You are on page 1of 10

PHASCI-03397; No of Pages 10

European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage:

Roller compaction scale-up using roll width as scale factor and laser-based determined
ribbon porosity as critical material attribute
Morten Allesø a,b,⁎, René Holm a, Per Holm a
H. Lundbeck A/S, Biologics and Pharmaceutical Science, Ottiliavej 9, DK-2500, Valby, Denmark
NNE Pharmaplan A/S, Process Technology Consulting, Nybrovej 80, DK-2820 Gentofte, Denmark

a r t i c l e i n f o a b s t r a c t

Article history: Due to the complexity and difficulties associated with the mechanistic modeling of roller compaction process for
Received 31 August 2015 scale-up, an innovative equipment approach is to keep roll diameter fixed between scales and instead vary the
Accepted 1 November 2015 roll width. Assuming a fixed gap and roll force, this approach should create similar conditions for the nip regions
Available online xxxx
of the two compactor scales, and thus result in a scale-reproducible ribbon porosity. In the present work a non-
destructive laser-based technique was used to measure the ribbon porosity at-line with high precision and high
Roller compaction
accuracy as confirmed by an initial comparison to a well-established volume displacement oil intrusion method.
Ribbon porosity The ribbon porosity was found to be scale-independent when comparing the average porosity of a group of rib-
Solid fraction bon samples (n = 12) from small-scale (Mini-Pactor®) to large-scale (Macro-Pactor®). A higher standard devi-
Volume displacement ation of ribbons fragment porosities from the large-scale roller compactor was attributed to minor variations in
Direct laser measurement powder densification across the roll width.
Microcrystalline cellulose With the intention to reproduce ribbon porosity from one scale to the other, process settings of roll force and gap
size applied to the Mini-Pactor® (and identified during formulation development) were therefore directly trans-
ferrable to subsequent commercial scale production on the Macro-Pactor®. This creates a better link between for-
mulation development and tech transfer and decreases the number of batches needed to establish the parameter
settings of the commercial process.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction process parameters as well as equipment design alterations; the latter
typically realistic only with new equipment as a part of its installation
Roller compaction for dry granulation has gained increasing interest and qualification. Proper understanding of scale-up parameters stems
in the pharmaceutical industry over the past decades. Improvements in from mechanistic insight into the given process technology and its var-
the technology such as better control over the gap, leading to more uni- ious scales. The aim of a successful scale up is therefore to create similar
form ribbon and granule properties, and improved roller side sealing process conditions for the powder material across scales through the
and thus containment of fines, are key reasons for its increased popular- adjustment of process parameters. For roller compaction, assuming a
ity (Kleinebudde, 2004; Miller, 2005). fixed granulator sieve size between pilot and commercial scale, ribbon
Roller compaction is a continuous unit operation, meaning that a pri- porosity is the most important quality descriptor of the granules' com-
mary batch may be split into several sub-batches and processed at var- pression properties and size distribution, as discussed by Allesø et al.
ious combinations of process parameters. This approach carries the (2013). The proper scale-factor for compaction pressure may be found
potential of efficient formulation development as well as flexibility to through empirical testing at the larger scale or through the application
shift between various scales by either adjusting press roller speed of powder mechanics theory (Johanson, 1965) combined with experi-
and/or gap size; however, only when their impact on granule quality mental data (Reynolds et al., 2010; Rowe et al., 2013). Mechanics theory
is fully understood. can be applied to determine the nip angle/in the compaction zone when
Scale-up from lab/pilot to commercial scale can pose challenges that switching to larger roll diameters. In roller compaction, the nip angle in-
are difficult to foresee as well as mitigate; not least because these issues dicates the compaction zone from start to end of densification, which
occur late in development where the formulation composition is often therefore dictates the total pressure applied to the powder (at a given
fixed. This restricts scale-up control strategies to the adjustment of roll force) and ultimately impacts ribbon porosity. There are studies
demonstrating the use of instrumented roller compactors to acquire
⁎ Corresponding author at: NNE Pharmaplan A/S, Nybrovej 80, DK-2820 Gentofte,
the pressure profile of the powder material in the nip region
Denmark. (Bindhumadhavan et al., 2005; Miguélez-Moran et al., 2008; Nesarikar
E-mail address: (M. Allesø). et al., 2012) as well as a recent example of roller compaction modeling
0928-0987/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: Allesø, M., et al., Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity
as critical material attrib..., European Journal of Pharmaceutical Sciences (2015),

where the volume binations of gap and roll force where investigated. An obvious drawback of this technique is the intention to sample an entire 100 mm ribbon from the Macro-Pactor® large sample size required for accurate porosity determination.0 mm and Environmentally friendly alternatives are intrusion with low viscosity roll force of 7. Four com- mensional method is perhaps the most simple of all. The study included a center point with a gap of 3. one may claim ribbon porosity to be scale inde.. The large number of ribbons needed for scale-up inves. Roller compaction techniques have also been described in the literature. 2010). Jona. i.e. the two most notable differences being 2. Khorasani et al.0 kN/cm.. Germany). 2015) or a fine-particulate ment techniques (laser and oil.0 kN/cm. A scale-comparable front-view of the two com. Gap size and roll force were varied according to The use of a laser-based direct volume measurement device is a re. Medium chain triglyceride a tablet.0 and since fragile and small ribbons. Germany). These methods include near-infrared stearate in a Bohle LM40 bin blender of suitable size (L.. Allesø et al. Every ribbon was first analyzed by the Nkansah et al.2. European Journal of Pharmaceutical Sciences (2015). Sample collection (see below) was performed once the roller compactor had reached its target set- 1 Porosity = (1 − solid fraction) ∗ 100% ting for gap and roll force. The fact that rollers have identical diameters between the two scales ensures an identical nip area and thus Microcrystalline cellulose (Avicel PH102) and croscarmellose sodium a similar impact of roll pressure and gap size on the conveyed powder (Ac-Di-Sol) were sourced from FMC Health and Nutrition (Philadelphia. 2012. however. using punch draw. the volume of the tablet can be calculated at a specific tablet thickness. in a certain process time frame and weighed. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib. which measures the effective refractive index of on its larger scale equivalent. USA) and used as filler/binder and superdisintegrant. Another aim of the present study and the Macro-Pactor®.0 and 4. Ribbons on the other hand are irregular in shape and a frequently used method to determine All samples consisted of microcrystalline cellulose (diluent) and the volume of the inter-particular voids is a volume displacement meth. pendent for these particular roller compactor types. oil intrusion method. Auger speed was et al.. expansion. Dry granulation was carried out on the small- et al. when roller compaction scale-up was done by means of a roll machine although the continuous nature of this unit operation makes width increase from 25 mm to 100 mm on Gerteis® compactors. 2013. M.. its major drawback being the toxicity of the mercury.0 rpm and the feed/tamp auger ratio was 250%.B. material. Switzerland) equipped with 25 mm wide rolls as well as has been described. A two-factor. respectively.KG. two level design was carried od where a medium of known density is absorbed into the pores of the out at each roller compactor equipment scale. automatically adjusted by the gap control feature in order to ensure tigations. has a very limited use when they get too fragile. As an alternative to first-principle understanding.75% magnesium stearate. though they do not directly measure the density of the compact but rather a property The microcrystalline cellulose was mixed with magnesium correlated with the density. The investigated settings were 2. Ribbons are collected in real-time with. For the comparison of the two porosity measure- oil (Allesø et al. Although the technique is commercially pressure on a larger scale compactor. The Mini-Pactor® is the R&D and pilot-scale rosity.. the sample material produced in that time frame is calculated from the process population was increased from six to n = 12. 6 × 5 × 2 = 60 ribbons in total. The Macro-Pactor® is Gerteis®' production scale compactor.1016/j. the Macro-Pactor®. Please cite this article as: Allesø. Materials and methods four times wider press rollers (25 mm vs. Furthermore the di- mensional approach does not take into account thickness variations of ribbons manufactured using knurled or serrated rollers. well-established volume et al.1.doi. Experimental design are known. Recently terahertz (THz) spectroscopy AG.11. It is very similar to the Mini-Pactor®. avoids the use of any consumable sup- Joint-Y PLS regression and applied it to predict the scale factor for roll plies and is simple to utilize. (2008) have described a simple throughput approach non-destructive laser method and finally subjected to the destructive to solid fraction1 determination. 100 mm) and a feeding system with two tamp augers instead of one in addition to the single 2. Souihi et al. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx by computational FEM combined with experimental data (Cunningham higher speed than the previously described. Acevedo et al. knowledge there are no published studies demonstrating this. but this time at rela- of the ribbon is determined by cutting the ribbon to a rectangular shape tively high roll pressures. When the tablet mass and the density of the particles 2.b). et al. A tablet has well defined dimensions and. The laser technique provides a result in ap- (2011) modeled historic data from a small-scale roller compactor by proximately 2 min per ribbon.. displacement approaches. Roll speed was cently introduced technique for determination of ribbon porosity (Iyer 6. Germany) was added as lubricant to reduce friction between the powder Determination of ribbon porosity is not as trivial to determine as for and metal surfaces of the roller compactor. having a larger roll diameter. literature demonstrating its performance is still sparse. The di. where the impact of light scattering er scale Mini-Pactor® (Gerteis® Maschinen + Processengineering on the baseline is considered.001 . Khorasani Ennigerloh.2015. (see below) and assess the within-ribbon porosity variability.ejps. which often are the case when API is 4. available. was obtained from Delios V (Illertissen.. Liu et al. one aim of this study was to compare the laser-based porosity averted through innovative designing of the different equipment scales. Spectroscopic 2.. A long-established technique in this category is mercury at levels was to investigate the hypothesized scale independence of ribbon po- pactors is shown in Fig. the additional six ribbons parameters gap and roll speed as well as a correction for viscoelastic only being analyzed by the laser method. kN/cm. 1. There- Parts of the scale-up challenges in roller compaction may also be fore.. Bohle.. (NIR) spectroscopy (Lim et al. MCT). the porosity can easily be acquired. This method. method against a volume displacement method using low viscosity oil The flagship models of Gerteis® compactors include the Mini-Pactor® (medium chain triglyceride. see below)... equipped with the ribbon that can be correlated to the density (Sullivan et al. (www. PA. 100 mm wide rolls.0 mm and the factor. roll force at levels 3. The mass and volume (per time unit) divided provide the An additional four experiments were conducted separately with the at-gap ribbon density. 2015.0 porosimetry. since whole ribbons are difficult to sample with a trimming knife. ings. 2015).2 M. http://dx. with the factor gap size Materials feed auger that both machines have. call for a method that efficiently analyzes ribbon porosity at a a stable gap during processing. 0. 2014a.3. The volume of ribbon For comparison of roller compactor equipment scales. present in the formulation. for example. Bad Münstereifel.micromeretics. To the best of our Magnesium stearate (Peter Greven GmbH & Co. A stable gap was confirmed for all in- dividual experiments of this study. the design described in the previous section. Consequently. it suitable for production scale also depending on the processing time. 2015). a total of six ribbons were solid material such as the Dry Flo® medium used for the GeoPyc® analyzed per experiment (five in total) per equipment scale (two). 2011. cannot be cut properly.0 and 11.0 and 11.0 mm gap and 7.

are four times wider for the Macro- Pactor® as shown in the bottom of the picture. Please cite this article as: Allesø.2. M. laser-based system. Sample collection during roller compaction 2. et al. Ribbon porosity determination 2. however. it was decided where U and L are the distances measured by the Upper and Lower la- to configure the granulator with an open star rotor. The Laser-based technique The process was run without granulator to avoid clogging of the The porosity of the ribbons was measured using a laser-based direct space between the rolls and the granulator region and to minimize volume measurement device (Solid Fraction Measurement Systems. 2. wear on the screen mesh.4. Mini-Pactor® (small scale) 2. trusion method (see below). operating sers to the top and the bottom of the sample. most were al.4.11. and thus produced ribbons. of the sample can be computed as ρribbon = mribbon/Vribbon. and then integrating of suitable size for the following porosity measurements by the laser this thickness data.5. Since the distance between the two lasers is known. 2. to run to steady-state (i. European Journal of Pharmaceutical Sciences (2015). Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib. and D is it without any screen support. That way the granulator broke the large the known distance between the lasers. the process was allowed movement during scanning of a ribbon. In practice.. tested..doi. it is possible to sample collection port. Most of these fragments were number of points on the surface of the sample. one can determine the volume of the sample being and oil intrusion method (see below). CT.1. Given the tation with a trimming knife and porosity determination by the laser mass of the ribbon and the volume of the sample.. the thickness of the sample (T) at the laser beam is given by: 2. USA). Scale-comparable front view of Gerteis®’ A) Mini-Pactor® and B) Macro-Pactor® roller compactors. The approach uses two laser displacement sen- ments where hard ribbons (i. To achieve this.ejps. A 100 mm wide ribbon attached ferred (automatically) to a precision weighing module within the to the left-side roller was manually removed and kept for later fragmen. where its mass (mribbon) is determined. located underneath the rollers and the left-side compute the thickness of the object being measured at the point of mea- scraper. The various distances relevant to the thickness measurement ready of a size suitable for porosity measurement by the laser and oil in.001 . the density (ρribbon) method. respectively. Since representative sampling is best ensured through sampling in real-time. are illustrated in Fig.4. 100 mm ribbons into smaller ribbon fragments that could be sampled By making measurements of the thickness of the sample at a large in real-time from the granulator outlet. low porosity) were produced.1.5. Centerbrook.1016/j. Of the collected ribbons from all five batches. the XY direction controlling the sampling of an entire ribbon. 1. http://dx. the ribbon is trans- thereafter brought to immediate stop. This was particularly relevant in the experi. the ribbon is held by a set of jaws attached to a com- The four experiments in the intra-ribbon variability study required puter controlled precision XYZ stage.2015.e. Samples sors capable of measuring the distances to the top and bottom of the rib- of ribbons from the Mini-Pactor® were collected in real-time from a bon. M. target settings of gap and roll force) and After the volume measurements are complete. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx 3 Fig.. Macro-Pactor® (large scale) T ¼ D−U−L Even after scraping off the left side roller.. surement. the ribbons were still too large to sample while the equipment was running.e. Allesø et al. Given this relationship.

0%. Given that the density of the oil (ρoil) and the density of the particles (ρp) were known. increased the powder densification and consequently reduced ribbon porosity from Fig. regarded as tion trials on a compactor with movable gap and inclined roll configura- exceptionally small in size. although ε ribbon ¼ 1−  100%: much less pronounced than the effect of force. which ultimately is linked to higher ribbon porosity. they are less common a total analysis time (from ribbon loading to display of result) of less than those related to roll force since many studies operate the roller than 2 min.5% to 21. European Journal of Pharmaceutical Sciences (2015). Statistical analysis Statistical comparison of the medians from two groups of samples (Mini-Pactor® vs Macro-Pactor®) was performed by the Mann–Whitney U test using SigmaPlot (ver. (2005) in- (approximately 1 mBar) in a desiccator and immersed in a low viscosity vestigated the effect of gap size using an instrumented roller compactor Please cite this article as: Allesø. a scan step size of 0. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx oil.. respec- tively.01 mm.6% (Macro-Pactor®). were scanned at a 0. Allesø et al. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib. a few ribbon fragments. et al. and found that an increase scan step size in the X-direction was fixed at 0. At a fixed gap of 2.5.11. the AccuPyc 1330 (Micromeritics. Illustration of the basic principle of the laser-based ribbon volume measurement.1. while the hardest ribbons (approximately 20% porosity) were found in batch #3. 11.5.. the volume of the oil (Voil) and the vol- ume of the particles (Vp) could be calculated as basis for the calculation of ribbon porosity (εribbon):   V oil εribbon ¼  100% V oil þ V p where Voil = moil/ρoil and Vp = mribbon/ρp.0 kN/cm The reader is referred to the work by Iyer et al.7 mm in the Y-direction tion as for any compaction process.1% to 21. http://dx.ejps. This negative effect of roll force on ribbon detailed description of the system.6.4 M.001 . increases the tensile ρp (true density) of the MCC and magnesium stearate mixture used in strength of the resulting tablets compressed from the roller compactor the current study was 1. While stud- (resolution) was applied to almost all ribbon measurements.. 2. (2014a.942 g/cm3 2. USA). M.0 to 4. The effect ! ρribbon of gap size was positively correlated to the ribbon porosity..1% (laser) to 25. providing ies investigating the effect of gap can be found. Increasing the roll force from 3. similar to the type used in this study. Rambali et al.2% (laser) on the Mini-Pactor® (Table 1).doi. approximately the computed as: same absolute loss in porosity irrespective of gap size suggesting that gap size and force do not interact for this placebo formulation. GA. which the laser-determined ribbon porosity from 47. Performance of the laser method for ribbon porosity determination Ribbon porosities measured by the laser and oil intrusion method are listed in Tables 1 and 2.e. In the Macro-Pactor® intra-ribbon variability study (see compactor at a fixed gap. Overall the laser and oil intrusion method provided very similar results and both corroborated the effect of gap size and roll force on ribbon porosity that one would The results were considered significant if P b 0.2. batch #1 including the most fragile ribbons with the highest porosity (approximately 50%). Oil intrusion method and thus a higher number of inter-particulate bonding sites available The sample ribbon (with known weight) was kept under vacuum for subsequent tablet compression.g. at a fixed gap of 4. with com- parable numbers on the Macro-Pactor® being 50.565 g/cm3 and was determined using an (RC) granules. respectively. the ribbon porosity (εribbon) can be Pactor®).b) for a more roll force (batches #3 and #4). Results and discussion 3. Although they did not assess the ribbon density.0 kN/cm. yet significantly. Medium chain triglyceride oil was used in this study and its density at room temperature is 0. tablet compression. Bindhumadhavan et al.. (2001) conducted dry granula- the Experimental Design section). and 44. reflecting samples processed on the small- scale Mini-Pactor® and large-scale Macro-Pactor®. the five batches span a relatively broad range of ribbon porosities.0 mm slightly. USA). The vacuum was then released and the sample was allowed to ab- sorb oil into the internal ribbon pores for 2 h. i. Inc.0 to 3. e. higher tablet strength suggests better recompactability of the MCC- based granules.0 mm. The value for in gap from 2.2% (Mini- were compressed to form the ribbon. The amount of oil absorbed (moil) was calculated by measuring the increase in mass of the sample. i.05 3.0 mm increased the porosity by approximately 6% at both a 3. 2. The tion.6% and 26.0 mm. San Jose CA.2015. Norcross. a similar increase in roll force decreased Given the operator's input of the density of the particles (ρp).2. Systat Software. 51.0 to 11..4 mm step size. porosity is logical and follows the same principles of powder densifica- In the current study. Excess oil was removed by wiping the ribbons carefully with a tissue.e.0 kN/cm roll force (batches #2 and #1) and at 11.0. Here an increased gap ρp size from 2. Due to varying combinations of roll force and gap size.1016/j.

0 34.5 27.6 −1.7 28.99) for both the Mini.1 −2.4 17.1 19.0 mm Laser 50. These samples were underestimated by as much as 3%. the latter being destructive. and the vol- of 3.0 mm Laser 46.9 34.4 44.0 27.2 49. the biggest underestimation of the oil-based method was ity data was excellent (R2 = 0.0 kN/cm and found that the average maximum pressure between the rolls de.6%) spective of the porosity variability within a batch.5 0.5 51.9 18.and the oil-based poros. Each ribbon was first measured by the non-destructive laser-based system. For the six samples from each batch manufactured on the Mini. then followed by an oil intrusion measurement. http://dx.3 25.4 32.0 mm Laser 50. although the lowest porosity ribbons (21.5 18.2 Force: 3.8 26. the nature of the vol- Although the two porosity determinations performed equally well in ume displacement method supports the hypothesis that this method is their ability to detect systematic and random porosity variations en. et al.0 mm Laser 35.5 28.5 45.0 16. Mini-Pactor® Batch # Technique Ribbon porosity.5 28. The porosity distri.1 27.2015.6 −2.3 30.0% Gap: 2.9 33.. ΔPorosity (%) (%) (oil-laser) 1 2 3 4 5 6 #1 Oil 50.0 Force: 7.4 0.5 19. M. The average value (n = 6) of the laser-based study.8 51.6 Force: 11.4 −2. such as the one used in the present the laser method.4 51.1% Gap: 3.1% Gap: 2. #5).1 51.3 22.7 43.doi.6 42.0 0. However.5 25.2 −2. The linear correlation between the laser. % (six ribbons) Avg. St.1016/j.0 kN/cm #2 Oil 44.5 39 40.2 18.7% Gap: 2. batch datasets.5 3.8 32.0 mm Laser 22. respectively. For the 25 mm ribbons (Mini- in Fig.7 47.5 52.5 2.9 32.2 0.. an overestimation by the laser method. St.3 32. the positive offset of the regression line the laser system were calibrated against certified standards.6 1. This impact of It can be argued as to whether the discrepancy between the two scale on the porosity distribution is further investigated in the following methods represents an underestimation of the oil intrusion method or section.6 −0.4 21. not a lack of accuracy of the porosity determinations.9 25.2 0.0 kN/cm Please cite this article as: Allesø.9 50.7 29.4 3. M.0 kN/cm #5 (CP) Oil 33.5 −1.6% Gap: 2.1 18.0 Force: 3.2 52. where the laser SD was slightly less than values) were generally underestimated by 2% (batches #2.7 21. eventually leading to more porous ribbons.0 0.0 53.8% and the oil-based SD was slightly less than 0.8 20.8 24.1 0. Macro-Pactor® dataset. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx 5 Table 1 Mini-Pactor® ribbon porosity results from the oil-based method and laser method.8 Force: 3. which was in the same re- higher SDs were in fact attributed to the roller compaction process and gion as for the ribbon batches of 28 and 45% porosity (average values).9 44.9 26. The excellent correlation between batches.9 48. confirmed that the were underestimated by approximately 2%.4 23.6 −1. Pactor®).dev.7 32.0 kN/cm #5 (CP) Oil 36.2%.0 47.6 34.8 51.4 −2.2 25.8 52.5 1.8 49.6 29.6 2.1 49.8 Force: 7. underestimating the results slightly. both porosity measurement techniques provided very low the samples in-between approximately 27 and 47% porosity (average standard deviations (SD's). ΔPorosity (%) (%) (oil-laser) 1 2 3 4 5 6 #1 Oil 51.2 25.1% for the laser agreement between the two techniques than with the other ribbon and oil-based methods.1 24.and Macro-Pactor® found for the strongest ribbons of low average porosity (21. Fig.7 45. % (six ribbons) Avg.9% to 3.5 53.2 Force: 11. #3).4 21. dev.2 27.0% Gap: 3.7 29.5 44..8 27.11. with SDs from 1.6%.0% and 1.0 mm Laser 21.4 52.0 18.9 50.0 2.7 46.5 0. Allesø et al.9 22.2 49. 3.1 21. then followed by an oil intrusion measurement.5 27.4 25.0 mm Laser 38.4 51.9 49.5 27. batch bution of the six samples collected from the large scale Macro-Pactor® #1) the differences in porosities were close to 0% confirming better was broader.9 Force: 11.6 2.1 22.8 45. A similar pattern of underestimations was identified in the the two methods for the Macro-Pactor® (same ribbons analyzed).0 32.5% Gap: 4.0 21.5% porosity.9 2.9 31.2% for the Mini.8 50.8 43.1 21.6% to 3. Both the lasers and the balance of countered during processing.1 31.0 kN/cm #4 Oil 25.4 Force: 11. porosities subtracted from the oil intrusion porosities are depicted in creases at higher roll gap.1 34.8 Force: 3.6 44.4 33.7 35.0 kN/cm #3 Oil 19.1% Gap: 4. In contrast.5 38 34.0% Gap: 4. for the most fragile ribbons (51.4 52.ejps.0 21.0 18.0 mm Laser 24.0 0.3 21.8 41.8 44. Each ribbon was first measured by the non-destructive laser-based system.5 47.8 32.0 mm Laser 46.0 kN/cm #4 Oil 22.2 49.2 46. 4 as quartiles (n = 6) in a box plot against the average ribbon poros- The ribbon porosities shown in Tables 1 and 2 are depicted graphically ity determined by the laser method.2 33.001 .2 21. rely on the complete absorption of the liquid medium into the Table 2 Macro-Pactor® ribbon porosity results from the oil-based method and laser method.3 42.9 45.5 29.9 34. #4 and 0.6 −2.0 mm Laser 27..0 kN/cm #2 Oil 44.0 46. whereas Pactor®.0% Gap: 4.3 45.and Macro-Pactor® suggests a tendency ume and mass provided can therefore be assumed to be highly accurate.7 50. of the oil intrusion method to underestimate the porosity relative to Volume displacement methods. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib.3 48.2 32.. Macro-Pactor® Batch # Technique Ribbon porosity.6% and 3. irre.1 23. the latter being destructive.5 2.7 26.0 kN/cm #3 Oil 18. European Journal of Pharmaceutical Sciences (2015).

/ European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx Fig.11.. able for further investigations of the Mini. the non-destructive nature of the laser method enabled the 3. to improve sampling in an plied to the ribbons. an analytical parameter that cannot be ac- quired for the oil intrusion method. the bons. which resulted in a broader porosity distribution.. six to twelve. M. the data suggested that the laser system was very suit- timation of the pore volume and thus porosity. This was seen as a reduction in to be a true measure of between-ribbon variability. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material fragments taken from the same 100 mm primary rib- quence of the jaws grasping the ribbon. Each sample coordinate in mined by oil intrusion and the laser method at given mean porosity (measured by the the plot represents the same ribbon analyzed first by the laser system and then by oil laser). The red dashed line is the mean value (n = 6). whilst for the the initial mass from start to end by approximately 6 mg for the fragile Macro-Pactor® the measurement variability represents a mixture of ribbons and 2 mg for the strong ribbons.2.e. explanation.6 M. although sampled 0..2% across the entire porosity range from 21 to 53%. Impact of roller compaction scale on ribbon porosity assessment of repeatability. Three single ribbons. Incomplete intrusion would result in an underes. from batches As evident from Tables 1 and 2. Each of these ribbons was measured six consecutive times with collected ribbon sample from the Mini-Pactor® represented the entire the laser system. With a standard deviation of merely population of n = 6 ribbons from the Macro-Pactor®.. consisted of random fragments from several 100 mm rib- clearly performs consistently with a high precision. Hence. 4 support this lowing scale-up investigations. bon) and between-ribbon variability. ically expected to be higher for hard compacts and lower for fragile hence it was the only porosity measurement technique used in the fol- compacts.2015. Data are shown for A) Mini-Pactor® and B) Macro-Pactor®. 1). Collectively. cates an underestimation by the oil-based method. Finally. This increase in population size was done for both the Please cite this article as: Allesø. does not impact the validity of the attempt to obtain a better measurement of the average porosity of the results since the ribbon volume correspondingly was measured to be 100 mm primary ribbon. and the band inside is the second quartile or median (50%). The variability raw data reveal a very minor wear of the ribbons during the repeated of 25 mm ribbons collected from the Micro-Pactor® may be considered performance of the six measurements.001 . European Journal of Pharmaceutical Sciences (2015).1016/j. A negative value indi- intrusion. et al. the laser system in real-time. This was not unexpected since every analysis. This wear was mainly a conse. were selected for repeatability collected from the Mini-Pactor®. Box plot showing the distribution of differences between ribbon porosities deter- for (A) Mini-Pactor® ribbons and (B) Macro-Pactor® ribbons. http://dx. including the raw data volume 25 mm roll width (see picture of ribbon in Fig. however. In contrast. Interestingly.and Macro-Pactor® ribbons. a sample and mass. are presented in Table 3. ribbon fragments from the Macro- which represented the edges of the experimental space (batches #1 Pactor® runs yielded a higher standard deviation compared to ribbons and #3) and the center point (batch #5). a probability that is log. the sample population was increased from lower by the lasers. 3. 4. and the liquid will not be able to enter closed pores of the compact. Allesø et al. pores of the compact. within-ribbon (i. This. The experimental results plotted in Fig.ejps. The bottom and top of each box are the first (25%) and third (75%) quartiles. as this was the only stress ap. The porosity results. Plot of ribbon porosities determined by oil intrusion against the laser-based results Fig.

038 20. The average which was comparable to the SDs of the all other ribbon batches from porosity of each of the three ribbons within a batch (n = 8) were very the Mini-Pactor®.2 35.4 528.272 Gap: 3. but rather slightly increased the standard porosities which ranged from approximately 28% to 37%.0 0. not to discard the outlier similar.0 1.996) between ribbon porosities obtained on were twelve fragments chosen randomly from the 100 mm primary rib. M.631 51. The porosity data are listed in Table 5 also showing the total measured rosity for batch #1 (n = 11) of 51.0 21. two compactor scales for batches #1.2 21.0 651.166 7.0% manually by hand.2 52. mg Average St.7 52.and Macro-Pactor®. manufactured on the Macro-Pactor®. While in gen- deviation (now up to 3. four additional batches were crease in sample population did not improve the ribbon porosity distri. 0.7 Force: 11.4 529.0 kN/cm #4 Mini-Pactor® 7.4 527.0 kN/cm #2 Mini-Pactor® 6. in spite of measurement. which was comparable to the SDs ob- rosity obtained on the Mini.5 652. and again the former twelve level.. which confirms that the roller compaction process.. The data of Table 4 are depicted graphically in Fig. Table 4 summarizes the average porosities from the 12 measure.1 385. with the ex. of complete 100 mm wide ribbons. et al.675 9.0 mm Macro-Pactor® 6. (11 out of 12 measurements fell within the 50.2 390.491 8.4 391. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx 7 Table 3 Results demonstrating the repeatability of the laser-based technique.3 3 527.088 10.. and 45.4 1030.3%). The in.152 45. Although not shown in Table 4.2 5 522. 7.4 St.3 Average 21.5%).9 1011. dev.1%.2 1028.3 52. As the sectioning was done 22.2 6 522.7 1029.8% for the Mini-Pactor® and Table 5 are presented graphically in Fig. 1.3% po.738 47.7 1020.0 kN/cm #5 (CP) Mini-Pactor® 10. The excellent samples were complete 25 mm ribbons.795 Gap: 4.3 2. 20.ejps.8 651.11. The P-value is derived from the Mann–Whitney U test for comparison of two medians. To investigate whether any pattern exists in the porosity distribution From a practical point of view the average ribbon porosities of batch #3 of a 100 mm ribbon.0 mm Macro-Pactor® 7. manufactured. dev.5% to 52.554 27. However.214 Gap: 2. based on 12 randomly compactor at identical settings of roll force and gap size.8 35. bon.623 8.9 651. dev. the porosity data for each ribbon fragment in were very similar for the two scales.206 6.4 0.940 5. volume per 100 mm ribbon and its corresponding mass.7 21.2015. #2. 6) and larger SD. with settings of force and gap providing average ribbon bution of the Macro-Pactor®. the porosity of each individual segment was determined using the rosity range. outlier from the dataset provides an average Mini-Pactor® ribbon po. was in the range of 1 to 51. 5.824 11. ments per batch and also shows the raw data.977 34.0 kN/cm Please cite this article as: Allesø. Six repeated measurements were performed on a single ribbon selected from each of three separate. Batch #1 Batch #3 Batch #5 (fragile ribbon) (intermediate ribbon) (hard ribbon) Volume Mass Porosity Volume Mass Porosity Volume Mass Porosity (cm3) (mg) (%) (cm3) (mg) (%) (cm3) (mg) (%) 1 528.017 Gap: 2. and the latter twelve samples linear correlation (R2 = 0.9 21. It was decided. Mini.5 1.and Macro-Pactor® when operating the tained in the experiments previously described. the two compactor scales confirm the expected 1:1 scalability of the bon. A statistical collected ribbon fragments.2 Mini-Pactor® and Macro-Pactor® samples.7 Force: 7. performed well and that gave similar porosities (45.7 0.8 1.1016/j.0 Force: 11. since triplicate measurements on that same ribbon some slight heterogeneity within a ribbon. dev.2 St.0 528.3.doi. There was no apparent relationship be- comparison of medians did not suggest any difference between the tween the chosen gap size and the within-ribbon standard deviation. European Journal of Pharmaceutical Sciences (2015).4 35. the dimensions of the individual fragments varied and 1.2 St.5 529.4 1030.5 52.013 35.4 1021.6%. mm3 Total mass. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib.9 35. Three whole ribbons per batch ment from Mini-Pactor® batch #1 was primarily responsible for the were sectioned into eight segments with a trimming knife (Fig.5 3.0 0.7 35. 45. Standard deviations were generally very eral 100 mm ribbons break up into fragments as they leave the roller.3 4 526.3 Force: 3.7%) they became statistically different at the 95% confidence and comparisons should be made with some caution.4 652. http://dx. the process was robust.8 388. low in the ribbons from the Mini-Pactor®. 0.1 651.. ribbon volume and To gain more insight into the variability between ribbons mass. with a single outlier measurement at 45. constituting the basis of the reported porosity results.7 0. however. #4 and #5 at 95% confidence.9 21.9 389.0% or lower.6 1031. for the 8 segments of the rib- There was an excellent agreement between the average ribbon po. The SDs (n = 8).0 mm Macro-Pactor® 6.005 27.195 22. the Table 4 Ribbon measurement results (n = 12) from the Mini.8 Average 35. 0.0 35. Allesø et al.1 0.8 Force: 3. Measurement no.3 533.2 0. % Total volume.8 1018.9 1019. All measurements were done by the laser system and the table shows raw data (volume and mass) and the calculated ribbon porosity. P-value (n = 12) (n = 12) (n = 12) (n = 12) #1 Mini-Pactor® 10.0 0.and Macro-Pactor® trials.0 mm Macro-Pactor® 10.0 1033. a single measure.3 Average 52.0 kN/cm #3 Mini-Pactor® 8.722 8. but due to low SDs on both scales (1..9% and brings the SD down to 0. Batch # Equipment scale Ribbon raw data Ribbon porosity. respectively.9 1019.0% for the Macro-Pactor®.0 2.0 mm Macro-Pactor® 12. Omitting the laser-based method giving a total of 3 × 8 measurements per batch.525 Gap: 4. M. the relatively low porosity of these special batches allowed collection ception of batch #1.646 10.2 21.5 385.0 52.5 52.6 3.8 2 523.001 .

However. fragments 3 to 6) and the edges tion to be periodical and linked to the period of the screw feeder. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx well as process conditions (gap size. (2015) investigated the difference in relative den- sity of an MCC-based formulation processed on a pilot scale roller com- pactor with rolls mounted in the vertical position and powder feeding with a single screw. 2008). For ex. Each ribbon segment has been numbered so that ribbons 1 and 8 represent the edges of the whole ribbon with 4–5 representing the center (approximately). difficult to interpret the density distribution Macro-Pactor® experiments. The authors attributed this to the friction between the powder and fixed side plate.ejps. (2015) demonstrated the application of terahertz spec- edges. The two dashed lines outline the 95% confidence interval of the regression line. Guigon and Simon (2003) investigated the heterogeneity in density of sodium chloride ribbons processed on a lab-scale roller compactor graphs are useful for revealing potential systematic differences in the equipped with 130 mm smooth rolls. but not the present.. Using NIR chemical imaging (at- line) Khorasani et al. two of the ribbons of batch #4 show this pattern while a third manufactured on a roller compactor with inclined roll configuration. RI) of ribbons ample. European Journal of Pharmaceutical Sciences (2015). Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib. http://dx.e. gap). however. They found the density distribu- porosity between the middle (e... direct comparisons described above. This observation was in contrast profiles into porosity generally revealed lower porosity in the middle to other studies. Photograph of a Macro-Pactor® ribbon before and after sectioning with a trimming knife. which improved the density distribution across the ribbon width (Miguélez-Moran et al. 2 and 7. similar to the one used in the present work. however. and compac- tor design) can influence the density distribution across the ribbon width. M. this pattern was not consistently observed within a batch. (2009) applied micro-indentation. Although there was a tendency timately the uneven distribution of feed pressure. among others.doi.8 M. In a very recent study for some ribbons to be more porous in the middle and less porous at the Sullivan et al. Therefore. roll speed. This difference was most pronounced for ribbons manufactured at the largest gap (here 1. The pattern.g. troscopy for density mapping (via refractive index. 8) of a ribbon. The importance of the friction between powder and side sealing material was confirmed by the addition of lu- bricant. The authors demonstrated a tendency of all ribbons to be most dense in the middle (least porous) relative to both edges. to measure the change in relative density across a 200 mm ribbon width manufactured on a lab scale instrumented roller compactor using gravity feeding. et al.g.and Macro-Pactor® (roll force and in the slope and shift of the NIR spectrum baseline (i. Please cite this article as: Allesø.. Information on process tal setup the heterogeneity in porosity seemed to be randomly defined parameter settings in this study is very sparse.001 . which revealed some repeatable patterns in ribbon relative to the edges. and ul- (e. which is in accordance with the research of others density distribution across the roll width. inhibiting powder pull-in at the edges and consequently less powder compaction at the edges since the gap was fixed. which is probably tion factors like filler compression properties and powder lubrication as due to the period of the feed screw described by Guigon and Simon Fig.11. Plot showing the relationship between ribbon porosities from the Mini-Pactor® and dense.1016/j.0 mm compared against 0. fragments 1. seem to be constant as a function of process time.. since formula.2015. ribbon shows the exact opposite pattern.9 mm) and high roll speeds (6 rpm). i. roll width. scatter effects).org/10.e. They found ribbons to be more heterogeneous when compacted at higher forces. 6. for this experimen. Ribbons from each of the five levels have been manufactured maps since these were provided as relative values based on the changes at identical process parameter settings on the Mini. did not with studies from different research groups are difficult. with the edges being the least Fig. It is. 5. Allesø et al.. Miguélez-Moran et al. but translation of the RI during the roller compaction process.

5 26.2 22. A direct and rapid laser-based method was used to determine Fig.5 19.4 21.2 22.4 4.5 37.4 37.1 21.5 21.8 7.2 32. Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib.2 30.267 (2003).8 34.4 2. Conclusion also explain the lack of a repeatable pattern in the porosity distribution of the 100 mm ribbons plotted in Fig.2 32. and all measurements were done by the laser system.4 31.4 36.3 6.5 1.1 31.ejps.5 37. Considering the inherent high The major aim of this study was to investigate the hypothesis of rib- sensitivity of a plastic material to deformation upon pressure. Results from 4 combinations of roll force and gap (both at 2 levels) are shown.4 31. mg (n = 8) (n =8) (n = 8) (n = 8) 1 2 3 4 5 6 7 8 (edge) (edge) #1 A 30.8 4.0 kN/cm C 32.0 32.318 6. Allesø et al.2015.1 23.668 8.910 7.3 28. covering the entire 100 mm width of a Macro-Pactor ribbon.5 32. that ribbon porosity is not affected by a from a practical point of view and the average porosity value of a change of roll width from 25 mm to 100 mm at constant RC parameter 100 mm ribbon was very repeatable. M. et al.4 34.8 3.912 4.0 mm B 22.2 38. Porosities of the eight ribbon segments.0 kN/cm C 16.4 27.2 29.168 #4 A 26.0 24.080 6.6 28.2 31...143 Force: 11.5 2.9 36.951 Force: 7.8 26. 7.1 29.989 6.0 19. like the bon porosity being scale-independent when using roll width as scale up MCC used here.3 2.1 6.8 2. mm3 Total mass.8 37.0 mm B 37..5 20. covering the entire 100 mm width of a Macro-Pactor ribbon.5 33.9 37.7 28.665 8.0 20.6 20.5 39.4 29.1 6.9 30.6 32. Total volume.6 30.9 23. In the present investigation.964 5.2 23.11. these within-ribbon variations were deemed minor parameter.3 38.741 Gap: 4.2 5.0 kN/cm C 29.2 31.082 7.854 #3 A 24. http://dx. more specifically.) The table shows raw data (volume and mass).001 ..8 3.7 1.8 21.4 32.783 Force: 7.5 2.0 mm B 27.849 5.9 36.6 18. % Ribbon raw data Ribbon segment # Average St.9 30.3 27. and the calculated ribbon porosity.3 32. Batch # Ribbon # Ribbon porosity.0 30. Results from 4 combinations of roll force and gap (both at 2 levels) are shown.7 24.578 6. European Journal of Pharmaceutical Sciences (2015).9 35.doi.616 Force: 11.5 29.437 Gap: 2.7 1. dev. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx 9 Table 5 Porosities of the eight ribbon fragments.7 35.8 27. Please cite this article as: Allesø.8 28.1016/j.1 28.5 32.5 5.0 38. M.6 24.9 31. the period of the feed screw may 4.1 21.9 0.1 35.3 6.7 29.0 21.0 36.9 21.8 5.6 31.0 30.8 6.9 37.7 36. (Three ribbons were sectioned and analyzed per experiment).5 31.0 37. (Three ribbons were fragmented and analyzed per experiment..9 23.779 #2 A 33.9 20.4 36.502 Gap: 2.0 6.8 27.8 35. settings.3 26.0 kN/cm C 37.1 22.0 18.1 26.2 31.5 0.4 28.0 25.0 mm B 30. 7.3 27.873 Gap: 4.8 34.0 30.483 6.

M. Olsen.2%.. C.S. Res.. Dave.. M. A. et al. 52–59.. Rathore.. Pharm. Int. J.. Anal.M.N. 10696–10706. Assessment of the Mini-Pactor® ribbons since each ribbon measured naturally represent. Zhou.. Comput..-Y.. A.. 41–48.. Pharm. process settings of roll force and gap size applied to the Scale-up of a pharmaceutical roller compaction process using a joint-Y partial least Mini-Pactor® (and identified during formulation development) are squares model. Pharm. J. Presenting a rational approach to sionless approach. Wu. Fitzpatrick. http://dx. 2012. Mechanis- troscopy. D. N. Wu.. J. tion on the Macro-Pactor® with same roll diameter. A. Khorasani.K. and Guigon. Massart. Pharm.M. MacGregor. J..L. J. Comput.. 2011. J. Josefson. 3586–3595. M. 72. 2001.W. 60.. (Ed.ejps. The authors are thankful to Uffe Teis Svare and Jacob West for their Nkansah. Wu. Seville. 2009. Souihi. Zavaliangos. Density mapping of roller-compacted ribbons using terahertz spec- Cunningham. Rantanen. Pharm. Int. V. Seville. N. P. fraction using a laser-based direct volume measurement device. C. P.. Sobotka.S. M. Powder Technol.. E. Pei. Kidder..J. 11–17. 58. H. P. 2015.. Pharm.. Roller Acknowledgments compaction process development and scale up using Johanson model calibrated with instrumented roll data.. approaches for real-time monitoring of roller compaction with near-infrared spec. Sprockel. Hoag. Chem.. 173–182. QbD-based pharmaceutical development: a roller compaction case study. Miguélez-Moran.. M. 2010. Torstenson. W. Appl.D.. 19..F.. L. New York. 2008. R. M. Rev.. A. G. 159–190. 102. W. Eng. Kothan. J. pp. D. R. Trygg... 2014a. J. K.. Jerzweski... Reynolds. S.. Roll press design—influence of force feed systems on compac- tion. Eur. 2011.M. Pharm. Sci. 200–211. 486–507. A. Hegde. W.K. E 32 (4). C. J. Sci.. Sonnergaard. Pharm.. J. paring the average porosity of five groups of ribbon samples (each with 19. 34. Bryder.. Holm. Eur.. 583–592. Characterisation of density link between formulation development and technology transfer. 6. Roll compaction/dry granulation: pharmaceutical applications... Roller compaction scale-up using roll width as scale factor and laser-based determined ribbon porosity as critical material attrib.. S. P. L. Early. Jans..J.. 50.. Sci. S. Nesarikar. Fahmy.... O.. S. Nikfar. O. A novel method for estimating solid fraction of roller-compacted ribbons. pp. Litster. Chem. Int. E. Champagne.V. R. 362.. McKay..W..-J. Heaps. J. Tablets & Capsules: granulation analysis. C. Bruwer. Please cite this article as: Allesø. Ser.M. theory for granular solids. the other. Singhal. R. Baert. 1058–1071. Iyer. P. 2015. Malick... Sci.H. Kleinebudde.M. 317–326. F. paction and tablet compression on physicomechanical properties of pharmaceutical proach to acquire ribbon porosity rapidly (less than 2 min per ribbon). A. S. Handbook of of the commercial process.P. Yamamoto. R. Influence of the roll compactor parame- ter settings and the compression pressure on the buccal bio-adhesive tablet proper- References ties. J. M. Patel. J. compaction through finite element-based process modeling.. With the intention to reproduce ribbon porosity from one scale to Liu. and is in agreement with reports from the literature. Greenwood. 2005. Carragher.. Technol. Pharm.P. The higher standard deviation of the measured po. H. Mech. CI) on roll compacted ribbons and tablets — multivariate mapping of physical and Bindhumadhavan.. 2013.R.. Ind. 3891–3897. Pharm. J. Adams. V..).R...-L. Reed.. S.. / European Journal of Pharmaceutical Sciences xxx (2015) xxx–xxx the porosity of the ribbons at-line.. decreases the number of batches needed to establish the parameters Miller. J. The impact of roller com- method..K. Ingale. 34.. C. 34.. DiNunzio. chael Hanisch are thanked for the technical discussions.. skilled contributions to the experimental work.. Chem. Sci. 2013. Roberts. S. Z.M.-J. n = 12 samples) from the small-scale (Mini-Pactor®) to the large-scale Johanson. Rambali. A rolling theory for granular solids.1016/j. R. Near-infrared chemical imaging (NIR- Pharm. Terry Chriss and Mi- 142–148. 220. Int.S. Nilsson. D. Amigo. Romanach. The effect of lubrication on density therefore directly transferrable to subsequent commercial scale produc. 842–848. Winstead. Giridhar. ASME J. 483. E. J. D. Vatsaraj.. J. AAPS PharmSciTech 13. Practical application of Acevedo.. R.. 17–24.. excellent agreement with a well-established volume displacement Iyer. D. and distributions in roller-compacted ribbons using micro-indentation and X-ray micro- computed tomography. 2008. 2014b.. Hegde. Shao.. Roll compaction of a pharmaceutical ingredient: experimental validation of rolling Sullivan. J. Evaluation of three roller compaction process modeling.. 410. D. Z. of using NIR chemical imaging to evaluate the porosity distribution. J. 1005–1012. Eur. 577–582.. This creates a better Miguélez-Moran. Kato. Malick. Eng. V.. European Journal of Pharmaceutical Sciences (2015). D. R. J.. Biopharm. Biomed. This within-ribbon variability was not assessed on the smaller scale Lim. Vatsaraj. Crison.. McCann. Understanding variation in roller troscopy. Singhal. 129–140. J.10 M. M.-Y. J. (Macro-Pactor®).W.. excipients. Muliadi... 1965.. A novel approach to determine solid The ribbon porosity was found to be scale-independent when com.M. Bertelsen. 2012. Pharm. B. S..-Y. chemical properties. 109. R.. Sci. Biopharm.C. Visu- rosities of ribbons segments from the large-scale roller compactor was alization and prediction of porosity in roller compacted ribbons with near-infrared attributed to minor variations in powder densification across the roll chemical imaging (NIR-CI). D. Pharm. J. Technol.. 2010. Eng. Dev.. Sci. distributions of roller compacted ribbons.. critical factors affecting the porosity of roller compacted ribbons and the feasibility ed the entire roll width (25 mm). J. Lewis.. 1049–1057. D. 2003.S. 1–8.J.. Simon. Taylor & Francis Group... Dev. G. 130.. tic insights into the scale-up of the roller compaction process: A practical and dimen- Allesø. Eng. 2005. P.2015.J.. R. X. Schiano.. Allesø et al.E. Ind.. the laser-based method was indeed a powerful and elegant ap. 26–32 (January). Zhang. Gururajan. Dong. Roller compaction technology. Chem. Wu. pharmaceutical granulation technology. 436. Seville. B.J.. 2015.11. T. Int. R.. In: Parikh. Rowe.-Y.doi. width (100 mm). S... Drug Dev..P.J.001 . 2004. With a repeatability of 0.