Trends in process analytical technology

Abstract

Since the promotion of Process Analytical Technology (PAT) by the U.S. Food and Drug Administration (FDA), there has been a flurry of activities happening across related fields. This excitement permeates regulatory agencies, professional societies, academia and industry worldwide. This review surveys the PAT related developments that have taken place in the period 2004–2009. It serves as an introduction to PAT, with highlights on the parallel advances and convergence points across various fields and applications. From this review, five common threads are identified from the underlying trends of the recent global PAT endeavor, namely, organisational objectives, enabling sciences, economic outlook, collaborative efforts and emerging trends. There are also six potential gaps that require further efforts to bridge. The overall PAT venture is promising for delivering an integrated systems approach for quality design, process analyses, understanding and control, continuous improvement, knowledge and risk-based management within the FDA 21st century pharmaceuticalcGMP initiative.

---

Introduction

Process Analytical Technology (PAT) has generated lively discussion in recent years within the pharmaceutical industry since its formal introduction through the U.S. Food and Drug Administration (FDA) guidance paper in 2004.¹PAT is intended to play an integral role in the delivery of a science-oriented current Good Manufacturing Practice (cGMP) in 21st century pharmaceutical manufacturing; it forms part of “a regulatory framework that will encourage the voluntary development and implementation of innovative approaches in pharmaceutical development, manufacturing, and quality assurance.”² The term “analytical” in PAT is “viewed broadly to include chemical, physical, microbiological, mathematical and risk analysis conducted in an integrated manner,” with the goal to “enhance understanding and control manufacturing process” that is consistent with FDA's current drug quality system ideology of “quality cannot be tested into products; it should be built-in or should be by design.”¹ Underlying this PAT thrust is a multi-disciplinary and multi-dimensional data paradigm that integrates and crosses several science and engineering sub-disciplines: process chemistry development, process analytical chemistry, spectroscopy, multivariate statistics, chemometrics, chemical engineering, process systems and control engineering.

PAT is also a methodological framework that sits within a cadre of related concepts that include risk-based regulatory approaches, process understanding, quality by design (QbD), critical quality attributes (CQAs), process monitoring, multivariate statistical process control (MSPC), quality risk management, real-time release (RTR), integrated systems approach, continuous improvement, and knowledge management. These concepts are manifestations of more rigorous and integrated attempts by the industry to ensure product quality by taking appropriate actions throughout the pharmaceutical product life cycle. Also, the PAT framework delineated in the FDA 2004 guidance specifically includes biological attributes and applicability to biologics.¹

This review traces the various threads of thinking, methods and applications in the PAT arena so as to highlight the trends, successes, challenges and gaps during its development thus far. Particular emphasis is given to an overview of the PAT framework, processes reported to incorporate some form of PAT, analytical techniques and technologies utilized for PAT applications, process chemometrics, modeling and statistical control, as well as activities occurring in professional bodies, regulatory agencies, consortia, academia and industry that are continuing to shape the subject. Some underlying commonalities and potential gaps in this global PAT endeavor are identified and discussed. The five identified common threads are organisational objectives, enabling sciences, economic outlook, collaborative efforts and emerging trends. The gaps include the requirement of a business value for the adoption of PAT solutions, reluctance of companies to divulge process analytics or PAT applications, integration of PAT technologies is still some distance from maturity, more demonstration of process control improvement through process innovation and PAT, the managing of a multi-disciplinary team required for implementing PAT processes, and the present lack of specific curriculum for teaching PAT in academia.

Why PAT?

The widely accepted definition of PAT for the pharmaceuticals sector originates from the U.S. FDA,¹ and has been adopted by ICH Q8 guidance and ASTM International Technical Committee E55.²

“A system for designing, analyzing, and controlling manufacturing through timely measurements (i.e., during processing) of critical quality and performance attributes of raw and in-process materials and processes with the goal of ensuring final product quality.”¹

The concept originates from the desire of the regulators to shift control of product quality towards a science-based approach that explicitly attempts to reduce the risk to patients by controlling the manufacturing based on understanding of the process. The situation before these concepts emerged (until the early years of the century) was more focused on post-production testing as a means to reject off-specification from processes that might be “out of control”. Control of the process aimed to attempt to avoid any batch-to-batch changes in the raw materials, process conditions and equipment. The accepted wisdom was that these processes could not be changed as there was limited understanding of the potential impacts of change and therefore re-registration would be required to demonstrate that a modified process still made the right material.

The new PAT focus emphasizes the following:

• Good measurements

• Linking those measurements to (causal) understanding of the process, and in particular those aspects of the process critical to control the risk to patients

• Anticipation of problems, or controlling them as detected rather than dealing with them post hoc.

The motivation for such an approach lies not only in the control of risk to patients, but also in the economics of production. Any reduction in the number of failed batches would have a direct economic benefit. Also, parametric release of product (approving the release of a material based on in-process testing) could enable significant reduction in finished materials awaiting testing, which would reduce the working capital requirements of a process. Further, a high level of confidence in process outcome would allow lower capital investment as there would be no need to build in “extra capacity” to make the off-specification product.

Historic perspectives

As PAT itself is a multi-dimensional paradigm, its historical development can be traced from multiple angles. This section provides some historical perspectives from the standpoints of regulatory agencies, professional organizations and consortia, process analyzers and instrumentation, process chemometrics, modeling and control, commercial companies, literature and publications.

Regulatory agencies

The PAT initiative evolved from the first public discussion in July 2001 organized by the Advisory Committee for Pharmaceutical Sciences (ACPS) of the U.S. FDA.³ Subsequently from 2002, the scope of PAT was worked out alongside a larger initiative to “enhance and modernize the regulation of pharmaceutical product quality”³ that resulted in the final report for Pharmaceutical cGMPs for the 21st century.² Between 2001 and the release of the PAT final guidance in September 2004,¹ extensive deliberations took place both within the various departments of FDA and with global stakeholders across government, regulatory, academia, industry, and consumer sectors. Draft PAT guidance was published in August 2003. Simultaneously, other activities were rolled out, which include internal training on PAT within FDA and communicating the PAT framework through external workshops, public meetings, conferences, and scientific literature. To ensure the latest science and technology are incorporated into policy making, training and regulatory implementation, FDA initiated research internally and through external collaborations. For example, FDA worked with companies such as Pfizer and Novartis, and academic institutions like Duquesne University Centre for Pharmaceutical Technology (DCPT), the Centre for Pharmaceutical Processing Research (CPPR) at Purdue University, and Engineering Research Centre for Structured Organic Composites (C–SOC) at Rutgers.³ Both CPPR and C–SOC consortia are funded by the National Science Foundation. Further, regulatory authorities across the world have collaborated with industry bodies and professional associations, contributing to the interpretation and development of the concept.

This global PAT endeavor and its counterpart in 21st century Pharmaceutical cGMPs continued to evolve after the release of the 2004 cornerstone documents. FDA expressed the belief that “the harmonization of international scientific standards on drug product quality will promote technological innovation for enhanced health promotion and protection.”² In working towards this harmonised “pharmaceutical quality standards or requirements to the fullest extent possible,”² FDA continued to forge collaborations with regulatory counterparts and participate in multilateral international forums such as the International Conference on Harmonisation of the Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH). By having FDA PAT Team representation in the ASTM International Technical Committee E55 looking into manufacture of pharmaceutical products and pharmaceutical application of PAT, there are to date six standards published with ten other work items under deliberation since its inception in 2003.² FDA also actively sought membership in the Pharmaceutical Inspection Cooperation Scheme (PIC/S) since expressing this intent explicitly in the Pharmaceutical cGMPs for 21st Century Final Report.^2,4

The European Union counterpart of FDA, the European Medicines Agency (EMEA or EMA), has also been active in the area of PAT. Prior to the release of FDAPAT guidance in 2004, EMEA and FDA had been deliberating the ICH topic Q2 on the use of analytics in pharmaceutical industries. Both agencies published guidelines on this area in the mid-1990s.⁵ Along the theme of parametric release, EMEA issued guidance in 2001, which included related topics like process monitoring and GMP compliance.⁶ In 2006, the EMEA PAT team published a reflection paper on chemical, pharmaceutical and biological information to be included in PAT implementations.⁷ This is a working document that highlighted, amongst other issues, specific points on quality risk management, design of experiment, data acquisition and chemometrics approach to establishing the design space, manufacturing process development and control of critical steps, intermediates, excipients and drug product.

In general, regulators are limited in the extent to which they can control the way in which technologies like PAT are adopted. Prescribing the particular sensors to be used, for example, would require resources beyond those available to the regulators, and would also lead to potential liability on their part. Instead, the approach has to be one of asking industry to show the regulator that the approach is workable.

Professional organizations and consortia

The FDA PAT initiative has grown to involve a network of multiple organizations and consortia engaged in activities to support the interpretation and delivery of PAT to real processes. Table 1 shows a list of some important contributors to PAT. Two organizations mentioned in the FDA 2004 documents^1,2 are the ICH⁸ and ASTM International Committee E55.⁹ The Committee E55 was initiated in 2003 between FDA and ASTM International. It presently comprises of ca. 290 members representing 17 countries, who actively deliberate on PAT and pharmaceutical manufacturing related standards under its four technical subcommittees E55.01, E55.02, E55.03 and E55.91.^9,10 Their current work items and published standards are shown in Table 2.⁹ The published standards by E55 are found in the Annual Book of ASTM Standards Volume 13.01.^10,11

Table 1 Professional Societies and Consortia

Organisation	Description	Website
ASTM International Committee E55	ASTM Committee E55 on Manufacture of Pharmaceutical Products, formed in 2003, was brought together following the major overhaul initiated by FDA for the pharmaceutical industry. The scope of this committee includes standardizing nomenclature, definition of terms, recommending practices, guides, test methods, specifications and performance standards for the manufacture of pharmaceutical products. Standards from E55 is published in the Annual Book of ASTM Standards, Volume 13.01.	http://www.astm.org/COMMIT/COMMITTEE/E55.htm http://www.astm.org/COMMIT/COMMITTEE/E55.htm http://www.astm.org/COMMIT/COMMITTEE/E55.htm http://www.astm.org/COMMIT/COMMITTEE/E55.htm http://www.astm.org/COMMIT/COMMITTEE/E55.htm
Center for Process Analytical Chemistry (CPAC)	CPAC is a consortium established at the University of Washington in 1984, comprising of members from industry, national laboratories and government agencies. It has been involved in developing new measurement approaches and promoting research, communications and partnership in areas related to PAT and process control.	http://www.cpac.washington.edu
Centre for Process Analysis and Control Technology (CPACT)	CPACT was established in 1997 and currently comprises of 7 universities and 18 companies in the UK. It seeks to be a platform for multidisciplinary exchanges between chemical and process engineers, analytical chemists, chemometricians, control systems engineers, etc. from both academia and industry to engender solutions to generic problems in process monitoring and control.	http://www.cpact.com
The International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH)	ICH is a collaborative endeavor involving worldwide regulatory authorities and experts from the pharmaceutical industry, in particular the three regions of Europe, Japan and the United States. The ICH work towards international harmonisation through discussions on scientific and technical aspects of new medicinal product registration.	http://www.ich.org
International Society of Automation (ISA)	ISA was founded in 1945 with the mission of setting global standards for automation via certification, education, publication, developing standards and hosting conferences and exhibitions. It currently has a global membership of more than 4000, and has published over 150 standards for automation and control systems.	http://www.isa.org
International Society of Pharmaceutical Engineering (ISPE)	Founded in 1980, ISPE is currently the world's largest not-for-profit society in pharmaceutical science and manufacturing with the purpose of educating and advancing pharmaceutical industry and professionals. It has over 25,000 members representing experts, technologists, regulators, consultants and students in 90 countries.	http://www.ispe.org
National Institute for Pharmaceutical Technology and Education (NIPTE)	NIPTE is a not-for-profit organization comprising of over ten member academic institutions in North America. It works closely with the U.S. FDA and disperses funding to research projects related to implementation of Quality by Design (QbD). NIPTE also produces roadmaps for research and pharmaceutical technology education.	http://www.nipte.org

---

Table 2 ASTM International Committee E55 Standards

E55 Technical Sub-committees	Published Standards	Work Items
E55.01 PAT System Management	• E2476-09 Standard Guide for Risk Assessment and Risk Control as it Impacts the Design, Development, and Operation of PAT Processes for Pharmaceutical Manufacture	• WK5930 Standard Practice for Risk Management as it Impacts the Design and Development of Processes for Pharmaceutical Manufacture
		• WK5935 Process Understanding Related to Pharmaceutical Manufacture and Control
		• WK9192 Standard Guide for the Application of Continuous Processing Technology to the Manufacture of Pharmaceutical Products
		• WK12892 Standard Guide for Process Sampling
E55.02 PAT System Implementation and Practice	• E2474-06 Standard Practice for Pharmaceutical Process Design Utilizing Process Analytical Technology	• WK9182 Standard Practice for Qualification of PAT Systems
		• WK9191 Standard Practices for Multivariate Analysis Related to Process Analytical Technology
		• WK20498 Multivariate Statistical Process Control for Manufacturing Processes
		• WK16888 Validation of PAT Methods
E55.03 General Pharmaceutical Standards	• E2500-07 Standard Guide for Specification, Design, and Verification of Pharmaceutical and Biopharmaceutical Manufacturing Systems and Equipment	• WK11898 On-line Total Organic Carbon (TOC) Method Validation in Pharmaceutical Waters
	• E2503-07 Standard Practice for Qualification of Basket and Paddle Dissolution Apparatus	• WK15778 Guide for Science-based and Risk-based Cleaning Process Development and Validation
	• E2537-08 Standard Guide for Application of Continuous Quality Verification to Pharmaceutical and Biopharmaceutical Manufacturing
E55.91 Terminology	• E2363-06a Standard Terminology Relating to Process Analytical Technology in the Pharmaceutical Industry

---

The ICH was officially launched in April 1990. Its emergence was driven by recognition of the need for harmonization of regulation and regulatory approach in pharmaceuticals. The trigger was from efforts in the European Community (now European Union) to create a single market for pharmaceuticals, but the involvement of the World Health Organisation (WHO) promoted a wider-reaching collaboration of the regulators and industry bodies from the US, Japan and Europe. ICH is a bilateral regulatory-industry endeavor that seeks to harmonise selected Topics under the scopes of Quality (Q), Safety (S), Efficacy (E), and Multidisciplinary (M), of which the first three are the “basis for approving and authorizing new medicinal products.”⁸ With this focus, various Expert Working Groups (EWGs), Implementation Working Group (IWG), and Informal Working Groups were formed to deliberate the science and technical aspects of each Harmonization Topic. Since most new medicines and drugs are developed in Western Europe, Japan and the United States, ICH is highly influential. Although in principle the ICH scope and guidelines are only applied to new medicinal product registration in these three geographical regions, its reach is in practice wider—with representatives of Canada, European Free Trade Association (EFTA) and the WHO acting as a link to non-ICH countries and regions, and also through the ICH Global Cooperation Group membership. There are presently over 50 harmonised tripartite guidelines with Common Technical Document (CTD) produced from the ICH. Electronic versions of these are available on the ICH website. These guidelines contain “agreed-upon state-of-the-art scientific guidance for meeting technical requirements for registration” within the three ICH regions. They are to be adapted and executed in each region according to national or regional requirements. More details on the ICH history, future, structure, process of harmonization, and published guidelines can be found on the ICH website.⁸ In relation to FDAPAT and 21st century cGMP, five ICH guidelines under scope on Quality (Q) are directly relevant, namely Q7 Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients,¹² Q8 Pharmaceutical Development,¹³ Q9 Quality Risk Management,¹⁴ Q10 Pharmaceutical Quality Systems,¹⁵ and Q11 Development and Manufacture of Drug Substances.¹⁶

The International Society of Pharmaceutical Engineering (ISPE) was founded in 1980 and has been active in educating and advancing professionals in pharmaceutical science and manufacturing.¹⁷ It has a global network of local affiliates and chapters, committees and Communities of Practices (COPs). Amongst the ISPE COPs, the Good Automated Manufacturing Practice (GAMP) and Process Analytical Technology communities are most relevant to the FDA PAT framework. ISPE also provides certification courses such as the Certified Pharmaceutical Industry Professional™ (CPIP™), organizes conferences, webinars and publishes regular articles in ISPE publications such as Journal of Pharmaceutical Innovation, Pharmaceutical Engineering, Pharmaceutical Research, technical documents and periodic newsletters. In June 2007, ISPE launched the Product Quality Lifecycle Implementation^® (PQLI^®) initiative for promoting “excellence in pharmaceutical production embracing product quality, efficiency, and sustainability as well as regulatory cooperation and compliance.”^18,19 This PQLI will be at least a five year endeavor for seeking practical approaches within a technical framework required for implementing the ICH guidelines Q8(R2), Q9, and Q10. It also aims to better the understanding of “enhanced Quality by Design (QbD) approach.” Working groups are set up to facilitate this PQLI initiative and their findings will be published as white papers and articles in the Journal of Pharmaceutical Innovation and Pharmaceutical Engineering. Some representative articles on the PQLI key topics of criticality, design space and control strategy were published in 2008.^19–23

Another nonprofit organization with activities closely relevant to PAT is the International Society of Automation (ISA). It was founded in 1945 and has the mission of becoming “the standard for automation globally by certifying industry professionals; providing education and training; publishing books and technical articles; hosting conferences and exhibitions for automation professionals; and developing standards for industry.”²⁴ ISA hosts divisions under its two major departments of Automation and Technology and Industries and Sciences, of these the Analysis Division (AD), Automatic Control Systems Division (ACOS), Computer Technology Division (COMPUTEC), Process Measurement and Control Division (PMCD), and Food and Pharmaceutical Industries Division (FPID) are of direct relevance to PAT and 21st century cGMP. Currently ISA has published over 150 standards for automation and control systems, which were produced through consensus building from a global expertise pool of more than 4000 industry experts. Two series of standards under ANSI/ISA-S.88 Batch Control and ANSI/ISA-S.95 Enterprise/Control System Integration contain concepts that overlap with the FDA PAT framework (see Table 3).

Table 3 ANSI/ISA-S.88 and ANSI/ISA-S.95 series of standards

ANSI/ISA Standard	Description
ISA-88.00.01-1995	Batch Control Part 1: Models and Terminology
ANSI/ISA-88.00.02-2001	Batch Control Part 2: Data Structures and Guidelines for Languages
ANSI/ISA-88.00.03-2003	Batch Control Part 3: General and Site Recipe Models and Representation
ANSI/ISA-88.00.04-2006	Batch Control Part 4: Batch Production Records
ANSI/ISA-95.00.01-2000	Enterprise/Control System Integration Part 1: Models and Terminology
ANSI/ISA-95.00.02-2001	Enterprise/Control System Integration Part 2: Object Model Attributes
ANSI/ISA-95.00.03-2005	Enterprise/Control System Integration Part 3: Activity Models of Manufacturing Operations Management

---

Besides the aforementioned professional organizations, there are a number of consortia that forge links between academia, industries, and regulatory agencies (see Table 1). The National Institute for Pharmaceutical Technology and Education (NIPTE) is a North American-based group of academic institutions actively developing QbD implementation with FDA through research partnerships, and also producing roadmaps for pharmaceutical research, technology and education. The Center for Process Analytical Chemistry (CPAC) was established at the University of Washington in 1984 as a National Science Foundation (NSF) Industry/University Cooperative Research Center (I/UCRC).²⁵ Over the years, it has been successful in “bridging the gap between basic research and full-scale process/product development through fostering interactions between academia, industry and national laboratories.”²⁵PAT has been a priority at CPAC as evidenced by its Chemometrics for On-line Process Analysis initiative (COPA) and also the most recent Summer Institute (SI) on the theme Micro Instrumentation for High Throughput Experimentation and Process Intensification: A Valuable Tool for Process Analytical Technology (PAT).²⁶ A similar consortium based in the UK is the Centre for Process Analytics and Control Technologies (CPACT), which was established in July 1997 through the Foresight Challenge with an EPSRC/DTI award.²⁷ It has successfully completed Phase I and II projects with a total grant of ca. £5M that supported research themes such as process analytics, process chemometrics, process modeling, inferential estimation, control and optimisation and the CPACT software. In embarking on Phase III, the CPACT membership structure is enlarged from its initial 3 universities to 7 universities with an added new aim to “support the entire Process Industries Supply Chain to enhance profitability, quality and competitiveness.”²⁷

Another consortium that is working towards “interoperability between analytical and data processing systems” is the OPC Analyzer Devices Integration (ADI) working group that was formed in 2008 under the OPC Foundation.^28,29 The OPC Foundation is a non-profit organization setup for establishing “standard OLE/COM interface protocols intended to foster greater interoperability between automation/control applications, field systems/devices, and business/office applications in the process control industry.”³⁰OPC is the acronym for “Object Linking and Embedding (OLE) for Process Control”, which is an open connectivity architecture based on Microsoft^®OLE and Component Object Model (COM) technology for defining objects, methods and properties for real-time information communication transfers between OLE/COM compliance enabled devices, servers and software applications (e.g. DCS, SCADA, PLC, smart field devices, analyzers, mathematical modeling software, process control systems, etc.). The scope of OPC includes specifications such as (i) standardized COM interfaces for client-server exchanges through its distributed version (DCOM™) provided by Microsoft, (ii) developing a set of OLE Automation interfaces for communications between business applications such as Microsoft Excel, Visual Basic supported by OPC clients.³⁰OPC functionalities include security, online and historic data access, batch and historic alarm and event handling, which in the PAT context is appropriate for meeting the FDA 21 CFR Part 11 requirements.⁷⁴ Furthermore, the OPC Unified Architecture (OPC UA) was released by the OPC Foundation as a new standard for data communication and information modeling catered to process automation and related IT needs. This OPC UA standard is anticipated to over time replace the Microsoft DCOM based specifications DA, HDA and A&E previously defined by the OPC Foundation.³¹ It contains all the functionalities of its forerunners and additional benefits for expediting wider usability; such as (i) platform independence, (ii) adopting state-of-the-art Web service technology, (iii) interoperability on devices, DCS, MES, ERP systems through standardized protocol and meta model, (iv) faster data exchange via binary encoding, (v) a secured address space model that allows complex metadata exposure through objected-oriented programming (OOP) concepts and full-meshed network of nodes related by multiple reference types, and (vi) scalability for varied server size and complexity. To facilitate deployment of classic OPC codes for clients and servers into the OPC UA, wrappers and proxies are provided by the OPC Foundation. The migration of codes to higher-level programming languages is achieved through the OPC UA server communication platform with the Microsoft .NET framework. The OPC UA technology is geared towards supporting other industry standards such as the ANSI/ISA-S.95, ANSI/ISA-S.88, EDDL, MIMOSA, OAGiS.²⁹

In November 2009, OPC Foundation announced the OPC ADI Specification to enable “true [analyzer] plug-and-play multivendor interoperability” so as to give “end-users a seamless experience in data management and integration for solutions such as Process Analytical Technology (PAT).”²⁸ The OPC UA technology is relied upon by the OPC ADI Information Model as a common platform for the goal of producing process data from varied analyzers (spectroscopic, chromatographic, imaging, particle size, etc.) after application of a chemometric model on the scaled data (i.e., measured raw data such as absorbance, intensity, etc.). The final process data is typically scalar values, such as chemical species concentration, moisture content, hardness measure, etc., which can be used for process control.²⁹ The ADI information passing of complex metadata (analyzer type, configuration, accessories, measurement channel, sampling point, spectral data, etc.) is facilitated using (i) stream concept, (ii) layered information modelling through the creation of the address space on the OPC UA server that instantiates all nodes and interconnecting them via references, and (iii) inherent object-oriented programming (OOP) paradigm for data abstraction, encapsulation, polymorphism and inheritance. More technical details on the aforesaid ADI Information Model implementation can be found in a white paper published by the OPC Foundation.²⁹ A significant number of commercial companies are supporting this OPC ADI Specification for their product development. They include ABB, Abbott, Arla Foods, Bristol-Myers Squibb, BRL Consulting, Bruker Optics, CAS, Foss, GlaxoSmithKline, Kaiser Optical Systems, Inc., Malvern Process Systems, Mettler-Toledo Autochem, Inc., Novartis, Pfizer Inc., Sentronic GmbH, Siemens AG, Software Toolbox, Inc., Sympatec GmbH, Thermo Fisher Scientific, Umetrics, Yokogawa.²⁸ Some activities of these companies in relation to classic OPC, OPC UA and ADI technologies will be mentioned in the later sections.

Industrial initiatives from individual companies

Alongside the efforts that have been shared through international bodies and consortia, individual pharmaceutical companies have devoted much effort to the interpretation of the PAT concept to their activities. The PAT concept is typically integrated with other ways of working. Companies face choices in the way they “roll out” the techniques—for example whether this is “technology push” from development or pull from manufacturing. Pfizer, for example, embedded PAT in their “Right First Time” (RFT) initiative.³² They recognized, also, that adoption would take time; the approach had three stages of capability development—firstly being able to obtain better process understanding, secondly being capable of identifying critical quality attributes and only then being able to replace traditional QA and QC. AstraZeneca reported the establishment of an internal centre of excellence in PAT, with developments aligned with the R&D activities rather than driven by manufacturing.³³

Process analyzers and instrumentation

Process analyzers constitute one of PAT tools mentioned in the FDA 2004 guidance.¹ From the perspective of process analyzer development, the forerunner of PAT is the sub-discipline under analytical chemistry termed as Process Analytical Chemistry (PAC). It originated from analysis needs from the oil and petrochemical industries,³⁴ and its history was reported to date back some 70 years.³⁵ Process analyzers³⁶ are incorporated in process streams and equipment to monitor process conditions for eventual quality control of chemical products. In its infancy, PAC was largely undertaken in a time consuming manual fashion involving off-line sampling, sample transportation, and following analysis protocols in a central analytical laboratory.³⁵ The lag time between sampling and analysis reporting was thus significantly long. Demand for achieving timely process monitoring became the impetus for innovation and automation of process analyzers, which over time led to modern day equivalents for real-time monitoring.^35,37 Many existing process analytical instruments were thus developed and applied within the oil and petrochemical industry before being deployed in other industrial processes.³⁴

Since the beginning of PAC, process states of chemical unit operations were ascertained indirectly from physico-chemical properties obtained via analytical measurements. This indirect analytical inference also holds true for PAT implementations and as such necessitates data analysis and modeling to extricate meaningful chemical and process information (see below).³⁸ The measured properties can be univariate (scalar) quantities such as process flowrate, localized temperatures, dissolved oxygen, pH, etc. Other more sophisticated analytical measurements that involve multivariate (vector and matrix) quantities, such as chromatography and spectroscopy, were steadily introduced over time. In general, PAC measurements can be performed off-, at-, on- or in-line, such as those using spectroscopic methods.³⁹ The time delay between sampling and analysis ranges from fast time-scales of one second for in-line analysis to as long as one day for off-line methods.⁴⁰ The recent two decades saw tremendous progress for the incorporation of on- and in-line (in situ or operando)† process monitoring using advanced spectroscopic instrumentation. These include: UV-Visible (UV-Vis),^41,42 near- and mid-infrared (NIR and MIR),^43–45 and Raman spectroscopy.^46–51 This rise was facilitated by related technological advances in solid-state detectors,^47,48,52 fiber-optics,^{37,47,53–55} and instrumentation innovations for in situ sampling.^{12,41,43,44,47–51} The introduction of commercially available advanced instrumentation in the last ten years that allows microspectroscopic measurements in the NIR, MIR and Raman spectroscopies opened another level of data complexity termed as hyperspectral data.^56–62 Such vibrational microspectroscopic techniques capture spatially and spectroscopically resolved information of microscopic samples via chemical mapping^56,58,59 or imaging^57–59,61 and were applied for PAC and PAT applications.^63–70 Hyperspectral data pose greater numerical challenges to their analysis and modeling as they contain voluminous multi-way data that are more than two-dimensional in variations.^71–73 The latest techniques employed for PAC/PAT applications include effusivity, tomography, NMR, acoustic and tetrahertz spectroscopies. Further mention of various aforesaid process analytical instrumentation and their applications to processes are provided in the section PAT Applications.

The development of analyzers suitable for use in industrial processes also poses some practical problems in ensuring that the devices are robust, meet safety requirements (for example for use in the presence of potentially flammable atmospheres) and the avoidance of interference by deposits on optics. A further issue that is a major challenge is ensuring that all software associated with PAT applications is suitable for use and can be validated. The FDA requirements are laid out in FDA 21 CFR Part 11, and guidance is available for the generation of compliant software.^74,75

Process chemometrics

Increasingly sophisticated process analyzer technologies produce large datasets that require appropriate numerical strategies to unravel chemical information (or process signature) and associated process states encoded within the analytical data. According to the FDA PAT guidance, “based on the level of process understanding, these signatures may also be useful for process monitoring, control, and end point determination when these patterns or signatures relate to product and process quality.”¹ With increasing know-how and affordability of spectroscopic and chromatographic instrumentation for on-line and in-line process analysis, PAC or PAT datasets are often intrinsically multivariate in nature. Furthermore, as aforementioned, the potential deployment of microspectroscopic analysis for PAT applications raises data complexity to multi-way dimensions. As all analytical measurements encode information about their respective samples (which may be off-line or real-time in situ for PAT applications), it is de rigueur to decode and extract such information using numerical methods.³⁸ Chemometrics methods are relied upon for this purpose.^38,76,77

Chemometrics evolved from the application of multivariate analytical and statistical techniques in chemistry. Such mathematical methods had previously started off within behavioral sciences and social sciences, such as psychometrics and econometrics. The word “Chemometrics” was initially coined by the Swedish chemist Svante Wold for his research grant application in 1971.⁷⁸ Over the past four decades, thriving developments in this sub-discipline of chemometrics under Analytical Chemistry saw the growth of diverse numerical techniques that were applied to varied analytical applications. The overall progress in chemometrics was aided by concurrent advancements in (i) analytical instrumentations for chemical research, (ii) computers and micro-electronics, and (iii) scientific programming languages and related computational platforms. Interestingly, industrial partnership played an important role in catalyzing the development of chemometrics.⁷⁹ At present, chemometrics encompasses many important fields of chemical data analysis, notably, statistical analysis, optimization, signal processing, multivariate curve resolution, multivariate calibration, regression, pattern recognition, parameter estimation, structure-property relationships, cluster analysis, discrimination analysis, multi-way analysis, factor analysis, library search, informatics, artificial intelligence, etc.^78,80

In a 2002 review article, Lavine and Workman, Jr. reported that chemometrics had over the years caused a paradigm shift towards multiple experimentation and multivariate analysis.⁸¹ They also mentioned the pharmaceutical industry had already embraced the chemometrics approach.⁸¹ Deliberate efforts were undertaken by three groups to promote regulatory compliance of chemometrics methods, namely, (i) FDA Subcommittee on Process Analytical Technologies (PAT), (ii) ASTM International Main Committee E13 on Molecular Spectroscopy and Chromatography, and (iii) Chemometrics for On-Line Process Analytics (COPA) initiative under CPAC.^81,82 It is therefore of no surprise that “multivariate mathematical approaches” were stated as one PAT tool in the FDA PAT guidance.¹ Though the term “chemometrics” is not explicitly mentioned in the same FDA guidance document, from the subsequent response in public literature it is clear that chemometrics has been taken as synonymous with the “multivariate tools” noted in the guidance. The implementation of chemometrics in PAC/PAT was recently reviewed by various authors, which offer a good mix of historical background,^79,83,84 assorted multivariate techniques utilized,^79,84 pitfalls to avoid,⁸⁵ and practical advice for developing in-house chemometrics expertise.⁸³

Process modeling and control

Another aspect of FDAPAT guidance is concerned with process control tools.¹ The ability to bring about control of unit operations in the manufacturing scenario presupposes a certain level of process understanding. It is highlighted in the PAT framework that there exists a “strong link between product design and process development” that is “essential to ensure effective control of all critical quality attributes.”¹ Furthermore, “process monitoring and control strategies are intended to monitor the state of a process and actively manipulate it to maintain a desired state.”¹ This practically means that the success or failure of PAT control strategies resides not only on the elucidation of pertinent process signatures from analyzer signals in real-time, but also on a workable theoretical or empirical process model that sufficiently describes the transformation process(es) occurring within a unit operation. To automate control, this process model would have to be a mathematical one that relates “quality attributes and measurements of critical material and process attributes.”¹

Historically, such process modeling and control can be achieved from two different domains of expertise. One is through process chemometrics and the other chemical process systems engineering. The former is the offspring between PAC and chemometrics, whilst the latter being a sub-discipline under chemical engineering. Broadly, process chemometrics comes from a statistical, data-driven perspective while chemical process systems engineering comes from a philosophy of bottom-up mathematical modeling ranging from first principles to semi-empirical and empirical models. Of course, there are opportunities to benefit from both approaches which should be seen as complementary. Qualitative modeling, too, has received considerable attention. This includes the modeling of business processes and logic^86,87 as well as the BRITEST qualitative process modeling tools.^88–90 All of these have useful roles to play in the representation of process understanding and thus in making the link between PAT and process control.

Since both PAC and chemometrics fall within the same knowledge domain of Analytical Chemistry and their advances were entwined, it is natural to find a large existing literature on process modeling and process control classified alongside process monitoring under the umbrella of process chemometrics.^35,82,91 As both on-line and in-line spectroscopic monitoring are increasingly popular, it is important to highlight process modeling via chemometrics analysis of spectroscopic data.⁹² Several good reviews in this area were published.^93–97 The chemometrics modeling of process spectroscopic data often involves data preprocessing, calibration, factor reduction and attributes prediction to unravel underlying chemical information encoded in the measured spectroscopic signals. The elucidated chemical information relates to process performance and is exploited by various process control strategies.^98–101

Taking the lead from the FDA PAT guidance in 2004, the chemical engineering community began to probePAT from the engineering design perspective, which is taking place within related disciplines such as process systems engineering (PSE) and computer aided process engineering (CAPE).^102–104 In the larger context, this community is not only forward looking in considering opportunities and challenges surrounding globalized markets and demands, sustainability, environmental protection, complex chemicals manufacturing and new processing methods, it is also positing a collaborative multi-disciplinary approach for model development involving partnership with other communities.¹⁰⁵ However, it is also worth noting that the PSE community has made only limited inroads into pharmaceutical processing; the tools that have been so successful in modeling petrochemical processes are as yet much less cost-effective for the complex, multiphase processes of the pharmaceutical industry. It will be many years before pharmaceutical process models can be developed and used with the same level of confidence that is current in the bulk chemicals sector. The proposed workplan of NIPTE reflects this gap and the opportunity to improve the value of modeling.¹⁰⁶

Commercial companies

The PAT initiative catalyzed several parallel commercial developments that complement PAT developments and investments within the pharmaceutical multi-national companies (MNCs). Furthermore, in recent years there is the orientation and inclusion of PAT and QbD in various commercial technological platforms offered. Table 4 lists some of the major players in the market and can be classified under companies that provide (i) process analytical instrumentation, (ii) chemometrics software, and (iii) process systems integration and automation. There are other companies that provide products as PAT solutions which are not mentioned for reasons of space.

Table 4 Commercial Companies offering PAT related technologies

Company	Technology	Website
Process Analytical Instrumentation
Bruker Optics	• Mid-infrared, near-infrared, and Raman Spectrometers, including the OEM IRcube	http://www.brukeroptics.com
	• Software: OPUS Suite
Foss	• XDS series of near-infrared (NIR) spectrometers	http://www.foss.dk
Kaiser Optical Systems, Inc.	• RamanRXN Systems with varied immersion and non-contact optics, PhAT, Airhead, Pilot probes	http://www.kosi.com
	• Software: HoloGRAMS, HoloPro, HoloReact and SDK for third party OEM integration
Malvern	• Insitec particle size analyzers	http://www.malvern.com/LabEng/industry/pat.htm
	• Spectral Dimensions SyNIRig near-infrared chemical imaging (NIR-CI) system
	• ISys software for NIR image analysis
Mettler Toledo	• Automated lab reactors and reaction calorimeters:	http://www.mt.com/autochem
	EasyMax™, MultiMax™, RC1™
	• In situ analytics: ReactIR™, FBRM^® and PVM^®
	• iC/iControl software suite, including iSafety, iC PAT
Thermo Fisher Scientific	• On-line and in-line process analytical instrumentation (elemental, gases, near-infrared, ultraviolet, mass spectrometry, etc.)	http://www.thermo.com
	• Software: Atlas CDS, EP Series, GRAMS, GRAMS/3D OMNIC, Spectral DB, Spectral ID
Process Enterprise System Providers
ABB	• Industrial^IT Technology (includes Extended Automation System 800xA and eXtended Process Analytical Technology (xPAT) platform)	http://www.abb.com
	• Analyze^IT Suite
GE Intelligent Platforms	• Proficy Process Systems and Software (includes Historian, HMI/SCADA (iFIX and CIMPLICITY), Batch Analysis, Real-Time Information Portal, RX modules)	http://www.ge-ip.com
	• Partnership with Symbion (includes RX Script Assistants, Tcl command set)
National Instruments	• LabVIEW (includes toolkits such as DSC, Real-Time, PID, Data Connectivity, OPC client, etc.)	http://www.ni.com
	• PAC hardware (includes PXI/CompactPCI, CompactRIO, Compact FieldPoint, Industrial PC, Wireless Sensor Networks, HMI, etc.)
	• ANSI C LabWindows/CVI
	• MathScripts RT
Siemens	• SIMATIC series (including, WinCC, sensors, HMI, controllers, etc.)	http://www.siemens.com
	• Data acquisition hardware
	• SIPAT (includes Base Station, High Level, analyzer instrument interface, offline Model builder, open interfaces, etc.)
Chemometrics Software
Applied Chemometrics, Inc.	• Chemometrics, Toolbox	http://www.chemometrics.com
	• Factor Analysis Toolbox
CAMO	• Unscrambler Software Suite (includes On-line Predictor OLUP, Optimizer, Classifier, GenX family)	http://www.camo.com
	• Quali-Sense
Eigenvector Research, Inc. (EVRI)	• PLS Toolbox	http://www.eigenvector.com
	• MIA Toolbox
	• Solo
Thermo Fisher Scientific	• GRAMS Suite 9.0 (includes GRAMS/AI, Spectral DB, Spectral ID, GRAMS/3D)	http://www.thermo.com
	• GRAMS IQ, IQ Predict, and PLSplus/IQ chemometrics packages
InfoMetrix	• Pirouette 4.0	http://www.infometrix.com
	• InStep
Umetrics	• SIMCA Software Suite (P/P+, Q/Q+, QP/QP+, QM/QM+, SBOL, 4000)	http://www.umetrics.com
	• MODDE 9.0 (including M-Link)

---

In the area of process analytical instrumentation, the companies surveyed herein are: Bruker Optics, Thermo Fisher Scientific, Foss, Kaiser Optical Systems, Malvern Instruments, and Mettler Toledo. Bruker Optics offers a range of spectroscopic PAT solutions, especially in vibrational spectroscopy, for process monitoring and control needs. The spectrometers are manufactured as compliant to regulatory requirements such as 21 CFR Part 11, USP 1119, IQ/OQ/PQ and GMP. Various industry standard communication protocols such as 4-20 mA, Ethernet, Modbus, Profibus DP, and OPC are also catered.¹⁰⁷ Its 21 CFR Part 11 compliant OPUS software 6.5 contains different interfaces for routine laboratory analyses (OPUS/LAB), spectra library search and identification (OPUS/IDENT), and the OPUS/PROCESS package that enables standardized OPC server/client protocols for integration with process control networks and communication with DSC.¹⁰⁷ There was a report in the use of OPUS for quantitative PLS analysis in PAT-based crystallization studies.¹⁰⁸ Thermo Fisher Scientific offers on-line and in-line process instrumentation for analysing elements, gases, sulfur, particle size, near-infrared (NIR), UV and mass spectrometry.¹⁰⁹ These analyzers when suitably combined form part of PAT solutions in manufacturing scenarios, e.g., the combination of in-line NIR with on-line mass spectrometry reveals process-related CQAs for a fermentation-based biotech process flow.¹¹⁰ The Thermo Fisher Scientific OMNIC and GRAMS suite of software platforms enable spectroscopic data processing and analyses. More details on GRAMS are provided in the commercial chemometrics software segment below. Foss has a long history of automating analytical methods in food and pharmaceutical industries.¹¹¹ Its XDS series of NIR spectrometers provides PAT solutions from at-line laboratory usage to real-time process analytics. The Foss NIR Process Analytics™ can cater for up to 4 or 9 sampling points for various in-line applications such as cell culture in bioreactors,¹¹² osmolality and pH,¹¹² and monitoring granulation drying.¹¹³ Bruker Optics, Foss and Thermo Fisher Scientific are adopting the OPC ADI Specification in their product development for PAT applications.²⁸

Kaiser Optical Systems, Inc. specializes in providing Raman spectrometers and applied holographic technology. Their RamanRXN Systems series of spectrometers are suitable for laboratory discovery, product and process development to chemical plant scenarios.¹¹⁴ There is also a range of Raman probe heads and optics that can be used in “plug-and-play” fashion with the various spectrometer analyzers. This includes (i) the PhATprobe head for solid sampling in pharmaceutical secondary unit operations such as drying, milling, granulation, blending, tabletting, coating, and QA/QC, (ii) the Airhead™ probe for in-line monitoring of gases such as nitrogen, fluorine, SO_x, CO_x and hydrocarbons, (iii) customizable Pilot™ Raman probe in combination with the RamanRXN3 analyzer can be deployed within an explosion-proof chemical plant environment for in situ process monitoring and control. The RamanRXN3 can also be multiplexed up to 4 sampling points. Kaiser Optical Systems also provides software that facilitates PAT and QbD implementation, and is incorporating the OPC ADI Specification for future products.²⁸ The HoloReact software can be used in-tandem with the aforesaid Raman spectrometers for real-time spectral visualisation and chemometrics analyses, and there is a Software Development Kit (SDK) for third party OEM deployment of RamanRXN analyzers. Kaiser Optical Systems has a partnership with Mettler Toledo in the iC Raman software development for integration with Mettler Toledo analytical technologies (see below). A good survey of pharmaceuticalPAT applications using Kaiser Raman technologies can be found in a recent publication.¹¹⁵

Malvern Instruments supplies laser diffraction and dynamic light scattering analyzers to the pharmaceutical industry for particle characterization.¹¹⁶ Its Insitec analyzers allow real-time in situ monitoring of particle size using laser diffraction in applications such as microsphere sizing control using spray drying,¹¹⁷ monitoring of suspensions during wet milling,¹¹⁸ and automation of milling.¹¹⁹ Insitec operates with particles in the size range 0.1 to 2000 microns and is designed to comply with GMP requirements. It is able to operate in hazardous areas through remote instrument control via interfaces such as SCADA. These systems offer an avenue to eliminate human exposure to dangerous substances and facilitate real time product release and continuous manufacturing. Another Malvern product series that is useful as a PAT tool is the SyNIRgi near-infrared chemical imaging (NIR-CI) system. The SyNIRgi is an advanced instrument that can capture more than 80,000 diffuse reflectance NIR microscopic spectra at different spatial locations in less than 2 min using 320 × 256 focal plane array (FPA) detectors. The NIR detector range covers either 1200–2400 nm (InSb FPA) or 950–1720 nm (InGaAs FPA); sample size can range from 3 × 2.5 mm (∼10 μm pixel magnification) as large as 40 × 32 mm (∼125 μm pixel magnification). The SyNIRgi software system follows GAMP guidelines and is 21 CFR Part 11 compliant, and Malvern also supplies the ISys data processing software that contains a suite of chemometrics tools for NIR image data analyses. This is very useful for revealing spatial composition variations at rapid speed. It has been applied heavily in Quality by Design,¹²⁰ to investigate the relationship between pharmaceutical tablet dissolution failure and its content distribution uniformity,¹²¹ and counterfeit recognition and sourcing.¹²² The Malvern Process Systems is also adopting the OPC ADI Specification.²⁸

Mettler Toledo is a supplier of analytical solutions for process scales ranging from laboratory to chemical plant.¹²³ Under its AutoChem division, which promotes the incorporation of product and service solutions for client processes, is a suite of instrumentation that covers parallel synthesis (MiniBlock^®), reaction calorimeters and automated laboratory reactors (EasyMax™, MultiMax™, RC1e™), and various in situ analytics such as ATR-FTIR (ReactIR™ product series), Focused Beam Reflectance Measurement (FBRM^®) and Particle Vision Measurement (PVM^®).¹²⁴ More recently, Mettler Toledo developed the iC and iControl software platforms to interface various analytical instrumentation and reactor systems so as to facilitate real-time integration of analytics, reactor system control, data visualization and analyses. The iC software includes iC Quant module, iC IR, iC FBRM, and the co-developed iC Raman with Kaiser Optical Systems. The iControl package covers iControl RC1e, iControl LabMax, iControl EasyMax. There is also the iSafety module that assists process safety assessment. This iC/iControl integration would enhance automation workflow, process reproducibility, and translation of innovative ideas into experimental design and scale-up, experimentation, data collection, process modeling and information generation. Mettler Toledo AutoChem announced the release of their iC PAT in 2010 that is based on the new OPC ADI Specification for their spectroscopy (ReactIR) and particle (FBRM) analyzers.²⁸

As noted above, chemometrics analyses are integral multivariate tools in the PAT framework. However, it is common to hear from the ground that such analyses can be rather abstract and difficult to apply. Interestingly, this perception was mentioned some years back by Lavine and Workman Jr., “Although the pharmaceutical industry has embraced the chemometric approach, few chemists in academia or industry actually take advantage of it. Chemometrics is considered to be too complex; the mathematics can be misinterpreted as esoteric and not relevant.”⁸¹ Fortunately, there are several companies that provide chemometrics software solutions that expedite the learning curve required (see Table 4). A number of these solutions are executed within the Matrix Laboratory (MATLAB^®) scientific programming environment from The MathWorks,¹²⁵ as MATLAB has been a development workhorse for chemometrics algorithms.¹²⁶ Applied Chemometrics, Inc. supplies toolboxes that contained MATLAB based chemometrics algorithms.¹²⁷ Its Chemometrics Toolbox includes over 40 functions such principal component analysis and regression (PCA and PCR), least squares codes (CLS, ILS, and PLS), and the Factor Analysis Toolbox offers methods for cross-validation (complete and binary) and various factor analysis (IKSFA, EFA, WFA, RAFA, and SIFA).

Another company that provides MATLAB based chemometrics packages is Eigenvector Research, Inc. (EVRI) which was founded in 1995. Its PLS Toolbox encapsulates the core set of chemometrics methods that can be expanded by other add-on solutions, such as the Multivariate Image Analysis (MIA) Toolbox and the Extended Multiplicative Scatter Correction (EMSC) Toolbox. The core PLS Toolbox contains a comprehensive suite of numerical tools that include standard signal preprocessing (smoothing, scaling, derivative, signal/scatter corrections, etc.), variable selection (purity based, genetic algorithms, etc.), pattern recognition and multi-way analyses (PCA, PARAFAC, MCR, Tucker models), classification (cluster and discriminate analysis), linear and non-linear regression (PLS, PCR, etc.), curve fitting, and graphical user interfaces (GUI) for data visualization and editing. Data is processed in the PLS Toolbox using the EVRI DataSet Data Standard Object (DSO) under the MATLAB environment. EVRI also provide stand-alone GUI version of PLS Toolbox called Solo that can be combined with their MIA Toolbox. There are also software modules called Solo Predictor and Model Exporter. The former allows connectivity between client and data management system viaTCP/IP, ActiveX, or Wait-for-File mode, and the latter exports developed chemometrics models and deploys them as generic XML format or other predictor formats usable in LabVIEW™, MATLAB and Symbion™ environments. PLS Toolbox algorithms can also be connected and deployed to Distributed Control Systems (DCS) via the MATLAB OPC Toolbox. Through the MATLAB compiler, codes written using PLS Toolbox can also be compiled into stand-alone executable files (EXE), dynamic linking libraries (DLL), or COM/Active X objects for integration with third party software.

Other non-MATLAB based commercial chemometrics packages are also available. The Thermo Fisher Scientific GRAMS software has been serving the spectroscopy community for over 20 years. Its GRAMS (.SPC) spectral format is now an industry standard which most software packages sold by commercial spectrometer companies would cater for. The latest GRAMS Suite 9.0 includes the GRAMS/AI™ a comprehensive spectral data processing package, Spectral ID qualitative spectral identification, Spectral DB new data management and GRAMS/3D for multi-dimensional visualization. It also contains the GRAMS IQ™ with IQ Predict™, with numerical techniques such as PCA, PLS, PCR, and discrimination analysis, for quantitative and qualitative chemometrics modelling and prediction.¹⁰⁹ Its PLSplus/IQ package allows the extension of GRAMS/AI for chemometric calibration modeling in both laboratory and manufacturing scenarios.

The company CAMO was established in 1984, and started providing the Unscrambler^® multivariate and chemometrics software in 1986. Presently, the Unscrambler had developed into a software suite of modules: the core Unscrambler, Unscrambler On-line Predictor (OLUP), Unscrambler Classifier, Unscrambler Optimizer, and Quali-Sense. Similar to the PLS Toolbox, the primary Unscrambler package includes essential chemometrics algorithms for standard data pretreatments (filtering, scaling, scatter corrections, etc.), design of experiment (F, FF, PB, CCD, BB, mixture, etc.), exploratory data analysis (PCA, MCR, ANOVA), regression (PLS, three way PLS-R, PCR, MLR), clustering and classification (K-means, SIMCA, PLS-DA).¹²⁸ CAMO software has been deployed in various industrial applications, such as analytical instrumentation, pharmaceutical, energy, chemicals and food and beverages. Moreover, the latest Unscrambler GenX family of products is 21 CFR Part 11 compliant and poised for PAT and QbD applications. CAMO also established business partnerships with Siemens, ABB and Symbion to integrate the Unscrambler in their respective PAT solutions through the OLUP (see below).¹²⁹

The company InfoMetrix was founded in 1978 and had since been developing various chemometrics solutions, e.g. AUTHUR which is written in FORTRAN, to support analytical instrument partners such as Agilent, Waters, Perkin Elmer and others.¹³⁰ Their current flagship chemometrics software product, Pirouette, was first released in 1990. The latest Pirouette 4.0 version contains algorithms for standard spectral data pretreatment (smoothing, baseline and signal corrections, etc.), multivariate analyses (PCA, HCA, K-nearest neighbor, CLS, PCR, PLS, PLS-DA, MCR, ALS, etc.) and graphics for data interactions. In the 1990s InfoMetrix worked with the FDA and Pfizer towards making Pirouette software 21 CFR Part 11 compliant. The companion InStep Visual Basic client software was also concurrently developed during that time to handle real-time data, automate routine prediction functions and facilitate hierarchical prediction strategies. An overhaul of Pirouette was undertaken in 2002 to allow cross platform compatibility via ActiveX COM so that third party software developer can access Pirouette algorithms using COM interfacing programming languages such as Visual Basic (VB), VB Scripting, C++, C#, MATLAB and LabVIEW.

The chemometrics group at Umeå University founded Umetri AB in 1987. The company name later had its name changed to Umetrics^®AB in 1999. Since 2006, Umetrics is owned by MKS Instruments, Inc.¹³¹ Umetrics is well-known for their SIMCA family of multivariate chemometrics algorithms, and the MODDE (acronym for MODdelling and DEsign) software package for design of experiment (DoE) and optimization. The latest versions of SIMCA 12.0 and MODDE 9.0 are 21 CFR Part 11 compliant. The SIMCA software contains the SIMCA-P core multivariate data analysis package, the SIMCA-P+ for batch analysis, SIMCA-Batch On-line (SBOL) for on-line batch monitoring and prediction (BSPC), SIMCA-4000 on-line multivariate statistical process control (MSPC) for continuous processes. The core SIMCA-P includes standard data preprocessing (signal corrections, filtering, wavelet transform, etc.), multivariate modelling (PCA, PLS, OPLS, O2PLS, with cluster and discriminant analyses), and also MSPC tools. The MODDE contains an assortment of DoE generation tools (FF, CCF, CCC, PB, BB, D-Optimal, etc.), a multidimensional Simplex method optimizer, and GUI with in-build wizards to provide guidance for analysis steps and plotting contour and prediction plots. Chemometrics models developed in SIMCA-P and SIMCA-P+ can be deployed as DLL for third party OEM software integration through predictor engines SIMCA-QP (continuous data) and SIMCA-QP+ (batch data) respectively. These are called through a COM interface by C++, C#, VB, MATLAB, Excel and Python, or direct calls via a fast low level C-interface using C++ or VB. The SIMCA-QM and SIMCA-QM+ modules are similar to their QP and QP+ counterparts, and possess added capability for automated multivariate model building. MODDE 9.0 is likewise interfaced viaCOM using M-Link. Furthermore, Umetrics provides a PAT Suite that enables deployment of different multivariate tools according to the four levels of PAT defined by Svante Wold.¹³² As early as 1991, Umetrics (then Umetri AB) collaborated with ABB Automation to develop an on-line version of Umetrics software.¹³¹ Presently, Umetrics have partnership alliances with GE and Siemens to incorporate Umetrics software in their integrated enterprise PAT solutions (see below).¹³¹

The PAT initiative created innovative developments across companies with core businesses in the area of process systems integration and automation. Four examples noted here are from the technologies of ABB, GE, National Instruments, and Siemens.

Under the ABB Analytical and Advanced Solutions Business is the ABB Industrial^IT ™ technology that includes several instrument and software platforms important for PAT implementations. The Analyze^IT suite comprises of industrial process analyzers, such as spectrometers (e.g.FTIR, FT-NIR, MS), UV/IR photometers, gas chromatographs and analyzers, physical property analyzers (e.g. for vapor pressure, H₂S, O₂),¹³³ are connected using OPC communications through network components and integrated analyzer systems to other ABB Industrial^IT applications or third-party OPC enabled platforms. These analyzers can also be linked to legacy DCS systems via MODBUS or ASCII communications. In 2004, ABB launched their latest automation platform known as Industrial^IT Extended Automation System 800xA. It has an open architecture that is built on ABB patented Aspect Object™ technology, which “relates plant data (the aspects) to specific plant assets (the objects).”¹³⁴ As such, System 800xA is able to integrate with all fieldbus standards and easily connect with plant devices and systems through OPC, Ethernet, TCP/IP, etc. Furthermore, System 800xA caters for flexible and scalable system configurations for small, medium and large process applications through its plug-and-produce architecture. The 800xA system and data security are compliant to FDA 21 CFR Part 11 requirements. For pharmaceuticalPAT applications, the System 800xA platform software can be coupled with the ABB Industrial^IT eXtended Process Analytical Technology (xPAT) platform for integration of process analytics, data historian, control systems and enterprise systems such as SAP, MES, LIMS, etc.PAT methods deployed can also be integrated with manufacturing batch through xPAT, with System 800xA hosting the batch control system and analyzer records stored in accordance with ANSI/ISA-S.88 guidelines against batch ID.¹³⁵ Multivariate analysis software from Umetrics (SIMCA-Q) and Thermo Fisher Scientific GRAMS can be deployed from within xPAT. The CAMO online version of the Unscrambler (OLUP) is integrated with the ABB FTSW1000 software used with the ABB Bomem FTIR and FT-NIR instruments for process monitoring. Further integration development of OLUP is underway with the xPAT system.¹³⁶ In a recent ABB conference, it was reported that various aforesaid ABB technologies were suitably combined for PAT applications such as reaction monitoring and control, crystallization, drying and blending, and extruder processes.¹³⁷ABB is also active in the OPC ADI working group with its staff members contributing substantially in both the OPC UA and ADI Specifications.^29,31

One of the key business focuses of GE Intelligent Platforms (previously GE Fanuc) is providing automation systems as industrial solutions.¹³⁸ Besides supplying products such as PLC, PAC systems, distributed I/O and operator interfaces, GE offers the scalable and integrated Proficy^® Process Systems for process automation and control. At the heart of this solution is the Proficy Process Software that comprises of a suite of different modules such as Proficy Historian, Proficy HMI/SCADA (iFIX and CIMPLICITY), Proficy Batch Analysis, Proficy Real-Time Information Portal, and others.^139,140 With Proficy, an “open and layered product development strategy that is based on industry standard data architecture models” is provided.¹⁴⁰ Through this, open technology interfaces such as Northern Dynamic OPCDA 3.0 C++ Toolkit and Microsoft COM, allows the Proficy historian, HMI/SCADA system, controls and I/O can be connected to third party software and enterprise systems. In 2006, GE Fanuc and Symbion™ Systems announced an agreement to develop the Proficy RX module based on Symbion technology, which will serve as a “comprehensive PAT solution” that couples with other GE process automation technologies.¹⁴¹ The aim of Proficy RX is for engendering process understanding through delivering (i) standardized control and networking of multiple instruments, (ii) preprocessing of captured analytical data, (iii) process data archiving and repository in a SQL server database, (iv) methods development platform, (v) interfacing multivariate analysis and chemometrics packages such as GRAMS™, PLS Plus/IQ, MATLAB, EVRI PLS Toolbox, Unscrambler OLUP, Pirouette, Umetrics SIMCA-P/QP, and (vi) compliance with FDA 21 CFR Part 11 through the Security Guard™ architecture.^141–143 For rapid method developments, Proficy RX Script Assistants pop-up windows and pull-down menus provide GUI access to all standard operational commands and mathematical functions, allowing the sequencing of links and controls of instruments, I/O devices, third party programs into operation methods. Alternatively, method configuration can be achieved through the Symbion language commands that form a superset of the Tcl command set.¹⁴⁴ The Tcl extension architecture enables the embedding of other scripting software packages to be incorporated into Symbion, for example Visual Basic, C, MATLAB, etc. Together with the Security Guard features of Symbion, these embedded codes are secured and traceable even though the executing program resides outside this shell of security. Currently, the Proficy RX contains drivers for spectroscopic instruments such as ABB Bomem FTIR, Bruker FTIR/NIR, Foss NIR, Axsun Technologies NIR, Kaiser RamanRXN spectrometers and UV-Vis/NIR spectrometers from Carl Zeiss.

National Instruments (NI) was founded in 1976 with the intention of supplying IEEE software and hardware. Since 1986, NI introduced LabVIEW™, a graphical programming development environment, for scalable test, measurement and control applications. LabVIEW contains modules, toolkits, and hundreds of graphical functionalities ranging from data acquisition, signal processing and conditioning, mathematical functions, control algorithms, and many others. Both NI hardware architecture such as PXI/CompactPCI, CompactRIO, Compact FieldPoint, Industrial PC, Wireless Sensor Networks, HMI, etc., and LabVIEW come under the Programmable Automation Controller (PAC) architecture.¹⁴⁵ This PAC architecture is an acronym coined by Automation Research Corporation (ARC), which seeks to describe a “new generation of industrial controllers that combine the functionality of Programmable Logic Controllers (PLCs) and PC-based control.”¹⁴⁵ Integrated enterprise process monitoring and control systems such as HMI/SCADA can be built by combining suitable NIPAC hardware and LabVIEW modules/toolkits such as Datalogging and Supervisory Control (DSC), Real-Time, PID, Database Connectivity, etc. A good example is an automated system for monitoring of chemical solution concentration for semiconductor manufacturing using the integration of spectrophotometer, liquid pump and FieldPoint modules through a LabVIEW program.¹⁴⁶ The combination of LabVIEW and DSC module allows creation of 21 CFR Part 11 compliant applications. The DSC writes/records to SQL-compliant historical database that is password protection enabled and prevents records from alteration.¹⁴⁷ An alliance partner of NI, Coleman Technologies, offers a LabVIEW toolkit named Data Integrity Toolkit that contains functionalities such as user access management, data integrity validation, RSA encryption, and validated audit trails.¹⁴⁸ Application software developed with LabVIEW can communicate via the OPC client/server COM interface to access other vendor OPC enabled applications, for example, the DSC module can act as an OPC client to access remote OPC servers.¹⁴⁹ In addition, hardware connectivity (to GPIB, USB, serial, Ethernet, PXI and VXI instruments) and data acquisition can be easily achieved through the NI LabWindows™/CVI, which is an ANSI C development environment. LabWindows/CVI also provides instrument control, signal processing and numerical analysis, SQL database connectivity, development of measurement and automation graphical user interfaces, and supports latest open technologies such as multithreading, .NET XML Web services and Internet-enabled applications.¹⁵⁰ Recently, the latest LabVIEW 2009 version included the MathScript RT module for deployment of custom MATLAB script files as a new hybrid approach to reap the benefits of both script-based numerical codes and graphical programming.

As with the aforementioned three companies, Siemens provides industrial process automation solutions as one of its major business areas. Siemens' industrial automation systems SIMATIC series provides a portfolio of products ranging from controllers, HMI, sensors, industrial PCs etc., and facilitates standard communication platforms for easy device integration for manufacturing processes. Connection with SCADA system can be achieved through the SIMATIC WinCC software package. WinCC can also be configured as a SCADA client web server for connection with different clients, remote OPC servers, and a central data archiving server viaLAN, PROFIBUS, and PROFINET. The Siemens flagship PAT product, SIPAT™, started development in 2006, and the first version was introduced to the market in March 2007. Since the beginning, Siemens worked together with pharmaceutical manufacturers in the SIPAT development. New versions are being released on a yearly basis, as driven by customer and market needs.¹⁵¹SIPAT version 3.0 was released in early 2010. Taking the cue from the PAT tools delineated in the FDA Guideline 2004,¹SIPAT seeks to provide a PAT system architecture for integrating PAT tools. It can be deployed under two scenarios: (i) SIPATBase Station in which the focus is placed on control of specific unit operations, and (ii) SIPAT High Level that includes the total manufacturing line batch quality for real-time-release of products.¹⁵² As such, SIPAT is designed for both integrating into industrial IT systems (e.g., MES, LIMS, EBR) and process automation systems (e.g., DCS, SCADA) to achieve product status transparency throughout the entire manufacturing chain, and interfacing varied process equipment, instrumentation and software platforms/packages for ease of configuration, model building, operations execution, data archiving, analyses and retrieval. SIPAT is compliant with the FDA 21 CFR Part 11 requirements. The SIPAT features includes (i) a common instrument interface for control of different process analyzers, (ii) multivariate/chemometrics data analyses using an offline Model builder and online data pre-processing and application of developed multivariate models, (iii) open technology platforms interfaces, such as OPC, COM, DLL, enable third party software for control of analyzers, process control systems, and generating predictions from PAT models, (iv) configurable objects for defining methods, models, user profiles, access rights, etc., that are under development testing and validated operation, (v) an open database structure, (vi) an integrated SIPAT graphical user interface for design of experiments, PAT model development and data retrieval from historic and online data. Various commercial multivariate data analysis chemometrics packages are integrated and embedded in SIPAT, such as MATLAB version R2007b, CAMO Unscrambler, Umetrics SIMCA-P/P+, SIMCA-QP/QP+ and MODDE. To date, SIPAT is being used in development lab applications (validated/non-validated environments) as well as in manufacturing facilities. Siemens has assisted in the implementation of SIPAT in small molecule and bio-manufacturing projects in companies such as GSK, Ferring, NVI, Merck and legacy Wyeth (now Pfizer).^151,153 Siemens also contributed to the OPC ADI endeavors in representing “the needs of process gas chromatographic applications in the OPC communication standard.”²⁸

Pharmaceutical industry

There are exciting PAT developments in the pharmaceutical industry. In 2005, GlaxoSmithKline (GSK) invested ca. S$116 million to setup a new R&D pilot plant at Singapore.¹⁵⁴ This pilot plant will be integral for engendering manufacturing innovation at the Global Manufacturing Supply (GMS) site in Jurong, Singapore. It is one of GSK's latest pilot plants for process development, scale-up and optimization of new chemical entities (NCE) that come out of its R&D pipeline. PAT instrumentation is implemented in this R&D pilot plant for facilitating real-time monitoring and process automation in the transfer of technology from laboratory to plant manufacturing.¹⁵⁴

It was reported that Merck (West Point, Pa.) had an NCE granted QbD real-time release status by FDA in 2006, which uses NIR PAT implementation for monitoring attributes such as potency, blending characteristics and tablet compression and allows for process prediction and control.¹⁵⁵ Besides adopting PAT for process monitoring, understanding and control, it is also used in part to save water and minimize environmental damage at a North Carolina facility.¹⁵⁶ Successful applications of Raman spectroscopic techniques by scientific teams in pharmaceutical industries are recently reported, e.g., monitoring content uniformity of a dry powder inhaler (Schering-Plough Research Institute),¹⁵⁷ quantitative transmission Raman analysis of pharmaceutical solids (AstraZeneca R&D),¹⁵⁸ and isolation of desired API polymorph (Merck & Co., Inc.).¹⁵⁹

In 2007, Novartis reported the use of OPC technology for their project at Novartis Singapore Pharmaceutical Manufacturing plant site.¹⁶⁰ Some important guiding principles for this project are: (i) manufacturer (vendor) independent concept design on the basis of use cases (for batch control, reporting and machine synchronization), (ii) compliance to GMP and 21 CFR Part 11 requirements, (iii) a central data storage historian for all shop floor data types, and (iv) mandatory use of OPC technology for mapping of communications and interface functions. The entire integration through OPC combined ten process unit (PU) classes from 25 manufacturers with ten different PLC/HMI, and batch recipe systems, and allowed remote batch control from MES level to PU. The same reference model was utilized for the OPCDA and A&E specifications to map functions, data structures and metadata functional description.¹⁶⁰ The number of DA and A&E tags ranges in the order of hundreds.

Publications, conferences and internet

With the flurry of diverse developments and activities happening under the banner of PAT, a fairly substantial amount of PAT-related information can now be found in regulatory guidance documents, the scientific literature, internet websites, company white papers and brochures, conferences and symposia. The references in this review provide a good sample. Table 5 tabulates a list of information sources, including specialized conferences, workshops and internet websites. With the rise of social networking websites, informal PAT discussion groups also can be found in virtual communities such as LinkedIn.

Table 5 Scientific Literature, Conferences, Symposia and Internet Resources related to PAT

Organizer/Publisher/Society/Field	Journal/Technical Magazines/Conference/Website
Journals
American Chemical Society (ACS)	• Analytical Chemistry
	• Biotechnology Progress
	• Crystal Growth & Design
	• Industrial & Engineering Chemistry Research
	• Organic Process Research & Development
American Association of Pharmaceutical Scientists/Springer	• AAPS PharmSciTech
American Pharmaceutical Online	• Journal of Process Analytical Technology (past issues)
Elsevier	• Advanced Drug Delivery Reviews
	• Analytica Chimica Acta
	• Biotechnology Advances
	• Chemometrics and Intelligent Laboratory Systems
	• Computers & Chemical Engineering
	• European Journal of Pharmaceutics and Biopharmaceutics
	• International Journal of Pharmaceutics
	• Journal of Biotechnology
	• Journal of Pharmaceutical and Biomedical Analysis
	• Talanta
	• Trends in Analytical Chemistry
	• Vibrational Spectroscopy
International Forum for Process Analytical Chemistry (IFPAC)	• Journal of Process Analytical Chemistry
International Society of Automation (ISA)	• ISA Transactions
	• InTech
International Society for Pharmaceutical Engineering (ISPE)	• Journal of Pharmaceutical Innovation
	• Pharmaceutical Engineering
IM Publications	• Journal of Near Infrared Spectroscopy
	• NIR News
Royal Society of Chemistry (RSC)	• Analyst
	• Analytical Methods
Society of Applied Spectroscopy	• Applied Spectroscopy
Wiley	• AIChE Journal
	• Biotechnology & Bioengineering
	• Biotechnology Progress
	• Journal of Chemometrics
	• Journal of Pharmaceutical Sciences
	• Journal of Raman Spectroscopy
	• Spectroscopy Europe
Conferences/Symposia
Analytical Chemistry, Spectroscopy and Chemometrics	• Chemometrics in Analytical Chemistry (CAC)
	• International Conference on Advanced Vibrational Spectroscopy (ICAVS)
	• International Conference on Near Infrared Spectroscopy (ICNIRS)
	• International Conference on Raman Spectroscopy (ICORS)
	• Pittsburgh Conference (Pittcon)
	• The Federation of Analytical Chemistry and Spectroscopy Societies (FACSS)
Chemical and Pharmaceutical Engineering	• AIChE Annual Meeting
	• European Symposium on Computer-Aided Process Engineering
	• ISPE Conference
	• Interfex
Process Analytical Technology	• International Forum for Process Analytical Chemistry (IFPAC)
	• Pharmaceutical Process Analytics Roundtable (PPAR)
	• CPAC Summer Institute
Internet Resources
American Association of Pharmaceutical Scientists	http://www.aapspharmaceutica.com
American Pharmaceutical Review	http://americanpharmaceuticalreview.com
Bioresearch Online	http://www.bioresearchonline.com
International Forum for Process Analytical Chemistry	http://www.ifpac.com
LinkedIN	http://www.linkedin.com (PATgroup)
PharmaAsia	http://www.pharmaasia.com
Pharmaceutical Online	http://www.pharmaceuticalonline.com
Pharmaceutical Technology	http://electronic.pharmtech.com
PharmaManufacturing.com	http://community.pharmamanufacturing.com
Process Analytical Technology	http://www.processanalyticaltechnology.com
R&D Directions	http://www.pharmalive.com
SpectroscopyNOW	http://www.spectroscopynow.com

---

There exist a regular stream of articles, reviews and commentaries that bear the term “PAT” in their title, abstract or keywords. In 2005, a series of articles in a special section of the journal Organic Process Research & Development (OPR&D) demonstrated the potential of PAT framework in various operations within pharmaceutical process development and manufacturing. These articles cover reaction monitoring^161–164crystallization,^161,165 granulation,¹⁶¹calorimetry and safety studies,¹⁶³ and in situ cleaning validation.¹⁶⁶ Since then, OPR&D had served as a regular platform for reporting new PAT technologies and applications. Other related works published under the American Chemical Society can be found in the journals Analytical Chemistry and Industrial & Engineering Chemistry Research. Two notable reviews closely related to PAT are published in Analytical Chemistry with the titles Process Analytical Chemistry^35,82,91 and Chemometrics,^81,167,168 and the latest developments in process chemometrics are found under the former review title.^35,82,91 Other journals that feature topical articles associated with PAT are primarily in the domains of analytical chemistry, applied spectroscopy, chemometrics, chemical engineering, biotechnology, pharmaceutics, pharmaceutical engineering, and automation (see Table 5).

In 2005, a compilation on PAT was published which focuses on spectroscopic analytical tools and their implementations within the chemical and pharmaceutical industries.¹⁶⁹ This milestone book surveys PAT from its background and history,³⁴ implementation,¹⁷⁰ use of varied spectroscopic methods, such as UV-Vis,⁴² infrared,⁴⁴ Raman,⁵¹ NIR,^64,171–173 and process chemometrics,⁷⁹ for applications in on-line analyses,¹⁷⁴chemical imaging,⁶⁴ process units in pharmaceutical^172,174 and chemical¹⁷³ manufacturing. The second edition of this book was recently published in 2010 by Wiley-Blackwell with all chapters updated and additional topics on process sampling, acoustic, fluorescence and NMR spectroscopies.¹³⁶

Since PAT interfaces several domains of knowledge, it is worth introducing some reference works in related disciplines such as analytical chemistry, spectroscopy, process analytical instrumentation, chemometrics and hyperspectral chemical imaging. Several handbooks are useful for background and technical details in analytical chemistry and spectroscopy.^175–179 The 15-volume Encyclopedia of Analytical Chemistry contains over 600 articles that cover every aspect of chemical analyses techniques from academia to industry.¹⁷⁵ There is a section on Process Instrumental Methods in its ninth volume with topics on sampling methods, chemometrics, chromatography, titration, flow and sequential injection analysis techniques, mass spectrometry and spectroscopic process analysis using UV-Vis, mid- and near-infrared, Raman, NMR and MRI.^41,46,180 Specialized articles for pharmaceuticals and drugs analysis expound on techniques such as gas and liquid chromatography, mass spectrometry, vibrational spectroscopy (including vibrational optical activity), and NMR.¹⁸¹

Among the optical spectroscopic methods, Raman and infrared (mid- and near-) are the most versatile for PAT applications as they provide both qualitative and quantitative information to chemical functional groups and signature spectrum for species identification. The Handbook of Vibrational Spectroscopy is a comprehensive 5-volume reference that detailed the fundamentals and applications of infrared and Raman spectroscopies classified under theory and instrumentation (volume 1), sampling techniques (volume 2), sample characterisation and spectral data processing (volume 3), applications (volumes 4 and 5).¹⁷⁶ There are two articles on process monitoring using mid- and near-infrared⁴⁵ and Raman⁵⁰ spectroscopies, and a section of articles on pharmaceutical applications in the fifth volume. Two handbooks that are dedicated to Raman and NIR techniques are the Handbook of Raman Spectroscopy¹⁷⁷ and the Handbook of Near-Infrared Analysis.¹⁷⁹ Both handbooks cover the basic spectroscopic principles, instrumentation, calibration issues, multivariate data analysis and various applications including pharmaceuticals.

There are two books dedicated to pharmaceutical applications using NIR and Raman spectroscopies.^182,183 The Pharmaceutical and Medical Applications of Near-infrared Spectroscopy addresses NIR instrumentation considerations, applications in blending, granulation, drying and coating, and validation issues.¹⁸² The Pharmaceutical Applications of Raman Spectroscopy covers different Raman techniques such as confocal microspectroscopy, imaging, surface enhanced resonance,¹⁸³ and a chapter on real-time monitoring in the pharmaceutical industry.¹¹⁵ Three books in the area of hyperspectral imaging provide a good selection of articles on fundamental theory, instrumentation, multivariate data analysis and applications. They are Spectrochemical Analysis using Infrared Multichannel Detectors,⁵⁹Techniques and Applications of Hyperspectral Image Analysis,⁶² and Infrared and Raman Spectroscopic Imaging.⁶⁶ Chapters directly related to PAT are those on chemometrics analyses,^72,73 industrial and pharmaceutical applications using mid- and near-infrared imaging^65,67–69 and Raman imaging.⁷⁰

Several books address process analytical instrumentation. They are Process Analyzer Technology,³⁶Handbook of Spectroscopy,¹⁷⁸Spectroscopy in Process Analysis,¹⁸⁴Analytical Instrumentation,¹⁸⁵ and Process Analytical Chemistry.¹⁸⁶ The first book is a classic book in the field that covers the design, installation and maintenance of on-line analyzer instrumentation in production operations. The second to fourth books include chapters on spectroscopic techniques like UV-Vis, mid- and near-infrared, Raman, NMR and MRI, fluorescence, chemiluminescence, with other topics on sampling, atomic and mass spectrometry, elemental analysis, chemometrics and data analysis. The fourth book further includes sections on process analyzer system implementation considerations (application justifications, IT interfacing, sample conditioning, calibration, SPC/SQC, validation, operations and maintenance), physical properties analyzers (instruments for measuring boiling, flash, cloud, freeze, moisture/dew and pour points, vapour pressure, refractive index, thermal conductivity, viscosity, different gases, etc.), electrochemical analyzers (pH, conductivity, redox, trace oxygen), and compositional analyzers (chromatographic, distillation % off, mass spectrometric).¹⁸⁵ The last book contains material on sampling systems, electrochemical and physical property analyzers, process chromatography, flow injection analysis, molecular spectroscopy (mass and optical), environmental monitoring technology, non-invasive techniques (ultrasound, acoustic emission, microwave, X-ray and infrared methods),¹⁸⁶ and a chapter on process chemometrics.¹⁸⁷

There are two categories of books published under the scope of chemometrics. The first are general texts that delineate theoretical background and key numerical methods across various aspects of chemometrics.^188–190 The second group are inclined towards specific group of methods or applications.^{62,97,191–193} In Chemometrics: Data Analysis for the Laboratory and Chemical Plant are five sections on experimental design, signal processing, pattern recognition, calibration, and evolutionary signals, together with appendices introducing the linear algebra and statistics, algorithms, and implementation of chemometrics in Microsoft Excel and MATLAB.¹⁸⁸ The Practical Guide to Chemometrics introduces concepts on multivariate statistics, principal component analysis (PCA) and multivariate calibration before delving into topics such as robust methods, nonlinear multivariate model estimation, multivariate curve resolution (MCR), classification and pattern recognition, and three-way calibration methods.¹⁸⁹ The book Applied Chemometrics for Scientists covers similar fundamental multivariate and chemometrics contents, and also include segments on statistics, multivariate image analysis, biological, medical and food applications, and two chapters discussing process chemometrics topics such as the PAT initiative, equilibria, reactions, spectroscopic process analytics, and the use of experimental design to improve processes.¹⁹⁰

The classic book Multivariate Calibration is a good resource for topics on statistical basis, data selection, pretreatment and linearisation, experimental design, outlier detection, various calibration methods, assessment, validation and choice of calibration method.¹⁹¹Factor Analysis in Chemistry delineates the historic background, philosophy, theory and applications of factor analysis methods for analytical measurements, including target factor analysis, evolutionary methods of window factor analysis and DECRA, multimode factor analysis, component analysis and partial least-squares regression. It covers applications such as NMR, chromatography, biomedical, food, environmental and fuel.¹⁹² Special focus was made in the book Chemometrics in Spectroscopy to highlight issues when applying chemometrics for spectroscopic analyses, e.g. linear algebra and statistics, experimental design, linearity in calibration, regression and good fit, spectral noise, use of derivatives, and limitations in analytical accuracy.⁹⁷Chemometrics for Pattern Recognition is a compilation of numerical approaches targeted for discrimination and classification. They comprise data preprocessing strategies, exploratory data analysis, different classifiers (one class, two class, and multiclass), optimization and validation, Bayesian methods and others.¹⁹³ For chemometrics analyses of hyperspectral images, the book Techniques and Applications of Hyperspectral Image Analysis is suited for both newcomers and advanced readers, with chapters introducing basic concepts, instrumentation, and data types in hyperspectral and multivariate image analysis (MIA), data conditioning, and various numerical methods for classification, factor analysis, regression and calibration, component curve resolution. Chapters are dedicated to specific applications in pharmaceutical, agricultural, and images from MRI, Positron Emission Tomography (PET) and NIR.⁶²

Elsevier recently published a reference work entitled Comprehensive Chemometrics: Chemical and Biochemical Data Analysis,¹⁹⁴ with articles that fall under the following areas: statistics, experimental design, optimisation (volume 1), data processing, linear soft-modelling, unsupervised data mining (volume 2), linear regression modeling, non-linear regression, classification, feature selection, multivariate robust techniques (volume 3) and applications (volume 4). There are a number of articles related to process chemometrics and PAT in this work.^195,196

From the above tracing of different historical perspectives related to the PAT endeavor, it is evident these parallel developments are not running independently, but intertwined synergistically in an organic fashion that contributes to the overall growth of PAT.

PAT applications

Reports of PAT applications can be found in both journals and also online websites (see Table 5 and section on Publications, Conferences and Internet). The applications mentioned in this segment are primarily from the period 2004–2009. Some legacy works that predates the terminology PAT are also mentioned due to their relevance. With the exception of review articles, the classification of reports on PAT applications is not always clear cut as often they interface several domains. The typical areas are process development and scale-up, process operation units (reaction, crystallization, bio-mediated chemical transformations, milling, granulation, etc.), process analytics (optical spectroscopies, physical properties, etc.), multivariate data analyses and chemometrics methods, process control and diagnosis. As it is not possible to include all publications herein, the intention is to introduce representative ones, especially those in review articles and book chapters.

Primary manufacturing

Primary manufacturing in pharmaceutical manufacturing typically consists of chemical synthesis, including catalysis and bio-mediated ones, batch processes, cell cultures and crystallization. A good survey of Raman applications in PAT was provided with specific examples in batch process monitoring, reaction analysis and process control.¹¹⁵ Important applications of Raman in primary manufacturing are catalysis,¹⁹⁷ bio-mediated transformations,^50,51 chemical reaction and synthesis.^50,51,198Infrared spectroscopy can be similarly applied in primary manufacturing processes using different sampling techniques in both mid- and near-infrared wavelength regions.^44,45 Some infrared PAT applications are in raw material analysis,¹⁹⁹ reaction monitoring (for synthesis,^174,198 catalysis,^43,92,200 product selectivity,¹⁶² safety studies,^163,198 reaction endpoint determination^162,164,198), in situ cleaning validation,¹⁶⁶ and bioprocesses.^201–203 It was noted that more than 90% of chemical reactions/transformations can be studied using Raman and mid-infrared spectroscopies.^115,204UV spectrometry was also used for monitoring reaction conversion.^92,205 Other novel process analytics used for monitoring biological cell cultures are dielectric spectroscopy, electric nose (for detecting gaseous emissions), and fluorescence spectroscopy.²⁰³

The PAT framework delineated in the FDA guidance specifically included microbiological analysis for engendering process understanding through measuring and controlling key biological attributes.¹ Vibrational spectroscopic techniques are popular in situPAT tools for bioprocesses. NIR spectroscopy has been used (i) in conjunction with chemometrics analyses for an industrial multi-stage API bioprocessing,²⁰¹ (ii) monitoring glycerol-boosted anaerobic digestion processes,²⁰² (iii) combined with image analysis, angle measure technique (AMT) and chemometrics modeling for heterogeneous anaerobic digestion processes in large scale biogas plants,²⁰⁶ (iv) on-line monitoring and control of mammalian cell cultures.²⁰³ Other bio-related applications of NIR can be found in a recent review.¹⁹⁹ The application of mid-infrared spectroscopy to bioprocesses is complicated by the strong absorption of water, which can be partially overcome by the use of the ATR-FTIR technique. This has been applied for (i) quantitative determination of cephalosporin C in fermentation broths,²⁰⁷ (ii) combined with PLS based calibration/prediction models for Escherichia colifermentation,²⁰⁸ (iii) real-time monitoring of important fungal fermentationanalytes for pilot scale operations (glucose, fructose, glutamate, proline and phosphate).²⁰⁹ As water has a weak Raman scatter, Raman spectroscopy is advantageous for bioreactor applications,⁵¹e.g. monitoring fermentation reactions.^50,51 The use of NIR excitation laser source can potentially minimized fluorescence effects in the Raman spectra,⁵⁰ and with a non-contact probe optics,^46,49,50in situ measurement of the reaction broth can be achieved through a viewing port of a suitable window without causing contamination issues. Other analytical measurements that were demonstrated to be efficacious for PAT applications in bioprocesses include visible photospectroscopy,²¹⁰ fluorescence microarrays,^210,211 and biomass related quantities of oxygen uptake rate (OUR), carbon dioxide production rate (CPR) and pH.²¹²

A FDA team published an overview on PAT applications to crystallization processes in 2004 that described the critical variables and thermodynamics of crystallization, available process sensors for real-time monitoring supersaturation, particle size and shape and polymorphic forms (NIR, ATR-FTIR, Raman and X-ray diffraction), chemometrics modeling, experimental design and multivariate analyses, and PAT case studies (progesterone, MK-A, trovafloxacin, glutamic acid and glycine).²¹³ A review paper on advances in crystallization control from an industrial perspective surveyed the use of different process analytics (ATR-FTIR, Raman, FBRM, and PVM) combined with feedback control strategies to regulate particle attributes (size and morphology) in crystallization processes in relation to the move from quality by testing (QbT) to quality by design (QbD) according to the PAT initiative.²¹⁴ There are other articles describing the use of PAT tools for monitoring crystallization phenomena, they include spectroscopic techniques such as Raman,^{51,108,115,165,215,216}ATR-FTIR,^{165,174,217,218} NIR,^{108,174,219,220} scalar measurements with FBRM,^{108,165,220,221} and turbidity,^220,222 and chemometrics methods such as PCA, PCR, and PLS.^{108,115,216,219,220} The external control of crystallization process with PAT tools and mathematical models using population density and moment equations was reported.²²³

Secondary manufacturing

Typical secondary manufacturing processes include purification, milling, granulation, extrusion, drying, powder blending, compression, capsulation, coating, tabletting, packaging and product quality assurance testing. Different analytical measurements can be used for PAT implementation in secondary manufacturing, and many of these are combined with multivariate analyses. A range of analytical methods for the purification of therapeutic monoclonal antibodies, which include electrophoresis, HPLC and immunoassays, was recently reviewed.²²⁴ It was suggested in the same review that FTIR and ultrasound have the greatest potential for in-line process monitoring and control. Raman is not suitable because its excitation laser may induce sample changes to the antibodies.²²⁴ NIR spectroscopy has proven to be a versatile PAT tool for secondary manufacturing operations. It can be used as process analytics for the control of granulation, drying, coating, blend uniformity analyses, packaging.^{174,179,182,199,225}Raman spectroscopy is readily deployed for monitoring processes involving blending, extrusion, granulation, coating, tableting, lyophilization, and monitoring compositions of polymorphs, films, slurries, emulsions and suspensions.^50,51,115 An in-line Raman PAT implementation for understanding the lyophilization or freeze-drying process of aqueous D-mannitol solution enabled the detection of solid mannitol polymorphs, i.e. α, β, δ forms or hemihydrate, during freeze-drying.²²⁶ The in-line Raman process data complemented with at-line NIR and powder XRD measurements, PCA and MCR multivariate analyses and factorial experimental design revealed the relationship between formulation factors and process variables.²²⁶ Furthermore, the use of hyperspectral chemical imaging using NIR, ATR-FTIR and Raman microspectroscopy and imaging instrumentation is also on the rise, with applications in powder and tablet preparation, blend homogeneity and measuring content uniformity using chemical distribution maps of solid dosage forms, detection of polymorphs, and monitoring the release/dissolution of polymer/drug formulations into water.^{64,66,183,227}UV-Vis spectroscopy is used for dissolution testing,⁴¹ monitoring of a processing unit during cleaning in place operation,⁴¹ and quantifying constituent concentrations in powder mixture blending.²²⁵

Process development

Processes from the laboratory bench undergo a developmental phase to scale up to pilot plant testing before eventually implemented in their actual primary and secondary manufacturing facilities. In the area of process development studies, the use of PAT is considered as the current industrial practice.²¹⁶ A perspective paper in 2006 addressed the question of making pharmaceutical process development high tech by looking into the frameworks and concepts surrounding 21st century pharmaceuticalcGMP, PAT, QbD, intrinsic process knowledge, adoption of ANSI/ISA-S.88 standard, and different levels of modeling (molecular, process and systems).²²⁸ The recommendation was for using the combination of S.88 platform, analytics, software and hardware standardization for integrating the three levels of modeling into high-throughput workflows that allows automated experimentation in process development.²²⁸ An overview of design of experiment (DoE) applications in pharmaceutical process R&D was provided in a book chapter.²²⁹ It introduces the basic concepts in DoE with discussion on its advantages and limitations, and provides examples in organic synthesis, e.g.hydrogenation, hydrolysis, Suzuki coupling, Brederick reaction, etc.²²⁹ Ascertaining the kinetic rates and thermochemistry of reactions is an important element of process development and scale-up. In situ vibrational spectroscopic techniques (MIR, NIR and Raman) have by far been the most successful in unraveling pertinent reaction kinetics,^{50,51,115,174,197,203} and can be employed in tandem with reaction calorimetric measurements.^163,198 The combination of DoE and multivariate analyses for modeling process and spectroscopic data was discussed in the context of applying PAT to batch reactor for process understanding and improvement.⁹² A review in 2006 delineated the factors associated with scalability in fine chemicals and pharmaceuticals processes. The issues, difficulties and problems in scale-up are discussed for single-phase reactions, multiphase reactions (solid-solid, gas-liquid, liquid–liquid), and catalysed reactions, with a set of potential scale-up obstacles listed.²³⁰ The use of in-line ATR-FTIR as a process development tool in bioprocesses has been demonstrated for a pilot scale fungal fermentation.²⁰⁹ A recent paper reported the development of a NIR analytical protocol for assuring quality control of the API and excipients contents in a pharmaceutical formulation from laboratory bench to process scales.²³¹ Another series of papers gave a detailed account of using NIR for process control of pharmaceutical powder blending through discussing experimental design for powder mixing characterization, qualitative modeling for blend homogeneity prediction, and quantitative calibration for prediction of blend homogeneity and powder mixing kinetics.²³²

Process analytics and chemometrics

Mid- and near-infrared monitoring can be achieved through different measurement techniques: transmission, transmission-reflection (transflection), retro-reflection, external reflection, diffuse reflection, attenuated total reflectance (ATR) techniques, and more recently microspectroscopy and imaging.^44,45,60,61 Technological advancements in the infrared spectrometry and accessories for process analyses were previously discussed in the section Process Analyzers and Instrumentation. Good overviews on these infrared techniques in pharmaceutical applications are available,^44,45,233 and more specific examples in pharmaceuticalR&D, manufacturing and also bioprocesses were discussed in the above paragraphs. A review on infrared spectroscopy for catalysis appeared in 2001, which is thus far the most comprehensive review covering transmission, ATR, DRIFTS, emission, PAS and hyphenated infrared techniques for investigating catalysed reactions, with schematic diagrams of different infrared cells, over 900 references cited, a list of acronyms and a glossary of important FTIR terms.⁴³ There is a recent review that surveyed the application of NIR and chemometrics analyses in pharmaceutical processes.¹⁹⁹ This review is sectioned by qualitative and quantitative analyses, and processes that NIR spectroscopy affords on-line control. The quantitative NIR pharmaceutical applications were tabulated with annotated notes on analyte identity and sample form, chemometrics methods used and summarized findings.¹⁹⁹ Furthermore, advanced MIR and NIR microspectroscopic instrumentation gave rise to hyperspectral chemical imaging approaches for pharmaceutical applications.^{64,68,69,199,227,234,235}

With the availability of commercial Raman process analyzers that produce high quality data and ease of usage, Raman spectroscopy has been an extremely successful PAT tool.^{49–51,115,183} Use of Raman spectrometry for real-time monitoring of pharmaceutical processes was reviewed, with discussions on its relation to the PAT initiative, instrumentation and technology, data analysis, and a rather comprehensive list of applications in both non-pharmaceutical and pharmaceuticalPAT.¹¹⁵ The pharmaceuticalPAT applications are helpfully summarised under primary and secondary manufacturing.¹¹⁵ The application of Raman spectroscopy to the study of crystals was the focus of a chapter in a Raman handbook published in 2001, which discussed the fundamentals in Raman scattering effect of crystals such as lattice vibrations, librations, group theory, selection rules, symmetry attributions, mode assignments and notations, effects from isomorphic and isotopic substitutions.²¹⁵ A recent review article on the use of in situRaman spectroscopy for in-line control of crystallization processes discusses the use of Raman in the PAT context.²¹⁶ It surveyed published literature on crystallization studies using Raman spectroscopy and methods of calibration for the period 2000–2006. There are sections that discuss the advantages and disadvantages of Raman when applied to pharmaceutical API crystallization, different calibration methods through direct mathematical formulae or chemometrics techniques, and in situ monitoring of solid API processing and crystallization processes.²¹⁶ As with its MIR and NIR counterparts, Raman microspectroscopy has been increasingly explored for various pharmaceutical applications in recent years.^70,227,235

Chemometrics and multivariate analyses in PAT applications can be categorized into the three main scopes of (i) factor reduction and analysis, (ii) calibration and (iii) prediction. The commonly used methods include PCA, PCR and PLS. Complementary methods for experimental design and signal preprocessing often accompany the aforesaid multivariate analyses. These numerical techniques serve different roles in their diverse applications. In process development and scale-up their primary goal is for process understanding, which is achieved through elucidating process related physicochemical correlations and information (e.g., kinetics, thermodynamics, product/process stability and safety, etc.). In manufacturing scenarios, the computational methods deployed are to be validated for ensuring numerical robustness when applied to multivariate statistical process control (MSPC), abnormal situation detection, fault diagnosis, and continuous process knowledge management over campaign time. There is a good number of reviews and articles that provide either an overview of utilizing chemometrics and multivariate analyses in PAT or specific categories of numerical methods in PAT applications. Most of the PAT reports mentioned in this review rely on spectroscopic measurements and utilize multivariate chemometrics analyses. Applying chemometrics to pharmaceutical and biotech processes can be a daunting task for practitioners who are not thoroughly acquainted with this field of multivariate statistical analyses. A short critical review published in 2006 that addressed the basics, computational approaches, use and misuse of chemometrics in PAT would be a helpful starter.⁸⁵ Other literature describes the various computational methods used and approaches to modeling multivariate data coming out from process analyzers.^{79,83,84,91,93} There are dedicated articles covering chemometrics strategies for PAT implementation,^85,132 process development and operations,^{79,84,93,231,232}spectroscopy,^93–97 hyperspectral microspectroscopic/imaging data analyses,^{62,73,199,227} multivariate calibration,^205,236,237 calibration transfer,^79,199,238 batch processes,^92,100,205 bioprocesses,²⁰¹ process control and diagnosis.^{84,132,196,239,240} It is common to find chemometrics modeling reported in PAT applications computed using commercial software, such as MATLAB,⁹² PLS Toolbox,^94,101SIMCA-P/P+,²²⁶ Unscrambler,²³² and GRAMS PLS/IQ.²⁴⁰

Novel applications

There are novel applications of analytics and data analyses in the PAT literature that are worth further mentioning. The first group comprises of applications that combine different analytical instrumentation to provide complementary chemical analyses. Some selected examples are as follows. In-line Raman and ATR-FTIR spectroscopies are combined with reaction calorimetric instrumentation for process understanding of reactions using toxic chemicals that are potentially unsafe at certain process conditions.¹⁹⁸ High-throughput experimentation using in-line ATR-FTIR, Raman, and FBRM measurements with off-line HPLC, microscopy and ¹H NMR analyses assist in finding the best crystallization conditions.¹⁶⁵In-line NIR spectrometry with in-line FBRM, on-line turbidity measurements and PCA were employed to study the co-precipitation of naproxen and polymer eudragit.²²⁰ UV and NIR spectroscopy with PCA were used for determining component concentrations in powder blend mixtures.²²⁵ The combination of in-line Raman process data with at-line NIR and powder XRD measurements and multivariate analyses reveal underlying polymorphic transitions of D-mannitol polymorphs during freeze-drying process.²²⁶

The second group includes new or less frequently used analytical measurements in PAT. Acoustic-Resonance Spectroscopy (ARS) with PCA or PCR multivariate analyses was reported to be useful for tablet identification and quantification of components in semi-solids.²⁴¹ Acoustic spectroscopy using ultrasound is also explored for in situ monitoring of particle size distribution and volume concentrations in batch crystallization processes.²⁴² Spectroscopic methods in the microwave and terahertz (THz) regions were tested for determining their utility in pharmaceutical applications. Microwave resonance technology (MRT) sensor was developed for implementation in fluidized-bed dryers to enable real-time continuous moisture determination of pharmaceutical granules.²⁴³ This is the first MRT implementation for such a purpose, and the accuracy of in-line granule moisture measurements using the novel MRT were verified with off-line methods of loss on drying using infrared (LOD/IR) and Karl Fisher titrations.²⁴³ A team from FDA collaborated with NIST to assess the feasibility of combining THzspectroscopy and chemometrics for quantifying pharmaceutical tablet concentrations.²⁴⁴ Three analysis methods were investigated, namely, spectral characteristics (i.e. unique THz spectral band peaks), superpositioning, and multivariate modeling (PCR and PLS1), and the results indicated promise for THz quantification, albeit tablet process history need to be accounted in the superpositioning method and the multivariate models yield high prediction errors for minor component in the formulation.²⁴⁴ A novel analytical indicator using effusivity for determining uniformity of powder blending has been investigated.²⁴⁵ The use of NMR flow cell for on-line monitoring of reactions (i.e., homogeneous sample) in pharmaceutical applications is reported.²⁴⁶ Recently, Dow Chemical Company published the first report on their in-house in situ fiber optic turbidity probe technology for monitoring and control of unseeded crystallization in commercial batch crystallizer.²²² This in situ turbidity technique has over the years of usage at Dow Chemical proven to be robust as many of them have been deployed for more than a decade without showing reduced performance or having fouling problems. Moreover, it is a lower cost alternative to more sophisticated in-line methods (e.g.ATR-FTIR, FBRM). With this turbidity probe measurements, stages in the unseeded crystallization process, namely nucleation, digestion and crystal growth can be monitored and crystallization is controlled by adjusting process variables to follow a predetermined turbidity signal profile over process time.²²² Other analytical methods that were reported as useful in PAT applications are electrophoresis and HPLC for purification of monoclonal antibodies,²²⁴ fluorescence microarrays for understanding the effect of oxygen limitation on gene expression²¹⁰ or investigating disturbing influences to cell physiology and vaccine product quality,²¹¹ and tomography for non-invasive imaging of pharmaceutical solid dosage forms at depth.²⁴⁷

Discussion

In this section, several common threads that underpin the global PAT endeavor are highlighted before identifying some potential gaps. These common threads can be broadly captured under five inter-related areas: organisational objectives, enabling sciences, economic outlook, collaborative efforts, and emerging trends (see Fig. 1). This collective perspective on PAT departs from the more usual science, technical, manufacturing and regulatory perspectives, which gives a useful alternative view for interpreting the global PAT development in the context of the variety of involvements from different organizations and sectors.

Fig. 1 Five Common Threads in Global PAT Endeavor.

Common threads in global PAT thrust

The organisational objectives for PAT manifests differently for each organization because of differing organizational goals. In practice, each organisation has unique organisational objectives that ultimately governs and shapes its internal PAT developments, implementations and external engagements. The organisational objective of the regulatory agencies to adopting PAT is for the mandate of safeguarding the interests and well-being of the end consumer or patient. As such, PAT efforts of these agencies are carried out through the means of consultation, collaborations, research funding, regulatory guidance, inspections and approvals. The ICH, with its organisational objective to harmonize regulation and regulatory approach in pharmaceuticals, has a unique contribution towards facilitating the outworking of the PAT framework. The ICH guidelines Q7, Q8, Q9, Q10, and Q11 surfaced the practical issues for all other parties—regulatory, professional bodies, academia, commercial and pharmaceutical companies—to engage and develop PAT approaches, technologies and implementations. Professional bodies and societies, such as ASTM, ISPE, ISA and others, are concerned with setting up best practices and guidelines that promote professionalism within their domains of expertise. Their PAT developments involve primarily member consultation and publication of standards, guidelines and articles on best practices in its science and technologies. Some of these professional organizations also get involved in education through holding conferences, training and certification. The academic institutions (universities and research centres) are in the business of training, generating inventions and novel methods to solve challenging problems under the scope of PAT. Their contributions are in the forms of education, innovations, and publications. Consortia, depending on their membership model (academia, commercial companies, regulatory agencies, professional bodies or mixed) and shared objectives, play a special role by bringing in various partners into platforms that cross-fertilize information and knowledge exchanges revolving around PAT and related issues. The various pharmaceutical manufacturing companies and technology vendors are profit-making businesses that exploit PAT know how and techniques for commercial purposes. Their organisational objective is primarily profit driven with decisions on technological developments strongly influenced by either private business owners or corporate shareholders. For companies that supply PAT solutions, the agenda is to develop technologies that serve the PAT requirements for their clientele of pharmaceutical MNCs and smaller biotech firms. As for the pharmaceutical and biotech companies that produce drugs, vaccines and other healthcare products, whether to adopt a validated PAT process very much depends on possible increase in profits of the final product through better controlled unit operations. Meeting stricter regulatory requirements from FDA, EMEA, or environmental agencies can also be a consideration for manufacturing companies to move towards PAT implementations, especially when failure to meet such requirements means decrease of profits or losing market share for some geographical regions.

The enabling sciences under PAT include an assortment of process analytical technologies and solutions that are an essential element of “understanding-based manufacture” of pharmaceuticals. Continuing developments that make instrumentation more sensitive and robust make the technologies usable in an ever-increasing range of applications. However, their application is far from straightforward, and there are pitfalls to avoid when analyzing multivariate data.⁸⁵ The full range of benefits can only be accessed when the analysis is combined with process understanding and a process that has been designed to be controlled. For example, the adoption of integrated enterprise solutions facilitates metadata record keeping and provides a vital link with DCS for process control. Such solutions are presently still at the early stage of testing and implementation in various pharmaceutical manufacturing scenarios. Also, the identification of suitable PAT analytics, multivariate data analyses and strategies should be carried out as early as practicable in the process development—so that they can be tested in the development experimentation. This brings the dual benefits of growing process understanding (assuming the right technologies are available) and giving confidence that the approach will work in the manufacturing environment.

The economic outlook that concerns PAT revolves around internal and external factors for organizations. Although dissimilar organizations have different PAT organisational objectives, their PAT initiatives will invariably be constrained by finances. The PAT guidance does not specifically require the use of advanced process analyzers that produce multivariate analytical data, but it is evident from the literature and industrial reports that such instrumentation and related integrated solutions are increasingly common in PAT applications; may it be for an individual process unit or an entire manufacturing line. The costs of development, implementation and maintenance of such technologies will be substantial considerations for both providers and users of PAT solutions. Externally, the combination of increased competition from generic products of rival companies, more stringent regulations and the unstable global economics squeezes the profit margins for pharmaceutical and biotech companies at present and in the foreseeable future. Also, the typical patent life of a novel NCE or formulation will not produce as much ROI as previously enjoyed. This in turn would restrict the company R&D funds available for new PAT technological developments or implementation of PAT to an existing pharmaceutical process unit/line that is generating healthy income. For the academia and consortia, the present global economic downturn has already resulted in either more cautious management or budget cuts in R&D funding from governments in Europe and the United States, which will impact the number and rate of new PAT technologies introduction, or more importantly, improving the implementation and integration of the vast array of tools currently available to produce commercially viable pharmaceuticalPAT processes.

Collaborative efforts in PAT can be observed from a few angles. Firstly, this is a crucial internal factor that determines the success of adopting PAT within pharmaceutical manufacturing. The implementation of PAT requires a range of organizational and cultural changes in a pharmaceutical company.²⁴⁸ While there has been much enthusiasm in the past for PAT as an “initiative” in itself, this has moved towards PAT being one of the key tools in the delivery of Quality by Design. There is no sense in trying to graft PAT onto the old thinking—much better to move towards a science- and understanding-based approach to all of the process development and manufacturing activities. Key areas requiring consideration include:

• Linking the analytical technology to those groups using process understanding—development chemists, process designers, engineers and operations staff

• Ensuring that the understanding from PAT is gathered, codified, stored and exploited; this involves having suitable archiving, data analysis/chemometrics, process understanding and modeling techniques available

• Shifting the production analytical culture from reactive (QC) to proactive (PAT and parametric release)

• Recognizing that decisions on analytical technique have far-reaching implications

• Balancing scientific curiosity and the need for robust techniques to apply in manufacture.

Secondly, it is also observed that the global PAT thrust promoted unprecedented collaborations between regulatory agencies, professional societies, academia, and commercial companies (instrument vendors, software providers, process and enterprise solution integrators, and pharmaceutical manufacturers). Such activities generally transpire through discussion forums, standardization efforts, collaborative projects and publications. Regulatory agencies such as FDA collaborate with universities and consortia to develop and test potential new technologies that can be harnessed for PAT applications. The ICH Q8, Q9 and Q10 documents⁸ have successfully raised the critical issues that were taken up by the FDA 21st century Pharmaceutical cGMPs guidance² and deliberations in professional societies (e.g., ASTM Committee E55⁹ and ISPE PQLI¹⁹). Other standards that are currently being deliberated for advancing pharmaceuticalcGMP in the pharmaceutical fraternity are the ANSI/ISA S.88 and S.95, OPC ADI Specification and related documents from the International Organization for Standardization (ISO).

Partnerships and collaborative efforts have also been undertaken among commercial companies for the common goal of implementing PAT in manufacturing scenarios. For example, Abbott Laboratory collaborated with Malvern Instruments to automate milling control using the on-line particle size analyzer Insitec. The result is a fully integrated system with a human-machine interface (HMI) which allows OPC remote control of the milling process through a wireless TCP/IP communications that feedback real-time particle size analysis.¹¹⁹ There are other examples similar to this. Also, companies developing enterprise systems (e.g., ABB, GE, and Siemens) work alongside with multivariate chemometrics software providers (CAMO, Symbion, and Umetrics) and analytical instrumentation manufacturers (Mettler Toledo, ABB, Kaiser Optical Systems Inc., etc.) to facilitate PAT implementations in manufacturing. The synergism that was achieved through the membership of OPC ADI working group—comprising of pharmaceutical manufacturers, process analyzer suppliers, enterprise solution integrators, and software providers—has in less than two years produced the OPC ADI Specification for hardware-software interoperability. The critical task of integrating PAT technologies to realize validated PAT processes will certainly be expedited through the working out of this standard.

Emerging trends at the global scale influence the paths in which PAT will continue to develop. One is the climate change deliberations and global green awareness that has prompted companies to “go green”. It was reported that the pharmaceutical industry as among the top in the Newsweek's recent Green Rankings.²⁴⁹ However, more can and should be done. The technologies utilized in PAT applications for CQA monitoring and QCPP controls are not only suited for QbD and 21st century risked based cGMP, they are potentially effective for realizing green efforts and environmental sustainability.²⁵⁰ Along this theme, GSK has recently committed $33 million R&D funding in Singapore to promote green and sustainable manufacture of pharmaceuticals. Another emerging trend in the industry is for continuous processing. This is being adopted as a means to obtain more robust and reproducible manufacture, and potentially to reduce costs. Continuous processing will represent a new set of challenges to PAT; the ability to control the outcome of a process on the basis of sophisticated PAT measurements is a real challenge. The changing economics for the pharmaceutical industry that arise from multiple factors such as price pressures, cost of drugs, and increasing refusal of approval from regulatory agencies, thinner development pipeline and faster rival competition, will affect decision making for or against PAT implementations and continuous improvements.

Potential gaps

From the survey of the multi-faceted global PAT efforts and also the industrial feedback received, several gaps are presently observed. Firstly, there must be a business value for the adoption of PAT solutions in existing manufacturing processes or new processes and plants. The bottom line is an economically viable ROI. This has to be supported by sufficient knowledge in the organization to be able to make the business case for investment in a timely manner.

Secondly, due to commercial interests, there is reluctance for companies to divulge process analytics or PAT applications, regardless of whether the internal attempts resulted in success or failure.⁵¹ A good case to note is the in-house in situ fiber optic turbidity probe technology that Dow Chemical Company had used for more than a decade for monitoring commercial batch crystallizer, which is proven to be a cheaper alternative to other in-line analytics.²²² Such technology, if made known early, would have saved cost and wastage across the entire industry. However, early information release of the technology, especially if it is not patent protected, would mean a loss of comparative advantage to the company that had invented it.

Thirdly, the seamless integration of PAT technologies—process analytics, multivariate analyses, process modeling and control—is still some distance from maturity. This is a critical area that requires more developments, and it is hopeful that the OPC ADI Specification can greatly facilitate this integration work.

Fourthly, the current pharmaceutical manufacturing processes are primarily based on batch processes that on average are operating at a 2–3 sigma level of process control. With the potential pharmaceutical process innovations that come with PAT and the 21st century cGMPs initiatives, it is with considerable expectation to see how the adoption of new process equipment (e.g., continuous processing), together with varied PAT tools, risk-based approaches, Six Sigma and Lean Manufacturing methods can lead to more robust and cost-saving QbD processes that will be right first time (RFT) in product delivery.

Fifth, the rallying and management of the multi-disciplinary team required for implementing PAT is not trivial. People from diverse training background and experience such as chemists, engineers, chemometricians, system developers, etc., have to learn to work in concert.

Lastly, as PAT traverses several existing domains of knowledge and is continuing to evolve in new technologies and applications, there is a present lack of specific education in academia to cater to this new trend in pharmaceutical engineering. It would be beneficial to have some form of curriculum developed in the near future to teach chemists and engineers working in this field.

Conclusion

The global PAT initiative has engendered unprecedented activities leading towards a synergistic integration of diverse expertise across multi-disciplines in industry, academia, professional societies and regulatory agencies. The goal is for promoting pharmaceutical innovations and raising the robustness of pharmaceutical cGMPs in the 21st century that will ensure final product quality for the patient.

The 5-year period of 2004–2009 witnessed many investments, collaborations, achievements and successes in deploying and working out PAT concepts, technologies and solutions. Much of the methods and applications in process analytics and process chemometrics for PAT stem from their forerunner sciences and technologies in the fields of process analytical chemistry (PAC) and multivariate chemometrics techniques (PCA, PLS, DoE, etc.). There are also new techniques that are gaining popularity, e.g., hyperspectral chemical imaging using vibrational microspectroscopy, and new or lesser applied in situ process analytics such as THz, NMR and acoustics spectroscopies, multi-way chemometrics, etc. Other technologies that are harnessed for PAT include process systems engineering and control, information technology, software engineering, industrial automation and open standard protocols for communication and device integration (e.g., OPC UA, OPC ADI, ANSI/ISA S.88 and S.95, etc.).

The convergence points that organically grew out of this PAT initiative had started new international platforms of interactions for the underlying science, technology, education, business, management and regulation. Within this global thrust, five common threads were identified within an alternative PAT paradigm, namely, organisational objectives, enabling sciences, economic outlook, collaborative efforts, and emerging trends. These common factors overlap and will not remain static with time or scope. In the foreseeable future, they will continue to evolve, interact and shape the global PAT venture. Also, there are presently existing gaps that would require continuing innovation and collaboration from all stakeholders to bridge. It is hopeful that the promised benefits from the PAT initiative can be reaped for both the present-day end consumer and generations to come.

Glossary

API OPC OPC ADI OPC A&E OPC DA OPC UA

	Active
	Pharmaceutical
	Ingredient (
	pharmaceutical
	usage)
API	Application Programming Interface (software usage)
COM	Component Object Model (Microsoft)
cGMP	Current Good Manufacturing Practice
CQA	Critical Quality Attributes
DCS	Distributed Control System
DCOM	Distributed COM (Microsoft)
DLL	Dynamic-link Library (Microsoft)
ERP	Electronic Batch Reporting/Record-Keeping
MCR	Multivariate Curve Resolution
MES	Manufacturing Enterprise System
MVA	Multivariate Analysis
NCE	New Chemical Entities
QbD	Quality by Design
QbT	Quality by Testing
OCS	Overall Consolidated SCADA
OEE	Overall Equipment Effectiveness
OLE	Object Linking Embedding (Microsoft)
OOP	Object-Oriented Programming
	OLE for Process Control (
	OPC
	Foundation)
	OPC Analyzer Device Integration (
	OPC
	Foundation)
	OPC Alarm & Event (
	OPC
	Foundation)
	OPC Data Access (
	OPC
	Foundation)
	OPC Unified Architecture (
	OPC
	Foundation)
RFT	Right First Time
ROI	Return On Investment
RTR	Real-Time Release
PAT	Process Analytical Technology (FDA)
PCA	Principal Component Analysis
PCR	Principal Component Regression
PLC	Programmable Logic Controller
PLS	Partial Least Squares or Projection to Latent Structures
PQLI	Product Quality Lifecycle Implementation (ISPE)
QCPP	Quality Critical Process Parameters
SCADA	Supervisory Control and Data Acquisition

Acknowledgements

The authors would like to thank all industrial and academic respondents who provided invaluable feedback on the various aspects and trends in PAT. This review is an outcome of the support for the project ICES/09-130B03.

References

1. U.S. Food and Drug Administration, Guidance for Industry, PAT – A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance, September 2004.
2. U.S. Food and Drug Administration, Pharmaceutical cGMPS for the 21st Century – A Risk-Based Approach (Final Report), September 2004.
3. D. C. Watts and J. E. Clark, Journal of Process Analytical Technology, 2006, 3, 6–9.
4. PIC/S, Committee Meeting 2–3 November 2009. Press Release, Uppsala (Sweden), 17 November 2009.
5. U.S. Food and Drug Administration, Guideline for Industry Text on Validation of Analytical Procedures, ICH-Q2A, March 1995; Emea, Ich Topic Q 2 (R1) Validation of Analytical Procedures: Text and Methodology, June 1995; U.S. Food and Drug Administration, Guidance for Industry, Q2B Validation of Analytical Procedures: Methodology, November 1996.
6. EMEA, Note for Guidance on Parametric Release, March 2001.
7. Emea Pat team, Reflection Paper: Chemical, pharmaceutical and biological information to be included in dossiers when Process Analytical Technology (PAT) is employed, EMEA/INS/277260/2005, March 2006.
8. http://www.ich.org/ .
9. http://www.astm.org/COMMIT/COMMITTEE/E55.htm .
10. ASTM International Committee E55, E55 Fact Sheet, January 2010.
11. http://www.astm.org/BOOKSTORE/BOS/TOCS_2009/13.01.html .
12. ICH, ICH Harmonised Tripartite Guideline, Q7: Good Manufacturing Practice Guide for Active Pharmaceutical Ingredients, Current Step 4 version, November 2000.
13. ICH, ICH Harmonised Tripartite Guideline, Q8 (R2): Pharmaceutical Development, Current Step 4 version, August 2009.
14. ICH, ICH Harmonised Tripartite Guideline, Q9: Quality Risk Management, Current Step 4 version, November 2005.
15. ICH, ICH Harmonised Tripartite Guideline, Q10: Pharmaceutical Quality System, Current Step 4 version, June 2008.
16. http://www.ich.org/cache/html/4522-616-1.html .
17. http://www.ispe.org/ .
18. http://www.ispe.org/cs/explore_by_topic/pqli_resources2 .
19. T. Garcia, G. Cook and R. Nosal, J. Pharm. Innov., 2008, 3, 60–68.
20. R. Nosal and T. Schultz, J. Pharm. Innov., 2008, 3, 69–78.
21. J. Lepore and J. Spavins, J. Pharm. Innov., 2008, 3, 79–87.
22. R. Bolton and S. Tyler, J. Pharm. Innov., 2008, 3, 88–94.
23. B. Davis, L. Lundsberg and G. Cook, J. Pharm. Innov., 2008, 3, 95–104.
24. http://www.isa.org/ .
25. http://www.cpac.washington.edu/ .
26. M. V. Koch, K. M. VandenBussche and R. W. Chrisman (eds), Micro Instrumentation for High Throughput Experimentation and Process Intensification – a Tool for PAT, Wiley-VCH, Weinheim 2007.
27. http://www.cpact.com/ .
28. OPC Foundation, OPC Unified Architecture Companion Specification for Analyser Devices, released November 2009.
29. M. Postol, V. J. Janisz, C. M. Lafond and J. Luth, White Paper: Implementation of the OPC UA Information Model for Analyzers, 2009, Version 1.00.
30. OPC Foundation, White Paper: OPC Overview, October 1998, version 1.0.
31. W. Mahnke, S.-H. Leitner, and M. Damm, OPC Unified Architecture, Springer-Verlag, Berlin/Heidelberg 2009.
32. J. E. Carroll, Pharmaceutical Technology, 2005, 1, electronic edition. Available at http://electronic.pharmtech.com.
33. Nils-Erik Andersson, Pittsburg Conference (Pittcon), Orlando, 2005.
34. E. Baughman, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005, ch.1, pp. 1–12.
35. J. Workman, Jr., M. Koch, B. Lavine and R. Chrisman, Anal. Chem., 2009, 81, 4623–4643.
36. K. J. Clevett, Process Analyzer Technology, Wiley-Interscience, New York, 1986.
37. K. R. Beebe, W. W. Blaser, R. A. Bredeweg, J. P. Chauvel, Jr., R. S. Harner, M. LaPack, A. Leugers, D. P. Martin, L. G. Wright and E. D. Yalvact, Anal. Chem., 1993, 65, 199R–216R.
38. K. Eckschlager, K. Danzer, Information Theory in Analytical Chemistry, Wiley-Interscience Publication, New York, 1994, ch. 1, pp. 5-18.
39. D. C. Hassell and E. M. Bowman, Appl. Spectrosc., 1998, 52, 18A–29A.
40. S. Kueppers and M. Haider, Anal. Bioanal. Chem., 2003, 376, 313–315.
41. R. S. Saltzman, in Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, ed. R. A. Meyers, John Wiley & Sons, WestSussex, 2000, vol. 9, pp. 8328–8334.
42. L. C. Baylor and P. E. O'Rourke, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005, ch. 6, pp. 170–186.
43. J. Ryczkowski, Catal. Today, 2001, 68, 263–381.
44. J. P. Coates, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005, ch. 4, pp. 91–132.
45. J. D. Tate, P. Chauvel, R. D. Guenard and R. Harner, in Handbook of Vibrational Spectroscopy, ed. J. M. Chalmers and P. R. Griffiths, John Wiley & Sons Ltd, WestSussex, 2002, vol. 4, pp. 2737–2769.
46. M. A. Leugers and E. D. Lipp, in Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, ed. R. A. Meyers, John Wiley & Sons, West Sussex, 2000, vol. 9, pp. 8281–8289.
47. F. Adar, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 2, pp. 11–40.
48. J. B. Slater, J. M. Tedescon, R. C. Fairchild and I. R. Lewis, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 3, pp. 41–144.
49. I. R. Lewis, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 23, pp. 919–974.
50. N. Everall, I. Clegg and B. King, in Handbook of Vibrational Spectroscopy, ed. J. M. Chalmers and P. R. Griffiths, John Wiley & Sons Ltd, West Sussex, 2002, vol. 4, pp. 2770–2801.
51. Nancy L. Jestel, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005, ch. 5, pp. 133–169.
52. J. M. Harnly and R. E. Fields, Appl. Spectrosc., 1997, 51, 334A–351A.
53. P. J. Melling and M. Thomson, in Handbook of Vibrational Spectroscopy, ed. J. M. Chalmers and P. R. Griffiths, John Wiley & Sons Ltd, West Sussex, 2002, vol. 2, pp. 1551–1559.
54. T. R. Todd, in Handbook of Vibrational Spectroscopy, ed. J. M. Chalmers and P. R. Griffiths, John Wiley & Sons Ltd, West Sussex, 2002, vol. 2, pp. 1574–1586.
55. I. R. Lewis and M. L. Lewis, in Handbook of Vibrational Spectroscopy, ed. J. M. Chalmers and P. R. Griffiths, John Wiley & Sons Ltd, West Sussex, 2002, vol. 2, pp. 1587–1597.
56. K. J. Baldwin, D. N. Batchelder and S. Webster, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 4, pp. 145–190.
57. P. J. Treado, M. P. Nelson, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 5, pp. 191–250.
58. I. W. Levin and R. Bhargava, Annu. Rev. Phys. Chem., 2005, 56, 429–474.
59. R. Bhargava and I. W. Levin (ed.), Spectrochemical Analysis Using Infrared Multichannel Detectors, Blackwell Publishing, Oxford, 2005.
60. R. Bhargava and I. R. Levin, in Spectrochemical Analysis Using Infrared Multichannel Detectors, ed. R. Bhargava and I. W. Levin, Blackwell Publishing, Oxford, 2005, ch. 1, pp. 1–24.
61. E. N. Lewis, L. H. Kidder, E. Lee and K. S. Haber, in Spectrochemical Analysis Using Infrared Multichannel Detectors, ed. R. Bhargava and I. W. Levin, Blackwell Publishing, Oxford, 2005, ch. 2, pp. 25–55.
62. H. F. Grahn and P. Geladi (ed.), Techniques and Applications of Hyperspectral Image Analysis, Wiley, West Sussex, 2007.
63. V. Baeten and P. Dardenne, in Spectrochemical Analysis Using Infrared Multichannel Detectors, ed. R. Bhargava and I. W. Levin, Blackwell Publishing, Oxford, 2005, ch. 13, pp. 283–301.
64. E. N. Lewis, J. W. Schoppelrei, E. Lee and L. H. Kidder, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005, ch. 7, pp. 187–225.
65. J. van der Weerd and S. G. Kazarian, in Techniques and Applications of Hyperspectral Image Analysis, ed. H. F. Grahn and P. Geladi, Wiley, West Sussex, 2007, ch. 10, pp. 221–260.
66. R. Salzer and H. W. Siesler (ed.), Infrared and Raman Spectroscopic Imaging, Wiley-VCH Verlag, Weinheim, 2009.
67. A. E. Dowrey, G. M. Story and C. Marcott, in Spectrochemical Analysis Using Infrared Multichannel Detectors, ed. R. Bhargava and I. W. Levin, Blackwell Publishing, Oxford, 2005, ch. 8, pp. 175–188.
68. E. Wessel, C. Vogel, O. Kolomiets, U. Hoffmann and H. W. Siesler, in Infrared and Raman Spectroscopic Imaging, ed. R. Salzer and H. W. Siesler, Wiley-VCH Verlag, Weinheim, 2009, ch. 9, pp. 297–346.
69. S. G. Kazarian, K. L. A. Chan, and F. H. Tay, in Infrared and Raman Spectroscopic Imaging, ed. R. Salzer and H. W. Siesler, Wiley-VCH Verlag, Weinheim, 2009, ch. 10, pp. 347–376.
70. E. Lee, in Infrared and Raman Spectroscopic Imaging, ed. R. Salzer and H. W. Siesler, Wiley-VCH Verlag, Weinheim, 2009, ch. 11, pp. 377–403.
71. P. Geladi, Chemom. Intell. Lab. Syst., 1989, 7, 11–30.
72. S. W. Huffman and C. W. Brown, in Spectrochemical Analysis Using Infrared Multichannel Detectors, ed. R. Bhargava and I. W. Levin, Blackwell Publishing, Oxford, 2005, ch. 4, pp. 85–114.
73. A. de Juan, M. Maeder, T. Hancewicz, L. Duponchel and R. Tauler, in Infrared and Raman Spectroscopic Imaging, ed. R. Salzer and H. W. Siesler, Wiley-VCH Verlag, Weinheim, 2009, ch. 2, pp. 65–111.
74. U.S. Food and Drug Administration, Guidance for Industry Part 11, Electronic Records; Electronic Signatures — Scope and Application, August 2003.
75. U.S. Food and Drug Administration, General Principles of Software Validation; Final Guidance for Industry and FDA Staff, January 2002.
76. K. Eckschlager, V. Štěpánek and K. Danzer, J. Chemom., 1990, 4, 195–216.
77. S. N. Deming, Anal. Chim. Acta, 1991, 249, 303–304.
78. R. G. Brereton, Chemometrics: Applications of Mathematics and Statistics to Laboratory Systems, Ellis Horwood Ltd., New York, 1990.
79. C. E. Miller, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, 2005. ch. 8, pp. 226–328.
80. E. R. Malinowski and D. G. Howery. Factor Analysis in Chemistry, Wiley, New York, 1980; M. A. Sharaf, D. L. Illman and B. R. Kowalski. Chemometrics, Wiley, New York 1986; D. L. Massart, B. G. M. Vandeginste, D. N. Deming, Y. Michotte and L. Kaufman. Chemometrics: A Textbook, Elsevier, Amsterdam, 2nd edn, 1988; R. R. Meglen, Chemom. Intell. Lab. Syst., 1988, 3, 17–19; S. D. Brown, T. S. Stephen, F. Despagne and B. K. Lavine, Anal. Chem., 1996, 68, 21R–61R; K. R. Beebe, R. J. Pell and M. B. Seasholtz. Chemometrics: A Practical Guide, Wiley, New York, 1998; E. R. Malinowski, Factor Analysis in Chemistry, Wiley-Interscience, New York, 3rd edn, 2002; P. Gemperline (ed.), Practical Guide to Chemometrics, CRC Press, Boca Raton, 2nd edn, 2006.
81. B. Lavine and J. Workman, Jr., Anal. Chem., 2002, 74, 2763–2770.
82. J. Workman, Jr., M. Koch and D. J. Veltkamp, Anal. Chem., 2003, 75, 2859–2876.
83. R. G. Brereton, Journal of Process Analytical Technology, 2005, 2, 8–11.
84. T. Kourti, Crit. Rev. Anal. Chem., 2006, 36, 257–278.
85. S. J. Doherty and A. J. Lange, TrAC, Trends Anal. Chem., 2006, 25, 1097–1102.
86. H. Aleem, S. Lord, T. McCarthy, P. Sharratt and Y. Zhao, Pharmaceutical Engineering, July/Aug 2003, 32–42.
87. NIST, Integration Definition for Function Modeling (IDEF0), Draft Federal Information Processing Standards Publication, December 1993, p. 183.
88. K. Wall, P. N. Sharratt, N. Sadr-Kazemi and J. N. Borland, Org. Process Res. Dev., 2001, 5, 434–437.
89. P. N. Sharratt, K. Wall and J. N. Borland, J. Chem. Technol. Biotechnol., 2003, 78, 156–160.
90. http://www.britest.co.uk/ .
91. J. Workman, Jr., D. J. Veltkamp, S. Doherty, B. B. Anderson, K. E. Creasy, M. Koch, J. F. Tatera, A. L. Robinson, L. Bond, L. W. Burgess, G. N. Bokerman, A. H. Ullman, G. P. Darsey, F. Mozayeni, J. A. Bamberger and M. S. Greenwood, Anal. Chem., 1999, 71, 121R–180R; J. Workman Jr., K. E. Creasy, S. Doherty, L. Bond, M. Koch, A. Ullman and D. J. Veltkamp, Anal. Chem., 2001, 73, 2705–2718; J. Workman, Jr., M. Koch and D. Veltkamp, Anal. Chem., 2005, 77, 3789–3806; J. Workman, Jr., M. Koch and D. Veltkamp, Anal. Chem., 2007, 79, 4345–4364.
92. J. Gabrielsson, H. Jonsson, J. Trygg, C. Airiau, B. Schmidt and R. Escott, AIChE J., 2006, 52, 3164–3172.
93. C. E. Miller, J. Chemom., 2000, 14, 513–528.
94. P. Geladi, Spectrochim. Acta, Part B, 2003, 58, 767–782.
95. P. Geladi, B. Sethson, J. Nyström, T. Lillhonga, T. Lestander and J. Burger, Spectrochimica Acta Part B, 2004, 59, 1347–1357.
96. J. Workman, Jr., in The Handbook of Organic Compounds. NIR, IR, Raman, and UV-Vis Spectra Featuring Polymers and Surfactants, Academic Press, San Diego, 2001, ch. 27, pp. 295–300; ch. 28, pp. 301–326; ch. 29, pp. 327–337; ch. 30, pp. 339–352.
97. H. Mark and J. Workman Jr., Chemometrics in Spectroscopy, Academic Press, London, 2007.
98. A. Ferrer, D. Aguado, S. Vidal-Puig, J. M. Prats and M. Zarzo, Applied Stochastic Models in Business and Industry, 2008, 24, 551–567.
99. V. Uraikul, C. W. Chan and P. Tontiwachwuthikul, Eng. Appl. Artif. Intell., 2007, 20, 115–131.
100. P. Nomikos and J. F. MacGregor, AIChE J., 1994, 40, 1361–1375; B. R. Bakshi, AIChE J., 1998, 44, 1596–1610; S. Wold, N. Kettaneh, H. Fridén and A. Holmberget, Chemom. Intell. Lab. Syst., 1998, 44, 331–340; S. P. Gurden, J. A. Westerhuis and A. K. Smilde, AIChE J., 2002, 48, 2283–2297.
101. P. Teppola and P. Minkkinen, J. Chemom., 2000, 14, 383–399.
102. Z. P. Chen, D. Lovett and J. Morris, Computer Aided Chemical Engineering, 2008, 18, 967–972.
103. S. Ravendra, K. V. Gernaey and R. Gani, Computer Aided Chemical Engineering, 2008, 25, 423–428.
104. S. Ravendra, K. V. Gernaey and R. Gani, Computers and Chemical Engineering, 2010, 34, 1108–1136.
105. R. Gani and I. E. Grossmann, Computer Aided Chemical Engineering, 2007, 24, 1–5.
106. http://www.nipte.org/ .
107. http://www.brukeroptics.com/ .
108. R. Kobayashi, Y. Fujimaki, T. Ukita and Y. Hiyama, Org. Process Res. Dev., 2006, 10, 1219–1226.
109. http://www.thermo.com/ .
110. B. Davies, The Development of PAT in biotech manufacturing, http://www.thermo.com/eThermo/CMA/PDFs/Product/productPDF_1113.pdf.
111. http://www.foss.dk/ .
112. R. Mattes, D. Root, M. A. Sugui, F. Chen, X. Shi, J. Liu and P.-A. Gilbert, BioProcess International, April 2009, 44–50.
113. R. A. Mattes, R. Schroeder, V. Dhopeshwarkar, R. Kowal and W. Randolph, Pharmaceutical Technology, Process Analytical Technology Supplement, September 2004.
114. http://www.kosi.com/ .
115. K. L. Davis, M. S. Kemper, and I. R. Lewis, in Pharmaceutical Applications of Raman Spectroscopy, ed. S. Šašić, John Wiley & Sons, New Jersey, 2008, ch. 5, pp. 117–162.
116. http://www.malvern.com/LabEng/industry/pharmaceutical.htm .
117. L. H. Tan, L. W. Chan, and P. W. S. Heng, Pharmaceutical Technology Europe, November 2008, http://pharmtech.findpharma.com/pharmtech/analytical%2Bequipment/Rapid-and-convenient-microsphere-sizing/ArticleStandard/Article/detail/566710.
118. Malvern Instruments, White Paper: Implementing PAT and QbD –; Real-time particle size measurement for the pharmaceutical industry. http://www.malvern.com.
119. A. Vaisman and Y. Hu, Pharmaceutical Technology, January 2010, http://pharmtech.findpharma.com/pharmtech/Article/A-PAT-Solution-for-Automated-Mill-Control/ArticleStandard/Article/detail/651078%3Fref%20%3D%2025.
120. Y. Roggo and M. Ulmschneider, in Pharmaceutical Manufacturing Handbook: Regulations and Quality, ed. S. C. Gad, John Wiley & Sons, New Jersey, 2008, ch. 4.3, pp. 411–432.
121. J. Dubois and E. N. Lewis, Pharmaceutical Solutions Update, Fall 2008, 38–45.
122. J. Dubois, J.-C. Wolff, J. K. Warrack, J. Schoppelrei and E. N. Lewis, Spectroscopy, 2007, 22, 40–50.
123. http://www.mt.com/ .
124. http://www.mt.com/autochem .
125. http://www.mathworks.com/ .
126. T. C. O'Haver, Chemom. Intell. Lab. Syst., 1989, 6, 95–103.
127. http://www.chemometrics.com/ .
128. http://www.camo.com/ .
129. http://www.camo.com/solutions/process-analytical-technology.html .
130. http://www.infometrix.com/ .
131. http://www.umetrics.com/ .
132. S. Wold, J. Cheney, N. Kettaneh and C. McCready, Chemom. Intell. Lab. Syst., 2006, 84, 159–163.
133. ABB brochure: AnalyzeIT for Integrated Solutions. http://www.abb.com.
134. ABB, Control System Evolution – A White Paper on evolving your ABB Open Control System to 800xA Extended Automation. http://www.abb.com.
135. ABB, Industrial IT eXtended Process Analytical Technology System Specifications. http://www.abb.com.
136. K. A. Bakeev, personal communication; K. A. Bakeev (ed.), Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, Blackwell, Chichester, 2010.
137. D. Funderburg, ABB Automation & Power World, Orlando, 2009.
138. http://www.ge-ip.com/ .
139. GE, White Paper: Proficy Software – A Powerful Production Solution that Delivers Results. http://www.ge-ip.com, 2010.
140. GE, White paper: Evolution, Not Revolution – The GE Fanuc HMI/SCADA Customer Protection Strategy, http://www.ge-ip.com, 2010.
141. GE, Press Release: GE Fanuc Automation Announces the Proficy^® RX Comprehensive PAT Solution for the Life Sciences Industry, Charlottesville, 2006.
142. M. A. Power and W. M. Doyle, Spectroscopy Supplement, Process Analytical Technologies, February 2006, 40–50.
143. GE, Proficy^® RX – Process Analysis Software for Process Analytical Technologies (PAT), http://www.ge-ip.com, 2010.
144. Symbion Systems Inc., Technical Note SD-501(204): An Introduction to Symbion™ The Language, 2010.
145. National Instruments, PACs for Industrial Control, the Future of Control. Programmable Automation Controllers White Paper. http://www.ni.com/pac/.
146. http://sine.ni.com/cs/app/doc/p/id/cs-439 .
147. http://zone.ni.com/devzone/cda/tut/p/id/4570 .
148. http://www.colemantech.com/ .
149. http://zone.ni.com/reference/en-XX/help/370246E-01/lvdscconcepts/accessing_opc_servers .
150. http://www.ni.com/lwcvi .
151. E. Yeo and R. Vangenechten, personal communication.
152. Siemens, White Paper: SIPAT – The Software Heart of PAT. http://www.siemens.com/SIPAT, 2010.
153. http://www.automation.siemens.com/w2/automation-technology-sipat-8994.htm .
154. http://www.sedb.com/edb/sg/en_uk/index/news/articles/glaxosmithkline_opens.html; http://www.pharmafocusasia.com/knowledge_bank/projects/glaxosmithkline_plant.htm.
155. http://www.pharmamanufacturing.com/articles/2008/022.html .
156. http://www.pharmamanufacturing.com/articles/2009/148.html .
157. C. R. Yonzon and B. A. Donovan, American Pharmaceutical Review, November/December 2009, 68–75.
158. A. Sparén, J. Johansson, O. Svensson, S. Folestad and M. Claybourn, American Pharmaceutical Review, January/February 2009, 62–71.
159. G. Zhou, A. Moment, A. Cote and T.-E. Hu, American Pharmaceutical Review, July/August 2009, 62–67.
160. O. Aurin, S. Himstedt and J. Zobel, Automation Technology in Practice, April 2007, 1–7.
161. F. Sistare, L. St. Pierre Berry and C. A. Mojica, Org. Process Res. Dev., 2005, 9, 332–336.
162. K. De Smet, J. van Dun, B. Stokbroekx, T. Spittaels, C. Schroyen, P. Van Broeck, J. Lambrechts, D. Van Cleuvenbergen, G. Smout, J. Dubois, A. Horváth, J. Verbraeken and J. Cuypers, Org. Process Res. Dev., 2005, 9, 344–347.
163. J. Wiss, M. Länzlinger and M. Wermuth, Org. Process Res. Dev., 2005, 9, 365–371.
164. A. Horváth, K. De Smet, D. Ormerod, D. Depré, C. Pérez-Balado, T. Govaerts, D. Van den Heuvel and I. Schorpion, Org. Process Res. Dev., 2005, 9, 356–359.
165. M. Birch, S. J. Fussell, P. D. Higginson, N. McDowall and I. Marziano, Org. Process Res. Dev., 2005, 9, 360–364.
166. M. L. Hamilton, B. B. Perston, P. W. Harland, B. E. Williamson, M. A. Thomson and P. J. Melling, Org. Process Res. Dev., 2005, 9, 337–343.
167. B. K. Lavine, Anal. Chem., 2000, 72, 91R–97R; B. Lavine and J. Workman, Jr., Anal. Chem., 2004, 76, 3365–3372; B. Lavine and J. Workman, Anal. Chem., 2006, 78, 4137–4145.
168. B. Lavine and J. Workman, Anal. Chem., 2008, 80, 4519–4531.
169. K. A. Bakeev (ed.), Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, Blackwell, Oxford, 2005.
170. R. Guenard and G. Thurau, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, ch. 2, pp.13–38.
171. M. B. Simpson, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, ch. 3, pp.39–90.
172. M. Blanco and M. Alcalá, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, ch. 10, pp. 362–391.
173. A. M. Brearley, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, ch. 11, pp. 392–423.
174. S. J. Doherty and C. N. Kettler, in Process Analytical Technology: Spectroscopic Tools and Implementation Strategies for the Chemical and Pharmaceutical Industries, ed. K. A. Bakeev, Blackwell, Oxford, ch. 9, pp. 329–361.
175. R. A. Meyers (ed.), Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, John Wiley & Sons, West Sussex, 2000.
176. J. M. Chalmers and P. R. Griffiths (ed.), Handbook of Vibrational Spectroscopy, John Wiley & Sons Ltd, WestSussex, 2002.
177. I. R. Lewis and H. G. M. Edwards (ed.), Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, Marcel Dekker, New York, 2001.
178. G. Gauglitz and T. Vo-Dinh (ed.), Handbook of Spectroscopy. Wiley-VCH, Weinheim, 2003.
179. D. A. Burns and E. W. Ciurczak (ed.), Handbook of Near-infrared Analysis, CRC Press, Boca Raton, 3rd edn, 2008.
180. References found in Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, ed. R. A. Meyers, John Wiley & Sons, West Sussex, 2000; J. W. Peterson and A. H. Ullman, vol. 9, pp. 8127–8145; K. S. Booksh, vol. 9, pp. 8145–8169; J. M. Clemons, vol. 9, pp. 8169–8193; J. H. Alstadt III, D. C. Olson, D. K. Wolcott, G. D. Marshall and S. W. Steig, vol. 9, pp. 8193–8217; J. P. Coates and P. H. Shelley, vol. 9, pp. 8217–8240; J. A. Olivares, vol. 9, pp. 8240–8255; D. S. Goldman, vol. 9, pp. 8256–8264; M. J. McCarthy and S. Bobroff, vol. 9, pp. 8264–8281; K. E. Creasy and T. W. Francisco, vol. 9, pp. 8289–8313; T. K. Fossum and D. K. Wolcott, vol. 9, pp. 8313–8327.
181. References found in Encyclopedia of Analytical Chemistry: Applications, Theory and Instrumentation, ed. R. A. Meyers, John Wiley & Sons, West Sussex, 2000; R. K. Dukor and L. A. Nafie, vol. 1, pp. 662–676; R. K. Gilpin, vol. 8, pp. 6989–6993; K. J. Duff, vol. 8, pp. 7158–7208; D. R. Goodlett, D. C. Gale, S. Guiles and J. B. Crowther, vol. 8, pp. 7208–7229; B. Dawson, vol. 8, pp. 7229–7242; S. A. Tomellini and J. W. Finn, vol. 8, pp. 7368–7390.
182. E. W. Ciurczak and J. K. Drennen III, Pharmaceutical and Medical Applications of Near-infrared Spectroscopy, Marcel Dekker, New York, 2002.
183. S. Šašić (ed.), Pharmaceutical Applications of Raman Spectroscopy, John Wiley & Sons, New Jersey, 2008.
184. J. M. Chalmers (ed.), Spectroscopy in Process Analysis, Sheffield Academic Press, Sheffield, 2000.
185. R. E. Sherman and L. Rhodes (ed.), Analytical Instrumentation, Instrument Society of America, 1996.
186. F. McLennan and B. R. Kowalski (ed.), Process Analytical Chemistry, Blackie Academic & Professional, London, 1995.
187. B. M. Wise and B. R. Kowalski, Process Chemometrics, in Process Analytical Chemistry, ed. F. McLennan and B. R. Kowalski, Blackie Academic & Professional, London, 1995, ch 8, pp. 259–312.
188. R. G. Brereton, Chemometrics: Data Analysis for the Laboratory and Chemical Plant, Wiley, West Sussex, 2003.
189. P. Gemperline (ed.), Practical Guide to Chemometrics, CRC Press, Boca Raton, 2nd edn, 2006.
190. R. G. Brereton, Applied Chemometrics for Scientists, John Wiley & Sons, New Jersey 2007.
191. H. Martens and T. Naes, Multivariate Calibration, John Wiley and Sons, Chichester, 1989.
192. E. R. Malinowski, Factor Analysis in Chemistry.Wiley-Interscience, New York, 3rd edn, 2002.
193. R. G. Brereton, Chemometrics for Pattern Recognition, John Wiley & Sons, Chichester, 2009.
194. S. D. Brown, R. Tauler, and B. Walczak (ed.), Comprehensive Chemometrics: Chemical and Biochemical Data Analysis, Elsevier Science, 2009.
195. References found in Comprehensive Chemometrics: Chemical and Biochemical Data Analysis, ed. S. D. Brown, R. Tauler and B. Walczak, Elsevier Science, 2009; F. F. Pitard, vol. 1, ch. 1, pp. 1–16; S. Wold, N. Kettaneh-Wold, J. F. MacGregor and K. G. Dunn, vol. 2, ch. 10, pp. 163–197; K. H. Esbensen and L. P. Julius, vol. 4, ch. 1, pp. 1–20; J. C. Menezes, A. P. Ferreira, L. O. Rodrigues, L. P. Brás and T. P. Alves, vol. 4, ch. 10, pp. 313–355.
196. T. Kourti, in Comprehensive Chemometrics: Chemical and Biochemical Data Analysis, ed. S. D. Brown, R. Tauler and B. Walczak, Elsevier Science, 2009, vol. 4, ch. 2, pp. 21–54.
197. V. S. Tumuluri, M. S. Kemper, A. Sheri, S.-R. Choi, I. R. Lewis, M. A. Avery and B. A. Avery, Org. Process Res. Dev., 2006, 10, 927–933.
198. J. Wiss and A. Zilian, Org. Process Res. Dev., 2003, 7, 1059–1066.
199. Y. Roggo, P. Chalus, L. Maurer, C. Lema- Martinez, A. Edmond and N. Jent, J. Pharm. Biomed. Anal., 2007, 44, 683–700.
200. L. O. Rodrigues, J. A. Lopes, J. P. Cardoso and J. C. Menezes, European Symposium on Computer-Aided Process Engineering, 2004, 14, 775–780.
201. J. A. Lopes, P. F. Costa, T. P. Alves and J. C. Menezes, Chemom. Intell. Lab. Syst., 2004, 74, 269–275.
202. J. B. Holm-Nielsen, C. J. Lomborg, P. Oleskowicz-Popiel and K. H. Esbensen, Biotechnol. Bioeng., 2008, 99, 302–313.
203. A. P. Teixeira, R. Oliveira, P. M. Alves and M. J. T. Carrondo, Biotechnol. Adv., 2009, 27, 726–732.
204. M. Warman and S. V. Hammond, Proceedings of the Pittsburgh Conference, Chicago, 2005.
205. J. Gabrielsson, H. Jonsson, C. Airiau, B. Schmidt, R. Escott and J. Trygg, J. Chemom., 2006, 20, 362–369.
206. J. B. Holm-Nielsen, C. K. Dahl and K. H. Esbensen, Chemom. Intell. Lab. Syst., 2006, 83, 114–126.
207. K. Kargosha, M. Khanmohammadi, M. Sarokhani, F. Ansari and M. Ghadiri, J. Pharm. Biomed. Anal., 2003, 31, 571–577.
208. D. L. Doak and J. A. Phillips, Biotechnol. Prog., 1999, 15, 529–539.
209. D. J. Pollard, R. Buccino, N. C. Connors, T. F. Kirschner, R. C. Olewinski, K. Saini and P. M. Salmon, Bioprocess Biosyst. Eng., 2001, 24, 13–24.
210. M. Streefland, B. van de Waterbeemd, J. Kint, L. A. van der Pol, E. C. Beuvery, J. Tramper and D. E. Martens, Biotechnol. Bioeng., 2009, 102, 161–167.
211. M. Streefland, B. van de Waterbeemd, H. Happé, L. A. van der Pol, E. C. Beuvery, J. Tramper and D. E. Martens, Vaccine, 2007, 25, 2994–3000.
212. S. Gnoth, M. Jenzsch, R. Simutis and A. Lübbert, J. Biotechnol., 2007, 132, 180–186.
213. L. X. Yu, R. A. Lionberger, A. S. Raw, R. D'Costa, H. Wu and A. S. Hussain, Adv. Drug Delivery Rev., 2004, 56, 349–369.
214. Z. Q. Yu, J. W. Chew, P. S. Chow and R. B. H. Tan, Chem. Eng. Res. Des., 2007, 85, 893–905.
215. D. C. Smith and C. Carabatos-Nédelec, in Handbook of Raman Spectroscopy: From the Research Laboratory to the Process Line, ed. I. R. Lewis, H. G. M. Edwards, Marcel Dekker, New York, 2001, ch. 9, pp. 349–422.
216. G. Févotte, Chem. Eng. Res. Des., 2007, 85, 906–920.
217. G. Févotte, Int. J. Pharm., 2002, 241, 263–278.
218. V. Liotta and V. Sabesan, Org. Process Res. Dev., 2004, 8, 488–494.
219. G. Févotte, J. Calas, F. Puel and C. Hoff, Int. J. Pharm., 2004, 273, 159–169.
220. H. Wu and M. A. Khan, J. Pharma. Sci., 2010, 99, 1516–1534.
221. N. Kail, W. Marquardt and H. Briesen, Ind. Eng. Chem. Res., 2009, 48, 2936–2946.
222. R. S. Harner, R. J. Ressler, R. L. Briggs, J. E. Hitt, P. A. Larsen and T. C. Frank, Org. Process Res. Dev., 2009, 13, 114–124.
223. S. Rohani, S. Horne and K. Murthy, Org. Process Res. Dev., 2005, 9, 873–883.
224. S. Flatman, I. Alam, J. Gerard and N. Mussa, J. Chromatogr., B: Anal. Technol. Biomed. Life Sci., 2007, 848, 79–87.
225. H. Wu, M. Tawakkul, M. White and M. A. Khan, Int. J. Pharm., 2009, 372, 39–48.
226. T. R. M. De Beer, M. Allesø, F. Goethals, A. Coppens, Y. Vander Heyden, H. Lopez De Diego, J. Rantanen, F. Verpoort, C. Vervaet, J. P. Remon and W. R. G. Baeyens, Anal. Chem., 2007, 79, 7992–8003.
227. C. Gendrin, Y. Roggo and C. Collet, J. Pharm. Biomed. Anal., 2008, 48, 533–553.
228. P. McKenzie, S. Kiang, J. Tom, A. E. Rubin and M. Futran, AIChE J., 2006, 52, 3990–3994.
229. J. E. Mills, in Chemical Process Research: The Art of Practical Organic Synthesis, ed. A. F. Abdel-Magid and J. A. Ragan, Oxford University Press/American Chemical Society, Washington, 2004, ch. 6, pp. 87–109.
230. G. Caygill, M. Zanfir and A. Gavriilidis, Org. Process Res. Dev., 2006, 10, 539–552.
231. M. C. Sarraguça and J. Al. Lopes, Vib. Spectrosc., 2009, 49, 204–210.
232. A. S. El-Hagrasy, F. D'Amico and J. K. Drennen III, J. Pharm. Sci., 2006, 95, 392–406; A. S. El-Hagrasy, M. Delgado-Lopez and J. K. Drennen III, J. Pharm. Sci., 2006, 95, 407–421; A. S. El-Hagrasy and J. K. Drennen III, J. Pharm. Sci., 2006, 95, 422–434.
233. E. W. Ciurczak and J. K. Drennen, in Near-infrared Applications in Biotechnology, ed. Ramesh Raghavachari, Marcel Dekker, New York, 2001. ch. 13, pp. 347–364.
234. G. Reich, Adv. Drug Delivery Rev., 2005, 57, 1109–1143.
235. A. A. Gowen, C. P. O'Donnell, P. J. Cullen and S. E. J. Bell, Eur. J. Pharm. Biopharm., 2008, 69, 10–22.
236. P. Geladi, Chemom. Intell. Lab. Syst., 2002, 60, 211–224.
237. J. Gabrielsson and J. Trygg, Crit. Rev. Anal. Chem., 2006, 36, 243–255.
238. Y. Sulub, R. LoBrutto, R. Vivilecchia and B. Wa Wabuyele, Anal. Chim. Acta, 2001, 611, 143–150.
239. T. Kourti and J. F. MacGregor, Chemometrics and Intelligent Laboratory Systems, 1995, 28, 3–21.
240. J. B. Cooper, Chemom. Intell. Lab. Syst., 1999, 46, 231–247.
241. J. Medendorp and R. A. Lodder, AAPS PharmSciTech, 2006, 7(1)(Article 25), E1–E9; J. Medendorp and R. A. Lodder, AAPS PharmSciTech, 2006, 7(3)(Article 59), E1–E8.
242. A. K. Hipp, B. Walker, M. Mazzotti and M. Morbidelli, Ind. Eng. Chem. Res., 2000, 39, 783–789.
243. C. Buschmüller, W. Wiedey, C. Döscher, J. Dressler and J. Breitkreutz, Eur. J. Pharm. Biopharm., 2008, 69, 380–387.
244. H. Wu, E. J. Heilweil, A. S. Hussain and M. A. Khan, J. Pharm. Sci., 2008, 97, 970–984.
245. G. Léonard, F. Bertrand, J. Chaouki and P. M. Gosselin, Powder Technol., 2008, 181, 149–159.
246. M. A. Bernstein, M. Štefinović and C. J. Sleigh, Magn. Reson. Chem., 2007, 45, 564–571; M. A. Bernstein, Spectroscopy Europe, 2007, 19, 11–13.
247. J. A. Zeitler and L. F. Gladden, Eur. J. Pharm. Biopharm., 2008, 71, 2–22.
248. P. Sharratt and W. Chew, PharmaAsia, October 2009, 14–16.
249. P. Thomas, Pharma's Green Evolution, 2009; http://www.pharmamanufacturing.com/articles/2009/149.html.
250. P. Thomas, Is It Time for PAT to Go Green?, 2009; http://www.pharmamanufacturing.com/articles/2009/148.html.

---

Footnote

† The terminologies in situ and operandospectroscopies are synonymously associated with in- and on-line analytical instrumentation, with the term operando often applied in the catalysis research community.

---