Introduction

Nanoporous gold (NPG) is formed by the selective dissolution of Ag from a binary solid solution of Ag and Au1,2,3. Dealloying produces a nanoporous structure, which has a high surface area-to-volume ratio, making it a candidate material for surface area-driven energy applications4,5,6. The functionality of NPG, however, is limited by the stability of its nanostructure, especially if used at elevated temperatures.

NPG-Pt, made from AgAuPt precursors with systematic variation of Pt content, was shown by Vega and Newman to exhibit limited coarsening of ligaments during the dealloying process, making them as small as 3 nm7. 3D Atom probe tomography (APT) analysis of the ligaments by the present authors have revealed Pt clustering on nanoligament surfaces, to which the feature size refinement is attributed8,9. The use of APT had been considered difficult for NPG and NPG-Pt owing to the high porosity. To address that, a electrodeposition method of Cu was developed to infiltrate the porosity and enable successful APT analysis10.

NPG-Pt was shown to coarsen significantly in reductive atmospheres, similar to NPG, at temperatures of 250 °C and above11. The coarsening of NPG-Pt by Ostwald ripening (OR) eventually leads to the complete annihilation of porosity (i.e., densification), especially when reaching temperatures of 400 °C and above11. This is mainly owing to the desegregation of Pt away from the surfaces of nanoligaments as shown by previous studies on Au-Pt nanosystems12,13.

An interesting property investigated by Vega and Newman in a previous study14 is that the thermal coarsening of NPG-Pt in air is impeded in the range of 200–500 °C. Oxygen-induced segregation of Pt acts to inhibit the surface diffusion of Au, which is considered as the main facilitator for coarsening to occur at temperatures as low as 200 °C15. The favorable interaction of Pt with O (relative to that of Au with O), induces the segregation of Pt atoms to the surface, replacing Au atoms16.

Changes in the nanoscale chemical composition of as-dealloyed ligaments might not be limited to Pt and Au. Several theoretical studies on Au-Ag nanoparticles have shown the tendency of Ag to segregate to the surface at elevated temperatures17,18. Preferential segregation of Ag to the surface was also shown for bulk AgAu alloy19. Investigating the changes to chemical and morphological properties of nanoligaments would help in understanding the thermal coarsening process and the functionality of NPG-Pt in reductive and oxidative atmospheres.

The present paper builds upon an atom probe study of dealloying in ternary AgAuPt alloys8. In this paper, different behaviors of Au, Ag, and Pt are correlated with the coarsening behavior of NPG-Pt3 (made from dealloying AgAuPt3) in Ar-H2 and air at 300 °C, using recently developed APT methods10 to measure the chemical structure of nanoligaments at the atomic-scale and the associated nanoligament morphologies.

Results and discussion

Thermal coarsening in reductive atmosphere

As is clear from the focused ion beam (FIB) cross-section images in Fig. 1c, the ligament sizes for the sample coarsened in the reductive atmosphere are significantly larger (~200 nm) than those observed previously in the APT study of as-dealloyed NPG-Pt38. Another observation in Fig. 1c, is the voids (dark feature highlighted in inset) present inside the large ligaments. A more recent in situ STEM study of coarsening in reductive atmosphere by the authors11 showed enclosed voids in NPG and NPG-Pt. Similar voids (or bubbles) were also observed previously using electron tomography20, and positron annihilation21. Kinetic Monte Carlo simulations by Erlebacher15 alluded to void formation being a result of solid-state Rayleigh instability, controlled by surface diffusion, during coarsening. Although atomistic simulations by Kolluri and Demkowicz22 suggested that the localized plasticity at nodes and within the ligaments leads to neighboring ligaments collapsing, and ultimately void formation. At this point, the source of these voids and their evolution during thermal coarsening is not yet fully concluded. Future studies focused on microstructural examinations of ligament/pore sizes after dealloying and at successive steps of thermal coarsening should clarify further the source of encased voids, and their role in coarsening.

Fig. 1: Cross-sectional analysis and APT specimen preparation of NPG-Pt thermally coarsened in reductive atmosphere.
figure 1

SEM images showing FIB sample preparation of an APT sample for Cu-filled NPG-Pt, heated at 300 °C in Ar-H2. a Sample surface (scale bar is 20 μm). b FIB cross-section cut showing the Cu-filled NPG layer, between the Cu deposition and bulk alloy (scale bar is 2 μm). c A high magnification image showing the coarsened ligaments and the inset showing an example of entrapped voids (V) within a ligament (L), completely independent from a Cu-filled pore (P) (scale bar is 200 nm). d The APT needle following annular milling (scale bar is 200 nm). e The final APT specimen after low-kV sharpening (scale bar is 200 nm).

Full size image

The shrinkage observed in the nanoporous layer depth owing to coarsening also agrees well with previous measurements, supporting further that shrinkage in the dealloyed layer during thermal coarsening does exist at temperatures where ordinary lattice diffusion is not expected to contribute to the mass transport process. Atom maps for typical samples of NPG-Pt coarsened in Ar-H2 are shown in Fig. 2, where coarsened ligaments and the surrounding Cu filling are shown.

Fig. 2: Atom maps of a single thermally coarsened nanoligament.
figure 2

3D APT atom maps of Cu-filled NPG-Pt heated at 300 °C in Ar-H2 showing Au, Ag, Pt, and Cu atoms. Scale bar is 10 nm. Refer to supplementary video for 3D reconstruction.

Full size image

A chemical profile across a single nanoligament coarsened at 300 °C in reducing atmosphere is presented in Fig. 3. The Ag and Au compositions are both observed to be uniform across the ligament. This result contrasts Ag and Au profiles for as-dealloyed NPG-Pt, as found in previous APT analysis8,9,10, which exhibited a clear core-shell structure. As the distribution of Ag and Au is now homogenous, the core-shell structure can be understood as having been eliminated by thermal coarsening. Nanoscale homogenization of originally segregated AgAu nanoparticles was observed previously in reducing environments at elevated temperatures23. As for Pt, a clear enrichment towards the core of the ligament, compared with being primarily surface-segregated in the as-dealloyed material, supports the desegregation of Pt from the surface in reductive atmospheres discussed by Vega and Newman14. The distribution of Pt is also not as uniform when compared with that of Ag and Au, showing evidence of clustering and segregation from the Ag and Au elements inside the nanoligament.

Fig. 3: Elimination of core-shell structure and homogenization of Au-Ag distribution.
figure 3

a APT atom map of a single ligament in a reconstructed volume, showing Au, Ag, Cu atoms. b 1D concentration profile through a ligament, as indicated on a. Scale bar is 25 nm.

Full size image

The surface diffusion dealloying model predicts a Au-rich nanoligament surface owing to the kinetic entrapment of Ag inside ligament cores during their formation10,24. More specifically to NPG-Pt, previous APT studies by the authors have observed the co-segregation of Au and Pt at nanoligament surfaces8. The current observation, suggesting a tendency of Ag to move to the ligament surface in reductive atmospheres during coarsening, is significant as it shows that Ag contributes to the coarsening process led by diffusion of atoms on nanoligament surfaces. This contribution could be eliminated by increasing the amount of Ag removed during dealloying, as that would decrease further the Ag retained inside nanoligaments by what is known as secondary dealloying7,25.

The considerable Ag surface enrichment observed at relatively low temperatures for NPG-Pt could be attributed to the role of curvature. A coarsening-mediated Ag surface segregation was shown to have accelerated Ag enrichment at the surface by Pia et al.26 considering NPG in an environment of CO and O2.

To monitor the change in Pt distribution across the nanoligaments, chemical mappings across a horizontal 2 nm-thick slice of a few interconnected ligaments are presented in Fig. 4. Large segregates of Pt are observed. In the bigger nanoligament, Pt clusters are located inside the ligaments, consistent with the chemical profile in Fig. 3. Yet other Pt segregates, on the relatively thinner ligaments seem to be at the surfaces. This observation might appear counterintuitive, as a non-favorable interaction with the reducing atmosphere is expected to drive Pt desegregation from the surface12,13. However, considering coarsening kinetics, we postulate that these Pt segregates found at the surfaces of thinner nanoligaments are a result of Pt having a much lower surface mobility than Au27,28. This causes the Pt to be left behind at the surfaces of very thin nanoligaments that are shrinking and on the verge of collapsing.

Fig. 4: Formation of subsurface large Pt clusters.
figure 4

Concentration maps taken from 2 nm-thick slices of APT volume of NPG-Pt heated at 300 °C in Ar-H2. Scale bar is 20 nm.

Full size image

Although it is expected that Pt has a high mobility when moving on a surface of Au29, it is important to note that Pt is already clustered in as-dealloyed NPG-Pt, making Pt rearrangement even slower30. Clusters of several Pt atoms would also have a higher activation energy for moving around as a cluster31. Meanwhile, the extremely immobile Pt segregates are increasing in size, due to the tendency to minimize interaction surface area with the reducing atmosphere32,33,34,35, especially at areas of the nanoligaments with high curvature (saddle points)36.

Inhibited thermal coarsening in oxidative atmosphere

In contrast to NPG-Pt annealed in Ar-H2, the ligament/pore sizes in the air-annealed NPG-Pt do not show the same coarsening. This can be seen in both the SEM images of the NPG-Pt (Fig. 5) and the APT data (Fig. 6). The cross-sectional image in Fig. 5c reveals a thicker layer of NPG-Pt, compared with H2-annealed NPG-Pt, and with small feature size, not dissimilar to that observed in as-dealloyed NPG-Pt9. This is further supported by a direct comparison of the APT data for the two respective ligament structures, as shown in Fig. 7. Clearly, the thermal coarsening of ligaments was inhibited owing to the oxidative environment, such that the ligament network remains visibly similar in size and scale to the as-dealloyed material.

Fig. 5: Cross-sectional analysis and APT specimen preparation of NPG-Pt thermally coarsened in oxidative atmosphere.
figure 5

SEM images showing FIB sample preparation of an APT sample for Cu-filled NPG-Pt, heated at 300 °C in air. a Sample surface (scale bar is 20 μm). b Cut out showing the Cu-filled NPG layer, between the Cu deposition and bulk alloy (scale bar is 2 μm). c A high magnification image showing the fine ligament-pore structure (scale bar is 200 nm). d The APT needle following annular milling (scale bar is 500 nm). e The final APT specimen after low-kV sharpening (scale bar is 200 nm).

Full size image
Fig. 6: Atom map of NPG-Pt layer along with the parent alloy (AgAuPt) after coarsening in air.
figure 6

3D APT atom maps of Cu-filled dealloyed layer on NPG-Pt, air-annealed at 300 °C for 2 h showing Au, Ag, Pt, and Cu atoms. Scale bar is 50 nm. Refer to supplementary video for 3D reconstruction.

Full size image
Fig. 7: Retained ligament-pore structure in NPG-Pt after annealing in air, owing to impeded thermal coarsening.
figure 7

a 3D atom maps showing ligaments of a NPG-Pt air-annealed at 300 °C for 2 h, and b as-dealloyed NPG-Pt (full methodology for spatially-determining the ligament boundaries and data for as-dealloyed NPG-Pt from ref. 8). Scale bar is 20 nm.

Full size image

To understand the effect of the oxidizing environment on the chemical distribution across nanoligaments, chemical mappings are examined. These are shown for a vertical cross-section in Fig. 8. Significant Pt enrichment is observed at ligament surfaces. Compared with the segregates observed in the Ar-H2-annealed NPG-Pt, the Pt segregates in this case are higher in number, yet considerably smaller. These observations agree with the anticipated adsorbate-induced segregation behavior of Pt which, owing to its favorable interaction with O, is expected to segregate to the ligament surfaces. As a result, Au atom movement on nanoligament surfaces is impeded, leading to mitigated thermal coarsening. Figure 8 also reveals a rather homogenized nanoligament composition, where the as-dealloyed core-shell structure appears to be annealed out. This suggests that while the movement of Au atoms on nanoligament surfaces is impeded due to the high surface area coverage of Pt segregates, the Au (and Ag) atoms within the ligament are sufficiently mobile as to eliminate the non-equilibrium core-shell structure.

Fig. 8: Formation of nanoscale Pt surface segregates and elimination of core-shell distribution of AgAu.
figure 8

Concentration maps taken from 5 nm-thick slices of APT volumes of NPG-Pt air-annealed at 300 °C for 2 h, along the longitudinal axis. Scale bar is 30 nm.

Full size image

A precise indication on the degree of coarsening for NPG-Pt in air is obtained by comparing its measured surface area-to-volume ratio with previously measured values for as-dealloyed NPG-Pt. As shown in Table 1, the volume fraction of the nanoligaments remains basically unchanged by air-annealing. However, a decrease of ~17% in the ligament surface-area-to-volume ratio suggests that some coarsening does occur. It is likely that the coarsening is not totally inhibited until the surfaces of the ligaments are saturated with enough Pt-O clusters to completely immobilize the Au at that temperature.

Table 1 Limited degradation of surface area-to-volume ratio after annealing in oxidative environment.
Full size table

A final observation made, requiring investigation in future work, is that the total Pt content in the dealloyed layer appeared to increase by as much as 4 at.% after heating in air. Composition profiles from the APT data, excluding Cu, that spans across the dealloying interface are presented in Fig. 9. These show the Pt content increasing moving across the dealloying interface towards the dealloyed layer. This allows us to suggest, for now, that the source of the extra Pt content might be the bulk layer of non-dealloyed AgAuPt3, which could be considered as a reservoir of Pt atoms that, at high enough temperature, could diffuse across the non-uniform dealloying interface.

Fig. 9: Overall Pt enrichment in NPG-Pt after coarsening in air.
figure 9

1D concentration profiles for NPG-Pt air-annealed at 300 °C for 2 h and as-dealloyed NPG crossing the dealloying interface.

Full size image

Diffusion across the dealloying interface at 300 °C, where minimal lattice diffusion is expected, could be assisted by excessive point defects37 injected at the dealloying front, as per previous discussions on the topic by Rösner and Weissmüller20. Nevertheless, this is yet to be validated through future in-depth studies that could reveal an interesting substrate effect on the functionality of NPG-Pt at high temperatures.

The findings presented in this paper highlight the different roles played by Ag, Au, and Pt during thermal coarsening and how such roles are closely dependent on chemical environment of coarsening. In reducing environments, nanoligaments increase in size due to the surface diffusion of Au. Ag segregates to nanoligament surfaces, highlighting the possible contribution of Ag to the coarsening of nanoligaments in nanoporous layers with residual Ag. Meanwhile, large clusters of Pt are formed on subsurfaces of coarsened ligament. Nanoligaments on the verge of collapsing had immobile Pt segregates at the surface, left behind by the more mobile Ag and Au.

In oxidative environments, thermal coarsening is inhibited owing to the enrichment of Pt at nanoligament surfaces, via the positive interaction between O and Pt. The nanoligament size showed little change as a result. The core-shell structure of as-dealloyed ligaments is also eliminated in oxidative air, even though the change in surface area-to-volume ratio is minimal and surface diffusion of Ag, Au is practically inhibited. Compositional changes across the entire dealloyed layer remain to be further investigated, with particular focus on local chemistry at the dealloying interface.

Methods

Dealloying

Two rectangular samples with nominal surface areas of ~0.8 cm2 were cut out of 200 μm foil of Ag77Au20Pt3 (at.%) procured from Goodfellow Metals, Cambridge, UK. All annealing was done at 975 °C for 15 h in a 2.5% H2-Ar atmosphere. The samples were then soldered to a copper electrical wire from one end (taking care not to alter the active surface area to be dealloyed). A fine coating of Microshield TM lacquer supplied by SPI was applied to the solder-copper-sample junction to insulate it from the solution used. Dealloying solution was prepared with 18 MΩ cm de-ionized water and de-aerated by high-purity N2 purging (min purity: 99.998%). The samples were then dealloyed potentiostatically, at room temperature, in a solution of 0.5 M HClO4 from Analar grade HClO4 (Alfa-Aesar, 62%), at a potential of 550 mV vs. MSE (Mercury/Mercurous Sulfate Electrode/sat. K2SO4), until the removal of ~0.5 C cm−2 was attained. The dealloyed samples were then immersed in de-ionized water for ~1 h to remove residual acid.

After cutting the NPG-Pt samples to detach them from the lacquered wire, they were annealed in two separate conditions, one for 2 h in a reducing atmosphere of Ar-2.5% H2 at 300 °C followed by quenching in furnace by purging Ar-2.5% H2, and the other for 2 h in oxidizing laboratory air at 300 °C followed by air quenching. Then the annealed samples were reattached to a copper electrical wire from one end and coated with an insulating layer of lacquer, as explained above, for pore infiltration by Cu deposition. Pore infiltration was then achieved by stepped potentiostatic deposition of Cu in CuSO4 + H2SO4 solution, following the deposition method explained in a previous APT analysis published recently10.

Atom probe tomography

APT samples were prepared from the Cu-filled dealloyed layer of NPG-Pt3 using a Zeiss NVision40 FIB and standard liftout procedures38. Target areas in the sample, with well-filled dealloyed layers, were identified by cross-sectional FIB cuts, and then extracted with a micro-manipulator. Specimens were then mounted to pre-sharpened Si posts using deposition of W. The samples were then sharpened into needles of suitable scale for APT using the FIB. Initial annular milling was conducted at 30 kV, with final cleaning/sharpening at 10 kV.

APT analysis was conducted using a Cameca local electrode atom probe (LEAP) ×4000 HR (Cameca Instruments, USA). Data were acquired while operating in laser-pulsing mode (λ = 355 nm) with a pulse of 40–60 pJ and a pulse rate of 100–125 kHz. The target evaporation rate was set to 0.003 or 0.005 ions per pulse (0.3 or 0.5%) by adjusting an applied DC voltage (typically ranging from ~2 to 5 kV). The base temperature for the specimen stage during analysis was ~60 K, and the chamber pressure was approximately 10−8 Pa. Reconstruction and analysis were performed using IVAS 3.8 software, using SEM images of the tips to assist with spatial calibration of the reconstructions.