Size-Induced Inversion of Selectivity in the Acylation of 1,2-Diols.

Relative rates for the Lewis base-catalyzed acylation of aryl-substituted 1,2-diols with anhydrides differing in size have been determined by turnover-limited competition experiments and absolute kinetics measurements. Depending on the structure of the anhydride reagent, the secondary hydroxyl group of the 1,2-diol reacts faster than the primary one. This preference towards the secondary hydroxyl group is boosted in the second acylation step from the monoesters to the diester through size and additional steric effects. In absolute terms the first acylation step is found to be up to 35 times faster than the second one for the primary alcohols due to neighboring group effects.

Introduction

In many organic syntheses the protection of functional groups play a strategically important role. In molecular targets of pharmaceutical or biological importance hydroxyl groups count among the most relevant functional groups, and their selective protection and deprotection thus dominates synthetic strategies. Despite all efforts that have already been invested in the development of chemo‐ and regioselective protection strategies, it is still today difficult to selectively address a single hydroxyl group in a polyol‐system. The most common reagents for hydroxyl group protection are carboxylic acid derivatives such as acid chlorides, acid cyanides, or acid anhydrides, and several selective protection strategies of polyols with these reagents can thus be found in the literature. One strategy focuses on the in situ formation of transient cyclic intermediates such as (dialkyltin) acetals,[ , , ] borinates, boronates, or cyclic ortho‐esters. Subsequent ring opening reactions of these transient intermediates usually lead to preferred formation of the primary reaction products in 1,n‐diols (except for the Lewis acid‐mediated opening of orthoesters). It should be added that this selectivity is also observed in reactions involving the transient formation of trialkyltin alkoxides.[ , ] A second strategy employs transient complexation of polyol substrates through anionic hydrogen bond acceptors (carboxylates, cyanide[ , ]) as a tool for directing acylation reactions. In selected cases, Lewis base catalysts have been combined with carboxylate side chains in order to exploit this effect.[ , ] A third (and fully complementary) strategy employs Lewis base catalysts with or without the additional aid of auxiliary bases. When using highly reactive acylation reagents such as acid chlorides, already the combination with sterically hindered auxiliary bases such as N,N‐diisopropyl‐N‐ethyl amine (DIPEA, Huenig base) may be sufficient for the effective acylation of 1,2‐diols. For butane‐1,2‐diol A as an example Yamamoto et al. reported an exceedingly high selectivity for transformation of the primary hydroxyl group under these conditions (Scheme 1a). A strong preference for the reaction of primary hydroxyl groups has also been reported by Dong et al. in reactions of polyol substrates with anhydrides mediated by DIPEA, DBU, or DBN.

Scheme 1

Intra‐ and intermolecular competition in acylation reactions of primary and secondary hydroxyl groups.

That the high preference for the acylation of primary hydroxyl groups is not a privilege of nitrogen bases has recently been illustrated by Toda, Suga et al. with phosphonium ylide B as a zwitterionic nucleophilic catalyst for the regioselective acylation of 1,n‐diols such as C (Scheme 1b). For higher polyols (such as carbohydrates) it is also known that acylation selectivity depends on the hydroxyl group hydrogen bonding network.[ , ] Very recently we have explored the use of anhydride size‐effects for favoring the acylation of secondary alcohol D in the presence of primary alcohol E (Scheme 1c). For substrates carrying sizeable aromatic side chains Ar, it was found that secondary alcohols D can react up to two times faster than primary alcohols E with anhydrides derived from 2‐substituted 1‐naphthoic acids. In this study we therefore address whether the same selectivity switch can be found for 1,2‐diols of general structure 1 decorated by aromatic side chains Ar of variable size. As shown in Scheme 1(d) this reaction potentially involves two mono‐acylated and one doubly acylated diol derivative. Rather than providing product distributions at a single conversion point,[ , , ] we have used three different quantitative approaches to determine the rate constants intrinsic to this type of reaction scheme. The resulting rate data can then be used to simulate the experimentally measured turnover curves and thus predict selectivities at any given turnover point.

Results and Discussion

As illustrated in Scheme 2, the acylation of 1,2‐diols 1a,b with acid anhydride reagents 5a–g is characterized by effective rate constants k 1–k 6, that can be combined into three relative rate constants reflective of primary/secondary selectivities. Relative rate constant k rel,a is defined as the ratio of the effective rate constants for the acylation of the primary and secondary hydroxyl groups of diols 1a,b. The ratio of effective reaction rates for acylation of secondary monoester 3(a,b)(a–g) and primary monoester 2(a,b)(a–g) is given by the relative rate constant k rel,b and thus characterizes the second acylation process. The third relative rate constant k rel,c describes the ratio between the migration rates of secondary monoester 3(a,b)(a–g) and primary monoester 2(a,b)(a–g) and thus the transesterification dynamics between monoesters 2 and 3. 1,2‐Ethanediols carrying aryl substituents such as phenyl (a) and 2‐naphthyl (b) were employed as diol substrates.

Scheme 2

General mechanistic scheme for the acylation of 1,2‐diols 1a,b with acid anhydrides 5a–g catalyzed by TCAP (6) in the presence of Et3N (7) at +23 °C.

As reagents we selected the seven anhydrides 5a–g shown in Scheme 2, two with a preference for primary alcohols (5a–b) and five with a preference for secondary alcohols (5c–g) based on our earlier studies. All acylation reactions were catalyzed by TCAP (6) in the presence of Et3N (7) at a temperature of +23 °C in CDCl3. The conversion of all hydroxyl groups involved in the acylation of diols 1 to doubly acylated diols 4 can be expressed by equation I (Figure 1), and the mole fraction of a given species as given by equation II (using diol substrate 1a,b as an example). Three different approaches have been employed for measuring the relative reaction rates for the first acylation of 1,2‐diol 1b, and two of these also lend themselves for the second acylation of the monoesters 2b(a–g) and 3b(a–g) to the diester 4. The results obtained with these three different approaches will in the following be discussed for the acylation of 1b with anhydride c catalyzed by TCAP (Figure 2). In the first approach we adapted our procedures developed for intermolecular turnover‐limited competition experiments[ , ] to the 1,2‐diol substrate 1b where anhydride 5c was added as the limiting reagent such that a maximum of 2 %–95 % of all hydroxyl groups in 1b can react in the presence of 0.1 equiv. TCAP (6) and 1.5 equiv. Et3N (7) (for details see Supporting Information). Effective rate constants k 1–k 4 were then determined so that the best possible agreement was achieved between experimentally measured and theoretically predicted concentrations of all compounds shown in Scheme 2 using the overall conversion of hydroxyl groups as the reaction variable (Figure 2a). In addition, the product distribution obtained at the 30 % conversion point was selected in order to compare the relative ratios of 2b(a–g) versus 3b(a–g).

Figure 1

Conversion (I) of all OH‐groups present in compounds 1–4 and mole fraction (II) of 1a,b for the acylation of 1a,b with 5a–g catalyzed by TCAP (6) in the presence of Et3N (7).

Figure 2

Relative rate constant k rel,a between the primary and secondary hydroxyl groups of diol 1b and relative rate constant k rel,b between primary alcohol 3bc and secondary alcohol 2bc with anhydride reagent 5c catalyzed by TCAP (6) quantified by three types of measurements: a) turnover‐limited competition experiments, b) turnover‐limited competition measurements at the 30 % conversion point, and c) absolute kinetics measurements. The lines represent turnover curves predicted by numerical kinetics simulations while the crosses or dots represent experimentally measured values in the turnover‐limited competition experiments or absolute kinetics measurements. The effective rate constants k 1–k 4 are given in L mol−1  s−1.

In kinetic resolution experiments and other 1 : 1 competition experiments it is common practice to select a single conversion point for the determination of relative reaction rates, often at conversions of around 50 %.[ , , , ] However, in order to limit the influence of the second acylation step on the ratio of the mono‐acylation, the 30 % conversion point seems more appropriate. From these measurements we derived the relative rate constant k rel,a30% defined as the ratio between the concentrations of monoesters 2b(a–g) and 3b(a–g) (Figure 2b, for more details see Supporting Information). In a third approach, adapted again from our earlier studies, we follow the reaction of diol 1b with an excess of anhydride 5 (1.5 equiv.) in the presence of 0.1 equiv. TCAP (6) and 2.0 equiv. Et3N (7) via absolute kinetics measurements (for details see Supporting Information). Effective rate constants k 1–k 4 were again determined such that the best possible agreement was achieved between experimentally measured and theoretically predicted concentrations of all compounds shown in Scheme 2 with numerical kinetics simulations (Figure 2c).

In the following, we first focus on the results obtained in the absolute kinetics studies. Results for the reaction of diol 1b with anhydrides 5a–g are shown in Figure 3. Anhydrides 5a,c based on the benzoate parent structure favor acylation of the primary hydroxyl group in diol 1 b. For benzoic anhydride 5a as the reference system we find a selectivity of k rel,a=3.38. Introduction of substituents in the 1‐ and 6‐positions as in anhydride 5c reduces the selectivity to k rel,a=1.47.

Figure 3

Relative rate constant k rel,a between primary and secondary hydroxyl groups of diol 1b and relative rate constant k rel,b between primary alcohol 3b(a–g) and secondary alcohol 2b(a–g) with anhydride reagents 5a–g catalyzed by TCAP (6). The lines represent simulated curves from numerical kinetics simulations and the dots represent experimentally measured values in the absolute kinetics study. [a] The acylation of 1b was catalyzed by 1 mol % TCAP (6).

Selectivities then invert (that is, switch to preferred acylation of the secondary hydroxyl group in diol 1b) with anhydrides derived from 2‐substituted naphthoic acids. The selectivities show only a minor dependence on the length of the 2‐substituent with selectivity factors of k rel,a=0.96 (5d), 0.96 (5e), 0.89 (5f), and 0.95 (5g). In absolute terms, however, the observed preference for acylation of the secondary hydroxyl group is very moderate.

The second acylation step shown in Scheme 2 shows, in part, quite different primary/secondary selectivities as compared to the first one. It is only for benzoic anhydride (5a) that both steps show a comparable preference for acylation of the primary hydroxyl group (k rel,a=3.38 versus k rel,b=3.22, Figure 3). Already for anhydride 5c we observe a preference for reaction of the secondary hydroxyl group with k rel,b=0.64, which is quite similar to earlier findings in intermolecular turnover‐limited competition reactions. Even higher preferences for reaction of the secondary hydroxyl group are found for the 2‐substituted naphthoic anhydrides with k rel,b=0.45 (5d), k rel,b=0.25 (5e), k rel,b=0.20 (5f), and k rel,b=0.18 (5g). This final value implies that acylation of 2bg is 5.6 times faster than acylation of 3bg. This inversion of selectivity is quite contrary to what is commonly found for significantly smaller substrates and can best be understood as a consequence of size effects between the π‐systems present in substrate 1b, reagents 5, and catalyst 6 in the respective transition states for acyl group transfer.

How the selectivity values presented in Figure 3 vary as a function of the method of determination is illustrated in Figure 4. The methodological influence on k rel,a is generally found to be quite moderate, except for anhydride 5c, where a larger difference can be noted between the turnover‐limited competition experiments and the absolute kinetics measurements. Additional experiments were therefore undertaken for this latter system in order to determine, whether interconversion between mono‐esters 3bc and 2bc are responsible for this observation (for details see Supporting Information). It was indeed found that this process becomes notable at the extended reaction times reached in some of the turnover‐limited competition experiments, but less so in the absolute kinetics experiments (which are more reliable at this point). For all other anhydrides studied here, the interconversion between monoesters 2b and 3b appears to be too slow to contribute. The values of k rel,b show a somewhat larger dependence on methodological choice as shown in Figure 5. It is now only for the most reactive anhydride 5c that we find consistent results for turnover‐limited competition experiments and absolute kinetics studies, which require reaction times on the order of eight days to run to completion. For all other substituted systems reaction times are much longer, which may, in part, be responsible for the fact that selectivities are found to be much closer to k rel,b=1.0 as compared to the k rel,b values from absolute kinetics experiments.

Figure 4

Relative rate constant k rel,a between the primary and secondary hydroxyl groups of diol 1b in its reaction with anhydrides 5a‐g catalyzed by TCAP (6). [a] The acylation of 1 b was catalyzed by 1 mol % TCAP (6). [b] No absolute kinetics study was performed.

Figure 5

Relative rate constants k rel,b between primary alcohols 3b(a–g) and secondary alcohols 2b(a–g) with anhydrides 5a–g catalyzed by TCAP (6). [a] The acylation of 1 b was catalyzed by 1 mol % TCAP (6).

The hypothesis of stronger attractive interactions between the π‐systems present in substrate 1b, reagents 5, and catalyst 6 in the acylation transition state for the secondary hydroxyl group in 1b can be tested by repeating the absolute kinetics experiments with anhydrides 5c,e for the smaller diol 1a. The effective reaction rates k 1–k 4 for the acylation of 1,2‐diols 1a,b are shown in Figure 6. In reactions of 2,6‐disubstituted anhydride 5c (Figure 6a) we see an increase in the preference for acylation of the primary hydroxyl group in diol 1a (k rel,a=5.32) as compared to 1b (k rel,a=1.47). The origin of this selectivity change through increasing the side chain size from Ph to 2‐Np can be found in a rather moderate increase in k (red line in Figure 6a) from k 1=29.7×10−3 L mol−1 s−1 (1a) to k 1=36.6×10−3 L mol−1 s−1 (1b), and a much larger increase in k 2 from k 2=5.6×10−3 L mol−1 s−1 (1a) to k 2=24.7×10−3 L mol−1 s−1 (1b). In the second acylation, both Ph and 2‐Np side chains prefer the acylation of secondary alcohols, whereby the preference for 2bc is larger in 1b (k rel,a=0.64) as compared to 1a (k rel,a=0.90). Again with increasing side chain of the monoesters 2(a,b)c/3(a,b)c the effective rate constants k 3 and k 4 (purple and orange lines, Figure 6a) increase from k 3=3.6×10−3 L mol−1 s−1 (2ac) to k 3=14.6×10−3 L mol−1 s−1 (2bc) and k 4=3.3×10−3 L mol−1 s−1 (3ac) to k 4=9.3×10−3 L mol−1 s−1 (3bc). The simple conclusion from all these rate measurements is that ALL rate constants increase when increasing the substrate side chain from Ph to 2‐Np, but that the gains are larger for the secondary hydroxyl group (dashed lines) as compared to the primary one (solid line).

Figure 6

Averaged effective rate constants k 1–k 4 for diols 1a and 1b in their reaction with anhydrides a) 5c and b) 5e catalyzed by TCAP (6) as obtained from absolute kinetics experiments. Solid lines correspond to transformations of the primary hydroxyl groups (k 1, k 4), and dashed lines correspond to transformations of the secondary hydroxyl groups (k 2, k 3).[a] The effective rate constants k 3 and k 4 could not be determined reliably for 1a due to technical problems.

This is also found to be the case for the sterically more demanding anhydride 5e, which shows an intrinsically larger preference for the conversion of secondary hydroxyl groups (dotted lines, Figure 6b). Unfortunately, the second acylation step for substrate 1a is so slow that conversions beyond 60 % (and thus the rate constants k 3 and k 4) could not be measured reliably. Despite this latter complication, the trends shown in Figure 6 for two anhydrides are largely similar to earlier results obtained in intermolecular competition experiments with the same anhydrides and are thus in full support of the hypothesis of size effects between the substituents present in substrates 1, reagents 5, and catalyst 6 in the acylation transition states.

The influence of the reaction temperature on the first acylation of diols 1a and 1b with anhydride 5c was studied in the range from −20 °C to +40 °C using the 30 % conversion point method (Figure 7). For diol 1a the relative rate constant k rel,a30% varies from k rel,a30%=2.2 (−20 °C) to k rel,a30%=7.0 (+40 °C). Analyzing these results with aid of the Eyring equation shows a difference in activation enthalpy and entropy of −12.1 kJ mol−1 and −54.1 J K−1 mol−1. Larger temperature effects were subsequently found for diol 1b with differences in activation enthalpy and entropy of −22.9 kJ mol−1 and −81.3 J K−1 mol−1 (see Supporting Information). From the difference of around ∼30 J K−1 mol−1 in the activation entropy values obtained for 1a and 1b it appears that reaction of the naphthyl‐substituted system involves a more highly ordered transition state. That the activation enthalpy difference for reaction of the primary and secondary hydroxyl groups is larger for 1b (ΔH ≠=−22.9 kJ mol−1) than for 1a (ΔH ≠=−12.1 kJ mol−1) is also in line with the presence of attractive size‐dependent interactions between the diol substrates, anhydride 5c, and catalyst 6 in the acylation transition states.

Figure 7

Eyring plot of ln k rel,a for the reaction of diols 1a and 1b with anhydride 5c mediated by TCAP (6).

The reaction of diol 1b with anhydride 5c was subsequently also studied in different solvents. A general trend of decreasing selectivity values k rel,a with increasing solvent polarity could be identified, but the range of solvents was, unfortunately, too limited due to solubility issues to allow for a systematic analysis (see Supporting Information). Similar reaction times for full conversion were also obtained when using different concentrations of the Et3N auxiliary base (from 0.02 to 0.08 M) in the acylation of 1b with anhydride 5c. However, based on changes in polarity caused through the different concentration of Et3N, the relative selectivities change from k rel,a=1.52 (0.02 M) to k rel,a=1.12 (0.08 M) and from k rel,b=0.62 (0.02 M) to k rel,b=0.69 (0.08 M). The background reactivity was studied for the acylation 1b with anhydride 5c and an excess of auxiliary base Et3N, whereby only traces of 2bc are detected.

The large selectivity differences between the first and the second acylation steps shown in Figure 3 can potentially arise from repulsive steric effects introduced through the first‐formed ester unit or from activating neighboring group effects of one hydroxyl group on the other. In order to differentiate these effects, the 1‐ and 2‐substituted propanol substrates 8b and 9b shown in Figure 8 were reacted with the same anhydrides 5 under the same reaction conditions as before. Formally, these two substrates replace one of the hydroxyl groups present in diol 1b by an unreactive methyl substituent unable to form intramolecular hydrogen bonds. In these experiments the relative rate constant k rel,prop is defined as the ratio between the acylation rate of primary alcohols 9b over the secondary alcohols 8b. A 1 : 1 mixture of alcohols 8b and 9b were therefore reacted with anhydrides 5 as the limiting reagents up to a final conversion of 20 %–70 %. The esterification reactions were catalyzed by TCAP (6) in the presence of Et3N (7) at a reaction temperature of +23 °C in CDCl3 (for details see Supporting Information). The experimentally measured turnover data were subsequently converted to relative rate constants k rel,prop using the same approach as in earlier studies.[ , ]

Figure 8

Relative rate constant k rel,prop between primary alcohols 9b versus secondary alcohols 8b in their reaction with anhydride reagents 5a–f and k rel,a of the acylation of 1b with 5a–f catalyzed by TCAP (6).[a] The acylation of 1 b was catalyzed by 1 mol % TCAP (6).

For the 2‐naphthyl substituted alcohol pair 8b/9b we note a general decrease in the selectivity values (towards a reduced preference for primary alcohol 9b) in the same range as could be seen before for the first acylation of diol substrate 1b in Figure 4. This can already be seen in the selectivity for benzoic anhydride 5a with k rel,prop=2.76, and becomes more prominent already in the 2,6‐disubstituted benzoic anhydride 5c with k rel,prop=0.87 and in the 2‐substitued naphthyl anhydrides such as 5f with k rel,prop=0.69. Thus, the propanol system shows the same trends found earlier in acylation reactions of ethanol systems. Comparing these results with the first acylation of diol substrate 1b (k rel,a), we see a small impact through activating neighboring group effects of one hydroxyl group on the other. This is illustrated trough the slightly reduced preference for the secondary OH group, for example 5c with k rel,prop=0.87 versus k rel,a=1.47 or 5f with k rel,prop=0.69 versus k rel,a=0.89. An alternative strategy for quantifying the neighboring group effects in acylation reactions of diol 1b exists in comparing the effective rate constants for the same hydroxyl group in the first and the second acylation steps shown in Scheme 2. A relative rate constant k prim can, for example, be defined as the ratio of effective rate constants k 1 (for acylation of the primary hydroxyl group in 1b) and k 4 (for acylation of the primary hydroxyl group in monoesters 3b(a–g)). Similarly, k sec can be defined as the ratio of effective rate constants k 2 and k 3 for reaction of the secondary hydroxyl groups in diol 1b and in monoesters 2b(a–g)). The results obtained for selected anhydrides are shown in Figure 9. For the unhindered anhydride 5a we see that k prim and k sec assume values larger than 10, which simply indicates that both hydroxyl groups in diol 1b are more reactive than the same hydroxyl groups in the intermediate monoesters 2ba or 3ba. For the 2,6‐disubstituted (and electronically most activated) benzoic anhydride 5c both values are reduced to k prim=4.0 and k sec=1.8. In particular this latter value indicates that the transition state for acylation of the secondary hydroxyl group in 1b can hardly benefit from its primary hydroxyl group neighbor in this case. For the series of 2‐substituted 1‐naphthoic anhydrides 5d–g, we find increasingly larger values of k prim=20.1 (5d), 25.0 (5e), 29.9 (5f), and 35.5 (5g) for increasingly long side chains.[ , ] In contrast, the values for k sec are much smaller and show an inverse trend with side chain size with k sec=9.2 (5d), 6.5 (5e), 6.8 (5f), and 6.8 (5g). These results again highlight that for all anhydrides studied here (except for 5c) the first acylation is significantly faster than the second, the effects being particularly large for the primary hydroxyl group in diol 1b.

Figure 9

Relative rate constant k rel,prim versus k rel,sec of the acylation of 1b with 5a–g catalyzed by TCAP (6). [a] The acylation of 1b was catalyzed by 1 mol % TCAP (6).

The acylation of diol 1b with anhydrides 5c–g is much slower in absolute terms as compared to that with benzoic anhydride 5a (see Supporting Information for more details). Whether this is a thermochemical or kinetic issue was tested by quantum mechanical calculation of the Gibbs free energy of reaction ΔG 298 for the formation of esters 2b(c,f) 3b(c,f), and 4b(c,f) at the SMD(CHCl3)/DLPNO‐CCSD(T)/def2‐TZVPP//SMD(CHCl3)/B3LYP‐D3/6‐31+G(d) level of theory. As expected from a previous study the first acylation of 1b to either 2b(c,f) or 3b(c,f) is similarly exothermic with reaction free energies of ΔG 298=−61.5 kJ mol−1 (2bc) versus −58.9 kJ mol−1 (3bc) and ΔG 298=−44.6 kJ mol−1 (2bf) versus −40.1 kJ mol−1 (3bf). The overall diacylation of 1b to 4(c,f) is almost exactly twice as exergonic with ΔG 298=−122.2 kJ mol−1 (4bc) and ΔG 298=−76.9 kJ mol−1 (4bf), which indicates that the acylation of 1b is under kinetic control. With respect to Marcus theory this implies that the intrinsic barrier for the reaction of 2b(c,f)/3b(c,f) with anhydride 5c,f mediated by 6 is much higher than the intrinsic barrier for the reaction of diol 1b under the same conditions. The results in Figures 3 and 6 for 1,2‐diol substrates 1a/1b can be rationalized with the same “triple” sandwich acylation transition state between catalyst 6 (red), the substrate aryl substituent (green), and the ortho‐substituent of the acid anhydrides 5c–g (blue) as shown in our previous study (Scheme 3). This structure provides a natural basis for the experimentally observed rate differences between diol 1a (phenyl side chain) and 1b (2‐naphthyl side chain) under the condition that side chain/catalyst interactions are attractive in nature through a combination of electrostatic and London dispersion interactions. The strong decrease in effective reaction rates from benzoic anhydride 5a to the bulky anhydrides such as 5f (by a factor of ca. 1000 for primary hydroxyl groups and <370 for secondary hydroxyl groups excluding 5c, see Supporting Information) point to strong differential steric hinderances in the acylation. These repulsive size effects are enhanced in the second acylation step and appear to be more substantial for transformations of primary as compared to secondary hydroxyl groups. Based on the transition states calculated earlier for the respective mono‐alcohol systems and the X‐ray solid state structure of 2bf, the TCAP‐mediated acylation of monoesters 2bf and 3bf to diester 4bf was explored at the SMD(CHCl3)/B3LYP−D3/6‐31+G(d) level of theory (Scheme 3a).[ , , , , , ] The Gibbs free energy of activation for the acylation of secondary alcohol 2bf is located +44.6 kJ mol−1 above the separated reactants 2bf, 5f and 6 and thus 7.6 kJ mol−1 lower than the transition state for the acylation of primary alcohol 3bf (ΔG 298 ≠=+52.2 kJ mol−1). The transition state for the secondary alcohol displays the “triple” sandwich motif found earlier in reactions of the respective mono‐alcohols, where the butyloxy side chain of anhydride 5f is located below and the naphthyl substituent of alcohol substrate 2bf is located on top of the TCAP‐derived pyridinium ring system. This latter structural characteristic is missing in the transition state for the primary alcohol 3bf, whose naphthyl substituent is oriented towards that of the carboxylate counter ion. That non‐covalent interactions (NCIs) exist between several of the three butyloxy side chains, the four naphthyl substituents, and the TCAP pyridinium ring system is readily seen in the NCI plots for these transition states (see Supporting Information for full details). Identification of a single interaction responsible for the barrier difference seen in Scheme 3 is, however, not possible.

Scheme 3

Computed transition state structures for the second acylation of monoesters 2bf and 3bf with anhydride 5f. Distances are given in pm.

Conclusion

We demonstrate here three possible ways to measure and quantify the acylation of different hydroxyl groups in a diol system. Best results for the first and second acylation steps of 1‐substituted ethylene‐1,2‐diols 1 were obtained using direct kinetics measurements. These were complemented by turnover‐limited competition experiments and single conversion point measurements for defining relative rate constants for the first acylation in cases, where internal acyl migration between monoesters 2 and 3 is minimal. Reactions of diol 1b carrying a large 2‐naphthyl side chain with sterically hindered anhydrides display a selectivity inversion favoring acylation of the secondary hydroxyl group in 1b. Variation of the size of the diol π‐system leads to systematic changes in reaction rates as well as selectivities best rationalized by assuming attractive interactions between the substrate side chains and the catalyst‐derived pyridinium core. Differential size effects strongly decrease the absolute reaction rates from benzoic anhydride 5a to bulky anhydrides 5c–5g and the acylation of the monoesters relative to those of the diols. Surprisingly, the effective acylation rates of primary alcohols respond more strongly to steric hinderance than those of secondary alcohols. The full exploitation of these effects may require catalysts optimally tailored for these structurally demanding reagents.

Experimental Section

Competition experiments

Three different CDCl3 stock solutions consisting of 0.02 m 1,2‐diol 1a,b (A), 0.04 m acid anhydride 5a–g (B), and a combination of 0.06 m Et3N and 0.004 m TCAP (C) are prepared under nitrogen. Afterwards stock solution B is diluted such that new solutions with 23 different concentrations are obtained. The concentration of these solutions has been fixed at 2, 5, 7, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90 and 95 % of the initial stock solution B, whereby the solutions of the 30 % concentration point are prepared three times. Subsequently 0.4 mL of stock solution A, 0.4 mL of stock solution C and 0.4 mL of diluted stock solution B1 are transferred under nitrogen to a GC vial equipped with a magnetic stir bar by use of a Hamilton syringe. The GC vial is then capped under nitrogen and placed in the GC vial holder with stirring for 3–4 weeks. The reaction is subsequently analyzed by 1H NMR recorded by Bruker Avance III 800 machine.

Kinetic studies

Three different CDCl3 stock solutions consisting of 0.02 m 1,2‐diol 1a,b (A), 0.06 m acid anhydride 5a–g (B), and a combination of 0.08 m Et3N and 0.004 m TCAP (C) are prepared under nitrogen. The reaction is analyzed by 1H NMR recorded on a Bruker Avance III 400 machine. NMR tubes are dried under vacuum using a special home‐made apparatus and flushed with nitrogen. 0.2 mL of stock solution A, 0.2 mL of stock solution C and 0.2 mL of stock solution B are transferred under nitrogen to an NMR tube by use of a Hamilton syringe. After closing the NMR tube, the reaction mixture is shaken and introduced into the NMR machine.

Computational details

All geometry optimizations and vibrational frequency calculations have been performed using the B3LYP‐D3 hybrid functional[ , , ] in combination with the 6‐31+G(d) basis set. Solvent effects for chloroform have been taken into account with the SMD continuum solvation model. This combination has recently been found to perform well for Lewis base‐catalyzed reactions.[ , ] Thermochemical corrections to 298.15 K have been calculated for all stationary points from unscaled vibrational frequencies obtained at this same level. Solvation energies have been obtained as the difference between the energies computed at B3LYP‐D3/6‐31+G(d) in solution and in gas phase. For the calculations of Gibbs free energies, the thermochemical corrections of optimized structures have been combined with single point energies calculated at the DLPNO‐CCSD(T)/def2‐TZVPP//B3LYP‐D3/6‐31+G(d) level.[ , , , ] Solvation energies have been added to the energy computed at DLPNO‐CCSD(T)/def2‐TZVPP//SMD(CHCl3)/B3LYP‐D3/6‐31+G(d) level to yield free energies G 298 at 298.15 K. Free energies in solution have been corrected to a reference state of 1 mol L−1 at 298.15 K through addition of RTln(24.46)=+7.925 kJ mol−1 to the free energies. All calculations have been performed with Gaussian 09 and ORCA version 4.0. Conformation search was performed with Maestro.

Deposition Number(s) 2085895 (for contain(s) the supplementary crystallographic data for this paper. These data are provided free of charge by the joint Cambridge Crystallographic Data Centre and Fachinformationszentrum Karlsruhe Access Structures service.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.