Final Report: Tailored Solvents for Pollution Source Reduction in Pharmaceutical and Fine Chemical ProcessingEPA Grant Number: R826121
Title: Tailored Solvents for Pollution Source Reduction in Pharmaceutical and Fine Chemical Processing
Investigators: Hatton, T. Alan
Institution: Massachusetts Institute of Technology
EPA Project Officer: Hahn, Intaek
Project Period: November 1, 1997 through October 31, 2000
Project Amount: $180,000
RFA: Technology for a Sustainable Environment (1997) RFA Text | Recipients Lists
Research Category: Nanotechnology , Sustainability , Pollution Prevention/Sustainable Development
The objective of this research project was to develop a new class of solvents that have solvation properties similar to those of solvents used conventionally in pharmaceutical and fine chemical processing but for which the potential for loss by environmentally-unfavorable air emissions or in aqueous discharge streams is minimized. Solvents are used widely in the chemicals processing industry (approximately a billion pounds per year) as they are necessary for the solvation of reactants and products in reaction processes. There is much interest in minimizing the use of these solvents to address safety, toxicity, emissions, and other aspects covered increasingly by the many government regulations promulgated over the past 2 decades. Indeed, much of the drive for reduction of solvent usage is in response to the U.S. Environmental Protection Agency (EPA) Pollution Prevention program to encourage pollution source reduction through "the use of materials, processes, or practices that reduce or eliminate the creation of pollutants or waste at the source. It includes practices that reduce the use of hazardous materials, energy, water, or other resources and practices that protect natural resources through conservation or more efficient use." Different approaches are being explored to meet these demands. New synthetic pathways are being sought to avoid the use of harsh solvents, a prime target development of water-based chemistries. The integration of process systems to minimize solvent usage, or the number of operations needed for solvent separation and recovery, can minimize energy usage and the potential for solvent emissions and losses. The use of replacement solvents, such as ionic liquids or supercritical fluids, is attractive in the chemicals industry, but in many cases the process requirements will be sufficiently different from the existing processes that retrofitting is not an economical alternative. There is a clear need, then, for alternative solvents for synthesis operations that are more environmentally friendly than the existing solvents that: (1) do not compromise chemical reactivity, selectivity, and yield of industrial synthesis operations; and (2) that can be used as drop-in solvent replacements for existing processes. This issue was addressed in the work reported here.
Tailored Solvents for Solvent Replacement and Recovery. A new concept was explored for the design and synthesis of a class of replacement solvents for tetrahydrofuran (THF), a widely used solvent in the chemical and pharmaceutical industries. THF is very volatile and is completely miscible with water, which makes aqueous workups problematic, owing to the need for subsequent recovery of the THF from the aqueous wastes, often by energy-intensive azeotropic distillation. There is a need to modify the physical properties of THF to reduce both its volatility and its water solubility dramatically, to minimize its propensity to enter the environment through vapor emissions and aqueous discharge streams, while retaining the solvation properties needed for facilitating reactions. This goal was achieved by tailoring THF through the attachment of hydrophobic moieties to modify its physical properties. The homologous series of modified THF solvents prepared and the physical properties determined for these solvents are listed in Table 1. The water solubility and water capacity of the n-alkyl tetrahydrofurfural ether solvents decreased as the alkyl tail length increased. The solvents remained stable after being exposed to strong bases during chemical syntheses, and their propensity to form peroxides was no different than that of THF. The optimum solvent for evaluation as a solvent replacement in the synthesis of two important pharmaceutical products, the HIV protease inhibitor Crixivan and the anti-triglyceride drug Gemfibrozil, was n-octyl tetrahydrofurfural ether (nOTE).
|Solvent||MW||Boiling Point(°C)||Solubility Solvent in H2O(g/L)||Solubility H2O in Solvent(mg/100mL)|
A concerted effort was made to develop alternative solvents for dimethylene chloride but was met with limited success. The combination of very low yields and very expensive reagents led to the abandonment of these efforts.
Synthesis of Crixivan, an HIV Protease Inhibitor. The series of reactions used in the commercial synthesis of the HIV protease inhibitor Crixivan was selected to demonstrate the potential benefits of using the tailored solvent, nOTE, as a replacement for THF in synthesis operations. Of these reaction steps, only one requires the use of an ethereal solvent such as THF. We have investigated the kinetics of this reaction to determine what effects the replacement solvent has on overall reactivity. The first order kinetic rate constants for a number of different solvents and solvent mixtures are shown in an Arrhenius plot in Figure 1. It is evident that the activation energy for the reaction appears to be unaffected, or only slightly affected, by the choice of solvent or solvent mixture. The reaction proceeds significantly slower in nOTE than in THF. This is demonstrated to be a result of the effective dilution of the THF by the attached tail, as the kinetic constants for nOTE are essentially the same as those for equimolar mixtures of THF and Me-O-C8 (the tail in nOTE) and of THF and decane. These results indicate that the attachment of the tail to THF does not modify the solvation capability of THF, and that the ether oxygen in the tail does not significantly contribute to the reaction process (cf results for Me-O-C8 and decane diluents). The importance of the strained ring structure in nOTE is evident from the lower kinetic rate constants obtained when the ring is opened, as in Me-O-C2-O-C8. The allylation of lithium acetonide yields two diastereomeric products, the selectivity for the major diastereomer decreasing with increasing temperature. A small decrease (less than 1 percent) in the diastereoselectivity was observed when using nOTE instead of THF as the solvent.
In the commercial production of Crixivan, isopropyl acetate (iPAc) is used as the solvent in all the intermediate steps shown in Figure 2, except in the allylation step, where THF is required. The iodination reaction that follows the allylation step cannot be allowed to proceed in the presence of THF because it leads to side reactions and low yields, and it is necessary to switch the solvent back to iPAc from THF. Each reaction is washed with water to remove byproducts,
Figure 1. Rate Constants for Allylation Reaction in Different Solvents
Figure 2. The nOTE Solvent Simplifies Overall Process.
with iPAc as the organic phase during the washes. The protection step and allylation reaction require anhydrous conditions; drying prior to the protection step is performed by distillation of an iPAc-water azeotrope. Demonstrations revealed that the solvent switch can be eliminated by using nOTE for the entire reaction sequence up to the allylation step (i.e., by eliminating iPAc) and that the drying of the nOTE solvent is a relatively easy task, as it has very limited capacity for water. Moreover, there is negligible loss of solvent to the aqueous wash streams, which eliminates the need for extensive treatment of the aqueous wastes to recover the solvent. The process block diagrams shown in Figure 2 illustrate the potential simplifications to the overall process flowsheet that can be gained by using nOTE as a replacement for both iPAc and THF in this process, both through the reduction in number of unit operations in the actual reaction sequence, and the number of offline separation and solvent recovery steps required.
The solvent usage was estimated to be significantly lower using nOTE as the solvent than it would be using iPAc and THF as in the existing process. This translates into lower energy usage through reduction and elimination of distillation and other separation requirements.
Synthesis of Gemfibrozil. Gemfibrozil is a drug used to treat high blood triglyceride levels linked to heart disease. It was originally marketed by Park Davis as Lopid with sales of $550 million in 1992, but since the expiration of its patent in 1993, it has been manufactured as a generic by many companies worldwide. In the patented manufacturing process, Gemfibrozil is prepared from isobutyric acid and 2,5-dimethylphenoxy-3-bromopropane (bromide). The double salt of the acid is alkylated by the bromide in THF to form the product, which is then extracted into basic water and recovered by acidification. The final product is purified by vacuum distillation, and THF must be recovered from the aqueous phase. This reaction was used to show that nOTE can be used as a drop-in replacement solvent for THF to reduce the need for the solvent recovery steps.
The deprotonation of isobutyric acid to form the enolate dilithio salt is slower in nOTE than in THF. In THF, deprotonation is done by refluxing with sodium hydride and then treating with lithium diisopropylamide (LDA) at 0 to 30°C. NaH is not used with nOTE owing to precipitation of sodium isobutyrate, but instead two equivalents of LDA are used. The kinetics of the second deprotonation of isobutyric acid are overall second-order, first-order in lithium isobutyrate, and first-order in LDA.
The formation of the dianion has a higher activation energy in nOTE than in THF. In practice, for the formation of the dianion the reaction mixture must be heated to 40°C for 1 hour to go to completion. In the THF solvent, the enolate was formed by refluxing with NaH and treating with LDA at 0 to 30°C. After alkylation with the bromide, the product was extracted into an aqueous base. The acidification of the aqueous phase yielded solid Gemfibrozil, which was purified by vacuum distillation. THF was dissolved in the aqueous waste and then recovered by distillation. In nOTE, the enolate was formed by reaction with two equivalents of LDA, and following alkylation with the bromide, the product was extracted into an aqueous base. Subsequent acidification of the aqueous phase yielded a solid Gemfibrozil precipitate that was purified by distillation. Small amounts of nOTE that were solubilized in the aqueous base solution by the Gemfibrozil were easily recovered by extraction with an organic solvent such as hexane.
The yields of Gemfibrozil using these different solvents were very similar. After distillation to isolate pure Gemfibrozil, the yield was 80 percent from nOTE and 78 percent from THF. The nOTE was easily recycled in this reaction, requiring only vacuum removal of trace amounts of water following the aqueous workup; recycling had no effect on yield.
The use of tailored or modified solvents as effective drop-in solvent replacements for the manufacture of pharmaceutical products has been investigated extensively. It has been demonstrated that significant savings in energy and solvent usage can accrue by appropriate solvent selection using the manufacture of two commercially important drugs, Crixivan and Gemfibrozil, as examples. The primary advantages of using these solvents are that their low water solubility ensures negligible loss of the solvent to the aqueous phase during workup and their capacity for water is sufficiently low so that drying of the solvent prior to reactions is relatively inexpensive.