High Temperature Corrosion Testing of HVOF Coatings

TWI, Cambridge, United Kingdom

Paper presented at ITSC 2012 International Thermal Spray Conference, Houston USA, 21-24 May 2012, and published in Journal of Thermal Spray Technology.

Abstract

This paper reports the corrosion behaviour of coatings deposited by high velocity oxy-fuel (HVOF) spraying and representative boiler substrate alloys in simulated high temperature biomass combustion conditions. Four commercially available oxidation resistant Ni alloy coating materials were selected: NiCrBSiFe, alloy 718, alloy 625 and alloy C-276. These were sprayed onto P91 substrates using a JP5000 spray system. The corrosion performance of the coatings varied when tested at 525, 625 and 725°C in K₂SO₄-KCl and gaseous HCl-H₂O-O₂ containing environments. Alloy 625, NiCrBSiFe and alloy 718 coatings performed better than alloy C-276 coating at 725°C, which had very little corrosion resistance resulting in degradation similar to uncoated P91. Alloy 625 coatings provided good protection from corrosion at 725°C, with the performance being comparable to wrought alloy 625, with significantly less attack of the substrate than uncoated P91. Alloy 625 performs best of these coating materials, with an overall ranking at 725°C as follows: alloy 625 > NiCrBSiFe > alloy 718 >> alloy C-276. Although alloy C-276 coatings performed poorly in the corrosion test environment at ~725°C, at lower temperatures (i.e. below the eutectic temperature of the salt mixture) it outperforms the other coating types studied.

Introduction

In 2007, the UK government agreed to an EU-wide target of 20% renewable energy by 2020^[1]. The UK share of this target would be the generation of 15% of the UK's energy from renewable sources by 2020, equivalent to almost a five-fold increase from 2009 levels^[2]. While such a target is ambitious, similar targets are also set outside UK to reduce environmental impact of energy generation. One way of meeting such targets is the combustion of biomass and municipal wastes for power generation. Utilisation of municipal wastes and biomass as fuel has two main advantages i.e. it reduces the dependence on landfill for waste disposal and produces carbon-neutral power in the case of biomass.

Several technological problems hinder the growth of power plants using biomass and municipal waste as fuel. These include high temperature corrosion via permeation of corrosive gases through the scale and/or corrosion, erosion and fouling of the underlying substrate via corrosive salts.

The use of biomass and waste-to-energy technology is dependent, to a great extent, on the development of high temperature corrosion resistant materials and processing technologies. The power plant component size and complexity as well as the rather limited ratio of improved corrosion resistance versus increased costs in most cases exclude solutions based on bulk corrosion resistant materials. Under such circumstances, solutions based on low-cost substrate materials (e.g. steels with sufficient strength at the temperatures envisaged); tailored coating and weld overlay compositions become technically and economically attractive. In addition to the development of new materials and coating systems, existing materials need testing by the industry to demonstrate that they are suitably corrosion resistant. Carrying out such testing in simulated waste or biomass incineration conditions will eliminate materials unsuited to such environments at low cost compared to full-scale industrial trials.

To address the above issues a high temperature (HT) corrosion test facility was designed and built. This was then used to assess the corrosion behaviour of coatings deposited by high velocity oxyfuel (HVOF) spraying and representative boiler substrate alloys in simulated high temperature biomass combustion conditions.

Figure 1: Schematic of the HT corrosion test set up

Experimental

Preparation of samples

Material selection: Two commercially available alloys (P91 and alloy 625) and four thermally sprayed coating compositions (NiCrBSiFe, and Ni alloys 718, 625 and C-276) supplied by Praxair (USA) were selected for the corrosion tests.

Coating production: HVOF coatings were deposited onto 20mm long, 20mm diameter P91 test coupons. In addition, uncoated P91 and alloy 625 test pieces were used to benchmark coating performance. Coatings were prepared using a standard, kerosene-fuelled, TAFA-JP5000 HVOF system (now Praxair). Prior to spraying, the substrates were blasted with alumina grit and cleaned with petroleum ether.

Table 1: Conditions used for the corrosion tests

Figure 2: Physical appearance of materials before (a) and after corrosion testing at 725°C (b), 625°C (c) and 525°C (d)

Corrosion testing

The test specimens were exposed to the corrosive environment given in Table 1. The specimens were embedded in a salt deposit comprising a KCl-45%K₂SO₄ mixture and then placed into the furnace in alumina crucibles (Fig.1). After placing the specimens in the furnace, it was heated to test temperature under N₂. Once the samples reached the test temperature, the test gas mixture (O₂, N₂, H₂O, HCl) was introduced at a flow rate of ~100ml/min. At the end of the test (after 168 hours), the test specimens were left to cool in the furnace under nitrogen.

Characterisation

Coated and uncoated test specimens, representative of before and after exposure, were photographed (Fig.2), and then cold mounted, sectioned and polished for metallographic examination. After the corrosion experiments, the mass change of the samples were recorded prior to sectioning (Fig.3).

Figure 3: Mass change of the samples after corrosion tests. Negative sign implies mass loss.

Corrosion of P91 and Alloy 625 Material

After exposure at ~725°C, P91 sample shows severe corrosion damage characterised by a very thick (>12mm), porous and dark grey coloured scale and a significant mass increase of ~0.81g/cm² (Fig.3). Cross-section indicates that the scale is characterised by poor adhesion, spalling and extensive cracking (Fig.4). The depletion of Cr in the outer layers, the presence of Cl and K mostly at the metal oxide interface, and S in the oxide scale are also evident (Fig.4a). Similar trends were also observed at lower temperatures (i.e. 525°C and 625°C) with milder corrosion damage (Fig.4b).

Figure 4: SEM micrographs with EDX maps of the scale formed on uncoated P91 after corrosion testing at
(a) 725°C and (b) 625°C

In the case of alloy 625, corrosion is less severe than P91. Corrosion and the formation of a porous layer is however evident at ~725°C (Fig.5a). The fragile nature of the thin corrosion scale formed on alloy 625 resulted in significant spalling and associated mass loss after corrosion testing. Element maps show depletion of Cr in the top surface (Fig.6). Very little oxide is seen in the mounted cross-section of the sample as a result of the oxide spalling during corrosion, leaving a porous layer underneath. Some Cl and Mo are seen on the surface. At lower temperatures (i.e. 525°C and 625°C) oxide spallation is not seen (Fig.2c and 2d).

Figure 5: SEM micrographs with EDX maps of the scale formed on wrought alloy 625 after corrosion testing at
(a) 725°C and (b) 625°C

Corrosion of Ni alloy coatings on P91 substrate

Alloy 625 coating

After testing at 725°C, the outer layer of the coating shows signs of corrosion similar to that seen in the wrought alloy 625 sample. The specimen shows very little mass gain after corrosion testing at all temperatures (Fig.3). Element mapping showed depletion of Cr in the coating, similar to the wrought alloy 625. Very little evidence of corrosion damage was observed at 625°C. At this temperature no evidence of penetration of the corrosive species was observed in the EDX maps (Fig.6b).

Figure 6: SEM micrographs with EDX maps of alloy 625 coated samples after corrosion at
(a) 725°C and (b) 625°C

Alloy 718 coating

Alloy 718 coating cross-section, after testing at 725°C, shows layers of corrosion product (Fig. 7a). EDX maps show a layer of oxide formed on top of the coating, with the top part of this oxide layer depleted of Cr, but Fe rich (Fig.7a). The presence of Cl can also be seen at the oxide-coating interface. The substrate adjacent to the interface with the coating has also been significantly depleted of Cr to the extent that voids are apparent in this region. At lower temperatures (i.e. 525°C and 625°C) the effect is milder, however, in some regions of the coating de-bonding and cracking was observed at 625°C (Fig.7b). At 525°C very little evidence of coating damage or de-bonding at the interface was observed (Fig.7c).

Figure 7: SEM micrographs with EDX maps of alloy 718 coated samples after corrosion testing at
(a) 725°C, (b) 625°C, (c) 525°C

Alloy C-276 coating

Extensive corrosion is evident in the samples coated with alloy C-276 as reflected by the mass gain of ~0.97g cm^-2 after testing at ~725°C (Fig.3). The specimen is characterised by the complete conversion of the coating to a very thick (>10mm) scale with considerable cracks and voids (Fig.8a). Element mapping shows extensive oxide formation with traces of chlorine present at the oxide-coating interface. The substrate adjacent to the interface with the coating has also been significantly depleted of Cr to the extent that extensive voids are visible in this region. This was not the case for samples tested at lower temperatures (i.e. 525°C and 625°C). The coatings were intact and the interface did not show any signs of de-bonding (Fig.8b and 8c).

Figure 8: SEM micrographs with EDX maps of alloy C-276 coated samples after corrosion testing at
(a) 725°C, (b) 625°C. (c) 525°C

NiCrBSiFe coating

The NiCrBSiFe coating is characterised by a small mass gain and the formation of very little corrosion product after testing at all temperatures (Fig.3 and 9). A porous layer has formed on the surface of the coating, but the test environment at 725°C has had no significant effect on the substrate. A corrosion front appears to have formed in the coating with an upper porous layer separated from the substrate by a relatively unaffected dense coating layer. Element mapping, after testing at 725°C, shows that a K, Cl, S, and Cr-rich corrosion product is formed on the surface of the coating and the presence of Si at the coating-corrosion product interface (Fig.9a). The substrate adjacent to the interface with the coating has also been depleted of Cr but with little evidence of void formation (in contrast to the behaviour of the alloy 718 and alloy C-276 coatings). At test temperatures below the eutectic (i.e. 525°C and 625°C) the effect is milder and there is no evidence of any damage (Fig.9b). The interface looks intact with very little evidence of corrosion. At all temperatures evidence of Si and O at the surface of the coatings can also be seen in the EDX maps hinting the formation of SiO₂ at the coating surface (Fig.9).

Figure 9: SEM micrographs with EDX maps of NiCrBSiFe coated samples after corrosion testing at
(a) 725°C, (b) 525°C

Corrosion mechanisms

Monolithic material

Alloy 625 coupons show void formation underneath the spalled corrosion product. In general, the formation and growth of certain metal oxides at the metal surface causes outward migration of the required metal atoms and inward migration of vacancies. The coalescence of vacancies causes voids to form (often called Kirkendall porosity)^[3].

For Cr₂O₃ forming steels, where during oxidation a Cr₂O₃ layer is expected to form, peeling of the top layer may be mitigated if the oxide is protective. However, in the presence of molten salts (K₂SO₄-KCl mixture), as used in the test at 725°C, dissolution (fluxing) and possible destruction of the oxide is expected. A molten deposit, rich in K and Fe (possibly mixed chloride or sulphate) is formed, which fluxes the oxides. This molten corrosion product converts to a mixed oxide on reaching regions of higher oxygen partial pressures (pO₂). The situation is exacerbated in an HCl-containing environment, where Cl₂ is generated by Weldon oxidation process.^[4] If the Cl₂ penetrates the oxide scale, it forms chlorides of iron and the alloying elements at the scale/metal interface where the oxygen partial pressure is low enough for the preferential formation of chloride over oxide.^[4-6] A metal chloride formed at the interface in this manner has a considerable vapour pressure (~10^-2bar for FeCl₂ at ~700°C), and it is expected that some evaporation will take place with volatile chloride diffusing outward through the porous oxide scale. On its way outwards, regions of higher oxygen partial pressure are reached where the chloride is oxidised, e.g. oxidation of FeCl₂(g) to form porous Fe₂O₃ in steels.^[4] Support for these mechanisms is seen in the formation of Cl-containing layers beneath the oxide seen in the element mapping (Fig.4).

Although P91 performs poorly when tested in HCl-containing environments, it was observed that at ~525°C the corrosion products were still adhered to the substrate similar to observations reported in ^[7]). The porous layer formed beneath the corrosion product, which has been separated from the rest of the corrosion product at temperatures >625°C, was possibly due to stresses generated. An adhesion loss, possibly due to cracking, can be clearly observed in an intermediate zone of the inner layer (Fig.4). Since the temperature used for one of the test (725°C) was above the eutectic point of the salt mixture used, it is likely that molten salt corrosion has occurred. The possible course of events can be summarised as melting of the salt mixture at ~684°C, reaction with the alloy components (primarily Fe), formation of chlorides and their evaporation leaving an oxide scale formed by the mechanism described previously. At temperatures lower than the eutectic of the salt used, the corrosion mechanism primarily involves gaseous corrosion assisted by the possible solid state diffusion of ions into the coating. This process is thermally activated and thus expected to be more profound at 625°C than at 525°C.

Coatings

The corrosion mechanisms involving alloy coatings are fairly complex, but some of the possible mechanisms are shown in Fig.10. The presence of inter-splat boundaries in HVOF coatings forms the weakest part of the coating and is where corrosion usually starts. With the onset of corrosion at the splat boundaries and formation of corrosion product, stresses are generated at the interface causing blistering and peeling of the corrosion layer. In cases where a protective oxide is expected, spalling of the top layer may be mitigated, but in the presence of molten salt and HCl containing environments the penetration of corrosive gases through the inter-connected, inter-splat porosity would increase defect concentration in the coating and is likely to accelerate the corrosion rate. Once the corrosive gases reach the substrate, the corrosion mechanism described previously will ensure further loss of material.

Figure 10: High temperature corrosion mechanisms in cases when an oxide layer is first formed

The depletion of Cr in the coating seen in the elemental maps is caused by the diffusion of Cr to form an oxide at the coating surface. As Cr₂O₃ is formed at a lower oxygen partial pressure than that required to form NiO, the NiO layer is mostly limited to the surface layers of the corrosion scale. At temperatures below the eutectic (i.e. <~684°C) the mechanism is mostly gaseous and vapour-induced corrosion which is much milder than molten salt corrosion, which involves fluxing of oxides, seen at temperatures exceeding the eutectic temperature of the salt used.

Alloy 718 coatings show an oxide layer, after testing at 725°C, which is depleted of Cr at the surface (Fig.7a). The Cr gradient seen in the oxide layer is possibly due to the evaporation of Cr-rich oxide in the presence of water vapour, which is known to cause evaporation of Cr₂O₃(s) in the form of CrO₂(OH)₂(g) from Cr-rich steels.^[8] It is possible that H₂O(g) may also act as an initiator for chlorine attack on Fe and Cr oxides above 500°C. The presence of Cl and K at the oxide-coating interface also indicates possible molten salt corrosion at ~725°C. The EDX maps show negligible Ni in the oxide scale suggesting the formation of oxides of Cr and Fe over that of Ni as expected from thermodynamics of oxide formation of these metals.

In the case of the alloy C-276 the reason for such a high rate of corrosion at 725°C is not entirely clear. As the samples were tested above the salt mixture's eutectic temperature (~684°C), it is possible that the Cl- ions penetrated the coating forming one or more particularly detrimental chloride of the alloy's constituent elements. For example, the disproportionation into gaseous chlorides of solid MoCl₃ which may have formed as a result of the reaction of the alloy with the molten salt mixture may have created significant voids in the coating.^[9] These voids provide an easy path for the molten salt and for gases to penetrate further into the coating. Once the Cl- ions reach the P91 substrate, extensive corrosion ensues. The element mapping also indicates the presence of oxides of Fe and Cr with traces of Cl, S and K possibly incorporated due to molten salt penetration (Fig.8). When a coating cathodic to the substrate is used molten salt penetration into the substrate can lead to further loss of anodic material (i.e. substrate) by galvanic interactions. At lower temperatures (i.e. <~684°C) when the corrosion mechanism changes from molten salt to gaseous corrosion the corrosion rate is much lower as the fluxing of the protective oxide does not take place.

The composition of the NiCrBSiFe coating suggests the possible formation of SiO₂ in-situ. The element maps confirms the presence of Si and O (and implies the formation of SiO₂) mostly at the coating-corrosion product interface (Fig.9). Depletion of Cr from the top layer of the coating is possible due to the diffusion of Cr to form its oxide. SiO₂ or a Si-containing compound is mostly seen below the chromium oxide layer. The coating seems to be fairly protective in the conditions used. It is known that the stability of oxides against chlorination is SiO₂> Al₂O₃> Cr₂O₃> Fe₃O₄> Fe₂O₃.^[10] The possible formation of SiO₂ from the Si-containing alloy might reduce the penetration of molten salt and act as a protective oxide film. The formation of a thin silica layer under a chromia scale, acting as a diffusion barrier, has also been postulated. ^[11]

Summary and conclusions

In common with other studies of substrate alloys and coatings, this work indicates that the corrosion behaviour of alloys differs widely with composition, but the effect of individual alloying elements is not entirely clear. However, it is widely accepted that the amount of chromium in a material will give some indication of the likely mechanisms responsible for their corrosion performance. Systematic studies of oxide formation as a function of Cr concentration in steels suggest that alloys containing <10wt% Cr mainly form multi-layered non-protective scales consisting of iron oxides. Above this concentration more protective Cr-rich oxides dominate the oxide scale. The minimum oxide growth rate is reported at 20wt% Cr. Thus, based on Cr concentration alone it might be expected that the coatings with high Cr content (21wt% for alloy 625 and ~19wt% for alloy 718) outperform coatings with lower Cr content (<16% for alloy C-276) at ~725°C. Although in molten salt environment fluxing of Cr₂O₃ is likely to reduce its otherwise protective function, nonetheless, some correlation between the Cr content and performance is evident. In case of high Mo content alloy coating i.e. C-276 (>16wt% Mo) chloride formation and its disproportionation could be contributing to its poor performance in HCl containing molten salt environment. This, however, is not the case at temperatures lower than the eutectic of the salt used and possibly gaseous and vapour-induced corrosion mechanisms are only operative at those temperatures. Thus, at lower temperatures alloy C-276 coatings perform well and show very little signs of degradation. In the case of NiCrBSiFe coating, the presence of SiO₂ forming Si and reasonably high Cr content (>17 wt%) could be the reasons for its lower corrosion rate.

It is possible to conclude a tentative ranking based on a combination of the amount of coating remaining and the impact on the substrate following chlorine-assisted corrosion at 725°C. Alloy 625 performs best of these coating materials at high temperatures, with an overall ranking as follows: alloy 625 > NiCrBSiFe > alloy 718 >> alloy C-276. At lower temperatures (i.e. <the eutectic point) the order changes and alloy C-276 coatings outperform the other coatings tested.

Acknowledgements

This work was funded by the Industrial Members of TWI as a part of the Core Research Programme. The authors would like to thank M.J. Bennett and A.K. Tabecki for technical support.

References

1. DTI, UK Biomass Strategy- Economic Analysis of Biomass Energy, DTI Report, 2007, URN 07/950, p 1-70.
2. DECC, The Renewable Energy Strategy (RES). Available from www.decc.gov.uk/en/content/cms/what_we_do/ uk_supply/energy_mix/renewable/res/res.aspx - [website no longer exists last checked on 27/02/2013].
3. E.O. Kirkendall, Diffusion of Zinc in Alpha Brass, Trans. AIME, 147, 1942, p 104-110.
4. H.J. Grabke, E. Reese, and M. Spiegel, The Effect of Chlorides, Hydrogen Chloride and Sulphur Dioxide in the Oxidation of Steels Below Deposits, Corros. Sci., 1995, 37(7), p 1023-1043.
5. S. Sroda and S. Tuurna, Laboratory Scale Tests on Corrosion Behaviour of Boiler Materials in Simulated Combustion Atmospheres (EU Project- OPTICORR), Mater. Corros., 2006, 57(3), p 244-251.
6. A. Zahs, M. Spiegel, and H.J. Grabke, Chloridation and Oxidation of Iron, Chromium, Nickel and their Alloys in Chloridizing Atmospheres at 400-700°C, Corros. Sci., 2000, 42(6), p 1093-1122.
7. D.A. Skobir and M. Spiegel, The Stability of Cast Alloys and CVD Coatings in a Simulated Biomass-Combustion Atmospheres, Mater. Tehnol., 2006, 40(6), p 247-251.
8. H. Asteman and M. Spiegel, Investigation of HCL(g) Attack on Pre-Oxidized Pure Fe, Cr, Ni and Commercial 304 Steel at 400°C, Corros. Sci., 2007, 49(9), p 3626-3637.
9. A. Zahs, M. Spiegel, and H.J. Grabke, The Influence of Alloying Elements on the Chlorine-Induced High Temperature Corrosion of Fe-Cr Alloys in Oxidizing Atmospheres, Mater. Corros., 1999, 50(10), p 561-578.
10. T. Ishitsuka and K. Nose, Stability of Protective Oxide Films in Waste Incineration Environment- Solubility Measurements of Oxides in Molten Chlorides, Corros. Sci., 2002, 44(2), p 247-263.
11. H.W. Grunling and R. Bauer, The Role of Silicon in Corrosion-Resistant High Temperature Coatings, Thin Solid Films, 1982, 95(1), p 3-20.