New Technologies for Process Research and Development: From in situ Spectroscopy to

1. New Technologies for Process Research and Development: From in situ Spectroscopy to Laboratory Automation Joel M. Hawkins Chemical Research and Development Pfizer Global Research and Development Siegfried Symposium Zürich, Switzerland October 14, 2004

2. New Technologies forProcess Research and Development • Technology in the Project Laboratory • In situ FT IR + heat flow + HPLC • “Manual” use of “engineering tools” • Automation for Process R&D • Collaborative project chemistry • Technology development and dissemination • New in situ spectroscopy (PAT) • In situ ATR UV-vis profiling of a “heterogeneous” palladium catalyzed Heck coupling

3. I. Technology in the Project Laboratory

7. But poor selectivity with one equiv of the expensive aldehyde under Evans’ conditions

8. New Titanium Aldol Conditions Developed which work with One Equivalent of Aldehyde

10. What is happening chemically here for the selectivity to jump up? Correlated HPLC, in situ FT IR, and Calorimetry …

11. In situ FT IR Profiles Correspond to: Rearrangement of the Ti aldolate Aldol Two Thermal Events 16 kcal/mol 5 kcal/mol

12. free bound aldol reaction 16 kcal/mol rearrangement of the Ti aldolate 5 kcal/mol

13. Aldolate Rearrangement Upon Warming Gives Higher Selectivity first aldolate enolate second aldolate

14. Pre-quench with a Lewis base to saturate coordination at titanium before the protic quench. Aldolate rearrangement upon warming gives higher selectivity. Observed by correlating HPLC, in situ FT IR, and calorimetry studies … and applied to a multi-hundred kg stereoselective process.

15. II. Automation for Process R&DBlend of Technology Development & Project Chemistry Pay Back (ROI) manually Time project (service) project (labs) break in

16. Electronic lab notebook-automation interface Overcome data analysis and visualization bottleneck Universal instrument ’wrapper’ to minimize number of packages Software Matrix of Automation in Pharmaceutical Process R&D Screen Optimize Characterize Many variables Many experiments Small scale Simulate flasks Key variables Fewer experiments Maximize information Simulate large scale Process safety Reaction engineering Process validation Pilots Solution Chemistry Many variables Many experiments Small scale Mixing still important Key variables Fewer experiments Maximize information Simulate large scale Process safety Reaction engineering Process validation Pilots Pressure Reactions Drug substance salt selection Intermediate resolution & purification Yield Purity Robustness Polymorph control Particle size Filtration rate Process validation Crystallizations

17. Software Matrix of Automation in Pharmaceutical Process R&D Screen Optimize Characterize 1 3 Solution Chemistry 2 Pressure Reactions Crystallizations

19. Base 1

20. Base 2

21. Optimization and Robustness Testing Starting point • 5 mol% Pd2(dba)3 / 2 equiv. olefin / 10 equiv. TEA / i-PrOH / 85 oC Goal to reduce stoichiometry of: • Pd (to ease Pd removal) • olefin (for cost) • TEA (for workup) Screened on the Anachem SK233

22. Screening on the Anachem • Decreased stoichiometries: • Pd2(dba)3 from 5 to 1 mol% • olefin from 2 to 1.1 equiv. • TEA from 10 to 5 equiv. • Stressed to test robustness • Water can be tolerated (at least to 20%). • Residual phase transfer catalyst not a problem – in fact, necessary! • Residual 2-MeTHF retards reaction, 10-20% tolerated. • Scaled successfully 3 x 50 kg

23. Specialist Group Project Lab (or walk up) Is the same technology optimal for both specialist groups and projects labs? Screen Optimize Characterize Solution Chemistry Pressure Reactions Crystallizations

24. Specialist Group Project Lab (or walk up) Is the same technology optimal for both specialist groups and projects labs? Screen Larger number of reactions in parallel (e.g. 10) Temperature programming Automated liquid additions and HPLC sampling via xyz robot Smaller number of reactions in parallel (e.g. 4) Temperature programming Manual operations, limited automation Reaction visibility / Simple on line analytics Solution Chemistry

25. Chemistry Screening Project Labs Profile Individual Temperatures Touch Screen Control Argonaut AS 2410

26. Chemistry Screening Project Labs Profile Heat Output Profile Internal Temperatures Argonaut AS 2410

28. Simple in situ Measurement Related to Extent of Reaction Pressure Transducer Hydrogen Uptake PV = nRT

29. “Out of the Box” 5 equiv. H2 3 equiv. H2

31. X X

33. Follow Kinetics Inverse Order Zero Order

34. Donna Blackmond, University of Hull

36. (3) Solution Reaction Optimization: Mettler MultiMax • Four Reactors • Heating and Cooling • Temperature Profiles • “Easy Calorimetry” from Tr-Tj • “A ‘Rate Meter’ for Every Flask”

37. Simple in situ Measurement Related to Extent of Reaction Two Accurate Temp. Probes Heat Flow Q = UA(Tr-Tj) Rate r = Q / V(-HR)

38. Qualitative Rate Analysis

39. Optimization of TMSCl Stoichiometry and Add. Time From 2.3 equiv. to 1.0 equiv. of TMSCl, 40 min. addition to avoid accumulation

40. Simple Test Case Diels-Alder Reaction Second Order First Order in Diene First Order in Dienophile Homogeneous Clean Good Rate Model Semi-Batch Reaction Fit k to shape of Tr-Tj curve Scale factor -HR/U Simple Numerical Methods (Excel) Use k to Predict Yield vs. Time for all Stoichiometries and all Addition Rates Quantitative Rate Analysis and Kinetics

41. Variable Addition Rates Dienophile dose 23 min 46 min Modeled Actual Calculated Actual 69 min 92 min Calculated Actual Calculated Actual Used First k to Predict Result of Different Add. Rates: Good Fits for Tr-Tj

42. Predicted Compositional Data: Diene + Dienophile (dosed) Diels-Alder Adduct 23 min addition 46 min addition 69 min addition 92 min addition

43. Simple Chemistry Known Kinetics Heat Flow Heat Flow continuous (seconds)  rate all events superimposed IR less frequent (minutes)  concentrations multiple peaks / component often overlapping peaks HPLC infrequent (10’s of minutes)  concentrations can show trace components one peak / component often one component / peak ? “Real” Chemistry Unknown Kinetics Simultaneous Mechanisms

44. Compositional Data fromHeat Flow / Kinetics, FT IR, and HPLC

45. Dissemination of Automated Parallel Lab Reactors Control experimental parameters Mimic scale up Minimize extraneous variables Collect more data, e.g. calorimetry: “Rate meter” Safety data during route development Shared back plane for parallel reactions a series for optimization or totally independent Greater Quantity and Quality of Data Argonaut AS 3400

46. AS 3400 Data for Diacid Reduction Heat Flow (W/L) Add’n of Diacid calibration macro Add’n of BF3 Heat Flow Rxn Temp Heat of Reaction (KJ)

47. Heat Flow for Chemists for Preliminary Safety Data and as a “Rate Meter” Argonaut AS3400

48. Dose controlled exotherm when base added at 70C

49. Argonaut AS 2410 Simple Qualitative Calorimetry base addition

50. Argonaut AS 2410 Simple Qualitative Calorimetry Variable temperature ramps See Tr-Tj on the screen Exotherms during base addition and while heating