The Effect of He on the Evolution of Radiation-Induced Dislocation Loops near W/Cu Interface

Abstract

In the current work, the distribution behaviors of irradiation-induced dislocation loops near the W-Cu interface (contains a thin W₂C transition layer) under self-interstitial atom diffusion-dominated conditions were investigated based on the comparative experiment of 3 MeV Fe ion and 100 keV He ion irradiation. The size distribution and number density of radiation-induced dislocation loops in both sides of the interface were characterized using Transmission Electron Microscopy with different two-beam conditions. The impact of the phase boundary on the dislocation loop distribution and the influence of He on this mechanism was discussed. The results showed that the phase boundary (PB) has a significant effect on the distribution of radiation-induced dislocation loops. In the Fe-irradiated sample, the proportion of b = 1/2<111> type dislocation loops near the phase boundary on the W side increases significantly, and b = 1/2<110> type dislocation loops dominate on the Cu side. He will significantly affect the loop distribution near the W/Cu phase boundary due to the strong binding of He with vacancies in W, which suppresses the recombination of SIA and vacancies and promotes the formation and growth of interstitial-type dislocations.

1. Introduction

The tungsten–copper divertor is one of the most critical components in future fusion reactors [1]. During operation in a fusion reactor, the divertor will face the mixed effects of high heat loads, high-flux plasma, and high-energy neutron irradiation [2], which will lead to significant microstructure evolution and performance degradation. For example, the formation of radiation-induced dislocation loops [3], voids [4], bubbles [5] in the materials, and the fuzz structure on the surface of tungsten [6], as well as brittleness [7] and swelling [3] after long-term service. One of the most critical structures in the divertor is the W-Cu interface, whose stability is directly correlated with the overall stability of the divertor, making it a key area for investigation. However, irradiation damage near the interface becomes more complex, involving factors such as the additional effects of hydrogen (H) and helium (He) [8], interface stress caused by different elastic modulus and thermal expansion coefficient, the interactions between the interface and irradiation-induced defects [9,10], and so on. The amalgamation of these complex factors complicates further analysis.

In neutron irradiation, the evolution of defects is primarily governed by the interaction between self-interstitial atoms (SIAs) and vacancies [11]. The majority of irradiation-induced SIAs and vacancies recombine, with only a few aggregating and further evolving. The migration of SIAs and vacancies is temperature-dependent, and thus the evolution of irradiation-induced defects is variable with respect to temperature [10,12,13]. Generally, the migration energy of vacancy is higher than that of SIAs [12], and significant vacancy migration occurs only at high temperatures. In the case of tungsten, this temperature corresponds to stage 3 of defect recovery, which occurs in the range of 700–900 K [14,15]. Therefore, using room temperature or low-temperature irradiation can suppress the effects of vacancy migration, simplifying the irradiation damage process to one primarily driven by SIAs.

On the other hand, irradiation damage in the divertor of fusion reactors is inevitably accompanied by interactions between H, He, and defects [8]. Among these interactions, the most significant is with vacancies, where the binding energy of He–vacancy pairs in W exceeds 4 eV, while the binding energy of H–vacancy pairs is about 1.2 eV [16,17]. As a comparison, the binding energies of H, He, and SIA are usually less than 1 eV. Therefore, He plays a more critical role in this process. Previous studies have shown that as the He atoms per million (appm) to displacements per atom (dpa) ratio (He-appm/dpa) increases from the range of 500–2000 to 40,000, the dislocation loop number density decreases, the size increases, and the proportion of interstitial-type loops increases from approximately 55% to 60%. As the He-appm/dpa ratio increases, both the number density and average size of He bubbles generated by He+ ion irradiation also increase, showing a pronounced temperature dependence [18]. This phenomenon is due to the strong interaction between He and vacancies, which fundamentally alters the diffusion behavior of point defects. A similar mechanism also exists in H [19,20]. In contrast to the irradiation of He and H, irradiation damage induced by heavy ions is more governed by SIAs and vacancies. This is because the ion concentration in heavy-ion irradiation is relatively low, and their diffusion rate is also slower.

Furthermore, dislocation loops are inherently complex. For example, in typical bcc metals, common irradiation-induced dislocation loops include b = 1/2<111> and b = <100> types [21,22], while in fcc metals, b = 1/2<110> and b = 1/3<111> are predominant [23]. These dislocation loops differ in their migration ability and interactions with point defects and interfaces [22,24]. Moreover, dislocation loops can be classified into interstitial-type and vacancy-type [25,26], which evolve in significantly different ways. In particular, distinguishing between interstitial-type and vacancy-type dislocation loops in typical Transmission Electron Microscopy (TEM) characterization is challenging. This requires using complex techniques, like inside–outside methods, and carefully comparing the dislocation loops in multiple images [27]. Fortunately, during room-temperature or low-temperature irradiation, defect evolution dominated by SIA migration primarily forms interstitial-type dislocation loops [28,29]. Distinguishing dislocation loops with different Burgers vectors in TEM is relatively simpler.

Based on the above considerations, a simplified experimental method for studying interface irradiation damage has been proposed: performing low-dose heavy-ion irradiation experiments, such as Fe ions, on cross-sectional samples of the interface. This approach shields the effects of H or He while ensuring that the dislocation loops formed are predominantly interstitial type. For comparison, parallel experiments using low-dose He ion irradiation are conducted to investigate the impact of He on dislocation evolution at the interface. In this study, we adopted this experimental approach, using room-temperature 3 MeV Fe ions and 100 keV He ions for irradiation. Through TEM characterization, we investigated the distribution behaviors of irradiation-induced dislocation loops on the W and Cu sides of the W-Cu interface under SIA diffusion-dominated conditions. We also discussed the impact of the W-Cu interface on the dislocation loop distribution and the influence of He on this mechanism.

2. Experimental

The tungsten–copper interface sample used in this experiment was prepared as follows: First, pure tungsten powder (purity > 99%) was shaped into a green body through cold isostatic pressing with a forming pressure of 200 MPa. The green body was then sintered for 6 h in a hydrogen atmosphere at a sintering temperature of 2300 °C. The heating rate during the sintering process was set to 2 °C/min, and the sample was cooled to room temperature using furnace cooling. After sintering, the sample was mechanically polished to remove the surface layer, and then it was hot-rolled at 2200 °C to a 50% reduction in thickness to obtain a tungsten plate. Finally, copper powder (purity > 99%) was heated to 1180 °C (melted) and then uniformly coated onto the preheated tungsten plate. The sample was then rolled again to obtain the W-Cu interface sample. Finally, rolled samples were cut into blocks of 10 mm × 3 mm × 1 mm.

The prepared W-Cu interface sample was polished and then subjected to high-energy Fe ion irradiation. The irradiation areas by Fe ion and He ion both are 10 mm × 3 mm. The Fe ion energy was 3 MeV, with an irradiation dose of 1 × 10¹⁴ ions/cm² and a dose rate of 6.9 × 10⁹ ions/cm²·s. The irradiation temperature was room temperature (25~100 °C). The background pressure was <1 × 10⁻⁵ Pa. The damage profile was estimated using The Stopping and Range of Ions in Matter (SRIM) 2013 software, with the “Quick Kinchin–Pease” mode selected. The dpa was calculated using the following equation:

$$\mathrm{d}\mathrm{p}\mathrm{a}={\displaystyle \frac{0.8}{2E_d}}\ast {\left({\displaystyle \frac{dE}{dx}}\right)}_n\ast {10}^8\ast {\displaystyle \frac{\theta }{N}}$$

(1)

where E is the displacement energy (eV), $\theta$ is the ion dose (ions/cm²), and N is the atomic density (eV/$Å$). Moreover, ${\displaystyle \frac{0.8}{2E_d}}\ast {\left({\displaystyle \frac{dE}{dx}}\right)}_n$ was given in the SRIM output file “VACANCY.txt” [30]. The displacement threshold energy for W and Cu was set as 90 eV and 25 eV [31]. The damage profile and the concentration of Fe ions are shown in Figure 1a. The peak damage on the W side was about 0.09 dpa at about 600 nm, and it was about 0.14 dpa at 1000 nm on the Cu side. The Fe concentration peak was about 0.003% at 800 nm, and it was about 0.002% at 1300 nm on the Cu side. Parallel irradiation experiments were conducted using 100 keV He ions at room temperature (25~50 °C, sensor temperature), with a dose of 5 × 10¹⁵ ions/cm² and a dose rate of 2.8 × 10¹¹ ions/cm²·s. The damage profile, calculated using SRIM 2013 software, is shown in Figure 1b. The peak damage on the W side was about 0.5 dpa at about 200 nm, and it was about 2.2 dpa at about 350 nm on the Cu side. The He concentration peak was about 3.5% at 250 nm, and it was about 2.5% at 400 nm on Cu side.

TEM samples were prepared using a Helios G4 UX DualBeam focused ion beam (FIB) system (Thermo Fisher Scientific, Waltham, MA, USA), using the Lift-Out model [32]: Firstly, a Pt protective layer was deposited using a 2 kV, 6.4 nA electron beam, and preliminary processing was performed using 30 keV Ge ions. Then, we gradually reduced the current from 1 nA to 0.1 nA for thinning. Finally, the FIB samples were cleaned using low-energy (0.5–5 keV) Ga ions to reduce FIB damage. The FIB sample was 4 μm × 4 μm in size. TEM observations were carried out using an FEI Titan Double Spherical Aberration Corrected TEM operated at 200 kV, coupled with an energy-dispersive X-ray spectroscopy (EDS). Samples for scanning electron microscope (SEM) characterization were prepared by mechanical cutting, grinding, and polishing. SEM characterization was carried out using Hitachi SU-8020 SEM (Tokyo, Japan) operated at 5 kV.

3. Results and Discussion

3.1. Microstructure Analysis

Figure 2 shows the SEM image, TEM bright field (BF) image, selected area electron diffraction (SAED) patterns, and elemental distribution from energy-dispersive spectrometer (EDS) of the W-Cu interface sample. Figure 2a,b display the SEM images of the W-Cu interface sample. The EDS elemental distribution indicated that there was no solid solution between W and Cu. Figure 2c shows the TEM BF image of the FIB sample. Due to the significant difference in hardness and sputter resistance between W and Cu, the copper side is easily damaged during the thinning process. Fortunately, the irradiated area of this sample remains largely intact. The TEM images show that a distinct interface exists between W and Cu, with no mutual solubility. This observation is consistent with the expected equilibrium phase diagram. The phases of bcc-W and fcc-Cu were identified using SAED, as shown in Figure 2e,f. However, at the interface, a different SAED pattern was observed, which did not correspond to any known tungsten or copper phase. Further diffraction analysis (Figure 2g) revealed that this transition layer has a W₂C structure [33]. The thickness of this transition layer is approximately several hundred nanometers, and it is likely formed by the accumulation of carbon impurities from the tungsten substrate or heating wires during the copper powder coating process. Therefore, the W-Cu interface consists of three distinct parts: the W matrix, the W₂C transition layer, and the Cu matrix. The effects of the W/W₂C interface on the tungsten side and the W₂C/Cu interface on the copper side will be investigated. It should be noted that the carbon in the W₂C transition layer observed in this experiment was not intentionally introduced. This implies that such a phenomenon may persist when applying the same fabrication techniques to divertor components. Therefore, it is essential to investigate the irradiation response of this interface to guide future optimization of divertor manufacturing processes.

3.2. Radiation-Induced Dislocation Loop After Fe Irradiation

Figure 3 shows the TEM BF images of radiation-induced dislocation loops on the W side after Fe ion irradiation. The TEM images were captured under two-beam conditions to highlight the contrast of dislocation loops. Two different two-beam conditions were applied, with the g vectors being (002) and (112), respectively. When observing dislocations in bcc crystals, the g = (110) or g = (002) vectors are typically chosen. However, due to the non-ideal sample orientation, it was not convenient to obtain g = (110), so g = (112) was used as a substitute. To further analyze the impact of the W/W₂C phase boundary on radiation-induced dislocation loops, we divided the sample into several regions using dashed lines parallel to the phase boundary. The distance of each zone to the phase boundary was 200 nm, 250 nm, and 300 nm, respectively. The dislocation size distribution and number density in each region were then statistically evaluated, as shown in Figure 4 and Figure 5. The thickness of each sample region was determined using convergent beam diffraction.

With a g vector of (002), the average dislocation loop size in Zone 1 was approximately 1.9 ± 0.5 nm, and the loop number density was about 6.2 × 10²² m⁻³. In Zone 2, the average dislocation loop size was approximately 1.9 ± 0.6 nm, with a lower loop number density of 2.6 × 10²² m⁻³. In Zone 3, the average loop size was similar to that in Zone 2 (1.9 ± 0.5 nm), with a number density of 2.1 × 10²² m⁻³, which was nearly identical to that in Zone 2. Clearly, the mean dislocation loop size in different zones was similar, with g = (002), but the loop density in Zone 1 was significantly higher. With a g vector of (112), a similar trend in dislocation loop size was observed, with mean loop sizes of 2.1 ± 0.7 nm, 2.0 ± 0.5 nm, and 2.2 ± 0.8 nm in Zones 1, 2, and 3, respectively. However, a significantly larger loop density was also observed, and the loop density with g = (112) was similar across zones. The loop density in Zones 1, 2, and 3 was 10.1 × 10²² m⁻³, 7.5 × 10²² m⁻³, and 9.7 × 10²² m⁻³, respectively.

Typically, in bcc metals, the primary radiation-induced dislocation loops are of the b = <100> and b = 1/2<111> types [22,34,35], where b represents the Burgers vector. If we assume that the Burgers vector of radiation-induced dislocation loops is uniformly distributed along equivalent crystallographic directions, meaning that the proportion of b = [001] and b = [010] dislocation loops is equal among all b = <100> dislocation loops. Then, we can estimate the proportion of different dislocation types by examining the dislocation density observed at different g vectors. At g = (002), b = <100> type dislocations have a 2/3 probability of diffraction extinction and, therefore, cannot be observed, while b = 1/2<111> dislocations cannot undergo diffraction extinction (i.e., they can all be observed). Similarly, at g = (112), b = <100> type dislocations do not undergo diffraction extinction, while b = 1/2<111> dislocations have a 3/4 probability of being observed. Therefore, we can estimate the number density of dislocations with different Burgers vectors near the interface using the following formula:

$$\left\{\begin{array}{c}{\rho }_{\left(002\right)}={\displaystyle \frac{1}{3}}{\rho }_{100}+{\rho }_{111}\\ {\rho }_{\left(11\overline{2}\right)}={\rho }_{100}+{\displaystyle \frac{3}{4}}{\rho }_{111}\end{array}\right.$$

(2)

where ${\rho }_{\left(002\right)}$ and ${\rho }_{\left(112\right)}$ are the loop density observed with g = (002) and g = (112); and ${\rho }_{100}$ and ${\rho }_{111}$ are the number densities of b = <100> and b = 1/2<111> type dislocations.

Table 1. The number density of different types of loops.

	ρ100(1022m−3)	ρ111(1022m−3)	ρ111/ρ100
Zone 1	7.2	3.8	0.53
Zone 2	7.4	0.1	0.01
Zone 3	10.7	/ *	/ *

* Due to negative values, data are unavailable.

presents the calculated dislocation loop density for different types of loops. In Zone 1, the number density of b = <100> type dislocation loops was about 7.2 × 10²² m⁻³, while that of b = 1/2<111> type dislocation loops was approximately half of the b = <100> type. In Zone 2, the number density of b = <100> type dislocation loops was similar (about 7.4 × 10²² m⁻³), while that of b = 1/2<111> type dislocation loops was nearly zero (only 0.1 × 10²² m⁻³), indicating that nearly all loops were of the b = <100> type. In Zone 3, a negative density of b = 1/2<111> dislocation loops was calculated, which could be attributed to measurement errors in the TEM images or to the uneven distribution of dislocation loops along equivalent crystallographic orientations. However, this result still indicates that the b = <100> dislocation loops dominate. A detailed discussion about the negative value can be found in Section 3.6. Therefore, in Zone 1, near the interface, the proportion of b = 1/2<111> type dislocation loops is relatively higher, whereas in Zones 2 and 3, further from the interface, b = <100> type dislocation loops dominate.

Figure 6 presents the TEM BF images of radiation-induced dislocation loops on the Cu side after Fe ion irradiation. The TEM images were acquired under two-beam conditions, with g = (002) and g = (111), which are commonly used g vectors for fcc crystals. Similar to the W side, the TEM images were divided into several regions, and the dislocation loop size distribution and number density in each region were statistically evaluated, as shown in Figure 7 and Figure 8. The distance of each zone to the phase boundary was 400 nm, 450 nm, 500 nm, and 550 nm, respectively. With a g vector of (002), the average dislocation loop size across all zones was similar (about 2.3–3.0 nm), while the number density varied. The number density in Zone 1 was approximately 13.7 × 10²² m⁻³, decreasing to 10.4 × 10²² m⁻³ in Zone 2, and further decreasing to 7.9 × 10²² m⁻³ in Zone 3. In Zone 4, the number density was about 10.4 × 10²² m⁻³, similar to that in Zone 2. With a g vector of (111), the loop size remained similar, ranging from 2.9 to 3.4 nm. The number density exhibited a similar trend to that observed with g = (002). In Zone 1, the loop density was about 6.7 × 10²² m⁻³, decreasing to 5.0 × 10²² m⁻³ in Zone 2 and 5.2 × 10²² m⁻³ in Zone 3. Zone 4 shows a slight increase in density (about 6.0 × 10²² m⁻³), remaining at the same level as Zone 2.

Table 2. Calculated dislocation loop density for different types of loops on the Cu side.

	ρ110(1022m−3)	ρ111(1022m−3)
Zone 1	42.0	/ *
Zone 2	32.2	/ *
Zone 3	16.6	/ *
Zone 4	26.4	/ *

* Due to negative values, data are unavailable.

presents the calculated dislocation loop density for different types of loops on the Cu side. The primary radiation-induced dislocation loops are of the b = 1/2<110> and b = 1/3<111> types [23]. The calculation method is the same as that used for the W side, with the only difference being the varying probabilities of dislocation extinction under different g vectors. The calculation formula is shown in Equation (2).

$$\left\{\begin{array}{c}{\mathsf{\rho }}_{\left(200\right)}={\displaystyle \frac{2}{3}}{\mathsf{\rho }}_{110}+{\mathsf{\rho }}_{111}\\ {\mathsf{\rho }}_{\left(1\overline{11}\right)}={\displaystyle \frac{1}{2}}{\mathsf{\rho }}_{110}+{\mathsf{\rho }}_{111}\end{array}\right.$$

(3)

where ${\mathsf{\rho }}_{\left(002\right)}$ and ${\rho }_{\left(111\right)}$ are the loop density observed with g = (002) and g = (111); and ${\rho }_{110}$ and ${\rho }_{111}$ are the number densities of b = 1/2<110> and b = 1/3<111> type dislocations. However, unlike the W side, negative dislocation densities were observed in all four regions on Cu. About this negative value, a detailed discussion can be found in Section 3.6. The cause of this issue may be that the Cu side of the W-Cu interface sample is more susceptible to damage during preparation, resulting in greater FIB damage. Since the interface sample could not be cleaned using conventional flash polishing to remove FIB damage, we employed low-energy Ga ions (500 eV) for cleaning. However, this process may still involve energy levels that are too high for Cu. Nevertheless, as shown in Figure 8, the dislocation density is significantly higher at g = (200) than at g = (111), indicating that the dislocations in Cu are predominantly of the b = 1/2<110> type.

3.3. Radiation-Induced Dislocation Loop After He Irradiation

Figure 9 shows the TEM BF images of radiation-induced dislocation loops on the W side after He ion irradiation. The TEM images were acquired under two-beam conditions, with g = (110). Unfortunately, at this orientation, the sample is close to the [111] zone axis of the W material, resulting in only three equivalent g = (110) vectors being available. The TEM images were divided into several regions, and the dislocation loop size distribution and number density in each region were statistically evaluated, as shown in Figure 10. The distances of each zone from the phase boundary were 25 nm, 75 nm, and 125 nm, respectively. The average dislocation loop size in Zone 1 was measured to be 2.4 ± 0.8 nm, and it slightly increased to about 3.0 nm in Zones 2 and 3. The trend of loop size was similar to the Fe-irradiated sample, where the closest zone (Zone 1) showed a slightly smaller loop size. The loop density in Zone 1 was measured as 1.3 × 10²³ m⁻³, and it slightly increased to 1.5 × 10²³ m⁻³ and 1.6 × 10²³ m⁻³ in Zone 2 and Zone 3. This trend was completely different from that observed in the Fe-irradiated sample, where the closest zone (Zone 1) showed the highest number density. However, in the He-irradiated sample, it showed the lowest dislocation density.

Figure 11 presents the TEM BF images of radiation-induced dislocation loops on the Cu side after He ion irradiation. The TEM images were acquired under two-beam conditions, with g = (002) and g = (111). The TEM images were divided into two regions, and the dislocation loop size distribution and number density in each region were statistically evaluated, as shown in Figure 12. The distance of each zone to the phase boundary was 10 nm and 80 nm, respectively. With a g vector of (002), the average dislocation loop size in Zone 1 was 3.1 ± 1.0 nm, and it was almost identical in Zone 2. The loop number density was measured as 5.8 × 10²² m⁻³ in Zone 1, and it decreased to 3.8 × 10²² m⁻³ in Zone 2. With a g vector of (111), the loop size remained similar, ranging from 2.4 to 2.7 nm. The loop density in Zone 1 was 5.1 × 10²² m⁻³, and it decreased to 1.7 × 10²² m⁻³ in Zone 2. It should be noted that the loop density in Zone 2 is too low to observe a sufficient number of loops; thus, the loop size distribution in Zone 2 appears somewhat irregular. It can be observed that the trend of dislocation density change on the Cu side is similar to that in the Fe-irradiated sample.

3.4. The Effect of the W-Cu Phase Boundary

Considering the very low Fe concentration in the Fe-irradiated sample, it can be assumed that the radiation-induced defects are primarily influenced by the evolution of self-interstitial atoms (SIAs) and vacancies, with the additional solute atoms introduced by the incident ions having a negligible effect on the evolution of radiation defects. Moreover, the vacancy migration energy in W is quite high (1.65 eV) [10], which means that vacancies hardly migrate during irradiation at room temperature. As a result, the evolution of dislocation loops is primarily governed by the migration of SIAs. As shown in Section 3.2, on the W side, the total dislocation density near the phase boundary increases, with a significant increase in the density of b = 1/2<111> dislocation loops, while the density of b = <100> dislocation loops does not seem to be significantly affected by the interface.

Generally, the high sink strength of the interface leads to the accumulation of SIAs and vacancies at the interface, promoting their annihilation. As a result, defect clusters, dislocation loops, and voids are expected to decrease in regions close to the interface. However, the opposite phenomenon was observed in this experiment. A reasonable explanation for this phenomenon is that during irradiation, the phase boundary acts as a defect trap, absorbing SIAs and SIA clusters. Previous studies have shown that almost all dislocation loops during room-temperature irradiation are interstitial-type [26,36]. Therefore, dislocation loops are attracted to the interface, leading to their movement towards it, such as one-dimensional migration of dislocation loops [25,37]. For free surfaces, dislocation loops will overflow due to surface effects, leading to a reduction in dislocation density. For high-temperature irradiation, the simultaneous absorption of vacancies near defect traps leads to more SIAs being annihilated by vacancies, also resulting in a reduction in dislocation density at the interface. However, in the case of room-temperature irradiation, due to the minimal migration of vacancies [12,13], interstitial-type dislocation loops migrating towards the phase boundary cannot be quickly annihilated, leading to an increase in dislocation density.

Among the two common types of radiation-induced dislocation loops in W, b = 1/2<111> dislocation loops have a lower migration energy and are more likely to slip [37]. As a result, more b = 1/2<111> dislocation loops accumulate near the phase boundary, leading to an increase in their proportion and the overall dislocation density. On the Cu side, a similar mechanism is at play, and the greater number of mobile b = 1/2<110> dislocation loops leads to a slight increase in dislocation density near the interface.

3.5. The Effect of He Atoms

However, the situation is different in the He-irradiated sample. As described in Section 3.3, the dislocation loop density near the interface is relatively lower on the W side, while it is higher near the critical interface on the Cu side. This result indicates that He significantly affects the formation and distribution of radiation-induced dislocation loops. This difference can be attributed to the additional bias introduced by He. Unlike vacancies, He has lower migration energy in W. Simulation results show a migration energy of only 0.06 eV, while experimental values range from 0.24 to 0.32 eV [38]. The binding energy of the He–vacancy pair is approximately 0.8 eV in Cu [39], whereas it is around 4.57 eV for W [40]. Therefore, the interaction between He and vacancies is stronger in W, making the He–vacancy pair more stable. A high He-appm/dpa ratio promotes the formation of a He–vacancy pair, consuming a large number of irradiation-induced vacancies and reducing the probability of recombination between SIAs and vacancies. The remaining SIAs are absorbed by the irradiation-induced dislocation loops, leading to a high density or large size for dislocation loops [16,18,41]. Thus, the key point is the He distribution.

Furthermore, due to the higher binding energy of the He–vacancy pair in W, He on the Cu side can cross the interface and transform into the W side, leading a migration towards the W side. Thus, the W side should have more He, and this is the reason why a large loop size and higher loop density were observed in Zone 2 and Zone 3 of the W side. In contrast, the He–vacancy pair binding energy is lower on the Cu side, and its effect is less pronounced. Moreover, He migration further weakens this mechanism. Consequently, the dislocation evolution is similar to that in the Fe-irradiated sample, with an increased dislocation density near the interface.

It is important to emphasize that, during this process, vacancies do not migrate, and the overall He concentration remains low, making it too insufficient to form He bubbles. Under high-temperature irradiation or with higher doses of He irradiation, the formation of He bubbles will alter the dislocation evolution mechanism, and the distribution should be different.

3.6. The Analysis of the Negative Dislocation Loop Density

As described above, Table 1 and Table 2 show some negative dislocation loop densities, which are physically impossible. Therefore, it is necessary to analyze the reasons that led to this phenomenon. Let us take bcc-W as an example for the discussion, in which the radiation-induced dislocation loops have two types: b = <100> type and b = 1/2<111> type. As we all know, the <100> direction has three equivalent crystallographic directions in cubic crystal, and the <111> direction has four. Correspondingly, the Burgers vector of the b = <100> type dislocation loop has three equivalent types, and the b =1/2<111> type has four. In an ideal situation, the distribution of b on the equivalent crystallographic directions is uniform. For example, there is a 1/3 probability that obtains a b = [100] loop in all b = <100> type dislocation loops, and the probability of b = [010] and [001] is also 1/3. In the b = 1/2<111> type loop, it is 1/4.

For a TEM observation under two beam conditions with a g vector, if the b vector of loop is perpendicular to g, g·b = 0, the loop is invisible. For g = (002), loops with b = [010] and [001] are invisible, and for g = (112), none of them is invisible. In an ideal situation, 2/3 of b = <100> type dislocation loop is invisible in g = (002), and all are visible in g = (112). This is the base of Equation (2). Obviously, in this case, all dislocation densities should be non-negative. However, the b vector of dislocations is not always uniformly distributed in each equivalent crystal direction. These may be due to the image force from the free surface or the sink strength from other defect traps.

For example, we have created a hypothetical dislocation distribution, as shown in

Table 3. A hypothetical dislocation distribution in W and its calculated results with Equation (2).

Item	b or g	Density (1021 m−3)	b or g	Density (1021 m−3)
Hypothetical dislocation distribution	b = [100]	3.00	b = 1/2 [111]	0.10
	b = [010]	3.00	b = 1/2 [11$\overline{1}$]	0.10
	b = [001]	1.00	b = 1/2 [1$\overline{1}$1]	0.10
			b = 1/2 [$\overline{1}$11]	0.10
	b = <100>	7.00	b = 1/2<111>	0.40
Measured loop density with different g vector	g = (002)	1.40	g = (112)	7.30
Calculated loop density	b = <100>	8.33	b = <111>	−1.38

, in which loops are mainly b = <100> type (about 95%), and the loops have preferred orientation; the b = [001] loop is much lower than that of b = [100] and b = [010]. This distribution will lead to a much lower density when observed with g = (002) and finally lead to a negative calculated value in b = 1/2<111> type loops. Thus, the negative b = 1/2<111> type loop density in Table 2 indicated that the radiation-induced dislocation loops in bcc-W are mainly of the b = <100> type, and the distribution of the Burgers vector has preferred orientation. Similarly, the negative loop density of b = 1/3<111> type loops in fcc-Cu suggests similar results: the radiation-induced dislocation loops in fcc-Cu are mainly b = 1/2<110> type, and they have preferred orientation.

4. Conclusions

The present study investigated the effect of the W/Cu phase boundary on defect evolution by characterizing dislocation loops induced by Fe ion and He ion irradiation at room temperature. Based on the results and analysis, the following conclusions can be drawn:

(1) In the Fe-irradiated sample, the W/Cu phase boundary has a significant effect on the distribution of radiation-induced dislocation loops. On the W side, the proportion of b = 1/2<111> type dislocation loops near the phase boundary increases significantly, with an overall increase in dislocation number density, while b = <100> type dislocations are almost unaffected. On the Cu side, b = 1/2<110> type dislocation loops near the phase boundary are dominated.

(2) In the He-irradiated sample, the influence of the W/Cu phase boundary is significantly affected by He. On the W side, the dislocation loop number density near the phase boundary decreases, while on the Cu side, the opposite trend is observed, with an increase in the number density near the boundary. The cause of this phenomenon can be attributed to the strong binding of He with vacancies in W, which suppresses the recombination of SIA and vacancies, thereby promoting the formation and growth of interstitial-type dislocations.