Oxidation at high temperatures in steam atmosphere and quench of silicon carbide composites for nuclear application

Nuclear Engineering and Design 295 (2015) 468–478

Contents lists available at ScienceDirect

Nuclear Engineering and Design journal homepage: www.elsevier.com/locate/nucengdes

Oxidation at high temperatures in steam atmosphere and quench of silicon carbide composites for nuclear application V. Angelici Avincola ∗ , M. Grosse 1 , U. Stegmaier 2 , M. Steinbrueck 3 , H.J. Seifert 4 Karlsruhe Institute of Technology, Hermann-von-Helmholtz-Platz 1, 76344 Eggenstein-Leopoldshafen, Germany

h i g h l i g h t s • SiC produces several times less hydrogen than Zircaloy in case of accident. • SiC tube maintained coolable shape after quench tests at 2000 ◦ C. • Oxidation tests in steam at 1600 ◦ C showed bubbles formation on SiC surface.

a r t i c l e

i n f o

Article history: Received 19 January 2015 Received in revised form 29 September 2015 Accepted 1 October 2015 Available online 17 November 2015

a b s t r a c t After the Fukushima accidents, the need for a fuel-cladding system with better performance in accident conditions has been raised. Silicon carbide and its composites are nowadays candidates for replacing zirconium alloys as cladding material for light water reactors. This paper reports about oxidation and quench tests of relevant silicon carbide samples at temperatures up to 2000 ◦ C in steam atmosphere. During the experiments the reaction gases were analyzed by mass spectrometry, and after the tests the samples were characterized by means of optical microscopy, scanning electron microscopy, as well as XRD and X-ray tomography. Oxidation results showed good resistance of SiC composite despite the formation of bubbles on the surface. The quench tests proved the samples retained their original shape. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Since the beginning of the nuclear industry, efforts to improve the fuel performance of light water reactors (LWR) have been made by industry and research institutions. After the severe accidents at the Fukushima Daiichi Nuclear Power Station in 2011, a need for enhanced system safety has been demonstrated. The actual fuel rods, which consist of zirconium alloy cladding coupled with uranium dioxide, or a mixture of uranium and plutonium dioxide fuel, are mature and reliable during operation. However, their behavior in the case of extremely rare accident events raises significant concerns. In particular, the strongly exothermic interaction of the cladding with steam produces free hydrogen. Combustion

∗ Corresponding author. Tel.: +49 721 60822872. E-mail addresses: [email protected] (V. Angelici Avincola), [email protected] (M. Grosse), [email protected] (U. Stegmaier), [email protected] (M. Steinbrueck), [email protected] (H.J. Seifert). 1 Tel.: +49 721 60823884. 2 Tel.: +49 721 60824981. 3 Tel.: +49 721 60822517. 4 Tel.: +49 721 60823895. http://dx.doi.org/10.1016/j.nucengdes.2015.10.002 0029-5493/© 2015 Elsevier B.V. All rights reserved.

of hydrogen worsened the Japanese accidents, causing explosions and destruction of the reactor building. This event underlined the evidence of the need for an improved cladding, which would have a better performance under accident conditions maintaining an excellent behavior during operation. By definition, accident tolerant fuels (ATF) are fuel-cladding systems, which could tolerate loss of active cooling in the core for longer periods of time than the currently adopted systems, while keeping an acceptable or even better performance during normal operation. Silicon carbide is a suitable candidate for substituting the actual cladding (Hallstadius et al., 2012): it combines high performance in extreme environments (Stempien, 2011; Alpettaz et al., 2012; Lee et al., 2013) and chemical compatibility with the fuel system (Katoh et al., 2011). For many years, silicon carbide has been tested as a possible material for nuclear power plants (Snead et al., 2007). Today it is being investigated as a structural material in both light water and high temperature reactors, and different designs are studied worldwide. The monolithic SiC behaves as a ceramic material, thus undergoing brittle fracture. The combination of monolithic SiC with SiC fibers in a SiC/SiCf composite leads to a pseudo-ductile behavior. For this reason, the use of a SiC/SiCf composite seems to be very promising. The SiC fibers have been

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

improved over the last years, going through three different fiber generations in order to reach an as close as possible stoichiometric composition (Bunsell and Piant, 2006). Indeed, the presence of the silicon oxycarbide phase, which was typical in the first fiber generation, caused instability of the material, in particular at high temperatures and under irradiation. Although the second fiber generation had overcome this problem, the presence of a large amount of free carbon was still affecting the oxidation and creep resistance (Bansal and Lamon, 2015). The third generation of SiC fibers produced was nearly stoichiometric and therefore the behavior was comparable to that of the SiC bulk material. To date, Hi-Nicalon Type-S and Tyranno SA SiC fibers are known to be stable under irradiation (Bansal and Lamon, 2015). Assessment of these composites at very high temperatures and under accident conditions is still on-going. In this paper, oxidation tests of SiC–SiCf cladding samples are described at temperatures between 1600 ◦ C and 1800 ◦ C in steam and quenching in water at temperatures of up to 2000 ◦ C. The response to these scenarios, including hydrogen production, has been measured. 2. SiC oxidation in steam The oxidation between silicon carbide and steam at high temperatures can be categorized into two different reactions, called passive and active oxidation. Active oxidation occurs in the case of very low water vapor partial pressures (Opila and Jacobson, 1995) (<10 Pa at 1427 ◦ C). In the case of higher oxidant partial pressure, the SiC undergoes reaction (Eq. (1)), developing a dense protective layer of silica (SiO2 ): SiC + 3H2 O = SiO2 + 3H2 + CO

(1)

In case passive oxidation occurs, the silicon carbide oxidation behavior follows the linear-parabolic model of Deal and Grove (Deal and Grove, 1965; Costello and Tressler 1986). At the same time, silica volatilization occurring via the following reactions degrades the oxide scale: SiO2(s) + 2H2 O(g) → Si(OH)4(g)

(2)

SiO2(s) + H2 O(g) → SiO(OH)2(g)

(3)

SiO2(g) +

1 1 H2 O(g) → SiO(OH)(g) + O2(g) 2 4

(4)

Volatilization is described by a linear constant, which is combined with the parabolic constant in the so-called para-linear behavior, which was formulated for Cr2 O3 by Tedmon (1966) and applied for the first time to model the oxidation of SiC by Opila and Hann (1997). This is defined by the paralinear equation (Eq. (5)), where kp is related to the parabolic oxidation of the SiC, whilst kl describes the volatilization. kp dx = − kl 2x dt

(5)

In this relation, x is the oxide thickness and t the time. The parabolic constant kp has been demonstrated to be proportional to P(H2 O) with a power law exponent of one, whilst the linear constant kl is proportional to the velocity (v) and the total pressure Ptot (Eq. (6)) (Opila and Hann, 1997). kl ∼

v1/2 P(H2 O)2 1/2 Ptot

(6)

Different authors have investigated the oxidation of SiC in steam. Jorgensen et al. (1961) tested SiC in water at up to 1514 ◦ C using a thermobalance. In his work, he stated that the oxidation rate of silicon carbide depends on the water vapor partial pressure. Narushima et al. (1990) tested CVD–SiC with a thermobalance

469

(TGA) at up to 1650 ◦ C in wet oxygen (10 kPa steam) and suggested that the rate-controlling step is oxygen diffusion in the parabolic regime based on the activation energy found. Opila (Opila and Smialek, 1999) developed the paralinear kinetic model and mapped out the conditions to apply it, elucidating the strong dependency of the volatilization on the external conditions (Opila, 2003). Oxidation tests at high pressures of up to 2 MPa and temperatures of as high as 1700 ◦ C have been performed by Oak Ridge National Laboratory (ORNL) (More et al., 2000; Terrani et al., 2014; Farmer et al., 2014). It was found that at high steam partial pressures, the oxide scale develops porosity due to the formation of CO gases, degrading the silica protective properties. This non-protective porous layer increases with time, leaving vitreous silica of constant thickness on the SiC. The oxidation of SiC is a diffusion-controlled reaction, although no agreement has been found regarding the species diffusing. It has been stated by many authors that the presence of water vapor enhances the oxidation rate. This has been explained considering that the H2 O molecule is smaller than the O2 molecule (276 versus 320 pm at 700 ◦ C) (Presser and Nickel, 2008), and it reacts with the silicon–oxygen network forming immobile SiOH groups (Doremus, 2001). Moreover, the H2 O molecule alters the SiO2 network and increases the diffusion of oxidant species (Irene, 1977). The plot of the logarithm of the parabolic rate constant kp versus the logarithm of the partial pressure of water provides information on the oxidant species. In the case SiC oxidizes in oxygen atmosphere, the slope indicates diffusion of molecular oxygen (n = 0.5). In the case of water, the n coefficient has higher value. The most reliable explanation yields to parallel diffusion of a molecular and an ionic hydrogen-containing phase (H2 O and OH− ) (Presser and Nickel, 2008). A third oxidation phase has been noted by different authors (Narushima et al., 1994; Terrani et al., 2014), which consists in the development of bubbles on the surface and therefore implies degradation of the protective silica scale.

3. Experimental procedure 3.1. Samples Samples of Ceramic Matrix Composite (CMC) silicon carbide/silicon carbide fibers from two producers (CEA and CTPLLC ) were used to perform oxidation and quench tests. A summary of the samples used is shown in Table 1. The cladding design developed by CEA laboratories (Bansal and Lamon, 2015; Buet et al., 2012) consists of three braided layers (with 45◦ of orientation) of Hi-Nicalon S fibers provided by Nippon Carbon, reinforced with SiC CVI matrix. To improve the mechanical behavior of the composite, which is controlled by the matrix–fiber bonding, the interface between the fiber and the matrix has been refined with pyrocarbon, which is known to be the best interphase; a layer ≈ 20–30 nm thick was deposited on the SiC fibers prior to the matrix infiltration. As protection, an external 50 ␮m thick CVD layer was applied. 21 tube samples of a length of 20 mm, an external diameter of 9.5 ± 0.1 mm and a wall thickness of 0.89 ± 0.01 mm have been provided by CEA (Table 1).

Table 1 Samples used for the experiments. Producer

Number of samples

Temperature ( ◦ C)

Quench

CEA CEA CEA CTP CTP

10 5 6 4 Monolith 4 Composite

1600 1700 1800 2000 2000

× × × √ √

470

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

Fig. 1. H2 O dissociation at temperature between 800 and 2400 ◦ C calculated with Thermo-Calc (SGTE dataset).

The typical major impurities are Si, C, O, Al, with carbon excess (C/Si 1.07 at.%) (Bansal and Lamon, 2015). SiC cladding tube samples were provided by Ceramic Tubular Products LCC, 25 mm length, 11 ± 0.5 mm external diameter and 1.2 ± 0.1 mm wall thickness (Table 1). The architecture is the socalled TRIPLEX (Feinroth, 2012), and it consists of three SiC layers: the inner layer is a monolith SiC tube operating as fission product release barrier and providing strength to the cladding; SiC fiber–SiC matrix composite is the middle layer, which gives pseudo-ductility. The external layer consists of a deposited SiC as environmental barrier coating. The denomination m indicates monolithic samples, which consist in the inner monolith SiC tube. The others are tri-layered samples. The sample CTP 8 has been quenched once at 1600 ◦ C, then re-heated at 2000 ◦ C and quenched again (CTP 8b). 3.2. Facilities The tests described in Section 3.3 were carried out in two different furnaces. For annealing at 1600 ◦ C, a horizontal alumina tube furnace, so-called BOX rig, with a suspended sample and an air lock allowing for the exchange of specimens at reaction temperature under a defined atmosphere were used (the facility is described in detail by Steinbrück et al., 2007). Studies conducted by Opila (1995) have highlighted the influence of the impurities which are present in alumina furnaces. These impurities, in particular sodium, increase the oxidation rate constant by about one order of magnitude, with respect to the same conditions in a quartz furnace. In the facility used in this work, the water was stored in a vessel under pressure. The gas supply system for steam and Ar consists of two gas flow controllers, one liquid flow controller and a so-called controlled evaporator mixer unit (CEM), where the liquid water was evaporated and mixed with the non-condensable gas. The whole system is delivered by Bronkhorst High-Tech B.V. The steam partial pressures were calculated considering the amount of moles, respectively of water and argon. This approach was checked by means of thermodynamic calculations: the water dissociation in the test rig was calculated at the experimental temperatures using Gibbs energy minimization techniques implemented in the Thermo-Calc software with the SGTE database (Scientific Group Thermodata, 2014). Agreement has been found comparing the results with the reviewed data in the work by Jellinek (1986). As a result from the calculation presented in Fig. 1, significant dissociation of water according to Eq. (7) starts only at temperatures beyond 1200◦ C, i.e., it is 0.1% at 1357 ◦ C. The water dissociation increases further

Fig. 2. Sketch of the silicon carbide plus the graphite stick used as susceptor and the SiC caps.

with temperature and it is around 0.5% at 1600 ◦ C and about 3.5% at 2000 ◦ C. 2H2 Og = 2H2g +O2g

(7)

Ionization of water is not expected at these temperatures. At temperatures higher than 1600 ◦ C, the samples were tested in a vertical quartz furnace (QUENCH-SR) heated by induction coils. The power was supplied to the coil by a 20 kW oscillator, at a frequency of up to 700 kHz. The rig allowed quenching the sample by water heated between 30 and 90 ◦ C. The temperature of the sample was controlled by means of a two-color pyrometer, connected to a controller, which adjusted the voltage and current accordingly. The device allowed the use of different gas mixtures, such as argon and steam. The samples were held by an alumina or zirconia sample holder and a graphite pellet was used as susceptor as shown in Fig. 2. To prevent oxidation of the graphite, alumina glue was used to cover the graphite extremities, and SiC caps were used to close the tube. During the quench procedure, the samples face thermal stresses due to radial and axial temperature gradients. After the test, samples were analyzed by X-ray tomography (GE v/tome/xs) using beam energy of 120 kV and a spatial resolution of 0.024 mm, in order to investigate the presence of cracks, which would compromise stability of the pin and the fission gas retention. This technique allows internal inspection of the sample depending on the total mass attenuation of the elements’ atomic number. Both furnaces were connected to a Balzers GAM 300 quadrupole mass spectrometer (MS) for quantitative analysis of the gaseous reaction products. The volumetric flow rates were calculated using argon as reference gas. The hydrogen production, which occurs via Eq. (1), was calculated referring the measured H2 concentration to the known argon mass flow rate according to the Eq. (8) V˙ H2 =

CH2 CAr

· V˙ Ar

(8)

V˙ H2 and V˙ Ar are the volumetric flows of hydrogen and argon per unit of time, CH2 and CAr are the concentrations of the same elements measured by the mass spectrometer. 3.3. Test conditions Isothermal oxidation tests were conducted at 1600 ◦ C in the BOX furnace at different steam partial pressures (10, 30 and 60 kPa) in pure argon in two different total gas velocities (0.1 m/s and 0.2 m/s). Each specimen of the first group of samples provided by CEA (see Table 1) was loaded in the furnace already heated at the

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478 Table 2 Overview of the SiC samples used for oxidation tests at 1600 ◦ C and quench tests.

471

Table 3 SiC oxidation experiment in argon and steam at 1600 ◦ C.

Sample

PH2 O (kPa)

Sample

Atmosphere

Sample

PH2 O (kPa)

Temperature (◦ C)

Time (h)

Total flow (l/h)

S1 S2 S3 S4 S5 S6 S7 S8 S9

10 10 60 30 60 30 60 60 60

CTP CTP CTP CTP CTP CTP CTP CTP

Argon Argon + Steam Argon Argon Argon Argon + Steam Argon + Steam Argon + Steam

S1 S2 S3 S4 S5 S6 S7 S8 S9

10 10 60 30 60 30 60 60 60

1600 1600 1600 1600 1600 1600 1600 1600 1600

5 16 22 22 22 25 22 64 22

66 66 64 64 133 133 133 64 133

1m 2m 3m 4m 5 6 7 8

prescribed temperature and were held in the reaction chamber for time period up to 64 h. Argon 6.0 (99.9999% purity) was used in all the experiments. The loading of the sample reduced the furnace’s temperature by approximately 5 ◦ C. In order to establish the stationary conditions, the steam was injected 10 min after the loading of the sample. The oxidation tests were performed for up to 64 h. Using a bench top balance, the mass of the samples was measured before each test, during the experiments at different times thanks to an extractable sample holder (air lock), and after the test. In Table 2, the list of samples oxidized at 1600 ◦ C at different steam partial pressures and the list of the samples quenched are presented. Experiments at temperatures beyond 1600 ◦ C were conducted in the QUENCH-SR with the remaining samples manufactured by CEA (Table 1). In this facility, the heating phase took place under inert atmosphere at up to 1400 ◦ C (argon 40 l/h), with a heating rate of 10 ◦ C/min. Once the temperature of 1400 ◦ C was reached, the steam was injected. The samples manufactured by CTP were used to study the behavior in the case of quenching. In these experiments, the temperature of the external SiC tube was raised to up to 2000 ◦ C and the samples were quenched by water at 90 ◦ C. The quench has been done using a quartz cylinder filled with water moved by means of a mechanical translation from the bottom of the facility at a velocity of 1 mm/s. The samples remained at the test’s temperature until the water quenched the SiC surface. The heating was automatically switched off by an inductive proximity switch attached to the mechanical translation which moved the water cylinder. After the tests, the samples were embedded in epoxy and were ground and polished using a series of diamond plates

with different coarsenesses. After this procedure, the samples were analyzed with optical microscopy (MeF3 Reichert-Jung), and scanning electron microscopy (Philips XL30S). XRD analysis (in Bragg–Brentano geometry, with CuK␣ radiation) and X-ray tomography (GE v/tome/xs) with a spatial resolution of 0.024 mm, were also conducted on some specific samples. 4. Results 4.1. Oxidation of SiC in steam at 1600–1800 ◦ C Isothermal oxidation tests have been performed at 1600 ◦ C up to 64 h in pure argon (6.0) and different steam partial pressures (10, 30 and 60 kPa). In Table 3 are summarized the oxidation experiments. An overview of the external surfaces of the samples oxidized at 1600 ◦ C at different steam partial pressures is shown in Fig. 3. The samples present some features on the surfaces, which are enhanced after 64 h oxidation at 60 kPa steam partial pressure, appearing as bubbles. After having been mounted in epoxy and polished, the samples were investigated by means of optical microscope. As example, Fig. 4 shows the section of sample S5. The external CVD layer is visible, as well as the fibers. Different kinds of porosity are present in the bulk material: although the layers are tightly braided, this kind of texture leads to bigger porosity when tows are crossing. As can be seen in Fig. 5, the influence of the partial pressure is visible considering the experiments at 10 and 30 kPa. Despite the uncertainties of the data, determination of the parabolic rate constant for oxidation (kp ) has been attempted. In order to calculate

Fig. 3. Overview of samples oxidized at 1600 ◦ C at different steam partial pressures.

472

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

Fig. 4. Micrograph made by means of optical microscope of the cross section of the sample S5 oxidized for 22 h with 60 kPa steam partial pressure.

Fig. 5. Mass gain of samples oxidized at 1600 ◦ C at 10 kPa and 30 kPa steam partial pressure in argon.

the surface specific oxidation rate, the geometric area has been measured. A parabolic rate constant of 4.8 × 10−4 mg cm−2 s−1/2 for 10 kPa steam partial pressure and 1.2 × 10−3 mg cm−2 s−1/2 for 30 kPa was calculated. As can be seen in Fig. 6, at higher steam partial pressures (i.e., 60 kPa) the oxidation behavior is different. The calculation of kp was not possible due to the scattering of the data. Considering oxidation carried out for longer period of time, the samples underwent severe mass loss and degradation due to the presence of bubbles. The mass loss due to the volatilization of silica (as in Eqs. (2)–(4)), has been considered negligible due to the thermal-hydraulic conditions (Opila, 2003). The mass spectrometer output signal highlighted the correlation between the steam partial pressure and the oxidation rate (Fig. 7). The mass spectrometer data at 60 kPa are more scattered due to the strong formation of bubbles on the silica surface. The SEM sections of the samples S2, S4, S5, and S8 are presented in Figs. 8 and 9. The samples oxidized at 10 kPa steam partial pressure have exhibited a dense, compact silica layer of about 5 micrometers. The samples oxidized at 30 kPa (e.g., Fig. 9a), show traces of bubble formation on the surface, but those are not visible on the section, whereas on the section of the samples oxidized at 60 kPa steam partial pressure, bubbles with at diameter of almost 200 ␮m are visible.

Fig. 6. Mass gain of samples oxidized at 1600 ◦ C at 10, 30, and 60 kPa steam partial pressure in argon.

Fig. 7. Hydrogen production during the oxidation phase at 1600 ◦ C of SiC samples in different steam partial pressures (each color corresponds to a sample S1, S2 at 10 kPa, S4 and S6 at 30 kPa, S3, S5, S7 and S8 at 60 kPa). The time interval is chosen since no extraction of the sample was done during this time.

Fig. 8. SEM micrograph of sample S2 showing the silica scale developed on silicon carbide samples oxidized at 1600 ◦ C at 10 kPa steam partial pressure.

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

473

Fig. 9. SEM micrographs of the silica scale developed on silicon carbide samples oxidized at 1600 ◦ C at different steam partial pressures; (a) steam 30 kPa after 22 h (Sample S4); (b) steam 60 kPa after 22 h (Sample S5); (c) steam 60 kPa after 64 h (Sample S8).

Fig. 10. XRD measurement of sample S6 oxidized for 22 h at 1600 ◦ C.

In Fig. 10 the XRD measurement of the section of sample S6 is presented. The XRD patterns of the uncoated Hi-Nicalon Type S SiC fibers can be recognized. They consist in 5 peaks at the 2␪ of 35.7◦ , 41.4◦ , 60.4◦ , 71.8◦ and 76◦ (Singh et al., 2012), even though the second and the last peaks are quite weak. A small amount of ␣-SiC can be due to microtwins and stacking in the ␤-SiC crystalline core (Buet et al., 2012). A broad peak at 2␪ ≈ 20◦ reveals the amorphous nature of the silica.

After 25 h, bubbles of about 70 ␮m diameter are easily produced in the SiO2 scale. The bubbles are developed onto a thin layer visible in Fig. 11 at the interface between the silica and the silicon carbide of the sample S5. A qualitative EDX analysis proves it to be a carbon layer. Backscattering images did not highlight the presence of impurities which could have enhanced the bubble formation. EDX analysis on the polished samples has shown the different layer composition of the section: the silicon carbide next to the C layer results to be depleted in C. After longer oxidation time a 10 ␮m thick carbon layer is clearly visible at the interface between the SiC and the SiO2 (S8 oxidized for 64 h in Fig. 12). Oxidation tests at 1700 ◦ C and 1800 ◦ C for 1 h were performed in the QUENCH-SR furnace described previously. On samples tested at these temperatures, it was not possible to measure the mass before and after the test, due to the inclusion of Al2 O3 glue on the surface. According to the mass spectrometer’s measurement, the hydrogen generation rate increased with increasing temperature. In addition, the hydrogen generation fluctuated to a greater degree at 1800 ◦ C than at 1700 ◦ C. The average hydrogen flow rate for each sample during the isothermal period was calculated. Considering the temperature dependence, the maximum value measured at 1600 ◦ C was 3 g m−2 h−1 . At 1700 ◦ C and 1800 ◦ C the hydrogen production was measured between 12 and 61 g m−2 h−1 . The values were scattered, maybe due to contamination by different materials, such as the graphite used to heat up the samples. The silica thickness of the samples oxidized at 1700 ◦ C and 1800 ◦ C was measured using SEM techniques. It resulted dense as shown in Fig. 13. Comparing to the measurements done at 1600 ◦ C, the silica layer at higher temperatures is considerably thinner, and no traces of bubbles in the

Fig. 11. SEM micrograph and elemental mapping of the section of sample S5 oxidized at 1600 ◦ C.

474

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

Fig. 12. SEM micrograph of the interface between silicon carbide and silicon oxide of sample S8. The carbon layer is visible and pointed by the arrows.

Fig. 14. Off-gas composition during heating and quench of the CTP 8 sample.

Fig. 13. SEM micrograph of a cross section at 10,000× magnification of a sample oxidized at 1700 ◦ C. The arrows show where the silica layer begins and ends. Table 4 Sample, atmosphere and temperature during quench tests. Sample

Atmosphere

T (◦ C)

Result

CTP CTP CTP CTP CTP CTP CTP CTP CTP

Argon Argon + Steam Argon Argon Argon Argon + Steam Argon + Steam Argon + Steam Argon + Steam

2000 2000 2000 2000 2000 2000 2000 1600 2000

Broken Broken Intact Intact Intact Broken Intact Intact Intact

1m 2m 3m 4m 5 6 7 8 8b

interlayer between SiO2 and SiC have been noticed. This can be due to the high temperature, which is close to the softening point of the silica and which can lead to the relocation of the excess silicon dioxide at the bottom of the samples. 4.2. Quenching of SiC samples from 2000 ◦ C Samples produced by CTP in Table 1 were oxidized and quenched. Table 4 listed the samples and the conditions used for the quench tests. A typical mass spectrometer measurement of the exhaust gases during the experiment is presented in Fig. 14. The hydrogen production can be correlated with the steam injection at 1400 ◦ C: after

a first peak, which corresponds to the steam injection and to the start of the oxidation, the H2 flow decreases. This can be seen as a parabolic type curve of the hydrogen production, due to the developing of SiO2 , which forms according to (Eq. (1)). Starting at 1700 ◦ C, the hydrogen production increases. This may be due to the loss of the protective oxide scale. The peaks of hydrogen at temperatures higher than 1800 ◦ C can be related to the reaction between graphite and steam, which could occur due to the weakening connection between the caps and the silicon carbide tube. At the end of the graph, the drop of the temperature corresponds to the moment when the quenching with water occurs. After quench, the surface of the samples showed traces of silica: on the surface of the monolith samples traces of bubbles can be noted (Fig. 15a), whilst on the surface of the composite samples a dense silica layer is visible (Fig. 15b). During the quench procedure, the sample faces high thermal stresses. In Fig. 16, the sample CTP 8 is shown after quenching at 1600 ◦ C. Under optical inspection, the tri-layered samples remained intact after each experiment, with the exception of one, which was broken. Post-test analyses by means of X-ray tomography have shown cracks in the inner monolith layer of sample CTP 7, as it can be seen in Fig. 17. It can be assumed that thermal stresses faced during quenching generated cracks along the inner monolith layer. A sequence of images extracted from the video recorded during the annealing in steam and quench in water of sample CTP 7 is presented in Fig. 18.

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

475

Fig. 15. SEM micrograph of the surface of sample CTP 2m and CTP 7 after quench.

The integral hydrogen release of sample CTP 7 was calculated and it is shown in Fig. 19. Comparison with hydrogen produced by Zircaloy-4 was done considering the integral corresponding to the temperature range between 1400 ◦ C and 1600 ◦ C. For the temperature range between 1400 and 1527 ◦ C a correlation determined at KIT was chosen to calculate the hydrogen release from Zircaloy-4 (Grosse, 2010), whereas for higher temperatures the Urbanic–Heidrick correlation was applied (Urbanic and Heidrick, 1978). A numerical code allowed to calculate the hydrogen production following the experimental conditions already described. The resulting hydrogen production shown in Fig. 19 is 0.003 kg/m2 from SiC and 0.12 kg/m2 from Zircaloy. According to these data, the amount of hydrogen produced by oxidation of SiC is 40 times lower than that produced by Zircaloy-4.

5. Discussion

Fig. 16. Sample CTP 8 after quenching from 1600 ◦ C with sample holder.

Fig. 17. X-ray tomography of sample CTP 7 after quenching at 2000 ◦ C.

In previous works, the influence of the steam partial pressure on the oxidation kinetics of silicon carbide has been studied at up to 1400 ◦ C (Opila, 1999). In Opila’s work, the log kp increased with increasing steam partial pressure. One explanation is that the water weakens the SiO2 original structure replacing the strong Si O Si bonding by the weaker Si OH OH Si (Wagstaff, 1969) increasing the oxidation rate. In Opila’s work (Opila, 1999), the kp derived from TG oxidation measurements of CVD SiC is 4.98 × 10−4 mg cm−2 s−1/2 in steam-oxygen atmosphere for 10 kPa and 6.07 × 10−4 mg cm−2 s−1/2 at 25 kPa at 1400 ◦ C, which is comparable with the result obtained in this work at 10 kPa (4.8 × 10−4 mg cm−2 s−1/2 ), while at 30 kPa, the value calculated in this work is slightly higher (1.2 × 10−3 mg cm−2 s−1/2 ). This can be due to the temperature and the 5 kPa steam partial pressure difference. In the case of silicon carbide oxidation, bubble formation has been frequently detected in the case of both dry and steam oxidation in the literature. Mieskowski et al. (1984) tested single crystal ␣ and ␤ SiC polycrystals in air and oxygen at temperatures between 1200 and 1400 ◦ C and bubbles were detected only on the polycrystalline samples. Kim and Moorhead (1990) studied the effects of oxygen partial pressure on the oxidation behavior of alpha-SiC at 1400 ◦ C in oxygen, where the presence of large pits suggested the eruption of bubbles. Schneider et al. (1998) performed experiments on CVD SiC as coating on a graphite substrate under low total pressures (100–800 Pa) and temperatures between 1300 and 1600 ◦ C using a microbalance. It was confirmed that two major parameters influence the development of the bubbles: the outside pressure and the silica properties. Therefore, Schneider inferred the external pressure threshold for bubble formation calculating the gas pressures at the SiC–SiO2 interface considering the

476

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

Fig. 18. Images from the video recorded during typical annealing in steam and quench in water (sample CTP 7).

Fig. 19. Hydrogen produced by oxidation of Zircaloy and SiC (Sample CTP 7) in steam under non-isothermal conditions.

different possible reactions, for both pure CVD and for the case of carbon inclusions. This would also explain the behavior found in the work by Mieskowski et al. (1984). Indeed, since polycrystals are more likely to have carbon inclusions, these are thought to be the cause of the pressure build-up in the SiO2 –SiC interface. Considering his approach, a gas pressure higher than atmospheric pressure would be necessary at the interface SiO2 and SiC to develop such bubbles. Opila (1999) investigated the SiC oxidation at different steam partial pressures of up to 1400 ◦ C, with O2 and argon as carrier gases. Bubbles were observed in the amorphous silica scale at 1100 ◦ C and 1200 ◦ C, increasing the quantity with increasing temperatures, exposure time and steam partial pressure. At 1400 ◦ C the silica scale appeared fully crystalline and no bubbles were detected. Ogura and Morimoto (2002) observed the production of bubbles in CVD SiC coating oxidized at low total pressure (0.8–440 Pa) between 1250 and 1700 ◦ C in oxygen. By means of a mass spectrometer, Ogura indicated the sharp rise of the channel 28 (CO) as the start of bubble formation. Goto (Goto and Homma, 2002; Goto, 2006) investigated CVD SiC using thermogravimetry at temperatures as high as 1717 ◦ C in Ar–O2 and N2 –O2 atmospheres. The output signal showed a zig-zag mass change, due to the

production and eruption of silica bubbles. In a more recent paper by the same author, the bubbles were detected at 1600 ◦ C after oxidation of CVD material for 42 h in oxygen. The samples did not exhibit the zigzag-shaped behavior, due to the smaller size of the bubbles (Katsui et al., 2014). Oak Ridge National Laboratories (Terrani et al., 2014) tested high purity CVD SiC and SiC/SiC composites5 in steam at temperatures between 1200 ◦ C and 1700 ◦ C, with environmental pressure in the range of 0.1–2 MPa for different flow velocities. Bubbles were detected at 1700 ◦ C at atmospheric pressure. In this work, the presence of bubbles was detected on the oxidized silicon carbide surface at every temperature tested. A dependence on the exposure time and steam partial pressure has been noted, confirming the work of Opila (1999). Considerations made for the bulk materials can be applied to the samples investigated in this work, since the external layer of the composite is made of CVD silicon carbide. Therefore, as long as the underlying composite layer is untouched, the theory expressed by Goto can be used to explain the development of the bubbles on the samples used in this work: in case that a dense SiO2 layer develops on the SiC external surface, the gas produced during the interaction between the SiC and the steam remains entrapped in the interface between the SiC and the SiO2 . When the gas pressure exceeds the external pressure, bubbles are formed. This can be summarized by three criteria: • Development of a dense silica scale. • Presence of higher gas pressure than external pressure at the interface. • Sufficiently low viscosity of the SiO2. Considering the first requirement, the solid silica, at the temperatures of interest in this paper, appears in three modifications: tridymite, cristobalite and amorphous. The transition between tridymite and cristobalite occurs at 1470 ◦ C, therefore at 1600 ◦ C, cristobalite is expected. Among the forms in which silica appears, its formation on SiC results in a first vitreous layer and in a

5 SiC/SiC composite resulted to behave identically to CVD SiC as long as the fiber–matrix interphase remains protected.

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

477

Fig. 21. Various fluxes during SiC oxidation in steam (Schneider et al., 1998).

Fig. 20. Vapor pressure of CO and SiO gas species in equilibrium with C–SiC–SiO2 and Si–SiC–SiO2 calculated with Thermo-Calc.

subsequent porous non-protective cristobalite scale (Tortorelli and More, 2003). It was proved that the steam, as well as the impurities coming from the reaction chamber, decrease the SiO2 crystallization temperature and enhance the crystallization process. Indeed, the crystallization follows a diffusion model in the case of dry oxidation, whilst in the case of steam oxidation it is linear. Despite the faster oxygen diffusion in vitreous silica than in beta-cristobalite (Rodri´ıguez-Viejo et al., 1993), the oxidation itself can be faster in the case of the formation of crystalline silica, since the crystalline form might lead to the development of cracks on the protective scale, and these cracks may act as shortcuts for oxidant transportation (National Research Council, 1970). XRD investigations on the samples oxidized in this work have shown the presence of amorphous silica, which support the bubble formation. The second condition necessary for explaining bubble formation is the presence of gases at the interface of SiO2 and SiC. They can either come from the oxidation of SiC or from the interaction between SiC and SiO2 . The first hypothesis implies that the permeability of the CO is lower than that of H2 O, in order to allow accumulation at the interface. Another source of CO production is the reaction between SiO2 and SiC (Narushima et al., 1994; Goto et al., 2002). Moreover, C impurities of the material can directly react with the H2 O and accumulate at the interface. The graph in Fig. 20 shows the vapor pressure at the interface between SiC and SiO2 calculated with the CALPHAD method, with the database from SGTE (Scientific Group Thermodata, 2014). In case that C is present, the bubbles can start to be produced at around 1507 ◦ C, when the gas internal pressure is equal to the external pressure. The presence of silicon raises this temperature to ≈1840 ◦ C. Following this model, the presence of C can justify the development of bubbles at relatively lower temperatures. The carbon layer has been already detected by Goto et al. (2014) and Katsui et al. (2014), after 43 h oxidation of CVD SiC in oxygen at 1600 ◦ C. This would suggest that at this temperature, the rate limiting factor would be the outward CO diffusion gas in the SiO2 , whereas H2 O diffuses in the opposite way from the external atmosphere to the SiC. This is supported by the diffusion coefficient of the two species CO and H2 O in silica. Data provided by Doremus (2001) stated that the values of the diffusion coefficients of molecular water follow Eq. (9).

 8360   cm2 

D = 1.3(10)−4 exp −

T

s

(9)

Eq. (9) yields to the value for the diffusion 3 × 10−7 cm2 /s at 1100 ◦ C. Compared with the value presented in the work by

Pongrácz (2009) at the same temperature for carbon monoxide (DCO = 1.8 × 10−9 cm2 /s), it is reasonable to assume that the CO accumulates at the interface and can be the limiting species. Regarding the presence of carbon in the material, the external layer of the oxidized samples was produced using a CVD method, which usually leads to a stoichiometric SiC. However, significant carbon excess was detected by the elemental analysis, thus a possible explanation for the carbon layer could be the migration of carbon toward the silica. The presence of carbon can be correlated with the flux of oxidant and products to explain bubble formation. Already Schneider et al. (1998) modeled three possible cases, shown in Fig. 21: in case 1, the flux of oxidant and products is the same, hence an equilibrium is established. In case 2, the CO and the H2 diffuse faster than H2 O. In case 3, the oxidant, i.e., H2 O, diffuses faster than the products. The third situation is the case where CO and H2 accumulate at the interface and, therefore, pressure build-up occurs. This is probably the situation that occurred during the experiments at 30 kPa and 60 kPa steam partial pressure in this work. Indeed, the stability diagram of the Si–C–O states that the carbon solid phase can exist with SiO2 solid and SiC solid when ac = 1. At 1873 K this corresponds to PCO = 2.3 × 105 Pa and PO2 = 2.3 × 10−10 (Katsui et al., 2014). This indicates that the outward diffusion of CO is the rate-limiting process of the oxidation of SiC. The last criterion is easily fulfilled, considering that the logarithm of the SiO2 viscosity in poise at 1600 ◦ C is 10.6. This value, according to the Urbain correlation (Doremus, 2002), is close to the softening point. Hence the SiO2 can easily expand producing bubbles. 6. Summary and conclusion In this paper studies of the oxidation in steam and quench in water of relevant SiC and SiC/SiCf designs are presented. • Samples provided by Ceramic Tubular Products have been heated up to 2000 ◦ C and then quenched in water at 90 ◦ C. The analysis of the hydrogen production between 1600 and 1800 ◦ C has shown hydrogen production of about 40 times lower compared to Zircaloy. 75% of the tri-layered samples survived the quench procedure, even though X-ray tomography has shown the presence of cracks. • Samples provided by the CEA laboratories have been oxidized at 1600 ◦ C for up to 64 h at different steam partial pressures, 10, 30 and 60 kPa. For the lower steam partial pressures, the values of the kp are in agreement with the data in the literature, presented at lower temperatures. At 60 kPa, the mass gain becomes irregular, and the kp cannot be calculated. This is mainly due to the formation of bubbles, which increases with exposure time and steam partial pressure. • Bubble formation was observed at 1600 ◦ C only at 30 and 60 kPa steam partial pressure. At higher temperatures, bubbles were detected during the experiments, but no traces were found in the silica scale in the post-analysis.

478

V. Angelici Avincola et al. / Nuclear Engineering and Design 295 (2015) 468–478

• Bubble formation can be explained considering that an amorphous silica layer develops on the silicon carbide surface, this has been confirmed by the XRD measurement. The raising of carbon activity at the interface between the SiC and the SiO2 increases the gas pressures at the interface and therefore promotes bubble formation. This suggests that under these conditions, the carbon monoxide accumulates at the interface and results in the rate-limiting step. • Oxidation at 1700 and 1800 ◦ C was performed for as long as 1 h and the silica layer was found to measure 2 ␮m in thickness. Bubbles were also noted during the experiments, but the post-test analysis is showing a dense silica layer. This can be explained considering that the temperature is close to or above the silica melting point. Therefore, during the cooling phase, relocation of silica could occur. • Overall, the silicon carbide performs very well at temperatures of up to 2000 ◦ C compared to zirconium alloys: in all the experiments, the samples maintained a coolable shape. From the oxidation experiments it can be stated that SiC tri-layered samples can give an additional safety margin regarding the beyond LOCA margin, maintaining the coolability in steam atmosphere for up to three days at 1600 ◦ C, and in the order of hours at 1700 ◦ C and 1800 ◦ C. Acknowledgment This work was supported by the Helmholtz program “NUKLEAR”. The authors wish to thank CEA and CTP for having provided the samples. Discussion with C. Sauder (CEA) has been of great help. Help in the lab from D. Mueller, A. Meier for the X-ray tomography, and from Dr. H. Leiste in the XDR analyses, has been appreciated. References Alpettaz, T., Gosse, S., Braun, J., Sauder, C., Chatain, S., Dugne, O., Guéneau, C., 2012. Investigation of the high temperature chemical interaction between UO2 fuel and Fe, B4C and SiC using Knudsen cell mass spectrometry. In: The Nuclear Material Conference 2012, Osaka, Japan. Bansal, N.P., Lamon, J., 2015. Ceramic Matrix Composites: Materials, Modeling and Technology. Wiley. Buet, E., Sauder, C., Poissonnet, S., Brender, P., Gadiou, R., Vix-Guterl, C., 2012. Influence of chemical and physical properties of the last generation of silicon carbide fibres on the mechanical behaviour of SiC/SiC composite. J. Eur. Ceram. Soc. 32, 547–557. Bunsell, A.R., Piant, A., 2006. A review of the development of three generations of small diameter silicon carbide fibres. J. Mater. Sci. 41, 823–839, http://dx.doi. org/10.1007/s10853-006-6566-z. Costello, J.A., Tressler, R.E., 1986. Oxidation kinetics of silicon carbide crystals and ceramics: I, in dry oxygen. J. Am. Ceram. Soc. 69, 674–681. Deal, B.E., Grove, A.S., 1965. General relationship for the thermal oxidation of silicon. J. Appl. Phys. 36, 3770. Doremus, R.H., 2001. Diffusion of Reactive Molecules in Solids and Melts. John Wiley & Sons. Doremus, R.H., 2002. Viscosity of silica. J. Appl. Phys. 92, 7619. Farmer, M.T., Leibowitz, L., Terrani, K.A., Robb, K.R., 2014. Scoping assessments of ATF impact on late-stage accident progression including molten core–concrete interaction. J. Nucl. Mater. 448, 534–540. Feinroth, H., 2012. Silicon carbide TRIPLEXTM fuel clad for accident resistance and durability. In: 1st ICMST Conference. Goto, T., 2006. High temperature passive oxidation mechanism of CVD SiC. Mater. Sci. Forum 522–523, 27–36. Goto, T., Homma, H., 2002. High-temperature active/passive oxidation and bubble formation of CVD SiC in O2 and CO2 atmospheres. J. Eur. Ceram. Soc. 22, 2749–2756. Goto, T., Homma, H., Hirai, T., 2002. Effect of oxygen partial pressure on the hightemperature oxidation of CVD SiC. Corros. Sci. 44, 359–370. Goto, T., Katsui, H., Oguma, M., 2014. High-temperature oxidation mechanism of chemical vapor deposited silicon carbide. In: International Symposium on HighTemperature Oxidation and Corrosion, Hakodate, Japan.

Grosse, M., 2010. Comparison of the high temperature steam oxidation kinetics of advanced cladding materials. Nucl. Technol. 170, 272–279. Hallstadius, L., Johnson, S., Lahoda, E., 2012. Cladding for high performance fuel. Prog. Nucl. Energy 57, 71–76. Irene, E.A., 1977. Silicon oxidation studies: the role of H2 O. J. Electrochem. Soc. 124, 1757. Jellinek, H.H.G., 1986. The thermal dissociation of water: the forgotten literature. J. Chem. Educ. 63, 1029. Jorgensen, P.J., Wadsworth, M.E., Cutler, I.B., 1961. Effects of water vapor on oxidation of silicon carbide. J. Am. Ceram. Soc. 44, 258–261. Katoh, Y., Nozawa, T., Snead, L.L., Ozawa, K., Tanigawa, H., 2011. Stability of SiC and its composites at high neutron fluence. J. Nucl. Mater. 417, 400–405. Katsui, H., Oguma, M., Goto, T., 2014. Carbon interlayer between CVD SiC and SiO2 in high-temperature passive oxidation. J. Am. Ceram. Soc. 97, 1633–1637. Kim, H.-E., Moorhead, A.J., 1990. Effects of active oxidation on the flexural strength of ␣-silicon carbide. J. Am. Ceram. Soc. 73, 1868–1872. Lee, Y., Kazimi, M., Yue, C., McKrell, T., 2013. Safety assessment of SiC cladding oxidation under loss-of-coolant accident conditions in light water reactors. Nucl. Technol. 183, 210–227. Mieskowski, D.M., Mitchell, T.E., Heuer, A.H., 1984. Bubble Formation in Oxide Scales on SiC., pp. 17–18. More, K.L., Tortorelli, P.F., Ferber, M.K., Keiser, J.R., 2000. Observations of accelerated silicon carbide recession by oxidation at high water-vapor pressures. J. Am. Ceram. Soc. 83, 211–213. Narushima, T., Goto, T., Iguchi, Y., Hirai, T., 1990. High-temperature oxidation of chemically vapor-deposited silicon carbide in wet oxygen at 1823 to 1923 K. J. Am. Ceram. Soc. 73, 3580–3584. Narushima, T., Goto, T., Yokoyama, Y., Takeuchi, M., Iguchi, Y., Hirai, T., 1994. Activeto-passive transition and bubble formation for high-temperature oxidation of chemically vapor-deposited silicon carbide in CO–CO2 atmosphere. J. Am. Ceram. Soc. 77, 1079–1082. National Research Council, 1970. High-Temperature Oxidation-Resistant Coatings: Coatings for Protection from Oxidation of Superalloys, Refractory Metals, and Graphite. National Academy of Sciences, Washington. Ogura, Y., Morimoto, T., 2002. Mass spectrometric study of oxidation of SiC in lowpressure oxygen. J. Electrochem. Soc. 149, J47–J52. Opila, E.J., 1995. Influence of alumina reaction tube impurities on the oxidation of chemically-vapor-deposited silicon carbide. J. Am. Ceram. Soc. 78, 1107–1110. Opila, E.J., 1999. Variation of the oxidation rate of silicon carbide with water-vapor pressure. J. Am. Ceram. Soc. 36, 625–636. Opila, E.J., 2003. Oxidation and volatilization of silica formers in water vapor. J. Am. Ceram. Soc. 86, 1238–1248. Opila, E.J., Hann, R.E., 1997. Paralinear oxidation of CVD SiC in water vapor. J. Am. Ceram. Soc. 80, 197–205. Opila, E.J., Smialek, J.L., 1999. SiC recession caused by SiO2 scale volatility under combustion conditions: II, thermodynamics and gaseous-diffusion model. J. Am. Ceram. Soc. 82, 1826–1834. Opila, Jacobson, N.S., 1995. SiO(g) formation from SiC in mixed oxidizing-reducing gases. Oxid. Met. 44, 527–544. Pongrácz, A., (Ph.D. thesis) 2009. SiC Nanocrystals on Si. Budabest University of Technology. Presser, V., Nickel, K.G., 2008. Silica on silicon carbide. Crit. Rev. Solid State Mater. Sci. 33, 1–99. Rodri´ıguez-Viejo, J., Sibieude, F., Clavaguera-Mora, M.T., Monty, C., 1993. 18O diffusion through amorphous SiO2 and cristobalite. Appl. Phys. Lett. 63, 1906. Schneider, B., Guette, A., Naslain, R., Cataldi, M., Costecalde, A., 1998. A theoretical and experimental approach to the active-to-passive transition in the oxidation of silicon carbide: experiments at high temperatures and low total pressures. J. Mater. Sci. 33, 535–547. 2014. Scientific Group Thermodata, www.sgte.org. Singh, D., Salem, J., Halbig, M., Mathur, S., 2012. Mechanical Properties and Performance of Engineering Ceramics and Composites VII. Wiley Online Library. Snead, L.L., Nozawa, T., Katoh, Y., Byun, T.-S., Kondo, S., Petti, D.A., 2007. Handbook of SiC properties for fuel performance modeling. J. Nucl. Mater. 371, 329–377. Steinbrück, M., Stegmaier, U., Ziegler, T., 2007. Prototypical experiments on air oxidation of Zircaloy-4 at high temperatures. In: Forschungszentrum Karlsruhe Report FZK. Stempien, J.D., 2011. Behavior of triplex silicon carbide fuel cladding designs tested under simulated PWR conditions. In: MIT MS Thesis. Tedmon, C.S., 1966. The effect of oxide volatilization on the oxidation kinetics of Cr and Fe–Cr alloys. J. Electrochem. Soc. 113, 766–768. Terrani, K.A., Pint, B.A., Parish, C.M., Silva, C.M., Snead, L.L., Katoh, Y., 2014. Silicon carbide oxidation in steam up to 2 MPa. J. Am. Ceram. Soc. 97, 2331–2352. Tortorelli, P.F., More, K.L., 2003. Effects of high water-vapor pressure on oxidation of silicon carbide at 1200 ◦ C. J. Am. Ceram. Soc. 86, 1249–1255. Urbanic, V.F., Heidrick, T.R., 1978. High-temperature oxidation of Zircaloy-2 and Zircaloy-4 in steam. J. Nucl. Mater. 75, 251–261. Wagstaff, F., 1969. Crystallization and melting kinetics of cristobalite. J. Am. Ceram. Soc. 49, 118–121.