Contents lists available atScienceDirect
International Journal of Pharmaceutics
journal homepage:www.elsevier.com/locate/ijpharmInfluence of binder attributes on binder effectiveness in a continuous twin
screw wet granulation process via wet and dry binder addition
L. Vandevivere
a, P. Denduyver
a, C. Portier
a, O. Häusler
b, T. De Beer
c, C. Vervaet
a,
V. Vanhoorne
a,⁎aGhent University, Laboratory of Pharmaceutical Technology, Ottergemsesteenweg 460, 9000 Ghent, Belgium bRoquette Frères, Rue de la Haute Loge, 62136 Lestrem, France
cGhent University, Laboratory of Pharmaceutical Process Analytical Technology, Ottergemsesesteenweg 460, 9000 Ghent, Belgium
A R T I C L E I N F O Keywords:
Continuous manufacturing Twin screw granulation Wet granulation Pharmaceutical binders Binder attributes Binder addition method
A B S T R A C T
The effect of a wide variety of binders on the quality of granules produced via continuous twin screw wet granulation was studied. Anhydrous dicalcium phosphate was used as poorly soluble filler and was granulated applying dry or wet addition of binders. Furthermore, dry and wet binder characteristics were determined and linked to the binder effectiveness. PVA 4–88 and starch octenyl succinate exhibited the lowest granule friability at low liquid-to-solid ratios, i.e. the highest binder effectiveness, which was attributed to fast binder activation based on the fast wetting kinetics of the binder, to efficient wetting of DCP particles, and to good spreading in the powder bed. The performance of wettability measurements in an early formulation development stage is therefore considered highly important. Additionally, an increased stickiness of the binder surface caused by high binder viscosity and slow dissolution kinetics also positively influenced the binder effectiveness. In conclusion, this study revealed which binder attributes have a critical impact on the granulation process of dicalcium phosphate. Additionally, dry binder addition proved successful for creation of high quality granules.
1. Introduction
Continuous twin screw wet granulation (TSG) is of great interest as it can be implemented into a fully continuous from-powder-to-tablet line. Therefore, several system suppliers have developed a continuous manufacturing line integrating TSG. As a switch to continuous manu-facturing is quickly gaining momentum in the pharmaceutical industry, the suitability of conventionally used pharmaceutical excipients needs to be evaluated towards this novel granulation process. However, as typical residence times during TSG are much shorter (5 – 20 s) com-pared to batch granulation processes (10 – 30 min), this may limit the transferability of excipients between batch and continuous processes. (De Leersnyder et al., 2018; El Hagrasy et al., 2013; Kumar et al., 2014; Vercruysse et al., 2014)
A binder is often added to the formulation to facilitate the granu-lation process. Ideally, a cohesive network between the formugranu-lation ingredients is created by the binder in order to produce granules and tablets that meet the predetermined quality target (e.g. in terms of
granule size distribution, granule friability, tablet tensile strength, and tablet content uniformity). The most common pharmaceutical binders are sugars (e.g. maltodextrins), natural polymers (e.g. starch), (semi-) synthetic polymers (e.g. polyvinylpyrrolidone (PVP)), and cellulose-based polymers (e.g. hydroxypropylmethylcellulose (HPMC)). Binders can be added as a solution prepared upfront (i.e. wet binder addition) or as a dry powder in the powder blend (i.e. dry binder addition). The effect of binder addition in TSG has been studied by several authors. El Hagrasy et al. mentioned a smaller proportion of fines and a more narrow size distribution when a larger proportion of the binder was added to the liquid phase during TSG. Fonteyne et al. also demonstrated a higher fraction of large granules, if the binder was added via the granulation liquid. A higher binder effectiveness via the wet addition method was also mentioned by Vercruysse et al. However, there is no consensus in literature concerning the preferred binder addition method as Keleb et al. observed no significant differences in granule properties when granulating α-lactose monohydrate with dry and wet addition of PVP.(El Hagrasy et al., 2013; Fonteyne et al., 2014; Keleb
https://doi.org/10.1016/j.ijpharm.2020.119466
Received 25 February 2020; Received in revised form 20 May 2020; Accepted 21 May 2020
Abbreviations: CA, Contact angle; DCP, Dicalcium phosphate; DE, Dextrose-equivalent; HP, Hydroxypropyl; HPMC, Hydroxypropyl methylcellulose; iGC, Inverse gas chromatography; L/D, Length-to-diameter; L/S, Liquid-to-solid; Min, Minutes; PCA, Principal component analysis; PTFE, polytetrafluoroethylene; PVA, Polyvinyl alcohol; PVP, Polyvinylpyrrolidone; Rpm, rotations per minute; S, seconds; SSA, Specific surface area; ST, Surface tension; TSG, Twin screw granulation
⁎Corresponding author.
E-mail address:[email protected](V. Vanhoorne).
International Journal of Pharmaceutics 585 (2020) 119466
Available online 25 May 2020
0378-5173/ © 2020 Elsevier B.V. All rights reserved.
et al., 2002; Vercruysse et al., 2012) From an industrial point of view, preparation of the binder solution and pumpability issues limit the processability of wet binder addition. Therefore, dry binder addition is preferred because of its ease of processing. Nevertheless, certain binders can exhibit cohesiveness and stickiness limiting accurate dry feeding of the binder.
The selection of a suitable binder to agglomerate a drug with ex-cipients is critical, as it influences the properties of granules (e.g. me-chanical strength, granule growth, particle size distribution).(Becker et al., 1997; Chitu et al., 2011; Parker et al., 1991; Ritala et al., 1986) Accordingly, an improved understanding of binder effectiveness is fundamental for more efficient formulation design during TSG. Will-ecke et al. studied the impact of fillers and binders on drug product quality using a systematic statistical approach. The scores of a principle component analysis (PCA) performed on binder characteristics were used as factors in a design of experiment. They concluded that binder properties (e.g. binder viscosity and wettability) impacted the granule quality. However, the binder concentration was not included as a factor in the experimental design.(Willecke et al., 2017b) The impact of binder properties on product quality was also examined by Dhenge et al., showing that the properties of the granulation liquid influenced the granule quality in TSG. Increasing binder viscosity yielded unim-odal granules with higher strength and better flow properties. However, these conclusions were based on experiments with a single binder type (hydroxypropylcellulose) used in different concentrations.(Dhenge et al., 2012a) These studies highlighted that the effect of different binders on granule quality was affected by binder attributes in a wet granulation process. Therefore, critical binder attributes need to be identified to allow a thorough and systematic binder selection.
Although the effect of several binders on granule quality for TSG has already been evaluated (Dhenge et al., 2012b, 2012a; Vercruysse et al., 2012; Willecke et al., 2017b), comparison between different binder types is hampered by the use of different formulations in the studies as other components also contribute to the binding performance (e.g. based on their solubility in the granulation liquid), affecting the gran-ulation process.(Vanhoorne et al., 2016a) Therefore, the correlation between binder effectiveness and binder attributes was examined in current study using a standard formulation in order to facilitate for-mulation design for TSG. Hereby, binder effectiveness was referring to the lowest amount of liquid needed to obtain a sufficiently low granule friability. Additionally, the impact of the dry and wet binder addition method and of binder concentration was studied. Next to the binder, the formulation consisted solely of dicalcium phosphate (DCP). As DCP is a poorly soluble excipient, this component does not contribute to the binding (Rowe et al., 2009), all interactions within the granule are therefore attributed to the binder. Accordingly, the use of DCP allowed to compare the intrinsic effect of a wide variety of binders on granule properties. As granule friability is an estimation for granule strength, the granule characterization focused on the granule friability.
2. Materials and methods
2.1. Materials
Anhydrous DCP (Calipharm® A, Innophos, Chicago Heights, USA) was used as poorly soluble model excipient. Maltodextrins with dex-trose-equivalent (DE) of 2 and 6, obtained from waxy maize starch (Glucidex® 2 and 6, Roquette Frères, Lestrem, France), maltodextrin with DE of 12, obtained from maize starch (Lycatab® DSH, Roquette Frères, Lestrem, France), starch octenyl succinate (Cleargum® CO 03, Roquette Frères, Lestrem, France), hydroxypropyl (HP) pea starch (Lycoat® RS 720, Roquette Frères, Lestrem, France), PVP K12, K30 and K90 (Kollidon® K12, K30 and K90, BASF, Ludwigshafen, Germany), HPMC E5 and E15 (Methocel® E5 and E15, Dow Chemical Company, Rheinmünster, Germany), and polyvinyl alcohol (PVA) 4–88 (Parteck® MXP, Merck, Darmstadt, Germany) were used as pharmaceutical biders. All reagents for inverse gas chromatography (iGC) (hexane, n-heptane, octane, nonane, ethyl acetate and dichloromethane) were obtained from Sigma Aldrich (USA), and were of high performance li-quid chromatography grade.
2.2. Preparation of granules
Granulation experiments were performed using the twin screw granulator of a continuous from-powder-to-tablet line (ConsiGmaTM-25 system, GEA Pharma Systems, Wommelgem, Belgium). This continuous granulator consists of two co-rotating screws with a length-to-diameter (L/D) ratio of 20/1. DCP was granulated applying dry or wet binder addition. Preblends of DCP (95% w/w) with the different binders (5% w/w) were prepared in a tumbling mixer (Inversina Bioengineering, Wald, Switzerland), blending for 15 min (min) at 25 rotations per minute (rpm). Each dry premix was transferred to the twin screw loss-in-weight feeder (KT20, K-Tron Soder, Niederlenz, Switzerland) and subsequently dosed into the barrel of the granulator. Demineralized water was used as granulation liquid when the binder was added via the preblend. The demineralized water was equilibrated at room tempera-ture before execution of the experiments.
For the wet binder addition method, an aqueous binder solution was prepared using a Typhoon mixer (Typhoon Roertechniek, Raamdonksveer, The Netherlands) and used as granulation liquid. The solutions were directly used after preparation. The concentration of the binders in the granules was 1.44% (based on dry powder). In addition, depending on the viscosity of the binder solution, granules with higher binder concentration (3 or 5%) were prepared for specific binders. To evaluate the impact of binder addition, granules were prepared via the dry binder addition method including the highest possible wet binder concentration in the granules (1.44, 3 or 5%). For both addition methods, the granulation liquid was gravimetrically pumped into the barrel via dual liquid feed ports, injecting the liquid on top of each screw. This addition was performed using two out-of-phase peristaltic
Flow direction
Powder addition Liquid addition
Fig. 1. Schematic overview of the screw configuration with conveying elements (green), kneading elements (orange) with a length-to-diameter ratio of ¼ and size control elements (blue). The horizontal arrow indicates the powder flow direction. Only one screw is illustrated.
pumps (Watson Marlow, Cornwall, UK) and two silicon tubings, both connected to a 1.6 mm nozzle. Due to processability, not all binders could be processed at similar liquid-to-solid (L/S) ratios. Consequently, in order to produce low friable granules with each binder, different L/S-ratios were used depending on the binder. The screw configuration contained two kneading zones of 6 kneading elements. These were se-parated by a conveying zone with an L/D of 1.5. The kneading elements were arranged at a forward stagger angle of 60°. Size control elements were present at the end of the screws in order to obtain breakage of oversized granules.(Portier et al., 2020b) The screw configuration is illustrated inFig. 1. The high density and poor aqueous solubility of DCP required a high filling degree of the granulator barrel to obtain sufficient densification. This high filling degree was achieved by fixing the screw speed and the throughput at 300 rpm and 20 kg/h, respec-tively. The barrel jacket was equipped with an active cooling system in order to maintain the temperature during processing at 30 °C. Active cooling of the barrel was necessary for experimental runs resulting in higher torque values. Torque was monitored by a built-in torque-gauge at 1-second intervals. After stabilization of the torque (maximal de-viation of 1.0 Nm during one minute prior to sampling), 800 g of the produced granules were sampled at the outlet of the granulator and tray dried at 40 °C until a loss-on-drying (LOD) value between 1 and 3% was achieved. The LOD was determined using a moisture analyzer (Mettler LP16, Mettler-Toledo, Zaventem, Belgium) including an infrared dryer and a balance. A sample of granules (1 g) was dried at 105 °C until the weight was constant for 30 s.
2.3. Methods for binder characterization 2.3.1. Particle size distribution
The particle size distribution of the starting materials was measured in triplicate by laser diffraction (Malvern Mastersizer 2000, Malvern Instruments, Worcestershire, UK). The measurements were performed via the dry dispersion method in volumetric distribution mode, using a 300F lens combined with a dry dispersion unit at a feed rate of 3.0 G and a jet pressure of 3.5 bar. The particle size is reported as a volume equivalent sphere diameter. For each volumetric distribution, the 10%, 50% and 90% cumulative undersize was reported as Dv_10, Dv_50, Dv_90, respectively. Furthermore, the span (d_span) was calculated to describe the particle size distribution and equals (dv_90-dv_10)/dv_50.
2.3.2. Specific surface area
Samples ( ± 1.0 g) were degassed for 24 h at 60 °C using the va-cuum mode of the VacPrep (Micromeritics, Norcross, USA) and subse-quently purged with nitrogen during one hour. Afterwards, N2-sorption experiments (Micromeritics Tristar I) were performed at −196 °C using a seven point method. The internal surface area was calculated making use of the Brunauer-Emmet-Teller theory. The determination of the specific surface area (SSA) was a single measurement for each binder. The SSA was reported in the PCA plot as SSA.
2.3.3. Dissolution kinetics
Binders were characterized by their dissolution rate in deminer-alized water. First, for each binder, a calibration curve was set up by measuring the refractive index of five different aqueous binder solu-tions. Second, the binder (10% w/w) was added to demineralized water, mixed for 30 s (s) (Bamix mixer, Mettlen, Switzerland) and fil-tered using a 110 mm diameter, cellulose based filter (Grade 2, Whatman, USA). Finally, the refractive index of the filtrate was de-termined and the amount of dissolved binder was calculated. The test was performed in triplicate for each binder and repeated for a mixing time of 60 and 90 s. Due to identical results obtained for the repetitions, no standard deviation could be calculated. The dissolved binder frac-tion at 30, 60 and 90 s was reported as DissRate_30, DissRate_60 and DissRate_90, respectively.
2.3.4. Wettability
Wettability of the binders was determined via contact angle mea-surements, using a Drop Shape Analyzer (DSA 30, KRÜSS, Hamburg, Germany) applying the sessile drop method. Three types of experiments were performed. (i) The binders were compressed in a 13 mm diameter die using a hydraulic press (Specac pellet press, Kent, United Kingdom) at a pressure of 98 kN for 1 min. Using this type of press, low porosity of the tablets (< 10%) was achieved to avoid bias of drop penetration. (Zhang et al., 2002) A 5 µl drop of demineralized water was placed on each tablet. When the droplet was placed on the surface of the tablet the contact angle (CA) (°) was measured immediately (CAbinder_t0), and after 30 s. The difference between the initial CA and the CA measured after 30 s indicated the wetting kinetics of the binder (CAbinder_(t0-t30)). Five replicate measurements were carried out. (ii) DCP was compressed with a compaction simulator (Styl’One Evolution, Medel-pharm, Lyon, France) at a pressure of 20 kN with a 10 mm diameter punch. As tableting of pure DCP resulted in high ejection forces, higher forces could not be applied, yielding tablets with a high porosity. Aqueous binder solutions were prepared (8% w/w) and 5 µl of each solution was placed on the DCP tablet. CAs (°) were measured im-mediately when the drop touched the surface of the tablet (CADCP). Due to high porosity, no measurement at 30 s could be made as the droplet penetrated into the tablet structure. Five replicate measure-ments were performed. (iii) A droplet of each binder solution (8% w/w) was placed on a polytetrafluoroethylene (PTFE) surface. As no inter-action took place between the binder solutions and the PTFE surface, CA did not change in function of time and CA (°) was only measured when the drop touched the surface (CAPTFE). Measurements were performed in triplicate.
2.3.5. Surface tension
Aqueous solutions of the binders (8% w/w) were prepared. The concentration of these solutions was also determined as m/V, where m is the mass of the binder liquid (g) and V is the binder liquid volume (ml). The surface tension (ST) of the solutions (mN/m) was obtained via the pendant drop method in air, using a Drop Shape Analyzer (DSA 30, KRÜSS GmbH, Hamburg, Germany). A liquid droplet was formed and held on the needle tip. Subsequently, the ST was calculated by using the curvature of the drop profile.(Winkler, 2010) Each measurement was conducted at room temperature in triplicate.
2.3.6. Viscosity
The dynamic viscosity of the aqueous binder solutions was mea-sured using a rotational rheometer (HAAKE MARS® III, Thermo Scientific, Massachusetts, USA). The plate-plate technique was applied at a temperature of 25 °C. The obtained data was corrected by the rheometer software for the non-homogeneous flow inherent to the plate-plate geometry. Solutions with three different concentrations (Table 1) were studied for each binder. This covered the range of binder concentrations which are typically applied in wet granulation pro-cesses.(Willecke et al., 2017a) A 8% w/w solution was measured for all Table 1
Concentrations of binder solutions for viscosity measurements.
Binder Binder concentrations (% w/w)
PVP K12 8 28 48 PVP K30 8 22 36 PVP K90 4 8 12 HPMC E5 4 8 12 HPMC E15 2 4 8 PVA 4-88 8 14 20 Maltodextrin 2 8 14 20 Maltodextrin 6 8 14 20 Maltodextrin DSH 8 14 20
Starch octenyl succinate 8 24 40
binders in order to have comparable values available for PCA, as it was possible to produce a solution for all binder types/grades at this specific concentration. 5 ml of the sample was loaded between the two plates. The upper plate was manually controlled to ensure an initial gap size, providing good contact between the measuring geometry and the sample (gap range: 0.9 – 1.1 mm). The shear rate ranged from 0.1 1/s to 1000 1/s, collecting 20 data points. The viscosity values (mPa.s) were derived from a comparable range of shear rates at which the binder solutions showed Newtonian behavior (20.7 to 143.8 1/s). These values measured at 8% w/w were used as descriptive values in PCA (Dyna-micViscosity).
The viscosity slope was calculated as an additional descriptor for the viscosity change as a function of the binder concentration. Therefore, the concentration of the binder solution was plotted against the average logarithmic viscosity, and the slope of the resulting linear graph was used for PCA (ViscositySlope).
2.3.7. Inverse gas chromatography
Inverse gas chromatography was conducted using the iGC Surface Energy Analyser 2000 instrument (Surface Measurement Systems, Wembley, UK). Samples were packed into individual iGC silanised glass columns. The injection of solvent vapours was controlled to pass a set volume of eluent through the column to give a predetermined fractional coverage of the sample in the column. The dispersive part of the surface energy was determined using the Dorris and Gray method.(Dorris and Gray, 1980) The neat retention time was obtained from peak maxima. The acid-base properties were determined using the della Volpe scale. (Della Volpe and Siboni, 1997) iGC was only performed for DCP, as the low SSA of certain binders limited the measurement of these binders.
2.4. Principal component analysis
PCA was performed on a dataset including the binders character-istics (listed inTable 2). SIMCA® 16 software (Sartorius Stedium Bio-tech, Umeå, Sweden) was used to analyze this dataset. Data were mean centered and scaled to unit variance. The latter was applied to make the analysis independent of the units used and to allow the simultaneous analysis of quantities which have different magnitudes. A logarithmical transformation was performed if needed to non-normally distributed responses. Block scaling was applied to Dv_10, Dv_50 and Dv_90, and to DissRate_30, DissRate_60 and DissRate_90. This scaling resulted in a variance of the block equaling to 1.
An overview of the characterization techniques with corresponding binder characteristic abbreviations used for PCA is provided inTable 2.
2.5. Granule characterization 2.5.1. Friability
The granule friability was determined in triplicate using a friabilator
(PTFE Pharma Test, Hainburg, Germany) at a speed of 25 rpm for 10 min, by subjecting 10 g (Iwt) of granules together with 200 glass beads (mean diameter 4 mm) to falling shocks. Prior to determination, the granule fraction < 250 µm was removed. Afterwards, the glass beads were removed and the mass retained on a 250 µm sieve (Fwt), was determined. The friability was calculated as [(Iwt– Fwt)/Iwt]*100.
2.5.2. Particle size analysis
The size distribution of 5 granule samples was determined via dy-namic image analysis using the QICPIC™ system with WINDOX 5 soft-ware (Sympatec, Clausthal-Zellerfeld, Germany). The system was equipped with a vibrating feeder system (Vibri/L™) for gravimetrical feeding of the granules.
3. Results and discussion
3.1. Dry binder addition 3.1.1. Granule friability
The granule friability as a function of L/S-ratio is shown inFig. 2. A higher L/S-ratio resulted in less friable granules for each binder. If a higher amount of granulation liquid was added, a larger fraction of the binder interacted with the granulation liquid and formed solid bridges after solidification during drying.(Djuric and Kleinebudde, 2010; Vanhoorne et al., 2016a; Vervaet et al., 1994) A friability below 30% was set as threshold to define the quality of the granules and was used to distinguish in binder effectiveness, as a binder was assumed to be more effective when a lower amount of liquid was needed to obtain the friability limit. This threshold is based on previous studies and is an indicator for attrition and breakage of granules during downstream processing.(Keleb et al., 2002; Portier et al., 2020a, 2020c; Van Melkebeke et al., 2008; Vandevivere et al., 2019; Vanhoorne et al., 2016b) A binder was considered activated when the friability limit was met. PVA and starch octenyl succinate allowed production of granules with low friability (≤30%) at lower L/S-ratios, compared to the other binders (Table 4). Other binders (e.g. HPMC E15 and PVP K90) resulted in very low granule friability at higher L/S-ratios.
Based on the L/S-ratios which were needed to obtain an acceptable granule friability, a difference in binder effectiveness was found. To evaluate which binder attributes contributed to a higher binder effec-tiveness when granulated with DCP, the dry binders were character-ized.
3.1.2. Principle component analysis
When a binder is added in a dry state, it needs to interact with the granulation liquid in order to activate its binding properties. This re-quires wetting of the binder with the granulation liquid to reach a hydrated state of the binder particles. Therefore, both binder solutions and dry binder particles were characterized. PCA was applied on the dataset including 11 observations (binders) and 15 variables (raw binder properties) to evaluate the variation between the included bin-ders. An overview of the included variables is given inTable 2. The raw binder characterization data is summarized inTable 3. A model with two PCs, explaining 48.7% and 21.6% of the variation, respectively, was fitted.
3.1.2.1. Loading scatter plot. The loading scatter plot indicates the
relationship between the binder properties showing which binder properties are positively or negatively correlated and which are not related to each other. Moreover, the plot represents the importance of binder properties towards a PC. Properties with strongly positive or negative loadings have a large influence on the PC, while properties located near the origin are not relevant for the respective PC.Fig. 3 illustrates the PC 1/PC 2 loading scatter plot. The dissolution kinetics, viscosity parameters, surface tension and wettability properties (CAbinder_t0, CAPTFE and CADCP) mainly influenced the first Table 2
Overview of characterization techniques with the corresponding abbreviation of the binder characteristics used in PCA.
Characterization technique Abbreviation binder characteristic Particle size distribution Dv_10, Dv_50, Dv_90, D_Span
Specific surface area SSA
Dissolution kinetics DissRate_30, DissRate_60, DissRate_90 Wettability
CA measured on tablet binder CAbinder_t0 Wetting kinetics of binder CAbinder_(t0-t30) CA binder solution measured on DCP CADCP CA binder solution measured on PTFE CAPTFE
Surface Tension ST
Viscosity parameters
Dynamic viscosity DynamicViscosity
component. The wettability properties (CAbinder_(t0-t30), CAPTFE and CADCP), surface tension and particle size distribution were driving the second component.
The viscosity parameters correlated positively with the initial wet-ting of the binder (CAbinder_t0), and negatively with the dissolution kinetics, the CA measured on PTFE and DCP, and the surface tension. This indicated that highly viscous binders were associated with a poor initial wetting (i.e. high CA values). These binders were also char-acterized by slow dissolution kinetics, which suggested swelling of the binders. The slow dissolution kinetics can be attributed to the high viscosity and the poor initial wetting of the binders. The inverse lationship between viscosity and dissolution rate is generally re-cognized in literature.(Jachowicz, 1987; Rades et al., 2015) The nega-tive correlation with the surface tension and with the CA measured on DCP and PTFE resulted in good spreading and wetting properties of the viscous binder solutions on these surfaces. In addition, the wetting ki-netics of binders were negatively correlated with the CA of binder so-lutions measured on DCP and PTFE, the surface tension and the particle size. Accordingly, the binders possessing fast wetting kinetics also had good wetting properties of DCP and PTFE (low CA), low surface tension and small particle size.
3.1.2.2. Score scatter plot. The score scatter plot illustrates the relation
between the different binders according to their principal properties, where binder clustering indicates similar binder properties. In contrast, binders located on opposite sides of the score scatter plot origin possess distinctly different properties.
The PC 1 versus PC 2 score scatter plot is illustrated inFig. 4. Three binder clusters could be distinguished. (i) PVP K90, HPMC E15 and HPMC E5, clustering at the left bottom corner of the scatter plot, ex-hibited high viscosity, a high initial CA with water, a low CA measured on DCP and PTFE, a low surface tension, and slow dissolution kinetics. (ii) On the right side of the plot a cluster was formed by PVP K12 and maltodextrin 2/6/DSH. Based on the loading plot, these binders were characterized by a fast dissolution rate, a low viscosity, a low initial CA with water, a high CA measured on DCP and PTFE, and a high surface tension. PVP K30 and HP pea starch were situated between these two binder clusters, indicating that their binder properties were inter-mediate compared to these clusters. (iii) Starch octenyl succinate and PVA were located at the top of the plot, mainly due to their wetting behavior: fast binder wetting kinetics and good wetting of PTFE and DCP surfaces.
3.1.3. Link between binder attributes and granule friability
Binder effectiveness differed amongst the binders when evaluating the granule friability (Fig. 2). As PVA and starch octenyl succinate produced granules with a low friability at the lowest L/S-ratios (0.1175
– 0.1300), these binders were considered more effective in comparison to the other binders. Binders which required the highest L/S-ratios (0.2050–0.2300) were defined as least effective, as more granulation liquid was needed to obtain good granule friability. Next to the com-parison of binder effectiveness, differences in absolute values of granule friability were observed. Several binders resulted in very low granule friability (< 10%) (PVA 4–88, HPMC E5/E15, PVP K30/K90, and HP pea starch), while other binders resulted in granule friability between 20 and 30% at their highest processing L/S-ratios (Maltodextrin 2/6/ DSH, starch octenyl succinate and PVP K12).
Binders were assigned a color on the score scatter plot in function of the L/S-ratio required to produce granules with a friability ≤ 30% (Fig. 4). The binders with corresponding color and L/S-ratio are shown inTable 4. Several binders required an L/S-ratio of 0.18 to meet the friability limit. However, at an L/S-ratio of 0.18, granules with sig-nificantly different friability were obtained depending on the binder: 7.4% for HPMC E15, 17.2% for HPMC E5, 22.1% for HP pea starch and 28.6% for PVP K30. To differentiate between these binders, different symbols were assigned to them: HPMC E15 (circle), HPMC E5 and HP pea starch (triangle), and PVP K30 (diamond). Binders needing a si-milar L/S-ratio for efficient granulation (i.e. sisi-milar color) clustered on the score scatter plot, indicating that they had a similar binder effec-tiveness. Furthermore, the binder effectiveness corresponded to the binder properties, because the clustered binders have similar values for binder attributes on the loading scatter plot. The arrow depicted on the score scatter plot (Fig. 4) indicates a flux towards the most optimal binder region as these binders were the most effective at low L/S-ratios when granulated with DCP.
PVA and starch octenyl succinate were the most effective, as these binders resulted in a good granule friability at low L/S ratios. These two binders were mainly influenced by the second PC, mainly defined by wettability, surface tension and particle size distribution. This is illu-strated by the score contribution plot of starch octenyl succinate and PVA (Fig. 5). The high effectiveness at low L/S-ratios of these binders is linked to their fast wetting which rapidly activated the binder (positive correlation with CAbinder_(t0-t30)). Afterwards, the activated binder could efficiently wet the DCP particles, which was attributed to the low CA measured on a DCP and PTFE surface. Compared to the other bin-ders, PVA and starch octenyl succinate resulted in the lowest CA measured on DCP, 36.8° and 32.8°, respectively (Table 3). PTFE is an accepted reference surface which is only capable of forming dispersive interactions, accounting for van der Waals interactions.(Masi, 1990) Moreover, iGC results showed that the dispersive component mainly contributed to the total surface energy of DCP: at low and high eluent coverage, the dispersive component contributed 93% and 81% to the total surface energy, respectively (the surface energy at low and high eluent coverage was 44 mJ/m2and 52 mJ/m2, respectively). The low
0.15 0.20 0.25 0 20 40 60 80 100 HPMC E15 HPMC E5 PVP K90 PVP K30 PVP K12 Maltodextrin 2 Maltodextrin 6 HP pea starch Maltodextrin DSH Starch octenyl succinate PVA 4-88
L/S-ratio
Friability (%)
Fig. 2. Granule friability in function of binder type and L/S-ratio. The dotted line represents a granule friability of 30%, set as threshold to identify granules having sufficient strength.
Table 3 Numerical values of descriptors used for PCA. (*Values could not be measured as it was not possible to produce tablets of PVP K12 due to high ejection forces.) (**Value was too low for accurate measurements.)
Descriptive bindercharacteristics (units)
Particle size distribution (µm) Span of particle size distribution Specific surface area (m 2/g) Dissolution kinetics (%) Abbreviation in PCA Dv_10 Dv_50 Dv_90 D_Span SSA DissRate_30 DissRate_60 DissRate_90 Binder PVP K12 13.5 (± 0.4) 49.3 (± 1.0) 100.4 (± 1.0) 1.76 (± 0.02) 0.12 9.6 9.9 9.9 PVP K30 29.2 (± 0.6) 83.0 (± 0.2) 144.1 (± 1.5) 1.38 (± 0.02) NA** 8 9.5 10 PVP K90 83.0 (± 2.1) 173.0 (± 4.7) 315.1 (± 7.1) 1.34 (± 0.01) 0.59 3 4 6.6 HPMC E5 34.8 (± 0.1) 84.0 (± 0.1) 161.4 (± 0.4) 1.51 (± 0.01) 0.26 2.3 5.1 7.2 HPMC E15 38.6 (± 0.3) 89.5 (± 0.5) 177.7 (± 2.0) 1.56 (± 0.02) 0.25 0.5 2.5 2.5 PVA 4-88 15.9 (± 5.4) 68.4 (± 2.5) 130.4 (± 12.0) 1.67 (± 0.10) 0.02 5 6.3 7 Maltodextrin 2 28.9 (± 2.3) 95.3 (± 1.2) 193.3 (± 5.7) 1.72 (± 0.06) 0.18 9.9 9.9 9.9 Maltodextrin 6 27.6 (± 3.5) 123.8 (± 2.4) 270.0 (± 8.7) 1.96 (± 0.05) 0.24 9.7 10 10 Maltodextrin DSH 15.7 (± 1.2) 122.5 (± 1.3) 246.1 (± 5.4) 1.88 (± 0.05) 0.23 9.7 10 10 Starch octenyl succinate 14.7 (± 0.2) 71.31 (± 0.5) 140.7 (± 2.8) 1.77 (± 0.03) 0.19 5 6.3 7.7 HP pea starch 105.8 (± 1.2) 216.9 (± 2.9) 374.9 (± 4.8) 1.24 (± 0.01) 0.05 8.7 10 10
Descriptive bindercharacteristics (units)
Wettability (°) Surface tension (mN/m) Dynamic viscosity (mPa.s) Viscosity slope Abbreviation in PCA CAbinder_t0 CAbinder_(t0-t30) CADCP CAPTFE ST DynamicViscosity ViscositySlope Binder PVP K12 NA* NA* 109.6 (± 3.1) 96.4 (± 1.5) 50.7 (± 0.5) 1.58 (± 0.01) 0.04 PVP K30 66.9 (± 3.2) 29.6 (± 3.3) 81.3 (± 2.9) 101.9 (± 0.9) 43.3 (± 0.2) 4.27 (± 0.14) 0.064 PVP K90 76.7 (± 2.9) 20.7 (± 1.9) 84.7 (± 1.3) 98.2 (± 1.5) 44.7 (± 0.4) 99.48 (± 1.55) 0.175 HPMC E5 82.5 (± 2.8) 9.9 (± 1.2) 90.8 (± 1.9) 83.8 (± 1.2) 46.9 (± 0.1) 52.85 (± 1.82) 0.178 HPMC E15 79.1 (± 1.7) 8.9 (± 1.1) 69.6 (± 4.4) 82.1 (± 1.1) 49.0 (± 0.2) 817.10 (± 2.75) 0.3 PVA 4-88 67.5 (± 1.3) 19.4 (± 1.2) 36.8 (± 2.5) 82.9 (± 0.4) 43.2 (± 0.2) 13.08 (± 0.16) 0.119 Maltodextrin 2 57.8 (± 3.8) 8.5 (± 3.4) 113.8 (± 4.9) 100.1 (± 1.3) 70.0 (± 0.1) 2.88 (± 0.02) 0.071 Maltodextrin 6 60.0 (± 3.9) 12.2 (± 2.8) 103 (± 5.3) 95.4 (± 1.0) 66.8 (± 0.2) 2.07 (± 0.07) 0.053 Maltodextrin DSH 56.2 (± 4.1) 15.7 (± 2.5) 101.1 (± 4.0) 99.2 (± 1.5) 68.4 (± 0.6) 1.54 (± 0.03) 0.037 Starch octenyl succinate 69.3 (± 3.1) 23.2 (± 2.6) 32.8 (± 3.9) 78.0 (± 0.6) 40.2 (± 0.1) 2.85 (± 0.10) 0.091 HP pea starch 67.0 (± 3.8) 9.6 (± 3.3) 95.3 (± 0.5) 94.9 (± 0.8) 57.7 (± 0.5) 15.67 (± 0.45) 0.11
CA on these surfaces showed a good wettability of dispersive surfaces by PVA and starch octenyl succinate solutions. Sufficient powder wettability is essential as wetting is the first step in a wet granulation process, spreading the granulation liquid over the powder bed.(Iveson et al., 2001) Furthermore, the low surface tension of PVA and starch octenyl succinate solutions indicated a low energy barrier between li-quid and air, facilitating the spreading of the droplet over the powder. (Cantor et al., 2008) As the particle size descriptors were located in the loading scatter plot opposite to the wetting kinetics (Fig. 3), the particle size only affected the wetting kinetics: a smaller binder particle size correlated with a faster binder wetting. In conclusion, the high binder effectiveness of PVA and starch octenyl succinate at low L/S-ratios was attributed to the good wetting and spreading properties of these binders on DCP, resulting in strong bonding formation.
Binders which were active at intermediate L/S-ratios (0.155 – 0.180) clustered in the lower left quadrant of the score plot: PVP K90 and HPMC E15/E5 (Fig. 4). These binders possessed high values for the viscosity parameters and initial binder wetting, and low values for the binder dissolution kinetics, the CA measured on DCP and PTFE, and the surface tension. In contrast, binders only active at higher L/S-ratios (PVP K12, maltodextrin 2/6/DSH) had opposite values for these ma-terial properties as they clustered on the right side of the score plot. The higher binder effectiveness of PVP K90 and HPMC E5/E15 was attrib-uted to the slow dissolution kinetics, high viscosity, good wetting properties of DCP and PTFE surfaces, and low surface tension. Upon wetting of these binders, a viscous gel layer was formed at the surface of the particles which probably increased the stickiness of the binder
surface. Hence, the consistency of the powder bed changed, resulting in a higher cohesiveness and more frictional resistance of the material to flow. The highest torque values were observed for PVP K90 and HPMC E15, supporting this hypothesis: 8 – 9 Nm compared to the other bin-ders which resulted in torque values of 3 – 6 Nm. According to Dhenge et al., the material residence time inside the TSG prolonged when a highly viscous binder solution was used (Dhenge et al., 2012a), im-proving binder distribution in the granulator barrel due to a longer mixing time. The longer residence time also caused a more intensive compaction and consolidation of the granules, allowing the binder to form more bridges with the powder particles. Furthermore, when hy-drated, these binders also showed good wetting of dispersive surfaces (negative correlation with CADCP and CAPTFE) and good spreading properties over the powder bed (negative correlation with ST). In ad-dition, different grades of a similar binder type (e.g. PVP K12 and K90) were found in different clusters. It is therefore important to consider the binder grade in formulation development.
Wettability attributes dominated the cluster of binders resulting in the highest binder effectiveness at low L/S-ratios. Nevertheless, a slightly lower L/S-ratio was needed for PVA to achieve the 30% granule friability limit in comparison to starch octenyl succinate (0.1175 versus 0.1300). PVA 4–88 also resulted in lower granule friability when comparing to starch octenyl succinate at their highest processable L/S-ratio. InFig. 6, binder attributes of PVA and starch octenyl succinate are compared by evaluating the ratio of the numerical values of PVA 4–88 and starch octenyl succinate properties. Compared to starch octenyl Fig. 3. Loading scatter plot of PC 1 and PC 2.
Fig. 4. The PC 1 versus PC 2 score scatter plot. The different colors refer to the L/S-ratio which was required to obtain granules with a friability ≤ 30%. The arrow indicates a flux towards the most optimal binder region. The binders clustered based on binder effectiveness are encircled.
Table 4
Overview of the binders with the L/S-ratio which was required to obtain a granule friability ≤ 30%, color on the PCA score plot and granule friability at this specific L/S-ratio.
Binder L/S-ratio PCA Color Code Friability (%)
PVA 4-88 0.1175 Light green 16.7 ( ± 0.2)
Starch octenyl succinate 0.1300 Dark green 23.5 ( ± 0.5)
PVP K90 0.1550 Light blue 17.5 ( ± 1.6)
HPMC E15 0.1800 Dark blue 7.4 ( ± 0.7)
HPMC E5 0.1800 Dark blue 17.2 ( ± 1.1)
HP pea starch 0.1800 Dark blue 22.1 ( ± 1.4)
PVP K30 0.1800 Dark blue 28.6 ( ± 0.6) Maltodextrin 2 0.2050 Red 21.5 ( ± 1.0) PVP K12 0.2050 Red 23.7 ( ± 2.0) Maltodextrin 6 0.2300 Purple 28.9 ( ± 0.9) Maltodextrin DSH 0.2300 Purple 26.2 ( ± 1.1) CADCP CAPTFE ST CAbinder_(t0-t30) Dv_10 Dv_50 Dv_90 D_Span DynamicViscosity ViscositySlope
DissRate_30 DissRate_60 DissRate_90 CAbinder_t0
SSA -4 -3 -2 -1 0 1 2
Starch octenyl succinate PVA 4-88 Sc or e con tr ibu tion (b ind er - ave ra ge )
Fig. 5. Score contribution plot of starch octenyl succinate (red) and PVA 4–88 (green).
succinate, PVA was characterized by a higher dynamic viscosity, re-sulting in slightly higher stickiness of the binder surface. According to Dhenge et al., this also increased the residence time and mixing effi-ciency of the material in the granulator barrel (Dhenge et al., 2012a), resulting in a slightly lower liquid requirement compared to starch octenyl succinate. Both binders resulted in a limited processing window of applicable L/S-ratios when granulated with DCP.
Furthermore, binders requiring higher L/S-ratios (e.g. HPMC E15, PVP K90 and HP pea starch) resulted in lower granule friability com-pared to starch octenyl succinate at its highest processing L/S-ratio. The drying capacity of the semi-continuous fluid bed dryer is limited, as the drying time is maximally six times the filling time of one single cell. Therefore, efficient granulation at lower L/S-ratios might be beneficial to obtain sufficiently dry granules at the end of a drying cycle. The presence of a limited amount of liquid might also be beneficial to result in a faster drying process. Nevertheless, the granule friability needs to be adequately to allow downstream processing.
3.2. Wet binder addition 3.2.1. Granule friability
For the wet addition of a binder, a solution with a specific binder concentration was prepared prior to granulation. However, a high viscosity of the solution could limit its processability. In order to allow comparison of the performance of different binders, granulation ex-periments were performed at the highest concentration which was achievable for each binder (i.e. 1.44% w/w in the dried granule). Additionally, granules with a binder concentration of 5% in the dry granule were produced for low-viscosity binders.
A high friability of all granules was observed at 1.44% binder concentration. Granules containing PVA were clearly less friable than other binders at low L/S-ratio. At an L/S-ratio of 0.28, HPMC E15 yielded the lowest granule friability (37.5%), followed by PVP K90 (67.2%) and HPMC E5 (70.7%) (Fig. 7a). For each binder, a higher binder content (5%) reduced the granule friability (Fig. 7b). A binder content of 5% in the dry granules could not be achieved for HPMC E15/ E5, PVP K90 and PVA due to the high viscosity of the binder solutions needed during wet granulation. At a 5% binder concentration in com-bination with an L/S-ratio of 0.13, starch octenyl succinate yielded a granule friability < 30%. HP pea starch resulted in low granule fria-bility at an L/S-ratio of 0.18. At high L/S-ratio (0.28), 5% binder could only be processed with PVP K30 and K12, which resulted in a very low granule friability for PVP K30 (6.9%).
3.2.2. Link between binder attributes and granule friability
As the binders were added as a solution, only the characteristics related to the hydrated state of the polymers in an aqueous environ-ment (i.e. wettability, surface tension and viscosity) were linked to the
granule friability. As this limited the dataset, no multivariate analysis was performed.
The friability of granules containing 1.44% binder was compared to evaluate the performance of the different binder types.Fig. 7c, illus-trating the granule friability at an L/S-ratio of 0.28 in function of the viscosity measured at 8% w/w binder solution, reflects an inverse correlation between granule friability and binder viscosity from 20 mPa.s. Friability of granules produced with binders containing a solution viscosity below 20 mPa.s was high (varying between 77.1 and 94.0%) and was not affected by the binder viscosity. Dhenge et al. described that a higher viscosity of the granulation liquid enhanced the stickiness of the wet powder mass when the liquid is squeezed to the granule surface during consolidation of wet granules. This allowed more particles to bind, resulting in a higher number of viscous bonds. The authors supported this observation with X-ray tomography images of granules showing an internal granule structure with less voids, yielding a higher granule strength.(Dhenge et al., 2012a) A higher stickiness was also observed by Chitu et al. when processing highly viscous binder solutions.(Chitu et al., 2011) The activity at lower L/S-ratio of PVA (and of starch octenyl succinate at 5% binder concentra-tion) was attributed to the low CA of the binder solution measured on a DCP surface (Fig. 7d). As previously mentioned, wettability is essential as it affects the granulation process.(Iveson et al., 2001) Higher surface tension values were measured for maltodextrin 2, 6 and DSH when compared to other binders (66 – 70 N/m versus 40 – 50 N/m).Fig. 7e shows the granule friability in function of the surface tension measured of the binder solutions. Starch octenyl succinate (40.2 N/m) and PVA (43.2 N/m) are not included, as these binders could not been processed at an L/S-ratio of 0.28 due to overwetting during the granulation pro-cess. Binder surface tension and granule friability showed no linear correlation. It can be concluded that the intrinsic binding capacity was higher for the highly viscous binders. However, a higher binder content in the granules could be achieved for lower viscosity binders, yielding granules with a lower friability.
The fast wetting of a binder and its ability to sufficiently wet the powder bed was crucial for high binder effectiveness at low L/S-ratios. Wettability is considered a key attribute in the wet granulation process as it affects the uniform liquid distribution in the powder bed. Moreover, the ability of the binder to sufficiently wet the powder bed is determinative for the formation of liquid bridges which influences the properties of the final granules and tablets.(Iveson et al., 2001; Jaiyeoba and Spring, 1980) Binders which sufficiently wet the powder bed required less granulation liquid to obtain low granule friability. Additionally, a high binder viscosity, low surface tension, and slow dissolution kinetics also positively influenced the binder effectiveness. As binder effectiveness cannot be attributed to only one binder char-acteristic, a multivariate approach to select appropriate binders during early stages of formulation development is recommended.
3.3. Effect of binder addition method
The effect of the binder addition method on granule friability was examined. For most binders, at lower L/S-ratio, wet binder addition yielded a lower granule friability compared to dry binder addition (Fig. 8a and 8b). As the binder was already dissolved for the wet method, it was fully hydrated, allowing its maximal efficiency. In contrast, for dry binder addition the binder still needed to interact with the granulation liquid in order to be activated. Nevertheless, at higher L/S-ratio, granules showed similar friability when processed with both addition methods as the higher amount of liquid available in the granulator barrel allowed to use the full potential of the dry binder.
Remarkably, dry binder addition resulted in lower granule friability for specific binders: HPMC E15, HPMC E5, and PVP K90 (Fig. 8c-e). According to Mort and Tardos, the degree of binder distribution throughout the powder bed was reflected by the granule size distribu-tion in fluidized bed granuladistribu-tion.(Mort and Tardos, 1999) The size
D issRa te_3 0 D issRa te_6 0 D issRa te_9 0 Dv_1 0 Dv_5 0 Dv_9 0 D_Spa n CAbinde r_ t0 CA bi nd er _(t 0-t 30 ) DynamicV iscos ity Vi scos ityS lop e ST CAPFT E CADC P 0 1 2 3 4 5 PV A 4-88 / st ar ch oc te ny l s ucc in at e
Fig. 6. Binder attributes in function of the ratio of the numerical values of PVA 4–88 and starch octenyl succinate properties.
distributions of granules prepared with the latter binders were therefore examined at different L/S-ratios for both binder addition methods. Fig. 9illustrates the size distribution of granules produced with HPMC E5 and E15 at an L/S-ratio of 0.1800 – 0.2050 and 0.1800 – 0.2175, respectively. At an L/S-ratio of 0.18 (Fig. 9a and 9c), both addition methods yielded a bimodal size distribution with a relatively high fines fraction (< 150 µm), which decreased or stayed similar at higher L/S-ratio. A strong decrease of the fines fraction was obtained with the dry addition method for both binders. Moreover, this resulted in a more unimodal granule size distribution whereas the distribution obtained by the wet binder addition remained bimodal (Fig. 9b and 9d). This was also reflected in the granule friability, because lower granule friability was obtained with the dry addition method compared to the wet ad-dition for HPMC E5 (71.3 versus 95.7%) and HPMC E15 (47.4 versus 82.7%). As a higher degree of bimodality was observed for the size distribution of granules produced with these specific wet added binders, it can be assumed that no even binder distribution was obtained. This might be attributed to the high viscosity of the binder solutions, as this can limit the spreading capacity of the granulation liquid throughout
the powder bed. Moreover, slightly higher torque values were obtained for the dry binder addition method. This also assumed a better binder distribution of the dry added binders, as the torque reflects the amount of energy introduced in the system during granulation.
From an industrial point of view, dry addition of binders is preferred because of its ease of manufacturing. However, stickiness and cohe-siveness of the dry binder can hamper accurate feeding. As the dry addition of binders resulted in similar granule quality at slightly higher L/S-ratios or even in better granule quality compared to wet binder addition, the efficiency of this processing strategy during continuous wet granulation was evident.
In current study, binder properties were correlated to friability of granules consisting of a poorly soluble excipient. Consequently, all granulation liquid was available for the activation of the (dry) binders. (Rowe et al., 2009) The high density and the poor aqueous solubility of DCP required a high filling degree of the granulator barrel to obtain sufficient densification. Moreover, the use of a highly viscous binder resulted in a higher degree of densification, as material residence time was assumed to be longer.(Dhenge et al., 2012a) This contributed to
0.15 0.20 0.25 0.30 20 40 60 80 100
(a) 1.44% of binder in dry granule
L/S-ratio Fr ia bili ty (% ) 0.0 0.1 0.2 0.3 0 20 40 60 80
100 (b) 5% of binder in dry granule
L/S-ratio Fr ia bili ty (% ) 0 20 40 60 80 100 0 20 40 60 80 100 500 1000 (c) L/S-ratio of 0.28 Viscosity (mPa.s) Fr ia bili ty (% ) PVP K12PVP K30PVP K90HPM C E5 HPM C E1 5 PVA 4 -88 Maltod extrin 2 Maltod extrin 6 Starc h octe nyl su ccinat e HP p ea sta rch Maltod extrin DS H 0 50 100 150 (d) Contact angle C on tac t a ng le (° ) 40 50 60 70 0 20 40 60 80 100 (e) L/S-ratio of 0.28 Surface tension (N/m) Fr ia bili ty (% )
Fig. 7. Granule friability of binders in function of L/S-ratio at (a) 1.44% binder concentration and at (b) 5% binder concentration. Granule friability in function of the viscosity (8% w/w) of binders is illustrated in (c). The contact angle measured on DCP surface for each binder is illustrated in (d). Granule friability in function of the surface tension measured of the binder solutions (8% w/w) is shown in (e).
0.10 0.11 0.12 0.13 0 20 40 60 80 100 (a) PVA - 3% L/S-ratio Friability (%) 0.18 0.20 0.22 0 20 40 60 80 100 (b) HP pea starch - 5% L/S-ratio Friability (%) 0.18 0.20 0.22 0 20 40 60 80 100 Wet addition Dry addition (c) HPMC E15 - 1.44% L/S-ratio Friability (%) 0.18 0.20 0.22 0 20 40 60 80 100 (d) HPMC E5 - 1.44% L/S-ratio Friability (%) 0.18 0.20 0.22 0 20 40 60 80 100 (e) PVP K90 - 3 % L/S-ratio Friability (%)
Fig. 8. Granule friability of (a) PVA – 3%, (b) HP pea starch – 5%, (c) HPMC E15 – 1.44%, (d) HPMC E5 – 1.44%, and (e) PVP K90 – 3% in function of L/S-ratio determined for wet and dry binder addition.
(a) L/S-ratio of 0.1800 (HPMC E5)
0 1000 2000 3000 4000 0 10 20 30 μm dQ 3 /%
(b) L/S-ratio of 0.2050 (HPMC E5)
0 1000 2000 3000 4000 0 10 20 30 Wet Addition Dry Addition μm dQ 3 /%(c) L/S-ratio of 0.1800 (HPMC E15)
0 1000 2000 3000 4000 0 10 20 30 μm dQ 3 /%(d) L/S-ratio of 0.2175 (HPMC E15)
0 1000 2000 3000 4000 0 10 20 30 μm dQ 3 /%more efficient granulation of the poorly soluble formulation studied. However, the effect of different binder types on a highly soluble for-mulation is also of interest as this forfor-mulation would also interact with the granulation liquid, resulting in less liquid available for binder ac-tivation. Next to the formulation aspect, the possibility of downstream processing should be studied. As wet granulation requires drying of granules, breakage and attrition during drying could be influenced by the binder type. Furthermore, the effect of different binders in function of the tableting process is of interest. This will allow to take the influ-ence of the binder on the final tablet properties into account, and thus, make a well-considered binder choice when composing a formulation for processing via TSG.
4. Conclusion
Current study investigated the correlation between binder attributes and binder effectiveness in a continuous wet granulation process. Binders were added via the dry and wet addition method with DCP as model excipient. Different binder types required different amounts of liquid in order to produce good quality granules. PVA and starch oc-tenyl succinate resulted in granule friability below 30% at low L/S-ratios in comparison to the other included binders. This higher binder effectiveness at low L/S-ratios correlated with good wetting properties of the powder bed. The performance of wettability measurements in an early formulation development stage is therefore considered highly important. Additionally, an increased stickiness of the binder surface caused by a high binder viscosity, slow dissolution kinetics, and a low surface tension also positively influenced the binder effectiveness. The efficient applicability of dry binder addition in continuous processes was shown. In conclusion, this research provided an initial insight in how binder attributes impacted binder effectiveness based on a selec-tion of binders in a continuous process with use of dicalcium phosphate as poorly soluble model excipient. Optimal binder selection should therefore focus on critical binder attributes which influence the gran-ulation process.
CRediT authorship contribution statement
L. Vandevivere: Conceptualization, Methodology, Formal analysis, Investigation, Writing - original draft, Writing - review & editing, Visualization, Project administration. P. Denduyver: Investigation. C.
Portier: Conceptualization, Methodology. O. Häusler:
Conceptualization, Resources, Writing - review & editing, Supervision, Project administration, Funding acquisition. T. De Beer: Methodology. C. Vervaet: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration, Funding acquisition. V. Vanhoorne: Conceptualization, Methodology, Writing - review & editing, Supervision, Project administration.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influ-ence the work reported in this paper.
Acknowledgements
The authors would like to thank Roquette Frères for funding of the study. Furthermore, Roquette Frères, BASF, Dow Chemical Company and Merck are acknowledged for donation of the binders. Also a note of thanks to Ms Barbora Blahová Prudilová (Regional Centre of Advanced Technologies and Materials, Department of Physical Chemistry, Palacky University, Czech Republic) for the performance of the iGC measure-ments. This work was supported by the INTERREG V 2 Mers Seas Zeeën
Crossborder Cooperation Programme 2014–2020 (Project number 2S01-059 – project acronym IMODE).
References
Becker, D., Rigassi, T., Bauer-Brandi, A., 1997. Effectiveness of binders in wet granula-tion: a comparison using model formulations of different tabletability. Drug
devel-ompent Ind. Pharm. 23 (8), 791–808.
Cantor, S.L., Augsburger, L.L., Hoag, S.W., Gerhardt, A., 2008. Pharmaceutical granula-tion processes, mechanism, and the use of binders, in: Pharmaceutical Dosage Forms: Tablets, Third Edition: Volume 1: Unit Operations and Mechanical Properties. pp. 261–302. https://doi.org/10.3109/9781420020281-9.
Chitu, T.M., Oulahna, D., Hemati, M., 2011. Wet granulation in laboratory scale high shear mixers: Effect of binder properties. Powder Technol. 206, 25–33.https://doi.
org/10.1016/j.powtec.2010.07.012.
De Leersnyder, F., Vanhoorne, V., Bekaert, H., Vercruysse, J., Ghijs, M., Bostijn, N., Verstraeten, M., Cappuyns, P., Van Assche, I., Vander Heyden, Y., Ziemons, E., Remon, J.P., Nopens, I., Vervaet, C., De Beer, T., 2018. Breakage and drying beha-viour of granules in a continuous fluid bed dryer: Influence of process parameters and wet granule transfer. Eur. J. Pharm. Sci. 115, 223–232.https://doi.org/10.1016/j.
ejps.2018.01.037.
Della Volpe, C., Siboni, S., 1997. Some reflections on acid-base solid surface free energy theories. J. Colloid Interface Sci. 195, 121–136.https://doi.org/10.1006/jcis.1997.
5124.
Dhenge, R.M., Cartwright, J.J., Hounslow, M.J., Salman, A.D., 2012a. Twin screw wet granulation: Effects of properties of granulation liquid. Powder Technol. 229, 126–136.https://doi.org/10.1016/j.powtec.2012.06.019.
Dhenge, R.M., Cartwright, J.J., Hounslow, M.J., Salman, A.D., 2012b. Twin screw gran-ulation: Steps in granule growth. Int. J. Pharm. 438, 20–32.https://doi.org/10.1016/
j.ijpharm.2012.08.049.
Djuric, D., Kleinebudde, P., 2010. Continuous granulation with a twin-screw extruder: Impact of material throughput. Pharm. Dev. Technol. 15, 518–525.https://doi.org/
10.3109/10837450903397578.
Dorris, G.M., Gray, D.G., 1980. Adsorption of n-alkanes at zero surface coverage on cel-lulose paper and wood fibers. J. Colloid Interface Sci. 77, 353–362.https://doi.org/
10.1016/0021-9797(80)90304-5.
El Hagrasy, A.S., Hennenkamp, J.R., Burke, M.D., Cartwright, J.J., Litster, J.D., 2013. Twin screw wet granulation: Influence of formulation parameters on granule prop-erties and growth behavior. Powder Technol. 238, 108–115.https://doi.org/10.
1016/j.powtec.2012.04.035.
Fonteyne, M., Fussell, A.L., Vercruysse, J., Vervaet, C., Remon, J.P., Strachan, C., Rades, T., De Beer, T., 2014. Distribution of binder in granules produced by means of twin screw granulation. Int. J. Pharm. 462, 8–10.https://doi.org/10.1016/j.ijpharm.
2013.12.008.
Iveson, S.M., Litster, J.D., Hapgood, K., Ennis, B.J., 2001. Nucleation, growth and breakage phenomena in agitated wet granulation processes: A review. Powder Technol. 117, 3–39.https://doi.org/10.1016/S0032-5910(01)00313-8. Jachowicz, R., 1987. Dissolution rates of partially water-soluble drugs from solid
dis-persion systems. I. Prednisolone. Int. J. Pharm. 35, 1–5.https://doi.org/10.1016/
0378-5173(87)90067-6.
Jaiyeoba, K.T., Spring, M.S., 1980. The granulation of ternary mixtures: the effect of the wettability of the powders. J. Pharm. Pharmacol. 32, 386–388.https://doi.org/10.
1111/j.2042-7158.1980.tb12948.x.
Keleb, E.I., Vermeire, A., Vervaet, C., Remon, J.P., 2002. Continuous twin screw extrusion for the wet granulation of lactose. Int. J. Pharm. 239, 69–80.https://doi.org/10.
1016/S0378-5173(02)00052-2.
Kumar, A., Vercruysse, J., Toiviainen, M., Panouillot, P.E., Juuti, M., Vanhoorne, V., Vervaet, C., Remon, J.P., Gernaey, K.V., De Beer, T., Nopens, I., 2014. Mixing and transport during pharmaceutical twin-screw wet granulation: Experimental analysis via chemical imaging. Eur. J. Pharm. Biopharm. 87, 279–289.https://doi.org/10.
1016/j.ejpb.2014.04.004.
Masi, C.G., 1990. So, you want to measure submicron dimensions. Test Meas. world
10, 1–8.
Mort, P.R., Tardos, G.I., 1999. Scale-up of agglomeration processes using transformations. KONA Powder Part. J. 17, 64–75.https://doi.org/10.14356/kona.1999013. Parker, M.D., York, P., Rowe, R.C., 1991. Binder-substrate interactions in wet
granula-tion. 2: The effect of binder molecular weight. Int. J. Pharm. 72, 243–249.https://
doi.org/10.1016/0378-5173(91)90114-4.
Portier, C., Pandelaere, K., Delaet, U., Vigh, T., Di Pretoro, G., De Beer, T., Vervaet, C., Vanhoorne, V., 2020a. Continuous twin screw granulation: A complex interplay be-tween formulation properties, process settings and screw design. Int. J. Pharm. 576, 119004.https://doi.org/10.1016/j.ijpharm.2019.119004.
Portier, C., Pandelaere, K., Delaet, U., Vigh, T., Kumar, A., Di Pretoro, G., De Beer, T., Vervaet, C., Vanhoorne, V., 2020b. Continuous twin screw granulation: influence of process and formulation variables on granule quality attributes of model formula-tions. Int. J. Pharm. 576, 118981.https://doi.org/10.1016/j.ijpharm.2019.118981. Portier, C., Vigh, T., Di Pretoro, G., De Beer, T., Vervaet, C., Vanhoorne, V., 2020c.
Continuous twin screw granulation: Impact of binder addition method and surfac-tants on granulation of a high-dosed, poorly soluble API. Int. J. Pharm. 577, 119068.
https://doi.org/10.1016/j.ijpharm.2020.119068.
Rades, T., Knopp, M.M., Olesen, N.E., Holm, P.E.R., Langguth, P., Ren, E., 2015. Influence of Polymer Molecular Weight on Drug – Polymer Solubility : A Comparison between Experimentally Determined Solubility in PVP and Prediction Derived from Solubility
in Monomer 2905–2912. https://doi.org/10.1002/jps.24410.
Ritala, M., Jungersen, O., Holm, P., Schæfer, T., Kristensen, H.G., 1986. A comparison between binders in the wet phase of granulation in a high shear mixer. Drug Dev. Ind. Pharm. 12, 1685–1700.https://doi.org/10.3109/03639048609042603.
Rowe, R.C.R., Sheskey, P.J.S., Cook, W., 2009. Handbook Pharmaceutical Excipients, Sixth Edition 1064. https://doi.org/http://dx.doi.org/10.1016/B978-012088479-7/ 50039-0.
Van Melkebeke, B., Vervaet, C., Remon, J.P., 2008. Validation of a continuous granula-tion process using a twin-screw extruder. Int. J. Pharm. 356, 224–230.https://doi.
org/10.1016/j.ijpharm.2008.01.012.
Vandevivere, L., Portier, C., Vanhoorne, V., Häusler, O., Simon, D., De Beer, T., Vervaet, C., 2019. Native starch as in situ binder for continuous twin screw wet granulation. Int. J. Pharm. 571.https://doi.org/10.1016/j.ijpharm.2019.118760.
Vanhoorne, V., Bekaert, B., Peeters, E., De Beer, T., Remon, J.P., Vervaet, C., 2016a. Improved tabletability after a polymorphic transition of delta-mannitol during twin screw granulation. Int. J. Pharm. 506, 13–24.https://doi.org/10.1016/j.ijpharm.
2016.04.025.
Vanhoorne, V., Janssens, L., Vercruysse, J., De Beer, T., Remon, J.P., Vervaet, C., 2016b. Continuous twin screw granulation of controlled release formulations with various HPMC grades. Int. J. Pharm. 511, 1048–1057.https://doi.org/10.1016/j.ijpharm.
2016.08.020.
Vercruysse, J., Córdoba Díaz, D., Peeters, E., Fonteyne, M., Delaet, U., Van Assche, I., De Beer, T., Remon, J.P., Vervaet, C., 2012. Continuous twin screw granulation:
Influence of process variables on granule and tablet quality. Eur. J. Pharm. Biopharm. 82, 205–211.https://doi.org/10.1016/j.ejpb.2012.05.010.
Vercruysse, J., Toiviainen, M., Fonteyne, M., Helkimo, N., Ketolainen, J., Juuti, M., Delaet, U., Van Assche, I., Remon, J.P., Vervaet, C., De Beer, T., 2014. Visualization and understanding of the granulation liquid mixing and distribution during con-tinuous twin screw granulation using NIR chemical imaging. Eur. J. Pharm. Biopharm. 86, 383–392.https://doi.org/10.1016/j.ejpb.2013.10.012. Vervaet, C., Vermeersch, H., Khotz, M.S., Massart, L., Remon, J.P., 1994. Parameters
influencing granule quality using a continuous granulator. Int. J. Pharm. 106, 157–160.https://doi.org/10.1016/0378-5173(94)90314-X.
Willecke, N., Szepes, A., Wunderlich, M., Remon, J.P., Vervaet, C., De Beer, T., 2017. Identifying overarching excipient properties towards an in-depth understanding of process and product performance for continuous twin-screw wet granulation. Int. J. Pharm. 522, 234–247.https://doi.org/10.1016/j.ijpharm.2017.02.028.
Willecke, N., Vervaet, C., De Beer, T., Szepes, A., Wunderlich, M., 2017b. Formulation design for continuous twin screw wet granulation.
Winkler, T., 2010. Determining the surface tension of liquids by measurements on
pen-dant drops. Tech. Note 1–5.
Zhang, D., Flory, J.H., Panmai, S., Batra, U., Kaufman, M.J., 2002. Wettability of phar-maceutical solids: Its measurement and influence on wet granulation. Colloids Surfaces A Physicochem. Eng. Asp. 206, 547–554.