Professional Documents
Culture Documents
RR3000和CMSX 4单晶高温合金的高温氧化行为 PDF
RR3000和CMSX 4单晶高温合金的高温氧化行为 PDF
is this layer which will eventually become a major cause SZ11 optical microscope. Higher spatial resolution
of TBC system failure.3,6 Upon thermal exposure and investigations were also carried out on metallographic
cycling, stresses develop within the TGO, and it is cross-sections in a JEOL JSM 840A SEM and on
thought that when the stored elastic strain energy microsections produced with an FEI focused ion beam
reaches a critical value, exceeding the TGO fracture (FIB) 201 system. The microsections were milled
toughness, scale failure occurs. In such events, the bond perpendicular to the metal/oxide interface following a
coat and ultimately the base metal will be directly procedure described elsewhere.7
exposed to the hot gas stream where the oxidation An automated X-ray diffractometer, Philips X’Pert,
behaviour of the latter becomes the main influencing was used with Cu Ka radiation at 40 kV and 20 mA to
factor for component integrity and durability. identify the phases present in the oxidised samples. X-
Therefore, data on the oxidation resistance of the coated ray analysis was performed at a scan rate of 2u min21
and uncoated blade materials are needed to help in over a 2h range of 20–100u. Phase identification from the
assessments of the safety and reliability of the turbine XRD spectra was made using the JCPDS-ICDD
engines. database (International Centre for Diffraction Data,
In this study, the high temperature oxidation beha- Newton Square, PA)
viour of two single crystal nickel based superalloys Shortly after each oxidation experiment, the residual
RR3000 and CMSX-4 has been investigated to under- stress in the oxide scale was measured at room
stand and compare their characteristic properties. For temperature using photostimulated Cr3z luminescence
this purpose, isothermal oxidation experiments were piezospectroscopy.6,8 This technique was implemented
performed by exposing these materials to air at the same in a laser Raman microprobe, Renishaw 2000, to
temperature and for equal durations. Post-oxidation determine the residual stress from scale regions identi-
investigations were carried out to characterise the fied in an optical microscope attached to the Raman
microstructures, compositions and residual stresses in system. For each alloy condition, a series of 15 measure-
the oxide scales as well as the development of these with ments were made on randomly selected areas of the
oxidation time at the test temperature. The results of sample surface. All areas were probed using a 633 nm
these investigations are presented first and then dis- He–Ne laser beam 5 mm in diameter. Stress values were
cussed with emphasis placed on the resistance of the derived from the spectral data according to a procedure
alloys studied to oxidation in gas turbine environments. described in Ref. 8 and by using the equation6
Dn(cm{1 )~5:07s(GPa) (1)
Experimental
where Dn represents the shift in frequency of the
The chemical compositions of the single crystal super- characteristic R2 luminescence line of a-Al2O3 and s
alloys investigated are listed in Table 1. Both materials the biaxial stress.
were supplied by Rolls-Royce plc. Alloy RR3000 was in
the form of 60630650 mm bar and alloy CMSX-4 Results
as a rod y12 mm in diameter. A series of duplicate
test samples, y3 mm thick, were cut from each alloy Initial microstructure of alloys
and then prepared with surfaces perpendicular to Figures 1 and 2 show the microstructures of the alloys
the solidification direction of the single crystal. After studied in the as received condition. The optical and
polishing to a 1 mm finish followed by rinsing in alcohol, electron optical images were taken after etching
they were separately placed in alumina boats inside a mechanically polished samples of each alloy in a
tube furnace and oxidised in air at 1100uC. All oxidation 1 : 1 : 1 vol.-% solution of acetic, nitric and hydrochloric
experiments were conducted in the following way: acids for 10 s. The microstructure of alloy RR3000 was
(i) heat the sample to 1100uC within 30 min similar to that of alloy CMSX-4, though it looked finer.
(ii) hold at this temperature for a fixed length of In each case, the alloy exhibited a dendritic structure
time (Fig. 1) consisting of a mixture of c matrix and c9 phase
(iii) cool to 300uC at a rate of 5 K min21 (Fig. 2). The c9 phase stood out from the c matrix as
(iv) retrieve the sample from the furnace and cool to cube like particles with a mean edge length of y300 nm.
room temperature in ambient air. The periods In addition to these main features, a small amount of
of oxidation at 1100uC are given in Table 2. c/c9 eutectic phase, exhibiting a ‘fan-like’ structure, was
Following the oxidation experiments, the microstruc- found in alloy RR3000.
ture, morphology of surface scales and cross-sections
through the metal/oxide interfaces were examined using
Oxidation behaviour
a Hitachi S-2300 scanning electron microscope equipped Surface scale development
with an energy dispersive X-ray (EDX) analysis system. The development of the surface scale was monitored by
Development of the surface scale with oxidation time taking a series of macrographs of the test samples for
was monitored for each alloy by taking low magnifica- each alloy. Figure 3 shows examples of the macrographs
tion photographs of the test samples on an Olympus recorded after test durations of 3, 30 and 100 h. The
scale formed on alloy RR3000 was initially light green
Table 1 Nominal compositions of alloys CMSX-4 and and contained numerous crater like features randomly
RR3000, wt-%
Table 2 Oxidation times (hours) used for testing alloys
Element Al Co Cr Mo Ti Re Ta W Hf Ni RR3000 and CSMX-4
CMSX-4 5.8 9.5 6.5 0.4 1.0 3.0 6.5 6.5 0.1 Bal. RR3000 0.25 0.5 1.0 1.5 2.0 3.0 5.0 20 30 50 100
RR3000 5.7 3.0 2.2 0.4 0.2 6.0 8.5 5.5 0.03 Bal. CSMX-4 0.25 0.5 1.0 1.5 2.0 3.0 5.0 10 30 … 100
distributed across the sample surface. These features were In contrast to alloy RR3000, a different develop-
of various sizes ranging between 10 and 250 mm in ment of the scale was observed with alloy CMSX-4.
diameter and exhibited, in general, a distinct blue In the early stages of oxidation, 0.25–1 hour, the
coloration. After exposure for 3 h, the surface of the scale appeared to grow preferentially on dendritic
scale darkened significantly and developed a pattern areas of the metal surface. After few hours of expo-
resembling that of the dendrites in the base alloy, though sure, its general appearance evolved into a distinct ‘criss-
at a larger scale. This pattern was visible for up to y30 h cross’ structure comprised of alternated dendritic and
of exposure and then became undiscernable as the interdendritic regions, as shown in Fig. 4. After longer
oxidation progressed. In the early stages of oxidation, exposures, this structure tended to fade slightly but
0.25–5 h, the scale appeared to adhere quite well to the remained visible under a light microscope. All the
substrate. However, spallation of the scale, during cool- test samples were found to exhibit good adhesion of
ing, subsequently occurred. The extent of spallation was the scale to the substrate. Spallation of the scale upon
found to increase gradually over 20 h of oxidation and cooling did not occur on any of these samples other than
then intensified to a point where only small portions of the near the sample edge following the 100 hour test
scale remained adherent to the underlying metal (Fig. 3). (Fig. 3).
5 a cross-sectional view of oxide scale formed on alloy RR3000 after oxidation for 30 h and b corresponding X-ray
images
6 a close up view of NiO top layer and b spalled area of surface oxide on alloy RR3000 after 50 h of oxidation
10 a cross-sectional view of oxide scale formed on alloy CMSX-4 after 100 h of oxidation and b corresponding EDX
images
a-alumina at 1100uC, as reported in the literature.9,10 It appeared, at first, to relax a little and then increase again
can be observed from the results shown in Fig. 11 that to reach a peak at ,4 GPa. The most likely reason for
over this transient period, the stress in the scale stress relaxation was believed to be microcracking in the
increased rapidly reaching a peak value of ,4 GPa alumina layer. Subsequently the cracking seemed to heal
after oxidation for only 1.5 h. The stress dropped from leading to a rapid increase in stress for exposures
its peak value after 2 h exposure rising again following between 5 and 20 h. After this time, the stress started to
the next exposure. This sudden change in the stress level decrease significantly attaining after 100 h a level well
over a short period of time was also seen with alloy below that measured after the first exposure. The onset
RR3000. It is believed to be due to the tensile stresses of the sharp decrease in stress coincided with that of the
developed in the scales by the large shrinkage in volume acceleration in scale spallation observed in the samples
(,9%) accompanying the transformation of alumina oxidised for periods between 30 and 100 h. The low
from the metastable h phase to the stable a phase. In stress levels associated with these samples were indica-
fact, both of these phases could be identified in the tive of the presence of large cracks in the scale, which
luminescence spectra from samples oxidised for short ultimately led to buckling and spallation of large
durations between 0.25 and 2 h. The spectra obtained portions of the oxide (Fig. 3).
from samples oxidised for longer periods, indicated that
only the a phase was present in the scale. After
completion of the alumina phase transformation, the
Discussion
compressive residual stress in the scale of CMSX-4 Studies of the oxidation behaviour of Ni–Cr–Al alloys
increased progressively until the last exposure (Fig. 11b). (often considered as the ternary equivalents of nickel
Such a steady increase was indicative of good adhesion based superalloys),11 have clearly identified a transient
of the scale to the substrate. phase at the commencement of oxidation followed by a
The development of residual stress in the scale of alloy steady state period.12,13 During the transient period,
RR3000 after the transient period differed markedly oxidation of the various alloy constituents takes place,
from that observed for alloy CMSX-4 (Fig. 11a). For leading to the nucleation of NiO, complex spinels, Cr2O3
periods of oxidation over 3–20 h, the residual stress and Al2O3. During the steady state period, different
11 Development of residual stress in oxide scale formed on a alloy RR3300 and b alloy CMSX-4 as function of oxida-
tion time
oxide scales may form and can be divided into three seen in alloy RR3000. The rapid establishment of a
distinct types, depending on the concentrations of Cr continuous a-Al2O3 inner layer immediately after the
and Al in the metal substrate. Type I is an external scale transient stage retarded the oxidation of alloy CMSX-4.
of NiO and a subscale of chromium and/or aluminium Because of the slow diffusion of reactive species in a-
oxides. Type II is an external layer of Cr2O3 and a Al2O3, the growth of CrTaO4 and other oxides was
sublayer of Al2O3. Type III is solely an external layer inhibited. This beneficial effect of the a-Al2O3 layer was
of Al2O3. According to the free energy of formation not evident in alloy RR3000 and was substantiated by
values, for the oxidation of Ni–Cr–Al alloys, a the fact that the overall thickness of the scale formed on
continuous Al2O3 scale should form if the Al content the latter was greater than that formed on alloy CMSX-
exceeds 4 wt-%, provided the Cr/Al concentration ratio 4. The results and observations of the present study are
is less than or equal to 4 (Ref. 12). Hence the alloys in agreement with those of a recent investigation aimed
investigated here, which contain nearly equal amounts at elucidating the effect of rhenium on the oxidation
of Al (,5.8 wt-%) with 2.2 and 6.5 wt-%Cr respectively resistance of Ni based alloys.18 For such a purpose, the
(Table 1), would be expected to form an Al2O3 outer Re content was varied in two series of Ni–8Cr–2Ti–
layer during oxidation. However, owing to the addition 10Al–Re and Ni–8Cr–2Ti–15Al–Re model alloys (the
of other elements, such as Co, Ti, Ta, W and Re, both of concentrations expressed in at.-%) and it was found that
the alloys under investigation exhibited a more complex the alloys with a Re/Al ratio >0.1 exhibited a much
behaviour than simple Ni–Cr–Al alloys. larger mass change than those with Re/Al ratio (0.1, as
The results of the XRD and SEM/EDX analyses a result of cyclic oxidation at 1100uC. This observation
presented in the previous section indicated that upon led to the suggestion that the Re/Al ratio is a good
exposure to air at 1100uC, a mixed oxide scale was indicator of the resistance to oxidation. Furthermore,
formed on the surface of the single crystal superalloys. examination of cross-sections showed that the a-Al2O3
During the initial stage of oxidation, simple oxides of layers that formed on the alloys with Re/Al ratio >0.1
the alloy elements, such as NiO, CoO, Al2O3, Ta2O5 and are highly porous owing to increased Re content and
Cr2O3 (although some of them were not detected), formation of more volatile Re2O7 oxide. Since the alloys
formed simultaneously (Table 3). At 1100uC, the much investigated in the present study have similar Al
higher growth rate of NiO as compared with that of the contents but different Re/Al ratios, it is possible that
other oxides resulted in the overgrowth of NiO crystals. the high Re content and the associated high vapour
This was more apparent in the case of alloy RR3000 pressure of its oxide renders the a-Al2O3 layer formed on
than in CMSX-4. Meanwhile, solid phase reactions alloy RR3000 porous and less protective.
among the initial simple oxides led to the formation of It is obvious from the results obtained from the test
more complex oxides, including NiCr2O4, CrTaO4, samples (Fig. 3) that spallation of the scale is the major
AlTaO4 and NiCoAl2O4. As a result, after 3 h of process impairing the resistance of alloy RR3000 to
oxidation the predominant oxide was identified to be oxidation at high temperatures. Under cyclic loading
(Ni,Co)O in the scale of alloy RR3000 and (Ni,Co)O, conditions, such as in gas turbines, alloy CMSX-4 would
(Ni,Co)Al2O4 and a-Al2O3 in that of alloy CMSX-4. be expected to be more resilient to oxidation owing to
The formation of CrTaO4 and AlTaO4 on the single the good adherence of its scale and resistance to spalling
crystal superalloys Rene N6 and CMSX-10 upon upon cooling. This is confirmed by the results of residual
isothermal and cyclic oxidation at temperatures in the stress measurements shown in Fig. 11, which indicated a
range 1000–1200uC was reported by other research- strong bonding of the scale with the substrate in the case
ers.14,15 In the present work, the XRD results indicated of alloy CMSX-4 and a weak bonding in the case of
the formation of Cr2O3 and W20O58 during the initial alloy RR3000. However, the oxidation of these alloys
stage of oxidation and their disappearance from the leads to the creation of a c9 denuded zone beneath the
scale on longer exposure (Table 3). This could be the scale and coarsening of the c9 precipitates (Fig. 7). The
result of the continued oxidation of Cr2O3 (s) to CrO3(g) reduction in the volume fraction of c9 precipitates in the
and W20O58 (s) to WO3 (g) at temperatures .1000uC.16 denuded zone may lead to a local decrease in strength.
Moreover, the XRD data and X-ray images did not This weakening effect may be extensive as the c9 volume
reveal the presence of a Re rich oxide or a Re enriched fraction is closely related to the stress to give a
layer in the scales formed on the superalloys studied. specified creep rate.19 The formation of a weak subsur-
This was believed to be due to the loss by vaporisation face zone could make the alloy susceptible to cracking
of the Re oxide, Re2O7, which has a high vapour under combined mechanical and thermal loading con-
pressure at high temperatures.17 ditions, of the type encountered in gas turbines. To
After the formation of a mixed oxide scale, the alleviate or better suppress the above mentioned
activity of oxygen at the metal/oxide interface may damaging effects owing to oxidation, it still appears
decrease, facilitating the selective oxidation of alumi- necessary to provide the superalloys studied with a
nium. With increasing oxidation time, the inner a-Al2O3 protective coating.
crystals grew laterally and eventually converged to form
a compact inner layer. The formation of this layer on
both alloys was ascertained by cross-sectioning some
Conclusions
samples which appeared to have readily reached steady The following conclusions can be drawn based on the
state conditions (Figs. 5 and 10). However, comparison present isothermal oxidation study of the single crystal
of samples oxidised for equal durations, particularly superalloys RR3000 and CMSX-4.
those exposed for short times, showed that the a-Al2O3 1. On exposure to air at 1100uC, alloy RR3000
layer had formed earlier on alloy CMSX-4, because this develops an oxide scale which is much thicker than that
layer proved to be thicker and more uniform than that which forms on alloy CMSX-4.