Controlled Dealloying of Alloy Nanoparticles toward Optimization ofElectrocatalysis on Spongy Metallic NanoframesGuangfang Grace Li,† Esteban Villarreal,† Qingfeng Zhang,† Tingting Zheng,†,‡ Jun-Jie Zhu,‡

and Hui Wang*,†

†Department of Chemistry and Biochemistry, University of South Carolina, Columbia, South Carolina 29208, United States‡State Key Laboratory of Analytical Chemistry for Life Science, School of Chemistry and Chemical Engineering, Nanjing University,Nanjing, Jiangsu 210093, China

*S Supporting Information

ABSTRACT: Atomic-level understanding of the structural trans-formations of multimetallic nanoparticles triggered by external stimuliis of vital importance to the enhancement of our capabilities to fine-tailorthe key structural parameters and thereby to precisely tune theproperties of the nanoparticles. Here, we show that, upon thermalannealing in a reducing atmosphere, Au@Cu2O core−shell nanoparticlestransform into Au−Cu alloy nanoparticles with tunable compositionalstoichiometries that are predetermined by the relative core and shelldimensions of their parental core−shell nanoparticle precursors. TheAu−Cu alloy nanoparticles exhibit distinct dealloying behaviors that aredependent upon their Cu/Au stoichiometric ratios. For Au−Cu alloynanoparticles with Cu atomic fractions above the parting limit,nanoporosity-evolving percolation dealloying occurs upon exposure ofthe alloy nanoparticles to appropriate chemical etchants, resulting in the formation of particulate spongy nanoframes with solid/void bicontinuous morphology composed of hierarchically interconnected nanoligaments. The nanoporosity evolution duringpercolation dealloying is synergistically guided by two intertwining structural rearrangement processes, ligament domaincoarsening driven by thermodynamics and framework expansion driven by Kirkendall effects, both of which can be maneuveredby controlling the Cu leaching rates during the percolation dealloying. The dealloyed nanoframes possess large open surfaceareas accessible by the reactant molecules and high abundance of catalytically active undercoordinated atoms on the ligamentsurfaces, two unique structural features highly desirable for high-performance electrocatalysis. Using the room temperatureelectro-oxidation of methanol as a model reaction, we further demonstrate that, through controlled percolation dealloying ofAu−Cu alloy nanoparticles, both the electrochemically active surface areas and the specific activity of the dealloyed metallicnanoframes can be systematically tuned to achieve the optimal electrocatalytic activities.

KEYWORDS: percolation dealloying, nanoporosity, electrocatalysis, nanoframes, undercoordinated surface atoms, alloy nanoparticles

■ INTRODUCTION

Multimetallic nanoparticles (NPs), either heteronanostructuresor homogeneous alloys, may undergo intriguing postsynthesisstructural transformations upon intraparticle atomic migrationstriggered by thermal,1 electrical,2 or chemical stimuli,3

providing a versatile pathway to deliberately tailor thegeometries and thereby fine-tune the optical, electronic, andcatalytic properties of the NPs. The chemical or electro-chemical dealloying of metallic alloy materials, which involvesselective leaching of the less-noble components from the alloymatrices accompanied by structural remodeling of the more-noble components, represents an intriguing structural rear-rangement that entangles multiple surface and bulk atomicdissolution and migration processes over the nanometer lengthscale.4 A prototypical system of particular interest has been thepercolation dealloying of bulk membranes of Au−Ag bimetallicalloys, which results in a unique nanoporous, solid/void

bicontinuous structure consisting of a three-dimensional (3D)network of hierarchically interconnected Au-rich nanoliga-ments.4−6 Strikingly distinct from the bulk Au films that arecatalytically inactive, the dealloyed nanoporous Au membranesexhibit exceptionally high catalytic activities commensurate withthose of the oxide-supported sub-5 nm Au NPs that have longbeen used for heterogeneous catalysis.6,7 The locally curvedsurfaces of the nanoligaments are essentially enclosed by highdensities of undercoordinated surface atoms, which serve as theactive sites for catalyzing a series of interfacial chemical andelectrochemical reactions.7−9

Alloy NPs may undergo dealloying-induced structuraltransformations that are substantially more complicated than

Received: June 16, 2016Accepted: August 25, 2016Published: August 25, 2016

Research Article

www.acsami.org

© 2016 American Chemical Society 23920 DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

those of their macroscopic bulk counterparts displaying a planarsurface to the electrolyte. Upon dealloying, an alloy NP mayevolve into a variety of distinct nanostructures, such as aheterostructure with an alloy core encapsulated by a noblemetal shell,10−13 a sponge-like particle with hierarchicalnanoporosity,10−13 or a skeletal nanoframe (NF) covered bya noble metal skin,14−16 depending on the size, crystallinestructure, compositional stoichiometry, and compositionalgradient of the starting alloy NPs as well as the conditionsunder which the dealloying occurs. While each one of thesestructural transformations gives rise to drastically enhancedcatalytic activities,10,13,14,17,18 the origin of the catalyticenhancements cannot be simply interpreted in the context ofa single unified mechanism because multiple effects interplayand contribute synergistically to the overall catalytic activities.The specific surface area of a NP accessible for catalysis maydrastically increase upon dealloying, especially when a solidalloy NP is converted into either a skeletal NF or a spongynanoporous particle with hollow interior and open surfacestructures.10,14,18 The specific catalytic activity of the NPs, onthe other hand, may also be greatly enhanced upon dealloyingas a consequence of two intrinsically interconnected effects,geometric and electronic effects, both of which are intimatelytied with the nature and density of the active sites on the NPsurfaces.2,19−21 Unraveling the detailed structure−composi-tion−activity correlations that underpin the intriguing catalyticbehaviors of the dealloyed metallic NPs, nevertheless, remainschallenging largely due to the intrinsic structural and composi-tional complexity and diversity of the materials systems as wellas lack of a versatile approach through which both the specific

surface area and specific catalytic activity of a dealloyed NP canbe precisely fine-tuned.Here, we endeavor to push the structural control of dealloyed

metallic nanocatalysts to a new level of precision and versatilitywith the goal of paving an avenue toward the rationaloptimization of electrocatalysis. Dealloyed spongy nanoframes(NFs) represent a particularly interesting geometry with uniquestructural characteristics highly desirable for electrocatalysis.The enormous surface-to-mass and surface-to-volume ratios ofthe spongy NFs are highly advantageous for achieving superiormass activities, which become especially crucial when preciousnoble metals, such as Au, Pt, and Pd, are used as thecatalysts.14,22−24 On the other hand, the locally curved surfacesof the nanoligaments are rich in catalytically active sitesoccupied by coordinatively unsaturated surface atoms.8,9 Usingthe methanol electro-oxidation as a model reaction, wedemonstrate that both the catalytically active surface area andthe density of active sites on the surfaces of spongy NFs can bedeliberately tuned toward the optimization of electrocatalysisthrough controlled percolation dealloying of Au−Cu alloy NPsunder mild conditions.

■ RESULTS AND DISCUSSION

Our synthetic approach to the electrocatalytically active spongyNFs involves two key steps (Figure 1A). We started from Au@Cu2O core−shell NPs, which underwent chemical reductionfollowed by intraparticle alloying25 to evolve into Au−Cu alloyNPs upon thermal annealing in a reducing atmosphere, such asH2. The Au−Cu alloy NPs further transformed into spongyNFs through percolation dealloying when exposed to a

Figure 1. Nanoparticle structural evolution during nanoscale alloying and dealloying. (A) Schemes illustrating the transformation of Au@Cu2Ocore−shell NPs into Au−Cu alloy NPs upon alloying and the transformation of Au−Cu alloy NPs into spongy NFs upon dealloying. (B) SEM imageof Au@Cu2O core−shell NPs (average core diameter: 104 nm; average shell thickness: 51 nm). The inset shows a TEM image of one Au@Cu2Ocore−shell NP. (C) SEM image of Au0.19Cu0.81 alloy NPs. The inset shows a TEM image of one alloy NP. (D) SEM and (E) TEM images of spongyAu0.72Cu0.28 alloy NFs obtained through dealloying of the Au0.19Cu0.81 alloy NPs in 0.5 M HNO3 for 3 h. (F) HAADF-STEM image (left panel) andEDS elemental distribution of Au (upper right panel) and Cu (lower right panel) of an individual Au0.72Cu0.28 alloy NF particle. The compositions ofthe alloy NPs and dealloyed NFs were quantified by EDS.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23921

chemical etchant, such as nitric acid (HNO3). We used acombination of scanning electron microscopy (SEM), trans-mission electron microscopy (TEM), energy-dispersive spec-troscopy (EDS), and powder X-ray diffraction (PXRD) tosystematically track the structural and compositional changes ofthe NPs during the nanoscale alloying and dealloying processes.The Au−Cu2O hybrid NPs exhibited a well-defined core−shellheterostructure clearly resolvable by electron microscopies(Figure 1B) and EDS elemental analysis (Figure S1). After thecore−shell NPs were thermally annealed in a flow of H2 (50sccm) under 100 Torr at 450 °C for 15 min, the contrastbetween the core and the shell in the electron microscopyimages completely disappeared (Figure 1C), indicating theformation of Au−Cu bimetallic alloy NPs. The EDS elementalmapping results (Figure S2) further verified that the Au and Cuatoms were homogeneously intermixed in the alloy NPs.During the percolation dealloying, the alloy NPs furtherevolved into compositionally Au-rich spongy NFs as Cu wasselectively dissolved from the alloy matrix. Figure 1D−Fshowed the SEM, TEM, and high-angle annular dark-fieldscanning electron microscopy (HAADF-STEM) images,respectively, of the spongy NFs obtained through dealloyingof Au0.19Cu0.81 alloy NPs in 0.5 M HNO3 for 3 h at roomtemperature. Each particle exhibited a bicontinuous nano-porous structure composed of nanocurved ligaments that were5 to 25 nm thick. The diameters of the pores were in the rangefrom 10 to 40 nm. While the Cu/Au atomic ratio significantlydecreased after percolation dealloying, the ligaments of thedealloyed NFs were still composed of Au−Cu alloys rather thansegregated monometallic domains (see STEM-EDS mappingresults in Figure 1F and PXRD results in Figure S3).The core and shell dimensions of the starting Au@Cu2O

core−shell NPs could be fine-controlled over a broad size rangeusing a seed-mediated growth method we recently developed,26

which enabled the fine-tuning of the alloy NP compositions.While the overall particle sizes shrank significantly uponalloying, both the quasi-spherical morphology and the Cu/Austoichiometric ratios of the NPs were well-preserved (FigureS4). Therefore, the Cu/Au stoichiometric ratios of the alloyNPs were essentially determined by the relative core and shelldimensions of the Au@Cu2O core−shell NP precursors andcould be systematically tuned over a broad range approximatelyfrom 0.3 to 4. Figure S5A showed the PXRD patterns of Au−Cu alloy NPs with various Cu/Au atomic ratios together withthose of monometallic Au NPs and Cu NPs (Figure S6). Theas-fabricated Au−Cu alloy NPs were essentially composed ofrandom alloys21 rather than ordered Au−Cu intermetallics withspecific Cu/Au stoichiometries, such as Au3Cu, AuCu, orAuCu3.

27,28 The lattice parameters of the face-centered cubic(fcc) Au−Cu alloy NPs were calculated using Bragg’s law

λθ

=d2sin( )hkl

hkl (1)

where λ is the wavelength of the incident X-ray (λ = 1.5406 Åfor Cu Kα), dhkl is the lattice spacing of the crystalline planewith Miller index of {khl}, and θhkl is the angle of incidence onthe {khl} plane. We calculated the Cu/Au stiochiometric ratiosof various Au−Cu alloy NPs based on the position of the (111)diffraction peak, assuming that the lattice parameters of a solidsolution agree with a linear relationship between the two endpoints of alloying elements, an empirical rule known asVegard’s law.29 Although some bimetallic alloys may deviate

significantly from Vegard’s law, Au−Cu binary random alloysonly exhibit slight deviations from Vegard’s law.30 The Cu/Auatomic ratios calculated from the PXRD results were in verygood agreement with those quantified by inductively coupledplasma mass spectrometry (ICP-MS) and EDS (Figure S5B).Colloidal Au@Cu2O core−shell NPs exhibited strong dipolarplasmon resonances that progressively red-shifted as thethickness of the Cu2O shell increased26,31,32 (Figure S7A).The transformation of core−shell NPs into alloy NPsintroduced drastic modifications to the resonance frequenciesand spectral line-shapes of the NP plasmons (Figure S7B). Asthe Cu/Au stoichiometric ratio and the particle size increased,the plasmon resonance of the alloy NPs red-shifted,accompanied by significant peak broadening due to theplasmon damping caused by alloying of Au with Cu.33−35

The key extinction spectral features observed experimentallywere well reproduced by the Mie scattering theory calculations(Figure S7C,D).The Au−Cu alloy NPs exhibited interesting composition-

dependent chemical dealloying behaviors (Figure S8). Uponexposure to 0.5 M HNO3 at room temperature, Cu-rich alloyNPs with a Cu atomic percentage (at %) above ∼70%underwent nanoporosity-evolving percolation dealloying, dur-ing which the Cu/Au atomic ratios progressively deceasedaccompanied by ligament thickening and pore volumeexpansion until complete leaching of Cu within a few hours.In striking contrast, Au-rich alloy NPs (Cu at % < ∼65%)exhibited unnoticeable morphological and compositionalchanges even over a few days under identical dealloyingconditions. Such composition-dependent dealloying behaviorsof Au−Cu alloy NPs can be interpreted in the context of theparting limit and critical potential for the percolationdealloying.11

For a macroscopic A1‑pBp binary alloy (A and B represent themore-noble and the less-noble elements, respectively, and p isatomic fraction of B), percolation dealloying occurs only whenthe p is higher than a threshold value known as the partinglimit. In Ag−Au alloys, this parting limit was measured to be∼55 at % Ag.4 The critical potential, Ec, which is theelectrochemical parameter signifying the onset of percolationdealloying, is related to the molar volume of A, ΩA, the B/electrolyte interfacial free energy, γB/elec, and the equilibriumpotential, Eeq, above which the surface dealloying at thetopmost atomic layer occurs. The relationship between Ec andEeq is given by36

γ

ξ = +

ΩE p E p

nF( ) ( )

4c eq

B/elec A

(2)

where n is the number of electrons transferred upon oxidationof 1 atom of B, F is the Faraday constant, and ξ represents thelocal radius of the surface where a cylindrical pit is created atthe initial stage of nanoporosity evolution. ξ is related to both pand the nearest-neighbor spacing, a, as shown by

ξ =+−

×pp

a11 (3)

For macroscopic bimetallic alloys, both the Ec and Eeq arefunctions of the alloy composition and the Ec values are morepositive than Eeq when the solid-state mass transport is slowerthan the imposed rate of dealloying. At a potential above the Ec,nanoporosity-evolving percolation dealloying occurs, whilebelow Ec only superficial dealloying can occur, forming a

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23922

surface passivating atomic layer of the more-noble componentwhich inhibits the nanoporosity formation.37,38

For a spherical A1‑pBp alloy NP of radius r, the criticalpotential becomes dependent not only on the composition butalso on the size of the NP. Accordingly, the Ec(p,r) of the alloyNP is given by11

γ= − Ω + Ω − Ω × ⎜ ⎟⎛⎝

⎞⎠E p r E p f

nFr( , ) ( ) [ ( ) ( )]

2c c Alloy A Alloy A

(4)

where γAlloy and fAlloy represent the free energy and the stress atthe alloy/electrolyte interface, respectively. ΩA is the partialmolar volume of A in the alloy, and ⟨Ω⟩ is the average molarvolume of the alloy. When r is greater than ∼5 nm, eqs 2 and 4generate virtually the same results because the maximum valuesof γAlloy and fAlloy are ∼2 and ∼6 J m−2, respectively.39

Therefore, alloy NPs beyond ∼10 nm in size typically undergodealloying-induced structural transformations analogous tothose of their bulk counterparts with the same compositions.It was shown that Au−Ag alloy NPs larger than 10 nm evolvedinto spongy nanoporous structures whereas their sub-10 nmcounterparts transformed into core−shell NPs under identicaldealloying conditions.11 Since the particle sizes under thecurrent investigations were far beyond 10 nm, the Au−Cu alloyNPs evolved into spongy NFs when the Cu content was higherthan the parting limit, which was determined to be ∼70 at % ofCu. Dealloying of the Au0.31Cu0.69 alloy NPs resulted in amixture of solid alloy NPs and spongy NFs owing to theintrinsic particle-to-particle compositional variations within thesample (Figure S8C).The composition-dependent electrochemical dealloying

behaviors of the Au−Cu alloy NPs were further investigatedby linear sweep voltammetry (LSV) measurements in 1.0 MKNO3 electrolyte (pH = 7) at a sweep rate of 5.0 mV s−1.During the positive potential sweep, the anodic currentremained at a low level until reaching the Ec, at which pointthe anodic current started to increase rapidly due to percolationdealloying. The onset potential for Cu dissolution from theAu0.19Cu0.81 NPs was positively shifted with respect to that ofmonometallic Cu NPs (Figure S9A) because the less-noble Cuwas significantly stabilized when it was alloyed with the more-noble Au, a well-known characteristic of bulk binary alloymaterials4 that is yet still less explored in alloy NP systems.40,41

For Au−Cu alloy NPs with Cu at % below the parting limit,however, only superficial dealloying could occur at the topmostatomic layer on the NP surfaces due to surface passivation by aAu-rich atomic layer or an oxide adlayer. As a consequence, theanodic peak currents decreased dramatically by more than 2orders of magnitude with the onset potential for dealloyingshifted to significantly more positive values. Consistent with thestructural evolutions upon chemical dealloying, the electro-chemical percolation dealloying resulted in nanoporous spongyNFs as well, whereas the surface dealloying of the alloy NPswith Cu at % below the parting limit did not introduce anysubstantial morphological changes observable by TEM (FigureS9B−D). The Ec was found to be dependent not only on theNP compositions but also on the pH of the electrolytes. In anacidic electrolyte, such as H2SO4, both the onset potential forpercolation dealloying and the anodic peak potential negativelyshifted with respect to those in the neutral electrolytes (FigureS10A). However, the alloy NPs became much more resistiveagainst percolation dealloying in an alkaline environment, e.g.,KOH electrolyte, possibly due to the surface passivation by the

oxide surface layers that formed at sufficiently high potentials(Figure S10B).The nanoporosity evolution during percolation dealloying is

essentially a multiscale structural rearrangement governed bytwo intertwining and competing processes, ligament domaincoarsening and framework expansion. The percolation deal-loying was initiated upon the dissolution of a Cu atom at thealloy surface, leaving behind a terrace vacancy coordinated withundercoordinated Cu atoms that were more susceptible tofurther dissolution. As the entire terrace was stripped, thecoordinatively unsaturated Au atoms leaving behind thedealloying frontiers underwent fast surface migration toagglomerate into Au-rich local islands. Therefore, the surfaceof the alloy upon initiation of dealloying was comprised of Au-rich domains that locally passivated the surface and patches ofundealloyed material directly exposed to electrolyte.4 As thepercolation dealloying further proceeded, the interfacialdissolution of Cu atoms and coarsening of the Au-rich domainscontinued, gradually evolving into bicontinuous spongystructures. Meanwhile, the outward migration of Cu atomswas faster than the inward migration of Au atoms in the alloymatrix,42 providing an additional contribution to the expansionof the pores during the nanoporosity evolution. This effect,known as the Kirkendall effect, has been harnessed to fabricatehollow NPs with fine-tailored interior structures.43−46 For analloy NP, the ligaments domain coarsening resulted in thethickening of the ligaments and shrinkage of the overall particlesize, whereas the framework expansion driven by the Kirkendalleffect led to increased overall particle size and pore volumes.The relative rates of the ligament domain coarsening andframework expansion could be maneuvered by tuning the ratesof Cu leaching during the percolation dealloying.We systematically studied the kinetics and thermodynamics

of the Cu leaching from Au0.19Cu0.81 alloy NPs duringpercolation dealloying in the presence of three differentchemical etchants, HNO3, Fe(NO3)3, and ammonia, in anaqueous environment under ambient conditions (Figure S11).Both HNO3 and Fe(NO3)3 are oxidative etchants that directlyoxidize metallic Cu and selectively dissolve the Cu from thealloy NPs, thereby giving rise to relatively fast Cu leaching uponpercolation dealloying. The leaching of Cu from the alloy NPsexposed to ammonia, however, is essentially a consequence ofoxidation of Cu by the oxygen dissolved in the aqueoussolution followed by the formation of water-soluble Cu-(NH3)4

2+ and Cu(NH3)2+ complexes.47 Therefore, the rate of

Cu leaching upon dealloying in ammonia appeared muchslower than those in HNO3 and Fe(NO3)3. The Cu/Au atomicratios progressively decreased as the percolation dealloyingproceeded until reaching a thermodynamic equilibrium point.Because of their different standard redox potentials, HNO3,Fe(NO3)3, and O2/ammonia resulted in different equilibriumCu/Au stoichiometries. While complete Cu leaching (less than3 at % residual Cu) was achieved after dealloying within 1 h in2.0 M HNO3, Fe(NO3)3 and O2/ammonia only resulted inpartial leaching of Cu when reaching the thermodynamicequilibria. These partially dealloyed NFs obtained in Fe(NO3)3or O2/ammonia underwent further Cu leaching upon exposureto HNO3 (Figure S12).Although the Cu leaching from the Au0.19Cu0.81 alloy NPs

under the nonequilibrium conditions was a continuous process,further dealloying of the partially dealloyed NFs could beeffectively inhibited by separating the particles from theetchants through centrifugation and redispersion in water.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23923

This allowed us to kinetically trap partially dealloyed NFs withfine-controlled pore volumes, ligament thicknesses, and Cu/Austoichiometries. As shown in Figure 2, the degree of Culeaching and the structural features of the dealloyed NFs couldbe systematically tuned by changing the etchants or theconcentration of each etchant at fixed dealloying times. Whenfixing the dealloying time at 1 h, increasing the HNO3

concentration resulted in more Cu leaching, larger averagepore sizes, thicker ligaments, and smaller overall sizes of thedealloyed NFs. In HNO3, the leaching of Cu from the alloyNPs was faster than the atomic diffusion of Cu in the alloymatrix. Therefore, the nanoporosity evolution was dominatedby the ligament domain coarsening rather than frameworkexpansion, leading to significant thickening of the ligaments andshrinkage of the overall size of the dealloyed NFs as anincreasing amount of Cu was dissolved. A similar volumeshrinkage of macroscopic Ag−Au alloy bulk materials by up to30 vol % was previously observed during fast electrochemicaldealloying as a consequence of the lattice defect formation and

local plastic deformation both associated with ligamentcoarsening.48 When the percolation dealloying occurred inammonia, however, the overall particle sizes were observed toincrease with the ammonia concentration when the dealloyingtime was fixed at 2 h. In ammonia, as Cu leaching becameslower than the atomic diffusion of Cu, the Kirkendall effectsstarted to dominate the nanoporosity evolution, which resultedin the expansion of the particle frameworks. In Fe(NO3)3, therates of Cu leaching and atomic diffusion of Cu werecomparable, and thus, the overall size of the dealloyed NFsremained almost unchanged regardless of the concentration ofFe(NO3)3 and the amount of Cu leached during thepercolation dealloying.We used high-resolution HAADF-STEM to fine-resolve the

surface and bulk structures of the dealloyed NFs at the atomiclevel. As shown in Figure 3A, a representative particle in thedealloyed Au0.97Cu0.03 NF sample (obtained through dealloyingof Au0.19Cu0.81 alloy NPs in 2.0 M HNO3 for 1 h) exhibited abicontinuous architecture consisting of hierarchically inter-

Figure 2. Compositional and structural evolution of Au0.19Cu0.81 alloy NPs during percolation dealloying. TEM images of spongy NFs obtained afterexposure of Au0.19Cu0.81 alloy NPs to (A) 0.2 M, (B) 0.5 M, (C) 1.0 M, and (D) 2.0 M HNO3 for 1 h. (E) Cu at % (quantified by EDS) and overallparticle diameters of spongy NFs obtained after exposure of Au0.19Cu0.81 alloy NPs to various concentrations of HNO3 for 1 h. TEM images ofspongy NFs obtained after exposure of Au0.19Cu0.81 alloy NPs to (F) 0.2 M, (G) 0.5 M, (H) 1.0 M, and (I) 2.0 M Fe(NO3)3 for 2 h. (J) Cu at % andoverall particle diameters of spongy NFs obtained after exposure of Au0.19Cu0.81 alloy NPs to various concentrations of Fe(NO3)3 for 2 h. TEMimages of spongy NFs obtained after exposure of Au0.19Cu0.81 alloy NPs to (K) 2.0 M, (L) 5.0 M, (M) 8.0 M, and (N) 14.0 M NH4OH for 2 h. (O)Cu at % and overall particle diameters of spongy NFs obtained after exposure of Au0.19Cu0.81 alloy NPs to various concentrations of ammonia for 2 h.All TEM images were taken with the same magnification. The error bars of the Cu atomic percentages represent the standard deviations of the EDSresults on 3 samples fabricated under exactly the same conditions. The average particle diameters and standard deviations were obtained from morethan 100 particles in the TEM images.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23924

connected nanoligaments whose compositions were dominatedby Au. Panels A-a and A-b show the high-resolution HAADF-STEM images of two regions in Panel A labeled as a and b,respectively, with the electron beam projected along the [011 ]zone axis of the crystalline domains. The corresponding fastFourier transform (FFT) patterns further confirmed theorientation of the crystalline domains in the ligaments. Thelattice fringes corresponding to the face center cubic phase ofAu were well resolved in the high-resolution HAADF-STEMimages. No segregated monometallic Cu domains wereobserved either in the bulk or on the surfaces of the ligaments,suggesting that the residual Cu was randomly intermixed withAu. When the atomic fraction of the residual Cu was below 3%,the Cu signals in EDS became indistinguishable from the noise,making it difficult to characterize the spatial distribution of Cuelements in the ligaments through EDS elemental mapping.However, the residual Cu was clearly resolvable by X-rayphotoelectron spectroscopy (XPS), which is a sensitive surfaceanalysis technique. While the binding energies of the Au 4fpeaks slightly up-shifted by ∼0.05 eV in comparison to those ofthe bulk Au, the Cu 2p XPS peaks exhibited more pronounced

down-shifts by ∼0.5 eV with respect to those of the bulk Cudue to the alloying of Cu with Au49 (Figure S13). The surfaceCu/Au atomic ratio was quantified to be ∼0.02 by XPS, whichwas in agreement with the bulk Cu/Au atomic ratio obtainedfrom EDS. The XPS and EDS results showed that the residualCu in the fully dealloyed NFs was randomly intermixed withthe Au matrix rather than being segregated on the ligamentsurfaces. The residual Cu was highly resistive against leaching.Further exposure of the Au0.97Cu0.03 NFs to 2.0 M HNO3 for 12h did not result in any observable decrease in the surface Cu/Au atomic ratios according to the XPS results (Figure S14).The surfaces of the ligaments were enclosed by a mixture of

both high-index and low-index local facets (Figure 3A). Thelocally flat surface regions of a monocrystalline domain weretypically enclosed by low-index {111} facets, whereas in thelocally curved regions, the surfaces were terminated with high-index facets with high densities of undercoordinated surfaceatoms at the atomic steps, kinks, and terrace edges.Coordinatively unsaturated surface atoms were also present atthe boundaries between two twinned crystalline domains. Thehigh abundance of undercoordinated surface atoms is a unique

Figure 3. Atomic-level surface structures and electrocatalytic activity of fully dealloyed spongy NFs. (A) HAADF-STEM image of an individualAu0.97Cu0.03 NF particle. High-resolution HAADF-STEM images showing the atomic-level structures of regions a and b are shown in panel A-a andpanel A-b, respectively. The insets in panels A-a and A-b are the FFT patterns of the regions labeled as i, ii, and iii, respectively. In the high-resolutionHAADF-STEM images, the crystalline domains were projected along the [011 ] zone axis. SEM images of (B) a Au SRNP with solid interior coreand (C) a Au QSNP. (D) CV curves of Au0.97Cu0.03 NFs, Au SRNPs, and Au QSNPs in 0.5 M H2SO4 at a potential sweep rate of 5.0 mV s−1. Theelectrochemical signals from SRNPs and QSNPs were multiplied by a factor of 5 for clearer comparison. (E) CV curves of Au0.97Cu0.03 NFs, AuSRNPs, and Au QSNPs in 1.0 M methanol and 0.5 M KOH at a potential sweep rate of 10 mV s−1. CV curve of Au0.97Cu0.03 NFs in the absence ofmethanol is shown as the dashed curve. (F) Mass activities (MAs), electrochemically active surface areas (ECSAs), and specific activities (SAs) ofAu0.97Cu0.03 NFs, Au SRNPs, and Au QSNPs in 1.0 M methanol and 0.5 M KOH at a potential sweep rate of 10 mV s−1.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23925

feature of NPs with highly curved surfaces, such as thecatalytically active sub-5 Au NPs50 and Au surface-roughenednanoparticles (SRNPs). We synthesized Au SRNPs with overallparticle diameters of 120 ± 7.2 nm using a seed-mediatednanocrystal growth method previously developed by ourgroup.51 The Au SRNPs exhibited nanoscale surface roughness(Figure 3B) with high densities of undercoordinated atoms atthe locally curved surface sites (see more detailed structuralcharacterizations in a previously published paper from ourgroup52). The Au SRNPs exhibited much smaller specificsurface areas than the dealloyed NF particles because of theirsolid interiors. In contrast to the Au SRNPs and dealloyed NFs,the surfaces multifaceted Au quasi-spherical nanoparticles(QSNPs, diameter of 104 ± 6.5 nm) were essentiallydominated by low-index facets bound with close-packed surfaceatoms (Figure 3C) with only a small fraction of under-coordinated surface atoms present at the particle corners, edges,crystalline boundaries, and surface defects.We used cyclic voltammetry (CV) as an electrochemical tool

to further characterize the atomic-level surface structures andspecific surface areas of the dealloyed NFs, Au SRNPs, and AuQSNPs. CV-based electrochemical oxide stripping is a methodcommonly used to study the surface structures of bulk Au53,54

and more recently Au NPs55−58 as well. Figure 3D showed theCV curves of various samples obtained in 0.5 M H2SO4 at asweep rate of 5.0 mV s−1. During the positive potential scans ofthe CV measurements, both the dealloyed Au0.97Cu 0.03 NFsand Au SRNPs exhibited oxidation peaks in the range of 0.95−1.25 V vs saturated calomel electrode (SCE), whereas AuQSNPs exhibited an oxidation peak in higher potential range

above 1.40 V (vs SCE). The negative shift of oxidation peaksobserved on the dealloyed NFs and SRNPs suggested that thesurfaces of the NFs and SRNPs were easier to get oxidized toform an oxide surface layer due to the presence of highlyabundant undercoordinated surface atoms while the close-packed surface atoms on Au QSNPs were more resistive againstoxidation. During the negative potential scans, the sharpreduction peak centered around 0.9 V (vs SCE) signified theelectrochemical stripping of the surface oxide layers formed inthe previous positive potential scan. Although the onsetpotentials of the surface atom oxidation were sensitivelydependent on the surface atomic coordination numbers, theelectrochemical stripping of the gold oxide surface layersoccurred in essentially the same potential range from ∼0.98 to∼0.85 V (vs SCE), which was in line with previous observationson Au bulk films53,54 and NPs.55−58 Assuming the specificcharge associated with gold oxide stripping to be 450 μCcm−2,59 the mass-specific electrochemically active surface area(ECSA) of the dealloyed NFs was estimated to be ∼20 m2 g−1,approximately 1 order of magnitude higher than that of SRNPsand 2 orders of magnitude higher than that of the QSNPs,respectively, despite the fact that the dealloyed NFs, SRNPs,and QSNPs all had similar overall particle sizes.We quantitatively compared the electrocatalytic activities of

the dealloyed NFs, SRNPs, and QSNPs using the room-temperature electrocatalytic methanol oxidation reaction(MOR) in an alkaline aqueous environment as a modelreaction. As shown in Figure 3E, while the Au QSNPs exhibitedpoor electrocatalytic activity toward MOR, both the dealloyedAu0.97Cu0.03 NFs and Au SRNPs were much more active. The

Figure 4. Electrocatalytic performance of dealloyed NFs for MOR. (A) CV curves of MOR on Au0.19Cu0.81 alloy NPs and various dealloyed NFs in1.0 M methanol and 0.5 M KOH at a potential sweep rate of 10 mV s−1. The NF samples labeled as NF-i, NF-ii, NF-iii, and NF-iv correspond to thesamples obtained through dealloying of Au0.19Cu0.81 alloy NPs in 0.2, 0.5, 1.0, and 2.0 M Fe(NO3)3 for 2 h, respectively. The sample labeled as NF-vwas obtained through dealloying of Au0.19Cu0.81 alloy NPs in 2.0 M HNO3 for 1 h. HAADF-STEM images of one representative particle for each NFsample are also shown as the insets. All the HAADF-STEM images share the same scale bar. (B) Cu atomic percentages (quantified by EDS andICP-MS), MAs, ECSAs, and SAs of Au−Cu alloy NPs and various dealloyed NFs. (C) CA curves collected on Au0.19Cu0.81 alloy NPs and variousdealloyed NFs for MOR at 0.1 V (vs SCE) and 0.3 V (vs SCE). All CA measurements were carried out in solutions containing 0.5 M KOH and 1.0M methanol deoxygenated with N2. The error bars of the particle compositions and the CA results represent the standard deviations obtained from 3samples. The error bars of the MAs and ECSAs represent the standard deviations obtained from 5 samples.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23926

onset oxidation potential and peak potential for MOR werearound 0.1 V (vs SCE) and 0.3 V (vs SCE), respectively. Bynormalizing the oxidation peak current against the mass of Auon each electrode, a mass activity (MA) of ∼12 μA μg−1 wasobtained at 0.3 V on the dealloyed Au0.97Cu0.03 NFs at a sweeprate of 10.0 mV s−1, which was about 1 order of magnitudehigher than that of the SRNPs. For Au, Pt, and Pd NPs, theundercoordinated surface atoms have been identified to be theelectrocatalytically active sites for MOR and other alcoholoxidation reactions.2,60 Therefore, the superior electrocatalyticactivity toward MOR observed on the dealloyed Au0.97Cu0.03NFs originated from both their large specific ECSA and highdensity of surface active sites. By normalizing the MAs againstECSAs, the specific activities (SAs) of the NPs were obtained,which were directly related to the active site density on the NPsurfaces. In Figure 3F, we compared the MAs, ECSAs, and SAsof the dealloyed Au0.97Cu0.03 NFs with those of the Au SRNPsand QSNPs. Interestingly, the Au SRNPs exhibited a SA onlyslightly lower than that of dealloyed NFs, indicating comparabledensities of surface active sites, i.e., undercoordinated surfaceatoms, on the dealloyed NFs and Au SRNPs. However, thedealloyed NFs exhibited much higher MA than Au SRNPsbecause of their significantly larger ECSA. For a low-indexfaceting Au QSNP, the undercoordinated surface atoms arelocated at the particle corners and edges and thus only accountfor a vanishingly small fraction of the surface atoms when theparticle size becomes larger than 10 nm.61 Therefore, the AuQSNPs exhibited diminished electrocatalytic activity for MORdue to their limited SA and ECSA.The controlled percolation dealloying of Au−Cu alloy NPs

allowed us to systematically tune both the ECSAs and the SAsof the dealloyed NFs to achieve the optimal MAs for MOR. InFigure 4A, we compared the MAs of 5 dealloyed NF sampleswith different degrees of Cu leaching. The NF samples labeledas NF-i, NF-ii, NF-iii, and NF-iv corresponded to the samplesobtained through dealloying of Au0.19Cu0.81 alloy NPs in 0.2,0.5, 1.0, and 2.0 M Fe(NO3)3 for 2 h, respectively. The samplelabeled as NF-v was obtained through dealloying of Au0.19Cu0.81alloy NPs in 2.0 M HNO3 for 1 h. The HAADF-STEM imageshighlighting the key structural features of individual NFparticles representing each sample were also shown as theinsets of Figure 4A. From NF-i to NF-v, the Cu/Austoichiometric ratios decreased (Figure 4B, top panel) whileboth the average ligament thicknesses and pore sizes increased.All the NF samples exhibited much higher MAs than the solidalloy NPs, and NF-ii exhibited the highest MA among the NFsamples. Dealloyed spongy NFs exhibited drastically higheractivities than the solid Au−Cu alloy NPs with similar Cu/Austoichiometric ratios (Figure S15), strongly indicating that theenhanced MAs of the dealloyed NFs were primarily determinedby the NF surface areas and the densities of surface active sitesrather than the Cu/Au stoichiometric ratios.We semiquantitatively assessed the ECSAs of various NF

samples based on the electrochemical oxide stripping results.For the Au−Cu alloy NFs, Cu leaching occurred during theelectrochemical oxide stripping experiments at relatively slowpotential sweep rates (e.g., 5 mV s−1), which prevented us fromobtaining meaningful ECSA results. Therefore, to accuratelymeasure the ECSAs of the Cu-containing alloy NFs, the CVmeasurements were carried out at much faster potential sweeprates to effectively suppress the kinetically slow Cu leachingprocess. As demonstrated on Au0.97Cu0.03 NFs (no further Culeaching during CV scans), the ECSA values measured at

various potential sweep rates in the range from 5 to 500 mV s−1

were highly consistent (Figure S16), though the multipeakedfeatures in the potential range of 1.0−1.3 V (vs SCE) signifyingthe electrochemical oxidation of the undercoordinated surfaceatoms became less well-resolved as the potential sweep rateincreased. At sufficiently fast potential sweep rates (e.g., 500mV s−1), the Cu leaching from Au0.30Cu0.70 alloy NFs waseffectively suppressed while the ECSA remained almostunchanged during multiple cycles of CV scans (Figure S17).In Figure 4B, we further compared the ECSAs and SAs ofvarious NF samples and the solid alloy NPs. Upon thepercolation dealloying, the ECSA first increased and thendecreased as an increasing amount of Cu was dissolved and NF-ii exhibited the largest ECSA among the NF samples. The SA ofthe dealloyed NFs, on the other hand, reached it’s maximumvalue upon initiation of nanoporosity formation and keptdecreasing as the Cu leaching further proceeded during thenanoporosity evolution. At the initiation stage of percolationdealloying, the surface pitting upon Cu atomic dissolutioncreated highly abundant undercoordinated surface atoms thatcontributed to the high SA of the NFs. As more Cu wasdissolved from the alloy matrix, the ligament domaincoarsening drove the migration of the undercoordinated surfaceatoms to form thermodynamically more stable local facets. As aconsequence, the active sites became less abundant, resulting indecreased SA. Therefore, the optimization of the MAs on thespongy NFs requires deliberate tuning of both the ECSA andSA of the NFs, which could be achieved through the controlledpercolation dealloying of Au−Cu alloy NPs as demonstrated inthis work.Among the various dealloyed NF samples, NF-ii exhibited

the highest MA and ECSA while NF-i exhibited the highest SA(Figure 4B). It was recently reported that nanoporous Aushells56 and Au bowls62 overgrown on sacrificial AgCl templatesexhibited MAs approximately 1.4 and 2.3 times higher than thatof the macroscopic dealloyed nanoporous Au foams,9

respectively, for the electrocatalytic MOR in alkaline electro-lytes. NF-ii was about 3 times better than the nanoporous Aubowls and 5 times better than the nanoporous Au shells,respectively, in terms of the MAs normalized to Au mass. TheECSA of NF-ii was about 2.5 times higher than the largestECSAs achievable on the nanoporous Au shells and bowls.56,62

The SAs of the dealloyed spongy NFs, macroscopic nano-porous Au foams, and the nanoporous Au shells and bowls,however, were all comparable to each other possibly due tosimilar densities of the undercoordinated atoms on their highlycurved surfaces.To assess the electrocatalytic durability of the spongy NFs,

chronoamperometry (CA) measurements were carried out atboth the onset oxidation potential (0.1 V, vs SCE) and theoxidation peak potential (0.3 V, vs SCE). As shown in Figure4C, the oxidation currents underwent a fast decay in the firstfew seconds followed by a much slower decay over minutesuntil reaching a steady-state plateau. The fast decay of theoxidation currents arose from the development of theelectrochemical double-layer after a potential was applied onthe samples until reaching the equilibrium after a few seconds.The slower current decay was found to be associated with thesurface structural remodeling of the NFs during electrocatalyticreactions, which resulted in the activity deterioration to certainextent. Although the nanoporous morphology of theAu0.30Cu0.70 NFs (NF-ii) was well-preserved with a limitedamount of Cu further leaching before the steady-state current

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23927

was reached, thickening of the ligaments and expansion of poresizes were clearly observed (Figure S18) as a consequence ofsurface atomic migrations. Such surface structural remodelingduring electrocatalytic reactions, which was previously alsoobserved on dealloyed macroscopic Au membranes,9 resultedin a decrease of both the ECSAs and densities of active sites onthe NF surfaces (Figure S19). The spongy NFs obtainedthrough electrochemical dealloying exhibited significantly lowerelectrocatalytic activities than their chemically dealloyedcounterparts (Figure S20), further indicating that the under-coordinated surface atoms located at the catalytic active sitescould migrate to form thermodynamically more stable butcatalytically less active surface structures when a potential biaswas applied to the NFs.

■ CONCLUSIONSThe percolation dealloying of Au−Cu alloy NPs is a multiscalestructural remodeling process involving nanoscale atomicdissolution and migration that led to the formation of spongyNFs with unique structural characteristics highly desired forelectrocatalysis. The solid-void bicontinuous NFs consisting ofhierarchically interconnected nanoligaments are found to be aunique structure generated from percolation dealloying of Au−Cu alloy NPs with Cu at % above the parting limit. Thenanoporosity evolution during percolation dealloying issynergistically guided by two intertwining and competingstructural rearrangement processes, ligament domain coarsen-ing driven by thermodynamics and framework expansion drivenby Kirkendall effects, both of which can be maneuvered bycontrolling the Cu leaching rates during the percolationdealloying. The controlled percolation dealloying of Au−Cualloy NPs provides a unique way to systematically tune both thecatalytically active surface areas and the surface densities ofactive sites of the dealloyed NFs such that the optimalelectrocatalytic activity can be achieved. The undercoordinatedsurface atoms, which serve as the catalytically active sites,undergo nanoscale surface migrations over a time scale ofminutes during electrocatalytic MOR until reaching the steady-state catalytic currents, resulting in structural remodeling of theNFs that causes partial deterioration of the catalytic activities.Development of new approaches to further stabilize the surfaceactive sites on the dealloyed NFs through either incorporationof structure-stabilizing residual components or deliberateintroduction of compositional gradients to the nanoligamentsis currently underway with the ultimate goal of retaining thesuperior mass-specific electrocatalytic activities of the spongyNFs over extended time periods for direct methanol fuel cell(DMFC) applications.

■ EXPERIMENTAL DETAILSChemicals and Materials. Cu(NO3)2, Fe(NO3)3, and poly-

vinylpyrrolidone (PVP, average molecular weight: 58 000) werepurchased from Alfa Aesar. Chloroauric acid (HAuCl4·4H2O),HNO3 (65%), N2H4·3H2O solution (35 wt %), H2SO4 (98%), andNafion perfluorinated resin solution (5 wt %) were purchased fromSigma-Aldrich. K2CO3 and 37% formaldehyde were purchased fromJ.T. Baker. Methanol, NaOH, KNO3, and KOH were purchased fromFisher Scientific. Ammonium hydroxide (28−30%) was purchasedfrom British Drug Houses. Ultrapure Milli-Q water with a resistivity of18.2 MΩ (Millipore) was used for all the experiments.Synthesis of Au QSNPs and SRNPs. Au QSNPs (diameter of

104 ± 6.5 nm) were synthesized through reduction of HAuCl4 byformaldehyde in the presence of K2CO3 at room temperature.26,31 AuSRNPs with overall particle diameters of 120 ± 7.2 nm were

synthesized through seed-mediated NP growth following a protocolwe recently published.51,52

Synthesis of Au@Cu2O Core−Shell NPs. Au QSNPs (diameterof 104 ± 6.5 nm) were used as the core materials for the fabrication ofAu@Cu2O core−shell NPs following a previously publishedprotocol.26 A varying amount (0.3−3 mL) of 0.1 M Cu(NO3)2solution was added into 300 mL of 2 wt % PVP aqueous solutioncontaining colloidal Au QSNPs. The reaction mixtures weretransferred into an ice bath, and then, 0.67 mL of 5 M NaOH and0.3 mL of N2H4·3H2O solution were added under magnetic stirring.After 10 min, the NPs were centrifuged (2000 rpm, 10 min), thenwashed with water, and finally redispersed in ethanol. The thickness ofthe Cu2O shells could be systematically tuned by adjusting the molarratios between Au QSNPs and Cu(NO3)2. More details of the core−shell NP synthesis can be found in a paper previously published by ourgroup.26

Synthesis of Au−Cu Alloy NPs. Au−Cu alloy NPs were preparedthrough thermal annealing of the Au@Cu2O core−shell NPs at 450°C in a flow of H2 flow (50 sccm) under 100 Torr for 15 min in a tubefurnace. The annealed samples were collected after cooling down toroom temperature and were redispersed in 10 mL of water. It wasrecently revealed by thermal gravimetric analysis that PVP in variousnanocomposites underwent rapid thermal degradation at temperaturesabove ∼420 °C.63−65 Therefore, thermal annealing at 450 °C enabledus not only to obtain fully alloyed NPs but also to effectively removethe residual PVP possibly present on the NP surfaces (the synthesis ofAu@Cu2O core−shell NPs involved the use of PVP).

Synthesis of Spongy NFs. Spongy NFs were fabricated throughchemical dealloying upon the introduction of 1 mL of a chemicaletchant, such as HNO3, Fe(NO3)3, or NH4OH, into 200 μL ofcolloidal Au−Cu alloy NPs at room temperature. After certaindealloying times, the dealloyed NFs were separated from the etchantsby centrifuging the NF particles and redispersing them in water.

Electrochemical Measurements. Electrochemical measure-ments, including CV, LSV, and CA, were carried out using a CHI660E workstation (CH Instruments, Austin, Texas) at roomtemperature. We used a standard three-electrode system, which wascomposed of a glassy carbon electrode (GCE) serving as the workingelectrode, a saturated calomel electrode (SCE) serving as thereference, and a Pt wire serving as the auxiliary. The GCE (3 mmdiameter, CH Instruments, Austin, Texas) was first polished with 0.3mm alumina slurry and then thoroughly washed with water andethanol before use. Dry powders of spongy NFs, Au−Cu alloy NPs, AuQSNPs, or Au SRNPs with certain total masses were first redispersedin ethanol to form colloidal suspensions (2.0 mg of Au in 1.0 mL ofwater), and then, 2 μL of the colloidal ink was drop-dried on eachpretreated GCE at room temperature. Finally, 2 μL of Nafion solution(0.2 wt %) was drop-dried to hold the NPs on the electrode surfaces.The Au mass of the NPs loaded on each GCE was kept at 4.0 μg forcomparison of the MAs and ECSAs of various samples. In a typicalelectrochemical test, CV scans were performed in a 0.5 M KOHsolution with or without 1.0 M CH3OH degassed with N2 at a sweeprate of 10 mV s−1. CV measurements for oxide stripping wereconducted in N2-purged 0.5 M H2SO4 solution at various potentialsweep rates in range of 5−500 mV s−1. The polarization trace wasnormalized against the Au mass of the NPs loaded on each GCE.

Structural and Compositional Characterizations. TEMimaging of the NP samples drop-dried on 400 mesh carbon-coatedCu grids was performed using a Hitachi H-8000 transmission electronmicroscope operated at an accelerating voltage of 200 kV. SEMimaging and EDS measurements were performed on NP samplessupported on silicon wafers using a Zeiss Ultraplus thermal fieldemission scanning electron microscope. HAADF-STEM imaging andEDS mapping of NFs drop-dried on Mo grids with ultrathin carbonsupport film were carried out using a JEOL 2100F 200 kV FEG-STEM/TEM microscope equipped with a CEOS CS corrector on theillumination system. The optical extinction spectra of colloidal NPsamples were measured using a Beckman coulter Du 640spectrophotometer. PXRD patterns were record on Bruker axs D8Discover (Cu Kα = 1.5406 Å). XPS measurements were carried out on

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23928

freshly prepared and vacuum-dried samples using a Krato AXIS UltraDLD XPS system equipped with a monochromatic Al Kα source. AFinnigan ELEMENT XR double focusing magnetic sector fieldinductively coupled plasma-mass spectrometer (SF-ICP-MS) wasused for the analysis of Cu (65, MR), Au (197, MR), and internalstandard Rh (103 MR). A 0.2 mL min−1 Micromist U-series nebulizer(GE, Australia), quartz torch, and injector (Thermo Fisher Scientific,USA) were used for sample introduction. Sample gas flow was at 1.08mL min−1. The forwarding power was 1250 W. The samples for ICP-MS measurements were prepared by adding 1 mL of nitric acid and 3mL of hydrochloric acid into Teflon digestion vessels containing theNP samples. The samples were digested using hot block at 180 °C for4 h. The digestates were brought to 50 mL with water. A 3-pointcalibration curve was used for Cu and Au. The calibration range wasfrom 50 to 600 ppb. The R2 values for the initial calibration curveswere greater than 0.995.

■ ASSOCIATED CONTENT*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acsami.6b07309.

Additional figures including TEM images, SEM images,EDS elemental analysis, PXRD patterns, ICP-MS results,optical extinction spectra, XPS results, Mie scatteringtheory calculations, and results of electrochemicalmeasurements as noted in the text (PDF)

■ AUTHOR INFORMATIONCorresponding Author*E-mail: wang344@mailbox.sc.edu. Phone: 1-803-777-2203.Fax: 1-803-777-9521.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported by the University of South Carolina(USC) Startup Funds, an ASPIRE-I Track-I Award from theUSC Office of Vice President for Research, and in part by aNational Science Foundation CAREER Award (DMR-1253231). E.V. was partially supported by a GAANNFellowship provided by the Department of Education throughGAANN Award P200A120075. Q.Z. was partially supported bya Dissertation Fellowship from USC NanoCenter and a SPARCGraduate Research Grant from the USC Office of the VicePresident for Research. As a visiting student from NanjingUniversity, T.Z. was partially supported by the ChinaScholarship Council. The authors would like to thank Dr.Douglas A. Blom for his help on high-resolution HAADF-STEM measurements, Dr. Andrew B. Greytak for the use of theatmosphere-controlled tube furnace, Dr. Lijian He for ICP-MSmeasurements, Dr. Ye Lin for XPS measurements, and the USCElectron Microscopy Center for instrument use.

■ REFERENCES(1) Wang, D. L.; Xin, H. L. L.; Hovden, R.; Wang, H. S.; Yu, Y. C.;Muller, D. A.; DiSalvo, F. J.; Abruna, H. D. Structurally OrderedIntermetallic Platinum-Cobalt Core-Shell Nanoparticles with En-hanced Activity and Stability as Oxygen Reduction Electrocatalysts.Nat. Mater. 2013, 12, 81−87.(2) Tian, N.; Zhou, Z. Y.; Sun, S. G.; Ding, Y.; Wang, Z. L. Synthesisof Tetrahexahedral Platinum Nanocrystals with High-Index Facets andHigh Electro-Oxidation Activity. Science 2007, 316, 732−735.(3) Tao, F.; Grass, M. E.; Zhang, Y. W.; Butcher, D. R.; Renzas, J. R.;Liu, Z.; Chung, J. Y.; Mun, B. S.; Salmeron, M.; Somorjai, G. A.

Reaction-Driven Restructuring of Rh-Pd and Pt-Pd Core-ShellNanoparticles. Science 2008, 322, 932−934.(4) Erlebacher, J.; Aziz, M. J.; Karma, A.; Dimitrov, N.; Sieradzki, K.Evolution of Nanoporosity in Dealloying. Nature 2001, 410, 450−453.(5) Ding, Y.; Kim, Y. J.; Erlebacher, J. Nanoporous Gold Leaf:″Ancient Technology″/Advanced Material. Adv. Mater. 2004, 16,1897−1900.(6) Wittstock, A.; Zielasek, V.; Biener, J.; Friend, C. M.; Baumer, M.Nanoporous Gold Catalysts for Selective Gas-Phase OxidativeCoupling of Methanol at Low Temperature. Science 2010, 327,319−322.(7) Wittstock, A.; Baumer, M. Catalysis by Unsupported SkeletalGold Catalysts. Acc. Chem. Res. 2014, 47, 731−739.(8) Fujita, T.; Guan, P. F.; McKenna, K.; Lang, X. Y.; Hirata, A.;Zhang, L.; Tokunaga, T.; Arai, S.; Yamamoto, Y.; Tanaka, N.; Ishikawa,Y.; Asao, N.; Yamamoto, Y.; Erlebacher, J.; Chen, M. W. AtomicOrigins of the High Catalytic Activity of Nanoporous Gold. Nat.Mater. 2012, 11, 775−780.(9) Zhang, J. T.; Liu, P. P.; Ma, H. Y.; Ding, Y. NanostructuredPorous Gold for Methanol Electro-Oxidation. J. Phys. Chem. C 2007,111, 10382−10388.(10) Gan, L.; Heggen, M.; O’Malley, R.; Theobald, B.; Strasser, P.Understanding and Controlling Nanoporosity Formation for Improv-ing the Stability of Bimetallic Fuel Cell Catalysts. Nano Lett. 2013, 13,1131−1138.(11) Li, X. Q.; Chen, Q.; McCue, I.; Snyder, J.; Crozier, P.;Erlebacher, J.; Sieradzki, K. Dealloying of Noble-Metal AlloyNanoparticles. Nano Lett. 2014, 14, 2569−2577.(12) Oezaslan, M.; Heggen, M.; Strasser, P. Size-DependentMorphology of Dealloyed Bimetallic Catalysts: Linking the Nano tothe Macro Scale. J. Am. Chem. Soc. 2012, 134, 514−524.(13) Wang, D. L.; Yu, Y. C.; Xin, H. L. L.; Hovden, R.; Ercius, P.;Mundy, J. A.; Chen, H.; Richard, J. H.; Muller, D. A.; DiSalvo, F. J.;Abruna, H. D. Tuning Oxygen Reduction Reaction Activity viaControllable Dealloying: A Model Study of Ordered Cu3Pt/CIntermetallic Nanocatalysts. Nano Lett. 2012, 12, 5230−5238.(14) Chen, C.; Kang, Y. J.; Huo, Z. Y.; Zhu, Z. W.; Huang, W. Y.;Xin, H. L. L.; Snyder, J. D.; Li, D. G.; Herron, J. A.; Mavrikakis, M.;Chi, M. F.; More, K. L.; Li, Y. D.; Markovic, N. M.; Somorjai, G. A.;Yang, P. D.; Stamenkovic, V. R. Highly Crystalline MultimetallicNanoframes with Three-Dimensional Electrocatalytic Surfaces. Science2014, 343, 1339−1343.(15) Ding, J. B.; Bu, L. Z.; Guo, S. J.; Zhao, Z. P.; Zhu, E. B.; Huang,Y.; Huang, X. Q. Morphology and Phase Controlled Construction ofPt-Ni Nanostructures for Efficient Electrocatalysis. Nano Lett. 2016,16, 2762−2767.(16) Ding, J. B.; Zhu, X.; Bu, L. Z.; Yao, J. L.; Guo, J.; Guo, S. J.;Huang, X. Q. Highly Open Rhombic Dodecahedral PtCu Nanoframes.Chem. Commun. 2015, 51, 9722−9725.(17) Gan, L.; Heggen, M.; Rudi, S.; Strasser, P. Core-ShellCompositional Fine Structures of Dealloyed PtxNi1-x Nanoparticlesand Their Impact on Oxygen Reduction Catalysis. Nano Lett. 2012,12, 5423−5430.(18) Snyder, J.; McCue, I.; Livi, K.; Erlebacher, J. Structure/Processing/Properties Relationships in Nanoporous Nanoparticles AsApplied to Catalysis of the Cathodic Oxygen Reduction Reaction. J.Am. Chem. Soc. 2012, 134, 8633−8645.(19) Bandarenka, A. S.; Koper, M. T. M. Structural and ElectronicEffects in Heterogeneous Electrocatalysis: Toward a Rational Designof Electrocatalysts. J. Catal. 2013, 308, 11−24.(20) Stamenkovic, V. R.; Fowler, B.; Mun, B. S.; Wang, G. F.; Ross,P. N.; Lucas, C. A.; Markovic, N. M. Improved Oxygen ReductionActivity on Pt3Ni(111) via Increased Surface Site Availability. Science2007, 315, 493−497.(21) Kim, D.; Resasco, J.; Yu, Y.; Asiri, A. M.; Yang, P. D. SynergisticGeometric and Electronic Effects for Electrochemical Reduction ofCarbon Dioxide using Gold-Copper Bimetallic Nanoparticles. Nat.Commun. 2014, 5, 4948.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23929

(22) Zhang, L.; Roling, L. T.; Wang, X.; Vara, M.; Chi, M. F.; Liu, J.Y.; Choi, S. I.; Park, J.; Herron, J. A.; Xie, Z. X.; Mavrikakis, M.; Xia, Y.N. Platinum-Based Nanocages with Subnanometer-Thick Walls andWell-Defined, Controllable Facets. Science 2015, 349, 412−416.(23) Huang, X. Q.; Zhao, Z. P.; Cao, L.; Chen, Y.; Zhu, E. B.; Lin, Z.Y.; Li, M. F.; Yan, A. M.; Zettl, A.; Wang, Y. M.; Duan, X. F.; Mueller,T.; Huang, Y. High-Performance Transition Metal-Doped Pt3NiOctahedra for Oxygen Reduction Reaction. Science 2015, 348, 1230−1234.(24) Mazumder, V.; Lee, Y.; Sun, S. H. Recent Development ofActive Nanoparticle Catalysts for Fuel Cell Reactions. Adv. Funct.Mater. 2010, 20, 1224−1231.(25) Connell, J. G.; Al Balushi, Z. Y.; Sohn, K.; Huang, J. X.; Lauhon,L. J. Growth of Ge Nanowires from Au-Cu Alloy NanoparticleCatalysts Synthesized from Aqueous Solution. J. Phys. Chem. Lett.2010, 1, 3360−3365.(26) Zhang, L.; Jing, H.; Boisvert, G.; He, J. Z.; Wang, H. GeometryControl and Optical Tunability of Metal-Cuprous Oxide Core-ShellNanoparticles. ACS Nano 2012, 6, 3514−3527.(27) Schaak, R. E.; Sra, A. K.; Leonard, B. M.; Cable, R. E.; Bauer, J.C.; Han, Y. F.; Means, J.; Teizer, W.; Vasquez, Y.; Funck, E. S.Metallurgy in a Beaker: Nanoparticle Toolkit for the Rapid Low-Temperature Solution Synthesis of Functional Multimetallic Solid-State Materials. J. Am. Chem. Soc. 2005, 127, 3506−3515.(28) Sra, A. K.; Ewers, T. D.; Schaak, R. E. Direct Solution Synthesisof Intermetallic AuCu and AuCu3 Nanocrystals and NanowireNetworks. Chem. Mater. 2005, 17, 758−766.(29) Vegard, L. Die Konstitution der Mischkristalle und dieRaumfullung der Atome. Eur. Phys. J. A 1921, 5, 17−26.(30) Lubarda, V. A. On the Effective Lattice Parameter of BinaryAlloys. Mech. Mater. 2003, 35, 53−68.(31) Zhang, L.; Blom, D. A.; Wang, H. Au-Cu2O Core-ShellNanoparticles: A Hybrid Metal-Semiconductor Heteronanostructurewith Geometrically Tunable Optical Properties. Chem. Mater. 2011,23, 4587−4598.(32) Liu, D. Y.; Ding, S. Y.; Lin, H. X.; Liu, B. J.; Ye, Z. Z.; Fan, F. R.;Ren, B.; Tian, Z. Q. Distinctive Enhanced and Tunable PlasmonResonant Absorption from Controllable Au@Cu2O Nanoparticles:Experimental and Theoretical Modeling. J. Phys. Chem. C 2012, 116,4477−4483.(33) Henkel, A.; Jakab, A.; Brunklaus, G.; Sonnichsen, C. TuningPlasmonic Properties by Alloying Copper into Gold Nanorods. J. Phys.Chem. C 2009, 113, 2200−2204.(34) Motl, N. E.; Ewusi-Annan, E.; Sines, I. T.; Jensen, L.; Schaak, R.E. Au-Cu Alloy Nanoparticles with Tunable Compositions andPlasmonic Properties: Experimental Determination of Compositionand Correlation with Theory. J. Phys. Chem. C 2010, 114, 19263−19269.(35) Thota, S.; Chen, S. T.; Zhou, Y. D.; Zhang, Y.; Zou, S. L.; Zhao,J. Structural Defect Induced Peak Splitting in Gold-Copper BimetallicNanorods during Growth by Single Particle Spectroscopy. Nanoscale2015, 7, 14652−14658.(36) Rugolo, J.; Erlebacher, J.; Sieradzki, K. Length Scales in AlloyDissolution and Measurement of Absolute Interfacial Free Energy.Nat. Mater. 2006, 5, 946−949.(37) Wagner, K.; Brankovic, S. R.; Dimitrov, N.; Sieradzki, K.Dealloying below the Critical Potential. J. Electrochem. Soc. 1997, 144,3545−3555.(38) Sieradzki, K.; Dimitrov, N.; Movrin, D.; McCall, C.; Vasiljevic,N.; Erlebacher, J. The Dealloying Critical Potential. J. Electrochem. Soc.2002, 149, B370−B377.(39) Cammarata, R. C. Surface and Interface Stress Effects in Thin-Films. Prog. Surf. Sci. 1994, 46, 1−38.(40) Huang, J. F.; Zhu, Y. H.; Liu, C. X.; Zhao, Y. F.; Liu, Z. H.;Hedhili, M. N.; Fratalocchi, A.; Han, Y. Fabricating a HomogeneouslyAlloyed AuAg Shell on Au Nanorods to Achieve Strong, Stable, andTunable Surface Plasmon Resonances. Small 2015, 11, 5214−5221.

(41) Gao, C. B.; Hu, Y. X.; Wang, M. S.; Chi, M. F.; Yin, Y. D. FullyAlloyed Ag/Au Nanospheres: Combining the Plasmonic Property ofAg with the Stability of Au. J. Am. Chem. Soc. 2014, 136, 7474−7479.(42) Liu, Y. L.; Walker, A. R. H. Preferential Outward Diffusion ofCu during Unconventional Galvanic Replacement Reactions betweenHAuCl4 and Surface-Limited Cu Nanocrystals. ACS Nano 2011, 5,6843−6854.(43) Yin, Y. D.; Rioux, R. M.; Erdonmez, C. K.; Hughes, S.; Somorjai,G. A.; Alivisatos, A. P. Formation of Hollow Nanocrystals through theNanoscale Kirkendall Effect. Science 2004, 304, 711−714.(44) Fan, H. J.; Gosele, U.; Zacharias, M. Formation of Nanotubesand Hollow Nanoparticles based on Kirkendall and DiffusionProcesses: A Review. Small 2007, 3, 1660−1671.(45) Gonzalez, E.; Arbiol, J.; Puntes, V. F. Carving at the Nanoscale:Sequential Galvanic Exchange and Kirkendall Growth at RoomTemperature. Science 2011, 334, 1377−1380.(46) Sun, Y. G.; Xia, Y. N. Shape-Controlled Synthesis of Gold andSilver Nanoparticles. Science 2002, 298, 2176−2179.(47) Luo, Q.; Mackay, R. A.; Babu, S. V. Copper Dissolution inAqueous Ammonia-Containing Media during Chemical MechanicalPolishing. Chem. Mater. 1997, 9, 2101−2106.(48) Parida, S.; Kramer, D.; Volkert, C. A.; Rosner, H.; Erlebacher, J.;Weissmuller, J. Volume Change during the Formation of NanoporousGold by Dealloying. Phys. Rev. Lett. 2006, 97, 035504.(49) Kim, M. J.; Na, H. J.; Lee, K. C.; Yoo, E. A.; Lee, M. Y.Preparation and Characterization of Au-Ag and Au-Cu AlloyNanoparticles in Chloroform. J. Mater. Chem. 2003, 13, 1789−1792.(50) Haruta, M.; Date, M. Advances in the Catalysis of AuNanoparticles. Appl. Catal., A 2001, 222, 427−437.(51) Zhang, Q. F.; Large, N.; Nordlander, P.; Wang, H. Porous AuNanoparticles with Tunable Plasmon Resonances and Intense FieldEnhancements for Single-Particle SERS. J. Phys. Chem. Lett. 2014, 5,370−374.(52) Zhang, Q. F.; Blom, D. A.; Wang, H. Nanoporosity-EnhancedCatalysis on Subwavelength Au Nanoparticles: a Plasmon-EnhancedSpectroscopic Study. Chem. Mater. 2014, 26, 5131−5142.(53) Hamelin, A. Cyclic Voltammetry at Gold Single-Crystal Surfaces0.1. Behaviour at Low-Index Faces. J. Electroanal. Chem. 1996, 407, 1−11.(54) Hamelin, A.; Martins, A. M. Cyclic Voltammetry at Gold Single-Crystal Surfaces 0.2. Behaviour of High-Index Faces. J. Electroanal.Chem. 1996, 407, 13−21.(55) Zhang, J. A.; Langille, M. R.; Personick, M. L.; Zhang, K.; Li, S.Y.; Mirkin, C. A. Concave Cubic Gold Nanocrystals with High-IndexFacets. J. Am. Chem. Soc. 2010, 132, 14012−14014.(56) Pedireddy, S.; Lee, H. K.; Tjiu, W. W.; Phang, I. Y.; Tan, H. R.;Chua, S. Q.; Troadec, C.; Ling, X. Y. One-Step Synthesis of Zero-Dimensional Hollow Nanoporous Gold Nanoparticles with EnhancedMethanol Electrooxidation Performance. Nat. Commun. 2014, 5, 4947.(57) Ma, Y. Y.; Kuang, Q.; Jiang, Z. Y.; Xie, Z. X.; Huang, R. B.;Zheng, L. S. Synthesis of Trisoctahedral Gold Nanocrystals withExposed High-Index Facets by a Facile Chemical Method. Angew.Chem., Int. Ed. 2008, 47, 8901−8904.(58) Ming, T.; Feng, W.; Tang, Q.; Wang, F.; Sun, L. D.; Wang, J. F.;Yan, C. H. Growth of Tetrahexahedral Gold Nanocrystals with High-index Facets. J. Am. Chem. Soc. 2009, 131, 16350−16351.(59) Trasatti, S.; Petrii, O. A. Real Surface-Area Measurements inElectrochemistry. J. Electroanal. Chem. 1992, 327, 353−376.(60) Lee, S. W.; Chen, S.; Suntivich, J.; Sasaki, K.; Adzic, R. R.; Shao-Horn, Y. Role of Surface Steps of Pt Nanoparticles on theElectrochemical Activity for Oxygen Reduction. J. Phys. Chem. Lett.2010, 1, 1316−1320.(61) Janssens, T. V. W.; Clausen, B. S.; Hvolbaek, B.; Falsig, H.;Christensen, C. H.; Bligaard, T.; Norskov, J. K. Insights into theReactivity of Supported Au Nanoparticles: Combining Theory andExperiments. Top. Catal. 2007, 44, 15−26.(62) Pedireddy, S.; Lee, H. K.; Koh, C. S.; Tan, J. M.; Tjiu, W. W.;Ling, X. Y. Nanoporous Gold Bowls: A Kinetic Approach to Control

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23930

Open Shell Structures and Size-Tunable Lattice Strain for Electro-catalytic Applications. Small 2016, 12, DOI: 10.1002/smll.201601371.(63) Pourjafari Amyab, S.; Saievar-Iranizad, E.; Bayat, A. PlatinumNanoparticles with Superacid-Doped Polyvinylpyrrolidone CoatedCarbon Nanotubes: Electrocatalyst for Oxygen Reduction Reaction inHigh-Temperature Proton Exchange Membrane Fuel Cell. RSC Adv.2016, 6, 41937−41946.(64) Mali, S. S.; Kim, H.; Jang, W. Y.; Park, H. S.; Patil, P. S.; Hong,C. K. Novel Synthesis and Characterization of Mesoporous ZnONanofibers by Electrospinning Technique. ACS Sustainable Chem. Eng.2013, 1, 1207−1213.(65) Jiang, C. L.; Nie, J.; Ma, G. P. A Polymer/Metal Core-ShellNanofiber Membrane by Electrospinning with an Electric Field, and itsApplication for Catalyst Support. RSC Adv. 2016, 6, 22996−23007.

ACS Applied Materials & Interfaces Research Article

DOI: 10.1021/acsami.6b07309ACS Appl. Mater. Interfaces 2016, 8, 23920−23931

23931