Structural, electronic, and magnetic properties of Fe _x Co _y Pd _z (x + y + z ≤ 7) clusters: a density functional theory study

Transition metal alloy nanoparticles are of interest both theoretically and experimentally, particularly due to their potential technological applications, and to their novel structural and magnetic properties in the subnanometer region. Here we compute structural parameters, chemical and magnetic properties, and the fragmentation channels of Fe\(_x\)Co\(_y\)Pd\(_z\) nanoparticles, for \(x+y+z\, \le \,7\), and compare our results with macroscopic systems whenever it is feasible. We carry out density functional theory calculations, as implemented in the SIESTA code, for all possible concentrations (i.e., all x-, y-, and z-values). The seeds for the possible homotops are built using a semiempirical Gupta potential; these, and additional low coordinated conformations, are thereafter subject to reoptimization by means of the SIESTA code. To the best of our knowledge, this is the first time that such kind of calculations are performed for all the possible compositions of up to 7 atom ternary nanoclusters. We find that the binding is strongest in the FeCo-rich region and weakest for pristine Pd for all the sizes we considered. Interatomic distances in general decrease monotonically, as the FeCo region is approached. The total magnetic moment varies almost continuously over the composition range, with the large Fe moment being quenched by the addition of Pd and/or Co; however, an almost continuous range of the moments magnitude can be achieved, which allows for fine tuning magnetism by controlling the composition. As far as the fragmentation channels are concerned, for neutral, cationic, and anionic clusters, the most likely path is through atomic Pd\(^{0}\), Pd\(^{+}\), and Pd\(^{-}\), when Pd is present in the cluster. However, in the absence of Pd, the most likely fragmentation channel is through the majority element. Molecular fragmentation channels are only observed for very small cluster sizes.