Kim
Alfonsi
a,
Juan
Colberg
b,
Peter J.
Dunn
*c,
Thomas
Fevig
d,
Sandra
Jennings
a,
Timothy A.
Johnson
b,
H. Peter
Kleine
d,
Craig
Knight
c,
Mark A.
Nagy
d,
David A.
Perry
*b and
Mark
Stefaniak
c
aPfizer Global Research and Development, Ann Arbor, Michigan, MI-48105, USA
bPfizer Global Research and Development, Groton, Connecticut, CT-06340, USA
cPfizer Global Research and Development, Sandwich, Kent, CT139NJ, UK
dPfizer Global Research and Development, Chesterfield, Missouri, MO63017, USA
First published on 16th November 2007
Influencing and improving the environmental performance of a large multi-national pharmaceutical company can be achieved with the help of electronic education tools, backed up by site champions and strong site teams. This paper describes the development of two of those education tools.
Early pioneers in green chemistry included Trost (who developed the atom economy principle)5 and Sheldon (who developed the E-Factor).6 These measures were introduced to encourage the use of more sustainable chemistry and provide some benchmarking data to encourage scientists to aspire to more benign synthesis. Later, green chemistry became formalised by the publication by Warner and Anastas7 of a holistic set of principles designed to raise awareness of the safe, environmentally sensitive and sustainable practice of chemistry. While many of these principles were second nature to process development chemists and their manufacturing colleagues in the wake of the pollution control legislation over the last 30 years, the same cannot be said of their medicinal chemistry colleagues. The modern practice of drug discovery relies heavily on speed of execution, which in turn relies on robust methodologies emphasising reliability rather than environmental impact. While the scale of the reactions conducted at the early stages of a program is usually small, the cumulative footprint generated by tens or hundreds of laboratories in a pharmaceutical company becomes significant. Moreover, the delay that may be incurred by the necessity to reengineer a ‘discovery route’ to achieve a scaleable process impacts the development timeline, as well as its cost. This paper describes ongoing initiatives in Pfizer to equip its medicinal chemists with a working knowledge of the principles of green chemistry, favouring restraint over constraint, and providing access to tools which guide the selection of greener solvents and reagents. We believe the success of these initiatives will reduce our environmental impact, improve worker safety and reduce the time taken to deliver important new medicines addressing major unmet medical needs.
(i) Worker safety10– including carcinogencity, mutagenicity, reprotoxicity, skin absorption/sensitisation, and toxicity
(ii) Process safety – including flammability, potential for high emissions through high vapour pressure, static charge, potential for peroxide formation and odour issues.
(iii) Environmental and regulatory considerations11 - including ecotoxicity and ground water contamination, potential EHS regulatory restrictions, ozone depletion potential, photoreactive potential. Of course compliance with regulations and company guidelines provide the baseline of Pfizer's environmental policy.
This detailed assessment was then translated into a simple 1 page guide which is shown in Fig. 1.12
![]() | ||
Fig. 1 Pfizer solvent selection guide for medicinal chemistry. |
A summary of why each solvent is placed in the red category is provided in Table 1.
Red solvent | Flash point | Reason |
---|---|---|
Pentane | –49 °C | Very low flash point, good alternative available. |
Hexane(s) | –23 °C | More toxic than the alternative heptane, classified as a hazardous airborne pollutant (HAP) in the US. |
Diisopropyl ether | –12 °C | Very powerful peroxide former, good alternative ethers available. |
Diether ether | –40 °C | Very low flash point, good alternative ethers available. |
Chloroform | N/A | Carcinogen, classified as a HAP in the US. |
Dichloroethane | 15 °C | Carcinogen, classified as a HAP in the US. |
Dimethyl formamide | 57 °C | Toxicity, strongly regulated by EU Solvent Directive, classified as a HAP in the US. |
Dimethyl acetamide | 70 °C | Toxicity, strongly regulated by EU Solvent Directive. |
N-Methyl pyrrolidinone | 86 °C | Toxicity, strongly regulated by EU Solvent Directive. |
Pyridine | 20 °C | Carinogenic/mutagenic/reprotoxic (CMR) category 3 carcinogen, toxicity, very low threshold limit value TLV for worker exposures. |
Dioxane | 12 °C | CMR category 3 carcinogen, classified as HAP in US. |
Dichloromethane | N/A | High volume use, regulated by EU solvent directive, classified as HAP in the US. |
Dimethoxyethane | 0 °C | CMR category 2 carcinogen, toxicity. |
Benzene | –11 °C | Avoid use : CMR category 1 carcinogen, toxic to humans and environment, very low TLV (0.5 ppm), strongly regulated in the EU and the US (HAP). |
Carbon tetrachloride | N/A | Avoid use : CMR category 3 carcinogen, toxic, ozone depleter, banned under the Montreal protocol, not available for large-scale use, strongly regulated in the EU and US (HAP). |
The list of solvents covered in Fig. 1 is not extensive but covers solvents commonly used in medicinal chemistry. Solvents, such as benzene and carbon tetrachloride, were included to reinforce the avoidance of their use.
In addition, the scientists in our green chemistry teams produced a simple solvent replacement table for each of the solvents in the red/undesirable category, with the philosophy of adopting a “use this instead” policy rather than a “don't use” policy. This replacement table is shown in Table 2. The replacements are either chemically similar (e.g., heptane as a replacement for the high flammable pentane) or functionally equivalent (e.g., ethyl acetate, methyl tert-butyl ether (MTBE) or 2-methyltetrahydrofuran (2-MeTHF) as alternative extraction solvents to dichloromethane).
Undesirable solvents | Alternative |
---|---|
Pentane | Heptane |
Hexane(s) | Heptane |
Di-isopropyl ether or diethyl ether | 2-MeTHF or tert-butyl methyl ether |
Dioxane or dimethoxyethane | 2-MeTHF or tert-butyl methyl ether |
Chloroform, dichloroethane or carbon tetrachloride | Dichloromethane |
Dimethyl formamide, dimethyl acetamide or N-methylpyrrolidinone | Acetonitrile |
Pyridine | Et3N (if pyridine used as base) |
Dichloromethane (extractions) | EtOAc, MTBE, toluene, 2-MeTHF |
Dichloromethane (chromatography) | EtOAc/heptane |
Benzene | Toluene |
There are a number of points that need further comment. Many of our scientists are surprised that dichloromethane is the recommended alternative to other chlorinated solvents, such as chloroform. All that Table 2 is indicating is that if a chlorinated solvent needs to be used, dichloromethane is the best choice out of the four.
All of the solvents have good replacements, with the exception of one group, which is the dipolar aprotic solvents dimethyl formamide, dimethyl acetamide and N-methylpyrrolidinone. For this group of solvents, acetonitrile is a relatively poor substitute, especially for reactions involving a strong base. Due to the lack of good alternatives, Pfizer, with a group of other pharmaceutical companies, has identified finding replacements for these solvents as a key target in green chemistry research.13
The guide and replacement table seem almost ridiculously simple but when used by our enthusiastic site teams they led to amazing results, including a 50% reduction in chlorinated solvent use across the whole of our research division (more than 1600 lab based synthetic organic chemists, and four scale-up facilities) during the time period 2004–2006. Even sites that had an increase in the number of chemists during that period were able to report a 50% reduction in chlorinated solvent use. In addition, we were able to reduce the use of an undesirable ether by 97% over the same two year period and substantially promote the use of heptane compared with hexane (more toxic) and pentane (much more flammable).
• To provide a balanced assessment of chemical methodologies, taking into account the many constraints that scientists have to take into account when making decisions in their work. To our mind the ideal reagent would have three ideal characteristics:
• (i) The ability to work in good yield in a wide variety of “drug like molecules” —this is a characteristic highly valued by medicinal chemists.
• (ii) The ability of a reagent to be used for scale-up to prepare multi-kilogram batches—a characteristic valued by our Chemical R and D, Kilo Lab and Pilot Plant chemists and engineers.
• (iii) To be as green as possible. Our green chemistry teams would like to introduce the greenest possible reagent as early as possible in the discovery/development process. The assessment of greenness included worker safety, ecotoxicity and atom economy.
• To provide easy access to the chemical literature or procedures for reagents that score well in the assessment. In the on-line Pfizer version of the guide, reagents that score well are linked directly through electronic links to key literature papers, internal procedures or both.
• To raise awareness of newer emerging green methodologies.
We decided to map the reagents onto a series of grids (or Venn diagrams), with each grid representing a commonly used chemical transformation. Each Venn diagram indicates which of the three ideal characteristics each reagent met. A breakdown of the grids and a discussion of the zones in the grid are shown in Fig. 2.
![]() | ||
Fig. 2 The zones in the Venn diagram (or grid) that form the basis of the reagent guide. |
Zone 1: reagents in this zone have all three desirable characteristics. These are reagents we would like our scientists in medicinal chemistry and chemical research and development to try first.
Zone 2: the reagents in this zone meet the wide applicability and scalability criteria but do not meet our greenness criterion. Reagents in this zone are still fully acceptable for use in late discovery/early development. Note that reagents with gross environmental issues, such as a thallium or tin reagent, would not be in this zone as they would fail the scalability criterion, but reagents with a slightly higher molecular weight and poor atom economy, such as EDCI for amide coupling, would make this zone.
Zone 3: this zone retains the positive attributes of scalability and greenness and reagents in this zone are good for our chemical research and development groups.
Zone 4: this zone has positive attributes for greenness and wide applicability but fails the scalability criterion, an example might be an electrolysis reaction where the company does not have access to large-scale electrolysis equipment.
Reagents in zones 5, 6 and 7 only meet one positive attribute and are less favoured. In the Pfizer electronic version of the guide, only reagents that fall in zones one to four are hypertext linked to in-house procedures or key references.
Two sample grids are shown to illustrate the reagent guide with a further two available in the electronic supplementary information.†
Fig. 3 14,15 shows the grid for the oxidation of alcohols to aldehydes.
![]() | ||
Fig. 3 Oxidation of primary alcohol to aldehyde. |
The three most common oxidants used by Pfizer's medicinal chemists for this transformation are Dess–Martin periodinane16 or its precursor IBX, tetrapropylammonium perruthenate (TPAP)17 and the Swern oxidation.18 All of these methods have significant scale-up issues, for example Dess–Martin periodinane is a high energy molecule14 that has poor atom economy and is prohibitively expensive for use on a multi-kilogram scale. The use of stoichiometric TPAP again has very poor atom economy and is also prohibitively expensive for large-scale use. A review of large-scale oxidations since 1980 revealed only one large-scale use of TPAP to catalyse an oxidation with a co-oxidant and no examples of stoichiometric use.15 The Swern oxidation is used at Pilot Plant scale but generates toxic by-products and the stench of dimethylsulfide. Hence, the purpose of the reagent guide is to influence the medicinal chemist away from the reliable but environmentally unfriendly methods to more friendly methods, such as the oxidation with bleach (NaOCl) catalysed by nitroxyl radicals, such as TEMPO19 and PIPO.20 In addition, there has been an explosion in the chemical literature of methods that use molecular oxygen as an oxidant, with more than 150 papers in the last 3 years. These methods carry some challenges on scale-up, as the use of molecular oxygen to aerate flammable solvents is a significant safety concern. These concerns can be reduced by using oxygen diluted with large volumes of nitrogen but still these methods21,22 lie on the edge of acceptability when judged against the scalability criteria. An improved safety profile and more acceptable scalability is obtained if the oxidation is performed in water.23 Again, the purpose of the reagent guide is to provide scientists with easy up-to-date access to developments in this exciting area of green oxidation. Other methods shown in Fig. 3 can be found in the following publications.24,25
A similar Venn diagram covering the oxidation of secondary alcohols to ketones can be found in the electronic supplementary information.†
Fig. 4 shows the grid for amide formation from acids (not prone to racemisation) and amines.
For the oxidation grids we were able to set strict criteria for greenness (reaction by-products should be either water or sodium chloride and there should be no major process safety issues). For amide formation, the majority of literature methods had very poor atom economy. We decided to set the greenness criteria for this transformation as the following.
• Side products should have a molecular weight less than 200.
• No major process safety issues.
• No major environmental issues.
The first of these criteria, based on atom economy, might seem overly generous but in fact 50% of the reagents in Fig. 4 fail this criterion.
Uronium salts, such as HATU26 and HBTU,27 have become widely used in research laboratories but have many green chemistry issues. Their by-products have molecular weights of 398 and 397, respectively, for accomplishing a dehydration reaction (removing a molecule of water with a molecular weight of 18). They are both highly energetic molecules and HATU is shock sensitive.28 The phosphorus based reagent BOP29 and PyBOP30 are again energetic molecules and have even worse atom economy. BOP has the further major disadvantage that its manufacture and use involve HMPA (a class 1 carcinogen).
Dicyclohexyl carbodiimide (DCC) and di-isopropyl carbodiimide fail our green criteria because of their very strong sensitisation properties and hence in recent years have become rarely used for scale-up in the pharmaceutical industry. Cyanuric chloride is similarly a very strong sensitiser. Oxalyl chloride does not meet our greenness criteria on account of its poisonous by-product carbon monoxide. 1-Chloro-4,6-dimethoxy-1,3,5-triazine (CDMT) is a sensitiser but has been used by some process groups for scale-up.31 EEDQ,32PPACA,33 and EDCI34 do not meet our greenness criteria on the basis of atom economy but are widely used for scale-up chemistry. Thionyl chloride and chloroformates are the most common reagents for this transformation used by the pharmaceutical industry,35N,N′-carbonyldiimidazole (CDI) is growing in popularity and was used in the commercial synthesis of sildenafil36 and sunitinib.37 We judged that thionyl chloride did not fully meet our greenness criteria because of its worker safety issues but was preferred to oxalyl chloride for acid chloride formation. Although reagents such as CDI and isobutyl chloroformate are described as green, they are not without issue, for example, the synthesis of CDI uses highly poisonous phosgene, our assessment simply says they are greener than some of the alternatives available at this time.
All of the reagents discussed so far are stoichiometric reagents but the real opportunity is in the development of catalytic reagents where the only by-product would be water. The use of boronic acids,38 and in particular boric acid,39 to catalyse amide formation is very exciting and works well in some substrates.40 In reality, boric acid is a poor catalyst for amide formation but it does help drive the reaction of acids and amines that undergo substantial uncatalysed reaction over to completion.41 For these substrates, boric acid catalysis represents a very green methodology. Enzymatic methods are another catalytic method where the only by-product is water.42
The boric acid and enzymatic methodology are active research areas and the regularly updated Pfizer reagent guide gives Pfizer scientists easy access to the latest green advances in these areas. The grid also gives references to other reagents that meet two out of the three criteria.43–45
A Venn diagram covering amide formation from acids, prone to racemisation, and amines can be found in the electronic supplementary information.†
Footnote |
† Electronic supplementary information (ESI) available: Grid 3–oxidation of secondary alcohols to ketones. Grid 4–amide formation from acids (prone to racemisation) and amines. See DOI: 10.1039/b711717e |
This journal is © The Royal Society of Chemistry 2008 |