Conversion of xylose into D-xylitol using catalytic transfer hydrogenation with formic acid as H-donor

Aigner, D. J., Hinterholzer, L., Almhofer, L., Bischof, R. H., and Wrodnigg, T. (2023). “Conversion of xylose into D-xylitol using catalytic transfer hydrogenation with formic acid as H-donor,” BioResources 18(4), 8631-8652.

Abstract

d-Xylitol, a biomass-derived sweetener, is increasingly used in cosmetics and pharmaceutical products. The raw material for d-xylitol production, d-xylose, is easily accessible from dissolving pulp production. d-xylitol production involves the heterogeneously catalyzed hydrogenation of d-xylose; this process is energy intensive, as the use of H₂ requires high pressure and temperature. This work examined catalytic transfer hydrogenation for xylose conversion into xylitol. Formic acid (FA) was used to replace H₂ as the H-donor, as it is easily available, inexpensive, may be obtained from renewable sources, and it avoids the risks associated with the use of high-pressure inflammable gas. A variety of commercially available catalysts were screened to reveal the one enabling the highest yield. The experiments were performed at 40, 80, and 140 °C, with pure xylose as a model compound. Triethylamine (Et₃N) was added to ensure sufficient conversion rates. Based on the preliminary studies an experimental design was created (Design Expert^®), including the two best performing catalysts Ru/Al₂O₃ and Ru/C, to investigate the influence of temperature and H-donor and base concentration on xylitol yield. Ru/C resulted in maximum d-xylitol yield of 73.2 % at 100 °C, FA to d-xylose ratio 5:1 and Et₃N to FA ratio 0.4.

Download PDF

Full Article

Conversion of Xylose into d-Xylitol using Catalytic Transfer Hydrogenation with Formic Acid as H-Donor

Danuta Joanna Aigner,^a,* Lena Hinterholzer,^{a, c} Lukas Almhofer,^a Robert H. Bischof,^b and Tanja Wrodnigg^c

DOI: 10.15376/biores.18.4.8631-8652

Keywords: Catalytical transfer hydrogenation; Xylose; Formic acid; D-Xylitol; Ruthenium based catalysts; Ru/C, Ru/Al₂O₃

Contact information: a: Kompetenzzentrum Holz GmbH, Altenberger Straße 69, 4040 Linz, Austria; b: Lenzing AG, Werkstraße 2, 4860 Lenzing, Austria; c: IBioSys Institute of Chemistry and Technology of Biobased Systems, Graz University of Technology, Stremayrgasse 9, 8010 Graz, Austria;

* Corresponding author: d.aigner@wood-kplus.at

INTRODUCTION

Bio-based alternatives must be substituted for as many as possible resources of fossil origin as a means to reach the climate goals and to preserve the quality of life on Earth for future generations. Many potential biomass-originating chemicals are still underexploited, due to higher resulting production and purification costs compared to their petrochemical counterparts. This might change with the increase of CO₂-pricing, thereby demanding low carbon emitting technologies. However, ongoing development of innovative production processes and sustainable resource management still have to be done to guarantee the most efficient use of the raw materials (Dean et al. 2005).

Lignocellulosic biomass is an ideal source for the production of chemicals and raw materials, as it lacks competition with the food industry, is inexpensive, and may be processed into a wide variety of products (Clements and Van Dyne 2005). Cellulose, hemicellulose and lignin, the main components of lignocellulosic feedstock, differ in chemical structure and properties, leading to manifold potential applications. Looking at textile fibers (viscose or lyocell) as the intended product, dissolving pulp with highest possible cellulose content is manufactured as a basic material, with the majority of lignin and hemicellulose remaining in the cooking liquor after the initial pulping step. The high energy content of the residual lignin, due to the aromatic structure, makes it well suited for energy generation, increasing the process sustainability. Hemicellulose, on the other hand, especially when originated from hardwood, is an excellent source of xylose and may be further processed into xylitol, which is in strong demand on the global market (Bozell and Petersen 2010; Felipe Hernández-Pérez et al. 2019; Delgado Arcaño et al. 2020). The content of hemicellulose in hardwood is 25 to 35%, dependent on species. Thus, it has great potential for valuable by-products. The worldwide demand for wood originated textile fibers is expected to constantly increase, due to population growth and limited availability of cotton cultivation areas. In parallel, the availability of hemicellulose as a renewable raw material will continue to grow (Gschwandtner 2022).

Besides wood pulp production effluents, xylose containing hemicellulose is present in several feedstocks (straw, reeds, grass, wood, paper waste, etc.). However, using effluents originating from an established process carried out on a large scale guarantees a constant supply of the required raw material and results in a significantly reduced environmental impact compared to the competitive biomass hydrolysis process. After xylose separation from the acidic pulping liquor, it is converted to xylitol by catalytic hydrogenation (Heikki et al. 1999; Delgado Arcaño et al. 2020). Although the production process is well-established, it involves several operational and economical drawbacks, due to laborious and energy-intensive purification and H₂-containing steps (Melaja et al. 1977). The hydrogenation step is typically carried out in batch reactors at 80 to 140 °C and pressures up to 50 bar H_2.The reaction conditions require cost-intensive infrastructure due to the hazard of H₂ handling. Further, highly selective metal catalysts have to be used, to prevent side reactions and guarantee maximum xylitol yield, implying intensive xylose purification steps. Traditionally, Raney nickel (<0.1 mm) is used as catalyst. However, this catalyst has some disadvantages in batch production (toxicity, deactivation due to leaching and instability, formation of by-products such as aldonic acids). Considering the fact that xylitol is mainly used in food industry and cosmetics, catalyst toxicity presents a considerable risk and entails cost-intensive purification steps such as ion exchange, filtration, and crystallization. In recent years, a variety of noble metal based (Pt, Pd, Ru) catalysts have been developed, showing good activities and selectivity and with the potential to replace Raney nickel (Sánchez-Bastardo et al. 2018). The catalytic activity of the noble metals decreases in the order Ru>Ni>Rh>Pd, with Ru catalysts showing high activities especially in the neutral and basic range.

Ru has been preferred due to its ability to effectively hydrogenate ketones into alcohols under relatively mild reaction conditions in aqueous solutions (Seretis et al. 2020). As is the case for almost every catalyst, some limitations due to undesired reactions and inactivation may occur. Reduced catalyst performance caused by CO adsorption during CO₂ hydrogenation was reported by Xu et al. (2020). A slightly reduced conversion velocity was observed when Ru/C was reused up to 5 times in hydrogenation of 1-methylpyrrole, due to nitrogen adsorption. Surprisingly, during the first reuse, significantly increased activity at room temperature was detected, compared to the fresh catalyst. The authors assigned the phenomenon to the removal of an initial RuO₂ layer (Hegedűs et al. 2018). Especially when using Al₂O₃ as a support, the presence of chlorine significantly reduces the amount of the adsorbed hydrogen, due to selective site blocking. Therefore, one should keep that in mind as well (Lin et al. 2011).

Another interesting issue was demonstrated by Hua et al. (2021). In their work, the influence of the catalyst supporting material on the product composition of 2,5-hexanedione hydrogenation was investigated. Dependent on the support structure, different reaction routes were favored, resulting in different products, both using Ru as a catalyst (Hua et al. 2021).

Attempts were recently made to reduce the resources and energy consumption within the xylitol production process. In the one-pot method proposed by Yi and Zhang (2012), hydrolysis and the subsequent hydrogenation of xylose-containing hemicellulose is completed within a single reaction step. The reaction requires the addition of an acid and a metal catalyst. The absence of acid in the reaction mixture led to significantly reduced xylitol yields.

The challenging handling of the easily diffusing and inflammable hydrogen gas, as well as the fact that it is mostly generated from fossil raw materials, provide reasons for its replacement by less problematic hydrogenating agents, which is a promising opportunity to create greener, more sustainable, efficient, and safe process conditions. Catalytic transfer hydrogenation (CTH) with low molecular mass organic molecules (e.g., alcohols or acids), especially if originating from bio-based sources, are particularly suitable for that purpose. Espro et al. (2018) described the use of short-chain alcohols and formic acid as appropriate H-donors in transfer hydrogenation reactions. The possibility of transfer hydrogenation of glucose into sorbitol, involving biomass-derived alcohols was demonstrated by Garcia et al. (2019, 2021). However, the problematic Ni-containing catalysts were used in the experiments. An advancement in the reduction of different functional organic compounds such as nitroarenes, olefins, and carbonyl compounds was recently presented by Goyal et al. (2023). In their work, methanol was used as a hydrogen source in the presence of a commercially available Pt/C catalyst.

Besides the alcohols, formic acid has been successfully applied as an H-donor in CTH for the reduction of sugar derivatives or various biomass-derived molecules (Jicsinszky and Iványi 2001; Wang et al. 2021; Sultana et al. 2023). The reactions require relatively mild conditions and exclude the risk of handling hydrogen gas. Commercial hydrogenation technology usually requires large investments in hydrogen gas infrastructure and safety equipment. On this account formic acid displays high potential for future applications, especially for small decentralized bio-refinery approaches. Formic acid is formed during biomass conversion, e.g. in the course of kraft pulping, the formation of furfural or catalytic biomass oxidation (Reichert and Albert 2017; Bulushev and Ross 2018; Preuster and Albert 2018; Valentini et al. 2019). Formic acid can also be produced from CO₂, which makes it a potentially carbon neutral reactant. However, the use of formic acid for transfer hydrogenation still requires metal catalysts. Noble metal catalysts, such as Ru, Au, Pd, and Pt, are preferentially used, due to their high activity in hydrogenation of different functional groups (Gilkey and Xu 2016).

In the literature, a combination of formic acid (FA) with organic/inorganic bases was shown to increase the efficiency of FA in CTH reactions (Gilkey and Xu 2016). In practice, it results in an increase of the complexity and costs of the process. Low-molecular-weight amines, such as triethylamine (Et₃N), are often chosen, whereby the molar ratio of FA/Et₃N has a considerable effect on the performance of the reaction (Zhou et al. 2012). Adding Et₃N favors the formation of [NEt₃H][CO₂H] and enhances the reactivity of formate over the metal center. CO₂ can then easily be liberated from the formate ion leaving the hydride ion on the surface of the metal. Subsequently, these active metal hydrides or dihydrogen complexes can hydrogenate the substrates (Nie et al. 2021). In Fig. 1, a reaction scheme of catalytic transfer hydrogenation of xylose using formic acid as the H-donor with Ru/C as a catalyst is shown.

Design of experiments (DoE) is an established tool to generate information with minimized experimental effort and is successfully applied in several research fields (Bowden et al. 2019; Almhofer et al. 2023a). Appropriate software supports the optimization of reaction conditions and statistical data evaluation in systems containing multiple variables and facilitates a graphical presentation of the results. A strategically planned and executed experiment can provide a great deal of information about the effects of one or more factors on a response variable. Therefore, it was an ideal solution for the optimization of the CTH reaction conditions to obtain the highest xylitol yield.

A more than fortyfold increase in xylitol production in the recent forty years illustrates the growing demand for the low-calorie sucrose substitute (Delgado Arcaño et al. 2020). Although d-xylitol is derived from cheap and abundant lignocellulosic biomass, its production process is expensive due to energy-intensive steps and the need for H₂ as the hydrogenation agent.

Herein, an alternative method of d-xylitol production via catalytic transfer hydrogenation (CTH) of xylose without the use of an undesired Ni-catalyst was investigated.

Fig. 1. Reaction scheme of catalytical transfer hydrogenation of xylose using formic acid as an H-donor and Ru/C catalyst

EXPERIMENTAL

Materials

The chemicals used in the catalytic transfer hydrogenation experiments are listed in Table 1.

Table 1. Chemicals Used in the HTC Experiments

Ru/HTC and Ru/Al₂O₃ catalysts were kindly provided by Heraeus (Hanau, Germany). All the other catalysts tested were commercially available and were purchased from Thermo scientific and Sigma Aldrich. A compilation of the catalysts used is displayed in Table 2.

Chemicals used for the analyses described in the corresponding section were of analytical grade. More detailed information may be found in the referred publications.

Design Expert® Software (version 13.0.4.0, Stat-Ease Inc., Minneapolis, MN, USA) was used to design the optimization experiment, statistical analysis, and the resulting model graph creation.

Table 2. Catalysts Used in the Experiments and the Related Metal Content

Methods

Analytical methods

Quantification of d-xylose and d-xylitol was done by HPLC (anion exchange chromatography with pulsed amperometric detection) using a Dionex ICS 5000+ system with a Dionex CarboPac SA10 4×50 mm as pre-column and a Dionex CarboPac SA10 4×250 mm separation column (Almhofer et al. 2023b).

Xylonic acid concentration was determined by ion exchange chromatography with a Dionex CarboPac SA10 4 × 50 mm as pre-column and a Dionex CarboPac SA10 4 × 250 mm column (Wolfsgruber et al. 2023).

The furfural and furfuryl alcohol concentrations were quantified with reversed phase HPLC with UV detection using a Thermo BDS Hypersil C8 250 x 4.6 mm 5 μL column, as described by Almhofer et al. (2023b). The analysis of formate was performed with anion exchange chromatography with conductivity detection and external calibration. A Dionex IonPac AS11-HC 4×250 mm pre-column and a Dionex IonPac AS11-HC 4×250 mm separation column were used in a system described by Almhofer et al. (2023b).

The pH values of the solutions were measured before the CTH experiments using a HI1230 pH electrode connected to a HI83141 pH meter (both Hanna Instruments GmbH, Graz, Austria).

Catalyst screening

The catalyst screening experiments were carried out in 450 mL Parr reactors (Stainless steel/T316) at constant pressure of 5 bar N₂, a temperature of 140 °C, and a constant stirring speed of 320 rpm. The stirring speed was chosen based on preliminary tests in which various catalysts were stirred in the reaction medium at room temperature. Stirring speed was chosen as the minimum speed at which the catalysts were homogeneously suspended. This was visually assessed in a transparent plastic reactor having the same size and geometry as the reaction vessels used in the main experiments. The temperature of 140 °C was chosen based on preliminary tests (data not shown).

To around 30 mL of deionized water in a 100 mL volumetric flask, 1.5 g FA, 0.6 g Et₃N and 5 g d-xylose were added and filled up with deionized water. 50 mL of the obtained reaction mixture was volumetrically transferred to the reactor. The heterogeneous catalyst was added in an amount of 1.5 % w/w to xylose. The reactor was twice evacuated and flushed with N₂ to remove air and the pressure was adjusted to 5 bar. After air removal, the reactor was heated, and the temperature was kept constant for 6 hours. Samples were taken after 3 hours through a cooling coil attached to the bottom drain valve. Before sampling, the coil was purged with double the volume of the sampling system. After the desired reaction time, the reactor was cooled with iced water. The catalyst was filtered off and washed with deionized water. The product solution was stored at 5 °C until analysis.

Optimization of hydrogenation conditions

Experiments involving the two best performing catalysts were designed using Design Expert® Software for the purpose of maximizing the xylitol yield by optimization of the reaction conditions. A Box-Behnken experimental design was suggested by the software to evaluate the influence of three parameters: temperature T, formic acid to xylose ratio FA:Xyl, and triethylamine to formic acid ratio Et₃N:FA. The investigated ranges of T, FA:Xyl, and Et₃N:FA were 70 to 130 °C, 1 to 5, and 0 to 0.4 respectively. The center point (T = 100 °C, FA:Xyl = 3, and Et3N:FA = 0.2) was replicated three times to reveal the reproducibility. Xylose conversion (X_Xyl), xylitol yield (Y_XylOH), and xylitol selectivity (S_XylOH) were chosen as responses. The response data was transformed using the arcsine square root function to limit the results to the physical boundaries of 0 and 1. Design Expert statistical software was used for the analysis of variance (ANOVA) and model evaluation. All CTH experiments were performed similar to catalyst screenings, with a reduced time of 3 h and the conditions given by the experimental design program.

Equations

The equations of xylitol yield calculation (Eq. 1) and the selectivity of xylose conversion into xylitol (Eq. 2) were as follows:

where c_t is the concentration after the reaction (gL^-1); c₀is the initial concentration (gL^-1), and M is the molar mass of the component (gmol^-1).

RESULTS AND DISCUSSION

Preliminary Catalyst Screening

Six commercially available Ru and Pd catalysts on various support materials were screened for their ability to convert xylose into xylitol via CTH using formic acid as an H-donor. Concentrations of the reactants and additives were chosen similar to those typically reported in the literature for analogous CTH reactions (Jicsinszky and Iványi 2001; Sánchez-Bastardo et al. 2018; García et al. 2021). To assess the catalyst performances, xylitol, residual xylose, and furans (sum of furfural and furfuryl alcohol were analyzed in the reaction solutions. Figure 1 shows the product yields dependent on the catalyst used. Most catalysts produced xylitol in relatively low yields between 4.1% and 22.7%. Interestingly, with the Ru/Al₂O₃ catalyst from Thermo Scientific a high amount of the original xylose (32.0%) was found, indicating insufficient catalytic activity under the applied conditions. A small amount of xylose (1.4%) was also found in the residual solution of the Pd/Al₂O₃ catalyst, whereas all other catalysts displayed complete conversion of xylose. The different behavior of both Ru/Al₂O₃ catalysts should be pointed out. Despite similar composition, a marked difference in the performance and product composition was found. A possible explanation may be the individual particle size and structure of the supporting material or distinct production processes (Jędrzejczyk et al. 2020; Vilcocq et al. 2021).

Samples taken during the reaction revealed that xylose conversion was completed within 3 h. Therefore, in the subsequent optimization experiments, the reaction time was reduced to prevent xylitol degradation.

Another observed issue was the formic acid consumption rate exceeding the xylose conversion rate. The resulting lack of the hydrogenation agent led to unpredictable side reactions and the additional change of the pH value, favoring various isomerization and degradation reactions of sugars. High pH values (≥ 8) favor isomerization of xylose in xylulose and the formation of xylonic acid (Yadav et al. 2012). Therefore, an excess of formic acid was used in the subsequent experiments.

Furfural is known to form during acid catalyzed dehydration of xylose at high temperatures, and furfuryl alcohol may be produced via hydrogenation of furfural; hence, their concentrations were determined (Wang et al. 2021). Furans were only detected in low amounts (<1 %), except for the Ru/Al₂O₃ catalyst from Thermo scientific, where 11.9 % furfural and 1.0 % furfuryl alcohol were found. Generally, only a small fraction of the reaction products could be identified with standard methods used in carbohydrate characterization. That fact indicates complex sugar and furan degradation, due to the harsh reaction conditions. This assumption was supported by the dark colors of the residual solutions, which is typical for high molar mass polymerization products. Although the identification of the degradation products would be beneficial and may contribute to the understanding of the reaction, it has not been covered within the current project, as it requires time-consuming method development.

Fig. 2. Yields of xylitol, residual xylose, and furans for six commercially available catalysts in the CTH of xylose using formic acid as H-donor. The catalyst screening was done at a temperature of 140 °C for 6 h, with 50 g L⁻¹ xylose, 15 g L⁻¹ FA, 6 g L⁻¹ Et₃N, and a catalyst loading of 1.5 % w/w on xylose.*Thermo scientific **Hereaus

Optimization of Reaction Conditions by DoE

Based on the catalyst screening results, Ru/C (Thermo scientific) and Ru/Al₂O₃ (Heraeus) were chosen to further investigate the influence of the reaction conditions and to maximize the resulting xylitol yield. An experimental design approach was applied, compiling a set of 15 runs with varying conditions for each catalyst. Temperature (Factor A) was chosen as one of the impact factors, as its reduction generally contributes to the energy balance and concurrently to process efficiency. Further, to minimize undesired side reactions and incomplete xylose conversion due to a reduced hydrogen availability, the concentrations of formic acid and triethylamine were adapted. Enhanced FA/Xyl ratios of 1 to 5 (Factor B) were tested, to ensure a surplus of formic acid during the entire reaction time. Additionally, the influence of different amounts of triethylamine on the expected xylitol yield was investigated as the ratio of formic acid to triethylamine (Factor C:FA/Et₃N). Considering the catalyst screening results, the reaction time was reduced to 3 h, to minimize the effect of xylitol degradation occurring at high temperatures and acidic conditions.

In Fig. 2, the response surfaces of the selectivity (a, b) and xylitol yield (c, d) obtained after data evaluation of the Ru/C experiments are shown. Comparing the diagrams, the influence of Et₃N addition on both response values was quite pronounced. Without Et₃N, the xylitol yields remained low (0 to 3.8 %), whereas at C = 0.4 actual yields of up to 73.2% were obtained with high selectivity and complete xylose conversion (Table 3). At the lowest temperature applied (70 °C) there was no d-xylitol formation. The highest d-xylitol yield and selectivity were obtained in experiment 15 (A = 100 °C, B = 5 eq. FA/Xyl, and C = 0.4).

The results of the experiments with C = 0.2 met the expectations placing in between those of the minimum and maximum C values (graphs not shown). According to the model, a further increase of xylitol yield appears possible with sufficiently high availability of H donor at temperatures slightly below 130 °C.

The reproducibility was calculated from the triplicates of the center point. An average of 56 ± 1.5 % resulted for both xylitol yield and selectivity, as xylose was entirely consumed during the reaction.

Fig. 3. Response surfaces after DoE using Ru/C as a catalyst. Selectivity and xylitol yield in dependence of Et₃N concentration are shown. Figures (a) and (c) illustrate selectivity and xylitol yield without the addition of Et₃N, figures (b) and (d) show the results with a Et₃N:FA ratio of 0.4.

The model related equations describe the behavior of the response variables within the boundaries and may be used to make predictions about the response for given levels of each factor.

Table 3. Parameters of the Experimental Runs and the Resulting Response Values for the CTH with Ru/C

The statistical analysis of variance (ANOVA) and diagnostics were performed using Design Expert® software. The results are presented in Table 4. The evaluation of the responses Y(Xylitol) in a significant model, A, B, C, AB, AC, BC, A², B² and C² as the significant model terms (p-values < 0.05) and a non-significant Lack of Fit values. Similar results were obtained for S(Xylitol) with A, B, C, AB, AC, BC, A² and C² as the significant model terms.

The correlation coefficients between predicted and actual values of d-xylitol yield and selectivity experiments are listed in Table 5. High coefficient of determination indicates a higher reliability of the relationship between predicted and experimentally determined values. If the difference between the predicted and the adjusted R² value is less than 0.2, the agreement is reasonable. A larger difference for d-xylitol yield may indicate a large block effect or a problem with the model or the provided data. Consideration must be given to model reduction, response transformation and outlier values. C.V. (coefficient of variation) describes the variation of the data and represents the accuracy and validity of the results. In this case, the C.V. of S(Xylitol) is significantly lower, compared to C.V. of Y(Xylitol), indicating less variation of the mean. A signal to noise ratio (Adeq. Precision) > 4 is indicating an adequate signal. The actual values by far exceed that level.

The highest d-xylitol yield of 36.2% was obtained with the Ru/Al₂O₃ catalyst at 130 °C, FA to d-xylose ratio of 3 and base to FA ratio 0.4. Similar to the Ru/C experiments, very low d-xylitol yields were obtained at 70 °C, demonstrating the necessity of Et₃N addition independent of the catalyst used.

Table 4. Analysis of Variance (ANOVA) for Quadratic Model with Arcsin Transformation for Ru/C Catalyst

Table 5. Results of the Statistical Analysis of the Models for Y(Xylitol) and S(Xylitol) with Ru/C as Catalyst

According to the evaluated model, a further increase of xylitol formation appears possible with sufficiently high availability of H donor and temperatures over 130 °C. However, hereby the maximum selectivity at around 110 °C should be considered. Again, the results of the experiments with C = 0.2 were found in between those of the minimum and maximum C values (graphs not shown). The average of xylitol yield calculated form the triplicates at the center point conditions resulted in 15 ± 1.1 %, indicating a good reproducibility of the experiments. Interestingly, a selectivity of 52 ± 5.2 % was found, which was in the similar range to the value obtained with Ru/C.

Fig. 4. Response surfaces after DoE using Ru/Al₂O₃ as a catalyst. Selectivity and xylitol yield in dependence of Et₃N concentration are shown. Figures (a) and (c) illustrate selectivity and xylitol yield without the addition of Et₃N, figures (b) and (d) show the results with a Et₃N:FA ratio of 0.4.

Table 6. Parameters of the Experimental Runs and the Resulting Response Values for the CTH with Ru/Al₂O₃

Table 7. Analysis of Variance (ANOVA) for Quadratic Model with Arcsin Transformation for Ru/Al₂O₃ Catalyst

Table 8. Results of the Statistical Analysis for the DoE with Ru/Al₂O₃ Catalyst

The d-xylitol yield and selectivity may be calculated using following equations:

The statistical analysis of variance (ANOVA) is shown in Table 7 and Table 8. Again, the evaluation of the responses Y(Xylitol) and S(Xylitol) resulted in a significant model and a non-significant Lack of Fit value. In case of Y, the significant model terms were A, B, C, AB, AC, BC and A², whereas A, B, C, AC, BC, A², B² and C² turned out to be significant for S.

High reliability of the relationship between predicted and experimentally determined values was attested, as the difference between the predicted and the adjusted R² value was < 0.2. C.V. (coefficient of variation) of S(Xylitol) and Y(Xylitol) were of similar magnitude, indicating comparable data accuracy and validity. A signal to noise ratio (Adeq. Precision) by far exceeded the minimum value of 4.

In summary, the results of the DoE experiments provide a good overview of various influencing factors on the CTH conversion from d-xylose to d-xylitol with a consistently good reproducibility and correspondence between the measuring point and the model. The tested factors temperature, formic acid/d-xylose ratio and triethylamine/formic acid ratio had a significant influence on the reaction. In particular, the experiments without triethylamine resulted in a rather low level of xylitol yield, confirming the necessity of a base addition, regardless the catalyst used. According to literature, the addition of basic components results in increased selectivity, improved reaction kinetics and reduced catalyst deactivation (Nie et al. 2021). Further, the use of high amounts of the H donor may be derived after data evaluation, however too acidic conditions may as well affect catalyst stability (leaching) or even lead to its deactivation, or result in high amount of undesired side products.

Xylose Loss Reactions

During the optimization experiment, d-xylose conversion increased with increasing temperature and the amount of formic acid as well as with enhanced concentration of triethylamine. In some runs xylose was completely consumed, but at the same time, the yields of xylitol were far from reaching the maximum. Therefore, a more detailed analysis of the residual liquors of the experiments with the highest d-xylitol yields was performed for both catalysts (Table 9). There was no residual xylose found in the final reaction medium. However, only 73% (Ru/C) and 36% xylitol yield (Ru/Al₂O₃) was detected respectively, indicating dissimilarly pronounced by-product formation. In case of Ru/C, xylonic acid made up the remaining 27 % of the products. Xylose itself may act as an H-donor, resulting in disproportionation into xylitol and xylonic acid (Mikkola et al. 2000). Preferably, this reaction route should be avoided, as it leads to considerable xylitol yield loss.

The use of Ru/Al₂O₃ as a catalyst resulted in significantly lower xylonic acid formation (6%), but also in half the xylitol yield compared to Ru/C. The 57% of the reaction products remained unidentified. Low pH values favor dehydration of d-xylose to furfural, which can be further hydrogenated to furfuryl alcohol and tetrahydrofuryl alcohol. However, none of these was determined in the solution. Further, the solutions were tested negative for the presence of levulinic acid, which may be formed as a furfuryl alcohol degradation product.

Table 9. Composition of the Product Solutions with the Highest Xylitol Yields after the CTH of Xylose with Formic Acid

The relatively harsh reaction conditions led to complex sugar and furan degradation and high molar mass polymerization products. The identification of the entity of degradation products would require time-consuming method development, which was not possible within the current project. However, in future, such studies may contribute to a more detailed understanding of the reactions.

When comparing both catalysts, Ru/C seemed to be highly active in forming metal hydride species. This high activity on the contrary led to the formation of xylonic acid via H-donation from xylose. The Ru/Al₂O₃ catalyst on the other hand showed less activity, thus leading to sugar degradation and low yields. Based on literature concerning Ru catalysts on alternating support materials, it is speculated that this difference is caused by the chemical nature and the morphology of the carrier, as it interacts with the metallic Ru (Vilcocq et al. 2021). Due to its amphoteric character, Al₂O₃ may interact with the species involved in the hydrogenation, changing the reaction rate and the overall reaction yield. The role of the catalyst support was confirmed in studies on hydrogenation reaction systems (Filonenko et al. 2016). König et al. (2014) described the different catalyst inactivation by sulphur containing components in dependence of the support properties. One should keep that in mind, especially when processing pulping derived streams, as sulphur is omnipresent in that field.

Generally, catalytic transfer hydrogenation with external H donors is a chemically efficient, environmentally friendly alternative to conventional hydrogenation. The process avoids several serious drawbacks associated with the use of hydrogen gas. CTH has attracted interest due to its advantageous properties, such as achieving high atomic yields under mild reaction conditions, lower energy intensity, and the use of H donors from renewable sources. Although yields are low compared to conventional hydrogenation of xylose, it must be pointed out that this is the first report on the CTH of xylose using formic acid. On the contrary, xylitol production using H₂ is done at industrial scale for decades and has been extensively optimized during this period. Thus, it is speculated that CTH yields will increase with future investigation, e.g. due to catalyst development.

Using formic acid as a hydrogen source results in an inevitable release of CO₂. It may be recycled to formic acid by hydrogenation; however, when fossil-based H₂ is used, the advantage of formic acid diminishes compared to direct hydrogenation. On the other hand, the resulting CO₂ is pure and may be easily removed. It may be directly used in CO₂ binding reactions or as a supercritical solvent, as the technology has been gaining attention in recent years (Munshi and Bhaduri 2010; Liu et al. 2015; Liu et al. 2022). Xylose and formic acid may be obtained as side-streams of biomass processing, e.g. kraft pulp production, and the processing to an attractive and valuable-added co-product would push pulp mills further towards wood biorefinery.

To live up to the idea of sustainability, a few words should be said on the recycling of additives and catalyst. The amine base is not consumed during CTH and thus needs to be separated from the product solution and recycled back to the process. Due to the large differences in physical properties compared to xylitol, this can be easily done, e.g. by distillation or liquid-liquid extraction. The catalyst remains solid during the reaction and may be easily recovered by filtration and reused. Literature describes poisoning of Ru with Cl, CO, and N-containing aromatics; however, they present less of a problem in the environment of a pulp mill. Sulphur was found to interact with the catalyst support, especially when activated carbon is used, resulting in diminished reactivation ability (König et al. 2014). That may be challenging, due to the presence of sulfur compounds in pulp production sites. In order to determine the maximum catalyst lifetime, the investigation of sensitivity to leaching and the additives under CTH conditions would be beneficial, especially when using components originating from a kraft mill.

Nevertheless, the best practice for producing xylitol is to use educts in highest possible purity, as the separation of the various co-products from xylitol may be challenging, compared to the established xylose separation techniques. Due to the fact that xylitol has been an established product for decades, numerous established purification methods may be applied, depending on the production process and the resulting concentration (Martínez et al. 2015). Crystallisation is a widely used xylitol purification method. By adjusting the crystallization parameters to the particular product composition, various impurities can be efficiently removed.

CONCLUSIONS

The production of xylitol via catalytical transfer hydrogenation of xylose, a sugar derived from lignocellulosic biomass, was shown to be feasible. The employment of formic acid as a bio-based H-donor improves the process sustainability.
Xylitol yield was significantly increased by the addition of triethylamine, due to the enhanced H-donor activity of the formate.
The catalysts Ru/C (Thermo scientific) and Ru/Al₂O₃ (Heraeus) showed the best performance in the preliminary tests and led to effective maximum xylitol yields of 73% and 36% under optimized conditions

ACKNOWLEDGMENTS

Financial support was provided by the Austrian government, the provinces of Lower Austria, Upper Austria, and Carinthia, as well as by Lenzing AG. We also express our gratitude to the Johannes Kepler University, Linz, the University of Natural Resources and Life Sciences (BOKU), Vienna, and Lenzing AG for their in-kind contributions.

We would like to extend our thanks to the Institute of Chemistry and Technology of Biobased Systems, Graz University of Technology, Graz for the constructive cooperation.

REFERENCES CITED

Almhofer, L., Paulik, C., Bischof, R. (2023a). “Simultaneous posthydrolysis and liquid-liquid extraction: A SIMPLLE process to detoxify eucalyptus prehydrolysis liquor,” Biomass Conversion and Biorefinery. DOI: 10.1007/s13399-023-04570-6

Almhofer, L., Bischof, R. H., Madera, M., and Paulik, C. (2023b). “Kinetic and mechanistic aspects of furfural degradation in biorefineries,” The Canadian Journal of Chemical Engineering 101(4), 2033-2049. DOI: 10.1002/cjce.24593

Bowden, G. D., Pichler, B. J., and Maurer, A. (2019). “A design of experiments (DoE) approach accelerates the optimization of copper-mediated ¹⁸F-fluorination reactions of arylstannanes,” Scientific Reports 9(1), article 11370. DOI: 10.1038/s41598-019-47846-6

Bozell, J. J., and Petersen, G. R. (2010). “Technology development for the production of biobased products from biorefinery carbohydrates—the US Department of Energy’s “Top 10” revisited,” Green Chemistry 12(4), 539-554. DOI: 10.1039/B922014C

Bulushev, D. A., and Ross, J. R. H. (2018). “Towards sustainable production of formic acid,” ChemSusChem, 11(5), 821-836. DOI: 10.1002/cssc.201702075

Clements, L. D., and Van Dyne, D. L. (2005). “The lignocellulosic viorefinery – A strategy for returning to a sustainable source of fuels and industrial organic chemicals,” Biorefineries‐Industrial Processes and Products 115-128.

Dean, B., Dodge, T., Valle, F., and Chotani, G. (2005). “Development of biorefineries–technical and economic considerations,” Biorefineries‐Industrial Processes and Products 67-83. DOI: 10.1002/9783527619849.ch3

Delgado Arcaño, Y., Valmaña García, O. D., Mandelli, D., Carvalho, W. A., and Magalhães Pontes, L. A. (2020). “Xylitol: A review on the progress and challenges of its production by chemical route,” Catalysis Today 344, 2-14. DOI: 10.1016/j.cattod.2018.07.060

Espro, C., Gumina, B., Szumelda, T., Paone, E., and Mauriello, F. (2018). “Catalytic transfer hydrogenolysis as an effective tool for the reductive upgrading of cellulose, hemicellulose, lignin, and their derived molecules,” Catalysts 8(8), 313. DOI: 10.3390/catal8080313

Felipe Hernández-Pérez, A., de Arruda, P. V., Sene, L., da Silva, S. S., Kumar Chandel, A., and de Almeida Felipe, M. d. G. (2019). “Xylitol bioproduction: State-of-the-art, industrial paradigm shift, and opportunities for integrated biorefineries,” Critical Reviews in Biotechnology 39(7), 924-943. DOI: 0.1080/07388551.2019.1640658

Filonenko, G. A., Vrijburg, W. L., Hensen, E. J. M., and Pidko, E. A. (2016). “On the activity of supported Au catalysts in the liquid phase hydrogenation of CO₂ to formates,” Journal of Catalysis 343, 97-105. DOI: 10.1016/j.jcat.2015.10.002

García, B., Moreno, J., Iglesias, J., Melero, J. A., and Morales, G. (2019). “Transformation of glucose into sorbitol on Raney Nickel catalysts in the absence of molecular hydrogen: Sugar disproportionation vs catalytic hydrogen transfer,” Topics in Catalysis 62(5), 570-578. DOI: 10.1007/s11244-019-01156-3

García, B., Orozco-Saumell, A., López Granados, M., Moreno, J., and Iglesias, J. (2021). “Catalytic transfer hydrogenation of glucose to sorbitol with Raney Ni catalysts using biomass-derived diols as hydrogen donors,” ACS Sustainable Chemistry & Engineering 9(44), 14857-14867. DOI: 10.1021/acssuschemeng.1c04957

Gilkey, M. J., and Xu, B. (2016). “Heterogeneous catalytic transfer hydrogenation as an effective pathway in biomass upgrading,” ACS Catalysis 6(3), 1420-1436. DOI: 10.1021/acscatal.5b02171

Goyal, V., Bhatt, T., Dewangan, C., Narani, A., Naik, G., Balaraman, E., Natte, K., and Jagadeesh, R. V. (2023). “Methanol as a potential hydrogen source for reduction reactions enabled by a commercial Pt/C catalyst,” The Journal of Organic Chemistry 88(4), 2245-2259. DOI: 10.1021/acs.joc.2c02657

Gschwandtner, C. (2022). “Outlook on global fiber demand and supply 2030,” Lenzinger Berichte 97, 11-19.

Hegedűs, L., Szőke-Molnár, K., Sajó, I. E., Srankó, D. F., and Schay, Z. (2018). “Poisoning and reuse of supported precious metal catalysts in the hydrogenation of N-heterocycles. Part I: Ruthenium-catalysed hydrogenation of 1-methylpyrrole,” Catalysis Letters 148(7), 1939-1950. DOI: 10.1007/s10562-018-2406-4

Heikki, H., Johanna, N., Marja-Leena, S., Håkan, G., Olli-Pekka, E., Julita, P., and Tammy, P. (1999). “Crystallization of xylitol, crystalline xylitol product and use thereof,” WO Patent 9959426

Jędrzejczyk, M., Soszka, E., Goscianska, J., Kozanecki, M., Grams, J., and Ruppert, A. M. (2020). “The influence of carbon nature on the catalytic performance of Ru/C in levulinic acid hydrogenation with internal hydrogen source,” Molecules 25(22), 5362. DOI: 10.3390/molecules25225362

Jicsinszky, L., and Iványi, R. (2001). “Catalytic transfer hydrogenation of sugar derivatives,” Carbohydrate Polymers 45(2), 139-145. DOI: 10.1016/S0144-8617(00)00319-2

König, C. F. J., Schuh, P., Huthwelker, T., Smolentsev, G., Schildhauer, T. J., and Nachtegaal, M. (2014). “Influence of the support on sulfur poisoning and regeneration of Ru catalysts probed by sulfur K-edge X-ray absorption spectroscopy,” Catalysis Today 229, 56-63. DOI: 10.1016/j.cattod.2013.09.065

Lin, B., Wang, R., Lin, J., Ni, J., and Wei, K. (2011). “Effect of chlorine on the chemisorptive properties and ammonia synthesis activity of alumina-supported Ru catalysts,” Catalysis Letters 141(10), 1557-1568. DOI: 10.1007/s10562-011-0658-3

Liu, X.-F., Zhang, K., Tao, L., Lu, X.-B., and Zhang, W.-Z. (2022). “Recent advances in electrochemical carboxylation reactions using carbon dioxide,” Green Chemical Engineering 3(2), 125-137. DOI: 10.1016/j.gce.2021.12.001

Liu, Q., Wu, L., Jackstell, R., and Beller, M. (2015). “Using carbon dioxide as a building block in organic synthesis,” Nature Communications 6(1), article 5933. DOI: 10.1038/ncomms6933

Martínez, E. A., Canettieri, E. V., Bispo, J. A. C., Giulietti, M., de Almeida e Silva, J. B., and Converti, A. (2015). “Strategies for xylitol purification and crystallization: A review,” Separation Science and Technology 50(14), 2087-2098. DOI: 10.1080/01496395.2015.1009115

Melaja, A. J., Hamalainen, L., and Heikkila, H. (1977). “Manufacture of xylose and xylitol,” (Suomen Sokeri Oy, Finland). Application: CA, 45 pp.

Mikkola, J. P., Salmi, T., Sjöholm, R., Mäki-Arvela, P., and Vainio, H. (2000). “Hydrogenation of xylose to xylitol: three-phase catalysis by promoted raney nickel, catalyst deactivation and in-situ sonochemical catalyst rejuvenation,” Studies in Surface Science and Catalysis, A. Corma, F. V. Melo, S. Mendioroz, and J. L. G. Fierro, (eds.), Elsevier, pp. 2027-2032. DOI: 10.1016/S0167-2991(00)80766-1

Munshi, P., and Bhaduri, S. (2010). “ChemInform abstract: Supercritical CO₂ — A twenty-first century solvent for the chemical industry,” ChemInform 41(11). DOI: 10.1002/chin.201011250

Nie, R., Tao, Y., Nie, Y., Lu, T., Wang, J., Zhang, Y., Lu, X., and Xu, C. C. (2021). “Recent advances in catalytic transfer hydrogenation with formic acid over heterogeneous transition metal catalysts,” ACS Catalysis 11(3), 1071-1095. DOI: 10.1021/acscatal.0c04939

Preuster, P., and Albert, J. (2018). “Biogenic formic acid as a green hydrogen carrier,” Energy Technology 6(3), 501-509. DOI: 10.1002/ente.201700572

Reichert, J., and Albert, J. (2017). “Detailed kinetic investigations on the selective oxidation of biomass to formic acid (OxFA process) using model substrates and real biomass,” ACS Sustainable Chemistry & Engineering 5(8), 7383-7392. DOI: 10.1021/acssuschemeng.7b01723

Sánchez-Bastardo, N., Delidovich, I., and Alonso, E. (2018). “From biomass to sugar alcohols: Purification of wheat bran hydrolysates using boronic acid carriers followed by hydrogenation of sugars over Ru/H-ZSM-5,” ACS Sustainable Chemistry & Engineering 6(9), 11930 – 11938. DOI: 10.1039/D0GC03931D

Seretis, A., Diamantopoulou, P., Thanou, I., Tzevelekidis, P., Fakas, C., Lilas, P., and Papadogianakis, G. (2020). “Recent advances in ruthenium-catalyzed hydrogenation reactions of renewable biomass-derived levulinic acid in aqueous media,” Frontiers in Chemistry 8. DOI: 10.3389/fchem.2020.00221

Sultana, A., Lomate, S., and Fujitani, T. (2023). “Highly active MgO catalysts for hydrogenation of levulinic acid to γ-valerolactone using formic acid as the hydrogen source,” Frontiers in Energy Research 11, 1133514. DOI: 10.1039/D1RA05804E

Valentini, F., Kozell, V., Petrucci, C., Marrocchi, A., Gu, Y., Gelman, D., and Vaccaro, L. (2019). “Formic acid, a biomass-derived source of energy and hydrogen for biomass upgrading,” Energy & Environmental Science 12(9), 2646 – 2664. DOI: 10.3389/fenrg.2023.1133514

Vilcocq, L., Paez, A., Freitas, V. D. S., Veyre, L., Fongarland, P., and Philippe, R. (2021). “Unexpected reactivity related to support effects during xylose hydrogenation over ruthenium catalysts,” RSC Advances 11(62), 39387-39398. DOI: 10.1039/D1RA08193D

Wang, T., Du, J., Sun, Y., Tang, X., Wei, Z.-J., Zeng, X., Liu, S.-J., and Lin, L. (2021). “Catalytic transfer hydrogenation of biomass-derived furfural to furfuryl alcohol with formic acid as hydrogen donor over CuCs-MCM catalyst,” Chinese Chemical Letters 32(3), 1186-1190. DOI: 10.1016/j.cclet.2020.07.044

Wolfsgruber, M., Rodrigues, B. V. M., da Cruz, M. G., Bischof, R. H., Budnyk, S., Beele, B., Monti, S., Barcaro, G., Paulik, C., and Slabon, A. (2023). “Electrocatalytic reduction of aldonic acids to aldoses on gold electrodes,” ACS Sustainable Chemistry & Engineering 11(1), 312-321. DOI: 10.1021/acssuschemeng.2c05576

Xu, S., Chansai, S., Xu, S., Stere, C. E., Jiao, Y., Yang, S., Hardacre, C., and Fan, X. (2020). “CO poisoning of Ru catalysts in CO₂ hydrogenation under thermal and plasma conditions: A combined kinetic and diffuse reflectance infrared Fourier transform spectroscopy–mass spectrometry study,” ACS Catalysis 10(21), 12828-12840. DOI: 10.1021/acscatal.0c03620

Yadav, M., Mishra, D. K., and Hwang, J.-S. (2012). “Catalytic hydrogenation of xylose to xylitol using ruthenium catalyst on NiO modified TiO₂ support,” Applied Catalysis A: General 425-426, 110-116. DOI: 10.1016/j.apcata.2012.03.007

Yi, G., and Zhang, Y. (2012). “One-pot selective conversion of hemicellulose (xylan) to xylitol under mild conditions,” ChemSusChem 5(8), 1383-1387. DOI: 10.1002/cssc.201200290

Zhou, X., Wu, X., Yang, B., and Xiao, J. (2012). “Varying the ratio of formic acid to triethylamine impacts on asymmetric transfer hydrogenation of ketones,” Journal of Molecular Catalysis A: Chemical 357, 133-140. DOI: 10.1016/j.molcata.2012.02.002

Article submitted: August 8, 2023; Peer review completed: August 26, 2023; Revised version received and accepted: October 22, 2023; Published: October 31, 2023.

DOI: 10.15376/biores.18.4.8631-8652