https://doi.org/10.5061/dryad.cc2fqz6n8

Description of the data repository

In the repository file Archive.zip exists a clear folder structure for each figure and related data that is described item by item below.

This data repository contains all raw data for plotting figures shown in the paper, 'Short-range excitonic correlations and enhanced excitonic
susceptibility revealed by few-femtosecond broadband XUV spectroscopy.' Unless otherwise noted, all figures were generated using Igor Pro (WaveMetrics) from the data files provided in this repository. The zip archive is organized into folders corresponding to individual figures (e.g., Fig_1, Fig_2, Fig_S1). Each folder contains the numerical data associated with the corresponding figure panels, as well as any fit files used to generate the plotted curves.

Specific content:

Fig. 1. Equilibrium core and valence spectra of 1T-TiSe2. (a) Static XUV absorption spectrum. (b) Core-level photoemission spectrum. (c) Orbital-projected density of states near the Fermi level for both cdw and normal states. (d) ARPES energy-momentum intensity maps at (260 K). (e) ARPES energy-momentum intensity maps at (40 K). For the 2D image data, the vertical axis corresponds to momentum, while the horizontal axis corresponds to energy. The spectral intensity is represented by the pixel value at each momentum–energy coordinate.

Fig. 2. Photoinduced dynamics of Ti and Se M edges in the ordered phase. (a) Transient XUV spectrum under high fluence excitation. (b) Transient XUV spectrum under low fluence excitation. The first column contains the photon energy values (eV), while the first row contains the pump–probe delay times (fs). The remaining matrix elements correspond to the transient spectral intensity at each photon-energy and delay-time coordinate. (c)--(f) Temporal traces under high and low photoexcitation. The file contains normalized transient traces and corresponding fit results for both high-fluence and low-fluence excitation conditions. The column labels in the first row specify the contents of each column.

Fig. 3. Spectroscopic signatures of the phase transition in time-resolved core-level absorption. (a) Temperature-dependent transient spectrum at 200 fs. The data contain the transient spectral response at a pump–probe delay of 200 fs as a function of temperature. The first column contains the photon energy values (eV), while the first row contains the temperatures (K). The remaining matrix elements correspond to the transient spectral intensity at each photon-energy and temperature coordinate. (b) Transient spectral response in the Ti edge region at 200 fs. (c) Transient spectral response in the Se edge region at 200 fs. The files contain transient spectral snapshots measured at a pump–probe delay of 200 fs for low- and high-temperature conditions. The first row provides the column headers, which specify the photon-energy and normalized transmission-change data for each temperature. (d) Temperature-dependent transmission change in the Ti spectral window. (e) Temperature-dependent transmission change in the Se spectral window. The files contain the temperature-dependent normalized transmission change integrated over the Ti spectral window (window ii) and the Se spectral window (window iii). The first column contains the temperature (K), and the second column contains the corresponding normalized transmission change.

Fig. 4. Persistent excitonic correlations above Tc and enhanced excitonic susceptibility at Tc. (a)--(c) Fluence-dependent transient dynamics. The files contain experimental transient traces and corresponding fit curves measured at different excitation fluences. Figures 4a, 4b, and 4c correspond to measurements performed at 23 K, 300 K, and 190 K, respectively. The first row provides the column headers, which identify the delay-time, normalized transmission-change, and fit data associated with each fluence. (d) Fluence dependence of the initial response time. The file contains the extracted initial response times as a function of excitation fluence at 23 K, 190 K, and 300 K. The first row provides the column headers, which identify the fluence values, initial response times, and their corresponding uncertainties for each temperature. The fitted square-root scaling curves shown in Fig. 4d are provided separately in the Fig_4_d_fit.txt file. (e) Temperature dependence of the initial response time. The file contains the initial response times extracted from transient-trace fitting as a function of temperature under low-fluence and high-fluence excitation conditions. The first row provides the column headers, which identify the temperature, initial response time, and corresponding uncertainty for each excitation fluence. The phenomenological model fits shown in Fig. 4e are provided separately in the Fig_4_e_ph_left.txt and Fig_4_e_ph_right.txt files.

Fig. S1. Normalized spectrum (a) and retrieved temporal profile (b) of the compressed NIR pulse.

Fig. S2. Typical XUV source spectra obtained from high harmonic generation in the in-house UBXAS beamline using either Kr (a) or Ar (b).

Fig. S3. Normalized transmitted XUV spectrum on the Ar 3s3p6np states (a) and the He doubly excited states (b).

Fig. S4. Long-time spectrum stability characterization. (a) Raw CCD spectra of the Ar 3s3p65p autoionizing state as a function of laboratory time lasting more than 16 hours. Data are organized in alternating column pairs, where each odd-numbered column contains the CCD pixel number and the subsequent even-numbered column contains the corresponding intensity. Each pixel–intensity column pair represents a single spectrum acquired at a specific laboratory time. The spectra are ordered chronologically from left to right, with later acquisition times appearing in successive column pairs. (b) Extracted peak positions of the Ar 3s3p65p autoionizing state from a. The first row provides the column headers, which identify the laboratory time, peak position, and corresponding fitting uncertainty.

Fig. S5. Characterization of UBXAS temporal resolution. (a) Transient transmission spectrogram of He two-electron excitation states. The file contains the transient transmission spectrogram of the He two-electron excitation states. The first column corresponds to the photon energy axis (eV), the first row corresponds to the pump–probe delay time axis (fs), and the remaining matrix elements contain the transient transmission values at each energy–delay coordinate. The static transmitted spectrum is provided in the 'Fig_S5_a_transmission.txt' file. (b) Transient signal of the He 2s2p state. The experimental result is provided in the 'Fig_S5_b.txt' file while the fit in the 'Fig_S5_b_fit.txt' file.

Fig. S6. Raw spectral intensity and signal-to-noise evaluation. (a) Raw static XUV spectra measured with and without the sample, and the dark counts measured under identical acquisition conditions. (b) Raw transmitted XUV spectra with pump on and off at a delay time of 200 fs in the Se M4,5 edge along with the associated transient spectrum.

Fig. S7. Oscillatory absorption transients dominated by the coherent A1g phonon. (a) Fluence-dependent transient dynamics in Se window iv. (b) Fluence-dependent transient dynamics in Ti window i. The files contain transient traces measured in Se Window iv and Ti Window i for different excitation fluences, together with the corresponding fit curves. The first row provides the column headers, which identify the delay-time, transmission-change, and fit data associated with each fluence. (c) Fluence dependence of the initial response time in Ti window i and Se window iv. The files contain the extracted initial response times as a function of excitation fluence. The first row provides the column headers, which identify the fluence values, initial response times, and their corresponding uncertainties.

Fig. S8. Comparison of fitting models for response time extraction. (a) Fluence-dependent transient dynamics in the Se M5 edge at 23 K. The file contains short-time transient XUV transmission traces measured in the Se M5 edge (spectral window iii in Fig. 2a) at 23 K for different excitation fluences, together with the corresponding fit curves. The first row provides the column headers, which identify the delay-time, normalized transmission-change, and fit data associated with each fluence. (b) Fluence-dependent initial response times extracted from (a). The square root scaling fit curves using the initial response times extracted from erf-exponential and biexponential models are provided in the 'Fig_s8_b_erf_fit.txt' and 'Fig_s8_b_biexp_fit.txt' files, respectively.

Fig. S9. Momentum-resolved electronic structure of 1T-TiSe2 across the phase transition. (a) Constant energy cut at 260 K. (b) Constant energy cut at 40 K. With the files, the first column represents the k_x coordinates, while the first row represents the k_y coordinates. The remaining matrix elements correspond to the spectral intensity at each (k_x, k_y) momentum coordinate. The corresponding energies are listed in the files' names. (c) ARPES energy-momentum intensity maps at (260 K). (d) ARPES energy-momentum intensity maps at (40 K). For the 2D image data, the vertical axis corresponds to momentum, while the horizontal axis corresponds to energy. The spectral intensity is represented by the pixel value at each momentum–energy coordinate. (e) Temperature-dependent spectral weight of the folded Se 4p bands taken from the dashed rectangle in (d). (f) Temperature-dependent gap size at Γ, as labeled in (d).

Fig. S10. Change of valence electronic structures across the phase transition. (a) at k_y = 0.745 (b) at k_y = 0. Within each file, odd-numbered columns correspond to momentum coordinates, while even-numbered columns correspond to the associated spectral intensity. Each momentum–intensity column pair represents a specific temperature. The data are ordered sequentially from low to high temperature.

Fig. S11. Absence of phase transition signals in equilibrium core-level spectroscopies. (a)(b) Temperature-dependent static XUV absorption spectra. Within each file, odd-numbered columns correspond to photon energy, while even-numbered columns correspond to the associated absorption. Each photon energy–absorption column pair represents a specific temperature. The data are ordered sequentially from low to high temperature. (c)--(j) Extracted peak positions along with the associated uncertainties as a function of temperature.

Fig. S12. XUV absorption spectrum of 1T-TiSe2 computed by TDDFT.

Fig. S13. Coherent phonon identification in spectral windows iii and iv. (a) Transient XUV transmission dynamics in spectral window iii at 23 K. The file contains transient XUV transmission traces measured in spectral window iii at 23 K under both low- and high-fluence excitation conditions, together with the corresponding fit curves. The first row provides the column headers, which identify the delay-time, transmission-change, and fit data for each fluence condition. (b) Fourier transform amplitudes of the residual components in windows iii and iv. The first row provides the column headers, which identify the frequency axis (THz) and the corresponding Fourier-transform amplitudes for each dataset.

Fig. S14. Frozen phonon calculation of the Se M4,5 edge absorption spectrum. Within each file, odd-numbered columns correspond to photon energy, while even-numbered columns correspond to the associated absorption as a function of the vertical displacement between Ti and Se atoms relative to the equilibrium position (gray), simulating coherent lattice motion. From left to right are 0%, -3%, -1.5%, +1.5%, +3% of the displacement.

Fig. S15. Integrated local density of states (ILDOS) in the b-c plane. (a) The ILDOS between 0 eV and 2 eV from the Fermi level. (b) The ILDOS between 2 eV and 4 eV. Within the files, the first row and first column represent the spatial coordinates (a.u.), and the remaining matrix elements correspond to the ILDOS intensity at each spatial coordinate.

Fig. S16. Fluence-dependent dynamics with respect to spectral window selection on Peak 2. (a)--(d) Reproduced from Fig. 4a–d in the main text, while the inset in (d) from Fig. 3c. The square root scaling curve in Fig. S16d is provided in the 'Fig_s16_d_fit.txt' file. (e)–(h) and (i)–(l) contain analogous datasets acquired using alternative spectral-window selections and follow the same data organization as panels (a)–(d). The first row of each file provides the column headers, which identify the recorded quantities.

Fig. S17. Temperature-dependent dynamics with respect to spectral window selection on Peak 2. (a)--(b) The files contain short-time transient XUV transmission traces measured in the Se M5 edge (corresponding to spectral window iii in Fig. 2a) as a function of temperature. (a) corresponds to measurements acquired at a pump fluence of 0.2 mJ/cm2, while (b) corresponds to measurements acquired at 2.0 mJ/cm² for all temperatures except 350 K, which was measured at 1.1 mJ/cm2 to avoid sample damage. The first row provides the column headers, which identify the delay-time, transmission-change, and fit data for each temperature. (c) Temperature dependence of the initial response time. The file contains the initial response times extracted from transient-trace fitting as a function of temperature under low-fluence and high-fluence excitation conditions. The first row provides the column headers, which identify the temperature, initial response time, and corresponding uncertainty for each excitation fluence. The phenomenological model fits shown in Fig. S17c are provided separately in the Fig_s17_c_ph_left.txt and Fig_s17_c_ph_right.txt files. (d)–-(f) and (g)–-(i) contain analogous datasets acquired using alternative spectral-window selections and follow the same data organization as panels (a)–-(c).

Fig. S18. Excitonic correlation signatures in Peak 3. (a)–-(b) The files contain short-time transient XUV transmission traces measured on Peak 3 at 23 K (a) and 190 K (b) for increasing excitation fluence. The first row provides the column headers, which identify the delay-time, normalized transmission-change, and corresponding fit data for each fluence. (c) Extracted initial response times along with the associated uncertainties. (d) Temperature-dependent transient dynamics at low fluence. The file contains short-time transient XUV transmission traces measured on Peak 3 as a function of temperature under low-fluence excitation, together with the corresponding fit curves. The first row provides the column headers, which identify the delay-time, normalized transmission-change, and fit data for each temperature. (e) Temperature-dependent transient dynamics at high fluence. The file contains short-time transient XUV transmission traces measured on Peak 3 as a function of temperature under high-fluence excitation, together with the corresponding fit curves. The first row provides the column headers, which identify the delay-time, normalized transmission-change, and fit data for each temperature. (f) Extracted initial response times with the corresponding uncertainties for high and low fluence excitation conditions.

Fig. S19. Transient change in XUV transmission near Se M4,5 edges. (a)--(b) Spectrally resolved early-time transient XUV transmission changes measured at 23 K (a) and 300 K (b) with an excitation fluence of 1.8 mJ/cm2. With in the files, the first column corresponds to the photon energy axis (eV), the first row corresponds to the pump–probe delay time axis (fs), and the remaining matrix elements contain the transient transmission values at each energy–delay coordinate. (c) Extracted temporal traces from (a) and (b). The first row provides the column headers, which identify the delay-time and transmission-change data for each temperature.