Abstract
Simulating the movements of individual atoms allows us to look at
and investigate the physical processes that happen in an experiment.
In this thesis I use simulations to support and improve experimental
studies of breaking gold nano-junctions.
By using molecular dynamics to study gold nanowires, I can investigate
their breaking forces under varying conditions, like stretching
rate or temperature. This resolves a confusion in the literature, where
the breaking forces of two different breaking structures happen to
coincide.
The correlations between the rupture and reformation of a gold
junction are studied, and when the same analysis methods are used
on both simulated and experimental results, we see a striking similarity.
The simulations are used to understand what atomic movements
could happen in the experiments.
Finally, I develop a classification method, using conductance traces
as input, to predict the structure of a gold junction just as it breaks.
This method is based on artificial neural networks and can be used
on experimental data, even when it is trained purely on simulated
data. The method is extended to other types of experimental traces,
where it is trained without the use of simulated data.
and investigate the physical processes that happen in an experiment.
In this thesis I use simulations to support and improve experimental
studies of breaking gold nano-junctions.
By using molecular dynamics to study gold nanowires, I can investigate
their breaking forces under varying conditions, like stretching
rate or temperature. This resolves a confusion in the literature, where
the breaking forces of two different breaking structures happen to
coincide.
The correlations between the rupture and reformation of a gold
junction are studied, and when the same analysis methods are used
on both simulated and experimental results, we see a striking similarity.
The simulations are used to understand what atomic movements
could happen in the experiments.
Finally, I develop a classification method, using conductance traces
as input, to predict the structure of a gold junction just as it breaks.
This method is based on artificial neural networks and can be used
on experimental data, even when it is trained purely on simulated
data. The method is extended to other types of experimental traces,
where it is trained without the use of simulated data.
| Original language | English |
|---|
| Publisher | Department of Chemistry, Faculty of Science, University of Copenhagen |
|---|---|
| Number of pages | 126 |
| Publication status | Published - 2017 |