isomer constitutes an unnecessary
`load to the body. Thus, a separation of isomers is
`almost a prerequisite for the production of a drug.
`Isomers can be separated by enzymatic resolution,
`chromatography or crystallization. Table 1 summar-
`izes and compares productivity, yield and develop-
`ment prerequisites of the three separation techniques.
`The success (or possibility) of the separation of
`isomers via crystallization depends on the phase dia-
`gram of the two compounds.
`Figure 2 shows typical phase diagrams of isomers,
`i.e. eutectics, solid solutions and partial solubility in
`the solid state. A separation of isomers in a single step
`is only feasible for eutectic systems. Systems forming
`solid solutions have to be puriRed in multiple steps, as
`for example in zone reRning which is only feasible
`if the substance is stable in the molten state. For
`systems exhibiting partial miscibility in the solid
`state, the separation cannot be better than the partial
`miscibility concentrations.
`In principle, isomers forming eutectics can be sep-
`arated directly via crystallization. However, without
`using special techniques, the crystallization can only
`be carried out until the concentration of the mother
`liquor has reached the composition of the eutectic
`mixture. To improve the yield two ways are often
`pursued:
`
`Figure 2 Typical phase diagrams for isomers: eutectics, complete miscibility in the solid state and partial miscibility in the solid.
`
`Page 5
`
`

`

`E Forming diastereomeric salt derivatives of
`the
`isomer will often direct the eutectic composition
`towards one of the isomers. This will increase the
`yield and productivity of the crystallization process
`considerably.
`E The desired isomer is enriched via preparative
`HPLC followed by crystallization. The chromato-
`graphic technique achieves a high degree of enrich-
`ment but the amount of solvent to be handled is
`considerable. Crystallization is carried out at high
`concentrations and thus more effectively. Attempts
`have been made to Rnd the optimal cost effective
`division between the separation via preparative
`HPLC and crystallization.
`
`Solid-State Forms
`
`Polymorphism Before a crystallization process is
`developed, the solid-state polymorphism of the sub-
`stance has to be elucidated. Less than 50% of the
`drug substances described in monographs crystallize
`in a single polymorphic form; the majority form
`polymorphs, pseudomorphs, or both. The polymor-
`phic form of a drug can inSuence a number of its
`properties such as the following:
`
`E The solubility and the dissolution rate and conse-
`quently the bioavailability. Although typical differ-
`ences in solubility between polymorphs are of the
`order of 42, the differences in solubility between
`pseudomorphs are somewhat higher. The largest
`differences exist between amorphous and crystal-
`line material.
`E The habit (the external appearance) of the crystals
`which in turn inSuences the mechanical properties
`of the drug during further processing such as the
`ease of comminution.
`E The chemical stability.
`
`Thus the regulatory agencies ask for the reproducible
`production of the speciRed polymorphic form. In the
`case that a drug can form more than one polymorph,
`a choice of the polymorphic form must be made.
`
`Amorphous compounds Amorphous solids are a
`metastable form of the drug substance that can cry-
`stallize at any instant. In this respect, an amorphous
`form can only be second choice as solid-state form for
`a drug substance.
`The stability of amorphous material can be char-
`acteriszed by its glass-transition temperature Tg,
`which can be determined by differential scanning
`calorimetry (DSC). Below the glass-transition tem-
`perature, the molecules are practically frozen; above
`it they have a Rnite mobility making the conversion
`into a crystalline form possible.
`
`III / PHARMACEUTICALS / Crystallization
`
`3731
`
`The glass-transition temperature and thus the stab-
`ility of amorphous materials can be decreased by
`residual solvent.
`
`Salts A number of properties of the chemical com-
`pound can call for the use of a salt as the drug
`substance. An insufRcient chemical stability of the
`parent compound can be overcome by the formation
`of a salt, e.g. amines sensitive to oxidation can be
`stabilized by forming a hydrochloride. Other proper-
`ties calling for salt formation include low melting
`points or unfavourable solid-state properties such as
`a tendency to form amorphous material or too many
`polymorphs.
`Finally, in case of an insufRcient solubility in water
`or gastro-enteric Suids it is sometimes tried to avoid
`this problem by the formation of salts.
`Salts of sparingly-soluble parent compounds can
`lead to the precipitation of the parent compound
`when the salt is dissolved in water. This poses con-
`siderable problems if it occurs during formulation
`like wet granulation.
`Salts are typically formed by precipitation or reac-
`tion crystallization, i.e. by adding an acid or a base to
`a solution of the drug substance. Of course, each salt
`constitutes a new drug substance, that has to be
`treated accordingly as a new chemical entity.
`
`Clathrates Chemical
`instability of parent com-
`pounds can also be overcome by the formation of
`clathrates typically of ♡-, ♢- or ♤-cyclodextrin. The
`parent molecule partially enters the large voids of the
`cyclodextrin (see Figure 3). It is thus shielded from
`the environment, especially the excipients.
`The clathrate is typically formed by precipitation
`i.e. by adding the parent compound dissolved in
`a water-miscible organic solvent to an aqueous solu-
`tion of the cyclodextrin.
`
`Choice of solid-state form The selection of the opti-
`mal solid-state form is an important step in the
`
`Figure 3 Schematic picture of the clathrate formation of a ster-
`oid by two ♢-cyclodextrin molecules.
`
`Page 6
`
`

`

`3732
`
`III / PHARMACEUTICALS / Crystallization
`
`Figure 4 Properties of the different solid-state forms of a drug
`to be considered when choosing the optimum form to be used in
`the final product.
`
`development of the crystallization process of a drug.
`The form, once decided upon, has to be used in all
`relevant clinical and pharmaceutical tests and it must
`be certain that this form can be crystallized in a repro-
`ducible way, and that it can be formulated into the
`drug product.
`A number of basic physico-chemical and pharma-
`ceutical properties of
`the different
`forms
`that
`are considered for selection are listed in Figure 4, and
`can be tested and used in the process of decision
`making.
`A drug substance that occurs only in a single poly-
`morphic form is preferred; if not available, the stable
`polymorph is preferred. The techniques to infer the
`relative stability of polymorphs include solubility
`measurements, storage in suspensions and DSC experi-
`ments to construct enthalpy}temperature diagrams.
`
`Crystal size and habit
`
`Crystal size and habit of a drug can vary consider-
`ably. Thin needles, platelets and rhombohedral crys-
`tals are found. The crystal habit can inSuence all
`processes after crystallization
`
`E during work-up, the de-watering in the centrifuge,
`the washing of the Rlter cake and the drying are
`affected;
`E in the formulation, behaviour during microniz-
`ation or Sow during direct tableting are affected.
`
`Micrographs of the drug substance and crystal size
`distributions can help in understanding the behaviour
`of the product crystallized under different conditions.
`
`Habit The external appearance of crystals is called
`habit. The crystal habit can be inSuenced by the
`growth conditions. For example, crystals of the A
`modiRcation of Abecarnil, a ♢-carboline derivative,
`grown after spontaneous nucleation at high super-
`saturations exhibit an avicular habit, while those
`grown at moderate supersaturations after seeding are
`still needle like but
`thicker and more rod-like
`(Figure 5).
`Other factors determining the crystal habit are the
`solvent and the impurity proRle of the material to be
`crystallized. The impurity level that inSuences the
`habit } and other properties } can be as low as ppm.
`Figure 6 shows the habits of a steroid crystallized
`from two different solvents, one more protic (solvent
`I) and the other more aprotic (solvent II). Of course,
`the different habits lead to a different behaviour in
`downstream processing.
`
`Crystal size distribution For the formulation of oral
`dosage forms, the desired crystal size distribution is
`
`Figure 5 (See Colour Plate 111) Habit of Abecarnil grown from methanol after spontaneous nucleation at relatively high supersatura-
`tions (left) and grown at moderately low supersaturations after the addition of seeds (right).
`
`Page 7
`
`

`

`III / PHARMACEUTICALS / Crystallization
`
`3733
`
`Figure 6 (See Colour Plate 112) Habit of a steroid crystallized from two different solvents.
`
`mainly imposed by the demands for blend and con-
`tent uniformity. For low-dosage formulations, such
`as
`steroids
`in contraceptives,
`the maximum
`crystal size is 410 ♯m. These crystal sizes cannot
`be attained via classical crystallization techniques.
`Thus, the standard procedure is crystallization and
`drying followed by a comminution, e.g. via jet-mill-
`ing. It has been reported that the milling process is
`also dependent on the crystal size and the homogen-
`eity of the crystal size to ensure a homogeneous mill-
`ing process.
`The size distribution obtained in a jet mill is deci-
`sively determined by the cut size of the cyclone of the
`mill. When the drug substance is to be used without
`comminution, the crystal size distribution will typi-
`cally be broader. Small crystals, 410}100 ♯m, can
`only be achieved via precipitation, larger crystals by
`evaporative or cooling crystallization. If large crystals
`are desired, the best way to control the process is via
`seeding. Here, care must be taken not to destroy the
`crystals during the crystallization process through
`the power input of the stirrer or during work-up,
`especially in agitation dryers.
`Care must be taken to avoid agglomeration during
`crystallization and drying, as this process is erratic
`and in precipitation processes agglomeration is al-
`most unavoidable. For the formulation of liquid dos-
`age forms,
`the dissolution rate can limit
`the
`permissible crystal size, although the requirements
`are not really strict.
`
`Development of the Crystallization
`Process
`In most cases, a drug substance will be crystallized
`from solution in a batch process. The techniques most
`often employed are cooling, evaporative, drowning-
`out and reaction crystallizations. Glass-lined as well
`as stainless-steel vessels are used. These classical crys-
`
`tallization techniques and their implications will be
`discussed in detail. Certain properties of the drug
`substance may call for a crystallization from the melt,
`via spray drying or through the use of novel tech-
`niques for particle formation. These techniques will
`be discussed more brieSy later.
`
`Laboratory Development of Crystallization
`Technique
`
`growth The
`and
`nucleation
`Supersaturation,
`crystallization involves two basic steps, nucleation
`and growth. The driving force for both processes is
`typically deRned as
`the relative supersaturation,
`♶"☭c/csat, where ☭c"c!csat, the difference be-
`tween the actual concentration c and the saturation
`concentration csat. Depending on the magnitude of the
`supersaturation, at which the nucleation occurs, nu-
`cleation and growth will have different importance:
`
`E at high supersaturation, nucleation will be the
`dominant process, the number of nuclei formed is
`large so that the increase in size via growth and
`thus the particles found are small;
`E at low supersaturation, the number of nuclei is
`small so that growth dominates and coarse crystals
`will be obtained.
`
`The Rrst process is usually called precipitation, the
`second one crystallization. In both processes, nuclea-
`tion can either be deliberate by seeding or involuntary
`by primary, spontaneous nucleation. As far as spon-
`taneous nucleation concerned, heterogeneous nuclea-
`tion by foreign particles in the solution will dominate.
`In case of easily or moderately soluble substances
`secondary nucleation caused by addition of the par-
`ent crystals in general dominates the nucleation pro-
`cess, once crystals of a sufRciently large size are
`present.
`
`Page 8
`
`

`

`3734
`
`III / PHARMACEUTICALS / Crystallization
`
`Figure 7 Typical solubility and metastability curves for a drug substance.
`
`Solubility and metastability Two basic pieces of in-
`formation on the system are necessary for a successful
`development of a crystallization process, the solubil-
`ity curve and degree a solution can be supersaturated
`before spontaneous nucleation occurs.
`The solubility deRnes the equilibrium state of the
`substance dissolved in the mother liquor, and the
`metastable zone is a concentration region, where a
`supersaturated mother phase can exist for a certain
`period of time without spontaneous nucleation. This
`latter region has typically a width of the order of
`103C and owing to the relatively large metastable
`zone, the amount of material coming out of solution
`under spontaneous nucleation without further cool-
`ing can be of the order of 10}30% of the entire mass,
`which is a considerable value. Figure 7 shows typical
`solubility curves and metastable lines. The left-hand
`system lends itself to cooling crystallization, the right
`hand one to a evaporative crystallization because of
`the slope of the solubility curve.
`For a drowning-out crystallization, a primary sol-
`vent in which the drug substance has a high solubility
`and a secondary, or anti-solvent, which has a negli-
`gible solubility are needed. Figure 8 shows three typi-
`cal curves for the mixing.
`If the solubility has a concave curvature, i.e. the
`secondary solvent acts as an anti-solvent a drowning-
`out crystallization is feasible as shown in curve B.
`For curves A and C, crystallization by addition of an
`anti-solvent is not possible.
`For precipitation by mixing of reactants or the
`formation of salts, the solubility is given by the solu-
`bility product, i.e. a#bPproduct and K"[a] [b]
`(Figure 9).
`For moderate formation constants, the solubility at
`[a]"[b] is quite low. This implies high to very high
`supersaturations upon mixing, which can lead to
`small or very small crystals, which in turn can cause
`
`problems in downstream processes, ♶"☭c/c. If
`ceq is very low, ♶ will become very high. But this on its
`own has nothing directly to do with small crystals.
`Small crystals result from the lack of growth limits in
`the solution (even at high ♶ values). Apart from that
`abundant nucleation also leads to small crystals, but
`this is also true for easily soluble compounds.
`
`InWuence of impurity and purity of material An of-
`ten underestimated effect is caused by changes that
`may be made during the lifetime of the synthesis and
`its implications for the crystallization process. Vari-
`ations in the impurity proRle can inSuence both the
`width of the metastable zone and thus the degree of
`supersaturation at which spontaneous nucleation oc-
`curs, and the polymorph obtained during via spon-
`taneous nucleation. It can also inSuence the rate of
`polymorphic transformation during work-up and the
`habit and crystal size. Although the effects of a
`changing impurity level on the behaviour of a drug
`substance are difRcult to anticipate during the Rrst
`stages of development, there is a need for a careful
`investigation of these effects.
`
`Figure 8 Typical solubility curves for a drug substance in a mix-
`ture of a primary and a secondary solvent.
`
`Page 9
`
`

`

`III / PHARMACEUTICALS / Crystallization
`
`3735
`
`miscible with water, anti-solvent precipitation with
`water can be used. Since a drowning-out crystalliza-
`tion leads to solvent mixtures that have to be treated
`afterwards, this technique should be avoided when
`possible.
`
`Rate of crystallization The rate of crystallization
`should not exceed values that cause too much uptake
`of solvent via inclusions. For a cooling crystallization,
`natural cooling proRles should be avoided, as most of
`the material crystallizes too fast. Linear cooling rates
`are a Rrst approximation. The rates can be classiRed
`as slow, realistic, fast and crash cooling for rates of
`(5, (10, (15 and '153C h\1, respectively.
`Instead of linear rates, a controlled rate of crystalli-
`zation should be used where possible; this is espe-
`cially important for drowning-out crystallizations,
`for neutralization reaction crystallizations or for the
`formation of salts. In these cases, even a moderate
`dosage of the anti-solvent or the acid or base can lead
`to very high supersaturation which in turn can cause
`the formation of oils or amorphous material. Often
`most of the yield is produced during the addition of
`the Rrst amount of the secondary solvent.
`Thus either a very slow dosing or when exactly
`stepwise dosing with time interval between are appro-
`priate tools to avoid too high supersaturation and
`the adverse effects usually observed for nucleation
`and growth at high supersaturation.
`At the point of addition, high local supersaturation
`is generated so that mixing has to be optimized (see
`Figure 10).
`Programmes for cooling, evaporative and drown-
`ing-out crystallization are derived by requiring the
`crystallization to proceed with a constant rate of
`deposition of mass or with a constant linear growth
`rate. Figure 11 presents curves for a drug substance
`having a temperature dependence of the solubility of
`40 kJ mol\1. The solubility decreases by a factor of
`10 with a change in temperature from 70 to 203C.
`However, the calculations yield only the proRle, the
`absolute times are not given.
`
`Figure 9 Solubility for a ionic reaction crystallization such as
`the formation of the salt. The supersaturation at moderate forma-
`tion constants reaches a high value.
`
`Choice of solvent and of the crystal