Comparison of a continuous ring layer wet granulation process with batch high shear and fluidized bed granulation processes

Powder Technology 275 (2015) 113–120

Contents lists available at ScienceDirect

Powder Technology journal homepage: www.elsevier.com/locate/powtec

Comparison of a continuous ring layer wet granulation process with batch high shear and fluidized bed granulation processes Maiju A. Järvinen a,⁎, Marko Paavola b, Sami Poutiainen a, Päivi Itkonen a, Ville Pasanen a, Katja Uljas a, Kauko Leiviskä b, Mikko Juuti c, Jarkko Ketolainen a, Kristiina Järvinen a a b c

School of Pharmacy, Faculty of Health Sciences, University of Eastern Finland, P.O. Box 1627, FI-70211 Kuopio, Finland Control Engineering Laboratory, Department of Process and Environmental Engineering, University of Oulu, P.O. Box 4300 FI-90014, Finland VTT Optical Instruments, VTT Technical Research Centre of Finland, P.O. Box 1199, FI-70211 Kuopio, Finland

a r t i c l e

i n f o

Article history: Received 28 August 2014 Received in revised form 28 January 2015 Accepted 31 January 2015 Available online 8 February 2015 Keywords: Wet granulation Continuous ring layer process High shear process Fluidized bed process PLS model

a b s t r a c t The traditional batch wet granulation processes encounter several challenges, such as problems in the scale-up step, batch-to-batch variability together with the multivariate and difficult to control nature of the process. A continuous wet granulation technique could be a possible solution for the scale-up problem, offering adjustable production volumes with the same equipment. In this study, a continuous ring layer wet granulation process (factors: shaft speed and binder flow rate) was compared with two batch granulation processes: high shear (factors: impeller speed and chopper speed) and fluidized bed (factors: inlet air temperature during granulation and binder flow rate) with formulations consisting of paracetamol, microcrystalline cellulose and polyvinylpyrrolidone. A quantitative PLS model was formed to assess the effects of the process parameters on the granule properties (the mean granule size and flowability). In the case of the continuous ring layer granulation process, the mean granule size increased linearly with increasing shaft speed and binder flow rate, and the granules resembled morphologically more the granules produced by the high shear granulation than by the fluidized bed granulation. It is notable that the continuous ring layer granulation process was easier to control than the fluidized bed and high shear granulation processes due to the linear responses towards changes in operation conditions. Both types of tablets, compressed either from the granules produced by the continuous ring layer granulation or by the high shear granulation, achieved an immediate drug release. In summary, the continuous ring layer granulation process was demonstrated to represent a promising tool for the production of pharmaceutical granules. © 2015 Published by Elsevier B.V.

1. Introduction Wet granulation is a crucial size enlargement process that improves powder flow properties, reduces dustiness and segregation in further processing such as during tableting [1]. Traditionally used batch wet granulation processes, such as high shear and fluidized bed granulation, suffer from the complexity of scale-up and batch-to-batch variability [2,3]. In addition, several process-related factors as well as equipment and material parameters can affect the batch wet granulation processes which means that they are difficult to control [3–5]. The scale-up of a batch process can be costly and complicated. Scale-up strategies of high shear granulation can vary from monitoring of some representative parameter (e.g. power consumption) to the usage of experimental design and population balance modeling. Modeling may require performance of many experiments to be carried out with both small scale and larger scale equipment. In the case of the fluidized bed granulation ⁎ Corresponding author. Tel.: +358 40 355 2041; fax: +358 17 162424. E-mail address: Maiju.Jarvinen@uef.fi (M.A. Järvinen).

http://dx.doi.org/10.1016/j.powtec.2015.01.071 0032-5910/© 2015 Published by Elsevier B.V.

process, the scale-up is rather complicated due to the importance of initial binder distribution together with other input parameters, such as inlet air and binder flow rate [4]. One possible solution to the scale-up issues in wet granulation is to use continuous processing, since here the size of equipment remains constant throughout the manufacturing process development but the manufacturing scale is defined by process throughput and running time [6]. Continuous processes confer advantages, such as decreased expenditure on equipment, premises and operation especially for large volume products [7]. The ability to utilize one-sized equipment minimizes waste during scale-up and the feasibility of testing different process conditions are conducted easily with a continuous process. At present, continuous processes are utilized widely in several industries e.g. in the chemical, food, pulp and paper industry [8–10]. There are reports describing continuous pharmaceutical processes; these include continuous mixing [8,9,11–14] and wet granulation. In fact, there are several continuous wet granulation techniques available for pharmaceutical applications, such as extrusion, instant agglomeration, spray drying and fluid bed agglomeration [15].

114

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

Single screw and twin-screw extruders for pharmaceutical granulation were introduced in the late 1980s and have been developed subsequently in recent years [16–20]. In one study, the semi-continuous extrusion wet granulation was compared with the high shear granulation using formulations containing varying lactose grades and an active pharmaceutical ingredient (paracetamol and cimetidine). In that study, extrusion granulation proved to be more efficient than the high shear granulation for the preparation of the paracetamol granules [21]. Nonetheless, rather few studies have compared the batch and continuous wet granulation processes. The aims of this study were to compare the continuous ring layer wet granulation process with the batch high shear and fluidized bed wet granulation processes and assess the effect of the process parameters on the granule properties. The tested formulations consisted of paracetamol as the drug, microcrystalline cellulose as the excipient and aqueous polyvinylpyrrolidone solution as the binder liquid. 2. Materials and methods 2.1. Materials Paracetamol (d50 (geometrical mean size) 15 ± 0.6 μm) was purchased from Xiamen Forever Green Source Biochem Tech. Co., Ltd (Xiamen, China) and microcrystalline cellulose (Avicel PH101, d50 62 ± 0.5 μm), was purchased from IMCD (Malmö, Sweden). Polyvinylpyrrolidone (Kollidon K25) was purchased from BASF (Ludwigshafen, Germany) and was used as a binder in aqueous solution. 2.2. Preparation of granules A 32 full factorial experimental design of experiments was applied to investigate the effects of the wet granulation process parameters on the properties of the granules (Table 1). All the process runs were conducted in a random order. The required concentration of polyvinylpyrrolidone in the granulation binder solution as well as the levels of the process parameters were determined for each different granulation process in pre-tests before fixing the final formulation (Table 1). 2.2.1. Fluidized bed granulation In the fluidized bed granulation process (Lödige LFP8, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany), the granulation chamber was pre-heated to reach the inlet air temperature of 55, 60 or 65 °C according to the experimental design (Table 1). After preheating, paracetamol and microcrystalline cellulose (total batch size 500 g) were poured into the granulation chamber and the fluidization of air began with an inlet air quantity of 50 m3/h. The inlet air flow was kept constant for the whole process from mixing to drying. Powders were mixed in fluidized bed for 5 min prior to the addition of the binder solution. The binder solution (12.5% (w/w) polyvinylpyrrolidone

in purified water) was added with a peristaltic pump (Watson-Marlow Bredel, UK) through a 1.2 mm nozzle at the rate of 18, 22 or 27 g/min (Table 1). The granulation phase lasted for 23–35 min depending on the run. Liquid to solid ratios in fluidized bed granulation process were 1.22 (18 g/min binder flow rate), 1.21 (22 g/min binder flow rate) and 1.24 (27 g/min binder flow rate), respectively. After the granulation, the granules were dried with the same equipment at a constant inlet air temperature of 60 °C. The drying phase continued until the moisture difference between the inlet and outlet air was under 1.5 g/kg, i.e. 25–71 min depending on the run and the conditions of the ambient room temperature and moisture. The end point moisture varied from 0.9 to 1.4 g/kg, corresponding to the theoretical end point moisture contents of 0.6–0.9% (w/w). 2.2.2. High shear granulation The weighed powder mass was poured into the granulation chamber (volume 5 l) and mixed in the high shear mixer (Lödige MGTL 5/15, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany) for 3 min with an impeller speed of either 400, 500 or 600 rpm corresponding the peripheral speeds of 4.5, 5.7 or 6.8 m/s, respectively (Table 1). Chopper speed was chosen for another factor, with levels of 0, 1500 or 3000 rpm. After mixing, the binder solution (7.5% (w/w) polyvinylpyrrolidone in purified water) was added with a peristaltic pump at a rate of 35 g/min. The granulation phase lasted for 10.5 min in each run. Liquid to solid ratio in high shear granulation process was constant 0.74. After granulation, the granules were transferred to pre-heated fluidized bed dryer with the help of a vacuum. The granules were dried with the fluidized bed apparatus (Lödige LFP8, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany) at the constant inlet air temperature of 60 °C until the moisture difference between inlet and outlet air was under 1.5 g/kg, i.e. 35–70 min. The end point moisture varied from 0.5 to 0.9 g/kg, corresponding to the theoretical end point moisture contents of 0.4–0.8% (w/w). The three best granulation conditions were chosen according to the values of mean granule size (d50, mean granule sizes around 500 μm being the most desirable) and granule flowability (evaluated by Carr's index) and total of three repetition batches for further characterization were made using each of those three best granulation conditions (total of 9 repetition batches from which the granule content uniformity of 6 batches was analyzed). 2.2.3. Continuous ring layer granulation The continuous ring layer granulator (Lödige CoriMix CM5, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany) used here had a high peripheral speed (max. 40 m/s depending on the size of the machine) creating a centrifugal force that caused the granulated material to form a layer on the wall of the granulator, i.e. there is a ring layer and mixing occurs with the high intensity [22]. Equipment properties, such as geometry, fill level and tooling of the rotating blade, affect the residence time of material inside the granulator. The throughput of the granulator can vary from 10 to 80 kg/h making this

Table 1 Wet granulation process conditions, based on 32 full factorial design. Parameter Formulation (% (w/w)) Paracetamol Microcrystalline cellulose Polyvinylpyrrolidone Factor 1

Factor 2 Constant parameters

Fluidized bed

High shear

Continuous ring layer

26.0 60.8 13.2 Granulation inlet Air temperature (55, 60, 65 °C) Binder flow rate (18, 22, 27 g/min) Batch size: 500 g Drying by fluidized bed 60 °C (25–71 min)

28.4 66.3 5.3 Impeller speed (400, 500, 600 rpm)

29.0 67.4–67.8 3.2–3.6 Shaft speed (1000, 1250, 1500 rpm)

Chopper speed (0, 1500, 3000 rpm) Batch size: 500 g Drying by fluidized bed 60 °C (35–70 min)

Binder flow rate (7.4, 8.1, 8.8 kg/h) Powder feed rate: 11.2 kg/h Drying by fluidized bed 60 °C (19–27 min)

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

equipment suitable for use with products being made with different production volumes. A screw-type feeder (Eberhard Bauer, Esslingen, Germany) was used in the continuous ring layer granulation process. Therefore, paracetamol and microcrystalline cellulose were premixed with the high shear mixer (Lödige MGTL 5/15, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany) in 2.5 kg portions with the mixing speed of 450 rpm and the mixing time of 7.5 min. The blended powder mix was then poured into the feeder hopper before starting the process runs in order to avoid disturbances in the powder flow during the granulation process. Due to the operation principle of the ring layer granulator [22], there was a 1.5 min period to allow the formation of a layer of material (i.e. ring layer) inside the granulator wall and attainment of steady state, thereafter the 1 min sample collection started. In the process runs, granulator shaft speed was set either 1000, 1250 or 1500 rpm corresponding the peripheral speeds of 5.0, 6.2 or 7.5 m/s, respectively, and the binder flow rate 7.4, 8.1 or 8.8 kg/h (Table 1). The granulator was operated with the standard blade configuration. The granulation binder used was 5% (w/w) aqueous polyvinylpyrrolidone solution. The powder feed rate was kept constant 11.2 kg/h in every run. Liquid to solid ratios in continuous ring layer granulation process were 0.66 (7.4 kg/h binder flow rate), 0.72 (8.1 kg/h binder flow rate) and 0.79 (8.8 kg/h binder flow rate). Wet granules were collected, weighed and transferred into the pre-heated fluidized bed dryer (Lödige LFP8, Gebrüder Lödige Maschinenbau GmbH, Paderborn, Germany) and dried at the constant inlet air temperature of 60 °C until the moisture difference between inlet and outlet air was under 1.5 g/kg, i.e. 19–27 min. The end point moisture varied from 0.5 to 0.9 g/kg, corresponding to the theoretical end point moisture contents of 0.38–0.43% (w/w). The three best granulation conditions were chosen according to the values of mean granule size (d50, mean granule sizes around 500 μm being the most desirable) and granule flowability (evaluated by Carr's index). 2.3. Characterization of granules 2.3.1. Flowability Bulk and tapped volumes (n = 2) were measured with a tapped density tester (Erweka SVM, Heusenstamm, Germany) weighing 27–80 g granules into 250 ml volumetric flask. The tapping time was 20 min (1260 taps). Bulk (ρbulk) and tapped (ρtapped) density was calculated thereafter, the results presented as Carr's index Eq. (1) [23]. 0

Carr s indexð%Þ ¼ 100% 

ρtapped −ρbulk ρtapped

ð1Þ

2.3.2. Particle size measurements Granules were sieved with Retsch sieve series (4.00; 2.38; 1.68; 1.00; 0.71; 0.42; 0.297; 0.21 and 0.149 mm), at the speed setting of 50 for 20 min (Retsch, Germany). After sieving, those fractions with a granule size of 0.210–1.00 mm (fluidized bed and high shear granulation) or 0.149–1.00 mm (continuous ring layer granulation) were chosen for the further studies. The granule size distributions (n = 6 per process run) were measured by laser diffraction (Mastersizer2000, Malvern Instruments Inc., Southborough, MA) using the dry measurement option (Scirocco2000). 2.3.3. Paracetamol content in granules Based on the flowability and particle size measurements, the paracetamol content of the three best granulation batches and their repetitions (from 6 batches of granules prepared by high shear and 3 runs of granules prepared by continuous ring layer, see Sections 2.2.2–2.2.3) were analyzed. Approximately 1 g (high shear granulation, fractions 0.21–1.00 mm, n = 10) or 400 mg (continuous ring layer granulation, fractions 0.149–1.00 mm, n = 5) sample of granules was dissolved in 12% (v/v) methanol in water binary solution and shaken for 10 min at

115

500 rpm (Kika Labortechnik, KS 125 basic, Kika®-Werke GmbH & Co. KG, Germany). Samples were then centrifuged for 5 min at 3000 rpm (Centrifuge B5, B. Braun International GmbH, Germany) and diluted to meet the concentration of the pre-determined standard curve (5–10 μg/ml). Samples were measured as a triplicate with UV/VIS spectrophotometer (UV-1800, Shimadzu, Shimadzu Suzhou Instuments Wfg. Co., Ltd, China) at the wavelength of 244 nm. The paracetamol concentration was calculated from the absorbance with the aid of predetermined standard curve that was linear within the concentration range from 5 to 10 μg/ml (correlation coefficient R2 = 0.995). 2.3.4. Granule morphology The morphology of the granules produced using the best granulation parameters was viewed with a scanning electron microscope (XL30 ESEM TMP, FEI Company, Netherlands) operated at 15 kV with high vacuum. Prior to scanning, the samples were sputter coated with a layer of gold using an automatic coater device (AGAR-Automatic sputter coater B7341, Agar Scientific Limited, Stansted, UK). 2.3.5. Experimental data analysis: partial least squares (PLS) model and box plots The analysis of the granulation runs was carried out using Modde© software (version 8.0.0.0). First, the collected data obtained by varying the factors, as described in Table 1, were used to estimate the regression coefficients of a PLS model which represented the relationship between the process factors and responses (i.e. d50 values and Carr's indexes). Before modelling, the data were scaled and centered. Interaction plots were examined in order to find out if the effect of one of the factors was dependent on the level of the other. For this, an interaction term was first included in the model. If a strong interaction could not be detected, the term was excluded from the model. Normal probability plots were then investigated to ensure that the residuals were normally distributed and to identify potential outliers (experimental runs, whose residuals were lying outside ±4 standard deviations). The residuals were also compared against the predicted response to detect possible non-constant variance that could justify transform of the response. Finally, the main effect plots were investigated in order to identify the effects of factors on the responses. The plot shows the response predicted by the model when one factor varies from a low to a high level with the other factors in the set-up being kept constant at their averages. Box plots were applied in the comparison of the distributions. In the plots, the tops and bottoms of each box represent the 25th and 75th percentiles of the samples. The line in the middle of the box refers to the sample median. The whiskers, i.e. the lines extending above and below each box, represent the extreme values of the sample. 2.4. Compression of tablets The granules prepared either with the high shear granulator or with the continuous ring layer granulator were compressed into 10 mm diameter flat-faced tablets with an eccentric single station Korsch EK0 DMS (Korsch Berlin, Germany) tablet press. The compression forces used for the granules prepared by high shear were 20–21 kN and for granules prepared by continuous ring layer 10–14 kN in order to achieve fracture force values between 70 and 100 N. The granules were compressed into the tablets as such, without any lubricant. 2.5. Evaluation of tablets 2.5.1. Fracture force and tensile strength measurement For the tensile strength measurements, 20 tablets (high shear granulation) or 10 tablets (continuous ring layer granulation) were picked up randomly. The tablets were weighed (Mettler-Toledo AG 245, Mettler-Toledo GmbH, Greifensee, Switzerland), and the diameter

116

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

(D) and height (h) of the tablets measured with a micrometer (NSK Digitrix, Japan). The radial fracture force (F) of each tablet was determined with the CT5 testing machine (Engineering Systems (NOTTM), Nottingham, UK). Since the tablets were cylindrical and flat-faced, tensile strength (σ, MPa) was calculated from fracture force (F) values with Eq. (2) [24]. σ¼

2F πDh

ð2Þ

2.5.2. Uniformity of dosage units In the uniformity of dosage unit test (Ph.Eur. 2.9.40), 10 tablets were randomly chosen from each entire tablet set which contained about 100 tablets. The tablets were dissolved individually in 5% (v/v) methanol in water binary solvent, shaken for 10 min, allowed to dissolve overnight, then centrifuged and diluted to meet the concentrations of predetermined standard curve solutions (5–10 μg/ml). The absorbance of each sample was measured with a UV/VIS spectrophotometer (Shimadzu UV-1800, Shimadzu Suzhou Instruments, Suzhou Jiangsu, China) at the wavelength of 244 nm. The paracetamol concentration was calculated from the absorbance with the aid of the predetermined standard curve which was linear over the concentration range from 5 to 10 μg/ml (correlation coefficient R2 = 0.993). The acceptance values were calculated according to Ph. Eur. 2.9.40 (based on the content uniformity): acceptance values should be ≤ 15% in order to fulfill the requirements of European Pharmacopoeia [23]. 2.5.3. Tablet dissolution Dissolution studies were performed in 900 ml phosphate buffer solution (USP) pH 7.4, at 37 °C, using the paddle method (Distek 2100C, Distek, North Brunswick, NJ, USA) with a paddle rotation speed of 50 rpm. Six tablets (compressed from the granules produced either by the high shear or the continuous ring layer granulator) were randomly picked out for dissolution testing. Samples (5 ml) were removed at 5, 10, 15, 20, 30, 45, 60, 120, 180 and 240 min time points and the sampling volume was replaced with an equal volume of the buffer solution. The samples were filtered through 0.45 μm membrane filters (Whatman FP30/0.45 CA-S, Whatman GmbH, Dassel, Germany) and diluted prior to the absorbance measurements so that they would lie within the concentration range of the standard curve (5–10 μg/ml). The absorbance of each sample was measured in a UV/VIS spectrophotometer (Shimadzu UV-1800, Shimadzu Suzhou Instruments, Suzhou Jiangsu, China) at the wavelength of 244 nm. The paracetamol concentration was calculated from the absorbance with the aid of predetermined standard curve that was linear within the concentration range from 5 to 10 μg/ml (correlation coefficient R2 = 0.998).

analysis, no data transformation was required and a strong interaction between the factors was not observed in case of either one of the responses. The modeling results imply that d50 depends on the granulation temperature (p = 0.06) (Fig. 1): an increase in the inlet air temperature from 55 to 65 °C increased the value of d50 from 280 to 305 μm. The binder flow rate did not seem to have any impact on d50 (p = 0.3). However, a large part of the observed variance remains unexplained by the model (R2 = 0.32 and Q2 = 0.15). Therefore, the effects of both the factors were examined more closely using box plots (Fig. 2). The results of the box plot (Fig. 2) are in accordance with the PLS analysis. Based on the distributions obtained from the experimental runs, the binder flow rate does not have any significant effect on the value of d50 (Fig. 2b). However, the result implies that the binder flow rate could influence the variation of the d50 obtained at the different levels: with the medium binder flow rate, the range of granule sizes obtained was approximately half of that obtained with the low and the high binder flow rates. From the process optimisation point of view, this implies that the binder flow rate could be used as a way of minimizing the variance in the values of d50. As shown in the box plot (Fig. 2a), the granule sizes obtained with the low granulation temperature were different from the cases with the medium and high granulation temperatures since the distributions are not overlapping. Moreover, the dependence between the granule size and the granulation temperature is non-linear for the applied temperature range and formulation. Interestingly, it seems that the growth of the granules reached a maximum at the medium granulation temperature level. The modeling metrics for the Carr's index were R2 = 0.54 and Q2 = 0.40. The increase in the temperature from 55 to 65 °C elevated Carr's index from 10.5% to 13% (p = 0.02) and the increase in the binder flow rate from 18 to 27 g/min reduced Carr's index from 13% to 11% (p = 0.05). The results were confirmed in box plots. To conclude, the range of d50 values (280–305 μm) means that all of the granules produced by fluidized bed process are usable in tableting. Moreover, all fluidized bed granulation batches yielded granules showing a good flowability (Carr's index 10.5–13%), except one batch granulated at 65 °C with the binder flow rate of 18 g/min yielded granules showing fair flowability (Carr's index 16%) [23]. Data analysis, as presented in Figs. 1 and 2, could represent a useful procedure in process optimization in order to save manufacturing costs. For example, it might be possible to perform fluidized bed granulation at a low temperature, using medium binder flow rates, while still achieving the required particle size and flowability with small variation between the batches. Based on the literature, the effect of the granulation temperature on granule growth in fluidized bed granulation is complex: both an increase [25] and a decrease [26] in granule size in response to an increase in granulation temperature have been reported. Instead, the mean

2.5.4. Statistical analysis A non-parametric, Mann–Whitney U test (IBM SPSS Statistics 19, IBM, USA) for independent samples was used to compare the fracture force, the tensile strength and the drug release data of two groups, i.e. tablets were prepared either from the granules produced by the continuous ring layer or by the high shear granulator. A significance level of 0.05 was used. 3. Results and discussion 3.1. Effect of the process parameters on the mean granule size and flowability 3.1.1. Granules prepared by the fluidized bed granulation The effects of the granulation temperature and the binder flow rate on the granule properties (9 batches, Table 1) were analyzed as described above (Section 2.3.5). The residuals were normally distributed and no potential outliers were detected. Additionally, based on the

Fig. 1. Main effects of the granulation inlet air temperature (presented in x-axis −1.0 = 55 °C, 0 = 60 °C, 1.0 = 65 °C) on the mean granule size (d50) (y-axis) in the fluidized bed granulation (confidence level = 0.9). The blue circular symbols represent replicated observations of the temperature, where the other factors (batch size and drying conditions) were held at their averages. For the applied model, R2 = 0.32 and Q2 = 0.15. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

Fig. 2. The effects of the fluidized bed granulation inlet air temperature (a) and the binder flow rate (b) on the granule size (d50) (n = 6). The sample sizes for the low (55 °C), medium (60 °C) and high (65 °C) granulation temperatures are n = 3, n = 5, and n = 7, respectively. The sample sizes for the low (18 g/min), medium (22 g/min) and high (27 g/min) binder flow rates are n = 7, n = 5, and n = 3, respectively. For each bar, the other factor was kept constant whereas the other varied from low to high.

granule size typically increased with increasing binder flow rate in fluidized bed granulation [25–27]. In this study, the binder flow rate did not significantly affect the valued of d50, whereas the temperature seemed to have a small non-linear increasing effect. The yield percentage of the best granule batch, prepared using the granulation inlet air temperature of 65 °C with the binder flow rate of 22 g/min, was low, 46 ± 2.5% (w/w) (mean ± SD, n = 3), that was due to tendency of the powder mass to stick to the walls of a granulator chamber. 3.1.2. Granules prepared by high shear granulation The effects of the impeller and the chopper speeds on the granule properties (9 batches, Table 1) were analyzed with the described procedure (Section 2.3.5). The residual analysis showed that the residuals were almost normally distributed and neither potential outliers nor strong interactions between the factors were detected. It was also concluded that no transformations of responses were required. As in the case of fluidized bed granulation, a large part of the observed variance remains unexplained by the model: R2 and Q2 values were 0.48 and 0.36, respectively, for d50 whereas both R2 and Q2 values were close to zero for Carr's index. The only statistically significant dependence was detected between the chopper speed and the granule size (p = 0.002), as d50 values increased from 370 to 470 μm with increasing the chopper speed from 0 to 3000 rpm (Fig. 3). All the granules produced by high shear possessed excellent flowability properties (Carr's index 7–9%) [23]. The effects of factors on granule properties were confirmed with box plots. The box plots supported the PLS analysis and provided also some additional information. First, the impeller speed seemed to have

Fig. 3. Effect of chopper speed (presented in x-axis −1.0 = 0 rpm, 0 = 1500 rpm, 1.0 = 3000 rpm) on mean granule size (d50) (y-axis) in high shear granulation (confidence level = 0.95). The blue circular symbols represent replicated observations of the chopper speed, where the other factors (the batch size and the drying conditions) are held at their averages. For the applied model, R2 was 0.48 and Q2 0.36. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

117

a significant effect on the variation of the d50 value (Fig. 4a), although the mean size of the granules was not affected. Second, the effect of the chopper speed may actually be non-linear with respect to the values of d50 (Fig. 4b). It was previously reported that with high shear granulation, coarser granules are formed as the impeller speed is increased (300–900 rpm), whereas the chopper speed (1000–2000 rpm) does not affect granule size [28]. In another study, it was also reported a shift in the granule size distribution towards larger granules with the increasing impeller speed (from 500 to 1200 rpm) [29]. In contrast, another study described a change in the granule size distribution towards smaller granules (b 250 μm) at increasing impeller speeds (1000–2000 rpm) [30]. Further, when the effect of a chopper on the granule size was investigated with different main impeller speeds (100–800 rpm), it was noted that the effect of chopper was important with low impeller speeds (100–400 rpm) whereas the effect of chopper was no longer apparent at the highest impeller speed (800 rpm) [31]. The utilization of a chopper might delay the formation of the largest granules [31]. In our study, the granulation process seems to be robust and to tolerate impeller speed changes from 400 to 600 rpm. The robustness of a high shear granulation process regardless of the impeller speed changes of 400–600 rpm can be due to the fact that there is a toroidal flow pattern, i.e. the particle surface velocities remain the same even though the impeller speed is elevated [32]. However, it must be noted that the variation in the values of d50 increased with the increasing impeller speed (Fig. 4a). The observed effect of chopper speed on granule size could be attributable to the low impeller speeds (400–600 rpm) used in our study [31]. All the granulation batches were evaluated for their mean granule size (d50), flowability and granule size distribution. Based on these results, the three best high shear granulation conditions (i.e. impeller speed/chopper speed) were as follows: 400 rpm/3000 rpm; 500 rpm/3000 rpm and 600 rpm/3000 rpm. Further, when the paracetamol content in the granules was included in the evaluation, the best high shear granulation process was performed using the impeller speed of 400 rpm with the chopper speed of 3000 rpm (yield percentage 83 ± 7% (w/w), mean ± SD, n = 3) (Table 2). 3.1.3. Granules from continuous ring layer granulation In the case of the continuous ring layer granulation process, the PLS modelling metrics for d50 (R2 = 0.85 and Q2 = 0.78) and Carr's index (R2 = 0.61 and Q2 = 0.42), were better than those obtained with fluidized bed and high shear granulations. The residuals were normally distributed and no interactions between factors or outliers were detected nor were any transformations needed. As shown in the effects plots, the value of d50 increases linearly with both the shaft speed (p = 0.007) (Fig. 5) and the binder flow

Fig. 4. The effect of impeller (a) and chopper (b) speeds on d50 in high shear granulation. The sample size for low (400 rpm), medium (500 rpm) and high (600 rpm) impeller speed is n = 6. The sample sizes for low (0 rpm), medium (1500 rpm) and high (3000 rpm) chopper speeds are n = 3, n = 3, and n = 12, respectively. For each bar, one factor is kept constant whereas the other factor varies from low to high.

118

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

Table 2 Properties of granules and tablets produced using the best granulation process parameters which are presented after each granulation type.

GRANULES

d50 (μm) ± SD

d10 (μm) ± SD

d90 (μm) ± SD

TABLETS

Carr's index (%) (n = 2) Paracetamol content in granules Limit: 85–115% of label claim Compression force (kN) Fracture force (N) Tensile strength (MPa) Uniformity of dosage units (Ph.Eur. 2.9.40) Acceptance value (%)

High shearImpeller speed 400 rpm, chopper speed 3000 rpm

Continuous ring layer Shaft speed 1250 rpm, binder flow rate 8.1 kg/h

490.8 ± 24.1a 471.3 ± 16.3a 512.6 ± 9.8a 302.6 ± 9.4a 300.6 ± 7.0a 324.1 ± 4.9a 888.5 ± 40.5a 840.6 ± 21.6a 870.8 ± 19.1a 8.8 (excellent) 5.0 (excellent) Passc

582.8 ± 74.8b

20–21 70.1 ± 4.2e⁎ 1.03 ± 0.06e⁎ 10.3f, g (pass)

10–14 93.2 ± 6.5f⁎ 1.38 ± 0.09f⁎ 13.7f, g(pass)

214.7 ± 19.1b

1131.3 ± 147.8b

6.4 (excellent) 6.5 (excellent) Passd

a

n = 6 (3 batch repetitions). n = 6 (1 run). c n = 10. d n = 5. e n = 20. f n = 10. g Acceptance values should be ≤15% to fulfill the requirements of the European Pharmacopoeia (EDQM 2010). ⁎ Statistically significant difference (p b 0.05). b

rate (p = 0.006) (Fig. 6). These results were confirmed in the box plots which showed a linear response (d50) for both the shaft speed and binder flow rate (data not shown). This could be based on the increased collisions and contact between powder particles with the ring layer inside the granulator wall. In the case of Carr's index, the linear dependency of the factors is less obvious. The results suggest that increasing both the binder flow (p = 0.06) and shaft speed (p = 0.08) improves flowability, which could also be observed from the box plots. Overall, Carr's index value decreased from 11.5% to 6% with both the increase in shaft speed from 1000 to 1500 rpm, and with the increase in binder flow rate from 7.4 kg/h to 8.8 kg/h indicating that all the granules met the specification of excellent or good flowability [23]. All the granulation runs were evaluated for their mean granule size (d50), flowability and granule size distribution. Based on these results, the three best continuous ring-layer granulation conditions (i.e. shaft speed/binder flow rate) were as follows: 1250 rpm/8.1 kg/h; 1500 rpm/8.8 kg/h and 1250 rpm/8.8 kg/h. Further, when the paracetamol content in the granules was included in the evaluation, the best continuous ring layer granulation process was performed using the shaft speed of 1250 rpm with the binder flow rate of 8.1 kg/h (Table 2). From a process control point of view, the main observation is that when compared to the fluidized bed and high shear granulation process, the continuous ring layer granulation process seems to be easiest to

Granules from continuous ring layer and high shear granulation processes exhibit somewhat similar surface morphologies as seen in SEM pictures (Fig. 7C–F). Compared to these granules, granules from fluidized bed granulation were more porous (Fig. 7A–B). Previously, similar tendency for the fluidized bed granulation technique to yield more porous granules in comparison to high shear granulation was reported [33]. Overall, the continuous ring layer granulation process yielded granules resembling more the granules produced by the high shear process than by the fluidized bed granulation process (Table 2, Fig. 7). Therefore, granules from high shear and continuous ring layer granulation processes were chosen for further tablet studies.

Fig. 5. Effect of shaft speed (presented in x-axis −1.0 = 1000 rpm, 0 = 1250 rpm, 1.0 = 1500 rpm) on mean granule size (d50) (y-axis) in continuous ring layer granulation (confidence level = 0.95). For the applied model, R2 was 0.85 and Q2 0.78.

Fig. 6. Effect of binder flow rate (presented in x-axis −1.0 = 7.4 kg/h, 0 = 8.1 kg/h, 1.0 = 8.8 kg/h) on mean granule size (d50) (y-axis) in continuous ring layer granulation (confidence level = 0.95). For the applied model, R2 was 0.85 and Q2 0.78.

control: the results demonstrate that continuous ring layer granulator provides almost linear responses of d50 to the manipulated variables over the widest size range. 3.2. Effect of granulation process on granule morphology

3.3. Tablet properties and drug release Table 2 indicates that both tablets, compressed either from the granules produced using the continuous ring layer granulator or from the high shear granulator, possessed acceptable fracture force, tensile

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

119

Fig. 7. Scanning electron microscopy pictures of granules produced using the best granulation process parameters. A–B: fluidized bed granulation (granulation temperature 65 °C, binder flow rate 22 g/min), C–D: high shear granulation (impeller speed 400 rpm and chopper speed 3000 rpm) and E–F: continuous ring layer granulation (shaft speed 1250 rpm and binder flow rate 8.1 kg/h). Scale bars are 100 μm (A, C), 200 μm (D, E, F) and 500 μm (B).

strength and uniformity of dosage units values. For comparison, the fracture forces values of commercial tablets have been reported to vary from 40 to 100 N [34]. The compression force required to form tightly bound tablets was higher for the granules produced by high shear granulation than that for the granules produced by continuous ring layer granulation (Table 2). It was early reported that high shear granulation produces denser granules that are more challenging to compress into tablets than granules produced by fluidized bed granulator [33]. Further, the porosity of microcrystalline cellulose pellets was demonstrated to influence the tablet tensile strength, as the tablets made from the low porosity granules (11%) were weaker than the tablets made from the high porosity granules (32%) [35]. In the high shear granulation, the densification phenomenon is based on the granulation mechanism: initially the liquid bridges are formed followed by cutting, compression and agglomerate formation that leads to the densification and thus to the formation of harder granules. In our study, tablets compressed either from granules produced using continuous ring layer granulation or from high shear granulation,

showed an immediate drug release, i.e. about 85% of a drug content was released within 30 min in both cases (Fig. 8). However, the drug release from the tablets compressed from the granules prepared by continuous ring layer granulation was statistically significantly faster at 5, 10 and 15 min. This could be explained by the higher polyvinylpyrrolidone concentration needed in the high shear granulation (5.3% (w/w)) than in the continuous ring layer granulation (3.2–3.8% (w/w)). Both the disintegration time of paracetamol tablets [36,37] and the time to reach 50% of the paracetamol content dissolved [37] have been reported to increase with an increase in polyvinylpyrrolidone concentration (1–10% (w/w)). 4. Conclusions Paracetamol granules were produced by three wet granulation processes i.e. a batch fluidized bed granulation and a high shear granulation and a continuous ring layer granulation process. Granules produced by the continuous ring layer granulation had as good pharmaceutical quality in terms of mean granule size, flowability and drug content as

120

M.A. Järvinen et al. / Powder Technology 275 (2015) 113–120

Fig. 8. Paracetamol release in USP phosphate buffer (pH 7.4, 37 °C) from tablets compressed from granules produced either by continuous ring layer granulation (●) or by high shear granulation (□). Mean values ± SD are shown (n = 6). * Statistically significant difference (p b 0.05).

the granules prepared with the batch processes. In addition, both types of tablets compressed either from granules from continuous ring layer granulation or from high shear granulation achieved an immediate drug release. Importantly, from a process control point of view, the continuous ring layer granulation process seems to be easiest to control. The continuous ring layer granulator provided nearly linear responses of mean granule size (d50) with the manipulated variables (shaft speed and binder flow rate) over the widest size range. Overall, the continuous ring layer granulation process was demonstrated to be an effective technique for producing pharmaceutical granules. Acknowledgement We would like to thank Simo-Pekka Simonaho, PhD (School of Pharmacy, University of Eastern Finland) and Jari Leskinen, PhD (SIB Labs, University of Eastern Finland) for the collaboration during the work. We thank the Finnish Funding Agency for Innovation (TEKES) and all collaborating companies in COCO project as well as North Savo Regional Fund (MJ) for supporting the work financially. In addition, we thank the PROMIS Centre consortium, funded by the TEKES ERDF and North Savo Centre for Economic Development, Transport and the Environment for providing excellent research facilities. References [1] S.M. Iveson, J.D. Litster, K. Hapgood, B.J. Ennis, Nucleation, growth and breakage phenomena in agitated wet granulation processes: a review, Powder Technol. 117 (2001) 3–39. [2] H. Leuenberger, New trends in the production of pharmaceutical granules: the classical batch concept and the problem of scale-up, Eur. J. Pharm. Biopharm. 52 (2001) 279–288. [3] R.P.J. Sochon, S. Zomer, J.J. Cartwright, M.J. Hounslow, A.D. Salman, The variability of pharmaceutical granulation, Chem. Eng. J. 164 (2010) 285–291. [4] A. Faure, P. York, R.C. Rowe, Process control and scale-up of pharmaceutical wet granulation processes: a review, Eur. J. Pharm. Biopharm. 52 (2001) 269–277. [5] P. Knight, Challenges in granulation technology, Powder Technol. 140 (2004) 156–162. [6] H. Leuenberger, Scale-up in the 4th dimension in the field of granulation and drying or how to avoid classical scale-up, Powder Technol. 130 (2003) 225–230. [7] S.D. Schaber, D.I. Gerogiorgis, R. Ramachandran, J.M.B. Evans, P.I. Barton, B.L. Trout, Economic analysis of integrated continuous and batch pharmaceutical manufacturing: a case study, Ind. Eng. Chem. Res. 50 (2011) 10083–10092. [8] P.M. Portillo, M.G. Ierapetritou, F.J. Muzzio, Characterization of continuous powder mixing processes, Powder Technol. 182 (2008) 368–378.

[9] R.K. Thakur, Ch. Vial, K.D.P. Nigam, E.B. Nauman, G. Djelveh, Static mixers in the process industries — a review, Trans. IChemE. 81 (2003) 787–826. [10] C. Vervaet, J.P. Remon, Continuous granulation in the pharmaceutical industry, Chem. Eng. Sci. 60 (2005) 3949–3957. [11] H. Berthiaux, K. Marikh, C. Gatumel, Continuous mixing of powder mixtures with pharmaceutical process constrains, Chem. Eng. Process. 47 (2008) 2315–2322. [12] M.A. Järvinen, J. Paaso, M. Paavola, K. Leiviskä, M. Juuti, F. Muzzio, K. Järvinen, Continuous direct tablet compression: effects of impeller rotation rate, total feed rate and drug content on the tablet properties and drug release, Drug Dev. Ind. Pharm. 39 (11) (2013) 1802–1808. [13] K. Järvinen, W. Hoehe, M. Järvinen, S. Poutiainen, M. Juuti, S. Borchert, In-line monitoring of the drug content of powder mixtures and tablets by near-infrared spectroscopy during the continuous direct compression tableting process, Eur. J. Pharm. Sci. 48 (4–5) (2013) 680–688. [14] A.U. Vanarase, F.J. Muzzio, Effect of operation conditions and design parameters in a continuous powder mixer, Powder Technol. 208 (2011) 26–36. [15] K. Järvinen, M. Toiviainen, M. Järvinen, M. Juuti, Continuous wet granulation process: granule properties and in-line process monitoring, Am. Pharm. Rev. 15 (6) (2012) 62–69 (URL: http://www.americanpharmaceuticalreview.com/1505Archives/122434-September-October-2012/). [16] M.J. Gamlen, C. Eardley, Continuous extrusion using a Raker Perkins MP50 (multipurpose) extruder, Drug Dev. Ind. Pharm. 12 (11–13) (1986) 1701–1713. [17] N.O. Lindberg, M. Myrenas, C. Tufvesson, L. Olbjer, Extrusion of an effervescent granulation with a twin screw extruder, Baker Perkins MPF 50D. Determination of mean residence time, Drug Dev. Ind. Pharm. 14 (5) (1988) 649–655. [18] D. Djuric, P. Kleinebudde, Impact of screw elements on continuous granulation with a twin-screw extruder, J. Pharm. Sci. 97 (11) (2008) 4934–4942. [19] M.R. Thompson, J. Sun, Wet granulation in a twin-screw extruder: implications of screw design, J. Pharm. Sci. 99 (4) (2010) 2090–2103. [20] M. Fonteyne, S. Soares, J. Vercruysse, E. Peeters, A. Burggraeve, C. Vervaet, J.P. Remon, N. Sandler, T. De Beer, Prediction of quality attributes of continuously produced granules using complementary pat tools, Eur. J. Pharm. Biopharm. 82 (2012) 429–436. [21] E.I. Keleb, A. Vermeire, C. Vervaet, J.P. Remon, Extrusion granulation and high shear granulation of different grades of lactose and highly dosed drugs: a comparative study, Drug Dev. Ind. Pharm. 30 (6) (2004) 679–691. [22] Gebrüder Lödige Maschinenbau GmbH, High-performance Ringlayer Mixer CoriMix® CMURL http://www.loedige.de/uploads/_downloads/BR-PRO-GB-CORIMIXGLP200605.pdf 2006. [23] EDQM, European Pharmacopoea 7.0th edition (and supplement 7.4), 1Council of Europe, Strasbourg Cedex, France, 2010. [24] J.T. Fell, J.M. Newton, Determination of tablet strength by diametral compression test, J. Pharm. Sci. 59 (1970) 688–691. [25] S.A. Cryer, P.N. Scherer, Observations and process parameter sensitivities in fluid-bed granulation, AICHE J. 49 (11) (2003) 2802–2809. [26] B. Rambali, L. Baert, D. Thoné, D.L. Massart, Using experimental design to optimize the process parameters in fluidized bed granulation, Drug Dev. Ind. Pharm. 27 (1) (2001) 47–55. [27] T. Närvänen, T. Lipsanen, O. Antikainen, H. Räikkönen, J. Yliruusi, Controlling granule size by granulation liquid feed pulsing, Int. J. Pharm. 357 (2008) 132–138. [28] T.K. Bock, U. Kraas, Experience with the Diosna mini-granulator and assessment of process scalability, Eur. J. Pharm. Biopharm. 52 (2001) 297–303. [29] M. Cavinato, E. Andreato, M. Bresciani, I. Pignatone, G. Bellazzi, E. Franceschinis, N. Realdon, P. Canu, A.C. Santomaso, Combining formulations and process aspects for optimizing the high-shear wet granulation of common drugs, Int. J. Pharm. 416 (2011) 229–241. [30] Sáska Zs, J. Dredán, O. Luhn, E. Balogh, G. Shafir, I. Antal, Evaluation of the impact of mixing speed on the compressibility and compactibility of paracetamol-isomalt containing granules with factorial design, Powder Technol. 213 (2011) 132–140. [31] T.M. Chitu, D. Oulahna, M. Hemati, Wet granulation in laboratory-scale high shear mixers: effect of chopper presence, design and impeller speed, Powder Technol. 206 (2011) 34–43. [32] J.S. Ramaker, M. Albada Jelgersma, P. Vonk, N.W.F. Kossen, Scale-down of a highshear pelletisation process: flow profile and growth kinetics, Int. J. Pharm. 166 (1998) 89–97. [33] J.Z.H. Gao, A. Jain, R. Motheram, D.B. Gray, M.A. Hussain, Fluid bed granulation of a poorly water soluble, low density, micronized drug: comparison with high shear granulation, Int. J. Pharm. 237 (2002) 1–14. [34] J.M. Sonnergaard, Distribution of crushing strength of tablets, Eur. J. Pharm. Biopharm. 53 (2002) 353–359. [35] J. Nordström, A.-S. Persson, L. Lazorova, G. Frenning, G. Alderborn, The degree of compression of spherical granular solids controls the evolution of microstructure and bond probability during compaction, Int. J. Pharm. 442 (2013) 3–12. [36] D. Becker, T. Rigassi, A. Bauer-Brandl, Effectiveness of binders in wet granulation: a comparison using model formulations of different tabletability, Drug Dev. Ind. Pharm. 23 (8) (1997) 791–808. [37] E.I. Okoye, A.O. Onyekweli, O.O. Kunle, Okra gum — an economic choice for the amelioration of capping and lamination in tablets, Ann. Biol. Res. 2 (2) (2011) 30–42.