Single Molecular Junctions and their Conductance Properties Yu Hui Tang(NCU), Jay Lin (NCU), Leo Zhang (NCU), Ji Su Song (SUNY) ConclusionReferencesAcknowledgements Introduction Theory and Methods Objective Results Single molecular junctions do not occur in nature, so they must be isolated from clusters. The two leads of a junction can be slowly pulled and relaxed to slowly rip off connected molecules of the channel. We can look at the density of states as molecules are disconnected so safely assume when we have a single molecular junction. Once we achieve this state, we can run a current through the junction and study the conductance properties. We were assigned different parts of the research project to look after. My role consisted of debugging a code for a program that models the pulling and relaxing process of the junction. Every cycle of stretching is followed by a relaxation of the molecules, which we accomplish by using the BFGS method. Our code did not converge the BFGS steps for some of the cycles, which proved to be a problem. I was given the task to fix this problem. The program to simulate the process of obtaining a single molecular junction seems to work now without much problems. I hope that I have done my part to further the research of Dr. Tang. With the functional program, the other students can proceed with their part of the program. Research in this field can be of interest in the scientific community. By studying the properties of certain materials, we can create much more efficient methods of data storage. This field is a pretty recent development and has much room for scientific exploration. Bingqian Q. Xu, Xiulan Li, Xiaoyin Xiao, Hiroshi Sakaguchi, and Nongjian Tao. Electromechanical and conductance switching properties of single oligothiophene molecules. Nano Letters, (2005) Vol. 5, No. 7, Yu-Hui Tang, V.M.K. Bagci, Jing-Han Chen, and Chao- Cheng Kaun. Conductance of Stretching Oligothiophene Single-Molecule Junctions: A First Principles Study. The Journal of Physical Chemistry (2011) Figure 1: Example of single molecular junction (Benzene ring) Figure 2: The flow of electrons due to different agendas of source and drain. The μ is the electrochemical potential. We use a program called Quantum Espresso, which has predefined methods and definitions for stretching and relaxation methods. It involves using a first principle calculation method, shown by Figure 3. The relaxation method that we chose to use is called the BFGS ( Broyden-Fletcher-Goldfarb-Shanno) method, which solves nonlinear optimization problems by looking for convergence. The BFGS method in Quantum Espresso has many definitions that can specify certain conditions. Input positions and potential Solve Schrödinger New Potential. Converged? New force ➔ New position (new input to loop) No Ye s Figure 3: First principle calculation method. The iteration process is called Self Consistent Field (SCF). If new potential does no converge, mix old and new potentials and solve again. Figure 4: Input file for Quantum Espresso Another relaxation method we could use is the damp method. This method uses damped dynamics for relaxation and can be used for constrained optimizations. Although the BFGS method is preferred, the damp method is also widely used for structural relaxation. We also include atomic positions in the input to define the channel. This definition should be very accurate because the atomic positions greatly affect the interactions between the molecules. With each stretching and relaxing cycle, the interactions are quite intricate. One of the benefits of using Quantum Espresso is that is it easy to define the crystal lattice structure of the leads. These input descriptions are shown in Figure 4. Figure 5 & 6: Results of decreasing convergence threshold. We tried to decrease the convergence threshold to increase the accuracy of the SCF convergence. Lowering the threshold would make the converging more strict. This change in the input led to worse results and did not fix the lack of BFGS steps. The damp method had similar results. The version of Quantum Espresso we use (v5.0) is a newer version with a different compiler. The problem with the relaxation has started ever since the newer compiler was used. The new compiler had optimized conditions, which were applied to reduce the running time of the program. These conditions seem to have led to a process that was “too optimized” for the input, resulting in a lack of details. In an attempt to fix the problem, we got rid of the optimizations and applied the older compiler to the newer versions of Quantum Espresso. Once we did this and ran the same input files, we got very promising results. The number of SCF cycles were always one more than the number of BFGS cycles, as it should be. The new results are shown in Figures 7 and 8. I hope that with these results, Dr. Tang may go on to do further research without too many problems with the program. Figure 7& 8: Example of fixed results