Homogeneously Catalyzed Gas–Liquid Reactions - Gas–Liquid Reactions - Microreactors in Organic Chemistry and Catalysis, Second Edition (2013)

Microreactors in Organic Chemistry and Catalysis, Second Edition (2013)

9. Gas–Liquid Reactions

9.5. Homogeneously Catalyzed Gas–Liquid Reactions

9.5.1 Asymmetric Hydrogenation of Cinnamic Acid Derivatives

The potential of microprocess technology for catalyst and ligand screening for multiphase flows was evaluated at the asymmetric hydrogenation of Z-methylacetamidocinnamate (mac) with enantiomerically pure rhodium diphosphine complexes [59]. A microstructured mesh reactor was used which separates gas and liquids by a microstructured porous support (mesh) and has a stable, well-defined gas–liquid interface through the openings. A test series of 20 enantiomerically pure diphosphines as homogeneous catalysts was carried out. Less than 1 min residence time in continuous-flow operation was sufficient to complete the reaction for the very active catalysts. For less active catalysts, longer residence times, (up to 30 min), were needed which could not be realized by continuous operation in a practical manner. Rather, the flow had to be stopped to allow for batch-wise operation. The enantiomeric excesses of the measurements and of published data were in good accordance. The impact of hydrogen pressure on the enantiomeric excesses was determined as well.

(9.37) equation

In an earlier investigation, another reactor concept was applied for the transient serial screening of the hydrogenation of a cinnamic methyl ester [156, 162, 163]. The microreactor screening setup contained valves for liquid pulse injection (200 μl volume) into the feed stream of an interdigital micromixer. The degree of pulse broadening was characterized in a later study and conditions were optimized [85]. The other feed of the mixer was for the gas stream. The surfactant sodium dodecyl sulfate was added to stabilize the dispersed flow. In this way, rigid foam flows were generated which were stable for several minutes. With bubble sizes of about 150–250 μm, large specific gas–liquid interfaces of the foams of up to 50 000 m2/m3 are achieved and improve mass transfer so that operation in a kinetically controlled regime is possible [156, 162, 163]. Foam stability increases with hydrogen pressure. By this, foams remain stable up to 12 min at 70 °C and no coalescence occurs within this period. It has to be mentioned, however, that the range of solvents to be used for a microreactor test unit is limited as a stable foam has to be established. So far, this was only possible for aqueous solutions.

These flows were guided into a wound delay tube for reaction with ethylene glycol/water (60/40 wt%) and hydrogen [156, 163]. Efficient mass transfer in a kinetically controlled manner was achieved by the large specific gas–liquid interfaces of the foams of up to 50 000 m2/m3.

Before using the new method for screening, some basic engineering studies were made. The rate of reaction is proportional to increasing catalyst concentration [156, 163]. This result suggests operation in a chemical regime. At higher temperatures, larger reaction rates were found. The rate of reaction is proportional to decreasing surfactant (sodium dodecyl sulfate) concentration. No change in the enantiomeric excess was observed which is another indication for operation in a chemical regime, that is, governed by the kinetics and not by transport issues.

The performance of the new screening method was compared to an established method in the same laboratory, using a well-behaved mini batch reactor (see Table 9.6). The microreactor testing was able to run in average 15 tests per day; at maximum, 40 tests were achieved. The average and the maximum number of the mini batch used routinely were 2 and 3 per day, respectively [164]. A total of 214 tests were carried during the microreactor campaign [162]. The consumption of noble metal and chiral ligands per test for a microreactor test unit was 5–20 μg and 10 μmol, respectively. This is much lower as compared to the mini batch approach with a usage of 500–1000 μg of the noble metal and 0.1 μmol of the chiral ligand [165].

Table 9.6 Benchmarking of figures of merit for the gas–liquid screening using a well-behaved mini-batch and a continuous foaming-flow microreactor operation.

Source: By courtesy of the Swiss Chemical Society [164].

Benchmarked property of the g/l screening

Mini batch

Microflow

Reaction volume (ml)

10

0.1

Average amount of Rh/experiment (μg)

500–1000

5–20

Typical amount of ligand/experiment (μmol)

10

0.1

Temperature range (°C)

20–100

20–80

Pressure range (bar)

1–100

1–11

Residence time

>10 min

1–30

Average TTF achieved during study (d−1)

2

15

Maximum actual achievable TTF (d−1)

3

40

Range of solvents

Large

Restricted

Automation of reagents/catalyst injection

No

Yes

Automation of sample collection

No

Yes

The deviation in the enantiomeric excess (ee) was small and 90% of all data were within 40–48% ee [162]. A kinetic analysis was made by fitting the experimental data to empirical models by parity diagrams (Figure 9.38). A statistical model with first-order kinetics for hydrogen gave the best fit (141 from 170 experiments were properly described within the ranges of deviations set). The few rejected data had higher conversion than theoretically predicted. The kinetic constant and the activation energy were lower for a microreactor test unit than for a batch reactor, which was explained by the very low inventory of material [162, 164].

Figure 9.38 Screening of five substrates (see listing in the text) at 8 min and 35 °C. Conv: conversion and ee: enantiomeric excess. Source: By courtesy of Elsevier [53].

img

9.5.2 Asymmetric Hydrogenation of Methylacetamidocynamate

Enantiomerically pure rhodium phosphine complexes have been evaluated for the asymmetric hydrogenation of the prochiral substrate methylacetamidocynamate using the helical falling film microreactor [53]. The ligands in these complexes were mostly diphosphines (P–P), some nitrogen containing ligands (P–N) and some monophosphines. This library is available from commercial resources. A Diop ligand is the most active under the operating conditions of the test (Diop: 2,3-o-diisopropylidene-2,3-dihydroxy-1,4-bis(diphenylphosphino)-butane).

(9.38) equation

The Rh/Diop catalytic system is one of the fastest catalyzed gas–liquid asymmetric hydrogenations. A (R,S)Cy-Cy-Josiphos ligand behaves almost as good as the Diop ligand and provides a better enantioselectivity of 75% (Josiphos: family of ferrocenyl diphosphine ligands; cy: cyclo). The latter is the most active of the Josiphos family (88% conversion). The reproducibility of the data obtained has been checked with the Rh/Diop catalytic system. Over five tests, the mean deviation was 2% for conversions and <1% for the enantiomeric excess, which proved the reliability of this new microdevice.

The catalyst precursor complex [Rh(COD)Diop]BF4 has been used for the screening of five substrates containing pro-chiral C=C double bond (COD = 1,5-cyclooctadiene) [53]. These were methylacetamidoacrylate (S1), Z-α-methylacetamidocinamate (S2), dimethylitaconate (S3), methone (S4), and rac-α-pinene (S5) (Figure 9.39). Activated C=C bonds such as in the two acetamido derivatives were more reactive. The most reactive molecule is the less sterically hindered substrate methylacetamidoacrylate. Reaction was less pronounced for unsubstituted and sterically hindered substrates such as methone. The reduction of C=O bond in α-pinene is more difficult. These results are in agreement with the general trends reported for asymmetric hydrogenations.

Figure 9.39 Reaction performance comparison of three reactors with the most active catalysts Rh/Josiphos and Rh/Diop. Caroussel (car), helical falling film microreactor (μ), and Parr (batch) reactor. 9% conv. and 46%conv. denotes a fixed conversion of 9% and 46%, respectively, which has to be achieved. Source: By courtesy of Elsevier [53].

img

Only very low catalyst concentrations down to 5 × 10−5 kmol/m3 are consumed which keeps also the catalyst inventory very small [53]. Only 0.08 mg of Rh and about 0.2 mg–13 μg of the very expensive chiral ligands (about 300–1000 €/g), depending on their molecular weight, are consumed. Finally, a performance comparison for three different reactors was made for the substrate methylacetamidocynamate and the two rhodium diphosphine complexes Rh/Josiphos and Rh/Diop (Figure 9.40). The first reactor was a commercial “Caroussel” reactor (Radleys Discovery Technologies) with a circular arrangement of 12 screw-capped gas-tied glass tubes agitated by magnetic stirring. The second reactor is a commercial pressure batch reactor with a turbine and baffles (Series 4590 from Parr Instruments). The third reactor was the helical falling film microreactor.

Figure 9.40 Fluorinations of m-benzaldehydes using a nine-channel microreactor, where —X is —NO2, —CN, CF3, or —CHO functional group. Source: Reprinted with permission from Ref. [121]. Copyright (2009) American Chemical Society.

img

Enantioselectivity was roughly the same for the three reactors, being 80–90% and 62–65% for the Rh/Josiphos and Rh/Diop catalysts, respectively [53]. Conversion was very different. For fixed reaction time, the batch reactor and the falling film microreactor had higher conversions than the Caroussel reactor. This was indicative of operation under mass transfer regime in the latter. Based on these data, it was concluded that the mass transfer coefficients kla of the helical falling film microreactor is in between the boundaries given by the known kla values of 1–2 1/s for small batch reactors and about 0.01 1/s for the Caroussel reactor.