Current understanding of radiation-induced degradation in light water reactor structural materials

Abstract

Current phenomenological knowledge and understanding of mechanisms are reviewed for radiation embrittlement of reactor pressure vessel low alloy steels and irradiation assisted stress corrosion cracking of core internals of stainless steels. Accumulated test data of irradiated materials in light water reactors and microscopic analyses by using state-of-the-art techniques such as a three-dimensional atom probe and electron backscatter diffraction have significantly increased knowledge about microstructural features. Characteristics of solute clusters and deformation microstructures and their contributions to macroscopic material property changes have been clarified to a large extent, which provide keys to understand in the degradation mechanisms. However, there are still fundamental research issues that merit study for long-term operation of reactors that requires reliable quantitative prediction of radiation-induced degradation of component materials in low-dose rate high-dose conditions.

Keywords:

• light water reactor
• neutron irradiation
• radiation embrittlement
• IASCC
• low-alloy steel
• stainless steels

1. Introduction

Structural components located near nuclear fuel assemblies in light water reactors (LWRs) are exposed to intense radiation fields. Neutron irradiation causes significant changes in material properties and in some cases results in degradation of structural integrity. Among various radiation-induced degradation phenomena, there are two major phenomena in LWRs. One is radiation embrittlement of reactor pressure vessels (RPVs) made of ferritic low alloy steels. This phenomenon appears as a decrease in fracture toughness and upper shelf energy and a shift of ductile brittle transition temperature to a higher temperature. The other is irradiation assisted stress corrosion cracking (IASCC) of reactor core structural components made of austenitic stainless steels (SSs). IASCC appears as an increase in cracking susceptibility and crack growth rate (CGR) in high temperature water. Since both degradation phenomena become more pronounced as the neutron fluence increases, these are now of increasing concern for the structural integrity of aged LWRs in long-term operations of 60 or 80 years.

There have been a large number of experimental and theoretical studies on radiation embrittlement and IASCC, and reviews have been already published [1–7]. However, to improve the reliability of structural integrity evaluation during long-term operation, continuous efforts are needed from both scientific and engineering aspects since experimental data and knowledge are still scarce at higher fluences. In radiation embrittlement, it is necessary to precisely predict changes in transition temperature and fracture toughness at high fluences and this requires improved understanding of mechanisms for a physical basis. Radiation embrittlement of RPVs is caused by microstructural changes in the material alone, while IASCC is a more complex phenomenon than radiation embrittlement since more processes relating to radiation, materials, water environment and stress–strain are simultaneously involved. Not only material damage processes induced by neutron irradiation, but also corrosion and water radiolysis processes induced by neutron and gamma radiation are involved in IASCC. Reliable predictions of crack initiation and growth inevitably require sound understanding of mechanisms.

Recent development of microanalysis techniques such as three-dimensional atom probe (3DAP) and electron backscatter diffraction (EBSD) techniques provides researchers with new knowledge of microstructural and microchemical characteristics under irradiation. Furthermore such microstructural and microchemical data are obtained from LWR-irradiated materials such as surveillance specimens of RPVs and removed materials of in-core components. These knowledge and data enable elucidation of the precise role of microscopic features on degradation mechanisms. In this paper, current phenomenological knowledge and understanding of mechanisms for radiation embrittlement and IASCC in pressurized water reactors (PWRs) and boiling water reactors (BWRs) are summarized and discussed. The major concern is focused on Mn–Mo–Ni steels for RPVs and type 304 and 316 SSs for core internals, with further emphasis being placed on microscopic material changes and their role in degradation processes.

2. Radiation environment

Figure 1 shows examples of the intensity distribution of neutrons and gamma rays and neutron energy spectrum within the RPV in a PWR [8]. Neutron flux on a structural component depends on its distance from the reactor core as well as the reactor type and power. The estimated maximum fluence or dose and typical temperature in the RPV and core internals of BWRs and PWRs for 40-year operation are shown in Figure 2. The dose reaches ∼100 dpa for PWR core internals, while it is ∼0.1 dpa for the PWR RPV. The neutron doses in BWRs are almost one or two orders lower than those in PWRs since BWRs have lower core power densities and larger distances between the core and components than PWRs.

Figure 1 Examples of (a) intensity distribution of neutrons and gamma rays and (b) neutron energy spectrum within the RPV in a PWR (figures drawn using radiation transport calculation data from [8])

Figure 2 Maximum fluence or dose and typical temperature for 40-year operation of a reactor pressure vessel and core internals

Neutron irradiation causes displacement damage and also the generation of He atoms through nuclear transmutation reactions of thermal neutrons mainly with B-10 and Ni-58. The He generation rate is estimated to be about 5–20 appmHe/dpa in core internals SSs. This rate is almost two orders higher than that in the core of fast breeder reactors (FBRs). Exposure to neutrons and gamma rays causes water radiolysis by decomposing water molecules in coolant water. The maximum dose rate absorbed in coolant water is estimated to be 10³–10⁴ Gy/s in the core region of LWRs.

3. Radiation embrittlement of pressure vessel steels

3.1. Radiation effects on embrittlement

The measure of radiation embrittlement is the decrease in fracture toughness, and in Charpy V-notch impact tests for surveillance programs, the shift of the ductile brittle transition temperature (hereafter transition temperature shift) at a specific absorbed energy. It is well known that the sensitivity to radiation embrittlement depends on material variables such as the chemical composition and microstructural variation with thermomechanical treatments, and irradiation variables such as temperature and neutron flux. Effects of these variables on the transition temperature or fracture toughness are briefly summarized in this section. Figure 3 shows schematic figures depicting embrittlement trends versus neutron fluence in BWRs and PWRs with variations of Cu and Ni contents and neutron flux.

Figure 3 Schematic embrittlement trends in (a) low-flux BWR and (b) high-flux PWR conditions

3.1.1. Influences of material variables

Material composition

Cu, Ni, P and Mn have been identified as solute elements that have a distinct influence on radiation embrittlement. The impurity Cu, typical contents of 0.05–0.3% in RPV steels, has a dominant effect on radiation embrittlement. The words ‘high-Cu’ and ‘low-Cu’ are often used in the literature, and in this paper both are used for Cu contents higher than 0.2% and lower than 0.07–0.08%, respectively. It is recognized that Cu enhances the rise and saturation of transition temperature shift (rise at a higher rate and saturation at a lower fluence for higher Cu contents) from model steel data with very high Cu content (up to 0.5%), while the Cu effects become insignificant for Cu contents lower than 0.07%. Recent surveillance data and material test reactor (MTR) data in commercial steels containing Cu at less than 0.2% indicate that Cu has a clear influence even at contents less than 0.07% [9,10]. The change in transition temperature shift per unit fluence seems to be relatively insensitive to Cu contents at high fluences [9–11].

Ni is an alloying element, typical contents of 0.5–1.5% in RPV steels, and is known to enhance hardening and embrittlement. The Ni effect is more significant for higher Cu steels, suggesting a combined effect with Cu [12,13]. The Ni effect seems to become less significant for lower Cu steels [14,15] since recent data on high-Ni commercial steel A508Cr4N (3.3%Ni) showed a similar transition temperature shift to that of low-Ni A508Gr2 steel (<1%Ni). Data of ion-irradiated model alloys showed that Ni enhanced hardening regardless of Cu content [16].

Mn is an alloying element, typical content of 0.5–1.5% in RPV steels, and is known from recent data to enhance radiation hardening and embrittlement [13,17–19]. A clear enhancement was observed in high-Cu model steels with increasing Mn from 0.01 to 1.63% [17]. In simple Fe–Mn alloys, significant enhancement of hardening by Mn addition up to 1.5% was observed after MTR irradiation [19].

P is an impurity element, typical contents below 0.03% in RPV steels, and is known to enhance radiation hardening and embrittlement in low-Cu steels [20,21]. Since P content generally varies with Cu content in commercial steels, the P effect becomes unclear for high-Cu commercial steels in which radiation embrittlement is dominated by Cu content. P is also known to enhance the sensitivity to intergranular fracture if P segregation becomes significant at grain boundaries (GBs). The intergranular fracture mode causes transition temperature shift without hardening. This effect is discussed in Section 3.2.3.

Microstructures

The sensitivity to embrittlement is known to depend on the pre-irradiation microstructure, which is determined by fabrication and thermomechanical processes, for example, by the plate, forge, weld and heat affected zone (HAZ). The microstructural variations include the amount and distribution of pre-existing dislocations, carbides, other precipitates, impurity segregation and so on. These microstructural variations result in differences in irradiation-induced microstructural changes and resultant mechanical property changes. For a similar chemical composition, the transition temperature shift is generally similar in plate and forged materials, and even in base metals and weld metals. However, for example, a relatively large shift was reported particularly in weld metals such as Linde 80. The exact reason of the large shift has not been clarified yet, although low upper shelf energy was considered as one of the possible reasons [22]. In the HAZ, the sensitivity to radiation embrittlement varies with the distance from the fusion line [23,24]. This generally reflects microstructural variation with the distance due to thermal history variations. It was reported that a coarse grain HAZ showed a higher transition temperature shift after MTR irradiation than others such as a fine grain HAZ [23]. The size distribution of fine precipitates such as Mo₂C and AlN was found to affect transition temperature shift in low-Cu A533B steels after MTR irradiation [25]. The transition temperature shift was smaller for steels containing finer and denser carbides [25]. The influence of the fine carbide distribution might be related to a recent observation under heavy ion irradiation that radiation hardening was larger in near-carbide regions compared to matrix regions without carbides [26]. Influences of the pre-irradiation microstructure and its mechanisms are not fully understood compared to effects of material composition.

3.1.2. Influences of irradiation variables

Irradiation temperature

Radiation embrittlement is generally lower for lower irradiation temperature. The influence of irradiation temperature on hardening or embrittlement is often expressed by the temperature term F_T using a linear function of temperature T: F_T = 1.869 – 4.57 × 10⁻³ T (K). This expression was derived from data of C–Mn steels in gas-cooled reactors over a wide temperature range [27]. Similar linear irradiation temperature dependence was also observed in weld metals and A533B steels irradiated in MTRs [25,28]. Temperature dependence was clearer for higher Cu content, while it seemed to be relatively small at temperatures lower than 300°C for high-Cu steels.

Neutron flux or dose rate

Influence of neutron flux or dose rate has been and is still a major concern in the prediction of radiation embrittlement. A number of studies on flux effects have been conducted mainly using MTR irradiation. Knowledge and understanding obtained until 1990s were discussed and summarized in 2001, the Workshop on Dose Rate Effects in Reactor Pressure Vessel Materials [29]. Additionally, there have been publications on kinetic modeling of microstructural evolution [30,31]. Understanding of flux effects (dose rate effects) on transition temperature shift based on these studies was as follows:

1.	Flux effect varies with Cu content in steels (higher or lower than 0.1%) and flux range (lower than 5 × 109 n/cm2/s, higher than 1011 n/cm2/s flux). In high-Cu steels, the flux effect also depends on whether the fluence is larger than the saturation fluence of embrittlement.
2.	In steels with higher Cu contents at fluences less than the saturation fluence, the flux effect appears both at lower and higher flux ranges. The saturation fluence is lower for lower flux in both lower and higher flux ranges. At fluences higher than the saturation fluence, no clear flux effect appears.
3.	In steels with lower Cu contents, no flux effect is observed at fluxes lower than 1012 n/cm2/s, while the flux effect may appear at higher fluxes depending on the contents of Ni and Mn.

Recently MTR irradiation experiments on archive materials of RPVs were conducted for direct comparison with surveillance data to provide a clear understanding of flux effects in both low- and high-flux ranges [32,33]. For the low-flux, low-fluence region relevant to BWRs, MTR irradiation of A302B steel (0.24%Cu) was conducted at a flux of 7 × 10¹¹ n/cm²/s and the data were compared with BWR surveillance data at 2 × 10⁹ n/cm²/s (vessel wall position) and 2 × 10¹⁰ n/cm²/s (accelerated position). The results clearly showed that lower flux irradiation enhanced radiation embrittlement (Figure 4a). For the high-flux, high-fluence region relevant to PWRs, PWR surveillance data at 1 × 10¹¹ n/cm²/s were compared with MTR irradiation data on two A533B steels (0.16 and 0.068%Cu) and one A508 steel (0.03%Cu) at a flux of 7 × 10¹² n/cm²/s. The transition temperature shift was higher for MTR irradiation in 0.16% and 0.068%Cu A533B steels, as shown in Figure 4(b) for 0.068%Cu steel, while the transition temperature shift was comparable in the low-Cu A508 steel. These data demonstrate that the general understanding of flux effects summarized in 2001 is valid in flux conditions relevant to RPVs in operating BWRs and PWRs. Such direct comparison of commercial surveillance materials at different fluxes is very effective for the confirmation of the current understanding.

Figure 4 Comparison of transition temperature shifts between surveillance data and MTR data: (a) high-Cu BWR surveillance material and (b) low-Cu PWR surveillance material. The data from references [32,33] are replotted and trends are shown

Thermal neutrons and gamma rays

Effects of thermal neutrons and gamma rays were discussed in the 1990s, in order to understand mechanisms of accelerated radiation embrittlement observed in surveillance data of the High Flux Isotope Reactor (HFIR) [34]. Gamma rays can cause displacement damage in materials through high-energy electrons and positrons that are produced through gamma–material interactions, Compton scattering, pair production and the photoelectric effect [35]. It was confirmed that electron irradiation with energies of several MeV causes hardening in low-alloy steels [36–38]. For thermal neutron effects, the contributions to displacement damage production and additional damage production through the nuclear reaction of boron were discussed. It was shown that thermal neutrons were almost two times more effective for causing radiation embrittlement than fast neutrons at the same dpa [39,40]. The contributions of thermal flux and gamma rays to radiation embrittlement of pressure vessel steels were assessed for the HFIR and LWRs considering the neutron and gamma-ray spectrum, core structure and water gap between the pressure vessel and core [34,41]. The enhanced embrittlement observed in the HFIR was attributed to additional damage induced by high-flux gamma rays [34]. On the other hand, the contributions of thermal neutrons and gamma rays to radiation embrittlement in LWRs were estimated to be negligible. The displacement damage induced by thermal neutrons and gamma rays was less than ∼5% in pressure vessels in BWRs and PWRs. In advanced BWRs (ABWRs), a large contribution (∼50%) from gamma rays was expected because of the larger water gap between the core and pressure vessel. However, the flux in ABWRs is very low and would cause negligible embrittlement [41].

Tensile stress

Stress during irradiation is one of the potential factors that can cause changes in radiation embrittlement. Stress effects are an issue related to the difference between surveillance irradiation without stress and pressure vessels under static tensile stress during operation. Data collected in the 1960s showed that the tensile stress had no effects or a suppressing effect on the transition temperature shift [42–44] For example, no effects for a stress of 75% yield strength were observed in A350 (0.08%Cu) and A302 (0.15%Cu) steels irradiated to (2–3) × 10¹⁹ n/cm² [42], while the transition temperature shift was smaller for a tensile stress of 20% yield strength in A302B steels irradiated to 5 × 10¹⁹ n/cm² [44]. In order to understand the mechanisms of stress effects, heavy ion irradiation experiments under stress were recently conducted on A533B steels [45,46]. The results showed that the tensile stress reduced radiation hardening and produced smaller solute clusters.

3.2. Microstructural evolution and embrittlement mechanisms

Understanding of radiation embrittlement mechanisms is directly connected with understanding microstructural features formed under irradiation and their role in hardening and the transition temperature shift. Application of advanced microanalysis techniques such as small-angle neutron scattering (SANS), 3DAP, positron annihilation (PA) and transmission electron microscopy (TEM) has provided significant improvement for mechanistic understanding [3]. Microstructural features having dominant roles in radiation embrittlement have been recognized as Cu precipitates and extended secondary defects such as microvoids in the 1980s [47]. Odette and Lucas [1] have categorized microstructural features causing hardening and embrittlement and noted Cu-rich precipitates (CRPs) and matrix damage (MD) as major features. The Mn–Ni precipitate (MNP) was proposed as the late blooming phase, which was thought to be thermodynamically formed after long-time irradiation of low-Cu, high-Ni steels. MD is divided into two types: stable MD (SMD) such as interstitial clusters and unstable MD (UMD) such as small vacancy clusters that may redissolve at irradiation temperatures [48], although the exact nature of MD has not been identified. Radiation-enhanced segregation at GBs and its influence on intergranular fracture were discussed in the 1990s [49]. Further improvement of analysis techniques such as local electrode atom probe and PA with coincidence Doppler broadening (PA-CDB) [50] and application of these techniques to surveillance specimens have brought detailed information on microstructural evolution.

At present, it is widely believed that there are three major microstructural features causing embrittlement: solute precipitates or clusters, MD and GB segregation. The first two features cause embrittlement through material hardening. The MD is a concept including all microstructural features other than CRPs. However, recent 3DAP analyses show that solute aggregates containing Fe, Mn, Ni, Si, Cu and P are commonly formed in commercial and model low-alloy steels regardless of the Cu content. The aggregates exhibit diffused images and have a wide range of Cu fractions, and are now often called ‘clusters’, rather than ‘precipitates’, in the recent literature. In this paper, thus, ‘solute cluster’ is used as a general term. ‘Cu-rich cluster’ is used for clusters with high Cu fraction formed in high-Cu steels and ‘Mn–Ni–Si cluster’ for clusters containing Mn, Ni and Si atoms with very low Cu fraction formed in low-Cu steels. Mn–Ni–Si clusters are not included as MD described in this paper, although their clusters with no or very low Cu fractions are sometimes treated as MD in the literature.

3.2.1. Solute clusters

Recent observations in surveillance materials

Recent 3DAP observations of surveillance materials have provided key data for understanding microstructural evolution in actual vessel materials under operating conditions. In surveillance specimens (16MND5 steels with 0.09 and 0.044%Cu) irradiated in French PWRs at 2 × 10¹¹ n/cm²/s, fine clusters (3–4 nm in diameter) were detected and the cluster density increased with fluence up to 1.7 × 10²⁰ n/cm² [51,52]. The averaged composition in the 0.044%Cu steel was Fe–6Mn–10Ni–8Si in at% and did not change with fluence up to 7.6 × 10¹⁹ n/cm². The clusters were uniformly formed in the material although some clusters were present along line dislocations. In high-Cu weld surveillance specimens from Belgian PWRs, Doel-1 (0.13%Cu, 5.9 × 10¹⁹ n/cm²) and Doel-2 (0.30%Cu, 5.1 × 10¹⁹ n/cm²), the formation of Cu-rich clusters was clearly identified [53]. The composition of the clusters, which varied with the distance from the cluster center, was found to be 40Cu–3Mn–2Ni–1Si–1P in at% at the center of the clusters. The Cu content in the bulk rapidly decreased and saturated at less than 0.1wt% in both steels at 1 × 10¹⁹ n/cm². The number density of the clusters decreased, while the diameter increased with fluence. These data indicated that the Cu-rich clusters rapidly nucleated at low fluences and were coarsened at high fluences. These behaviors were well correlated with the results of PA-CDB measurements. Positron lifetime measurements showed that vacancy clusters with—three to four vacancies appeared and accumulated with fluence. In VVER surveillance specimens, formation of Mn–Ni–Si clusters was found. Mn–Ni–Si clusters were formed in high-Ni, low-Cu steels (0.03Cu–1.89Ni weld and 0.04Cu–1.0Ni plate) [54], while Mn–Si clusters were found in low-Ni steels (0.06Cu–0.07Ni) [55]. These findings indicated that the clusters without Cu were commonly formed.

Systematic analyses were conducted on surveillance specimens from Japanese PWRs [0.03–0.16%Cu, 1 × 10¹¹ n/cm²/s, (3–6) × 10¹⁹ n/cm²] and BWRs (0.24%Cu, 1 × 10⁹ n/cm²/s, 1 × 10¹⁸ n/cm²) [32,33,56–59]. The clusters containing Fe, Cu, Mn, Ni and Si were commonly formed. The density, volume fraction and composition depended on the Cu content, while the cluster diameter was not sensitive to it. In PWR surveillance specimens [59], the number density and averaged atomic composition of the clusters were 23 × 10²² /m³ and 6Cu–10Ni–6Mn–7Si for 0.12%Cu steels, 12 × 10²² /m³ and 2Cu–11Ni–5Mn–9Si for 0.07%Cu steels, and 2 × 10²² /m³ and 0Cu–12Ni–6Mn–12Si for 0.04%Cu steels, respectively. In this range of Cu contents (0.03–0.12%Cu), Cu atoms were not dominant in the clusters, while the Cu content still had a strong effect on cluster density. In the clusters, the Si fraction increased as the Cu fraction decreased, while the fractions of Ni and Mn remained unchanged. It was found that the clusters in PWR surveillance specimens were larger and less dense than those for MTR irradiation regardless of the Cu content [33,56].

In high-Cu BWR surveillance specimens [32,58], Cu-rich clusters with an average composition of 11Cu–9Ni–7Mn–3Si in at% were formed at a density of 43 × 10²² /m³. This density was higher than densities in a higher flux surveillance specimen (1 × 10¹⁰ n/cm²/s) and an MTR-irradiated specimen (7 × 10¹¹ n/cm²/s), indicating a clear enhancement effect of flux on cluster formation. The enhancement of Cu-rich cluster formation in low-flux surveillance irradiation compared to high-flux MTR irradiation was also reported in other high-Cu BWR surveillance specimens (0.22%Cu weld) [60] and high-Cu gas-cooled reactor surveillance specimens (0.14–0.19%Cu C–Mn steels, 240°C) [61]. In the latter case, at the same hardening level, Cu-rich clusters were dominant in the surveillance specimens at 4.2 × 10⁸ n/cm²/s, while no clusters were found in MTR-irradiated specimens at 3.6 × 10¹² n/cm²/s.

Characteristics of solute clusters

In addition to the surveillance specimen data mentioned above, a large sample of data from 3DAP, SANS and PA measurements have been reported on cluster formation in commercial steels and model alloys irradiated in MTRs and with charged particles [62–87]. Summaries of recent observations of solute cluster formation in low-alloy steels are given below.

Figure 5 Examples of 3DAP atom maps showing (a) Cu-rich clusters and (b) Mn–Ni–Si clusters in A533B steels irradiated with heavy ions to 1 dpa at 290ºC (courtesy of Katsuhiko Fujii of the Institute of Nuclear Safety System, Inc.). It should be noted that Mn–Ni–Si clusters are also found in (a)

Figure 6 CDB spectra in 0.12%Cu A533B steels irradiated in a MTR at 290°C [37]

Figure 7 3DAP data of density and diameter for solute clusters in MTR-irradiated A533B steels with various Cu and Ni contents [77]. Reprinted, with permission, from the Journal of ASTM International, Volume 6, Issue 7, copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428

1.	The solute clusters are commonly formed in low-alloy steels regardless of alloy compositions. Figure 5 shows the examples of 3DAP atom maps indicating clusters formed by heavy ion irradiation in high-Cu and low-Cu steels [87]. The clusters identified by 3DAP generally have neither a clear interface with the matrix nor a specific lattice structure. The clusters are composed of Fe, Mn, Ni, Si, Cu and occasionally P atoms and the composition of the clusters depends on material composition. Although the composition of the clusters from 3DAP analysis largely depends on the cluster definition method [76,88], data analysis based on a recursive search algorithm shows that Fe is the major element and that the Fe atomic fraction often exceeds 50%. The fraction of Cu depends on the Cu content and is typically less than 20% in commercial steels. The typical fractions of Ni, Mn and Si in the clusters formed in commercial steels are around 10%, 10% and 5%, respectively.
2.	Cu-rich clusters (Cu-enriched clusters in some references) are formed in Cu-containing steels and alloys. The Cu-rich clusters also contain Mn, Ni, Si and P, and are composed of the core of Cu atoms and the surrounding shell region enriched in Mn, Ni and Si. The formation of Cu-rich clusters, measured as the number density of the cluster, is enhanced by Cu content. The fraction of Cu atoms in the clusters increases with increasing Cu content. It is recognized that the formation of Cu-rich clusters is governed not by Cu content, but by solute Cu content. The Cu content is decreased with dose, due to Cu-rich cluster formation, and tends to saturate at 0.05–0.15%. The formation of Cu-rich clusters was detected at fluences of the order of 1017 n/cm2 by using the PA-CDB technique. Figure 6 shows an example of CDB spectra in 0.12%Cu A533B steel irradiated in a MTR at 290°C, showing the rise of the Cu peak at 1 mdpa (∼5 × 1017 n/cm2) [37]. The coarsening of Cu-rich clusters (increase in size and decrease in density) is known to occur in high-Cu model alloys, and was also recently reported in surveillance specimens (0.3%Cu) at high fluences [53]. However, detailed conditions such as dose and dose rate for coarsening have not been given.
3.	Mn–Ni–Si clusters without Cu atoms or with a low Cu fraction are now recognized as common microstructural features formed under irradiation since they have been widely observed in steels and alloys with very low Cu and also in Cu-free model alloys. The clusters are composed of Fe, Mn, Ni and Si atoms in commercial Mn–Mo–Ni low-alloy steels. Mn–Ni and Mn–Si clusters are also reported in a Mn–Ni model steel and a very low Ni steel, respectively [53]. The formation of Mn–Ni–Si clusters was detected at fluences of the order of 1019 n/cm2 in low-Cu commercial steels by using 3DAP. Small Mn–Ni–Si clusters coexist with Cu-rich clusters in high-Cu commercial steels, and the relative density of the Mn–Ni–Si clusters to Cu-rich clusters increases with dose under high dose ion irradiation in high-Cu commercial steels [87].
4.	The clusters observed in commercial steels are in the ranges of 2–4 nm in average diameter and of 1022–1023 /m3 in number density. The cluster size and density increase with increasing dose in most cases. Figure 7 shows the 3DAP data of the density and size of solute clusters formed in A533B steels with various Cu and Ni contents irradiated in a MTR [77]. The Cu content has weak influence on the size but strong influence on the density. The density and volume fraction are almost linearly dependent on the Cu content in a wide range of 0.02–0.25%. Obtained 3DAP and SANS data indicate Ni enhances the formation of Cu-rich clusters, higher in density and smaller in size. This effect is more significant for higher Cu materials. Some SANS data suggest that P and Mn enhance ‘precipitation’ in high-Cu steels [62,82], while more recent 3DAP data show no such enhancement of cluster formation [77].
5.	The clusters are homogeneously formed in the matrix and also heterogeneously formed along dislocations and GBs. The latter suggests that the cluster formation is related to cascade formation and point defect flow near sinks. In high-P steels, which often contain high Cu content, the clusters often contain P atoms. Since P segregates on dislocations and GBs, the cluster on such areas contains both Cu and P.
6.	Recent 3DAP studies on surveillance specimens comparing MTR irradiation provide clear knowledge on dose rate (flux) effects on the cluster formation. The Cu-rich clusters have a higher number density with a larger size and a larger volume fraction at lower fluxes in high-Cu steels under irradiation conditions relevant to BWR surveillance specimens [32,60]. On the other hand, the clusters have a lower number density with a larger size at lower fluxes in medium- to low-Cu steels (0.03–0.15%Cu) under irradiation conditions relevant to PWR surveillance specimens [33]. Previous knowledge, mostly from MTR irradiation, is that Cu-rich clusters formed in high-Cu steels and model alloys are larger and less numerous with higher volume fraction for lower flux [64]. Recent data seem to be consistent with previous knowledge in terms of volume fraction, while the formation of Cu-rich clusters is enhanced at very low fluxes (108–109 n/cm2/s) and low fluences (<1018 n/cm2). At high fluxes (1011–1013 n/cm2/s) and high fluences (>1019n/cm2), the formation of both Cu-rich clusters and Mn–Ni–Si clusters is enhanced for higher flux irradiation. Such enhancement of solute cluster formation at higher fluxes is not consistent with the previous knowledge that Cu-rich cluster formation is enhanced at lower fluxes. This suggests that there may be different mechanisms in solute cluster formation in the high-flux region.
7.	Data on effect of irradiation temperature on cluster formation are rare. However, a compilation of various data shows that the cluster size is insensitive to an irradiation temperature less than 300°C, while the cluster size rapidly increases with temperature above 300°C [89].

Contribution to mechanical property changes

The solute clusters act as obstacles for gliding dislocations and cause hardening (increase in yield strength Δσ _y ). It is well known that the transition temperature shift (ΔT) is proportional to the increase in yield strength (ΔT = (0.5–0.7)Δσ _y ) [1]. For Cu-rich clusters, based on knowledge from Cu precipitates in Fe–Cu model alloys that have a lower shear modulus than the matrix, the Russell–Brown model is used to correlate Δσ _y with cluster size and density [90]. The ratio of shear modulus values (G1/G2) between the cluster (G1) and matrix (G2) is estimated to be 0.6–0.8 in Cu-rich clusters which can explain the hardening in medium- and high-Cu commercial steels [56,63,78]. For Mn–Ni–Si clusters, the suitable hardening model is unclear.

On the other hand, it is known that the cluster volume fraction has a good correlation with radiation hardening and transition temperature shift. The root of the volume fraction determined by 3DAP analysis is proportional to the transition temperature shift and the increase in hardness or yield strength. This relationship is widely found in surveillance data [59,69,71], MTR irradiation data [77,79] and also various changed particle irradiation data [37], and is irrespective of the Cu content and irradiation conditions. Figure 8 shows an example of the linear relationship between the cluster volume fraction and transition temperature shift [59]. This relationship supports the belief that the cluster formation is the dominant factor in hardening and embrittlement. It also suggests that the contribution of the clusters to hardening is insensitive to the composition of the clusters, namely that Cu-rich clusters and Mn–Ni–Si clusters have almost the same hardening efficiency. However, some difference in relationship is found between surveillance data and MTR data, suggesting that there are some differences in cluster nature and/or there is a contribution from microstructural features other than clusters [33]. It was suggested that the Mn–Ni phase in irradiated low-Cu, high-Ni model alloys might have different hardening efficiencies from those of Cu-rich clusters [2]. Influence of the cluster composition on the hardening efficiency is still an issue to be investigated and it is also related to experimental techniques for quantifying diffuse clusters in 3DAP measurements.

Figure 8 Relationship between cluster volume fraction (V _f ^1/2) and transition temperature shift (ΔRT _NDT) in surveillance specimens [59]. Reprinted, with permission, from the Journal of ASTM International, Volume 7, Issue 3, copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428

Formation processes of solute clusters

Recent 3DAP observations on solute clusters suggest that the distinction between Cu-rich clusters and Mn–Ni–Si clusters might be expediential and that it might be reasonable to consider them as the same type of clusters. This is mainly because the clusters have similar ranges of size and number density in spite of wide variations in material compositions and irradiation conditions, and they have compositions that change continuously with chemical composition. The cluster number density and composition variations can be attributed to the effects of the Cu content. However, this does not mean that the clusters are formed through one common mechanism. Since the solubility of Cu in Fe is very low, it is widely accepted that Cu atoms form clusters or precipitates through radiation-enhanced vacancy diffusion. The aggregation mechanism of Mn, Ni and Si atoms with Cu atoms in Cu-containing steels has been proposed as Cu-catalyzed precipitation of Mn and Ni, based on thermodynamical considerations [48]. The clustering process was well simulated by kinetic Monte Carlo calculations based on vacancy diffusion [91,92]. A similar calculation was also conducted considering both vacancy and interstitial diffusion mechanisms [93]. These calculations gave dilute cluster compositions and cluster structures of the Cu core and Mn–Ni–Si shell which were very similar to the actual composition and structure of the clusters in commercial steels observed by using 3DAP.

Influences of the dose rate and temperature on Cu-rich cluster formation were discussed through radiation-enhanced diffusion and diffusion controlled growth [1,48]. This consideration predicts that the formation of Cu-rich clusters in high-Cu steels is suppressed at higher flux, while the flux effects disappear at fluxes lower than 5 × 10¹¹ n/cm²/s, and it is also expected that thermal diffusion becomes important at very low fluxes. The dose rate effect on vacancy-driven clustering in a wide range of dose rates was reproduced by using the total number of vacancy jumps calculated from kinetic Monte Carlo calculations considering both radiation-induced and thermal vacancies, as clearly shown in Figure 9 [94].

Figure 9 Effect of dose rate on the number of vacancy jumps from kinetic Monte Carlo calculations [94]

For Mn–Ni–Si clusters in very low Cu or no Cu steels, an aggregation mechanism other than the vacancy-driven thermodynamical process is important. In 3DAP maps, Ni, Mn and Si often segregate along pre-existing line dislocations and GBs in commercial steels regardless of the Cu content [52,58,75,76]. It was reported that the composition of Mn–Ni–Si clusters near dislocations was very similar to that of a segregated region along a GB in a low-Cu surveillance specimen (0.044%Cu) [52]. These observations suggest that Ni, Mn and Si segregate to sinks during irradiation and that a non-equilibrium segregation process can be a potential mechanism for aggregation of Mn, Ni and Si atoms to form Mn–Ni–Si clusters. Migration mechanisms of these solute atoms have been discussed [95–97]. Si and Ni atoms are supposed to have strong binding with self-interstitials and migrate as mixed dumbbells. Dislocation loops formed due to irradiation are expected to be segregation sites of Mn, Ni and Si atoms and then nucleation sites of Mn–Ni–Si clusters. If this process is dominant in low-Cu steels, the observed enhancement of Mn–Ni–Si cluster formation at higher fluxes might be explained by this process since the formation of dislocation loops (self-interstitial clusters) is expected to be enhanced for higher fluxes [94]. This process of Mn–Ni–Si formation is expected to occur after the Cu content in the bulk becomes very low due to Cu-rich cluster formation in high-Cu steels. Such an example is seen in Figure 5(a) showing both Cu-rich clusters and Mn–Ni–Si clusters in 0.16%Cu steels after high-dose ion irradiation. However, neither data nor an understanding on the solute behavior is sufficient in low-flux and high-fluence conditions.

3.2.2. Matrix damage

Dislocation loops

MD is microstructural feature including point defect cluster, clusters with point defects and solute atoms and fine precipitates such as carbides and nitrides [1] Full understanding of the exact nature of MD has not been obtained yet. However, recent microstructural analyses have provided new knowledge mainly from TEM observations and PA measurements. TEM observations with improved specimen preparation techniques using focused ion beam techniques confirmed that dislocation loops are commonly formed in irradiated commercial steels [58,59,77,98–101]. Figure 10 shows example TEM images indicating interstitial dislocation loops in A533B steels irradiated with heavy ions [98]. The Burgers vector of the loops in A533B steels was reported to be b = a <100> and the loops were thought to be of interstitial type since they did not dissociate by annealing at 450°C [98]. The dislocation loops tend to form preferentially near pre-existing line dislocations. In the literature, the diameter of loops is in the range of 2–10 nm. The number density is of the order of 10²¹–10²²/m³, which is almost one order lower than the number density of clusters. However, since the minimum size of the dislocation loops is almost comparable to the resolution limit of TEM (∼1 nm), the reported size and density should be recognized as the values for ‘visible’ dislocation loops. As the dose increases, the number density shows a clear increase, while the size shows a very slow increase [100].

Figure 10 TEM images showing interstitial dislocation loops in A533B steels irradiated with heavy ions to 1 dpa at 290°C: weak-beam images with diffraction vectors (a) g = 011 and (b) g = 200 close to the [011] pole [98]

The influence of material composition on dislocation loop formation has been examined in simple model alloys, but data in commercial steels have not yet been reported in the literature. Compared to pure Fe, dislocation loop formation was reported to be enhanced by Ni and Mn addition [16,21,102], but it was reported to be suppressed by Mn, Ni and Cu [103]. A few references have studied influences of irradiation conditions such as flux and temperature on dislocation loop formation in low-alloy steels, and enhanced formation for a higher flux and a lower temperature may occur. Interstitial dislocation loops are thought to be one of the microstructural features of SMD. It was pointed out in several papers that the contribution of visible dislocation loops to hardening estimated using an Orowan-type hardening model was smaller than that of solute clusters [77,100,104]. However, since the estimated contribution of dislocation loops to hardening or transition temperature shift exceeded 10% of the total value, dislocation loops might have non-negligible contribution to radiation embrittlement.

Vacancy components

Vacancy components such as vacancies and microvoids can be examined by using PA techniques. Several studies using positron lifetime measurements showed that microvoids containing more than 10 vacancies were formed in binary or ternary model alloys [105,106]. A recent study showed that only vacancies and divacancies existed in irradiated Fe–Mn–Ni and Fe–Mn–Ni–Cu model alloys, while vacancy clusters containing more than 10 vacancies were formed in Fe–Cu alloys [104]. Vacancy clusters or microvoids containing more than one vacancy were not reported in commercial steels [37,77,106–108]. Recent data on weld surveillance specimens (0.13, 0.3%Cu) showed that vacancy clusters containing—three to four vacancies were formed and that their density increased gently with increasing dose, as shown in Figure 11 [53]. These vacancy clusters are formed after initial Cu-rich cluster formation and their contribution to hardening might be small. From experiments on binary model alloys, it is known that the formation of vacancy clusters is enhanced by P addition [109]. However, the influence of material composition on vacancy cluster formation is unclear in commercial steels.

Figure 11 Positron lifetime data in weld surveillance specimens (0.13%Cu for Doel-1 and 0.30%Cu for Doel-2) [53]

The combination of post-irradiation annealing (PIA) and PA measurements (change in the shape parameter that is proportional to the amount of positron-trapping sites) can provide indirect information about the contribution of vacancy components on hardening. Data in several references [37,57,58,108,110] showed that vacancy components with short lifetimes have a negligible contribution to hardening since the annealing temperature where the recovery of hardening occurs is not coincident with that of vacancy components. Comparisons of PIA responses between surveillance irradiation and higher flux MTR irradiation showed that vacancy components in the MTR irradiation started to recover at lower temperatures with a larger amount of recovery than in surveillance irradiation in 0.12%Cu and 0.06%Cu A533B steels [57,110,111]. This suggests that the vacancy components formed at higher fluxes are more thermally unstable and that these might correspond to UMD.

Interrelationship between matrix damage and solute clusters

As mentioned above, knowledge on dislocation loops has been significantly expanded. Dislocation loops, mostly the interstitial type, are formed as a stable microstructural feature in commercial steels regardless of the material composition and irradiation conditions. The dislocation loops have a definite but smaller contribution to radiation hardening than solute clusters. The detailed nature of vacancy-type MD is still unknown in commercial steels. From current data by PA measurements, it is reasonable to conclude that vacancy clusters containing more than one or two vacancies are not formed, which make a definite contribution to hardening. However, it is necessary to carefully consider interactions between point defects and solute atoms. It is evident that Cu-rich clusters in high-Cu model alloys are formed in a manner that Cu atoms cover the inner surface of microvoids [84,105]. Some fraction of vacancies must be assumed for explaining SANS data (mass density of the clusters) of the clusters formed in commercial steels [3]. These facts suggest that vacancies are contained in the clusters and involved in cluster formation processes.

Dislocation loops can be sinks for point defects. Thus, dislocation loops are expected to be decorated with segregated solute atoms such as Ni, Si and Mn and therefore can act as nucleation sites of clusters. Enhanced formation of both dislocation loops and solute clusters was observed along pre-existing line dislocations. Unfortunately, no analytical methods are available which can directly identify the state and distribution of point defects and solute atoms at the same time. Detailed correspondence of solute clusters with dislocation loops or vacancy clusters is still unknown. Vacancy clusters, if they exist as UMD, might act as point defect sinks and change the concentration of point defects, resulting in a change in the diffusion coefficient of solute atoms and nucleation of solute clusters. Furthermore, the interrelationship between MD and solute clusters might relate with hardening processes. In the case that enhanced formation of dislocation loops and solute clusters near pre-existing line dislocations is dominant, source hardening might be a dominant hardening process. Continuing research efforts to identify MD and its relation to solute cluster evolution and hardening are essential to understand embrittlement mechanisms.

3.2.3. Grain boundary segregation

GB embrittlement due to P segregation was initially examined for understating effects of long-term thermal ageing on embrittlement of weld HAZ [112]. Since intergranular fracture was found in an irradiated high-P C–Mn weld [112], the influence of irradiation on P segregation and GB embrittlement has been studied. Knowledge on the segregation behavior and its influence on embrittlement, obtained until the early 2000s, has been summarized in the literature [49,114], and new studies have increased understanding of the critical condition for embrittlement due to P segregation. Major points are summarized below.

Figure 12 Irradiation hardening (Δσ_y ) versus transition temperature shift (ΔT _41J) in MTR-irradiated A533B steels with various levels of P segregation at GBs (ΔC_P ^gb) [121]. Reprinted, with permission, from the Journal of ASTM International, Volume 6, Issue 7, copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428

1.	P segregation occurs both during post-weld heat treatment of the fabrication process and during service at operation temperatures and under neutron irradiation. P segregation at GBs during thermal ageing is treated as thermal equilibrium segregation. Model estimation shows that the P concentration at GBs remains almost unchanged due to thermal aging alone for 60 years at around 300°C in commercial low-alloy steels [49,115].
2.	Neutron irradiation significantly enhances P segregation at GBs. This was confirmed by a sample of data in irradiated commercial steels [113,116–121] and model alloys [122,123]. The enhancement is thought to be caused by accelerating thermal segregation due to radiation-enhanced diffusion (vacancy dragging) and also radiation-induced diffusion (mobile P-interstitial dumbbells) [114,124–126]. P segregation at GBs becomes higher for higher P content. In commercial Mn–Mo–Ni steels, P segregation seems to saturate at high fluences and the saturation level seems to be about 0.2 monolayer coverage at 1 × 1020 n/cm2 in steels containing P less than <0.02%. P segregation in C–Mn steels and VVER-type vessel steels is found to be higher than that in Mn–Mo–Ni steels [114] although the reason is unclear.
3.	Segregation of other elements, C, Ni, Mn, Si, Mo and Cu, is also known to exist at GBs both before and after irradiation in commercial steels [57,121]. Irradiation enhances segregation of Ni, Mn and Si at GBs [52,57,121]. C desegregation during irradiation has been discussed since P and C are competitive segregants in steels and C enhances the GB cohesion [57,121–123]. The increase in P coverage is expected to cause a decrease in C coverage. However, in commercial A533B steels which typically contains less than 0.03% P, changes in C concentration at GBs due to irradiation are so small that the trend is not clear [121]. The influence of segregation of other elements on the irradiated mechanical behavior has not been well examined.
4.	It is well established that high-level P segregation at GBs enhances intergranular fracture and transition temperature shift in thermally aged steels. In such cases, the transition temperature shift is proportional to the P segregation level [112,116]. Recent studies carefully examined the correlation between transition temperature shift, radiation-induced hardening and P segregation in A533B-type steels containing a wide range of P contents (0.008–0.57%P), and found that P segregation affects the transition temperature shift only when the P concentration at GBs exceeds approximately 0.3 monolayer coverage [120,121]. Figure 12 shows irradiation hardening (Δσy ) versus transition temperature shift (ΔT 41J) in MTR-irradiated A533B steels with various levels of P segregation at GBs (ΔCP gb) [121]. Excess transition temperature shift was observed only in the specimen with very high P segregation (ΔCP gb = 0.32).
5.	In general, the influence of P segregation on embrittlement does not seem to be significant even for long-term operation for Mn–Mo–Ni steels containing less than 0.03% P; it might be important in high-P steels irradiated to very high fluences and/or in steels subjected to annealing.

3.3. Prediction and modeling of embrittlement

3.3.1. Transition temperature shift

Evaluation of pressure vessel structural integrity for long-term operation requires a reliable prediction of radiation embrittlement. Predictive models or correlation models for transition temperature shift (reference temperature shift) from Charpy V-notch impact testing have been recently revised in some countries. This was to improve predictability at high fluences based on the expanded understanding of embrittlement mechanisms.

A new embrittlement correlation method was developed in Japan and adopted in the 2007 revision of the code JEAC4201 [127]. This method is a kind of mechanism-guided correlation, which models microstructural changes by a series of rate equations and correlates the predicted microstructural changes with mechanical properties based on the knowledge from extensive microstructural analyses of surveillance materials [32,58,59]. The model considers three microstructural components: radiation-enhanced solute clusters (related to Cu clustering), radiation-induced solute clusters (related to heterogeneous solute clustering) and MD (related to point defect clustering). The model has Cu and Ni contents, temperature and neutron flux as parameters, and other constants that are optimized for the Japanese surveillance materials database. The model assumes that the radiation-induced solute clusters are formed from MD, and that MD is formed in proportion to the neutron flux. Based on the proportional relationship between the square root of the solute cluster volume fraction and the transition temperature shift, the reference temperature shift is calculated by the square root sum squared of the solute cluster contribution and MD contribution. Detailed descriptions are given in [58,59]. The predicted shifts using this method are in good agreement with measured data from both low-flux BWR surveillance data and high-fluence PWR surveillance data [10,32,128]. This correlation model will be modified based on the accumulation of high-fluence data for further improving its reliability at high fluences.

Mechanism-guided prediction equations have been proposed in the USA [129–131]: they are known as the EWO model [129], ASTM E900 model [130] and EONY model [131]. The last model was adopted in the evaluation rule of pressurized thermal shock (PTS) in 2010 [132]. These models consider two microstructural components that contribute to hardening: stable MD and CRP. The transition temperature shift is expressed by a simple sum of MD and CRP contributions which are expressed as the product of a fluence term and a chemical term. In the EONY model, the MD contribution is a function of temperature, P and Mn contents and the square root of fluence. The CRP contribution is a function of the Cu, Ni and P contents and fluence. The effect of flux ϕ is incorporated as the effective fluence, which is modified by the factor (4.39 × 10¹⁰/ϕ)^0.259 at fluxes lower than 4.39 × 10¹⁰ n/cm²/s. The model equations are optimized for the surveillance materials database in the USA, and different constants in the equations are prepared for plate, forged and weld materials.

The mechanism-guided trend curve has also been discussed for Belgian reactors considering MD and CRPs [133]. In this curve, the Cu content for CRPs, and Ni content and temperature for MD are included. Modification of the prediction equations or trend curves is being discussed in other European countries based on an empirical formulation using fluence and chemical factors [134–137].

Current prediction methods are optimized to the surveillance materials database used in each country. On the other hand, long-term operation requires a reliable embrittlement prediction at fluences much higher than the fluence ranges of current surveillance databases. Thus, there have been discussions on extending the materials’ embrittlement databases, using MTR data and constructing an international database [137]. Prediction using MTR data requires careful considerations on high-flux effects. It was proposed that the contribution of UMD, which is assumed to be vacancy-type clusters and formed more densely at higher fluxes, should be added to the contributions to MD and Cu-rich clusters [2,138].

3.3.2. Attenuation through the pressure vessel wall

For the evaluation of pressure vessel integrity, the fracture toughness value at a given depth is required based on an attenuation trend of mechanical properties. The fracture toughness at a given depth is usually estimated using calculated attenuation trends of neutron fluence or dpa and embrittlement prediction methods. Reliability of such estimations has been examined using irradiation defect production calculations and mechanical property tests on decommissioned pressure vessels [139]. For example, molecular dynamics calculations of freely migrating point defects with a neutron transport calculation through pressure vessel steels revealed that the dpa attenuation trend was suitable to predict the attenuation of damage production [140]. Data from the decommissioned French PWR Chooz A pressure vessel demonstrated that the change in measured transition temperature shift was in agreement with the trend of the predicted shift assuming a dependence of (fluence)^0.48 [141]. The attenuation issue has been continued to be examined in the 2000s. The attenuation of fracture toughness was examined in weld trepan of the decommissioned WWER-440, and the results indicated nonsystematic changes and large data scatter [142]. A MTR irradiation program organized by IAEA has been conducted. In this program, 180 mm stacks of 10mm thick impact and compact tension specimens for commercial steels (0.14%Cu and 0.06%Cu) and high-Cu weld (0.3%Cu) were irradiated while being monitored by many dosimeters [143–145]. The results showed that the trends of the transition temperature shift and the master curve T ₀ shift are generally in agreement with the dpa attenuation prediction although the degree of agreement varied with material.

3.3.3. Fracture toughness

In fracture mechanics evaluation of pressure vessel integrity using transition temperature shift determined by Charpy V-notch impact tests, the basic assumption is that the transition temperature shift is equivalent to the reference temperature shift determined in fracture toughness tests. Application of the master curve approach to integrity evaluation requires precise assessment on irradiated fracture toughness, which includes the shape of the fracture toughness curve (temperature dependence of fracture toughness) and the equivalence of the master curve reference temperature to the reference transition temperature in a wide range of fluence and material conditions. A detailed review of these macroscopic mechanical issues is beyond the purpose of this paper. Thus, only a brief description of some critical technical subjects for continued study is made here.

Figure 13 Plots of master curve reference temperature shift ΔT ₀ versus Charpy transition temperature shift ΔT _41J for A533B steel data sets containing high-shift data. The dotted line shows the correlation given by Sokolov and Nanstad [148]

1.	Equivalence of the master curve reference temperature shift (ΔT 0) with transition temperature shift defined at the absorbed energy of 41 J (ΔT 41J) has been extensively studied especially for data about high embrittlement such as obtained with high-fluence, high-Cu and high-Ni steels. The regression analyses of datasets showed a trend that ΔT 0 is slightly higher than ΔT 41J: ΔT 0 = (1.03–1.08)·ΔT 41J [146, 147]. Several high ΔT 0 data (100–200°C) are in good agreement with corresponding ΔT 41J data [146–148], while some recent ΔT 0 data showed about 40°C higher values [9]. Figure 13 shows an example graph of the correlation between ΔT 0 and ΔT 41J [9]. The equivalence of the conventional reference temperature (RTNDT) and master curve reference temperature (RTT0) in ASME code cases has been examined in various irradiated commercial steels such as surveillance materials, MTR-irradiated archive materials and decommissioned pressure vessel steels [149–153]. It was confirmed that the master curve reference temperature generally gives conservative evaluation of fracture toughness. However, some issues should be noted. One is the change in fracture mode from cleavage to intergranular, which might have a significant influence in some cases. In an A302B HAZ material, intergranular fracture induced by thermal aging (460°C for 168 h) changed the shape of the temperature versus fracture toughness curve and gave almost 50°C higher T 41J than T 0 [154].
2.	Influence of specimen size has long been one of the essential issues for determining fracture toughness. For the master curve application, IAEA coordinated programs extensively examined influences of specimen type and size (CT vs. pre-cracked Charpy V-notch), pre-crack depth and strain rate [155, 156]. These factors affect the master curve reference temperature T 0 and bring a bias to T 0 determination compared to data from large CT specimens. Considering actual practice for determining fracture toughness from surveillance specimens, applicability of 10 mm CT and 3.3 mm precracked Charpy V-notch specimens was examined [157–159]. The fracture toughness obtained with 10 mm CT specimens was almost comparable to those with full-sized pre-cracked Charpy V-notch specimens in A508 surveillance material irradiated up to 3.8 × 1019 n/cm2 [157].
3.	Some determination methods of fracture toughness master curves from surveillance Charpy data have been proposed. One method calculates the temperature dependence of fracture toughness from Charpy absorbed energy data using material properties such as work hardening coefficient and tensile strength [160]. Another determines the reference temperature of the master curve of Charpy absorbed energy. Although this method does not determine the temperature dependence of fracture toughness, the reference temperature of Charpy absorbed energy is almost of the same value as T 0 obtained from fracture toughness tests [161, 162]. Further studies are needed to assess the applicability of these methods to fracture toughness evaluation of RPVs.
4.	The scatter of the fracture toughness value is one of the key issues for both probabilistic and deterministic evaluation of structural integrity. The scatter initiates from microscopic variation such as carbide distribution even in a homogeneous material, and also from macroscopic variation of microstructure and material properties within the volume of a pressure vessel. Influence of macroscopic variation was recently reported using full size vessel products and trepans from pressure vessels of decommissioned plants. Influence of macro-segregation in ingots on fracture toughness was examined in French A508 forged rings and the scatter of surveillance transition temperature shifts and some high shifts data were explained by C macro-segregation in ingots [163, 164]. Other studies showed that the master curve reference temperature at the surface of unirradiated actual VVER440 pressure vessel steels was 35–70°C lower than those of the mid-thickness region and that this difference came from the microstructural difference induced by cooling rates [165]. However, fracture toughness distribution in pressure vessels of a decommissioned VVER, which had been annealed after 13 years in operation, showed no clear trend with large scatter, although the reason for the observed large scatter was unclear [166].
5.	Effects of warm pre-stress (WPS) on fracture toughness have been extensively studied in European countries. This is to establish a reasonable margin for PTS evaluation since WPS is expected to increase fracture toughness at low temperatures. A sample of fracture toughness data have been reported after various stress–temperature procedures, for example, load, cooling and fracture, and load, unload, cooling and fracture [167, 168]. It was confirmed that WPS effects appear in unirradiated pressure vessel steels. The WPS effect was also confirmed in irradiated steels. The increase of fracture toughness at 90°C from 100 MPa√m to 238 MPa√m after WPS (160°C) was observed in high-Cu weld (0.22%Cu) irradiated to 1.1 × 1019 n/cm2 [169]. The increase of fracture toughness after WPS was also reported in 1/2TCT surveillance specimens irradiated to 4.6 × 1019 n/cm2 [170].

3.3.4. Multiscale modeling

A large number of computer simulation works on radiation embrittlement have been conducted and have contributed to understanding the mechanisms of embrittlement and fracture mechanics. Multiscale modeling is a combination of physical and mechanical modeling in various time and space scales. It consists of modeling, for example, damage production processes by molecular dynamics calculations, diffusion processes by ab initio calculations, clustering processes by kinetic Monte Carlo calculations, dislocation interactions by dislocation dynamics calculations, and macroscopic deformation and fracture processes by crystalline plasticity calculations. In European countries, a multiscale approach was organized as the PERFECT project and has been followed by the PERFORM60 project [171–173]. The major results of the PERFECT project were summarized in a special publication [174].

The multiscale approaches provide both basic knowledge of microstructural evolution and micromechanics of fracture, although current modeling has not yet provided a prediction of fracture toughness changes which is applicable to actual reactor evaluation. Some examples of knowledge that has come from using the approaches are: the interaction and diffusion energies of point defects and various solutes [175] and the dislocation pinning strength of various clusters [176]. For micromechanics, several models for crack initiation at carbides and cleavage propagation were proposed to explain the shape and shift of the fracture toughness curve [177–181]. These efforts are still of essential importance to establish the physical basis of embrittlement prediction.

3.4. Summary of radiation embrittlement

Knowledge of microstructural features and understanding of mechanisms on radiation embrittlement have been significantly improved by recent development of nano-scale analytical techniques such as 3DAP and PA. The formation of solute clusters of Cu, Mn, Ni and Si, and dislocation loops has been confirmed in low-alloy steels irrespective of the material composition and irradiation condition. These features are now believed to be dominant for radiation hardening and embrittlement, and are incorporated into recently developed mechanism-guided embrittlement predictions used in some countries, which give much better correlation with surveillance data than previous empirical correlations. However, there are still unresolved issues to obtain a sound understanding of radiation embrittlement at higher fluences in long-term operations. The issues include identifying microstructural features that are dominant at high fluences for various material compositions and neutron fluxes, and formation and development processes of the features. As a physical basis, studies on the interaction between point defects and solute atoms, the interrelationship between MD and solute clusters, and the contribution of each microstructural feature to embrittlement are essential. Especially, the relationship between dislocation loops and Mn–Ni–Si clusters is important to understand the nucleation process and high-dose behavior of solute clusters. The acquired knowledge will be properly modeled and incorporated into embrittlement prediction or correlation.

4. IASCC of core structural materials

4.1. Radiation effects on SCC

IASCC is an intergranular-type SCC (IGSCC) that occurs under the radiation environment in the core of LWRs. The susceptibility, initiation and propagation of IGSCC are enhanced by neutron irradiation. Failures of core structural SS components were found in both BWRs and PWRs [6]. The IASCC susceptibility is mostly determined by the following primary factors: neutron fluence of materials, electrochemical corrosion potential (ECP) in water and applied tensile stress level. Many secondary factors are also known to affect IASCC: material conditions including alloy type, composition and cold work (CW); irradiation conditions such as temperature, flux and spectrum; environmental conditions such as temperature, water purity and dissolved oxygen (DO) and dissolved hydrogen (DH) concentrations; and stress–strain conditions such as strain rate.

The fundamental cause of IASCC has been considered to be irradiation-induced changes in materials. Neutron irradiation causes displacement damage in materials, which consequently causes changes in microstructure, microchemistry and macroscopic properties. Neutron irradiation also causes changes in stress–strain states through irradiation creep and stress relaxation. These changes are accumulated as the fluence increases. Furthermore, neutron irradiation together with gamma-ray irradiation causes radiolysis in water. The radiolysis produces various active chemical products causing ECP change in water containing DO, while the products disappear through reactions with hydrogen causing no ECP change in water containing DH. IASCC behavior has long been studied for two different water conditions. In this paper, the following terms are used for water conditions: BWR water conditions [288°C pure water containing DO, known as normal water chemistry (NWC), and hydrogen water chemistry (HWC) adding hydrogen to NWC]; and PWR water conditions [290–360°C water containing 500–1500 ppm B as H₂BO₃, ∼2 ppm Li as LiOH and ∼30 cc H₂/kgH₂O (∼2.7 ppmDH)]

IASCC data are obtained by conducting reactor irradiation in LWRs and MTRs followed by post-irradiation examinations (PIEs) using hot cell facility and involve large costs and long times; thus, systematic databases are still scarce. However, in the past two decades, initiation and growth data at high fluences have become available. In this section, current phenomenological knowledge on IASCC behavior is briefly summarized. In Sections 4.2, 4.3 and 4.4, radiation effects on materials, water and stress–strain, and mechanisms are discussed.

4.1.1. Susceptibility

IASCC susceptibility of LWR-irradiated materials has been examined mostly using slow strain rate tensile (SSRT) tests. This method has provided knowledge on influences of various parameters although the susceptibility means only the qualitative relative easiness of IASCC occurrence. Major points have been summarized in the literature [4–7]. IASCC susceptibility appears at ∼1 dpa in both oxygenated BWR water and hydrogenated PWR water conditions. This dose has been often recognized as the threshold dose for IASCC occurrence, but it does not mean substantial immunity to IASCC at doses lower than 1 dpa. The factors that enhance IASCC susceptibility are: higher DO, higher ECP, and lower strain rate in BWR water conditions [182]; and lower DH, higher water temperature and lower strain rate in PWR water conditions [183–185]. With respect to material factors, many elements have been examined by using SSRT tests and swelling tube tests. In BWR water conditions, IASCC susceptibility was suppressed by the addition of Mo, Cr, C, N, Hf and Ti, and enhanced by the addition of Si, P, Nb and Ti [186–190]. Data showing effects of material composition on IASCC in PWR water conditions are very limited.

4.1.2. Initiation

IASCC initiation has been investigated using constant load tests. The stress for IASCC initiation is decreased with increasing dose and tends to saturate at high doses [191–197]. Figure 14 shows the constant load SCC test data in the stress ratio to yield strength (σ/σ_y ) versus dose maps in BWR water conditions (288°C, 32 ppmDO) and PWR water conditions (320–340°C, 2.7 ppmDH). The tests were conducted using tensile type, O- or C-ring type, and bend beam type specimens for test duration up to 5000 h. The initiation stress was decreased to ∼0.4σ_y at doses less than 10 dpa in high-DO BWR water conditions [191, 192], while it decreased to ∼0.4σ_y at 30–40 dpa in PWR water conditions [194, 195,197]. The lower bound of stress for IASCC occurrence was almost the same for BWR and PWR water conditions. The crack initiation time in PWR water conditions decreased with increasing applied stress and dose. At 60–75 dpa, the initiation time became as short as 1–10 h.

Figure 14 Constant load SCC test data in the stress ratio to yield strength (σ/σ_y ) versus dose maps in (a) BWR water conditions (288°C, 32 ppm DO) and PWR water conditions (320–340°C, 2.7 ppm DH)

No systematic data showing influence of material variables on IASCC initiation have been reported, while there are data showing influence of water conditions. In BWR water conditions, decreasing DO from 32 ppm to 0.02 ppm suppressed IASCC initiation [191]. In PWR water conditions, IASCC initiation was enhanced by increasing water temperature [197], increasing DH concentration [198] and increasing Li concentration [199]. These trends are generally consistent with those of IASCC susceptibility as found by SSRT tests.

4.1.3. Crack growth

IASCC CGR dominantly depends on dose, stress and water conditions. Most crack growth data have been obtained using irradiated SSs taken from in-core components of LWRs; thus, CGR data have wide variations of material composition and irradiation conditions. Data compilation showed that very large scatter was found in CGR databases, with ranges of three to four orders [7]. This makes it difficult to find clear trends in some cases. Figure 15 shows the dose dependence of CGR in BWR water conditions at both high ECP (>150 mV_SHE) and low ECP (<–200 mV_SHE). This figure contains a series of MTR irradiation data [200] and other PIE and in-reactor data [201, 202]. The CGR clearly increased with increasing dose up to ∼5 dpa and tended to saturate at certain levels. CGR generally shifted to lower values by lowing ECP, while it is slightly increased with increasing dose at higher doses at low ECP. The saturation dose for CGR seemed to be lower than that for initiation stress in BWR water conditions as shown in Figure 14. The dependence of CGR on stress intensity K is expressed as the nth power of K. The n values in the literature were within 1.5–2.5 in BWR water conditions [7], while some data showed lower n values of 1.2 at higher doses in MTR data [200,203]. These data suggested that K dependence might be altered by irradiation and became smaller at higher dose.

Figure 15 Dose dependence of CGR in BWR water conditions at both high ECP (>150 mV_SHE) and low ECP (<–200 mV_SHE)

In PWR water conditions, CGR was examined in in-reactor tests and PIEs of LWR-irradiated or FBR-irradiated SSs at doses up to 40 dpa [204–206]. Some data showed very high CGRs (>10⁻⁸ m/s). Very large scatter was found, ranging by three orders. The upper bound of data became higher for higher doses [7]; thus, CGR seemed to be higher for higher doses. The n value in K dependence in PWR water conditions showed wide variation (1–4.5) at doses higher than 10 dpa. The observed large variation in CGR data is likely related to technical problems such as specimen size effects on K validity and K change with crack growth, as well as differences in material and irradiation conditions.

4.1.4. In-core IASCC

Since water radiolysis causes ECP changes in the core region of BWRs, in-reactor IASCC behavior is of essential importance for understanding the IASCC behavior of BWR core internals. One unique CGR data set on sensitized 304 SSs measured in an operating BWR showed that much higher CGRs were observed at an in-core position than at an out-of-core position and these corresponded to higher ECP for the in-core position [207]. In-core tests (in-pile or in-reactor tests) of SCC growth of irradiated SSs have been conducted in the Halden reactor [202,204] and the Japan Material Testing Reactor [208]. Figure 16 compares CGR data of 304 SSs from in-pile tests and PIEs in BWR NWC water conditions [208]. Comparison of the in-core data with corresponding PIE data (ex-core, out-of-core or out-of-pile data) showed that the in-core CGRs were similar to or slightly higher than PIE data at high ECP in BWR water conditions. Current data are insufficient to get quantitative understanding of the in-core effects on IASCC initiation and growth.

Figure 16 CGR data of 304 SSs from in-pile tests and PIEs in BWR NWC water conditions [208]

4.1.5. Simulation irradiation

Irradiation with ions and in FBRs has been used to study the IASCC behavior. Ion irradiations have advantages such as low activation and flexibility of irradiation conditions. Irradiations with 60 MeV He ions [209] and with 2–3 MeV protons (H ions) [210] were applied to investigate IASCC susceptibility. In the case of proton irradiation, irradiation temperature was selected to produce almost the same radiation-induced microstructure and GB segregation as those in LWR-irradiated SSs. The fact that the ion irradiation caused IASCC indicates that the displacement damage is essential for IASCC and that other factors specific to neutron irradiation such as transmutation are secondary effects.

Use of FBR-irradiated materials [211, 212] and irradiation in FBRs at temperatures lower than 370°C [213] have been tried to study the IASCC behavior at higher doses. Recent SSRT test results in PWR water conditions showed that the IASCC susceptibility of FBR-irradiated type 316 SSs was much lower than PWR-irradiated data [213]. Since no significant differences in mechanical properties and microstructures were found, helium effects were proposed as one of the causes for such difference. Helium generation was almost two orders higher in PWR-irradiated type 316 SSs. The influence of He on IASCC is still not clear and requires further studies in order to determine whether FBR data can be used for understanding the IASCC behavior in an LWR environment.

4.2. Radiation effects on materials

4.2.1. Microstructural evolution

Microstructural evolution in SSs under neutron irradiation has long been investigated and several reviews have already been published [214–216]. Recent careful TEM observations of LWR-irradiated SSs [217–223] revealed that interstitial dislocation loops are the dominant component of irradiated microstructures, that fine cavities or voids are formed at higher doses (higher He concentration), and that nickel silicide (Ni₃Si, γ’ phase) was found as radiation-induced precipitates. Figure 17 shows example TEM micrographs showing these features observed in type 316 SSs irradiated in a PWR to 53 dpa. The following subsections summarize the microstructural features.

Figure 17 TEM images showing microstructural features observed in CW type 316 SSs PWR-irradiated to 53 dpa: (a) rel-rod dark field image of dislocation loops; (b) dark field image of dislocation loops and black dots; (c) dark field image of Ni₃Si precipitates; and (d) defocused image of bubbles [220]

Interstitial dislocation loops

Interstitial dislocation loops (unfaulted Frank loops) are very fine and dense (5–20 nm in size and 10²²–10²³/m³ in density) in LWR irradiation conditions. Such fine loop microstructure remains almost unchanged even at very high doses (>70 dpa). The density increases with dose and saturates at doses of 1–5 dpa. Since the loops are very fine and dense at high doses, effects of material variables and irradiation conditions on dislocation loops have not been clearly identified. However, at low dose, these effects can be identified. The effect of flux was clearly found in type 304 SSs at very low flux irradiation [224]. A number of studies have been conducted on dislocation loop formation under LWR-relevant irradiation conditions using MTR or FBR irradiation [225, 226] and rate equation modeling [227]. Compiling LWR, MTR and FBR data, it was confirmed that the formation of dislocation loops is enhanced for lower temperature and higher flux in LWR irradiation conditions.

Since the size of dislocation loops formed in LWR irradiation conditions is comparable to the resolution limit of TEM, the nature of small defect clusters is difficult to fully identify. The formation of black dots was reported, which could not be identified as dislocation loops [220]. It was suggested that, since the black dots showed slightly faster recovery during PIA than the dislocation loops, the black dots contained some fraction of thermal unstable clusters, which might be the vacancy type and induced from cascades [220].

Cavities

The formation of cavities was reported in LWR-irradiated SSs and was thought to be related to the high He generation rate (5–20 appmHe/dpa). The cavity formation was very sensitive to irradiation conditions such as irradiation temperature and flux as well as material variables. The formation of fine bubbles was recently reported in type 304 SSs irradiated in BWRs to 5–10 dpa at 288°C [223]. The formation of fine bubbles and/or voids was commonly observed in PWR-irradiated SSs [218–220]. Voids were observed only at temperatures higher than 320°C, while fine bubbles were observed at temperatures lower than 320°C. The size of voids did not exceed 10 nm. The fine bubbles (1–2 nm in diameter) were homogeneously formed and their density was of the order of 10²³ /m³. The calculated volume swelling was within a range of 0.01–0.5%. The cavity formation directly relates to volumetric swelling of core internal components which results in the source of unexpected stresses. Swelling data are still scarce in actual conditions of PWR core internals (10⁻¹⁰–10⁻⁸ dpa/s, >20 dpa, 300–370°C).

Recently the fine bubble formation on GBs was evidenced in PWR-irradiated type 316 SSs to 33 and 70 dpa [219]. Such GB bubbles might affect the sensitivity to GB cracking. It was proposed that such an effect might have a link to the difference in IASCC susceptibility between FBR and PWR irradiations [213] since GB bubble formation was not observed in FBR irradiation with lower He production than PWR irradiation.

Precipitates

The precipitate commonly identified in LWR-irradiated SSs by using TEM is the Ni₃Si phase [218,220,223]. This phase is well known as radiation-induced in SSs [216]. The size and density of Ni₃Si precipitates determined by TEM were 3–5 nm and of the order of 10²¹ /m³, respectively. Recent 3DAP observations of irradiated SSs provided more detailed knowledge of radiation-induced phases [228–233]. In CW type 316 SSs irradiated to 12 dpa in a PWR, Si aggregates and Ni–Si clusters were identified [228]. The Ni–Si clusters were ∼10 nm in diameter and ∼6 × 10²³ /m³ in number density, and their atomic composition was 50Ni–40Si. Mo and P were enriched at the interface of the clusters. In type 304 SSs irradiated to 24 dpa at 300°C in a PWR, Ni–Si clusters were found, which were ∼10 nm in diameter and 4 × 10²³ /m³ in density [232]. The averaged atomic composition was 32Fe–40Ni–14Si–11Cr. The ratio of Ni/Si was consistent with that of the Ni₃Si phase. Mn was enriched in some of the clusters and P was segregated at the interface of the clusters. Both observations confirmed the formation of Ni–Si clusters with similar size and density. It is notable that the reported density of the Ni–Si clusters by 3DAP was almost one order higher than the typical density of Ni₃Si precipitates in the above TEM analyses [218,220,223] and also higher than the density of dislocation loops. Another detailed 3DAP analysis on proton-irradiated type 304 SSs clearly showed that Ni and Si atoms segregated at dislocation loops, the same as at line dislocations and GBs [231]. It is reasonable to think that the Ni–Si clusters are the precursors of Ni₃Si phases and nucleate preferentially at interstitial dislocation loops. Si is an undersized atom and preferentially migrates to sinks via a fast interstitial diffusion mechanism of mixed dumbbells. Interestingly, no evidence was reported for Ni₃Si phase formation at GBs in spite of a sufficient level of Ni and Si segregation.

As to clusters or phases other than the Ni₃Si phase, Si aggregates [228], Cu-rich clusters [231], Mn-rich Ni–Si clusters [232] and carbides [234] were observed in irradiated SSs. However, the generality of these features is unclear.

4.2.2. Mechanical properties

Tensile properties

The formation of radiation-induced microstructural features consequently causes changes in stress–strain response under tensile loading. These include: hardening, increase in yield strength and tensile strength, and decrease in ductility (uniform elongation and total elongation). Radiation-induced tensile property changes in SSs have been well studied in wide temperature ranges [235]. Tensile property data in LWR irradiation conditions have been accumulated in the last two decades. Figure 18 shows the change in yield strength and total elongation of LWR-irradiated SSs [194,196,236–239]. The yield strength increases rapidly with increasing dose and saturates to 800–1000 MPa at doses of 1–10 dpa. The increase in yield strength accompanies the decrease in ductility and work hardening coefficient. At an yield strength of 600–700 MPa, the uniform elongation becomes very low and plastic instability appears. As the yield strength saturates, the elongation saturates to ∼5%. Hardness is often used as a measure of mechanical properties. The increase in hardness (DHv) is almost linear with the increase in yield strength (Dsy), and a correlation of Δσ_y (MPa) = 3.03ΔHv (kg/mm²) was derived based on data of LWR-irradiated SSs [240].

Figure 18 Change in (a) yield strength and (b) fracture elongation in LWR-irradiated SSs

Tensile property changes originate from the elastic interaction of moving dislocations and microstructural features such as dislocation loops. The increase in yield strength (Δσ_y ) has a good correlation with the evolution of microstructural features predicted by the barrier hardening model [235]: , where N_i and d_i are the number density and diameter of the ith microstructural feature, respectively. This relation was confirmed in LWR-irradiated SSs assuming the proper hardening coefficient for each feature [218–220,223, 224].

Factors that affect microstructural evolution consequently cause variation of tensile property changes. Material composition and CW, and irradiation temperature and flux are known as such factors. Variations in these factors are probably the main reason for the observed variations in yield strength at low doses in Figure 18. Since the dominant microstructural feature is dislocation loops in LWR irradiation conditions, a faster increase and saturation in yield strength are caused by factors that enhance dislocation loop formation: for example, higher flux, lower irradiation temperature, and addition of C, Si, P and Nb [196,224,238, 239]. However, effects of these factors on a saturation level of yield strength have not been clearly identified.

Deformation mode and microstructure

Radiation-induced microstructural changes cause significant changes in deformation mode as well as hardening. The deformation mechanisms in austenitic SSs are slip formation at higher temperature and twinning at lower temperature. Slip formation is dominant at around 300°C. The deformation mode in irradiated SSs becomes heterogeneous and planar from the homogeneous one in unirradiated conditions. This change occurs due to the formation of dislocation channels in which moving line dislocations clear away dislocation loops, forming a path for subsequent slips. In such a deformation mode, plastic strain is localized within channels. In the last decade, channel formation and flow localization in irradiated materials have been extensively studied [241–244] and their influences on intergranular fracture and IASCC have also been examined [245–249].

A detailed observation of deformation microstructure in LWR-irradiated SSs was reported in several references [242,245,246,249]. The channel formation was commonly observed in LWR-irradiated SSs after deformation at around 300°C. Figure 19 shows an example of dislocation channels observed near the surface region in PWR-irradiated type 316 SSs after being slowly deformed to 3% at 300°C [245]. The width of the channels was found to be 20–100 nm and the estimated strain within a channel might exceed 100%. It was confirmed from crystalline plasticity calculations that the dislocation channels were formed by the local shear component of tensile stress [250]. The twin formation was observed and seemed to be pronounced after deformation at room temperature or at a fast strain rate, and also in the mid-thickness region of the specimens where high constrain and multiaxiality exist [245]. Since the formation of channels (slips), twins and ɛ martensites occurs on the (111) planes in SSs, these features may coexist in a grain and in a slip band in irradiated and deformed commercial SSs [241,245,246,249,251].

Figure 19 Near-surface deformation microstructure in CW type 316 SSs deformed to 3% at 320°C: (a) irradiated to 35 dpa showing coarse slips and surface steps with an enlarged image of the dislocation channel and (b) unirradiated showing tangled dislocations [245]

Dislocation channels terminate at GBs or free surfaces. The interaction of dislocation channels with GBs causes local concentration of stress–strain. There are three different interactions as schematically shown in Figure 20. Channel transfer occurs when the misorientation between dominant slip planes in two adjacent grains is small. This interaction produces steps on the GB, as shown in Figure 20 (c). The absorption of dislocations at GBs causes GB sliding. The absorption also produces local strain in GBs. When channel transfer cannot occur with the neighboring grain, dislocation pileup occurs at the GB and produces a high stress field near the intersection, as shown in Figure 20 (b). Detailed morphology of these interactions was examined by SEM and TEM observations not only in neutron-irradiated SSs, but also in ion-irradiated SSs [252–255]. Recent EBSD studies also confirmed the existence of high local misorientation at dislocation pileups near GBs [256, 257].

Figure 20 Schematic illustration depicting (a) three types of interactions of dislocation channels with GBs; (b) SEM image showing dislocation pileups in irradiated CW type 316 SSs after slow deformation to 13% at 320°C; and (c) TEM image showing channel transfer in irradiated CW type 316 SSs after slow deformation to ∼3% at 320°C

Deformation mode and microstructure are affected by material variables and irradiation conditions through effects on microstructural evolution. With respect to material variables, stacking fault energy (SFE) is well known as an essential parameter for deformation microstructure in non-irradiated conditions. It was confirmed that SFE also affected deformation mode after irradiation [253, 254,257]. From data of average strain in each channel obtained from surface step measurements on proton-irradiated austenitic alloys with different Cr and Ni contents, a higher degree of strain localization was found for lower SFE alloys. It is also known that the formation of dislocation pileups results in microcracking along GBs. This is described in Section 4.4.2.

4.2.3. Grain boundary segregation

Radiation-induced segregation (RIS) is a common phenomenon in various alloys under irradiation. The phenomenology and underlying mechanisms of RIS at GBs are generally well understood for SSs in LWR irradiation conditions [5–7]. Figure 21 shows typical solute distribution at GBs (PWR-irradiated type 316 SSs to 35 dpa) and changes in GB segregation (changes from the in-grain level) of Cr, Ni and Si in LWR-irradiated commercial SSs [182,218,220,223,258, 259]. Depletion of Cr and enrichment of Ni and Si are common RIS behaviors in irradiated SSs. RIS rapidly evolves at doses lower than 5–10 dpa and its change becomes very gradual at higher doses. For other elements, depletion is known for Mo and Mn, and enrichment for P and S. The width of the RIS region is very narrow, typically less than 10 nm from GBs, as shown in Figure 21. Recent 3DAP analyses [228–233] confirmed that the RIS behavior is quantitatively consistent with previous data from TEM-EDS (energy dispersive X-ray spectroscopy) analyses. Furthermore, 3DAP analyses confirmed enrichment of B, C and S in proton-irradiated type 304 SSs [231], while these elements might be segregated before irradiation. RIS to dislocation loops has been clearly confirmed by 3DAP analyses. Si and Ni co-segregation at dislocation loops is believed to be the cause of Ni₃Si phase formation as described in Section 4.2.1.

Figure 21 Typical solute distributions across a GB and changes in GB segregation of Cr, Ni and Si with dose in LWR-irradiated SSs

RIS is affected by material variables and irradiation conditions such as irradiation temperature and dose rate. The degree of RIS has a peak temperature which is determined by the balance of RIS and thermal back-diffusion induced by the concentration gradient. LWR-relevant temperature (290–370°C) is probably lower than the peak temperature of Cr and Ni segregation at LWR-relevant dose rates (10⁻¹⁰–10⁻⁸ dpa/s). The RIS of major elements induced by a vacancy mechanism is generally enhanced at a lower dose rate since the freely migrating vacancy concentration might be higher. This trend was confirmed by recent comparisons between low-flux LWR data, high-flux FBR data, and high-flux proton irradiation data in type 304 and 316 SSs [260, 261]. Effects of material composition and alloying elements on RIS in SSs have been well examined [262–264]. In general, the addition of minor elements such as C, N, Zr and Ti suppresses RIS by reducing diffusivity of vacancies and interstitials by trapping.

It is widely accepted that Cr depletion and Ni enrichment occur mainly through an inverse Kirkendall mechanism in the Fe–Cr–Ni system [265]. In this mechanism, the faster diffuser Cr preferentially diffuses away from GBs via the vacancy flux to GB sinks, and then the slower diffuser Ni is accumulated near GBs. The depletion of oversized Mo and Mn atoms may be explained by this mechanism. The mechanism of enrichment of undersized elements such as Si and P is considered to be the migration of solute-interstitial mixed dumbbells [266]. Based on these mechanisms, several RIS models have been developed and applied to measured data for discussing the influence of various factors: for example, the contribution of Ni-interstitial binding, and influences of temperature, dose rate and material compositions [267–269]. Modeling efforts still continue, for example, to develop multi-element RIS models for commercial SSs [270, 271], and as an ab initio approach for determining reasonable physical parameters and the contribution of interstitial diffusion [272].

4.3. Radiation effects on the water environment and stress–strain

4.3.1. Water radiolysis and H₂O₂ effects

In a radiolysis scheme, water is decomposed into various radicals and molecules with the final radiolytic products of H₂O₂, H₂ and O₂. The effects of radiolysis on the water environment have been well reviewed [4,6]. ECP increases significantly at a low-DO condition, mainly due to the formation of H₂O₂. ECP is recognized as a direct measure of water radiolysis effects on IASCC and then one of the key factors for IASCC in BWR water condition [6]. In general, effects of bulk water chemistry on SCC (IASCC) for different DO and DH concentrations are recognized as effects of ECP. ECP at a given temperature is dependent on concentrations of DO, DH and other oxidants, which is dominated by the radiation dose rate and water flow rate. In BWR water conditions, ECP changes with position in the core.

Water conditions within narrow cracks and crevices may differ from bulk water conditions. The ECP at a crack tip remains low due to oxygen consumption and is almost insensitive to bulk water chemistry. While it was pointed out that the influence of radiation on ECP at crack tips may be negligible [6], the specific role of H₂O₂ in oxidation, crack growth and ECP has been studied in the last decade [273–277]. It was found that the energy deposition by radiation was enhanced within a crack due to back-scattered radiation from the surrounding material [273]. This suggested that water radiolysis might be enhanced within a crack compared to bulk water radiolysis. It was pointed out that the difference in ECP behavior of SSs exposed to water containing DO or H₂O₂ was related to the difference in thickness and the evolution of inner layer oxides. These findings suggested that H₂O₂ has an additional role in corrosion and IASCC under radiation field. However, quantitative estimation or data showing the influence of H₂O₂ on initiation and growth of IASCC in an LWR environment have not been found. This issue is still important for quantitative understanding of the difference between the IASCC behavior in LWRs and out-of-core PIE data as described in Section 4.1.4.

4.3.2. Irradiation creep and stress relaxation

Creep and stress relaxation are enhanced under LWR irradiation conditions [215]. Stress relaxation causes the decrease in stress under constant strain such as weld residual stress and bolt or spring load, and thus may suppress IASCC initiation and growth. This results in beneficial effects on LWR core internal components. On the other hand, irradiation creep enhances plastic strain at crack tips under constant stress and might result in accelerating crack growth.

Understanding on irradiation creep has obtained mainly from data at temperatures higher than 400°C [278] and experimental data at LWR-relevant temperatures (280–370°C) are scarce [279, 280]. Recently data on irradiation creep and stress relaxation at temperatures around 300°C have been accumulated by MTR irradiation experiments in the last decade [281–287]. Figure 22 shows the data of irradiation stress relaxation in type 304 and 316L SSs at 288°C [280–284]. The ratio of the measured stress (σ) to the initial applied stress (σ ₀) was decreased with dose and became less than 0.5 at ∼5 dpa. Measurements of weld residual stress were conducted by neutron diffraction techniques [283, 284] and showed that the tensile stress was relaxed at a similar rate to data from ring or tensile type measurements. Large scatter observed in Figure 22 was due to scatter of transition relaxation (corresponding to primary creep), which appeared due to the relaxation of preexiting internal strains. The degree of transition relaxation depends on the thermomechanical treatment and stress states [288]; methods for quantitative estimation have not been established yet.

Figure 22 Data of stress relaxation in type 304 and 316L SSs under MTR irradiation at 288°C

Neglecting the swelling contribution to creep in LWR irradiation conditions, irradiation creep strain ɛ under stress σ at dose f is generally expressed by the sum of transition creep and steady creep:

where A ₁, A ₂, B and n are constants. At low stresses less than 200 MPa, typical values are: n ∼ 1 and B ∼ 1 × 10⁻⁶/MPa/dpa. At higher stresses around 500 MPa in CW type 316 SSs, the n value of 3 was reported in an in-pile experiment [286]. In-beam experiments with 17 MeV protons at 288°C and 2 × 10⁻⁷ dpa/s showed that the n value was 5 at stresses higher than 500 MPa in CW type 316 SSs [289]. Effects of material variables and irradiation conditions such as temperature and dose rate are largely unknown in LWR irradiation conditions since systematic in-reactor experiments are difficult to conduct. Model calculations based on creep mechanisms at low temperatures (<300°C) are effective tools to understand effects of material variables and irradiation conditions [290]. Dominant mechanisms of creep are considered to be stress-induced preferential nucleation of interstitial dislocation loops, dislocation climb by stress-induced preferential absorption of point defects, and preferred absorption and glide. Creep strain becomes larger for low temperature and high dose rate. These trends appear by interstitial dominant creep mechanisms in low temperatures where vacancy mobility is not high.

With respect to material variables, the creep rate is known to be suppressed by increasing CW level, and addition of P, N, Si and Mo from data at higher temperatures (>400°C). Recent well-controlled proton beam experiments and accompanying model calculations showed that the creep rate of type 316L SSs was lower for 25%CW than 5%CW and that the stress relaxation of type 316L SSs was higher than that of type 304 SSs [289,291].

4.4. Mechanisms of IASCC

4.4.1. Factors controlling IASCC

Correlations of IASCC susceptibility with irradiated material properties have been discussed in the last two decades to find key factors for IASCC occurrence under given water and applied stress conditions. Hardening as a mechanical factor and GB segregation as a chemical factor have been the main concerns for IASCC occurrence. Major results of such correlations are summarized below.

Figure 23 Dose-dependent changes in IASCC, mechanical properties and RIS in PWR-irradiated CW type 316 SSs [298]

1.	Compilation of LWR-irradiated data revealed that IASCC susceptibility, as examined by SSRT tests, generally increased with decreasing GB Cr concentration [5,182,186] in both BWR and PWR water conditions. Similar correlation was found in SSs containing Mo (Cr + Mo effects) and having GB Cr enrichment before irradiation [292]. Recent CGR data in BWR water conditions also showed a good correlation with GB Cr concentration [293]. GB Cr concentration is a key factor for the passivation behavior and oxide characteristics near GBs.
2.	IASCC susceptibility appeared at the yield strength of 500–600 MPa and increased with increasing yield strength in both BWR and PWR water conditions [5,294]. CGRs of irradiated SSs were slightly higher than the linear trend seen between CGR and yield strength in non-irradiated CW SSs [295, 296]. Higher CGRs of irradiated SSs might be due to GB RIS effects.
3.	Dislocation microstructure in irradiated SSs (dense dislocation loops) was quite different from that in non-irradiated CW SSs (tangled line dislocations). To identify the influence of this difference, IASCC susceptibility in BWR water conditions was examined in type 304 SSs by changing the CW level and proton irradiation dose while keeping the same total hardness [297]. IASCC occurred only in high-dose proton irradiated SSs with null or low CW level. This result indicated that irradiation hardening has a specific role for IASCC initiation and that the deformation mode is more important for IASCC than for macroscopic hardness.
4.	IASCC, mechanical properties and RIS generally show similar dose-dependent changes, as shown in Figure 23 for PWR-irradiated CW type 316 SSs [298]. These show rapid change at doses lower than 10 dpa and very slow change or saturation at higher doses. To differentiate the roles of the mechanical factor (hardening) and chemical factor (GB segregation) on IASCC, PIA and high temperature irradiation were applied to produce specific material states that are not achieved in LWR irradiation conditions [294,299,300]. These experiments showed that hardening might have larger effects on IASCC than GB segregation in PWR water conditions and vice versa in BWR water conditions; importantly, neither hardening nor GB segregation alone could explain IASCC susceptibility.

Adding to the knowledge on the correlations, recent TEM observations of IASCC crack tips have brought forth important information on processes involved in IASCC [301–304]. In type 316 SS bolts removed from a PWR after a 17 cycle operation, cracks were filled with oxides and porous Cr-rich spinel oxides on the crack wall extended to the crack tips [301]. The oxides had steps at the intersections with slip bands. In type 304 SS components removed from a BWR after a 23 year operation, similar things were seen: cracks were filled with oxides and thin Cr-rich spinel oxides on the crack wall extended to the crack tips [302]. The crack tip oxides were finger-like with Ni enrichment ahead of the tips. These cracks in plant components were highly oxidized and seemed not to be active. The observations of growing cracks at very high rates during PIE SCC tests in PWR-irradiated type 316 SSs showed that oxidation near crack tips did not precede so much but that oxygen was detected near crack tips in both BWR and PWR water conditions [304]. Figure 24 shows the TEM image and solute distributions near crack tips in CW type 316 SSs PWR-irradiated to 38 dpa after a constant load test at stress of 750 MPa in PWR water conditions [304]. The oxygen concentration near the crack tip is very low and the crack has steps on the wall. Such steps were also found in PWR-irradiated type 316 SSs [301] and corresponded to the intersection with slip bands or deformation twins. Although crack tip observations of growing IASCC cracks are still scarce, the available observations have indicated that crack tip oxidation or oxygen penetration occurs within very narrow cracks, typically less than 5 nm, in both BWR and PWR water conditions, and that the crack propagation relates to deformation microstructure such as slip bands or twins near GBs.

Figure 24 TEM image and solute distribution near the IASCC crack tip in CW type 316 SSs PWR-irradiated to 38 dpa after the constant load test at 750 MPa in PWR water conditions [304]

To date it is widely accepted that IASCC occurrence cannot be attributed to a single static material property and is the result of combined effects of multiple dynamic processes. The important factors are likely to be oxidation and deformation. The oxidation and corrosion kinetics near GBs in a water environment are controlled by GB chemistry and water conditions, and also determine the fracture strength of oxidized or corroded GBs. GB chemistry includes not only RIS, but also the existence of other elements such as He generated under irradiation and H generated by corrosion processes. He and H are likely to have influences on GB cohesion or strength. The deformation controls local stress–strain exerted on GBs and GB fracture processes. Material variables and irradiation conditions affect deformation and oxidation processes through microstructural evolution and RIS. Figure 25 shows a schematic illustration depicting various processes related to IASCC. Recent studies for IASCC mechanisms have placed the focus on clarifying the role of deformation and oxidation processes.

Figure 25 Schematic illustration depicting various processes related to IASCC

4.4.2. Role of deformation

As described in Section 4.2.2, dislocation channeling is a distinguishing characteristic of the deformation microstructure in irradiated SSs. The interactions between dislocation channels and GBs cause high local strain–stress near GBs. When dislocation pileups occur, microcracks can be initiated as Zener–Stroh type cracks by coalescing dislocations piled up along a slip plane. GB separation also can occur if the local stress exceeds GB cohesion strength. Such IG fracture can occur without any environmental factors and have been found in LWR-irradiated SSs. IG fracture in an argon gas atmosphere was found in type 304 and 316 SSs LWR-irradiated to 7–73 dpa after tensile tests at low strain rates (<10⁻⁶ /s) and temperatures of 300–340°C [184,194,197,305,306]. The IG cracking initiated from the specimen surface and the sensitivity to IG fracture decreased with increasing tensile strain rate [305]. Similar IG cracks in an argon atmosphere were not found at 288°C in type 304 SSs irradiated to 18 dpa [307] and in type 316 SSs irradiated in a FBR to 50 dpa [212], suggesting that the IG fracture is easier at higher temperatures (300–340°C vs. 288°C). The IG fracture without an environmental factor might be a common phenomenon in highly irradiated SSs at high temperatures and a low strain rate [308]. In such conditions, the dominant deformation mode is slip/channeling in irradiated SSs.

Characteristics of IG cracking have been examined by surface observations after tensile tests were interrupted at low elongation levels. A typical example of IG cracks formed in PWR-irradiated CW type 316 SSs after slow deformation to ∼3% at 300°C is shown in Figure 26 [245]. The cracks were observed on GBs with angles of 60°–90° to the tensile direction and channel-induced surface steps were observed for only one side of the grain. These observations indicated that the cracks were formed along GBs with high normal stress and with dislocation pileups. Direct correlation between localized deformation and IASCC in a water environment has been examined in proton-irradiated SSs while changing material compositions and irradiation conditions [247,253,257]. As shown in Figure 27 [247], it was clearly confirmed that IASCC susceptibility (crack length per unit area) became higher for a higher degree of deformation localization (average step height of the slip channel on the surface). It was also confirmed that IASCC susceptibility was found to correlate with the slip continuity at GBs [309], and relative spatial arrangement between tensile direction, GB, Schmidt factors and channel directions for both sides for grains [245,309,310].

Figure 26 IG cracks formed in PWR-irradiated CW type 316 SSs after slow deformation to ∼3% at 300°C in an argon atmosphere [245]

Figure 27 Contribution of localized deformation as measured by the weighted average channel height to cracking in SSs irradiated with protons to 1–5 dpa and SCC-tested in BWR water conditions [247]

IASCC initiation stress under constant load SCC tests was in the yield strength range of 0.4–0.5 at higher doses in both BWR and PWR water conditions (see Figure 14). These stress levels were comparable to the estimated stress required to cause coarse slips on the planes with the most preferable orientation (Schmidt factor of 0.5) [245], assuming that the local stress within a grain can reach two times the applied stress based on crystalline plasticity calculations [311]. Recent EBSD measurements of misorientations in PWR-irradiated CW type 316 SSs showed that macroscopic elastic deformation at ∼50% of the yield strength caused detectable strains near GBs [312]. Recent modeling studies for channel–GB interactions also showed a similar role to that of dislocation pileups [313, 314]. Although more evidence is needed to firmly correlate IASCC initiation with dislocation pileups in channels, these findings support the idea that the dislocation pileup in dislocation channels might be one of the key triggers for the IASCC initiation site. Since there are incalculable variations in channel–GB interactions in a specimen due to variations in the spatial configuration of grains and GBs, a probabilistic approach might be useful for a quantitative understanding of the role of localized deformation on IASCC initiation.

Channel transfer to the neighboring grain and dislocation absorption can cause GB steps and GB sliding, respectively, as described in Section 4.2.2. Such steps and sliding may break surface oxides if applied strain is sufficiently larger than the ductility of oxides. When this process continues in a channel and a GB, accelerated oxidation is expected to occur at the intersection and along the GB, and might result in triggering IASCC initiation. However, these type of interactions might not be dominant since IASCC cracks were more frequently found for GBs without slip continuity in BWR water conditions [308].

4.4.3. Role of oxidation

Oxidation kinetics and the nature of oxides are key information for the understanding of mechanisms of IASCC as well as IGSCC in non-irradiated SSs. The oxidation rate and structure of non-irradiated SSs in LWR water conditions have been well examined [315–317]. The surface oxide is a double-layer structure: an inner Cr-rich spinel layer and an outer Fe- and Ni-rich spinel layer in PWR water conditions, and an inner Fe–Cr–Ni spinel layer and an outer hematite layer in BWR water conditions. There are proposed IGSCC mechanisms of austenitic SSs in LWR water conditions such as slip dissolution or slip oxidation [207,318], creep and GB sliding [319], and hydrogen embrittlement [320]. In these mechanisms, oxidation plays essential roles for not only the oxidation itself, but also as a source of vacancies and hydrogen atoms. Enhancement of oxidation results in enhanced sensitivity to cracking in any mechanism. While a similar role of oxidation is expected for IASCC occurrence, experimental studies on the role of oxidation in IASCC of irradiated SSs are scarce.

Morphology of oxides formed on irradiated SSs has been examined in only a few papers. In type 316 SSs irradiated to 5 dpa with protons and exposed to BWR NWC water for 70 h, neither enhancement of oxidation nor a difference in oxide structure was found compared to unirradiated SSs [321]. In CW type 316 SSs irradiated up to 73 dpa and exposed to PWR water conditions for 1000 h, slight enhancement of the inner oxide layer thickness was found [322]. Preferential oxidation and Ni enrichment ahead of the oxides were also found along several GBs, as shown in Figure 28 [322]. This indicated that the oxidation of both the matrix and GBs might be enhanced in irradiated SSs. Although more data are needed for conclusive understanding, GB oxidation might be enhanced by RIS at GBs, possibly as Cr depletion and Si enrichment. It was confirmed that as the Cr content in material decreased, the inner oxide layer became thicker with lower Cr fraction and then became less protective as a diffusion barrier [316]. A Si-enriched area might be easily oxidized and more soluble and then would enhance GB oxidation and GB fracture [323, 324]. This might be one of the reasons for enhancement of IASCC susceptibility by Si addition reported in BWR-irradiated and proton-irradiated SSs [188,325]. Preferential and accelerated oxidation at GBs may be a key factor for IASCC. However, detailed knowledge of GB oxidation in irradiated SSs is still limited, especially considered in combination with water radiolysis effects.

Figure 28 Distribution of Fe, Ni, Cr and O in the cross section of the surface oxide layer in type 316 SSs PWR-irradiated to 20 dpa after immersion in PWR water conditions for 1000 h [322]

4.4.4. Other contributors

Hydrogen effects

Hydrogen is known to affect cracking processes through various metallurgical factors. There are a number of studies concerning hydrogen effects on cracking, mechanical properties, and microstructure in various non-irradiated materials. However, knowledge on hydrogen effects on irradiated materials and IACC is limited. Several measurement results of hydrogen concentration were reported for LWR-irradiated SSs [213,218,220,326,327]. The measured values ranged from 10 to 200 wtppm with very large scatter and showed no clear trend against dose. Since the measured values were generally inconsistent with calculated generation amounts from nuclear reactions [220], hydrogen in LWR-irradiated SSs is likely to originate from the water environment probably through corrosion reactions.

The relationship between hydrogen and IG cracking in irradiated SSs has been discussed in a limited number of papers. The hydrogen measurements in type 316 SSs irradiated to 6 and 53 dpa after SSRT tests in PWR water conditions revealed that the hydrogen concentration near the IG fracture surface was slightly higher than that away from the fracture surface [184]. This provided evidence that hydrogen is associated with cracking processes in irradiated SSs. Hydrogen charging effects on mechanical properties were examined in type 304 SSs BWR-irradiated up to 20 dpa [328]. After hydrogen charging and discharging at 100°C, IG fracture was found after tensile tests at room temperature and its fraction increased with increasing dose. The cause of IG fracture was attributed to combined effects of hydrogen-induced martensite formation and radiation-induced GB Cr depletion, based on the similarity to hydrogen-induced IG cracking in sensitized type 304 SSs. Hydrogen charging effects were also examined in type 304 SSs BWR-irradiated to 2–3 dpa [329]. IG cracking was found by bending at 23°C after hydrogen charging, indicating that hydrogen caused IG cracking. Similar IG cracking was found by bending at 23°C after SSRT tests in BWR water conditions. However, steels that exhibited higher IASCC susceptibility by SSRT tests in BWR water conditions showed lower sensitivity to IG cracking at 23°C after SSRT tests and hydrogen charging. Thus, these results seem to suggest that processes related to hydrogen might not be involved in IASCC in BWR water conditions. Hydrogen embrittlement or hydrogen-induced cracking might be a possible effect of hydrogen on IASCC. High yield strength and well-developed RIS in highly irradiated SSs might provide more preferable conditions for hydrogen embrittlement. However, based on current irradiated data, it is still uncertain whether hydrogen is the cause or the result of the cracking.

Another possible effect of hydrogen is the interaction of hydrogen with the deformation microstructure [330, 331]. It is known that hydrogen, if the concentration is sufficiently high, can relax elastic interactions between moving dislocations and obstacles enhancing dislocation gliding and it can reduce SFE and cross-slip enhancing the planar dislocation microstructure. These effects are expected to enhance dislocation channel formation in irradiated materials. This effect was confirmed by slow tensile tests on type 316 SSs irradiated to 5 dpa with Fe ions [332]. The average spacing between surface slips after slow tensile tests at 300°C was larger in an hydrogen atmosphere than in an argon atmosphere. TEM observations on the specimen cross sections confirmed that the surface slips were formed by dislocation channels which could penetrate the near-surface damaged area containing dense dislocation loops. High-concentration hydrogen is expected to enhance deformation localization and might have an enhancement effect on IASCC initiation and growth.

GB cohesion or strength

Among various microstructural and microchemical changes, some changes result in the decrease in cohesion or strength of GBs. This might not be a direct cause of IASCC but might have an important role in crack initiation and propagation. Possible factors are hydrogen, helium, RIS and oxygen. As for hydrogen, it causes GB decohesion which is one of the causes of hydrogen embrittlement.

As described in the previous sections, He is generated under LWR irradiation and helium bubbles are formed on GBs in PWR-irradiated SSs to high doses [219]. The high generation rate of He in PWR irradiation compared to FBR irradiation was considered as one of the reasons that PWR-irradiated SSs exhibited higher IASCC susceptibility than FBR-irradiated SSs [213]. It is well known that He induces IG fracture and significantly reduces ductility in slow tensile and creep conditions at high temperatures (typically higher than 400–450°C) [235]. At LWR-relevant temperatures, high-temperature He embrittlement is not believed to be important. However, dense fine bubbles on GBs might reduce stresses that would lead to GB fracture. The criteria for IG cracking in He-implanted SSs were reported to be ∼30 nm for cavity spacing and ∼4 nm for the cavity diameter at temperatures lower than 500°C [333]. He bubbles observed in PWR-irradiated SSs seem to satisfy these conditions. In recent micro-tensile tests on helium-implanted type 316 SSs, preferential fracture on GBs was found [334]. Fine bubbles on GBs might provide a preferential path of fracture with probably lower strength than the matrix, but the exact role of helium and bubbles is not yet understood.

GB segregation of solute elements is a potential factor enhancing GB fracture. GB embrittlement due to segregation of impurities such as P has been extensively studied for body centered cubic (BCC) alloys and steels. Recent ab initio calculations of GB cohesive energy showed GB decohesion by some elements in face centered cubic (FCC) systems [335–337]. For Ni Σ5 (012) symmetrical tilt GBs, most of the solute elements including helium have embrittling potency, while B, C and Si can enhance the GB cohesion [335]. Other recent ab initio calculations showed that significant GB decohesion occurs for segregation of S and O atoms in Ni [336, 337]. It is known that S enhanced IASCC susceptibility in BWR water [188,329]. S-induced decohesion might assist IG cracking in an argon atmosphere and also assist IASCC susceptibility, while Ni–S film formation at GBs due to S and Ni segregation has been proposed [329]. In LWR water conditions, O atoms are preferentially consumed by forming oxides rather than being segregated at GBs. However, in the observations of rapidly growing IASCC crack tips, O atoms penetrated into the GBs ahead of the tips [304]. In such cases, O atoms at GBs might have some role in GB separation ahead of the crack tip and CGR might have relation with O diffusion along GBs. The basic physical knowledge of GB cohesion or strength is helpful to understand mechanisms of IASCC from a wider viewpoint of IG cracking phenomena in irradiated materials.

4.5. Prediction and modeling of IASCC

4.5.1. Crack initiation and crack growth rate

Empirical deposition curves for IASCC data have been proposed for IASCC initiation and CGR [7,196,338]. The curves of CGR (da/dt) are usually a function of stress intensity factor (K): for example, da/dt = AKⁿ , where A and n are the constants depending on the irradiation dose, and material and water conditions. The constants were fitted to the CGR data from PIEs and in-rector experiments. For IASCC initiation, empirical deposition curves in PWR conditions have also been discussed [7,196]. The curves for IASCC initiation were basically determined by the lower limit of applied stress for failure in post-irradiation constant load SCC tests within 2000 h. While there are two curve types, initiation time and initiation dose, limitation of data made it difficult to set curves applicable to high doses and long times. Since IASCC data have limitations in the number and quality and have large scatters, the reliability of empirical curves depends on such limitations and needs to be improved by better understanding of the mechanisms.

Mechanism-based modes for IASCC CGR have been proposed based on slip dissolution or slip oxidation mechanisms of IGSCC, in which CGR is modeled by the oxide rupture rate due to crack tip strain [207,318]. Assuming that IASCC and non-irradiated IGSCC are controlled by the same slip oxidation processes, the models have been applied to IASCC CGR by considering radiation-induced changes, that is, the increase of yield strength for crack tip strain and RIS for oxidation or repassivation rate. These models successfully predict IASCC crack growth behavior. For IASCC initiation, no mechanism-based modeling has been reported yet. To establish reliable prediction methods for IASCC initiation, continued efforts for modeling and data creation are needed. For example, trial modeling of an IASCC initiation curve was reported by correlating IASCC with predicted increase in yield strength and GB Cr depletion [339]. Furthermore, methods for correlating PIE data with IASCC initiation in operating LWRs are required for realistic prediction, where materials are simultaneously exposed to irradiation, coolant water and stress–strain for long time and where the loading condition changes with time.

4.5.2. Other related properties

Evaluation of the structural integrity of LWR core internals requires not only precise prediction methods for IASCC initiation and growth rate, but also for other materials properties related to stress conditions. Under irradiation, the stress levels of components change with dose due to irradiation-induced mechanical property changes and irradiation creep and stress relaxation. Swelling is also involved in stress increase for bolt components in PWRs. Fracture toughness is a key property, when the integrity is evaluated by fracture mechanics. Irradiation creep and stress relaxation are discussed in Section 4.3.2 based on recently obtained irradiated data. The current understanding of swelling and fracture toughness is briefly described here.

Swelling

Since swelling is very sensitive to irradiation conditions, such as temperature and dose rate, swelling data are very scarce in PWR-relevant low-dose-rate, low-temperature conditions (10⁻¹⁰–10⁻⁸ dpa/s, >20 dpa, 300–370°C). Figure 29 shows the swelling data from PWR-irradiated SSs and MTR-irradiated SSs at LWR-relevant temperatures [218,340–347]. The data show different trends in swelling evolution with morphology (bubble vs. void), reactor type (PWR vs. FBR) and materials (304 vs. others). A prediction curve of swelling for type 304 SSs was proposed based on low-temperature, FBR-irradiated type 304 SS data [348]; it is an empirical one and a function of dose, dose rate and temperature. Since swelling is known to be lower for higher dose rate, lower temperature and higher CW level [215], MTR-irradiation data at higher dose rates, and data from removed bolts and thimble tubes which were made of CW type 316 SSs and irradiated at temperatures less than 320°C, are not directly applicable to swelling prediction in PWRs. A swelling model calculation for evolution of cavities observed in PWR-irradiated CW type 316 SSs to 53 dpa at 290–340°C showed that swelling might not exceed 0.2% at 100 dpa [340]. However, correlation studies for bridging different condition data are needed based on theoretical modeling and data acquisition.

Figure 29 Swelling data in various SSs irradiated at LWR-relevant temperatures of 290–390ºC

Fracture toughness

It is well known that irradiation causes reduced fracture toughness and that the reduction is related to loss of ductility [235]. Fracture toughness data of LWR-irradiated SSs have been accumulated in the last decade [223,236,349–352]. Data analysis and assessment of failure analysis have been reported [353, 354]. Figure 30 shows the data of fracture toughness in LWR-irradiated SSs. Fracture toughness generally decreased rapidly to 5–10 dpa and the lower bound of the data seems to saturate to ∼50 kJ/m². Current databases have large data scatters, probably due to large variation of irradiation and material conditions, and also due to test methods, and specimen size and shapes. Fracture toughness depends on material and irradiation conditions as well as test methods and conditions, while influences of these parameters in LWR irradiation conditions are not well understood. Fracture toughness has a close link to tensile properties and correlations have been proposed [235,355,356]. Development of mechanistic understanding and database improvement are needed for reliable prediction.

Figure 30 Fracture toughness data in LWR-irradiated SSs tested at ∼300ºC

4.6. Summary of IASCC

Databases on IASCC initiation and CGR in LWR-irradiated SSs have been significantly expanded in the last decade and they now provide phenomenological knowledge on the general response of IASCC initiation and CGR to fluence, stress and water conditions. However, the knowledge is still insufficient for a full description of effects of such parameters on IASCC. While initiation stress and CGR tend to saturate at certain doses, it is unclear where complete saturation occurs and at what saturation level: these might depend on material and water conditions. Based on current databases, empirical trend curves of IASCC initiation and CGR have been developed but are applicable to only limited conditions. To establish a realistic prediction of IASCC in LWR conditions from available databases, relationships between PIE data and in-reactor data, and between LWR-irradiated data and FBR-irradiated data, must be clarified. To overcome limitations of databases on irradiated materials, mechanistic understanding is believed to be essential. Understanding of IASCC mechanisms has been improved in this decade, although exact causes of IASCC are yet uncertain. It has been confirmed that IASCC susceptibility can correlate with material property changes such as RIS, dislocation microstructure and hardening but it cannot be explained by a single static material property change. Current consensus is that IASCC is likely to be caused by combined effects of multiple dynamic processes. Recent studies focused on the role of deformation and oxidation behaviors as essential processes for IASCC. Characteristics of dislocation channels and their interactions with GBs continue to be examined, and the importance of localized deformation on IG cracking has been confirmed. However, changes in GB properties by oxidation and other processes in irradiated SSs and changes under irradiation are not well understood. Degradation of GB cohesion or strength due to irradiation and oxidation needs to be clarified considering effects of RIS and other factors such as hydrogen and helium.

The evaluation of structural integrity of core internals requires prediction methods of stress relaxation, swelling and fracture toughness as well as IASCC initiation and CGR. While empirical prediction curves for these properties have been proposed, further improvement is needed not only by accumulation of irradiated data, but also by mechanistic modeling.

5. Conclusions

In this paper, current phenomenological knowledge and understanding of mechanisms on radiation embrittlement and IASCC in PWRs and BWRs were reviewed, placing emphasis on microscopic material changes and their role in degradation processes. Knowledge of microstructural evolution in LWR irradiation conditions has been significantly improved by recent development of nano-scale analytical techniques such as 3DAP and by applying such techniques to LWR-irradiated materials. In RPV low-alloy steels, solute clusters of Cu, Mn, Ni and Si and dislocation loops were identified as features causing hardening and embrittlement. Mechanism-guided correlations of transition temperature shift have been developed based on improved understanding. In SSs for core internals, deformation and oxidation have been recognized as potential processes causing IASCC initiation and growth. However, full understanding of the mechanisms has not been achieved yet and there are still many issues for improving the understanding of phenomenology and mechanisms of radiation embrittlement and IASCC. Several important issues were pointed out in the review.

Radiation-induced degradation is a complex phenomenon involving multiple physical, chemical and mechanical processes, and it is controlled by nano-scaled features that are very difficult to identify even by state-of-the-art techniques. Furthermore, macroscopic tests and microscopic analyses of neutron-irradiated materials require special technical efforts and facilities, and make it difficult to obtain sufficient and systematic databases. To overcome these difficulties, theoretical modeling and simulation experiments for fundamental processes have been and will be important tools for developing prediction methods in operating LWRs and for mechanistic understanding.