Origin of the metal-insulator transition of indium atom wires on Si(111)

Abstract

As a prototypical one-dimensional electron system, self-assembled indium (In) nanowires on the Si(111) surface have been believed to drive a metal-insulator transition by a charge-density-wave (CDW) formation due to electron-phonon coupling. Here, our first-principles calculations demonstrate that the structural phase transition from the high-temperature 4×1 phase to the low-temperature 8×2 phase occurs through an exothermic reaction with the consecutive bond-breaking and bond-making processes, giving rise to an energy barrier between the two phases as well as a gap opening. This atomistic picture for the phase transition not only identifies its firstorder nature but also solves a long-standing puzzle of the origin of the metal-insulator transition in terms of the ×2 periodic lattice reconstruction of In hexagons via bond breakage and new bond formation, not by the Peierls instability-driven CDW formation. PACS numbers: 73.20.At, 68.35.Md, 71.30.+h Low-dimensional electronic systems are of great interest in contemporary condensed-matter physics because of their susceptibility to charge density wave (CDW) instability [1], non-Fermi liquid behavior [2], spin ordering [3,4], and superconductivity [5,6] at low temperatures. Specifically, metalatom adsorption on semiconductor surfaces provides a unique playground for the exploration of such exotic physical phenomena [7,8]. We here focus on a prototypical example of quasi-one-dimensional (1D) systems, self-assembled indium (In) atom wires on the Si(111) surface [9][10][11]. Each In wire consists of two zigzag chains of In atoms [see the left panel of Fig. 1(a)] [10]. Below ∼120 K, this quasi-1D system undergoes a reversible phase transition initially from a 4×1 structure to a 4×2 one, then to an 8×2 structure [9,12], showing a period doubling both parallel and perpendicular to the In wires. This (4×1)→(8×2) structural phase transition is accompanied by a metal-insulator (MI) transition [9,13,14]. For the explanation of such a MI transition, the CDW mechanism due to a Peierls instability was initially proposed [9,[13][14][15], but subsequently other mechanisms based on an orderdisorder transition [19,20] and many-body interactions [21] have been proposed. Despite such debates, the CDW mechanism has been most widely believed to drive the observed MI transition [9,[13][14][15][16][17][18]. It is noted that the CDW formation invokes the strong coupling between lattice vibrations and electrons near the Fermi level E F , caused by Fermi surface nesting with a nesting vector 2k F = π/a x (a x : the 4×1 lattice constant along the In wires) [9,22]. The resulting Peierls dimerization was believed to occur on each chain, and the two dimerized chains further interact with each other, leading to a coupled double Peierls-dimerized chain model [23] [see Fig. 1(b)].Regarding the phase transition of the In/Si(111) system, there are still some unsettled issues. Although it is well established that the structural model of the 8×2 phase is constituted by the basic building block of In ...

Efficient PAW‐based bond strength analysis for understanding the In/Si(111)(8 × 2) – (4 × 1) phase transition

“…Thereby we confirm that the bonds proposed recently for a SSH model of the surface are indeed instrumental for the phase change. Conversely, taking advantage of the performance of the present approach, we find drastic bond strength changes on phase transition, in agreement with the interpretation recently proposed by Kim and Cho …”

“…The drastic bond strength changes calculated here between the 8 × 2 and the 4 × 1 phase, conversely, supports the more chemical interpretation of the phase transition in terms of bond‐breaking and bond‐formation …”

“…Above about 120 K the In atoms arrange in metallic zigzag chains with 4 × 1 symmetry, see Figure . Whether or not this insulator‐metal transition can be classified as Peierls instability driven or not, and if it is a first‐ or second‐order phase transition is controversially discussed up to now …”

“…The reduction of the In bonding configuration to these bonds allows for decoupling the nanowire system into three independent chains with Su–Schrieffer–Heeger (SSH) Hamiltonians that suggest to interpret the 4 × 1 → 8 × 2 phase change as first‐order grand canonical Peierls transition . Kim and Cho, conversely, stress the importance of bond‐breaking and bond‐making processes during the metal‐insulator transition of the indium atom wires and classify the latter as an exothermic surface reaction.…”

“…The mechanism and driving force for this phase transition has been controversially discussed since it's discovery. In particular, it is presently discussed whether the phase transition is of order‐disorder or displacive type, and whether it can be characterized as Peierls transition or should rather be addressed as an exothermic reaction with the consecutive bond‐breaking and bond‐making processes . The investigation of these questions will certainly profit from a quantitative investigation of the relevant bond strengths.…”