Dimethyl ether steam reforming to produce H₂ over Ga-doped ZnO/γ-Al₂O₃ catalysts

Abstract

Herein, we report the performance of a series of gallium-doped zinc oxide (GDZ) catalysts mechanically mixed with γ-Al₂O₃ (GDZ/γ-Al₂O₃) for the dimethyl ether (DME) steam reforming (SR) to produce H₂. Compared with the ZnO catalyst mechanically mixed with γ-Al₂O₃, the doping of ZnO with gallium can significantly improve the conversion of DME and the yield of hydrogen. The catalyst with a Zn : Ga molar ratio of 9 : 1 has the highest conversion of DME (95.4%) and yield of hydrogen (95.0%) at 450 °C. The carbon dioxide selectivity of GDZ/γ-Al₂O₃ catalysts is higher than 95%, which is much higher than that of the CuZnAlO/γ-Al₂O₃ catalyst. Moreover, the GDZ/γ-Al₂O₃ catalysts have better stability than the CuZnAlO/γ-Al₂O₃ catalyst. Doping of ZnO with gallium generates a large number of oxygen vacancies on the catalyst, which is beneficial to the DME SR reaction. Our results indicate that HCOO– is an intermedium of DME SR over GDZ/γ-Al₂O₃ catalysts, and the transformation from HCOO– to CO₂ is the rate-controlling step. The results of conductivity indicate that the rate-controlling step is an n-type reaction.

---

1. Introduction

The increasing consumption of fossil fuels inevitably leads to both energy shortage and various environmental pollutions. So, the exploration and application of new clean energies are becoming more and more important. In recent years the fuel cells using H₂ as an energy source, such as the proton exchange membrane fuel cells (PEMFCs), have attracted a lot of attention due to their multiple advantages, e.g. high energy density, rapid start response, compact size and zero pollution.¹ To more cleanly and efficiently use the limited fossil fuels, the traditional fuels including hydrocarbons, alcohols and ethers are often converted to hydrogen through reforming processes. Among the reported reforming techniques, steam reforming (SR) operating at relatively low temperatures can generate more H₂ due to the efficient utilization of the hydrogen element in water. For instance, the dimethyl ether steam reforming (DME SR) can generate 6 times molar amounts of H₂ per molar DME, as described in eqn (1):

(CH₃)₂O + 3H₂O (g) → 6H₂ + 2CO₂, ΔH₂₉₈ = 135 kJ mol⁻¹ (1)

As a H₂ carrier material, DME possesses a lot of advantages, such as high H/C ratio, high energy density, no toxicity and no C–C bonds in DME molecule. The absence of C–C bonds in DME greatly decreases the carbon deposition on catalysts during DME SR. All of these advantages make DME SR technique very promising in areas of onboard hydrogen production. The DME SR reaction consists of two successive steps, namely DME hydrolysis to methanol (DME H) and methanol steam reforming (MSR) as expressed in eqn (2) and (3), respectively.

(CH₃)₂O + H₂O (g) → 2 CH₃OH, ΔH₂₉₈ = 37 kJ mol⁻¹ (2)

CH₃OH + H₂O → 3H₂ + CO₂, ΔH₂₉₈ = 49 kJ mol⁻¹ (3)

The DME hydrolysis is catalyzed by solid acid catalysts, such as bayerite-derived Al₂O₃, γ-Al₂O₃, ZSM-5, H-mordenite and TiO₂. The alumina-based solid acid catalysts show higher catalytic stability than the zeolite-based ones, since the strong acidity of zeolites can readily lead to the formation of coke, and decrease the activity and durability of the catalysts.^2–4

The MSR reaction requires redox catalysts, such as copper and zinc-based catalysts, especially the Cu-containing one.^5–9 Our previous work has reported that the CuZnAl_1−xM_xO (M = Zr or Ce) catalysts were active for MSR during DME SR.^5,6 The partial substitution of Al by Zr or Ce can increase the dispersion of copper species, improve the reducibility of the catalysts, and inhibit the sintering of Cu crystallites in the DME SR reaction.^5,6 Shimoda et al. reported that the CuFe₂O₄ spinel mixed with γ-Al₂O₃ exhibited high activity and stability for DME SR because of formation of a new active phase of Cu–Fe–AlO₄ for the MSR step.⁷ Although the reported Cu-based catalysts displayed high performance for DME SR, they also had some shortcomings, such as high CO selectivity, low thermal stability and short catalyst durability. Thus, it is necessary to develop copper-free catalyst systems for DME SR.

In methanol conversion reactions the CO₂ selectivity was highly related to the oxidation of formaldehyde. The literature reported that zinc oxide without copper was a good CO₂-selective catalyst in MSR, which gave the CO₂ selectivity as high as 95–99.6%.¹⁰ ZnO was beneficial to oxidization of the formaldehyde and the formaldehyde-derived intermediates toward CO₂, resulting in the high CO₂ selectivity.¹⁰ However, another study reported that considerable amounts of CH₄ and CO could be identified during DME SR over ZnO–Al₂O₃ catalysts by a co-precipitation method. The presence of vast Al³⁺ ions probably enhanced the direct decomposition of methanol and DME, producing more CO and CH₄.¹¹ Our previous work reported that the partial substitution of Al³⁺ (20%) by Ce⁴⁺ in ZnAlO mixed oxide catalyst can improve the performance of ZnAlO catalyst for DME SR, and the modified ZnAlCeO catalyst exhibited higher CO₂ selectivity and better start-off durability than ZnO and ZnAlO catalysts.¹² The introduction of Ce not only enhanced the dispersion of Zn species but also inhibited the transformation of ZnO to spinel ZnAl₂O₄. In addition, the interaction between ceria and zinc oxide may produce new active sites, such as the Ce⁴⁺–O–Zn²⁺ linkages.¹²

Intrinsic ZnO is an n-type semiconductor.¹³ According to the electron theory of catalysis on semiconductors, the catalytic activity of semiconductor is determined by the position of the Fermi level on the surface of the crystal, which can be varied to some extent by doping the surface or the bulk of the crystal with the properly chosen impurities in proper concentrations.¹⁴ Ga atoms can substitutionally enter the lattice of ZnO acted as donors,¹⁵ and it can raise the Fermi level on the surface of ZnO. Zinc oxide catalysts modified by gallium have been proved to be active for many reactions including water splitting,^16,17 DME SR¹⁸ and MSR reactions.¹⁹

Based upon the above analysis, we expected the doping of gallium could improve the catalytic performance of ZnO-based catalysts for DME SR. In this work a series of gallium-doped ZnO (GDZ) catalysts physically mixed with γ-Al₂O₃ (GDZ/γ-Al₂O₃) with different zinc/gallium molar ratios were designed and prepared. Our results show that the GDZ/γ-Al₂O₃ catalysts do display the remarkably improved catalytic performance, such as high DME conversion, high CO₂ selectivity and long high-temperature durability, as compared with the traditional CuZnAlO catalyst physically mixed with γ-Al₂O₃. By using multiple techniques including XRD, TPD-MS-CH₃OH, in situ FTIR of CH₃OH, X-ray photoelectron spectroscopy (XPS) and four-probe direct current (DC) method, the catalysts were reasonably characterized, and the structure–performance relationship and the reaction pathways of CH₃OH reforming were discussed, as well.

2. Experimental

2.1 Catalyst preparation

The GDZ catalysts with different Zn/Ga atomic ratios were prepared by co-precipitation method, using sodium carbonate as precipitant. Zn(NO₃)₂·6H₂O and Ga(NO₃)₃ solution (purchased from Alfa Aesar) were used as source materials. A Na₂CO₃ solution with [CO₃²⁻] = 0.5 M and a mixed solution of Zn and Ga nitrates with the total cations concentration of 1.0 M were prepared, respectively. Then these two solutions were simultaneously dropped into 100 mL deionized water by keeping the pH around 7.5 under vigorous mechanical stirring at the temperature of 60 °C. After aged at the same temperature for 2 h, the obtained precipitate was filtered and washed by deionized water for three times, then dried overnight at 120 °C and finally calcined at 500 °C for 4 h. According to the Zn/Ga molar ratio, the prepared catalysts were denoted as ZnO, Zn_9.5Ga_0.5O, Zn₉Ga₁O, Zn₈Ga₂O and Zn₇Ga₃O, respectively. For comparison, a conventional CuZnAlO catalyst was also prepared, the details of which can be found elsewhere.⁶

Before used for the measurements of activity and durability, the GDZ and CuZnAlO catalysts were mechanically mixed with γ-Al₂O₃ (supplied by Tianjin Chemical Research & Design Institute, calcined at 600 °C for 2 h, SSA = 189 m² g⁻¹) with the mass ratio of 2 : 1 (GDZ or CuZnAlO to γ-Al₂O₃), and the catalysts were made into small pellets with the size of 40–60 mesh. The total acidity of γ-Al₂O₃ calculated by the temperature-programmed desorption of ammonia (NH₃-TPD) is 50 μmol g⁻¹. The catalysts were denoted as GDZ/γ-Al₂O₃ or CuZnAlO/γ-Al₂O₃.

2.2 Catalyst characterization

The specific surface area (SSA) of the catalysts was measured by nitrogen adsorption at −196 °C on a Quantachrome QuadraSorb SI instrument. Before the measurements, the catalysts were degassed under vacuum at 300 °C for 3 h. The specific surface area was calculated using standard BET method in 0–0.3 partial pressure range.

The crystal structure of the catalysts was characterized by X-ray diffraction (XRD). The tests were performed on Rigaku D/Max-2500 X-ray diffractometer operating at 200 mA and 40 kV, using Cu Kα (λ = 0.15418 nm) as radiation source. The diffraction data in the range of 2θ from 10° to 90° were collected at a speed of 8° min⁻¹, and the slow scan were collected at a speed of 1° min⁻¹ from 30° to 40°.

The temperature-programmed desorption of CH₃OH (TPD-MS-CH₃OH) was performed on a TPDRO instrument (Xian Quan TP-5079) connected with a mass spectrometer (Hiden HPR20). 50 mg catalyst powder (40–60 mesh) was put into the quartz tube reactor and pretreated at 400 °C for 1 h under 8% H₂/N₂ stream (30 mL min⁻¹), then the feed gas was switched to pure helium flow (99.999%, 30 mL min⁻¹) for 0.5 h at the same temperature to purge the gaseous and adsorbed H₂. After the temperature cooled down to 60 °C in the same helium flow, pulses of CH₃OH (1.0 μL) were injected into the catalyst bed until the MS signal got stable. Subsequently, the temperature-programmed desorption was conducted from ambient temperature to 700 °C at a heating rate of 10 °C min⁻¹ in the same helium flow (99.999%, 30 mL min⁻¹).

To evaluate the quantity of methanol adsorbed on the surface of the catalysts, the CH₃OH pulse adsorption was also performed on the TP-5079 with a thermal conductivity detector. Similar to the test of TPD-MS-CH₃OH, the catalysts were first pretreated with 8% H₂/N₂ and purified with helium before measurement. Then, the pulses of CH₃OH (1.0 μL) were rapidly injected into the tube every 0.5 h until the detector signal got stable at 60 °C.

The in situ Fourier-transform infrared (FTIR) spectroscopies of CH₃OH were performed on a Nicolet Nexus spectrometer equipped with a MCT detector cooled by liquid nitrogen. 20 mg catalyst was ground into fine powder and pressed into a thin wafer, which was then placed in a quartz IR cell connected to a high vacuum system. The catalyst was in situ pretreated for 1 h at 400 °C in 5% H₂/N₂ mixture gas flow with a rate of 50 mL min⁻¹. The background spectra were collected in vacuum. Subsequently, the catalyst was exposed to CH₃OH at 50 °C in N₂ flow to achieve adsorption saturation. The FTIR spectra were recorded at the desired temperatures from 50 to 400 °C.

The conductivities of the catalysts were measured with a four-probe direct current (DC) method with an electrochemical workstation (VersaSTAT 3, Ametek). The powders were pressed at 18 MPa and made into rectangular bars with a typical size of 25 × 7 × 1.5 mm and then calcined at 500 °C for 3 h to form dense bars. Four silver wires with silver paste were used as electrodes. Before the test, the bars were reduced in H₂ (50 mL min⁻¹) at 400 °C for 0.5 h, and then the conductivities of the catalysts were measured in H₂ atmosphere.

The X-ray photoelectron spectra were analyzed using an X-ray photoelectron spectroscopy (XPS) (K-alpha X, Thermo Fisher co., USA). The binding energies were referenced to the C_1s line at 284.6 eV from adventitious carbon.

The procedures of NH₃-TPD and thermal gravimetric analysis (TG) can be obtained in ESI.†

2.3 Catalytic activity and durability tests

The evaluation of catalytic activity and durability of the catalysts was performed in a continuous flow fixed-bed reactor under atmospheric pressure. Before activity evaluation, 500 mg catalyst was loaded in the quartz reactor and in situ reduced at 400 °C for 0.5 h in 10% H₂/N₂ (50 mL min⁻¹). The feedstock for activity and durability tests is composed of N₂ (40%), H₂O vapor (50%) and DME (10%) with the steam/carbon molar ratio (S/C) fixed at 2.5, showing a space velocity of 12 000 mL h⁻¹ g_cat⁻¹. The external and internal mass transfer limitations have already been excluded by variation of the linear flow rate over the catalyst and the catalyst particle size, respectively. A mixture gas of DME and N₂ was directly fed to the reactor, and the water was vaporized at 150 °C before entering the reactor. For activity measurements the reaction temperature varied in the range of 370–450 °C, while for durability tests the reaction temperature was fixed at 440 °C. The compositions of the influent and effluent gases containing DME, CH₄, CH₃OH, N₂, CO, CO₂ and H₂ were analyzed by an Agilent 7890A chromatograph (GC). DME, CH₄ and CH₃OH were separated by a PoraPak Q capillary column and analyzed by a flame ionization detector (FID), while N₂, CO and CO₂ were separated by a PoraPak N column and a MoleSieve 5A column, and analyzed by a thermal conductivity detector (TCD). H₂ was analyzed by TCD with N₂ as carrier gas. The DME conversion (X_DME), H₂ yield (y_H2) and the selectivity (S_Ci) to C_i (including CO₂, CO and CH₄) are defined as follows:here, DME_in, and DME_out stand for the molar flow rate of DME in feed gas and product gas. F_H2 represents the effluent molar flow rate of H₂, and F_Ci is the molar flow rate of the C_i products.

3. Results and discussion

3.1 Catalysts characterization

3.1.1 Texture and bulk structure of the catalysts. The XRD patterns of the catalysts are displayed in Fig. 1(a). The pattern of the ZnO catalyst shows peaks matching with the zincite structure [JCPDS 36-1451]. The same peaks were also observed on the GDZ catalysts, and no other phase was detected. Additionally, compared with the pure ZnO catalyst, the intensity of the XRD peaks are noticeably reduced with the increase of the Ga content in the catalysts. It suggests the gradually decreased particle size and crystallinity. By applying the Scherrer equation, the average crystallite size has been estimated to be 45.5 nm for the pure ZnO, 11.7, 9.0, 4.4 and 3.9 nm for the catalysts Zn_9.5Ga_0.5O, Zn₉Ga₁O, Zn₈Ga₂O and Zn₇Ga₃O, respectively.

Fig. 1 XRD patterns in the region of (a) 10° to 75° with a speed of 8° min⁻¹ and (b) 30° to 40° with a speed of 1° min⁻¹: (1) ZnO, (2) Zn_9.5Ga_0.5O, (3) Zn₉Ga₁O, (4) Zn₈Ga₂O and (5) Zn₇Ga₃O.

Furthermore, Fig. 1(b) shows the slow scan XRD patterns. Doping of ZnO with Ga causes the shift of the XRD peaks towards small angles. The peak corresponding to (002) plane shifted from 34.48 to 34.18°, and the peak corresponding to (101) plane shifted from 36.31 to 36.15°. The shift of the diffraction peaks indicates the GDZ catalysts still keep the zincite structure and Ga is successfully incorporated into the crystal lattice causing the lattice distortion. Wei et al. reported that the interplanar spacing of the (101) plane in Ga-doped ZnO catalysts shifted from 0.2468 to 0.2473 nm.²⁰ According to the semiconductor theory, the Ga cations in the zincite structure probably exist in the form of Ga³⁺·e, which can increase the conductivity and the electron-donating ability of the catalyst. A weak diffraction peak of gallium oxide (Ga₂O₃ [JCPDS 06-0509]) appears in the Zn₈Ga₂O pattern, and it becomes stronger in the Zn₇Ga₃O pattern. It indicates that partial of Ga exists in the form of Ga₂O₃ on the GDZ catalysts.

The results of the specific surface areas of the catalysts are summarized in Table 1. The surface areas are enlarged with the Ga contents increased.

Table 1 Specific surface (SSA), average crystallite size (ACS), CH₃OH adsorption amounts (MAA), apparent activation energy (E_a) and TOF of different catalysts

Catalyst	SSA (m^2 g^−1)	ACS (nm)	MAA^a (mmol g^−1)	Ea^b (kJ mol^−1)	TOF^c (s^−1)
a Methanol adsorption amounts (MAA) was calculated from the area of the peaks in the CH3OH pulse adsorption profiles.b Apparent activation energy (Ea) was estimated from Arrhenius equation based on the activity over all the catalysts (50 mg) at 1 atm total pressure (space velocity of 12 000 mL h^−1 gcat^−1. 10% DME, 50% H2O and 40% N2).c Turnover frequency (TOF) was estimated based on the MAA results and the activity tests over all the catalysts (50 mg) at 430 °C and 1 atm total pressure (space velocity of 12 000 mL h^−1 gcat^−1. 10% DME, 50% H2O and 40% N2).
ZnO	9.2	45.5	0.05	—	—
Zn9.5Ga0.5O	57.0	11.7	0.31	213.6 ± 4.4	0.013
Zn9Ga1O	68.8	9.0	0.34	197.4 ± 2.7	0.019
Zn8Ga2O	80.2	4.4	0.33	203.6 ± 11.0	0.016
Zn7Ga3O	93.2	3.9	0.35	210.2 ± 5.9	0.015

---

3.1.2 Electrical conductivity of the catalysts. Doping of semiconductor can change the position of the Fermi level on the surface and in the bulk. The position of Fermi level will affect both the electrical conductivity and the catalytic activity, simultaneously.¹⁴ Therefore, there is a correlation, which is related to the type of reaction (n-type or p-type reaction) and the type of semiconductor (n-type or p-type), between the catalytic activity and the electrical conductivity.¹⁴ ZnO is an n-type semiconductor. Generally, Ga can be doped into the ZnO crystal lattice as a donor impurity and be considered in the form of Ga³⁺·e, which will increase the conductivity of GDZ catalysts. However, excessive Ga will also lead to the formation of Ga₂O₃, reducing the conductivity of GDZ catalysts. The conductivity data of the catalysts are displayed in Fig. 2. The conductivity of all of the GDZ catalysts increase with the rising temperature, and the Zn₉Ga₁O catalyst shows the highest conductivity. The data of the ZnO catalyst are not given, because the undoped ZnO has unstable electrical properties.¹⁵ These results indicate that Ga atoms have been doped into ZnO crystal lattice and lead to the increase of conductivity, which is in accordance with the XRD results.

Fig. 2 The conductivity of the GDZ catalysts.

3.1.3 TPD-MS-CH₃OH. Methanol is an intermediate product in DME SR, and its adsorption and activation on catalyst surface is rather important. The methanol adsorption behaviour on the catalyst surface was investigated by TPD-MS-CH₃OH. Table 1 lists the methanol adsorption amounts (MAA) of the catalysts measured by pulse adsorption. In the TPD-MS-CH₃OH tests, the MS signals of the desorbed CH₃OH, H₂, CO₂ and H₂O were simultaneously monitored, as shown in Fig. 3(a–d), respectively. Fig. 3(a) shows that the desorption of CH₃OH from the pure ZnO takes place at around 100 °C with a small peak area, suggesting that only small amount of CH₃OH can be adsorbed on ZnO, similarly to the previous report.¹² The desorption signals of CH₃OH from the GDZ catalysts exhibits two desorption peaks with much larger peak areas than that from ZnO. The peak in the range of ∼60–200 °C is attributed to the desorption of molecularly adsorbed methanol, and the top of the peak gradually increase from 98 to 115 °C with the increase of Ga in the catalyst, indicating doping of Ga raises the ability for methanol adsorption on the catalyst. The signal of desorption of CH₃OH above 200 °C mainly comes from the rehydrogenation of methoxy species by hydroxyl groups.²¹ No methanol was desorbed on ZnO above 200 °C in Fig. 3(a). Probably, there is no chemically adsorbed methanol on ZnO under the present experimental conditions. It can also be verified by the CO₂ desorption (Fig. 3(b)) and H₂ desorption (Fig. 3(c)) that there is also no CO₂ and H₂ desorption peak for ZnO. For comparison, we carried out a TPD-MS-CH₃OH experiment with adsorption of CH₃OH on ZnO at 100 °C. The corresponding results are plotted in Fig. S1.† In Fig. S1,† the desorption signal of CH₃OH, CO₂ and H₂ can be clearly observed. It indicates that methanol is hardly chemically adsorbed over ZnO at 60 °C until the adsorption temperature increases to 100 °C. Nevertheless, doping Ga into ZnO is an efficient way to enhance the chemical adsorption of methanol, which occurs even at 60 °C, as shown in Fig. 3. The desorption profiles of CO₂ (Fig. 3(b)) show that the initial desorption temperature of CO₂ of the GDZ catalysts is at about 260–280 °C. With the increase of Ga in the catalyst, the temperature of the desorption peak shifted from 336 (Zn_9.5Ga_0.5O) to 326 °C (Zn₉Ga₁O) and then raises to 338 °C (Zn₇Ga₃O). In Fig. 3(c), the temperature for H₂ desorption from the catalysts has a similar trend with the CO₂ desorption. The H₂ desorption of the Zn₉Ga₁O catalyst shows the lowest temperature. The initial temperature of H₂ desorption is at about 230 °C for the GDZ catalysts, which is lower than that of CO₂ (about 260–280 °C). Moreover, the temperature of the H₂ desorption peak is also lower than CO₂. Thus, it can be deduced that the formation or desorption of CO₂ might be the rate-controlling step of DME SR. Two H₂O desorption peaks are observed in Fig. 3(d). The peak below 400 °C is attributed to the desorption of molecular H₂O formed during CH₃OH adsorption, and the peak above 400 °C is ascribed to the water produced from the interaction of surface hydroxyls.¹² The surface hydroxyls play a vital role in the reaction from CH₃OH to CO₂ and H₂. Chadwick and Zheng found that over the dehydroxylated ZnO catalyst the dominant reaction was the decomposition of methanol to form CO and H₂ rather than the MSR to form CO₂ and H₂.²² In addition, the formation of hydrogen often takes place via the interaction of surface hydroxyl groups.²³ Herein, above 400 °C the GDZ catalysts show the larger peak area of water desorption signals than that of ZnO, indicating that the former may have higher catalytic activities for DME SR.

Fig. 3 The signals of desorption of (a) CH₃OH, (b) CO₂, (c) H₂ and (d) H₂O in TPD-MS-CH₃OH over (1) ZnO, (2) Zn_9.5Ga_0.5O, (3) Zn₉Ga₁O, (4) Zn₈Ga₂O and (5) Zn₇Ga₃O.

3.2 Activity and stability

3.2.1 Activity on DME SR. Fig. 4 shows the conversion of DME, the yield of hydrogen and the selectivity of CO and CO₂ of the catalysts as a function of the reaction temperature. Zn₉Ga₁O/γ-Al₂O₃ exhibits the best catalytic performance as compared with others. At 400 °C, the GDZ/γ-Al₂O₃ catalysts exhibit higher DME conversion (13–20%) than that of pure ZnO/γ-Al₂O₃ (∼10%). With the temperature increasing to 450 °C, Zn₉Ga₁O/γ-Al₂O₃ displays a DME conversion as high as 95%, which is much higher than that for pure ZnO (65%). Several assumptions to estimate the apparent activation energy of DME SR over different catalysts were put up by Eguchi et al.²⁴ namely: (1) it is a pseudo-first-order reaction, (2) the reactor is an isobaric plug flow reactor, and (3) side reactions are neglected. Based on these assumptions, corresponding Arrhenius plots for DME SR over this series of catalysts are obtained (Fig. S2†). The good linearity of Arrhenius plots indicates that the assumptions are acceptable for the GDZ/γ-Al₂O₃ catalysts. The obtained apparent activation energy (E_a) and turnover frequency (TOF) of different catalysts are listed in Table 1. Zn₉Ga₁O/γ-Al₂O₃ shows the lowest E_a (197.4 kJ mol⁻¹) and highest TOF (0.019 s⁻¹) values, which is in accordance with the highest DME conversion. The hydrogen yield (y_H2) exhibits the same trend with the DME conversion as seen in Fig. 4(b), the values of the H₂ yield and the DME conversion are similar, indicating that DME almost transform to H₂ and CO₂. During DME SR, some side reactions might occur,^4–6 and the main by-products are CO and CH₄. As shown in Fig. 4(c), all catalysts exhibit the CO selectivity less than 5% in the whole reaction temperature range, except Zn₇Ga₃O. The selectivity of CO is much lower than that of our previously reported CuZnAlO based catalysts.^5,6 The small amount of CO may come from some potential side reactions such as reverse water gas shift reaction (r-WGS) and methanol decomposition. Moreover, CH₄ was not found in the product. Thus, except Zn₇Ga₃O, all catalysts exhibit higher CO₂ selectivity than 95% in the whole reaction temperature range, as shown in Fig. 4(d).

Fig. 4 (a) Conversion of DME, (b) yield of H₂, (c) selectivity of CO and (d) selectivity of CO₂ for the DME SR reaction.

With the increase of Ga content, the conversion of DME and the yield of hydrogen increased and then decreased. The XRD results showed that Ga was doped into the crystal lattice of ZnO. As shown in Fig. 2, doping of Ga had caused the change of electrical conductivity. The relationship between electrical conductivity and catalytic performance is displayed in Fig. 5. Compared with the change of the conductivity of GDZ catalysts, the change of the conversion of DME and the yield of hydrogen displayed a similar trend. The electron theory of catalysis on semiconductor indicated in the case of an n-type reaction on an n-type semiconductor, the electrical conductivity and the activity in the same direction.¹⁴ Our results indicate that the rate-controlling step of DME SR over GDZ/γ-Al₂O₃ catalysts is an n-type reaction. It can be accelerated by doping donor in the lattice of n-type semiconductor of ZnO, which can enhance the donor ability. The rate-controlling step may change with the increase of donor. It indicates that a proper electrical conductivity is helpful to improve the activity and selectivity of the catalyst. In this work, the highest electrical conductivity corresponds to the highest catalytic activity. It indicates that increasing the conductivity of GDZ catalysts probably further enhance the activity of the catalyst. Moreover, as listed in Table 1, with the increase of the amount of gallium, the specific surface areas of the GDZ catalysts significantly enlarged. Thus, although the conductivity of Zn₇Ga₃O and Zn₈Ga₂O was lower than that of Zn_9.5Ga_0.5O, the formers still presented the higher catalytic activity than the later, since the specific surface areas of Zn₈Ga₂O (80.2 m² g⁻¹) and Zn₇Ga₃O (93.2 m² g⁻¹) are much larger than that of Zn_9.5Ga_0.5O (57 m² g⁻¹).

Fig. 5 The trend of conversion of DME, yield of H₂ and conductivity.

Understanding the formation and desorption of surface species is helpful to the improvement of catalyst. According to the results of TPD-MS-CH₃OH, the initial desorption temperature of hydrogen on the GDZ catalysts is about 230 °C, which is lower than that of CO₂ (about 260–280 °C). The top temperature of the H₂ desorption peak is also lower than that of CO₂. We found that the trend of the conversion of DME, as well as the yield of hydrogen, is similar to the trend of the CO₂ desorption temperature. The lower desorption temperature of CO₂ matches the higher activity and selectivity. Thus, we deduced that the formation of CO₂ might be the rate-controlling step of DME SR using GDZ/γ-Al₂O₃ as catalysts. The signals of CO₂ and H₂ suggest that Zn₉Ga₁O has the highest activity to convert CH₃OH to H₂ and CO₂, as shown in Fig. 4(b and c). Enhancing the ability of GDZ catalysts to convert CH₃O– into CO₂ will contribute to the improvement of the catalytic activity of DME SR.

3.2.2 Durability test on the Zn₉Ga₁O/γ-Al₂O₃ and CuZnAlO/γ-Al₂O₃ catalysts. To investigate the stability of this series of catalysts for DME SR, the Zn₉Ga₁O/γ-Al₂O₃ catalyst was selected for the prolonged durability test, which is carried out at 440 °C for 48 h. Meanwhile, an identical experiment was performed on the conventional CuZnAlO/γ-Al₂O₃ catalyst for a direct comparison. The results are shown in Fig. 6. For the CuZnAlO/γ-Al₂O₃ catalyst, the DME conversion (around 98%) shows little change in the first 8 h, but after this period it declines rapidly, giving the final conversion of 62% at 48 h. For the catalyst Zn₉Ga₁O/γ-Al₂O₃, it possesses an initial DME conversion of ∼90%, a little lower than that of CuZnAlO/γ-Al₂O₃, but it decreases more slowly, showing the conversion of 73% in the end. The total decreased DME conversion over the catalyst Zn₉Ga₁O/γ-Al₂O₃ during 48 h test is about 17%, much lower than that on the CuZnAlO/γ-Al₂O₃ catalyst (36%). We deduce that the good stability of the GDZ/γ-Al₂O₃ catalysts probably originates from its high thermal stability and the ability of resistance of coke deposition. As for the CuZnAlO/γ-Al₂O₃ catalyst, the catalytic deactivation mainly results from the readily sintered metallic Cu species and the coke deposition on the catalyst at the temperature above 300 °C, as indicated by our XRD and TG results (Fig. S3 and S4†) in ESI.

Fig. 6 The stability of DME SR over the Zn₉Ga₁O/γ-Al₂O₃ and CuZnAlO/γ-Al₂O₃ catalysts.

3.3 Mechanism of DME SR over GDZ/γ-Al₂O₃ catalyst

3.3.1 XPS results. In order to identify the change of the distribution and content of surface oxygen species caused by Ga doping into zinc oxide, X-ray photoelectron spectroscopy (XPS) was employed. Fig. 7 shows the oxygen 1s (O 1s) peaks of ZnO and Zn₉Ga₁O catalysts. The O 1s signals of the Zn₉Ga₁O catalyst pre-reduced in 5% H₂/N₂ for 0.5 h before XPS test is also given in Fig. 7. The typical O 1s peak can be consistently fitted by three nearly Gaussian, centered at 532.3, 530.8 and 531.5 eV, respectively.^25–27 The high binding energy component located at 532.3 eV is usually attributed to the adsorbed O²⁻.²⁵ The component on the low binding energy side of O 1s spectrum (530.8 eV) is attributed to the bulk O²⁻.²⁶ The medium binding energy component centered at 531.5 eV is associated with O²⁻ ions in the oxygen deficient regions with the matrix of ZnO.²⁷ So, changes in the intensity of this component may be connected in part to the variations in the concentration of oxygen vacancies. In this work, Fig. 7 shows that the O 1s peaks of the Zn₉Ga₁O catalyst slightly shifted towards higher binding energy, as compared with ZnO. The peak shift is attributed to the difference in the electronegativity of Zn and Ga. The valence electron density of Zn and O in the Zn–O–Ga bond in the Zn₉Ga₁O catalyst becomes lower than that in the Zn–O–Zn bond in the ZnO catalyst because of the higher electronegativity of Ga than that of Zn. As a result, the interaction between Zn and O is weakened, which leads to the increases in the binding energy of O 1s peaks.^26,27

Fig. 7 The XPS spectra in the O 1s region of the ZnO and Zn₉Ga₁O catalysts: (1) ZnO, (2) Zn₉Ga₁O and (3) Zn₉Ga₁O pretreated in 8% H₂/N₂ for 0.5 h before XPS test.

In order to compare the cases of the GDZ and ZnO catalysts, the relative intensities of the Zn₉Ga₁O, pretreated Zn₉Ga₁O and ZnO catalysts are calculated and are listed in Table 2. As described in Table 2, the Zn₉Ga₁O and pretreated Zn₉Ga₁O have higher values of the medium binding energy component than ZnO, confirming that the oxygen vacancy increased after the Ga element incorporated into zinc oxide crystal lattice. After the Zn₉Ga₁O catalyst was pretreated by H₂, the percentage of lattice oxygen decreased continuously, and the oxygen vacancy increased. The oxygen vacancy plays an important role in the adsorption of reactants and it is very important for DME SR.

Table 2 Relative intensity of O 1s peaks of ZnO and Zn₉Ga₁O catalysts, evaluated by XPS

Catalyst	Fraction O state
	532.3 eV (%)	531.5 eV (%)	530.8 eV (%)
a The Zn9Ga1O was reduced in 8% H2/N2 for 0.5 h before characterized by XPS measurement.
ZnO	44.5	0	55.4
Zn9Ga1O	42.9	9.3	47.8
Zn9Ga1O^a	56.8	21.4	21.8

---

To confirm the effect of oxygen vacancies, the catalytic tests of the Zn₉Ga₁O/γ-Al₂O₃ pretreated in different atmospheres were also evaluated. As shown in Fig. 8, at 425 °C the conversion of DME over the pre-reduced Zn₉Ga₁O/γ-Al₂O₃ catalyst can reaches 67.8%, while it is only 48.5% and 44.3% over the Zn₉Ga₁O/γ-Al₂O₃ catalyst pretreated in pure N₂ and the fresh one, respectively. Apparently, the pre-reduction treatment can significantly improve the catalytic activity of Zn₉Ga₁O/γ-Al₂O₃. So, it indicates that the increase of the number of oxygen vacancies can improve the catalytic activity for DME SR over the GDZ/γ-Al₂O₃ catalyst.

Fig. 8 The conversion of DME over the Zn₉Ga₁O/γ-Al₂O₃ catalyst under different pretreatments: (1) pre-reduced in 8% H₂/N₂ mixture gas, (2) pretreated in pure N₂, (3) un-pretreated.

3.3.2 In situ FTIR of CH₃OH results. In order to infer the reaction route, the thermal evolution of the adsorbed CH₃OH and other species on the Zn₉Ga₁O catalyst surface was monitored by in situ FTIR spectroscopy. The temperature-dependent IR spectra and the assignments of bands in the spectra are shown in Fig. 9 and Table 3, respectively. At 50 °C, a series of the characteristic peaks corresponding to molecularly adsorbed methanol can be observed. The peaks at 3709 cm⁻¹ and 3664 cm⁻¹ are assigned to the O–H stretching vibration [ν (OH)] of methanol molecules.²⁸ The signal at 3006 cm⁻¹ belongs to the C–H stretching vibration in the methyl group [ν (CH₃)] of methanol molecules,²⁹ and the band around 1032 cm⁻¹ is attributed to the C–O stretching vibration [ν (CO)] of CH₃OH molecule.²¹ Additionally, methoxy species are also detectable, with the bands appearing at 3000–2800 cm⁻¹ (2961, 2926, 2868, 2821 cm⁻¹), ∼1453 cm⁻¹ and 1100–1000 cm⁻¹ (1070 and 1030 cm⁻¹), which correspond to ν (CH),^29,30 δ_s (CH₃)²⁹ and ν (CO),^21,30 respectively. The detected methoxy species at the low temperature of 50 °C are mainly assigned to the methoxy in methanol molecule. With the increase of temperature to 100 °C, the absorption intensity of adsorbed methanol molecule is weakened. A broad band in the region of 3600–3300 cm⁻¹ assigned to the ν (OH) of adsorption H₂O²⁹ and an absorption band at ∼1610 cm⁻¹ assigned to the δ (HOH) of adsorption H₂O appeared.³¹ Negative signals between 3750 and 3580 cm⁻¹ also appeared, indicating that a perturbation or a consumption of the surface –OH groups on the catalyst occurred after the methanol adsorption. These results indicate that the hydroxyl groups on the surface of GDZ catalyst react with methanol to give methoxy and water molecules, which is in accordance with the results of TPD-MS-CH₃OH. With the temperature further increased to 150 °C or higher, the methoxy species gradually decrease. When the temperature reaches 200 °C, some new bands emerge at the positions of 2961, 1580, and 1369 cm⁻¹, which can be attributed to formate species. More specifically, the peak at 2961 cm⁻¹ is ascribed to stretching of carboxylate and bending vibration of C–H mode (HCOO–).³⁰ The bands at about 1580 cm⁻¹ can be assigned to ν_as (COO) and 1369 cm⁻¹ assigned to ν_s (COO), respectively.³² With the temperature increasing to 250 °C the intensity of the absorption bands for formate species is strengthened, suggesting the transformation of more CH₃O– to HCOO–. At this temperature the peaks of HCOO– shift to lower wavenumbers, and wavenumber of the ν_as (COO) shifts from 1580 cm⁻¹ to 1568 cm⁻¹. Gaseous CO₂ peaks centered at 2372 and 2337 cm⁻¹ emerged at 300 °C.³¹ This means that the formate species and carbonate species turned into carbon dioxide, and this is coincident with the TPD-MS-CH₃OH. The appearance of gaseous CO₂ absorption bands further demonstrates that the formed intermedium HCOO– species prefers to turn into CO₂. The amount of HCOO– species increases with the temperature elevated from 250 to 300 °C. Above 300 °C, the amount of HCOO– species decreases with the temperature further increasing, but the intensity of gaseous CO₂ is continuously strengthened, which suggests that more and more HCOO– species are transformed into CO₂.

Fig. 9 In situ FTIR spectra of adsorbed CH₃OH at different temperatures over the Zn₉Ga₁O catalyst: (1) 50 °C, (2) 100 °C, (3) 150 °C, (4) 200 °C, (5) 250 °C, (6) 300 °C, (7) 325 °C, (8) 350 °C, (9) 375 °C and (10) 400 °C.

Table 3 In situ FTIR assignments of surface species of adsorbed CH₃OH over Zn₉Ga₁O catalyst

Groups	Vibrational mode	Wavenumber (cm^−1)	Ref.
H2O	δ (H2O)	1610	31
–OH	ν (OH)	3664, 3709	28
		3600–3300	29
CH3O–	ν (CO)	1032	21
		1100–1000	21 and 30
	νas (CH3)	3006	29
		2961, 2926	30
	νs (CH3)	2868	29
		2821	30
	δs (CH3)	1453	29
HCOO–	δ (CH)	2961	30
	νs (COO–)	1369	32
	νas (COO–)	1580	32
CO2	Physisorbed CO2	2372, 2337	31

---

3.3.3 The pathway of DME SR over the GDZ/γ-Al₂O₃ catalyst. The GDZ catalysts mechanically mixed with γ-Al₂O₃ have been proved to be active for the DME SR reaction, especially the extremely high CO₂ selectivity. In order to understand DME SR over the catalysts, we deduce a reaction mechanism according to the results of XPS, in situ FTIR of CH₃OH and the TPD-MS-CH₃OH. The proposed mechanism is shown in Fig. 10. It has been proved that the DME SR reaction consists of two successive steps, namely DME hydrolysis to methanol (DME H) (eqn (2)) and methanol steam reforming (MSR) (eqn (3)). In this work, the DME hydrolysis is catalyzed by γ-Al₂O₃, which acted as a solid acid catalyst. We mainly focus on the MSR step. As shown in XPS, the reduced GDZ catalyst has a lot of oxygen vacancies under reducing atmosphere, and these oxygen vacancies usually are considered as the active sites for the chemisorption of methanol and water. The in situ FTIR shows that when the temperature reaches 100 °C, methoxy species and water appear. The signals of desorbed water also have been observed in TPD-MS-CH₃OH above 50 °C. We infer that methanol molecules are adsorbed on the oxygen vacancies of GDZ catalyst, and it dissociate to methoxy species and adsorbed hydrogen atoms (H_a) (eqn (7)), then the adsorbed hydrogen atoms react with the surface hydroxyls to form water molecules. The dissociation of methanol molecules to methoxy species is a fast step since it can occur at a low temperature. We infer that the adsorbed water molecules can also dissociate to form the adsorbed hydroxyls (HO_a) and adsorbed hydrogen atoms (eqn (8)) on the oxygen vacancies, and this process is a fast reversible process due to the easy desorption of water. As shown in Fig. 4(d), we infer that the adsorbed hydroxyls can further dissociate to form the adsorbed hydrogen atoms and adsorbed oxygen atoms (eqn (9)). The adsorbed hydrogen atoms can combine to form the hydrogen molecules (eqn (10)), which have been detected by TPD-MS-CH₃OH above 220 °C.

CH₃OH + □ → CH₃O_a + H_a (7)

H₂O + □ → HO_a + H_a (8)

HO_a → O_a + H_a (9)

2H_a → H₂ (10)

□: oxygen vacancy.

Fig. 10 The proposed mechanism for DME SR over the GDZ/γ-Al₂O₃ catalyst.

The appearance of signals of HCOO– in in situ FTIR at 200 °C proved that the CH₃O– species transformed into HCOO–. We infer the equations (eqn (11) and (12)) for this process.

The CO₂ signals appeared in TPD-MS-CH₃OH test at about 270 °C. Moreover, at 300 °C, new bands corresponding to gaseous CO₂ (2372, 2337 cm⁻¹) are observed in in situ FTIR spectra. The appearance of the gaseous CO₂ demonstrates that the formed intermedium HCOO– prefers to turn into CO₂ (eqn (13)), which can provide a strong evidence for the high CO₂ selectivity of DME SR on the GDZ/γ-Al₂O₃ catalysts (Fig. 4(d)).

We found that CO₂ was generated at the highest temperature in all steps. Meanwhile, the temperature of CO₂ desorption showed in TPD-MS-CH₃OH has a similar trend with the conversion of DME. Thus, we deduce that the transformation from HCOO– to CO₂ is the controlling step for MSR over the GDZ catalyst. As discussed in 3.2.1, the rate-controlling step of DME SR over GDZ/γ-Al₂O₃ catalysts is an n-type reaction. Therefore, we deduce the transformation from HCOO– to CO₂ is an n-type reaction, which can be accelerated by doping donor in the lattice of n-type semiconductor, and we have accelerated the transformation from HCOO– to CO₂ by the doping of ZnO with Ga.

4. Conclusion

A series of gallium-doped zinc oxide (GDZ) mixed with γ-Al₂O₃ catalysts for DME SR to produce H₂ was prepared. Compared with the conventional catalyst, GDZ/γ-Al₂O₃ catalysts exhibited higher carbon dioxide selectivity and better stability. XRD results indicated that Ga atoms incorporated into the crystal lattice of ZnO. With the increase of gallium content, the conductivity of GDZ catalysts increased and then decreased, similar with the conversion of DME trend and the yield of hydrogen trend. When the molar ratio of Ga : Zn is 1 : 9, the GDZ catalyst has the highest conductivity, and the corresponding Zn₉Ga₁O/γ-Al₂O₃ catalyst has the highest conversion of DME (95.4%) and yield of hydrogen (95%). The results of conductivity indicate that the rate-controlling step of DME SR over the GDZ catalysts is an n-type reaction. XPS results indicated that doping of gallium introduced a large number of oxygen vacancies into the catalyst, and the vacancies were beneficial to the reaction. According to the results of XPS, TPD-MS-CH₃OH, in situ FTIR of CH₃OH and the proposed mechanism with HCOO– as an intermedium, we speculated that the formation of carbon dioxide might be the rate-controlling step for DME SR over GDZ/γ-Al₂O₃ catalysts. We deduce the transformation from HCOO– to CO₂ is an n-type reaction. The good conversion of DME, high selectivity of CO₂ and high stability of GDZ/γ-Al₂O₃ catalysts makes it promising for DME SR both in research and industry.

Acknowledgements

This work is financially supported by National Natural Science foundation of China (No. 21476159, 21476160), the Natural Science Foundation of Tianjin (No. 15JCZDJC37400), the 973 program (2014CB932403), and the Program for Introducing Talents of Discipline to Universities of China (No. B06006).

Notes and references

1. B. Chutichai, S. Authayanun, S. Assabumrungrat and A. Arpornwichanop, Energy, 2013, 55, 98–106.
2. K. Faungnawakij, R. Kikuchi, T. Matsui, T. Fukunaga and K. Eguchi, Appl. Catal., A, 2007, 333, 114–121.
3. K. Faungnawakij, Y. Tanaka, N. Shimoda, T. Fukunaga, S. Kawashima, R. Kikuchi and K. Eguchi, Appl. Catal., A, 2006, 304, 40–48.
4. T. A. Semelsberger, K. C. Ott, R. L. Borup and H. L. Greene, Appl. Catal., B, 2006, 65, 291–300.
5. Z. Sun, M. Meng, L. Zhang, Y. Zha, X. Zhou, Z. Jiang, S. Zhang and Y. Huang, Int. J. Hydrogen Energy, 2012, 37, 18860–18869.
6. L. Zhang, M. Meng, S. Zhou, Z. Sun, J. Zhang, Y. Xie and T. Hu, J. Power Sources, 2013, 232, 286–296.
7. N. Shimoda, K. Faungnawakij, R. Kikuchi, T. Fukunaga and K. Eguchi, Appl. Catal., A, 2009, 365, 71–78.
8. X. Zhou, M. Meng, Z. Sun, Q. Li and Z. Jiang, Chem. Eng. J., 2011, 174, 400–407.
9. D. Feng, Y. Wang, D. Wang and J. Wang, Chem. Eng. J., 2009, 146, 477–485.
10. H. Lorenz, M. Friedrich, M. Armbrüster, B. Klötzer and S. Penner, J. Catal., 2013, 297, 151–154.
11. M. Yang, Y. Men, S. Li and G. Chen, Int. J. Hydrogen Energy, 2012, 37, 8360–8369.
12. L. Zhang, M. Meng, X. Wang, S. Zhou, L. Yang, T. Zhang, L. Zheng, J. Zhang and T. Hu, J. Power Sources, 2014, 268, 331–340.
13. T. Minami, MRS Bull., 2000, 25, 38–44.
14. T. Wolkenstein, Adv. Catal., 1960, 12, 189–264.
15. W. Göpel and U. Lampe, Phys. Rev. B: Condens. Matter Mater. Phys., 1980, 22, 6447.
16. K. Maeda, K. Teramura and K. Domen, J. Catal., 2008, 254, 198–204.
17. X. Wang, S. Shen, S. Jin, J. Yang, M. Li, X. Wang, H. Han and C. Li, Phys. Chem. Chem. Phys., 2013, 15, 19380–19386.
18. T. Mathew, Y. Yamada, A. Ueda, H. Shioyama, T. Kobayashi and C. S. Gopinath, Appl. Catal., A, 2006, 300, 58–66.
19. W. Tong, K. Cheung, A. West, K.-M. Yu and S. C. E. Tsang, Phys. Chem. Chem. Phys., 2013, 15, 7240–7248.
20. H. Wei, M. Li, Z. Ye, Z. Yang and Y. Zhang, Mater. Lett., 2011, 65, 427–429.
21. S. E. Collins, L. E. Briand, L. A. Gambaro, M. A. Baltanás and A. L. Bonivardi, J. Phys. Chem. C, 2008, 112, 14988–15000.
22. D. Chadwick and K. Zheng, Catal. Lett., 1993, 20, 231–242.
23. K. A. Pokrovski and A. T. Bell, J. Catal., 2006, 241, 276–286.
24. K. Faungnawakij, T. Fukunaga, R. Kikuchi and K. Eguchi, J. Catal., 2008, 256, 37–44.
25. P.-T. Hsieh, Y.-C. Chen, K.-S. Kao and C.-M. Wang, Appl. Phys. A: Mater. Sci. Process., 2008, 90, 317–321.
26. G. C. Park, S. M. Hwang, J. H. Lim and J. Joo, Nanoscale, 2014, 6, 1840–1847.
27. X. Duan, C. Song, F. Yu, D. Yuan and X. Li, Appl. Surf. Sci., 2011, 257, 4291–4295.
28. A. Vimont, J. Lavalley, A. Sahibed-Dine, C. Otero Areán, M. Rodríguez Delgado and M. Daturi, J. Phys. Chem. B, 2005, 109, 9656–9664.
29. M. M. Branda, S. E. Collins, N. J. Castellani, M. A. Baltanás and A. L. Bonivardi, J. Phys. Chem. B, 2006, 110, 11847–11853.
30. P. H. Matter and U. S. Ozkan, J. Catal., 2005, 234, 463–475.
31. R. W. Stevens Jr, R. V. Siriwardane and J. Logan, Energy Fuels, 2008, 22, 3070–3079.
32. M. Calatayud, S. Collins, M. Baltanas and A. Bonivardi, Phys. Chem. Chem. Phys., 2009, 11, 1397–1405.

---

Footnote

† Electronic supplementary information (ESI) available: TPD-MS-CH₃OH with the adsorption of CH₃OH at 100 °C on ZnO; Arrhenius plots for DME SR reaction over GDZ/γ-Al₂O₃ catalysts; XRD of the CuZnAlO/γ-Al₂O₃ and Zn₉Ga₁O/γ-Al₂O₃ catalysts before and after stability test; TG of the CuZnAlO/γ-Al₂O₃ and Zn₉Ga₁O/γ-Al₂O₃ catalysts before and after stability test; catalyst characterization of NH₃-TPD and TG. See DOI: 10.1039/c6ra07940g

---