Surface Composition and Crystallinity of Coalescing Silver–Gold

Nov 7, 2017 - Bimetallic nanoparticles exhibit catalytic, optical, electronic, and magnetic synergy between their constituent metals. Typically, that ...
0 downloads 4 Views 5MB Size
Download PDF
Surface Composition and Crystallinity of Coalescing Silver−Gold Nanoparticles Eirini Goudeli and Sotiris E. Pratsinis* Particle Technology Laboratory, Institute of Process Engineering, Department of Mechanical and Process Engineering, ETH Zürich, Sonneggstrasse 3, CH-8092 Zürich, Switzerland S Supporting Information *

ABSTRACT: Bimetallic nanoparticles exhibit catalytic, optical, electronic, and magnetic synergy between their constituent metals. Typically, that synergy is traced to the domain structure and surface characteristics of such particles. Here these characteristics of coalescing Ag−Au nanoparticles of various initial sizes and morphologies (segregated or alloys) are investigated by atomistic molecular dynamics (MD) at different temperatures. Silver atoms exhibit increased mobility over Au and occupy gradually the surface of the coalesced (or sintered) bimetallic particle, consistent with scanning electron microscopy and selective O2 chemisorption experiments for heterogeneous catalysis of ethylene oxidation. The characteristic sintering time of equally sized Ag−Au nanoparticles is similar to that of pure Au but shorter than that of Ag nanoparticles. When the latter coalesce with substantially bigger Au ones, a patchy Ag layer is formed at the Au particle surface. However, when Ag nanoparticles are bigger, then Au is rather embedded into Ag, consistent with microscopy data. Most notably, X-ray diffraction (XRD) patterns of Ag−Au nanoparticles are obtained by MD, distinguishing segregated from alloyed ones. The latter exhibit a weaker XRD reflection of the (200) crystalline plane and, most distinctly, form smaller crystal size (highly polycrystalline) than coalescing pure and segregated Ag and Au nanoparticles, quantitatively explaining the structure of flame-made Ag−Au nanoparticles for biomaterial applications. KEYWORDS: bimetallics, crystallinity, sintering rate, gold−silver alloys, XRD, molecular dynamics

B

bimetallic nanoparticles in energy applications. Such interactions dictate whether the catalyst is poisoned (strong binding of reaction species) or nonreactive (weak binding of reactants). Adsorption, however, depends on nanoparticle surface morphology and composition. Silver nanoparticles exhibit high antibacterial activity, having applications in biomedical products, such as catheters7 and wound dressings. These nanoparticles, however, could be cytotoxic, rendering their use as therapeutic agents rather challenging.8 Alloyed silver−gold is an attractive alternative with potential in theranostic applications, as it has improved biocompatibility without degrading the plasmonic activity of nanosilver.9 Silver exhibits a stronger plasmonic performance than gold, but its surface oxidizes easily at ambient conditions, increasing nanosilver toxicity. When alloyed with gold, the cytotoxicity of silver decreases drastically due to suppressed Ag+ ion release without losing its superior plasmonic properties.9 Furthermore, Au−Ag nanoparticles are used as active centers for surface-enhanced Raman scattering (SERS), where particle

imetallic nanoparticles have frequently superior electronic, chemical, and plasmonic properties than monometallic nanoparticles.1 As a result, they have many biomedical, sensory, or catalytic applications, such as ammonia decomposition2 and oxygenate reforming.3 For example, gold nanoparticles exhibit reduced electron transfer on their (111) surface due to their high work function, hindering O2 adsorption. Gold-based bimetallic (e.g., with Ag) nanoparticles, however, overcome this by their enhanced affinity with O2.4 Also bimetallic particles with shell monolayers often exhibit distinct properties attributed to the lattice mismatch (strain effect) and charge transfer between layers (ligand effect), resulting in different electronic properties.5 Furthermore, such an admetal may affect reactant adsorption (ensemble effect) and extent of the admetal coverage and diffusion into the bulk, altering the adsorbate binding energy.5 Even slight changes in nanoparticle size, structure, or composition can influence the physicochemical properties and, thus, the performance of bimetallic nanoparticles.6 Yet, there is only a limited understanding of the structure and mixing of the two constituent metals. Furthermore, understanding metal−adsorbate interactions is key to controlling and improving the functionality of © 2017 American Chemical Society

Received: September 21, 2017 Accepted: November 7, 2017 Published: November 7, 2017 11653

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

www.acsnano.org

Cite This: ACS Nano 2017, 11, 11653-11660

Article

ACS Nano composition and anisotropy at the nanoscale are key issues.10 More specifically, the SERS substrate is affected by hot spots at sharpened corners and edges11 that lead to a highly localized electromagnetic field. The conjunctions in Ag−Au aggregates also contribute vastly to SERS sensitivity, as they produce coupled plasmon bands and electromagnetic hot spots.12 Molecular dynamics (MD) allows tracking the detailed motion of atoms and can provide valuable insight into early stages of Ag−Au particle formation, such as nucleation, sintering, and coalescence, which are hard to follow experimentally. This advances their quantitative understanding and can facilitate the assembly of process models from firstprinciples for their synthesis.13 Here, the effects of temperature and initial particle morphology on sintering rate, aggregation, and structure of coalescing Ag−Au nanoparticles are investigated by MD. The surface composition and crystallinity of the resulting nanoparticles are elucidated, as they greatly affect performance in catalysts and biomaterials. Most importantly, the X-ray diffraction (XRD) patterns of alloyed and segregated Ag−Au nanoparticles are calculated by molecular dynamics,14 revealing distinct XRD characteristics for bimetallic alloys and segregated morphologies. This allows the quantitative distinction of such structures in biomedical and catalytic applications. Most specifically it is shown that (a) silver preferentially occupies the surface of its alloys with gold during their formation by coalescence, explaining experimental data with such catalysts,15 and (b) alloys form smaller crystallites than pure metals and their segregated composites, in agreement with data from flame-made Ag−Au nanoparticles immobilized on silica for biomaterial applications.9

Figure 1. Evolution of (a) normalized surface area of two freestanding Au (orange line) and Ag nanoparticles (gray line), as well as segregated (black line) and highly mixed (alloyed) Ag and Au atom structures (blue line) and (b) their average disorder variable, D, during sintering or coalescence at T = 800 K. All nanoparticles have an initial diameter dp,0 = 3 nm. The horizontal line in Figure 2a indicates the characteristic sintering time, τs (the time needed for the excess surface area to decrease by 67%). All particles exhibit an increased degree of disorder during adhesion (t ≈ 10−3−10−1 s) but alloyed ones maintain their low crystallinity at all times.

RESULTS AND DISCUSSION Coalescence Rate. Figure 1a shows the evolution of (a) the normalized surface area for two free-standing Au (orange lines), Ag (gray lines), Ag−Au segregated (black lines), and alloyed (blue lines) nanoparticles with dp,0 = 3 nm, coalescing at T = 800 K in the canonical (NVT) ensemble. The lightly colored shades around each line show the variability of multiple MD simulations. The horizontal line indicates the characteristic sintering time, τs, defined as the time needed for the sinter neck length to reach 83% of the initial primary particle radius.16 This corresponds to about 67% reduction of the excess surface area of the equivalent fully coalesced spherical particle18 (or to about 86% of the total initial surface area of the two coalescing particles).19 The τs can be defined also as the time for reduction of the excess surface area by (1 − 1/e) = 63% (or 87% of the initial surface area) by exponential phenomenological coalescence models,20 as has been used to describe the coalescence rate of Ag nanoparticles.21 The MD simulations had been validated by calculating the melting point of silver and gold nanoparticles as a function of particle size by monitoring the evolution of the Lindemann index.22 They were in excellent agreement with MD21,22 and experimental23 literature for pure Au and Ag. The melting point of Ag−Au nanoparticles was largely in between that of the pure metals (Figure S1). For pure Ag and Au as well for their nanocomposites, sintering starts by adhesion (t < 0.1 ns), and a neck is formed between the two coalescing nanoparticles. The resulting concave region around the sinter neck is gradually filled by surface diffusion,19 while the total surface area decreases. The overall particle shape evolves from oval to spherical-like19 (within tenths of nanoseconds). The evolution of the normalized (to initial one) surface area is similar for all

particles, while that of monometallic ones is identical to that in the literature for Ag21 and Au.22 The initial sintering rates of segregated and alloyed Ag−Au nanoparticles follow that of Au, although the Ag−Au alloy (blue line) attains a spherical-like shape faster than the segregated and even that of pure Au nanoparticles. It is worth noting that increasing the Au content (5%, 10%, or 50%) of these composites gradually shifts their surface area evolution (and sintering rate) toward that of pure Au (Figure S2), as shown in Figure 1a. Increasing the initial particle size increases all τs, consistent with the literature,19 while Figure S3 shows exemplarily the effect of particle size on τs for the segregated Ag−Au nanoparticles. Figure 1b shows the average crystallinity of the above pure and bimetallic nanoparticles through their average disorder variable, D. Small D ( 0.1 ns) than all other orientations. In contrast, when such fusion takes place between a Ag (100) terrace and a Au corner atom (Figure 2, squares), the attainment of the spherical-like shape is faster than all other facet orientations. Similarly, alloyed nanoparticles exhibit almost the same surface area evolution and identical characteristic sintering time, regardless of their initial facet orientation (Figure S5). In contrast to segregated Ag−Au (Figure 2), they exhibit smaller differences in long-term particle reshaping among the various orientations than segregated nanoparticles. Increasing initial particle size does not change the above picture for the various orientations. Figure S6 shows the evolution of surface area of the two extreme (with respect to coalescence rate) facet orientations for larger nanoparticles (dp,0 = 4 nm) at the same T. Increasing particle diameter by 33% doubled τs, indicating that particle size is more important for the coalescence rate than facet orientation. Surface Composition. Figure 3 shows the evolution of the surface Ag fraction during sintering of two free-standing Ag and

distortion of the initial crystalline structure as the outer layers come in contact. Later on (t > 10−1 ns), particles recrystallize but at a different rate, depending on starting structure. Then, D decreases and gradually levels off, consistent with the evolution of the normalized surface area (Figure 1a) when coalescing particles become rather elongated and oval-like.19 The alloyed nanoparticles exhibit the largest degree of disorder (Figure 1b) and sinter faster than more crystalline ones, as atoms in the former are more mobile, allowing quick restructuring and neck formation between the coalescing nanoparticles. Furthermore, the average disorder variable, D, of sintered nanoparticles (t > 1 ns) is lower than that initially, as particle size has increased by sintering, leading to more crystalline nanoparticles after prolonged coalescence. The Lindemann index follows the evolution of the disorder variable, D, although it does not capture fine changes of crystallinity as well as D does (Figure S4). Figure 1a shows also that the initial surface area drops sharply during adhesion (t < 0.1 ns). As a result, the initial contact point between coalescing particles and subsequently their facet (terrace, edge, or corner) orientation can affect their coalescence rate. Figure 2 shows the coalescence rate of

Figure 2. Evolution of normalized surface area by coalescence of two segregated Ag−Au nanoparticles with dp,0 = 3 nm at T = 800 K that adhere to each other at different initial facet orientations: (1) Ag (100) terrace with Au (100) terrace (circles), (2) Ag (100) terrace with Au corner atoms (squares), (3) Ag corner atoms with Au corner atoms (triangles), (4) Ag edge with Au corner atom (diamonds), (5) Ag edge with Au terrace (crosses) or Ag edge with Au edge (side triangles), and (6) Ag edge with Au edge (side triangles). The characteristic sintering time (the time needed to reduce initial surface area40 to 86%) is hardly affected by the initial facet orientation, while there are significant differences in the time needed to attain the final spherical-like shape (79% of initial area).

Figure 3. Evolution of the surface Ag fraction of coalescing Ag−Au segregated nanoparticles with dp,0 = 4 nm at T = 600 (blue line), 700 (green line), and 800 K (red line). The surface Ag composition increases from initially 0.5 up to about 0.63 (for T = 800 K) as Ag diffuses onto the surface of the Au nanoparticle. The Ag surface fraction of the sintered particle increases with temperature. The shaded region represents the variation of three simulations.

Au nanoparticles (insets) with dp,0 = 4 nm at T = 600 (blue lines), 700 (green lines), and 800 K (red lines). The shaded regions around the 600, 700 and 800 K lines represent exemplarily the variation of three replicate MD simulations. When sintering starts (t ≈ 10−3 ns), the Ag surface fraction increases rapidly from about 0.5 as Ag diffuses over the Au nanoparticle surface. At longer times, that fraction attains asymptotically a value that depends on temperature. Increasing that temperature increases the Ag surface fraction. For example, Ag atoms occupy about 58% of the composite particle surface after 30 ns at T = 800 K compared to 53% at 600 K (Figure 3).

segregated Ag−Au nanoparticles of 3 nm initial diameter (Figure 1a black line) at various facet orientations: terrace− terrace (circles), terrace−corner (squares), corner−corner (triangles), edge−corner (diamonds), edge−terrace (crosses), and edge−edge (side triangles). Also a broken line at 86% of their normalized surface area is shown, the definition of the characteristic time16 for coalescence or sintering, τs. Clearly τs is not affected by the facet orientation. If the exponential model of coalescence20 is used for the definition of τs, the broken line would be a bit higher, at 0.87, instead of 0.86, in the ordinate of 11655

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

ACS Nano

thin Ag monolayer (t = 10 ns). In contrast, sintering of a 4 nm Ag with a 2 nm Au nanoparticle (Figure 4b) results in segregated structures, as the latter is embedded into the large Ag particle, consistent with HRTEM images of Ag−Au sintering by electron beam irradiation (Liu and Sun:24 Figure 6). The present results are consistent also with global optimization methods25 of 34- and 38-atom Ag−Au clusters that show the prevalence of Ag on the composite cluster surface. Such patchy Au−Ag core−shell structures exhibit enhanced electrocatalytic activity compared to their monometallic counterparts.26 Figure 5 shows the surface Au fraction of alloyed Ag−Au nanoparticles as a function of bulk Au fraction by the present

The inset snapshots show the surface composition of the nanoparticles having coalesced for t = 100 ns at 600 and 800 K. At that time (t = 100 ns), the Ag atoms occupy 63% of the composite particle surface at 800 K but only 53% at 600 K. The Au nanoparticle morphology changes less than that of Ag due to its lower cohesive energy, surface tension, and heat of sublimation, consistent with MD simulations and electron microscopy images of Au nanoparticles deposited on Ag nanorods.17 Segregation is not induced by strain, as Ag and Au have almost identical lattice constants. Decreasing the nanoparticle size accelerates the spread of Ag over Au. For example, at 30 ns for nanoparticles with dp,0 = 3 nm the above fractions are 54% and 61% at 600 and 800 K (Supporting Information: Figure S7). Figure 4 shows cross-section snapshots of unequally sized Ag (blue) and Au (yellow) nanoparticles of (a) dp,0,Ag = 2 and

Figure 5. Variation of surface Au fraction of alloyed Ag−Au nanoparticles with bulk Au fraction by MD simulations at temperature T = 523 K (diamonds) and electron spectroscopy for chemical analysis (ESCA: circles) and combined scanning electron microscopy (SEM) and selective O2 chemisorption experiments (triangles).15 The MD-obtained surface Au composition at T = 523 K is in excellent agreement with experiments for most bulk Au compositions (below 0.7), indicating surface enrichment in Ag.

MD simulations (filled diamonds) at T = 523 K. Also the electron spectroscopy for chemical analysis (ESCA: circles) and combined scanning electron microscopy (SEM) and selective O2 chemisorption (triangles) data obtained by impregnation onto an α-alumina support and calcination at T = 250 °C for 24 h15 are shown in Figure 5. The simulations are in excellent agreement with the data up to about 70% bulk Au content and well below the 1:1 line for homogeneous Ag/Au mixing, quantitatively proving the surface enrichment in Ag of these Ag−Au composite particles. At larger bulk Au fractions (>70%), even though MD simulations are consistent with the data, they overpredict them, as simulations exhibit large variation at high bulk Au fractions. Nanoparticle Crystallinity. Figure 6 shows the XRD pattern (orange line) of two spherical Au primary particles both with dp,0 = 4 nm at point contact (t = 0 ns). They exhibit four characteristic peaks at 2θ = 38.2°, 44.4°, 64.6°, and 77.5° corresponding to (111), (200), (220), and (311) of bulk Au atomic planes, respectively. The MD-obtained XRD pattern is in excellent agreement with experiments27 (Uppal et al., 2013: black pattern) of Au dendritic structures that consist of polydisperse primary particles with dp,0 = 4−10 nm at room temperature. Even though the diffraction angles of the MDderived Au peaks are identical with experimental ones, the

Figure 4. Cross-section snapshots of Ag−Au nanoparticles with initial (a) dp,0,Ag = 2 and dp,0,Au = 4 nm and (b) dp,0,Ag = 4 and dp,0,Au = 2 nm coalescing at T = 800 K at t = 0, 0.1, 1, and 10 ns. The Ag atoms are blue, while the Au atoms are yellow. Initially (t ≤ 0.1 ns), adhesion takes place and practically the same particle morphology is obtained. Later (t > 0.1 ns), however, when the small Ag nanoparticle coalesces with a bigger Au nanoparticle, the Ag atoms diffuse on the Au surface, forming a thin layer (t = 10 ns). On the other hand, sintering of a large Ag with a smaller Au nanoparticle results in segregated particle domains with little diffusion of Au over the Ag surface.

dp,0,Au = 4 nm and (b) dp,0,Ag = 4 and dp,0,Au = 2 nm coalescing at T = 800 K and t = 0, 0.1, 1, and 10 ns, while their corresponding 3D or full-size snapshots are shown in Figure S8. Initially (t ≤ 0.1 ns), adhesion and sintering take place, forming rapidly compact composite particles, regardless of the Ag to Au particle size ratio: the smaller particle (or droplet here) fuses to the bigger one, which hardly changes shape. Later on (t > 0.1 ns), however, that size ratio matters: when the small Ag nanoparticle coalesces with a bigger Au one (Figure 4a), the Ag atoms diffuse quickly onto the rather rigid Au surface, forming a 11656

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

ACS Nano

most of the surface atoms tend to be amorphous (green to red atoms), especially at T = 800 K. Both Au and Ag nanoparticles exhibit characteristic peaks at 2θ = 38.2°, 44.4°, 64.6°, and 77.5° (triangles) corresponding to (111), (200), (220), and (311), respectively, of bulk Au and Ag atomic planes.31 At T = 800 K (Figure 7b), the peaks broaden compared to T = 600 K, indicating less crystalline nanoparticles, consistent with pure Ag21 and pure Au.22 Most importantly, the XRD patterns of alloyed nanoparticles (Figure 7a) exhibit the (111), (220), and (311) characteristic peaks of Au and Ag but hardly the (200) peak. The reason that the (200) peak is weaker for Ag−Au nanoalloys is probably their lower crystallinity than that of segregated nanoparticles, as indicated by the local disorder variable (Figure 7, snapshots) and smaller crystal size due to formation of many crystalline grains (Figure 7a), in contrast to segregated ones, which exhibit three distinct crystalline grains (Figure 7b). So further research is needed to better understand the weaker reflection of (200) in alloyed than segregated nanoparticles. Nevertheless this can be used to qualitatively distinguish segregated from alloyed bimetallics. For example, Sotiriou et al.9 generated bimetallic Ag−Au immobilized on SiO2. In their XRD pattern (Figure S2a in ref 9) for 50% Au (equivalent to those in Figure 7), the (200) peak at 44° hardly appears and is comparable to that of the (220) plane at about 65°. Their pattern is similar to those of alloyed rather than segregated nanoparticles of Figure 7, for which the (200) peak distinctly appears and is much stronger than the (220) peak at all T. This comparison indicates that the experimentally made Ag−Au were alloyed rather than segregated, consistent with their low release of Ag+ ions compared to mechanically mixed (and segregated) Ag and Au nanoparticles.9 Alloyed nanoparticles (Figure 7a) form crystallites of notably smaller size than segregated ones (Figure 7b) at all temperatures. For example, at 800 K alloyed and segregated nanoparticles develop crystals of 1.72 and 3 nm average size, respectively, even though they were formed by coalescence of two particles with identical starting size dp,0 = 4 nm. Figure 8 shows the evolution of dXRD during coalescence of pure Ag (gray line) and Au (orange line) as well as Ag−Au segregated

Figure 6. MD-obtained XRD pattern of a Au dimer (orange line) consisting of two spherical primary particles with dp,0 = 4 nm in point contact compared to experiments27 (black line) of Au dentritic structures that consist of polydisperse primary particles with dp,0 = 4−10 nm. Excellent agreement between the two patterns is obtained at the main reflections from the (111), (200), and (220) atomic planes.

former are more detailed. This difference can be attributed to the primary particle polydispersity in the experiments, contrary to the simulated Au nanoparticles, which consist of two monocrystalline and equally sized spheres. Figure 7 shows XRD patterns of (a) alloyed and (b) segregated (Figure 3) Ag−Au nanoparticles with dp,0 = 4 nm

Figure 7. XRD patterns of (a) alloyed and (b) segregated nanoparticles with dp,0 = 4 nm having coalesced for t = 50 ns at 600 (blue patterns), 700 (green pattern), and 800 K (red patterns) with the corresponding cross-section images and average crystal sizes.

that had coalesced for t = 50 ns at 600 (blue patterns), 700 (green patterns), and 800 K (red patterns) along with the average crystallite size. The particle cross-section snapshots (insets) are colored based on the local disorder variable that refers to each atom in the particle.28−30 Its average over all atoms, D, indicates the particle crystalline structure (Figure 1b). FCC-like structured atoms are blue (high crystalline order), while disordered ones are red (low crystalline order). The nanoparticles are crystalline (blue atoms) in the bulk, while

Figure 8. Evolution of the XRD diameter of Ag (gray lines), Au (orange lines, inset), Ag−Au segregated (black lines), and alloyed (blue lines) nanoparticles with dp,0 = 4 nm coalescing at 600 K. Alloyed nanoparticles exhibit the smallest crystallite size of all nanoparticles. 11657

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

ACS Nano (black line) and alloyed (blue line) nanoparticles with dp,0 = 4 nm at 600 K. Silver nanoparticles reveal the largest dXRD, while that of segregated nanoparticles is bracketed in between those of pure Ag and Au, similarly to the evolution of the normalized surface area and disorder variable (Figure 1). Nanoalloys, however, exhibit the smallest crystallite size, dXRD = 2.55 nm. At T = 800 K (Supporting Information: Figure S9), the dXRD drops to about 3 nm for segregated and to 1.6 nm for alloyed nanoparticles, at all times, due to the larger fraction of amorphous atoms at the particle surface than at 600 K, as discussed in Figure 7. For alloys the dXRD is about 50% (Figure 8) and 70% smaller (Figure S9) at 600 and 800 K, respectively, than that of the corresponding spherical nanoparticle of the same volume. As a result, alloyed Ag−Au nanoparticles have higher polycrystallinity and smaller crystallites than segregated ones. This indicates the potential of discerning the attainment of an alloyed state by bimetallic nanoparticles through XRD or high-resolution TEM image counting. Most importantly, these simulations reveal that Ag−Au nanoparticles made in flames and immobilized on SiO2 supports were alloyed, as they had crystal sizes much smaller than pure Ag or Au on SiO2.9 Focusing on crystallite size provides a quantitative criterion for discerning the alloyed or segregated state of bimetallics, which is much stronger than the qualitative analysis of XRD patterns (Figure 7). Figure 9 shows the evolution of dXRD during coalescence of unequally sized Ag−Au nanoparticles (Figure 4). For initial Ag

Therefore, the atomic composition or structure of Ag−Au nanoparticles affects significantly the XRD patterns and crystallite size of the resulting bimetallic nanocomposite.

CONCLUSIONS The surface composition and crystallinity during sintering or coalescence of free-standing Ag and Au nanoparticles are investigated by molecular dynamics at 300−800 K. The surface composition and anisotropy of Ag−Au nanoparticles is a key issue that affects their electronic, chemical, and plasmonic properties. Equilibrium of Ag−Au alloys at 523 K leads to nanoparticle surface enrichment in Ag, consistent with catalytic characterization experiments.15 Similarly, sintering of equally sized Ag and Au nanoparticles results in segregated nanostructures with a Ag-enriched surface, where Au remains almost rigid, consistent with MD of fusion of Au rods with Ag nanoparticles. Enhanced surface coverage by Ag or embedded Au domains are created when Au nanoparticles coalesce with smaller or larger Ag particles, respectively. Most importantly, XRD patterns during particle sintering or coalescence of Ag− Au are obtained by MD. These patterns reveal that the (200) peak is weaker for Ag−Au nanoalloys that have also the smallest crystal size (dXRD) of all other configurations of equal initial size. This offers a possibility for distinction of the alloyed state of bimetallic particles by XRD measurements as with flame-made Ag−Au nanoparticles immobilized on silica that had exhibited substantially lower release of Ag+ ions than mechanically mixed and segregated Ag and Au nanoparticles.9 The present MD results can facilitate process design13 for synthesis of alloyed or segregated Ag−Au nanoparticles for heterogeneous catalysis, bioimaging, and bactericidal applications. For example, Figure 3 nicely shows that increasing the process temperature facilitates particle sintering (a well-known fact16,20) but at the same time increases the surface coverage of the resulting Ag−Au composites with silver and not gold! Also, Figure 4 demonstrates that fusing unevenly sized Ag and Au nanoparticles increases the Ag surface coverage and not that of gold due to the above tendency of silver. This was corroborated nicely with experiments, at least up to 0.6 bulk Au fraction15 (Figure 5). Furthermore, the synthesis of bimetallic alloys or segregated Au−Ag nanoparticles is a long-standing challenge in the community, especially in catalysis and biomaterial engineering. The present MD simulations show the prominence of a distinct XRD reflection at 44° (corresponding to the (200) plane) for segregated bimetallics compared to alloyed ones (Figure 7). Most important (and more accessible proof addressing the above challenge) is the formation of systematically smaller crystallites in alloyed than in segregated Ag−Au (Figures 7−9 and S9). This introduces a new way of distinguishing Ag−Au heterostructures and could guide process design for synthesis of alloyed or segregated Au−Ag nanoparticles.

Figure 9. Evolution of crystallite size of Ag−Au segregated nanoparticles with initial dp,0,Ag = 2 and dp,0,Au = 4 nm (blue line, triangles) and dp,0,Ag = 4 and dp,0,Au = 2 nm (black line, circles) at 800 K. For Ag particles smaller than Au ones (triangles), the composite particle has a larger average crystallite size than those of particles consisting initially of Ag particles larger than Au (circles).

particles smaller than Au (triangles), the composite particle has larger crystal size than initially larger Ag than Au particles (circles). This is attributed to the formation of the Ag layer around Au that hardly changes the shape and crystallinity of the underlying Au particle (Figure 4a and S8a). However, when Au is smaller than Ag, it is embedded into Ag, changing its shape, as discussed in Figures 4b and S8b. This embedding reduces the Ag crystallinity, resulting in smaller crystal domains (Figure 9, insets), as shown by their distinctly smaller average dXRD (circles). Even though, initially (t = 10−4 ns), the dXRD of the two configurations are comparable (about 2.5−2.7 nm), their difference increases after adhesion and sintering (t > 10−3 ns).

METHODS Molecular Dynamics. Spherical Au, Ag, and Ag−Au nanoparticles were extracted from a large cubic FCC crystal with a cutoff (particle) diameter dp,0 = 2−4 nm. These FCC crystals initially were perfect spheres and were equilibrated at various temperatures, T = 600−800 K, for 1 ns, resulting in facets on their surface. The equations of motion are integrated by the velocity-Verlet algorithm32 with a time step of 1 fs. Two unsupported nanoparticles are placed next to each other for sintering in a vacuum with a separation distance of about 0.4 nm between the closest atom centers and equilibrium. Sintering 11658

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

ACS Nano N

simulations between two Ag−Ag, Au−Au, or Ag−Au nanoparticles are carried out in the canonical ensemble (NVT: constant number, volume, and temperature). The Ag−Au alloyed nanoparticles are formed by heating above the bulk melting point (up to 1600 K) and then equilibrated at constant T. Here the many-body embedded-atom method potential33,34 is employed with parametrization based on the Finnis−Sinclair formulation35 as obtained by Ward et al.,36 while Brink et al.37 have found good agreement of the potential of Ward et al.36 with that by Mendelev et al.38 for Cu−Zr alloys. The total energy, Etot, is Etot =

∑ Fi(ρh ,i ) + i

1 2

∑

ϕij(R ij)

i,j≠i

D̅ (t ) =

i=1

dXRD

1 Nb(i)

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.7b06727. Figures showing the melting point of Ag, Au, and Ag−Au nanoparticles, the evolution of the normalized surface area of Ag−Au nanoparticles for different Au compositions, the detailed characteristic sintering times of segregated particles for several sizes and temperatures, the temporal evolution of the Lindemann index during coalescence of Ag, Au, and Ag−Au nanoparticles, the evolving surface area for different initial particle orientation, the surface Ag fraction of coalescing Ag and Au nanoparticles, 3D snapshots of coalescing unevenly sized Ag and Au particles, and the evolution of XRD sizes of coalescing Ag, Au, and Ag−Au nanoparticles with a diameter of 4 nm (PDF)

(1)

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Ph: +41 44 632 31 80. Fax: +41 44 632 15 95. ORCID

Eirini Goudeli: 0000-0001-8056-6941 Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS The research was supported by the Swiss National Science Foundation (grant no. 200021_149144) and the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013, ERC grant agreement no. 247283).

(2)

REFERENCES (1) Sinfelt, J. H. Bimetallic Catalysts: Discoveries, Concepts and Applications; John Wiley & Sons: New York, 1983. (2) Hansgen, D. A.; Vlachos, D. G.; Chen, J. G. Using First Principles to Predict Bimetallic Catalysts for the Ammonia Decomposition Reaction. Nat. Chem. 2010, 2, 484. (3) Skoplyak, O.; Menning, C. A.; Barteau, M. A.; Chen, J. G. Reforming of Oxygenates for H2 Production on 3d/Pt(111) Bimetallic Surfaces. Top. Catal. 2008, 51, 49−59. (4) Nakatsuji, H.; Hu, Z.-M.; Nakai, H.; Ikeda, K. Activation of O2 on Cu, Ag, and Au Surfaces for the Epoxidation of Ethylene: Dipped Adcluster Model Study. Surf. Sci. 1997, 387, 328−341. (5) Mavrikakis, M.; Hammer, B.; Nørskov, J. K. Effect of Strain on the Reactivity of Metal Surfaces. Phys. Rev. Lett. 1998, 81, 2819−2822. (6) Zhai, H. J.; Li, J.; Wang, L. S. J. Icosahedral Gold Cage Clusters: M@Au12 (M = V, Nb, and Ta). J. Chem. Phys. 2004, 121, 8369−8374. (7) Samuel, U.; Guggenbichler, J. P. Prevention of Catheter-Related Infections: the Potential of a New Nano-Silver Impregnated Catheter. Int. J. Antimicrob. Agents 2004, 23, 75−85. (8) Sotiriou, G. A.; Pratsinis, S. E. Engineering Nanosilver as an Antibacterial, Biosensor and Bioimaging Material. Curr. Opin. Chem. Eng. 2011, 1, 3−10.

Nb(i)

∑ Ylm(rij) (3)

j=1

where Nb(i) is the number of nearest neighbors around particle i and Ylm(rij) are the spherical harmonic functions of degree l and −l ≤ m ≤ l. Here, l = 6 as Au and Ag nanoparticles are arranged in the FCC crystal structure.30 From this equation the local crystal structure is effectively described by the average local bond order parameters:29

qlm ̅ (i) =

1 Nb(i)

Nb(i)

∑ qlm(j) j=1

(4)

Thus, the extent of system disorder is given by the disorder variable:30 Di(t ) =

1 Nb(i)

Nb(i)

l

∑ ∑ j = 1 m =−l

|qlm(i , t ) − qlm(j , t )|2

(6)

ASSOCIATED CONTENT

where λ = 0.154 nm is the wavelength of the X-ray and θmax is the diffraction angle corresponding to the maximum of the largest peak. The first two peaks of the XRD pattern are fitted using a first- or second-order Gaussian model. Crystallinity Dynamics. The crystallinity of the system is characterized here by its degree of disorder quantified by the bond order (or Steinhardt) parameters,28 qlm, which are measures of the local and extended orientational symmetries of the particles: qlm(i) =

Di(t ) N − Nb(i)

where N is the total number of atoms. Small values indicate an FCClike crystal structure and big ones a more disordered environment. D̅ is close to unity for glassy or liquid materials and decreases significantly for crystal structures.30

where Fi is the embedding term accounting for the local electron density, ρh,i, ϕij(Rij) is the potential for pairwise interactions between atoms, and Rij is the position vector of atoms i and j. The above potential36 is a match between well-established elemental potentials (Ag and Au) based on a small database of DFT calculations. The simulations are carried out using the LAMMPS39 MD code. Characteristic Sintering Time and Nanoparticle Characterization. When particles coalesce, they minimize their free energy by reduction of their surface area, resulting in elongated and eventually spherical structures. The evolution of the surface area is determined using MSMS 6.2.1.40 The probe radius is set equal to 2.25 Å, corresponding to the N2 molecule,19 while the van der Waals radius of gold and silver atoms is 1.66 and 1.72 Å, respectively.41 This method of surface area calculation corresponds to the standard technique of specific surface area measurement by N2 adsorption.42 The atomic structure and degree of disorder of the Ag−Au nanoparticles are quantified by the local and average disorder variable as shown in Goudeli and Pratsinis22 (see Methods: Crystallinity Dynamics), while the particle crystallinity is described by the simulated XRD intensity as calculated from the structure factor based on Coleman et al.,13 which is different than that of Ahmed et al.,43 who had obtained the XRD patterns of various CuO crystal sizes. The nanoparticle crystal size is calculated based on the XRD patterns by the full-width at half-maximum of the fitted largest peak using Scherrer’s equation:44

0.9λ = (FWHM)cos(θmax )

∑

(5)

and the overall disorder variable, D̅ , is the average Di over all particles: 11659

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660

Article

ACS Nano

Densities in Three Dimensions. J. Chem. Phys. 2011, 135, 174109− 1−174109−8. (31) Venudhar, Y. C.; Iyengar, L.; Krishna Rao, K. V. Temperature Dependence of the Lattice Parameter and the Thermal Expansion of Ag-Au (50 at.%) Alloy by an X-ray Method. J. Less-Common Met. 1978, 60, 41−46. (32) Swope, W. C.; Anderson, H. C.; Berens, P. H.; Wilson, K. R. A Computer Simulation Method for the Calculation of Equilibrium Constants for the Formation of Physical Clusters of Molecules: Application to Small Water Clusters. J. Chem. Phys. 1982, 76, 637− 649. (33) Daw, M.; Baskes, M. Semiempirical, Quantum-Mechanical Calculation of Hydrogen Embrittlement in Metals. Phys. Rev. Lett. 1983, 50, 1285−1288. (34) Daw, M.; Baskes, M. Embedded-Atom-Method − Derivation and Application to Impurities, Surfaces and Other Defects in Metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1984, 29, 6443−6453. (35) Finnis, M. W.; Sinclair, J. E. A Simple Empirical N-Body Potential for Transition Metals. Philos. Mag. A 1984, 50, 45−55. (36) Ward, L.; Agrawal, A.; Flores, K. M.; Windl, W. Rapid Production of Accurate Embedded-Atom Method Potentials for Metal Alloys. preprint arXiv:12090619, 2012. (37) Brink, T.; Şopu, D.; Albe, K. Solid-State Amorphization of Cu Nanolayers Embedded in a Cu64Zr36 Glass. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91, 184103−1−184103−10. (38) Mendelev, M. I.; Kramer, M. J.; Ott, R. T.; Sordelet, D. J.; Yagodin, D.; Popel, P. Development of Suitable Interatomic Potentials for Simulation of Liquid and Amorphous Cu-Zr Alloys. Philos. Mag. 2009, 89, 967−987. (39) Plimpton, S. Fast parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1−19. (40) Sanner, M. F.; Olson, A. J.; Spehner, J. C. Reduced Surface: An Efficient Way to Compute Molecular Surfaces. Biopolymers 1995, 38, 305−320. (41) Bondi, A. van der Waals Volumes and Radii. J. Phys. Chem. 1964, 68, 441−451. (42) Brunauer, S.; Emmett, P. H.; Teller, E. Adsorption of Gases in Multimolecular Layers. J. Am. Chem. Soc. 1938, 60, 309−319. (43) Ahmed, A.; Elvati, E.; Violi, A. Size- and Phase-Dependent Structure of Copper(II) Oxide Nanoparticles. RSC Adv. 2015, 5, 35033−35041. (44) Rau, R. C. Routine Crystallite-Size Determination by X-ray Diffraction Line Broadening. Adv. X-Ray Anal 1962, 5, 104−116.

(9) Sotiriou, G. A.; Etterlin, G. D.; Spyrogianni, A.; Krumeich, F.; Leroux, J.-C.; Pratsinis, S. E. Plasmonic Biocompatible Silver−Gold Alloyed Nanoparticles. Chem. Commun. 2014, 50, 13559−13562. (10) Westcott, S. L.; Jackson, J. B.; Radloff, C.; Hallas, N. J. Relative Contributions to the Plasmon Line Shape of Metal Nanoshells. Phys. Rev. B: Condens. Matter Mater. Phys. 2002, 66, 155431−1−155431−5. (11) Rycenga, M.; Langille, M. R.; Personick, M. L.; Ozel, T.; Mirkin, C. A. Chemically Isolating Hot Spots on Concave Nanocubes. Nano Lett. 2012, 12, 6218−6222. (12) Jana, N. R.; Pal, T. Anisotropic Metal Nanoparticles for Use as Surface-Enhanced Raman Substrates. Adv. Mater. 2007, 19, 1761− 1765. (13) Pratsinis, S. E. Aerosol-based Technologies in Nanoscale Manufacturing: from Functional Materials to Devices through Core Chemical Engineering. AIChE J. 2010, 56, 3028. (14) Coleman, S. P.; Spearot, D. E.; Capolungo, L. Virtual Diffraction Analysis of Ni [010] Symmetric Tilt Grain Boundaries. Modell. Simul. Mater. Sci. Eng. 2013, 21, 055020. (15) Toreis, N.; Verykios, X. E.; Khalid, S. M.; Bunker, G. B. Bulk and Surface Characterization of Supported Ag-Au Alloy Catalysts. Surf. Sci. 1988, 197, 415−429. (16) Kobata, A.; Kusakabe, K.; Morooka, S. Growth and Transformation of TiO2 Crystallites in Aerosol Reactor. AIChE J. 1991, 37, 347. (17) Grouchko, M.; Roitman, P.; Zhu, X.; Popov, I.; Kamyshny, A.; Su, H.; Magdassi, S. Merging of Metal Nanoparticles Driven by Selective Wettability of Silver Nanostructures. Nat. Commun. 2014, 5, 1−5. (18) Xiong, Y.; Pratsinis, S. E. Formation of Agglomerate Particles by Coagulation and Sintering − Part I. A Two-Dimensional Solution of the Population Balance Equation. J. Aerosol Sci. 1993, 24, 283−300. (19) Buesser, B.; Gröhn, A.; Pratsinis, S. E. Sintering Rate and Mechanism of TiO2 Nanoparticles by Molecular Dynamics. J. Phys. Chem. C 2011, 115, 11030−11035. (20) Koch, W.; Friedlander, S. K. The Effect of Particle Coalescence on the Surface Area of a Coagulating Aerosol. J. Colloid Interface Sci. 1990, 140, 419. (21) Buesser, B.; Pratsinis, S. E. Morphology and Crystallinity of Coalescing Nanosilver by Molecular Dynamics. J. Phys. Chem. C 2015, 119, 10116−10122. (22) Goudeli, E.; Pratsinis, S. E. Crystallinity Dynamics of Gold Nanoparticles During Sintering or Coalescence. AIChE J. 2016, 62, 589−598. (23) Buffat, P.; Borel, J. P. Size Effect on the Melting Temperature of Gold Particles. Phys. Rev. A: At., Mol., Opt. Phys. 1976, 13, 2287. (24) Liu, Y.; Sun, Y. Electron Beam Induced Evolution in Au, Ag, and Interfaced Heterogeneous Au/Ag Nanoparticles. Nanoscale 2015, 7, 13687−13693. (25) Rossi, G.; Ferrando, R.; Rapallo, A.; Fortunelli, A.; Curley, B. C.; Lloyd, L. D.; Johnston, R. L. Global Optimization of Bimetallic Cluster Structures. II. Size-Matched Ag-Pd, Ag-Au, and Pd-Pt Systems. J. Chem. Phys. 2005, 122, 194309. (26) Song, Y.; Liu, K.; Chen, S. AgAu Bimetallic Janus Nanoparticles and their Electrocatalytic Activity for Oxygen Reduction in Alkaline Media. Langmuir 2012, 28, 17143−17152. (27) Uppal, M. A.; Kafizas, A.; Ewing, M. B.; Parkin, I. P. The Room Temperature Formation of Gold Nanoparticles from the Reaction of Cyclohexanone and Auric Acid; a Transition from Dendritic Particles to Compact Shapes and Nanoplates. J. Mater. Chem. A 2013, 1, 7351− 7359. (28) Steinhardt, P. J.; Nelson, D. R.; Ronchetti, M. BondOrientational Order in Liquid and Glasses. Phys. Rev. B: Condens. Matter Mater. Phys. 1983, 28, 784−805. (29) Lechner, W.; Dellago, C. Accurate Determination of Crystal Structures Based in Average Local Bond Order Parameters. J. Chem. Phys. 2008, 129, 114707−1−114707−5. (30) Kawasaki, T.; Onuki, A. Construction of a Disorder Variable from Steinhardt Order Parameters in Binary Mixtures at High 11660

DOI: 10.1021/acsnano.7b06727 ACS Nano 2017, 11, 11653−11660