The Effect of the Physical States of Binders on High-Shear Wet Granulation and Granule Properties: A Mechanistic Approach Toward Understanding High-Shear Wet Granulation Process. Part II. Granulation and Granule Properties

The Effect of the Physical States of Binders on High-Shear Wet Granulation and Granule Properties: A Mechanistic Approach Toward Understanding High-Shear Wet Granulation Process. Part II. Granulation and Granule Properties JINJIANG LI,1 LI TAO,1 MANDAR DALI,1 DAVID BUCKLEY,1 JULIA GAO,1 MARIO HUBERT2 1

Biopharmaceutics R&D, Bristol-Myers Squibb Company, 1 Squibb Drive, PO Box 191, New Brunswick, New Jersey 08903-0191

2

Analytical R&D, Bristol-Myers Squibb Company, 1 Squibb Drive, PO Box 191, New Brunswick, New Jersey 08903-0191

Received 16 February 2010; revised 15 April 2010; accepted 6 May 2010 Published online 23 June 2010 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jps.22261 ABSTRACT: The objective is to provide mechanistic understanding of a preferred wet granulation process that a binder is added in a dry state. Blends of CaCO3 and binders were prepared and used as model systems, and they were exposed to either 96% RH (rubbery/solution state) or 60% RH (glassy state) at room temperature to control the physical state of the binders, followed by high-shear granulation and particle size measurement. The blends of PVP K12, PVP K29/32, and HPC showed a significant increase in particle size after exposure to 96% RH. An increase of aspect ratio was also observed for the blend of HPC. In contrast, the blends being exposed to 60% RH did not exhibit any increase in particle size or aspect ratio. Regarding the effect of binder molecular weight on the mechanical strength of granules, granules of PVP K29/32 had higher strength than granules of PVP K12. This can be explained using polymer entanglement theory, in which the degree of polymerization (DP) of (N
440–540) of PVP K29/32 is above the critical value (Nc
300–600) for entanglement; while DP of PVP K12 (N
20–30) is below it. Finally, a water sorption-phase transition-diffusion induced granule growth model for granulation has been suggested. ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 100:294– 310, 2011

Keywords: wet granulation; binder; phase change; glassy state; rubbery/gel/solution state; granule strength; compactibility; entanglement; polymer adhesion; diffusion; mechanism

INTRODUCTION High-shear, wet granulation processes, in which powder particles are agglomerated to increase their size and density, as well as to alter their shape through the addition of a granulating liquid such as a binder solution to the powder bed with the aid of mixing, is a key unit operation in the pharmaceutical industry and other industries such as those pertaining to food and fertilizer production.1–3 Wet granulation is a flexible and versatile process compared to other particle agglomeration technologies such as dry granulation. In addition, granules prepared by wet granulation generally have better particle size dis-

Mandar Dali’s present address is PTC Therapeutics, South Plainfield, NJ 07080. Correspondence to: Jinjiang Li (Telephone: 732-227-6584; Fax: 732-227-3784; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 100, 294–310 (2011) ß 2010 Wiley-Liss, Inc. and the American Pharmacists Association

294

tribution than those prepared by dry granulation,4 a normal particle size distribution. The binder is a key component in wet granulation formulations and plays a critical role in the formation of granules.5–8 Polymers, including synthetic, semi-synthetic and natural, are commonly used as wet granulation binders in the pharmaceutical industry.9,10 It has been long recognized in the pharmaceutical industry that the method of binder incorporation is an important factor influencing granulation processes.11 The polymeric binders are incorporated into a granulation process either in a dry state or in a solution state. In the former process, binders are first dry-mixed with API and other excipients, and then activated by spraying water onto powder particles.12 In particular, addition of binder in a dry state has the advantage of being simple and easy to operate from a manufacturing point of view, as compared to adding polymer solution to a powder bed. The agglomeration mechanisms involved in the wet granulation process, in which binder is added in a

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

solution state, have been extensively investigated. The results were summarized by Lister and Ennis.13 According to Lister and Ennis, the wet granulation process can be divided into three different phases: (1) wetting and nucleation, (2) consolidation and granule growth, and (3) breakage and attrition.13,14 It is commonly regarded that the distribution of granulation liquid (binder solution) is a critical factor to control wetting and nucleation; while mixing dynamics affect consolidation and granule growth, as well as breakage and attrition.15–17 Lately, Hapgood and coworkers, introduced the nucleation regime map, in which the liquid penetration time correlated with the spray flux of the granulation liquid.15,17 By doing this, the region, which is controlled by either the droplets or the mechanical dispersion, was identified. In the pharmaceutical industry, it is becoming increasingly popular to incorporate binder in a dry state. Therefore, there is a need to study the mechanisms involved. The hypothesis of this article is that the physical state of binders is the major factor affecting the granulation processes and the resulting granule properties. The physical state (G to R–S) change of binders after exposure to 95% RH or liquid water has been reported in Part I.18 The focus of Part II is to investigate the effect of the physical state and other properties of a binder on the granulation process as well as the mechanical properties of the resulting granules. For this purpose, CaCO3 is selected as filler because CaCO3 is inert and it has less or no interaction with water so that the interaction between water and a binder can be focused on. Research on a complex and more realistic system including an API and lactose monohydrate is ongoing and the results will be reported as Part III. The binary blends of CaCO3 and each binder were made at binder concentrations of 5% (w/w), 10% (w/w), and 20% (w/w) and after exposing to either 60% RH or 96% RH at room temperature the equilibrated blends were mixed in a Diosna high-shear granulator for 2 min. The particle size distribution, as well as the aspect ratio distribution of the resulting granules was then measured using a Sympatec dynamic microscope with image analysis capability. Granule friability (as a surrogate test for granule strength) and the compaction properties of granules were evaluated using a sonic sifter and a compaction simulator, respectively. A mechanistic model for particle agglomeration, based on the effect of the physical state of binders, is proposed. Finally, the effect of the molecular weight of the binders on the mechanical properties of the granules has been evaluated from a polymer dynamic point of view, in which the formation of chain entanglement only occurs at or above the critical values of the degree of polymerization (Nc). Entanglement is also proportional to the DOI 10.1002/jps

295

molecular weight of a polymer. Although a high-shear granulation process was used in this study, the conclusions from this study should be applicable to fluid bed processing as well, given that the binder is added in a dry state. This is because the current study has taken an equilibrium approach to investigate the effect of phase transition on granulation, and the kinetics of phase change involved in granulation processes, which can vary from one process to another, is not considered.

EXPERIMENTAL Materials Four commonly used polymeric binders (simply referred to as binder in the following text), including polyvinylpyrrolidone (PVP) K12 and K29/32, hydroxypropylcellulose (HPC EXF), and hydroxypropylmethylcellulose (HPMC E5), were used in this study. PVP K12 was obtained from BASF Chemicals, Inc. (Florham Park, NJ) and PVP K 29/32 was purchased from ISP Technologies, Inc. (Wayne, NJ), separately. HPC EXE was purchased from Hercules Chemicals, Inc. (Wilmington, DE), and HPMC E5 was obtained from Univar Chemicals, Inc. (Redmond, WA), respectively. Calcium carbonate (CaCO3) from Konoshima Chemicals Co. (Osaka, Japan) was used as filler in this study. Potassium sulfate (K2SO4) and sodium bromide (NaBr), were both purchased from Sigma– Aldrich Chemicals, Inc. (St. Louis, MO). They were used to control the relative humidity (RH) (K2SO4 for 96% RH and NaBr for 60% RH) for preparing granulation samples. Ziplock plastic bags (4 mil, 12 in.  18 in.) purchased from VWR International (Wester Chester, PA), were used as humidity chambers. Sample Preparation Preparation of Dry Blends Binary blends of CaCO3 with each binder at binder concentrations of 5% (w/w), 10% (w/w), and 20% (w/w) were prepared as follows. To prepare a 400 g batch of a binary blend of CaCO3 with PVP K12 at a binder concentration of 5% (w/w), CaCO3 and PVP K12 were weighed out in a ratio of 95% (w/w) to 5% (w/w), transferred into a P 1/6 Diosna high shear granulator (Diosna, Dieerks and Sohne Gmbn, Osnabruck, Germany), and then mixed for 2 min with an impeller speed of 700 rpm and a chopper speed of 1200 rpm. For preparing the binary blends (400-g batches) of CaCO3 with PVP K12 at binder concentrations of 10% (w/w) and 20% (w/w), CaCO3 and PVP K12 were weighed out in ratios of 90% (w/w) to 10% (w/w) and 80% (w/w) to 20% (w/w), respectively. The same mixing parameters used to prepare the blend with 5% JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

296

LI ET AL.

(w/w) PVP K12 were used for these blends. The same procedure was used for preparing binary blends of CaCO3 with the other three binders at concentrations of 5% (w/w), 10% (w/w), and 20% (w/w).

particle size. Particle shape is represented by aspect ratio. Aspect ratio is defined as a ratio of the minimum and maximum Feret diameters where Feret diameter is the distance of two tangents to the contour of the particle.

Preparation of Wet Blends

Scanning Electron Microscopy (SEM)

To produce granules, the binary blends of CaCO3 and binder at binder concentrations of 5% (w/w), 10% (w/w), and 20% (w/w) were first exposed to either 60% RH or 96% RH at room temperature for 4 weeks in sealed Ziplock plastic bags. The humidity levels in these plastic bags were controlled by saturated salt solutions, NaBr for 60% RH and K2SO4 for 96% RH at room temperature. In addition, a small amount (about 50–100 mg) of each binder was placed in the Ziplock bag of the corresponding blend to observe the change in the physical state (G to R–S) of the binder after exposure to either 60% RH or 96% RH. All the binary blends in the Ziplock bags were redistributed manually throughout the bags after 2 weeks exposure to ensure that particles of CaCO3 and binder were uniformly exposed to the specified humidity. After exposure to either 60% or 96% RH for 4 weeks, the exposed binary blends were then transferred to a P1/6 Diosna high-shear granulator with a 1 L bowl, and mixed for 2 min with an impeller speed of 700 rpm and a chopper speed of 1200 rpm. This procedure was repeated for all of the binary blends. After mixing, all samples were dried in an oven at 508C for 24 h, and then manually screened through #18 mesh.

A scanning electron microscope (SEM) (Philips XL-30-ESEM, Lab 6 and FEG) from FEI (Hillsboro, OR) was used to acquire the micrographs of binders and CaCO3 in a high vacuum using a secondary electron detector. All samples were coated with platinum to eliminate possible sample charging.

Characterization of Granules Loss on Drying (LOD) Prior to mixing the wet blends in a high-shear granulator, a Mettler-Toledo LJ16 Moisture Analyzer from Mettler-Toledo, Inc. (Columbus, OH) was used to measure the water content of the exposed blends. Approximately 2–3 g of the wet blends was placed on the moisture analyzer, and heated to 1058C. The weight loss was then recorded. Granule Size Analysis A Sympatec Qicpic Instrument (Lawrenceville, NJ) dynamic microscope with image analysis capability was used to measure the particle size of the granules, and to determine their shape distribution. The instrument is equipped with a compressed air dispersion system (Rodos), and employs dynamic image analysis to derive particle size and shape information. The measurement procedure included an M6 lens with a range of 5–1705 mm, a dispersion pressure of 2 Bar and an image acquisition rate of 400 images/s. A 4 mm injection port was used for the analysis. The diameter of a circle of equal projection area as the measured particle was used to represent JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

Mechanical Measurement Granule Friability Test. Granule friability was conducted by first preparing a sieve fraction of the bulk granules. The sample was prepared by pouring a portion of the bulk granules onto a sieve stack consisting of 20 and 40 mesh screens. The sieve stack with sample was placed into the sonic sifter and run at sift ¼ 5 (equivalent to 30 Hz), pulse ¼ 5 for 5 min. The granules that passed through the 20 mesh screen, and retained on the 40 mesh screen were then used in the granule friability test. Approximately 2 g of accurately weighed and recorded sample was placed into a 20-mL scintillation vial followed by the addition of
5 g of glass balls (4 mm diameter, Pyrex, Corning, NY). The vial was then placed into a 60-mL glass bottle, supported by article towels, to avoid any movement of the vial within the glass bottle. After supporting the vial within the bottle, the bottle was placed in a Turbula Mixer (Type T2C, Glen Mills, Inc., Clifton, NJ), and rotated for 20 min at
56 rpm. For measuring the loss due to friability, the contents of the vial (granules and glass beads) was poured onto a sieve stack consisting of 20 and 40 mesh screens. The glass balls were retained on the 20 mesh screen, and removed for reuse. The sieve stack with sample was placed into the sonic sifter, and run at sift ¼ 5, pulse ¼ 5 for 5 min. Finally, the granules retained on the 40 mesh screen were accurately weighed and recorded. The friability index was taken as (1  Y/X)100, where X is the original sample retained on the screen (425–710 mm) and Y is the sample retained by a 40 mesh (425 mm) sieve after testing. Compaction Test. Compactibility measurement of CaCO3-binder granules at a 10% binder concentration was conducted using an ESH compaction simulator. Flat-faced, round upper and lower punches with a diameter 0.96 cm were used. The crosssectional area of the round dies cavity was 0.713 cm2. The profile duration was set at 2 s. Each result is an average of three readings. The target weight of each compressed tablet was 0.300 g (within DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

Table 1. Water (w/w) Content of CaCO3/Binder Binary Blends Exposed to 96% RH by LOD

Sample Name PVP K12 PVP K29/32 HPC HPMC

Water Uptake at 5% Binder Concentration

Water Uptake at 10% Binder Concentration

Water Uptake at 20% Binder Concentration

5.5% 5.5% 3.26% 4.0%

8.7% 8.68% 4.0% 4.39%

15.00% 14.98% 7.11% 7.78%

297

had higher water content than those with cellulosic binders (HPC and HPMC). The blends of PVP K12 and PVP K29/32 had similar water content at the same binder concentration as compared to the blends with cellulosic binders, which is consistent with the results from water sorption analysis reported in Part I. Moreover, after exposure to 96% RH, all binders except for HPMC, were observed to undergo phase transition from the glassy state to the rubbery or solution state. Particle Size and Particle Morphology of Granules

3%). The weight, thickness (Mitutoyo Absolute, Model ID-S1012EBS, Mitutoyo Corp., Kanagawa, Japan), and hardness (Vanderkamp, VK200 Tablet Hardness Tester, Vankel, Edison, NJ) of each compressed tablet was determined at each compressional force. The compression force and in-die thickness were recorded as displayed by the instrument (Compaction Simulator, Phoenix Calibration and Services Ltd., Brierley Hill, UK). Compactibility was determined as the slope of tensile strength (kPa) versus compression force (MPa).

RESULTS Water Content of Binary Blends Table 1 shows the water content of the CaCO3-binder blends for all four binders, in which the binder concentration varied from 5% (w/w) to 10% (w/w), and up to 20% (w/w). As shown in Table 1, the amount of water in the blends generally increased with an increase in binder concentration, although the blends of cellulosic binders (HPC and HPMC) showed less increase in water compared with the blends of PVP binders, in particular in the concentration range of 5% (w/w) to 10% (w/w). Furthermore, the blends of PVP binders (PVP K12 and PVP K29/32) consistently

Effect of Binders Figure 1a displays the particle size (D50 ¼ 61 mm for PVP K12, D50 ¼ 106 mm for PVP K29/32, D50 ¼ 85 mm for HPC, and D50 ¼ 106 mm for HPMC) distribution of individual binders as well as CaCO3 (D50 ¼ 39 mm). As shown in Figure 1a, all four binders have a monomodal distribution of particle size, and all of them, except for PVP K12, have the highest percentage of particles sized around 100 mm (mode), as indicated by their D50 values, while, in comparison, PVP K12 had smaller particles with a mode of around 60 mm (D50 ¼ 61 m). However, CaCO3 particles had a bimodal distribution with two modes around 5–10 mm (D10 ¼ 7 m) and 40–50 mm (D50 ¼ 39 m). Figure 1b shows the aspect ratio distribution of individual binders, and CaCO3, in which PVP binders have a narrower and higher aspect ratio (AR50 ¼ 0.81 for PVP K12 and AR50 ¼ 0.77 for PVP K29/32) distribution compared with cellulosic binders (AR50 ¼ 0.54 for HPC and AR50 ¼ 0.63 for HPMC), indicating that PVP binders are more spherical than cellulosic binders because the AR 50 values of PVP binders are close to 1 (the aspect ratio of a perfect sphere is 1). Figure 2 shows the SEM micrographs of individual binders and CaCO3. In general, binders have much bigger particles than CaCO3 based on both particle size distribution and SEM micrographs. The

Figure 1. The particle size (a) and aspect ratio (b) distribution of individual binders (PVP K12, PVP K29/32, HOC, and HPMC) and CaCO3. DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

298

LI ET AL.

Figure 2. SEM micrographs of individual binders (PVP K12, PVP K29/32, HPC, and HPMC) and CaCO3.

large particles observed for the CaCO3 sample (D90 ¼ 250 m) is attributed to the agglomeration of small or individual particles, as displayed in the SEM micrograph (Fig. 2). In addition, it was also confirmed that PVP K12 had more small particles compared to PVP K29/32 (see Fig. 2 for details), which is consistent with the results from particle size measurement. Furthermore, as seen in the SEM micrographs, the

particles of PVP binders appear to be round and the particles of cellulosic binders (HPC and HPMC) are irregular in shape with some elongated particles. Regarding CaCO3 particles, most of them appeared to be clustered together so that the shape of the individual particles of CaCO3 could not be defined even though the clusters appeared to be round (AR50 ¼ 0.71). Figure 3 shows the particle size distribution of the binary blends of CaCO3-binder at a binder concentration of 10% for all four binders after exposure to either 96% RH or 60% RH and being wet massed for 2 min. As shown in Figure 3a, the particle size (D50 ¼ 287 mm for PVP K12, D50 ¼ 264 mm for PVP K29/32 and D50 ¼ 202 mm for HPC) of the blends, except for the blend of HPMC, increased significantly compared with that of individual binders after exposure to 96% RH and wet massing. In contrast, there was no observed increase in particle size (D50 ¼ 84 m for PVP K29/32, D50 ¼ 83 m for HPC and D50 ¼ 88 m for HPMC) for these three blends after exposure to 60% RH and wet massing for 2 min, as compared to that of individual binders (see Figs. 1a and 3b). The blend of PVP K12 exposed to 60% RH is excluded. PVP K12 binder could potentially undergo a physical state change from the glassy state to the rubbery/solution state. Based on the results shown in Figure 3, the blends of PVP K12, PVP K29/32, and HPC, after exposure to 96% RH and wet massing, formed granules while the blends of PVP K29/23, HPC, and HPMC, being exposed to 60% RH and wet massed, remained un-granulated (look powder-like). This is consistent with the observation that binders being exposed to 96% RH (except for HPMC) undergo a phase change from a glassy state to a rubbery/ solution state, while binders being exposed to 60% RH remain in the glassy state (powder-like). Further examination of Figure 3a indicates that the granulated blends of PVP binders (D90 ¼ 697 mm for PVP

Figure 3. Effect of binder on particle size distribution: (a) the binary blends of CaCO3/ binder (10% binder concentration) for binders PVP K12, PVP K29/32, HPC, and HPMC after exposed to 96% RH and (b) the binary blends of CaCO3/binder for binders PVP K29/ 32, HPC, and HPMC after exposed to 60% RH. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

299

Figure 4. Effect of binder on aspect ratio distribution: (a) the binary blends of CaCO3/ binder (10% binder concentration) for binders PVP K12, PVP K29/32, HPC, and HPMC after being exposed to 96% RH and (b) the binary blends of CaCO3/binder for binders PVP K29/32, HPC, and HPMC after exposed to 60% RH.

K12 and D90 ¼ 699 mm for PVP K29/32) have larger particles than the blend of HPC (D90 ¼ 390 m), although they have similar amounts of fines (D10 ¼ 42 mm for PVP K12, D10 ¼ 57 mm for PVP K29/23, and D10 ¼ 62 mm for HPC). The blends of HPMC, which were exposed to 96% RH and 60% RH as well as wet massed for 2 min, showed particle size distributions that were identical to the distribution of the HPMC binder. This suggests that no granules formed in these blends. This can be explained by the fact that the HPMC binder did not undergo a phase change from the glassy state to the rubbery/solution state even after exposure to 96% RH (see Part I for details). Figure 4 shows the aspect ratio distribution of the blends of all four binders after being exposed to either 96% RH or 60% RH, and being wet massed for 2 min. After exposure to 96% RH, all blends, except for the blend of HPMC, exhibited a high and narrow aspect ratio distribution (see Fig. 4a); a typical indication of round particles; while the blend of HPMC showed a low and broad aspect ratio distribution since the physical state of HPMC remained unchanged, as indicated previously. Compared with their respective individual binders, the aspect ratio distribution (AR50 ¼ 0.79 for PVP K12 and AR50 ¼ 0.84 for PVP K29/32) of PVP blends is similar to those of the binders alone (AR50 ¼ 0.81 for PVP K12 and AR50 ¼ 0.77 for PVP K29/32) (see Fig. 1b). The aspect ratio (AR50 ¼ 0.79) distribution of the HPC blend is higher and narrower compared with that of HPC binder alone (AR50 ¼ 0.54) (see Fig. 1b), which implies the sample is granulated. For the cellulosic binder blends exposed to 60% RH (Fig. 4b), the aspect ratio distribution is broad, which is the same as those of the binders alone; a typical indication of nongranulation. The blend of PVP K29/32 had a narrower aspect ratio distribution, which is similar to that of PVP K29/32 alone. DOI 10.1002/jps

Therefore, aspect ratio distribution is not a good indicator for agglomeration for the blends of PVP binders; whereas for the blends of HPC and HPMC binders, it could be indicative of granulation. Effect of the Physical State of Binders As previously discussed, it is assumed in this article that agglomeration in the wet granulation processes is controlled by the physical state of the binders used. Experimentally, the physical state of a binder is controlled by exposing the binder to different levels of humidity; 60% RH for the glassy state and 96% RH for the rubbery/solution state. To further elaborate on this, particle size data for the binary blends at a binder concentration of 10%, after exposure to moisture and being wet massed for 2 min, is presented. Figure 5 shows the effect of the physical state of binders on particle size, and thereby the granulation state of the blends, in which the particle size distributions of the exposed blends are displayed for all four binders. In addition, the particle size distributions of the dry blends are included for comparison. To reiterate, the data for the blend of PVP K12 after exposure to 60% RH, is excluded for the reason that PVP K12 can potentially undergo a physical state (G to R–S) change at the specified humidity level based on the water sorption data reported in Part I. It is clearly shown in Figure 5 that all the blends being exposed to 96% RH, except for the blend of HPMC, have particles which are larger in size (D50 ¼ 286.8 mm for PVP K12, D50 ¼ 264.3 mm for PVP K29/32, and D50 ¼ 201.8 mm for HPC) than those being exposed to 60% RH (D50 ¼ 84 mm for PVP K29/ 32 and D50 ¼ 83 m m for HPC), and also the dry blends (D50 ¼ 42 mm for PVP K12, D50 ¼ 91 mm for PVP K29/ 32, and D50 ¼ 76 mm for HPC). The latter two have a similar particle size distribution. This indicates that the blends being exposed to 96% RH, with exception of JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

300

LI ET AL.

Figure 5. Effect of physical state (96% RH vs. 60% RH) on the particle size distribution of the binary blends of CaCO3/binder (a) PVP K12, (b) PVP K29/32, (c) HPC, and (d) HPMC.

the blend of HPMC, were granulated, while the blends being exposed to 60% RH were not. This is in line with the observation that the binders being exposed to 96% RH, with the exception of HPMC, underwent a phase change from the glassy state to the rubbery-solution state, suggesting that the physical state of binders plays a significant role in granulation. Regarding the blends of HPMC, as indicated in Figure 5d, the blend being exposed to 96% RH (D50 ¼ 70 mm) had a similar particle size distribution as the one exposed to 60% RH (D50 ¼ 88 mm), as well as the dry blend (D50 ¼ 88 mm), which is consistent with the observation that the physical state of the HPMC binder remains unchanged after exposure to moisture as discussed in Part I. It is also noted that the blends made with PVP binders (D90 ¼ 697 mm, D90 ¼ 699 mm) have more large particles than the blend of HPC (D90 ¼ 390 mm), which may be attributed to the physical-chemical properties of binders in the rubbery or solution state (see Part I for more in-depth discussion). The impact of the physical state of binders on granulation of these blends is also reflected in their aspect ratio distribution. Figure 6 shows the effect of the physical state of binder on aspect ratio distribution of the exposed blends, in which the aspect ratio distribution of the blends being exposed to 96% RH and 60% RH (excluding the blend of PVP K12 being exposed to 60% RH), as well as the dry blends are exhibited. As shown in the previous section, the aspect ratio distribution of the PVP JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

blends being exposed to 96% RH (AR50 ¼ 0.79 for PVP K12 and AR50 ¼ 0.84 for PVP K29/32) is high and narrow, which is similar to that of both dry blends (0.75 for PVP K12 and AR50 ¼ 0.76 for PVP K29/32), and the blend being exposed to 60% RH (AR50 ¼ 0.8), as in the case of the PVP K29/32 binder. This is again due to the fact that PVP binders have spherical particles, whereby the formation of spherical granules does not increase the aspect ratio of the blends. For the blend of HPC, the aspect ratio of the one being exposed to 96% RH is larger than both the blend being exposed to 60% RH, and the dry blend because HPC binder is irregular in shape; the formation of spherical granules increases the aspect ratio. Regarding the blends with HPMC, the aspect ratio distribution of the one being exposed to 96% RH is the same as that of the blend being exposed to 60% RH as well as the dry blend. The distribution is low and broad owing to the irregular shape of the particles of HPMC binders. The HPMC binder remained in the glassy state after exposure to moisture so there was no formation of granules observed in the HPMC blend. This further confirms that aspect ratio is a good indicator of granulation for the blends with HPC and HPMC, but not for the blends with PVP binders. Effect of Binder Concentration The amount of a binder in a formulation affects both the granulation rate and the physical properties of granules made, since both the formation of granules DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

301

Figure 6. Effect of physical state (96% RH vs. 60% RH) on the aspect ratio distribution (a) PVP K12, (b) PVP K29/32, (c) HPC, and (d) HPMC.

and the granule properties depend on the interactions of binders with API and other excipients. In this article, the binder concentration in the CaCO3-binder blends was varied from 5% (w/w) to 10% (w/w) to 20% (w/w). The effect of binder concentration on granule

formation, particle size, and aspect ratio was investigated using particle size and aspect ratio measurements. The physical state of the binders was controlled by exposing the blends to 96% RH at room temperature. Figure 7 shows the impact of binder

Figure 7. Effect of binder concentration on particle size distribution of binary blends of CaCO3/binder after being exposed to 96% RH (a) PVP K12, (b) PVP K29/32, (c) HPC, and (d) HPMC. DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

302

LI ET AL.

Table 2. Effect of Binder Concentration on Particle Size (EQPC) Distribution of Granulated Blends Binder PVP K12 PVP K29/32 HPC HPMC PVP K12 PVP K29/32 HPC HPMC PVP K12 PVP K29/32 HPC HPMC

D10 (mm) D50 (mm) D90 (mm) 36.9 64.5 42.9 33.8 41.6 56.5 61.5 29.0 93.3 250.7 126.3 30.5

181.7 334.5 233.2 89.4 286.8 264.3 201.8 70.4 188.1 381.7 279.5 74.9

694.0 794.6 464.4 206.7 696.7 698.5 390.4 145.9 399.2 563.4 674.9 168.9

Binder Concentration (%) 5 5 5 5 10 10 10 10 20 20 20 20

concentration on the particle size distribution for the blends: Figure 7a for PVP K12, Figure 7b for PVP K29/23, Figure 7c for HPC, and Figure 7d for HPMC. It is also noted in Figure 7 that the blends with 5% (w/ w) binder as well as the blends with 10% (w/w) binder were wet-massed for 2 min; while the blends with 20% (w/w) binder were only wet-massed for 50 s to avoid over granulation. In general, all the granulated blends (except for the blends of HPMC), regardless of the binder concentration, had a larger particle size than their corresponding dry blends (see Table 2 for D50 values), indicating that the physical state of the binders dictates the initiation of granulation. Specifically, as shown in Figure 7, the amount of fines decreases with increasing binder concentration for the blends of all binders except for HPMC. This is also reflected as an increase of the D10 values (see Tab. 2), in particular at 20% (w/w) binder concentration. In addition, the particle size distribution narrows with increasing binder concentration for the granulated blends, especially for the blends with PVP binders at a 20% (w/w) concentration. The particle size distribution for the blends with HPMC remained the same even as the binder concentration increased from 5%

(w/w) up to 20% (w/w). The distributions were also same as that of dry blend. This is again due to the fact that the physical state of HPMC remained in the glassy state, even after exposure to 96% RH. There was no formation of granules in these blends. This conclusion is also supported by the results from aspect ratio determination. Figure 8 shows the effect of binder concentration on the aspect ratio of the blends of HPC and HPMC, in which the aspect ratio distribution of HPC blends at all three binder concentrations is higher and narrower compared with that of the dry blends, whereas the aspect ratio of HPMC blends at all three binder concentrations is low and broad, similar to that of its dry blend. This is in line with the particle size distribution results in which the blends with HPC binder are granulated, as indicated by the formation of round particles from irregular/elongated particles. Additionally, their aspect ratio increased. The blends with HPMC are not granulated. The HPMC binders remain in their original, irregular/elongated shape, and therefore, their aspect ratio stays unchanged. In addition to binder concentration, mixing is also an important parameter in a wet granulation process, which will be discussed in the next section. Effect of Mixing In a traditional, wet granulation process, mixing is critical for the distribution of granulation liquid, nucleation, granule growth, and granule breakup. The effect of wet massing (mixing) on granulation was investigated for the blends of CaCO3-binder at a 10% (w/w) binder concentration given that all the blends were exposed to 96% RH to control their physical state (rubbery/solution), and is discussed in this section. Figure 9 shows the effect of wet massing on the particle size of CaCO3-binder granules, in which the particle size distribution of the blends of all four binders, before and after wet massing for 2 min, are displayed. In addition, the particle size distribution of the dry blends is included for comparison. It is noted

Figure 8. Effect of binder concentration on aspect ratio distribution of binary blends of CaCO3/binder after being exposed to 96% RH (a) HPC and (b) HPMC. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

303

Figure 9. Effect of mixing on the particle size distribution of binary blends of CaCO3/ binder being exposed to 96% RH (a) PVP K12, (b) PVP K29/32, (c) HPC, and (d) HPMC.

from Figure 9 that the blends which were wet massed for 2 min (D50 ¼ 287 mm for PVP K12, D50 ¼ 264 mm for PVP K29/32, D50 ¼ 202 mm for HPC, and D50 ¼ 70 mm for HPMC) have similar particle size distribution compared to those that were not wet massed (D50 ¼ 530 mm for PVP K12, D50 ¼ 255 mm for PVP K29/32, D50 ¼ 199 mm for HPC and D50 ¼ 77 mm for HPMC). A significant particle size increase was observed for the wet massed blends compared to their corresponding dry blends (D50 ¼ 42 mm for PVP K12, D50 ¼ 91 mm for PVP K29/32, D50 ¼ 75 mm for HPC, and D50 ¼ 88 mm for HPMC). This was true for blends with all the binders except HPMC. The blend with HPMC had a similar particle size distribution compared to its dry blend for to the reason explained

in the previous section. For the blends with PVP K12, it is noted that the one that was wet massed for 2 min had smaller particles than the blend not wet massed. This is probably because wet massing broke some granules of CaCO3–PVP K12. The fact that wet massing did not significantly influence particle size is interesting since the conventional view is that mixing is a critical step to increase particle size in agglomeration. Some possible mechanisms will be discussed later. Similarly, the effect of wet massing on the aspect ratio of the blends of HPC and HPMC has also been examined. Figure 10 shows the effect of mixing on the aspect ratio distribution of HPC and HPMC blends. For the blends of HPC (Fig. 10a), the one which was wet massed (AR50 ¼ 0.79) had a

Figure 10. Effect of mixing on aspect ratio distribution of binary blends of CaCO3/ binder being exposed to 96% RH: (a) HPC and (b) HPMC. DOI 10.1002/jps

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

304

LI ET AL.

similar aspect ratio distribution as the blend which was not wet massed (AR50 ¼ 0.80). Both blends had a higher and narrower aspect ratio distribution compared to the dry blend (AR50 ¼ 0.56), suggesting there is formation of spherical granules in the blends. This is consistent with results obtained during particle size analysis. For the blends with HPMC, both the wet-massed (AR50 ¼ 0.7), and nonwetmassed (AR50 ¼ 0.69) blends had similar aspect ratio distribution compared to the dry blend (AR50 ¼ 0.64), indicating that there is no change of particle shape, and therefore, no formation of granules. As discussed before, this is because HPMC remains in the glassy state after exposure to 96% RH. In general, based on the results obtained so far, mixing does not appear to be a critical factor in affecting granule size if the binder is well mixed with other excipients, and kept in the rubbery/gel/solution state for certain period. The possible mechanism for this will be discussed later. Besides granulation or granulation rate, binders can also affect the granule properties. In the following section, the effect of binders and binder concentration on the mechanical properties of granules will be discussed. Mechanical Properties Friability Mechanical properties of CaCO3-binder granules prepared with PVP K12, PVP K29/32, and HPC binders were studied by measuring their friability and compaction properties. Since the blends of HPMC did not form granules after exposure to 96% RH, there is no friability and compactibility data available for CaCO3–HPMC granules. In Table 3, the friability index values, which are the percentage of weight loss after 20-min rotations in a vial, for granules of PVP K12, PVP K29/32, and HPC at a 10% binder concentration are reported. As indicated in Table 3, the granules of PVP K29/32 have much less weight loss than those of PVP K12 after 20 min rotation. This is due to the fact that the average molecular weight of PVP K29/32 is higher than that of PVP K12, which leads to different physical mechanisms of interaction (see Discussion Section for details). Therefore, PVP K29/32 granules are relatively stronger. Regarding the granules made with HPC, their weight loss is less Table 3. Friability Index Values for CaCO3/Binder Granules of All Four Binders at a Binder Concentration of 10% Binder PVP K12 PVP K29-32 HPC

Percentage of Loss 70 14 39

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

Table 4. Friability Index Value for CaCO3/PVP K29/32 Granules at Binder Concentrations of (w/w) 5%, 10%, and 20% Binder Concentration

Percentage of Loss

5% 10% 20%a a

55 14 30

Fifty seconds granulation time to avoid over granulation.

than those prepared with PVP K12, and more than those made with PVP K29/32 after 20-min rotations. Therefore, HPC granules are stronger than PVP K12 granules, but weaker than PVP K29/32 granules. This is because HPC is chemically a different polymer from the PVP types; it exhibits a different hydration behavior, and has different flow properties as well (see Part I for details). As shown in Part I, HPC polymer changes to the rubbery-gel state rather than the solution state after exposure to 96% RH. It stays in the gel state until much further dilution. Overall, granules prepared with PVP K29/32 binder are the strongest of the three. As expected, the binder concentration can also significantly impact granule strength. Table 4 shows the effect of binder concentration on granule strength for PVP K29/32 granules, in which the friability index values are reported for granules of three binder concentrations: 5%, 10%, and 20%. As indicated in Table 4, the weight loss decreased from 55% to 14% when the binder concentration increased from 5% (w/w) to 10% (w/ w), indicating that granules made with a higher level of binder had greater mechanical strength. For the granules made with 20% (w/w) binder, the mechanical strength values were between those for granules made with 10% and 5% of PVP K29/32 since their weight loss was higher than those made with 10% (w/ w) of PVP K29/32, and lower than those made with 5% (w/w) of PVP K29/32. This is due to the fact that a short mixing time (50 s) was used for the blends with 20% (w/w) binder to avoid over granulation. This is rather interesting since mixing does not affect granule size according to the results reported in the previous section. However, based on the above results, it appears that mixing can significantly impact the strength of granules. Compactibility Similar to the friability test, compaction properties were characterized for granules prepared with three different binders: PVP K29/23, PVP K12, and HPC at a 10% binder concentration. In Table 5, compactibility (the slope of tensile strength vs. compression force plot) is listed for granules made with these three DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

Table 5. Compaction Properties of CaCO3/Binder Granules at a Binder Concentration of 10% Compatibility (kPa/MPa)

Binder PVP K12 PVP K29/32 HPC

8.34 9.17 4.70

binders, respectively. As shown in Table 5, granules made with PVP K29/32 binder are most compactable, followed by granules made with PVP K12 and HPC, although the granules made with the three binders have similar powder properties (see Tab. 6).

DISCUSSION The results in this article show that the physical state of a binder controls the granulation (nucleation and granule growth) in the high-shear wet granulation process when a binder is added in a dry state. The chemical nature of a binder, and the amount of the binder in the formulation influence both the granulation rate, and the physical and mechanical properties of the granules prepared. Theoretical Background As shown in the results section, the change of the physical state of a polymeric binder from the glassy state to the rubbery-solution state in a wet granulation process is determined by the water activity and the subsequent water sorption. This controls both the granulation and granule properties. Therefore, it is pertinent to discuss polymer adhesion mechanisms in this section. Based on the Bueche–Cashin–Debye theory,19,20 Lee19 proposed three distinct types of polymer adhesion on the basis of the physical state of adhesion and adherence: (1) rubbery polymer–rubbery polymer (R–R adhesion), (2) rubbery polymer– glass polymer (R–G adhesion), and (3) rubbery polymer–nonpolymer substrate (R–S adhesion). According to this classification, the physical state of the polymers determines the adhesion mechanism involved. Diffusion of polymers in various states can be expressed in the following equation: Dh ¼



 ArkT R2 36 M

(1)

305

where D, h, A, r, k, T, M, and R2 are the molecular diffusion constant, the bulk viscosity, Avogadro’s number, the density, Boltzmann’s constant, the absolute temperature, the molecular weight, and the mean-square end-to-end distance of a single polymer chain, respectively.19,21,22 As indicated in Eq. (1), the self diffusion coefficient of polymers is inversely proportional to viscosity. In the glassy state, since the viscosity of polymers is around 10 13 poise, the diffusion coefficient is un-measurable (
1021 cm2/s) for most of the polymers.20 On the other hand, the self diffusion coefficient of polymers in the rubbery state is
1011–1013 cm2/s,21,22 which is much higher than that in the glassy state. As suggested by Voyutskii,23 molecular diffusion is the major driving force for polymer autohesion or heteroadhesion given that a strong interaction between polymers and an adherent surface is expected. Therefore, polymers need to be at the rubbery or gel or solution state to act as an adhesive. As shown above,24 the diffusion of a polymer is inversely proportional to the viscosity of the polymer, which is related to the molecular weight of the polymer. Based on polymer physics, for polymer chains to form entanglement structures, the molecular weight needs to pass a critical molecular mass, Mc, which can also be expressed as a critical degree of polymerization, Nc. The value of Nc is typically between 300 and 600 for linear polymers. Regarding the viscosity of polymer in solution, it is proportional to the degree of polymerization in the following ways: h1ðN < Nc Þ and

h1N 3:4 ðN > Nc Þ

(2)

The self-diffusion of polymers in concentrated and undiluted solutions has been described by at least three theories: entanglement coupling, repetition, and cooperative motion.25–27 A common feature of these theories is that the self-diffusion coefficient, D, is inversely proportional to the degree of polymerization, N, and also the molecular weight. The strength of an adhesion bond depends on many factors such as inter-diffusion across the interface and average monomer interpenetration depth. Practically, tack and green strength are both important material properties which are related to the molecular interdiffusion across an interface. The green strength, which is the ability of a material to resist deformation and fracture, is an important material property for granules made with polymeric binders. Based

Table 6. Compression Properties of CaCO3/Binder Granules Based on Carr Index Property Bulk density (g/cc) Tap density (75 taps) (g/cc) Carr index (%) DOI 10.1002/jps

CaCO3 þ 10% HPC

CaCO3 þ 10% PVP K-12

CaCO3 þ 10% PVP K-29/32

0.401  0.055 0.487  0.057 17.9  1.4

0.414  0.015 0.507  0.016 18.3  0.7

0.413  0.061 0.505  0.066 18.3  1.4

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

306

LI ET AL.

on molecular theories such as inter-diffusion and interpenetration, the green strength of granules (the strength of polymer adhesion, s1) is proportional to the square root of a degree of polymerization (s1
N1/2) of the binder used. In other words, if the same kind of polymer is used in a wet granulation process, the binder with higher molecular weight produces stronger granules. Based on the general theory of polymer adhesion described above, the following analysis can be made for the roles of polymeric binders in wet granulation processes. Water, a plasticizer, is used to lower the glass transition temperature of polymeric binders to below room temperature, which is typically the operating temperature for many wet granulation processes. Prior to water sorption, the blends consist of polymer particles in the glassy state and CaCO3 particles. After the binder absorbs water, and subsequently undergoes a phase transition to the rubbery/solution state from the glassy state, two major types of adhesion exist in the blends: (1) Rubbery polymer (solution)–CaCO3 (R–C), (2) rubbery polymer (solution)–rubbery polymer (solution) (R–R). The R–C adhesion describes the situation when a binder adheres to the surface of CaCO3 particles, and the R–R adhesion represents the case when CaCO3 particles with binder molecules adhere at their surface, and agglomerate through polymer– polymer autohesion. The effect of polymer adhesion on the mechanical properties of granules for the polymer binders in this article will be discussed later. Effect of the Physical State Granulation As shown in the Results Section, the initiation of granulation in the blends with PVP and cellulosic binders is controlled by the physical state change (G to R–S) of the binders in the blends. As discussed in the Theoretical Background Section, polymeric binders start to act as adhesives only when they undergo (G to R–S) phase change. This is in line with the results presented. As discussed in Part I, the binders in Ziplock bags being exposed to 96% RH, PVP K12 (MW ¼ 2000–3000 Da), PVP K29/32 (MW ¼ 44000– 54000 Da), and HPC, were liquefied. The PVP K12 solution was less viscous than PVP K29/32, and HPC solutions. In this case, polymers were first plasticized by water and changed to the rubbery/solution state. Solutions of polymers start to wet CaCO3 particles and adhere to the surface of these particles after phase transition. This is referred to as R–C adhesion. Simultaneously, polymer chains in the solution start to inter-penetrate with each other, which is R–R adhesion. PVP K12 and PVP K29/32 changed to solution quickly after exposure to moisture, thereby indicating that their molecules have higher mobility JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

than HPC. It is expected that both PVP K12 and PVP K29/32 binders can quickly migrate to CaCO3, and start interpenetration. This is supported by the observation that granules made with PVP K12 and PVP K29/32 are larger than those prepared with HPC (see Tab. 2). As shown in Part I, HPC stays at the gel state after exposure to moisture, indicating that HPC has less mobility than PVP K12 and PVP K29/32 in solution. Therefore, granules made with HPC are smaller (see Table 2). For the blends being exposed to 60% RH, it is not expected that they are granular since the binders remain at the glassy state, and the molecules of the polymers are not mobile. The blends with HPMC remain un-granulated even after exposure to 96% RH and being wet massed since HPMC remains at the glassy (powder) state. This is because the intermolecular hydrogen bonding in HPMC prevents the disruption of the chains in the polymer to water vapor due to a lack of interaction sites.27 However, HPMC can become a gel when it is exposed to enough liquid water. This was discussed in Part I. Regarding the effect of binder concentration, the amount of binder in the blends affects both the granulation rate and granule size if they are in the rubbery/solution state. The amount of fines can be reduced by increasing the binder concentration, and decreasing the granulation time. Both criteria are required for the blends with high binder concentration (20%). Interestingly, it is noted that mixing has no significant effect on the particle size distribution of granules, indicating that once binders are in the rubbery/solution state, there is enough molecular mobility for the binder to migrate to the surface of CaCO3 particles, start inter-penetration, and initiate nucleation and granule growth in 4 weeks. The particle size of the PVP K12 blends was reduced by mixing due to breakage resulting from the lack of strength of adhesion of PVP K12. However, mixing can influence the mechanical properties of granules. This will be discussed in the next section. Granule Properties Based on the data presented in this article, the physical properties of granules can be significantly impacted by the physical state of the binders as well as other binder properties such as binder concentration, average molecular weight, chemical composition, and mixing. As indicated in Table 3, granules made with PVP K29/32 are much stronger than those made with PVP K12. This can be explained using the adhesion theory discussed previously. PVP K29/32 is a larger molecule with a degree of polymerization (N
440–540) which is close or above the critical value (Nc
300–600) for entanglement formation. Therefore, PVP K29/32 binders can form entanglements in the rubbery/solution state. In other words, DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

PVP K29/32 chains can be entangled to form a physical network which would result in high adhesive strength for binding given that the molecular penetration of PVP K29/32 is same as that of PVP K12. On the other hand, the degree of polymerization of PVP K12 (N
20–30) is far below the critical value (Nc
300–600) for formation of entanglement. Therefore, it is expected that PVP K12 chains would not entangle with each other, forming a physical network, as well as produce strong adhesion bonds. This is supported by the results in which the granules of PVP K12 are much weaker than those prepared with PVP K29/32. In practice, the degree of polymerization of a binder, rather than the solution viscosity of the binder, should be a factor to be considered when selecting a granulation binder. HPC binder is chemically different from PVP28 and, therefore, granules made with HPC binder have different mechanical strength from those made with PVP binders as shown in Table 3. HPC granules are stronger than PVP K12 granules, but they are weaker than PVP K29/32 granules. This is because HPC has a different water sorption and flow behavior, such as viscosity, upon exposing to moisture/water. After exposure to 96% RH, HPC appears to be gel-like. This may render HPC less mobile compared to PVP K29/ 32. However, the gel-like nature may also give HPC granules better strength than PVP K12 granules. Regarding compaction properties, as indicated in Table 5, granules made with PVP K29/32 binder are more compactable than those made with PVP K12 and HPC. Again, granules made with HPC are less compactable due in part to its physical behavior in the rubbery/solution state. Mixing can affect the granule strength as well (see Tab. 4). Granules experiencing longer mixing times are typically stronger since the kinetic energy from mixing increases particle contact. Mechanistic Model Historically, the first theory for wet granulation was proposed by Newitt and Conway-Jones.29 It describes the state of liquid saturation with increasing liquid content, and identifies the stages of water distribution in a wet granulation process as: (1) pendular (liquid bridges), (2) funicular (transitional), and (3) capillary state (fully saturated state).30–34 The fourth state was identified as the ‘‘droplet state’’.29 Based on this theory, the formation of liquid bridges and capillary pressure in the pendular state are mainly responsible for the formation of granules. The contributing factors are the surface tension and negative suction pressure due to curvature of the liquid (negative capillary pressure). In the capillary state, the total pore space is filled by liquid, and interfacial forces and capillary pressure are again involved. At ‘‘the droplet state’’, all pores, including DOI 10.1002/jps

307

inter- and intra-particulate voids, are filled, and there is extra liquid to surround solid particles. In practice, this is often the case in over-granulation. The cohesive forces involved in these states were investigated by Rumpf,35,36 Conway-Jones,37 and Newitt and Papadopoulos.38 According to their models, the cohesive forces, which are responsible for agglomeration through the liquid bridges between solid particles, can be described as follows: The strength of the cohesive force ( F) depends on the surface tension (g) of the granulating liquid, wetting angle (u), the distance between the particles (x), and the particle diameter (a) (see Eq. 3). In an idealized situation with x ¼ 0 (contact) and d ¼ 08, the force between the spherical particles of a diameter, a, is equal to: ! pag   F¼ (3) 1 þ tg f2 where f is the center angle describing the extension of liquid bridges. Furthermore, as discussed above, the formation of liquid bridges depends on the amount of liquid added.37 The physical states are distinguished by the relative amount of liquid occupying the interstitial pore volume, S. The pendular state is characterized by the formation of liquid bridges between particles with S < 25%. Further increasing the liquid content to S > 25% results in the formation of the funicular state. It reaches the capillary state when S is greater than 80%. For all these states, the cohesive stress, sc, is equal to: s c ¼ Spc

and

pc ¼ A

1  " gcos u " a

(4)

where S is the degree of liquid saturation, e is the porosity of the powder mass, and A is the proportionality constant depending on the geometry of packing of the particles, as well as others. In particular, A ¼ 9/ 4 for the pendular state and 6 for the capillary state. In the pharmaceutical industry, the amount of solution or water required in a wet granulation process is often used to indicate the end point of the process. Lately, an engineering approach to the understanding of wet granulation was taken by Lister and Ennis, in which the granulation process was modeled using three kinetic rate processes: wetting and nucleation, growth and consolidation, and attrition and breakage.39,40 In this model, wetting and nucleation, and attrition and breakage are controlled by the distribution of granulation liquid and mechanical mixing, respectively, while granule growth and consolidation are determined by both the binder properties and the mechanical energy applied. Growth of granules by consolidation is modeled by the Stoke number, Stv, which is the ratio of the initial kinetic energy to the energy dissipated JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

308

LI ET AL.

by viscous effect. It is expressed as follow: Stv ¼

4ru0 D 9m

(5)

where r, u0, D, and m are, respectively, the granule density, half the initial velocity of impact, the granule diameter, and the liquid viscosity. This model predicts the effect of binder viscosity to reduce the rate of granule consolidation. In order for two granules to coalesce, their relative velocity needs to become zero. In practice, the initial kinetic energy may be too small to overcome the viscous lubrication resistance in the liquid layer. To solve this problem, Ennis introduced the concept of the critical Stoke number, Stv , which is given by the following equation:

 1 h Stv ¼ 1 þ ln (6) e ha where e, h, and ha are, respectively, the coefficient of restitution, the thickness of the surface layer, and the characteristic height of the surface asperities. According to this model, two granules will coalesce if Stv  Stv , and they will bounce off if Stv > Stv . Clearly in this model, the effect of binder viscosity, which is related to the binder molecular weight, is taken into account as a viscous force. Models proposed by Newitt et al., as well as by Lister and Ennis, are specific for the situation in which the granulation liquid is added in a liquid state, although it was suggested that it can be applied to a granulation process in which a binder is added in a dry state. As shown by the data presented in this article, for a wet granulation process in which a binder is added in a dry state, swelling and phase change of the binder are responsible for the initiation of nucleation and granule growth. As discussed in Part I, water taken up in the binary blends of CaCO3-binder can be divided into to two portions: water sorption (swelling) of the polymeric binder and water adsorbed at the

surface of the CaCO3 particles. Microscopically, the binder viscosity is related to both molecular weight of a binder and the interactions of polymer chains (entanglement or interpenetration). As discussed in the theoretical section, the entanglement of polymer chains can significantly increase the solution viscosity. As indicated by the results, granules strength is proportional to the molecular weight of the binder used, which can be attributed to the formation of chain entanglement, and thereby the physical network. A molecular mechanism, based on polymer adhesion, is proposed for the wet granulation processes in which the binder is added in a dry state. In this model, the water sorption and phase transition of a binder, and diffusion of the binder in the blends are proposed as the critical factors influencing the agglomeration of particles. In particular, the water sorption of polymeric binders and subsequent change from a glassy state to rubbery/solution state upon contacting water is the critical step for the initiation of nucleation. Since polymer diffusion in the rubbery/ solution state is actually a fast process given the time frame used in the experiment, the granules grow readily once nucleation is initiated. This is supported by the fact that mixing is not a critical parameter influencing the formation of granules once binder and CaCO3 are predispersed. However, in an actual wet granulation process, when water is sprayed onto powder particles, mechanical mixing is of great importance since it can help distribute water evenly among powder particles. Furthermore, the granule strength is affected by both the binder used and the mixing dynamics. An entanglement theory is proposed for explaining the effect of binder molecular weight on granule strength. In addition, when a chemically different binder is used, its interaction with water and solution properties can affect granulation rate and granule properties significantly. This is supported by the results of HPC blends.

Figure 11. A schematic showing that water sorption-phase transition-diffusion and polymer adhesion mechanism: (A) dry blend, (B) polymeric binder absorbed water and underwent phase change and (C) polymeric binders diffuse to CaCO3 particle surface and polymers start to interpenetrate each other. JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

DOI 10.1002/jps

THE EFFECT OF THE PHYSICAL STATES OF BINDERS ON GRANULATION: PART II

Figure 11 shows a schematic diagram for the proposed mechanism where CaCO3 and binder particle are first randomly mixed, then the binder particles swell upon being exposed to water, and finally particles adhere to each other through CaCO3binder and CaCO3/binder-Binder/CaCO3 adhesion due to the diffusion of polymer chains and mixing action. Based on the current mechanism, it is the physical state change (G to R–S) of the polymer binders that not only triggers the nucleation, but also affects the granulation rate, as well as the mechanical properties of granules. In addition, the granule strength can be affected by the molecular weight of the polymers or the degree of polymerization, since the change (G to R–S) of the physical state of a binder depends on both the chemical nature of the binder, and its molecular weight. As shown in the Results Section, high molecular weight polymer PVP K29/23 yields stronger granules than PVP K12 since its polymer chains can be significantly entangled (forming networks), and can therefore, create strong bonds between particles. Similarly, when a large amount of binder is present in a formulation, consequently, there are more interactions between the binder and the other components, and more binder–binder interactions. Therefore, granules prepared with a high amount of binder are stronger compared to those prepared with less binder.

CONCLUSIONS In this article, four polymers: PVP K12, PVP K29/32, HPC, and HPMC were selected as high-shear, wet granulation binders. The binary blends of CaCO3 with each binder were prepared as model systems. After being exposed to either 96% RH or 60% RH, the blends were mixed for 2 min in a high-shear granulator to form granules. Based on the results of particle size as well as aspect ratio, it is concluded that the physical state of polymeric binders dictates the initiation of granulation. For the blends of PVP K12, PVP K29/32, and HPC, the results suggest that the blends only form granules after binders changed from the glassy state to the rubbery/solution state. It is also found that mixing is not critical for granule formation as particle size remains unchanged before and after mixing for the blends being exposed 96% RH. This is due to the fact that the blends were exposed to 96% RH for 4 weeks which gave enough time for diffusion of binder molecules to CaCO3. In practice, mixing not only helps distribute water but also increases the chances of a binder contacting API and other excipients. The concentration of binder in the blends can also affect the particle size distribution in which the amount of fines was found to decrease with increasing binder concentration. As expected, DOI 10.1002/jps

309

both the chemical composition of a binder and the concentration of the binder in the blends influence the mechanical properties of granules. The granules prepared with PVP K29/32 are stronger compared to granules made with PVP K12, indicating that the molecular weight of the binder can affect the strength of the granules. This is consistent with the prediction from polymer physics of adhesion based on entanglement theory. In addition, the results show that the strength of the PVP K29/32 granules increases with an increase in the binder concentration in the blends. The data have also been analyzed using the theories of polymer adhesion, including autohesion and heteradhesion, based on the diffusion of the polymers in various states (glassy state vs. rubbery state). Finally, a ‘‘water sorption-phase transition-diffusion’’ mechanism is proposed for a high-shear, wet granulation process, in which a binder is added in a dry state followed by water addition. This is also consistent with theoretical prediction because based on the difference of diffusion coefficient of polymers between the glass state and the rubbery-solution state, it is impossible to have enough molecular mobility for polymers to have adhesive properties in the glass state with respect to the time frame of a granulation process even high shear force is applied. However, the dynamics of phase transition of binders in the high shear wet granulation process is important, which is not considered since an equilibrium approach has been taken in this study. Although the impact of shearing force on the kinetics of binder phase transition can affect granulation rate and granule properties, the conclusion, in which the phase transition of a binder initiate nucleation and granulation, remains the same. In the future, a ternary system including an API and lactose monohydrate, which closely represents real pharmaceutical formulations, will be examined for the effect of binders on the high-shear, wet granulation process in the presence of API and lactose monohydrate. Granulation rate, the physical and mechanical properties of granules, and dissolution of granules will be examined.

ACKNOWLEDGMENTS The authors thank the management of Biopharmaceutics R&D and Analytical R&D for encouragement and support, and Ms. D. Chiappetta for proof reading, and Prof. Zografi for discussion.

REFERENCES 1. Snow RH, Allen T, Ennis BJ, Lister J. 1998. Size reduction and size enlargement. In: Green D, Perry R, editors. Perry’s cheJOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

310

2.

3.

4.

5.

6.

7.

8. 9.

10.

11.

12.

13.

14.

15.

16.

17. 18.

LI ET AL.

mical engineers’ handbook. 7th edition. New York: McGrawHill, Section 20, pp 56–89. Ennis BJ, Witt W, Weinekotter R, Sphar D, Commeran E, Snow RH, Allen T, Raymus CJ, Lister JD. 2007. Solid-solid operations and processing. In: Green D, Perry R, editors. Perry’s chemical engineers’ handbook. 8th edition. New York: McGraw-Hill, Section 21, pp 117–127. Parikh DM. 2005. Handbook of pharmaceutical granulation technology. 2nd edition. Roca Raton, FL: Taylor & Francis Group. Salman AD, Hounslow MJ, Seville JPK. 2007. Handbook of powder technology: Granulation. Amsterdam, The Netherlands: Elsevier. Hancock BC, York P, Rowe RC. 1994. An assessment of substrate-binder interactions in model wet masses. 1: Mixers torque rheometry. Int J Pharm 102:167–176. Parker MD, York P, Rowe RC. 1991. Binder-substrate interactions in wet granulation. 2: The effect of binder molecular weight. Int J Pharm 72:243–249. Parker MD, York P, Rowe RC. 1992. Binder-substrate interactions in wet granulation. 3: The effect of excipient source variation. Int J Pharm 80:179–190. Rowe RC, Sheskey PJ, Weller PJ. 2003. Handbook of pharmaceutical excipients. Chicago: Pharmaceutical Press. Bu¨hler V. 2005. Polyvinylpyrrolidone excipients for pharmaceuticals: Povidone, crospovidone and copovidone. New York: Springer. Li J, Mei X. 2006. Applications of cellulose derivatives in immediate release solid dosage. In: Marchessualt RH, Ravenelle F, Zhu XX, editors. Polysaccharides for drug delivery and pharmaceutical applications. Washington, DC: ACS Symposium Series 934, American Chemical Society. pp. 19–55. Knight PC, Instone T, Pearson JMK, Hounslow MJ. 1998. An investigation into the kinetics of liquid distribution and growth in high shear mixer agglomeration. Powder Technol 97:246–257. Iveson SM, Lister JD, Hapgood K, Ennis BJ. 2001. Nucleation, growth and breakage phenomena in agitated wet granulation processes: A review. Powder Technol 117:3–39. Lister J, Ennis BJ. 2004. The science & engineering of granulation processes. Dordrecht, The Netherlands: Kluwer Academic Press. Michaels JN, Farber L, Wong GS, Hapgood K, Heidel SJ, Farabaugh J, Chou JH, Tardos GI. 2009. Steady states in granulation of pharmaceutical powders with application to scale-up. Powder Technol 189:295–303. Lister JD, Hapgood KP, Michaels JN, Sims A, Robarts M, Kameneni SK, Hsu T. 2001. Liquid distribution in wet granulation: Dimensionless spray flux. Powder Technol 114: 32–39. Hapgood KP, Lister JD, White ET, Mort PR, Jones DG. 2004. Dimensionless spray flux in wet granulation: Monte-Carlo simulations and experimental validation. Powder Technol 141:20–30. Hapgood KP, Lister JD, Smith R. 2003. Nucleation regime map for liquid bound granules. AIChE J 49:350–361. Li J, Tao L, Dali M, Buckley D, Gao Z, Hubert M. in press. The effect of the physical states of binders on high-shear wet granulation and granule Properties: A mechanistic approach towards understanding high-shear wet granulation process. Part I. Physical characterization of binders. J Pharm Sci 9999:1–18.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 100, NO. 1, JANUARY 2011

19. Lee LH. 1967. Adhesion of high polymers. I. Influence of diffusion, adsorption, and physical state on polymer adhesion. J Poly Sci Part A-1: Polym Chem 5:751–760. 20. Bueche F, Cashin WM, Debye P. 1952. The measurement of self-diffusion in solid polymers. J Chem Phys 20:1956–1958. 21. Lee LH. 1989. New perspectives on polymer adhesion mechanisms. In: Lee LH, editor. New trends in physics and physical chemistry of polymers. New York: Plenum. pp. 95–127. 22. Lee LH. 1991. Recent studies in polymer adhesion mechanisms. In: Lee LH, editor. Adhesive bonding. New York: Plenum Press. pp. 1–30. 23. Voyutskii SS. 1963. Polymer reviews. Vol. 4. Autohesion and adhesion of high polymers. New York: Interscience. 24. Rubinstein M, Colby RH. 2003. Polymer physics. New York: Oxford University Press. 25. De Gennes PG. 1975. Scaling concepts in polymer physics. New York: Cornell University Press. 26. De Gennes PG, Leger L. 1982. Dynamics of entangled polymer chains. Ann Rev Phys Chem 33:49–61. 27. Klein J. 1979. The self-diffusion of polymers. Contemp Phys 20:611–629. 28. Kondo T. 1998. Hydrogen bonds in cellulose and cellulose derivatives. In: Dumitriu S, editor. Polysaccharides: Structural diversity and functional versatility. New York: Marcel Dekker. pp. 131–172. 29. Schwartz JB. 1990. Theory of granulation. In: Kadam KL, editor. Granulation technology for bioproducts. Boca Ratob, Florida: CRC Press. pp. 9–28. 30. Pietsch W. 2002. Agglomeraton processes: Phenomena, technologies, equipment. Weinheim, Germany: Wiey-VCH. 31. Leuenberger H. 1994. Moist agglomeration of pharmaceutical powders (size enlargement of particulate material)-the production of granules by moist agglomeration of powders in mixers/ kneaders. In: Chulia D, Deleuil M, Pourcelot Y, editors. Handbook of powder technology: Powder technology and pharmaceutical processes. Amsterdam, The Netherlands: Elsevier. pp. 377–389. 32. Benali M, Gerbaud V, Hemati M. 2009. Effect of operating conditions and physico-chemical properties on the wet granulation kinetics in high shear. Powder Technol 190:160–169. 33. Fournier Z, Geromichalos D, Herminghaus S, Kohonen MM, Mugele F, Scheel M, Schulz M, Schulz B, Schier Ch, Seemann R, Skudelny A. 2005. Mechanical properties of wet granular materials. J Phys: Condens Matter 17:S477–S502. 34. Soulie F, El Youssoufi MS, Cherblanc F, Saix C. 2006. Capillary cohesion and mechanical strength of polydisperse granular materials. Eur Phys J E: Soft Matter 21:349–357. 35. Rumpf H. 1958. Basic principles and methods of granulation. I. II. Chem Ing Tech 30:144–158. 36. Rumpf H. 1958. Basic principles and methods of granulation. III. Survey of technical granulation processes. Chem Ing Tech 30:329–336. 37. Conway-Jones JM. 1956. Capillary cohesion of granules. J Imp Coll Chem Eng Soc 10:66–84. 38. Newitt DM, Papadopoulos AL. 1959. Mechanism of granule formation. Fertiliser Soc (Engl) Proc 55:35. 39. Liu LX, Lister JD, Iveson SM, Ennis BJ. 2000. Coalescence of deformable granules in wet granulation processes. AIChE J 46:529–539. 40. Ennis BJ, Tardos G, Pfeffer R. 1991. A microlevel-based characterization of granulation phenomena. Powder Technol 65:257–272.

DOI 10.1002/jps