- Email: [email protected]
Development of chromium barrier coatings for solid oxide fuel cells
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
Development of chromium barrier coatings for solid oxide fuel cells Dilip Chatterjee a,b,*, Samir Biswas a a b
Corning Incorporated, Corning, NY 14831, USA Director of Research, Oxane Materials Inc, 467 W 38th Street, Houston, TX 77018, USA
article info
abstract
Article history:
The performance of electrolyte supported solid oxide fuel cell is impaired primarily due to
Received 22 December 2009
poisoning of electrodes due to contaminants generated from the metallic components of
Received in revised form
the stack.
30 March 2010
Ferritic stainless steels are commonly used as stack material under severe operating
Accepted 18 April 2010
conditions of SOFC environment. However, the high chromium content in this type of
Available online 3 June 2010
steels tends to form gaseous oxides and/or hydroxides which volatilize and condense on various components of stack assembly, particularly cathodes, resulting in performance
Keywords:
degradation of the system.
SOFC
Two types of barrier coatings have been developed to minimize the chromium volatiliza-
Cathode poisoning
tion. In one case, coatings of oxide species were deposited by processes such as thermal
Chromium volatilization
and plasma spraying, and the other is by diffusion coating process such as aluminizing.
Diffusion coating
This presentation will describe various barrier coatings, barrier properties provided by the
Aluminide
coatings, and transpiration measurements adopted to evaluate the efficiency of those coatings.
Plasma spray
Copyright ª 2010, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Spinel
1.
Introduction
In electrolyte supported solid oxide fuel cell (SOFC) stack, a frame capable of withstanding SOFC operating environments is required to support the solid oxide electrolyte sheet of yttria-stabilized zirconia (YSZ). Various types of ferritic stainless steel such as AISI 446, 430 and E-Brite (Allegheny Ludlum) were under evaluation as potential candidate frame material (Fig. 1). These steels posses the coefficient of thermal expansion in the range of 11.2e11.6 ppm/ C, which is compatible with that of the 3 mol% YSZ electrolyte, in addition to their stability in both oxidizing and reducing environments at the operating temperature around 750 C. However, all these steels mentioned above have chromium content in the range of 18e27 wt.%. All types of stainless steels form Cr2O3 during oxidation at elevated temperature which acts as a base
protective scale. But under fuel cell operating conditions of high humidity and elevated temperature, Cr2O3 scale is oxidized to gaseous species such as CrO3 or CrO2(OH)2), through the following reactions: 2Cr þ 1.5O2 (g) ¼ Cr2O3 (s)
Cr2O3 (s) þ 1.5O2 (g) ¼ 2CrO3 (g)
Cr2O3 (s) þ 2H2O (g) þ 3/2O2 (g) ¼ 2 CrO2(OH)2 (g)
According to Ebbinghaus’ calculated and measured thermodynamic data [1], CrO2(OH)2 (g) is the most dominant
* Corresponding author. E-mail address: [email protected] (D. Chatterjee). 0360-3199/$ e see front matter Copyright ª 2010, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2010.04.114
4531
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 1 e A typical electrolyte supported SOFC stack assembly.
volatile component at moderate temperature in the presence of O2 and H2O such as in the fuel cell operating environment. However, the above species are reduced at the cathode side causing degradation of the cell performance known as “cathode poisoning” according to the following reactions [2].
2CrO2(OH)2 (g) þ 6e / Cr2O3 (s) þ 2H2O (g) þ 3O2
2CrO3 (g) þ 6e / Cr2O3 (s) þ 3O2 In cathode poisoning, gaseous CrO3 (g) and CrO2 (OH)2 (g) deposit onto the cathode as solid Cr2O3 (s) and degrades SOFC performances over time. Abatement of chromium evaporation from ferritic stainless steels by surface modification of the frame material or by developing a barrier layer is of significant interest in SOFC development. An effective barrier layer for electrolyte supported SOFC system should possess a few ideal attributes, such as (a) minimum mismatch of thermal expansion coefficients (CTE) between the glass seal materials and the frame materials; (b) electrically non-conductive; and (c) low diffusivity of chromium in it. This paper summarizes approaches in development of suitable barrier coatings and evaluation of their effectiveness.
2. Approach and details of various coating processes Considering the advantages and disadvantages of various candidate materials for coating on ferritic stainless steel, it was determined that alumina and/or other mixed oxides (spinel) would meet the criteria of an effective barrier coating on ferritic stainless steels, except for its considerable mismatch of CTE with stainless steels (as noted below in Table 1).
Alumina is dielectric, and dense. It possesses low diffusivity of chromium at high temperature (estimated w1022 m2/s at 800 C) [5]). Another important parameter for the barrier coating development is determination of its optimum thickness for suppression of the volatile species migration through the barrier layer. Theoretical calculations [6] predicted that alumina barrier layer on AISI 446 ferritic stainless steel should be at least in the range of 20 mm thickness to ensure chrome concentration to be well below 5 wt.% at the outer surface of the barrier layer after five years (w 40,000 h) of fuel cell operation at 800 C. Depending on the methods for application of barrier coating on stainless steel substrate and the desired coating thickness, the coating applications were classified as (a) direct deposition of alumina or other mixed oxides, such as spinel and (b) in-situ formation of thermally grown alumina from diffusion coated substrate as described below.
2.1. Direct deposition of alumina or other oxides as Cr barrier layer on ferritic stainless steels Thermal and Plasma Spray e These are process in which metallic and nonmetallic materials are deposited in a molten or semi-molten state on a prepared substrate imparting properties that the substrate would not otherwise possess. In these methods either an electric or gas source (for thermal spray) or plasma sources (for plasma spray) helps melting alumina powder feedstock. As the molten powder particles impinge with high velocities on to a ferritic stainless steel substrate or on to a SOFC frame, they form a dense coating of
Table 1 e Coefficient of thermal expansion. Material Ferritic Stainless Steele(E-Brite) Alumina a from reference [3]. b from reference [4].
CTE at 750 C (ppm/ C) 11.7a 8.2b
4532
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
oxides (alumina, zirconia or other mixed oxides) on steel. Plasma spraying of alumina and other oxides (Zirconia or spinel) on a number of ferritic stainless steels coupons and SOFC frames was investigated. The details of barrier layer microstructures, its advantages and disadvantages are described in a later section.
2.2. Diffusion coatings for in-situ formation of alumina (thermally grown oxides) as Cr barrier layer on ferritic stainless steels In this approach, intermetallic coatings with alumina forming characteristics are formed through a high temperature diffusion process known as aluminizing, whereby aluminum species diffuse into the surface of the base metal forming new metallurgical aluminide alloys (depending on process temperature, reaction time, and aluminum activity of the starting aluminum precursors). The aluminide intermetallics formed on the surface typically contain w 20% (by weight) aluminum which acts a reservoir of aluminum. Under thermodynamically suitable conditions, the coating layer is expected to form thermally grown oxide in-situ when the part is subjected to high temperature in service. Typical depth of the aluminide layer forming on stainless alloy is in the range of about 35e150 mm. There are different variations of the aluminizing process such as: a) b) c) d)
Pack aluminizing Slurry aluminizing (vapor phase) Out of pack aluminizing (vapor phase) CVD aluminizing
Pack aluminizing e the parts to be aluminized are put inside a pack consisting of aluminum halide vapor producing precursors (e.g. mixture of an aluminum source material such as powdered aluminum and or aluminum chloride, an inert material such as alumina, and an activator such as ammonium chloride) and heated at high temperatures. During the process aluminum melts, reacts with the halide to form aluminum halide gas, the gas contacts the part surfaces and is reduced and the aluminum reacts on the surfaces. Slurry aluminizing (vapor phase) e In this process slurries comprising of an aluminum alloy pigment and a halide compound in an organic binder is sprayed on the substrate. When this slurry is heated, the halide reacts with aluminum in the alloy pigment to form aluminum halide vapors. These vapors migrate through the slurry layer to the substrate where they react to release aluminum. The parts are subsequently annealed at high temperatures to diffuse aluminum into the metal surface. Out of Pack aluminizing e Out of Pack aluminizing process is essentially same as that of pack aluminizing, however, in this case the aluminum vapor is generated in-situ. The pack material (aluminum source material, activator, and inert filler, if required) described above are placed into a special multichamber reactor that has gas flow tubes to interconnect the chambers and to provide aluminum gas to the parts to be aluminized. The chambers are heated and an aluminum rich gas forms and the balance of the process is essentially as for the pack aluminizing.
The advantages of ‘out of pack’ aluminizing over ‘pack aluminizing’ include is the cleanliness of the process and aluminized coatings formed by this process are contaminant free. This process is capable of producing aluminized layers all over the part, including internal cavities (i.e. non-line of sight process). CVD aluminizing e This process offers a more environmentally friendly, better forming alternative to older pack and out of pack aluminizing technologies. This process utilizes gas phase aluminum species from the beginning and produces a highly controllable, extremely homogeneous aluminide diffusion layer and the coating process is also a ‘non-line of sight’ process. There exist subtle differences between all the above three aluminizing processes. The pack process involves submersing the parts to be processed in a powdered pack. In out of pack process, solid powders are required to produce aluminum in gaseous form, but those are not in contact with the part to be aluminized. The powders are made to react to produce the gas, which is carried forward to the part to be coated. However, for CVD aluminizing, no powder is required to generate the gas; also the gas has a flow/pressure characteristics that is suitable for coating internal surfaces of the parts too. Additional advantage of CVD aluminizing is that it has more throwing power and produces uniform coatings on the substrates. Apart from different aluminizing processes described above, another diffusion coating process, such as aluminum and nickel plating followed by thermal diffusion under protective atmosphere and oxidation in air, was also considered as a cost effective alternative surface alloying technique to form aluminide coating layers on stainless steel substrate.
3.
Characterization of coatings
3.1.
Plasma sprayed alumina
Both SS 446 and E-Brite coupons coated with 6 mil thick alumina were used for investigation of Cr volatilization rate. Typical cross sections of SS 446 coupons in as-coated and after the Cr volatilization test are shown in Fig. 2. Alumina coating appears to have become slightly denser when subjected to heating at 800 C during the Cr volatilization test. As such there was no evidence of spalling observed after the Cr volatilization test (Fig. 2). However, spalling was observed occasionally after thermal cycling to 750 C from RT repeated 3 times in a separate test (Fig. 3).
3.2. (PSS)
Plasma sprayed zirconia (PSZ) and MgeAl spinel
Spallation of PSA coating prompted investigation of plasma spraying of zirconia (PSZ) and MgeAl Spinel (PSS) because thermal expansion coefficients of these materials are somewhat closer to those of ferritic stainless steels and zirconia electrolyte. Plasma spraying of these materials was done adopting exactly the same procedures that were followed for plasma sprayed alumina.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4533
Fig. 2 e Cross section of SS 446 coupon e as-coated (PSA) and after Cr volatilization test.
Ferritic stainless steel substrates coated with PSS were subjected to 10, 30, and 50 thermal cycles (TC) between 200 and 750 C.
Evidence of grain growth of E-Brite substrate during pack aluminizing is obvious. No detailed characterization of the phases was performed due to unsatisfactory coating quality.
3.3.
3.3.2.
Diffusion coating processes
Slurry aluminizing (vapor phase)
Bare E-Brite substrates were pack aluminized using both low and high activity processes. However, the coating quality was unsatisfactory irrespective of the processes used due to presence of cracks, voids and entrapment of contaminants (Fig. 5).
Slurry aluminizing of 12 mm Ni plated E-Brite substrate produced satisfactory coating microstructure as shown in Fig. 6. EPMA of the coating layer was done to characterize the phases formed in as-coated condition (Fig. 7). As-coated samples, which were aged at 800 C for 4 h, showed very low level of Cr up to 30 mm depth of the coating layer (Fig. 7), indicating that the aluminide coating layer might act as an effective barrier layer to Cr migration. In addition, it was expected that the aluminide coating when oxidized would produce a thin aluminum oxide layer on the top surface that might further reinforce the Cr barrier properties of the intermetallic coating. Therefore, aluminide coated coupons were further oxidized for 48 h at 800 C. Microscopic observations revealed that the voids at the coating/substrate interface became more abundant after oxidation (Fig. 8).
Fig. 3 e SS 446 e spalling of PSA coating after thermal cycling.
Fig. 4 e SS 446 plated with Al followed by thermal diffusion.
Diffusion of Al into SS 446 stainless steels does not produce satisfactory coating microstructure due to nitrogen being present in the steel, producing acicular AlN phase (Fig. 4) which is extremely brittle. This phase has been reported to promote spallation of brittle intermetallic as-deposited coatings in high N containing (790 ppmw) commercial 304 L stainless steel [7]. As a result, further diffusion coatings were done using E-Brite substrates only, bare or Ni plated.
3.3.1.
Pack aluminizing of E-Brite steels
4534
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 5 e Pack aluminized bare E-Brite substrate e presence of cracks, voids and contaminants (marker length e 50 mm).
Fig. 7 e EPMA data of as-coated sample.
It seems that the oxidation temperature of 800 C was too low for formation of significant a-alumina. Oxidation of NiAl and FeAl has been extensively discussed in many literature related to alumina forming high temperature alloys [8e10]. According to Grabke [8] presence of Cr, Y and Ce etc. in NiAl helps nucleation of a-alumina at around 1000 C while suppressing formation of transition g or qalumina which is metastable oxides. Brumm and Grabke proposed that the presence of Cr accelerates the transformation to a-alumina through transient Cr2O3, which is isostructural with a-alumina, providing nucleation sites [11]. In this case, most probably the Cr content in the coating layer (Fig. 7) was not high enough for nucleation of sufficient a-alumina during oxidation of NiAl coating at 800 C (Fig. 9).
Fig. 11 displays the surface XRD pattern with peak identification for as-coated VPA E-Brite. After comparing with the diffraction pattern of uncoated E-Brite, it was concluded that the coating phase was most likely FeAl [12]. Surface X ray diffraction of oxidized VPA coated E-Brite, bare or Ni plated, failed to detect presence of any a-alumina probably due to low oxidation temperature (800 C) as discussed earlier.
3.3.3.
“Out of pack” aluminizing (vapor phase)
Both Ni plated as well as bare E-Brite substrates were aluminized using this process (Fig. 10).
3.3.4.
CVD aluminizing
CVD aluminized E-Brite coupon showed that the outer layer consists of FeAl intermetallic with about 10 mol% chromium solubility (Fig. 12). FeAl phase extends up to about 60 mm towards the interior of the steel substrates from the outer surfaces. The predominance of the cracks extending from the outer layer to the interior of the substrates or voids at the substrate/coating interface was less common in CVD aluminized samples compared to samples from other diffusion coating processes.
3.3.5.
Ni and Al plating followed by thermal diffusion
By adjusting the Ni to Al plating thickness ratio, it was possible to form mostly NiAl type aluminide (Figs. 13 and 14) after thermal diffusion process. It was observed that there was very little Cr uptake in NiAl layer (Fig. 14) even after 48 h oxidation at 800 C. This indicates that NiAl coating has the potential of functioning as a Cr barrier layer. However, this intermetallic coating was prone to cracking (Fig. 13) due to its inherent brittleness.
4.
Fig. 6 e E-Brite with 12 mm Ni plating after diffusion coating (marker length e 50 mm).
Chromium volatilization tests
A few types of coating, after initial evaluation of their microstructures, were tested for their effectiveness as barrier to the Cr volatilization. Diffusion coated coupons were oxidized at 800 C for 48 h before the Cr volatilization test in order to stabilize their microstructures and help formation of alumina layer, if any, on the top surface.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4535
Fig. 8 e Voids at coating/substrate interface.
The apparatus as shown in Fig. 15 used for measurement of the chromium volatilization rate was set up according to the transpiration method followed by Gindorf, Singheister and Hilpert [13]. The carrier gas used for this experiment was commercial air from a cylinder passing through a flow rate controller and humidifier set to a dew point of 17.5 C. This dew point setting would load the air with water vapor equivalent to
pH2O ¼ 0.02 atm (RH 60% at room temperature). The chromium volatilization rate was determined in the regime where it is independent of the flow rate of the carrier gas, defined as the “non-equilibrium” regime by Gindorf et al. [13]. The test conditions were as under: Temperature: 700, 750 and 800 C pH2O: 0.02 atm
Fig. 9 e Surface XRD of oxidized coating (48 h at 800 C).
4536
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 10 e VPA process e Ni plated (left) and bare (right) E-Brite in as-coated condition (marker length e 50 mm).
Flow rate: 4 std. liter per min. Duration: 48 h The quantity of Cr volatilized from the coupons was determined using the Inductively Coupled Plasma Mass Spectroscopy (ICPMS) technique. Chromium volatilization rates from the substrate materials (ferritic stainless steels) with respect to various barrier coatings at specific exposure conditions are shown in Table 2. Chromium retention criteria (calculated in %) are also included in Table 2. From the above data it is evident that thermal cycling of substrates with PSS coatings has adverse effects on chromium retention. The stability and efficiency of all the barrier coatings investigated in this study need to be evaluated by subjecting them to prolonged thermal cycling. Another crucial area that needs to be investigated is, whether the growth and
formation of voids at the coating/substrate interface for diffusional NiAl type coating can be an issue, particularly if the coatings are subjected to SOFC operating temperature over a longer duration.
5.
Discussions and conclusions
Chromium volatilization from high chromium ferritic stainless steels degrades the efficiencies of solid oxide fuel cell system by poisoning the cathodes. Design of an efficient barrier layer to prevent chromium transport requires consideration of several important factors such as: diffusivity of chromium through the barrier layer at the operating environment, thermal expansion mismatch between the barrier material and the steel substrates, and the nature of interface between the barrier coating and the steel.
Fig. 11 e Surface XRD pattern with peak identification for as-coated VPA E-Brite.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4537
Fig. 12 e CVD aluminized E-Brite substrate.
It was concluded that aluminides are good barriers against chromium evaporation. Two types of barrier coatings, characterized by their interfaces, are developed and described in this paper. They are non-diffusional coatings where the interfaces between the barrier layers and the substrates are discrete and diffusional coatings where the interfaces are continuous and the chemical composition gradually changes between the barrier layer and the substrate. Coatings produced by both processes have their merits and demerits. For non-diffusional process (e.g. plasma spray), CTE mismatch is a major issue and cause spallation of the barrier layer. However, in diffusional coating process (e.g. aluminizing), the CTE mismatch at the coating/substrate interface is more gradual and may not appear to be a major issue in respect of coating adherence. But creation of voids at or near
the interface during prolonged exposure to high temperature may contribute to mechanical degradation of the barrier layer. As described earlier, the interfaces between the steel and the plasma sprayed layers are discrete and not diffusionally bonded. In this work it was shown conclusively that the PSA layer, although has very low diffusivity of chromium, shows poor survivability under thermal cycling during cell operating environments caused by spallation of the layer. This is primarily attributed to a large thermal expansion mismatch and discrete nature of the interface. Of all types of plasma sprayed barrier layer studied, PSZ showed the worst barrier property, presumably due to the higher diffusivity of chromium species in zirconia. Indication of higher chromium retention values for PSS was obtained from the transpiration tests. The chromium retention value for long term aged
Fig. 13 e BSE image of the coating after oxidation for 48 h.
Fig. 14 e EMPA profile of the coating after oxidation.
4538
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 15 e The Cr volatilization test set up. (1500 h at 750 C) steel sample with PSS barrier layer is higher than those of thermally cycled (10, 30, and 50 cycles) steel samples with PSS barrier layer due to cracking of the PSS layer during thermal cycling. From various diffusion coating trials, it was concluded that AISI 446 type ferritic stainless steel does not produce a good quality aluminide coating microstructure owing to high nitrogen content in the steel. Both nickel aluminide (NiAl) prepared by slurry coating as well as vapor phase aluminizing
(VPA) and iron aluminide (FeAl) barrier coatings prepared by VPA from E-Brite stainless steel substrate were of reasonably good quality. Both types of coatings showed excellent chromium retention characteristics. The aluminides are considered to be the source of aluminum supply for creation of alumina barrier layer (more particularly, a-alumina) on top of the aluminides during high temperature oxidation of these aluminides. It is presumed that continuous formation of the alumina layer would proceed during the high temperature
Table 2 e Chromium Volatilization Rate for Various Barrier Coatings. Cr Volatilization Test Data (pH2O ¼ 0.02 atm) Alloy
Condition
Time (h)
Temp ( C)
Cr Vol Rate (Kg/m2/s)
446 446 446 446 446 446 446 446 446 446 446
Bare PSA coating Bare PSA coating Bare PSA coating PSZ Coating PSS Coating PSS Coating 10 TC PSS Coating 30 TC PSS Coating 50 TC
48 48 48 48 48 48 48 48 48 48 48
700 700 750 750 800 800 750 750 750 750 750
8.41927E-11 3.63305E-12 2.72726E-10 1.04107E-11 4.01078E-10 6.33643E-11 8.00942E-11 7.61024E-12 4.04302E-11 6.22697E-11 7.1266E-11
E-brite E-brite E-brite E-brite
Bare NiAl (Slurry coating) NiAl (Out of Pack) FeAl (Out of Pack)
48 48 48 48
750 750 750 750
1.94605E-10 1.51994E-11 4.72571E-12 9.63688E-12
Cr Ret (%)
Remarks
95.68
125 mm
96.18
125 mm
84.20 70.63 97.21 85.18 77.17 73.87
125 mm 125 mm 125 mm 125 mm 125 mm 125 mm
92.19 97.57 95.05
After Oxidation After Oxidation After oxidation
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
exposure of the aluminide coating in fuel cell operation. In this investigation attempts were made to form alumina barrier layers by heat treating the aluminide coated coupons at 800 C for 48 h. But due to low heat treatment temperature, only a very thin layer of a-alumina was formed. It is postulated that lack of enough chromium in the aluminide layer might fail to provide selective oxidation process for formation of aalumina [11]. It remains to be investigated if a prior heat treatment of aluminide coating carried out ex-situ at a higher temperature (w1000 C) would produce a thicker and denser a-alumina top layer that would continue to remain adherent after thermal cycling as encountered in SOFC operation and provide additional barrier to Cr migration. Whatever the case may be, it is concluded in this study that sufficiently thick aalumina layer on top of the aluminides would certainly enhance the barrier characteristics of such layers. Most interesting observation of this investigation with NiAl barrier layer was very low chromium solubility in as-coated condition, which indicates it is a potential barrier layer for chromium migration at high temperature even without the top a-alumina layer.
[3] [4]
[5]
[6]
[7]
[8] [9]
[10]
references
[1] Ebbinghaus BB. Thermodynamics of gas phase chromium species: the chromium oxides, the chromium oxyhydroxides and volatility calculations in waste incineration processes. Combustion and Flame 1993;93:119e1137. [2] Hilpert K, Das D, Miller M, Peck DH, Weiss R. Chromium vapor species over solid oxide fuel cell
[11]
[12] [13]
4539
interconnect materials and their potential for degradation processes. Journal of the Electrochemical Society 1996;143(11):3642e7. Allegheny ludlum technical data blue sheet e E-Brite alloy for solid oxide fuel cells. Touloukian YS, Kirby RK, Taylor RE, Lee TYR. Thermal expansion of non-metallic solids in "thermo physical properties of matter", vol. 13. New York, NY: Plenum Co.; 1977. p. 176. Moya F, Moya EG, Sami A, Juve D, Treheux D, Grattepain C. Diffusion of chromium in alumina single crystals. Phil Magazine A 1995;72:861e70. Roy S, Wusirika Raja. Modeling reactive diffusion of Cr through aluminum layer on 446 steel, Private communication dated September 05, 2003. Zhang Y, Pint BA, Cooley KM, Haynes JA. Effect of nitrogen on the formation and oxidation behavior of iron aluminide coatings. Surface and Coatings Technology 2005;200: 1231e5. Grabke HJ. Oxidation of NiAl and FeAl. Intermetallics 1999;7: 1153e8. Zhang XF, Thaidigsmann K, Ager J, Hou PY. Alumina scale development on iron aluminides. Journal of Materials Research June 2006;21(6):1409e19. Birks N, Meier GH, Pettit FS. Introduction to the high temperature oxidation of metals. Cambridge University Press; 2006. p. 122e6. Brumm MW, Grabke HJ. The oxidation behaviour of NiAl-I. Phase transformations in the alumina scale during oxidation of NiAl and NiAl-Cr alloys. Corrosion Science 1992; 33:1677. Wheaton Brian. XRD results of VPA coated E-Brite, Private communication dated November 26, 2007. Gindorf C, Singheister L, Hilpert K. Chromium vaporization from FeeCr base alloys used as interconnect in fuel cell. Steel Research 2001;72(11e12):528e33.
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/he
Development of chromium barrier coatings for solid oxide fuel cells Dilip Chatterjee a,b,*, Samir Biswas a a b
Corning Incorporated, Corning, NY 14831, USA Director of Research, Oxane Materials Inc, 467 W 38th Street, Houston, TX 77018, USA
article info
abstract
Article history:
The performance of electrolyte supported solid oxide fuel cell is impaired primarily due to
Received 22 December 2009
poisoning of electrodes due to contaminants generated from the metallic components of
Received in revised form
the stack.
30 March 2010
Ferritic stainless steels are commonly used as stack material under severe operating
Accepted 18 April 2010
conditions of SOFC environment. However, the high chromium content in this type of
Available online 3 June 2010
steels tends to form gaseous oxides and/or hydroxides which volatilize and condense on various components of stack assembly, particularly cathodes, resulting in performance
Keywords:
degradation of the system.
SOFC
Two types of barrier coatings have been developed to minimize the chromium volatiliza-
Cathode poisoning
tion. In one case, coatings of oxide species were deposited by processes such as thermal
Chromium volatilization
and plasma spraying, and the other is by diffusion coating process such as aluminizing.
Diffusion coating
This presentation will describe various barrier coatings, barrier properties provided by the
Aluminide
coatings, and transpiration measurements adopted to evaluate the efficiency of those coatings.
Plasma spray
Copyright ª 2010, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.
Spinel
1.
Introduction
In electrolyte supported solid oxide fuel cell (SOFC) stack, a frame capable of withstanding SOFC operating environments is required to support the solid oxide electrolyte sheet of yttria-stabilized zirconia (YSZ). Various types of ferritic stainless steel such as AISI 446, 430 and E-Brite (Allegheny Ludlum) were under evaluation as potential candidate frame material (Fig. 1). These steels posses the coefficient of thermal expansion in the range of 11.2e11.6 ppm/ C, which is compatible with that of the 3 mol% YSZ electrolyte, in addition to their stability in both oxidizing and reducing environments at the operating temperature around 750 C. However, all these steels mentioned above have chromium content in the range of 18e27 wt.%. All types of stainless steels form Cr2O3 during oxidation at elevated temperature which acts as a base
protective scale. But under fuel cell operating conditions of high humidity and elevated temperature, Cr2O3 scale is oxidized to gaseous species such as CrO3 or CrO2(OH)2), through the following reactions: 2Cr þ 1.5O2 (g) ¼ Cr2O3 (s)
Cr2O3 (s) þ 1.5O2 (g) ¼ 2CrO3 (g)
Cr2O3 (s) þ 2H2O (g) þ 3/2O2 (g) ¼ 2 CrO2(OH)2 (g)
According to Ebbinghaus’ calculated and measured thermodynamic data [1], CrO2(OH)2 (g) is the most dominant
* Corresponding author. E-mail address: [email protected] (D. Chatterjee). 0360-3199/$ e see front matter Copyright ª 2010, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.ijhydene.2010.04.114
4531
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 1 e A typical electrolyte supported SOFC stack assembly.
volatile component at moderate temperature in the presence of O2 and H2O such as in the fuel cell operating environment. However, the above species are reduced at the cathode side causing degradation of the cell performance known as “cathode poisoning” according to the following reactions [2].
2CrO2(OH)2 (g) þ 6e / Cr2O3 (s) þ 2H2O (g) þ 3O2
2CrO3 (g) þ 6e / Cr2O3 (s) þ 3O2 In cathode poisoning, gaseous CrO3 (g) and CrO2 (OH)2 (g) deposit onto the cathode as solid Cr2O3 (s) and degrades SOFC performances over time. Abatement of chromium evaporation from ferritic stainless steels by surface modification of the frame material or by developing a barrier layer is of significant interest in SOFC development. An effective barrier layer for electrolyte supported SOFC system should possess a few ideal attributes, such as (a) minimum mismatch of thermal expansion coefficients (CTE) between the glass seal materials and the frame materials; (b) electrically non-conductive; and (c) low diffusivity of chromium in it. This paper summarizes approaches in development of suitable barrier coatings and evaluation of their effectiveness.
2. Approach and details of various coating processes Considering the advantages and disadvantages of various candidate materials for coating on ferritic stainless steel, it was determined that alumina and/or other mixed oxides (spinel) would meet the criteria of an effective barrier coating on ferritic stainless steels, except for its considerable mismatch of CTE with stainless steels (as noted below in Table 1).
Alumina is dielectric, and dense. It possesses low diffusivity of chromium at high temperature (estimated w1022 m2/s at 800 C) [5]). Another important parameter for the barrier coating development is determination of its optimum thickness for suppression of the volatile species migration through the barrier layer. Theoretical calculations [6] predicted that alumina barrier layer on AISI 446 ferritic stainless steel should be at least in the range of 20 mm thickness to ensure chrome concentration to be well below 5 wt.% at the outer surface of the barrier layer after five years (w 40,000 h) of fuel cell operation at 800 C. Depending on the methods for application of barrier coating on stainless steel substrate and the desired coating thickness, the coating applications were classified as (a) direct deposition of alumina or other mixed oxides, such as spinel and (b) in-situ formation of thermally grown alumina from diffusion coated substrate as described below.
2.1. Direct deposition of alumina or other oxides as Cr barrier layer on ferritic stainless steels Thermal and Plasma Spray e These are process in which metallic and nonmetallic materials are deposited in a molten or semi-molten state on a prepared substrate imparting properties that the substrate would not otherwise possess. In these methods either an electric or gas source (for thermal spray) or plasma sources (for plasma spray) helps melting alumina powder feedstock. As the molten powder particles impinge with high velocities on to a ferritic stainless steel substrate or on to a SOFC frame, they form a dense coating of
Table 1 e Coefficient of thermal expansion. Material Ferritic Stainless Steele(E-Brite) Alumina a from reference [3]. b from reference [4].
CTE at 750 C (ppm/ C) 11.7a 8.2b
4532
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
oxides (alumina, zirconia or other mixed oxides) on steel. Plasma spraying of alumina and other oxides (Zirconia or spinel) on a number of ferritic stainless steels coupons and SOFC frames was investigated. The details of barrier layer microstructures, its advantages and disadvantages are described in a later section.
2.2. Diffusion coatings for in-situ formation of alumina (thermally grown oxides) as Cr barrier layer on ferritic stainless steels In this approach, intermetallic coatings with alumina forming characteristics are formed through a high temperature diffusion process known as aluminizing, whereby aluminum species diffuse into the surface of the base metal forming new metallurgical aluminide alloys (depending on process temperature, reaction time, and aluminum activity of the starting aluminum precursors). The aluminide intermetallics formed on the surface typically contain w 20% (by weight) aluminum which acts a reservoir of aluminum. Under thermodynamically suitable conditions, the coating layer is expected to form thermally grown oxide in-situ when the part is subjected to high temperature in service. Typical depth of the aluminide layer forming on stainless alloy is in the range of about 35e150 mm. There are different variations of the aluminizing process such as: a) b) c) d)
Pack aluminizing Slurry aluminizing (vapor phase) Out of pack aluminizing (vapor phase) CVD aluminizing
Pack aluminizing e the parts to be aluminized are put inside a pack consisting of aluminum halide vapor producing precursors (e.g. mixture of an aluminum source material such as powdered aluminum and or aluminum chloride, an inert material such as alumina, and an activator such as ammonium chloride) and heated at high temperatures. During the process aluminum melts, reacts with the halide to form aluminum halide gas, the gas contacts the part surfaces and is reduced and the aluminum reacts on the surfaces. Slurry aluminizing (vapor phase) e In this process slurries comprising of an aluminum alloy pigment and a halide compound in an organic binder is sprayed on the substrate. When this slurry is heated, the halide reacts with aluminum in the alloy pigment to form aluminum halide vapors. These vapors migrate through the slurry layer to the substrate where they react to release aluminum. The parts are subsequently annealed at high temperatures to diffuse aluminum into the metal surface. Out of Pack aluminizing e Out of Pack aluminizing process is essentially same as that of pack aluminizing, however, in this case the aluminum vapor is generated in-situ. The pack material (aluminum source material, activator, and inert filler, if required) described above are placed into a special multichamber reactor that has gas flow tubes to interconnect the chambers and to provide aluminum gas to the parts to be aluminized. The chambers are heated and an aluminum rich gas forms and the balance of the process is essentially as for the pack aluminizing.
The advantages of ‘out of pack’ aluminizing over ‘pack aluminizing’ include is the cleanliness of the process and aluminized coatings formed by this process are contaminant free. This process is capable of producing aluminized layers all over the part, including internal cavities (i.e. non-line of sight process). CVD aluminizing e This process offers a more environmentally friendly, better forming alternative to older pack and out of pack aluminizing technologies. This process utilizes gas phase aluminum species from the beginning and produces a highly controllable, extremely homogeneous aluminide diffusion layer and the coating process is also a ‘non-line of sight’ process. There exist subtle differences between all the above three aluminizing processes. The pack process involves submersing the parts to be processed in a powdered pack. In out of pack process, solid powders are required to produce aluminum in gaseous form, but those are not in contact with the part to be aluminized. The powders are made to react to produce the gas, which is carried forward to the part to be coated. However, for CVD aluminizing, no powder is required to generate the gas; also the gas has a flow/pressure characteristics that is suitable for coating internal surfaces of the parts too. Additional advantage of CVD aluminizing is that it has more throwing power and produces uniform coatings on the substrates. Apart from different aluminizing processes described above, another diffusion coating process, such as aluminum and nickel plating followed by thermal diffusion under protective atmosphere and oxidation in air, was also considered as a cost effective alternative surface alloying technique to form aluminide coating layers on stainless steel substrate.
3.
Characterization of coatings
3.1.
Plasma sprayed alumina
Both SS 446 and E-Brite coupons coated with 6 mil thick alumina were used for investigation of Cr volatilization rate. Typical cross sections of SS 446 coupons in as-coated and after the Cr volatilization test are shown in Fig. 2. Alumina coating appears to have become slightly denser when subjected to heating at 800 C during the Cr volatilization test. As such there was no evidence of spalling observed after the Cr volatilization test (Fig. 2). However, spalling was observed occasionally after thermal cycling to 750 C from RT repeated 3 times in a separate test (Fig. 3).
3.2. (PSS)
Plasma sprayed zirconia (PSZ) and MgeAl spinel
Spallation of PSA coating prompted investigation of plasma spraying of zirconia (PSZ) and MgeAl Spinel (PSS) because thermal expansion coefficients of these materials are somewhat closer to those of ferritic stainless steels and zirconia electrolyte. Plasma spraying of these materials was done adopting exactly the same procedures that were followed for plasma sprayed alumina.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4533
Fig. 2 e Cross section of SS 446 coupon e as-coated (PSA) and after Cr volatilization test.
Ferritic stainless steel substrates coated with PSS were subjected to 10, 30, and 50 thermal cycles (TC) between 200 and 750 C.
Evidence of grain growth of E-Brite substrate during pack aluminizing is obvious. No detailed characterization of the phases was performed due to unsatisfactory coating quality.
3.3.
3.3.2.
Diffusion coating processes
Slurry aluminizing (vapor phase)
Bare E-Brite substrates were pack aluminized using both low and high activity processes. However, the coating quality was unsatisfactory irrespective of the processes used due to presence of cracks, voids and entrapment of contaminants (Fig. 5).
Slurry aluminizing of 12 mm Ni plated E-Brite substrate produced satisfactory coating microstructure as shown in Fig. 6. EPMA of the coating layer was done to characterize the phases formed in as-coated condition (Fig. 7). As-coated samples, which were aged at 800 C for 4 h, showed very low level of Cr up to 30 mm depth of the coating layer (Fig. 7), indicating that the aluminide coating layer might act as an effective barrier layer to Cr migration. In addition, it was expected that the aluminide coating when oxidized would produce a thin aluminum oxide layer on the top surface that might further reinforce the Cr barrier properties of the intermetallic coating. Therefore, aluminide coated coupons were further oxidized for 48 h at 800 C. Microscopic observations revealed that the voids at the coating/substrate interface became more abundant after oxidation (Fig. 8).
Fig. 3 e SS 446 e spalling of PSA coating after thermal cycling.
Fig. 4 e SS 446 plated with Al followed by thermal diffusion.
Diffusion of Al into SS 446 stainless steels does not produce satisfactory coating microstructure due to nitrogen being present in the steel, producing acicular AlN phase (Fig. 4) which is extremely brittle. This phase has been reported to promote spallation of brittle intermetallic as-deposited coatings in high N containing (790 ppmw) commercial 304 L stainless steel [7]. As a result, further diffusion coatings were done using E-Brite substrates only, bare or Ni plated.
3.3.1.
Pack aluminizing of E-Brite steels
4534
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 5 e Pack aluminized bare E-Brite substrate e presence of cracks, voids and contaminants (marker length e 50 mm).
Fig. 7 e EPMA data of as-coated sample.
It seems that the oxidation temperature of 800 C was too low for formation of significant a-alumina. Oxidation of NiAl and FeAl has been extensively discussed in many literature related to alumina forming high temperature alloys [8e10]. According to Grabke [8] presence of Cr, Y and Ce etc. in NiAl helps nucleation of a-alumina at around 1000 C while suppressing formation of transition g or qalumina which is metastable oxides. Brumm and Grabke proposed that the presence of Cr accelerates the transformation to a-alumina through transient Cr2O3, which is isostructural with a-alumina, providing nucleation sites [11]. In this case, most probably the Cr content in the coating layer (Fig. 7) was not high enough for nucleation of sufficient a-alumina during oxidation of NiAl coating at 800 C (Fig. 9).
Fig. 11 displays the surface XRD pattern with peak identification for as-coated VPA E-Brite. After comparing with the diffraction pattern of uncoated E-Brite, it was concluded that the coating phase was most likely FeAl [12]. Surface X ray diffraction of oxidized VPA coated E-Brite, bare or Ni plated, failed to detect presence of any a-alumina probably due to low oxidation temperature (800 C) as discussed earlier.
3.3.3.
“Out of pack” aluminizing (vapor phase)
Both Ni plated as well as bare E-Brite substrates were aluminized using this process (Fig. 10).
3.3.4.
CVD aluminizing
CVD aluminized E-Brite coupon showed that the outer layer consists of FeAl intermetallic with about 10 mol% chromium solubility (Fig. 12). FeAl phase extends up to about 60 mm towards the interior of the steel substrates from the outer surfaces. The predominance of the cracks extending from the outer layer to the interior of the substrates or voids at the substrate/coating interface was less common in CVD aluminized samples compared to samples from other diffusion coating processes.
3.3.5.
Ni and Al plating followed by thermal diffusion
By adjusting the Ni to Al plating thickness ratio, it was possible to form mostly NiAl type aluminide (Figs. 13 and 14) after thermal diffusion process. It was observed that there was very little Cr uptake in NiAl layer (Fig. 14) even after 48 h oxidation at 800 C. This indicates that NiAl coating has the potential of functioning as a Cr barrier layer. However, this intermetallic coating was prone to cracking (Fig. 13) due to its inherent brittleness.
4.
Fig. 6 e E-Brite with 12 mm Ni plating after diffusion coating (marker length e 50 mm).
Chromium volatilization tests
A few types of coating, after initial evaluation of their microstructures, were tested for their effectiveness as barrier to the Cr volatilization. Diffusion coated coupons were oxidized at 800 C for 48 h before the Cr volatilization test in order to stabilize their microstructures and help formation of alumina layer, if any, on the top surface.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4535
Fig. 8 e Voids at coating/substrate interface.
The apparatus as shown in Fig. 15 used for measurement of the chromium volatilization rate was set up according to the transpiration method followed by Gindorf, Singheister and Hilpert [13]. The carrier gas used for this experiment was commercial air from a cylinder passing through a flow rate controller and humidifier set to a dew point of 17.5 C. This dew point setting would load the air with water vapor equivalent to
pH2O ¼ 0.02 atm (RH 60% at room temperature). The chromium volatilization rate was determined in the regime where it is independent of the flow rate of the carrier gas, defined as the “non-equilibrium” regime by Gindorf et al. [13]. The test conditions were as under: Temperature: 700, 750 and 800 C pH2O: 0.02 atm
Fig. 9 e Surface XRD of oxidized coating (48 h at 800 C).
4536
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 10 e VPA process e Ni plated (left) and bare (right) E-Brite in as-coated condition (marker length e 50 mm).
Flow rate: 4 std. liter per min. Duration: 48 h The quantity of Cr volatilized from the coupons was determined using the Inductively Coupled Plasma Mass Spectroscopy (ICPMS) technique. Chromium volatilization rates from the substrate materials (ferritic stainless steels) with respect to various barrier coatings at specific exposure conditions are shown in Table 2. Chromium retention criteria (calculated in %) are also included in Table 2. From the above data it is evident that thermal cycling of substrates with PSS coatings has adverse effects on chromium retention. The stability and efficiency of all the barrier coatings investigated in this study need to be evaluated by subjecting them to prolonged thermal cycling. Another crucial area that needs to be investigated is, whether the growth and
formation of voids at the coating/substrate interface for diffusional NiAl type coating can be an issue, particularly if the coatings are subjected to SOFC operating temperature over a longer duration.
5.
Discussions and conclusions
Chromium volatilization from high chromium ferritic stainless steels degrades the efficiencies of solid oxide fuel cell system by poisoning the cathodes. Design of an efficient barrier layer to prevent chromium transport requires consideration of several important factors such as: diffusivity of chromium through the barrier layer at the operating environment, thermal expansion mismatch between the barrier material and the steel substrates, and the nature of interface between the barrier coating and the steel.
Fig. 11 e Surface XRD pattern with peak identification for as-coated VPA E-Brite.
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
4537
Fig. 12 e CVD aluminized E-Brite substrate.
It was concluded that aluminides are good barriers against chromium evaporation. Two types of barrier coatings, characterized by their interfaces, are developed and described in this paper. They are non-diffusional coatings where the interfaces between the barrier layers and the substrates are discrete and diffusional coatings where the interfaces are continuous and the chemical composition gradually changes between the barrier layer and the substrate. Coatings produced by both processes have their merits and demerits. For non-diffusional process (e.g. plasma spray), CTE mismatch is a major issue and cause spallation of the barrier layer. However, in diffusional coating process (e.g. aluminizing), the CTE mismatch at the coating/substrate interface is more gradual and may not appear to be a major issue in respect of coating adherence. But creation of voids at or near
the interface during prolonged exposure to high temperature may contribute to mechanical degradation of the barrier layer. As described earlier, the interfaces between the steel and the plasma sprayed layers are discrete and not diffusionally bonded. In this work it was shown conclusively that the PSA layer, although has very low diffusivity of chromium, shows poor survivability under thermal cycling during cell operating environments caused by spallation of the layer. This is primarily attributed to a large thermal expansion mismatch and discrete nature of the interface. Of all types of plasma sprayed barrier layer studied, PSZ showed the worst barrier property, presumably due to the higher diffusivity of chromium species in zirconia. Indication of higher chromium retention values for PSS was obtained from the transpiration tests. The chromium retention value for long term aged
Fig. 13 e BSE image of the coating after oxidation for 48 h.
Fig. 14 e EMPA profile of the coating after oxidation.
4538
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
Fig. 15 e The Cr volatilization test set up. (1500 h at 750 C) steel sample with PSS barrier layer is higher than those of thermally cycled (10, 30, and 50 cycles) steel samples with PSS barrier layer due to cracking of the PSS layer during thermal cycling. From various diffusion coating trials, it was concluded that AISI 446 type ferritic stainless steel does not produce a good quality aluminide coating microstructure owing to high nitrogen content in the steel. Both nickel aluminide (NiAl) prepared by slurry coating as well as vapor phase aluminizing
(VPA) and iron aluminide (FeAl) barrier coatings prepared by VPA from E-Brite stainless steel substrate were of reasonably good quality. Both types of coatings showed excellent chromium retention characteristics. The aluminides are considered to be the source of aluminum supply for creation of alumina barrier layer (more particularly, a-alumina) on top of the aluminides during high temperature oxidation of these aluminides. It is presumed that continuous formation of the alumina layer would proceed during the high temperature
Table 2 e Chromium Volatilization Rate for Various Barrier Coatings. Cr Volatilization Test Data (pH2O ¼ 0.02 atm) Alloy
Condition
Time (h)
Temp ( C)
Cr Vol Rate (Kg/m2/s)
446 446 446 446 446 446 446 446 446 446 446
Bare PSA coating Bare PSA coating Bare PSA coating PSZ Coating PSS Coating PSS Coating 10 TC PSS Coating 30 TC PSS Coating 50 TC
48 48 48 48 48 48 48 48 48 48 48
700 700 750 750 800 800 750 750 750 750 750
8.41927E-11 3.63305E-12 2.72726E-10 1.04107E-11 4.01078E-10 6.33643E-11 8.00942E-11 7.61024E-12 4.04302E-11 6.22697E-11 7.1266E-11
E-brite E-brite E-brite E-brite
Bare NiAl (Slurry coating) NiAl (Out of Pack) FeAl (Out of Pack)
48 48 48 48
750 750 750 750
1.94605E-10 1.51994E-11 4.72571E-12 9.63688E-12
Cr Ret (%)
Remarks
95.68
125 mm
96.18
125 mm
84.20 70.63 97.21 85.18 77.17 73.87
125 mm 125 mm 125 mm 125 mm 125 mm 125 mm
92.19 97.57 95.05
After Oxidation After Oxidation After oxidation
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 3 6 ( 2 0 1 1 ) 4 5 3 0 e4 5 3 9
exposure of the aluminide coating in fuel cell operation. In this investigation attempts were made to form alumina barrier layers by heat treating the aluminide coated coupons at 800 C for 48 h. But due to low heat treatment temperature, only a very thin layer of a-alumina was formed. It is postulated that lack of enough chromium in the aluminide layer might fail to provide selective oxidation process for formation of aalumina [11]. It remains to be investigated if a prior heat treatment of aluminide coating carried out ex-situ at a higher temperature (w1000 C) would produce a thicker and denser a-alumina top layer that would continue to remain adherent after thermal cycling as encountered in SOFC operation and provide additional barrier to Cr migration. Whatever the case may be, it is concluded in this study that sufficiently thick aalumina layer on top of the aluminides would certainly enhance the barrier characteristics of such layers. Most interesting observation of this investigation with NiAl barrier layer was very low chromium solubility in as-coated condition, which indicates it is a potential barrier layer for chromium migration at high temperature even without the top a-alumina layer.
[3] [4]
[5]
[6]
[7]
[8] [9]
[10]
references
[1] Ebbinghaus BB. Thermodynamics of gas phase chromium species: the chromium oxides, the chromium oxyhydroxides and volatility calculations in waste incineration processes. Combustion and Flame 1993;93:119e1137. [2] Hilpert K, Das D, Miller M, Peck DH, Weiss R. Chromium vapor species over solid oxide fuel cell
[11]
[12] [13]
4539
interconnect materials and their potential for degradation processes. Journal of the Electrochemical Society 1996;143(11):3642e7. Allegheny ludlum technical data blue sheet e E-Brite alloy for solid oxide fuel cells. Touloukian YS, Kirby RK, Taylor RE, Lee TYR. Thermal expansion of non-metallic solids in "thermo physical properties of matter", vol. 13. New York, NY: Plenum Co.; 1977. p. 176. Moya F, Moya EG, Sami A, Juve D, Treheux D, Grattepain C. Diffusion of chromium in alumina single crystals. Phil Magazine A 1995;72:861e70. Roy S, Wusirika Raja. Modeling reactive diffusion of Cr through aluminum layer on 446 steel, Private communication dated September 05, 2003. Zhang Y, Pint BA, Cooley KM, Haynes JA. Effect of nitrogen on the formation and oxidation behavior of iron aluminide coatings. Surface and Coatings Technology 2005;200: 1231e5. Grabke HJ. Oxidation of NiAl and FeAl. Intermetallics 1999;7: 1153e8. Zhang XF, Thaidigsmann K, Ager J, Hou PY. Alumina scale development on iron aluminides. Journal of Materials Research June 2006;21(6):1409e19. Birks N, Meier GH, Pettit FS. Introduction to the high temperature oxidation of metals. Cambridge University Press; 2006. p. 122e6. Brumm MW, Grabke HJ. The oxidation behaviour of NiAl-I. Phase transformations in the alumina scale during oxidation of NiAl and NiAl-Cr alloys. Corrosion Science 1992; 33:1677. Wheaton Brian. XRD results of VPA coated E-Brite, Private communication dated November 26, 2007. Gindorf C, Singheister L, Hilpert K. Chromium vaporization from FeeCr base alloys used as interconnect in fuel cell. Steel Research 2001;72(11e12):528e33.