Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Analysis of particle deformation mechanisms and compact expansion during compaction on a high speed rotary… Dwivedi, Sarvajna Kumar 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1992_fall_dwivedi_sarvajna_kumar.pdf [ 3.52MB ]
Metadata
JSON: 831-1.0086553.json
JSON-LD: 831-1.0086553-ld.json
RDF/XML (Pretty): 831-1.0086553-rdf.xml
RDF/JSON: 831-1.0086553-rdf.json
Turtle: 831-1.0086553-turtle.txt
N-Triples: 831-1.0086553-rdf-ntriples.txt
Original Record: 831-1.0086553-source.json
Full Text
831-1.0086553-fulltext.txt
Citation
831-1.0086553.ris

Full Text

ANALYSIS OF PARTICLE DEFORMATION MECHANISMS ANDCOMPACT EXPANSION DURING COMPACTION ONA HIGH SPEED ROTARY TABLET PRESSbySARVAJNA KUMAR DWIVEDIB.Pharrn., Banaras Hindu University, 1984M.Pharm., Banaras Hindu University, 1986M.Sc., The University of British Columbia, 1988A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIES(Pharmaceutical Sciences)(We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAOctober, 1992© Sarvajna Kumar Dwivedi, 1992In presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.______________________Department of P1 AMACV]1CAL SCI(’JCCcThe University of British ColumbiaVancouver, CanadaDate o€fq2DE-6 (2/88)iiABSTRACTPharmaceutical tablets are the most widely used dosage form and areprepared by the high speed compaction of powders or granules in a die on arotary tablet press. The tablets produced must be coherent and capable ofwithstanding the stresses of handling and transportation. Successfulcompact formation depends on the ability of the particles to deform andform interparticulate bonds during compression in the die and the abilityof these bonds to withstand elastic expansion during decompression andejection from the die.Compaction on the high speed rotary presses used in thepharmaceutical industry normally occurs in less than 50 ms. The mechanismsof particle deformation during compaction are often analysed using pressesoperating at slow speeds or by using specialised equipment which performtests on either preformed compacts or on single crystals.In the present work a sixteen-station rotary tablet press - a ManestyBetapress- was used to analyse high speed compaction. The analysisinvolved a study of the relationship between punch force and machinedeformation, and its use in understanding the deformation of powderparticles during compression and the elastic expansion of compacts duringdecompression. Rotary presses have been used to analyse compactionpreviously but either the materials were compressed under staticconditions, or the results were obtained by complex viscoelastic modeling.These results were in error because the machine deformation was ignoredwhen the punch displacement was calculated from the machine and punch-headgeometry.In contrast, the relationship between machine deformation and punchforce was used in this work to calculate punch displacements on theiiiBetapress from the measurements of upper punch force. A previouslyreported method of calculating the punch displacement (Oates and Mitchell,1989, 1990) was refined and simplified. This requires only the forceversus time data where the force was measured by a strain-gauged upper rollsupport pin. Since it is relatively easier to measure punch force on arotary press than make direct measurements of punch displacement, thismethod offers accurate punch displacement analysis without using complexinstrumentation and/or geometric calculations.Over forty solids were studied on the Betapress. The solids werecharacterized for various physicochemical properties including true andbulk densities, and melting and/or decomposition temperatures. Powder Xray diffraction, melting points and a two-component melting point-composition phase diagram of R and 5-ibuprofen showed that racemicibuprofen is a one phase ‘racemic compound’ as opposed to a two phase‘racemic mixture’. Thus, the USP description of ibuprofen as a ‘± mixture’is misleading. All solids, including the commercially available Sibuprofen and racemic ibuprofen, were compressed on the Betapress underspeed.The force signals from the upper roll support pin were collected as afunction of time on a desk-top computer via an analog to digital converter.The compaction cycle was divided into compression and decompression phasesby the dead centre position at which the punches are vertically alignedwith the centres of the compression roll support pins. The force-time datawere analysed using specially written software to obtain several parametersfrom the compaction cycle. Parameters obtained from the compression phaseincluded peak offset time, decrease in punch pressure during peak offsettime, porosity changes, work of compression and yield values of the solids.ivThese parameters related particle deformation to pressure. Therelationship between force and machine deformation was used to subtract themachine recovery from the total recovery during decompression. This gave anovel method of estimating tablet expansion during decompression. Thetablet expansion data was used to calculate the work of tabletdecompression and to estimate the Young’s modulus of several pharmaceuticalsolids.Since particle deformation during compression is dependent on strainrate, the strain rate during compaction was approximated by using thedecrease in volume of the powder bed in the die during compaction, andthis, along with the above parameters, was used to ascertain thedeformation mechanism of each solid.The solids were categorised into groups ranging from low yield-strength ductile solids such as acetylsalicylic acid, ibuprofen and theirformulations to high yield-strength brittle solids such as the variouscalcium phosphates. The tableting parameters of formulated drugs andprocessed excipients were different from the parent samples. Racemicibuprofen and S-ibuprofen showed little difference in their tabletingparameters, hence a decision to use S-ibuprofen instead of racemicibuprofen in tableting should be based on differences in other propertiessuch as solubilities and pharmacokinetic differences.A simple and inexpensive method of analysing the behaviour of powderparticles during compaction on a high speed rotary press using onlyforce-time measurements is presented. This method is potentiallyapplicable to any tablet press, and can be used for the in-processvalidation of compaction, for the quality control of raw materials, and forthe development of new tablet formulations.VLIST OF CONTENTSPageABSTRACT iiLIST OF SYMBOLS ixLIST OF TABLES xiLIST OF FIGURES xiiACKNOWLEDGEMENTS xviINTRODUCTION 11.1 THE COMPRESSION PHASE 11.1.1 Elastic, viscous and viscoelastic behaviour of 2solids1.1.2 Creep and stress relaxation in pharmaceutical 4solids under pressure1.2 THE DECOMPRESSION PHASE 81.2.1 Analysis of tablet expansion during 9decompression1.2.2 Use of tablet expansion in material 10characteri sati on1.3 PARTICLE DEFORMATION UNDER PRESSURE 111.3.1 Viscoelastic modeling of the compression data 131.3.2 Modeling of compaction using mechanical prope- 15rties from single crystal microindentationmethods1.3.3 Use of indices of performance in characteris- 16ation of materials.1.4 EQUIPMENT USED FOR ANALYSIS OF POWDER BEHAVIOUR UNDER 17PRESSURE1.5 PRELIMINARY WORK LEADING TO THE PRESENT RESEARCH 181.6 OBJECTIVES OF THE PRESENT RESEARCH 19viPage2 MATERIALS AND METHODS 212.1 MATERIALS 212.2 CHARACTERIZATION OF MATERIALS 242.2.1 Determination of bulk and true density 242.2.2 Differential scanning calorimetry 242.2.3 Determination of binary phase diagram for 25characterization of ibuprofen2.2.4 Powder X-ray diffraction patterns of ibuprofen 26samples2.3 COMPRESSION OF THE MATERIALS ON A ROTARY PRESS 262.3.1 Equipment 262.3.2 Instrumentation and calibration of Betapress 27to obtain force and displacement2.3.3 Data collection and analysis 382.3.4 Preparation of the materials for compression 392.3.5 Compression protocol 412.3.6 Machine speed 412.3.7 Relationship between fraction of turret revol- 42ution and turret time2.4 ANALYSIS OF THE COMPRESSION PHASE ‘ 442.4.1 Calculation of stress, strain and strain rate 442.4.2 Determination of tablet porosity 442.4.3 Calculation of yield stress 452.4.4 Determination of peak offset time 452.4.5 Factors affecting t0ff 472.4.6 Determination of decrease in pressure during 48t0ff2.4.7 Calculation of work of compression 482.6 ANALYSIS OF THE DECOMPRESSION PHASE 492.6.1 Determination of machine deformation during 49decompressi on2.6.2 Selection of the incompressible material 512.6.3 Determination of tablet expansion 522.6.4 Determination of work of decompression 542.6.5 Determination of Young’s modulus 582.6.6 Determination of porosity at the end of 60decompressionviiPage2.7 DETERMINATION OF TABLET STRENGTH 613 RESULTS AND DISCUSSION 623.1 MATERIAL PROPERTIES 623.2 THE SOLID STATE OF IBUPROFEN 643.2.1 DSC and powder X-ray diffraction of ibuprofen 66samples3.2.2 The binary phase diagram of ibuprofen 693.3 ANALYSIS OF POWDER COMPACTION 743.3.1 Machine deformation during powder compaction 763.3.2 Particle deformation during powder compaction 773.3.3 Stress, strain and strain rate during compaction 773.3.4 Influence of strain rate on particle deformation 783.3.5 Approximation of stress, strain and strain rate 79during compaction3.4 PARAMETERS FROM THE COMPRESSION PHASE 803.4.1 A general view of the events during the 80compression phase3.4.2 The rate of compaction profiles 823.4.3 Porosity-stress relationship 933.4.4 Position of the peak punch pressure: Peak offset 94time3.4.4a Effect of max on t0ff 983.4.4b Effect of machine speed on t0ff 1013.4.4c Effect of punch type on t0ff 1053.4.4d Effect of formulation variables on t0ff 1053.4.5 Decrease in punch pressure during the phase 112of constant strain3.4.6 Work of compression 1143.5 PARAMETERS FROM THE DECOMPRESSION PHASE 1143.5.1 Elastic expansion and work of decompression 1163.5.2 Determination of Young’s modulus 1183.5.3 Effect of formulation and processing on 125Young’s modulus3.5.4 The use of decompression analysis and of 129Young’s modulusviii3.9. COMMENTS ON INTERPARTICULATE BOND FORMATION UNDER 155PRESSURE4 SUMMARY 157Page3.6.33.6.43.6 ENERGY CHANGES DURING COMPACTION 1303.6.1 Consumption of energy during the compression 130phase3.6.2 Temperature dependence of the mechanisms of 131deformation and its correlation with the energychanges during compactionRelationship between tablet strength and WC-WD 134Relationship between WC-WD and porosity changes 1363.7 POSSIBLE MECHANISMS OF PARTICLE DEFORMATION3.8 DEFORMATION MECHANISMS OF THE VARIOUS CATEGORIESOF SOLIDS3.8.1 Deformation mechanism of acetylsalicylic acid,ibuprofen and their formulationsDeformation mechanism of calcium phosphatesDeformation mechanism of cellulosesDeformation mechanism of acetaminophen and itsformulationsDeformation mechanism of lactosesDeformation mechanism of sucrosesDeformation mechanism of polyolsDeformation mechanism of miscellaneous othersubstances3.8.23.8.33.8.43.8.53.8.63.8.73.8.81381401401411431441471491501525 REFERENCES 162ixLIST OF SYMBOLSa StressStress at the points of interparticulate contact: true stressUy Yield stress or yield valueE StrainE Strain rateE(fr) Strain as a function of frq ViscosityRelative densitytxDm Change in machine deformation in response to AFADm(fr) ADm as a function of frADmfrO ADm at fr = 0, i.e., the dead centre position of punchesADmFO ADm at F = 0, i.e., at the end of decompressionDm Machine deformationD Distance between the punch faces during compactionD(fr) Distance between the punch faces during compaction as afunction of frD(F) Distance between the punch faces during compaction as afunction of punch forceD(t) Distance between the punch faces during compaction as afunction of time during compressionE Young’s modulusE Modulus of elasticity of a tablet with a certain porosityfr Fraction of a turret revolutionfrFO fr at F = 0, i.e., at the end of decompressionChange in punch force during the compaction cycleF Punch force during the compaction cycleFfrO F at fr = 0, i.e., at the dead centre position of punchesxChange in the height of compacts during decompressionAH(fr) Change in the height of compacts during decompression as afunction of frAHR Heat of fusion of a racemic compoundHmax Value of the Heckel term ln(1/p) at maxHFO Compact height at F = 0, i.e., at the end of decompressionHfrO Compact height at fr = 0, i.e., at the dead centre position ofpunchesK’ Machine deformation constant (2.3 x i06 cm/N)p porosityAP/VT Decrease in pressure during peak offset time normalised fortrue volume of the materialP Pressuremax Maximum pressure or peak pressureR Gas constantt Time during the compaction cycleT Absolute temperature on liquidus curve of the binary phasediagram of an enantiomeric systemTR Absolute melting point of racemic compoundTm Melting pointt0ff Peak offset timeVT True volumeWC Work of compressionWD Work of decompressionx mole fraction enantiomeric compositionxiLIST OF TABLESTable PageI Various parameters calculated automatically by the 40software used to analyse the force versus timedata from the BetapressII Physical constants of the solids compressed on 63BetapressIII The Uy values determined from Heckel Plots of the 96various solidsIV Range of t0ff and AP/VT values corresponding to 100a range of maxV Range of W and WD values corresponding to a range 115of maxVI E values of various solids determined from the 122decompression analysisVII Literature values of E 123VIII Range of WC-WD, porosity and Ff values corresponding 132to the max range given in Table IVIX Classification of the deformation behaviour of 155various solids on the Betapress.xiiLIST OF FIGURESFigure Page1 Schematic illustration of the differences between 28the punch head profiles of Manesty and IPT typepunches2 Drawing illustrating the attachment of a LVDT to a 30punch for the direct measurement of punch displacements using an LVDT-slip ring system3 Representative upper and lower punch displacement 32profiles measured using an LVDT-slip ring system4a Schematic representation of punch displacement at 34various fractional turret positions4b Schematic representation of a machine deformation 35plot4c Schematic representation of the contribution of 37acceleration of the punches from an initial restingposition to their full velocity during thecompression phase5 Pressure-time curves for Avicel PH1O2, Spray-dried 43Lactose, and Emcompress6 Pressure and punch displacement-time curves showing 46stress relaxation at constant strain and peak offsettime for Avicel PH1O27 Upper punch force versus fraction of turret revolu- 50tion curves for Avicel PH1O2, Spray-dried Lactose,Emcompress and ‘steel+Emcompress’ tablets8 An array of evenly spaced decompression curves for 53‘steel+Emcompress’ tabl ets9 Diagrammatic representation of calculation of tablet 55expansionxiiiFigure Page10 Tablet expansion for various materials as a function 56of fraction of turret revolution during decompression11 Decrease in upper punch force during decompression 57as a function of compact expansion12 Schematic representation of tablet expansion at 59various stages during decompression (this figure isused to calculate the Young’s modulus)13 Chemical structure of ibuprofen 6514 Representative DSC curves of various compositions of 67ibuprofen15 Powder X-ray diffraction patterns of S-ibuprofen and 68racemic ibuprofen16 Isobaric binary phase diagram of ibuprofen enantiomers 70showing a racemic compound formation between S andR- i buprofen17 Test of Prigogine Defay equation used to calculate 73the phase diagram in Fig. 1618 Volumetric rate of compaction profiles during the 83compression phase19 Change in duration of the acceleratory phase of 86volumetric rate of compaction profiles with max forthree direct compression excipients20a Volumetric rate of compaction profiles superimposed 88on the corresponding stress profiles during thecompression phase for Avicel PH1O220b Volumetric rate of compaction profiles superimposed 89on the corresponding stress profiles during thecompression phase for Spray-dried LactosexivFigure Page20c Volumetric rate of compaction profiles superimposed 90on the corresponding stress profiles during thecompression phase for Emconipress20d Volumetric rate of compaction profiles superimposed 91on the corresponding stress profiles during thecompression phase for acetylsalicylic acid21 Plots of the maximum value of Heckel term (Hmax) at 95max during the compression phase against the corresponding max for different materials22 Variation in peak offset times with peak pressure for 99three direct compression excipients23 Variation in peak offset time with peak pressure for 102two methods of changing peak pressure24a Effect of turret revolution time on peak offset times 103for Avicel PH1O2 when compressed using Manesty punches24b Effect of turret revolution time on peak offset times 104for Emcompress when compressed Manesty punches25 Effect of punch type on peak offset times of Avicel 106PH1O2 and Emcompress26a Variation in peak offset times with peak pressure for 108various particle sizes of acetaminophen26b Variation in peak offset times with peak pressure for 109acetaminophen and its selected formulations27 Variation in force of failure with peak pressure 110for crystalline and direct compression forms ofacetami nophen28 Comparison of peak offset times for crystalline 111ibuprofen and a direct compression formulationxvFigure Page29 Decrease in punch stress during t0ff normalised for 113true volume (AP/VT) of different materials as afunction of t0ff30 Change in work of decompression with upper punch 117peak force31 Variation in modulus of elasticity of tablets with 119tablet porosity. Extrapolation of this plot to zeroporosity gives the Young’s modulus32a Variation in force of failure of tablets with upper 126punch peak force for sucrose and its directcompression forms32b Variation in force of failure of tablets with upper 127punch peak force for various celluloses32c Variation in force of failure of tablets with upper 128punch peak force for various lactoses33 A plot of the difference in the lost work (WC-WD) 133at highest and lowest max for several organicsubstances against their homologous temperatures34 A plot of Ff of the strongest intact tablet produced 135when a material was compressed over a range of max’against the corresponding lost work (WC-WD)35 A plot of the difference in the lost work (WC-WD) 137at highest and lowest max for several substances,against the corresponding change in compact porosity36 Compact volume of celluloses determined from compact 145dimensions measured immediately post-ejection as afunction of maxxviACKNOWLEDGEMENTSI am thankful to the following people:My research supervisor Prof. A.G. Mitchell, for his continuoussupport, encouragement and advice during my studies at The University ofBritish Columbia, and for providing a laboratory atmosphere congenial toindependent thinking and research. His guidance over the years has been aninvaluable experience for me. I will cherish his friendship for a longtime to come.Dr. J. A. Lund of the Dept. of Metals and Materials Engineering atUBC for critical evaluation of most of the work presented in this thesisand for his thoughtful suggestions at various times during the course ofthis work. His continued support as a member of my supervisory committeeis specially acknowledged.Dr. J. G. Sinclair, Dr. H.M. Burt, and Dr. A.H.L. Chow (now at GlaxoCanada Inc.) for their valuable support as members of my supervisorycommittee throughout the course of this work.My friend and research colleague Mr. R.J. Oates for insightfuldiscussions and ideas, and for writing the excellent computer softwarewithout which much of this work would be incomplete.My colleague Mr. Ibrahim El-Bagory for his help when needed, andMr. Charles Winternitz and Ms. Marion Wong for their sincere technicalassistance as summer students during the initial part of this work.Gifts of samples from the various suppliers, and Graduate Awards fromthe Berlex Foundation and the Novopharm Group (Stanley Drugs) aregratefully acknowledged.AlHSVVS0!TTAX11. INTRODUCTIONTableting by powder compaction can be described as a process ofincreasing the density of a powder bed with the intent of creating acertain degree of bonding between the powder particles, thereby obtaining acoherent compact capable of withstanding the stresses of handling andtransportation. Compaction is the general process of application ofmechanical force to a powder bed. The densification of the powder bedduring compaction is called compression and an increase in the mechanicalstrength of the powdered material due to bonding is called consolidation(Marshall, 1986). The compaction process involves application of force, orpressure, on the powder bed by confining the powder to a graduallydiminishing volume between two punches within a die (the compressionphase), and then releasing this force (the decompression phase) to allowthe removal of the tablet from the die. The most commonly used machines inthe manufacture of tablets by powder compaction are rotary tablet presseswhich compress the powders by a biaxial compression process. On a rotarytablet press the complete compaction cycle usually occurs in less than50 ms. Industrial scale tableting is therefore a high speed process.1.1 THE COMPRESSION PHASEDuring the compression phase, the powder particles rearrange underthe applied force and, above a certain force, undergo deformation. Theprocess of rearrangement causes an initial densification of the powder bedduring which some particle deformation may occur. The success of powdercompaction in producing strong, coherent tablets depends, at least in part,on the extent of deformation of the particles (crystals, crystallineaggregates, or granules containing some type of binder) during the2compression phase. Deformation exposes new, uncontaminated surfaces wherebonding readily occurs since these surfaces are in close proximity to eachother.Train (1956), Toure et al. (1980), and Parrott (1985) discussed thevarious stages which a powder bed goes through during compression. Theactual events are complex since the behaviour of the powder particlesduring compression is governed by factors such as pressure (or stress), theamount of deformation (or strain), and the rate of deformation (or strainrate) at the particulate level. Hence knowledge of stress, strain andstrain rate is important in understanding particle behaviour duringcompression. The nature of the equipment used, the physicochemicalproperties of the solid, and the ambient conditions of temperature andhumidity, are among the factors that can influence the stress-strainbehaviour of the particles during compression.1.1.1. Elastic, viscous and viscoelastic stress-strain behaviourTwo important mechanisms by which materials may deform under stressare elastic deformation and viscous flow. These are described by thefollowing theories:(1) The theory of elasticity, according to which the stress in elasticsolids is always directly proportional to strain but is independentof the rate of strain. This is the basis of the Hooke’s law ofelasticity, which can be expressed as,u = E•c (1)3where, a = stress, e = elastic strain and E is the modulus ofelasticity, or the Young’s Modulus, of the solid assuming anisotropic nature. The elastic strain is completely recoverable uponthe removal of stress.(2) The theory of hydrodynamics, according to which the stress in viscousmaterials (fluids) is always directly proportional to the rate ofstrain but is independent of the strain itself. This is described byNewton’s law of flow of viscous fluids:u= n.E (2)where, E = rate of strain, and q is the coefficient of viscosity.The strain in viscous materials is non-recoverable upon the removalof stress.The above two categories of deformation represent ideal stress-strainbehaviour. Real solids deviate from ideality and combine the elasticsolid-like behaviour with the viscous liquid-like behaviour. Suchmaterials, unlike elastic solids, do not maintain a constant strain under aconstant stress, but go on slowly deforming under this stress. If deformedto a constant strain, these materials will require gradually diminishingstress to maintain the deformation. The phenomenon of gradually increasingstrain under constant stress is called creep, while the phenomenon ofgradually diminishing stress under a constant strain is called stressrelaxation. Materials which show these deviations from ideality may not beable to recover their deformation instantaneously and/or completely upon4the removal of stress and are called viscoelastic materials (Popov, 1968;Ferry, 1980).The nature of viscoelastic materials is described by the theory ofviscoelasticity which is based on models combining one or more idealelastic solid element (a spring) with one or more ideal viscous fluidelement (a dashpot). The most simple is the combination of a spring and adashpot in series, giving a Maxwell model, or in parallel, giving a VoigtKelvin (or Kelvin) model. A combination of these simpler models in seriesor in parallel with each other and/or with the ideal elastic or viscouselements gives more complex models. Equations relating the stress, strain,and the strain rate in these models can be derived for each configurationof the elastic and viscous elements (Bland, 1960; Flugge, 1975). Theequations can be derived assuming unidimensional linear viscoelasticbehaviour, or can be based on the more complex three dimensional linearviscoelastic behaviour. The general purpose of these equations is toexplain the rate and extent of creep and stress relaxation phenomena whenmaterials deviating from the ideal elastic or viscous behaviour aresubjectedto stress.1.1.2. Creep and stress relaxation in pharmaceutical solids under pressureThe ability of a material to undergo creep or stress relaxation isrelated to its ability to deform by flowing under stress. Therefore anestimate of the rate and extent of creep or stress relaxation undercompression can be used as an estimate of the ability of the powderparticles to deform by a flow process. Creep and stress relaxation inpharmaceutical powders have been studied by compressing the powders forvarying lengths of time using various types of compression equipment.5The phenomenon of creep in pharmaceutical powders has beeninvestigated by measuring the change in strain as a change in the powderbed height under a constant stress, measured as the stress on the punches.Among the first reports was that by Okada and Fukumori (1975) who examinedthe creep properties of a number of inorganic salts by maintaining aconstant upper punch pressure on an isolated punch and die assembly for upto 10 hours. The thickness of the powder bed decreased and levelled offduring this period of time, suggesting that the particles of these solidsundergo some type of time-dependent deformation by flow under stress.Travers et al. (1983) and Celik and Travers (1985) used a hydraulic pressand recorded creep as ‘strain movements’, i.e., change in the height of thecompacts within the die, when the stress was maintained at a constant levelfor up to 60 s. Based on the amount of strain movement, selectedpharmaceutical solids were divided into compressible and poorlycompressible classes. Staniforth et al. (1987) used a similar testingprocedure to measure the creep behaviour of microcrystalline cellulose whenit was formulated with different amounts of water. They concluded that,while it differentiated the formulated material from the parent material,the creep analysis could not explain the differences between the tabletstrengths before and after formulation with water.Contrary to the assumption that the peak punch pressure occurs atmaximum punch displacement, Ho and Jones (1988a) reported the phenomenon of‘punch travel beyond peak force’ on a compaction simulator and related itto the plastic flow of materials during compression. This can beinterpreted as creep provided the peak force (or the stress) remainsconstant while the punches travel beyond the peak force. Doroudian (1991)recorded punch travel beyond peak pressure on a hydraulic press, but this6observation was complicated by a decrease in the pressure when the punchestravelled beyond the peak pressure. Hence, it seems that the punch travelbeyond peak pressure neither represents creep nor stress relaxation, sinceneither stress nor strain is constant.The study of creep during powder compaction requires the conditionsof constant stress, which can be attained under static conditions on slowspeed compaction equipment such as a hydraulic press. Creep analysis bythe above methods therefore differentiates the flow properties of materialsover time intervals uncharacteristic of real tableting conditions. Duringthe highly dynamic process of tableting on a rotary tablet press it will bealmost impossible to obtain the condition of constant stress. It ispossible that the differences in the flow properties, as indicated bydifferences in the creep behaviour under static conditions may eitherdisappear, or may become more pronounced, under dynamic conditions.During high speed tableting, stresses change rapidly in response tothe rapidly changing volume between the punch faces within the die.Deformation of powder particles by flow will occur within the confines ofthe die and the punch faces over a time period which is determined by thetype of compression equipment used. Since, during tableting, it isrelatively easier to measure the change in pressure as a function of time,the analysis of the stress relaxation phenomenon has received greaterattention than has the analysis of creep. For example, Shlanta andMilosovich (1964) observed a decrease in punch stress with time for anumber of pharmaceutical solids on an instrumented hydraulic press. Coleet al. (1975) reported differences in the extent of stress relaxation ofsodium chloride, potassium chloride, lactose and potassium citrate whenthese substances were compressed on a device specially developed to7simulate a rotary tablet machine. Using a hydraulic press similar to theone used by Shlanta and Milosovich (1964), Hiestand et a!. (1977) foundthat materials, the tablets of which fail by capping or lamination, showedslow stress relaxation. David and Augsburger (1977) quantitated stressrelaxation on a Stokes RB-2 rotary tablet machine under static conditionsby recording the decrease in pressure over a period of several seconds forsome direct compression excipients. Rees and Rue (1978) measured stressrelaxation on a Wilkinson STD 1 reciprocating tablet machine. They alsoindicated that a rotary tablet machine, such as the one used by David andAugsburger (1977), may not be a deflection-free system and therefore thestrains during the period of stress relaxation may not be constant. Pelegand Moreyra (1979) studied the effect of moisture on the stress relaxationpattern of powders under pressure using an Instron Universal TestingMachine. Caspar and Muller (1984) used a single punch eccentric press toobtain the stress relaxation data over a period of up to 60 s and founddifferences between the behaviour of selected pharmaceutical materials.More recently, using an Instron Physical Testing Instrument, Cutt et al.(1987) demonstrated the effect of wet granulation on the stress relaxationof a model system consisting of glass bellotini granulated with differentbinders. The granulated glass showed significantly enhanced stressrelaxation over a period of 6 minutes compared with non-granulated glass.Ho and Jones (1988b) studied the stress relaxation of a number ofpharmaceutical materials on a modern compaction simulator.These reports indicate both qualitative and quantitative differencesin the stress relaxation behaviour of various substances. Relative to thehigh speed operating conditions of rotary presses, the decrease in stresswas followed over long periods of time ranging from a few seconds to8several minutes. Only some of these studies provided evidence of constantstrain.To establish the relevance of stress relaxation in tableting onrotary presses, where compression normally occurs in less than 50 ms,evidence is required that stress relaxation indeed occurs at the speedsencountered on such machines and that the condition of constant strain issatisfied over the period of stress relaxation. To provide such evidenceit is necessary to investigate the relationships between stress and strainon a rotary tablet machine or on a compaction simulator. Stress relaxationstudies on a compaction simulator require an input of the punchdisplacement profile from the rotary tablet press in order to simulate itscompaction cycle. If the deformation of the tablet press under load isignored (section 2.3.2), the resultant errors in the displacement profileswill lead to inaccuracies in the estimation of strain during the period ofstress relaxation.1.2. THE DECOMPRESSION PHASEDuring compression, the distance between the punch faces decreases asthe tablet press does work on both the powder bed and the press (includingthe punches and die). The work done on the press deforms it elasticallysuch that the elastic energy is recovered during decompression. Bycontrast, only a small portion of the work done to the powder bed isrecoverable. The rest of the work is lost to friction, particledeformation, heat and other irreversible processes in forming a tablet.During the decompression phase the stresses built up during the compressionphase are rapidly released and the tablets are subsequently ejected fromthe die. The length of decompression phase on high speed presses is9normally less than 20 ms, during which time a tablet expands in an axialdirection within the confines of the die while still in contact with thepunches. The rapid decompression and the resultant expansion may causetablet failure by capping and lamination if the interparticulate bondsformed during the compression phase are too weak to withstand the stressesinduced by the decompression (Ritter and Sucker, 1980).1.2.1. Analysis of tablet expansion during decompressionTablet expansion during decompression has been studied using isolatedpunch and die assemblies mounted in stress-strain analysers (Travers etal., 1983; Malamataris et al., 1984; Bangudu and Pilpel, 1985; Celik andTravers, 1985) and compaction simulators (Yu et al., 1988). Single punchpresses fitted with linear variable differential transformers (LVDTs) havebeen used to measure punch displacements and press deformation (Ho et al.,1979; Juslin and Paronen, 1980; Lammens et al., 1980; Kaneniwa et al.,1984; Cook et al., 1988). Since tablet expansion during decompression isvery small relative to the elastic recovery of the press, it is essentialthat the LVDTs are mounted and calibrated such that tablet recovery can bedifferentiated from that of the press.It is technically more difficult to measure the punch displacement ona rotary tablet press (Ridgway-Watt, 1983, 1988; Walter and Augsburger,1986; Oates and Mitchell, 1990). As an alternative to direct measurements,Ripple and Danielson (1981) and Charlton and Newton (1984) calculated punchdisplacement from machine and punch head geometry assuming no machinedeformation and recovery.101.2.2. Use of tablet expansion in material characterisationRipple and Danielson (1981) and Danielson et al. (1983) analysed thedecompression of several pharmaceutical solids on a rotary tablet press andconcluded that the recovery of tablets during decompression is elastic innature. If this conclusion is extended further, it is possible that theelastic expansion of the tablets can be analysed by using the theory ofelasticity to obtain the Young’s modulus of the tablet material. Young’smodulus is an important fundamental material constant.The Young’s modulus, and other elastic moduli, of composites havebeen derived from physical models consisting of spherical or nonsphericalpores homogeneously distributed in an isotropic matrix (Rossi, 1968).Tablets are anisotropic, heterogeneous bodies in which the pore shape andstructure differ from the ideal nature required by the theoretical models.Therefore, their recovery and modulus at a given porosity, must bedetermined experimentally for each formulation and each set of tabletingconditions. The E of various compacted pharmaceutical materials has beendetermined in flexure tests (Church and Kennerley, 1982, 1983; Mashadi andNewton, 1987; Bassam et al., 1988; Agbada and York, 1990) and compressivetests (Kerridge and Newton, 1986). Roberts and Rowe (1987a, b) calculatedE using yield strengths estimated from Heckel plots (Heckel, 1961) and thetablet indentation hardness values, previously determined by Jetzer et al.(1983), in an equation given by Marsh (1964). The modulus at zeroporosity, E, was estimated by Kerridge and Newton (1986), Roberts and Rowe(1987a), and Bassam et al. (1988) using E in the equation of Spriggs(1961), and by Roberts and Rowe (1987a) using E in the equation ofWachtman (1969).11Problems associated with the comparison of E obtained from compactsof differing porosities and uncertainties in values of E obtained byextrapolating the values of E to zero porosity are avoided by singlecrystal tests. Ridgway et al. (1969) used a microtensile testing machinemodified for use in compression and estimated Young’s modulus from thestress-strain curves of a number of substances. Single crystalmicroindentation measurements were made by Duncan-Hewitt (1988), Duncan-Hewitt and Weatherly (1989b) and Wong and Aulton (1989) to determine E forsucrose and a-lactose nionohydrate, respectively. The modulus was found tovary with crystal face illustrating the anisotropic nature of thesecrystals. Similar conclusions can be drawn on the basis of an earlier workby Bridgman (1948) in which compressibility data for different faces ofsucrose crystals were obtained by a hydrostatic compression procedure, fromwhich the Young’s modulus for different faces can be estimated.It would be convenient if the data from a high speed rotary tabletpress could be used to characterise the decompression phase by a simpleanalytical procedure without having to resort to extensive viscoelasticmodeling. Such analysis would not only be useful in characterisingmaterials, but should also help in developing better formulations forproblem drugs.1.3. PARTICLE DEFORMATION UNDER PRESSUREThere are five possible mechanisms by which a particle may deformduring compaction:1. Elastic deformation which occurs by either stretching, bending, orcompression of the bonds in the crystal structure. This normally12occurs below the pressure at which permanent deformation by theprocesses 2-5 occurs.2. Plastic flow which occurs mainly by a crystallographic slipphenomenon, i.e. by the movement of one dimensional defects(dislocations) in the crystal structure. The presence ofdislocations in crystalline solids promotes plastic flow and flowgenerates new dislocations with increasing strain. Dislocations ondifferent crystallographic planes interfere with each other’smovement and the stress needed to continue deformation increases.This is known as work-hardening.3. Viscous flow which occurs normally by a bulk movement or displacementof molecules in the particle structure. Dislocations are not aprerequisite for this type of flow. Under load, solids exhibitingviscous flow behave as fluids with high viscosity coefficients.4. Viscoplastic flow which occurs partly by plastic flow and partly by aviscous flow process. Solids exhibiting viscoplastic deformation areless readily deformed than viscous materials.5. Fracture by a crack initiation and propagation process which occursas a consequence of the separation of certain lattice planes in thecrystals. Materials which deform by fracture are called brittle.There may also be non-crystalline materials, e.g., polymers belowtheir glass transition temperature, which will exhibit a brittlebehaviour at ambient temperature. Fracture is normally preceded by acertain degree of permanent deformation by plastic flow. Workhardening, and the related high localised stresses, due to plasticflow is one of the primary causes of fracture. Under certain13conditions of loading, some materials may fracture without priorpermanent deformation.The particles of most solids exhibit an initial elastic deformationphase followed by permanent deformation of some type as the powder bed issubjected to increasing loads during compaction. It is likely that a givenmaterial may exhibit a mixture of flow and fracture phenomena depending onthe conditions of stress, strain and strain rate.1.3.1. Viscoelastic modeling of the compression dataIn several of the reports mentioned in section 1.1.2 the stressrelaxation data has been fitted to the equations for a given type ofviscoelastic model. For example, David and Augsburger (1977) applied theMaxwell model of linear viscoelasticity to the stress relaxation data andreported a characteristic viscoelastic constant for each material. Reesand Rue (1978) found an exponential decay of stress, contrary to the lineardecay reported by David and Augsburger (1977). They argued that therecould be more than one characteristic viscoelastic constant for the powderscompressed. Peleg and Moreyra (1979) used stresses at the end of stressrelaxation normalised for the initial stress in a modification of theMaxwell model, and obtained easily interpreted linear plots, the slopes andintercepts of which were characteristics of a given material. Caspar andMuller (1984) fitted the stress relaxation data of some pharmaceuticalsolids to a model with at least 5 Maxwell elements.As mentioned above, viscoelastic materials are also characterised bystrain recovery which does not occur instantaneously and which may or maynot be complete. Ripple and Danielson (1981) and Danielson et al. (1983)analysed several pharmaceutical solids from this standpoint using the- 14punch-stress versus time data from the decompression phase of thecompaction cycle on a rotary press, and the stress-relaxation data from thepost-compression phase by leaving the tablet in a die instrumented tomeasure the residual die wall pressure. They fitted these data withequations derived from the three-dimensional linear viscoelastic theory,and concluded that the deformation under load could be best characterizedby the Voigt-Kelvin (or Kelvin) solid model in contrast with the Maxwellmodel. The recovery of the compacts during decompression was elastic,while the post-compression recovery, when the upper punch was no longer incontact with the compact, was viscoelastic.It can be seen that although the theory of viscoelasticity provides asound theoretical ground for analysing the deformation behaviour of solidsunder compression, there is a lack of agreement in the literature overwhich particular model, if any, can best describe the behaviour ofpharmaceutical solids during compression. Since most pharmaceutical solidsare expected to possess some time-dependent deformation, one source ofdisagreement could be the difference in the rates of deformation as aresult of different types of compression equipment used in the variousstudies. The theories of elasticity or viscoelasticity were developed fornon-porous, isotropic solid bodies. Thus another, mostly ignored,potential source of variation in the results and their interpretation, isthe effect of porosity of the powder bed and the effect of changes inporosity on particle deformation during compaction. A realistic analysisof the deformation behaviour should include deformation speeds (i.e., thestrain rates) comparable to those encountered on a rotary press, and theeffect of changes in porosity during the compaction process. One solutionwould be to use a high speed rotary press for the analysis.151.3.2. Modeling of compaction using mechanical properties from singlecrystal microindentation methodsThe actual deformation mechanism of crystals can be ascertained byusing single crystal microindentation techniques, but the particledeformation behaviour in a powder bed during compaction will be differentowing to the particle-particle interactions which are absent in themicroindentation tests. Nevertheless, the mechanical properties of singlecrystals derived from such tests can be used to model the compaction ofpowders. Modeling of compaction using mechanical properties fromindentation tests is well established in the area of material sciences, andhas been recently applied to the compaction of pharmaceutical powders.Duncan-Hewitt and Weatherly (1990a, 19Ob) modeled the densification ofpharmaceutical powders during uniaxial compaction using single crystalproperties derived from microindentation. The models differed for ductileand brittle materials, although in both cases the models required aknowledge of the change in the number of particle contacts duringdensification, the condition for local yielding at particle-particlecontact, the average area of interparticulate contact (which depends on theparticle geometry and the powder bed density), and the relationship between‘far-field stress and the local stresses at each contact’. Each of thesefactors were analysed using the mechanical properties, and thedensification as a function of pressure was calculated and compared withthe actual densification profiles according to the already establishedHeckel equation (Heckel, 1961).Although the modeling of powder compaction is mathematicallyappealing and scientifically rigorous, its use in the routine analysis ofpowder behaviour under pressure may be limited. Large, well-formedcrystals are required and specialised equipment is used to obtain the16single crystal properties. It may not be possible to obtain suitablesingle crystals and hence the mechanical properties from indentation forthose pharmaceutical powders that are either non-crystalline, or aregranulations containing more than one component in the same particle.1.3.3. Use of indices of performance in characterisation of materials.A large body of literature on the tableting of pharmaceutical solidsis related to the characterisation of materials by determining variousparameters which relate some measure of tablet strength to either thedeformation properties of materials under pressure or to their elasticnature under given conditions of stress. These parameters are called‘indices’ of performance of powders. The most commonly used indices oftableting performance are due to Hiestand (Hiestand and Smith, 1984;Hiestand, 1985). These indices, namely, ‘Strain Index’, ‘Bonding Index’and ‘Brittle Fracture Index’, represent ‘the relative strain during theelastic recovery that follows plastic deformation’, ‘the relative survivalduring decompression of the areas of true contact that formed at maximumcompression’, and the ratio of ‘the tensile strength of tablets with andwithout a hole at their center’. The wide application of these indices isrestricted because they require the use of specialized equipment, such as apendulum impact indentation hardness tester and a special die and punchassembly to form large rectangular compacts with a small cylindrical holein the centre. Moreover, it has been found difficult to reproduce theseindices from one laboratory to another (Duncan-Hewitt and Grant, 1987a).Malamataris et al. (1984) used ratios of two parameters, namely,‘elastic recovery’ and ‘plastic compression’ as indicators of the balancebetween the elastic and plastic behaviour of pharmaceutical powders on a17Dartec universal tester, and related this ratio to tablet strength. Celikand Travers (1985) proposed an ‘elastic recovery index’ defined as theratio of the elastic recovery of compacts to the strain movements (section1.1.2). This index was used to ascertain the compaction behaviour of somedirect compression bases on a hydraulic press. Guyot et al. (1986)proposed a ‘tabletability index’ and a ‘cohesion index’ to analyse thebehaviour of powder mixtures on a single punch eccentric press. Ho andJones (1988b) proposed ‘rise time’ as a new index of tablet compressionusing a compaction simulator. This time is the interval between initiationof compression and the point of maximum force.Most of the above indices or parameters are determined on slow speedcompaction equipment and therefore their conclusions may not be easilyextrapolated to the high speed rotary presses.1.4. EQUIPMENT USED FOR ANALYSIS OF POWDER BEHAVIOUR UNDER PRESSUREWhile rotary tablet presses are universally used for tabletmanufacturing, various other types of equipment have been used to analysethe behaviour of powders under pressure. These include various types ofhydraulic presses, mechanical stress-strain analysers such as InstronPhysical Testing Equipment, single station tablet presses, and the moremodern and expensive compaction simulators. The most obvious problems withthe use of hydraulic and single punch presses are their uniaxial compactioncycle as opposed to the biaxial cycle of the rotary presses, and theirinability to reproduce the rates of compaction obtained on the rotarypresses. The compaction simulators are designed to simulate the compactioncycle of the rotary presses (Cole et a!., 1971; Rees et al., 1972; Hunter18et al., 1976; Baternan, 1988; Bateman et al., 1989; Celik and Marshall,1989). While the advantage of compaction simulators is that it should bepossible to mimic the compaction cycle of any rotary tablet press byfeeding the punch displacement cycles to the computer controlling thesimulator, this very aspect of their use becomes their drawback. This isbecause the punch displacement profiles of rotary presses are difficult todetermine (Ridgway Watt and Rue, 1979; Walter and Augsburger, 1986; Oatesand Mitchell, 1990). The punch displacement profiles have been calculatedfrom the machine and punch geometry (Rippie and Danielson, 1981; Charletonand Newton, 1984). However, there is a significant error in thesecalculations because the contribution to the total displacement by themachine deformation is ignored. Oates and Mitchell (1989, 1990) proposed asystematic way of deducing the punch displacement profiles using arelationship between the machine deformation and applied force. Unless thecorrect profiles are fed to the compaction simulator, the results generatedare in serious error.1.5. PRELIMINARY WORK LEADING TO THE PRESENT RESEARCHExperiments in our laboratory indicated that the peak punch pressureson a Manesty Betapress occur prior to the vertical alignment of the puncheswith the centres of the upper and lower pressure rolls (the dead centreposition). The time by which the peak pressures were set off from such analignment was called the ‘peak offset time’ (t0ff), the duration of whichis apparently a characteristic of the compaction behaviour of the materialunder stress (Oates and Mitchell, 1989). Recently, Morehead and Rippie(1990) confirmed this observation on a Colton 216 rotary tablet press. The19punch stress profiles were not symmetrical about the punch displacementprofiles, and this asymmetry was attributed to time-dependent viscoelasticprocesses occurring within the compact. For substances such as Avicel,Klucel and Mannitol, they reported ‘maxima lead times’ analogous to thet0ff on the Manesty Betapress. The duration of t0ff is presumably relatedto time-dependent flow, and it is pertinent to investigate whether thisoccurs under conditions of constant strain.An observation of the force-time profiles on the Manesty Betapresssuggested that the small differences between the decompression curves ofvarious pharmaceutical materials might be used to calculate the in-dietablet expansion and to estimate the Young’s modulus, E, of the materialfrom this expansion by using the data obtained under normal operatingconditions. As mentioned above, the machine deformation has been ignoredby previous workers when calculating punch displacement on rotary presses.By contrast, the present work extends the relationship between punch forceand machine deformation developed for the compression phase of thetableting cycle of a Manesty Betapress to punch force and machine recoveryduring the decompression phase.1.6. OBJECTIVES OF THE PRESENT RESEARCHIn their reviews of the various techniques employed for investigationand interpretation of powder compaction, Krycer et al. (1982a, 1982b)indicated that ‘a great deal of confusion and discrepancy’ exists in theliterature. This was primarily because of the ‘many varied, and oftenunsubstantiated, techniques’ used. The brief review of the literaturepresented in the above sections indicates that the situation has notimproved greatly since these remarks were published.20In the present work a high speed rotary tablet press was used undernormal operating conditions with the following specific objectives:1. To develop an inexpensive and simple method of analysing thecompaction cycle which, in principle, can be applied to the analysisof powder compaction on any rotary press.2. To obtain parameters, under normal operating conditions, from thecompression and decompression phase of the compaction cycle, whichrelate to permanent deformation and elastic expansion.3. To characterise the deformation mechanism of various pharmaceuticalsolids using the information obtained in objective 2 with particularreference to the effect of strain rate.4. To demonstrate the influence of processing and formulation of variousdirect compression excipients and drugs on their deformationmechanism.21Anhydrous dicalcium phosphate, RhonePoulenc.Crystalline acetaminophen USP (20% particleslarger than 38 jim), Mallinckrodt.Crystalline acetaminophen USP (95% particleslarger than 45 jim), Mallinckrodt.Crystalline acetarninophen USP (60% particleslarger than 38 jim), Mallinckrodt.Crystalline sample, Monsanto.Anhydrous dicalcium phosphate, EdwardMendell.Roller-dried fl-lactose containing 73.9%fl-anomer, Sheffield.Direct compression formulation, Monsanto.Microcrystalline cellulose (average particlesize 200jim), FMC Corporation.Microcrystalline cellulose (average particlesize 5Ojim), FMC Corporation.Microcrystalline cellulose (average particlesize 90jim), FMC Corporation.Microcrystalline cellulose (average particlesize 2Ojim), FMC Corporation.Crystalline sample, MCB ManufacturingChemists.Dicalcium phosphate dihydrate, FMCCorporation.2. MATERIALS AND METHODS2.1. MATERIALSThe materials listed below were used as received from the varioussources:A-TabAcetaminophen Fine PowderAcetami nophen GranularAcetaminophen PowderAcetylsalicylic AcidAnhydrous EmcompressAnhydrous LactoseAsagranAvicel LargeAvicel PFI1O1Avicel PH1O2Avicel PH1O5CaffeineCal -Star22Cellactose A direct compression formulation with25% cellulose and 75% lactose , Meggle.Compap CG 90% Acetarninophen USP with 10% excipients,Mall inckrodt.Compap Coarse 73 L 73% Acetaminophen USP with 27% excipientsincluding lubricant, Mallinckrodt.Compap Coarse L 90% Acetaminophen USP with 10% excipientsincluding lubricant, Mallinckrodt.Compap L 90% Acetaininophen USP with 10% excipientsincluding lubricant, Mallinckrodt.DCI-63 Direct compression formulation containing63% ibuprofen with excipients, Mallinckrodt.Di-Pac Sucrose (98%) co-crystallized with dextrins(2%), Amstar.Di-Tab Dicalcium Phosphate Dihydrate, Rhone-PoulencElcema G250 Powdered cellulose, Degussa.Emcocel Microcrystalline cellulose, Edward Mendell.Emcompress Dicalcium Phosphate Dihydrate, EdwardMendell.Emdex Dextrates, Edward Mendell.Fast-Flo Lactose Spray-dried agglomerates of fine a-lactosecrystals bound by amorphous lactose (94.6%- anomer), Foremost.Lactose DCL 21 Roller-dried a-lactose containing 76.7%-anonier, De Melkindustrie Veghel.Mannitol Crystalline sample, Atlas.Mannitol M.G. Direct compression form, Roquette.Neosorb Direct compression form of sorbitol,Roquette.a-Lactose monohydrate Crystalline sample containing 98.2%a-anomer, BDH.23Potassium Chloride Crystalline sample, Fisher.R-ibuprofen Stereochemically 97% pure R-enantiomer ofibuprofen, Sepracor.Racemic ibuprofen Crystalline sample, Apotex.Racemic ibuprofen Crystalline sample, Upjohn Co.Rhodapap DC-P3 Direct compression form containing 97%acetaminophen and 3% PVP, Rhone-Poulenc.S-ibuprofen Stereochemically 96% pure S-enantiomer ofibuprofen, Ethyl Corporation.S-ibuprofen 99.7% pure S-enantiomer of ibuprofen, EthylCorporation.Sodium Chloride Crystalline sample, Allied Chemicals.Spray-dried lactose Spray-dried slurry of large a-lactosecrystals with some amorphous lactose(approx. 95% a-lactose), Foremost.STA-Rx-1500 Pregelatinized starch, Colorcon.Sucrose Crystalline sample, BDH.Sugartab Sucrose (90-93%) agglomerated with invertsugar (7-O%), Edward Mendell.Tn -Tab Tricalcium Phosphate, Rhone-Poulenc.Xylitol Crystalline sample, Roquette.242.2. CHARACTERIZATION OF MATERIALS2.2.1. Determination of bulk and true densityThe bulk densities of various materials were determined by aprocedure used in the Handbook of Pharmaceutical Excipients (1986). Aknown quantity (approximately 25g) of the powder was poured from a beakerinto a 100 mL measuring cylinder held at an angle of approximately 450 fromthe horizontal. The cylinder was brought to a vertical position and gentlyrocked sideways to level the top surface of the powder. The volume of thepowder to closest mL was noted. Dividing the mass of the powder by thisvolume gave the bulk density of the material.The true density (or the particle density) of various materials wasdetermined by helium-displacement pycnometry using a QuantachromeMultipycnometer. A 149.59 cm3 sample cell was used for all powders as thisgave density values closest to the values determined from a suspensiondensity method (Dwivedi, 1988).2.2.2. Differential scanning calorimetry (USC)The melting temperature of the various solids was determined byheating the samples in open pans using a Du Pont model 910 differentialscanning calorimeter controlled by a Du Pont Series 99 thermal analyzer.The signals from the DSC were fed to an Apple 11+ computer through avariable amplifier and an analog to digital converter. The DSC curves wererecorded and analyzed using a data analysis software. The melting pointswere recorded as intersections of the extrapolated leading edge of theendotherms with the baseline extrapolated from before the endotherm.Enthalpies of fusion were determined from the DSC curves using indium asthe calibration standard.252.2.3. Determination of binary phase diagram for characterization ofibuprofenDetermination of a melting point versus composition phase diagrambetween the constituent enantiomers provides the most fundamentalverification of the nature of the racemic substances. Since both, racemicibuprofen and S-ibuprofen, are available for tableting the binary phasediagram of ibuprofen was determined using the nearly pure samples ofR-ibuprofen and S-ibuprofen to ascertain the nature of the solid state ofracemic ibuprofen.Accurately known masses of R-ibuprofen and S-ibuprofen were weighedout directly into standard open aluminum pans to obtain enantiomericmixtures with evenly spaced mole fractions of S-ibuprofen between 0.04 and0.96. The total mass of each mixture was about 5 mg. The pans were heatedat 10°C/mm from 20°C to 90°C under a stream of nitrogen flowing at 138kPa. The samples were cooled to room temperature and reweighed. No weightloss was detected indicating that the volatility of ibuprofen (Ertel etal., 1990) was not a problem in the interpretation of the DSC results.Standard open pans were used instead of hermetically sealed volatile samplepans for quantitative DSC because of the greater contact area between thepan bottom and the constantan sample platform in the DSC cell. Nosignificant difference was found in the enthalpies of fusion (AH)determined in standard or volatile sample pans confirming that no solidvapor transition occurred under the experimental conditions. All samplesrecrystallized within 2 h of cooling, but were annealed at room temperaturefor 36 h before the DSC scan was repeated. Peak temperatures from themelting endotherms were plotted against the enantiomeric composition togive the binary phase diagram. The peak temperatures changed negligiblywhen the same samples were reheated after a storage for 6 months at room26temperature, indicating that the initial annealing of 36 h was adequate.Recrystallizing from the melt was preferred over recrystallization fromsolution which would require complete removal of the solvent and thepossibility of solvate formation would have to be ruled out.The phase diagram was verified by the thermal analysis of mixtures ofracemic ibuprofen with R-ibuprofen and S-ibuprofen and confirmed bycalculating the melting point at various enantiomeric compositions using anequation derived from the Prigogine and Defay equation (Prigogine andDefay, 1954).2.2.4. Powder X-ray diffraction patterns of ibuprofen samplesPowder X-ray diffraction provides information supplemental to thephase diagram. Powder X-ray diffraction patterns of the racemic ibuprofenand enantiomers of ibuprofen were obtained on a Rigaku Geigerflex X-raydiffraction system. The system was operated by an IBM compatible computervia a Rigaku D/MAX-B controller. The diffraction patterns were recorded ata continuous scanning rate of 5°29/min using CuK radiation (40 kV, 20 mA)with the intensity of diffracted X-rays being collected at intervals ofO.05°20. A Ni-filter was used to remove CuK radiation.2.3. COMPRESSION OF THE MATERIALS ON A ROTARY PRESS2.3.1. EquipmentA Manesty Betapress was used for all experiments. This is a sixteenstation rotary tablet press. One of the sixteen stations of the press wasfitted with a die with 1/2” (1.270 cm) flat-faced upper and lower punches,and the remaining fifteen stations were blanked off. No force was applied27to the punches by the precompression pressure rolls. The hopper and feed-frame were removed for easy access to the tooling.Two types of punches, namely, IPT and Manesty, were used. TheManesty punches are supplied by the makers of the machine, and conform tothe European standards. The IPT punches conform to the specifications ofthe Industrial Pharmaceutical Technology Section of the Academy ofPharmaceutical Sciences, American Pharmaceutical Association (Swartz,1969). The Betapress can be fitted with appropriate lower cam tracks toaccommodate the differences in the geometrical design of the two types ofpunches. The most important difference between these punches is in thegeometry of their heads. The IPT punches have a larger flat portion on thehead relative to the Manesty punches (Fig. 1).2.3.2. Instrumentation and calibration of Betapress to obtain force anddisplacementAnalysis of the tableting compaction cycle requires measurements ofboth punch force and the distance between the upper and lower punches(punch displacement) during powder compaction and consolidation as afunction of time, t, or fractional turret position, fr. Punch forcemeasurements are relatively easy to make using strain gauges or load cellsmounted either directly on the punches or on other, remote parts of themachine. Measurement of punch displacement, however, poses seriousproblems, arising mainly from the technical difficulties and expense ofrecovering signals from rapidly moving punches. An important objective ininstrumenting the Betapress, therefore, was to develop a method ofdetermining both punch force and displacement from measurements of punchforce only.28—2.540 2.5401.538 0.794oà3MANESTY PUNCH HEAD IPT PUNCH HEADFIG. 1. Schematic illustration of geometrical differencesbetween punch head profiles of Manesty and IPT type punches.The IPT punch heads have a relatively larger flat area. Alldimensions are in cm.29This required an analysis of the relationship between the appliedforce and punch displacement during the compaction cycle. The proceduresdeveloped by Oates and Mitchell (1989, 1990) were refined and simplified asoutlined below to measure force and then calculate displacement as afunction of the force exerted on the powder bed during the compressionphase of the compaction cycle. The method used to analyse expansion of thecompacted powder during the decompression phase is described in section2.6.The roll pin supporting the upper pressure roll and the cross beamsupporting the lower pressure roll were strain gauged to determine theupper and lower punch forces respectively (Oates and Mitchell, 1989). Thesignals from the strain gauged upper compression roll pin were calibratedagainst known loads before placing the pin in position. The signals fromthe strain gauges on the lower cross-beam were calibrated against the upperroll pin.To establish the relationship between punch force and displacement,punch displacements during the compression of selected materials using bothManesty and IPT punches were measured directly by a linear variabledifferential transformer (LVDT) whilst running the machine under speed.The LVDT was inserted in an empty punch hole in the turret, next to thehole in which the punches were placed, and its actuator arm was connectedto the punches by a specially designed linkage (Fig. 2). The signals fromthe LVDT were retrieved through a set of slip-rings mounted on the outersurface of the upper flange of the turret. To eliminate errors due topossible tilting of the punches as they travel between the upper and lowerpressure rolls, the LVDT was mounted in leading and trailing positions onboth the upper and lower punches. Thus, the net measured punch30Upper PunchTurret rotationto slip ringS—LinkageAdjustable platenFIG. 2. Drawing illustrating attachment of a linear variabledifferential transformer (LVDT) to an upper punch for directmeasurement of punch displacements using an LVDT-slip ringsystem on the Betapress.The armature of the LVDT rests on the adjustable platenwhich is a part of a linkage between LVDT and punches. Signalsfrom the LVDT were retrieved through slip rings mounted on theupper flange of the turret. The linkage is shown in a trailingposition relative to the punch as it rotates with the turret,but was also mounted in a leading position. Measurements inboth leading and trailing positions were used to annul errors indisplacement due to tilting of the punches with respect to theirvertical axis as they passed between the upper and lowercompression rolls.LVDT31displacement was a combination of four separate measurements. Typicalupper and lower punch displacements profiles obtained in this manner aregiven in Fig. 3.The net punch displacement, expressed as the distance, D, between thetwo punch faces during the compression phase was found to be a function ofthree independent variables:(a) the fractional turret position, fr, of the punches as they moveagainst the respective compression rolls under a negligible force,and hence under a negligible machine deformation,(b) the punch force, F, during compression, which causes significantmachine deformation (punch contractions + machine deflections;section 3.3.1) throughout the compression phase, and(c) the time, t, during which the punches accelerate from their initialresting positions. Hence there is a time-dependent change indisplacement during the initial part of the compression phase.The contributions to the displacement from these variables can berepresented as D(fr), D(F), and 0(t), respectively. D is then given by:D = D(fr) + D(F) + 0(t) (3)The terms on the right hand side of equation 3 were determined asfollows:(1) To obtain D(fr), displacement was measured using the LVDT-slip ringsystem under speed with the die cavity completely filled with a highviscosity oil. The oil was squeezed out of the die duringcompression, while maintaining sufficient force ( 1 kN correspondingto a pressure of 8 MPa) for the punches to follow the contours of32Punch Displacement (cm)perPuriIeO6Tablet Thickness- -0.8rPunchprofiIe1-0.04 -0.03 -0.02 -0.01 0Time during the compression phase (s) DEADCENTREFig. 3. Representative upper and lower punch displacementprofiles measured using an LVDT-slip ring system. Compressiontimes are expressed as negative values since the dead centreposition is assigned a value of zero time.The initial part of the upper punch displacement profileshows an upward movement because the upper punch resting on thepowder bed under gravity is initially pushed up by the lowerpunch until it comes in contact with the upper compression rolland subsequently begins moving down along the roll.33the compression rolls, but not to cause significant machinedeformation. The measured displacement profiles (Fig. 4a) werecurve-fitted using polynomial regression and the constants of thepolynomial were used to calculate the D(fr) component of D duringcompression.Rippie and Danielson (1981) and Charleton and Newton (1984)calculated the punch displacement profile from the geometry ofmachine and the shape of the punch heads. This is equivalent tosaying that D = D(fr) and is therefore in considerable error sincemachine deformation is neglected.(2) When a solid is compressed under speed, the displacement not onlychanges with fr, but also with F since the punches contract andbearings carrying the compression rolls, as well as the lower crossbeam, are deflected away from the powder bed together with the upperand lower punches. Thus additional changes in displacement, D(F),occur as a result of the machine deformation when solids arecompressed. To separate D(F) from D(fr), the D(fr) curves fromFig. 4a were subtracted from the total punch displacement profiles ofvarious particulate solids measured using the LVDT-slip ring system.The resultant difference, D(F) = D - D(fr), was plotted against Fduring the compression phase (Fig. 4b). At higher forces this plotwas linear indicating a linearity of the machine deformation with Fat these forces. The slope of the linear portion was the machinedeformation constant, K’ = 2.3 x i06 cm/N.(3) The initial part of the plot in Fig. 4b was curved due to a change inthe displacement as a result of acceleration of the punches fromtheir initial resting position, i.e., due to a time-dependent change34D(fr)B A4-toTime during the compression phaseFIG. 4a. Schematic representation of punch displacement atvarious fractional turret positions, D(fr), as the punchesfollow the contours of the compression rolls, expressed as afunction of time during the compression phase.D(fr) was obtained by directly measuring the upper andlower punch displacements with an LVDT-slip ring system whilstrunning the Betapress under speed. A negligible force wasapplied by filling the die with a high viscosity oil, which wassqueezed out during compression. Most solids only partiallyfill up the die, and, hence, show displacements during theshorter time interval ‘A’. Since the die was completely filledwith oil, punch displacements were recorded over an additionaltime interval ‘B’, during which time the punches alsoaccelerated to full velocity. Therefore, the curve during ‘A’corresponds solely to the change in punch displacement with fr,and can be used as the D(fr) curve to calculate punchdisplacements during compression of the powder bed.35D-D(fr)FIG. 4b. Schematic representation of a machine deformation plotobtained by subtracting the D(fr) curve (Fig. 4a) from themeasured punch displacement curve, D, during the compression ofa solid.An LVDT-slip ring system was used to measure D. The plotis linear at higher forces indicating that the machinedeformation (sum of punch contractions and machine deflections)is linearly related to the punch force during compression. Theslope of the linear part is the machine deformation constant,K1, and was used to calculate the contribution of the machinedeformation to the total punch displacement when solids arecompressed. The initial curvature in the plot is due toacceleration of the punches to their full velocity from aninitial resting position.K1 2.3 x 106 cm/NForce during the compression phase36in the displacement, D(t), during the initial part of the compressionphase. The term D(t) was obtained by extrapolating the linearportion of the plot in Fig. 4b to F = 0 and by subtracting theextrapolated plot from the plot in Fig. 4b. The difference{D-D(fr)]-K1.F was expressed as a function of time to obtainFig. 4c. The initial curved part of this figure is 1O ms long andcontributes O.O5 cm to D.(4) The results from the above experiments were combined to calculate thenet punch displacement, D. The calculated displacements for selectedsolids were verified by measuring the displacement profiles using theLVDT-slip ring system at different peak forces. The calculateddisplacements for both Manesty and IPT punches agreed with themeasured displacements.(5) The effect of force on machine deformation (the linear portion of theplot in Fig. 4b) was also confirmed by experiments under staticconditions in which different thicknesses of feeler gauges wereplaced between the punch faces at a position at which the puncheswere aligned with an imaginary line joining the centres of thecompression rolls, i.e., at t = 0 in Fig. 4a or 4c. This changed thedistance, D, between the punch faces thereby increasing F at a fixedfr. Plotting D against F gave a linear relationship with the slopebeing almost identical to the K’ value of 2.3 x i06 cm/N.Once the initial experiments to calculate punch displacement and itsverification by direct measurements were complete, the LVDT-slip ringsystem was abandoned and the results from the above experiments werecombined in displacement analysis. This analysis requires only an input of37Time during the compression phaseID-D(fr)]-K1 FFIG. 4c. Schematic representation of the contribution ofacceleration of the punches from an initial resting position totheir full velocity during the compression phase.The linear portion of the plot in Fig. 4b was extrapolatedto F = 0 and the extrapolated plot was subtracted from the plotin Fig. 4b. The difference [D-D(fr)]-K1.F was expressed as afunction of time to obtain Fig. 4c. The displacement of zO.05cm is the displacement during the initial time of 1O ms. Thecurve almost levels off after this initial time, indicating thatthe punches require 10 ms to accelerate to their full velocity.Approx.lOmstxO38the experimental F versus fr (or F versus time) curves measured using thestrain gauges. The computer software automatically calculates thecorresponding displacement.2.3.3. Data collection and analysisThe data collection was triggered by actuating a reed switch by asmall bar magnet affixed to the upper flange of the turret. The magnet waspositioned on the turret such that the data collection was triggered a fewmilliseconds before the punches came in contact with the compression rolls.The signals from the strain gauges (or the LVDT) were collected for aperiod of O.ls which was a sufficient interval to accommodate thecompaction cycle of all materials tested even at the slowest turret speedsused.Methods used for data collection, analysis and storage have beenupgraded since the reports by Oates and Mitchell (1989, 1990). Afteramplification and filtering, the analogue signals from the strain gauges onthe Betapress were converted to digital form using a Metrabyte 12-bit fastA/D converter, and collected by an IBM compatible computer at a rate of2500 readings per second for each of the two (upper and lower) punch forcechannels. The raw data were collected using a data acquisition softwareand were analyzed using a data analysis software with the aid of a mathcoprocessor fitted in the computer. Separate sets of software were writtenfor the analysis of compression and decompression phases. Each permitteddata analysis in two modes: one data file at a time in an individualanalysis mode or several data files together in a bulk analysis mode. Thetotal time taken for compression of a tablet and collection, storage andanalysis of the corresponding data in the individual analysis mode was39between 2-3 minutes. A number of parameters routinely computed by thesoftware are listed in Table I. The software could produce the results intwo formats, (a) the total value of a given parameter during thecompression or decompression phase, and (b) the profiles of theseparameters as a function of the fractional turret position of the punchesduring compression or simply the time during compression. The results weresaved as ASCII files which could be imported by various commerciallyavailable statistics and graphics software.2.3.4. Preparation of the materials for compressionTablets were made after mixing each material with the lubricantmagnesium stearate previously screened through a fine cloth sieve. Thegeometric dilution technique was used for obtaining a uniform distributionof the lubricant. In this technique, 0.5% of the sieved lubricant wasfirst mixed with an equal amount of the tableting material on a piece ofpaper using a spatula. The amount of mixture so produced was doubled inthe next step, then quadrupled, and so on, until all of the material wasused up. The total amount was finally mixed on a Fisher Kendell Mixer for5 minutes in closed jars.The internal lubricant was avoided in the case of materials whichpossess poor tableting properties, e.g., acetaminophen and ibuprofen. Thisis because the internal lubricant would compound the poor tabletingproperties by contaminating the potential bonding sites, if any, therebyinhibiting the bonding. These drugs, and some of their formulations whichdo not have a previously added lubricant, were tableted after lubricatingthe die wall and punch faces with a 5% solution of stearic acid inchloroform. The solution was applied to the punch faces and the die wall40TABLE I. Various parameters calculated automatically by the software usedto analyse the force versus time data from the Betapress.Parameter Absolute’ Profile2ValuePARAMETERS FROM THE COMPRESSION PHASEForce related parameters:Upper and lower punch force yaUpper and lower punch pressure Y VRatio of lower to upper force V -Displacement related parameters:Total punch displacement Y VPowder bed height at the onset of compression V -Tablet thickness at peak pressure V -Maximum and minimum relative density of the tablet V VMaximum and minimum value of the Heckel term Y VThermodynamic parameters:Work done to the powder bed during compression Y VPower V VMaximum power VTurret position where maximum power occurs YMachine speed related parameters:Compression time VDecompression time YDeformation related parameters:Peak offset time (t0ff) YDecrease in punch stress during t0ff VStrain rate related parameter:Rate of compaction - VPARAMETERS FROM THE DECOMPRESSION PHASETablet expansion during decompression Y VWork of decompression V -Elastic modulus of tablets V -1. The absolute values of the various parameters can be used to calculatecertain physical constants by either plotting these parameters againstpressure during compression or by using them in known equations fromthe literature.2. The software also provides the profiles of several parameters withrespect to time during compression or fraction of turret revolutionduring compressiona. V = yes41in a smooth motion using a small cotton swab wrapped around a thin woodenstraw, which produced a uniform layer of stearic acid in a few seconds whenthe chloroform evaporated. The punch faces and die wall were cleaned andthe solution was reapplied before each tablet was made.2.3.5. Compression protocolMethod I: All materials were compressed over a range of peak pressures byvarying the mass of the material in the die keeping the thickness settingon the Betapress fixed. Thus, an increasing amount of material wassqueezed in a volume which, in an empty die, would have otherwise stayedconstant, and this increased the punch pressure.Method II: A fixed mass of selected material was subjected to increasingpressures by changing the thickness setting on the Betapress such thatthe distance between the punch faces decreased.A series of about 20 tablets, covering a range of peak pressures(max) between 2O MPa to =210 MPa, were made with each material. A newaliquot of the material was weighed out for each pressure. Unlessotherwise stated, all results correspond to the variation in pressure byMethod I.2.3.6. Machine speedThe machine speed was expressed as ‘turret time’ which is the timetaken for one complete revolution of the turret. All materials werenormally compressed at a turret time of is. At this speed the machinewould manufacture approximately 1000 tablets per minute if all 16 stationsof the Betapress were used. Turret times of is, 0.88s, 0.75s and 0.65swere used for experiments in which the effect of the machine speed on42tableting parameters was studied. These turret times corresponded withaverage punch velocities in a vertical direction of about 21 cm/s, 23 cm/s,29 cm/s and 32 cm/s, respectively, during the initial part of thecompression phase.2.3.7. Relationship between fraction of turret revolution and turret timeThe position of the occurrence of various events during thecompaction cycle can be described either in terms of time (t) at a giventurret time relative to a reference position during the compaction cycle,or in terms of the turret position during its revolution (given as thefraction, fr, of the turret revolution) relative to the same referenceposition. The compression phase of the compaction cycle begins when theforce on the upper and lower punches is first measurable and ends when thepunches are vertically aligned with the axes of the two pressure rollsalong an imaginary vertical line called the line of dead centre. In termsof time this line was specified as t = 0 (Fig. 5), or as fr = 0 in terms offraction of turret revolution. The position fr = 0 or t = 0 was thereference position used throughout this work to divide the compaction cycleinto the compression and decompression phases, and to obtain the accurateposition of various events during compaction. Profiles of variousparameters during the compression and decompression phases were obtainedrelative to this reference position as a function of either t or fr.A multiplication by the turret time converted fr to t. Thus, at aturret time of is, fr = t. Unless otherwise specified all experiments wereconducted at a turret time of is, hence, the fr-axis on plots showingchange in a given parameter with respect to fr can be also be read as at-axis.43403020100Pressure (MPa)FIG. 5. Pressure-time curves for: (1) Avicel PH1O2, (2) Spray-dried Lactose, and (3) Emcompress. Turret time = is, IPTpunches.Line ofA Dead Centre1-40 -30 -20 -10 0 10Time (ms)44At different turret times, the positions of t = 0 after triggering ofdata collection were different. However, to maintain the uniqueness of thereference position at each machine speed, the position of fr = 0 was keptthe same as t = 0. The determination of position of t = 0 (or fr = 0)after the triggering of data collection is given in section 2.4.4.2.4. ANALYSIS OF THE COMPRESSION PHASE2.4.1. Calculation of stress, strain and strain rateStress was obtained as the punch force given by the strain gauges,divided by the cross sectional area of the powder bed within the die.During the compression phase, strain was defined as a decrease in theheight of the powder bed relative to its height at the instant before thedecrease. The strain between any two time fractions was calculated as thechange in the height of the compact between the two fractions divided byits instantaneous height at the earlier of the two fractions. Dividing bythe average height between these two fractions gave the average strain onthe powder bed. Using the volume of the powder bed in the calculationsinstead of its height gave the volumetric strain.Strain at a given time during compression was calculated using thedistance between the upper and lower punch faces. This distance wascalculated automatically by the data analysis software using thedisplacement analysis procedure. Volumetric strain rate was calculatedfrom the volumetric strain according to the above definition.2.4.2. Determination of tablet porosity (p)Porosity, p, of a tablet within the die at any time point during thecompression phase is given by p = 1- r’ where r = relative density45calculated as mass per unit volume of the material in the die divided bythe true density of the compact material. The data analysis softwarecalculated the volume of material in the die by multiplying the distancebetween the upper and lower punch faces by the cross-sectional area of thedie.2.4.3. Calculation of yield stress fry)The yield stress was determined from the equation in (l/p) = K.P + c(Heckel, 1961), where in (l/p) is the Heckel term and K = l/(3u). Aspecial case of the Heckel plot was used to determine °y The dataanalysis software gave the maximum value of the Heckel term (Hmax), at maxduring the compression phase. The yield stress for each material was givenby 1/(3K), where K is the slope of a plot of Hmax versus max over a rangeof max These plots were linear with r2 values ranging between 0.850 forS-ibuprofen and 0.995 for Di-Pac. The r2 values were generally above 0.95.2.4.4. Determination of peak offset timePeak offset time, t0.ff, was defined as the interval between the timeto reach max and t = 0 (Fig. 6). For an accurate determination of t0ff,it was necessary to know when t = 0 occurred following the triggering ofdata collection. A prefabricated hardened steel tablet (1.269 cm diameterX 0.3 cm thickness) within the confines of a die was virtuallyincompressible under the pressures used in this work and showed symmetricalpressure-time profiles with peak pressures, max’ at t = 0 (Oates andMitchell, 1989). The position of max was given by the data analysissoftware as the point where the derivative of pressure with respect to time46FIG. 6. Pressure and punch displacement-time curves showingstress relaxation at constant strain and peak offset time forAvicel PH1O2. Turret time = is, IPT punches.Pressure (MPa)Stress relaxationat constant strainDisplacement (cm)4030201000-0.1-0.2-0.3-0.4DEADCENTRE-20 -15Time (ms)47was 0. Thus, the position of t = 0, on an experimental pressure-timeprofile at a given machine speed, could be determined by compressing thesteel tablet at that speed and finding the position of max However, analternative and safer procedure was adopted as described below.When repeatedly compressed without ejection at pressures above 280MPa, Emcompress showed pressure-time profiles almost identical with thoseof the steel tablet. Therefore, the position of t = 0 at different speedswas found by determining the positions of max for Emcompress tablets madeby repeated compression, without ejection, at pressures above 280 MPa. Theaverage position of max for ten tablets at each turret time was used togive the average position of t = 0 at that speed. The coefficient ofvariation of these averages was less than 0.4% at each turret timeindicating high precision in the position of t = 0. When provided with theturret time and the position of t = 0, the data analysis softwareautomatically determined the value of t0ff.2.4.5. Factors affecting t0ffThe following experiments were performed to study various factors likely toaffect t0ff:Experiment I: Using IPT punches at a turret time of 1 s, tablets were madefrom Avicel PH1O2, Emcompress and Spray-dried Lactose over a range of maxby (a) varying the mass of the tablet, with the tablet thickness setting onthe machine fixed (Method I, page 41), and (b) fixing the mass of thetablet and varying the thickness setting (Method II, page 41).48Experiment II: Experiment 1(a) was repeated at different turret timesusing Manesty punches in order to study the effect of machine speed on thet0ff for Avicel PHJO2 and Emcompress.Experiment III: To study the effect of different punch types on t0ff,tablets were made over a range of max at a turret time of 1 s using bothIPT and Manesty punches.Experiment IV: To study the effect of formulation on t0ff for drugs withpoor compression properties such as acetaminophen and ibuprofen, tabletsfrom crystalline samples and the directly compressible commercialformulations of these drugs were made at a turret time of 1 s using IPTpunches. The change in t0ff with increasing max was recorded for eachsample.2.4.6. Determination of decrease in pressure during t0ff (Al’)Figure 6 shows that the pressure decreases during t0ff. The decreasein pressure between the point where t0ff occurred and t = 0 was termed liP.The software automatically calculated liP by subtracting the pressure att = 0 from max To compare different materials, the value of liP for eachmaterial at different pressures was normalised for the corresponding truevolume (mass of the tablet divided by the true density of the material) ofthe materials in the die.2.4.7. Calculation of work of compression (We)The method of calculating W was reported by Oates and Mitchell(1989, 1990). This method determined W as a product of force exerted by49the punches to form the compact and the corresponding displacementcorrected for the machine deformation during the compression phase. It wasshown that ignoring the machine deformation, and calculating the punchdisplacement solely on the basis of machine geometry, as proposed byprevious authors (Rippie and Danielson, 1981; Chariton and Newton, 1984),resulted in errors in WC of up to 40% (Oates and Mitchell, 1990).2.6. ANALYSIS OF THE DECOMPRESSION PHASEThe decompression phase begins at fr = 0 and ends when the appliedvertical force, F, experienced by the tablet drops to zero, i.e., when thefraction at which F = 0 (frFO) is reached (Fig. 7). During this phase themachine deformation is recovered and the tablet expands. Thus, the F-frcurve corresponds to the total recovery of the machine and the tablet. Toobtain the expansion of the tablet, the machine recovery must be subtractedfrom the total recovery. This idea was the basis for analysing thedecompression phase to obtain tablet expansion and related parameters.2.6.1. Determination of machine deformation during decompressionThe force applied by the press during the compression phase causesthe punches and the press to deform. The sum of punch contractions andpress deflections is the total machine deformation, Dm, which is totallyrecovered during decompression. At any given fr during the decompressionphase, D is directly proportional to F:ADm = K1AF (4)501612840Upper Punch Force (kN)FIG. 7. Upper punch force versus fraction of turret revolutioncurves for: (1) Avicel PH1O2, (2) Spray-dried Lactose,(3) Emcompress, and (4) steel+Emcompress tablets. At a turrettime of is, the x-axis also gives the time for compression anddecompression in fractions of a second.LINE OFDEAD CENTRE123-0.04 -0.03 -0.02 -0.01 0 0.01Fraction of Turret Revolution51where, IxDm = the change in machine deformation, K1 = the machinedeformation constant (section 2.3.2) and AF = the change in the verticalforce.Throughout the compaction cycle, the upper and lower punch heads arepressed against their respective pressure rolls. The pressure rollsconstrain the punches and, during powder compaction, force them togetherthereby decreasing the distance between the two punch faces. If the powderbed could be replaced by an ‘incompressible’ solid, the distance betweenthe punch faces would be constant throughout the compaction cycle. Hence,a F-fr curve for an incompressible solid would be proportional to thecorresponding Dm-fr curve according to equation 4.2.6.2. Selection of the incompressible materialThe ideal incompressible solid would fill the space between the punchfaces, i.e., it would be completely non-porous, and would have a negligiblechange in dimensions during the compression and decompression phases.Theoretically any liquid would be suitable as an incompressible material,but only if it were not squeezed through the space between the punches andthe die wall. After initial experiments with plasticine and silly putty inmuslin bags, it was decided to use steel, which could be machined intotablets which fitted the die cavity snugly, and then hardened. However, aseries of tablets with varying thicknesses would have been required to varythe punch force at a fixed tablet thickness setting on the press, and therewere potential dangers in the impact between the punches and the steeltablets, especially at high forces. Therefore, it was decided to use thehardened steel tablet described in section 2.4.4, which gave almost52negligible punch force when the press was operated at a thickness settingof about 0.3 cm, and to cushion the impact of collision of the punches withthe tablet by putting a layer of Emcompress on its top surface. Asmentioned earlier (section 2.4.4), repeated compression makes Emcompressbehave as an incompressible solid. Thus, ‘steel+Emcompress tablets’ wereused as the incompressible solid for the decompression analysis.2.6.3. Determination of tablet expansionA series of F-fr curves was determined by ‘compressing’ theincompressible solid under running conditions. The ejection cam wasremoved and the thickness setting was fixed to give a distance of about0.3 cm between the punch faces at fr = 0 in an empty die. The steel tabletwas inserted into the die cavity and various weighed amounts of Emcompresswere added to increase the peak force. After each addition, the steelEmcompress tablet was repeatedly compressed without ejection to minimizeany Emcompress recovery. By increasing the mass of Emcompress from 80 to430 mg in 10 mg increments, about thirty five F-fr curves were recordedwith peak forces ranging from 2 to 35 kN. A computer program converted thedata from the decompression phase into an array of F-fr curves for evenlyspaced values of F at fr = 0 (Ffr0) ranging from 0 to 35 kN (Fig. 8).From this array, a F-fr decompression curve could be computed byinterpolation for any Ffr,0 in the range 0 to 35 kN (0 to z270 MPa).A tablet which does not expand axially upon decompression would havea F-fr curve which coincides with the curve computed from the array at amatching Ffr0. When a tablet expands axially it would cause additionalmachine deformation, ADm, which would correspond to the expansion of thetablet and results in a corresponding increase in force, AF. To determine--‘001001I..)0r\)C)Ci)010aL,z--—<CDGCD•-il-jo+-mC) CDC) 0 -S CD-50,0-II-O—‘JCD0 .+-I,0 •CDCDCD-CD-no=—-0-••0CD‘CD00.—CDon0CD -<.-CD- CD—I0noc-f-0—‘•—-0o- CD•0•0 0CD 0 -fc-f-0o-C CD -U C 0 ‘-I 0 1 C) CD z0 p 0-rio-01C) 0 0 -I’ -I. CD 0 C 00 -a UI p 0 r.354AF, the F-fr curve derived from the array was subtracted from the F-frcurve of the test material. A ADm-fr curve was obtained by converting AFto ADm using equation 4. The range of turret positions over which ADm wasdefined began at fr=0 and ended where the force on the array curve droppedto zero. The procedure for deriving ADm from AF is illustrated in Fig. 9.Some representative plots of ADm-fr for various materials are shown inFig. 10.Any deformation and elastic recovery in the steel tablet would leadto errors in the estimation of ADm. Errors in ADm would increase withincreases in compression force and would be greatest for tablets undergoingthe least expansion during decompression. Thus, using an elastic modulusfor steel of 200 GPa (Popov, 1968), it was found that the maximum error inestimates of ADm would be about 2.4 percent for Avicel PH1O2, which had ahigh elastic recovery on decompression, and about 5.0 percent forEmcompress, which had low elastic recovery.2.6.4. Determination of work of decompressionThe F-fr (Fig. 7) and ADm-fr (Fig. ) plots were combined to obtainplots of FADm (Fig. 11). The F-ADm curves were linear, with r2 valuesusually above 0.95, and hence could be extrapolated to F = 0 to obtain aquantity called AD at F = 0 (ADmFo). The area under the curve from ADm =0 to ADmFO was the work done by the expanding tablet on the machine duringdecompression. The work of decompression, WD, is the negative of thisarea, where the negative sign indicates that energy is lost by the tabletduring expansion.5586420Upper Punch Force (kN)FIG. 9. Diagrammatic representation of calculation of tabletexpansion. AF is the difference in force between thedecompression curve of a test material and an array curve fromFig. 8, and is proportional to AD, the expansion of the testAvicel PH1O21EF Steel+Emcompress0 0.004 0.008 0.012 0.016Fraction of Turret Revolutionmaterial.56ADm (i’m)100-Avicel PHIO20 Spray-dried Lactose80- A Acet. GranularC Emcompress60 -40-AêQ u020- Q0 C_H0 I0 0.004 0.008 0.012Fraction of Turret RevolutionFIG. 10. Tablet expansion for various materials as a functionof fraction of turret revolution during decompression.57Upper PUflCh Force (kN)84100FIG. 11. Decrease in Upper Punch force during decompressj0Each plot represents the expansion of a single tablet and thearea from ADm 0 to 0mF—O gives the work of decompressjo201612AvlceI PHIO20Cspray_dried LactoseAcef. Granularmcompresg0 20 40 60LDm (Lim)80582.6.5. Determination of Young’s ModulusAssuming tablet expansion during decompression was elastic, the forceexerted by the expanding tablet on the upper punch was expressed by thefollowing form of Hooke’s law:a = E.e (5)where, a = the axial stress given by F divided by the cross-sectional areaof the punch face, E = the modulus of elasticity of a tablet with acertain degree of porosity, and e = axial strain which was obtained asfollows:a. Relative to compact thickness at the end of decompression, HFO(Fig. 12), the change in the thickness at any fraction duringdecompression, AH(fr), was given by:AH(fr) = ADmF=O - ADm(fr) (6),where the values of ADm(fr) and ADmFO were obtained from Figs. 10 and11, respectively.b. Tablet thickness at the end of decompression, HFO, was calculated from:HFO HfrO + ADmFO (7),where, HfrO, the tablet thickness at fr = 0, was given by,Hfr0 = K FfrO + 0.314 cm (8).59Maximum Stress___________Stress • 0Maximum Strain Elastic Expansion ‘ Elastic Strain - 01AH(fr)1 11AHF..o JADm(fr)ADFOI HF.oIHtr.ofr-O Any fr End ofdecompressionDECOMPRESSIONFIG. 12. Schematic representation of tablet expansion atvarious stages during decompression (drawing not to scale). Thequantities shown are used to calculate the modulus oftablets.60The intercept of 0.314 cm is the distance between the punch faces in anempty die at fr = 0 and was close to the machine thickness setting ofabout 0.3 cm. Equation 5 was derived from an independent experiment inwhich increasing weights of lead shot (#4, Winchester) were compressedby hand turning the press to give different values of FfrO. Thethickness of the ejected lead tablets was measured. Linear regressionof this thickness against Ffr,0 (r2 = 0.985) gave K1 = 2.2 x io6 cm/Nwhich was close to the value of 2.3 x i06 cm/N determined under staticconditions using feeler gauges (section 2.3.2). This indicates thatthe elastic recovery of compacts of lead shot was negligible upondecompression.c. Combining equations 6 and 7 gave strain as a function of fraction,c(fr):c(fr) = AH(fr) / HF,0 (9)Using equation 9, the plots of ADm-fr were converted to plots ofc-fr. Dividing F by the cross-sectional area of the punch face convertedthe plots of F-fr to u-fr. The u-fr and c-fr plots were then combined toobtain stress-strain (u-c) curves, the slopes of which gave E for a giventablet according to equation 5. Extrapolation of the plots of E versus p(porosity at the end of decompression) to p = 0 gave the values of Young’smodulus for the tablet material.2.6.6. Determination of porosity at the end of decompressionThe compact porosity at the end of decompression was calculated fromthe mass and true density of the material, and the distance between the61punch faces at the end of decompression (hence the in-die volume of thecompact at the end of decompression). At the end of decompression thedistance between the punch faces equals HFO, where HF,O was calculatedfrom equations 7 and 8. The porosity at the end of decompression waswithin 95% confidence intervals of the minimum porosity during thecompression phase. Hence, the minimum porosities during the compressionphase were used to construct the plots of E versus p.2.7. DETERMINATION OF TABLET STRENGTHThe strength of the tablets was determined in a diametral compressiontest using a CT-40 tablet strength tester (Systems Engineering). On thistester a tablet is placed vertically on a small round platform, an uppercross-beam with a platen moves down onto the tablet and applies forceacross its diameter. The platform is connected to a strain gauged loadcell the analog signals from which were recorded via an analog to digitalconverter on an Apple 11+ computer and analysed to obtain a force-timecurve up to tablet failure. All tablets were stored at about 21°C under arelative humidity of 30-40% for 24 h before the determination of the forceof failure.623. RESULTS AND DISCUSSION3.1. MATERIAL PROPERTIESThe true and bulk densities of various solids are given in Table II.These values were verified by comparing with other, independentmeasurements, or by comparison with the values in the literature. Thevalues of true densities obtained from the helium pycnometry were verifiedby using suspension density measurements on selected solids (Dwivedi,1988). The bulk densities for most excipients are close to those reportedin the Handbook of Pharmaceutical Excipients (1986).The melting points of several solids are given in Table II. In mostcases these correspond to the melting of crystalline samples of each solid.With certain formulations (e.g., for some Compaps), multiple peaks wereobtained on the DSC curves, and the temperature on the leading edge of thefinal endotherm was taken as the melting point. The melting points of someother solids (e.g., sucrose, lactose) correspond to melting withdecomposition. The melting points for celluloses are decompositiontemperatures. The melting points of samples containing cr-lactosemonohydrate (e.g. a-Lactose monohydrate, Fast-Flo Lactose and Spray-driedLactose) correspond to the melting of the sample after dehydration. Thedehydration causes a small degree of anomeric conversion from the a-anomerto fl-anomer in open pans (Dwivedi, 1988).The melting points were used to calculate the homologous temperatureof each solid where the homologous temperature is the ratio of absoluteroom temperature (293 K during tableting on the rotary press) to theabsolute melting temperature of the solid (Table II). The homologoustemperature indicates the proximity of room temperature to melting63TABLE II. Physical constants of the solids compressed on Betapress.Material True Bulk Melting3 Homologous4Density1 Density2 Temperature Temperature(g/cm3) (g/cm3) (K)A-Tab 2.774 0.68 - -Acetaminophen Fine Powder 1.301 0.26 443 0.661Acetaminophen Granular 1.294 0.64 443 0.661Acetaminophen Powder 1.296 0.35 443 0.661Acetylsalicylic Acid 1.350- 432 0.678Anhydrous Emcompress 2.780 0.74 - -Anhydrous Lactose 1.564 0.51 508 0.577Asagran 1.385 0.59 432 0.678Avicel Large 1.555 0.32 538 0.545Avicel PH1O1 1.556 0.28 538 0.545Avicel PH1O2 1.549 0.33 538 0.545Avicel PH1O5 1.556 0.25 538 0.545Caffeine 1.458 0.29 511 0.573Cal-Star 2.316 0.86 - -Cellactose 1.542 0.37 471 0.622Compap CG 1.300 0.43 441 0.664Compap Coarse 73 L 1.395 0.56 439 0.667Compap Coarse L 1.290 0.54 439 0.667Compap L 1.306 0.44 441 0.664DCI-63 1.233 0.57 342 0.857Di-Pac 1.543 0.66 460 0.638Di-Tab 2.330 0.83 - -Elcema G250 1.525 0.34 538 0.545Emcocel 1.539 0.31 538 0.545Emcompress 2.353 0.81 - -Emdex 1.504 0.64 421 0.697Fast-Flo Lactose 1.533 (0.69) 488 0.600Lactose DCL 21 1.561 0.59 508 0.577Mannitol (Crystalline) 1.490 0.52 431 0.680Mannitol M.G. 1.482 0.66 428 0.685Neosorb (Sorbitol) 1.487 0.59 369 0.795a-Lactose monohydrate 1.538 0.43 488 0.600Potassium Chloride 1.980 - - -Racemic ibuprofen 1.120 0.38 348 0.842Rhodapap DC-P3 1.295 0.39 441 0.664S-Ibuprofen 1.096 0.45 323 0.907Sodium Chloride 2.170 (0.93) - -Spray-dried Lactose 1.538 0.64 489 0.599STA-Rx-1500 1.480 0.65 - -Sucrose (crystalline) 1.584 0.78 453 0.647Sugartab 1.560 0.61 443 0.661Tn-Tab 2.883 0.78 - -Xylitol (Crystalline) 1.533 0.71 364 0.8051. Determined by helium pycnometry (coefficient of variation < 0.15%,n = 8); 2. Determined using a procedure adopted from the Handbook ofPharmaceutical Excipients (1986). The values in parenthesis are taken fromthis Handbook; 3. Determined using DSC (see text in section 3.1 forcomments on these values); 4. Ratio of room temperature (293 K) to theabsolute melting temperature.64temperature, and is an useful parameter in explaining the deformationbehaviour of the solids (section 3.6.2).3.2. THE SOLID STATE OF IBUPROFENCurrently, there is great interest in the significance of drugchirality in pharmaceutical development and regulation (Hutt, 19l) andincreasing emphasis is being placed on the use of single enantiomers,rather than the racemic drug, in dosage forms. Ibuprofen is a chiral drug.The chemical structure of ibuprofen in Fig. 13 shows that the molecule hasone chiral center. Hence, there are two enantiomers. Racemic ibuprofencontains equal amounts of these enantiomeric molecules. Both S-ibuprofenand racemic ibuprofen are commercially available for use in tableting.These samples were characterized to establish the differences in the natureof their solid state.Racemic drugs may be classified into three types based on a meltingpoint phase diagram. The most common type is the racemic compound whichcontains an equal number of molecules of each enantiomer in the unit cellof the crystal. The solid therefore is a one phase crystalline additioncompound, and the binary phase diagram of the two enantiomers shows twoeutectic points. The melting point of a racemic compound may be eitherbelow or above the melting point of the pure enantiomers.The second less common type of racemic drug is the racemic mixture orconglomerate in which the two enantiomers crystallize separately. Thesolid therefore is a two phase physical mixture. In the binary phasediagram there is only one eutectic point which corresponds to the meltingpoint of the racemic mixture.65COOHCH3HCOOH(S)- ( +) - lb up ro fe(R )— ( - ) - lb up ro fe nFIG. 13. Chemical structure of ibuprofen.66The third type of racemic drug is obtained when the two enantiomersform a continuous series of solid solutions (mixed crystals). A solidsolution of the two enantiomers with equimolar enantiomeric composition isdesignated a racemic solid solution or pseudoracemate.The literature sometimes refers to racemic ibuprofen as a racemiccompound and sometimes as a racemic mixture. The USP describes ibuprofenas a ‘(±) mixture’ which is misleading. To identify the racemicmodification of ibuprofen, DSC and powder X-ray diffraction were used andthe two component phase diagram between R and S-ibuprofen was constructedbased on the DSC results.3.2.1. USC and powder X-ray diffraction of ibuprofen samplesDSC scans indicate that the melting points of R-ibuprofen andS-ibuprofen are nearly identical. Some representative DSC curves ofvarious samples of ibuprofen are given in Fig. 14. The melting point ofS-ibuprofen is much lower than that of racemic ibuprofen and samples ofintermediate composition show a clearly defined eutectic melting endotherm.This suggests that racernic ibuprofen is a racemic compound. The finalmelting of DCI-63, the formulation containing 63% ibuprofen, occurred at atemperature close to the melting point of racemic ibuprofen which indicatesthat racemic ibuprofen was used to prepare this formulation.The powder X-ray diffraction patterns (Fig. 15) of the twoenantiomers were identical, but different from the pattern of racemicibuprofen. This confirms that racemic ibuprofen is a racemic compoundcapable of existing as a separate phase in the solid state independent ofits constituent enantiomers. The diffraction pattern of a racemic mixturewould be identical with that of its enantiomers, while that of a racemic67120 30 40 50 60TEMPERATURE (°C)80 90FIG. 14. Representative DSC curves of various compositions ofibuprofen: (1) 96% S-ibuprofen, (2) 27.6 % S-ibuprofen,(3) 46 % S-ibuprofen, (4) racernic ibuprofen.ENDO17068cnaU>-FzUFzS-I bu pro tenRacemic Ibuprofen1 0.0 1 5.0 20.0 25.0 30.02 THETA65005200390026001 30005.0 35.0FIG. 15. Powder X-ray diffraction patterns of S-ibuprofen andracemic ibuprofen. The pattern of R-ibuprofen was identical tothat of S-ibuprofen. The pattern of S-ibuprofen is shown with abaseline offset.69solid solution would show slight shifts in the peak positions dependingupon the nature of the solid solution (whether it is a substitutional or aninterstitial solid solution). The term racemic mixture, often used in theliterature to describe crystalline racemic ibuprofen, is thereforeincorrect and misleading.3.2.2. The binary phase diagram of ibuprofenFundamental confirmation that racemic ibuprofen is a racemiccompound, and not a racemic mixture, was provided by the binary phasediagram (Fig. 16). The phase diagram is characteristic of a eutecticsystem with a binary addition compound formation. The melting points, Tm,determined by extrapolating the leading edge of the melting endotherm tothe baseline, were 46°C, 46.5°C and 71°C for R-ibuprofen, S-ibuprofen andracemic ibuprofen, respectively. The eutectic temperature was 37°C and theeutectic points occurred at approximately 0.18 and 0.82 mole fractions ofS-ibuprofen. The extrapolation method gives accurate values of Tm forsingle phases such as R-ibuprofen, S-ibuprofen and racemic ibuprofen butbecomes unreliable for mixtures of intermediate composition due tobroadening in the eutectic and melting endotherms (Fig. 14). Hence, forconstructing the phase diagram the temperatures at the peaks of theeutectic and melting endotherms (i.e., the end of fusion) were used. Thepeak temperatures are a function of mass of the sample, but theapproximately constant sample mass (5 mg) reduces the possibility of massrelated effects when comparing results from different experiments. Theeutectic melting peaks could not be observed for samples with compositionsbelow about 0.18 mole fraction S-ibuprofen and above about 0.82 molefraction S-ibuprofen due to their close proximity to the terminal70PEAK TEMP (°C)80C70U6050CCU U uLi u40 I I I I0 0.2 0.4 0.6 0.8 1MOLE FRACTION S-IBUPROFENFIG. 16. Isobaric binary phase diagram of ibuprofen enantiomersshowing a racemic compound formation between S and R-ibuprofen.Open squares: fused, recrystallized, and annealed mixtures ofR-ibuprofen and S-ibuprofen; Crossed squares: fused,recrystallized and annealed mixtures of R-ibuprofen andS-ibuprofen with racemic ibuprofen; Circles: points calculatedusing equation 11; Diamond: 99.7% S-ibuprofen from EthylCorporation.71enantiomeric melting peaks. The melting point of a 1:1 fused mixture ofR-ibuprofen and S-ibuprofen coincided with the melting point of racemicibuprofen. Additional points on the phase diagram were obtained byplotting the melting points of fused samples containing various proportionsof racemic ibuprofen with either R-ibuprofen or S-ibuprofen (Fig. 16).These points were in accordance with the fact that admixture of a racemiccompound with either of its component enantiomers will depress its meltingpoint. The melting point of a racemic mixture would be the minimumtemperature on a binary phase diagram, i.e., the eutectic point, andadmixture with either pure enantiomer would elevate it to highertemperatures.In the phase diagram of a racemic compound, the absolutetemperatures, T1, on the liquidus curve, under which the racemic compoundexists as a stable phase, can be predicted as a function of mole fractionenantiomeric composition, x, from its heat of fusion (AH) and absolutemelting point (TR, end of fusion) by using these quantities in theequation:2AHR 1 1ln 4x(1-x) = ( ) (11)R TR Twhere AHR is assumed to be constant over the range of compositions forwhich T is predicted. This equation can be derived from that of Prigogineand Defay (1954). Figure 16 shows that the experimental liquidus curvedetermined from the DSC observations on ibuprofen was in good agreementwith the liquidus curve predicted using equation 11. The prediction,however, is accurate only in the region of the congruent melting point72(dystectic or indifferent point), but fails outside this region because theassumption of constancy of AHR fails. The Prigogine-Defay equation wasused by Bettinetti et al. (1990) and Pitre and Stradi (1990) to predict thebinary phase diagrams of sobrerol and dropropizine, respectively.A test of equation 11 is to plot in x(1-x) against 1/T, where T isdetermined experimentally. A linear plot should be obtained the slope ofwhich should give AHR after division by 2/R. For ibuprofen the plot waslinear (Fig. 17) and the slope gave a AHR value of 26.4 ± 1.8 kJ/mol whichagreed with the value of 26.9 ± 1.0 kJ/mol obtained directly from the areaunder the DSC melting endotherm of racemic ibuprofen.The phase diagram and X-ray analysis indicate that ibuprofen USP is aracemic compound and that the USP XXII (1990) monograph description ofibuprofen as a ‘(±) mixture’ is ambiguous.The mechanical properties of S-ibuprofen and racemic ibuprofen are,in fact, very similar (section 3.8.1). Hence the question whether or notS-ibuprofen should be used instead of racemic ibuprofen in tabletformulations would have to be based primarily on the differences in theirsolid state, in particular differences in their solubilities, and also onthe basis of previously reported pharmacokinetic differences (e.g., Jamaliet al., 1988; Ahn et al., 1991, Beck et al., 1991).73In x(1-x)-1.4--1.5--1.6--1.7--1.8-- I2.87 2.89 2.91 2.931/Tf (1E3)FIG. 17. Test of Prigogine Defay equation. The Tf values wereexperimentally determined from DSC. The slope affords the valueof AHf’.743.3. ANALYSIS OF POWDER COMPACTIONThe tableting process is divisible into a compression and adecompression phase. The overall (punch or die) stress during thecompression phase increases mainly due to a build up of elastic stresses inthe powder bed. These stresses arise as the particles are progressivelyconfined into a rapidly decreasing volume between the two punches and thedie wall. Any permanent deformation of the particles by flow, or byfracture of some kind, will relieve these stresses. At a certain stageduring the compression phase the stresses within the powder bed rise to amaximum corresponding to a punch pressure max The stresses then decreaseas the punches are separated during the decompression phase.The magnitude of the stress during tableting depends on the strainwithin the powder bed. This strain can be defined as a change in thedimension of any given particle under stress relative to its originaldimensions.The rate of strain during tableting is also important since it mayinfluence both the mode of deformation and the stress required to causedeformation. To understand the deformation of particles by flow duringcompression, changes in microscopic strain rates at the points ofinterparticulate contact should be studied. Mathematical modeling of themicroscopic strain rate changes is possible, but would be very complex on arotary press. Modeling these changes would include factors such asindentation hardness and elastic modulus derived from single crystalmicroindentation, initial particle shape and changes in shape duringcompression, the number of particles surrounding each individual particle(the coordination number) and changes in this number, and particle75reorientation due to angular momentum induced by unequal stresses on eachparticle (Duncan-Hewitt, 1988; Duncan-Hewitt and Weatherly, 1990a, 1990b).The changes in the microscopic areas of inter-particulate contact, changesin the porosity of the powder bed and a non-uniform distribution of thisporosity, and work hardening of particles if the deformation ispredominantly due to dislocation induced crystallographic slip willcomplicate the model further.For simplicity, strain on a powder bed can be approximated bycalculating strain as the decrease in the height of the powder bed relativeto its original height. The height of the powder bed at any instant duringcompression can be obtained from an accurate knowledge of the punchdisplacement. When data are collected at several small fractions of thetotal compression time, the strain between any two time intervals can becalculated as the fractional change in the height of the compact. Dividingby the average height between these two intervals (i.e., the instantaneousaverage powder bed height) gives the average strain on the powder bed.Using the instantaneous volume of the powder bed in the calculationsinstead of height gives the instantaneous volumetric strain.Similarly, the strain rate calculated from an accurate knowledge ofpunch displacement can be used as an approximation of the microscopicstrain rates. To include both the movement of the particles and changes inthe pore space, the calculated strain rate should be based on the volume ofthe powder bed and not just the height.It is shown in the following sections that the characterization ofdeformation behaviour of pharmaceutical materials during tableting can beaccomplished by using punch pressure and macroscopic volumetric strain rateas approximations of stress and strain rate respectively at the points of76interparticulate contact. Supporting information from the compressionphase such as the porosity of the compact (p), the yield value of thesolids (Uy), the work of compression (We), t0ff and the decrease in themax during t0ff (AP), as well as information from the decompression phasesuch as Young’s modulus (E) and the work of decompression (WD) can be usedin such a characterization.3.3.1. Machine deformation during powder compactionThe design of the Manesty Betapress is such that it deforms alongwith the powder particles during compaction. Both the upper and lowercompression rolls are mounted on bearings, and the lower beam carrying thelower compression roll is supported by a spring. Under pressure, thebearings allow an upward displacement of the upper roll and a downwarddisplacement of the lower roll. The spring supporting the lower cross-beamallows a downward deflection of the cross-beam which further displaces thelower compression roll in a downward direction. The punches contract underpressure. The machine deflections constitute up to 80% of the totalmachine deformation, while the remaining 20% is due to the contraction ofthe punches. The downward deflection of the lower roll constitutes about65% of the total machine deflections, the remaining 35% resulting due tothe upward deflection of the upper roll (Oates and Mitchell, 1989).The Betapress can therefore be considered as a large elastic body inwhich the punch contractions and roll deflections constitute the totalmachine deformation under load. Machine deformation has usually beenignored in the tableting literature. Oates and Mitchell (1989, 1990)established that the machine deformation is significant and cannot beoverlooked when analysing the overall punch displacement. The use of the77relationship between machine deformation and punch force in calculating thepunch displacement was demonstrated in section 2.3.2. The discussion ofthe compression and decompression phases in the following sections showshow this relationship can be further used to analyse the compaction cycleby obtaining only the force versus time data on a Manesty Betapress. Thisanalysis results in parameters related to particle deformation duringcompression, compact expansion during decompression, and energy consumptionduring compression, which are useful in interpreting the deformationbehaviour of solids under pressure.3.3.2. Particle deformation during powder compactionThe possible mechanisms by which powder particles can deform underpressure during compaction are described in section 1.3. Whether or not aparticle will deform by one of these mechanisms, and the relative amountsof the deformation by a given mechanism during compaction, will dependprimarily on the conditions of stress, strain and strain rate. Thepermanent deformation may occur by only one of the mechanisms 2-5 (section1.3), or by a mixture of these mechanisms depending on the conditions ofstress, strain and strain rate.The stress on the particles at which permanent deformation by flow isinitiated is the yield stress, Oy• If the particles continue to deform bya flow process, the stress required to sustain the flow after itsinitiation is the flow stress (Courtney, 1990).3.3.3. Stress, strain and strain rate during compactionThe particle deformation during compaction occurs at the points ofinterparticulate contact or in the regions in the immediate vicinity of78these contact points. The ‘stress’ during compaction is the stress at thepoints of the interparticulate contact, or the true stress (ST) as definedearlier. The strain is the amount of deformation of the particles underthis stress relative to their original dimensions, and the strain rate isthe rate of such deformation. The terms stress, strain, and strain rate,therefore, apply strictly to the deformation at the microscopic level, and,throughout this discussion, are used to describe events at the microscopiclevel.3.3.4. Influence of strain rate on particle deformationFor materials showing viscous/viscoplastic flow, the flow stressincreases directly with strain rate at the points of contact. Thus theamount of viscous/viscoplastic flow during compaction will vary directlywith the strain rate. The strain rate has a negligible influence if theparticles deform by plastic flow or if they undergo fracture since theseprocesses are not strain-rate dependent. Plastic flow and fracture arestrain-rate dependent only when loading is of the impact type, i.e., whenthe rate of loading is much more rapid than the speed of dislocationinduced crystallographic slip in the case of plastic flow, or of crackinitiation and propagation in the case of fracture. During tableting, thecompaction occurs over a period of time which, although very short, islonger than the instantaneous impact phenomenon. Hence, during compaction,plastic deformation and fracture can be considered to be independent ofstrain-rate.Duncan-Hewitt and Weatherly (1989a) indicated the importance ofstrain rate in understanding the deformation phenomenon in single crystalmicroindentation experiments. In an earlier study, Roberts and Rowe (1985)79reported the influence of the variation in punch speed to obtain varyingstrain rates (or rates of compaction) on tableting parameters ofpharmaceutical solids during powder compaction on a compaction simulator.3.3.5. Approximation of stress, strain and strain rate during compactionThere is an inhomogeneous distribution of the stress and strain indifferent parts of the powder bed during compaction. In reality, it isalmost impossible to determine the stress, strain and strain rate duringthe compaction of a powder bed. Nevertheless, an approximation of averagevalues of these parameters by using similar parameters determined at amacroscopic level would be useful in understanding the particle deformationbehaviour during powder compaction. The punch pressure, P, during thecompaction cycle (both the compression and decompression phases) can beused to approximate the stress. The punch pressure is the punch forcedivided by the cross-sectional area of the punch face, where the punchforce is recorded by strain gauges affixed to some remote part of thepress. The decrease in powder bed height during the compression phaserelative to instantaneous bed height, or an increase in the compact heightduring the decompression phase relative to the instantaneous compactheight, can be used to approximate the strain. Hence, the rate ofcompaction calculated from an accurate knowledge of punch movement duringthe compression phase can be used as an approximation of the microscopicstrain rates during this phase. To include both the movement of theparticles and changes in the pore space, the calculated rate of compactionshould be based on the volume of the powder bed and not just the height.The above approximations will be in error during initial part of the80compression phase when the reduction in the powder bed volume isrearrangement dominated.A discussion of the compression and decompression phases, and adescription of the various parameters obtained from their analysis, isgiven in the following sections. These parameters, and the above mentionedgeneral concepts relating deformation, stress and strain rates, were usedto ascertain the deformation behaviour of particles during compaction. Adiscussion of the deformation behaviour of a number of general categoriesof solids is presented at the end of the discussion.3.4. PARAMETERS FROM THE COMPRESSION PHASEThe compression phase of the compaction cycle begins when the straingauges first record an increase in pressure above a certain baseline value.This phase ends when the punches are vertically aligned with the centres ofthe compression rolls (Fig. 5, page 43).3.4.1. A general view of the events during the compression phaseAt the beginning of the compression phase, the areas of contactbetween the particles in the powder bed are usually microscopic in size.The stress at these ‘points’ of contact, i.e., the true stress (CT), willbe much higher than the punch pressure (P). Even if P is relatively small,UT may be high enough to overcome Uy and the particles will yield anddeform by some type of flow. After yielding, particle deformation by flowwill continue if T is higher than the flow stress.The flow will cause the area of interparticulate contact to graduallyincrease. Hence P gradually increases because the structure formed due toincreased interparticulate contact increasingly resists the punch movement.81The particulate material in the die is able to influence the punch movementto a significant degree because, as mentioned above, the machine itselfbehaves as an elastic body. Thus, punch movement is defined partially bythe machine geometry, and partially by the nature of the material in thedie. The increase in P tends to increase T’ while the increase in thearea of interparticulate contact tends to decrease Thus, the actualvalue of T is governed by the relative rates of increase in the value of Pand the degree of interparticulate contact. The o will nevertheless behigher than P throughout the compression phase because the compactsinvariably have a certain degree of pore space. The values of UT and Pwill converge to equality as the area of interparticulate contact becomesvery large. There will be an unequal distribution of T within thecompacts due to differences in the nature of interparticulate contacts andthe density gradients within the compact.After a certain degree of flow the particles will fracture if andwhen a T value equal to fracture stress is achieved. Fracture stress isthe stress required to cause the initiation and propagation of a crack inthe particles. For materials having high Jy, which, at room temperature,is normally true for materials with high melting temperature, fracture mayoccur even before the particles have yielded. This is because, in theprocess of building up the UT to a value equal to Cy, the particles maydevelop regions of very high stress concentrations, e.g., at microscopicflaws in their structure, or develop triaxial stresses on the particles.Under such conditions it will be energetically preferable for the particlesto fracture without any prior permanent deformation. Therefore, solidswith high Uy may behave as intrinsically brittle materials at roomtemperature under normal conditions of powder compaction.823.4.2. The rate of compaction profilesOn a rotary press, the rate of compaction is the rate at which theupper and lower punches are brought together. Between any two timefractions during compression the rate of compaction can be calculated bydividing the decrease in the powder bed height or volume by the timeinterval between the two fractions. This rate is expected to be highest atthe beginning of the compression phase. It will decrease as the stressesin the powder bed offer increasing resistance to the punch movement, andwill be negligible when the flat top of the punch heads comes in contactwith the compression rolls.As the compression phase progresses, the initially high rate ofcompaction, and hence the high flow stress (assuming that the rate ofcompaction is an approximation of the strain rate), will decrease as thepunches slow down due to an increase in the resistance offered by thecompact. If Uy has been exceeded and if T is higher than the flow stress,flow will continue as long as enough voids are available to accommodate theflow. As the umax approaches, the rate of compaction and the flow stresseswill reach a minimum, and the particles will flow with even greater ease,provided sufficient pores are still available. The flow, and theconsequent stress relief, will continue during the dwell time, if any. Ifthe particles fracture, the fracture will continue to occur if T is higherthan the fracture stress, and the change in the rate of compaction wouldhave a negligible effect on this process.The volumetric rate of compaction profiles during the compressionphase for various materials, when compressed to the same max’ are shown inFig. 18. Although the initial magnitude of the rate of compaction isdifferent for different materials, the shape of the plots is similar. This830-50-100-150Vol. rate of compaction (cm3/s)Line ofDeadCentreFIG. 18. Volumetric rate of compaction profiles during thecompression phase. Each profile shows an initial acceleratoryphase, an intermediate deceleratory phase and the final phase ofconstant powder bed volume where punch head flats are in contactwith the compression rolls of the press. Each material wascompressed to nearly the same max using IPT type punches at aturret time of 1 s.-35 -30 -25 -20 -15 -10 -5 0Compression time (ms)84suggests that while the magnitude of the initial rate of compaction isgoverned by the material, the pattern of change in the rate of compactionis a machine characteristic.The rate of compaction profiles of all materials show an initialacceleratory phase, a deceleratory phase and a phase of zero rate ofcompaction. The latter phase can be described as the dwell phase of thecompression time during which the flat heads of the punches are in contactwith the compression rolls and the distance between the punch faces isconstant. According to geometric calculations, and at a turret speed of 1revolution per second, the IPT punch head flats should have a dwell time of6 ms. The duration of 7 ms of the constant strain phase agrees with thistime.The initial differences in the rate of compaction can be related todifferences in the bulk density of the different materials. Those withsmaller bulk density (e.g., Avicel PH1O2, O.3 g/cm3) have a much higherinitial rate of compaction than materials with higher bulk density (e.g.,Emcompress, O.8 g/cm3). This follows from the fact that the rate ofcompaction is another way of expressing the punch velocity. At thebeginning of the compression phase, materials with a low bulk density offerless resistance to the punches, hence the initial punch velocity or therate of compaction is much higher than when a material with a higher bulkdensity is compressed. The lowest initial rates of compaction occur when aprefabricated steel tablet, with a thin cushioning layer of Emcompress onits top, is compressed. Since the steel+Emcompress tablet is an almostnon-porous body, the punches decelerate almost instantaneously, and therates of compaction decrease very rapidly to a nearly negligible level.85For all materials the initial rate of compaction increases as thepunches accelerate into the powder bed, which initially offers littleresistance to the punch movement. As the powder bed begins to resist thepunch movement more firmly, a few milliseconds into the compression time,this acceleratory phase ends and the rate of compaction decreases. Theduration of the acceleratory phase appears to be characteristic of thematerial compressed (Fig. 19). Materials with highest initial rate ofcompaction, e.g. Avicel PH1O2, have the longest acceleratory phase. Thisagain reflects the differences between the bulk densities of the variousmaterials. A material with a low bulk density allows little initialresistance and takes longer to become sufficiently dense to decelerate thepunch movement than a material with a higher bulk density (Fig. 19).As mentioned in section 3.3.5, the rate of compaction during thecompression phase can be used as an approximation of strain rate at themicroscopic points of contact. Whether or not this approximation is validduring the initial acceleratory phase of the rate of compaction change isquestionable. During this initial phase, when the particles undergopredominantly a rearrangement phenomenon and when they just begin to bedeformed at their contacts, the strain at these contacts is small relativeto the size of the individual particles. On these grounds, the strain rateis expected to be intrinsically low at the start of compression. The rateof compaction, on the other hand, is high because of the low powder beddensity and due to the acceleration of the punches from their initialresting positions. The initial part of the compression phase is thereforea very dynamic stage of compaction, and it will be difficult to stateclearly whether the strain rate at the particulate level is high, or, if infact, it is low while the rate of compaction is high. This dilemma is86Duration of Accel. Phase (ms/cm3)45O AvIcel PHIO240 - Emcompre8ao Spray-drIed Lactose35 -0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 19. Change in duration of the acceleratory phase ofvolumetric rate of compaction profiles with max for threedirect compression excipients. The duration of the acceleratoryphase has been normalised for true volume of the three materialsat each max•U87further emphasised by the rate of compaction profiles of the‘steel+Emcompress’ tablets. Since a ‘steel+Emcompress’ tablet is an almostnon-porous body, the rate of compaction is very low. But, due to the samenon-porous nature, there must actually be a very high rate of elasticstraining of this tablet at the start of the compression phase.The above arguments indicate that it may not be correct toapproximate the strain rate in a powder bed during the initial part of thecompression phase by using the rate of compaction. The rate of compactionmay be a valid approximation once the initial rearrangement dominated phaseof compaction is over, and once the powder bed has entered into adeformation dominated phase. The latter phase would probably start at theend of the initial acceleratory phase of the rate of compaction change.The profiles of rate of compaction with respect to time during thecompression phase, when superimposed on the profiles of P (Fig. 20a-20d),are helpful in a mechanistic interpretation of events during thecompression phase. The initial parts of the P versus t profiles inFig. 20a-20d show an inflection point between approximately 7 and 14 MPa.This point occurs during the acceleratory phase of punch movement, and therates of increase in P before and after this point are evidently different.The rate of increase in P for Avicel PH1O2 is higher before this point dueto a sudden build up of T mainly in response to friction at the points ofinterparticulate contact during the initial particle rearrangementphenomenon. The friction may cause a certain degree of bonding, and somedeformation and, possibly even fracture, of the particles at theircontacts. These phenomena would make the powder bed somewhat resistant tocollapsing in the initial part of the compression phase. After a certainlevel of T is built up, which will depend on a complex interplay of the-50-100-150Vol. Rate of Compaction (cm3ls) Pressure (MPa)352821147088FIG. 20a. Volumetric rate of compaction profiles superimposed onthe corresponding stress profiles during the compression phasefor Avicel PH1O2. Turret time = 1 s, IPT punches.0-35 -30 -25 -20 -15 -10 -5 0Compression time (ms)890-50-100Vol. rate of compaction (cm3/s) Pressure (MPa)352821147-150-35 -30 -25 -20 -15Compression Time (ms)0FIG. 20b. Volumetric rate of compaction profiles superimposedon the corresponding stress profiles during the compressionphase for Fast-Flo Lactose. Turret time = 1 s, IPT punches.A Pressure7 Strain RateV/-10 -5 0900-50-100-150Vol. Rate of Compaction (cm3/s) Pressure (MPa)3528211470FIG. 20c. Volumetric rate of compaction profiles superimposedon the corresponding stress profiles during the compressionphase for Emcompress. Turret time = 1 s, IPT punches.-35 -30 -25 -20 -15 -10 -5 0Compression Time (ms)91Vol. Rate of Compaction (cm3/s) Pressure (MPa)* 420* PressureE Strain Rate-50 28-100 / 217I I14-1500-35 -30 -25 -20 -15 -10 -5 0Compression Time (ms)FIG. 20d. Volumetric rate of compaction profiles superimposedon the corresponding stress profiles during the compressionphase for acetylsalicylic acid. Turret time = 1 s, IPT punches.92rearrangement and friction related phenomena, a failure of the initiallyresilient structure of the powder bed takes place. This appears as theinflection point, after which, the rates of compaction enter thedeceleratory phase, and the increase in stress is more gradual because theparticles continue to deform by flow during this phase.Thus, the inflection point can be considered to be a transition pointbetween a phase of rearrangement dominated compaction and a phase ofparticle deformation dominated compaction. Once the deformation dominatedphase has begun, the punch pressure and the rate of compaction can beassumed to be estimates respectively of the stress and strain rate at theparticulate level.Fast-Flo Lactose has apparently a lower ability to flow than Aviceland therefore has a smaller difference between the rates of increase instress before and after the inflection point (Fig. 20b). The P versus tprofile of Emcompress (Fig. 20c) shows no change in the rates of increasein P before and after this point which indicates that Emcompress particlesdo not deform by a flow process during the deceleratory phase.Acetylsalicylic acid has the opposite pattern of increase in P (Fig. 20d)to Avicel. The rate of increase in P is slower before the inflectionpoint, because acetylsalicylic acid is a very easily deformed substance andits particles fill up the voids at low pressures. As the inflection pointis approached the powder bed becomes very dense (i.e., the porosity becomesvery low), and the compact becomes resistant to the punch movement, hence Pincreases more rapidly.933.4.3. Porosity-stress relationshipA material with a high propensity to fracture will have negligibledeformational flow, thus maintaining a large porosity throughout theacceleratory, the deceleratory and the constant rate phases of the rate ofcompaction profile. On the other hand a material with a certain degree offlow becomes compacted to relatively lower porosity and reaches constantporosity only during the latter phase of the rate of compaction profilewhen the powder bed volume (or strain) is almost constant.Even among those materials that show particle flow, one would expecta varying degree of change in compact porosity depending upon the actualtype of flow. It should therefore be possible to ascertain the mechanismof deformation of the particles of various materials during compressionwith a reasonable degree of confidence by using some type of porosity-stress relationship. Several equations correlating the powder bed volume,and hence porosity, with applied stress are available (MacLeod, 1983). Acommonly employed relationship of this type is given by the Heckel plots ofln(1/p) against overall stress during compression (Heckel, 1961). Based onchanges in Heckel plots due to changes in particle size, Hersey and Rees(1971) classified solids into two categories: Type 1 materials, whichdeform mainly by plastic deformation, and Type 2 materials which deformmainly by fragmentation. These plots alone have been used to characterizethe compaction mechanisms on a single punch press (Duberg and Nystrom,1986). Yield values are estimated from the slope of the Heckel plots, butdifficulties arise where the plots are not linear (Jones, 1978; Marshall,1989). York (1979) reported differences in the estimated Uy values whenHeckel plots were constructed using different methods under differentexperimental conditions. An alternative method of obtaining a3, was used in94this work. The values of ln(1/p) at max (1max) were plotted against max•Values of Uy were obtained from the slopes of these plots which, as shownin Fig. 21 for a representative selection of materials, are linear. Thevalues of cry of various materials are listed in Table III. The differentmaterials shown on this plot demonstrate the wide range of changes inporosity among the various pharmaceutical materials. Materials such as thecalcium phosphates show little change in Hmax, whereas drugs such asibuprofen and acetylsalicylic acid show a much larger change over a similarrange of max The Uy values listed in Table III reflect these differencesand indicate the ease with which a given material is compacted to a givenporosity. These cry values are useful in comparing the different materialson a relative scale, and in ascertaining the deformation mechanisms ofthese solids (section 3.8).3.4.4. Position of the peak punch pressure: Peak offset timeIf the powder bed were replaced by a nearly nonporous, and hence analmost incompressible substance, e.g. the steel tablet, the stress recordedvia the strain gauges will correspond mainly to the machine deformation,and, to a very small extent, to the deformation of the steel tablet. Sincethe machine deformation is completely recoverable, the compression anddecompression phases will be mirror images of each other, and the stressprofiles will be symmetrical about the position of vertical alignment ofthe punches with the centres of the compression rolls. The peak punchpressure, max’ for an incompressible solid will therefore occur at thedead centre position.When a porous powder bed is compressed, the max does not necessarilycoincide with this dead centre position. The position of max relative to95H 1max7-6-5-34-43-2- 61-0- I0 100 150 250Upper Punch Peak Pressure (MPa)FIG. 21. Plots of the maximum value of Heckel term (Hmax) atmax during the compression phase against the corresponding maxfor different materials. Each data point corresponds to onetablet. Slopes obtained by linear regression of these plotsgive the yield strength (Uy) of each material.(1) Acetylsalicylic acid, (2) S-Ibuprofen, (3) AcetaminophenGranular, (4) Avicel PH1O2, (5) a-Lactose monohydrate,(6) Emcompress, (7) Tn-Tab.z2550 20096TABLE III. The Uy values determined from Heckel Plots of the varioussolids.Material c:7y1(MPa)A-Tab 165 (23)Acetaminophen Fine Powder 40.3 (4.9)Acetaminophen Granular 33.1 (3.6)Acetaminophen Powder 39.3 (3.9)Acetylsalicylic Acid 11.5 (4.1)Anhydrous Emcompress 162 (23)Anhydrous Lactose 58.7 (3.6Asagran 13.8 (3.2)Avicel Large 30.5 (3.3)Avicel PH1O1 25.8 (1.9).Avicel PH1O2 26.7 (2.8)Avicel PH1O5 28.2 (3.0)Caffeine 44.3 (4.6)Cal-Star 89.3 (16)Cellactose 43.7 (3.8)Compap CG 36.3 (3.0)Compap Coarse 73 L 36.0 (2.8)Compap Coarse L 29.6 (2.0)Compap L 38.3 (3.5)DCI-63 12.5 (2.5)Di-Pac 49.3 (2.5)Di-Tab 92.3 (15)Elcema G250 32.2 (1.7)Emcocel 25.4 (1.6)Emcompress 95.0 (16)Emdex 34.7 (2.6)Fast-Flo Lactose 49.0 (3.6)Lactose DCL 21 55.0 (4.2)Mannitol (Crystalline) 54.7 (3.8)Mannitol M.G. 45.3 (4.2)Neosorb (Sorbitol) 38.0 (3.2)a-Lactose monohydrate 69.0 (5.1)Potassium Chloride 20.0 (4.1)Racemic ibuprofen 11.2 (3.2)Rhodapap DC-P3 42.7 (3.0)S-Ibuprofen 16.2 (5.9)Sodium Chloride 35.3 (3.9)Spray-dried Lactose 53.7 (4.9)STA-Rx-1500 18.1 (1.6)Sucrose (crystalline) 55.3 (16)Sugartab 35.7 (3.2)Tn-Tab 221 (35)Xylitol (Crystalline) 52.0 (7.5)1. Values (standard error): determined from the slope of Hmaxversus max plots.97the dead centre position will depend on the rate at which the materialdeforms in response to the rate of compaction under a given set of machineconditions which are defined by the punch type and machine speed. Hence,any change in the material which affects its ability to deform will bereflected in the position of max Both too much or too little flow willbring the position of max closer to the dead centre position. This isbecause a material which readily flows becomes compacted to very lowporosities earlier on in the compression phase and will behave almost as anincompressible solid when the umax is reached, while a material which doesnot deform by flow will have no strain rate dependency. Both of theseconditions will cause the max to occur relatively close to the dead centreposition. For materials with a degree of flow somewhere in between thesetwo extremes the max will precede the dead centre position of the punches.Some representative pressure-time profiles for Avicel PH1O2, Spray-dried Lactose, and Emcompress are shown in Fig. 5 (page 43). All threematerials show an offset of max with respect to the vertical alignment ofthe punches with the pressure rolls at t = 0. The interval between theposition where max occurs and the position where the dead centre occurs isdefined as the peak offset time, t0f.f. The duration of t0ff is differentfor each material. The net punch displacement profile (Fig. 6, page 46)shows that the displacement and therefore the distance between the upperand lower punch faces is approximately constant during t0ff. Assuming thatthe distance between punch faces, especially in the region where maxoccurs, is an indication of strain in the particles, and that the punchpressure is an indication of stress at the particles, it can be seen thatthe stress decreases during t0ff while the strain remains essentially98constant. Hence t0ff is indicative of stress relaxation since the stressis decaying at constant strain albeit for a very short time.Factors that can affect the degree of deformational flow duringtableting are likely to affect the magnitude of t0ff. At least three suchmachine-related factors can be readily identified, namely, max’ turrettime and punch type. A fourth factor is the addition of one or morecomponents which impart an increased degree of flow to the formulation.Each of these factors is discussed below.3.4.4a. Effect of nax Ofl t0ffFigure 22 shows the changes in t0ff for Avicel PH1O2, Spray-driedLactose and Emcornpress with changes in the umax using IPT punches at aturret time of 1 s. Each of these excipients has excellent tabletingproperties, and the differences in t0ff reflect differences in theirdeformation mechanisms. Table IV gives the range of t0ff values forseveral solids when these solids were compressed over a range of maxAvicel PH1O2 has the longest t0ff with values up to about 8 ms at thelowest niax studied, suggesting that stress relief during the compressionphase is achieved by some type of flow into the voids of the tablet.Emconipress has much shorter t0ff values than Avicel PH1O2 and Spray-driedLactose, suggesting that it does not flow during compression. Spray-driedLactose is manufactured from a suspension of lactose crystals and consistsof aggregates of amorphous lactose with loose or embedded lactose crystals(Kussendrager et al., 1981). Amorphous lactose deforms by flow whilecrystalline lactose undergoes particle fracture (Morita et al., 1984;Vromans et al., 1986). Thus, Spray-dried Lactose exhibits deformation99Peak Offset Time (ms)10-AvIcel P1-1102G Spray-dried Lactose8- 0 Emcompress6-4b..0 G -----.5--0I0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 22. Variation in peak offset times with peak pressure forthree direct compression excipients. Turret time = is, IPTpunches.100TABLE IV. Range of t0ff, and AP/VTmaxvalues* corresponding to a range ofMaterial max t0ff AP/VT.(MPa) (ms) (MPa/cm)A-Tab 28.3-214 3.2-0.9 6.0-1.1Acetaminophen Fine Powder 26.9-213 3.1-0.4 3.6-0.23Acetaminophen Granular 18.7-224 2.6-0 5.0-0Acetaminophen Powder 17.8-220 3.5-0.6 2.7-2.1Acetylsalicylic Acid 16.6-211 3.0-0.1 3.0-0.88Anhydrous Emcompress 28.3-213 3.0-1.0 6.4-0.31Anhydrous Lactose 33.5-215 3.0-1.5 6.8-0Asagran 31.5-212 1.5-0 1.0-0Avicel Large 20.9-225 7.2-0.4 12 -0.51Avicel PH1O1 18.7-211 7.7-0.8 12 -0.52Avicel PH1O2 18.9-226 8.2-0.6 20 -0.11Avicel PH1O5 16.5-228 7.1-0.1 12 -0.47Caffeine 17.8-203 7.1-0.7 8.5-0.46Cal-Star 25.3-215 4.3-0.3 5.5-0Cellactose 19.5-227 6.9-1.0 4.5-1.0Compap CG 23.4-210 6.9-0.5 6.5-0.26Compap Coarse 73 L 22.8-214 5.0-1.0 6.3-0.47Compap Coarse L 26.9-211 3.7-0.7 4.3-0.27Compap L 23.1-216 4.3-0.8 5.0-0.44DCI-63 27.9-212 4.0-0.1 3.5-0Di-Pac 39.1-222 5.0-1.3 9.6-1.8Di-Tab 27.6-215 3.1-0.3 3.3-0.09Elcema G250 29.4-212 6.4-0.9 10 - 0.79Emcocel 37.8-215 6.4-1.2 12 - 1.1Emcompress 27.6-214 6.4-0.4 8.0-0.21Emdex 15.2-221 6.4-1.1 12 - 0.41Fast-Flo Lactose 32.1-221 4.5-0.4 7.1-0Lactose DCL 21 26.9-210 4.5-0.8 7.2-0Mannitol (Crystalline) 39.3-218 4.5-1.5 9.9-2.6Mannitol M.G. 24.5-218 6.0-0.8 8.1-0.71Neosorb (Sorbitol) 32.8-223 6.0-1.2 17 -1.7a-Lactose monohydrate 43.9-214 3.4-1.3 6.9-1.4Potassium Chloride 38.5-208 3.4-0.2 6.2-0Racemic Ibuprofen 29.2-224 2.0-0 3.5-0Rhodapap DC-P3 32.1-219 3.0-0.9 5.2-0.12S-Ibuprofen 28.3-214 3.0-0.0 1.4-0Sodium Chloride 39.7-215 3.0-0.6 5.4-0.37Spray-dried lactose 26.8-223 3.0-1.1 8.2-1.4STA-Rx-1500 38.8-221 4.8-0.3 11 -0.12Sucrose (Crystalline) 27.7-219 4.8-1.3 5.2-0.68Sugartab 25.8-211 6.5-1.2 11 -1.8Tn-Tab 19.9-214 4.4-0.8 6.2-0.61Xylitol (Crystalline) 18.8-220 5.9-0.3 7.8-0.1* IPT Punches; turret time is101characteristics that are intermediate between those of Avicel PH1O2 andEmcompress, with t0ff of up to about 5 ms at the lowest max studied.The ability of particles to relieve stress within the confines of adie cavity will become increasingly restricted as the porosity of a powderbed decreases with increase in the punch pressure. This is illustrated bythe results in Fig. 22 which, for each material, show that the duration oft0ff decreases asymptotically towards a lower limit as umax is increased.The umax on a rotary tablet press can be increased either byincreasing the mass of the material in the die at a fixed thickness settingon the press (Method I, page 41), or by changing the thickness setting toreduce tablet thickness keeping the mass constant (Method II, page 41).Figure 23 shows that the duration of t0ff is independent of the method ofincreasing max It was easier to change the mass of the material in thedie at a fixed thickness setting, and this procedure for changing max wasused in all further experiments.3.4.4b. Effect of machine speed on t0ffThe t0ff for Avicel PH1O2 at different turret times are shown inFig. 24a. Decreasing the turret time reduces t0ff showing that at fastermachine speeds, i.e., at faster rates of compaction or faster strain rates,Avicel PH1O2 has less time to relieve stress by deforming by flow than atslower speeds. Situations can sometimes occur where a formulation producesgood tablets on slow machines but fails when transferred to higher speedmachines. One reason may be a decrease in the extent of deformation of thematerial by flow as indicated by a decrease in t0ff at faster machinespeeds. For Emcompress, t0ff is independent of turret time (Fig. 24b)indicating that the rate of stress relief for Emcompress does not depend on102Peak Offset Time (ms)10U Avicel PH1O2‘a 0 Emcompress86U.”,4-r3U00 . “2- 0’.00 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 23. Variation in peak offset time with peak pressure.Turret time = is, IPT punches. Open symbols, peak pressurevaried by changing the mass at fixed thickness setting; closedsymbols, peak pressure varied by changing the thickness settingkeeping the mass constant.103Peak Offset Time (ms)6Turret time - 1.00 85-0 Turret time - 0.75 s0 Turret time • 0.65 84.302-1-00---0—1- I I I0 50 100 150 200Uppper Punch Peak Pressure (MPa)FIG. 24a. Effect of turret time on peak offset times for AvicelPH1O2. Manesty punches.Turret time - 1.00 8Turret time - 0.88 8o Turret time • 0.75 8O Turret time • 0.65 850 100 150Upper Punch Peak Pressure (MPa)FIG. 24b.Emcompress.Effect of turret time on peak offset times forManesty punches.Peak Offset Time (ms)10465.4.3.2-1-0cO--o ,• 0—10I-J 6oI200105the rate of at which it is compressed. In other words, Emcompress probablyhas a deformation mechanism which is independent of the rate of compaction.3.4.4c. Effect of punch type on t0ffThe IPT punches have a flatter punch head profile than the Manestypunches (Fig. 1, page 28) thereby providing a longer dwell time for stressrelaxation. Figure 25 shows the t0ff for Avicel PH1O2 and Emcompress whencompressed using the two punch types. At max below 110 MPa, the t0ff forAvicel PH1O2 becomes progressively longer when compressed using IPT punchesthan with Manesty punches. For Emcompress, differences in t0ff due topunch type were not significant. These observations suggest that thetableting behaviour of materials deforming by a strain rate dependentprocess is more likely to be affected than that of materials with non-strain rate dependent deformation when the tooling is changed from IPT toManesty type. Above about 110 MPa, equivalent to about 1.5 tons, there islittle difference in t0ff for Avicel PH1O2 between the two punch typessuggesting that differences in dwell time are less significant than thedecrease in porosity above this pressure.3.4.4d. Effect of formulation variables on t0ffDrugs such as acetaminophen and ibuprofen which cap or laminateduring or after decompression are modified commercially by formulating themwith polymeric substances which, at a given temperature, are more likelydeform by flow under stress than crystalline materials. An examination ofRhodapap DC-P3 (97% acetaminophen + 3% polyvinylpyrrolidone, PVP) under ascanning electron microscope indicated that it is made by spray drying aslurry of acetaminophen crystals in a solution of the polymer. Compap CG106Peak Offset Time (ms)8LI IPT Punches0 Manesty Punches‘ci6U4-“-p‘.0N2-N.•‘- —-•.•0- I . -•.----- - F1_0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 25. Effect of punch type on peak offset times. Turret time= is. Open symbols, Avicel PH1O2; closed symbols, Emcornpress.107(90% acetaminophen + 10% excipients) appears to be made by a similarprocess.Figure 26a shows the relationship between t0ff and max for threesamples of crystalline acetaminophen. The t0ff values for the threesamples are highly variable and are indistinguishable from one anotherdespite major differences in their particle size distributions. The shortt0ff values indicate that crystalline acetaminophen undergoes only a smallamount of deformational flow during compression.It is evident from Fig. 26b that the addition of a small amount ofPVP as in Rhodapap DC-P3 increases flow sufficiently to prolong t0ff byabout 2 ms for max up to about 90 MPa. Rhodapap DC-P3 forms strongtablets below this pressure but shows lamination at higher pressures whereits t0ff approaches that of crystalline acetaminophen. Compap CG containsa greater amount of excipients and forms intact tablets up to max of about210 MPa. These excipients presumably deform by flow, and t0ff isincreased. The tablet strengths of Compap CG are comparable to those ofthe tablets of Rhodapap DC-P3 (Fig. 27).Similar results were obtained for ibuprofen (Fig. 28). The directlycompressible ibuprofen, DCI-63, has longer t0ff relative to crystallineibuprofen at max up to about 90 MPa. There are no differences in t0ffabove this pressure. Like the direct compression forms of acetaminophen,DCI-63 forms tablets only below about 90 MPa, whereas above this pressurelamination occurs. Crystalline ibuprofen, on the other hand, does not formtablets at any pressure.The increase in t0ff for the direct compression formulations ofacetaminophen and ibuprofen suggests that these formulations permit adegree of deformation by flow which leads to interparticulate bonding.108Peak Offset Time (ms)4AcetaminophenGranular30 PowderFine PowderLi2 LiLILILI LI-QLI Li•LILI0- I I0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 26a. Variation in peak offset times with peak pressure forvarious particle sizes of acetaminophen. Turret time = is, IPTpunches.109Peak Offset Time (ms)6-0 Compap OGRhodapap DC-P30” Acet. Granular4-2-: ;--___•_z0- I I0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 26b. Variation in peak offset times with peak pressure foracetaminophen and its selected formulations. Turret time = is,IPT punches.110Force of Failure (N)100 150Upper Punch Peak Pressure (MPa)FIG. 27. Variation in force of failure with peak pressure forcrystalline and direct compression forms of acetaminophen.Turret time = is, IPT punches.250 -200 -150-100-50 -0-0 50Rhodapap DC-P3o Compap CGAcet. Granular200 250111Peak Offset Time (ms)4El DCI-630 Racemic IbuprofenEl’.‘El2-El ElEl_1- 0 El jDD0 -00 0i:0 00- I 10 D0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 28. Comparison of peak offset times for crystallineibuprofen and a direct compression formulation. Turret time= is, IPT punches.112Thus, it appears that the strength of tablets is, at least in part,dependent on the degree of deformational flow under pressure.3.4.5. Decrease in punch pressure during the phase of constant strainWhen the rate of compaction achieves a minimum value towards the endof the compression phase, then the powder bed is under a state of almostconstant strain. Since the rate of compaction is close to zero, the flowstresses achieve a minimum value and T will be readily relieved byparticle flow into the pores, if any pores still exist at this stage. Thisis when t0ff also occurs and, as mentioned earlier, any pressure drop (AP)during t0ff due to the relief of T comprises the phenomenon of stressrelaxation at constant strain. The values of AP normalized for true volume(tkP/V--) for some materials when compressed over a range of max are givenin Fig. 29. The differences in the slopes of these plots indicate thedifferent rate of stress relief for each material, and, also, to a certainextent, suggest the mechanism of deformation. A material showing a smallchange in AP/VT over the range of max used (e.g. Emcompress) has amechanism of deformation with little strain-rate dependency (i.e., fractureor plastic flow). On the other hand, a material with large change in AP/VTover a similar pressure range (e.g. Avicel PH1O2) has a mechanism ofdeformation which is strain-rate dependent (i.e., viscous or viscoplasticflow).The range of AP/VT corresponding to the range of max for the solidscompressed on the Betapress is given in Table IV. The AP/VT is a measureof the ability of various materials to deform by a viscous/viscoplasticflow mechanism. It appears therefore that t0ff and AP/VT can be used toascertain the deformation mechanisms during compression. These parametersFIG. 29. Decrease in punch stress during t0ff normalised fortrue volume (AP/VT) of different materials as a function oft0ff. The difference in the slopes of these plots indicate adifference in the rate of stress relief during t0ff of thedifferent materials.113*AP/VT (MPa/cm3)0-5-10-15-20* Acet. Granular00Compap Coarse 73LFast-Plo Lactose0 EmcocelNeosorb0 1 2 3 4 5 6Peak Offset Time (ms)7114were used in conjunction with other parameters from the compression anddecompression phases to analyse the deformation behaviour of various solidsduring compaction on the Betapress (section 3.8).3.4.6. Work of compressionA large amount of energy is imparted to the powder bed by the pressduring the compression phase. This energy can be expressed as the work ofcompression, WC. The range of values of W corresponding to a range ofmax during the compression phase is given in Table V. These values arecorrected for the work of machine deformation as described by Oates andMitchell (1990) and are norrnaltsed for tablet mass to account fordifferences in the tablet masses between the different materials. The workof compression is a useful parameter in analysing the energetics ofparticle rearrangement and deformation during compaction. Along with thework of decompression (section 3.5.1), it can be used in understanding theparticle deformation behaviour during powder compaction (section 3.6).3.5. PARAMETERS FROM THE DECOMPRESSION PHASEThe decompression phase begins at the end of the constant rate ofcompaction part of the compression phase when the punches are at the deadcentre position. For all powders, the punch pressure decreases morerapidly during decompression than the rate at which it increases duringcompression. This suggests that only a part of the energy expended duringcompression is recovered during decompression. The recovered energycorresponds to the recovery of the machine and also to the recovery (axialexpansion) of the compact formed during compression. The pressure-time115TABLE V. Range of WC and WD values* corresponding to a range of maxMaterial max W WD(MPa) (J/g) (J/g)A-Tab 28.3-214 3.80-22.2 0.11-2.4Acetaminophen Fine Powder 26.9-213 2.60-19.4 0.09-2.5Acetaminophen Granular 18.7-224 1.50-14.4 0.10-2.1Acetaminophen Powder 17.8-220 1.20-17.1 0.12-2.2Acetylsalicylic Acid 16.6-211 1.80-8.00 0.06-2.2Anhydrous Emcompress 28.3-213 3.90-22.2 0.09-2.4Anhydrous Lactose 33.5-215 3.10-23.0 0.10-2.9Asagran 31.5-212 3.50-12.5 0.16-4.2Avicel Large 20.9-225 9.40-44.8 0.19-4.5Avicel PH1O1 18.7-211 7.70-40.8 0.12-3.7Avicel PH1O2 18.9-226 3.30-30.9 0.07-3.4Avicel PH1O5 16.5-228 5.60-36.8 0.12-4.8Caffeine 17.8-203 2.10-21.1 0.15-1.5Cal-Star 25.3-215 3.37-15.7 0.05-2.8Cellactose 19.5-227 5.80-36.0 0.10-4.4Compap CG 23.4-210 6.00-31.5 0.11-2.6Compap Coarse 73 L 22.8-214 3.70-25.4 0.08-2.9Compap Coarse L 26.9-211 5.20-26.1 0.12-3.2Compap L 23.1-216 6.60-31.9 0.12-3.4DCI-63 27.9-212 5.90-19.6 0.11-3.0Di-Pac 39.1-222 5.80-23.5 0.08-2.8Di-Tab 27.6-215 3.30-15.2 0.04-2.1Elcema G250 29.4-212 5.40-26.6 0.15-2.8Emcocel 37.8-215 14.3- 41.6 0.32-3.1Emcompress 27.6-214 3.40-15.0 0.08-1.5Emdex 15.2-221 2.20-29.3 0.06-2.0Fast-Flo Lactose 32.1-221 5.30-27.0 0.18-3.9Lactose DCL 21 26.9-210 1.90-21.3 0.09-3.0Mannitol (Crystalline) 39.3-218 5.70-25.8 0.14-2.0Mannitol M.G. 24.5-218 3.44-21.2 0.08-4.7Neosorb (Sorbitol) 32.8-223 5.88-25.5 0.16-3.6a-Lactose moriohydrate 43.9-214 3.70-18.4 0.15-2.2Potassium Chloride 38.5-208 5.20-15.5 0.07-0.9Racemic Ibuprofen 29.2-224 4.70-15.8 0.11-3.3Rhodapap DC-P3 32.1-219 7.70-31.3 0.15-4.0S-Ibuprofen 28.3-214 2.80-11.5 0.06-3.7Sodium Chloride 39.7-215 3.40-15.4 0.05-1.3Spray-dried lactose 26.8-223 4.10-23.2 0.07-2.6STA-Rx-1500 38.8-221 3.80-24.6 0.28-3.4Sucrose (Crystalline) 27.7-219 3.30-16.9 0.09-1.7Sugartab 25.8-211 3.40-18.7 0.04-1.4Tn-Tab 19.9-214 1.70-15.4 0.07-3.0Xylitol (Crystalline) 18.8-220 2.07-14.6 0.22-3.6* IPT Punches; turret time is116curve for a given material during the decompression therefore correspondsto the recovery of the machine plus the compact expansion. To isolate thecompact expansion from the total recovery during decompression, it isnecessary to subtract the machine recovery from the total recovery. Usingthis idea, the tablet expansion, work of decompression, and the Young’smodulus were determined from the decompression phase (section 2.6).3.5.1. Elastic expansion and work of decompression (Wü)Figure 11 (page 57) shows FADm plots for a number of pharmaceuticalmaterials. The linearity of these plots for all materials tested is strongevidence in support of the hypothesis that in-die tablet expansion duringdecompression is elastic. During decompression, force is registered onlyfor as long as the expanding tablet is in contact with the receding upperpunch face. Since tablet expansion may continue both during and afterejection from the die (Aulton et a!., 1973; York and Baily, 1977), ADmFOobtained from Fig. 11 may not represent the total tablet expansion.Representative ADm-fr plots shown in Fig. 10 give a direct indicationof tablet expansion during decompression. Such plots are useful forshowing differences in the expansion of various materials, or formulations,after compression to the same peak force. Microcrystalline cellulosesshowed the greatest expansion during the decompression phase of thecompaction cycle, while brittle substances such as Emcompress and sucroseshowed the least. A similar rank order was found in plots of WD as afunction of peak force, in which the greatest amount of work duringdecompression is done by the microcrystalline celluloses (e.g., AvicelPH1O2, Fig. 30). The range of WD values corresponding to a range of umaxare given in Table V for the various materials compressed on the Betapress.117WD (Nm/g)0-El()El/ El-0.5- A El ElAA°0-1- 0 o0 c)-1.5- A El D-2A0-2.5- El EmcompressA Acet. Granular0 Spray-dried LactoseAvicel PH1O2—3.5 I I I I I0 5 10 15 20 25 30Upper Punch Peak Force (kN)FIG. 30. Change in work of decompression with upper punch peakforce. Each data point is obtained from the decompressionanalysis of a single tablet.1183.5.2. Determination of Young’s ModulusThe modulus of elasticity of a tablet with a certain porosity wasdetermined from the stress-strain plots during decompression. The stressand strain used on such plots were determined using the punch pressure andthe change in the compact height respectively, and are therefore onlyapproximations of the stress and strain at the particulate level. Thisapproximation is however quite reliable since the compact duringdecompression, unlike the powder bed during compression, is a well definedobject and changes in its dimensions during decompression are very small.With increasing peak force, tablet porosity, p, decreases and as itapproaches 0, the modulus of elasticity of compacts, E, will approach theYoung’s modulus, E, of the fully dense material. Some representative plotsof E against p are shown in Fig. 31. Extrapolation of these plots to p =o provides an estimate of E.For low p values, E can be estimated from the least squares solutionof the following linear equation (Wachtman, 1969):E=E+b.p (12)Estimates of E are commonly obtained by least squares solutions of equation13 which is based on an exponential relationship between E and p (Spriggs,1961):lnE = lnE + b.p (13)The application of least squares solutions to nontransformed datarequires the values of E corresponding to each p value to be normallydistributed and randomly selected from such distributions. In addition,119E (GPa)20U EmcompressO> SucroseAvicel PH1O215 \7 STA-Rx-15001o0010203:4PorosityFIG. 31. Variation in modulus of elasticity of tablets withtablet porosity. Extrapolation to zero porosity gives theYoung’s modulus.120the distributions of E at each p value must have equal variances, i.e. thevariances must be homoscedastic. While the least squares analysis is quiterobust with respect to deviations from the requirements of normality andrandomness, the variances must meet the requirement of homoscedasticity(Zar, 1984). The latter requirement can be tested by plotting theresiduals of a linear least squares model, e.g. equation 12, against p(Draper and Smith, 1966; Zar, 1984). A plot showing nearly the same valueof residuals at all p values indicates homoscedasticity, and a simplelinear relationship can be used. If the value of residuals increases withincreasing values of p, the variances are heteroscedastic such that alogarithmic transformation in the form of equation 13 can be applied tomake them homoscedastic (Carroll and Ruppert, 1988). Residual plots forall materials tested in the present study were curved implying acurvilinear relationship between E and p. Linear regression onnontransformed or log-transformed data was therefore invalid. Sinceneither equation 12 nor 13 could be used, the E versus p plots were curve-fitted using polynomial regression. The degree of polynomial was selectedon the basis of a t-test at a critical probability level of 0.05. For mostmaterials a quadratic expression yielded a satisfactory fit.The values of slope, b, obtained from linear regression are “measuresof the rate of change in the modulus relative to the change in porosity”and were related by Spriggs (1961) to the “proportions of closed and openpores, or the proportions of continuous solid-phase structure andcontinuous pore-phase structure”. The need for polynomial regression inour work suggests that factors other than those mentioned by Spriggs(1961), are responsible for changes in E with p. These factors may121include the effect of increasing load leading to differences in thepreferred orientation of certain crystal faces during consolidation.For easily deformed substances such as Avicel PH1O2 and STA-Rx-1500,the extrapolation to zero porosity using polynomial regression is shortproducing reliable E values. Compacts made from substances such asEmcompress and sucrose were much more porous than compacts made from AvicelPH102 and STA-Rx-1500, even after compression at high pressures, and showeda sharp inflection in values. Therefore, the extrapolated estimates ofE are less reliable. Nevertheless, Table VI shows that the E valuesobtained from the Betapress are, in most cases, reasonably close to therange of published values given in Table VII. This is encouraging for thefollowing reasons:a. All tablets contained 0.5 percent magnesium stearate as lubricant andare not therefore a single component. Measurements of E usingnonlubricated materials in a die previously lubricated with a 5 percentsolution of stearic acid in chloroform showed increased scatter.b. Compressive and flexure tests are made on preformed, unconfinedcompacts. In contrast, the tableting process can be likened to theformation of a spring (or rather an assembly of springs with anassociated viscous component) where the spring is formed duringcompaction and confined within the die cavity during the decompressionanalysis of elastic recovery.c. An elastic modulus implies that an equilibrium exists. Compressive andflexure tests are normally performed at low rates of deformation ofpre-equilibrated compacts whereas on the Betapress, the modulus isdetermined under high rate of expansion on a non-equilibrated tablet.122TABLE VI. E values of determined from the decompression analysis.Material E*(GPa)A-Tab 8.1 (2.4)Acetaminophen Fine Powder 11 (1.9)Acetaminophen Granular 9.7 (2.7)Acetaminophen Powder 10 (3.4)Acetylsalicylic Acid 7.2 (1.9)Anhydrous Emcompress 10 (1.4)Anhydrous Lactose 8.1 (3.0)Asagran 4.6 (1.2)Avicel Large 4.1 (0.7)Avicel PH1O1 4.5 (0.5)Avicel PH1O2 5.8 (1.1)Avicel PH1O5 4.4 (0.5)Caffeine 15 (7.2)Cal-Star 9.0 (2.2)Cellactose 4.6 (0.5)Conipap CG 8.5 (1.3)Compap Coarse 73 L 7.5 (0.9)Compap Coarse L 7.2 (1.5)Compap L 8.0 (1.3)DCI-63 6.0 (1.7)Di-Pac 8.2 (1.8)Di-Tab 8.1 (1.5)Elcema G250 7.4 (0.4)Emcocel 6.1 (1.0)Emconipress 27 (5.4)Emdex 9.7 (3.7)Fast-Flo Lactose 6.0 (0.7)Lactose DCL 21 8.5 (1.1)Mannitol (Crystalline) 25 (4.7)Mannitol M.G. 5.0 (1.8)Neosorb (Sorbitol) 5.2 (1.4)a-Lactose monohydrate 36 (3.9)Potassium Chloride 14 (6.9)Racemic ibuprofen 5.9 (1.7)Rhodapap DC-P3 6.8 (2.6)S-Ibuprofen 5.4 (1.3)Sodium Chloride 18 (6.9)Spray-dried Lactose 17 (3.4)STA-Rx-1500 6.6 (0.8)Sucrose (crystalline) 54 (6.1)Sugartab 20 (3.4)Tn-Tab 16 (4.1)Xylitol (Crystalline) 7.2 (2.5)* Values (95% confidence intervals of the estimate of E).123TABLE VII. Literature values of E*.Material E Reference#(CPa)Acetaminophen 8.4 5Acetylsalicylic Acid 8.8 9Avicel PH1O1 0.01-9.0 3,4b,8Avicel PH1O2 4.7, 8.5 2,6Avicel PH1O5 10 2Elcema 5.4-8.6 4b, lOa, lObEmcocel 9.0-9.4 2Emcompress 7.0-182 4b, 10, lOa, lObMannitol (Crystalline) 24 10Mi crocrystal 1 1 ne cellulose(unspecified type) 8.3-16 1, 10, lOa, lOba-Lactose monohydrate 0.84-53 12Potassium Chloride 9.2-26 6, 11Rhodapap DC-P3 (Acetaminophen DC) 5.7 10Sodium Chloride 7.8-186 4, 7, 9, 11Spray-dried Lactose 5.3-14 4b, 10, lOaSTA-Rx-1500 1.4-6.1 4b, lOa, lObSucrose (crystalline) 19 -73 5* These values correspond to different methods of determining E. Thesemethods include four-point flexure testing of large rectangularcompacts, compressive testing of single crystals, single crystalmicroindentation, compressive testing of large cylindrical compacts,and the calculation of E using tablet indentation hardness and yieldvalue from Heckel plots. Please see the references for individualmethods corresponding to each value.# References are numbered as follows:1. Atkins and Mal 1985; 2. Bassam et al., 1988; 3. Church and Kennerley1982; 4. Church and Kennerley 1983; 5. Duncan-Hewitt 1988; 6. Kerridgeand Newton 1986; 7. Lawn and Wilshaw 1975; 8. Mashadi and Newton 1987;9. Ridgway et al., 1969; 10. Roberts and Rowe 1987a; 11. Simmons andWang 1971; 12. Wong and Aulton 1989.a. Values reported by Church (1984) extrapolated to zero porosity usingSpriggs’ equation (Spriggs 1961).b. Values reported by Church (1984) extrapolated to zero porosity usingWachtman’s equation (Wachtnian 1969).124d. Total tablet recovery during unloading and postcompression includesboth elastic and time-dependent viscoelastic components (Rippie andDanielson, 1981). Although viscoelastic expansion can be a significantpart of the total recovery (Celik and Travers, 1985), its contributionwill be negligible on the Betapress where the decompression time wasalways less than 20 ms at a turret revolution time of 1 s. As statedabove, the linearity of the F-ADm plots (Fig. 11, page 57) alsoindicates that expansion in this time period is purely elastic.Errors in the calculation of ADm values due to expansion of the steeltablet leads to corresponding error in the estimation of E. Thus themaximum error in E was about 4.5 percent for Avicel PH1O2 and about 3percent for Emcompress. Extrapolation of E values, corrected for theexpansion of the steel tablet, to p = 0 gave E values which were within the95 percent confidence intervals of the E values in Table VI.For a given material compacted to a specific porosity, some of thescatter in the values of E can be attributed to the interrelated effectsof crystal anisotropy and variation in powder orientation when filling thedie cavity. This is reflected in the 95 percent confidence intervalsreported in Table VI. The E values reported in this table are ‘average’Young’s moduli for the various materials tested. The actual moduli forindividual faces of the single crystals of these materials will differ dueto crystal anisotropy. For instance, the E value of 54 GPa for sucrose isbetween the values of 19 GPa for the (100) surface and 73 GPa for the (001)surface found by single crystal microindentation work (Duncan-Hewitt,1988). It is also within the range of 48 to 97 GPa calculated from theinitial linear portion of the stress-strain data for different125crystallographic axes of a sucrose crystal obtained by a hydrostaticcompression procedure (Bridgman, 1949).3.5.3. Effect of formulation and processing on Young’s modulusThe E values of the processed forms of various excipients such asDi-Pac/Sugartab, microcrystalline celluloses, and Anhydrous Lactose/FastFlo Lactose/Lactose DCL 21 are lower than sucrose, powdered cellulose(Elcema G250) and a-lactose monohydrate, respectively (Table VI). A lowerE value means greater recovery during decompression on the Betapress. Thissuggests that tablets made from formulated or processed materials might beexpected to possess a greater tendency to laminate or cap. In general,however, the tablets are stronger. Thus, tablets of Di-Pac (Fig. 32a),microcrystalline celluloses (Avicel PH1O2 and Emcocel, Fig. 32b), andAnhydrous Lactose/Fast-Flo Lactose/Lactose DCL 21 (Fig. 32c) were muchstronger than sucrose, powdered cellulose, and a-lactose monohydrate,respectively. Similarly, the direct compression forms of acetarninophengave much stronger tablets than crystalline acetaminophen (Fig. 27, page110), even though the E values are lower (Table VI).The apparent paradox arising from the ability of the directcompression forms to produce stronger tablets, despite their greaterelasticity, is explained by the tendency of these forms to deform byviscoplastic flow during compression, which leads to more extensive bondformation. An indication of the degree of viscoplastic deformation duringcompression is given by peak offset time, t0ff, described earlier. Thus, Evalues should be used in conjunction with t0ff to characterise thebehaviour of materials during high speed compression. An interestingsituation is presented by crystalline ibuprofen and its direct compression126Force of FaHure (N)FIG. 32a. Variation in force of failure of tablets with upperpunch peak force for sucrose and its direct compression forms.LI Di-PacV SugartabG Sucrose20015010050LILIU00 5 10 15 20 25 30Upper Punch Peak Force (kN)12710 15 20Upper Punch Peak Force (kN)FIG. 32b. Variation in force of failure of tablets with upperpunch peak force for various celluloses.Force of Failure (N)Avicel PH1O2El EmcocelA Elecma G250600450 -300 -150-0-0-r5AA25 30128FIG. 32c. Variation in force of failure of tablets with upperpunch peak force for various lactoses.Force of Failure (N)Anhydrous LactoseU Lactose DCL 21Fast-Flo Lactoseo Alpha Lactose.H20200150100-50 -0-000 5 10 15 20 25Upper Punch Peak Force (kN)30129formulation, DCI-63. The expansion of both materials is identical and asgreat as Avicel PH1O2. DCI-63 contains 63 percent ibuprofen, lubricant andother USP/NF grade excipients suggesting that the major component(s) of theexcipients have similar elastic properties to ibuprofen. Compared withcrystalline ibuprofen which cannot be compacted into tablets, DCI-63 showsan increase in the t0ff (Fig. 28, page 111) indicating that the excipientsimpart viscoplastic flow to the formulation and facilitate interparticulatebonding.3.5.4. The use of decompression analysis and of Young’s modulusThe analysis of machine recovery during decompression using therelationship between punch force and machine deformation provides areliable estimate of the in-die tablet expansion and E. Together with themeasurement of t0ff and work of compression (Oates and Mitchell, 1989), thecalculation of tablet expansion, work of decompression and E, using thedecompression analysis, could be useful in preformulation studies on newdrugs and excipients, in the quality control of in-coming tabletingmaterials, and for in-process validation of compaction.The E values of pharmaceutical materials determined by compressive orflexure testing of compacts, tablet indentation, and single crystalmicroindentation require specialized equipment. The present method is anovel application of a high speed rotary press in that the Young’s modulus,a fundamental material constant, is readily obtained under normal tabletingconditions without the use of specialized equipment.The Young’s modulus is an indicator of the stiffness of a materialand can be used on its own to describe the deformability of a solid underload in the elastic regime of deformation. A material with a low E value130would tend to deform more easily than a material with a high E value.Young’s modulus alone is of little significance in identifying the flow orfracture characteristics of particles during compaction, but can be of usein this respect in conjunction with WD and other parameters, such as Uy,t0ff, AP/VT and WC derived from the compression phase.3.6. ENERGY CHANGES DURING COMPACTIONThe following section refers to the energy changes during compactionin terms of the WC and the WD values. The WC and WD values (bothnormalized for tablet mass) corresponding to the range of max over whicheach material was compressed, are given in Table V (page 115).. Thecorrelation between possible mechanisms of deformation, the homologoustemperature of various materials, and the energy changes during thecompaction of these materials, are discussed below.3.6.1. Consumption of energy during the compression phaseThe increases in stress at the points of interparticulate contactinvolve an increase in the energy of the powder bed as the machine doeswork on the powder bed. The process of permanent deformation of particlestends to decrease this energy and the excess energy thus released increasesthe temperature of the powder bed, as well as that of the punches and thedie. Other processes such as friction also contribute to this energychange. These events are accompanied by bond formation during thecompression phase. During decompression the converse is true since theexpanding tablet does work on the machine. Thus, the difference WC-WD isan approximation of the work lost in particle deformation and bondformation during the compression phase. A material with low WC-WD values131would deform more readily and require very little energy from the machine,or alternatively will have a strain-rate independent deformation mechanismand thus would not require a continuous input of energy by the machineuntil the max is reached during the compression phase. The range of WC-WDvalues, corresponding to the range of max values given in Table V, arepresented in Table VIII.3.6.2 Temperature dependence of the mechanisms of deformation and itscorrelation with the energy changes during compactionThe deformation mechanism of any given material may be temperaturedependent. A material which fractures at room temperature may flow at ahigher temperature closer to its melting point. This concept can beexpanded to compare different materials in the sense that a material with alow melting point would probably have a greater tendency to flow at roomtemperature, and hence deform with greater ease, than a material with ahigher melting point. A plot of WC-WD for various organic materialsagainst their homologous temperature, i.e. the ratio of absolute roomtemperature (293 K) to their absolute melting temperature, seems to agreereasonably well with this contention (Fig. 33). The scatter of W-W onthis plot is probably due to differences between particle size, particleshape, crystal system, type and strength of bonds within and in between theparticles, and true and bulk densities of the various materials. Also themelting points of certain materials on this plot (e.g. the celluloses andsucroses) do not really correspond to melting, but to decomposition ormelting with decomposition. Nonetheless, a more or less general trend onFig. 33 indicates that the energy of deformation (reflected by W-W) is,in general, related to the inherent energy of bonding within the particles1323.69-19.82. 51-16.91.40-12.31.08-14.91.74- 5.793.81-19.83.00-20.13.34- 8.359. 21-40.37. 58-37. 13.23-40.25.48-32.01.95-19.63.32-12.95. 70-31.65.89-28.93.62-22.55.08-22.96.48-28.55.79-16.65. 72-20. 73.26-13.15. 25-23 .814.0- 38.53.32-13.52. 14-27 .35. 12-23. 11.81-18.35. 56-23 .83.36-16.55. 72-21. 93.55-16.25. 13-14.64.59-12.57. 55-27.32.74- 7.763.35-14.14.03-20.63. 52-21.23.21-15.23.36-17.31.63-12.41. 85-11. 00.516-0.3480. 225-0. 0410. 184-0. 0240. 220-0. 04001010007a0.518-0.3490. 292-0. 09801100001a0.447-0.0490.443-0.0340.430-0.0380.439-0.0380.334-0.0980.352-0.1670. 401-0. 0830. 315-0. 0540.305-0.0470.268-0.03 10. 296-0. 06001490002a0. 248-0. 0680.341-0.1700.341-0.0400. 281-0. 0280. 281-0. 1770.378-0.0470.294-0.0800.310-0.0920.257-0.0880. 257-0. 0620.263-0.0520. 2 18-0. 0940.158-0.009•106011a0. 279-0. 064010500010. 211-0. 0450. 287-0. 0790.223-0.0070. 220-0. 0700. 211-0. 0360.527-0.3840. 226-0. 0548.88-145bb11. 0-21. 99.33-14716.5-19429.3-1286.l4223304...>5 0d32 .0->500500...d624-100C13.2-1740.73-28919.l-92c4.61- 1586.29-13718.3-19811.7-1400-55.896.8->5009.60-1355.81-38014.8-1909.60-19010-31c7.92-14829. 4-3 536.07-1356. 73-61. 7b34..9-l2lcb14•0325e6.00-1083•0088e0-31.72.80-41.04.99-1072.86-55.9* (IPT Punches; turret time is). a. Compacts fully dense at high pressuresand transmit the stress radially; b. No tablets at any pressure; c. Tabletslaminate at high pressures; d. Tablets at high pressures overload the CT-40load cell; e. No tablet at low pressures.TABLE VIII. Range of WC-WD, porosity and Ffrange given in Table IV.values* corresponding to the maxMaterial WC-WD Porosity Ff(J/g) (N)A-TabAcetaminophen Fine PowderAcetami nophen GranularAcetaminophen PowderAcetyl sal icyl Ic AcidAnhydrous EmcompressAnhydrous LactoseAsagranAvicel LargeAvicel PH1O1Avicel PH1O2Avicel PH1O5CaffeineCal -StarCell actoseCompap CGCompap Coarse 73 LCompap Coarse LCompap LDCI-63Di-PacDi -TabElcema G250EmcocelEmcompressEmdexFast-Flo LactoseLactose DCL 21Mannitol (Crystalline)Mannitol M.G.Neosorb (Sorbitol)a-Lactose monohydratePotassium ChlorideRacemic IbuprofenRhodapap DC-P3S- IbuprofenSodium ChlorideSpray-dried LactoseSTA-Rx- 1500Sucrose (crystalline)SugartabTn -TabXylitol (Crystalline)1330.7Homologous TemperatureFIG. 33. A plot of the difference in the lost work (WC-WD) athighest and lowest max (215 MPa and 2O MPa respectively) forseveral organic substances against their homologous temperatures(the ratio of absolute room temperature, 293 K, to the absolutemelting temperature), when each material was compressed over asimilar max range. Each point represents one material.WcWD (J/g)403020 -10-0-0.5 0.6 0.8 0.9134(reflected by the homologous temperature). This plot does not establishthe actual deformation mechanism, but can be used for this purpose inconjunction with other properties obtained from the compaction cycle.3.6.3. Relationship between tablet strength and WC-WDThe lost energy represented by WC-WD is also an indication of theenergy stored in the tablets partly in the form of bonds between theparticles and partly in the form of stressed but unyielded particles.After decompression the stress on these unyielded particles will berelieved by a slow stress-relaxation phenomenon, thereby releasing theexcess energy in the form of heat. The energy stored in the form of bondswithin the tablets can be approximated by various tablet strength tests,e.g., tensile fracture test by diametral compression (Fell and Newton,1968, 1970), impact test (Hiestand et al., 1971), flexure test (David andAugsburger, 1974), tablet indentation hardness test (Jetzer et al., 1983),or the more recently proposed impact fracture wear test (Duncan-Hewitt andGrant, 1987b).All materials were tableted over a similar range of max and theforce of failure, Ff, was measured using a diametral compression test. Aplot of Ff of the strongest intact tablet of each material against thecorresponding WC-WD shows a rough correlation between tablet strength andWC-WD (Fig. 34). Various formulations and processed materials which formintact and/or much stronger tablets relative to their parent compounds havemuch higher W-W, e.g. various Compaps compared with crystallineacetaminophen, Di-Pac compared with sucrose, Fast-Flo Lactose compared witha-lactose monohydrate, and DCI-63 compared with racemic ibuprofen. Thus,WC-WD is also an indication of the tablet strength of various materials1356255003752501250Force of Failure (N)45FIG. 34. A plot of Ff of the strongest intact tablet producedwhen a material was compressed over a range of max’ against thecorresponding lost work (WC-WD). Each point represents onematerial.5 15 25 35W-W0 (Jig)136regardless of the mechanism of deformation of their particles during thecompression phase. A similar quantity called net work was used byRagnarsson (1985) to characterize materials, and was shown to be sensitiveto friction and bonding.3.6.4. Relationship between WC-WD and porosity changesThe aim of compaction is to decrease the porosity of a powder bed bycompression, thereby increasing the probability of the particles forming acohesive, compact structure. The larger the change in porosity, the largeris the amount of work consumed during compression. A plot of thedifference between WC-WD at max 215 MPa and at max 2O MPa against thecorresponding porosity change for the materials listed in Table VIII islinear (Fig. 35) indicating a direct relationship between the amount ofenergy consumed and the reduction in porosity. A bulky solid (high bulkvolume, e.g. Avicels) has a greater reduction in porosity, relative to aless bulky solid (low bulk volume, e.g. Emcompress, Tn-Tab). Hence theamount of energy consumed per unit mass is higher. The probability of thisenergy being stored as bonds is also likely to be higher, and the tabletsshould be stronger. This seems to be true as indicated by the generaltrend in Fig. 34.It follows from Figs. 34 and 35 that, to increase the tablet strengthof a particular solid, it is necessary to increase the bulk volume of thesolid either by mixing it with a more bulky substance, or by processing itin a manner that increases the bulk volume. More often than not, this willbe accompanied by a change in the mechanism of particle deformation, whichwill be reflected by a change in the various parameters obtained from thecompression and the decompression phases. Examples of this can be seen in137WC-WD (J/g)402010-0- I0.05 0.15 0.25 0.35 0.45Maximum Porosity ChangeFIG. 35. A plot of the difference in the lost work (WC-WD) athighest and lowest umax (215 MPa and 2O MPa respectively) forseveral substances, against the corresponding change in compactporosity, when each material was compressed over a similar maxrange. Each point represents one material.138Table VIII. Cellactose and microcrystalline celluloses form much strongertablets than lactose and powdered cellulose respectively.3.7. POSSIBLE MECHANISMS OF PARTICLE DEFORMATIONHypothetically, if a given material could have different deformationmechanisms, then the ease of deformation at the room temperature would havethe following order: viscous > viscoplastic among the strain-rate dependentmechanisms, and plastic > fracture (if Uy is low) and fracture > plastic(if Uy is high) among the non-strain-rate dependent mechanisms.Materials which deform by viscous flow, especially those having lowcoefficients of viscosity, may sustain a large amount of strain withoutfracturing. Such behaviour is observed in ductile metals. Pharmaceuticalmaterials may not exhibit a truely ductile behaviour because most of thesematerials are organic compounds with more complex crystal lattices than theductile metals. Nevertheless, those pharmaceutical materials that readilydeform at low stresses can be termed ‘ductile’ for the lack of a betterterm. Such materials will deform much more at a given strain rate andunder a given stress than viscoplastic materials. Both ductile andviscoplastic materials will have relatively low Uy values, but the valuesof ductile materials will be much lower than those of viscoplasticmaterials. Both will compact to very low porosities, and will have low Evalues indicating that they are very easily deformable even under lowstresses. The primary difference between them will be in their strain-ratedependency, which will be reflected in the values of t0ff and AP/VTobtained from the compression phase (sections 3.4.4 and 34.5). Ductilematerials will be compacted to very low porosities at lower pressures thana viscoplastic material and, in effect, will behave as almost139incompressible solids with a very short t0ff and almost no AP/VT. On theother hand, viscoplastic materials, being slightly stiffer, will havelonger t0ff values which will decrease with increase in max’ and have amuch higher AP/V-j-. Due to differences in the ease of their deformabilitythere will also be differences in the energy consumed during compression,i.e., WC-WD for ductile materials will be much lower than for viscoplasticmaterial s.The of plastic materials will be greater than ductile orviscoplastic materials but smaller than intrinsically brittle materials.Both plastic and intrinsically brittle materials will have low t0ff valueswith small AP/VT, but for different reasons. A plastic material would flowduring compression phase to fill the voids which would result in lowporosities, whereas a brittle material would fracture without filling upthe voids thereby leaving a porous compact at the end of the compressionphase with a structure capable of resisting flow. The E values of bothplastic and brittle materials would tend to be higher than those of theductile or viscoplastic materials indicating their greater stiffness undersimilar conditions of compression. Since the element of strain-ratedependency is absent in plastic flow/fracture mechanisms, the energyconsumption by these materials (WC-WD) during compression will be muchlower than in the case of viscoplastic materials because the machine doesnot have to work continuously during the compression phase.The above discussion provides a general background by which one canattempt to identify the principal particle deformation mechanisms ofvarious materials. The materials listed in Tables Il-VI have been140categorised on the basis of their stress and strain rate behaviour and onthe properties derived from the compaction cycle.3.8. DEFORMATION MECHANISMS OF THE VARIOUS CATEGORIES OF SOLIDS3.8.1. Deformation mechanism of acetylsalicylic acid, ibuprofen and theirformulationsDrugs such as acetylsalicylic acid, racemic ibuprofen, S-ibuprofen,and formulations such as Asagran and DCI-63, all have very short t0ff andvery low AP/VT preliminarily suggesting that their deformation is notstrain-rate dependent. The very low values of Uy and E indicate that theparticles of these materials are easily deformed. The low values of WC-WDshow that only a small amount of energy is consumed in deforming thesematerials. At high max’ the porosity becomes less than zero indicatingthat the compacts become fully dense and undergo a hydrostatic compression,which causes significant radial expansion of the die wall. Thus, thevolume between the punch faces, if not corrected for radial expansion,becomes lower than the true volume of the powder resulting in a negativeporosity.The low values of E, Uy and WC-WD suggest that these materials arereadily deformed and can be classified as low yield-strength ductilesolids. Due to their viscous nature they become almost nonporous at verylow stresses, and then behave as almost incompressible solids. Thisresults in low t0ff and correspondingly low AP/VT, and can give rise to theerroneous impression that the deformation of these materials is independentof rate of compaction.The low melting temperatures of acetylsalicylic acid, racemicibuprofen and 5-ibuprofen are consistent with a viscous deformation141mechanism. At room temperature (293 K) their homologous temperatures are0.68, 0.84 and 0.91 respectively. The ease of their deformation becomesobvious considering that almost all materials tend to show some degree offlow above a homologous temperature of about 0.5. Any rise in temperatureduring tableting will further increase the degree of the viscous flow ofthese drugs and formulations.3.8.2. Deformation mechanism of calcium phosphatesThree types of calcium phosphate are available commercially:dicalcium phosphate dihydrate (Cal-Star, Di-Tab, Emcompress), anhydrousdicalcium phosphate (Anhydrous Emcompress, A-Tab) and tricalcium phosphate(Tn-Tab). All have short t0ff values, low AP/VT, high to very high Uy,and higher E compared with other materials. The higher o, and E valuessuggest that these materials are very stiff and not easily deformed. Thecompacts remain very porous at the end of the compression phase. At highmax’ the residual porosity values of =0.17 for the dihydrates, =0.35 forthe anhydrous samples and =0.38 for tricalcium phosphate suggest anincreased resistance to flow among these samples which correlates well withtheir respective Uy values of =90 MPa, =163 MPa and 221 MPa. Theseobservations, coupled with the short t0ff suggest that the calciumphosphates do not deform by a flow-based mechanism and that there is anegligible dependency of their deformation on strain rate. The most likelymechanism of deformation is therefore fracture of some type. The calciumphosphate powders have a low bulk volume, and the machine has to work for amuch shorter period of time during compression (Fig. 5, page 43), theEmcompress curve), causing a relatively small energy expenditure in theircompaction, hence the low WC-WD. Still, the compacts are fairly strong142presumably due to strong interparticulate bonds formed between the newsurfaces exposed upon fracture.In general, calcium phosphates can be classified as high yield-strength intrinsically brittle solids (see section 3.4.1 for explanation ofintrinsically brittle behaviour). This mechanism allows a large porevolume in the compacts, and does not require the machine to expend a largeamount of energy in continuously deforming the particles throughout thecompression phase. Fracture only requires sufficient energy for crackinitiation and propagation. Their intrinsically brittle nature wouldpreferentially allow fracture rather than flow during compaction. Thereis, however, some likelihood of flow at low max’ because at low max theporosities are very high, hence T could be very high causing some flowconsistent with the short t0ff and small AP/VT at these pressures.It is not possible to correlate the deformation mechanism of calciumphosphates with their homologous temperature, because only the meltingpoint of tricalcium phosphate can be determined. Other calcium phosphatesundergo phase transitions such as dehydration or condensation of theorthophosphate groups to pyrophosphate groups (Rhone-Poulenc ProductInformation Manual, 1989), but at temperatures much higher than the meltingpoints of organic substances listed in Table II. If the melting point oftricalcium phosphate (1943 K) and the transition temperatures of dicalciumphosphates (>670 K for the condensation reaction) are any indication of thestrength of their structure, it is reasonable to expect that, at roomtemperature, their particles would not deform by a flow process.1433.8.3. Deformation mechanism of cellulosesCelluloses are characterised by low Uy, long t0ff values and a largeAP/VT, high WC-WD, and relatively low E values (Table III, IV, VI, VIII).The low E and Uy values indicate that celluloses deform easily, albeit notas easily as acetylsalicylic acid or ibuprofen. This is supported by theobservation that the decrease in the porosity of cellulose compacts overthe range of max applied is not as sudden as with acetylsalicylic acid oribuprofen, even though the decrease is much larger (Table VIII). At lowmax the compacts of celluloses are quite porous (p O.4), but withincrease in max the porosity gradually decreases to very low values (pO.O4). The change in t0ff, and AP/VT, with max are also gradual. Thehomologous temperature of celluloses is 0.54, indicating that there shouldbe a certain degree of viscous flow, but not as significant as that ofacetylsalicylic acid or ibuprofen. This, and the higher Uy, indicates thatthe celluloses are much more viscous, i.e. less easily deformed, than thesedrugs. On the other hand they are much more readily deformed thanmaterials such as lactose, sucrose or the calcium phosphates, which havegenerally much higher Uy and E, and have comparable or lower homologoustemperatures (O.60 for lactoses, 0.66 for sucroses, and 0.15 fortricalcium phosphate). The celluloses can therefore be classified asintermediate yield-strength viscoplastic solids, which have a significantdegree of strain-rate dependent flow (hence long t0ff), and also a certaindegree of stiffness (hence intermediateOf the two types of celluloses commercially available for directcompression, the microcrystalline celluloses (different Avicels andEmcocel) form much stronger tablets than powdered cellulose (Elcema G250).There are small but obvious differences between the tableting parameters of144powdered cellulose and the microcrystalline celluloses. Powdered cellulosehas slightly shorter t0ff, and slightly higher ay and E suggesting that itis less easily deformed than microcrystalline cellulose. Powderedcellulose also consumes much less energy (lower WC-WD) during compression,which is partly why it forms much weaker tablets. Although the elasticexpansion of powdered cellulose during decompression is similar to that ofthe microcrystalline celluloses, it has a significantly higher volumetricexpansion during ejection (Fig. 36), which further weakens its tablets.Cellulose has very strong hydrogen bonds between adjacent n-glucosechains (Talman, 1977). The strong inter-chain bonding, their partiallyamorphous nature, and the large amount of energy consumed (highest WC-WD)during compression of the microcrystalline celluloses results in hightablet strength. In fact, microcrystalline cellulose tablets are thestrongest of all the materials used in this study. The viscoplasticbehaviour and the high strength of their tablets makes the microcrystallinecelluloses excellent direct compression agents.3.8.4. Deformation mechanism of acetaminophen and its formulationsCrystalline acetaminophen has a short t0ff and a small AP/VTcomparable to those of the calcium phosphates. The Uy and E values arehigher than those of the celluloses but much lower than those of thecalcium phosphates. It is compacted to porosities almost as low as thecelluloses. Tablets are formed only when large particles are compressed atlow pressures. The amount of energy consumed (WC-WD) during compression iscomparable to that of the calcium phosphates, and is much lower than thecelluloses. The homologous temperature of acetaminophen (=0.66) is similarto that of the celluloses but is much higher than that of the tricalcium145Compact Volume (cm3)0.545AAA AAAA A A u0.495 AA i=i0 50 100 150 200 250Upper Punch Peak Pressure (MPa)FIG. 36. Compact volume of celluloses determined from compactdimensions measured immediately post-ejection as a function ofmax The solid line represents the compact volume at the deadcentre position. This line was determined in a separateexperiment by compressing lead shot (Section 2.6.5).146phosphate. Thus, acetaminophen appears to be fairly deformable at roomtemperature (hence the low porosities) and has a certain degree of flow(short t0ff), but beyond a certain max the ability to flow diminishes(negligible t0ff). Large crystals of Acetaminophen Granular readily yieldat low pressures, at which the porosities are low, probably because the UTis very high, and eventually fail by fracture exposing new surfaces. Somebond formation occurs between these new surfaces and intact, but weak,tablets are formed. At high max’ or when the particle size is decreased(as with Acetaminophen Fine Powder or Powder), these events do not occurand no tablet is formed. The brittle behaviour of acetarninophen isconsistent with the observation that the fine particles of acetaminophen donot form tablets and have a slightly higher Uy, because it is known thateven the most brittle materials will not fracture once the particle size isreduced to below a critical value (Kendall, 1978; Roberts and Rowe, 1987).Crystalline acetaminophen can therefore be classified as an intermediatestrength plastic/brittle solid, where the plastic flow results in lowporosities and short t0ff, and the plastic/brittle behaviour resulting in arelative lack of strain rate dependence after the initial yielding. Sincethe overall porosity changes are small, the machine does not have to expenda large amount of energy and WC-WD is low.Crystalline acetaminophen has a higher E than celluloses, i.e., itselastic recovery during decompression is lower than that of the celluloses.Hence the inability of acetaminophen to form tablets is apparently not dueto a high recovery as is commonly believed, but due to its inability toform bonds during compression, or due to the low strength of the bondsformed, if any. Using a flexible die to prolong decompression, Arnidon eta!. (1981) obtained intact tablets of an acetaminophen formulation. They147concluded that this process enabled the weak interparticulate bonds towithstand the tablet expansion. However an alternative explanation is thatthe longer contact time (time during which the die contents are underpressure; Jones, 1977) leads to increased deformation and an increase inthe number of bonds.When crystalline acetaminophen is formulated with other substances,t0ff and AP/VT are increased especially at low max’ °y values are eithersimilar or slightly decreased, and E is decreased. The porosity changesare much larger. Therefore, WC-WD increases significantly and the compactsare much stronger despite an increase in the elastic recovery duringdecompression as reflected by the low E values. An increase in t0ff and inAP/VT at low max suggests that the degree of strain rate dependency of theformulations is higher and the mechanism of deformation is more viscousthan plastic/brittle. Some formulations do not form intact tablets at highmax’ at which their t0ff values are very close to those of crystallineacetaminophen. This indicates a decrease in the degree of viscous flow,and this decrease, coupled with the greater recovery of the formulationsduring decompression, causes failure of the compacts at high max3.8.5. Deformation mechanism of lactosesLactose in its processed forms is commonly used as a directcompression excipient. The processing results in either a-anomer richproducts (e.g., Fast-Flo Lactose, Spray-dried Lactose) or fl-anomer richproducts (e.g., Anhydrous Lactose and Lactose DCL 21) (Dwivedi andMitchell, 1989). There are distinct differences between the tabletingparameters of these forms. Alpha-lactose monohydrate has the shortest t0ffand highest Uy and E among all lactoses tested. The AP/VT and the residual148porosities of the different lactoses are very similar. The processed formshave higher WC-WD values than a-lactose monohydrate, and the strength oftheir tablets is also higher. The small difference in W-W values,despite very similar porosities, indicates a slight degree of difference inthe deformation mechanism of the various forms.The Uy and F values of the lactoses are higher than those of thecelluloses, but much smaller than those of the calcium phosphates. Theirt0ff values, AP/VT, and W-W are much smaller than those of the cellulosesbut comparable to those of the calcium phosphates. This suggests thatalthough the lactoses have a certain degree of flowability (short t0ff),their deformation mechanism is predominantly fracture of some type. Thisis especially true of a-lactose monohydrate which has the highest ay amongthe various lactoses. According to the rank order of the Oy values therank order for the ease of deformation is: a-rich processed samplesrich processed samples > a-lactose monohydrate. Alpha-lactose monohydratecan be classified as an intermediate strength plastic/brittle solid, whilethe processed samples have a greater degree of plastic flow. The increaseddeformation by flow is probably due to the amorphous content of theprocessed samples.The homologous temperatures of the lactoses are in the intermediaterange. Although the higher melting point and slightly higher Uy values offl-rich samples indicate that they should be less readily deformed than thea-rich samples, they have comparable t0ff, indicating that they deformequally well. The fl-rich samples form stronger tablets presumably becausethe lattice bonds are much stronger than those of a-rich samples.1493.8.6. Deformation mechanism of sucrosesRelative to a-lactose monohydrate, the of sucrose is lower, E ishigher, the t0ff values and AP/VT are comparable, the homologoustemperature is slightly higher, and WC-WD is lower. The similarity of thevarious tableting parameters of crystalline sucrose and a-lactosemonohydrate suggests a similar deformation mechanism. Therefore, sucroseis also classified as an intermediate yieldstrength plastic/brittle solid,with perhaps a slightly greater ease of deformation than a-lactosemonohydrate. Sucrose, however, forms much weaker tablets than a-lactosemonohydrate. This can be correlated with the lower bond strength ofsucrose as reflected by its lower temperature of melting with decomposition(175°C) relative to that of anhydrous a-lactose (dehydration of themonohydrate at 15O°C followed by melting at z210°C).The tableting parameters of crystalline sucrose are quite differentfrom those of Sugartab and Di-Pac, the two types of compressible sugar usedin this study. Sugartab is an agglomerated sugar product containingz90-93% sucrose with the rest being invert sugar, whereas Di-Pac is acrystallized product containing 98% sucrose with 2% dextrins (Handbook ofPharmaceutical Excipients, 1986). Compared with crystalline sucrose, thet0ff, AP/VT and WC-WD for the compressible sugars are much higher, butE and the minimum residual porosities are lower, while the homologoustemperature are almost identical. This suggests that the compressiblesugars deform more readily. The agglomeration and inclusion of invertsugar, or co-crystallization with higher saccharides, evidently imparts acertain degree of viscoplastic flow. The original plastic/brittlebehaviour is probably still maintained to a certain degree, the differencebeing that the samples perhaps fail by a fracture mechanism that shows some150degree of ductility. The degree of viscoplastic behaviour is not aspronounced as in the case of celluloses because WC-WD, although higher thanthat of crystalline sucrose, is much lower than that of the celluloses, andalso the porosity changes are not as marked as with the celluloses. Thatincreased flowability does not necessarily result in stronger compacts isevidenced by the weaker compacts of Sugartab relative to those of Di-Pac.This difference is presumably because the Di-Pac particles bond much morestrongly due to the small content of polymeric dextrins, whereas theinclusion of invert sugar in Sugartab does not have a similar effect.3.8.7. Deformation mechanism of polyolsThree monosaccharide-derived polyols, namely, mannitol, sorbitol andxylitol, are used in the crystalline state or in processed form as directcompression excipients. The tableting parameters of crystalline mannitoland xylitol, and direct compression forms Mannitol M.G. (mannitol) andNeosorb (sorbitol) are given in Table Il-VI, and Table VIII. The t0ff,AP/VT, and W-W of all samples are higher than those of calciumphosphates, but lower than those of celluloses, suggesting a certain degreeof flowability. The homologous temperatures of xylitol, sorbitol andmannitol are zO.8O, =0.77 and =0.66 respectively, which also suggests adeformation mechanism involving flow. The 0y and porosities are higherthan those of celluloses but much lower than those of calcium phosphates.The E values are generally higher than those of celluloses, and are similarto those of calcium phosphates, indicating a fair degree of stiffness intheir lattice, which may cause fracture once a certain degree of flow hasoccurred. Thus, the polyols seem to have a certain degree of viscoplasticflow, although not as much as that of the celluloses, and also a certain151propensity to fracture, but not as much as of the calcium phosphates. Thefracture is probably ductile and occurs only after an initial viscoplasticflow. The initial flow permits a rapid reduction in porosity, but thesubsequent fracture does not allow this reduction to continue and locks afair proportion of pore space in the compacts. The polyols used in thisstudy can therefore be classified as intermediate yield-strengthviscoplastic/brittle solids.The toff of the various polyols do not differ much from each other.The AP/VT has the rank order: Neosorb > crystalline mannitol z MannitolM.G. > crystalline xylitol, which implies a difference in the rate ofstress relief during t0ff. The of mannitol and xylitol are similar, butthe E value of mannitol is higher than that of xylitol. The and E ofthe direct compression formulations are lower than those of the crystallinesamples. The Uy of Neosorb is lower than that of Mannitol M.G.. The WCWDvalues have the following sequence: crystalline mannitol > Neosorb >Mannitol M.G. > crystalline xylitol. Neosorb forms the strongest tablets,xylitol the weakest, and mannitol with intermediate strength. Smalldifferences in their tableting parameters and also differences in otherphysical properties are the underlying causes of their different tabletstrengths. For example, mannitol and sorbitol are isomers of each other,therefore they might be expected to produce tablets of nearly equalstrength. However, the lower Uy, slightly longer t0ff, much higher P/VT,and a higher homologous temperature of sorbitol indicate that it deformswith greater ease, which increases the probability of bond formation andhence sorbitol produce stronger tablets than inannitol. This suggests thatdifferences in crystal structure, and the associated differences in theability of deformation, can cause significant differences in the tablet152characteristics of materials with similar chemical structure. Xylitoltablets are weakest presumably due to the very low amount of energyconsumed during its compaction and also due to the weaker nature (low m.p.)of its crystal lattice.3.8.8. Deformation mechanism of miscellaneous other substancesCellactose is a formulation of 25% microcrystalline cellulose with75% lactose (Garr and Rubinstein, 1991). Its tableting parameters areintermediate between those of its constituent components. The t0ff, AP/V-jand WC-WD of Cellactose are slightly smaller, and Uy slightly higher, thanthe microcrystalline cellulose, indicating a slight reduction in the degreeof viscoplastic flow due to the inclusion of lactose. The E. value issimilar to that of microcrystalline cellulose. A reduction in theviscoplastic flow and an equal degree of elastic recovery of Cellactoseleads to the formation of weaker tablets compared with microcrystallinecellulose, but the tablets are still stronger than lactose alone.Cellactose can be classified as an intermediate yield strength viscoplasticsolid which is essentially in the same category as microcrystallinecellulose.Caffeine has a short t0ff and a low AP/VT indicating a lowstrain-rate dependency. It has intermediate Uy, E, WC-WD, and porositychange suggesting it is a fairly stiff solid with the degree of stiffnessvery similar to that of the lactoses. This suggestion is supported by thesimilarity of its homologous temperature (O.57) to that of the lactoses(=0.59). Hence it can be classified as an intermediate-strengthplastic/brittle solid, with a lower tendency to flow than lactoses as itst0ff and AP/VT are lower than that of the lactoses. The compacts of153caffeine are slightly stronger than those of the lactoses, but theylaminate at pressures above 127 MPa. This is perhaps due to the resistanceto flow of caffeine which presumably forms fewer interparticulate bonds.Emdex is made up of microcrystals of dextrose as porous spheres,with some amount of higher polysaccharides (Seugling, 1980). Its tabletingparameters are slightly different from those of the celluloses. Comparedwith the celluloses the t0ff are shorter, AP/V-j- is lower, the porositychange is smaller, hence WC-WD is lower, and ay and E values are slightlyhigher. These differences suggest that Emdex is a viscoplastic materialwith a lower degree of rate of compaction dependency and higher stiffnessthan the microcrystalline celluloses, and can be classified as anintermediate yield-strength viscoplastic solid. The tablets of Emdex areweaker than those of microcrystalline celluloses, which suggests lessviscoplastic flow into the voids and consequently a lower degree ofbonding.Starch is a polymer made up of partially linear and partiallybranched chains of a-glucose (The Merck Index, 1976). STA-Rx 1500 is acommercial form of pregelatinised starch. Its tableting parameters areintermediate between those of ibuprofen/acetylsalicylic acid and themicrocrystalline celluloses. Its low Uy and E values suggest that it iseasily deformed and hence forms compacts of very low porosities. Its t0ffand AP/VT are low, which indicate that the compacts become very denseearlier on during the compression phase, and behave as a nearlyincompressible solid. Thus STA-Rx 1500 can be regarded as a low yieldstrength ductile solid. The WC-WD is intermediate, which means that thecompacts should be fairly strong. In fact, they are very weak, implyingthat not much of the lost work is stored in the form of interparticulate154bonds, but is probably lost in other irreversible processes. Also, thebonds between the a-glucose chains may not be very strong. Evidence tosupport this is the difference in solubilities of starch and cellulose.Starch dissolves in boiling water, whereas cellulose does not. This wouldbe true if a-glucose chains are much more weakly bonded in starch than thefl-glucose chains in cellulose. The weak nature of the bonds coupled with ahigh elastic recovery of its compacts during decompression weaken the STARx 1500 compacts. Lubrication with magnesium stearate is also known toweaken the compacts of STA-Rx 1500 (Ragnarsson and Sjorgen, 1985).• Two ionic salts, namely, potassium chloride and sodium chloride, werealso studied on the Betapress. Both have low homologous temperatures(z 0.28), which suggests that at room temperature the preferred mechanismof their deformation should be fracture, which would be consistent withtheir low WC-WD. Potassium chloride has longer t0ff values and a greaterAP/VT than sodium chloride. It also has lower Uy and E, and is compactedto much lower porosities. Therefore potassium chloride can be classifiedas a ductile solid, and sodium chloride as a plastic/brittle solid, both ofwhich fail by a fracture process during compression at room temperature.The longer t0ff, and larger AP/VT, of potassium chloride at low pressuresare manifestations of its greater ductility.The various materials and their general deformation behaviour islisted in Table IX. The above discussion of the various general classes ofmaterials shows that it is possible to use a rotary press for acomprehensive analysis of the deformation mechanisms of pharmaceuticalmaterial s.155TABLE IX. Classification of deformation behavioura of various solids onthe Betapress.Material Classification of deformationbehaviourAcetylsalicylic acid, ibuprofen Low yield-strength ductileand their formulations, solidsSTA-Rx- 1500Potassium chloride Low yield-strengthductile/brittle solidCellactose, Celluloses, Emdex Intermediate yield-strengthductile/viscoplastic solidsFormulations of acetaminophen Intermediate strengthplastic/brittle solids withsome viscous flowPolyols and their processed forms Intermediate yield-strengthviscoplastic/brittle sol idsThe processed forms of a-Lactose Intermediate strengthmonohydrate and of Sucrose plastic/brittle solids withsome viscoplastic flowAcetaminophen, caffeine, Intermediate strengtha-Lactose monohydrate, plastic/brittle solidSodium chloride, SucroseCalcium phosphates High yield-strength brittlesolidsa. This classification provides a simplified view of the generaldeformation behaviour of various solids. There are subtle differencesin the extent of deformation within a given class. For a detailedanalysis please see the discussion in section 3.8.1563.9. COMMENTS ON INTERPARTICULATE BOND FORMATION UNDER PRESSUREAn examination of the differences between the force of failure valuesof various tableted solids (Table VIII) and the differences in theirdeformation mechanisms (section 3.8, Table IX) indicate that, although thepermanent deformation of a material under stress is considered aprerequisite for bond formation during compression, it does not guaranteethe formation of strong tablets. For example, acetylsalicylic acid andibuprofen are very readily deformed (section 3.8.1). Nevertheless, theyform very weak tablets or no tablets at all. This is presumably becauseinsufficient bonds are formed in spite of intimate interparticulate contactas indicated by the negligible residual porosities. Their highly elasticnature further aggravates the problem since the small number of bondscannot withstand the substantial elastic expansion during decompression.It appears therefore that, along with an understanding of the deformationmechanisms under high speed compaction as outlined in this work, it is alsonecessary to understand the factors controlling bond formation between thesurfaces newly created during compression.1574. SUMMARY1. A high speed rotary tablet press, the Manesty Betapress, was used toanalyse the powder compaction process while operating the press underspeed.2. After an initial calibration of the punch displacement profiles usingan LVDT-slip ring system, the punch displacements on the Betapress werecalculated using a relationship between machine deformation and punchforce. Thus the analysis of powder compaction on the Betapressrequired only the measurement of upper and lower punch forces, and theposition of the punches relative to the dead centre position where thepunches are vertically aligned with the centres of the compression rollsupport pins.3. Two types of punches, Manesty and IPT, were used at various machinespeeds to study the influence of these variables on powder compactioncharacteristics over a range of pressures.4. Several parameters from the compression phase were obtained and wererelated to particle deformation. These parameters included:i. Peak offset time (t0ff): the time interval by which the positionof the peak pressure preceded the dead centre position of thepunches. The t0ff is related to the ability of materials toundergo stress relaxation at constant strain during the time whenthe punch head flats are in contact with the compression rolls.ii. Decrease in pressure during t0ff (AP/VT): this is an indicationof the ease with which a compact material can flow into theresidual pore spaces after max is achieved.158iii. The work of compression (We): this was corrected for the work ofmachine deformation and gives the energy expended by the machinein compressing the powder bed during the compression phase.iv. Porosity change determined from the punch displacement analysis.The porosity change was used to calculate the maximum value ofthe Heckel term, Hmax, at max Heckel plots were constructed byplotting the values of Hmax against max These plots werelinear and their slopes were used to calculate the yield stress(Uy) of the material under pressure.5. A new method of estimating compact expansion from the small differencesbetween the decompression profiles of the various materials wasdeveloped by using the relationship between the machine deformation andforce as follows:i. The machine recovery was obtained by ‘compressing’ anincompressible solid, a ‘steel+Emcompress’ tablet. The forceapplied during the compression phase deforms only the machine.Hence the force during the decompression phase correspondsprimarily to the recovery of the press.ii. When a powder bed is compressed an additional force over andabove the force due to machine recovery is recorded duringdecompression. This force, which causes an additional machinedeformation during decompression, was obtained by subtracting theincompressible solid decompression profiles from the powder beddecompression profiles. The difference was translated intocompact expansion by using the machine deformation constant.159iii. The expansion was also used to obtain stress-strain plots, thelinear slopes of which provide a direct evidence that theexpansion of a compact during decompression is elastic in nature.Hence there is no need for complex viscoelastic modeling to drawthis conclusion.iv. The slopes of the stress-strain plots gave the elastic modulus ofthe tablets at a given porosity, E. Plotting E againstporosity and extrapolating these plots to zero porosity using apolynomial regression gave the Young’s modulus of the solidscompressed. This was a novel way of determining the Young’smodulus of solids using a high speed rotary press, without theneed of tests on single crystals or on large preformed compacts.v. The compact expansion was used to calculate the work ofdecompression (W0).6. Over forty pharmaceutical solids were selected to cover a wide range ofmaterials including direct compression excipients and their processedforms, and poorly compressible drugs and their formulations.7. The solids were characterized by determining true and bulk densities,and melting and/or decomposition temperatures. Based on DSC, powder Xray diffraction and a melting point-composition phase diagram,ibuprofen was found to be a ‘racemic compound’, and not a ‘racemicmixture’. The liSP description of ibuprofen as a ‘± mixture’ istherefore misleading.1608. All solids, including S-ibuprofen and racemic ibuprofen, werecompressed on the Betapress, and the various parameters mentioned abovewere obtained for each solid.9. The parameters from the compression phase were related to the abilityof the powder particles to deform permanently under pressure, and theparameters from the decompression phases were used to estimate theelastic behaviour of the solids under pressure.10. A combination of these parameters with rate of compaction profiles wasused to elucidate the deformation mechanisms of various categories ofsolids. These categories ranged from low yield-strength ductile solidssuch as acetylsalicylic acid, ibuprofen and their formulations to highyield-strength brittle solids such as the various calcium phosphates.11. The difference between W and W0, i.e., the work lost in irreversibleprocesses during compression is an indication of the energy stored asbonds in the tablets, and correlated reasonably well with the force offailure of tablets in a diametral compression test.12. The lost work WC-WD also correlated well with porosity change duringcompression. Materials with large porosity change, or high W-W,produced strong tablets, e.g., the microcrystalline celluloses relativeto lactoses or calcium phosphates.13. The lost work WC-WD was also correlated with the homologoustemperatures of the various solids. Materials with a high homologoustemperature had a low WC-WD indicating that solids are more easilycompressed at temperatures close to their melting points.16114. The processed forms of the direct compression agents and directcompression formulations of poorly compressible drugs formed strongertablets relative to the parent substances. The improvements in tabletstrength were reflected in the parameters obtained from the compressionand decompression phases.15. S-ibuprofen and racemic ibuprofen have similar tableting properties,hence the decision to replace racemic ibuprofen with S-ibuprofen incommercial formulations should be based on the differences in theirsolubility and pharmacokinetics.16. A simple and inexpensive system of analysis of powder compaction usingonly the force and time measurements on a rotary tablet press has beendeveloped. This system is potentially applicable to any rotary press.Since the method uses data obtained under normal operating conditions,it can be used for the in-process monitoring of high speed compactionusing all the tooling stations, provided the compaction cycles do notoverlap. On the Betapress, overlapping seemed likely to be a problemonly with fluffy materials such as the microcrystalline celluloseswhen, to obtain high max’ the die was completely filled.17. The method can be used for fundamental research on deformationbehaviour, for the quality control of materials, for the development oftablet formulations and for the in-process monitoring of compaction.1625. REFERENCESAgbada, CO. and York, P. (1990) Theophylline hydrate/anhydrous system:effects of water of hydration on mechanical properties of compactedbeams. J. Pharm. Pharmacol. 42(Suppl.): 76P.Ahn, H.-Y., Amidon, G.L. and Smith, D.E., (1991) Stereoselective systemicdisposition of ibuprofen enantiomers in the dog. Pharm. Res.8: 1186-1190.Amidon, G.E., Smith, D.P. and Hiestand, E.N. (1981) Rotary pressutilizing a flexible die wall. J. Pharm. Sd. 70: 613-617.Atkins, A.G. and Mai, Y.W. (1985) Elastic and Plastic Fracture: Metals,Polymers, Ceramics, Composites, Biological Materials. Ellis HorwoodLtd., Chichester, West Sussex. pp. 798-800.Aulton, M.E., Travers, D.N., and White, P.J.P. (1973) Strain recovery ofcompacts on extended storage. J. Pharm. Pharmacol. 25(Suppl.): 79P-86PBangudu, A.B. and Pilpel, N. (1985) Effects of composition and stearicacid on the plasto-elasticity and tableting of paracetamolmicrocrystalline cellulose mixtures. Ibid. 37: 289-293.Bassam, F., York , P., Rowe, R.C. and Roberts, R.J. (1988) Effect ofparticle size and source on variability of Young’s modulus ofmicrocrystalline cellulose powders. Ibid. 40(Suppl .): 68P.Bateman, S. (1988) High speed compression simulators in tablettingresearch. Pharm. J. May: 632-633.Bateman, S., Rubinstein, M.H., Rowe, R.C., Roberts, R.J., Drew, P. andHo, A.Y.K. (1989) A comparative investigation of compressionsimulators. mt. J. Pharm. 49: 209-212.Beck, W.S., Geisslinger, G., Engler, H. and Brune, K., (1991)Pharmacokinetics of ibuprofen enantiomers in dogs. Chirality 3: 165-169.Bettinetti, G., Giordano, F., Fronza, G., Italia, A., Pellegata, R., Villa,M. and Ventura, P., (1990) Sobrerol enantiomers and racemates: solidstate spectroscopy, thermal behavior, and phase diagrams.J. Pharm. Sd. 79: 470-475.163Bland, D.R. (1960) The theory of linear viscoelasticity. Pergamon Press,New York, N.Y.Bridgman, P.W. (1949) Linear compression to 30,000 kg/cm2, includingrelatively incompressible substances. Proc. Am. Soc. Arts. Sci.77: 187-234.Carroll, R.J. and Ruppert, 0. (1988) Transformation and weighting inregression. Chapman and Hall, New York. pp 29-30.Caspar, U. and Muller, F. (1984) Viskoelastizitat von Tabletten beiStatischer Belastung, Pharm. Acta Helv. 59: 329-334.Celik, M. and Marshall, K. (1989) Use of a compaction simulator system intabletting research. I. Introduction to and initial experiments with thesystem. Drug. 0ev. md. Pharm. 15: 759-800.Celik, M. and Travers, D.N. (1985) The use of an elastic recovery index asa criterion of compactional behaviour of some directly compressionbases, Ibid. 11: 299-314.Charlton, B. and Newton, J.M. (1984) Theoretical estimation of punchvelocities and displacements of single-punch and rotary tablet machines.J. Pharm. Pharmacol. 36: 645-651.Church, M.S. (1984) Mechanical characterization of pharmaceutical compacts,Ph.D. Thesis, Nottingham.Church, M.S. and Kennerley, J.W. (1982) Flexure testing of compactedrectangular beams. J. Pharm. Pharmacol. 34(Suppl.): 50P.Church, M.S. and Kennerley, J.W. (1983) A comparison of the mechanicalproperties of pharmaceutical materials obtained by the flexure testingof compacted rectangular beams. Ibid. 35(Suppl): 43P.Cole, E.T., Rees, J.E and Hersey, J.A. (1971) Preliminary compactionstudies using a device to simulate a rotary tabletting machine. Ibid.23(Suppl): 258S.Cole, E.T., Rees, J.E and Hersey, J.A. (1975) Relations between compactiondata for some crystalline pharmaceutical materials. Pharm. Acta Helv.50: 28-32.164Cook, G.D., Duffield, P.J. and Oliver, A. (1988) Mounting an inductivedisplacement transducer on a single punch tablet machine. J. Pharm.Pharmacol. 40(Suppl.): 119P.Courtney, T.H. (1990) Mechanical behaviour of materials. McGraw-Hill, Inc.McGraw-Hill Series in Materials Science and Engineering. Eds. M.B.Bever and C.A. Wert. New York, U.S.A. p. 8, 16, 382.Cutt, T., Fell, J.T., Rue, P.J. and Spring, M.S. (1987) Granulation andcompaction of a model system II. Stress Relaxation. mt. J. Pharm.39: 157-161.Danielson, D.W., Morehead, W.L. and Rippie, E.G. (1983) Unloading andpostcompression viscoelastic stress versus strain behavior ofpharmaceutical solids J Pharm Sci 72 342-345David, S.T. and Augsburger, L.L. (1974) Flexure test for determination oftablet tensile strength. Ibid 63: 933-936.David, ST. and Augsburger, L.L. (1977) Plastic flow during compression ofdirectly compressible fillers and its effect on tablet strength. Ibid66: 155-159.Doroudian, A. (1991) Determination of compaction parameters ofpharmaceutical powders with an instrumented hydraulic press. M.Sc.Thesis, The University of British Columbia, Vancouver, B.C., Canada.Draper, N.R. and Smith, H. (1966) Applied Regression Analysis. John Wiley &Sons, Inc., New York. pp 88-90.Duberg, M. and Nystrom, C. (1986) Studies on direct compression oftablets XVII. Porosity-pressure curves for the characterization ofvolume reduction mechanisms in powder compression. Powder Technol. 46:67-75.Duncan-Hewitt, W.C. and Grant, D.J.W. (1987a) Use of Hiestand’s indicesof tableting performance and the impact fracture wear test to evaluatethe mechanical behaviour of labetalol.HC1 compacts. Pharm. Res. 4: 22s.Duncan-Hewitt, W.C. and Grant, D.J.W. (1987b) The impact fracture weartest: a novel method of tablet evaluation. Powder Technol. 52: 17-28.165Duncan-Hewitt, W.C. (1988) The use of microindentation techniques toassess the ability of pharmaceutical crystals to form strong compacts.Ph.D. Thesis, University of Toronto, Toronto, Canada.Duncan-Hewitt, W.C. and Weatherly, G.C. (1989a) Evaluating thedeformation kinetics of sucrose crystals using microindentationtechniques. Pharm. Res. 6: 1060-1066.Duncan-Hewitt, W.C. and Weatherly, G.C. (1989b) Evaluating the hardness,Young’s modulus, and fracture toughness of some pharmaceutical crystalsusing microindentation techniques. J. Mat. Sd. Letters 8: 1350-1352.Duncan-Hewitt, W.C. and Weatherly, G.C. (1990a) Modeling the uniaxialcompaction of pharmaceutical powders using the mechanical properties ofsingle crystals. I. Ductile materials. J. Pharm. Sci. 79: 147-152.Duncan-Hewitt, W.C. and Weatherly, G.C. (1990b) Modeling the uniaxialcompaction of pharmaceutical powders using the mechanical properties ofsingle crystals. I. Brittle materials. Ibid 79: 273-278.Dwivedi, S.K. (1988) Anomeric composition and solid state properties oflactose. M.Sc. Thesis, The University of British Columbia, Vancouver,B.C.Dwivedi, S.K. and Mitchell, A.G. (1989) Gas-chromatographic analysis ofanomeric composition of lactose. J. Pharm. Sd. 78: 1055-1056.Ertel, K.D., Heasley, R.A., Koegel, C., Chakrabarti, A. and Carstensen,J.T. (1990) Determination of ibuprofen vapor pressure at temperaturesof pharmaceutical interest. Ibid. 79: 552.Fell, J.T. and Newton, J.M. (1968) The tensile srength of lactosetablets. Ibid. 20: 657-659.Fell, J.T and Newton, J.M. (1970) Determination of tablet strength by thediametral-compression test. Ibid 69: 688-691.Ferry, J.D. (1980) Viscoelastic properties of polymers. John Wiley andSons, Inc., New York, N.Y.Flugge, W. (1967) Viscoelasticity. A Blaisdell Book in Solid Mechanics,Blaisdell Publishing Co., Waltham, Massachusetts.166Garr, J.S.M. and Rubinstein, MH. (1991) Compaction properties of acellulose-lactose direct-compression excipient. Pharm. Technol. mt. 3:76, 77, 80.Guyot, J.C., Delacourte, A., and Marie, B. (1986) Computer determinationand comparison of the compression behaviour of powder mixtures. DrugDev. md. Pharm. 12: 1869-1884.Handbook of Pharmaceutical Excipients (1986) published by AmericanPharmaceutical Association, Washington, DC, USA and The PharmaceuticalSociety of Great Britain, London, England. p. 285, 309.Heckel, R.W. (1961) Density-pressure relationships in powder compaction.Trans. Metall. Soc. A.I.M.E. 221: 671-675.Hersey, J.A. and Rees, J.E. (1971) Deformation of particles duringbriquetting. Nature 230: 96.Hiestand, E.N. (1985) Dispersion forces and plastic deformation in tabletbond. J. Pharm. Sd. 74: 768-770.Hiestand, E.N., Bane, J.M., Jr. and Strzelinski, E.P. (1971) Impact testfor hardness of compressed powder tablets. J. Pharm. Pharmacol.60: 758-763.Hiestand, E.N. and Smith, D.P. (1984) Indices of tableting performance.Powder Technol. 38: 145-149.Hiestand, E.N., Wells, J.E., Peot, C.B. and Ochs, J.F. (1977) Physicalprocess of tableting. J. Pharm. Sd. 66: 510-519.Ho, A., Barker, J.F., Spence, J. and Jones, T.M. (1979) A comparison ofthree methods of mounting a linear variable displacement transducer onan instrumented tablet machine. J. Pharm. Pharmacol. 31: 471-472.Ho, A.Y.K. and Jones, T.M. (1988a) Punch travel beyond peak force duringtablet compression. Ibid 40S: 75P.Ho, A.Y.K. and Jones, T.M. (1988b) Rise time: a new index of tabletcompression. Ibid 40S: 74P.Hunter, B.M., Fisher, D.G., Pratt, R.M. and Rowe, R.C. (1976) A highspeed compression simulator. Ibid 28(Suppl.): 65P.167Hutt, A.J. (1991) Drug chirality: impact on pharmaceutical regulation.Chirality, 3: 161-164.Jamali, F., Singh, N.N., Pasutto, F.M., Russel, A.S. and Coutts, R.T.,(1988) Pharmacokinetics of ibuprofen enantiomers in man following oraladministration of tablets with different absorption rates. Pharm. Res.,5: 40-43.Jetzer, W.E., Leuenberger, H. and Sucker, H. (1983) The compressibility andcompactibility of pharmaceutical powders. Pharm. Technol. 7: 33-39.Jones, T.M. (1977) Formulation factors: Drugs given by oral route. InFormulation and Preparation of Dosage Forms: Proceedings of the 37thInternational Congress of Pharmaceutical Sciences of F.I.P. Eds. J.Polderman. The Hague, The Netherlands. Elsevier/North Holland,Amsterdam, The Netherlands. pp. 29-44.Jones, T.M. (1978) Preformulation studies to predict the compactionproperties of materials used in tablets and capsules. Acta Pharm.Technol. Supplement 6: 141-159.Juslin, M.J. and Paronen, T.P. (1980) On the accuracy of displacementmeasurements by instrumented single-punch machines. J. Pharm.Pharmacol. 32: 796-798.Kaneniwa, N., Imagawa, K. and Otsuka, M. (1984) Compression properties ofcephalexin powder and physical properties of the tablet. Chem. Pharm.Bull. 32: 4986-4993.Kendall, K. (1978) The impossibility of comminuting small particles bycompression. Nature 272: 710-711.Kerridge, J.C. and Newton, J.M. (1986) The determination of thecompressive Young’s modulus of pharmaceutical materials. J. Pharm.Pharmacol. 38(Suppl.): 79P.Krycer, I., Pope, D.G. and Hersey, J.A. (1982a) An evaluation of thetechniques employed to investigate powder compaction behaviour. mt. J.Pharm. 12: 113-134.168Krycer, I., Pope, D.G. and Hersey, J.A. (1982b) The interpretation ofpowder compaction data- a critical review. Drug. Dev. md. Pharm. 8:307-342.Kussendrager, K., De Hoog, P. and Van Leverink, J. (1981) Some physicalproperties of spray-dried lactose with respect to stability,compression. Acta Pharm. Suec. 18: 94-95.Lammens, R.F., Polderman, J., De Blaey, C.J. and Armstrong, N.A. (1980)Evaluation of force-displacement measurements during powder compaction.Part II: Precision and accuracy of powder height and punch displacementmeasurements. mt. J. Pharm. Tech. & Prod. Mfr. 1: 26-35.Lawn, B.R. and Wilshaw, T.R. (1975) Fracture of Brittle Solids. CambridgeUniversity Press, Cambridge, London. p 77.MacLeod, H.M. (1983) Compaction of ceramics. In: Enlargement andCompaction of Particulate Solids. Ed. N.G. Stanley-Wood. ButterworthsMonographs in Chemical Engineering. Series ed. J.W. Mullin.Butterworths & Co, London, England. P. 259.Malamataris, S., Bin Baie, S. and Pilpel, N. (1984) Plasto-elasticity andtableting of paracetamol, Avicel and other powders. J. Pharm.Pharmacol. 36: 616-617.Marsh, D.M. (1964) Plastic flow in glass. Proc. R. Soc. London.A279: 420-435.Marshall, K. (1986) Compression and consolidation of powdered solids. InThe Theory and Practice of Industrial Pharmacy, L. Lachman, H.A.Lieberman, J.L. Kanig, Eds., Lea & Febiger, Philadelphia, PA, p. 66.Marshall, K. (1989) Monitoring punch forces and punch movements as an aidto developing robust tablet formulations. Drug. Dev. md. Pharm.15: 2153-2176.Mashadi, A.B. and Newton. J.M. (1987) The characterization of themechanical properties of microcrystalline cellulose: a fracturemechanics approach. J. Pharm. Pharmacol. 39: 961-965.Morehead, W.T. and Rippie, E.G. (1990) Timing relationships among maximaof punch and die-wall stress and punch displacement during compaction ofviscoelastic solids. J. Pharm. Sci. 79: 1020-1022.169Morita, M., Nakai, Y., Fukuoka, E. and Nakajima, S. (1984) Physicochemicalproperties of crystalline lactose. II. Effect of crystallinity onmechanical and structural properties. Chem. Pharm. Bull. 32: 4076-4083.Oates, R.J. and Mitchell, A.G. (1989) Calculation of punch displacementand work of powder compaction on a rotary tablet press. J. Pharm.Pharmacol. 41: 517-523.Oates, R.J. and Mitchell, A.G. (1990) Comparison of calculated andexperimentally determined punch displacement on a rotary tablet pressusing both Manesty and IPT punches. Ibid. 42: 388-396.Okada, J. and Fukumori, Y. (1975) Compaction of powders. II. Comparisonof compaction behaviours among various powders, Chem. Pharm. Bull., 23:326-332.Parrott, E.L. (1985) Compression. In Pharmaceutical Dosage Forms:Tablets. H.A. Lieberman, L. Lachman, Eds., vol. 2., Marcell Dekker,Inc., New York, N.Y.Peleg, M. and Moreyra, R. (1979) Effect of moisture on the stressrelaxation pattern of compacted powders. Powder Technol. 23: 277-279.Pitre, D. and Stradi, R., (1990) Racemic modification of (R,S)-3-(4-Phenyl-1-piperazinyl)-1,2-propandiol and melting point diagram.Arch. Pharm. (Weinheim), 323: 23-25.Popov, E.P. (1968) Introduction to mechanics of solids. Prentice-Hall, Inc.Englewood Cliffs, N.J., pp. 116-120, p 554.Prigogine, I. and Defay, R. (1954) Chemical Thermodynamics, D.H. Everett,translator, Longman Group Limited, London, pp. 373-375.Ragnarsson, G. (1985) Evaluation of tabletting properties in preformulationand early formulation work. Acta Pharm. Suec. 22: 173Ragnarsson, G. and Sjorgen, J. Force-displacement measurements intableting. J. Pharm. Pharmacol. 37: 145-150.Rees, J.E., Hersey, J.A. and Cole, E.T. (1972) Simulation devices forpreliminary tablet compression studies. J. Pharm. Sd. 61: 1313-1315.170Rees, J.E. and Rue, P.J. (1978) Time-dependent deformation of some directcompression excipients. J. Pharm. Pharmacol. 30: 601-607.Rhone-Poulenc Product Information Manual (1989) Calcium phosphatepharmaceutical ingredients. Rhone-Poulenc Basic Chemical Co., Shelton,CT.Rhone-Poulenc Product Information Manual (1989) Calcium PhosphatePharmaceutical Ingredients, Rhone-Poulenc Basic Chemicals Co., Shelton,CT, USA, p. 11.Ridgway Watt, P. (1983) Tablet press instrumentation. Mnfg. Chem. 54:42-43, 45.Ridgway Watt, P. (1988) Tablet Machine Instrumentation in Pharmaceutics.Ellis Horwood Ltd., Chichester, West Sussex. pp 305-309.Ridgway Watt, P. and Rue, P.J. (1979) The design and construction of afully instrumented rotary tablet machine. International Conference onPharmaceutical Technology and Product Manufacture, Copenhagen.Ridgway, K., Shotton, E. and Glasby, J. (1969) The hardness and elasticmodulus of some crystalline pharmaceutical materials. J. Pharm.Pharmacol. 21(Suppl.): 19S-23S.Rippie, E.G. and Danielson, D.W. (1981) Viscoelastic stress/strainbehavior of pharmaceutical tablets: analysis during unloading and postcompression periods. J. Pharm. Sci. 70: 476-482.Ritter, A. and Sucker, H.B. (1980) Studies of variables that affect tabletcapping. Pharm. Technol. 4: 56-65, 128.Roberts, R.J. and Rowe, R.C. (1985) The effect of punch velocity on thecompaction of a variety of materials. J. Pharm. Pharmacol. 37:3 77-384.Roberts, R.J. and Rowe, R.C. (1987) Brittle/ductile behaviour inpharmaceutical materials used in tabletting. mt. J. Pharm. 36: 205-209.Roberts, R.J. and Rowe, R.C. (1987a) The Young’s modulus of pharmaceuticalmaterials. Ibid 37: 15-18.171Roberts, R.J. and Rowe, R.C. (1987b) The compaction of pharmaceutical andother model materials- a pragmatic approach. Chem. Eng. Sci. 42:903 -911.Rossi, R.C. (1968) Prediction of the elastic moduli of composites. J. Am.Ceram. Soc. 51: 433-439.Seugling, E.W. (1980) The contribution of tableting aids topharmaceutical development. Pharm. Technol. 4: 27-35, 62.Shianta, S. and Milosovich, G. (1964) Compression of pharmaceuticalpowders I. Theory and instrumentation. J. Pharm. Sd. 53: 562-564.Simmons, G. and Wang, H. (1971) Single Crystal Elastic Constants andCalculated Aggregate Properties: A Handbook. The M.I.T. Press,Cambridge, Massachusetts. p 271.Spriggs, R.M. (1961) Expression for effect of porosity on elastic modulusof polycrystalline refractory materials, particularly aluminum oxide.J. Am. Ceram. Soc. 44: 628-629.Staniforth, J.N., Baichwal, A.R. and Hart, J.P. (1987) Interpretation ofcreep behaviour of microcrystalline cellulose powders and granulesduring compaction, mt. J. Pharm. 40: 267-269.Swartz, C.J. (1969) Development of IPT standards for tableting tools.J. Pharm. Sci. 58: 473-477.Talman, F.A.J. (1977) Formulation. In Bentley’s Textbook of PharmaceuticsEd. E.A. Rawlins. Bailliere Tindall, London, England. p. 644.The Merck Index (1976) Starch. M. Windholz, Ed., 9th ed., Merck & Co.,Inc., Rahway, N.J., p. 8574.Toure, P., Puisieux, F., Duchene, D. and Carstensen, J.T. (1980) Energyterms in tablet compression cycles. Powder Technol. 26: 213-216.Train, D. (1956) An investigation into the compaction of powders.J. Pharm. Pharmacol. 8: 745-760.Travers, D.N., Celik, M. and Buttery, T.C. (1983) A computer aidedinvestigation on strain movements in compacts under constant stresswithin the die. Drug. Dev. Ind. Pharm. 9: 139-157.172USP XXII (1990) Ibuprofen, Official Monographs. p. 682.Vromans, H., Boihuis, G.K., Lerk, C.F., Kussendrager, K.D. and Bosch, H.(1986) Studies on tableting properties of lactose. VI. Consolidationand compaction of spray dried amorphous lactose. Acta Pharm. Suec. 23:23 1-240.Wachtman, Jr., J.B. (1969) Elastic deformation of ceramics and otherrefractory materials. In Wachtman Jr., J.B. (Ed.), Mechanical andThermal Properties of Ceramics, National Bureau of Standards SpecialPublication 303. National Bureau of Standards, Washington, D.C.pp 139-168.Walter, J.T. and Augsburger, L. (1986) A computerized force-displacementinstrumentation system for a rotary press. Pharm. Technol. 10: 26-34.Wong, D.Y.T. and Aulton, M.E. (1989) Modification of the mechanicalproperties of alpha lactose monohydrate crystals by adjustment of growthconditions. J. Pharm. Pharmacol. 41(Suppl.) 124P.York, P. (1979) A consideration of experimental variables in the analysisof powder compaction behaviour. Ibid 31: 244-246.York, P. and Baily, ED. (1977) Dimensional changes of compacts aftercompression. Ibid. 29: 70-74.Yu, H.C.M., Rubinstein, M.H., Jackson, I.M. and Elsabbagh, H.M. (1988)Multiple compression and plasto-elastic behaviour of paracetamol andmicrocrystalline cellulose. Ibid. 40: 669-673.Zar, J. H. (1984) Biostatistical Analysis. Prentice-Hall, Inc., EnglewoodCliffs, New Jersey. pp. 268, 285-289.

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0086553/manifest

Comment

Related Items