Comprehensive X-ray Observations of the Exceptional Ultra-long X-ray and Gamma-ray Transient GRB 250702B with Swift, NuSTAR, and Chandra: Insights from the X-ray Afterglow Properties

TL;DR

GRB 250702B challenges simple classifications as either an ultra-long GRB or a relativistic TDE due to its prolonged central engine activity, early soft X-ray precursor, and wind-like environment. Using Swift, NuSTAR, and Chandra data from ~0.5 to 65 days, the authors model the X-ray afterglow with a forward+reverse shock in a wind, plus late-time engine-driven flares, requiring $\Gamma_0 \gtrsim 100$ and a narrow jet. Spectral modeling reveals a synchrotron cooling break near the X-ray band and substantial host dust extinction, with $N_{\rm H,z}/A_{V,z} \approx (4.5\pm1.0)\times10^{21}\ \text{cm}^{-2}\ \text{mag}^{-1}$. The authors favor a hybrid stellar-mass black hole progenitor, such as a micro-TDE or helium-star merger, over a WD-IMBH scenario, highlighting the potential of Einstein Probe to uncover similar events.

Abstract

GRB 250702B is an exceptional transient that produced multiple episodes of luminous gamma-ray radiation lasting for $>25$ ks, placing it among the class of ultra-long gamma-ray bursts (GRBs). However, unlike any known GRB, the \textit{Einstein Probe} detected soft X-ray emission up to 24 hours before the gamma-ray triggers. We present comprehensive X-ray observations of the transient's afterglow obtained with the Neil Gehrels Swift Observatory, the Nuclear Spectroscopic Telescope Array, and the Chandra X-ray Observatory between 0.5 to 65 days (observer frame) after the initial high-energy trigger. The X-ray emission decays steeply as $\sim t^{-1.9}$, and shows short timescale X-ray variability ($ΔT/T < 0.03$) in both Swift and NuSTAR, consistent with flares superposed on an external shock continuum. Serendipitous detections by the Swift Burst Alert Telescope (BAT) out to $\sim$0.3 days and continued NuSTAR variability to $\sim$2 days imply sustained central engine activity; including the precursor, the required engine duration is $\gtrsim 3$ days. Afterglow modeling favors the combination of forward and reverse shock emission in a wind-like ($k \approx 2$) environment. These properties, especially the long-lived engine and early soft X-ray emission, are difficult to reconcile with a collapsar origin, and GRB 250702B does not fit neatly with canonical ultra-long GRBs or relativistic tidal disruption events (TDEs). A hybrid scenario in which a star is disrupted by a stellar-mass black hole (a micro-TDE) provides a plausible explanation, although a relativistic TDE from an intermediate-mass black hole remains viable.

Figures (22)

• Figure 1: Left: X-ray lightcurve of GRB 250702B combining Swift and Chandra in the $0.3$$-$$10$ keV energy range. The start time $T_0$ is the GBM "D" burst. The red lines show the best-fit decay from the initial trigger time of GRB 250702D (Table \ref{['tab:triggertimes']}). The dashed red line ($t^{-1.9}$) includes the first XRT orbit in the fit, whereas the solid red line excludes the first XRT orbit and provides a better fit to the late-time data ($t^{-1.8}$). The observation windows of our NuSTAR data are shown as orange shaded regions. For reference we show the expected slopes for a relativistic TDE, corresponding to both complete ($t^{-5/3}$) and partial ($t^{-9/4}$) disruption. The bottom panel shows the $0.3$$-$$10$ keV lightcurve fit residuals relative to the solid red line. Right: Histogram showing the posterior probability density of the best-fit $T_0$ time for the X-ray afterglow emission. The gray histogram shows the $T_0$ posterior obtained when fitting all XRT data, and the red histogram shows the $T_0$ posterior when excluding the first XRT orbit. The trigger times reported by Fermi (red) and EP (blue) are shown as vertical lines (Table \ref{['tab:triggertimes']}). The July 1 window in which EP reports an initial detection is to the left of the vertical blue line. We also show as vertical black lines the times of Swift/BAT gamma-ray detections. All times are relative to the trigger of GRB 250702D at MJD 60858.548.
• Figure 2: The $3$$-$$79$ keV NuSTAR lightcurve (FPMA+FPMB) in 300 s bins (purple). The lightcurve is not background subtracted, but the background lightcurve (gray) is also shown for comparison. The dashed black line shows the mean count rate. The Bayesian Blocks results are shown as larger black circles. The bottom panel shows the residuals with respect to the mean count rate to underscore the source's short timescale variability.
• Figure 3: Zoom on the early Swift/XRT lightcurve using 25 s bins. The start time $T_0$ is the GBM "D" burst. The red lines is the same as the temporal fit shown in Figure \ref{['fig:temporalfit']} (left panel). The hardness ratio (HR) between $2$$-$$10$ and $0.3$$-$$2$ keV is shown in the bottom panel. The dashed black line shows the mean hardness ratio of the full lightcurve.
• Figure 4: Left: X-ray lightcurve comparing NuSTAR (FPMA+FPMB) data to the XRT lightcurve and best-fit temporal powerlaw as shown in Figure \ref{['fig:temporalfit']}. The start time $T_0$ is the GBM "D" burst. The solid red line is the best-fit ($t^{-1.7}$) when including only XRT at $>2.5$ d, and the red dashed line is the best-fit ($t^{-1.9}$) when including all XRT data. NuSTAR data is in 150 s bins and XRT in 100 s bins. The XRT count rate and powerlaw fit are re-scaled to their expected NuSTAR$3$$-$$79$ keV count rate. We emphasize that this scale factor is not arbitrary and is computed directly based on the best-fit spectral shape (§ \ref{['sec:spec']}). Right: Same as the left panel but zoomed in on the NuSTAR data in linear-linear space. Note that the temporal axis is in units of ks in this panel.
• Figure 5: Left:Swift/BAT survey mode ($14$$-$$195$ keV) on emission from GRB 250702B from the start of July 1, 2025 to the end of July 4, 2025. Black points show the Swift/BAT detections ($14$$-$$195$ keV; Tables \ref{['fig:batlimits']} and \ref{['tab:guanofluxes']}) from both GUANO and survey data. Gray downward triangles represent $5\sigma$ upper limits from Swift/BAT in the $14$$-$$195$ keV energy range. The trigger times reported by Fermi (red) and EP (blue) are shown as vertical lines (Table \ref{['tab:triggertimes']}). The July 1 window in which EP reports an initial detection is shown as a blue shaded region. The time window of our initial NuSTAR observation (starting 1.3 d after the initial Fermi trigger) is shown as an orange shaded region. Right: Best-fit temporal powerlaw to the NuSTAR lightcurve in $3$$-$$79$ keV. The $1\sigma$ error on the fit is shown as a shaded orange region. Swift/BAT upper limits are shown as gray triangles and detections as black circles. The start time $T_0$ is taken as the trigger time of GRB 250702D. The vertical red lines show the Fermi trigger times of GRBs 250702B, 250702C, and 250702E (see Table \ref{['tab:triggertimes']}).