Oxidation of Metals

, Volume 69, Issue 3, pp 233–247

Effect of Al on High-Temperature Oxidation of Cr–W Alloys

Authors

    • DOE National Energy Technology Laboratory
Original Paper

DOI: 10.1007/s11085-008-9095-0

Cite this article as:
Doğan, Ö.N. Oxid Met (2008) 69: 233. doi:10.1007/s11085-008-9095-0

Abstract

The effect of Al on the high temperature oxidation behavior of Cr–10 wt.%W alloy was investigated using a cyclical oxidation test at 1,000 °C in dry air. First, Al was added into the Cr–W alloy as an alloying element up to 8 weight percent. Although alloying with Al reduced the spalling, it did not eliminate it. Secondly, Al was applied to the surface using an aluminizing process. Forming an Al–Cr layer on the Cr–W alloy reduced oxidation rate significantly and eliminated spalling completely.

Keywords

Chromium–tungsten alloysHigh-temperature oxidationAluminizing

Introduction

There is an increasing demand for new materials with higher strength, creep resistance, corrosion resistance, thermal-fatigue resistance, and wear resistance in energy systems to increase generation efficiency, to reduce environmental pollution, and, in some cases, to make new energy-generating technologies economically feasible. New technologies such as ultra super-critical steam plants, integrated-gasification-combined cycle, pressurized fluidized-bed combustion, oxy-fuel turbines, and solid-oxide fuel cells are being developed to meet this demand. One common barrier in the development of these different technologies is the insufficiency of existing materials. Even the highly alloyed and high-cost Ni-base superalloys do not have the desired properties at the temperatures that some parts of the future energy generating systems will be exposed. Therefore, there is an urgent need to develop structural alloys with desirable properties using elements with high melting point (refractory metals).

Refractory metals such as tungsten, molybdenum, chromium, tantalum, and rhenium have high strength at room temperature and the decrease in strength with increasing temperature is gradual compared to the conventional structural alloys such as steels and superalloys [1]. Furthermore, the strength of chromium can be increased with alloying and other strengthening techniques [2]. While the Cr–W alloys are very attractive strength wise, they have limitations with room-temperature ductility and elevated temperature oxidation. The reasons behind the low room-temperature ductility of Cr are not very well understood. Whether low room-temperature ductility arises from intrinsic (due to high Peierls stresses in the BCC structures) or extrinsic (due to grain-boundary segregation of impurities) properties is a matter of debate. There is theoretical and experimental work supporting both points of view [312]. Oxidation resistance of some refractory alloys, e.g. Cr and Mo, is low because they form volatile oxide surface layers at elevated temperatures in oxidizing atmospheres [1, 13]. Chromium forms Cr2O3 scale when exposed to oxidizing environments at elevated temperatures; however, the scale spalls under cyclical conditions or significantly oxidized to volatile CrO3 above 800 °C [1415]. Some type of surface modification or coating technique is needed in order to use these alloys in oxidizing environments at elevated temperatures. Nitridation of Cr and its alloys in air was studied by several researchers [1417]. Recently, the effect of additions of Fe, La, and MgO to Cr was shown to be effective in reducing the extent of Cr2N formation [17].

In this study, the effect of Al on the oxidation behavior of Cr alloyed with 10 weight percent W was studied in dry air at elevated temperatures. Aluminum was applied in two ways: as an alloying addition and as a surface coating (aluminizing).

Experimental Procedure

Cr, Cr–W and Cr–W–Al ingots were prepared by melting high-purity elemental charge materials in a water-cooled copper hearth arc furnace partially filled with argon. A commercial Ni-base alloy IN 718 was used in the cyclical tests as a comparison material. Designations of the experimental alloys used in this study indicate their nominal chemical compositions in weight percent. Analyzed chemical compositions are given in Table 1. The alloys were analyzed using X-ray fluorescence (XRF) for all elements except N and O. These elements were analyzed using a combustion-gas-analysis technique. Aluminizing of Cr–10W alloys was accomplished by dipping the specimens (25 mm dia.) for 5 min in a molten aluminum bath at 750 °C held in a crucible heated by a resistance furnace. Excess aluminum on the surface of the Cr–W specimens was removed mechanically. Only the reaction layer between aluminum and the substrate was left on the surface.
Table 1

Chemical composition of the experimental alloys (in wt.%)

 

W

Al

V

Fe

O

N

Cr

Cr

nd

nd

nd

nd

0.02

0.001

Bal.

Cr–10W

10.3

nd

0.5

nd

0.02

0.001

Bal.

Cr–10W–2Al

10.5

1.5

nd

0.3

na

na

Bal.

Cr–10W–5Al

10.5

4.2

nd

0.3

na

na

Bal.

Cr–10W–8Al

10.4

7.3

nd

0.3

na

na

Bal.

na—Not analyzed; nd—Not detected

Oxidation specimens with 25 mm diameter and 2 mm thickness were cut from the ingots. The specimens had a 6.4 mm diameter hole for hanging on a holder. Surfaces of the specimens were ground using 600-grit SiC abrasive paper. After dimensions were measured, the specimens were cleaned in ethanol and weighed before being placed in the furnace.

A thermogravimetric instrument was used for the continuous oxidation experiments. After obtaining a vacuum, dry air was introduced and flowed through the specimen chamber. The flow rate was sufficiently slow so it would not affect the microbalance. The specimens were heated to 800 °C at 5 C/min. and held at 800 °C for 99 h. They were cooled at 5 C/min until this rate was too fast for the furnace to keep up. The rest of the cooling to room temperature was uncontrolled. The mass change was recorded continuously using a data acquisition system. The buoyancy effect over the temperature range which was determined by a dry run (without samples) was subtracted from the data.

Cyclical-oxidation experiments were performed in a tube furnace with a controlled atmosphere. These tests were conducted in a flow of dry air at 1,000 °C. The specimens were placed on a quartz rack in the furnace tube. The specimens were taken out of the furnace at irregular intervals and weighed on a microbalance. They were replaced in the furnace after weighing.

The oxidized specimens were characterized using optical microscopy, scanning-electron microscopy (SEM), X-ray diffraction (XRD), and microchemical analysis using wave-length-dispersive (WDX) and energy-dispersive (EDS) spectroscopy.

Results and Discussion

Base Material Microstructure

As-solidified Cr–10W alloy has a dendritic microstructure of a bcc phase. This phase is a solid solution of W in Cr. The addition of W to Cr increases the lattice spacing in bcc Cr and provides solid-solution strengthening. The addition of up to 8 weight percent Al to Cr–10W did not change the single phase character of this alloy.

Continuous Oxidation of the Cr–10W Alloy

A continuous isothermal-oxidation test was performed using a thermogravimetric instrument. A Cr–10W specimen was exposed to a slow flow of dry air at 800 °C for 100 h. The mass change was recorded continuously as a function of time, as shown in Fig. 1. A parabolic function (Eq. 1) represents the mass gain of the Cr–10W specimen well.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig1_HTML.gif
Fig. 1

Mass change of Cr–10W oxidized continuously for 100 h in dry air at 800 °C

$$ \Updelta m = 1.75\;t^{{0.5}} $$
(1)

Equation 1 represents a faster oxidation kinetics compared to the previous work summarized in an article by Caplan et al. [18]. This previous work was conducted on unalloyed Cr specimens in oxygen.

From this result, it may appear that the oxidation of the Cr–10W alloy is dominated by the growth of an outer oxide scale at 800 °C. However, microstructural analysis performed on the oxidized specimens reveal a more-complicated oxidation mechanism as discussed below. But the first, following additional observations were made: Spalling of the scale did not take place during the isothermal hold, however, the scale spalled partially during cooling to room temperature. Also, some evidence of buckling of the scale was observed in the cross section using SEM (Fig. 2), although it was not clear whether the buckling occurred during the isothermal hold or during cooling.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig2_HTML.jpg
Fig. 2

Back-scattered-electron image of cross-section of Cr–10W oxidized in dry air at 800 °C for 100 h and cooled to room temperature

XRD done on the surface of the specimen with a partial oxide scale revealed that the scale is Cr2O3 as shown in Fig. 3. In this XRD data, there was no indication of tungsten oxides or other compounds forming in the outer scale. After removing the loose outer scale completely, the specimen was subjected to X-ray diffraction analysis again. This time, Cr2N and WO3 peaks were identified in addition to the Cr2O3 peaks in the X-ray spectrum. Formation of Cr2N in Cr and Cr–Nb alloys oxidized in dry and moist air at 950–1,100 °C was previously reported [1417].
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig3_HTML.gif
Fig. 3

XRD spectra acquired on the surface of the specimens oxidized in dry air at 800 °C for 100 h and cooled to room temperature

X-ray elemental mapping of the cross-sections of the oxidized specimens in SEM revealed distribution of various elements in and underneath the scale as shown in Fig. 4. These maps show that the outermost layer of the oxidation scale contains Cr, O, and N. As the XRD results show, Cr and O are combined as Cr2O3. Since there is no nitrogen compound detected in the outer scale, the nitrogen concentration observed in the maps is thought to be the dissolved nitrogen in the oxide scale. A gap separates the outer scale from the rest of the specimen. The thickness of the gap varies, and it can be as large as 1 μm in some locations. Below the gap, a region that contains W, Cr, N, and O exists. Again XRD results obtained from the spalled specimen suggest that this is a mixed-oxide layer comprised of Cr2O3, and WO3. Below the mixed-oxide layer, nearly continuous porosity was observed. Below that, a Cr- and N-rich layer of about 10 μm thickness is shown in the elemental maps. The morphology of this phase and XRD results suggest that this layer is composed of Cr2N in a matrix of Cr–W solution.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig4_HTML.jpg
Fig. 4

SEM image and X-ray elemental maps acquired from the cross-section of Cr–10W oxidized in dry air at 800 °C for 100 h and cooled to room temperature

Examining the phase-stability diagram of the Cr–W–N–O system at 800 °C (Fig. 5) in the light of the above observations suggests that the mixed oxide layer with Cr2O3 and WO3 formed first upon exposure of fresh alloy surface to dry air. This is concluded because the diagram shows that a relatively high partial pressure of oxygen is needed to form WO3. This is only possible in the beginning of the experiment at the location where WO3 is observed. WO3 was not observed anywhere else. This also indicates that the outward diffusion of W is negligible, and inward diffusion of oxygen is not fast enough to maintain high activity below the outer scale. After the initial formation of the mixed-oxide layer, oxidation reactions take place above and below this mixed layer simultaneously. Above the initial oxidation layer, the Cr2O3 layer grows by diffusion of Cr ions from the substrate alloy to the scale surface in the absence of W atoms. Porosity below the outer Cr2O3 scale is thought to form due to cation vacancies diffusing inward to facilitate outward cation diffusion [18]. Below the initial mixed-oxide layer, low-oxygen activity and high nitrogen activity promotes Cr2N formation as suggested by the phase diagram. The widespread Cr2N formation internally near the original alloy surface (Figs. 2 and 3) and high N concentration throughout the oxide scales (Fig. 4) indicate sufficient nitrogen transport through the oxide scales and pores to maintain a high activity of nitrogen below the scales during oxidation. The transportation of nitrogen through the Cr2O3 scale and pores to maintain a high activity at the scale—alloy interface was also observed earlier [14, 15].
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig5_HTML.gif
Fig. 5

Phase-stability diagram of the Cr–W–N–O system at 800 °C as calculated using FactSage Thermochemical Software (FT53 database)

The significance of the evaporation of the oxide scale was not deduced from this experiment. However, the stability diagram of the Cr–W–N–O system (Fig. 5) suggests that oxidation of Cr2O3–CrO3 does not take place at 800 °C.

Cyclical Oxidation of Cr–W alloys

These tests were performed in a flow of dry air in a tube furnace at 1,000 °C. Figure 6 shows the mass changes of unalloyed Cr, Cr–10W, and Alloy 718 specimens. In general, all three specimens spalled resulting in a mass loss. The oxidation rate of Cr and Cr–10W was unpredictable due to spalling of the oxide scale in large pieces (Fig. 7). The effect of expected evaporation of the oxide scale at 1,000 °C was masked by the large magnitude of the mass change due to scale spallation. The mass loss in alloy 718 was more gradual due to different spallation characteristics. The oxide scale spalled from alloy 718, producing a fine powder.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig6_HTML.gif
Fig. 6

Mass change of Cr, Cr–10W, and commercial alloy 718 during cyclical oxidation in dry air at 1,000 °C

https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig7_HTML.jpg
Fig. 7

Cr–10W oxidized in dry air at 1,000 °C

Effect of Alloying with Aluminum on the Oxidation Behavior of Cr–W Alloys

In order to increase the high-temperature oxidation resistance of Cr–W alloys by promoting the formation of an alumina scale instead of a chromia scale, three different levels of Al were added to Cr–10W. The resulting alloys were tested for their oxidation resistance under the same cyclical conditions as the Cr–10W alloy. Figure 8 compares the mass change of Cr–10W to the alloys containing Al. Increasing Al content of Cr–10W alloys decreases the mass loss at 1,000 °C in dry air. Although Cr–10W–5Al and Cr–10W–8Al recorded mass gains, spallation was observed during the tests. There were two significant effects of alloying with Al on the oxidation behavior of Cr–10W: first, the oxidation resistance of these alloys improved significantly due primarily to reduced spalling. Second, the scale spalled in the form of fine particulates instead of large pieces.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig8_HTML.gif
Fig. 8

Effect of alloying with Al on the mass change of Cr–10W under cyclical conditions at 1,000 °C in dry air

Figure 9a shows a cross section of Cr–10W–8Al after exposure to dry air at 1,000 °C. The outer oxidation scale was partially spalled while the part of the remaining scale was buckled above another layer of oxide scale on the specimen. This specimen also showed an internal-oxidation layer several hundred microns thick. A closer look at the same region using back-scattered-electron imaging reveals compositionally different formations near the outer scale as shown in Fig. 9b. Microchemical analysis using WDX shows that the outer layer is (Cr,Al)2O3. Below this outer scale, there is a layer that contains two oxides of the same type with different compositions. The phase appearing bright in the back-scattered-electron image is (Cr,W)2O3. The other phase which appears gray is the same type of oxide as that in the outer scale. Below this mixed scale is a layer of Cr2N. The particles in the internal oxidation layer were primarily AlN. Aluminum oxides containing nitrogen were also observed near the top of the internal-oxidation zone. Elemental x-ray map supporting these findings is also shown in Fig. 10. These observations indicate that Al additions enhanced internal oxidation under the outer oxide scale. So, the mass gain recorded on the Cr–W specimens containing Al in these experiments was a result of both reduced spalling and enhanced internal oxidation.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig9_HTML.jpg
Fig. 9

Back-scattered-electron images of a cross-section of the Cr–10W–8Al oxidized at 1,000 °C in dry air. (a) Image showing the scale and the entire thickness of the internal-oxidation region and (b) a close up of a region near the external scale

https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig10_HTML.jpg
Fig. 10

X-ray elemental maps acquired on a cross-section of Cr–10W–8Al oxidized at 1,000 °C in dry air

Effect of Aluminizing on the Oxidation Behavior of Cr–W Alloys

Aluminizing Cr–10W alloys was accomplished by dipping the specimens in a molten aluminum bath held in a crucible heated by a resistance furnace. Excess aluminum on the surface of the Cr–W specimens was removed mechanically. Only the reaction layer between aluminum and the substrate was left on the surface. This layer was about 10 μm thick. Figure 11 shows the reaction layer on top of the substrate alloy. It is evident from the back-scattered-electron image and WDX analysis that the reaction zone between the substrate and aluminum composed of several compositionally different layers. The Cr and W content of the layers increases and the Al content decreases as the probe gets closer to the substrate.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig11_HTML.jpg
Fig. 11

(a) A Cr–10W coated with aluminum showing a reaction zone. (b) A closeup image showing compositionally different layers within the reaction zone (Compositions are given in atomic percent)

To test the effectiveness of aluminizing Cr–W alloys against oxidation, a cyclical-oxidation test was conducted. The specimens were oxidized in a flow of dry air at 1,000 °C up to 432 h. Specimens of Cr–10W, Cr–10W–2Al, Cr–10W–5Al, Cr–10W–8Al, aluminized Cr–10W, and IN718 were tested. The results of this test are shown in Fig. 12. The Cr–10W specimen, the Al alloyed Cr–10W specimens, and IN718 specimen in this test oxidized in a similar fashion to the previous experiments discussed earlier (Figs. 6 and 8). The mass loss of Cr–10W was high due to extensive spalling. Alloying with Al decreased the amount of spallation. As a result, these specimens gained mass during the test, although some spallation was evident. The aluminized Cr–10W specimens gained mass as a function of exposure time according to Eq. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig12_HTML.gif
Fig. 12

Mass change of various materials exposed to dry air at 1,000 °C

$$ \Updelta m = 1.94\;t^{{0.2}} $$
(2)
This is compared to the mass gain of the uncoated Cr–10W specimen in dry air at 800 °C during a continuous oxidation in Fig. 13. Although the test conditions were more aggressive due to higher exposure temperature and cyclical exposure in the case of the aluminized specimens, the rate of oxidation was much lower. The oxidation rate of the uncoated Cr–10W is approaching a value an order of magnitude higher than the rate of the aluminized Cr–10W at 10,000 h.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig13_HTML.gif
Fig. 13

Comparison of oxidation rates of uncoated Cr–10W at 800 °C and aluminized Cr–10W at 1,000 °C in dry air. The fit to data was extrapolated to 10,000 h

The aluminized specimens showed no sign of spallation during the cyclical oxidation test. Although the oxide-scale surface appeared full of cracks as shown in Fig. 14, the scale was very adherent to the substrate. A closer look at the surface of the scale revealed micron-size grains which were determined to be Al1.7Cr0.3O3 using WDX. XRD done on the surface of the scale gave similar d-spacing for this compound to Cr2O3.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig14_HTML.jpg
Fig. 14

Aluminized Cr–10W after being oxidized at 1,000 °C in dry air for 432 h

Examination of a cross-section of an aluminized Cr–10W specimen after 432 h exposure to dry air at 1,000 °C was done using WDX as shown on a back-scattered image in Fig. 15 and X-ray mapping using WDX and EDS, shown in Fig. 16. This analysis revealed that the outer scale was composed of an oxide of aluminum and chromium as shown in Fig. 15. The Al and Cr content of this scale varied as a function of distance into the scale as determined using WDX. This oxide can be expressed as AlxCr2−xO3−y. The observed values of x and y are 0.9 < x < 2 and 0 < y < 0.005. At the top of the scale, the oxide was essentially Al2O3 with a very small amount of Cr. Below that layer, there was another oxide layer that contains almost equal amounts of Al and Cr. A layer containing a high volume fraction of porosity lied below this layer. Below the porosity layer, another oxide layer with less Al and more Cr existed. A layer of Al2O3 a few grains thick constituted the final oxide layer which had a semi-continuous character. Throughout the oxide layers described above, up to 0.3 atomic percent of nitrogen was consistently detected. A dispersion of AlN particles in the metal matrix was another significant effect of the oxidation process into the substrate. This AlN dispersion layer was about 30 μm thick. The last oxidation effect observed farthest into the substrate was the needle-like precipitates. The X-ray elemental maps (Fig. 16) suggest that these were Al2O3 particles.
https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig15_HTML.jpg
Fig. 15

Cross-section of aluminized Cr–10W after oxidation in dry air at 1,000 °C for 432 h (Compositions are given in atomic percent)

https://static-content.springer.com/image/art%3A10.1007%2Fs11085-008-9095-0/MediaObjects/11085_2008_9095_Fig16_HTML.jpg
Fig. 16

X-ray elemental maps acquired on a cross-section of aluminized Cr–10W oxidized in dry air at 1,000 °C for 432 h

It should be mentioned that instead of having an aluminum–chromium oxide phase with different Al and Cr ratios in the scale as described above, possibility of having a fine mixture of Cr2O3 and Al2O3 also exists. The XRD results were not conclusive to make this distinction.

These observations suggest that there is significant diffusion of Al, N, and O into the substrate material during the exposure. Outward diffusion of Cr and W is very limited if any.

Conclusions

Continuous oxidation of the Cr–W alloys follows a parabolic rate in dry air at 800 °C. Observations suggest that a mixed-oxide layer composed of Cr2O3 and WO3 forms first upon exposure of fresh alloy surface to dry air. After initial formation of the mixed oxide layer, oxidation reactions take place above and below this layer simultaneously. Above the initial oxidation layer, the Cr2O3 layer grows by diffusion of Cr atoms from the substrate alloy to the scale surface. Below the initial scale, low oxygen activity and high nitrogen activity promotes Cr2N formation.

The oxidation rate of Cr and Cr–10W specimens is dominated by spalling of the oxide scale in large pieces during the cyclical tests in dry air at 1,000 °C.

The oxidation resistance of the Cr–W alloys in dry air at 1,000 °C improves significantly by alloying with Al due primarily to the reduced spalling which takes place in the form of fine particulates instead of large pieces.

Aluminizing improves the oxidation resistance of the Cr–10W alloys significantly. The aluminized specimens show no sign of spallation during the cyclical oxidation test in dry air at 1,000 °C.

Acknowledgements

I would like to thank Keith Collins and Steve Matthes for the SEM observations and microchemical measurements, David Smith for the XRD results, Glen Soltau for aluminizing the oxidation samples, Ed Argetsinger for arc melting Cr–W alloys. I would also like to thank Dr. Kyei-Sing Kwong for generating the relevant predominance diagrams and Dr. Gordon Holcomb for reviewing the manuscript. All of the above personnel are with the NETL. In addition, I would like to acknowledge the work on the continuous oxidation experiments of Julie Flores of University of Texas at El Paso during her internship on the Mickey Leland Energy Fellowship Program.

Copyright information

© Springer Science+Business Media, LLC 2008