Kinetic study of hydrolytic hydrogenation of cellulose
First, we investigated the time course of the hydrolytic hydrogenation of cellulose using 2 wt % Ru/AC(N) catalyst to reveal the reaction pathways (Figure 1). The reaction was performed at 463 K under an H2 pressure of 0.7 MPa as the optimized conditions as shown in Figure S1 in the Supporting Information. The products are classified into four groups: residual cellulose and oligomers, glucose, sugar alcohols (sorbitol and mannitol), and other by-products. Oligomers are included in residual cellulose in this case, because the separation of both compounds is complicated and virtually useless for the kinetic study (see Scheme 2). The solid lines represent the fitting results based on kinetic equations (see below). The amount of cellulose was gradually decreased with increasing the reaction time, but the conversion rate declined after 3 h and the first-order approximation as shown in Figure S2 did not agree with the experimental results. This result implies that either the nature of the catalyst or cellulose changed during the reaction. Regarding the nature of the catalyst, reuse experiments (Figure S3) indicated that the catalyst was durable without decreasing cellulose conversion or yields of sugar alcohols at least three times. Hence, the catalyst did not change during the reaction. As to the nature of the cellulose, Essayem et al. have demonstrated that the cellulose particles aggregate in the hot-compressed water at 463 K, which may decrease the reactivity of cellulose.14 In our system, the particle size of cellulose was indeed increased from 22 μm to 48 μm in 3 h under the reaction conditions (Figure S4). Note that the quick transformation of amorphous cellulose to cellulose IV was observed by XRD in the first several minutes (Figure S5), a result that does not correlate with the gradual reduction of the hydrolysis rate in several hours. Accordingly, the aggregation of cellulose particles might reduce their reactivity and suppress the cellulose conversion.
Figure 1. Time course of the hydrolytic hydrogenation of cellulose using Ru/AC(N) catalyst at 463 K under an H2 pressure of 0.7 MPa.
Download figure to PowerPoint
The product initially formed was glucose (Figure 1, inset) and the yield was maximized at 1 h (4 %). The amount of glucose decreased to less than 1 % at 3 h, demonstrating that glucose is an intermediate. Subsequently, sugar alcohols were produced in 36 % yield (sorbitol 30 %, mannitol 6 %) within 6 h but diminished to 4.3 % at 48 h. The sugar alcohols underwent decomposition, and the amount of by-products such as ethylene glycol, propylene glycol, and CH4 continuously increased during the reaction. The successive decomposition of sugar alcohols was further studied in the conversion of sorbitol (Table 1). Only 17 % of sorbitol remained under an H2 pressure of 0.7 MPa in the presence of Ru/AC(N) catalyst (entry 1), whereas 82 % and 96 % of sorbitol were recovered in the absence of H2 or catalyst, respectively (entries 2 and 3). These results show that the degradation of sugar alcohols is catalyzed by Ru/AC(N) especially under H2 pressure. As a mechanism of the sorbitol hydrogenolysis, Shanks and co-workers have proposed that the decomposition proceeds through the terminal CC scission by dehydrogenation–decarbonylation.28 They observed the formation of various C2–C6 polyols in the hydrogenolysis of sorbitol using Ru catalyst. In our case, CH4 (25 %), CO2 (9.3 %), ethylene glycol (1.5 %), propylene glycol (3.3 %), glycerol (3.1 %), erythritol (1.6 %), and xylitol (4.7 %) were formed during the reaction (entry 1), which supports the dehydrogenation–decarbonylation mechanism. These results indicate that the hydrogenolysis pathway strongly affects the yield and selectivity of sugar alcohols in the hydrolytic hydrogenation of cellulose.
Table 1. Decomposition of sorbitol by Ru/AC(N) catalyst.[a]
| || ||Recovery and yield [% C]|
The proposed reaction pathway based on the above results is shown in Scheme 2. Original cellulose, shown as celluloseA, is converted to glucose (rate constant: k1) and the less reactive cellulose, denoted as celluloseB, (k0) in parallel. CelluloseB is slowly hydrolyzed to glucose (k1′). The produced glucose is then reduced to sugar alcohols (k2), and decomposition of glucose also occurs (k3) because of its thermal instability. Finally, the sugar alcohols undergo decomposition to by-products (k4). The hydrogenation step usually takes place by the Langmuir–Hinshelwood or the Eley–Rideal mechanism29 but both can be treated as a pseudo-first-order reaction because of the low concentration of glucose (<0.1 g L−1) and excess H2. The hydrolysis of cellulose also roughly follows first-order kinetics.17, 28 Although the reaction orders of other steps were not revealed, we applied pseudo first-order approximation to determine the rate constants that reproduce the actual experimental data [Eqs. (1)–(5)]. The total amount of celluloseA and celluloseB was regarded as the quantity of cellulose. The simulation well reproduced the experimental results as shown in Figure 1. The rate constant of celluloseA hydrolysis (k1=0.27 h−1, Scheme 2) is 3.3 times higher than that without using the catalyst (0.082 h−1),17 indicating that Ru/AC(N) promotes the cellulose hydrolysis. The hydrolysis of celluloseB (k1′=0.050 h−1) is significantly slower than that of celluloseA. These rate constants are 1–2 orders of magnitude smaller than that of glucose hydrogenation (k2=3.5 h−1), and, accordingly, the rate-determining step for the formation of sorbitol is the hydrolysis. In addition, the rate constant of glucose degradation (k3=1.5 h−1) is significantly higher than that of decomposition of sugar alcohols (k4=0.060 h−1), because the aldehyde group of glucose in its linear form readily undergoes many types of reactions. However, the theoretical yield of sugar alcohols is 70 % at 100 % conversion using the ratio of k2/(k2+k3), if k4=0. This assumed value is remarkably higher than the maximum yield in the real experiment (38 %). Hence, the degradation of sugar alcohols is not negligible to obtain good yields of sugar alcohols.(1), (2), (3), (4), (5)
To investigate the effect of k4 on the production of sugar alcohols, two simulations were performed at k4=0 h−1 and 0.06 h−1 (Figure 2 a). These two calculations exhibit a large gap of 61 % yield of sugar alcohols at 48 h. In contrast, the yields in the respective simulations are almost the same at 0–3 h, which indicates that the decomposition of sugar alcohols is limited if the reaction is finished within 3 h. The improvement of k1, the rate-determining step, would be the promising strategy to shorten the reaction time. Therefore, we quantitatively evaluated the effect of k1 value on the yield of sugar alcohols in a computation. By increasing the k1 values as shown in Figure 2 b, the reaction time that gives the maximum yield of sugar alcohols is reduced and the maximum yield is raised up to nearly 70 % as expected.
Figure 2. Simulative yield of sugar alcohols (sorbitol+mannitol) with different a) k4 and b) k1 values. Other rate constants were used as follows; (a) k0=0.10 h−1, k1=0.27 h−1, k1′=0.050 h−1, k2=3.5 h−1, k3=1.5 h−1, (b) k0=0.10 h−1, k1′=0.050 h−1, k2=3.5 h−1, k3=1.5 h−1, k4=0.060 h−1.
Download figure to PowerPoint
Design of an efficient catalytic system for the hydrolytic hydrogenation of cellulose
The kinetic study motivated us to accelerate the cellulose hydrolysis with high yields of sugar alcohols. Possible embodying ways are change of the pretreatment method or addition of mineral acids (Table 2). In this study, reaction conditions of 463 K, 3 h and an H2 pressure of 0.9 MPa were chosen to minimize the degradation of glucose and sugar alcohols. Ru/AC(N) without any modifications gave 51 % conversion of cellulose and 25 % yield of sugar alcohols (entry 4). By-products including sorbitan (0.1 %), ethylene glycol (0.4 %), and propylene glycol (0.6 %) were also obtained in the reaction. Mix-milling method, which is a pretreatment to make good contact between a catalyst and cellulose for accelerating the hydrolysis of cellulose (see Table 2, footnote [f]),31 gave high cellulose conversion (89 %) and yield of sugar alcohols (68 %, entry 5). This yield is the highest ever achieved under low H2 pressures (absolute pressure <1 MPa at room temperature) by solid catalysts without adding soluble components. Amounts of by-products such as sorbitan (0.2 %), ethylene glycol (0.5 %), and propylene glycol (1.1 %) were similar to those provided without the mix-milling pretreatment (entries 4, 5). Thus, the mix-milling method selectively accelerated the hydrolysis step. The turnover numbers per surface Ru (8.5 μmol, determined by extended X-ray absorption fine structure (EXAFS) methods, see Table 4) and bulk Ru for the formation of sugar alcohols (1.3 mmol) were 150 and 129, respectively, which means Ru/AC(N) works as a catalyst. H4SiW12O40 and HCl were also used to accelerate the hydrolysis as they were effective in the hydrolytic hydrogenation of cellulose under high H2 pressure (≥5 MPa).15, 18 In our case, the addition of H4SiW12O40 surely increased the cellulose conversion to 89 %, but the yield of sugar alcohols was not improved (entry 6). Instead, by-products such as sorbitan (4.1 %), ethylene glycol (9.5 %), and propylene glycol (6.2 %) formed. In addition, the solution after the reaction turned to blue, revealing the reduction of W6+ to W5+ species. Zhang and co-workers have found that soluble W species such as tungsten bronze (HxWO3) containing W5+ cleave CC bonds of sugar intermediates.32 The addition of HCl similarly provided a high conversion of cellulose (91 %) and a low yield of sugar alcohols (24 %, entry 7). Cl− or H+ could enhance the degradation of the sugar products.17 Thus, the use of mineral acids is not useful for the selective conversion of cellulose to sugar alcohols under our reaction conditions. Accordingly, we conclude that the mix-milling pretreatment is the best way to gain high yields and selectivity of sugar alcohols in the low-pressure reactions.
Table 2. Acceleration of hydrolysis step on the hydrolytic hydrogenation of cellulose by Ru/AC(N) catalyst.[a]
| || ||Yield [% C]||Conversion[e]|
|Entry||Method||Sorbitol||Mannitol||Sorbitol+ mannitol[b]||Glucose||C2–C6 polyols[c]||Other[d]||[%]|
|4||no modification||17||7.9||25 (50)||0.2||5.9||20||51|
Table 4. Summary of curve fittings for the EXAFS spectra of Ru/AC(N) catalyst in the hydrolytic hydrogenation of cellobiose.[a]
|Sample||CN[b]||R[c] [Å]||σ2 [d] [10−3 Å2]||Ru particle size[e] [nm]|
Characterization of active Ru species during the reaction
We focused on the physical and chemical structure of Ru/AC(N) catalyst before and after the reaction to reveal the active Ru species. N2 adsorption–desorption isotherms of AC(N) support and Ru/AC(N) catalyst were almost of identical type I shape (Figure S6), and they had similar specific surface areas and total pore volumes (Table 3). The morphology of AC(N) was maintained during the catalyst preparation. Ru particles of 1–2 nm were observed by TEM (Figure S7 a), and the average size was 1.4 nm. Notably, the size of Ru was almost retained (1.7 nm) after the mix-milling and the hydrolytic hydrogenation reaction (Figure S7 b). The results is in good agreement with the reusability of the catalyst as already shown in Figure S3.
Table 3. Textural and physicochemical properties of Ru/AC(N) catalyst.
|Sample||SBET[a] [m2 g−1]||VT[b] [cm3 g−1]||Average Ru particle diameter[c] [nm]|
|2 wt % Ru/AC(N)||940||0.89||1.4±0.2|
|2 wt % Ru/AC(N)[d]||n.d.[e]||n.d.[e]||1.7±0.3|
We further performed in situ Ru K-edge X-ray absorption fine structure (XAFS) analysis of Ru/AC(N) to observe the active Ru species during the reaction. Cellobiose and 2-propanol were chosen as a substrate and a hydrogen source, respectively, which were convenient model reagents to handle in a synchrotron facility. Moreover, both 2-propanol and the low-pressure H2 similarly work as the reductant in the hydrolytic hydrogenation by Ru/AC(N), as demonstrated in our previous report.27 In Figure 3, the in situ X-ray absorption near-edge structure (XANES) spectra of Ru/AC(N) catalyst recorded at 313–413 K with the heating rate of 10 K min−1 are shown. RuO2⋅2 H2O [edge energy: 22 125 eV, spectrum (i)] and Ru metal powder [22 117 eV, spectrum (ix)] are also shown as references. Ru/AC(N) catalyst exhibited an edge energy of 22 124 eV with a peak at 22 148 eV at 313 K [spectrum (ii)]. The shape of this spectrum is almost the same as that of RuO2⋅2 H2O, and this result is consistent with our precious report.33 Namely, the Ru species is RuO2⋅2 H2O after the passivation step in the catalyst preparation, because Ru metal particles less than 2 nm are easily oxidized to RuO2⋅2 H2O in air at room temperature. The edge energy of Ru/AC(N) shifted to 22 117 eV at 413 K [spectrum (vii)], and characteristic peaks for Ru metal appeared at 22 143 eV and 22 165 eV. The shape of XANES spectrum did not change any more on extension of the reaction time [spectrum (viii)]. Therefore, RuO2⋅2 H2O is quickly reduced to Ru metal during the heating, and the metallic Ru particles would be the actual catalysts. This result does not conflict with the formation of Ruδ+ observed in the hydrolytic disproportionation of cellobiose under similar reaction conditions except for the absence of 2-propanol.34 The distinction owes to the stronger reducing reagent derived from 2-propanol (this study) than that from reducing terminals of sugars (the previous study). In a related work, Davis and co-workers reported the in situ reduction of oxidized Ru species to Ru metal during the hydrogenation of glucose under an H2 pressure of 4 MPa.35 Consequently, the Ru catalyst under working conditions is the metal. We previously proposed that oxidized Ru species were more active than zero-valent metals for the hydrolysis of cellulose;33 however, the metallic Ru species are still active as described in the kinetic study.
Figure 3. Time course of the in situ Ru K-edge XANES spectra of Ru/AC(N) catalyst in the hydrolytic hydrogenation of cellobiose during heating from 313 K to 413 K with a heating rate of 10 K min−1. The measurement conditions were ii) 313 K, iii) 333 K, iv) 353 K, v) 373 K, vi) 393 K, vii) 413 K, viii) kept 30 min at 413 K. i) RuO2⋅2H2O and ix) Ru metal were also measured at 413 K as Ru standards.
Download figure to PowerPoint
As depicted in Figure 4 (solid line), Fourier transform of an in situ EXAFS spectrum of Ru/AC(N) kept for 30 min at 413 K in the reaction mixture exhibited a peak at 2.3 Å attributed to RuRu shell of Ru metal. The fitting result of Ru/AC(N) shown as a dashed line provided reasonable parameters (Table 4); The RuRu bond length (2.66 Å) is similar to that of standard Ru metal (2.67 Å) and the R factor is as low as 0.69 %. The average particle size of Ru on AC(N) calculated from the coordination number36 is 1.5 nm, which is almost the same as that observed by TEM (1.4 nm). The number of surface Ru atoms is estimated to 170 μmol gcat−1, and the dispersion of Ru particles is 86 %.37 Therefore, highly dispersed Ru metal particles on AC(N) are the active species for the low-pressure hydrolytic hydrogenation of cellulose. Interestingly, Ru particles (9 nm) on Al2O3 are inactive in the reaction under <1 MPa H2,27 whereas this catalyst improves the productivity of sugar alcohols under 5 MPa H2.11 In contrast, Ru/AC(N) catalyst in our work exhibited a negative effect in pressurizing H2 (see Figure S1 b). Moreover, Liu and co-workers investigated the effects of Ru particle size in the cellulose conversion using Ru/C catalyst under an H2 pressure of 6 MPa. They found that the Ru particles of 1.5 nm gave a low selectivity of sugar alcohols (36 %) compared with that with 3–4 nm Ru (56–61 %).12 These results indicate that highly dispersed Ru particles are not preferable under the high-pressure hydrolytic hydrogenation of cellulose but effective for the reaction under low-pressure H2. According to these works, we proposed that highly dispersed small Ru particles might be essential for the efficient production of sugar alcohols from cellulose under low H2 pressure.
Figure 4. Fourier transform of an in situ Ru K-edge EXAFS spectrum of Ru/AC(N) catalyst in the hydrolytic hydrogenation of cellobiose with 2-propanol at 413 K for 30 min. The dashed line represents the fitting result.
Download figure to PowerPoint