LUX-ZEPLIN

Abstract:

Metal halide perovskite (MHP) semiconductors have driven a revolution in optoelectronic technologies over the last decade, in particular for high-efficiency photovoltaic applications. Low-dimensional MHPs presenting electronic confinement have promising additional prospects in light emission and quantum technologies. However, the optimisation of such applications requires a comprehensive understanding of the nature of charge carriers and their transport mechanisms. This study employs a combination of ultrafast optical and terahertz spectroscopy to investigate phonon energies, charge-carrier mobilities, and exciton formation in 2D (PEA)₂PbI₄ and (BA)₂PbI₄ (where PEA is phenylethylammonium and BA is butylammonium). Temperature-dependent measurements of free charge-carrier mobilities reveal band transport in these strongly confined semiconductors, with surprisingly high in-plane mobilities. Enhanced charge-phonon coupling is shown to reduce charge-carrier mobilities in (BA)₂PbI₄ with respect to (PEA)₂PbI₄. Exciton and free charge-carrier dynamics are disentangled by simultaneous monitoring of transient absorption and THz photoconductivity. A sustained free charge-carrier population is observed, surpassing the Saha equation predictions even at low temperature. These findings provide new insights into the temperature-dependent interplay of exciton and free-carrier populations in 2D MHPs. Furthermore, such sustained free charge-carrier population and high mobilities demonstrate the potential of these semiconductors for applications such as solar cells, transistors, and electrically driven light sources.

Abstract:

The CRESST experiment employs cryogenic calorimeters for the sensitive measurement of nuclear recoils induced by dark matter particles. The recorded signals need to undergo a careful cleaning process to avoid wrongly reconstructed recoil energies caused by pile-up and read-out artefacts. We frame this process as a time series classification task and propose to automate it with neural networks. With a data set of over one million labeled records from 68 detectors, recorded between 2013 and 2019 by CRESST, we test the capability of four commonly used neural network architectures to learn the data cleaning task. Our best performing model achieves a balanced accuracy of 0.932 on our test set. We show on an exemplary detector that about half of the wrongly predicted events are in fact wrongly labeled events, and a large share of the remaining ones have a context-dependent ground truth. We furthermore evaluate the recall and selectivity of our classifiers with simulated data. The results confirm that the trained classifiers are well suited for the data cleaning task.Comment: 12 pages, 8 figures, 6 table