Phase-separated PVcD materials

Metastable solid solutions of low-solubility element pairs, such as Cu-Nb,^{Reference Banerjee, Puthucode, Bose and Ayyub24} Cu-W,^{Reference Vullers and Spolenak25} or Cu-Mo,^{Reference Derby, Cui, Baldwin and Misra26} may be synthesized by PVcD. When deposited at room temperature, these alloys often take on a noncrystalline (amorphous) structure. Even in the absence of crystallization, subnanometer patterning into compositionally enriched domains is observed in these materials.^{Reference Banerjee, Puthucode, Bose and Ayyub24,Reference Baumer and Demkowicz27} Upon annealing, these amorphous alloys undergo simultaneous crystallization and phase separation via the formation and growth of spherulitic domains consisting of near-equilibrium crystalline phases embedded in a metastable amorphous solid solution.^{Reference Puthucode, Devaraj, Nag, Bose, Ayyub, Kaufman and Banerjee28} The two constituent phases in these domains take on complex, interconnected shapes, resulting in the formation of a morphologically complex composite upon complete devitrification.

Vüllers and Spolenak conducted a comprehensive investigation of the phase-separated morphologies of room-temperature codeposited Cu-W films after annealing at 750°C.^{Reference Vullers and Spolenak25} They investigated a wide range of film compositions, from 5 to 67 at.% W, and annealing times from 10 min to 10 h. Figure 2 illustrates the range of microstructure morphologies they observed in cross sections obtained by ion-beam milling. For films with W content beneath 48 at.%, annealing leads to the formation and coarsening of numerous, discrete, W-rich particles embedded in a Cu matrix. For W content between 48 and 67 at.%, this situation reverses, giving rise to Cu-rich precipitates embedded in a W-rich matrix. The W-rich films exhibit a pronounced columnar structure, giving rise to an anisotropic morphology with elongated Cu-rich domains oriented perpendicular to the sample surface.

Figure 2.

Montage of cross-sectional views of Cu-W films of varying composition codeposited at room temperature and subsequently annealed over a range of times at 750°C. Arrows indicate the Cu-rich capping layer in the 31 at.% W sample. Note: ad, as-deposited. Reprinted with permission from Reference Reference Vullers and Spolenak25. © 2015 Elsevier.

Derby et al. investigated the microstructural morphology and mechanical response of PVcD Cu-Mo films and obtained results that contrasted with the previous finding.^{Reference Derby, Cui, Baldwin and Misra26} Rather than anneal samples that had been deposited at room temperature, they performed codeposition at elevated temperatures between 400°C and 800°C and deposition rates between 0.12 and 1.4 nm/s. Their investigation was confined to a single composition with 45 at.% Cu. By examining microstructural morphology both along sample surfaces and cross sections, they identified three distinct morphology types, as shown in Figure 3—vertical composition modulations (VCMs), lateral composition modulations (LCMs), and random composition modulations (RCMs).

Figure 3.

Three canonical morphology types observed in Cu-Mo films codeposited at elevated temperatures: vertical composition modulations (VCMs), lateral composition modulations (LCMs), and random composition modulations (RCMs). Reprinted with permission from Reference Reference Derby, Cui, Baldwin and Misra26. © 2018 Elsevier.

The primary factor that determines the type of morphology that forms during PVcD of Cu-Mo is the deposition temperature, with the lowest temperature giving rise to VCMs and the highest temperature yielding RCMs. These observations agree with the findings of Lu et al., who carried out two-dimensional phase field simulations of morphology evolution in phase separating PVcD alloys.^{Reference Lu, Wang, Gao, Shi, Liu and Wang29} They demonstrated the emergence of VCM and LCM morphologies, the former at lower and the latter at higher temperatures. Their simulations did not observe the RCM morphology because it is inherently a three-dimensional (3D) structure and therefore could not be captured by their model. Interestingly, Adams et al. report the opposite trend in PVcD Al-Ge films, with VCMs occurring at higher temperatures than LCMs.^{Reference Cui, Derby, Li, Mara and Misra30} In the experiments of Derby et al., the effect of deposition rate appears to be limited to determining the characteristic dimensions of the compositionally enriched domains, with the lowest rates giving rise to ∼100-nm-sized features and the highest yielding ∼10-nm-scale features.^{Reference Derby, Cui, Baldwin and Misra26} This trend agrees with the Monte Carlo simulations of Adams et al.^{Reference McCue, Karma and Erlebacher31}

Lu et al. furthermore found that VCMs appear to emerge by a local rearrangement in the immediate vicinity of the sample surface, suggesting that this morphology requires high-temperature deposition and cannot form during post-deposition annealing. Indeed, there is no sign of VCMs in the Cu-W films in Figure 2. Meanwhile, the cross-sectional views of LCMs shown in Figure 3 appear comparable to the W-rich columnar composite structures shown in Figure 2, suggesting that the latter might also exhibit a LCM morphology. However, the formation of a Cu-rich capping layer on all the Cu-W films discussed in Reference Reference Vullers and Spolenak25 prevented the acquisition of plan-view surface micrographs, precluding definitive conclusions concerning the 3D morphology of these films. It remains to be determined whether the range of morphologies shown in Figure 3 is only accessible via high-temperature deposition or if they also arise in post-deposition annealing. Vüllers and Spolenak conducted annealing at only one temperature, so it is not clear whether other morphologies may emerge at other annealing temperatures in Cu-W.

Vüllers and Spolenak used nanoindentation to investigate the modulus and hardness of their Cu-W films.^{Reference Vullers and Spolenak25} For all annealing times, both quantities exhibited the same monotonic variation with composition, in reasonable agreement with the rule of mixtures. It appears that the differences in microstructure morphology exhibited by these materials do not have a discernible effect on indentation modulus or hardness. Cui et al. performed indentation and nanopillar compression tests^{Reference Cui, Derby, Li, Mara and Misra30} on PVcD Cu-Mo materials synthesized by Derby et al.^{Reference Derby, Cui, Baldwin and Misra26} and compared them to the mechanical response of nanolaminate composites synthesized by alternating deposition of pure Cu and pure Mo. They found that both the PVcD materials and the multilayers had comparable hardness and flow strength under compression, confirming that microstructure morphology does not have any clear effect on the strength of these materials.

Cui et al., however, found a clear difference in the approach to steady-state flow exhibited by multilayers and PVcD films. The former displayed an extended elastic range followed by a rapid transition—spanning 1–2% plastic strain—to nearly perfect-plastic flow. Moreover, the multilayer exhibited localization of flow into sharp shear bands, leading to early and sudden failure of the microcompression specimen. In contrast, the PVcD material showed a prolonged transition to steady-state flow spanning nearly 10% plastic strain. Departure from linear elastic behavior occurred at substantially lower stress than in the multilayered specimen and was followed by extensive hardening behavior. The PVcD microcompression specimen exhibited a greater degree of uniform plastic deformation than the multilayered specimen, without formation of shear bands and without sudden failure. These results suggest that while microstructural morphology may not have a pronounced effect on the ultimate strength of the nanocomposite, it has a marked influence on the hardening behavior and consequently on the uniformity of plastic flow. In particular, the complex, intertwined morphologies of PVcD materials appear to be more resistant to flow localization than the multilayer nanocomposites.

LMD composites

In dealloying, one component of an initially uniform parent alloy is dissolved away into a surrounding solvent while the remaining component self-organizes into a morphologically complex microstructure. For example, nanoporous metals made by dealloying in aqueous solvents have garnered great interest over the past two decades,^{Reference McCue, Karma and Erlebacher31,Reference McCue, Benn, Gaskey, Erlebacher and Clarke32} with nanoporous gold being the most extensively investigated.^{Reference Erlebacher, Aziz, Karma, Dimitrov and Sieradzki33,Reference McCue, Stuckner, Murayama and Demkowicz34} More recently, it has been shown that dealloying can also be carried out using liquid metals^{Reference Wada, Yubuta, Inoue and Kato35,Reference Wada and Kato36} or even solid metals^{Reference Wada, Yubuta and Kato37,Reference McCue and Demkowicz38} as solvents. Unlike dealloying in aqueous media, LMD and solid metal dealloying may be used not only to synthesize porous materials,^{Reference Yu, Yubuta, Wada and Kato39–Reference Okulov, Weissmuller and Markmann43} but also fully dense, multiphase metal composites.^{Reference McCue and Demkowicz38,Reference Geslin, McCue, Gaskey, Erlebacher and Karma44,Reference McCue, Gaskey, Geslin, Karma and Erlebacher45}

One distinct advantage of composite synthesis via LMD is the wide range of composite morphologies that can be produced. For example, Geslin et al. investigated composites made by dealloying Ta_xTi_1–x parent alloys in liquid Cu.^{Reference Geslin, McCue, Gaskey, Erlebacher and Karma44} As illustrated in Figure 4, by changing the parent alloy composition, they observed a wide range of morphologies, from particles in a matrix (Figure 4d), to lamellar (Figure 4c), to interpenetrating bicontinuous (Figure 4a–b). The authors also demonstrated a similar range of morphologies in concurrent phase-field simulations of the dealloying process. By changing the dealloying time and temperature, the characteristic dimensions of these composite structures may be altered.^{Reference McCue, Gaskey, Geslin, Karma and Erlebacher45} Moreover, because LMD composites can be fabricated in bulk form, they are amenable to post-processing treatments, such as cold rolling.^{Reference Mokhtari, Le Bourlot, Adrien, Dancette, Wada, Duchet-Rumeau, Kato and Maire46}

Figure 4.

Micrographs and corresponding phase-field models of metal composites made by dealloying of Ta_xTi_1–x alloys in liquid Cu: (a) x = 0.35, (b) x = 0.25, (c) x = 0.15, (d) x = 0.05. The color map in the phase-field models shows the concentration of the parent alloy component that is immiscible with the solvent. Reprinted with permission from Reference Reference Geslin, McCue, Gaskey, Erlebacher and Karma44. © 2015 Nature Publishing Group.

McCue et al. used nanoindentation^{Reference McCue, Gaskey, Crawford and Erlebacher47} as well as uniaxial compression and tension testing^{Reference McCue, Ryan, Hemker, Xu, Li, Chen and Erlebacher48} to characterize the mechanical properties of LMD composites. As expected, these materials exhibited marked differences in local indentation modulus and hardness, depending on which phase in the composite was being indented.^{Reference McCue, Gaskey, Crawford and Erlebacher47} Uniaxial compression tests on larger-scale specimens showed a clear dependence of strength on the characteristic dimensions of the domains making up the composite, with smaller features yielding stronger materials, as demonstrated by the stress–strain curves in Figure 5. Since domain dimensions were on the micron level in all of the composites tested, this variation was interpreted as being analogous to the classical Hall–Petch effect in single-phase polycrystals. Interestingly, concurrent with the increase in strength with decreasing microstructure dimensions, these materials also exhibited increasing rates of strain hardening. It remains to be determined whether these trends persist as component dimensions decrease to the submicron and nanometer level.

Figure 5.

Stress–strain curves for uniaxial compression of composites made by dealloying of Ta_xTi_1–x alloys in liquid Cu. All curves are labeled by the characteristic dimensions of the composite constituents. The strains to failure for each test are given on the right side of the figure. Reprinted with permission from Reference Reference McCue, Ryan, Hemker, Xu, Li, Chen and Erlebacher48. © 2016 Wiley.

Challenges and opportunities

While phase-separated PVcD and LMD composites both provide new opportunities for studying the effects of microstructure morphology on plastic flow behavior, investigations of these materials are still in the early stages with questions greatly outnumbering answers. In particular, in both cases, complete descriptions of microstructural morphology are rarely possible because detailed 3D reconstructions of the microstructure morphology are often lacking. Depending on characteristic microstructure dimensions, different approaches to obtaining such reconstructions may be pursued. Banerjee et al. showed that atom probe tomography is suitable for investigating phase distributions in PVcD composites with nanometer-scale dimensions.^{Reference Banerjee, Puthucode, Bose and Ayyub24,Reference Puthucode, Devaraj, Nag, Bose, Ayyub, Kaufman and Banerjee28} On the micrometer scale, Zhao was able to use x-ray nanotomography reconstructions to reconstruct 3D microstructure morphology.^{Reference Zhao, Wada, De Andrade, Williams, Gelb, Li, Thieme, Kato and Chen-Wiegart42} Once acquired, 3D reconstructions may be analyzed to quantify the topology, connectivity, and interface curvatures of the composite morphology.^{Reference Chen, Chu, Yi, McNulty, Shen, Voorhees and Dunand49,Reference Mendoza, Thornton, Savin and Voorhees50}

Mechanical testing of these materials is also at an early stage, yet a variety of techniques has already been applied to them, including nanoindentation,^{Reference Vullers and Spolenak25,Reference Cui, Derby, Li, Mara and Misra30,Reference McCue, Gaskey, Crawford and Erlebacher47} micropillar compression,^{Reference Cui, Derby, Li, Mara and Misra30} and conventional uniaxial loading.^{Reference McCue, Ryan, Hemker, Xu, Li, Chen and Erlebacher48} One area of potential future opportunity is to combine small-scale mechanical testing with nondestructive morphology reconstruction via x-ray nanotomography.^{Reference Zhao, Wada, De Andrade, Williams, Gelb, Li, Thieme, Kato and Chen-Wiegart42} Comparisons of microstructure morphologies before and after deformation may provide insights into the mechanisms responsible for the deformation behavior of these materials, especially their apparent high hardening rates. Due to the low frequency of dislocation–dislocation interactions in nanocomposite solids, explaining the high rates of hardening seen in these materials may require the postulation of novel, nondislocation-based mechanisms. Improved tracking of microstructure evolution during deformation may shed light on these mechanisms.

Finally, dedicated atomistic simulations are needed to understand the mechanisms responsible for the deformation behavior of phase-separated PVcD and LMD composites. Atomic-scale models may be constructed using experimentally acquired 3D microstructure reconstructions. Alternatively, they may also be created using phase-field simulations as a template, in analogy to the approach commonly used in the modeling of nanoporous solids.^{Reference Beets and Farkas51} With improved understanding of the connections between microstructural morphology and mechanical deformation, new types of designer morphologies optimized for peak mechanical performance may be designed and synthesized, hastening the transition of morphology-controlled metal composites from the lab to technological use.