Influence of Composition and Chemical Arrangement on the Kinetic

Subscriber access provided by UNIV MASSACHUSETTS BOSTON

Article

The Influence of Composition and Chemical Arrangement on the Kinetic Stability of 147-atom Au-Ag Bimetallic Nanoclusters Anna Louise Gould, Andrew J. Logsdail, and C. Richard A. Catlow J. Phys. Chem. C, Just Accepted Manuscript • DOI: 10.1021/acs.jpcc.5b03577 • Publication Date (Web): 17 Sep 2015 Downloaded from http://pubs.acs.org on September 22, 2015

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

The Journal of Physical Chemistry C is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The Journal of Physical Chemistry

The Influence of Composition and Chemical Arrangement on the Kinetic Stability of 147-atom Au−Ag Bimetallic Nanoclusters Anna L. Gould,*ab Andrew J. Logsdail,*b C. Richard A. Catlowab a

UK Catalysis Hub, Research Complex at Harwell (RCaH), Rutherford Appleton Laboratory, Harwell

Oxford Didcot, Oxon, OX11 0FA, UK. b

Kathleen Lonsdale Materials Chemistry, Department of Chemistry, University College London, 20

Gordon Street, London, WC1H 0AJ, UK. *

Corresponding authors: [email protected]; [email protected]

KEYWORDS: Bimetallic nanoparticles, nanoclusters, nanoalloys, Au, Ag, core@shell, kinetic stability, melting, martensitic transformation, molecular dynamics.

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract High calcination temperatures are often required in nanoparticle synthesis and an understanding of how heating affects the structure and chemical arrangement of bimetallic nanoclusters is essential to design efficient fabrication processes. We have investigated the kinetic stability of 147-atom Au−Ag nanoalloys with varying composition and chemical ordering using ensemble molecular dynamics simulations to replicate these high temperature conditions. Ag-rich nanoalloys undergo a diffusion-less ‘martensitic’ structural transition from cuboctahedral to icosahedral; the melting temperature (Tm) of the subsequent icosahedra is dependent on the Au:Ag stoichiometry. Core@shell chemical arrangements do not behave in a similar manner: Tm strongly depends on the shell component, with Au55@Ag92 exhibiting increased stability as a result of its icosahedral Au core. We also report a novel dependence of nanocluster phase transitions on the chemical arrangement, as shown by low temperature non-martensitic atomic diffusion of Ag atoms to the surface for Ag55@Au92. This finding establishes why Ag@Au chemical arrangements, in particular, are difficult to maintain experimentally. Overall, the kinetic phenomena observed helps to explain why particular morphologies and chemical arrangements are more abundant in experimental synthesis, and how post-processing affects structure and chemical arrangement.


Page 2 of 38

Page 3 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Bimetallic nanoclusters (“nanoalloys”) are nanometre-sized aggregates that exhibit distinctive properties from those associated with single atoms or bulk matter.1 A variety of sizes and chemical arrangements2-3 (mixed alloy, core@shell, segregated subcluster and onion-like A-B-A) are possible for nanoalloys, and this variation makes them suitable for widespread application in engineering, magnetism, and catalysis.4-7 Au−Ag nanoalloys are of current interest as they have been shown to improve catalytic activity for oxidation of carbon monoxide and H2 production in comparison to monometallic Au nanoclusters.8-9 Additionally, the tuneable optical properties, as a result of the localized surface plasmon resonance (SPR), allow the chemical composition to be monitored experimentally through ultraviolet-visible (UV-Vis) spectroscopy.10-13 Experimental synthesis of core@shell bimetallic nanoclusters often requires the use of a sequential deposition/reduction process, beginning from an initial “seed” core nanocluster, but synthesis is not always straightforward due to the competing formation of alloys. Jiang et al.14 synthesised Au@Ag nanoparticles by immersing a desolvated metal organic framework (MOF) in an aqueous precursor Au solution, followed by an Ag precursor, with respective reduction and drying at each stage. However, the reverse deposition sequence resulted in Au@AuAg nanoclusters, with an alloy surface, rather than Ag@Au. Mallin and Murphy15 also attempted synthesis of sub-10 nm nanoclusters, but these were determined to be alloys rather than core@shell arrangements. Spontaneous inter-diffusion at the boundary between the two metals in an initial core@shell has been reported by Shibata et al.,16 and Chen et al.,17 highlighting the susceptibility of Au-Ag nanoclusters to alloy formation.


3


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 38

In addition to the possible variation in chemical arrangement, many different low- and high-energy geometric configurations are accessible during synthesis including those with high-index surface morphologies.18-19 Consequently, an extra step is often implemented to maximise uniformity in sample preparation, achieved through capping agents such as polyvinyl pyrrolidone (PVP), which favours surfaces primarily being capped by (100) facets.20-21 These capping agents are then typically removed using a calcining treatment and thus, should nanocluster reactivity be dependent on the chemical arrangement, an understanding of the stability under elevated temperatures is essential; if chemical structure is vulnerable to operating conditions, the occurrence of interatomic diffusion and/or phase segregation may render the reaction futile. Theoretical investigations with molecular dynamics (MD) using interatomic potentials have shown that the thermal properties of bimetallic nanoclusters are more complex than those observed for monometallic equivalents. Mottet et al.22 found that the introduction of a single impurity dopant can elevate melting temperatures (Tm), as shown by the incorporation of a single Ni or Cu atom into icosahedral (Ih) Ag clusters of n atoms (n = 55, 147, 309, 561). However, the incorporation of an Au atom did not yield similar behaviour; thus they concluded that only atoms smaller than Ag tend to be incorporated inside the nanocluster, allowing the strained Ih to relax and enhancing its thermal stability. Likewise, canonical Monte Carlo (MC) simulations showed the addition of a single Cu atom to Au55 (Ih) increased Tm by up to 150 K.23 The thermodynamic stability of small Au-Ag clusters has been thoroughly investigated theoretically, using conformational space searches to attain the global minimum.24-28 However, the influence of kinetics is less well understood and, due to the high temperature calcination processes involved in wet chemistry synthesis,


4

Page 5 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


knowledge of the changes in shape and composition under prolonged heating would be highly beneficial for the improvement of fabrication processes. MD simulations of Ih AuxAg55−x by Chen et al.,29 where x is the number of Au atoms, identified a correlation between Tm and Ag content, in line with the melting properties of the monometallic nanoclusters (Ag55 = 525-600 K; Au55 = 290-460 K). Additionally, glass transitions were prominent in the Au-rich nanoalloys, with smooth transitions between structural isomers. Chen and Johnston30 also performed a selective MD simulation on a core@shell Au55@Ag92 bimetallic nanocluster, where a spontaneous cuboctahedron (CO) to Ih transformation was observed, with characteristics of a diffusion-less martensitic transformation (MT).31 Large (AuAg)x nanoalloys have also been studied by Qi and Lee, with x = 561 for (CO) and (Ih), and x = 584 for a decahedral motif (Dh);32 while shape had little effect on the melting process, its influence during a sustained anneal (600 K for 58 ns)

becomes significant for

core@shell arrangements: the CO forms an alloy surface quickest, followed by the Ih. The Dh retains its core@shell arrangement during annealing, and thus the alloy rate of formation is CO > Ih > Dh. In this work, we conduct a thorough investigation into the effects of heating on AuxAg147−x nanoclusters (where x is the number of Au atoms), as a function of composition (x = 0, 1, 13, 55, 92, 134, 146, 147) and chemical arrangement (alloy vs. core@shell). We aim to simulate the annealing process that is performed on nanoparticles during synthesis in order to remove protecting agents. We have started with shape-controlled nanoparticle geometries that are experimentally viable and desirable for applications in e.g. heterogeneous catalysis; we are particularly interested in the stability of these kinetically obtained structures rather than that of equilibrium polydisperse samples produced in the absence of any stabilising ligands.


5


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 38

Both gold and silver have a bulk face-centred cubic (FCC) crystal structure, and thus we use, among other motifs, the cuboctahedron as an experimentally obtainable fragment of the FCC crystal.31-33. Our choice of nuclearity, however, means that several other high-symmetry geometrical motifs exist (decahedron, icosahedron) and are energetically competitive with the cuboctahedron.

2. Computational Method The MD package DL_POLY_433 was used to perform simulations in the canonical ensemble (NVT), using the velocity Verlet algorithm34 to solve Newton’s equations of motion, with a timestep (∆t) of 0.001 ps. A two-stage equilibration process is used for the initial starting structures, first using the Berendsen thermostat,35-36 followed by the Nosé-Hoover thermostat,37-38 for a total relaxation time of 200 ps. The Nosé-Hoover thermostat gives the desired canonical ensemble but must be started close to equilibrium to prevent long-lived oscillations during thermalization;36 hence we attain our initial equilibrium using the Berendsen thermostat. Starting from a CO arrangement, AuxAg147−x nanoclusters were then subjected to a calcination heat treatment, increasing from 1 – 900 K, in increments of 10 K, with temperature controlled by the Nosé-Hoover thermostat. Each temperature step has a duration of 1000 ps, resulting in a heating rate of 10 K / ns. Previous literature values tend to vary between 0.8-100 K / ns, as this is slow enough to be close to the thermodynamic limit.29-30,39-40 The potential model used in this study is the Gupta potential,41 which is derived from the second moment approximation to tight binding theory.42-43 It is known to reproduce kinetics in AuAg systems, having been used previously by Chen et al. and Calvo et al..29,44 The general form of the Gupta potential consists of an attractive


6

Page 7 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


many body (Vim) and a repulsive pair (Vir) term, obtained by summing over all N atoms, to give a total energy: N

Vclus = ∑ (Vri − Vm i )

.

[1]

i=0

Vri and Vm i are defined as: N 𝑟𝑖𝑗 −p( −1) r Vi (rij )= ∑ A e 𝑟0

[2]

j≠i

and: N −2q(

2 Vm i (rij )= [∑ ξ e

𝑟𝑖𝑗 −1) 𝑟0 ]

1 2

[3]

j≠i

where rij is the interatomic distance between atoms i and j, and r0 is the equilibrium bond length. All other potential parameters (A, p, ξ and q) are fitted to reproduce experimental values for the cohesive energy, lattice parameters and independent elastic constants for the bulk crystal at 0 K. The parameters used in this study are taken from the work of Cleri and Rosato,45 presented in Table I. As we are focusing on the trends rather than absolute energies, we have chosen to remain with the original Cleri-Rosato parameters, given their previous use,46-48 and these parameters were also used to create the initial alloy arrangements in our own previous work.49 An interaction cutoff of 10.5 Å is enforced throughout; we consider that this does not artificially alter our results as the energy of the interactions at this distance are extremely small: -0.2974 and -0.04208 meV for Ag and Au, respectively. We consider that maintenance of the high miscibility between the two metals, as predicted by our chosen potential, is important for the study of kinetics; however we note that ideally we would include charge effects. Cerbelaud et al.50 have


7


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 38

previously remarked on the need to include charge transfer for AuAg systems in order to attain better agreement with DFT calculations, but the methodology of Ferrando and Fortunelli is unstable for non-equilibrium bond distances as encountered at high temperatures during molecular dynamics simulations.51 Table I. Gupta potential parameters as given by Cleri & Rosato.45 Bimetallic Au−Ag parameters are taken as the arithmetic mean of the pure parameters. Parameter

A / eV

r0 / Å

q

p

ξ / eV

Au−Au

0.2061

2.8843

4.036

10.229

1.790

Au−Ag

0.15445

2.8864

3.5875

10.5785

1.484

Ag−Ag

0.1028

2.8885

3.139

10.928

1.178

2.1 Analysis of Simulation Results As the nanoclusters are heated, fluctuations in the internal energy commonly correspond to morphological changes. The melting points were determined through the plotting of caloric curves, calculated by determining the change in total internal energy (E) relative to that of the starting arrangement: 𝐸diff = 𝐸𝑡 − 𝐸𝑡=0

[4]

where Et is the total internal energy at time t, Et=0 is the energy of the starting structure when t=0. Ediff is then averaged over the ten independent simulations to give . has the advantage that energetic transitions, such as the martensitic transformations that occur for Ag-rich nanoclusters, can be identified much more easily.22,39,52 However, the relative energies of different isomers may be of interest and thus the total internal energy is also plotted as a function of temperature in the supporting information (SI), in Figs. S1-S4.


8

Page 9 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Chemical and Structural Characterisation Chemical ordering within the nanocluster is analysed by mapping the radial chemical distribution function (RCDF). Working from the nanocluster’s centre of mass outwards, the molar ratio of Au is monitored within the volume delimited by two spherical surfaces, with radii rRCDF and rRCDF + ε, respectively (Fig. 1), where ε is the thickness. rRCDF is measured relative to the nanocluster’s centre of mass and ranges from zero to the maximum pair distance of the cluster in integer multiples of ε, which is fixed at 0.3 Å. Fig. 1. Schematic diagram of the division of the nanocluster into concentric spheres, used to map the RCDF. The distance between spherical surfaces is ε. No atoms are found within the volume delimited between the centre of mass and the first surface, at ε, which would result in a region of white space in the RCDF.

Structure type assignment for face centred cubic (FCC), body centred cubic (BCC), hexagonal close packed (HCP) and other phases is implemented as in the software package Ovito,53 using the adaptive common neighbour analysis (a-CNA).5455

Correct selection of the nearest neighbor radius cutoff parameter can be difficult

and so the a-CNA works without a fixed cutoff radius. The a-CNA is given as a percentage, which is averaged over the ten independent simulations. The percentages provided in this work concern only FCC and HCP structure types (referred to hereafter as and , respectively); percentages for all “other” structure types are collectively referred to as .


9


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 38

3. Results and Discussion 3.1 Pure and single dopant clusters During the calcination process, Ag147 undergoes a solid-solid MT from CO to Ih at 226 K (hereafter referred to as the transition temperature, Tt), identified by a reduction in the energy of the nanocluster (Fig. 2) and a change in the predominant crystal structure in the a-CNA analysis (Fig. 3). This solid-solid transition occurs as a rapid, diffusion-less process, occurring through cooperative displacement of atoms in the crystal structure, and is driven by the difference in energy between the parent and the product phases.30 We emphasise that our aim is to identify the trends due to differing chemical arrangements, rather than exact transition and/or melting points for specific systems, and so sustained annealing for longer periods of 100 ns was only considered at 150 K and 200 K for Ag147 in order to verify observations with respect to martensitic transformations (MTs). The newly-formed Ih is stable until melting at 688 K (Table II), as indicated by a sudden increase in the energy of the system (Fig. 2) and decrease in (Fig. 3). Fig. 3. shows that the perfect 147-atom CO geometry initially has = 37.4%, and no HCP features at 0 K; in contrast, the perfect Ih geometry has = 20.4%. As T is increased, there is a sharp increase in (and decrease in ) for Ag147 when T=210-230 K, indicative of the sudden MT to the Ih geometry. In contrast, for Au147 does not immediately go to zero when begins to increase at T = 300 K, but rather the FCC and HCP structure types appear to be in competition with one another. The competition between the two crystal structures increases the extent of deformation Au147 is able to undergo, and allows the maintenance of a facetted crystalline geometry during the heat treatment (Fig. 4, inset). The FCC and HCP phases are present until T ≈ 420-460 K, about 100


10

Page 11 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


K below Tm(Au147), after which only HCP features remain, with the solid-liquid transition occurring at 555 K.

Fig. 2. as a function of T for: pure Ag147, and the addition of a single gold dopant atom in (i) a low energy position [Au1Ag146] and (ii) a high energy (HE) position [Au1Ag146 (HE)]. A key is provided; the transition temperature, Tt, and the melting temperature, Tm, have been highlighted with upward and downward arrows, respectively. Inset: Ag147 at (a) 0 K, (b) 230 K, after a martensitic transition, and (c) 690 K, after melting.


11


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 38

Fig. 3. Average a-CNA percentages as a function of T for Ag147 and Au147. Solid lines represent and dashed lines represent ; a key is provided. A perfect CO geometry consists of = 37.4%, = 0%, and = 62.6%, whereas a perfect Ih geometry does not contain any FCC features ( = 20.4% and = 79.6%). Tt and Tm are indicated by dash-dotted and dotted lines, respectively.

Fig. 4. as a function of T for Au147 and the addition of a single Ag dopant atom in (i) a low energy (thermodynamically favourable at 0 K) position [Au146Ag1] and (ii) a high energy (HE) position (thermodynamically unfavourable at 0 K) [Au146Ag1 (HE)]. A key is provided and Tm is highlighted by a downward arrow. Inset: Au147 undergoes a step dislocation as it transitions from (a) CO at 0 K to (b) a facetted


12

Page 13 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


arrangement at 340 K. However, no significant energy change is noted before it melts (c), after which there are are energy fluctuations. The simplest combination of the two metals is achieved through the introduction of a single dopant into the monometallic nanocluster, resulting in Au1Ag146 and Au146Ag1 compositions. There are several permutations for the position of a dopant atom, with each arrangement resulting in different cluster energies, as assessed in our previous work,49 so we have used both the lowest and highest energy (HE) arrangements of these singly doped systems. All the single dopant structures we have used are provided in Fig. S5 of the SI. Au1Ag146 undergoes an MT similar to Ag147, with little difference to Tt and Tm, and the Au atom does not diffuse from its original position until close to the melting point. Similarly, a single Ag atom in Au147 (Au146Ag1) has little effect on Tt or Tm, (Table II) and the Ag atom is immobile until Tm is reached.


13


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 38

Table II. Summary of the transition and melting temperatures for pure and single dopant structures, where (CO) = cuboctahedral, (Ih) = icosahedral, (Dh) = decahedral, and (HE) = high energy. Tm[Ih] is averaged from 10 ensembles and refers to the Ih melting point, although the starting geometry is also provided in square parentheses for clarity, given that not all clusters undergo an MT. Transition/melting ranges over the ensemble are provided in parentheses. For the nanoclusters that do not undergo an MT, the melting point of the CO, Tm[CO], is also available. Composition & Structure

Tt [CO to Ih] / K

Tm [Ih] / K [starting geometry]

Tm [CO] / K

Ag147 [CO]

226 (220-230)

688 (660-700) [CO]

Au1Ag146 [CO]

228 (200-230)

696 (690-700) [CO]

Au1Ag146 (HE) [CO]

221 (190-240)

691 (680-700) [CO]

Ag147 [Dh]

289 (240-310)

693 (680-710) [Dh]

Au1Ag146 [Dh]

294 (270-310)

693 (680-710) [Dh]

Au1Ag146 (HE) [Dh]

274 (240-300)

688 (680-710) [Dh]

Au147 [CO]

-

555 (540-570) [Ih*]

555 (540-560)

Au146Ag1 [CO]

-

-

551 (540-560)

Au146Ag1 (HE) [CO]

-

-

554 (540-570)

Au147 [Dh]

285 (270-300)

550 (540-560) [Dh]

Au146Ag1 [Dh]

286 (250-310)

555 (540-570) [Dh]

Au146Ag1 (HE) [Dh]

256 (210-280)

553 (540-570) [Dh]

*

As this cluster does not undergo an MT, a stoichiometrically equivalent Ih is used as

the starting arrangement to obtain Tm(Ih). The influence of single dopant atoms was also extended to the Dh geometry. Pure Ag147 (Dh) and Au147 (Dh) nanoclusters undergo MTs at Tt = 289 K and 285 K, respectively, which suggests that the kinetic stability of Ag147 is CO < Dh < Ih. However, the kinetics for Au147 are more complicated due to the lack of structural transitions, e.g. Dh < CO ≈ Ih. In general, the introduction of dopants again has little effect on the properties of the Dh nanoclusters, with the exception of Au146Ag1 (Dh): when the dopant atom is in a high energy configuration, Tt is lowered by 30 K


14

Page 15 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


suggesting that the barriers to transformation are dependent on the position of a dopant in this instance. Of the pure and singly doped systems, only Au147 (CO) and Au146Ag1 (CO) do not have a structural transition to the Ih. Previous work by Garzon et al.56-57 has shown that small Au clusters (x = 38, 75) with a high degree of deformation are just as stable as highly ordered FCC morphologies, and thus these high and low symmetry structures are in close competition energetically. There is no preferred structural motif for the CO to adopt during the annealing process and many local basins are visited instead before melting, as shown in Fig. 3. The Ih window is practically absent for Au clusters, as established by Baletto et al.,58-59 with the Ih to Dh transition occurring at approximately 100 atoms, and FCC/ Dh arrangements strongly competing when x ≃ 400. Given that our Au-rich Dh clusters do transform to the Ih, and similar thermal evolutions have been observed by Landman et al.60 for Au clusters that begin as Marks-Decahedra (m-Dh), this observation highlights the need to explore further kinetic influences in this area during future work. As the Ih is the most stable structure prior to melting for the majority of systems investigated, we have repeated the MD simulations for all systems starting from this arrangement, thus allowing direct comparison of Tm in the following work for equivalent morphologies (Table II). As noted already, the melting points of Ag147 (Ih) and Au147 (Ih) are 688 and 555 K, respectively. In experiment, the Au and Ag surface energies are 1.5 and 1.25 Jm-2, respectively,61-62 and at first glance the observed trends in Tm match these observations i.e. Au has a greater surface energy and so is less stable than Ag, therefore melting at a lower T. However, and rather interestingly, the overall trends in Tm are inconsistent with the actual calculated surface energies for our chosen Cleri-Rosato parameterisation. We performed calculations with the GULP


15


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 38

software package63 where we found that the surface energy for Au (111), (100) and (110) planes is 0.491, 0.578 and 0.614 J/m2 respectively; for Ag the same surfaces planes have formation energies of 0.626, 0.704 and 0.764 J/m2, respectively. Alternative potential parameterisations, such as those of Mottet et al.59 yielded a similar hierarchy of surface energies. These lower surface energies for Au, accompanied by the greater cohesive energy, mean that it should have a greater thermal stability than Ag in our simulations; so the question then becomes why are the Au nanoparticles so unstable? We tentatively suggest that this is an artefact of the specific geometric configurations, and relates to an inherent strain in the Au geometries that is not present in the Ag systems – i.e. the competition between FCC and HCP structure types, as demonstrated in Fig. 3. Further investigation will be pursued in future work. 3.2 Alloy arrangements for AuxAg147−x (x = 13, 55, 92, 134) Low-energy mixed alloy AuxAg147−x nanoclusters, with x = 13, 55, 92 and 134, were subjected to a calcination simulation. As highlighted in our previous work, the Gupta potentials favour Au in a subsurface position and in an arrangement that maximises the number of Au-Ag bonds.49 As x is increased, and the cluster becomes Au-rich, the majority of the additional Ag atoms are positioned at the surface. The atomic mismatch between Au and Ag is small;64 however, with the tendency for Au147 to undergo deformation and Ag147 to perform MTs, it is reasonable to think that the route taken during the heat treatment process will depend on the stoichiometric ratio of the two elements. As it is difficult to determine these structural changes from aCNA alone, any local basins visited pre-melting have been visualized. Indeed, we find that only some nanoalloys stoichiometries undergo MTs, and thus direct comparison


16

Page 17 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


of Tm is made only for the Ih geometry, and this topic is discussed separately below for clarity. 3.2.1 Martensitic transformations (MTs) Both Au13Ag134 and Au55Ag92 perform MTs from the CO to the Ih. Au13Ag134 (x = 13) has a higher Tt than Ag147 (Tt = 254 and 226 K, respectively); the resulting Ih has a pseudo-onion chemical arrangement with an alloy surface: Ag@AuAg. For x = 55, Tt increases to 331 K (Table III), and for x ≥ 92, no clear-cut structural transformation is observed, but rather a multi-facetted face-centered cubic (FCC) arrangement forms (Fig. 5). As the MT is a diffusion-less rearrangement, the RCDF remains unchanged, and can only be observed through (Fig. 6). As Tt increases with Au content up to x = 55, it is reasonable to interpret that the energetic barrier to transform to an Ih geometry increases with x, and hence only Ag-rich nanoclusters exhibit an MT (Table III).

Fig. 5. Snapshot images of the alloy Au92Ag55 (CO) at (a) 350 K and (b) 450 K, illustrating partial transformation of the nanoalloy structure. Grey spheres represent Ag atoms, while yellow spheres represent Au atoms.


17


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 38

3.2.2. Melting of the Ih The Ih structure is used for direct comparison of Tm since neither Au92Ag55 nor Au134Ag13 transform from CO to Ih. Thus, Au92Ag55 (Ih) and and Au134Ag13 (Ih) are also used as initial starting arrangements, and subjected to the same heat treatment. The melting points of the alloy nanoclusters are given in Table III. These icosahedral alloy configurations are obtained using the methods in Section 2 and their structures are presented in Fig. S6 of the SI. Figures 6 and 7 represent the changes that occur in the alloy nanocluster as a function of T for and the RCDF, respectively. The RCDF is monitored in 0.3 Å increments, which can result in regions of ‘white space’ where no atoms are present within a particular radial segment (Fig. 1). However, as the nanocluster is heated, the extent of structural disorder increases, and the number of ‘white space’ regions decreases. Au13Ag134 begins in an onion-like arrangement, with Au atoms located at subsurface positions at 0 K,49 which can be observed in Fig. 7. (a), where green markers represent regions that are Au-rich and blue markers represent segments that are Au-poor. The RCDF only begins to change when close to Tm (Tm = 724 K) as shown by an increase in the number of red segments, which is indicative of an Au/Ag ratio ≈ 0.4 - 0.6 i.e. a greater degree of mixing between the two metals. Au55Ag92 displays similar behaviour, although there is a greater degree of initial mixing in the subsurface layers than found in Au13Ag134 due to the starting chemical arrangement. Atoms remain stationary until close to Tm, at 629 K (Fig. 7). Au92Ag55 (Ih) and Au134Ag13 (Ih) also do not exhibit any significant RCDF changes until close to their melting points of 624 and 559 K, respectively.


18

Page 19 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 6. as a function of temperature for alloy nanoclusters; a key is provided for the differing chemical arrangements. Tt and Tm are highlighted by upward and downward arrows, respectively. Inset, left to right: Nanoclusters at 0 K, Tt (where applicable), and Tm. Inset, top to bottom: Au13Ag134, Au55Ag92, Au92Ag55, Au134Ag13.


19


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 38

Fig. 7. RCDF (for a one simulation) for AuxAg147−x alloy nanoclusters: (a) Au13Ag134, (b) Au55Ag92, (c) Au92Ag55 [Ih] and (d) Au134Ag13[Ih]. The colour scale represents the Au ratio, where 1.0 (green) represents an Au-rich radial segment, and 0.0 (blue) represents an Au-poor radial segment. (a) and (b) have undergone an MT from CO to Ih earlier in the heating process. Au13Ag134 has a particularly high Tm of 724 K. However, as x is further increased beyond 13 Au atoms, there is a decrease in Tm: Tm(Au55Ag92) = 629 K, Tm(Au92Ag55) = 624 K, Tm(Au134Ag13) = 559 K, Tm(Au147) = 555 K. Phase transitions are clear-cut for Ag-rich alloys, such as for Au55Ag92 in Fig. 6. When x ≥ 92, i.e. Au-rich, the nanoclusters become a facetted FCC-fragment during the heating, and Tm is more difficult to distinguish due to smoother melting transitions (Fig. 6).


20

Page 21 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table III. The transition temperature (Tt) from CO to Ih geometries and averaged melting temperature (Tm) of AuxAg147−x alloy nanoclusters. Tm[Ih] is averaged from 10 ensembles and refers to the Ih melting point, although the starting geometry is also provided in square parentheses for clarity, given that not all clusters undergo an MT. Transition/melting ranges over the ensemble are provided in parentheses. For the nanoclusters that do not undergo an MT, the melting point of the CO, Tm[CO], is also available. Composition

Tt [CO to Ih] / K


Tm [CO] / K

Ag147

226 (220-230)

688 (660-700) [CO]

Au13Ag134

254 (220-270)

724 (710-730) [CO]

Au55Ag92

331(300-340)

629 (610-640) [CO]

Au92Ag55

-

624 (610-640) [Ih*]

552 (540-570)

Au134Ag13

-

559 (540-580) [Ih*]

553 (540-570)

Au147

-

555 (540-570) [Ih*]

555 (540-560)

*

As these Au-rich nanoalloys do not undergo an MT, a stoichiometrically equivalent Ih is used as the starting arrangement to obtain Tm(Ih). Overall, two melting-types are observed: Ag-rich nanoclusters exhibit a sharp transition to the melted amorphous structure, whereas Au-rich nanoclusters exhibit smoother transitions, developing defects during the annealing process. These observations support the work of Chen et al., who also observed glass-like transitions to dominate the thermal stability of AuxAg55−x nanoclusters when x ≥ 30 i.e. Au-rich.29 The anomalously higher Tm for Au13Ag134 (724 K) in comparison to Ag147 (688 K) is perhaps due to improved stability from accommodating a few Au atoms within the Ag particle, though this would not be intuitive given the similar atomic radius of Ag and Au. We also note that Tm(Au55Ag92) and Tm(Au92Ag55) are very similar, despite the different Au:Ag stoichiometries. This result indicates that the chemical arrangement may influence Tm; Au55Ag92 still has Ag-rich regions at the surface and in the very core at its melting point, whereas Au92Ag55 (Ih) appears to be a completely random alloy (Fig. 7). Whilst one could expect the melting point of Au55Ag92 to be higher as a


21


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 38

result of the Ag core, both nanoclusters have similar AuAg shells and so it is more likely that in fact the chemical arrangement in the shell of the nanocluster dominates the melting properties. Au134Ag13 (Ih) has the highest Au content of the alloys and exhibits the lowest Tm (559 K), which is very close to the melting point of Au147 (Tm = 555 K). 3.3 Core@shell arrangements for (Au@ Ag)147 and (Ag@ Au)147 Core@shell arrangements are thought to have considerable potential for catalytic applications, as the ability to dictate the chemical arrangement in experiment can provide tailor-made catalysts. None of the aforementioned nanoalloys spontaneously form a core@shell arrangement upon heating in our calculations, and so we have repeated the simulated calcinations for core@shell arrangements with identical metallic ratios, namely: Au13@Ag134, Au55@Ag92, Ag55@Au92, and Ag13@Au134. 3.3.1 Au@Ag - Martensitic transformation (MTs) and melting Both Au-core nanoclusters, Au13@Ag134 and Au55@Ag92, exhibit MTs from CO to Ih at 196 and 221 K, respectively, as shown in Fig. 8. The RCDFs of Au@Ag systems (Fig. 9) show that the nanoclusters remain chemically segregated with no alloying (i.e. no red segments), and the first indication of atomic diffusion occurs ~40 K below Tm. After melting, it is noticeable for Au13@Ag134 that the Au atoms remain predominantly in the sub-surface layer, forming a pseudo-onion arrangement. Au13@Ag134 has a marginally lower Tm than Ag147 (682 and 688 K, respectively) and when the Au core is increased to 55 atoms we find that Tm also increases to 704 K, i.e. the nanocluster is stabilised.


22

Page 23 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 8. as a function of temperature for core@shell nanoclusters; a key is provided for the differing chemical arrangements. Tt and Tm are highlighted by upward and downward arrows, respectively. Inset, left to right: Nanoclusters at 0 K, Tt (where applicable; the pseudo-spherical arrangement is shown for Ag55@Au92), and Tm. Inset, top to bottom: Au13@Ag134, Au55@Ag92, Ag55@Au92, Ag13@Au134; core atoms have been enlarged for clarity.

Fig. 9. RCDFs (for one simulation) for (a) Au13@Ag134 and (b) Au55@Ag92; both systems undergo an MT at 196 and 221 K, respectively. The colour scale represents the Au ratio, where 1.0 (green) represents an Au-rich radial segment, and 0.0 (blue) represents an Au-poor radial segment.


23


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 38

3.3.2 Ag@Au - Martensitic transformation (MTs) and melting At 192 K, Ag55@Au92 exhibits a structural transition but rather than undergoing an MT to the Ih, it forms a pseudo-spherical geometry. This deformation of the Au shell occurs as a result of a surface step dislocation, similar to that observed for Au147 (but less structured), and is triggered by the low temperature outward migration of Ag atoms to the surface. This migration occurs over a temperature range of 140-180 K, which is more than 350 K below the melting point (551 K) and well below room temperature, implying spontaneous diffusion would occur in experiment. By definition, this is a non-MT due to the non-cooperative diffusion of the core Ag atoms; no other core@shell arrangements exhibit core-atom migration until ~20 K below Tm, implying that migration is associated with melting in all other cases. As Ag@Au structures do not transform to Ih, the calculations were repeated with Ag13@Au134 (Ih) and Ag55@Au92 (Ih), for comparison with other melting points. However, it was noted that Ag55@Au92 (Ih) also exhibits the novel non-MT, with Ag atoms diffusing to the nanocluster surface at even lower temperatures than for the CO morphology (T = 20-30 K); the resulting structure is again pseudo-spherical (Fig. 10). For Ag55@Au92, Tm is very similar to that of Au147 (Tm = 551 and 555 K, respectively), perhaps due to the lower crystallinity in the morphology after the nonMT; Ag13@Au134 (Ih), maintains a core@shell arrangement until close to the Tm of 548 K, which is very similar to Tm(Au147). After melting, we again notice that the Au atoms tend to migrate to the sub-surface area to form a pseudo-onion chemical arrangement; this is especially noticeable for Ag55@Au92 in Fig. 11.(a). Melting points of all analysed nanoclusters are given in Table IV.


24

Page 25 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 10. (a) Ag55@Au92 (CO) and (b) Ag55@Au92 (Ih) at: (i) 0 K, (ii) the temperature at which diffusion is first observed, as labelled, (iii) 300 K, and (iv) a cross section of the nanoclusters at 300 K. Colours are the same as in Fig. 5.

Fig. 11. RCDF (for one simulation) for Ag@Au nanoclusters (a) Ag55@Au92 [Ih] and (b) Ag13@Au134 [Ih]. The colour scale represents the Au ratio, where 1.0 (green) represents an Au-rich radial segment, and 0.0 (blue) represents an Au-poor radial segment.


25


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 38

Table IV. The transition temperature (Tt) from CO to Ih geometries and melting temperature (Tm) for AuxAg147−x core@shell nanoclusters. Tm[Ih] is averaged from 10 ensembles and refers to the Ih melting point, although the starting geometry is also provided in square parentheses for clarity, given that not all clusters undergo an MT. Transition/melting ranges over the ensemble are provided in parentheses. For the nanoclusters that do not undergo an MT, the melting point of the CO, Tm[CO], is also available. Structure & composition

Tt [CO to Ih] / K


Ag147

226 (220-230)

688 (660-700) [CO]

Au13@Ag134

196 (180-210)

682 (670-690) [CO]

Au55@Ag92

221 (200-240)

704 (700-710) [CO]

Ag55@Au92

†192 (170-230)

†551 (530-570) [Ih*]

Ag13@Au134 Au147 *

-

†543 (530-570)

*

556 (540-570)

*

555 (540-560)

548 (540-570) [Ih ]

-

Tm [CO] / K

555 (540-570) [Ih ]

As these Au-rich nanoalloys do not undergo an MT, a stoichiometrically equivalent

Ih is used as the starting arrangement to obtain Tm(Ih). †Ag55@Au92 transforms to a spherical arrangement regardless of the initial geometry. 3.3.3 Detailed analysis of Ag55@Au92 It is particularly interesting that only Ag55@Au92 undergoes non-MTs to pseudo-spherical

arrangements,

irrespective

of

starting

structure,

whereas

Ag13@Au134 and Au147 do not transform despite all possessing a gold shell. The outward migration of Ag for Ag55@Au92 can be perhaps rationalized because Ag55@Au92 has a thinner Au shell than Ag13@Au134, and so fewer Au−Au bonds need to be broken during the diffusion process, meaning that Ag migrates at low temperature to the surface, indicating that the shell thickness, as well as the shell composition, plays a significant role in determining the adopted geometry upon annealing. As a consequence of the migration of Ag atoms, any pre-defined starting geometry and core@shell chemical arrangements are compromised very early in the simulation: the Ag migration stimulates the formation of facets in the Au shell and the structure re-stabilises in a pseudo-spherical arrangement once Ag atoms reach the surface.


26

Page 27 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 12. as a function of temperature for (a) cuboctahedral Ag55@Au92 and alloy Au92Ag55; (b) icosahedral Ag55@Au92 and alloy Au92Ag55. A key is provided. is represented as solid lines and, for the nanoclusters that begin with a CO geometry, is also given, as dashed coloured lines. Tt and Tm are given as dash-dotted and dotted lines, respectively. is compared for Ag55@Au92 with that of its stoichiometric equivalent alloy, Au92Ag55 for both CO [Fig. 12.(a)] and Ih [Fig. 12(b)] geometries, at increasing temperatures. For the nanoclusters that begin in a CO arrangement, is also provided. As shown previously in Fig. 3, a perfect CO geometry consists of = 37.4%, = 0%, and = 62.6%, whereas a perfect Ih geometry does not contain any FCC features ( = 20.4% and = 79.6%). Let us first


27


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 38

consider Ag55@Au92 (CO): When T = 150-190 K, the cluster transforms to a pseudospherical arrangement as a result of the Ag migration to the surface, reflected in a corresponding increase in (and decline in ). Following this, is maintained at a fairly constant level, between 4% and 10%, until the cluster melts (and = 0%). is similar to that of Au147 just below its Tm (Fig. 3), and thus we suggest that the migration of core Ag atoms to the surface illustrates a larger degree of structural fluctuation for this nanocluster at low temperatures; the same is also true for Ag55@Au92 (Ih), with the non-MT occurring at ~30 K. We also note that during the non-MT for the CO arrangement both FCC and HCP features are present. Ag55@Au92 (Ih) does not contain any FCC structural signatures initially ( = 0%) and migration of Ag atoms seems to occur much more readily. Similarly, if we compare the Ag55@Au92 (CO) and Au92Ag55 (CO), the core@shell arrangement loses its FCC elements much more readily than the alloy. The drive to form a solid solution equilibrium phase is strong for Ag@Au arrangements, and has been shown in the work of Calvo et al..44 Due to the already mixed nature of the alloy systems (both CO and Ih), there is no similar driving force towards the solid solution as in the core@shell systems (Fig. S7). For the RCDFs of these systems (Fig. 13), Ag atoms that originate in the core can be observed to migrate to the surface, with full migration first identifiable at 400 K in Fig. 13. (a). This process is more difficult within the CO and thus is not observed until the melting point has been reached. The energetic drive of Ag@Au arrangements to attain alloy arrangements is further highlighted in Fig. 14. Au92Ag55 (Ih) has a higher melting point than the other three arrangements (Tm = 624 K), and we suggest that this difference is due to the CO comprising of the less stable (100) faces as well as (111) faces: the surface energy of


28

Page 29 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Au (111) and Au (100) are 0.491 and 0.614 Jm-2, respectively, for the Cleri-Rosato parameters. As Au92Ag55 (Ih) comprises only (111) faces, it melts at a higher temperature – snapshot images for Au92Ag55 (CO, Ih) at 550 K are given in the SI, Fig. S7. Finally, we note that diffusion of the Ag atoms from the core to the surface occurs for Ag13@Au134, as well as Ag55@Au92, but closer to Tm in the former case; atomic diffusion of Ag to the surface is slower due to the increased thickness of the Au shell. Therefore, for Ag13@Au134, Tm is related to the Au shell, and not the migrating Ag atoms, as the difference in Tm compared to Au147 is < 10 K.

Fig. 13. RCDF as a function of temperature for one simulation of the following: core@shell (a) Ag55@Au92 [Ih], (b) Ag55@Au92 [CO], and alloy (c) Au92Ag55 [Ih] and (d) Au92Ag55 [CO)]. The colour scale represents the Au ratio, where 1.0 (green) represents an Au-rich radial segment, and 0.0 (blue) represents an Au-poor radial segment.


29


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 38

Fig. 14. Energetic comparison of Au92Ag55 and Ag55@Au92; a key is provided. Tt and Tm are highlighted by upward and downward arrows, respectively. 3.4 Discussion Overall, it is observed that the shell component dictates Tm, with Au@Ag nanoclusters exhibiting melting points that are similar to Ag147, and Ag@Au close to that of Au147. The calculated Tm is very similar for Ag55@Au92, Ag13@Au134 and Au147, despite the spherical arrangement for Ag55@Au92. Furthermore, given the difference in Tm of the pure 147-atom nanoclusters, where Au147 is less stable than Ag147, one would expect Tm to decrease with increasing Au content; however, Au55@Ag92 has a higher Tm than Au13@Ag134. This difference might be attributed to the stability of the core atoms, where previous literature has shown that core@shell arrangements can actually improve stability in comparison to pure nanoclusters, depending on the core atom geometry.65-66 Global optimization methods have shown Au55 to be particularly stable as a heavily distorted and rearranged icosahedron,58 and Au nanoclusters also have higher binding energies than corresponding Ag nanoclusters of the same size,59 and so the core Au atoms in Au55@Ag92 may allow the structure to maintain an Au@Ag arrangement to higher temperatures than other


30

Page 31 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


systems. The stability of the Au55 core may also explain why Au55@Ag92 has a higher Tm than the corresponding alloy arrangement, Au55Ag92; alloy arrangements are more stable for all the other compositions investigated. Our extended analysis of the core@shell structures, and the kinetic diffusion of core Ag atoms to the surface, may help to explain why the formation of Ag@Au nanoclusters is more difficult than Au@Ag: in addition to the aforementioned reports by Jiang et al.14 and Mallin and Murphy,15 the colloidal nanoclusters created by Srnová-Šloufová et al.67 also had an Ag@AuAg chemical arrangement, rather than the intended Ag@Au. Sustained annealing (600 K for 58 ns) of Ih core@shell nanoclusters by Qi and Lee32 resulted in the migration of core Ag atoms to the surface in Ag@Au nanoclusters, whereas no Au atoms were observed to migrate in Au@Ag; clearly the driving force behind alloy formation is greater for Ag@Au arrangements than Au@Ag. As Calvo et al.44 obtained a solid solution to be the most stable phase (below the melting temperature) from parallel tempering studies, this suggests that the alloy formation is highly dependent on the starting arrangement – our study of Au55@Ag92 exhibits a ‘super stability’, whereas diffusion occurs very readily for Ag55@Au92. Our observations also support the work of Chen et al.,17 who found spontaneous surface alloying for colloidal Ag@Au nanoclusters with a thin Au shell (Ag:Au = 0.4:0.1 mol fraction), but not for when the Au content ≥ 0.2 mol fraction. The extended x-ray absorption fine structure spectroscopy (EXAFS) spectra oscillation of Ag:Au = 0.4:0.1 (Au LIII edge) was similar to that obtained for alloy samples, indicating the presence of first nearest-neighbour Ag atoms rather than a uniform Au environment, as obtained for the other core@shell arrangements. Chen et al. attribute this spontaneous alloying to an increasing diffusion coefficient with


31


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 38

decreasing nanocluster size, and ease of alloy formation due to the similar atomic sizes of Au and Ag. However, as neither of these are contending factors in our work, we suggest an alternative explanation: namely that the stability of the Ag@Au arrangement is highly dependent on the Au shell thickness. Straightforward analysis from ultraviolet-visible (UV-Vis) spectroscopy may not be possible in the outlined cases, because the elements in the shell of the nanocluster tend to dominate incident light interactions, and electron plasmon oscillations will be dampened by any interactions between the two metallic constituent species.68 Thus, if the shell thickness determines the rate of core diffusion in Ag@Au nanoclusters, as seen in our work, this adds an experimental requirement to the synthesis process, which must be deliberately controlled; with increasing shell thickness, however, we also generate larger particles, and the high reactivity that is associated with small cluster sizes becomes vulnerable. 4. Summary and Conclusions The kinetic stability of 147-atom cuboctahedron Au−Ag bimetallic nanoclusters has been investigated using molecular dynamics simulations, with melting temperatures identified through energy transitions and adaptive common neighbor analysis. During the simulated heating treatment, Au-rich nanoalloys were observed to become multi-facetted FCC fragments, whereas Ag-rich nanoalloys underwent a martensitic transformation to the icosahedron morphology before melting. The icosahedron melting point is dependent on the Au:Ag ratio; small concentrations of Au dopants cause an initial increase in Tm, but when Au content was further increased beyond 50% we observed that Tm decreased in accordance. For core@shell arrangements, the shell component is the dominating factor in


32

Page 33 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


determining Tm; however, the kinetic stability may also be further increased depending on the stability of the core, as demonstrated by Au55@Ag92. We have also identified a novel non-martensitic transformation to a pseudo-spherical geometry for Ag55@Au92, which has not been documented thus far. The dependency we have shown of Tm on chemical arrangement may also help to explain why there are few successful experimental reports of Ag@Au synthesis, as their stability is highly dependent on shell thickness, and they are more susceptible to interatomic diffusion at ambient temperatures. We suggest thicker Au shells will provide greater long-term stability, but caution must be exercised to maintain the high reactivity that is associated with small nanoclusters. Acknowledgements We are grateful to A. J. O’Malley for his input and advice, and acknowledge financial support from EPSRC (EP/G036675/1), (EP/I030662/1) and (EP/K038419/1) and Diamond Light Source, (as well as computational resources from Research Complex at Harwell and the Catalysis Hub). The authors acknowledge the UK's HEC Materials Chemistry Consortium (EP/L000202), which provided access to the ARCHER UK National Supercomputing Service, as well as the UCL Legion High Performance Computing Facility (Legion@UCL), and associated support services, which were used in the completion of this work. AJL is grateful to the Ramsay Memorial Trust and University College London for the provision of a Ramsay Fellowship. Supporting Information Figures S1-S4: Total internal energy as a function of temperature for varying nanocluster arrangements; Fig. S5: Initial starting positions for single dopant atoms;


33


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 38

Fig. S6: Starting arrangements for alloy nanoclusters at 0 K, comparing the CO and Ih; Fig. S7: Au92Ag55 (CO and Ih) snapshot at 550 K.

References 1. Ferrando, R.; Jellinek, J.; Johnston, R. L., Nanoalloys: From Theory to Applications of Alloy Clusters and Nanoparticles. Chem. Rev. 2008, 108, 845-910. 2. Johnston, R. L., Atomic and Molecular Clusters; CRC Press, 2002. 3. Johnston, R. L.; Wilcoxon, J. P., Metal Nanoparticles and Nanoalloys; Elsevier, 2012; Vol. 3. 4. Hsieh, C.-T.; Lin, J.-Y., Fabrication of Bimetallic Pt–M (M= Fe, Co, and Ni) Nanoparticle/Carbon Nanotube Electrocatalysts for Direct Methanol Fuel Cells. J. Power Sources 2009, 188, 347-352. 5. Zhong, C.-J.; Luo, J.; Fang, B.; Wanjala, B. N.; Njoki, P. N.; Loukrakpam, R.; Yin, J., Nanostructured Catalysts in Fuel Cells. Nanotechnology 2010, 21, 062001. 6. Liu, J. P.; Fullerton, E.; Gutfleisch, O.; Sellmyer, D. J., Nanoscale Magnetic Materials and Applications; Springer, 2009. 7. Toshima, N.; Yonezawa, T., Bimetallic Nanoparticles—Novel Materials for Chemical and Physical Applications. New J. Chem. 1998, 22, 1179-1201. 8. Liu, X.; Wang, A.; Yang, X.; Zhang, T.; Mou, C.-Y.; Su, D.-S.; Li, J., Synthesis of Thermally Stable and Highly Active Bimetallic Au−Ag Nanoparticles on Inert Supports. Chem. Mater. 2008, 21, 410-418. 9. Déronzier, T.; Morfin, F.; Lomello, M.; Rousset, J.-L., Catalysis on Nanoporous Gold–Silver Systems: Synergistic Effects toward Oxidation Reactions and Influence of the Surface Composition. J. Catal. 2014, 311, 221-229. 10. Wilcoxon, J., Optical Absorption Properties of Dispersed Gold and Silver Alloy Nanoparticles†. J. Phys. Chem. B 2008, 113, 2647-2656. 11. Barcaro, G.; Broyer, M.; Durante, N.; Fortunelli, A.; Stener, M., Alloying Effects on the Optical Properties of Ag–Au Nanoclusters from Tddft Calculations. J. Phys. Chem. C 2011, 115, 24085-24091. 12. Logsdail, A. J.; Johnston, R. L., Predicting the Optical Properties of Core– Shell and Janus Segregated Au–M Nanoparticles (M= Ag, Pd). J. Phys. Chem. C 2012, 116, 23616-23628. 13. López Lozano, X. c.; Mottet, C.; Weissker, H.-C., Effect of Alloying on the Optical Properties of Ag–Au Nanoparticles. J. Phys. Chem. C 2013, 117, 3062-3068. 14. Jiang, H.-L.; Akita, T.; Ishida, T.; Haruta, M.; Xu, Q., Synergistic Catalysis of Au@Ag Core−Shell Nanoparticles Stabilized on Metal− Organic Framework. J. Am. Chem. Soc. 2011, 133, 1304-1306. 15. Mallin, M. P.; Murphy, C. J., Solution-Phase Synthesis of Sub-10 Nm Au-Ag Alloy Nanoparticles. Nano Lett. 2002, 2, 1235-1237. 16. Shibata, T.; Bunker, B. A.; Zhang, Z.; Meisel, D.; Vardeman, C. F.; Gezelter, J. D., Size-Dependent Spontaneous Alloying of Au-Ag Nanoparticles. J. Am. Chem. Soc. 2002, 124, 11989-11996. 17. Chen, H.; Liu, R.; Jang, L.-Y.; Lee, J.-F.; Hu, S., Characterization of Core– Shell Type and Alloy Ag/Au Bimetallic Clusters by Using Extended X-Ray Absorption Fine Structure Spectroscopy. Chem. Phys. Lett. 2006, 421, 118-123.


34

Page 35 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


18. Zhang, J.; Langille, M. R.; Personick, M. L.; Zhang, K.; Li, S.; Mirkin, C. A., Concave Cubic Gold Nanocrystals with High-Index Facets. J. Am. Chem. Soc. 2010, 132, 14012-14014. 19. Barnard, A. S., Direct Comparison of Kinetic and Thermodynamic Influences on Gold Nanomorphology. Acc. Chem. Res. 2012, 45, 1688-1697. 20. Rycenga, M.; Cobley, C. M.; Zeng, J.; Li, W.; Moran, C. H.; Zhang, Q.; Qin, D.; Xia, Y., Controlling the Synthesis and Assembly of Silver Nanostructures for Plasmonic Applications. Chem. Rev. 2011, 111, 3669-3712. 21. Pietrobon, B.; Kitaev, V., Photochemical Synthesis of Monodisperse SizeControlled Silver Decahedral Nanoparticles and Their Remarkable Optical Properties. Chem. Mater. 2008, 20, 5186-5190. 22. Mottet, C.; Rossi, G.; Baletto, F.; Ferrando, R., Single Impurity Effect on the Melting of Nanoclusters. Phys. Rev. Lett. 2005, 95, 035501. 23. Cheng, D.; Huang, S.; Wang, W., Thermal Behavior of Core-Shell and ThreeShell Layered Clusters: Melting of Cu1au54 and Cu12au43. Phys. Rev. B 2006, 74, 064117. 24. Paz-Borbón, L. O.; Mortimer-Jones, T. V.; Johnston, R. L.; Posada-Amarillas, A.; Barcaro, G.; Fortunelli, A., Structures and Energetics of 98 Atom Pd–Pt Nanoalloys: Potential Stability of the Leary Tetrahedron for Bimetallic Nanoparticles. Phys. Chem. Chem. Phys. 2007, 9, 5202-5208. 25. Ismail, R.; Johnston, R. L., Investigation of the Structures and Chemical Ordering of Small Pd–Au Clusters as a Function of Composition and Potential Parameterisation. Phys. Chem. Chem. Phys. 2010, 12, 8607-8619. 26. Nunez, S.; Johnston, R., Structures and Chemical Ordering of Small Cu− Ag Clusters. J. Phys. Chem. C 2010, 114, 13255-13266. 27. Heiles, S.; Logsdail, A. J.; Schäfer, R.; Johnston, R. L., Dopant-Induced 2d– 3d Transition in Small Au-Containing Clusters: Dft-Global Optimisation of 8-Atom Au–Ag Nanoalloys. Nanoscale 2012, 4, 1109-1115. 28. Calvo, F.; Fortunelli, A.; Negreiros, F.; Wales, D. J., Communication: Kinetics of Chemical Ordering in Ag-Au and Ag-Ni Nanoalloys. J. Chem. Phys. 2013, 139, 111102. 29. Chen, F.; Curley, B. C.; Rossi, G.; Johnston, R. L., Structure, Melting, and Thermal Stability of 55 Atom Ag-Au Nanoalloys. J. Phys. Chem. C 2007, 111, 91579165. 30. Chen, F.; Johnston, R. L., Martensitic Transformations in Ag-Au Bimetallic Core-Shell Nanoalloys. Appl. Phys. Lett. 2008, 92, 023112. 31. Otsuka, K.; Kakeshita, T., Science and Technology of Shape-Memory Alloys: New Developments. MRS Bull. 2002, 27, 91-100. 32. Qi, W.; Lee, S., Phase Stability, Melting, and Alloy Formation of Au−Ag Bimetallic Nanoparticles. J. Phys. Chem. C 2010, 114, 9580-9587. 33. Todorov, I.; Smith, W.; Cheshire, U., The Dl Poly 4 User Manual. STFC, STFC Daresbury Laboratory, Daresbury, Warrington, Cheshire, WA4 4AD, United Kingdom, v. 4 2011. 34. Swope, W. C.; Andersen, 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. 35. Berendsen, H. J.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J., Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81, 3684-3690.


35


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 38

36. Hünenberger, P. H., Thermostat Algorithms for Molecular Dynamics Simulations. In Advanced Computer Simulation, Springer: 2005; pp 105-149. 37. Nosé, S., A Unified Formulation of the Constant Temperature Molecular Dynamics Methods. J. Chem. Phys. 1984, 81, 511-519. 38. Nosé, S., A Molecular Dynamics Method for Simulations in the Canonical Ensemble. Mol. Phys. 1984, 52, 255-268. 39. Kuntová, Z.; Rossi, G.; Ferrando, R., Melting of Core-Shell Ag-Ni and Ag-Co Nanoclusters Studied Via Molecular Dynamics Simulations. Phys. Rev. B 2008, 77, 205431. 40. Mottet, C.; Goniakowski, J.; Baletto, F.; Ferrando, R.; Treglia, G., Modeling Free and Supported Metallic Nanoclusters: Structure and Dynamics. Phase Transit. 2004, 77, 101-113. 41. Gupta, R. P., Lattice Relaxation at a Metal Surface. Phys. Rev. B 1981, 23, 6265-6270. 42. Sutton, A. P.; Finnis, M. W.; Pettifor, D. G.; Ohta, Y., The Tight-Binding Bond Model. J. Phys. C 1988, 21, 35-66. 43. Tomanek, D.; Aligia, A.; Balseiro, C., Calculation of Elastic Strain and Electronic Effects on Surface Segregation. Phys. Rev. B 1985, 32, 5051-5056. 44. Calvo, F.; Cottancin, E.; Broyer, M., Segregation, Core Alloying, and Shape Transitions in Bimetallic Nanoclusters: Monte Carlo Simulations. Phys. Rev. B 2008, 77, 121406. 45. Cleri, F.; Rosato, V., Tight-Binding Potentials for Transition Metals and Alloys. Phys. Rev. B 1993, 48, 22-33. 46. Lobato, I.; Rojas, J.; Landauro, C. V.; Torres, J., Atomic and Electronic Structure Transformations of Silver Nanoparticles under Rapid Cooling Conditions. J. Phys. Condens. Matter 2009, 21, 055301. 47. Sauceda, H. E.; Salazar, F.; Pérez, L. A.; Garzón, I. L., Size and Shape Dependence of the Vibrational Spectrum and Low-Temperature Specific Heat of Au Nanoparticles. J. Phys. Chem. C 2013, 117, 25160-25168. 48. Qi, W. H.; Huang, B. Y.; Wang, M. P.; Liu, F. X.; Yin, Z. M., Freezing of Silver Cluster and Nanowire: A Comparison Study by Molecular Dynamics Simulation. Comp. Mater. Sci. 2008, 42, 517-524. 49. Gould, A. L.; Heard, C. J.; Logsdail, A. J.; Catlow, C. R. A., Segregation Effects on the Properties of (Auag)147. Phys. Chem. Chem. Phys. 2014, 16, 2104921061. 50. Cerbelaud, M.; Ferrando, R.; Barcaro, G.; Fortunelli, A., Optimization of Chemical Ordering in Agau Nanoalloys. Phys. Chem. Chem. Phys. 2011, 13, 1023210240. 51. Ferrando, R., Private Communication. 52. Pavan, L.; Baletto, F.; Novakovic, R. Multiscale Approach for Studying Melting Transitions in Cupt Nanoparticles Phys. Chem. Chem. Phys. 2015, Advance Article, DOI: 10.1039/c5cp01096a. 53. Stukowski, A., Visualization and Analysis of Atomistic Simulation Data with Ovito–the Open Visualization Tool. Modell. Simul. Mater. Sci. Eng. 2010, 18, 015012. 54. Honeycutt, J. D.; Andersen, H. C., Molecular Dynamics Study of Melting and Freezing of Small Lennard-Jones Clusters. J. Phys. Chem. 1987, 91, 4950-4963. 55. Stukowski, A., Structure Identification Methods for Atomistic Simulations of Crystalline Materials. Modell. Simul. Mater. Sci. Eng. 2012, 20, 045021.


36

Page 37 of 38

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


56. Garzón, I.; Michaelian, K.; Beltrán, M.; Posada-Amarillas, A.; Ordejón, P.; Artacho, E.; Sánchez-Portal, D.; Soler, J., Lowest Energy Structures of Gold Nanoclusters. Phys. Rev. Lett. 1998, 81, 1600-1603. 57. Michaelian, K.; Rendon, N.; Garzón, I., Structure and Energetics of Ni, Ag, and Au Nanoclusters. Phys. Rev. B 1999, 60, 2000-2010. 58. Baletto, F.; Ferrando, R., Structural Properties of Nanoclusters: Energetic, Thermodynamic, and Kinetic Effects. Rev. Mod. Phys. 2005, 77, 371-423. 59. Baletto, F.; Ferrando, R.; Fortunelli, A.; Montalenti, F.; Mottet, C., Crossover among Structural Motifs in Transition and Noble-Metal Clusters. J. Chem. Phys. 2002, 116, 3856-3863. 60. Cleveland, C. L.; Luedtke, W. D.; Landman, U., Melting of Gold Clusters. Phys. Rev. B 1999, 60, 5065-5077. 61. De Boer, F. R.; Mattens, W.; Boom, R.; Miedema, A.; Niessen, A., Cohesion in Metals. 1988. 62. Vitos, L.; Ruban, A. V.; Skriver, H. L.; Kollár, J., The Surface Energy of Metals. Surf. Sci. 1998, 411, 186-202. 63. Gale, J. D.; Rohl, A. L., The General Utility Lattice Program (Gulp). Molecular Simulation 2003, 29, 291-341. 64. Yaws, C. L., Chemical Properties Handbook. McGraw-Hill, New York: 1999. 65. Logsdail, A. J.; Li, Z.; Johnston, R. L., Faceting Preferences for Aun and Pdn Nanoclusters with High-Symmetry Motifs. Phys. Chem. Chem. Phys. 2013, 15, 83928400. 66. Logsdail, A. J.; Johnston, R. L., Interdependence of Structure and Chemical Order in High Symmetry (Pdau)N Nanoclusters. RSC Advances 2012, 2, 5863-5869. 67. Srnová-Šloufová, I.; Lednický, F.; Gemperle, A.; Gemperlová, J., Core-Shell (Ag)Au Bimetallic Nanoparticles: Analysis of Transmission Electron Microscopy Images. Langmuir 2000, 16, 9928-9935. 68. Mallik, K.; Mandal, M.; Pradhan, N.; Pal, T., Seed Mediated Formation of Bimetallic Nanoparticles by Uv Irradiation: A Photochemical Approach for the Preparation of “Core-Shell” Type Structures. Nano Lett. 2001, 1, 319-322.


37


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 38

TOC


38

Influence of Composition and Chemical Arrangement on the Kinetic

Recommend Documents