The increasing consumption of fossil fuels has led to the energy crisis and various environmental pollutions. Thus, it is necessary to search for clean and sustainable energy sources. As a new clean energy, H₂ is getting more attention from world governments and researchers ^{[1, 2]}. In the past several years, H₂-based fuel cells such as the proton exchange membrane fuel cells have become research hotspots due to their high efficiency in chemical-to-electric energy conversion ^[3]. However, as a renewable gaseous energy source, H₂ is hard to be stored, compressed and transported ^[1], which restricts the wide application of H₂-based fuel cells on vehicles. So, techniques for onboard H₂ production from H₂-rich carriers must be developed. Recently, dimethyl ether (DME) as a potential onboard H₂ source material has attracted increasing attention due to its intrinsic advantages such as a high H/C ratio, high energy density, no toxicity and no carcinogenicity ^[4]. DME can be stored and handled like liquid petroleum gas (LPG) because of their similar physical properties. By using reforming techniques, DME can be efficiently converted to H₂. The reforming techniques include partial oxidation reforming, autothermal reforming, dry reforming and steam reforming (SR). SR can produce the largest amount of H₂ per mole DME due to the additional utilization of the hydrogen available in water. So, DME SR is now regarded as a promising route for onboard H₂ production and supply for H₂-based fuel cells.

The overall reaction of DME SR can be expressed as Eq. (1), which consists of two successive moderately endothermic reactions: DME hydrolysis to methanol (Eq. (2)) catalyzed by a solid acid catalyst and the subsequent steam reforming of methanol to H₂ and CO₂ (Eq. (3)) catalyzed by a metal catalyst.

DME SR:

DME hydrolysis:

MeOH SR:

During DME SR, the formation of CO cannot be avoided due to the presence of reverse water gas shift reaction (r-WGS) shown as Eq. (4).

r-WGS:

It was reported that DME hydrolysis was the rate limiting step for DME SR in many catalytic systems ^{[5, 6, 7]}, so, the acceleration of this step is significant for the increase of the overall reaction rate. DME hydrolysis is usually catalyzed by a solid acid catalyst such as γ-Al₂O₃ and zeolites. Up to now, γ-Al₂O₃ has been studied extensively due to its good activity and stability for DME hydrolysis and it has few side reactions. However, γ-Al₂O₃ needs a high operating temperature (300-400 °C) for DME hydrolysis, leading to the sintering of Cu species and deactivation of Cu-based catalysts ^[7]. Compared to γ-Al₂O₃, zeolites have much stronger acidity and can catalyze DME hydrolysis at a lower temperature (< 300 °C) ^{[8, 9]}. Nevertheless, the strong acid sites on zeolites also result in the formation of coke ^{[10, 11]}, thus decreasing the performance of the catalyst. So, the modification of the acidity of the zeolite is necessary. Based on this analysis, the present research aims to tune the acidity of zeolite HZSM-5 to remove the unnecessary strong acid sites.

The H-type of the zeolite HZSM-5 has been widely employed in many catalytic reactions due to its high activity and low cost ^{[12, 13, 14, 15]}. In order to further improve the performance of HZSM-5, many elements or compounds have been employed to modify the structure and acidity of HZSM-5, such as La ^{[10, 16]}, Mg ^[17], Ga ^[18], B ^[15] or P ^{[14, 19, 20, 21]} and an alkaline treatment ^[22]. After P modification of H-ZSM-5, the coke deposition in many reactions was dramatically decreased ^{[14, 23]}. So, in this work, phosphorus was selected and added to HZSM-5 for the optimization of its acidity. It is found that the CuO-ZnO-Al₂O₃ (CuZnAlO, CuO/ZnO/Al₂O₃ mass ratio was 3/1/6) catalysts mixed with the modified P-HZSM-5 exhibited higher activity and also better stability and higher CO₂ selectivity in DME SR compared to those consisting of CuZnAlO and pure HZSM-5. By using ³¹P magic angle spinning nuclear magnetic resonance (³¹P MAS NMR), NH₃ temperature- programmed desorption (NH₃-TPD) and Fourier transform infrared spectroscopy (FT-IR), the catalysts were well characterized, and the relationship between the catalyst structure and catalytic performance was clearly revealed.

HZSM-5 with a SiO₂/Al₂O₃ molar ratio of 25 was purchased from the Catalyst Co. Ltd of Nankai University and calcined at 550 °C for 3 h to remove impurities. The P modified samples (P/HZSM-5 mass ratios: 1%, 3%, 5% or 7%) were prepared by impregnating the HZSM-5 zeolite in an aqueous solution of diammonium hydrogen phosphate with the desired concentration. After evaporating at 50 °C and drying at 120 °C for 12 h, the dry precursor was obtained, which was further calcined in air at 500 °C for 4 h. The final solid acid catalysts were denoted as xP/HZSM-5, where x represents the mass ratio of P to HZSM-5. Before use in the DME SR reaction, the solid acid catalyst was mechanically mixed with a CuZnAlO reforming catalyst (mass ratio: 1/1) and made into particles of 40-60 mesh. The details for the preparation of the CuZnAlO catalyst (CuO/ZnO/Al₂O₃ mass ratios are 3/1/6) by the co-precipitation method can be found elsewhere.

The specific surface area and pore structure of the catalysts were determined by N₂ adsorption at -196 °C on a volumetric adsorption apparatus (Quantachrome QuadraSorb SI instrument). The specific surface area of the catalyst was calculated according to BET equation. The average pore diameter was measured by the BJH method.

The X-ray diffraction (XRD) patterns were recorded on a D8 diffractometer (Bruker Company) using a cobalt anode X-ray tube (Cu K_α radiation, λ = 0.15418 nm) operated at 40 kV and 40 mA. Diffraction data in the range of 2θ from 10° to 50° were collected with a step size of 0.02°.

Before the NH₃-TPD test, the catalyst (100 mg each time) was first pretreated at 500 °C for 1 h in He (99.999%, 80 mL/min). The sample was cooled to 100 °C in the same atmosphere. Then, the atmosphere was switched to 5% NH₃ in N₂ for 2 h NH₃ adsorption at 100 °C. After flushing with pure He, the NH₃-TPD test was carried out in the range of 100−600 °C at a heating rate of 10 °C /min. The signal was monitored by a gas chromatograph equipped with a TCD detector (XianQuan TP-5079 instrument).

Temperature-programmed oxidation (TPO) was performed on the spent catalysts using the same XianQuan TP-5079 instrument. Each time, 100 mg spent catalyst was heated from room temperature to 700 °C at a heating rate of 10 °C /min in the atmosphere of 6% O₂/He (20 mL/min).

The FT-IR spectra of the catalysts after pyridine adsorption were recorded on a Nicolet Nexus IR spectrometer to characterize the surface acid sites (Brönsted or Lewis). Fresh catalyst (30 mg) was ground and pressed into a very thin wafer, which was placed in a quartz IR cell with CaF₂ windows. The sample was evacuated at 300 °C for 1 h to remove impurities adsorbed on the catalyst. The catalyst was exposed to pyridine under a N₂ stream at room temperature until the adsorbed amount of pyridine reached saturation. Finally the IR cell was evacuated at 150 and 300 °C for 30 min for the collection of IR spectra at these two temperatures.

The spectra of ³¹P MAS NMR were recorded on a Varian InfinityPlus 300 NMR spectrometer at 121.38 MHz with a spinning rate of 8.0 kHz. The pulse width and pulse delay were 1.0 μs and 35.0 s, respectively.

The activity of the catalysts (500 mg) for DME SR was measured in a fixed-bed continuous flow reactor with a constant space velocity of 12000 mL/(h·g_cat) at atmospheric pressure. Prior to each test, the mixed catalyst was reduced at 300 °C for 1 h in 10% H₂/N₂ mixture gas. After the temperature dropped to 200 °C, the stream was switched to the feed gas consisting of 10% DME, 40% N₂ and 50% H₂O (g) (S/C = 2.5), and kept at this same temperature for 0.5 h, and the DME SR activity was evaluated at different temperatures with an interval of 25 °C. The effects of external and internal mass transfer limitation on the catalytic reaction were confirmed as excluded by varying the linear flow rate over the catalyst and the catalyst particle size, respectively. The products were analyzed using an online gas chromatograph (Agilent 7890A) equipped with PoraPak N, molecular sieve 5A and PoraPak Q capillary columns.

DME conversion and selectivity to C1 species were defined as follows:

DME conversion=(F_DME,in-F_DME,out)/F_DME,inx100%

Selectivity to C_i species=F_Ci/∑F_Cix100%

Hydrogen yield was defined as the ratio of the molar amount of DME converted to hydrogen to the total molar amount of DME fed to the reactor, and was calculated using

H₂yield=F_H2/6F_DME,inx100%

The catalytic stability of HZSM-5 or 5%P-modified HZSM-5 mixed with CuO/ZnO/Al₂O₃ for DME SR was investigated by a continuous isothermal activity measurement at 350 °C for 8 h.

The specific surface area, pore volume and pore diameter of P-modified HZSM-5 are summarized in Table 1. With the increase of P content, both the BET surface area and total pore volume of the modified HZSM-5 samples gradually decreased. However, there was no significant difference in the pore diameter among all the catalysts, which suggested no pore blocking by the added P.

The XRD patterns of the P-modified samples are shown in Fig. 1. The P-modified catalysts displayed similar diffraction peaks at the same positions as pure HZSM-5, suggesting the maintaining of the framework structure of HZSM-5 zeolite. However, the intensity of the diffraction peaks of the P-modified catalysts gradually decreased with the increase of P content, which may be due to the decreased crystallite size of the HZSM-5 zeolite. In addition, after P modification, no diffraction peak of P species was observed even for the sample containing 7%P, which reflects the high dispersion of the P species.

Fig. 1. XRD patterns of the parent HZSM-5 and the P-modified HZSM-5 catalysts. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

The NH₃-TPD profiles of the parent HZSM-5 and P-modified HZSM-5 are presented in Fig. 2. There were two desorption peaks for the parent HZSM-5. The peak at high temperature corresponded to strong acid sites and can be attributed to the desorption of NH₃ from bridging hydroxyl groups (Si-OH-Al) on the zeolite. The other peak at low temperature corresponded to the desorption of NH₃ from silanol groups or extra framework aluminum species ^[24]. After modification by P, the low temperature desorption peaks showed small changes in position and intensity. The high temperature desorption peaks were much decreased. Moreover, as the P content increased to 5%, the high temperature desorption peak almost disappeared. In order to give the acid amounts on these catalysts, we calculated the amounts of desorbed NH₃ from the desorption peaks areas, as listed in Table 2. It is clear that the P modification mainly reduced the amounts of strong acid sites, which is favorable for catalyst stability in the DME SR reaction because strong acid sites catalyze the formation of coke. By careful observation, it was also found that with the increase of P content, the desorption peak of weak acid sites slightly shifted to lower temperature, implying the subtle weakening of the acid strength.

Fig. 2. NH3-TPD profiles of the different catalysts. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

Based on these results, a schematic model for P-modified HZSM-5 was proposed and shown in Fig. 3. As the content of P increased, the strong acid sites decreased sharply and the weak acid sites increased slowly. When the amount of P reached 5%, the strong acid sites had almost disappeared. Further increase of P content to 7% resulted in the decrease of the weak acid sites as reflected by the decreased intensity of the desorption peak below 300 °C.

Fig. 3. Schematic model of P-modified HZSM-5.

To investigate the correlation between P modification and coke formation during DME SR, O₂-TPD tests were performed on the spent catalysts. The results are presented in Fig. 4. With the increase of P content, the amount of coke on the catalyst surface sharply decreased, suggesting the obvious decrease of strong acid sites as confirmed by NH₃-TPD. It is known that the coke formed can cover both the metal sites and the acid sites, leading to the decrease of the catalytic activities for DME hydrolysis and the following methanol reforming. The P induced decrease or disappearance of strong acid sites is helpful for catalytic activity and catalyst stability.

Fig. 4. O2-TPO profiles of spent catalysts. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

Because NH₃-TPD cannot distinguish Brönsted acid sites and Lewis acid sites, the FT-IR spectra of the catalysts after pyridine adsorption were collected. Fig. 5(a) shows the FT-IR spectra of pyridine adsorbed on the parent HZSM-5 and P-modified HZSM-5 at 150 °C. Both the bands at 1450 cm^-1 corresponding to Lewis acid sites and that at 1540 cm^-1 corresponding to Brönsted acid sites ^{[25, 26]} were clearly identified. By calculating the areas of these two peaks, the relative amounts of Lewis acid sites and Brönsted acid sites as well as the ratios of Lewis acid sites to Brönsted acid sites (L/B) were obtained, as listed in Table 3. As the P content increased, both the Lewis acid sites and Brönsted acid sites decreased. Meanwhile, the L/B ratios also decreased sharply, which means the larger decreasing extent of Lewis acid sites, as depicted in Fig. 6(a). The FT-IR spectra of pyridine adsorption at the higher temperature of 300 °C were also recorded. The results are shown in Fig. 5(b), Table 3 and Fig. 6(b). By a comparison of the data in Table 3, it is found that as the content of P increased, both the Brönsted acid sites and Lewis acid sites decreased quickly. The remaining band at 1540 cm^-1 in Fig. 5(b) corresponded to the strong Brönsted acid sites because adsorbed pyridine on weak Brönsted acid sites can be removed by vacuum evacuation at 300 °C. For the samples containing 1% to 5% P, the quantities of strong Brönsted acid sites measured at 300 °C decreased rapidly with the increase of P content, while the quantity of total Brönsted acid sites measured at 150 °C changed little. Further increase of P content to 7% decreased both the strong acid sites and weak acid sites. Based on these results, it is suggested that the strong Lewis acid sites can be readily covered by P, while the strong Brönsted acid sites can be modulated by P and were transformed to weak acid sites, as demonstrated by the increasing difference of Brönsted acid sites listed in Table 3.

Fig. 5. FT-IR spectra after pyridine adsorption on the parent and P-modified HZSM-5 recorded at different temperatures. (a) 150 °C; (b) 300 °C. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

Fig. 6. Acid sites of P-modified HZSM-5 catalysts measured by pyridine IR at different temperatures. (a) 150 °C; (b) 300 °C. (1) L acid sites; (2) B acid sites; (3) Total acid sites.

In order to reveal the potential change of the acidity of these catalysts during reactions, the FT-IR spectra of pyridine adsorbed on the spent 5%P-HZSM-5 catalysts at different temperatures were also recorded and compared with those for the fresh catalysts, as shown in Fig. 7. It was found that the quantities of the acid sites on this catalyst decreased after use in the DME hydrolysis reaction, especially the strong acid sites, which was reflected by the obvious difference in the spectra measured at 300 °C. This accounts well for the decrease of catalytic activity.

Fig. 7. FT-IR spectra of pyridine adsorption over fresh or spent 5%P/HZSM-5 at 150 or 300 °C. (1) Fresh, 150 °C; (2) Spent, 150 °C; (3) Fresh, 300 °C; (4) Spent, 300 °C.

It is well known that a progressive dehydration of acidic solid catalysts can occur during heating, inducing the consequent transformation of Brönsted acid sites to Lewis acid sites, as described by the following reaction ^[25]:

However, in the presence of H₂O, the reverse chemical reaction can also occur. So, to reveal the influence of steam (one of the reactants) on the acid sites of the catalysts, the FT-IR spectra of pyridine adsorption on 5% P-modified HZSM-5 previously treated by steam at 350 °C for 0.5 h were collected, as displayed in Fig. 8. By the comparison with the spectra of the untreated samples, it was found that steam treatment increased the Brönsted acid sites but decreased the Lewis acid sites for both the parent HZSM-5 and P-modified HZSM-5. So it was deduced that during the DME SR reaction, most of the Lewis acid sites were transformed to Brönsted acid sites due to the presence of the large amount of steam in the feed and the long reaction time. In other words, the Brönsted acid sites play the major role for DME hydrolysis during the DME SR reaction.

Fig. 8. FT-IR spectra of pyridine adsorption at 150 °C over fresh HZSM-5 (1), HZSM-5 treated by steam for 0.5 h (2), fresh 5% P-modified HZSM-5 (3), and 5% P-modified HZSM-5 treated by steam for 0.5 h (4).

Fig. 9 shows the ³¹P MAS NMR spectra of the P-modified HZSM-5 catalysts. The resonance peak at the chemical shift position of d = 0 corresponds to the free monomeric [PO₄] orthophosphate species group ^[19]. From the signal at d = 0, it can be inferred that as 1% or 3% P was added to HZSM-5, few free monomeric [PO₄] orthophosphate species were formed, while the addition of larger amounts of P resulted in a remarkable increase of such species. The other two peaks at d = −6 and −12 can be attributed to the terminal P atoms in pyrophosphoric acid species, and the middle P atoms in polyphosphates or some polymerized short-chain polyphosphates, respectively, according to the assignments in the literature ^{[20, 21, 27]}. The characteristic resonance peaks at d = -20 and -24 corresponding to the bidentate middle chain groups of polyphosphates with an Al-O-P structure ^[27] were also observed. The resonance peak at d = -40, which increased with the increase of P content in HZSM-5, can be assigned to the groups in highly condensed branching polyphosphates like P₄O₁₀, and it was also correlated well with aluminum ^[27]. The above pyridine FT-IR results indicated that the modification of HZSM-5 by the increased amounts of P gradually modulated the strong Brönsted acid sites to form weak acid sites. So, by combining the results of ³¹P MAS NMR and pyridine FT-IR, a mechanism for the modification of HZSM-5 by phosphorus can be derived, which is similar to that reported in the literature ^[27], as shown in Scheme 1.

Scheme 1. A mechanism for the modification of HZSM-5 by phosphorus.

Fig. 9. 31P MAS NMR spectra of the P-modified HZSM-5 catalysts. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

The catalytic activity of the catalysts including DME conversion and H₂ yield are shown in Fig. 10(a) and (b). Both DME conversion and H₂ yield increased with the increase of reaction temperature. The modification with P exhibited different influence on the catalytic performance in different temperature regions. Below 290 °C, the DME conversion and H₂ yield decreased with the increase of P content. Above 300 °C, P modification showed a positive influence on the DME SR performance of the catalysts with the amounts of P between 1% and 5%. Further increase of P content to 7% led to decreased DME conversion and H₂ yield. At the lower temperature below 290 °C, the DME SR performance was mainly determined by the DME hydrolysis activity of the solid acid catalyst. Since the parent HZSM-5 has a larger amount of acid sites than the catalysts modified by P, it naturally exhibited better performance for DME hydrolysis. As a result, better DME SR performance including higher DME conversion and higher H₂ yield were achieved over this sample. However, at relatively higher temperatures (> 300 °C), DME hydrolysis is not the determining step for DME SR. In this situation, the catalytic performance of the catalysts was mainly determined by methanol reforming over the Cu-based metallic catalyst. The NH₃-TPD and pyridine FT-IR results indicated that the densities of strong acid sites decreased with the increase of P doped in HZSM-5, which is favorable for the inhibition of coke formation. Nevertheless, an excessive amount of P in HZSM such as 7% would result in too weak acidity and too low activity of the catalyst for DME hydrolysis, thus decreasing the whole DME SR performance. So, the optimized content of P in HZSM-5 is 5%.

Fig. 10. DME SR performance of the P-modified HZSM-5 catalysts mixed with the CuZnAlO catalyst. (a) DME conversion; (b) H2 yield; (c) CO selectivity; (d) CO2 selectivity. (1) HZSM-5; (2) 1%P/HZSM-5; (3) 3%P/HZSM-5; (4) 5%P/HZSM-5; (5) 7%P/HZSM-5.

Fig. 10(c) and (d) illustrates that as the reaction temperature increased, the CO₂ selectivity decreased and the CO selectivity increased, which were caused by the side reactions including the r-WGS reaction, DME decomposition, MeOH decomposition and DME to syngas reforming at high temperatures. From this figure, it was also found that at 275 °C or higher, the sample containing 5% P exhibited the lowest CO selectivity.

As mentioned above, acid sites facilitate DME hydrolysis, increasing the DME SR reaction rate. However, the acidity also enhanced the formation of coke, decreasing the catalytic performance of the catalysts. To evaluate the catalytic stability of the P-modified HZSM-5 catalysts, the parent HZSM-5 and 5%P-modified HZSM-5 were selected for an 8 h durability test at 350 °C. The results are displayed in Fig. 11. DME conversion over HZSM-5 decreased quickly in the first 4 h, which resulted from coke formation or the transformation of strong acid sites to weak acid sites in the presence of steam. After modification by P, the decline of DME conversion over 5%P-modified HZSM-5 was effectively inhibited. In the whole process, 5% P-modified HZSM-5 displayed higher catalytic stability.

Fig. 11. Time-on-stream stability tests on HZSM-5(1) and 5% P-modified HZSM-5(2) catalysts.

Compared to HZSM-5, a 5% P-modified HZSM-5 catalyst mixed with CuZnAlO exhibited higher CO₂ selectivity and better catalytic stability for DME SR due to the presence of the appropriate amounts of P and strength of acidity. The framework structure of H-ZSM-5 was maintained after P modification. As the P content increased, the strong acid sites decreased sharply. 5 wt% P in HZSM-5 can completely cover the strong acid sites, transforming them to weak acid sites. On the basis of ³¹P MAS NMR and pyridine FT-IR data, a mechanism of P modification on the acidity of HZSM-5 was derived. HZSM-5 modified by 5% P exhibited better stability due to less coke deposition.