Journal of Power Sources 156 (2006) 224–231
Effects of sputtering parameters on the performance of electrodes
fabricated for proton exchange membrane fuel cells
Kuo-Lin Huang
a,∗, Yi-Chieh Lai
b, Cheng-Hsien Tsai
caDepartment of Environmental Engineering and Science, National Pingtung University of Science and Technology,
1 Hseuh Fu Road, Neipu, Pingtung 91201, Taiwan
bDepartment of Environmental Engineering, National Cheng Kung University, Tainan 70101, Taiwan cDepartment of Chemical and Material Engineering, National Kaohsiung University of Applied Sciences,
Kaohsiung 807, Taiwan
Received 19 April 2005; received in revised form 24 May 2005; accepted 27 May 2005 Available online 22 July 2005
Abstract
One way to alleviate the emission of air pollutants and CO2due to burning fossil fuels is the use of fuel cells. Sputter deposition techniques
are good candidates for the fabrication of electrodes used for proton exchange membrane fuel cells (PEMFCs). Input power and sputtering-gas pressure are two important parameters in a sputtering process. However, little is known about the effects of these sputtering parameters on the performance of PEMFC electrodes. Therefore, this study applied a radio frequency (RF) magnetron sputter deposition process to prepare PEMFC electrodes and investigated the effects of RF power and sputtering-gas pressure in electrode fabrication on electrode/cell performance. At a Pt loading of 0.1 mg cm−2, the electrode fabricated at 100 W, 10−3Torr was found to exhibit the best performance mainly due to its lowest kinetic (activation) resistance (dominating the cell performance) in comparison to those fabricated by 50 and 150 W at 10−3Torr, as well as by 10−4and 10−2Torr at 100 W. In the tested ranges, the control of sputtering-gas pressure seems to be more critical than that of RF power for the activation loss. In addition to electrochemically active surface area, electrode microstructure should also be responsible for electrode/cell polarization, particularly the activation polarization.
© 2005 Elsevier B.V. All rights reserved.
Keywords: Proton exchange membrane fuel cells; RF sputtering; Pt deposition; Electrode fabrication
1. Introduction
The use of fuel cells may alleviate the emission of air pollutants and CO2 resulted from utilizing fossil fuels[1]. Presently, the greatest research interest throughout the world has focused on PEMFCs and solid oxide cell stacks[2]; fur-thermore, the PEMFCs have reached the stage of being in the forefront among the different types of fuel cells[3]. The most studied fuels for the PEMFCs are hydrogen and methanol; however, the former, without any methanol crossover con-cern, has a higher power density output than the latter[4,5]. A
∗Corresponding author. Tel.: +886 8 770 3202x7092;
fax: +886 8 774 0256.
E-mail address: [email protected] (K.-L. Huang).
single H2–O2/air PEMFC is chiefly comprised of three types of components: a membrane-electrode assembly (MEA), two bipolar plates (having flow fields or separators), and two seals; in its simplest form, the MEA consists of a membrane, two dispersed catalyst layers, and two gas diffusion layers (GDL)[6].
The MEA is the heart of a PEMFC [3] and there are two modes of MEA fabrication: (I) application of the cat-alyst layer to the GDL followed by membrane addition or (II) application of the catalyst layer to the membrane fol-lowed by GDL addition[6]. MEAs rely on different catalyst preparation methods such as impregnation reduction, spread-ing, sprayspread-ing, catalyst powder deposition, catalyst decalspread-ing, painting, electro-deposition, and sputter deposition in combi-nation with hot-pressing procedures to bond disparate carbon 0378-7753/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
K.-L. Huang et al. / Journal of Power Sources 156 (2006) 224–231 225
cloth electrodes to an electrolytic membrane[6–10]. (The last two methods mentioned belong to the Modes I or II whereas the others are the Mode I for MEA fabrication.) Improved processes use thinner active layer thickness (i.e., from 100 to 25m) and smaller carbon-supported nanometer size Pt par-ticles (<10 nm, i.e., 2.5 nm average for E-TEK) to enhance the Pt utilization and reduce the Pt amount in gas diffusion electrodes (GDEs)[9,11]. Therefore, the Pt loading in the conventional GDEs has been lowered from early 4 to recent 0.4 mg cm−2[8,12], very crucial for H2–O2/air PEMFCs to be commercially realizable[6,8,9,13,14].
Among the electrode fabrication methods aforemen-tioned, electro-deposition [13–16] and sputter deposition [8,12,17]techniques can be used to localize Pt within the (membrane carbon interface) regions responsible for elec-trochemical reaction in MEAs; the purpose of this design is to increase the electrode Pt utilization by diminishing inactive catalyst sites when compared to most of the other methods with Pt more uniformly dispersed in the whole GDE. The plasma sputter deposition technique has a high potential for MEA fabrication to reduce cell costs by achiev-ing ultra-low levels of Pt catalyst loadachiev-ing[8]. For example (Mode I type mentioned above for MEA fabrication), it has been found that the oxygen electrode with an ultra-low Pt loading (0.1 mg cm−2) electrocatalyst layer (1m thick) sputter-deposited on uncatalyzed GDE exhibits comparable performance to that of a standard E-TEK electrode[17]; addi-tionally, the sputtering of a 50 nm Pt film (0.05 mg cm−2) on the front surface of a fuel cell electrode may dimin-ish overpotentials to improve the electrode/cell performance [11,18]. Recently, using the Mode II methods for MEA fabri-cation, researchers have found that the electrode fabricated by plasma-sputtering Pt (0.043 mg cm−2) directly on the surface of Nafion electrolyte displays ten times higher efficiency than those prepared by conventional methods[12]; moreover, the performance of a sputter-deposited Nafion membrane with a Pt loading level of 0.04 mg cm−2is comparable to a commer-cial MEA with a Pt loading of 0.4 mg cm−2[8]. This work did not specifically focus on the electrode Pt localization and loading. Instead, we are concerned about the electrode/cell performance (mainly regulated by the gas/proton/electron (triple access) continuity) achieved at various sputtering con-ditions[8,12].
However, little is known about the effects of RF power and sputtering-gas pressure on the performance of MEAs/electrodes fabricated by sputtering processes. Both RF power and sputtering-gas pressure influence the deposi-tion rates on substrates and the microstructures (microcrys-tallizationon) of films sputtered because the plasma density and electron kinetic energy vary with the sputter conditions [19–23]. The microstructures of electrode catalysts prepared by sputtering for fuel cells may affect the electrode/cell per-formance. In this study, therefore, an RF magnetron sput-ter process was used to sputsput-ter Pt in GDEs to prepare the electrodes in MEAs (Mode I type) and they were tested using cell polarization, cyclic voltammetry, ac impedance,
and surface analysis techniques to explore how the exper-imental parameters including RF power and sputtering-gas pressure employed in electrode fabrication influenced the per-formances of electrodes/MEAs.
2. Experimental
2.1. Preparation of electrodes and MEAs
An RF sputtering system was employed to deposit Pt cat-alyst onto the carbon particles (ranging ∼30–90 nm with an average size of 56 nm; also see Section 3.2) supported by carbon cloth (uncatalyzed Gore gas diffusion electrodes (CARBEL CL, thickness = 0.4 mm), Gore, USA) with a pro-jected active area of 5 cm2. The base vacuum pressure was set at 4× 10−5Torr in each sputtering operation. The Argon flow rates of 15, 30, and 65 sccm were employed to achieve the pressures of 10−4, 10−3, and 10−2Torr, respectively. Three input RF powers (50, 100, and 150 W) were tested.
A␣-step surface profiler (Alpha-Step, Tencor 200) was used to measure the thickness of Pt films sputtered onto a planar Si substrate and the Pt loading of a substrate (Pt amount on Pt film projected area in mg cm−2) was calculated by Pt film thickness × Pt density. The measured Pt film thickness divided by sputtering time gave Pt deposition rate (in nm min−1). For convenience, we controlled sputtering time to achieve different Pt loadings on substrates at various operational conditions. By this way, the loadings of Pt sputtered onto uncatalyzed Gore gas diffusion electrodes were estimated. For each test, the anode and cathode Pt loadings were 0.4 and 0.1 mg cm−2, respectively. The polymer membrane electrolyte was Nafion 117 (Dupont, USA). Prior to use, each membrane was first boiled in 3% hydrogen peroxide, then washed with de-ionized water, and finally boiled in 1 M sulfuric acid[8]. The catalyzed anode and cathode and the pretreated membrane were hot-pressed at 120◦C and 5000 lb for 1 min to form the MEA. Fig. 1 schematically depicts the structures of the MEA (the Nafion membrane sandwiched/hot-pressed with the anode and cathode) and catalyzed electrode (with Pt sputtered onto carbon particles supported by carbon cloth) (also see Section3.2).
2.2. Cell polarization measurements and electrochemical analysis
Screws and nuts were used to assemble a cell with the prepared MEA symmetrically sandwiched by two Teflon gas-kets, two carbon (graphite) blocks with gas flow channels, and two copper current collectors. The single cell was then installed on a standard fuel cell test station. After the activa-tion of each MEA in cell, the cell polarizaactiva-tion was conducted at 65◦C (80 and 70◦C for the anode and cathode humidifiers, respectively) using the test station. The airflow rates were controlled at 1.5 and 2 times of the stochiometric
require-230 K.-L. Huang et al. / Journal of Power Sources 156 (2006) 224–231
less, all the five cases exhibited a slope increase around 0.6 V in the non-kinetically controlled regions of cell polarization curves, suggesting the increasing influence from mass trans-fer, possibly due to the increasing water accumulation (but not yet the flooding) within the three-phase zone. This phe-nomenon is also associated with the dense Pt films sputtered on carbon particles in electrode fabrication in this study.
Although the electrode Pt loadings were controlled to be the same for the electrodes prepared at three different pressures, the electrodes made at 10−3Torr exhibited more available sites for hydrogen desorption in CV (Fig. 3) and thus had an evidently greater EAS area than those made at 10−2 and 10−4Torr (Table 2), partially responsible for the better electrode/cell performance for the 10−3Torr case than for the 10−2and 10−4Torr cases. The finding implies that the inter-connection (contact) of carbon-supported Pt was better for the 10−3Torr case than for the other two cases, and this is possi-bly associated with the different Pt deposition rates controlled by gas pressure. As stated in Section3.1, the sputtering sys-tem operating at 10−3Torr displayed a higher deposition rate of Pt and needed a shorter sputtering time than those operating at 10−2and 10−4Torr to obtain the same Pt loading. Possi-bly, the operation using a lower deposition rate and a longer sputtering time enhances the formation of a denser Pt film (or more concentrated configuration of Pt grains) on carbon particles to decrease the EAS area. It is also possible that the operation at the lower sputtering-gas pressure (the 10−4Torr case) had a better collimation to conduct more deposition of Pt in the micropores of carbon particles whereas a higher collision possibility for the Pt atoms bombarded away from a target at the higher sputtering-gas pressure (the 10−2Torr case) might lead to the aggregation of Pt grains to decrease EAS area. The higher EAS area or Pt roughness is also asso-ciated with the lower Rpof the 10−3Torr case. Note that the EAS area only accounted for partial activation loss which was also associated with the surface concentrations of the species involved in the electrode reactions, as stated in Section3.2. The magnitudes of EAS area and reactant concentration on active sites should be related to the microstructure of Pt grains sputtered on carbon particles.
Accordingly, to obtain the best electrode/cell performance at the Pt loading of 0.1 mg cm−2, the optimal RF power and the sputtering-gas pressure were 100 W and 10−3Torr, respectively, in this study. The cathode prepared at the optimal RF power and sputtering-gas pressure exhibited the low-est kinetic resistance. The activation polarization (varying with electrodes) dominated the cell performance because the polarization trends in ohmic-controlled (or concentration-controlled) regions were similar for all the electrodes tested. The results are consistent with that activation overvoltage is the most important irreversibility and cause of voltage drop (mainly occurring at the cathode) in low and medium temperature (hydrogen–oxygen/air) fuel cells[32]. Further-more, it seems that the control of sputtering-gas pressure is more important than that of RF power in the tested ranges of this work to lower the activation loss. However, the sputter
deposition of Pt onto porous carbon particles is complicated, and the electrode microstructure, varying with RF power and sputtering-gas pressure, is crucial to improve the elec-trode/cell performance.
4. Conclusions
The effects of RF power and sputtering-gas pressure in electrode fabrication on MEA/cell performance were inves-tigated in this study. It was found that the deposition rate of Pt increased with increasing RF power, to reach 0.047, 0.084, and 0.107 mg cm−2min−1, at 50, 100, and 150 W, respec-tively at 10−3Torr. At 100 W, the deposition rates of Pt were 0.060 and 0.070 mg cm−2min−1for the operations at 10−4 and 10−2Torr, respectively.
At a Pt loading of 0.1 mg cm−2and 10−3Torr, the perfor-mance of MEAs with the Pt/C cathode prepared at 50 and 150 W were similar but lower than that of the MEA prepared at 100 W, which is mainly attributed to the higher kinetic resistance in oxygen reduction for the former electrodes than for the latter electrode although an opposite trend of their EAS areas was observed for the three electrodes. However, at the same Pt loading operating at the 100 W, in comparison to the electrodes prepared at 10−4and 10−2Torr, the electrode pre-pared at 10−3Torr exhibited a higher EAS area and a lower kinetic resistance responsible for a better electrode/cell per-formance. It seems that the control of sputtering-gas pressure is more important than that of RF power in the tested ranges to lower the activation loss. All the electrodes having dense Pt films sputtered on carbon particles exhibited a similar trend in slope increase around 0.6 V in cell polarization curves, possibly due to the increasing water accumulation at active sites.
We believe that the microstructures of the electrodes fabri-cated at different RF powers and gas pressures are important for the improvement of MEA/cell performance and need to be further studied.
Acknowledgments
The authors would like to thank Professor Wen-Jhy Lee at National Cheng Kung University and Dr. Yi-Yie Yan at Industrial Technology Research Institute for their helps, and the National Science Council of the Republic of China, Tai-wan for financially supporting this research under Contract No. NSC-92-2211-E-309-002.
References
[1] R.A. Ristinen, J.J. Kraushaar, Energy and the Environment, Wiley, New York, 1999, p. 293.
[2] A.B. Stambouli, E. Traversa, Renew. Sust. Energ. Rev. 6 (2002) 297. [3] P. Costamagna, S. Srinivasan, J. Power Sources 102 (2001) 242.
K.-L. Huang et al. / Journal of Power Sources 156 (2006) 224–231 231
[4] C.K. Witham, W. Chun, T.I. Valdez, S.R. Narayanan, Electrochem. Solid-State Lett. 3 (2000) 497.
[5] W.C. Choi, J.D. Kim, S.I. Woo, J. Power Sources 96 (2001) 411. [6] V. Mehta, J.S. Cooper, J. Power Sources 114 (2003) 32. [7] M.S. Wilson, S. Gottesfeld, J. Appl. Electrochem. 22 (1992) 1. [8] R. O’Hayre, S.-J. Lee, S.-W. Cha, F.B. Prinz, J. Power Sources 109
(2002) 483.
[9] S. Gamburzev, A.J. Appley, J. Power Sources 107 (2002) 5. [10] G. Bender, T.A. Zawodzinski, A.P. Saab, J. Power Sources 12 (2003)
114.
[11] S. Mukerjee, S. Srinivasan, A.J. Appleby, Electrochim. Acta 38 (1993) 1661.
[12] S.Y. Cha, W.M. Lee, J. Electrochem. Soc. 146 (1999) 4055. [13] E.J. Taylor, E.B. Anderson, N.R.K. Vilambi, J. Electrochem. Soc.
139 (1992) 45.
[14] M.W. Verbrugge, J. Electrochem. Soc. 141 (1994) 46.
[15] K.H. Choi, H.S. Kim, T.H. Lee, J. Power Sources 75 (1998) 230. [16] H. Kim, B.N. Popov, Electrochem. Solid-State Lett. 7 (2004) 71. [17] S. Hirano, J. Kim, S. Srinivasan, Electrochim. Acta 42 (1997) 1587. [18] E.A. Ticianelli, J.G. Beery, S. Srinivasan, J. Appl. Electrochem. 21
(1991) 597.
[19] W.-T. Li, D.R. McKenzie, W.D. McFall, Q.-C. Zhang, Thin Solid Films 384 (2001) 46.
[20] A. Nagata, H. Okayama, Vacuum 66 (2002) 523. [21] H.W. Kim, N.H. Kim, Mater. Sci. Eng. B103 (2003) 297. [22] Z.H. Liu, N.M.D. Brown, Thin Solid Films 349 (1999) 78. [23] M. Jana, D. Das, Sol. Energ. Mater. Sol. Cells 79 (2003) 519. [24] T. Biegler, D.A.J. Rand, R. Woods, J. Electroanal. Chem. 29 (1971)
269.
[25] M. Kawamura, T. Mashima, Y. Abe, K. Sasaki, Thin Solid Films 377–378 (2000) 537.
[26] B. Chapman, Glow Discharge Processes: Sputtering and Plasma Etching, Wiley, New York, 1980, pp. 9–13.
[27] V.I. Shulga, Nucl. Instrum. Meth. Phys. Res. B 174 (2001) 423. [28] C.H. Hsu, C.C. Wan, J. Power Sources 115 (2003) 268.
[29] L. Giorgi, E. Antolini, A. Pozio, E. Passalacqua, Electrochim. Acta 43 (1998) 3675.
[30] E.A. Ticianelli, C.R. Derouin, A. Redonod, S. Srinivasan, J. Elec-trochem. Soc. 135 (1988) 2209.
[31] A. Damjanovic, V. Brusic, Electrochim. Acta 12 (1967) 615. [32] J. Larminie, A. Dicks, Fuel Cell Systems Explained, Wiley, England,
