Article pubs.acs.org/cm
Cite This: Chem. Mater. XXXX, XXX, XXX−XXX
Mechanistic Study of Galvanic Replacement of Chemically Heterogeneous Templates Alexander N. Chen,† Sophia M. McClain,†,§ Stephen D. House,‡ Judith C. Yang,‡ and Sara E. Skrabalak*,† †
Department of Chemistry, Indiana University, 800 East Kirkwood Avenue, Bloomington, Indiana 47405, United States Department of Chemical and Petroleum Engineering and Department of Physics, University of Pittsburgh, Pittsburgh, Pennsylvania 15261, United States
‡
Chem. Mater. Downloaded from pubs.acs.org by UNIV OF CAMBRIDGE on 01/31/19. For personal use only.
S Supporting Information *
ABSTRACT: Galvanic replacement is a useful method for synthesizing architecturally complex multimetallic nanostructures from monometallic templates. The oxidation of monometallic templates by ions of a more noble metal is well-studied; however, chemically heterogeneous templates offer more than one type of reaction site and potentially structurally more complex materials. Yet, mechanistic studies with such templates are limited. Here, the reactivities of Ag and Pd in Janus-style AgPd dimers are compared when ions capable of oxidizing both metals are introduced. This study reveals (1) the selectivity of galvanic replacement with dimer templates, (2) cooperativity between galvanic replacement in one dimer domain and solid-state diffusion in another domain, and (3) similar replacement mechanisms for different redox pairs. Specifically, Au ions almost exclusively replace Ag before exchange with Pd is evident, and Ag oxidation facilitates diffusion of the remaining Ag into the Pd domain. These results provide mechanistic insight into the kinetically linked processes involved in the galvanic replacement of complex multimetallic templates and demonstrate the importance of understanding these interactions to achieve structural and compositional control over the resulting nanostructures.
■
INTRODUCTION The properties of metal nanocrystals depend on particle composition, shape, and architecture.1,2 Thus, to produce nanocrystals with specific properties, one must be able to predictably synthesize materials with the appropriate compositional and structural features. This requirement is more easily met for relatively isotropic, high symmetry materials3,4 than for anisotropic, low symmetry materials, as the latter suffer from less detailed understanding of their mechanisms of formation. The galvanic replacement reaction with nanoscale templates is a robust route to hollow multimetallic nanostructures, often using Ag nanocrystals as sacrificial templates. In this comparatively simple model system, the mechanism by which Ag nanostructures transform into multimetallic cages has been well-studied, and this mechanism may be used to reliably synthesize a variety of architecturally complex Agbased multimetallic nanostructures with effectively the same procedure.5−9 Synthesis of analogous architecturally complex nanostructures using multimetallic sacrificial templates with spatially separated domains promises to be less straightforward. Instead of a single reaction between the metal ion oxidant and the metallic nanocrystal reductant, each ionic species is now expected to react with every metal that possesses a lower reduction potential. This situation leads to many different © XXXX American Chemical Society
replacement reactions, likely with significantly different rates. Thus, while the Nernst equation predicts that the metals with the highest reduction potentials will eventually oxidize those with lower reduction potentials, the reaction pathway, the intermediates that lie along it, and the relative reaction rates of different pairs remain unclear. Previous work has shown that for a heterodimeric template composed of one Ag compartment and one core−shell Au−Pd compartment, Ag alone is oxidized upon addition of AuCl4−, PdCl42−, or PtCl62−, and the inertness of the Pd surface is attributed to binding of Cl− ions.10 This result is useful in that some selectivity is shown; however, overly selective surface passivation by Cl− and the possible effect of the Au core on the thin Pd shell make it difficult to extract mechanistic insights. Here, results of galvanic replacement with chemically anisotropic nanostructures and mixtures of chemically different nanostructures are presented. The templates studied include: Pd nanocubes, AgPd heterodimers with one Ag compartment and one Pd compartment, and a mixture of Ag and Pd nanocubes. These reductants are exposed to the oxidants HAuCl4, NaAuCl2, and KAuBr4 to study the effect of reaction Received: November 4, 2018 Revised: January 12, 2019
A
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
size of 1. High-resolution TEM (HRTEM) imaging of the particle interfaces was performed using a Hitachi H-9500 Environmental TEM, equipped with a LaB6 electron source and operated at 300 kV. Elemental compositions were determined using Oxford Aztec energy dispersive X-ray detectors interfaced to the SEM and to the JEOL JEM 3200FS TEM operating in STEM mode. Absorption spectra were obtained using a Cary 100 Bio UV−visible spectrometer. Powder X-ray diffraction patterns were obtained on a Siemens/Bruker D-5000 X-ray diffractometer using Cu Kα radiation (λ = 0.15418 nm).
stoichiometry and reduction potential difference. The standard reduction potentials of the AuCl4−/Au pair (1.002 V vs the standard hydrogen electrode, SHE),11 the AuCl2−/Au pair (1.22 V vs SHE),6 and the AuBr4−/Au pair (0.854 V vs SHE)11 are all higher than those of the Ag+/Ag pair (0.7996 V vs SHE)11 and of the PdCl42−/Pd pair (0.591 V vs SHE)11 and should be able to oxidize both Ag and Pd. As we find, replacement of Ag occurs more rapidly than for Pd, and the reduction−oxidation reactions taking place at the Ag compartment promote concurrent solid-state diffusion of Ag into the Pd compartment.
■
■
RESULTS AND DISCUSSION Titration of AgPd Heterodimers. Janus AgPd dimers were synthesized by depositing Ag on cubic Pd seeds (Figure S1) in a sulfide-mediated synthesis to produce particles that consistently display a cubic Pd domain, corresponding to the original seed, and a rounder Ag domain, as confirmed by TEM and scanning transmission electron microscopy coupled to energy dispersive X-ray spectroscopy (STEM-EDS) in Figures 1A−F. Notably, elemental mapping of particles by STEM-EDS
EXPERIMENTAL SECTION
Chemicals. Chloroauric acid (HAuCl4·3H2O, 99.9%), gold(I) chloride (AuCl, 99.9%), potassium tetrabromoaurate (KAuBr4·xH2O, 99.9%)), palladium(II) chloride (PdCl2, 99.98%), silver nitrate (AgNO3, 99.9999%), sodium chloride (NaCl, 99%), sodium sulfide nonahydrate (Na2S·9H2O, 98%), cetyltrimethylammonium bromide (CTAB, 99%, lot nos. BCBS1424 V, BCBT1510), cetyltrimethylammonium chloride (CTAC, 0.78 M, lot nos. STBG5554 V, STBH2336), polyvinylpyrrolidone (PVP55k, MW ≈ 55 000 g/mol, and PVP29k, MW ≈ 29 000 g/mol), L-ascorbic acid (L-AA, C6H8O6, 99%), and ethylene glycol (99.8%, lot no. SHBH2712) were purchased from Sigma-Aldrich. Hydrochloric acid (HCl, 1 M) was purchased from Macron. Nanopure water (18.2 MΩ·cm) was used in all experiments. An aqueous 10 mM H2PdCl4 solution was prepared by stirring dissolved PdCl2 (44.6 mg) in 25 mL of HCl (pH 1.70) for 1 h while heating at 70 °C. Synthesis of Ag Cubes. Ag cubes were synthesized according to a literature procedure.5 Briefly, 6 mL ethylene glycol were heated in a 150 °C oil bath for 1 h, before adding 70 μL Na2S (3 mM), and then 1.5 mL PVP29k (20 mg/mL) and 0.5 mL AgNO3 (48 mg/mL). Solutions were left in the oil bath for 10−15 min before being collected by centrifugation. Synthesis of Pd Cubes. Pd cubes were synthesized according to a modified literature procedure.12 Five-hundred microliters of H2PdCl4 (10 mM) was added to 10 mL of CTAB (12.5 mM). After heating to 95 °C in an oil bath, 80 μL of L-AA (100 mM) was added, and the solution was stirred for 30 min. Two-hundred microliters of the stirred solution was added to another solution made of 5 mL of CTAB (50 mM), 125 μL of H2PdCl4 (10 mM), and 25 μL of L-AA (100 mM). The second solution was left in a 40 °C oil bath for 14 h before being centrifuged for 15 min at 7291.8g and then redispersed in 1 mL of water. Synthesis of AgPd Heterodimers. AgPd heterodimers were synthesized according to a modified literature procedure.13 Four hundred and seventy-two microliters of CTAC (0.78 M), 6.908 mL of H2O, 50 μL of Na2S (3 mM), 1 mL of the previously synthesized Pd cubes, 572 μL of AgNO3 (5 mM), and 1 mL of L-AA (300 mM) were mixed together and left in a 95 °C oil bath for 3 h under darkness before being centrifuged for 8.5 min at 3832.7g and then redispersed in 1 mL of water. Galvanic Replacement. Galvanic replacement experiments were based on a literature procedure.5 Typically, solutions of 0.1 mM HAuCl4, AuCl, or KAuBr4 were injected using a syringe pump at a rate of 0.25 mL/min into refluxing solutions composed of 2.5 mL of PVP55k (10 mg/mL) and 1 mL of seeds. Ten minutes after the oxidant solution was injected, products were collected by centrifuging for 15 min at 7291.8g and then redispersed in 1 mL of water. When mixing together Ag cubes and Pd cubes, the seed solutions were composed of 830 μL of Pd cubes to 170 μL of Ag cubes. AuCl was dissolved in saturated aqueous solutions of NaCl according to previously reported work.6 Characterization. Scanning electron microscopy (SEM) images were taken using a FEI Quanta 600F ESEM operated at 30 kV with a spot size of 3. Transmission electron microscopy (TEM) images were taken using a JEOL JEM 1010 TEM operate at 80 kV with a spot size of 1 and a JEOL JEM 3200FS TEM operated at 300 kV with a spot
Figure 1. (A, G) TEM images and (B−F, H−L) STEM-EDS elemental maps of (A--F) AgPd heterodimers and (G--L) products after adding 8 mL of HAuCl4 to AgPd heterodimers. (F) Overlays of panels C−D. Red, Pd (Lα); green, Ag (Lα); yellow, Au (Mα); cyan, S (Kα).
sometimes reveals a significant S signal, although the lowerresolution and larger-scale EDS scans done in the scanning electron microscope (SEM-EDS) do not automatically detect S, indicating that S should occupy very little of the sample; thus, S-containing compounds should account for very little of the particles’ reactivity. Also, powder X-ray diffraction (XRD) of AgPd heterodimers shows only peaks corresponding to Ag and Pd (Figure S2). While the small presence of S in the heterodimers prevents them from being a perfect comparison B
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
Figure 2. (A) UV−vis spectra, (B) plot of localized surface plasmon resonance peak positions from (A) as a function of HAuCl4 added, and (C− G) selected TEM images of galvanic replacement products from adding (C) 0, (D) 1, (E) 8, (F) 18, and (G) 45 mL of HAuCl4 to AgPd heterodimers.
rods by HAuCl4 has previously been shown to result in spheroidal Au phases growing at one tip of a rod, without hollowing;14 however, the exposed Pd surfaces in Figure 1 appear to be completely covered by a Au-rich phase. The heterodimeric template must follow a different mechanism than that undergone by monometallic templates. Figure 2 shows key stages in the replacement of AgPd heterodimers over time, as increasing amounts of HAuCl4 are titrated into the solution containing the AgPd heterodimers. Changes in the plasmonic properties of the increasingly hollowed and Au-rich nanostructures are observed by UV− visible spectra of replacement products in Figure 2A. When HAuCl4 was injected into the AgPd heterodimer solution, the absorption peak red-shifted gradually and is consistent with what has previously been shown for continually hollowing Ag particles with HAuCl4 addition.15 When more than 8 mL of HAuCl4 were added, however, the absorption peak blueshifted, signifying a decrease in the sizes of plasmonic metal phases that was attributed to cage fragmentation for monometallic Ag templates.15 Shifts in the absorbance peak associated with the main plasmonic mode are tracked in Figure 2B. Prominent narrow peaks appearing in the UV region after adding 45 mL indicate the appearance of small, isolated plasmonic phases, presumably Au or AuAg.16 Corresponding TEM images support the spectral data. As shown in Figures 2C−D, replacement of AgPd heterodimers starts with the appearance of pits in the surface of the Ag domain, but mostly
to individual Ag and Pd nanocrystals, the low amount suggested by SEM-EDS combined with identification of crystalline Ag and Pd phases by XRD and segregation of the Ag, Pd, and S signals in STEM-EDS suggests that most of the Ag and Pd phases should display the same chemistry as their monometallic counterparts. Galvanic replacement of AgPd heterodimers yields an architecturally complex product that suggests the action of multiple competing processes. The TEM image in Figure 1G shows an original Pd cube encapsulated by a shell of higher atomic number, connected to a partially hollowed spheroidal Ag phase. STEM-EDS elemental mapping confirms that the cube is the only source of Pd in the product (Figures 1H−I). As expected, the Au signal forms a shell around both the Pd and the Ag domains (Figure 1K). The Ag signal is mostly concentrated in the spheroid; however, some overlap with the Pd signal from the nanocubes is surprisingly evident (Figure 1J). This Ag−Pd signal overlap contrasts with the complete phase segregation seen in the original heterodimers (Figures 1B−F). S is generally not detected, possibly due to the overlap between Au and S EDS signals or its removal during the replacement process. Overall, Figure 1 shows unexpected behavior for both Ag and Pd compared to previously published work for monometallic templates. Where the replacement of monometallic Ag typically leads to simple hollowing,5−7 the heterodimeric system sees both hollowing and migration of Ag inside a previously pure Pd phase. Galvanic replacement of Pd C
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
Figure 3. (A−D, F−I) STEM-EDS elemental maps and (E, J) corresponding line scans following the arrows for products obtained from (A−E) AgPd heterodimers stirred in boiling water for 3 h and (F−J) 4 mL of HAuCl4 added to AgPd heterodimers. Red, Pd (Lα); green, Ag (Lα); yellow, Au (Mα); purple, Sm (Lα).
is unsurprising, as an increase in size is expected from hollowing due to the Kirkendall effect.9 In addition, Table S1 displays the bulk elemental composition of the heterodimers and galvanic replacement products as determined by SEMEDS. The Ag:Pd ratio continually decreases from 0 to 18 mL of added HAuCl4, before increasing again between 18 and 45 mL. The evolution of bulk composition implies that mostly Ag is oxidized in the beginning of the titration and confirms that Pd is eventually oxidized as well in the later stages of titration. Oxidation and removal of Pd from the nanocrystal demands access to the surface; thus, the alloyed surface of the cubic section of the nanocrystal in Figure 2D should be trimetallic. The separate AgAu and Pd peaks in Figure S2 suggest that the Pd presence in the shell is small enough not to disturb an essentially AgAu lattice.
unchanged morphology, consistent with the little change observed in the UV−visible spectra. There is no apparent reaction of the Pd phase as the Ag phase begins to be replaced. However, as more HAuCl4 is added, visible deposition occurs on the Pd phase before the Ag domain is completely hollowed out (Figure 2E). Eventually, replacement progresses to the point that Ag domains are completely hollowed out (Figure 2F). At this point, the original Ag phase is still visible in TEM, but now as a thick, hollow shell, presumably composed of mostly Au. The original Pd phase is still cubic, but dotted with spheroids, which account for the small, isolated plasmonic phases indicated by the UV−vis spectrum. Notably, the Ag phase still does not fragment, unlike in monometallic Ag templates, suggesting that the resulting alloy is difficult to oxidize. Comparing Figures 2C and 2G, the final product is also significantly larger than the original seed. This observation D
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
Pd does eventually get replaced. Figure 3 as a whole argues that galvanic replacement provokes increased vacancy diffusion, allowing for migration between the two compartments of the heterodimer. The Ag−Pd interface is important in the observed diffusion of Ag; therefore, further structural characterization was undertaken. Figure S3A is a representative HRTEM image of the interface−of the AgPd dimer shown in Figure S3B, in which the lattices are clearly resolved. The image has been average background subtracted filtered (ABSF)27 and Fourier filtered to enhance visibility of the lattice. Fast Fourier transforms (FFTs) from each side of the interface are included as insets (Figures S3C and D). The image reveals that the interface is epitaxial. The visible lattice spacings perpendicular to the interface, as measured from the FFTs for the Pd and Ag, were 1.96 and 2.04 Å, respectively, corresponding to the {200} planes for each of the metals. Misfit dislocations, arising from this ∼3.9% lattice mismatch, are visible on both sides of the interface, with the unambiguous misfit planes indicated by “T”s. Likewise, the interfacial planes themselves in both the Ag (measurable from the FFT) and Pd (inferable from the visible planes and the cubic particle shape) are also of the {200} type. HRTEM imaging was more ambiguous for galvanic replacement products. Figure S4 shows a phase with lattice spacing corresponding to Ag, Au, or an alloy stretching well beyond the face of the Pd cube, whose lattice remains intact and of the same orientation, but the former Ag−Pd interface could not be resolved. Moiré fringe analysis performed on these particles is consistent with the shell being alloyed rather than a pure metal (Figure S5). Overall, the HRTEM analysis agrees with the structural characteristics implied by Figure 3 and the XRD data and adds that Ag grows epitaxially on Pd. While the above discussion assumes that Ag and Pd are the only significant reducing agents, the surfactant used to maintain colloidal stability during the galvanic replacement reactions is polyvinylpyrrolidone (PVP), whose terminal hydroxyl groups make it a weak reducing agent.28 Previous studies show that combination of galvanic replacement with a coreducing agent yields different products compared to galvanic replacement alone,29,30 to the point of potentially suppressing replacement.31 Control experiments wherein PVP was replaced with 9 mM cetyltrimethylammonium chloride (CTAC) make it clear that PVP does not compete with Ag or Pd as a reducing agent: when conducting analogous galvanic replacement reactions on AgPd heterodimers (Figure S6), similar products are obtained after replacing PVP with CTAC. Minor differences in morphology can be attributed to the differing capping effects of the surfactants. In addition, because neither Ag nor Pd can reduce Ag+, and PVP is a weaker reductant, it is unlikely that oxidized Ag+ species are being reduced in solution. Titration of Mixture of Monometallic Templates. As previously stated, diffusion of Ag into Pd has no analogue in the galvanic replacement of monometallic templates. Figure 3 shows that the diffusion process requires galvanic replacement, but the physical contact of Ag and Pd domains in the Janus dimers should also be important in diffusion. Monometallic Ag seeds mixed with monometallic Pd seeds provided a chemically heterogeneous sacrificial template in which Ag and Pd phases were unlikely to be in close proximity with each other. Diffusion of Ag across the comparatively great distances between particles is difficult; therefore, mixing different
Migration of Ag into the Pd compartment merits further analysis as a phenomenon without analogues when monometallic templates are used. The Kirkendall effect, whereby one metal diffuses by vacancy diffusion into a neighboring metal phase more quickly than the reverse process,17−19 has previously been cited to explain how nanoparticles become hollowed out during galvanic replacement: the difference in diffusion rates results in a net flux of vacancies, increasing vacancy concentration until voids nucleate.20 Moreover, the Kirkendall effect has been used to explain anisotropic hollowing and interdiffusion during galvanic replacement of AuCu3 alloy templates.21−23 Hollowing has also more generally been attributed to a competition between minimizing intermetallic bond energy and minimizing surface energy.24 While the ultimate cause may be unclear, solid-state diffusion is not unheard of in galvanic replacement, although it behaves isotropically in these examples of hollowing initially monometallic or single-phase templates. In addition, Ag templates are known to form hollow, alloyed products when replaced by H2PdCl4,25 Na2PdCl4,7 or PdCl2.26 As the hollow products formed after replacement of Ag by HAuCl4 are alloyed, this suggests that, given the Ag−Pd interface in the heterodimer, Ag might readily migrate into Pd. Figures 3A−E, wherein a AgPd heterodimer solution was refluxed for 3 h, shows that alloying at the titration temperature without galvanic replacement is unlikely. STEM-EDS elemental mapping along with the accompanying line scan (Figure 3E) shows that the Ag and Pd domains remain segregated. The Sm signal used to represent the background argues that small Ag signals in the Pd domain are not significant. There is no obvious movement of the Ag−Pd interface like that seen in the original demonstration of the Kirkendall effect.18 Thus, the migration of Ag into Pd must be intimately linked with galvanic replacement. The nature of the hollowing process provides the necessary link: for voids to form, vacancies need to exist and coalesce, and when oxidation of Ag increases the concentration of vacancies, the rate of vacancy diffusion should increase.17 So, galvanic replacement can induce both hollowing and an increased rate of vacancy diffusion, the latter of which may be less evident in monometallic templates, but causes noticeable migration in the Pd domain of the AgPd heterodimer. Figures 3F−J, showing the elemental mapping of another galvanic replacement product, support this reasoning. The line scan in Figure 3J passes over three important phases. First, a AgAu signal just outside the bulk of the Pd signal represents a AgAu shell, possibly containing some Pd. Second, a AgPd signal corresponds to the now alloyed cube. The weaker Au signal can be attributed to the AgAu shell surrounding the AgPd phase and not Au diffusing inside the Pd phase. Third, a dip followed by an increase in the AgAu signal just outside of the Pd signal represents a hollow shell around the former Ag phase. Figure 3J confirms that Ag is indeed migrating inside the Pd phase and not merely forming a shell on the surface. Powder XRD of the product from Figures 1G−L also support the existence of two phases: overlapping Ag and Au peaks can correspond to a AgAu phase, and a peak with a slightly lower 2θ values relative to the calculated Pd reference suggests that a predominantly Pd phase is alloyed with either Ag or Au (Figure S2). When considered next to the bulk elemental compositions in Table S1, this compositional distribution also suggests that much of the Au deposition surrounding the Pd cube may be due to replacement of Ag in the AgPd alloy; however, the 45 mL point in Table S1 shows that at least some E
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
cover the surrounding face and finally form thin Au layers on other faces. Hollow cages do not form, underlining an evident difference between the galvanic replacement mechanisms of Ag and Pd. As with heterodimers, PVP does not compete with Pd as a reducing agent (Figure S9). Segregation of Au may be rationalized in terms of surface energy, where the formation of an isolated Au phase is more favorable than that of a large Au− Pd interface. Previous reports have shown that core−shell architectures can be energetically disfavorable when compared to Janus particles, allowing for the conversion of some core− shell particles to heterodimers upon annealing.32,33 As Au monomers are produced slowly, monomers produced later in the synthesis may then be expected to deposit on or diffuse to existing Au phases rather than form new Au−Pd interfaces. Replacement of Pd at 25 °C, where surface diffusion and accompanying restructuring is expected to be slower, supports this idea, as deposited Au phases are less localized (Figure S10). Also, hollowing has previously been attributed to the Kirkendall effect.20 Under the experimental conditions reported in this work, Pd and Au do not appear to diffuse into each other quickly enough for either phase to become alloyed or hollowed out according to the Kirkendall effect. Comparison with Different Oxidants. To show the generality of the mechanism elaborated above for HAuCl4, similar galvanic replacement reactions were conducted where HAuCl4 was replaced by other oxidants. AuCl, or NaAuCl2 when dissolved in a NaCl solution,6 provided a comparison between Au(I) and Au(III), showing the effect of the redox reaction’s stoichiometry on product morphology. The watersoluble NaAuCl2 complex is necessary, as AuCl has poor solubility in H2O.6 Previous work shows that Ag cubes are galvanically replaced by Au(I) to form hollow boxes rather than the cages obtained from Au(III) due to more Au being deposited per oxidized Ag atom, allowing the pinhole to close earlier in the reaction.6 The pinhole is a small, localized hole that has been reported to develop at the beginning of the replacement of Ag cubes, allowing for quick evacuation of replaced Ag from the interior. Replacement of Pd cubes by Au(I) contrasted with Au(III). Au phases developed equally on each face of the Pd cubes, although the template still was not hollowed (Figure S11). When using a mixture of Ag cubes and Pd cubes, Ag cubes were hollowed out first, before Pd cubes developed Au-rich shells (Figure S12). When using AgPd heterodimers, similar motifs are observed for both NaAuCl2 and HAuCl4: Ag is hollowed, spheroids start growing on the Pd domain to eventually cover the cube (Figures 5A−C), and a AgPd alloy forms (Figures 5D−I). The influence of the saturated NaCl solution used to dissolve AuCl was minimal, as injecting saturated NaCl alone into the seed solution did not produce obvious changes in morphology (Figure S13). Stoichiometry does account for some differences, such as the relatively thick walls and lesser amounts of Au deposition seen in previous work;6 however, the stoichiometry changes do not bring forward markedly different considerations in the galvanic replacement of heterogeneous templates. KAuBr4 provided a comparison between two Au(III) oxidants with different reduction potentials. Much like for HAuCl4, KAuBr4 oxidizes Ag cubes more rapidly than Pd cubes when both are present, causing hollowing of Ag nanostructures and anisotropic replacement of Pd nanostructures (Figure S14). Similarly to AuCl, the use of KAuBr4 results in more isotropic Au deposition on Pd cubes than HAuCl4 (Figure S15). Interestingly, the reduction potential of
monometallic templates together should yield information about selectivity without interference from solid-state diffusion. Oxidant solutions were slowly injected into aqueous suspensions of Ag cubes (Figure S7) mixed with Pd cubes (Figure S1). Figures 4A−D show TEM images of the products
Figure 4. (A−D) TEM images of galvanic replacement products from adding (A) 1.5, (B) 2.5, (C) 6, and (D) 8 mL of HAuCl4 to mixtures of Ag cubes and Pd cubes, and (E−H) STEM-EDS elemental map of panel B. Red, Pd (Lα); green, Ag (Lα); yellow, Au (Mα).
obtained when adding increasing amounts of HAuCl4, where the Ag particles are galvanically replaced far more quickly than their Pd counterparts. At low volumes, in Figures 4A and B, Pd cubes appear to be completely unaffected, retaining their original shapes while Ag cubes are hollowed out to form AgAu cages. Elemental mapping by STEM-EDS in Figures 4E−H confirms that the hollow structures are indeed composed of alloyed AgAu, while the Pd cubes remain unchanged, with no accompanying Ag or Au signals beyond the background. Further addition of HAuCl4 in Figures 4C and D results in fragmentation of increasingly Au-rich cages, as reported previously,5,15 and also in the highly anisotropic growth of spheroidal phases on some regions of the otherwise still cubic Pd templates. Overall, titrating the mixture of monometallic Ag cubes and Pd cubes with HAuCl4 shows that Ag is oxidized much more rapidly than Pd and reacts almost to completion before Pd oxidation is evident. These results mostly agree with those presented for AgPd heterodimers in Figures 1−3 while eliminating the possibility of Ag interacting with Pd due to the large distances separating monometallic particles in solution. Product morphologies are different in that Au does not deposit nearly as uniformly on Pd cubes as it does on the Pd domains of heterodimers, further supporting the argument that Ag in the AgPd alloys serves as a reductant. The apparently anisotropic galvanic replacement of Pd demands further study as the literature suggests that Ag is replaced isotropically, forming hollow structures with the same shape as the original template.5−9 The replacement of Pd rods by HAuCl4 has previously been noted to form tadpole-like structures, with Au-rich spheroids perched at one end of each Pd rod. Ostwald ripening between Au nuclei explained the deposition on one tip over another,14 and the presence of higher-energy sites at rod tips may explain why nucleation preferentially occurred at tips rather than sides of the rods. The galvanic replacement of Pd cubes by HAuCl4, shown in Figure S8, shows more clearly the anisotropic growth seen in Figure 4. As HAuCl4 is added, spheroidal Au phases develop at apparently random sites on the cubes before they expand to F
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
Figure 6. Scheme detailing proposed pathway by which AgPd heterodimers evolve during galvanic replacement by Au precursors.
still faster, but oxidation of Pd is complicated by vacancy diffusion of Ag into Pd, likely resulting in different reduction potentials for Ag and Pd inside the new AgPd alloy. Using oxidants with different reduction potentials and oxidation states yields similar results, suggesting that vacancy diffusion and selectivity of the reduction−oxidation reaction should be expected staples of galvanic replacement in heterogeneous templates. This comparison between a bimetallic heterodimer template and a mixture of its monometallic analogues provides a simple route to probing the interactions between distinct phases in atomically heterogeneous materials during a galvanic replacement reaction with the potential for application in the syntheses of other complex multimetallic systems.
Figure 5. TEM images of galvanic replacement products obtained from adding (A) 1, (B) 4, and (C) 8 mL of AuCl to AgPd heterodimers, and (D−I) STEM-EDS elemental mapping of (B), where (I) overlays panels E−G. Red, Pd (Lα); green, Ag (Lα); yellow, Au (Mα); cyan, S (Kα).
the AuBr4−/Au pair (0.854 V vs SHE)11 is smaller than that of the Pd2+/Pd pair (0.951 V vs SHE)11 but larger than that of the PdCl42−/Pd pair (0.591 V vs SHE),11 implying that Pd must be oxidized to a ligand-containing complex rather than Pd2+. Galvanic replacement of AgPd heterodimers also shows hollowed Ag domains and some deposition on Pd domains; however, unusually sharp features do appear on the Ag domain, and voids are not as evident in TEM (Figure S16). Table S2 shows that the proportions of Au, Ag, and Pd over time, as determined by SEM-EDS, remain nearly the same between KAuBr4 and HAuCl4. As elemental composition and morphological trends were essentially preserved, the differences observed in TEM images were attributed to the different binding abilities of bromide and chloride ions34 rather than the lower reduction potential. The similar trends observed in the galvanic replacement of heterogeneous templates by different oxidants are summarized in Figure 6. Replacement of a AgPd template by Au precursors is selective for Ag at first. As with monometallic Ag, a pinhole forms in the surface and Au replaces Ag on the surface, forming an alloy. More Ag is replaced around the pinhole, allowing vacancies to form inside the Ag phase and eventually coalesce into voids. Then, unlike with monometallic Ag, coalescence of vacancies competes with vacancy diffusion of Ag into the still intact Pd phase, resulting in a AgPd alloy. Ag is then oxidized from both the Ag phase and the AgPd phase, allowing Au alloys to form in the entirety of the Ag phase and on the surface of the Pd phase. Some Pd is eventually oxidized as well, but the AgPd domain is not hollowed.
■
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.chemmater.8b04630. Additional characterization of reaction products by SEM, SEM-EDS, TEM, STEM-EDS, and XRD (PDF)
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Stephen D. House: 0000-0003-2035-6373 Sara E. Skrabalak: 0000-0002-1873-100X Present Address §
S.M.M.: Department of Chemistry, University of Illinois at Urbana−Champaign, 600 South Mathews Avenue, Urbana, IL 61801, United States. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We thank Meredith R. Kunz for preliminary work preceding this study and Hannah M. Ashberry for collection of XRD patterns. A.N.C. and S.E.S. acknowledge the National Science Foundation (NSF CHE 1602476) for financial support. S.D.H. and J.C.Y. acknowledge the financial support of NSF-DMREF under Contract CHE-1534630 and the Environmental TEM Catalysis Consortium (ECC), which is supported by the University of Pittsburgh and Hitachi High Technologies. The
■
CONCLUSIONS The replacement results from the three different oxidants used all indicate that replacement of monometallic Ag is faster than monometallic Pd. Furthermore, when Ag and Pd phases are connected in a heterodimer, replacement of the Ag domain is G
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX
Article
Chemistry of Materials
nanorods during galvanic replacement reaction. Chem. Commun. 2016, 52, 5593−5596. (22) Thota, S.; Zhou, Y.; Chen, S.; Zou, S.; Zhao, J. Formation of bimetallic dumbbell shaped particles with a hollow junction during galvanic replacement reaction. Nanoscale 2017, 9, 6128−6135. (23) Li, G. G.; Sun, M.; Villarreal, E.; Pandey, S.; Phillpot, S. R.; Wang, H. Galvanic Replacement-Driven Transformations of Atomically Intermixed Bimetallic Colloidal Nanocrystals: Effects of Compositional Stoichiometry and Structural Ordering. Langmuir 2018, 34, 4340−4350. (24) da Silva, A. G. M.; Rodrigues, T. S.; Haigh, S. J.; Camargo, P. H. C. Galvanic replacement reaction: recent developments for engineering metal nanostructures towards catalytic applications. Chem. Commun. 2017, 53, 7135−7148. (25) Jing, H.; Wang, H. Structural Evolution of Ag−Pd Bimetallic Nanoparticles through Controlled Galvanic Replacement: Effects of Mild Reducing Agents. Chem. Mater. 2015, 27, 2172−2180. (26) Sutter, E.; Jungjohann, K.; Bliznakov, S.; Courty, A.; Maisonhaute, E.; Tenney, S.; Sutter, P. In situ liquid-cell electron microscopy of silver-palladium galvanic replacement reactions on silver nanoparticles. Nat. Commun. 2014, 5, 4946. (27) Kilaas, R. Optimal and near-optimal filters in high-resolution electron microscopy. J. Microsc. 1998, 190, 45−51. (28) Washio, I.; Xiong, Y.; Yin, Y.; Xia, Y. Reduction by the End Groups of Poly(vinyl pyrrolidone): A New and Versatile Route to the Kinetically Controlled Synthesis of Ag Triangular Nanoplates. Adv. Mater. 2006, 18, 1745−1749. (29) Weiner, R. G.; Smith, A. F.; Skrabalak, S. E. Synthesis of hollow and trimetallic nanostructures by seed-mediated co-reduction. Chem. Commun. 2015, 51, 8872−8875. (30) Ahn, J.; Wang, D.; Ding, Y.; Zhang, J.; Qin, D. Site-Selective Carving and Co-Deposition: Transformation of Ag Nanocubes into Concave Nanocrystals Encased by Au-Ag Alloy Frames. ACS Nano 2018, 12, 298−307. (31) Yang, Y.; Liu, J.; Fu, Z. W.; Qin, D. Galvanic replacement-free deposition of Au on Ag for core-shell nanocubes with enhanced chemical stability and SERS activity. J. Am. Chem. Soc. 2014, 136, 8153−8156. (32) Gu, H.; Zheng, R.; Zhang, X.; Xu, B. Facile one-pot synthesis of bifunctional heterodimers of nanoparticles: a conjugate of quantum dot and magnetic nanoparticles. J. Am. Chem. Soc. 2004, 126, 5664− 5665. (33) Grodzińska, D.; Pietra, F.; van Huis, M. A.; Vanmaekelbergh, D.; de Mello Donegá, C. Thermally induced atomic reconstruction of PbSe/CdSe core/shell quantum dots into PbSe/CdSe bi-hemisphere hetero-nanocrystals. J. Mater. Chem. 2011, 21, 11556−11565. (34) Carrasquillo, A., Jr.; Jeng, J.-J.; Barriga, R. J.; Temesghen, W. F.; Soriaga, M. P. Electrode-surface coordination chemistry: ligand substitution and competitive coordination of halides at well-defined Pd(100) and Pd(111) single crystals. Inorg. Chim. Acta 1997, 255, 249−254.
HRTEM interfacial imaging was performed at the Petersen Institute of Nanoscience and Engineering (PINSE) Nanoscale Fabrication and Characterization Facility (NFCF) at the University of Pittsburgh.
■
REFERENCES
(1) Motl, N. E.; Smith, A. F.; DeSantis, C. J.; Skrabalak, S. E. Engineering plasmonic metal colloids through composition and structural design. Chem. Soc. Rev. 2014, 43, 3823−3834. (2) Gamler, J. T. L.; Ashberry, H. M.; Skrabalak, S. E.; Koczkur, K. M. Random Alloyed versus Intermetallic Nanoparticles: A Comparison of Electrocatalytic Performance. Adv. Mater. 2018, 30, 1801563. (3) Lim, B.; Jiang, M.; Tao, J.; Camargo, P. H. C.; Zhu, Y.; Xia, Y. Shape-Controlled Synthesis of Pd Nanocrystals in Aqueous Solutions. Adv. Funct. Mater. 2009, 19, 189−200. (4) Personick, M. L.; Mirkin, C. A. Making sense of the mayhem behind shape control in the synthesis of gold nanoparticles. J. Am. Chem. Soc. 2013, 135, 18238−18247. (5) Skrabalak, S. E.; Au, L.; Li, X.; Xia, Y. Facile synthesis of Ag nanocubes and Au nanocages. Nat. Protoc. 2007, 2, 2182−2190. (6) Au, L.; Lu, X.; Xia, Y. A Comparative Study of Galvanic Replacement Reactions Involving Ag Nanocubes and AuCl(2) or AuCl(4). Adv. Mater. 2008, 20, 2517−2522. (7) Chen, J.; Wiley, B.; McLellan, J.; Xiong, Y.; Li, Z.-Y.; Xia, Y. Optical Properties of Pd-Ag and Pt-Ag Nanoboxes Synthesized via Galvanic Replacement Reactions. Nano Lett. 2005, 5, 2058−2062. (8) Cobley, C. M.; Campbell, D. J.; Xia, Y. Tailoring the Optical and Catalytic Properties of Gold-Silver Nanoboxes and Nanocages by Introducing Palladium. Adv. Mater. 2008, 20, 748−752. (9) Moreau, L. M.; Schurman, C. A.; Kewalramani, S.; Shahjamali, M. M.; Mirkin, C. A.; Bedzyk, M. J. How Ag Nanospheres Are Transformed into AgAu Nanocages. J. Am. Chem. Soc. 2017, 139, 12291−12298. (10) Lutz, P. S.; Bae, I. T.; Maye, M. M. Heterostructured Au/Pd-M (M = Au, Pd, Pt) nanoparticles with compartmentalized composition, morphology, and electrocatalytic activity. Nanoscale 2015, 7, 15748− 15756. (11) Vanýsek, P. Electrochemical Series. In CRC Handbook of Chemistry and Physics, 99th ed.; Rumble, J. R., Ed.; CRC Press/Taylor & Francis: Boca Raton, FL, 2018. (12) Zhang, J.; Zhang, L.; Xie, S.; Kuang, Q.; Han, X.; Xie, Z.; Zheng, L. Synthesis of Concave Palladium Nanocubes with HighIndex Surfaces and High Electrocatalytic Activities. Chem. - Eur. J. 2011, 17, 9915−9919. (13) Lee, S. U.; Hong, J. W.; Choi, S. I.; Han, S. W. Universal sulfide-assisted synthesis of M-Ag heterodimers (M = Pd, Au, Pt) as efficient platforms for fabricating metal-semiconductor heteronanostructures. J. Am. Chem. Soc. 2014, 136, 5221−5224. (14) Camargo, P. H. C.; Xiong, Y.; Ji, L.; Zuo, J. M.; Xia, Y. Facile Synthesis of Tadpole-like Nanostructures Consisting of Au Heads and Pd Tails. J. Am. Chem. Soc. 2007, 129, 15452−15453. (15) Sun, Y.; Xia, Y. Mechanistic Study on the Replacement Reaction between Silver Nanostructures and Chloroauric Acid in Aqueous Medium. J. Am. Chem. Soc. 2004, 126, 3892−3901. (16) Haiss, W.; Thanh, N. T. K.; Aveyard, J.; Fernig, D. G. Determination of size and concentration of gold nanoparticles from UV-vis spectra. Anal. Chem. 2007, 79, 4215−4221. (17) Mehrer, H. Diffusion in Solids; Springer: 2007. (18) Smigelskas, A. D.; Kirkendall, E. O. Zinc Diffusion in Alpha Brass. Trans. Am. Inst. Min. Metall. Eng. 1947, 171, 130−142. (19) Darken, L. S. Diffusion, Mobility and Their Interrlation through Free Energy in Binary Metallic Systems. Trans. Am. Inst. Min. Metall. Eng. 1948, 175, 184−201. (20) González, E.; Arbiol, J.; Puntes, V. F. Carving at the Nanoscale: Sequential Galvanic Exchange and Kirkendall Growth at Room Temperature. Science 2011, 334, 1377−1380. (21) Thota, S.; Chen, S.; Zhao, J. An unconventional mechanism of hollow nanorod formation: asymmetric Cu diffusion in Au-Cu alloy H
DOI: 10.1021/acs.chemmater.8b04630 Chem. Mater. XXXX, XXX, XXX−XXX