Biocatalysis at extreme temperatures: enantioselective 2 synthesis of both enantiomers of mandelic acid by 3 transterification catalyzed by a thermophilic lipase in 4 ionic liquids at 120 °C. 5

: The use of biocatalysts in organic chemistry for catalysing chemo-, regio- and stereo- 11 selective transformations has become an usual tool in the last years, both at lab and industrial scale, 12 not only because of their exquisite precision, but also due to the inherent increase in the process 13 sustainability. Nevertheless, most of the interesting industrial reactions involve water-insoluble 14 substrates, so that the use of (generally not green) organic solvents is generally required. Although 15 lipases are perfectly capable of maintaining their catalytic precision working in those solvents, 16 reactions are usually very slow and consequently not very appropriate for industrial purposes. 17 Thus, the use of thermophilic enzymes at high temperatures can help in accelerating reaction rates. 18 In this paper we describe the use of lipase from Geobacillus thermocatenolatus as catalyst in the 19 ethanolysis of racemic 2-(butyryloxy)-2-phenylacetic to furnish both enantiomers of mandelic acid, 20 an useful intermediate in the synthesis of many drugs and active products. The catalytic 21 performance at high temperature in a conventional organic solvent ( iso octane) and four 22 imidazolium-based ionic liquids has been assessed. Best results were obtained using 1-ethyl-3- 23 methyl imidazolium tetrafluoroborate (EMIMBF 4 ) and 1-ethyl-3-methyl imidazolium 24 hexafluorophosphate (EMIMPF 6 ) at temperatures as high as 120°C, observing in both cases very fast 25 and exquisite enantioselective kinetic resolutions, respectively leading exclusively to the ( S ) or to 26 the ( R )-enantiomer of mandelic acid, depending on the anion component of the ionic liquid. 27


31
The employ of biocatalysts in organic chemistry, either alone [1,2] or combined with chemical 32 catalysts [3,4] for developing selective transformations has become an usual tool in the last years [5,6].

38
In fact, one of the green credentials of biocatalysis derives from the fact that biotransformations 39 can be conducted under very mild reaction conditions, e.g., atmospheric pressure, room temperature 40 or aqueous media. Anyhow, harsh conditions required for many industrial processes, such as high 41 temperature and/or the use of organic (co)solvents may impede the use of some enzymes. For solving

124
Nevertheless, hydrolysis is not a good alternative for checking the real thermotolerance of a 125 lipase, as the stability of these enzymes is much higher when they are working on organic solvents 126 [33,41,94]. Thus, we decided to use a water-free reaction media, selecting EtOH as nucleophile instead 127 of water and a water-insoluble organic solvent (Figure 1(a)). Opting for EtOH was based on our

160
As can be seen from Fig.1 1, the reaction at 40°C (Figure 1a) proceeds very slowly, reaching a 161 global conversion of around 10% for (S)-2 after 400 h and 3-4% for the (R)-2 counterpart. Also, a very 162 strong lag-time is observed for both enantiomers, not detecting any trace of ethanolysis of (R, S)-1 in 163 the first 75 hours. A similar behavior can be found for the generation of (R)-2 at 70°C (Figure 1b), 164 while for (S)-2 at that temperature and for both enantiomers at 90°C (Figure 1c)

189
Overall, best results are those obtained at 70°C at short reaction times. In fact, inside the time 190 interval from 0 to 75h, the enantioselectivity is almost perfect, although at the expenses of a low 191 overall conversions (around 10%, Figure 1b). Nevertheless, as these results are quite unsatisfactory, 192 the use of room-temperature ionic liquids (RTILs) as reaction media was tested.

251
Another interesting aspect to be taken into account is that, while the generation (R)-2 remains 252 constant after a certain reaction time (around 30% after 50 h, according to the sigmoid fitting), the 253 production of (S)-2 is slowly increasing after that time, although at a lower reaction rate than that 254 observed at the early stages; that is the reason why the overall (S)-2 production follows a double 255 exponential fitting. This slower second reaction rate could be caused by an inhibition promoted by 256 the increasing amounts of (R)-2 present in the reaction media, as the slope change in the (S)-2 257 production is observed only after a certain accumulation of (R)-2.

298
As can be seen, the behavior is somehow similar to that observed using BMIMBF4, hence, when 299 performing the ethanolysis at 90°C (Figure 3a), the (S) enantiomer of mandelic acid is more quickly 300 produced, once again following a double exponential fit, and displaying a smaller initial rate than 301 that one obtained using BMIMBF4., but also higher if compared to that calculated using isooctane 302 (Table 1). Similarly, the inversion in the stereobias at 120°C is also detected, but now the reaction 303 proceeded very poorly compared to that depicted in Figure 2b, as only a maximum of 6% conversion   Figure 4a, the lag-time for the detection of (R)-2 is slightly higher than that observed using 328 BMIMPF4, and also the initial rate is somewhat higher (4,08 vs 3.85 mM h -1 , Table 2), so that it was 329 possible to detect only (S)-2 in the first 12 hours, reaching a concentration of 18.2 mM (around 30% 330 conversion) with a perfect enantioselectivity Once again, as the reaction proceeded and the (R)-2 331 enantiomer is being produced (lag-time and single exponential models almost similar), the rate of 332 production of (S)-2 was reduced, so that once again the overall behavior for (R)-2 can be fitted to a 333 double exponential curve.

334
When the reaction temperature is increased up to 120°C (Figure 4b), a fantastic and fast kinetic 335 resolution can be observed. In fact, the initial rate in the generation of (S)-2 (VS =53.4 mM h -1 , Table 2) 336 is one order of magnitude higher than that obtained at 90°C; moreover, no traces of (R)-2 were 337 detected during the whole reaction time, so that the shape of the progress curve fits to a single 338 exponential plot leading to around 50% conversion (the maximum for a kinetic resolution) in only 5 339 hours.

340
Results obtained using EMIMPF6 as solvents are depicted in Figure 5. As can be seen, at both 341 temperatures only small traces of (R)-2 were detected, while the generation of (S)-2 follows single 342 exponential kinetics. At 90°C, the kinetic resolution observed using this solvent (Figure 5a) is slower 343 (VS around one half, See Table 2, entries #8 vs #10) than that obtained with EMIMBF4 (Figure 4a),

344
although this fact is compensated with an absence of production of (R)-2, so that the kinetic resolution 345 is considerably better in terms of enantioselectivity, allowing around a 30% conversion after 250 h.

346
When increasing the temperature to 120°C (Figure 5b), the kinetic resolution was slightly slower and 347 the maxinum conversion was half that obtained at 90°C.

400
It has been also described that polar solvents (water and short-chain alcohols) lead to enhanced As can be seen from Table 3, the molar composition of the reaction mixture is not exactly the 411 same, because of the differences in the molecular weight and density of the four RTILs, although 412 average values of (0.56±0.04) and (0.44±0.04) for XRTIL and XEtOH can be considered.

413
The most hydrophobic RTIL, water-insoluble BMIMPF6, is definitively not the best option for 414 the ethanolysis of (R,S)-1, as shown in Figure 3, neither at 90 nor at 120°C; in fact, the initial rate in

459
When using EtOH/EMIMBF4 at 90°C (Figure 4a), the situation is quite similar to that obtained 460 with BMIMBF4 (Figure 2a) or BMIMPF6 ( Figure 3a); nevertheless, the kinetic resolution is surprisingly 461 perfect when using EtOH/EMIMBF4 at 120° (Figure 4b), as in only 2.5 h a perfect 50% conversion in