Impact of roll compactor scale on ribbon density

PTEC-12383; No of Pages 12 Powder Technology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

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

Impact of roll compactor scale on ribbon density Ana Pérez Gago a, Gavin Reynolds b, Peter Kleinebudde a,⁎ a b

Institute of Pharmaceutics and Biopharmaceutics, Heinrich-Heine-University, Düsseldorf, Germany Pharmaceutical Technology and Development, AstraZeneca, Silk Road Business Park Charter Way, Macclesfield, Cheshire SK10 2NA, UK

a r t i c l e

i n f o

Article history: Received 3 November 2016 Received in revised form 14 February 2017 Accepted 16 February 2017 Available online xxxx Keywords: Roll compaction Scale-up Modelling Ribbon relative density Gerteis L.B. Bohle

a b s t r a c t Limited work has been performed regarding the scalability in the roll compaction process. Most of those studies available focus their efforts on developing models to successfully scale-up the process and only few of them strive to analyse the effect of the roll compaction scale on the product's properties. Therefore, in this work a double evaluation is performed focusing on process understanding and modelling application. In order to achieve this aim, ribbons of MCC, mannitol and a binary 1:1 mixture were roll compacted on 2 scales of compactors developed by Gerteis and L.B. Bohle, respectively. All compactors have a roll diameter of 250 mm in common but they differ in the roll width. The production was carried out following a common design of experiments in which the effect of the specific compaction force, the gap width and the roll speed were also investigated. The ribbons obtained were collected and characterized regarding their relative density. After statistical evaluation, it was found that the relative density of the mannitol and the mixture's ribbons produced using the Gerteis and L.B. Bohle compactors, are significantly affected by the scale, i.e. the roll width. For MCC, the impact of the compactor scales on the process was not so critical. The data collected was also modelled using the approach developed by Reynolds et al. 2010 in order to successfully scale-up the process. Excellent prediction was found for MCC, and although for mannitol and the mixture, the quality of the models decreased, they are still in good agreement, indicating the great utility of this approach when scaling-up a roll compaction process. © 2017 Elsevier B.V. All rights reserved.

1. Introduction The production of tablets in the pharmaceutical industry often requires a granulation process in order to achieve adequate flowability and compactability of the material by size enlargement. Roll compaction/dry granulation is a continuous process which popularity has increased in recent years [1]. This process comprises compaction of feed powder by passing through the gap formed between two counter-rotating rolls. The denser product ‘ribbon’ is subsequently milled into granules that can be later compacted into tablets. Many parameters and process conditions can be modified in order to obtain the desirable product characteristics. The properties of the compacted material also have a significant impact on the process. Two types of compaction behavior are relevant: plastic deformation and fragmentation (brittle character). A plastic material like microcrystalline cellulose (MCC) is able to remain deformed once the compaction stress is removed. A brittle material such as mannitol, will suffer fragmentation of its particles by the stress applied [2]. These two materials were chosen for this study as they are both widely used as diluents in the pharmaceutical industry [3,4].

⁎ Corresponding author. E-mail address: [email protected] (P. Kleinebudde).

As for any process of high interest for the pharmaceutical industry, it is necessary to investigate its scale-up, i.e. the transfer of the process from smaller to larger scale. For achieving this objective, the critical parameters involved on this procedure must be identified and their influence understood, so that it is possible to adapt them in order to obtain the same product quality at both scales. This is important to consider in research and a critical step in the industry, as it is desirable that the results obtained in the laboratory can also be transferred to pilot, production or commercial scale. Roll compaction process vendors use two scale-up strategies based on how the size of the rolls are modified between scales: by changing the roll diameter together with the roll width or by just varying the roll width while keeping the diameter constant. Despite the importance of the roll compaction process, limited scale-up work has been reported in the literature. Several authors have performed multiple studies in the roll compaction field which resulted in different tools considered for the investigators as useful for scale-up [5–7]. Sheskey et al. [8] performed a scale-up study using three compactors from Freund-Vector (TF-Mini, TF-156 and TF-3012) to transfer a drug-containing formulation where production conditions were optimized for the TF-Mini. Freund-Vector Corporation applies a change in roll diameter and width scale-up strategy for its equipment. It was observed that the roll force (expressed as tons/in.) and speed can be easily scaled by adapting those parameters in order to obtain

http://dx.doi.org/10.1016/j.powtec.2017.02.045 0032-5910/© 2017 Elsevier B.V. All rights reserved.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

2

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

the same total roll force (depending on the roll width) and the same linear speed (depending on the roll diameter) in all compactors. The screw and its relationship with the roll speed caused minimum problems. Recently, Allesø et al. [9] investigated the scale-up between two compactors from Gerteis (Mini-Pactor® and Macro-Pactor®) which have a common roll diameter, thus, only the width of the rolls is different between both scales. A design of experiments (DOE) was performed in the two scales using mainly MCC. Ribbon porosity was determined using two different methods (laser-based technique and an oil intrusion method) and the values obtained for the same process conditions in the two scales were compared and an excellent correlation was found. They concluded that ribbon porosity was scale-independent when the roll width is used as the scale factor and the specific compaction force (SCF) is constant, i.e. when only the roll width differs between both pieces of equipment. Many of the scale-up studies more recently published include the development of models which allow prediction of the density of the ribbons and the parameter settings required for achieving a target ribbon density value. These models are interesting tools as they can be used for the diverse scales. The models for scale-up can be primarily classified as mechanistic models or dimensionless variables. Nevertheless, there are many other papers in which statistical and multivariate models are described [6,10–21]. However, these approaches were not described to be applicable for the scalability of the process and for this reason, they will not be considered. The mechanistic models are based on physical principles and considerations of the roll compaction process and most of them rely on the rolling theory developed by Johanson [22]. On the other hand, the dimensional analysis to develop a scaling variable and therefore allows transfer of the process. Johanson [22] developed a model which considers the geometry of the roll compactor and it could be useful for scale-up predictions. His theory is based on predicting the density of the ribbon by calculating the differences on volume between the nip area and the gap. The nip area is the zone where the densification starts and it is defined by the nip angle [23]. In order to calculate the nip angle, Johanson equates the normal stress generated in both regions. Once this value has been determined, the volume of the nip region is calculated by determination of the roll force. Then, the density of the ribbon is obtained by calculating the differences in volume between the nip region and the gap. Nevertheless, in the Johanson's model, the pressure on the nip angle (nip pressure) is used to calculate the roll force. However, this value cannot be accurately estimated and several authors have developed different approaches to overcome this limitation. Reynolds et al. [24] proposed a practical approach in which the need for the nip pressure is avoided by using a preconsolidation relative density that can be used directly to relate the ribbon density to the modelled peak pressure, Pmax, between the rollers. Another novel point is an alternative form of the model that includes the feed screw speed, showing how this has an effect on the incoming material. The model was validated using experimental data obtained using Alexanderwerk WP 120. For this compactor two roll widths are available: 25 and 40 mm but both of 120 mm diameters. The relative density of the ribbons was obtained, proving that the model provides excellent correlation between the predicted and experimental values. This approach is not only applicable for scale-up but also for other types of roll compactor. Souihi et al. [25] used the Reynolds's model to examine a formulation using two roll compactors with different feeding systems: Alexanderwerk WP 120 and Pharmapaktor C250 (from Hosokawa Bepex company). Results showed an excellent prediction of the ribbon porosity with overall root-meansquare error (RMSE) between 1.0 and 1.5%, and therefore, they confirmed the applicability of the model not only for scale-up but also for different systems. However, only two formulations were tested in total, although they involved plastic and brittle excipients and commonly used active pharmaceutical ingredients (API). Another mechanistic model for scale-up was developed by Nesarikar et al. [26] in which the nip pressure requirement is eliminated. This

approach calculates the relative density as a function of the gap and the roll force per unit of roll width (RFU) which are two variables independent on the roll compactor. An instrumented Alexanderwerk WP 120 (120 mm roll diameter and 40 mm roll width), which allows measuring the normal stress on the ribbon and hence the nip angle, was used. In this manner, the nip angle is calculated as a function of gap and RFU which is subsequently used for calculating the roll force, and therefore, the ribbon density. Placebo was used to calibrate the model and three active blends to validate this approach, obtaining reasonably accuracy for the ribbon density. Then, an uninstrumented Alexanderwerk WP 200 (200 mm and 75 mm of roll diameter and width respectively) was used for the scale-up study. As no data regarding normal stress or nip angle was collected for the WP 200, the equation obtained for the WP 120 was then used. Considering this assumption, the ribbon densities for a placebo mixture were calculated for the WP 200 and these values compared well with those obtained experimentally using the WP 200. The limitations of this approach are the requirement of extensive calibration data using materials with different compaction behaviors, and the need of an instrumented roll in order to obtain the nip angle. Other authors applied dimensionless analysis for scale-up. Rowe et al. [27] developed the so-called modified Bingham number (Bm*), and proposed a model based on the Johanson's theory and including some modifications from Reynolds's approach, specifically the consideration of the feeding zone. The same experimental data collected in Nesarikar's work [26] was used. The Bm* is plotted against the relative density and a linear correlation was obtained for the two scales considered. As in the case of the previous model, it covers a broad range of materials and process combinations, however, in this case, the authors stressed the need of parameters which are easy to measure or obtain. More recently, Boersen et al. [28] proposed another dimensionless relationship, called as dimensionless variable (DV) which considered the roll pressure, roll speed, feed screw speed, the true density of the material and the roll diameter. This variable was correlated to the ribbon density obtaining a linear relationship. This DV was used to evaluate the roll compaction process using an Alexanderwerk WP 120 and a Fitzpatrick IR220. Only a small percentage of error was found when comparing the ribbon relative density in both compactors for those conditions resulting in similar values of DV. Although this model was applied to different sieve fractions of MCC and their mixture with APIs and requires further validation, it looks promising, in particular because of the limited effort required to calculate the DV. The authors concluded that it could be a good approach for scale-up. Nevertheless, other techniques or approaches can be also used in order to predict the ribbon relative density at larger scales, and therefore, scale the process up. For instance, Shi et al. [29] developed a practical approach to scale-up the roll compaction process from the Alexanderwerk WP 120 to the Alexanderwerk WP 200 roller compactor. This methodology consists of the development of a linear equation which describes a property and is later adjusted for the large scale. The authors adapted the equation considering that the slopes of the previous equations are powder-dependent. The success of this methodology was later confirmed and it has as great advantage of being simple to implement. Another tool was proposed by Liu et al. [30], in which a new statistical methodology is used to scale-up a roll compaction process between the Fitzpatrick IR220 and the Fitzpatrick IR520: joint-Y partial least squares. The results confirmed that this methodology could be used for scaling-up. However, this model requires a large amount of data to perform its calibration. 2. Objectives In this work, the scale-up with two roll compactors from Gerteis and two produced by L.B. Bohle is investigated by producing ribbons of MCC, mannitol and a binary mixture of both, following a common DOE. These compactors are seldom reported in the literature, and for the specific

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

case of the L.B. Bohle compactors and according to the authors' knowledge no scale-up study has been published. MCC and mannitol were selected in this study as they show different compaction behaviors and therefore it is possible to see how the change in the scale and the roll compaction conditions affects a typical plastic and brittle material. Ribbon relative density was chosen as it is a key property which correlates to other properties of the ribbons, but also granules and tablets [24,31]. Results using both systems under the same process conditions are compared in order to study the effect of the scale. These datasets are later used to calibrate the Reynolds' model and evaluate the accuracy of this modelling approach. 3. Materials and methods 3.1. Materials MCC (Avicel® PH 101, FMC Bio Polymer, USA) and mannitol (Pearlitol® 200 SD, Roquette, France) were used to perform the experiments. No lubricant was used in any of the studies as it can drastically affect the process. Due to the hygroscopicity of MCC, the materials were stored in a climate room at 21 °C and 45% of relative humidity (RH). 3.2. Methods 3.2.1. Powder characterization: Flow properties The frictional properties of the starting materials were measured using a ring shear tester (RST-XS, Dietmar Schulze, Germany). A standard shear testing procedure [32] was used to measure the effective angle of internal friction (δE) and angle of wall friction (ϕW). The latter was measured against a stainless steel coupon with Ra value of 0.4. The measurements were made using preconsolidation loads of 4 kPa and pre-shear stress of 1 kPa, 1.4 kPa, 2 kPa and 2.6 kPa. 3.2.2. Design of experiments A full factorial design consisting of 3 factors at 2 levels together with 3 repetitions of the center point was carried out in order to investigate how the different combinations affect the resulting products in both scales. The factors considered were roll gap, roll speed and SCF in the levels described in Table 1. The levels that appear on the table were chosen according to the limitations of the process as some combinations of gap width and roll speed were not feasible. Furthermore, those values are within the typical levels used for pharmaceutical purposes. Setting higher SCFs can lead to overcompaction of MCC while lower values make it difficult to obtain mannitol ribbons. This DOE was performed for MCC, mannitol and 1:1 mixture (50% MCC and 50% mannitol) in the four compactors used in this study and the statistical evaluation of the DOEs was carried out in Modde 9.0 (Umetrics, Sweden). The runs were sorted randomly in order to remove any systematic errors. 3.2.3. Roll compaction – Gerteis: Mini-Pactor® vs Polygran® The experiments using the Gerties compactors were performed on a Mini-Pactor® 250/25 and a 3-W-Polygran® 250/50/3 (Gerteis Machinen + Processengineering AG, Switzerland). The Mini-Pactor® has a roll diameter of 250 mm and a roll width of 25 mm (small scale), while the Polygran® has an identical diameter but a width of

Table 1 Description of the DOE performed in this study. Factors

Gap (mm) Roll speed (rpm) Specific compaction force (kN/cm)

Levels −1

0

+1

1.5 2 4

2.25 3 6

3 4 8

3

50 mm (large scale). The position of the rolls in both cases is inclined. In order to ensure that only the scale has an effect on the resulting ribbons, all process conditions were kept as constant as possible. In this respect a knurled roll surface and side-sealing system (cheek plates) were used in both systems. The transport of a powder from the hopper to the gap is performed by a feeding auger (FA) and a tamping auger (TA) which, as well as the rolls, is inclined. Gap control was activated, so the speed of both screws was automatically adjusted. The ratio FA:TA was preset to 1:2 for the Mini-Pactor® and 1:3.5 to the Polygran®, which ensured that the correct gap value was achieved. The Gerteis compactors have a floating roll design, whereby the roll force will remain constant, but in order to keep the gap width constant, the feedback control modifies the powder feed rate. The roll force is controlled by the hydraulic pressure. The mixture compacted on the Polygran® was prepared using a Rhönrad mixer (RRM 100, J. Engelsmann AG, Germany) equipped with a Sew-Eurodrive motor (RF40DT80K4BMG/TF, Germany). The blend utilized to perform the experiments on the Mini-Pactor® was prepared using a L.B. Bohle mixer (LM40, L.B.Bohle Maschinen + Verfahren GmbH, Germany). A minimum of 500 g of ribbons were collected at the moment the steady-state conditions were achieved. Samples were stored in the climate room for at least 24 h before any further characterization. 3.2.4. Roll compaction – L.B. Bohle: BRC 25 vs BRC 100 The experiments for the L.B. Bohle compactors were performed using a BRC 25 and BRC 100 (L.B. Bohle Maschinen + Verfahren GmbH, Germany), which both have a diameter of 250 mm. The BRC 25 has a roll width of 25 mm (small scale) while for the BRC 100 is 100 mm (large scale). The roll position is horizontal. The two systems were assembled with smooth surface rolls and rim rolls as sealing system. Powder feeding is also carried out using a FA and a TA, however the BRC 100, due to the size, requires a double TA. During production, gap control was used, meaning that the speed of the screws was automatically adjusted according to a preset ratio FA:TA of 1:2.5. The L.B. Bohle compactors have also a movable or floating roll and a gap control based on the feeding rate, however, in contrast to Gerteis compactors, the machines are equipped with a spindle motor which is used to apply the roll force. The binary mixture was prepared using a L.B. Bohle mixer (LM40 NIR, L.B. Bohle Maschinen + Verfahren GmbH, Germany). The temperature and RH were controlled during collection of 500 g ribbons, after achieving the steady-state conditions. Before characterizing the samples, they were stored at 21 °C and 45% RH for at least 24 h. 3.2.5. Ribbon relative density The ribbon relative density or solid fraction (expressed in percent) was calculated using Eq. (1). It was obtained based on the porosity measured using the envelope and T.A.P. density analyser (GeoPyc® 1360, Micromeritics Instrument Corp., USA). γR ¼ 100−ε

ð1Þ

where γR is the relative density and ε the porosity obtained from the GeoPyc®. This machine was assembled using a 25.4 mm chamber with a consolidation force of 51 N and a conversion factor of 0.5153 cm3/mm. The true density of the starting material was measured with a helium pycnometer (AccuPyc 1330, Micromeritics Instrument Corp., USA) resulting in values of 1.59, 1.53 and 1.47 g/cm3 for MCC, the binary mixture and mannitol respectively. A minimum of 5 ribbons were randomly sampled and manually cut into pieces. The aim of this sampling procedure was to have a representative value of the mean density, and therefore, pieces from all parts of the ribbon (edges and centre) were taken. A 3-cycle blank with only the medium (DryFlo™) and a 3-cycle measurement were performed for each of the 3 replicates from each batch analysed.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

4

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

4.1. Ribbon relative density

3.2.6. Modelling The model of Reynolds [24] was used to calculate the peak pressure (Pmax) imposed by the roll compactors and provide a mechanistic analysis of the influence of the different governing factors on ribbon density. This model is based on that proposed by Johanson [22] to solve the pressure gradients in the slip and nip regions between the rolls. The relative density of the ribbon leaving the rolls (γR), is related to the calculated Pmax at minimum gap by the following power law presented in Eq. (2): 1 γ R ¼ γ0 P max =K

Relative density of all ribbons of MCC, mannitol and the 1:1 mixture produced in the compactors from both suppliers (Gerteis and L.B. Bohle) were characterized and presented in Tables 2 and 3, respectively. In these tables, the minimal and maximum values were highlighted with light and dark grey correspondingly. According to Zinchuk et al. [7], the ribbon relative density for obtaining acceptable products (granules and tablets) regarding quality should be between 60 and 80%. For the whole scale-up study, the majority of the ribbons fulfill this requirement, only few exceptions occurred. In the case of the compactors from Gerteis, mainly the MCC ribbons produced in both compactors with 4 kN/cm of SCF did not reach this minimum value. For the study involving the compactors from L.B. Bohle, not only MCC ribbons but also some of the batches from the mixture produced at 4 kN/cm had a relative density below 60%. Despite the differences in the sealing system and roll surfaces, some comparisons can be also established between the two families of compactors. When comparing the MCC produced using the four compactors under the same conditions, in general lower relative densities are obtained for the Gerteis equipment than for the BRCs. However, for mannitol the density was normally lower in the case of the BRCs. An intermediate behavior can be observed for the binary mixture, which shows smaller values for the BRC 25 than for the Mini-Pactor ® and also lower densities for the Polygran ® than for the BRC 100.

ð2Þ

where γ0 is the preconsolidation relative density and K is the material compressibility, both are compaction parameters depending on the material. From Eq. (2), Pmax can be described as presented in Eq. (3): P max ¼ θ¼α ðδ ;ϕ ;K Þ WD∫ θ¼0 E W

2R f



S= D

ð3Þ

K

ð1þS =D − cosθÞ cosθ

cos θ dθ

where Rf is the roll force, W is the roller width, D is the roll diameter, θ is the angular roll position, α is the nip angle and S is the gap or roll separation. 4. Results and discussion

4.1.1. Effect of scale and material The three repetitions of the centre point were used as reference conditions for general data discussion. Furthermore, these batches also give information about reproducibility. Average values of the ribbon relative density for the centre point conditions (6 kN/cm SCF, 2.25 mm gap and 3 rpm roll speed) for MCC, mixture and mannitol were depicted in Fig. 1a for the Gerteis compactors and in Fig. 1b for those from L.B. Bohle. The average and standard deviation presented in these graphs were calculated considering the mean value of the 3 centre points. If the influence of the material is taken into consideration, the lowest value is always obtained for the MCC ribbons while the relative density increases for the mixture and reaches the maximum value for the samples of mannitol. This means that the plastic MCC has lower relative densities while

Prior to presenting the results, it is noted that under some conditions considerable variability in the ribbon relative density measurements were found. The source of this variability could be due to inherent variability in the ribbon properties being produced under those conditions or unrepresentative sampling of the ribbon, which will naturally exhibit a lateral density distribution. As described in the methodology, considerable effort was taken to obtain a representative sample and the low variability in some of the measurements would suggest it is suitable, implying that the differences in some of the measurements are likely to be attributed to variability in the ribbon properties as the equipment seeks to maintain a steady state. Furthermore, this method has been widely used for the characterization of ribbon relative density in the literature [12,25,26,33–40].

Table 2 Ribbon relative density values together with the production conditions for the Gerteis compactors (minimum and maximum values were marked with light and dark grey respectively). Conditions SCF (kN/cm)

Average relative density (%)

Gap width Roll speed (mm) (rpm)

MCC MiniPactor

MCC Polygran

Mixt 50% MiniPactor

Mixt 50% Polygran

Mannitol MiniPactor

Mannitol Polygran

4

1.5

2

58.3

55.7

63.6

62.5

72.3

73.6

4

3

2

54.3

54.9

61.4

59.7

69.6

71.4

4

1.5

4

56.6

55.4

63.0

62.2

71.8

73.2

4

3

4

52.4

53.0

61.8

60.0

71.0

71.7

8

1.5

2

66.8

65.8

71.1

70.1

77.9

80.1

8

3

2

63.2

65.6

72.5

69.8

76.7

78.4

8

1.5

4

68.1

65.3

68.6

70.6

75.7

80.5

8

3

4

64.4

63.6

71.9

68.7

77.0

77.7

6

2.25

3

62.1

62.1

67.6

66.7

75.6

76.7

6

2.25

3

58.2

61.9

67.2

67.1

74.4

76.3

6

2.25

3

61.4

61.6

64.6

65.2

75.5

76.0

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

5

Table 3 Ribbon relative density values together with the production conditions for the L.B. Bohle compactors (minimum and maximum values were marked with light and dark grey respectively). Conditions SCF (kN/cm)

Average relative density (%)

Gap width Roll speed (mm) (rpm)

MCC BRC 25

MCC BRC 100

Mixt 50% BRC 25

Mixt 50% BRC 100

Mannitol BRC 25

Mannitol BRC 100

4

1.5

2

59.1

57.9

65.5

63.5

67.8

71.7

4

3

2

55.6

54.5

59.6

59.8

62.4

67.4

4

1.5

4

58.9

58.4

59.5

62.6

67.2

65.5

4

3

4

55.3

53.4

57.1

59.6

67.1

65.1

8

1.5

2

67.8

70.7

70.9

72.5

73.3

78.0

8

3

2

66.4

65.2

67.7

70.6

72.8

75.1

8

1.5

4

68.0

70.3

67.5

73.4

74.1

74.3

8

3

4

66.9

65.7

68.8

69.5

72.6

75.4

6

2.25

3

63.8

62.8

66.9

67.6

67.5

69.5

6

2.25

3

63.0

64.9

65.3

69.1

71.9

72.4

6

2.25

3

62.9

62.3

64.0

68.2

69.9

67.1

mannitol as a brittle material leads to higher values. This is in good agreement with what has been described in the literature. Golchert et al. [37] observed that, for the same formulation in which only the type of excipient is changed, the ribbon relative density is lower when using a plastic material like MCC, than when including other brittle powders like lactose or dicalcium phosphate anhydrous. In fact, Chang et al. [34] found similar ribbon solid fractions when for a particular formulation containing brittle materials, mannitol or lactose are used. Similarly, Iyer et al. [41] investigated the ribbon relative density for Avicel ® DG (co-processed excipient consisting of a 75% of MCC and a 25% of anhydrous dibasic calcium phosphate), lactose and their 1:1 mixture compacted also at the Mini-Pactor®. When comparing all materials, for the same conditions the relative density tends also to increase from the minimum observed for Avicel® DG to a maximum for lactose. However, if the same material is compared in both scales, the relative density values for all the powders compacted in the Gerteis machines are statistically equal for a significance level of α = 0.05. For the BRC 25 and 100, a statistically significant difference was found only for the mixture, however, when compacting MCC and mannitol at these conditions, the values obtained are considered to be identical (α = 0.05). Standard deviation was found to be slightly higher for the samples produced with the compactors from L.B. Bohle.

4.1.2. Coefficient plots The DOE presented in Table 1 which was performed for all the materials and compactors was evaluated using Modde 9.0. The coefficient plot can be found in Fig. 2a for the Mini-Pactor® and Polygran® and in Fig. 2b for the compactors from L.B. Bohle. In these two figures, it is possible to evaluate separately every combination of material and compactor. In general, models with a relatively high significance were obtained. All those show R2 value over 0.7, although most of them exceed 0.9, which was the case for all powders compacted in the Gerteis machines. For the Gerteis compactors, in all the cases the SCF has a direct effect on the relative density (the higher the SCF, the denser the ribbons). For MCC produced using the Mini-Pactor®, and the mixture and mannitol compacted using the Polygran®, the gap has also a significant inverse influence (the lower the gap, the denser the ribbons). For the 50% MCC mixture using the Mini-Pactor®, also the interaction between SCF and gap has a proportional effect. On the other hand, for the L.B. Bohle compactors, the SCF and gap were the main factors affecting directly and reversely the ribbon relative density respectively. The SCF is significant for all the combinations of scales and materials, while gap was found only to be significant for the MCC compacted at both BRCs and the mixture prepared at the BRC 100. In all cases, the non-significant factors were deleted from the model. This general effect of increasing the ribbon density while rising the SCF and decreasing the gap is presented in Fig. 3 with a contour plot as an example.

Fig. 1. Relative density for MCC, mannitol and the mixture for the centre point conditions comparing the two scales of Gerteis (a) and L.B. Bohle (b) suppliers, mean ± sd (n = 3).

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

6

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 2. Coefficient plots for the ribbon relative density analysing every combination of material and scale separately for the of Gerteis (a) and L.B. Bohle (b) compactors.

The impact of the SCF and the gap in the ribbon density was also identified by Allesø et al. [9]. They observed also that the higher the specific compaction force and the lower the gap, the denser (or less porous) the ribbons. Some other authors observed similar influence of the SCF [42,43] and the gap width [31] in the ribbon relative density, although these studies included other formulations and factors in their DOEs.

It is also important to recognise that the interaction between SCF and gap on ribbon density is quantitatively captured by the modelling approach [24]. In order to evaluate if the change in scale has an effect on the results, a new DOE was prepared including the scale as a quantitative factor, expressed as the roll width (mm) in the model. In Fig. 4, the coefficient

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

7

The scale becomes a significant direct factor for the blend and its interaction with the roll speed is also a proportional significant factor for the mannitol. It is important to point out that for both suppliers, the compaction performed with MCC is not significantly affected by the scale or an interaction involving this factor. The excellent compactability of MCC may explain why the material is exempted from the impact of the scale, however, the mixture and mannitol are probably affected due to poorer compaction properties of mannitol.

Fig. 3. Example of a contour plot for the mixture compacted at the Polygran® in which the effect of SCF and the gap on the relative density is shown.

plots obtained after analysing this new design are presented. In all the cases good models were obtained, with minimum R2 of 0.8, although a value of 0.9 was exceeded in most cases. For the Gerteis compactors, the SCF is significant for all the mixtures, as well as the gap for the pure materials. However, depending on the powder compacted, the profile is different. The scale by itself, as well as its interaction between gap and scale is indirectly significant for the mixture, while the gap-SCF interaction has a direct influence. For mannitol, the scale has a significant direct effect. Regarding the compactors from LB. Bohle, the SCF is the most significant factor affecting the ribbon density in all cases. Gap is inversely significant for MCC and the mixture.

4.1.3. Scale-up correlation In a recent publication, Allesø et al. [9] plotted the porosity of the ribbons produced with the same material and under the same conditions in the Mini-Pactor® (small scale compactor) against the MacroPactor® (large scale) and an excellent correlation was found, indicating a successful direct scale-up. This method was applied to the relative density data collected in the present study and depicted in Fig. 5. The average ribbon density was plotted together with the standard deviation and the best fit lines as well as the correlation coefficient, R. The best correlations are obtained for MCC, and then they tend to become worse as the proportion of this material decreases. For mixture and mannitol, the higher R is achieved for the Gerteis' family of compactors, while for MCC, the best correlation is found for the BRCs, being as well the best of all those evaluated. These results indicate that the higher the proportion of MCC, the better compactability of the material and the easier the handling of the material. However, the quality of these correlations may be related to the characterization method and the sample preparation. Contrarily to the comfortable manipulation of MCC, mannitol, as a brittle material, generates ribbons which tend to easily break or even shatter back into powder. Increased handling of these samples result in greater loss of material. Furthermore, another fact that can be of importance is the sealing system used. When

Fig. 4. Coefficient plots for the new design including the scale as a factor for the of Gerteis (a) and L.B. Bohle (b) compactors.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

8

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

assembling the rim rolls, the ribbons tend to break into small pieces independently of the nature of the material compacted. These fragments could have an impact on the results, as the sampling will be favourable to the larger pieces, what could lead to the exclusion of other pieces. For this reason, the results seem to be more reliable when higher proportions of MCC are used and when cheek plates are assembled. Independent of the differences between materials, the correlations obtained, even for MCC, are poorer in comparison to the excellent relationship found by Allesø et al. [9], meaning that in the present study difficulties for the scale-up are found. However, a possible explanation for the differences between their study and the present work may be the number of points considered. In the current evaluation 9 points (an average was prepared for the 3 centre points) were used instead of the 5 considered by these investigators Allesø et al. [9].

4.2. Modelling The approach developed by Reynolds et al. [24] was used in order to try to model the scalability of the roll compaction process by investigating the capacity of this approach to predict the relative density of the ribbons in both scales, and therefore, successfully scale-up the process. As a prior requirement, this methodology needs the shear cell data of the feed powder, i.e. the δE and the ϕW, which values can be found in Table 4. This model allows calculation of a single parameter, Pmax, which describes the extent of compacting stress on the material as a result of differences in geometry and operating parameters, providing a basis for process understanding between different roller compactors. In the methodology, the material compaction parameters, K and γ0, are estimated using the ribbon density data. Once estimated, these parameters can be used in the model to predict ribbon densities for different scales of roll compactor. To explore this in more detail and to

Fig. 5. Correlation for relative density of ribbons produced at the same manufacturing conditions between the bigger scale in the Y-axis (Polygran® or BRC 100) and the smaller scale in the X-axis (Mini-Pactor® or BRC 25) for the Gerteis compactors (left) and L.B. Bohle (right) regarding the compaction of MCC (a), the 50% mixture (b) and mannitol (c). The equation of the best fit line as well as the correlation coefficient are presented.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx Table 4 Measured frictional angles for the starting materials. Material

δE (°)

ϕW (°)

MCC Mixture 50% Mannitol

43.6 41.5 38.0

11.6 14.4 14.8

understand any deviation from the model assumptions, the scale-up within the Gerteis and the L.B. Bohle compactors for MCC, mannitol and the 1:1 mixture were examined. To mimic a typical development workflow, the model was calibrated using the data from the small scale compactor (the Mini-Pactor® for the Gerteis case and the BRC 25 for L.B. Bohle compactors) and then applied to predict the ribbon densities on the large scale. Thus, the material compaction parameters (K and γ0) are estimated for each small scale equipment and the model predictions compared with the large scale equivalents. In Table 5, the K and γ0 values are presented, as well as the relative density RMSE and the bias expressing the percentage of error. The root-mean-square error or RMSE refers to the quality of the correlation between the predicted and the observed values. However, the bias represent the differences in percentage between the model average prediction and the experimental data, and depending on the sign, the estimation of the model is higher (positive percent) or lower (negative value) than the experimental data measured. In Fig. 6, the graphical comparisons between the experimental data obtained for both scales and the predictions given by the model are collected for the different materials and compactors. The first conclusion that can be extracted from the Table 5 is that the RMSEs are in general low, which means a good fit of the model. Nevertheless, these error values are mostly higher than those obtained in the previous studies in which this model was applied [24,25]. It is important to note that the RMSEs are primarily affected by the degree of scatter within the raw data rather than deviation from model. If the bias is analysed, for the case of MCC using both families of compactors, the values are very low. This means an excellent prediction of the larger scale compactor by the model as can be seen in Fig. 6a with overpredictions of only 0.1%. This illustrates again how easy is to work with this material. However, when increasing the proportion of mannitol, the quality of the prediction decreases. For the mixture (Fig. 6b), underprediction of up to 4.0% is obtained for the L.B. Bohle compactors, while for the Gerteis compactors an overestimation of 0.9% takes place instead. A similar result is observed for pure mannitol (Fig. 6c), where a higher percentage of error is again found for the L.B. Bohle, but in this case, with overprediction of 4.1%, while the Gerteis is underestimated by 1.7%. This latter value, although higher than for the binary mixture, is marginal compared to the variation in the raw dataset. Therefore, for the binary mixture and mannitol, the model predicts more accurately when compactors from Gerteis are used. However, it is important to point out that the scatter in the results can be also explained by the differences between the two scale-up studies. For the Gerteis compactors, the increase in roll width is twice that of the small scale (from 25 mm to 50 mm) while for the L.B. Bohle machines, the Table 5 Values of material compaction parameters together with model results regarding RMSE and the percentage of error for each material produced at the both scales of compactors from Gerteis and L.B. Bohle. Material

Model

K

γ0

Relative density RMSE (%)

Relative density bias (%)

MCC

Gerteis L.B. Bohle Gerteis L.B. Bohle Gerteis L.B. Bohle

4.19 4.59 6.75 5.68 10.16 6.72

0.215 0.240 0.338 0.294 0.462 0.359

1.7 1.4 1.8 4.4 1.6 5.2

0.1 0.1 0.9 −4.0 −1.7 4.1

Mixture Mannitol

9

width of the large scale is 4 times larger (from 25 mm to 100 mm). This could explain the greater differences between the experimental data and model prediction. Another reason for these deviations between the two suppliers can be due to differences in the sealing systems and roll surfaces used, which can have an impact in the ribbon relative density. These different configurations can affect the nip angle which defines where the compaction starts. As the nip angle increases, the roll force is distributed over a larger compaction zone and therefore the compacting pressure applied to the powder will decrease, leading to lower densification and, therefore, lower ribbon relative density. Therefore, changes in the nip angle caused by differences in the powder interaction with different roll configurations may be the cause of some of the deviations seen. The model uses powder frictional properties obtained using a shear cell at 4 kPa preconsolidation stress to determine the nip angle. Within the roll compactor, the stresses on the powder in the nip region will be significantly higher and furthermore, the different roll surface finishes will affect the frictional interaction between the powder and rolls. In practise, measuring the frictional response of the powders under the relevant conditions is difficult, but deviations in the bias of the model may point to disparity in the location of the nip angle. Nevertheless, it cannot be clearly stated if the model works better for the Gerteis compactors as a result of the differences between the machines regarding design, the roll surface and sealing system assembled or the different scales used. Apart from the differences between the two families of compactors, the diverse behavior of the mixture and mannitol in respect to the prediction results is unexpected. When the Gerteis equipment is used, the mixture is underpredicted and the pure mannitol overestimated, while the opposite is seen for the L.B. Bohle compactors. However, the model highlights the trends in the raw data. If the whole data set is analysed, i.e. all the ribbon densities obtained for the three materials, approximately half of the batches show a higher density for the small scale for Gerteis ribbons as well as for L.B. Bohle. This means, that the distribution of results is more or less homogenous, i.e. there is not a clear tendency of the small scale (neither the larger) to always present denser ribbons. However, if the densities obtained are classified according to the material, an opposing effect is observed. For MCC, denser ribbons are achieved when compacting with the Mini-Pactor® and the BRC 25 than the respective large scales. However, although the mixture compacted at the Mini-Pactor® also results in denser ribbons, the larger BRC 100 produces samples with a higher relative density, compared with the smaller BRC 25. For mannitol, the ribbons compacted in the Polygran® are denser than when using the small scale, but for the L.B. Bohle compactors, the BRC 25 is producing the higher density samples compared with the large scale. In other words, there is no consistent tendency between both compactor families on how the mixture and mannitol behaves when using the small or large scale, and this is also reflected on the model prediction. In order to further understand this problem, the same scales for the different materials were compared, i.e. the Mini-Pactor ® vs the BRC 25 and Polygran® vs BRC 100. This allows observing if there are similarities on the tendency of the relative density according to the roll configuration. In Table 6 and in order to facilitate the interpretation of results, it is collected the values obtained when deducting the relative density of the L.B. Bohle compactor from the same value collected at the Gerteis piece of equipment (Result = Gerteis – L.B. Bohle). This deduction is always performed for specific compaction conditions and the same material. In other words, the objective is to compare both small and large scales in order to see which family of compactors generates higher densities, and if the tendency is the same for all the materials. When calculating the differences between small scales and large scales for the three materials, the same trend is observed for MCC and for mannitol, what means that normally for both scales, one family of compactors presents higher density. This means that for MCC, in most of the cases the highest relative density is obtained using the L.B.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

10

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

Fig. 6. Comparison of the model with the experimental data by presenting the Pmax estimation vs the relative density of the ribbons for the Gerteis compactors (left) and L.B. Bohle (right) regarding the compaction of MCC (a), the 50% mixture (b) and mannitol (c).

Bohle compactors, while for mannitol, Gerteis tends to produce denser ribbons. However, this is not the case for the mixture, which shows a different tendency when comparing the small scales than when relating the large ones. This means that for the small scales, Gerteis systems produce denser ribbons, while for the large scale, L.B. Bohle systems lead to a higher density. Therefore, what leads the model to decrease in quality may be the compaction of the mixture. A possible reason of these differences could be a partial segregation of the blend or a non-appropriate mixing procedure before its compactions. Overall, it appears that as the material becomes more brittle, the effect of scale becomes more pronounced in a way that is not currently captured by the model. This is likely to be caused by complexity in the way the roll configuration (surface and sealing system) interacts with the powder, which could cause changes in the nip angle, leading to increase or decrease in the peak pressure and therefore density. These effects may also be exacerbated at increased scale (roll width) due to changes in the lateral pressure distribution, again influencing the size

and stress distribution in the compacting zone. The manipulation of the samples could be a source of variability for the mannitol containing ribbon relative density measurements. Nevertheless, it can be concluded that the model predicts, with a small material dependent bias, the ribbon relative density in the larger scale making it a useful tool to scale up the roll compaction process. 5. Conlusions Two individual scale-up studies using MCC, mannitol and a binary mixture of both were performed for two compactors from Gerteis and two compactors from L.B. Bohle. In all cases, the machines used in this work have the same roller diameter, but differ in roll width. A design of experiments was used to study the compaction process in the four compactors. The evaluation of the experimental data gave no statistically significant differences when the relative density of the centre point ribbons is compared for all the materials in both scales, except for the binary

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

11

Table 6 Differences in ribbon relative density when deducting from L.B. Bohle compactors the value obtained for Gerteis for the same conditions (production and material). Please note that in this case the small and the large scales are compared between each other. Values positive are marked, meaning that in that case, the density of the Gerteis compactor is higher.

Batch number

Differences ribbon density (%) Mini–Pactor–BRC 25

Differences ribbon density (%) Polygran–BRC 100

MCC

Mixt 50%

Mannitol

MCC

Mixt 50%

Mannitol

1

–0.79

–1.93

0.62

–2.17

–0.98

5.83

2

–1.32

1.87

2.10

0.37

–0.11

9.02

3

–2.32

3.55

6.29

–3.02

–0.38

5.99

4

–2.84

4.69

5.85

–0.40

0.47

4.58

5

–0.94

0.21

–0.19

–4.84

–2.46

6.80

6

–3.24

4.84

1.61

0.32

–0.73

5.62

7

0.05

1.04

1.35

–5.00

–2.73

6.36

8

–2.57

3.10

1.63

–2.13

–0.85

5.14

9

–1.70

0.78

6.16

–0.70

–0.88

9.23

10

–4.79

1.93

2.03

–3.00

–1.97

4.38

11

–1.46

0.54

8.39

–0.67

–3.02

6.13

mixture compacted in the BRC compactors. The impact of scale is an effect area of concern as the manufacturers invest many effort in order to enable the scalability between their compactors. Due to its importance, the influence of the roll width on the relative density of the ribbons was further investigated by statistical analysis of the DOE. From the coefficient plots the impact of scale (directly or belonging with an interaction) was elucidated for the binary mixture and mannitol for the compactors of both suppliers. This effect of scale was also observed when plotting the ribbon relative density of the small scale vs the large scale, by the absence of a strong linear correlation. Therefore, in order to scale-up, the approach developed by Reynolds was used. The model was fitted using the material data and calibrated with the ribbon relative density values of the smaller scale. Good fits between the observed and predicted data were found. For MCC, the model showed excellent prediction with less than 1% error in the estimations. Although the quality of the model decreased for the binary mixture and mannitol, the predictions were still in good agreement with a maximum margin of error of 4.1%. The model was found to predict more accurately between the Gerteis compactors, in which case the maximum bias fell to 1.7%. Declaration of interest This work was supported by the IPROCOM Marie Curie initial training network, funded through the People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007–2013/ under REA grant agreement No. 316555. Acknowledgement The authors would like to thank Bayer Pharma AG located in Berlin (Germany) and the team of Dr. Susanne Skrabs, especially Dr. Sarah Just, Dr. Dejan Djuric and Christian Nienerza for their different roles in this work. A special acknowledge also to L.B. Bohle in Ennirgeloh (Germany) for kindly offering the installations and machines to perform this study, and particularly to Dr. Hubertus Rehbaum, Andreas Altmeyer

and Philipp Harbaum for their different ways of helping to carry out this work. References [1] P. Kleinebudde, Roll compaction/dry granulation: pharmaceutical applications, Eur. J. Pharm. Biopharm. 58 (2) (Sep 2004) 317–326. [2] R.C. Rowe, R.J. Roberts, Mechanical Properties, in: G. Alderborn, C. Nystrom (Eds.), Pharmaceutical Powder Compaction Technology, New York, Marcel Dekker, Inc, 1996. [3] N.A. Armstrong, Mannitol, in: R.C. Rowe, P.J. Sheskey, M.E. Quinn (Eds.), Handbook of Pharmaceutical Excipients, sixth ed.Pharmaceutical Press and American Pharmacists Association, London 2009, pp. 424–428. [4] Guy A. Cellulose, Microcrystalline, in: R.C. Rowe, P.J. Sheskey, M.E. Quinn (Eds.), Handbook of Pharmaceutical Excipients, sixth ed.Pharmaceutical Press and American Pharmacists Association, London 2009, pp. 129–133. [5] A. Gupta, G.E. Peck, R.W. Miller, K.R. Morris, Nondestructive measurements of the compact strength and the particle-Size distribution after milling of roller compacted powders by near-infrared spectroscopy, J. Pharm. Sci. 93 (4) (2004) 1047–1053. [6] V. Esnault, D. Heitzmann, M. Michrafy, D. Oulahna, A. Michrafy, Numerical simulation of roll compaction of aerated powders, Chem. Eng. Sci. 104 (2013) 717–726. [7] A.V. Zinchuk, M.P. Mullarney, B.C. Hancock, Simulation of roller compaction using a laboratory scale compaction simulator, Int. J. Pharm. 269 (2) (2004) 403–415. [8] P. Sheskey, K. Pacholke, G. Sackett, L. Maher, J. Polli, Roll compaction granulation of a controlled-release matrix tablet formulation containing HPMC: effect of process scale-up on robustness of tablets, tablet stability, and predicted in vivo performance, Pharm. Technol. 24 (11) (2000) 30–52. [9] M. Allesø, R. Holm, P. Holm, Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attribute, Eur. J. Pharm. Sci. 87 (2016) 69–78. [10] M.D. Bi, F. Alvarez-Nunez, F. Alvarez, Evaluating and modifying Johanson's rolling model to improve its predictability, J. Pharm. Sci. 103 (7) (Jul 2014) 2062–2071. [11] J.C. Cunningham, D. Winstead, A. Zavaliangos, Understanding variation in roller compaction through finite element-based process modeling, Comput. Chem. Eng. 34 (7) (2010 Jul) 1058–1071. [12] J. Hilden, G. Earle, E. Lilly, Prediction of roller compacted ribbon solid fraction for quality by design development, Powder Technol. 213 (1–3) (11/10/2011) 1–13. [13] Y. Liu, C. Wassgren, Modifications to Johanson's roll compaction model for improved relative density predictions, Powder Technol. 297 (2016) 294–302 (9//). [14] R.F. Mansa, R.H. Bridson, R.W. Greenwood, H. Barker, J.P.K. Seville, Using intelligent software to predict the effects of formulation and processing parameters on roller compaction, Powder Technol. 181 (2) (Feb 2008) 217–225. [15] A. Michrafy, H. Diarra, J.A. Dodds, M. Michrafy, L. Penazzi, Analysis of strain stress state in roller compaction process, Powder Technol. 208 (2) (3/25/2011) 417–422. [16] A.R. Muliadi, J.D. Litster, C.R. Wassgren, Modeling the powder roll compaction process: comparison of 2-D finite element method and the rolling theory for granular solids (Johanson's model), Powder Technol. 221 (May 2012) 90–100. [17] P. Nkansah, S.J. Wu, S. Sobotka, K. Yamamoto, Z.J. Shao, A novel method for estimating solid fraction of roller-compacted ribbons, Drug Dev. Ind. Pharm. 34 (2) (Feb 2008) 142–148.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045

12

A. Pérez Gago et al. / Powder Technology xxx (2017) xxx–xxx

[18] S. Peter, R.F. Lammens, K.-J. Steffens, Roller compaction/dry granulation: use of the thin layer model for predicting densities and forces during roller compaction, Powder Technol. 199 (2) (4/23/2010) 165–175. [19] J.L.P. Soh, F. Wang, N. Boersen, R. Pinal, G.E. Peck, M.T. Carvajal, et al., Utility of multivariate analysis in modeling the effects of raw material properties and operating parameters on granule and ribbon properties prepared in roller compaction, Drug Dev. Ind. Pharm. 34 (10) (2008) 1022–1035. [20] B.M.J. Tan, L.W. Chan, P.W.S. Heng, Determination of the nip angle in roller compactors with serrated rolls, J. Pharm. Sci. 105 (6) (2016) 1967–1975 (6//). [21] Z.X. Zheng, W. Xia, Z.Y. Zhou, Experimental and numerical modeling for powder rolling, Rev. Adv. Mater. Sci. 33 (4) (Aug 2013) 330–336. [22] J.R. Johanson, A rolling theory for granular solids, J. Appl. Mech. 842-8 (1965). [23] P. Guigon, O. Simon, K. Saleh, G. Bindhumadhavan, M.J. Adams, J.P.K. Seville, Chapter 5 Roll pressing, in: A.D. Salman MJH, Seville JPK (Eds.), Handbook of Powder Technology, Elsevier Science B.V. 2007, pp. 255–288. [24] G. Reynolds, R. Ingale, R. Roberts, S. Kothari, B. Gururajan, Practical application of roller compaction process modeling, Comput. Chem. Eng. 34 (7) (Jul 2010) 1049–1057. [25] N. Souihi, G. Reynolds, P. Tajarobi, H. Wikström, G. Haeffler, M. Josefson, et al., Roll compaction process modeling: transfer between equipment and impact of process parameters, Int. J. Pharm. 484 (1–2) (2015) 192–206. [26] V.V. Nesarikar, C. Patel, W. Early, N. Vatsaraj, O. Sprockel, R. Jerzweski, Roller compaction process development and scale up using Johanson model calibrated with instrumented roll data, Int. J. Pharm. 436 (1–2) (2012) 486–507. [27] J.M. Rowe, J.R. Crison, T.J. Carragher, N. Vatsaraj, R.J. McCann, F. Nikfar, Mechanistic insights into the scale-up of the roller compaction process: a practical and dimensionless approach, J. Pharm. Sci. 102 (10) (Oct 2013) 3586–3595. [28] N. Boersen, D. Belair, G.E. Peck, R. Pinal, A dimensionless variable for the scale up and transfer of a roller compaction formulation, Drug Dev. Ind. Pharm. 42 (1) (2015) 60–69. [29] W. Shi, O.L. Sprockel, A practical approach for the scale-up of roller compaction process, Eur. J. Pharm. Biopharm. (2016). [30] Z. Liu, M.J. Bruwer, J.F. MacGregor, S.S.S. Rathore, D.E. Reed, M.J. Champagne, Scaleup of a pharmaceutical roller compaction process using a joint-Y partial least squares model, Ind. Eng. Chem. Res. 50 (18) (Sep 2011) 10696–10706. [31] A.M. Miguelez-Moran, C.Y. Wu, H.S. Dong, J.P.K. Seville, Characterisation of density distributions in roller-compacted ribbons using micro-indentation and X-ray micro-computed tomography, Eur. J. Pharm. Biopharm. 72 (1) (May 2009) 173–182.

[32] I.A.S.Z. Peschl, H. Colijn, New rotational shear testing technique, J. Powder Bulk Solids Technol. 1 (3) (1977) 55–60. [33] J. Austin, A. Gupta, R. McDonnell, G.V. Reklaitis, M.T. Harris, The use of near-infrared and microwave resonance sensing to monitor a continuous roller compaction process, J. Pharm. Sci. 102 (6) (Jun 2013) 1895–1904. [34] C.K. Chang, F.A. Alvarez-Nunez, J.V. Rinella, L.E. Magnusson, K. Sueda, Roller compaction, granulation and capsule product dissolution of drug formulations containing a lactose or mannitol filler, starch, and talc, AAPS PharmSciTech 9 (2) (Jun 2008) 597–604. [35] M. Dumarey, H. Wikstrom, M. Fransson, A. Sparen, P. Tajarobi, M. Josefson, et al., Combining experimental design and orthogonal projections to latent structures to study the influence of microcrystalline cellulose properties on roll compaction, Int. J. Pharm. 416 (1) (Sep 2011) 110–119. [36] J.F. Gamble, M. Tobyn, A.B. Dennis, T. Shah, Roller compaction: application of an ingap ribbon porosity calculation for the optimization of downstream granule flow and compactability characteristics, Pharm. Dev. Technol. 15 (3) (Jun 2010) 223–229. [37] D. Golchert, E. Bines, A. Carmody, Evaluation of some compression aids in tableting of roller compacted swellable core drug layer, Int. J. Pharm. 453 (2) (Sep 2013) 322–328. [38] Iyer RM, Hegde S, Singhal D, Malick W. A novel approach to determine solid fraction using a laser-based direct volume measurement device. Pharm. Dev. Technol. Aug 2014a;19(5):577-82. [39] H. Lim, V.S. Dave, L. Kidder, E. Neil Lewis, R. Fahmy, S.W. Hoag, Assessment of the critical factors affecting the porosity of roller compacted ribbons and the feasibility of using NIR chemical imaging to evaluate the porosity distribution, Int. J. Pharm. 410 (1–2) (2011) 1–8. [40] M.A.P. McAuliffe, G.E. Omahony, C.A. Blackshields, J.A. Collins, D.P. Egan, L. Kiernan, et al., The use of PAT and off-line methods for monitoring of roller compacted ribbon and granule properties with a view to continuous processing, Org. Process. Res. Dev. 19 (1) (2015) 158–166. [41] Iyer RM, Hegde S, Dinunzio J, Singhal D, Malick W. The impact of roller compaction and tablet compression on physicomechanical properties of pharmaceutical excipients. Pharm. Dev. Technol. Aug 2014b;19(5):583-92. [42] M. Khorasani, J.M. Amigo, C.C. Sun, P. Bertelsen, J. Rantanen, Near-infrared chemical imaging (NIR-CI) as a process monitoring solution for a production line of roll compaction and tableting, Eur. J. Pharm. Biopharm. 93 (2015) 293–302 (6//). [43] N. Souihi, M. Josefson, P. Tajarobi, B. Gururajan, J. Trygg, Design space estimation of the roller compaction process, Ind. Eng. Chem. Res. 52 (35) (2013) 12408–12419.

Please cite this article as: A. Pérez Gago, et al., Impact of roll compactor scale on ribbon density, Powder Technol. (2017), http://dx.doi.org/ 10.1016/j.powtec.2017.02.045