286847 GPU-Enabled Simulations of Size Effects On the Elongation and Rupture of Metallic Nanowires Using Many-Body Potentials

Wednesday, October 31, 2012: 10:10 AM
415 (Convention Center )
William R. French1, Christopher R. Iacovella1 and Peter T. Cummings2, (1)Chemical and Biomolecular Engineering, Vanderbilt University, Nashville, TN, (2)Center of Nanophase Materials Sciences, Oak Ridge National Laboratory, Oak Ridge, TN

Understanding the formation of atomic scale point contacts via the mechanical deformation of metallic nanowires is important to the development of future nano- and molecular-electronic devices.  Molecular simulation has been used extensively to explore this behavior; however, simulations often use smaller system sizes, higher strain rates, and fewer independent measurements than experiment. Most simulations have relied on many-body potentials, such as the embedded atom model (EAM) and second-moment approximation to the tight-binding potential (TB-SMA). The use of many-body potentials, as opposed to pair potentials, is critical for the accurate description of surface properties, defects, and elastic moduli of metallic systems.1 The EAM potential has recently been ported to run on graphical processing units (GPUs) with great success.2 However, previous work from our group has shown that the TB-SMA potential provides a more accurate description of the structure and energy of the elongation of nanowires than EAM, when compared to density functional theory.3

Here, we discuss our recent porting of the TB-SMA potential to the GPU by extending the highly optimized object-oriented many-particle dynamics (HOOMD) package.4  We find that the GPU-based TB-SMA potential provides significant speedups relative to parallel CPU-based simulations for nanowire systems.  As a test of this work, we use the GPU-enabled TB-SMA potential to investigate the elongation and rupture of Au nanowires over a range of parameters.  The speed-up associated with the GPU enables us to more closely model experimental conditions, with larger wires, slower elongation rates, and more total state points available for study.

[1] F. Cleri and V. Rosta, Tight-binding potentials for transition metals and alloys, Physical Review B, 1993, 48(1) 22-33.

[2] I. V. Morozov, A. M. Kazennova, R.G. Bystryia, G.E. Normana, V.V. Pisareva, and V.V. Stegailova. Molecular dynamics simulations of the relaxation processes in the condensed matter on GPUs Computer Physics Communications 182(9): 1974-1978, 2011.

[3] Q. Pu, Y. Leng, L. Tsetseris, H. S. Park, S. T. Pantelides, and P. T. Cummings, “Molecular Dynamics Simulations of Stretched Gold Nanowires: the Relative Utility of Different Semiempirical Potentials,” J. Chem. Phys., 2007, 126, 144707.

[4] J. A. Anderson, C. D. Lorenz, and A. Travesset, “General Purpose Molecular Dynamics Simulations Fully Implemented on Graphics Processing Units,” J. Comp. Phys., 2008, 227(10), 5342-5359. http://codeblue.umich.edu/hoomd-blue

Extended Abstract: File Not Uploaded