Electron Transport Mechanism in Single-Molecule Junctions
概要
With the rapid expansion of semiconductor technologies, the characteristic size of transistors has been greatly shrunken into < 20nm and is progressively approaching the miniaturization limits of Si-based technology. For years, researchers have been inspired by the vista of utilizing molecules as functional units in electronic circuit.1–3 Unlike conventional electronic devices that controlling electrons within solid crystals structures, in single-molecule devices, electrons are transported by electronic states generated from single molecule structures. Single-molecule devices thus would diminish dimensions beyond classic potential barriers of convention semiconductor devices. As long ago as 1974, Aviram and Ratner had foreseen that single molecules would be employed as circuit units in electronic devices (Aviram-Ratner rectifier). Today, single molecule electronics symbolizes the ultimate goal that we are marching on.
The ability to calculate and control electron transport through a molecule is one of the major challenges in fabricating electronic devices. Substantial efforts have been assigned to integrating nanometer-sized molecules to macroscopic circuits and reliable controlling since the first current-voltage (I-V) characteristic measurements of benzene-1,4,-dithiol (BDT) using a break junction that mechanically controlled. Two approaches for electrode bridges with molecules are usually available. One is to create top-contact junction which includes scanning probe microscopy (STM or CAFM), cross wire junctions, mercury drop electrodes and nanopore.4–6 The second approach is building vertical nanogap metal/molecule/metal (MMM) junctions.7,8 Various fabrication techniques have been used in building nanogap junctions including MCBJ, electrochemical deposition of nanosized junction, electromigration breakdown junction.9,10 Among these methods, MCBJ technique, which was originally designed in 1985 by J. Moreland and J. W. Ekin and further improved by C. Muller in 1988~1992,11,12 provides two mechanically-stable electrodes within an adjustable gap, and bring two uncontaminated atomic nanoelectrodes to characterize single molecules.
Hence, crafting an authentic nano-sized pair of electrodes is important in mechanical break junction experiments. Different methods, such as mechanical cutting, electron beam lithography, and electrochemical etching were investigated to refine the measuring results under the MCBJ’s structure. Finally, electron beam lithography was overmatching other methods by its atomic-scale resolution, which guarantees a nano-sized electrode apex and laid a solid foundation for a high-stability MCBJ chip. Once the MCBJ chips were well fabricated, single molecules were inserted into the gap of electrodes and electron transport properties of molecule junction can be acquired after repetitive tensile break-up of nano-junctions by poking them into and out of contact.
After the Au monoatomic chain ruptured during the extension process, plateau-like conductance features can be detected fortuitously, which is naturally considered as signatures of established Au-Molecule-Au structure. Unavoidably, sorting out involved noise during the measurements was the key issue for any kind of nanodevices. For single molecule junctions, what noise properties bring is not only the hints on dynamic electron transport processes but also affects devices’ performance and the reliability. Typical noise such as Generation- Recombination (G-R) noise was initially observed in MCBJ test, but root causes of G-R noise and of electronic characteristics of integrated molecules had never been revealed.13,14 Although simple and effective to measure electron transport by applying denoising hardware, the stochastic nature of the contact mechanics inevitably involves random changes in the data through various reasons, such as direct tunneling through electrode gap and fluctuated binding motifs of unpredictable molecules. For obtaining the most likely conductance value, classical methods records thousands of repeated traces from automatic measurements and binning into conductance histograms.8,15–17 By marking molecule dependent peaks in the wide distribution of conductance, one may indeed statistically estimate the averaged bridged molecule conductance (GM) of a specific kind at the price of neglecting other subtle but probably meaningful molecular features in the histograms. In Chapter 3, we are trying to solve this dilemma by proposing a nonparametric hybrid machine learning approach. Our algorithm can straightforwardly monitor single molecular conductance behavior arose over time via a simple two-step strategy: Analysis and examining/pre-screening through a denoised classification by means of Grid- based DBSCAN (Density-based clustering) and successive reconstructing of dynamic histograms from classified trace groups.
Thereafter, the electron transport route of the alkanedithiol junction was investigated. The research on electron transport through a junction of metal/molecule/metal(MMM) has been an active area of research for over a decade.1-5 Today, single molecule measurements is increasingly feasible thanks to the break junction methods wherein most electron transport mechanics through MMM structures were tested in organic solvents.16 From practical viewpoints, these measurements are preferred to be conducted in dry environments considering that any electronic components would be used in ambient conditions. Despite the fact that a variety of laboratory experiments have been carried out in non-liquid environments. Taking the advantages of the machine-learning classifying algorithm, we systematically investigated the conductance behaviors of alkanedithiols in vacuum and evaluated solvents’ influence on detecting results. In Chapter 4, we inspected the details of the molecular junction formation mechanism using MCBJ's microfabricated to gather single-molecule signatures during conductance measurements with an emphasis on lower molecular concentration limits.
Finally, a more complicated environment at high temperature (range from 300K to 420K) was implemented on single alkanedithiol measurements for querying potential transport route changes at short tunneling distance beyond moderate temperature. Although there a lot of theoretical study about thermoelectricity in molecular junctions, the single-molecule charge transport mechanism at high temperature has not been detailed investigated experimentally. According to the Simmons model11,18 under an assumption of symmetric coupling strength, a large offset between the electrode Fermi level and the HOMO of the bridged molecule makes the non-resonant tunneling model (temperature-independent) sufficient to describe the charge transport via short alkanedithiol junctions. However, temperature-dependent conducting behaviors were often observed in short alkanedithiol experiments and its mechanisms are still under debating and need to be clarified. This thesis (in Chapter 5) shed light on this issue by contributing to multi-roles of intra-molecular and inter-molecular charge transport mechanism of short molecules at high temperatures.
The thesis is outlined as follows: First I will introduce the theory and a description of the setup and materials used in single molecule measurement, see Chapter 2. This will be followed by the designing of machine-learning methodology and its performance in dissecting conductance behavior of single molecule junctions evolved over time, see Chapter 3. The results and comprehensive descriptions of conductance behaviors of alkanedithiols in vacuum and the study of charge transport mechanism at high temperatures are given in Chapter 4 and Chapter 5, respectively. Finally, we conclude the thesis with a summary and give an outlook on possible research that may be pursued on the basis of this work.