The effect of vessel material on granules produced in a high-shear mixer

European Journal of Pharmaceutical Sciences 23 (2004) 169–179

The effect of vessel material on granules produced in a high-shear mixer Anneke M. Bouwmana,∗ , Marinella R. Vissera , Anko C. Eissensa , Johannes A. Wesselinghb , Henderik W. Frijlinka a

Department of Pharmaceutical Technology and Biopharmacy, Groningen University Institute for Drug Exploration (GUIDE), Ant. Deusinglaan 1, 9713 AV Groningen, The Netherlands b Department of Chemical Engineering, University of Groningen, Nijenborg 4, 9747 AG Groningen, The Netherlands Received 11 March 2004; received in revised form 16 June 2004; accepted 5 July 2004

Abstract In this study the effect of different vessel wall materials on the granule size distributions obtained during high-shear granulation of different materials is investigated. The distributions obtained in glass and stainless steel vessels differ from those obtained in PMMA (polymethyl methacrylate) and PTFE (polytetrafluoroethylene) vessels. The high contact angle of PMMA forces all liquid immediately into the more easily wetted powder bed. In this vessel a fast liquid absorbing powder nucleates in the droplet controlled regime, leading to a narrow particle size distribution. In a vessel with a low contact angle (glass or stainless steel) a liquid layer can be formed on the wall surface. This liquid causes an inhomogeneous distribution of liquid over the powder bed; a broader granule size distribution is the result. With a powder that slowly absorbs liquid, local overwet areas can be created, resulting in large granules. This results in broader granule size distributions as well. In conclusion; the contact angle of the vessel material and the wetting rate of the powder used determine the granule growth process and the resulting granule size distribution. © 2004 Elsevier B.V. All rights reserved. Keywords: High-shear granulation; Excipients; Contact angle; Water sorption; Vessel wall material

1. Introduction 1.1. Granulation process High-shear granulation is a commonly used unit operation to produce larger granules of primary particles (Holm, 1997). The granulation process has been described to consist of different growth regimes. Several authors have described these regimes (Hoornaert et al., 1998; Iveson and Litster, 1998a; Scott et al., 2000; Iveson et al., 2001). Roughly, the first stage is wetting and nucleation, followed by consolidation and coalescence, and finally attrition and breakage. The initial growth stage is the nucleation regime, where wetted particles stick together and form primary nuclei. Schæfer and Mathiesen propose two nucleation mechanisms; SMIH (Schæfer and Mathiesen’s Immersion Hypothesis) and

SMDH (Schæfer and Mathiesen’s Distribution Hypothesis) (Schæfer and Mathiesen, 1996; Scott et al., 2000). In the first mechanism particles are immersed in a binder droplet, one droplet forms one nucleus; the nucleation mechanism is droplet controlled. In the second mechanism (SMDH) binder droplets are smaller than powder particles. The droplets are dispersed over the bed, coat the primary particles and nuclei are formed by coalescence of the wetted primary particles. According to the authors, droplet size in relation to the primary particle size determines which nucleation mechanism will occur. Hapgood et al. (2003) and Litster (2003) have proposed a nucleation regime map in which the drop penetration time is used as a parameter. The drop penetration time τ p is defined as: 2/3

∗

Corresponding author. Tel.: +31 50 363 2397; fax: +31 50 363 2500. E-mail address: [email protected] (A.M. Bouwman).

0928-0987/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2004.07.008

τp = 1.35

µ Vo ε2eff Reff γLV cos θ

(1)

170

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Vo (total drop volume), εeff (effective bed porosity), and Reff (effective pore radius) describe the bed properties, whereas µ (liquid viscosity), γ LV (liquid surface tension), and θ (solid–liquid contact angle) describe the drop properties. Eq. (1) shows that the drop penetration time τ p is low if the droplet is absorbed quickly by the powder bed. The dimensionless spray flux Ψ A is defined as the ratio of the liquid area flux to the solid area flux: 3V (2) ΨA = 2Add In this equation the powder traverses the spray zone with a flux A (m2 /s), V the volumetric spray rate (m3 /s) and dd the drop diameter (m). At a low dimensionless spray flux the wetted powder is quickly replaced by non-wetted powder. One droplet can form one nucleus, leading to a droplet controlled nucleation regime. If the liquid is only slowly absorbed by the powder bed or the wetted powder is not removed fast enough from the spray area, the liquid has to be dispersed mechanically to prevent local overwetting. In Fig. 1 this nucleation regime map is shown in combination with the SMIH and SMDH. Although the latter two mechanisms originate from melt pelletization, where the amount of liquid binder depends on the temperature, the nucleation regimes are comparable to those mentioned in the nucleation regime map of Hapgood and Litster. In the drop controlled region, powder particles are immersed in a droplet, which forms the primary nucleus. As the droplets are uniform in size, a narrow nuclei size distribution can be obtained. In the mechanical dispersion region, liquid has to be dispersed through the powder bed to allow primary particles to coalesce. There can be an excess

liquid (slow liquid absorption or high spray flux) or primary particles that are too large to be immersed in the droplet. Due to the poor liquid distribution, the nuclei size distribution in general will be broader. The second growth stage of granules is the consolidation or coalescence of nuclei. According to Scott et al. (2000) the granule growth mechanism depends on the nucleation mechanism. The SMIH (immersion mechanism) leads to preferential nucleation, where nuclei do not grow much. This nucleation mechanism is promoted by pouring-on liquid; all water is immediately distributed over the powder bed. A small granule size distribution is expected. The SMDH (distribution mechanism) leads to preferential growth. The nuclei formed grow by coalescence with primary particles and other granules. Since coalescence promotes growth of the large granules, a broader granule size distribution is expected. This nucleation mechanism is promoted by using a melt-in liquid addition method. This second growth stage can also be described using a Stokes deformation number (Ennis et al., 1991; Tardos et al., 1997). This deformation number is defined as the ratio between the externally applied kinetic energy and the energy dissipated by the liquid bonds between the particles. The small size of the primary nuclei and consequently the relative thick binder layer results in a smaller viscous Stokes number than the critical viscous Stokes number. This means that all collisions result in coalescence. At the end of the second growth stage, when the viscous Stokes number becomes equal to the critical viscous Stokes number, growth stops and compaction starts. Due to the collisions, the packing of the particles becomes denser and liquid moves to the granule surface. When enough binder liquid is present at the granule surface, further growth by coalescence occurs. After a successful collision, particle rearrangement to a sphere may occur. Coalescence promotes the growth of larger granules. At the end of the coalescence regime, crushing and layering will be the predominant process (Hoornaert et al., 1998). The last growth stage is now reached: granules become too large to withstand the high-shear forces. Breakage and attrition takes place, the broken pieces can be layered around existing granules, or can coalesce. Vonk et al. (1997) showed that an equilibrium between growth and breakage is obtained. 1.2. Vessel wall material

Fig. 1. Nucleation regime map according to Schæfer and Mathiesen’s Immersion and Distribution Hypothesis. The dimensionless spray flux Ψ A is the ratio of the area flux of the droplets to the area flux of powder surface through the spray zone. The drop penetration time τ p is predicted from the capillary pressure driven flow of liquid into pores in the powder bed (Hapgood et al., 2003; Litster, 2003; Schæfer and Mathiesen, 1996; Scott et al., 2000).

Laboratory scale granulator vessels can be made from a variety of materials, whereas industrial (large) scale granulator vessels are generally made of stainless steel, which may however be coated to change the material properties. The aim of this study was to investigate whether different vessel wall materials affect the outcome of the granulation process. This is relevant for scale up since it will mean that the vessel wall properties of laboratory equipment need to match that of (future) production equipment. In this study, the granulation process is investigated in 250 mL vessels, using glass,

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

171

Table 1 Process conditions used for granulation of MCC, MFC, and ␣-lactose MCC MFC ␣-Lactose

Amount of powder (g)

Amount of liquid (ml)

Liquid addition rate (ml/min)

Impeller speed (rpm)

Chopper speed (rpm)

24 10 24

24 18 4.3

48 40 12

1000 1000 750

1500 1500 750

PMMA, stainless steel, and PTFE as vessel wall materials. Glass and PMMA are often chosen for their transparency; the granulation process can be observed. Stainless steel is a logical choice since it is the material mostly used in large scale granulators, and PTFE was chosen for its low friction. This could prevent powder and granules sticking to the wall, thereby altering the granulation process.

2. Materials and methods Microcrystalline cellulose (MCC) (Pharmacel® 101, DMV International, Veghel, The Netherlands), microfine cellulose (MFC) (Elcema P100, Degussa AG, Frankfurt, Germany), and ␣-lactose 450 M (Pharmatose® 450 M, DMV International, Veghel, The Netherlands) are granulated using water as a binding agent. Liquid was added as a continuous flow through a tube with a 1 mm orifice by a computer controlled dosimat (765 Dosimat, Metrohm Ltd., Hensau, Switzerland). The process conditions for the granulation procedures are given in Table 1. To exclude production influences (e.g. increase in impeller speed might change granule size), we chose to change only the vessel material. The process conditions chosen, with respect to the amount of liquid, liquid addition rate, impeller speed, and chopper speed, were optimised per powder. All particles grow in the steady regime; granule size increases linearly with time. Vessels of four different materials are used, all with similar size and shape and all polished; glass, PMMA (polymethyl methacrylate), stainless steel, and PTFE (polytetrafluoroethylene). High-shear granulation is performed in a MI-PRO 250 (Pro-C-epT, Zelzate, Belgium). The produced granules are dried overnight at 50 ◦ C. Granule size distribution is determined by sieving the whole batch of granules using ASTM standard sieves. From the mass based distribution the d50 is extracted. Primary particle size (as shown in Table 2) was measured by laser diffraction, type HELOS KA/LA,

dispersed with dispersion system RODOS (Sympatec GmbH, Clausthal-Zellerfeld, Germany). Water sorption measurements were carried out in the setup shown in Fig. 2. This setup was previously used for measuring water uptake of tablets by Kamp van et al. (1986). Experiments are carried out five times, using 5 g of powder. The sorption of water by the powder bed takes place over a surface of 380 mm2 . During the high-shear granulation process, powder in the wetting zone is continuously renewed, leading to non-wetted powder in the wetting zone. Therefore, we focus on the initial sorption rate of the powder; water sorption rate is calculated from the amount of water taken up during the first 50 s. It is found that sorption is linear during this period for the materials tested.

Table 2 Physicochemical characteristics of the primary powders used d50 (␮m) MCC 101 MFC P100 ␣-Lactose 450 (M)

58 42 16

Bed porosity (–)

Sorption rate (g/s) ± S.D.

0.78 0.84 0.70

0.149 ± 0.005 0.079 ± 0.007 0.019 ± 0.004

Fig. 2. The sorption measurement setup. On the glass filter (1) a glass cylinder filled with powder (2) is placed. This column of powder is into contact with water in a beaker (3) The sorption rate of the powder is measured by the decrease of mass in the beaker, measured by a balance (4) and recorded by a computer. Water to fill the tubes can be taken from the storage (5).

172

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Fig. 3. Granules produced in the different vessels. Bar represents 2 mm.

3. Results Fig. 3 shows pictures of the different granules after drying. Clearly the size and size distribution is different when different vessels are used. Fig. 4 shows that granulating MCC in different vessels leads to different granule size distributions. Glass (Fig. 4A) and stainless steel (Fig. 4C) result in similar granule size distributions. Granulation in PMMA (Fig. 4B) results in six out of eight experiments in similar size distributions. However, granule size was smaller in two experiments. The batches giving the small granules are those in which all powder took part in the granulation process immediately from the start. The other batches are those in which the initial amount of powder taking part in the process was incomplete. When the remaining fraction enters the process later on, this will result in larger granules (Bouwman et al., 2004). Finally, granulation in the PTFE vessel (Fig. 4D) gives irreproducible

results. Two of the five distributions even show a large fraction of powder which is hardly granulated, in spite of the fact that it was observed that powder was taking part in the process. The poor reproducibility of the PTFE results is also reflected in the high average width and large difference between the highest and lowest d50 obtained (Table 3). The size distributions obtained after granulating MFC in different vessels are shown in Fig. 5. The last graph (Fig. 5E) shows a comparison of MFC granule size distributions from the four different vessels. The distributions of lactose are shown in Fig. 6. Again different vessels result in different size distributions. For MFC and lactose, granulation in the PMMA vessel (Figs. 5B and 6B) results in the largest granules, whereas granulation of MCC in PMMA (Fig. 4B) can give the smallest granules. This is even clearer in Table 3, in which the average d50 of the granule size distributions made in the different vessels are listed. The different behaviour is also seen when the width of the size distribution (difference between d90 and d10 ) is taken into consideration. MCC has the narrowest size distribution, whereas the size distributions of lactose and MFC are wide. In glass and steel vessels the granulation behaviour of MFC (Fig. 5A and C, respectively) is similar and reproducible, which makes the granulation behaviour of MFC comparable to that of MCC in these vessels. The granulation of lactose in glass or stainless steel vessels (Fig. 6A and C, respectively) is also reproducible but results in size distributions broader than for the comparable MFC or MCC granulations. In the size distributions of granules prepared in the glass or steel vessel two distinct peaks are visible. The second peak has about double the size of the first peak. This is an indication for the occurrence of preferential growth occurring for only a limited number of granules. Obviously, the growth occurs according to a mechanism in which on granule binds about seven other granules (which results in a size increase of 2).

Table 3 The average granule size (d50 ) ± the average distribution widths (d90 –d10 ), the number of granulations (n), followed by the lowest and highest obtained d50 of all granule size distributions in the four different vessels Vessel material

Average d50 (mm)

Average width (d90 –d10 ) (mm)

n

Lowest d50 obtained (mm)

Highest d50 obtained (mm)

MCC

Glass PMMA Steel PTFE

0.83 0.69 0.77 0.57

0.54 0.25 0.32 0.36

5 8 4 5

0.82 0.58 0.75 0.36

0.85 0.74 0.82 0.74

MFC

Glass PMMA Steel PTFE

0.78 1.98 0.84 1.29

0.95 1.16 0.84 1.38

3 3 4 5

0.75 1.93 0.82 1.12

0.80 2.01 0.86 1.56

␣-Lactose

Glass PMMA Steel PTFE

1.19 1.57 0.88 1.72

4.47 2.17 3.43 3.26

3 3 4 5

0.86 1.51 0.81 0.47

1.47 1.56 1.05 3.79

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

173

Fig. 4. (A) Granule size distributions of MCC granulated in a glass vessel. The curve marked with * was observed three times. (B) Granule size distributions of MCC granulated in a PMMA vessel. The curve marked with * was observed four times. (C) Granule size distributions of MCC granulated in a stainless steel vessel. (D) Granule size distributions of MCC granulated in a PTFE vessel.

174

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Fig. 4. (Continued ).

Table 4 Contact angles of the vessel materials obtained from literature (Borruto et al., 1998; Radelczuk et al., 2002) Vessel material

Contact angle (◦ )

Glass Stainless steel PMMA PTFE

30 50 72 126

In the PTFE vessel no reproducible size distributions could be obtained (Fig. 6D). Moreover, the size distributions of the granular materials produced in the PTFE vessel are irreproducible irrespective of the powder used. Table 2 lists physicochemical characteristics of the primary powders used in this study. The MCC powder bed sorbs water almost twice as fast as the MFC powder bed, and seven times faster than the lactose powder bed. The bed porosities are all between 70 and 80%. A possible explanation for the variations seen in the size distributions of the granulations prepared in the different vessel materials may be found in differences in the contact angles of the vessel wall. The contact angles of the four vessels with water differ from each other. Table 4 lists the contact angles of the four materials. Glass and stainless steel show the lowest contact angles, indicating a hydrophilic behaviour. PMMA is just slightly hydrophilic, whereas PTFE is highly hydrophobic (Borruto et al., 1998; Radelczuk et al., 2002).

4. Discussion We hypothesise that the granulation behaviour (especially the initial nucleation) is determined by the balance between the contact angle of the vessel wall and the sorption rate of the powder material. A low contact angle (for example glass and stainless steel) allows the formation of a layer of liquid

on the wall. This liquid is not immediately taking part in the nucleation process. For a more slowly imbibing powder, nuclei swept against the wall take up the water, causing rapid coalescence growth of these nuclei. Due to the uneven liquid distribution over the powder, nucleation takes place according to the distribution mechanism. The size distribution of the nuclei is broad, and the final granule size distribution will probably also be broad. Only those nuclei that are swept against the wall will take up water and show growth, which explains the bimodal distributions seen with lactose granulated in glass or steel. When the vessel wall material possesses a high contact angle (PMMA and PTFE), water rather stays in the powder mixture, since the powder mixture is more easily wetted. If the liquid is absorbed in the powder bed quickly, nucleation starts immediately and all nuclei are approximately of the same size (droplet controlled nucleation). The following slow consolidation process is called preferential nucleation (Iveson and Litster, 1998b; Scott et al., 2000). In this case, the final granule size distribution is likely to be narrow. However, with a low contact angle of the vessel wall (glass, stainless steel) and a fast imbibing powder (MCC), a minor fraction of the liquid will not immediately take part in the process. Therefore, in the second growth stage some nuclei are more wetted than others, leading to a somewhat broader granule size distribution. This is in agreement with the finding of Holm et al. (1983), who concluded, that when moisture distribution is not optimal, larger aggregates are formed due to excessive wetting, resulting in a broader granule size distribution. Hoornaert et al. (1998) also have shown that a relatively small change in the amount of binder can significantly change the granule size distribution obtained by granulation. A different situation arises when the powder cannot take up the liquid quickly (MFC, lactose), but a high contact angle of the wall (PMMA) still forces all water into the powder.

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

175

Fig. 5. (A) Granule size distributions of MFC granulated in a glass vessel. (B) Granule size distributions of MFC granulated in a PMMA vessel. The curve marked with * was observed two times. (C) Granule size distributions of MFC granulated in a stainless steel vessel. The curve marked with * was observed two times. (D) Granule size distributions of MFC granulated in a PTFE vessel. (E) Granule size distributions of MFC granulated in the four different vessels.

176

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Fig. 5. (Continued ).

Nuclei are formed, but due to the slow uptake of water in the powder bed there is still unbound liquid present. This results in a situation of local overwetting in the powder bed, leading to a relatively low number of nuclei with a high water content. Coalescence growth occurs, which results in larger granules (Holm, 1997). This explains why large granules are formed from MFC and lactose granulated in a PMMA vessel. When a powder cannot take up liquid quickly, a low contact angle of the vessel wall will slow down the liquid penetration into the powder bed. In the nucleation regime map, shown in Fig. 1, this means a movement to the left towards a narrower nuclei size distribution. The liquid spreads better over the powder. Nuclei are wetted more evenly, which results in smaller granules. In Fig. 7 the different granulation mechanisms found for the different vessels and different materials are depicted schematically. MCC and MFC are similar materials; both consisting of cellulose. However, due to a different treatment, they pos-

sess different material properties. MFC is more amorphous than MCC; therefore more binder fluid is necessary to saturate the water–powder mixture (Kothari et al., 2002; Parker and Rowe, 1991). The effects are visible in Table 1, which shows the process conditions used. For granulation of MCC the mass of powder equals the mass of water used, whereas for granulation of MFC the mass of water was almost twice the mass of powder. MFC and lactose are slowly imbibing powders, which means that water is slowly transported through the powder bed. These powders nucleate and grow via mechanism C in Fig. 7 when granulated in a vessel with a high contact angle, like PMMA. If granulated in glass or stainless steel, nucleation and growth takes place via mechanism D. Both powders show a broader granule size distribution than MCC. MCC is a fast imbibing powder, which nucleates via mechanism A (in a vessel with a high contact angle, like PMMA) or via mechanism B (in a vessel with a low contact angle, like glass and stainless steel). MCC needs a fast liquid distribution over

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

177

Fig. 6. (A) Granule size distributions of lactose granulated in a glass vessel. (B) Granule size distributions of lactose granulated in a PMMA vessel. (C) Granule size distributions lactose granulated in a stainless steel vessel. (D) Granule size distributions of lactose granulated in a PTFE vessel.

178

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Fig. 6. (Continued ).

the powder bed to obtain a narrow particle size distribution; this can be achieved in the PMMA vessel. PTFE is a material which is used because of its low friction. When water is present, friction is decreased even more (Borruto et al., 1998). Granule size distributions of all three powders tested show irreproducible results. In high-shear granulation high-shear forces are needed. Shear is caused by the rotation of the impeller, by granule–granule collisions, and by wall friction. In the small scale used, the wall friction is important. Clearly, friction of PTFE with the powder bed is too low to give a reproducible granulation process. It can be concluded that the type of vessel wall material chosen significantly affects the granulation mechanism

and the final product obtained in high-shear granulation. The contact angle of the vessel material and the sorption rate of the powder used determine the nucleation process and the final granule size distribution. Glass and stainless steel vessels produce similar granules, whereas PMMA and PTFE vessels give different granule size distributions. In this study we have used small laboratory equipment which makes the effects of the wall of the vessel large. In large equipment the effects may be smaller. However, when scaling down it is important that the properties of the vessel wall of the small scale laboratory granulator match the properties of the wall of the large scale production granulator, especially with regard to contact angle and friction properties.

Fig. 7. The different granulation mechanisms. Liquid binder is added to the moving powder bed, the liquid is distributed over the bed and nucleation starts. Addition to the powder bed is similar for all situations. A difference in nucleation originates in differences in the contact angle of the vessel wall material and the liquid sorption rate of the powder.

A.M. Bouwman et al. / European Journal of Pharmaceutical Sciences 23 (2004) 169–179

Acknowledgement We thank Pro-C-epT (Zelzate, Belgium) for providing us with the stainless steel and the PTFE vessel. References Borruto, A., Crivellone, G., Marani, F., 1998. Influence of surface wettability on friction and wear tests. Wear 222, 57–65. Bouwman, A.M., Eissens, A.C., Meesters, G.M.H., Wesselingh, J.A., Frijlink, H.W., 2004. In: Pratsinis, S.E., Schulz, H., Strobel, R., Schreglmann, C. (Eds.), PARTEC. Nuremberg, Germany. Ennis, B.J., Tardos, G., Pfeffer, R., 1991. A microlevel-based characterization of granulation phenomena. Powder Technol. 65, 257–272. Hapgood, K.P., Litster, J., Smith, R., 2003. Nucleation regime map for liquid bound granules. AIChE J. 49, 350–361. Holm, P., 1997. High shear mixer granulators. In: Parikh, D.M. (Ed.), Handbook of Pharmaceutical Granulation Technology, vol. 81. Marcel Dekker, New York, pp. 151–204. Holm, P., Jungersen, O., Schæfer, T., Kristensen, H.G., 1983. Granulation in high-speed mixers. Part I. Effects of process variables during kneeding. Pharm. Ind. 45, 806–811. Hoornaert, F., Wauters, P.A.L., Meesters, G.M.H., Pratsinis, S.E., Scarlett, B., 1998. Agglomeration behaviour of powders in a L¨odige mixer granulator. Powder Technol. 96, 116–128. Iveson, S.M., Litster, J.D., 1998a. Growth regime map for liquid-bound granules. AIChE J. 44, 1510–1518. Iveson, S.M., Litster, J.D., 1998b. Liquid-bound granule impact deformation and coefficient of restitution. Powder Technol. 99, 234–242.

179

Iveson, S.M., Litster, J.D., Hapgood, K.P., Ennis, B.J., 2001. Nucleation, growth and breakage phenomena in agitated wet granulation processes: a review. Powder Technol. 117, 3–39. Kamp van, H.V., Bolhuis, G.K., Boer de, A.H., Lie, A., Huen, L., 1986. The role of water uptake on tablet disintegration. Pharm. Acta Helv. 61, 22–29. Kothari, S.H., Kumar, V., Banker, G.S., 2002. Comparative evaluations of powder and mechanical properties of low crystallinity celluloses, microcrystalline celluloses, and powdered celluloses. Int. J. Pharm. 232, 69–80. Litster, J.D., 2003. Scaleup of wet granulation processes: science not art. Powder Technol. 130, 35–40. Parker, M.D., Rowe, R.C., 1991. Source variation in the wet massing (granulation) of some microcrystalline celluloses. Powder Technol. 65, 273–281. Radelczuk, H., Holysz, L., Chibowski, E., 2002. Comparison of the Lifschitz–van der Waals/acid–base and contact angle hysteresis approaches for determination of solid surface free energy. J. Adhes. Sci. Technol. 16, 1547–1568. 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 heterogeneity during high-shear granulation. Powder Technol. 113, 205– 213. Tardos, G.I., Khan, M.I., Mort, P.R., 1997. Critical parameters and limiting conditions in binder granulation of fine powders. Powder Technol. 94, 245–258. Vonk, P., Guilaume, C.P.F., Ramaker, J.S., Vromans, H., Kossen, N.W.F., 1997. Growth mechanisms of high-shear pelletisation. Int. J. Pharm. 157, 93–102.