Stability of Ge 12 C 48 and Ge 20 C 40 heterofullerenes: A first principles molecular dynamics study Carlo Massobrio a,⇑ , Duval Mbongo Djimbi a , Masahiko Matsubara b , Roberto Scipioni c , Mauro Boero a a Institut de Physique et de Chimie des Matériaux de Strasbourg, CNRS-University of Strasbourg, UMR 7504, 23 Rue du Loess, BP43, F-67034 Strasbourg, France b EMAT, University of Antwerp, Groenenborgerlaan 171, B-2020 Antwerp, Belgium c Department of Earth Sciences, University College London, Gower Street, London WC1E 6BT, United Kingdom article info Article history: Received 26 July 2012 In final form 16 November 2012 Available online 27 November 2012 abstract By using first-principles molecular dynamics, we address the issue of structural stability for the C 60m Ge m family of doped heterofullerenes through a set of calculations targeting C 48 Ge 12 and C 40 Ge 20 . Three kinds of theoretical tools are employed: (a) static structural optimization, (b) a bonding analysis based on local- ized orbitals (Wannier wavefunctions and centers) and (c) first-principles molecular dynamics at finite temperature. This latter tool allows concluding that the segregated form of C 40 Ge 20 is less stable than its Si-based counterpart. However, the non-segregated forms of C 40 Ge 20 and C 40 Si 20 have comparable sta- bilities at finite temperatures. Ó 2012 Elsevier B.V. All rights reserved. 1. Introduction Doping fullerenes via the replacement of C atoms with other atomic species has been the object of a wealth of experimental and theoretical studies aimed at understanding the structural and electronic properties of these new nanosystems, with special emphasis on the changes in the chemical reactivity [1–22]. Over the past fifteen years, a well targeted effort has been devoted to C 60m Si m by relying on first principles molecular dynamics at finite temperatures (FPMD in what follows) and structural optimization techniques (at T = 0 K) in the framework of the density functional theory (DFT) [23–33]. This research was stimulated by early exper- iments based on mass spectroscopy and photofragmentation, showing the existence of m = 12 as an upper limit to the number of doped silicon atoms in C 60m Si m heterofullerenes [10,11,18– 20]. For systems belonging to this family, we recall that a first char- acterization was performed in the case of one or two silicon atoms doping C 60 (C 59 Si, C 58 Si 2 ). Then, the number of Si atoms was grad- ually increased up to m = 12, thereby providing a first milestone in the search of a theoretical counterpart to the experimental predic- tions [11–13,28]. The quest for the highest possible content of Si atoms compatible with the existence of the fullerene network was at the origin of structural optimizations (i.e. T = 0 K results) performed on C 40 Si 20 , C 36 Si 24 and C 30 Si 30 [27]. As a main common feature, it was found that the binding energy is lower when two neighboring and yet spatially segregated regions are formed, each one populated exclusively by silicon or carbon atoms. In what follows, we shall label segregated and non non-segregated arrangements ‘SE’ and ‘NSE’, respectively. Consideration of thermal effects via FPMD provided a new insight into the properties of Si-doped heterofullerenes. By focusing on SE structures only, an upper limit to the stability was found (m = 20) and rationalized on the basis of the specific atomic-scale interac- tions [29]. The conclusions of Ref. [29] have been recently improved and extended by considering the case of the NSE C 30 Si 30 structures and their finite temperature behavior. We could show that the ther- mal stability of NSE heterofullerenes is much larger than the one of the SE counterpart [25]. This can be explained by invoking charge transfer effects from the more electronegative C atoms to the less electronegative Si atoms. In the presence of thermal effects, the elec- trostatic attraction between Si and C atoms is able to preserve the integrity of the NSE cage, since (at least for m < 30) no Si–Si bonds featuring neighboring positive charges are present. The opposite sit- uation is encountered in SE structures. In this case, for increasing m, the Si–C interactions at the Si–C frontier are not able to offset the energetic cost related to: (a) the unfavorable Si–Si sp 2 bonds inher- ent in the Si rich part of the network and (b) the repulsive effect of charged Si–Si pairs of neighbors within the Si populated part of the cage. As a result, on a time scale of about 20 ps, no thermal insta- bility was found for NSE C 30 Si 30 at 2000 K, while fragmentation occurred readily in the SE C 30 Si 30 case. In view of the general scenario obtained for C 60m Si m , it is of interest to determine whether or not other chemical elements hav- ing the same number of valence electrons as C and Si can become good candidates for fullerene doping upon C atoms replacement. To this aim, we selected Ge and carried out a series of structural optimizations for two values of m, specifically C 48 Ge 12 and 0009-2614/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cplett.2012.11.033 ⇑ Corresponding author. Fax: +33 388107249. E-mail addresses: Carlo.Massobrio@ipcms.unistra.fr (C. Massobrio), mbongo @ipcms.unistra.fr (D.M. Djimbi), matsubara.masahiko@gmail.com (M. Matsubara), r.scipioni@ucl.ac.uk (R. Scipioni), Mauro.Boero@ipcms.unistra.fr (M. Boero). Chemical Physics Letters 556 (2013) 163–167 Contents lists available at SciVerse ScienceDirect Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett