This is an author-produced version of the following Elsevier-published article: Belokon, Clegg , Harrington, Young and North, Asymmetric synthesis using heterobimetallic catalysts obtained from titanium and vanadium complexes of chiral and achiral salen ligands, Tetrahedron, 63 (24), 5287-5299 doi:10.1016/j.tet.2007.03.140

Graphical Abstract

To create your abstract, type over the instructions in the template box below. Fonts or abstract dimensions should not be changed or altered.

Asymmetric cyanohydrin synthesis using Leave this area blank for abstract info. heterobimetallic catalysts obtained from titaniumO andMe 3SiCN OX OX vanadium complexes of chiral and achiral salen + or 1 (0.1-1 mol%) + ligands R H KCN / Ac2O R CN R CN Yuri N. Belokon’, b William Clegg, a Ross W. X = SiMe3: ratio = 84 : 16 Harrington, a Carl Young a and Michael North a* X = Ac; ratio = 16 : 84 a) School of Natural Sciences, Bedson Building, + - 1 = [(R,R-salen)Ti(µ-O)2V (S,S-salen') EtOSO3 ] Newcastle University, Newcastle upon Tyne, NE1 7RU, UK b) A.N. Nesmeyanov Institute of Organo-Element Compounds, Russian Academy of Sciences, 117813, Moscow, Vavilov 28, Russian Federation.

Tetrahedron 1

TETRAHEDRON

Pergamon

Asymmetric cyanohydrin synthesis using heterobimetallic catalysts obtained from titanium and vanadium complexes of chiral and achiral salen ligands

Yuri N. Belokon’, b William Clegg, a Ross W. Harrington, a Carl Young a and Michael North a*

a School of Natural Sciences, Bedson Building, Newcastle University, Newcastle upon Tyne, NE1 7RU, UK. b A.N. Nesmeyanov Institute of Organo-Element Compounds, Russian Academy of Sciences, 117813, Moscow, Vavilov 28, Russian Federation

Abstract— Titanium(IV)(salen) and vanadium(V)(salen) complexes are both known to form catalysts for asymmetric cyanohydrin synthesis. When a mixture of titanium and vanadium complexes derived from the same or different salen ligands is used for the asymmetric addition of trimethylsilyl to benzaldehyde, the absolute configuration of the product and level of asymmetric induction can only be explained by in situ formation of a catalytically active heterobimetallic complex, and is not consistent with two monometallic species acting cooperatively. Combined use of complexes containing chiral and achiral salen ligands demonstrates that during the asymmetry inducing step of the mechanism, the is coordinated to the vanadium rather than the titanium ion. The titanium complexes also catalyse the asymmetric addition of ethyl cyanoformate to , a reaction for which vanadium(V)(salen) complexes are not active. For this reaction, use of a mixture of titanium and vanadium(salen) complexes resulted in a complete loss of catalytic activity, a result which again can only be explained by in situ formation of a heterometallic complex. Both the titanium and vanadium based catalysts also induce the asymmetric addition of / acetic anhydride to aldehydes. For this reaction, combined use of chiral and achiral complexes indicated that during the asymmetry inducing step of the mechanism, the aldehyde was coordinated to titanium rather than vanadium, a result which contrasts with the observed results when trimethylsilyl cyanide was used as the cyanide source. © 2007 Elsevier Science. All rights reserved

1. Introduction and carbonates, 9 and insoluble, polymeric versions of the catalyst have also been developed and used as catalysts for are key synthetic intermediates for the asymmetric cyanohydrin synthesis. 10,11 Closely related synthesis of pharmaceuticals and natural products. As a titanium(salen) complexes have also been used as catalysts result, the development of asymmetric catalysts for the for the enantioselective ring-opening of meso -epoxides, 12 addition of various cyanide sources to carbonyl compounds hetero-Diels-Alder reactions, 13 the asymmetric addition of has increased significantly over the last ten years. 1,2 dialkylzinc reagents to α-ketoesters, 14 the asymmetric Metal(salen) complexes are amongst the most versatile addition of organogallium reagents to aldehydes, 15 and the catalysts for a wide range of reactions, 3 and we have asymmetric oxidation of sulfides. 16 previously reported 4 the development of bimetallic titanium complex 1 as a highly effective catalyst for the asymmetric Subsequently, we have shown that vanadium(V)(salen) addition of trimethylsilyl cyanide to aldehydes 5 and complex 2 is an even more enantioselective catalyst for the ketones 6 at room temperature, and for the asymmetric asymmetric addition of trimethylsilyl cyanide to addition of ethyl cyanoformate 7 and potassium cyanide 8 to aldehydes, 17 and that complex 2 also catalyses the aldehydes at -40 oC. Complex 1 has also been used by other asymmetric addition of potassium cyanide to aldehydes. 8 workers for the asymmetric synthesis of cyanohydrin esters Polymeric, 11 and supported 18 versions of this catalyst have ——— Key words: Asymmetric catalysis; Cyanohydrin; Titanium; Vanadium; Heterometallic complexes * Corresponding author. Tel.: +44-191-222-7128; fax: +44-870-131-3783; e-mail: [email protected]. 2 Tetrahedron been prepared and vanadium(V)(salen) complexes have shown in Figure 1A. However, an alternative mechanism also been used as catalysts for the asymmetric oxidation of involving sequestration of titanium ions by vanadium (for sulfides. 19 A related vanadium(V)(salen) complex was example by formation of a catalytically inactive oligomeric shown to catalyse the asymmetric transfer of hydrogen complex 23 ) leaving only vanadium(salen) complexes to cyanide from acetone cyanohydrin to aldehydes. 20 For both catalyse the reaction through the bimolecular transition complexes 1 and 2, the catalyst derived from ( R,R )- state shown in Figure 1B could not be completely ruled out. cyclohexane diamine always catalyses the formation of the (S)-enantiomer of the cyanohydrin. (salen)M CN OSiMe3 (salen)M CN tBu tBu O (salen)M O H R (salen)M O H R O tBu tBu CN A B N O O O N Figure 1 Ti Ti O N O O N In this manuscript, we report for the first time results on tBu tBu heterobimetallic titanium and vanadium complexes derived from two different salen ligands, one of which is achiral and the other chiral. The catalysis of the addition of various tBu tBu cyanide sources to benzaldehyde using these 1 heterobimetallic systems is investigated, and the results can only be explained by in situ formation of a catalytically active heterobimetallic complex as illustrated in Figure 1A.

O EtOSO3 The structure and stereochemistry of the heterobimetallic N N complexes are analysed in detail and can be used to predict V not only the sense and magnitude of asymmetric induction, tBu OO tBu but also the rate of catalysis observed using the complexes. O H H tBu tBu 2. Results and Discussion 2

Mechanistic studies on catalysts 1 and 2 demonstrated that 2.1 Synthesis of achiral salen complexes of titanium and two metal(salen) units were involved in the catalysis and vanadium were consistent with either of the transition state structures shown in Figure 1, where one metal ion acts as a chiral To allow further investigation of the mechanism of Lewis acid, and the other as a chiral cyanide source, with asymmetric cyanohydrin synthesis induced by metal(salen) cyanide then being transferred between the two metals onto complexes, it was decided to combine the use of both 21 enantiomers of catalysts 1 and 2 with the corresponding the coordinated carbonyl compound. Whilst catalyst 2 is 24 more enantioselective than catalyst 1, it accelerates the rate complexes derived from achiral salen ligand 3. Ligand 3 of addition of trimethylsilyl cyanide to aldehydes to a much was converted into the corresponding bimetallic- lower extent. Reactions involving catalyst 1 are typically titanium(salen) complex 4 using the same synthetic route complete in 30 minutes, whereas those involving catalyst 2 developed for the synthesis of catalyst 1 (Scheme 1). require at least 16 hours to reach completion under However, treatment of ligand 3 with vanadyl sulfate in the identical reaction conditions. presence of air, using the methodology developed for the synthesis of catalyst 2,8 did not result in spontaneous We recently reported 22 however, that when the enantiomer oxidation of the vanadium ion and formation of the desired vanadium(V)(salen) complex 5; rather the corresponding of catalyst 1 (( S,S )-1) was mixed with catalyst 2, so as to 25 prepare a solution containing equimolar amounts of vanadium(IV)(salen) complex was isolated and could then be oxidized to the desired vanadium(V) complex 5 by titanium and vanadium ions, the resulting mixture catalysed 19 the formation of the ( S)-enantiomer of mandelonitrile treatment with cumene hydroperoxide. trimethylsilyl ether, a result which was inconsistent with the two catalysts acting independently or with scrambling 2.2 Structural analysis of bimetallic salen complexes of the salen ligand between the two metals. These results could only be explained by catalysis involving at least two In monometallic octahedral complexes, the planar, trans - configuration of a salen ligand is usually energetically metal ions, and indicated that catalysis by two different 26 metal ions was more favourable than catalysis by either preferred (Figure 2). However, in bis -bridged complexes homometallic system. Kinetics experiments indicated that as occurs in complexes 1 and 4, this is not possible as the two bridging oxygens must adopt positions cis to one catalysis by the heterometallic system was slower than 27 22 another. The salen ligand can then adopt either cis -α or catalysis using catalyst 1 alone, which was consistent β 28 with the in situ formation of a heterobimetallic complex as cis - configurations, with the latter being preferred in all Tetrahedron 3 cases known to date. 29,30 The cis -β configuration of a salen A X O O ligand is chiral, and two enantiomeric structures (∆ or Λ) are possible. For bimetallic complexes bearing two N X X O M N N M X N M X different salen ligands, there are thus four possible O N O stereoisomeric configurations ( ∆∆, ΛΛ, ∆Λ and Λ∆ ) of the two salen ligands. However, each of these complexes also X O N exists as syn - and anti -isomers as shown in Figure 2, thus X O O giving a total of eight possible stereoisomers. 29 O N N X N X M M M tBu tBu O N N X O X X O N trans cis-α cis-β tBu tBu B N O O O N O O O O Ti Ti O O N N O O O NO O N M M M M N O O O O N tButBu N N N N syn-∆∆ syn-ΛΛ

O O O O tBu tBu O N 4 O O N O M M M M N O N O O O 1. TiCl4 2. phosphate buffer N N N N syn-∆Λ syn-Λ∆ O N O N NN O O O N O N M M M M tBu OH HO tBu N O N O O O N O N O tBu tBu anti-∆∆ anti-ΛΛ 3 O N O N 1. VO(SO ) / O , EtOH, ∆ O O N N O O 4 2 M M M M 2. PhC(Me) OOH, HCl 2 N O O O O N N O N O N O N Cl anti-∆Λ anti-Λ∆ V tBu OO tBu Figure 2: (A) The three configurations of a salen ligand in a O monometallic complex. (B) The eight stereoisomeric complexes H H of a bimetallic complex [(salen)M(µ-O)2M(salen')] in which the tBu tBu 5 salen ligand adopts a cis-β configuration. Scheme 1 In contrast, the crystal structure of complex 9 which differs In chiral complexes 6-8 (complexes 6 and 7 are active from catalyst 4 only by the absence of tert -butyl groups on catalysts for the addition of trimethylsilyl cyanide to the aromatic rings has been determined by Tsuchimoto, 33 aldehydes 5 and differ from catalyst 1 only by the absence of and stereochemically analysed by Katsuki. 34 In this tert -butyl groups on the aromatic rings), X-ray complex derived from a salen ligand which is devoid of crystallography showed that the two metal(salen) units stereocentres, the two metal frameworks were found to be were homochiral, with both possessing ( R)-stereocentres heterochiral with one possessing the ∆-configuration and and ∆-configuration. 5,31,32 The two salen ligands were also the other the Λ-configuration, and the two ligands were oriented syn to one another, thus giving selectively the syn - oriented anti to one another ( anti -∆Λ ). In this case, since ∆(RR )∆(RR ) stereoisomer. The ∆-configuration of the the 1,2-diaminoethane unit is unsubstituted, the δ- and λ- metal(salen) units in complexes 6 and 7 is a direct conformations of an individual five membered ring chelate consequence of the presence of ( R,R)-1,2-disubstituted 1,2- will be of equal energy, and there will be no intrinsic diaminoethanes within the salen ligands. The absolute preference for the formation of a ∆- or Λ-configuration. configuration of the two stereocentres induces a λ- Therefore, the selective formation of the ∆Λ -configuration conformation on the five membered chelate ring, and this in within the bimetallic complex suggests that this is due to turn induces the overall ∆-configuration on the metal- the heterochiral ( ∆Λ ) configuration being intrinsically more ligand unit. stable than the homochiral ( ∆∆ or ΛΛ ) configuration. As complex 9 was the only achiral bimetallic titanium(salen) 4 Tetrahedron complex whose structure had been determined, we obtained O bond as this gives the least distorted structure for the the crystal structure of complex 4 (Figure 3). This complex, central Ti 2O2 unit. If the two salen ligands have the same which is exactly centrosymmetric with a planar central configuration ( ∆∆ or ΛΛ ) as in complexes 6-8, then only a Ti 2O2 ring, was also found to exist in a heterochiral, anti - syn -dimer allows both bridging Ti-O-Ti units to have one ∆Λ configuration in which each salen ligand adopted a cis - long and one short bond. In contrast, if the two salen β configuration, exactly analogous to the known structure ligands have opposite configurations ( ∆Λ ), then only the of complex 9. The preferential formation of heterochiral anti -dimer allows both bridging Ti-O-Ti units to have one (∆Λ ) complexes also appears to be generally true for other long and one short bond. In the case of complex 4, these bimetallic metal(salen) complexes derived from achiral 35 were found to be 1.8253(16)Å and 1.8978(16)Å ligands, though there are two related examples of bimetallic salen complexes (derived from aluminium 36 and respectively, lengths which fall within the ranges manganese 37 ) with bridging OMe groups which crystallize previously found for complexes 6-9. A consequence of this as a racemic mixture of syn -∆∆ and syn -ΛΛ configurations analysis is that the four structures shown in the box in instead. Figure 2 will be thermodynamically less stable than the other four structures, and only four ( syn -∆∆ , syn -ΛΛ , anti - ∆Λ and anti -Λ∆ ) of the eight possible configurations of a

(salen)Ti( µ-O) 2Ti(salen’) complex are likely to be formed. R R There is only one crystal structure of a bimetallic salen 38 N O O O N complex which exists in the anti -∆∆ configuration, and in Ti this case all of the bridging M-O bonds have the same O Ti NO O N length (1.823(2)Å). The formation of an anti -∆∆ R R configuration in this case may be due to the lack of a dipole moment if the two salen ligands are arranged anti- to one another. If the two salen ligands are identical (as in the 6 (R = H) cases of complexes 4, 6-9), then the two syn -dimers form a 7 (R = tBu) pair of enantiomers whilst the two anti -dimers correspond to a single achiral meso-compound. Based on the above analysis, the structures of (salen)M( µ-O) M’(salen’) Ph 2 2 Ph2 complexes can be predicted on the basis of three rules: P P O Ph N O O N Ph 1. An ( R,R )-1,2-disubstitued-1,2-ethanediamine within the Ti I2Pd PdI2 Ti salen ligand forces the ligand to adopt the ∆- Ph NO O O N Ph configuration whilst an ( S,S )-1,2-disubstitued-1,2- P P diamine within the salen ligand forces the ligand to Ph2 Ph2 adopt the Λ-configuration. ∆Λ 8 2. The heterochiral -dimer is thermodynamically more stable than the corresponding homochiral ( ∆∆ or ΛΛ ) dimers.

3. Heterochiral dimers will have an anti -arrangement of the O N N O O two ligands and homochiral dimers will have a syn - Ti Ti O arrangement of the two ligands, unless all of the NO O N bridging M-O and M’-O bonds have the same length in which case an anti -arrangement is preferred.

Thus, by analogy with the known structures of complexes 9 6-8, and by application of the above rules, complex 1 will be forced to exist as the syn -∆(RR )∆(RR ) stereoisomer. The formation of a syn -complex in the case of compounds 6-8 but an anti -complex in the case of compounds 4 and 9 The solution-state structure of complex 4 was also can be explained by considering the bridging Ti-O bond investigated using variable concentration and variable lengths. 29 For complexes 6-9, the four central Ti-O bonds temperature NMR studies in different solvents. We have are not of equal length; in each case, the two bridging Ti-O previously reported that complex 1 dissociates in both bonds which are trans to a nitrogen atom of the salen chloroform and and exists as an equilibrium mixture of bimetallic and monometallic ligand have a length of 1.80-1.84Å, whilst the two Ti-O [(salen)Ti=O] species. In contrast, in benzene no bonds which are trans to an oxygen atom of the salen dissociation could be detected and only resonances due to ligand are longer (1.86-1.92Å). Thus, in these complexes, 39 the dimeric complex were present. However, for complex each Ti-O-Ti bridge consists of one long and one short Ti- 4, similar studies showed only one species present in either Tetrahedron 5 dichloromethane or benzene at concentrations between : 1 ((salen)Ti=O), whilst at a concentration of 0.0008M the 0.1M and 0.004M. At 25 oC, this species gave rise to a very corresponding ratio is 0.9 : 0.7 : 1, and at a concentration of simple 1H NMR spectrum which included just one imine 0.000016M, only peaks corresponding to the monometallic CH peak and two aromatic CH’s. This implied that all four complex are present. The detection of the monometallic aromatic rings and imines within structure 4 were and syn -∆∆ / syn -ΛΛ dimers in chloroform but not in equivalent, but this is not consistent with any single dichloromethane is consistent with a relatively polar structure shown in Figure 2B. The spectrum (in deuterated solvent being required to stabilize the dipole moment dichloromethane) was however found to be highly present in these two complexes. temperature dependent (Figure 4); as on cooling, the aromatic and imine signals first broadened at temperatures 2.3 Catalysis of the asymmetric addition of down to -45 oC, and then reappeared as two imine and four trimethylsilyl cyanide to benzaldehyde aromatic signals at temperatures down to -70 oC. This behaviour was attributed to the existence of an equilibrating The use of complexes 1, 2, 4, 5 as catalysts for the addition mixture of the anti -∆Λ / anti -Λ∆ dimers. For complex 4, of trimethylsilyl cyanide to benzaldehyde was studied these two structures are equivalent and would give rise to under the standard conditions (0.1 mol% total catalyst, two imine CH’s and four aromatic CH’s, as observed at the CH 2Cl 2, room temperature, 24 hours) that have been lowest temperatures. However, equilibration between the developed for asymmetric cyanohydrin trimethylsilyl ether two dimeric structures has the effect of interconverting the synthesis using catalysts 1 and 2.5 To allow the two imines and the aromatic rings and if this occurs rapidly enantiomeric excess of the product to be determined, it was on the NMR timescale, then an averaged spectrum converted into O-acetyl mandelonitrile using acetic consisting of one imine and two aromatic CH’s would be anhydride and scandium triflate, under conditions observed, as seen at temperatures above -55 oC. The developed by Kagan 40 and proven not to cause any equilibration of each salen ligand between ∆ and Λ- racemisation (Scheme 2). The O-acetyl mandelonitrile configurations presumably occurs via an undetectably small could then be easily analysed by chiral gas amount of the monomeric [(salen)Ti=O] complex. chromatography. The results of reactions using complexes 1, 2, 4, 5 either individually or in combination are summarized in Table 1.

catalyst 1,2,4,5 (0.1 mol%) OSiMe3 PhCHO + Me3SiCN room temperature, CH2Cl2 Ph CN

Ac2O / Sc(OTf)3 (1 mol%) / MeCN OAc

Scheme 2 Ph CN

Entries 1-4 in Table 1 provide control experiments which demonstrate the activity of catalysts 1, 2, 4, 5 when used individually. Although not apparent from the data in Table 1, we have previously shown 21,22 that reactions catalysed by complex 1 are faster than those catalysed by complex 2 by about a factor of 10 and this is important for the analysis of the data obtained for mixed complexes. Whilst the achiral titanium complex 4 did catalyse the addition of trimethylsilyl cyanide, the reaction was very slow compared to the corresponding chiral catalyst 141 (Table 1: entries 1 and 3). To quantify this reactivity difference, the Figure 3. Molecular structure of complex 4 with 40% probability ellipsoids and selected atom labels. Hydrogen atoms have been omitted. kinetics of the addition of trimethylsilyl cyanide to o benzaldehyde at 0 C using catalysts 1 and 4 were studied. In chloroform, the spectrum of complex 4 was however Reactions catalysed by complexes 1 and 4 both obeyed first very concentration dependent and showed evidence of the order kinetics under these conditions (Figure 6) and the presence of three different species (Figure 5). Two of these first order rate constants were 2.4x10 -3 s -1 for complex 1 species can be attributed to the anti -∆Λ dimer (major and 1.7x10 -6 s -1 for complex 4. Thus, the reaction catalysed complex ‘A’ present at high concentration) and the by complex 4 was a thousand times slower than that monometallic [(salen)Ti=O] complex (minor complex ‘C’ catalysed by complex 1. As discussed above, complex 1 present at high concentration). The third species present in will be forced to exist as the syn -∆(RR )∆(RR ) stereoisomer, solution ‘B’ is assumed to be the syn -∆∆ / syn -ΛΛ dimer. whilst complex 4 is able to form the thermodynamically At a concentration of 0.1M, these three species are present more stable anti -∆Λ configuration and in dichloromethane, ∆Λ ∆∆ ΛΛ in a ratio of 22.5 ( anti - dimer): 2.6 ( syn - / dimer) no evidence for significant populations of other 6 Tetrahedron

-4 -1 configurations was detected by NMR spectroscopy. measured in CDCl 3 (Figure 6) were 1.2 x 10 s for Provided the process by which precatalysts 1 and 4 enter complex 1 (a factor of twenty lower than the corresponding the catalytic cycle is reversible, it is then reasonable that rate constant in dichloromethane) and 1.7 x 10 -5 for the more stable catalyst precursor 4 would be less active complex 4 (ten times greater than corresponding rate than complex 1. constant in dichloromethane). This suggests that for complex 1, less of the catalytically active complex is In chloroform solution however, NMR spectroscopy present in chloroform than in dichloromethane, whilst for suggests that complex 4 does exist in the syn -∆∆ / syn -ΛΛ complex 4 the reverse is true. This is consistent with the configuration as well as the anti -∆Λ configuration. syn -∆∆ / syn -ΛΛ configuration of the complexes being the Therefore, the catalytic activity of complexes 1 and 4 was catalytically active complex. For complex 1, the syn -∆∆ also compared for cyanohydrin synthesis in chloroform configuration is the only species detected by NMR solution (Table 1: entries 5 and 6). The chiral catalyst 1 was spectroscopy in dichloromethane, whilst in chloroform less reactive in chloroform (compare Table 1: entries 1 and monomeric species are also present. In contrast, for 5), whilst the achiral catalyst 4 was more reactive in complex 4, the syn -∆∆ / syn -ΛΛ configuration is not chloroform than in dichloromethane (compare Table 1: detectable by NMR spectroscopy in dichloromethane, but is entries 3 and 6). Similarly, the first order rate constants present in chloroform.

1 Figure 4: Variable temperature H NMR spectra of compound 4 in CD 2Cl 2. Expansion of the aromatic and imine region

Tetrahedron 7

Table 1. Use of catalysts 1, 2, 4, 5 in the synthesis of mandelonitrile Entry 7 of Table 1 confirms that mixing equimolar amounts trimethylsilyl ether. a of catalysts 1 and 2 with opposite absolute configurations results in formation of a species in which the ligand Conversion Enantiomeric excess (%) entry Catalyst b attached to the vanadium ion determines the absolute (%) (absolute configuration) configuration of the O-trimethylsilyl mandelonitrile 1 (R,R )-1 100 82 ( S) formed, whilst mixing catalysts 1 and 2 with the same 2 (S,S )-2 100 c 91 ( R) configuration (Table 1: entry 8) also gives an active catalyst. 22 Table 1: entries 9 and 10 then illustrate the effect 3 4 34 (80 c) 0 of mixing an achiral-salen complex of one metal with the 4 5 86 0 chiral-salen complex of a different metal. In both cases, the 5 (R,R )-1d 45 62 ( S) results are inconsistent with the two metal complexes 6 4d 56 0 acting independently (entry 10 in particular would have given a product with a high enantiomeric excess in favour 7 (R,R )-1 + ( S,S )-2 62 72 ( R) of the ( S)-enantiomer if this were the case), but are entirely 8 (R,R )-1 + ( R,R )-2 79 90 ( S) consistent with the in situ formation of a heterobimetallic 9 4 + ( S,S )-2 68 79 ( R) complex (Figure 1A) as the catalytically active species. 10 (R,R )-1 + 5 100 18 ( S) Similarly, the measured first order rate constants (Figure 6) for these two reactions (3.1 x 10 -4 s -1 for a mixture of 11 (R,R )-1 + 4 20 34 ( S) complexes 4 and ( S,S )-2; and 2.3 x 10 -5 s -1 for a mixture of a) Unless otherwise stated, all reactions were conducted at room complexes ( R,R )-1 and 5) were incompatible with reactions temperature in dichloromethane for 24 hours using 0.1 mol% of catalyst. involving just the more reactive species present in the b) Whenever a mixture of two catalysts was used, the mixture was reaction mixture, as is the previously reported rate data 22 equimolar in each metal ion. c) After 48 hours. d) Reaction in chloroform. obtained using different ratios of ( R,R )-1 and ( S,S )-2.

1 Figure 5: Variable concentration H NMR spectra of compound 4 in CDCl 3. Expansion of the aromatic and imine region. The imine signals of the three species present have been labeled A, B and C

8 Tetrahedron

Figure 6 : First order kinetics plots for the asymmetric addition of trimethylsilyl cyanide to benzaldehyde using various catalysts.

Complex 1 in CH2Cl2 0 Complex 4 in CH2Cl2 0 2 4 6 8 10 12 14 -0.79

-0.5 0 50 100 150 200 -0.795 y = -0.1435x - 0.7721 -1 y = -0.0001x - 0.7898 R2 = 0.999 -0.8 R2 = 0.9880

-1.5 -0.805 ln[PhCHO] ln[PhCHO] -2 -0.81

-2.5 -0.815

-0.82 -3 time (minutes) time (minutes)

Complex 1 in CDCl3 -0.6 Complex 4 in CDCl3 -0.5 0 5 10 15 20 25 30 0 50 100 150 200 250 300 -0.65 -0.55

y = -0.001x - 0.5629 -0.6 -0.7 R2 = 0.9559

y = -0.0071x - 0.6509 -0.65 -0.75 R2 = 0.9798 -0.7 ln[PhCHO] ln[PhCHO] -0.8 -0.75

-0.85 -0.8

-0.9 -0.85 time (minutes) time (minutes)

Complexes 4 and SS-5 in CH2Cl2 Complexes RR-1 and 5 in CH2Cl2 -0.7 -0.8 0 10 20 30 40 0 50 100 150 -0.8 -0.85 y = -0.0185x - 0.7322 -0.9 2 R = 0.9884 y = -0.0014x - 0.8069 -0.9 -1 R2 = 0.9813

-1.1 ln[PhCHO] ln[PhCHO] -0.95

-1.2 -1 -1.3

-1.4 -1.05 time (minutes) time (minutes)

Assuming that the salen ligand closest to the newly forming similar to the situation when only complex 2 is used as stereocenter has the largest influence on the catalyst (compare Table 1: entries 2 and 9). In contrast, enantioselectivity of the reaction, these two results also when the salen ligand attached to the vanadium ion is indicate that in the transition state, the aldehyde is bound to achiral, an almost racemic product is formed, with just a the vanadium rather than the titanium ion. When the salen small enantiomeric excess in favour of the ( S)-enantiomer ligand attached to the vanadium ion is chiral and has ( S)- due to the remote influence of the chiral salen ligand with configuration (Table 1: entry 9), then a product with high (R)-configuration attached to the titanium ion. If the new enantiomeric excess and ( R)-configuration is formed stereocenter in the product had been created whilst the Tetrahedron 9 aldehyde was coordinated to titanium, a high enantiomeric able to catalyse the reaction. However, this result can be excess comparable to that obtained using catalyst 1 alone explained by in situ formation of a heterobimetallic would have been expected in this case (compare Table 1: complex (Figure 1A), which in this case is catalytically entries 1 and 10). inactive. Essentially complete formation of the heterobimetallic complex leaves no complex 1 available to Use of equimolar amounts of chiral 1 and achiral 4 titanium catalyse the reaction. catalysts (Table 1: entry 11) gave a very low conversion to a product with an enantiomeric excess of just 34%. Again, only the in situ formation of a bimetallic complex (Figure Table 2. Use of catalysts 1, 2, 4 in the synthesis of mandelonitrile ethyl 1A) can explain this result. In the absence of such a carbonate at room temperature. complex, catalytically active species derived from complex Conversion Enantiomeric excess (%) 1 would have dominated the catalysis leading to the entry Catalyst a formation of ( S)-mandelonitrile trimethylsilyl ether in high (%) (absolute configuration) yield and with high enantiomeric excess. The low 1 (R,R )-1 100 75 ( S) conversion observed is however consistent with the 2 (S,S )-2 0 formation of a bimetallic complex in which the achiral 3 (R,R )-1 + ( S,S )-2 0 ligand preferentially adopts the Λ-configuration, thus forming the thermodynamically preferred anti - 4 4 20 0 ∆(RR )Λ heterochiral complex with catalytic activity 5 4c 58 0 comparable to that observed with complex 4. The moderate 6 (R,R )-1 + 4 19 35 ( S) value for the enantiomeric excess of the product is also 7 (R,R )-1b 100 68 ( S) consistent with little or no preference for the aldehyde to bind to either titanium ion in the transition state, leading to 8 4 + (S,S )-2 0 a product with enantiomeric excess approximately mid-way a) Whenever a mixture of two catalysts was used, the mixture was between that observed using complexes 1 and 4 equimolar in each metal ion, and a total of 5 mol% of catalyst relative to individually (compare Table 1: entries 1, 3 and 11). benzaldehyde was used. b) 2.5 mol% of catalyst relative to benzaldehyde was used. c) Reaction in chloroform. 2.4 Catalysis of the asymmetric addition of ethyl cyanoformate to benzaldehyde The achiral catalyst 4 is catalytically active for the Complex 1 will also catalyse the asymmetric addition of synthesis of mandelonitrile ethyl carbonate (Table 2: entry ethyl cyanoformate to benzaldehyde, 7 giving mandelonitrile 4), though with much reduced reactivity compared to ethyl carbonate as shown in Scheme 3. Optimal complex 1, exactly as observed for catalysis of the addition enantioselectivity in this reaction is observed at -40 oC, and of trimethylsilyl cyanide to benzaldehyde. The achiral under these conditions, complex 2 is totally inactive as a catalyst was however significantly more active when used catalyst for this reaction. However, in this study we chose in chloroform (Table 2: entry 5), again consistent with to work at room temperature to increase the reaction rates more catalytically active species being present in the more observed with various bimetallic complexes. Table 2 polar solvent as discussed above for reactions involving summarizes the key results obtained using various trimethylsilyl cyanide. Use of an equimolar mixture of bimetallic complexes to catalyse the reaction shown in complexes 1 and 4 (Table 2: entry 6), gave a conversion Scheme 3. which was comparable to that obtained by complex 4 alone, and a product with enantiomeric excess approximately half O way between those obtained using complexes 1 and 4 alone catalyst 1 (5.0 mol%) (compare Table 2: entries 1, 4 and 6). That this result could room temperature, PhCHO + EtOCOCN CH Cl O OEt not be due to catalysis just by complex 1 was confirmed by 2 2 a control experiment using 2.5 mol% of complex 1 (Table Scheme 3 Ph CN 2: entry 7). This is the same amount of complex 1 that was added to the mixed catalyst reaction (Table 2: entry 6), but Entry 1 of Table 2 confirms that catalyst 1 is active for this the enantiomeric excess of the product was far higher in the reaction, though the enantioselectivity observed at room latter case. The qualitative trend shown by entries 1, 4, 6 temperature is about 20% lower than that obtained at -40 and 7 of Table 2 was confirmed by a study of the kinetics oC. 7 Table 2: entry 2 similarly confirms, that even at room of these three reactions (Figure 7), though in this case the temperature, complex 2 displays no catalytic activity for reactions could not be fitted to a simple reaction order due to the co-catalysis by cyanide ion liberated during the the synthesis of mandelonitrile ethyl carbonate. The key 42 result is that shown in Table 2: entry 3, where combined reaction. use of complexes 1 and 2 resulted in complete loss of all of the catalytic activity associated with complex 1. This result Again, these results are not consistent with the model cannot be explained by the two metal(salen) complexes shown in Figure 1B, but are entirely compatible with the acting separately but cooperatively (Figure 1B), since if formation of a catalytically active bimetallic complex this was the case then complex 1 would still be present and (Figure 1A), and with preferential formation of a heterochiral complex in which the chiral salen ligand 10 Tetrahedron adopts the ∆-configuration, whilst the achiral salen ligand opposite of the result obtained using trimethylsilyl cyanide, adopts the Λ-configuration, to give a bimetallic complex and is again totally inconsistent with a mechanism with low intrinsic catalytic activity. 34 Finally, combined use involving intermolecular cyanide transfer between of achiral titanium complex 4 and chiral vanadium complex monometallic species (Figure 1B) as that would have been 2 resulted in formation of a species which was catalytically predicted to produce nearly racemic product in this case. inactive (Table 2: entry 8), a result which is also consistent This result is also inconsistent with sequestration of with formation of a heterobimetallic complex. titanium ions by vanadium ions through oligomer formation 23 as this would leave an excess of vanadium ions present to catalyse the reaction resulting in preferential Figure 7 : Conversion / time plots for the asymmetric addition of ethyl cyanoformate to benzaldehyde catalysed by complexes 1 and 4. formation of ( R)-mandelonitrile. The low conversion observed in this case is however fully consistent with the in 50% situ formation of a heterobimetallic complex, which in this 40% case would be heterochiral ( anti -Ti ∆(RR )-VΛ(SS )) and 30% hence have low intrinsic catalytic activity as discussed above. 20% Conversion % Conversion 10% The achiral complexes 4 and 5 were found to be very poor 0% catalysts for this reaction (Table 3: entries 4 and 5), with 0 500 1000 1500 2000 2500 3000 only a trace of product (<10%) being formed after an 18 o Time (min) hour reaction at -40 C. This is again consistent with both

Mixed catalyst (1+4) Catalyst 1 of these species forming dimeric complexes in which the 34 Catalyst 4 Catalyst 1 (2.5 mol%) ∆Λ -dimer is formed preferentially and is less catalytically active than the homochiral dimers formed from complexes 1 and 2. Use of complexes 4 and 5 in chloroform rather 2.5 Catalysis of the asymmetric addition of potassium than dichloromethane gave significantly higher conversions cyanide to benzaldehyde (Table 3: entries 6 and 7), which is consistent with a greater proportion of the catalytically active ∆∆ -dimer being A unique feature of catalysts 1 and 2 is their ability to 1 catalyse the asymmetric addition of potassium cyanide to present in the more polar solvent, as suggested by H NMR aldehydes in the presence of an anhydride, thus providing a spectroscopy in the case of complex 4. one-pot synthesis of non-racemic cyanohydrin esters. 8 This reaction also requires the presence of both water and tert - butanol as additives with dichloromethane as solvent, and Table 3. Use of catalysts 1, 2, 4, 5 in the synthesis of mandelonitrile acetate. in view of the different reaction conditions for this transformation, it was decided to investigate the efficacy of Conversion Enantiomeric excess (%) entry Catalyst a mixtures of complexes 1, 2, 4, 5 using benzaldehyde and (%) (absolute configuration) acetic anhydride as substrates (Scheme 4). 1 (R,R )-1 100 93 ( S) 1,2,4, or 5 (1.0 mol%) 2 (S,S )-2 70 94 ( R) o -40 C, CH2Cl2 / OAc 3 (R,R )-1 + ( S,S )-2 24 72 ( S) H2O / tBuOH 4 4 <10 0 PhCHO + KCN + Ac O Ph CN 2 5 5 <10 0 Scheme 4 6 4b 27 0 7 5b 47 0 Key results for this reaction are shown in Table 3. As entries 1 and 2 illustrate, both catalysts 1 and 2 are active 8 (R,R )-1 + 5 23 44 ( S) for this transformation, and as for all other reactions 9 4 + (S,S )-2 21 1 ( S) involving these catalysts, the ( R,R )-enantiomer of the 10 (R,R )-1 + 4 73 73 ( S) catalyst leads to the ( S)-enantiomer of mandelonitrile 11 (S,S )-2 + 5 40 46 ( R) acetate. However, unlike the addition of trimethylsilyl a) Whenever a mixture of two catalysts was used, the mixture was cyanide to aldehydes, for this reaction, catalysts 1 and 2 equimolar in each metal ion, and a total of 1 mol% of catalyst relative to result in similar reaction rates and give mandelonitrile benzaldehyde was used. b) reaction carried out in chloroform. acetate with 93-94% ee. Remarkably, when a mixture of catalysts ( R,R )-1 and ( S,S )-2 was used for this reaction When the almost inactive, achiral complexes were mixed (Table 3: entry 3), it was the stereochemistry of the salen with catalysts 1 and 2, they were able to exert a marked ligand attached to the titanium ion that determined the influence on the asymmetric induction for reactions carried configuration of the product; ( S)-mandelonitrile acetate out in dichloromethane (Table 3: entries 8-11). Thus, being obtained with 72% enantiomeric excess. This is the mixing chiral titanium complex 1 with achiral vanadium Tetrahedron 11 complex 5, significantly reduced both the rate of reaction determining the absolute configuration of the product. and the enantioselectivity compared to the use of complex Thus, all catalyst mixtures containing at least 30% of ( R,R )- 1 alone. However, the absolute configuration of the product titanium complex 1 gave predominantly the ( S)-enantiomer was still that expected based on the use of ( R,R )-1 (Table 3: of mandelonitrile acetate. entry 8). Similarly, mixing chiral vanadium complex 2 with achiral titanium complex 4 also significantly reduced the The potassium cyanide system is unique in that it employs rate of reaction and this time the reaction gave essentially an ionic cyanide source, and it may be that the negatively racemic product (Table 3: entry 9). This clearly indicates charged cyanide ion is strongly attracted to the positively that for the asymmetric synthesis of cyanohydrin acetates, charged vanadium(V) ion within a heterobimetallic it is the stereochemistry of the ligand attached to titanium complex. This would then force the aldehyde to coordinate that determines the absolute configuration of the product. to the titanium ion. In contrast, for the addition of When the achiral and chiral complexes of the same metal trimethylsilyl cyanide to aldehydes discussed above, the were mixed (Table 3: entries 10 and 11), in both cases the most polarized bond present is the aldehyde carbonyl and enantiomeric excess of the mandelonitrile acetate was this will be most attracted to the positively charged lower than that obtained from use of just the chiral catalyst vanadium(V) ion, thus accounting for the difference in 1 or 2 and the absolute configuration of the product was binding regiochemistry and hence the difference in also that expected based on the stereochemistry of the asymmetric induction observed when the two cyanide chiral ligand. In view of the different reactivities of the sources are used with a heterobimetallic catalyst. chiral and achiral complexes, these results are again not consistent with the two catalysts acting independently, but 3. Conclusions are what would be expected for a bimetallic catalyst of ∆Λ - configuration formed in situ from one metal ion The stereochemistry of salen ligands present in bimetallic coordinated to a chiral salen ligand and one metal ion complexes has been analysed in detail, and used to derive a µ coordinated to an achiral salen ligand. set of rules which allow the stereochemistry of (salen)M( - O) 2M’(salen)’ complexes to be predicted. The X-ray crystal structure of a bimetallic titanium complex containing the 100 achiral salen ligand derived from N,N -bis -(3,5-di-tert-butyl-

80 2-hydroxybenzylidene)-1,2-diamino-ethane has been determined and the stereochemistry of the complex was as 60 expected on the basis of the rules. 40 When homo-metallic(salen) complexes derived from 20 titanium(IV) and vanadium(V) are mixed and used as 0 catalysts for asymmetric cyanohydrin synthesis, the results

ee (+ = S) = (+ ee -20 can only be explained by the in situ formation of heterobimetallic complexes. The stereochemistry of these -40 heterobimetallic complexes can be predicted and correctly -60 explains both the rates of catalysis and the asymmetric induction. For reactions involving the use of trimethylsilyl -80 cyanide as the cyanating agent, the results indicate that the -100 aldehyde is bound to the vanadium(V) ion during the 0 20 40 60 80 100 asymmetry inducing transition state, whilst cyanide is % (S,S)-V(Salen) 2 coordinated to titanium. However, when potassium cyanide is used as the cyanide source, the ionic cyanide coordinates to the positively charged vanadium(V) ion and the aldehyde Figure 8. Variation of the enantiomeric excess of mandelonitrile acetate is bound to the titanium ion. In all cases, complexes in as the ratio of complexes ( R,R)-1 and ( S,S )-2 is varied. Hashed line is which the two salen ligands have different configurations predicted enantiomeric excess if the catalysts acted independently and ∆ Λ with equal rates, black points and solid curve are the experimentally ( or ) are less catalytically active than complexes in observed results. which the two ligands have the same configuration.

To further investigate the influence of titanium and 4. Experimental vanadium containing species in this reaction, a series of reactions were carried out at different titanium to vanadium 4.1 General methods ratios, using complexes 1 and 2 as the metal ion sources. 1 13 The results of this study are shown graphically in Figure 8, H and C NMR spectra were run on a Bruker Avance 300MHz or JEOL 500MHz spectrometer in the specified which clearly indicates that for the potassium cyanide 13 system, the stereochemistry of the salen ligand attached to solvent. C NMR assignments were made using DEPT the titanium ion is more important than the stereochemistry editing. Infrared spectra were recorded on a Perkin Elmer FT-IR spectrometer. Only significant absorptions are listed. of the salen ligand attached to the vanadium ion in Low and high resolution mass spectra were recorded by the 12 Tetrahedron

48 + EPSRC national mass spectrometry service using LSIMS 1109.6061; C 65 H95 N4O6 Ti 2 (M+Me) requires 1123.6205 48 + with meta -nitrobenzyl alcohol (NOBA) as matrix or ESI and C 64 H93 N4O6 Ti 2 (MH ) requires 1109.6049. ionization techniques. Optical rotations were measured using a Polaar 2001 Optical Activity automatic polarimeter 4.4 Synthesis of complex 5 and are reported with a concentration value in g / 100mL. Melting points were obtained using a Barnstead 4.4.1 Synthesis of the V(IV) complex of ligand 3 Electrothermal 9100 system. Gas chromatography was carried out using a Hewlett Packard 5890 series II fitted Vanadyl sulphate hydrate (0.65 g, 4.0 mmol) was dissolved with a thermal conductivity detector and using hydrogen as in warm ethanol (70 mL). The resulting solution was added the carrier gas. to a solution of N,N -bis -(3,5-di-tert-butyl-2- hydroxybenzylidene)-1,2-diaminoethane 24 (1.8 g, 3.65 4.2 Synthesis of the titanium dichloride complex of mmol) in ethanol (50 mL) and the reaction was heated to ligand 3 reflux under an inert atmosphere for three hours. The solvent was removed in vacuo and the residue was purified N,N -bis -(3,5-di-tert-butyl-2-hydroxybenzylidene)-1,2- by flash chromatography eluting with dichloromethane) to diaminoethane 24 (2.0 g, 4.06 mmol) was dissolved in give the vanadium(IV) complex (1.3 g, 64%) as a 25 o dichloromethane (50 mL) and a solution of titanium paramagnetic green solid. Mp 184-188 C; υmax (CHCl 3) tetrachloride (2.0 M in dichloromethane, 4 mL, 1.1 eq.) was 3019 (s), 2975 (w), 1619 (w) and 1523 cm -1 w. added. The solution was stirred at room temperature for 30 minutes, then transferred to a separating funnel and washed 4.4.2 Oxidation of V(IV) to V(V) complex 5 with water (200 mL) and brine (100 mL). The organic layer was dried (Na 2SO 4) and evaporated in vacuo to leave the Cumene hydroperoxide (1.02 mL, 10 eq.) was added to a title compound (2.47 g, 100%) as a red / brown powder. solution of the vanadium(IV) complex of ligand 3 (0.39 g, o Mp >370 C (decomp.); υmax (CHCl 3) 3429 (br), 3019 (s), 0.7 mmol) in dichloromethane (50 mL) and the reaction -1 2966 (s), 2871 (s), 1621 (s) and 1561cm (s); δH (300 mixture was stirred at room temperature for 18 hours. The MHz, CDCl 3) 8.26 (2H, s, CH=N), 7.54 (2H, s, ArH), 7.24 dichloromethane was evaporated in vacuo and the residue (2H, s, ArH), 4.12 (4H, s, NCH 2CH 2N), 1.45 (18H, s, dissolved in acetonitrile (20 mL) and treated with aqueous 2xC(CH 3)3), 1.25 (18H, s, 2xC(CH 3)3); δ C (75 MHz, hydrochloric acid (2 M, 40 mL). The resulting dark green CDCl 3) 164.7, 161.0, 145.0, 137.5, 131.7, 129.8, 125.8, solution was diluted with water (100 mL), and extracted 58.8, 35.9, 34.8, 31.7, 30.3; m/z (LSIMS) 690 (M+NOBA- with dichloromethane (200 mL). The organic phase was Cl-HCl) +, 573 (M-Cl) +, 539; Found (LSIMS, NOBA): washed with water (100 mL) and brine (100 mL), dried 48 690.3380 and 573.2725; C 39 H52 N3O5 Ti (M+NOBA-Cl- over Na 2SO 4 and passed through a plug of silica gel (4 cm 48 + HCl) requires 690.3381 and C 32 H46 ClN 2O2 Ti (M-Cl) deep by 3 cm diameter) eluting with hexane (100 mL), requires: 573.2722. ethyl acetate (100 mL) and finally methanol (200 mL). The methanol fraction was evaporated in vacuo , giving complex 4.3 Synthesis of complex 4 5 (0.11 g, 28%) as a dark green powder. Mp 277-285 oC; υmax (CHCl 3) 3400 (br), 3019 (s), 2966 (s), 1621 (s) and -1 The titanium dichloride complex of ligand 3 (1.0 g, 1.5 1547 cm ; δH (300 MHz CDCl3) 8.66 (2H, s, CH=N), 7.66 mmol) was dissolved in dichloromethane (500 mL). (2H, s, ArH), 7.42 (2H, s, ArH), 4.54 (2H, br s, Aqueous pH7 sodium phosphate buffer solution (100 mL) NCH 2CH 2N), 4.23 (2H, br s, NCH 2CH 2N), 1.45 (18H, s, was added and the mixture stirred vigorously at room 2xC(CH 3)3), 1.29 (18H, s, 2xC(CH 3)3); δC (75 MHz, temperature for one hour, after which time the aqueous CDCl 3) 165.1, 163.8, 144.5, 136.4, 132.5, 128.7, 121.5, layer was removed and replaced by a fresh portion (100 61.0, 36.0, 34.8, 31.7, 30.3; m/z (ESI, CH 2Cl 2) 557 (M-Cl- + mL) of the buffer. The reaction mixture was stirred for a H2O) , 279, 207; Found (ESI, CH 2Cl 2): 557.2937; 51 + further one hour, the aqueous layer was again removed and C32 H46 N2O3 V (M-Cl-H2O) requires 557.2943. replaced by a final portion (30 mL) of the buffer. The reaction mixture was stirred for a further 30 minutes, and 4.5 General procedure for the asymmetric synthesis of was then washed with water (2 x 150 mL) and brine (100 O-trimethylsilyl mandelonitrile mL). The organic phase was dried (Na 2SO 4) and evaporated in vacuo . The solid residue was washed with ether (3 x 5 Catalyst (0.1 mol%) was dissolved in dichloromethane (5 mL) to leave compound 4 (0.81 g, 97%) as a yellow mL). Benzaldehyde (0.1 mL, 0.94 mmol) was added o powder. Mp >280 C (decomp.); υmax (CHCl 3) 3430 (br), followed by trimethylsilyl cyanide (0.15 mL, 1.1 eq.). The 3019 (s), 2963 (s), 2906 (s), 2869 (s), 1628 (s) and 1556 resulting solution was allowed to stir at room temperature -1 cm (s); δH (300 MHz, C 6D6) 7.75 (4H, d J 2.5 Hz, ArH), for 24 hours and then filtered through a silica filled pipette, 7.61 (4H, s, CH=N), 7.02 (4H, d J 2.5 Hz, ArH), 4.25-4.23 eluting with dichloromethane. The solvent was evaporated (4H, m, NCH 2CH 2N), 2.92-2.90 (4H, m, NCH 2CH 2N), 1.84 in vacuo to leave the title compound as a pale yellow oil. 1 (36H, s, 4xC(CH 3)3), 1.43 (36H, s, 4xC(CH 3)3); δC(125 The % conversion was determined by H NMR MHz, C 6D6) 164.88, 162.38, 138.61, 138.36, 128.91, spectroscopy by integration of the aldehyde signal of 127.37, 122.47, 65.86, 35.87, 34.08, 31.64, 30.66; m/z unreacted benzaldehyde, and the CHCN signal of the (ESI, MeOH) 1123.7 (M+Me) +, 1109.7 (MH +), 569.4 product. To determine the enantiomeric excess, a sample of (M+2Me) 2+ ; Found (ESI, MeOH): 1123.6196 and the product (0.1 g, 0.53 mmol) was dissolved in acetonitrile Tetrahedron 13

(5 mL) and scandium(III) triflate (2.6 mg, 0.01 eq.) was 4.8 Kinetics of the addition of ethyl cyanoformate to added followed by acetic anhydride (108 mg, 2.0 eq.). The benzaldehyde reaction mixture was allowed to stir at room temperature for 30 minutes before being filtered through a silica filled To a stirring solution of catalyst (5 mol%) in pipette, eluting with acetonitrile. The solvent was dichloromethane (6 mL) was added ethyl cyanoformate evaporated in vacuo to leave O-acetyl mandelonitrile as a (0.18mL, 1.96mmol) followed by benzaldehyde (0.1mL, colourless oil. The enantiomeric excess could then be 0.98mmol). At selected time intervals, a sample of the determined by chiral GC using a γ-CD butyryl, fused silica reaction mixture (0.5mL) was removed and filtered through capillary column (30 m x 0.25 mm) with hydrogen (1.6 mL a silica column (ca 6 cm) eluting with further per minute) as the carrier gas. With an initial temperature dichloromethane (ca. 5mL). The resulting colorless eluent of 100 oC for 2 minutes, followed by a ramp rate of 5 oC was evaporated in vacuo and then analysed by 1H NMR per minute to a final temperature of 180 oC, O-acetyl (R)- spectroscopy using the aldehyde signal (10 ppm) and mandelonitrile has a retention time of 15.7 minutes and O- cyanohydrin ethyl carbonate signal (6.2 ppm) to calculate acetyl (S)-mandelonitrile has a retention time of 15.9 the % conversion. minutes. 4.9 General procedure for the asymmetric synthesis of 4.6 Kinetics of the addition of trimethylsilyl cyanide to mandelonitrile acetate benzaldehyde A sample of catalyst (1 mol%) was dissolved in To a dry sealed round-bottomed flask fitted with a stirrer dichloromethane (10 mL) and cooled to –78°C. Potassium was added a solution of the appropriate catalyst (0.00196 cyanide (0.245g, 3.6 mmol), tert -butanol (0.135 mL) and mmol) in dry CH 2Cl 2 or CHCl 3 (1.75 mL). The solution water (0.01 mL) were added, and the reaction transferred to was cooled to 0 oC and an aliquot (0.7 µL) was taken and a cryostatically cooled bath at –40 °C. Benzaldehyde (0.1 diluted with dry CH 2Cl 2 (3.0 mL). The absorbance of the mL, 0.94 mmol) and acetic anhydride (0.356 mL, 3.6 diluted sample was measured at 240-260 nm, this reading mmol) were added and the resulting mixture stirred being used for baseline calibration. Freshly distilled vigorously at –40°C overnight. The resulting mixture was benzaldehyde (0.1 mL, 1.0 mmol) was added to the filtered through a silica filled pipette washing with further reaction mixture which was stirred for 10 minutes and the dichloromethane (5 mL) and evaporated in vacuo and the first aliquot (t = 0) was taken as described above. Then, residue analysed without further purification. The % trimethylsilyl cyanide (0.15 mL, 1.1 mmol) was added and conversion was determined by 1H NMR spectroscopy by aliquots were taken at intervals of 1, 2, 3, 4, 5, 7.5, 10, 15, integration of the aldehyde signal of unreacted 22, and 33 minutes. The absorbance data were used to benzaldehyde, and the CHCN signal of the product. The construct a first order kinetics plot. enantiomeric excess of the product was determined by chiral GC as discussed above for the product of 4.7 General procedure for the asymmetric synthesis of trimethylsilyl cyanide addition to benzaldehyde. mandelonitrile ethyl carbonate 4.10 X-ray crystallography. Crystal data for 4 To a solution of catalyst (5 mol%) in dichloromethane (3 mL) was added benzaldehyde (0.05 mL, 0.47 mmol) and C64 H92 O6N4Ti 2⋅2CH 2Cl 2, M = 1279.1, monoclinic, space ethyl cyanoformate (0.1 mL, 0.94 mmol). The reaction group P21/c, a = 13.124(2), b = 22.928(4), c = mixture was stirred at room temperature for 16 hours, then 11.6837(18)Å, β = 92.838(2) °, V = 3511.3(9)Å 3, Z = 2 the solution was passed through a Pasteur pipette (crystallographically centrosymmetric dimers), T = 150 K, − containing silica, eluting with dichloromethane (ca 3 mL). µ(Mo Kα, λ = 0.71073Å) = 0.43 mm 1. 25105 measured The eluent was evaporated in vacuo and the residue data (Bruker SMART diffractometer, θ < 25 °, corrected analysed without further purification. The % conversion semi-empirically for absorption), 6184 unique ( Rint = 1 2 was determined by H NMR spectroscopy by integration of 0.031), 383 refined parameters, R ( F, F > 2 σ) = 0.045, Rw the aldehyde signal of unreacted benzaldehyde, and the (F2, all data) = 0.115, goodness-of-fit = 1.12, final − CHCN signal of the product. The enantiomeric excess of difference map features within ±0.70 eÅ 3. H atoms were the product was determined by chiral GC using a γ-CD constrained. Standard software (Bruker SMART, SAINT, butyryl, fused silica capillary column (30 m x 0.25 mm) SADABS and SHELXTL) and procedures were used. with hydrogen (1.6 mL per minute) as the carrier gas. With CCDC-612829 contains supplementary crystallographic an initial temperature of 100 oC for 2 minutes, followed by data for this paper. These data can be obtained free of a ramp rate of 0.2 oC per minute to a final temperature of charge via www.ccdc.cam.ac.uk/conts/retrieving.html (or 130 oC, ( R)-mandelonitrile ethyl carbonate has a retention from the Cambridge Crystallographic Data Centre, 12 time of 120 minutes and ( S)-mandelonitrile ethyl carbonate Union Road, Cambridge CB2 1EZ, UK; fax (+44) 1223- has a retention time of 123 minutes. 336-033; or [email protected]).

Acknowledgements

The authors thank the EPSRC for funding through a DTA studentship (to CY) and partial postdoctoral funding for 14 Tetrahedron

RWH, and the EPSRC national mass spectrometry service 13 Wang, B.; Feng, X.; Huang, Y.; Liu, H.; Cui, X.; Jiang, Y. J. at the University of Wales, Swansea for low and high Org. Chem. 2002 , 67 , 2175-2182. resolution mass spectra. Use of the EPSRC's Chemical 14 DiMauro, E.F.; Kozlowski, M.C. Org. Lett. 2002 , 4, 3781- Database Service at Daresbury is also gratefully 3784. acknowledged. 15 Dai, Z.; Zhu, C.; Yang, M.; Zheng, Y.; Pan, Y. Tetrahedron: Asymmetry 2005 , 16 , 605-608. 16 (a) Saito, B.; Katsuki, T. Tetrahedron Lett. 2001 , 42 , 3873- References 3876. (b) Saito, B.; Katsuki, T. Tetrahedron Lett. 2001 , 42 ,

8333-8336. 1 For recent reviews of the synthesis and applications of chiral 17 (a) Belokon, Y.N.; North, M.; Parsons, T. Org. Lett. 2000 , 2, cyanohydrins see: (a) North, M. Tetrahedron: Asymmetry 1617-1619. (b) Belokon, Y.N.; Maleev, V.I.; North, M.; 2003 , 14, 147-176. (b) Brunel, J.-M.; Holmes, I.P. Angew. Usanov, D.L. Chem. Commun. 2006 , 4614-4616. Chem., Int. Ed. 2004 , 43, 2752-2778. 18 (a) Baleizão, C.; Gigante, B.; Garcia, H.; Corma, A. J. Cat. 2 For leading references to other work on asymmetric 2003 , 215, 199-207. (b) Baleizão, C.; Gigante, B.; Das, D.; cyanohydrin synthesis see: (a) Steele, R.M.; Monti, C.; Alvaro, M.; Garcia, H.; Corma, A. Chem. Commun. 2003 , Gennari, C.; Piarulli, U.; Andreoli, F.; Vanthuyne, N.; 1860-1861. (c) Baleizão, C.; Gigante, B.; Garcia, H.; Corma, Roussel, C. Tetrahedron: Asymmetry 2006 , 17 , 999-1006. (b) A. Tetrahedron 2004, 60 , 10461-10468. (d) Baleizão, C.; Baeza, A.; Casas, J.; Nájera, C.; Sansano, J.M.; Saá, J.M. Gigante, B.; Garcia, H.; Corma, A. J. Cat. 2004 , 221, 77-84. Euro. J. Org. Chem. 2006 , 1949-1958. (c) Baeza, A.; Casas, 19 Nakajima, K.; Kojima, K.; Kojima, M.; Fujita, J. Bull. Chem. J.; Nájera, C.; Sansano, J.M. J. Org. Chem. 2006 , 71 , 3837- Soc. Jpn. 1990 , 63 , 2620-2630. 3848. (d) Yamagiwa, N.; Abiko, Y.; Sugita, M.; Tian, J.; 20 Watanabe, A.; Matsumoto, K.; Shimada, Y.; Katsuki, T. Matsunaga, S.; Shibasaki, M. Tetrahedron: Asymmetry 2006 , Tetrahedron Lett. 2004 , 45 , 6229-6233. 17 , 566-573. (e) Xiong, Y.; Huang, X.; Gou, S.; Huang, J.; 21 (a) Belokon, Y.N.; Green, B.; Ikonnikov, N.S.; Larichev, Wen, Y.; Feng, X. Adv. Synth. Catal. 2006 , 348 , 538-544. (f) V.S.; Lokshin, B.V.; Moscalenko, M.A.; North, M.; Orizu, Liu, Y.; Liu, X.; Xin, J.; Feng, X. Synlett 2006 , 1085-1089. C.; Peregudov, A.S.; Timofeeva, G.I. Eur. J. Org. Chem. (g) Moreno, R.M.; Rosol, M.; Moyano, A. Tetrahedron: 2000 , 2655-2661. (b) Belokon, Y.N.; Green, B.; Ikonnikov, Asymmetry 2006 , 17 , 1089-1103. (h) Rodríguez, B.; Pastó, M. N.S.; North, M.; Parsons T.; Tararov, V.I. Tetrahedron 2001 , Jimeno, C. Pericàs, M.A. Tetrahedron: Asymmetry 2006 , 17 , 57 , 771-779. 151-160. (i) Li, Q.; Liu, X.; Wang, J.; Shen, K.; Feng, X. 22 Belokon, Y.N.; North, M.; Maleev, V.I.; Voskoboev, N.V.; Tetrahedron Lett. 2006 , 47 , 4011-4014. (j) Kim, S.S.; Kwak, Moskalenko, M.A.; Peregudov, A.S.; Dmitriev, A.V.; J.M. Tetrahedron 2006 , 62 , 49-53. Ikonnikov, N.S.; Kagan, H.B. Angew. Chem. Int. Ed. 2004 , 3 For reviews of asymmetric catalysis using M(salen) 43 , 4085-4089. complexes see: (a) Canali, L.; Sherrington, D.C. Chem. Soc. 23 For examples of oligomeric vanadium(salen) complexes see: Rev. 1999 , 28 , 85-93. (b) Katsuki, T. Synlett 2003 , 281-297. (a) Mathew, M.; Carty, A.J.; Palenik, G.J. J. Am. Chem. Soc. (c) Cozzi, P.G. Chem. Soc. Rev. 2004 , 33 , 410-421. (d) 1970 , 92 , 3197-3198. (b) Nakajima, K.; Kojima, M.; Azuma, Katsuki, T. Chem. Soc. Rev. 2004 , 33 , 437-444. S.; Kasahara, R.; Tsuchimoto, M.; Kubozono, Y.; Maeda, H.; 4 For a review of the development and applications of catalysts Kashino, S.; Ohba, S.; Yoshikawa, Y.; Fujita, J. Bull. Chem. 1 and 2 see: Achard, T.R.J.; Clutterbuck, L.A.; North, M. Soc. Jpn. 1996 , 69 , 3207-3216. (c) Kasahara, R.; Tsuchimoto, Synlett 2005 , 1828-1847. M.; Ohba, S.; Nakajima, K.; Ishida, H.; Kojima, M. Inorg. 5 Belokon, Y.N.; Caveda-Cepas, S.; Green, B.; Ikonnikov, Chem. 1996 , 35 , 7661-7665. (d) Tsuchimoto, M.; Hoshina, N.S.; Khrustalev, V.N.; Larichev, V.S.; Mosckalenko, M.A.; G.; Yoshioka, N.; Inoue, H.; Nakajima, K.; Kamishima, M.; North, M.; Orizu, C.; Taravov, V.I.; Tasinazzo, M.; Kojima, M.; Ohba, S. J. Solid State Chem. 2000 , 153 , 9-15. Timofeeva, G.I.; Yashkina, L.V. J. Am. Chem. Soc. 1999 , 24 Mucha, F.; Haberecht, J.; Bohme, U.; Roewer, G. Monatsh. 121 , 3968-3973. Chem. 1999 , 130 , 117-132. 6 Belokon, Y.N.; Green, B.; Ikonnikov, N.S.; North, M.; 25 Liu, Z.; Anson, F.C. Inorg. Chem. 2001 , 40 , 1329-1333. Tararov, V.I. Tetrahedron Lett. 1999 , 40 , 8147-8150. 26 Throughout this analysis, the term ‘salen’ is used in its most 7 Belokon, Y.N.; Blacker, J.; Clutterbuck, L.A.; North, M. Org. general sense to refer to ligands derived from any diamine, Lett. 2003 , 5, 4505-4507. not just ethylene diamine. 8 Belokon, Y.N.; Carta, P.; Gutnov, A.V.; Maleev, V.; 27 For a review of the stereochemistry of metal complexes see: Moskalenko, M.A.; Yashkina, L.V.; Ikonnikov, N.S.; Knof, U.; von Zelewsky, A. Angew. Chem. Int. Ed. 1999 , 38 , Voskoboev, N.V.; Khrustalev, V.N, North, M. Helv. Chim. 302-322. Acta 2002 , 85 , 3301-3312. 28 For the definition of this terminology see: Sargeson, A.M.; 9 (a) Lundgren, S.; Wingstrand, E.; Penhoat, M.; Moberg, C. J. Searle G.H. Inorg. Chem. 1965 , 4, 45-52. Due to the structure Am. Chem. Soc. 2005 , 127, 11592-11593. (b) Wingstrand, E.; of the salen ligand, cis -β complexes containing salen ligands Lundgren, S.; Penhoat, M.; Moberg, C. Pure Appl. Chem. are usually distorted compared to the ideal geometry, but this 2006 , 78 , 409-411. is not relevant to the stereochemical analysis. 10 Huang, W.; Song, Y.; Bai, C.; Cao, G.; Zheng, Z. 29 For a detailed discussion of the stereochemistry of bimetallic Tetrahedron Lett. 2004 , 45 , 4763-4767. salen complexes see: Tararov, V.I.; Larichev, V.S.; 11 (a) Huang, W.; Song, Y.; Wang, J.; Cao, G.; Zheng, Z. Moscalenko, M.A.; Yashkina, L.V.; Khrustalev, V.N.Antipin, Tetrahedron 2004 , 60, 10469-10477. (b) Khan, N.H.; M.Y.; Börner, A.; Belokon, Y.N. Enantiomer 2000 , 5, 169- Agrawal, S.; Kureshy, R.I.; Abdi, S.H.R.; Mayani, V.J.; Jasra, 173. R.V. Tetrahedron: Asymmetry 2006 , 17 , 2659-2666. 30 The cis -β configuration is also found in octahedral 12 (a) Wu, J.; Hou, X.-L.; Dai, L.-X.; Xia, L.-J.; Tang, M.-H. monometallic complexes containing a salen ligand and a Tetrahedron: Asymmetry 1998 , 9, 3431-3436. (b) Yang, M.; bidentate ligand: (a) Carroll, K.M.; Schwartz, J.; Ho, D.M. Zhu, C.; Yuan, F.; Huang, Y.; Pan, Y. Org. Lett. 2005 , 7, Inorg. Chem. 1994 , 33 , 2707-2708. (b) Lloret, F.; Julve, M.; 1927-1930. Mollar, M.; Castro, I.; Latorre, J.; Faus, J.; Solans, X.; Morgenstern-Badarau, I. J. Chem. Soc., Dalton Trans. 1989 ,

Tetrahedron 15

729-738. (c) Malfant, I.; Morgenstern-Badarau, I.; Philoche- Levisalles, M.; Lloret, F. Chem. Commun. 1990 , 1338-1340. (d) Vergopoulos, V.; Jantzen, S.; Rodewald, D.; Rehder, D. Chem. Commun. 1995 , 377-378. (e) Gullotti, M.; Pasini, A.; Zanderighi, G.M.; Ciani, G.; Sironi, A. J. Chem. Soc., Dalton Trans. 1981 , 902-908. (f) Bailey, N.A.; Higson, B.M.; McKenzie, E.D. J. Chem. Soc., Dalton Trans. 1972 , 503-507. (g) Lauffer, R.B.; Heistand (II), R.H.; Que (Junior), L. Inorg. Chem. 1983 , 22 , 50-55. (h) Calligaris, M.; Manzini, G.; Nardin, G.; Randaccio, L. J. Chem. Soc., Dalton Trans. 1972 , 543-546. (i) Ohba, M.; Tamaki, H.; Matsumoto, N.; Okawa, H. Inorg. Chem. 1993 , 32 , 5385-5390. (j) Agustin, D.; Rima, G.; Gornitzka, H.; Barrau, J. J. Organomet. Chem. 1999 , 592 , 1-10. (k) Maroney, M.J.; Day, R.O.; Psyris, T.; Fleury, L.M.; Whitehead, J.P. Inorg. Chem. 1989 , 28 , 173-175. (l) Evans, W.J.; Fujimoto, C.H.; Ziller, J.W. Polyhedron 2002 , 21 , 1683-1688. 31 Davis, A.; Kilner, C.A.; Kee, T.P. Inorganica Chimica Acta 2004 , 357 , 3493-3502. 32 Kless, A.; Lefeber, C.; Spannenberg, A.; Kempe, R.; Baumann, W.; Holz, J.; Borner, A. Tetrahedron 1996 , 52 , 14599-14606. 33 Tsuchimoto, M. Bull. Chem. Soc. Jpn. 2001 , 74 , 2101-2105. 34 Matsumoto, K.; Sawada, Y.; Saito, B.; Sakai, K.; Katsuki, T. Angew. Chem. Int. Ed. 2005 , 44 , 4935-4939. 35 (a) Torayama, H.; Nishide, T.; Asada, H.; Fujiwara, M.; Matsushita, T. Polyhedron 1998 , 17 , 105-118. (b) Blaauw, R.; van der Baan, J.L.; Balt, S.; de Bolster, M.W.G.; Klumpp, G.W.; Kooijman, H.; Spek, A.L. Inorganica Chemica Acta 2002 , 336 , 29-38. (c) Negoro, S.; Asada, H.; Fujiwara, M.; Matsushita T. Inorg. Chem. Commun. 2003 , 6, 357-360. (d) Gohdes, J.W.; Armstrong, W.H. Inorg. Chem. 1992 , 31 , 368- 373. (e) Si, S.-F.; Tang, J.-K.; Liao, D.-Z.; Jiang, Z.-H.; Yan, S.-P. J. Molecular Structure 2002 , 606 , 87-90. (f) Aurangzeb, N.; McAuliffe, C.A.; Pritchard, R.G.; Watkinson, M. Acta Cryst. 1993 , 49C , 1945-1947. (g) Maslen H.S.; Waters T.N. J. Chem. Soc., Chem. Commun. 1973 , 760-761. 36 Atwood, D.A.; Jegier, J.A.; Rutherford, D. Inorg. Chem. 1996 , 35 , 63-70. 37 Larson, E.J.; Pecoraro, V.L. J. Am. Chem. Soc. 1991 , 113 , 3810-3818. 38 Law, N.A.; Kampf, J.W.; Pecoraro, V.L. Inorganica Chemica Acta 2000 , 297 , 252-264. 39 Belokon, Y.N.; Blacker, A.J.; Carta, P.; Clutterbuck, L.A.; North, M. Tetrahedron 2004 , 60 , 10433-10447. 40 Norsikian, S.; Holmes, I.; Lagasse, F.; Kagan, H.B. Tetrahedron Lett. 2002 , 43 , 5715-5717. 41 A titanium complex of unknown structure derived from ligand 3, titanium tetra -isopropoxide and N,N - dimethylaniline-N-oxide has in contrast been shown to be a highly effective catalyst for the addition of trimethylsilyl cyanide to ketones: Chen, F.; Feng, X.; Qin, B.; Zhang, G.; Jiang, Y. Synlett 2003 , 558-560. 42 Belokon, Y.N.; Ishibashi, E.; Nomura, H.; North, M. Chem. Commun. 2006 , 1775-1777.