INSTITUTE OF PHYSICS PUBLISHING J. Phys. D: Appl. Phys.

39 (2006) R101R124

JOURNAL OF PHYSICS D: APPLIED PHYSICS doi:10.1088/0022-3727/39/6/R01

TOPICAL REVIEW

Elevated temperature erosive wear of metallic materials

Manish Roy
Defence Metallurgical Research Laboratory, PO Kanchanbagh, Hyderabad 500 258, India

Received 6 July 2005, in nal form 9 December 2005 Published 3 March 2006 Online at stacks.iop.org/JPhysD/39/R101 Abstract Solid particle erosion of metals and alloys at elevated temperature is governed by the nature of the interaction between erosion and oxidation, which, in turn, is determined by the thickness, pliability, morphology, adhesion characteristics and toughness of the oxide scale. The main objective of this paper is to critically review the present state of understanding of the elevated temperature erosion behaviour of metals and alloys. First of all, the erosion testing at elevated temperature is reviewed. This is followed by discussion of the essential features of elevated temperature erosion with special emphasis on microscopic observation, giving details of the erosionoxidation (EO) interaction mechanisms. The EO interaction has been elaborated in the subsequent section. The EO interaction includes EO maps, analysis of transition criteria from one erosion mechanism to another mechanism and quantication of enhanced oxidation kinetics during erosion. Finally, the relevant areas for future studies are indicated.

Nomenclature
E En Mn Mn t mn V K1 p o Kp m Ao Q R T Z C Kp o Erosion rate Erosion rate at nth exposure Mass loss suffered due to erosion on nth exposure Mass gain experienced due to oxidation on nth exposure Time of exposure to eroding conditions Cumulative mass of erodent for nth exposure Impact velocity Constant Velocity exponent of erosion rate Parabolic rate constant Incremental mass gain due to oxidation Arrhenius constant Activation energy for oxidation Universal gas constant Absolute temperature Thickness of the oxide scale Constant Scaling constant Density of the oxide
2006 IOP Publishing Ltd

F L W H U mp r N tb tim Zb

Particle ux rate Depth to which the deformed zone extends Indentation diameter Constant Hardness Crater volume Mass of single erodent Radius of the erodent Number of erodents Semi included angle of conical erodent Time between impacts Time of impact Thickness of the oxide scale grown between two successive impacts

1. Introduction
Erosive wear or solid particle erosion is sometimes known as impact wear. Solid particle erosion is dened as material degradation due to the impact of particles travelling with some signicant velocity. It is mechanistically different from other forms of erosion such as liquid impact erosion, slurry
Printed in the UK

0022-3727/06/060101+24$30.00

R101

Topical Review

Table 1. Examples of systems subjected to elevated temperature erosion. System Combustion systems Coal gasication systems Coal liquefaction system Gas turbines Components Burner nozzles, reheater, super heater, economizer tube banks, boiler heat exchanger, in bed tubes, tube banks, etc Turbine, lock Hopper valves Valve to throttle the ow of product stream Blades References [5, 6] [7] [7] [8]

erosion and cavitation erosion, etc. A tribosystem suffering from erosive wear can be characterized as an open system. In such a system, the counterbody is continuously replaced. Another important feature of solid particle erosion is that the wear of the counterbody is completely uninteresting. In solid particle erosion, the contact time between the erodent and the target material is only momentary. In this respect, erosion is different from other related processes such as sliding wear, abrasive wear, grinding and machining, etc, in which the contact between the tool/abrasive and the target/work piece is continuous. Room temperature erosion is an important problem in several engineering applications. Rocket motor trail nozzles [1], the engine of a helicopter operating in dusty terrain [2], equipment in oil and mining industries [3, 4], etc are subjected to solid particle erosion at ambient temperature. Several engineering components are degraded due to solid particle erosion at elevated temperature. A number of industrial systems, which undergo erosion at elevated temperature, are summarized in table 1 [58]. At the same time, it should be mentioned that erosion could also be used constructively, as in the case of shot peening, sand blasting, mining, rock drilling and cutting applications, etc [911]. Erosion behaviour of metallic materials at room and elevated temperatures has been reviewed in several publications [1220]. The information contained in these reviews will be repeated only to the extent of impressing upon the readers the essential features of the erosion processes. It is not the purpose of this work to provide a comprehensive review of the status of research in the eld of solid particle erosion at elevated temperature; rather the aim is to discuss and review some of the recent results which have enhanced our understanding in the areas of elevated temperature erosion of metallic materials. This review will be conned only to metals and alloys. Elevated temperature erosion of ceramics, glasses and composites will not be considered here. For the purpose of convenience, the review is divided into several sections. Section 2 consists of elevated temperature erosion tests. The salient features of elevated temperature erosion constitute section 3. Examination of the eroded surfaces is considered in section 4. Section 5 deals with erosion oxidation interaction. Erosion enhanced oxidation kinetics is analysed in section 6. Future research areas for elevated temperature erosion are discussed in section 7. This is followed by concluding remarks in section 8.

and assessing the extent of erosion. These tests can broadly be divided into two categories, namely (1) those simulative tests that are designed to simulate a specic type of erosion and (2) those that are intended to be used for fundamental studies. The main problem of the simulative test is that it is very expensive and it is difcult to conduct fundamental studies for material development or for understanding the mechanisms of erosion. In order to avoid these problems, various laboratory tests have been developed. The most common laboratory test involves blasting a stream of airborne particles against the target as standardized by G 76-83 [21]. In this type of test, a known quantity of erodent is fed into an air stream, accelerated through a converging nozzle and directed towards the test specimen. A cleaned and weighed sample is exposed to the particle-laden air stream for a predetermined time and weighed after interrupting the test. The ratio of the weight loss suffered by the sample to the weight of erodent gives the dimensionless erosion rate. The air jet-type elevated temperature erosion rigs can be classied into two broad groups. The rst group comprises erosion rigs, which are designed in such a fashion that both the uid stream carrying the particles and the eroding target are heated to the same test temperature. These are called isothermal erosion rigs. On the other hand, the second group of erosion rigs, called non-isothermal rigs, have the facility to heat the target material alone, while the uid stream with the particles is not preheated before allowing it to enter the erosion chamber. In these types of rigs, the colder uid stream cools the target materials to some extent on impact. Nevertheless, the target material still attains a steady-state temperature. The above classication of the erosion rig as isothermal and nonisothermal is largely articial for reasons stated below. 1. The test specimen attains a constant temperature and remains at that temperature throughout the test in both types of rigs. 2. The erodents are not heated and the cold particles impinge the target material. However, unlike in the case of abrasion or sliding wear, the contact between the particles and the target materials in the case of solid particle erosion is momentary, and hence negligible transfer of heat takes place between the particles and the substrate. Thus, it is immaterial as to whether the particles are preheated or not. The non-isothermal erosion rigs are easy to fabricate but they fail to simulate the erosion conditions. Hence, such rigs are not popular at present. In contrast, isothermal type erosion rigs can simulate erosion conditions but they are difcult to fabricate. The schematic diagram of one such erosion rig, fabricated by the author at DMRL, is shown in gure 1. The unique feature of this rig is its ability to alter the particle feed rate by over 100 times. Its particle feeding system is

2. Elevated temperature erosion test

Over the last few decades several test techniques or methodologies have been developed for studying mechanisms R102

Topical Review

Temperature Indicator of Hot Air Particle Feeding System

Air Pressure Indicator

Fluidized Chamber Outlet Temperature Indicator of The Sample

Air Flow Rate Indicator

Air Heating System Inlet Test Chamber Temperature Controller For Air Heating System

Sample Holding Devices

Temperature Controller of the Test Chamber

Figure 1. Schematic representation of jet type elevated temperature erosion rig [67].

a miniaturized conveyer belt system and particle feed rate is controlled by controlling the speed of the motor of the system. Further description of this rig is available elsewhere [22]. The test procedure involves heating the compressed air to the required temperature and then heating and soaking the test sample to that temperature. The heated samples are then exposed to compressed, heated, uidized and accelerated air streams carrying the particles. The elevated temperature erosion test is a multiple specimen test procedure. Cleaned, dried and weighed samples are exposed to erodents for various time intervals (say t1 , t2 , t3 , . . . , where t1 < t2 < t3 ) corresponding to various mass of erodents (m1 , m2 , . . . , mn ). A similar number of samples are again exposed to an air stream without carrying particles. If M1 , M2 , . . . , Mn represent the mass loss suffered by n samples when exposed to erodents for time intervals of t1 , t2 , . . . , tn and if M1 , M2 , . . . , Mn are the mass gains experienced by n samples when exposed to the plain air stream without erodents for time intervals of (t1 , t2 , . . . , tn ) then the incremental erosion rate E1 , E2 , . . . , En can be computed as En = (Mn Mn ) (Mn1 Mn1 ) . mn mn1 (1)

This procedure is repeated until En1 is equal to En2 and this E is considered to be the incremental erosion rate. The main problem of the jet type of erosion rig is the measurement of impact velocity. There are three different types of velocity measurement techniques available. In the rst method, known as the photographic method, a highspeed camera is used to photograph the successive positions of a single particle as a function of time and thus compute the velocity. The second method, known as the rotating

disc method, was developed by Ruff and Ives [23]. In this method velocity is determined by estimating the time of ight of particles between two discs xed on a common shaft rotating at some specied velocity. A modied version of the rotating disc is the paddle wheel technique [24]. This method gives a more reliable and statistically more accurate velocity. The third method uses the laser Doppler velocitimeter (LDV). This technique is an accurate, non-interactive and on-line velocity measuring device. The LDV uses the wellknown Doppler effect to measure the velocity of the particles. When light is scattered from a moving object, the stationary observer will see a change in the frequency of scattered light proportional to the velocity of the object. A laser is used as a light source because it is easily focused and is coherent. In many applications, for example, pipe bends in a slurry transportation system, the impact is primarily by big particles with very low impact velocity. To simulate such a system, samples are impacted by dropping particles under gravity. Such a system was rst introduced by Bitter [25], a schematic presentation of which is shown in gure 2. The system consists of a ball dispenser unit, the velocity measuring system, ball counting unit and the sample holder. The ball dispenser and the sample holder can be moved up and down so as to alter the height over which the eroding particle falls. Before the steel ball impacts the sample it passes through a multiple photodiode unit, which measures the velocity of the passing ball and, in addition, keeps track of the total number of balls passing through. Whirling arm rigs were developed to enable tests to be carried out at precisely controlled velocity over a range of impact conditions. The target specimens are attached to the tips of the rotor arms and whirled through a certain or a R103

Topical Review

Ball Feeder Velocity Measuring System

3. Salient features of elevated temperature erosion of metallic material

Over the last decade or so, a substantial amount of work on elevated temperature erosion of metals and alloys has been carried out [2762]. A compilation of these investigations, providing details of materials subjected to elevated temperature erosion and the test conditions, is made in table 2. These tests cover a wide range of test conditions and test materials. Based on the work of the investigators compiled in table 2, the important factors which inuence the solid particle erosion behaviour of metallic materials at elevated temperature are described below. 3.1. Effect of temperature
Sample Holder

Timer

Figure 2. Schematic diagram of low velocity erosion rig.

The variation of erosion rate with temperature for a number of metals and alloys is depicted in gures 4 and 5 [27, 28]. The erosion data presented in these gures pertain to high impact velocities and mostly with angular particles. The observed temperature dependence of the erosion rate can be classied into three groups. In the rst group, the erosion rate initially decreases with the increase of temperature, reaches a minimum and then starts increasing with increasing temperature. Materials such as 5.0Cr0.5Mo, 17-4 PHSS, 410 SS, Alloy 800, Ti-6Al-4V, and tungsten belong to this group. The second group comprises metals such as Ta and lead (for oblique impact) and alloys such as 310 SS (for oblique impact), 1018 steel and 1100 aluminium (for normal impact) which exhibit a temperature independent erosion rate up to a critical temperature followed by an increase of the erosion rate with increasing temperature. Finally, group three materials show a monotonically increasing erosion rate with increasing temperature. Inco 600, carbon steel, 12Cr1MoV steel, and 2.25Cr1.0Mo steel, lead and 20245 Al are some of the typical examples in this group. 3.2. Effect of impact velocity The velocity dependence of erosion rate (E) is characterized by the velocity exponent, p, given by E = K1 V p , (2)

Figure 3. Schematic diagram of whirling arm erosion rig.

narrow band of erosive particles. These rigs are very noisy and consume considerable power. A whirling arm erosion rig is schematically shown in gure 3. These kinds of rigs simulate the degradation conditions prevalent in the uidized bed combustor. The whirling arm erosion rig was initially developed by Tilly and Sage [26]. One of the advantages of this type of rig is that the erodent velocity can be controlled precisely as the velocity is governed by the rotating speed of the arms. It also permits testing to be carried out over a wide range of impact velocities. It makes efcient use of the erodent as the entire amount of erodent is delivered to the target sample. R104

where K1 is a constant and V is the impact velocity. The velocity exponents (p) obtained by various investigators are plotted in the velocitytemperature regime in gure 6. The velocity exponent decreases with an increase in erosion test temperature for 304 SS to values as low as 0.9 at low impact velocities. At relatively higher impact velocities p appears to lie in the range 23. Levy and Man [29, 30] reported the inuence of erodent size on the velocity exponent for erosion of 9Cr1Mo steel at 923 K using angular SiC particles. 3.3. Effect of impact angle The available data related to erosion rate and impact angle at different temperatures are presented in this section. However, most of the metallic materials, irrespective of temperature of erosion, exhibit a ductile behaviour, i.e. a maximum erosion

Topical Review

Table 2. Compilation of the work of various investigators on elevated temperature erosion of metallic material. Materials 310 SS, 304 SS, 1018, 2.25Cr1.0Mo, 5.0Cr0.5Mo, 410 SS, 17-4 PHSS 1100 Al, 310 SS 9Cr1.0Mo, 2.25Cr1.0Mo, 5.0Cr10.5Mo, 304 SS, 410 SS 410 SS 5.0Cr0.5MoSi Inco 600 Ti6Al-4V, 2074 Al, 410 SS, W, Ta, Pb and 304 SS Ni, Co Ni, Co AISI 303 304 SS, alloy 800, C-steel, 2.25Cr1.0Mo, 12.0Cr1.0MoV 304 SS, 316 SS and 410 SS In 738, X 40, MA 754 and HA 188 Stellite 6B and 1, In-100, MA-754, HA 8077, AISI 446 SS, FeCrAlY, AISI 446 9.0Cr1.0Mo, 304 SS 304 SS, 416 SS, 430 SS and 17-4 PH SS 2.25Cr1.0Mo Alloy 800 HT, 310 SS Haynes 188, Waspaloy Ti6Al-4V Carbon steel FeCrC cast steel Ni, Ni20Cr Fe3 Al base alloy, Ni, Co FeCrC Name of the investigators Levy et al [27] Finnie et al [32] Levy and Man [29, 30, 47] Levy et al [45] Levy and Wang [40, 48] Tabakoff and Vittal [34] Gat and Tabakoff [28] Kang et al [50] Chang et al [38, 47] Shayler and Yee [31] Shida and Fujikawa [42] Singh and Sundararajan [43] Barklow et al [51] Wright et al [52] Sethi and Wright [53] Zhou and Bahadur [35] Sethi and Carey [54] Stott et al [55] Chinadurai and Bahadur [56] Zhou and Bahadur [57] Xie and Walsh [58] Drotlew et al [59] Manish Roy et al [60] Yu et al [61] Hayashi et al [62] Test conditions Nitrogen gas, 240 m SiC, V = 30 m s1 , T = RT to 1173 K, = 30 and 90 Nitrogen gas, 250 m SiC, V = 3060 m s1 , T = RT to 0.8 Tm, = 30 Air, 130 m SiC and SiO2 , V = 35 m s1 , T = 9231123 K, = 90 and 30 Air, 5 m y ash, 130 m Al2 O3 , T = 1223 K, V = 5 m s1 , = 45 Air, 90 m alumina, V = 30, 40 and 70 m s1 , T = 1123 K, = 9 Air, 100800 m quartz, V = 60250 m s1 , T = 544, 666 and 846 K, = 1070 Air, 86 m quartz, V = 180300 m s1 , T = 3001023 K, = 1590 also 164 m quartz, V = 120 m s1 , T = 300483 K, = 20 , 60 and 90 Nitrogen, air, 20 m alumina, V = 90 and 140 m s1 , T = 9231073 K, = 90 Air, 20 m alumina, V = 140, 123 and 70 m s1 , T = 8731053 K, = 90 , 60 , 30 and 20 Air, 4766 m y ash, V = 150300 m s1 , T = 300773 K, = 35 Argon gas, 120 m quartz, V = 40120 m s1 , T = 673923 K, = 2090 Air, 160 m SiC, V = 55110 m s1 , T = 300773 K, = 30 , 60 and 90 Burner rig, 20 m alumina, V = 200275 m s1 , T = 1148 K Argon gas, air, 12 m alumina, V = 43 m s1 , T = 1033 K, = 30 Air, 1 m alumina, V = 2.7 and 4.3 m s1 , T = 7331100 K, = 30 Air, 120 grit SiC, V = 65 m s1 , T = 9231073 K, = 30 Air, 1 m alumina, V = 2.7 m s1 , T = 763863 K, = 30 Air, 4766 m y ash, V = 150300 m s1 , T = 300773 K, = 35 Air, 150 m SiC, V = 50 m s1 , T = 3001073 K, = 30 Air, 4766 m y ash, V = 50 m s1 , T = 3001073 K, = 1090 Air, 43 m coal ash, V = 530 m s1 , T = 423673 K V = 65 m s1 , T = 723 K, = 1560 Air, 200 m SiO2 , V = 35, 65 and 105 m s1 , T = 3001073 K, = 30 , 45 , 60 and 90 SO2 , air, 2050 m SiO2 , V = 160190 m s1 , T = 8731073 K, = 090 Air, 23 m cold silica, V = 80 m s1 , T = 973 K

Note: T = test temperature, V = impact velocity and = impact angle.

rate at oblique impact angles (1030 ). The universality of such an observation is shown in gures 7 and 8, wherein data from a large number of investigators have been compiled [3136]. Levy [37] obtained a higher erosion rate at normal impact than at oblique impact for 9Cr1Mo steel at 1123 K using rounded Al2 O3 (130 m) erodent for a range of impact velocities (3070 m s1 ), as shown in gure 9. But at a low impact velocity of 20 m s1 , a maximum in the erosion rate occurred at oblique impact angle. Observations by Chang et al [33, 38] shown in gure 10 indicate that the peak erosion rate of Co at a test temperature of 1053 K occurs at an impact angle of 60 when impacted with 20 m angular alumina particles at

impact velocities of 70140 m s1 . However, when the erosion test is carried out at 873 K and at impact velocity of 140 m s1 , the erosion rate peaks at an impact angle of 30 . Thus, there exist apparently conicting observations regarding the erosion rateimpact angle behaviour. 3.4. Effect of particle size Tabakoff and Vittal [34] carried out erosion tests on Inco 600 alloy using quartz particles having sizes between 70 and 800 m. Their work indicates that the erosion rate increases marginally with the increase of particle size, as shown in gure 11. Zhou and Bahadur [35] investigated the effect R105

Topical Review

of the particle size of SiC on the erosion rate of 304 SS at 923 K (impact angle: 30 and impact velocity: 65 m s1 ). These results indicate that erosion rate increases with the increase in particle size up to 40 m and thereafter it becomes independent of the particle size (gure 12). Levy et al [29, 30, 37, 39], however, noted only the increase of erosion rate with the increase of particle size for 9Cr1Mo steel eroded at a temperature of 923 K and that for 1018 steel eroded at 723 K.

3.5. Effect of particle shape Levy et al [29, 30, 37, 39, 40] investigated the erosion rate of a number of Cr containing steels at 1123 K with angular SiC and spherical Al2 O3 as erodent particles. A typical result for 9Cr1Mo steel is presented in gure 13. The erosion rate is signicantly higher when SiC is used as the erodent. 3.6. Effect of particle feed rate Zhou and Bahadur [41] have investigated the inuence of the particle feed rate on the erosion rate of 304 and 430 SS over a large temperature range. The reported results as illustrated in gure 14 suggest that up to a temperature of about 773 K increasing the feed rate by a factor of 16 has no inuence on the erosion rate. However, beyond 773 K, a lower feed rate results in a substantially higher erosion rate. Roy et al [60] also noted a decrease of erosion rate with the increase of the particle feed rate, especially at low impact velocity. 3.7. Effect of eroded material characteristics

120 100 Erosion Rate x 10 6 (Kg/Kg) 80 60 40 1018 Steel 20 310 SS 0 0 200 400 600 800 Temperature (K) 1,000 1,200 Impact Velocity : 30 m/s Impact Angle : 30o Erodent : SiC (250 m) Gas : Nitrogen 410 SS 5.0Cr-0.5Mo

2.25Cr-0.5Mo 17-4 PHSS 304 SS

Figure 4. Variation of erosion rate with temperature for a number of alloys [27].

The interpretation of the available data on the effect of eroded material characteristics on its erosion rate is complicated by the fact that the behaviour of the oxide scale under erosion conditions needs to be considered in addition to the behaviour of the metallic material per se. The oxidation characteristic of the eroded material plays a more important role than the mechanical properties of the eroded materials at elevated temperature.

120 Erosion Rate x10 4 (Kg/Kg)

Erosion Rate x 10 4 (Kg/Kg)

140 120 100 80 60 Tungsten 40 20 0 Impact Angle : 90o Impact velocity : 120 m/s Erodent : Quartz ( 130 m) 283 303 323 343 363 383 403 423 Lead

110 100 90 80 70 60 50 40 30 20 283 303 323 343 363 383 403 423 Temperature (K)
10

Lead

Impact Angle : 20o Impact velocity : 120 m/s Erodent : Quartz ( 138 - 164 m)

Temperature (K)
5

Erosion rate x 104 (Kg/Kg)

9 8 7 6 5 4 3 2 1 0 283 333 383 Temperature (K) 433 483 Impact Angle : 20o Impact velocity : 120 m/s Erodent : Quartz ( 138 - 164 m) Ti-6Al-4V 410 SS 2024 Al Ta
Erosion Rate x10 4 (Kg/Kg) 4 4 3 3 2 2 283 333

Ta 410 SS Ti-Al-4V

2024 Al o Impact Angle : 90 Impact velocity : 120 m/s Erodent : Quartz ( 138 - 164 m) 383 Temperature (K) 433 483

Figure 5. Effect of test temperature on the erosion rate of a number of metals and alloys [28].

R106

Topical Review

P values given alongside data points

Figure 6. Velocity exponent obtained by various investigators, plotted in velocitytemperature regime.

Even at elevated temperature, if one considers the erosion behaviour of metallic material at high impact velocities and feed rates, the oxidation plays an insignicant role and the erosion behaviour is essentially metal erosion behaviour. The experimental data of Shida and Fujikawa [42] pertaining to 1.25 Cr1MoV, 2.25CrMo, 12Cr1MoV and plain carbon steel (up to 923 K), that of Singh and Sundararajan [36, 43] pertaining to 304, 316, 410 stainless steel (up to 773 K) and that of Levy et al [44, 45] pertaining to 2.25Cr1.0Mo steel, 5Cr 0.5Mo steel, 1018 steel, 304 SS, 310 SS, 410 SS and 17-4 PHSS can be considered elevated temperature erosion of metals with minimum or negligible oxidation. Under such conditions the dependence of the strength of the material on temperature is a reasonable indicator of the temperature dependence of erosion resistance of the materials. Erosion data further indicate that austenitic stainless steels have superior resistance to elevated temperature erosion than ferritic steels. The 410 stainless steel having a tempered martensitic matrix exhibits an erosion resistance comparable to that of austenitic stainless steels. The oxidation effect is reported to be important for elevated temperature erosion tests conducted at low impact velocities and using rounded Al2 O3 as an erodent. The inuence of Cr content on the erosion rate of steel is illustrated in gure 15. It is indicated that the erosion rate decreases to a very low value when Cr content in the steel exceeds 1012% [35]. In the case of steel having Cr less than 10%, thick Fe2 O3 scale was formed during erosion leading to

45 Erosion Rate x 10 4 (Kg/Kg)

6 Erosion Rate x 10 4 (Kg/Kg)

40 35 30 25 20 15 10 5 0 0

850 K

5 4 3 291 K 2 1 0 10 20 30 40 50 60 70 80 90 Impact Angle (Degrees)

6

1023 K

766 K 644 K Material : INCO 600 Impact Velocity : 183 m/s 10 20 30 40 50 60 70 80

Material : 303 SS Impact Velocity : 130 m/s

Impact Angle (Degrees)

0.050 0.045 Maximum Thickness Loss (m/s) 0.040 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 0 20 40 60 80 100 Impact Angle (Degrees)

Erosion Rate x 10 4 (Kg/Kg)

Test Temperature = 573K Impact Velocity = 120 m/s Erodent = quartz (120 m) 304 SS

5 1093 K 4 3 2 1 0

1255 K

Al 310 SS

Impact Velocity = 61 m/s Impact Velocity = 31 m/s Erodent = SiC (250 m) Fluidizing Gas = Nitrogen

c
C Steel

d
743 K 372 K 293 K 293K 558 K

Cr-Mo-V

20

40

60

80

100

Impact Angle (Degrees)

Figure 7. Variation of erosion rate with impact angle at various test temperatures: (a) INCO 600 [34], (b) 303 SS [31], (c) CrMoV, carbon and 304 SS [42] and (d) aluminium and 310 SS [32].

R107

Topical Review
45 Erosion Rate x 10 4 (Kg/Kg) Material : 304 SS Impact Velocity : 305 m/s 1173 K
Erosion Rate x 10 4 (Kg/Kg) 16 14 12 10 8 6 4 2 0 10 20 30 40 50 60 70 80 90 873 K 1173 K Material : 304 SS Impact Velocity : 183 m/s

40 35 30 25 20 15 10 5 0 10 20 30 40 50 60 70 80 90 Impact Angle (Degrees) 873 K

Impact Angle (Degrees)

14 Erosion Rate x 10 4 (Kg/Kg)

3 Erosion Rate x 10 4 (Kg/Kg) 3 2 2 1 1 0 0 20 40 60 80 100 Impact Angle (Degrees) 297 K 773 K Material : 304 SS Impact Velocity : 70 m/s

13 12 11 10 9 8 7 6 5 4 20 30 40 50 60 70 80 90 100 Impact Angle (Degrees) Impact Velocity : 129 m/s 316 SS 410 SS 548 K RT 763 K

Figure 8. Inuence of impact angle and test temperature on erosion rate of (a) 304 SS, (b) 304 SS, (c) 304 SS [35] and (d) 410 SS and 316 SS [43].

high erosion rates [46]. The experimental results of Chang et al [47], presented in gure 16, also indicate the importance of the nature of the scale that forms during erosion at elevated temperature. It is also noted that materials with high scaling rate such as nickel and cobalt exhibit the highest erosion rates, while the alumina-forming alloys such as CoCrAlY and NiCrAlY exhibit intermediate erosion rates. The superiority of the Al2 O3 forming alloy stems from the fact that the Al2 O3 forming scale forms much more slowly than the Cr2 O3 scale. Extensive work by Levy and co-workers [37,39,44,48,49] implies that the morphology of the oxide scale that forms during erosion is important. Segmented scales have a better erosion resistance than thick, continuous and dense scale since the spalled area is conned to oxide crystalline only in the case of erosion of the segmented scale. A striking illustration of the above fact is obtained when Si is added to steels. Addition of Si to low chromium steel results in the formation of a segmented scale even at high impact velocity and thereby reduces the erosion rate substantially as compared with the same steel without Si [49]. The above discussion clearly brings out certain features of elevated temperature erosion of metallic materials. Almost all metallic materials exhibit ductile erosion response at room temperature, whereas at elevated temperature both brittle and ductile erosion responses are noted [31, 36, 37]. The velocity exponents for metallic materials are 2.5 during ambient temperature erosion. At elevated temperature the velocity exponent of metals and alloys varies over a wide range from 0.9 to even more than 3.0 [3133]. It is established that the erosion rate at room temperature increases with the increase of the particle size up to 50 m and beyond such magnitude R108

45 40 Metal Thickness Loss (m) 35 30 25 20 15 10 5 0 0 20 40 60 8 00 Impact Angle (Degrees) 25 m/s 35 m/s 70 m/s 45 m/s

Figure 9. Variation of metal thickness loss with the impact angle for 9.0 Cr1.0 Mo steel at 1123 K [37].

the particle size has no effect on the erosion rate. Reported literature indicates that the erosion rate increases with the increase of particle size at high temperature [29,30,34,39]. At ambient temperature, changing the particle shape from angular to spherical results in altering the erosion response from brittle to ductile [63, 64]. At elevated temperature, brittle to ductile response is noted irrespective of the particle shape [29,30,37]. Particle feed rate has a negligible effect on room temperature erosion rate [6567]. A remarkable effect of the particle feed rate has been noticed at elevated temperature [29]. The characteristics of mechanical properties of eroding material particles have nominal inuence on room temperature erosion behaviour [56, 6875]. In contrast, this aspect at elevated temperature is hardly explored.

Topical Review
3.0 1053 K, 140 m/s
Weight Loss x 10-6 (Kg/Kg) 300 9Cr 1 Mo Steel Nozzle Tester Corrosion Erosion Air Particle Size: 130 m Test Temperature : 1123K Test Time: 2 hr 90o 150 30o 90o 50 Al2O3 30o SiC

Erosion Rate ( x 10 -1, Kg/m2 sec)

2.5

250

2.0 873 K, 140 m/s 1.5

200

1.0

100

0.5

873 K, 70 m/s
0

0.0 0 20 40 60 8 00 Impact Angle (Degrees)

10

20

30 40 50 Impact Velocity (m/s)

60

Figure 10. Variation of erosion rate of Co with the impact angle at 1053 K [38].
7

Figure 13. Variation of weight loss with impact velocity for 9.0 Cr1.0 Mo steel at 1123 K showing the effect of particle feed rate on erosion behaviour [37].
450

400 Erosive Mass Loss Rate (Kg/Kg) 350 300 250 200 150 100 50
0 100 200 300 400 500 600 700 800 900

Erosion Rate (m3/Kg) x 10

5 4 3 2 1 0 Particle Size (m)

Erodent Concentration 2.6x10-6 Kg/m2s 42.0x10-6 Kg/m2.s Material: 304 SS 430 SS

0 0 200 400 600 800 Test Temperature (K) 1,00 ,200

Figure 11. Inuence of particle size on the erosion rate of Inco 600 alloy [34].
400 350 Erosion Rate (Kg/Kg) 300 250 200 150 100 50 0 0 20 40 60 80 100 120 140 160 Particle Size (m)

Figure 14. Variation of erosion rate of 304 SS and 430 SS with test temperature showing the effect of feed rate on erosion rate [41].

Figure 12. Effect of particle size on the erosion rate of 304 SS at 923 K [35].

4. Examination of eroded surfaces

At elevated temperature, the material removal is governed by the synergistic effect of erosion and oxidation (EO). Detailed examination of the eroded target material using a scanning electron microscope (SEM), a transmission electron microscope (TEM) and an optical microscope (OM) has been carried out. The advent of single particle experiment is

particularly fruitful since such experiments allow a detailed examination of large, individual craters. On the basis of the extensive literature [57, 67, 7678], four different types of EO mechanisms can be envisaged: in the rst case, at low temperatures, at high impact velocities and feed rates, there is no oxide scale. Even if there is any oxide scale, it will be very thin and it will be able to deform in the same manner as that of the substrate target. Under such circumstances, erosion takes place from the metallic surface and this mechanism of erosion is called metal erosion. The erosion behaviour in this regime is similar to the ambient temperature erosion behaviour of metallic materials. The erosion response in the metal erosion regime is ductile, the velocity exponent of the erosion behaviour is between 2 and 3 and the erosion rate is independent of the particle feed rate. The metal erosion mechanism is schematically shown in gure 17(a). In the metal erosion regime, there are two modes by which materials can be removed. These modes are ploughing and cutting. In general, when a particle is in contact with a target at positive rake angle, the cutting mode operates. On the other hand, the ploughing mechanism operates at negative rake angle. Cutting mechanisms result in generation of new surfaces while the ploughing mechanism involves the displacement and extrusion of the material with no new surface generation. In addition, the cutting mode is more efcient than the ploughing mode R109

Topical Review
120 Normal Impact Oblique Impact 2.25Cr 5Cr 60 9Cr 40 410SS 304 SS 20 310 SS Test Temperature : 1123 K Impact velocity : 35 m/s Erodent: Al2O3, 130 m Fluidizing Medium: Air

100 Metal Thickness Loss (m)

80

0 0 5 10 15 20 25 30 Chromium Content (wt %)

Figure 15. Inuence of Cr content on the erosion rate of steel [35].

when considered in terms of energy consumed per unit volume removal of the target material. The work carried out by a large number of investigators [7982] has revealed that almost all metals and alloys lose material by the formation of a lip and its subsequent fracture. In metals and alloys, during erosion, once the lip is formed, it is fractured by several modes. In the case of ductile metals like copper [83], brass [83], aluminium [84] and iron [85] the lip fracture occurs by necking, and the resulting fracture is ductile, as exemplied by the dimpled fracture surface. This mode of lip fracture is shown in gure 18(a). In the case of a high strength alloy such as CuBe in age hardened condition [83], 301 SS [85] and TD nickel [85] the lip removal is greatly aided by the formation of adiabatic shear bands at the base of the lip as shown in gure 18(b) and subsequent easy separation/fracture across this band. In this case also the fracture is ductile, as indicated by the presence of shear dimples on the fracture surface. These two modes of material removal considered above involve the fracture of pre-existing lips. On the other hand, a new mode termed adiabatic shear induced spalling involves the formation of intersecting adiabatic shear bands at the base of the crater and subsequent removal of chunks of material as illustrated in gure 18(c). This mode of weight loss, which is highly efcient in terms of energy expended per unit volume of target material removed, is important only at normal impact where maximum resistance is offered to curtail spreading of deformation. The erosion response under such circumstances will be similar to that observed for ceramic materials. But the underlying mechanism is entirely different. In the case of ceramic materials, material removal occurs with the formation of intersecting cone or radial cracks, which nucleate from pre-existing aws once a critical tensile stress is exceeded. Further, these cracks are essentially brittle in nature. On the other hand, formation of an adiabatic shear band requires critical strain [86]. In addition, the fracture surface resulting from adiabatic shear induced spalling exhibits shear dimples, implying an essentially ductile fracture. On the other extreme, at very high temperatures and low velocities and particle feed rates, erosion takes place from R110

the oxide scale only, as shown in gure 17(b). Under such conditions, a thick oxide scale is formed on the target material during erosion and the deformed zone formed due to impact is conned within the oxide scale. The erosion behaviour from the oxide scale is characterized by a brittle erosion response, strong velocity dependence and particle feed rate independent of the erosion rate. This erosion mechanism is termed oxide erosion. In oxide erosion, material removal occurs with the formation of intersecting cones and radial cracks, which nucleate from pre-existing aws once a critical tensile stress is exceeded. At an intermediate temperature, impact velocity and particle feed rate, an oxide scale of intermediate thickness is formed. However, the depth of the deformed zone extends to the metallic substrate beyond the oxide scale. Consequently, the oxide scale beneath the eroding particle tends to crack, gets pushed down into much softer base material and in the process the softer base material gets squeezed out onto the top surface through the cracks in the oxide scale. Over a period of time, the repetition of such a process during each impact causes the formation of a composite layer comprising the bulk metal and broken pieces of oxide scale. Erosion takes place from this composite layer. This mechanism, presented schematically in gure 17(c), is termed oxidation affected erosion. The interesting aspect of oxidation affected erosion is that the volume fraction of the oxide in the composite layer is a function of erosion conditions such as temperature, impact velocity and particle feed rate. As a result, the erosion behaviour in the oxidation affected erosion regime can vary from a ductile to a brittle response depending on the amount of oxide scale present in the composite layer. Further, unlike in the case of metal erosion or oxide erosion, the oxidation affected erosion rate depends strongly on the test temperature and particle feed rate. The nal erosion mechanism is oxidation controlled erosion and this is illustrated in gure 17(d). At relatively higher temperatures and lower particle feed rates and impact velocities, the oxide scale that forms during erosion is brittle and non-adherent. In such cases, the oxide scale gets removed after it attains a critical thickness. The erosion behaviour in this regime exhibits a brittle erosion response, weak velocity dependence and particle feed rate dependent erosion rate. Figures 19 and 20 show SEM images of the morphologies of the eroded surfaces and transverse sections of the eroded surfaces, obtained after exposing commercially pure Ni to elevated temperature erosion. All the four mechanisms described schematically in gure 17 can be seen under SEM.

5. Erosion oxidation interaction

As mentioned previously, the erosion behaviour of metallic materials at elevated temperature is governed by the nature of interaction between erosion and oxidation. The nature of interaction between erosion and oxidation in turn depends on the thickness, morphology, adherence and the toughness of the oxide scales that form in these materials. Before going into the details of mechanisms it is important to deal with the theoretical aspects.

Topical Review

(a) 4.0
3.5 Ni 3.0 Erosion Rate (Kg/m2s)x106 2.5 2.0 MA-754 1.5 CoCrAlY 1.0 Ni-20Al 0.5 0.0 Co Test Temperature : 1053 K Impact Velocity : 140 m/s Impact Angle : 30o Erodent : Al2O3 (20 m)

Ni-30Cr

(b) 3.0
Ni 2.5 Erosion Rate (Kg/m2s)x106 Test Temperature : 873 K Impact Velocity : 140 m/s Impact Angle : 30o Erodent : Al2O3 (20 m)

2.0

Co MA-754

1.5

Ni-30Cr

1.0

CoCrAlY

0.5

Ni-20Al

0.0

Figure 16. Bar diagram showing the erosion rates of various alloys forming different types of scale under oblique impact at (a) 1073 K, (b) 873 K [48].

5.1. Theoretical aspects The details of the interrelationship between erosion conditions and erosion mechanisms has been presented earlier. Further discussion on this interrelationship will be carried out subsequently. Thus, only the salient theoretical aspects will be considered in the following section. 5.1.1. Steady state oxide scale thickness. If it is assumed that the oxide scale which forms on the eroding material during erosion is adherent and sufciently ductile to withstand repeated impacts without developing cracks, steady state oxide scale thickness can be dened. It can be assumed that the oxidation of the eroding material follows the parabolic kinetics given by equation (3) m =
2 o Kp t,

parabolic rate constant and t is the time of exposure. The parabolic rate constant is usually expressed in the form:
o Kp = Ao exp

Q , RT

(4)

where Ao is the Arrhenius constant, Q is the activation energy for oxidation, R is the gas constant and T is the absolute temperature. In order to represent EO interaction in mathematical terms one needs the rate of growth of the oxide scale thickness with time rather than the weight gain given by equation (3). As noted by Lim and Ashby [87], once the composition of the oxide scale is known, equation (1) can be transformed to give Z 2 = 2Kp t, Kp = 0.5 C
2 o Kp ,

(5) (6)

(3)

where m is the mass gain experienced by the metal per unit o area due to intake of oxygen to form oxide scale, Kp is the

where C is a constant for a given oxide composition (unit is m3 kg1 ). Kp is usually referred to as the scaling constant. R111

Topical Review
Erodent Deformed Region Lm Lip
Deformed Region Lm V Erodent Lip Oxide Scale

Metal

Metal

Metal Erosion with Thin Adherent Ductile Oxide Scale

Metal Erosion with No Oxide Scale

a (Metal Erosion)
V Damaged Zone Z Lo Oxide

Metal

b (Oxide Erosion)
V Deformed Region Oxide Scale Erodent

V Erodent Lip Oxide Scale

Metal

Composite Layer

Metal

Metal

c (Oxidation Affected Erosion)

Oxide Scale
Newly Formed Scale
Oxide Debris

Oxide Scale

Metal

Metal

Metal

d (Oxidation Controlled Erosion)

Figure 17. Schematic presentation of various erosion mechanisms at elevated temperature.

A value appropriate to the erosion conditions should be chosen for Kp since Levy et al [88] have clearly demonstrated that the oxide scales grow much more rapidly under erosion conditions as compared with static conditions. From equation (5), the rate of increase of oxide scale thickness with time is given as Kp dZ = . (7) dt Z If Eo is assumed to be the erosion rate of oxide scale and F be the particle ux rate given by the ratio of particle feed rate R112

(f ) to the eroded area then the rate of decrease of the oxide scale thickness due to its erosion is obtained as dZ Eo F , = dt o where o is the density of the oxide. Finally, a situation will arise when the oxide growth by oxidation (equation (7)) will be equal to the oxide removal by erosion (equation (8)). Under such conditions, the steady state (8)

Topical Review

Figure 18. Schematic diagrams of various metal removal mechanisms during metal erosion.

Figure 19. SEM images showing different morphologies of eroded surfaces of Ni exposed at elevated temperature: (a) metal erosion, (b) oxidation affected erosion, (c) oxidation controlled erosion and (d) oxide erosion [67].

oxide thickness (Zss ) can be obtained as Zss = K p o . Eo F (9)

rate and decreasing particle ux rate. It should be mentioned that Eo in equation (9) represents the erosion rate of pure oxide. 5.1.2. Depth of deformed zone in oxide and metal. A large number of experiments have consistently shown that the depth to which the deformed zone extends is usually of the order of R113

Hence, the steady state oxide thickness increases with increasing temperature (through Kp ), decreasing oxide erosion

Topical Review

Figure 20. SEM images of the transverse section of the eroded surfaces of Ni shown in gure 19. (a) Metal erosion, (b) oxidation affected erosion, (c) oxidation controlled erosion and (d) oxide erosion [67].

indentation diameter, or in other words L = W, (10)

where W is the indentation diameter and is a constant of the order of unity. The magnitude of W depends on whether the erodent is spherical or angular. When a conical particle of mass mp as shown in gure 21 impacts an eroding material of hardness H with an impact velocity V , then the incident energy of the impacting particle is given as 0.5 mp V 2 and the energy consumed in forming the crater of volume U is H U where H is the hardness of the target material. Equating these two energies, we get H U = 0.5 mp V 2 . (11)

Substituting equation (15) in equation (11) and solving for W one obtains 2r 1/3 V 2/3 . (16) W = H 1/3 Thus the depth of deformation for a conical particle can be obtained by putting the value of W in equation (10) as L(con) = 2.0 r 1/3 V 2/3 . H 1/3 (17)

For a spherical erodent having radius r, U is given by U= W 4 64r (12)

Equations (14) and (17) are valid irrespective of whether the target material is metallic material or oxide scale. However, it should be mentioned that the hardness H in equations (14) and (17) represents the hardness of the oxide scale Ho in the case of thick oxide scale as target material and the hardness of the base metal in cases where the thickness of the oxide scale is very small. 5.1.3. Critical oxide thickness (Zc ). An important factor that should be considered to understand the EO interaction is that the oxide scale usually exhibits a ductile to brittle transition as a function of both thickness and temperature. Stephenson et al [89] have demonstrated this phenomenon under impact conditions. Saunders and Nicholls [90] have also noted similar ductile to brittle transition for chromia and alumina coatings. Important points related to this phenomenon are given as follows. (1) At all temperatures, the scale becomes brittle beyond a critical thickness (Zc ) and thus can be removed easily by spalling or by cracking and chipping due to particle impacts. (2) This value of Zc changes discontinuously over narrow temperature ranges usually in the range of 700800 C. (3) Below and above this temperature range Zc is independent of temperature.

since crater depth can be considered to be considerably smaller than the diameter of the erodent. Substituting equations (12) in equation (11) and solving for W one obtains W = 2.56 rV 1/2 H
1/4

(13)

where is the density of the erodent. Hence, the depth of deformation for a spherical erodent can be obtained by putting equation (13) in equation (10) as L(Sphe) = 2.56rV 1/2 H
1/4

(14)

For conical particles having a half angle of = 30 , U is given as W3 . (15) U= 24 tan R114

Topical Review

Figure 21. Variation of the calculated ratio of the time between impacts to the time of impact with hardness.

ts/

5.1.4. Time between the impacts and time of impact. If a spherical erodent of radius r and mass m is considered, then the number of particles impacting the unit area of the eroding material every second equals F /m where F is the particle feed rate. If it is assumed that each impact event causes damage over an area A, where A is the impact crater area and is a constant of the order of unity, then the number of particles (N ) impacting an area of A every second is given by N= or N= F A m (18)

assumption is reasonable because the erosion oxidation interaction becomes prominent only at a low particle feed rate. It is also important to consider the contact duration between the particle and the eroding material during each impact. This duration known as time of impact (tim ) is given as [91, 92] 1.28r tim (sph) = , (23) H 1/2 where r is the radius of the spherical particle having hardness H and density . Thus for spherical particles the ratio of time between impacts and time of impact can be obtained by dividing equation (21) by equation (23) as tb 0.64H . (sph) = tim V F (24)

F 0.25 W 2 , m

(19)

where A = 0.25 W 2 , W is the impact crater diameter. The time between two impacts can now be obtained as 4mp . tb = F W 2 (20)

In the case of conical particles the time of impact can be given to a good approximation as tim (con) = 2.8r 1/3 . H 1/3 V 1/3 (25)

Substituting the value of W obtained from equation (13) in equation (20) and noting that mp = 4/3r 3 , tb is obtained as 0.82r 1/2 H 1/2 . (21) tb (sph) = F V In the case of conical particles with a hemispherical top of size 2r, mass m and half angle , the time between impacts can be obtained from equation (20) if is 30 . This is because the mass of such a particle is equal to that of a sphere of diameter 2r. For conical particles W is given by equation (16) and substituting equation (16) in equation (20) and putting m = 4/3r 3 , tb is obtained as tb (con) = 4r 1/3 H 2/3 . 3F V 4/3 (22)

Thus the ratio of tim and tb is obtained from equations (22) and (25) as tb 0.48H 1/3 (con) = . (26) tim V F In the above equations H represents the hardness of the eroding material. The calculated values of the ratios tb /tim in the case of spherical (continuous lines) and conical erodents (broken lines) are illustrated in gure 21 as a function of hardness of the eroding material and for four values of the product of impact velocity and feed rate. It can be noted from gure 21 that the time between impacts is several orders of magnitude higher than the time of impact. Hence for the erosion process at elevated temperature, time of impact has negligible inuence compared with time between impacts. R115

In the above expression it is assumed that the rebounding particle does not interfere with the incident particle. This

Topical Review

5.1.5. Oxide scale growth between impacts. On the assumption that oxidation is governed by a parabolic law, the growth of oxide scale thickness (Z) is described by equation (5). Thus the thickness of the oxide scale (Zb ) grown between two successive impacts is given by
2 Zb = 2Kp tb .

Table 3. Assumed values of various important parameters. Constant/variable Hardness of oxide Hardness of metal Parabolic rate constant Arrhenius constant Activation energy Erosion rate of oxide Erosion rate constant Reference velocity Velocity exponent Density of oxide Density of particles Erodent radius Coversion factors for o transforming Kp to Kp Particle ux rate Particle shape Critical oxide thickness Impact velocity Test temperature Symbol Ho Hm o Kp Ao Q Eo Eoo Vo n o p r C F Zc V T Values 3.0 2.0 105 , 104 , 103 210 105 107 10 3 5400 3200 100(10) 1.3 104 0.1, 1.0, 10.0 1, 10 5100 8731173 Units GPa GPa kg2 m4 s1 kg2 m4 s1 kJ mol1 kg kg1 m s1 kg m3 kg m3 m m3 kg1 kg m2 s1 m m s1 K

(27)

Substituting the value of tb from equations (21) and (22) in equation (27), one obtains the following equation for Zb for spherical and conical particles. Zb (sph) = 1.26 and Zb (con) = 1.63 Kp r F
1/2

Kp r V F

1/2

(H )1/4

(28)

1/6 H 1/3 V 2/3

(29)

While deriving the above two equations it is assumed that the growth of the oxide scale is controlled by the metal and/or the oxygen ion diffusion through the scale. Thus, it is assumed that the oxide scale does not crack as it grows, thereby providing an easy path for the oxygen diffusion to the metal oxide interface. This assumption is reasonable because it concerns the growth of the oxide scale at a very local region surrounding the impact point wherein prior impact has removed the scale. Thus, the scale needs to be adherent and uncracked only in this local region for the parabolic kinetics to be valid. It does not matter that the scale is heavily cracked or spalled at the macroscopic level. The magnitude of Zb in comparison with steady state oxide scale thickness (Zss ) has an important bearing on the erosion oxidation interaction. Therefore, the variation of Zss and Zb with impact velocity is given in gure 22 for two different temperatures and two different feed rates. The magnitude of Zb is indicated for both spherical and conical erodents. The values of various parameters used for calculating Zb and Zss are given in table 3. Since most of the work on elevated temperature erosion is done with steel, the values appropriate to steel are chosen for various parameters. A value of 105 for Ao and 210 kJ mol1 for Q are chosen on the basis of the data reported by Quinn [93]. The value of C is calculated based on the formation of Fe2 O3 scale as observed by Levy and coworkers [29, 45, 59] for a variety of ferritic steel undergoing erosion oxidation degradation. The hardness of the oxide scale and the metal is assumed to be 3.0 GPa and 2.0 GPa, respectively. Zc is kept within the range of 110 m, consistent with values reported by various investigators [89, 90]. A perusal of the erosion literature indicates that the particle feed rate mostly lies in the range 0.110 kg m2 s1 . The erosion rate of the oxide scale is assumed to be proportional to V 3 . The test temperature is chosen in the range of 8731173 K since the oxidation effect becomes important at these temperatures in the case of steels. It is clear from gure 22 that Zss decreases much more rapidly with impact velocity than Zb . Thus a transition velocity beyond which Zss < Zb can be obtained for a given temperature and feed rate. R116

5.2. Conditions for prevalence of various EO mechanisms If there is no oxide scale on the metal surface or if the thickness of the oxide scale is very small compared with the depth to which the deformation extends in the metal surface, the dominant erosion mechanism will be metal erosion. On the other extreme, if the steady state thickness (Zss ) of the oxide scale is less than the critical thickness of spalling (Zc ) and the depth of deformation is lower than the steady state thickness, oxide erosion is the prevalent erosion mechanism. If, however, Zss is lower than Zc and the depth of the deformed zone is bigger than Zss , the oxidation affected erosion mechanism is observed. Finally, if the steady state thickness of the oxide scale is greater than the critical thickness, the oxide scale will never attain steady state thickness. The erosion will take place by spalling of the scale and oxidation controlled erosion will be operative. Roy et al [94], employing a new methodology, examined the transition criteria from the metal erosion regime to the oxide erosion regime. In their work, they eroded pre-oxidized Ni samples having varying thicknesses of oxide scale at ambient temperature. The ratio of the erosion rate of the oxide scale (Eo ) to the erosion rate of the substrate (E) is plotted against the ratio of the initial thickness of the oxide scale (t) to the depth of deformation (L) due to impact. Their observation is portrayed in gure 23. This gure reveals three distinct regimes. In region one, the value of t/L is higher than 4.0. The erosion rate of the oxide scale assumes a relatively high but constant value. In region two, t/L is higher than 0.5 but lower than 4.0. In this regime, there is a smooth change of the relative erosion rate from a low value to a high value. Finally, in regime three, t/L is lower than 0.5. In this regime, the rate of change of the relative erosion rate is slow. In addition, there appears to be a peak in the relative erosion rate at approximately t/L = 6. Regime one can be deduced as an oxide erosion regime on the basis of the erosion response, which is brittle, i.e. higher erosion rate at normal impact and on the basis of the velocity exponent, which is 3.0, of the erosion rate. Further, the material in this regime was removed by brittle chipping.

Topical Review

Figure 22. Variation of the calculated values of the steady state thickness and the thickness of the oxide scale grown between two successive impacts.

Figure 23. Variation of the ratio of the erosion rate of the oxide scale to the erosion rate of the substrate with the ratio of the thickness of the oxide scale to the depth of deformation [94].

In contrast, regime three is characterized by pure metallic erosion because the thickness of the oxide scale was negligibly low. In addition, the erosion response, the material removal mechanisms and the velocity exponent of the erosion rate in this regime are consistent with those of the erosion of metals and alloys at ambient temperature. Regime two represents the transition from the oxide erosion regime to the metal erosion regime. It also shows that the transition is not sharp but rather smooth. Thus as long as the depth of deformation is conned within the oxide scale, erosion takes place from the oxide scale only and erosion behaviour similar to oxide erosion is prevalent. When there is no oxide scale, metal erosion is

dominant. In the intermediate regime, erosion may take place from the oxide layer, but the deformation will extend to the substrate also. 5.3. Models for erosion oxidation interaction mechanisms Several investigators [57, 61, 95103] have tried to identify various possible mechanisms of interaction between erosion and oxidation over the last few years. For example, Hogmark et al [95] have identied six different mechanisms of interaction ranging from pure oxidation to pure erosion. Such mechanisms are compiled in the report of Stack et al [96] and Sundararajan and Roy [16]. Wellman and Nicholls [97] R117

Topical Review

presented an excellent review on this aspect. Barklow and Petit [98] have identied four different mechanisms such as metallic erosion, oxidation modied erosion, erosion modied oxidation and oxidation on the basis of the combined effect of kinetic energy of the impacting particles, oxide growth rate and particle ux rate. According to them the effect of the substrate becomes negligible when the particle impact energy and particle ux rate are less. The inuence of the substrate becomes important at high particle ux rate and high impact energy. Wright et al [52] classied the interaction between corrosion and erosion into two regimes on the basis of kinetic energies of the impacting particles. When the kinetic energy is low, erosion behaviour exhibits brittle erosion mechanisms characterized by a maximum erosion rate under normal impact and a velocity exponent in excess of 3.0. When the kinetic energy is high, erosion behaviour is ductile, maximum erosion rate occurs at oblique impact angle and the velocity exponent lies in the range 22.5. This classication was subsequently modelled by Natesan and Liu [99] and later modied by Stack et al [100]. Rishel et al [101] extended the idea originally proposed by Kang et al [50] and Chang et al [102]. They dened various mechanisms of erosionoxidation interaction using four parameters: instantaneous scale thickness ( ), parabolic rate constant under combined attack of erosion and oxidation Kce , erosion rate constant of oxidation product K and parabolic rate constant for oxidation (Kp ) only. According to these investigators the rate of change of instantaneous scale thickness is given by K(ce) d = (30) K . dt If d/dt is positive erosion of the oxide scale takes place. Erosion under such conditions will be characterized by a brittle erosion response. If d/dt is negative oxidation affected erosion becomes dominant, i.e. erosion takes place from a composite layer consisting of oxide scale, metallic substrate and erodent. If, however, d/dt is equal to zero erosion enhanced oxidation plays a dominant role. Within this regime three different modes are possible; Type I erosion enhanced oxidation will be operative if K(ce) is equal to Kp . Type II erosion enhanced oxidation can be seen if K(ce) is greater than Kp and Type III erosion enhanced oxidation occurs when the oxide scale spalls. Finally, if there is no corrosion, metal erosion will be prevalent. This model, in principle, can explain the effect of feed rate, particle size, etc on the observed erosion rate. However, the predictive capability of this model is limited. Sethi and Corey [54], with the help of oxide scale growth kinetics, have demonstrated that the temperature dependence of erosion rate shows three different regimes. These regimes are (1) a low temperature regime where erosion rate is independent of temperature, (2) an intermediate temperature regime where erosion rate increases with the increase of temperature and (3) a high temperature regime. Stephenson and Nicholls [103] have dened three different regimes on the basis of the ratio of contact radius (a) and scale thickness (z). If this ratio is less than 0.1 a substrate dominated regime can be seen. If this ratio is between 0.1 and 1.0, oxide modied behaviour will be prevalent. If this ratio is more than 1.0 oxidation dominated erosion will be the R118

operating mechanism. They also proposed the presence of pure oxidation in the case of the negligible presence of the erodent. Another classication due to Stack et al [104] proposes the presence of three regimes, namely, erosion dominated, erosioncorrosion dominated and corrosion dominated. These regimes are dened based on kinetic energy, temperature and critical oxide thickness. The transition to corrosion dominated behaviour can be attributed to the formation of critical oxide thickness. Above the temperature at which this occurs, the oxide formed in a given time interval cannot be removed from the scale metal interface by erosion. Below the temperature at which the critical scale thickness is attained, the oxide formed can be removed during impact. The corrosion dominated regime can be further divided into two more sub-regimes depending on the velocity dependence of these regimes. The corrosion dominated regimes can be corrosion dominated-1 and corrosion dominated-2 regimes. In the processes of identifying these mechanisms, Stack assumed that oxidation behaviour of the target material follows a parabolic oxidation behaviour and the oxidation that occurs during each impact is negligible. Sundararajan [78] has broadly proposed two different regimes of EO interaction, namely, erosion controlled regimes and oxidation controlled regimes. In erosion controlled regimes three different mechanisms can be envisaged. These mechanisms are (a) metal erosion, (b) oxidation affected erosion and (c) oxide erosion. In oxidation controlled regimes there are two different mechanisms, namely (a) oxidation controlled erosion continuous and (b) oxidation controlled erosion spalling. These zones can be identied on the basis of several erosion related parameters. The most important feature of this model is the ability of the model to predict the prevalent mechanisms of erosion once the conditions of erosion and the thermo-physical properties of the erodent and target material are known. 5.4. Erosion oxidation interaction map While establishing the operative mechanisms for erosion oxidation, it is noted that erosion conditions signicantly inuence the oxidation. The factors which inuence the operating mechanisms are impact velocity, impact angle, feed rate and temperature. Similarly, the particle size and shape also have a profound effect on this map. The combined inuence of all these factors on erosion rate remained qualitative for a considerably long time. Specic erosion mechanisms for various metallic materials have been described by Hogmark et al [95], Wright et al [52], Kang et al [50] and Sundararajan [78]. Barkalow and Petit [98] for the rst time have tried to organize such information in the form of an erosion oxidation map where the prevalence of various mechanisms is shown in the domain of the particle kinetic energy and the scale growth rate. Sundararajan [78] has attempted to organize erosion oxidation maps in the domain of impact velocity and temperature. Stephenson and Nicholls [77, 105] have plotted particle velocity versus oxide thickness for a specic particle size and temperature to construct such maps while Stack and Pena [106] plotted velocity versus temperature. But the attempt has remained limited with the consideration of theoretically postulated

Topical Review

values of erosion rates. Numerous experiments conducted by the author can be considered sufcient to attempt making erosion oxidation interaction maps from the experimental data. Because of the signicance of various erosion conditions on erosion rate and erosion mechanisms, an erosion oxidation interaction map can be represented in a multidimensional conguration. A simplied representation of the erosion oxidation map can be made in a two-dimensional plot of impact velocity and temperature for a given impact angle and feed rate. The inuence of feed rate on the EO map at normal impact for commercially pure Ni is shown in gure 24 whereas the inuence of impact angle on the EO map of Ni is illustrated in gure 25. Similarly the effect of Cr addition on the EO map can be noted in gure 26. In general all these maps describe the transitional boundary between regimes of various operative mechanisms of erosion at elevated temperature. These maps are able to clearly delineate the metal erosion, oxidation affected erosion, oxidation controlled erosion and oxide erosion. The extent of each of these regimes depends on temperature, feed rate, impact angle and impact velocity. An examination of all these maps shows that: (a) a low temperature imparts the metal erosion regime. With an increase in temperature the metal erosion regime shifts to the oxide erosion regime via the oxidation affected erosion and oxidation controlled erosion regimes, (b) a higher feed rate extends the metal erosion regime and in turn alters the existence and the extent of the other regimes and (c) oblique impacts tend to reduce the extent of the metal erosion regime and promote other regimes. The existence and extent of a particular regime is governed by the oxide scale growth and the depth of deformation due to impact. These two factors are opposing in nature with respect to erosion conditions. Thus for a given condition, keeping the impact velocity constant, as the temperature of erosion is increased, the inuence of oxide scale becomes dominant. As a result, the oxide scale starts growing at a rate faster than erosion. This leads to a situation where the thickness of the oxide scale becomes signicant compared with the depth of deformation and at still higher temperature where the thickness of the scale attains critical thickness of spalling or attains a thickness where the rate of growth of oxide scale is equal to the rate of erosion. This results in transition from metal erosion to oxidation affected erosion or oxidation controlled erosion or even to oxide erosion. This can be seen in gure 25 at impact angle of 30 and at impact velocity of 35 m s1 and at feed rate of 0.2 g min1 . Similarly keeping the temperature of erosion constant as the impact velocity is increased, it not only becomes more difcult for the oxide scale to grow but the thickness of the oxide scale also becomes less compared with the depth of deformation. This situation causes prevalence of metal erosion in preference to oxidation affected erosion and oxidation affected erosion in preference to oxidation controlled erosion, as shown in gure 24 (at impact angle of 90 , feed rate of 0.2 g min1 and temperature of 673 K). The effect of the higher particle feed rate is analogous to that of lower temperature. At a specic condition, higher particle feed rate does not permit the oxide scale to grow as the time intervals between impinging particles are shortened. This

(a)

(b)

Figure 24. Erosion oxidation interaction map for Ni, (a) for feed rate of 3.3 106 kg s1 and (b) for feed rate of 3.3 104 kg s1 .

condition results in an expansion of the metal erosion regime and the shifting of other erosion regimes to higher temperature or lower impact velocities. The effect of impact angle can be considered in terms of impact velocity. As the impact angle is decreased the normal component of the impact velocity decreases. Thus, oblique impact gives rise to constricted metal erosion regimes. Consequently, the oxidation affected erosion regime and the oxidation controlled erosion regimes appear at higher impact velocity and lower test temperature. At oblique impact, it is possible to see the presence of oxide erosion, which is not prevalent at normal impact in many cases. It is noted in gures 2426 that the erosion rate tends to increase and the erosion oxidation interaction mechanism shifts from metal erosion to oxide erosion via oxidation affected erosion and oxidation controlled erosion with the increase of temperature. It can also be inferred that the erosion rate is higher (1) in oxidation affected erosion than in metal erosion, (2) in oxidation controlled erosion than in oxidation affected erosion and (3) in the oxide erosion regime than in the oxidation controlled erosion regime [107]. It is important to R119

Topical Review
(a)

(a)

(b)

(b)

Figure 25. Erosion oxidation interaction map for Ni, (a) for impact angle of 30 and (b) for impact angle of 90 .

Figure 26. Erosion oxidation interaction map for (a) Ni and (b) Ni20 Cr alloy.

mention that the erosion oxidation map has been constructed experimentally for the rst time. The inuence of various erosion conditions on such a map can be explained on the basis of oxidation characteristics and mechanical properties of the eroding materials. Stack et al [108] constructed a similar map previously. However, such a map was available only in the low velocity regime. It also failed to highlight the inuence of other parameters such as particle feed rate, impact angle, alloy composition, etc.

6. An analysis of enhanced oxidation kinetics due to erosion induced roughness

The oxidation kinetics of metals and alloys are found to alter during erosion or wear. The observation on enhanced oxidation kinetics is primarily centred on iron base alloys. The inuence of sliding wear on oxidation was initially proposed by Quinn [109]. According to Quinn, the activation energy for the parabolic rate constant (KP ) remains the same during R120

static or wear induced oxidation. However, the magnitude of KP of wear induced oxidation is higher than that for static oxidation. A similar contention is also made by Lim and Ashby [87] for wear induced oxidation and by Levy and co-workers [29, 45, 59] for erosion induced oxidation. Roy et al [76] for the rst time attempted to estimate the altered oxidation kinetics during erosion and subsequently tried to formulate a phenomenological model to explain such a result. The observations of Roy et al [76] are presented in gures 27 and 28. The variations of mass gain per unit area as a function of time at 1073 K for as-received and eroded Ni are given in gure 27 whereas gure 28 depicts similar data at 1173 K for Ni20Cr alloy. It is clear that the oxidation rate of Ni increases signicantly for eroded samples when compared with as-received samples. The oxidation rate increases with the increase of impact velocity. The increase of oxidation rate with increase of impact velocity is higher at normal impact than at oblique impact. With regard to the Ni20Cr alloy it is noted

Topical Review

0.008 0.007 Mass Gain / Unit Area x 10 3 (Kg/m 2) 0.006 0.005 0.004 0.003 0.002 0.001 0.000 0 10 000 20 000 30 000 40 000 Time (sec) 50 000 60 000 70 000 80 000 Material: Ni V= Impact Velocity = Impact Angle Temperature : 1073 K As Received V = 35 m/s, = 30o V = 65 m/s, = 30o V = 35 m/s, = 90o V = 65 m/s, = 90o

Figure 27. Variation of mass gain per unit area of Ni with time at 1073 K [76].

18 16
3

14 12 10 8 6 4 2 0 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 Time (sec) Material: Ni-20Cr V= Impact Velocity = Impact Angle Temperature : 1173 K As Received V = 35 m/s, = 30 o V = 65 m/s, = 30 o V = 65 m/s, = 90 o V = 35 m/s, = 90 o

that the oxidation rate decreases with the increase of impact velocity. Further, the inuence of impact angle on oxidation kinetics is negligible. The above observation for Ni is modelled by assuming that the parabolic rate constant is related to the surface roughness (Ra ) or impact velocity as K = KO1 (Ra )x1 exp Q , RT (31)

Mass Gain per Unit Area x 10 (Kg / m 2)

Figure 28. Variation of mass gain per unit area of Ni20Cr alloy at 1173 K [76].

Q , (32) RT where KO1 and KO2 are constants, V is the impact velocity and is the impact angle, Q is activation energy, R is the universal K = KO2 (V sin )x2 exp

gas constant, T is test temperature and x1 and x2 are roughness exponent and velocity exponent, respectively. In order to establish the relation between the parabolic rate constant and impacting condition, the average activation energy for the parabolic rate constant is determined by plotting ln K (parabolic rate constant) against (1000 T 1 ) in K1 . Using the average activation energy, the average parabolic rate constants at different eroding conditions are estimated. The natural logarithm of the average parabolic rate constant is then plotted against ln Ra and ln V sin . The slopes of best-t straight lines are then computed as constants KO1 and KO2 whereas the intercepts of the lines with the ordinate are calculated as exponents x1 and x2. The expressions for rate constants are R121

Topical Review

Figure 29. Variation of the parabolic rate constant with (V sin )2.08 exp(Q/RT ) [76].

then obtained as K = 4.67 109 (V sin )2.08 exp K = 9.62 107 (V sin )4.26 exp 22 400 , RT 22 400 . RT (33)

(34)

In order to assess the ability of the expression to predict the parabolic rate constant, the estimated rate constants are plotted against (V sin )2.08 exp(22 400/RT ) in gure 29. A straight line with a good t suggests the suitability of the expression. A similar exercise cannot be carried out for Ni20Cr alloy because of two factors. First, the nature of the oxide scale that forms on Ni20Cr alloy can have a dual grain size distribution [110]. Second, higher roughness in these alloys results in lower sizes of the globular oxide grains. Hence even though the roughness changes the amount of grain boundary also gets altered and this in turn provides different extents of short circuit paths. Thus the pre-exponential factors in the Arrhenius type equation should be a function of the roughness, grain size and the nature of the oxides that are present in the oxide scales of Ni20Cr alloy. Hence a similar modelling for Ni20Cr alloy cannot be achieved.

7. Areas of future research

In spite of signicant progress in elevated temperature erosion of metals and alloys, certain areas still need to be addressed. The material ow behaviour during solid particle erosion is at high strain, high strain rate, under adiabatic deformation and under multiaxial stress condition. The material ow behaviour under such conditions, particularly at elevated temperature is not well understood. The estimation of mechanical properties of materials under such conditions is required for effective modelling of the elevated temperature erosion behaviour of materials. Thus there is a clear need to develop a simplied test technique which will simulate the erosion conditions in a R122

test sample and evaluate the mechanical behaviour under such conditions. For example the depth of deformation estimated using equation (14) would be more accurate if dynamic hardness is used instead of static hardness. The deformation and fracture behaviour of the oxide scale under erosion conditions is poorly understood. Further, the high strain rate ow behaviour of a layered structure containing the oxide scale and the substrate material also needs to be addressed. Such an understanding is required to model the spalling behaviour of the oxide scale once it reaches the critical thickness of spalling. The dramatically different spalling behaviour of the segmented scale and the compacted scale can be explained more accurately only when concepts related to such deformation behaviour get crystallized. The kinetics of oxidation during erosion are an order of magnitude higher than that under static conditions. An effort to estimate the increased oxidation rate quantitatively either through careful experiment or by rigorous theoretical analysis is hitherto unexplored. All the reasons for such accelerated oxidation are not yet revealed. Thus, there is a clear need to carry out organized experiments to evaluate the oxidation behaviour of materials under erosion conditions. Attempts should be made to separate out the inuence of possible factors responsible for enhancing the oxidation kinetics of metallic materials during erosion. Erosion oxidation interaction maps have been constructed experimentally for electrolytic Ni and Ni20Cr alloys. Such maps are to be constructed for a variety of engineering materials that are subjected to similar degradation through generation of a comprehensive database. Construction of such maps for several engineering materials subjected to erosion oxidation conditions is considered to be vital in future research in this area. One important drawback of erosion oxidation maps is that to characterize a system one needs a large number of maps to cover the whole range of erosion conditions. Stack and Pena [111] have recently produced a family of wear maps of velocity versus particle size for three different compositions, for four different temperatures, for three different particle sizes and three different WC percentages for NiCr/WC MMCs. One major challenge for future work is to develop these maps under suldizing and chlorinating environments. A serious deciency of our understanding of elevated temperature erosion is related to the lack of experimental data or theoretical analysis of the oxide scale behaviour under erosion conditions. For example, the inuence of the oxide thickness, the test temperature, the hardness of the substrate metal and the particle velocity on the plasticity and ductility of various oxides is least understood. The behaviour of oxide scale in the oxidation affected erosion regime needs to be fully understood. The conditions that lead to the formation of a composite unprotective layer on the one hand and the protective, pliable oxide skin on the other have to be investigated. In the oxidation controlled erosion regime, exhaustive experiments to characterize the effect of particle ux rate on the erosion rate should be undertaken. The elevated temperature erosion has to be carried out in a corrosive environment also. Signicant work has been done on erosion of materials in room temperature aqueous conditions. An excellent review on this work can be seen in [112]. However, experiments are to be carried out at

Topical Review

elevated temperature in aqueous conditions or in other gaseous corrosive conditions also in order to assess the synergy of erosion and corrosion at elevated temperature. An important advance in tribology and corrosion is the development of mapping techniques as shown in gures 2426. Based on rationale, mechanistic diagrams are constructed to delineate various tribocorrosion regimes as a function of tribocorrosion conditions [113117]. To date no effort is made to develop a parameter, which takes care of several tribocorrosion conditions. Further, there is no unied system to extrapolate the tribocorrosion map from one condition or processes to another condition or processes.

8. Concluding remarks
Solid particle erosion has been an area of increased scientic attention for the past few decades. Technological advancements in several high temperature applications such as gas turbine engines, power plants, etc lead to an increased interest in elevated temperature erosion. However, investigations on elevated temperature erosion of metals and alloys are very much limited. One reason for such a scarcity of published work in this discipline is the non-availability of a suitable test facility. The presence of oxide scale is responsible for the existence of a variety of material removal mechanisms during elevated temperature erosion of metallic material. These mechanisms have been carefully veried by experimental work. Four distinct material removal mechanisms, namely, metal erosion, oxidation affected erosion, oxidation controlled erosion and oxide erosion have been established experimentally. These experimental observations have been substantiated by theoretical consideration. A suitable criterion for transitions from one erosion regime to another regime is yet to evolve. Attempts have been made to analyse such transition using a pre-oxidized sample. The prevalence of various mechanisms has been found to be dependent on eroding conditions. The best way to demonstrate the inuence of these conditions on prevailing erosion mechanisms is to present erosion oxidation interaction mechanism maps. Though attempts have been made earlier to construct such a map theoretically, an experimental construction of such a map has been attempted only recently. The inuence of various eroding conditions on such mechanisms has been studied quite extensively for Ni and Ni20Cr alloys.

Acknowledgments
The author is grateful to Dr A M Sriramamurthy, Director, DMRL, for his constant encouragement and for granting permission to publish this work.

References
[1] [2] [3] [4] Neilson J H and Gilchrist A 1968 Wear 11 123 Tilly G P 1972 8th ICAS Cong. (Amsterdam) Finnie I 1962 Am. Soc. Test. Mater. STP 307 70 Forse C and Ball A 1983 Proc. 6th Int. Conf. on Erosion by Liquid and Solid Impact (Cambridge, UK, September 1983) ed J E Field and N S Corney

[5] Wright I G 1987 Mater. Sci. Eng. 88 261 [6] Tabakoff W and Hamed A 1984 Proc. 6th Int. Conf. on Erosion by Liquid and Solid Impact (Cambridge, UK, 1984) paper no. 54 [7] Sage W and Tilly G P 1969 Aeronaut J. 73 429 [8] Bitter J G A 1963 Wear 6 169 [9] Black P H 1972 Met. Eng. Q. 12 46 [10] Finnie I and Oh H 1966 Proc. 1st Cong. Int. Society of Rock Mechanics vol 2, p 99 [11] Summers D A 1979 Treatise on Material Science and Technology vol 16, ed C M Preece (New York: Academic) p 395 [12] Finnie I 1963 Wear 3 87 [13] Finnie J and McFadden D H 1978 Wear 48 18 [14] Sundararajan G 1983 Trans. Indian Inst. Met. 36 474 [15] Engel P A 1975 Wear 32 126 [16] Sundararajan G and Manish Roy 1997 Tribol. Int. 30 339 [17] Tilly G P 1979 Treatise on Material Science and Technology vol 13, ed D Scott (New York: Academic) p 287 [18] Kosel T H 1992 Metals Handbook vol 18 (Metals Park, OH: ASM International) p 199 [19] Levy A V 1995 Solid Particle Erosion and Erosion Corrosion of Materials (Materials Park, OH: ASM International) [20] Hutchings I M 1979 Proc. Conf. on Corrosion/ Erosion of Coal Conversion System Materials (NACE, Houston) ed A V Levy, p 393 [21] 1990 Metal Test Methods and Analytical Procedures, Annual Book of ASTM Standards (vol 03.02, Wear and Erosion, Metal Corrosion) (West Conshohocken, PA: ASTM) [22] Roy M and Sundararajan G 1998 DMRL Technical Report DMRL TR 98241 [23] Ruff A W and Ives L K 1975 Wear 35 195 [24] Hovis S K, Anand K, Conrad H and Scattergood R O 1985 Wear 107 69 [25] Bitter J G A 1963 Wear 6 5 [26] Tilly G P and Sage W 1968 Wear 16 447 [27] Levy A V, Yan J and Patterson J 1985 Proc. Int. Conf. on Wear of Material (Vancouver, Canada) ed K C Ludema (New York: ASME) p 708 [28] Gat N and Tabakoff W 1978 Wear 50 85 [29] Levy A V and Man Y F 1986 Wear 111 173 [30] Levy A V and Man Y F 1989 Wear 131 53 [31] Shayler P J and Yee K H 1985 Wear 98 127 [32] Finnie I, Levy A V and McFadden D H 1976 ASTM STP 664 ed W F Adler (West Conshohocken, PA: ASTM) p 36 [33] Kang C T, Chang S L, Petit F S and Birks N 1986 Proc. 3rd Berkeley. Conf. On Corrosion-Erosion-Wear of Materials at Elevated Temperature (Berkeley, CA) ed A V Levy [34] Tabakoff W and Vittal B V R 1983 Wear 86 89 [35] Zhou J and Bahadur S 1989 Corrosion and Particle Erosion at High Temperature ed V Srinivasan and K Vedula (Materials Park, OH: Minerals, Metals and Materials Society) p 315 [36] Singh T 1952 PhD Thesis Banaras Hindu University, Varanasi, India [37] Levy A V 1989 Corrosion and Particle Erosion at High Temperature ed V Srinivasan and K Vedula (Metals Park, OH: Metals and Materials Society) p 207 [38] Chang S L, Petit F S and Birks N 1990 Oxid. Met. 34 47 [39] Levy A V, Wang B W, Man Y F and Zee N 1989 Wear 131 85 [40] Levy A V and Wang B Q 1989 Wear 131 71 [41] Zhou J and Bahadur S 1991 Proc. CorrosionErosion Wear of Materials at Elevated Temperature (Berkeley, CA) ed A V Levy, p 13-1 [42] Shida Y and Fujikawa H 1985 Wear 103 281 [43] Singh T and Sundararajan G 1990 Metal. Trans. A 21 3187 [44] Wang B Q, Geng G Q and Levy A V 1992 Wear 159 233 [45] Levy A V, Slamovich E and Zee N 1996 Wear 110 117 [46] Levy A V and Man Y F 1989 Wear 131 39 [47] Chang S L, Petit F S and Birks N 1990 Oxid. Met. 34 71 [48] Levy A V and Wang B 1987 Surf. Coat. Technol. 33 285 [49] Geng Q, Wang B, Hou P and Levy A V 1991 Wear 150 89

R123

Topical Review

[50] Kang C T, Pettit F S and Birks N 1987 Metall. Trans. A 18 1785 [51] Barklow R H, Geobel J A and Petit F S 1979 Erosion: Prevention and Useful Application, ASTM, STP 664 ed W Edler (Philadelphia, PA: ASTM) p 163 [52] Wright I G, Nagarajan V and Stringer J 1986 Oxid. Met. 25 175 [53] Sethi V K and Wright I G 1989 Corrosion and Particle Erosion at High Temperature ed V Srinivasan and K Vedula (Warrendale, PA: TMS) p 245 [54] Sethy V K and Corey R G 1983 Proc. 7th Int. Conf. on Erosion by Liquid and Solid Impact (Cambridge, UK, September 1983) ed J E Field [55] Stott F H, Jordan M P, Lekatos S, Stack M M and Wood G C 1995 Wear 186187 291 [56] Chinnadurai S and Bahadur S 1995 Wear 186187 299 [57] Zhou J and Bahadur S 1995 Wear 186187 332 [58] Xie J and Walsh P M 1995 Wear 186187 256 [59] Drotlew A, Christodonlou P and Gutowsky V 1997 Wear 211 120 [60] Roy M, Ray K K and Sundararajan G 2001 Metall. Mater. Trans. A 32 143151 [61] Yu X Q, Fan M and Sun Y S 2002 Wear 253 604 [62] Hayashi N, Hasezaki K and Takaki S 2000 Wear 242 54 [63] Cursons A K and Hutchings I M 1983 Proc. 6th Int. Conf. On Erosion by Solid and Liquid Impact (Cambridge, UK, September, 1983) ed J E Field and N S Corney [64] Reddy A V and Sundararajan G 1986 Wear 111 313 [65] Mills D and Mason J S 1983 Proc. 6th Int. Conf. on Erosion by Solid and Liquid Impact (Cambridge, UK, September, 1983) ed J E Field and N S Corney [66] Young J P and Ruff A W 1977 J. Eng. Mater. Technol. Trans. ASME 99 121 [67] Anand K, Hovis S K, Conrad H and Scattergood R O 1987 Wear 118 243 [68] Sahoo M, Lui A, Morin G and Sikka V K 1981 Mechanical properties and erosion behaviour of nickel aluminide Report MTL 89-68 (J), Canada Centre for Mineral and Energy Technology (CANMET) [69] Kleis I 1984 Z Werkstofftech. 15 69 [70] Soderberg S, Hogmark H, Engman U and Swahn H 1981 Tribol. Int. 14 333 [71] Goretta K C, Arroyo R C, Wu C T and Routbort J L 1991 Wear 147 145 [72] Foley T and Levy A V 1983 Wear 91 45 [73] Behrendt A 1970 Proc. 3rd Int. Conf. on Erosion and Allied Phenomenon (Royal Aircraft Establishment, UK) ed A A Fyall and R B King [74] Roy M, Tirupataiah Y and Sundararajan G 1993 Mater. Sci. Eng. A 165 51 [75] Roy M, Tirupataiah Y and Sundararajan G 1995 Mater. Sci. Technol. 11 791 [76] Roy M, Ray K K and Sundararajan G 1999 Oxid. Met. 51 25772 [77] Stephenson D J and Nicholls J R 1995 Wear 186187 284 [78] Sundararajan G 1991 Wear 145 251

[79] [80] [81] [82] [83] [84] [85] [86] [87] [88] [89] [90] [91] [92] [93] [94] [95]

[96] [97] [98] [99] [100] [101] [102] [103] [104] [105] [106] [107] [108] [109] [110] [111] [112] [113] [114] [115] [116] [117]

Ives L K and Ruff A W 1978 Wear 46 149 Hutchings I M and Winter R E 1974 Wear 27 121 Hutching I M 1977 Int. J. Mech. Sci. 19 45 Bellman R and Levy A 1981 Proc. Int. Conf. on Wear of Materials (San Francisco, CA) (New York: ASME) p 564 Quadir T and Schewmon P G 1981 Metall. Trans. A 12 1163 Chrisman T and Schewmon P G 1979 Wear 52 57 Chrisman T and Schewmon P G 1979 Wear 53 145 Timothy S P and Hutching I M 1981 Proc. 7th Int. Conf. on High Energy Rate Fabrication (Leeds) Lim S C and Ashby M F 1987 Acta. Metall. 35 1 Levy A V, Slamovich E and Jee N 1986 Wear 110 117 Stephenson D J, Nicholls J R and Handcock P 1983 Proc. 6th Int. Conf. on Erosion by Liquid and Solid Impact (Cambridge, UK) ed J E Field and N S Corney, p 43 Saunders S R J and Nicholls J R 1989 Mater. Sci. Eng. 5 780 Hutchings I M 1997 J. Phys. D.: Appl. Phys. 10 L179 Tabor D 1951 The Hardness of Metals (Oxford: Clarendon) Quinn T F J 1983 Tribol. Int. 16 305 Roy M, Ray K K and Sundararajan G 1998 Wear 217 312 Hogmark S, Hammarsten A and Soderberg S 1979 Proc. 6th Int. Conf. on Erosion by Liquid and Solid Impact (Cambridge, UK, September 1979) ed J E Field and N S Corney Stack M M, Lekatos S and Stott F H 1995 Tribol. Int. 28 445 Wellman R G and Nicholls J R 2004 Wear 256 907 Barkolow R H and Petit F S 1979 Corrosion Erosion of Coal Conversion System Materials (Berkeley, CA) (Houston: NACE) Natesan K and Liu Y Y 1980 Mater. Sci. Eng. A 121 571 Stack M M, Stott F H and Wood G C 1993 Wear B 162164 706 Rishel M, Petit F S and Birk N 1987 Mater. Sci. Eng. A 143 1785 Chang S L, Petit F S and Birk N 1990 Oxid. Metal 34 23 Stephenson D J and Nicholls J R 1995 Wear 186187 284 Stack M M and Stott F H 1993 Corros. Sci. 35 1015 Levy A V and Yong-Fa Man 1989 Wear 131 53 Stack M M and Pena D 1999 Surf. Coat. Technol. 113 5 Roy M, Ray K K and Sundararajan G 1997 Proc. Int. Conf. on Corrosion (CORCON-97) (Bombay, India, December, 1997) p 588 Stack M M, Stott F H and Wood G C 1991 Mater. Sci. Technol. 7 1128 Quinn T F J, Sullivan J L and Rowson D M 1984 Wear 94 175 Caplan D and Sproule G I 1975 Oxid. Met. 9 459 Stack M M and Pena D 2001 Wear 251 1433 Stack M M 2005 Int. Met. Rev. 50 1 Stack M M, Zhou S and Newman R C 1995 Wear 186187 523 Stack M M, Corlett N and Zhou S 1990 Wear 225229 190 Stack M M, Corlett N and Turgoose S 2003 Wear 255 225 Stack M M, Corlett N and Zhou S 1998 Wear 214 175 Stack M M and Jena B D 2004 Wear 256 986

R124