Machine Learning for Analysis of Time-Resolved Luminescence Data

Đorđević, Nikola; Beckwith, Joseph S.; Yarema, Maksym; Yarema, Olesya; Rosspeintner, Arnulf; Yazdani, Nasser; Leuthold, Juerg; Vauthey, Eric; Wood, Vanessa

doi:10.1021/acsphotonics.8b01047

Cited by 34 publications

(39 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We therefore applied machine-learning algorithms to distinguish the emission states in the measured spectra for our NPLs with minimal user bias. 26 We employed a specific implementation of the k-means clustering algorithm 27 available in the Python programming language. 28 This allowed us to sort the 1800 individual frames shown in Figure 2a, where the number of clusters was the only user input.…”

mentioning

confidence: 99%

Observation of Electron Shakeup in CdSe/CdS Core/Shell Nanoplatelets

et al. 2019

View full text Add to dashboard Cite

While ensembles of CdSe nanoplatelets (NPLs) show remarkably narrow photoluminescence line widths at room temperature, adding a CdS shell to increase their fluorescence efficiency and photostability causes line width broadening. Moreover, ensemble emission spectra of CdSe/CdS core/shell NPLs become strongly asymmetric at cryogenic temperatures. If the origin of these effects were understood, this could potentially lead to stable core/shell NPLs with narrower emission, which would be advantageous for applications. To move in this direction, we report timeresolved emission spectra of individual CdSe/CdS core/shell NPLs at 4 K. We observe surprisingly complex emission spectra that contain multiple spectrally narrow emission features that change during the experiment. With machine-learning algorithms, we can extract characteristic peak energy differences in these spectra. We show that they are consistent with electron "shakeup lines" from negatively charged trions. In this process, an electron-hole pair recombines radiatively, but gives part of its energy to the remaining electron by exciting it into a higher single-electron level. This "shakeup" mechanism is enabled in our NPLs due to strong exciton binding and weak lateral confinement of the charge carriers. Time-resolved single-photon-counting measurements and numerical calculations suggest that spectral jumps in the emission features originate from fluctuations in the confinement potential caused by microscopic structural changes on the NPL surface (e.g., due to mobile surface charges). Our results provide valuable insights into line width broadening mechanisms in colloidal NPLs.

show abstract

mentioning

confidence: 99%

Observation of Electron Shakeup in CdSe/CdS Core/Shell Nanoplatelets

et al. 2019

View full text Add to dashboard Cite

show abstract

“…Inspired by recent studies on inverse design of polymers and inorganic solids [23][24][25] , as well as on using machine learning to understand PSCs' properties [26][27][28] , we present a machine-learning framework to investigate LD organic-inorganic perovskites serving as a capping layer for MAPbI 3 . We elucidate which properties of capping layers are responsible for enhancing stability, and the underlying mechanisms whereby they work.…”

mentioning

confidence: 99%

How machine learning can help select capping layers to suppress perovskite degradation

et al. 2020

View full text Add to dashboard Cite

Environmental stability of perovskite solar cells (PSCs) has been improved by trial-and-error exploration of thin low-dimensional (LD) perovskite deposited on top of the perovskite absorber, called the capping layer. In this study, a machine-learning framework is presented to optimize this layer. We featurize 21 organic halide salts, apply them as capping layers onto methylammonium lead iodide (MAPbI3) films, age them under accelerated conditions, and determine features governing stability using supervised machine learning and Shapley values. We find that organic molecules’ low number of hydrogen-bonding donors and small topological polar surface area correlate with increased MAPbI3 film stability. The top performing organic halide, phenyltriethylammonium iodide (PTEAI), successfully extends the MAPbI3 stability lifetime by 4 ± 2 times over bare MAPbI3 and 1.3 ± 0.3 times over state-of-the-art octylammonium bromide (OABr). Through characterization, we find that this capping layer stabilizes the photoactive layer by changing the surface chemistry and suppressing methylammonium loss.

show abstract

“…For Tikhonov regularisation, LASSO and elastic net there are multiple popular methods to select the best hyper-parameter. These include the generalised- [93,120] and k-fold [131] cross validation, the C p statistic, [120,122] the L-curve [93,120] and the minimal product method. [93] Their use and a comparison of their effectiveness is more thoroughly described in the paper of Dorlhiac et al, [120] and the paper by Slavov et al [93] For the MEM, the subject of the regularisation parameter is often ill-discussed (it is hardly mentioned in the paper by Kumar et al) [128] however this was remedied in a paper by Lórenz-Fonfría and Kandori [132] where they discuss the effectiveness and use of the generalised cross validation, L-curve, Bayesian inference methods (also discussed previously by Skilling and Gull [133]) and the so-called 'discrepancy' criterion (optimising the α parameter until the χ 2 N value reaches 1).…”

Section: The Hyper-parametermentioning

confidence: 99%

“…[93] Recently Ðorđević et al also published an open-source implementation in Python, validated using data taken by (some of) the authors of this review. [131] For the MEM, Steinbach et al [138] has published a freeware program that is still currently available and was last updated in late 2017 (it is also closed source). It is noted by us that all of these do (at present) appear to be significantly less used by the transient spectroscopy community than the implementations of Global Kinetic Models discussed in Section 3 -89 citations for the Steinbach freeware as of early 2020 according to Google Scholar, 5 for the 'PyLDM' program and 51 for the OPTIMUS program.…”

Section: Published Implementationsmentioning

confidence: 99%

Data analysis in transient electronic spectroscopy – an experimentalist's view

Beckwith

Rumble

Vauthey

2020

International Reviews in Physical Chemistry

Self Cite

View full text Add to dashboard Cite

Time-resolved electronic spectroscopy has grown into a technique that provides hundreds to thousands of electronic spectra with femtosecond time resolution. This enables complex questions to be interrogated, with an obvious cost that the data are more detailed and thus require accurate modelling to be properly reproduced. Data analysis of these data comes in a variety of forms, starting with a variety of assumptions about how the data may be decomposed. Here, four different types of analysis commonly used are discussed: band-shape analysis, global kinetic analysis, lifetime distribution models, and soft-modelling. This review provides a 'user's guide' to these various methods of data analysis, and attempts to elucidate their successes, domains in which they may be useful, and potential pitfalls in their usage.

show abstract

Machine Learning for Analysis of Time-Resolved Luminescence Data

Cited by 34 publications

References 45 publications

Observation of Electron Shakeup in CdSe/CdS Core/Shell Nanoplatelets

Observation of Electron Shakeup in CdSe/CdS Core/Shell Nanoplatelets

How machine learning can help select capping layers to suppress perovskite degradation

Data analysis in transient electronic spectroscopy – an experimentalist's view

Contact Info

Product

Resources

About