Renewable hydrogen by aqueous-phase reforming of glucose

Abstract

Hydrogen can be produced from aqueous solutions containing 10 wt% glucose with high selectivities through the combined use of a hydrogenation reactor for conversion of glucose to sorbitol, followed by a reforming reactor for conversion of sorbitol to H₂ and CO₂ and then a gas–liquid separator for the removal of high-pressure H₂-rich reformate gas, ready for use in a fuel cell.

---

Biomass is a promising source for the sustainable production of hydrogen in an age of diminishing fossil fuel reserves; however, conversion of biomass to hydrogen remains a challenge, since processes such as enzymatic decomposition of sugars, steam-reforming of bio-oils and gasification suffer from low hydrogen production rates and/or complex processing requirements.¹

We have recently reported² that hydrogen can be generated by catalytic reforming of oxygenated hydrocarbons in liquid water at temperatures near 500 K. Of particular interest in this class of reactants is glucose (C₆O₆H₁₂), because this sugar makes up the major energy reserves in plants and animals. While the selectivity for hydrogen production is insensitive to the liquid-phase concentration of sugar-alcohols such as sorbitol (C₆O₆H₁₄), the hydrogen selectivity from reforming of glucose decreases as the liquid concentration increases from 1 to 10 wt% because of undesired hydrogen-consuming side reactions that occur in the liquid phase.³ This decrease in selectivity is an important limitation, because processing dilute aqueous solutions involves the processing of excessive amounts of water.

Fig. 1 depicts various reaction pathways that take place during aqueous-phase reforming of glucose and sorbitol. Production of H₂ and CO₂ from glucose (G) and sorbitol (S) takes place on metal catalysts such as Pt or NiSn alloys (pathways G1 and S1) via cleavage of C–C bonds followed by water-gas shift processes.² Undesired alkanes are formed by cleavage of C–O bonds on the metal catalyst and by dehydration processes on acidic catalyst supports (pathways G2, S2).⁴ In the case of glucose, undesired reactions can also take place in the aqueous phase to form organic acids, aldehydes and carbonaceous deposits (pathway G3).³ These undesirable homogeneous decomposition reactions are first order in glucose concentration, whereas the desirable reforming reactions on the catalyst surface are fractional order;⁵ therefore, high concentrations of glucose lead to low hydrogen selectivities. Another reaction that links the aforementioned pathways is the hydrogenation of glucose to sorbitol (pathway G–S), which takes place on metal catalysts with high selectivity at low temperatures (e.g., 400 K) and high H₂ pressures.⁶

Fig. 1 Reaction pathways involved in glucose and sorbitol reforming.

The rates of pathways G1, G2, G3 increase more rapidly with temperature than does the rate of pathway G–S. Accordingly, a strategy for improving the hydrogen selectivity from aqueous-phase reforming of glucose is to employ a dual-reactor system involving a low-temperature hydrogenation step followed by a higher-temperature reforming process. To implement this two-step process, it is important to understand how the pressure of H₂ depends on the system pressure. Aqueous-phase reforming leads to the production of gas bubbles containing H₂, CO₂ and alkanes. The pressure in these bubbles is approximately equal to the total pressure of the system, and the partial pressures of the gaseous reaction products and water vapour depend on the composition of the feed stream and the saturation pressure of water at the reactor temperature.⁷ For this study, the partial pressures of the gas phase products in the reactor were varied by co-feeding a gas stream (N₂ or H₂) with the liquid feed at the inlet of the reactor. The performance of the reactor in this mode can then be compared to the case where only liquid is fed to the reactor and N₂ sweep gas is combined with the effluent stream at the exit of the reactor.

Table 1A and B shows the results of aqueous-phase reforming of 10 wt% sorbitol and glucose solutions at 538 K and 52.4 bar over a Pt/Al₂O₃ catalyst. All runs were conducted at high conversions of the feed to gas-phase products, from which the reaction selectivities were calculated. We report the hydrogen selectivity, which is defined as the number of moles of H₂ in the effluent gas normalized by the number of moles of H₂ that would be present if each mole of carbon in the effluent gas had participated in the sorbitol or glucose reforming reaction to give 13/6 or 2 moles of H₂, respectively. When H₂ is co-fed with the liquid reactant stream to the reactor, the hydrogen selectivity reported is based on the calculated moles of hydrogen produced, using the composition of the carbon-containing products in the effluent gas stream. We also report the alkane selectivity, which is defined as the moles of carbon in the gaseous alkane products normalized by the total moles of carbon in the gaseous effluent stream.

Table 1 Experimental results for the aqueous-phase reforming of 10 wt% sorbitol and glucose at 538 K under different reactor configurations^a

Experiment	A: Sorbitol reforming			B: Glucose reforming			C: Hydrogenation + reforming
Gas flow at reactor inlet/outlet	N2 sweep at outlet	N2 fed at inlet	H2 fed at inlet	N2 sweep at outlet	N2 fed at inlet	H2 fed at inlet	Sorbitol + H2 fed at inlet	Glucose + H2 fed at inlet
a Catalyst: 3 wt% Pt on nanofibers of γ-Al2O3 (BET area 500 m^2 g^−1) having a Pt dispersion of 70% (see ESI^1).
Pressure/bar	52.4	52.4	52.4	52.4	52.4	52.4	54.8	54.8
Hydrogenation/reforming WHSV/g(feed) g(cat)^−1 h^−1		—/0.0963			—/0.0224		0.091/0.065
% Carbon in gas-phase effluent	86	96	91	91.2	92.4	95.8	100	93
% H2 selectivity	50	62	63	10.5	13.4	11.6	57.3	62.4
% Alkane selectivity	22	20	19.1	49.5	47.5	46.8	22.7	21.3
α = moles H2/mole glucose fed (αmax = 12)	4.4	6.7	6.6	1.3	1.5	1.5	6.5	5.8

---

The highest rates of aqueous-phase reforming were obtained for runs where N₂ was combined with the liquid feed at the reactor inlet. Also, since glucose is less reactive than sorbitol for aqueous-phase reforming over Pt/Al₂O₃,² lower weight hourly space velocities (WHSV; grams of reactant per gram of catalyst per hour) of glucose were employed to achieve high conversions. Table 1 shows that the hydrogen selectivities observed for aqueous-phase reforming of glucose (∼10–13%) are much lower compared to reforming of sorbitol (∼60%) under similar conditions. Also higher alkane selectivities of 47–50% were observed for reforming of glucose, indicating that aqueous-phase reforming of glucose, even with hydrogen co-fed with the liquid reactant stream at the inlet of the reactor, is not selective for production of H₂. The above results indicate that co-feeding hydrogen with liquid reactants into the reforming reactor at 538 K (and 52.4 bar) does not lead to rapid hydrogenation of glucose into sorbitol and its subsequent reforming to give high H₂ selectivities. Similarly, sorbitol does not undergo rapid dehydrogenation to glucose under conditions of aqueous-phase reforming where H₂ is not co-fed with the liquid reactant stream to the reforming reactor, a desirable result.

Experiments were conducted by co-feeding gaseous H₂ with aqueous solutions containing 10 wt% sorbitol or glucose into a dual-reactor system consisting of a hydrogenation reactor at 393 K followed by a reforming reactor. The presence of the hydrogenation reactor did not affect the aqueous-phase reforming of sorbitol. However, when 10 wt% glucose was co-fed with gaseous H₂ to the dual-reactor system, high hydrogen selectivity (62.4%) and low alkane selectivity (21.3%) were obtained (see Table 1C), these values being similar to the selectivities obtained from the reforming of sorbitol. These results indicate that glucose was first completely hydrogenated to sorbitol before being sent to the reformer in which the sorbitol was converted with high selectivity to H₂ and CO₂.

Table 1 compares the net moles of hydrogen produced per mole of glucose (α), where the maximum theoretical value of α is 12. For sorbitol feeds, α is calculated by subtracting 1 from the moles of H₂ produced per mole of feed (to account for the extra mole of H₂ in sorbitol compared to glucose). For sorbitol reforming, α is ca. 6.6 when N₂ or H₂ are co-fed with the liquid reactant stream to the reactor. In contrast, α is only 1.5 for the glucose feed. The addition of a hydrogenation reactor upstream of the reformer leads to α equal to nearly 6 for glucose reforming, showing an improvement of 290% in the net production of hydrogen per mole of glucose.

A novel aspect of the approach outlined here for the generation of hydrogen from aqueous solutions of glucose is that we have identified a beneficial synergy formed by operating the hydrogenation reactor, the reformer, and the gas–liquid separator (situated downstream of the reformer) at different temperatures while maintaining the total pressure of the system, P_tot, at a constant value. This situation is depicted schematically in Fig. 2. An aqueous solution of glucose is co-fed with gaseous H₂ to the hydrogenation reactor, which is operated at a relatively low temperature, T₁, to minimize glucose decomposition reactions in the liquid phase. The partial pressure of hydrogen, P_H2, in the reactor is equal to P_tot − P_H2O(T₁), where P_H2O(T₁) is the vapor pressure of water at T₁. For example, if P_tot is 54.8 bar and T₁ is 393 K, then P_H2 = 52 bar. This relatively high pressure of hydrogen is favorable for the conversion of glucose to sorbitol. The aqueous solution of sorbitol and gaseous H₂ are then fed to the reforming reactor, which is operated at the higher temperature, T₂ (e.g., T₂ = 538 K), necessary to convert sorbitol to H₂ and CO₂. Finally, the liquid and gaseous effluents from the reformer are cooled and sent to a separator, which is maintained at a low temperature, T₃. The sum of the partial pressures of H₂ and CO₂ in the separator is equal to P_tot − P_H2O(T₃). If the separator is at room temperature, then the sum of the H₂ and CO₂ pressures is essentially equal to P_tot, which is 54.8 bar. This high pressure of the reformate facilitates further removal of CO₂ from H₂ by pressure-swing adsorption. A fraction of the purified H₂ at high pressure can then be recycled to the hydrogenation reactor, and the remaining hydrogen may be directed to a fuel cell for conversion to electrical power.

Fig. 2 Schematic representation of the recycle loop for generation of hydrogen by aqueous-phase reforming of glucose.

Notes and references

1. J. Woodward, M. Orr, K. Cordray and E. Greenbaum, Nature, 2000, 405, 1014; L. Garcia, R. French, S. Czernik and E. Chornet, Appl. Catal., A, 2000, 201, 225.
2. R. D. Cortright, R. R. Davda and J. A. Dumesic, Nature, 2002, 418, 964; G. W. Huber, J. W. Shabaker and J. A. Dumesic, Science, 2003, 300, 2075; R. R. Davda, J. W. Shabaker, G. W. Huber, R. D. Cortright and J. A. Dumesic, Appl. Catal., B, 2003, 43, 13.
3. B. M. Kabyemala, T. Adschiri, R. M. Malaluan and K. Arai, Ind. Eng. Chem. Res., 1999, 38, 2888.
4. S. P. Bates and R. A. Van Santen, Adv. Catal., 1998, 42, 1.
5. J. W. Shabaker, R. R. Davda, G. W. Huber, R. D. Cortright and J. A. Dumesic, J. Catal., 2003, 215, 344.
6. P. Gallezot, P. J. Cerino, B. Blanc, G. Fleche and P. Fuertes, J. Catal., 1994, 146(1), 93.
7. R. R. Davda and J. A. Dumesic, Angew. Chem., Int. Ed., 2003, 42(34), 4068.

---

Footnote

† Electronic Supplementary Information (ESI) available: catalyst preparation and characterization; reaction kinetics studies; acknowledgements. See http://www.rsc.org/suppdata/cc/b3/b310152e/

---