New Preprint about Unsupervised Machine Learning on Diverse Measurement Data

Already since a few years our research group is exploring approaches for the automated and – most importantly – unbiased analysis of experimental data. In particular Anton Vladyka and Maria El Abbassi have been driving forces in that respect.

My colleague Mickael Perrin has taken this further by developing a generalised approach for the analysis of various kinds of measurement data and performing an extensive benchmark of a large range of algorithms. Together, we have extended this to my particular field of interest, Raman spectroscopy.

Our discussion of the various methods for dimensionality reduction, clustering and cluster validation are available in the following preprint: https://arxiv.org/abs/2004.14271
We are happy about any feedback!



Raman Spectroscopy Confirms Nanoscale Alignment of GNRs

Graphene nanoribbons (GNRs) are a novel class of quasi-onedimensional conductors which are just one atom thick, few atoms (~ 1 nm) wide and currently a few tens of nanometers long. After their synthesis on gold substrates in ultra-high vacuum one obtains films of GNRs with a size of a few square millimeters. The individual GNRs can be randomly oriented or aligned in parallel, depending on the surface structure of the substrate. In the latter case, this allows for the development of polarisation-dependent optical methods to investigate the properties of GNR-films on a larger scale.
Raman spectroscopy is a relatively well-established technique that enables the determination of direction and degree of alignment of GNRs and other elongated nanostructures – also via the polarization of the laser used for excitation and on a variety of different substrates.
We have therefore used this method to characterize 7-atom wide GNRs on their gold substrate and after transfer to glass. Afterwards, our collaborators from the University of Berkeley could show that they can use polarization-dependent absorption spektroscopy on just a monolayer of GNRs to verify their optical bandgap. Congrats Sihan!

Sihan Zhao, Gabriela Borin Barin, Ting Cao, Jan Overbeck, Rimah Darawish, Tairu Lyu, Steve Drapcho, Sheng Wang, Tim Dumslaff, Akimitsu Narita, Michel Calame, Klaus Müllen, Steven G. Louie, Pascal Ruffieux, Roman Fasel, and Feng Wang. Optical Imaging and Spectroscopy of Atomically Precise Armchair Graphene Nanoribbons. Nano Letters 2020, 20 (2), 1124-1130 DOI: 10.1021/acs.nanolett.9b04497



Optimized Chip-Design for Raman Spectroscopy on Graphene Nanoribbons

Graphene nanoribbons (GNRs) are the current focus of my work, with Raman spectroscopy playing a major role in the characterization of their properties and material quality.

GNRs are synthesized from precursor molecules under ultra-high vacuum conditions and have to be transferred to silicon chips with nanostructured electrods for electrical characterization. As the fabrication of these chips is quite time-intensive we want to routinely make sure that the quality of GNRs to be transferred is as expected and that we can assess this after transfer and independent of their electrical properties. Towards that end, we employ Raman spectroscopy that allows us to obtain precise information about the structure of the GNRs based on their spectroscopic fingerprint – in particular also on insulating substrates and without the need for vacuum.

To be able to routinely apply this technique on all chips, we employ a layered structure of materials optimized for constructive interference which maximizes the Raman signal. This structure combines the technological benefits of using widely available silicon chips without disturbance by their background signal. This allows us to detect particularly informative Raman modes that are otherwise masked by that background.

We have revcently published a detailed desciption of this approach alongside optimized measurement routines:

Overbeck, J., Borin Barin, G., Daniels, C., Perrin, M.L., Liang, L., Braun, O., Darawish, R., Burkhardt, B., Dumslaff, T., Wang, X., Narita, A., Müllen, K., Meunier, V., Fasel, R., Calame, M. and Ruffieux, P. (2019), Optimized Substrates and Measurement Approaches for Raman Spectroscopy of Graphene Nanoribbons. Phys. Status Solidi B, 256: 1900343. doi:10.1002/pssb.201900343
The preprint can be found on arXiv: https://arxiv.org/abs/1907.01797

Length-Dependent Mode in Graphene Nanoribbons

Graphene Nanoribbons (GNRs) are quasi-onedimensional materials based on carbon. They can be synthesized from suitable molecular precursors under ultra-high vacuum conditions. Thanks to this approach, it is possible to create GNRs with a width and edge-structure defined to the last atom. Their length, however, is statistically distributed and influenced by the parameters of on-surface synthesis. Importantly, it is a key parameter for their integration into functional electronic circuits that would allow to characterize their novel properties. In this publication we show how Raman spectroscopy can be used to characterize the length of GNRs and investigate the interaction with their environment.

https://pubs.acs.org/doi/10.1021/acsnano.9b05817

Controlled Straining of Heterostructures

Graphen is a material that has attracted interest for several reasons: it is just 1 atom thick, electrically conducting and nearly transparent. What is more, graphene in its pristine, low defect form is extremely stable – at least considering it’s volume or mass. This has led to fascinating comparisons.

Colorized Scanning Electron Micrograph of a freely suspended graphene membrane (light blue) between two platin electrodes (grey). The electrodes are separated from the golden bottom gate by a 700 nm-thick polymer layer. J.O. 2016.

In order to investigate the mechanical properties of graphene it is advantagous to perform emasurements in a freely suspended geometry (see picture above), as this allows to reduce uncertainties resulting from inhomogenous attachment to the support layer.
For the investigation of its electric properties as a function of strain, however, there is a strong benefit from encapsulating graphene between two layers of isolating (hexagonal) boron nitride, as we report in a recent publication. Congratualtions, Lujun!

Wang, L.; Zihlmann, S.; Baumgartner, A.; Overbeck, J.; Watanabe, K.; Taniguchi, T.; Makk, P.; Schönenberger, C. (2019): In Situ Strain Tuning in hBN-Encapsulated Graphene Electronic Devices. Nano letters 19, 4097–4102; doi: 10.1021/acs.nanolett.9b01491.

Excursion into Political Speeches

Listening to the morning radio I made an interesting discovery: A feature about speeches given by Barack Obamas throughout his presidency. What I found particularly striking was the analysis into the tone of his speeches on gun legislation. It was pointed out that over time – as he had to comfort families of gun violence time and again – the level of hope faded and a certain amout of fatigue and resignation became apparent. I had to agree with this from my own listening experience but interestingly this can also be objectified, as discussed here. (German audio).

Große Reden, große Redner? (2/3) – Obamas Reden: Politik als Storytelling

Results of my Diploma Thesis Published in ACS Nano

In April 2013 I finished my Diploma thesis at TU Munich in the group of Prof. Alexander Holleitner . The results were published in September 2015 in ACS Nano:
Photocurrents in a Single InAs Nanowire/Silicon Heterojunction

Andreas Brenneis, Jan Overbeck, Julian Treu, Simon Hertenberger, Stefanie Morkötter, Markus Döblinger, Jonathan J. Finley, Gerhard Abstreiter, Gregor Koblmüller, and Alexander W. Holleitner ACS Nano 2015 9 (10), 9849-9858, doi: 10.1021/acsnano.5b03017

The topic is the investigation of electrical currents in a semiconductor under illumination – similar to a solar cell. This particular semiconductor is a single nanowird made of the III-V semiconductor indium arsenide. Notably, this could be vertically integrated on commercially available silicon substrates. The direct band gap of indium arsenide makes this material a promising candidate for future optoelectronic circuits.