- Email: [email protected]
Development of a niobium clad PEM fuel cell bipolar plate material
International Journal of Hydrogen Energy 32 (2007) 3724 – 3733 www.elsevier.com/locate/ijhydene
Development of a niobium clad PEM fuel cell bipolar plate material K. Scott Weil ∗ , Gordon Xia, Z. Gary Yang, Jin Yong Kim Pacific Northwest National Laboratory, P.O. Box 999, Richland, WA 99352, USA Available online 4 October 2006
Abstract Reported in this paper are results obtained on a niobium clad material that is being developed for use in polymer electrolyte membrane fuel cell (PEMFC) stacks. A series of materials evaluation tests were initially conducted on niobium coupons to determine if this material was suitable as an external cladding layer exposed to a prototypic PEMFC environment. Results from corrosion testing conducted in 80 ◦ C, 1 M H2 SO4 (with 2 ppm HF) display no measurable weight loss in the niobium specimens out to 2000 h of exposure. Interfacial contact resistance measurements conducted on as-received and post-exposed niobium indicate that it exhibits excellent surface conductivity ( 10 m cm2 ) under low clamping forces in both conditions. Polarization testing carried out under both prototypic anodic and cathodic PEMFC operating conditions suggest that the electrochemical behavior of niobium is comparable to that of platinum, with current densities of 2.7 × 10−5 and 6.3 × 10−9 A/cm2 at half cell potentials of −0.1 and 0.6 V versus a saturated calomel reference electrode, respectively, under prototypic anode and cathode environmental conditions. Subsequent contact resistance and polarization testing of niobium clad stainless steel coupons yielded results similar to those found in monolithic niobium testing. 䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: PEM fuel cell; Bipolar plate; Clad metal
1. Introduction There has been mounting public and political pressure in a number of industrialized nations, including the United States, to improve fuel utilization in power transportation applications and thereby reduce reliance on foreign oil supply and lessen pollution emissions. Coupled with continued growth in the worldwide demand for power, these factors have led to increasing interest in alternative methods of power generation, such as fuel cells, for various stationary and mobile applications, including off-power-grid residential power, auxiliary semi-truck power, electric automobiles, and portable personal power sources [1,2]. The primary feature of the fuel cell is that it directly converts the chemical energy of the incoming fuel into electrical energy via an electrochemical reaction. Because there is no intermediate thermal conversion step, and therefore no attendant Carnot cycle limitation, fuel cells provide a highly efficient means of energy conversion [3]. However, despite the significant technical progress that has been made in recent years toward ∗ Corresponding author. Tel.: +1 509 376 6796; fax: +1 509 375 2186.
E-mail address: [email protected] (K.S. Weil).
developing a commercially viable polymer electrolyte membrane fuel cell (PEMFC) system, the device currently finds use only niche applications [4]. Among the key reasons that PEMFC technology has not become more popular are the high cost of manufacture and the steady loss in power output during long-term, continuous operation. An additional factor that has hindered the acceptance of PEMFC systems in transportation markets is their current size and weight [5]. One of the most bulky components in the stack with respect to both weight and volume is the bipolar plate. In addition, it is one of the most expensive to manufacture. This component serves not only as the electrical junction between serially connected cells in the stack, but also performs several other key functions in the overall device, including [6]: • Distribute fuel and oxidant uniformly over the active areas of the cells. • Facilitate water management of the membrane to keep it humidified, yet mitigate flooding. • Act as an impermeable barrier between the fuel and oxidant streams (particularly H2 ) to maintain the hydrogen gradient across the membrane necessary for high power output.
0360-3199/$ - see front matter 䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2006.08.041
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
• Provide structural support for the stack. • Remove heat from the active areas of the cells. In the early development of PEMFCs, graphite was the material of choice for use in the bipolar plate because of its excellent corrosion resistance and surface conductivity in the low pH/hot aqueous environment (60–80 ◦ C) of the stack [2]. However, the high cost and low mechanical strength of high purity graphite and the expense associated with machining the individual plates necessitated the development of lower cost alternative bipolar plate materials for commercial applications. To date, carbon composites, various coated metals, and uncoated stainless steel and titanium have been considered [7–12]. Although carbonbased composites appear to be the leading candidates, there remain several challenges with these materials that are yet to be resolved, particularly for automotive and other mobile applications, including high manufacturing cost, insufficient mechanical strength, and poor barrier to hydrogen permeability [6]. While the use of metal in PEMFC bipolar plates appears to solve these issues, this class of materials is generally plagued by surface corrosion. The recent desire of stack designers to increase the operating temperature will only exacerbate this problem [13]. Corrosion of the bipolar plate leads to a release of metal ions that can contaminate the electrolyte membrane and poison the electrode catalysts, thereby exaggerating current concerns with long-term degradation in stack performance. In addition, the formation of a passivating oxide or oxyhydroxide product on the surface of the metal tends to increase the contact resistance between the bipolar plate and the adjacent graphite electrode backing layer, often by many orders of magnitude, which both limits the amount of power that can be generated by the stack and serves as an additional source of heat that must be removed during operation. A number of research groups have investigated various schemes for protecting metallic bipolar plates, most of which rely on a thin, inert yet electrically conductive coating [14–16]. While several of these look quite promising at the lab-scale, there are many practical problems with deploying a surface coating strategy in mass bipolar plate production, including the incorporation of flaws during processing, concerns with handling damage (chipping and scratching) during subsequent manufacturing steps, poor adhesion between the coating and underlying substrate during stack assembly, and the additional costs associated with the coating process. Ideally desired is a bipolar plate material that incorporates the advantages of metal, but undergoes little or no corrosion and is not susceptible to the potential manufacturing problems associated with coatings. Early transition metal elements such as Ti, Zr, Mo, V, and Nb are known to form a thin (sub-micron), adherent native oxide layer that passivates the underlying metal against corrosion in low pH, mineral acid environments. Of these niobium exhibits the best resistance to sulfuric acid, nominally the composition of the liquid environment within the PEMFC stack, with a reported rate of metal loss of < 0.2 m/yr in boiling 1 M H2 SO4 [17]. At issue is the relative cost of niobium compared to alternative, albeit less corrosion resistant, bipolar plate material candidates such as 304SS. Presented here are the results of
3725
testing to evaluate a clad metal concept that takes advantage of the beneficial properties of niobium, while minimizing the amount of material required and therefore its cost. 2. Experimental 2.1. Low-cost, clad metal bipolar plate concept The approach presently under development is to fabricate a metal laminate sheet that consists of a metallic core roll bonded to a thin sheet of a transition metal alloy which will form a passivating surface when exposed to the low pH aqueous environment within the stack. The manufacture of this type of clad material product is well-established commercially, with a number of producers capable of routinely fabricating various multilayer clad products in 50–500 m thick sheets. Ideally the material selected for the core, which will form the thickest layer, is chosen based primarily on material cost, formability, durability, and thermal conductivity. On the other hand, the material used in the cladding layer is selected based on corrosion resistance, surface contact resistance, formability, and cost. In this way, the bipolar plate can be tailored to take advantage of the merits of each material, while minimizing material and processing costs. Shown schematically in Fig. 1 is one of the clad metal concepts currently under study. The bipolar plate consists of a composite laminate structure formed by roll bonding a thin layer of niobium to one side of a stainless steel core and a braze filler metal layer on the other. The latter material facilitates the joining of two mated plates after stamping to form a bipolar plate component that contains an internal water cooling channel. As a first step in developing and evaluating this material, commercial purity niobium (cp-Nb) was clad to 430 stainless steel (430SS). 430SS was chosen because it is an inexpensive stainless steel that displays excellent formability. In the annealed condition cp-Nb also displays very good formability and ductility (up to 80% cold reduction) and is readily roll bonded to 430SS [18]. It is anticipated that once the concept viability is validated through ex-situ testing, the stainless core could be eventually replaced with an even lower cost material such as 1008 carbon steel. 2.2. Materials Characterization studies were initially conducted on 200 m thick, 99.5% purity niobium sheet (Alfa Aesar, Waltham, MA) to determine the feasibility of niobium as a potential cladding material for the bipolar plate application. The material was cut into 25 × 25 cm coupons for corrosion testing and interfacial contact resistance measurements, and 25 × 12.5 cm coupons for polarization testing. Prior to characterization, the specimens were degreased by ultrasonically cleaning in acetone and methanol followed by rinsing with distilled water. The bilayer Nb/430SS laminate was fabricated by Engineered Materials Solutions Inc. (EMS; Waltham, MA) via roll bonding to the following target dimensions: 450 m thick 430SS core with 50 m niobium to form a sheet measuring 0.5 mm in total thickness.
3726
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
Braze two halves to form final plate with internal cooling water channels
Flow channel (electrolyte side)
Cooling water channel
Core metal (e.g. stainless steel) Clad braze alloy
Clad passivation layer (e.g. niobium)
Fig. 1. An illustration of the stamped clad metal bipolar plate concept. In this example, the bipolar plate component is formed from two stamped pieces that are joined via a brazing layer to form an internal water channel. Alternatively the bipolar plate can be formed without the water channel from a single laminated metal piece that is clad on both exposed surfaces with a nitridable layer.
As with the monolithic niobium, the clad material was cut into 25 × 25 cm and 25 × 12.5 cm coupons for characterization, the surfaces of which were thoroughly degreased prior to testing. 2.3. Characterization The samples were analyzed by XRD before and after corrosion and polarization testing to identify any changes in the surface of the niobium due to exposure. The analysis was carried out with a Philips Wide-Range Vertical Goniometer and a Philips XRG3100 X-ray Generator over a scan range of 20–80◦ 2 with a 0.04◦ step size and 2 s hold time. XRD pattern analysis was conducted using Jade 6+ (EasyQuant) software. SEM/EDX analysis of the niobium coupons was conducted to examine the evolution in surface microstructure as a function of exposure time. The work was carried out using a JEOL JSM5900LV equipped with an Oxford Energy Dispersive X-ray Spectrometer (EDS) system that employs a windowless detector for quantitative detection of both light and heavy elements. Exposure testing of the monolithic coupons was conducted at 80 ◦ C in static 1 M H2 SO4 containing 2 ppm HF, an aqueous solution equivalent to the most aggressive pH condition under which the bipolar plate could be expected to operate [19]. H2 SO4 was selected because the PEMFC electrolyte membranes are pretreated with this acid and maintain this
environment during stack operation. The high acid concentration employed is rather aggressive, simulating the potential for direct contact between the bipolar plate and the polymer electrolyte, which is an accidental condition not desired under normal operation. The inclusion of 2 ppm HF was based on a previously reported analysis of water chemistry in an operating PEMFC system [11]. It is also noted that 80 ◦ C represents the upper range of operating temperature for most aqueous-based PEMFC systems. Thus the test conditions employed in the present study are accelerated relative to those typically encountered in the actual stack application. Coupons were removed after 100, 300, 400, 700, 1100, and 2000 h of exposure, dried and weighed, and compared with their initial pre-exposure weight to determine the amount of metal loss that occurred. In addition, the effluent to which each sample was exposed was analyzed by inductively coupled plasma mass spectrometry (ICP-MS; Agilent 4500) to determine the amount of niobium and other elements that dissolved in the solution over the period of exposure. Interfacial contact resistance was measured between the niobium and carbon paper, a typical gas diffusion layer (GDL) material, using a technique similar to that described by Mishra et al. [20]. A square piece of conductive carbon paper (Toray, Inc.) measuring 27 mm on each side was sandwiched between two niobium coupons. An electrical current of 200 mA was
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
applied from one coupon to the other via spot welded leads using a Hewlett-Packard 3263 A DC power source. The resulting voltage drop across the niobium/carbon/niobium sandwich was measured through a separate set of spot welded leads using a Hewlett-Packard 34 401 A multimeter, thereby establishing a four point measurement that eliminates lead resistance. The area specific contact resistance attributable is calculated by the following expression: R=
V QAs , 2I
(1)
where R is the interfacial contact resistance (assuming that the bulk resistance of the niobium and carbon are several orders of magnitude smaller than the contact resistance), V is the voltage drop through the sandwich, As is the area of contact between the niobium coupons and the carbon paper, and I is the applied current. Note that factor of 2 in the divisor for Eq. (1) averages the effects of both interfacial surfaces in the test specimen. The dependency of contact resistance on compressive hold down force was determined by carrying out the electrical measurements while the sandwich specimen was compressed under a known uniform load applied by a hydraulic ram. The amount of force applied was monitored using a load cell and the resulting average compressive stress was determined based on the area of the niobium coupons. Polarization testing was conducted at 80 ◦ C in 1 M H2 SO4 containing 2 ppm HF through which air or hydrogen was sparged to simulate the cathode or anode environment of the cell. A standard three-electrode electrochemical measurement technique was employed, with a 25 × 12.5 mm coupon of platinum as the counter electrode, a saturated calomel electrode (SCE) as the reference electrode, and the niobium specimen as the working electrode. A computer controlled potentiostat (Solatron 1287 A) was used to collect polarization data on the niobium coupons. Prior to conducting linear sweep voltammetry measurements, the test specimen was stabilized in solution at open circuit for 5 min. The scan was initiated from this potential in the anodic direction at a rate of 1 mV/s. In addition, potentiostatic experiments were conducted to gauge the performance of the niobium under prototypic PEMFC operating conditions. After allowing the test specimen to stabilize in solution, a potential was applied and the resulting current/time curves were recorded. Cathodic testing was conducted in sparged air at a potential of 0.6 V vs. SCE, while anodic tests were run in sparged hydrogen at −0.1 V vs. SCE. 3. Results 3.1. Niobium testing Shown in Table 1 are the results from aging tests conducted in the 80 ◦ C sulfuric acid solution. Weight measurements taken directly on the specimens before and after exposure indicate that niobium undergoes essentially no change in weight and is quite resistant to corrosion under these test conditions. These results are validated by the ICP-MS measurements of the amount of niobium dissolved in the corresponding effluent solutions,
3727
provided in column 3 of Table 1 and converted to an equivalent weight loss in column 4. According to this second set of data, the maximum amount of niobium loss observed was 0.022% over 2000 h of exposure and it appears that much of this weight loss is achieved within the first 100 h of exposure, after which little additional loss is observed. Like other early transition metals, niobium derives its corrosion resistance from a readily formed adherent oxide surface film, typically Nb2 O5 , on the order of a few nanometers thick [17,21]. The micrographs shown in Figs. 2(a)–(d) are planar images of the niobium specimen surface in the as-received condition and after H2 SO4 exposure for 100, 400, and 1126 h. It is apparent from this sequence of figures that there is no change in surface morphology that can be observed via SEM. In addition, results from EDS and XRD analyses indicate that the surface is 100% Nb, with no evidence of corrosion products. Due to the limitations of both techniques, the presence of a thin passivation layer could not be observed but is entirely expected based on the aging data in Table 1. However, a more sensitive surface analysis technique such as X-ray photoelectron spectroscopy (XPS) or Auger electron spectroscopy (AES) that is capable of detecting nanoscale thick surface layers would be required to probe the surface chemistry of the niobium as a function of exposure time. Nevertheless, the present results indicate no significant change that occurs in the niobium up to 1100 h of exposure. Measurements of the area specific interfacial contact resistance between niobium and a representative GDL carbon paper were recorded as a function of compaction force to determine the magnitude of loading required to establish an acceptably low level of contact resistance between the two materials. Plotted in Fig. 3 are the corresponding results for niobium in the as-received condition and after 300 h of exposure in 1 M H2 SO4 +2 ppm HF at 80 ◦ C. For each clamping pressure, three independent samples were tested and the variation of contact resistance is denoted by the vertical error bars. Note that contact resistance decreases dramatically with increasing compaction pressure, a behavior that is typically observed in these types of measurements for metal bipolar plate candidate materials [13,14,22,23]. In addition, the magnitude of contact resistance compares quite well with previously reported values for nitrided Ni-50Cr, 349SS, and 446SS [22,23], all of which are only slightly higher than that reported for carbon paper to graphite bipolar plate contact [20]. The low levels of contact resistance measured with these metals are presumably due to the nature of the native passivating oxide layer; i.e. semi-conducting in the case of chromia and niobia and measuring only a few tens of nanometers thick. Shown in Figs. 4(a) and (b) are anodic polarization curves for niobium in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under sparged hydrogen and air, respectively. For comparison, similar curves generated for a monolithic platinum electrode are plotted along with those for niobium. In addition, the anode and cathode potentials under typical PEMFC operation are marked. The current densities for niobium at −0.1 V in sparged hydrogen and 0.6 V in sparged air are quite low, 2.7 × 10−5 and 6.3 × 10−9 A/cm2 , respectively. Under actual operation, the
3728
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
Table 1 Results from aging tests conducted at 80 ◦ C in 1 M H2 SO4 with 2 ppm HF Time of exposure (h)
Measured wt loss (%)
[Nb] in effluent (ppm)
Estimated wt loss (%)
100 300 400 700 1100 2000
0 0 0 0 0 0
3.15 3.76 4.75 4.97 3.40 4.11
0.014 0.016 0.021 0.022 0.015 0.018
Fig. 2. Planar SEM micrographs of niobium coupons in (a) the as-received state and after exposure in 1 M H2 SO4 (2 ppm HF) at 80 ◦ C for (b) 400, (c) 700, and (d) 1126 h.
Contact Resistance, R (m Ω • cm2)
18 As-received condition
16
300hrs corrodant exposure
14 12 10 8 6 4 2 0 0
0.5
1
1.5
2
2.5
3
3.5
Clamping Pressure, P (MPa) Fig. 3. Interfacial contact resistance between niobium and carbon paper as a function of compaction pressure.
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
3729
0.0012 0.001 Current Density (A/cm2)
Pt
0.0008 0.0006 Anode potential in PEMFC operation
0.0004 0.0002 Nb
0 -0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
-0.0002 Voltage vs SCE (V)
(a) 4.00E-05
Current Density (A/cm2)
3.50E-05 Pt
3.00E-05 2.50E-05 Cathode potential in PEMFC operation
2.00E-05 1.50E-05 Nb
1.00E-05 5.00E-06 0.00E+00 0
0.2
0.4
(b)
0.6
0.8
1
1.2
1.4
Voltage vs SCE (V)
Fig. 4. Anodic behavior of niobium and platinum in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under (a) sparged hydrogen and (b) sparged air.
bipolar plates are exposed to a range of potentials at each electrode rather than simply a single set value. Both current density curves for niobium in Fig. 4 are relatively flat over respective potential ranges of −0.2 to 0.2 V and 0.5 to 0.9 V, although a short spike is observed at ∼ 0.84 V (j = 1.04 × 10−6 A/cm2 ) under sparged air. It is noted that the response under the prototypic anode condition is somewhat higher than the US DOE target of 10−6 A/cm2 , but as mentioned previously the test was conducted in an aggressive environment. It is anticipated that if tested under a more typical solution concentration of 0.1 M H2 SO4 the measured current density would drop correspondingly. Also note that the measured values for niobium are nearly equivalent to those of platinum at each potential. In fact, the general behavior of the two metals under linear polarization is very similar, indicating that niobium is quite noble under simulated PEMFC operating conditions. Further evidence of this is seen in the potentiostatic results shown in Figs. 5(a) and (b). Under the hydrogen purged anode environment, the transient current observed with both
metals quickly decays, reaching a stable current of approximately −4 A/cm2 in the case of platinum and approximately −10 A/cm2 for niobium. The decay in current is related to passive film formation. The current undergoes a transition from positive to negative after approximately 5 min, indicating that the surface passivation process is complete within this period. The negative current indicates that the film is immune to active dissolution under cathodic current flow [11]. In the air-sparged simulated PEMFC environment, platinum displays almost no transient current and a subsequent current of essentially zero throughout testing for 550 min. The niobium exhibits a transient current that decays to approximately zero after ∼ 300 min of testing. In this case the current curve does not change from anodic to cathodic, which is likely due to the fact that the applied potential lies in the passivation region for niobium, as seen in Fig. 4(b). The micrographs in Figs. 6(a)–(d) show a comparison of planar SEM images obtained on niobium specimens in the as-received condition and after potentiostatic testing in purged
3730
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
6.00E-05 -0.1V vs SCE in H2
Current Density (A/cm2)
5.00E-05 4.00E-05 3.00E-05 2.00E-05 1.00E-05
0 -1.00E-05 (a)
Time (min)
Pt 0.00E+00 Nb
100
200
300
400
500
600
300
400
500
600
-2.00E-05 3.00E-07 0.6V vs SCE in air
Current Density (A/cm2)
2.00E-07 1.00E-07 Pt 0.00E+00 0 -1.00E-07
Nb
100
200
Time (min)
-2.00E-07 -3.00E-07
(b)
-4.00E-07
Fig. 5. Behavior of niobium and platinum in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under (a) simulated anode operating conditions of −0.1 V and sparged hydrogen and (b) simulated cathode operating conditions of 0.6 V and sparged air.
hydrogen, which based on the polarization test results is the more corrosive environment for niobium. Although the coupons employed in polarization testing were of the same composition as those used in corrosion testing, they were rolled from a different lot of material. Thus the surface morphology differs somewhat from that indicated in Fig. 2(a), with the polarization specimens exhibiting a small amount of surface porosity measuring 1–5 m in size. However, comparison of the low magnification images in Figs. 6(a) and (b) and high magnification images in Figs. 6(c) and (d) shows virtually no difference in surface morphology or surface pore size between the two specimen conditions. In addition, results from EDS and XRD analyses indicate that the surfaces of the tested coupons are 100% Nb, although the previously discussed limitations of these measurement techniques are noted. 3.2. Nb/430 SS testing From the micrograph shown in Fig. 7(a), it is apparent that roll bonding process forms a metallurgical bond between the niobium cladding and underlying stainless with no interfacial porosity present. The bondline between the two materials
is quite distinct. Results from EDS characterization demonstrate only a minor amount of iron diffusion into the niobium cladding during warm rolling. Measurements of local chemistry at the points indicated in Fig. 7(a) are given in Table 2 and indicate that diffusion is limited to a ∼ 1 m thick region on either side of the bondline. An elemental line scan of iron, niobium, and chromium across the core/clad interface shown in Fig. 7(b) confirms this result. Unlike previous results obtained with a titanium clad stainless steel material [24], the diffusion zone is free of brittle intermetallic phases that could potentially limit the amount of forming that can take place in the clad sheet during subsequent stamping operations. Displayed in Fig. 8 are the results of area specific interfacial contact resistance measurements for the niobium clad material in the as-received condition as a function of compaction pressure. Note that the voltage and current leads were spot welded to the backside, i.e. 430SS side, of the coupons so that the niobium faces made contact with the carbon paper. As with monolithic niobium, the amount of compression required to achieve a low level of contact resistance is quite small and the magnitude of resistance is again quite comparable to that observed in surface treated and graphitic bipolar plate materials [14,20,22,23].
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
3731
Fig. 6. Planar SEM micrographs of niobium coupons in the as-received state at (a) 200× and (c) 1000× of magnification and after potentiostatic testing at (b) 200× and (d) 1000× of magnification. Testing was conducted in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under simulated anode conditions of −0.1 V and sparged hydrogen for 550 min.
Fig. 7. (a) Cross-sectional SEM micrograph of the clad Nb/430SS material in the as-received condition. (b) Elemental EDS analysis across the Nb/430SS interface in the clad specimen.
Shown in Fig. 9 is the anodic polarization curve for niobium clad 430SS material in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under sparged hydrogen. The curve is very similar to that obtained for monolithic niobium, Fig. 4(a), displaying a very low current density of 4.1 × 10−5 A/cm2 measured at the −0.1 V vs. SCE anode half cell potential. In addition, the trendline of the curve in Fig. 9 is similar to that recorded for platinum in Fig. 4(a), indicating that the niobium cladding layer displays a noble electrochemical behavior under simulated PEMFC operating conditions and effectively passivates and protects the underlying stainless steel from corrosion.
Table 2 EDX results for points marked in Fig. 7(a) Point
1 2 3 4 5 6
Composition at (%) Fe
Cr
Si
Nb
81.66 83.11 62.07 14.46 1.50 0.89
17.57 16.10 13.27 3.90 — —
0.77 0.79 0.79 — — —
— — 23.87 81.64 98.50 99.11
3732
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
4. Conclusions
Contact Resistance, R (m Ω • cm2)
The corrosion, polarization, and contact resistance properties of niobium were investigated for potential use in a new clad PEMFC bipolar plate material concept. Results from exposure testing in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C indicated that the corrosion resistance properties of niobium are exceptional under the conditions of this application. No weight loss was directly measured in the exposure coupons and subsequent ICP-MS analysis of the test effluents demonstrate less than 0.02% weight loss over a 2000 h period. Measurements of contact resistance between GDL carbon paper and niobium in the as-received or passivated conditions were quite low and comparable to previously reported values for modified stainless and Ni–Cr alloys, as well as for graphite. Linear voltammetry testing conducted in the same aqueous environment
16 14 12
employed in the corrosion study with the introduction of either sparged hydrogen or air indicated very low passivation currents. Potentiostatic testing conducted as a function of time demonstrated stable passivation under both simulated anode and cathode operating conditions and no observable change in surface morphology due to passivation, although a more sensitive surface measurement technique such as XPS or AES is required to identify the thin passive film that forms and understand the kinetics of its formation. Niobium clad 430SS specimens were fabricated by roll cladding and both contact resistance and linear sweep voltammetry measurements indicated that this material behaves in a manner that is nearly identical to that of monolithic niobium. Given these results, future work will focus on reducing the thickness of the niobium cladding layer to demonstrate the commercial viability of the material. It is anticipated that the overall thickness of the material can be reduced to 200 m or less with the Nb cladding layer representing 10% or less of this thickness, although the processing of this type of material in a cost-effective manner (i.e. single pass rolling) still needs to be demonstrated. Acknowledgments
10 8 6 4 2 0 0
0.5
1
1.5
2
3
2.5
3.5
Clamping Pressure, P (MPa) Fig. 8. Interfacial contact resistance between Nb/430SS and carbon paper as a function of compaction pressure.
The authors thank Steve Chang, Michael Hardy, and Rob Weiermair at Engineered Materials Solutions, Inc. for providing the clad materials and for their technical assistance on the study. In addition, the authors thank Nat Saenz, Shelly Carlson, and Jim Coleman for their assistance in metallographic preparation and SEM analysis. This work was supported by the US Department of Energy, Office of Energy Efficiency, and Renewable Energy. The Pacific Northwest National Laboratory is operated by Battelle Memorial Institute for the United States Department of Energy (US DOE) under Contract DE-AC0676RLO 1830.
1.00E-03 9.00E-04
Current Density (A/cm2)
8.00E-04 7.00E-04 6.00E-04 5.00E-04 4.00E-04
Anode potential in PEMFC operation
3.00E-04 2.00E-04 1.00E-04 0.00E+00 -0.4 -1.00E-04
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Voltage vs SCE (min) Fig. 9. Linear voltammetry scans of the niobium/430SS material under anodic conditions. Note that the 430SS portion of the coupon is covered by epoxy.
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
References [1] Chalk SG, Miller JF, Wagner FW. Challenges for fuel cells in transport applications. J Power Sources 2000;86(1–2):40–51. [2] Mehta V, Cooper JS. Review and analysis of PEM fuel cell design and manufacturing. J Power Sources 2003;114(1):32–53. [3] Donitz W. Fuel cells for mobile applications, status, requirements and future application potential. Int J Hydr Energy 1998;23(7):611–5. [4] Lin M, Cheng Y, Lin M, Yen S. Evaluation of PEMFC power systems for UPS base station applications. J Power Sources 2005;140(2): 346–9. [5] Li X, Sabir I. Review of bipolar plates in PEM fuel cells: flow-field designs. Int J Hydr Energy 2005;30(4):359–71. [6] Hermann A, Chaudhuri T, Spagnol P. Bipolar plates for PEM fuel cells: A review. Int J Hydr Energy 2005;30(12):1297–302. [7] Hornung R, Kappelt G. Bipolar plate materials development using Febased alloys for solid polymer fuel cells. J Power Sources 1998;72(1): 20–1. [8] Scholta J, Rohland B, Trapp V, Focken U. Investigations on novel lowcost graphite composite bipolar plates. J Power Sources 1999;84(2): 231–4. [9] Cunningham N, Guay D, Dodelet JP, Meng Y, Hlil AR, Hay AS. New materials and procedures to protect metallic PEM fuel cell bipolar plates. J Electrochem Soc 2002;149(7):A905–11. [10] Davies DP, Adcock PL, Turpin M, Rowen SJ. Bipolar plate materials for solid polymer fuel cells. J Appl Electrochem 2000;30(1):101–5. [11] Wang H, Sweikart MA, Turner JA. Stainless steel as bipolar plate material for polymer electrolyte membrane fuel cells. J Power Sources 2003;115(2):243–51. [12] Nikam VV, Reddy RG. Corrosion studies of a copper–beryllium alloy in a simulated polymer electrolyte membrane fuel cell environment. J Power Sources 2005;152(1):146–55. [13] Aukland N, Boudina A, Eddy DS, Mantese JV, Thompson MP, Wang SS. Alloys that form conductive and passivating oxides for proton exchange membrane fuel cell bipolar plates. J Mater Res 2004;19(6): 1723–9.
3733
[14] Lee S-J, Huang C-H, Chen Y-P. Investigation of PVD coating on corrosion resistance of metallic bipolar plates in PEM fuel cell. J Mater Process Techn 2003;140(1–3):688–93. [15] Brady MP, Weisbrod K, Paulauskas I, Buchanan RA, More KL, Wang H. et al. Preferential thermal nitridation to form pin-hole free Cr-nitrides to protect proton exchange membrane fuel cell metallic bipolar plates. Scr Mater 2004;50(7):1017–22. [16] McClure JC, Joseph S, Chianelli R, Pich P, Sebastian PJ. Conducting polymer-coated stainless steel bipolar plates for proton exchange membrane fuel cells (PEMFC). Int J Hydr Energy 2005;30(12):1339–44. [17] Robin A, de Almeida ME, Nunes CA. Corrosion behavior of niobium and Nb-25 wt.% Ta alloy in sulfuric acid solutions. Corrosion 1991;47(6): 443–8. [18] Jha SC, Delagi RG, Forster JA, Krotz PD. High-strength highconductivity Cu–Nb microcomposite sheet fabricated via multiple roll bonding. Metall Trans A 1993;24A(1):15–20. [19] Chu D, Jiang R. Comparative studies of polymer electrolyte membrane fuel cell stack and single cell. J Power Sources 1999;80(1–2):226–34. [20] Mishra V, Yang F, Pitchumani R. Measurement and prediction of electrical contact resistance between gas diffusion layers and bipolar plate for applications to PEM fuel cells. J Fuel Cell Sci Technol 2004;1(1): 2–9. [21] Robin A. Corrosion behavior of niobium, tantalum and their alloys in boiling sulfuric acid solutions. Int J Refr Metals Hard Mater 1997;15(5–6):317–23. [22] Wang H, Brady MP, Teeter G, Turner JA. Thermally nitrided stainless steels for polymer electrolyte membrane fuel cell bipolar plates, Part 1: model Ni–50Cr and austenitic 349 alloys. J Power Sources 2004;138(1–2):86–93. [23] Wang H, Brady MP, More KL, Meyer HM, Turner JA. Thermally nitrided stainless steels for polymer electrolyte membrane fuel cell bipolar plates Part 2: beneficial modification of passive layer on AISI446. J Power Sources 2004;138(1–2):86–93. [24] Weil KS, Kim JY, Xia GG, Yang ZG, Chang S, Hardy M. Development of low-cost, clad metal bipolar plates for PEM fuel cells. DOE-EERE (HFCIT) Report; July 2005.
Development of a niobium clad PEM fuel cell bipolar plate material K. Scott Weil ∗ , Gordon Xia, Z. Gary Yang, Jin Yong Kim Pacific Northwest National Laboratory, P.O. Box 999, Richland, WA 99352, USA Available online 4 October 2006
Abstract Reported in this paper are results obtained on a niobium clad material that is being developed for use in polymer electrolyte membrane fuel cell (PEMFC) stacks. A series of materials evaluation tests were initially conducted on niobium coupons to determine if this material was suitable as an external cladding layer exposed to a prototypic PEMFC environment. Results from corrosion testing conducted in 80 ◦ C, 1 M H2 SO4 (with 2 ppm HF) display no measurable weight loss in the niobium specimens out to 2000 h of exposure. Interfacial contact resistance measurements conducted on as-received and post-exposed niobium indicate that it exhibits excellent surface conductivity ( 10 m cm2 ) under low clamping forces in both conditions. Polarization testing carried out under both prototypic anodic and cathodic PEMFC operating conditions suggest that the electrochemical behavior of niobium is comparable to that of platinum, with current densities of 2.7 × 10−5 and 6.3 × 10−9 A/cm2 at half cell potentials of −0.1 and 0.6 V versus a saturated calomel reference electrode, respectively, under prototypic anode and cathode environmental conditions. Subsequent contact resistance and polarization testing of niobium clad stainless steel coupons yielded results similar to those found in monolithic niobium testing. 䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. Keywords: PEM fuel cell; Bipolar plate; Clad metal
1. Introduction There has been mounting public and political pressure in a number of industrialized nations, including the United States, to improve fuel utilization in power transportation applications and thereby reduce reliance on foreign oil supply and lessen pollution emissions. Coupled with continued growth in the worldwide demand for power, these factors have led to increasing interest in alternative methods of power generation, such as fuel cells, for various stationary and mobile applications, including off-power-grid residential power, auxiliary semi-truck power, electric automobiles, and portable personal power sources [1,2]. The primary feature of the fuel cell is that it directly converts the chemical energy of the incoming fuel into electrical energy via an electrochemical reaction. Because there is no intermediate thermal conversion step, and therefore no attendant Carnot cycle limitation, fuel cells provide a highly efficient means of energy conversion [3]. However, despite the significant technical progress that has been made in recent years toward ∗ Corresponding author. Tel.: +1 509 376 6796; fax: +1 509 375 2186.
E-mail address: [email protected] (K.S. Weil).
developing a commercially viable polymer electrolyte membrane fuel cell (PEMFC) system, the device currently finds use only niche applications [4]. Among the key reasons that PEMFC technology has not become more popular are the high cost of manufacture and the steady loss in power output during long-term, continuous operation. An additional factor that has hindered the acceptance of PEMFC systems in transportation markets is their current size and weight [5]. One of the most bulky components in the stack with respect to both weight and volume is the bipolar plate. In addition, it is one of the most expensive to manufacture. This component serves not only as the electrical junction between serially connected cells in the stack, but also performs several other key functions in the overall device, including [6]: • Distribute fuel and oxidant uniformly over the active areas of the cells. • Facilitate water management of the membrane to keep it humidified, yet mitigate flooding. • Act as an impermeable barrier between the fuel and oxidant streams (particularly H2 ) to maintain the hydrogen gradient across the membrane necessary for high power output.
0360-3199/$ - see front matter 䉷 2006 International Association for Hydrogen Energy. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2006.08.041
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
• Provide structural support for the stack. • Remove heat from the active areas of the cells. In the early development of PEMFCs, graphite was the material of choice for use in the bipolar plate because of its excellent corrosion resistance and surface conductivity in the low pH/hot aqueous environment (60–80 ◦ C) of the stack [2]. However, the high cost and low mechanical strength of high purity graphite and the expense associated with machining the individual plates necessitated the development of lower cost alternative bipolar plate materials for commercial applications. To date, carbon composites, various coated metals, and uncoated stainless steel and titanium have been considered [7–12]. Although carbonbased composites appear to be the leading candidates, there remain several challenges with these materials that are yet to be resolved, particularly for automotive and other mobile applications, including high manufacturing cost, insufficient mechanical strength, and poor barrier to hydrogen permeability [6]. While the use of metal in PEMFC bipolar plates appears to solve these issues, this class of materials is generally plagued by surface corrosion. The recent desire of stack designers to increase the operating temperature will only exacerbate this problem [13]. Corrosion of the bipolar plate leads to a release of metal ions that can contaminate the electrolyte membrane and poison the electrode catalysts, thereby exaggerating current concerns with long-term degradation in stack performance. In addition, the formation of a passivating oxide or oxyhydroxide product on the surface of the metal tends to increase the contact resistance between the bipolar plate and the adjacent graphite electrode backing layer, often by many orders of magnitude, which both limits the amount of power that can be generated by the stack and serves as an additional source of heat that must be removed during operation. A number of research groups have investigated various schemes for protecting metallic bipolar plates, most of which rely on a thin, inert yet electrically conductive coating [14–16]. While several of these look quite promising at the lab-scale, there are many practical problems with deploying a surface coating strategy in mass bipolar plate production, including the incorporation of flaws during processing, concerns with handling damage (chipping and scratching) during subsequent manufacturing steps, poor adhesion between the coating and underlying substrate during stack assembly, and the additional costs associated with the coating process. Ideally desired is a bipolar plate material that incorporates the advantages of metal, but undergoes little or no corrosion and is not susceptible to the potential manufacturing problems associated with coatings. Early transition metal elements such as Ti, Zr, Mo, V, and Nb are known to form a thin (sub-micron), adherent native oxide layer that passivates the underlying metal against corrosion in low pH, mineral acid environments. Of these niobium exhibits the best resistance to sulfuric acid, nominally the composition of the liquid environment within the PEMFC stack, with a reported rate of metal loss of < 0.2 m/yr in boiling 1 M H2 SO4 [17]. At issue is the relative cost of niobium compared to alternative, albeit less corrosion resistant, bipolar plate material candidates such as 304SS. Presented here are the results of
3725
testing to evaluate a clad metal concept that takes advantage of the beneficial properties of niobium, while minimizing the amount of material required and therefore its cost. 2. Experimental 2.1. Low-cost, clad metal bipolar plate concept The approach presently under development is to fabricate a metal laminate sheet that consists of a metallic core roll bonded to a thin sheet of a transition metal alloy which will form a passivating surface when exposed to the low pH aqueous environment within the stack. The manufacture of this type of clad material product is well-established commercially, with a number of producers capable of routinely fabricating various multilayer clad products in 50–500 m thick sheets. Ideally the material selected for the core, which will form the thickest layer, is chosen based primarily on material cost, formability, durability, and thermal conductivity. On the other hand, the material used in the cladding layer is selected based on corrosion resistance, surface contact resistance, formability, and cost. In this way, the bipolar plate can be tailored to take advantage of the merits of each material, while minimizing material and processing costs. Shown schematically in Fig. 1 is one of the clad metal concepts currently under study. The bipolar plate consists of a composite laminate structure formed by roll bonding a thin layer of niobium to one side of a stainless steel core and a braze filler metal layer on the other. The latter material facilitates the joining of two mated plates after stamping to form a bipolar plate component that contains an internal water cooling channel. As a first step in developing and evaluating this material, commercial purity niobium (cp-Nb) was clad to 430 stainless steel (430SS). 430SS was chosen because it is an inexpensive stainless steel that displays excellent formability. In the annealed condition cp-Nb also displays very good formability and ductility (up to 80% cold reduction) and is readily roll bonded to 430SS [18]. It is anticipated that once the concept viability is validated through ex-situ testing, the stainless core could be eventually replaced with an even lower cost material such as 1008 carbon steel. 2.2. Materials Characterization studies were initially conducted on 200 m thick, 99.5% purity niobium sheet (Alfa Aesar, Waltham, MA) to determine the feasibility of niobium as a potential cladding material for the bipolar plate application. The material was cut into 25 × 25 cm coupons for corrosion testing and interfacial contact resistance measurements, and 25 × 12.5 cm coupons for polarization testing. Prior to characterization, the specimens were degreased by ultrasonically cleaning in acetone and methanol followed by rinsing with distilled water. The bilayer Nb/430SS laminate was fabricated by Engineered Materials Solutions Inc. (EMS; Waltham, MA) via roll bonding to the following target dimensions: 450 m thick 430SS core with 50 m niobium to form a sheet measuring 0.5 mm in total thickness.
3726
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
Braze two halves to form final plate with internal cooling water channels
Flow channel (electrolyte side)
Cooling water channel
Core metal (e.g. stainless steel) Clad braze alloy
Clad passivation layer (e.g. niobium)
Fig. 1. An illustration of the stamped clad metal bipolar plate concept. In this example, the bipolar plate component is formed from two stamped pieces that are joined via a brazing layer to form an internal water channel. Alternatively the bipolar plate can be formed without the water channel from a single laminated metal piece that is clad on both exposed surfaces with a nitridable layer.
As with the monolithic niobium, the clad material was cut into 25 × 25 cm and 25 × 12.5 cm coupons for characterization, the surfaces of which were thoroughly degreased prior to testing. 2.3. Characterization The samples were analyzed by XRD before and after corrosion and polarization testing to identify any changes in the surface of the niobium due to exposure. The analysis was carried out with a Philips Wide-Range Vertical Goniometer and a Philips XRG3100 X-ray Generator over a scan range of 20–80◦ 2 with a 0.04◦ step size and 2 s hold time. XRD pattern analysis was conducted using Jade 6+ (EasyQuant) software. SEM/EDX analysis of the niobium coupons was conducted to examine the evolution in surface microstructure as a function of exposure time. The work was carried out using a JEOL JSM5900LV equipped with an Oxford Energy Dispersive X-ray Spectrometer (EDS) system that employs a windowless detector for quantitative detection of both light and heavy elements. Exposure testing of the monolithic coupons was conducted at 80 ◦ C in static 1 M H2 SO4 containing 2 ppm HF, an aqueous solution equivalent to the most aggressive pH condition under which the bipolar plate could be expected to operate [19]. H2 SO4 was selected because the PEMFC electrolyte membranes are pretreated with this acid and maintain this
environment during stack operation. The high acid concentration employed is rather aggressive, simulating the potential for direct contact between the bipolar plate and the polymer electrolyte, which is an accidental condition not desired under normal operation. The inclusion of 2 ppm HF was based on a previously reported analysis of water chemistry in an operating PEMFC system [11]. It is also noted that 80 ◦ C represents the upper range of operating temperature for most aqueous-based PEMFC systems. Thus the test conditions employed in the present study are accelerated relative to those typically encountered in the actual stack application. Coupons were removed after 100, 300, 400, 700, 1100, and 2000 h of exposure, dried and weighed, and compared with their initial pre-exposure weight to determine the amount of metal loss that occurred. In addition, the effluent to which each sample was exposed was analyzed by inductively coupled plasma mass spectrometry (ICP-MS; Agilent 4500) to determine the amount of niobium and other elements that dissolved in the solution over the period of exposure. Interfacial contact resistance was measured between the niobium and carbon paper, a typical gas diffusion layer (GDL) material, using a technique similar to that described by Mishra et al. [20]. A square piece of conductive carbon paper (Toray, Inc.) measuring 27 mm on each side was sandwiched between two niobium coupons. An electrical current of 200 mA was
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
applied from one coupon to the other via spot welded leads using a Hewlett-Packard 3263 A DC power source. The resulting voltage drop across the niobium/carbon/niobium sandwich was measured through a separate set of spot welded leads using a Hewlett-Packard 34 401 A multimeter, thereby establishing a four point measurement that eliminates lead resistance. The area specific contact resistance attributable is calculated by the following expression: R=
V QAs , 2I
(1)
where R is the interfacial contact resistance (assuming that the bulk resistance of the niobium and carbon are several orders of magnitude smaller than the contact resistance), V is the voltage drop through the sandwich, As is the area of contact between the niobium coupons and the carbon paper, and I is the applied current. Note that factor of 2 in the divisor for Eq. (1) averages the effects of both interfacial surfaces in the test specimen. The dependency of contact resistance on compressive hold down force was determined by carrying out the electrical measurements while the sandwich specimen was compressed under a known uniform load applied by a hydraulic ram. The amount of force applied was monitored using a load cell and the resulting average compressive stress was determined based on the area of the niobium coupons. Polarization testing was conducted at 80 ◦ C in 1 M H2 SO4 containing 2 ppm HF through which air or hydrogen was sparged to simulate the cathode or anode environment of the cell. A standard three-electrode electrochemical measurement technique was employed, with a 25 × 12.5 mm coupon of platinum as the counter electrode, a saturated calomel electrode (SCE) as the reference electrode, and the niobium specimen as the working electrode. A computer controlled potentiostat (Solatron 1287 A) was used to collect polarization data on the niobium coupons. Prior to conducting linear sweep voltammetry measurements, the test specimen was stabilized in solution at open circuit for 5 min. The scan was initiated from this potential in the anodic direction at a rate of 1 mV/s. In addition, potentiostatic experiments were conducted to gauge the performance of the niobium under prototypic PEMFC operating conditions. After allowing the test specimen to stabilize in solution, a potential was applied and the resulting current/time curves were recorded. Cathodic testing was conducted in sparged air at a potential of 0.6 V vs. SCE, while anodic tests were run in sparged hydrogen at −0.1 V vs. SCE. 3. Results 3.1. Niobium testing Shown in Table 1 are the results from aging tests conducted in the 80 ◦ C sulfuric acid solution. Weight measurements taken directly on the specimens before and after exposure indicate that niobium undergoes essentially no change in weight and is quite resistant to corrosion under these test conditions. These results are validated by the ICP-MS measurements of the amount of niobium dissolved in the corresponding effluent solutions,
3727
provided in column 3 of Table 1 and converted to an equivalent weight loss in column 4. According to this second set of data, the maximum amount of niobium loss observed was 0.022% over 2000 h of exposure and it appears that much of this weight loss is achieved within the first 100 h of exposure, after which little additional loss is observed. Like other early transition metals, niobium derives its corrosion resistance from a readily formed adherent oxide surface film, typically Nb2 O5 , on the order of a few nanometers thick [17,21]. The micrographs shown in Figs. 2(a)–(d) are planar images of the niobium specimen surface in the as-received condition and after H2 SO4 exposure for 100, 400, and 1126 h. It is apparent from this sequence of figures that there is no change in surface morphology that can be observed via SEM. In addition, results from EDS and XRD analyses indicate that the surface is 100% Nb, with no evidence of corrosion products. Due to the limitations of both techniques, the presence of a thin passivation layer could not be observed but is entirely expected based on the aging data in Table 1. However, a more sensitive surface analysis technique such as X-ray photoelectron spectroscopy (XPS) or Auger electron spectroscopy (AES) that is capable of detecting nanoscale thick surface layers would be required to probe the surface chemistry of the niobium as a function of exposure time. Nevertheless, the present results indicate no significant change that occurs in the niobium up to 1100 h of exposure. Measurements of the area specific interfacial contact resistance between niobium and a representative GDL carbon paper were recorded as a function of compaction force to determine the magnitude of loading required to establish an acceptably low level of contact resistance between the two materials. Plotted in Fig. 3 are the corresponding results for niobium in the as-received condition and after 300 h of exposure in 1 M H2 SO4 +2 ppm HF at 80 ◦ C. For each clamping pressure, three independent samples were tested and the variation of contact resistance is denoted by the vertical error bars. Note that contact resistance decreases dramatically with increasing compaction pressure, a behavior that is typically observed in these types of measurements for metal bipolar plate candidate materials [13,14,22,23]. In addition, the magnitude of contact resistance compares quite well with previously reported values for nitrided Ni-50Cr, 349SS, and 446SS [22,23], all of which are only slightly higher than that reported for carbon paper to graphite bipolar plate contact [20]. The low levels of contact resistance measured with these metals are presumably due to the nature of the native passivating oxide layer; i.e. semi-conducting in the case of chromia and niobia and measuring only a few tens of nanometers thick. Shown in Figs. 4(a) and (b) are anodic polarization curves for niobium in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under sparged hydrogen and air, respectively. For comparison, similar curves generated for a monolithic platinum electrode are plotted along with those for niobium. In addition, the anode and cathode potentials under typical PEMFC operation are marked. The current densities for niobium at −0.1 V in sparged hydrogen and 0.6 V in sparged air are quite low, 2.7 × 10−5 and 6.3 × 10−9 A/cm2 , respectively. Under actual operation, the
3728
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
Table 1 Results from aging tests conducted at 80 ◦ C in 1 M H2 SO4 with 2 ppm HF Time of exposure (h)
Measured wt loss (%)
[Nb] in effluent (ppm)
Estimated wt loss (%)
100 300 400 700 1100 2000
0 0 0 0 0 0
3.15 3.76 4.75 4.97 3.40 4.11
0.014 0.016 0.021 0.022 0.015 0.018
Fig. 2. Planar SEM micrographs of niobium coupons in (a) the as-received state and after exposure in 1 M H2 SO4 (2 ppm HF) at 80 ◦ C for (b) 400, (c) 700, and (d) 1126 h.
Contact Resistance, R (m Ω • cm2)
18 As-received condition
16
300hrs corrodant exposure
14 12 10 8 6 4 2 0 0
0.5
1
1.5
2
2.5
3
3.5
Clamping Pressure, P (MPa) Fig. 3. Interfacial contact resistance between niobium and carbon paper as a function of compaction pressure.
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
3729
0.0012 0.001 Current Density (A/cm2)
Pt
0.0008 0.0006 Anode potential in PEMFC operation
0.0004 0.0002 Nb
0 -0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
-0.0002 Voltage vs SCE (V)
(a) 4.00E-05
Current Density (A/cm2)
3.50E-05 Pt
3.00E-05 2.50E-05 Cathode potential in PEMFC operation
2.00E-05 1.50E-05 Nb
1.00E-05 5.00E-06 0.00E+00 0
0.2
0.4
(b)
0.6
0.8
1
1.2
1.4
Voltage vs SCE (V)
Fig. 4. Anodic behavior of niobium and platinum in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under (a) sparged hydrogen and (b) sparged air.
bipolar plates are exposed to a range of potentials at each electrode rather than simply a single set value. Both current density curves for niobium in Fig. 4 are relatively flat over respective potential ranges of −0.2 to 0.2 V and 0.5 to 0.9 V, although a short spike is observed at ∼ 0.84 V (j = 1.04 × 10−6 A/cm2 ) under sparged air. It is noted that the response under the prototypic anode condition is somewhat higher than the US DOE target of 10−6 A/cm2 , but as mentioned previously the test was conducted in an aggressive environment. It is anticipated that if tested under a more typical solution concentration of 0.1 M H2 SO4 the measured current density would drop correspondingly. Also note that the measured values for niobium are nearly equivalent to those of platinum at each potential. In fact, the general behavior of the two metals under linear polarization is very similar, indicating that niobium is quite noble under simulated PEMFC operating conditions. Further evidence of this is seen in the potentiostatic results shown in Figs. 5(a) and (b). Under the hydrogen purged anode environment, the transient current observed with both
metals quickly decays, reaching a stable current of approximately −4 A/cm2 in the case of platinum and approximately −10 A/cm2 for niobium. The decay in current is related to passive film formation. The current undergoes a transition from positive to negative after approximately 5 min, indicating that the surface passivation process is complete within this period. The negative current indicates that the film is immune to active dissolution under cathodic current flow [11]. In the air-sparged simulated PEMFC environment, platinum displays almost no transient current and a subsequent current of essentially zero throughout testing for 550 min. The niobium exhibits a transient current that decays to approximately zero after ∼ 300 min of testing. In this case the current curve does not change from anodic to cathodic, which is likely due to the fact that the applied potential lies in the passivation region for niobium, as seen in Fig. 4(b). The micrographs in Figs. 6(a)–(d) show a comparison of planar SEM images obtained on niobium specimens in the as-received condition and after potentiostatic testing in purged
3730
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
6.00E-05 -0.1V vs SCE in H2
Current Density (A/cm2)
5.00E-05 4.00E-05 3.00E-05 2.00E-05 1.00E-05
0 -1.00E-05 (a)
Time (min)
Pt 0.00E+00 Nb
100
200
300
400
500
600
300
400
500
600
-2.00E-05 3.00E-07 0.6V vs SCE in air
Current Density (A/cm2)
2.00E-07 1.00E-07 Pt 0.00E+00 0 -1.00E-07
Nb
100
200
Time (min)
-2.00E-07 -3.00E-07
(b)
-4.00E-07
Fig. 5. Behavior of niobium and platinum in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under (a) simulated anode operating conditions of −0.1 V and sparged hydrogen and (b) simulated cathode operating conditions of 0.6 V and sparged air.
hydrogen, which based on the polarization test results is the more corrosive environment for niobium. Although the coupons employed in polarization testing were of the same composition as those used in corrosion testing, they were rolled from a different lot of material. Thus the surface morphology differs somewhat from that indicated in Fig. 2(a), with the polarization specimens exhibiting a small amount of surface porosity measuring 1–5 m in size. However, comparison of the low magnification images in Figs. 6(a) and (b) and high magnification images in Figs. 6(c) and (d) shows virtually no difference in surface morphology or surface pore size between the two specimen conditions. In addition, results from EDS and XRD analyses indicate that the surfaces of the tested coupons are 100% Nb, although the previously discussed limitations of these measurement techniques are noted. 3.2. Nb/430 SS testing From the micrograph shown in Fig. 7(a), it is apparent that roll bonding process forms a metallurgical bond between the niobium cladding and underlying stainless with no interfacial porosity present. The bondline between the two materials
is quite distinct. Results from EDS characterization demonstrate only a minor amount of iron diffusion into the niobium cladding during warm rolling. Measurements of local chemistry at the points indicated in Fig. 7(a) are given in Table 2 and indicate that diffusion is limited to a ∼ 1 m thick region on either side of the bondline. An elemental line scan of iron, niobium, and chromium across the core/clad interface shown in Fig. 7(b) confirms this result. Unlike previous results obtained with a titanium clad stainless steel material [24], the diffusion zone is free of brittle intermetallic phases that could potentially limit the amount of forming that can take place in the clad sheet during subsequent stamping operations. Displayed in Fig. 8 are the results of area specific interfacial contact resistance measurements for the niobium clad material in the as-received condition as a function of compaction pressure. Note that the voltage and current leads were spot welded to the backside, i.e. 430SS side, of the coupons so that the niobium faces made contact with the carbon paper. As with monolithic niobium, the amount of compression required to achieve a low level of contact resistance is quite small and the magnitude of resistance is again quite comparable to that observed in surface treated and graphitic bipolar plate materials [14,20,22,23].
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
3731
Fig. 6. Planar SEM micrographs of niobium coupons in the as-received state at (a) 200× and (c) 1000× of magnification and after potentiostatic testing at (b) 200× and (d) 1000× of magnification. Testing was conducted in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under simulated anode conditions of −0.1 V and sparged hydrogen for 550 min.
Fig. 7. (a) Cross-sectional SEM micrograph of the clad Nb/430SS material in the as-received condition. (b) Elemental EDS analysis across the Nb/430SS interface in the clad specimen.
Shown in Fig. 9 is the anodic polarization curve for niobium clad 430SS material in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C under sparged hydrogen. The curve is very similar to that obtained for monolithic niobium, Fig. 4(a), displaying a very low current density of 4.1 × 10−5 A/cm2 measured at the −0.1 V vs. SCE anode half cell potential. In addition, the trendline of the curve in Fig. 9 is similar to that recorded for platinum in Fig. 4(a), indicating that the niobium cladding layer displays a noble electrochemical behavior under simulated PEMFC operating conditions and effectively passivates and protects the underlying stainless steel from corrosion.
Table 2 EDX results for points marked in Fig. 7(a) Point
1 2 3 4 5 6
Composition at (%) Fe
Cr
Si
Nb
81.66 83.11 62.07 14.46 1.50 0.89
17.57 16.10 13.27 3.90 — —
0.77 0.79 0.79 — — —
— — 23.87 81.64 98.50 99.11
3732
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
4. Conclusions
Contact Resistance, R (m Ω • cm2)
The corrosion, polarization, and contact resistance properties of niobium were investigated for potential use in a new clad PEMFC bipolar plate material concept. Results from exposure testing in 1 M H2 SO4 + 2 ppm HF at 80 ◦ C indicated that the corrosion resistance properties of niobium are exceptional under the conditions of this application. No weight loss was directly measured in the exposure coupons and subsequent ICP-MS analysis of the test effluents demonstrate less than 0.02% weight loss over a 2000 h period. Measurements of contact resistance between GDL carbon paper and niobium in the as-received or passivated conditions were quite low and comparable to previously reported values for modified stainless and Ni–Cr alloys, as well as for graphite. Linear voltammetry testing conducted in the same aqueous environment
16 14 12
employed in the corrosion study with the introduction of either sparged hydrogen or air indicated very low passivation currents. Potentiostatic testing conducted as a function of time demonstrated stable passivation under both simulated anode and cathode operating conditions and no observable change in surface morphology due to passivation, although a more sensitive surface measurement technique such as XPS or AES is required to identify the thin passive film that forms and understand the kinetics of its formation. Niobium clad 430SS specimens were fabricated by roll cladding and both contact resistance and linear sweep voltammetry measurements indicated that this material behaves in a manner that is nearly identical to that of monolithic niobium. Given these results, future work will focus on reducing the thickness of the niobium cladding layer to demonstrate the commercial viability of the material. It is anticipated that the overall thickness of the material can be reduced to 200 m or less with the Nb cladding layer representing 10% or less of this thickness, although the processing of this type of material in a cost-effective manner (i.e. single pass rolling) still needs to be demonstrated. Acknowledgments
10 8 6 4 2 0 0
0.5
1
1.5
2
3
2.5
3.5
Clamping Pressure, P (MPa) Fig. 8. Interfacial contact resistance between Nb/430SS and carbon paper as a function of compaction pressure.
The authors thank Steve Chang, Michael Hardy, and Rob Weiermair at Engineered Materials Solutions, Inc. for providing the clad materials and for their technical assistance on the study. In addition, the authors thank Nat Saenz, Shelly Carlson, and Jim Coleman for their assistance in metallographic preparation and SEM analysis. This work was supported by the US Department of Energy, Office of Energy Efficiency, and Renewable Energy. The Pacific Northwest National Laboratory is operated by Battelle Memorial Institute for the United States Department of Energy (US DOE) under Contract DE-AC0676RLO 1830.
1.00E-03 9.00E-04
Current Density (A/cm2)
8.00E-04 7.00E-04 6.00E-04 5.00E-04 4.00E-04
Anode potential in PEMFC operation
3.00E-04 2.00E-04 1.00E-04 0.00E+00 -0.4 -1.00E-04
-0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Voltage vs SCE (min) Fig. 9. Linear voltammetry scans of the niobium/430SS material under anodic conditions. Note that the 430SS portion of the coupon is covered by epoxy.
K.S. Weil et al. / International Journal of Hydrogen Energy 32 (2007) 3724 – 3733
References [1] Chalk SG, Miller JF, Wagner FW. Challenges for fuel cells in transport applications. J Power Sources 2000;86(1–2):40–51. [2] Mehta V, Cooper JS. Review and analysis of PEM fuel cell design and manufacturing. J Power Sources 2003;114(1):32–53. [3] Donitz W. Fuel cells for mobile applications, status, requirements and future application potential. Int J Hydr Energy 1998;23(7):611–5. [4] Lin M, Cheng Y, Lin M, Yen S. Evaluation of PEMFC power systems for UPS base station applications. J Power Sources 2005;140(2): 346–9. [5] Li X, Sabir I. Review of bipolar plates in PEM fuel cells: flow-field designs. Int J Hydr Energy 2005;30(4):359–71. [6] Hermann A, Chaudhuri T, Spagnol P. Bipolar plates for PEM fuel cells: A review. Int J Hydr Energy 2005;30(12):1297–302. [7] Hornung R, Kappelt G. Bipolar plate materials development using Febased alloys for solid polymer fuel cells. J Power Sources 1998;72(1): 20–1. [8] Scholta J, Rohland B, Trapp V, Focken U. Investigations on novel lowcost graphite composite bipolar plates. J Power Sources 1999;84(2): 231–4. [9] Cunningham N, Guay D, Dodelet JP, Meng Y, Hlil AR, Hay AS. New materials and procedures to protect metallic PEM fuel cell bipolar plates. J Electrochem Soc 2002;149(7):A905–11. [10] Davies DP, Adcock PL, Turpin M, Rowen SJ. Bipolar plate materials for solid polymer fuel cells. J Appl Electrochem 2000;30(1):101–5. [11] Wang H, Sweikart MA, Turner JA. Stainless steel as bipolar plate material for polymer electrolyte membrane fuel cells. J Power Sources 2003;115(2):243–51. [12] Nikam VV, Reddy RG. Corrosion studies of a copper–beryllium alloy in a simulated polymer electrolyte membrane fuel cell environment. J Power Sources 2005;152(1):146–55. [13] Aukland N, Boudina A, Eddy DS, Mantese JV, Thompson MP, Wang SS. Alloys that form conductive and passivating oxides for proton exchange membrane fuel cell bipolar plates. J Mater Res 2004;19(6): 1723–9.
3733
[14] Lee S-J, Huang C-H, Chen Y-P. Investigation of PVD coating on corrosion resistance of metallic bipolar plates in PEM fuel cell. J Mater Process Techn 2003;140(1–3):688–93. [15] Brady MP, Weisbrod K, Paulauskas I, Buchanan RA, More KL, Wang H. et al. Preferential thermal nitridation to form pin-hole free Cr-nitrides to protect proton exchange membrane fuel cell metallic bipolar plates. Scr Mater 2004;50(7):1017–22. [16] McClure JC, Joseph S, Chianelli R, Pich P, Sebastian PJ. Conducting polymer-coated stainless steel bipolar plates for proton exchange membrane fuel cells (PEMFC). Int J Hydr Energy 2005;30(12):1339–44. [17] Robin A, de Almeida ME, Nunes CA. Corrosion behavior of niobium and Nb-25 wt.% Ta alloy in sulfuric acid solutions. Corrosion 1991;47(6): 443–8. [18] Jha SC, Delagi RG, Forster JA, Krotz PD. High-strength highconductivity Cu–Nb microcomposite sheet fabricated via multiple roll bonding. Metall Trans A 1993;24A(1):15–20. [19] Chu D, Jiang R. Comparative studies of polymer electrolyte membrane fuel cell stack and single cell. J Power Sources 1999;80(1–2):226–34. [20] Mishra V, Yang F, Pitchumani R. Measurement and prediction of electrical contact resistance between gas diffusion layers and bipolar plate for applications to PEM fuel cells. J Fuel Cell Sci Technol 2004;1(1): 2–9. [21] Robin A. Corrosion behavior of niobium, tantalum and their alloys in boiling sulfuric acid solutions. Int J Refr Metals Hard Mater 1997;15(5–6):317–23. [22] Wang H, Brady MP, Teeter G, Turner JA. Thermally nitrided stainless steels for polymer electrolyte membrane fuel cell bipolar plates, Part 1: model Ni–50Cr and austenitic 349 alloys. J Power Sources 2004;138(1–2):86–93. [23] Wang H, Brady MP, More KL, Meyer HM, Turner JA. Thermally nitrided stainless steels for polymer electrolyte membrane fuel cell bipolar plates Part 2: beneficial modification of passive layer on AISI446. J Power Sources 2004;138(1–2):86–93. [24] Weil KS, Kim JY, Xia GG, Yang ZG, Chang S, Hardy M. Development of low-cost, clad metal bipolar plates for PEM fuel cells. DOE-EERE (HFCIT) Report; July 2005.