Direct Deoxydehydration of Cyclic trans‐Diol Substrates: An Experimental and Computational Study of the Reaction Mechanism of Vanadium(V)‐based Catalysis

Abstract

The deoxydehydration of carbohydrates represents a key target to leverage renewable biomass resources chemically. Using a vanadium(V)‐based catalyst, it was possible to directly deoxydehydrate cyclic trans‐diol substrates. Accompanying mechanistic characterisation of this process by density functional calculations pointed to an energetically tractable route for deoxydehydration of cyclic trans‐diol substrates involving stepwise cleavage of the diol C−O bonds via the triplet state; experimentally, this was supported by light dependence of the reaction. Calculations also indicated that cyclic cis‐diols and a linear diol substrate could additionally proceed by a concerted singlet DODH mechanism. This work potentially opens a new and cost‐effective way to efficiently convert carbohydrates of trans‐diol stereochemistry into alkenes.

Removing vicinal hydroxyls with a twist: Experimental and computational studies indicate that cyclic trans‐diol compounds can be deoxydehydrated to alkenes. The problem presented by the trans stereochemistry of the diol groups is overcome using a vanadium(V)‐based catalyst. The computed mechanism suggests stepwise cleavage of the diol C−O bonds, allowing a key twist in substrate‐catalyst orientation after the first C−O bond cleavage to yield alkene product via a metallacycle intermediate.

Introduction

With the rapid depletion of fossil fuel resources, there is an urgency to leverage biomass for high‐value chemicals as a replacement for petrochemical‐based products, from fuels, dispersants and detergents, to rheology modifiers and flocculants. ^[1] Carbohydrates such as cellulose and starch constitute the largest source of biomass substrate. However, the synthesis and selective modification of saccharides is challenging. Enzymatic modification approaches show promise but face significant challenges in terms of titre yield, process accessibility and eventual industry applicability, often requiring further chemical or physical processing. ^[2]

Carbohydrate conversion using selective non‐enzymatic catalysis has also received considerable attention. Characterised by the large number of hydroxyl groups in carbohydrate systems, most modifications are restricted to acetylation and esterification, to crosslink or incorporate additional functionalities. To improve the value of carbohydrates as chemical precursors, it is desirable to reduce their oxygen content; unfortunately, there are relatively few reactions known to be successful for reducing the oxygen content of biomass‐derived polyols. ^[3] The most widely employed deoxygenation reactions are acid‐catalysed dehydration, high‐temperature pyrolysis and hydrogenolysis. However, hydrodeoxygenation and deoxydehydration (DODH) reactions have become increasingly used, producing alkanes and alkenes, respectively. ^[4] The formation of alkenes as opposed to alkanes is advantageous to the chemical industry, as the olefin functional group can be further manipulated via addition reactions and oligomerisation to yield polymeric materials. ^[5]

DODH reactions convert diols to alkenes (Figure 1) and generally require a metal catalyst, a reductant and a solvent. ^[6] Metal catalysts that have been successfully explored are rhenium, ^[7] vanadium ^[8] and molybdenum. ^[9] Rhenium has provided the most efficient and highest yielding catalysts but is the most expensive of the three metals. The substrate scope of the reaction includes simple aliphatic diols, cyclic diols and more recently, sugar alcohols containing many vicinal diol moieties. ^[7f] In general, it is observed experimentally for rhenium‐catalysed DODH reactions that (i) terminal diols undergo DODH faster than internal diols; ^[10] and (ii) cyclic cis‐diols are able to undergo DODH reactions whereas cyclic trans‐diols cannot.[ ^7c , ^7k , ¹¹ ]

Figure 1.

Rhenium‐catalysed (left) and vanadium‐catalysed (right) DODH reactions of ethylene glycol. Steps 1, 2 and 3 refer to condensation of the diol, reduction and olefin‐extrusion, respectively. Catalysts are shown in red.

As an example of (ii), Abu‐Omar and co‐workers reported a cyclohexene yield of 60 % when cis‐1,2‐cyclohexanediol underwent methyltrioxorhenium (MTO)‐catalysed DODH using hydrogen as the reductant. ^[7b] However, the trans isomer did not undergo DODH. Shiramizu and Toste reported the MTO‐catalysed DODH reaction of cyclic sugars, some of which contained trans‐diol moieties. ^[7f] The findings supported the preference for the cis stereochemistry; however, the trans‐diol sugars did react, albeit providing low alkene yields. Speculation as to the underlying mechanism of this cyclic trans‐diol DODH included pyranose ring opening to provide an acyclic substrate capable of the syn stereochemistry for olefin extrusion. ^[7f] While cis‐DODH is suited to sugars such as d‐mannose, this is unfortunate for trans‐diol substrates including the key biomass residue, d‐glucose.

Insight into the inability of MTO to process trans‐cyclic diol substrates was provided by quantum chemical calculation of the DODH reaction mechanism by Wang and co‐workers. ^[12] This density functional theory (DFT) study compared DODH of trans‐ and cis‐1,4‐anhydroerythritol (denoted here as S1 and S1’ respectively, Figure S1a) using the M06/SDD/6‐311++G** level of theory and a continuum model for the 3‐octanol solvent. Using this model, the key olefin extrusion barrier energy for the trans‐diol S1 substrate was found to be 50 kcal mol⁻¹ higher than the 19.9 kcal mol⁻¹ barrier for the cis isomer S1’. The substantial difference in barrier height was attributed to the stereospecific requirements of the retrocycloaddition reaction: the concerted olefin extrusion transition state was formed with the requirement for a syn approach of the diol's OH groups to the Re centre. Here, due to its diol stereochemistry, the cyclic trans‐diol could not adopt the planar arrangement in the concerted breakage of both Re−O bonds between the catalyst and the diol, required for a low energy barrier to this extrusion step (Figure S1b).

As an alternative to rhenium‐based DODH, recent work has demonstrated the ability of homogeneous vanadium complexes to catalyse DODH of vicinal diols in comparable yields: ^[8] an example is the DODH of 1,2‐propanediol to form propene, using [n‐Bu₄N]‐(dipic)VO₂ catalyst 1, triphenylphosphine as the reductant and benzene as the solvent. For this reaction, computational work by Jiang et al. ^[13] and Poutas et al. ^[14] reported novel spin‐crossover mechanisms. The spin‐crossover, involving vanadium's triplet state, provided a lower energy path than a previously proposed one‐step olefin extrusion mechanism. ^[15] In the proposed crossover mechanisms, the key olefin‐extrusion step occurs stepwise rather than via concerted cleavage of the diol C−O bonds required by MTO‐catalysed DODH (Figures 1 and 2).

Figure 2.

Comparison of (left) rhenium‐catalysed DODH key olefin‐extrusion transition state, where C−O cleavage is concerted; and (right) vanadium‐catalysed DODH key olefin‐extrusion transition state, where C−O cleavage is stepwise. Numbering of diol C−O atoms indicated.

In this study, we explore the possibility that the preferred stepwise catalytic route exploited by vanadium catalysts could allow sufficient flexibility for DODH of a cyclic diol of trans stereochemistry (Figure 2). Vanadium DODH has been shown to occur for a cyclic substrate of cis‐stereochemistry: specifically, it was reported that [n‐Bu₄N]‐(dipic)VO₂ catalysed the DODH of cis‐1,2‐cyclohexanediol, producing a cyclohexene yield of 15 %. ^[8a] However, to our knowledge, vanadium‐catalysed DODH of cyclic trans‐diol substrates remains unexplored. Indeed, prior mechanistic studies on the vanadium‐catalysed DODH of vicinal diols have focused on model linear substrates with no exploration of the cis or trans preference of cyclic diol substrates. Therefore, in this work we investigate by experiment and computation the ability of vanadium(V) species to catalyse the DODH reaction of cyclic trans‐diols and compare with their cis isomers.

Results and Discussion

Cyclic trans‐diol substrates can undergo direct DODH

To confirm our hypothesis, first we investigate whether [n‐Bu₄N]‐(dipic)VO₂ 1 is able to catalyse the DODH of trans‐1,2‐cyclopentanediol (S2) and compare with its cis isomer (S2’) (Table 1). 10 mol% of 1 was used, in chlorobenzene, with triphenylphosphine as the reductant. We find that S2 gives a cyclopentene yield of 20 %, and that this is 2 % higher than that of its cis‐isomer S2’ (Table 1). Similarly, we see comparable yields for cis and trans isomers for a substrate with a six‐membered ring: trans‐1,2‐cyclohexanediol S3 gives a cyclohexene yield of 14 %, in comparison to 16 % for its cis‐isomer S3’ (Table 1). As a further comparison, we carried out DODH of 1‐phenyl‐1,2‐ethanediol S4 (styrenediol, Table 1), a linear substrate that is widely used as substrate for DODH reactions. S4 is able to undergo DODH efficiently to give the olefin product, styrene, due to the driving force of generating a product with extended conjugation. Here, we report a styrene yield from S4 of 36 % (Table 1). Given the lack of conversion for cyclic trans‐diol substrates via rhenium‐catalysed DODH, the reactivity of trans‐diol substrates S2 and S3 observed here using a vanadium‐based catalyst is notable, as is the approximate parity in yield with their cis‐isomers.

Table 1.

Oxo‐vanadium catalysed DODH of substrates in normal light and dark conditions.^[a]

Substrate	Product	Yield [%]		Fold Δ[d]
		light[b]	dark[c]
		20	6	3.3
S2	P2
		18	4	4.5
S2’	P2
		14	6	2.3
S3	P3
		16	3	5.3
S3’	P3
		36	14	2.6
S4 [e]	P4

[a] Reaction conditions: 0.1 mmol substrate, 0.15 mmol PPh₃, and 10 mol% of the catalyst. Reaction time is 72 h unless otherwise stated. Yields are determined by ¹H NMR spectroscopy using mesitylene as the internal standard and based on an average of three repeats. [b] Olefin yield when the reaction was carried out in normal light. [c] Olefin yield in the dark. [d] Fold difference (fold Δ) is calculated as (yield in light)/(yield in dark).[e] Reaction time is 24 h.

As mentioned above, preceding computational work ^[13] suggested the involvement of a spin‐crossover mechanism in vanadium‐catalysed DODH. Specifically, Jiang et al. predict that the [n‐Bu₄N]‐(dipic)VO₂‐catalysed DODH reaction proceeds via intersystem crossings (ISC) involving singlet and triplet states. ^[13] A DFT study by Poutas et al. described a variant of this mechanism, also utilising vanadium's triplet state. ^[14] Both of these stepwise mechanisms have a lower computed energy pathway than a previously proposed concerted route. ^[15]

Therefore, given the potential involvement of ISC events, we carried out the reaction in the presence and absence of visible light, in order to identify any role for photocatalysis in this reaction. As borosilicate glassware was used for this reaction, the effects of UV radiation were precluded. When DODH was performed for the 1,2‐cyclopentanediols S2 and S2’, 1,2‐cyclohexanediols S3 and S3’ and styrenediol S4 under these conditions, we do indeed find that light has a significant effect on this reaction, with enhancements in yield ranging from 2.3‐fold to 5.3‐fold (Table 1). We observe that the disparity in yield was most pronounced for the cis‐cyclic diols S2’ and S3’, where the presence of light increases the yield 4.5‐ and 5.3‐fold, respectively. Although ISC events can be thermally activated in some instances, we observe here that the presence of light results in enhanced yields of olefin products. Thus, we propose that the role of light in this reaction is providing additional energy for this ISC event to take place. However, the absence of light does not completely halt the reaction; this may indicate the presence of alternative routes to one involving ISC, such that when light is excluded, the reaction proceeds via less favourable singlet pathways. This may also hint that thermal activation is suitable for this ISC event to take place, but at a significantly lower efficiency.

Mechanistic insights from density functional calculations

To further investigate the mechanistic basis of our observed DODH reactions and their light dependence, using DFT we compute the free energy surfaces of vanadium‐catalysed DODH of cyclic trans‐ and cis‐diols, and compare with the energetics for the linear substrate, 1‐phenyl‐1,2‐ethanediol S4.

The previously proposed ^[13] spin‐crossover mechanism of vanadium‐catalysed DODH of vicinal diols to alkenes involves six steps: (a) condensation of the diol onto the vanadium catalyst; (b) reduction of the vanadium‐diolate from its (V) to (III) state; (c) crossover from the singlet to triplet state of vanadium; (d) scission of one of the diol C−O bonds, forming a carbon‐centred radical on the resulting vanadium(IV) species; (e) a second spin‐crossover to form a vanadium metallacycle in the singlet state and finally, (f) concerted C−O and C−V bond cleavage to form the alkene product and regenerate the catalyst in its starting singlet state. Thus, the olefin extrusion stage comprises steps (c) to (f).

Five‐membered ring substrates

We first consider the spin‐crossover DODH mechanism for the five‐membered carbocyclic trans‐diol substrate S2 (Figure 3). In step (a), there is condensation of S2 with the vanadium catalyst 1 (Figure 3); during this, there are two consecutive proton transfers from substrate to catalyst to yield intermediate 5s; here, two V−O bonds are formed between the substrate diol and the vanadium catalyst. The highest barrier found in step (a) is the second proton transfer and formation of the second V−O bond, which occurs via TS2 s, with a calculated free energy of 30.8 kcal mol⁻¹ relative to the starting catalyst and substrate diol. Step (b) involves the reduction of the vanadium‐diolate intermediate 6 s by triphenylphosphine (Figure 3). Water is first expelled from complex 5s, exergonically forming complex 6 s. The phosphorus atom of the triphenylphosphine reagent attacks the oxo ligand of 1 and reduces the vanadium complex from a V^V to V^III oxidation state, giving the catalytically active form 7 s via TS3 s. This barrier is calculated as 40.3 kcal mol⁻¹ at the M06‐L/SDD/6‐311+G**/SMD level of theory. As vanadium(III) complexes possess triplet ground states, intermediate 7 s converts to intermediate 7 t via a minimum‐energy crossover point, MECP1, in a spin‐crossover process (Figure 4); this is in accord with preceding quantum chemical studies for DODH of linear substrates.[ ¹³ , ¹⁴ ]

Figure 3.

Proposed reaction mechanism for DODH of trans‐1,2‐cyclopentanediol (S2), catalysed by 1, utilising PPh₃ as reductant. Free energies [kcal mol⁻¹] computed at the M06‐L/SDD/6‐311+G**/SMD level of theory. Steps (a) and (b) denote condensation of the substrate onto catalyst, and reduction of the metal‐diolate species, respectively. For clarity, dipicolinate ligand (inset) is defined as L. Also note that vanadium dipicolinate complexes have a net charge of −1.

Figure 4.

Proposed reaction mechanism for DODH of S2, showing steps (c–f). Free energies [kcal mol⁻¹] computed at the M06‐L/SDD/6‐311+G**/SMD level of theory.

Potentially the order of condensation [step (a)] and reduction [step (b)] could be reversed as pointed out by Poutas et al. ^[14] Considering the possibility of reduction preceding condensation, here we find that, using catalyst 1 and PPh₃ reductant, reversal of (a) and (b) is disfavoured by 23.0 kcal mol⁻¹ at the M06‐L/SDD/6‐311+G**/SMD level of theory, effectively excluding this sequence of reaction (see Figure S2 where the full DODH pathways are compared for substrate 1,2‐hexanediol). The initial loading of the substrate S2, via TS1 s, is 14.8 kcal mol⁻¹ ; this barrier is lower in energy than the barrier associated with immediate reduction of 1, via TS‐1 (Figure S1b) which is 37.8 kcal mol⁻¹. Additionally, comparison of the two reduction transition states suggests that the reduction transition state in which the reduction precedes condensation is 6.7 kcal/mol higher in energy than TS3s in which the condensation precedes reduction. Therefore, we predict that the condensation of the substrate diol is likely to occur before the reduction of 1.

The final phase of the mechanism is olefin extrusion. This has been the problematic stage for DODH of cyclic trans‐diol substrates in previous computational studies using rhenium catalysts, where only one‐step concerted extrusion mechanisms were considered. ^[12] These mechanisms invoke an E_i syn elimination, which requires a co‐planar arrangement of catalyst and substrate atoms that the trans‐diol is unable to readily adopt (Figure S1). Here, however, we consider the possibility of stepwise cleavage of the diol's C−O bonds, applied to the cyclic trans‐diol intermediate in its triplet state, 7 t (Figure 4). Firstly, step (d) involves cleavage of one diol C−O bond, forming radical 8 t. This is accompanied by the shortening of the V−O bond distance from 1.91 Å in intermediate 7 t to 1.62 Å in intermediate 8 t, suggesting the formation of a V=O double bond, as depicted in intermediate 8 t (Figure 4). This step passes through TS4 t (Figure 5a) and is rate‐determining in the overall mechanism; here we find a substantial energetic barrier of 41.5 kcal mol⁻¹ (Figure 4). However, this activation energy is considerably lower than that computed for the rhenium‐catalysed DODH of S2 (71.0 kcal mol⁻¹), with respect to MTO and S2, calculated at M06/SDD/6‐311++G**/SMD according to the level of theory employed by Wang and co‐workers. ^[12] Although this estimate may vary in magnitude according to density functional and basis set, it is nevertheless expected to remain highly disfavoured due to the trans‐diol stereochemistry.

Figure 5.

Optimised structures of species (a) TS4 t, (b) 9 t and (c) 9 s for DODH of S2, computed at the M06‐L/SDD/6‐311+G**/SMD level of theory. Bond distances in Å. Dihedral angles in ° for (a) O₁−C₁−C₂−O₂ and (b,c) V−O₁−C₁−C₂.

From TS4 t, species 8 t is formed in the vanadium(IV) oxidation state; this oxidation state is implied by the spin density concentrated on the substrate carbon atom C₂ (1.07) and on the vanadium centre (1.14). As such, carbon C₂ is trigonal planar (for atom labelling, see Figure 2). From complex 8 t, triphenylphosphine oxide is eliminated to generate 9 t (Figure 4). Next, step (e) involves a second spin‐crossover to form 9 s in the singlet state (Figure 4), where the intermediate 9 t collapses to form metallacycle 9 s. The formation of the metallacycle is enabled by a rotation around the V−O₁−C₁−C₂ dihedral angle, with a value of −89.9° in 9 t (Figure 5b) to −21.6° in 9 s (Figure 5c). The rotation enables alkylation of the V centre by carbon C₂ of the trans‐diol substrate, leading to the syn stereochemistry of the aliphatic C₁ and C₂ protons in 9 s (Figure 5c) required for concerted elimination in the final step (f): here, the alkylvanadium metallacycle proceeds through a [2+2] retrocycloaddition to give product P2 (Figure 4). ^[14] This concerted cleavage of C−O and C−V bonds in TS5 s also regenerates the starting catalyst, 1. The 9 t–9 s spin‐crossover reaction ensures the starting catalyst is in its singlet form.

Thus, we have described here a stepwise mechanism of DODH catalysed by 1 which can process cyclic trans‐diol S2 leading to alkene product. We note that during our analysis of DODH of S2, additional pathways for extrusion of the olefin were identified. However, pathways that did not involve the triplet state of vanadium had computed energetic barrier heights of more than 50 kcal mol⁻¹ (Figure S3). Also, no stationary points associated with a mechanism of concerted C−O cleavage were obtained.

In order to further probe the role of light on the reaction mechanism, we used time‐dependent DFT to compute the excited states and associated absorption spectra of the first intermediate involved in intersystem crossing, i.e. structure 7 s, for all substrate diols studied experimentally. As 1 is not photoactive, that is, does not absorb in the visible region, it was thought that this key intermediate 7 s may have an absorption with a wavelength in the visible region. For all substrates, at the CAM−B3LYP/SDD/6‐311+G**/SMD level of theory, we find that their corresponding 7 s structures exhibit strong predicted absorption in the cyan‐green light region, with wavelengths ranging from 502–518 nm (Figure S4). Absorption of green light would result in a purple solution, which corresponds to the colour observed experimentally in this study (Figure S5).

Interestingly, the strongest absorption in the visible region for the 7 s singlet state appears to correspond to an excitation mainly composed of a transition from the highest occupied molecular orbital (HOMO) of 7 s to its lowest unoccupied molecular orbital (LUMO) (Figures 6 and S6; Tables S1 and S2). The HOMO has significant metal 3d character with additional contributions from the dipic ligand and to a lesser extent the substrate diol groups. The LUMO has a strong π‐character and is located on the triphenylphosphine atoms. This suggests the singlet excitation has the nature of a metal‐to‐ligand charge transfer (MLCT). The excited singlet state then crosses to a triplet state, potentially with assistance from vibrational overlap. For the triplet ground state of 7 t, a computed spin density of 1.89 is found on the vanadium centre, the unpaired electrons occupying d‐orbitals of the metal. This is prior to C−O bond cleavage and redistribution of spin density in species 8 t across the metal centre (1.14) and the reacting substrate diol carbon C₂ (1.07).

Figure 6.

Key molecular orbitals involved in proposed transition of intermediate 7 s from ground‐state singlet to the excited singlet (7 s HOMO to LUMO). Calculations performed at the CAM−B3LYP/SDD/6‐311+G**/SMD level of theory (although we note similar orbital features were obtained via M06‐L/SDD/6‐311+G**/SMD level).

We also compute the DODH pathway for corresponding cis‐isomer of S2, namely S2’ (Table 1, Figure S7). For substrate S2’, the transition state for the reduction by triphenylphosphine, TS3 s’, is 6.3 kcal mol⁻¹ lower than for S2 (Figure S7). The minimum energy crossing point for the transition of 7 s’ to 7 t’ is also more facile for this cis‐diol substrate, by 6.4 kcal mol⁻¹. Interestingly, however, the transition state for C−O cleavage prior to elimination of triphenylphosphine oxide, TS4 t’, could not be located. Instead, we find a route where triphenylphosphine oxide leaves before C−O scission, through transition state TS6 t’ to form radical intermediate 9 t’. The computed relative free energy for TS6 t’ is 30.7 kcal mol⁻¹ (Figure S7); this is 12.8 kcal mol⁻¹ more stable than for the equivalent of its isomeric TS6 t of trans‐diol S2 (Figure S4) where the trans‐diol displays a non‐ideal geometry.

In addition we note that a low‐lying singlet pathway to DODH is available to S2’, via transition state TS8 s’, with a calculated free‐energy barrier of 30.5 kcal mol⁻¹, similar to that of TS6 t’ (Figure S7). This extrusion route involves concerted C−O cleavage due to a syn orientation of the cis‐diol moiety. Thus, both singlet and triplet pathways are predicted to coexist for cyclic cis‐diol S2’. Overall, for singlet and triplet pathways, the rate‐determining step is the reduction via TS3 s’, with a barrier of 34.0 kcal mol⁻¹. Experimentally, we find similar yields for both S2 and S2’ isomers at the elevated temperature of 170 °C for 72 h. The disparity in the rate‐determining steps in the reaction pathways of S2, via TS4 t at 41.5 kcal mol⁻¹ and S2’ through TS3 s at 34.0 kcal mol⁻¹ would suggest that potentially additional unexplored pathways could be available to the cyclic trans‐diol.

We briefly also consider the computed reaction pathways of cis and trans isomers of 1,4‐anhydroerythritol, namely S1’ and S1 (Figure S1a). These are analogues of S2 and S2’ and allow consideration of the effect of including an oxygen in the ring. Here, unsurprisingly, we find trans‐diol substrate S1 follows a pathway analogous to S2, with stepwise cleavage of the diol C−O bonds via the triplet state (Figure S8). Indeed, the computed energies for the corresponding transition state TS4 t differ by only 2.7 kcal mol⁻¹ between S1 and S2. As for S2, no concerted routes to extrusion are accessible. The reduction step via TS3 s is rate‐determining for S1, with a relative energy of 39.8 kcal mol⁻¹, very similar to a value of 40.3 kcal mol⁻¹ for the equivalent TS3 s structure of S2 (Figure 3).

By contrast, cis‐diol S1’, is predicted to have accessible concerted as well as stepwise pathways for extrusion (Figure S9). As for S2’, the concerted routes are via the singlet state (TS7 s’ and TS8 s’, Figure S9); and the stepwise pathway is through the triplet state (TS6 t’), where the first diol C−O bond is cleaved after elimination of triphenylphosphine oxide. Both concerted transition states TS7 s’ and TS8 s’ satisfy the requirement for syn co‐planar arrangement in E_i elimination with O₁−C₁−C₂−O₂ dihedral angles of 6.1 and 4.8°, respectively. The energy barriers associated with the stepwise and concerted transition states are similar, ranging from 31.2–32.6 kcal mol⁻¹, and constitute the rate‐determining step.

Overall, the highest barriers for the DODH pathways of S1 and S1’ are similar in free energy to those of their respective S2 and S2’ analogues, differing at most by 2.7 kcal mol⁻¹ for the trans substrates and 4.3 kcal mol⁻¹ for the cis reactants. This suggests only a moderate effect on energetics by the presence of a ring oxygen in the substrate. Of significant note, for cyclic trans‐diol S1, the energy barrier for its V‐catalysed rate‐determining step is 39.8 kcal mol⁻¹ (TS3 s, Figure S8); this is over 30 kcal mol⁻¹ lower in energy than the rate‐determining step calculated previously for the DODH of S1 via the rhenium catalyst MTO. ^[12] Although the latter calculation differs in using the M06 rather than M06‐L functional employed here, the difference in computed barrier height is substantial.

Linear substrate

An interesting comparison for the DODH profiles of constrained cyclic diols is with that of a linear acyclic substrate, where the latter's diol moiety can in principle orient optimally with the catalyst. For this, we consider the commonly studied DODH substrate, 1‐phenyl‐1,2‐ethanediol S4 (Table 1). For S4, we find that concerted singlet pathways (via TS7 s and TS8 s) and stepwise pathways involving the triplet state (through TS4 t and TS6 t) are predicted as feasible mechanisms of DODH (Figure 7). For the stepwise routes of S4, these energy barriers are 2.2 to 10.8 kcal mol⁻¹ lower than for its concerted pathways. The rate‐determining step is reduction, with an energy barrier for TS3 s of 30.1 kcal mol⁻¹, more similar to that of cis‐diol S2’ (34.0 kcal mol⁻¹) than trans‐diol S2 (40.3 kcal mol⁻¹).

Figure 7.

DODH reaction pathway for 1‐phenyl‐1,2‐ethanediol (S4). Free energies [kcal mol⁻¹] computed at the M06‐L/SDD/6‐311+G**/SMD level of theory.

Indeed, both the reduction and extrusion barriers of S4 were the lowest calculated of all substrates considered in our study, with energy values of 30.1 and 22.6 kcal mol⁻¹, respectively (Figure 7). Correspondingly, S4 had the highest olefin yield of all the substrates observed experimentally, of 36% in visible light and 14% in its absence (Table 1). Nevertheless, comparison of the TS4 t geometries for the DODH reaction of S4 compared to trans‐diol substrate S2 shows similarities in bond angles and distances (Figure S10): in the fully flexible linear substrate S4, the orientation of the diol carbons C₁ and C₂ gives a O₁−C₁−C₂−O₂ torsion angle value of 45.1° (Figure S10b), similar to a value of 52.8° in the trans‐cyclic diol S2 (Figure 5a). Likewise, the C₂…O₂=V distance is 1.99 Å in S4 and 2.07 Å in S2 (Figure S10). This similarity suggests that the trans stereochemistry of the aliphatic CH protons in S2 is compatible with a productive presentation of the diol O₂ oxygen to the V centre of the DODH catalyst.

Six‐membered ring substrates

Our observation that [n‐Bu₄N]‐(dipic)VO₂ can catalyse the DODH of both five and six‐membered ring trans diols experimentally (Table 1) is encouraging for the potential processing of carbohydrate compounds. As an indication of the feasibility of processing trans‐diol hexapyranose saccharides, we compute the DODH reduction and extrusion steps for sugar analogues trans‐ and cis‐1,5‐anhydro‐2‐deoxypentitol S5 and S5’ respectively (Figures S11 and S12). As for other trans‐diol substrates S1 and S2, the six‐membered pyranose S5 can proceed through a stepwise extrusion DODH mechanism via the triplet state (Figure S11). Interestingly, the extrusion step through TS4 t is significantly more favoured for S5 than five‐membered ring substrates, with energy barriers lower than for the corresponding S1 and S2 structures by 8.9 and 11.6 kcal mol⁻¹, respectively; and for extrusion via the TS6 t transition state, energy barriers lower by 16.5 and 16.3 kcal mol⁻¹, respectively. Interestingly, a singlet route for stepwise extrusion is also possible for the more flexible six‐membered ring system of S5: the transition state TS6 s is lowered in energy from 49.1 kcal mol⁻¹ for S1 to 28.6 kcal mol⁻¹ for S5. While stepwise singlet and triplet pathways are accessible to S5, no concerted pathway is found, as for other the trans‐diol substrates we considered. The rate‐determining step for DODH is predicted to be the reduction: transition state TS3 s has a computed relative free energy of 31.7 kcal mol⁻¹ (Figure S11), which is also reduced compared to the strained rings of S1 (39.8 kcal mol⁻¹, Figure S8) and S2 (40.3 kcal mol⁻¹, Figure 3).

The six‐membered ring cis‐diol S5’ is able to proceed through a concerted route, with a barrier of 33.2 kcal mol⁻¹ via transition state TS8 s’ (Figure S12). Interestingly, S5’ is also able to follow a stepwise spin‐crossover mechanism such that TS4 t’ could be located, representing a barrier of 23.2 kcal mol⁻¹ for formation of intermediate 8 t’ (Figure S12). This suggests that the additional degrees of freedom of the tetrahydropyran ring facilitate the geometry required for formation of TS4 t’, in contrast to five‐membered ring substrates such as S1’ and S2’. Indeed, substrate S5’ possesses a O₁−C₁−C₂−O₂ dihedral angle of −30.6° in intermediate 7 t’, in contrast to the almost co‐planar arrangement of −2.2° for the 7 t’ intermediate of substrate S1’. Correspondingly, substrate S5’ and catalyst 1 can suitably orientate for elimination, forming a O₁−C₁−C₂−O₂ torsion angle of 42.5° in TS4 t’. The rate‐determining step for DODH of S5’ is the reduction, with an energetic barrier of 32.1 kcal mol⁻¹ through TS3 s’. In this case, the barrier is slightly higher than that found for the other cyclic cis‐diols, for example 30.7 kcal mol⁻¹ for TS3 s’ of S1’ (Figure S9). The difference between the rate‐determining step through TS3 s for S5 and S5’ is 0.4 kcal mol⁻¹. Therefore, it is unsurprising that for model substrates cis‐ and trans‐1,2‐cyclohexanediol, we observe comparable yields experimentally where we obtain 14 % of cyclohexene from the DODH reaction of the trans isomer in comparison to 16 % for the cis form (Table 1).

Conclusions

The feasibility of the vanadium‐catalysed DODH reaction of cyclic trans‐diols has been demonstrated experimentally and explored computationally. Here, we show that direct DODH of cyclic trans‐diols can occur in the presence of catalyst, [n‐Bu₄N]‐(dipic)VO₂ and triphenylphosphine reductant. Furthermore, for the five substrates considered here, product yield shows a dependence on light. This observation suggests the involvement of the triplet state of vanadium in the DODH mechanism involving these substrates, although the generality of this finding to a wider range of diol substrates remains to be established. Computational study of the mechanism proposes a two‐step olefin extrusion for the five‐membered ring trans‐diol substrates S1 and S2 via the triplet state; for the six‐membered ring S5 trans‐diol, in addition to this pathway, a stepwise route on the singlet surface is also predicted as possible. For cis‐diols S1’, S2’ and S5’ and linear substrate S4, concerted pathways on the singlet surface are identified as well as stepwise routes via intersystem crossing. We note that the computed rate‐determining free energy barrier for the DODH of six‐membered ring S5 trans‐diol is 31.7 kcal mol⁻¹; for the five‐membered trans‐diol substrates S1 and S2, these barriers are 39.8 and 41.5 kcal mol⁻¹ respectively. While this is consistent with the observation that DODH of trans‐diols is not a very efficient reaction, the calculated estimates for S1 and S2 are at the limit of what is acceptable for the generation of alkenes under the reported experimental conditions (150–170 °C, 24–72 h). The rate‐determining energetic barriers calculated here for stepwise DODH of S1 and S2 correspond to estimated rates of approximately 10⁻⁸ s⁻¹ at the elevated temperatures required for the reaction. This may point to the potential existence of additional DODH routes for these five‐membered ring systems. Nevertheless, in this work we have demonstrated the feasibility of DODH of five‐ and six‐membered cyclic trans‐diol substrates using V catalysis. In particular, the ability of this vanadium‐based DODH catalyst to process six‐membered trans‐diol compounds suggests that further optimisation of its efficiency could provide a new cost‐effective route to converting biomass‐derived substrate diols such as glucose into high‐value alkenes.

Experimental Section

All reagents were purchased from Sigma Aldrich and used without further purification. DODH reactions were repeated three times and average yields are reported here. We describe here the general DODH reaction set up: under N₂, the substrate diol (0.1 mmol), PPh₃ (0.15 mmol), [n‐Bu₄N]‐(dipic)VO₂ (0.01 mmol) and chlorobenzene (0.5 mL) were combined in a thick‐walled ace pressure tube, equipped with a stirrer bar. The sealed tube was placed in a preheated oil bath (temperature 170 °C) and the reaction was left to stir for 24–72 h. The reaction was cooled in an ice bath and the internal standard mesitylene (18 μL, 0.13 mmol) was added and the reaction was stirred. A small aliquot was taken and diluted with CDCl₃ and the yield was determined by ¹H NMR spectroscopy. [n‐Bu₄N]‐(dipic)VO₂ 1 was synthesised from [n‐Bu₄N]VO₃ according to the procedure reported by Chapman and Nicholas. ^[8a] [n‐Bu₄N]VO₃ was synthesised according to the procedure of Day et al. ^[16]

Computational methods

All density functional calculations were performed using the Gaussian 09 electronic structure package. ^[17] We use the Minnesota suite of functionals: specifically, energy and geometry calculations employed the M06‐L functional, ^[18] suited to computing thermodynamics of transition metal systems. ^[19] For example, the M06‐L functional was found to perform well in the challenging task of predicting the ground spin states of Fe^II, Fe^III and Fe^IV systems. ^[20] An SDD effective core potential ^[21] was employed on V and a 6‐311+G** basis on all other atoms. Solvent was modelled using the SMD solvation model of benzene. ^[22] We note that this level of theory reproduces well the bond distances of the crystallographic structure of [(dipic)VO₂]⁻ anion, with an average deviation in bond distance of 0.01 Å (Table S3). The largest differences are for V−O distances, which are predicted to within 0.05 Å; this arises from the involvement of these oxygen atoms in a hydrogen‐bonding interaction with neighbouring ammonium counterions in the crystal structure. ^[23]

Stationary points were characterised as minima or transition states by calculation of vibrational frequencies and intrinsic reaction coordinate (IRC) analyses. Thermal corrections for free energies were carried out at 298.15 K and 1 atm using harmonic frequencies obtained above. Restricted and unrestricted DFT calculations were used for singlet and triplet states, respectively. The wavefunctions for all stationary points on the reaction pathway of trans‐1,2‐cyclopentanediol were tested and confirmed as stable. Minimum energy crossing points (MECP) between singlet and triplet states were calculated using the program developed by Harvey et al. ^[24] Excited state calculations using time‐dependent DFT were carried out, solving for the first twelve low‐lying excited states. This was performed using the CAM−B3LYP functional ^[25] employing the SDD pseudopotential for vanadium atoms and 6–311+G** for all other atoms. The solvent was modelled via SMD model for benzene. These were performed using geometries obtained at the M06‐L/SDD/6‐311+G**/SMD level of theory.

Supporting information

The Supporting Information includes computed reaction schemes, energetics, geometries, molecular orbitals, predicted excitations and UV/Vis spectra, as well as further experimental details.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

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.

Supplementary

E. Aksanoglu, Y. H. Lim, R. A. Bryce, ChemSusChem 2021, 14, 1545.