From Lab to Production

Transcription

1 From Lab to Production "In-situ FTIR monitoring of solvent- drug matrices during displacements and crystallizations of potentially hazardous intermediates." Kyle Leeman Pfizer Inc. Groton, Ct.

2 Reasons for exploring PAT Process Understanding Single point vs. continuous monitoring Rxn mechanisms/kinetics Hidden Intermediates Many recent intermediates/apis required additional containment Additional safety measures were implemented Worker exposure became a larger concern Time Taking too long to obtain analytical results Many hours wasted cooling/warming tanks to pull reaction samples Decision Monitor processes (Rxn( Rxn,, Displacements, Crystallizations)

3 Objective Apply PAT at scale Focus on FTIR Goal Quantitative analysis Applications explored Reaction endpoints Solvent ratios in displacements Solvent/H 2 O ratios during crystallizations Product concentrations during crystallizations

4 On-line vs. Off-line analysis Online testing through FTIR technology prevents workers from being exposed to harmful intermediates, in which they would have to be outfitted in PPE, as shown above.

5 On-line monitoring of product crystallizations Must initially scan product, solvents and/or H 2 O to determine feasibility Find distinct regions Evaluate sensitivity Analyze standards to generate QuantIR curve Create model to quantitate H 2 O/solvent levels during crystallization Monitor crystallizations and report solvent /H 2 O ratios during real time

6 Isolation conditions for highly potent intermediate Once cyclization is complete, the reaction concentration is atmospherically striped to 5.2L/Kg and cooled to 20-25C 25C to effect crystallization. The slurry is then treated with 15.2L/Kg H 2 O over a period of thirty minutes. The target isolation conditions are ~25% Isopropanol in H 2 O with an overall concentration of roughly 20-22L/Kg. 22L/Kg. Note: Isopropanol/H 2 O ratio is critical to a successful crystallization

7 2-propanol (IPA) and H 2 O Bands These regions will be used to determine solvent-h 2 O ratios during crystallizations of intermediate IPA 946 cm -1 H 2 O 1640 cm -1

8 IR spectra depicting H 2 O additions to a H 2 O/IPA solution H 2 O concentrations range from 50-80% IPA concentrations range from 50-20% Spectra show increase of H 2 O band and decrease of IPA band after each H 2 O addition H 2 O peak IPA peak

9 Crystallization spectra analyzed versus standard IPA/H 2 O QuantIR calibration curve %IPA %IPA Actual %H2O %H2O Actual % Comparison of Actual and Calculated addition y = x R 2 = y = x R 2 = %IPA %IPA Actual %H2O %H2O Actual y = x R 2 = y = x R 2 =

10 Spectra were collected of H 2 O additions to a H 2 O/IPA solution containing 10g product. The H 2 O concentration was increased from 50% to 80%, forcing the material to crystallize from the solution. H 2 O peak Product region IPA peak

11 Product in IPA/H 2 O Solution 5-50mg/mL Product dilutions in IPA/H 2 O solution Close-up of product band at 1225 cm -1

12 QuantIR data for product concentration QuantIR data from known concentrations of the product in H 2 O/IPA solution Region tested 1235 cm -1 to 1215 cm -1 Concentrations ranged from 50mg/ml to 5mg/ml Average percent error was 2.86%

13 Product Solubility Concentration of product in IPA/H 2 O was compared to known solubility values, determined at ºC Data generated from 10g lab scale crystallization (No sampling required) Product Solubility Crystallization Result 15 Solubility Value H2O:IPO Ratio

14 New project: Crystallization in 2-β Ethanol / H 2 O Project intermediates potential safety issues Tested positive for mutagenicity test Sensitizer Can solvent / H 2 O ratio be tested on-line during real time Many lab runs performed to determine appropriate regions to monitor In addition, investigate determination of product concentration during crystallization

15 Isolation conditions for highly potent intermediate Once the alkylation is complete, DCM is displaced with 2β-ethanol2 and concentrated to ~6.25L/Kg The solution is adjusted to 9-11%water(w/w), 9 seeded and stirred overnight to initiate granulation Once a seed bed is established, The slurry is adjusted to 28-32% water(w/w) and granulated for several hours Note: Ethanol/H 2 O ratio is critical for a successful crystallization

16 IR spectra of H 2 O additions to product rich 2β ethanol solution. As H 2 O amounts increase, product precipitates out of solution. Can monitor H 2 O, ethanol, and product concentrations. Saves Time and Resources, No worker exposure Ethanol 885 cm - 1 H 2 O 1640cm -1 Product cm -1

17 Determination of solvent/h 2 O ratios by IR during a crystallization of API. Results calculated during real time ; thus minimizing sampling and exposure Ethanol/Water ratios for product crystallizations % # of H 2 O additions %Ethanol(Know n) %Ethanol(Calc) %H2O(Know n) %H2O(Calc)

18 API standards in 2β Ethanol (Used to generate QuantIR curve) 108 mg/ml 96 mg/ml 84mg/ml 72 mg/ml 60 mg/ml 48 mg/ml 36 mg/ml 24 mg/ml 12 mg/ml

19 Evaluation of product concentrations on-line. In-situ IR can determine appropriate solvent/h 2 O ratios for optimum crystallizations. PRODUCT CRYSTALLIZATION (IR vs HPLC) mg/ml LC IR %H2O

20 ProcessIR in pilot plant

21 Scale Up Pilot Plant Determination of H 2 O during crystallization in 2β ethanol (15 45%) 44.6% 40.5% 33.8% 29.2% 22.3% 14.7% Standards generated in tank prior to run Using ProcessIR Minimizes error

22 Monitoring H 2 O additions during the crystallization 2 nd H 2 O addition 1 st H 2 O addition FTIR was able to monitor H 2 O levels during real time Can determine when desired H 2 O content is achieved

23 Real Time plot vs. External KF FTIR real time profile 2 nd addition 30.2% off-line 29.6% 1 st addition 10.8% off-line 9.5% Initial - 0% Off-line 0% Results show good agreement throughout crystallization

24 Solvent displacements Frequently displacements are required to isolate or proceed with subsequent chemistry In-situ IR can monitor these solvent ratios throughout the displacement. Results are calculated during the displacement Eliminating worker exposure Decreasing analysis time Removal of the initial solvent is critical. IR has the capability of monitoring these mixtures down to low levels (0.1%)

25 Typical Solvent displacement Product rich mixture in 5 volumes of toluene 20 volumes methanol added and vacuum distilled down to 5 volumes An additional 10 volumes added and distilled down to 5 volumes Final mixture should contain ~1% toluene

26 IR monitoring of final displacement MeOH 1023cm -1 Toluene 733cm -1 96% MeOH 97.2% MeOH 4% Toluene 1.3% Toluene

27 Comparisons with off-line testing Sample %MeOH (IR; on-line) %MeOH (GC; off-line) %Toluene (IR; on-line) %Toluene (GC; off-line) After 1 st strip After 2 nd strip Final strip Equivalent results Off-line Analysis time 1hour On-line Ratios determined during the distillation

28 Scale Up EDSF Determination of toluene during displacement in methanol Toluene standard spectra 1 50% in methanol Out of plane =C-H bend at 733cm -1 Standards generated in tank prior to run Using ProcessIR Minimizes error

29 Monitoring of Displacement in EDSF FT IR profile for solvent exchange Toluene 733cm -1 Real-time analysis of Toluene content ~18% - on-line FTIR 17.9% - off-line GC ~3% - on-line FTIR 2.6% - off-line GC Comparable results with FTIR. Saves time while limiting exposure

30 Robustness of QuantIR model in Pilot Plant Due to safety concerns and miscibility must perform displacement between methylene chloride and acetonitrile Neat solutions of starting material presented a high thermal hazard Use FTIR to monitor the solvent exchange and determine when CH 2 CL 2 was below acceptable levels (< 2%) Demonstrate robustness of QuantIR model Use on multiple runs 5-66 weeks between analysis

31 FTIR was able to monitor displacement and determine solvent amounts present in lab pilot ACN 1375cm -1 CH 2Cl cm -1

32 Real time profile and comparison vs. off-line direct inject GC 85% ACN by FTIR 77.1% by GC 19% CH 2 CL 2 by FTIR 17.9% by GC 90% ACN by FTIR 89.4% by GC 3.6% CH 2 Cl 2 by FTIR 2.6% by GC Due to the success in the lab, technology was transferred to pilot plant

33 Pilot plant Real time results Run 1 vs Run 2 Run #1 CH 2 Cl cm -1 Run #2 CH 2 Cl cm % CH 2 Cl 2 Sample pulled for off-line testing (2.1% by GC) %CH 2 Cl 2 % CH 2 Cl 2 2.2% CH 2 Cl 2 -Sample pulled for off-line testing (2.3% by GC) Results calculated using same QuantIR file Demonstrates robustness of model

34 Through in situ IR monitoring we can: Quantitatively detect ratios of solvent/h 2 O matrices Crystallizations Displacements Inversely monitor crystallization by evaluating solution spectra Results are determined real time, therefore: Reducing worker exposure Increase process efficiency Eliminate off-line testing Increase yield

35 Future Plans Investigate fiber optics to improve robustness. Improved transfer of data Lab Plant (Universal application) Determine feasibility of quantitative analysis for reaction completions. Incorporate with Lasentec to provide additional data during crystallizations solubility supersaturation particle size Use technique solely to proceed forward.

36 Acknowledgements Crystallization work Matt Weekly / Anne Serdakowski Solvent Displacements Greg Withbroe Bob Handfield / Bob Mclaughlin Pilot plant/edsf engineers Paul Ahlijanian; Sandeep Kedia Scott Lennon / Tom Staigers Technical support John Tucker Cliff Meltz