Pith. sign in

REVIEW 2 major objections 4 minor 46 references

Line shapes in time- and angle-resolved photoemission spectroscopy explored by machine learning

T0 review · 2 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Finite energy resolution in TR-ARPES creates an apparent momentum dependence in decay curves and an apparent time-dependent band dispersion, even when the underlying electron dynamics is purely energy-dependent.

desk verdict A clean analytic warning about energy-resolution artifacts in TR-ARPES, but the decisive simulation uses twice the stated resolution and needs a re-run before the empirical claim is solid. read the letter →

arxiv 2506.02137 v1 pith:6KQ5VUIA submitted 2025-06-02 cond-mat.str-el

Tami C. Meyer , Gesa-R. Siemann , Paulina Majchrzak , Thomas Seyller , Jennifer Rigden , Yu Zhang , Emma Springate , Charlotte Sanders
show 1 more author
Philip Hofmann
This is my paper
classification cond-mat.str-el
keywords time-andangle-resolvedphotoemissionspectroscopyTR-ARPESk-meansclusteringtimedistributioncurvesenergyresolutiongrapheneultrafastcarrierdynamicsapparentdispersion
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper claims that a finite energy resolution in time- and angle-resolved photoemission spectroscopy (TR-ARPES) can change the shape of measured decay curves in a momentum-dependent way, even when the underlying electron dynamics depends only on energy. The authors reach this conclusion by applying $k$-means clustering to the full set of time distribution curves (TDCs) from quasi-free-standing monolayer graphene and to a minimal model of a linearly dispersing band with a Fermi-Dirac distribution whose temperature decays exponentially. The same resolution effect is shown to produce an apparent time dependence of the measured band dispersion above the Fermi level. If correct, analyses that extract decay times or band velocities from a few selected regions of interest can be biased by experimental resolution rather than by the material's physics.

What carries the argument

The load-bearing machinery is a Gaussian convolution in energy applied to a static Lorentzian spectral function $A(k,\omega)=\frac{1}{\pi}\frac{|\Sigma''|}{(\hbar\omega-\hbar v k)^2+\Sigma''^2}$ multiplied by a time-dependent Fermi-Dirac factor $f(\omega,T(t))$, with $T(t)=T_0+(T_{\max}-T_0)H(t-t_0)e^{-t/\tau}$; $k$-means clustering then groups normalized TDCs by line shape without assuming a fit function. The energy convolution is what mixes fast-decaying high-energy and slow-decaying low-energy intensity at a fixed detection energy, turning a purely energy-dependent decay into an apparent $k$-dependent one. The clustering serves as an unsupervised probe that exposes the resulting wave-like pattern in cluster boundaries.

What would settle it

Measure the same graphene sample under identical pump conditions at two or more analyzer energy resolutions, for example 110 meV and 55 meV, while keeping the time resolution fixed; if the mechanism is right, the amplitude of the wave-like cluster pattern and the fitted band-slope shift over the first 100 fs should decrease when the energy resolution is improved. If the $k$-dependence is unchanged or grows, resolution alone is not the cause.

Watch

Extended reading notes

Core claim

The paper shows that the apparent $k$-dependence of TDC decay times in graphene is an instrumental artifact with a concrete mechanism: when the photoemission intensity is convolved with a Gaussian energy resolution, intensity at a chosen energy receives contributions from both higher-energy states, which decay faster because the Fermi-Dirac distribution is nonlinear, and lower-energy states, which decay slower. Across the two sides of the linearly dispersing band this produces an asymmetric, wave-like pattern of TDC line shapes in the $(k,\omega)$ plane. The paper argues this explains the previously unnoticed $k$-dependence in the graphene data and predicts that the same effect shifts momentum distribution curves, making the fitted band slope time-dependent on a scale of tens of femtoseconds. It also shows that a simple exponential decay of the intensity is only a valid approximation at high energy and high temperature early in the decay, which is why exponential fits often appear to work.

Load-bearing premise

The argument depends on the assumption that the real experiment smears energy twice as much as the stated 110 meV resolution, because only a 220 meV smearing in the simulation reproduces the wave-shaped cluster pattern; if the actual smearing is 110 meV, the predicted momentum dependence may be weaker than observed.

Editorial extensions

If this is right

  • TDCs taken at the same binding energy but different $k$ can have measurably different decay shapes; a single exponential fit over such a region averages multiple decay times and broadens the extracted TDC.
  • The fitted slope of the dispersion above $E_F$ can change considerably over the first roughly 100 fs purely from energy resolution, so time-dependent band renormalization claims must first exclude this effect.
  • Exponential-decay analysis is justified only in the high-energy, high-temperature, early-time regime, where the intensity approximately follows $I(\omega,t)\propto e^{-\hbar\omega t/\tau k_B T_{\max}}$.
  • The $k$-means cluster centroids provide averaged, high-signal-to-noise TDCs and can expose line-shape trends that are invisible when only a few regions of interest are inspected.
  • Reported apparent $k$-dependences of decay constants in other TR-ARPES studies could be partly explained by finite energy resolution rather than intrinsic physics.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The mechanism predicts the strength of the $k$-dependence should grow with the band slope and with the ratio of the energy resolution to that slope; re-measuring the same sample with deliberately degraded energy resolution would give a quantitative scaling test.
  • If the effect is generic, published lifetime maps for other materials that show $k$-dependent decay constants should be re-analyzed with an energy-resolution kernel before assigning them to many-body physics.
  • Because the apparent time-dependent dispersion arises from a symmetric convolution but an asymmetric Fermi-Dirac weighting, the effect should be strongest close to $E_F$ and negligible at high energy, giving a falsifiable spatial signature within a single dataset.
  • A practical extension would be to use the amplitude of the cluster-boundary wave as an in-situ estimate of the effective energy resolution, turning the artifact into a calibration tool.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 4 minor

Summary. The paper applies k-means clustering to TR-ARPES data from quasi-free-standing monolayer graphene and to a one-dimensional model with a linear band and exponential electronic-temperature decay in order to study TDC line shapes. The central claim is that finite energy resolution mixes contributions from different energies across the band, producing an apparent k-dependence of TDC decay times and an apparent time-dependent band slope above the Fermi level, even when the underlying dynamics are purely energy-dependent and k-independent. The paper also analyzes how finite time resolution shifts TDC maxima and derives the conditions under which a single-exponential decay fit is a reasonable approximation, summarized in Eq. (5).

Significance. If the central claim holds, the paper identifies a generic and important resolution artifact in TR-ARPES analysis: measured TDC decay times and apparent dispersions above EF can depend on experimental energy resolution even for simple, k-independent dynamics. Strengths of the manuscript include a transparent minimal model, a clear schematic explanation of the proposed mechanism, a clean analytic derivation of Eq. (5) with stated assumptions, and the use of unsupervised clustering to survey line shapes without imposing a fit model. The main weakness is empirical validation: the simulation that reproduces the experimental wave-like pattern uses an energy resolution of 220 meV, whereas the stated experimental resolution is 110 meV, and the agreement itself is only qualitative. The mechanism is plausible, but the comparison to experiment does not yet establish that the quoted instrument resolution is sufficient to produce the observed effect.

major comments (2)
  1. [Methods and Fig. 4(a)] The simulation that reproduces the experimental wave-like cluster pattern uses ΔE = 220 meV, while the Methods section states the combined experimental energy resolution as ΔE = 110 meV and the experimental preprocessing additionally applies a 106 meV FWHM Gaussian smoothing in energy. No justification is given for 220 meV. If the 106 meV smoothing is intended to be part of the effective resolution, the standard quadrature sum would be about 153 meV, not 220 meV; if the smoothing was not applied to the simulated data, the comparison is not apples-to-apples. Because the magnitude of the k-dependent mixing grows with the energy broadening, the present comparison shows only that a larger, unstated effective broadening can produce the wave pattern, not that the quoted 110 meV resolution suffices. Please justify the 220 meV value explicitly or rerun the simulation at 110 meV (or 153 meV, if the smoothing is included) and compare quantitatively with the experimental cluster map.
  2. [Finite energy resolution effects (Figs. 4 and 5)] The claimed agreement between simulation and experiment is qualitative: the cluster boundaries and centroid line shapes are compared by eye, and no metric is given for the similarity of the wave-like patterns. In addition, the apparent time-dependent band slope in Fig. 5(b) is extracted from fits to MDC maxima without error bars or a noise model, so it is not possible to assess whether the slope variation is statistically significant in the experimental data. Please provide a quantitative comparison between simulated and experimental cluster maps (e.g., overlap of cluster regions or a decay-time difference metric) and report uncertainties for the fitted band slopes, for example by bootstrapping. This is necessary to support the claim that the resolution-induced effect is experimentally relevant at the stated instrument parameters.
minor comments (4)
  1. [Methods, Eq. (4)] Eq. (4) contains a Heaviside function H(t − t0) but the exponential is written as e^{−t/τ} rather than e^{−(t−t0)/τ}; please clarify the time origin used for the temperature decay.
  2. [Fig. 1] The experimental spectrum is shown near peak excitation at t − t0 = 20 fs, while the simulation is shown at t − t0 = 78 fs; the choice of the simulated time is not explained and may confuse readers.
  3. [k-means clustering] The number of clusters is fixed to six without explanation; a brief statement about how this value was selected (e.g., cluster stability or silhouette analysis) would strengthen the presentation.
  4. [Methods and code availability] The text says the analysis is based on the code in Ref. [43] accompanying Ref. [24]; please make the specific simulation and clustering code used for the present model explicitly available, or state clearly that it is a minor extension of the existing repository.

Circularity Check

1 steps flagged · score 4.0 of 10

Validation of the resolution-induced k-dependence uses ΔE=220 meV in the simulation while the stated experimental resolution is 110 meV; the analytic mechanism is independent, but the empirical match is conditioned on an unstated parameter choice.

  1. fitted input called prediction [Methods (experimental resolution) and 'Finite energy resolution effects' / Fig. 4(a)]
    "The combined energy resolution was ∆E = 110 meV... The wave-shaped cluster distribution observed in the experimental data in Fig. 2(b) can be reproduced in the simulation when a finite energy resolution is taken into account. This is shown in Fig. 4(a), where we present the results of our clustering analysis after applying a Gaussian broadening with ∆E = 220 meV to the simulated data."

    The empirical case for 'finite energy resolution introduces a k-dependence in the TDC shape' rests on the match between the simulated cluster map (Fig. 4a) and the experimental cluster map (Fig. 4b). That match is obtained with a Gaussian broadening of 220 meV, not the stated experimental resolution of 110 meV, and also not the quadrature combination of 110 meV with the 106 meV energy smoothing used before clustering (~153 meV). No justification for this factor-of-two increase is given. The simulation parameter is therefore effectively chosen to reproduce the target pattern, and the claimed agreement does not independently demonstrate that the quoted experimental resolution produces the effect. The analytic mechanism explained in Fig.

full rationale

The paper's central analytic mechanism — that Gaussian energy broadening mixes faster-decaying higher-energy states and slower-decaying lower-energy states asymmetrically across a linear band, producing a k-dependent TDC shape and an apparent time-dependent dispersion — is derived from the stated model (Eqs. 2–4) and is internally consistent. That part is not circular. The circularity concern is confined to the empirical validation. The only comparison that ties the mechanism to the experimental data is the reproduction of the wave-like cluster pattern in Fig. 4(a), which uses ΔE = 220 meV although the Methods section quotes the combined experimental energy resolution as ΔE = 110 meV. The text supplies no explanation for this discrepancy, and the additional 106 meV Gaussian smoothing applied to the experimental TDCs prior to clustering does not, under standard quadrature addition, yield 220 meV. Thus the simulation's key input is not an independently measured quantity; it is a value that makes the simulated cluster map resemble the experimental one. This makes the 'prediction' of the wave pattern a demonstration that a larger effective resolution can produce the pattern, not an independent confirmation that the quoted resolution does so. The self-citation to the authors' earlier k-means code (Ref. [24]/[43]) is methodological and not load-bearing for the physics. Overall, the analytic derivation stands on its own, but the decisive empirical validation is weakened by a parameter effectively fit to the target observation, giving a partial circularity score of 4 rather than 0 or 2.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The central claim rests mainly on the factorization of the photoemission intensity, the simplified graphene model with a static spectral function and exponential temperature decay, and the clustering preprocessing choices. No new physical entities are introduced. The most consequential free parameter is the simulated energy resolution of 220 meV, which is not tied to the stated experimental value of 110 meV.

free parameters (7)
  • Fermi velocity v = 1.9e6 m/s
    Set by hand to mimic graphene dispersion; not fitted to the TR-ARPES data.
  • Imaginary self-energy Sigma'' = 100 meV
    Set by hand for the model spectral function.
  • Equilibrium temperature T0 = 80 K
    Matches experimental sample temperature; model input.
  • Peak electronic temperature Tmax = 2500 K
    Chosen model parameter; affects the effective decay time in Eq. (5) but not the qualitative resolution artifact.
  • Temperature decay time tau = 500 fs
    Chosen model parameter; explicitly not an excited-state lifetime.
  • Simulated energy resolution Delta E = 220 meV
    Hand-selected, twice the stated experimental 110 meV; used to reproduce the wave-like cluster pattern. This is the most consequential free parameter.
  • Clustering preprocessing parameters = energy FWHM 106 meV, k FWHM 0.02 Å^-1, time FWHM 70 fs, k=6 clusters
    User-chosen preprocessing and clustering settings; not optimized against a benchmark.
assumptions (6)
  • domain assumption Photoemission intensity factorizes as A(k,omega,t)|M|^2 f(omega,T_t) convolved with Gaussian resolution (Eq. 1).
    Standard TR-ARPES model; the paper assumes this factorization and Gaussian resolution for all simulations.
  • domain assumption For quasi-free-standing monolayer graphene, the spectral function A and matrix element M are time-independent; the only time dependence enters through the Fermi-Dirac distribution f.
    Justified by graphene's simple band structure, but it is an assumption that excludes pump-induced band renormalization or matrix element effects.
  • ad hoc to paper The spectral function is a Lorentzian with constant imaginary self-energy Sigma'' and linear dispersion epsilon(k)=v k (Eq. 2).
    Minimal model chosen to isolate resolution effects; it neglects band curvature, k-dependent self-energy, and matrix elements.
  • ad hoc to paper The electronic temperature decays as a single exponential after instantaneous excitation (Eq. 4).
    Chosen for simplicity; real graphene requires a three-temperature model, as the paper notes.
  • standard math At high energy the Fermi-Dirac distribution reduces to a Boltzmann factor, and for t << tau a first-order Taylor expansion is valid (Eq. 5).
    Provides the exponential-decay approximation; assumptions are stated explicitly by the authors.
  • domain assumption k-means clustering of normalized TDCs with user-selected k captures meaningful line-shape trends even though it ignores spatial proximity.
    The paper argues the contiguous cluster regions indicate reliability, but no statistical validation or stability check is given.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Line shapes in time- and angle-resolved photoemission spectroscopy explored by machine learning." pith.science (2026). https://pith.science/paper/6KQ5VUIA

@misc{pith2026250602137,
  author       = {Pith},
  title        = {Pith review of: Line shapes in time- and angle-resolved photoemission spectroscopy explored by machine learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6KQ5VUIA}},
  note         = {Machine review of arXiv:2506.02137}
}
abstract

Time- and angle-resolved photoemission spectroscopy is a powerful technique for investigating the dynamics of excited carriers in quantum materials. Typically, data analysis proceeds via the inspection of time distribution curves (TDCs), which represent the time-dependent photoemission intensity in a region of interest -- often chosen somewhat arbitrarily -- in energy-momentum space. Here, we employ $k$-means, an unsupervised machine learning technique, to systematically investigate trends in TDC line shape for quasi-free-standing monolayer graphene and for a simple analytical model. Our analysis reveals how finite energy and time resolution can affect the TDC line shape. We discuss how this can be taken into account in a quantitative analysis, and under what conditions the time-dependent photoemission intensity after laser excitation can be approximated by a simple exponential decay.

Figures

Figures reproduced from arXiv: 2506.02137 by the authors.

Figure 1
Figure 1. FIG. 1. (a) Photoemission spectrum from QFMLG at peak [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. (c) shows the cluster centroids. As already observed in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. (a) Selected TDCs from the spectral function simu [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4. (a,b) Cluster distribution resulting from applying [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Apparent changes in the of the measured electronic [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. (a) Effect of the energy resolution on the TDC line [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

46 extracted references · 44 canonical work pages

  1. [1]

    Xu and A

    C. Xu and A. Zong, npj Quantum Materials 10 (2025)

    work page 2025

  2. [2]

    Perfetti, P

    L. Perfetti, P. A. Loukakos, M. Lisowski, U. Bovensiepen, H. Eisaki, and M. Wolf, Phys. Rev. Lett. 99, 197001 (2007)

    work page 2007

  3. [3]

    J. C. Johannsen, S. Ulstrup, F. Cilento, A. Crepaldi, M. Zacchigna, C. Cacho, I. C. E. Turcu, E. Springate, F. Fromm, C. Raidel, et al., Phys. Rev. Lett.111, 027403 (2013)

    work page 2013

  4. [4]

    Rohwer, S

    T. Rohwer, S. Hellmann, M. Wiesenmayer, C. Sohrt, A. Stange, B. Slomski, A. Carr, Y. Liu, L. M. Avila, M. Kallane, et al., Nature 471, 490 (2011)

    work page 2011

  5. [5]

    Hellmann, T

    S. Hellmann, T. Rohwer, M. Kall¨ ane, K. Hanff, C. Sohrt, A. Stange, A. Carr, M. M. Murnane, H. C. Kapteyn, L. Kipp, et al., Nature Communications 3, 1069 (2012)

    work page 2012

  6. [6]

    C. L. Smallwood, W. Zhang, T. L. Miller, G. Affeldt, K. Kurashima, C. Jozwiak, T. Noji, Y. Koike, H. Eisaki, D.-H. Lee, et al., Phys. Rev. B 92, 161102 (2015)

    work page 2015

  7. [7]

    Zonno, F

    M. Zonno, F. Boschini, E. Razzoli, S. K. Y. Dufresne, M. Michiardi, M. X. Na, T. M. Pedersen, S. Gorovikov, S. Gonzalez, G. Di Santo, et al., Phys. Rev. B 103, 155109 (2021)

    work page 2021

  8. [8]

    S. Duan, W. Xia, C. Huang, S. Wang, L. Gu, H. Liu, D. Xiang, D. Qian, Y. Guo, and W. Zhang, Phys. Rev. Lett. 130, 226501 (2023)

    work page 2023

Show all 46 references
  1. [9]

    Y. H. Wang, H. Steinberg, P. Jarillo-Herrero, and N. Gedik, Science 342, 453 (2013)

    2013

  2. [10]

    Budden, T

    M. Budden, T. Gebert, M. Buzzi, G. Jotzu, E. Wang, T. Matsuyama, G. Meier, Y. Laplace, D. Pontiroli, M. Ricc` o, et al., Nature Physics17, 611 (2021)

    2021

  3. [11]

    de la Torre, D

    A. de la Torre, D. M. Kennes, M. Claassen, S. Gerber, J. W. McIver, and M. A. Sentef, Rev. Mod. Phys. 93, 041002 (2021)

    2021

  4. [12]

    Boschini, M

    F. Boschini, M. Zonno, and A. Damascelli, Rev. Mod. Phys. 96, 015003 (2024)

    2024

  5. [13]

    J. A. Sobota, S.-L. Yang, D. Leuenberger, A. F. Kemper, J. G. Analytis, I. R. Fisher, P. S. Kirchmann, T. P. De- vereaux, and Z.-X. Shen, Phys. Rev. Lett. 113, 157401 (2014)

    2014

  6. [14]

    Crepaldi, B

    A. Crepaldi, B. Ressel, F. Cilento, M. Zacchigna, C. Grazioli, H. Berger, P. Bugnon, K. Kern, M. Grioni, and F. Parmigiani, Phys. Rev. B 86, 205133 (2012)

    2012

  7. [15]

    Hajlaoui, E

    M. Hajlaoui, E. Papalazarou, J. Mauchain, G. Lantz, N. Moisan, D. Boschetto, Z. Jiang, I. Miotkowski, Y. P. Chen, A. Taleb-Ibrahimi, et al., Nano Letters 12, 3532 (2012)

    2012

  8. [16]

    C. L. Smallwood, J. P. Hinton, C. Jozwiak, W. Zhang, J. D. Koralek, H. Eisaki, D.-H. Lee, J. Orenstein, and A. Lanzara, Science 336, 1137 (2012)

    2012

  9. [17]

    J. A. Sobota, S. Yang, J. G. Analytis, Y. L. Chen, I. R. Fisher, P. S. Kirchmann, and Z.-X. Shen, Phys. Rev. Lett. 108, 117403 (2012)

    2012

  10. [18]

    Ulstrup, J

    S. Ulstrup, J. C. Johannsen, F. Cilento, J. A. Miwa, A. Crepaldi, M. Zacchigna, C. Cacho, R. Chapman, E. Springate, S. Mammadov, et al., Phys. Rev. Lett.112, 257401 (2014)

    2014

  11. [19]

    J. C. Johannsen, S. Ulstrup, A. Crepaldi, F. Cilento, M. Zacchigna, J. A. Miwa, C. Cacho, R. T. Chapman, E. Springate, F. Fromm, et al., Nano Letters 15, 326 (2015)

    2015

  12. [20]

    S.-L. Yang, J. A. Sobota, D. Leuenberger, Y. He, M. Hashimoto, D. H. Lu, H. Eisaki, P. S. Kirchmann, and Z.-X. Shen, Phys. Rev. Lett. 114, 247001 (2015)

    2015

  13. [21]

    Majchrzak, S

    P. Majchrzak, S. Pakdel, D. Biswas, A. J. H. Jones, K. Volckaert, I. Markovi´ c, F. Andreatta, R. Sankar, C. Jozwiak, E. Rotenberg, et al., Physical Review B 103 (2021)

    2021

  14. [22]

    C. Bao, Q. Li, S. Xu, S. Zhou, X.-Y. Zeng, H. Zhong, Q. Gao, L. Luo, D. Sun, T.-L. Xia, et al., Nano Letters 22, 1138 (2022)

    2022

  15. [23]

    Zhong, C

    H. Zhong, C. Bao, T. Lin, F. Wang, X. Cai, P. Yu, and S. Zhou, Physical Review B 109 (2024)

    2024

  16. [24]

    Majchrzak, C

    P. Majchrzak, C. Sanders, Y. Zhang, A. Kuibarov, O. Suvorov, E. Springate, I. Kovalchuk, S. Aswartham, G. Shipunov, B. B¨ uchner, et al., Physical Review Re- search 7, 013025 (2025)

    2025

  17. [25]

    Riedl, C

    C. Riedl, C. Coletti, T. Iwasaki, A. A. Zakharov, and U. Starke, Physical Review Letters103, 246804 (pages 4) (2009)

    2009

  18. [26]

    Speck, M

    F. Speck, M. Ostler, J. R¨ ohrl, J. Jobst, D. Waldmann, M. Hundhausen, L. Ley, H. B. Weber, and T. Seyller, Materials Science Forum 645-648, 629 (2010)

    2010

  19. [27]

    Gierz, J

    I. Gierz, J. C. Petersen, M. Mitrano, C. Cacho, I. C. E. Turcu, E. Springate, A. St¨ ohr, A. K¨ ohler, U. Starke, and A. Cavalleri, Nature Materials 12, 1119 (2013)

    2013

  20. [28]

    Ulstrup, J

    S. Ulstrup, J. C. Johannsen, M. Grioni, and P. Hofmann, Review of Scientific Instruments 85, 013907 (2014)

    2014

  21. [29]

    M. X. Na, A. K. Mills, F. Boschini, M. Michiardi, B. Nosarzewski, R. P. Day, E. Razzoli, A. Sheyerman, M. Schneider, G. Levy, et al., Science 366, 1231 (2019)

    2019

  22. [30]

    Curcio, S

    D. Curcio, S. Pakdel, K. Volckaert, J. A. Miwa, S. Ul- strup, N. Lanat` a, M. Bianchi, D. Kutnyakhov, F. Pres- sacco, G. Brenner, et al., Phys. Rev. B 104, L161104 (2021)

    2021

  23. [31]

    E. L. Shirley, L. J. Terminello, A. Santoni, and F. J. Himpsel, Phys. Rev. B 51, 13614 (1995). 9

    1995

  24. [32]

    Mucha-Kruczynski, O

    M. Mucha-Kruczynski, O. Tsyplyatyev, A. Grishin, E. McCann, V. I. Fal’ko, A. Bostwick, and E. Rotenberg, Physical Review B 77, 195403 (2008)

    2008

  25. [33]

    Butscher, F

    S. Butscher, F. Milde, M. Hirtschulz, E. Malic, and A. Knorr, Applied Physics Letters 91, 203103 (pages 3) (2007)

    2007

  26. [34]

    Tse and S

    W.-K. Tse and S. Das Sarma, Phys. Rev. B 79, 235406 (2009)

    2009

  27. [35]

    J. C. Johannsen, S. U. Ulstrup, M. Bianchi, R. Hatch, D. Guan, F. Mazzola, L. Hornekær, F. Fromm, C. Raidel, T. Seyller, et al., Journal of Physics: Condensed Matter 25, 094001 (2013)

    2013

  28. [36]

    Gauthier, J

    A. Gauthier, J. A. Sobota, N. Gauthier, K.-J. Xu, H. Pfau, C. R. Rotundu, Z.-X. Shen, and P. S. Kirch- mann, Journal of Applied Physics 128 (2020)

    2020

  29. [37]

    Speck, J

    F. Speck, J. Jobst, F. Fromm, M. Ostler, D. Waldmann, M. Hundhausen, H. B. Weber, and T. Seyller, Applied Physics Letters 99, 122106 (pages 3) (2011)

    2011

  30. [38]

    Thir´ e, G

    N. Thir´ e, G. Chatterjee, Y. Pertot, O. Albert, G. Karras, Y. Zhang, A. S. Wyatt, M. Towrie, E. Springate, G. M. Greetham, et al., Scientific Reports 13 (2023)

    2023

  31. [39]

    Freericks and A

    J. Freericks and A. F. Kemper, Journal of Electron Spec- troscopy and Related Phenomena 251, 147104 (2021)

    2021

  32. [40]

    J. B. MacQueen, in Proc. of the fifth Berkeley Sympo- sium on Mathematical Statistics and Probability , edited by L. M. L. Cam and J. Neyman (University of California Press, 1967), vol. 1, pp. 281–297

    1967

  33. [41]

    G. H. Ball and D. J. Hall, Behavioral Science 12, 153 (1967)

    1967

  34. [42]

    Xu and D

    R. Xu and D. C. Wunsch, Clustering (Wiley IEEE Press, 2008)

    2008

  35. [43]

    URL https://github.com/PhilipHofmann/PtBi2_ kmeans

  36. [44]

    Hofmann, I

    P. Hofmann, I. Y. Sklyadneva, E. D. L. Rienks, and E. V. Chulkov, New Journal of Physics 11, 125005 (2009)

    2009

  37. [45]

    Kevan, Physical Review B 33, 4364 (1986)

    S. Kevan, Physical Review B 33, 4364 (1986)

    1986

  38. [46]

    Y. H. Wang, D. Hsieh, E. J. Sie, H. Steinberg, D. R. Gardner, Y. S. Lee, P. Jarillo-Herrero, and N. Gedik, Phys. Rev. Lett. 109, 127401 (2012)

    2012

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.