S Shrestha and A Sturgeon, Cambridge, UK; T Hodgkiess, Glasgow, UK; A Neville, Edinburgh, UK
Paper 95 presented at ITSC 2002 International Thermal Spray Conference, 4-6 March 2002, Essen, Germany
The microstructure and aqueous corrosion characteristics of two types of HVOF sprayed coatings of the same generic Ni-Cr-Si-B self-fluxing alloy have been investigated in the as-sprayed condition. One of the coatings was studied also after post sealing with polymer impregnation in a vacuum and after vacuum-furnace fusion. Vacuum fusion was found to have a significant effect on the coating microstructure, by increasing the type and concentration of hardphase particles. The principal hardphase in the as-sprayed condition and polymer-sealed condition is chromium carbide, whereas molybdenum-containing boride phases are also found to have formed after vacuum fusion. The vacuum fusionpost-treatment eliminates splat boundaries, which can act as sites where preferential corrosion can occur, hence the dominant corrosion mechanisms change. In the as-sprayed and polymer-sealed coatings, localised attack at splatparticle boundaries and crevice corrosion dominate, whereas in the vacuum-fused coating the principal mechanism of corrosion is via micropitting as a result of hardphase loss.
Many potential applications of thermal spray coatings (TSCs) involve offering protection to components exposed to wear in an aqueous system. In such circumstances the corrosion characteristics of the coating must be considered. This is important for two reasons; firstly, when the component is required to remain immersed in an aqueous solution for periods during plant shutdown and secondly, it is known from studies on metallic materials [1,2] that corrosion can play a significant role in material degradation in aqueous environments where wear occurs.
The key issues relating to the corrosion behaviour of TSCs are the interconnected porosity and penetration of the liquid phase into the coating/substrate interface.  However, thermal sprayed coatings comprising ceramic hardphases embedded in a metallic matrix represent extremely complex materials from an electrochemical point of view and even in the absence of interconnected porosity there is scope for complex corrosion mechanisms to occur which will lead to deterioration of the coating. Indeed several studies by the authors have demonstrated on a variety of TSCs that the inherent corrosion resistance of the coating is a factor which must be considered when coatings are to be employed in aqueous systems. [4,5]
In this paper the corrosion characteristics of two coatings of the same generic Ni-Cr-Si-B self-fluxing type applied by the High Velocity Oxy-Fuel (HVOF) process are assessed in the as-sprayed condition. One of the coatings was also investigated after post spray treatment with polymer sealing and after vacuum furnace fusion. The paper focuses on the effect of corrosion behaviour in terms of electrochemical measurements and observed mechanisms in relation to the coating microstructure.
2 Experimental procedures
Two cermet coatings (designatd A and B) of the same generic type Ni-Cr-Si-B supplied by two commercial spraying contractors have been studied in this work. Both coatings were prepared using the JP-5000 HVOF system. The specified composition of coating A is Bal.Ni 17Cr 4Si 3.5B 1C and that of coating B is Bal.Ni 16Cr 3.5Si 3.5B 3Mo 3Cu 4Fe 0.8C.
Coating A was deposited onto a BS EN 10083-1 C40E carbon-steel and a stainless steel (UNS S31603) substrate. Coating B was deposited only onto a BS EN 10083-1 C40E carbon steel substrate. Microstructural characteristics and the static corrosion behaviour of the coatings were investigated. Coating A was studied in the as-sprayed condition, whereas coating B was studied in the as-sprayed condition, after sealing in vacuum (vacuum-sealed) using a polymeric sealant and also after a post-spray fusion in vacuum furnace (vacuum-fused). Vacuum fusion was conducted between 1050-1150°C, which was carried out commercially.
2.2 Corrosion tests
Specimens of 10x10mm were cut from the larger coated plates. Electrical wires were soldered to the back (steel) face of the samples prior to encapsulation in an epoxy mounting. The specimens were mounted in plane (i.e. with the coated face exposed) and cleaned with methanol prior to any testing.
Corrosion testing for coating A was carried out in a 3.5% NaCl solution. Coating B was tested in an artificial seawater solution. Electrolytes considered herein i.e. NaCl solution and the artificial seawater both yielded salinity of35,000ppm. The specimen-resin interface of the encapsulated specimen was painted with a sealing lacquer, 'Lacomit' to prevent interference in the electrochemical measurements from the substrate material. The test specimen was immersed in the electrolyte solution and the rest potentilal 'E corr' was allowed to settle for one hour prior to electrochemical testing, which utilised DC anodic polarisation scans using an electrochemical cell as described elsewhere.  The scan rate was 15mV/min. The electrode potential was measured using the saturated calomel electrode (SCE). The applied potential was reversed back to the rest potential once a current density of 500µA/cm 2 was reached. Additionally, free-corrosion experiments were conducted in which specimens were exposed to the electrolyte solution for periods of up to four weeks and subsequently examined without any accelerated electrochemical testing. All tests were performed at ambient temperature of 18°C.
2.3 Microstructural characterisation
Microstructural examinations of as-received coatings and specimens after corrosion tests were performed using optical and scanning electron microscopy (SEM). Quantification of the porosity from 100x100µm area was performed using an image analysis software. Energy dispersive x-ray (EDX) analyses were undertaken using the facility attached to the SEM, which also provided quantitative correction procedures for spot analyses. Spot analyses were undertaken on hardphase particles and matrix areas. X-ray diffraction (XRD) patterns were obtained from the coatings to identify crystalline phases present in the coating. Microhardness was measured using the Buehler Micromet ®-II hardness tester. Coating density was measured by weighing the samples before and after spraying and then calculating from the volume of the coating deposited. Surface roughness was measured using theTaylor Hobson surface profilometer.
3.1 Coating characteristics
The coating characteristics are given in Table 1.
Table 1. Coating characteristics
| ||Coating type||Coating porosity (%)||Vickers Hardness (HV 0.5)||Thickness (µm)||R a (µm)||Density (g/cm 3)|
Coating A displayed a dense microstructure shown by its low porosity amounting up to 1.9% and a microhardness in the range of 527-587HV. The surface roughness and density values measured for the as-sprayed coating A are similar to the values measured for the as-sprayed and vacuum-sealed coatings B, which are reported in the following section.
It can be seen that the as-sprayed and vacuum-sealed coatings displayed similar characteristics, Table 1. In contrast, the vacuum-fused coating resulted in a much lower porosity, a higher hardness and a higher density. The as-received surface of the coating after vacuum fusion is smoother (lower surface roughness, Ra)than the non-fused (as-sprayed and vacuum-sealed) coatings giving an indication of the consolidation effect of the post-treatment.
3.2 Coating microstructures
Examination of the polished cross section displayed coating formation with layers of lamellae, well bonded to the underlying steel substrate with little porosity as shown by the back-scattered SEM image, Fig.1a. The back-scattered SEM image of the polished coating surface showed abundance of black (low atomic number) globular and smaller grey hardphase particles, Fig.1b. EDX analysis of these particles, intervening matrix and bulk area of 100x100µm are shown in Table 2.
Fig.1. SEM back-scattered images of the cross section and polished surface of:
a, b) Coating A-sprayed
c, d) Coating B-sprayed
e, f) Coating B-fused
Table 2. Microanalysis on polished surface of as-sprayed coating A (values in wt.% taken from 5 analyses).
|Element||Matrix||Tiny grey hard phase||Black globular hard phase||Bulk|
XRD patterns confirmed the matrix to be a solid solution NiCrFe. The hardphase particles comprised nickel chromium silicide (Ni 16 Cr 6Si 7) and chromium carbide (Cr 23C 6).
As-sprayed/vacuum-sealed coating. Both as-sprayed and vacuum-sealed coatings showed similar microstructures and hence hereafter will be referred to as sprayed/sealed implying both types of coatings. Examination of the sprayed/sealed coating cross-section revealed low coating porosity and low interfacial porosity as shown by the back-scattered SEM image of the cross section, Fig.1c. The SEM image of the polished plane surface showed a clear abundance of black (low atomic number) globular hardphase particles, Fig.1d. EDX micro-analyses on the particles and the matrix are listed in Table.3.
Table 3. Microanalysis on polished surface of as-sprayed coating B (values in wt.% taken from 5 analyses).
|Element||Matrix||Globular hard phase||Bulk area|
The XRD confirmed the principal hardphase particles to be chromiun carbide (Cr 23 C 6) and no Mo-containing crystalline phase was identified. The metal matrix was nickel rich, identified by the XRD to be a solid solution of NiCrFe.
Vacuum-fused coating. The vacuum-furnace fusion altered the coating microstructure significantly, which was clearly discernible in both cross-section and planar views Fig.1e and f.
The more dense abundance of larger dark acicular hardphases than in the as-sprayed or vacuum-sealed coatings was quite clear in the back-scattered image of the coating, Fig.1f. In addition to these dark paricles, abundance of blocky white (high atomic number) hardphase particles were visible, which were found to be rich in molybdenum. The XRD confirmed these particles to be carbides (Cr 23C 6) and also borides (Cr 3NiB 6, FeMo 2B 2, Fe 2MoB 4), which were not present in the as-sprayed coating. Significant diffusion at the coating-substrate interface has resulted in the removal of the sharp coating/metal interface, which was visible in the as-sprayed or vacuum-sealed coatings and a reduction in interfacial porosity, Fig.1e. The coalescing of splat particles due to the vacuum-fusion process also resulted in removal of inter-splat boundaries in the main coating.
Vacuum-fused coating displayed an interdiffusion (mixing white) layer, where coating and substrate constituents have been mixed to a large extent Fig.1e. EDX results at spot positions of the coating after the fusion process are detailed in Table 4.
Table 4. Microanalysis on polished surface of vacuum fused coating B (values in wt.% taken from 5 analyses)
|Element||Matrix||Dark acicular phase||Blocky white phase||Main coating layer||Mixing white layer|
3.3 Electrochemical corrosion tests
Anodic polarisation plots of the as-sprayed coating A deposited on stainless steel and carbon steel substrates exhibited very similar passive behaviour shown by the low currents within a wide range of potential values positive from'E corr', Fig.2.
Fig.2. Anodic polarisation of the as-sprayed coating A in static 3.5% NaCl solution at 18°C
Coatings with stainless steel and carbon steel substrates displayed similar breakdown potentials 'E b' about +90 to +100mV (SCE). Beyond this potential a rapid increase of the current was observed. The hysterisis loops for both coatings are very much alike with very similar loop area. The maximum current density recorded was about 700µA/cm 2.
The typical anodic polarisation plots for the as-sprayed, vacuum-sealed and vacuum-fused coatings are displayed in Fig.3. It is found that there is no difference between the response of the three coatings in a saline solution at 18LC. All coatings exhibited very low currents on shifting the potential from 'E corr' for a potential range up to +100mV (SCE).
Fig.3. Anodic polarisation of the sprayed, sealed and fused coatings in static artificial seawater at 18°C
After the breakdown potential, 'E b' a rapid rise in current was observed indicating initiation of the corrosion attack. The scan reversed (i.e. the potential was shifted to more negative values) when a current density of 500µA/cm 2 was attained and the potential moved towards the rest potential. The maximum current reached on reversal of the potential scan is indicative of the extent to which corrosion can propagate once corrosion has initiated. It is apparent that although there is no difference in 'E b' and hence the ability for corrosion to initiate, it would appear that the vacuum-fused coating may be less susceptible to corrosion propagation as shown by the lower current density 'i max' attained during anodic polarisation, Fig.3.
3.4 Free corrosion tests
Polished specimens were exposed in the naturally corroding condition without any electrical connection with examination after 1, 2 and 4 weeks.
Corrosion products in the form of brown circular patches were observed at various areas of the exposed surface after 1-week immersion. This was further increased after 2-week exposure. After 4-week exposure, the entire exposed surface displayed corrosion attack over the surface with dull appearance and also evidence of severe crevice corrosion at the coating/lacomit interface. Severe pitting attack with loss of partcile splats at the crevice region is shown in Fig.4a.
Microscopic observation after 1-week immersion showed visible corrosion product on the exposed surface of the as-sprayed and vacuum-sealed coatings, but the vacuum-fused coating exhibited no significant change of its exposed polished surface. After 2 weeks of immersion, a little attack had initiated on the vacuum-fused coating. Specimen observed after 4-week period showed heavy corrosion product on both as-sprayed and vacuum-sealed coatings, but the vacuum-fused coating exhibited relatively much less corrosion attack. Corrosion attack at the crevice region on the sprayed and sealed coatings with pitting attack due to splat loss as mentioned for the coating A is shown in Fig.4b. Similar crevice area examined on the vacuum-fused displayed significantly less corrosion attack, Fig.4c.
3.5 Attack mechanisms
Analysis of the electrochemical measurements (anodic polarisation) by consideration of the parameter E b, which is most commonly used to indicate resistance to the onset of corrosion, does not show any difference in the relative performance of the as-sprayed, vacuum-sealed and vacuum-fused coatings. However, close examination of the extent of corrosion attack and the attack mechanisms shows that there is a difference, which accounts for the variations in 'i max' depending on the post-treatment of the coating.
As stated earlier, after fusion, no splat boundaries were detected and this has been shown in this study to have implications for the corrosion resistance of the coating. As-sprayed coating A and coating B both displayed macro pitting as a result of splat removal after 4-week exposure at 'E corr', Fig.4a and b. In addition to the corrosion attack at the splat boundaries, another mechanism observed was the attack at the hardphase/matrix interface, which results in micropitting due to loss of hardphase particles.
In relation to the vacuum-fused coating there were no visible splat boundaries and hence no corrosion initiation associated with individual splat regions occurred. In addition, there was visibly less severe crevice corrosion observed on the vacuum-fused coating compared with the sprayed/sealed coating. On the vacuum-fused coating it was evident that the most dominant mechanism of corrosion was in the form of very localised micropitting, which was clearly associated with the individual hardphase particles and their removal once corrosion of the surrounding matrix had occurred. Exposure of the vacuum-fused coating at 'E corr' after 4-week period revealed both corrosion of the matrix and preferential attack at the hardphase/matrix interface that resulted in micropitting, Fig.4c.
Fig.4. SEM back-scattered images showing corrosion attack at the crevice region:
Fig.4a) as-sprayed coating A
Fig.4b) as-sprayed coating B
Fig.4c) vacuum-fused coating B after 4-week exposure in static saline solution
Thermal spraying is becoming widely used in aqueous erosion-corrosion environments and because of this, there is impetus to improve the corrosion resistance of the coatings, [7-10] which were typically developed for mechanical wear resistance. In the early development, the principal concern regarding the corrosion behaviour of TSCs was considered to be their ability to provide a barrier between the corrosive fluid and the underlying substrate, which is often a low-grade material with poor corrosion resistance.
As a result, techniques were developed to post-treat the coating surface as a means of reducing, or even eliminating interconnected porosity such as polymer impregnation  and laser treatment.  As reported by Neville and Hodgkiess  and Shrestha et al.  , in current HVOF coatings, the level of interconnected porosity is very low and therefore corrosion by penetration of the coating is not the main concern. This was also shown in this work by similar anodic polarisation plotsobtained for the sprayed coatings deposited onto a carbon steel and a stainless steel substrates.
In the work described in the present paper, Ni-Cr-Si-B self-fluxing coating applied by the HVOF process was examined in the as-sprayed, vacuum-sealed and vacuum-fused conditions. The aspect of the coatings' corrosion behaviour underscrutiny was its inherent corrosion resistance, which due to the network of carbides and matrix, has the potential to be very complex. In agreement with other workers,  it has been shown that application of polymer sealant has no beneficial or detrimental effect on the corrosion behaviour of the coating. All the coatings possessed very low porosity (< 3%). Fusion of a thermal sprayed coating is normally performed using an oxyacetylene torch, where the residual porosity is high and there is a genuine need to seal and consolidate the coating. Fusion of HVOF coatings is less common due to their generally high quality. However, in this work the modification of the microstructure of the coating by the fusion process has been shown to affect the microstructure, micro-chemical composition and density and also some aspects of the corrosion behaviour and mechanisms.
Electrochemical corrosion behaviour
The anodic polarisation experiments exhibited passive behaviour for all coatings when they were initially exposed to the saline environment. However, it was apparent that complete passivity does not persist for long (1-2 weeks)under natural-corrosion exposure before localised corrosion is initiated.
These coatings, therefore, possess a considerably superior corrosion behaviour than carbon steels. The coatings indeed, behave in a rather similar manner to UNS S31603 stainless steel but have reduced passive potential range and fairly rapid initiation of localised corrosion attack in saline aqueous environment. The fairly early onset of corrosion on the coatings is clearly correlated with their complex microstructures as detailed below.
Corrosion on the non-fused coatings, which here include the as-sprayed and vacuum-sealed coatings, occurred by a mixture of mechanisms, where corrosion initiated at the microstructural features in a manner, which was essentially identical in the anodic polarisation and free-corrosion experiments.
Initiation of corrosion at splat boundaries of thermal sprayed coatings has been reported previously, where oxide 'stringers' were found to be responsible for providing a corrosion initiation site.  In the corrosion of the coating by the mechanism above, it would appear that, once corrosion is initiated on a particular splat, there is then a micro-galvanic action, which prevents subsequent corrosion of the nearest adjoining splats. The surface then corrodes non-uniformly at two levels: micropitting where the hardphase is lost due to corrosion of the supporting matrix and macropitting when the individual splat is removed. It is possible, in long-term exposures, that this macropitting mechanism might provide paths for the corrosive fluid to penetrate through the coating thickness to cause corrosion and coating adherence problems at the coating/substrate interface - especially with substrate materials of low corrosion resistance such as carbon steel.
Effect of post-spray fusion
The differences in electrochemical behaviour of the coating with and without fusion were very small. In terms of the resistance of the coating to corrosion initiation, denoted by the breakdown potential, there was no difference. More subtle differences were apparent in terms of the maximum currents attained during anodic polarisation. Corrosion of the vacuum-fused coating was contrasting in nature mainly due to the fact that there are no visible splat boundaries, which as discussed, above act as individual initiation sites for corrosion. Although this did not change the apparent resistance to passivity breakdown, as manifested by the similar 'E b' values, it could account for the slightly increased times for the initiation of corrosion of fused specimens in the free corrosion tests and apparent reduced intensity of attack up to 4 weeks as well as the lower maximum currents in the anodic polarisation experiments.
In the fused coatings, corrosion initiation was associated primarily with the hardphase particles. There was a higher density of hardphase particles after fusion and their interfaces with the matrix provided sites for corrosion to initiate. A change in the corrosion behaviour after fusion is not unexpected given the absence of splat boundaries and also the fact that fusion promotes formation of C and B-rich phases. These primarily tie up the key species, which confer corrosion resistance (Cr and Mo) and as clearly demonstrated by a comparison of Tables 2-4, thus leave the matrix denuded in these elements.  Nevertheless, such features clearly do not have a major impact on the corrosion resistance.
- Very similar corrosion behaviour was observed on as sprayed Ni-Cr-Si-B coatings supplied by two commercial spraying contractors. The coatings exhibited passive characteristics during electrochemical testing upon initial exposure to saline solution but suffer corrosion attack upon extended exposure.
- Corrosion of the Ni-Cr-Si-B coating without vacuum-fusion post-treatment occurs by a series of micropitting and macropitting mechanisms associated with the loss of hardphase particles and individual splats.
- Post-treatment of HVOF sprayed Ni-Cr-Si-B coating by polymer impregnation has been shown to have no effect on the corrosion resistance.
- Vacuum fusion treatment of the as-sprayed coating is beneficial in many aspects such as lowering both the coating and interfacial porosities, increasing the coating density and hardness and reducing the roughness of the as-sprayed surface.
- Vacuum fusion also alters the microstructure and microchemistry of the coating.
- Vacuum-fused coatings corrode by different mechanisms such as matrix corrosion and micropitting than do the as-sprayed or vacuum-sealed coatings but these mechanistic differences do not result in any significant difference in corrosion resistance.
- Neville, A, Hodgkiess, T and Dallas, J T: A study of the erosion-corrosion behaviour of engineering steels for marine pumping applications. Wear, 186-187 (1995), pp. 497/507.
- Neville, A and Hodgkiess, T: Study of effect of liquid corrosivity in liquid-solid impingement on cast iron and austenitic stainless steel, Brit. Corr. J., 32 (1997), No. 3, pp. 197/205.
- Ashary, A and Tucker, R C: Corrosion characteristics of several thermal spray cermet-coating/alloy systems. Surface & Coatings Technology, 49 (1991), pp. 78/82.
- Neville, A and Hodgkiess, T: Corrosion behaviour and microstructure of two thermal spray coatings. Surface Engineering, 12 (1996), No. 4, pp. 303/312.
- Neville, A and Hodgkiess, T: The Corrosion behaviour of a cermet coating applied by HVOF spraying. Proc. 1 st UTSC, 15-18 Sept., Indianapolis (1997), pp. 161/166.
- Shrestha, S, Neville, A and Hodgkiess, T: The effect of post-treatment of a high-velocity oxy-fuel Ni-Cr-Mo-Si-B coating. Part I: Microstructure/corrosion behaviour relationships. JTST, 10 (2001), No.3, pp. 470/479.
- Shrestha, S, Hodgkiess, T and Neville, A: The effect of post-treatment of a high-velocity oxy-fuel Ni-Cr-Mo-Si-B coating. Part II: Erosion-corrosion behaviour. JTST in press.
- Shrestha, S, Hodgkiess, T and Neville, A: The effect of post-treatment on the erosion-corrosion performance of HVOF sprayed coatings. 12 th Intl. Conf. on Wear of Materials (1999), Atlanta, USA.
- Hodgkiess, T, Neville, A and Shrestha, S: Mechanical and electrochemical interactions during erosion-corrosion of a high velocity oxy-fuel coating and a stainless steel. Wear, 233-235 (1999), pp. 623/634.
- Hodgkiess, T, Thye, N S and Shrestha, S: Some comparisons between the aqueous corrosion behaviour of a cermet coating and UNS S31603 stainless steel in alternate static and solid/liquid impingement conditions. Corrosion Odyssey 2001, NACE (UK) & The Inst. of Corrosion, 18-20 Sept., UK.
- Lugscheider, E, Jokiel, P, Messeschmidt, V and Beckschulte, G: Subsequent sealing of thermally sprayed coatings to increase corrosion resistance. IoM, London (1993), pp. 46/51.
- Garcia-Alonso, M C, Escudero, M L, Lopez, V and Macias, A: Characterisation and corrosion behaviour of laser surface alloyed Ni-Cr-W-Mo-Cu-C-B coatings. ISIJ Intl., 37 (1997), No. 2, pp. 161/168.
- Smith, M F, Dykhuizen, R C, and Neiser, R A: Oxidation of HVOF-sprayed steel. Proc. 1 st UTSC, ASM Intl., 15-18 Sept. Indianapolis (1997), pp. 885/893.
- Kretschmer, I, Heimgartner, P, Polak, R and Kammer, P A: New fusible alloys with enhanced corrosion resistance. Proc. 1 st UTSC, ASM Intl., 15-18 Sept. (1997), Indianapolis, pp. 199/202.