Oxide Growth Stresses Research Articles

Scale formation mechanisms of martensitic steels in high CO2/H2O-containing gases simulating oxyfuel environments

In oxyfuel power plants, metallic components will be exposed to service environments containing high amounts of CO2 and water vapour. Therefore, the oxidation behaviour of a number of martensitic 9–12%Cr steels in a model gas mixture containing 70% CO2–30% H2O was studied in the temperature range 550–700°C. The results were compared with the behaviour in air, Ar–CO2 and Ar–H2O. It was found that in the CO2- and/or H2O-rich gases, the mentioned steels tended to form iron-rich oxide scales with significantly higher growth rates than the Cr-rich surface scales formed during air exposure. The iron-rich scales were formed as a result of a decreased flux of chromium in the bulk alloy toward the surface because of enhanced internal oxidation of chromium in the H2O-containing gases and carbide formation in the CO2-rich gases. Additionally, the presence of water vapour in the exposure atmosphere led to buckling of the outer haematite layer, apparently as a result of compressive oxide growth stresses. The Fe-base oxide scales formed in CO2(–H2O)-rich gases appeared to be permeable to CO2 molecules resulting in substantial carburization of the steel.

Scale formation mechanisms of martensitic steels in high CO2/H2O-containing gases simulating oxyfuel environments

In oxyfuel power plants, metallic components will be exposed to service environments containing high amounts of CO2 and water vapour. Therefore, the oxidation behaviour of a number of martensitic 9–12%Cr steels in a model gas mixture containing 70% CO2–30% H2O was studied in the temperature range 550–700°C. The results were compared with the behaviour in air, Ar–CO2 and Ar–H2O. It was found that in the CO2- and/or H2O-rich gases, the mentioned steels tended to form iron-rich oxide scales with significantly higher growth rates than the Cr-rich surface scales formed during air exposure. The iron-rich scales were formed as a result of a decreased flux of chromium in the bulk alloy toward the surface because of enhanced internal oxidation of chromium in the H2O-containing gases and carbide formation in the CO2-rich gases. Additionally, the presence of water vapour in the exposure atmosphere led to buckling of the outer haematite layer, apparently as a result of compressive oxide growth stresses. The Fe-base oxide scales formed in CO2(–H2O)-rich gases appeared to be permeable to CO2 molecules resulting in substantial carburization of the steel.

Why the Growth Rates of Alumina and Chromia Scales on Thin Specimens Differ from those on Thick Specimens

For a number of chromia and alumina forming high temperature alloys and coatings, recent studies revealed, that in some cases the specimen/component or coating thickness may substantially affect the growth rates of the surface oxides. For the alumina formers the thickness dependence is mainly governed by depletion of oxygen active elements such as Y, Zr, Hf, Mg which are either intentionally added alloying elements or manufacturing related alloy impurities. In the case of the chromia forming materials, which tend to exhibit a more substantial dependence of oxidation rate on specimen/component thickness, depletion of minor alloying additions is also an important factor to be considered. However, for these alloys relaxation of oxide growth stresses by plastic deformation of the metallic substrates seems to be the dominant parameter which governs the observed behaviour.

The effect of water vapor in increasing growth stresses in the oxide scale on martensitic steam plant alloys

Investigation of the oxidation behavior of 9% Cr steels at 650°C in dry air and air + 10% H2O have shown that oxide scale growth stresses may play a significant role in breakaway. For this reason, preliminary studies on the development and characteristics of growth stresses in oxides on these materials have begun. These studies include in situ acoustic emission (AE) for monitoring scale cracking, deflection testing in monofacial oxidation (DTMO) and in situ X–ray measurements. The measurements were complemented by detailed post-experimental metallographic investigations.From the DTMO measurements the stresses in the oxide on the specimens tested in humid environment are increased by about a factor of 5 compared to dry air at the beginning of oxidation for P91. In a humid environment the stresses decrease with oxidation time while in dry air they remain almost constant. Significant acoustic emission (AE) occurs in humid air for oxidation coupons while virtually no AE is observed for thin foils in a humid environment and for coupons in dry air. These first results seem to be in good agreement with weight gain data characterizing the breakaway behavior in these environments, indicating that, indeed, oxide growth stresses play a key role for oxidation resistance and, thus, component lifetimes of such steels.The first X–ray results indicate that for E911 the scale structure and composition changes completely between the two environments. Furthermore, in humid air, a breakaway effect is observed with a change from protective spinel type oxide to locally non-protective Fe–rich oxide.

Residual stresses in alumina scales grown on different types of Fe–Cr–Al alloys: effect of specimen geometry and cooling rate

Residual stress measurements in alumina scales grown on Fe–Cr–Al alloys were conducted using the ruby fluorescence technique and the results compared with common theoretical considerations. The oxidation experiments were carried out at 1200 °C using different Fe–Cr–Al-based materials, different substrate thickness values and varying cooling rates. In case of a conventional wrought alloy the cooling rate significantly influences the residual stress state, whereas in case of a stronger, dispersion strengthened (ODS) alloy hardly any effect is visible. The results indicate at the very beginning of the oxidation process a high oxide growth stress level (up to −1.6 GPa), which does not substantially depend on the alloy type or specimen geometry. The growth stresses relax with time and this relaxation at oxidation temperature is faster in case of the wrought alloy than for the ODS alloy. Additionally, for all studied systems except the 2 mm thick PM 2000 specimens a transition from compressive to apparent tensile growth stresses after longer exposure times occurs. For very thin Aluchrom YHf specimens (hM = 50 μm) this transition might be explained by nearly total Al-depletion of the substrate material and the resulting decrease of the thermal expansion coefficient. However, the validity of the theoretical calculations based on an ideal flat surface cannot be assured at least for such thin substrates. In fact, geometrical irregularities, a non-even relaxation of the stress or micro-cracking in the scale are likely to affect the calculated residual and/or growth stresses even for thicker specimens.

Residual stresses in alumina scales grown on different types of Fe–Cr–Al alloys: effect of specimen geometry and cooling rate

Abstract Residual stress measurements in alumina scales grown on Fe–Cr–Al alloys were conducted using the ruby fluorescence technique and the results compared with common theoretical considerations. The oxidation experiments were carried out at 1200 °C using different Fe–Cr–Al-based materials, different substrate thickness values and varying cooling rates. In case of a conventional wrought alloy the cooling rate significantly influences the residual stress state, whereas in case of a stronger, dispersion strengthened (ODS) alloy hardly any effect is visible. The results indicate at the very beginning of the oxidation process a high oxide growth stress level (up to −1.6 GPa), which does not substantially depend on the alloy type or specimen geometry. The growth stresses relax with time and this relaxation at oxidation temperature is faster in case of the wrought alloy than for the ODS alloy. Additionally, for all studied systems except the 2 mm thick PM 2000 specimens a transition from compressive to apparent tensile growth stresses after longer exposure times occurs. For very thin Aluchrom YHf specimens ( h M = 50 μm) this transition might be explained by nearly total Al-depletion of the substrate material and the resulting decrease of the thermal expansion coefficient. However, the validity of the theoretical calculations based on an ideal flat surface cannot be assured at least for such thin substrates. In fact, geometrical irregularities, a non-even relaxation of the stress or micro-cracking in the scale are likely to affect the calculated residual and/or growth stresses even for thicker specimens.

Oxide Growth Stresses in an Austenitic Stainless Steel Determined by Creep Extension

Oxide Growth Stresses in an Austenitic Stainless Steel Determined by Creep Extension

Thermal Shock Testing of Thermal Barrier Coating/Bondcoat Systems

Various methods of thermal shock testing are used by aircraft and industrial gas turbine engine (IGT) manufacturers to characterize new thermal barrier coating systems in the development stage as well as for quality control. The cyclic furnace oxidation test (FCT), widely used in aircraft applications, stresses the ceramic/bondcoat interface, predominantly through thermally grown oxide (TGO) growth stress. The jet engine thermal shock (JETS) test, derived from a burner rig test, creates a large thermal gradient across the thermal barrier coating (TBC), as well as thermomechanical stress at the interface. For IGT applications with long high-temperature exposure times, a combination of isothermal preoxidation and thermal shock testing in a fluidized bed reactor may better represent the actual engine conditions while both types of stress are present. A comparative evaluation of FCT, JETS, and a combined isothermal oxidation and fluidized bed thermal shock test has been conducted for selected ceramic/bondcoat systems. The results and the failure mechanisms as they relate to the TBC system are discussed. A recommendation on the test method of choice providing best discrimination between the thermal shock resistance of the ceramic layer, the ceramic/bondcoat interface, and even substrate related effects, is given.

The effect of cerium on the oxidation resistance of Fe -20Cr-1.7C-lSi

Co is the major contributor to the radiation build-up in nuclear power plants. Many studies have been done to develop Co-free hardfacing alloys to replace Co-based Stellite 6 which is used for valve hardfacing. The Co-free alloys should have equivalent wear resistance, cavitation erosion resistance, and oxidation resistance to Stellite 6. Recently, we developed Fe-20Cr-1.7C-lSi alloy to replace Stellite 6. The new alloy has equivalent sliding wear resistance, and cavitation erosion resistance to Stellite 6. However, the high temperature oxidation resistance of the Fe-based hardfacing alloy is not as good as much as that of Stellite 6 due to the Cr deficiency in the Fe-based hardfacing alloy matrix. Ce is known to improve oxidation resistance of Fe-based alloys at elevated temperatures due to the beneficial effects of Ce such as formation of fine oxide layers and reduction of oxide-growth stresses. Therefore, in present study, the effect of Ce on the oxidation resistance was investigated in pressurized-water at 300? under 7 MPa for 200 days in an autoclave. The concentration range of Ce was up to 0.2 wt.%. The specimens containing 0.2 wt.% Ce showed lower weight loss compared to that of the Ce-free specimen. With increasing Ce concentration, high temperature oxidation resistance of new alloy increased. The increased high temperature oxidation resistance may be due to the formation of adhesive oxide layers.

Growth Stresses in the Oxide Scales on TiAl Alloys at 800 and 900°C

In the oxidation of TiAl alloys, the role of scale-growth stresses formed during oxidation has, thus far, been unknown. In the present paper the oxide-growth stresses were investigated by the deflection-test method in monofacial oxidation (DTMO) accompanied by acoustic-emission measurements. On unmodified surfaces the growth stresses are compressive and reach levels of around −100 MPa. At the same time, significant acoustic emission occurs indicating that even under isothermal conditions, stresses are relieved by a scale-cracking mechanism. For oxide scales on TiAl surfaces, which had been ion implanted with chlorine before oxidation, a very thin protective alumina layer is formed which, however, develops growth stresses in the range of several GPa, accompanied by intensive acoustic emission. In all stress–time curves, a dynamic situation is observed. This consists of phases of stress relief by scale microcracking and phases of stresses increase due to crack healing and further oxide growth. As a result, the level of stress as a function of oxidation time, is characterized by an oscillating course.

Phase transformation and bond coat oxidation behavior of plasma-sprayed zirconia thermal barrier coating

ZrO 2–CeO 2–Y 2O 3 and ZrO 2–Y 2O 3 thermal barrier coatings were prepared using the air plasma spray process. Phase transformation in the ceramic top coating, bond coat oxidation and thermal barrier properties were investigated to compare ZrO 2–CeO 2–Y 2O 3 with ZrO 2–Y 2O 3 at 1300°C under high temperature thermal cycles. In the as-sprayed condition, both coatings showed a 7∼11% porosity fraction and typical lamellar structures formed by continuous wetting by liquid droplets. In the ZrO 2–CeO 2–Y 2O 3 coating, the phase ratio of tetragonal to cubic phase was 75:25 and ZrO 2–Y 2O 3 coating had a 100% non-transformable tetragonal phase. There was no monoclinic phase in either coating. However, the phase ratio of the coatings was changed after 1300°C thermal cycles. In the ZrO 2–CeO 2–Y 2O 3 coating, the ratio of tetragonal to cubic was changed to 88:12 and a monoclinic phase was still not detected, but a 10∼19% monoclinic phase had formed in the ZrO 2–Y 2O 3 coating. The life of the coatings was found to be strongly dependent on the temperature which the bond coat experienced during exposure to a peak temperature of 1300°C. When the bond coat experienced a temperature higher than 1100°C, the useful life of both thermal barrier coatings was decreased drastically and this was related to the oxidation behavior of the bond coat. Al 2O 3 formed preferentially along the bond and top coat interface and the other oxides such as NiO and Ni(Cr,Al) 2O 4 spinel, which were believed to decrease the coating life by oxide growth stress, formed rapidly at the top coat side of the interface at temperature higher than 1100°C. The thermally shocked ZrO 2–Y 2O 3 coating exhibited a non-linear thermal expansion curve which was probably due to a reversible tetragonal–monoclinic transformation. The ZrO 2–CeO 2–Y 2O 3 coating was found to have better thermal cycling behavior than the ZrO 2–Y 2O 3 coating. The reason for this could be that there was no phase transformation from tetragonal to monoclinic phase, which induces volume expansion. The thermal expansion mismatch is smaller and the effect of oxide growth stress is relatively small because of better thermal insulation.

Relation between oxide growth direction and stress on silicon surfaces and at silicon-oxide/silicon interfaces

The effects of stress on oxide growth direction are found to be important on Si(100) surfaces and at SiO 2/Si(100) interfaces from the energetics of the oxide growth. The results of first-principles calculations show that the oxide nuclei on the clean Si(100) surfaces preferentially grow vertically into the substrate. The growth direction is the same even for dihydride Si(100) surfaces. In contrast, the initial oxide nuclei at SiO 2/Si(100) interfaces preferentially grow laterally, parallel to the interface. These results can be explained by the difference in the distribution of the induced stress. The results also show that the oxide growth direction is governed by the stress remaining in the oxidized region.

Determination of oxide growth stress by a novel deflection method

A new technique has been developed to monitorthe growth stresses in oxide scales by measuringspecimen deflection. Compared to the conventionaldeflection method, the new technique does not require acoating for protecting one side of the specimen fromoxidation. This new method does not have temperaturelimits and has improved precision for stress measurementin alumina scales. Two alumina-forming alloys, Fe-22Cr-5Al-0.3Y and Fe-22.6Cr4.42Al-0.21Ti,were studied with this method. During oxidation at1000°C in air, the average compressive stresses inthe alumina scales were found to decrease from ~1000 to ~100 MPa with increasing scale thickness on thetwo alloys. The growth stress in alumina scales wasdirectly proportional to the oxide growth rates. Thestresses in the substrate alloys were much lower, on the order of 1 MPa. Creep occurred in thesubstrate alloys and was taken into consideration in thestress analysis. It is believed that creep in the alloysubstrates allowed stresses in the oxide scales to relax. Annealing tests showed that thestresses relaxed much easier in thin than in thickscales.

Evidence of stress relaxation in thermally grown oxide layers—experiments and modelling

High temperature X-ray diffraction was used for determining the stresses created in chromia layers formed on Ni–Cr as a function of various parameters: substrate and oxide layer thickness, grain size, yttrium doping, cooling rate. After oxidation, stresses at room temperature in the oxide layer are compressive. During further heating, their values decrease with annealing temperature and they tend to zero at 900°C, indicating that the oxide growth stresses are negligible. Besides, relaxation phenomena were experimentally evidenced during heat-treatments after previous oxidation. A numerical model was developed to calculate the stress and the relaxation phenomena. It takes into account the mechanical behaviour of both the substrate and the oxide, and parameters such as substrate thickness, oxide grain size, cooling rate or heat-treatment duration, etc. This elasto-visco-plastic model correlates the experimental results with the various parameters of the study and agrees with previous results obtained for NiO layers. Finally, using a finite element model, it was shown that the stress gradient in the oxide is negligible.

Plastic deformation of oxide scales at elevated temperatures

The atomic force microscope (AFM) has been used to observe and characterize for the first time surface steps and grooves on the faces of Cr2O3 grains formed as an oxide scale on Ni−30Cr and Ni−30Cr−0.5Y alloys during high temperature oxidation. The very high spatial resolution of the AFM is required to characterize these features. We propose that these surface features, whose dimensions are in the range of nanometers and tens of nanometers, may be interpreted as evidence of highly localized plastic deformation of the oxide scale. The size and spacing of the steps and grooves are consistent with models of plastic deformation based on slip bands derived from dislocation climb or dislocation glide. Mechanical twinning and the models for stress-driven surface instability are also possibly responsible for some surface features. The addition of yttrium to the alloy seemed to enable enhanced plastic deformation of the scale. The strain corresponding to the observed features, estimated by simple models, could relax a significant part of oxide growth and thermal stresses.

A creep analysis of the deflection test for evaluating oxide growth stresses

During isothermal oxidation of one surface of a metal strip, growth-induced stresses will produce curvature of the metal/oxide composite sample. This observation has been used extensively elsewhere to evaluate the average stress developed within the oxide layer, but previous estimates have been based on an elastic stress analysis of the bending composite. At typical oxidation temperatures, creep deformation of both oxide and metal phases is much more likely. The present paper provides a stress analysis assuming such non-linear creep deformation. It considers the situation where the oxide experiences a net compressive stress and the metal a net tensile stress. The oxide neutral axis may be located either within the oxide layer or outside it, in the gas phase, depending on the thickness of the oxide and the relative creep strengths of the oxide layer and metal substrate. Evaluation of the stress distribution and average stress values in both the metal and oxide phases is provided for notional mechanical properties, within which the metal creep strength is varied widely. It is also shown how the assumption of elastic deformation can lead to large overestimates of stress values.

Stress effects in high temperature oxidation of metals

In this review the role of stress in the high temperature oxidation of metals is examined. In the introductory sections the Wagner theory of oxide growth and its modification to account for short circuit diffusion are outlined. The direct influence of oxide growth stresses on defect diffusion is then examined and shown to be a significant factor when constrained volume changes accompany the growth of the oxide layer. Methods for evaluating growth stresses are critically reviewed and the possible origins of such stresses considered in detail. The response of the oxide layer to thermally and mechanically induced stresses is considered in the last part of the review. Under tensile conditions, through-thickness cracks can readily be produced but may also heal with fresh oxide if imposed strain rates are relatively low. Spallation of the oxide is not usually found under tensile conditions but occurs frequently under compressive stresses by either a ‘buckling’ or ‘wedginq’ process. Detailed descriptions of these are provided and their use as the basis of spallation maps is described.

An approach to a global model of the mechanical behaviour of oxide scales

The protective effect of oxide scales is strongly linked with their behaviour under mechanical loading. Scale cracking may occur at relatively low strains and a decisive part is played by the process of scale crack healing due to ongoing oxidation. It can even be maintained that many technical high temperature materials only survive in service for a long time due to the good healing capability of the protective oxide scales formed on them. This illustrates the fact that the characterization of the mechanical behaviour of protective oxide scales also has to include oxidation-induced effects. In order to characterize the behaviour of protective scales under mechanical loading quantitatively a number of parameters can be used. On the one hand, there are the conventional parameters used for bulk materials as well, such as elastic and plastic or creep properties as well as fracture mechanical properties. On the other hand, for growing oxide scales parameters describing the oxidation-related mechanical properties should be used in addition. The latter describe effects like the superimposed build-up of oxide growth stresses, pseudo-plasticity due to crack-healing processes and final loss of the protective effect after a certain loading history due to subsurface zone depletion of protective scale forming elements. In the first part of this paper the conventional parameters are discussed with respect to their application to a quantitative description of the mechanical behaviour of oxide scales. The second part focuses on the oxidation-related parameters of scale deformation. In all cases where relevant model equations exist, they are discussed in some detail. In the third part an attempt is made to combine the model equations of the first two parts with a global representation in the form of a scale-failure diagram. Three principal types of scale-failure diagram are introduced and discussed, and in conclusion it is recommended to extend these diagrams by a third dimension in the form of the physical defect size in the scale. A comparison of a theoretical scale-failure diagram with the experimentally determined borderlines for the scale on Alloy 800 H shows that the theoretical models yield conservative values. The models can, therefore, be used to assess conditions for the safe operation of the materials.

A model for the oxide growth stress and its effect on the creep of metals

A model for the oxide growth stress and its effect on the creep of metals

The development of Cr 2O 3 scales on iron-chromium alloys containing reactive elements

An investigation has been undertaken into the oxidation behaviour of Fe-28Cr and similar alloys containing cerium, yttrium or gadolinium additions, in oxygen at 1000°C. Following the initial transient oxidation period, a healing layer of Cr 2O 3 is established on the four alloys. This develops a convoluted configuration on the binary alloy, with large areas growing out of contact with the alloy surface. Such feature results from lateral growth of the scale, following formation of new oxide in the bulk matrix, and the generation of considerable oxide growth stresses. These stresses are sufficient to cause isothermal fracture of the convoluted scale, allowing penetration of oxygen to the alloy substrate. The scales developed on Fe-28Cr-0.8Y and Fe-27Cr-1.3Gd maintain good contact with the alloy and thicken more slowly than on the binary alloy. These effects can be associated with a reduction in the rate of transport of chromium across the scale and with the formation of intergranular and internal oxides in the underlying alloy substrate. Keying of the scale to the alloy probably does not improve scale adhesion but the precipitates prevent alloy grain-boundary sliding and reduce alloy-deformation processes. The scale formed on Fe-23Cr-0.5Ce also thickens more slowly than that on the binary alloy. However, like Fe-28Cr, it develops a convoluted configuration in the early stages. This is possibly associated with the fact that very little intergranular or internal oxidation occurs for this alloy and alloy grain-boundary sliding may initiate loss of contact and facilitate the establishment of a convoluted scale.