Effects of interactions between powder particle size and binder viscosity on agglomerate growth mechanisms in a high shear mixer

Download Effects of interactions between powder particle size and binder viscosity on agglomerate growth mechanisms in a high shear mixer

Post on 17-Sep-2016




4 download

Embed Size (px)


<ul><li><p>European Journal of Pharmaceutical Sciences 12 (2001) 297309www.elsevier.nl / locate /ejps</p><p>Effects of interactions between powder particle size and binder viscosityon agglomerate growth mechanisms in a high shear mixer</p><p>*Anita Johansen, Torben SchferThe Royal Danish School of Pharmacy, Department of Pharmaceutics, 2 Universitetsparken, DK-2100 Copenhagen, Denmark</p><p>Received 3 July 2000; received in revised form 5 September 2000; accepted 12 September 2000</p><p>Abstract</p><p>A study was performed in order to elucidate the effects of the interactions between powder particle size and binder viscosity on themechanisms involved in agglomerate formation and growth. Calcium carbonates having mean particle sizes in the range of 5214 mm andpolyethylene glycols having viscosities in the range of approximately 50100 000 mPas were melt agglomerated in a high shear mixer.Agglomerate growth by nucleation and coalescence was found to dominate when agglomerating small powder particles and binders with alow viscosity. Increasing the binder viscosity increased the formation of agglomerates by immersion of powder particles in the surface ofthe binder droplets. With a larger powder particle size, an increasing binder viscosity was necessary in order to obtain an agglomeratestrength being sufficient to avoid breakage. Due to a low agglomerate strength, a satisfying agglomeration of very large particles (214mm) could not be obtained, even with very viscous binders. The study demonstrated that the optimum agglomerate growth occurred whenthe agglomerates were of an intermediate strength causing an intermediate deformability of the agglomerates. In order to producespherical agglomerates (pellets), a low viscosity binder has to be chosen when agglomerating a powder with a small particle size, and ahigh viscosity binder must be applied in agglomeration of powders with large particles. 2001 Elsevier Science B.V. All rightsreserved.</p><p>Keywords: Melt agglomeration; High shear mixer; Binder viscosity; Powder particle size; Polyethylene glycols; Agglomerate growth mechanisms</p><p>1. Introduction is distributed on the surface of the powder particles, andformation of nuclei occurs by coalescence between the</p><p>Agglomeration is an important and established technolo- wetted particles. By the immersion mechanism, nuclei aregy in many industrial processes. However, the knowledge formed when the powder particles are captured on theof the mechanisms and kinetics involved in agglomeration surface of the binder droplets and immersed. When theis still sparse, and the need of additional research in the binder droplet size is larger than the size of the powdermatter is evident. particles, the dominating agglomerate formation mecha-</p><p>Some of the important early contributions to theory on nism will tend to be the immersion mechanism. In a highagglomeration were made by Newitt and Conway-Jones shear mixer, the binder droplet size will normally become(1958), Rumpf (1962), Capes and Danckwerts (1965), and reduced by comminution in the initial stage of the processKapur (1978). More recently, Ennis et al. (1991) de- because of the shearing forces. A low binder viscosity willveloped a new agglomeration model, and this work has enhance this size reduction of the binder droplets, as will abeen followed by several other investigations into ag- higher impeller speed. Therefore, a small initial binderglomeration mechanisms and kinetics (Knight, 1993; Kris- particle size, a low binder viscosity, and/or a high impellertensen, 1996; Iveson et al., 1996; Tardos et al., 1997; speed promote the distribution mechanism. On the otherHoornaert et al., 1998; Iveson and Litster, 1998). hand, a large initial binder particle size, a high binder</p><p>In melt agglomeration, the formation of agglomerates viscosity, and/or a low impeller speed promote the immer-can proceed by two different mechanisms (Schfer and sion mechanism (Schfer and Mathiesen, 1996a).Mathiesen, 1996a). By the distribution mechanism, binder The nuclei formed initially in the process might grow in</p><p>size by coalescence between nuclei. This agglomerategrowth is determined by a balance between coalescence*Corresponding author. Tel.: 145-35-306-000; fax: 145-35-306-031.</p><p>E-mail address: ts@mail.dfh.dk (T. Schfer). and breakage. Coalescence will be the dominant growth</p><p>0928-0987/01/$ see front matter 2001 Elsevier Science B.V. All rights reserved.PI I : S0928-0987( 00 )00182-2</p></li><li><p>298 A. Johansen, T. Schfer / European Journal of Pharmaceutical Sciences 12 (2001) 297 309</p><p>mechanism if the agglomerates possess a high agglomerate Powder particles that are too large will result in a wide sizestrength (Tardos et al., 1997). If the agglomerates are too distribution and agglomerates with a non-uniform shapeweak to resist the impact and shearing forces in the mixer, due to breakage. The optimum particle size of a rawa marked agglomerate breakage will occur simultaneously material to be melt pelletized has been indicated to be inwith growth by coalescence (Knight et al., 1998; Eliasen et the size range of 2025 mm (Schfer et al., 1992b).al., 1998, 1999). As a result, particles and small fragments Keningley et al. (1997) have shown that to formformed by breakage might participate in the growth by a granules with calcium carbonate and silicone oils, alayering mechanism by which small particles or fragments minimum binder viscosity was required, which increasedare layered on the surface of surviving agglomerates with the size of the powder particles from 1 mPas with(Linkson et al., 1973; Eliasen et al., 1999). 8-mm particles to 1 Pas with 230-mm particles. This</p><p>The agglomerate strength becomes increased by a indicates that an interaction between powder particle sizesmaller powder particle size, a higher binder viscosity, and and binder viscosity might be important in order to controlby a densification of the agglomerates (Kristensen et al., the agglomerate growth. The present work is based on the1985a; Keningley et al., 1997). In a high shear mixer, the hypothesis that a controllable agglomerate growth isshearing forces become increased by a higher impeller attainable with powders having a small as well as a largerotation speed. Further, the shearing forces depend on the particle size by combining the powder particle size and thesize and shape of the impeller blades (Schfer et al., binder viscosity in a way that gives rise to a suitable1993b). agglomerate strength. The purpose of this work is to test</p><p>According to Ennis et al. (1991), growth by coalescence this hypothesis by carrying out melt agglomeration experi-occurs until a critical agglomerate size has been reached. ments in a high shear mixer.This critical size becomes increased by, e.g., a smaller sizeof the powder particles /agglomerates, a higher binderviscosity, and a lower impeller speed (Ennis et al., 1991; 2. Materials and methodsTardos et al., 1997). However, at the same time a higherviscosity will reduce the growth potential by decreasing 2.1. Materialsthe deformability and the rate of densification of theagglomerates. The effect of viscosity on agglomerate Six different grades of calcium carbonate powders withgrowth will therefore depend on the balance of these two different mean particle size, manufactured from a whitecounteracting effects (Schfer and Mathiesen, 1996c). marble, and produced by comminution and classificationAgglomerates can only resist deformation and breakage (Omya, France), were used as starting material. Poly-when below a critical agglomerate size, which depends on ethylene glycol (PEG) 1500 S, 3000 S, 6000 S, 10000 S,the externally applied energy and on the agglomerate 20000 S, or 35000 S (Hoechst, Germany) was used asstrength (Tardos et al., 1997). meltable binder. S indicates that the PEGs are used as</p><p>Agglomeration of powders with a mean particle size flakes. Butylated hydroxyanisole (BHA) (Merck-Schuch-below approximately 10 mm has traditionally been trouble- ardt, Germany) was used as an antioxidant in order tosome and has often led to an uncontrollable agglomerate prevent thermal decomposition of the binder during thegrowth. This is because of the high liquid saturation agglomeration process (Schfer and Mathiesen, 1996b).needed in order to obtain a sufficient deformability to The size distribution by volume of the calcium carbon-counteract for the high agglomerate strength caused by the ates was determined by a Malvern 2601Lc laser diffractioncohesiveness of the small particles (Schfer et al., 1992b; particle sizer (Malvern Instruments, UK). The span isSchfer, 1996a; Schfer and Mathiesen, 1996b; Knight et calculated as the difference between the diameters at 90al., 1998). Also, the agglomeration of large powder and 10 percentage points relative to the median diameter,particles is problematic. Breakage will often dominate, and D(v;0.5).no agglomeration can occur because of a low agglomerate A Gemini 2375 Surface Area Analyzer (Micromeritics,strength (Newitt and Conway-Jones, 1958; Capes and USA) was used for the determination of the BET multi-Danckwerts, 1965). In melt agglomeration experiments, it point surface area of the calcium carbonates.has previously been shown that a powder with a mean The true density of the calcium carbonates and of theparticle size of 100 mm was unable to agglomerate solid PEGs was determined by an Accupyc 1330 gas(Thomsen, 1994). In agglomeration experiments with a displacement pycnometer (Micromeritics, USA) usingcoarse lactose monohydrate (127 mm) agglomerated with helium purge. The poured and tapped densities of thewater, weak granules were produced (Mackaplow et al., calcium carbonates were determined according to the test2000). for apparent volume (European Pharmacopoeia, 1999), and</p><p>The powder particle size is especially critical if pellets the interparticular porosities were calculated from thehave to be produced. Powder particles that are too small tapped densities.will result in oversized agglomerates and a wide agglomer- The size distributions of the PEGs were estimated byate size distribution because of an uncontrollable growth. sieve analysis with a series of 12 ASTM standard sieves in</p></li><li><p>A. Johansen, T. Schfer / European Journal of Pharmaceutical Sciences 12 (2001) 297 309 299</p><p>the range of 754000 mm. A sample of about 100 g was 752000 mm was vibrated by a Fritsch analysette 3sieved for 5 min at low vibration level by a Fritsch vibrator (Fritsch, Germany) for 10 min. The mass mediananalysette 3 vibrator (Fritsch, Germany). The mass median diameter and the span were calculated.diameter and the span were calculated.</p><p>The densities of molten PEGs were estimated at 70, 80,and 908C as previously described (Eliasen et al., 1998). 2.4.2. Intragranular porosity</p><p>The melting range and the peak temperature of the PEGs The intragranular porosity of the agglomerates waswere estimated by a Perkin-Elmer DSC 7 differential estimated by a mercury immersion method similar to thatscanning calorimeter (Perkin-Elmer, USA) as previously described by Strickland et al. (1956). A sample of 34 gdescribed (Schfer and Mathiesen, 1996b). from the agglomerate size fraction 2502000 mm was</p><p>The viscosities of the molten PEGs were estimated at placed in a glass pycnometer with an approximate volume70, 80, and 908C by a Rotovisco RV 12 rotation viscome- of 30 ml having a calibrated scale. Mercury was sucked upter (Haake, Germany) as previously described (Schfer into the pycnometer by means of vacuum. The apparentand Mathiesen, 1996c). volume of the sample was estimated by displacement of</p><p>mercury after increasing the intrusion pressure to 98.7 kPa2.2. Equipment (740 mmHg). At this intrusion pressure, mercury will</p><p>penetrate into pores greater than approximately 20 mm inThe agglomeration experiments were performed in an diameter. The corrected porosity and the liquid saturation</p><p>8-l Pellmix PL 1/8 laboratory scale high shear mixer were calculated as described by Eliasen et al. (1998). The(Niro, Denmark) (Schfer et al., 1993a). corrected porosity reflects the porosity of the wet granules</p><p>in the agglomeration phase in which the molten binder acts2.3. Mixing procedure like a liquid. All porosity analyses were performed in</p><p>duplicate.The heating jacket was preheated to 508C. Calcium The actual binder concentrations (%, m/m, of calcium</p><p>carbonate (1500 g), the amount of PEG, and 3% BHA (%, carbonate) of the fractions (2502000 mm) were estimatedm/m, of PEG) were dry mixed at an impeller speed of indirectly from the true densities of the milled agglomerate1300 rpm. The amount of PEG (%, m/m, of calcium size fraction, determined by helium pycnometry as earliercarbonate) used was 15.5% with the 5-mm powder, 15.0% described (Schfer and Mathiesen, 1996a). In addition, forwith the 13-mm powder, 13.5% with the 34-mm powder, a representative group of fractions, the actual binder12.0% with the 39-mm powder, 9.5% with the 80-mm concentration of the fractions was estimated from apowder, and 4.0% with the 214-mm powder. Because of quantitative determination of the calcium carbonate byformation of frictional heat caused by the impeller rotation, titration (Pharmacopoea Nordica, 1963). The two tech-the product temperature increased during mixing to a niques produced similar results, and the actual concen-temperature exceeding the melting point of the PEG. The trations of the fractions were found to be close to themelting point was observed as an inflection point on the nominal concentration. The density of the agglomerate sizerecorded product temperature curve. This inflection point fractions, therefore, was calculated from the nominal PEGwas defined as the start of massing time. Two minutes after concentration and used in the calculations of the intra-the melting point was observed on the temperature curve, granular porosity.the impeller speed was lowered to 800 rpm. After 8 min ofadditional massing time, the agglomeration procedure wasterminated. 2.4.3. Scanning electron microscopy</p><p>At the end of each experiment, the agglomerates were Photographs of agglomerate size fractions 2502000sieved on a 4-mm Jel-Fix 50 vibration sieve (J. En- mm were taken by a scanning electron microscope (SEM)gelsmann, Germany) for about 10 s, until the fraction finer (Jeol JSM 5200, Japan).than 4 mm had passed. The agglomerates were then spreadout in thin layers on trays allowing them to cool at ambienttemperature. The adhesion of mass to the bowl was 2.4.4. Binder distribution in size fractionsestimated as previously described (Schfer, 1996a). A sample was drawn by scooping approximately 200 g</p><p>from the cooled fraction finer than 4 mm. A series of 72.4. Agglomerate characterisation ASTM standard sieves in the range of 1801400 mm was</p><p>vibrated by a Fritsch analysette 3 vibrator (Fritsch, Ger-2.4.1. Size distribution many) for 10 min, and the single size fractions were</p><p>The size distributions of the agglomerates were esti- collected. The actual PEG concentration of the single sizemated by sieve analysis of a sample drawn by sc...</p></li></ul>


View more >