- Email: [email protected]
Determination of breakage rate and breakage mode of roller compacted pharmaceutical materials
Powder Technology 298 (2016) 99–105
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Determination of breakage rate and breakage mode of roller compacted pharmaceutical materials Andreja Mirtič, Gavin K. Reynolds ⁎ Pharmaceutical Technology and Development, AstraZeneca, Macclesfield SK10 2NA, UK
a r t i c l e
i n f o
Article history: Received 14 December 2015 Received in revised form 9 April 2016 Accepted 24 April 2016 Available online 26 April 2016 Keywords: Oscillating mill Milling kinetic Breakage mode Fraction of fines
a b s t r a c t Roller compaction is a common unit operation in the manufacture of oral solid dose pharmaceutical products. The roller compactor can produce intermediate ribbons of compacted material that exhibit a range of mechanical properties. Both the breakage rate and the breakage mode of these intermediates will have an effect on the final particle size distribution (PSD), and are therefore important parameters in determining the performance of the final product in pharmaceutical manufacturing. The breakage rates of roller compacted ribbons of two pharmaceutical excipients, microcrystalline cellulose (MCC) and mannitol, were determined with two different approaches, i.e. by analysing the milling mass throughput and the change in mass in the largest granule size class. The self-similar solution of the breakage population balance equation provided an insight into size dependencies of breakage rates and the prevailing breakage mode for specific process parameters. The breakage rate was found to be dependent on the impeller speed, milling mesh aperture size and ribbon porosity, but not on grinding media fill level. For MCC ribbons the milling kinetic profile changed with the mesh aperture size, suggesting dependence of the breakage mode on screen size. Our work shows clear correlation between the amount of fines produced during milling and the underlying breakage mode. For the materials studied, it is evident that mannitol ribbons tend more toward an attrition breakage mode than MCC ribbons of equal porosity when milling with the same process settings. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Milling is one of the most extensively used unit operations in pharmaceutical manufacturing. Understanding how milling parameters affect size reduction of particles is extremely important in achieving a desired particle size distribution, which allows better product uniformity, optimization of formulation dissolution properties and improves bioavailability of the product. Despite extensive literature on the subject of milling, mechanistic insight into the milling process of pharmaceutical material still remains poorly understood. It is already established that the size reduction behaviour of particles not only depends on the material properties, but also on the milling type and process settings. In order to develop a predictive methodology for modelling size reduction processes of pharmaceutical material, the correlation among those characteristics have to be understood and quantified [1,2], in particular with respect to breakage rate and particle size distribution (PSD) of milled material. In dry granulation processes of pharmaceutical materials the oscillating mill is a commonly used method to produce desired granules from roll compacted ribbons [3]. The principle of an oscillating granulator is to mechanically pass compacted material through a wire mesh screen using an oscillating rotor. Sakwanichol et al. [4] evaluated the granule properties obtained from an oscillating granulator using ⁎ Corresponding author. E-mail address: [email protected] (G.K. Reynolds).
http://dx.doi.org/10.1016/j.powtec.2016.04.033 0032-5910/© 2016 Elsevier B.V. All rights reserved.
different oscillating-granulator parameters, i.e. rotor speed, oscillating angle, aperture of mesh screen and rotor type. They concluded that all the examined parameters significantly affect the granule size distribution. Moreover, it was reported that screen aperture size has the largest effect on milling time and work as well as the degree of particle size reduction whereas the impeller side arm shape has the largest effect on overall milling performance in terms of milling efficiency and energy usage [5]. In general, breakage can be achieved by impact or attrition, depending on material properties, equipment design and process parameters. Any or both of the above breakage modes may occur independently or simultaneously [6]. Bazin and Lavoie [7] found that milling proceeds by attrition when operating in the rolling and cascading regime, as the low energy intensity collisions chip off weaker edges and corners of particles. Several methods have been developed for determination of particle breakage behaviour during milling process [6,8,9]. The aim of the current study was to determine the breakage rate of roll compacted ribbons on the underlying breakage modes during oscillating batch milling using several different approaches. In addition, the aim was to correlate the influence of process parameters and material properties, such as ribbon porosity, tensile strength, impeller speeds, input ribbon mass and screen size on the milling performance of two pharmaceutical excipients that exhibit very different mechanical properties. Specifically, microcrystalline cellulose (MCC) was selected as a material that exhibits ductile behaviour and mannitol was selected as it exhibits more brittle behaviour.
100
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
literature the exponential mass throughput was reported for batch screen milling using following equation [10]: N m ¼ m∞ 1− exp − Nc
ð1Þ
where N is the number of cycles, m and m∞ are the mass of granules for N = N and N = ∞, respectively. Nc is a characteristic number of milling cycles at which m = 1 − e−1. Smaller value of Nc corresponds to more rapid breakup of the ribbons, and a larger value corresponds to a lower fracture rate of the ribbons [10]. 2.4. Study of topmost size class of PSD plots Fig. 1. Schematic representation of the oscillating granulator.
2. Experimental methods and methodology 2.1. Ribbon production Microcrystalline cellulose (MCC) Avicel PH101 (FMC BioPolymer) and mannitol Pearlitol 200SD (Roquette)Were used to manufacture ribbons using an AlexanderWerk BT120 (Alexanderwerk, Germany) roller compactor with roll width 25 mm and smooth roll surface. To achieve different ribbon porosities, the roll force and screw speed of the roller compactor were changed. Roll gap (2.5 mm) and roll speed (3 rpm) were kept constant during experiments. Ribbon density was measured with a Geopyc (Micromeritics Ltd., US) with three repetitions. To calculate ribbon porosity the following true densities of materials were used: 1.57 and 1.47 g/cm3 for MCC and mannitol, respectively, as determined by AccPyc 1330 Pycnometer. 2.2. Milling and particle size analysis Before milling experiments all fines were removed from the ribbons. Ribbons were cut into smaller pieces and 50 g of ribbons were introduced into the mill for each experimental setup. The milling process was performed with a Frewitt screen mill (OscilloWitt) using oscillating mode that is schematically presented in Fig. 1. The particle size distribution (PSD) was measured using a QicPic (Sympatec) with 0.2 bar dispersion pressure. QicPic images were processed using an EQPC (diameter of circle of equal projection area) calculation mode. In the subsequent analysis, particles of size less than 200 μm and 360 μm for MCC and mannitol, respectively, were considered as fines, based on consideration of the unprocessed material size. 2.3. Mass throughput experiment The mass throughput of granules as a function of time was measured using a computerised balance that was placed under the mill. From the
Fig. 2. Experimental (●) and fitted (
Particle size distributions were measured at several time points during one batch milling process and the mass of granules that exited the mill was used to calculate the associated time of milling for each experiment. Breakage rate was calculated for the largest granules based on the change of mass in the topmost size class (from 3.9 to 8.0 mm) using following equation [11] mi ðt Þ ¼ m f expð−Si t Þ
ð2Þ
where mf is the feeding mass at the beginning of milling, Si is breakage rate of i-th size class and t is time of milling. Based on the evolution of the PSD at different milling times, the breakage mode can be determined [12]. 2.5. Self-similar solution Kapur [8] used a self-similar solution of the breakage population balance equation to study the shape of the breakage rate function and the breakage mode where self-preserving distributions of diverse dynamic distributions collapse into a single time-invariant curve. Cumulative distribution of scaled particle size by the particle size at 80% of cumulative distribution was fitted with the following function: " !a # b xΓ 1þb a Γ ; a Γ ba ∫ zðxÞ ¼ 1− b Γ a
ð3Þ
where parameters a and b describe the size dependency of the breakage rate function, Si = Axa, and fragment distribution function, bi,j = (xi/xj)b, respectively, where xi and xj represent daughter and mother particle size. If parameter b is higher than 1 there is more impact breakage while if parameter b is smaller than 1 the abrasion mode is predominant. From the values of breakage rates obtained from the topmost size class analysis using Eq. (2) the parameter A in breakage rate function was calculated. These parameters can be used directly in a population balance model to simulate the milled size distribution.
) milling mass throughput for MCC with 24% porosity when batch milling with 150 rpm impeller speed and a) 2 mm and b) 1 mm screen size.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
101
Fig. 3. Milling time for MCC (○) and mannitol (×) when milling ribbons with a) 24% porosity, 2 mm screen size and different impeller speeds; b) 24% porosity, 150 rpm impeller speed and different screen sizes; c) different porosities, 150 rpm impeller speed and 1.5 mm screen sizes; and d) 24% porosity, 150 rpm impeller speed, 1.5 mm screen sizes and different input ribbon mass.
2.6. Three point bend test (3PBT)
3. Results
To understand the influence of porosity on the mechanical properties of the ribbons, rectangular specimens were manufactured that mimic ribbons geometrically, while having a more homogeneous density distribution. Rectangular specimens with the dimensions: 15 mm in length, 6 mm in width and 2–3.5 mm in thickness of different porosities were produced with a compaction simulator (ESH Compaction Simulator). The tensile strength of specimens of different porosities was determined using a three point bend test (3PBT, Texture Analyser). The loading cell of the Texture analyser measured the force that was needed to break the specimen and was used to calculate tensile strength with the following equation:
3.1. Study of breakage rate
σ¼
3Fl 2
2bh
ð4Þ
where F is force applied to break the specimen, l is the length between the supports (10.4 mm), b is the width of specimen (6 mm) and h is the thickness of the specimen.
To study breakage rate, several different approaches were used: measurement of the mass throughput and analysis of the change of mass in the topmost size class in PSD plots during one milling process from where the exact values of breakage rate were obtained, and fitting of the self-similar solution that gave information about the size dependency of breakage rate. 3.2. Determination of breakage rate from mass throughput experiment The milling kinetic profile for MCC ribbons was found to be dependent on screen size of milling mesh and can be divided into two regimes: when screen size was 1.25 mm or larger mass throughput was predominantly linear in time before tending towards the plateau value when the milling process ended (Fig. 2a), whereas when milling with a 1 mm screen the mass throughput was described using Eq. (1) (Fig. 2b), where m∞ is 42 g and NC is 407 when milling MCC with 24% porosity and 150 rpm impeller speed. However, milling of mannitol
Fig. 4. a) Dependences of milling time and amount of fines on screen size, where ◊ represents milling time and + represents fraction of fines and b) dependences of d90 on screen size during milling of MCC ribbons with 24% porosity and 150 rpm impeller speed.
102
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
Fig. 5. a) PSD plots of particles inside and outside the mill at several time points during the milling process and b) decreasing fraction of granules from the topmost size class in time with fitted curve of Eq. (2) for MCC with 22% porosity, and milling process condition: 1 mm screen size and 150 rpm impeller speed.
ribbons resulted only in linear mass throughput for all sizes of mesh apertures. We defined the milling time as the time when 90% of milled material has exited the mill. From Fig. 3 it is evident that milling time is longer when MCC ribbons are milled compared to mannitol ribbons with the same porosity and the same milling settings. The most significant difference in milling time between MCC and mannitol ribbons was observed for the 1 mm screen size mesh where milling time was 7.3 times longer for MCC than for mannitol ribbons with the same porosity when milling with 150 rpm impeller speed (Fig. 3b). This is due to a different kinetic profile at 1 mm screen size between both materials, for mannitol ribbons there is linear mass throughput (i.e. a constant rate) whereas milling kinetic of MCC ribbons is described with Eq. (1) (i.e. the rate is proportional to the mass of material in the mill). For both materials milling time decreases with increasing impeller speed and screen size of milling mesh (Fig. 3a and b). Milling time does not change significantly when milling the mannitol ribbons of different porosities while it decreases with increasing porosity for MCC ribbons (Fig. 3c). Milling time significantly increases with increasing input ribbon mass (Fig. 3d). Fig. 4a indicates a nonlinear decrease of milling time with increasing screen sizes of milling mesh for MCC ribbons. The behaviour is similar for amount of fines produced during milling with the same process settings (Fig. 4a), where a significantly larger amount of fines is produced when milling with 1 mm screen size compared to larger screen sizes. However, at the same milling conditions the d90 (the diameter of particle at 90% in the cumulative distribution) is linearly dependent on screen size (Fig. 4b).
Breakage rates of the topmost size class disappearance plots were calculated for milling processes of MCC and mannitol ribbons with different porosities. From Fig. 6 it is evident that breakage rate increases with increasing ribbon porosity and it is higher for mannitol than for MCC ribbons for equivalent porosity. 3.4. Study of breakage rate with self-similar solution The self-similar solution of the breakage population balance equation presented by Kapur [8], uses a size dependent breakage rate function with parameter a representing the breakage rate size dependency power. Fitting Eq. (3) to the cumulative PSDs, parameter a was determined for MCC and mannitol with 95% confidence bounds. The values were below one and decreased with increasing porosity (Fig. 7), suggesting that with higher ribbon porosity smaller particles break more frequently than bigger particles. 4. Studying breakage mode If the breakage rate can be exactly determined with various experimental approaches like milling mass throughput experiment and from the changes in topmost size class in PSD plots, determination of the breakage mode is more challenging. With the approaches that are described here only an estimate of the breakage mode can be achieved, giving us information which breakage mode (abrasion or impact) is predominant or how different process parameters or material properties influence breakage mode. 4.1. Studying breakage mode from changes in PSD during milling
3.3. Determination of breakage rate from the changes in the topmost size class of PSD plots For MCC ribbons with 22% porosity and milling parameters of 1 mm screen size and 150 rpm impeller speed, the breakage rate of topmost size class was 0.03 s− 1 that is determined from fitting Eq. (2) to the points in Fig. 5b.
Fig. 6. Breakage rate dependency on porosity of MCC (○) and mannitol (×) ribbons for milling at 150 rpm impeller speed and 1 mm screen size.
The breakage mode was studied by examining the evolution of particles and fragments inside the mill during the milling process. From PSD plots in Fig. 5a we can observe the breakup of initial MCC ribbons of approximately 6 mm size into intermediate sized particles with a diameter of approximately 3 mm as a result of a impact breakage mode.
Fig. 7. Dependence of parameter a from Eq. (3) on ribbon porosity for MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
103
Detailed insight into the breakage mode was obtained by examining how different fill levels of the mill chamber affect the mill performance where the experiments with different input ribbon mass were preformed. From Fig. 10 it is evident that both the fraction of fines and d90 of milled granules don't depend on the media fill level whereas the milling time linearly increases with increasing input ribbon mass (Fig. 3d). 4.3. Studying breakage mode with self-similar solution
Fig. 8. PSD plots of particles inside plus outside the mill at several time points during milling process for mannitol ribbons with 28% porosity, and milling process condition of 1 mm screen size and 150 rpm impeller speed.
From fitting the cumulative PSDs to the self-similar solution of the breakage population balance equation (Eq. (3)), the parameter b was determined that describes the breakage mode. From Fig. 11 it is evident that parameter b is decreasing with increasing ribbon porosity for both MCC and mannitol, suggesting that with higher ribbon porosities there is more abrasion breakage during milling. When milling MCC ribbons the parameter b always has a value above 1, suggesting the fracture breakage mode is predominant. However, when milling mannitol ribbons with 28% porosity the parameter b has a value of 0.86, therefore during this milling process more abrasion took place. 5. Tensile strength Tensile strength was determined using Eq. (4) for MCC and mannitol for different porosities as shown in Fig. 12. The values of tensile strength of different porosities of specimen followed exponential behaviour. 6. Discussion
Fig. 9. Dependence of fraction of fines on different porosity of MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
There were also some fines produced, as observed from the PSD in small size classes, that is characteristic of the abrasion mode. The broad range of particle sizes suggests that breakage is caused by complex breakage events where both breakage modes, abrasion and impact, are involved. When comparing the PSD plots of the all the particles during one milling run (inside plus outside the mill) between mannitol (Fig. 8) and MCC (Fig. 5a) more particles of an intermediate size are observed when milling MCC ribbons, suggesting more impact mode.
4.2. Studying breakage mode from amount of fines produced during milling Production of fines during the milling process is characteristic for an abrasion mode. From Fig. 9 it is evident that the amount of fines doesn't change when milling MCC ribbons of different porosities at 1 mm screen size and 150 rpm impeller speed, whilst the amount of fines increases significantly with increasing porosities of mannitol ribbons when milling at the same milling conditions.
Different experimental and analytical approaches allowed us to study the breakage rate and breakage mode of two pharmaceutical materials, i.e. MCC and mannitol. With the milling mass throughput method the milling times were obtained for batch screen milling processes using different settings and different mechanical properties of ribbons. The breakage rate that was determined by mass throughput that was found to follow first order kinetics when milling MCC ribbons with 1 mm screen, similarly as found in many studies of short grinding times [11,13,14]. However, for larger screen sizes milling kinetics was constant (Fig. 2a). The finer mesh not only increases the residence time by particles having to undergo further size reduction before passing through but also presents a more abrasivecontact surface for particles due to the reduced size of holes in the metal surface increasing the probability of hitting a metal edge. For MCC ribbons, a significant increase in milling time was observed when milling with 1 mm screen size compared to the larger screen sizes and a similar trend was observed for the fraction of fines (Fig. 4a). The linear dependency of d90 over screen size, in conjunction with the non-linear change of milling time and fraction of fines over screen size, suggests a change in the overall breakage mode and PSD. Production of fines during the milling process is characteristic of an abrasion mode, therefore we can suggest that there is more abrasive breaking when milling MCC ribbons with smaller
Fig. 10. Effect of changing input ribbon mass on a) amount of fines and b) d90 during milling MCC (○) and mannitol (×) ribbons with 24% porosity, 1.5 mm screen size and 150 rpm impeller speed.
104
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
Fig. 11. Dependence of parameter b on ribbon porosity using Kapur function to fit PSD data of MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
Fig. 13. A volume averaged value of sphericity when milling MCC ribbons with 24% porosity, 150 rpm impeller speed and different screen sizes.
screen sizes compared to the larger screens and it takes a longer time to mill ribbons with abrasion compared to impact breakage. Sphericity describes the ratio between the area of the particle image and its perimeter, where for an ideal sphere its value reaches 1 otherwise it is smaller than 1. Fig. 13 shows a volume averaged value of sphericity when milling MCC ribbons at different screen sizes. It is evident that with decreasing screen size and therefore increasing residence time of particles inside the mill the sphericity of particles that exit the mill increases. This suggests that prolonged exposure to the milling environment leads to rounding of the particles and creates a distribution of smaller sized particles or fines due to attrition, similarly as found by Kaya et al. [15]. Moreover, the amount of fines was larger for mannitol granules than for MCC granules with the same ribbon porosity and when milling with the same milling settings (Fig. 10a), which is consistent with the fitted values of parameter b for both materials that indicates more attrition breakage for mannitol ribbons compared to MCC ribbons of 24% porosity. At the same porosity, MCC ribbons have a higher tensile strength than mannitol (Fig. 12), suggesting higher tensile strength contributes to lower amount of fines produced during milling. When milling MCC ribbons the milling time decreased with increasing ribbon porosity (Fig. 3c) while the breakage rate of the topmost size class increases with increasing ribbon porosity (Fig. 4). This indicates that milling time and breakage rate of the topmost size class describe breakage rate in similar manner and suggests a consistent breakage mode across the range of porosities studied. However, when milling mannitol ribbons the milling time does not depend significantly on ribbon porosity (Fig. 3c), while the breakage rate of the topmost size class increases significantly with increasing ribbon porosity (Fig. 6). Additionally, parameter a from the self-similar model decreases with increasing porosity, suggesting that larger mannitol particles break more slowly with increasing ribbon porosity compared to MCC particles (Fig. 7). These observations suggests a change of breakage mode towards abrasion when milling mannitol ribbons of higher porosities. The change of breakage mode was confirmed by observing the increase
in amount of fines with increasing ribbon porosity for mannitol (Fig. 9) and the change in parameter b of the self-similar model that indicates a change of breakage mode from impact at lower ribbon porosities toward abrasion at higher ribbon porosities (Fig. 11). In contrast, for MCC ribbons the parameter a decreases only slightly with increasing ribbon porosity, and the fraction of fines and parameter b are nearly constant, suggesting that the breakage mode is not changing across the ribbon porosities studied. It was proposed by Berthiaux et al. [11] that the fragment distribution function (described using the parameter b in the case of the Kapur function) can be derived from the breakage rate function for short grinding times. Our study shows that analysis of breakage rate using the approaches described can be used to infer the underlying breakage mode. In this study we also investigated the temporal change of fines during the milling process. We measured the PSD of granules that exit the mill at several time points during an experiment. From Fig. 14 a much higher fraction of fines is observed for mannitol granules at the begining of the milling process compared to the final PSD. This can be explained with a longer residence time for granules to reach a size that is small enougth to exit the mill whilst the fines that are generated by attrition have minimal residence time and therefore represents a higher proportion exiting of the mill at the begining of milling process. This behaviour is expected to be more characteristic for abrasion breakage where more fines are produced and size reduction for coarse particles takes a longer time. However, for MCC ribbons the fraction of fines do not change durring milling, indicating that the breakage mode is predominantly impact. We showed that different occupancy of the mill chamber doesn't affect the granule size distribution (Fig. 10), suggesting that most of the breakage occurs with ribbon-impeller contacts and not with ribbonribbon contacts. It may be concluded that the load in the mill has little influence on the milling kinetics over the range of loads used in our experiments, similar to the findings of Dodds et al. [16]. Contrary to this, Shoji et al. [17] found that there exists an optimal ball fill for maximum breakage within the ball mill, where below this value too few grinding
Fig. 12. Tensile strength vs. porosities for a) MCC and b) mannitol specimens.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
105
Actions) of the European Union's Seventh Framework Programme FP7/2007-2013/ under REA grant agreement No. 316555.
References
Fig. 14. Changes of amount of fines produced at several time points during milling of MCC (○) and mannitol (×) ribbons with 24% porosity and 150 rpm impeller speed and 1 mm screen size.
balls limit the number of contacts between media and material, and above this value contacts between media and material are limited by too many contacts between media and other media. 7. Conclusion A batch oscillating screening mill was used to evaluate the effect of various processing parameters on the breakage rate and breakage mode for the milling of roller compacted ribbons of two different pharmaceutical materials, MCC and mannitol. Information about milling kinetics and breakage modes was obtained using different experimental approaches. From the milling mass throughput experiment the milling time for specific process conditions was determined and correlated to different material properties and process settings. Furthermore, the PSD plots were analysed at several time points during milling process, giving detailed insight to the breakage mode and how different materials behave during milling. Acknowledgements This work was supported by the IPROCOM Marie Curie initial training network, funded through the People Programme (Marie Curie
[1] L. Vogel, W. Peukert, Breakage behaviour of different materials - construction of mastercurve for the breakage probability, Powder Technol. 129 (2003) 101–110. [2] M. Meier, E. John, D. Wieckhusen, W. Wirth, W. Peukert, Influence of mechanical properties on impact fracture: prediction of the milling behaviour of pharmaceutical powders by nanoindentation, Powder Technol. 188 (2009) 301–313. [3] M.G. Herting, P. Kleinebudde, Roll compaction/dry granulation: effect of raw material particle size on granule and tablet properties, Int. J. Pharm. 338 (2007) 110–118. [4] J. Sakwanichol, S. Puttipipatkhachorn, G. Ingenerf, P. Kleinebudde, Roll compaction/ dry granulation: comparison between roll mill and oscillating granulator in dry granulation, Pharm. Dev. Technol. 17 (2012) 30–39. [5] J.E. Byers, G.E. Peck, The effect of mill variables on a granulation milling process, Drug Dev. Ind. Pharm. 16 (1990) 1761–1779. [6] C. Varinot, S. Hiltgun, M.-N. Pons, J. Dodds, Identification of the fragmentation mechanisms in wet-phase fine grinding in a stirred bead mill, Chem. Eng. Sci. 52 (1997) 3605–3612. [7] C. Bazin, G. Lavoie, Ball-mill rotation speed and rate of particle breakage: results for a full-scale unit, Miner. Process. Extr. Metall. (Trans. Inst. Min. Metall. C) 109 (2000) 161–164. [8] P.C. Kapur, Self-preserving size spectra of comminuted particles, Chem. Eng. Sci. 27 (1972) 425–431. [9] J.M. Menacho, Some solutions for the kinetics of combined fracture and abrasion breakage, Powder Technol. 49 (1986) 87–95. [10] S. Yu, M. Adams, B. Gururajan, G. Reynolds, R. Roberts, C.-Y. Wu, The effects of lubrication on roll compaction, ribbon milling and tabletting, 5th International Granulation Workshop, 86 2013, pp. 9–18. [11] H. Berthiaux, C. Varinot, J. Dodds, Approximate calculation of breakage parameters from batch grinding tests, Chem. Eng. Sci. 51 (1996) 4509–4516. [12] S.L.A. Hennart, W.J. Wildeboer, P. van Hee, G.M.H. Meesters, Identification of the grinding mechanisms and their origin in a stirred ball mill using population balances, Chem. Eng. Sci. 64 (2009) 4123–4130. [13] C. Varinot, H. Berthiaux, J. Dodds, Prediction of the product size distribution in associations of stirred bead mills, Powder Technol. 105 (1999) 228–236. [14] P.C. Kapur, Kinetics of batch grinding, Trans. Metall. Soc. AIME 247 (1970) 309–313. [15] E. Kaya, R. Hogg, S.R. Kummar, Particle shape modifications in comminution, Kona 20 (2002) 188–195. [16] H. Berthiaux, J. Dodds, Modelling fine grinding in a fluidized bed opposed jet mill: part I: batch grinding kinetics, Powder Technol. 106 (1999) 78–87. [17] K. Shoji, S. Lohrasb, L.G. Austin, The variation of breakage parameters with ball and powder loading in dry ball milling, Powder Technol. 25 (1980) 109–114.
Contents lists available at ScienceDirect
Powder Technology journal homepage: www.elsevier.com/locate/powtec
Determination of breakage rate and breakage mode of roller compacted pharmaceutical materials Andreja Mirtič, Gavin K. Reynolds ⁎ Pharmaceutical Technology and Development, AstraZeneca, Macclesfield SK10 2NA, UK
a r t i c l e
i n f o
Article history: Received 14 December 2015 Received in revised form 9 April 2016 Accepted 24 April 2016 Available online 26 April 2016 Keywords: Oscillating mill Milling kinetic Breakage mode Fraction of fines
a b s t r a c t Roller compaction is a common unit operation in the manufacture of oral solid dose pharmaceutical products. The roller compactor can produce intermediate ribbons of compacted material that exhibit a range of mechanical properties. Both the breakage rate and the breakage mode of these intermediates will have an effect on the final particle size distribution (PSD), and are therefore important parameters in determining the performance of the final product in pharmaceutical manufacturing. The breakage rates of roller compacted ribbons of two pharmaceutical excipients, microcrystalline cellulose (MCC) and mannitol, were determined with two different approaches, i.e. by analysing the milling mass throughput and the change in mass in the largest granule size class. The self-similar solution of the breakage population balance equation provided an insight into size dependencies of breakage rates and the prevailing breakage mode for specific process parameters. The breakage rate was found to be dependent on the impeller speed, milling mesh aperture size and ribbon porosity, but not on grinding media fill level. For MCC ribbons the milling kinetic profile changed with the mesh aperture size, suggesting dependence of the breakage mode on screen size. Our work shows clear correlation between the amount of fines produced during milling and the underlying breakage mode. For the materials studied, it is evident that mannitol ribbons tend more toward an attrition breakage mode than MCC ribbons of equal porosity when milling with the same process settings. © 2016 Elsevier B.V. All rights reserved.
1. Introduction Milling is one of the most extensively used unit operations in pharmaceutical manufacturing. Understanding how milling parameters affect size reduction of particles is extremely important in achieving a desired particle size distribution, which allows better product uniformity, optimization of formulation dissolution properties and improves bioavailability of the product. Despite extensive literature on the subject of milling, mechanistic insight into the milling process of pharmaceutical material still remains poorly understood. It is already established that the size reduction behaviour of particles not only depends on the material properties, but also on the milling type and process settings. In order to develop a predictive methodology for modelling size reduction processes of pharmaceutical material, the correlation among those characteristics have to be understood and quantified [1,2], in particular with respect to breakage rate and particle size distribution (PSD) of milled material. In dry granulation processes of pharmaceutical materials the oscillating mill is a commonly used method to produce desired granules from roll compacted ribbons [3]. The principle of an oscillating granulator is to mechanically pass compacted material through a wire mesh screen using an oscillating rotor. Sakwanichol et al. [4] evaluated the granule properties obtained from an oscillating granulator using ⁎ Corresponding author. E-mail address: [email protected] (G.K. Reynolds).
http://dx.doi.org/10.1016/j.powtec.2016.04.033 0032-5910/© 2016 Elsevier B.V. All rights reserved.
different oscillating-granulator parameters, i.e. rotor speed, oscillating angle, aperture of mesh screen and rotor type. They concluded that all the examined parameters significantly affect the granule size distribution. Moreover, it was reported that screen aperture size has the largest effect on milling time and work as well as the degree of particle size reduction whereas the impeller side arm shape has the largest effect on overall milling performance in terms of milling efficiency and energy usage [5]. In general, breakage can be achieved by impact or attrition, depending on material properties, equipment design and process parameters. Any or both of the above breakage modes may occur independently or simultaneously [6]. Bazin and Lavoie [7] found that milling proceeds by attrition when operating in the rolling and cascading regime, as the low energy intensity collisions chip off weaker edges and corners of particles. Several methods have been developed for determination of particle breakage behaviour during milling process [6,8,9]. The aim of the current study was to determine the breakage rate of roll compacted ribbons on the underlying breakage modes during oscillating batch milling using several different approaches. In addition, the aim was to correlate the influence of process parameters and material properties, such as ribbon porosity, tensile strength, impeller speeds, input ribbon mass and screen size on the milling performance of two pharmaceutical excipients that exhibit very different mechanical properties. Specifically, microcrystalline cellulose (MCC) was selected as a material that exhibits ductile behaviour and mannitol was selected as it exhibits more brittle behaviour.
100
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
literature the exponential mass throughput was reported for batch screen milling using following equation [10]: N m ¼ m∞ 1− exp − Nc
ð1Þ
where N is the number of cycles, m and m∞ are the mass of granules for N = N and N = ∞, respectively. Nc is a characteristic number of milling cycles at which m = 1 − e−1. Smaller value of Nc corresponds to more rapid breakup of the ribbons, and a larger value corresponds to a lower fracture rate of the ribbons [10]. 2.4. Study of topmost size class of PSD plots Fig. 1. Schematic representation of the oscillating granulator.
2. Experimental methods and methodology 2.1. Ribbon production Microcrystalline cellulose (MCC) Avicel PH101 (FMC BioPolymer) and mannitol Pearlitol 200SD (Roquette)Were used to manufacture ribbons using an AlexanderWerk BT120 (Alexanderwerk, Germany) roller compactor with roll width 25 mm and smooth roll surface. To achieve different ribbon porosities, the roll force and screw speed of the roller compactor were changed. Roll gap (2.5 mm) and roll speed (3 rpm) were kept constant during experiments. Ribbon density was measured with a Geopyc (Micromeritics Ltd., US) with three repetitions. To calculate ribbon porosity the following true densities of materials were used: 1.57 and 1.47 g/cm3 for MCC and mannitol, respectively, as determined by AccPyc 1330 Pycnometer. 2.2. Milling and particle size analysis Before milling experiments all fines were removed from the ribbons. Ribbons were cut into smaller pieces and 50 g of ribbons were introduced into the mill for each experimental setup. The milling process was performed with a Frewitt screen mill (OscilloWitt) using oscillating mode that is schematically presented in Fig. 1. The particle size distribution (PSD) was measured using a QicPic (Sympatec) with 0.2 bar dispersion pressure. QicPic images were processed using an EQPC (diameter of circle of equal projection area) calculation mode. In the subsequent analysis, particles of size less than 200 μm and 360 μm for MCC and mannitol, respectively, were considered as fines, based on consideration of the unprocessed material size. 2.3. Mass throughput experiment The mass throughput of granules as a function of time was measured using a computerised balance that was placed under the mill. From the
Fig. 2. Experimental (●) and fitted (
Particle size distributions were measured at several time points during one batch milling process and the mass of granules that exited the mill was used to calculate the associated time of milling for each experiment. Breakage rate was calculated for the largest granules based on the change of mass in the topmost size class (from 3.9 to 8.0 mm) using following equation [11] mi ðt Þ ¼ m f expð−Si t Þ
ð2Þ
where mf is the feeding mass at the beginning of milling, Si is breakage rate of i-th size class and t is time of milling. Based on the evolution of the PSD at different milling times, the breakage mode can be determined [12]. 2.5. Self-similar solution Kapur [8] used a self-similar solution of the breakage population balance equation to study the shape of the breakage rate function and the breakage mode where self-preserving distributions of diverse dynamic distributions collapse into a single time-invariant curve. Cumulative distribution of scaled particle size by the particle size at 80% of cumulative distribution was fitted with the following function: " !a # b xΓ 1þb a Γ ; a Γ ba ∫ zðxÞ ¼ 1− b Γ a
ð3Þ
where parameters a and b describe the size dependency of the breakage rate function, Si = Axa, and fragment distribution function, bi,j = (xi/xj)b, respectively, where xi and xj represent daughter and mother particle size. If parameter b is higher than 1 there is more impact breakage while if parameter b is smaller than 1 the abrasion mode is predominant. From the values of breakage rates obtained from the topmost size class analysis using Eq. (2) the parameter A in breakage rate function was calculated. These parameters can be used directly in a population balance model to simulate the milled size distribution.
) milling mass throughput for MCC with 24% porosity when batch milling with 150 rpm impeller speed and a) 2 mm and b) 1 mm screen size.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
101
Fig. 3. Milling time for MCC (○) and mannitol (×) when milling ribbons with a) 24% porosity, 2 mm screen size and different impeller speeds; b) 24% porosity, 150 rpm impeller speed and different screen sizes; c) different porosities, 150 rpm impeller speed and 1.5 mm screen sizes; and d) 24% porosity, 150 rpm impeller speed, 1.5 mm screen sizes and different input ribbon mass.
2.6. Three point bend test (3PBT)
3. Results
To understand the influence of porosity on the mechanical properties of the ribbons, rectangular specimens were manufactured that mimic ribbons geometrically, while having a more homogeneous density distribution. Rectangular specimens with the dimensions: 15 mm in length, 6 mm in width and 2–3.5 mm in thickness of different porosities were produced with a compaction simulator (ESH Compaction Simulator). The tensile strength of specimens of different porosities was determined using a three point bend test (3PBT, Texture Analyser). The loading cell of the Texture analyser measured the force that was needed to break the specimen and was used to calculate tensile strength with the following equation:
3.1. Study of breakage rate
σ¼
3Fl 2
2bh
ð4Þ
where F is force applied to break the specimen, l is the length between the supports (10.4 mm), b is the width of specimen (6 mm) and h is the thickness of the specimen.
To study breakage rate, several different approaches were used: measurement of the mass throughput and analysis of the change of mass in the topmost size class in PSD plots during one milling process from where the exact values of breakage rate were obtained, and fitting of the self-similar solution that gave information about the size dependency of breakage rate. 3.2. Determination of breakage rate from mass throughput experiment The milling kinetic profile for MCC ribbons was found to be dependent on screen size of milling mesh and can be divided into two regimes: when screen size was 1.25 mm or larger mass throughput was predominantly linear in time before tending towards the plateau value when the milling process ended (Fig. 2a), whereas when milling with a 1 mm screen the mass throughput was described using Eq. (1) (Fig. 2b), where m∞ is 42 g and NC is 407 when milling MCC with 24% porosity and 150 rpm impeller speed. However, milling of mannitol
Fig. 4. a) Dependences of milling time and amount of fines on screen size, where ◊ represents milling time and + represents fraction of fines and b) dependences of d90 on screen size during milling of MCC ribbons with 24% porosity and 150 rpm impeller speed.
102
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
Fig. 5. a) PSD plots of particles inside and outside the mill at several time points during the milling process and b) decreasing fraction of granules from the topmost size class in time with fitted curve of Eq. (2) for MCC with 22% porosity, and milling process condition: 1 mm screen size and 150 rpm impeller speed.
ribbons resulted only in linear mass throughput for all sizes of mesh apertures. We defined the milling time as the time when 90% of milled material has exited the mill. From Fig. 3 it is evident that milling time is longer when MCC ribbons are milled compared to mannitol ribbons with the same porosity and the same milling settings. The most significant difference in milling time between MCC and mannitol ribbons was observed for the 1 mm screen size mesh where milling time was 7.3 times longer for MCC than for mannitol ribbons with the same porosity when milling with 150 rpm impeller speed (Fig. 3b). This is due to a different kinetic profile at 1 mm screen size between both materials, for mannitol ribbons there is linear mass throughput (i.e. a constant rate) whereas milling kinetic of MCC ribbons is described with Eq. (1) (i.e. the rate is proportional to the mass of material in the mill). For both materials milling time decreases with increasing impeller speed and screen size of milling mesh (Fig. 3a and b). Milling time does not change significantly when milling the mannitol ribbons of different porosities while it decreases with increasing porosity for MCC ribbons (Fig. 3c). Milling time significantly increases with increasing input ribbon mass (Fig. 3d). Fig. 4a indicates a nonlinear decrease of milling time with increasing screen sizes of milling mesh for MCC ribbons. The behaviour is similar for amount of fines produced during milling with the same process settings (Fig. 4a), where a significantly larger amount of fines is produced when milling with 1 mm screen size compared to larger screen sizes. However, at the same milling conditions the d90 (the diameter of particle at 90% in the cumulative distribution) is linearly dependent on screen size (Fig. 4b).
Breakage rates of the topmost size class disappearance plots were calculated for milling processes of MCC and mannitol ribbons with different porosities. From Fig. 6 it is evident that breakage rate increases with increasing ribbon porosity and it is higher for mannitol than for MCC ribbons for equivalent porosity. 3.4. Study of breakage rate with self-similar solution The self-similar solution of the breakage population balance equation presented by Kapur [8], uses a size dependent breakage rate function with parameter a representing the breakage rate size dependency power. Fitting Eq. (3) to the cumulative PSDs, parameter a was determined for MCC and mannitol with 95% confidence bounds. The values were below one and decreased with increasing porosity (Fig. 7), suggesting that with higher ribbon porosity smaller particles break more frequently than bigger particles. 4. Studying breakage mode If the breakage rate can be exactly determined with various experimental approaches like milling mass throughput experiment and from the changes in topmost size class in PSD plots, determination of the breakage mode is more challenging. With the approaches that are described here only an estimate of the breakage mode can be achieved, giving us information which breakage mode (abrasion or impact) is predominant or how different process parameters or material properties influence breakage mode. 4.1. Studying breakage mode from changes in PSD during milling
3.3. Determination of breakage rate from the changes in the topmost size class of PSD plots For MCC ribbons with 22% porosity and milling parameters of 1 mm screen size and 150 rpm impeller speed, the breakage rate of topmost size class was 0.03 s− 1 that is determined from fitting Eq. (2) to the points in Fig. 5b.
Fig. 6. Breakage rate dependency on porosity of MCC (○) and mannitol (×) ribbons for milling at 150 rpm impeller speed and 1 mm screen size.
The breakage mode was studied by examining the evolution of particles and fragments inside the mill during the milling process. From PSD plots in Fig. 5a we can observe the breakup of initial MCC ribbons of approximately 6 mm size into intermediate sized particles with a diameter of approximately 3 mm as a result of a impact breakage mode.
Fig. 7. Dependence of parameter a from Eq. (3) on ribbon porosity for MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
103
Detailed insight into the breakage mode was obtained by examining how different fill levels of the mill chamber affect the mill performance where the experiments with different input ribbon mass were preformed. From Fig. 10 it is evident that both the fraction of fines and d90 of milled granules don't depend on the media fill level whereas the milling time linearly increases with increasing input ribbon mass (Fig. 3d). 4.3. Studying breakage mode with self-similar solution
Fig. 8. PSD plots of particles inside plus outside the mill at several time points during milling process for mannitol ribbons with 28% porosity, and milling process condition of 1 mm screen size and 150 rpm impeller speed.
From fitting the cumulative PSDs to the self-similar solution of the breakage population balance equation (Eq. (3)), the parameter b was determined that describes the breakage mode. From Fig. 11 it is evident that parameter b is decreasing with increasing ribbon porosity for both MCC and mannitol, suggesting that with higher ribbon porosities there is more abrasion breakage during milling. When milling MCC ribbons the parameter b always has a value above 1, suggesting the fracture breakage mode is predominant. However, when milling mannitol ribbons with 28% porosity the parameter b has a value of 0.86, therefore during this milling process more abrasion took place. 5. Tensile strength Tensile strength was determined using Eq. (4) for MCC and mannitol for different porosities as shown in Fig. 12. The values of tensile strength of different porosities of specimen followed exponential behaviour. 6. Discussion
Fig. 9. Dependence of fraction of fines on different porosity of MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
There were also some fines produced, as observed from the PSD in small size classes, that is characteristic of the abrasion mode. The broad range of particle sizes suggests that breakage is caused by complex breakage events where both breakage modes, abrasion and impact, are involved. When comparing the PSD plots of the all the particles during one milling run (inside plus outside the mill) between mannitol (Fig. 8) and MCC (Fig. 5a) more particles of an intermediate size are observed when milling MCC ribbons, suggesting more impact mode.
4.2. Studying breakage mode from amount of fines produced during milling Production of fines during the milling process is characteristic for an abrasion mode. From Fig. 9 it is evident that the amount of fines doesn't change when milling MCC ribbons of different porosities at 1 mm screen size and 150 rpm impeller speed, whilst the amount of fines increases significantly with increasing porosities of mannitol ribbons when milling at the same milling conditions.
Different experimental and analytical approaches allowed us to study the breakage rate and breakage mode of two pharmaceutical materials, i.e. MCC and mannitol. With the milling mass throughput method the milling times were obtained for batch screen milling processes using different settings and different mechanical properties of ribbons. The breakage rate that was determined by mass throughput that was found to follow first order kinetics when milling MCC ribbons with 1 mm screen, similarly as found in many studies of short grinding times [11,13,14]. However, for larger screen sizes milling kinetics was constant (Fig. 2a). The finer mesh not only increases the residence time by particles having to undergo further size reduction before passing through but also presents a more abrasivecontact surface for particles due to the reduced size of holes in the metal surface increasing the probability of hitting a metal edge. For MCC ribbons, a significant increase in milling time was observed when milling with 1 mm screen size compared to the larger screen sizes and a similar trend was observed for the fraction of fines (Fig. 4a). The linear dependency of d90 over screen size, in conjunction with the non-linear change of milling time and fraction of fines over screen size, suggests a change in the overall breakage mode and PSD. Production of fines during the milling process is characteristic of an abrasion mode, therefore we can suggest that there is more abrasive breaking when milling MCC ribbons with smaller
Fig. 10. Effect of changing input ribbon mass on a) amount of fines and b) d90 during milling MCC (○) and mannitol (×) ribbons with 24% porosity, 1.5 mm screen size and 150 rpm impeller speed.
104
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
Fig. 11. Dependence of parameter b on ribbon porosity using Kapur function to fit PSD data of MCC (○) and mannitol (×) ribbons after milling with 150 rpm impeller speed and 1 mm screen size.
Fig. 13. A volume averaged value of sphericity when milling MCC ribbons with 24% porosity, 150 rpm impeller speed and different screen sizes.
screen sizes compared to the larger screens and it takes a longer time to mill ribbons with abrasion compared to impact breakage. Sphericity describes the ratio between the area of the particle image and its perimeter, where for an ideal sphere its value reaches 1 otherwise it is smaller than 1. Fig. 13 shows a volume averaged value of sphericity when milling MCC ribbons at different screen sizes. It is evident that with decreasing screen size and therefore increasing residence time of particles inside the mill the sphericity of particles that exit the mill increases. This suggests that prolonged exposure to the milling environment leads to rounding of the particles and creates a distribution of smaller sized particles or fines due to attrition, similarly as found by Kaya et al. [15]. Moreover, the amount of fines was larger for mannitol granules than for MCC granules with the same ribbon porosity and when milling with the same milling settings (Fig. 10a), which is consistent with the fitted values of parameter b for both materials that indicates more attrition breakage for mannitol ribbons compared to MCC ribbons of 24% porosity. At the same porosity, MCC ribbons have a higher tensile strength than mannitol (Fig. 12), suggesting higher tensile strength contributes to lower amount of fines produced during milling. When milling MCC ribbons the milling time decreased with increasing ribbon porosity (Fig. 3c) while the breakage rate of the topmost size class increases with increasing ribbon porosity (Fig. 4). This indicates that milling time and breakage rate of the topmost size class describe breakage rate in similar manner and suggests a consistent breakage mode across the range of porosities studied. However, when milling mannitol ribbons the milling time does not depend significantly on ribbon porosity (Fig. 3c), while the breakage rate of the topmost size class increases significantly with increasing ribbon porosity (Fig. 6). Additionally, parameter a from the self-similar model decreases with increasing porosity, suggesting that larger mannitol particles break more slowly with increasing ribbon porosity compared to MCC particles (Fig. 7). These observations suggests a change of breakage mode towards abrasion when milling mannitol ribbons of higher porosities. The change of breakage mode was confirmed by observing the increase
in amount of fines with increasing ribbon porosity for mannitol (Fig. 9) and the change in parameter b of the self-similar model that indicates a change of breakage mode from impact at lower ribbon porosities toward abrasion at higher ribbon porosities (Fig. 11). In contrast, for MCC ribbons the parameter a decreases only slightly with increasing ribbon porosity, and the fraction of fines and parameter b are nearly constant, suggesting that the breakage mode is not changing across the ribbon porosities studied. It was proposed by Berthiaux et al. [11] that the fragment distribution function (described using the parameter b in the case of the Kapur function) can be derived from the breakage rate function for short grinding times. Our study shows that analysis of breakage rate using the approaches described can be used to infer the underlying breakage mode. In this study we also investigated the temporal change of fines during the milling process. We measured the PSD of granules that exit the mill at several time points during an experiment. From Fig. 14 a much higher fraction of fines is observed for mannitol granules at the begining of the milling process compared to the final PSD. This can be explained with a longer residence time for granules to reach a size that is small enougth to exit the mill whilst the fines that are generated by attrition have minimal residence time and therefore represents a higher proportion exiting of the mill at the begining of milling process. This behaviour is expected to be more characteristic for abrasion breakage where more fines are produced and size reduction for coarse particles takes a longer time. However, for MCC ribbons the fraction of fines do not change durring milling, indicating that the breakage mode is predominantly impact. We showed that different occupancy of the mill chamber doesn't affect the granule size distribution (Fig. 10), suggesting that most of the breakage occurs with ribbon-impeller contacts and not with ribbonribbon contacts. It may be concluded that the load in the mill has little influence on the milling kinetics over the range of loads used in our experiments, similar to the findings of Dodds et al. [16]. Contrary to this, Shoji et al. [17] found that there exists an optimal ball fill for maximum breakage within the ball mill, where below this value too few grinding
Fig. 12. Tensile strength vs. porosities for a) MCC and b) mannitol specimens.
A. Mirtič, G.K. Reynolds / Powder Technology 298 (2016) 99–105
105
Actions) of the European Union's Seventh Framework Programme FP7/2007-2013/ under REA grant agreement No. 316555.
References
Fig. 14. Changes of amount of fines produced at several time points during milling of MCC (○) and mannitol (×) ribbons with 24% porosity and 150 rpm impeller speed and 1 mm screen size.
balls limit the number of contacts between media and material, and above this value contacts between media and material are limited by too many contacts between media and other media. 7. Conclusion A batch oscillating screening mill was used to evaluate the effect of various processing parameters on the breakage rate and breakage mode for the milling of roller compacted ribbons of two different pharmaceutical materials, MCC and mannitol. Information about milling kinetics and breakage modes was obtained using different experimental approaches. From the milling mass throughput experiment the milling time for specific process conditions was determined and correlated to different material properties and process settings. Furthermore, the PSD plots were analysed at several time points during milling process, giving detailed insight to the breakage mode and how different materials behave during milling. Acknowledgements This work was supported by the IPROCOM Marie Curie initial training network, funded through the People Programme (Marie Curie
[1] L. Vogel, W. Peukert, Breakage behaviour of different materials - construction of mastercurve for the breakage probability, Powder Technol. 129 (2003) 101–110. [2] M. Meier, E. John, D. Wieckhusen, W. Wirth, W. Peukert, Influence of mechanical properties on impact fracture: prediction of the milling behaviour of pharmaceutical powders by nanoindentation, Powder Technol. 188 (2009) 301–313. [3] M.G. Herting, P. Kleinebudde, Roll compaction/dry granulation: effect of raw material particle size on granule and tablet properties, Int. J. Pharm. 338 (2007) 110–118. [4] J. Sakwanichol, S. Puttipipatkhachorn, G. Ingenerf, P. Kleinebudde, Roll compaction/ dry granulation: comparison between roll mill and oscillating granulator in dry granulation, Pharm. Dev. Technol. 17 (2012) 30–39. [5] J.E. Byers, G.E. Peck, The effect of mill variables on a granulation milling process, Drug Dev. Ind. Pharm. 16 (1990) 1761–1779. [6] C. Varinot, S. Hiltgun, M.-N. Pons, J. Dodds, Identification of the fragmentation mechanisms in wet-phase fine grinding in a stirred bead mill, Chem. Eng. Sci. 52 (1997) 3605–3612. [7] C. Bazin, G. Lavoie, Ball-mill rotation speed and rate of particle breakage: results for a full-scale unit, Miner. Process. Extr. Metall. (Trans. Inst. Min. Metall. C) 109 (2000) 161–164. [8] P.C. Kapur, Self-preserving size spectra of comminuted particles, Chem. Eng. Sci. 27 (1972) 425–431. [9] J.M. Menacho, Some solutions for the kinetics of combined fracture and abrasion breakage, Powder Technol. 49 (1986) 87–95. [10] S. Yu, M. Adams, B. Gururajan, G. Reynolds, R. Roberts, C.-Y. Wu, The effects of lubrication on roll compaction, ribbon milling and tabletting, 5th International Granulation Workshop, 86 2013, pp. 9–18. [11] H. Berthiaux, C. Varinot, J. Dodds, Approximate calculation of breakage parameters from batch grinding tests, Chem. Eng. Sci. 51 (1996) 4509–4516. [12] S.L.A. Hennart, W.J. Wildeboer, P. van Hee, G.M.H. Meesters, Identification of the grinding mechanisms and their origin in a stirred ball mill using population balances, Chem. Eng. Sci. 64 (2009) 4123–4130. [13] C. Varinot, H. Berthiaux, J. Dodds, Prediction of the product size distribution in associations of stirred bead mills, Powder Technol. 105 (1999) 228–236. [14] P.C. Kapur, Kinetics of batch grinding, Trans. Metall. Soc. AIME 247 (1970) 309–313. [15] E. Kaya, R. Hogg, S.R. Kummar, Particle shape modifications in comminution, Kona 20 (2002) 188–195. [16] H. Berthiaux, J. Dodds, Modelling fine grinding in a fluidized bed opposed jet mill: part I: batch grinding kinetics, Powder Technol. 106 (1999) 78–87. [17] K. Shoji, S. Lohrasb, L.G. Austin, The variation of breakage parameters with ball and powder loading in dry ball milling, Powder Technol. 25 (1980) 109–114.