A Material with Electrically Tunable Strength and Flow Stress

See allHide authors and affiliations

Science  03 Jun 2011:
Vol. 332, Issue 6034, pp. 1179-1182
DOI: 10.1126/science.1202190


The selection of a structural material requires a compromise between strength and ductility. The material properties will then be set by the choice of alloy composition and microstructure during synthesis and processing, although the requirements may change during service life. Materials design strategies that allow for a recoverable tuning of the mechanical properties would thus be desirable, either in response to external control signals or in the form of a spontaneous adaptation, for instance in self-healing. We have designed a material that has a hybrid nanostructure consisting of a strong metal backbone that is interpenetrated by an electrolyte as the second component. By polarizing the internal interface via an applied electric potential, we accomplish fast and repeatable tuning of yield strength, flow stress, and ductility. The concept allows the user to select, for instance, a soft and ductile state for processing and a high-strength state for service as a structural material.

Environmental exposure and in-service wear influence the mechanical performance of engineering materials. Environmental effects are often adverse, as exemplified by stress corrosion cracking (1). Immersion in corrosive media may also impair strength and flow stress without immediate failure (24). More recently, nanoindentation studies have revealed a decisive effect of the surface state on the hardness (5). Although the microscopic processes that couple the plasticity to the environment have not been conclusively determined, the observations demonstrate that a material’s mechanical performance can vary depending on the environment to which it is exposed during service. Here, we exploit these observations in designing a material with controllable strength and ductility. Our approach rests on two principles. First, we maximize the impact of surface processes by working with nanomaterials with an extremely large surface area. Second, we design the material as a hybrid in which an electrolyte becomes an inherent part of the microstructure. Interfacial properties and processes can be controlled via an electric potential, with consequences for the macroscopic behavior of the nanocomposite. In this way, the yield strength and flow stress of our material can be recoverably varied by as much as a factor of 2.

Our samples are made by dealloying, a corrosion process that selectively dissolves the less noble component from an alloy and leaves behind a monolithic body with a uniform, nanometer-scale structure composed of a contiguous skeleton of “ligaments” of the more noble component interpenetrated by an equally contiguous pore space (6, 7). Previous work has revealed extremely large local strength of the individual ligaments (8) and excellent compressive ductility of macroscopic nanoporous gold (npg) samples (9). It has also been shown that nanoporous metal wetted by electrolyte may serve as an actuator material with high stroke and high work density (10, 11).

The hybrid material is made by imbibition of the pores of npg with 1 M HClO4, an electrolyte exhibiting only weak adsorption on gold in a wide range of the electrode potential E. Figure 1 displays the microstructure of the porous metal (Fig. 1B), along with a schematic illustration of the compression test and of the in situ electrochemical control setup (Fig. 1A). The electrochemical characteristics are exemplified in the cyclic voltammogram, showing the dominantly capacitive polarization of the internal interfaces at more negative potential and the adsorption and desorption of about one monolayer of oxygen species (12) at the metal surface at more positive potential (Fig. 1C). Bulk gold oxide is not formed under these conditions (13).

Fig. 1

(A) Schematic illustration of compression of bulk npg samples in situ with electrochemical control. WE, working electrode; RE, reference electrode; CE, counterelectrode. (B) SEM image of a npg sample, showing uniform and small structure size (L = 20 nm). (C) Cyclic voltammogram (five successive scans) of current I versus potential E for npg in 1 M HClO4 solution at the potential scan rate of 5 mV/s. Potential is specified versus the standard hydrogen electrode, SHE.

We started out by compressing the hybrid material at constant potential. Figure 2A shows the compression stress-strain curves of a sample with ligament diameter L = 20 nm at two values of E, 1.03 V and 1.48 V, as measured versus the standard hydrogen electrode (SHE). The potentials correspond to “clean” and “oxygen-covered” surfaces of the ligaments, respectively. The two different surface states lead to distinctly different compression performance. At the lower potential, the sample is ductile up to high strain. The inset in Fig. 2A shows photographs of the initial and final sample shapes, testifying to deformation by uniform densification at negligible transverse plastic strain. Compression at the higher potential brings a quite different behavior. Not only does the yield strength increase by 36%, from 22 MPa to 30 MPa, but there is also a noticeable loss of ductility. The differences in mechanical behavior, despite the practically identical microstructure, demonstrate the importance of changes of state at the internal interfaces for yielding and plastic flow.

Fig. 2

Potential dependence of strength and flow stress of npg. (A) Compressive stress-strain curves of engineering stress, σ, versus engineering strain, ε, measured in situ at constant potentials. Ligament surface is covered with submonolayer-thick oxygen when E = 1.48 V and is clean when potential is held at 1.03 V. The inset is a photograph of npg samples with clean surface before and after compression [adapted from (9)]. (B) Responses of plastic flow (red) to potential jumps. Data obtained at a constant potential of 1.03 V (gray) were plotted for comparison. (C) Summary of strength and flow stress increase (σf,1.48f,1.03) induced by surface oxygen adsorption. Note that the relative density, ρ, increases during compression. (D) Response of flow stress to potential jumps in double-layer region. Ligament diameter is 20 nm for samples in (A), (B), and (D).

The recoverability of the mechanical property changes can be verified by implementing cyclic potential jumps during compression tests. Figure 2B shows a typical stress-strain curve for such a test, superimposed on a curve recorded at a constant E = 1.03 V. Both samples have L = 20 nm. The segments of both graphs agree well when E = 1.03 V, exemplifying the excellent sample-to-sample reproducibility of the mechanical behavior. When the potential is increased to 1.48 V, the flow stress is seen to increase rapidly. This change can be reversed, because the flow behavior at the lower voltage is recovered when the potential is switched back to 1.03 V.

Collecting data for different samples, Fig. 2C shows that the flow stress variation increases with increasing strain. At the higher strains, the flow stress can be as much as doubled by varying the potential. The figure also contains data for samples with larger ligament diameters, L = 45 and 200 nm. The results confirm the trend of larger flow stress at larger potential, while the amplitude of the variation decreases with increasing L.

The recoverable changes in plastic behavior concur with the formation and removal of the surface oxygen adsorbate layer. However, compression tests with potential jumps within the regime of capacitive charging demonstrate that adsorption is not a requirement. Figure 2D shows the recoverable change of flow stress when E is varied between 1.03 and 0.08 V, avoiding oxygen adsorption. The flow stress variation is here smaller, 10 to 15% for L = 20 nm, but still appreciable. Remarkably, the sign of the flow stress–potential response is reversed compared to that in the oxygen adsorption/desorption regime. Figure 2D also shows that the flow stress change is considerably more pronounced when varying the potential between 0.08 V and 0.53 V than between 0.53 V and 1.03 V. The observations are consistent with a roughly parabolic variation of flow stress with potential, with a minimum near E = 0.53 V. This is close to the potential of zero charge (pzc), which is in the range 0.24 to 0.55 V for gold in HClO4 (14).

Dry npg samples were also tested to verify the extent to which the surface states and the corresponding mechanical response are preserved when the electrolyte and the potential control are removed. Samples with L = 20 nm were held at 1.03 V and 1.48 V, respectively, for 30 min, then rinsed in pure water and dried at room temperature in open air for 8 to 9 days. Their stress-strain curves are shown in Fig. 3. The dry sample with oxygen-covered surface (conditioned at 1.48 V) was less brittle than that measured in situ at 1.48 V in electrolyte, but the yield strengths were almost identical. By contrast, at E = 1.03 V, the yield stress of dry samples with clean surface was lower than that measured in situ in the “wet” state (<10 MPa versus 22 MPa). This finding can be attributed to the structure coarsening to L = 40 nm during drying, as evidenced by scanning electron microscopy (SEM) observation. The dependence of the flow properties on the conditioning, even after drying, implies that the surface oxygen coverage and the corresponding material strengthening can be maintained, for at least a few days, after switching off the potential and even after removing the electrolyte.

Fig. 3

Compression behavior of “dry” npg samples with and without surface oxygen, corresponding to pretreatment potential of 1.48 V and 1.03 V, respectively. Graphs show engineering stress, σ, versus engineering strain, ε. Data measured in “wet” state with potentiostatic control are shown for comparison. Ligament diameter is L = 20 nm except for the E = 1.03 V (dry) sample, which experienced coarsening to L = 40 nm during drying.

The potential-controlled plasticity we present here has parallels to the “Rehbinder effect” (24) and to other observations in relation to the impact of the environment on the plasticity of engineering materials (1, 15, 16). However, reports in that context document an effective weakening of metals (lesser flow stress at given strain rate or enhanced deformation rate at given stress) when exposed to electrolyte (24, 15, 16), whereas the present material shows the opposite behavior: Our samples are weakest when dry and with clean surfaces or near the pzc, and are strengthened by exposure to the environment. Moreover, the Rehbinder effect is suppressed by even a small addition of impurities, because solid solution hardening in the bulk dominates over surface effects on dislocation movement (17). Yet the present material exhibits a large effect of potential on the mechanical behavior, despite 5 to 10 atomic percent Ag impurities. The impact of the potential variation on the mechanical behavior even increases at larger strain, despite dislocation cell structure formation (9) and the concomitant work hardening. Obviously, the correlation between potential and strength in our nanomaterial is distinctly different from the results reported for the Rehbinder effect in bulk metals.

It seems premature to try to conclusively identify a single mechanism responsible for our observations. However, experiments similar to the ones presented here connect to important issues in the general context of nanomaterial deformation and provide new ways of probing the underlying mechanisms. As a context for inspecting some of the related aspects, let us make a rough estimate of the jump in the local flow stress, σL, of the ligaments. By means of example, we inspect the first jump in the data of Fig. 2B, where the macroscopic flow stress, σM, varies by ~30%, from ~29 MPa to >38 MPa for a material with L = 20 nm and an initial solid volume fraction, υ0, of 0.25. At this point, the material has undergone ~15% plastic compression. Because the compression does not entail transverse plastic expansion (9), the solid volume fraction, υ, scales with the plastic engineering compression strain, ε, according to υ = υ0/(1 – ε). Thus, υ in the example has increased to 0.29. The Gibson-Ashby foam scaling equation, σM = 0.3σLυ3/2, provides an approximate relation among σM, υ, and σL (18). This implies that during the potential jump, the ligament strength increases by 190 MPa, from 620 to 810 MPa. We discuss possible origins of that jump, starting with the impact of two relevant capillary parameters (19): surface stress and surface tension.

The surface stress, f, imposes a bulk stress in the solid that scales inversely with the size and varies with the electrode potential (10, 11). At extremely small ligament size and f > 0, the surface-induced bulk stress may be large enough to trigger the spontaneous shear of a nanometer-size solid (20). Previous work has found f to decrease with increasing potential throughout the potential interval under study here (21). Decreasing f requires increasing bulk stress and, consequently, elastic expansion of the porous solid. The in situ dilatometry experiments of Fig. 4 verify this effect, with a macroscopic elastic strain Δl/l0 (in the absence of external load) on the order of 1.0 × 10−4 when cycling npg with a ligament size of 55 nm through the oxygen adsorption-desorption region. With a 1/L scaling, the above finding implies 2.8 × 10−4 strain for the sample with L = 20 nm. Using the value Y = 79 GPa for Young’s modulus of gold, we thus find that oxygen adsorption induces an axial stress increase of ~22 MPa. This stress may act as a prestress that increases the external load required for yielding. The magnitude of this stress is clearly too small to account for the ligament strength increase of 190 MPa. Furthermore, the changes in flow stress and surface stress are not consistent when attention is turned to the capacitive regime. Here, the surface stress still decreases with increasing potential (see the strain data of Fig. 4), but—contrary to what is found in the OH-adsorption regime—the flow stress now decreases. In summary, our findings suggest that surface stress does not play an important role in potential-dependent strengthening.

Fig. 4

In situ dilatometry trace (solid blue line) of outer dimension change, Δl/l0, of a npg sample (L = 55 nm) in response to potential sweeping in 1 M HClO4 solution at 10 mV/s. Dashed black line shows cyclic voltammogram measured during the same scan.

We next consider the effect of changes in surface tension. When a ligament is sheared by the motion of a dislocation over a cross-sectional glide plane, the step edge created at the surface increases the total surface area. The condition that the mechanical work done by the Peach-Köhler forces acting on the moving dislocation is at least equal to the increase in total surface energy (extra area × surface tension) couples the yield stress with changes in the surface tension. In support of that concept, a minimum in creep rate and a maximum in strength have been reported for macroscopic metal samples near their electrocapillary maximum of γ (3, 15, 16). The mechanism is specifically relevant for npg, because the ligaments are so small that almost all dislocations traveling therein may be expected to have one end or two ends moving on the surface (22, 23). Yet the concept is incompatible with our observations: The electrocapillary maximum agrees with the pzc (19), which is near the center of our experimental potential interval, where the flow stress is found to be at minimum. Environmental effects on the plasticity of ionic crystals have been discussed in terms of a different link between surface tension and flow stress, which does predict a flow stress minimum near the pzc: Image forces may induce cross-slip and, consequently, pinning for screw dislocation segments near free surfaces (24). The step line tension counteracts the cross slip. A reduction in surface tension and line tension would thus reduce the barrier to cross-slip, promoting pinning and ultimately strengthening the material.

As an additional interaction mechanism, we consider the possible impact of adsorption on the elastic interaction of dislocations with the surface. The drag exerted by adsorbates on moving dislocation endpoints has been termed “adsorption locking” (17). Similar to Cottrell clouds in the bulk of conventional solid solutions, adsorbed anions decorate stress fields at the surface (25). It is conceivable that the adsorbed species tend to pin the dislocation, enhancing the flow stress. Adsorption may also affect the local surface excess elastic constants and, through them, the interaction of dislocations with the surface. The modified compliance of the surface would result in a size- and potential-dependent macroscopic elastic response. In nanoporous metals, the surface effects on the effective compliance are open to experimental investigation (26).

The relevance of the dislocation endpoint drag picture of electrochemical strengthening—whether related to cross-slip or to adsorption locking—is actually supported by the results in Fig. 2C. We observed the largest flow stress change at the highest strain. In the initial nanoporous structure there are few dislocations, so the very small volumes of the ligaments may be deformed without dislocation interaction (27, 28). Yet electron backscatter diffraction imaging on npg has shown that large deformation does lead to dislocation interaction, as evidenced by the formation of subgrain boundaries by increased strain rate sensitivity (9). The creation of internal pinning points in deformed samples implies that the free arms of the dislocations that end at the surface will become shorter, so that more stress is required to active them. The flow stress may then be dominated by operation of many single-arm dislocations with one end locked in the interior of the material and another end traveling—with friction—on the surface. The maximum arm length is here shorter than the ligament diameter, which amplifies the surface effects. This mechanism would be characteristic for nanoscale porous materials, whereas coarser microstructures would favor coarser subgrain structures and hence would have a lesser impact of dislocation storage on the variation in flow stress. This notion agrees with the observation that the flow stress variation becomes less strain-dependent as the ligament size increases, as shown in Fig. 2C.

The action of surface dislocation endpoint drag is also consistent with the transient undershoot of the flow stress during oxygen desorption (Fig. 2B). The stress here initially falls below the constant-potential flow stress at 1.03 V, but it recovers that behavior as the deformation proceeds. The observation cannot be explained by stored elastic energy. Instead, the removal of surface oxygen apparently eases the motion of mobile dislocations or, alternatively, acts to unpin dislocations that were immobilized in the oxygen-covered state. Thus, the minimum flow stress during desorption would correspond to a transient peak in mobile dislocation density. Subsequent egression similar to what is known as “mechanical annealing” (29), or pinning by dislocation interaction, can recover the steady-state flow behavior.

We have demonstrated a hybrid material concept that affords control over mechanical performance. The material’s strength and ductility can thus be matched to altering requirements in service. For instance, the material may be switched to a soft and ductile state for processing, or to a high-strength state for service as a structural material. The recoverable changes in the flow stress approach a factor of 2. On a more speculative level, hybrid materials similar to the one investigated may even adapt their performance spontaneously. The response of the internal stresses to surface charging is analogous to the response of the local electrode potential to elastic strain (30, 31). In a properly designed electrochemical environment, this change may prompt local adsorption processes at stress concentrations (25, 32). As we have shown, such processes may drastically enhance the strength. This provides a mechanism for selectively removing weak spots from a microstructure, which is an essential part of a self-healing strategy. Essential to this strategy is our demonstration that deposition of submonolayer adsorbate layers is sufficient to largely enhance the strength of nanoscale hybrid materials.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

References and Notes

  1. Acknowledgments: Supported by Deutsche Forschungsgemeinschaft grant We1424/14 and by the Hundred Talents Program of the Chinese Academy of Sciences (H.-J.J.). We thank L. Kurmanaeva and Y. Ivanisenko for assistance with the compression experiments, K. Sieradzki and H. Gleiter for stimulating discussions, and K. Lu for critical reading of the manuscript.

Stay Connected to Science

Navigate This Article