Silicon - single molecule - silicon circuits

Abstract

In 2020, silicon - molecule - silicon junctions were fabricated and shown to be on average one third as conductive as traditional junctions made using gold electrodes, but in some instances to be even more conductive, and significantly 3 times more extendable and 5 times more mechanically stable. Herein, calculations are performed of single-molecule junction structure and conductivity pertaining to blinking and scanning-tunnelling-microscopy (STM) break junction (STMBJ) experiments performed using chemisorbed 1,6-hexanedithiol linkers. Some strikingly different characteristics are found compared to analogous junctions formed using the metals which, to date, have dominated the field of molecular electronics. In the STMBJ experiment, following retraction of the STM tip after collision with the substrate, unterminated silicon surface dangling bonds are predicted to remain after reaction of the fresh tips with the dithiol solute. These dangling bonds occupy the silicon band gap and are predicted to facilitate extraordinary single-molecule conductivity. Enhanced junction extendibility is attributed to junction flexibility and the translation of adsorbed molecules between silicon dangling bonds. The calculations investigate a range of junction atomic-structural models using density-functional-theory (DFT) calculations of structure, often explored at 300 K using molecular dynamics (MD) simulations. These are aided by DFT calculations of barriers for passivation reactions of the dangling bonds. Thermally averaged conductivities are then evaluated using non-equilibrium Green's function (NEGF) methods. Countless applications through electronics, nanotechnology, photonics, and sensing are envisaged for this technology.

Fig. 1. Sketches of (a) the blinking experiment, and (b) the STMBJ experiment. Blinking experiments are performed on pre-formed regular SAMs (at ca. 75% coverage), bringing up an STM tip that is assumed to be flat on the atomic scale, detecting single-molecule junctions forming and breaking. STMBJ experiments involving crashing an STM tip into the substrate, and withdrawing it until first single-atom-wide silicon–silicon junctions form and then further until single molecules from solution bridge the formed gap.

Fig. 2. Modelling the blinking experiment. (a) 2D (3 × 3) model of two flat Si(111)–H surfaces spanned by S(CH₂)₆S at 1 : 9 coverage; bulk silicon is emulated by H-terminating the outer layer of each silicon slab, with this and the next inner layer frozen at bulk coordinates. The vertical height inside the junction is taken to be the distance z from the uppermost frozen atom in the lower electrode. The junction length is taken to be the vertical extent, d = 24.6 Å, of the region containing optimised atoms. Si-brown, S-yellow, C-cyan, H-white. (b) PDOS (in (eV Å)⁻¹) from NEGF calculations of the conductivity, showing the relative density of electronic states (colour coded) as a function of the vertical height z inside the junction and the electron energy E. See also ESI Fig. S1.

Fig. 3. Modelling the STMBJ experiment assuming regular H-terminated silicon tips form on each electrode. (a) 2D (3 × 3) model of two flat Si(111)–H surfaces with two added H-terminated Si layers in the shape of a tip, spanned by S(CH₂)₆S, at d = 36.6 Å; Si-brown, S-yellow, C-cyan, H-white. (b) Conductivity at zero voltage evaluated along a 800 fs MD trajectory for P-type silicon. (c) Average energy of each MD simulation. (d) Conductivity at zero voltage for P-type and N-type silicon for static geometries (T = 0 K) and averaged over T = 300 K MD trajectories; the shaded region shows the observed conductance found throughout extensions of 3–10 Å for P-type Si at 300 K. See ESI Fig. S2 and S3 for more information.

Fig. 4. Modelling the STMBJ experiment, using regular unterminated silicon tips on each unterminated electrode. (a) 2D (3 × 3) model for tips spanned by S(CH₂)₆S at d = 37.37 Å; Si-brown, S-yellow, C-cyan, H-white. (b) Conductivity at zero voltage for P-type and N-type silicon at 0 K; the shaded region shows the observed conductance found throughout extensions of 3–10 Å for P-type Si. See ESI Fig. S4 for more information. (c) Comparison of the transmission for P-type Si, as a function of electron energy from the Fermi energy, at d = 32.17 Å, from this unterminated series to that for the analogous terminated structure (Fig. 3, ESI Fig. S3).

Fig. 5. Modelling the STMBJ experiment, following MD at 300 K on regular unterminated silicon tips made by simulating a tip-surface crash and withdrawal (see ESI Fig. S5). (a) Conductivity at zero voltage for P-type and N-type silicon at 300 K; the shaded region shows the observed conductance found throughout extensions of 3–10 Å for P-type Si. (b) 2D (3 × 3) model for tips spanned by S(CH₂)₆S; Si-brown, S-yellow, C-cyan, H-white. See ESI Fig. S6 for more information.

Fig. 6. Modelling the STMBJ experiment, following MD at 300 K on SAMs made by simulating a tip-surface crash and withdrawal and then chemisorption of HS(CH₂)₆SH solute molecules, with one molecule bridging top and bottom. (a) 2D (3 × 3) model for tips spanned by S(CH₂)₆S; Si-brown, S-yellow, C-cyan, H-white. (b) Conductivity variations at zero voltage for P-type and N-type silicon at 300 K along a MD trajectory. (c) Conductivity at zero voltage for P-type silicon at 0 K. (d) Conductivity at zero voltage for N-type silicon at optimised geometries (T = 0 K). (e) Conductivity at zero voltage for P-type and N-type silicon averaged over T = 300 K MD trajectories. The shaded region shows the observed conductance found throughout extensions of 3–10 Å for P-type Si. See ESI Fig. S5 and S9–S12 for more information.