- Email: [email protected]
The role of catalysis for the clean production of fine chemicals
Applied Catalysis A: General 189 (1999) 191–204
The role of catalysis for the clean production of fine chemicals Hans-Ulrich Blaser ∗ , Martin Studer Novartis Services AG, Catalysis and Synthesis Services, WRO 1055.6, CH-4002 Basel, Switzerland
Abstract The role of catalysis for the production of fine chemicals is reviewed. The following topics are discussed on a general level: characteristics of the manufacture of fine chemicals, opportunities opened up by catalysis, critical factors for the application of catalysts and the tools that are available to the catalytic chemist. The general part is illustrated by specific examples from the catalysis group of Ciba-Geigy/Novartis such as chemoselective hydrogenations of aromatic nitro groups, the combination of a homogeneous and heterogeneous Pd catalyzed reaction for the alkylation of aromatic systems, catalytic systems for the enantioselective reduction of an ␣-keto ester, different routes to an N-alkylated hindered aniline including the (S)-metolachlor process, and the use of on-line monitoring of catalytic hydrogenations with ATR-probes. A short outlook on future developments is also presented. ©1999 Elsevier Science B.V. All rights reserved. Keywords: Technical catalysis; Fine chemicals production; Catalyst development; Process optimization; On-line monitoring
1. Introduction Traditionally, fine and specialty chemicals have been produced predominantly using non-catalytic organic synthesis. At least in the opinion of R. Sheldon, this is one important reason why besides the desired product, between 20 and 100 times as much waste is produced [1]. Of course many other factors such as the complexity of the molecules (and consequently, the number of synthetic steps), as well as the short development time for a (registered) technical synthesis and also the high requirements concerning purity of many fine chemicals are responsible for the unfavorable ecological situation. Nevertheless, the application of catalytic methods in the fine and specialty chemicals industry has increased in recent years in part because both produc∗ Corresponding author. E-mail address: [email protected] (H.-U. Blaser)
tion costs and waste minimization are of growing importance even for high value pharma and especially agro chemicals. In this contribution we will first give a short characterization of the problems in fine chemical production. Then we will describe the tools that are available to solve some of these problems with the help of catalysis and finally, we will present examples from our laboratory to illustrate some of these points.
2. Problems and opportunities 2.1. Characteristics of the manufacture of fine chemicals The manufacture of fine chemicals and especially of pharmaceuticals and agrochemicals can be characterized as follows (typical numbers are given in parenthesis):
0926-860X/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 2 7 6 - 8
192
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
• Rather complex molecules (isomers, stereochemistry, several functional groups) with limited thermal stability. • Production via multistep syntheses (5– > 10 steps for pharmaceuticals and 3–7 for agrochemicals) with short product lives (often < 20 years). Usually classical organic reactions, catalysis as exception. • Production usually in solution, at ambient pressure and low to medium temperature in relatively small (500 l–10 m3 ) multipurpose batch equipment. • Relatively small scale products (1–1000 t/year for pharmaceuticals, 500–10 000 t/year for agrochemicals) • High purity requirements (usually >99% and <10 ppm metal residue and ee > 98% in pharmaceuticals). • High value added and therefore, tolerant to higher process cost (especially for very effective, small scale products). • Short development time for the production process (
2.2.3. Combining several transformation in one step • Reductive alkylation of amines with carbonyl compounds (imine not isolated) • Hydrogenation–acylation of nitro arenes to acylanilines • Direct alkylation of amines with alcohols via a dehydrogenation–condensation–hydrogenation sequence. 2.2.4. Replacing toxic or problematic reagents (and reactants) • Alkylation of amines or aromatics with alcohols instead of alkyl halides (reduction of salt production) • Use of H2 instead of metals, metal hydrides or sulfides • Use of H2 O2 or O2 instead of metal oxides or peracids • Solid acids and bases to replace soluble ones. 2.3. Critical factors for the application of catalysts An impressively large number of highly selective catalytic transformations is recorded in the literature that in principle can be applied to the synthesis of fine chemicals. However, quite many prerequisites must be fulfilled in order to render a catalytic process technically viable and to really profit from the opportunities described in the preceding chapter. 2.3.1. Catalyst performance The selectivity of a catalyst is probably its most cited property. Besides chemoselectivity, regio-, stereo- and enantioselectivity play a role for the synthesis of fine chemicals. Due to often high cost of starting materials and intermediates as well as of separation steps, selectivities >95% are usually required to make a catalytic method attractive. The catalyst productivity, given as turnover number (ton) or as substrate/catalyst ratio (s/c), determines catalyst costs. Here it is more difficult to give general numbers because both catalyst prices and the value added by the catalytic transformation play an important role. For enantioselective catalysts, tons ought to be >1000 for high value products and >50 000 for large scale or less expensive products (catalyst re-use increases the productivity). The catalyst activity (turnover frequency (tof), h−1 ), affects the production capacity. As a rule of thumb, tofs ought to
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
193
Fig. 1. Transformations that are only possible with metal catalysts.
be > 500 h−1 for small and >10 000 h−1 for large-scale products. 2.3.2. Substrate specificity Catalytic methods are often more substrate specific than stoichiometric ones, i.e., even small changes of the structure of the starting material can strongly affect the catalyst performance of a given catalyst. This is especially true for highly optimized stereo- and enantioselective catalysts, thereby leading to a low predictability for new substrates. This fact renders synthesis planning difficult for the synthetic chemist who by definition must be a generalist and often does not know the scope and limitations of catalytic methodologies. In our experience, catalytic methods are usually abandoned if they do not succeed almost at first try because development time and costs for fine chemicals are limited. 2.3.3. Commercial availability of catalysts and ligands In many cases this is still a major problem for products where relatively small amounts of catalysts are needed — usually not enough to make it worthwhile for a catalyst manufacturer to develop a tailor-made catalyst that might be required. For homogeneous catalyst both (chiral) ligands and many metal precursors can be expensive. Ligands often have unusual structures that are prepared via many step syntheses.
Heterogeneous catalysts can not be characterized on a molecular level. The catalytic properties of two catalysts with the same description, e.g., 5% Pt/C, can vary considerably. It is well known that even small variations in the preparation procedure or impurities can alter the structural and chemical properties of a heterogeneous catalyst significantly. This might or might not affect its catalytic or chemical activity. It is precisely this ‘might or might not’ that leads to problems of reproducibility and frustration. 2.3.4. Practical problems Especially homogeneous catalysts are very sensitive to air, reactor materials, handling etc. Most catalytic processes are also sensitive to the presence of impurities that can act as poisons or can modify the selectivity of a catalyst. This is due to the fact that the catalyst is present in very small concentration, that the active species are very reactive and that quite often oxygen sensitive ligands and complexes are employed. The quality of the starting material as well as of the reagents (solvent, gasses, etc.) is therefore, absolutely crucial. Catalyst separation from the reaction mixture is a problem that has to be addressed for every homogeneous catalyst. Usually, the solution has to be tailored to the substrate, solvent, catalyst type etc. Distillation, crystallization, extraction or immobilization of the soluble catalyst are often successful approaches. Solid catalysts can be separated from the reaction mixture
194
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
by simple filtration. This allows easier work up and isolation of the desired product and is the most obvious advantage of a heterogeneous catalyst. Leaching of the active species such as noble metals can contaminate products and require an expensive purification. 2.4. The toolbox of the catalytic chemist In the following chapter we sketch the tools that are available for developing catalysts that make a desired transformation technically feasible. We use the example of homogeneous and heterogeneous metal catalysts for this endeavor. 2.4.1. Design parameters for catalytic systems 2.4.1.1. Homogeneous catalysts For homogeneous catalysts the most important elements are the type of metal (very often used are Pd, Rh, Ru, Ir, Ni, Cu, Ti) and the specific precursor complex, the type (mono- and bidentate phosphines, amines, alcohols) and structure of the organic (chiral) ligand, and in some cases the presence of additional additives. Especially for the design of chiral ligands there is practically no limit and as illustration a number of important chiral PP-ligands used for preparing soluble hydrogenation and isomerization catalysts is depicted in Fig. 2. Most (chiral) ligands are not commercially available. Therefore, chemists in research and development have to tap various sources to get the needed amounts of a desired ligand. Collaboration with university labs or industrial research groups is often useful for getting small amounts of experimental ligands. Own custom synthesis of ligand, especially of (chiral) phosphines requires considerable know how in organic synthesis and the handling of phosphor derivatives. For large-scale applications, ligand synthesis will at the moment be part of process development. Collaboration with an external manufacturer might be the method of choice for small and large quantities of ligand for long term projects.
2.4.1.2. Heterogeneous catalysts Of the many parameters of a heterogeneous hydrogenation catalyst that affect its catalytic performance, the following are
the most important ones: Type of metal (most often used Pd, Pt, Ni, Cu, Rh, Ru) ; type of catalyst (supported, powders, skeletal); metal loading of supported catalysts; type of support (active carbon, alumina, silica). Important parameters for the active metal are the surface area, the dispersion (typically only 10–60% of the metal atoms are exposed), the size of the crystallites (typically in the range 20 to > 200 Å), the location in the pores of the support and oxidation state (reduced or unreduced). Important support parameters are the particle size (for slurry catalysts typically 1–100 m), the surface area (typically in the range of 100–1500 m2 /g), the pore structure (pore volume, pore size distribution) and acid–base properties. Many types of heterogeneous catalysts are now available on a commercial basis. In our experience, it is of advantage to develop a close working relationship with several catalyst producers that specialize in catalysts for the fine chemical industry. 2.4.2. Catalyst modifiers or promoters In cases where a commercially available catalyst lacks a desired property or selectivity, the addition of a modifier is an interesting option. Both organic molecules (e.g. amines, chiral modifiers such as cinchona alkaloids or tartaric acid) or inorganic salts / metals are known for this purpose. The modifier can either be added to the catalysts before it is introduced into the reaction (often done with inorganic compounds) or added directly to the reaction mixture as process modifier. Factors that may be influenced are catalyst selectivity, activity, reduction of intermediate / side product formation and catalyst recovery. 2.4.3. Reaction conditions The catalyst performance can be optimized by choosing a suitable reaction system and the proper reaction conditions. Important parameters are the solvent, the temperature, in case of hydrogenation the hydrogen pressure, the concentration of the substrate and the catalyst, and process modifiers. Very often, the choice of the solvent is the most important of these parameters. The optimal solvent can improve catalyst performance. Furthermore, the separation of the catalyst and if necessary of the undesired enantiomer or racemate can be facilitated by the proper solvent system (see below). Last but not
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
195
Fig. 2. Structural characteristics of important chiral diphosphine ligands (P = PAr2 or PAlk2 ).
least the solvent also has to fit into the sequence of reactions. 2.4.4. Reaction control (end point) Monitoring the progress of a catalytic reaction can be difficult, especially if the catalyst is air-sensitive or the reaction carried out in an autoclave. Nevertheless, in the laboratory it is usually possible to find a suitable solution. This is by no means the case under the conditions of large scale production. There, one very often has to rely on relatively inaccurate measurements or defined reaction times. On-line monitoring of substrate and/or product concentrations could be of great help, especially for reaction where a precise end point control is crucial for high yield and / or selectivity. Here, ATR- and other in-line probes can sometimes be very useful for on-line spectroscopy.
3. Examples and short case studies 3.1. Chemoselective hydrogenations of aromatic nitro groups [2] 3.1.1. Development of the new catalytic systems Until recently, only the Béchamp reduction was available for the technical reduction of nitroarenes with additional, easily reducible functional groups (e.g., halide, C=C or C=O). When the catalysis process development team of Ciba-Geigy was faced with the task of developing a process for the reduction of the nitroallyl ester shown in Fig. 3, a completely new solution had to be found. Finally, two new catalytic systems that are both technically feasible were developed.
Both methods make use of catalyst modifiers. In the first process, a newly developed Pt–Pb–CaCO3 catalyst was used. In the second process, H3 PO2 was used as process modifier for a commercial Pt–C catalyst. The Pt–Pb-system gives the best results in methyl ethyl ketone (MEK) in presence of FeCl2 and tetrabutylammonium chloride. The hypophosphorous acid modified Pt–charcoal catalyst works best in toluene-water in the presence of VO(acac)2 as promoter. The iron or vanadium promoters serve to suppress the accumulation of hydroxylamines. The Béchamp process was used to make the first kg of material but was not developed further due to problems with filtration and work-up. In addition, large amounts of product remained adsorbed on the iron oxide that would present an enormous waste problem. Comment: As shown in Table 1, for the hydrogenation of the nitroallyl ester, the Pt–C/H3 PO2 system is clearly the most advantageous. It combines low waste with high selectivity, activity and space time yield. Furthermore, a relatively cheap commercially available Pt–C catalyst can be used and because toluene is the solvent used in the steps before and after reduction, no solvent change is necessary. The Pt–Pb–CaCO3 catalyst is of advantage when more polar solvents are required.
3.1.2. Scope of the new systems Both catalyst systems have a wide scope for related substrates (see Fig. 4), for details see [2]. They are for example able to hydrogenate aromatic nitro compounds containing functional groups such as iodide, C=C, C≡N and even C≡C with high yield and high selectivity.
196
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 3. Chemoselective hydrogenation of a nitroallyl ester.
Table 1 Comparison of important features of three reduction methods of nitro allyl esters
Reducing agent Solvent Reaction conditions Catalyst Modifiers Selectivity for allyl Yield Reaction time Important features
Critical factors
Pt–Pb–CaCO3
Pt–C with H3 PO2
B´echamp
Hydrogen Aprotic polar solvent, best results with MEK 140◦ C and 15 bar H2 Custom catalyst Lead, FeCl2 and N(Bu)4 Cl 99.8% >90% 5h low space time yield by-product formation change of solvent
Hydrogen apolar solvents, best results in toluene 100◦ C and 5 bar H2 Commercial catalyst H3 PO2 and VO(acac)2 99.9% >98% 2h no solvent change high space time yield
Iron EtOH/HCl/H2 O
catalyst preparation recycling of MEK
Fig. 4. Scope of the modified hydrogenation catalysts.
80◦ C – – 100% ca. 90% 18 h Contaminated iron oxide sludge Two filtrations Change of solvent EtOAc extraction Waste disposal
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
197
Fig. 5. Important substrates for nitro reduction.
3.2. Lowering of hydroxylamine accumulation in the hydrogenation of aromatic nitro groups [3,4] The catalytic hydrogenation of aromatic nitro groups proceeds via several intermediates and the most important is the corresponding hydroxylamine. Nitro compounds with electron withdrawing substituents, e.g., sulfonamide or halogen, accumulate hydroxylamines in large amounts during the reaction. This is especially critical when the hydrogenation is carried out in batch reactors at low or medium temperature because hydroxylamines are in many respects problematic (thermally unstable, explosions; strong carcinogens; by-product formation). The maximum concentration of hydroxylamines is notoriously difficult to predict. Here we show how with a process modifier/promoter the hydroxylamine problem can often be solved for a number of industrially important substrates (Fig. 5). A broad screening of transition metal salts showed that the addition of vanadium salts drastically reduced the accumulation of the hydroxylamine intermediate without increasing the reaction time in the hydrogenation of 1 (See Table 2 and [3,4]). This ‘vanadium’ effect has a relative wide scope and NH4 VO3 also works for Pt/C in the case of 2 or for the nitroallyl ester described in 3.1.3.. In the case Table 2 Hydroxylamine accumulation and reaction time for the hydrogenation of 1a Additive
Hydrogenation time (min)
Hydroxylamine accumulation
None NH4 VO3 V on Pd/C catalyst VOSO4
150 120 120 180
41% <1% <1% 2%
a 5% Pd/C, AcOH, 0.006 mol% promoter relative to 1, 120◦ C, 20 bar.
of 3, where a Pt/Cu/C catalyst was used, the vanadium had to be deposited on carbon for the desired effect. Without this measure, the catalyst was deactivated. Comment: This example shows how process-promoters allow to modify the properties of a catalyst to get a safer reaction, less by-products and in some cases shorter reaction times. Especially advantageous is the fact the commercially available catalysts can be used. 3.3. Alkylation of aromatic ring: combination of a homogeneous and heterogeneous Pd catalyzed reaction [5,6] 3.3.1. Process development Sodium 2-(3,3,3,-trifluoropropyl)-benzenesulfonate is a key intermediate for the sulfonylurea herbicide Prosulfuron® (Fig. 6). This and analogous building blocks were prepared in the Central Research Laboratories of Ciba-Geigy by the three step sequence diazotation, Matsuda arylation, hydrogenation. Attempts failed to find a classical synthetic method such as a Friedel-Crafts alkylation of benzenesulfonic acid. Therefore, the process development team had to develop a technically feasible production process using a technology that until that time was only used on a laboratory scale. In the end, a process was developed starting with 2-aminobenzenesulfonic acid and ending with sodium 2-(3,3,3-trifluoropropyl)-benzenesulfonate without isolation of the diazonium or olefin intermediates, producing only 2 kg wastes/kg product over the three consecutive synthetic steps equal to an E-factor of 2. Moreover, the yield over these three steps is 93%, i.e., an average of 98% per step (Fig. 7). 3.3.2. Important factors The solvent had to be compatible with three different chemical reactions, have high solubility for
198
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 6. Key intermediate and formula of Prosulfuron® herbicide.
Fig. 7. Three step one pot reaction for the synthesis of sodium 2-(3,3,3-trifluoropropyl)-benzenesulfonate.
trifluoropropene and be easy to regenerate. Pentan-1-ol showed these properties to a high degree. The cost and separation of the Pd catalyst was another crucial factor. By careful optimization of the reaction conditions catalyst loading for the arylation was lowered to 0.5–1.5% and the catalyst precursor Pd(dba)2 was prepared from readily available PdCl2 . The most crucial idea however, was to add charcoal after completion of the arylation reaction and to produce a heterogeneous hydrogenation catalyst in-situ that on the one hand is able to catalyze the hydrogenation of the C=C bond and at the same time allows an efficient separation of the Pd by simple filtration. Comment: By linking homogeneous and heterogeneous catalysis, and by using a one pot procedure for three consecutive steps, it was possible to develop an economically and ecologically sound process for the production of Prosulfuron®.
precursor, a chiral diphosphine ligand and H2 (20 bar) in MeOH. 3. The ␣-keto acid was reduced to the corresponding a-hydroxy acid with a d-lactate dehydrogenase (d-LDH), a formate/formate dehydrogenase (FDH) for the co-enzyme NAD regeneration and formate in an aqueous buffer using an enzyme membrane reactor. 4. The ␣,-unsaturated ␣-keto acid was reduced to the corresponding ␣-hydroxy acid with the microorganism Proteus mirabilis (immobilized), a co-factor and formate. The resulting ␣, -unsaturated ␣-hydroxy acid was hydrogenated with Pd/C to the saturated acid. Comment: The over-all performance and economy for (1), (3) and (4) are comparable, but each process has its specific problem(s) associated. Except for the homogenous hydrogenation, they are all technically feasible on a 1–100 kg scale.
3.4. Enantioselective hydrogenation of an ␣-keto ester [7] 3.5. Routes to an N-alkylated hindered aniline [8] (R)-2-Hydroxy-4-phenyl-butyric acid ethyl ester, the so-called HPB-ester, is an important intermediate for the synthesis of several ACE inhibitors. It can be synthesized from the corresponding ␣-keto acid derivatives via chemical or biochemical catalytic methods as depicted in Fig. 8 (Table 3). 1. The ␣-keto ester was catalytically hydrogenated to the corresponding a-hydroxy derivative with a Pt on Al2 O3 catalyst, a cinchona alkaloid (chiral modifier) and H2 (60 bar) in toluene or AcOH. 2. The ␣-keto ester was catalytically hydrogenated to the corresponding a-hydroxy ester with a Rh(I)
3.5.1. Background Metolachlor is the active ingredient of Dual® , one of the most important grass herbicides for use in maize and a number of other crops. It is an N-chloroacetylated, N-alkoxyalkylated ortho disubstituted aniline. In this context it can serve as an illustration how the requirements for the technical production of a large scale agrochemical can evolve (Table 4). The commercial product was introduced to the market in 1976 as a mixture of four stereoisomers (Fig. 9) and is produced via a Pt catalyzed reductive
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
199
Fig. 8. Catalytic routes to HPB ester. Table 3 Comparison of systems 1–4 for the synthesis of HPB ester Pt-Cinchona
Rh-PP
Dehydrogenase
Proteus mirabilis
‘Catalytic’ system Reducing agent % ee s/c (w/w) s/c (mol/mol) Ton tof (h−1 ) Space/time yield Cat. cost ($/kg p) Problems
Commercial heterogeneous catalyst H2 80–92 200 4000 4000 1000 210 1–2 Catalyst activation
Commercial enzymes HCOOH, NAD >99.9 25 000 N/A. High N/A. 7 Not known Long development time Complicated work-up
Immobilized microorganism HCOOH, co-factor >99 50–100 N/A. Living N/A. 5–12 Not known Long development time
Economy
process very sensitive to substrate purity Ok
Custom diphosphine H2 96 50 100 100 5 53 140–400 ee’s drops at higher s/c ratios optimization not successful bad
Ok
Ok
Complicated work-up
Table 4 Milestones in the history of metolachlor 1970 1978 1982 1983 1987 1993 1993/19944 1995/1996 1996
Discovery of the biological activity of metolachlor (patent for product and synthesis) Full-scale plant with a production capacity >10 000 t/year in operation Synthesis and biological tests of the four stereoisomers of metolachlor First unsuccessful attempts to synthesize (S)-metolachlor catalytically Ca–Sn–Pt catalyst for direct alkylation of MEA in the gas phase developed Ir/ferrocenyl diphosphine catalysts and acid effect discovered Patents for racemic-metolachlor expire Pilot results for (S)-metolachlor: ee 79%, ton 1 000 000, tof >200 000 h, first 300 t produced Full-scale plant for production of >10 000 t/year (S)-metolachlor starts operation
alkylation of 2-methyl-5-ethyl-aniline (MEA) with aqueous methoxyacetone (MOA, made by dehydrogenation of methoxyisopropanol, MOIP) in presence of traces of sulfuric acid followed by chloroacetylation [9] (see Fig. 10). A redox catalyst for the direct alkylation of MEA with MOIP was developed but not
used commercially (see Fig. 9) [10]. Already in 1982 it was found that about 95% of the herbicidal activity of metolachlor lies in the two (1’S)-diastereomers [11]. In 1997, after years of intensive research [12], Dual Magnum® with a content of approximately 90% (10 S)-diastereomers and with the same biologi-
200
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 9. Structure and stereoisomers of metolachlor.
Fig. 10. The production process for racemic metolachlor.
Fig. 11. Enantioselective MEA imine hydrogenation.
cal effect at about 65% of the use rate was introduced in USA. To make this ’chiral switch’ possible, a new technical process had to be found that allowed the economical production of the enantiomerically enriched precursor of metolachlor. Key step of this new synthesis is the enantioselective hydrogenation of N-(2-
ethyl-6-methylphenyl)-N-(10 -methoxymethyl)-ethylidene-amine (see Fig. 11). The optimized process operates at 80 bar hydrogen and 50◦ C with a catalyst generated in-situ from [Ir(cod)Cl]2 and the ferrocenyldiphosphine ligand xyliphos at a substrate to catalyst ratio (s/c) of >1 000 000. Complete conver-
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
201
Table 5 Comparison of important features of three methods to make hindered N-alkyl aniline
Catalyst Technology Status Reaction conditions Modifiers Conversion Chemoselectivity Important features Critical factors
Reductive alkylation
Direct alkylation
Enantioselective hydrogenation
5% Pt/C, commercial catalyst Liquid phase, batch production 50◦ C, 5 bar H2 >98% >98% High space time yield High catalyst recycling rate catalyst filtration recycling of starting material
Pt/SiO2 , custom catalyst Gas-phase, continuous laboratory 200◦ C, 1 bar H2 Sn, Ca 66% >98% one step reaction
Custom diphosphine Liquid phase, batch production 50◦ C, 80 bar H2 Acid, iodide 100 100% (80% ee) high space time yield extremely active catalyst imine quality
sion is reached within 3–4 h, the initial tof exceeds 1 800 000 h−1 and enantioselectivity is approximately 80%. 3.5.2. The production process for racemic metolachlor [8,13] For the reductive alkylation, several problems had to be solved: Competing reduction of MOA, especially when imine formation was slow; trace hydrogenation of the aromatic ring and re-use and separation of the Pt/C catalyst from the two-phase reaction mixture. After intensive development work, first in the laboratory (comprising >1500 experiments) and later in the pilot plant (also used for the production of the first commercial quantities), the production process depicted in Fig. 10 was established in early 1978. 3.5.3. Alternatives processes [10] Acid catalyzed alkylation of MEA with MOIP gave complex mixtures of the desired product (NAA), as well as N-methyl-, N-dimethyl, N-propyl- and N-isopropyl-MEA as by-products due to cleavage of the ether group of MOIP. More successful was the development of a multifunctional bimetallic platinum tin catalyst on a calcium treated silica support. This catalyst is able to catalyze the dehydrogenation of MOIP to MOA, the condensation to the MEA imine and the subsequent hydrogenation to produce NAA. It was shown that all three components as well as the silica support are absolutely necessary to get a good catalysts performance, i.e. there is a remarkable synergy between the various elements. At 200◦ C in the gas phase NAA was produced with >98% selectivity
catalyst preparation process control
at ca. 66% conversion over more than 1000 h. The process was not commercialized because large investments would have been necessary and also because the development of the catalyst would have been very difficult. 3.5.4. (S)-Metolachlor process [12] On the way to a successful process for the enantioselective hydrogenation of the imine intermediate (see Fig. 11) the following milestones are worthy of note: A rhodium diphosphine catalyst was found that was able to produce S-NAA with about 70% ee — quite close to the desired selectivity of 80%. However, low temperatures and large amounts of catalyst were needed to accomplish this. Iridium catalysts with ‘classical’ diphosphines proved to be as selective as the best Rh catalyst, were about 100 times more active but tended to deactivate relatively fast. The final breakthrough came in 1993: A new class of Ir ferrocenyl diphosphine complexes (see Fig. 11) turned out to be stable and the discovery of an extraordinary acid effect led to extremely active and productive catalysts. Scale up presented little problems and now the reaction is carried out on a scale of >10 000 t/year (Table 5). Comment: The three case histories demonstrate that catalytic methods are very suitable for the production of medium to large volumes fine chemicals even in an enantiomerically enriched form. It is worthy of note that the time for process development depends very much on the state of the art of a given catalytic technology. It was quite short for the reductive alkylation, a well known method already in 1972, whereas it took
202
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 12. Intermediates and by-products for the hydrogenation of 3-cyanopyridine.
Fig. 13. Dependence of 3-cyanopyridine concentration and ATR-response on hydrogen uptake.
more than 10 years to develop a suitable catalyst for the enantioselective imine hydrogenation. 3.6. On-line monitoring of catalytic hydrogenations with ATR-probes [14] As described in Section 2.4.4., accurate end point determination is often the key for obtaining high selectivity and therefore, high yield. Here, we describe how this can be done on-line for catalytic hydrogenations with the help of ATR-probes under difficult technical conditions. In both examples the technical hydrogenation was carried out under pressure with heterogeneous catalyst in steel autoclaves, where sampling is time consuming and complicated. ATR probes can also be very helpful for the development and optimization of a reaction and for identifying interme-
diates and by-products. Contrary to the examples below, an in-depth analysis of the data gathered is then necessary. 3.6.1. 3-Cyanopyridine to 3-Pyridinecarboxaldehyde 3-Pyridinecarboxaldehyde is a valuable intermediate for a number of industrial processes but expensive to buy. An inexpensive precursor is 3-cyanopyridine which can be converted to the desired product according to Fig. 12. However, many side- and consecutive reactions, e.g. the formation of primary and secondary amines are possible. Furthermore, the reaction does not stop on the aldehyde stage but slowly continues to give the corresponding alcohol. To obtain a high yield it is, therefore, essential to have a good method for end point determination. Comparing the change in adsorption at 240 nm with the concentration of the start-
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
203
Fig. 14. Hydrazone hydrogenation.
ing material reveals that a maximum in adsorption is observed when the starting material has disappeared completely (Fig. 13). This maximum coincides with the maximum concentration of the desired aldehyde. Therefore, the optimal end point can be determined very easily. 3.6.2. C= N double-bond hydrogenation in a acetyl-hydrazone derivative The hydrogenation of the hydrazone depicted in Fig. 14 is an important step in the production of a Novartis fungicide. One key problem is the determination of the end point of the reaction because prolonged hydrogenation leads to a cleavage of the N–N bond in the desired hydrazine. Again, the difference observed at 240 nm was shown to correlate with the concentration of the hydrazone and was directly related to the hydrogen uptake (data not shown). Therefore, the end point of the reaction was reached when the adsorption reached a plateau. If the reaction was not stopped at this point, the hydrogen uptake continued slowly, leading to undesired cleavage products. Comment: In both examples, the relative simple measurement at a single wavelength provided enough information for end point determination and yield optimization. Due to the high sensitivity of the method it is very important to carefully maintain the equipment and to critically analyze the data produced.
4. Conclusions and outlook We have illustrated with several examples how catalysis can help to optimize existing processes or
to open up new syntheses. We are convinced that the application of catalytic methods for the production of fine chemicals will increase in the coming years. It is clear, however, that due to the rather conservative nature of most people involved in applying chemical technology, on should not expect a radical change. Some new developments that are already being described in the literature might help to accelerate the rate of application: • Automated high through-put screening of catalysts and reaction conditions • More readily accessible catalyst families (homogeneous, heterogeneous, enzymes) with tunable properties and well defined scope and limitations • In-situ process control and improved (micro)- analytical techniques • Specialized (small) companies offering catalyst and process development to fine chemicals manufacturers [15].
References [1] R.A. Sheldon, Chem. Ind. (1992) 906. [2] U. Siegrist, P. Baumeister, H.U. Blaser, M. Studer, Chem. Ind. (Dekker) 75 (1998) 207. [3] P. Baumeister, H.U. Blaser, M. Studer, Catal. Lett. 49 (1997) 219. [4] P. Baumeister, M. Studer, WO 96/36597, 1996, assigned to Ciba-Geigy AG. [5] P. Baumeister, W. Meyer, K. Oertle, G. Seifert, U. Siegrist, H. Steiner, Stud. Surf. Sci. Catal. 108 (1997) 37. [6] P. Baumeister, G. Seifert, H. Steiner, EP 5840 043, 1992, assigned to Ciba-Geigy AG. [7] E. Schmidt, H.U. Blaser, P.F. Fauquex, G. Sedelmeier, F. Spindler, in: S. Servi (Ed.), Microbial Reagents in Organic Synthesis, Kluwer Academic Publishers, Dordrecht, 1992, p. 377.
204
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
[8] R.R. Bader, H.U. Blaser, Stud. Surf. Sci. Catal. 108 (1997) 17. [9] C. Vogel, R. Aebi, DP 23 28 340, 1972, assigned to Ciba-Geigy AG. [10] M. Rusek, Stud. Surf. Sci. Catal. 59 (1991) 359. [11] H. Moser, G. Ryhs, H. Sauter, Z. Naturforsch. 37b (1982) 451. [12] F. Spindler, B. Pugin, H.P. Jalett, H.P. Buser, U. Pittelkow, H.U. Blaser, Chem. Ind. (Dekker) 68 (1996) 153.
[13] R. Bader, P. Flatt, P. Radimerski, EP 605363-A1, 1992, assigned to Ciba-Geigy AG. [14] H. Danigel, N. Graber, M. Länzinger, M. Studer, H. Thies, A. Zilian, Chemical processing technology international, Special Achema Issue, 1997, p. 99. [15] H.-U. Blaser, M. Studer, Chimia. 53 (1999) 261.
The role of catalysis for the clean production of fine chemicals Hans-Ulrich Blaser ∗ , Martin Studer Novartis Services AG, Catalysis and Synthesis Services, WRO 1055.6, CH-4002 Basel, Switzerland
Abstract The role of catalysis for the production of fine chemicals is reviewed. The following topics are discussed on a general level: characteristics of the manufacture of fine chemicals, opportunities opened up by catalysis, critical factors for the application of catalysts and the tools that are available to the catalytic chemist. The general part is illustrated by specific examples from the catalysis group of Ciba-Geigy/Novartis such as chemoselective hydrogenations of aromatic nitro groups, the combination of a homogeneous and heterogeneous Pd catalyzed reaction for the alkylation of aromatic systems, catalytic systems for the enantioselective reduction of an ␣-keto ester, different routes to an N-alkylated hindered aniline including the (S)-metolachlor process, and the use of on-line monitoring of catalytic hydrogenations with ATR-probes. A short outlook on future developments is also presented. ©1999 Elsevier Science B.V. All rights reserved. Keywords: Technical catalysis; Fine chemicals production; Catalyst development; Process optimization; On-line monitoring
1. Introduction Traditionally, fine and specialty chemicals have been produced predominantly using non-catalytic organic synthesis. At least in the opinion of R. Sheldon, this is one important reason why besides the desired product, between 20 and 100 times as much waste is produced [1]. Of course many other factors such as the complexity of the molecules (and consequently, the number of synthetic steps), as well as the short development time for a (registered) technical synthesis and also the high requirements concerning purity of many fine chemicals are responsible for the unfavorable ecological situation. Nevertheless, the application of catalytic methods in the fine and specialty chemicals industry has increased in recent years in part because both produc∗ Corresponding author. E-mail address: [email protected] (H.-U. Blaser)
tion costs and waste minimization are of growing importance even for high value pharma and especially agro chemicals. In this contribution we will first give a short characterization of the problems in fine chemical production. Then we will describe the tools that are available to solve some of these problems with the help of catalysis and finally, we will present examples from our laboratory to illustrate some of these points.
2. Problems and opportunities 2.1. Characteristics of the manufacture of fine chemicals The manufacture of fine chemicals and especially of pharmaceuticals and agrochemicals can be characterized as follows (typical numbers are given in parenthesis):
0926-860X/99/$ – see front matter ©1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 6 - 8 6 0 X ( 9 9 ) 0 0 2 7 6 - 8
192
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
• Rather complex molecules (isomers, stereochemistry, several functional groups) with limited thermal stability. • Production via multistep syntheses (5– > 10 steps for pharmaceuticals and 3–7 for agrochemicals) with short product lives (often < 20 years). Usually classical organic reactions, catalysis as exception. • Production usually in solution, at ambient pressure and low to medium temperature in relatively small (500 l–10 m3 ) multipurpose batch equipment. • Relatively small scale products (1–1000 t/year for pharmaceuticals, 500–10 000 t/year for agrochemicals) • High purity requirements (usually >99% and <10 ppm metal residue and ee > 98% in pharmaceuticals). • High value added and therefore, tolerant to higher process cost (especially for very effective, small scale products). • Short development time for the production process (
2.2.3. Combining several transformation in one step • Reductive alkylation of amines with carbonyl compounds (imine not isolated) • Hydrogenation–acylation of nitro arenes to acylanilines • Direct alkylation of amines with alcohols via a dehydrogenation–condensation–hydrogenation sequence. 2.2.4. Replacing toxic or problematic reagents (and reactants) • Alkylation of amines or aromatics with alcohols instead of alkyl halides (reduction of salt production) • Use of H2 instead of metals, metal hydrides or sulfides • Use of H2 O2 or O2 instead of metal oxides or peracids • Solid acids and bases to replace soluble ones. 2.3. Critical factors for the application of catalysts An impressively large number of highly selective catalytic transformations is recorded in the literature that in principle can be applied to the synthesis of fine chemicals. However, quite many prerequisites must be fulfilled in order to render a catalytic process technically viable and to really profit from the opportunities described in the preceding chapter. 2.3.1. Catalyst performance The selectivity of a catalyst is probably its most cited property. Besides chemoselectivity, regio-, stereo- and enantioselectivity play a role for the synthesis of fine chemicals. Due to often high cost of starting materials and intermediates as well as of separation steps, selectivities >95% are usually required to make a catalytic method attractive. The catalyst productivity, given as turnover number (ton) or as substrate/catalyst ratio (s/c), determines catalyst costs. Here it is more difficult to give general numbers because both catalyst prices and the value added by the catalytic transformation play an important role. For enantioselective catalysts, tons ought to be >1000 for high value products and >50 000 for large scale or less expensive products (catalyst re-use increases the productivity). The catalyst activity (turnover frequency (tof), h−1 ), affects the production capacity. As a rule of thumb, tofs ought to
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
193
Fig. 1. Transformations that are only possible with metal catalysts.
be > 500 h−1 for small and >10 000 h−1 for large-scale products. 2.3.2. Substrate specificity Catalytic methods are often more substrate specific than stoichiometric ones, i.e., even small changes of the structure of the starting material can strongly affect the catalyst performance of a given catalyst. This is especially true for highly optimized stereo- and enantioselective catalysts, thereby leading to a low predictability for new substrates. This fact renders synthesis planning difficult for the synthetic chemist who by definition must be a generalist and often does not know the scope and limitations of catalytic methodologies. In our experience, catalytic methods are usually abandoned if they do not succeed almost at first try because development time and costs for fine chemicals are limited. 2.3.3. Commercial availability of catalysts and ligands In many cases this is still a major problem for products where relatively small amounts of catalysts are needed — usually not enough to make it worthwhile for a catalyst manufacturer to develop a tailor-made catalyst that might be required. For homogeneous catalyst both (chiral) ligands and many metal precursors can be expensive. Ligands often have unusual structures that are prepared via many step syntheses.
Heterogeneous catalysts can not be characterized on a molecular level. The catalytic properties of two catalysts with the same description, e.g., 5% Pt/C, can vary considerably. It is well known that even small variations in the preparation procedure or impurities can alter the structural and chemical properties of a heterogeneous catalyst significantly. This might or might not affect its catalytic or chemical activity. It is precisely this ‘might or might not’ that leads to problems of reproducibility and frustration. 2.3.4. Practical problems Especially homogeneous catalysts are very sensitive to air, reactor materials, handling etc. Most catalytic processes are also sensitive to the presence of impurities that can act as poisons or can modify the selectivity of a catalyst. This is due to the fact that the catalyst is present in very small concentration, that the active species are very reactive and that quite often oxygen sensitive ligands and complexes are employed. The quality of the starting material as well as of the reagents (solvent, gasses, etc.) is therefore, absolutely crucial. Catalyst separation from the reaction mixture is a problem that has to be addressed for every homogeneous catalyst. Usually, the solution has to be tailored to the substrate, solvent, catalyst type etc. Distillation, crystallization, extraction or immobilization of the soluble catalyst are often successful approaches. Solid catalysts can be separated from the reaction mixture
194
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
by simple filtration. This allows easier work up and isolation of the desired product and is the most obvious advantage of a heterogeneous catalyst. Leaching of the active species such as noble metals can contaminate products and require an expensive purification. 2.4. The toolbox of the catalytic chemist In the following chapter we sketch the tools that are available for developing catalysts that make a desired transformation technically feasible. We use the example of homogeneous and heterogeneous metal catalysts for this endeavor. 2.4.1. Design parameters for catalytic systems 2.4.1.1. Homogeneous catalysts For homogeneous catalysts the most important elements are the type of metal (very often used are Pd, Rh, Ru, Ir, Ni, Cu, Ti) and the specific precursor complex, the type (mono- and bidentate phosphines, amines, alcohols) and structure of the organic (chiral) ligand, and in some cases the presence of additional additives. Especially for the design of chiral ligands there is practically no limit and as illustration a number of important chiral PP-ligands used for preparing soluble hydrogenation and isomerization catalysts is depicted in Fig. 2. Most (chiral) ligands are not commercially available. Therefore, chemists in research and development have to tap various sources to get the needed amounts of a desired ligand. Collaboration with university labs or industrial research groups is often useful for getting small amounts of experimental ligands. Own custom synthesis of ligand, especially of (chiral) phosphines requires considerable know how in organic synthesis and the handling of phosphor derivatives. For large-scale applications, ligand synthesis will at the moment be part of process development. Collaboration with an external manufacturer might be the method of choice for small and large quantities of ligand for long term projects.
2.4.1.2. Heterogeneous catalysts Of the many parameters of a heterogeneous hydrogenation catalyst that affect its catalytic performance, the following are
the most important ones: Type of metal (most often used Pd, Pt, Ni, Cu, Rh, Ru) ; type of catalyst (supported, powders, skeletal); metal loading of supported catalysts; type of support (active carbon, alumina, silica). Important parameters for the active metal are the surface area, the dispersion (typically only 10–60% of the metal atoms are exposed), the size of the crystallites (typically in the range 20 to > 200 Å), the location in the pores of the support and oxidation state (reduced or unreduced). Important support parameters are the particle size (for slurry catalysts typically 1–100 m), the surface area (typically in the range of 100–1500 m2 /g), the pore structure (pore volume, pore size distribution) and acid–base properties. Many types of heterogeneous catalysts are now available on a commercial basis. In our experience, it is of advantage to develop a close working relationship with several catalyst producers that specialize in catalysts for the fine chemical industry. 2.4.2. Catalyst modifiers or promoters In cases where a commercially available catalyst lacks a desired property or selectivity, the addition of a modifier is an interesting option. Both organic molecules (e.g. amines, chiral modifiers such as cinchona alkaloids or tartaric acid) or inorganic salts / metals are known for this purpose. The modifier can either be added to the catalysts before it is introduced into the reaction (often done with inorganic compounds) or added directly to the reaction mixture as process modifier. Factors that may be influenced are catalyst selectivity, activity, reduction of intermediate / side product formation and catalyst recovery. 2.4.3. Reaction conditions The catalyst performance can be optimized by choosing a suitable reaction system and the proper reaction conditions. Important parameters are the solvent, the temperature, in case of hydrogenation the hydrogen pressure, the concentration of the substrate and the catalyst, and process modifiers. Very often, the choice of the solvent is the most important of these parameters. The optimal solvent can improve catalyst performance. Furthermore, the separation of the catalyst and if necessary of the undesired enantiomer or racemate can be facilitated by the proper solvent system (see below). Last but not
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
195
Fig. 2. Structural characteristics of important chiral diphosphine ligands (P = PAr2 or PAlk2 ).
least the solvent also has to fit into the sequence of reactions. 2.4.4. Reaction control (end point) Monitoring the progress of a catalytic reaction can be difficult, especially if the catalyst is air-sensitive or the reaction carried out in an autoclave. Nevertheless, in the laboratory it is usually possible to find a suitable solution. This is by no means the case under the conditions of large scale production. There, one very often has to rely on relatively inaccurate measurements or defined reaction times. On-line monitoring of substrate and/or product concentrations could be of great help, especially for reaction where a precise end point control is crucial for high yield and / or selectivity. Here, ATR- and other in-line probes can sometimes be very useful for on-line spectroscopy.
3. Examples and short case studies 3.1. Chemoselective hydrogenations of aromatic nitro groups [2] 3.1.1. Development of the new catalytic systems Until recently, only the Béchamp reduction was available for the technical reduction of nitroarenes with additional, easily reducible functional groups (e.g., halide, C=C or C=O). When the catalysis process development team of Ciba-Geigy was faced with the task of developing a process for the reduction of the nitroallyl ester shown in Fig. 3, a completely new solution had to be found. Finally, two new catalytic systems that are both technically feasible were developed.
Both methods make use of catalyst modifiers. In the first process, a newly developed Pt–Pb–CaCO3 catalyst was used. In the second process, H3 PO2 was used as process modifier for a commercial Pt–C catalyst. The Pt–Pb-system gives the best results in methyl ethyl ketone (MEK) in presence of FeCl2 and tetrabutylammonium chloride. The hypophosphorous acid modified Pt–charcoal catalyst works best in toluene-water in the presence of VO(acac)2 as promoter. The iron or vanadium promoters serve to suppress the accumulation of hydroxylamines. The Béchamp process was used to make the first kg of material but was not developed further due to problems with filtration and work-up. In addition, large amounts of product remained adsorbed on the iron oxide that would present an enormous waste problem. Comment: As shown in Table 1, for the hydrogenation of the nitroallyl ester, the Pt–C/H3 PO2 system is clearly the most advantageous. It combines low waste with high selectivity, activity and space time yield. Furthermore, a relatively cheap commercially available Pt–C catalyst can be used and because toluene is the solvent used in the steps before and after reduction, no solvent change is necessary. The Pt–Pb–CaCO3 catalyst is of advantage when more polar solvents are required.
3.1.2. Scope of the new systems Both catalyst systems have a wide scope for related substrates (see Fig. 4), for details see [2]. They are for example able to hydrogenate aromatic nitro compounds containing functional groups such as iodide, C=C, C≡N and even C≡C with high yield and high selectivity.
196
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 3. Chemoselective hydrogenation of a nitroallyl ester.
Table 1 Comparison of important features of three reduction methods of nitro allyl esters
Reducing agent Solvent Reaction conditions Catalyst Modifiers Selectivity for allyl Yield Reaction time Important features
Critical factors
Pt–Pb–CaCO3
Pt–C with H3 PO2
B´echamp
Hydrogen Aprotic polar solvent, best results with MEK 140◦ C and 15 bar H2 Custom catalyst Lead, FeCl2 and N(Bu)4 Cl 99.8% >90% 5h low space time yield by-product formation change of solvent
Hydrogen apolar solvents, best results in toluene 100◦ C and 5 bar H2 Commercial catalyst H3 PO2 and VO(acac)2 99.9% >98% 2h no solvent change high space time yield
Iron EtOH/HCl/H2 O
catalyst preparation recycling of MEK
Fig. 4. Scope of the modified hydrogenation catalysts.
80◦ C – – 100% ca. 90% 18 h Contaminated iron oxide sludge Two filtrations Change of solvent EtOAc extraction Waste disposal
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
197
Fig. 5. Important substrates for nitro reduction.
3.2. Lowering of hydroxylamine accumulation in the hydrogenation of aromatic nitro groups [3,4] The catalytic hydrogenation of aromatic nitro groups proceeds via several intermediates and the most important is the corresponding hydroxylamine. Nitro compounds with electron withdrawing substituents, e.g., sulfonamide or halogen, accumulate hydroxylamines in large amounts during the reaction. This is especially critical when the hydrogenation is carried out in batch reactors at low or medium temperature because hydroxylamines are in many respects problematic (thermally unstable, explosions; strong carcinogens; by-product formation). The maximum concentration of hydroxylamines is notoriously difficult to predict. Here we show how with a process modifier/promoter the hydroxylamine problem can often be solved for a number of industrially important substrates (Fig. 5). A broad screening of transition metal salts showed that the addition of vanadium salts drastically reduced the accumulation of the hydroxylamine intermediate without increasing the reaction time in the hydrogenation of 1 (See Table 2 and [3,4]). This ‘vanadium’ effect has a relative wide scope and NH4 VO3 also works for Pt/C in the case of 2 or for the nitroallyl ester described in 3.1.3.. In the case Table 2 Hydroxylamine accumulation and reaction time for the hydrogenation of 1a Additive
Hydrogenation time (min)
Hydroxylamine accumulation
None NH4 VO3 V on Pd/C catalyst VOSO4
150 120 120 180
41% <1% <1% 2%
a 5% Pd/C, AcOH, 0.006 mol% promoter relative to 1, 120◦ C, 20 bar.
of 3, where a Pt/Cu/C catalyst was used, the vanadium had to be deposited on carbon for the desired effect. Without this measure, the catalyst was deactivated. Comment: This example shows how process-promoters allow to modify the properties of a catalyst to get a safer reaction, less by-products and in some cases shorter reaction times. Especially advantageous is the fact the commercially available catalysts can be used. 3.3. Alkylation of aromatic ring: combination of a homogeneous and heterogeneous Pd catalyzed reaction [5,6] 3.3.1. Process development Sodium 2-(3,3,3,-trifluoropropyl)-benzenesulfonate is a key intermediate for the sulfonylurea herbicide Prosulfuron® (Fig. 6). This and analogous building blocks were prepared in the Central Research Laboratories of Ciba-Geigy by the three step sequence diazotation, Matsuda arylation, hydrogenation. Attempts failed to find a classical synthetic method such as a Friedel-Crafts alkylation of benzenesulfonic acid. Therefore, the process development team had to develop a technically feasible production process using a technology that until that time was only used on a laboratory scale. In the end, a process was developed starting with 2-aminobenzenesulfonic acid and ending with sodium 2-(3,3,3-trifluoropropyl)-benzenesulfonate without isolation of the diazonium or olefin intermediates, producing only 2 kg wastes/kg product over the three consecutive synthetic steps equal to an E-factor of 2. Moreover, the yield over these three steps is 93%, i.e., an average of 98% per step (Fig. 7). 3.3.2. Important factors The solvent had to be compatible with three different chemical reactions, have high solubility for
198
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 6. Key intermediate and formula of Prosulfuron® herbicide.
Fig. 7. Three step one pot reaction for the synthesis of sodium 2-(3,3,3-trifluoropropyl)-benzenesulfonate.
trifluoropropene and be easy to regenerate. Pentan-1-ol showed these properties to a high degree. The cost and separation of the Pd catalyst was another crucial factor. By careful optimization of the reaction conditions catalyst loading for the arylation was lowered to 0.5–1.5% and the catalyst precursor Pd(dba)2 was prepared from readily available PdCl2 . The most crucial idea however, was to add charcoal after completion of the arylation reaction and to produce a heterogeneous hydrogenation catalyst in-situ that on the one hand is able to catalyze the hydrogenation of the C=C bond and at the same time allows an efficient separation of the Pd by simple filtration. Comment: By linking homogeneous and heterogeneous catalysis, and by using a one pot procedure for three consecutive steps, it was possible to develop an economically and ecologically sound process for the production of Prosulfuron®.
precursor, a chiral diphosphine ligand and H2 (20 bar) in MeOH. 3. The ␣-keto acid was reduced to the corresponding a-hydroxy acid with a d-lactate dehydrogenase (d-LDH), a formate/formate dehydrogenase (FDH) for the co-enzyme NAD regeneration and formate in an aqueous buffer using an enzyme membrane reactor. 4. The ␣,-unsaturated ␣-keto acid was reduced to the corresponding ␣-hydroxy acid with the microorganism Proteus mirabilis (immobilized), a co-factor and formate. The resulting ␣, -unsaturated ␣-hydroxy acid was hydrogenated with Pd/C to the saturated acid. Comment: The over-all performance and economy for (1), (3) and (4) are comparable, but each process has its specific problem(s) associated. Except for the homogenous hydrogenation, they are all technically feasible on a 1–100 kg scale.
3.4. Enantioselective hydrogenation of an ␣-keto ester [7] 3.5. Routes to an N-alkylated hindered aniline [8] (R)-2-Hydroxy-4-phenyl-butyric acid ethyl ester, the so-called HPB-ester, is an important intermediate for the synthesis of several ACE inhibitors. It can be synthesized from the corresponding ␣-keto acid derivatives via chemical or biochemical catalytic methods as depicted in Fig. 8 (Table 3). 1. The ␣-keto ester was catalytically hydrogenated to the corresponding a-hydroxy derivative with a Pt on Al2 O3 catalyst, a cinchona alkaloid (chiral modifier) and H2 (60 bar) in toluene or AcOH. 2. The ␣-keto ester was catalytically hydrogenated to the corresponding a-hydroxy ester with a Rh(I)
3.5.1. Background Metolachlor is the active ingredient of Dual® , one of the most important grass herbicides for use in maize and a number of other crops. It is an N-chloroacetylated, N-alkoxyalkylated ortho disubstituted aniline. In this context it can serve as an illustration how the requirements for the technical production of a large scale agrochemical can evolve (Table 4). The commercial product was introduced to the market in 1976 as a mixture of four stereoisomers (Fig. 9) and is produced via a Pt catalyzed reductive
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
199
Fig. 8. Catalytic routes to HPB ester. Table 3 Comparison of systems 1–4 for the synthesis of HPB ester Pt-Cinchona
Rh-PP
Dehydrogenase
Proteus mirabilis
‘Catalytic’ system Reducing agent % ee s/c (w/w) s/c (mol/mol) Ton tof (h−1 ) Space/time yield Cat. cost ($/kg p) Problems
Commercial heterogeneous catalyst H2 80–92 200 4000 4000 1000 210 1–2 Catalyst activation
Commercial enzymes HCOOH, NAD >99.9 25 000 N/A. High N/A. 7 Not known Long development time Complicated work-up
Immobilized microorganism HCOOH, co-factor >99 50–100 N/A. Living N/A. 5–12 Not known Long development time
Economy
process very sensitive to substrate purity Ok
Custom diphosphine H2 96 50 100 100 5 53 140–400 ee’s drops at higher s/c ratios optimization not successful bad
Ok
Ok
Complicated work-up
Table 4 Milestones in the history of metolachlor 1970 1978 1982 1983 1987 1993 1993/19944 1995/1996 1996
Discovery of the biological activity of metolachlor (patent for product and synthesis) Full-scale plant with a production capacity >10 000 t/year in operation Synthesis and biological tests of the four stereoisomers of metolachlor First unsuccessful attempts to synthesize (S)-metolachlor catalytically Ca–Sn–Pt catalyst for direct alkylation of MEA in the gas phase developed Ir/ferrocenyl diphosphine catalysts and acid effect discovered Patents for racemic-metolachlor expire Pilot results for (S)-metolachlor: ee 79%, ton 1 000 000, tof >200 000 h, first 300 t produced Full-scale plant for production of >10 000 t/year (S)-metolachlor starts operation
alkylation of 2-methyl-5-ethyl-aniline (MEA) with aqueous methoxyacetone (MOA, made by dehydrogenation of methoxyisopropanol, MOIP) in presence of traces of sulfuric acid followed by chloroacetylation [9] (see Fig. 10). A redox catalyst for the direct alkylation of MEA with MOIP was developed but not
used commercially (see Fig. 9) [10]. Already in 1982 it was found that about 95% of the herbicidal activity of metolachlor lies in the two (1’S)-diastereomers [11]. In 1997, after years of intensive research [12], Dual Magnum® with a content of approximately 90% (10 S)-diastereomers and with the same biologi-
200
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 9. Structure and stereoisomers of metolachlor.
Fig. 10. The production process for racemic metolachlor.
Fig. 11. Enantioselective MEA imine hydrogenation.
cal effect at about 65% of the use rate was introduced in USA. To make this ’chiral switch’ possible, a new technical process had to be found that allowed the economical production of the enantiomerically enriched precursor of metolachlor. Key step of this new synthesis is the enantioselective hydrogenation of N-(2-
ethyl-6-methylphenyl)-N-(10 -methoxymethyl)-ethylidene-amine (see Fig. 11). The optimized process operates at 80 bar hydrogen and 50◦ C with a catalyst generated in-situ from [Ir(cod)Cl]2 and the ferrocenyldiphosphine ligand xyliphos at a substrate to catalyst ratio (s/c) of >1 000 000. Complete conver-
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
201
Table 5 Comparison of important features of three methods to make hindered N-alkyl aniline
Catalyst Technology Status Reaction conditions Modifiers Conversion Chemoselectivity Important features Critical factors
Reductive alkylation
Direct alkylation
Enantioselective hydrogenation
5% Pt/C, commercial catalyst Liquid phase, batch production 50◦ C, 5 bar H2 >98% >98% High space time yield High catalyst recycling rate catalyst filtration recycling of starting material
Pt/SiO2 , custom catalyst Gas-phase, continuous laboratory 200◦ C, 1 bar H2 Sn, Ca 66% >98% one step reaction
Custom diphosphine Liquid phase, batch production 50◦ C, 80 bar H2 Acid, iodide 100 100% (80% ee) high space time yield extremely active catalyst imine quality
sion is reached within 3–4 h, the initial tof exceeds 1 800 000 h−1 and enantioselectivity is approximately 80%. 3.5.2. The production process for racemic metolachlor [8,13] For the reductive alkylation, several problems had to be solved: Competing reduction of MOA, especially when imine formation was slow; trace hydrogenation of the aromatic ring and re-use and separation of the Pt/C catalyst from the two-phase reaction mixture. After intensive development work, first in the laboratory (comprising >1500 experiments) and later in the pilot plant (also used for the production of the first commercial quantities), the production process depicted in Fig. 10 was established in early 1978. 3.5.3. Alternatives processes [10] Acid catalyzed alkylation of MEA with MOIP gave complex mixtures of the desired product (NAA), as well as N-methyl-, N-dimethyl, N-propyl- and N-isopropyl-MEA as by-products due to cleavage of the ether group of MOIP. More successful was the development of a multifunctional bimetallic platinum tin catalyst on a calcium treated silica support. This catalyst is able to catalyze the dehydrogenation of MOIP to MOA, the condensation to the MEA imine and the subsequent hydrogenation to produce NAA. It was shown that all three components as well as the silica support are absolutely necessary to get a good catalysts performance, i.e. there is a remarkable synergy between the various elements. At 200◦ C in the gas phase NAA was produced with >98% selectivity
catalyst preparation process control
at ca. 66% conversion over more than 1000 h. The process was not commercialized because large investments would have been necessary and also because the development of the catalyst would have been very difficult. 3.5.4. (S)-Metolachlor process [12] On the way to a successful process for the enantioselective hydrogenation of the imine intermediate (see Fig. 11) the following milestones are worthy of note: A rhodium diphosphine catalyst was found that was able to produce S-NAA with about 70% ee — quite close to the desired selectivity of 80%. However, low temperatures and large amounts of catalyst were needed to accomplish this. Iridium catalysts with ‘classical’ diphosphines proved to be as selective as the best Rh catalyst, were about 100 times more active but tended to deactivate relatively fast. The final breakthrough came in 1993: A new class of Ir ferrocenyl diphosphine complexes (see Fig. 11) turned out to be stable and the discovery of an extraordinary acid effect led to extremely active and productive catalysts. Scale up presented little problems and now the reaction is carried out on a scale of >10 000 t/year (Table 5). Comment: The three case histories demonstrate that catalytic methods are very suitable for the production of medium to large volumes fine chemicals even in an enantiomerically enriched form. It is worthy of note that the time for process development depends very much on the state of the art of a given catalytic technology. It was quite short for the reductive alkylation, a well known method already in 1972, whereas it took
202
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
Fig. 12. Intermediates and by-products for the hydrogenation of 3-cyanopyridine.
Fig. 13. Dependence of 3-cyanopyridine concentration and ATR-response on hydrogen uptake.
more than 10 years to develop a suitable catalyst for the enantioselective imine hydrogenation. 3.6. On-line monitoring of catalytic hydrogenations with ATR-probes [14] As described in Section 2.4.4., accurate end point determination is often the key for obtaining high selectivity and therefore, high yield. Here, we describe how this can be done on-line for catalytic hydrogenations with the help of ATR-probes under difficult technical conditions. In both examples the technical hydrogenation was carried out under pressure with heterogeneous catalyst in steel autoclaves, where sampling is time consuming and complicated. ATR probes can also be very helpful for the development and optimization of a reaction and for identifying interme-
diates and by-products. Contrary to the examples below, an in-depth analysis of the data gathered is then necessary. 3.6.1. 3-Cyanopyridine to 3-Pyridinecarboxaldehyde 3-Pyridinecarboxaldehyde is a valuable intermediate for a number of industrial processes but expensive to buy. An inexpensive precursor is 3-cyanopyridine which can be converted to the desired product according to Fig. 12. However, many side- and consecutive reactions, e.g. the formation of primary and secondary amines are possible. Furthermore, the reaction does not stop on the aldehyde stage but slowly continues to give the corresponding alcohol. To obtain a high yield it is, therefore, essential to have a good method for end point determination. Comparing the change in adsorption at 240 nm with the concentration of the start-
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
203
Fig. 14. Hydrazone hydrogenation.
ing material reveals that a maximum in adsorption is observed when the starting material has disappeared completely (Fig. 13). This maximum coincides with the maximum concentration of the desired aldehyde. Therefore, the optimal end point can be determined very easily. 3.6.2. C= N double-bond hydrogenation in a acetyl-hydrazone derivative The hydrogenation of the hydrazone depicted in Fig. 14 is an important step in the production of a Novartis fungicide. One key problem is the determination of the end point of the reaction because prolonged hydrogenation leads to a cleavage of the N–N bond in the desired hydrazine. Again, the difference observed at 240 nm was shown to correlate with the concentration of the hydrazone and was directly related to the hydrogen uptake (data not shown). Therefore, the end point of the reaction was reached when the adsorption reached a plateau. If the reaction was not stopped at this point, the hydrogen uptake continued slowly, leading to undesired cleavage products. Comment: In both examples, the relative simple measurement at a single wavelength provided enough information for end point determination and yield optimization. Due to the high sensitivity of the method it is very important to carefully maintain the equipment and to critically analyze the data produced.
4. Conclusions and outlook We have illustrated with several examples how catalysis can help to optimize existing processes or
to open up new syntheses. We are convinced that the application of catalytic methods for the production of fine chemicals will increase in the coming years. It is clear, however, that due to the rather conservative nature of most people involved in applying chemical technology, on should not expect a radical change. Some new developments that are already being described in the literature might help to accelerate the rate of application: • Automated high through-put screening of catalysts and reaction conditions • More readily accessible catalyst families (homogeneous, heterogeneous, enzymes) with tunable properties and well defined scope and limitations • In-situ process control and improved (micro)- analytical techniques • Specialized (small) companies offering catalyst and process development to fine chemicals manufacturers [15].
References [1] R.A. Sheldon, Chem. Ind. (1992) 906. [2] U. Siegrist, P. Baumeister, H.U. Blaser, M. Studer, Chem. Ind. (Dekker) 75 (1998) 207. [3] P. Baumeister, H.U. Blaser, M. Studer, Catal. Lett. 49 (1997) 219. [4] P. Baumeister, M. Studer, WO 96/36597, 1996, assigned to Ciba-Geigy AG. [5] P. Baumeister, W. Meyer, K. Oertle, G. Seifert, U. Siegrist, H. Steiner, Stud. Surf. Sci. Catal. 108 (1997) 37. [6] P. Baumeister, G. Seifert, H. Steiner, EP 5840 043, 1992, assigned to Ciba-Geigy AG. [7] E. Schmidt, H.U. Blaser, P.F. Fauquex, G. Sedelmeier, F. Spindler, in: S. Servi (Ed.), Microbial Reagents in Organic Synthesis, Kluwer Academic Publishers, Dordrecht, 1992, p. 377.
204
H.-U. Blaser, M. Studer / Applied Catalysis A: General 189 (1999) 191–204
[8] R.R. Bader, H.U. Blaser, Stud. Surf. Sci. Catal. 108 (1997) 17. [9] C. Vogel, R. Aebi, DP 23 28 340, 1972, assigned to Ciba-Geigy AG. [10] M. Rusek, Stud. Surf. Sci. Catal. 59 (1991) 359. [11] H. Moser, G. Ryhs, H. Sauter, Z. Naturforsch. 37b (1982) 451. [12] F. Spindler, B. Pugin, H.P. Jalett, H.P. Buser, U. Pittelkow, H.U. Blaser, Chem. Ind. (Dekker) 68 (1996) 153.
[13] R. Bader, P. Flatt, P. Radimerski, EP 605363-A1, 1992, assigned to Ciba-Geigy AG. [14] H. Danigel, N. Graber, M. Länzinger, M. Studer, H. Thies, A. Zilian, Chemical processing technology international, Special Achema Issue, 1997, p. 99. [15] H.-U. Blaser, M. Studer, Chimia. 53 (1999) 261.