Ciara L.
Pollock
,
Kevin J.
Fox
,
Sophie D.
Lacroix
,
James
McDonagh
,
Patricia C.
Marr
*,
Alanna M.
Nethercott
,
Annie
Pennycook
,
Shimeng
Qian
,
Linda
Robinson
,
Graham C.
Saunders
* and
Andrew C.
Marr
*
School of Chemistry and Chemical Engineering, Queen's University Belfast, David Keir Building, Belfast, BT9 5AG, UK. E-mail: a.marr@qub.ac.uk
First published on 7th September 2012
Chemoenzymatic dynamic kinetic resolution (DKR) of rac-1-phenyl ethanol into R-1-phenylethanol acetate was investigated with emphasis on the minimization of side reactions. The organometallic hydrogen transfer (racemization) catalyst was varied, and this was observed to alter the rate and extent of oxidation of the alcohol to form ketone side products. The performance of highly active catalyst [(pentamethylcyclopentadienyl)IrCl2(1-benzyl,3-methyl-imidazol-2-ylidene)] was found to depend on the batch of lipase B used. The interaction between the bio- and chemo-catalysts was reduced by employing physical entrapment of the enzyme in silica using a sol–gel process. The nature of the gelation method was found to be important, with an alkaline method preferred, as an acidic method was found to initiate a further side reaction, the acid catalyzed dehydration of the secondary alcohol. The acidic gel was found to be a heterogeneous solid acid.
Scheme 1 Dynamic kinetic resolution of rac-1-phenyl ethanol into R-1-phenylethanol acetate. |
Fig. 1 Chemocatalysts for chemoenzymatic DKR. |
Scheme 2 Hydrogen transfer mechanism for the racemization of a secondary alcohol by catalyst MLn. |
The chemoenzymatic DKR of secondary alcohols to yield chiral esters is susceptible to oxidation of the alcohol by dehydrogenation or Oppenauer oxidation10–12 to form ketone side products. This side reaction is expected to vary as the organometallic catalyst varies since the relative rate of oxidation versus racemization should be chemocatalyst dependent.
A further side reaction that may occur in chemoenzymatic DKR is the reaction between the chemical catalyst and biocatalyst. Enzymes such as CalB have a wide variety of ionic domains and regions rich in groups that may act as donors for the dissolved metal ion. In instances in which the chemocatalyst becomes associated with the protein structure racemization will be prevented, or at least, severely reduced; and the interaction may also disrupt the enzyme's activity by altering the protein structure. The prevalence of this type of interaction explains why many active hydrogen transfer catalysts exist which are inactive in the presence of lipase.5
In order to validate the work-up and analysis methods kinetic resolutions were first carried out in the absence of an organometallic complex. As expected the procedure gave 50% conversion to R-1-phenylethanol acetate, the percentage enantiomeric excess (% ee) was 94% at 20 °C. DKR results were obtained for a variety of Ru(II) (Table 1), Rh(III) and Ir(III) (Table 2) complexes. The structure of 1, 2 and 3 is given (Fig. 1).
Catalyst | % Estera (% ee) | % Ketone |
---|---|---|
Toluene (2.4 mL), catalyst (0.0072 mmol), rac-1-phenyl ethanol (7.2 mmol), isopropenyl acetate (7.2 mmol), Novozym® 435 (40.5 mg), potassium carbonate (0.26 g), temperature 70 °C, reaction time 18 h. 1 is ½[(η6-mesitylene)Ru(μ-Cl)3RuCl(dfppm)].a Percentage conversion to 1-phenylethanol acetate by NMR.b 5 h reaction. Mesitylene is 1,3,5-trimethylbenzene, p-cymene is 1-methyl-4-(1-methylethyl)benzene, DPEN is 1,2-diphenyl-1,2-ethylenediamine (R,R), PHEN is 2-phenylglycinol (S), dppe is bis(diphenylphosphino)ethane, dppp is 1,3-bis(diphenylphosphino)propane, dfppm is (C6F5)2PCH2P(C6F5)2. | ||
½[(η6-Mesitylene)RuCl(μ-Cl)]2 + DPEN | 70 (97) | 3 |
½[(η6-p-Cymene)RuCl(μ-Cl)]2 + DPEN | 73 (97) | 2 |
½[(η6-Mesitylene)RuCl(μ-Cl)]2 + PHEN | 80 (97) | 1 |
½[(η6-p-Cymene)RuCl(μ-Cl)]2 + PHEN | 86 (94) | <1 |
[(η6-Mesitylene)RuCl(dppe)]BF4 | 94 (85) | <1 |
[(η6-Mesitylene)RuCl(dppp)]BF4 | 83 (98) | <1 |
[(η6-Mesitylene)RuCl(dfppm)]BF4 | 89 (92) | — |
[(η6-p-Cymene)RuCl(dfppm)]BF4 | 75 (91) | <1 |
1 | >99 (96) | <1 |
1 b | >99 (98) | <1 |
Catalyst | % Ester (% ee) | % Ketone |
---|---|---|
Toluene (2.4 mL), catalyst (0.0072 mmol), rac-1-phenyl ethanol (7.2 mmol), isopropenyl acetate (7.2 mmol), Novozym® 435 (40.5 mg), potassium carbonate (0.26 g), temperature 70 °C, reaction time 18 h. 2 is [{η5,κP-C5Me4CH2C6F4-2-P(C6F5)CH2P(C6F5)2)}RhCl2]. 3 is [Cp*IrCl2(1-benzyl,3-methyl-imidazol-2-ylidene)].8a 48 h reaction. | ||
½[Cp*IrCl(μ-Cl)]2 + DPEN | 82 (84) | 2 |
½[Cp*IrCl(μ-Cl)]2 + PHEN | 86 (91) | 3 |
[Cp*IrCl(dfppm)]BF4 | 84 (83) | 1 |
½[Cp*RhCl(μ-Cl)]2 + DPEN | 71 (96) | 12 |
½[Cp*RhCl(μ-Cl)]2 + PHEN | 88 (97) | 4 |
[Cp*RhCl(dppe)]BF4 | 70 (83) | 2 |
[Cp*RhCl(dfppm)]BF4 | 36 (86) | 5 |
[Cp*RhCl2(PPh3)] | >99 (90) | <1 |
[Cp*RhCl2(p-C6H4F)3] | 87 (98) | 2 |
2 | 63 (96) | — |
2 a | 97 (96) | <3 |
3 | >99 (97) | <1 |
Most of the organometallic complexes tested afforded reasonable conversion over 18 h. The enzyme was affected by the presence of the metal in many cases, affording lower enantioselectivity than obtained in the absence of the complex. Of the Ru(II) complexes tested bimetallic complex 113 gave optimum activity and fully conserved selectivity. The synthesis of similar complexes with non-fluorinated ligands has been reported.14 Comparing 1 and Shvo's catalyst (Fig. 1)7 both these active catalysts have hindered bimetallic cores and produce two different complexes on dissociation.
Loss of stereoselectivity was more common for Rh(III) and Ir(III) complexes (Table 2), and this points towards an increased interaction between the metal and lipase B. Complex 2 was selective, but exhibited lower activity, requiring longer reaction times; 97% conversion and good enantioselectivity could be obtained after 48 h. The slow rate of reaction could be attributed to steric crowding as the metal ion is surrounded by fluorinated groups. Such bulk is expected to retard interactions between the bio- and chemo-catalyst, but this must be balanced with the rate of racemization, as very bulky ligands will slow the approach of the organic substrate to the metal centre.
Levels of ketone formation were low for Ru(II) and Ir(III) catalysts and higher for Rh(III) catalysts. [Cp*RhCl(μ-Cl)]2 + DPEN was the most active system for the dehydrogenative oxidation of 1-phenyl ethanol under these conditions.
Overall the results suggest that the activity and selectivity can be tuned, we suggest that the key to a more active chemoenzymatic system is the positioning of enough steric bulk to prevent interaction between the enzyme and chemocatalyst, but not too much, as this will slow down the overall rate of DKR by retarding racemization.
The most active Ru(II) catalyst 1 was tested as a racemization catalyst on the aliphatic substrate rac-pinacolyl alcohol (3,3-dimethyl-2-butanol). Under similar conditions 81% ester at 82% ee was obtained with 3% ketone, Ir(III) catalyst 3 achieved 99% ester, 99% ee.8
The activity of 3 was compared to that of [(η5-Ph5C5)RuCl(CO)2], a highly active racemization catalyst (Fig. 1),15 both catalysts facilitated good conversion (>99%) and good enantioselectivity (>97% ee) in the presence of base, but only 3 was found to be active in the absence of base. Liquid sampling of a scaled up reaction employing 3 (containing base) revealed the reaction was complete after 2 h 30 min. Isolation of the ester by distillation from a repeated reaction gave a 70% isolated yield of R-1-phenylethanol acetate (97% ee).
A materials approach to the prevention of mutual poisoning of bio- and chemo-catalysts has been reported by Blum and Avnir.16 The catalysts were compartmentalized by entrapping them inside porous silica by sol–gel processes. This method physically imprisoned the catalyst preventing free diffusion through the solvent and therefore making processes that result from the interaction between the bio- and chemo-catalyst unlikely.17 Entrapping both catalysts in chemoenzymatic DKR is likely to lead to a highly diffusion limited process, therefore we investigated the entrapment of just the biocatalyst (CalB) in silica gels.
The activity of a batch (different from that above) of lipase acrylic resin from Candida Antarctica (Novozym® 435) and 3 was evaluated. Kinetic resolution gave the expected result of 50% 1-rac-phenyl ethanol conversion to the R-1-phenylethyl acetate. The reaction was repeated as a DKR in the presence of 3, after 48 h only 47% conversion to 1-rac-phenyl acetate was observed. This drop in performance is indicative of an interaction between the chemocatalyst and enzyme. In order to compartmentalize the catalyst and reduce poisoning the Novozym® 435 was entrapped in silica by sol–gel methods. This provides a porous silica coating around the lipase acrylic resin. Entrapment methods have been successfully applied by Hoyos et al.18 as an enzyme stabilising method in the dynamic DKR of benzoins by Pseudomonas stutzeri employing Shvo's catalyst. In order to entrap the enzyme a basic sol–gel method was applied catalyzed by benzylamine; this method was based on an organogel-templated method previously described.19 Upon moderate heating and cooling organogelator 2,5-di-O-methanesulfonyl-1,4:3,6-dianhydro-D-sorbitol gels a range of solvents, including alcohols. In the presence of a silica precursor and promoter the organogel formed acts as a template transcribing silica formation and dictating structure on the micro- and nano-scale. This method was capable of providing a stable matrix that entrapped the enzyme. The distribution of the enzyme throughout the matrix is important to the activity of the resultant material. A base catalysed method of gelation and transcription was used as base promotes the hydrogen transfer reaction. AOT (dioctyl sodium sulfosuccinate) and PEG (polyethylene glycol) 400 were used to stabilize the enzyme and create the correct environment within the matrix.20 The charged head group on AOT has the potential to interact with the enzyme, and the PEG forms stable microemulsions within a gel matrix.21
An acidic sol–gel method was also used to entrap lipase B employing PEG 400 to stabilise the enzyme and phosphoric acid to catalyse the gelation.
In order to investigate the effects of the silica support, the entrapped enzyme catalysts were tested in kinetic resolution. The activity of the doubly immobilized basic gel was approximately 1/10th that of the acrylic resin Novozym® 435 (Table 3). The activity of the acidic gel could not be measured as the resulting 1H NMR spectrum contained many additional resonances, indicating that new side reactions had occurred; these were further investigated and the results reported below.
Catalyst | Active/g | % Ester |
---|---|---|
Toluene (2.4 mL), rac-1-phenyl ethanol (7.2 mmol), isopropenyl acetate (7.2 mmol), temperature 70 °C, reaction time 4 h, NA = not applicable due to side reactions. | ||
Novozym® 435 | 0.02 | 36 |
Novozym® 435 | 0.04 | 50 |
Basic gel | 0.33 | 39 |
Acidic gel | 0.35 | NA |
In order to assess the extent to which the lipase B in the basic gel and in Novozym® 435 was heterogeneous and recyclable, the following procedure was carried out. The enzyme-containing material was exposed to reaction conditions for 2 h, the solution was then filtered off and returned to reaction conditions for a further 2 h, the lipase-containing material was added to a fresh reactant solution and run for 4 h (Table 4).
Catalyst | Initial conversion/% | Filtrate conversion/% | Recycled conversion/% |
---|---|---|---|
Toluene (2.4 mL), rac-1-phenyl ethanol (7.2 mmol), isopropenyl acetate (7.2 mmol), temperature 70 °C. | |||
Novozym® 435 | 32 | 36 | 9 |
Basic gel | 10 | 15 | 27 |
Under these conditions this batch of Novozym® 435 did not recycle well. A significant further conversion (from 32 to 36%) was observed when the filtrate was exposed to reaction conditions for a further 2 h, indicating leaching of active material. Furthermore the used enzyme aggregate gave only 9% conversion after 4 h, indicating little active material was left. The basic gel performed better but not ideally. Initial conversion was reduced relative to Novozym® 435, this was expected as the reactant must now diffuse through the silica gel to reach the entrapped lipase. Upon separation the filtrate was still active, and this reveals that active enzyme was leaching, the activity was 50% of that observed in the presence of the gel. The used and recycled entrapped enzyme gave 27% conversion after 4 h, consistent with conservation of enzyme activity.
Comparing the DKR activity of the entrapped enzyme with 3 and the activity of the parent Novozym® 435 with 3, the entrapped enzyme enabled a 5% improvement in conversion, affording 52% ester (Table 5).
Catalyst | Active/g | Time/h | % Conversion | % Ester |
---|---|---|---|---|
Toluene (2.4 mL), rac-1-phenyl ethanol (7.2 mmol), isopropenyl acetate (7.2 mmol), Novozym® 435 (41 mg), potassium carbonate (0.26 g), temperature 70 °C. | ||||
Novozym® 435 | 0.041 | 20 | 50 | 50 |
Novozym® 435 + 3 | 0.020 + 0.0022 | 48 | 47 | 47 |
Basic gel + 3 | 0.15 + 0.0022 | 48 | 52 | 52 |
We propose that the DKR reaction of the basic gel and 3 failed to achieve full conversion for two reasons. Firstly there is a reduction in the activity of the enzyme due to the additional diffusion processes necessary to approach the active site. Secondly there is measureable leaching from the gel, the active enzyme in solution is believed to interact with the chemocatalyst, retarding activity and preventing full conversion of the substrate. As the entrapment of lipase B was only partially successful and the basic gel exhibited some leaching, higher but not complete conversion was observed in the resultant DKR. We suggest that a better entrapment method would lead to a more robust and recyclable DKR.
In order to investigate the side-reactions occurring when the acidic gel was employed, a gel was prepared consisting of PEG 400, phosphoric acid and silica, with no entrapped lipase. The gel was heated to 105 °C (unstirred) in a toluene solution containing rac-1-phenylethanol. The products detected correlated with the side products during kinetic resolution and were identified as the products of dehydration [1-(1-phenylethoxylethyl)benzene] and styrene (Scheme 3). In order to test whether the catalyst behaved as a homogeneous or heterogeneous Brønsted acid catalyst, the following procedure was carried out. The dehydration was performed for 2 h and the solution and gel separated by filtration. The filtrate was returned to the reaction conditions for a further 2 h. For the final period of 2 h the gel was returned to the solution. The reaction was followed by 1H NMR and showed straightforward heterogeneous behaviour (Table 6, Chart 1).
Scheme 3 The dehydration of rac-1-phenyl ethanol. |
Chart 1 Heterogeneous solid acid behaviour in the dehydration of rac-1-phenyl ethanol. |
The most detrimental and unpredictable side reaction in chemoenzymatic DKR is the interaction between the chemo-catalyst and bio-catalyst, this reduces the activity of the enzyme, can hamper enantioselectivity, and frequently stops the racemization catalyst working altogether. The application of bulky groups on the metal catalyst can help to avoid this interaction, but this can be at the detriment of the overall rate, as steric bulk will slow down racemization. In instances where considerable poisoning occurs it is difficult to prevent, and extremely active hydrogen transfer catalysts can be rendered inactive. We suggest that the further protection of the enzyme in a porous material, or alternatively, the adoption of a heterogeneous hydrogen transfer catalyst, can reduce poisoning. A sol–gel entrapped dehydrogenation catalyst was recently reported by Oded et al.22
Solvents were dried and degassed using standard methods.
Complexes [(η6-mesitylene)RuCl(μ-Cl)]2, [(η6-p-cymene)RuCl(μ-Cl)]223 [Cp*IrCl(μ-Cl)]2, [Cp*RhCl(μ-Cl)]224 and [(η5-Ph5C5)RuCl(CO)2]15 were prepared by literature methods. The synthesis of [(η6-mesitylene)RuCl(dfppm)]BF4, [(η6-p-cymene)RuCl(dfppm)]BF4, 1 [(η6-mesitylene)Ru(μ-Cl)3RuCl(dfppm)], [Cp*IrCl(dfppm)]BF4, [Cp*RhCl(dfppm)]BF4 and 2 [{η5,κP-C5Me4CH2C6F4-2-P(C6F5)CH2P(C6F5)2)}RhCl2] were conducted as previously reported.13 [(η6-Mesitylene)RuCl(dppe)]BF4, [(η6-mesitylene)RuCl(dppp)]BF4 were prepared by methods analogous to [(η6-mesitylene)RuCl(dfppm)]BF4 by substituting dfppm with the equivalent amount of dppe and dppp. [Cp*RhCl(dppe)]BF4 was prepared in a method analogous to [Cp*RhCl{(C6F5)2PCH2CH2(C6F5)2}]BF4 by substituting dfppm with the equivalent amount of dppe.25
[Cp*RhCl2(PPh3)] was prepared by a method analogous to [Cp*RhCl2P(C6H4F-4)].
[Cp*RhCl(μ-Cl)]2 (0.071 g, 0.11 mmol) and P(C6H4F-4)3 (0.073 g, 0.23 mmol) were dissolved in dichloromethane (20 mL). The resulting red solution was kept at ambient temperature overnight. Concentration by rotary evaporation afforded the product as a red solid, which was filtered and dried in vacuo. Yield 0.116 g (81%).
1H NMR (CDCl3): δ 7.80 (6H, m, Ho or Hm), 7.08 (6H, m, Ho or Hm), 1.38 [15H, d, 4J(P–H) 3.5, CH3]. 19F NMR (CDCl3): δ −108.43 (2F, br), −110.45 (1F, br). 31P NMR (CDCl3): δ 28.9 [d, 1J(Rh–P) 146]. EI-MS: m/z 589 [M − Cl]+. (Found: C, 53.4; H, 4.2. Calc. for C28H27Cl2F3PRh: C, 53.8; H, 4.35%).
3 [Cp*IrCl2(1-benzyl,3-methyl-imidazol-2-ylidene)] was prepared as previously reported.26,27 2,5-Di-O-methanesulfonyl-1,4:3,6-dianhydro-D-sorbitol was prepared as previously reported.19
The following reagents were used as supplied; Novozym® 435, isopropenyl acetate, DPEN, PHEN, benzylamine, Tetraethylorthosilicate (TEOS), PEG 400 and AOT (Sigma-Aldrich), Rac-1-phenyl ethanol (Fluka and Sigma-Aldrich), potassium carbonate (Lancaster), phosphoric acid (85% aqueous solution Lancaster).
Footnote |
† This paper is dedicated to Prof. David J. Cole-Hamilton on occasion of his retirement in celebration of his distinguished (and ongoing) contribution to transition metal catalysis, and his role in introducing ACM to organometallic chemistry for homogeneous catalysis, his enthusiasm was truly infectious. |
This journal is © The Royal Society of Chemistry 2012 |