Optimizing Complex Organic Chemical Process Schemes

The KiloMentor Blog concerns itself with process development. It assumes that an adequate process scheme has been chosen for the development. A process scheme is the sequence of reaction transformations that take commercially available substances and create from them a desired material. KiloMentor’s concern with choosing the best process scheme to be scaled up is to increase the likelihood that considerations important in scale-up are properly weighted when selecting the most preferred route.

Steps to make this important selection of the process route are amplified below.  

1. Prepare alternative schemes for the synthesis of the desired product from commercial materials. The methods for preparing process schemes will not be discussed here. Write balanced equations for each reaction step if a reagent is known. Prepare preliminary costings for the candidate routes.

2. Consider selecting two different routes: one to meet the demand for timely delivery of the final product to be used in other parts of the project and a second to be developed into the path for a commercially viable product.

3. Rate the alternative schemes prepared in 1 versus the criteria, musts, wants, ruggedness, analogs in the literature, risk of failure, convergence, phase shifting of intermediates, overall cost, throughput, waste, safety, and atom economy. Use some ranking methodology such as in the Kepner-Tregoe method.

4. If you are going to use two schemes: one to deliver material in a timely fashion for other departments and one for the commercial route, the last step should be the same for both. If the last step is not the same, the last purification should be very rugged and capable of removing many impurities. It is a major problem if the final steps are going to be different and the final isolation-purification is not rugged. The reason is that specifications and pharmacological and sometimes clinical data is collected using material from the research synthesis and this work might have to be redone if the products from both syntheses do not have the same impurity profiles.

5. Choice of whether 1 or 2 routes will be worked.

6. If a timely research route will be used to supply material for development run through that route to give the first look at the product. Often the route used to supply early batches to other teams working towards the commercial product is the laboratory procedure that has been modified as needed to increase scale and throughput. Operations such as column chromatography which cannot be adapted for scaling are worked around but changes are as few as possible. The idea is not to optimize just to make it workable for larger batches. 
7. Gather literature about the steps of the planned commercial route. The objective is to obtain as complete an understanding as possible of everything in the literature, which seems pertinent to the transformations you are going to try. This would include substructure searches on all the intermediates to discover the closest analogous structures existing in the literature; the closest analogous reactions including reagents, interfering or non-interfering groups, and solvents; reviews and mechanism studies of the transformations being studied.

8. Route for research products should only be improved as much as necessary to deliver the commitment for the material. If the ingredients are inexpensive work only on the isolations. If the ingredients are expensive only address the big-ticket items and changes that are substantial and quick.

9. A quick run through the proposed reaction sequence in one of its embodiments. In this work, the intermediates are split and one portion is purified before going on but the other portion is taken on without purification to see whether the subsequent reactions will (purge) purify the mixtures and, if there is no purification, indicate to the analytical department the impurities that are likely to most difficult to remove. At the end of the process, analytical and spectroscopic methods are used to evaluate whether the desired substance has been created and in what purity and yield. This is possible because the product is available from the research route. This exercise will provide a level of confidence that the desired product can be produced by the route. There is no point in optimizing a route that cannot be made to work!

10. Do a revised costing of the commercial proposed route to determine priorities for optimization. Remember optimization can go on forever but at some point, time will run out. The object is to have the best process and the lowest cost possible before the time expires!

11. The starting points for optimization are the paper reaction scheme, the best analogous transformations in the literature, your literature file, your costing, and your quick preliminary process run.

12. Make a preliminary assessment of what reactions might be telescoped together. The criteria for considering telescoping in good process work is that no intermediate is isolated unless its isolation contributes significantly to the purification. In multipurpose plants, there is an additional reason for isolating intermediates and that is to provide convenient and stable stopping points so that the flexibility of the plant is maintained. Stopping points are important as patches to protect against problems. Consecutive reactions are only candidates to be telescoped together in the process if there are no phase shifts in the purification of the first of the candidate reactions, which substantially contribute to purification. This is so because two telescoped reactions are normally done consecutively in the same solvent, phase shifts are usually done during the work-up. Since the phase shifts in the first work-up are lost in the process of telescoping, the purification, which occurs in these phase shifts will also be lost.

13. Choose the reactions to begin optimization on. Optimization of the early steps gains better access to material for the later steps. Attack the “must fix” problems first since if these can’t be addressed the route must be abandoned.

14. It is very advisable to fix the last step early on. If you can decide at the outset what the last step will be, you can design it to be as chemically simple as possible and optimize conditions to get the same polymorph every time. If you can’t fix the final step then add a recrystallization step. It may seem redundant or wasteful of material but for highly active compounds whose activity is very dependent on physical form, it may be the best route.

15. Assess the dilution (D) in L/Kg at which each proposed process step is run in the better literature examples and assign a rough numerical value. Assign a time (H) in hours for the completion from beginning to end of each process step. The conversion of starting material X_i  Kg to Y_i Kg of product in step I can be expressed in the expression X_i/Y_i Kg of X_i giving rise to 1.0 Kg of Y_i.

16. Validate the method of “in situ” analysis for the intermediate that is being optimized. Remember that each process transformation is divided into the reaction transformation yield, which measures the quantity of intermediate in the reaction solution as determined by assay divided by the theoretical quantity of material in solution expected AND the isolation yield which is the isolated weight divided by the quantity of intermediate in the reaction solution as determined by assay.

17. Select among the discontinuous variables the most promising reagent and solvent. This is a mouthful and includes some enormous assumptions. One is more likely to make better choices if one performs a reaction-based electronic literature search for the transformation required in the presence of all the non-reacting functional groups that will be present in the particular process substrate. The sample conditions that you unearth must be such that your particular substrate will be soluble. Identify as many solvents as possible that have worked with the preferred reagent. Searching in monographs like Organic Reactions and Fieser and Fieser can also give leads to start conditions. Aim to use a reagent and condition close to a literature condition. There are no special bonuses for being unusual unless the usual fails. Only if there is a very substantial advantage to be gained by using a less tested or untested condition should it be considered. Such advantages are cheap reagents, convenient solvents, high-load solvents, improved stability, safety/lower toxicity, and solvent convergence. If these conditions apply then try a “wish” reaction using the desired conditions or reagent being particularly careful and aiming for a positive result. If you do not immediately get a measurable yield of product, give up. There is a tremendous waste of time possible in 0 yield reaction space!

18. Select the screening parameters (up to 7), which will be used in the optimization. An understanding of the mechanism of the reaction can be very useful in choosing the screening parameters. If you have extra space for screening parameters choose one to be the effect of absolute concentration on the reaction. This parameter is crucial to getting the optimal throughput of mass in the process. Additionally the higher concentration reaction may proceed substantially faster. Certain reaction variables are only important when one of the reactions or associations is fast compared to the time of addition. Direct or inverse addition, mixing method and rate, addition time…. Sampling the reaction mixture at half-addition time and checking for product formation and/or starting material consumption can give a good clue as to whether these variables are likely to be important.

19. Run the eight screening runs collecting samples at staged time intervals. Quench the samples and then analyze them all to determine the reaction transformation yields and the time profile of the product/by-product productions. Half-addition analysis can also be done when just half of a material has been added. The purpose is to see whether the reaction is so quick that a substantial amount of reaction has occurred at the half-addition point. Using statistics determine the significant variables from those tested. In the screening be vigilant for conditions that substantially increase particular impurities. It may be critical later to identify these impurities and it will be much easier to isolate these impurities using reaction conditions, which give much larger yields of them.

20. 20.  Using half-log paper predict the likelihood that the optimization will meet the target of the yield improvement that is desired. If it is unlikely, choose some new screening parameters including the best of the old ones, and rerun the eight experiments and reevaluate.

21. Compare your best result from the screening reactions. If it is already close to the target or if the yield is good enough to practically produce material then switch your attention to the isolation method. If the isolation method has no major problems switch to the next target reaction and do its screening parameters.

22. Optimization is a cost optimization, not a yield optimization. Remember optimization time is wasted time if the route has any irremediable flaws.

23. Suppose we are continuing with the optimization using the identified statistically significant parameters. Simplex optimization is the simplest methodology and it is psychologically most satisfying. It is also the easiest method for dividing up the work so that everyone has a satisfying activity. As the start point in the simplex choose those conditions that had the highest reported reaction yield but also include lower yield conditions which showed a rising trend for product that had not peaked at the final time. Another advantage of Simplex is that if during the work any new significant parameter gets identified it can immediately be included in further optimization. Again in Simplex runs surprising increases in the level of any impurity may provide a method to synthesize and identify that impurity.

24. Work on the isolation is best performed on larger batches of material prepared by the best available reaction conditions. Isolation experiments are evaluated by the weight of clean intermediate obtained. Isolation studies are preferably done using split runs. That is, a reaction mixture is divided into equal portions, and alternative isolations are tested against one another. Besides yields, the purity profiles of the isolated product are examined and compared. Analyses of the fractions during the isolation development are useful because these form the basis for the creation of in-process checks required in the process. During the isolation particular attention must be paid to the maximum volume. The point of maximum volume in the entire reaction step very often occurs during the isolation and this maximum volume limits the throughput of the process step. The volumes, which are typically used in development research and literature preparations are enormously excessive and many of the treatments are precautionary more than necessary. Large washes contribute to the waste, which must be processed and the expense. Remember to include waste disposal in the costing.

25. Developing a scheme for robust isolation is not easy. Although there are reaction databases there are no isolation databases. Each isolation should be considered de-novo and a think tank approach taken to gather a wide range of ideas, which could form the basis for a rugged isolation. The objective should be to achieve what is called “practical purity”. By this I mean that there is no point in removing impurities, which will automatically be removed by the further chemical transformations of the material or by the phase shifts in subsequent steps. Practical purity is a purity that has removed impurities that interfere in further processing. Therefore it may be a good idea to take some of the crude unpurified product directly from the evaporation of the reaction liquors and perform the next reaction (or reactions) to see what impurities must be cleaned up and what impurities are taken care of by the later transformations.

26. Swish TLC on the product of these isolations can also provide concentrated samples of impurities. Multiple recrystallizations of small samples of product reusing the recrystallization solvent can also provide impurity samples. An impurity that does not change its proportion to the main product during screening or/and optimization is probably an impurity coming from a reactant. It is probably an isomer or a homologue. Check the quality of your commercial ingredients. Use materials of a quality available in kilogram quantities.

27. Solvent changes may be required in the work-up. Use azeotropes and other methods to efficiently do solvent switches without waste.

28. Practical concerns such as filter speed, filter media, emulsions, crystal form interfaces number of reactors, and reactor transfers are of concern, particularly in isolation. These and the elimination of chromatography must be solved.

Reagent Selection

Certain reagents are clearly disfavoured based on cost. 

A reagent that has not been shown to function in a solvent, which will dissolve the required substrate is a very poor reagent choice. This is true because functional group transformations are very sensitive to solvent choice and information about a transformation is not generally transferable between different solvents. This is particularly true when the change is between radically different solvent types.

 A reagent that has not been shown to be compatible with all the functional groups, which are supposed to be unreactive during a transformation cannot be accepted uncritically. This is not as serious a problem as the solvent because often a potential interfering reaction turns out to be kinetically uncompetitive with the desired reaction or the reaction conditions can be adjusted to make it uncompetitive. Selectivity in the reaction is better than using extra protection-deprotection steps. 

Most “optimization” done in reaction method publications by academic authors are one-variable-at-a-time studies with a single model substrate and are useless in assessing the scope of a reaction.

Solvent Choice for Optimization

A pure solvent is preferred so that the recovery of solvent for recycling is possible. Very often solvent recycling is not practiced in early development. Solvent recovery is uneconomic in multipurpose plants. If for solubility reasons the dilution using a pure solvent must be high, then a solvent mixture, which greatly enhances throughput could be considered. Solvent mixtures made up of readily separable components are promising. If the more polar component is more volatile than the less polar solvent there is a good likelihood that the product can be crystallized upon removing the more volatile component.  A pure solvent is preferred for crystallization otherwise getting the proper solvent ratio reproducibly can be a problem.

The solubility of the substrate and product should be determined in solvents with the biggest possible differences in principal properties. Are the different solvents all miscible together? Consider using no solvent. This provides the highest throughput but the least temperature control.

Solvents differing in their principal properties are the most likely to optimize differently.