- Email: [email protected]
alumina membranes prepared by high power impulse magnetron sputtering for hydrogen separation
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
PdAg/alumina membranes prepared by high power impulse magnetron sputtering for hydrogen separation S. Barison a,*, S. Fasolin a, S. Boldrini a, A. Ferrario a, M. Romano a,b, F. Montagner a, S.M. Deambrosis a, M. Fabrizio a, L. Armelao a,b a
Institute of Condensed Matter Chemistry and Technologies for Energy (ICMATE), National Research Council (CNR), Corso Stati Uniti 4, 35127 Padova, Italy b University of Padova, Dept. of Chemical Sciences, Via Marzolo 1, 35131, Padova, Italy
article info
abstract
Article history:
The development of hydrogen purification membranes that meet market demands such as
Received 16 January 2018
high purity, dynamic hydrogen production even at small scale, and reduced costs is still an
Received in revised form
open question. With this view, the present study aims at developing, for the first time, a
9 March 2018
method based on high power impulse magnetron sputtering for the deposition of Pd77Ag23
Accepted 11 March 2018
(wt%) films onto porous alumina substrates to achieve composite membranes with high
Available online 1 April 2018
hydrogen permeability and stability. This technique allows the deposition of films also on complex geometries and can be easily scaled up, thus making this technology a potential
Keywords:
candidate for preparing high performing membranes. Membranes made by stable and
PdAg
porous alumina supports and metallic, dense and crystalline Pd77Ag23 layers, from 3.5 mm
Hydrogen
to 17 mm thick, have been prepared and tested. The membranes showed good hydrogen
Membrane
permeability values, showing flux values up to a maximum of 0.62 molH2 m2 s1 at 450 C
High power impulse magnetron
and DP of 300 kPa. The resistance to hydrogen embrittlement and the chemical inertness to
sputtering
syngas were also demonstrated.
Alumina
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Hydrogen is a valuable energy vector and a largely consumed reagent in many industrial processes as ammonia production, hydrocarbon hydrogenation, etc. [1]. Most of hydrogen is currently produced by steam reforming of natural gas [2], although hydrogen production from biomass gasification could accelerate the H2 utilization as a future sustainable fuel [3,4]. In both processes, an H2-rich gas mixture is produced, and pure H2 is obtained by some chemical processes carried
out in a number of reaction units followed by separation/purification (mostly by pressure swing adsorption) [5]. The large number of steps to produce and separate hydrogen represents efficiency and cost limits of traditional reactors, that can be circumvented by using integrated systems such as membrane reactors [5], where both reaction and separation are carried out in the same device. Membrane technology is nowadays increasingly considered, thanks to several advantages including low energy consumption, ability to carry out separation continuously, and simple scaling up [6]. However, membranes will achieve commercial targets
* Corresponding author. E-mail address: [email protected] (S. Barison). https://doi.org/10.1016/j.ijhydene.2018.03.065 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
only when suitable hydrogen flux, cost, durability and tolerance to pollutants are optimized [7e9]. The most diffused membranes for hydrogen separation/ purification are nowadays based on palladium due to its very high hydrogen diffusion and permeability, although pure palladium suffers membrane poisoning and embrittlement. The shift, upon hydrogen uptake, from a-Pd(H) to b-Pd(H) phase, with a consequent volume increase, produces internal stresses, deformation and failure of the membrane. When compared to pure palladium, PdeAg alloys show important advantages, such as embrittlement resistance through the avoidance of a to b phase transition, robustness during thermal cycling, high permeability, and resistance to fouling [10e12]. Among these alloys, the Pd77eAg23 wt% composition has been largely investigated and tested due to its high permeance and resistance to hydrogen embrittlement [13]. Most commercial purification units contain conventional membranes consisting of relatively thick (>20 mm) Pd-alloy sheets or tubular membranes. However, to reduce the noble metal content, an effective strategy is needed to prepare dense membranes in the form of mm films deposited on suitable porous substrates. Various Pde and Pd alloys-based films have been deposited onto porous alumina, nickel or stainless steel [14], but they still lack of sufficient long-term stability and selectivity to H2. Electroless plating is commonly employed to deposit palladium-based films for this purpose. However, this technique is limited to few compositions/alloys [15]. Conversely, magnetron sputtering is effective in depositing alloys with variable compositions; it also allows the one stage deposition of multilayers in vacuum. In this work, we exploited a recent evolution of this technique, the High Power Impulse Magnetron Sputtering (HiPIMS). HiPIMS, introduced by Kouznetsov et al., in 1999 [16], is a successful technique for improving magnetron sputtering by pulsed power technology. Its main feature is the combination of sputtering from standard magnetrons and pulsed plasma discharges, with the aim of generating highly ionized plasma with large quantities of ionized sputtered material [17]. The high degree of ionization of the sputtered species is combined to a bias voltage applied to the substrate. This leads to the growth of smooth and dense films, to a good control on composition and microstructure, and allows improving film adhesion and uniformity also on substrates of complex shapes [18]. This feature could allow the deposition of films also on tubular membranes or other geometries useful for commercial plants. In this work, the most investigated composition, Pd 77 wt% e Ag 23 wt%, was chosen to evaluate HiPIMS influence on the preparation of membranes based on a Pd alloy widely studied in literature. Films with thickness ranging from 3 to 17 mm were deposited by a combined HiPIMS/Direct Current (DC) magnetron sputtering process onto porous alumina. Alumina has been chosen as substrate being mechanically and chemically stable in operating conditions. Alumina substrates present also the advantage of preventing interdiffusion phenomena at high temperatures, typical of steel substrates, an issue that can reduce hydrogen permeation. A procedure was set up to achieve substrates having a fine surface porosity to allow the deposition of a dense metallic layer with no need of any interlayer, that would increase the processing time, costs and complexity.
7983
Morphology, composition and structure of membranes were fully characterized. Hydrogen permeability was measured between 300 C and 450 C and at pressure difference up to 300 kPa. The resistance to hydrogen embrittlement and the chemical inertness to syngas were also investigated.
Experimental Materials and methods Alumina-based porous supports have been prepared by mixing 35 vol% of a-Al2O3 (Alfa Aesar, 99.9%) with 65 vol% of polymethyl methacrylate powders (PMMA) as pore former (Soken Chemical & Engineering, average size 1.5 mm). A wet ball milling process (absolute ethanol, Sigma Aldrich ACS reagent 99.8%) in zirconia jars was carried out in a planetary mill (Fritsch Pulverisette 7) at 350 RPM for a total of 2 h for each mixture. As obtained mixtures were uniaxially pressed (Nannetti Mignon SS/EA), in a 2.5 cm diameter mold by a 140 MPa load. The disks were then sintered at 1500 C in a high temperature furnace (Nabertherm HT 04/17), with an isotherm step of 1 h at the burning temperature of PMMA (386 C) and a very slow heating rate above 1000 C (30 C/h), when the shrinkage is maximum, to avoid pellet bending. The sintered disks were then polished and cleaned in an ultrasonic bath. The porosity was estimated by measuring the geometrical density and comparing it with the theoretical value of 3.98 g/ cm3 for alumina density. The membranes were deposited by a combination of HiPIMS and DC magnetron sputtering techniques. The high vacuum chamber was preliminary evacuated to a base pressure 1 104 Pa, while depositions were conducted in Argon atmosphere (Ar, 99.999% purity) at 1 Pa. During each sputtering run, the substrates were rotated at 5 rpm to improve homogeneity. The alloy was deposited by properly regulating the power on palladium and silver targets to achieve the Pd 77 wt % e Ag 23 wt% composition. By virtue of a deep preliminary set-up of the process, the palladium target (99.95% purity, 102 mm diameter) was driven by a HiPIMS power supply (Trueplasma HighPulse 4002, Hu¨ttinger Electronic, Germany) at an average power of 800 W (~10 Wcm2, pulse length 50 msec, frequency 500 Hz), while the silver target (99.99% purity, 102 mm diameter) was driven by a DC power supply (Trueplasma DC 4001, Hu¨ttinger Electronic, Germany) at 330 W (4 Wcm2). The substrate-target distance was set at 120 mm for Pd and 150 mm for Ag to achieve the selected composition. Moreover, a 100 V negative bias (Trueplasma Bias 3018, Hu¨ttinger Electronic, Germany) was applied to the substrates, which were heated up to 350 C. Temperature was maintained throughout the process by ad hoc heaters and thermocouple monitoring. Prior to perform Pd 77 wt% e Ag 23 wt% alloy deposition, a thin PdAg film was deposited at room temperature using the sputtering parameters listed above. The aim was to produce a conductive layer for a proper bias application during the next alloy deposition by using a properly designed sample holder. Only in this way it is possible to fully exploit the combined effect of the large number of target material ions produced during HiPIMS process with a bias voltage applied to the substrate.
7984
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Characterization Structural and microstructural information on the crystalline phases of the membranes were obtained by XRD and Rietveld refinement. The patterns were recorded at room temperature using a Philips PW 1830 diffractometer with Bragg-Brentano geometry, employing a Cu anode X-ray tube operated at 40 kV and 30 mA (angular range from 30 to 100 , 0.02 step, 6 s per step). Rietveld refinements were performed using MAUD software [19]. Surface and fractured surfaces of samples were observed by field emission scanning electron microscopy (FESEM) with a SIGMA Zeiss instrument (Carl Zeiss SMT Ltd, UK), operating in high vacuum conditions at an accelerating voltage of 20 kV and the composition was determined by Energy Dispersive Spectroscopy (EDS, Oxford X-MAX, UK). The surface roughness of alumina substrates was measured by a mechanical profiler (KLA Tencor P10, USA). Membrane permeability measurements were carried out by means of a custom-built stainless steel test station. A scheme of the permeation test apparatus is reported elsewhere [20]. Membranes were clamped and sealed in a stainless steel module using graphite gaskets. The module consists of two parts, feed side and permeate side, connected by a channel of about 1 cm in diameter, closed by the membrane to be tested, and placed in a furnace (Nabertherm N11/HR). The membrane housing temperature was monitored by a K-type thermocouple inserted directly in the module test. The gas flows at feed and permeate sides were set by independent mass flow controllers (1179A, 1179B and 647C, MKS). The feed side pressure was controlled by a Baratron pressure transducer (722B, MKS). Nitrogen (99.999% purity) was used to test membrane selectivity, while high purity hydrogen was produced by an electrolyzer (Perkin Elmer PGX Plus H2 160). During permeability tests, the feed side pressure was varied among absolute pressures of 110e400 kPa, while the permeate side was set at atmospheric pressure by a sweep gas flow. Pressure, flows and temperature were controlled and monitored by a Labview interface. Prior to each permeation test, the hydrogen flux was stabilized and monitored for few hours to ascertain or achieve a stable flux. After reaching stationary conditions, the hydrogen permeation was measured under typical operating conditions for these membranes, that is at 4 temperatures, from 300 C to 450 C, and at 16 DP values, from 300 kPa to 10 kPa (descending). For each
Fig. 2 e SEM backscattered electron image of the PdAg/ Al2O3 membrane with a 8.7 mm film thickness.
pressure point, the value was recorded after reaching a stable flux and the permeance at different pressures was tested twice for each temperature. The measures were repeated for membranes deposited in the same conditions to verify reproducibility. The selectivity was evaluated by the ratio between hydrogen and nitrogen permeances. Embrittlement tests were performed by fixing a DP value and by measuring the hydrogen flux as a function of temperature progressive reduction. Stability tests in commercial syngas (CO2 15 mol%, CO 15 mol%, H2 10 mol%, CH4 3.025 mol% and N2) were performed by exposing the membrane for various cycles to DP ¼ 200 kPa of syngas at 400 C, and by both monitoring the permeate flux by GC chromatography (Agilent Micro-GC 490) during syngas exposure and measuring the hydrogen flux after each cycle.
Results Mechanically and chemically stable porous substrates were prepared with a well-engineered porosity to reduce pressure drops in the substrate and a surface porosity sufficiently fine to deposit dense films, thus avoiding the need for interlayers [21]. The criteria for the selection of the pore former included decomposition with minimal residues, a resulting fine and
Fig. 1 e Surface SEM micrographs at different magnifications of a Pd77Ag23/Al2O3 membrane.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
well interconnected porosity and good mechanical stability. PMMA was chosen as pore former, with a mean powder size of 1.5 mm. Planar and porous substrates were obtained, with ~2 cm diameter (20e25% shrinkage), thickness between 500 and 1200 mm, porosity >35%, homogenous pore size distribution and a mean surface pore size of about 500e600 nm (estimated by SEM micrographs and by nitrogen and helium permeation measurements in substrates [20]). The hydrogen flux in porous substrates was measured and a minimal difference in the permeance was detected by changing their thickness, and this contribution was taken into account in elaborating data of permeation [20]. A preliminary deposition of Pd77 wt%-Ag23 wt% (4 min) was performed at room temperature prior to the deposition of the membrane, with the manifold scope of evaluating the
7985
influence of this film on the coating adhesion and creating a suitable conductive layer for the following application of bias during the membrane deposition. This layer resulted with columnar structure and about 1.5 mm thick. Once this layer was deposited, the sample was mounted in a specially designed sample holder that allowed the electrical contact with the conductive surface in order to have an effective bias application during the following alloy deposition. The HiPIMS deposition of the alloy was performed at 350 C. Various deposition durations were investigated for the alloy (from 30 to 180 min), leading to films ranging from about 3.5 to 17 mm (with a mean deposition rate of about 93 nm/min). Surface SEM micrographs of a membrane with a film thickness of about 8.7 mm (Fig. 1) show a very homogeneous, compact, dense and crystalline layer.
Fig. 3 e XRD patterns of the membrane a) as-deposited and b) after thermal treatment at 450 C for 20e24 h; in the figure * refers to Pd and D to Ag.
7986
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 2 shows a backscattered electron image of the cross sectional view of a membrane with a total thickness of about 8.7 mm. The micrograph highlights the PdAg porous interlayer deposited at room temperature and shows a dense and compact PdAg layer on top. EDS analyses were performed on surfaces and cross sections of all samples. The composition slightly varied among all samples and the mean value was Pd 78 wt% and Ag 22 wt%, with a standard deviation within all samples of ±2%, that is very close to the sought value. An example of EDS spectrum is reported in Figure S1 of Supplementary Information. XRD analysis of as-deposited membranes (see Fig. 3a) revealed that in as-prepared films the alloy was not formed, but crystalline Pd (cubic Fm-3m, cell parameter of 3.876 A, mean crystallite size of z140 nm) and Ag (cubic Fm-3m, cell parameter of 3.952 A, mean crystallite size of z123 nm) peaks
were still clearly distinguishable. Therefore, prior to any permeation test the membranes were conditioned for 20e24 h at 450 C in nitrogen to achieve the crystalline PdAg alloy (cubic Fm-3m, cell parameter of 3.916 A, mean crystallite size of z77 nm), as evidenced in Fig. 3b.
Hydrogen fluxes The hydrogen fluxes were measured between 300 and 400 C for all membranes with film thickness from 3.5 to 17 mm, in the 10e300 kPa ranges. Fig. 4 reports, as an example, the H2 fluxes measured at various temperatures as a function of the pressure difference between feed and permeate side for membranes with different PdAg film thickness. The flux values are higher for thinner films (up to 0.62 molH2 m2 s1 at DP ¼ 300 kPa and 450 C for the 3.5 mm film as indicated in
Fig. 4 e H2 flux measured between 300 and 450 C and as a function of the pressure difference between feed and permeate sides in membranes with PdAg film a) 3.5 mm or b) 8.7 mm thick.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 4a), and they decrease at increasing thickness (e.g. up to 0.28 molH2 m2 s1 at DP ¼ 300 kPa and 450 C for the 8.7 mm film reported in Fig. 4b). These values perfectly agree with literature data of PdAg films deposited with other methods (typically electroless plating). They are similar to flux values recorded for films with similar or higher thickness [10,22] and higher than values recorded for thicker films (25 mm [23]), while in some cases are lower than fluxes recorded for thinner films (1.2 mm [24]), thus demonstrating the effectiveness of this technique to deposit dense films with proper hydrogen flux values. To compare membranes with different PdAg layer thickness, the fluxes measured at DP ¼ 300 kPa as a function of temperature for samples with various thicknesses are reported in Fig. 5. A clear flux reduction, from a maximum of 0.62 molH2 m2 s1 for the 3.6 mm thick film to a minimum of 0.12 molH2 m2 s1 for the 17 mm thick film can be observed. By analyzing the trend of flux as a function of temperature, a general flux increase with temperature was observed for all samples, although in some membranes the flux decreased at 450 C. By extrapolating the data at various temperatures from permeances, apparent activation energies were found to decrease with film thickness, going from 7.3 kJ mol1 H2 for the thinnest PdAg film to 4.9 kJ mol1 H2 for the thickest sample. These values are analogous to those of fully formed and crystalline PdAg alloy/ceramic composite membranes with the same composition (see Ref. [25] and references therein). Ward and Dao deeply modeled the hydrogen permeation in Pd membranes [26] and found that for thin membranes (1 mm) at temperatures higher than 300 C the flux trend with temperature shows a diffusion-limited flow slightly influenced by the desorption-limited process, and they identified high activation energies for absorption or desorption processes (>42 kJ mol1 H ). However, considering a mass transfer resistance at low pressure side, typical for thin films deposited on porous supports, they found a significant reduction of fluxes and a trend with temperature that could explain the flux reduction observed at 450 C. Considering the thickness of membranes and the values of activation energies, we can hypothesize a permeation process mainly controlled by a
Fig. 5 e H2 flux values measured at DP ¼ 300 kPa vs T for samples with different thicknesses (indicated in the legend in mm).
7987
combination of hydrogen diffusion in bulk and mass transfer resistance in the alumina support, besides further possible transport phenomena at grain boundaries. Considering the Fick's first law to describe the atomic hydrogen diffusion through a homogeneous metal phase as a function of concentration gradient and diffusion coefficient, and the Sievert's law that may be used under certain conditions to describe the relationship between the concentration and pressure, Eq. (1) was derived to describe how the hydrogen flux (J) varies with temperature and pressure [8]. f qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi f0 eRT qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi Pfeed Pperm ¼ Pfeed Pperm L L Ea
j¼
(1)
where F is the permeability, that is defined as the rate of gas permeation per unit area, per unit driving force (pressure in this case), per unit membrane thickness, Ea is the activation energy for the whole process, F0 is the pre-exponential factor related to the permeability of the metal, R the gas constant, L the membrane thickness, Pfeed and Pperm are the hydrogen
Fig. 6 e H2 flux values at DP ¼ 300 kPa and permeability values at 400 C for all samples.
Fig. 7 e H2 flux at 400 C and DP ¼ 300 kPa before and after various cycles of exposure to syngas (DP ¼ 200 kPa) in a 17 mm thick sample. The membrane was exposed to N2 before and after exposure to syngas for few min. The red bars indicate each syngas exposure duration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
7988
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 8 e a) Surface SEM micrograph after permeability tests and syngas exposure of a 17 mm thick sample (the inset shows the morphology at the same magnification of the same sample before testing); b) section at high magnification of the subsurface region of the same sample (the inset shows total section of the sample, similar to as prepared sample).
pressure at the feed and at the permeate sides, respectively. However, this equation is valid for self-sustained thick membranes (commonly >10 mm). A more general formula has been typically used especially to estimate permeability or permeance (F/L) of thin palladium-based membranes, as in Eq. (2) [8]. j¼
f n P Pnperm L feed
(2)
Although with some limitations, this formula represents a general approach to take into account some phenomena that can lead to a best fit of flux data vs (Pnfeed e Pnperm) with an exponent n higher than 0.5. Among them, there are surface resistance to H2 absorption and desorption, mass transfer resistance in porous substrates, hydrogen diffusion as molecular H2 along grain boundaries and/or Knudsen or viscous flow through pores and film defects and leaks [8,27,28]. In thin Pd-based membranes (thickness from 1 to 7 mm), n was shown to approach 1 and this was typically attributed to surface phenomena as the rate-limiting step [24]. For all our samples the best fit of data was identified for n ¼ 0.8 (see an example of flux vs (Pnfeed e Pnperm) for 3 values of n in Figure S2 of Supplementary Information). This deviation from Sievert's law could be due to a combination of interface and grain boundaries controlled phenomena and mass transfer resistance in the porous supports at the low-pressure side. However, to compare these data with literature, the permeability and the permeance (F/L) were calculated by Eq. (2) with n ¼ 1. The permeances thus calculated at 400 C ranged between 5.6 107 for the thickest sample (17 mm) to 2.3 106 mol m2 s1 Pa1 for the thinnest sample (3.5 mm). The last value is similar or slightly smaller than permeance values identified in Pd and PdAg membranes on alumina having thickness around 2 mm (see Ref. [24] and references therein). Fig. 6 reports the flux at DP ¼ 300 kPa and the permeability values at 400 C for all samples. The permeability was in the 7 to 12 1012 mol m1 s1 Pa1 range for all samples. The selectivity values, calculated as the ratio between hydrogen and nitrogen permeances, ranged from about 30 for thinnest samples to about 580 for samples having thickness 8.7 mm (corresponding to nitrogen fluxes 4 104 mol m2 s1 at DP ¼ 300 kPa and 400 C, which is probably the sealing limit of our apparatus, as already observed in Ref. [20]). By analyzing data
with n ¼ 0.5, the permeability values at 400 C where between 6.6 and 8.4 109 mol m1 s1 Pa0.5, that is in agreement with the literature [22,29].
Membrane stability In this study, each membrane has been tested for at least 2 weeks, at temperatures ranging from 300 to 450 C and up to DP ¼ 300 kPa, showing flux stability over the time and no failure due to embrittlement. By decreasing the temperature down to room temperature, no abrupt increase of flux due to film embrittlement was ascertained, showing no failure of the film. Only in few cases, a small change in slope at temperatures below 200 C (see Fig. S3 as an example) could indicate the appearance of very small holes in the coating. SEM analyses after permeability tests and embrittlement tests showed very similar morphology to that of as-prepared samples. The stability of these membranes in commercial syngas was also tested. In particular, the hydrogen flux was measured at 400 C and DP ¼ 300 kPa before and after various cycles of exposure to syngas (Fig. 7). The hydrogen flux was almost constant and had no significant changes even after a total exposure of 13 h to syngas, suggesting the stability of the membrane. The chromatograms recorded during syngas tests (Fig. S4) detected only a peak due to hydrogen permeation and no other peaks within the gas chromatography uncertainty. The morphological characterization of membranes after syngas tests showed surface and fractured morphologies very similar to as-prepared samples (Fig. 8), with only a nanoporosity identified in fractured sections, but limited to outer membrane layers (100e200 nm close to the surface) (Fig. 8b). XRD analyses after embrittlement and stability tests did not evidence any difference with respect to diffractograms recorded before these tests (Fig. 3b).
Conclusions The deposition of PdAg films onto porous alumina substrates by a combined HiPIMS/DC technique was successfully performed. Membranes made by stable and porous alumina supports and metallic, dense and crystalline Pd77Ag23 layers,
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
from 3.5 mm to 17 mm thick, have been prepared and tested. Membranes showed good hydrogen permeability values, comparable or higher than literature values, showing flux values up to a maximum of 0.62 molH2 m2 s1 at 450 C and DP of 300 kPa. The resistance to hydrogen embrittlement and to syngas was also demonstrated. This technique allows the deposition of films on complex geometries, such as tubes, and can be easily scaled up, thus making this technology suitable for preparing high performing hydrogen selective membranes.
Acknowledgements The authors are grateful to Dr. Rosalba Gerbasi (CNR ICMATE) for XRD analyses. This work has been founded by Italian Industria 2015 project “Produzione di energia rinnovabile con il minimo impatto da un mix di biomasse e rifiuti speciali non pericolosi attraverso processi innovativi”.
Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.ijhydene.2018.03.065.
references
[1] Derwent R, Simmonds P, O'Doherty S, Manning A, Collins W, Stevenson D. Global environmental impacts of the hydrogen economy. Int J Nucl Hydrogen Prod Appl 2006;1:57e67. [2] Dincer I, Acar C. Review and evaluation of hydrogen production methods for better sustainability. Int J Hydrogen Energy 2015;40:11094e111. [3] Hosseini SE, Abdul Wahid M, Jamil MM, Azli AAM, Misbah MF. A review on biomass-based hydrogen production for renewable energy supply. Int J Energy Res 2015;39:1597e615. [4] Demirbas A. Biohydrogen generation from organic waste. Energy Sources Part A 2008;30(5):475e82. [5] Gallucci F, Fernandez E, Corengia P, Annaland M. Recent advances on membranes and membrane reactors for hydrogen production. Chem Eng Sci 2013;92:40e66. [6] Al-Mufachi NA, Rees NV, Steinberger-Wilkens R. Hydrogen selective membranes: a review of palladium-based dense metal membranes. Renew Sustain Energy Rev 2015;47:540e51. [7] Sjardin M, Damen KJ, Faaij APC. Techno-economic prospects of small-scale membrane reactors in a future hydrogenfuelled transportation sector. Energy 2006;31:2523e55. [8] Phair JW, Donelson R. Developments and design of novel (non-palladium-based) metal membranes for hydrogen separation. Ind Eng Chem Res 2006;45(16):5657e74. [9] Dolan MD. Non-Pd BCC alloy membranes for industrial hydrogen separation. J Membr Sci 2010;362(1e2):12e28. [10] Pizzi D, Worth R, Baschetti MG, Sarti GC, Noda K. Hydrogen permeability of 2.5mm palladiumesilver membranes deposited on ceramic supports. J Membr Sci 2008;325:446e53.
7989
[11] Okazaki J, Tanaka DAP, Tanco MAL, Wakui Y, Mizukami F, Suzuki TM. Hydrogen permeability study of the thin Pd-Ag alloy membranes in the temperature range across the alphabeta phase transition. J Membr Sci 2006;282:370. [12] Cheng YS, Pena MA, Fierro JL, Hui DCW, Yeung KL. Performance of alumina, zeolite, palladium, PdeAg alloy membranes for hydrogen separation from Towngas mixture. J Membr Sci 2002;204:329. [13] Karakaya I, Thompson WT. The Ag Pd (Silver-Palladium) system. Bull Alloy Ph Diagr 1988;9(3):237e43. [14] Yun S, Oyama ST. Correlations in palladium membranes for hydrogen separation: a review. J Membr Sci 2011;375:28e45. [15] Den Exter MJ. Palladium membrane technology for hydrogen production, carbon capture and other applications, 3. Elsevier; 2014. p. 43e67. k K, Schneider JM, Helmersson U, [16] Kouznetsov V, Maca Petrov I. A novel pulsed magnetron sputter technique utilizing very high target power densities. Surf Coating Technol 1999;122:290e3. [17] Sarakinos K, Alami J, Konstantinidis S. High power pulsed magnetron sputtering: a review on scientific and engineering state of the art. Surf Coating Technol 2010;204:1661e84. [18] Lundin D, Sarakinos K. An introduction to thin film processing using high-power impulse magnetron sputtering. J Mater Res 2012;27(5):780e92. [19] Lutterotti L, Mattheis S, Wenk HR. MAUD: a friendly Java program for material analysis using diffraction, 21. Newsletter of the CPD; 1999. p. 14e5. [20] Fasolin S, Barison S, Boldrini S, Ferrario A, Romano M, Montagner F, et al. Hydrogen separation by thin vanadiumbased multi-layered membranes. Int J Hydrogen Energy 2018;43(6):3235e43. [21] Zheng L, Li H, Xu H. “Defect-free” interlayer with a smooth surface and controlled pore-mouth size for thin and thermally stable Pd composite membranes. Int J Hydrogen Energy 2016;41:1002e9. [22] Wang L, Yoshiie R, Uemiya S. Fabrication of novel PdeAgeRu/Al2O3 ternary alloy composite membrane with remarkably enhanced H2 permeability. J Membr Sci 2007;306:1e7. [23] Wieland S, Melin T, Lamm A. Membrane reactors for hydrogen production. Chem Eng Sci 2002;57:1571e6. [24] Maneerung T, Hidajat K, Kawi S. Ultra-thin (<1 mm) internally-coated PdeAg alloy hollow fiber membrane with superior thermal stability and durability for high temperature H2 separation. J Membr Sci 2014;452:127e42. [25] Zeng G, Goldbach A, Shi L, Xu H. Compensation effect in H2 permeation kinetics of PdAg membranes. J Phys Chem C 2012;116(34):18101e7. [26] Ward TL, Dao T. Model of hydrogen permeation behavior in palladium membranes. J Membr Sci 1999;153:211e31. [27] Morreale BD, Ciocco MV, Enick RM, Morsi BI, Howard BH, Cugini AV, et al. The permeability of hydrogen in bulk palladium at elevated temperatures and pressures. J Membr Sci 2003;212(1e2):87e97. [28] Guazzone F, Ma YH. Leak growth mechanism in composite Pd membranes prepared by the electroless deposition method. AIChE J 2008;54:487e92. [29] Peters TA, Kaleta T, Stange M, Bredesen R. Hydrogen transport through a selection of thin Pd-alloy membranes: membrane stability, H2S inhibition, and flux recovery in hydrogen and simulated WGS mixtures. Catal Today 2012;193:8e19.
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.elsevier.com/locate/he
PdAg/alumina membranes prepared by high power impulse magnetron sputtering for hydrogen separation S. Barison a,*, S. Fasolin a, S. Boldrini a, A. Ferrario a, M. Romano a,b, F. Montagner a, S.M. Deambrosis a, M. Fabrizio a, L. Armelao a,b a
Institute of Condensed Matter Chemistry and Technologies for Energy (ICMATE), National Research Council (CNR), Corso Stati Uniti 4, 35127 Padova, Italy b University of Padova, Dept. of Chemical Sciences, Via Marzolo 1, 35131, Padova, Italy
article info
abstract
Article history:
The development of hydrogen purification membranes that meet market demands such as
Received 16 January 2018
high purity, dynamic hydrogen production even at small scale, and reduced costs is still an
Received in revised form
open question. With this view, the present study aims at developing, for the first time, a
9 March 2018
method based on high power impulse magnetron sputtering for the deposition of Pd77Ag23
Accepted 11 March 2018
(wt%) films onto porous alumina substrates to achieve composite membranes with high
Available online 1 April 2018
hydrogen permeability and stability. This technique allows the deposition of films also on complex geometries and can be easily scaled up, thus making this technology a potential
Keywords:
candidate for preparing high performing membranes. Membranes made by stable and
PdAg
porous alumina supports and metallic, dense and crystalline Pd77Ag23 layers, from 3.5 mm
Hydrogen
to 17 mm thick, have been prepared and tested. The membranes showed good hydrogen
Membrane
permeability values, showing flux values up to a maximum of 0.62 molH2 m2 s1 at 450 C
High power impulse magnetron
and DP of 300 kPa. The resistance to hydrogen embrittlement and the chemical inertness to
sputtering
syngas were also demonstrated.
Alumina
© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
Introduction Hydrogen is a valuable energy vector and a largely consumed reagent in many industrial processes as ammonia production, hydrocarbon hydrogenation, etc. [1]. Most of hydrogen is currently produced by steam reforming of natural gas [2], although hydrogen production from biomass gasification could accelerate the H2 utilization as a future sustainable fuel [3,4]. In both processes, an H2-rich gas mixture is produced, and pure H2 is obtained by some chemical processes carried
out in a number of reaction units followed by separation/purification (mostly by pressure swing adsorption) [5]. The large number of steps to produce and separate hydrogen represents efficiency and cost limits of traditional reactors, that can be circumvented by using integrated systems such as membrane reactors [5], where both reaction and separation are carried out in the same device. Membrane technology is nowadays increasingly considered, thanks to several advantages including low energy consumption, ability to carry out separation continuously, and simple scaling up [6]. However, membranes will achieve commercial targets
* Corresponding author. E-mail address: [email protected] (S. Barison). https://doi.org/10.1016/j.ijhydene.2018.03.065 0360-3199/© 2018 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
only when suitable hydrogen flux, cost, durability and tolerance to pollutants are optimized [7e9]. The most diffused membranes for hydrogen separation/ purification are nowadays based on palladium due to its very high hydrogen diffusion and permeability, although pure palladium suffers membrane poisoning and embrittlement. The shift, upon hydrogen uptake, from a-Pd(H) to b-Pd(H) phase, with a consequent volume increase, produces internal stresses, deformation and failure of the membrane. When compared to pure palladium, PdeAg alloys show important advantages, such as embrittlement resistance through the avoidance of a to b phase transition, robustness during thermal cycling, high permeability, and resistance to fouling [10e12]. Among these alloys, the Pd77eAg23 wt% composition has been largely investigated and tested due to its high permeance and resistance to hydrogen embrittlement [13]. Most commercial purification units contain conventional membranes consisting of relatively thick (>20 mm) Pd-alloy sheets or tubular membranes. However, to reduce the noble metal content, an effective strategy is needed to prepare dense membranes in the form of mm films deposited on suitable porous substrates. Various Pde and Pd alloys-based films have been deposited onto porous alumina, nickel or stainless steel [14], but they still lack of sufficient long-term stability and selectivity to H2. Electroless plating is commonly employed to deposit palladium-based films for this purpose. However, this technique is limited to few compositions/alloys [15]. Conversely, magnetron sputtering is effective in depositing alloys with variable compositions; it also allows the one stage deposition of multilayers in vacuum. In this work, we exploited a recent evolution of this technique, the High Power Impulse Magnetron Sputtering (HiPIMS). HiPIMS, introduced by Kouznetsov et al., in 1999 [16], is a successful technique for improving magnetron sputtering by pulsed power technology. Its main feature is the combination of sputtering from standard magnetrons and pulsed plasma discharges, with the aim of generating highly ionized plasma with large quantities of ionized sputtered material [17]. The high degree of ionization of the sputtered species is combined to a bias voltage applied to the substrate. This leads to the growth of smooth and dense films, to a good control on composition and microstructure, and allows improving film adhesion and uniformity also on substrates of complex shapes [18]. This feature could allow the deposition of films also on tubular membranes or other geometries useful for commercial plants. In this work, the most investigated composition, Pd 77 wt% e Ag 23 wt%, was chosen to evaluate HiPIMS influence on the preparation of membranes based on a Pd alloy widely studied in literature. Films with thickness ranging from 3 to 17 mm were deposited by a combined HiPIMS/Direct Current (DC) magnetron sputtering process onto porous alumina. Alumina has been chosen as substrate being mechanically and chemically stable in operating conditions. Alumina substrates present also the advantage of preventing interdiffusion phenomena at high temperatures, typical of steel substrates, an issue that can reduce hydrogen permeation. A procedure was set up to achieve substrates having a fine surface porosity to allow the deposition of a dense metallic layer with no need of any interlayer, that would increase the processing time, costs and complexity.
7983
Morphology, composition and structure of membranes were fully characterized. Hydrogen permeability was measured between 300 C and 450 C and at pressure difference up to 300 kPa. The resistance to hydrogen embrittlement and the chemical inertness to syngas were also investigated.
Experimental Materials and methods Alumina-based porous supports have been prepared by mixing 35 vol% of a-Al2O3 (Alfa Aesar, 99.9%) with 65 vol% of polymethyl methacrylate powders (PMMA) as pore former (Soken Chemical & Engineering, average size 1.5 mm). A wet ball milling process (absolute ethanol, Sigma Aldrich ACS reagent 99.8%) in zirconia jars was carried out in a planetary mill (Fritsch Pulverisette 7) at 350 RPM for a total of 2 h for each mixture. As obtained mixtures were uniaxially pressed (Nannetti Mignon SS/EA), in a 2.5 cm diameter mold by a 140 MPa load. The disks were then sintered at 1500 C in a high temperature furnace (Nabertherm HT 04/17), with an isotherm step of 1 h at the burning temperature of PMMA (386 C) and a very slow heating rate above 1000 C (30 C/h), when the shrinkage is maximum, to avoid pellet bending. The sintered disks were then polished and cleaned in an ultrasonic bath. The porosity was estimated by measuring the geometrical density and comparing it with the theoretical value of 3.98 g/ cm3 for alumina density. The membranes were deposited by a combination of HiPIMS and DC magnetron sputtering techniques. The high vacuum chamber was preliminary evacuated to a base pressure 1 104 Pa, while depositions were conducted in Argon atmosphere (Ar, 99.999% purity) at 1 Pa. During each sputtering run, the substrates were rotated at 5 rpm to improve homogeneity. The alloy was deposited by properly regulating the power on palladium and silver targets to achieve the Pd 77 wt % e Ag 23 wt% composition. By virtue of a deep preliminary set-up of the process, the palladium target (99.95% purity, 102 mm diameter) was driven by a HiPIMS power supply (Trueplasma HighPulse 4002, Hu¨ttinger Electronic, Germany) at an average power of 800 W (~10 Wcm2, pulse length 50 msec, frequency 500 Hz), while the silver target (99.99% purity, 102 mm diameter) was driven by a DC power supply (Trueplasma DC 4001, Hu¨ttinger Electronic, Germany) at 330 W (4 Wcm2). The substrate-target distance was set at 120 mm for Pd and 150 mm for Ag to achieve the selected composition. Moreover, a 100 V negative bias (Trueplasma Bias 3018, Hu¨ttinger Electronic, Germany) was applied to the substrates, which were heated up to 350 C. Temperature was maintained throughout the process by ad hoc heaters and thermocouple monitoring. Prior to perform Pd 77 wt% e Ag 23 wt% alloy deposition, a thin PdAg film was deposited at room temperature using the sputtering parameters listed above. The aim was to produce a conductive layer for a proper bias application during the next alloy deposition by using a properly designed sample holder. Only in this way it is possible to fully exploit the combined effect of the large number of target material ions produced during HiPIMS process with a bias voltage applied to the substrate.
7984
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Characterization Structural and microstructural information on the crystalline phases of the membranes were obtained by XRD and Rietveld refinement. The patterns were recorded at room temperature using a Philips PW 1830 diffractometer with Bragg-Brentano geometry, employing a Cu anode X-ray tube operated at 40 kV and 30 mA (angular range from 30 to 100 , 0.02 step, 6 s per step). Rietveld refinements were performed using MAUD software [19]. Surface and fractured surfaces of samples were observed by field emission scanning electron microscopy (FESEM) with a SIGMA Zeiss instrument (Carl Zeiss SMT Ltd, UK), operating in high vacuum conditions at an accelerating voltage of 20 kV and the composition was determined by Energy Dispersive Spectroscopy (EDS, Oxford X-MAX, UK). The surface roughness of alumina substrates was measured by a mechanical profiler (KLA Tencor P10, USA). Membrane permeability measurements were carried out by means of a custom-built stainless steel test station. A scheme of the permeation test apparatus is reported elsewhere [20]. Membranes were clamped and sealed in a stainless steel module using graphite gaskets. The module consists of two parts, feed side and permeate side, connected by a channel of about 1 cm in diameter, closed by the membrane to be tested, and placed in a furnace (Nabertherm N11/HR). The membrane housing temperature was monitored by a K-type thermocouple inserted directly in the module test. The gas flows at feed and permeate sides were set by independent mass flow controllers (1179A, 1179B and 647C, MKS). The feed side pressure was controlled by a Baratron pressure transducer (722B, MKS). Nitrogen (99.999% purity) was used to test membrane selectivity, while high purity hydrogen was produced by an electrolyzer (Perkin Elmer PGX Plus H2 160). During permeability tests, the feed side pressure was varied among absolute pressures of 110e400 kPa, while the permeate side was set at atmospheric pressure by a sweep gas flow. Pressure, flows and temperature were controlled and monitored by a Labview interface. Prior to each permeation test, the hydrogen flux was stabilized and monitored for few hours to ascertain or achieve a stable flux. After reaching stationary conditions, the hydrogen permeation was measured under typical operating conditions for these membranes, that is at 4 temperatures, from 300 C to 450 C, and at 16 DP values, from 300 kPa to 10 kPa (descending). For each
Fig. 2 e SEM backscattered electron image of the PdAg/ Al2O3 membrane with a 8.7 mm film thickness.
pressure point, the value was recorded after reaching a stable flux and the permeance at different pressures was tested twice for each temperature. The measures were repeated for membranes deposited in the same conditions to verify reproducibility. The selectivity was evaluated by the ratio between hydrogen and nitrogen permeances. Embrittlement tests were performed by fixing a DP value and by measuring the hydrogen flux as a function of temperature progressive reduction. Stability tests in commercial syngas (CO2 15 mol%, CO 15 mol%, H2 10 mol%, CH4 3.025 mol% and N2) were performed by exposing the membrane for various cycles to DP ¼ 200 kPa of syngas at 400 C, and by both monitoring the permeate flux by GC chromatography (Agilent Micro-GC 490) during syngas exposure and measuring the hydrogen flux after each cycle.
Results Mechanically and chemically stable porous substrates were prepared with a well-engineered porosity to reduce pressure drops in the substrate and a surface porosity sufficiently fine to deposit dense films, thus avoiding the need for interlayers [21]. The criteria for the selection of the pore former included decomposition with minimal residues, a resulting fine and
Fig. 1 e Surface SEM micrographs at different magnifications of a Pd77Ag23/Al2O3 membrane.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
well interconnected porosity and good mechanical stability. PMMA was chosen as pore former, with a mean powder size of 1.5 mm. Planar and porous substrates were obtained, with ~2 cm diameter (20e25% shrinkage), thickness between 500 and 1200 mm, porosity >35%, homogenous pore size distribution and a mean surface pore size of about 500e600 nm (estimated by SEM micrographs and by nitrogen and helium permeation measurements in substrates [20]). The hydrogen flux in porous substrates was measured and a minimal difference in the permeance was detected by changing their thickness, and this contribution was taken into account in elaborating data of permeation [20]. A preliminary deposition of Pd77 wt%-Ag23 wt% (4 min) was performed at room temperature prior to the deposition of the membrane, with the manifold scope of evaluating the
7985
influence of this film on the coating adhesion and creating a suitable conductive layer for the following application of bias during the membrane deposition. This layer resulted with columnar structure and about 1.5 mm thick. Once this layer was deposited, the sample was mounted in a specially designed sample holder that allowed the electrical contact with the conductive surface in order to have an effective bias application during the following alloy deposition. The HiPIMS deposition of the alloy was performed at 350 C. Various deposition durations were investigated for the alloy (from 30 to 180 min), leading to films ranging from about 3.5 to 17 mm (with a mean deposition rate of about 93 nm/min). Surface SEM micrographs of a membrane with a film thickness of about 8.7 mm (Fig. 1) show a very homogeneous, compact, dense and crystalline layer.
Fig. 3 e XRD patterns of the membrane a) as-deposited and b) after thermal treatment at 450 C for 20e24 h; in the figure * refers to Pd and D to Ag.
7986
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 2 shows a backscattered electron image of the cross sectional view of a membrane with a total thickness of about 8.7 mm. The micrograph highlights the PdAg porous interlayer deposited at room temperature and shows a dense and compact PdAg layer on top. EDS analyses were performed on surfaces and cross sections of all samples. The composition slightly varied among all samples and the mean value was Pd 78 wt% and Ag 22 wt%, with a standard deviation within all samples of ±2%, that is very close to the sought value. An example of EDS spectrum is reported in Figure S1 of Supplementary Information. XRD analysis of as-deposited membranes (see Fig. 3a) revealed that in as-prepared films the alloy was not formed, but crystalline Pd (cubic Fm-3m, cell parameter of 3.876 A, mean crystallite size of z140 nm) and Ag (cubic Fm-3m, cell parameter of 3.952 A, mean crystallite size of z123 nm) peaks
were still clearly distinguishable. Therefore, prior to any permeation test the membranes were conditioned for 20e24 h at 450 C in nitrogen to achieve the crystalline PdAg alloy (cubic Fm-3m, cell parameter of 3.916 A, mean crystallite size of z77 nm), as evidenced in Fig. 3b.
Hydrogen fluxes The hydrogen fluxes were measured between 300 and 400 C for all membranes with film thickness from 3.5 to 17 mm, in the 10e300 kPa ranges. Fig. 4 reports, as an example, the H2 fluxes measured at various temperatures as a function of the pressure difference between feed and permeate side for membranes with different PdAg film thickness. The flux values are higher for thinner films (up to 0.62 molH2 m2 s1 at DP ¼ 300 kPa and 450 C for the 3.5 mm film as indicated in
Fig. 4 e H2 flux measured between 300 and 450 C and as a function of the pressure difference between feed and permeate sides in membranes with PdAg film a) 3.5 mm or b) 8.7 mm thick.
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 4a), and they decrease at increasing thickness (e.g. up to 0.28 molH2 m2 s1 at DP ¼ 300 kPa and 450 C for the 8.7 mm film reported in Fig. 4b). These values perfectly agree with literature data of PdAg films deposited with other methods (typically electroless plating). They are similar to flux values recorded for films with similar or higher thickness [10,22] and higher than values recorded for thicker films (25 mm [23]), while in some cases are lower than fluxes recorded for thinner films (1.2 mm [24]), thus demonstrating the effectiveness of this technique to deposit dense films with proper hydrogen flux values. To compare membranes with different PdAg layer thickness, the fluxes measured at DP ¼ 300 kPa as a function of temperature for samples with various thicknesses are reported in Fig. 5. A clear flux reduction, from a maximum of 0.62 molH2 m2 s1 for the 3.6 mm thick film to a minimum of 0.12 molH2 m2 s1 for the 17 mm thick film can be observed. By analyzing the trend of flux as a function of temperature, a general flux increase with temperature was observed for all samples, although in some membranes the flux decreased at 450 C. By extrapolating the data at various temperatures from permeances, apparent activation energies were found to decrease with film thickness, going from 7.3 kJ mol1 H2 for the thinnest PdAg film to 4.9 kJ mol1 H2 for the thickest sample. These values are analogous to those of fully formed and crystalline PdAg alloy/ceramic composite membranes with the same composition (see Ref. [25] and references therein). Ward and Dao deeply modeled the hydrogen permeation in Pd membranes [26] and found that for thin membranes (1 mm) at temperatures higher than 300 C the flux trend with temperature shows a diffusion-limited flow slightly influenced by the desorption-limited process, and they identified high activation energies for absorption or desorption processes (>42 kJ mol1 H ). However, considering a mass transfer resistance at low pressure side, typical for thin films deposited on porous supports, they found a significant reduction of fluxes and a trend with temperature that could explain the flux reduction observed at 450 C. Considering the thickness of membranes and the values of activation energies, we can hypothesize a permeation process mainly controlled by a
Fig. 5 e H2 flux values measured at DP ¼ 300 kPa vs T for samples with different thicknesses (indicated in the legend in mm).
7987
combination of hydrogen diffusion in bulk and mass transfer resistance in the alumina support, besides further possible transport phenomena at grain boundaries. Considering the Fick's first law to describe the atomic hydrogen diffusion through a homogeneous metal phase as a function of concentration gradient and diffusion coefficient, and the Sievert's law that may be used under certain conditions to describe the relationship between the concentration and pressure, Eq. (1) was derived to describe how the hydrogen flux (J) varies with temperature and pressure [8]. f qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi f0 eRT qffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffi Pfeed Pperm ¼ Pfeed Pperm L L Ea
j¼
(1)
where F is the permeability, that is defined as the rate of gas permeation per unit area, per unit driving force (pressure in this case), per unit membrane thickness, Ea is the activation energy for the whole process, F0 is the pre-exponential factor related to the permeability of the metal, R the gas constant, L the membrane thickness, Pfeed and Pperm are the hydrogen
Fig. 6 e H2 flux values at DP ¼ 300 kPa and permeability values at 400 C for all samples.
Fig. 7 e H2 flux at 400 C and DP ¼ 300 kPa before and after various cycles of exposure to syngas (DP ¼ 200 kPa) in a 17 mm thick sample. The membrane was exposed to N2 before and after exposure to syngas for few min. The red bars indicate each syngas exposure duration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
7988
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
Fig. 8 e a) Surface SEM micrograph after permeability tests and syngas exposure of a 17 mm thick sample (the inset shows the morphology at the same magnification of the same sample before testing); b) section at high magnification of the subsurface region of the same sample (the inset shows total section of the sample, similar to as prepared sample).
pressure at the feed and at the permeate sides, respectively. However, this equation is valid for self-sustained thick membranes (commonly >10 mm). A more general formula has been typically used especially to estimate permeability or permeance (F/L) of thin palladium-based membranes, as in Eq. (2) [8]. j¼
f n P Pnperm L feed
(2)
Although with some limitations, this formula represents a general approach to take into account some phenomena that can lead to a best fit of flux data vs (Pnfeed e Pnperm) with an exponent n higher than 0.5. Among them, there are surface resistance to H2 absorption and desorption, mass transfer resistance in porous substrates, hydrogen diffusion as molecular H2 along grain boundaries and/or Knudsen or viscous flow through pores and film defects and leaks [8,27,28]. In thin Pd-based membranes (thickness from 1 to 7 mm), n was shown to approach 1 and this was typically attributed to surface phenomena as the rate-limiting step [24]. For all our samples the best fit of data was identified for n ¼ 0.8 (see an example of flux vs (Pnfeed e Pnperm) for 3 values of n in Figure S2 of Supplementary Information). This deviation from Sievert's law could be due to a combination of interface and grain boundaries controlled phenomena and mass transfer resistance in the porous supports at the low-pressure side. However, to compare these data with literature, the permeability and the permeance (F/L) were calculated by Eq. (2) with n ¼ 1. The permeances thus calculated at 400 C ranged between 5.6 107 for the thickest sample (17 mm) to 2.3 106 mol m2 s1 Pa1 for the thinnest sample (3.5 mm). The last value is similar or slightly smaller than permeance values identified in Pd and PdAg membranes on alumina having thickness around 2 mm (see Ref. [24] and references therein). Fig. 6 reports the flux at DP ¼ 300 kPa and the permeability values at 400 C for all samples. The permeability was in the 7 to 12 1012 mol m1 s1 Pa1 range for all samples. The selectivity values, calculated as the ratio between hydrogen and nitrogen permeances, ranged from about 30 for thinnest samples to about 580 for samples having thickness 8.7 mm (corresponding to nitrogen fluxes 4 104 mol m2 s1 at DP ¼ 300 kPa and 400 C, which is probably the sealing limit of our apparatus, as already observed in Ref. [20]). By analyzing data
with n ¼ 0.5, the permeability values at 400 C where between 6.6 and 8.4 109 mol m1 s1 Pa0.5, that is in agreement with the literature [22,29].
Membrane stability In this study, each membrane has been tested for at least 2 weeks, at temperatures ranging from 300 to 450 C and up to DP ¼ 300 kPa, showing flux stability over the time and no failure due to embrittlement. By decreasing the temperature down to room temperature, no abrupt increase of flux due to film embrittlement was ascertained, showing no failure of the film. Only in few cases, a small change in slope at temperatures below 200 C (see Fig. S3 as an example) could indicate the appearance of very small holes in the coating. SEM analyses after permeability tests and embrittlement tests showed very similar morphology to that of as-prepared samples. The stability of these membranes in commercial syngas was also tested. In particular, the hydrogen flux was measured at 400 C and DP ¼ 300 kPa before and after various cycles of exposure to syngas (Fig. 7). The hydrogen flux was almost constant and had no significant changes even after a total exposure of 13 h to syngas, suggesting the stability of the membrane. The chromatograms recorded during syngas tests (Fig. S4) detected only a peak due to hydrogen permeation and no other peaks within the gas chromatography uncertainty. The morphological characterization of membranes after syngas tests showed surface and fractured morphologies very similar to as-prepared samples (Fig. 8), with only a nanoporosity identified in fractured sections, but limited to outer membrane layers (100e200 nm close to the surface) (Fig. 8b). XRD analyses after embrittlement and stability tests did not evidence any difference with respect to diffractograms recorded before these tests (Fig. 3b).
Conclusions The deposition of PdAg films onto porous alumina substrates by a combined HiPIMS/DC technique was successfully performed. Membranes made by stable and porous alumina supports and metallic, dense and crystalline Pd77Ag23 layers,
i n t e r n a t i o n a l j o u r n a l o f h y d r o g e n e n e r g y 4 3 ( 2 0 1 8 ) 7 9 8 2 e7 9 8 9
from 3.5 mm to 17 mm thick, have been prepared and tested. Membranes showed good hydrogen permeability values, comparable or higher than literature values, showing flux values up to a maximum of 0.62 molH2 m2 s1 at 450 C and DP of 300 kPa. The resistance to hydrogen embrittlement and to syngas was also demonstrated. This technique allows the deposition of films on complex geometries, such as tubes, and can be easily scaled up, thus making this technology suitable for preparing high performing hydrogen selective membranes.
Acknowledgements The authors are grateful to Dr. Rosalba Gerbasi (CNR ICMATE) for XRD analyses. This work has been founded by Italian Industria 2015 project “Produzione di energia rinnovabile con il minimo impatto da un mix di biomasse e rifiuti speciali non pericolosi attraverso processi innovativi”.
Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.ijhydene.2018.03.065.
references
[1] Derwent R, Simmonds P, O'Doherty S, Manning A, Collins W, Stevenson D. Global environmental impacts of the hydrogen economy. Int J Nucl Hydrogen Prod Appl 2006;1:57e67. [2] Dincer I, Acar C. Review and evaluation of hydrogen production methods for better sustainability. Int J Hydrogen Energy 2015;40:11094e111. [3] Hosseini SE, Abdul Wahid M, Jamil MM, Azli AAM, Misbah MF. A review on biomass-based hydrogen production for renewable energy supply. Int J Energy Res 2015;39:1597e615. [4] Demirbas A. Biohydrogen generation from organic waste. Energy Sources Part A 2008;30(5):475e82. [5] Gallucci F, Fernandez E, Corengia P, Annaland M. Recent advances on membranes and membrane reactors for hydrogen production. Chem Eng Sci 2013;92:40e66. [6] Al-Mufachi NA, Rees NV, Steinberger-Wilkens R. Hydrogen selective membranes: a review of palladium-based dense metal membranes. Renew Sustain Energy Rev 2015;47:540e51. [7] Sjardin M, Damen KJ, Faaij APC. Techno-economic prospects of small-scale membrane reactors in a future hydrogenfuelled transportation sector. Energy 2006;31:2523e55. [8] Phair JW, Donelson R. Developments and design of novel (non-palladium-based) metal membranes for hydrogen separation. Ind Eng Chem Res 2006;45(16):5657e74. [9] Dolan MD. Non-Pd BCC alloy membranes for industrial hydrogen separation. J Membr Sci 2010;362(1e2):12e28. [10] Pizzi D, Worth R, Baschetti MG, Sarti GC, Noda K. Hydrogen permeability of 2.5mm palladiumesilver membranes deposited on ceramic supports. J Membr Sci 2008;325:446e53.
7989
[11] Okazaki J, Tanaka DAP, Tanco MAL, Wakui Y, Mizukami F, Suzuki TM. Hydrogen permeability study of the thin Pd-Ag alloy membranes in the temperature range across the alphabeta phase transition. J Membr Sci 2006;282:370. [12] Cheng YS, Pena MA, Fierro JL, Hui DCW, Yeung KL. Performance of alumina, zeolite, palladium, PdeAg alloy membranes for hydrogen separation from Towngas mixture. J Membr Sci 2002;204:329. [13] Karakaya I, Thompson WT. The Ag Pd (Silver-Palladium) system. Bull Alloy Ph Diagr 1988;9(3):237e43. [14] Yun S, Oyama ST. Correlations in palladium membranes for hydrogen separation: a review. J Membr Sci 2011;375:28e45. [15] Den Exter MJ. Palladium membrane technology for hydrogen production, carbon capture and other applications, 3. Elsevier; 2014. p. 43e67. k K, Schneider JM, Helmersson U, [16] Kouznetsov V, Maca Petrov I. A novel pulsed magnetron sputter technique utilizing very high target power densities. Surf Coating Technol 1999;122:290e3. [17] Sarakinos K, Alami J, Konstantinidis S. High power pulsed magnetron sputtering: a review on scientific and engineering state of the art. Surf Coating Technol 2010;204:1661e84. [18] Lundin D, Sarakinos K. An introduction to thin film processing using high-power impulse magnetron sputtering. J Mater Res 2012;27(5):780e92. [19] Lutterotti L, Mattheis S, Wenk HR. MAUD: a friendly Java program for material analysis using diffraction, 21. Newsletter of the CPD; 1999. p. 14e5. [20] Fasolin S, Barison S, Boldrini S, Ferrario A, Romano M, Montagner F, et al. Hydrogen separation by thin vanadiumbased multi-layered membranes. Int J Hydrogen Energy 2018;43(6):3235e43. [21] Zheng L, Li H, Xu H. “Defect-free” interlayer with a smooth surface and controlled pore-mouth size for thin and thermally stable Pd composite membranes. Int J Hydrogen Energy 2016;41:1002e9. [22] Wang L, Yoshiie R, Uemiya S. Fabrication of novel PdeAgeRu/Al2O3 ternary alloy composite membrane with remarkably enhanced H2 permeability. J Membr Sci 2007;306:1e7. [23] Wieland S, Melin T, Lamm A. Membrane reactors for hydrogen production. Chem Eng Sci 2002;57:1571e6. [24] Maneerung T, Hidajat K, Kawi S. Ultra-thin (<1 mm) internally-coated PdeAg alloy hollow fiber membrane with superior thermal stability and durability for high temperature H2 separation. J Membr Sci 2014;452:127e42. [25] Zeng G, Goldbach A, Shi L, Xu H. Compensation effect in H2 permeation kinetics of PdAg membranes. J Phys Chem C 2012;116(34):18101e7. [26] Ward TL, Dao T. Model of hydrogen permeation behavior in palladium membranes. J Membr Sci 1999;153:211e31. [27] Morreale BD, Ciocco MV, Enick RM, Morsi BI, Howard BH, Cugini AV, et al. The permeability of hydrogen in bulk palladium at elevated temperatures and pressures. J Membr Sci 2003;212(1e2):87e97. [28] Guazzone F, Ma YH. Leak growth mechanism in composite Pd membranes prepared by the electroless deposition method. AIChE J 2008;54:487e92. [29] Peters TA, Kaleta T, Stange M, Bredesen R. Hydrogen transport through a selection of thin Pd-alloy membranes: membrane stability, H2S inhibition, and flux recovery in hydrogen and simulated WGS mixtures. Catal Today 2012;193:8e19.