Accepted Manuscript On-line monitoring of fluid bed granulation by photometric imaging Ira Soppela, Osmo Antikainen, Niklas Sandler, Jouko Yliruusi PII: DOI: Reference:

S0939-6411(14)00250-1 http://dx.doi.org/10.1016/j.ejpb.2014.08.009 EJPB 11697

To appear in:

European Journal of Pharmaceutics and Biopharmaceutics

Received Date: Accepted Date:

13 May 2014 19 August 2014

Please cite this article as: I. Soppela, O. Antikainen, N. Sandler, J. Yliruusi, On-line monitoring of fluid bed granulation by photometric imaging, European Journal of Pharmaceutics and Biopharmaceutics (2014), doi: http:// dx.doi.org/10.1016/j.ejpb.2014.08.009

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

On-line monitoring of fluid bed granulation by photometric imaging Ira Soppela*a, Osmo Antikainena, Niklas Sandlerb, Jouko Yliruusia

a

Division of Pharmaceutical Chemistry and Technology, Faculty of Pharmacy, University of Helsinki, Helsinki, Finland

b

Pharmaceutical Sciences Laboratory, Department of Biosciences, Abo Akademi University, Turku, Finland *Corresponding author [email protected] Tel. +358 41 5506622 Division of Pharmaceutical Chemistry and Technology, Faculty of Pharmacy, P.O. Box 56 (Viikinkaari 5E), FI-00014 University of Helsinki, Finland

Keywords: fluid bed granulation, image analysis, photometric imaging, on-line monitoring, process analytical technology, powder processing

Abstract This paper introduces and discusses a photometric surface imaging approach for on-line monitoring of fluid bed granulation. Five granule batches consisting of paracetamol and varying amounts of lactose and microcrystalline cellulose were manufactured with an instrumented fluid bed granulator. Photometric images and NIR spectra were continuously captured on-line and particle size information was extracted from them. Also key process parametres were recorded. The images provided direct real-time information on the growth, attrition and packing behaviour of the batches. Moreover, decreasing image brightness in the drying phase was found to indicate granule drying. The changes observed in the image data were also linked to the moisture and temperature profiles of the processes. Combined with complementary process analytical tools, photometric imaging opens up possibilities for improved real-time evaluation fluid bed granulation. Furthermore, images can give valuable insight into the behaviour of excipients or formulations during product development.

1. Introduction Fluid bed granulation is a common unit operation in solid dosage form manufacturing. Its performance is influenced by several factors including process parameters, powder characteristics and binder properties. Granulation consists of three consecutive and partly overlapping phenomena, i.e. wetting and nucleation followed by consolidation and growth and finally breakage and attrition [1]. Controlling changes during granulation is important for achieving a smooth process and the required end-product quality. Fluid bed processes are conventionally monitored using process parameters, e.g., process air flow, volume and humidity [2]. Information on the typical behaviour of different materials can be obtained by combining temperature and humidity data obtained from the process [3]. However, to monitor the process effectively and reliably in real-time, direct measurements on key product properties such as moisture and particle size distribution can be beneficial. The endeavour of the industry to increase in-process controls and e.g. to shift from batch to continuous processing with further process automation requires reliable, continuous and automated solutions for process monitoring. In this context, on-line, in-line or at-line measurements are preferred over off-line measurements [4].

Monitoring moisture during granulation and pelletization is important in terms of product quality and numerous methods, such as NIR and acoustic emission have been developed and studied [5-9]. Furthermore, controlling particle size during granulation is critical due to the great importance of granule size in the subsequent tabletting process. A few systematic realtime methods for measuring particle size in fluid bed granulators have been developed, including an imaging probe, spatial filtering velocimetry (SFV), acoustic emission, NIR, Focused Beam Reflectance Measurement (FBRM) and Particle Image Velocimetry (PIV) [10-15]. However, the movement of the sample, probe fouling and insufficient fluidisation close to the sampling probe make the particle size measurements challenging. An attempt to overcome these problems and to gain direct visual information on the granulation process by measuring particle size of a motionless sample with an on-line or at-line surface imaging approach have also been studied [16-19].

In addition to particle size, enormous untapped potential lies in the images collected during processing [18]. The most common problems encountered in granulation include oversized granules, excessive amount of fines, poor fluidization and heterogeneity of the finished product [20]. Continuous imaging could help to direct the process towards optimised particle size and improved product uniformity as images provide simultaneous numerical and visual real-time information on material characteristics. The need to develop effective methods to monitor and control fluid bed processes in real-time can be met by well selected continuous in-line or on-line analysers that can be used in combination with process data. The combination of particle size measured by SFV and moisture information collected through NIR spectroscopy has earlier been shown to be advantageous in real-time monitoring and the control of a fluid bed granulation process [4].

This paper aims to investigate if image information together with simultaneously collected process data can visualise and provide insight into the phenomena taking place during fluid bed granulation. Specifically, the goal is to study the usefulness of the generated image information in real-time monitoring of granule formation and drying during fluid bed granulation.

2. Materials and methods

2.1. Formulation and granulation

Granules consisting of paracetamol (Mallinckrodt Inc, Raleigh, NC, USA), microcrystalline cellulose (Avicel PH101, FMC BioPolymer, Little Island, Ireland), lactose monohydrate (Pharmatose 200M, DMV Pharma, Veghel, The Netherlands) and polyvinylpyrrolidone (Plasdone K25, ISP Technologies Inc, Wayne, USA) were manufactured. The drug amount was kept constant at 5% (w/w) in all formulations and the ratios of the fillers are shown in table 1.

Five granule batches, referred to as I–V, were manufactured with an instrumented benchscale fluidized bed granulator (Glatt, WSG 5, Glatt Gmbh, Binzen, Germany). The batch size was three kilograms and 1500g of 15% aqueous povidone solution was used as a binder. The aim was to produce granules with varying particle size and moisture properties. The batches and respective formulations are shown in table I. The spraying rate was 77 g/min, atomization pressure was 0.15 MPa and the nozzle height 45 cm from the distributor plate. The inlet air temperature was 40°C during mixing and spraying and 60°C during drying. The inlet air flow rate was adjusted depending on the formulation to obtain optimal fluidisation.

The 1) inlet air flow rate, 2) inlet air humidity, 3) inlet air temperature, 4) outlet air humidity, 5) mass temperature, and 6) outlet air temperature were continuously recorded during processing of the batches. The water amounts of the inlet and outlet air were calculated from the measured relative humidity and air temperature. The total inlet water amount of each process phase was calculated for each batch by multiplying the inlet air water amount by the process time. Weight loss on drying of samples obtained at the end of the mixing, spraying and drying phase was measured by IR-drying (Sartorius Thermocontrol MA 100; Sartorius, Göttingen, Germany). The samples were measured in 105°C and the sample weight ranged from three to five grams.

2.2. Photometric imaging

Each granulation was recorded by a 3D surface imaging device prototype, consisting of a camera connected to an automated sampling double-cuvette attached to the granulator vessel (Flashsizer FS3D, Intelligent Pharmaceutics Ltd, Turku, Finland) (Figure1). The dimensions of the cuvette are 5*4*1.3 cm and the size of the measurement field is 1.5×1.1 cm. The sampling interval was five seconds and 300 to 450 images were taken per batch depending on the length of the granulation. The number of particles per image measured ranged from 600 to 1700. In the sampling cuvette a pulsed air pressure is used to return the sample to the process between each imaging time-point. The air pulse also cleans the glass window of the cuvette, preventing window fouling. The camera is situated horizontally to the window and sample surface. The viewing direction is kept constant, but the direction of the incident illumination is varied. The resulting gradient fields contain direct information about surface normal in xz plane and indirect information about surface normal in yz plane. Line integration was used in horizontal direction to obtain a 3D surface.

The numerical particle size values were obtained from the 3D images by assuming the peaks on the 3D surface to be particles. The volume (V)-based particle size (d) is then calculated from the area of peaks (a) in xy direction:

d = √a * c

(Equation 1)

V = d3 c in Eq. 1 is calibration constant, calibrated with six different-sized (100–1,400 μm) spherical cellulose particles, cellets (Syntapharm, Mülheim an der Ruhr, Germany). The operating principle of the 3D imaging system has been described in more detail earlier [21]. The arithmetic average of the brightness profile from each image was extracted from the image data and used to describe the granule surface brightness during drying as described by Burggraeve and colleagues [22].

2.3. NIR spectroscopy

NIR spectra were continuously collected from each granulation process through the doublecuvette sampler with a NIR spectrophotometer over the spectral range 1081-2250 nm (Control Development, South Bend, USA). The median particle size at each granulation time point was extracted from the untreated NIR spectra by plotting the spectral height (i.e. counts) at 1288 nm against time. The spectral baseline shifts at this wavelength are attributed to particle size changes due to minimal chemical absorption. Moreover, this wavelength has been used for particle sizing earlier [23]. The particle size corresponding to each spectral height was obtained by referencing the NIR particle size curves to the image-based particle size curves.

2.4. Characterisation of the granules The particle size distributions of the final granule batches were measured in triplicate by photometric imaging. The entire batches were fed through a hopper connected to the camera based on a procedure described earlier [21]. The particle size distributions of samples (n=3) obtained from each batch were also measured by Helos laser diffractometer with a Rodos disperser unit and vibrational feeder (Sympatec Gmbh, Clausthal, Germany). The bulk and tap densities of the final granules were measured by the European pharmacopeia method (Erweka Apparatebau GmbH, Germany). Carr’s index and Hausner ratio were calculated from the bulk and tap densities.

3. Results and discussion 3.1. On-line photometric imaging and NIR spectroscopy

3.1.1. Granule growth and size reduction

The granule growth curves plotted from the image data show that batch I forms granules very rapidly (Figure 2a). The process images reveal that batch I has started to agglomerate when liquid has been sprayed for one minute but the majority of the formulation remains powdery. The major growth occurs during the first 2-3 minutes of spraying, after which moderate increase in particle size is observed. This along with the spherical appearance of the batch I

granules can be distinguished in the images. However, introduction of MCC into the formulation slows down the granule growth rate leading to a rather constant growth throughout the spraying phase (Figure 2b-e). The batch V remains very powdery after one minute of spraying (Figure 2e). After three minutes of spraying, batch V has formed a fluffy yet powdery mass while the batch I is mainly granular (Videos 1-2). In the end of the spraying phase, MCC has formed very weak agglomerates (Figure 2e-f). The batches consisting of both fillers show slower growth with a higher MCC proportion. The granule growth reducing effect of MCC can clearly be seen in the images captured from batch II: relatively decent granules are formed but the growth is slow (Figure 2b). The granulation behaviour of all binary formulations is relatively similar even though the particle size reducing effect of increasing amounts of MCC can be seen in the images. The particle size growth of all the formulations is clearly visible in both the process images and plots. Surprisingly, the growth plots suggest that the particle size of both batch I and V granules reaches approximately 550 µm at the end of the spraying phase (Figure 2a and e) . However, the corresponding images indicate that the particle size of batch V granules is significantly smaller. The overestimation arises from the poor packing of wet MCC which leads to shading effects that create areas that are erroneously calculated as particles. In the drying stage, the batch I granules retain their particle size with only slight diminution in the end, which is also apparent from the process images (Figure 2a). By contrast, a rapid decrease in particle size is observed in batches containing MCC (II-V) after the granulation liquid feed is stopped (Figure 2b-e). The breakage of the loose MCC agglomerates of batch V during drying leading to perfectly powdery end product is also clearly visible in the process images (Figure 2e). Even the end product in batch II appears to have a substantial amount of fines (Figure 2b). The size reduction accelerates with growing MCC proportion. The final granule size of batches containing MCC (II-V) ranges from 180 to 200 µm compared to 470 µm for batch I. The final particle size in batches that contain at least 50% MCC i.e. batches III-V is also very close to that of the initial powder (Figure 2c-e). Attrition of formulations containing a large amount of MCC compared to lactose was earlier explained to result from the longer drying time of MCC [3]. However, the current surface imaging approach revealed that the MCC granule breakage takes place immediately when drying is started. The particle size of batches IV and V decreases rapidly during drying until almost the initial starting material particle size is reached (Figure 2d-e). Approximately at the same time, the mass temperature begins to increase faster (Figure 3) and the outlet humidity begins to decrease

after the initial increase (Figure 4). Thus, fast breakage of the granules appears to occur as long as water is rapidly removed from the mass. Fines have been reported to form when the bulk water content is below 6 % [24]. Further drying also accelerates the formation of fines. By contrast, at water contents higher than 6% fines formation is negligible. Surprisingly, the particle size of batches II and IV appears to increase slightly during the last 300 seconds of drying (Figure 2a and d). However, based on the granule images the measurement is erroneous and in reality results from inhomogeneous granule packing. The uneven powder packing is likely to result from the dry fines becoming static, which was visually observed during the processes. This hypothesis is also supported by the fact that the batches II and IV had the lowest inlet air humidity during processing (Table 2). Compared to the photometric method, NIR spectroscopy suggests smaller particle sizes in the spraying phase and faster size reduction of batch I during drying (Figure 5). The attrition behaviour of the other formulations is similar in both methods. In the end of the spraying phase, the median batch I granule size measured by NIRS and images is very similar. However, images suggest larger particle sizes for the other batches and the difference between the methods grows with increasing MCC proportion. The final particle sizes of batches I-III measured by NIR correlate well with the image data but the images give again larger results for the batches IV and V. Based on these findings, the surface imaging method can calculate the particle size of dense and well-packing granules. However, the accuracy of the particle size measurements decreases when poorly-packing powders are measured. Yet, the particle size trends can be followed by the imaging method throughout the granulation process. The granule growth and size reduction disparities between lactose and MCC reflect their behaviour in contact with water. MCC absorbs water due to its amorphous regions while lactose keeps the added water at the surface and has higher wettability than MCC. Based on a granulation regime map proposed by Iveson and colleagues [25], the material differences and location of water could be explained in more detail by pore saturation. In the current study, the faster outlet air humidity increase in the beginning of lactose granulation could indicate more extensive pore saturation. Quick pore saturation could be critical in terms of granule growth as it reflects the amount of water available for liquid bridge formation. Also granule breakage during drying has been explained by liquid volumes required for adequate liquid

bridge formation [26]. Below a minimum liquid bridge volume substantial size reduction occurs, observed as rapid changes in granule size and formation of fines. Above a minimum liquid bridge volume, sufficient amount of surface water enables granule growth. The current results suggest that already a minor amount of MCC absorbs the granulation liquid so rapidly that adequate liquid bridge formation between the colliding particles is hindered. Thus, the liquid bridges are broken during drying before the particles can form solid bridges. Moreover, in formulations containing water soluble filler and binder, the solid bridges are formed by coprecipitation of the filler and polymer [27]. Thus, the disparity between lactose and MCC in the drying step is likely to arise partly from lactose and PVP coprecipitating to form strong solid bridges unlike the non-water soluble MCC. There is also variation in the ability of binders to bind different fillers [27]. It has been proposed that granule growth behaviour can be divided into two groups: steady or induction growth [28]. Steady growth is typical for deformable and weak granules that have a large contact area. By contrast, slowly-consolidating granules are not able to form a strong bond due to insufficient deformation. Thus, the collided granules break apart rapidly leading to an induction period with little or no granule growth. In the light of this theory, rapid growth appears to be typical for lactose and induction growth for MCC.

3.1.2. Granule surface brightness Increase in pellet surface brightness has been shown to correlate with drying [22]. In the current study, however, surface brightness decreased during the drying of the batches II-V, finally reaching a plateau (Figure 6). Similar behaviour was observed in the batch I after an initial brightness increase. A shift in the brightness slopes of batches II and III occurred at time points 1600 s and 1700 s, respectively. Moreover, the image brightness of batches IV and V remained unchanged for approximately 250 seconds in the middle of drying, simultaneously with the transition in mass temperature kinetics (Figure 3). The results suggest that decreasing granule surface brightness is an indicator of drying for lactose monohydrate and MCC granules. This results from reduced granule surface reflectivity at lower water amounts. The initial increase in lactose surface brightness is related to a rapid increase in free surface water due to quick free water removal from the matrix when drying is started. Image brightness has previously been shown to increase as long as water was

removed from pellets and monohydrous theophylline was converted into the anhydrous form [22]. Comparison between the earlier and current results shows that change in the image brightness is connected to granule or pellet drying. However, the direction of the change appears to depend on the formulation, dosage form, the location of water in the product and the occurrence of polymorphic changes. The impact and mechanims of these individual variables needs further investigations.

3.2. Final granule characteristics The particle sizes and the bulk and tapped densities of the final granules are shown in table 3. Surface imaging gives larger particle sizes than laser diffractometer, excluding batch I. The d50 values of batch I obtained from images and laser diffraction are rather similar, 827 µm and 881 µm, respectively. The deviation between the methods generally grows with increasing MCC proportion and presumably originates from the breakage of the fragile MCCcontaining granules during handling, sampling and laser measurements. Vibratory impact can reduce particle size and high laser diffractometer dispersion pressure leads to size reduction of fragile granules but not pellets [19, 30]. Also different measuring principles and the assumption of the particles being smooth and spherical contribute to the observed differences [31-32]. Thus, a considerable strength of surface imaging compared to laser diffraction is that the reliability of the results can easily be evaluated from the images.

The bulk density of the final batch V granules falls into the same range with the starting material. Moreover, Carr’s index and Hausner ratio imply poor to fair flowability of the batches III, IV and V. Together with the surface images, the final granule characteristics confirm that large amounts of MCC prevented the formation of granules with sufficient quality for downstream processability.

3.3. Prospects of surface imaging in monitoring fluid bed granulation

The current results demonstrate that images from the entire granulation process provide valuable information on material characteristics and behaviour during manufacturing. Thus, in addition to auspicious process monitoring application, research and development could

benefit enormously from increased use of visual information. For example, images could give valuable insight into the behaviour of new excipients or formulations during processing. Another great advantage of recording process video and taking images during granulation is the creation of an image library. If formulation-related problems arise in subsequent manufacturing processes, such as tabletting, it is possible to explore the image library with e.g. a content-based image retrieval approach to find the explanation [33]. Together with other recorded in-line or on-line data from the process the image information also opens up novel data mining possibilities. The increased use of systematic data analysis from large databases could lead to e.g. process improvements, facilitated design space creation and easier scale-up. The significance of creating data libraries from processes will become even more pronounced in the future when the manufacturing processes become automated to a larger extent.

Conclusions

This paper has presented a surface imaging approach for on-line monitoring of fluid bed granulation. The images revealed that already a minor proportion of MCC dominated the granulation behaviour of formulations consisting of lactose and MCC. The batches containing MCC exhibited also pronounced size reduction upon drying. Moreover, decreasing image brightness during drying reflected the removal of water from the granules. The unrivalled feature of photometric imaging is that the continuously captured images provide direct visual information coupled with numerical data. Combined with complementary process analytical tools, e.g. NIR spectroscopy, surface imaging is a value-adding tool for monitoring fluid bed granulation.

Acknowledgements

FinPharma Doctoral Program (Ministry of Education, Finland) and Orion Research Foundation are acknowledged for funding the study. Heikki Räikkönen (University of Helsinki, Finland) is greatly acknowledged for technical support.

References

[1] BJ Ennis, JD Litster, Particle Size Enlargement, in: R. Perry and D. Green, (Eds.) Peny’s Chemical Engineers’ Handbook, 7th ed., 1997 , p. 20. [2] DM Parikh, JA Bonck, M. Mogavero, in: D.M. Parikh (Ed.), Handbook of Pharmaceutical Granulation Technology, Marcel Dekker, Inc., New York, , 1997, p. 227. [3] J Rantanen, A Jørgensen, E Räsänen, P Luukkonen, S Airaksinen, J Raiman, K Hänninen, O Antikainen, J Yliruusi, Process analysis of fluidized bed granulation, AAPS PharmsciTech . 2(4) (2001), article 21 http://dx.doi.org/10.1007/BF02830561

[4] A Burggraeve, AFT Silva, T Van Den Kerkhof, M Hellings, C Vervaet, JP Remon, Y Van der Heyden, T De Beer, Development of a fluid bed granulation process control strategy based on real-time process and product measurements. Talanta 100 (2012) 293–302 http://dx.doi.org/10.1016/j.talanta.2012.07.054

[5] P Frake, D Greenhalgh, SM Grierson, JM Hempenstall, DR Rudd, Process control and end-point determination of a fluid bed granulation by application of near infra-red spectroscopy. Int J Pharm. 151 (1997) 75–80 http://dx.doi.org/10.1016/S0378-5173(97)04894-1

[6] J Rantanen, S Lehtola, P Rämet, J-P Mannermaa, J Yliruusi, On-line monitoring of moisture content in an instrumented fluidized bed granulator with a multi-channel NIR moisture sensor. Powder Technol. 99(2) (1998) 163–170 http://dx.doi.org/10.1016/S0032-5910(98)00100-4

[7] J Mantanus, E Ziemons, P Lebrun, E Rozet, R Klinkenberg, B Streel, B Evrard, PH. Hubert, Moisture content determination of pharmaceutical pellets by near infrared spectroscopy: Method development and validation. Anal Chim Acta 642 (2009) 186–192 http://dx.doi.org/10.1016/j.aca.2008.12.031 [8] K Ravikanth, H Qu, R Mattes, B Jancsik, RM Fahmy, SW Hoag, Application of in-line near infrared spectroscopy and multivariate batch modeling for process monitoring in fluid bed granulation, Int J Pharm. 452 (2013) 63–72 http://dx.doi.org/10.1016/j.ijpharm.2013.04.039 [9] S Matero, S Poutiainen, J Leskinen, K Järvinen, K Ketolainen, S-P Reinikainen, M Hakulinen, R Lappalainen, A Poso, The feasibility of using acoustic emissions for monitoring of fluidized bed granulation. Chemometrics and Intelligent Laboratory Systems 97 (2009) 75–81 http://dx.doi.org/10.1016/j.chemolab.2008.11.001

[10] S Watano, K Miyanami: Image processing for on-line monitoring of granule size distribution and shape in fluidized bed granulation. Powder Tech 83 (1995) 55–60 http://dx.doi.org/10.1016/0032-5910(94)02944-J

[11] A Burggraeve A, T Van Den Kerkhof, M Hellings, JP Remon, C Vervaet, T De Beer, Evaluation of in-line spatial filter velocimetry as PAT monitoring tool for particle growth during fluid bed granulation. Eur J Pharm Biopharm. 76 (2010) 138–146 http://dx.doi.org/10.1016/j.ejpb.2010.06.001

[12] J Rantanen, J Yliruusi, Determination of Particle Size in a Fluidized Bed Granulator With a Near Infrared Set-up. Pharm Pharmacol Commun. 4 (1998) 73-75 http://dx.doi.org/10.1111/j.2042-7158.1998.tb00509.x [13] X Chen, D Chen, Measuring average particle size for fluidized bed reactors by employing acoustic emission signals and neural networks. Chem Eng Technol. 31 (2008) 95– 102 http://dx.doi.org/10.1002/ceat.200700176 [14] X Hu, JC Cunningham, D Winstead, Study growth kinetics in fluidized bed granulation with at-line FBRM. Int J Pharm. 347 (2008) 54–61 http://dx.doi.org/10.1016/j.ijpharm.2007.06.043. [15] M Börner, M Peglow, E Tsotsas, Derivation of parameters for a two compartment population balance model of Wurster fluidised bed granulation, Powder Tech. 238 (2013) 122–131 http://dx.doi.org/10.1016/j.powtec.2012.04.014 [16] N Laitinen, J Rantanen, O Antikainen, J Yliruusi, New perspectives for visual characterization of pharmaceutical solids. J Pharm Sci 93 (2004) 165–176 http://dx.doi.org/10.1002/jps.10529 [17] T Närvänen, K Seppälä, O Antikainen, J Yliruusi, A new rapid on-line imaging method to determine particle size distribution of granules, AAPS PharmSciTech 9 (2008) 282-287 http://dx.doi.org/10.1208/s12249-008-9043-y [18] N Sandler, Photometric imaging in particle size measurement and surface visualization, Int J Pharm 417 (2011) 227– 234 http://dx.doi.org/10.1016/j.ijpharm.2010.11.007

[19] AFT Silva, A Burggraeve, Q Denon, P Van der Meeren, N Sandler, T Van Den Kerkhof, M Hellings, C Vervaet, JP Remon, JA Lopes, T De Beer, Particle sizing measurements in pharmaceutical applications: Comparison of in-process methods versus off-line methods, Eur J Pharm Biopharm. 85 (2013) 1006 -1018 http://dx.doi.org/10.1016/j.ejpb.2013.03.032

[20] A Burggraeve, T Monteyne, C Vervaet, JP Remon, T De Beer, Process analytical tools for monitoring, understanding, and control of pharmaceutical fluidized bed granulation: A review, Eur J Pharm Biopharm 83 (2013) 2–15 http://dx.doi.org/10.1016/j.ejpb.2012.09.008

[21] I Soppela, S Airaksinen, J Hatara, H Räikkönen, O Antikainen, J Yliruusi, N Sandler, Rapid Particle Size Measurement Using 3D Surface Imaging, AAPS PharmSciTech. 12 (2011) 476–484 http://dx.doi.org/10.1208/s12249-011-9607-0.

[22] A Burggraeve, N Sandler, J Heinämäki, H Räikkönen, JP Remon, C Vervaet, T De Beer, J Yliruusi, Real-time image-based investigation of spheronization and drying phenomena using different pellet formulations, Eur J Pharm Sci. 44 (2011) 635–642 http://dx.doi.org/10.1016/j.ejps.2011.10.011

[23] BG Osborne, S Douglas, T Fearn, Assessment of wheat grain texture by near infrared reflectance measurements on bühler-milled flour, J Sci Food Agric. 32 (1981) 200–202 [24] F Nieuwmeyer, K van der Voort Maarschalk, H Vromans, The consequences of granulate heterogeneity towards breakage and attrition upon fluid-bed drying, Eur J Pharm Biopharm. 70 (2008) 402–408 http://dx.doi.org/10.1016/j.ejpb.2008.03.003

[25] S Iveson, J Litster, K Hapgood, B Ennis: Nucleation, growth and breakage phenomena in agitated wet granulation processes: a review, Powder Tech. 117 (2001) 3–39 http://dx.doi.org/10.1016/S0032-5910(01)00313-8

[26] F Nieuwmeyer, K van der Voort Maarschalk, H Vromans, Granule breakage during drying processes, Int J Pharm. 329 (2007) 81–87 http://dx.doi.org/10.1016/j.ijpharm.2006.08.017

[27] D Bika, GI Tardos, S Panmai, L Farber, J Michaels, Strength and morphology of solid bridges in dry granules of pharmaceutical powders, Powder Technol. 150 (2005) 104–116 http://dx.doi.org/10.1016/j.powtec.2004.11.024

[28] SM Iveson, JD Litster, Growth Regime Map for Liquid-Bound Granules, Particle Technology and Fluidization AIChE Journal 44 (1998) 1510–1518 http://dx.doi.org/10.1002/aic.690440705 [29] G Zografi, M Kontny, The interactions of water with cellulose-and starch-derived pharmaceutical excipients, Pharm Res. 3 (1986) 187–194 http://dx.doi.org/10.1023/A:1016330528260

[30] S Antonyuk, M Khanal, J Tomasa, S Heinrich, L Mörl, Impact breakage of spherical granules: Experimental study and DEM simulation. Chem Eng Process. 45 (2006) 838–856 http://dx.doi.org/10.1016/j.cep.2005.12.005 [31] BY Shekunov, P Chattopadhyay, HHY Tong, AHL Chow, Particle size analysis in pharmaceutics: principles, methods and applications, Pharm Res. 24 (2007) 203-227 http://dx.doi.org/10.1007/s11095-006-9146-7 [32] C Andrès, P Réginault, MH Rochat, B Chaillot, Y Pourcelot, Particle-size distribution of a powder: comparison of three analytical techniques, Int J Pharm. 144 (1996) 141-146 http://dx.doi.org/10.1016/S0378-5173(96)04737-0

[33] N Laitinen, O Antikainen, J-P Mannermaa, J Yliruusi, Content-Based Image Retrieval: A Promising Technique in Powder Technology, Pharm Dev Tech. 5 (2000) 171-179

Table I. Filler proportions in the batches Batch

Lactose (%)

MCC (%)

I

100

0

II

75

25

III

50

50

IV

25

75

V

0

100

Table II. Amount of water in the incoming process air in each batch.

I

II

III

IV

V

Total water amount of the process air flow (g) Spraying

355

250

241

166

212

Drying

290

158

211

183

335

Total

645

408

452

349

547

Table III. The median particle sizes and densities of final granules.

Batch

d50 image (µm)

d50 laser (µm)

Dbulk (g/ml)

Dtap (g/ml)

I

827

881

0.52

0.60

II

400

280

0.38

0.44

III

321

163

0.32

0.39

IV

213

97

0.29

0.37

V

147

74

0.28

0.36

FIg 1

Fig 2a

Fig 2b

Fig 2c

Fig 2d

Fig 2e

Fig 2f

Fig 3

Fig 4

Fig 5

Fig 6

Graphical abstract

Highlights - Photometric imaging in on-line monitoring of fluid bed granulation is discussed - Images provide direct real-time information on granule growth, attrition and packing - Decreasing image brightness in the drying phase is connected to granule drying - Images give valuable insight into the behaviour of formulations during processing