Hydrogen production via silica membrane reactor during the methanol steam reforming process: experimental study

Abstract

The main purpose of this work is to present an experimental operating analysis of a silica membrane reactor (MR) for hydrogen production during the methanol steam reforming (MSR) reaction. To implement this performance analysis, a microporous silica membrane is synthesized by a polymeric sol–gel method. To achieve a high quality silica membrane, a new strategy is used for surface modification of the ceramic support, in which particle size control of the boehmite sol is applied. After evaluation of modified alumina supports, the synthesized silica membrane is characterized and its performance is investigated. The performance analysis of the silica membrane in hydrogen purification shows that the H₂/N₂, H₂/CO₂ and H₂/Ar permselectivity increase sufficiently to 26.18, 22.13 and 29.42, at 200 °C, respectively, so that hydrogen permeance is measured around 1.1 × 10⁻⁶ mol m⁻² s Pa at the corresponding conditions. These promising results indicate a high quality of silica membrane for hydrogen production in comparison with the literature data. To achieve the initial purpose of this study, the synthesized silica membrane performance is investigated in the MR set up during the MSR reaction for hydrogen production. In this case, silica MR performance is compared with a traditional reactor (TR) in terms of the methanol conversion, hydrogen yield, hydrogen recovery and CO selectivity. The effects of several operating parameters are also investigated on silica MR performance. In general, the silica membrane presents a higher performance with respect to the TR in all ranges of operating parameters. As a specific consequence, a 6% performance enhancement can be obtained by silica MR in compared to the TR system.

---

1. Introduction

In recent years, there has been a growing interest in developing technologies considering the benefits of clean energy sources. The reduction of atmospheric pollution, namely, the emission of greenhouse gases has become imperative and, among the new technologies for mitigating these emissions, fuel cells have the ability to convert chemical into electrical energy efficiently. In particular, proton exchange membrane fuel cells (PEMFCs) are zero-pollutant emission systems since they transform the chemical energy of the electrochemical reaction between hydrogen and oxygen into clean electrical power.^1,2

On the other hand, it should be noted that the high purity hydrogen applications are not restricted to PEMFCs. Indeed, large quantities of hydrogen are required in the petroleum and chemical industries, so that the most applications of hydrogen are for the processing (“upgrading”) of fossil fuels and for the production of ammonia. The major consumers of hydrogen in the petrochemical plant include hydrodealkylation, hydrodesulfurization, and hydrocracking processes.

Hydrogen is produced industrially as a hydrogen-rich stream mainly via steam reforming of natural gas in traditional reactors (TRs).³ In sequence, hydrogen is purified to reach the desired purity for various applications. Indeed, the stream coming out from TRs commonly contains hydrogen, CO₂, CO, CH₄ and other byproducts. As a consequence, hydrogen consumers such as PEMFC impose a purification processes on hydrogen.^3,4 On the other hand, the conventional stages of hydrogen purification influence negatively the overall process in terms of costs and efficiency.³ Hence, scientific attention has increased in the development of alternative technologies to produce high purity hydrogen. Among them, membrane reactor (MR) technology plays an important role as an alternative solution to conventional systems (TRs + further stages of hydrogen purification) in terms of combination in a single stage of the reforming reaction for hydrogen production and its purification without demanding any further process/treatment.⁵ As shown in Fig. 1, the interest towards this technology is testified by the growing number of scientific publications in the specialized literature.

Fig. 1 Number of scientific papers on H₂ production by MR technology versus year. Scopus database: http://www.scopus.com.

As a particular aspect regarding membrane technology, the utilization of inorganic MRs makes possible several advantages over TRs.^3,6–8 Concerning the type of membrane to be housed in an MR, both MR cost and performance need to be considered. Specifically, several studies have focused on the application Pd-based MRs.^9–13 The Pd-based membranes are highly selective for hydrogen permeation and allow a high purity hydrogen stream to be obtained. However, these membranes suffer from cracking during thermal cycling and readily evidence surface contamination by sulfur or carbon monoxide species.¹⁴ Moreover, Pd-based membranes are very expensive and their applications are limited owing to low hydrogen permeability.^15,16 Therefore, an alternative and cheaper solution is strongly needed. On this way, microporous silica membranes are cheaper and present higher permeance, but, on the contrary, they show lower selectivity to hydrogen permeation compared to Pd-based membranes.^17–19

On the other hand, in terms of processing temperature and CO content, with respect to other feed sources, methanol utilization shows various advantages as a hydrogen carrier for fuel cell applications and, namely, it can be produced from renewable sources.⁶ In the meanwhile, the methanol reforming reaction occurs at relatively low temperatures of 240–300 °C, compared to the methane reforming reaction which is normally performed at 800–1000 °C. Therefore, the MSR reaction has been seen as a very attractive and promising process for hydrogen production according to the scientific literature on the argument.

Nevertheless, using silica MRs for carrying out MSR reaction has not been extensively studied. Indeed, to the best of our knowledge, few studies have been presented for silica MR performance in the MSR reaction.^20–26 According to our previous modeling works,^23–26 promising results have justified a necessity for a comprehensive experimental study of the potential advantages achievable using a silica MR.

Therefore, the present work aims to evaluate and compare the performance attainable in silica MR by carrying out the MSR reaction. Moreover, a comparison with a TR, working at the same operating conditions of MR, was realized. To perform this study, as a first approach, a silica membrane is synthesized with high hydrogen permeance on the modified γ-alumina support, in which the support surface is modified using a new strategy of size distribution control of sol particles. After analyzing the synthesized silica membrane efficiency for hydrogen purification, the silica membrane performance in the MR set up is investigated for the MSR process. A set of experimental results are then provided which illustrate some key points about the silica MR performance in terms of the methanol conversion, hydrogen recovery, total hydrogen yield and CO selectivity versus some important operating parameters such as reaction temperature, feed molar ratio and gas hourly space velocity (GHSV).

2. Experimental

2.1. Materials

In this experimental study, the material sources used were as follows: tetraethylorthosilicate (TEOS, 98%, Sigma Aldrich) as silicon source; nitric acid (HNO₃, 65%, Merck) as catalyst for silica sol preparation; ethanol (EtOH, 99.9% Merck) as solvent; methanol (MeOH, 99.9% Merck) and deionized water as feed for steam reforming; and polyethylene glycol (PEG, Merck, molecular weight: 35 000) as stabilizer for boehmite sol preparation; and also, aluminum-tri-sec-butylate (ATSB, 97%, Merck) as source of γ-alumina.

2.2. Surface modification of ceramic support

2.2.1. Support preparation. According to the authors’ previous works,²⁷ the home-made supports applied for membrane synthesis were α-alumina tubes with a thickness of 4 mm, diameter of 11 mm, length of 70 mm, average pore size of ∼570 nm and average porosity of 47.2%. Before the γ-alumina synthesis, the supports were cleaned in deionized water by an ultrasonic regenerator for 10 min and then dried at 40 °C for 12 h.

2.2.2. Support modification. In order to modify the pore structure of porous α-alumina substrates, a boehmite sol as the material for the intermediate layers were prepared. The boehmite sol was prepared by the method of Uhlhorn et al.,²⁸ although an innovation was applied for sol preparation in this study. In this procedure, the boehmite sol preparation was carried out by adding ATSB dropwise to deionized water, in which about 1.5 L of water was added per mole alkoxide at 80 °C and under vigorous stirring. Then, a white solution was obtained, which was peptized with nitric acid. The resulting colloidal suspension was kept boiling until the most of the butanol was evaporated. The PEG solution was made by dissolving PEG (1 wt% of sol) in deionized water under vigorous stirring and then was added to the sol. It should be noted that the nitric acid was added to decrease the pH to the range 3–4 and after spending 1 h time in this step, the sol was refluxed for 16 h to form a stable boehmite sol.

The dip coating process was performed at room temperature, in which the substrate speed and dip-time were 1 mm s⁻¹ and 10 s, respectively. After the dipping step, the membranes were dried at 40 °C for at least 24 h. Subsequently, the γ-alumina layer was formed by calcining at 650 °C for 3 h in atmospheric condition with a heating and cooling rate of 0.5 °C min⁻¹. The whole processes of dipping, drying and calcining was repeated 4 times. As first approach, the particle size distribution of the boehmite sol was altered for each step of coating. Varying the particle size distribution of the sol was provided by changing the nitric acid and aging time. In fact, the aging time means the period time before adding the nitric acid to the boehmite sol. The conditions of boehmite sol preparation for different intermediate modified layers are presented in Table 1.

Table 1 The conditions of boehmite sol preparation and its particle size distribution for different layers

The intermediate layer or coating sol	Molar ratio of nitric acid/alkoxide	Aging time/min	Particle size distribution/nm
Boehmite sol of layer 1	0.026	1440	500–600
Boehmite sol of layer 2	0.026	5	200–300
Boehmite sol of layer 3	0.053	5	100–200
Boehmite sol of layer 4	0.070	5	40–100

---

2.3. Silica membrane preparation

The microporous silica membrane top-layer was prepared by dipping the modified γ-alumina mesoporous substrate in the silica standard solution, and followed by calcining. In this method, a mixture of acid and water is carefully added to a mixture of TEOS and ethanol under vigorous stirring, while during the addition of the acid/water mixture, the TEOS/ethanol mixture is placed in an ice-bath to avoid partial hydrolysis.²⁹ Then, the reaction mixture was refluxed for 3 h at 60 °C in the oil bath under uniform stirring. The reaction mixture had a final molar TEOS/ethanol/water/acid ratio of 1/3.8/6.4/0.085. The reacted mixture was cooled and diluted 19 times with ethanol to obtain the final dip solution. The modified γ-alumina substrate was coated by prepared silica solution, so that the coating speed was 1 mm s⁻¹ and the dip-time was 10 s. After dipping step, the membrane was calcined at 600 °C for 3 h at atmospheric conditions with a heating and cooling rate of 1 °C min⁻¹. The whole processes of dipping and calcining was repeated 5 times to eliminate defects in the silica membrane layer structure.

2.4. Characterization and permeance tests

The meso/microstructure and morphology of the synthesized membrane were studied by scanning electroscope microscope (SEM) and energy dispersive X-ray spectrometry (EDAX, Phoenix) measurements.

The permeation tests were carried out using a custom-made stainless steel module, designed for 70 mm tubular membranes. A schematic of the permeation set up is shown in Fig. S1,† in which gas permeation was measured based on the constant pressure method.

The membrane ends were sealed in the module using Witon and Teflon bronze O-rings, which allow measurements at high temperatures (up to 300 °C). For a modified γ-alumina mesoporous membrane, only pure gas permeance (H₂ and N₂) tests were performed at ambient temperature, while for silica membranes, pure-gas (H₂, N₂,CO₂ and Ar) tests were carried out at different temperatures (25, 100, 200 and 300 °C). Moreover, all the permeance tests were investigated at different gradient pressures (1, 2, 3 and 4 bar) and the permselectivity (F_α) was obtained by the ratios of single gases permeances.

2.5. Silica MR tests

In this case, the silica MR performance was compared with TR. Fig. 2 shows a schematic of the experimental set up used for the MSR reaction.

Fig. 2 Schematic of the permeation test set-up.

The experimental setup of the MR consists of a tubular stainless steel module (length 260 mm, I.D. 24 mm) housing and a tubular microporous silica membrane that is selective to hydrogen (O.D. 11 mm, total length 70 mm and active length 50 mm) which was produced during this study. As shown in Fig. 3, two zones can be identified in the MR: a first zone, inside the membrane (lumen side), where permeate flow (mostly hydrogen) is collected; and a second one corresponding to the annulus of the MR (shell side), where the reaction takes place. Regarding to flow hydrodynamics in TR, the silica membrane was replaced with a metal profile tube.

Fig. 3 Schematic flow-sheet of the silica MR set-up.

A schematic of the silica membrane module is illustrated in Fig. 3. For the silica MR and TR in the MSR reaction, 1 g of commercial Cu/ZnO/Al₂O₃ catalyst (ICI 83-3, furnished by Synetix), was filled into the shell side and glass spheres (1 mm diameter) were mixed and also placed into both end sides of the module to prevent the motion of the catalyst due to gas flow. Two Teflon-bronze O-rings ensure that permeate and retentate streams don’t mix with each other in the module at high temperatures.

Before the reaction, the catalyst was pretreated for 3 h with hydrogen and nitrogen flow (1.1 × 10⁻² mol min⁻¹), at atmospheric pressure and a temperature of 320 °C to reduce copper oxide to metal Cu. A flat temperature profile along the furnace was confirmed during the reaction by means of a three-point thermocouple placed in the furnace.

During the MSR process, the H₂O/MeOH mixture was evaporated in a preheating line, and diluted by argon carrier gas, in which the argon carrier gas was used with a flow rate of 25 ml min⁻¹. The concentration of products and reactants in the retentate and permeate side was analyzed using gas chromatography equipped with a packed Porapak Q column. Each experimental point obtained in this work represents an average value of 7 experimental points taken in 180 min at steady-state conditions.

The following definitions were used for describing the silica MR/TR performances:

where CH₃OH_in is the methanol molar feed flow rate and CH₃OH_out is the methanol flow rate in the MR outlet.where H_2-permeate is the hydrogen molar flow rate that permeates through the silica membrane, H_2-retentate is the hydrogen molar flow rate in the retentate side and X can be H₂, CO₂, CO. Among the mentioned definitions, eqn (3) was only related to the MR.

3. Result and discussion

3.1. Modified membrane support

In the preparation of nanostructure silica membranes, the quality of the support has a great effect on the membrane layer integrity. The surface roughness and homogeneity of the support determine not only the integrity of the selective silica layer but also the minimal thickness of the silica layer for complete surface coverage.^30,31 The use of thin intermediate layers is an attractive alternative which can be used to generate a smooth surface, to improve the chemical adhesion of the silica layer to the support, to limit the effect of differential thermal expansion coefficients, and finally, to limit the diffusion of the silica sol in the support pores.

According to literature,³² the γ-alumina layer is mostly used as an intermediate membrane layer for the development of a gas separation membrane. These layers are not susceptible to crack-formation and peeling-off effects during the firing process. Hence, to modify the homemade α-alumina tubular supports, the γ-alumina layer was prepared as intermediate layer in synthesizing the nanostructure silica membrane. However, in order to improve the surface modification of the ceramic support, a new technique for the synthesis of intermediate γ-alumina layers was suggested, so that size distribution control of sol particles in each step of coating was applied (Fig. 4).

Fig. 4 SEM images of the modified γ-alumina support prepared by control of particle size: (a) surface and (b) cross-section.

Fig. 5 shows the SEM micrographs of the surface and the cross-section of the γ-alumina/α-alumina support. The pore size and surface roughness of the α-alumina support were clearly reduced, in which the pore sizes of such γ-alumina layers are recognized to be in the 3–5 nm range. Moreover, after coating 4 times, as shown in Fig. 5, the thickness of the γ-alumina layer is around 10 μm.

Fig. 5 SEM images of (a) cross section and (b) surface of silica membrane.

Furthermore, the gas permeation tests were conducted to observe the effect of modification using the γ-alumina layers. As presented in Fig. S2,† the slope of the N₂ and H₂ permeances were decreased after modifying the support with respect to the unmodified support. On the other hand, the pressure dependence of gas permeance was decreased. This result indicated that the gas permeance mechanism was approximately changed from viscous flow to Knudsen diffusion.

As reported in Table S1,† the permselectivity of H₂/N₂ (higher than 3.2, and relatively near to the Knudsen diffusion mechanism (3.74)) absolutely validates the formation of a mesoporous layer on the α-alumina support. The H₂/N₂ permselectivity is decreased by increasing the pressure difference and this trend is probably related to the greater effects of macropores or micro-defects on permeance at high pressure differences.

Corresponding to the obtained results, a suitable uniformity and roughness of the γ-alumina membrane was achieved by using control of the particle size distribution in the boehmite sol preparation in comparison with the literature.³¹

3.2. Silica membrane

According to the high stability and quality of the modified γ-alumina support, the silica membrane was synthesized on the modified support via the sol–gel method. The SEM observation results of the silica membrane surface and cross section are shown in Fig. 5.

As depicted, the α-alumina support, γ-alumina layer and silica membrane layer are approximately recognized and the homogenous crack-free silica layer was formed.

However, the SEM image does not clearly distinguish the top layer from the γ-alumina intermediate layer or α-alumina substrate. Therefore, an EDAX analysis was also carried out to investigate the quality of the synthesized silica membrane. An EDAX analysis is depicted in Fig. S3;† the formation of a silica layer on modified γ-alumina is confirmed, so diffusion of the silica particles into the support is negligible.

The activated transport or molecular sieving mechanism as reported in the literature³¹ has a temperature dependency flux as follows:

J₀ is a temperature independent coefficient and E_a is the apparent activation energy. E_a is the sum of two contributions: the heat of sorption of the molecule, which is a negative number because adsorption is an exothermic process; and the positive activation energy of mobility of the permeating molecule inside the membrane matrix. Since these two terms have opposite signs, the apparent activation energy can be positive or negative depending on their relative magnitudes.³³

Fig. 6 shows the observed pure gas (H₂, CO₂ and Ar) permeance versus increasing pressure difference for the synthesized silica membrane at different temperatures of 25, 100 and 200 °C. By referring to the obtained results, increasing the gas permeance by pressure difference due to the enhancement of driving force is reasonable.

Fig. 6 Pure gases permeance versus pressure difference at various temperatures: (a) H₂, (b) CO₂ and (c) Ar.

According to the dual trend of gas permeance versus temperature, as depicted in Fig. 6 and S4,† the permeance of H₂ through the silica membrane is activated and increases with temperature. Therefore, regarding Fig. S4† and eqn (5), the activation energy of H₂ permeation in synthesized silica membrane is calculated to be positive value around 10.1 kJ mol⁻¹, and consequently the molecules have a larger adsorption on the silica surface at higher temperatures, while the permeations of CO₂ and Ar decrease with temperature. Hence, the activation energies of CO₂ and Ar have negative values equal to −3.1 and −1.9 kJ mol⁻¹, respectively.

Moreover, as reported in Table 2, the permselectivity of H₂/CO₂, H₂/N₂ and H₂/Ar increased remarkably by enhancement of the temperature due to higher H₂ permeance and lower CO₂, N₂ and Ar permeances at higher temperatures. This result is another confirmation of the activated transport mechanism in the synthesized silica membrane.

Table 2 H₂/CO₂, H₂/N₂ and H₂/Ar permselectivity for synthesized silica membrane at different temperatures (Δp = 2 bar)

Temperature	25 °C	100 °C	200 °C
H2/CO2	2.35	7.68	22.13
H2/N2	3.76	10.41	26.18
H2/Ar	4.21	11.76	29.42

---

A direct comparison among all the experimental data from the literature reported in Table 3 is not possible owing to the different operating conditions adopted by each author. Nevertheless, from a qualitative point of view it is possible to observe that most of the H₂/N₂ and H₂/CO₂ permselectivity values for silica membrane from the literature are concentrated between 100 and 500 °C. Hence, it is found that the permselectivity values are increased by the enhancement of the temperature and pressure difference.

Table 3 Comparing performance of synthesized silica membrane in this work with literatures data

Membrane	H2/CO2	H2/N2	H2/Ar	Permeance H2/mol m^−2 s^−1 Pa^−1	T/°C	Pressure gradient/bar	Ref.
Silica	3.9	—	—	11.6 × 10^−7	100	1	32
Silica	6.8	—	—	17.4 × 10^−7	200	1	32
Silica	15.5	—	—	20.9 × 10^−7	400	0.2–1	34
Silica	41	—	—	4.04 × 10^−7	200	2	32
Silica	8	—	—	1.85 × 10^−7	200	1	32
Silica	—	32	—	10.7 × 10^−7	500	—	35
Silica	—	17	—	0.57 × 10^−7	220	—	36
Silica	—	8	—	20 × 10^−7	250	—	32
Silica	19.9	23.8	26.9	9.5 × 10^−7	200	1	This work
Silica	22.2	26.2	29.5	11.7 × 10^−7	200	2	This work

---

This aspect visualizes to the reader a scenario in which a great performance is achievable by the synthesized silica membrane. According to the use of a similar sol–gel method for the synthesis of a silica layer, the higher performance of the silica membrane is probably is to the good surface modification of the α-alumina support by the γ-alumina intermediate layers.

3.3. Silica MR performance in MSR process

To understand the ability of the synthesized silica membrane, an experimental study has been carried out to evaluate the effect of the most important operating conditions on silica MR performance compared to TR, in terms of methanol conversion, total hydrogen yield, CO selectivity and hydrogen recovery. In particular, the effect of feed molar ratio, reaction temperature and gas hour space velocity (GHSV) was investigated for both the silica MR and TR.

Table 4 reports the operating conditions used to carry out the MSR reaction in both silica MR and TR. In general, the experimental analysis can be divided into three parts, in which the feed molar ratio, reaction temperature and GHSV are changed.

Table 4 The investigated conditions for the silica MR and TR

Operating parameters	Feed molar ratio (steam/MeOH) effect	Temperature effect	GHSV effect
Pressure/bar	1.5	1.5	1.5
GHSV/h^−1	6000	6000	6000–10 000
Temperature/°C	300	240–300	300
H2O/CH3OH	1–3	3	3

---

3.3.1. Evaluation of reaction temperature effect. The influence of temperature on both silica MR and TR performances in terms of methanol conversion (eqn (1)), total hydrogen yield (eqn (2)), hydrogen recovery (eqn (3)) and CO selectivity (eqn (4)) was evaluated. As reported in Table 6, the experiments were carried out by maintaining the reaction pressure equal to 1.5 bar, H₂O/CH₃OH = 3 and GHSV = 6000 h⁻¹. The reaction temperature has been varied between 240 and 300 °C.

Fig. 7 shows methanol conversion versus reaction temperature for two cases, namely silica MR and TR. For each case, methanol conversion increases with rising temperature owing to the endothermic character of the reaction system. Moreover, it is evident that a higher methanol conversion is achievable by using silica MR with respect to TR due to the products’ removal from the reaction zone through the silica membrane, which shifts the reactions towards further product formation with consequent methanol consumption.

Fig. 7 Methanol conversion versus reaction temperature for silica MR and TR (at 1.5 bar, H₂O/CH₃OH = 3 and GHSV = 6000 h⁻¹).

Furthermore, Fig. 8 shows total hydrogen yield versus reaction temperature for both silica MR and TR. The hydrogen yield is improved by increasing the reaction temperature in both cases. In fact, by considering the eqn (2), a higher temperature can result in a higher hydrogen production rate during MSR reaction. On the other hand, it is clear that higher hydrogen yield is achievable by using silica MR with respect to TR owing to the products’ removal from reaction zone through the silica membrane which can shift the MSR reaction towards further hydrogen formation.

Fig. 8 Total hydrogen yield versus reaction temperature for the silica MR and TR (at 1.5 bar, H₂O/CH₃OH = 3 and GHSV = 6000 h⁻¹).

A further comparison between the silica MR and TR is presented in Table 5, where the CO selectivity values are reported as function of the reaction temperature.

Table 5 CO selectivity and hydrogen recovery for silica MR and TR versus reaction temperature (at feed molar ratio 3, 1.5 bar, Ar-flow rate = 25 ml min⁻¹ and GHSV = 6000 h⁻¹)

Temperature/°C	CO-selectivity (%)		Hydrogen-recovery (%)
	Silica-MR	TR	Silica-MR
240	0.49	0.68	30.83
270	1.35	1.72	34.71
300	1.61	2.01	37.95

---

This table shows the lower CO selectivity in the silica MR with respect to the TR outlet composition at a feed molar ratio of 3, GHSV equal to 6000 h ⁻¹ and reaction pressure of 1.5 bar. This result can be ascribed to the “shift effect”, which induced the shift of the WGS reaction towards the products, allowing a greater CO consumption in the silica MR.

Moreover, it’s evident that CO selectivity is increased by enhancement of reaction temperature for both cases. It can be related more to the role of the methanol decomposition (MD) reaction at higher temperatures.

Meanwhile, as reported in Table 5, the hydrogen recovery is improved by increasing the reaction temperature in silica MR. In fact, regarding eqn (3), a higher temperature results in a higher hydrogen permeation flux involving a higher hydrogen stream recovered in the permeate side. However, it should be noted that lower values of hydrogen recovery can be improved by increasing the reaction pressure.^23–26

3.3.2. Evaluation of feed molar ratio effect. Another important operating parameter taken into account in the present study is the feed molar ratio (steam/MeOH). The influence of feed molar ratio on the both silica MR and TR performances in terms of methanol conversion and hydrogen recovery was evaluated. According to the MSR reaction, an enhancement of steam/MeOH can shift the reaction toward the products. Therefore, regarding Le Châtelier’s principle, methanol conversion and hydrogen production can be improved by increasing the feed molar ratio.

As reported in Table 4, the experimental tests were carried out at 300 °C, 1.5 bar and GHSV = 6000 h⁻¹ by varying the steam/MeOH between 1 and 3.

Fig. 9 shows that the methanol conversion is increased by enhancement of the feed molar ratio. In particular, at a feed molar ratio of 3, a methanol conversion of 88.7% was achieved for silica MR, whereas an 84.4% methanol conversion was obtained for TR. It is clear that the higher performance grade of silica MR with respect to TR at a feed molar ratio of 3 is lower than that which can be achieved with a feed molar ratio of 1. This phenomenon is probably related to the steam concentration polarization effect on the silica membrane.

Fig. 9 Methanol conversion versus feed molar ratio (steam/MeOH) for the silica MR and TR (at 1.5 bar, 300 °C and GHSV = 6000 h⁻¹).

Moreover, as depicted in Fig. 10, increasing the feed molar ratio from 1 to 3 can increase hydrogen production and, as a consequence, the total hydrogen yield. It’s obvious that the higher total hydrogen yield is achievable by using silica MR with respect to TR owing to the shift effect. By comparing Fig. 9 and 10, a similar trend of methanol conversion was obtained for the hydrogen yield at a higher feed molar ratio with respect to the lower molar ratio. It means that the higher steam value can affect silica performance, nevertheless, performance loss is only 3% for the higher feed molar ratio (steam/MeOH = 3).

Fig. 10 Total hydrogen yield versus feed molar ratio (steam/MeOH) for the silica MR and TR (at 1.5 bar, 300 °C and GHSV = 6000 h⁻¹).

Another comparison terms between the silica MR and TR is shown in Table 6, where the CO selectivity values and hydrogen recovery are reported as a function of the feed molar ratio.

Table 6 CO selectivity and hydrogen recovery for silica MR and TR versus feed molar ratio (at 300 °C, 1.5 bar, Ar-flow rate = 25 ml min⁻¹ and GHSV = 6000 h⁻¹)

Feed molar ratio	CO-selectivity (%)		Hydrogen-recovery (%)
	Silica-MR	TR	Silica-MR
1	3.3	5	43.1
2	2.49	3.21	41.3
3	1.61	2.01	37.95

---

Table 8 indicates the lower CO selectivity in the silica MR with respect to the TR outlet composition at a reaction temperature of 300 °C, GHSV equal to 6000 h ⁻¹ and reaction pressure of 1.5 bar. This result, which can be referred to as the “shift effect”, induces the shift of the WGS reaction towards the products (CO₂ and H₂) and allowed a greater CO consumption in the silica MR. In addition, the CO selectivity is decreased by increasing the steam ratio in the case of both silica MR and TR.

Furthermore, as illustrated in Table 6, the hydrogen recovery is decreased by enhancement of the feed molar ratio. The trend observed in this table is probably due to the steam concentration polarization effect on the silica membrane and consequently the lower hydrogen permeation inside the silica MR at high feed molar ratios.

3.3.3. Evaluation of GHSV effect. A further parameter that can strongly affect the silica MR and TR performance is the GHSV. In this case, the experiments were accomplished at 300 °C, 1.5 bar and steam/MeOH = 3 by varying the GHSV between 6000 h⁻¹ and 10 000 h⁻¹. As reported in Table 7, the methanol conversion is decreased by increasing the GHSV for the silica MR and TR. By decreasing the GHSV, a higher residence or contact time in the reaction zone is favored. In fact, the lower values of GHSV can favour hydrogen formation in the reaction side. Therefore, a greater methanol conversion and hydrogen yield can be justified by decreasing the GHSV. Indeed, this effect produces a higher retentate hydrogen partial pressure that enhances the hydrogen permeation driving forces with a consequent more-effective shifting of the MSR reaction towards the products. Therefore, this gives more methanol consumption and a greater hydrogen production as well as a higher hydrogen stream permeating the membrane. Hence, the higher performance of silica MR with respect to TR in terms of methanol conversion and hydrogen yield can be realized (see Table 7).

Table 7 Methanol conversion and total hydrogen yield for silica MR and TR versus GHSV (at 300 °C, 1.5 bar, Ar-flow rate = 25 ml min⁻¹ and feed molar ratio 3)

GHSV/h^−1	Conversion (%)		Total yield (%)
	Silica-MR	TR	Silica-MR	TR
6000	88.50	84.1	85.1	82.3
10 000	82.61	79.10	79.62	76.8

---

Table 8 specifies the trend of CO selectivity and hydrogen recovery versus GHSV values. It is found that hydrogen recovery is improved by decreasing GHSV. In fact, higher retentate hydrogen partial pressure due to higher hydrogen production rate in lower GHSV values is the main specification of hydrogen recovery improvement.

Table 8 The CO selectivity and hydrogen recovery for silica MR and TR versus GHSV (at 300 °C, 1.5 bar, Ar-flow rate = 25 ml min⁻¹ and feed molar ratio 3)

GHSV/h^−1	CO-selectivity (%)		Hydrogen-recovery (%)
	Silica-MR	TR	Silica-MR
6000	1.61	2.01	37.9
10 000	1.91	2.41	27.8

---

Moreover, the CO selectivity is decreased by the enhancement of GHSV. Indeed, this trend can result in a lower conversion of the WGS reaction at higher GHSV values. On the other hand, the lower CO values in silica MR with respect to TR can a result of the extractor role of hydrogen through the silica membrane.

A further investigation was carried out to analyze the molar composition during the MSR reaction for silica MR with respect to TR. As reported in Table S2,† regarding the feed and permeate flow rates, it can be concluded that there is a higher hydrogen and lower carbon monoxide flow rate in the silica MR when compared to the TR.

4. Conclusion

In the present work, a high quality microporous silica membrane was synthesized by the polymeric sol–gel method showing a temperature-dependent flux of activated transport. In the synthesis of the composite silica membrane, a new successful strategy was used for the surface modification of the α-alumina support, in which a particle-size-control of the boehmite sol was applied. The permeance test results strongly suggest that the higher quality of the surface modification of supports can directly affect the silica membrane performance. Hence, the improvement of the silica membrane performance can extend its application. Regarding the main purpose of work, the analysis of methanol conversion, hydrogen yield, hydrogen recovery and CO selectivity was performed in a silica MR during the MSR reaction. According to the results, it was concluded that the methanol conversion and total hydrogen yield improved by increasing the reaction temperature and feed molar ratio and also by decreasing the GHSV, while CO selectivity decreased by the variations mentioned above. However, increasing the feed molar ratio had negative effect in terms of hydrogen recovery. As a consequence, a 6% enhancement of performance can be achieved by the silica MR in comparison with TR. As a future direction, this study indicates that silica MRs can be introduced as a promising option for hydrogen production in comparison with competitive processes.

Acronyms list

ATSB	Aluminum-tri-sec-butylate
Ethanol	EtOH
GHSV	Gas hourly space velocity
Methanol	MeOH
MR	Membrane reactor
MSR	Methanol steam reforming
PEMFC	Proton exchange membrane fuel cell
MD	Methanol decomposition
SEM	Scanning electronic microscope
TEOS	Tetraethylorthosilicate
TR	Traditional reactor
WGS	Water gas shift

Acknowledgements

The authors wish to thank Sahand University of Technology (SUT) for the financial support of this work. Also, we thank our co-workers and the technical staff in the department of chemical engineering, the ITM-CNR and nanostructure materials research center (NMRC) of SUT for their help during various stages of this work.

References

1. J. Wee, Renewable Sustainable Energy Rev., 2007, 11, 1720.
2. S. Bose, T. Kuila and T. X. H. Nguyen, Prog. Polym. Sci., 2011, 36, 813.
3. A. Iulianelli, P. Ribeirinha, A. Mendes and A. Basile, Renewable Sustainable Energy Rev., 2014, 29, 355.
4. R. Y. Chein, Y. C. Chen, Y. S. Lin and J. N. Chung, Int. J. Hydrogen Energy, 2012, 36, 466.
5. G. Q. Lu, J. C. Diniz de Costa, M. Duke, S. Giessler, R. Socolow and R. H. Williams, J. Colloid Interface Sci., 2007, 314, 589.
6. S. Xiu and A. Shahbazi, Int. J. Hydrogen Energy, 2012, 16, 4406.
7. A. Basile, A. Iulianelli, T. Longo, S. Liguori and M. De Falco, Pd-Based selective membrane state of the art, in Membrane Reactors for Hydrogen Production Processes, ed. L. Marrelli, M. de Falco and G. Iaquaniello, Springer, London Dordrecht Heidelberg NewYork, 2011, ch. 2, pp. 21–55.
8. A. S. Damle, J. Power Sources, 2009, 186, 167.
9. A. Iulianelli, T. Longo and A. Basile, J. Membr. Sci., 2008, 323, 235.
10. B. K. R. Nair and M. P. Harold, Chem. Eng. Sci., 2006, 61, 6616–6636.
11. A. Basile, A. Parmaliana, S. Tosti, A. Iulianelli, F. Gallucci, C. Espro and J. Spooren, Catal. Today, 2008, 13, 717.
12. F. Gallucci, A. Basile, S. Tosti, A. Iulianelli and E. Drioli, Int. J. Hydrogen Energy, 2007, 32, 1201.
13. C. H. Fu and J. C. S. Wu, Int. J. Hydrogen Energy, 2007, 32, 4830.
14. L. Shao, B. T. Low, T. S. Chung and A. R. Greenberg, J. Membr. Sci., 2009, 327, 18.
15. A. F. Ismail and L. I. B. David, J. Membr. Sci., 2001, 193, 1.
16. G. A. Sznejer, I. Efremenko and M. Sheintuch, AIChE J., 2004, 50, 224.
17. B. N. Nair, W. J. Elferink, K. Keizer and H. Verweij, J. Colloid Interface Sci., 1996, 178, 565.
18. R. M. de Vos, W. F. Maierb and H. Verweij, J. Membr. Sci., 1999, 158, 277.
19. D. W. Lee, S. J. Park, C. Y. Yu, S. K. Ihm and K. H. Lee, J. Membr. Sci., 2007, 302, 265.
20. A. Jabbari, K. Ghasemzadeh, P. Khajavi, F. Assa, M. A. Abdi, A. A. Babaluo and A. Basile, Int. J. Hydrogen Energy, 2014, 39, 18585,  DOI:10.1016/j.ijhydene.2014.05.056.
21. D. W. Lee, S. J. Park, C. Y. Yu, S. K. Ihm and K. H. Lee, J. Membr. Sci., 2008, 316, 63.
22. D. W. Lee, S. E. Nam, B. Sea, S. K. Ihm and K. H. Lee, Catal. Today, 2006, 118, 198.
23. K. Ghasemzadeh, P. Morrone, A. A. Babalou and A. Basile, Int. J. Hydrogen Energy, 2015, 40, 3909,  DOI:10.1016/j.ijhydene.2014.04.010.
24. K. Ghasemzadeh, P. Morrone, S. Liguori, A. A. Babalou and A. Basile, Int. J. Hydrogen Energy, 2013, 38, 16698.
25. K. Ghasemzadeh, P. Morrone, A. Iulianelli, S. Liguori, A. A. Babalou and A. Basile, Int. J. Hydrogen Energy, 2013, 38, 10315.
26. K. Ghasemzadeh, S. Liguori, P. Morrone, A. Iulianelli, V. Piemonte, A. A. Babaluo and A. Basile, Int. J. Hydrogen Energy, 2013, 38, 16685.
27. A. A. Babaluo, M. Kokabi, M. Manteghian and R. Sarraf-Mamoory, J. Eur. Ceram. Soc., 2004, 24, 3779.
28. R. J. R. Uhlhorn, K. Keizer and A. J. Burggraaff, J. Mater. Sci., 1992, 27, 527.
29. R. M. de Vos and H. Verweij, J. Membr. Sci., 1998, 143, 37.
30. R. S. A. Lange, K. Keizer and A. J. Burggraaf, J. Membr. Sci., 1995, 99, 57.
31. A. Ayral, A. Julbe, V. Rouessac, S. Roualdes and J. Durand, Membr. Sci. Technol., 2008, 13, 33.
32. T. V. Gestel, D. Sebold, F. Hauler, W. A. Meulenberg and H. P. Buchkremer, J. Membr. Sci., 2010, 359, 64.
33. V. Boffa, J. E. ten Elshof, R. Garcia and D. H. A. Blank, Microporous Mesoporous Mater., 2009, 118, 202.
34. M. Pakizeh, M. R. Omidkhah and A. Zarringhalam, Int. J. Hydrogen Energy, 2007, 32, 1825.
35. J. P. Collins, R. W. Schwarts, R. Sehgal, T. L. Ward, C. J. Brinker, G. P. Hagen and C. A. Udovich, Ind. Eng. Chem. Res., 1996, 35, 4398.
36. R. Bredesen, K. Jordal and O. Bolland, Chem. Eng. Process., 2004, 43, 1129.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra14002a

---