Mechanistic examination of AuIII-mediated 1,5-enyne cycloisomerization by AuBr2(N-imidate)(NHC)/AgX precatalysts – is the active catalyst AuIII or AuI?

3524 | Catal. Sci. Technol., 2014, 4, 3524–3533 This journal is © The R Department of Chemistry, University of York, Heslington, York YO10 5DD, UK. E-mail: ian.fairlamb@york.ac.uk GlaxoSmithKline, New Frontiers Science Park, Third Avenue, Harlow, Essex CM19 5AW, UK † Electronic supplementary information (ESI) available: Representative NMR spectra and other characterization data. See DOI: 10.1039/c4cy00617h ‡ Novartis Pharmaceuticals UK Limited, Horsham Research Centre, Wimblehurst Road, GB-Horsham, West Sussex RH12 5AB, U.K. Scheme 1 Top: 1,5-enyne cycloisomer direct selectivity (3 & 4 representative prod a tandem nucleophilic substitution-1,5-enyn 1,3-rearrangements are possible for 8 (R/R Cite this: Catal. Sci. Technol., 2014, 4, 3524


Introduction
Organic transformations catalysed by Au species are of current interest within both the synthetic chemistry and catalysis communities. 1 Of the reaction classes, 1,n-enyne cycloisomerization processes (n = 5 and 6) have attracted notable attention. 2 While several transition metals are able to affect 1,n-enyne substrates, Au occupies a special position in this arena, 2,3 as an eclectic array of structurally diverse organic products can be formed, depending on the catalyst and substrate identities, in addition to the reaction conditions used.
1,6-Enyne cycloisomerization processes generally involve 6-exo-dig cyclization and are mechanistically well understood. 4 1,5-Enyne cycloisomerization processes, involving 5-endo-dig cyclization, have not been examined to the same extent, despite the numerous synthetic possibilities. Nevertheless, several catalyst manifolds are known to deliver different products depending on the electronic character of the Au centre (see Scheme 1 for selected examples). This prompted Sun and Lee to explore the mechanisms of 1,5-enyne cycloisomerization, mediated by Au I , by DFT calculations (B3LYP). 5 In an earlier study we developed a tandem nucleophilic substitution-1,5-enyne cycloisomerization process (5→8), 6 starting from propargyl alcohol (5) and allyl silane (6) starting materials, to deliver bicyclo[3.1.0]hexene products (8). For this process, Au III Br 2 (N-TFS)(I t Pe) in combination with the noncoordinating anion, Ag[Al(OC(CF 3 ) 4 ], proved an active and selective catalyst for the tandem process. By comparison with Au I Br(I t Pe) (note: near-inactive for the tandem process), it was shown that a Au III species was required for the initial nucleophilic substitution reaction. However, a question that arose from the study was the nature of the active catalyst species in the cycloisomerization step, i.e. is Au III -or Au I responsible for cycloisomerization? Scheme 1. Top: 1,5-enyne cycloisomerization showing how L can direct selectivity (3 & 4 representative products; others possible). Bottom: a tandem nucleophilic substitution-1,5-enyne cycloisomerization -thermal 1,3rearrangements are possible for 8 (R 1 /R 2 = aryl groups).
In this paper we examine the reaction kinetics of 1,5-enyne cycloisomerization, mediated by AuBr 2 (N-imidate)(NHC) catalysts {where N-imidate = N-tetrafluorosuccinimide (N-TFS) or N-phthalimide (N-phthal) and NHC = N,N'-di-tertpentylimidazol-2-ylidene (I t Pe)} in the presence of AgOTf. The findings allow an understanding of 1,5-enyne substrate turnover, indicating that catalyst decomposition is an issue for certain catalysts. It has been found that the N-imidate anion plays a stabilizing role for effective catalysis. NMR spectroscopic investigations have been used to examine the degradation of Au III to Au I , which support the kinetic studies.

Results and discussion
The study is divided into two parts: (a) kinetic investigations; (b) NMR spectroscopic investigations.

(a) Kinetic investigations
The 1,5-enyne cycloisomerization reaction (9→10) examined in these kinetic investigations is shown in Scheme 2, which has previously been validated as a useful benchmark reaction 6b for testing Au catalyst efficacy (the reaction is selective for 10 and negligible side-reactions are seen).
The catalysts used in this study are collated in Figure 1, which were selected from a wider library of catalysts previously reported by our group. 6a,b The reaction of 9→10 was monitored by HRGC analysis. The response factors are near identical for 9 and 10 (authenticated by 1 H NMR spectroscopic analysis), which is expected as they are isomers. The conversion of pure 9 (ascertained by 1 H NMR spectroscopy, see section b) and associated kinetic profile is shown in Figure 2 for the reaction mediated by 1 mol% AuBr 2 (N-TFS)(I t Pe), in the presence of either 1 or 2 mol% AgOTf in DCE at 25 °C. The first equivalent (relative to Au cat.) serves to abstract bromide from AuBr 2 (N-TFS)(I t Pe) forming [AuBr(N-TFS)(I t Pe)]OTf and AgBr in situ. The second equivalent of AgOTf could in principle abstract another bromide.
Examination of the kinetic profiles in Figure 2 indicate that there is no benefit in adding an additional equivalent of AgOTf. Both AgOTf and AuBr 2 (N-TFS)(I t Pe) are necessary for active catalysis (confirmed by control experiments). The data generated by HRGC could be fitted to a pseudo-first order rate equation using Dynafit TM (by curve fitting; regression analysis of ln[enyne] vs time afforded similar data), which allows the observed rate (k obs ) and initial rate to be determined. 7 For the reaction where the AuBr 2 (N-TFS)(I t Pe):AgOTf was 1:1, it was determined that k obs = 3.14x10 -4 s -1 and the initial rate (T 0 ) = 6.28x10 -5 moldm -3 s -1 (standard error = 2.5 %). A marginally faster reaction was recorded when the ratio was 1:2 (k obs = 4.16x10 -4 s -1 ; T 0 = 8.32x10 -5 moldm -3 s -1 ; standard error = 3.3%). Based on these data the ratio of the Au catalyst to Ag additive was maintained at 1:1 for all subsequent reactions, which is in-keeping with the ratio commonly used for many related cycloisomerization processes.
The kinetic profiles for the conversion of 9 (by HRGC), catalyzed by five different Au catalysts, are shown in Figure 3. The kinetic data is collated in Table 1. AuBr(I t Pe), AuBr 2 (N-TFS)(I t Pe), AuBr 2 (N-phthal)(I t Pe) and AuBr 2 (N-phthal)(IMes) show reasonable first order behaviour in 9. The catalyst showing the highest observed rate is AuBr(I t Pe) (k obs = 13.98x10 -4 s -1 ), which has an initial rate commensurate with AuBr 3 (I t Pe) (T 0 = 27.96x10 -5 moldm -3 s -1 and 21.87x10 -5 , respectively). It is important to note that AuBr 3 (I t Pe) did not display clean first order behaviour and catalyst deactivation is apparent after ca. 60 substrate turnovers. The kinetic profile could, however be fitted to a second order rate equation by curve-fitting, in which it was assumed that the Au catalyst decomposed over time and that the rate depended on both the concentration of 9 and Au catalyst. phthal)(IMes)). 1,5-Enyne conversion monitored by HRGC analysis -each sample point was quenched by addition of (n-Bu)4NBr. The kinetic curves were fitted using Dynafit TM with standard errors < 5%. , k2obs = 9.07 x10 -4 s -1 (error 3.0%).
Catalyst decomposition does not appear to hinder catalysis by the other Au III complexes tested in this series. Comparison of k obs for the AuBr 2 (N-phthal)(I t Pe) and AuBr 2 (Nphthal)(IMes) catalysts, shows that the former is twice as active as the latter (Table 1). AuBr 2 (N-TFS)(I t Pe) has a marginally higher activity than AuBr 2 (N-phthal)(I t Pe). It is evident that the N-imidate ligand could be playing a stabilizing role, when comparing N-imidate-containing catalysts directly with AuBr 3 (I t Pe).
In the course of these kinetic studies it was determined that 'aged' 1,5-enyne significantly affects the kinetic profiles recorded. Aldehyde decomposition products are formed from 9 (observed by 1 H NMR, ca. 5%, after several weeks), which could interfere with the catalysis (i.e. slowing catalytic turnover). It was therefore found essential to employ either freshly prepared or purified 9 for the kinetic studies, which gave reproducible kinetic data.
Further examination of the catalytic behaviour of AuBr 2 (Nphthal)(I t Pe) by variation of [9] reveals an interesting trend ( Figure 4, Table 2).  [9] in a reaction mediated by AuBr2(N-phthal)(I t Pe) (1 mol%), AgOTf (1 mol%), in DCE at 25 °C (key: ○ = 0.1 M 9; • = 0.2 M 9; □ = 0.5 M 9). 1,5-Enyne conversion monitored by GC analysiseach sample point was quenched by addition of (n-Bu)4NBr. The kinetic curves were fitted using Dynafit TM . Increasing [9] causes a decrease in the observed rate constants, with the initial rates increasing with concentration, relating to the change in [9] over time. At 0.5 M [9] it was necessary to fit the kinetic curve to a second order rate equation, again factoring in the Au catalyst deactivation step. The observation is consistent with increasing [Au], which is concomitant with increasing [9]. Given these data, product inhibition and Au aggregation / decomposition are likely at higher [9]. A question arising from these studies is the nature of the active catalyst species. Therefore, a competition experiment was devised to discern any subtle differences in catalyst behaviour. Dimethyl allylpropargylmalonate 11 was selected as a viable 1,6-enyne benchmark substrate, which principally affords the cyclic products 12a and 12b (Scheme 3).
Firstly, AuBr(I t Pe) is a competent catalyst for the cycloisomerization of 11→12 (k obs = 1.08x10 -4 s -1 ; T 0 = 2.15x10 -5 moldm -3 s -1 ; standard error = 2.2%). The kinetic analysis of the competition experiments is complicated by the presence of two substrates and two competing catalytic cycles, which is evident in the error analysis. This complication aside, the kinetic profiles are similar in all cases (see Fig. 5). Catalyst performance is reduced for AuBr(I t Pe) as compared to the Au III catalysts -the presence of 12 slows catalytic turnover in the case of AuBr(I t Pe). A plot of the ratio of 12a:12b over the duration of the 1,6enyne cycloisomerization reaction, reveals an interesting trend as a function of catalyst ( Figure 6). Firstly, the independent 1,6-enyne cycloisomerisation of 11→12a/12b mediated by AuBr(I t Pe) gives a ratio of ca. 1:3, which does not alter appreciably over the 6 h reaction time. The same trend is revealed in the competition experiment in the presence of both 9 and 12, using AuBr(I t Pe) as the catalyst. The product distribution is different for the Au III catalysts, with AuBr 2 (N-TFS)(I t Pe) showing a smaller ratio than AuBr 2 (Nphthal)(I t Pe). Crucially, at ca. 1.5 h the ratio of 12a/12b is identical to the AuBr(I t Pe) catalyst. This finding, coupled with our previous observation that dual Au III /Au I catalyst behaviour is necessary for a successful tandem nucleophilic / 1,5-enyne cycloisomerization reaction, 6a indicates that Au III to Au I degradation under the working reaction conditions, affects the ratio of products depending on distribution of Au III /Au I species that are present at any given time. It may be conjectured that liberated N-imidate anion could affect the distribution of 12a/12b under the reaction conditions. However, in control experiments we have detected liberation of Br 2 from these Au III catalysts (see section b, below).

(b) NMR spectroscopic investigations
NMR spectroscopic analysis allows us to discern whether Au III could be playing a role in the catalysis detailed earlier. It has been suggested previously that Au III can be reduced to Au I under the reaction conditions. 8 Nolan and co-workers 8a stated that the active catalyst in the hydration of alkynes and polymerisation of styrene mediated by AuBr 3 (NHC) type complexes is a Au I species, formed by reductive elimination of bromide ligands. Further examination of the behaviour of the Au catalysts described herein was deemed necessary, especially whether any evidence for the formation of a cationic Au III catalyst species, and degradation to Au I , could be gathered. 1 H NMR spectroscopy was used to assess whether cationic Au III species were formed by a stoichiometric reaction of AuBr 2 (N-TFS)(I t Pe) with AgOTf in acetone-d 6 (this solvent was selected to increase solubility of all the salts). The I t Pe imidazole proton signals were the most characteristic and sensitive to changes in the Au electronic configuration. It was necessary to compare NMR spectra with reference materials, namely AuBr(I t Pe) and Au(N-TFS)(I t Pe), as shown in Figure 7.
The AuBr(I t Pe) imidazole signal shifts 0.15 ppm downfield on treatment with AgOTf, giving Au(OTf)(I t Pe) quantitatively {compare spectra (h) and (f), Figure 7}. AuBr 2 (N-TFS)(I t Pe) reacts with AgOTf to afford Au(N-TFS)(I t Pe) and trace Au(OTf)(I t Pe), with no cationic Au III species observed after 4.5 h {compare spectra (d), (e) and (i). The outcome is consistent with the liberation of Br 2 (as noted by a brown colouration).  The binding of the AuBr 2 (N-TFS)(I t Pe) and AgOTf toward 1-hexene in CD 2 Cl 2 was evaluated to reveal evidence for any alkene-Au III interaction (Table 4), in addition to showing whether liberated Br 2 could be sequestered by 1-hexene. For comparison, the reported spectroscopic data 9 of [Au(µ 2 -H 2 C=CHC 4 H 9 )(IPr)][SbF 6 ] is included in Table 4. A reduction in the alkene proton coupling constants, relative to free 1hexene, is evidence for π-back donation, i.e. an increase in p character for the orbitals at carbon. The 1 H NMR spectral data show no evidence for 1-hexene binding to neutral AuBr 2 (N-TFS)(I t Pe), in the absence of AgOTf (Table 4). Upon mixing AuBr 2 (N-TFS)(I t Pe) with 1 equiv. of AgOTf, the solution turns a cloudy colour (due to AgBr formation). Addition of 1-hexene (1 equiv.) results in a shift of ∆δ = 0.44 ppm (relative to free 1-hexene). Coupling constants of 18.0 Hz (∆J E-H = +1 Hz) for the trans-proton and 8.7 Hz (∆J Z-H = -1.4 Hz) for the cis-coupling were recorded. Addition of 2 equiv. of 1-hexene leads to a reduction in the chemical shift (∆δ = 0.18 ppm), with little change in coupling constants observed. The excess 1-hexene is therefore in exchange with the alkene-Au III species or alkene-Ag I species.
After 24 h, the 1 H NMR spectrum had changed significantly, with a number of small ItPe imidazole signals corresponding to Au I and Au III species, in addition to a large broad signal (ca. 80% of the total imidazole proton signal) in the Au I region.
Another broad signal at δ 4.45-4.09 ppm was recorded, which we tentatively suggest is the alkene coordinated to Au I ; the broadness in the proton signal is due to fluxionality of the alkene as either Au(µ 2 -H 2 C=CHC 4 H 9 )(N-TFS)(I t Pe) or the naked cationic species [Au(µ 2 -H 2 C=CHC 4 H 9 )(I t Pe)] + . Signals corresponding to free 1-hexene and a new set of proton signals corresponding to 1,2-dibromohexane, ca. 50% abundance compared to free 1-hexene, i.e. 1 equivalent. After 24 h, there was no evidence of 1-hexene binding to Ag or Au. The 19 F NMR spectrum showed a very large broad signal (-182.2 ppm) suggesting the observed alkene complex is that of Au(µ 2 -H 2 C=CHC 4 H 9 )(N-TFS)(I t Pe). In the absence of 1-hexene, the decomposition of AuBr 2 (N-TFS)(I t Pe) was not observed. It is important to note that AgOTf alone binds to 1-hexene (∆δ = 0.39 ppm; ∆J E-H = +0.7 Hz and ∆J Z-H = -1.2 Hz), which is similar to the binding of AuBr 2 (N-TFS)(I t Pe) / 1 equiv. AgOTf. Whilst AgBr is formed, we cannot rule out the involvement of chemical equilibria with AuBr 2 (N-TFS)(I t Pe) and AgOTf. Moreover, under the catalytic conditions there is an excess of substrate and solvent relative to the catalyst system, therefore solubilisation of all the metal species is possible.

(c) Mechanistic discussion
The experimental evidence from this study indicates that AuBr(I t Pe) is the most active catalyst for 1,5-enyne cycloisomerization in the series tested, showing first order behaviour. The Au III catalyst, AuBr 3 (I t Pe), mediates a slower reaction, which was successfully modelled kinetically as a second order process. The inclusion of a catalyst deactivation step was found necessary for the analysis. The 1 H NMR spectroscopic analysis revealed that AuBr 3 (I t Pe) degrades to AuBr(I t Bu), even in the absence of AgOTf.
The presence of an N-imidate anion led to higher catalyst efficacy (defined as leading to full substrate conversion) in comparison with AuBr 3 (I t Pe). 10 Of the N-imidate series, AuBr 2 (N-TFS)(I t Pe), is the most catalytically active. The nature of the NHC ligand exhibits a significant effect. For example, AuBr 2 (N-phthal)(I t Pe) is twice as active as AuBr 2 (Nphthal)(IMes).
Increasing the 1,5-enyne [9], and therefore the concentration of Au catalyst {AuBr 2 (N-phthal)(I t Pe)}, leads to catalyst deactivation. Optimal catalyst performance was recorded at [9] between 0.1-0.2 M. At 0.5 M [9] it was necessary to model the kinetics to a second order process, again involving a catalyst deactivation step. Product inhibition is also possible at higher concentrations, and formation of Au colloids is visible by the naked eye (purple colouration, relating to the Au Plasmon band).
It was possible to assess subtle differences in catalyst behaviour by a concurrent cycloisomerization competition experiment, involving 1,5-enyne 9 and 1,6-enyne 11. The former gives 10 selectively, whereas 11 afford one of two products, 12a or 12b. Three catalysts were tested, namely AuBr(I t Pe), AuBr 2 (N-TFS)(I t Pe) and AuBr 2 (N-phthal)(I t Pe). In all cases, 1,5-enyne 9 was found to be more reactive than 1,6enyne 11. Further analysis of the product ratios 12a:12b from the 1,6-enyne cycloisomerization reaction revealed key differences for the Au III catalysts when compared to AuBr(I t Pe). After ca. 1.5 h, the ratios of 12a:12b for AuBr 2 (N-TFS)(I t Pe) and AuBr 2 (N-phthal)(I t Pe) mirrored the reaction mediated by AuBr(I t Pe). Therefore, while Au III is still present, 12a is formed to a greater extent than in reactions solely mediated by Au I .
Our findings on the 1,5-enyne cycloisomerization process (9→11) are in-keeping with the DFT calculations reported by Sun and Lin. 5 In that study, it was shown that the 1,5-enyne cycloisomerization reaction mediated by AuCl is both kinetically and thermodynamically favoured for the formation of the bicyclo[3.1.0]hexene product, with key catalytic steps (TS-I comp carbene formation and TS-II comp hydrogen migration) being of similar energy (Scheme 4).
From our experimental findings, we propose that Br 2 is liberated from cationic Au III -NHC catalyst species, affording Au I -NHC catalyst species in situ. The finding is supported by the liberation and trapping of the Br 2 by a sacrificial alkene (1hexene) in the NMR spectroscopic investigations. Decomposition of the Au I -NHC catalyst species affords Au colloids, which are most likely a moribund-form, especially as reduced catalyst performance is seen at higher [Au] and [9].
The binding experiments which explored the interaction of 1-hexene with AuBr 2 (N-TFS)(I t Pe) / AgOTf, and AgOTf alone, demonstrates that 1-hexene binding is influenced in both cases. Crucially, 1-hexene mediates the decomposition of AuBr 2 (N-TFS)(I t Pe) to Au(N-TFS)(I t Pe), which was not observed in the absence of 1-hexene in CD 2 Cl 2 . The results taken together demonstrate that unsaturated substrates are involved in: (i) catalyst activation to give Au I , with concomitant decomposition of Au III ; (ii) solubilisation of the Ag salt, which is likely noninnocent in the catalysis. Subsequently, the nature of the coordinating ligand (and substrate and any impurities, as shown by substrate-aging experiments) is likely to affect catalytic turnover. Finally, it is important to emphasise that other products can result from the cycloisomerization of different 1,5-enynes, catalyzed by either AuBr(I t Pe) or AuBr 2 (N-TFS)(I t Pe). In keeping with previous observations, 3a,3d we find that a 1,5enyne containing a highly substituted carbon tether gives a cyclohexadiene product, both in the tandem nucleophilic substitution-1,5-enyne cycloisomerization (13→14→15) and the latter independent reaction (14→15) (Scheme 5). The outcome serves to highlight that the catalysts described here, and elsewhere, mirror related observations from other studies.

Conclusions
In this paper the ability of both Au III and Au I catalysts to mediate a 1,5-enyne cycloisomerization reaction has been examined. It was necessary to understand whether Au III was being reduced under working catalyst conditions, especially as the success of the tandem nucleophilic substitution-1,5-enyne cycloisomerization relies on Au III for the initial nucleophilic substitution step. 6a The reaction kinetics for 1,5-enyne cycloisomerization, mediated by AuBr 2 (N-imidate)(NHC) catalysts {where Nimidate = N-tetrafluorosuccinimide (N-TFS) or N-phthalimide (N-phthal) and NHC = N,N'-di-tert-pentylimidazol-2-ylidene (I t Pe)}, in the presence of AgOTf, revealed that the nature of the N-imidate anionic ligand influences catalyst efficacy. A direct comparison was made with Au III Br 3 (NHC), which exhibited significant catalyst deactivation in the absence of the stabilizing N-imidate ligand. In a concurrent cycloisomerization of 1,5-and 1,6-enynes (9 and 11) we were able to tease out subtle differences in the product distribution (11→12a:12b) during the early stages of catalytic turnover by the Au III catalysts. NMR spectroscopic investigations have allowed the ease of reduction of AuBr 2 (N-TFS)(NHC) to Au I X(NHC) (where X = N-TFS or Br) to be examined. In these experiments, 1-hexene was shown to mediate Au III decomposition. The liberation of Br 2 from the Au III centre was trapped by this sacrificial alkene.
Taking together all of the results, we conclude, that under working catalyst conditions, cationic Au III is reduced to Au I . The role of Ag I has been more difficult to understand (i.e. beyond simple bromide abstraction). The interplay of Au and Ag salts in 1,n-enyne cycloisomerization processes is currently being studied within our laboratories.

Experimental
General experimental details. All reactions involving silver salts were carried out in the absence of light. Deuterated and non-deuterated dichloromethane and acetonitrile were dried by passing through a column of activated alumina. Where necessary dichloromethane Infra-red spectra were recorded on a Unicam Research Series FT-IR spectrometer. Mass spectrometry was carried out using a Fisons Analytical (VG) Autospec instrument. 1 H and 13 C NMR spectra were collected on a JEOL ECX400 spectrometer operating at 400 and 101 MHz, respectively, and referenced to residual solvent signals. 13 C NMR signals are singlets unless otherwise stated. All column chromatography was performed using silica-gel (mesh 220-440) purchased from Fluka Chemicals with the solvent systems specified within the text. TLC analysis was performed using Merck 5554 aluminium backed silica plates, compounds were visualised using UV light (254 nm) and a basic aqueous solution of potassium permanganate. Melting points were measured in open capillary tubes using a Stuart SMP3 Digital Melting Point Apparatus and are uncorrected. 1-Phenyl-2propyn-1-ol and AgOTf were purchased from Alfa Aesar. All other chemicals were purchased from Sigma Aldrich Inc. and used without further purification unless otherwise stated. All of the Au catalysts used in this study have been reported; see our previous paper for full details. 6b

Kinetic Experiments
High resolution gas chromatographic (HRGC) analysis. GC was carried out on a Varian 430 instrument with a Factor Four Capillary column (VF-1ms, 15 m, 0.25 mm) and a flame ionisation detector. Samples of 10 µl were taken via syringe from the reaction mixtures at the specified time points. The samples were immediately quenched by addition of the aliquots to a solution of tetra-n-butylammonium chloride (8 mM, 20 µl) in CH 2 Cl 2 . Conversion was determined via gas chromatography using a 1 µl sample injected directly into the instrument via syringe. Further sampling was carried out on selected aliquots over a period of time to ensure no further conversion occurred after quenching. The errors on the kinetic curves are recorded in the Tables detailed in the main text (see also Tables 1-3).
Values for the integrated areas of the GC signals corresponding to starting materials and products were inputted into Microsoft Excel spreadsheet software. This software was used to calculate percent conversions by comparison of starting material and product signal areas. The concentration of the species in solution was determined from the percent conversions and the initial starting material concentration. Selected aliquots were analysed by 1 H NMR spectroscopy to ensure consistency with conversions measured by GC. 11 Kinetic analysis. The reaction kinetics were calculated assuming a first order kinetic model by plotting ln[enyne] against time. Linear least-squares regression was used to calculate observed rate constants, initial rates and standard errors. The calculated concentrations of starting materials and products during the course of the reactions were also inputted into Dynafit TM software (reported by Biokin). Nonlinear leastsquares regression was used to fit the experimental kinetic data to predetermined molecular mechanisms as described in the main body of the paper. This was used to determine the order of reaction, observed rate constants, initial rates and standard errors and to fit curves to the kinetic data.