High energy solid particle erosion mechanisms of superhard CVD coatings

High energy solid particle erosion mechanisms of superhard CVD coatings

Wear 259 (2005) 135–144 High energy solid particle erosion mechanisms of superhard CVD coatings K. Bose, R.J.K. Wood ∗ , D.W. Wheeler School of Engin...

723KB Sizes 1 Downloads 97 Views

Wear 259 (2005) 135–144

High energy solid particle erosion mechanisms of superhard CVD coatings K. Bose, R.J.K. Wood ∗ , D.W. Wheeler School of Engineering Sciences, University of Southampton, Highfield, Southampton SO17 1BJ, UK Received 3 August 2004; received in revised form 24 January 2005; accepted 2 February 2005 Available online 10 May 2005

Abstract The high hardness of boron carbide (B13 C2 ) and diamond make them attractive candidates for use as erosion- and abrasion-resistant coatings in applications such as valves and pumps used in the off-shore oil industry. This paper compares the dominant solid particle erosion mechanisms of boron carbide and diamond coatings produced by chemical vapour deposition (CVD), when subjected to high energy particle impacts. To generate a range of impact damage features, a variety of erodents were used with differing hardness and shape. The erosion tests were performed on a gas blast erosion rig using spherical soda-lime glass beads, angular quartz silica sand and diamond grit, at impingement velocities between 130 and 270 m s−1 and an erodent particle flux of 0.5 kg m−2 s−1 . A range of techniques including optical interferometry, scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS), identified the damage mechanisms. For boron carbide, erosion occurred via removal of the coating through lateral/radial cracks generated by particle impacts. For the diamond coatings, the damage was in the form of stress-wave induced circumferential crack formation at delaminated regions of the coating leading to ejection of material within the cracks. The coatings were found to have a high threshold velocity for the initiation and propagation of the above damage features. Diamond was found to be highly resistant to the propagation of lateral–radial crack systems, which reflected in its superior erosion performance compared to boron carbide. © 2005 Elsevier B.V. All rights reserved. Keywords: Solid particle erosion; Superhard; CVD coatings; High energy particle impact

1. Introduction Rapid advances in coating deposition techniques such as physical and chemical vapour deposition (PVD and CVD) have led to the availability of a variety of superhard coatings with enhanced wear resistance for tribological applications. The high hardness and strength of CVD diamond have led to it being exploited as an erosion-resistant coating for aircraft and missile domes [1]. Similarly, the superior mechanical properties of CVD boron carbide, which has excellent neutron absorption properties, have led to it being used in nuclear and defence applications [2]. However, although hard coatings are ideal candidates for applications in highly erosive environments, their application has been limited by a lack of ∗

Corresponding author. Tel.: +44 238 059 4881; fax: +44 238 059 3230. E-mail address: [email protected] (R.J.K. Wood).

0043-1648/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.wear.2005.02.043

comprehensive and reliable understanding of their behaviour under these conditions. One particular environment with great potential for the use of these superhard wear resistant coatings is as valve materials in the off-shore industry where sand erosion can severely reduce the lifetime of valve components. The sand produced from oil and gas wells cause extensive wear of hydrocarbon transport and processing equipment. Choke valves, which regulate the flow of hydrocarbon fluids, are particularly susceptible to solid particle erosion. The dissipative nature of the choke valve renders it susceptible to sand erosion owing to the high flow rates that are generated, which can lead to sonic impingement velocities depending on whether the process stream is partially or wholly gaseous [3]. The flow pressure across the valve drops from about 700–300 bar in high pressure wells and although typical flow velocities at the inlet are about 30 m s−1 , veloci-

136

K. Bose et al. / Wear 259 (2005) 135–144

ties up to 500 m s−1 are reached inside the valve ports [4]. In oilfields that are in the latter stages of their operating lives the conditions are more severe owing to the increased quantities of sand particles that are carried in the hydrocarbon fluid as it is extracted from the reservoir. This paper reviews the high energy solid particle impact damage mechanisms that result in the erosion and failure of two superhard CVD coatings: boron carbide and diamond. In particular, it concentrates on establishing a unified approach towards the erosion performance of the two superhard coated systems by examining similarities in their underlying erosion mechanisms with a view to establishing a more generic understanding towards predicting the high energy solid particle erosion performance of superhard coating systems.

2. Erosion mechanisms of brittle materials: recent advances in the understanding An early study of the brittle fracture processes in glass was made by Wiederhorn and Lawn [7], who used hardened steel and tungsten carbide spheres at impact velocities of between 2 and 300 m s−1 . They found that at low velocities, well-defined cone cracks were produced upon impact; with increasing impact velocity, radial cracking was seen, eventually leading to lateral cracks observed on the glass surface. At high impact velocities (greater than ∼50 m s−1 for glass), the damage pattern was essentially independent of the impacting particle geometry and hardness as well as the impacted glass surface characteristics. A more recent study was made by Ballout et al. [8] who studied the erosion of S-glass using both sharp alumina and spherical glass particles. They reported that the predominant damage mechanism for both types of erodent was the formation of lateral/radial crack systems resulting in material removal, although the extent of damage and the resultant erosion rates was more severe with the sharper alumina particles. In a different approach, Gorham and Salman [9], studied these damage features using Hertzian quasi-static indentation studies on soda-lime glass using a spherical indenter. They also illustrated that under certain contact conditions both ‘sharp’ and ‘blunt’ damage were observed: Hertzian cone cracking was seen to coexist with lateral, radial and median cracks and with increasing contact stress the damage pattern was dominated by lateral/radial cracking. Erosion maps for brittle materials presented by Hutchings [10,11] contained general relationships for the elastic plastic transition and the transition to Hertzian fracture based on the impact conditions derived by Wiederhorn and Lawn [7]. In a separate map, the transition from the plastic regime to lateral/radial fracture based on the indentation studies of Chiang et al. [12] and Marshall [13] was presented with two transition lines derived, one for angular particles and one for rounded particles. However, Verspui et al. [14], from theoretical derivation and experimental studies on AF45 borosilicate

glass using zirblast B205 erodent of various sizes, show that the particle shape has no influence on the transition from plastic to lateral/radial fracture regime. Their derivation is based on work by Chiang et al. [12], which indicated that the only parameter governing complete crack pattern was the volume, δV, that the indenter (erodent) penetrates into the target. Verspui et al. [15] have since developed a Monte Carlo simulation model for the erosion of brittle materials under oblique impact, and the results show good correlation with their experimentally derived results. Slikkerveer et al. [16], in a recent study of ductile–brittle transitions in the erosion of silicon, have shown that the transition is related to the kinetic energy associated with impacts rather than the erodent shape. Subsequently, Wensik and Elwenspoek [17] have verified the existence of a threshold kinetic energy for the formation of lateral cracks on three brittle materials. These were soda lime glass, Pyrex glass and single crystalline 1 0 0 silicon using alumina erodent of between 3 and 29 ␮m in size for particle impact velocities between 75 and 200 m s−1 . They observed that the tendency to form lateral cracks was greater at higher impact kinetic energies. Moore and Wood [18] had earlier shown that it is useful to represent erosion rates in terms of the unit volume loss per impact (Vu , expressed as ␮m3 impact−1 ) as a function of the erodent kinetic energy, Ek . This enables the erosion performance to be mapped across a wide range of particle energies independent of particle size. From the above discussion, it is evident that the solid particle erosion mechanisms and the resulting erosion rates are predominantly governed by the kinetic energy of the impacting particles, and that at high erodent kinetic energy conditions the erodent shape and its hardness may not significantly affect the erosion mechanisms.

3. Experimental work 3.1. Coatings tested The erosion study was carried out on boron carbide and diamond coatings, which were produced by chemical vapour deposition. Fig. 1a and b shows typical SEM micrographs of representative boron carbide and diamond coatings, respectively. The boron carbide specimens were in the form of 25 mm × 25 mm squares, while the diamond specimens were 50 mm diameter specimens. Details of these coatings can be found in Table 1. The boron carbide coatings were deposited at a temperature of 1273 K and 1 × 103 Pa pressure. However, no details of the deposition conditions were provided by the coating suppliers for the diamond coatings. The boron carbide coatings were deposited onto tungsten carbide substrates while the diamond coatings were deposited onto tungsten substrates. On the boron carbide specimens, a 1 ␮m TiC interlayer was present between the coating and the substrate. No such interlayer was present between the diamond coating and substrate. One further difference between the two

K. Bose et al. / Wear 259 (2005) 135–144

137

available grades of quartz silica sand, which were blended together in order to reproduce the size distribution of the sand that is found in the North Sea Forties oil field. The erosion tests were performed using an air solid particle erosion rig the details of which can be found in an earlier publication by Wood and Wheeler [19]. Photographs of the Southampton solid particle erosion rig are shown in Fig. 2a and b. The erodents are accelerated in an air stream down a 16 mm stainless steel nozzle, 1 m in length, to impact on the samples. The rig is capable of generating particle impingement velocities of up to 300 m s−1 for particles upto 400 ␮m in size. The stand-off distance, from the end of the nozzle to the surface of the specimen, was 30 mm. The tests were periodically interrupted in order to weigh the samples to evaluate the cumulative mass loss against time and to calculate the resulting steady state erosion rates. All tests were performed at an erodent particle flux rate of 0.5 kg m−2 s−1 , corresponding to a feed rate of 6 g min−1 . This flux rate is below the level at which particle-particle interactions begin to significantly reduce the erosion rate [10]. For tests with the commercial blocky diamond erodents, short 30 s blasts were performed in order to generate single particle impact damage sites on the coating surfaces. The rationale for these short tests is discussed in the following section.

Fig. 1. SEM micrographs showing the CVD growth surfaces of (a) 18 ␮m boron carbide and (b) a lapped 33 ␮m diamond coating.

coatings is that the boron carbide coatings were tested in the as-grown condition, while the diamond coatings were lapped to a surface roughness (Ra ) of 0.2 ␮m following deposition. 3.2. Erosion test conditions In this study, three different types of erodent were used, details of which can be seen in Table 2. Both spherical and angular erodents were used in order to study the effect of particle morphology on the erosion damage features. The angular erodent used was a mixture of three commercially

4. Results and discussion 4.1. Impact damage and solid particle erosion mechanisms of CVD boron carbide and CVD diamond 4.1.1. Damage mechanisms induced by diamond grit The blocky diamond particles were used as the erodent in order to generate maximum damage in the boron carbide and diamond coatings as well as minimise fracture of the erodent. The greater hardness of the diamond grit, compared with the soda lime glass and silica sand (see Table 1), would result in a more severe damage to the coating surfaces and a corresponding reduction in fracture and fragmentation of the erodent particles.

Table 1 Details of the boron carbide and diamond coatings Coating

Nominal thickness (␮m)

Hardness (GPa)

Surface roughness (Ra ) (␮m)

Grain size (␮m)

Boron carbide

13–18

42–45 [5]

0.1–0.5 (as grown)

2–5

Diamond

10–60 600a

65–75 [6]

0.2 (lapped) 0.08 (polished)

10–30 200

a

Free-standing CVD diamond brazed onto tungsten carbide.

Table 2 Details of the erodents used in the erosion tests Erodent type

Hardness (GPa)

Mean size (␮m)

Particle size range (␮m)

Ek (␮J) at 250 m s−1

Spherical soda lime glass Sub-angular silica sand Blocky synthetic diamond grit

7 13 70

310 194 200

250–400 90–355 177–210

360 360 450

138

K. Bose et al. / Wear 259 (2005) 135–144

would result in strain energies that are sufficient to induce plastic deformation and subsequent fracture in the coatings. Hence, the impact damage caused by the ‘softer’ soda lime glass and silica sand erodents can be expected to be similar to those induced by the diamond grit, though the extent of damage observed would be greatest with the diamond grit for reasons explained above. The damage induced by the diamond grit could thus be considered to be representative of the most severe damage possible to the coatings in the impact energy regime of interest here. The tests consisted of short 10 s blasts of the diamond erodents at 250 m s−1 (Ek = 450 ␮J), which induced predominantly single impact damage sites. Fig. 3a and b shows two typical damage sites on the surface of an 18 ␮m boron carbide coating formed by the impact of a single diamond particle. Fig. 3a reveals typical radial–lateral brittle damage with lateral “c-type” cracks emanating outwards from the centre of the impact site with limited material removal being observed. Fig. 3b shows a fully developed lateral–radial crack damage

Fig. 2. (a) Photograph of the erosion test facility, showing the air reservoir, erodent hopper, erodent injector nozzle and the erosion chamber. (b) A view of the interior of the erosion chamber showing the 16 mm diameter nozzle and specimen holder.

Traditional approaches to modelling the solid particle erosion mechanisms in brittle materials [20–22], based on elastic and elastic plastic contact, suggest that the mechanisms of erosive wear may be completely different depending on whether the erodent is ‘softer’ than the target. However, the high impact velocity regimes of interest to the current study

Fig. 3. (a and b) SEM micrographs from the centre of typical impact damage sites formed on 18 ␮m CVD boron carbide after just 30 s by the blocky diamond grit, impacting at 250 m s−1 , showing significant brittle fracture, with possible exposure of the interface.

K. Bose et al. / Wear 259 (2005) 135–144

139

Fig. 4. Nanoindentation loading–unloading curves showing the less hard phases in the CVD boron carbide coating compared with those from the B13 C2 phase.

system with significant material loss observed at the centre of the impact pit. There are two possible explanations for the greater degree of damage observed in Fig. 3b. They are that the impact has occurred on (i) an area of the coating where the compressive residual stress is lower; (ii) an area rich in one of the lower hardness phases of boron carbide, such as B50 C2 . Supporting evidence for the latter possibility is provided by nanoindentation data (see Fig. 4), which show variations in the hardness of the coating. These variations in hardness have been attributed to deviations from the ideal stoichiometric composition of B13 C2 (13.3% C). These regions are rich in either

Fig. 5. A region of the CVD boron carbide surface formed by impacting diamond grit showing overlap of emanating lateral cracks from neighbouring impact pits resulting in the formation of ‘channel-like’ regions.

boron or carbon resulting in a lower hardness and, therefore, lower resistance to plastic deformation and failure on impact. The lower hardness is attributed to diminishing bond strength with excess boron, while increasing carbon content leads to the formation of free graphitic carbon at grain boundaries. Another distinct feature of the impact damage on the boron carbide surface is that of “c-cracks”, which radiate from some of the impact sites. When two or more of these cracks from adjacent impact sites intersect, material is removed leaving

Fig. 6. EDS elemental map from a typical impact pit formed by impacting diamond grit at 250 m s−1 on 18 ␮m CVD boron carbide (a) showing exposure of (b) the TiC interlayer and (c) the tungsten substrate.

140

K. Bose et al. / Wear 259 (2005) 135–144

that high energy impacts from diamond grit can generate elastic–plastic damage in diamond. In both cases, radial and lateral cracks can be seen radiating from the centre of the impact sites.

Fig. 7. Scanning electron micrograph of the surface of 600 ␮m CVD diamond impacted by blocky synthetic diamond grit at 268 m s−1 .

‘channel-like’ features in the remaining coating. An example of these features can be seen in Fig. 5. The formation of these features on the boron carbide eroded surface contributes further to material loss from the lateral crack systems as erosion progresses. Interestingly, the depth of the impact pits were found to be of the order of the coating thickness and at their centre delamination from the interlayer-substrate interface was observed. This was confirmed by the energy dispersive spectroscopy (EDS) surface maps of the impact pits (shown in Fig. 6), where the presence of titanium and tungsten peaks from the exposed interlayer and substrate was observed. Twodimensional (2D) optical interferometry profiles (using a Veeco NT 3100 Optical Profiler) from several impact pits confirmed that none of the impact pits extend beyond the interface. It appears that the impact pits are arrested at the interface and further damage occurs predominantly through a lateral expansion of these pits. Figs. 7 and 8 show electron micrographs of the surface of diamond following impact from blocky synthetic diamond grit at 268 m s−1 . They show that, in contrast to silica sand,

Fig. 8. Scanning electron micrograph of the surface of 600 ␮m CVD diamond impacted by blocky synthetic diamond grit at 268 m s−1 .

4.1.2. Damage induced by the glass and sand erodents Prolonged erosion tests using the two different sized spherical glass and angular sand erodents at an identical impingement velocity of 250 m s−1 showed similar impact damage features on the boron carbide surface to those generated by the diamond grit. However, whereas radial/lateral cracks were formed from single diamond impacts, when the soda lime glass and silica sand were used large numbers of repetitive impacts around the site of damage were required for their initiation. Coatings eroded at 250 m s−1 for 30 s by both the glass and sand erodents revealed no damage of this nature. Instead, the impact pits begin to form after approximately 5 min of the erosion test. Fig. 9 shows one such impact pit formed on an 18 ␮m boron carbide coating after 10 min from erosion by 310 ␮m glass beads at 250 m s−1 . The progressive formation of impact pits from repeated impacts by particles of inferior hardness (i.e. soda lime glass and silica sand) is a consequence of the fragmentation of these particles. A higher proportion of the impact kinetic energy of these erodents is dissipated through fracture of the erodent themselves, thus reducing the energy available for crack initiation and propagation. Used glass particles were collected from the erosion chamber following impact at 250 m s−1 on an 18 ␮m boron carbide. An electron micrograph of a sample of these particles can be seen in Fig. 10 and shows extensive brittle fracture of the erodents. Laser diffraction based size analysis using a Malvern Mastersizer 2000 indicated that the mean erodent size (initially 310 ␮m) had reduced to approximately 130 ␮m. A similar degree of degradation was noted for the silica sand erodents.

Fig. 9. SEM micrograph of an impact site formed on an 18 ␮m CVD boron carbide during the early stages of a test, after 10 min on erosion by 250–400 ␮m spherical glass at 250 m s−1 showing emanating lateral cracks.

K. Bose et al. / Wear 259 (2005) 135–144

141

Fig. 12. 3D optical profile from the centre of an 18 ␮m thick eroded B13 C2 after 150 min, eroded by 250–400 ␮m spherical glass at 250 m s−1 showing several impact sites along the surface. Fig. 10. SEM micrograph of the post-test 250–400 ␮m spherical glass erodent showing severe brittle fracture after 10 min at 250 m s−1 impact velocity.

As erosion progresses, the impact pits expand laterally within the coating, which results in a single stage erosion material loss. Fig. 11 shows a typical impact pit formed after 150 min of erosion showing a fully developed impact erosion pit resulting from radiating lateral cracks. The site has a diameter of approximately 150 ␮m and with the overall area of particle impingement on the test specimen being 18–20 mm in diameter the number of erodents impacting within a 150 ␮m erosion pit is approximately 2000 after 2.5 h (using a simple erodent mass flux calculation). The overlapping of adjacent lateral cracks emanating from neighbouring impact pits, similar to those in Fig. 5, are also observed on the eroded surface from impacting 310 ␮m glass erodents. The resultant ‘channels’ of erosion material loss increase in intensity as erosion progresses. Fig. 12 shows a 3D optical profile of a region from the eroded surface of an 18 ␮m boron carbide coating after 150 min of erosion by 310 ␮m glass at 250 m s−1 . The site shows extensive formation of

Fig. 11. Scanning electron micrograph of a typical erosion impact site after 150 min by 250–400 ␮m glass beads impacting a 250 m s−1 showing typical lateral “c-cracks” propagating outwards from the centre of the impact. The specimen has been tilted at an angle of 30◦ .

channels on the surface exaggerating progressive material loss from the coating. 2D optical profiles across several erosion pits revealed that even on prolonged erosion, the impact pits are arrested at the interlayer/substrate interface, where delamination occurs. Further erosion occurs only though the lateral expansion of individual impact pits and through the formation of channels between interacting pits, i.e. erosion occurs wholly within the coating and the tungsten carbide substrate is protected from erosion damage. However, subsequent exposure to high energy erodent will result in substrate damage. The appearance of the impact pits generated on the surface by impacting glass and sand erodents suggested that the interfacial delamination was a precursor to the lateral expansion of impact pits. This was verified by EDS, which revealed the presence of titanium and tungsten peaks from the interlayer and substrate in a similar manner to those seen in Fig. 6. At high impact energy erosion conditions the formation and appearance of the erosion impact pits was found to be independent of the erodent shape. Fig. 13 shows an erosion impact pit formed by impacting 194 ␮m angular sand erodents on an 18 ␮m boron carbide surface after 10 min of erosion at

Fig. 13. A typical impact pit formed after 10 min or erosion at 250 m s−1 using the 90–355 ␮m angular silica sand showing a similar damage mechanism to those observed with spherical erodents.

142

K. Bose et al. / Wear 259 (2005) 135–144

250 m s−1 . The pit is almost identical to that in Fig. 9, formed by 310 ␮m glass particles impacting at the same velocity. However, the observed erosion rates were slightly higher for the angular sand compared to those for spherical glass (see Section 4.5, Fig. 16) which suggests a greater extent of damage by these impacts pits as a result of higher contact stresses generated by the angular nature of the sand erodents. No elastic–plastic damage was seen on the surfaces of the diamond coatings eroded by the silica sand particles. Instead, the diamond coatings underwent a three-stage process of damage accumulation starting with the build-up of delamination at the coating-substrate interface, which was caused by sub-surface shear stresses generated by the particle impacts [23]. This delamination has been shown using scanning acoustic microscopy [24]. Once significant delamination had taken place, the next stage was the formation of circumferential cracks caused by stress-wave reflection at these locally debonded regions [25]. An example of a circumferential crack can be seen in Fig. 14. Further particle impacts then cause material contained within these circumferential cracks to be ejected, resulting in the formation of a “pin-hole”; a typical pin-hole is seen in Fig. 15. These pin-holes increase in depth until the coating is penetrated through to the substrate. The pin-holes, which are visible to the naked eye, thus provide a useful visual indication that the diamond coating is becoming delaminated. 4.2. Summary of erosion mechanisms Solid particle impact erosion for both coatings occurs by elastic–plastic damage resulting in the formation of brittle radial–lateral crack systems when impacted by the diamond erodents, i.e. for low target to erodent hardness ratios, HT /HE (HTBC /HEDIA = 0.67, HTDIA /HEDIA = 1). For impacting glass and sand erodents on CVD boron carbide at high energies, erosion still occurs through the formation of radial–lateral cracks, when the HT /HE ratios are HTBC /HEG =

Fig. 15. Electron micrograph taken from a 60 ␮m diamond coating tested at 268 m s−1 using 90–355 ␮m sand for 9 h showing a pin-hole.

6 and HTBC /HESAND = 3.3 for glass and sand erodents, respectively. However, for sand impacts on diamond, where the HTDIA /HESAND = 5, limited elastic plastic deformation in diamond is observed and erosion occurs by stress-wave reinforced delamination of the coating from the surface leading to the formation of ‘pin-hole’ type damage. Previous work on brittle materials [26,27] indicates that at target-toparticle hardness ratios around unity, a marked transition in the damage mechanism is observed. At high energy impacts, lateral/radial cracks in the two CVD coatings are observed even at high target-to-particle hardness ratios of 6, except in the case of impacting sand on CVD diamond. This indicates that the damage mechanisms are not simply determined by the target-to particle hardness ratio, but also depend on the actual hardness value of the target itself. 4.3. A unified energy based model of solid particle erosion of the two CVD coatings The high energy solid particle erosion of both CVD boron carbide and CVD diamond have been shown to occur by a single stage damage mechanism, independent of erodent shape, when brittle elasto-plastic damage, by the formation of radial–lateral crack systems, of the coating is observed. This single stage damage mechanism make it convenient to represent the solid particle erosion process by a unified energy based model describing the synergistic nature of damage in the erodent-target combination. If EEP is taken to be the total energy dissipated from elastic–plastic damage from a C and EE are the fractions dissipated single impact, while EEP EP through coating and erodent damage, respectively: C E EEP = EEP + EEP

Fig. 14. Electron micrograph taken from a 60 ␮m lapped diamond coating tested at 268 m s−1 using 90–355 ␮m angular silica sand for 5 h showing a circumferential crack.

E comA diamond erodent impact will have a lower EEP C pared to glass or sand and hence a higherEEP . Assuming a similar elastic–plastic damage mechanism with all erodents, it follows that a greater number of repeated erodent impacts would be needed to induce a similar level of erosion loss for a softer erodent. Obviously, for the case of CVD diamond

K. Bose et al. / Wear 259 (2005) 135–144

Fig. 16. Erosion rates expressed in terms of Vu of B13 C2 for two erodent kinetic energies Ek (360 and 110 ␮J), compared with other previously tested candidate erosion resistant valve materials, 45 ␮m CVD diamond on W substrate, A WC–7% Ni, and stainless steel type AISI 316, respectively.

eroded by the softer sand erodent, owing to a high target to erodent hardness ratio (∼5), limited elastic–plastic damage is induced in CVD diamond and brittle fracture of the coating is suppressed under these conditions. 4.4. Erosion rates of the boron carbide and diamond coatings Details of the erosion rates of boron carbide and diamond for the range of impact velocities and erodent types tested on the erosion rig have been reported elsewhere [28–30]. Optimum coating thicknesses for maximum erosion life were also discussed. Fig. 16 summarises the erosion performance of the two coatings (18 ␮m boron carbide and 45 ␮m diamond), expressed in terms of Vu at two different erodent kinetic energies, Ek (360 and 110 ␮J). The erosion rates are compared with other candidate erosion resistant valve materials tested on the erosion rig. These candidate materials include a sintered WC–7% Ni and AISI stainless steel type 316. Both CVD coatings outperform both the stainless steel 316 and sintered WC, offering a significant improvement in erosion resistance to WC and stainless steel. 4.5. Effect of impact velocity on the erosion mechanism The effect of lower particle velocities was also examined in order to see if any differences in erosion mechanism could be seen. Erosion tests were conducted using velocities of 200, 150, 100 and 50 m s−1 . However, the erosion mechanism appeared to be unchanged at these lower velocities. However, below 150 m s−1 little material loss was observed; the pits that were observed on the eroded surfaces were smaller in size than those formed on impact at 250 m s−1 . The number of impact erosion pits formed at lower erodent impingement velocities was also much reduced, although the number of impacts required to initiate the pits was similar. This was confirmed by visual examinations by interrupting tests after an initial erodent exposure of 2 min at each velocity.

143

Fig. 17. The number of erosion impact sites formed on 18 ␮m CVD boron carbide and their average diameter after 150 min in a 2.5 mm × 1.9 mm area in the centre of the erosion scar as a function of the impacting erodent velocity.

Both these factors, i.e. smaller size of the average impact erosion pits and the reduced number of impact pits formed, contribute to the lower erosion rates at lower impact velocities. Fig. 17 shows the number of erosion impact sites formed, together with the average impact pit diameter, after 150 min in a 2.5 mm × 1.9 mm area in the centre of the erosion scar as a function of the impacting particle velocity.

5. Conclusions This paper has examined the erosion of boron carbide coatings on tungsten carbide in an attempt to understand the damage mechanisms by which these coatings degrade in erosive environment. The performance and behaviour of diamond coatings on tungsten and tungsten carbide has also been considered and compared with those of boron carbide. Examination of the eroded coatings has revealed that on impact by diamond grits, both boron carbide and diamond show significant elasto-plastic damage from single impacts. However, when the softer soda lime glass and silica sand were used to impact boron carbide, while similar damage features were observed, a greater number of impacts was required to accumulate this damage, owing to extensive fragmentation of the erodents on impact. The higher hardness of diamond, compared to boron carbide, results in greater particle fragmentation on impact. Consequently, no elastic–plastic impact damage was observed on diamond coatings eroded by glass and silica sand. Instead, the coatings degrade by the build-up of delamination at the coating-substrate interface caused by sub-surface shear stresses generated by the impacting particles. The delamination enables the formation of circumferential cracks at these regions, which are caused by reflection of stress waves. On the basis of the results it is proposed that the solid particle impact can be described on the basis of an energy approach so that: C E EEP = EEP + EEP

144

K. Bose et al. / Wear 259 (2005) 135–144

where EEP is taken to be the total energy dissipated from C elastic–plastic damage caused by a single impact, whileEEP E and EEP are the fractions dissipated through coating and eroC is less than dent damage, respectively. For diamond, EEP for boron carbide: as a result, the brittle elastic–plastic (lateral–radial) damage is suppressed.

Acknowledgments The authors would like to thank the School of Engineering Sciences at the University of Southampton for funding the studentship for this work. In addition the assistance of the technical staff in maintaining the erosion test facilities is also gratefully acknowledged.

References [1] E.J. Coad, C.S.J. Pickles, G.H. Jilbert, J.E. Field, Aerospace erosion of diamond and diamond coatings, Diamond Relat. Mater. 5 (1996) 640–643. [2] K.W. Lee, S.J. Harris, Boron carbide films grown from microwave plasma chemical vapour deposition, Diamond Relat. Mater. 7 (1998) 1539–1543. [3] A. Forder, M. Thew, D. Harrison, A numerical investigation of solid particle erosion within oilfield control valves, Wear 216 (1998) 184–193. [4] K. Haugen, O. Kvernvold, A. Ronald, R. Sandberg, Sand erosion of wear-resistant materials: erosion in choke valves, Wear 186–187 (1995) 179–188. [5] L. De Fazio, S. Syngellakis, R.J.K. Wood, F.M. Fugiuele, G. Scium´e, Nanoindentation of CVD diamond: comparison of an FE model with analytical and experimental data, Diamond Relat. Mater. 10 (2001) 765–769. [6] K. Bose, Tribological characterisation of CVD boron carbide, Ph.D. Thesis, University of Southampton, 2004. [7] S.M. Wiederhorn, B.R. Lawn, Strength degradation of glass impacted with sharp particles, J. Am. Ceram. Soc. 62 (1979) 66–70. [8] Y. Ballout, J.A. Mathis, J.E. Talia, Solid particle erosion mechanism in glass, Wear 196 (1996) 263–269. [9] D.A. Gorham, A.D. Salman, Indentation fracture of glass and mechanisms of material removal, Wear 233–235 (1999) 151–156. [10] I.M. Hutchings, Ductile–brittle transitions and wear maps for erosion and abrasion of brittle materials, J. Phys. D 25 (1992) A212– A221.

[11] I.M. Hutchings, Transitions, threshold effect and erosion maps, Key Eng. Mater. 71 (1992) 75–92. [12] S.S. Chiang, D.B. Marshall, A.G. Evans, The response of solids to elastic/plastic indentation I: stresses and residual stresses, J. Appl. Phys. 53 (1) (1982) 298–311. [13] D.B. Marshall, Geometrical effects in elastic/plastic indentation, J. Am. Ceram. Soc. 67 (1) (1984) 57–60. [14] M.A. Verspui, P.J. Slikkerveer, G.J.E. Skerka, I. Oomen, G. de With, Validation of the erosion map of spherical particle impacts on glass, Wear 215 (1998) 77–82. [15] M.A. Verspui, G. de With, A. Corbijn, P.J. Slikkerveer, Simulation model for the erosion of brittle materials, Wear 233–235 (1999) 436–443. [16] P.J. Slikkerveer, P.C.P. Bouten, F.H. in’t Veld, H. Scholten, Erosion and damage by sharp particles, Wear 217 (1998) 237–250. [17] H. Wensink, M.C. Elwenspoek, A closer look at ductile brittle transition in solid particle erosion, Wear 253 (2002) 1035–1043. [18] A.J. Moore, R.J.K. Wood, Erosive wear mapping of pipeline materials. Plastic Pipes VIII, The Plastic and Rubber Institute, Koninshof, The Netherlands, 21–24 September 1992, pp. 1–10, paper E1/4. [19] R.J.K. Wood, D.W. Wheeler, Design and performance of a high velocity air–sand jet impingement erosion facility, Wear 220 (1999) 95–112. [20] J.E. Goodwin, W. Sage, G.P. Tilly, Study of erosion by solid particles, Proc. Inst. Mech. Eng. 184 (1969) 279–292. [21] P.H. Shipway, I.M. Hutchings, The role of particle properties on the erosive wear of sintered boron carbide, Wear 149 (1991) 85–98. [22] P.H. Shipway, I.M. Hutchings, The role of particle properties in the erosion of brittle materials, Wear 193 (1996) 105–113. [23] D.W. Wheeler, R.J.K. Wood, CVD diamond: erosion-resistant hard material, Surf. Eng. 19 (2003) 466–470. [24] D.W. Wheeler, R.J.K. Wood, High velocity sand impact damage on CVD diamond, Diamond Relat. Mater. 10 (2001) 459–462. [25] D.W. Wheeler, Solid particle erosion on CVD diamond coatings, Ph.D. Thesis, University of Southampton, 2001. [26] S. Wada, W. Watanabe, Solid particle erosion in brittle materials. Part 3. The interaction with material properties and that of the impingement particle on erosive wear mechanism, Yogyo-Kyokai Shi (1987) 573–578. [27] S. Srinivasan, R. Scattergood, Effect of erodent hardness on erosion of brittle materials, Wear 128 (1988) 139–152. [28] R.J.K. Wood, D.W. Wheeler, D.C. Lejeau, B.G. Mellor, Sand erosion performance of CVD boron carbide coated tungsten carbide, Wear 233–235 (1999) 134–150. [29] D.W. Wheeler, R.J.K. Wood, Erosive wear behaviour of thick chemical vapour deposited diamond coatings, Wear 225–229 (1999) 523–536. [30] K. Bose, R.J.K. Wood, High velocity solid particle erosion behaviour of CVD boron carbide on tungsten carbide, Wear 258 (2005) 366–376.