Spontaneous S-Si bonding of alkanethiols to Si(111)-H: towards Si-molecule-Si circuits

Abstract

We report the synthesis of covalently linked self-assembled monolayers (SAMs) on silicon surfaces, using mild conditions, in a way that is compatible with silicon-electronics fabrication technologies. In molecular electronics, SAMs of functional molecules tethered to gold via sulfur linkages dominate, but these devices are not robust in design and not amenable to scalable manufacture. Whereas covalent bonding to silicon has long been recognized as an attractive alternative, only formation processes involving high temperature and/or pressure, strong chemicals, or irradiation are known. To make molecular devices on silicon under mild conditions with properties reminiscent of Au-S ones, we exploit the susceptibility of thiols to oxidation by dissolved O₂, initiating free-radical polymerization mechanisms without causing oxidative damage to the surface. Without thiols present, dissolved O₂ would normally oxidize the silicon and hence reaction conditions such as these have been strenuously avoided in the past. The surface coverage on Si(111)-H is measured to be very high, 75% of a full monolayer, with density-functional theory calculations used to profile spontaneous reaction mechanisms. The impact of the Si-S chemistry in single-molecule electronics is demonstrated using STM-junction approaches by forming Si-hexanedithiol-Si junctions. Si-S contacts result in single-molecule wires that are mechanically stable, with an average lifetime at room temperature of 2.7 s, which is five folds higher than that reported for conventional molecular junctions formed between gold electrodes. The enhanced "ON" lifetime of this single-molecule circuit enables previously inaccessible electrical measurements on single molecules.

Fig. 1. (a) Molecular and SAM structures: 1 enables electrochemical studies while 2 enables STM-junction studies and 3, without a thiol, is used as a control (a) molecules are reacted with H-terminated silicon surfaces (b) via a thiyl free-radical polymerization (c), as illustrated for propanethiol, to form covalently bonded SAMs (d). (e) Si–molecule–Si junction formed from 2 using single-molecule STM-junction technology (e). S – yellow, Si – fawn, C – cyan, H – white.

Fig. 2. Cyclic voltammetry. (a) Cyclic voltammetry at different scan rates for a SAM of 1 on p-type Si(111)–H of resistivity 0.001 Ω cm. The estimated surface coverage is 2.53 × 10¹⁴ molecules cm⁻². (b) Current versus scan rate for the SAM formed of 1. The current increases linearly with scan rate, indicating a surface redox reaction. (c) Cyclic voltammetry for a SAM of 1 on Si(111)–H at 50 mV s⁻¹ on n-type phosphorus doped of resistivity 0.001 Ω cm. Coverage is dopant independent, the results leading to coverages of 2.50 × 10¹⁴ molecules cm⁻² and is comparable to that obtained on p-type boron doped silicon shown in (a). (d) Cyclic voltammetry at 50 mV s⁻¹ for SAMs of 1 on Au(111), showing very similar appearances to those reported for Si in (a) at a similar scan rate. The deduced coverages are 2.20 × 10¹⁴ molecules cm⁻², slightly less than that for the corresponding monolayers on Si(111)–H. (e) Cyclic voltammetry at 50 mV s⁻¹ for SAMs of 1 on Si(100)–H showing a similar appearances to those reported for Si(111)–H; however the surface coverage is lower and estimated at 9.3 × 10¹³ molecules cm⁻². (f) Cyclic voltammetry for SAMs of 1 on Si(111)–H prepared from fresh solid of 1 dissolved in deoxygenated DCM. Oxygen was removed by bubbling Ar for 60 min in the DCM solution containing 1 using a septum sealed vessel and a needle as a vent. The SAM reaction flask was kept under a positive pressure of Ar during the 24 h reaction time. The surface coverage is estimated to be 2.10 × 10¹³ molecules cm⁻² and is about 10% of that produced under ambient conditions.

Fig. 3. Surface characterization. (a) 10 × 10 μm² of a p-type Si(111)–H surface covered by dithiol 2 on Si(111). (b) XRR measurements of 2 on Si(111)–H for various silicon dopant levels: high p-doped (green), low n-doped (red), low p-doped (yellow) and high n-doped (blue). The symbols with error bars are the collected data and the solid lines are the fits to each data set. The data is offset for clarity (c) XPS spectra showing the absence of Si–O_x at 102–104 eV for SAMs of 2 on high p-doped (green), low p-doped (yellow), low n-doped (red) and high n-doped (blue). The data is offset for clarity. (d) S 2p narrow XPS scan of monolayers of 2 on Si (p-type, 0.001 Ω cm). Two sets of spin–orbit-split S 2p peaks (2p_1/2 and 2p_3/2, high and low binding energy, respectively), held 1.16 eV apart, are evident between 162 and 165 eV. The intensity ratio between the 3/2 and 1/2 emission is set to two, and values of full width at half maximum are 1.3 eV. The S 2p emission centred at 164.7 eV is ascribed to thiols in a R–SH configurations, and the emission 163.3 eV is associated to thiols bound to Si (RS–Si).

Fig. 4. Proposed reaction scheme for the consumption of ambient oxygen as an initiator leading to surface free-radical polymerization. (a) Known solution reactions of thiols (RSH). (b) DFT mechanism for surface SAM formation starting at low coverage. Calculations indicate that thiyl radicals (RS˙, with here R = C₃H₇) react with Si(111)–H to abstract hydrogen to form thiol physisorbed to a silicon surface radical (black dot). Reaction over a barrier then leads to chemisorption and radical regeneration. This provides initiation for a free-radical polymerization reaction that then covers the surface with adsorbate. Some critical bond lengths are shown, in Å; only one copy of the used 3 × 3 supercell is shown. These results show Gibbs free energy changes for surface reactions, while ESI Fig. S3 provides analogous electronic energy changes, with detailed internal reaction coordinate descriptions provided in ESI Fig. S6. Also, ESI Fig. S4 and S5 provide analogous energies for a model compound.

Fig. 5. DFT calculated SAM structures for 2 on Si(111)–H as a function of coverage, showing each four copies of the 3 × 3 supercell used in the calculations. (a) 1 : 9, (b) 7 : 9, and (c) 1 : 1.

Fig. 6. (a) Cartoons of the junctions studied. (b) Representative blinks for single molecules of 2 bonded to two Au(111) electrodes (blue), one Au(111) and one Si(111)–H electrode (yellow), and two Si(111)–H electrodes (red). (c) Blinking histograms for the Au–2–Au, Au–2–Si and Si–2–Si junctions. The blinking histograms were constructed by the accumulation of 300 blinks for each system. Each blinking histogram was normalised to the total number of blinks. (d) Representative current-distance traces for Si–2–Si junctions. (e) Representative current-distance traces for Au–2–Au junctions. (f) Current-distance traces histograms constructed from the accumulation of 3000 curves for each system. Each histogram was normalised to the total number of traces.