This article is one of eight papers to be presented exclusively on the web as part of the January 2000 JOM-e the electronic supplement to JOM.
	The following article appears as part of JOM-e, 52 (1) (2000), http://www.tms.org/pubs/journals/JOM/0001/Das/Das-0001.html JOM is a publication of The Minerals, Metals & Materials Society

Functional Coatings: Research Summary

The Cyclic Oxidation Performance of Aluminide and Pt-Aluminide Coatings on Cast Ni-Based Superalloy CM-247

D.K. Das, Vakil Singh, and S.V. Joshi

The cyclic-oxidation performance of plain aluminide-coated and platinum-aluminide-coated CM-247 nickel-based cast superalloy was evaluated at 1,000°C, 1,100°C, and 1,200°C in atmospheric air. The cyclic-oxidation resistance of platinum-aluminide coatings depended strongly on microstructure. The presence of a single-phase PtAl₂ outer layer is highly undesirable under cyclic-oxidation conditions because of the extreme brittleness of this phase; the PtAl₂ phase must remain distributed in a matrix of more strain-tolerant NiAl phase. The bare alloy possessed reasonably good resistance against cyclic oxidation at 1,000°C; however, it underwent severe weight loss due to oxide spallation at 1,100°C and 1,200°C. The platinum-aluminide coating provided superior cyclic-oxidation protection to CM-247 at all three temperatures as compared to its plain aluminide counterpart; this superiority became more prominent as the cyclic-oxidation temperature increased. Both alumina and spinel formed during oxidation in the bare alloy and the coatings. Oxide spalling under cyclic oxidation in the coatings was thought to be a result of spinel formation as the additional oxide phase during the progress of oxidation.

INTRODUCTION

Nickel-based superalloys are extensively used as gas-turbine engine components. These components need protection against high-temperature oxidation during operation, and diffusion aluminide coatings are widely used for this purpose. In the past few decades, it has been established that modifying plain aluminide coatings with platinum brings about drastic improvements in their high-temperature oxidation resistance.^1-4 This improvement has been primarily attributed to the corresponding enhancement of the adherence of the protective oxide layer (alumina) to the coated substrate in the presence of platinum. Several possible mechanisms have been proposed to explain how the presence of platinum in aluminide coatings aids retention of the alumina scale.^1,5-8 It has been suggested that platinum improves the spalling resistance of the alumina scale by reducing the stresses in the scale through an enhanced diffusional-creep process or through enhanced grain-boundary sliding.⁵ The role of platinum in affecting the development of pegs underneath the scale to enhance its resistance to spallation during use has also been considered.^6,7

The microstructural aspects of platinum-aluminide coatings produced under various processing conditions have also been widely reported.^9-13 Das et al. have described the evolution process of these coatings on CM-247 nickel-based superalloy during aluminizing.¹³ Several studies have also been reported pertaining to the oxidation properties of both aluminide and platinum-aluminide coatings on various superalloys.^1-3,14-20 Goward et al.¹⁵ studied the degradation of high-activity aluminide coatings on superalloy MAR-M200 under isothermal oxidation at 1,200°C and found Al₂O₃ and NiAl₂O₄ (spinel) to be the principal oxidation products. Further, the study concluded that continuous depletion of aluminum from the coating causes the b-NiAl, which constitutes the primary coating phase in the unexposed condition, to transform sequentially to b-NiAl + g¢-Ni₃Al, g¢-Ni₃Al + g-Ni solid solution, and, finally, to g-Ni solid solution. Goward et al. have also observed that, with the above gradual degradation of coating phases, the coating system, from the oxidation point of view, eventually behaves in a manner characteristic of the uncoated alloy.

Recently, the various stages of degradation in a high-activity plain aluminide coating on CM-247 cast nickel-based alloy under isothermal oxidation at 1,100°C have been reported by Das et al.²¹ This study identified three distinct stages of degeneration of the coating microstructure during the 500 hour oxidation study. A three-stage degradation process of the microstructure was also observed in the corresponding high-activity platinum-aluminide coating under identical oxidation conditions. However, the nature of the microstructural changes in each of these stages and the time scales over which these changes occurred were found to be somewhat different for the two coatings.²¹

In a recent publication, Chen and Little evaluated the degradation sequence of commercial platinum-aluminide coating RT22LT on single-crystal alloy CMSX4 during exposure to isothermal oxidation at 1,100°C up to 600 h.¹⁸ Although the same sequence of phase transformation was noted in the platinum-aluminide coating as reported by Goward et al. for plain aluminide coating,¹⁵ the coating transformed to a g¢ + g structure beyond 150 h of exposure instead of the b + g¢ structure observed by Goward et al.¹⁵ Chen and Little concluded that platinum in the coating does not act as a diffusion barrier against aluminum diffusion, but diffuses into the substrate, forming brittle topologically close-packed phases rich in refractory elements, such as tungsten and rhenium. In their oxidation study on platinum-aluminide-coated IN-738 LC industrial gas turbine blades, Aurecoechea et al.¹⁷ also observed a similar nature of coating degradation.¹⁸

Despite the well-known fact that the presence of platinum in diffusion aluminide coatings drastically improves their oxidation properties, very few studies have systematically compared the high-temperature oxidation performance of plain aluminide and platinum-aluminide coatings. The study reported here involves a comparative evaluation of a high-activity platinum-aluminide coating, its plain aluminide counterpart, and the bare alloy under cycling oxidation conditions at 1,000°C, 1,100°C, and 1,200°C.

MICROSTRUCTURE

As expected,^9-11,13 the as-formed plain aluminide coating obtained using the high-activity pack possesses a three-layer structure (Figure 1). While the outer layer of the coating contains numerous tungsten-rich precipitates in a NiAl matrix, the intermediate NiAl layer is distinctly thin in these precipitates. The inner layer of the coating is the typical interdiffusion layer obtained in aluminide coatings due to the outward diffusion of nickel from the substrate.^9-11,13 The details of the coating structure and its formation mechanism have been reported elsewhere by Das et al.¹³

The platinum-aluminide coatings obtained by the two previously mentioned prior diffusion treatments are presented in Figures 2a and 2b. Both coatings show three-layer structures; however, while the platinum-rich outer layer corresponding to prior diffusion at 850°C for 0.5 h is a single-phase structure consisting of PtAl₂ (Figure 2a), the sample for the prior diffusion treatment at 1,034°C for 5 h possesses a two-phase structure of PtAl₂ in a matrix of NiAl (Figure 2b). The other two layers of both coatings are the same (i.e., the intermediate b-NiAl layer and the inner interdiffusion layer).

The structures of the outer-coating layers are a direct result of the extent of dilution of the as-plated platinum layer in the prior diffusion treatments.^13,23 It has been demonstrated that the dilution of the platinum layer caused by interdiffusion with the substrate during the diffusion treatment at 850°C for 0.5 h is only about 4%; platinum concentration in the diffusion layer is ~96 wt.%.¹³ In diffusion at 1,034°C for 5 h, on the other hand, dilution is much higher at 46% (platinum concentration of 54 wt.%). Thus, these two extreme cases of dilution of the initial platinum layer during the course of the corresponding diffusion treatment have resulted in coatings with vastly different structures of the platinum-rich outer layer. The details of the formation mechanisms of the two platinum-aluminide coatings can be found elsewhere.¹³

THE EFFECT OF COATING STRUCTURE ON OXIDATION RESISTANCE

Both platinum-aluminide coatings were subjected to conditions of cyclic oxidation. The coating corresponding to prior diffusion at 850°C for 0.5 h (Figure 2a) underwent severe spalling when subjected to thermal cycling. In fact, for the test conducted at 1,200°C, the top layer of the coating spalled over a significant portion of the surface of the sample after only one cycle of heating and cooling. Figure 3 shows the samples for this case before and after cycling; the considerable part of the surface over which the coating spalled is clearly evident from Figure 3b. The top layer of the coating detached from the sample surface cleanly in the form of flakes during the cooling period. The coating corresponding to prior diffusion at 1,034°C for 5 h did not show any such coating failure. In fact, spalling of the surface oxides that occurs in the advanced stages of oxidation in the latter case takes place uniformly as a fine powder over the entire surface of the sample. This aspect can be seen in Figure 4, which shows the as-coated sample and samples after 2.5 h and 15 h of oxidation at 1,200°C.

The PtAl₂ phase fully constituted the outer coating layer when prior diffusion was done at 850°C for 0.5 h. This intermetallic phase is known to be extremely brittle, perhaps much more so than the b-NiAl phase. Therefore, the mismatch strain between the top PtAl₂ and the underlying NiAl layer caused by the difference in their coefficients of thermal expansion during cyclic heating and cooling causes cracking and the subsequent spalling of the coating. In contrast, the coating corresponding to prior diffusion at 1,034°C for 5 h possesses an outer layer having the brittle PtAl₂ phase distributed in the more strain-tolerant NiAl matrix. Therefore, this layer does not exhibit premature failure under thermal-cycling conditions as in the case of single-phase PtAl₂.

No data exist in open literature to enable a comparison of the ductility/toughness of the PtAl₂ phase with that of the NiAl phase. However, the NiAl phase, being more strain tolerant than the former, can be concluded from the fact that failure by flaking in plain aluminide coatings primarily consisting of the NiAl phase have not been reported, as mentioned for the single-phase PtAl₂ layer. In fact, the plain aluminide coating did not show any such failure mode. This is evident in Figure 5, which shows the surface condition of the as-formed coated sample and that of the coated samples progressively withdrawn after various durations of cyclic oxidation at 1,200°C. The observed uniform condition of the surface of the plain aluminide-coated specimens after oxidation indirectly confirms the fact that the NiAl phase is considerably more strain tolerant than the PtAl₂ phase.

Platinum usually remains in the form of the PtAl₂ phase in platinum-aluminide coatings.^9-13,23 However, the coating structure with a single-phase PtAl₂ outer layer is highly undesirable from the point of view of their utility under cyclic-oxidation conditions. Therefore, the result clearly establishes the importance of adopting an appropriate prior-diffusion temperature-time schedule that primarily decides the platinum-aluminide coating structure. Ideally, the diffusion-treatment schedule should be such that just enough dilution of the plated platinum layer takes place, and, consequently, the outer layer of the coating develops the two-phase PtAl₂ + NiAl structure instead of single-phase PtAl₂. In light of these results, the platinum-aluminide coating corresponding to prior diffusion at 850°C for 0.5 h was excluded from further testing because of its questionable practical relevance on account of poor shock resistance under cyclic heating and cooling.

PHASE CONSTITUTION CHANGES AFTER OXIDATION

Bare Alloy

Subjecting the bare CM-247 alloy to cyclic oxidation at 1,000°C results in the formation of both alumina and spinel on the sample surface from the very beginning. The same two oxide phases are found throughout the 350 h period of oxidation, with their amount increasing with exposure time (Figure 6). The 5 wt.% aluminum originally present in the alloy contributes to the formation of these two oxide phases during oxidation. The loss of aluminum from close to the sample surface as the oxidation progresses causes the original g + g¢ structure of the alloy to transform to g-Ni. Thus, a layer of g forms below the oxide layer (Figure 7), with the thickness of the g-Ni layer expectedly increasing with oxidation time. No significant spalling of the oxide layer that formed on the bare alloy is observed at 1,000°C. This fact, which is also reflected in the weight change measurements, is indicative of the fact that the mismatch strains between the oxide layer and the coated substrate (due to the difference in their coefficients of thermal expansion) are not high enough to cause oxide spalling.

Alumina and spinel form even in oxidation at 1,100°C and 1,200°C. The difference, however, is that severe spalling of the oxide layer in the form of fine powder occurs due to cyclic heating and cooling. The XRD of the spalled oxide powder indicates it is mainly spinel, with virtually no alumina present. The severe spalling of oxide in these two cycling cases indicates not only the increased oxidation kinetics with an increase of oxidation temperature, but also the inability of the oxide layer to withstand the mismatch strains generated due to thermal cycling at these temperatures.

Coatings

The phase constitution of the coatings was determined using XRD. The primary phase in the as-formed plain aluminide coating, as expected, was found to be b-NiAl. Cyclic oxidation at 1,000°C resulted in the formation of Al₂O₃ on the sample surface and, as anticipated, its amount increased with exposure duration. This is evident from Figure 8, which shows the x-ray diffractograms of the plain aluminide coating corresponding to various exposure durations. Also, no oxide spalling is noticed during the entire oxidation period. It is, however, interesting to note from this figure that the bulk coating phase b-NiAl does not transform to lower aluminum-containing phases, such as g¢-Ni₃Al and g-nickel phases,¹⁵ even after 350 h of exposure. This is a clear indication of the lower rate of aluminum loss from the coating at 1,000°C due to the lack of spalling of the protective alumina layer. Thus, the plain aluminide coating appears to be fairly adequate in providing cyclic-oxidation protection to CM-247 alloy at 1,000°C.

The platinum-aluminide coating, exposed to similar cyclic-oxidation conditions, also produces an alumina scale due to oxidation. The PtAl₂ + NiAl phase structure of the outer coating layer in this case remains stable up to nearly 150 h of exposure, beyond which it transforms to single-phase NiAl (with platinum remaining in solid solution),²¹ as indicated in the diffractograms of Figure 9.

Cyclic oxidation performed at 1,100°C and 1,200°C, however, resulted not only in the transformation of the bulk coating phases, but also in the formation of spinel as the additional oxide phase in both coatings. For example, when the plain aluminide coating is thermally cycled at 1,200°C, the bulk-coating phase transforms to g¢-Ni₃Al + g-Ni, with the accompanying formation of spinel in addition to alumina by 15 h of exposure. After 50 h, the coating consists of only g-Ni because of the excessive loss of aluminum toward the formation of the oxide phases and also into the substrate via diffusion.^15,21 Further oxidation does not change the coating phase structure, and by 160 h, the oxide phase on the sample surface consists of only spinel. This transformation and oxide formation for 1,200°C cyclic oxidation are evident in diffractograms shown in Figure 10.

In a study by Goward et al., a high-activity plain aluminide coating on MAR-M200 is reported to retain b + g¢ as the bulk coating phase for as long as 200 h of isothermal oxidation at 1,200°C, although a surface layer of g is also said to be present from 86 h onward.¹⁵ As the composition of MAR-M200 is very similar to that of CM-247, it can be concluded that the loss of aluminum from coating under cyclic oxidation is much quicker than that under isothermal oxidation. This is expected because of the additional stresses experienced by the oxide layer under cyclic heating and cooling, which lead to faster oxide spallation and, hence, faster aluminum consumption from the coating.

The diffractograms corresponding to cyclic oxidation at 1,200°C for the platinum-aluminide coating are presented in Figure 11. The two-phase PtAl₂ + NiAl structure transforms to single-phase b-NiAl after only 15 h and to g¢ + g after 50 h of exposure at 1,200°C. Although alumina remains the major oxide phase over at least 50 h, spinel also forms beyond that time. By 200 h of exposure, the bulk-coating phase is transformed to g-Ni because of the loss of aluminum.

The platinum-aluminide coating structures after 15 h and 200 h of oxidation at 1,200°C are presented in Figures 12a and 12b. The transformation of the two-phase PtAl₂ + NiAl structure (Figure 2b) to single-phase NiAl (Figure 12a) and, subsequently, to g-Ni (Figure 12b) is evident. In fact, several Kirkendall porosities can be seen in the coating after 200 h, which is a clear indication of excessive loss of nickel and aluminum toward the formation of oxides (both NiAl₂O₄ and Al₂O₃). These trends are also observed for cyclic oxidation of both coatings at 1,100°C, with the times for the transformation of the bulk coating phases and the formation of spinel as the additional oxide phase being longer than those observed at 1,200°C. In fact, theoretically, one would also observe the same trends for cyclic oxidation at 1,000°C at much longer times.

THE COMPARATIVE OXIDATION PERFORMANCE OF BARE AND COATED CM-247 ALLOY

Figures 13a, 13b, and 13c present the weight change data due to cyclic oxidation for both bare alloy and coated samples tested at 1,000°C, 1,100°C, and 1,200°C. The weight change of a sample due to oxidation was determined with respect to its initial weight (i.e., by subtracting the weight prior to oxidation from the one measured after oxidation). This weight change was normalized with the initial surface of the sample to determine the specific weight change (in mg cm^-2), which has been plotted in the figure. A comparative evaluation of the two coatings was carried out with reference to the bare alloy based on the weight-change data obtained at the three temperatures.

1,000°C

Only weight gain (i.e., positive weight change) is registered over the entire 350 h period of oxidation in the bare alloy and the two coatings. Further, the weight-gain values of the bare alloy and the plain aluminide coating are quite similar; however, the values for the platinum-aluminide coating are comparatively much smaller. Positive weight change during oxidation is desirable because it is indicative of the retention of the protective oxide layer formed on the sample surface. Weight loss, on the other hand, implies that the material from the sample surface is lost in the form of oxides due to spallation of the oxide layer. As mentioned, there is virtually no loss of oxides in either the plain aluminide coating or the platinum-aluminide coating during the 350 h test period. However, the higher weight gain noted in the former coating is possibly because of the higher diffusivity of the oxygen through the oxide scale. In this context, it is relevant to mention that the presence of platinum in the coating has been reported to reduce the diffusivity of oxygen through the oxide scale.²⁴

Based on the positive weight-change data (Figure 13a), the bare CM-247 alloy appears to be fairly resistant to oxide spallation at 1,000°C despite the fact that spinel forms on the surface with alumina throughout the exposure (Figure 6). However, even if a protective coating is provided on this alloy, a plain aluminide coating appears quite adequate for this purpose for two reasons. First, the most desirable alumina protective oxide layer forms on the sample surface during oxidation (Figure 8), and this oxide layer has sufficient adherence with the coated substrate to withstand the imposed cyclic heating and cooling environment. This is clearly evident from the continuous weight gain noted during oxidation (Figure 13a). Second, the b-NiAl phase of the coating remaining untransformed over the entire exposure period (Figure 8) indicates that the aluminum loss from the coating during oxidation is very slow because of the formation of a spall-resistant alumina layer at 1,000°C. Therefore, based on the weight-change data (Figure 8) and the nature of the oxide that forms (Figure 9), the platinum-aluminide coating may warrant consideration only when very long durations of oxidation are involved, over which weight loss of the plain aluminide coating is eventually expected.

1,100°C and 1,200°C

The negative weight change (weight loss) observed in the bare alloy from the beginning of exposure at 1,100°C and 1,200°C (Figure 13b and 13c) indicates that the alloy possesses very poor resistance to cyclic oxidation at these higher temperatures. The high rate of weight loss at both temperatures also confirms the poor oxidation resistance of this alloy at these temperatures. Severe spalling of the spinel is observed during the oxidation of CM-247 at these temperatures, and this is reflected in the weight-change plots. Hence, it is clear that the superalloy needs protection at 1,100°C and 1,200°C, although it appears to be reasonably resistant to oxidation at 1,000°C.

The application of aluminide coatings enhances the cyclic-oxidation resistance of the alloy, with the platinum-aluminide coating providing superior protection as compared to the plain aluminide. For instance, while the platinum-aluminide coated sample shows weight loss after 100 h of oxidation at 1,200°C, its plain aluminide counterpart does the same after 10 h only (Figure 13c). The superiority of the platinum-aluminide coating is even more clearly demonstrated by the weight-change results obtained at 1,100°C (Figure 13b). The data show that the plain aluminide coating lasts much longer (440 h) before registering weight loss. The platinum-aluminide coating does not lose weight over the entire 1,000 h exposure at this temperature (Figure 13b).

Although there is no doubt about the superiority of the platinum-aluminide coating over the plain aluminide at all three oxidation temperatures based on the weight-change data, it must be noted that the superiority becomes more prominent as the oxidation temperature increases. The plain aluminide coating is adequate in protecting CM-247 alloy at 1,000°C, and a platinum-aluminide coating may not be required for this purpose. Similarly, at 1,100°C, the plain aluminide coating offers protection up to a considerable period of 200 h, beyond which oxide spalling begins, resulting in weight loss by about 440 h. Thus, although the platinum-aluminide coating has the capability to protect for the entire 1,000 h exposure duration, as indicated by its continuous weight gain, the plain aluminide coating can also be used for this purpose for at least 200 h. For cyclic oxidation at 1,200°C, however, the platinum-aluminide coating is absolutely necessary for protection because of the inability of the plain aluminide coating to survive even beyond 5 h. The platinum-aluminide coating under these conditions lasts for about 80 h before showing oxide spallation. By this time, the bulk coating phase transforms to g¢ + g, with the formation of spinel in addition to alumina. As the severity of the cyclic oxidation conditions increases with increasing oxidation temperature, the superiority of the platinum-aluminide coating over the plain aluminide coating in providing protection becomes more prominent.

OXIDE LAYER FAILURE DURING OXIDATION

During cyclic heating and cooling, mismatch strains are generated between the oxide layer and the coated sample due to the difference in their coefficients of thermal expansion. These strains contribute to spalling of the oxide layer during cyclic oxidation, which is reflected in the weight-change values. Continuous weight gain during the oxidation test at 1,000°C, even for the plain aluminide coating, indicates that the thermal shock created by heating and cooling between 1,000°C and room temperature was not severe enough to cause any major oxide spalling. Even the bare alloy appears to be resistant to oxide spalling for the thermal cycling (Figure 13a). The kinetics of oxidation at this temperature also appears to be quite slow for both bare and coated CM-247 alloy.

The situation, however, worsens with increasing exposure temperature during cyclic oxidation because of the corresponding increase in the mismatch strains as well as the faster rate of oxidation. The oxide layer in the plain aluminide coating appears to be adherent enough to survive about 200 h of exposure at 1,100°C, beyond which oxide spalling begins. The oxide layer surviving for 200 h at this temperature indicates that the mismatch strains are not high enough to overcome the oxide adherence in any significant way. It is over this period that alumina constitutes the oxide layer. Beyond this duration, however, the lack of adequate aluminum in the coating prevents the continuous generation of the alumina layer. Instead, the less protective spinel begins forming beyond 200 h. This oxide phase is possibly not as adherent as alumina, because of which it begins spalling under the same mismatch strains at 1,100°C. In cycling at 1,200°C, not only are the mismatch strains much higher, but the aluminum loss from the coating toward the formation surface oxide layer and the consequent change of oxide phase from alumina to spinel occur much earlier (Figure 11). As a result, the loss of oxide due to spalling in the plain aluminide coating is noticed as early as after 5 h of exposure (Figure 13c). Further, the weight loss by oxide spallation for cycling at 1,200°C occurs at a much higher rate than that for the 1,100°C cycling for the same reasons.

The improved oxidation performance of the platinum-aluminide coating observed in this study is primarily because of its ability to retain alumina as the only oxide phase and prevent spinel formation for much longer periods during oxidation than its plain aluminide counterpart. For example, while the plain aluminide coating shows alumina as the major oxide phase over 200 h for 1,100°C cyclic oxidation, the platinum-aluminide coating has this oxide layer over the entire 1,000 h exposure. Further, platinum in platinum-aluminide coatings is also known to enhance the adherence of the Al₂O₃ layer formed on the coated substrate during oxidation.^1,5-8 These two factors enable the platinum-aluminide coating to be more resistant to cyclic oxidation, especially at 1,100°C and 1,200°C, than the plain aluminide coating.

References
1. G.J. Tatlock and T.J. Hurd, Oxid. Met., 22 (1984), pp. 201-226.
2. M. Gobel, A. Rahmel, and M. Schutze, Oxid. Met., 3/4 (1993), pp. 231-261.
3. J.H. Sun, H.C. Jang, and E. Chang, Surf. Coat. Technol., 64 (1994), pp. 195-303.
4. P.C. Patanaik, R. Thamburaj, and T.S. Sudarshan, Surface Modification Technologies III, ed. T.S. Sudarshan and D.G. Bhat (Warrendale, PA: TMS, 1990), pp. 759-776.
5. J.G. Fountain et al., Oxid. Met., 10 (1976), pp. 341-345.
6. I.M. Allam, H.C. Akuezue, and D.P. Whittle, Oxid. Met., 14 (1980), pp. 517-530.
7. E.J. Felten and F.S. Pettit, Oxid. Met., 10 (1976), pp. 189-223.
8. E.J. Felten, Oxid. Met., 10 (1976), pp. 23-28.
9. R. Streiff, O. Cerclier, and D.H. Boone, Surf. Coat. Technol., 32 (1987), pp. 111-126.
10. D.K. Das and J. Annapurna, Development of Pt-Aluminide Coatings on CM-247 Ni-Base superalloy : II. A Preliminary Study on Mechanism of Coating Formation, DMRL TR 96206 (Hyderabad, India: Defence Metallurgical Research Laboratory, May 1996).
11. P.C. Pattanaik, R. Thamburaj, and T.S. Sudarshan, Surface Modification Technologies III, ed. T.S. Sudarshan and D.G. Bhat (Warrendale, PA: TMS, 1990), pp. 759-776.
12. M.R. Jackson and J.R. Rairden, Metall. Trans. A, 8A (1977), pp. 1697-1707.
13. D.K. Das, Vakil Singh, and S.V. Joshi, Metall. Mater. Trans. A (in press).
14. T.K. Redden, Trans. AIME, 242 (1968), pp. 1695-1702.
15. G.W. Goward, D.H. Boone, and C.S. Giggins, Trans. ASM, 60 (1967), pp. 228-241.
16. N.R. Lindblad, Oxid. Met., 1 (1969), pp. 143-170.
17. J.M. Aurrecoechea, L.L. Hsu, and K.G. Kubarych, Mater. Manuf. Process., 10 (1995), pp. 1037-1051.
18. J.H. Chen and J.A. Little, Surf. Coat. Technol., 92 (1997), pp. 69-77.
19. H.M. Twancy, N.M. Abbas, and T.N. Rhys-Jones, Surf. Coat. Technol., 49 (1991), pp. 1-7.
20. H.M. Twancy, N.M. Abbas, and T.N. Rhys-Jones, Surf. Coat. Technol., 54/55 (1992), pp. 1-7.
21. D.K. Das, Vakil Singh, and S.V. Joshi, Mater. Sci. Technol. (1999).
22. D.K. Das, Vakil Singh, and S.V. Joshi, Metall. and Mater. Trans. A, 29A (1998), p. 2173.
23. G. Ravi Krishna et al., Mater. Sci. Eng. A, 251A (1998), pp. 40-47.

D.K. Das and S.V. Joshi are with the Defence Metallurgical Research Laboratory. Vakil Singh is with the Department of Metallurgical Engineering, Benaras Hindu University.

For more information, contact D.K. Das, Defence Metallurgical Research Laboratory, Surface Engineering Group, Kanchanbagh, Hyderabad, 500 058 India; telephone 91-040-444-0051; fax 91-040-444-0683; e-mail dkd@dmrl.ernet.in.