Surface Triple Junctions Govern the Strength of a Nanoscale Solid

Surface triple junctions (STJs), i.e., the termination lines of grain boundaries at solid surface, are the common line defects in polycrystalline materials. Compared with planar defects such as grain boundaries and surfaces, STJ lines are usually overlooked in a material’s strengthening although abundant atoms may reside at STJs in many nanomaterials. In this study, by in situ compression of coarse-grained and nanocrystalline nanoporous gold samples in an electrochemical environment, the effect of STJs on the strength of nanoporous gold was successfully decoupled from grain-boundary and surface effects. We found that the strength of nanoporous gold became sensitive to STJ modification when ligament size was decreased to below $\sim 100\mathrm{nm}$, indicating that STJs started to influence ligament strength at sub-100 nm scale. This STJ effect was associated with the emission of dislocations from STJs during plastic deformation. Our findings strongly suggest that the structure and chemistry at STJs should be considered in understanding the mechanical response of sub-100 nm scale materials.

Figure 1

NPG with high-population STJs. (a) TEM image and an inset selected area electron diffraction (SAED) pattern, (b) PED orientation map, and (c) a schematic illustration of NC-NPG structure. (d) TEM image and an inset SAED pattern, (e) EBSD orientation map, and (f) a schematic illustration of the structure of CG-NPG. Grain boundary (GB).

Figure 2

In situ compression of NPG in an electrochemical environment. (a) Schematic illustration of an electrochemical cell for in situ compression. Working electrode (WE). Reference electrode (RE). Counter electrode (CE). (b) Cyclic voltammogram of CG-NPG and NC-NPG in 1 M ${\mathrm{HClO}}_4$ at $5\mathrm{mV}/\mathrm{s}$. Compressive engineering stress-strain curves of (c) a typical CG-NPG ($L=32\mathrm{nm}$) and (d) a typical NC-NPG ($L=43\mathrm{nm}$) measured in situ under electrochemical control, with potential jumping from 1.13 to 1.53 V at a strain of approximately 0.10. The strain rate is ${10}^{-4}{\mathrm{s}}^{-1}$. (e) Monolayer oxide induced increase in flow stress, $\mathrm{\Delta }\sigma /{\sigma }^{\mathrm{CLN}}$, plotted as a function of ligament size $L$, for CG-NPG, NC-NPG1 dealloyed from ${\mathrm{Al}}_2\mathrm{Au}$, and NC-NPG2 obtained by compressing CG-NPG to a strain of 0.55 (see more details in Figs. S3 and S5 [17]). Grain size (D). Some data of CG-NPG are adapted from Refs. [18, 28]. Inset table shows major defects and their sensitivity to electrochemical modifications (on/off: sensitive; off: insensitive; –: this defect is very few or does not exist).

Figure 3

Strain-rate sensitivity and activation volume. (a) Typical engineering stress-strain curves of NPG under compression obtained by strain-rate jump tests (NC-NPG with clean surface). (b) Strain-rate sensitivity $m$ and (c) activation volume for CG- and NC-NPG samples with similar ligament sizes ($\sim 45\mathrm{nm}$) but different surface states.

Figure 4

Traces of dislocation nucleation in NC-NPG. HR-TEM images showing stacking faults (indicated by yellow lines) bounded by (a-b) a grain boundary or surface triple junction and (c) the surface. The other ends of stacking faults terminate in Shockley partial dislocations. Ligament size is approximately 45 nm. Compression strain is approximately 0.10. Most of these stacking faults were eventually removed by trailing partials to form full dislocations, as frequently seen in the deformed structures of CG- and NC-NPGs ($L=45\mathrm{nm}$) with different surface states (Figs. S9–S12 [17]).