Preparation and evaluation of Pd membrane on supports activated by PEG embedded Pd nanoparticles for ATR membrane reactor

Preparation and evaluation of Pd membrane on supports activated by PEG embedded Pd nanoparticles for ATR membrane reactor

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx Contents lists available at ScienceDirect Chemical Engineering & Processi...

3MB Sizes 0 Downloads 1 Views

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

Contents lists available at ScienceDirect

Chemical Engineering & Processing: Process Intensification journal homepage: www.elsevier.com/locate/cep

Preparation and evaluation of Pd membrane on supports activated by PEG embedded Pd nanoparticles for ATR membrane reactor S. Jamshidia, A.A. Babaluob,c,* a

Department of Chemical Engineering, Tabriz Branch, Islamic Azad University, P.O. Box 1655, Tabriz, IR, Iran Nanostructure Material Research Center (NMRC), Sahand University of Technology, P.O. Box 51335-1996, Sahand New Town, Tabriz, Iran c Faculty of chemical engineering of Chemical Engineering, Sahand University of Technology, P.O. Box 51335/1996, Tabriz, Iran b

A R T I C LE I N FO

A B S T R A C T

Keywords: Pd nanoparticles PEG template Surface roughness Hydrogen purification Membrane reactor Autothermal reforming

Stresses and formation of voids are the main origins of pinholes in palladium (Pd) composite membranes at high temperatures. The stresses are associated with the Pd layer-substrate interaction and the formation of the voids is due to the gas bubble trapped inside the deposits. Pd embedded polyethylene glycol (PEG) was used during the activation step of the electroless plating (ELP) method. Also, multiple plating and annealing steps were employed to suppress void formation. Three plating and annealing steps rendered the prepared Pd membrane quite dense and defect-free. The surface roughness of modified support was found to play a crucial role in the stability of the Pd membrane. The resulting membrane showed an infinite selectivity for H2/Ar with H2 flux of 1-6.6 m3 m-2 h-1. Besides, H2 permeate flux kept constant and the separation factor of H2/Ar over infinite for 240 h at different temperatures, indicating the high potential of the prepared membrane for H2 purification. Furthermore, membrane reactor experiments showed the CH4 conversion is about 15 % higher than that in the traditional reactor at the same condition.

1. Introduction

energy needed for the SMR reactions. This process is called the autothermal reforming of methane (ATR, Reaction (4)) [4,5].

Hydrogen is widely used in several industries, such as petroleum, metallurgical, electronics, etc. It is also lately being employed as a fuel in proton exchange membrane (PEM) fuel cell, this is because of its efficient energy conversion [1]. Among all the potential sources of hydrogen, natural gas (especially methane) has been considered a good option because it is clean, abundant and it can be easily converted to hydrogen [2]. The steam reforming of methane (SRM) has become the most important chemical process for the production of hydrogen from natural gas by using the following equations [3].

CH4 + H2 O ↔ CO + 3H2 ΔH °298 = 206 KJ / mol

(1)

CO + H2 O ↔ CO2 + H2 ΔH °298 = −41 KJ / mol

(2)

Steam reforming reaction (SRM, Reaction (1)) is strongly endothermic and water-gas shift reaction (Reaction (2)) is moderately exothermic. Due to its extremely endothermic nature, the process requires high operational temperatures to obtain reasonable conversions. The coupling of steam reforming and partial oxidation of methane (Reaction (3)) caused to give rise to a high amount of heat to supply the

CH4 +

1 KJ O2 → CO + 2H2 ΔH °298 = −36 2 mol

3CH4 + O2 + 4H2 O ↔ 3CO2 + 10H2 ΔH °298 = 11 KJ / mol

(3) (4)

The equilibrium graphs of ATR and SRM as a function of temperature were presented in Fig. 1. It’s clear that at high temperatures in the range of 700−800 K, the methane conversion is lower than 90 % even the ATR reactions because of equilibrium thermodynamics limitations. To separate H2 from the reaction medium, catalytic membrane reactors (CMR, the combination of membranes into ATR reactors) can be a new operational solution for overcoming the equilibrium limitations of ATR traditional reactors (TR). Indeed, by separation of H2 (one of the products), the equilibrium reaction shifts to the products side based on Le Chatelier’s principle [7]. Therefore, many attempts have been made to apply Pd-based membranes as one of the promising alternatives for H2 separation and purification [8–10]. Several techniques, such as ELP [11], sputtering [12] and electroplating (EPD) [13] and have been thus far used to fabricate Pd and/or its alloy membranes.

⁎ Corresponding author at: Nanostructure Material Research Center (NMRC), Sahand University of Technology, P.O. Box 51335-1996, Sahand New Town, Tabriz, Iran. E-mail address: [email protected] (A.A. Babaluo).

https://doi.org/10.1016/j.cep.2019.107736 Received 31 March 2019; Received in revised form 3 November 2019; Accepted 7 November 2019 0255-2701/ © 2019 Published by Elsevier B.V.

Please cite this article as: S. Jamshidi and A.A. Babaluo, Chemical Engineering & Processing: Process Intensification, https://doi.org/10.1016/j.cep.2019.107736

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

reducing agent plays a crucial role in the activation step of the ELP method. Unfortunately, there are a few reports on applying the combined organic and inorganic approach for the synthesis of high-quality Pd membranes. To the best of our knowledge, poly 2,6-dimethyl-1,4phenylene oxide (PPO) has been the only organic template used in the combined organic and inorganic method [16,19,20]. Therefore, a study on the potential of different polymer templates in the Pd membrane synthesis is appreciated. In this study, thin and dense Pd membranes with high stability were synthesized via electroless plating method. To prevent the formation of defects and pinholes of the as-deposited Pd film and to avoid the decline of the membrane selectivity during annealing, a couple of procedures were followed. First, a combined organic and inorganic method was used during the activation step where polyethylene glycol (PEG) was used, for the first time, as the polymer template in this work. Furthermore, we previously showed that PEG acts as a high potential reducing agent for the synthesis of Pd nanoparticles [21]. The effect of relative roughness of support in the presence of a small gap between the Pd layer and modified support on the Pd membrane stability was also investigated. In the second procedure, multiple plating and annealing steps were applied to render the membrane defect-free with high separation and purification performance. Then, the morphology, hydrogen permeability and thermal test through the prepared Pd membranes were studied. The prepared Pd composite membrane has been then applied for ATR tests in a membrane reactor.

Fig. 1. Equilibrium thermodynamic methane conversions of SRM and ATR as a function of temperature [6].

Nevertheless, the high-performance membrane with excellent thermal and chemical stability is a challenge for palladium-based membranes for exploitation in the industry. Namely, self-supporting palladium membranes need to be thick for sufficient mechanical strength. This raises the material cost and decreases the flux of hydrogen. Therefore, a more cost-effective approach has been employed by synthesizing thin composite membranes (with a thickness of several microns). Porous glass, ceramic and stainless steel have typically been utilized as the mechanically-stable supports of these membranes. However, the alumina substrate is very often used in lab-scale work, in particular on ELP [14]. Besides, the vacuum electroless plating (VELP) method is also used to improve the hydrogen permeation and the stability of the thin Pd composite membranes. This occurs owing to the higher deposition rate of palladium resulting from the vacuum on both sides of the tubular substrate. Therefore, a dense Pd composite membrane with a finer and more uniform microstructure can be obtained [15]. The stability of a membrane can also be improved by creating a small gap between the Pd-based layer and the support. Indeed, as reported in the literature [16–18], this small space might result in the mobility of the whole Pd membrane during thermal cycles, leading to a uniform force distribution on the Pd layer rather than imposing a local force on the small Pd anchor in the pores. For this purpose, some researchers prepared Pd-based foil (by the cold rolling method [17]) mechanically covered on the porous substrate to make a small gap between the Pd-based membrane and the porous substrate. However, the large membrane thickness (around 50 μm for this case), leads to a quite low hydrogen permeability and costly membrane fabrication. Deposition of Pd film with a thickness ranging from 0.5 and 8 μm via sputtering and then placing the as-deposited thin film on a porous substrate surface is another method for forming a gap [18]. In this method, high hydrogen permeance was attained but the respective preparation procedure is rather troublesome where the welding zone of the tubular membrane could be vulnerable to form defects. Recently, a combined organic and inorganic approach has been employed to coat the thin and defect-free Pd based membrane while creating a small gap between the Pd-based layer and the porous substrate [16,19,20]. In this technique, Pd (used as the catalyst for the fabrication of the membrane by ELP) embedded in a polymer template film is uniformly deposited on a porous substrate. Then, the polymer template is removed during the heat treatment, resulting in a small interstice between the porous substrate and the metal layer. In organic and inorganic approach, the polymer template as a

2. Experimental 2.1. Membrane fabrication The preparation of the Pd composite membrane mainly involved the modification of porous alumina support, activation of the modified support and deposition of the palladium layer on the activated support by the ELP method. 2.1.1. The modification of porous alumina support The α-alumina tubular supports were prepared by the gel-casting technique [22]. Introducing an intermediate layer between the top Pdbased membrane and the porous support caused reduction of any inherent defects of the support and prevention of the top layer material infiltration into the pores of the support. So, after cleaning of nonmodified support in acetone solution by ultrasonic bath for 10 min, two types of intermediate layers were applied. As reported in our previous work [23], the first intermediate layer was obtained by dipping the support in the TiO2-boehmite suspension for 30 s. The coated layer was dried for 24 h at ambient temperature and sintered vertically at 1273 K for 3 h in an electric furnace. This procedure was repeated two times. The TiO2-bohemite modified support was dipped into the γ-AlOOH sol for 10–30 s and then dried in air at room temperature for 12 h. Finally, the modified support was calcined in air at 873 K for 3 h [24]. This procedure was also repeated two times. 2.1.2. Activation of the modified porous alumina support As reported in our previous work [21], for the fabrication of palladium nanoparticles, palladium acetate was added into the melting PEG matrix by magnetic stirring. Then the solution was sonicated for 1 h. The activation process was done by a coating of modified support in a polymer solution containing Pd nanoparticles. 2.1.3. The deposition of palladium on the activated support Finally, a Pd layer was deposited on the outside surface of the activated support during two stages. Pre-electroless plating was carried out using plating solution including palladium chloride (2 g L-1, Merck Co.) and EDTA sodium (Merck Co.) as a stabilizer at 343–353 K. In this stage, PEG acted as a reducing agent and no dense Pd layer was formed. 2

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

the main cell. The pressure in the system was controlled with a back pressure control valve in the retentate side, whereas the permeate stream was kept at atmospheric conditions. The ATR CMR consists of H2 selective tubular membrane (10 mm outer diameter and 75 mm length), 10 g of commercial ATR catalyst (Ni/MgO/Al2O3, KATALCOJM- catalyst-23 8Q) and 20 g of inert material of the same size. The catalysts and inert materials were packed in the annular space between the shell and membrane. Every annealing step was done under an inert gas atmosphere to protect both catalyst and membrane. The catalyst was reduced at 773 K for 5 h in a stream of hydrogen. All the experiments were carried out under isothermal conditions at 773 K and trans-membrane pressures in the range 1.013–2.026 × 105 Pa. The space velocity 480 (GHSV, L/h•g catalyst) was maintained for all the ATR CMR tests. A feed gas composition was used: air to carbon = 2.4 (dry basis), and steam to carbon = 2. An inert gas, N2, was supplied to the permeation side (inside the membrane) in a cocurrent manner as a sweep gas. After eliminating the remaining water by a cold-trap, the product streams were investigated on-line by using a gas chromatograph (GC, Tief Gostar, Ind. Co.) equipped with a thermal conductivity detector (TCD) and flame ionization detector (FID) with two analytical columns. Each analysis was repeated three times to minimize experimental error. The methane conversion, hydrogen recovery (HR) and hydrogen yield (HY) were defined as follows:

Then, the polymer layer was eliminated at high temperature (873–973 K) in an air atmosphere. This caused to form a small gap among the support and the thin preplated Pd layer. The obtained layer was reduced at 773−823 K for 3 h to be activated for formal electroless plating. After preplating and the formation of a thin Pd layer, a dense Pd layer was deposited by the VELP method. In this step, the plating solution included PdCl2 (3.6 g L-1, Merck Co.), Na2EDTA (76 g L-1, Merck Co.), ammonia (650mlL-1, Merck Co.) and hydrazine (10mlL-1, Acros Organics Co.). The whole bath system was placed in a thermostat chamber maintained at 318 K. The vacuum pump was also applied for creating of vacuum inside of the tubular support. The prepared membrane was washed with ammonia solution and hot deionized water in sequence and then dried at 393 K. Finally, the palladium membrane was annealed at 823 K for 3 h in a hydrogen atmosphere. 2.2. Characterization techniques The size and uniformity distribution of Pd nanoparticles in polymer solution was examined by particle size analyzer (FRITSCH, Analysette 22 NanoTec). Also, the palladium particle size in the activation solution was determined by transmission electron microscopy (TEM, CM-200 FEG Philips microscope). TEM sample was prepared by solving Pd/PEG in CH2Cl2 and then drops off the prepared colloidal solution were placed onto a carbon-coated copper grid. To the investigation of surface morphology for activated supports and palladium composite membranes and also to estimate the thickness of the Pd membrane, the scanning electron microscopy (SEM, Cam Scan MV2300 Czech Republic) was applied. The distribution of palladium nanoparticles in the activated support was analyzed by EDX-SEM. The crystalline structure of the Pd membrane was characterized by X-ray diffraction (XRD, Siemens D500). The gas permeation experiments were done by a permeation device. The side edges of the Pd composite membrane was sealed by hightemperature graphite gaskets. Feed gases flowed to the outside of the tubular membrane (retentate side) and the permeated gases flowed to the inside of the tubular membrane (permeate side). Permeate side was kept atmospheric and no sweep gas was applied. The measuring of permeated gas flow rate was done by a soap bubble flow meter. Before the high temperatures permeation examinations, Ar permeation tests up to 2.026 × 105 Pa were carried out at room temperature to define possible pinholes in the dense Pd layer. If no bubble arose in the permeate side after multi-step plating and annealing at 823 K, the membrane was appropriate for membrane and membrane reactor testing. For membrane testing at high temperature, the Pd composite membrane was heated in the Ar atmosphere up to 673 K and then was switched to pure H2 for hydrogen activation treatment. After the activation process was completed and hydrogen permeance reached a constant value, gas permeation tests were conducted with single gases (pure H2 and Ar) and H2–Ar binary mixtures over different pressures and temperatures. A bubble flow meter and an on-line gas chromatograph (Propack Q column, TCD detector, Tief Gostar, Ind. Co.) were applied for determining the permeated flow rate and components of the permeated gases, respectively. Finally, the thermal stability of the membrane was estimated by the hydrogen permeation flux versus the operation time at 700 and 773 K with a pressure difference of 1.013 × 105 Pa.

CH 4 Conversion (%) =

H 2 Recovery (%) =

H 2 Yield (%) =

nCH 4, in − nCH 4, out × 100 nCH 4, in

n H2, Perm × 100 n H2, Perm + n H2, Ret

(5)

(6)

nH 2, Per × 100 (7) nCH 4, in

Where nCH4 represents the methane molar flow in the inlet (subscript ‘in’) or the outlet (subscript ‘out’) of the system and nH2 the hydrogen molar flow in permeate (subscript ‘perm’) or retentate (subscript ‘ret’) side. 3. Results and discussions 3.1. Activated modified support Fig. 2a shows the particle size analysis of polymer solution containing Pd nanoparticles with three times of repetition. The average size of Pd nanoparticles is less than 20 nm with a narrow size distribution. The presence of narrow size distributed palladium nanoparticles was also visualized by transmission electron microscope (TEM) analysis (Fig. 2b). The synthesis of a Pd-polymer solution with narrow nanoparticles size distribution is a vital step to occur the plating reaction onto the support surface uniformly. The polymer layer containing Pd nanoparticles was obtained by three times of the dip-coating process. SEM images of the surface and cross-section of activated support after coating with polymer solution containing Pd nanoparticles (1 wt. % Pd) are presented in Fig. 3. As shown in Fig. 3a and b, a uniform, smooth and crack-free polymer layer was formed on the modified support with γ-alumina twice. Fig. 3c shows the palladium nanoparticles distribution on the activated surface. The density of the white points qualitatively shows the high concentration of palladium confirming a uniform distribution and large amount attendance of Pd nanoparticles on the activated surface. It is noticeable that the activated support involving the high concentration of palladium nanoparticles with uniform distribution has a strong effect during the electroless plating. It can control the relative rates of Pd nucleation that causes fewer defects and pinholes in the deposition of the Pd membrane. Therefore, Pd embedded in the PEG matrix can be one of the best activation procedures for electroless plating.

2.3. ATR CMR studies H2, N2, CH4, and O2 gases were fed by individually calibrated mass flow controllers (Unit MFC, 1660 Series) and then were mixed based on the desired composition. A Syringe pump was used for pumping of distilledwater to mix with gas feed stream. The mixture was further introduced in an evaporator to fully vaporize the water before to get in 3

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

Fig. 2. (a)Particle size distribution of Pd (II)-polymer solutions with three times repetition And (b) TEM micrograph of Pd-polymer solution.

method. Because, a small gap between the Pd layer and the substrate caused higher acting length (L) related to the traditional method [12]. So by applying organic and inorganic method, the acting length will be equal to the membrane tube length and shear stress decreases considerably (Eq. 8) [16]. While by increasing the modified support roughness in the presence of the small gap between Pd layer and substrate, the acting length enhanced due to the formation of a corrugated Pd layer (marked as the red line in Fig.4(a))., resulting in lower shear stress and then more stable Pd composite membrane. Also, there are many different types of internal stresses during annealing that can be controlled with mechanical interlocking. There are Failure modes for compressed films which susceptible to interface debonding or delamination. Buckling-driven delamination can occur if stresses are compressive. These failure modes can be analyzed with 2D solutions. Robust failsafe criteria for ensuring that the respective failure modes will not occur, is [25]:

3.2. Palladium membranes In most cases, the palladium membranes annealed at high temperature show the presence of a large number of pinholes that causes a decline in the membrane selectivity. On the other hand, the formation of pinholes can depend on the support, the coating process, the atmosphere and temperature during annealing, etc. Two effective parameters in the formation of these pinholes are the stresses associated with the Pd layer-support interaction and the formation of the voids due to the gas bubble trapped inside the deposits. To prevent the defects and pinholes formation in as-deposited Pd film and avoid the decline in the membrane selectivity during annealing, two procedures were applied. 3.2.1. First procedure: applying combined organic and inorganic method in the activation step Stresses in thin films and substrate have three primary origins: intrinsic, thermal and mechanical [20]. Thermal stresses increase by temperature changing when the film and substrate have different coefficients of thermal expansion (CTE). This type of stress is unlikely for Pd composite membrane on a ceramic support. A small gap between the Pd-based layer and ceramic support can omit the stresses associated with the Pd layer-support interaction. For this purpose, a combined organic and inorganic method (PEG containing Pd nanoparticles in the activation step) was applied to the modified supports without and with γ-alumina layers to create a small gap. After the first conventional vacuum ELP (Section 2.1.3), the Ar permeation measurements indicated that the prepared membranes were gas-tight and no leakage was found during the permeation test. However, after annealing in a hydrogen atmosphere at 823 K, the palladium membrane deposited on the activated support modified with γ-alumina layers peeled off. But a uniform Pd layer was observed when no γ-alumina layer was not applied during the modification of support. On the other hand, as illustrated in Fig. 4, SEM images of the support surface modified with and without modification by γ-alumina layers and back of peeled off Pd layer suggested that the relative surface roughness of modified support plays an important role in the stability of the Pd membrane. Based on the literature published [26], the shear stress can be calculated from the following equation:

τ=

F 2πrL

Ω=

hσ 2 EΓ

(9)

Where h is a film thickness, σ is a stress, E is a modulus and Γ is a toughness. If the "cracking number", Ω is less than 2 (critical value), the failure mode is excluded [25]. The toughness would be lowest for very smooth surfaces, because a very smooth substrate would result in a continuous brittle intermetallic layer that would minimize the fracture path length [27]. Therefore, relative roughness caused higher toughness and lower stress which results in a lower cracking numbers. Finally, it can be concluded that the relative roughness is needed to obtain a more stable Pd composite membrane. So for the following investigations in the next sections, the activated modified support without the γ-alumina layer was applied. 3.2.2. Second procedure: applying multi-step plating and annealing The N2H4 oxidation during the conventional ELP method results in the formation of N2 gas that can be trapped inside the deposits and formed a large number of voids. On the annealing, the voids agglomerated and formed a continuous crack across the Pd layer. The effect of sequence annealing and plating on the formation of the voids and Ar leak is presented in Fig. 5. After the first conventional vacuum ELP, the Ar gas-tight deposited Pd layer was annealed at 823 K and the voids in the Pd layer were agglomerated, resulting in the Ar leak after first annealing. The 2nd Pd layer was plated and annealed again. The second annealing caused again the formation of pinholes in the second deposited layer, but only a few pinholes align with the first layer pinholes. So the Ar leak after second annealing is declined. After the third plating and annealing, no Ar leak was observed that confirms

(8)

where τ is shear stress, F shear force, r the tube radius and L a characteristic length in which the shear stresses are active. Under thermal cycling and hydrogen loading, the shear stress for Pd-based supported membrane synthesized by the combined organic and inorganic method is smaller than that for the membrane synthesized by the traditional 4

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

the prepared Pd membrane is defect-free. The SEM images and XRD pattern of Pd composite membranes after three steps of plating and annealing are presented in Fig. 6. The prepared Pd membrane has finer and more uniform microstructure with tightly packed together to form the dense Pd layer with any pinholes as illustrated in Fig. 6a. Also, Fig. 6b shows the cross-section image of the Pd membrane that shows a thin Pd layer (3−4 μm) deposited tightly on the surface of the modified support. The peaks of the XRD pattern (Fig. 6c.) are assigned to the crystalline cubic form of metallic palladium. 3.3. The gas permeation of palladium composite membrane Generally, the permeation rate can be explained as follow:

J = Q (PRn − PPn )

(10)

Where J is hydrogen permeation flux, Q hydrogen permeance, PR and PP the hydrogen partial pressures in the retentate and permeates side respectively and n the pressure dependence factor (0.5–1) [14]. As shown in Fig. 7, hydrogen flux is linearly proportional to the transmembrane pressure at the different temperatures and then n is very close to 0.5. The factor n is equal to 0.5 suggests that the bulk diffusion through dense Pd layer is the rate-controlling step. The correlation between temperature and hydrogen permeance can be expressed by an Arrhenius-type relation as follow:

Q = Q0exp(

−Ea ) RT

(11)

where Q0 is a pre-exponential factor, Ea the activation energy, R the universal gas constant and T the absolute temperature. The activation energy Ea of the Pd membrane is calculated to be 20.43 kJ mol-1. The value of the activation energy for hydrogen permeation was similar to that reported in the literature [10]. Finally, the hydrogen flux at 700 and 773 K and pressure difference of 1.013 × 105 Pa versus the operating time was measured and the obtained results are shown in Fig. 8. No significant changes in hydrogen flux are observed within the durability operating time of 240 h. The gas exchange cycle between hydrogen and other inert gas (Ar) was used during operating time to check the stability of the membrane. The Ar leak measurement during time operation progressively proved that the hydrogen separation factor still was infinite under experimental conditions. So the results show that the prepared Pd composite membrane has good stability. Finally, Table 1 summarizes the performance of Pd composite membranes prepared by the organic and inorganic method as reported in the previous literature [16,19] and this work under similar conditions. From the table, it can be seen that the values of activation energy and permselectivity are similar. Permeate flux is another important parameter for the membrane selection which the overall flow resistance of gases permeating through the Pd composite membrane (Rmembrane=ΔP/Fmembrane) is equal to the sum of resistances in the support (Rsupport= ΔP/ Fsupport) and the Pd top layer (RPd-layer= ΔP/FPdlayer) [28].

ΔP ΔP ΔP = + Fmembrane FPd − layer Fsupport

(12)

Where F is flux (m m h ) and ΔP is pressure difference (Pa). It can be claimed that in the prepared membranes by organic and inorganic method, permeate flux reduces considerably after applying Pd layer on the support surface where the relative resistance of the Pd layer (RPdlayer/Rmembrane) in our work is comparatively less than that in the previous literature [16,19]. However, the main reason for the lower permeation of the prepared Pd composite membrane is high support resistance (Rsupport), so that the Rsupport in this work is one order of magnitude higher than that reported in the literature [16,19]. 3

Fig. 3. SEM-EDX images of activated support: (a) surface, (b) cross-section and (c) Pd distribution.

5

-2

-1

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

Fig. 4. Schematic and SEM Images of Pd membrane on modified support (a) without and (b) with γ-alumina layer.

Fig. 5. Schematic of multi-step plating and annealing and Ar leak progression at room temperature and 2.026 × 105 Pa. 6

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

Fig. 7. Hydrogen flux versus operating pressure difference at different temperatures.

Fig. 8. Hydrogen flux versus operating time with gas exchanging cycles at 700 and 773 K with a pressure difference of 1.013 × 105 Pa.

3.4. Enhancement of CH4 conversion in ATR CMR Before applying the prepared membrane for ATR CMR tests, the effect of several gases on H2 permeation flux has been examined to estimate the effect of other gases in the membrane reactor zone (Fig. 9). The obtained results showed a decrease in hydrogen flux compared to a similar tests in pure H2. The low hydrogen flux with hydrogen/nitrogen is due to the dilution of nitrogen and the reduction of the hydrogen partial pressure as a driving force for hydrogen permeation on the feed side. The dilution effect and the adsorption ability of steam on the Pd surface is the possible reason for the lower hydrogen permeation flux than for the addition of N2. The flux drop given by CH4 has been more than N2. This trend may be related to decomposing to carbon or metal carbides consequently of catalytic activity at the palladium surface. The adsorption mechanism of CO and CO2 at the surface of palladium are often chemical. The strong interactions between CO and palladium atoms caused a fast declining in the hydrogen permeation flux even at low CO concentration. These results are also in agreement [29,30].

Fig. 6. (a) SEM micrograph of the surfaces-section, (b) SEM-EDS micrograph of the cross-section, and (c) XRD pattern of as-prepared Pd composite membranes after third step platting-annealing.

7

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx This work

Deriving Force (bar)

1

1

21.3

20.43

H2/He =∞

H2/Ar=∞ 3.25 4.30 10.5 9.9

69.1 56.5

> 10 75.39 62.3 26.57 38.70 108 102.6

700 773

700 773 Organic- inorganic (PPO)

Organic- inorganic (PEG)

Pd/Alumina

Pd/Alumina

3.5

These facts could be considered in the analysis of ATR CMR tests. The CH4 conversion for a membrane reactor (MR) and a traditional reactor (TR) versus the pressure at 1.013–2.026 × 105 Pa was presented in Fig. 10. The number of moles of reactants is less than that of the products in ATR reaction leads to the reaction will go backward by increasing pressure in the reaction according to the Le Chatelier’s principle. So the reduction in CH4 conversion for the TR when there is an increasing pressure was observed. The CH4 conversion in MR is higher than in TR. H2 separation from the products in ATR reaction zone improved the forward reaction according to the Le Chatelier’s principle. So the CH4 conversion in MR was higher than that in TR. Fig. 11 shows the CO and CO2 percent in the reaction zone for the TR and MR at various pressures. The CO percent decreases when pressure is increased in both the TR and MR, while the CO percent in the MR is always ca.5 % lower than in the TR. Concomitantly, the CO2 percent increases when there is an increase in pressure, both in the TR and MR. The percent of CO2 in the TR is about 3 % higher than that in the MR. These observations indicate that the water-gas shift reaction was improved with pressure increasing and the presence of H2 selective

*With Ar leak test in intervals more than.20 h

RPd layer (%) RMembrane

Pd composite Support

5

Temperature (K) Method

Thickness (μm)

Fig. 9. Hydrogen permeation flux for 10 % other gases: 90 % H2 feed mixture at T = 773 K and ΔP = 9 × 104 Pa.

Fig. 10. CH4 conversion versus operating pressure difference for the MR and the TR at 773 K.

Membrane

Table 1 Comparison of Pd-based membranes prepared by organic-inorganic techniques.

Flux (m3m-2.h-1)

RSupport , this work Rsupport , Ref 11, 14

Selectivity

Ea (Kjmol-1)

Durability tests

T=773K ΔP=1.013×105Pa Time= 200h T=773K ΔP=1.013×105Pa Time= 240h*

Ref

[16,19]

S. Jamshidi and A.A. Babaluo

8

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

4. Conclusion The embedding Pd nanoparticles in PEG as the polymer template in the activation step and multiple plating and annealing steps were proposed to plate thin and defect-free Pd-based membrane on the modified nanostructure support by ELP. The composite membrane prepared by organic and inorganic method had uniform thickness but for improving stability, relative roughness of the modified support was necessary. On the other hand, the formed small interstice between modified support with relative roughness and membrane layer can greatly enhance stability by increasing characteristic length in which the shear stresses are active. Also, relative roughness decreases cracking number of failure mode by increasing toughness and decreasing stress. The results showed that after the third step plating and annealing, the obtained membrane was dense and defect-free with no decline in the membrane selectivity. The hydrogen permeation flux of 6.6 m3 m−2 h−1 and the infinite H2/Ar selectivity was attained at the temperature of 773 K with a pressure difference of 1.013 × 105 Pa for the synthesized Pd membrane. The membrane kept the stable performance of hydrogen permeation at high temperatures for operating time (240 h). The effect of other gases on H2 permeation flux has been shown that have a negative effect. The membrane reactor was able to reach high CH4 conversion to TR. An increase in the hydrogen recovery and hydrogen yield has been obtained for ATR CMR by increasing the pressure difference. The MR results demonstrated that the thermodynamic effect (negative effect) overcomes the permeation one (positive effect) at low pressures.

Fig. 11. Comparison of CO and CO2 percent for the MR and the TR at operating pressure difference at 773 K.

Acknowledgements The authors gratefully acknowledge Sahand University of Technology (SUT) for complementary support of this work. Also, thank co-workers and technical staff in the department of chemical engineering and nanostructure materials research center (NMRC) of SUT for their help during various stages of this work. References [1] K. Hemmes, G. Barbieri, Y. MooLee, E. Drioli, H. (J.H.W.) De Wit, Process intensification and fuel cells using a Multi-Source Multi-Product approach, Chem. Eng. Process. 51 (2012) 88–108, https://doi.org/10.1016/j.cep.2011.09.010. [2] M.B. Noureldin, N.O. Elbashir, M.M. El-Halwagi, Optimization and selection of reforming approaches for syngas generation from Natural/Shale gas, Ind. Eng. Chem. Res. 54 (2014) 1841–1855, https://doi.org/10.1021/ie402382. [3] T. Chompupun, S. Limtrakul, T. Vatanatham, Ch. Kanhari, P.A. Ramachandran, Experiments, modeling and scaling-up of membrane reactors for hydrogen production via steam methane reforming, Chem. Eng. Process. 134 (2018) 124–140, https://doi.org/10.1016/j.cep.2018.10.007. [4] A. Indarto, D.R. Yang, J. Palgunadi, J. Choi, H. Lee, H.K. Song, Partial oxidation of methane with Cu–Zn–Al catalyst in a dielectric barrier discharge, Chem. Eng. Process. 47 (2008) 780–786, https://doi.org/10.1016/j.cep.2006.12.015. [5] M.A. Murmura, M. Diana, R. Spera, M.C. Annesini, Modeling of autothermal methane steam reforming: comparison of reactor configurations, Chem. Eng. Process. 109 (2016) 125–135, https://doi.org/10.1016/j.cep.2016.08.019. [6] C.N. Ávila-Neto, S.C. Dantas, F.A. Silva, T.V. FrancoL, L. Romanielo, C.E. Hori, A.J. Assis, Hydrogen production from methane reforming: thermodynamic assessment and autothermal reactor design, J. Nat. Gas Sci. Eng. 1 (2009) 205–215, https://doi.org/10.1016/j.jngse.2009.12.003. [7] Y. Yan, H. Li, L. Li, L. Zhang, J. Zhang, Properties of methane autothermal reforming to generate hydrogen in membrane reactor based on thermodynamic equilibrium model, Chem. Eng. Process. 125 (2018) 311–317, https://doi.org/10. 1016/j.cep.2018.01.010. [8] M.R. Rahimpour, F. Samimi, A. Babapoor, T. Tohidian, S. Mohebi, Palladium membranes applications in reaction systems for hydrogen separation and purification: a review, Chem. Eng. Process. 121 (2017) 24–49, https://doi.org/10.1016/j. cep.2017.07.021. [9] G. Straczewski, J. Völler-Blumenroth, H. Beyer, P. Pfeifer, M. Steffen, I. Felden, A. Heinzel, M. Wessling, R. Dittmeyer, Development of thin palladium membranes supported on large porous 310L tubes for a steam reformer operated with gas-toliquid fuel, Chem. Eng. Process. 81 (2014) 13–23, https://doi.org/10.1016/j.cep. 2014.04.002. [10] G. Marigliano, G. Barbieri, E. Drioli, Equilibrium conversion for a Pd-based membrane reactor. Dependence on the temperature and pressure, Chem. Eng. Process. 42 (2003) 231–236, https://doi.org/10.1016/S0255-2701(02)00092-2.

Fig. 12. Hydrogen yield and Hydrogen recovery versus operating pressure difference at 773 K.

membrane. Also, this suggests that the MR could more easily avoid CO formation at a higher pressure. Fig. 12 shows the hydrogen yield and hydrogen recovery for MR at various pressures. An increase in the hydrogen recovery and hydrogen yield has been obtained for ATR CMR by increasing the pressure difference due to the enhancement of hydrogen permeation. At high reaction pressure, the reaction conversion reduces (negative effect) while the permeation driving force and then the hydrogen recovery increases (positive effect). On the other hand, these results confirmed that the permeation effect overcomes the thermodynamic one. To identify whether the H2 permeation flux from the used membrane decrease or not, the single gas experiments were repeated during ATR CMR tests. The tested Pd composite membrane indicated no decline in the ideal H2/Ar selectivity. This demonstrates that the synthesized Pd composite membrane in this work had high performance even during the CMR experiments.

9

Chemical Engineering & Processing: Process Intensification xxx (xxxx) xxxx

S. Jamshidi and A.A. Babaluo

[11] X. Li, T. Liu, D. Huang, Y. Fan, N.P. Xu, Preparation and characterization of ultrathin palladium membranes, Ind. Eng. Chem. Res. 48 (2009) 2061–2065, https:// doi.org/10.1016/j.memsci.2017.01.011. [12] C. Su, T. Jin, K. Kuraoka, Thin palladium film supported on SiO2-modified porous stainless steel for a high-hydrogen-flux membrane, Ind. Eng. Chem. Res. 44 (2005) 3053–3058, https://doi.org/10.1021/ie049349b. [13] W. Chen, X. Hu, R. Wang, Y. Huang, On the assembling of Pd/ceramic composite membranes for hydrogen separation, Sep. Purif. Technol. 72 (2010) 92–97, https:// doi.org/10.1016/j.seppur.2010.01.010. [14] S. Yuna, S. Ted Oyama, Correlations in palladium membranes for hydrogen separation: a review, J. Membr. Sci. 375 (2011) 28–45, https://doi.org/10.1016/j. memsci.2011.03.057. [15] X. Zhang, G. Xiong, W. Yang, A modified electroless plating technique for thin dense palladium composite membranes with enhanced stability, J. Membr. Sci. 314 (2008) 226–237, https://doi.org/10.1016/j.memsci.2008.01.051. [16] J. Tong, L. Su, K. Haraya, H. Suda, Thin Pd membrane on α-Al2O3 hollow fiber substrate without any interlayer by electrolessplating combined with embedding Pd catalyst in polymer template, J. Membr. Sci. 310 (2008) 93–101, https://doi.org/ 10.1016/j.memsci.2007.10.053. [17] S. Tosti, A. Basile, L. Bettinali, F. Borgognoni, F. Chiaravalloti, F. Gallucci, Longterm tests of Pd–Ag thin wall permeator tube, J. Membr. Sci. 284 (2006) 393–397, https://doi.org/10.1016/j.memsci.2006.08.006. [18] H. Klette, R. Bredesen, Sputtering of very thin palladium-alloy hydrogen separation membranes, Membr. Technol. 5 (2005) 7–9, https://doi.org/10.1016/S09582118(05)70414-6. [19] J.H. Tong, L.L. Su, K. Haraya, H. Suda, Thin and defect-free Pd-based composite membrane without any interlayer and substrate penetration by a combined organic and inorganic process, Chem. Commun. 10 (2006) 1142–1144, https://doi.org/10. 1039/B513613J. [20] M. Mukaida, N. Takahashi, K. Hisamatsu, M. Ishitsuka, S. Hara, H. Suda, K. Haraya, Preparation for defect-free self-supported Pd membranes by an electroless plating method, J. Membr. Sci. 365 (2011) 378–381, https://doi.org/10.1016/j.memsci. 2010.09.025. [21] P. Ahmadian Namini, A.A. Babaluo, B. Bayati, Palladium nanoparticles synthesis

[22]

[23]

[24]

[25] [26]

[27]

[28]

[29]

[30]

10

using polymeric matrix: poly (ethylene glycol) molecular weight and palladium concentration effects, Int. J. Nanosci. Nanotechnol. 3 (2007) 37–43. A.A. Babaluo, M. Kokabi, M. Manteghian, R. Sarraf-Mamoory, A modified model for alumina membranes formed by gel-casting followed by dip-coating, J. Eur. Ceram. Soc. 24 (2004) 3779–3787, https://doi.org/10.1016/j.jeurceramsoc.2004.01.007. B. Bayati, Y. Bayat, N. Charchi, M. Ejtemaei, A.A. Babaluo, M. Haghighi, E. Drioli, Preparation of crack-free nanocompositeceramic membrane intermediate layers on α-alumina tubular supports, Sep. Sci. Technol. 48 (2013) 1930–1940 doi/abs/ 10.1080/01496395.2013.786728. A. Jabbari, K. Ghasemzadeh, P. Khajavi, F. Assa, M.A. Abdi, A.A. Babaluo, A. Basile, Surface modification of α-alumina support in synthesis of silica membrane for hydrogen purification, Int. J. Hydrogen Energy 39 (2014) 18585–18591, https://doi. org/10.1016/j.ijhydene.2014.05.056. J.W. Hutchinson, Stresses and Failure Modes in Thin Films and Multilayers. Notes for a Dcamm Course, Technical University of Denmark, 1996. S. Tosti, L. Bettinali, S. Castelli, F. Sarto, S. Scaglione, V. Violante, Sputtered, electroless, and rolled palladium-ceramic membranes, J. Membr. Sci. 196 (2002) 241–249, https://doi.org/10.1016/S0376-7388(01)00597-X. E.I. Stromswold, R.E. Pratt, D.S. Quesnel, The effect of substrate surface roughness on the fracture toughness of Cu/96.5Sn-3.5Ag solder joints, J. Korean Inst. Electr. Electron. Mater. Eng. 23 (1994) 1047–1053, https://doi.org/10.1007/ BF02650374. R.S.A. De Lange, J.H.A. Hekkink, K. Keizer, A.J. Burggraaf, Formation and characterization of supported microporous ceramic membranes by sol–gel modification techniques, J. Membr. Sci. 99 (1995) 57–75, https://doi.org/10.1016/03767388(94)00206-E. A.L. Mejdell, M. Jondahl, T.A. Peters, R. Bredesen, H.J. Venvik, Effects of CO and CO2 on hydrogen permeation through a ∼3 μm Pd/Ag 23 wt.% membrane employed in a microchannel membrane configuration, Sep. Purif. Technol. 68 (2009) 178–184, https://doi.org/10.1016/j.seppur.2009.04.025. F. Gallucci, F. Chiaravalloti, S. Tosti, E. Drioli, A. Basile, The effect of mixture gas on hydrogen permeation through a palladium membrane: experimental study and theoretical approach, Int. J. Hydrogen Energy 32 (2007) 1837–1845, https://doi. org/10.1016/j.ijhydene.2006.09.034.