Cover
Title Page
List of Contributors
Preface
1 Introduction
1. 1.1 Quality by Design Overview
2. 1.2 Pharmaceutical Industry
3. 1.3 Quality by Design Details
4. 1.4 Chapter Summaries
5. References
2 An Overview of the Role of Mathematical Models in Implementation of Quality by Design Paradigm for Drug Development and Manufacture
1. 2.1 Introduction
2. 2.2 Overview of Models
3. 2.3 Role of Models in QbD
4. 2.4 General Scientific Considerations for Model Development
5. 2.5 Scientific Considerations for Maintenance of Models
6. 2.6 Conclusion
7. References
3 Role of Automatic Process Control in Quality by Design
1. 3.1 Introduction
2. 3.2 Design of Robust Control Strategies
3. 3.3 Some Example Applications of Automatic Feedback Control
4. 3.4 The Role of Kinetics Modeling
5. 3.5 Ideas for a Deeper QbD Approach
6. 3.6 Summary
7. Acknowledgments
8. References
4 Predictive Distributions for Constructing the ICH Q8 Design Space
1. 4.1 Introduction
2. 4.2 Overlapping Means Approach
3. 4.3 Predictive Distribution Approach
4. 4.4 Examples
5. 4.5 Summary and Discussion
6. Acknowledgment
7. References
5 Design of Novel Integrated Pharmaceutical Processes: A Model‐Based Approach
1. 5.1 Introduction
2. 5.2 Problem Description
3. 5.3 Methodology
4. 5.4 Application: Case Study
5. 5.5 Conclusions
6. References
6 Methods and Tools for Design Space Identification in Pharmaceutical Development
1. 6.1 Introduction
2. 6.2 Design Space: A Multidisciplinary Concept
3. 6.3 Integration of Design Space and Control Strategy
4. 6.4 Case Studies
5. 6.5 Conclusions
6. Acknowledgment
7. References
7 Using Quality by Design Principles as a Guide for Designing a Process Control Strategy
1. 7.1 Introduction
2. 7.2 Chemical Sequence, Impurity Formation, and Control Strategy
3. 7.3 Mass Transfer and Reaction Kinetics
4. 7.4 Optimal Processing Conditions
5. 7.5 Predicted Product Quality under Varied Processing Conditions
6. 7.6 Conclusions
7. Acknowledgments
8. Notation
9. References
8 A Strategy for Tablet Active Film Coating Formulation Development Using a Content Uniformity Model and Quality by Design Principles
1. 8.1 Introduction
2. 8.2 Content Uniformity Model Development
3. 8.3 RSD Model Validation and Sensitivity Analysis for Model Parameters
4. 8.4 Model‐Based Design Space Establishment for Tablet Active Film Coating
5. 8.5 Summary
6. Notations
7. References
9 Quality by Design: Process Trajectory Development for a Dynamic Pharmaceutical Coprecipitation Process Based on an Integrated Real‐Time Process Monitoring Strategy
1. 9.1 Introduction
2. 9.2 Experimental
3. 9.3 Data Analysis Methods
4. 9.4 Results and Discussion
5. 9.5 Challenges and Opportunities for PCA‐Based Data Analysis and Modeling in Pharmaceutical PAT and QbD Development
6. 9.6 Conclusions
7. Acknowledgments
8. References
10 Application of Advanced Simulation Tools for Establishing Process Design Spaces Within the Quality by Design Framework
1. 10.1 Introduction
2. 10.2 Computer Simulation‐Based Process Characterization of a Pharmaceutical Blending Process
3. 10.3 Characterization of a Tablet Coating Process via CFD Simulations
4. 10.4 Overall Conclusions
5. References
11 Design Space Definition: A Case Study—Small Molecule Lyophilized Parenteral
1. 11.1 Introduction
2. 11.2 Case Study: Bayesian Treatment of Design Space for a Lyophilized Small Molecule Parenteral
3. 11.3 Results
4. 11.4 Conclusions
5. Appendix 11.A Implementation Using WinBUGS and R
6. Notation
7. Acknowledgments
8. References
12 Enhanced Process Design and Control of a Multiple‐Input Multiple‐Output Granulation Process
1. 12.1 Introduction and Objectives
2. 12.2 Population Balance Model
3. 12.3 Simulation and Controllability Studies
4. 12.4 Identification of Existing “Optimal” Control‐Loop Pairings
5. 12.5 Novel Process Design
6. 12.6 Conclusions
7. References
13 A Perspective on the Implementation of QbD on Manufacturing through Control System: The Fluidized Bed Dryer Control with MPC and NIR Spectroscopy Case
1. 13.1 Introduction
2. 13.2 Theory
3. 13.3 Materials and Methods
4. 13.4 Results and Discussion
5. 13.5 Continuous Fluidized Bed Drying
6. 13.6 Control Limitations
7. 13.7 Conclusions
8. Acknowledgment
9. References
14 Knowledge Management in Support of QbD
1. 14.1 Introduction
2. 14.2 Knowledge Hierarchy
3. 14.3 Review of Existing Software
4. 14.4 Workflow‐Based Framework
5. 14.5 Drug Substance Case Study
6. 14.6 Design Space
7. 14.7 Technical Challenges
8. 14.8 Conclusions
9. References
Index
End User License Agreement

List of Tables

Chapter 05
1. Table 5.1 Structural constraints.
2. Table 5.2 Options after screening by structural constraints.
3. Table 5.3 Results of process simulation at same initial substrates amounts (n_GlcNAc,0 = 1.3 mol, n_Pyr,0 = 2.6 mol) and same enzyme concentrations (C_epi = 1500 U/l, C_ald = 24 000 U/l).
4. Table 5.4 Calculation of the objective function of the feasible alternatives.
Chapter 06
1. Table 6.1 Input variables studied ranges (knowledge space) and output variables acceptable ranges.
Chapter 07
1. Table 7.1 Thresholds according to ICH guidelines.
2. Table 7.2 Henry’s law coefficient as determined using MS and FTIR (bold) and by pressure decay experiments.
3. Table 7.3 Matrix compositions.
4. Table 7.4 Mass transfer coefficients for CO₂ in ethanol and the deprotection reaction matrix.
5. Table 7.5 Summary of 6 degradation kinetic parameters.
6. Table 7.6 Summary of 1 deprotection kinetic parameters.
7. Table 7.7 Calculated deprotection reaction Arrhenius constants.
8. Table 7.8 Matrix compositions used to determine the mass transfer and Henry’s law coefficients in the coupling reaction.
9. Table 7.9 Step 2 matrix calculated k_la values.
10. Table 7.10 Henry’s law coefficient as determined by pressure decay experiments.
11. Table 7.11 Summary of total mole fraction of CO₂ in the reaction system (as a fraction of the total amount of compound 1 at the start of the reaction) for the various CO₂ removal options considered.
12. Table 7.12 Deprotection chemistry process parameters.
13. Table 7.13 Deprotection process acceptable in situ impurity levels.
14. Table 7.14 PAR conditions.
15. Table 7.15 PAR simulation results (area %).
16. Table 7.16 Screening design terms and relative statistical significance.
Chapter 08
1. Table 8.1 Key operational parameters for various size pan coaters.
2. Table 8.2 Process variables for a tablet active film coating process at pilot and commercial scales.
Chapter 09
1. Table 9.1 Formulations used, process stream introduced, and subsequent process events observed via process NIR system (visual observation for the coprecipitation system during the course of adding water into the ternary system of naproxen, Eudragit L100, and alcohol to form a four‐component coprecipitation system).
Chapter 10
1. Table 10.1 Risk assessment of the blending process step using FMEA according to Adam et al. [1].
2. Table 10.2 Design levels for the potentially critical process parameters.
Chapter 11
1. Table 11.1 Marginal distributions of shelf life at 30°C and centered covariable levels.
2. Table 11.2 Summary of the data used for analysis.
Chapter 12
1. Table 12.1 Nominal values of material properties and process/design parameters.
2. Table 12.2 Fundamental material properties and less sensitive adjustable parameters.
Chapter 13
1. Table 13.1 Nominal values of the FBD and MPC tuning parameters.
2. Table 13.2 Initial conditions of the granulated lactose.
3. Table 13.3 Non‐nominal values for the FBD and corresponding MPC tuning parameters.
4. Table 13.4 Initial conditions of the granulated lactose for the non‐nominal case.
Chapter 14
1. Table 14.1 CPPs for processing steps and CMAs for process materials.

List of Illustrations

Chapter 02
1. Figure 2.1 Knowledge pyramid for developing mathematical models.
2. Figure 2.2 Schematic of types of models.
3. Figure 2.3 Approach for developing a design space based on DOE data. (a) Choose experimental design (e.g., full factorial, d‐optimal), (b) conduct randomized experiments, (c) analyze data (determine significant factors), and (d) define design space (e.g., contour surface).
4. Figure 2.4 Approach for defining a design space based on DOE and MVA.
5. Figure 2.5 Approach for developing LV models for feed‐forward control.
Chapter 03
1. Figure 3.1 A photograph of a bench‐scale crystallizer with a jacket for temperature control. Supersaturation in pharmaceutical crystallizers is usually induced by cooling and/or antisolvent addition.
2. Figure 3.2 A photograph of some pharmaceutical crystals produced from a crystallizer that was inadequately controlled, showing potential variations in size and shape.
3. Figure 3.3 (a) α‐Polymorph (metastable form, prismatic shape) and (b) β‐polymorph (stable form, needlelike platelet shape) crystals of L‐glutamic acid. The two polymorphs show different crystal morphologies.
4. Figure 3.4 An example of a potential target CSD and polymorphic form.
5. Figure 3.5 Example simulation results of the spatial distribution of turbulent Reynolds number in a confined DIJ mixer. Such mixers can be used to generate sub 10‐µm crystals for many solute/solvent systems.
6. Figure 3.6 A microfluidic platform that uses evaporation to generate supersaturation in 2–10 µl‐sized droplets in microwells. The rate of evaporation (J), and thereby the supersaturation profile, is set by the dimensions of the channel (cross‐sectional area A and length L) that connects the well with the ambient environment.
7. Figure 3.7 A comparison of induction times measured in wells in a microfluidic platform (crosses) with predictions of a first‐principles model (circles) with nucleation kinetics estimated from the microfluidic data at high supersaturation. A detailed study demonstrated that these nucleation kinetic parameters resulted in accurate predictions of induction times in subsequent experiments.
8. Figure 3.8 A high‐throughput microfluidic chip for protein and pharmaceutical crystallization that enables 144 induction time experiments to be conducted in parallel with different precipitants and different solute concentrations: (a) a micrograph of the microfluidic chip, which contains 8 inlets for pressurized control lines, 48 inlets for different precipitants, and one inlet for the target protein or pharmaceutical, and (b) a micrograph of the microfluidic components used for metering solutions and controlling evaporation of solvent [29]. The target molecule and precipitant are metered into serpentine channels, and the ratio of each component is controlled by the placement of a pneumatic valve. After filling, the solutions are dispensed into a well that is connected to an evaporation channel. The dimensions of the evaporation channel control the rate of evaporation.
9. Figure 3.9 Temperature profiles for seeded cooling crystallizations. Optimal control that ignores uncertainties predicts a 21% improvement in n/s (from 10.7 to 8.5) that can be completely lost due to uncertainties (from 10.7 to 14.6).
10. Figure 3.10 Phase diagram for crystal dissolution, growth, and uncontrolled nucleation showing robust and nonrobust operational trajectories. A control strategy that is insensitive to model uncertainties must keep the operational trajectory between the solubility curve and metastable limit.
11. Figure 3.11 (a) A photograph and (b) a schematic of crystallization laboratory setup with state‐of‐the‐art in situ sensor technologies that enable the development of robust control strategies. FBRM, focused beam reflectance measurement (laser backscattering), and PVM, particle vision and measurement (in situ video microscopy).
12. Figure 3.12 Automatic feedback control of the batch crystallization of a Merck compound: (a) concentration control at constant absolute and relative supersaturation; (b) secondary nucleation monitored by FBRM total counts per second.
13. Figure 3.13 Powder X‐ray diffraction patterns for the metastable α (upper curve) and stable β (lower curve) polymorphic forms of L‐glutamic acid.
14. Figure 3.14 Crystallization phase diagram for L‐glutamic acid. For such a monotropic system, β‐form crystals grow and α‐form crystals dissolve in the region between the solubility curves of the α‐ and β‐forms.
15. Figure 3.15 Crystallization phase diagram for L‐glutamic acid, showing metastable limits at various cooling rates (0.1, 0.5, and 1.0°C/min) with corresponding polymorphic portions of nuclei.
16. Figure 3.16 Controlled operations to manufacture the L‐glutamic acid stable β‐polymorph. The operation begins with a quick cooling from 65 to 25°C to generate nuclei, followed by holding the temperature at 25°C for 1 h to deplete supersaturation to maximize surface area of crystals, then a quick heating to 50°C (while operating between the two solubility curves) to reach a much higher temperature to increase the growth rate of the β‐form crystals. The last step is to apply concentration feedback control (solid lines) to grow larger β‐form product crystals.
17. Figure 3.17 Controlled operations to manufacture a pure batch of metastable α‐form crystals of L‐glutamic acid. α‐Form seeds (75–180 µm in size) are grown to large sizes (270–450 µm in size) by following an operational trajectory between the α‐form solubility curve and the α‐form metastable limit.
18. Figure 3.18 Implementations of (a) temperature control and (b) concentration feedback control for cooling crystallization. The implementations for antisolvent crystallization are similar, with T replaced by % solvent.
19. Figure 3.19 A manufacturing‐scale crystallizer couples phenomena (micromixing, macromixing, nucleation, and growth) over a wide range of length scales (submicron to meter scale).
20. Figure 3.20 A schematic of antisolvent crystallization occurring within a confined DIJ mixer operating at high supersaturation that continuously produces seed crystals that enter a stirred tank. The high supersaturation in the DIJ mixer is generated in a mixing zone where opposing streams from two jets (containing saturated solution and antisolvent, respectively) collide. A feedback control system operates the stirred‐tank crystallizer at lower supersaturation so that only crystal growth (no nucleation) occurs.
21. Figure 3.21 (a) Comparison of a unimodal target CSD (line) with the CSD (dash star) achieved by optimization of continuously varying inlet velocities to a DIJ mixer coupled to an stirred tank (simulation results). (b) The time profile of inlet jet velocities for the simulated achievable CSD, with the jets constrained to have the same inlet velocity.
22. Figure 3.22 (a) Comparison of a uniform target CSD (line) with the CSD (dash star) achieved by optimization of continuously varying inlet velocities to a DIJ mixer coupled to an aging tank (simulation results). (b) The time profile of inlet jet velocities for the simulated achievable CSD, with the jets constrained to have the same inlet velocity.
23. Figure 3.23 A nearly uniform crystal size distribution (CSD) for L‐asparagine monohydrate obtained by coupling a cooling DIJ mixer with a stirred‐tank crystallizer controlled to follow a preset constant supersaturation profile. The product crystals are shown in the photograph (scale bar = 200 µm). The experimentally measured CSD (histogram) is very close to the CSD predicted by Woo et al. [5] (line star) as being the closest that is achievable to a uniform distribution. (The width of the CSD depends on the crystallization system (e.g., compound/solvent combination and growth rate), whereas the overall shape does not.)
Chapter 04
1. Figure 4.1 The overlapping response surface plot appearing in the ICH Q8 Annex.
2. Figure 4.2 Nucleophilic aromatic substitution of aromatic chloride ArCl with CPA.
3. Figure 4.3 The shaded regions bordered by the 0.5 contour lines represent the OMR DS. The shaded regions bordered by the 0.99 contour lines represent the parametric bootstrap and Bayesian posterior predictive DSs.
4. Figure 4.4 The light‐gray regions represent the OMR DS. The black region represents the Bayesian DS with reliability value R = 0.8. The black and darker gray regions combined represent the parametric bootstrap DS with reliability value R = 0.8.
Chapter 05
1. Figure 5.1 Superstructure for reaction–separation configuration of enzymatic processes (symbols: streams F (mol), temperature T (K), pressure P (bar), total number o moles n (mol), separation factors σ (–), binary existence variable ζ [0,1]; subscripts: bottom flow α, top flow β, product p).
2. Figure 5.2 Workflow diagram of the model‐based methodology.
3. Figure 5.3 Synthesis of N‐acetyl‐neuraminic acid from GlcNAc in two steps (A, N‐acetyl‐D‐glucosamine (GlcNAc); B, N‐acetyl‐D‐manosamine (ManNAc); C, pyruvic acid (Pyr); D, N‐acetylneuraminic acid (NeuAc); epimerase (EPi), N‐acylglucosamine‐2‐epimerase (E.C. 5.1.3.8); aldolase (Ald), N‐acetylneuraminic acid aldolase (E.C. 4.1.3.3)).
4. Figure 5.4 Conventional process scheme for chemo‐enzymatic synthesis of Neu5Ac.
5. Figure 5.5 Reaction–separation options for the chemo‐enzymatic synthesis of Neu5Ac.
6. Figure 5.6 Process flowsheet (network) for option 3 (OPR‐CRYST).
7. Figure 5.7 Process flowsheet for option 2 (OPRS‐CHRO).
Chapter 06
1. Figure 6.1 Link between knowledge space, design space, and normal operating ranges.
2. Figure 6.2 Estimated optimum RSD within the range of investigated flow rate and impeller rotation rate values predicted by (a) RSM, (b) Kriging, (c) HDMR, and (d) RSM, Kriging, and HDMR for flow rate of 30 kg/h in 2D.
3. Figure 6.3 Design space (flow rate and impeller rotation rate ranges) of continuous mixer for achieving RSD_max = 10%.
4. Figure 6.4 Design space produced by (a) RSM approach, (b) Kriging approach, and (c) HDMR approach showing the optimum design configurations for each point.
5. Figure 6.5 Schematic representation of the roller compaction process.
6. Figure 6.6 Design space of roller compactor for quality specifications of Table 6.1.
7. Figure 6.7 Normalized by mean values of inlet density, outlet ribbon density, and ribbon thickness for 100 random simulations.
8. Figure 6.8 Comparison of deterministic and stochastic DS with respect to the operating conditions.
9. Figure 6.9 Comparison of DS with and without control of P_h.
10. Figure 6.10 Optimal values of P_h as a control variable based on the solution of problem defined by Equation 6.11.
11. Figure 6.11 Comparison of DS with and without control of flow rate u_in.
12. Figure 6.12 Optimal values of u_in as a control variable based on the solution of problem defined by Equation 6.11.
13. Figure 6.13 Comparison of DS with and without control of angular velocity ω.
14. Figure 6.14 Optimal values of ω as a control variable based on the solution of problem defined by Equation 6.11.
15. Figure 6.15 Stochastic design space when ω is considered a manipulative variable (control strategy 3).
Chapter 07
1. Figure 7.1 Process design and development flow chart.
2. Figure 7.2 Reaction scheme for generating the active pharmaceutical ingredient 3.
3. Figure 7.3 Semagacestat GMP sequence.
4. Figure 7.4 Reaction mechanism for the EDCI‐mediated peptide coupling.
5. Figure 7.5 (a–c) Key impurities that can form during step 2.
6. Figure 7.6 Detailed list of impurities resulting from extended heating of 2.
7. Figure 7.7 Possible pathways for the formation of urea impurity, 4.
8. Figure 7.8 Propagation of 6 and alanine through the step 2 to coupling.
9. Figure 7.9 Impurities associated with HOVal derivatives.
10. Figure 7.10 Impurity purge plot for ent‐6 across the entire semagacestat GMP sequence.
11. Figure 7.11 Reactor schematic.
12. Figure 7.12 Experimental configuration for multidimensional analysis of parallel processes at small scale.
13. Figure 7.13 Adsorption/desorption curve for CO₂ in ethanol (left) and the reaction matrix (right) at 5°C.
14. Figure 7.14 Mass transfer rates in ethanol and the reaction mixture.
15. Figure 7.15 Experimental setup used for determining the k_la at 8000 l scale.
16. Figure 7.16 Desorption of CO₂ from ethanol at 8550 l scale, 40°C, liquid volume = 950 l, gas volume = 7600 l, purge gas flow = 260 l/min. Black arrow indicates secondary axis for partial pressure.
17. Figure 7.17 Elementary chemical expression for intermediate 2 formation and associated degradation impurities.
18. Figure 7.18 Predicted and measured concentrations of 6, 9, and 10 at 56°C.
19. Figure 7.19 Predicted and measured concentrations of 6, 9, and 10 at 80°C (near reflux).
20. Figure 7.20 Predicted and measured concentrations of 1 at 54°C. Compounds 6 and 7 are near 0.
21. Figure 7.21 Predicted and measured concentrations of 1 at 74°C. Compound 1 is near 0.
22. Figure 7.22 Crystal dimensions.
23. Figure 7.23 Coupling reaction kinetic data for the desired product, compound 3, formation, experimental (open circle) versus predictive (line).
24. Figure 7.24 Compound 4 reaction kinetics equipment configuration.
25. Figure 7.25 CO₂ uptake during the urea formation reaction (+ = actual, line = predicted).
26. Figure 7.26 Comparison of actual and predicted urea concentration at the end of the reaction when spiked with various amounts of CO₂ in the headspace.
27. Figure 7.27 Predicted impurity levels as a function of degassing temperature and headspace degassing flow rate.
28. Figure 7.28 Predicted individual impurity levels as a function of degassing temperature and headspace degassing flow rate.
29. Figure 7.29 Effect of vacuum cycles on concentration of CO₂ in the reactor prior to the coupling reaction (option 1).
30. Figure 7.30 Effect of pressure and vacuum cycles on concentration of CO₂ in the reactor prior to the coupling reaction (option 2).
31. Figure 7.31 Effect of headspace purge and vacuum cycles on concentration of CO₂ in the reactor prior to the coupling reaction (option 3).
32. Figure 7.32 Effect of headspace purge, vacuum, and pressure cycles on concentration of CO₂ in the reactor prior to the coupling reaction (option 4).
33. Figure 7.33 Effect of four pressure cycles on concentration of CO₂ in the reactor prior to the coupling reaction (option 5).
34. Figure 7.34 Mass fraction of urea, compound 4, in the product as a function of CO₂ concentration in solution at the start of the coupling reaction and as a function of the hydroxyvaline addition rate.
35. Figure 7.35 Contour plot of total impurity level (TRS) as a function of the reaction time and temperature.
36. Figure 7.36 Individual impurities levels at the end of the step 2a reaction as a function of reaction temperature and reaction time.
37. Figure 7.37 Acceptable operational region for the deprotection reaction as a function of reaction time and reaction temperature. Areas outside of the filled region do not indicate an edge of failure, but instead a limit of knowledge. Unreacted starting material is represented by the region in gray, Compound 8 in excess by the region in white at the top right.
38. Figure 7.38 Fit model summary for screening design.
39. Figure 7.39 Screening design multivariate experiment contour plots.
Chapter 08
1. Figure 8.1 Schematic of an active film coated tablet.
2. Figure 8.2 The linear relationship between the actual API potency of coated tablets and the theoretical amount of API in the coating liquid sprayed.
3. Figure 8.3 Schematic of a perforated pan coating process.
4. Figure 8.4 Analogy between a perforated pan coater and ideal chemical reactors.
5. Figure 8.5 Schematic of the RSD model development.
6. Figure 8.6 Links between model parameters and measurable conditions of the system.
7. Figure 8.7 Installation of a Pro‐Watch^® camera in a commercial pan coater.
8. Figure 8.8 Tablet imagine (number density) on the surface of the pan coater.
9. Figure 8.9 Tablet velocity vector on the surface of the pan coater.
10. Figure 8.10 Representative tablet velocity versus pan speed in a commercial pan coater.
11. Figure 8.11 Empirical correlation of the tablet velocity, peripheral velocity (pan speed), and pan radius and particle size.
12. Figure 8.12 Set up of the OSIER SprayWatch system.
13. Figure 8.13 A typical image of three‐pulse backlight droplets for velocity and size analysis.
14. Figure 8.14 A typical spray image from the profile camera.
15. Figure 8.15 Effect of AA/PA ratio on the spatial distribution of spray volume flux and spray zone width (SLPM, standard liter per minute).
16. Figure 8.16 Comparison between model predicted and experimental content uniformity.
17. Figure 8.17 Distribution of potency predicted by the model and compared with the measured values in the 200 kg batch.
18. Figure 8.18 Effects of spray zone width on tablet content uniformity.
19. Figure 8.19 Effects of tablet velocity on tablet content uniformity.
20. Figure 8.20 Effects of tablet size on tablet content uniformity.
21. Figure 8.21 Effects of pan load on spray zone width.
22. Figure 8.22 Effects of pan load on tablet content uniformity.
23. Figure 8.23 Process map for tablet active film coating.
24. Figure 8.24 Correlation between DOE factors and RSD model parameters.
25. Figure 8.25 Design space for pan coating by model.
26. Figure 8.26 Schematic DOE to establish a design space for a tablet film coating process with the aid of the RSD model.
27. Figure 8.27 A decision tree illustrating how to scale up a tablet active film coating process using a perforated pan coater and with the aid of the RSD model.
Chapter 09
1. Figure 9.1 Schematic diagram of experimental setup and process flow with an integrated PAT monitoring system for API (naproxen)/polymer coprecipitation process using in‐line process NIR and online turbidity measurement simultaneously.
2. Figure 9.2 PCA‐based process trajectory for binary powder blends of naproxen and Eudragit L100. The binary powder blends were made by mixing naproxen and Eudragit L100 with various weight ratios as shown in the figure by E/N values. After mixing well, the blends were transferred to vials that were then scanned by FOSS NIRS offline. The PCA was conducted on NIR spectra of the binary blends with various E/N values.
3. Figure 9.3 PCA‐based process trajectory for binary liquid mixtures of alcohol and water with various volume ratios at room temperature. The binary liquid mixtures were made by mixing alcohol and water with various volume ratios. After mixing well, the blends were measured by real‐time Brimrose in‐line NIR spectroscopy. The PCA was conducted on NIR spectra of the binary blends with various alcohol/water ratios.
4. Figure 9.4 Graphical illustration of the movement of equilibrium line and metastable line zone. Concentrations C₀, C₁, and C₂ represent the solubility of solute at temperature T_lim, T₁, and T_2, respectively.
5. Figure 9.5 Three‐dimensional map of the process state over the entire process course of adding water to the ternary system. An inflection point occurs among process spectra window between spectral numbers #60 and #67. New phase is detected between spectral numbers #60 and #67, which maybe an indication of nucleation and growth.
6. Figure 9.6 PCA score plot for process NIR spectra of the process batch started with formulation components of 4.7617 g naproxen and 1.205 g Eudragit L100.
7. Figure 9.7 Turbidity profile during the process of sequentially adding water to the ternary system of API (0.1320 g), Eudragit L100 (6.0970 g), and alcohol (320 ml) to form the coprecipitate.
Chapter 10
1. Figure 10.1 Systematic QbD approach, modified based on Schousboe and Hirsh [10].
2. Figure 10.2 Graphical representation of the considered double‐cone blender.
3. Figure 10.3 Evolution of RSD value for binary mixtures with 10, 30, and 50 wt% API content, respectively.
4. Figure 10.4 Evolution of Lacey’s index for binary mixtures with 10, 30, and 50 wt% API content.
5. Figure 10.5 Blending unit operation experimental space, containing various possible design spaces as subset (rectangles on the right).
6. Figure 10.6 Schematic representation of a coating apparatus. A tablet emerges on the top, traverses the spray zone, and reenters the tablet bed at the bottom. In this design, drying air is introduced and extracted through the perforated drum.
7. Figure 10.7 Basic steps in an aqueous coating process: application of a partial film by spraying, evolution of the film on the surface, and finally drying of the partial coating.
8. Figure 10.8 Graphical representation of the simulation setup. The rectangular box marks the simulation region; part of the front walls is set transparent. The dark gray region in the middle of the top wall denotes the air inlet. In the center, the tablet is white and the spray droplets are black. The velocity distribution of the airstream is shown in the rectangular cut surface surrounding the tablet.
9. Figure 10.9 Evolution of the coating film on the top surface of the tablet. The coverage increases during the spray phase and decreases again during drying until the shown final state.
10. Figure 10.10 Number‐weighted diameter distribution of spray droplets generated with three atomizing air pressure settings.
11. Figure 10.11 Time evolution of mean film thickness over the coating event for constant spray pressure level 0.
12. Figure 10.12 Time evolution of mean film thickness over the coating event for constant temperature level 0.
13. Figure 10.13 Time evolution of RSD over the coating event for constant spray pressure level 0.
14. Figure 10.14 Time evolution of RSD over the coating event for constant temperature level 0.
15. Figure 10.15 Contour plot of the mean film thickness, averaged over the entire tablet after the spray phase, as a function of atomizing air pressure and air temperature. The rectangle shows a possible choice of the design space.
16. Figure 10.16 RSD of the film thickness over the whole tablet after the spraying phase as a function of atomizing air pressure and air temperature.
17. Figure 10.17 RSD of the film thickness over the entire tablet after the drying phase as a function of atomizing air pressure and air temperature. The dashed region is the design space in the case when only coating thickness is considered, and the solid region is a possible final design space with RSD optimized and over‐wetting considered.
Chapter 11
1. Figure 11.1 Degradation model.
2. Figure 11.2 Representative potency time profile.
3. Figure 11.3 Mixing of parameter θ₁.
4. Figure 11.4 Distribution of predicted shelf life.
5. Figure 11.5 Wireframe representation of design space.
6. Figure 11.6 Contour plot representation of design space.
Chapter 12
1. Figure 12.1 Compartmentalized model of the granulator.
2. Figure 12.2 Time profiles for (a) average diameter, (b) distribution width, (c) average moisture content, and (d) average bulk density for different spray nozzle positions under start‐up conditions.
3. Figure 12.3 A block diagram of the process and its control objectives.
4. Figure 12.4 A comprehensive systems representation of the granulation process.
5. Figure 12.5 Proposed design and control configuration.
Chapter 13
1. Figure 13.1 MPC computed trajectories of process variables.
2. Figure 13.2 Schematic of a fluidized bed dryer.
3. Figure 13.3 Design space of the air inlet temperature and flow rate for proper operation of the FBD and desired quality of the dried material.
4. Figure 13.4 Diagram of the system integration for sensing and control.
5. Figure 13.5 Moisture profile predicted by the NIR under MPC and nominal model values.
6. Figure 13.6 Manipulated variables computed by the MPC (top) and its values applied through local PIDs (bottom), nominal model values, run #1, and #2.
7. Figure 13.7 Moisture profile predicted by the NIR under MPC for non‐nominal model values.
8. Figure 13.8 Manipulated variables computed by the MPC (top) and process its values applied through local PIDs (bottom), non‐nominal model values, run #1.
9. Figure 13.9 A continuous fluid bed dryer conceptual schematic.
Chapter 14
1. Figure 14.1 Pharmaceutical quality system [3].
2. Figure 14.2 The knowledge pyramid.
3. Figure 14.3 Workflow for the project to control impurities in a reaction system.
4. Figure 14.4 Workflow for the laboratory study for reaction kinetics.
5. Figure 14.5 Example of experiment workflow.
6. Figure 14.6 Example of computational workflow.
7. Figure 14.7 Diagram of manufacturing process.
8. Figure 14.8 Define design space for tablet compression.
9. Figure 14.9 Critical process variables and critical material attributes.
10. Figure 14.10 Workflow for the simulation‐based determination of design space.
11. Figure 14.11 Workflow for the experimental verification of the design space.

List of Contributors

Siegfried Adam
Research Center Pharmaceutical Engineering GmbH
Graz, Austria

Prabir K. Basu
QbD Consultant
Mt Prospect, IL, USA

Richard D. Braatz
Department of Chemical Engineering
Massachusetts Institute of Technology
Cambridge, MA
and
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA

Christopher L. Burcham
Small Molecule Design and Development
Eli Lilly and Company
Indianapolis, IN, USA

Shih‐Ying Chang
Drug Product Science and Technology
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Sharmista Chatterjee
Office of New Drugs Quality Assessment
Food and Drug Administration
Silver Spring, MD, USA

Wei Chen
Drug Product Science and Technology
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Divyakant Desai
Drug Product Science and Technology
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Mitsuko Fujiwara
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA

Rafiqul Gani
Department of Chemical and Biochemical Engineering
Technical University of Denmark
Lyngby, Denmark

A. Giridhar
Davidson School of Chemical Engineering
Purdue University
West Lafayette, IN, USA

Li May Goh
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA

Neil Hodnett
GlaxoSmithKline Pharmaceuticals
Brentford, UK

Marianthi G. Ierapetritou
Department of Chemical and Biochemical Engineering
Rutgers University
Piscataway, NJ, USA
Fernando J. Muzzio
Department of Chemical and
Biochemical Engineering
Rutgers University
Piscataway, NJ, USA

Mo Jiang
Department of Chemical Engineering
Massachusetts Institute of Technology
Cambridge, MA, USA

G. Joglekar
Davidson School of Chemical Engineering
Purdue University
West Lafayette, IN, USA

Nicholas C. S. Kee
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA
and
National University of Singapore, Block E5
and
Institute of Chemical and Engineering Sciences
Jurong Island, Singapore

Paul J. A. Kenis
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA

Mansoor A. Khan
Division of Product Quality Research (DPQR, HFD‐940)
OTR, Office of Pharmaceutical Quality, Center for Drug Evaluation and Research, Food and Drug Administration
Silver Spring, MD, USA
Current affiliation
Rangel College of Pharmacy, Texas A&M University Health Science Center
College Station, TX, USA

Johannes G. Khinast
Research Center Pharmaceutical Engineering GmbH
and
Institute for Process and Particle Engineering
Graz University of Technology
Graz, Austria

San Kiang
Drug Product Science and Technology
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Mark LaPack
Mark LaPack & Associates Consulting, LLC, Lafayette IN, USA

David LeBlond
Applied Statistics consultant
Wadsworth, IL, USA

Kevin Lief
GlaxoSmithKline Pharmaceuticals
Brentford, UK

Olav Lyngberg
Chemical Development, Research and Development
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Joseph R. Martinelli
Small Molecule Design and Development
Eli Lilly and Company
Indianapolis, IN, USA

Neil McCracken
Small Molecule Design and Development
Eli Lilly and Company
Indianapolis, IN, USA
and
Bioproduct Research and Development
Eli Lilly and Company
Indianapolis, IN, USA

Linas Mockus
Davidson School of Chemical Engineering
Purdue University
West Lafayette, IN, USA

Christine M. V. Moore
Global CMC Policy
Merck, Inc.
Philadelphia, PA, USA

Fani Boukouvala
Department of Chemical and Biochemical Engineering
Rutgers University
Piscataway, NJ, USA

Zoltan K. Nagy
School of Chemical Engineering
Purdue University, Forney Hall of Chemical Engineering
West Lafayette, IN, USA

Steven L. Nail
Pharmaceutical Development
Baxter Pharmaceutical Solutions, LLC
Bloomington, IN, USA

Moheb M. Nasr
GlaxoSmithKline Pharmaceuticals
Brentford, UK

Luis Obregón
Pharmaceutical Engineering Research laboratory, Chemical Engineering Department
University of Puerto Rico at Mayaguez
Mayaguez, Puerto Rico

Tim Paul
Pharmaceutical Development
Baxter Pharmaceutical Solutions, LLC
Bloomington, IN, USA

Nathan Pease
Pharmaceutical Development
Baxter Pharmaceutical Solutions, LLC
Bloomington, IN, USA

John J. Peterson
GlaxoSmithKline Pharmaceuticals
Brentford, UK

Leonel Quiñones
Pharmaceutical Engineering Research laboratory, Chemical Engineering Department
University of Puerto Rico at Mayaguez
Mayaguez, Puerto Rico

Rohit Ramachandran
Department of Chemical and Biochemical Engineering
Rutgers, The State University of New Jersey
Piscataway, NJ, USA

Gintaras V. Reklaitis
Davidson School of Chemical Engineering
Purdue University
West Lafayette, IN, USA

Alicia Román‐Martínez
Department of Chemical and Biochemical Engineering
Technical University of Denmark
Lyngby, Denmark
and
Facultad de Ciencias Químicas
Universidad Autónoma de San Luis Potosí
San Luis Potosí, Mexico

Christine Seymour
Global Regulatory Chemistry and Manufacturing Controls
Pfizer Inc.
Groton, CT, USA

Daniele Suzzi
Research Center Pharmaceutical Engineering GmbH
Graz, Austria

Reginald B. H. Tan
National University of Singapore, Block E5
and
Institute of Chemical and Engineering Sciences
Jurong Island, Singapore

Joshua D. Tice
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL, USA

Gregor Toschkoff
Research Center Pharmaceutical Engineering GmbH
Graz, Austria

Carlos Velázquez
Pharmaceutical Engineering Research laboratory, Chemical Engineering Department
University of Puerto Rico at Mayaguez
Mayaguez, Puerto Rico

Jennifer Wang
Drug Product Science and Technology
Bristol‐Myers Squibb Company
New Brunswick, NJ, USA

Xing Yi Woo
The Jackson Laboratory
Bar Harbor, ME, USA

John M. Woodley
Department of Chemical and Biochemical Engineering
Technical University of Denmark
Lyngby, Denmark

Huiquan Wu
Division of Product Quality Research (DPQR, HFD‐940)
OTR, Office of Pharmaceutical Quality, Center for Drug Evaluation and Research, Food and Drug Administration
Silver Spring, MD, USA
Current affiliation
Process Assessment Branch II, Division of Process Assessment 1
Office of Process and Facilities, Office of Pharmaceutical Quality, Center for Drug Evaluation and Research, Food and Drug Administration
Silver Spring, MD, USA

Mohammad Yahyah
GlaxoSmithKline Pharmaceuticals
Brentford, UK

Lifang Zhou
Department of Chemical Engineering
Massachusetts Institute of Technology
Cambridge, MA, USA

Charles F. Zukoski
Department of Chemical and Biomolecular Engineering
University of Illinois at Urbana‐Champaign
Urbana, IL
and
Department of Chemical and Biological Engineering
University at Buffalo
Buffalo, NY, USA