Hirokazu
Kobayashi
,
Hidetoshi
Ohta
and
Atsushi
Fukuoka
*
Catalysis Research Centre, Hokkaido University, Kita 21, Nishi 10 Kita-ku, Sapporo, Hokkaido 001-0021, Japan. E-mail: fukuoka@cat.hokudai.ac.jp; Fax: +81 11 706 9139
First published on 6th February 2012
Conversion of lignocellulose into renewable chemicals and fuels has received great attention for building up sustainable societies. However, the utilisation of lignocellulose in the chemical industry has largely been limited to paper manufacturing because of the complicated chemical structure and persistent property of lignocellulose. Heterogeneous catalysis has the potential to selectively convert lignocellulosic biomass into various useful chemicals, and this methodology has rapidly progressed in the last several years. In this perspective article, we outline our recent approaches on the heterogeneous catalysis for this challenging subject with related literatures.
Lignocellulose in woods consists of cellulose (40–50%), hemicellulose (20–40%) and lignin (20–30%).2 Cellulose is a water-insoluble polymer composed of glucose linked by β-1,4-glycosidic bonds (Fig. 1(a)) and forms robust crystal structures with inter- and intra-molecular hydrogen bonds, possessing high chemical stability.3 Hemicellulose is also a polysaccharide, but includes pentoses and hexoses connected by several forms of glycosidic bonds, which results in amorphous structure.4 The components of hemicelluloses are varied depending on plants and their parts, and accordingly they have different names based on the feature, e.g. “glucuronoarabinoxylan” in grasses, consisting of xylose chain with glucuronic acid and arabinose residues, and “arabinan” in beet fibre, bearing a backbone of arabinose (Fig. 1(b)). Lignin is a three-dimensional aromatic polymer generated by the radical polymerisation of p-coumaryl alcohol, coniferyl alcohol and sinapyl alcohol (Fig. 1(c)).5 First of all, we focused on the catalytic conversion of cellulose, accounting for the largest part in lignocellulose and having a simple chemical structure like starch (Fig. 1(d)).
![]() | ||
Fig. 1 Chemical structures of (a) cellulose, (b) hemicellulose (top: glucuronoarabinoxylan, bottom: arabinan), (c) lignin and (d) starch. |
An extensive number of works have been devoted to the hydrolysis of cellulose to glucose,6 and sulfuric acid is known to be a typical catalyst for this reaction; however, this process suffers from the corrosive property of the acid and the product separation. Although cellulase enzymes can be used to selectively convert cellulose to glucose under ambient conditions,7 the driving down of the cost of enzymes is a major issue in this field. Sub- and super-critical water has been applied for the hydrolysis of cellulose,8 but the low product selectivity can be pointed out as a problem. In contrast, heterogeneous catalysis has the potential to overcome these problems.9 In this scope, the catalytic gasification of cellulose to syngas and to pure hydrogen was reported,10 but the direct production of ≥C2 compounds would be more effective in order to leverage the chemical structure of cellulose, i.e. C6 polymer. In our work, we have studied the conversion of cellulose to sugar alcohols as well as glucose by supported metal catalysts and expanded to those of hemicellulose and lignin derivatives.
In the 1950s, Ballandin and co-workers reported the hydrolytic hydrogenation of cellulose to sorbitol and sorbitan by mineral acids and supported Ru catalysts under H2 pressure of 7 MPa (Scheme 1).12 The important point for this methodology is that the in situ hydrogenation of glucose to sugar alcohols having higher chemical stabilities resolves the trade-off problem by preventing the decomposition of glucose. Indeed, only 11% of glucose remained in water at 463 K for 3 h because of the decomposition to 5-hydroxymethylfurfural, furfural, and other compounds, whereas 99% of sorbitol was recovered under the same conditions in our tests. However, development of the process using only solid catalysts remained a challenge at that time, which had been due to the limited collision between the solid cellulose and the solid catalyst. On the other hand, Jacobs filed a patent for the one-pot conversion of starch to sorbitol using Ru/USY in 1989, in which USY worked as a solid acid to hydrolyse soluble starch and Ru catalysed the reduction of glucose to sorbitol.13
![]() | ||
Scheme 1 Sorbitol as a platform chemical produced by the hydrolytic hydrogenation of cellulose. |
Following on from this, we reported the first hydrolytic hydrogenation of cellulose to sugar alcohols under aqueous conditions by supported metal catalysts without using soluble acids.14 Avicel (microcrystalline cellulose, Merck) was converted to sugar alcohols (total 31%; sorbitol 25% and mannitol 6%) by a γ-Al2O3-supported Pt catalyst prepared from H2PtCl6, denoted as Pt(Cl)/γ-Al2O3, in the presence of H2 pressure of 5 MPa at 463 K for 24 h. The turnover number (TON) of bulk Pt for the production of sugar alcohols was 34. Ru/γ-Al2O3 was also active, whereas Rh, Ir and Pd catalysts gave only small amounts of sugar alcohols.
The produced sugar alcohols can be converted into various useful chemicals (Scheme 1).15 In the dehydrated products, sorbitan is a precursor to surfactants, and isosorbide is a feedstock to plastics16 including poly(ethylene-co-isosorbide) terephthalate (PEIT) as a remarkable PET analogue having a high glass transition temperature up to 470 K and also a commercial medicine for Ménière's disease. The nitric acid esters of isosorbide are used to treat angina pectoris. Hexane is produced from sorbitol by hydrodeoxygenation using a metal–solid acid combined catalyst (Pt/SiO2–Al2O3),17 and the product can be served as gasoline after isomerisation to increase the octane number.
At first, we performed the hydrolytic hydrogenation of the milled cellulose by using 2 wt% Pt(Cl)/γ-Al2O3 catalyst (Table 1 entry 1). Although the conversion of cellulose reached 89% because of the higher reactivity, yield of the sugar alcohols was 39% (sorbitol 32% and mannitol 7%). Accordingly, the selectivity for the sugar alcohols based on the conversion was 43%. The major by-products were 1,4-sorbitan (16%) and C2–C3 polyols (8%) including ethylene glycol, propylene glycol and glycerol. This result shows that the selective reduction of the hemiacetal/aldehyde group of glucose to alcohol is also important to achieve good yields of sugar alcohols, because the side-reactions, e.g. C–C hydrogenolysis, easily occur at the high temperatures required for the depolymerisation of cellulose. We prepared the catalyst from H2PtCl6 as described above, but Cl in the catalyst might affect the catalytic activity,20 which prompted us to study the effect of catalyst precursors on the reaction. We utilised Cl-free γ-Al2O3 (JRC-ALO-2, Catalysis Society of Japan) as the support in this work. The catalyst prepared from Pt(NH3)2(NO2)2, denoted as Pt(N)/γ-Al2O3, gave higher yields of sorbitol (46%) and mannitol (5%) (entry 2), and the total selectivity (70%) was nearly doubled from that (43%) by the Pt(Cl)/γ-Al2O3 catalyst.21 It is implied that Cl-free precursors are suitable for the reaction. To elucidate the origin of the selectivity difference, we performed energy dispersive X-ray spectroscopy (EDX) analyses of the Pt catalysts to determine the content of Cl. Residual Cl in the Pt(Cl)/γ-Al2O3 catalyst was 1.1 wt%, whereas that in Pt(N)/γ-Al2O3 was less than the detection limit (<0.01 wt%). Regarding the morphological property, both catalysts provided almost the same XRD patterns, CO uptake and N2-adsorption isotherms. Consequently, a clear difference is the content of Cl, and it appears that the residual Cl induces the side-reactions by being adsorbed on the metal and/or by the formation of acidic species like HCl. In fact, the addition of HCl (Cl 1.1 wt% to Pt(N)/γ-Al2O3) in the reaction mixture reduced the selectivity for the sugar alcohols (entry 3).
Entry | Catalyst | Conv. of cellulose/% | Sugar alcohol yields/%C | ||
---|---|---|---|---|---|
Sorbitol | Mannitol | Total (Sel.b/%) | |||
a Cellulose (Merck, Avicel, ball-milled for 2 days) 324 mg, catalyst 195 mg (Pt 2.0 wt%), water 40 mL, reaction time 24 h, p(H2) 5.0 MPa at r.t. b Based on the conversion. c The residue of the previous entry number was used as the catalyst. The conversion and yields are based on fresh cellulose (324 mg). d γ-Al2O3: JRC-ALO-2, Catalysis Society of Japan. e BP2000: carbon black, Black Pearls 2000, Cabot. f “101%” was due to the degradation of cellulose that had remained in the previous reaction. See also footnote c. | |||||
1 | Pt(Cl)/γ-Al2O3d | 89 | 32 | 7 | 39 (43) |
2 | Pt(N)/γ-Al2O3d | 72 | 46 | 5 | 51 (70) |
3 | Pt(N)/γ-Al2O3d + HCl | 98 | 38 | 10 | 47 (48) |
4 | Pt(N)/BP2000e | 82 | 49 | 9 | 58 (70) |
5c | Pt(N)/BP2000e | 101f | 53 | 11 | 64 (63) |
6c | Pt(N)/BP2000e | 101f | 53 | 12 | 65 (64) |
Next, we focused on the durability of the catalysts: Pt(N)/γ-Al2O3 catalyst selectively converts cellulose to sorbitol and mannitol (entry 2); however, the catalytic activity does not sustain in the reuse experiments owing to the transformation of γ-Al2O3 support to boehmite (AlO(OH)). In the screening tests of water-tolerant supports (TiO2, ZrO2 and C), we found that an easily available carbon black (BP2000, Cabot)-supported Pt(N) catalyst gives the sugar alcohols in 58%, 64% and 65%, respectively, in the reuse experiments of 3 times (entries 4–6). A total TON of Pt for the formation of sorbitol and mannitol was 175 in these reactions. In another method, Zhang et al. proposed an ordered mesoporous γ-Al2O3-C composite for the catalyst support affording both good interaction between Pt and Al2O3 and water-tolerant property derived from C.22 The γ-Al2O3 phase in the composite was thermally more stable than regular γ-Al2O3, but the reusability was not mentioned. A kinetic study of the Pt(N)/BP2000-catalysed reaction indicated that this reaction consists of two steps: the hydrolysis of cellulose to glucose via water-soluble oligosaccharides and the hydrogenation of glucose to sorbitol as expected.21 Although the former reaction, which is the rate determining step, proceeds even without catalysts under hydrothermal conditions,23 it was found that the Pt catalyst unpredictably accelerates this step threefold. The catalyst has the hydrolysis activity and furthermore might produce protons by the heterolysis of H2.14,24 The kinetic study also suggested that mannitol is predominantly derived from fructose generated by the isomerisation of glucose.
In the last few years, several groups have applied Ru/C catalysts to the hydrolytic hydrogenation of cellulose under aqueous conditions because of the superior activity of Ru/C for the hydrogenation of glucose to sorbitol.25 Liu et al. employed a Ru/activated carbon (AC) catalyst for the degradation of microcrystalline cellulose and gained the sugar alcohols in 39% yield (Table 2 entry 8).26 Wang et al. showed that the products are obtained in 73% yield from H3PO4-pretreated cellulose by using Ru/carbon nanotube catalyst (entry 10).27 Besides, Cs-substituted phospho- or silico-tungstates were utilised as solid acids to improve the hydrolysis step.28 The reactions took place at lower temperatures (433–443 K) than those of usual cases (≥458 K), and the yield of sugar alcohols was up to 70% (entries 11, 12). Although the stability of the heteropolyacids in water and the effect of dissolved species are of concern, Sels et al. indicated that the hydrothermal treatment of the acid measurably improves the water-tolerance.28a In addition, Wang and co-workers proposed that Cs3PW12O40 also works as an acid catalyst by the heterolysis of H2,28b which might be a similar step to that proposed for Pt-SO4/ZrO2 in the production of alkylate gasoline.29 However, W6+ species in heteropolyacids might be reduced to W5+ during the reaction (e.g., [PW12O40]3− + e− → [PW12O40]4−, E° + 0.22 V).
Entry | Catalyst | Cellulose | T/K | P(H2)/MPa | Reaction time/h | Cellulose conv./% | Sugar alcohol yields/%C | Ref. | ||
---|---|---|---|---|---|---|---|---|---|---|
Sorbitol | Mannitol | Total (Sel./%) | ||||||||
a BP2000: carbon black, Cabot. b AC: activated carbon. c CNT: carbon nanotube. d Pre-treated under hydrothermal conditions. e CNF: carbon nano-fibre. | ||||||||||
7 | 2.0 wt% Pt(N)/BP2000a | Microcrystalline | 463 | 5.0 | 24 | 66 | 39 | 4 | 43 (66) | 21 |
4 | 2.0 wt% Pt(N)/BP2000a | Ball-milled | 463 | 5.0 | 24 | 82 | 49 | 9 | 58 (70) | 21 |
8 | 4.0 wt% Ru//ACb | Microcrystalline | 518 | 6.0 | 0.5 | 86 | 30 | 10 | 39 (46) | 26 |
9 | 1.0 wt% Ru/CNTc | Microcrystalline | 458 | 5.0 | 24 | — | 36 | 4 | 40 | 27 |
10 | 1.0 wt% Ru/CNTc | H3PO4-treated | 458 | 5.0 | 24 | — | 69 | 4 | 73 | 27 |
11 | 5 wt% Ru/C, Cs3.5H0.5SiW12O40d | Ball-milled | 443 | 5.0 | 48 | 100 | — | — | 70 (70) | 28a |
12 | 1.0 wt% Ru/Cs3PW12O40 | Ball-milled | 433 | 2.0 | 24 | — | 43 | 2 | 45 | 28b |
13 | 16 wt% Ni2P/ACb | Microcrystalline | 498 | 6.0 | 1.5 | 100 | 48 | 5 | 53 (53) | 30 |
14 | 3.0 wt% Ni/CNFe | Microcrystalline | 503 | 6.0 | 4.0 | 94 | 23 | 5 | 28 (29) | 31 |
15 | 3.0 wt% Ni/CNFe | Ball-milled | 463 | 6.0 | 24 | 93 | 50 | 6 | 57 (61) | 31 |
Recently, base metal catalysts have been studied for the conversion of cellulose as alternatives for noble metals. Zhang et al. demonstrated that 16 wt% Ni2P/AC catalyst produces sorbitol (48% yield) and mannitol (5%) from microcrystalline cellulose (entry 13).30 Although the activity of the fresh catalyst was superior to those for 16 wt% Ni/AC catalyst and the mixture of 16 wt% Ni/AC and H3PO4 (ca. <10% yield), the catalyst was not durable in reuse experiments because of the phosphorus leaching. Sels and co-workers reported that a reshaped Ni metal catalyst at the tip of carbon nano-fibres on γ-Al2O3 (3 wt% Ni/CNF),31 produced in methane decomposition,32 provides 50% yield of sorbitol and 6% of mannitol in the conversion of ball-milled cellulose (entry 15), whereas 3 wt% Ni/AC catalyst prepared by simple impregnation methods is ineffective for the production of the sugar alcohols. They suggested that the reshaping of Ni crystals lifted up from the original support (Al2O3) by the formation of CNF in the decomposition of methane is the major factor for inhibiting the bond-breaking side-reactions. It is known that Ni crystals grow at the initial stage of the methane decomposition.32b In addition, Ni/CNF catalyst was stable in the reuse reactions.
Also in the combination of metal catalysts and homogeneous acids, many works have been devoted to improve the yield and to overcome the drawbacks. Sels et al. demonstrated the quantitative degradation of cellulose to the sugar alcohols (85%) and sorbitan (15%) in only 1 h by using H4SiW12O40.33 Mizuno and co-workers also utilised heteropolyacids for this reaction, and 54% yield of sorbitol was obtained at a low temperature (333 K) and hydrogen pressure (0.7 MPa).34 Heteropolyacids are recoverable by recrystallisation or extraction.35 Moreover, Sels et al. reported that highly-diluted HCl (0.0177 wt%, pH = 2.3) accelerates the hydrolysis step, and 60% yield of sugar alcohols and 33% of sorbitan are obtained.36 Some types of stainless steel can be used as a reactor even in the presence of HCl at low concentrations.
Hence, the hydrolytic hydrogenation method is also useful for the conversion of hemicellulose. Moreover, C5 sugar alcohols such as xylitol are not only sweetners that do not induce dental caries, but also potential precursors to ethylene glycol and propylene glycol.39 However, homogeneous acids or multi-step reactions have been required for this reaction as far as we know.12a,40 Thus, we applied the heterogeneous catalysis to the one-pot conversion of hemicellulose.41 In this study, we chose beet fibre (Nippon Beet Sugar Manufacturing) as an actual agricultural waste in the production of sucrose from sugar beet, whose annual production has been about 200–300 Mt. The beet fibre contains 23 wt% hemicellulose principally composed of arabinan, which is a polymer of arabinose mainly linked by 1,5-α-glycosidic bonds (Scheme 2). Accordingly, the hydrolytic hydrogenated product from beet fibre is arabitol. We found that 2 wt% Ru/AC catalyst converts beet fibre to arabitol in 83 wt% yield based on the weight of hemicellulose at 428 K for 24 h under H2 pressure of 5 MPa.41b Assuming that the hemicellulose part is approximately pure arabinan, the carbon-based yield is 72%.
Very recently, Murzin et al. reported a Pt/MCM-48 catalyst for the conversion of a bleached birch kraft pulp, and they obtained small amounts of sorbitol and xylitol.42
As a substitute for H2 molecules, sources supplying 2H+ and 2e− can be used for the reduction of sugar intermediates to sugar alcohols like the conversion of glucose to sorbitol on cathodes by electrochemical potential (2e−) with water (2H+).
Zhu et al. reported ionic liquid-stabilised Ru nanoparticles and a reversible binding boron agent for the conversion of cellulose to sorbitol, in which formates were used for the reduction step.43 As a simple and solid-catalytic system, we investigated the conversion of cellulose to sugar alcohols in water using 2-propanol as a reductant by supported metal catalysts (Scheme 3), which is the hydrolytic transfer hydrogenation of cellulose.44 In the screening tests of various catalysts, 2 wt% Ru/C catalysts (C: AC(N), CMK-3, C-Q10)45 prepared from RuCl3 by a typical impregnation method with H2-reduction at 673 K gave the highest yields of the sugar alcohols (43–46%) with the generation of H2 from 2-propanol (vide infra). In contrast, the Ru/γ-Al2O3 catalyst, which is active for the hydrolytic hydrogenation under 5 MPa of H2, was completely inactive for the transfer hydrogenation (0% yield), implying the difference of Ru active species. Regarding the durability, the Ru/AC(N) catalyst was reusable only twice (total TON = 170), which might be due to strong adsorption of by-products as EDX and XRD analyses showed no change of the structure of the Ru species before and after the deactivation. In addition, no Ru leaching was detected by inductively coupled plasma atomic emission spectroscopy (ICP-AES) measurement, and the filtrate after the first reaction did not catalyse the transfer hydrogenation reaction. We concluded that the Ru/carbons are heterogeneous catalysts for the hydrolytic transfer hydrogenation of cellulose.
![]() | ||
Scheme 3 Transfer hydrogenation of cellulose to sorbitol. |
When the Ru/C catalysts were used for the reaction, H2 was evolved during the reaction and the pressure became constant at 0.8 MPa in the initial period of 0.5 h, meaning a chemical equilibrium (eqn (1)). Considering this phenomenon, we expected that the hydrolytic hydrogenation of cellulose occurs with 0.8 MPa of H2 without using 2-propanol. In fact, the Ru/AC(N) catalyst provided 38% yield of the sugar alcohols in this reaction, whereas the Ru/γ-Al2O3 catalyst was inactive, suggesting that the active Ru species on C and Al2O3 are different. Note that very recently Mizuno and co-workers reported the hydrolytic hydrogenation of cellulose to sorbitol by a Pt nanoparticle catalyst under 0.7 MPa of H2 as described above, although a concentrated solution of H4SiW12O40 and a large amount of Pt (24 mol% to the substrate based on the number of glucose units) were required.34
(CH3)2CH–OH ⇌ (CH3)2C![]() | (1) |
![]() | ||
Fig. 2 Possible structure of Ru/CMK-3 catalyst. |
RuO2·2H2O might be reduced to RuOx species (x < 2) during the reaction,49 due to the reducing atmosphere. The in situ generated species should be the real active species, whose characterisation with in situ XAFS is under our investigation. We propose that the reaction mechanism is as follows (Scheme 4): (i) 2-propanol is transformed to acetone to give active hydrogen species on the Ru oxide active species. (ii) The active hydrogen species is in chemical equilibrium with H2 molecules, which allows both 2-propanol and H2 gas to be used as reductants. (iii) Cellulose is hydrolysed to glucose by hot compressed water and by the Ru/C catalysts. (iv) Glucose is reduced to sorbitol by the active hydrogen species.
![]() | ||
Scheme 4 Possible reaction pathway for the transfer hydrogenation of cellulose. |
![]() | ||
Scheme 5 Hydrolysis of cellulose to glucose and the potential chemicals derived from glucose. |
Immobilised sulfonic acids have been well studied for the hydrolysis of cellulose as outlined above, whereas the catalysis of supported metals has not been elucidated. Since we found that supported metal catalysts accelerate the hydrolysis step in the hydrolytic hydrogenation of cellulose, we retried the hydrolysis of cellulose by using this type of catalyst.58 In our study, we used 4-day-milled cellulose (Avicel, Merck) as the substrate. To achieve both the high reaction rate and the inhibition of glucose decomposition, rapid heating and cooling conditions were utilised,23bi.e. the reactor was heated from 298 K to 503 K in 15 min and then cooled down to 298 K by blowing air. In the screening tests of catalysts, we found that 2 wt% Ru/CMK-3 catalyst prepared from RuCl3 with H2-reduction at 673 K hydrolyses cellulose to glucose in 24% yield (Fig. 3). This catalytic system can produce glucose over a short period. The TON of Ru for the formation of glucose was 15 with excluding the glucose yield obtained by CMK-3 (see below). The conversion of cellulose was 56%, and thereby the selectivity for glucose was 43%. Other products were water-soluble oligosaccharides (16%) and glucose derivatives such as fructose (2%), mannose (1%), levoglucosan (1%), 5-hydroxymethylfurfural (5-HMF; 3%) and furfural (0.4%). The sugar compounds should be carefully analysed because they sometimes overlap in the chromatographs.9e This catalyst was reusable at least 5 times with retaining the activity.
![]() | ||
Fig. 3 Hydrolysis of cellulose by Ru/CMK-3 catalyst under rapid heating and cooling conditions. Cellulose (Merck, Avicel, ball-milled for 4 days) 324 mg, catalyst 50 mg, water 40 mL. “Blank” entry shows the reaction without catalysts. “Others” = (conversion of cellulose) − (“Glucose” + “Oligomers”). |
We conducted control experiments to clarify the detail of the catalysis. A blank test without catalysts gave small amounts of glucose (5%) and oligosaccharides (14%), but CMK-3 gave 16% yield of glucose and 22% yield of oligosaccharides, indicating that the carbon has hydrolytic activity. By increasing Ru loading from 0 wt% to 10 wt%, the yield of glucose was smoothly raised from 16% up to 31%, whereas that of oligosaccharides decreased from 22% to 5%. The conversion of cellulose was slightly improved by increasing the content of Ru. It is thus indicated that the Ru species hydrolyses both cellulose and oligosaccharides, and especially shows high activity for the latter substrate. In fact, the hydrolysis of cellobiose to glucose by the Ru/CMK-3 catalyst took place 5 times more quickly than that by CMK-3 at 393 K. The catalytic activity of the Ru species is intriguing, and we determined that the Ru species is RuO2·2H2O by means of several analytical techniques (Fig. 2, vide supra).47 Regarding the mechanism for the hydrolysis, one possibility is that the RuO2·2H2O species on CMK-3 might desorb the hydrated water to give a Lewis acid site (Scheme 6a), as it has been proposed that a Ru polyoxometalate depolymerises cellulose to glucose.35a Another hypothesis is that the Ru species works as Brønsted acid by the heterolysis of water molecules on Ru (Scheme 6b), because it is known that [Ru(H2O)6]3+ has a low pKa (2.9 at 298 K).59
![]() | ||
Scheme 6 Speculated mechanisms for the formation of acid sites on Ru/CMK-3 catalyst. |
![]() | ||
Scheme 7 Conversion of cellulose to poly(3-hydroxybutyrate) using chemical and biological systems. |
There are two key factors for the synthesis of cellulose hydrolysate for the fermentation process. (1) The E. coli requires at least 2 g L−1 glucose in the culture medium to accumulate P(3HB). This point can be achieved by increase of the concentration of cellulose. (2) The growth of the microbes is mainly inhibited by 5-HMF,63 indicating that a high molar ratio of glucose to 5-HMF is required. Since 5-HMF was formed by the successive thermal decomposition of glucose, we expected that a lower conversion of cellulose by dropping the reaction temperature would give a higher glucose/5-HMF ratio. As the first trial, we used Ru/γ-Al2O3 as a catalyst and carried out the hydrolysis of cellulose at a cellulose concentration of 130 g L−1 using the rapid heating to 488 K (15 K lower than the typical hydrolysis temperature) and cooling conditions, where substrate/catalyst (S/C) ratio based on mol-glucose units to g-atom-Ru was 740. The major products were glucose (12% yield), corresponding to the concentration of 16 g L−1, oligosaccharides (5.2%), fructose (2.1%) and 5-HMF (1.1%). The concentration of glucose and the glucose/5-HMF ratio of 10 were sufficiently high for the fermentation, which prompted us to conduct the microbial synthesis of P(3HB) using the hydrolysate by collaborating with biochemists.
E. coli (JM109 strain) harbouring a plasmid pGEM-CAB, which bore the P(3HB) synthesis genes from Ralstonia eutropha that encoded β-ketothiolase, acetoacetyl-CoA reductase and PHA synthase, was cultured for 72 h at 303 K in LB media containing the reaction solution of the cellulose hydrolysis in various fractions to be aimed concentrations of glucose (2–10 g L−1) (Table 3). Terminal growth of the E. coli improved from 2.1 g L−1 to 3.9 g L−1 based on the dried cell weight with increasing the concentration of glucose in the range of 2–7 g L−1, and the accumulated weight of P(3HB) was also raised from 7 wt% to 42 wt% in the dried E. coli cells at the same time (entries 1–6). Therefore, the production amount of P(3HB) reached 1.6 g L−1 at 7 g L−1 glucose, which corresponds to 49% efficiency based on the theoretical maximum yield (3.4 g L−1) calculated from eqn (2). A higher concentration of glucose gains a better productivity of P(3HB). However, the E. coli did not grow at ≥8 g L−1 glucose (entries 7, 8) because the concentrations of harmful by-products for the bacteria, in which 5-HMF was the major one, were proportional to that of glucose and reached the threshold for inhibiting the cell growth. If we can improve the glucose/5-HMF ratio of the hydrolysate, the fermentation should proceed at higher concentrations of glucose.
Glucose (C6H12O6) → P(3HB) unit (C4H6O2) + 2CO2 + 6H+ + 6e− | (2) |
Entry | Glucoseb/g L−1 | E. coli c/g L−1 | P(3HB) contentd/% | P(3HB) amount/g L−1 (Yielde/%) |
---|---|---|---|---|
a 303 K, incubation time 72 h. b The cellulose hydrolysate containing glucose 86 mM (16 g L−1) was added to condensed LB media for each concentration of glucose by changing the mixing ratio. c Terminal growth of E. coli based on the dried weight. d In dried cells of E. coli. e P(3HB) yield based on eqn (2). | ||||
1 | 2.0 | 2.1 | 7.0 | 0.15 (15) |
2 | 3.0 | 2.3 | 24 | 0.55 (38) |
3 | 4.0 | 3.1 | 23 | 0.71 (37) |
4 | 5.0 | 3.4 | 26 | 0.88 (37) |
5 | 6.0 | 3.9 | 31 | 1.2 (42) |
6 | 7.0 | 3.9 | 42 | 1.6 (49) |
7 | 8.0 | 0.0 | — | 0.0 (0.0) |
8 | 10 | 0.0 | — | 0.0 (0.0) |
The above strategy was applied to synthesise poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (P(3HB-co-3HV)),64 which has higher strength and flexibility than P(3HB).65 P(3HB-co-3HV) is producible by using propionate based on the mechanism for the formation of P(3HB), whose key step is the condensation of propionyl-CoA with acetyl-CoA to 3-ketovaleryl-CoA by β-ketothiolase. Since it was found that E. coli LS5218 is more resistant to 5-HMF than JM109, the concentration of the cellulose hydrolysate in the LB media was increased to 10 g L−1 of glucose. 0.5–3.8 g L−1 of sodium propionate was added into the mixture, and then the pH of the solution was adjusted to 7 because the concentration of the base (propionate) was different in each experiment, which could affect the results. The microbes converted glucose and propionate to a plastic, and the amount was up to 2.1 g L−1. The plastic surely contained 3HV fraction in the range of 5.6–40 mol%.
Catalytic hydrodeoxygenation (HDO) is one of the most important reactions for converting lignin and lignin-derived materials such as bio-oils into hydrocarbons.66,71 HDO studies often employ lignin models, especially phenol derivatives, rather than lignin and bio-oil themselves to get better understanding of the reaction pathway and intermediates in lignin conversion. The HDO of phenols involves two reaction pathways; (i) one is the direct hydrogenolysis of the C–O bond in phenols leading to deoxygenated aromatic products, and the other provides aliphatic hydrocarbons via (ii) the hydrogenation of aromatic ring, (iii) the dehydration of cycloalkanols to cycloalkenes and (iv) the hydrogenation of cycloalkenes (Scheme 8).71,72 The former route is important not only for bio-fuel production but also as an alternative to petrochemical-based routes for the production of bulk aromatic compounds such as BTX.73 The latter route consumes more hydrogen to produce aliphatic hydrocarbons but is more promising for the production of biofuels with lower aromatics content satisfying an environmental demand.74
![]() | ||
Scheme 8 Reaction pathway for the HDO of phenol. |
Entry | Catalyst | Substrate | Reactor | P(H2)/MPab | T/K | Reaction time/h | Conv. of the substrate/% | Main product (Sel./%) | Ref. |
---|---|---|---|---|---|---|---|---|---|
a Selected examples from each reference. b Initial H2 pressure at r.t. c H2 pressure at reaction temperature. d Weight hourly space velocity (WHSV/h−1). e W/F (h) = catalyst mass (g)/feed rate (g/h). f Mesoporous ZSM-5. g Not specified. h Liquid hourly space velocity (LHSV/h−1). i 1-Methyl-3-(4-sulfobutyl)imidazolium trifluoromethanesulfonate. j AC: activated carbon. k Ni-W/AC prepared using H3[P(W3O10)4]·xH2O as a tungsten precursor. l The Ni particle diameter is 3.7 nm. m Cu-PMO: hydrotalcite-derived porous metal oxide doped with Cu. n Methanol as a solvent and H2 source. o AC(N): activated carbon (Norit). | |||||||||
1 | Rh/ZrO2 | Guaiacol | Batch | 8c | 573 | 3 | ca. 100 | Benzene | 78 |
2 | Pt/HY | Phenol | Fixed-bed | 4c | 523 | 20d | 100 | Cyclohexane (94) | 79 |
3 | Pt/HBeta | Anisole | Fixed-bed | 0.1 | 673 | 0.33e | 100 | Benzene (51), toluene (28), xylene (11) | 80 |
4 | Pt/m-ZSM-5f | Dibenzofuran | Fixed-bed | —g | 473 | 6h | —g | Bicyclohexyl (83) | 81 |
5 | Rh/Ru NPs, ILi | 4-Ethylphenol | Batch | 4 | 403 | 4 | 99 | Ethylcyclohexane (99) | 82 |
6 | Pt/Al2O3/SiO2 | Pine pyrolytic oil | Trickle-bed | 10c | 673 | 2d | —g | Phenols, aromatics, alkanes | 83 |
7 | Ru/ACj | Alcell lignin fast pyrolytic oil | Batch | 10 | 623 | 1 | —g | Cycloalkanols, cycloalkanes, linear alkanes | 84 |
8 | Ru/ACj | Beech fast pyrolytic oil | Batch | 2 | 623 | 4 | —g | Phenols, aromatics, alkanes | 85 |
9 | γ-Mo2N | Benzofuran | Microreactor | 0.1 | 723 | —g | ca. 100 | Benzene (37), toluene (30), ethylbenzene (33) | 86 |
10 | VN | Benzofuran (in a model liquid feed mixture) | Trickle-bed | 3.1c | 643 | 5h | 63 | Ethylbenzene | 87 |
11 | Ni–Mo/γ-Al2O3 | Phenol | Batch | 2.8 | 723 | 1 | 100 | Benzene (60), cyclohexane (16) | 88 |
12 | Ni–Mo/γ-Al2O3 | 2,3-Dihydrobenzofuran | Fixed-bed | —g | 553 | —g | ca. 100 | Ethylcyclohexane (ca. 70), methylcyclohexane (ca. 30) | 89 |
13 | Ni–Cu/CeO2 | Anisole | Fixed-bed | 1c | 573 | 1h | 100 | Cyclohexane (>99) | 90 |
14 | Co–Mo–B | Phenol | Batch | 4c | 548 | 10 | 100 | Cyclohexane (>95) | 91 |
15 | Ni–Mo–B | Phenol | Batch | 4c | 498 | 7 | 81 | Cyclohexane (81), benzene (12) | 92 |
16 | La–Ni–W–B | Phenol | Batch | 4c | 498 | 4 | 95 | Cyclohexane (36), cyclohexene (37) cyclohexanol (25) | 93 |
17 | Ni-W(P)/ACj,k | Phenol | Fixed-bed | —g | 523 | 0.5d | 100 | Cyclohexane (>95) | 94 |
18 | MoO3 | 4-Methylphenol | Batch | 4.8c | 648 | 1.7 | 100 | Toluene (ca. 60) | 95 |
19 | Ni/SiO2l | Phenol | Fixed-bed | 0.1 | 573 | 0.73e | —g | Benzene (99) | 96 |
20 | Ni2P/SiO2 | Guaiacol | Packed-bed | 0.1 | 573 | —g | 80 | Benzene (60), phenol (30), anisole (10) | 97 |
21 | Cu-PMOm | 2,3-Dihydrobenzofuran | Batch | —n | 573 | 13 | 100 | 2-Ethylcyclohexanol (68), methylethylcyclohexanol (24) | 98 |
22 | Cu-PMOm | Poplar organosolv lignin | Batch | —n | 573 | 24 | 100 | Monomeric cyclohexyl derivatives | 99 |
23 | Pd/AC,j H3PO4 | 4-Propylphenol | Batch | 5 | 523 | 0.5 | 100 | Propylcyclohexane (96) | 101 |
24 | RANEY® Ni, Nafion/SiO2 | 4-Propylphenol | Batch | 4 | 473 | 0.5 | 100 | Propylcyclohexane (99) | 102 |
25 | Pt/AC or Pd/AC,j H3PO4 | White birch wood lignin | Batch | 4 | 473–523 | 2–4 | —g | C8–C9 alkanes, C14–C18 alkanes, methanol | 103 |
26 | Pt/AC(N)o | 4-Propylphenol | Batch | 4 | 553 | 2 | 100 | Propylcyclohexane (>99) | 104 |
Recent efforts have been focused on developing low-cost, base metal catalysts with high activity and selectivity in the HDO process. Bell reported that γ-Mo2N was an effective catalyst for the gas-phase HDO of benzofuran at 723 K under 1 atm H2 to give a mixture of aromatic products (ethylbenzene, toluene and benzene) (entry 9).86 The reaction involves hydrogenation of the furan ring, hydrogenolysis of the C–O bonds and hydrogenolysis/dealkylation of the alkyl group. With a series of metal carbide and nitride catalysts (VN, Mo2N, TiN, VC, Mo2C, WC, NbC), Oyama studied the hydroprocessing of a model liquid feed mixture (3000 ppm dibenzothiophene, 2000 ppm quinoline, 500 ppm benzofuran, 20 wt% tetralin and balance aliphatics) in a trickle-bed reactor at 643 K.87 Among the catalysts tested, VN showed the best activity to give ethylbenzene (entry 10), while the conventional sulfided Ni–Mo/γ-Al2O3 catalyst produced ethylcyclohexane as a major product.
Recently, it is presumed that the presence of coordinatively unsaturated sites in non-sulfided catalysts is important for the HDO of lignin-related compounds. Kallury et al. tested the HDO activity of a non-sulfided Ni–Mo/γ-Al2O3 catalyst for various phenol derivatives in a fixed-bed reactor at 623–773 K under 2.8 MPa H2.88 The reaction of phenol at 723 K afforded benzene and cyclohexane in 60% and 16% yields, respectively (entry 11). The deoxygenation activity was greatly decreased in the presence of water (14% conversion, benzene in <1% yield), suggesting the interaction of the oxygen atoms with the active sites. Ozkan et al. investigated the use of Ni–Mo/γ-Al2O3 catalysts for the gas-phase HDO of benzofuran derivatives (entry 12).89 The reduced Ni–Mo/γ-Al2O3 showed good hydrogenation activity to give saturated hydrocarbons, suggesting the formation of Ni- and Mo-associated anionic vacancies during the reaction. In contrast, hydrogenation activity of the sulfided catalyst was decreased by adding a sulfur source. Yakovlev et al. found that Ni–Cu catalysts supported on metal oxides were effective for the gas-phase HDO of anisole and biodiesel.90 When Ni–Cu/CeO2 was used as a catalyst at 573 K under 1 MPa H2, anisole was fully converted to cyclohexane (entry 13). The authors explain that Cu facilitates the reduction of nickel oxide and catalyst support, and the reduced Ni activates hydrogen whereas the oxide support serves for an additional activation of the C–O group in oxygen-containing compounds. Yang et al. studied the HDO of phenol in dodecane with amorphous Co–Mo–B,91 Ni–Mo–B92 and La–Ni–W–B93 catalysts (entries 14, 15 and 16). The reactions may involve the adsorption of phenol on the acid sites of MoO2 and WO3via the donation of a lone pair of the C–O group as well as the activation of hydrogen by Ni and Co. Similarly, Echeandia et al. found that the Ni-W/AC catalysts prepared from heteropolyacids were active in the gas-phase HDO of phenol at 523 K to afford cyclohexane with high selectivity (entry 17).94 Smith examined the gas-phase HDO of 4-methylphenol at 648 K under 4.8 MPa H2 over unsupported MoO3, MoO2, MoS2 and MoP catalysts. The MoO3 catalyst showed the highest activity to yield toluene as a major product (entry 18).95 Keane reported the HDO of a methanolic solution of phenol using a Ni/SiO2 catalyst (Ni particle diameter: 3.7 nm) at 573 K under atmospheric H2 flow in a fixed-bed reactor to give benzene in 99% selectivity (entry 19).96 The product ratio of benzene/cyclohexane greatly depends on the Ni particle size and the reaction temperature. Oyama et al. investigated the gas-phase HDO of guaiacol at 573 K under atmospheric H2 using SiO2-supported metal phosphides (Ni2P, Fe2P, MoP, Co2P, WP) as catalysts.97 Among the catalysts tested, Ni2P/SiO2 showed the highest activity (80% conversion) to form benzene, phenol and anisole in 60%, 30% and 10% selectivities, respectively (entry 20). Under the conditions, commercial sulfided Co–Mo/γ-Al2O3 deactivated quickly by coking and showed almost no activity for the HDO of guaiacol. As outlined above, various heterogeneous catalysts have been developed for the conversion of lignin derivatives, providing valuable information about designing a new catalyst for HDO reaction.
The Pt/carbons were employed as catalysts in the HDO of 4-propylphenol (4-PrPhOH) in water at 553 K under 4 MPa H2 for 1 h (Table 5). Readily available Pt/AC(N) showed excellent catalytic activity and gave propylcyclohexane (PrCyH) in 97% yield (entry 1). Pt/MWCNT and Pt/CMK-3 were also effective in this reaction, affording PrCyH in slightly lower yields (entries 6 and 7). Pt/AC(W) yielded PrCyH and 4-propylcyclohexanol (4-PrCyOH) in 73% and 16% yields, respectively (entry 5). In contrast, Pt/BP2000 gave much lower conversion (16%) and yield of PrCyH (9%) probably due to the reduction of catalytic activity by the sulfur impurity of BP2000 support (entry 8).105 Since Pt/AC(N) showed high activity, a reaction of 4-PrPhOH on a gram-scale was performed with a high substrate to catalyst molar ratio (S/C) of 1000 at 553 K for 2 h, which gave PrCyH in >99% yield (entry 2). Pt/AC(N) also promoted the large-scale reaction even at 513 K, affording PrCyH in excellent yield (>99%) after 10 h (entry 3). Furthermore, 4-PrPhOH was efficiently converted into PrCyH without water solvent (entry 4), which shows that the catalyst can be used both in water and under water-free conditions. Besides, oxide-supported Pt/CeO2, Pt/TiO2 and Pt/ZrO2 were also examined as catalysts for this reaction, giving PrCyH in 66–83% yield (entries 9–11). In these reactions, propylbenzene (PrPhH) was also formed in 3–10% yield probably via the direct hydrogenolysis of 4-PrPhOH. In stark contrast, Pt/γ-Al2O3 showed no catalytic activity because of the structural change of γ-Al2O3 into boehmite (AlO(OH)) during the reaction (entry 12). Ru/AC(N), Rh/AC(N) and Pd/AC(N) were not very effective for the reaction of PrPhOH, affording lower yields of PrCyH than the Pt/AC(N) catalyst (entries 13–15).
Entry | Catalyst | Yield/% | Conv./% | ||||
---|---|---|---|---|---|---|---|
PrCyHn | 4-PrCyOHo | 4-PrCyOp | PrPhHq | Othersr | |||
a Reaction conditions: 4-PrPhOH 272 mg (2.0 mmol), catalyst (2 wt% metal, S/C = 200), water 40 mL, 553 K, reaction time 1 h, p(H2) 4.0 MPa at r.t. b 4-PrPhOH 1.36 g (10 mmol), S/C = 1000. c 553 K, 2 h. d 513 K, 10 h. e 553 K, 4 h, without water. f Activated carbon, SX Ultra (Norit). g Activated carbon (Wako). h Multi-walled carbon nanotube (TCI). i Carbon black Black Pearls 2000 (Cabot). j JRC-CEO-2 (JRC: Japan Reference Catalyst, Catalysis Society of Japan). k P-25 (Deggusa). l JRC-ZRO-2. m JRC-ALO-2. n Propylcyclohexane. o 4-Propylcyclohexanol. p 4-Propylcyclohexanone. q Propylbenzene. r Gaseous hydrocarbons, etc. | |||||||
1 | Pt/AC(N)f | 97 | <1 | 0 | 0 | 3 | 100 |
2b,c | Pt/AC(N)f | >99 | 0 | 0 | 0 | 0 | 100 |
3b,d | Pt/AC(N)f | >99 | 0 | 0 | 0 | 0 | 100 |
4b,e | Pt/AC(N)f | >99 | 0 | 0 | 0 | 0 | 100 |
5 | Pt/AC(W)g | 73 | 16 | <1 | 0 | 11 | 100 |
6 | Pt/MWCNTh | 92 | 1 | <1 | 0 | 7 | 100 |
7 | Pt/CMK-3 | 92 | 0 | 0 | 0 | 8 | 100 |
8 | Pt/BP2000i | 9 | 1 | 0 | 1 | 5 | 16 |
9 | Pt/CeO2j | 83 | <1 | 0 | 8 | 9 | 100 |
10 | Pt/TiO2k | 77 | 2 | 0 | 3 | 18 | 100 |
11 | Pt/ZrO2l | 66 | <1 | 0 | 10 | 24 | 100 |
12 | Pt/γ-Al2O3m | 0 | 0 | 0 | 0 | 0 | 0 |
13 | Ru/AC(N)f | 55 | <1 | 0 | 0 | 45 | 100 |
14 | Rh/AC(N)f | 83 | <1 | 0 | 0 | 17 | 100 |
15 | Pd/AC(N)f | 22 | 57 | 2 | <1 | 3 | 84 |
Pt/AC(N) was also effective for the aqueous-phase HDO of guaiacols (4-propylguaiacol, 4-allylguaiacol and 4-acetonylguaiacol) and a syringol derivative (4-allyl-2,6-dimethoxyphenol) under the standard conditions in Table 5, giving PrCyH (58–78% yield based on carbon) and methanol (2–3%) after 6 h. In these cases, propylcyclopentane (5–10%) was also formed possibly via skeletal isomerisation and hydrogenolysis of cycloalkane.101,102 These results suggest that this water-based catalyst system is applicable to the HDO of various phenolic compounds, including lignin and bio-oil.
The reuse experiments of Pt/AC(N) were carried out for the HDO of 4-PrPhOH at 553 K in water with the S/C ratio of 1000 for the reaction time (10 min) that provided ca. 60% yield of PrCyH in the first run to check the durability (Fig. 4(a)). The catalyst was recovered by centrifugation and was reused twice in water, which afforded PrCyH in stable chemical yields (61–65%). The XRD analysis of the Pt/AC(N) after the 3rd run showed that the Pt particles retained their small sizes (< 2 nm). In addition, the pH of the aqueous solution after the 1st run was neutral (pH = 7.0), showing no leaching of acid components from the catalyst. These results indicate that the Pt/AC(N) catalyst is durable under the catalytic conditions. In contrast, the reuse experiments under the water-free conditions showed that the catalytic activity was reduced during the repeated catalytic runs probably due to coke formation (Fig. 4(b)). Hence, the aqueous conditions are suitable to keep the catalytic activity.
![]() | ||
Fig. 4 Reuse experiments of Pt/AC(N) catalyst in the HDO of 4-PrPhOH (a) in water and (b) water-free conditions. Reaction conditions: 4-PrPhOH 1.36 g (10 mmol), 2 wt% Pt/AC(N) 98 mg (S/C = 1000), 553 K, 10 min (in water) or 30 min (water-free conditions). Abbreviations: see Table 5. |
The HDO of 4-PrCyOH was investigated to reveal the reaction pathway from 4-PrPhOH to PrCyH (Fig. 5). In the absence of catalyst, 4-PrCyOH was dehydrated to give 4-propylcyclohexene (4-PrCHE) in 11% yield (15% conversion). It is suggested that the dehydration is catalysed by in situ generated protons from hot compressed water.100,106,107 Hence, the reaction of 4-PrCyOH into 4-PrCHE was completely suppressed by the addition of a base (CaCO3, pH = 10). On the other hand, Pt/AC(N) easily facilitated the conversion of 4-PrCyOH to PrCyH in 95% yield. Even under the basic conditions, Pt/AC(N) afforded PrCyH in a good yield (58%). Besides, NH3-TPD profile showed that Pt/AC(N) had no acidity. These results suggest that this reaction is promoted not by acid-catalysis but by the direct C–O bond hydrogenolysis of 4-PrCyOH with Pt/AC(N).
![]() | ||
Fig. 5 HDO of 4-PrCyOH in water by Pt/AC(N). 4-PrCyOH 2.0 mmol, 2 wt% Pt/AC(N) 98 mg, CaCO3 4.0 mmol, water 40 mL, 553 K, 1 h, 4 MPa H2 at r.t. “Blank” entry shows the reaction without catalyst. PrCHE: 4-propylcyclohexene. Other abbreviations: see, Table 5. |
Scheme 9 shows the proposed reaction pathways for the HDO of 4-PrPhOH to PrCyH by Pt/AC(N) catalyst. (i) First, the aromatic ring of 4-PrPhOH is hydrogenated to give 4-PrCyOH. (ii) 4-PrCyOH is converted to PrCyH by the direct C–O bond hydrogenolysis, which is the main pathway from 4-PrCyOH to PrCyH. (iii) When the HDO is performed in water, the reaction includes the dehydration, as a minor pathway, of 4-PrCyOH to 4-PrCHE by in situ generated protons from hot compressed water. (iv) Subsequent hydrogenation of 4-PrCHE affords PrCyH. Therefore, the addition of acid catalysts is not necessary in this reaction.
![]() | ||
Scheme 9 Putative reaction mechanism for the HDO of 4-PrPhOH into PrCyH by Pt/AC(N) without additional acid catalysts. |
In the utilisation of the lignocellulose conversion, the separation of heterogeneous catalysts from a solid mixture containing lignin is a problem. Since the extraction of lignin has been established in the paper industry, one possible way is the separation of lignocellulose into carbohydrates and lignin followed by the successive catalytic reactions. Another route is the direct transformation of whole lignocellulose to chemicals using multi-functional catalytic steps.
As various types of solid catalysts can be designed and wide ranges of reaction conditions are applicable to the catalytic systems, we believe that further progress of solid catalysis overcomes this tough issue for practical applications. One such catalyst would be different from previous ones possessing strong acidic sites. Very recently, it has been demonstrated that silanols (average pKa = 7) can cleave glycosidic bonds by optimising the configurations.108
This journal is © The Royal Society of Chemistry 2012 |