Vivek - IJP Paper

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages8International Journal of Pharmaceutics xxx (2011) xxxxxx

Contents lists available at ScienceDirect

International Journal of Pharmaceutics

journa l homepage: www.e lsev ier .com

Assessm poribbon alporosit

Hanpin L , Raaa School of Pha atesb Lubrizol Advac Ofce of New ited Sd Malvern Instr

a r t i c l

Article history:Received 17 AReceived in reAccepted 21 FAvailable onlin

Keywords:Roller compactionNIR chemical imagingPorosityDensityRoller compacFeed screw spspeed, Quality

ss thion, tamineroll sp

decreased as RP increasedwith the exception of ribbons produced by the combination of high RS and lowFSS where increasing RP increases the porosity of the ribbons. Lower RS was found to produce ribbonswith lower porosity and the porosity increases as the RS increased. Increased FSS will decrease ribbonporosity at higher RS while it slightly increase the ribbon porosity at lower RS. A simple linear regressionmodel showed NIR-CI was able to predict the ribbon porosity with a correlation of 0.9258. NIR-CI is ableto characterize differences in porosity as a function of position on the ribbon where regions with lower

1. Introdu

Roller cogranulationity withoutwhen worksensitive anFeng et al.,process, rolder blend isby passing iThe resultinthenmilledThe continution, time shas the potChu, 2007).

The fundplex, and lik

CorresponE-mail add

0378-5173/$ doi:10.1016/j.this article in press as: Lim, H., et al., Assessment of the critical factors affecting the porosity of roller compacted ribbons and theof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Pharm. (2011), doi:10.1016/j.ijpharm.2011.02.028

ted ribboneed, Roller pressure, Rollerby design, NIR-CI

porosity show higher absorbance. Nevertheless, NIR-CI is able to show sinusoidal variation in intensitiesalong the roller compacted ribbon among all settings studied.

2011 Elsevier B.V. All rights reserved.

ction

mpaction has gained popularity in recent years as a drymethod to improve material ow and compressibil-the use of heat or solvents. This is especially benecialing with active ingredients that are heat or moistured therefore cannot bewet granulated (Ende et al., 2007;2008; Ghorab et al., 2007). Unlike the wet granulationler compaction is a continuous process where a pow-compacted and consolidated into a sheet of solid masst between two counter-rotating rollers under pressure.gproduct is called the roller compacted ribbon,which isinto granules of desired particle size (Peck et al., 2008).ous process has many advantages, consistent produc-calability and fewer pieces of equipment, and thereforeential to reduce manufacturing costs (Daugherity and

amental mechanisms of roller compaction are com-e othermanufacturing techniques, product quality and

ding author. Tel.: +1 410 706 6865; fax: +1 410 706 0346.ress: [email protected] (S.W. Hoag).

performance depend upon raw material properties, machine con-struction and process variables. Material properties such as theparticle size andmorphology of the rawmaterials have been shownto affect the compaction properties of the ribbons, granule particlesize distribution, owability, content uniformity and compactionproperties of the tablets (Bacher et al., 2007, 2008). Studies haveshown that the tap and bulk density of the resulting milled granu-lation was consistently higher when made with smooth rolls thanwith a linear knurl (serrated) roll surface (Daugherity and Chu,2007; Sheskey and Hendren, 1999). The volume of the serrated rollsurface can signicantly affect the ribbon thickness, as serrationdraws powder into the roll surface; the greater the serration vol-ume, the thicker the ribbon (Daugherity and Chu, 2007). Processvariables such as feed screw speed (FSS), roll speed (RS), roll pres-sure (RP), roll gapandmilling conditioncanalso impact theporositydistribution on roller compacted ribbons and the properties of theresulted granules (Daugherity and Chu, 2007; Peck et al., 2008).

Maintaining constant process parameters throughout the entireroller compaction operation does not always guarantee a com-pletely homogenous ribbon. For example, the motion of the lastight of the spiral feed screw has been shown to create periodi-cal sinusoidal density variation across the ribbon width and alongthe ribbon length (in the direction of ribbon output motion) as it

see front matter 2011 Elsevier B.V. All rights reserved.ijpharm.2011.02.028ent of the critical factors affecting thes and the feasibility of using NIR chemicy distribution

ima, Vivek S. Daveb, Linda Kidderc, E. Neil Lewisc

rmacy, University of Maryland Baltimore, 20N Pine St., Baltimore, MD 21201, United Stnced Materials, 9911 Brecksville Road, Brecksville, OH 44141, United StatesAnimal Drug Evaluation, Center for Veterinary Medicine, FDA, Rockville, MD 20855, Unuments Inc.,7221 Lee Deforest Drive, Suite 300, Columbia, MD 21046, United States

e i n f o

ugust 2010vised form 24 January 2011ebruary 2011e xxx

a b s t r a c t

The purpose of this study was to assedifferent process parameters; in additto evaluate porosity variationswas exusing a range of roll pressures (RP),/ locate / i jpharm

rosity of roller compactedimaging to evaluate the

fat Fahmyd, Stephen W. Hoaga,

tates

e porosity variation of roller compacted ribbons made usinghe feasibility of using near-infrared chemical imaging (NIR-CI)d. Ribbons of neatmicrocrystalline cellulosewere compactedeeds (RS) and feed screw speeds (FSS). The ribbon porosity

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages82 H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx

Table 1Roller compaction operating parameters for this study.

Batch # Roller speed (RPM) Feed screw speed (RPM) Roller pressure (MPa) Feed screw speed/roller speed (RPM/MPa)

1 4 20 4 523456789

delivered p2003). Alsofriction bettion in pro(Ghorab etdevelop a bthe roller cothe formulaunderstandmaintain deof environm(FDA, 2006)

Based ongranules arsolid fractioporosity intion in porand densitytion (Migue2006). As oporosity ofticle size da greater voporosity westrength dumaterials.

The rstbulk porosiparameterstrols the mrolls. Contithe rolls is iwhile overcause a blotime of thecertain dwand bondinlead to lamcan lead toand laminacompactionsure on plasmethycellurespondingdue to work

The Proof the USAingredientcontrol stragenerated imonitor ansingle poinuniformity,

s morediccempmef calnt manyationult. Aa maulk cthe ss havstree tecIt canendtweethe mesolvraphrollethe

mposovereasue featy usin pod toality, 200erstndesthouposige (sividu(depwougenetiontiveteris4 20 64 28 44 28 66 24 58 20 48 20 68 28 48 28 6

owder to the compaction region (Guigon and Simon,, heat may be generated in the compaction region fromween the rolls and compacted powder, causing varia-perties of the compacts during long compaction runsal., 2007). Therefore, it is important to identify andetter understanding of the critical factors that affectmpaction process, so that they can be accounted for intion design and be monitored during the process; thising of the process is needed to reliably and consistentlysired quality and product performance across a rangeents as part of the Quality-by-Design (QbD) approach.published research, the quality of the roller compactede inuenced by the porosity (i.e., relative density orn) distribution of the roller compacted ribbons, becauseuences the ribbon strength and consequently varia-osity will lead to a variation in the size distributionof the granules produced for a given milling condi-

lez-Moran et al., 2008, 2009; Tye et al., 2005; Wu et al.,bserved by Gamble et al. (2010) increased in ribbonplastically deforming materials results in broader par-istribution and reduce granule ow properties due tolume of ne particles. Conversely, ribbons with lowerre shown to produce tablets with lower tablet crushinge to the work hardening of the plastically deforming

part of this paper focuses on the characterization ofty variation of roller compacted ribbons when processsuch as the FSS, RS and RP are changed. The FSS con-aterial feeding rate into the compaction region of thenuous ow of material into the compaction region ofmportant as under feeding can lead to less compactionfeeding can cause over compression, which may evenck in the pre-compression zone. RS controls the dwellmaterial under pressure in the compaction region. A

ell time is necessary to allow particle rearrangementg. Hence, short dwell times, for a given pressure, canination and capping (in tablets) while long dwell timesover compression causing the ribbon to atten out

te (Peck et al., 2008). In large part, the RP controls thepressure applied to the material. Higher roller pres-tically deforming materials such as the hydroxypropyllose was found to produce larger granules and the cor-tablets were found to have a lower crushing strength

YoungNIRS preferendeveloation oconstawith minformgle resan areuct, a bproberibbontensilebut th2007).it is depand beacrosstially rtomogtion inused inical co

Tobulk mgate thporosiencescan leauct quSimonter undand nomanytion ofan imathe indscopicwhichheterocalibraqualitacharacthis article in press as: Lim, H., et al., Assessment of the critical factors afof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Ph

hardening (Peck et al., 2008).cess Analytical Technology (PAT), which is a partFDAs QbD initiative, seeks to better understand

process relationships and to establish appropriatetegies, which includes real-time release testing, hasncrease interest in using nondestructive techniques tod control the roller compaction process. These includet near infrared spectroscopy (NIRS) to look at contentmoisture content, relative density, tensile strength and

insights areribbons as tsample, butacross the snometer mof determinerogeneityreported inerogeneity57742.52.53.53.5

dulus. A strong correlation has been found between theted values andmeasurements obtained using a suitableethod (Gupta et al., 2004, 2005a,b,c). However, methodnt for NIRS can be quite extensive, requiring the cre-ibration samples outside of normal ranges, and theaintenance of the model as it is applied in practice. Asexisting analytical techniques, NIRS averages spectralfrom the entire data collection area, producing a sin-

s the variation in chemical and physical properties overy actually inuence the quality of the resulting prod-haracterization method may not be able to adequatelyample. Thermal effusivity results from roller compactede also been shown to correlate with the solid fraction,ngth and Youngs modulus of roller compacted ribbonshnique suffers from several drawbacks (Ghorab et al.,only accuratelymeasure ribbonsmadebysmooth rolls,

enton theabilityof thematerial to transferheat throughn the particles, and aswithNIRS it derives a single valueeasurement sampling area, instead of providing spa-

ed data. Other method includes X-raymicro-computedy where the authors determined the density distribu-r compacted ribbons; however, it is inappropriate to bedensity distribution study when more than one chem-ition is included (Miguelez-Moran et al., 2008, 2009).come some of the inherit limitations associated withrements, the second part of this paper will investi-sibility of determining the spatial dependence of ribboning near infrared-chemical imaging (NIR-CI), as differ-rosity across and along the ribbons width and lengthvariations in compact properties which can affect prod-(Gamble et al., 2010; Ghorab et al., 2007; Guigon and3). NIR-CI can be used as a tool to determine and bet-and CQAs as part of a QbD approach. NIR-CI is a rapidtructive analytical method that allows the collection ofsands of NIR spectra (chemical information) as a func-tion in the sample. These spectra can be compiled intopatial information). In other words, in a single datasetal spectra are associated with a microscopic or macro-ending upon image resolution) point in the sampleld enable a spatial description of porosity distribution inous roller compacted ribbonswithout running separatesamples. Subsequent analysis of NIR-CI data enablesand quantitative insight in the chemical and physicaltics of heterogeneous samples (Lewis et al., 2006). Thesefecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

valuable whenmeasuring porosity of roller compactedhe pycnometer method averages the density across theit does not give information about the density variationample. Using the porosity value measured by the pyc-ethod together with NIR-CI analysis, NIR-CI is capableing the spatial porosity distribution, i.e., sample het-across the whole sample. To date no studies have beenthe literature that have used NIR-CI to study the het-of roller compacted ribbons; hence, if successful this

Please cite rs afffeasibility . J. Ph

ARTICLE IN PRESSGModelIJP-11709; No.of Pages8H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx 3

study will provide a novel tool for roller compaction users to bet-ter understand the roller compaction process. Information aboutthe chemistry and physical organization of the samples providedby NIR-CI is valuable for formulation development, and promotesprocess undprocess suc

In this sNIR-CI to exby monitorlength for dposed of a sin this studof porositychanges in tchanges inthe greaterwill have prNIR illuminanalogous ret al., 2008;

2. Materia

2.1. Materi

The Avicwas obtaine

2.2. Prepara

2.2.1. PrepaRibbons

AlexanderwHorsham, Pface rolls ofIn order tovariation intings achievlong compaspanning thimental desis typicallyhowever, thin porosityically adjusresponse throller gap a

To collements, theFSS of 20, 2tions werefactors andmeasureme76cm in lenwas collectecut with a dand storedreach steadafter 12mples from earotation of texcept for t24 RPM: 5Mrepresent tstudy the repaction.

2.2.2. Porosity measurementsThe true density of bulkAvicel PH102powderwas determined

using a fully automatic gas displacement pycnometer-AccuPyc

1330 helium displacement pycnometer (Micromeritics Instrumentrcroinede of tplacey thred.l presig/menvs weicrompycnhe enl diasioncm3

masthe As perequ

1

P isrue isineds front roan pmplen or

NIR cine Nollecrn InDatam wper

y 40m/pix

ect flowcubehigh

NIR-CNIR-Cre (Mas ge

D

D

S isanceima

ance

10

(this article in press as: Lim, H., et al., Assessment of the critical factoof using NIR chemical imaging to evaluate the porosity distribution. Int

erstanding, especially for complex, spatially varyingh as the roller compaction.tudy, we have attempted to extend the application ofamine and understand the roller compaction process

ing the porosity variation across the ribbon width andifferent process parameters. The ribbons were com-ingle ingredient, Avicel PH 102 and no API was usedy, as the goal was to isolate the spatial distributionas a function of processing parameters. The porosityhe roller compaction ribbon are reected in absorbancethe NIR spectra. The denser (less porous) a sample is,the absorbance will be, while a very porous sampleoportionally less material interacting with the incidentation, andwill have correspondingly lower absorbance;esults were found for tablets (Cantor et al., 2011; TabasiTatavarti et al., 2005).

ls and methods

als

el PH 102 grade of microcrystalline cellulose (MCC)d from FMC Corporation (Newark, DE).

tion

ration of ribbonsof neat Avicel PH 102 were prepared using anerk WP120V roller compactor (Alexanderwerk Inc.,A) tted with two counter rotating diamond knurl sur-12 cm diameter and 4 cmwide and a single feed screw.produce ribbons with largest possible inter- and intra-density, combinations of the highest and lowest set-able that could still produce a continuous ribbon, i.e., acted strip without breaks, were determined, and valuesis broad range of settings were selected for the exper-ign. To produce more consistent ribbons the roller gapcontrolled via feedback circuits in the roller compactor;ese feedback circuits would damp out the differencesbecause to achieve a consistent gap the FSS is dynam-ted during operation. To avoid this damping out of thee roller gapwasnotxed in this study, i.e., theautomaticdjustment controls were turned off.ct samples for the bulk and NIR-CI porosity measure-ribbons were compacted using a RS of 4, 6 or 8 RPM,4 or 28 RPM, and RP of 4, 5 or 6MPa; these combina-studied using a factorial design with two levels threea center point as shown in Table 1. For the porositynts, a single strip of roller compacted ribbon, at leastgth (about two times of the circumference of the roll)d for each roller compaction setting. The stripwas thenisposablemicrotome blade into smaller 3 cm sectionsfor further analysis. To allow the roller compactor toy state, samples were only cut from strips producedin of continuous operation. Thirteen consecutive sam-ch roller compaction settings which represent a singlehe rolls (circumference of rolls = 38 cm)weremeasuredhe center point roller setting of RS: FSS: RP of 6 RPM:Pa where twenty-six samples were measured which

wo rotations of the rolls; these samples were used topeatability of two consecutive cycles during roller com-

Co., Nodetermvolumgas disically bmeasurun l0.005p

Theribbonter (M1360lates tinternaconver1.1476mediusity ofreportlowing

P% =(

whereand tdetermsampleter poithe metive sathirtee

2.2.3.Of

were c(Malvemode.2400nimagesimatel125mto corrusing aimageusing a

2.2.4.All

softwadata w

R = S B

wherereectgroundabsorb

A = logecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

ss, GA). Weight of Avicel PH 102 powder were pre-prior to the analysis. The pycnometer determines the

he Avicel PH 102 powder by measuring the volume ofd by the powder. The true density is derived automat-e quotient of the sample weight entered and volumeThe number of purges and runs were 10, purge andssures were 19.5psig and the equilibration rate wasin.

elope density of the Avicel PH 102 roller compactedre determined using the GeoPyc 1360 pycnome-eritics Instrument Co., Norcross, GA). The GeoPyc

ometer measures the envelope volume and calcu-velope density of the roller compacted ribbons. Themeter of the chamber used, consolidation force andfactor selected in this study was 38.1mm, 90N and/mm, respectively. DryFlo is used as the dry-uidrecommendedby themanufacturer. Using the trueden-vicel PH 102 powder, the GeoPyc 1360 pycnometercentage porosity of roller compacted ribbons using fol-ation:

envelopetrue

) 100

the ribbon porosity, envelope is the envelope densitythe true density. The single point bulk pycnometermean porosity are the average of thirteen consecutivem each roller compaction settings except for the cen-ller setting of RS:FSS:RP of 6 RPM:24 RPM:5MPa whereorosity are the average porosity of twenty-six consecu-s. The standard deviations are calculated based on thesetwenty-six consecutive samples.

hemical imaging measurementsIR-CI measurements on the roller compacted ribbonstedwith aMalvern SyNIRgiTM chemical imaging systemstruments, Westborough, MA) in diffuse reectancesets of each ribbon were collected from 1300 to

ith a spectral point spacing of 10nm and 16 co-addedwavelength. The sampleeld of view (FOV)was approx-m32mm corresponding to amagnication of about

el (an image is 320256pixels).Dark imagecubesusedor stray light and detector dark current were recorded-diffuse reectancematerial (amirror), and backgrounds used to determine sample absorbance were recorded-diffuse reectance target (a white ceramic plate).

I data analysisI data were analyzed using ISys 5.0 chemical imagingalvern Instruments,Westborough,MA). Reectance (R)nerated using the following equation:

the reectance of a sample image cube, D is theof a dark image cube and B is the reectance of a back-ge cube. The resulting sample data were converted to(A) using the following equation:

1R

)

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages84 H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx

Fig. 1. Mean %102 prepared uof 6:14:5 whe

Non-samplthe image peach of thepoint baseltion, the goeffects,whiall image culter (lterture for MConly of neavariability acharacteriztional to thintensity dithe porositear regressithe Avicel r2100nm.

The resufrom each rinter- andfrom differepattern on ttral were suderivativesical variatiothe wafe pvariation.

3. Results

3.1.1. Poros

Using thdata analysbulk pycnotion resultsfactors, RS,are indicatecle, the centadjacent tecompaction

As show4MPa to 6

to the higher RP exerted on the powder. An exception to thistrend was seen for RS of 8 RPM and FSS of 20 RPM where increas-ing RP from 4MPa (porosity =45.53%) to 6MPa (porosity =47.77%)actually increased the ribbon porosity by 2.24%. These settings

0FSSt por, theteriarran, themeapowntraduriped aimeimeas witbonsM anor FSch ingemo berositesea). WhPM),of 4Mlessed ted, hhasher

Pa ant thee entsumme ex0RPherh the)whrositFSS ibon0 RPporosity and corresponding standard deviation (SD) of Avicel PHsing different RS, FSS and RP (n=13 for each setting except RS:FSS:RP

re n=26). Values determined using helium pycnometry.

e areas were masked to eliminate non-sample data andlane at 1310nm of each sample was subtracted fromdata cubes. This is equivalent to performing a singleine subtraction where there is no spectral contribu-al of which is to minimize scattering and other baselinelepreservingoverall spectral intensity. The spectra frombes were smoothed using a SavitskyGolay smoothinglength 9, lter order 3). As there is a strong spectral fea-C at 2100nm, and the roller compacted ribbons consistt Avicel PH 102, no spectral changes due to chemicalre expected, the resulting linear shift in the baselinee by the mean intensity at this wavelength is propor-e amount of Avicel PH 102 present. Hence, the meanfference within and between samples is correlated toy of the roller compacted ribbons using a simple lin-on. As described in Section 1, the denser (less porous)ibbon is, the greater the spectral absorbance will be at

lting color mapping with intensity scaling at 2100nmoller compaction settings were compared to study theintra- porosity variation of roller compacted ribbonsnt roller compaction settings. To conrm if the wafehe intensitymapwas due to density variation, the spec-bjected to second derivative pre-processing as secondis known to minimize baseline effects caused by phys-n in the sample. The wafe pattern will disappear if

(8RS-2highes8 RPMfor macle rea20 RPMwhichof themore eregionentrapdwell tdwell tribbon

Ribof 4 RP28.3% fRSwhirearrancan alsthe poother r2005bRS (8 Rfor RPdue todeliverincreasthe FSSRPM) wof 4MRPM. Aand th

Towith thtings (2RP. Higity wit(4MPabonpoas thethe ribfrom 2this article in press as: Lim, H., et al., Assessment of the critical factors afof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Ph

attern shown in the intensity map is a result of density

and discussion

ity of ribbons

e DOE described above and shown in Table 1 and theis techniques described in Section 2.2.2, the single pointmeter determined mean porosity and standard devia-for the ribbons are shown in Fig. 1. In Fig. 1 the threeFSS and RP are each plotted as an axis and the two levelsd by the text adjacent to the porosity values in the cir-er point also contains the experiment conditions in thext. The mean porosity of ribbons from different rollersettings ranged from 26% to 48%.n in Fig. 1, the porosity decreased as RP increased fromMPa at different RS and FSS settings as expected due

3.2. NIR che

Fig. 2(a)intensity scity variatiovariation wcal variatio2100nm ditive pre-prominimize ting the che

To detera calibratiopoint bulkribbons (n=absorbancevalue of theare shownfecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

-4RPand8RS-20FSS-6RP) alsoproduce ribbonswith theosity values for all the settings studied. At higher RS ofrollers rotate faster, reducing compaction dwell timel under pressure which in turn, reduces time for parti-gement and bonding. In addition, at the slower FSS offeed delivers less powder into the compaction region,

ns there is a lower bulk density and higher porosityder as it enters the compaction region, which enablespped air to remain inside the powder in the compactionng roller compaction. Combined with the higher RS, their has a more difculty leaving the powder mass as theis reduced; hence, the faster RSwith shorter compactionndslowerFSSwitha slowerpowderdelivery rateyieldsh the highest porosity among all settings studied.with the lowest porosity value were produced with RSd RP of 6MPa where the porosity value are 26.9% andS of 20 RPMand 28 RPM, respectively, due to the slowerturn, allows a longer compactiondwell time for particleent and bonding under the higher RP. The effect of RSseen when it is increased from 4 RPM to 8 RPM wherey of the ribbons are increased which is consistent withrcher due to lower compaction dwell time (Gupta et al.,en FSS is increased from 20 RPM to 28 RPM at the highthe porosity of the ribbons decreased by 8.8% and 14.4%Pa and 6MPa, respectively. The authors believe this ispermeation of entrapped air occurs when powder is

o the compaction region at a higher rate when FSS isence, decreased the porosity of the ribbons. However,minimal effect on the ribbons porosity at lower RS (4e the ribbons porosity increased by 1.6% and 1.4% for RPd 6MPa when the FSS is increased from 20 RPM to 28lower RS, the longer dwell time effect is predominantrapped air becomes less affecting.arize, ribbon porosity decreases as the RP increased

ception of settings at high RS (8 RPM) with low FSS set-M)where ribbonporositywas increasedwith increasedRS was found to produce ribbons with higher poros-exception of high FSS (28 RPM) with low RP settings

ere increasingRSdoes not signicantly affecting the rib-y. At higher RS (8RPM), the ribbonporositywas reduceds increased from 20 RPM to 28 RPM. On the other hand,porosity was slightly increased as the FSS is increasedM to 28 RPM at lower RS (4 RPM).

mical imaging

is the color mapping of a roller compacted ribbon withaling at 2100nm, which enables us to observe poros-n across the roller compacted ribbons. The porosityithin the Avicel PH 102 ribbons are due to physi-n rather than chemical differences. Image contrast atsappeared (Fig. 2(b)) when subjecting to second deriva-cessing, where this pre-processingmethod is known tohe impact of physical characterization while highlight-mical differences.mine the feasibility of NIR-CI to predict ribbon porosity,n curve was created by plotting the measured singlepycnometer porosity values of the roller compacted100) against the NIR-CI mean absorbance. The meanis calculated by averaging the absorbance intensitybulk roller compacted ribbon at 2100nm. The resultsin Fig. 3(a), along with a simple linear regression t

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages8H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx 5

Fig. 2. (a) Color mapping of a neat Avicel PH 102 ribbon. The colors of this image are based on absorbance at 2100nm. (b) The image contrast disappeared when subjectingto second derivative pre-processing as it represents the porosity variation instead of chemical differences. (For interpretation of the references to color in this gure legend,the reader is referred to the web version of this article.)

yielding a calibration curve:

P = 280.27A + 136.99 (R2 = 0.9044)

where P is the porosity of the roller compacted ribbons and A is themean absorbance at 2100nm. The model was tested by predictingporosity resof ribbons (calibration,of 0.9258, sity of the ronot at, as trolls. The poincreasingity (higherincident NIRature wherpath length

In additiribbons, NIRwhich can bsystemmagresentativeribbon showprised of a snegligible ware appliedintra ribbon

condently be attributed to physical (porosity) rather than otherfactors which could cause spectral differences. Fig. 4 shows animagewhose contrast is based on the peak height (NIR absorbance)of the 2100nm spectral band. Red areas have the lowest porosity(highest density), blue areas correspond to highest porosity (lowestdensity), and green areas are of intermediate porosity and density.

s frosorbathea ismurfaare

anceribbouardgionpectrin Fim ism frn in

ancesisteed uchanet alteriz.

Fig. 3. (a) Sim0.9258.ults from NIR absorbance data on an independent setn=30) whose values were not used in determining thesee Fig. 3(b). The prediction yielded a strong correlationuggesting that NIR-CI can be used to predict the poros-ller compacted ribbon even when the ribbon surface ishese ribbonswere allmade fromdiamond knurl surfacerosity and NIR absorbance are highly correlated whereNIR absorbance was observed with decreasing poros-density) as there is more material interacting with theillumination. This nding is consistencewith the liter-

e density can affect absorbance and diffuse reectance(Tabasi et al., 2008).on to predicting the bulk porosity of roller compacted-CI inherently provides spatially resolved information,e used to show porosity variation on the order of thenication (125m/pixel). As illustrated in Fig. 4, rep-spectra from different spatial locations within a singlepeak height variations. Because the ribbon is com-

ingle chemical species, the spectral differences becomehen pre-processing steps that remove physical effects, the data can be used to predict ribbon porosity anddifferences (see below), these spectral differences can

Regionest abdue totial areroller ssurfaceabsorbof theedge gthis repixel sshown2100nspectruis showabsorbare conobservtabletsTabasicharacribbonthis article in press as: Lim, H., et al., Assessment of the critical factors affof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Ph

ple linear regression model used to correlate the mean intensity at 2100nm with the labm within the diamond-shaped areas show the high-nce at 2100nm (highest density or lowest porosity)diamond knurl surface of the rollers where this spa-ore tightly compacted than the depressed areas on the

ce surrounding it. Spectra from these depressed rolleras surrounding the diamonds should show lower NIR(higher porosity), and this is reected in Fig. 4. The edgen is expected to have the highest porosity as the Teons were not in place allowing material to leaks out fromduring the compaction process. Representative singlea corresponding to the three distinct spatial regions areg. 4. The red spectra with the highest NIR absorbance atfrom the raised diamond area on the roller surface, theom a surrounding depressed area of the roller surfacegreen, and an edge spectrum is in blue. The NIR-CIscans showed spectral shift as the porosity changesnt with the literature where spectral shifts were alsosing single point NIRS as the crushing strengths of theges as a result of densitydifferences (Cantor et al., 2011;., 2008; Tatavarti et al., 2005). NIR-CI is therefore able toe differences in porosity as a function of position on theecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

measured porosity. (b) Prediction model yielding R-squared value of

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages86 H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx

Fig. 4. The color mapping of the roller compacted ribbon at 2100nm where red spectrum corresponding to the regions within the raised diamond area, green spectrumcorresponding to the regions within the depressed area and blue spectrum corresponding to the regions around the edge. (For interpretation of the references to color in thisgure legend, the reader is referred to the web version of this article.)

To assess the porosity variation across the roller compacted rib-bon width, and also between the roller surface raised diamondareas and depressed areas, three samples with known bulk aver-age porosity (26.35%, 36.56% and 50.27%) were compared. The lineprojection plot (see Fig. 5(a)) was used to show the ribbons inten-sity with respect to the location where the blue line was drawn onthe intensity images (see Fig. 5(b)). Both the line projection plotand intensity images show porosity variation across each ribbonswidth, the center showing higher intensity than the edges as seenby the slight curvature of the line plots, suggesting lower porosityvalues at the central regions. More variations are found betweenthe raised diamond area and depressed area, seen as the high fre-quency changes in the plots. The rst sample shown in Fig. 5(b)showed highest intensity (mean absorbance=0.3947) correspond-ing to the lthird samplcorrespondThe porositregions is dmore matethe raised dlower poros

The edges of the ribbons showed highest porosity due to powderleakage.

To assess the axial porosity variation of the roller compactedribbons, individual sample producedby each roller compaction set-tingwere concatenated (digitally stitched together) to examine theporosity variation across thewidth and along the length of an entireroller compacted ribbon and between different roller compactionsettings, the individual 3 cm images of the ribbon were concate-nated to produce a single image of the entire ribbon. The intensitymap shown in Fig. 6 shows sinusoidal variation in intensities alongthe roller compacted ribbon among all settings studied, indicativeof a corresponding sinusoidal variation in the porosity. This nd-ing is consistent with other researchers (Guigon and Simon, 2003;Simon and Guigon, 2003) where sinusoidal density variation was

in a cthe dicityofpow

t igve mspac

Fig. 5. (a) Linewith known mmean absorbaowest measured bulk porosity (26.35%), whereas theewith the lowest intensity (mean absorbance=0.3285)ing to the highest measured bulk porosity (50.27%).y variation between the raised diamond and depressedue to the diamond knurl surface rolls where it allowsrial to deposit per volume on the depressed area thaniamond area. As a result, the diamond regions showedity (or higher density) comparing to thedepressed area.

foundsuringperiodmotiondeliverthe lasnot moin thethis article in press as: Lim, H., et al., Assessment of the critical factors afof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Ph

projection plot shows porosity variation across the sample width, and between the raiseeasured porosity at 2100nm. The measured bulk porosity for the sample on the left, cence are 0.3947, 0.3674 and 0.3285, respectively.ompacted sodium chloride ribbon, determined bymea-istribution of light transmitted through the ribbon. Theof variation is due to the periodicity of the FSS and thethe last ight of the spiral feed screw as it rotates toder to the compaction region. The powder located at

ht of feed screw will get densied and the powder doesuch if it has higher porosity than the powder located

e between the rolls, whereas if the powder at the lastfecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

d diamonds and depressed area. (b) Intensity images of three samplesnter and right are 26.35%, 36.56% and 50.27%, respectively, and the

Please citefeasibility

ARTICLE IN PRESSGModelIJP-11709; No.of Pages8H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx 7

ared b

Fig. 7. Intensi

ight of scspace betwSimon and Gcation provariation alshows increthe same FSto study thewidth andpaction as swere identirepeatable.

4. Conclus

In this s(RS, FSS anstudyshowthe reductiorangementribbon porobon porositRS, the comobserved thhigher prestion, the stutool to deteused to invalong the leFig. 6. Intensity map of roller compacted ribbon prepthis article in press as: Lim, H., et al., Assessment of the critical factors affof using NIR chemical imaging to evaluate the porosity distribution. Int. J. Ph

ty map of roller compacted ribbon prepared by two continuous roller compaction cycle w

rew has equal porosity as the powder located in theeen the rolls, it will move (Guigon and Simon, 2003;uigon, 2003). Thus, the non-continuous pulsing densi-cess from the feed screw results in sinusoidal porosityong the roller-compacted ribbons. In fact, the NIR-CIased sinusoidal periodicity as the FSS is increased andS show same periodicity. In addition, NIR-CI allowed usrepeatability andvariationacross andalong the ribbonslength made by two consecutive cycles of roller com-een in Fig. 7 where the periodicity porosity variationcal for the two continuous cycle showing the process is

ions

tudy, the effects of various roller compaction settingsd RP) on the ribbon porosity were investigated. Thethat ribbonporosity increases as theRS increaseddue ton of compaction dwell time for particles undergo rear-and bonding. FSS has negative effect in changing thesity at higher RS where increased FSS will decrease rib-ywhile it hasminor positive effect at lower RS. At lowerpaction dwell time and RP predominant. It was alsoat increased RP will decrease ribbon porosity, becausesures are exerted on the powder by the rollers. In addi-dy demonstrated the use of NIR-CI as a nondestructivermine the ribbon porosity. The technique can also beestigate the porosity distribution across the width andngth of the ribbons.

Acknowled

The authInc. and Maand advice.

References

Bacher, C.,Olsethe compa342, 115

Bacher, C., OlsmogeneityPharm. 34

Cantor, S., Hoaspectroscosystem fo9580-z.

Daugherity, P.ribbon thi

Ende,M.T.A.,MR.J., 2007.through ro

FDA, 2006. IQ9., httpInformatio

Feng, T., WangvariabilityFourier tra

Gamble, J.F., Toan in-gapule ow an

Ghorab, M.K.,sivity as aroller comy various RS, FSS and RP.ecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

ith RS, FSS and RP settings of 6 RPM, 24 RPM and 5MPa, respectively.

gements

ors gratefully acknowledge Nick Hayes of Volkmann,lvern Instrument, Inc., and the FDA for their support

n, P.M., Bertelsen, P., Kristensen, J., Sonnergaard, J.M., 2007. Improvingction properties of roller compacted calcium carbonate. Int. J. Pharm.123.en, P.M., Bertelsen, P., Sonnergaard, J.M., 2008. Granule fraction inho-of calcium carbonate/sorbitol in roller compacted granules. Int. J.

9, 1923.g, S.W., Augsburger, L.L., D. Ellison, C., Khan,M.A., Lyon, R.C., 2011. NIRpy applications in the development of a compacted multiparticulater modied release. Pharm. Sci. Technol., doi:10.1208/s12249-010-

D., Chu, J.H., 2007. Investigation of serrated roll surface differences onckness during roller compaction. Pharm. Dev. Technol. 12, 603608.oses, S.K., Carella, A.J., Gadkari, R.A., Graul, T.W., Otano, A.L., Timpano,Improving the content uniformity of a low-dose tablet formulationller compaction optimization. Pharm. Dev. Technol. 12, 391404.CH Harmonised Tripartite Guideline: Quality Risk Management://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatory-n/Guidances/ucm073511.pdf., F., Pinal, R., Wassgren, C., Carvajal, M.T., 2008. Investigation of theof NIR in-line monitoring of roller compaction process by using Fastnsform (FFT) analysis. AAPS Pharm. Sci. Technol. 9, 419424.byn,M., Dennis, A.B., Shah, T., 2010. Roller compaction: application ofribbon porosity calculation for the optimization of downstream gran-d compactability characteristics. Pharm. Dev. Technol. 15, 223229.Chatlapalli, R., Hasan, S., Nagi, A., 2007. Application of thermal effu-process analytical technology tool for monitoring and control of thepaction process. AAPS Pharm. Sci. Technol. 8, 7.

Please cite rs affeasibility t. J. Ph

ARTICLE IN PRESSGModelIJP-11709; No.of Pages88 H. Lim et al. / International Journal of Pharmaceutics xxx (2011) xxxxxx

Guigon, P., Simon, O., 2003. Roll press design - inuence of force feed systems oncompaction. Powder Technology. 130, 4148.

Gupta, A., Peck, G.E., Miller, R.W., Morris, K.R., 2004. Nondestructive measure-ments of the compact strength and the particle-size distribution after millingof roller compacted powders by near-infrared spectroscopy. J. Pharm. Sci. 93,10471053.

Gupta, A., Peck, G.E., Miller, R.W., Morris, K.R., 2005a. Effect of the variation in theambient moisture on the compaction behavior of powder undergoing roller-compaction and on the characteristics of tablets produced from the post-milledgranules. J. Pharm. Sci. 94, 23142326.

Gupta, A., Peck, G.E., Miller, R.W., Morris, K.R., 2005b. Inuence of ambient moistureon the compaction behavior of microcrystalline cellulose powder undergoinguni-axial compression and roller-compaction: a comparative study using near-infrared spectroscopy. J. Pharm. Sci. 94, 23012313.

Gupta, A., Peck, G.E., Miller, R.W., Morris, K.R., 2005c. Real-time near-infraredmonitoring of content uniformity, moisture content, compact density, tensilestrength, and Youngs modulus of roller compacted powder blends. J. Pharm.Sci. 94, 15891597.

Lewis, E.N., Kidder, L.H., Lee, E., 2006. NIR chemical imaging as process analyticaltool. Innovations Pharm. Technol., 107111.

Miguelez-Moran, A.M., Wu, C.Y., Dong, H.S., Seville, J.P.K., 2009. Characterisation ofdensity distributions in roller-compacted ribbons using micro-indentation andX-ray micro-computed tomography. Eur. J. Pharm. Biopharm. 72, 173182.

Miguelez-Moran, A.M., Wu, C.Y., Seville, J.P.K., 2008. The effect of lubrication ondensity distributions of roller compacted ribbons. Int. J. Pharm. 362, 5259.

Peck, G.E., Soh, J.L.P., Morris, K., 2008. Dry granulation. In: Augsburger, L.L., Hoag,S.W. (Eds.), PharmaceuticalDosage Forms: Tablets. InformaHealthcareUSA, Inc.,New York, pp. 303336.

Sheskey, P.J., Hendren, J., 1999. The effects of roll compaction equipment variables,granulation technique, and HPMC polymer level on a controlled-release matrixmodel drug formulation. Pharm. Technol. 23, 90106.

Simon, O., Guigon, P., 2003. Correlation between powder-packing properties androll press compact heterogeneity. Powder Technology. 130, 257264.

Tabasi, S.H., Fahmy, R., Bensley, D., OBrien, C., Hoag, S.W., 2008. Quality by design.Part I. Application of NIR spectroscopy tomonitor tabletmanufacturing process.J. Pharm. Sci. 97, 40404051.

Tatavarti, A.S., Fahmy, R.,Wu, H., Hussain, A.S.,Marnane,W., Bensley, D., Hollenbeck,G., Hoag, S.W., 2005. Assessment ofNIR spectroscopy fornondestructive analysisof physical and chemical attributes of sulfamethazine bolus dosage forms. AAPSPharm. Sci. Technol. 06, E91E99.

Tye, C.K., Sun, C.C., Amidon, G.E., 2005. Evaluation of the effects of tableting speedon the relationships between compaction pressure, tablet tensile strength, andtablet solid fraction. J. Pharm. Sci. 94, 465472.

Wu, C.Y., Best, S.M., Bentham, A.C., Hancock, B.C., Boneld, W., 2006. Predicting thetensile strength of compacted multi-component mixtures of pharmaceuticalpowders. Pharm. Res. 23, 18981905.this article in press as: Lim, H., et al., Assessment of the critical factoof using NIR chemical imaging to evaluate the porosity distribution. Infecting the porosity of roller compacted ribbons and thearm. (2011), doi:10.1016/j.ijpharm.2011.02.028

Assessment of the critical factors affecting the porosity of roller compacted ribbons and the feasibility of using NIR che...IntroductionMaterials and methodsMaterialsPreparationPreparation of ribbonsPorosity measurementsNIR chemical imaging measurementsNIR-CI data analysis

Results and discussionPorosity of ribbonsNIR chemical imaging

ConclusionsAcknowledgementsReferences