process analytical technology

Edited by Katherine A. Bakeev
Process
Analytical
Technology
Process Analytical Technology
This page intentionally left blank
Process Analytical Technology
Spectroscopic Tools and Implementation
Strategies for the Chemical and
Pharmaceutical Industries
Edited by
Katherine A. Bakeev
Ó 2005 by Blackwell Publishing Ltd
Editorial Offices:
Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK
Tel: +44 (0)1865 776868
Blackwell Publishing Professional, 2121 State Avenue, Ames, Iowa 50014–8300, USA
Tel: +1 515 292 0140
Blackwell Publishing Asia Pty Ltd, 550 Swanston Street, Carlton, Victoria 3053, Australia
Tel: +61 (0)3 8359 1011
The right of the Author to be identified as the Author of this Work has been asserted in
accordance with the Copyright, Designs and Patents Act 1988.
All rights reserved. No part of this publication may be rep roduced, stored in a retrieval
system, or transmitted, in any form or by any means, electronic, mechanical, photocopying,
recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988,
without the prior permission of the publisher.
First published 2005
Library of Congress Cataloging-in-Publication Data
Process analytical technology /edited by Katherine A. Bakeev.
p. cm.
Includes bibliographical references and index.

ISBN-10: 1–4051–2103–3 (acid-free paper)
ISBN-13: 978–1–4051–2103–3 (acid-free paper)
1. Chemical process control—Industrial applications. 2. Chemistry, Technical. 3. Chemistry,
Analytic—Technological innovations. 4. Chemistry, Analytic—Technique. 5. Spectrum
analysis. 6. Pharmaceutical chemistry. I. Bakeev, Katherine A.
TP155.75.P737 2005
660
0
.2—dc22
2004065962
ISBN-10: 1–4051–2103–3
ISBN-13: 978–1–4051–2103–3
A catalogue record for this title is available from the British Library
Set in 10/12pt Minion
by Integra Software Services Pvt. Ltd, Pondicherry, India
Printed and bound in India
by Gopson Papers, Noida
The publisher’s policy is to use permanent paper from mills that operate a sustainable forestry
policy, and which has been manufactured from pulp processed using acid-free and elementary
chlorine-free practices. Furthermore, the publisher ensures that the text paper and cover
board used have met acceptable environmental accreditation standards.
For further information on Blackwell Publishing, visit our website:
www.blackwellpublishing.com
Contents
Contributors xiii
Preface xv
List of Abbreviations xvii
1 Process Analytical Chemistry: Introduction and Historical
Perspective Ernest Baughman 1
1.1 Historical perspective 2

1.2 Early instrument develo pment 4
1.3 Sampling systems 7
1.4 Examples 8
References 11
2 Implementation of Process Analytical Technologies
Robert Guenard and Gert Thurau 13
2.1 Introduction to implementation of process analytical
technologies (PATs) in the industrial setting 13
2.1.1 Definition of process analytics 14
2.1.2 Differences between process analyzers and
laboratory analysis 14
2.1.3 General industrial drivers for process analytics 15
2.1.4 Types of applications (R&D vs. Manufacturing) 16
2.1.5 Organizationa l considerations 17
2.2 Generalized process ana lytics work process 20
2.2.1 Project identification and definition 22
2.2.2 Analytical application development 24
2.2.3 Design, specify and procure 24
2.2.4 Implementatio n in production 26
2.2.5 Routine operation 27
2.2.6 Continuous improvement 28
2.3 Differences between implementation in chemical and
pharmaceutical indus tries 29
2.3.1 Introduction 29
2.3.2 Business model 29
2.3.3 Technical differences 30
2.3.4 Regulatory differences 32
2.4 Conclusions 37
References 37
3 Near-Infrared Spectr oscopy for Process Analytical Chemistry: Theory,

Technology and Impleme ntation Michael B. Simpson 39
3.1 Introduction 39
3.2 Theory of near-infrar ed spectroscopy 44
3.2.1 Molecular vibrations 44
3.2.2 Anharmonicity of the potential well 45
3.2.3 Combination and overtone absorptions in the near-infrared 47
3.2.4 Examples of useful near-infrared absorption bands 48
3.3 Analyser technologies in the near-infrared 51
3.3.1 The scanning grating monochromator 51
3.3.2 Light sources and detectors for near-infrared analysers 55
3.3.3 The polychromator photodiode-array a nalyser 62
3.3.4 The acousto-optic tunable (AOTF) analyser 63
3.3.5 Fourier transform near-infrared analys ers 69
3.4 The sampling interface 77
3.4.1 Introduction 77
3.4.2 Further discussion of sampling issues 84
3.4.3 The use of fibre-optics 86
3.5 Conclusion 88
Bibliography 89
4 Infrared Spectroscopy for Process Analytical Applications
John P. Coates 91
Abstract 91
4.1 Introduction 92
4.2 Basic IR spectroscopy 95
4.3 Instrumentation design and technology 97
4.4 Process IR instrumentation 100
4.4.1 Commercially available IR instruments 101
4.4.2 Important IR component technologies 108
4.4.3 New technologies for IR comp onents and instruments 112
4.4.4 Requirements for process infrared analyzers 114

4.4.5 Sample handling for IR process analyzers 121
4.4.6 Issues for consideration in the implementation
of process IR 124
vi Contents
4.5 Applications of process IR analyzers 126
4.6 Process IR analyzers: A review 127
4.7 Trends and directions 129
References 130
5 Process Raman Spectroscopy Nancy L. Jestel 133
5.1 How Raman spectroscopy works 133
5.2 When Raman spectroscopy works well and when
it does not 136
5.2.1 Advantages 136
5.2.2 Disadvantages and risks 138
5.3 What are the special design issues for process R aman
instruments? 140
5.3.1 Safety 141
5.3.2 Laser wavelength selection 142
5.3.3 Laser power and stability 142
5.3.4 Sample interface/probes 143
5.3.5 Spectrometer 144
5.3.6 Communications 146
5.3.7 Maintenance 147
5.4 Where Raman spectroscopy is being used 147
5.4.1 Reaction monitoring 147
5.4.2 In-process aid or quality-monitoring tool 155
5.4.3 Product properties 161
5.4.4 Mobile or field uses 161
5.5 What is the current state of Raman spectroscopy? 161
5.5.1 Publication reluctance 162

5.5.2 Technique maturity and long-term performance 163
5.5.3 Lack of widespread knowledge and experience 163
References 163
6 UV-Vis for On-Line Analysis Lewis C. Baylor and Patrick E. O’Rourke 170
6.1 Introduction 170
6.2 Theory 171
6.2.1 Chemical concentration 171
6.2.2 Color 172
6.2.3 Film thickness 173
6.3 Instrumentation 173
6.4 Sample interface 174
6.4.1 Cuvette/vial 174
6.4.2 Flow cells 175
6.4.3 Insertion probe 176
6.4.4 Reflectance probe 177
Contents vii
6.5 A complete process analyz er 177
6.6 Applications 178
6.6.1 Gas analysis – toluene 178
6.6.2 Liquid analysis – nickel 180
6.6.3 Solid analysis – extruded plastic color 181
6.6.4 Film thickness – polymer 182
6.6.5 Dissolution testing 183
6.6.6 Liquid analysis – vessel cleaning 185
References 186
7 Near-Infrared Chemical Imaging as a Process Analytical Tool
E. Neil Lewis, Joseph W. Schoppelrei, Eunah Lee, and Linda H. Kidder 187
7.1 The process analytical technology (PAT) initiative 187
7.2 The role of near-infrared chemical imaging (NIR-CI)
in the pharmaceutical industry 188

7.2.1 Characterization of solid dosage forms 188
7.2.2 ‘A picture is worth a thousand words’ 189
7.3 The development of imaging spectroscopy 190
7.3.1 Spatially resolved spectroscopy – mapping 190
7.3.2 The infrared focal-plane array (FPA) 190
7.3.3 Wavelength selection 191
7.3.4 The benefits of NIR spectroscopy 191
7.3.5 NIR imaging instrumentation 192
7.4 Chemical imaging principles 194
7.4.1 The hypercube 195
7.4.2 Data analysis 196
7.4.3 Spectral correction 197
7.4.4 Spectral pre-processing 197
7.4.5 Classification 198
7.4.6 Image processing 200
7.5 PAT applications 201
7.5.1 ‘Self-calibrating’ high-throughput content uniformity
measurements 201
7.5.2 High-throughp ut applications: Counterfeit
screening/quality assurance 204
7.5.3 Defeating sample dilution: Finding the needle
in the haystack 206
7.5.4 Advanced dosage delivery systems 209
7.6 Processing case study one: Estimating ‘abundance’ of sample
components 210
7.6.1 Experimental 211
7.6.2 Spectral correction and pre-processing 211
7.6.3 Analysis 211
7.6.4 Conclusions 217
viii Contents

7.7 Processing case study two: Determining blend homoge neity
through statistical analysis 217
7.7.1 Experimental 218
7.7.2 Observing visual contrast in the image 219
7.7.3 Statistical analysis of the imag e 219
7.7.4 Blend uniformity measurement 221
7.7.5 Conclusions 222
7.8 Final thoughts 223
Acknowledgements 223
References 223
8 Chemometrics in Process Analytical Chemistry Charles E. Miller 226
8.1 Introduction 226
8.1.1 What is chemometric s? 226
8.1.2 What does it do for analytical chemistry? 227
8.1.3 What about process analytical chemistry? 228
8.1.4 Some history 228
8.1.5 Some philosophy 229
8.2 The building blocks of chemometrics 230
8.2.1 Notation 230
8.2.2 A bit of statistics 231
8.2.3 Linear regression 233
8.2.4 Multiple linear regression (MLR) 236
8.2.5 Data pre-treatment 237
8.2.6 Data compression 243
8.2.7 Spatial sample representation 249
8.2.8 Experimental desig n 250
8.3 Quantitative model building 254
8.3.1 ‘Inverse’ multiple linear regression 254
8.3.2 Classical least squares (CLS) 257
8.3.3 Principal component regression (PCR) 259

8.3.4 Projection to latent struc tures (PLS) regression 262
8.3.5 Artificial neural networks (ANN) 264
8.3.6 Other quantitative model-building tools 267
8.3.7 Overfitting and model validation 267
8.3.8 Improving quantitative model performance 274
8.4 Outliers 277
8.4.1 What are they, and why should we care? 277
8.4.2 Outliers in calibration 277
8.4.3 Outliers in prediction 283
8.5 Qualitative model building 285
8.5.1 Space definition 286
8.5.2 Measures of distance in the space 287
Contents ix
8.5.3 Classification rule development 289
8.5.4 Commonly encountered classification method s 289
8.6 Exploratory analysis 297
8.6.1 Model parameters from PCA, PCR, and PLS 297
8.6.2 Self-modeling curve resolution (SMCR) 303
8.6.3 ‘Unsupervised’ learning 307
8.7 The ‘calibration sampling paradox’ of process analytical
chemistry 309
8.8 Sampl e and variable selection in chemo metrics 311
8.8.1 Sample selection 311
8.8.2 Variable selection 313
8.9 Calibration transfer 316
8.9.1 Slope/intercept adjustme nt 317
8.9.2 Piecewise direct standardization (PDS) 318
8.9.3 Shenk-Westerhaus method 319
8.9.4 Other standardization methods 320
8.10 Non-technical issues 320

8.10.1 Problem definition 320
8.10.2 Where does all the time go? 321
8.10.3 The importance of documentation 322
8.10.4 People issues 323
8.11 The final word 324
References 324
9 On-Line Applications in the Pharmaceutical Industry
Steven J. Doherty and Charles N. Kettler 329
9.1 Background 329
9.2 Reaction monitoring and control 332
9.3 Crystallization monitoring and control 338
9.4 Drying 341
9.5 Milling 344
9.6 Cleaning validation 345
9.7 Solid dosage form manufacture 345
9.8 Granulation and blending of API and excipients 345
9.8.1 Granulation 347
9.8.2 Blending 348
9.9 Detection of drug polymorphism 349
9.10 Spectroscopic techniques for tablets 352
References 356
10 Use of Near-Infrared Spectroscopy for Off-Line Measurements in the
Pharmaceutical Industry Marcelo Blanco and Manel Alcala
´
362
10.1 Introduction 362
10.1.1 Operational procedures 363
10.1.2 Instrument qualification 364
x Contents
10.2 Qualitative analysis 365

10.2.1 Foundation of identification (authentication) 366
10.2.2 Construction of NIR libraries 367
10.2.3 Identification of raw materials and pharmaceutical
preparations 370
10.2.4 Determination of homogeneity 372
10.2.5 Characterization of polymorphs 373
10.3 Quantitative analysis 374
10.3.1 Selection of samples 375
10.3.2 Determination of reference values 376
10.3.3 Acquisition of spectra 377
10.3.4 Construction of the calibration model 377
10.3.5 Validation of the model 378
10.3.6 Prediction of new samples 378
10.4 Method validation 379
10.5 Matching models 379
10.6 Pharmaceutical applications 381
10.6.1 Determination of physical parameters 381
10.6.2 Determination of moisture 382
10.6.3 Determination of active pharmaceutical ingredients 384
References 384
11 Applications of Near-Infrared Spectroscopy (NIR) in the Chemical
Industry Ann M. Brearley 392
11.1 Introduction 392
11.2 Successful process analyzer implementation 393
11.2.1 A process for successful process analyzer implementation 393
11.2.2 How NIR process analyzers contribute to
business value 396
11.2.3 Issues to consider in setting technical requirements
for a process analyzer 398
11.2.4 Capabilities and limitations of NIR 399

11.2.5 General challenges in process analyzer implementation 401
11.2.6 Approaches to calibrati ng an NIR analyzer on-line 403
11.2.7 Special challenges in NIR moni toring of polymer melts 406
11.3 Example applications 407
11.3.1 Monitoring monomer conversion during emulsion
polymerization 408
11.3.2 Monitoring a diethylbenzene isomer separation process 410
11.3.3 Monitoring the comp osition of copolymers and
polymer blends in an extruder 411
11.3.4 Rapid identification of carpet face fiber 415
11.3.5 Monitoring the comp osition of spinning solution 417
11.3.6 Monitoring end groups and viscosity in polyester melts 419
References 423
Contents xi
12 Future Trends in Process Analytical Chemistry Katherine A. Bakeev 424
12.1 Introduction 424
12.2 Sensor development 426
12.2.1 Micro-analytical systems 426
12.2.2 Micro-instrumentation 427
12.2.3 Biosensors 427
12.3 New types of PAC hardware tools 428
12.3.1 Thermal effusivity 428
12.3.2 Acoustic spectroscopy 429
12.3.3 Light-induced fluo rescence (LIF) 430
12.3.4 Frequency-domain photon migration 431
12.3.5 Focused beam reflectance measurement (FBRM) 431
12.3.6 Multidimensional analytical technology 432
12.4 Data communicati on 432
12.4.1 Data management 432
12.4.2 Regulatory issues in data management 434

12.4.3 Digital communication 435
12.4.4 Wireless communication 436
12.5 Data handling 436
12.5.1 Chemometric tools 437
12.5.2 Soft sensors and predictive modeling 439
12.6 New application areas of PAC 440
References 442
Index 445
The Colour Plate Section appears after page 204
xii Contents
Contributors
Mr Manel Alcala
´
Department of Chemistry, Faculty of Sciences, Autonomous
University of Barcelona, E-08193 Bellaterra, Spain
Dr Katherine A. Bakeev Glaxo SmithKline, 709 Swedeland Rd, King of Prussia, PA
19406
Dr Ernest Baughman PO Box 1171, Rancho Cucamonga, CA 91729-1171
Dr Lewis C. Baylor Equitech International Corporation, 903 Main St. South,
New Ellenton, SC 29809
Prof Marcelo Blanco Department of Chemistry, Faculty of Sciences, Autonomous
University of Barcelona, E-08193 Bellaterra, Spain
Dr Ann M. Brearley 2435 Shadyview Lane N, Plymouth MN 55447
Dr John P. Coates Coates Consulting, 12 N. Branch Rd, Newtown, CT 06470-
1858
Dr Steven J. Doherty Eli Lilly and Company, Lilly Corporate Center, Indiana polis,
IN 46285
Dr Robert Guenard Merck and Co, Inc, PO Box 4, West Point, PA 19486
Dr Nancy L. Jestel General Electric, Advanced Materials, 1 Noryl Ave, Selkirk,
NY 12158

Dr Charles N. Kettler Eli Lilly and Company, Lilly Corporate Center, Indianapolis,
IN 46285
Dr Linda H. Kidder Spectral Dimensions, 3416 Olandwood Ct, #210, Olney, MD
20832
Dr Eunah Lee Spectral Dimensions, 3416 Olandwood Ct, #210, Olney, MD
20832
Dr E. Neil Lewis Spectral Dimensions, 3416 Olandwood Ct, #210, Olney, MD
20832
Dr Charles E. Miller Dupont Engineering Technologies, 140 Cypress Station
Drive, Ste 135, Houston, TX 77090
Dr Patrick E. O’Rourke Equitech International Corporation, 903 Main St. South,
New Ellenton, SC 29809
Dr Joseph W. Schoppelrei Spectral Dimens ions, 3416 Olandwood Ct, #210, Olney,
MD 20832
Dr Michael B. Simpson ABB Analytical and Advanced Solutions, 585 Charest
Boulevard East, Ste 300, Quebec GIK 9H4, Canada
Dr Gert Thurau Merck and Co, Inc, PO Box 4, West Point, PA 19486
xiv Contributors
Preface
A subject as broad as Process Analytical Technology (PAT) is difficult to capture in one
volume. It can be covered from so many different angles, covering engineering, analytical
chemistry, chemometrics, and plant operations, that one needs to set a perspective and
starting point. This book is presented from the perspective of the spectroscopist w ho is
interested in implementing PAT tools for any number of processes. Not all spectroscopic
tools are covered. Included in this book are UV-vis spectroscopy, mid-infrared and near-
infrared, Raman, and near-infrared chemical imaging. Hopefully the treatment of these
subjects and discussion of applications will provide information useful to newcomers to
the field and also expand the knowledge of existing practitioners. The chapter on
implementation is written in a general way making it applicable to those who have
worked in the field, and to those who are new to it. It discusses some of the issues that

need to be addressed when undertaking any type of process analytical project. In the
applications of NIR chapters, many diverse applications are presented. Implementatio n in
terms of business value and team work is discussed, along with applications in the
chemical industry in Chapter 11. Chemometrics is a critical tool in PAT and is discussed
thoroughly in Chapter 8, with salient points relev ant to process analytical data analysis.
Since the field of Process Analytical Technology is currently a very active field in
various industries, the contents of such a book will rapidly become historical, requiring
revision to reflect the current state of the work. It is hoped that the discussions presented
here will serve as a basis for those seeking to understand the area better, and provide
information on many practical aspects of work done thus far.
I would like to thank all those who contributed to this book. Writing a book chapter
is an activity that is taken on in addition to one’s work, and I appreciate the commitment
of all of the authors who took on this task. I would also like to thank my friends who
helped me by reading through items for me including John Hellgeth, Bogdan Kurtyka,
and Glyn Short.
This page intentionally left blank
List of Abbreviations
2D second derivative
ALS alternating least squares
ANN artificial neural network
ANOVA analysis of varia nce
AOTF acousto-optic tunable filter
APC all possible combinations
APCI atmospheric pressure chemical ionization
API active pharmaceutical ingredient
ASTM American Society for Testing of Materials
ATR attenuated total reflectance
BEST Bootstrap Error-adjusted Single-sample Technique
CCD charge-coupled device
CDER Center for Drug Evaluation and Research

CENELEC Comite
´
Europe
´
en de Normalisation Electrotechnique
CLS classical least squares
CMC chemical manufacturing and controls
CPAC Center for Process Analytical Chemistry
CPACT Center for Process Analytics and Control
CVF circular variable filter
DA discriminant analysis
DCS distributed control system
DOE design of experiments
DS direct standardization
DSC differential scanning calorimetry
DTGS deuterated triglycine sulfate
EMEA European Medicines Agency
EPA Environmental Protection Agency
EP-IR encoded photometric IR
ESI electrospray ionization
FDA Food and Drug Administration
FDPM frequency domain photon migration
FIR finite impulse response
FM factory mutual
FOV field of view
FPA focal-plane array
FPE Fabry-Perot e
´
talon
FTIR Fourier transform infrared

FTNIR Fourier transform near-infrared
GA gen etic algorithm
GC gas chromatography
GLP good laboratory practice
GMP good manufacturing practice
HCA hierarchical cluster analysis
HDPE high-density polyethylene
InGaAs Indium-Gallium-Arsenide
IQ installation qualification
ISE ion-selective electrode
ITTFA iterative target transformation factor analysis
IVS interactive variable selection
KNN K-nearest neighbor
LCTF liquid crystal tunable filter
LDA linear discriminant analysis
LDPE low-density polyethylene
LED light emitting diode
LEL lower explosive limit
LIF light-induced fluorescence
LVF linear variable filter
LVQ learning vector quantization
MCT mercury cadmium telluride
MEMS micro-electrical mechanical systems
MIR mid-infrared
MLR multiple linear regression
MSC multiplicative scatter correction or multiplicative signal correction
MST minimal spanning tree
MSZW metastable zone width
MWS multivariate wavelength standardization
NDIR non-dispersive infrared

NIR near-infrared
NIR-CI near-infrared chemical imaging
NIST National Institute of Standards and Technology
NMR nuclear magnetic resonance
OQ operational qualification
PAC process analytical chemistry
PAT process analytical technology
PbS lead sulfide
PbSe lead selenide
PC principal component or personal computer
PCR principal component regression
PDA photodiode array
PDS piecewise direct standardization
xviii List of Abbreviations
PFM potential function method
PLS projection to latent structures or partial least squares
PLSR partial least squares regression
PQ performance qualification or personnel qualification
QA quality assurance
QC quality control
QFD quality function deployment
RMSEE root mean square error of estimate
RMSEP root mean square error of prediction
ROI return on investment
SEC standard error of calibration
SEP standard error of prediction
SIMCA soft independent modeling of class analogies
SMCR self-modeling curve resolution
SMLR stepwise multiple linear regression
SMV spectral match value

SMZ sulfamethoxazole
SNR signal to noise ratio (also abbreviate d S/N)
SOM self-organizing maps
SPC statistical process control
SNV standard normal variate
SVM support vector machines
SWS simple wavelength standardization
TE thermo-electric
UL Underwriters Laboratory
USDA United States Department of Agriculture
USP United States Pharmacopeia
UV-vis ultraviolet-visible
List of Abbreviations xix
This page intentionally left blank
Chapter 1
Process Analytical Chemistry: Introduction
and Historical Perspective
Ernest Baughman
Process analytical chemistry (PAC) is a field that has existed for several decades in many
industries. It is now gaining renewed popularity as the pharmaceutical industry begins to
embrace it as well. PAC encompasses a combination of analytical chemistry, process
engineering, pro cess chemistry, and multivariate data analysis. It is a multidisciplinary
field that works best when supported by a cross-fun ctional team incl uding members from
manufacturing, analytical chemistry, and plant maintenance.
The use of PAC enables one to gain a deeper understanding of the process. This in turn
can lead to more consistent product, reduced waste, improved manufacturing efficiencies,
overall improvement in the use of resources, improved safety, and the reduced costs that
can be garnered from each of these. In the most rudimentary application of PAC, one can
gain a descriptive knowledge of a process. The process signature that is measured permits
a determination of trending in process parameters. Measurements can be made to give a

direct indication of reaction progress or the composition of a mixture at a given time.
This information about the process can be used to make changes to keep a process
running within some set limits. With additional process information, and kno wledge,
there is the possibility of understanding the parameters that impact a process and the
product quality. Full process understanding includes the identification of key parameters
that can be used to control a process.
In this chapter I mainly discuss my experience over the time period from 1978 to 1994
during which I worked at Amoco Oil and Amoco Chemical companies, now part of
British Petroleum (BP). The last 10 years of that period were spent in a group charged
with finding, modifying or inventing the on-line analytical tools to fit Amoco’s needs.
Among the instruments used were on-line gas chromatographs (GCs), filter spec-
trometers, pH meters, density meters, flash point analyzers, and both infrared and near-
infrared (NIR) spectrometers (both filter photometers and full wavelength ones). Later in
this chapter, some of the tools that were developed will be discussed.
This chapter deals with two main issues: Why PAC is done and how it is done. The
history of PAC is discussed from the perspective of its early origins in petrochemicals.
Further, aspects on early instrument developm ent are presented. Why PAC is done is easy
to answer: to increase the bottom line via improved production efficiency. The benefits of
PAC have already been mentioned – all of which impact a company’s bottom line. How
PAC is done is much more complex, and is difficult to generalize. Many of those who
work in this field continually discover new ways to glean information that allows for
optimization of processing parameters and better control of the process. Better control is
the prime goal as it will improve product quality, result in less waste, increase the safety of
operations, and thus increase profitability.
This chapter is written in first person because I am trying to communicate to YOU, the
person working in this world of PAC who is involved in selecting the projects as well as
the people to work on these process analytical projects. Let me introduce myself as my
view of PAC is colored by what I have seen and done. My PhD is in chemical kinetics, the
measurement of speed of reactions. Since concentration of the reactants plays a role in
governing the speed of the reaction, one needs to know accurately what is present and at

what amounts. In the laboratory setting, this is reasonably easy to do. In the real world of
manufacturing it can be much more involved. Sometimes the label may not match what is
in the barrel. In a batch process, one may not know when a process has reached
completion. In continuous processes, one never knows exact ly what is where in the
process without a means to measure it directly. Add in the uncertainty of the composition
of something like crude oil and it is amazing that the refinery, with its wide variety of
feedstocks, operates at all. (The refinery feed depends on which crude oil can be delivered
and at what price, normally the cheaper the better: a parameter which frequently
changes.)
1.1 Historical perspective
Process analytical che mistry is a field that has developed over many years. Over the past
decade there has been a 5% annual growth in the use of process analytical instruments,
and this growth is expected to continue.
1
More recently, the term process analytical
technology (PAT) has been used to describe this area that is the application of analytical
chemistry and process chemistry with multivariate tools for process understanding. Much
of the industrial PAC work is considered proprietary and does not appear in published
literature until after it has become standard. Some work is found in the patent literature,
especially that dealing with analyzer development. A classic book on process analyzers by
Clevett
2
does a good job of comparing laboratory and process analyzers that use the same
name and is very useful as a general overview. An article
3
by Callis et al. in 1987 defined
five different eras of PAC. These eras discuss an evolution from off-line and at-line
analysis, to intermittent on-line analysis, continuous on-line analysis, then in-line
analysis, and non-invasive analysis. The terminology defining the different eras or
applications of PAC is still in use today, though the generic term of on-line analysis is

often used without distinction from the specific type of interface u sed. A review
4
of the
applications and instrumentation for PAC with 507 references covers advances until
1993. This was the first in what have come to be biannual reviews on PAC publis hed in
Analytical Chemistry. In this series the authors cover the important developments in the
area that occurred in the previous 2 y ears – a demanding task. The current review
5
has
119 references to the literatu re and covers topics including biosensors, chromatography,
2 Process Analytical Technology
optical spectroscopy, mass spectrometry, chemometrics, and flow injection analysis (FIA).
To really help a company to be first-rate in PAC, it is best to take a leadership role in
developing on-line analyzers that will solve the problems specific to your company. It is
important to focus attention on PAC projects that have tangible benefits to the company.
Another approach is to work closely with an instrument manufacturer to find optima l
measurement solutions.
Process analytical chemistry has been performed in the petrochemical industry for
several decades. Some of the earliest univariate tools such as pH meters, oxygen sensors,
and flow meters are still in use today. Along with these, there are many more tools
available including on-line chromatography, spectroscopy (NIR, mid-infrared, mass
spectroscopy, nuclear magnetic resonance spectroscopy, and others), viscosity measure-
ments, and X-ray analys is. An early work
2
on process analyzers has a thorough discussion
of the state-of-the-art of process analyzer technology in 1986, and is still a valuable
reference for those new to the field. It covers analyzers, as well as sampling systems, data
systems, analyzer installation and maintenance, and analyzer calibration – all relevant
topics today when undertaking a PAC project.
Many of the process analytical instruments currently in use were developed by the oil and

petrochemical industries and then adopted by other industries. The very high throughput in
many of these industries requires quick information attainable only by direct on-line
measurements. To define ‘very high throughput,’ let me give an example from one of the
refineries where I have had the pleasure of working. This typical refinery makes about one
million gallons of gasoline per hour, that is about 15 000 gallons per minute. If something
goes wrong and no corrective action is taken for a few hours, thousands of dollars in
product can be lost. In the specialty chemical or pharmaceutical industries where raw
material costs may be orders of magnitude higher, the cost of lost product is proportionally
larger. The typical laboratory response time can be in the range of 4–6 hours. Why so long?
The operator must manually extract a sample from the process and take it to the laboratory,
which may be far from the actual sampling point. The laboratory will then need to calibrate
the instruments because one does not want to give misleading information to the process
operators. After instrument calibration or validation, the sample analysis is started, which
may require more than an hour to prepare and run for each sample. Under such conditions
it is not unusual for the results to be available only 6 hours after the sample was taken. In
situations where an unexpected result is reported, the operator may not take the result to be
accurate and another process sample is taken to the laboratory for analysis before any action
is taken on any process adjustments.
This type of case was observed when we were starting up a new distillate desulfurization
unit at one refinery, and a new type of analyzer was installed. The analyzer reported a
sharp change in the sulfur level of the product. As this was an unexpected result, even
though the laboratory had confirmed it, it required about 15 hours for the operations
people to believe the result. Their argument was that ‘big units couldn’t change that fast.’
If they had accepted the analyzer result, several thousand gallons of diesel fuel would not
have needed to be reproces sed. The change in sulfur level was later traced to a change in
the crude oil feedstock and had been accurately dete cted by the process analyzer.
So, it takes some time. What does this cost in terms of productivity, utilities or work
force? The only way to be sure the petroleum product stays on specification is to make the
Introduction and Historical Perspective 3
product that is over specification, so the ‘bumps’ in the process never put the product

under specification. This amounts to product giveaway. To show how expensive this can
be, one refinery implemented faster on-line tools and saved about US$14 000 000/year
in the first year, new spectroscopic tools were installed, merely by reducing product
giveaway. The product must be manufactured to, at least, the minimum specification
value. In the oil world no one worries about getting more for their money than was
anticipated. In industries where products have defined specification ranges, material that
is not manufactured within the range for a premium-grade product may be sold at a lower
price as a lower grade. In many cases, materials that are not manufactured to the
specification may be harmful to the consumer and must be discarded or, perhaps, can
be reprocessed to be within acceptable limits.
How is the incentive calculated for installation of process analytical instrumentation?
My approach was to determine the current costs of the operation, then find the costs of
the operation when everything was at the optimum and were normally significantly less.
I would take about half the difference to determine my potential savings. Calculating the
economic benefits is an important step in choosing projects to pursue. In the project
planning, it is necessary to do this to secure funding for a project with costs that will
include the time for choosing an analyzer, the analyzer, sampling system, calibration
development, and installation with post-installation training. The business justification of
process analytical projects is critical to the success of such projects and this is discussed
further in Chapters 2 and 11 of this book. No operation that I ever had the pleasure of
working with was perfect. However, the closer one ran to perfection, the greater the
economic payout of the operation. In Clevett’s book,
2
he notes, ‘Success that can be
attributed to the use of process analyzers includes savings in production, product give-
away, operating work force, and energy conservation.’ Other authors
1
note that in the
chemical industry one plant went from 2 to 80 process analyzers and was able to achieve a
35% capacity increase without making any expansions to the plant.

Savings on implementation of real-time analysis can come from the better use of raw
material, less energy consumption, higher throughput or any combination of the abov e.
Reduced raw material usage results in reduced waste. A better-controlled process yields
more products within the specification limits. In batch operations, one frequently ‘holds’
the batch in the reactor awaiting laboratory analysis. With on-line equipment, hold time
is reduced or eliminated. If operating at capacity, eliminating that hold time can con-
tribute to increased manufacturing capacity, which can have a large economic impact.
1.2 Early instrument development
The differences between laboratory and on-line instruments are huge to the point where
often one does not see that they are related. What are some of the commonalities and
differences between an on-line analyzer and a laboratory piece of equipment? Usually the
name is the same. The laboratory is a temperature-controlled, reasonably safe environment
with instrumentation operated by trained chemists. In contrast, the process is frequently
outdoors so temperature control is what Mother Nature gives you. Many processes also
4 Process Analytical Technology