Project supported by the National Natural Science Foundation of China (Grant No. 21301112) and the Ph. D. Program Foundation of the Education Ministry of China (Grant No. 20131404120001).
Abstract
Abstract
The configurations, stabilities, electronic, and magnetic properties of FenAu (n = 1−12) clusters are investigated systematically by using the relativistic all-electron density functional theory with the generalized gradient approximation. The substitutional effects of Au in Fen+1 (n = 1, 2, 4, 5, 10−12) clusters are found in optimized structures which keep the similar frameworks with the most stable Fen+1 clusters. And the growth way for FenAu (n = 6−9) clusters is that the Au atom occupies a peripheral position of Fen cluster. The peaks appear respectively at n = 6 and 9 for FenAu clusters and at n = 5 and 10 for Fen+1 clusters based on the size dependence of second-order difference of energy, implying that these clusters possess relatively high stabilities. The analysis of atomic net charge Q indicates that the charge always transfers from Fe to Au atom which causes the Au atom to be nearly non-magnetic, and the doped Au atom has little effect on the average magnetic moment of Fe atoms in FenAu cluster. Finally, the total magnetic moment is reduced by 3 μB for each of FenAu clusters except n = 3, 11, and 12 compared with for corresponding pure Fen+1 clusters.
Nanoclusters have aroused considerable interest recently, mainly due to their novel electronic and magnetic properties which are different from those of the individual atoms and molecules or bulk matter. Of transition metal clusters, Fe clusters are the most widely studied 3d transition-metal clusters both experimentally[1–5] and theoretically.[6–14] The popularity of iron cluster is attributed to its distinctive reactivity, electronic, magnetic, and catalytic properties[15–21] as cluster size varies. Besides these properties, iron is well known for its high magnetization and an easy conversion to biocompatible oxide, which has received much attention in the biomedical areas.[22–24] At the same time, gold has been regarded as a special research material because of its unique features such as catalytic activity and bio-compatibility.[25–31] Therefore, much attention has been paid to studying the nanostructures comprised of iron and gold in both basic science research and promising application.
The bimetallic Fe/Au nanoparticles have become more popular, which is due mainly to the outstanding physical and chemical properties. Of particular interest is good bio-compatibility associated with biological applications. Naturally, based on this reason, a great number of researches have been done. Recently, Liu et al.[32] have reported that iron–gold (FeAu) alloy nanoparticles with the optical and magnetic properties present a promising prospect in biological applications. Sun et al.[33] studied that the gold-coated iron nanoclusters prevent the iron core from oxidizing, perhaps making a chance to functionalize drug delivery for the magnetic particles. Obviously, Wijaya et al.[34] performed magnetic heating experiments of Fe-doped Au nanoparticles and found that the nanoparticles exhibit notable magnetic behavior, presenting a great potential for use in hyperthermia application. Besides, experimental researches about Au–Fe alloys also have been performed in order to explore new nanomaterials in many functional material applications. The optical investigation of AuFe alloys are studied with the Fe concentration range by Gorshunov et al.[35] And they observed that the ferromagnetic AuFe clusters give rise to partial localization of conduction electrons. Via a combination of x-ray magnetic circular dichroism (XMCD) experiments and superconducting quantum interference device (SQUID) magnetometries, Wilhelm et al.[36] have demonstrated that the Fe in fcc disordered Au–Fe alloy exists in the form of the high-spin state and found the maximum value of magnetic Au moment in bcc Au–Fe alloys. Meanwhile, Crespo et al.[37] reported a crucial experiment, showing that Fe impurities weaken the ferromagnetic behavior in Au-contained nanoparticles. Moreover, theoretically, Sternik and Parlinski[38] calculated bilinear coupling constant for Fen/Aum multilayers and observed the increased magnetism of Fe atom, where there is a slight polarization of Au layers. To sum up, the extent to which size, shape, and composition of iron and gold are controlled will be a cruical factor of its flexible applications in magnetic field and biological area.
Based on the above analysis, a considerable number of theoretical researches about Fe–Au alloy clusters have been carried out. Up to now, these investigations mainly focused on the gold-rich Fe–Au alloy clusters aiming at obtaining the desired structural, magnetic, and chemical properties for potential applications. For example, Die et al.[39] have investigated single Fe doped gold clusters by using the density-functional theory and summarized the low-energy isomers tending to a two-dimensional structure. Interestingly, Zhang et al.[40] revealed doping 3d transition-metal Fe atom into gold cluster that is of planar hexagonal structure where it could stabilize the Au6 ring and promote the formation of new binary alloy clusters. The stabilities of gold clusters doped with open 3d-shell Fe atom have been investigated by Torres,[41] demonstrating that the electronic and magnetic structure depend on the dopant atom and the geometrical environment. Notably, Yang et al.[42] obtained the result that Fe@Au24 cluster still keeps the tubular Au24 structure with a slight distortion and retains the atom-like magnetism of 4 μB, which might be used as new nanomaterials with tunable magnetic properties in the near future. Later, Wang et al.[43] found that forms a new class of endohedral golden cage cluster which maintains its high spin. In addition, there has been existed apparently on a considerable amount of theoretical work which is focused on the 3d transition metal atoms doped with gold-rich clusters.[39–47] However, to the best of our knowledge, the studies of Au doped with TM-rich along with its dependence of their structures and properties are seriously lacking. It is therefore imperative to obtain a comprehensive understanding of Au doped with TM-rich alloy clusters.
Motivated by the above research background, we carry out the present study on the gold-doped iron clusters FenAu (n = 1−12) in order to systematically explore their geometric structures, relative stabilities, electronic and magnetic properties. We attempt to clarify the following points: (i) how the Au doping influences the geometrical structures of the Fen cluster, (ii) what the size-dependent growth behavior for FenAu is, (iii) whether some novel properties could be found when one Au atom is doped into iron clusters, (iv) how much is the degree to which the magnetic behavior is changed through incorporating the Au atom and what is the magnetic physics origin responsible for this change. It is expected that in the future, our present work can provide some forceful theoretical guidance in the Stern–Gerlach (SG) experiment, the development of good catalysts for synthesis and the biological and biomedical applications.
2. Computational details
All calculations are performed at the DFT level with the DMol3 package in the Materials Studio of Accelrys Inc.[48,49] The electron density functional is treated with the generalized gradient approximation (GGA) corrected exchange potential of Perdew, Burke, and Ernzerhof (PBE).[50] The double numerical basis set augmented with d-polarization and p-polarization functions (DNP) is utilized. Relativistic calculations are performed with scalar relativistic corrections to valence orbitals relevant to atomic bonding properties via a local pseudopotential (VPSR).[51] For the numerical integration, a fine quality mesh size is used, and the real space cutoff of the atomic orbital is set to be 5.5 Å. The convergence criteria for structure optimization and energy calculations are set to be fine with the tolerance for electron density and total energy convergence in self-consistent field (SCF), energy, gradient, and displacement of 1 × 10−6 a.u. (The unit a.u. is short for atomic unit), 1.0 × 10−5 a.u., 0.002 a.u., and 0.005 Å, respectively. In the geometry optimization procedure, we consider a number of initial structures including linear chains, planar and three-dimensional structures in this work to maximize our chance to find the ground state configurations of the alloy clusters. First, we identify the low-lying structures of pure Fen clusters based on earlier theoretical work on iron[6–14] and other TM clusters,[52–,55] and choose these low-lying structures as various reasonable initial structures. Second, on the basis of these initial structures, we substitute the Fe atom with the Au atom at different positions of Fen+1 cluster, and we place the Au atom on each possible site of Fen cluster, and also test some structures through referring to ConMn/ConV,[56,57] ScnAl,[58] FenMn,[59] FenCr,[60] ConAu,[61,62] ConFe,[63] FenPt,[64] and YnAl[65] clusters. In this way, we consider more than 400 candidates altogether. Moreover, without any symmetry constrains, we relax the geometric structure to find out the true ground state. Third, for each geometry structure, the magnetic moment is first allowed to optimize automatically into the favored state (Sz) in DMol3, then, we consider the neighboring spin states (Sz±2) or longer-range spin states and optimize them by fixing the spin state. The calculations are implemented until the minimum energy is reached. Finally, subsequent frequency calculations are performed to confirm the optimized geometry corresponding to the minimum value. In addition, for confirming that our obtained low-lying isomers are the most stable isomers in the FenAu clusters, a global search of Fe7Au cluster based on a particle swarm optimization algorithm within an evolutionary scheme is carried out by using the CALYPSO package.[66] As a result, the most stable structure and metastable state obtained from CALYPSO are consistent with those from our methods mentioned above. The other low-lying isomers are also similar, but only in the order energy there exist some differences. This indicates that the most stable and low-lying isomers our confirmed are indeed in ground state and they are the most stable isomers. Based on the above steps, the net charge and the magnetic moments are evaluated via the Mulliken population analysis.
In order to test the validity of the computational method for the description of FenAu clusters, we first optimize Fe2 and Au2 dimers. All these results are in good agreement with the reported theoretical and experimental results.[1–3,8,13,14,67–71] As listed in Table 1, for Fe2 the bond length d is 2.02 Å, which is very close to the experimental result of 2.02 ± 0.02 Å[2] (1.87 ± 0.13 Å[1]) and previous theoretical data a of 1.98 Å[13] and 2.008 Å.[8] At the same time, Eb, ū, ω, and IP corresponding to 1.621 eV, 3 μB, 418.34 cm−1, and 6.50 eV respectively, are also in good agreement with previous experimental values of 1.64 eV,[13] 3.3 ± 0.5 μB,[4] 415 cm−1,[8] and 6.30 eV, respectively.[3] Similarly, for Au2, our calculated d = 2.49 Å and Eb = 1.189 eV are also in consistent with previous theoretical values (2.49 Å and 1.18 eV[67]). Meanwhile, our caculated ω (181.83 cm−1) and IP (9.41 eV) are similar to the experimental values (191 cm−1 and 9.5 eV[69]) and are also in good accordance with previous theoretical calculations (184.2 cm−1, 9.44 eV[67]). All these results indicate that our approach and accuracy are enough to describe the structures and properties of FenAu clusters.
Table 1.
Table 1.
Table 1.
Calculated values of bond length d (Å), average binding energy Eb (eV), average magnetic moment ū (μB), vibrational frequency ω (cm−1), and ionization potential IP (eV) of the Fe2 and Au2 clusters.
Calculated values of bond length d (Å), average binding energy Eb (eV), average magnetic moment ū (μB), vibrational frequency ω (cm−1), and ionization potential IP (eV) of the Fe2 and Au2 clusters.
.
3. Results and discussion
3.1. Geometrical structures
Using the computational scheme described in Section 2, we have explored a number of low lying isomers and determined the lowest energy structures of FenAu clusters up to n = 12. The obtained ground state structures which include these isomers within 0.4 eV of the most stable structure are shown in Fig. 1, together with the symmetries, magnetic moment, and relative energy (ΔE). The point-group symmetry of each cluster is determined using a tolerance of 0.001 Å between the atomic position in the computed structure and the atomic position in the symmetrized reference structure. The lowest energy structures for pure Fen+1 clusters are also considered for comparison. The small Fen+1 clusters up to n = 12 are extensively investigated computationally by using DFT. In this work, the geometries for pure iron clusters obtained from the published literatures are reoptimized by using the spin-polarized DFT as described in Section 2. With few exceptions, our calculated structures for n ≤ 6 agree with published theoretical results,[8–10,13] similar to recently predicted equilibrium structures.[6,9] This reoptimization leads to a slight change in the geometry of the iron clusters. The values of average binding energy (Eb) and the average bond length (d) for the lowest energy structures of FenAu clusters are calculated and listed in Table 2, where the corresponding values for Fen+1 clusters are given in the brackets.
Fig. 1. Lowest-energy structures and low-lying isomers with different relative energies (in unit eV) of FenAu (n = 1−12) clusters. The ground state geometries of the corresponding bare Fen+1 clusters are also given on the left. The blue and pink balls represent Fe and Au, respectively. Distances are given in unit Å.
Table 2.
Table 2.
Table 2.
Values of average binding energy (Eb), average bond length (d), average magnetic moment at Fe atom (ūatom), HOMO–LUMO (highest occupied molecular orbital–lowest unoccupied molecular orbital) gaps in the lowest energy structures of FenAu clusters. Corresponding values for Fen+1 cluster are given in the brackets.
.
Cluster
Eb/(eV/atom)
d/Å
ūatom
Gap/eV
FeAu
1.552
(1.621)
2.398
(2.020)
3.040
(3.000)
0.040
Fe2Au
2.001
(2.089)
2.394
(2.271)
3.301
(3.333)
0.485
Fe3Au
2.433
(2.516)
2.366
(2.353)
3.010
(3.500)
0.506
Fe4Au
2.736
(2.848)
2.450
(2.398)
3.667
(3.600)
0.404
Fe5Au
2.970
(3.103)
2.450
(2.400)
3.341
(3.333)
0.376
Fe6Au
3.144
(3.252)
2.448
(2.391)
3.163
(3.143)
0.475
Fe7Au
3.251
(3.337)
2.446
(2.426)
3.000
(3.000)
0.452
Fe8Au
3.318
(3.387)
2.429
(2.393)
2.885
(2.889)
0.423
Fe9Au
3.395
(3.445)
2.455
(2.434)
2.980
(3.000)
0.393
Fe10Au
3.453
(3.516)
2.477
(2.454)
3.086
(3.091)
0.120
Fe11Au
3.521
(3.576)
2.484
(2.476)
2.983
(3.167)
0.204
Fe12Au
3.586
(3.657)
2.514
(2.504)
3.076
(3.385)
0.126
Table 2.
Values of average binding energy (Eb), average bond length (d), average magnetic moment at Fe atom (ūatom), HOMO–LUMO (highest occupied molecular orbital–lowest unoccupied molecular orbital) gaps in the lowest energy structures of FenAu clusters. Corresponding values for Fen+1 cluster are given in the brackets.
.
FeAu is C∞v symmetry dimer, whose bond length and average binding energy are 2.398 Å and 1.552 eV/atom, respectively. These results lie between Fe2 (2.020 Å, 1.621 eV/atom) and Au2 (2.488 Å, 1.189 eV/atom). Compared with it of pure Fe2 (6 μB), the magnetic moment of FeAu dimer is reduced by 3 μB. Our computed values are in good agreement with the values (2.470 Å and 3 μB) obtained in previous paper.[39]
Energetically, the most favorable structure of the Fe2Au cluster with a total magnetic moment of 7 μB is in the form of an isoscele triangle (C2v), whose Fe–Fe distance is 2.073 Å. This may be seen as replacing an Fe atom with an Au atom. Our results about Fe2Au are proved by Sun et al.,[33] whose results show the Fe–Fe distance of 2.020 Å and the same magnetism (7 μB). Its metastable state shares the linear chain structure, and the third stable state (9 μB) is similar to the ground state. They are energetically less stable by 0.020 eV and 0.124 eV, respectively.
A planar structure (C2v) is the most stable Fe3Au cluster with 9 μB of total magnetic moment. The second energetically degenerate structure (9 μB), which is of a turnup rhombus (four atoms are not on the same plane) structure (Cs), is energetically less favorable than the most stable structure in energy by 0.062 eV. The third lowest-energy structure is similar to the most stable structure, but they differ by 11 μB of total magnetic moment. A distorted tetrahedronal configuration with Cs symmetry is the fourth stable isomer. The energy is 0.382 eV less stable in energy compared the lowest energy structure, which means that there is the turning from the planar (2D) structure to the three-dimensional (3D) structure.
In the case of Fe5, we find the C2v triangular bipyramid structure to have the lowest energy. For Fe4Au, the overall structures are totally retained, but they are slightly distorted triangular bipyramid with the Au atom at the vertex. The most stable isomer is of Cs symmetry and it has a total magnetic moment of 15 μB, which is reduced by 3 μB in comparison with the pure Fe5 (18 μB). By contrast, the second stable structure (Cs) with a total magnetic moment of 13 μB is 0.120 eV evidently higher in energy. The other two stable isomers (C3v) have total magnetic moments of 15 μB and 13 μB, and 0.164 eV and 0.166 eV above the lowest-energy structure, respectively.
The C2v configuration of Fe5Au with the magnetism (17 μB) is found to be the most stable isomer, which may be viewed as a deviation from substitution of a Fe atom by an Au atom in the octahedron Fe6 (D4h) structure. The other distorted octahedron geometrical isomer with C4v symmetry is considered to be the second stable structure at a slightly higher energy (0.037 eV). Differently, the following two stable isomers prefer to share a capped triangle bipyramid with Au atom being at the cap position, which are located at 0.094 eV and 0.286 eV in energy above the most stable conformation.
For the Fe6Au clusters, we consider a lot of initial geometries including the bicapped triangular bipyramid, capped octahedron, pentagonal bipyamid, bicapped square pyramid, etc. At last, an Au-capped octahedron structure with Cs symmetry is found to have the lowest energy. However, the metastable state is a distorted pentagonal bipyramid (C2v) with the impurity Au atom in the middle pentagonal ring plane, and is found to be only 0.091 eV less stable than the ground state. The third state also favors a capped octahedron (C3v) and is energetically less favorable than the most stable structure by 0.137 eV. The fourth state is still a capped octahedron (Cs) with the Au atom at the square site, and is 0.382 eV higher in total energy. Moreover, we notice that these four stable structures of Fe6Au have the equal total magnetic moment (19 μB).
Energetically the most favorable structure for Fe7Au (21 μB) is an Au-capped distorted pentagonal bipyramid with C1 symmetry. The metastable state with a classic bicapped octahedron structure is 0.134 eV in energy higher than the lowest-energy structure. As seen in Fig. 1, the third lowest-energy structure is still a bicapped octahedron with an Au atom at the vertex position of the octahedron. It has the same magnetism as the ground state structure and is 0.136 eV higher in energy. A capped-pentagon bipyramid (21 μB), where Au atom becomes a part of the pentagonal ring, is regarded as the fourth stable isomer and has a high relative energy of ΔE = 0.263 eV.
In the case of n = 8, the most stable Fe8Au (23 μB) with C1 symmetry is capped one Au atom on the Fe8 ground-state geometry. The second stable isomer and the third stable isomer have similar geometric configurations with the lowest-energy structure but Au atoms occupy the different capped positions, whose total energies are only 0.046 eV and 0.094 eV above the lowest-energy isomer. The last stable isomer (C1) with a total magnetic moment of 23 μB can be viewed as the Fe atom in the ground state structure of Fe9 replaced by one Au atom.
For the case of Fe9Au, the lowest-energy structure with Cs symmetry prefers one capping Au atom on the surface of the ground-state Fe9 cluster. As shown in Fig. 1, Au atom substituting one Fe atom of a three-capped pentagonal bipyramid structure of Fe10 cluster yields the second and third stable structure of the Fe9Au cluster with the Cs symmetry, and are 0.038 eV and 0.158 eV in energy less favorable than the most stable structure, respectively. The last stable structure with C2v symmetry is formed by capping one Au atom on tri-capped triangular prism, which is 0.167 eV higher in energy than the ground state.
In terms of Fe10Au (Cs), the most stable structure is similar to that of the Fe11 cluster which is a tetra-capped pentagonal bipyramid with the Au atom at the vertex position. It has a total magnetic moment 31 μB, and is 0.155 eV in energy lower than that of the metastable state. As seen in Fig. 1, the third and the fourth lowest energy structure which can be regarded as Au replacing one Fe atom of the Fe11 cluster have the same geometry but different magnetisms, and are 0.181 eV and 0.339 eV above the lowest-energy structure, respectively.
Just like the Fe12 cluster, the lowest-energy structure of Fe11Au cluster takes an icosahedral growth pattern where Au atom substitutes the Fe site at the apex with a total magnetic moment of 33 μB as shown in Fig. 1. Differently, an Au atom capping a peripheral position of the Fe11 cluster produces the second and the fourth lowest-energy structure, where relative energies are 0.068 eV and 0.223 eV, respectively. In comparison, the third lowest-energy structure takes a similar geometry to the most stable structure but the increased magnetic moment (35 μB) and is located at 0.114 eV in energy above the ground state.
With regard to Fe12Au cluster, we consider the icosahedral, hexagonal close packed (hcp), rhombic dodecahedral (bcc) and cuboctahedral (fcc) structures as initial geometries. Overall, the most stable configurations we have finally obtained can be looked upon as a substitutional Au impurity in the pure Fe13 cluster, in which Au atom lies at the one vertex site of the icosahedral structure. As shown in Fig. 1, the other low-lying isomers (a, b, c) have notable instablities (0.077, 0.145, and 0.194 eV, respectively).
The overall evolutionary trends for the FenAu series show that for n = 1−2, 4−5, 10−12, the Au doped geometries are similar to those of Fen+1 clusters, where the Au atom occupies a substitutional surface site accompanied with a slight distortion in the host cluster. And the way of growing FenAu (n = 6−9) clusters is that the Au atom occupies a peripheral position of Fen cluster. This phenomenon can be attributed to the two aspects: on the one hand, Au atom prefers to occupy peripheral location, which can be explained by the cohesive energy (CE) and the surface energy (SE) of the metal atom. To minimize the total energy, the atom with the smaller surface energy and cohesive energy tends to occupy the surface, while the atom with a higher surface energy and cohesive energy favors the interior[72–74] The surface energy and average cohesive energy of the bulk Au are 1.506 J·m−2 and 3.81 eV/atom, which are smaller than those of the bulk Fe 2.417 J·m−2 and 4.28 eV/atom respectively.[75,76] On the other hand, the atomic size effect also accounts for this. The smaller atoms tend to occupy the more sterically confined core, especially in icosahedral clusters, exhibiting the compression effects finally forming a stable cluster.[72] The atomic radius of Au is larger than that of Fe atom.[75] Thus, the Fe atoms prefer to occupy central position, whereas the Au atom prefers to be located at the site of surface. Similar phenomena and explanations also appeared in previous studies.[61,76–79]
3.2. Relative stabilities
In this subsection, we will investigate the size-dependent physical properties of these clusters. The variation of average binding energy Eb(n) with cluster size is shown in Fig. 2(a). For comparison, we also show the variation of Eb(n) with cluster size for pure iron cluster. The average binding energy Eb(n) is defined as
where E is the total energy of the respective atoms or clusters. From Fig. 2(a), it can be seen that the Eb rises monotonically with increasing the sizes of both Fen+1 and FenAu clusters. This is because both the average coordination number and the effective hybridization increase as cluster size n increases. Especially, the average binding energy of Fen+1 clusters increases towards the cohesive energy of bulk Fe solid (4.280 eV/atom) with cluster size growing. Meanwhile, in the beginning stage for n = 1−6, Eb of FenAu clusters increases rapidly; when n goes from 7 to 12, the Eb increases with size n increasing, but finally tend to saturation and convergence. Despite all this, the Eb of the FenAu cluster is 0.050 eV−0.133 eV smaller than that of Fen+1, indicating the decrease of the stability of iron clusters after Au doping. We infer that a reduction of Eb could be attributed to the following two reasons: (i) the cohesive energy of bulk Fe (4.28 eV) is significantly larger than bulk Au (3.81 eV), indicating that the stability of pure Fen+1 cluster is weakened after one Au atom doping. Based on the above analysis, the trend of Eb of our calculated dimers is Fe2 (1.621 eV) > FeAu (1.552 eV) > Au2 (1.189 eV), which is consistent with the cohesive energy of bulk; (ii) much longer average bond lengths of doping clusters, which generates a weaker Fe–Au bond than the Fe-Fe bond (as seen in Table 1).
Fig. 2. Size dependences of averaged atomic binding energies (a), fragmentation energies (b), and second-order difference energies (c) for the lowest energy structures of FenAu and Fen+1 (n = 1−12) clusters.
In cluster physics, the fragmentation energies and second-order difference of energies are sensitive indicators of the relative stability. Thus, the size dependence of the fragmentation energies and second-order difference of energies for FenAu (n = 1−12) clusters are further investigated. The fragmentation energy Δ1E(n) and the second-order difference energy Δ2E (n) are calculated using the following formulas:
where E (·) is the respective atoms or clusters. Based on the above formulas, the evolutions of Δ1E (n) and Δ2E (n) are plotted in Figs. 2(b) and 2(c). Obviously, the local peaks of the fragmentation energies and second-order difference of energies are localized at n = 6 and 9, indicating that Fe6Au and Fe9Au clusters are more stable than the other isomers. The high stability of Fe6Au may mainly stem from the Au capping a similar compact octahedral Fe6 core. Meanwhile, it is clearly seen that both curve of the fragmentation energies and curve of the second-order difference of energies reach their maxima at n = 5 and 10, indicating that the clusters of Fe6 and Fe11 are the most stable structures among our investigated Fen+1 (n = 1−12) clusters. The magic stability of Fe6 cluster has appeared in previous studies.[8,10]
In addition, the HOMO–LUMO gap is a characteristic quantity of electronic structure in clusters and is commonly used to measure the ability for clusters to undergo activated chemical reactions with small molecules. A big energy gap usually corresponds to a high chemical stability. As shown in Fig. 3, FeAu cluster has the narrowest HOMO–LUMO gap, which is only 0.040 eV. However, the gap change trend for FenAu (n = 2−9) clusters which are bigger in energy gap (0.376 eV−0.506 eV) almost keeps a relatively stable value, indicating their high chemical inertness. There are three dramatic drops at n = 10, 11, and 12. The decreases of the gaps (0.120 eV−0.204 eV) suggest that more atoms are involved in the metal-metal bonding.[80,81] Thus, these clusters could have potential utility in new cluster catalysis nanomaterials as building block with reactive and metallic properties.
Fig. 3. HOMO–LUMO gaps of the most stable structures of FenAu clusters as function of cluster size.
The magnetic stability of these clusters can be verified by calculating the spin gaps as a function of the cluster size. The spin gaps for a magnetic cluster are defined as
where and are the HOMO level and LUMO level of the majority spin density, and are the HOMO level and LUMO level of the minority spin density, respectively. The magnitude of spin gap is a measure of the chemical activeness of cluster: the higher the spin gap, the more stable the cluster is. As seen from Fig. 4, both δ1 and δ2 of FenAu (n = 1−12) clusters are positive, indicating that those clusters are magnetically stable. Moreover, the δ1 and δ2 values are found to decrease as size increases. The higher spin gaps of the small size clusters are attributed to the quantum size effect.[82]
Fig. 4. Size dependences of the spin gaps of FenAu (n = 1−12) clusters.
3.3. Magnetic moments
The total magnetic moments of the ground-state FenAu (n = 1−12) clusters are calculated and the results are presented in Fig. 5(a). These results show that the total magnetic moments of the ground-state FenAu (n = 1−12) clusters are 3 μB smaller than those of the pure Fen+1 with three exceptional jumps of 5 μB (Fe3Au, Fe11Au, and Fe12Au). The valence electron configurations of the free atoms Fe and Au are 3d64s2 and 4f145d106s1, respectively. For an isolated Fe (Au) atom, the magnetic moment is 4 μB (1 μB) based on the Hund’s rule. Taking the ferromagnetic alignment into account on the Fen+1 and most of FenAu clusters, the unpaired electrons of Fen+1 and FenAu clusters will be 4(n+1) and 4n+1, respectively. Therefore, the magnetic moments of the FenAu clusters are supposed to decrease by 3 μB. A similar explanation can be found in the studies of Con−1Mn and Con−1V clusters.[56] In addition, the average magnetic moments of Fe atom for ground state Fen+1 and FenAu clusters are also plotted in Fig. 5(b). It can be seen that the Fe atomic average magnetic moment remains almost unchanged for the case of Au doping, which still keeps about 3 μB of magnetism of pure Fen+1 clusters. However, the magnetism of FenAu clusters is reduced by 5 μB compared with that of the corresponding pure Fen+1 clusters, and Fe atomic average magnetic moment of FenAu clusters decreases largely.
Fig. 5. Size dependences of the magnetic moments (a) and the average magnetic moments of Fe atom (b) for FenAu and Fen+1 (n = 1−12) clusters.
As shown in Fig. 6, the average bond lengths and average coordination numbers display a similar trend as a function of the cluster size, which is due to their similar geometrical structures. Evidently, in the initial stage for n = 1−4, their average bond lengths increase rapidly, when n goes from 4 to 8, become relatively stable, as n goes from 8 to 12, they increase gradually again. And yet, the average coordination numbers always increase gradually. As is well known, the larger the bond length, the larger the magnetic moment of cluster is, on the contrary, the larger the coordination numbers, the smaller the magnetic moment of cluster is. The consequence of mutual competition of the two factors ultimately determines the most stable structure of each cluster size.[83,84] Thus, as shown in Fig. 5(b), the average magnetic moment of the Fen and FenAu presents an increasing trend in n = 1−4 decreases rapidly in n = 4−8 increases gradually again in n = 8−12. This indicates that the average bond length is dominant in influencing the magnetism of the clusters.
Fig. 6. Variations of average bond length (a) and average coordination number (b) with the cluster size for FenAu and Fen+1 (n = 1−12) clusters.
The reason why the Fe atomic average magnetic moments of n = 3, 11, and 12 alloy clusters obviously decrease compared with that of pure Fe clusters is explained in detail. It can be seen from Fig. 6 that for Fe12Au and Fe13, the average bond lengths remain almost unchanged and average coordination numbers are the same. So the reason most likely is the result of the symmetry. Researches about clusters pointed out that the higher the symmetry, the bigger the magnetic moment is.[85,86] Generally, the higher symmetry of a cluster forms a narrow d electronic state, which leads to spin splitting more easily, and become a more parallel d electronic spin, consequently generating larger magnetism. Since one Fe atom is substituted by one Au atom at the apex of the icosahedron, the symmetry Cs of Fe12Au is lower than D5d of Fe13. Thus, the average magnetic moment of Fe atom obviously decreases with respect to pure Fe cluster. This explanation is suitable for Fe11Au cluster. However, Fe3Au does not have the close-packed tetrahedron geometric structure, but has the complete plane rhombic geometric construction where the dihedral angle is 0 degree. Therefore, Fe4 with its relative close-packed tetrahedron geometric structure exhibits the stronger spin polarization, while the spin polarization of Fe3Au is sharply weakened, finally leading to the fact that the average magnetic moment of Fe atom of Fe3Au cluster is attenuated compared with that of the pure Fe4 cluster.
For the optimized geometry structure, the total magnetic moments of FenAu clusters, the local moments of Fe and Au atoms, the atomic net charges and the local moments of 3d, 4s, and 4p states for Fe and Au atoms are calculated and summarized in Table 3. The local magnetic moment of per Fe (Au) is the summation of the differences in number between spin-up and spin-down 3d/4s/4p (5d/6s/6p) electrons. The spin magnetic moment of an electron is 1 μB.[83] The atom label is presented in Fig. 7. It can be seen from Table 3 that all of the atoms in the ground states of FenAu alloy clusters are aligned ferromagnetically except for FeAu, Fe3Au, and Fe8Au. At the same time, the magnetic moments of the alloy clusters are mainly manipulated by the 3d electrons whereas the s and p electrons contribute only a small amount of net spin. For FenAu clusters, the 4s state of Fe and 5d state of Au all lose some electrons, meanwhile 3d, 4p states of Fe and 6s, 6p states of Au gain extra electrons (see in Table 3). It therefore means that there happens s-pd hybridization in the Fe and Au atoms. As a result, the local magnetic moments of Fe and Au in FenAu clusters are attenuated compared with those of their free atoms. From the analysis of atomic net charge Q, we find that the charge always transfers from Fe to Au atom, which is consistent with the empirical Pauling electro-negativities (χp(Fe) = 1.83 eV, χp(Au) = 2.54 eV). Thus, this trend of charge transfer leads to the Au atom being nearly non-magnetic. The above results imply that the charge transfer and the strong hybridization between Fe 4s, 3d and Au 6s, 6p, 5d states might be one major reason for inducing magnetism of FenAu cluster.
Table 3.
Table 3.
Table 3.
Atom labels and values of atomic net charge Q (e), local magnetic moment μatom (μB) of the atom, Mulliken charge (e), and local magnetic moment of s, p, and d states μL (μB) for the lowest energy structures of FenAu (n = 1−12) clusters.
.
Clusters
Atom label
Q
μatom
Mulliken charge/μL
M/μB
3d/5d
4s/6s
4p/6p
1/C∞v
Au(1)
-0.258
−0.04
9.812/0.062
1.411/−0.117
0.039 / 0.016
3
Fe(2)
0.258
3.04
6.837/3.099
0.838/−0.064
0.067/0.005
2/C2v
Au(1)
−0.322
0.399
9.678 /0.120
1.496/0.220
0.153 /0.059
7
Fe(2,3)
0.161
3.300
6.757/3.100
0.834/0.148
0.244/0.057
3/C2v
Fe(1,3)
0.162
2.973
6.815/2.989
0.698/−0.008
0.319/−0.002
9
Fe(2)
−0.001
3.084
6.761/3.035
1.067/0.076
0.170/−0.022
Au(4)
−0.323
−0.030
9.716/0.035
1.468/ −0.047
0.144/−0.018
4/Cs
Fe(1,5)
0.065
3.769
6.576/3.282
0.821/0.262
0.534/0.232
15
Au(2)
−0.220
0.333
9.620/0.158
1.315/0.048
0.291/0.127
Fe(3)
0.044
3.653
6.636/3.193
0.782/0.249
0.533/0.218
Fe(4)
0.045
3.476
6.699/3.171
0.831/0.104
0.422/0.207
5/C2v
Fe(1)
−0.020
3.191
6.776/3.010
0.735/0.089
0.505/0.099
17
Fe(2,3)
0.057
3.361
6.672/3.158
0.821/0.101
0.446/0.108
Au(4)
−0.224
0.295
9.579/0.155
1.265/0.088
0.387/0.053
Fe(5,6)
0.065
3.396
6.693/3.121
0.725/0.161
0.512/0.121
6/Cs
Fe(1,2)
0.013
3.208
6.743/3.060
0.826/0.069
0.415/0.086
19
Fe(3)
0.033
2.972
6.805/2.859
0.594/0.049
0.562/0.071
Fe(4)
−0.048
3.116
6.777/2.955
0.721/0.085
0.562/0.071
Fe(5,6)
0.127
3.237
6.711/3.074
0.697/0.071
0.546/0.083
Au(7)
−0.264
0.022
9.677/0.070
1.341/−0.059
0.253/0.011
7/C1
Fe(1)
−0.019
3.069
6.788/2.972
0.825/0.050
0.401/0.055
21
Fe(2)
−0.047
2.888
6.826/2.847
0.836/0.027
0.381/0.020
Fe(3)
−0.019
3.071
6.787/2.974
0.826/0.049
0.402/0.055
Fe(4)
0.077
3.074
6.784/2.961
0.706/0.046
0.427/0.074
Fe(5)
0.078
3.069
6.785/2.958
0.705/0.045
0.426/0.074
Fe(6)
0.175
2.887
6.787/2.826
0.583/0.010
0.449/0.048
Fe(7)
0.040
2.953
6.777/2.891
0.646/0.033
0.532/0.036
Au(8)
−0.283
−0.001
9.684/0.054
1.377/−0.055
0.229/0.001
8/C1
Fe(1)
−0.069
2.802
6.863/2.785
0.835/0.013
0.367/0.010
23
Fe(2)
−0.036
3.004
6.807/2.948
0.834/0.020
0.391/0.043
Fe(3)
0.146
2.904
6.805/2.843
0.600/0.029
0.445/0.040
Fe(4)
0.033
2.875
6.849/2.824
0.684/0.014
0.429/0.044
Fe(5)
0.034
3.005
6.783/2.927
0.734/0.053
0.446/0.032
Fe(6)
0.183
2.713
6.837/2.686
0.578/0.008
0.398/0.026
Fe (7)
0.061
2.845
6.823/2.816
0.630/0.013
0.482/0.024
Au(8)
−0.286
−0.077
9.682/0.046
1.370/−0.117
0.241/−0.006
Fe(9)
−0.065
2.928
6.859/2.834
0.802/0.041
0.400/0.060
9/Cs
Fe(1,3)
0.031
3.003
6.791/2.957
0.730/0.018
0.443/0.035
27
Au(2)
−0.320
0.179
9.655/0.081
1.457/0.061
0.214/0.038
Fe(4,6)
−0.014
3.088
6.795/2.961
0.781/0.060
0.434/0.075
Fe(5)
0.014
2.949
6.778/2.886
0.681/0.024
0.523/0.047
Fe(7)
0.225
2.698
6.778/2.685
0.629/−0.022
0.363/0.041
Fe(8)
0.169
2.902
6.800/2.841
0.612/0.011
0.415/0.056
Fe(9,10)
−0.061
3.044
6.794/2.946
0.866/0.037
0.415/0.056
10/Cs
Au(1)
−0.346
0.142
9.652/0.095
1.473/−0.006
0.227/0.054
31
Fe(2,9)
−0.076
3.233
6.747/3.052
0.937/0.103
0.389/0.084
Fe(3,10)
−0.068
3.191
6.748/2.979
0.845/0.10
0.471/0.113
Fe(4)
0.001
3.072
6.797/2.913
0.703/0.036
0.494/0.131
Fe(5,6)
0.164
2.930
6.810/2.868
0.640/0.006
0.382/0.063
Fe(7)
0.311
2.731
6.803/2.768
0.600/−0.068
0.280/0.037
Fe(8)
0.073
3.026
6.783/2.915
0.647/0.018
0.491/0.100
Fe(11)
−0.078
3.322
6.725/3.036
0.920/0.148
0.429/0.144
11/Cs
Fe(1,3)
0.009
3.031
6.790/2.897
0.737/0.050
0.457/0.092
33
Fe(2,4)
−0.083
3.142
6.757/2.998
0.854/0.065
0.469/0.087
Fe(5)
−0.007
3.021
6.820/2.917
0.754/0.040
0.428/0.071
Fe(6)
−0.040
3.126
6.759/2.914
0.773/0.085
0.504/0.135
Fe(7,9)
−0.053
3.093
6.763/2.967
0.836/0.040
0.450/0.092
Fe(8,10)
−0.021
3.016
6.790/2.882
0.730/0.033
0.495/0.110
Fe(11)
0.681
2.107
6.981/2.263
0.579/−0.098
−0.249/−0.052
Au(12)
−0.338
0.182
9.596/0.076
1.326/0.073
0.425/0.032
12/Cs
Fe(1,11)
−0.055
3.221
6.745/2.994
0.771/0.069
0.535/0.166
39
Fe(2)
−0.067
3.342
6.703/3.037
0.817/0.123
0.542/0.190
Fe(3,7)
−0.053
3.395
6.681/3.090
0.829/0.142
0.540/0.171
Fe(4,6)
−0.032
3.393
6.692/3.085
0.813/0.157
0.523/0.159
Fe(5)
−0.013
3.466
6.660/3.116
0.801/0.161
0.549/0.197
Fe(8)
−0.041
3.464
6.651/3.113
0.817/0.148
0.570/0.211
Fe(9,12)
−0.026
3.190
6.754/2.981
0.759/0.066
0.508/0.150
Au(10)
−0.331
0.237
9.553/0.096
1.271/0.073
0.519/0.068
Fe(13)
0.783
2.093
7.073/2.388
0.640/−0.122
−0.509/−0.168
Table 3.
Atom labels and values of atomic net charge Q (e), local magnetic moment μatom (μB) of the atom, Mulliken charge (e), and local magnetic moment of s, p, and d states μL (μB) for the lowest energy structures of FenAu (n = 1−12) clusters.
Fig. 7. Atom labels of the ground state of FenAu (n = 1−12) clusters.
To gain an insight into the magnetism of the FenAu cluster, we also explore the spin densities of the alloy clusters. In Fig. 8, for proper comparison, we plot the differences between the charge density for spin-up electrons and that of spin-down electrons for Fen+1 (n = 1−12) clusters. As seen from Fig. 8, the contributions to magnetism of the cluster mainly originate from Fe atoms in FenAu and the Au atom has no influence on magnetism. Meanwhile, the spin density volume of the Fe atom has almost no change after substituting an Fe atom with an Au atom, which means that the local magnetic moment of Fe in FenAu is nearly equal to those in Fen+1 (n = 1−12) clusters. Meanwhile, the doping-Au atom has no spin density, which is consistent with the fact that the Au atom is almost nonmagnetic. In addition, to further ascertain the reason for reducing the magnetic moment, we perform detailed analyses of spin polarization (P) of Fe atomic d orbital for Fen+1 and FenAu alloy clusters in the Supporting Information Table S1. The results show that the average spin polarizations of d orbital of Fe atom for Fen+1 and FenAu alloy clusters are almost identical. Therefore, doping with the Au atom results in reducing the total magnetic moment of FenAu clusters, while it has no effect on the local magnetic moment of Fe atom.
Fig. 8. spin densities of FenAu and Fen+1 (n = 1−12) clusters. The density isosurfaces are drawn at 0.5 a.u.
In order to explore the magnetic origin further, In Fig. 9, we present the s-, p-, and d-projected partial densities of states (PDOSs) and local densities of states (LDOSs) of Fe11 and Fe10Au clusters as representatives. (See the Supporting Information about the PDOS and LDOS contributions of Fe13 and Fe12Au clusters). It is clearly seen that the magnetic moments of these clusters mainly come from d states, while the s and p states only contribute a small amount of net spin, which is in good agreement with the above Mulliken population analyses of the on-site atomic charges and the atomic spin density. For Fe11, we find that the up- and down-spin channels of the d PDOS are seriously asymmetric, which has a magnetic moment of 34 μB. Substituting Fe atom by Au reduces the up-spin channel, and a small peak of the down-spin d PDOS is broadened around −4.2 eV below the Fermi level, which consequently enhances significantly the depletion of down-spin d states, finally leading to the total magnetic moment reducing 3 μB. In addition, the up-spin channels of Fe10Au/Fe10 is evidently larger than that of down-spin, leading to the characteristic ferromagnetic exchange splitting with the local magnetic moment of 30.858 μB. For Fe10Au/Au, the distributions of the up- and down-spin d electrons are nearly equal, resulting in lower spin polarization. So the local magnetic moment of Au in Fe10Au clusters is only 0.142 μB. This further confirms that the Fe atomic magnetism is the main contribution to the Fe10Au cluster magnetic moment.
Fig. 9. PDOSs and LDOSs of Fe10Au and Fe11 clusters.
4. Conclusions
In this work, we perform DFT–GGA computations of FenAu (n = 1−12) alloy clusters. The size-dependent structural, electronic, magnetic properties are investigated. The main conclusions are summarized as follows.