Autonomous Single-Molecule Manipulation Based on Reinforcement Learning

Bernhard Ramsauer; Grant J. Simpson; Johannes J. Cartus; Andreas Jeindl; Victor García-López; James M. Tour; Leonhard Grill; Oliver T. Hofmann

doi:10.1021/acs.jpca.2c08696

Autonomous Single-Molecule Manipulation Based on Reinforcement Learning

Bernhard Ramsauer, Grant J. Simpson, Johannes J. Cartus, Andreas Jeindl, Victor García-López, James M. Tour^*, Leonhard Grill^*, Oliver T. Hofmann^*

^*Corresponding author for this work

Institute of Solid State Physics (5130)

Research output: Contribution to journal › Article › peer-review

Abstract

Building nanostructures one-by-one requires precise control of single molecules over many manipulation steps. The ideal scenario for machine learning algorithms is complex, repetitive, and time-consuming. Here, we show a reinforcement learning algorithm that learns how to control a single dipolar molecule in the electric field of a scanning tunneling microscope. Using about 2250 iterations to train, the algorithm learned to manipulate the molecule toward specific positions on the surface. Simultaneously, it generates physical insights into the movement as well as orientation of the molecule, based on the position where the electric field is applied relative to the molecule. This reveals that molecular movement is strongly inhibited in some directions, and the torque is not symmetric around the dipole moment.

Original language	English
Pages (from-to)	2041-2050
Number of pages	10
Journal	Journal of Physical Chemistry A
Volume	127
Issue number	8
DOIs	https://doi.org/10.1021/acs.jpca.2c08696
Publication status	Published - 2 Mar 2023

ASJC Scopus subject areas

Physical and Theoretical Chemistry

Cooperations

NAWI Graz

Access to Document

10.1021/acs.jpca.2c08696Licence: CC BY 4.0

acs.jpca.2c08696Final published version, 4.14 MBLicence: CC BY 4.0

Cite this

@article{5886b0b87d144bca9675a33f406ef5a1,

title = "Autonomous Single-Molecule Manipulation Based on Reinforcement Learning",

abstract = "Building nanostructures one-by-one requires precise control of single molecules over many manipulation steps. The ideal scenario for machine learning algorithms is complex, repetitive, and time-consuming. Here, we show a reinforcement learning algorithm that learns how to control a single dipolar molecule in the electric field of a scanning tunneling microscope. Using about 2250 iterations to train, the algorithm learned to manipulate the molecule toward specific positions on the surface. Simultaneously, it generates physical insights into the movement as well as orientation of the molecule, based on the position where the electric field is applied relative to the molecule. This reveals that molecular movement is strongly inhibited in some directions, and the torque is not symmetric around the dipole moment.",

author = "Bernhard Ramsauer and Simpson, {Grant J.} and Cartus, {Johannes J.} and Andreas Jeindl and Victor Garc{\'i}a-L{\'o}pez and Tour, {James M.} and Leonhard Grill and Hofmann, {Oliver T.}",

note = "Publisher Copyright: {\textcopyright} 2023 The Authors. Published by American Chemical Society.",

year = "2023",

month = mar,

day = "2",

doi = "10.1021/acs.jpca.2c08696",

language = "English",

volume = "127",

pages = "2041--2050",

journal = "Journal of Physical Chemistry A",

issn = "1089-5639",

publisher = "American Chemical Society",

number = "8",

}

TY - JOUR

T1 - Autonomous Single-Molecule Manipulation Based on Reinforcement Learning

AU - Ramsauer, Bernhard

AU - Simpson, Grant J.

AU - Cartus, Johannes J.

AU - Jeindl, Andreas

AU - García-López, Victor

AU - Tour, James M.

AU - Grill, Leonhard

AU - Hofmann, Oliver T.

PY - 2023/3/2

Y1 - 2023/3/2

N2 - Building nanostructures one-by-one requires precise control of single molecules over many manipulation steps. The ideal scenario for machine learning algorithms is complex, repetitive, and time-consuming. Here, we show a reinforcement learning algorithm that learns how to control a single dipolar molecule in the electric field of a scanning tunneling microscope. Using about 2250 iterations to train, the algorithm learned to manipulate the molecule toward specific positions on the surface. Simultaneously, it generates physical insights into the movement as well as orientation of the molecule, based on the position where the electric field is applied relative to the molecule. This reveals that molecular movement is strongly inhibited in some directions, and the torque is not symmetric around the dipole moment.

AB - Building nanostructures one-by-one requires precise control of single molecules over many manipulation steps. The ideal scenario for machine learning algorithms is complex, repetitive, and time-consuming. Here, we show a reinforcement learning algorithm that learns how to control a single dipolar molecule in the electric field of a scanning tunneling microscope. Using about 2250 iterations to train, the algorithm learned to manipulate the molecule toward specific positions on the surface. Simultaneously, it generates physical insights into the movement as well as orientation of the molecule, based on the position where the electric field is applied relative to the molecule. This reveals that molecular movement is strongly inhibited in some directions, and the torque is not symmetric around the dipole moment.

UR - http://www.scopus.com/inward/record.url?scp=85147814649&partnerID=8YFLogxK

U2 - 10.1021/acs.jpca.2c08696

DO - 10.1021/acs.jpca.2c08696

M3 - Article

C2 - 36749194

AN - SCOPUS:85147814649

SN - 1089-5639

VL - 127

SP - 2041

EP - 2050

JO - Journal of Physical Chemistry A

JF - Journal of Physical Chemistry A

IS - 8

ER -

Autonomous Single-Molecule Manipulation Based on Reinforcement Learning

Abstract

ASJC Scopus subject areas

Cooperations

Access to Document

Other files and links

Fingerprint

Cite this