European Journal of Pharmaceutical Sciences 21 (2004) 525–531

Effects of powder particle size and binder viscosity on intergranular andintragranular particle size heterogeneity during high shear granulation

Torben Schæfer∗, Dorrit Johnsen, Anita Johansen

Department of Pharmaceutics, The Danish University of Pharmaceutical Sciences, 2 Universitetsparken, DK-2100 Copenhagen, Denmark

Received 28 August 2003; received in revised form 6 November 2003; accepted 1 December 2003

Abstract

A study was performed in order to elucidate the effects of powder particle size and binder viscosity on intergranular and intragranularparticle size heterogeneities. Granules were produced by melt granulation in a high shear mixer from each of four calcium carbonates havingmean particle sizes in the range of 5.5–63.1�m. Each of three polyethylene glycols (PEGs) having viscosities in the range of approximately40–14000 mPa s were applied as meltable binders. The size distribution of the calcium carbonate particles in three granule size fractions(125–250, 355–500, and 800–1000�m) was measured after disintegration of the granules. Intragranular particle size heterogeneities wereevaluated qualitatively by means of scanning electron microscopy. A preferential growth of the smaller particles was found to give rise toa higher content of small particles in large granules when calcium carbonates with mean particle sizes of 11.7, 34.5, and 63.1�m weregranulated with a binder of low viscosity. The use of a binder of medium or high viscosity leads to a marked reduction of these heterogeneities.A preferential growth of larger particles was seen when calcium carbonates with mean particle sizes of 5.5 and 11.7�m were granulated witha highly viscous binder. The use of a binder with low or medium viscosity resulted in an increased homogeneity. Intragranular particle sizeheterogeneities were primarily seen when 5.5 and 11.7�m calcium carbonate particles were granulated with a highly viscous binder.© 2004 Elsevier B.V. All rights reserved.

Keywords: Powder particle size; Binder viscosity; Melt granulation; Particle size distribution; Particle size heterogeneity; Granule growth mechanisms

1. Introduction

Segregation of multi-component mixtures during granu-lation is a common problem, which is of particular concernto the pharmaceutical industry (Iveson et al., 2001). When amixture of a drug and one or more excipients is granulated,large differences in the drug content of different granulesize fractions were found (Ojile et al., 1982; Egermann andReiss, 1988; Vromans et al., 1999; van den Dries andVromans, 2002). Demixing during handling and process-ing of such granules might give problems in meeting thedemands on the content uniformity of the final dosage form.

A systematic approach to granule inhomogeneity mightbe to relate the problems to the three fundamental agglom-eration mechanisms described byIveson et al. (2001): (i)wetting and nucleation, (ii) consolidation and growth by co-alescence, and (iii) attrition and breakage. Poor wetting willcause weak liquid bonds between the primary particles. This

∗ Corresponding author. Tel.:+45-35-306-000; fax:+45-35-306-031.E-mail address: ts@dfuni.dk (T. Schæfer).

might result in intergranular heterogeneity if a mixture of ahydrophobic drug and hydrophilic excipients is granulated(Crooks and Schade, 1978; Miyamoto et al., 1998). A vis-cous binder liquid is difficult to disperse uniformly in thepowder, and this might delay or counteract a nucleation ofthe primary particles (Iveson et al., 2001). If the initial nu-clei are strong enough to resist the shearing forces involvedin the process, they will gain further strength by consoli-dation and will grow in size by coalescence. Breakage willoccur in the wet state of the process if nuclei or granules aretoo weak to survive, whereas attrition is related to handlingof dried granules. If attrition occurs after a migration of asoluble drug to the granule surface during drying (Selkirk,1976), the granulation will contain fines with a high drugcontent.

According to van den Dries and Vromans (2002), thetensile strength of a granule under dynamic conditions isdirectly proportional to the viscosity of the binder liquidand inversely proportional to the particle size of the pri-mary particles and to the intragranular porosity. This meansthat primary particle size and binder viscosity are essential

0928-0987/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.ejps.2003.12.002

526 T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531

formulation variables that might determine whether coa-lescence or breakage will be the dominant agglomerationmechanism. Particle size and binder viscosity were foundto interact since a low viscosity resulted in sufficientlystrong granules when the particle size was small, whereas ahigher viscosity was necessary in order to obtain a strengthbeing sufficient to avoid breakage at a larger particle size(Keningley et al., 1997; Johansen and Schæfer, 2001).

Since smaller particles cause stronger granules, a prefer-ential growth of small particles might occur giving rise toa higher content of small particles in the largest granules(Scott et al., 2000). This might explain why a higher drugcontent was found in the larger granule size fractions whena drug with a small particle size was granulated with excip-ients of a larger particle size (Egermann and Reiss, 1988;Vromans et al., 1999; van den Dries and Vromans, 2002).A very large particle size difference between drug and ex-cipient was found to prevent demixing (van den Dries andVromans, 2002). This was explained by a continuous break-age and coalescence preventing a preferential growth of thesmall particles. A different explanation might be that the mi-cronized drug particles adhere to the surface of the large ex-cipient particles forming an ordered mixture, which preventsdemixing (Yip and Hersey, 1977; McGinity et al., 1985).The role of breakage in preventing demixing seems to becomplex since an increased demixing at a higher impellerspeed in a high shear granulator was ascribed to breakage(Vromans et al., 1999). A different mechanism of demixingis a preferential layering of small primary particles onto thesurface of existing granules (Tardos et al., 1997; van denDries and Vromans, 2002).

In order to study the effect of primary particle size ondemixing, it will be more useful to investigate differencesin the size distribution of the primary particles in differentgranule size fractions instead of differences in drug con-tent. A demixing of small and large drug particles mightalso contribute to intergranular heterogeneity. When study-ing demixing of a mixture of drug and excipient it might bedifficult to distinguish the effect of a particle size differencefrom possible effects of differences in wettability and solu-bility. It might be more convenient, therefore, to use a singlemodel substance when particle size effects have to be stud-ied. Scott et al. (2000)granulated calcium carbonate withpolyethylene glycol (PEG) and measured the size distribu-tion of the primary calcium carbonate particles in differentgranule size fractions after dissolving the PEG in water.

The present study is based on previous experiments inwhich calcium carbonates of different particle sizes weremelt granulated with PEGs of different viscosities (Johansenand Schæfer, 2001). The size distributions of the primaryparticles in selected granule size fractions are measured. Theaim of this study is to investigate whether intergranular par-ticle size heterogeneities can be reduced by a proper com-bination of particle size and binder viscosity and to relateintergranular heterogeneities to the agglomeration mecha-nisms.

2. Materials and methods

2.1. Materials

Four different grades of calcium carbonate powders withdifferent mean particle size (Durcals 5, 10, 40, and 65,Omya, France) were used as starting material. Polyethy-lene glycol 1500, 6000, or 20000 (Clariant, Germany) wasused as meltable binder. The PEGs were used as flakes.The calcium carbonates as well as the PEGs were identi-cal to those applied in previous experiments (Johansen andSchæfer, 2001).

The size distribution by volume of the calcium carbonateswas determined by a Malvern Mastersizer S laser diffrac-tion particle sizer (Malvern Instruments, UK). Before themeasurement, the calcium carbonate was dispersed in a fil-tered saturated aqueous calcium carbonate solution contain-ing 7% (m/v) of povidone (Kollidon® 90, BASF, Germany)and agitated by a magnetic stirrer. A sample of the disper-sion was taken with a pipette and added to the small vol-ume sample preparation unit containing the same dispersionmedium. The sample as well as the dispersion medium wastreated with ultrasound for 20 min before the measurement.The povidone was added to the dispersion medium in or-der to increase the viscosity. Sedimentation of the largestparticles was observed when a dispersion medium of lowerviscosity was used.

2.2. Granule production

The granules used in the present investigation were pro-duced in an 8 l Pellmix PL 1/8 laboratory scale high shearmixer (Niro, Denmark) at an impeller speed of 800 rpm(Johansen and Schæfer, 2001).

2.3. Granule characterisation

The granulations were previously characterized by sizedistribution, intragranular porosity, scanning electron mi-croscopy, and binder distribution in size fractions (Johansenand Schæfer, 2001).

In the present investigation, the size distribution by vol-ume of the primary particles of calcium carbonate withingranules of different size fractions was determined by dis-persing a sample of 1 g of the granule size fraction in 50 mlof the above-mentioned dispersion medium. After completedissolution of the PEG causing a disintegration of the gran-ules, the dispersion was agitated with a magnetic stirrer andtreated with ultrasound for 20 min. Then the particle sizedistribution was measured according to the procedure de-scribed inSection 2.1. The measurements were carried outin duplicate.

Further scanning electron micrographs of selected sizefractions of the granules were taken by an environmentalscanning electron microscope (SEM) (Philips XL30ESEM,the Netherlands) using backscattered electron emission.

T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531 527

2.4. Experimental design

Particle size distributions of the primary particles of cal-cium carbonate within granule size fractions of 125–250,355–500, and 800–1000�m were measured for granulesproduced from Durcal 5, 10, 40, and 65 with each of thethree types of PEG mentioned inSection 2.1. All data onparticle size distributions within granule size fractions aremean values of two repeated experiments.

Intragranular particle size heterogeneities were evaluatedfrom SEM micrographs of size fractions of selected granu-lations. The selection included granulations produced withthe four particle size grades of calcium carbonate as well asthe three viscosity grades of PEG.

3. Results and discussion

3.1. Raw material properties

The mean particle sizes,D (v; 0.5), of Durcal 5, 10, 40,and 65 were measured to be 5.5 (0.3), 11.7 (0.9), 34.5 (0.8),and 63.1 (0.7)�m, respectively, the values between paren-theses showing standard deviations of three repeated mea-surements. The size distributions of the Durcals are shownin Figs. 1–4.

Data on true density, specific surface area, and interpartic-ular porosity of the Durcals and particle size, density, melt-ing range, and viscosity of the PEGs were presented in aprevious paper (Johansen and Schæfer, 2001). The viscos-ity at 90◦C was found to be 38 mPa s for the PEG 1500,454 mPa s for the PEG 6000, and 13955 mPa s for the PEG20000.

3.2. Intergranular heterogeneity

Table 1shows the median particle size of the primary par-ticles in different size fractions. The reproducibility of theparticle size measurement is generally good. The size frac-tions were chosen in order to represent granules of a small,medium and large size. Further, the choice was limited tosize fractions, which were present in an amount being suffi-cient for a particle size measurement for all combinations ofDurcals and PEGs. For most of the combinations, the me-dian particle size is seen to be clearly different in differentsize fractions indicating an intergranular particle size het-erogeneity. In order to elucidate the reasons for these het-erogeneities, an evaluation of the particle size distributionsin different size fractions is necessary (Figs. 1–4).

For the Durcal with the largest particle size (Durcal 65),the smallest median particle size is observed in the largestgranule size fraction for all the PEGs (Table 1). This in-dicates a preferential growth of the smaller Durcal parti-cles. Accordingly, the size distributions (Fig. 1) show anincreased content of small particles and a markedly lowercontent of large particles in the 800–1000�m size fraction.

0

2

4

6

8

10

12

14

01 0.1 1 10 100 10

% v

/v

0

2

4

6

8

10

12

14

0.01 0.1 1 10 100 1000

% v

/v

0

2

4

6

8

10

12

14

0.01 0.1 1 10 100 1000

Particle diameter, µm

% v

/vPEG 20 000

PEG 6000

PEG 1500

Fig. 1. Particle size distributions of the primary particles of Durcal 65in selected size fractions of granules produced with PEGs of differentmolecular mass: (×) ungranulated Durcal 65. Granule size fractions: (�)125–250�m, (�) 355–500�m, and (�) 800–1000�m.

Table 1Median particle size,D (v; 0.5) (�m), of the primary calcium carbonate(Durcal) particles in selected size fractions of granules produced fromDurcals of different particle size with PEGs of different molecular mass

Sizefraction(�m)

PEG 1500 PEG 6000 PEG 20000

Durcal 5 125–250 5.3 (±0.3) 4.3 (±0.0) 3.4 (±0.2)355–500 5.2 (±0.2) 5.6 (±0.3) 4.7 (±0.3)800–1000 6.2 (±0.1) 6.7 (±0.5) 7.6 (±0.1)

Durcal 10 125–250 27.9 (±0.5) 10.6 (±0.8) 10.0 (±0.1)355–500 11.0 (±0.2) 9.7 (±0.3) 8.2 (±0.1)800–1000 10.0 (±0.2) 12.1 (±0.0) 13.6 (±0.2)

Durcal 40 125–250 68.5 (±0.8) 41.7 (±1.7) 30.6 (±0.8)355–500 49.9 (±0.4) 33.0 (±0.9) 30.7 (±0.1)800–1000 20.0 (±1.0) 28.7 (±0.3) 31.3 (±0.4)

Durcal 65 125–250 87.4 (±0.8) 95.0 (±0.9) 170.7 (±1.9)355–500 83.8 (±0.5) 90.8 (±0.0) 65.7 (±0.1)800–1000 29.6 (±0.4) 45.7 (±1.0) 56.9 (±0.1)

The values between parentheses are the range of two repeated experiments.

528 T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531

0

2

4

6

8

10

12

0.01 0.1 1 10 100 1000

% v

/v

0

2

4

6

8

10

12

0.01 0.1 1 10 100 100

% v

/v

0

2

4

6

8

10

12

0.01 0.1 1 10 100 1000Particle diameter, µm

% v

/v

PEG 20 000

PEG 6000

PEG 1500

Fig. 2. Particle size distributions of the primary particles of Durcal 40in selected size fractions of granules produced with PEGs of differentmolecular mass: (×) ungranulated Durcal 40. Granule size fractions: (�)125–250�m, (�) 355–500�m, and (�) 800–1000�m.

This is more pronounced with the decrease of the viscosityof the PEG.

Due to the large particle size of Durcal 65, the granuleswill be rather weak, and breakage is likely to occur. Thiswas seen with PEG 1500 as well as PEG 6000 (Johansenand Schæfer, 2001). The smaller Durcal particles result instronger granules. It is more likely, therefore, that granulescontaining small particles will resist the impact forces ex-erted by the rotating impeller. This explains the preferentialgrowth of the smaller particles.

A higher binder viscosity will increase the granulestrength. Consequently, PEG 20000 will be able to granulatelarger primary particles than will PEG 6000, which againwill be able to granulate larger particles than PEG 1500.Thus, a higher viscosity reduces the particle size hetero-geneity. PEG 20000 is still unable to granulate the largestparticles. The granule size fraction 125–250�m, therefore,is supposed primarily to consist of ungranulated Durcalparticles of a very large size. This granule size fraction,however, contributes scarcely to particle size heterogeneitysince it makes up only 0.2% of the granulate. With PEG

0

2

4

6

8

10

12

0.01 0.1 1 10 100 1000

% v

/v

0

2

4

6

8

10

12

0.01 0.1 1 10 100 1000

% v

/v

0

2

4

6

8

10

12

0.01 0.1 1 10 100 1000

Particle diameter, µm

% v

/v

PEG 1500

PEG 6000

PEG 20 000

Fig. 3. Particle size distributions of the primary particles of Durcal 10in selected size fractions of granules produced with PEGs of differentmolecular mass: (×) ungranulated Durcal 10. Granule size fractions: (�)125–250�m, (�) 355–500�m, and (�) 800–1000�m.

1500 and 6000, the particle size distributions in the granulesize fractions 125–250 and 355–500�m are narrower thanthat of the initial Durcal 65 raw material because of thepreferential growth of the smaller particles.

For Durcal 40 (Table 1andFig. 2), a preferential growthof the smaller particles are seen with PEG 1500 and PEG6000 although less pronounced with the PEG 6000. Simi-lar results with Durcal 40 and PEG 1500 were obtained byScott et al. (2000). With PEG 20000, the particle size dis-tributions in different granule size fractions are very sim-ilar indicating no preferential growth. The smaller particlesize of Durcal 40 increases the granule strength compared toDurcal 65. Consequently, PEG 20000 was able to granulateeven the largest Durcal 40 particles. A pronounced break-age of granules occurred only with PEG 1500 (Johansen andSchæfer, 2001). This is why marked intergranular particlesize heterogeneities are seen for PEG 1500 only. Contrary toDurcal 65, clear particle size differences between the smalland the middle granule size fractions are seen for PEG 1500and PEG 6000. This indicates that only the largest Durcal40 particles are difficult to granulate.

T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531 529

0

2

4

6

8

0.01 0.1 1 10 100 1

% v

/v

0

2

4

6

8

01 0.1 1 10 100 10

% v

/v

0

2

4

6

8

0.01 0.1 1 10 100 1000

Particle diameter, µm

% v

/v

PEG 20 000

PEG 6000

PEG 1500

Fig. 4. Particle size distributions of the primary particles of Durcal 5in selected size fractions of granules produced with PEGs of differentmolecular mass: (×) ungranulated Durcal 5. Granule size fractions: (�)125–250�m, (�) 355–500�m, and (�) 800–1000�m.

For Durcal 10, the small particle size makes the granulesso strong that a preferential growth of the smaller particlesoccurs with PEG 1500 only (Table 1andFig. 3). With PEG6000 and especially with PEG 20000, an opposite tendencyis seen, the particle size being slightly larger in the largestgranule size fraction (Table 1). For Durcal 5, a preferentialgrowth of larger particles is seen for all PEGs (Table 1andFig. 4). No breakage of granules produced from either Dur-cal 10 or Durcal 5 was observed (Johansen and Schæfer,2001).

The preferential growth of larger particles seen for Dur-cal 10 and 5 is ascribed to an agglomerate formation by theimmersion mechanism (Johansen and Schæfer, 2001). Thisis supposed to cause a higher content of large particles inthe larger granules as will be discussed inSection 3.3. Im-mersion is promoted by a small powder particle size and ahigh binder viscosity (Schæfer and Mathiesen, 1996). There-fore, a higher binder viscosity is seen to increase the inter-granular particle size heterogeneity for Durcal 10 as well asDurcal 5.

3.3. Intragranular heterogeneity

Particle size heterogeneities within a single granule aresupposed to be less critical for the content uniformity thanintergranular heterogeneities. Intragranular heterogeneities,however, might affect granule properties such as the granulestrength and the dissolution of drug from the granules. Thereare no simple ways of obtaining quantitative measures ofintragranular particle size differences. Instead, SEM micro-graphs can give some qualitative indications of intragranularheterogeneities.

Fig. 5shows a cross section of a granule that is formed byimmersion. In the case of melt granulation, it is characteristicof granules formed by immersion that their shape partlyresembles the original shape of the binder particles (Schæferand Mathiesen, 1996; Johansen and Schæfer, 2001). Thus,the platelike shape of the PEG flakes will cause oblong andrather flat granules.

Granules formed by immersion are typically formed froma single binder particle or flake. This is supported by the fol-lowing approximate calculation. It was shown that the sizeof PEG flakes is reduced by about 50% during dry mixing(Schæfer and Mathiesen, 1996). After this size reduction, themedian size of the flakes will be approximately 800�m, thelargest flakes being approximately 2000�m. The mean massof flakes in the size fractions 710–800 and 1400–2000�mwas estimated by weighing to be 0.20 and 0.92 mg, respec-tively. From these mass values, it was calculated by meansof previous data (Johansen and Schæfer, 2001) on density ofthe molten binder, binder content, and intragranular poros-ity that typical flakes from the size fractions 710–800 and1400–2000�m are able to form granules having diametersof approximately 1200 and 2000�m, respectively, assumingspherical granules. Although the granules are not spherical,the calculation illustrates that a single flake is able to forma large granule.

It is seen fromFig. 5 that the primary particles in thecentral part of the granule are markedly smaller than theparticles in the rest of the granule. This indicates that smallparticles entered a volume that was initially occupied by thePEG flake. When the flake begins to melt, small particleswill be preferentially immersed into the surface layer of theflake since inertial forces acting to prevent immersion willincrease with particle size (Scott et al., 2000). During mass-ing, the small particles will penetrate into the molten flake,and molten binder will become squeezed to the surface dueto densification. This will make immersion of further par-ticles possible but small particles will still be preferentiallyimmersed since it is much easier to bind a small particle to alarge one (Tardos et al., 1997). When most of the small par-ticles have been immersed, the granule will be able to cap-ture larger particles as long as a sufficient amount of moltenbinder becomes squeezed to the surface. This means thatsmall binder particles will form small granules by preferen-tially capturing small Durcal particles, because the amountof binder is insufficient for further growth by capturing larger

530 T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531

Fig. 5. SEM micrograph of a cross section of a granule produced from Durcal 10 and PEG 20000.

Fig. 6. SEM micrograph of the surface of a granule produced from Durcal 10 and PEG 20000.

T. Schæfer et al. / European Journal of Pharmaceutical Sciences 21 (2004) 525–531 531

particles. This will result in a higher content of large parti-cles in the larger granules, i.e. immersion will cause a pref-erential growth of large particles.

A similar concentration of small particles in the centerof the granule was also seen in granules made from Dur-cal 5 and PEG 20000 and from Durcal 10 and PEG 6000.This intragranular particle size heterogeneity is supposed tobe closely related to the immersion mechanism and conse-quently to the use of a combination of a small particle sizeand a viscous binder.

Fig. 6 shows the surface of a granule made from Durcal10 and PEG 20000. Surface areas of a high concentrationof small particles as well as areas of a high concentrationof large particles can be identified. Similar variations inparticle size were only seen with Durcal 5 and PEG 20000.Such variations might be caused by random variations inthe liquid saturation of the surface layer since a higherliquid saturation will increase the probability of capturinglarge particles. Random variations in liquid saturation aremore likely to occur in the case of viscous binders, whichare more difficult to move by capillary forces. A differentexplanation of the surface heterogeneities seen inFig. 6might be that the granule has been formed by coalescencebetween one large and more small granules all formed byimmersion. As mentioned above, small granules formed byimmersion are supposed primarily to contain small particles.

4. Conclusions

Intergranular particle size heterogeneities due to a prefer-ential growth of the smaller particles will occur if the largerparticles of the material are unable to form granules that aresufficiently strong to resist the impact forces exerted by thegranulation equipment. This means that intergranular parti-cle size heterogeneities can be reduced either by increasingthe granule strength or by decreasing the impact forces. Inthe present investigation, the impact forces in the high shearmixer were kept constant since the granulations were pro-duced at a constant impeller speed. Granule strength can beincreased by using a raw material with a smaller particlesize and/or a binder liquid with a higher viscosity.

On the other hand, intergranular particle size hetero-geneities due to a preferential growth of the larger particleswill occur if a cohesive material is granulated with a highlyviscous binder liquid. This is ascribed to a preferentialgrowth of the larger particles caused by the immersionmechanism. Consequently, a binder liquid of low viscosityis preferable for cohesive materials.

It was found to be possible to minimize intergranular par-ticle size heterogeneities for raw materials having mean par-ticle sizes in the range of 5.5–63.1�m by a proper choiceof binder liquid. For granulation of materials having meanparticle sizes of 34.5 and 63.1�m, it was necessary to applya highly viscous (approximately 14000 mPa s) binder liquidin order to avoid intergranular particle size heterogeneities.

For a material with a mean particle size of 11.7�m, a binderliquid of medium viscosity (approximately 500 mPa s) wasfound to be the optimum choice, whereas a binder liquidof low viscosity (approximately 40 mPa s) was the optimumfor a mean particle size of 5.5�m.

Qualitative studies of intragranular particle size hetero-geneities show clear heterogeneities for small particles gran-ulated with a viscous binder liquid.

Acknowledgements

The authors wish to thank Robert A. Gault and Paul B.Hamilton, Canadian Museum of Nature, for helpful assis-tance with the scanning electron microscopy, Omya, France,for supplying the calcium carbonates, and Clariant, Ger-many, for supplying the PEGs.

References

Crooks, M.J., Schade, H.W., 1978. Fluidized bed granulation of a micro-dose pharmaceutical powder. Powder Technol. 19, 103–108.

Egermann, H., Reiss, W., 1988. Effect of particle size of drug and diluenton drug distribution in granule size fractions. Acta Pharm. Technol.34, 5S.

Iveson, S.M., Litster, J.D., Hapgood, K., Ennis, B.J., 2001. Nucleation,growth and breakage phenomena in agitated wet granulation processes:a review. Powder Technol. 117, 3–39.

Johansen, A., Schæfer, T., 2001. Effects of interactions between powderparticle size and binder viscosity on agglomerate growth mechanismsin a high shear mixer. Eur. J. Pharm. Sci. 12, 297–309.

Keningley, S.T., Knight, P.C., Marson, A.D., 1997. An investigation intothe effects of binder viscosity on agglomeration behaviour. PowderTechnol. 91, 95–103.

McGinity, J.W., Ku, C.T., Bodmeier, R., Harris, M.R., 1985. Dissolutionand uniformity properties of ordered mixes of micronized griseofulvinand a directly compressible excipient. Drug Dev. Ind. Pharm. 11,891–900.

Miyamoto, Y., Ryu, A., Sugawara, S., Miyajima, M., Ogawa, S., Matsui,M., Takayama, K., Nagai, T., 1998. Simultaneous optimization of wetgranulation process involving factor of drug content dependency ongranule size. Drug Dev. Ind. Pharm. 24, 1055–1065.

Ojile, J.E., Macfarlane, C.B., Selkirk, A.B., 1982. Drug distribution duringmassing and its effect on dose uniformity in granules. Int. J. Pharm.10, 99–107.

Schæfer, T., Mathiesen, C., 1996. Melt pelletization in a high shear mixer.IX. Effects of binder particle size. Int. J. Pharm. 139, 139–148.

Scott, A.C., Hounslow, M.J., Instone, T., 2000. Direct evidence of hetero-geneity during high-shear granulation. Powder Technol. 113, 205–213.

Selkirk, A.B., 1976. The effect of solute migration on the distributionof borax throughout a batch of granules. J. Pharm. Pharmacol. 28,512–514.

Tardos, G.I., Khan, M.I., Mort, P.R., 1997. Critical parameters and limitingconditions in binder granulation of fine powders. Powder Technol. 94,245–258.

van den Dries, K., Vromans, H., 2002. Relationship between inhomo-geneity phenomena and granule growth mechanisms in a high-shearmixer. Int. J. Pharm. 247, 167–177.

Vromans, H., Poels-Janssen, H.G.M., Egermann, H., 1999. Effects ofhigh-shear granulation on granulate homogeneity. Pharm. Dev. Tech-nol. 4, 297–303.

Yip, C.W., Hersey, J.A., 1977. Ordered or random mixing: Choice ofsystem and mixer. Drug Dev. Ind. Pharm. 3, 429–438.