Performance of FeCrAl for accident-tolerant fuel cladding in high-temperature steam

1 Introduction

After the station blackout event in March 2011 at the Fukushima Daiichi power plant (Gauntt et al., 2012; Robb et al., 2014), considerable research has focused on developing a more accident-tolerant light water reactor (LWR) fuel system than the current Zr alloy (e.g. Zircaloy)/UO₂ fuel rods. Under such a beyond design basis accident (BDBA) scenario, the fuel rods burst at 700–1000°C and the cladding rapidly oxidizes in the steam environment forming hydrogen. The self-catalytic enthalpy production due to the reaction significantly increases the heat production rate in the core. Thus, the metric for selecting alternative cladding materials has been a ≥100× reduction in the steam oxidation rate at 1200°C compared to Zr (Pint et al., 2013, 2015a; Ott et al., 2014; Zinkle et al., 2014).

A leading accident-tolerant fuel (ATF) cladding material is FeCrAl (Cheng et al., 2012; Pint et al., 2013, 2015a,b; Terrani et al., 2014a; Rebak, 2015; Robb, 2016). While it has some issues with neutronics and tritium permeation, it is far less expensive than Zr and more compatible than SiC with the LWR operating conditions (Terrani et al., 2015, 2016). Also, while cladding or coatings on the current Zr-based cladding is a potential near-term solution, the effectiveness is questionable if the cladding bursts or interacts with the Zr (Terrani et al., 2013). Obviously, SiC has the potential to be protective above 1500°C in steam (Terrani et al., 2014b), well above the liquidus temperature of ferrous alloys; however, it is also difficult to fabricate and has lower ductility than a metallic cladding (Katoh et al., 2012). The properties of FeCrAl alloys for this application have been evaluated since 2011 (Field et al., 2014, 2015; Yan et al., 2014; Hu et al., 2015; Yamamoto et al., 2015; Edmondson et al., 2016; Massey et al., 2016), including alloy development to examine 10–13% Cr wrought and oxide dispersion strengthened (ODS) (Pint et al., 2014; Unocic et al., 2015) versions with better radiation tolerance than conventional ~20% Cr FeCrAl compositions. Initial work to evaluate the steam oxidation resistance examined behavior at 1200°C with isothermal exposures (Cheng et al., 2012; Pint et al., 2013, 2015a,b; Unocic et al., 2017). However, modeling has shown that sustaining low reaction kinetics to 1500°C is critical to maximizing the benefit. Therefore, more recent work has begun to focus on higher temperatures (≥1400°C) and exploring non-isothermal conditions to better simulate the LWR accident scenario where the temperature is continuously increasing (Pint et al., 2014, 2015b; Unocic et al., 2015, 2017). While relevant to the BDBA scenario, these conditions have not been considered previously; thus, there are no prior data for comparison, including the effect of steam on the oxidation reaction in this regime. As the next phase in the evaluation, the steam oxidation behavior of commercially fabricated FeCrAl alloy tubing has been investigated from 1400°C to 1700°C in steam and air, and compared to Zircaloy-4 (Zr-4) tubing.

2 Materials and methods

Specimens in this study were typically ~12.5 mm lengths of ~9.5 mm diameter (Zr-4 was 11 mm diameter) tubing with 0.3–0.7 mm wall thickness and 6–9 cm² surface area. The alloy compositions are shown in Table 1. The new compositions were developed at Oak Ridge National Laboratory (Yamamoto et al., 2015) and commercially fabricated. The alumina specimens were typically 1.5×10×20 mm. Oxidation exposures were conducted in two systems: (i) a high-temperature (1700°C) test rig consisting of a vertical alumina tube with two resistively heated furnaces, and (ii) a magnetic suspension thermogravimetric analyzer (TGA) using a DynTHERM LP-HT-II instrument (Rubotherm GmbH, Bochum Germany) where the alumina test chamber was fully isolated with dry air or 100% steam at up to 1500°C (Terrani et al., 2014b). In the first system, where the majority of work was performed, steam or air entered the bottom of the reaction tube and was preheated to 1000–1300°C by the first furnace. The tube specimen was held in the second furnace in an alumina holder shown in Figure 1 designed to catch any oxidation debris. The specimen was heated to temperature in 1.5 h (15–19°C/min) and held for 1 min at temperature before cooling to room temperature in flowing argon. Initially, Ar (500 cm³/min) was used to purge the system while it was heated to 600°C, then either steam or dry air (500 cm³/min) was introduced. In the TGA experiments, the specimen was suspended with a Pt-Rh wire, which experienced little evaporation in steam. In this system, the specimens were ramped at 5°C/min and, similarly, the steam was introduced at 600°C after an Ar purge. With the TGA monitoring mass gain, the ramp was stopped if the total mass gain exceeded ~10 mg or the temperature reached 1500°C. The water used to generate steam in both rigs was deionized but not Ar-bubbled or filtered, as is typically done for ~600°C testing. The gas velocity in the TGA experiments was typically 1 cm/s, while the steam velocity was 50–60 cm/s (~200 ml/h) in the high-temperature furnace. The mass change of all specimens was measured using a Mettler Toledo (Columbus, OH, USA) model XP205 balance with ±0.04 mg or <0.01 mg/cm² accuracy. After exposure, specimens were metallographically sectioned and examined by light microscopy.

Figure 1:

Specimen holder and tube specimen used for ramp testing in air and steam in the high-temperature furnace.

3 Results

Figure 2 summarizes some of the mass change data from these ramp experiments in 1 bar steam and air. The mass gain is plotted versus the maximum ramp temperature. For the higher-temperature FeCrAl experiments, it was difficult to obtain mass change data as the specimen often disintegrated and/or reacted with the alumina holder. For the Zr-4 specimens, the mass gain increased from 1400°C to 1600°C, as expected. At 1700°C, no further increase was noted in air likely because at both 1600°C and 1700°C, the specimen was fully reacted. As ZrO₂ and Al₂O₃ form no compounds (Jerebtsov et al., 2000), there was no problem with interactions with the alumina holder. However, the lower mass gain after 1700°C in steam reflects that part of the specimen was lost during handling. Figure 3 shows images after each of the tests, indicating the damage after the exposure to 1700°C in air. Previously, it was reported that oxidation of Zr-based alloys in air was more severe than in steam (Steinbrück & Boettcher, 2011). These results confirm that observation both in mass change in Figure 2 and based on the cross-sections in Figure 4. For ramping to 1400°C, the oxide was much thicker in air than in steam (Figure 4A and B). For 1600°C, a similar difference was noted (Figure 4C and D), although the reaction front was almost at the center of the specimen after this exposure.

Figure 2:

Tube specimen mass change after ramp testing in steam and air in the high-temperature furnace.

Figure 3:

Images of Zr-4 tube specimens ramped to increasing temperatures in air and steam in the high-temperature furnace.

Figure 4:

Light microscopy of polished cross-sections of Zr-4 specimens ramped to 1400°C in the high-temperature furnace using (A) steam and (B) air and to 1600°C in (C) steam and (D) air.

The initial comparison with FeCrAl used existing Kanthal alloy APM tubing. This alloy is similar to the Kanthal APMT alloy (Jönsson & Svedberg, 1997; Jönsson et al., 2004, 2013) used in earlier studies (Pint et al., 2013, 2015a,b; Unocic et al., 2017) (Table 1). Figure 2 shows that the mass gains for APM tubes were very low after ramping to 1400° and 1500°C in steam and air, reflecting the formation of a protective α-Al₂O₃ scale and dramatically different than the behavior of Zr-4. Figure 5 shows that these specimens were relatively unaffected by these exposures below the solidus temperature of ~1520°C. Figure 6 shows that thin oxides formed on the outer diameter (OD) of these tubes after exposure. The alumina scale was only a few microns thick and mainly adherent. The mass gains were also modest after ramping to 1600°C in steam and air; however, significant attack occurred as shown in Figure 5. No mass change data could be obtained after the 1700°C ramps in steam or air as the specimen was heavily attacked (Figure 5), and the Al₂O₃ holder and FeO_x reaction product are prone to react to form a mixed oxide similar to alumina and NiO forming NiAl₂O₄ (Pettit et al., 1966).

Figure 5:

Images of commercial FeCrAl (APM) tube specimens ramped to increasing temperatures in air and steam in the high-temperature furnace.

Figure 6:

Light microscopy of polished cross-sections of the scale formed on the OD of commercial FeCrAl (APM) tube specimens ramped to 1400°C in the high-temperature furnace using (A) steam and (B) air and to 1500°C in (C) steam and (D) air.

The first commercial low-Cr FeCrAl tube evaluated was C135M. The results shown in Figures 2 and 7 were disappointing. In both air and steam, the tube specimens were completely oxidized. Specimens were then tested in the TGA in order to track the mass gain during ramping and prevent the specimen from completely oxidizing. It was possible to ramp the C135M specimen in low-velocity steam to 1500°C without complete failure. The cross-section of the specimen is shown in Figure 8A. While a thin alumina scale formed on the OD, the tube inner diameter (ID) formed a much thicker Fe-rich oxide. Figure 8B shows the C135M specimen ramped in air in the TGA. In this case, because of the high mass gain, the specimen was stopped at ~1190°C and cooled to room temperature in argon. Again, a thick oxide formed on the ID and a thin scale on the OD, suggesting an issue with the ID surface. For comparison, the thin OD scale formed on APM is shown in Figure 8C after ramping in steam to 1500°C. The oxide was similar to that formed on C135M. Figure 8D shows the oxide formed on Zr-4. In this case, the ramp was stopped at ~1030°C because of high mass gain.

Figure 7:

Images of FeCrAl tube specimens (A,B) C135M and (C,D) B136Y ramped in steam in the high temperature furnace at (A,C) 1400° and (B,D) 1500°C.

Figure 8:

Light microscopy of polished cross-sections of tube specimens ramp tested in the TGA using low-velocity steam or air: (A) C135M in steam to 1500°C, (B) C135M in air to 1190°C, (C) APM to 1500°C in steam, and (D) Zr-4 to 1030°C in steam.

A second FeCrAl tube batch designated B136Y was also evaluated. In Figure 2, the mass gains after ramping to 1400°C in steam and air were similar to those for APM. For example, the mass gain in steam was 0.32 mg/cm² for APM and 0.48 mg/cm² for B136Y. This B136Y tube specimen is shown in Figure 7. Increasing the temperature to 1500°C caused a catastrophic attack (Figure 7), and the mass gain could not be measured. Inexplicably, the ceramic catch pan ended the test above the specimen (compare Figure 7C and D). The test was repeated, and a similar result was obtained. Repeating the test without the pan resulted in a similar oxidation attack to the specimen. The suggestion is that there was a liquid reaction product at one point. To support that point, Figure 9 shows the C106M tube specimen after ramp testing to 1490°C in steam in the TGA. Clearly, some type of melting occurred during this test with a large mass gain during the last 10°C of heating. Unfortunately, only small quantities of the C106M were available for oxidation testing, and no further experiments could be conducted to study this behavior.

Figure 9:

Images of C106M tube specimens ramped to 1490°C in low-velocity steam in the TGA.

Finally, a B136Y specimen was ramp tested in the TGA to ~1480°C in steam, and a mass gain of only 0.66 mg/cm² was observed after the exposure. The ramp was inadvertently stopped before reaching 1500°C. This result combined with other comparisons between the two test rigs has led to the hypothesis that steam velocity was affecting the behavior. An obvious possibility is that Al(OH)₃ evaporation (Opila, 2004; Opila & Myers, 2004) was occurring and affecting behavior. Initial experiments were conducted for 4–50 h in the high-temperature furnace to quantify the evaporation using specimens of nominally pure alumina that were previously annealed at 1600°C to remove any impurities. Figure 10 shows an Arrhenius plot of the measured linear evaporation rates with the same water flow rate used in these experiments. An activation energy was identified from the curve fit. The 181 kJ/mol energy is similar to literature values of 190–230 kJ/mol for Al(OH)₃ formation at 1250–1550°C in O₂-50% H₂O (Opila & Myers, 2004).

Figure 10:

Arrhenius plot of Al(OH)₃ evaporation rates from bulk alumina specimens in high-velocity steam in the high-temperature furnace.

4 Discussion

These results are part of the evolution of evaluating the steam oxidation behavior of FeCrAl alloys for ATF application and the need for information to input into the various accident models available (Vierow et al., 2004; Ott et al., 2014; Robb et al., 2014; Robb, 2016). Prior work focused on alloy coupons with polished surfaces, while these results are from tube specimens using the as-fabricated surfaces. Surface finish may be an issue with the accelerated ID attack on C135M (Figure 8A and B), and explains the poor ramp testing results (Figure 2). As commercial tube processing is developed, this should be less of an issue. An alternative explanation for the poor performance of C135M is that it has reduced wall thickness, as shown in Figure 8A and D. However, the OD had no problem forming protective alumina; only the ID was an issue.

As mentioned initially, these test conditions are rather unique as there are few applications with steam environments above 1000°C, and most applications for power generation or transportation require long-term durability (Pint et al., 2006a). Typically, FeCrAl lifetime modeling focuses on alumina scale growth kinetics and Al depletion (Quadakkers & Bennett, 1994; Pint et al., 2012); however, those aspects are not relevant for these very short exposures. Increasing the hold time from 1 min in this study to 4 h did not result in breakaway due to Al depletion (Unocic et al., 2017). Practically speaking, alumina-forming alloys can only be used up to ~1100–1150°C and expected to operate for multiple years. Thus, there is no prior experience with these conditions and few applications except an accident scenario that would be expected to operate so near the solidus temperature. Unfortunately, it appears that the steam environment and temperature range of interest for modeling represent some complications that are not easily understood. First, it is apparent that different results were obtained in the two test rigs, the clearest example being B136Y at >1400°C. Protective behavior was observed in the low-velocity steam TGA, and non-protective behavior in the high-velocity steam experiment. The initial hypothesis is that Al(OH)₃ evaporation may affect behavior, which is velocity dependent. Evidence of evaporation has been observed in prior studies of Al-forming alloys and coatings in air with H₂O (Pint et al., 2006b). The measured rates were rather low to affect these short experiments, and modeling is in progress to determine how much these rates could affect the oxide thickness. An alternative hypothesis is that the heating rate is affecting these results, and that difference in heating rate between these two experiments (5°C/min vs. 15–19°C/min) is currently being evaluated.

A second complication is the clear indications of melting observed in Figures 7 and 9, which appear to be more prevalent in steam. Calorimetry experiments of mixed FeCrAl oxides (pieces of failed coupons) have shown no indication of melting in this temperature range (Robb, 2016). However, those post-exposure analyses are missing two important features: steam and the oxidation reaction. It is not possible with current equipment to perform that analysis in a steam environment. Also, the heat of reaction, particularly the rapid formation of FeO_x, could locally cause heating, resulting in melting, similar to the heat generated by the reaction of Zr in steam. Comparisons of literature data show that the reaction rates of Fe and Zr in steam are of the same order of magnitude (Pint et al., 2015a).

Neither melting nor accelerated oxidation was observed with the 20% Cr APM tube specimens until >1500°C. One implication is that dropping the Cr to improve radiation tolerance (i.e. inhibit α′ formation; Field et al., 2015; Edmondson et al., 2016) will significantly drop the maximum temperature where the cladding can remain protective in steam. The accident modeling will determine if this debit is worth the benefit. It is well known that Cr, through a “third element” effect (Stott et al., 1995), assists in the formation of alumina in FeCrAl. Prior work showed that higher Al contents were favored in steam at 1200°C when the Cr content was reduced (Pint et al., 2015b). However, this effect of Cr content has never been studied at such high temperatures. Previously, it was shown at lower temperatures that higher Cr content increased the lifetime in long-term experiments due a decrease in the critical Al content remaining at breakaway oxidation (Gurrappa et al., 2000). However, the breakaway in these experiments is not likely associated with Al depletion due to the short exposure times.

Rather than the higher Cr content, it could be argued that the better performance for APM and APMT is due to their higher dispersion-strengthened creep properties compared to wrought FeCrAl alloys (Jönsson et al., 2004, 2013). However, similar decreases in maximum temperature capabilities have been observed for ODS versions of these low-Cr FeCrAl compositions (Pint et al., 2014; Unocic et al., 2015). While a mechanical aspect cannot be completely ruled out at this time, there does appear to be some consistency in behavior between wrought and ODS low-Cr FeCrAl.

Ultimately, alloy composition is a key variable in these studies. The “C” alloys with Mo and Si (Table 1) were designed to improve the tensile properties of FeCrAl (Pint et al., 2015a; Yamamoto et al., 2015). It may be that these additions either mechanically or chemically have a negative effect on oxidation resistance, and the potential benefits will need to be compared to the potential degradation in maximum use temperature. More work is needed to clearly demonstrate if such a debit exists. Other differences among the low-Cr alloys include the Y content. The B136Y tube has a lower Y content than the other alloys (Table 1). The optimum Y content has not been determined. Prior work has focused on minimizing growth rates and improving scale adhesion (Pint, 2003; Naumenko et al., 2016), not improving the maximum use temperature in steam.

One aspect of testing in steam appears to be confirmed by these results: when protective alumina forms, the growth rate is similar in steam and air. This was first observed using isothermal experiments on APMT up to 1475°C (Pint et al., 2015b). This is not surprising as alumina is a very stable oxide that can form at very low oxygen partial pressures. One difference between these environments is the equilibrium oxygen partial pressure, which continues to increase in steam from ppb levels at 600°C to ppm levels at ≥1200°C.

Finally, it has taken years of development to get to this stage of testing various FeCrAl compositions in tube form, similar to that used for fuel cladding. Until recently, it was difficult to obtain sufficient material to conduct these experiments. Only a few pieces of the C tubing compositions were available for steam oxidation testing with other experiments requiring material. In particular, tube burst testing has yet to be conducted to confirm an increase in burst temperature (Massey et al., 2016) due to the higher tensile strength of the C materials compared to the B materials (Pint et al., 2015a; Yamamoto et al., 2015). With the identification of several different hot and cold processing methods, tube availability should no longer be an issue. Assuming that the technology implementation plan for FeCrAl alloys continues to make progress, a commercial reactor demonstration in 2018 appears feasible.

5 Summary

The very high temperature (≥1400°C) steam and air oxidation behavior of various FeCrAl tubes was compared to Zircaloy-4 to confirm the benefit in reduced oxidation rate near the solidus temperature for these alloys. A commercial FeCrAl alloy, APM, with ~20% Cr was used as a baseline for ramp testing with 1 min hold times at peak temperatures up to 1700°C. At 1400–1500°C, excellent oxidation resistance was observed with thin external alumina scales formed in contrast to the thick oxides formed on Zr-4 under similar conditions. For lower-Cr FeCrAl compositions, one batch performed poorly at 1400°C, while a second batch formed a protective scale at 1400°C but was fully oxidized at 1500°C. The differences in results between the two test rigs suggest a role of gas velocity, and alumina evaporation rates were measured on bulk alumina specimens to support future modeling work. Also, several results suggest that melting is occurring well below the solidus temperature. Further work is needed to fully understand the oxidation mechanisms at these high temperatures in steam, and to determine the effect of FeCrAl composition on the maximum use temperature for this ATF cladding candidate.