Tilt-Engineered Molecular-Scale Selector for Enhanced Learning in Artificial Neural Networks

There have been studies of understanding and controlling the charge transport through molecular junction wince it gives fundamental grounds that helps to gain a novel idea to realize numerous potential molecular functional devices. [1,2] There are some key factors considered in molecular charge transport, such as molecular barrier which is directly related to molecular orbital levels, coupling strength, asymmetric factor, and molecular configuration and conformation. Much research has been reported to obtain deeper insights of such key factors under certain controlled conductions with the effect of the external stimuli applied on the molecular junction such as light irradiation [3], electrochemical gating [4], electrical field [5], and mechanical stress such by tip loading force [6]. Molecular heterojunction with two-dimensional (2D) semiconductor provides various key factors for the charge transport through the heterojunction; however, the effect of molecular tilt configuration on the molecular heterojunction have never been studied. Here, we present the effect of molecular tilt configuration on the molecule/mono-layer (1_L) 2D semiconductor heterojunction by tip loading force with conductive atomic force microscopy (C-AFM) technique. With various tip loading force (1-30 nN), the molecular tilt configuration affects differently for the three different interfaces in Au/molecule/1_L-2D semiconductor/Au heterojunction that the transition voltage spectroscopy (TVS), which is a conventional tool to analyze charge transport through molecular junction, rectification ratio (<i>RR</i>), and pattern recognition accuracy is simulated with MNIST data images. Molecule/2D semiconductor heterojunction system has been reported as molecular diode (with <i>RR</i>) and rectifying selector and it is a novel platform owning so much various potential in, that understanding the charge transport by the effect of external stress on the heterojunction will help developing such future molecular scale selector implementation for the crossbar array architecture.<br/>[1] L. A. Bumm, J. J. Arnold, M. T. Cygan, T. D. Dunbar, T. P. Burgin, L. Jones, D. L. Allara, J. M. Tour, P. S. Weiss, <i>Science</i> 271, 1705 <b>(</b><b>1996)</b>,.<br/>[2] H. Song, Y. Kim, Y. H. Jang, H. Jeong, M. A. Reed, T. Lee, <i>Nature</i> 462, 1039 <b>(</b><b>2009)</b>.<br/>[3] G. Reecgt, C. Lotze, D. Sysoiev, T. Huhn, K. J. Franke, <i>ACS Nano</i> 10, 10555 <b>(2016)</b>.<br/>[4] R. A. Wong, Y. Yokta, M. Wakisaka, J. Inukai, Y. Kim<i>, Nat. </i><i>Commun.</i> 11, 4194 <b>(2020)</b>.<br/>[5] J. Shin, S. Yang, Y. Jang, J. S. Eo, T.-W. Kim, T. Lee, C.-H. Lee, G. Wang, <i>Nat. Commun.</i> <i>11</i>, 1 <b>(2020)</b>.<br/>[6] G. Wang, T. -W. Kim, G. Jo, T. Lee, <i>J. Am. </i><i>Chem. Soc.</i> 131, 5980 <b>(2009)</b>.