Acta Materialia 160 (2018) 211e223

Contents lists available at ScienceDirect

Acta Materialia

journal homepage: www.elsevier.com/locate/actamat

Full length article Defect-interface interactions in irradiated Cu/Ag nanocomposites

* Min Wang a, Irene J. Beyerlein b, Jian Zhang c, Wei-Zhong Han a, a Center for Advancing Materials Performance from the Nanoscale (CAMP-Nano), State Key Laboratory for Mechanical Behavior of Materials, Xi'an Jiaotong University, Xi'an, 710049, China b Mechanical Engineering Department, Materials Department, University of California, Santa Barbara, CA, 93106-5070, USA c College of Energy, Xiamen University, Xiamen, 361005, China article info abstract

Article history: In this work, we employ transmission electron microscopy and helium ion irradiation to study the Received 20 July 2018 response of biphase interfaces to radiation induced point defect fluxes from the two adjoining phases. Received in revised form Analysis of interface-affected defect accumulation was carried out over a wide range of radiation damage 23 August 2018 levels from near zero displacement per atom (dpa) to 16 dpa and helium concentrations of 0 at.% to Accepted 3 September 2018 8 at.%. Results show a strong interface density dependence in which Cu/Ag interfaces in the nanolayered Available online 5 September 2018 regions spaced <500 nm were remarkably microstructural stable over the entire range without accu- mulating micro-scale defects, while those spaced >1 mm apart were destroyed. We report the concom- Keywords: Interface itant development of a bubble-free zone in Cu that was independent of defect levels and interface- fi Radiation defects contacting bubbles zone in Ag. This nding is explained by bias segregation to the interface of in- Helium bubbles terstitials from Ag and vacancies to misfit dislocation nodes in the interface from Cu. The point defect Vacancy pump transfer across the interface can be explained by the spatial variation in interface pressure within the Dislocations interface and gradient in pressure across the interface, both originating from the lattice mismatch and surface energy difference between the two crystals. © 2018 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction Interface engineering is becoming a recognized and widely adopted method for designing radiation tolerant materials [14e22]. Structural materials used in nuclear reactors are subjected to a Under the same service conditions, material radiation tolerance can high level of irradiation, an extreme environment that over time be dramatically enhanced by introducing a large number of in- causes defects to form and accumulate inside the material and terfaces, either homophase or biphase, into its microstructure. The eventually lead to internal damage [1e4]. Radiation-induced de- basic strategy exploits the idea that interfaces are efficient defect fects are first produced in the form of atomic scale point defects, “sinks”; that is, they are preferable regions within the material, vacancies and interstitials, which evolve into larger point defect where the interstitial and vacancy combination rates can be clusters, such as dislocation loops, voids, and bubbles [1e4]. significantly enhanced relative to the adjoining bulk crystals Accumulation of these radiation defects degrades mechanical per- [14e22]. The sink properties of free surfaces [23e26], grain formance typically in the form of significant increases in hardening boundaries [27e34] and interfaces [16,35e37] have been studied and embrittlement [5e10]. In order to reduce radiation damage, the extensively. By microscopic quantification of the width of defect- key is to enhance the recombination/annihilation rate of radiation free-zone formed along these interfaces, the sink efficiency of defects as soon as they are produced [10e18]. It is well known that different interfaces or grain boundaries can be measured, and the recombination processes of radiation defects are influenced by therefore, they can be ranked [16,33]. It becomes clear from such a number of factors, such as, radiation dose, the nature of energized analyses that not all these interfaces exhibit similar sink effi- particles, radiation flux, radiation temperature, the diffusion and ciencies, and that the differences can be rationalized based on migration behavior of defects, the intrinsic properties and micro- differences in their atomic structures and interaction energies with structures of materials, etc [1e4]. point defects. In addition to selecting or engineering the optimal interface structure, in order to achieve high radiation tolerance, it is also of high importance to increase the volume fraction of in- * Corresponding author. terfaces. Several methods have been used to fabricate interface- E-mail address: [email protected] (W.-Z. Han). https://doi.org/10.1016/j.actamat.2018.09.003 1359-6454/© 2018 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. 212 M. Wang et al. / Acta Materialia 160 (2018) 211e223 dominated materials, such as physical vapor deposition [38], interface froms a bubble-free zone in Cu and interface-contacting accumulative roll bonding [39] and sintering [40] etc. With these bubbles zone in Ag. We rationalize this interface-driven phenom- techniques, the radiation tolerance of multilayers consisting of enon as evidence of point defect transfer across the interface, either two alternating immiscible or miscible phases have been driven by spatial variation in interface pressure within the interface widely studied either by in situ or ex situ radiation techniques. The and change in compression/tensile pressure across the interface high volume fraction of interfaces gives rise to superior irradiation from Cu to Ag. tolerance and mechanical properties [17,22]. With this success, in the past decade, a large number of in- 2. Experimental design and procedures vestigations have been dedicated to the radiation behavior of face- centered cubic (FCC)/body-centered cubic (BCC) nanolaminates Bulk Cu/Ag nanocomposites were prepared through quenching [41e43]. BCC metals usually have higher melting temperature and of a molten Cu/Ag alloy with a eutectic composition, i.e., a 2:3 better radiation resistances than FCC metals, and therefore, the atomic ratio. The starting materials were high-purity Cu (99.99%) FCC/BCC nanolaminates demonstrate greater radiation tolerance and Ag (99.99%). They were first placed into an Al2O3 tube, and then and thermal stability than its FCC component. As such, the contri- sealed with pure Argon to a pressure slightly higher than 1 atm. To bution of BCC phase on radiation resistance of BCC/FCC interface is melt and mix them, the tube was heated slowly to 1150 Cina significant, and much experimental evidence exist demonstrating vertical furnace. After holding at 1150 C for 1 h, the tube containing that these interfaces can heal the radiation defects in FCC layers the liquid mixture of Cu and Ag was immediately quenched in a [16]. In addition, the defect dynamics in BCC and FCC metals are water tank. The dimensions of the ingots produced are 10 mm in markedly different. The mobility of defects in an FCC phase are diameter and 40 mm in length. Two types of samples were made higher than that in a BCC phase [41e43]. Consequently, the FCC/BCC from the ingots: “bulk” disk samples and transmission electron interface under such radiation conditions actually only interact microscopy (TEM) samples. The bulk samples were cut from the with a defect flux from the FCC layer. This may be one explanation ingot by spark-cutting technique, then ground and polished to a why the FCC/BCC interfaces show much better radiation resistance mirror surface. The TEM foils were cut from the ingots and ground and stability than their grain boundaries. However, how a bimetal to about 50 mm in thickness. These thinned TEM foils were dimpled interface would respond to the defect fluxes from the two dissim- by an M200 Dimpling Grinder, and further thinned using an ilar, adjoining crystals has not been given as much study to date. In Ar þ ion milling on an M1050 TEM Mill, operated at 3.5 kV and with order to achieve such a model system, FCC/FCC nanolaminates is a final polishing beam angle of 4. good choice. Several FCC metals have a similar melting tempera- Helium implantation was performed on these two types of tures around 1000 C, such as Cu, Ag and Au, etc., and thus, would samples using helium ions with an energy of 400 keV at 400 Cby have similar point defect dynamics in response to radiation. Hence, using a NEC 400kV Implanter. The ion fluence is 2 Â 1017 ions/cm2, biphase FCC/FCC interfaces found in these nanolaminates made and the implantation lasted for 200 min (corresponding to a flux rate with these combinations of materials would experience similar of 1.67 Â 1013 ions/cm2/s). The helium beam was tuned to implant defect fluxes from both sides. In this case, the radiation behavior of near perpendicular (7 off to avoid channeling effect) to the top FCC/FCC bimetal interface should be different from the FCC/BCC surface of the bulk samples and to the surface of the TEM foils, as interfaces and from homo-phase, i.e., grain boundaries, studied in illustrated in Fig. 1(a) and (b). The radiation damage (in units of dpa) previous works. The irradiation response of biphase interfaces in for each phase and the helium concentration as a function of depth such cases remains largely unexplored [44e46]. from the top surface can be estimated by Monte Carlo simulation, In this study, we aim to obtain a better understanding of the role specifically using the Stopping and Range of Ions in Solids (SRIM) FCC/FCC biphase interfaces play in radiation damage development [52]. Fig. 1 shows the depth variation in radiation damage (dpa) for when exposed to comparable radiation defect fluxes from both Cu (the magenta line) and Ag (the red line) and the helium con- sides. Further, by changing the concentration of radiation damage centration (black line) when using an average threshold displace- and helium concentration, we pursue a second objective to probe ment energy of 29 eV for Cu and 39 eV for Ag, respectively [2]. how defect/interface interactions change with amounts of point To examine over a broad length scale range from the micron to defects created in each adjoining crystal and with different types of the atomic scale, the microstructures of Cu/Ag nanocomposites radiation defects. To this end, we choose Cu/Ag nanocomposites as before and after helium implantation and the defect structures a model system. Two types of interfaces, with nearly equivalent after implantation, we employed a suite of microscopy techniques, interface energies, prevail in this system: the cube-on-cube inter- including optical microscopy, scanning electron microscopy (SEM, face ([100]Cu//[100]Ag,[010]Cu//[010]Ag and (111)Cu//(111)Ag) and SU6600) and transmission electron microscopy (TEM, a JEOL hetero-twin interface ([111]Cu//[111]Ag,[211]Cu//[112]Ag and 2100F). First, in order to study radiation defects across the whole (112)Cu//(313)Ag)[47e49]. The bulk Cu/Ag nanocomposites can be range of implantation, cross-section TEM-transparent samples, easily fabricated via quenching of a eutectic Cu/Ag alloy from the oriented parallel to the implantation direction, were lifted out us- melt [50,51]. Helium ion implantation is used to radiate the Cu/Ag ing a focused ion beam (FIB) manipulator on the bulk sample. nanolaminates in order to probe the defect-interface interactions in Second, to characterize the radiation defects as a function of helium a wide range of radiation damage (displacement per atom (dpa)) in concentration, TEM foils were cut at different locations using FIB, as one piece of sample. In these experiments, it is not possible to shown in Fig. 1(b). The ion beam current was reduced to 9.7 pA for distinguish whether the cavity is a helium bubble or a void. As a the final thinning to minimize possible FIB-induced surface dam- reference, a second set of experiments is performed in which he- age. The TEM sample at a selected implantation depth D corre- lium irradiation is applied to Cu/Ag thin foils (with thickness of ~ sponds to a fixed radiation damage and helium concentration. The 100 nm). The high surface area enables the helium to leave the foil, thicknesses of the TEM samples were measured by SEM and giving the opportunity to study the effects of Cu/Ag interfaces on determined to be about 100 nm, and therefore, the sample at a managing the vacancies and interstitials without helium. It is found depth D means that it spans depths from D to D þ 100 nm. To that the Cu/Ag interfaces in the nanolayered regions were examine radiation-induced defect agglomeration and defect- remarkably microstructural stable over the entire range of defect interface interaction at each D, a two-beam diffraction contrast levels without accumulating micro-scale defects. Independent of condition, as well as defocus imaging were employed. At least two the radiation damage and helium concentration, the Cu/Ag TEM foils were cut for defect observation at each depth. M. Wang et al. / Acta Materialia 160 (2018) 211e223 213

Fig. 1. Schematic of helium implantation in TEM foils and bulk samples using a helium ion energy of 400 keV to a fluence of 2 Â 1017 ions/cm2 at 400 C. Helium beam was tuned to implant near perpendicular (7-off) to the surface of TEM foils and the top surface of bulk samples. (a) TEM foils and corresponding helium concentration and damage distribution estimated by SRIM. (b) Bulk implantation and corresponding helium density and damage distribution estimated by SRIM. The inside slices represent TEM transparent samples that were cut at certain radiation depths to study the defect-interface interactions. The radiation depth D means the sample is cut from a depth of D to Dþ100 nm.

3. Results 300e800 nm in size) embedded in an Ag phase. An image of the microstructure of non-layered region can be found in the following 3.1. Microstructure of Cu/Ag nanocomposites section (Fig. 4). Fig. 2(b) displays an enlarged image of the nanolayered struc- Due to negligible mutual solubility of Cu and Ag [53], the tures. The white phase is Ag and the black one is Cu. Previous microstructure of the eutectic Cu/Ag alloy separates into two studies have shown that eutectic Cu/Ag alloy has two types of in- alternative phases and forms a nanolayered structure with a high terfaces: cube-on-cube interfaces and hetero-twin interfaces, and density of Cu/Ag interfaces. The as-made material developed two MD simulation indicates that their interfacial energies correspond parts with very different phase morphologies: a nanolayered part to local minima and are nearly equal [48]. In surveying many and a non-layered part. Fig. 2(a) shows the nanolayered structure of distinct areas of the sample, we found that the interfaces in the this material, with layer thicknesses in the range of submicrons, is nanolayered region are predominantly hetero-twin type. Fig. 2(c) maintained for several microns. The average thickness of the Ag displays the bright-field TEM micrograph of a hetero-twin Cu/Ag and Cu nanolayers are 150 nm and 50 nm, respectively. The non- interface according to the selected area diffraction pattern (SADP) layered portion consists of submicron Cu particles ( ~ in the inset. The hetero-twin interface has an orientation

Fig. 2. Microstructure of the as-received bulk Cu/Ag nanocomposites after water-quenching. (a) and (b) SEM micrographs at different magnifications; the black layers are Cu and the white layers are Ag. (c) Bright-field TEM micrograph and SADP of hetero-twin Cu/Ag interfaces. (d) A typical high-resolution TEM image of part of the hetero-twin Cu/Ag interface.

The zone axis is [110], and the interface plane is (111)Cu//(111)Ag. The lattices of the Cu and Ag layers exhibit a twin-like symmetry along the interface plane. 214 M. Wang et al. / Acta Materialia 160 (2018) 211e223 relationship of [110]Cu//[110]Ag and (111)Cu//(111)Ag. The tiny black (dislocation loops) were detected. Fig. 3(b) and (c) show enlarged dots in these TEM images are very likely FIB-induced defects, and images of these areas. First, many stacking fault tetrahedra (SFTs) therefore, different from the helium radiation produced defects we have formed in Ag layers. These can be recognized in Fig. 3(b) and study in detail below. (c) by their triangular shape. Second, a few isolated voids have To fully characterize the interface, we analyzed several in- formed in the Ag layers, which are also marked in Fig. 3(b) and (c). terfaces via high-resolution TEM (HR-TEM). Fig. 2(d) shows a Since the helium concentration is zero, these cavities are voids, not typical HR-TEM image of a hetero-twin Cu/Ag interface in the helium bubbles. The size of these SFTs and voids range from several nanolayered material. The zone axis is [1 1 0]. With this higher nanometers to about 12 nm. In the Cu layers numerous faceted resolution, it is easy to see that the interface plane is (111)Cu// voids bounded by low energy planes of {111} and {100} have (111)Ag. An analysis of many parts of the sample finds that the Cu formed, as shown in Fig. 3(d). Unlike the Ag layer, no SFTs develop and Ag interfaces in this material consistently exhibit a twin-like in the Cu layer. The SAED pattern in Fig. 3(b) indicates that interface symmetry about the Cu/Ag interface plane, and thus, are hetero- crystallographic character of the hetero-twin Cu/Ag interface is twin interfaces. The atomic structure of this part of the Cu/Ag preserved, even after such a high level of helium implantation. interface is in accord with the previous TEM reports of hetero-twin In order to study the radiation defects in Cu/Ag interface, the interfaces in rapidly quenched eutectic Cu/Ag [54,55]. The differ- TEM foil was tilted off the axis about 15 to incline the interface ence in lattice parameter of Cu and Ag leads to about 12.2% misfit plane at an angle with respect to the electron beam. Some typical strain on interface. To relieve the misfit strain, a set of three non- TEM images are shown in Fig. 3(e) and (f). Under this diffraction parallel misfit dislocation arrays are expected to form within this contrast (under focus of 1.5 mm), only the nanovoids are visible hetero-twin interface [45]. From the same calculation it is expected within the Ag and Cu layers. The interesting observation is the to be planar. Analysis of the interfaces in the present study finds presence of many voids, about 1 nm in diameter, as marked by the that these interfaces contain not only a periodic array of misfit red boxes and arrows in Fig. 3(f). dislocations, but also many steps of height two atomic steps, making it slightly non-planar (Fig. 2(d)). These extrinsic defects likely develop from the nonequilibrium solidification due to the 3.3. Radiation behavior of bulk Cu/Ag nanocomposites rapid cooling by water quenching. Also using TEM, a survey of the interfaces (low density) in the non-layered region find that they Next, we discuss damage development in the helium- consistently have a cube-on-cube orientation relationship. implanted bulk Cu/Ag nanocomposites, in which the helium concentration is clearly non-zero. Recall that the helium ions are implanted vertically on the top surface of the sample. The two 3.2. Radiation defects in Cu/Ag TEM foils parts of the material, the nanolayered region and the non-layered region developed two distinct surface morphologies, as shown in To help elucidate the effects of helium, we first discuss results Fig. 4(a). Surface holes form in the non-layered region whereas from the helium-implanted Cu/Ag TEM foils at 400 C. According to surface bumps form on the nanolayered region (Fig. 4(b) and (c)). the SRIM simulation (see Fig. 1(a)), due to lack of recoil atoms Using SEM, we determined that the submicron-scale bumps (0.28/ during radiation, the average radiation damage in the Cu/Ag TEM mm2, average diameter of 700 nm) are a mixture of Cu and Ag with foils should be about 2.5 dpa and the helium concentration zero. 84 at. % Cu. This is unexpected since the composition of the Fig. 3 shows the defect structures that develop in the nanolayered nanolayered composites underneath consists of 40 at. % Cu. The material after irradiation. The standard contrast under a zone axis disproportionately large amount of Cu suggests that the nano- near [110] of Cu and Ag layers is used to visualize the defect layered material response to radiation has caused the mass structure in the layers. Between the Cu/Ag interfaces, in the layers, a transport significantly in favor of Cu over Ag, a point to which we density of vacancy clusters can be seen, while no interstitial clusters will return in the discussion.

Fig. 3. Radiation defects in Cu/Ag TEM foils after helium implantation. (a) and (b) TEM images taken near [110] beam direction. Some voids with an average diameter of 2 nm and many SFTs form in the Ag layers. Some bigger voids with an average diameter of 7 nm are found in the Cu layers. (c) SFTs in the Ag layer and corresponding High-Resolution (HR) TEM image in the inset. Some isolated voids have nucleated in the Ag layers as well. (d) Numerous faceted voids bounded by low energy planes of {111} and {100} have formed in the Cu layers. Imaging here is with an under focus of 1.5 mm (e) and (f) TEM foils (defocus of À1.5 mm) was tilted off axis about 15 to incline the interface at an angle with respect to the electron beam. Many voids with sizes of about 1 nm appear on the Cu/Ag hetero-twin interface, as marked by the red boxes and arrows in (f). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) M. Wang et al. / Acta Materialia 160 (2018) 211e223 215

Fig. 4. (a) Typical SEM image of the bulk Cu/Ag after helium irradiation; (b) Numerous micron-scale bumps have formed in the nanolayered region in the Cu/Ag nanocomposites. The bumps are a mixture of Cu and Ag but enriched with Cu (84 at.%). (c) Radiation-induced cavities have formed on the non-layered region; (d) SEM image shows the typical morphology of the helium-implanted bulk Cu/Ag nanocomposites; (e) and (f) are enlarged images corresponding to the white rectangles in (d). Some of the cavities are attached to the interface in the nanolayered region, but a large spallation with width of 620 nm and length of several micrometers has formed in the region with the non-layered structures. (g) Typical cross-section TEM image showing the distribution of radiation induced cavities within the helium implanted region (This TEM foil was cut parallel to the implanted ion beam). The helium concentration and damage distributions are simulated by SRIM.

To associate these surface defects with the subsurface defect Fig. 4(g) investigates the radiation damage formed in the development underneath, a 30 mm  30 mm cross-section of the nanolayered region marked in Fig. 4(d) by using TEM. Under the helium-implanted bulk Cu/Ag nanocomposites is taken to include current low magnification TEM observation, only radiation induced the entire damage region, starting from the top surface, as shown in large helium bubbles are visible in the peak helium and damage Fig. 4(d)e(f). This depth was selected to include not only the concentration region. Numerous bubbles with sizes of tens of damage region predicted by SRIM but well beyond. According to nanometers have formed within the Cu and Ag layers, as shown in SRIM, the highest radiation damage of 16 dpa and a maximum Fig. 4(g). Some of the Ag bubbles are attached to the interface. The helium concentration of 8 at. % in Cu/Ag nanocomposites are ach- helium ion dose at this depth likely marks the limit at which the ieved at a depth of 900 nm. Fig. 4(d) shows the typical morphology interface can positively mitigate irradiation defects. However, in the of a cross-section of the top surface damaged region of the helium- other regions experiencing lower doses, the interface is clearly implanted bulk Cu/Ag nanocomposites. The nanolayered and non- mitigating defect accumulation. In addition, the structure marked layered regions exhibit significantly different responses to the same by the red box in Fig. 4(g) is the cross-section of the surface bump. It irradiation conditions. At this magnification, the nanolayered ma- is clear that the surface bump has a solid interior and is different terial sustained no change in the layered morphology. A region of from surface blistering. cavities at the peak radiation damage region (~900 nm), according to the SRIM prediction, appear in the nanolayered region, as 3.4. Depth-dependent radiation defects distribution under low highlighted in Fig. 4(e). In contrast, at this same peak radiation helium concentrations depth (~900 nm), the non-layered region developed a large spall- ation with a length of several micrometers and a height of 620 nm, The foregoing analysis suggests that the high density of in- as marked in Fig. 4(f). The spallation forms via the coalescence of terfaces in the nanolayers are resisting macro-scale damage outside radiation-induced helium bubbles and explains the development of of the peak damaged region. To analyze at a finer resolution, the the holes in the surface above the non-layered region observed in effect of the interface on radiation defect evolution and defect- Fig. 4(c). interface interactions under the lower helium concentrations, 216 M. Wang et al. / Acta Materialia 160 (2018) 211e223

TEM analysis is performed at different depths from the surface with The density and size of the helium bubbles in the D ¼ 400 nm relatively low helium concentration. Specifically, we select three region are even higher. At this depth, the damage and helium regions with depths in the range from 0 to 400 nm from the surface, concentration are 4.5 dpa and 0.4 at. % respectively. Despite the wherein the radiation damage ranges from 2 to 4 dpa, and the higher dpa and helium concentrations, the interface remains helium concentration continuously increases from zero to 0.4 at.%. undefective as in lower helium concentrations. Similarly, the Ag TEM foils from the helium-implanted bulk Cu/Ag nanocomposites bubbles are just contacting the interface and not spreading in the were cut at depths of D ¼ 100 nm, 200 nm and 400 nm (see the interface, and a bubble-free zone clearly forms along the interface schematic in Fig. 1(b)). Because the helium concentration in these from the Cu side over an approximate distance of 20 nm from the regions is not zero, we will refer to the observed cavities as helium interface. Numerous faceted helium bubbles are observed within bubbles. the Ag layer, and only slightly smaller faceted helium bubbles than Fig. 5(a) shows typical images of the defects accumulated within those in Ag have formed with the Cu layer. Faceted helium bubbles the first 100 nm. At this depth, the damage and helium concen- are an indication that the internal gas pressure is low [1]. tration according to the SRIM calculation are low, being 2.3 dpa and 0.05 at. % respectively. Only radiation induced helium bubbles can 3.5. Depth-dependent radiation defects distributions under high be seen; no SFTs and dislocation loops appear. The interface does helium concentrations not contain any bubbles or other obvious defects. One striking feature found is that the helium bubbles in Ag are all attached to Defect-interface interactions in the nanolayered region are fi < Cu/Ag interface at a ne ( 5 nm) contact point on the surface of the analyzed at even higher helium concentrations than the foregoing bubble, and those in the Cu layer are located away from the inter- by repeating the TEM analysis at depths in the range of face, lying only in the center of the Cu layer (Fig. 5(a) and (b)). The 600 nme1000 nm. At depths ranging from 400 nm to 1000 nm, the ± average size of the helium bubbles in Ag and Cu is 8.1 3.0 nm and damage and helium concentrations vary non-linearly, increasing ± 6.9 1.4 nm, respectively. from 400 nm to 800 nm (from 5 to 16 dpa and from 1 at.% to 8 at.% ¼ Fig. 5(c) shows the radiation defects in D 200 nm of bulk Cu/ in helium concentration), and both decreasing precipitously at Ag nanocomposites. The damage and helium concentration are 2.5 depths greater than 950e1000 nm (Fig. 1(b)). dpa and 0.13 at. % respectively. We observe that the number density Fig. 6 presents collectively typical TEM images focused on the of helium bubbles in the Cu and Ag layers is higher than that in the radiation defects in the nanolayered region at depths of 600 nm, ¼ fi D 100 nm region. Helium bubbles are still attached at a ne point 800 nm and 1000 nm. Compared to the low damage depths to the Cu/Ag interface and have enlarged by growing into the Ag (D  400 nm), the number density and size of helium bubbles in- layer rather than spreading in the interface or into the Cu layer. crease, as to be expected. Further, the density of helium bubbles is Apart from these contacting bubbles, the interface still remains higher and their sizes are larger in Ag layers than that in the Cu undefective. Unlike at the lower helium concentrations, some of the layers in this range. Bubbles form in the interior of the Cu and Ag bubbles can be seen embedded fully inside the Ag layer and not layers of approximate sizes 14.5 ± 1.1 nm and 19.5 ± 2.3 nm, attached to the interface. At this higher defect density, a well- respectively. More interestingly, the interface morphology remains fi de ned bubble-free zone of width 30 nm from the Cu/Ag inter- intact in all depths, despite the higher helium ion doses. As another face on the Cu layer side can be seen, an example of which is shown notable feature, in the 600 nm and 1000 nm regions, the helium in Fig. 5(d). The largest helium bubble in Cu layer is less than 11 nm, bubble interface distribution also remains unchanged. At the while the helium bubbles in Ag layer can reached a size of 22 nm. interface, bubble-free zones of width 20 nm form in the Cu layers Bubbles (and voids) are expected to be larger in Ag than in Cu due and bubbles contacting the interface from the Ag side. In the ± 2 to the lower surface energy in Ag than Cu (19 3 mJ/m for Ag 1000 nm region, more of the helium bubbles are faceted, an indi- e 2 e compared to 36 80 mJ/m for Cu [56 59]). cation of lower helium/vacancy ratio than at 600 nm and 800 nm

Fig. 5. Typical radiation defects in the bulk Cu/Ag nanocomposites at depths of (a) D ¼ 100 nm, (c) D ¼ 200 nm and (e) D ¼ 400 nm, with under focus of 1.5 mm (b), (d) and (f) are the enlarged images of the white rectangles in (a), (c) and (e), respectively. Helium bubbles adhered on the Ag side of the Cu/Ag interface in Ag layers. A bubble-free zone has formed along Cu/Ag interface on the Cu layer side in (f). M. Wang et al. / Acta Materialia 160 (2018) 211e223 217

Fig. 6. TEM images of typical radiation defects in the bulk Cu/Ag nanocomposites at depths of (a) D ¼ 600 nm, (c) D ¼ 800 nm and (e) D ¼ 1000 nm, taken using a defocus of À1.5 mm (b), (d) and (f) are the enlarged images corresponding to the white rectangles in (a), (c) and (e), respectively. With increasing radiation damage (3e16 dpa) and helium concentration (1 at.% to 8 at.%), the number density and size of helium bubbles have increased; however, the distribution character of helium bubbles is unchanged.

[1]. The notable difference in the 1000 nm region is shown in form in the narrow Cu layer. To study the defects structures in the Fig. 6(f); the helium bubbles formed along the Cu/Ag interface in Cu/Ag interface, we tilt the sample, as shown in Fig. 7(d). Similar to the Ag layer side are no longer just touching the interface but the Cu layers, parallel dislocation lines have also developed in the appear to be flattened, with a greater part of the bubble surface area Cu/Ag interface, likely accumulating along the misfit dislocation in contact with the interface. lines on the interface. In contrast, in the Ag layer, curved dislocation The defect distribution at D ¼ 800 nm corresponds to the region lines without a preferential orientation and small faulted intersti- experiencing the peak radiation damage (16 dpa) and highest he- tial plates of diameter 10 nm have formed in the Ag layer, as marked lium concentration (8 at. %) region. Here both Ag and Cu layers by white arrows in Fig. 7(d). In the deeper region of D ¼ 2000 nm contain a high density of helium bubbles, overlapping with one shown in Fig. 8(a), only dislocation structures in the Ag layer can be another. No Cu bubble-free region can be seen. At this high amount observed. No defects are obvious in the Cu layers. It is well known of irradiation, the interface no longer appears to affect damage that the interstitials have higher mobility than vacancies in metals; accumulation. This observation is consistent with the macroscopic therefore, the radiation defects observed in D ¼ 1800 nm and observation made in Fig. 4(d) and (e), where bubbles had formed. D ¼ 2000 nm of bulk Cu/Ag nanocomposites are largely interstitial clusters. Finally, Fig. 8(b) confirms that in the nanostructure at D ¼ 3000 nm, no radiation defects are detected. 3.6. Radiation defects beyond the SRIM-calculated helium- implanted region 4. Discussion According to the SRIM simulation in Fig. 1(b), the helium ions irradiate the sample from top surface to a depth of 1200 nm in the 4.1. Effect of Cu/Ag interface density on damage accumulation Cu/Ag nanocomposites and the radiation damage and helium concentration in the region beyond 1200 nm should be zero. Yet The unique microstructures in the region of bulk Cu/Ag nano- this is an approximation and defects can still possibly migrate at composites provide an opportunity to witness the effect of inter- greater depths and interact with the Cu/Ag interface. To examine face density and morphology on their radiation tolerance. Both interface effects on those migrating radiation defects, TEM analyses regions contain low energy Cu/Ag interfaces. While the orientation are performed at deeper regions in the range of 1400e2000 nm. relationships of their interfaces are different, being hetero-twin in Figs. 7 and 8 show common TEM images of these regions. Despite the nanolayered and cube-on-cube in the non-layered, their the SRIM prediction, overall, numerous radiation defects are found. interface energies are similar, and both interface variants contain Fig. 7(a) and (b) show many cavities (likely containing helium due similar misfit networks [47]. In the non-layered material, however, to diffusion) at D ¼ 1400 nm in the Ag and Cu layers. The observa- the interfaces are spaced five to ten times greater than the nano- tion of cavities in 1400 nm region indicates that the radiation- layered composites and hence contain a much lower interface induced vacancies in the helium-implanted region are mobile and density than the nanolayered composite. As a general, well diffuse beyond the limit of the radiated area to form cavities. accepted notion, the Cu/Ag interfaces mitigate radiation damage Similar to the helium bubbles in the first 100 nm region in Fig. 5(a), and the more interfaces, the better the radiation tolerance [16]. the cavities in the Ag layer at this depth are attached to the Cu/Ag Fig. 4, however, is direct evidence of this. Due to the high density of interface. Cavities in the Cu are smaller than those in the Ag layers. Cu/Ag interfaces, only small cavities are formed in the nanolayered The Cu cavities are mainly distributed in the interior of layer and Cu/Ag region, but a micron-sized spallation is produced in the non- not near the interface. layered Cu/Ag region. In what follows, we focus the discussion on In the D ¼ 1800 nm region, a high density of dislocations is also the interaction of the Cu/Ag interfaces with the different radiation produced in both the Cu and Ag layers, as shown in Fig. 7(c) and (d). defects produced during helium implantation as analyzed in the Parallel dislocation lines with two ends pinned by Cu/Ag interface nanolayered composites. 218 M. Wang et al. / Acta Materialia 160 (2018) 211e223

Fig. 7. TEM images of typical radiation defects in the bulk Cu/Ag nanocomposites at depths of (a) D ¼ 1400 nm, and (c) D ¼ 1800 nm, using a defocus of À1.5 mm (b) is the enlarged images corresponding to the area marked by the white rectangles in (a). The voids are attached to the Cu/Ag interface in Ag, and distributed in the interior in Cu. (c) Parallel dislocation lines with two ends pinned by Cu/Ag interface are formed in the narrow Cu layer, but curved dislocation lines without prefer orientation are formed in Ag layer. (d) The sample was tilted to study the defect structures in the Cu/Ag interface. Parallel dislocation lines have also formed in the Cu/Ag interface, and small faulted interstitial plates appear in the Ag layers.

Fig. 8. TEM images of typical radiation defects in the bulk Cu/Ag nanocomposites at depths of (a) D ¼ 2000 nm and (b) D ¼ 1800 nm using an under focus of -1.5 mm. (a) Only dislocation structures appear in the Ag layer, and no obvious defects in the Cu layers at D ¼ 2000 nm. (b) No radiation defects in the Cu/Ag nanocomposites at D ¼ 3000 nm.

4.2. Combined effect of surfaces and helium on radiation defects (2.5e3 dpa) and the bulk sample contains approximately 0.05 at% accumulation helium. However the TEM foils contain no helium. Consequently, we can expect that vacancy clusters form in the helium irradiated In order to assess the effect of helium on defect development, TEM foils but helium bubbles would form in the bulk sample. The we use TEM to analyze the radiation defects of the nanolayered Cu/ TEM foil has two free surfaces that will guide the point defect flux Ag TEM foils (Fig. 3(a)) 100 nm thick and its bulk nanolayerd sample migrating from the center to these surfaces. The bulk sample, within the 100 nm range from the surface (Fig. 5(a)). According to however, only has one surface, which gives rise to radiation point the SRIM simulation in Fig. 1, both regions experience similar dpa defect flux flow from the irradiated region to the outside. Two M. Wang et al. / Acta Materialia 160 (2018) 211e223 219 important differences in radiation damage response are revealed. tests inside the TEM, must be performed under a suitable radiation First, many SFTs and few voids form in the Ag layer and make conditions to remove the significant surface sink effect. For contact with the interface in the TEM foils, whereas no SFTs and instance, the in situ radiation test could be performed at a low to only interior bubbles and interface contacting-helium bubbles are intermediate temperature (e.g., lower than 400 C in this case) to found in the Ag layer in the irradiated bulk sample. Second, a high slow done point defect migration and avoid the strong surface ef- density of voids form in the Cu layer in TEM foils compared to a very fect. From a more design perspective, the strong surface effect can low helium bubble density in the Cu layer in the bulk sample and be exploited to design radiation tolerant materials, such as nano- no SFTs form. porous metals, etc. [23e26]. Due to the strong sink effect of the surfaces in thin foil radiation, the interstitials in both Cu and Ag layers migrate to the sample 4.3. Interface stress-induced segregation m surfaces and are removed immediately. The vacancies ( ECu ¼ m ¼ fl 0.69 eV and EAg 0.66 eV), have nearly one order of magnitude One of the main effects of interfaces and surfaces is to in uence m ¼ m e lower mobility than interstitials ( ECu 0.084 eV and EAg the segregation of point defects [14 27]. When adjacent interfaces ¼ 0.088 eV) [60e63], and consequently are largely trapped inside exist in close proximity, with spacing of nanometer dimension, they the TEM foils and cluster. Atomic-scale simulations suggest that a can have a marked influence on radiation defect dynamics. In this critical cluster size is needed to form SFTs or via a Silcox-Hirsch work, we investigate the interaction of a high density of Cu/Ag mechanism [64,65]. This cluster size is smaller in Ag than in Cu, interfaces to radiation defects. These interfaces are semi-coherent due to the low stacking fault energy of Ag (19 ± 3 mJ/m2)as and contain a misfit dislocation network [45]. Using dislocation compared to Cu (36e80 mJ/m2)[56e59]. With all else being equal, theory and atomic-scale simulation, this network was character- SFT formation would be expected to occur earlier in Ag than Cu. ized by three sets of non-parallel misfit dislocation arrays, aligned However, due to interface-affected segregation, the vacancies in the along the 〈110〉 directions lying in the {111} interface plane and Cu crystal would segregate to the interface but those in the Ag spaced approximately 10 lattice spacings apart [45]. would remain in the Ag layers (More details are given in following The misfit dislocation array develops a heterogeneous interfa- part). The preferential supersaturation of vacancies in the Ag layers cial stress state, where the local stresses are particularly intense at in the irradiated TEM foils prefer to cluster into SFTs and only some the nodes where the three misfit arrays intersect [45,69]. Because of them randomly agglomerate into few voids. Taken together, the the Ag lattice parameter is larger than the Cu lattice parameter, If relatively lower vacancy concentration in Cu than Ag due to the the two metals attempt to form a coherent interface, then the Ag interface and surface, and the lower stacking fault energy of Ag will shrink and the Cu expand so their atomic positions match. compared to Cu explains the observation of SFTs in Ag and not in However, for relatively thick layers (as in this study), locally due to Cu, as illustrated in Fig. 9(a). the misft dislocation network that has formed in the interface, the Yet in the helium irradiated bulk Cu/Ag nanocomposites, espe- stress state and hence their influence on point defect segregation cially within D ¼ 100 nm range, SFTs did not form in either Ag or Cu. energies can be expected to vary inhomogeneously. At the edge In the case, the top surface of the bulk plays a role of an unlimited interface misfit dislocation, the extra half plane lies on the Cu layer sink for interstitials at this low depth. The surface attracts an in- side due to the smaller lattice constant than Ag. Therefore terstitials flux from peak damaged region at D ¼ 800 nm to the compressive stress of the misfit lies on the Cu layer side and tensile surface, while a large fracture of vacancies are trapped by helium stress on the Ag side. Using atomic-scale simulation, the effect of via the formation of helium-vacancy complexes, and as a result, the local interfacial stresses on vacancy and interstitial segregation radiation-induced vacancies are mostly eliminated by the direc- energies was studied for the cube-on-cube interface in much earlier tional interstitials flux. The small fraction of vacancies that survive work by Meunier et al. [69]. At that time, the misfit dislocations are cluster with the implanted-helium interstitials and form into were not characterized but the three arrays can be discerned from low density of helium bubbles in both Ag and Cu layers, as shown in their atomic structures and interfacial nodes they refer to in their Fig. 5(a). Consequently, a lower density of vacancy clusters in the calculations, correspond to the misfit dislocation intersections bulk sample compared to the TEM sample can be expected, (MDIs). The nodal points in the interface were found to have explaining the lower density of bubbles in the Cu layer at exceptionally high pressure on the Cu side and low on the Ag side D ¼ 100 nm than voids in the TEM sample. Further, experimental [69]. Similarly, they computed the chemical segregation energy studies have shown that the helium ions promote the formation of map on the Ag layer of the interface, which is the change in energy cavities rather than SFT [66]. Considering the surface segregation of when an Ag atom in bulk Cu is exchanged with a Cu atom at a given interstitials and the suppressing effect of helium on SFT formation interface site [69]. Based on atomic-scale calculations of vacancy would explain why numerous SFTs in the Ag layered in TEM foil and chemical segregation energies for the pristine interface, the Cu irradiation (with no helium), while only bubbles or cavities are vacancy formation energy is much lower in the high pressure detected in the Ag layers in bulk irradiation with a non-zero helium (compression) interfacial regions at these nodes than in the bulk. concentration. Therefore, the energy for Cu vacancies to segregate to the interface The TEM analysis in Fig. 3(a) shows that the SFTs are distributed is particularly high at these nodal points. On the Ag side, the in the middle of Ag layer and not adjacent to Cu/Ag interface, interface pressure is tensile, except at the nodes. Consequently, the suggesting the formation of an SFT-free zone in the Ag layer. It was Ag interstitial segregation energy, in contrast, was found negative observed under in situ radiation [67,68] and predicted by kinetic for most of the interface except at these nodal/MDI sites. Thus, the Monte Carlo simulation that the SFTs are mobile under radiation Ag interstitials would tend to segregate to the remaining portions [65]. The sink effect of interface can cause SFTs close to the Cu/Ag of the interface and not to the MDIs. The contribution of interface interface to migrate away from the interface or to annihilate with stress-affected segregation, for this interface, would suggest that Cu nearby segregating interstitials. As a result, only those formed in vacancies would preferentially segregate to the interfaces, espe- the middle of Ag layers remained stable, as shown in Fig. 3(a). cially in the particular relatively small regions in the interface The above analyses collectively demonstrate a significant effect around the nodes, leaving an excess of Cu interstitials in the layers of sample surface on the radiation defect dynamics and their for- (Fig. 10(a)). In contrast, the Ag interstitials would segregate over a mation mechanism. From a practical viewpoint, the radiation larger interfacial area (away from the nodal points) leaving an even experiment performed with TEM foils, for example in situ radiation greater excess of vacancies in the Ag layers than interstitials in the 220 M. Wang et al. / Acta Materialia 160 (2018) 211e223

Fig. 9. Schematic of the radiation-induced defects distribution seen in the irradiated (a) TEM foils (no helium concentration) and (b) the bulk samples of Cu/Ag nanocomposites.

Fig. 10. Schematic of the vacancy transfer across the Cu/Ag interface under irradiation leading to a bubble free zone in Cu and Ag bubbles in contact with the interface. (a) Cu vacancies segregate to the MDI due to compression stress, and Ag interstitials migrate to non-MDI area because of tensile stress; (b) Cu vacancies at MDI transfer into Ag layer and attach to interface; (c) Cu vacancies can be further transferred into Ag layer interior; (d) A state of interstitials enriched in Cu and vacancies enriched in Ag (Cavities form along interface or inside the layer) is achieved due to the vacancy pump effect of Cu/Ag interface. M. Wang et al. / Acta Materialia 160 (2018) 211e223 221

Cu layers (Fig. 10(a)). During this process, the implanted helium can that any vacancies in the Cu side of the interface would not prefer to slow down vacancy migration through the formation of helium- remain there but move into Ag, as Ag atoms segregate even more vacancy complexes [1,66]. In light of the above analysis, many of preferentially to all areas of the interface (Fig. 10(b)). If indeed the interface effects on radiation damage found in our TEM analysis excess vacancies would prefer to move into Ag, apart from the 3e8 can be rationalized on the basis of interface-stress assisted vacancies at the interfacial MDIs, more vacancies would cluster as segregation. the voids or bubbles and into Ag (see Fig. 10(c)). As we have seen, We should note briefly that these segregation energy estimates the bubbles (and voids) grow into Ag and not further within the Cu/ are based on a single point defect within a pristine interface. This Ag interface or Cu. segregation energy naturally becomes adjusted as vacancies and As we have seen in the TEM analysis, even as more vacancies are interstitials become trapped in the interface. We will return to this produced with higher dpa, the bubbles only get bigger. Beyond this, point, particularly in explaining the formation of interface- it cannot be stated how these cavities would grow from the MDIs contacting bubbles, but despite this, we find that much basic into Ag and where the vacancies to support this growth may insight on interface-stress induced segregation still holds. originate. We offer some speculation that they come from not only the Ag layers but also the Cu layers. Ag atoms would segregate to 4.4. Interface-affected zones in nanolaminates the interface but due to the positive heat of mixing between Ag and Cu (6.8 kJ/mol at RT) [70], they would tend not to move into Cu once An in-depth TEM analysis of this Cu/Ag nanolayered material they are at the interface. Due to lower vacancy formation energy shows a strong influence of the interface on defect accumulation, 111 ¼ and lower surface energy of Ag compared to Cu (Ag ( gAg 0.92 J/ the details of which depends significantly on dpa and helium m2, g100 ¼ 0.99 J/m2) compared to Cu ( g111 ¼ 1.06 J/m2, g100 concentration. Fig. 9(b) summarizes in the form of an illustrative Ag Cu Cu ¼ 2 > map, the defects observed and radiation defect dynamics. 1.13 J/m )[71]), excess ( Nv) Cu vacancies that are absorbed into An unusual and important phenomenon observed here is the the Cu/Ag interface could be driven energetically to transmit across formation of interface-affected zones on either side of the interface the interface into the interior of Ag layers rather than remaining that persist over the range of dpa and helium concentration stud- near the interface in the Cu side (Fig. 10(d)). Fig. 10 provides a ied. In the Ag layer, a number of voids and bubbles form that just schematic of the transfer. This idea introduces the notion of the contact and do not overlap the interface. The Cu side, in contrast, interface MDIs, acting as a vacancy pump from the Cu layers to the develops a void-free and bubble-free zone. Interface-contacting Ag layers. Time accelerated MD calculations would help to provide voids on the Ag side were found to attach to the misfit dislocation support for this notion of vacancy transfer across the interface. intersections (MDIs) only on the Ag side and not the Cu side [45]. We should emphasize that the aforementioned analysis applies The relative values of surface energy to interface energy at the MDIs to regions outside the peak region in which the defect density has meets the wetting criterion, enabling the void to just “wet” the not overwhelmed the crystalline layers. Clearly, as we have seen in ¼ interface, and not spread within the interface [45]. By studying the the TEM analysis of D 800 nm, if the dpa reaches a high level of 16 response of the interface to increasing dpa and helium concentra- dpa, a threshold is reached for this Cu/Ag interface in which just tion, we find that similarly interface-contacting bubbles form in the interface-contacting bubbles form. bulk helium implanted samples. In addition, these interface- contacting bubbles remain and only grow in size with increasing 4.5. Radiation accelerated defect diffusion dpa and helium concentration. At the same time, a bubble-free zone that forms on the Cu side of the interface, however, remains According to the SRIM simulation in Fig. 1, the radiation damage at 20e22 nm (Fig. 6). should be zero for depths D beyond 1200 nm from the implanted The formation mechanism for both the interface-contacting Ag surface. However, many voids and dislocation structures are bubbles and the bubble-free zone in Cu have yet to be clarified. observed at depths ranging from 1400 nm to 2000 nm. For com- Here we propose that the stress-assisted interface segregation parison, we can separately estimate the diffusion distance of point pffiffiffiffiffiffiffiffiffiffiffi likely plays a role. Because of the high-compressive pressure points defects L using Fick's diffusion law L ¼ 4DDt, and DD ¼ D0exp À at the MDIs, Cu vacancy formation energy is lower there compared  to the bulk and consequently segregate preferentially to the MDIs Q , where D is the diffusion coefficient, Q is activation energy of (Fig. 10(a)). In Meunier et al. [69], considering the favorability of the RT D nodes for segregation, vacancies were added to the nodes (MDIs) diffusion, R is molar gas constant, T is temperature, D0 is diffusion and the calculations of segregation energy were repeated. When constant, t is radiation duration, and L is the diffusion distance. By the number of vacancies per node, Nv, reaches 3 (up to 5), the Ag taking the duration of radiation t to be 200 min [72] and using the segregation energy becomes negative (favorable) homogeneously diffusivity of interstitials in Ag and Cu at 400 C, the diffusion dis- everywhere in the interface. For Nv in excess of 8, the Cu vacancy tance for interstitials in Cu and Ag are 13 nm and 78 nm, respec- formation energy, however, increased to the point that it reached tively. These distances are much lower than the depths at which the same value as the bulk. Accordingly, as a result of a small cluster dislocation structures are observed (see Figs. 7 and 8). The analysis of vacancies at the MDIs, Cu vacancy segregation energy to the indicates that the Ag interstitials migrated at least for a distance of interface would no longer be biased at the MDIs, and more 800 nm (10 times the prediction), while the Cu interstitials moved a importantly, more vacancies will not collect at the MDIs. distance of more than 600 nm (46 times the prediction). The sig- The assessment above implies that a critical 3e5Nv at the MDIs nificant difference in the experimental measurement and calcula- is favorable, explaining the fine contact point of the bubble (or void) tion suggests radiation-enhanced diffusion; that is, the radiation at the interface but the spreading vacancies along the interface in accelerates defects diffusion by at least one order of magnitude excess of Nv is not favorable. The question then remains whether [73,74]. From the experimental results in Figs. 7 and 8, dislocations there is a more favorable region for which additional vacancies at are seen at greater depths than voids, indicating that the in- the interface could move and the bubble could expand from this terstitials are moving faster than the vacancies, and further, that the interface contact point. According to the calculation, with a critical Ag interstitials are diffusing further than the Cu interstitials. A Nv of vacancies at the MDIs, the interface area over which Ag second point worth noting is the larger discrepancy for Cu than for interstitial segregation energy is negative [69]. This alone indicates Ag. Because interface-stress assisted segregation is occurring at the 222 M. Wang et al. / Acta Materialia 160 (2018) 211e223 numerous Cu/Ag interfaces, the interstitial concentration in the Cu in-depth TEM analysis of this Cu/Ag nanolayered material shows a layer is higher than that in Ag layer, which promotes the diffusion of strong influence of the interface on defect accumulation, the details Cu interstitials into the deeper regions. This effect would explain of which depends significantly on dpa and helium concentration. why the Cu interstitials demonstrate a much larger discrepancy, 46 The important findings from the TEM analysis are: times in the diffusion distance, from that predicted by the diffusion law compared to Ag. Both radiation-enhanced diffusion and the 1) For the same 2 to 3 dpa, the Cu/Ag nanocomposite develops a marked difference in the mobility of vacancies and interstitials different defect configuration with helium than without. Helium between the two phases are harmful to metals during radiation. suppresses the formation of stacking fault tetrahedra and an Together, they hinder the recombination of point defects and pro- SFT-free zone forms near the Cu/Ag interface. mote the accumulation of radiation damage. 2) Study of irradiated TEM foils demonstrates a strong surface ef- fect, resulting in a higher vacancy concentration in the Cu and 4.6. Effect of interfaces on surface bumps formation Ag layers than in the irradiated bulk composite under the same dpa. The result is a completely different set of defects for the The surface bumps that form on the surface above the nano- same material: from larger SFT in Ag compared to smaller layered material had a disproportionate amount of Cu, much larger bubbles in the bulk and a higher void concentration in Cu than than the eutectic composition. Due to the top surface sink effect, bubble concentration in the bulk. radiation induced interstitials in both Cu and Ag layers will flow 3) Although the Cu/Ag interfaces receive defect fluxes from both from the irradiated region to the surface. However, due interface- the adjoining crystals, they remain remarkably stable micro- stress assisted segregation of Cu/Ag interface, there is a higher structurally at the nanometer scale over the broad range of concentration of vacancies in Ag than in the Cu layers and some of concentrations studied, with a stability limit of 16 dpa and 8 at. the Ag interstitials are largely annihilated at the interface before %. arriving the surface. In the Cu layer, the vacancies segregate to the 4) Apart from the highest, peak dpa and helium concentration, the interface and are annihilated there, leaving the Cu layer enriched interface is found to create a zone of interface-contacting bub- with interstitials. These Cu interstitials have a greater chance to bles in Ag and a bubble-free zone in Cu, which persists over the migrate to the top surface to form the surface Cu-enriched bumps entire range of dpa. This finding is rationalized based on inter- or diffuse into deeper regions to cluster into dislocations. Interface- face stress-assisted segregation, which induces vacancy transfer stress assisted segregation of the high density of Cu/Ag interface across the interface at the interface misfit dislocation nodes. Due can explain both the lower amount of helium bubbles in the Cu to the compressive pressure on the Cu interface layer and tensile layer (Fig. 5(a)), especially for the top 100 nm range, and the pressure on the Ag interface layer generated by their lattice enrichment of Cu in the surface bumps (Fig. 4(e)) seen above the mismatch, Cu vacancies are biased to segregate to the misfit nanolayered composite. nodes in the interface and Ag atoms to segregate to the other portions of the interface. 4.7. Final remarks on radiation tolerant materials design 5) With the accumulating of point defects, vacancies can be transferred from Cu to Ag, leading to an enrichment of in- Voids and bubbles have been shown in numerous works to terstitials in Cu and vacancies in Ag, which further accelerate affect dislocation motion in single phase materials [9]aswellasin diffusion to form surface bumps and dislocations structures threading dislocations in nanolayered materials [38]. The devel- beyond the helium-irradiated region. opment of a bias bubble-free and void-free zone on the Cu side and interface-contacting bubbles and voids on the Ag side would sug- Acknowledgements gest that differences in the way dislocations would move in the two adjoining crystals across the interface as well as how dislocation This work was supported by the National Key Research and glide would potentially transfer across the interface. These findings Development Program of China (2017YFB0702301) and the Na- imply that for future radiation tolerant materials design, only tional Natural Science Foundation of China (Grant Nos.51471128 having interfaces on one direction is insufficient. Parallel interfaces, and 51621063). W.Z.H. would like to acknowledge the support of although in high density, still appear to be somewhat insufficient to Youth Thousand Talents Program of China and the Young Talent heal the radiation-induced vacancies and interstitials. Three- Support Plan of XJTU. I.J.B. acknowledges financial support from the dimensional interface network of closely spaced interfaces, pha- National Science Foundation Designing Materials to Revolutionize ses of nanoscale dimension, could be incorporated to obstruct the and Engineer our Future Program (NSF CMMI-1729887). long-range migration of vacancies and interstitials and confine them inside a small volume in order to enhance their annihilation References rate. In such nanostructured systems, we show that role of interface-stress assisted segregation cannot be neglected. Material [1] G.S. Was, Fundamentals of Radiation Materials Science: Metals and Alloys, choices could be made to minimize or enhance the bias generated Springer Verlag, Berlin, 2007. [2] M. Nastasi, J.W. Mayer, J.K. Hirvonen, Ion-solid Interactions: Fundamentals from the interface-stress assisted segregation. and Applications, Cambridge University Press, Cambridge, 1996. [3] S.J. Zinkle, J.T. Busby, Structural materials for fission & fusion energy, Mater. 5. Conclusions Today 12 (2009) 12e19. [4] S.J. Zinkle, G.S. Was, Materials challenges in nuclear energy, Acta Mater. 61 (2013) 735e758. Radiation-induced point defect interactions with an FCC/FCC [5] T.D. de la Rubia, H.M. Zbib, T.A. Khraishi, B.D. Wirth, M. Victoria, M.J. CaturIa, interface in a wide range of dpa (zero dpa to 16 dpa) and helium Multiscale modelling of plastic flow localization in irradiated materials, Na- e concentration (0e8 at. %) have been investigated by helium im- ture 406 (2000) 871 874. [6] M. Victoria, N. Baluc, C. Bailat, Y. Dai, M.I. Luppo, R. Schaublin, B.N. Singh, plantation in both bulk Cu/Ag nanocomposites and its TEM foils. Microstructure and associated tensile properties of irradiated fcc and bcc Direct comparison of the irradiation response of a nanolayered and metals, J. Nucl. Mater. 276 (2000) 114e122. ultra-fine (submicron) Cu/Ag composite with similar interfaces [7] G.R. Odette, G.E. Lucas, Embrittlement of nuclear reactor pressure vessels, JOM (J. Occup. Med.) 53 (2001) 18e22. demonstrates that the nanolayered structure, bearing a high den- [8] B.D. Wirth, How does radiation damage materials? Science 318 (2007) sity of Cu/Ag interfaces drastically improve radiation tolerance. An 923e924. M. Wang et al. / Acta Materialia 160 (2018) 211e223 223

[9] M.S. Ding, J.P. Du, L. Wan, S. Ogata, L. Tian, E. Ma, W.Z. Han, J. Li, Z.W. Shan, [41] Q.M. Wei, Y.Q. Wang, M. Nastasi, A. Misra, Nucleation and growth of bubbles Radiation-induced helium nanobubbles enhance ductility in submicron-sized in He ion-implanted V/Ag multilayers, Phil. Mag. 91 (2010) 553e573. single-crystalline copper, Nano Lett. 16 (2016) 4118e4124. [42] X. Zhang, J.A. Beach, M. Wang, P. Bellon, R.S. Averback, Precipitation kinetics of [10] M.S. Ding, L. Tian, W.Z. Han, J. Li, E. Ma, Z.W. Shan, Nanobubble fragmentation dilute Cu-W alloys during low-temperature ion irradiation, Acta Mater. 120 and bubble free-channel shear localization in helium irradiated submicron- (2016) 46e55. sized copper, Phys. Rev. Lett. 117 (2016) 215501. [43] L.F. Zeng, R. Gao, Z.M. Xie, S. Miao, Q.F. Fang, X.P. Wang, T. Zhang, C.S. Liu, [11] R.W. Grimes, R.J.M. Konings, L. Edwards, Greater tolerance for nuclear mate- Development of interface-dominant bulk Cu/V nanolamellar composites by rials, Nat. Mater. 7 (2008) 683e685. cross accumulative roll bonding, Sci. Rep. 7 (2017) 40742. [12] X.M. Bai, A.F. Voter, R.G. Hoagland, M. Nastasi, B.P. Uberuaga, Efficient [44] S.J. Zinkle, J.T. Busby, Structural materials for fission & fusion energy, Mater. annealing of radiation damage near grain boundaries via interstitial emission, Today 12 (2009) 12e19. Science 327 (2010) 1631e1634. [45] S. Zheng, S. Shao, J. Zhang, Y. Wang, M.J. Demkowicz, I.J. Beyerlein, N.A. Mara, [13] G. Ackland, Controlling radiation damage, Science 327 (2010) 1587e1588. Adhesion of voids to bimetal interfaces with non-uniform energies, Sci. Rep. 5 [14] M.J. Demkowicz, R.G. Hoagland, J.P. Hirth, Interface structure and radiation (2015) 15428. damage resistance in Cu-Nb multilayer nanocomposites, Phys. Rev. Lett. 100 [46] K.Y. Yu, Y. Liu, E.G. Fu, Y.Q. Wang, M.T. Myers, H. Wang, L. Shao, X. Zhang, (2008) 136102. Comparisons of radiation damage in He ion and proton irradiated immiscible [15] E.G. Fu, J. Carter, G. Swadener, A. Misra, L. Shao, H. Wang, X. Zhang, Size Ag/Ni nanolayers, J. Nucl. Mater. 440 (2013) 310e318. dependent enhancement of helium ion irradiation tolerance in sputtered Cu/V [47] J. Wang, I.J. Beyerlein, N.A. Mara, D. Bhattacharyya, Interface-facilitated nanolaminates, J. Nucl. Mater. 385 (2009) 629e932. deformation twinning in copper within submicron AgeCu multilayered [16] W.Z. Han, M.J. Demkowicz, N.A. Mara, E. Fu, S. Sinha, A.D. Rollett, Y. Wang, composites, Scripta Mater. 64 (2011) 1083e1086. J.S. Carpenter, I.J. Beyerlein, A. Misra, Design of radiation tolerant materials via [48] Y.Z. Tian, Z.F. Zhang, Bulk eutectic CueAg alloys with abundant twin bound- interface engineering, Adv. Mater. 25 (2013) 6975e6979. aries, Scripta Mater. 66 (2012) 65e68. [17] I.J. Beyerlein, A. Caro, M.J. Demkowicz, N.A. Mara, A. Misra, B.P. Uberuaga, [49] B.P. Eftink, A. Li, I. Szlufarska, N.A. Mara, I.M. Robertson, Deformation response Radiation damage tolerant nanomaterials, Mater. Today 16 (2013) 443e449. of AgCu interfaces investigated by in situ and ex situ TEM straining and MD [18] X.C. Liu, H.W. Zhang, K. Lu, Strain-induced ultrahard and ultrastable nano- simulations, Acta Mater. 138 (2017) 212e223. laminated structure in nickel, Science 342 (2013) 337e340. [50] T.D. Shen, X. Zhang, K. Han, C.A. Davy, D. Aujla, P.N. Kalu, R.B. Schwarz, [19] V.Y. Gertsman, R. Birringer, On the room-temperature grain growth in Structure and properties of bulk nanostructured alloys synthesized by flux- nanocrystalline copper, Scripta Mater. 30 (1994) 577e581. melting, J. Mater. Sci. 42 (2007) 1638e1648. [20] R. Bullough, M.R. Hayns, M.H. Wood, Sink strengths for thin-film surfaces and [51] T.D. Shen, R.B. Schwarz, X. Zhang, Bulk nanostructured alloys prepared by flux grain-boundaries, J. Nucl. Mater. 90 (1980) 44e59. melting and melt solidification, Appl. Phys. Lett. 87 (2005) 141906. [21] Y. Guerin, G.S. Was, S.J. Zinkle, Materials challenges for advanced nuclear [52] J.F. Ziegler, J.P. Biersack, U. Littmark, The Stopping and Range of Ions in Soilds, energy systems, MRS Bull. 34 (2009) 10e19. Ion Implantation Science and Technology, Pergamon Press, New York, 1985. [22] X. Zhang, K. Hattar, Y. Chen, L. Shao, J. Li, C. Sun, K. Yu, N. Li, M.L. Taheri, [53] R.K. Linde, Lattice parameters of metastable silver-copper alloys, J. Appl. Phys. H. Wang, J. Wang, M. Nastasi, Radiation damage in nanostructured materials, 37 (1966) 934. Prog. Mater. Sci. 96 (2018) 217e321. [54] J.B. Liu, Y.W. Zeng, L. Meng, Interface structure and energy in Cue71.8wt.% Ag, [23] E.M. Bringa, J.D. Monk, A. Caro, A. Misra, L. Zepeda-Ruiz, M. Duchaineau, J. Alloy. Comp. 464 (2008) 168e173. F. Abraham, M. Nastasi, S.T. Picraux, Y.Q. Wang, D. Farkas, Are nanoporous [55] I.J. Beyerlein, N.A. Mara, D. Bhattacharyya, D.J. Alexander, C.T. Necker, Texture materials radiation resistant? Nano Lett. 12 (2012) 3351e3355. evolution via combined slip and deformation twinning in rolled silverecopper [24] C. Sun, D. Bufford, Y. Chen, M.A. Kirk, Y.Q. Wang, M. Li, H. Wang, S.A. Maloy, cast eutectic nanocomposite, Int. J. Plast. 27 (2011) 121e146. X. Zhang, In situ study of defect migration kinetics in nanoporous Ag with [56] H. Suzuki, C.S. Barrett, Deformation twinning in silver-gold alloys, Acta Metall. enhanced radiation tolerance, Sci. Rep. 4 (2014) 6e11. 6 (1958) 156e165. [25] C. Sun, B.P. Uberuaga, L. Yin, J. Li, Y. Chen, M.A. Kirk, M. Li, S.A. Maloy, H. Wang, [57] E.B. Tadmor, N. Bernstein, A first-principles measure for the twinnability of C. Yu, X. Zhang, Resilient ZnO nanowires in an irradiation environment: an in FCC metals, J. Mech. Phys. Solid. 52 (2004) 2507e2519. situ study, Acta Mater. 95 (2015) 156e163. [58] C.B. Carter, I.L.F. Ray, Stacking fault energies of copper alloys, Philos. Mag. A 35 [26] J. Li, C. Fan, J. Ding, S. Xue, Y. Chen, Q. Li, H. Wang, X. Zhang, In situ heavy ion (1977) 189e200. irradiation studies of nanopore shrinkage and enhanced radiation tolerance of [59] J.P. Hirth, J. Lothe, T. Mura, Theory of dislocations, J. Appl. Mech. 50 (1983) nanoporous Au, Sci. Rep. 7 (2017) 1e10. 476. [27] B.N. Singh, Effect of grain-size on void formation during high-energy electron- [60] R. Rizk, P. Vajda, F. Maury, A. Lucasson, P. Lucasson, Stage IE recovery of irradiation of austenitic stainless-steel, Phil. Mag. 29 (1974) 25e42. electron-irradiated pure silver and of its dilute alloys with cadmium and in- [28] R.W. Siegel, S.M. Chang, R.W. Balluffi, Vacancy loss at grain-boundaries in dium, J. Appl. Phys. 48 (1977) 481e486. quenched polycrystalline gold, Acta Metall. 28 (1980) 249e257. [61] R.W. Balluffi, Vacancy defect mobilities and binding energies obtained from [29] Y. Chimi, A. Iwase, N. Ishikawa, A. Kobiyama, T. Inami, S. Okuda, Accumulation annealing studies, J. Nucl. Mater. 69e70 (1978) 240e263. and recovery of defects in ion-irradiated nanocrystalline gold, J. Nucl. Mater. [62] P.L. Williams, Y. Mishin, J.C. Hamilton, An embedded-atom potential for the 297 (2001) 355e357. Cu eAg system, Model. Simulat. Mater. Sci. Eng. 14 (2006) 817e833. [30] P.A. Thorsen, J.B. Bilde-Sorensen, B.N. Singh, Bubble formation at grain [63] M.J. Demkowicz, R.G. Hoagland, B.P. Uberuaga, A. Misra, Influence of interface boundaries in helium implanted copper, Scripta Mater. 51 (2004) 557e560. sink strength on the reduction of radiation-induced defect concentrations and [31] N. Nita, R. Schaeublin, M. Victoria, R.Z. Valiev, Effects of irradiation on the fluxes in materials with large interface area per unit volume, Phys. Rev. B 84 microstructure and mechanical properties of nanostructured materials, Phil. (2011) 104102. Mag. 85 (2005) 723e735. [64] J. Silcox, P.B. Hirsch, Direct observations of defects in quenched gold, Philos. [32] M.J. Demkowicz, O. Anderoglu, X.H. Zhang, A. Misra, The influence of sigma 3 Mag. A 4 (1959) 72e89. twin boundaries on the formation of radiation-induced defect clusters in [65] B.P. Uberuaga, R.G. Hoagland, A.F. Voter, S.M. Valone, Direct transformation of nanotwined Cu, J. Mater. Res. 26 (2011) 1666e1675. vacancy voids to stacking fault tetrahedra, Phys. Rev. Lett. 99 (2007) 1e4. [33] W.Z. Han, M.J. Demkowicz, E.G. Fu, Y.Q. Wang, A. Misra, Effect of grain [66] I. Mukouda, Y. Shimomura, T. Iiyama, Y. Harada, Y. Katano, T. Nakazawa, boundary character on sink efficiency, Acta Mater. 60 (2012) 6341e6351. D. Yamaki, K. Noda, Microstructure in pure copper irradiated by simultaneous [34] M.A. Tschopp, K. Solanki, F. Gao, X. Sun, M.A. Khaleel, M. Horstemeyer, Probing multi-ion beam of hydrogen, helium and self ions, J. Nucl. Mater. 283e287 grain boundary sink strength at the nanoscale: energetics and length scales of (2000) 302. vacancy and interstitial absorption by grain boundaries in a-Fe, Phys. Rev. B [67] Y. Matsukawa, S.J. Zinkle, One-dimensional fast migration of vacancy clusters 85 (2012), 064108. in metals, Science 318 (2007) 959e962. [35] M.J. Demkowicz, A. Misra, A. Caro, The role of interface structure in controlling [68] K.Y. Yu, D. Bufford, C. Sun, Y. Liu, H. Wang, M.A. Kirk, M. Li, X. Zhang, Removal high helium concentrations, Curr. Opin. Solid State Mater. Sci. 16 (2012) of stacking-fault tetrahedra by twin boundaries in nanotwinned metals, Nat. 101e108. Commun. 4 (2013) 1e7. [36] S. Mao, S. Dillon, R.S. Averback, The influence of CueNb interfaces on local [69] I. Meunier, G. Treglia, B. Legrand, R. Tetot, B. Aufray, J.-M. Gay, Molecular vacancy concentrations in Cu, Scripta Mater. 69 (2013) 21e24. dynamics simulations for the Ag/Cu(111) system: from segregated to consti- [37] W.Z. Han, N.A. Mara, Y.Q. Wang, A. Misra, M.J. Demkowicz, He implantation of tutive interfacial vacancies, Appl. Surf. Sci. 162e163 (2000) 219e226. bulk CueNb nanocomposites fabricated by accumulated roll bonding, J. Nucl. [70] H. Sheng, E. Ma, Enhanced solubility on surfaces: molecular-dynamics simu- Mater. 452 (2014) 57e60. lations of an Ag overlayer on Cu(100), Phys. Rev. B 61 (2000) 9979e9982. [38] N. Li, M. Nastasi, A. Misra, Defect structures and hardening mechanisms in [71] B.D. Todd, R.M. Lyndenbell, Surface and bulk properties of metals modeled high dose helium ion implanted Cu and Cu/Nb multilayer thin films, Int. J. with sutton-chen potentials, Surf. Sci. 281 (1993) 191e206. Plast. 32e33 (2012) 1e16. [72] N.L. Peterson, Self-diffusion in pure metals, J. Nucl. Mater. 69e70 (1978) 3e37. [39] S. Ohsaki, S. Kato, N. Tsuji, T. Ohkubo, K. Hono, Bulk mechanical alloying of Cu- [73] R. Sizmann, The effect of radiation upon diffusion in metals, J. Nucl. Mater. Ag and Cu/Zr two-phase microstructures by accumulative roll-bonding pro- 69e70 (1968) 386e412. cess, Acta Mater. 55 (2007) 2885e2895. [74] V. Naundorf, Diffusion in metals and alloys under irradiation, Int. J. Mod. Phys. [40] G.R. Odette, M.J. Alinger, B.D. Wirth, Recent developments in irradiation- 6 (1992) 2925e2986. resistant steels, Annu. Rev. Mater. Res. 38 (2008) 471e503.