Jari T.T. Leskinen Acoustic Techniques for · 2014. 1. 8. · JARI T. T. LESKINEN Acoustic...

Publications of the University of Eastern Finland Dissertations in Forestry and Natural Sciences

Publications of the University of Eastern Finland

Dissertations in Forestry and Natural Sciences

isbn 978-952-61-1169-8

Jari T.T. Leskinen

Acoustic Techniques for Pharmaceutical Process MonitoringMeasurements in Tablet Manufacturing and Quality Control

The tablet is probably the most

common solid dosage form for

orally administered drugs. In this

thesis, acoustic techniques were

tested for pharmaceutical process

monitoring and tablet quality control

purposes. An acoustic emission

method was found to be suitable for

real-time particle size estimation in

a granulation process. Ultrasound

(US) methods were found to be

good for real-time monitoring of the

tabletting, as well as detecting the

integrity of the tablet. Additionally, a

developed US technique was capable

for determining the formed gel layer

thickness on immersed tablets.

dissertatio

ns | 112 | Ja

ri T

.T. L

eskin

en | A

coustic T

echniques for P

harm

aceutical Process M

onitoring – M

easurements in T

ablet...

Jari T.T. LeskinenAcoustic Techniques for Pharmaceutical Process

MonitoringMeasurements in Tablet Manufacturing

and Quality Control

JARI T. T. LESKINEN

Acoustic Techniques forPharmaceutical Process


and Quality Control

Publications of the University of Eastern FinlandDissertations in Forestry and Natural Sciences

No 112

Academic DissertationTo be presented by permission of the Faculty of Science and Forestry for publicexamination in the Auditorium L22 in Snellmania Building at the University of

Eastern Finland, Kuopio, on August, 7, 2013,at 12 o’clock noon.

Department of Applied Physics

Kopijyvä Oy

Kuopio, 2013

Editor: Prof. Pertti Pasanen, Prof. Kai-Erik Peiponen,

Prof. Matti Vornanen, Prof. Pekka Kilpeläinen

Distribution:

University of Eastern Finland Library / Sales of publications

P.O. Box 107, FI-80101 Joensuu, Finland

tel. +358-50-3058396

http://www.uef.fi/kirjasto

ISBN: 978-952-61-1169-8 (printed)

ISSN: 1798-5668

ISSNL: 1798-5668

ISBN: 978-952-61-1170-4 (pdf)

ISSN: 1798-5676

Author’s address: University of Eastern FinlandDepartment of Applied PhysicsP.O.Box 1627FI-70211 KuopioFINLANDemail: [email protected]

Supervisors: Professor Reijo Lappalainen, Ph.D.University of Eastern FinlandDepartment of Applied PhysicsKuopioFINLAND

Professor Jarkko Ketolainen, Ph.D. (Pharm.)University of Eastern FinlandSchool of PharmacyKuopioFINLAND

Mikko Hakulinen, Ph.D.Kuopio University HospitalDepartment of Clinical Physiologyand Nuclear MedicineImaging CenterKuopioFINLAND

Reviewers: Associate Professor Göran Frenning, Ph.D.Uppsala UniversityDepartment of PharmacyUppsalaSWEDEN

Adjunct Professor Simo Saarakkala, Ph.D.University of OuluDepartment of Medical TechnologyInstitute of BiomedicineOuluFINLAND

Opponent: Professor Michiel Postema, Ph.D.University of BergenDepartment of Physics and TechnologyBergenNORWAY

ABSTRACT

Pharmaceutical manufacturing has traditionally been considered asseveral consequent unit processes. Each unit is processed by opera-tors having experience they have obtained with trial-and-error andqualitative or quantitative measurements. This type of training maynot always help operators reach the optimal goal of each unit pro-cess. Therefore, development for better unit operation monitoringtools is needed.

This work contains several experimental studies on monitoringpharmaceutical manufacturing unit processes and theoretical andnumerical analysis of the obtained results. Acoustic emission spec-troscopy (AES) was used for granulation particle size studies, par-allel to near infrared (NIR) spectroscopy and digital camera flashtopography (TOPO) during fluidized bed granulation process. Ul-trasound (US) based applications for tabletting process monitoringand post-compaction tablet defect determination were introduced.Additionally, a developed US window method for immersed poly-mer tablet swelling process monitoring and gel layer thickness mea-surements was presented.

The mean granule size was quantitatively measured during flu-idization and granulation processes was monitored both qualita-tively and quantitatively. The particle sizing methods proved tobe accurate as the relative root-mean-square (RMS) error of AEand TOPO method was 6.7 and 14.4 %, respectively. Pharmaceu-tical tablet mechanics were studied during tablet compression. Themeasurement system was tested in an actual manufacturing envi-ronment and found to be capable of measuring the US responseof the tabletting process from bulk to tablet. Manufactured tabletswere tested for quality control in order to determine mechanicalintegrity with the US technique. Each tablet was measured and94.5 % were correctly identified as intact or defected.

The tested US techniques proved to be promising tools for phar-maceutical tablet manufacturing and quality control unit opera-tions. It is concluded that the acoustic techniques presented in this

study could be useful for the pharmaceutical industry to optimizeprocesses and minimize the variation of the end product quality.

Universal Decimal Classification: 534-8, 534.6, 543.422.3-74, 615.453.6

National Library of Medicine Classification: QV 778, QV 786.5.T3

Library of Congress Subject Headings: Acoustic emission; Acoustic spec-troscopy; Ultrasonics; Ultrasonic waves; Pharmaceutical technology; Tablets(Medicine); Tableting; Granulation; Particle size determination; Near in-frared spectroscopy; Quality control

Yleinen suomalainen asiasanasto: akustiikka; ultraääni; farmasian teknolo-gia; tabletit; rakeistus; topografia; spektroskopia; laadunvalvonta

To Tiina, Juska, Jerri and Jonni

Acknowledgements

This thesis summarizes the studies carried out in the Department ofApplied Physics and at the School of Pharmacy of University at theEastern Finland (formerly known as University of Kuopio) duringthe years 2006–2012.

I express my deepest gratitude to my supervisor Professor ReijoLappalainen, Ph.D., for his supervision and enthusiastic attitudetowards science, and I also want to thank him for giving me an earlyopportunity to work in his laboratory as a young student since thesummer of 2000.

I am very grateful to Professor Jarkko Ketolainen, Pharm.D., forsupervision and scientific discussions in the field of pharmaceuticaltechnology. This thesis is published due to this opportunity givento me in order to study pharmaceutical processes with you since2005.

I also owe thanks to Adjunct Professor Mikko Hakulinen, Ph.D.,who has earned my very deep gratitude for guiding me during thestudy, giving me helpful comments and answering all my ques-tions; both small and big ones. He has offered me a lot of positivesupport and attitude during the years, in addition to his supervi-sion.

All my co-authors, Ph.D. Simo-Pekka Simonaho , M.Sc. Matti-Antero Okkonen, M.Sc Maunu Toiviainen, Ph.D. Sami Poutiainen,Ph.D. Mari Tenhunen, Ph.D. Pekka Teppola, Professor Kristiina Järvi-nen, Pharm.D., M.Sc. Marko Kuosmanen, Ph.D. Susanna Abrahmsén-Alami, are acknowledged for their valuable scientific contributions.

I extend my sincere thanks to Associate Professor Göran Fren-ning, Ph.D., and Adjunct Professor Simo Saarakkala, Ph.D., for theirreview of the thesis and for giving their constructive comments forits improvement. I also thank James Fick, Ph.D., for the linguisticreview of this thesis.

I want to thank M.Sc. Maiju Järvinen from School of Pharmacy

for practical work during laboratory granulations. B.Sc. TuomoSilvast is thanked for his help with microCT imaging and 3D-re-construction. I want to thank M.Sc. Jarkko Leskinen and Ph.D.Mikko Laasanen for endless help with the practical things con-cerning acoustic and ultrasonic research. Adjunct Professor OssiKorhonen, Pharm.D., is thanked for the scientific discussions andcomputational fluid dynamics simulations. I want to acknowledgeB.Sc. Päivi Tiihonen for help and guidance during powder materialexamination. B.Sc. Matti Timonen and Mr. Olli-Matti Hanhinenare acknowledged for their helpful work with LabView program-ming. The help of M.Sc. Heikki Hyvärinen from Waltti ElectronicsLtd., Kuopio, Finland in solving technical solutions with the tablet-ting machine is greatly appreciated. The help of Professor JukkaJurvelin, Ph.D., offering his research groups laboratory facilities forthe ultrasound measurements is gratefully appreciated.

I gratefully acknowledge the PROMIS Centre consortium, whichis funded by the Finnish Funding Agency for Technology and In-novation, TEKES, ERDF and State Provincial Office of Eastern Fin-land, for providing excellent research facilities for the work thathas been made in VARMA, ORPAT, PATKIVA, PROMET and PRO-TONS projects. AstraZeneca is acknowledged with gratitude forfunding a part of the work. The National Doctoral Programme ofMusculoskeletal Disorders and Biomaterials (TBDP) and The NorthSavo Regional Fund of the Finnish Foundation of Culture are ap-preciated for the financial support during the studies.

I am most grateful to all colleagues in the School of Pharmacy,the Department of Applied Physics and SIB Labs for the pleas-ant working atmosphere. Special thanks to Ph.D. Markku Tiittaand Ph.D. Laura Tomppo. Especially, Ph.D. Arto Koistinen andM.Sc. Mikko Selenius are thanked for enlightening conversationsand friendship during these years. I am also thankful to Ritva Sor-munen and Virpi Miettinen for their practical help and laboratoryassistance. Thanks for Mr. Juhani ’Sorsaveden Sulttaani’ Hakala andMr. Jukka Laakkonen for invaluable help with his technical skills,but also for the true stories. I also appreciate the work done by the

personnel in the offices of the departments and offering the helpwhenever needed. I would also like to thank all of my nearestfriends for giving me something else than science to think about.

My warmest thanks go to ’Queen, Mermaid and the Catwomanpacked as one’ a.k.a. my wife Tiina and our three marvellous sonsJuska, Jerri and Jonni being always sincerely interested in every-thing and showing the power of unlimited imagination. Thanks forbeing there for me.

Lopuksi haluan kiittää vanhempiani Leenaa ja Seppoa kaikestatuesta ja kannustuksesta näiden vuosien aikana.

Kuopio 9 July, 2013 Jari Leskinen

LIST OF PUBLICATIONS

This thesis consists of the present review of the author’s work inthe field of applied physics and pharmaceutical technology and thefollowing selection of the author’s publications referred to in thetext by Roman numerals:

I J. Leskinen, M.-A. Okkonen, M. Toiviainen, S. Poutiainen, M.Tenhunen, P. Teppola, R. Lappalainen, J. Ketolainen, K. Järvi-nen, "Labscale fluidized bed granulator instrumented withnon-invasive process monitoring devices,” Chem. Eng. J. 164,

268–274 (2010).

II J. Leskinen, S.-P. Simonaho, M. Hakulinen, J. Ketolainen, “Real-time Tablet Formation Monitoring with Ultrasound Measure-ments in Eccentric Single Station Tablet Press,” Int. J. Pharm.442, 27–34 (2013).

III J. Leskinen, S.-P. Simonaho, M. Hakulinen, J. Ketolainen, “In-line ultrasound measurement system for detecting tablet in-tegrity,” Int. J. Pharm. 400, 104–113 (2010).

IV J. Leskinen, M. Hakulinen, M. Kuosmanen, J. Ketolainen, S.Abrahmsén-Alami, R. Lappalainen, “Monitoring of swellingof hydrophilic polymer matrix tablets by ultrasound techniques,”Int. J. Pharm. 404, 142–147 (2011).

The original articles have been reproduced with permission of thecopyright holders. The thesis also includes previously unpublisheddata.

AUTHOR’S CONTRIBUTION

The publications selected in this dissertation are original researchpapers on acoustic applications in pharmaceutical tablet manufac-turing and quality testing. The research ideas from the Publicationshave arisen on discussions between the author and co-authors dur-ing the years 2005–2011.

In Publication I the author has carried out all acoustic measure-ments and their off-line analyses. In Publications II–IV the authorhas carried out all numerical calculations and the selection of usedmeasurement methods as well as the numerical development of so-lution methods.

The author has written the manuscript to all the Publicationsexcept a part of Publication I, where the expertise with optical mea-surement methods belonged to Matti-Antero Okkonen and MaunuToiviainen; in all the Publications the co-operation with the co-authors has been significant.

ABBREVIATIONS

A/D analog-to-digitalAE acoustic emissionAPI active pharmaceutical ingredientASTM American Society of Testing and MaterialsCA caffeineCG crystallization and crystal growthDIA digital image analysisDCP dibasic calcium phosphateDFT discrete fourier transformDT destructive testingFB, FBG fluidized bed, FB granulationFDA United States food and drug administrationFTIRi Fourier transform infrared imagingGI gastrointestinalGMP good manufacturing practiceHPMC hydroxypropyl methylcelluloseHS hydrated silicaHSG high shear granulationLM lactose monohydrateMCC microcrystalline celluloseMS magnesium stearateNDT nondestructive testingNIR near infraredNMR nuclear magnetic resonancePAT process analytical technologyPA photoacousticPE pulse echoPEO polyethylene oxidePh.Eur. European PharmacopoeiapH measure of the activity of H+ ions in a solutionPVP polyvinylpyrrolidonePRC paracetamolQbD quality by designRC roller compactionRH relative humiditySPL sound pressure levelT tabletsTC tablet compressionTOPO flash topographic cameraTT through transmissionUS ultrasound

SYMBOLS

A amplitudeα attenuationc speed of soundD diameterE Young’s modulusε strainf frequencyF loadFC crushing forceg signalG Fourier transformΓ discrete Fourier transformG shear modulush thicknessK bulk modulusL, L0 length, length before loadingλ wavelengthn number of samplesN length of the nearfieldν Poisson’s ratioω angular frequencyp pressureP porosityρ densityσ stressσt tensile strength of a tabletS, S0 cross sectional area, cross sectional area before loadingτ period of oscillationt temporal coordinate, timeθi, θt angle of incidence and transmission, respectivelyu displacement of an oscillating particlex spatial coordinate, locationzi, zi particle diameter, particle mean diameterZi acoustic impedance of i

Contents

1 INTRODUCTION 1

2 PHARMACEUTICAL TABLET MANUFACTURING AND

’PAT’ 3

2.1 Process Analytical Technology (PAT) . . . . . . . . . . 32.1.1 Real-time measurements . . . . . . . . . . . . . 5

2.2 Per oral administration route . . . . . . . . . . . . . . 52.3 From fine powder to end user tablet . . . . . . . . . . 6

2.3.1 Granulation and drying . . . . . . . . . . . . . 72.3.2 Monitoring of particle size enlargement process 92.3.3 Tablet compression . . . . . . . . . . . . . . . . 112.3.4 Elasticity . . . . . . . . . . . . . . . . . . . . . . 132.3.5 Mechanical failure of tablets . . . . . . . . . . 13

2.4 Quality of tablets . . . . . . . . . . . . . . . . . . . . . 152.4.1 Mechanical strength testing of tablets . . . . . 152.4.2 Tablet swelling test as a simulator of controlled

release . . . . . . . . . . . . . . . . . . . . . . . 17

3 INTRODUCTION TO ACOUSTICS 21

3.1 Basics of Acoustics . . . . . . . . . . . . . . . . . . . . 213.1.1 Pressure and Wave Motion . . . . . . . . . . . 213.1.2 Speed of Sound . . . . . . . . . . . . . . . . . . 233.1.3 Acoustic Impedance, Transmission and Reflec-

tion . . . . . . . . . . . . . . . . . . . . . . . . . 233.1.4 Attenuation . . . . . . . . . . . . . . . . . . . . 24

3.2 Acoustic measurement techniques . . . . . . . . . . . 253.2.1 Acoustic emission . . . . . . . . . . . . . . . . 253.2.2 Ultrasound . . . . . . . . . . . . . . . . . . . . 293.2.3 Fourier Analysis of Signals . . . . . . . . . . . 303.2.4 Acoustic Transducers . . . . . . . . . . . . . . 31

3.3 Acoustics in Pharmaceutics . . . . . . . . . . . . . . . 33

3.3.1 Passive (AE) Techniques . . . . . . . . . . . . . 343.3.2 Active (US & PA) Techniques . . . . . . . . . . 34

4 AIMS 37

5 MATERIALS AND METHODS 39

5.1 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . 395.1.1 Binder . . . . . . . . . . . . . . . . . . . . . . . 395.1.2 Excipients . . . . . . . . . . . . . . . . . . . . . 395.1.3 Model drugs . . . . . . . . . . . . . . . . . . . 395.1.4 Ready-to-use materials and formulations . . . 40

5.2 Sample preparation . . . . . . . . . . . . . . . . . . . . 405.2.1 Granules (I) . . . . . . . . . . . . . . . . . . . . 415.2.2 Binary tablets (II) . . . . . . . . . . . . . . . . . 415.2.3 Monolithic tablets (III,IV) . . . . . . . . . . . . 42

5.3 Experimental Setups . . . . . . . . . . . . . . . . . . . 435.3.1 Fluidized bed studies (I) . . . . . . . . . . . . . 445.3.2 Tablet formation monitoring (II) . . . . . . . . 505.3.3 Tablet integrity testing (III) . . . . . . . . . . . 535.3.4 US monitoring of swellable matrix tablet fronts

movement (IV) . . . . . . . . . . . . . . . . . . 56

6 RESULTS 61

6.1 Acoustic emission as footprint of moving particles (I) 616.2 Ultrasound measurement during tablet formation (II) 636.3 Ultrasound determination of tablet integrity (III) . . 666.4 Polymer tablet swelling monitoring with ultrasound

echo (IV) . . . . . . . . . . . . . . . . . . . . . . . . . . 68

7 DISCUSSION 71

7.1 Granulation process monitoring togetherwith AE, TOPO and multi-point NIR . . . . . . . . . 71

7.2 US transmission for tablet formation monitoring . . . 747.3 US transmission as a tool for tablet integrity testing . 777.4 US echo as a tool for polymer swelling monitoring . 79

8 SUMMARY AND CONCLUSIONS 83

8.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 838.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 83

BIBLIOGRAPHY 85

1 Introduction

Oral administration is the dominant method of delivering drugs tothe human systemic blood circulation [11] and the tablet is probablythe most common solid dosage form for orally administered drugs.Tablets are popular for many reasons, e.g. they are easy to handleand administer, and their cost per dose is relatively low. Also, byusing industrial tabletting machines, it becomes possible to quicklymanufacture large amounts of tablets. Pharmaceutical tablet man-ufacturing process from stored powders to a dense compact withmultiple components including the active pharmaceutical ingredi-ent can be a very complicated process with numerous possibilitiesfor failure to occur. Capping and lamination are common problemsin pharmaceutical tablet manufacturing.

In the beginning of 2012, "The Truly Staggering Cost Of Invent-ing New Drugs" was represented [61]. For the 10 largest companiesin the pharmaceutical industry, the total cost to develop and ap-prove a new drug to the market would be at least 3.7 billion US dol-lars. A more efficient production chain is imminent due to climbingexpenses of the drug development, because the price should be aslow as possible and reachable to customers. However, new drugsneed to be developed. The old developed drug products may haveunwanted side effects, or they might not be as effective for curingthe specific diseases, that they are intended to target.

To get the production efficiency as high as possible, the vari-ation in quality must be minimized. This can be achieved onlyby automated manufacturing processes. In order to control theseautomated processes, different production steps have to be mon-itored. For comprehensive quality assurance monitoring of soliddosage forms in the pharmaceutical industry, in 2004 the UnitedStates Food and Drug Administration (FDA) initiated a currentlywell-known guidance program entitled the Process Analytical Tech-nology (PAT) [153].

Dissertations in Forestry and Natural Sciences No 112 1

Jari Leskinen: Acoustic Techniques for Pharmaceutical ProcessMonitoring

A system for designing, analyzing, and controlling manufactur-ing through timely measurements with the goal of ensuring finalproduct quality, as PAT was defined [153], it encourages the devel-opment and implementation of new technologies and proceduresin pharmaceutical manufacturing. Therefore, its purpose was toencourage industry to attempt to make improvements in the under-standing and the control of different manufacturing processes. Asa result, better knowledge of processes, higher level of understand-ing would show up in efficacy, safety and higher product reliability.This in turn would lower the costs associated with pharmaceuticalmanufacturing.

The PAT guidelines strongly encourage a "quality by design"(QbD) approach in pharmaceutical research and development work,which requires an increased mechanistic understanding of criticalraw material properties that determine product functionality. Prod-uct testing confirms the product quality. As quality cannot be testedinto products, it should be built-in or should be by design. [154,171]

The aim of the studies presented in this thesis was to developand apply acoustic techniques in small scale pharmaceutical tabletmanufacturing and to monitor manufacturing and quality assur-ance unit operations. The thesis studied the feasibility of used tech-niques in pharmaceutical granulation, tabletting and product qual-ity control. The acoustic methods offer, passive and active, non-destructive PAT tools for granule size measurement with acousticemission and tabletting process monitoring with ultrasound. Ad-ditionally, the ultrasound methods were used for quality controlpurposes, i.e. for tablet integrity determination and for swellingprocess monitoring of hydrophilic polymer tablets.

2 Dissertations in Forestry and Natural Sciences No 112

2 Pharmaceutical TabletManufacturing and ’PAT’

The pharmaceutical industry is a highly regulated industry and allproduction must be carried out in accordance with good manu-facturing practice (GMP). Traditionally, virtually all manufacturingoperations have been carried out batch wise in spite of cost dis-advantages and the fact that in many cases, continuous process-ing could lead to the manufacture of variation-free products. Thepharmaceutical industry is dominated by batch processes so muchthat naturally continuous process such as in-line milling or semi-continuous processes like tablet compression are modified to makethem into batch processes. The history for this is that the regulatoryquality requirements are easier to comply with if the products aremanufactured by batch processes. [110]

In pharmaceutical research and development, there has been aninterest in shifting processing methods from batch to continuousforms during the recent years [67, 99, 108, 110, 166]. However, thebatch-orientated manufacturing techniques have been the commonapproach of doing in the pharmaceutical industry for decades andthe change will not take place instantaneously.

2.1 PROCESS ANALYTICAL TECHNOLOGY (PAT)

In the early 2000’s, the pharmaceutical industry was poor in pro-duction efficiency. It had received a low-end grade in productionefficiency when the high-end grade was reserved by the microchipindustry: The statistical defective percentage was 4.5 and 0.0003 forpharmaceutical and microchip industry, respectively. [29]

The concept of PAT has been introduced to improve our under-standing of the pharmaceutical process and to monitor and control



critical process parameters [153]. PAT stands for Process AnalyticalTechnology and it aims to change the present thinking and opera-tion within the pharmaceutical industry towards real-time processcontrol or monitoring instead of the intermediate or end producttesting off-line [100, 130].

The present idea seems to be that the quality cannot be testedinto products, but it should be designed into processes beforehand.Finally, evaluating not only the final product but the whole pro-duction process leads to a more comprehensive understanding ofthe production chain. The ideal is real-time quality control anda capability of process control throughout the manufacturing pro-cess. Pharmaceutical formulations are complex systems and evennowadays are often developed on the basis of "trial-and-error" ex-periments. A process is well understood if only all critical sourcesof variability are identified and accounted for.Many pharmaceuti-cal processes are poorly understood and their manufacturing per-formance is low. The goal of FDA’s PAT iniative is to achieve sci-entifically based decisions. To design the quality of the productand to ’test-in’ quality by eliminating unwanted items at the end ofproduction does not achieve the desired outcome creating a wasteof time and money. Process monitoring and control strategies areintended to monitor the state of a process and actively manipu-late it to maintain a desired state. Optimization of manufactur-ing processes includes designing a process measurement systemwhich allow real-time or near real-time monitoring of critical at-tributes. [81, 85]

There are many tools available that enable processes to be un-derstood for scientific, risk-managed pharmaceutical development,manufacture, and quality control. The PAT guidance [153] catego-rizes these tools as:

1. Multivariate data acquisition and analysis tools

2. Process analyzers (sensors)

3. Process control tools


Pharmaceutical Tablet Manufacturing and ’PAT’

4. Continuous improvement and knowledge management

2.1.1 Real-time measurements

The physical state of a process can be monitored, for example bymeasuring its variables, such as temperature. However, the stateof a process is usually determined by several factors and there canalso be interactions between the variables. Based on the FDA’s PATinitiative [153], the following real-time measurements can be per-formed:

At-line The sample is removed and analyzed in close proximity tothe process stream.

On-line The sample is diverted from the manufacturing process,and may be returned to the process stream.

In-line The sample is not removed from the process stream andcan be invasive or noninvasive.

The PAT initiative boosted the installation of additional in-processcontrol units in the manufacturing departments for optimizing thequality of pharmaceuticals. Several European pharmaceutical com-panies have introduced at-line, on-line or in-line near-infrared (NIR)spectroscopy control tools for nearly all process steps such as rawmaterial identification, blending, drying and tabletting [29, 93].

2.2 PER ORAL ADMINISTRATION ROUTE

The per oral route is the simplest, most convenient and safest meansof drug administration. The most popular oral dosage forms aretablets, capsules, suspensions and emulsions. Tablets are preparedby compression and contain drugs and formulation additives in-cluding specific functions such as disintegrants promoting break-up into granules and particles in the gastrointestinal (GI) tract. Thisfacilitates drug dissolution and absorption. [16]

There is an equilibrium between bioavailability of the product,its chemical and physical stability and the technical feasibility of



producing it. This must be taken into account while formulatingthe pharmaceutical dosage form. Virtually all solid dosage formsare manufactured from powders [16]. The understanding of theunique properties of powders is necessary for rational formulationand manufacture. Requirements for the most common solid dosageforms, i.e. tablets and capsules, are: the flow of the correct weight ofmaterial into a certain volume, the behavior of the material underpressure and the wetting of the powder. This is particularly criticalfor both granulation and subsequent disintegration and dissolutionof the dosage form. [41]

One of the most practical thing for tablets is that they can beadministered by patients themselves. Therefore, it is likely thattablets and capsules will remain one of the most common usedmethods of delivering drugs to patients in the future.

2.3 FROM FINE POWDER TO END USER TABLET

Practically, all pharmaceutical products contain active pharmaceu-tical ingredients (APIs) with a therapeutic effect and excipients, (i.e.the pharmaceutically inactive substances) which are necessary toensure the final dosage form to act as intended. Water, lactose andsugar are typical excipients.

In Fig. 2.1, an example of a manufacturing unit processes formaking pharmaceutical tablets is shown: A period of mixing a com-

Blending Granulation Drying Mixing Tabletting Coating End Product Testing

Figure 2.1: An example pathway from powder to end product by manufacturing unitprocesses. Processes that are shaded grey were studied in this thesis.

position of powders occurs before the wetting stage of the granu-lation process. Increasing moisture content together with mixing



wetted powders gets formulation to agglomerate. When the gran-ules have achieved the wanted properties such as proper granulesize and density, drying is started by removing the moisture fromthe formulation by heated air flow simultaneously. Blending of gran-ules with lubricant fines may be necessary before tabletting. The useof lubrication helps to avoid the jamming the tablet formulation inthe die. The tablets are often coated with some polymer film coat-ing, e.g. in order to be administered as enjoyably as possible. Thetabletted product is then ready to be tested for quality before can bereleased into the market.

2.3.1 Granulation and drying

Before tablets are manufactured, pharmaceutical materials requireprocessing. Granulation improves flow properties and compactioncharacteristics of the mix [16] and reduces the risk of hazards, e.g.explosion [63]. The pharmaceutical granulation process includesseveral processes and their subprocesses. In this thesis, the gran-ulation was considered as three stages: mixing, wet granulation,i.e. agglomeration and drying. During the wet granulation stagejust after premixing of the powder formulation, the moisture orwater content should be increased during agglomeration starting1–2 %(w/w) of dry stored fine powders to over 10 %(w/w) of wetpowder bed. The drying stage is started after adequate particlesize is obtained by mixing of powder bed into granules. The opti-mal endpoint is always a compromise between elapsed time, mois-ture and granule size/breakage. There are four main types of wet-agitated granulator types: 1) drum, 2) pan, 3) mixer and 4) fluidizedbed (FB) granulators. Mixer granulators, i.e. high shear granulatorsare used widely in pharmaceutical, detergent and agrochemical in-dustries and they are less sensitive to operating conditions thanother granulator types. [139]

FB technology was established in 1922 for coal gasification [169].Nowadays, FBs are used in various fields of industry for physicalprocesses, e.g. mixing, classifying, drying, coating, granulation (ag-



glomeration), heating and cooling of solids. Chemical FB processesare familiar from gasification, pyrolysis, combustion, water purifi-cation and catalytic or gas–solid reactions. [18, 64, 150, 151]

FB granulation in particular is a very common size enlargementprocess. The moisture content of solids is one of the most importantparticle properties in controlling the FB granulation process [161].Particle size is also strongly influenced by the thermal conditions inthe FB. FB particle size enlargement process as a simplified process,is driven by wet binder addition, whereas the drying of particles isdriven by massive heat flow into the FB chamber.

The fluidization of the particle bed can be hard to control through-out the premixing, agglomeration and drying stages. If the particlebed is of narrow size and density distribution of particles, thenflow occurs uniformly. In the case of a wide particle size distri-bution bed, the small particles usually tend to get too much lift asheavy particles accumulate on the bottom screen of the granula-tor. Eventually, small particles get lifted into the particle filters inthe upper part of the chamber or wet agglomerates jam into thesurfaces and quit mixing. Practically, the FB granulation processrequires maintenance in order to clean the particle filters and peelthe over-wetted powder paste off the granulator wall, between eachfluidization process ’trial’.

Normally, the FB granulation of particles involves different ki-netics such as the formation of seeds, growth breakage and ag-glomeration. The property in combination with continuous prod-uct classification and recycling of particle fractions can lead to self-sustained oscillations of particle size distribution, temperature andconcentration progressions of both the gas and solid phases withinthe FB [128]. Therefore, it is important that one can analyze fluidiz-ing conditions, such as pneumatic behavior, particle growth andwetting as they have an influence on the fluidized bed operationand product performance at the end of production.



2.3.2 Monitoring of particle size enlargement process

Recently, several methods have been successfully utilized for themonitoring of the fluidized bed granulation process such as nearinfrared (NIR) spectroscopy [47, 119–121], Raman spectroscopy [1,60, 76], triboelectric probes [112, 113], imaging techniques [106, 160,162], and acoustic emissions (AE) [28, 58, 91, 92, 115, 149].

However, methods that require an optical pathway are problem-atic if the probe or window becomes clogged by the wet, agitatedpowder as in FB granulation. Technical solutions have been pub-lished for these problems, although, only for particle imaging ap-plications [106, 162]. Parallel responses of three inline techniques,namely focused beam reflectance measurement, a single-point NIRspectroscopy and AE, were reported as being applied to monitor apilot-scale FB granulation process [146]. When compared to a singleprocess analytical technique, simultaneous measurements providedbetter process understanding and reduced the need for precaution-ary system set-ups.

Near infrared (NIR) spectroscopy

Near infrared (NIR) spectroscopy and imaging are fast and nonde-structive analytical techniques that provide chemical and physicalinformation for virtually any matrix [122]. It utilizes the near in-frared region of the electromagnetic spectrum (wavelength λ: 780–2500 nm or wave number: 12821–4000 cm−1) [33]. NIR has verygood specificity for water, and has found widespread use for thisanalysis. If technically feasible, the same spectra used to confirmtablet identify can be re-purposed for the determination of watercontent. [123]

The important molecules for NIR measurements have most of-ten been water (O-H stretch), proteins, carbohydrates, fats, and hy-drocarbon classes including pharmaceuticals. The NIR spectra con-sist of overtones and combination bands of the fundamental molec-ular absorptions of covalently bonded polar groups such as O-H,N-H, S-H and C=O [135].



Multivariate analysis techniques are usually needed to extractthe desired chemical information. Careful development of a setof calibration samples and application of multivariate calibrationtechniques is essential for NIR analytical methods.

Flash topographic camera

One way to obtain the particle size information from a process in-line, is to shoot still images of particles. Images of moving particlescan be obtained, e.g. through a transparent sheet of glass. Whileparticles are gliding on the glass surface, the images can be cap-tured. The principle of this technique is to project a collimated lightpattern into the objects to be measured and the pattern is capturedfrom a different angle. Thus, the surface of the objects modulates

Figure 2.2: A) Illustration of topographic evaluation of particles. B) an example of apattern used for the illumination of particles. C) Illuminated particles. D) Reconstructed3D view of C).

the pattern and the height map can be computed [19] by extract-ing the modulation component. An illustration of this is shown inFig. 2.2.



2.3.3 Tablet compression

The application of a compaction force to a powder requires severalmechanisms to allow the powder bed to compact into a compressedtablet. To push the particles closer together initial compression isneeded and a sequence of deformation by brittle fractures and plas-tic flow of closely packed particles. These processes are needed,usually as a combination, to form the compacted tablet. How-ever, the surface contact and bond strength is not always enoughto withstand the fracture due to the elastic recovery of the mate-rial. [22, 32, 68–71, 105]

The three deformation mechanisms that can occur to particleswithin the powder bed during compression are: elastic deforma-tion, plastic deformation and fragmentation. Elastic deformationis reversible. However, a material with time-dependent propertiescan store elastic energy and may relax only after a period of time orafter ejection from the die. The energy required to cause plastic de-formation or fragmentation cannot be recovered and the structureof the particles changes permanently.

The compression properties of most drug powders alone areextremely poor [16]. A good formulation of tablet material shouldbe plastic, i.e. capable of permanent deformation and should exhibita degree of brittleness. If the API is plastic, the excipient should befragmenting, and vice versa. The pharmaceutical powders can bedivided to three material types by their deformability:

Elastic material Some materials, e.g. paracetamol, are elastic andvery little permanent change is caused by compression. Ifbonding is weak, the tablet will loose its top (capping), or thewhole tablet cracks into distinct layers (lamination).

Plastic material As there is no fracture, no new surfaces are gen-erated during compression. This leads to poorer bonding inthe material. Because the bonding mechanism is time depen-dent (viscoelastic deformation), the increasing the dwell timeat compression will increase strength of the material.



Fragmenting material Fragmenting material particles tend to breakinto smaller particles during compression. The forming ofnew particle surfaces enables more bondings and ultimatelyleads to stronger tablets. Material that predominantly frag-ments should not have an effect on tablet strength, e.g. bylubricant mixing time or dwell time during compression.

The formation of tablets is fundamentally an interparticulate bond-ing process. The assembled particles are bonded together increas-ing the strength of the compacted powder. A powder with a highcompactability forms tablets with a high resistance towards fractur-ing and also does not exhibit a tendency to cap or laminate.

There are several technical challenges that can occur duringtabletting and the most important issues are listed in Table 2.1.

Table 2.1: The most important problems during tabletting [12].

Problem definition

A high variation in tablet weight and doseB low mechanical strength of tabletC capping and lamination of tabletsD adhesion or sticking of powder to punchesE high friction during tablet ejection

The structure of the compacted powder, or tablet, is filled withair pores. The compression force and material compression proper-ties affect on the air content that will be entrapped in the preparedtablet, i.e. the porosity of the tablet. The porosity of the tablet can becalculated using the density of the tablet (ρt) based on the measureddimensions and weight of the tablet and the measured density ofpowder (ρm) [141]:

P = 1 − ρt

ρm(2.1)



2.3.4 Elasticity

For one dimensional (1D) situation in solids, a bar of length L andcross-section S stress σ is

σ =FS

(2.2)

where F is the force applied along L. The stress applied to a ma-terial causes a compression (or expansion) to the material. The re-sponse is called strain ε and it is defined as

ε =ΔLL0

(2.3)

where ΔL and L0 are the change and the original bar length, respec-tively. Hooke’s law (in 1D) states that the stress is proportional tothe strain:

σ = Eε (2.4)

This is true in case of elastic isotropic material in an unconfined1

geometry and material elasticity or elastic, i.e. Young’s modulus, E,can be generalized and applies to whole tested sample.

Instead of using mechanical testing (Fig. 2.3), the mechanicalproperties of a material can be determined, e.g. with acoustic mea-surements. This method utilizes the stress σ produced from apply-ing an acoustic wave to a medium. The theory of acoustics andacoustic measurements are described in chapter 3.

2.3.5 Mechanical failure of tablets

When the material is loaded axially in a die, the shearing forceis applied to the die wall through the generation of radial stress.As the powder bed thickness is reduced under compression, thepressure developed within the powder in the die varies with depth.The final tablet contains density variations due to friction betweenthe die wall and powder compact and is well known phenomenon[30, 48, 66, 147]. If the material is unable to relieve stresses present

1The movement of the material is restricted mechanically only in the directionof loading. Therefore, it can freely expand in the other directions.



��y

�u

Strain � =��L�L�

Yield point

�u

Ultimate strength

Y��

Figure 2.3: The principle of the determination of mechanical properties of materials bystress–strain curve Y(ε). The strain ε is proprtional to normalized change of the samplelength. σy, σu and εu are the yield strength, the ultimate strength and strain, respectively.Young’s modulus equals the slope of the linear part (elastic region) of Y(ε).

within a compacted tablet, capping and lamination (Fig. 2.4) canfollow compression by plastic deformation [62].

Figure 2.4: Two typical examples: an intact and a laminated tablet.

The compressive tabletting load, the speed of punches and therate of tabletting are known to affect the physical tablet character-istics such as mechanical strength [13]. The tensile strength of thetablet is theoretically dictated by the number and bonding force ofthe interparticulate bonds, which are affected by the particle size ofthe original powder [42].



2.4 QUALITY OF TABLETS

The use of solid tablets for medicinal purposes has a history ofusage spanning thousands of years [54] and the tablets have becomethe most popular dosage form due to their safety and simplicity[11]. Safe drug products can be perceived as free of contaminationand consistent in delivering the therapeutic benefit. To be safe, thequality of tablets must be controlled. Before a medicinal productis released for sale, the Qualified Person responsible for its releaseshould take into account, among other aspects, the conformity ofthe product to its specifications [43].

The pharmaceutical industry has been under strict regulariza-tion by authorities for decades. This has lead to a situation inwhich the pharmaceutical industry has reached only a basic levelunderstanding from certain manufacturing processes. This level ofknowhow has been enough for the manufacturers to continue pro-ducing pharmaceuticals. However, investing for a better, slightlymore effective way of processing might not have been tempting be-cause of the risks of failure.

2.4.1 Mechanical strength testing of tablets

The mechanical strength of a tablet is associated with the resistancetowards fracturing and erosion. An acceptable tablet must remainintact during handling between production and administration. [12]

A typical example of force-displacement data obtained from asingle compression event is shown in Fig. 2.5. The force trans-ducer must be instrumented axially in the line of loading in orderto gather correct loading values through compression cycle. Theinformation obtained from the loading cycle (Fig. 2.5) includes theenergy (work) due to tabletting (compression and decompression).The friction between the die wall and powder bed is energy lostdue to compression. Therefore, it can be used for adjusting thelubrication and maximum loading for tabletting to obtain strongtablets and minimize the risk of jamming. The rise of ejection forceindicates increasing friction in the die wall and helps to adjust the



process before getting stuck.

Upper punch displacement

Work of compaction

Compression Decompression

Work recovered during decompression

A C D

B

Figure 2.5: Force-displacement during powder compression. The total and elastic worksare defined by ABC and DBC, respectively.

The most common way to assess powder compactability is tostudy compaction pressure on the resulted tablet strength. The ten-sile strength of a tablet σt can be measured with diametrical com-pression (Fig. 2.6) as suggested by Fell and Newton [45]:

APPLIED LOAD APPLIED LOAD

Compression failure

Tension failure

Shear failure

APPLIED LOAD APPLIED LOAD

TENSION LOAD

TENSION LOAD

Figure 2.6: Typical tablet fractures by diametrical testing. Modified from [29].

σt =2FC

πDh(2.5)



where FC is crushing force, D and h are the tablet diameter andthickness, respectively. The crushing force is equal to the maxi-mum value of the measured force in the diametrical comression test(corresponding to the ultimate stress) prior to tablet breaking. Formany pharmaceutical products, compression is the last processingstep, only to be followed by coating, release testing and packaging.The coating is sometimes considered undesirable from a cost andcycle time perspective. [148]

2.4.2 Tablet swelling test as a simulator of controlled release

One example of a post production tablet quality test is the swellingand erosion front testing. The hydrophilic polymer matrix tabletis immersed to a buffer solution and swelling is monitored as afunction of time. The test simulates conditions in the GI tract anda pH value can be modified to be, for example, as in the humanstomach (pH ≈ 1.5) or in the small intestine (pH ≈ 6.5) [44], inwhich the proper location for drug release would be.

Swellable matrix tablets have become popular as drug deliv-ery technologies because of their ability to regulate drug releasekinetics and relatively simple manufacturing process [158]. Themonolithic systems can be prepared by compression of a powderedmixture of a drug and additional excipients. Drug release fromswellable matrix tablets is strongly associated with the swellingand dissolution characteristics of the hydrophilic polymer, i.e. theformation and erosion of an outer gel layer on the matrix sur-face [27, 35–38, 59, 133, 134].

Exposure to biological fluids in the gastrointestinal tract causesthe liquid penetration into the dry tablet matrix evoking an abruptchange of the hydrophilic polymer from the glassy to the rubberystate. At the time, a sharp boundary appears between the glassyand rubbery regions, i.e. the swelling front (Fig. 2.7). The total vol-ume of the tablet increases due to polymer swelling and a boundarybetween the polymer matrix and the surrounding medium, calledthe erosion front, becomes detectable [83]. These two physically



Swellable matrix

Erodible matrix (size reduction)

Hydrophobic matrix

Swelling front

Diffusion front

Eroding front

Dry matrix with API

Figure 2.7: Schematic representation of the polymer matrix tablet types during immersion.Liquid penetration forms eroding (outer) and swelling (inner) fronts, where the gel layeris formed in between. Modified from [35, 104].

evident fronts define the tablet gel layer. During the drug releaseprocess the gel layer thickness as well as its structure and com-position experiences a continuous change. With time, the swollengel layer becomes sufficiently hydrated for erosion or dissolution totake place. The swelling behavior of the tablet matrix can be de-scribed by the movement of the swelling and erosion fronts. Insidethe gel layer, a third front, called diffusion front, may also exist sep-arating the undissolved drug from the dissolved [84]. In Table 2.2,different studies of immersed tablets’ front detection are listed. The

Table 2.2: Published methods to monitor the get front movements.

Method Reference

Visual [17, 34, 46, 50, 173]FTIRi [73]NMR [2, 20, 39, 49, 78, 98, 118, 143]US [77]

drug release is controlled by the dissolved drug diffusion throughthe gel layer and/or by erosion of the gel layer. Therefore, therehas been an increasing interest in focused on objective methods for



qualitative and quantitative analysis of erosion and swelling frontcharacteristics in the research of pharmaceutical tablet manufactur-ing technology.




3 Introduction to Acoustics

3.1 BASICS OF ACOUSTICS

Acoustics is the science of sound and is considered to have its originin ancient Greece [88]. Sound in general, is a propagating (mechan-ical) oscillating motion of particles. When particles are oscillatingin parallel to the direction of propagation the wave is longitudinal.If the oscillation is perpendicular to the direction of oscillation, it iscalled a transverse or shear wave (Fig. 3.1). Also other wave types,e.g. torsional, surface and plane waves, may occur.

λ

Direction of wave propagation

λ

Direction of particle vibration

a)

b)

Figure 3.1: a) Longitudinal and b) transverse wave. The wavelength λ and the directionof the particle vibrations are shown. [142, 145]

3.1.1 Pressure and Wave Motion

The most basic definition of a wave is a disturbance that propagatesthrough a medium [25]. Waves are generally described by the pres-sure variations in the medium due to the wave. The total pressure



in the medium is given in location x and time t by

pT(x, t) = p0(x, t) + p1(x, t), (3.1)

where p0 and p1 represent the ambient pressure of the mediumand pressure fluctuation caused by the acoustic field. The floorlevel of an acoustic pressure is considered by the smallest pressurea human ear can sense, 20 μPa. The highest pressure for a humanear to withstand can be over 200 Pa. Because of very large dynamicrange, it is convenient to work with a relative measurement scalerather than an absolute scale [25]. The sound pressure level, SPL, isdefined by

SPLdB = 20 log10p

pref(3.2)

where pref is the reference pressure, e.g. 20 μPa. Usually in acousticapplications, only the longitudinal and transverse waves are ap-plied. They are utilized for estimating the physical characteristicsof a medium (Table 3.1) such as elasticity. In gases and fluids, trans-verse, i.e. shear wave do not occur [25].

A function that repeats itself exactly after certain intervals oftime is called periodic. The simplest case of periodic motion is theharmonic (or sinusoidal) that can be defined mathematically by asine or cosine function:

u(x, t) = u0 cos 2π f( x

c− t

)(3.3)

where u0 is peak amplitude, x, and t, are coordinates in space andtime, respectively. c is the speed of a wave in the medium, f , isfrequency and 2π f is the angular frequency. The time between twoidentical conditions of oscillation is defined as its period, τ, andit is the inverse of wave frequency, f . In terms of the period andwavelength, λ = c/ f , (3.3) can be expressed as

u(x, t) = u0 cos 2π( x

λ− t

τ

)(3.4)

The resonant frequency, fr, of the standing wave is determined bythe wavelength and the speed of sound: fr = c/λ. Solid particles or


Introduction to Acoustics

a body can absorp energy, e.g. mechanical vibration most efficientlyif the frequency of the vibration equals to resonant frequency of thebody.

3.1.2 Speed of Sound

The elapsed time Δt for the wave traveling the distance Δx betweentwo points, e.g. locations of the sending and receiving transducer,give relation for the speed of sound:

c =ΔxΔt

, (3.5)

The transmission of the wave (and speed c) is dependent on themedium properties, e.g. density and elasticity. In isotropic solids,the shear rigidity of medium couples the longitudinal and trans-verse wave components together [168]. Therefore, the speed ofsound depends on both the bulk and shear modulus of the mediumitself.

3.1.3 Acoustic Impedance, Transmission and Reflection

The energy of the oscillation, or vibration, is transmitted through amedium via a progressing wave. The material (medium) has a char-acteristic property to transport the energy of a mechanical wave.The property that represents the ability to resist the mechanical en-ergy transportation, is called acoustic impedance of the material. Itcan be expressed as a material specific parameter Zi:

Zi = ρici, (3.6)

where ρi and ci are the density and the speed of sound of themedium i. When a wave is transmitted through the interface of twomedia, the physical properties of the materials surrounding the in-terface determine how much of the energy is transmitted throughthis junction, i.e. interface. The efficiency of the energy transferfrom one medium into the next is given by the ratio of the two



impedances. The relative amplitude of the wave that is reflectedback and its magnitude can be expressed as reflection coefficient R

R =Z2 cos θi − Z1 cos θt

Z2 cos θi + Z1 cos θt(3.7)

where θi and θt are the angle of incidence and transmission, respec-tively. Z1 and Z2 are the impedances of material 1 and 2, respec-tively. The transmission coefficient T is determined as:

T =2Z2 cos θi

Z2 cos θi + Z1 cos θt(3.8)

If the impedances are identical, the transmission coefficient T = 1and all the acoustic energy will pass through the interface. Veryoften acoustic boundary conditions apply as the impedances arenot equal and the acoustic mismatch between two media exists [25].

In ultrasonic material testing, the speed of sound is utilized togather information about the test material. The typical mechani-cal characteristics of material related to the acoustic (ultrasound)parameters are shown in Table 3.1. cL and cT are the speed of

Table 3.1: Basic mechanical parameters in relation to the acoustic material testing forisotropic materials.

Parameter Equation

Poisson’s ratio ν = εSεL

= ΔS/S0ΔL/L0

= 1−2(cT/cL)2

2−2(cT/cL)2

Bulk modulus K = c2Lρ

Young’s modulus E = c2Lρ

(1+ν)(1−2ν)1−ν

Shear modulus G = c2Tρ (Exists only in solids.)

longitudinal and transverse wave in (solid) medium, respectively.Poisson’s ratio ν is defined as the ratio of transverse to longitudinalstrains of a loaded specimen. ν is needed in calculating of the elasticproperties of a solid medium with longitudinal US measurement.

3.1.4 Attenuation

In an ideal isotropic material, the acoustic pressure of a travelingsound wave remains constant and, hence, the energy is conserved.



However, if the material has some discontinuity inside, the atten-uation phenomenon takes place within the material. A defect isa typical example of a discontinuity in pharmaceutical tablet. Itoccurs, e.g. due to capping or lamination after compression. Theattenuation can occur by means of absorption, scattering and beamspreading. Mathematically, attenuation by absorption or scatteringof wave can be represented as a decaying exponential. Acousticsignal attenuation coefficient α is dependent on the frequency ofthe signal [132, 163]. The attenuation coefficient α measured on fre-quency f can be calculated:

α( f ) =8.686

hln

Aref( f )A( f )

(3.9)

where h is the thickness of the material specimen, A( f ) is the trans-mission amplitude and Aref( f ) is the amplitude of the measuredreference (a measurement without the sample) on frequency, f .

3.2 ACOUSTIC MEASUREMENT TECHNIQUES

Acoustic techniques can be categorized into two types: acousticemission (AE, passive mode) and ultrasound (US, active mode)techniques. The technique of acoustic emission is based on the de-tection and analysis of sound produced by a process or system,whereas in the ultrasound method the transducer is used for pro-ducing a mechanical pulse using voltage excited piezoelectric crys-tals. Usually, the same transducers can be used for transmittingand receiving a signal. The frequency of the sound can be infra-sonic (i.e. subsonic), audible or ultrasonic having frequency bandsof f < 20 Hz, 20 < f < 20000 Hz or f > 20000 Hz, respectively(Fig. 3.2). [125, 155]

3.2.1 Acoustic emission

Acoustic emission (AE) is the name given to the transient elas-tic stress waves that are generated by the rapid release of energy.Sources of AE can be localized to crack growth and many other



f [Hz]

Figure 3.2: Acoustic spectrum. The typical fields of applications (blue) are named abovethe three frequency ranges (red). [107]

types of material degradation. AE is also emitted during materialshitting or rubbing together. In recent decades, acoustic emissionhas been used as a nondestructive testing (NDT) method [172]. It iswidely used as a NDT technique and is applied routinely for the in-spection of aircraft wings, pressure vessels, load-bearing structures,mechanical integrity of bridges and components. Acoustic emissionis also used in our daily lives, for example, for the automatic ad-justment of the ignition timing in car engines and for screening ofheart and lung functions by a physician with a stethoscope.

As a passive technique no stimulus is transmitted to the objectin AE. Therefore, AE can be considered as a recording of acousticevents with a special microphone. Most commonly for AE testing,frequencies 100–300 kHz, are used [172]. However, applicationswith frequencies of range 0.03–1 MHz are not exceptional [149,164,165].

The AE technique is highly sensitive and the measured AE sig-nal may contain a high number of transients from sources locatedboth in the object and the environment. In Fig. (3.3), the AEof poured dry powder flowing along fluidized bed stainless steelchamber wall is shown. The first particles hit the surface in 2.5 sec-onds from the start of acquisition and the signal level increases untilplateau is reached in six seconds. The object sources include surfacevibration, collisions and friction between particles or particles and



Figure 3.3: An example AE signal lasting 10 seconds. The event starts at 2.5 seconds.High transients can be detected from the signal.

surface and changes in the the material matrix due to heating, etc.Furthermore, certain AE waves can be masked by the AE generatedfrom friction and rubbing [152].

To calibrate the AE sensors, a standard [14] by the American So-ciety of Testing and Materials (ASTM) can be used. The standard isbased on the emitting sound of breaking a special thin graphite rodagainst a plate that is used as a waveguide for AE sensing. How-ever, executing the standard method may lead to variable results inthe calibration of transient numbers and amplitudes [97]. Newermethods have been published to enhance the repeatability of mea-surements [15] and AE sensor response calibrations [53]. However,if the exact source of AE is not known, it could be useful to moni-tor changes in the process qualitatively. For quantitative measure-ments, the response in AE to phenomenon must be known in orderto have correct values.

The AE signals can be separated to be bursts or continuous in na-ture. In continuous AE, the emission does not "shut off" in contrastwith bursts of AE, that consists of short events (sounds) generatedby, e.g. fractures and the cracking of solid crystals. Some typicalparameters that are used in AE analysis are shown in Table 3.2.

The advantage in the use of AE is its noninvasive nature as atechnique. The instrumentation can be done by attaching the sen-



Table 3.2: Parameters in AE analysis.

Parameter Definition

Threshold Signal level over background noise.Peak amplitude The maximum of AE signal.Duration The time from the first threshold crossing to

the end of the last threshold crossing.Energy Integral of the rectified signal over the

duration of the AE hit.Counts The number of AE signal exceeding threshold.Count rate The number of counts per time unit.Average frequency The average frequency over the entire AE hit.Rise time The time from the first threshold crossing

to the maximum amplitude.Frequency spectrum Frequency contents of a signal.Histogram Distribution of magnitudes of AE signal

impulses.

sor to a surface that has an acoustic connection to the event to berecorded. A route is considered as connected if the mechanicalwave from the acoustic event can propagate to the sensor. Thus,an example of a good acoustic connection would be an undamagedpiece of metal with an AE sensor attached to the other end and ob-ject/events hitting/occurring at the other end. The rod would act asa connecting waveguide for recording the events. The quality of theconnection is measured with acoustic impedances of counterpartmaterials. This contact can be enhanced by adding a layer of someacoustic couplant in order to match material counterparts for wavetransmission over the contact. Practical AE sensor instrumentationis as easy as the AE sensor put in contact with a silicon grease toa surface guiding the observable vibrations. The silicon grease forvacuum use works as a good waveguide up to 400 kHz [144] andits properties stay constant because it does not vaporize.



3.2.2 Ultrasound

Active acoustic techniques differ from AE by the stimulus given tothe objects. The stimulus, e.g. a mechanical impact or powerfullight impulse. In the case of using light, the method is called a pho-toacoustic (PA) technique, (Fig. 3.4). Thus, active methods can not

LASER

PE TT PA

Figure 3.4: US measurement: pulsed echo (PE), through-transmission (TT) and photoa-coustic (PA) method.

be automatically considered as non-invasive and the technique isalso used intensionally for destructive purposes [24, 111]. For non-destructive applications, the stimulus must be limited in energy forminimizing the risk of causing permanent changes to the object.

In the case of ultrasound (US), the instrument is designed togenerate ultrasound waves across a defined frequency range. Thesewaves travel through the medium (sample) and are measured us-ing a receiver. The detected ultrasound reflects the changes in thespeed of sound or sound attenuation due to the interaction withthe medium. In pulsed echo (PE), i.e. pitch-catch measurementgeometry, sending and the receiving of US is done with the sametransducer. In through-transmission (TT) geometry, the US signal istransmitted through an object and received after transmission fromother side with another transducer. The advantage in the use of PEis that only one transducer is needed. The advantage of TT geome-try is that, most likely, the measurement of the speed of sound in amedium succeeds with TT geometry, if it is measurable at all.



Probably the most famous use of US is in ultrasonic imaging,in medicine (sonograms) as well as in the ocean (sonar). Ultrasonicwaves are used in many medical diagnostic procedures such as de-tecting malignancies and hemorrhaging in various organs. US isalso used to monitor real-time movement of heart valves and largeblood vessels [125]. US waves are directed toward a patient’s bodyand reflected when they reach boundaries between tissues of differ-ent densities. These reflected waves are detected and displayed ona monitor.

3.2.3 Fourier Analysis of Signals

The signal holds information of the measured phenomenon. It canbe analyzed, e.g. in time or frequency domain. The mathemati-cal tool for analyzing the frequency domain of the signals acquiredfrom the vibration system is through the usew of a Fourier trans-form [75]. A signal g(t), which is a function of time, t, is trans-formed to a frequency domain. The (continuous) Fourier transformG(ω) is defined as

G(ω) =∫ ∞

−∞g(t)ejωtdt (3.10)

where j =√−1, ω is the angular frequency and jωt is the phase

angle. The (continuous) inverse Fourier transform is g(t):

g(t) =1

2π

∫ ∞

−∞G(ω)e−jωtdω (3.11)

The measured signals are (usually) intrinsically discrete. In fourieranalysis of the measured signals, the components in different fre-quencies can be revealed (Fig. 3.5). The frequency spectrum ofa discrete signal can be calculated using (discrete) Fourier trans-form/series (DFT):

Γn =K

∑k=1

gke−j 2πK (k−1)(n−1) (3.12)

where K is the length (integer) of the sample. It is very common that



g(t)

g(t)

g(t)

g(t) GG (ω)

G (ω)

G (ω)

G (ω)

Fourier transform

Figure 3.5: Meaning of Fourier transform. Modified from [75].

a spectrum contains various kinds of information, e.g. mechanicalresonances due to fractures of a solid body, its density or knowl-edge of the light wavelengths it absorbed, encoded in signals. Forexample, speech is a result of vibration of the human vocal cords,ship’s propellers generate periodic displacement of the water, pres-sure, and so on. The shape of the time domain waveform is notimportant in these signals; the key information is in the frequency,phase and amplitude of the component sinusoids. The DFT is usedto extract this information. [138]

3.2.4 Acoustic Transducers

Ultrasound can be generated with an oscillating voltage over apiezoelectric crystal, when it starts expanding and contract alongwith the voltage. Hence, the voltage generates the mechanical vi-bration. Conversely, if a piezoelectric crystal is exposed to mechanicvibration, a respective oscillating voltage is generated over a mate-



rial [51, 163]. In Fig. 3.6, a schematic structure of a planar piezo-electric transducer is shown. Basically, there is a similarly workingactive element and wear plate in most of the acoustic transduc-ers. The diameter and thickness of the element are chosen in orderto have a certain resonant frequency for the crystal and the ma-terial of the wear plate can be tailored to have a certain acousticimpedance for acoustic matching against the vibration leading ma-terial. According to the similar structures of the active (US) and pas-

Figure 3.6: Schematic structure of piezoelectric acoustic transducer. [107]

sive (AE) transducers, also both types of transducers can be usedfor transmitting and receiving mechanical waves. This property istypically utilized, for calibrating AE sensors [156].

For materials research, the measurement distance can be critical,in order to get reliable attenuation results. The transducers havecharacteristic beam fields defined by the diameter and working fre-quency of the active piezoelectric element (Fig. 3.6). The beamintensity within the nearfield is not monotonic (Fig. 3.7) and cannot be used in attenuation measurements. However, adding properwave guide, i.e. a ’delay line’, between the sample and the US trans-ducer, farfield can be obtained. The length N of the nearfield can



Figure 3.7: Pressure field of a planar acoustic transducer. [107]

be estimated for a planar, non-focused transducer using (3.13)

N = D2 f4c

= D2 14λ

(3.13)

where D is the diameter of the transducer’s active element (Fig.3.6). In the attenuation measurements, the distance between thetransducer and the receiver should be ≥ N.

3.3 ACOUSTICS IN PHARMACEUTICS

The problem of the mechanical failure of tablets can be solved bymaterial engineering using theoretical simulations and experimen-tal measurements. One possible method that can be used to mea-sure mechanical properties of a tablet is ultrasound. It utilizes alow intensity mechanical wave to tablet material and the elasticwave propagation does not harm the material, i.e. the test is non-destructive. Recently, the ultrasound measurements of tablets havebeen studied intensively: the negative correlation of the speed ofsound and the porosity of the tablet has been shown [57] with USTT technique and the mechanical properties, such as Young’s mod-ulus, were estimated with ultrasound [7, 74]. US transducers wereimplemented to a hydraulic press and used US PE measurementsto study the feasibility of the pulse-echo method with high acousticmismatching boundary of the punch-tablet interface [89]. Tablets



were tested with US TT measurement as a function of tablet poros-ity [136] and the positive correlation between the speed of soundand tensile strength was found. However, none of these studieshave been used in a real tabletting environment that has continu-ously moving punches and changing mechanics.

3.3.1 Passive (AE) Techniques

In terms of pharmaceutical applications, the dependence of theacoustic measurements on physical properties such as particle size,mechanical strength, and cohesivity of solid materials allows thetechnique to be used for the control and endpoint detection of pro-cesses such as high shear granulation, fluid bed drying, milling,and micronization. [155]

Acoustic emission (AE) is a technique that has also been studiedin pharmaceutical applications and it can be also found in the phar-macopoeia [155]. Manufacturing processes cause vibrations thatcarry embedded information concerning both physical and chem-ical parameters (e.g. composition, mixing progress, flow density,particle size). These vibrations can be measured by AE sensors.AE can be applied as a noninvasive in-line technique. Therefore,it is also appropriate for a FBG environment. Some reported AEapplications in pharmaceutics are listed in Table 3.3.

3.3.2 Active (US & PA) Techniques

Active acoustic techniques have been used in various research prob-lems pertaining to pharmaceutical research. Previously reportedsolutions are listed in Table 3.4. One might say that, the US tech-niques have been particularly popular in tablet testing and com-pression studies, instead of process monitoring which has been themostly used target with passive acoustic testing methods (Table3.3).



Table 3.3: Reported AE studies succesfully applied in pharmaceutical processes. The ap-plication abbreviations stand for HSG: high shear granulation, T: tablets, TC: tablet com-pression, RC: roller compaction, HS: hydrated silica, FBC: fluidized bed coating, FBG:fluidized bed granulation and CG: crystallization/crystal growth.

Appl. Implementation Ref.

HSG Particle size of different elasticity granules. [82]HSG Monitoring changes in size, flow and compression [165]

properties during granulation. Detection ofgranulation endpoint.

HSG State of granulation process. [26]HSG Particle size and moisture content. [109]HSG End point detection [40]

T Capping of compacts after compression. [126]T NaCl tablets’ strength increase during AE decay. [127]T AE of post-compression relaxation. [96]

TC Powder AE during compression. [159]TC Correlation between the AE and compression work. [167]TC Tablet screening of capping and lamination [65]

during compression.RC Useful spectra up from 15 kHz. [129]RC Compaction AE of three different pharmaceutical [55]

materials.RC, T MCC powder during roller compaction and from [56]

single tablets after compaction by a single-punchtablet machine.

HS Monitoring of water content and grouping of [21]different particles.

FBC Coating film formation. [101]FBG Particle fluidization. [149]FBG Capable of detecting unwanted lump formation [58]

already at an incipient stage as well as bottomplate clogging.

FBG AE spectral response of granule size distribution. [91]FBG Discrimination of good and bad yielded FBGs. [92]FBG Evolution of size distribution with modelling of [115]

AE data.CG AE characteristics during crystallization. [52]



Table 3.4: Publications of active acoustic measurements in pharmaceutical research. Theapplication abbreviations stand for T: tablets, TC: tablet compression and RC: roller com-paction.

Appl. Implementation Ref.

T Photoacoustic (PA) evaluation of elasticity and [74]integrity of tablets.

T Eroding front movement of swelling tablets. [77]T US of tablet coating stiffness during layer [72]

formation.T Acoustic Resonance Spectroscopy of tablet [94]

identification and characterization.T Coating thickness estimations using air-coupled [4]

ultrasonics.T Tablet thickness estimations using [6]

photo(air-coupled)acoustics.T Tablets’ mechanical properties in x-, y- and [7]

z-directions with US TT.T US PE determination of Young’s moduli of [8]

the coat and core of a tablet.T Mechanical properties of bilayered tablets. [9]T US TT assessment of mean grain size in [137]

pharmaceutical compacts.T US PE of the tablet coat–core interface. [89]T US TT for tensile strength evaluation of tablets. [136]T US PE real-time in-die monitoring of the tablet [140]

compaction process.T US PE coating thickness microscopy. [23]

TC US assisted compaction. [124]TC US assisted compaction. [86]TC US characterization of tablet’s mechanical [3]

properties.RC Density distributions of ribbons. [10]


4 Aims

The present studies for this thesis included implementations of pas-sive and active acoustic techniques for pharmaceutical applications.In the first part of the work, in-line methods were developed for themanufacturing processes aiming to:

• estimate particle size throughout granulation processes withnoninvasive passive acoustic system (I)

• monitor real-time tablet formation and evaluate mechanicalproperties of tablets with an active acoustic system (II).

In the second part of this research, the active acoustic methods wereused for quality control purposes to:

• determine the mechanical integrity of the manufactured tabletswith nondestructive ultrasound measurements (III)

• to investigate the feasibility of ultrasound pulse echo scanningtechniques in monitoring the swelling behavior in hydrophilicmatrix tablets without disturbing the processes (IV).




5 Materials and methods

5.1 MATERIALS

The pharmaceutical manufacturing studies proceeded during man-ufacturing unit operations and quality assurance processes. Thematerial described in this chapter cover aims I and II for manufac-turing and III and IV for quality assurance processes.

5.1.1 Binder

Liquid binder was prepared for the wetting stage of granulationprocess (I) by mixing 16.7% polyvinylpyrrolidone (PVP, KollidonK30, BASF, Germany) and 83.3% of purified water.

5.1.2 Excipients

Several common pharmaceutical excipients were used for the stud-ies. Two lactose monohydrates (LM), Pharmatose R©200M and 90Mwere obtained from DMV-Fronterra Excipients (Veghel, The Nether-lands) to investigate research aims (I) and (III), respectively. Twomicrocrystalline cellulose (MCC) powders, Avicel R©PH200 and PH101were obtained from FMC Biopolymers (Cork, Ireland) to investigateresearch aims (II) and (III), respectively. To investigate researchaim (III), Emcompress R© Premium dibasic calcium phosphate di-hydrate (DCP), was utilized (JRS Pharma, Budenheim, Germany).The excipients were used as received without sieving or any otherpre-processing.

5.1.3 Model drugs

Two common active pharmaceutical ingredients (APIs) were usedin the studies. Anhydrous caffeine (CA, I), was obtained fromScharlau (Barcelona, Spain). Paracetamol (PRC, II), was obtainedfrom Xiamen Top Health Biochem Tech Co. Ltd. (Xiamen City,



China). The PRC was used after sieving with 0.75 mm sieve in or-der to break the large aggregates of PRC.

5.1.4 Ready-to-use materials and formulations

Commercial granules, Cellets R© (Harke Pharma, Germany) and Pro-tease A and B granules (Genencor Inc., Finland) were used as drymaterials to test particle size measurement methods (I). Cellets R©are spherical granules made of MCC (sphericity degree > 0.9 re-ported by the manufacturer). The studied granules included eightsize fractions between 150–1200 μm in diameter. Additionally, threeequally proportioned Cellets R© mixtures were prepared (Table 5.1).Ph.Eur. grade magnesium stearate (MS) powder, (II,III), was ob-

Table 5.1: The components for prepared Cellets R© mixtures.

Mixture Cellets fraction100 200 350 500 1000

Cellets100/1000 X XCellets350/500 X XCellets200/350/500 X X X

tained from Orion Pharma (Espoo, Finland) and used as received.Several hydrohilic polymer formulations were used for polymerswelling studies (IV). Hydroxypropyl methylcellulose (HPMC) usedwas of USP grade 2910 (type 60 SH) produced by Shin-Etsu Chem-ical Co. Ltd. (Tokyo, Japan). The polyethylene oxide (PEO) poly-mers used were Polyox WSR N-10 (PEO0.1) and Polyox WSR N-60K (PEO2.0) by Dow (Wien, Austria). The powders were used asreceived without sieving or any other pre-processing.

5.2 SAMPLE PREPARATION

The materials in the studies were weighed with an analytical bal-ance (A200S, Sartorius, Goettingen, Germany) and the mean den-sity values of powders were measured by five parallel determina-


Materials and methods

tions with a Multi-pycnometer (Quanta Chrome, NY, USA) usinghelium as the measuring gas.

5.2.1 Granules (I)

The granulated granules were measured during the fluidized bedtop spray granulation experiment. The formulation consisted of80 %(w/w) of LM and 20 %(w/w) of CA. The total dry mass of0.2 kg was used. The powders were manually premixed for twominutes prior to the granulation process.

5.2.2 Binary tablets (II)

Binary mixtures of MCC and PRC were prepared using binary mix-tures (Table 5.2): The A1 and A2 formulations are chemically simi-

Table 5.2: Recipes for binary tablets. MCC, PRC and MS stand for microcrystallinecellulose, paracetamol and magnesium stearate, respectively. One %(w/w) of MS wasadded to A1 and A2 and mixed for 2 and 10 minutes, respectively.

Recipe MCC (% w/w) PRC (% w/w) MS (% w/w)

A1 69.3 29.7 1.02min

A2 69.3 29.7 1.010min

B1 100.0 0.0 0.0B2 95.0 5.0 0.0B3 90.0 10.0 0.0

lar. Their only difference is the mixing time of MS. The B mixtureswere made by adding 0, 5 and 10 %(w/w) of PRC into MCC. Thebinary powder mixtures were to study real-time US response oftablet formation. Korsch EK0 (Korsch AG, Berlin, Germany) ec-centric tablet press was used to make flat-faced, 10 mm diameter,tablets and filling of the die was done using an automated fillingshoe. Different compression depths were used between test sets toachieve variation in the maximum compression force.



5.2.3 Monolithic tablets (III,IV)

Monolithic powders (Table 5.3) were directly compressed with acompaction simulator (PCS-1, Puuman Oy, Kuopio, Finland) usingflat-faced punches of 10 mm and 13 mm, in diameter, for swelling(IV) and integrity (III) tests, respectively. Prior to the compaction,

Table 5.3: Recipes for monolithic tablets. Tablet diameter (D), mass (m) and resulted tabletheight (h) are shown.

Study Material D (mm) m(mg) h(mm)

III DCP 13 800 3.1MCC 13 600 3.2LM 13 600 3.3

IV HPMC 10 300 4.0PEO0.1 10 315 3.7PEO2.0 10 315 3.7

the powder was manually poured into the die for each tablet (III,IV).Tablets for integrity testing (III) were prepared by employing a sin-gle sided triangle for the upper punch. The lower punch was keptstationary. Minimal lubrication was used by covering the die withfine MS powder using a brush. Additional sample sets with an’artificial’ defect were prepared by using parchment paper with a46 μm of thickness. Folded paper patches of 10 mm × 10 mm insize. The paper was laid horizontally in the vertical centre level onaxis of the powder bed of the die before compaction. In Fig. 5.1, acomputed tomograph image of an intact tablet is shown with tablethaving the paper patch enclosed inside by compaction. The result-ing tablet was considered to have a significant defect of 92 μm inthickness. The swellable polymer tablets prepared for the immer-sion tests (IV) were directly compressed with sinusoidal loadingprofiles for both the upper and lower punches with no lubricationused.



Figure 5.1: A computed tomograph of an intact and artificially defected 13 mm diameter(D) tablet. The violet and black colored areas in the image do not belong to the tablet.

5.3 EXPERIMENTAL SETUPS

Several different methods were used throughout the studies. Thestudies presented in I and II attempt to improve techniques ofprocess control and offer data obtained from implemented acous-tic monitoring techniques. Studies III and IV are focus on testingmethods for quality assurance. Table 5.4 summarizes the measure-ment methods used in each paper. Passive acoustic (AE) measure-

Table 5.4: Methods overview. "Visual" includes the topographic camera (TOPO) anddigital image analysis (DIA).

Study AE NIR US TT US PE Visual DT

I X X XTOPO

II X XIII XIV X XDIA

ments were done in paper I simultaneously with eight-point NIRspectroscopy and topographic camera. Active acoustic (US) mea-surements of two different geometries, through-transmission (TT)and pulse echo (PE), were used in papers II, III and IV. Destruc-tive testing (DT) to determine tablet crushing force measurementswere executed with a mechanical strength testing apparatus (CT5,Engineering systems, Nottingham, England).



There was eight different models of AE and US transducersused in the acoustic measurements of this thesis. The types of trans-ducers are listed in the Table 5.5.

Table 5.5: The acoustic transducer manufacturers and types used in the measurements.Focal Distance (FD) for focused transducers, effective element diameter (D) and nominalfrequency ( f ) are shown. ∗) Used together in a given study.

Study aManuf./Type Model FD (mm) f (MHz) D (mm)

I Vallen/AE V150-M No 0.15 20II Panametrics/US XMS-310 No 10.00 2.0III Panametrics/US XMS-310 No 10.00 2.0

Panametrics/US V133-RM No 2.25 6.0* Panametrics/US C110-RM No 5.00 6.0* Panametrics/US V110-RM No 5.00 6.0

Panametrics/US V112-RM No 10.00 6.0IV Panametrics/US V307 50 5.00 19

aPanametrics Ltd. is nowadays part of Olympus NDT Inc., Olympus Corp.

5.3.1 Fluidized bed studies (I)

The study was carried out using a modified fluidized bed granula-tor (STREA-1, Aeromatic-Fielder AG, Switzerland) equipped with atop spray unit and a custom made granulation chamber of 485 mmin height (Fig. 5.2). During all AE measurements air spraying wasapplied through a D = 0.8 mm nozzle and with a 0.6 bar atomizingpressure to get similar background spectrum for all AE tests.

The AE sensor was attached onto the outer surface of the gran-ulation chamber with silicon grease (Dow Corning DC 976 highvacuum grease, Kurt J. Lesker Company, Pittsburgh, USA). The to-pographic images were captured through the window seen in Fig.5.2.



Sampling spoon

Window

Topspray unit

Humidity and temperature sensors

NIR probes R1...4

NIR probes L1...4

AE

Figure 5.2: Instrumentation of FBG used in the study. TOPO is removed and does notshow on the figure. AE sensor location is behind the NIR probes R1...4. Air inlet andoutlet is marked with green and red arrows, respectively.

Fluidization of commercial granules

Three minute fluidization experiments were conducted with 300gof ready Cellets R© or Protease granules. The measurements weredone in-line during fluidizing with TOPO and AE.

Granulation process monitoring

The process had three major stages: (1) Five minutes of premix-ing started the process when the fluidizing air flow was kept in 18m3/h. (2) The wetting stage was continuing for 17 minutes withincreased air flow up to 36 m3/h to maintain adequate fluidizationduring the wet granulation. (3) Drying was conducted for 10 min-utes and fluidizing air flow was set to 25 m3/h. During the wettingstage of the granulation process, 55 g of binder was sprayed on thepowder bed through the nozzle with a constant pumping speedfrom a peristaltic pump.

PAT sensors were instrumented into the fluidized bed granu-



lator for real-time non-invasive process monitoring purposes. Themeasurement setup consisted of a five-module granulator chamberinstrumented with an AE module, TOPO camera and eight NIRprobes. Both the NIR probes and TOPO were positioned to capturedata through glass windows whose surfaces were smoothly alignedwith the inner metal surface of the granulation chamber.

The FBG was also equipped with inlet and outlet air tempera-ture and humidity sensors (EE21, E+E Elektronik, Austria), a bedmass temperature sensor (T-type thermocouple, Amestec Oy, Fin-land) and an outlet air velocity sensor (MiniAir 6 anemometer,SchiltknechtMesstechnik AG, Switzerland). A Grant Squirrel SQ800 data log-ger (Grant Instruments Ltd., U.K.) was used for acquisition of thetemperature, relative humidity (RH) and air flow data.

AE spectroscopy

The acoustic recording module consisted of a 16-bit A/D card (USB-6251M, National Instruments, Austin, TX, USA), AEP4 amplifier,AEP3 preamplifier and VS-150M AE transducer (Vallen-SystemeGmbH, Germany). The signals were recorded using a custom madeprogram and Matlab 7.3 with the Data acquisition toolbox (Math-works Inc, USA). The signal acquisition was carried out using arecording length of 100 ms with sampling frequency of 1.25 MHzand 1.6 s intervals between each recording point. Frequency in-formation was calculated from the AE signal using the fast Fouriertransform (FFT). The frequency band of 50–300 kHz was segmentedinto 32 frequency segments. Particle size distribution occurringduring formulation was measured by a sieving test before and afterthe granulation process. Mean particle size values of the sampleswere used as the reference values for calculations.

Particle size estimation using AE spectra

Particle size was estimated in conditions of constant moisture witha linear combination of three frequency bands (low: 50–100 kHz,



middle: 110–150 kHz and high: 200–300 kHz). These frequencybands were used for an instant measurement point ’i’ to estimatethe particle size (the mean diameter), zi:

zi = (θ1 Ilow,i + θ2 Imid,i + θ3 Ihigh,i + θ4) (5.1)

where θ1 , θ2 and θ3 are linear coefficients and θ4 is a constant. Thecoefficients were obtained from the fluidization measurements byleast squares fitting

θLS = (HTH)−1HTz (5.2)

where H = [IlowImidIhigh1] and z is the vector containing the mea-sured particle size information. Vectors Ilow, Imid and Ihigh containthe measured and normalized AE mean intensity values related toeach particle size. The normalization was performed by dividingthe intensity of each AE band recorded as function of reference sizeby the intensities of the AE bands recorded from the Cellets100 test.

For the granulation experiment, where the bed moisture changesdynamically, granule size was estimated with an simplified algo-rithm based on a comparison of two AE frequency bands F1 : {96.9–120.3} kHz and F2 : {175.0–198.4} kHz. Ratio ’R’ of the mean AEmagnitudes Mi = 20 log10 Ii, where i = {1, 2}, was used as:

R =M2

M1(5.3)

The particle size estimate at measurement number i was estimatedusing zi = Rizre f , where zre f is the particle size of unprocessedpowder.

TOPO image analysis

The module for particle size and topography consisted of a pro-totype device (VTT, Oulu, Finland) including a LED light source,projection optics, control and synchronizing electronics and a CCDcamera. The collimated pattern was captured with an exposuretime of 50 μs and an image resolution of 1280 × 960 pixels withpixel dimensions of 8μm×8 μm.



Particle size estimation was made by capturing the images ofthe fluidizing powder through a glass window of 25 mm in widthand 175 mm in height in the conical section of the granulator. Theparticle size in this approach was estimated using an imaging basedmethod, where the 3-D structure of the particles was evaluated. Thealgorithm is based on ellipse fitting. First, the particles are detect-ing using scale invariant features [87]. Detection gives estimates ofthe spatial coordinates of each detected particle, along with a roughestimate of its size. Using this as an initial approximation, an iter-ative ellipse fitting method is used to refine the size distribution.The similar sinusoidal pattern approach as used previously [116],was used in study I.

NIR spectroscopy

The used multi-point NIR module consisted of spectral camera(Specim Oy, Oulu, Finland) capable of full NIR range of 1000–2500nm, and eight probes (VTT, Oulu, Finland) with an internal refer-ence. The probes R1...4 stand for right-hand-side from top (R1) tobottom (R4) and L1...4 stand for left-hand-side from top to bottom(Fig. 5.2).

The measured data is transformed into absorbance units by tak-ing a 10-base logarithm of the inverse of the reflectance spectra. Inaddition to the strong increase in absorbance at the water peak lo-cations, the presence of water is observed as increased baseline inthe absorbance spectra which is caused by the decrease in the in-tensity of the backscattered light. In the granulation experiment,no reference moisture values were measured, and the moisture ofthe granules was only studied qualitatively. The purpose was toshow that the moisture content can be successfully extracted fromthe measured data. The NIR spectra were collected at the rate of3 Hz with the exposure time of 8 ms. Three consecutive spectrawere averaged to obtain one spectrum per second.



Water content monitoring with NIR

To permit the estimation of temporal moisture profiles, the aug-mented linear model (5.4) was assumed for the measured spectralsignals [31, 90].

xi = ai1 +3

∑j=1

cijsj + diλ + eiλ2 + εi (5.4)

A Column vector xi is the measured log10(1R ) spectrum at any of

the channels at the time instant i. The second term on the right-hand-side contains the pure analyte spectra of lactose, caffeine andgranulation liquid, which were measured off-line as the vectorssj, j = {1, 2, 3}. The elements of the vectors 1, λ and λ2 follow con-stant, linear and quadratic functions of the wavelength, respectively,and they attempt to explain changes in the offset, tilt and curvatureof the spectral baseline. The unmodelled residuals are representedhere by the vector εi. The scalar coefficients ai, cij, di and ei were es-timated for each measured spectrum in the least squares (LS) senseas:

[ ai ci1 ci2 ci3 di ei ] = xTi P(PTP)

−1(5.5)

where P = [1 s1 s2 s3 λ λ2]. The estimated coefficient ci3 (e.g. theweight of granulation liquid spectrum) was assumed to be propor-tional to the moisture level of the measured powder.

Other reference methods

The reference particle size used for the in-line estimation was de-termined by sieving. The reference size for Protease samples wasanalyzed by their donor with commercial video imaging based an-alyzer (PartAn, Sci-Tec Inc., Sandy Hook, CT, USA).

Error of estimations

For the method accuracy and precision evaluation, the root meansquared error (RMSE) and bias of estimation [102] were used. RMSE



was calculated using:

RMSE =

√1n

n

∑i=1

(zi − zi)2 (5.6)

where zi and zi are the estimated and true value, respectively.

5.3.2 Tablet formation monitoring (II)

Making of powder blends for tablet formation monitoring

The first mixture was prepared by mixing of MCC and 30 %(w/w)

of PRC with Turbula R© mixer (Willy A. Bachofen Machinenfabrik,Basel, Switzerland). The mixing time was 10 min at 22 rpm afterwhich the mixture was divided into two different sub-mixtures, A1and A2. In order to study the effect of extended MS mixing time,1 %(w/w) of MS was added in the middle of the both powder bedsand mixed again at 22 rpm. The extra mixing time for mixtures A1and A2 was two and 10 min, respectively. Another set of binarymixtures, B1, B2 and B3, was prepared from MCC and PRC. Usedconcentrations are listed in Table 5.2.

Instrumented tabletting measurement system

The tabletting was done with an instrumented single station tablet-ting apparatus (EK0, Korsch Pressen, Berlin, Germany) with custom-made tabletting punches (Fig. 5.3B). One Panametrics XMS-310miniature US transducer was implemented inside of each punch.The position and loading of the upper and lower punches weremonitored with linear displacement sensors (LP30FQJ, Midori Pre-cisions Co. Ltd., Tokyo, Japan) and strain gages. The tablettingdynamics were measured with Data Acquisition and AnalysingSystem (DAAS, Waltti Electronics Ltd., Kuopio, Finland) and thesampling rate of 200 Hz was used. A metallic bolt was attachedto the flywheel of eccentric tabletting apparatus for triggering theinductive sensor (IF 5297, IFM electronic GmbH, Essen, Germany)



A) B)

Figure 5.3: A) Schematic overview of the US measurement system. B) 3D cross-sectionalview of a punch. The miniature US transducers (model XMS-310) are attached withbicomponent polymer glue in the bottom end of D = 5 mm cavity drilled along the axis ofpunches. The electrical connector is removed in the picture.

connected indirectly to an arbitrary waveform generator (WFG, Ag-ilent model 33220A, Malaysia). A schematic illustration of the sys-tem is shown in Fig. 5.3A. An arbitrary signal including 40 squarewaves was generated with WFG to control the US pulser/receiver(PSR, Olympus model 5077PR, Olympus NDT Inc., Waltham, MA,USA). The PSR voltage was set at 200 V, transducer frequency set-ting was 10 MHz and the receiver gain was +20 dB with 1 MHzhigh pass and 10 MHz low pass filters. The transmitted US pulsewas recorded after amplifying (+40 dB gain, pass band frequen-cies: 0.05–10.00 MHz) with a preamplifier (PRE, Olympus Ultra-sonic Preamp, USA). The US measurements were done with a dig-itizing oscilloscope (LeCroy Wavesurfer 42Xs-B, LeCroy Corp., NY,USA) using sequence mode. 40 sequences were measured in a rowfor each tablet. The measurement system was controlled with a PCcomputer and custom-made LabviewTM (version 10.0, National In-struments, Austin, TX, USA) program. No averaging was used. Aand B formulations were prepared in experiments.



Real-time US measurements during tabletting

Formulation A was used to study the effect of extended MS mixingtime in tablet compression properties and formulation B was usedto study the effect of PRC concentration in tablet compression. Ineach test, 10 replicates were compressed. In the extended MS mix-ing experiment, the transmitted US was measured for each tabletin 40 time points with 8 ms intervals starting at 300 ms after trig-gering by the flywheel. In the other experiment, the effect of PRCconcentration on the mechanical properties of tablets as a functionof time was studied. The transmitted US was measured for eachtablet in 40 time points with 1 ms intervals starting at 535 ms afterflywheel triggering.

Analyses from the measurements

The off-line analyses of US measurements were done using Mat-lab R2007a software (Mathworks Inc., Natick, MA, USA). Time offlight (TOF, [117]) was determined from the measured waveformsby calculating the Hilbert Transform envelope of the recorded sig-nals. The TOF was calculated from the point exceeding 2

3 of themaximum value of the signal. The corrected TOF was obtained bysubtracting the reference TOF measured with the empty die. Thereference measurement was performed with the tabletting punchestogether without any additional acoustic couplant between them.The applied load for the reference measurement was 10 kN. Thespeed of sound (c) was calculated using Eq. (3.5), using the thick-ness of the tablet in compression and time difference between TOFof tablet measurement and TOF of empty die. The frequency spec-trum of the US transmission was determined using the fast Fouriertransform (FFT) algorithm. Only the first pulse of the measured USsignal was used.



Tensile strength and porosity of tablets

After the elastic recovery, the crushing force, FC, of each tablet wasmeasured with destructive diametrical crushing test (Eq. (2.5)). Theporosity of tablets was calculated using the measured dimensions,weights and density values of tablets and Eq. (2.1).

5.3.3 Tablet integrity testing (III)

US transmission along the tablet axis through the manufacturedpharmaceutical tablets was measured in order to determine theirmechanical integrity, nondestructively. The tests were done usingthree pairs of commercial flat faced contact US transducers (Tab.5.5) and a pair of instrumented punches (Fig. 5.3B) installed in totabletting press (Korsch EK0). US pulser-receiver (model 5077PR,Olympus-NDT Inc., Waltham, MA, USA) was used as a source forsignals measured with LeCroy Wavesurfer 42Xs-A digital oscillo-scope (LeCroy Corp., NY, USA).

Tablets for integrity testing

Three common pharmaceutical excipient powders with the meandensities of 2.389 (DCP), 1.668 (MCC) and 1.538 g/cm3 (LM), wereused as received without sieving or any other pre-processing toprepare sets of intact and artificially defected tablets (Table 5.3).After compaction, the tablets were stored with anhydrous silica.The tablet dimensions were measured with a micrometer (Digitrix,NSK, Japan) and tablet weights were determined with the analyticalbalance within 24 h after compaction. The total number of sampleswas n = 55, with the number of intact DCP, MCC and LM tabletsprepared was 10, 6 and 10, respectively. The number of defectedDCP, MCC and LM tablets prepared was 11, 8 and 10, respectively.

US measurements with the contact transducers

The intact and the artificially defected tablets were tested with theUS transmission technique under constant uniaxial loading. Con-



tact ultrasound transducers were used to measure the speed ofsound in three different pharmaceutical tabletting materials at threedifferent frequencies. All the measurements were made withoutany additional acoustic couplant. Transmitted US signals of tabletswere measured with three different pairs of commercial contactUS transducers (Table 5.5). The samples were axially compressedbetween two US transducers and the loading force was manuallydriven by a bolt (Fig. 5.4). The axial force was monitored witha load cell (LPM560, Cooper Instruments, Warrenton, VA, USA).Transducers of the nominal frequencies of 2.25 MHz (Olympusmodel V133), 5 MHz (Olympus model C110 sending and V110 re-ceiving) and 10 MHz (Olympus model V112) were used. To obtainthe constant state for acoustic dry coupling, a uniaxial load of 12 Nwas selected. No acoustic couplant or any other material that mightcontaminate the samples during the measurements was used. TheUS pulser voltage of 200 V (excluding 100 V during 5 MHz mea-surements) with the repeat rate of 1 kHz and receiver gain of 10 dBwas utilized. Averaging of 512 consecutive measurements, withoutany filtering, was used during the measurements. The samplingfrequency was 50 MHz during the 2.25 MHz measurements and250 MHz during the 5 and 10 MHz measurements.

US measurements with the instrumented tablet press punches

An eccentric single station tablet press similar as used in Paper II

was used with instrumented punches (Fig. 5.3B). The US trans-mission signals of tablets were measured during static compressionwith 210 N. The tabletting die was removed from the apparatus inorder to measure tablets having a diameter of 13 mm. The load-ing was established by manipulating the flywheel of the tablettingmachine manually and the load was monitored continuously. Thediameter of flat-faced punches was 10 mm. Any special acousticcouplants or other materials that could contaminate the samplesduring the measurements were not used. The same US pulser-receiver, with a voltage of 200 V and receiver gain of +30 dB, was



Figure 5.4: The component configuration used for the US transmission measurementswith contact transducers.

used throughout the study. No filtering was used during the mea-surements. The repeat rate of the pulser was 200 Hz. The averagingof 512 consecutive measurements was used during the measure-ments.

Speed of US and US Attenuation

The sample tablets were measured with US. Transmitted US speedwas measured with every transducer pair for different materials.The attenuation coefficient of the US was calculated with (3.9) fromthe transmitted signal for eight frequencies: 1.2, 1.8, 2.4, 3.0, 3.6,4.2, 4.8 and 5.4 MHz. The transmission amplitude of cylindrical,30.005 mm in diameter and 9.929 mm in length, piece of AISI316Lstainless steel rod was used as reference.

Statistical analyses

The statistical tests were made with SPSS 14.0.1 software (LeadTechnologies, Inc., Chicago, IL, USA). Normality of distributionswas determined through the use of the Shapiro-Wilk test. The sta-tistical significance was determined through the use of t-test of in-



dependent samples and Kolmogorov-Smirnov Z-tests for normallydistributed parameters and others, respectively. All the preparedsamples (n = 55) were tested.

Analyses of the measurements

The off-line analyses were performed through the use of MatlabR2008a software (Mathworks Inc., Natick, MA, USA). The TOF wascalculated from the maximum value of the signal using the Hilberttransformation. The speed of sound (c) was calculated by dividingthe measured tablet thickness by the TOF. The frequency spectrumof the US transmission was determined by using the Fast FourierTransform (FFT) algorithm. Only the first pulse of the measuredUS signal was used. The US attenuation of tablets was also ana-lyzed. To be sure that the attenuation measurements were donein the farfield, the near field distance ’N’ was estimated by using(3.13), as given by the manufacturer of the US transducers, wherethe diameter of the US transducer element is 3 mm, the frequency10 MHz and the speed of sound in the steel is 5800 m/s. Thesevalues result in N = 39 mm. The punch shaft works as a delay linefor the US wave. The length of both shafts was 20 mm (one perpunch), thus the sample is situated in the farfield as the minimumlength for wave to propagate is 40 mm.

5.3.4 US monitoring of swellable matrix tablet fronts

movement (IV)

Eroding front determination

Six identical samples were measured simultaneously in each test.Samples were set into a line with a distance of 30 mm betweenthe center points of the samples. Polished stainless steel was usedas the reference for amplitude normalization. The sample surfacewas adjusted to a distance similar to the focal length of the UStransducer. The reflected US echo was measured in 0, 10, 30, 60,120,180, 240 and 480 min. The speed of sound (c) was measured



in 0.1 M phosphate buffer (PB, pH 6.8), as specified in the Ph.Eur.The tablets were immersed into degassed PB. Acoustic measure-ments were conducted at room temperature and the measurementtime, i.e., total of 8 h, was used for all tests. All the tablets wereline scanned along the tablet diameter using the ultrasound pulse-echo measurement mode to investigate the movement of the erod-ing front. The distance between the samples and the US transducerwas 50 mm at the beginning of the immersion. The measurementwere done with a custom made LabViewTM 6.5 (National Instru-ments, Austin, TX, USA) software and the data were stored foroff-line analysis. The signal envelope was determined by utiliz-ing the Hilbert transform. The maximum of the signal envelope, i.e.the reflection from the interface between two materials (PB-erodingfront), was used to define the correct TOF. The distance (d) betweenthe sample surface and the transducer was determined by usingthe recorded TOF and the predetermined speed of sound (c) in themedium, using (3.5): d = c

2TOF . The PE measurement data wasused to determine the tablet surface displacement. The vertical lo-cation of the eroding front was determined from the sample surfaceat each time point. An average of 10 subsequent signal echoes wasused at each measurement point.

Swelling front determination

The acoustic determination of the movement of both the swellingand eroding fronts were investigated for all polymer tablets. Thesample holder was made of a polymethyl metacrylate (PMMA) rod,30 mm in diameter and 70 mm in length (Fig. 5.5). PMMA waschosen as a transparent material with appropriate acoustic prop-erties [114, 170]. A hole with 10 mm diameter was drilled axiallythrough the rod. The upper surface of the rod was flat milled toobtain a flat faced interface for the US measurements and visualcamera monitoring and also in order to minimize the ultrasoundscattering within the US window.

The distance between the transducer and the sample was set to



US

A B A(x)

Tablet

Figure 5.5: A) The US window sample holder made of PMMA. Steel bolts are used asposition manipulators and sinkers. B) The principle of US window measurement, whichresults the amplitude A as a function of location.

the focal length of the transducer to achieve maximal intensity fromthe distance of PMMA-sample interface. The US scanning step sizewas 0.1 mm and the US beam orientation was set to be parallel tothe sample radius. The US PE along the tablet axis was measuredand fast Fourier transform (FFT) was used for spectral calculationas a function of the scanning location. Eroding and swelling frontdisplacement were determined by using the first order derivative ofthe US echo intensity at 3.4 MHz frequency, which was observed toprovide the highest sensitivity for the application, this being chosenafter preliminary experiments conducted with the transducer.

Finally, the US window measurement was compared to frontdisplacement data obtained by optical monitoring. The optical es-timation of the swelling and eroding front location was obtainedby microphotography. The sample was photographed through anOlympus SZ-60 stereomicroscope (Olympus Optical Co. Ltd., Tokyo,Japan) with an Olympus C-5050 digital camera (Olympus OpticalCo. Ltd., Tokyo, Japan) with remote triggering in order to get ridof the mechanical operator based vibrations. The photographs atthe set time points were analyzed by using the measurement toolof Gimp v2.1–software (http://www.gimp.org). The thickness ofthe tablet structures was measured from those microphotographs



from which the gel layer and dry core of the tablet could be clearlyseen (Fig. 5.6). The optical measurement was made and used as areference method for the US window method.

Figure 5.6: Immersed swelling polymer tablet. Liquid penetration forms eroding (outer)and swelling (inner) fronts. The gel layers are visible on each side of the dry core of theswelling tablet of total thickness equal to two gel layers and a core.

Acoustic measurements

All acoustic measurements in study IV were conducted using anUltraPACsystem (Physical Acoustics Corporation, Princetown, NJ,USA) which consisted of a tank and 3D-scanning drives, a high fre-quency A/D-board (PAC-AD-500) and focused ultrasound 5 MHztransducer (Panametrics V307, Panametrics Inc., Waltham, MA, USA)with a beam diameter of 0.6 mm. The measurements were doneusing the pulse-echo (PE) geometry with 62.5 MHz sampling fre-quency. The detected US signal was digitally filtered with a thirdorder (18 dB/octave) Butterworth 0.05–10 MHz band-pass filter.




6 Results

This chapter summarizes the results from studies I–IV. The com-plete results can be found in the original papers attached to end ofthis thesis.

6.1 ACOUSTIC EMISSION AS FOOTPRINT OF MOVING PAR-

TICLES (I)

Flash topography, multi-point NIR techniques and acoustic emis-sion method were applied simultaneously to monitor fluidized bedprocesses and measure, directly or indirectly, particle size and mois-ture content changes in a labscale top spray granulator. The flu-idized particles granule size was estimated with AE and TOPOtechniques in a test with commercial dry granules. The estimatedmean particle sizes are shown in Table 6.1. The RMS error of esti-

Table 6.1: The reference mean diameter DRef vs. estimated diameters DAE and DTOPOmeasured with acoustic emission (AE) method and flash topography (TOPO), respectively.

Label DRef(μm) DAE(μm) DTOPO (μm)

Cellets100 150 136 111Cellets200 255 297 210Cellets350 425 425 381Cellets500 605 579 531Cellets700 850 899 775Cellets1000 1200 1172 1068Cellets100/1000 675 740 -Cellets350/500 515 503 -Cellets200/350/500 436 450 -Protease A 481 451 -Protease B 534 552 515

mated size values was calculated with Eq. (5.6). RMSE for AE andTOPO methods was RMSEAE = 9.8 μm and RMSETOPO = 26.8 μm.



The AE overestimated and TOPO underestimated the particle sizethroughout the measurements as the estimation biases [102] were4.8 μm and -55.6 μm, for AE and TOPO, respectively.

The granulation process was monitored with AE and multi-point NIR. The mean granule size as a function of time is shownin Fig. 6.1. In the caffeine granulation process experiment, the

AE

NIR

Figure 6.1: The mean granule size estimated with AE and four moisture profiles monitoredwith NIR as a function of time during pharmatose/caffeine granulation process. The periodmarked with two red vertical lines in the figure is the wetting stage.

mean particle size obtained with sieving was 129 μm at the begin-ning of agglomeration and 530 μm at the end of drying, as theywere estimated using AE to be approximately 100 μm and 500 μm,respectively.


Results

The particles began to grow rapidly after wetting was started.The first size estimates of particle size were about 100 μm. Themaximum mean particle size was 650 μm. The granules reachedtheir maximum size slightly after the end of the wetting stage, in1200 seconds after the start of the process. It was observed, thatdrastic changes in the process’ particle size values, e.g. in 800 and1200 seconds, were caused by wet powder mass that dropped offthe chamber wall during the process.

Four of eight measured NIR channels are shown in Fig. 6.1. In-terestingly, there were an asymmetry between the moisture profileswhen comparing the right and left sides to one another. Addition-ally, the end level of the moisture estimated with NIR was observedto be greater than in the beginning of the process, except the chan-nel L2.

6.2 ULTRASOUND MEASUREMENT DURING TABLET FOR-

MATION (II)

An ultrasound (US) measurement system for tablet compressionmeasurements in an actual tabletting environment to evaluate thefeasibility of the system was introduced. US measurements withdifferent powder formulations were done during the tablet com-pression process. The measurement system is based on the UStransducers implemented inside flat-faced punches. In Fig. 6.2, thein-line US and compression force measurements are shown. The to-tal time of flight including the 13 μs of wave propagating throughthe steel punches was approximately 14 μs. Therefore, the time offlight in tablet material under compression was ≈ 1μs. The time de-pendence of the speed of sound can be seen clearly in Fig. 6.3. Thecompaction of powder decreases the void fraction in the compactedvolume and the US transmission can be measured.

The dependence between the tensile strength and the maximumspeed of sound, cmax, is shown in Fig. 6.4. cmax in the tablets de-pends on the tensile strength of tablets. However, the dependencewas decreased when the MS mixing time was increased (Fig. 6.4A).



Figure 6.2: (A) The real-time ultrasound during dynamic compression process ofparacetamol–avicel tablet with 2 min mixing time of magnesium stearate. (B) US sig-nal for 10 min mixing time. Measured forces of upper (solid) and lower punch (dashed)during compression for (C) 2 and (D) 10 min of MS mixing. The peak forces used were15175 and 19865 N, respectively.

In the Fig. 6.4B, cmax as a function of the tensile strength for tabletswithout MS is shown. The different PRC concentration groups sep-arate in tensile strength and cmax.

In Fig. 6.5, the correlation between the maximum speed ofsound (cmax) and the compression force for all tablet formulationsused in the study is illustrated. The correlation is very strong(r = 0.99).


Results

Figure 6.3: Mean of measured speed of sound and the distance between punches (solid line)as a function of time during compression with (A) low (7.5(◦) or 8.6(×) kN ) and (B) high(14.8(◦) or 19.0(×) kN) compression load. Measurements after 2 min MS mixing (◦) andafter 10 min of MS mixing (×). The gray area stands for the dwell-time of compression.



2.5 3 3.5 4 4.5 5 5.52200

2300

2400

2500

2600

2700

2800

Tensile strength, � [MPa]

Spe

ed o

f sou

nd, c

[m/s

]

B1: 0%B2: 5%B3: 10%

1 1.5 2 2.5 3 3.52200

2400

2600

2800

3000

3200

3400

3600

3800

Tensile strength, � [MPa]

Spe

ed o

f sou

nd, c

[m/s

]

A1LA1HA2LA2H

A) B)

Figure 6.4: The maximum speed of sound as a function of tensile strength of tablets A)with and B) without MS.

0.6 0.8 1 1.2 1.4 1.6 1.8 2

x 104

2200

2400

2600

2800

3000

3200

3400

3600

3800

Spe

ed o

f sou

nd, c

[m/s

]

Compression force, F [N]

r = 0.99

A1LowA1HighA2LowA2HighB1: 0%B2: 5%B3: 10%

Figure 6.5: The maximum speed of sound as a function of compression force. The tabletswithout MS are plotted with red color.

6.3 ULTRASOUND DETERMINATION OF TABLET INTEGRITY

(III)

The tablet integrity determination using US transmission was stud-ied. The used technique based on the attenuation of US signal bytablet medium. In Fig. 6.6, US measurements of an example intactand defected DCP tablet is shown. The US transmission is strongerthrough (Fig. 6.6a) an intact than (Fig. 6.6b) defected tablet. Thespectrum was calculated from the measured signals and the attenu-


Results

Figure 6.6: US signal and frequency spectra measured through a,c) intact and b,d) defectedDCP tablets. The nominal frequency of the instrumented ’punch’ transducer is about 4.4MHz (spectrum is shown in Paper III).

ation on each frequency band was calculated using Eq. (3.9). The at-tenuation values are shown in Table 6.2. The tablet integrity was de-termined by attenuation difference between the intact and defectedtablets. The determination of tablet integrity could be done for DCP,MCC and LM tablets. However, it was found that the most efficientfrequency for defect recognition was different for each excipients.The most suitable frequency range for DCP, MCC and LM integritytesting using US transmission were 2.4–4.2 MHz, 4.2–5.4 MHz and1.2–1.8 MHz, respectively. Using these frequencies, the significancelevel of difference between intact and defective tablets were 0.001for DCP and MCC and 0.01 for LM.



Table 6.2: Attenuation difference by frequency, Δα f , between intact and defected tablets.∗, ∗∗ and ∗∗∗ as difference significance level of 0.05, 0.01 and 0.001, respectively.

f (MHz) Δα f (dB/mm)DCP MCC LM

1.2 1.4∗∗∗ 0.0 1.8∗∗1.8 0.7 0.3 1.6∗∗2.4 2.2∗∗∗ 0.9 1.33.0 2.3∗∗∗ 1.0∗ 0.13.6 2.4∗∗∗ 1.1∗ -0.94.2 2.9∗∗∗ 1.3∗∗ 0.04.8 3.2∗∗ 1.3∗∗ 0.15.4 3.5∗∗ 1.6∗∗ -0.2

6.4 POLYMER TABLET SWELLING MONITORING WITH UL-

TRASOUND ECHO (IV)

The feasibility of ultrasound pulse echo measurement techniquesfor investigating swelling behavior of immersed hydrophilic matrixtablets was tested. For this purpose, two different US measure-ment methods and one reference method based on optical exam-ination were used. In first test, the direct echo measurement oferoding front displacement was tested. The eroding front displace-ment during tablet immersion was evaluated for HPMC and PEObased tablets as shown in Fig. 6.7A-C. The dry core and the gel

Figure 6.7: Eroding front displacement measured with direct US echo method. The solidline is the mean of six replicates and the dashed line is the mean±1 S.D.


Results

layer of the HPMC tablets remained intact at the last measurementpoint, where the total eroding front displacement was determinedto 2.5 ± 0.4 mm. The PEO0.1 tablets swelled rapidly and dissolvedtotally within 120 min. The PEO2.0 tablets sustained during the ob-servation time and had an eroding front displacement of 1.7 ± 0.5mm measured after 8 h. PEO0.1 tablet size decreased continuouslyas a result of tablet hydration (Fig. 4C) indicating a very fast disso-lution process.

In the second experiment, the feasibility of the developed ultra-sound window technique was tested. The US window was usefulfor measurement of the swelling front location of immersed poly-mer tablets. The technique was evaluated with an optical methodbased on digital image analysis (DIA). In Fig. 6.8A-C, the swellingand eroding front location results are shown. The US windowswelling results have good correlation with measured reference (Fig.6.8D). The eroding front was more difficult to measure correctlywith US window method (Fig. 6.8B,C).



A B

C D

Figure 6.8: Eroding and swelling front locations as a function of immersion time measuredwith US echo method for A) HPMC, B) PEO2.0 and C) PEO0.1 polymer tablets. D) Theswelling front location determined with US window vs. reference optical imaging method.


7 Discussion

The information obtained with these techniques, needs interpreta-tion. This is discussed in order to couple the process variables intothe analytical knowledge behind the measurements.

7.1 GRANULATION PROCESS MONITORING TOGETHER

WITH AE, TOPO AND MULTI-POINT NIR

Both the topographic camera and the acoustic emission spectroscopydemonstrated their potential as real-time measurement systems pro-viding precise particle size analysis over a wide particle size range.Further, multi-point NIR (using eight probes) and acoustic emissionmethods were able to detect the three granulation process phases,mixing, agglomeration and drying.

In the present study, the mean particle diameter of Cellets R©and Protease granule samples was successfully determined with amethod based on the normalized acoustic intensity values of threefrequencies calibrated with different sizes of Cellets R© granules.Both of the AE spectrometry and TOPO were successful methodsfor determining particle size of the studied samples as the predictedparticle size of the sample was comparable to the particle size mea-sured by the off-line reference method.

According to RMSE and bias for AE and TOPO methods, AEwas more accurate and precise compared with TOPO in this par-ticular study. The TOPO underestimated the results approximately55 μm according to the bias, while AE overestimated the particlesize approximately 5 μm. However, particles must move duringparticle estimation with AE method in order to emit sound by dryparticles.

The advantage of AE as pharmaceutical process monitoring toolis the ease of instrumentation. The signal can be detected from onespecific location although a signal is emitted from the whole cham-



ber surface. The acoustic emission is a very powerful method in theparticle size estimation of moving particles. However, the moisturecontent affects the acoustic response. As the moisture changes inthe wetting and drying processes, the particle mechanical proper-ties, e.g. mass, density and elasticity change. The acoustic emis-sion technique is known to be sensitive to the system mechanicalchanges [82] and the association between AE and the particle sizeof the sample is reported [21]. However, studies considering the re-lationship between the size of a pharmaceutical granule and acous-tic emission in fluidized bed are rare [91,115,149]. Particle behaviorin a fluidized bed granulator can be monitored and characterizedby assessing the sounds, once the correlation between particularsounds and particle motion is established [149].

Acoustic emission is traditionally considered as a sensitive toolto monitor changes in processes. Different stages of the granula-tion process could be clearly distinguished with the AE measure-ment. In an earlier study, correlation between measured AE inten-sity without spectral analyses and the granule properties was notevident [146]. However, the granule size and water content of gran-ules could be determined during the process by the multivariatemethods able to preserve and extract physical information from theacoustic emission spectra of fluidization and granulation [91].

The granule size estimation was challenging due to changingmoisture content in this study, also. The drastic changes in granulesize were observed during the wet granulation process. Most likely,this was due to wetting induced powder mass sticking on and dry-ing induced peeling off the chamber walls because the formulationhad a tendency to stick on the granulator walls throughout the pro-cess.

It is previously reported, that the granule properties, such asparticle size and water content, may be totally heterogeneousthroughout the granulation process [103]. It was perceived afterpublishing the Paper I, that the used fluidized bed granulator hasa particular geometric design that has an asymmetry of input airflow (unpublished data). The fluid dynamics simulation image of the


Discussion

empty (air filled) chamber is shown in Fig. 7.1.The multi-point NIR of fluidized bed granulation was detecting

large differences between channels, e.g. they exhibit very differentbaseline offsets which may be attributed to the varying light inten-sity measured by the different probes. Variations in both the pow-der packing density in front of the probe window and the moisturecontent of the powder influence the intensity of backscattered light.As partly responsible for it, the used fluidized bed granulator hada geometric design with asymmetry with input flow. It probablycaused uneven distributions of moisture and particles in fluidiza-tion. Although the simulations could not confirm NIR estimations,it was assumed that the differences in conditions between channelsexisted during measurements.

Figure 7.1: Simulated empty chamber fluid dynamics of fluidized bed used in the study I.

Moisture is one of the most important parameters in fluidizedbed granulation: liquid binders are needed to produce granulesbut excess moisture may cause bed collapse during the process. Tooptimize the granule properties by adjusting process parameters,the moisture content of the granules should be measured with suf-ficient accuracy in-line. The estimated relative temporal moistureprofiles are illustrated for the four lowest channels. As the datawere not calibrated, the moisture values are given in arbitrary units



(A.U.). Changes in the water content were the most apparent inthese four channels which are located 10–15 cm above the bottomscreen of the granulator. The moisture level is seen to remain rela-tively constant in all channels during the mixing phase. The binderspraying started to moisten the fluidized powder leading the pow-der bed inhomogeneously or excessively wet. It is thus expectedthat the NIR measurement is fully noninvasive and the presence ofprobes do not disturb the flow properties of the system. This is alsothe case with the AE, because the probes are outside the granulatorchamber and do not interact with the materials or conditions insidethe chamber.

The present instrumentation involves simultaneous moisturemeasurement at different locations at which different moisture con-tent is expected and calibration should be performed off-line anda calibration transfer should be used in order to evaluate actualwater content in-line. However, the power of multi-point NIR spec-troscopy lies in the fact that estimates for the moisture level maybe gathered simultaneously at multiple locations in the granulationchamber. The fluidized powder mass may be very heterogeneouswith respect to the moisture level due to the process parametersand granule properties. Multi-point NIR enables a quick detectionof such heterogeneity.

7.2 US TRANSMISSION FOR TABLET FORMATION MONI-

TORING

The ultrasound transmission was measured as a function of timeduring tablet formation process. It was shown that the instrumen-tation using polymer glue as acoustic couplant between the trans-ducer and the tablet press punch (Fig. 5.3B) for attachment wasfeasible.

The compression process was monitored for each tablet usingDAAS system. During the compression cycle, the thickness of thepowder bed compression forces and the speed of sound were mea-sured. The reference US was measured by pressing the punches


Discussion

together. The measured reference TOF was 13 μs including the de-lays of the system. The measured US signal of tablets mixed for 2and 10 minutes is shown in (Fig. 6.2). The acoustic mismatch be-tween powder and metal used in the punches has been recognizedas the main problem of in-die compression monitoring [89]. As canbe seen in (Fig. 6.2A,B), this acoustic mismatch was not a prob-lem using the presented measurement system that is based on thethrough transmission technique.

The US pulse was detectable when loading was high enoughfor adequate acoustic coupling. The loading for proper transmis-sion was around 2 kN in this experiment. Interestingly, it wasfound that speed of sound changes, not only when the punchesare moving, but also during static compression. During the timebetween 535 and 575 ms, the gray area in Fig. 6.3, the speed ofsound changes constantly without the moving of punches. This in-dicates the mechanical properties of the powder bed are changing(i.e., the formation of the tablet) as a function of time.

In addition, the tensile strength and speed of sound were mea-sured (Fig. 6.4). The extension in MS mixing time probably filledin pores between the granules and possibly coated the granules inthe powder bed and, therefore, increased the speed of sound values(Fig. 6.4A). The filling of pores due to extended MS mixing andfilm formation on the granules, lead to improved acoustic couplingbetween the granules during compression. As a side effect of MSovermixing, the tablets mixed for 10 minutes became mechanicallyweaker. Nevertheless, the tensile strength changed as a function ofthe speed of sound within the formulation under investigation (Fig.6.4A).

The method was also found to be sensitive to the MCC contentin the formulation (Fig. 6.4B). Since the speed of sound was foundto be sensitive to the mixing time of MS, the effect of PRC concen-tration to the speed of sound was studied without MS. The speed ofsound was measured as a function of time during the dwell-time of40 ms at the interval of 1 ms. During the measurement, the speedof sound increased from 2431, 2274 and 2064 m/s to 2719, 2589 and



2309 m/s for 0, 5 and 10%(w/w) PRC concentration tablets, respec-tively. This happened during static compression, i.e., between 535and 575 ms of the tabletting process as the volume (density) of thecompact was constant. Therefore, the change in the speed of soundwas caused by the change of mechanical properties due to bondformation and the deformation of the powder during compression.The speed of sound was different for different PRC concentrationspartly due to different porosities. Thus, the speed of sound mea-sured during the dwell time is sensitive to the PRC concentration.

The speed of sound during compression increased with the ten-sile strength of tablet. The effect of MS mixing time to speed ofsound were calculated at 538 ms of tabletting cycle, i.e. in the low-est position of the upper punch. The diametrical tensile strengthof tablets with 10 minutes mixing period was lower than with twominutes of mixing. The speed of sound increased with the ten-sile strength. However, the final tensile strength of the tablets cannot be predicted only with the (longitudinal) speed of sound mea-surements. This finding is obvious and shown in Fig. 6.5. Themaximum compression in tabletting strongly correlates (r = 0.99)with the speed of sound of tablets. Thus, the comparison of thespeed of sound is not straightforward, because the speed of soundis known to be inversely proportional to the porosity [57, 136].

The measurement time window and interval are adjustable mak-ing the measurement system flexible and suitable for monitoringthe different phases of the compression process. The ultrasoundwas found to be sensitive to the changes of the mechanical proper-ties of the powder bed. Thus, the ultrasound measurements madeduring the tablet compression give more information on the tabletformation, especially made with the force-distance and/or in-diepressure measurements.


Discussion

7.3 US TRANSMISSION AS A TOOL FOR TABLET INTEGRITY

TESTING

An in-line ultrasound measurement system was introduced for thedetection of tablet defects. The measurement system is based on ul-trasound transducers that are positioned inside the flat-faced tablet-ting machine punches. One transducer emits a short ultrasoundpulse that propagates through the tablet while the other transducerreceives the transmitted pulse. This system was implemented inan eccentric single station tabletting machine. Its performance wastested by using both intact and defective tablets that were madewith three different excipients. The speed of sound and ultrasoundattenuation was determined from the transmitted ultrasound sig-nal.

During compression, particles go through various deformationphases, namely fragmentation and both elastic and plastic deforma-tions. If the amount of elastic deformation is high, elastic recoverymight break existing permanent interparticle bonds and cause thetablet to cap or laminate. Capping generally refers to the lid of abiconvex tablet that is separated from the compacted tablet [42]. Inlamination, small cracks generate layers within a tablet, parallel tothe punch face. Recently, many studies have been carried out tofind measurement systems for defect detection in tablets. The pur-pose behind such studies is to better understand the processes oftabletting for online control, which is emphasized in the FDA’s PATguidance monograph [153].

This goal can be achieved by using real time measurement sys-tems in these processes. As capping and lamination change themechanical properties of the tablet, acoustic measurement systems,which are sensitive to mechanical changes, have been studied exten-sively. One of the first monitoring systems for defect tablet detec-tion was a measurement system based on AE [65, 131, 159]. Mecha-nisms of deformation generate AE signals that can be detected, andthese acoustic responses may give information on the compressionprocess during tablet formation. In these measurement systems,



AE sensors were attached to the side of the upper punch and thesystem was tested in a single tablet production machine. From theacquired data, probability distribution was calculated and used toclassify capped and non-capped tablets. As classification is basedon probability, the system correctly classified 95% of the cappedtablets [65].

Ultrasound is a mechanical wave that includes certain wave-lengths, propagating in a medium. It is widely used for nonde-structive testing in many different areas. Recently, ultrasound basedmeasurement systems have also been introduced in pharmaceuti-cal research. Since ultrasound is a mechanical wave, the speed ofsound is sensitive to mechanical properties and thus it has beenused to determine the porosity and elastic modulus of tablets andcoating thickness [4, 7, 57, 74]. Other acoustics based techniques fortablet defect measurements have been published recently includ-ing photo-acoustic [157], air-coupled acoustic [4, 5] and acousticresonance spectroscopy [94, 95] measurements. However, none ofthese techniques have not been suitable for embedding them in to atabletting apparatus as thay were, and therefore their suitability asPAT tools should be evaluated.

The speed of sound values were found to be independent of theapplied frequency. Moreover, it was observed that an increase infrequency did not increase the high frequency content in the trans-mitted frequency spectrum. Thus, in couplant-free ultrasound mea-surements, high frequencies are attenuated. The speed of soundmeasurements were also made by using instrumented tablet presspunches. Results were the same as those obtained with contacttransducers.

Thus, the instrumented tablet press punches can be used for ul-trasound measurements, despite the fact that their centre frequencyshifted from 10 to 4.5 MHz. One should note that the force betweenthe punch and the surface of tablet was higher than in contact trans-ducer measurements, so the speed of sound values are more co-herent in the punch measurements than in the contact transducermeasurements. This observation verifies that ultrasound measure-


Discussion

ments can be made without additional couplants when the force be-tween transducer and tablet is sufficient. The speed of sound valuesfrom defective tablets were also measured by using instrumentedpunches, and comparing the speed of sound values between theintact and defective tablets show that they are different. However,based on the statistical analyses, the speed of sound values cannotreliably discriminate defective from intact tablets.

Ultrasound attenuation was found to be a very selective andsensitive analysis method that discriminated between intact and de-fective tablets. Only two samples were discriminated incorrect. Apossible reason for this may be related to the poor signal-to-noiseratio for LM samples. In addition, the ultrasound attenuation (Table6.2) was found to have different behaviors that were apparently de-pendent on the physical properties of the material being measured.All the measured materials had different US frequency responses.However, a possibility for identifying different powdered materialsusing ultrasound attenuation measurements need further investiga-tion.

In addition, the signal analysis needed for discrimination isbased on relatively simple Fast Fourier Transform calculations. Thesecalculations can be made in real time, using real time data acquisi-tion hardware. Thus, this application fulfills all the required char-acteristics as a PAT device for in-line ultrasound detection duringthe compaction of pharmaceutical tablets. [153]

7.4 US ECHO AS A TOOL FOR POLYMER SWELLING MON-

ITORING

The US echo techniques were used for detecting hydrophilic poly-mer swelling front movement in phosphorus buffered solution im-mersion. The eroding front displacement during tablet immersionwas evaluated for HPMC and PEO based tablets as shown in theFig. 6.7.

The results from the US echo measurement suggest that themethod is suitable to be used non-invasively to study the swelling



of hydrophilic polymers and give real-time size information. How-ever, the structural investigation was restricted to the outer surface,in agreement with an earlier study [77].

A novel method, US window, was developed in order to probeboth the swelling and the erosion fronts simultaneously. The frontsdefining the gel layer was determined by using the derivative ofthe US amplitude measured as a function of time. The 3.4 MHzfrequency was the most sensitive for the detection.

Systematic differences was found in the HPMC eroding andswelling front movement, as measured with US window and op-tical reference methods (Fig. 6.8A). The positions of the fronts de-termined by US are 0.6–1.0 mm smaller than those of determinedwith the optical method. The HPMC gel layer thickness at 4 hwas 4.9 ± 1.0 mm and 5.5 ± 1.0 mm (Fig. 6.8A) by the US windowand optical methods, respectively. Even though, there were difficul-ties to measure the eroding front displacement with both methods,the swelling front movement results were found to coincide. Inthis case, the eroding front movement was faster during the opti-cal measurement compared with the US method and the gel layerthickness measured with US was nearly two times larger than thatof measured with the optical method.

The geometry used in US window method was found challeng-ing to use with low viscosity grade PEO0.1, even for the opticalinvestigation used as the reference, because the gel layer formed byPEO0.1 resembled more a viscous fluid than a rubbery gel. The re-sults obtained with PEO0.1 (Fig. 6.8B) are similar to the previouslypublished data [79], where PEO0.1 tablets dissolved totally within2–3 h in distilled, 25 ◦C water. It was shown with the PEO2.0 mea-surement (Fig. 6.8C), that the US window method gave values forthe erosion front positions that were significantly lower than thosedetermined with the optical method and the gel layer thickness wasunderestimated by one third (Fig. 6.8D). However, the US windowmeasured swelling front displacement results were nearly similarto the results with the optical method for all polymers used in thestudy. There was a systematic difference of 0.6–0.8 mm between US


Discussion

window and optical measurement values at the beginning of themeasurements. This was likely result of the transducer characteris-tic axial resolution, i.e. the beam diameter of 0.6 mm.

In the US window, the geometry was restricted and blockedthe solution to penetrate only into the flat faces of polymer tabletwhich was not the case in the original US echo measurement. Thus,the measurement geometries of the methods were different and theresults cannot be compared to one another. However, qualitativecorrespondence between the methods was obtained, and strong cor-relation (r > 0.95) was found between the values obtained with theultrasound window and optical method in the detection of swellingfront for all polymers used in the study (Fig. 6.8D). Therefore, it issuggested that the US window method is suitable technique for de-termining the inner structure of hydrophilic swelling polymers.

Another point to be considered is related to the detection ofthe interface between the inner (glassy) core and the swelling poly-mer gel. The spatial resolution of the US beam at a frequency of3.3–6.7 MHz is approximately 250–500 μm in a water-like medium.The thickness of the gel layer is only a few millimeters at its max-imum. Although it is possible to measure detectable echoes fromboth interfaces (medium-gel and gel-core), echoes may (partiallyor completely) overlap due to limited spatial resolution, especiallyduring the early stages of polymer swelling when only a thin gellayer is observed.

In addition, a part of the signal is reflected and scattered de-pending on the differences in the acoustic impedance between thematerials, the heterogeneity of the gel layer and the propagationdirection of the beam compared to the interface. These are factorsthat are dependent on the investigated material and may limit theuse of the method in some applications. Using transducers withproper frequency range and focusing parameters, one might over-come these limitations and tune the transducer properties to be ad-equate for any particular application.




8 Summary and conclusions

8.1 SUMMARY

Advanced process-control technologies are required for process mon-itoring in the pharmaceutical industry. The present study aimed atdeveloping suitable active and passive acoustic techniques in orderto measure, monitor and evaluate processes involved in pharma-ceutical tablet manufacturing. The used tools were implementedusing commercial devices. The transducer and sensor instrumenta-tions were tailored and custom-made softwares were used to fulfillneeds of achieving certain physical parameter from each manufac-turing unit process including granulation, tabletting, tablet defectdetection and swelling.

8.2 CONCLUSIONS

The following can conclude the findings of this research:

• Passive acoustic methods for monitoring of a pharmaceuticalfluid bed granulation process combined with other methodsas flash topography and multi-point NIR offer valuable in-formation about the particle properties. The techniques canbe used simultaneously to receive in-line moisture and sizeinformation of fluidizing particles.

• Ultrasound transmission can be used for real-time tablettingformation monitoring. Instrumented tabletting apparatus wasused to measure the speed of sound in the binary mixturesof pharmaceutical powder materials during tabletting. Thespeed of sound was found to be sensitive to the mixing timeof magnesium stearate. In addition, the speed of sound wasobserved to change during the dwell-time of the compres-sion cycle. Therefore, this experimental study showed that



the change of the material mechanical properties can be mea-sured during compaction by US.

• Developed couplant-free ultrasound techniques were used asa detecting tool of the mechanical integrity errors of tabletsand showed great potential for in-line quality control appli-cation. As transducers are located inside the press punches,and measurements are made within a few microseconds, thissystem can be used in the real time defect detection of phar-maceutical tablets.

• Detection of the swelling front is more challenging than theeroding front. It was found that the US window techniqueintroduced in the study was a promising method for simulta-neous multi-front detection. The direct US pulse echo methodwas highly challenging when used for simultaneous detectionof both the eroding and swelling fronts, whereas US windowis more suitable for this due to the design of measurement ge-ometry where all the layers are detectable with US beam. TheUS was found feasible for the swelling process monitoring ofhydrophilic polymer tablets.

The studied techniques have large potential and provide tools forpharmaceutical process monitoring and nondestructive quality con-trol. They fulfill the definition of the PAT tools and can offer infor-mation about the mechanical properties of processed materials. Itwas found that the acoustic emission technique can give direct in-formation from particle size evolution during the wet granulationprocess as the ultrasound transmission can be used in real-timetabletting process monitoring or tablet integrity determination. Fi-nally, the use of the ultrasound window technique provided previ-ously unreachable information for acoustic methods about the gellayer thickness of the immersed hydrophilic polymer tablets.


Bibliography

[1] Aaltonen J, Kogerman K, Strachan CJ, Rantanen J. In-line monitoring of solid-state transitions during fluidisation. Chem Eng Sci 62 (1–2), 408–415, 2007.

[2] Abrahmsén-Alami S, Körner A, Nilsson I, Larsson A. New release cell forNMR microimaging of tablets swelling and erosion of poly(ethylene oxide).Int J Pharm 342, 105–114, 2007.

[3] Akseli I, Libordi C, Cetinkaya C. Real-time Acoustic Elastic Property Moni-toring of Compacts During Compaction. J Pharm Innov 3, 134–140, 2008.

[4] Akseli I, Cetinkaya C. Air-coupled non-contact mechanical property determi-nation of drug tablets. Int J Pharm 359, 25–34, 2008.

[5] Akseli I, Mani GN, Cetinkaya C. Non-destructive acoustic defect detection indrug tablets. Int J Pharm 360, 65–76, 2008.

[6] Akseli I, Cetinkaya C. Drug tablet thickness estimations using air-coupledacoustics. Int J Pharm 351, 165–173, 2008.

[7] Akseli I, Hancock BC, Cetinkaya C. Non-destructive determination ofanisotropic mechanical properties of pharmaceutical solid dosage forms. IntJ Pharm 377, 35–44, 2009.

[8] Akseli I, Becker DC, Cetinkaya C. Ultrasonic determination of Young’s moduliof the coat and core materials of a drug tablet. Int J Pharm 370, 17–25. 2009.

[9] Akseli I, Dey D, Cetinkaya C. Mechanical Property Characterizationof Bilayered Tablets using Nondestructive Air-Coupled Acoustics. DOI:10.1208/s12249-009-9352-9. AAPS PharmSciTech 2009.

[10] Akseli I, Iyer S, Lee HP, Cuitino AM. A Quantitative Correlation of the Ef-fect of Density Distributions in Roller-Compacted Ribbons on the MechanicalProperties of Tablets Using Ultrasonics and X-ray Tomography. AAPS Pharm-SciTech 12(3), 834–853, 2011.

[11] Alderborn G and Nyström C. (Ed.) Pharmaceutical Powder CompactionTechnology, Marcel Dekker, Inc., 1996. New York, NY, USA.

[12] Alderborn G. Tablets and compaction, chapter in: M.E. Aulton (Ed.), Phar-maceutics: the Science of Dosage Form Design, 2nd Edition, Churchill Liv-ingstone, Edinburgh, 2002.



[13] Anderson N, Palfrey L. The effect of machine speed on the consolidation offour directly compressible tablet diluents. J Pharm Pharmacol 41, 149–151,1989.

[14] ASTM Standard method for primary calibration of acoustic emission sensorsASTM Standard E1106, 492–501, 1992.

[15] ASTM Standard guide for determining the reproducibility of acoustic emis-sion sensor response ASTM Standard E976, 380–385, 1994.

[16] Aulton ME, Summers M. Granulation, chapter in: M.E. Aulton (Ed.), Phar-maceutics: The Science of Dosage Form Design, 2nd ed., Elsevier Limited,Churchill Livingstone, 364–378, 2002.

[17] Bajwa GS, Hoebler K, Sammon C, Timmins P, Melia CD. Microstructuralimaging of early gel layer formation in HPMC matrices. J Pharm Sci 95, 2145–2157, 2006.

[18] Banks M, Aulton ME. Ind Pharm 17 (11), 1437–1463, 1991.

[19] Batlle J, Mouaddib E, Salvi J. Recent progress in coded structured light as atechnique to solve the correspondence problem: a survey. Pattern Recogn 31(7), 963–982, 1998.

[20] Baumgartner S, Lahajnar G, Sepe A, Kristl J. Quantitative evaluation of poly-mer concentration profile during swelling of hydrophilic matrix tablets using1H NMR and MRI methods. Eur J Pharm Biopharm 59, 299–306, 2005.

[21] Belchamber RM, Betteridge D, Collins MP, Lilley T, Marczewski CZ, WadeAP. Quantitative study of acoustic emission from a model chemical process.Anal Chem 58, 1873–1877, 1986.

[22] Benbow JJ. Chapter in: Stanley-Wood, NG, (Ed.) Enlargement and com-paction of particulate solids. London: Butterworths, 171, 1983.

[23] Bikiaris D, Koutri I, Alexiadis D, Damtsios A, Karagiannis G. Real time andnon-destructive analysis of tablet coating thickness using acoustic microscopyand infrared diffuse reflectance spectroscopy. Int J Pharm 438, 33–44, 2012.

[24] Brannen GE, Bush WH. Ultrasonic Destruction of Kidney Stones. West J Med140 (2), 227–232, 1984.

[25] Breazeale MA, McPherson M. Physical Acoustics, chapter in: Rossing TD(ed.) Springer Handbook of Acoustics, Springer Science+Business Media, LLCNew York, 2007.


Bibliography

[26] Briens L, Daniher D, Tallevi A. Monitoring high-shear granulation usingsound and vibration measurements. Int J Pharm 331, 54–60, 2007.

[27] Borgquist P, Körner A, Piculell L, Larsson A, Axelsson A. A model for thedrug release from a polymer matrix tablets effects of swelling and dissolution.J Control Release 113, 216–225, 2006.

[28] Briongos JV, Aragon JM, Palancar MC. Fluidised bed dynamics diagnosisfrom measurements of low-frequency out-bed passive acoustic emissions,Powder Tech 162 (2), 145–156, 2006.

[29] Celik M. Pharmaceutical powder compaction technology 2nd Ed. ISBN(print): 0360-2583, Informa Healthcare, London, UK, 2011.

[30] Charlton B, Newton JM. Application of gamma-ray attenuation to the deter-mination of density distributions within compacted powders. Powder Tech41, 123–134, 1985.

[31] Chen ZP, Morris J, Martin E. Extracting chemical information from spectraldata with multiplicative light scattering effects by optical path-length estima-tion and correction, Anal Chem 78 (22), 7674–7681, 2006.

[32] Chowhan ZT, Chow YP. Compression properties of granulations made withbinders containing different moisture contents. J Pharm Sci 70 (10), 1134–1139,1981.

[33] Ciurczak EW. Principles of near-infrared spectroscopy, chapter in: Burns DA,Ciurczak EW (Eds.), Handbook of Near-Infrared Analysis, 2nd ed., pp. 7–18.Marcel Dekker Inc., New York/Basel, 2001.

[34] Colombo P, Gazzaniga A, Caramella C, Come U, La Manna A. In vitro pro-grammable zero-order release drug delivery system. Acta Pharm Technol 33,15–20, 1987.

[35] Colombo P, Bettini R, Massimo G, Catellani PL, Santi P, Peppas NA. Drugdiffusion front movement is important in drug release control from swellablematrix tablets. J Pharm Sci 84, 991–997, 1995.

[36] Colombo P, Bettini R, Santi P, De Ascentiis A, Peppas NA. Analysis of theswelling and release mechanisms from drug delivery systems with emphasison drug solubility and water transport. J Control Release 39, 231–237, 1996.

[37] Colombo P, Bettini R, Peppas NA. Observation of swelling process and diffu-sion front position during swelling in hydroxypropylmethylcellulose (HPMC)matrices containing a soluble drug. J Control Release 61, 83–91, 1999.



[38] Colombo P, Bettini R, Santi P, Peppas NA. Swellable matrices for controlleddrug delivery: gel-layer behavior, mechanisms and optimal performance.Pharm Sci Technol To 3, 198–204, 2000.

[39] Dahlberg C, Fureby A, Schuleit M, Dvinskikh SV, Fur’o I. Polymer mobi-lization and drug release during tablet swelling A 1H NMR and NMR mi-croimaging study. J Control Release 122, 199–205, 2007.

[40] Daniher D, Briens L, Tallevi A. End-point detection in high-shear granulationusing sound and vibration signal analysis. Powder Tech 181, 130–136, 2008.

[41] Davies P. Chapter in: Gibson M. (Ed.) Pharmaceutical Preformulation andFormulation, ISBN: 978-4-4200-7317-1, Informa Healthcare USA, Inc., NY,2009.

[42] Eriksson M, Alderborn G. The effect of particle fragmentation on the inter-particulate bond formation process during powder compaction. Pharm Res12, 1031–1039, 1995.

[43] European Council. EU Directive 2003/94/EC, article 11.

[44] Evans DF, Pye G, Bramley R, Clark AG, Dyson TJ, Hardcastle JD. Measure-ment of gastrointestinal pH profiles in normal ambulant human subjects. Gut29, 1035–1041, 1988.

[45] Fell JT, Newton JM. Determination of tablet strength by the diametrical com-pression test. J Pharm Sci 59, 688–691, 1970.

[46] Ferrero C, Munoz-Ruiz A, Jimenez-Castellanos MR. Fronts movement as auseful tool for hydrophilic matrix release mechanism elucidation. Int J Pharm202, 21–28, 2000.

[47] Frake P, Greenhalgh D, Grierson SM, Hempenstall JM, Rudd DR. Processcontrol and end-point determination of a fluid bed granulation by applicationof near infrared spectroscopy, Int J Pharm 151 (1), 75–80, 1997.

[48] Frenning G. Analysis of pharmaceutical powder compaction using multi-plicative hyperelasto-plastic theory, Powder Tech 172, 103–112, 2007.

[49] Fyfe CA, Blazek AI. Investigation of hydrogel formation from hydroxypropy-lmethylcellulose (HPMC) by NMR spectroscopy and NMR imaging tech-niques. Macromolecules 30, 6230–6237, 1997.

[50] Gao P, Meury RH. Swelling of hydroxypropyl methylcellulose matrix tablets1. Characterization of swelling using a novel optical imaging method. J PharmSci 85, 725–731, 1996.


Bibliography

[51] Gatti PL, Ferrari V. Applied Structural and Mechanical Vibrations: TheoryMethods and Measuring Instrumentation. 815 p, ISBN: 0-419-22710-5, E &FN Spon Press, London, UK, 1999.

[52] Gherras N, Serris E, Fevotte G. Monitoring industrial pharmaceutical crys-tallization processes using acoustic emission in pure and impure media. Int JPharm 439, 109–119, 2012.

[53] Goujon L, Baboux JC. Behaviour of acoustic emission sensors using broad-band calibration techniques. Meas Sci Technol 14, 903, 2003.

[54] Griffenhagen GB. Pharm. Technol. 4, 45, 1980.

[55] Hakanen A, Laine E, Jalonen H, Linsaari K, Jokinen J. Acoustic Emissionduring powder Compaction and its Frequency Spectral Analysis. Drug DevInd Pharm 19, 2539–2560, 1993.

[56] Hakanen A, Laine E. Acoustic characterization of a microcrystalline cellulosepowder during and after its compression. Drug Dev Ind Pharm 21 (13), 1573–1582, 1995.

[57] Hakulinen MA, Pajander J, Leskinen J, Ketolainen J, van Veen B, Niinimäki,K, Pirskanen K, Poso A, Lappalainen R. Ultrasound transmission techniqueas a potential tool for physical evaluation of monolithic matrix tablets. AAPSPharmSciTech 9, 267–273, 2008.

[58] Halstensen M, de Bakker P, Esbensen KH. Acoustic chemometric monitor-ing of an industrial granulation production process–a PAT feasibility study.Chemometr Intell Lab 84 (1–2), 88–97, 2006.

[59] Harland RS, Gazzaniga A, Sangalli ME, Colombo P, Peppas NA.Drug/polymer matrix swelling and dissolution. Pharm Res 5, 488–494, 1988.

[60] Hausman DS, Cambron RT, Sakr A. Application of on-line Raman spec-troscopy for characterizing relationships between drug hydration state andtablet physical stability. Int J Pharm 299 (1–2), 19–33, 2005.

[61] Herper M. The Truly Staggering Cost Of Inventing New Drugs. Feb 2012.http://www.forbes.com/sites/matthewherper/2012/02/10/......the-truly-staggering-cost-of-inventing-new-drugs/ (Valid November 92012).

[62] Hiestand EN, Smith DP. Three indexes for characterizing the tabletting per-formance of materials. Adv Ceram 9, 47–57, 1984.

[63] Hjemsted K, Schneider T. 1996. Dustiness from powder materials. J AerosolSci 27, Abstracts of the 1996 European Aerosol Conference, Supplement 1,485–486, 1996.



[64] Iveson SM, Litster JD, Hapgood K, Ennis BJ. Powder Tech 117, 3–39, 2001.

[65] Joe Au YH, Eissa S, Jones BE. Receiver operating characteristic analysis forthe selection of threshold values for detection of capping in powder compres-sion. Ultrasonics 42, 149–153, 2004.

[66] Kamm R, Steinberg MA, Wulff J. Lead-grid study of metal powder com-paction. Trans Am Inst Mining Met Eng, 180, 649–706, 1949.

[67] Karande AD, Wan Sia Heng P, Liew CV. In-line quantification of micronizeddrug and excipients in tablets by near infrared (NIR) spectroscopy: Real timemonitoring of tabletting process, Int J Pharm 396 (1–2), 63–74, 2010.

[68] Karehill PG, Börjesson E, Glazer M. Bonding Surface area and BondingMechanism-Two Important Factors fir the Understanding of Powder Com-parability. Drug Dev Ind Pharm 19 (17-18), 2143–2196, 1993.

[69] Karehill PG, Nyström C. Studies on direct compression of tablets XXI. In-vestigation of bonding mechanisms of some directly compressed materials bystrength characterization in media with different dielectric constants (relativepermittivity). Int J Pharm 61 (3), 251–260, 1990.

[70] Karehill PG, Nyström C. Studies on direct compression of tablets. XXII. In-vestigation of strength increase upon ageing and bonding mechanisms forsome plastically deforming materials. Int J Pharm 64 (1) 27–34, 1990.

[71] Karehill PG, Glazer M, Nyström C. Studies on direct compression of tablets.XXIII. The importance of surface roughness for the compactability of somedirectly compressible materials with different bonding and volume reductionproperties. Int J Pharm 64 (1), 35–43, 1990.

[72] Karppinen T, Lassila I, Haeggström E. Ultrasonic Monitoring of Paper andTablet Coating Stiffness During Layer Formation. 0-7803-9383-X/05, 2005IEEE Ultrasonics Symposium, 1488–1495, 2005.

[73] Kazarian S, van der Weerd J. Simultaneous FTIR spectroscopic imaging andvisible photography to monitor tablet dissolution and drug release. PharmRes 25, 853–860, 2008.

[74] Ketolainen J, Oksanen M, Rantala J, Stor-Pellinen J, Luukkala M, Paronen P.Photoacoustic evaluation of elasticity and integrity of pharmaceutical tablets.Int J Pharm 125, 45–53, 1995.

[75] Kim Y-H. Sound Propagation. An Impedance Based Approach. John Wiley &Sons (Asia) Pte Ltd, 2010.


Bibliography

[76] Kogerman K, Aaltonen J, Strachan CJ, Pöllänen K, Heinämäki J, Yliruusi J,Rantanen J. Establishing quantitative in-line analysis of multiple solid-statetransformations during dehydration. J Pharm Sci 97 (11), 4983–4999, 2008.

[77] Konrad R, Christ A, Zessin G, Cobet U. The use of ultrasound and penetrom-eter to characterize the advancement of swelling and eroding fronts in HPMCmatrices. Int J Pharm 163, 123–131, 1998.

[78] Kowalczuk J, Tritt-Goc J, Pislewski N. The swelling properties of hydrox-ypropyl methyl cellulose loaded with tetracycline hydrochloride: magneticresonance imaging study. Solid State Nucl Magn 25, 35–41, 2004.

[79] Körner A, Larsson A, Andersson Å, Piculell L. Swelling and polymer erosionfor poly(ethylene oxide) tablets of different molecular weights polydispersi-ties. J Pharm Sci 99, 1225–1238, 2009.

[80] Körner A, Larsson A, Piculell L, Wittgren B. Molecular information onthe dissolution of polydisperse polymers: mixtures of long and shortpoly(ethylene oxide). J Phys Chem B 109, 11530–11537, 2005.

[81] Lakio S. Towards real-time understanding of processes in pharmaceuticalpowder technology. Academic dissertation. ISSN:1795-7079, Helsinki, 2010.

[82] Leach MF, Rubin GA, Williams JC. Particle size determination from acousticemissions. Powder Tech 16, 153–157, 1977.

[83] Lee PI, Peppas NA. Prediction of polymer dissolution in swellable controlled-release systems. J Control Release 6, 207–215, 1987.

[84] Lee PI, Kim C. Probing the mechanisms of drug release from hydrogels. JControl Release 16, 229–236, 1991.

[85] Leuneberger H, Betz G. Granulation Process Control – Production of Phar-maceutical Granules, chapter in: Salman AD, Hounslow MJ, Seville JPK(Eds.) Handbook of Powder Technology, vol. 11: Granulation, ISBN: 978-0-444-51871-2, Elsevier BV, Oxford, UK, 2007.

[86] Levina M, Rubinstein MH, Rajabi-Siahboomi AR. Principles and Applicationof Ultrasound in Pharmaceutical Powder Compression. Pharm Res 17 (3),257–265, 2000.

[87] Lindeberg T. Shape measurement of small objects using LCD fringe projec-tion with phase shifting, Int J Comput Vision 30 (2), 77–116, 1998.

[88] Lindsay RB. Acoustics: Historical and Philosophical Development. Dowden,Hutchinson & Ross, Stroudsburg, PA, p. 88, 1973.



[89] Liu J, Stephens JD, Kowalczyk BR, Cetinkaya C. Real-time in-die compactionmonitoring of dry-coated tablets. Int J Pharm 414, 171–178, 2011.

[90] Martens H, Nielsen JP, Engelsen SB. Light scattering and light absorbanceseparated by extended multiplicative signal correction. Application to nearin-frared transmission analysis of powder mixtures, Anal Chem 75 (3), 394–404,2003.

[91] Matero S, Poutiainen S, Leskinen J, Järvinen K, Ketolainen J, ReinikainenS-P, Hakulinen M, Lappalainen R, Poso A. The feasibility of using acousticemissions for monitoring of fluidized bed granulation, Chemometr Int Lab97 (1), 75–81, 2009.

[92] Matero S, Poutiainen S, Leskinen J, Reinikainen S-P, Ketolainen J, Järvinen K,Poso A. Monitoring the wetting phase of fluidized bed granulation processusing multi-way methods: The separation of successful from unsuccessfulbatches. Chemometr Int Lab 96, 88–93, 2009.

[93] Maurer L, Leuenberger H. Applications of near infrared spectroscopyin thefull-scale manufacturing of pharmaceutical solid dosage forms. Pharm Ind71, 672–678, 2009.

[94] Medendorp J, Lodder RA. Acoustic-resonance spectrometry as a process an-alytical technology for rapid and accurate tablet identification. AAPS Pharm-SciTech 7, E1–E9, 2006.

[95] Medendorp JP, Fackler JA, Douglas CC, Lodder RA. Integrated sensing andprocessing acoustic resonance spectrometry (ISP-ARS) for sample classifica-tion. J Pharm Innov 2, 125–134, 2007.

[96] Mellin V, Salonen J, Laine E. The stimulated acoustic relaxation emission ofmaize starch tablets. Int J Pharm 220, 85–90, 2001.

[97] Miettinen J. Condition Monitoring of Grease Lubricated Rolling Bearings byAcoustic Emission Measurements. Academic dissertation, 70 p. Tampere Uni-versity of Technology Publications 307, 2000.

[98] Mikac U, Demsar A, Demsar F, Sersa I. A study of tablet dissolution bymagnetic resonance electric current density imaging. J Magn Reson 185, 103–109, 2007.

[99] Mortier STFC, De Beer T, Gernaey KV, Vercruysse J, Fonteyne M, Remon JP,Vervaet C, Nopens I. Mechanistic modelling of the drying behaviour of singlepharmaceutical granules, Eur J Pharm Biopharm 80 (3), 682–689, 2012.

[100] Munson J, Stanfield CF, Gujral B. A review of process analytical technology(PAT) in the US pharmaceutical industry. Curr Pharm Anal 2, 405–414, 2006.


Bibliography

[101] Naelapää K, Veski P, Pedersen JG, Anov D, Jörgensen P, Kristensen HG, Ber-telsen P. Acoustic monitoring of a fluidized bed coating process. Int J Pharm332, 90–97, 2007.

[102] Naes T, Isaksson T. SEP or RMSEP, which is best? NIR News 2, 16, 1991.

[103] Nieuwmeyer F, van der Voort Maarschalk K, Vromans H. The consequencesof granulate heterogeneity towards breakage and attrition upon fluid-beddrying, Eur J Pharm Biopharm 170 (1), 402–408, 2008.

[104] Nokhodchi A, Raja S, Patel P, Asare-Atto K. The Role of Oral ControlledRelease Matrix Tablets. BioImpacts 2, 175–187, 2012.

[105] Nyström C, Karehill PG. Studies on direct compression of tablets XVI. Theuse of surface area measurements for the evaluation of bonding surface areain compressed powders. Powder Tech 47 (3), 201–209, 1986.

[106] Närvänen T, Seppälä K, Antikainen O, Yliruusi J. A new rapid on-linemethod to determine particle size distribution of granules, AAPS Pharm-SciTech 9 (1), 282–287, 2008.

[107] Ultrasonics Transducers Technical Notes, 2010. http://www.olympus-ims.com/data/File/panametrics/UT-technotes.en.pdf (valid August 12,2010).

[108] Page TG, Waldron MS, Boeckx J. Tablet Production Module and Method forContinuous Production of Tablets, 2012. Patent WO2010128359 (A1) World-wide Database 5.7.44.6; 93p

[109] Papp MK, Pujara CP, Pinal R. Monitoring of High-shear Granulation usingAcoustic Emission: Predicting Granule Properties. J Pharm Innov 3, 113–122,2008.

[110] Plumb K. Continuous Processing in the Pharmaceutical Industry: Changingthe Mind Set, Chem Eng Res Des 83, 6, 730–738, 2005.

[111] Postema M, Smith AJ. Tablet Processing Unit, 2012. WIPO Patent Applica-tion WO/2010/055337.

[112] Portoghese F, Berruti F, Briens C. Use of triboelectric probes for online mon-itoring of liquid concentration in wet gas-solid fluidized beds. Chem Eng Sci60 (22), 6043–6048, 2005.

[113] Portoghese F, Berruti F, Briens C. Continuous on-line measurement of solidmoisture content during fluidized bed drying using triboelectric probes. Pow-der Tech 181 (2), 169–177, 2008.



[114] Pouet B, Rasolofosaon NJP. Ultrasonic Intrinsic Attenuation MeasurementUsing Laser Techniques. IEEE 1989 Ultrasonics Symposium, 545–549, 1989.

[115] Poutiainen S, Matero S, Hämäläinen T, Leskinen J, Ketolainen J, JärvinenK. Predicting granule size distribution of a fluidized bed spray granulationprocess by regime based PLS modeling of acoustic emission data. PowderTech 228, 149–157, 2012.

[116] Quan C, He XY, Wang CF, Tay CJ, Shang HM. Feature detection with auto-matic scale selection. Opt Commun 198 (1), 21–29, 2001.

[117] Ragozzino M. Analysis of the error in measurement of ultrasound speedin tissue due to waveform deformation by frequency-dependent attenuation.Ultrasonics 19, 135–138, 1981.

[118] Rajabi-Siahboomi AR, Bowtell RW, Mansfield P, Henderson A, Davies MC,Melia CD. Structure and behavior in hydrophilic matrix sustained releasedosage forms Part 2. NMR-imaging studies of dimensional changes in thegel layer and core of HPMC tablets undergoing hydration. J Control Release31, 121–128, 1994.

[119] Rantanen J, Lehtola S, Ramet P, Mannermaa JP, Yliruusi J. Online monitoringof moisture content in an instrumented ïnCuidized bed granulator with amultichannel NIR moisture sensor. Powder Tech 99 (2), 163–170, 1998.

[120] Rantanen J, Räsänen E, Tenhunen J, Känsäkoski M, Mannermaa JP, YliruusiJ. In-line moisture measurement during granulation with a four-wavelengthnear infrared sensor: an evaluation of particle size and binder effects. Eur JPharm Biopharm 50 (2), 271–276, 2000.

[121] Rantanen J, Räsänen E, Antikainen O, Mannermaa JP, Yliruusi J. In-linemoisture measurement during granulation with a four-wavelength near-infrared sensor: an evaluation of process-related variables and a developmentof nonlinear calibration model. Chemometr Int Lab 56 (1), 51–58, 2001.

[122] Reich G. Near-infrared spectroscopy and imaging: Basic principles andpharmaceutical applications. Adv Drug Deliver Rev 57, 1109–1143, 2005.

[123] Reid GL, Ward II HW, Palm AS, Muteki K. Process Analytical Technol-ogy (PAT) in Pharmaceutical Development. Am Pharm Rev 15 (4), 2012.http://www.americanpharmaceuticalreview.com/Featured-Articles/115453-Process-Analytical-Technology-PAT-in-Pharmaceutical-Development/ (validDecember 21 2012).

[124] Rodriguez L, Cini M, Cavallari C, Passerini N, Saettone MF, Fini A, Ca-puto O. Evaluation of theophylline tablets compacted by means of a novelultrasound-assisted apparatus. Int J Pharm 170, 201–208, 1998.


Bibliography

[125] Rossing TD. A Brief History of Acoustics, chapter in: Rossing TD. (Ed.)Springer Handbook of Acoustics. Springer Science+Business Media, LLCNew York, 2007.

[126] Rue PJ, Barkworth PMR, Ridgway-Watt P, Rough P, Sharland DC, Seager H,Fisher H. Analysis of tablet fracture during tabletting by acoustic emissiontechniques. Int J Pharm Tech 1 (1), 2–5, 1979.

[127] Rue PJ, Barkworth PM. The mechanism of time-dependent strength in-creases of sodium chloride tablets. Int J Pharm Tech 1 (2), 2–3, 1980.

[128] Salman AD, Hounslow MJ, Seville JPK (Eds.) Handbook of Powder Tech-nology, vol. 11: Granulation, ISBN: 978-0-444-51871-2, Elsevier BV, Oxford,UK, 2007.

[129] Salonen J, Salmi K, Hakanen A, Laine E, Linsaari K. Monitoring of acousticactivity of a pharmaceutical powder during roller compaction. Int J Pharm153, 257–261, 1997.

[130] Scott P. Process Analytical Technlogy: Applications to the pharmaceuticalindustry. Dissolution Technologies 9, August 2002.

[131] Serris E, Perier-Camby L, Thomas G, Desfontaines M, Fantozzi G. Acousticemission of pharmaceutical powders during compaction. Powder Tech 128,296–299, 2002.

[132] Shull PJ and Tittmann BR. Ultrasound, chapter in: Shull PJ (ed.) Nonde-structive Evaluation: Theory, Techniques, and Applications, pages 63–192.Marcel Dekker Inc., New York, USA, 2002.

[133] Siepmann J, Peppas NA. Hydrophilic matrices for controlled drug deliver:an improved mathematical model to predict the resulting drug release kinet-ics (the sequential layer model). Pharm Res 17, 1290–1298, 2000.

[134] Siepmann J, Streubel A, Peppas NA. Understanding and predicting drugdelivery from hydrophilic matrix tablets using the sequential layer model.Pharm Res 19, 306–314, 2002.

[135] Siesler HW, Ozaki O, Kawata S, Heise HM. Near-Infrared Spectroscopy -Principles Instruments Applications, 1st ed., Wiley-VCH, Weinheim, 2002.

[136] Simonaho SP, Takala TA, Kuosmanen M, Ketolainen J. Ultrasound transmis-sion measurements for tensile strength evaluation of tablets. Int J Pharm 409,104–110, 2011.

[137] Smith CJ, James D. Stephens JD, Hancock BC, Vahdat AS, Cetinkaya C.Acoustic assessment of mean grain size in pharmaceutical compacts. Int JPharm 419, 137–146, 2011.



[138] Smith SW. The Scientist and Engineer’s Guide to Digital Signal Processing.http://www.dspguid.com, California Technical Publishing, San Diego, CA,2011. (Valid December 20 2012.)

[139] Snow RH, Allen T, Ennis BJ, Litster BJ. Size reduction and size enlargement,chapter in: Perry RH, Green DW. (Eds.), Perry’s Chemical Engineers’ Hand-book, McGraw-Hill, USA, 1997.

[140] Stephens JD, Kowalczyk BR, Hancock BC, Kaul G, Cetinkaya C. Ultrasonicreal-time in-die monitoring of the tablet compaction process proof of conceptstudy. Int J Pharm 442 (1–2) 20–26, 2012.

[141] Sun CC. Quantifying errors in tableting data analysis using the Ryshkewitchequation due to inaccurate true density. J Pharm Sci 94, 2061–2068, 2005.

[142] Szilard J. (Ed.) Ultrasonic Testing: Non-conventional testing techniques,ISBN: 0-471-27938-2, John Wiley & Sons, Ltd., NY, USA, 1982.

[143] Tajarobi F, Abrahmsén-Alami S, Carlsson AS, Larsson A. Simultaneousprobing of swelling, erosion and dissolution by NMR-microimaging effectof solubility of additives on HPMC matrix tablets. Eur J Pharm Sci 37, 89–97,2009.

[144] Theobald P, Zeqiri B and Avison J. Couplants and Their Influence on AESensor Sensitivity. J Acoustic Emission, 26, 91–97, 2008.

[145] Thompson RB, Thompson DO. Ultrasonics in Nondestructive Evaluation.Proceedings of the IEEE 73 (12), 1985.

[146] Tok AT, Goh X, Ng WK, Tan RBH. Monitoring granulation rate processesusing three PAT tools in a pilot-scale fluidized bed, AAPS PharmSciTech 9(4),1083–1091, 2008.

[147] Train D. Agglomeration of solids by compaction. Trans Ins Chem Engrs, 35,258–262, 1957.

[148] Troup GM, Georgakis C. Process systems engineering tools in the pharma-ceutical industry. Comput Chem Eng 51, 157–171, 2013.

[149] Tsujimoto H, Yokoyama T, Huang CC, Sekiguchi I. Monitoring particle flu-idization in a fluidized bed granulator with an acoustic emission sensor. Pow-der Tech 113 (1–2), 88–96, 2000.

[150] Turton R, Tardos GI, Ennis BJ. Fluidized Bed Coating and Granulation, Flu-idization, Solids Handling and Processing, chapter in: Yang WC (Ed.) NoyesPublications, Westwood, NJ, 331–434, 1999.


Bibliography

[151] Uhlemann H, Mörl L. Wirbelschichtsprühgranulation, ISBN: 3-540-66985-X,Springer-Verlag, Berlin, 2000.

[152] Unnbórsson R. Identifying and Monitoring Evolving AE Sources, chapterin Acoustic Emission by Sikorsky W. (Ed.), ISBN 978-953-51-0056-0, InTech,2012.

[153] U.S. Food and Drug Administration. "PAT" A Framework for Innova-tive Pharmaceutical Development, Manufacturing, and Quality Assurance",Sept 2004. http://www.fda.gov/cder/guidance/6419fnl.htm (Valid October10 2012)

[154] U.S. Food and Drug Administration. "Guidance for In-dustry Q8(R2) Pharmaceutical Development", May 2006.http://www.fda.gov/downloads/Drugs/Guidances/ucm073507.pdf(Valid Dec 11 2012)

[155] US Pharmacopeial Convention, "United States Pharmacopeia 32 and 27thNational Formulary, <1005> Acoustic Emission", 2011.

[156] Vallen Systeme GmbH. AMSY-6 System Description. www.vallen.de. (Validin November 31 2011)

[157] Varghese I, Cetinkaya C. Noncontact photo-acoustic defect detection in drugtablets. J Pharm Sci 96, 2125–2133, 2007.

[158] Varma MVS, Kaushal AM, Garg A, Garg S. Factors affecting mechanismand kinetics of drug release from matrix-based oral controlled drug deliverysystems. Am J Drug Deliv. 2, 43–57, 2004.

[159] Waring MJ, Rubinstein MH, Howard JR. 1987. Acoustic emission of phar-maceutical materials during compression. Int J Pharm 36, 29–36, 1987.

[160] Watano S, Miyanami K. Image processing for online monitoring of granulesize distribution and shape in fluidized bed granulation, Powder Tech 83 (1),55–60, 1995.

[161] Watano S, Fukushima T, Miyanami K. Chem Pharm Bull 44, 572–576, 1996.

[162] Watano S. Direct control of wet granulation processes by image processingsystem. Powder Tech 117 (1–2), 163–172, 2001.

[163] Wells PNT. Physical Principles of Ultrasonic Diagnosis. Academic Press,London, 1969.

[164] Wentzell PD. and Wade AP. 1989. Chemical Acoustic Emission Analysis inthe Frequency Domain. Anal Chem 61, 2638–2642, 1989.



[165] Whitaker M, Baker GR, Westrup J, Goulding PA, Rudd DR, Belchamber RM,Collins MP. Application of acoustic emission to the monitoring and end pointdetermination of a high shear granulation process. Int J Pharm 205, 79–91,2000.

[166] Witzleb R, Kanikanti V-R, Hamann H-J, Kleinebudde P. Solid lipid extrusionwith small die diameters - Electrostatic charging, taste masking and continu-ous production, Eur J Pharm Biopharm, 77 (1), 170–177, 2011.

[167] Wong DY, Waring MJ, Wright P, Aulton ME. Acoustic emission during thedeformation of α-lactose monohydrate and anhydrous α-lactose monocrys-tals. J Pharm Pharmacol 43 (9), 659-661, 1991.

[168] Wood AB. Textbook of Sound. G. Bell and Sons Ltd., London, UK, 1932.

[169] Winkler F. Verfahren zum Herstellen von Wasserglas, German patent DE437970, 1922.

[170] Yochev B, Kutzarov S, Ganchev D, Staykov K. Investigation of ultrasoundproperties of hydrophilic polymers for dry-coupled inspection. ECNDT 2006Proceedings, Berlin, Germany. 2006.

[171] Yu LX. Pharmaceutical quality by design: product and process develop-ment, understanding, and control. Pharm Res 25, 781–791, 2008.

[172] Ziehl P, Pollock A. Acoustic Emission Acoustic Emission for Civil Structures,chapter in Acoustic Emission by Sikorsky W. (Ed.), ISBN 978-953-51-0056-0,InTech, 2012.

[173] Zuleger S, Fassihi R, Lippold BC. Polymer particle erosion controlling drugrelease II. Swelling investigations to clarify the release mechanism. Int JPharm 247, 23–37, 2002.


Publications of the University of Eastern Finland Dissertations in Forestry and Natural Sciences

Publications of the University of Eastern Finland

Dissertations in Forestry and Natural Sciences

isbn 978-952-61-1169-8

Jari T.T. Leskinen

Acoustic Techniques for Pharmaceutical Process MonitoringMeasurements in Tablet Manufacturing and Quality Control

The tablet is probably the most

common solid dosage form for

orally administered drugs. In this

thesis, acoustic techniques were

tested for pharmaceutical process

monitoring and tablet quality control

purposes. An acoustic emission

method was found to be suitable for

real-time particle size estimation in

a granulation process. Ultrasound

(US) methods were found to be

good for real-time monitoring of the

tabletting, as well as detecting the

integrity of the tablet. Additionally, a

developed US technique was capable

for determining the formed gel layer

thickness on immersed tablets.

dissertatio

ns | 112 | Ja

ri T

.T. L

eskin

en | A

coustic T

echniques for P

harm

aceutical Process M

onitoring – M

easurements in T

ablet...

Jari T.T. LeskinenAcoustic Techniques for Pharmaceutical Process


and Quality Control

Date post:	14-Sep-2020
Category:	Documents
Upload:	others
View:	1 times
Download:	0 times

Jari T.T. Leskinen Acoustic Techniques for · 2014. 1. 8. · JARI T. T. LESKINEN Acoustic...

Documents