Theoretical and Experimental Investigations of Acetal Formation During the Oxidation of Xylose to Formic Acid Catalysed by H₅PV₂Mo₁₀O₄₀ in Methanolic‐Aqueous Solution

Abstract

The BioValCat (Enhanced Biomass Valorization by Engineering of Polyoxometalate Catalysts) project aims at developing homogeneously catalysed, selective biomass transformation technologies ultimately leading to an industrial viable biomass valorisation process. Herein, investigations of the conversion pathways for selected model substrates is one of the major tasks. Besides the catalyst‐solvent interactions, the substrate‐solvent interactions also play an important role. In this study, we aimed at answering the fundamental research question: what are the key mechanisms in aqueous‐alcoholic solutions preventing decarboxylation and where are the limits with respect to process conditions and solvent composition? Therefore, a special focus was placed on the influence of the functional groups in different reaction intermediates from HPA‐2 (H₅PV₂Mo₁₀O₄₀) catalysed xylose oxidation on the way to formic acid. We found that the protection of reactive functional groups of the model carbohydrates depends on the formation and stability of several reaction intermediates, which are formed via acetalization or esterification. This is strongly dependent on solvent composition and reaction conditions. The present findings were supported by DFT‐calculations explaining the different effects of the solvent composition on both substrate and catalyst, emphasising the catalyst‐solvent interactions are of superior importance for catalysis.

The substrate‐solvent interactions play a major role in the modified OxFA process. Herein, the formation and stability of several reaction intermediates via acetalization depending on solvent composition and reaction conditions is investigated.

1. Introduction

Biomass conversion is one of the most relevant strategies for the sustainable production of platform chemicals [1, 2, 3]. This includes formic acid (FA), which can be applied in textile, pharmaceutical, and leather production as well as chemical synthesis, or even as a hydrogen carrier [4, 5]. The most common industrial production pathway to date (annual production 995 kilotons in 2024) is the carbonylation of fossil‐based methanol followed by hydrolysis of the respective ester [6].

A more sustainable alternative is the OxFA process whereby formic acid is produced through the selective catalytic oxidation of different kinds of abundant biomass. This process is catalysed by H₅[PV₂Mo₁₀O₄₀] (HPA‐2), a Keggin‐type vanadium‐substituted polyoxometalate (POM) [7, 8, 9]. POMs are known for their variety of molecular and electronic structures, which make them attractive in the field of catalysis [10, 11, 12, 13, 14, 15]. The catalyst oxidises the biogenic substrate, while undergoing reduction from V⁵⁺ to V⁴⁺ and is subsequently reoxidised by molecular oxygen via a per‐oxo intermediate [16]. This redox behaviour of the catalyst can be predicted by density functional theory (DFT) calculations and followed using UV‐Vis spectroscopy, specifically observing the inter‐valence charge transfer (IVCT) of the reduced form [17].

When in situ extraction was used, yields of up to 85% of FA could be achieved using glucose as a substrate in an aqueous‐alcoholic solution and even 61% FA from raw lignocellulosic biomass like beech or spruce wood [18]. Other processes, such as the wet oxidation of carbohydrates like xylose with H₂O₂ as the oxidant, can reach FA‐yields of up to 89% but require the addition of alkali salts and achieve only low yield when using raw biomass [19]. Therefore, the direct catalytic conversion using molecular oxygen as oxidant is a more promising approach [20]. The OxFA process can be further improved by manipulating the solvent system. This has a huge impact on the selectivity and the kinetics by inhibiting undesired side reactions by forming more stable intermediates [21], hydrogen bonding, solubility of the substrate or intermediates [22], and catalyst‐solvent interactions [23] and even facilitating different reaction pathways [24, 25, 26]. Using cosolvents such as methanol, ethanol, iso‐propanol, or dimethyl sulfoxide (DMSO) in the oxidative conversion of biogenic substrates increases the yield of FA significantly, suppressing the formation of CO₂ as an undesired side reaction [27, 28, 29, 30, 31].

For the modified OxFA process, using pure methanolic solution increases the selectivity towards methyl format to almost 99% with H₈PV₅Mo₇O₄₀ (HPA‐5) as a catalyst under mild conditions of below 100°C and oxygen pressures below 10 bar but drastically decreases the reaction rate [32]. This effect has been attributed to the superior catalyst‐solvent interactions. Specifically, alcohol‐activated vanadium‐containing (POM) complexes were observed by ⁵¹V‐nuclear magnetic resonance (NMR) and electron paramagnetic resonance spectroscopy (EPR) spectroscopy [33]. A similar effect was observed by Wu et al., who achieved an FA yield of 89 % using HPA‐2 + H₂SO₄ as the catalytic system in an aqueous solution containing 20 vol% methanol [34, 35]. In this case, the effect was attributed to the scavenging of hydroxyl radicals that form during the reoxidation process by methanol, thereby hindering overoxidation to CO₂ [28, 34].

DFT calculations suggest that the POM‐catalyst interacts with the solvent competitively and this interaction might affect both the selectivity and the reaction rate [36]. On the other hand, substrate‐solvent interactions are also postulated to play a crucial role, where the use of DMSO in combination with sodium vanadate (NaVO₃) and H₂SO₄ as the catalyst stabilises the reactive aldehyde groups via hydrogen bonding [27]. It was further discovered that the use of alcohols leads to the formation of acetals, which may protect reactive functional groups [29, 35]. Furthermore, it is known that POMs can be used for acetalization reactions, such as glycerol acetalization to solketal [37].

Since both, the catalyst and the substrate may interact with the solvent in a competitive manner, influencing both selectivity and kinetics, we herein studied the formation of acetals and their influence. We also investigate how the catalyst interacts with the different intermediate products, in order to establish, which of these interactions are crucial to the modified OxFA process.

2. Experimental Details

The H₅PV₂Mo₁₀O₄₀ (HPA‐2) catalyst was synthesised according to a previously reported procedure [38]. All details regarding synthesis and characterisation can be found in the Supporting Information.

2.1. Reduction of HPA‐2 with Aqueous Hydrazine

Four grams of (2.3 mmol) HPA‐2 was added to 50 ml of demineralised water. Once the solid had been completely dissolved, 2 ml of an aqueous hydrazine solution (0.8 M) was added to the solution, changing its colour from orange to dark green. As hydrazine is a highly reactive and toxic substance, it should only be handled by suitably qualified professionals. The solution was then heated to 70°C with stirring for 1 h, during which time the colour changed to dark blue. After evaporating the solvent and drying under argon for 2 h at 150°C, a dark blue solid was obtained. The characterisation data can be found in the Supporting Information (see Figures S1, S2).

2.2. Setup and Experimental Procedure for Catalytic Experiments

The kinetic experiments were performed in a threefold parallel high pressure stirred tank reactor setup (see Figure S3) in a batch mode. For each experiment, a 50 mL measuring cylinder was filled with 0.5 mmol catalyst, 7.5 mmol substrate, and 50 mL of solvent. 45 mL of the solution was transferred into a glass liner. The remaining solution was used for analysis. For experiments under nitrogen, the solvent was degassed for 30 min with a porous stone and nitrogen. A mixture of 50 vol.% methanol (MeOH) or 50 vol.% ethanol (EtOH) in water was used as a solvent. The glass liners were then inserted into the 100 mL autoclaves. After closing and ensuring leak‐tightness, the autoclave was purged three times with the gas used in the experiment at 15 bar. The autoclaves were pre‐pressurised to 40.5 bar. The reaction temperature was then set to 100°C, with a stirrer speed of 300 rpm. Once the reaction solutions had reached the desired temperature, a liquid sample was taken from the valve. The stirrer speed was then adjusted to 1000 rpm to mark the start of the reaction time and to start gas entrainment. Samples of the liquid phase were taken every 30 min, cooled down on ice for a few minutes, and then prepared for further analysis. The samples were generally handled under ambient conditions. Due to mass transfer limitations, exposure to air did not result in significant reoxidation during the short handling time.

In the case of reactions under nitrogen atmosphere, UV‐Vis measurements were carried out, using an Agilent Technologies Cary 60 UV‐Vis spectrometer with a quartz cuvette of 10 mm path length. Afterwards, the samples were filtered with syringe filters (0.45 μm) and prepared for gas chromatography coupled with mass spectrometry (GC–MS) measurements and high‐pressure liquid chromatography (HPLC). For the GC–MS analysis, an Agilent system with a DB‐WAX UI column was used, using 1 vol.% chloroform as an internal standard. Subsequently, HPLC measurements were performed using a Nexera 40 system from SHIMADZU with a polymer phase organic acid column (300 × 8 mm) from Chromatography Service GmbH. For nuclear magnetic resonance (NMR) spectroscopy, 0.5 mL of unfiltered reaction solution was mixed with 0.1 mL D₂O containing 10 wt.% tert‐butanol as an internal standard. Additionally, the pH value of the reaction solution was measured before and after the reaction. The gas phase was analysed using a VARIAN 450‐GC equipped with a SHIN‐CARBON‐ST‐COLUMN (2 m x 0.75 mm). The gas sample was transferred directly from the reactor at room temperature into a gas sampling bag, then injected through a 0.25 mL sample loop. Samples were analysed for CO, CO₂, N₂, and O₂ using a thermal conductivity detector (TCD) where no other side‐products could be detected. Quantification was performed using stored pure substance calibration data with the Galaxy Chromatography systems software. Example chromatograms of all the previously described analysis are provided in the Supporting Information (see Figures S4‐S6).

For long‐term experiments (24 h), as well as pressure and solvent composition variations, a 10‐fold parallel screening plant was used (see Figure S7). The autoclaves were filled with 10 mL of the respective components, as previously described before for the 3‐fold plant. PTFE‐coated magnetic stirring bars were added to each autoclave. After sealing the reactor, it was connected to a steel tube with a rupture disc and placed in a heating block. The autoclaves were then purged with the gas used in the respective experiment. Subsequently, the pre‐pressure was adjusted, the temperature set to 100°C, and the stirring speed to 300 rpm. After the heating plate reached 100°C, the stirring speed was adjusted to 1000 rpm. This point in time was marked as starting point of the reaction (0 min). After 100 min or 24 h, the reaction was stopped by lowering the stirring speed to 0 rpm, removing the autoclaves from the heating block and allowing them to cool to ambient temperature. The reaction solution was prepared for analysis as described above after reaching room temperature.

2.3. Calculations

The conversion (X) of each substrate was calculated using Equation (1), where n is the molar amount and t the time the sample was taken.

$$X=\hspace{0.17em}\frac{n_{\mathit{\text{substrate}},t=0\hspace{0.17em}}-\hspace{0.17em}{n}_{\mathit{\text{substrate}},t}}{n_{\mathit{\text{substrate}},t=0}}$$ (1)

The yield (Y) of the products (i) in the liquid phase and in the gas phase were calculated by Equation (2).

$$Y_i=\frac{n_{\mathit{\text{carbon}}\hspace{0.17em}\mathit{\text{atoms}},\mathit{\text{product}},i}}{n_{\mathit{\text{carbon}}\hspace{0.17em}\mathit{\text{atoms}},\mathit{\text{substrate}}\hspace{0.17em}t=0}}$$ (2)

2.4. Computational Details for DFT Calculations

For DFT modelling, the α−1,4 isomer of HPA‐2 was chosen, as it has been shown to be the most relevant with respect to catalytic activity [16].

Geometry optimisations of HPA‐2 (with and without intermediates) were calculated using the r²SCAN‐3c functional implemented in Orca 6, with intermediates located at the different symmetry unique sites of HPA‐2 (see Figure S8). Solvation effects were accounted for using the conductor‐like polarisable continuum model (C‐PCM), using either water or methanol. Stationary points were considered converged using the default options in Orca 6 (energy changes <5 x 10⁻⁶ Hartree; max. gradient and RMS gradient <3 x 10⁻⁴ and 1 x 10⁻⁴ Hartree/Bohr, respectively; max. displacement and RMS displacement <4 x 10⁻³ and 2 x 10⁻³ Bohr, respectively). The nature of the stationary points was determined using harmonic vibrational frequency analysis. The binding energies were calculated as the difference in the Gibbs free energy between the optimised bound state and a supermolecule calculation, in which HPA‐2 and the intermediate were separated by at least 10 Å. For the supermolecule calculations, imaginary frequencies less than  − iω = 20 cm⁻¹ were neglected. Time‐dependent DFT (TD‐DFT) was performed for the top 4 binding motifs of glycolaldehyde, as well as methanol, using the same functional. The calculated binding energies for all positions are shown in Table S3‐S6.

3. Results and Discussion

In order to study the influence of acetalization on the oxidation of a sugar‐based substrate, we focussed on the conversion of xylose using HPA‐2 as a catalyst, as this has not yet been thoroughly investigated, whereas the catalytic conversion of glucose has been extensively researched. However, most of the intermediates of the oxidation of glucose or other hexoses are similar to those of xylose, which makes the results transferable to some degree [39]. We used HPA‐2 as the catalyst, as it is known to effectively catalyse the aerobic oxidation reaction. Although H₈[PV₅Mo₁₀O₄₀] (HPA‐5) is more efficient for aerobic oxidation of carbohydrates, reducing the degree of substitution helped to drastically reduce the complexity of the model system from 42 to only five isomers [28, 40].

In an aqueous solution, xylose undergoes a C2‐C3 cleavage via retro‐aldol condensation in the presence of a POM such as HPA‐2 leading to glyoxal and glyceraldehyde [41, 42]. Due to further oxidation in the presence of the POM catalyst, other intermediates like formaldehyde and glycol aldehyde are formed. Finally, further oxidative conversion results in the formation of FA, with CO₂ being produced as an undesired by‐product (Figure 1). Using methanol as a cosolvent or additive shifts the selectivity to FA/MF as seen in Table 1.

FIGURE 1.

Proposed catalytic conversion of xylose to formic acid in the presence of oxygen and HPA‐2 in aqueous‐methanolic solution [35, 36].

TABLE 1.

Selectivity of catalytic conversion of xylose in the presence of oxygen and HPA‐2 in aqueous vs aqueous‐methanolic solution.

Substratea	Solvent	Conversion, %	Yield, %
			FA/MF	CO2	Intermediates
Xylose	H2O	100	52	36	12
	50 vol.% MeOH	100	87	3	10

^a Reaction conditions: c (substrate) = 150 mM, c (HPA‐2) = 10 mM p(O2) = 45 bar, T = 100°C, t = 24 h, V = 10 mL, stirrer = 1000 rpm.

As expected from previous studies, MeOH addition efficiently suppressed decarboxylation, and therefore, only 3% of CO₂ was formed. This is in agreement with previous work investigating the influence of additives where the addition of methanol leads to higher selectivity to C1 products but a decrease in conversion rate [36]. The overoxidation of glycol aldehyde and glyceraldehyde to their respective carboxylic acids and the latter further to CO₂ is described in the publications of Wu et al. [31]. As Ramani et al. detected glycol aldehyde dimethyl acetal, they proposed the idea that aldehyde groups form acetals in the alcoholic solutions (see Figure 1) during the oxidation of glucose. After 10 min at 160°C, the yield of the C1 products reached 70%, with complete glucose conversion [35].

In this contribution, the addition of 50 vol.% MeOH had only a little influence on the intermediates formed, as well as on their further conversion except that the overoxidation to the carboxylic acid might be inhibited via acetal formation. For this reason, we conclude that the intermediates formed in the oxidation of xylose in an aqueous/methanolic solution are generally the same as those in a purely aqueous system.

We therefore used the above‐mentioned intermediates individually as substrates to study their acetalization in detail, varying the reaction conditions and the nature of the HPA‐2 catalyst.

One of the main intermediates of xylose oxidation is glycol aldehyde, which we used as a substrate to determine whether detectable acetal formation occurs under typical OxFA process conditions. Glycol aldehyde conversion under oxygen atmosphere is relatively slow, taking up to 150 mins for complete conversion; C1 (FA + MF) are formed as the main products with a yield of 65% at full glycol aldehyde conversion (Figure 2). Only a very small amount of glycolic acid (5%) is formed, and small quantities of CO₂ (<1%) were detected. The glycol aldehyde dimethyl acetal could also be detected in small quantities via GC–MS (see Figure 3, Figure S9) and NMR (see Figure S10‐S11). Surprisingly, the acetal begins to form during the heating phase (14% yield) and is steadily converted with the start of the reaction, where the gas entrainment is increased. Methanol is partly oxidised to formaldehyde, which then forms dimethoxymethane (DMM) [32, 34]. As we have previously demonstrated, under the applied reaction conditions, the C1 products originate predominantly from xylose and not from the oxidation of methanol [32]. Partial methanol oxidation occurs as a side reaction in all of the applied reactions, accompanied by acetalization of formaldehyde [43].

FIGURE 2.

Conversion of glycol aldehyde and yields of intermediates vs reaction time in aqueous‐methanolic solution under oxygen atmosphere. Reaction conditions: c (glycol aldehyde) = 150 mM, c (HPA2) = 10 mM, p(O2) = 45 bar, T = 100°C, t = 150 min, V = 45 mL, stirrer = 1000 rpm, solvent composition: 50 vol % MeOH in H₂O.

FIGURE 3.

GC–MS chromatogram of reaction solution under oxygen atmosphere at 0 min reaction time (right) and the GC–MS fingerprint (left).

No significant changes to the HPA‐2 catalyst could be observed in the NMR spectra during the oxidation of glycol aldehyde (see Figure S12). The ⁵¹V‐NMR spectrum shows the typical pattern for HPA‐2: a strong signal at −531 ppm with smaller signals at −539 and −541 ppm. A small low field shift (0.3 ppm) was observed at the end of the heating phase and after the addition of oxygen. No broadening of the signals was detected. Despite the signals being shifted compared to a pure aqueous solution, all signals can be assigned to the catalytically active isomers of the twofold vanadium‐substituted HPA‐2 catalyst [44]. To determine whether glycol aldehyde dimethyl acetal is an intermediate, the reaction time was increased to 24 h, after which the selectivity of glycol aldehyde and glycol aldehyde dimethyl acetal was compared. Both substrates were fully converted, and 65–66% of C1 products were found in the liquid phase, along with 2% CO₂. Therefore, we conclude that acetal formation did not lead to any different products compared to the oxidation in pure aqueous solution and that the glycol aldehyde dimethyl acetal is an intermediate in this reaction network. During the time‐resolved experiments, no other side products were detected. Thus, we suspect an equilibrium between the aldehyde and the acetal groups, which could function like a protective group.

To study the influence of the oxidative conditions (45 bar O₂) on acetal formation, we repeated the experiment using the same the reaction conditions but replacing the oxygen atmosphere with nitrogen (see Figure S13). Here, the conversion of glycol aldehyde reached only 36% after 150 mins reaction time. As the catalyst cannot be reoxidised under anaerobic conditions, the oxidative conversion of the substrate is limited, and only 3% of C1 products were formed. In contrast to the reaction under oxygen, significant amounts of glyoxal could be detected. Also, glycol aldehyde dimethyl acetal is formed during the heating phase. However, it is converted very slowly, indicating that the acetal might be less stable under the applied reaction conditions.

In order to study the influence of the nature of the HPA‐2 catalyst on acetal formation, we repeated both experiments without catalyst. Acetal formation was also observed without HPA‐2, but it is much slower than in the experiment with HPA‐2. Using either oxygen or nitrogen in the gas phase leads to a yield of 19% glycolaldehyde dimethyl acetal (see Figure S14). It is clear that the acetal formation is influenced by HPA‐2. To elucidate whether this is due to its Brønsted acidity, phosphomolybdic acid (HPMo) and H₂SO₄ were employed under pure nitrogen atmosphere. Again, similar yields of acetals were observed even though the pH value differed slightly (see Table 2, Figure S15).

TABLE 2.

Comparison of glycol aldehyde conversion using different catalysts.

Catalysta	Conversion, %	Acetal, %	Glyoxal, %	C1, %	pH (end)
HPA‐2	35	15	10	4	2.01
HPA‐2*	30	14	11	2	1.99
H2SO4	15	13	—	—	1.97
HPMo	25	15	10	—	1.68

^a Reaction conditions: c (glycol aldehyde) = 150 mM, c (HPA‐2/HPMo) = 10 mM, V (94‐96 % H₂SO₄) = 0.038 ml, p(N₂) = 45 bar, T = 100°C, t = 150 min, V = 45 mL, stirrer = 1000 rpm, solvent composition: 50 vol % MeOH in H₂O.

Interestingly, no glyoxal formation was observed in the experiment using H₂SO₄ as a catalyst despite the latter being observed for using HPMo. Furthermore, in both cases, no C1 products could be detected. In case of HPMo, the redox potential was high enough to convert some of the glycol aldehyde to glyoxal. Although substitution of molybdenum with vanadium is required to achieve high redox activity in the POM catalyst [45, 46], the consistent observation of acetal formation with the different Brønsted acids suggests that it is mainly the H⁺ concentration that promotes acetal formation.

To investigate the effect of the catalyst's oxidation state further, we reduced the HPA2 catalyst with hydrazine prior to the experiment (details can be found in the corresponding section of the Supporting Information). After evaporating excess hydrazine, we tested the reduced HPA‐2* catalyst for the oxidation of xylose and found that the pretreatment had no significant impact on the oxidation of xylose under oxygen atmosphere. Afterwards, we used the reduced HPA2* catalyst under nitrogen atmosphere to see if it has any effect on the acetal formation with glycol aldehyde as a substrate. Prior to this, we degassed the solvent by purging it with nitrogen. In addition, we measured the V⁴⁺ concentration ex situ using UV/Vis spectroscopy. The maxima of V⁴⁺ shifted from 750 to 780 nm during the reaction [47, 48, 49]. We plotted the 780 nm absorption peak against the time samples which were taken for both used HPA‐2 catalysts (see Figure 4). TD‐DFT calculations reveal this band to have predominantly V(d) → V + Mo (d) character. We observed other maxima at 705 and 875 nm, respectively, after 30 min which most likely belong to the Mo^V+ species in both reactions [50]. Based on the Beer–Lambert law, we used the extinction as a semiquantitative measure for the amount of reduced HPA‐2*. As expected, the concentration of V⁴⁺ was higher at the beginning of the reaction when using the prereduced catalyst compared to the untreated HPA‐2. This led to lower conversion of glycol aldehyde and slightly less formation of C1 products. After 60 min, the concentration of V⁴⁺ was similar in both experiments with only minor deviations. The same progression was observed for the Mo^V+ species (see Figure S16). Notably, the acetal formation is barely affected by the pretreatment, which leads to the conclusion that only the Brønsted acidity affects the acetal formation and the oxidation state of the catalyst does not play a significant role here (Figure S17, S18).

FIGURE 4.

UV–vis spectra of diluted reaction samples in the presence of reduced HPA‐2* (left). The extinction at 780 nm vs the corresponding reaction time (right) for reaction samples of reduced HPA‐2* (black) and HPA‐2 (red) as used catalysts for the conversion of glycol aldehyde‐dimer under nitrogen atmosphere.

To gain a deeper understanding of how glycol aldehyde interacts with both forms of the HPA2 catalyst, we investigated the binding of glycol aldehyde (and other intermediates and their acetals) with reduced HPA‐2* using DFT calculations for determination. Glycol aldehyde binds less favourably than methanol, with binding energies at the vanadium site of 35.9 kJ mol⁻¹ for glycol aldehyde and 42.0 kJ mol⁻¹ for methanol, both binding via H‐bonds between their O‐H group and the bridging oxygen between the two vanadium atoms. Glycol aldehyde dimethyl acetal binds more strongly to HPA‐2 than glycol aldehyde (or the solvent), with a binding energy of 47.8 kJ mol⁻¹ at the favoured binding site (at site ‘1^’, bound to the V‐O‐V motif via H‐bonding). This suggests that the acetal inhibits binding at this site once formed, slowing the oxidation processes and preventing overoxidation to CO₂. Glyoxal and formaldehyde as well as glyceraldehyde and their respective acetals bind less favourably (often by ∼10 kJ mol⁻¹), while glyceraldehyde binds more favourably than methanol, with a maximum binding energy of 47.5 kJ mol⁻¹ at the vanadium site; the acetal of glyceraldehyde binds less favourably (36.7 kJ mol⁻¹).

We also investigated the effect of the solvent composition on the conversion of glycol aldehyde dimer under nitrogen atmosphere. For this experiment, we used the 10‐fold multiparallel screening plant with a reaction time of 100 min, which has a lower gas entrainment due to magnetic stirring. Nevertheless, at a solvent composition of 50 vol%. MeOH in H₂O, we detected 12% of glycol aldehyde dimethyl acetal which makes this system comparable for this investigation. The amount of C1 products was lower by comparison, which we suspect is caused by differences in gas entrainment in the employed equipment. Since our above‐described experiments showed no influence of the gas phase on acetal formation, this is not of concern for this investigation. We observed an increase in acetal formation and a decrease in the amount of glyoxal with an increased methanol content (Figure 5, left). Only traces of glycolic acid could be detected. When pure methanol was used (with a water content of 0.045 wt.% as measured by Karl Fischer titration), almost all of the glycol aldehyde was converted into its acetal with a yield of 91%. The glyoxal formed was converted to the glyoxal dimethyl acetal with a yield of 4%. As Hao et al. demonstrated, the dominant component in methanol at room temperature is the glycol aldehyde hemiacetal, whereas in an aqueous solution it is mostly glycol aldehyde hydrate [51]. As the conversion of glycol aldehyde to glyoxal may involve protons, the formation of glyoxal decreases with the addition of relatively alkaline methanol (pKs = 16 in water) [52]. Furthermore, the presence of water as a solvent and a low pH value appear to suppress acetal formation. In addition, glyoxal dimethyl acetal was detected. Before and after the reaction, we measured the UV/Vis spectra of the solution and observed that absorbance correlated with the solvent composition (see Figure 5, right).

FIGURE 5.

Conversion of glycol aldehyde‐dimer and yields of intermediates in the presence of HPA‐2 in aqueous‐methanolic solutions under nitrogen atmosphere (left) and the UV/Vis spectra of the respective diluted reaction solutions after reaction. Reaction conditions: c (glycol aldehyde‐dimer) = 150 mM, c (HPA‐2) = 10 mM, p(N₂) = 45 bar, T = 100°C, t = 100 min, V = 10 mL, stirrer = 1000 rpm, solvent composition: 50 vol % MeOH in H₂O.

With increasing methanol content, the intensity of the bands at 705 and 875 nm decreases drastically. Using structures with glycol aldehyde bound to HPA‐2 at the top binding positions (1, 16, 7), these bands are only present when glycol aldehyde is not bound at vanadium. The same is observed with methanol; thus, as the concentration of methanol increases, binding of methanol to vanadium becomes more likely, leading to reduced intensity. As previously mentioned, these bands may be related to the concentration of Mo^V+ species. If glyoxal formation mainly occurs via molybdenum and not vanadium in this case, the conversion of glycol aldehyde might be inhibited either by acetal formation or by methanol addition. DFT calculations revealed that the binding energy of glycol aldehyde to HPA‐2 is −39.1 kJ mol⁻¹ in methanol compared to −22 kJ mol⁻¹ in water, while the binding energy of methanol to the POM is −42 kJ mol⁻¹. As the methanol content increases, glycol aldehyde and methanol compete for the binding position, which could result in reduced glycol aldehyde conversion to glyoxal. TD‐DFT calculations demonstrate that the bands between 705 and 875 nm are V(d) → V + Mo(d) transitions when no substrate is bound at vanadium; these bands then disappear once a substrate is bound at vanadium.

Finally, we used ethanol instead of methanol to see if acetal formation is influenced by the carbon chain length. Thirty‐eight percent of glycol aldehyde was converted after 150 min, which is comparable to the 36% conversion of glycol aldehyde in methanol/water as a solvent. The amount of glyoxal formed was also very similar. Unlike Gui‐Hua et al., no formation of acetal of glycol aldehyde could be observed either with GC–MS or with ¹HNMR [29]. Perhaps, the equilibrium is shifted further towards the aldehyde or hydrate, as adding ethyl groups is more challenging in terms of the stereochemistry. Nevertheless, the glycol aldehyde conversion rate in an aqueous‐methanolic solution was similar to that in an aqueous‐ethanolic solution, despite the fact that no acetal was detected in the latter.

Unlike with glycol aldehyde, we did not observe any acetal formation during the oxidation of glyoxal (Figure 6). This is most likely due to its fast conversion, reacting even at room temperature with the HPA‐2 catalyst. The binding of glyoxal to the POM is competitive with MeOH (as for glycol aldehyde; the binding energy is −40.7 kJ mol⁻¹ at the favoured position); given the experimental data, intrinsic reaction of glyoxal is probably just fast. When nitrogen atmosphere was applied, both aldehyde groups undergo acetal formation (see Figure 6). We were able to detect all expected compounds using GC–MS and could quantify the amount of glyoxal dimethyl acetal (see Figures S19‐S22). Compared to glycol aldehyde dimethyl acetal, the formation is relatively slow. Only a small amount (2%) of acetal is formed during the warm‐up phase. However, the concentration of acetals continues to increase as the reaction progresses reaching up to a maximum yield of 24%, which is higher in comparison to the yield of glycol aldehyde dimethyl acetal after 150 min with only 14%. Furthermore, it appears to be more stable under the applied reaction conditions, steadily increasing, whereas glycol aldehyde dimethyl acetal is slowly converted.

FIGURE 6.

Conversion of glyoxal and yields of intermediates in the presence of HPA‐2 in aqueous‐methanolic solutions under oxygen atmosphere (left) and nitrogen atmosphere (right). Reaction conditions: c (glyoxal) = 150 mM, c (HPA‐2) = 10 mM, p(O₂/N₂) = 45 bar, T = 100°C, t = 100 min, V = 10 mL, stirrer = 1000 rpm, solvent composition: 50 vol % MeOH in H₂O.

Without the HPA‐2 catalyst, no acetal formation could be detected. Once again, we used H₂SO₄ as catalyst in a control experiment and observed similarities to the acetal formation of glycol aldehyde which is primarily related to Brønsted acidity (see Figure 7), which confirms our previous findings.

FIGURE 7.

Conversion of glyoxal and yields of intermediates in the presence of H₂SO₄ in aqueous‐methanolic solutions under nitrogen atmosphere (left). The extinction at 780 nm vs the corresponding reaction time (right) for reaction samples of reduced HPA‐2* (black) and HPA‐2 (red) as used catalysts for the conversion of glyoxal under nitrogen atmosphere.

Using the reduced HPA‐2* as a catalyst, we found 27% of the glyoxal dimethyl acetal was formed, compared to 24% in the presence of HPA‐2 (Figure S23). After 150 min, the conversion of glyoxal was lower with 38% compared to the untreated conversion with 49%, where more C1 products (16% compared to 7%) were formed. Since less glyoxal is oxidised at the beginning, we assume that more glyoxal is available as a substrate to form the corresponding acetals. Using UV/Vis spectroscopy, we observed that some of the glyoxal had been converted prior to the reaction reaching equilibrium, resulting in an increase in V⁴⁺ species even at room temperature (see Figure 6). After that, the amount slowly increased further as more glyoxal was oxidised. In summary, acetal formation for glyoxal is relatively slow compared to the oxidative cleavage, therefore it is only detected under nitrogen atmosphere and mainly as a function of Brønsted acidity. With less glyoxal being oxidatively converted, more acetals are formed.

No acetal formation could be detected for glyceraldehyde, either under oxygen or nitrogen atmosphere. When oxygen was used the reaction was relatively fast, reaching full conversion after already 40 min. Glycol aldehyde as an intermediate was formed, intrinsically also the acetal (see Figure S24). The binding energies are 67 kJ mol⁻¹ higher than MeOH at V sites (and generally), thus not competitive. Therefore, it is likely that glyceraldehyde remains bound to HPA‐2 and does not react to form acetals. Under nitrogen atmosphere the conversion rate slows down drastically and dihydroxyacetone is formed due to the higher temperature [53]; methylglyoxal by dehydration [54, 55] (see Figure S25). Oxidation and dehydration appear to be favoured here, with no acetal formation observed. Without a catalyst, glyceraldehyde is only converted to dihydroxyacetone (see Figure S26). No significant differences were observed when the reduced HPA‐2* was used, since the dehydration of glyceraldehyde mainly involves Brønsted acidity.

Although formaldehyde is mainly formed from the oxidation of methanol and not from xylose oxidation, we investigated the formation of its acetal DMM as well. Under both anaerobic and aerobic conditions, DMM was formed during the heating phase where the gas atmosphere has no influence on the product composition (see Figure S27). After 30 min, 56% of the formaldehyde had been converted, with conversion occurring faster under aerobic conditions than under anaerobic conditions. In solution, the binding energy of MeOH to HPA‐2 is twice that of formaldehyde (−42 vs −21 kJ mol⁻¹), and this may inhibit the interaction. Therefore, formaldehyde is stable under these reaction conditions. In contrast to reactions involving other substrates, the reaction solution did not turn dark blue, and no changes could be identified using UV/Vis spectroscopy. Furthermore, no conversion of formaldehyde and thus no DMM was detected in the absence of HPA‐2. In this case, acetal formation is catalysed by the Brønsted acidity of HPA‐2. Therefore, it is not surprising that, as soon as methanol is used as an additive, DMM is also produced through the formation of formaldehyde. When the reaction time was prolonged in an aqueous solution and an aqueous methanolic solution, only a small quantity of formaldehyde was converted into C1 products. (see Table 3).

TABLE 3.

Selectivities of main intermediates in aqueous and aqueous‐methanolic solution in the presence of HPA‐2 under aerobic conditions.

Substratea	Solvent	Conversion, %	Yield, %
			FA/MF	CO2	others
Glycol aldehyde	H2O	100	53	13	9
	50 V.% MeOH	100	72	—	16
Glyoxal	H2O	100	56	16	0.3
	50 V.% MeOH	100	72	—	9
Glyceraldehyde	H2O	100	42	33	6
	50 V.% MeOH	100	51	3	32
Formaldehyde	H2O	3	2.3	—	—
	50 V.% MeOH	57	3	0.3	47

^a Reaction conditions: c (substrate) = 150 mM, c (HPA‐2) = 10 mM, p(O₂) = 45 bar, T = 100°C, t = 24 h, V = 10 mL, stirrer = 1000 rpm.

Based on these results, we performed an additional experiment using xylose as a substrate to improve our understanding of the influence of acetal formation on the catalytic oxidative conversion of xylose (see Figure 8). Herein, xylose is almost completely converted after 150 min reaction time. The main intermediates are glyoxal, which is formed very quickly, with a yield of 16% after 30 min and glycol aldehyde with a yield of 17% after 60 min. Subsequently, FA and MF are formed reaching a maximum yield of 65%, whereas again only traces of CO₂ (1.6%) could be detected. Glyceraldehyde and glycolic acid are steadily formed and further oxidised to C1 products. Formaldehyde is formed from solvent decomposition and as a result of acetalization to DMM [34].

FIGURE 8.

Conversion of xylose and yields of intermediates in the presence of HPA‐2 in aqueous‐methanolic solutions under oxygen atmosphere. Reaction conditions: c (xylose) = 150 mM, c (HPA‐2) = 10 mM, p(O₂) = 45 bar, T = 100°C, t = 100 min, V = 45 mL, stirrer = 1000 rpm, solvent composition: 50 vol % MeOH in H₂O.

Except for DMM, glycol aldehyde dimethyl acetal is the only acetal formation observed for all intermediates. Therefore, the results obtained using glycol aldehyde dimer can be transferred to the oxidation of xylose. Although glyoxal conversion is slower when glyoxal is used as the only substrate, no acetal formation was detected. As seen in the experiments, acetal formation is much slower under anaerobic conditions compared to the decomposition of glyoxal under aerobic conditions. Therefore, it is unlikely that acetal formation affects the conversion rate of glyoxal. The glyoxal‐HPA‐2 interaction (−40.7 kJ mol⁻¹ at the favoured binding site) competes not only with the MeOH‐HPA‐2 interaction but also with the other intermediates, leading to a decrease in the conversion rate. No acetal formation could be detected for glyceraldehyde, as it was not observable in previous experiments.

The inhibitory effect of adding alcohol on the conversion of glycol aldehyde is clearly evident. The chemical behaviour of glycol aldehyde is generally quite complex in alcoholic solutions, with acetal formation accounting for part of this behaviour. Therefore, acetal formation may impact the conversion of glycol aldehyde. Indeed, the binding energies of the various acetals considered in this study (from glyceraldehyde, glycol aldehyde, and glyoxal) with HPA‐2 are all competitive with methanol, which will also have a significant impact [36]. The yield of 72% of C1 products at a conversion of 100% is consistent with the yield after 150 min (Figure 2) with 69% at a conversion of 98%. Examining the selectivity of the oxidation of these intermediates in different solvents reveals that CO₂ formation is strongly inhibited for all intermediates, regardless of whether acetals are formed. For glyoxal, for which no acetal formation was observed under aerobic conditions, the production of CO₂ is completely suppressed. For glycol aldehyde dimer, where acetal formation occurs, the conversion to CO₂ is also suppressed. Other products were glycolic acid, formaldehyde and its acetal, respectively. Adding methanol for the glyceraldehyde conversion yielded only 3% CO₂. Therefore, the glycol aldehyde dimethyl acetal is formed as an intermediate, but it is unlikely to have a significant impact on selectivity.

The acetal formation does play a role in the formation of DMM as part of the decomposition of the solvent. After 24 h, nearly 80% of formaldehyde is converted to DMM due to the Brønsted acidity, and only a fraction of the C1 products are formed. Furthermore, DMM is stable under the applied reaction conditions. To increase the conversion of formaldehyde to formic acid, the pH value can be lowered by the addition of sulphuric acid leading the formation of the more active species VO²⁺. Overall, we conclude that acetalization has a minor influence on the catalytic oxidation of xylose and its intermediates in a methanolic‐aqueous solution, where the interaction between the solvent and the catalyst seems to have a greater impact.

4. Conclusion

The formation of acetals from individual intermediates of the catalytic oxidative conversion of xylose to formic acid/methyl formate in an aqueous‐methanolic solution containing HPA‐2 was investigated in order to determine its influence on selectivity and reaction rates. Glycol aldehyde dimethyl acetal was formed in different solvent compositions and various gas atmospheres. Modifying the redox state of the catalyst through pretreatment of HPA‐2 with the reducing agent hydrazine had no effect on the acetal formation. The use of HPA‐2, HPMo, and sulphuric acid confirmed that acetal formation is catalysed by Brønsted acidity. However, acetal formation for glycol aldehyde even occurred in the absence of any catalyst. Acetalization was only observed for glyoxal under anaerobic conditions and was very slow compared to the oxidative conversion under aerobic conditions. Finally, no acetal was detected for glyceraldehyde. Under anaerobic conditions, it appears that dehydration to pyruvaldehyde occurs more quickly. Formaldehyde, which was mainly derived from the oxidation of methanol, forms DMM as acetal, which is stable under the applied reaction conditions. During the oxidative conversion of xylose, only glycol aldehyde dimethyl acetal was formed, which may affect the conversion rate of glycol aldehyde. Regarding the selectivity of individual intermediates in pure aqueous solution, CO₂ suppression and acetal formation were uncorrelated compared to the addition of methanol. The interaction of the intermediates and methanol with HPA‐2 was DFT calculated with and compared. Based on these calculations, it can be concluded that the addition of methanol leads to competitive inhibition of the catalyst. This decreases the conversion rate of each individual intermediate, which may dominate in terms of solvent effects where acetalization tends to play a minor role.

Supporting Information

Additional supporting information can be found online in the Supporting Information section. The materials used, description of synthesis of the used HPA‐2 catalyst, catalyst characterization, together with respective HPLC chromatograms and lists of kinetic results can be found in the Supporting Information file.

Declaration of Generative AI and AI‐Assisted Technologies in the Writing Process

When writing this article, the authors used DeepL Write and Microsoft Word to improve the linguistic formulation. The authors reviewed and edited the content as needed and are fully responsible for the content.

Funding

This study was supported by the European Research Council (101086573).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supplementary Material

Data Availability Statement

The data that support the findings of this study are openly available in Zenodo at https://doi.org/10.5281/zenodo.17598491, reference number 17598491.