Comparative Thermal Aging Effects on PM-HIP and Forged Inconel 690

Abstract

This study compares thermal aging effects in Inconel 690 (IN690) produced by forging and powder metallurgy with hot isostatic pressing (PM-HIP). Isothermal aging is carried out over 400–800°C for up to 1000 h and then metallography and nanoindentation are utilized to relate grain microstructure with hardness and yield strength. The PM-HIP IN690 maintains a constant grain size through all aging conditions, while the forged IN690 exhibits limited grain growth at the highest aging temperature and longest aging time. The PM-HIP IN690 exhibits comparable mechanical integrity as the forged material throughout aging: hardness and yield strength are unchanged with 100 h aging, but increase after 1000 h aging at all temperatures. In both the PM-HIP and forged IN690, the Hall–Petch relationship for Ni-based super-alloys predicts yield strength for 0–100 h aged specimens, but underestimates yield strength in the 1000 h aged specimens because of thermally induced precipitation.

INTRODUCTION AND BACKGROUND

Alloy fabrication by powder metallurgy with hot isostatic pressing (PM-HIP) provides many notable advantages over conventional alloy forging, including structural uniformity, chemical homogeneity, superior mechanical properties, and enhanced weldability.¹ Additionally, PM-HIP components are produced near-net shape, requiring little machining and simplifying inspectability. Because of these attributes, PM-HIP is an attractive processing pathway for heavy components used in pressure-retaining applications throughout the electric power industry.^2–4 High-temperature Ni-base alloys, including Inconel 690 (IN690), are leading candidates for PM-HIP processing, especially for applications such as pressure vessels, heat exchanger tubing, steam generators, and nuclear core components.³ However, PM-HIP materials must perform comparable to or better than forgings throughout the service lifetime. Fossil and nuclear power generation service environments include 40+ year service at ~ 290–315°C.^5,6

Thermal aging and its resultant microstructural and mechanical changes must be understood to qualify PM-HIP components for high-temperature service. Further, the PM-HIP components must perform comparably, if not better than their forged counterparts under thermal aging conditions. The microstructural effects of thermal aging have been studied extensively on forged and cast IN690,^7–14 leading to a comprehensive understanding of thermally induced chromium (Cr) carbide precipitation along grain boundaries. However, most of these studies have focused on the relationship between these Cr carbides and the susceptibility to intergranular attack and intergranular stress corrosion cracking (IGSCC).^11–14 Fewer studies focus on the mechanical implications (without concurrent corrosion) of thermal aging. Even fewer studies have considered thermal aging effects in PM-HIP produced IN690. Differences, if any, in the thermal aging behavior between PM-HIP and forged IN690 are specifically not well known and are the focus of this study.

The objective of this study is to elucidate differences in the microstructural and mechanical effects of thermal aging of PM-HIP compared with forged IN690. A series of anneals is carried out from 400°C to 800°C for up to 1000 h. Then, metallography with optical microscopy is utilized to quantify grain microstructures, and nanoindentation is used to quantify hardness and estimate yield strength. Finally, differences between the PM-HIP and forged specimens are discussed by relating grain size and yield strength through the Hall–Petch formulation.

METHODS

Materials and Aging

Ingots of PM-HIP heat Y2532B and forged IN690 (nominally Ni-30Cr-9Fe-0.04C-0.3Si-0.7 Mn-0.5Cu, in wt.%) were obtained from the Electric Power Research Institute. HIPing was carried out at 1121.1°C and 15 ksi. The forged IN690 plate was hot rolled and subsequently annealed at 1037.8°C for 2 h 40 min, followed by air cooling. For both the PM-HIP and forged materials, the final two-step heat treatment involved solution annealing at 1075°C for 30 min with water quenching and then thermal treatment at 700°C for 15 h with air cooling.

To emulate thermal aging over the 300°C, the 80-year service lifetime of an IN690 steam generator tube, this study employed an accelerated aging approach. There is also an interest in studying the effects of aging in advanced reactor designs, which will operate at temperatures as high as 600–650°C for the same lifetime. Accelerated aging times were determined by setting the Larson–Miller parameters (LMP) equal to one another for the actual (T₁, t₁) and simulated (T₂, t₂) aging conditions. Use of the LMP is a widely accepted method for establishing accelerated aging conditions:¹⁵

$$LMP=T\left(\log t+C\right)$$ (1)

$$T_1\left(\log t_1+C\right)=T_2\left(\log t_2+C\right)$$ (2)

where T is the temperature (Kelvin), t is aging time (h), and C is a material constant set to 20 for Ni-base alloys.¹⁶ Using this approach, aging times were established at 100 h and 1000 h. At 400°C, 100 h aging simulated the 300°C, 80-year steam generator tubing service life. At 800°C, the range of 100–1000 h aging simulated the 600–650°C, 80-year advanced reactor component service life. An intermediate aging temperature of 600°C was also selected. Hence, a thermal aging experiment matrix was carried out at 400°C, 600°C, and 800°C, to 100 h and 1000 h at each temperature. This expersiment matrix enabled the study of thermal aging effects as a function of both time and temperature. Aging was conducted in Carbolite CWF 12/5 ovens in air, and temperature was continuously monitored using thermocouples. Following aging, specimens were prepared to metallographic standards using mechanical polishing up to 1200-grit SiC paper and then 1-μm diamond suspension. Finally, specimens were etched using a solution of 6.3 mL H₂O + 2.6 mL HNO₃ + 5.6 mL HCl + 2.6 mL H₂O₂ to reveal grain boundaries; this etchant was recommended through collaborative trial and error with Pace Technologies.

Microscopy

Light optical microscopy was conducted using an Olympus BX series upright metallurgical microscope to observe the grain microstructure. Grain sizes were measured using the intercept method on light optical micrographs, in accordance with ASTM Standard E112.

Nanoindentation

Nanoindentation was conducted using a Nanomechanics iMicro equipped with a diamond Berkovich tip. The iMicro was operating in dynamic nanoindentation mode using the Oliver–Pharr method¹⁷ to calculate nanohardness from the indent depth and contact area. The target indentation load was 500 mN; indent depths varied based on the specimen hardness, but ranged ~ 3–4 μm for the IN690 specimens tested herein. A minimum of 13 indents were repeated on each alloy/condition to provide > 90% confidence in measurements. Indents were 50 μm apart to prevent plastic zone interactions and overlap. To reduce system creep, a three-segment loading curve was defined with a 35-s load period, 1-s hold period, and 1-s unload period.

Nanoindentation is sensitive to surface effects,¹⁸ wherein residual roughness and plasticity on the surface cause exaggerated nanohardness measurements. The nanohardness will decrease with increasing indent depth, plateauing to bulk hardness for indents deeper than ≳ 200 nm for materials with comparable strength as IN690.¹⁹ Dynamic nanoindentation provides a high-fidelity depth profile of hardness, making it easy to observe the surface effects and plateau to bulk hardness. Throughout the remainder of this article, plateau bulk hardness values have been taken from each indent; the averages and standard deviations of these values were taken over all indents from a condition to determine average hardness and error bars.

RESULTS

Experimental results include average grain size and nanohardness. All results are summarized in Table I for all aging conditions and are discussed in greater detail in the following sub-sections.

Table I.

Experimental matrix with average grain size measurements, nanohardness, and yield stress and modulus extracted from nanoindentation

Method	Aging temperature (0C)	Aging time (h)	Grain size (μm)	Hardness (GPa)	Yield stress (MPa)
PM-HIP	–	–	37.5 ± 7.5	1.87 ± 0.17	541 ± 50
	400	100	39.9 ± 5.7	1.98 ± 0.05	572 ± 16
		1000	35.2 ± 6.4	3.14 ± 0.45	908 ± 129
	600	100	35.3 ± 3.1	1.93 ± 0.20	558 ± 57
		1000	33.4 ± 5.5	2.95 ± 0.21	854 ± 62
	800	100	32.6 ± 4.3	1.82 ± 0.09	526 ± 27
		1000	37.5 ± 4.7	2.40 ± 0.09	694 ± 27
Forged	–	–	70.6 ± 5.5	1.97 ± 0.10	569 ± 29
	400	100	67.0 ± 8.3	1.78 ± 0.07	514 ± 20
		1000	62.7 ± 5.6	3.55 ± 0.32	1026 ± 93
	600	100	74.0 ± 6.9	1.93 ± 0.06	558 ± 18
		1000	71.5 ± 5.1	2.68 ± 0.33	775 ± 95
	800	100	60.5 ± 8.1	1.82 ± 0.07	526 ± 21
		1000	82.3 ± 4.5	3.01 ± 0.36	871 ± 105

Grain Size

Average grain sizes in the PM-HIP and forged IN690 specimens were measured from optical micrographs (Fig. 1). Prior to aging, the average grain size of the PM-HIP IN690 is 37.5 ± 7.5 μm, as compared with 70.6 ± 5.5 μm for the forged IN690. Throughout aging, PM-HIP IN690 maintains a smaller grain size than forged IN690 (Fig. 2a). With aging, PM-HIP IN690 exhibits a statistically unchanged grain size, ranging from 32.6 μm to 39.9 μm for all aging conditions studied. Forged IN690 also exhibits a statistically unchanged grain size for most aging conditions, ranging from 60.5 μm to 74.0 μm. The only condition with statistically significant grain growth (compared to the as received) is the 800°C, 1000-h forged IN690 specimen, which has average grain size 82.3 ± 4.5 μm. This result is unsurprising given that grain growth is not often observed after only 100–1000 h aging at temperatures ⪅ 0.5*T_m.

Fig. 1.

Representative optical micrographs from PM-HIP IN690 (a) unaged, (b, c) aged at 400°C for 100 h and 1000 h, (d, e) aged at 600°C for 100 h and 1000 h, and (f, g) aged at 800°C for 100 h and 1000 h. Forged IN690 (h) unaged, (i, j) aged at 400°C for 100 h and 1000 h, (k, l) aged at 600°C for 100 h and 1000 h, and (m, n) aged at 800°C for 100 h and 1000 h. Scale bars are applicable to all PM-HIP or forged micrographs.

Fig. 2.

Effect of temperature on (a) grain size and (b) nanohardness for 100 h and 1000 h aging.

Nanohardness

Nanohardness of the as-received PM-HIP and forged specimens is self-consistent at 1.87 ± 0.17 GPa and 1.97 ± 0.10 GPa, respectively. Hardnesses of PM-HIP and forged specimens remain comparable with one another at all aging conditions; the aging process has a similar influence on the hardness of both the PM-HIP and forged materials (Fig. 2b). After 100 h aging, no statistically significant changes in hardness in either the PM-HIP or forged specimens are apparent. However, after 1000 h aging, hardening is observed in both the PM-HIP and forged specimens. For the PM-HIP 1000 h specimens, hardness is greatest following 400°C aging (3.14 GPa); the extent of hardening decreases with increasing aging temperature. For the forged 1000 h specimens, hardness is again greatest after 400°C aging (3.55 GPa), although there is no statistically significant difference in hardness between all 1000-h aging temperatures.

DISCUSSION

Yield strength, σ_y, is estimated from nanohardness measurements using the empirical relationship for austenitic alloys:²⁰

$${\sigma }_y\left[\mathrm{MPa}\right]=3.03H_{\text{V}}\left[\mathrm{kg}/{\mathrm{mm}}^2\right]$$ (3)

where the Vickers hardness, H_V (kg/mm²), is directly related to the measured Berkovich nanohardness, H_B (GPa), through the Fischer–Cripps relationship:²¹

$$H_{\text{V}}\left[\mathrm{kg}/{\mathrm{mm}}^2\right]=94.495H_{\text{B}}\left[\mathrm{GPa}\right]$$ (4)

Although Eq. 4 was initially developed for microhardness measurements, it has been shown to be applicable to nanohardness measurements.^19,22

Estimated yield strengths for the unaged PM-HIP and forged specimens are 541 and 569 MPa, respectively (Table I). Aging for 100 h does not induce statistically significant change in estimated yield strength in both the PM-HIP and forged materials, regardless of aging temperature. The 1000-h aged specimens exhibit estimated yield strength increases, consistent with the trends noted in “Nanohardness” section for hardness (since yield strength and hardness are linearly related).

The increased estimated yield strength after 1000 h aging is not accompanied by a reduction in grain size. Hence, this strengthening cannot likely be attributed to grain boundary strengthening. The grain boundary strengthening mechanism is commonly formulated through the Hall–Petch relationship:²³

$${\sigma }_y={\sigma }_0+k{d}^{-1/2}$$ (5)

where σ₀ is the friction stress (i.e., a materials constant representing the resistance of the lattice to dislocation motion), k is the Hall–Petch coefficient, and d is the average grain diameter. The Hall–Petch relationship is calculated by assuming the friction stress and Hall–Petch coefficient for Ni-base superalloys: k = 400 MPa√m²⁴ and σ₀ = 450 MPa.²⁵

Experimental data points obtained in this study are overlaid onto the Hall–Petch curve (Fig. 3) and then categorized as either aged for 0–100 h or aged for 1000 h. Regardless of the processing method, the Hall–Petch relationship underestimates the yield strength for 0–100 h aged specimens by only ~ 40–110 MPa, but severely underestimates the yield strength for 1000 h aged specimens by ~ 210–560 MPa. Differences between the Hall–Petch prediction and experimental measurements for the 0–100-h-aged specimens fall within 7–20% of the Hall–Petch prediction. This difference is likely attributed to the approach used to determine yield strength, which is based on estimation from highly localized nanohardness measurements that are likely contained within a single grain (whereas the Hall–Petch relationship is inherently polycrystalline). Other potential factors contributing to this difference include nanoindentation surface effects and chemistry variations between the test specimens and those used in the empirical determination of the k and σ₀ values.

Fig. 3.

Hall-Petch relationship for Ni superalloys reasonably predicts yield strength for specimens aged 0–100 h, but underpredicts yield strength for 1000 h aging

Differences between experimental measurements and Hall–Petch predictions for 1000-h-aging specimens may be attributed to precipitation strengthening and extensive precipitation during aging. Again, while specific investigation into the type and morphology of these precipitates is beyond the scope of this work, there are numerous reports of thermally induced precipitation in IN690, which can explain the results reported herein.^{7–9,11–13,26,27}

Thermal aging of IN690 at 600–800°C has been shown to induce grain boundary Cr depletion by way of Cr diffusion to Cr-rich M₂₃C₆ carbides, which extensively cover the grain boundaries.^7–9 These carbides form preferentially at random high-angle boundaries.^8,9 Reports show Cr depletion and carbide coarsening as rapidly as 0.1–72 h of aging at 700–800°C, although at these lower aging times, the carbides have little influence on grain size or hardness.^11,13,26,27 This observation is consistent with the 100-h-aging results, for which no significant change in grain size or hardness is measured.

At isothermal aging times > 100 h, the Cr-rich M₂₃C₆ carbides begin to precipitate on imperfections such as dislocations and stacking faults.¹³ Carbides on random high-angle boundaries continue to coarsen and become increasingly discrete with further aging.¹³ Additional TiN and CrS particles have also been found in the matrix and on grain boundaries after long aging times.¹² The continued nucleation and coarsening of carbides, together with additional long-aging-time phase nucleation and growth, can explain the hardening observed after 1000 h aging—extensive precipitation can sufficiently change the matrix composition such that the friction stress and Hall–Petch coefficient for Ni-base superalloys no longer hold. These phases can also explain the grain size stability throughout this study—carbides decorating grain boundaries obstruct thermally activated grain boundary migration.

It should be noted that the same reactions driving precipitation at temperatures > 600°C are also active at lower temperatures, but the reaction kinetics are slower. Thermal precipitation occurs at all temperatures investigated, given enough time. Notably, the 400°C, 1000-h specimens exhibit consistent hardening—presumably due to thermal precipitation—as the 600°C and 800°C 1000-h specimens. Thus, the accelerated aging is a meaningful performance predictor for lower-temperature applications. Confirmatory study of the precipitate chemistry and morphology is future work planned beyond the scope of this article.

CONCLUSION

PM-HIP and forged IN690 respond similarly to thermal aging at 400°C, 600°C, and 800°C for 100 h and 1000 h. Grain sizes remain smaller in the PM-HIP material than in the forged material, across all aging conditions. Besides slight grain growth in the forged IN690 after 800°C, 1000 h aging, both alloys exhibit statistically invariant grain sizes at all other aging conditions. Both PM-HIP and forged IN690 also exhibit similar hardness and yield strength measurements, and these properties remain statistically invariant through 100 h aging at all temperatures. Consistent increases in hardness and yield strength are observed in both the PM-HIP and forged IN690 after 1000 h aging. Because of the lack of accompanying grain growth, hardening is attributed to thermally induced precipitation.

In both the PM-HIP and forged IN690, the Hall–Petch relationship for Ni-base superalloys reasonably predicts yield strength for unaged and 100-h-aged specimens, but underestimates yield strength by hundreds of MPa in the 1000-h-aged specimens. Deviation from Hall–Petch behavior in the 1000-h specimens may be attributed to precipitation and matrix composition changes