Using Dynochem to determine a suitable sampling endpoint in a DoE. David Place.
0 likes711 views
The document discusses using Dynochem software to determine suitable sampling endpoints for design of experiments (DOEs) investigating chemical reactions. It provides a case study of a Finkelstein alkylation reaction where an impurity forms. Dynochem is used to fit rate constants and activation energies to the reaction mechanism. This allows simulating different experimental conditions to identify suitable reaction times that control impurity formation before committing resources to a DOE. The kinetic model can then refine the factor ranges investigated in the DOE to efficiently establish critical process parameters.
Using Dynochem to determine a suitable sampling endpoint in a DoE. David Place.
- 1. Using Dynochem to determine a suitable sampling endpoint in a DOE David W. Place, Ph. D. 401 N Middletown Rd B222/2149 Pearl River, NY 10965 May 13-14, 2009
- 2. Outline I. Comments on DoE Assessment Process II. Case Study: Finkelstein activated alkylation Establish control over impurity formation that carries through to API III. Importance of sampling endpoint Understand kinetics in order to remove time as a factor from the DoE IV. Data Fitting: Establishing k’s and Ea’s V. Simulating Alternate Design Points Refine Factor/CPP ranges based on Dynochem solved kinetic model VI. Summary 2 D. Place
- 3. I. DOE Investigation Assessment process Increasing Process Predictability Fractional Response Reproducibility Kinetic Factorial Surface Model Assessment Assessment DOE DOE Validation that Understanding of Establish/identify Generate predictive Parameters NOT Factor ranges to Most important CPPs Equation for CQA investigated are establish suitable And their rank order/ based on important being controlled process endpoints interactions CPPs 3 D. Place
- 4. II. Case Study: Finkelstein activated alkylation N y NaI R N R Cl + x 50-82 C N N H z parts Solvent “substrate” “amine” “product” Parts Experiment Amine NaI Temp Solvent mol equiv mol equiv degC mL/g A (low) 2 0 50 4 B (centerpoint) 3.5 0.5 66 5.5 C (repeat) 3.5 0.5 66 5.5 D (High) 5 1 82 7 Issue: Reaction Conditions lead to formation of a quaternary salt (0.1- 2%) impurity that carries through into the API and is difficult to remove. Approach: Use a Fractional Factorial Res V design to determine critical process parameters (CPPs) to control Quat Salt formation Problem: Reaction mixture is heterogeneous requiring sacrificial quench of entire reaction mixture to determine impurity profile 4 D. Place
- 5. Dynochem Models used n Model: Dynochem’s Yield loss from side reactions (batch) n Data: HPLC assay data for substrate and product converted to mmol via calibration curves n Assumption: Use simplest mechanism to describe conversion N R N k1 R Cl + x N + HCl N 50-82 C H z parts Solvent “substrate” “amine” “product” k2 R Cl + y NaI 50-82 C R I + NaCl z parts Solvent “intermediate” N R N k3 R I + x + HI 50-82 C N N H z parts Solvent R N R N R + k4 R Cl N 50-82 C N z parts Solvent Cl- “impurity” 5 D. Place
- 6. III. Importance of Sampling Endpoint Removing time as a factor from the DoE If an adequate Kinetic model of the mechanism can be elucidated: Dynochem Simulator can be used to scope out suitable endpoints DoE factor (CPP) ranges can be investigated prior to committing time/resources 3.5 mol% <1 mol% from Previous batch experience 9h 15 h * Simulated using Dynochem’s Yield loss from side reactions (batch) Model 6 D. Place
- 7. IV. Data Fitting: Procedure to fit rate constants and Ea k1 substrate + amine > product + HCl k2 substrate + NaI > intermediate + NaCl Process sheet k3 intermediate + amine > product + HI k4 product + substrate > impurity Parts Experiment Amine NaI Temp Solvent mol equiv mol equiv degC mL/g A (low) 2 0 50 4 Scenario Sheet B (centerpoint) 3.5 0.5 66 5.5 C (repeat) 3.5 0.5 66 5.5 D (High) 5 1 82 7 1. Translate mechanistic proposal into process sheet 2. Translate design factors into the scenarios sheet 3. HPLC Area count data converted to mmol for substrate and product using reference standard calibration curves 4. Use Dynochem fitting engine to solve 4 k’s and 4 Ea’s 7 D. Place
- 8. Yield loss from side reactions (batch) Modified to model suspected reaction mechanism Data Sheet Scenario Sheet 8 D. Place
- 9. Dynochem Fits of Experimental Data Low factors, 50C No NaI 2 equiv amine 4 parts 7.5 Centerpoint values, 66C 0.5 equiv NaI 3.5 equiv amine Solution.product (Exp) (mmol) 5.5 parts Solution.substrate (Exp) (mmol) 6.0 A 5.0 Solution.impurity (mmol) Solution.product (mmol) High factors, 82C 1 equiv NaI 5 equiv amine 7 parts Solution.product (Exp) (mmol) 7.5 Solution.substrate (Exp) (mmol) Solution.substrate (mmol) B&C Solution.impurity (mmol) Solution.product (Exp) (mmol) Solution.substrate (Exp) (mmol) 4.0 Solution.product (mmol) Solution.impurity (mmol) D Process profile (see legend) Solution.substrate (mmol) Solution.product (mmol) 6.0 4.5 Process profile (see legend) Solution.substrate (mmol) 3.0 Process profile (see legend) 4.5 3.0 2.0 3.0 1.5 1.0 1.5 0.0 0.0 361.2 722.4 1.084E+3 1.445E+3 1.806E+3 Time (min) 0.0 0.0 336.2 672.4 1.009E+3 1.345E+3 1.681E+3 Time (min) 0.0 0.0 336.2 672.4 1.009E+3 1.345E+3 1.681E+3 S/P R2 = 0.97/0.98 S/P R2 = 0.99/0.98 Time (min) S/P R2 = 0.99/0.97 n Model fits substrate loss fairly well over the set of data n Model overestimates impurity content – model refinement necessary 9 D. Place
- 10. Fitting Summary SCENARIO 4 k1 substrate + amine > product + HCl k2 substrate + NaI > intermediate + NaCl k3 intermediate + amine > product + HI k4 product + substrate > impurity k 10-5 L/mol s Ea kJ/mol kcal/mol k1 1.1 +/-0.2 Ea1 40 +/- 9 9 +/- 2 k2 34 +/- 6 Ea2 - - k3 3.7 +/- 0.7 Ea3 100 +/- 30 24 +/- 6 k4 5.0 +/- 0.8 Ea4 90 +/- 10 23 +/- 3 n k values reported at T(ref) = 66 C 10 D. Place
- 11. V. Simulating Alternate Design Points n Criteria for the reaction: Reaction completed to <1% substrate Reactions time <30h. n Question: Which ranges of CPPs will fit the criteria? Process sheet Variables Yield % molpctSM % Calculate Yield:= solution.product / solution.substrate.Y0 molpctSM:= solution.substrate / solution.substrate.Y0 End time:= if(molpctSM<0.01,time,14400) 11 D. Place
- 12. Searching for New CPP Ranges Use the Dynochem Simulator Endpoint at 1 mol% substrate 12 D. Place
- 13. Reaction Endpoint Predictions n Initial Design points predict Reaction endpoints between 8h and 136h @ <1% substrate Parts Reaction Experiment Amine NaI Temp Solvent Endpoint mol equiv mol equiv degC mL/g h A (low) 2 0 50 4 136 B&C (centerpoint) 3.5 0.5 66 5.5 30 D (High) 5 1 82 7 8 A' (corner point) 2 0 50 7 261 D' (corner point) 5 1 82 4 4 n Simulation of alternate design points actually suggests that the reaction endpoint will vary between 4 and 261 h within the design space – the CPP ranges need to be altered to meet the criteria 13 D. Place
- 14. Influence of CPP “decrease” on Reaction Endpoint Parts Reaction Experiment Amine NaI Temp Solvent Endpoint mol equiv mol equiv degC mL/g h B&C (centerpoint) 3.5 0.5 66 5.5 29 Scenario 1 2 0.5 66 5.5 47 Scenario 2 3.5 0 66 5.5 43 Scenario 3 3.5 0.5 50 5.5 70 Scenario 4 3.5 0.5 66 7 38 n Reaction Temperature is the most influential parameter governing reaction endpoint. n Rank order: Rxn Temp > Amine > NaI mol > Parts Solvent 14 D. Place
- 15. Identifying a new "All-factors-low" design point Temperature effects Parts Reaction Experiment Amine NaI Temp Solvent Endpoint mol equiv mol equiv degC mL/g h A (low) 2 0 50 4 136 Scenario 1 2 0 66 4 52 Scenario 2 2 0 66 7 99 Scenario 2A 2 0 68 7 89 Scenario 2B 2 0 70 7 81 Scenario 2C 2 0 72 7 73 Scenario 2D 2 0 82 7 44 n In order to preserve Amine CPP range between 2-5 mol equiv and NaI CPP range to 0-1: Parts solvent CPP would need to be set to <4 parts to meet criteria -OR- Reaction Temperature CPP would need to have its lowest value set to > 82 degC to meet criteria 15 D. Place
- 16. Identifying a new "All-factors-low" design point NaI effects Parts Reaction Experiment Amine NaI Temp Solvent Endpoint mol equiv mol equiv degC mL/g h A (low) 2 0 50 4 136 Scenario 1 2 0 66 4 52 Scenario 2 2 0 66 7 99 Scenario 2E 2 0.1 66 7 91 Scenario 2F 2 0.2 66 7 86 Scenario 2G 2 0.3 66 7 78 Scenario 2H 2 0.5 66 7 61 Scenario 2I 2 0.9 66 7 28 n In order to preserve Amine CPP range between 2-5 mol equiv and Reaction Temperature CPP range to 66-82 degC NaI CPP would need to have its lowest value set to 0.9 mol equiv to meet criteria 16 D. Place
- 17. Conclusion: The Trade-Off Parts Reaction Experiment Amine NaI Temp Solvent Endpoint mol equiv mol equiv degC mL/g h A' (low) - Old Design Point 2 0 50 7 261 New "low" Design Point 2 0.5 72 6 29 New and Recommended CPP ranges for DoE based on kinetic assessment CPP Unit Low Centerpoint High Amine mol equiv 2 3.5 5 NaI mol equiv 0.5 1.5 2.5 Temp degC 72 77 82 Parts Solvent ml/g 5 5.5 6 17 D. Place
- 18. VI. Summary n A Kinetic assessment of the reaction prior to running a DoE is essential to ensure proper choice of design factor ranges. n If an adequate Kinetic model of the mechanism can be elucidated DoE factor (CPP) ranges can be investigated prior to committing time & resources n Dynochem is a powerful tool that enables the process chemist to leverage data collected from 4 “shake- down” runs. 18 D. Place
- 19. Acknowledgements Jianxin Ren Michael O’Brien Marty Guinn Peter Clark 19 D. Place