CHNH 3
3.4. Discussion
3.4.1. Isotope Effects
Recall that the Strecker synthesis can be divided into the three steps depicted in Figure 3.2: (1) the equilibration between reactants and α-aminonitrile (‘Equilibrium’), (2) the hydrolysis of the α-aminonitrile into an α-aminoamide (‘H-1’), and (3) the subsequent hydrolysis of the α-aminoamide into an α-amino acid (‘H-2’) (Figures 3.1 and 3.2). Of these steps, the first occurs in equilibrium and the latter two are irreversible (Van Trump, 1975).
In circumneutral-to-basic hydrolysis conditions, the equilibrium step continues during the hydrolysis steps. The acidic hydrolysis conditions used in this study inhibit significant α-aminonitrile (α-APN) creation or decomposition following the equilibrium step. These conditions simplify the reaction pathway such that following the equilibrium step, the α-aminonitrile can be treated as an isolated reactant pool for the irreversible hydrolysis reactions. This simplification permits us to consider each of the three steps
independently.
3.4.2. The Equilibrium Step
In the equilibrium step (Figure 3.2a), nitrogen from ammonia and carbon from cyanide and acetaldehyde’s carbonyl carbon increase in bond order when they convert to α-APN (i.e., the product α-APN has more bonds at the cyanide-, acetaldehyde- and ammonia- derived sites than do the reactants). Generally speaking, assuming a similarity of bond
types and bonding partners, when reactions are at equilibrium, we expect the compound with higher bond order to have a higher proportion of heavy-to-light isotopes.
Consequently, we predict inverse equilibrium isotope effects will occur during the
equilibrium step between cyanide carbon and the C-1 site in α-APN; between ammonia to the amine nitrogen site in α-APN, and between acetaldehyde’s carboxyl carbon and the C-2 site in α-APN. Because the methyl carbon on acetaldehyde, which becomes the C-3 carbon in α-APN, is not involved in the reaction, we predict that it will be approximately equal in δ13C in reactant and product.
Figure 3.2a displays the measured isotope effects, isotope ratios, and concentrations at equilibrium for reactants, α-APN, and side products. The δ15N measurements agree with this scenario: we find a +56.6 ‰ inverse EIE between of α-APN’s amine site and
ammonia (Table 3.3). The calculated EIE is heavily impacted by the 29 ‰ EIE value we used for the reaction NH3 (aq) « NH4+ (aq) (Kirshenbaum et al., 1947; Walters et al., 2019), because at equilibrium roughly one-third of the initial ammonia pool is in the form of ammonium (29 ± 1 % to 38 ± 1 %) and only 3 % is in the form of ammonia. The large ratio of the concentration of ammonium to ammonia means that ammonium can sequester large amounts of 13C and errors in this concentration or in the εNH4-NH3 value could result in errors in the εαAPN-NH3 . We note that a recent study measured an εNH4-NH3 of 45 ‰ (Li et al., 2012) and past studies have found εNH4-NH3 values of 20 ‰ (Hogberg, 1997);
however, 29 ‰ remains the εNH4-NH3 with the clearest consensus in both past and recent theoretical and experiments work (Walters et al., 2019). The 56.6 ‰ inverse EIE we infer agrees with our expectations and is likely due to the stronger bonding environment in α-APN (sp3 hybridization) relative to ammonia (sp2 hybridization).
In contrast, the estimated equilibrium isotope effect between acetaldehyde (lower in δ13C) and α-APN (higher in δ13C ) did not follow our expectations, but instead exhibited a negative EIE of -10.0 ‰, averaged across the two relevant sites (Table 3.3, Figure 3.6a). As with the ammonia system, the residual acetaldehyde pool is split between two species, acetaldehyde (20 ± 1 % to 15 ± 1 %) and acetaldehyde hydrate (7 ± 1 % to 17 ± 1 %) (Figure 3.2a, Table S3.1). However, in this case, there is no experimentally verified
EIE for CH3CHO (aq) « CH3CH(OH)2 (aq). Consequently, we use values for the hydration of CO2 (Marlier and O’Leary, 1984), to estimate an approximate εace(OH)2-ace
value of 6.9 ‰ at the carbonyl site with the acetaldehyde hydrate enriched relative to acetaldehyde. This value could be incorrect especially in light of the past theoretic work that estimated εace(OH)2-ace as high as 40 ‰ at the carbonyl site (Hogg, 1980). Because the C-3 site is a spectator atom during all equilibrium steps (i.e., it does not form or change bonds), we infer that the conversion of the acetaldehyde’s carbonyl carbon to α-APN’s C-2 site carries the entire EIE for the C-2 and C-3 sites, and thus that at equilibrium the carbonyl site of a-APN is 20 ‰ lower than that of acetaldehyde. This value does not support our hypothesis of an inverse EIE, but does agree with past work that also found the change from a double to single bond on the carbonyl carbon can lead to normal isotope effects up to -30 ‰ (O’Leary and Marlier, 1979; Marlier, 1993). Like Robins et al. (Robins et al., 2015), we argue the need for further SSIR work on the α-APN
intermediate to elucidate the causes for the negative EIE at the C-2 site. Furthermore, future work that measures the EIE between acetaldehyde and acetaldehyde-hydrate would aid interpretation of our results.
3.4.3. The Hydrolysis Steps
Following equilibration, α-APN undergoes hydrolysis into alaninamide and then to alanine. These hydrolysis steps are irreversible and in a closed system, so associated isotope effects should be KIEs and the reactions are modeled as Rayleigh distillations. Of all sites in the final alanine, the two hydrolysis steps only directly involve the C-1 site: in H-1, the nitrile is converted into an amide and in H-2 the amide is converted into a carboxyl group. These steps are associated with significant bond reordering: during H-1, the C-1 site changes hybridization from sp to sp2, and during H-2, the C-1 changes from sp2 to sp3 then back to sp2. Theoretical and experimental work also suggests indirect involvement of the amine N during the two hydrolysis steps, by stabilizing the water lattice via hydrogen bonding; but this study predicts no involvement of the C-2 or C-3 sites (Yamabe et al., 2014). Given that the hydrolysis steps are irreversible and involve the formation of new bonds only at the C-1 site, we expect a primary normal KIE on the
C-1 site for each hydrolysis step and no isotope effects at the C-2 and C-3 sites. While we predict no primary isotope effects on the amine N, due to its potential in stabilizing intermediates we suspect an inverse secondary nitrogen isotope effect could occur, on the order of a few per mil.
The model calculations of d13C changes for the molecular average and the C-1 site of α- APN during reaction step H-1 are inconclusive, suggesting two competing fractionating processes (with one direction dominating early and the other late) and no consistent overall change; in contrast, our findings provide statistically strong and straightforward support for the expected primary KIEs during reaction step H-2 (Table 3.3). No
significant variations in δ15N and the average δ13C of the C-2 and C-3 sites are observed during either hydrolysis step, suggesting they are spectator atoms, as expected (Figures 3.4b, 3.4c, 3.6b, and S3.1).
During H-1, the molecular-average δ13C value has an overall positive trend with
increasing reaction progress for both the δ13C of product amide and the difference in δ13C between the reactant α-APN and product amide (Figures 3.4b and 3.5b). The C-1 site appears to have these trends, too, although they are obscured by the high error bars for the C-1 measurements (Figures 3.4b and 3.7a). The difference in δ13C between the reactant α-APN and product amide has a minimum value for the amide δ13C subtracted from α-APN δ13C at -15 ‰. However, the difference is resolvable above zero for some but not all syntheses at high (70 % - 80 %) reaction extents (Figure 3.5b), and the change in molecular average 13R for α-APN versus the fraction of residual α-APN (Figure S3.2) have trends inconsistent with a single Rayleigh distillation process.
The discrepancies from a Rayleigh fractionation trend and potential for opposing isotope effects acting during reaction step H-1 lead us to conclude that H-1 included one or more side reactions and that dominated the reaction progress variable in some reactions. The reaction progress variable here is the fraction of remaining α-APN and was measured by the change in the initial moles of alaninamide (calculated from directly measured final moles of alaninamide and alanine) relative to the estimated moles of initial α-APN.
Assuming that a side reaction consumed α-APN, less alaninamide would be made from α-APN, which would appear in our model as a higher fraction of residual α-APN or a lower amide yield in that reaction. If the side reaction had a normal KIE, it would also enrich the residual α-APN reservoir and make the product alaninamide appear to be more enriched, especially when it has high yields (and thus does not express its own KIE as strongly). We suggest that a real understanding of the elementary isotope effects associated with the H-1 reaction step of the Strecker chemistry will require direct measurements of the amounts and of the isotopic compositions of both α-APN and alaninamide, including site-specific effects and will also require a wider sampling of reaction progress. For these reasons, our considerations of broader implications of this work focus on the more confident conclusions we can reach regarding the equilibrium and H-2 steps and the overall net isotope effects of the Strecker chemistry.
Data from the H-2 step affirms our expectations regarding KIE’s associated with hydrolysis. Using the directly measured molecular average values for alaninamide and alanine, we find δ13C normal 5.2 ‰ KIE on the molecular average δ13C during H-2 (Figure 3.5c, Table 3.3) and a normal 15.4 ‰ KIE on the C-1 site (Figure 3.7b, Table 3.3) (in both cases alanine being 13C-depleted relative to alaninamide). The normal 15.4 ‰ KIE on the C-1 site is roughly thrice that of the molecular average and both the models for the C-1 site, and equal to the minimum value for the δ13C of amide’s C-1 site subtracted from that of α-APN. We note that in Rayleigh plots (isotope ratio vs. amount of residual reactant) for both the molecular average and C-1 site carbon isotope values conform to the Rayleigh law with high correlation coefficients, suggesting a single mechanism with a constant KIE (Figures 3.5c and 3.7b). These facts support the assertion the molecular average KIE associated with converting amide to alanine is solely due the isotope effect associated with hydrolysis at the C-1 site (Figures 3.5c and 3.7b). These data also agree with past amide hydrolysis studies, which argue that the normal KIE results from the hydroxyl addition to the amide and the ensuing conversion from sp2 to sp3 hybridization prior to ammonia leaving (Robins et al., 2015) (Figure 3.2b).