High Temperature Material Selection for Powder Processing

TWI Ltd, Granta Park Great Abington, Cambridge, CB21 6AL, UK

Copyright for this published paper belongs to TWI. It was also published in the March 2017 edition of NACE Materials Performance, and is based on material previously issued by TWI as part of its Core Research Programme.

Abstract

A high temperature powder processing cell experienced significant corrosion at 1100°C within 30 days. To select appropriate materials for service, trials were carried out on engineering alloys at 1100°C for 336h. Two simulated environments were used; 10%CO-1%CO₂-Ar without oxide powder and 90%CO-2.5%H₂-Ar while covered in TiO2 powder. Surface finish did not affect resistance to either environment, but more oxide powder adhered to ground surfaces. UNS N06625 and UNS N06230 showed the greatest resistance to attack, UNS N06600 showed moderate performance. Aluminide diffusion coatings on UNS N06601, UNS N06025 and UNS N07214 were protective but any defects yielded heavy localised corrosion.

1. Introduction

A key application of high temperature corrosion testing is to aid intelligent materials selection for use in challenging environments, through acquisition and interpretation of data relating to corrosion modes and mechanism, and methods of mitigating said corrosion. A case study is presented, involving a proprietary industrial powder processing cell.

The cell contains a complex environment which reduces or partially reduces metal oxide powder. The materials of construction are in direct contact with the powder and include high temperature nickel alloys and moving components. Measurements indicated the gas composition changes along the length of the cell due to variations in temperature, reaction with the metal oxides and interconversions in the gaseous environment. The temperature varies within the cell but approaches 1100°C in areas where the greatest corrosion and degradation was observed within 1 month.

This case study considers material selection by investigation of candidate alloys and coatings that might resist the proprietary service environment and to obtain further data on the corrosion mechanisms, corrosion rate and any other related factors which might accelerate or retard damage in service.

2. Approach

A series of UNS N06601 coupons were cut from the same rolled sheet and prepared with four different surface finishes. One set was ground to a 600 grit finish using silicon carbide paper, another was polished to a ¼µm finish using diamond paste and the third set was heat treated for 12 hours at 1050°C in air to form a protective oxide layer. The fourth set remained in the as-received condition. The as-polished coupon was stored under anhydrous toluene (at 20°C) to prevent oxidation. Alumina crucibles were ultrasonically degreased in acetone then heated in air to 1100°C for 24 hours. UNS N06601 coupons from each set were air dried, loaded into crucibles, and placed into the test cell.

The cell was purged with flowing nitrogen (2h, 1L.min^-1) and heated to 150°C for 2 hours to remove all toluene (nickel alloys are corroded at less than 0.05mm/year by boiling toluene1). The cell was then heated to 1100°C at 10°C/min and a test gas (10%CO-1%CO₂-Ar) flowed over the samples at 0.15L.min^-1 for 336 hours to simulate service conditions. Following a further nitrogen purge (2h, 1L.min-1) the samples were extracted, ultrasonically cleaned in acetone then weighed. The samples were cross-sectioned and examined by light microscopy (in the etched and unetched conditions), scanning electron microscopy (SEM) and energy-dispersive x-ray analysis (EDXA).

Initial inspection indicated that the behaviour of the UNS N06601 did not match that experienced in the proprietary industrial cell. Combined with updated data from this cell, conditions were modified to better simulate service. Coupons of UNS N06600, UNS N06601, UNS N06025, UNS N06625, UNS N06693, UNS N06002, UNS N07214 and UNS N06230 (some with a 50µm thick beta-aluminide diffusion coating) were prepared and their compositions determined by ICP-OES, Table 1. UNS N06601 was tested in the four surface conditions described above, all other alloys in the as-received state.

Table 1 Chemical compositions of parent materials as determined by ICP-OES. All materials were within standard manufacturer specifications. Some elements were not analysed, these are noted.

The coupons were ultrasonically degreased in acetone and loaded into alumina crucibles prepared as described above, then covered with fine rutile (TiO2) powder to further simulate the microenvironment experienced in service. Toluene was not used. The test cell was purged with nitrogen (2h, 1L.min-1) and then heated to 600°C. A test gas with higher carbon activity (90%CO-2.5%H2-Ar) was introduced and the test cell then heated to 1100°C. The test gas was flowed over the samples at 0.15L.min-1 for 336 hours at 1100°C. Four coupons were tested at a time. The test cell was then purged with nitrogen and the samples extracted and cross-sectioned for microstructural analysis by light microscopy and SEM/EDX. This included measurement of penetration depth, i.e. the deepest point at which the microstructure had been affected by the environment. Mass change results are not presented for these trials due to varying amounts of partially reduced/reacted Ti oxides tenaciously adhering to the surface, as shown in Figure 1. The powder was analysed by x-ray diffraction.

Figure 1: UNS N06601 (tested in the as-received surface condition) after 336h at 1100°C in flowing 90%CO-2.5%H2-Ar while under TiO2 powder. A millimetre scale is included.

3. Results and Discussion

3.1 UNS N06601 in 10%CO-1%CO₂-Ar at 1100°C

When tested under 10%CO-1%CO₂-Ar, UNS N06601showed a mixture of carburisation and oxidation, with black particles visible along grain boundaries (example in Figure 2) at depths of up to 150µm. Mass gains varied from 3.6-4.9g.cm^-2 depending on surface finish. EDX confirmed the presence of complex corrosion products on the UNS N06601 surface rich in Cr, O, presumably chromium oxide. Chromium and carbon-rich regions were nearer the surface, alongside detached pieces of substrate. Intergranular attack was present below, with dark particles at the grain boundaries primarily consisting of Al, O presumably Al2O3. This indicates minimal oxygen diffusion along the substrate into the grain boundaries; under a more oxidising environment the formation of chromium oxides would be expected to dominate² (p[O₂] for formation of Cr₂O₃ is ~10^-19atm at 1100°C vs ~10^-31atm for Al₂O₃). No microstructural difference was observed between the four different surface finishes post‑testing, but in all cases the surfaces and cross‑sections appeared less corroded and had a different appearance than the proprietary industrial cell. This gas mixture was not judged to be a suitable simulant for this service.

Figure 2: Light micrograph of UNS N06601 cross-section (tested in the polished surface condition) after 336h at 1100°C in flowing 10%CO-1%CO₂-Ar.

3.2 Uncoated alloys in 90%CO-2.5%H₂-Ar at 1100°C

Testing of multiple engineering alloys under TiO₂ powder in 90%CO-2.5%H₂-Ar yielded reduction of TiO₂ to Ti₃O₅, likely releasing oxidising species into the local environment. Different amounts of oxide powder adhered to alloy surfaces even after multiple ultrasonic cleaning cycles and mechanical removal. Metallic flecks were observed in the powder indicating reasonably strong adherence to the surface, with pieces of the substrate removed as the powder was detached. For UNS N06601, more powder adhered to surfaces ground to 600 grit finishes than heat treated, polished or as-received surfaces, likely due to mechanical keying and greater surface area. No correlation between alloy composition and oxide adherence was observed.

All alloys were significantly corroded by this modified environment, but mechanism and corrosion product microstructure varied substantially. For most alloys, corrosion was general and did not vary across different sides of the same coupon, exceptions are noted below. Overall corrosion behaviour was similar enough to that experienced in the proprietary industrial cell (including a partially ‘melted’ appearance on the surface of UNS N06601 and several other alloys) to consider the combination of 90%CO-2.5%H₂-Ar and TiO₂ powder a good simulant of service and thus suitable for material selection and further testing.

The surface finish of UNS N06601 did not significantly affect the extent or mechanism of corrosion, with penetration depths ranging from 91µm (heat treated) to 105µm (as-received), likely the same within the range of experimental error. Figure 3 shows an example. Below the affected region the composition was unchanged but Al and O (indicating Al₂O₃) were present along the grain boundaries. The metallic region was depleted in chromium and the surface oxide layer correspondingly enriched with both chromium and titanium. Local aluminium enrichment was observed in some regions of the scale.

Figure 3: Scanning electron micrograph of UNS N06601 cross-section (tested in the polished surface condition) after 336h at 1100°C in flowing 90%CO-2.5%H₂-Ar while under TiO₂ powder.

UNS N06025 exhibited similar intergranular corrosion to UNS N06601 with Al, O (again presumably Al₂O₃) at the grain boundaries, though with a thicker attacked layer at the surface. UNS N06600, N06625 and N06002) corroded mainly by void formation beneath a coherent scale layer of mixed titanium-chromium oxides. Limited intergranular attack was observed in UNS N06600. UNS N06693 and UNS N07214 both formed multi-phase surface layers; for UNS N06693, the layer was primarily chromium-depleted metallic fragments surrounded by carbon-rich oxide corrosion product whereas for UNS N07214 the concentrations of titanium were much higher. UNS N06230 formed a coherent layer of chromium oxide, mixed titanium-chromium oxide and a carbon-rich chromium phase, likely chromium carbide. This indicated direct reaction between titanium oxide(s) and the substrate or corrosion scale on the surface. Intergranular precipitation with Al₂O₃ was also observed in the substrate. Below the scale layers, no voids or intergranular attack were visible in either case.

3.3 Coated alloys in 90%CO-2.5%H₂-Ar at 1100°C

Examination of aluminide diffusion coated samples pre-test verified that the coatings were coherent, non-porous and covered the entire outer surface with no visible cracking. The microstructure showed a rough outer layer, a middle layer with voids at its base and an inner layer with filamentary penetration into the substrate, including diffusion of aluminium and formation of a metallurgical bond between coating and substrate. The observed attack on coated substrates depended on local coating behaviour. For UNS N07214 and UNS N06025, there was little or no attack where the coating was present. Where significant regions of the coating had been damaged, the attack was similar to that on uncoated coupons of the same materials (Figure 4). The coating microstructure had also changed, with the formation of a mixed aluminide layer with voids remaining in the coating. For UNS N06601, only a small flaw was present in the coating, this lead to massive grain boundary precipitation radiating outwards from this region. For all three coated alloys, the attack was deeper than observed on uncoated substrates, but only in regions where the coating was damaged.

Figure 4: Light micrograph of beta-aluminide coated UNS N06601 cross-section after 336h at 1100°C in flowing 90%CO-2.5%H₂-Ar while under TiO₂ powder. Grain boundary precipitation is visible for several hundred µm below this region.

3.4 Overall alloy performance

Difficulties were encountered in accurately measuring penetration depth perpendicular to the surface, as the original substrate surface could not be clearly discerned. Where this was the case, measurements were taken from the visible remaining surface. All alloys and penetration depths are summarised in Figure 5.

Figure 5 Maximum penetration depths for all alloys tested for 336h at 1100°C in flowing 90%CO-2.5%H₂-Ar while under TiO₂ powder. Some depths are above the scale of this graph, these are noted.

Based on the above, UNS N07214 was deemed unsuitable for this service. UNS N06230 experienced low penetration with little intergranular attack, though a Cr-depleted layer formed below the scale surface. UNS N06625 had moderately promising performance; low penetration combined with void formation and no intergranular attack. The surface did not display a significant melted/corroded appearance. This may be due to the presence of niobium which acts as a carbide stabiliser and may lead to preferential formation of niobium carbides over chromium carbides. These two materials appeared to be the most promising. UNS N06600 did have the lowest penetration into the substrate; voids and carbides had formed below the substrate surface but the original surface could not be definitely discerned and so was regarded as promising but with caution. Other alloys displayed intermediate performance, with higher levels of molybdenum and/or chromium generally leading to increased resistance, but this was qualitative only.

4. Conclusions

In 90%CO-2.5%H₂-Ar at 1100°C with TiO2 powder present (a suitable simulant for the proprietary industrial cell), attack was attributed to a complex mixture of oxidation and carburisation resulting in formation of chromium oxides, aluminium oxides and multiple carbides. Attack was either general/uniform (with void formation) corrosion or intergranular corrosion with associated precipitation along grain boundaries, depending on alloy composition.

No quantitative correlation between alloy composition and resistance to attack was observed, but qualitative trends were present. UNS N06230 and UNS N06625 showed reasonable resistance to this environment with UNS N06600 showing good, but slightly lesser performance. UNS N07214 showed the poorest performance. Beta-aluminide coatings were partially protective, but any local failure led to extremely heavy localised attack. Material lifetime could potentially be improved by preventing the metal oxide powder from directly contacting the substrate by means of a coating or tiling.

5. Acknowledgements

The author gratefully acknowledges the Core Research Programme of TWI for supporting this and subsequent research.

6. References

1. P. A Schweitzer Corrosion Resistance Tables: Metals, Plastics, Nonmetallics and Rubbers, 2^nd ed. (Marcel Dekker Inc, 1986), p1166
2. L Coudurier, D. W. Hopkins, I. Wilkomirsky Fundamentals of Metallurgical Processes, 2^nd ed, International Series on Materials Science and Technology Volume 27, (Pergamon Press 1985), p74