Insight-HXMT observations of jet-like corona in a black hole X-ray binary MAXI J1820+070

Black hole X-ray binaries (BHXRBs) undergo spectral state transitions during outbursts, alternating between hard and soft states characterized by differing dominance of Comptonized corona emission versus thermal disk emission. MAXI J1820+070, discovered in March 2018, exhibited extensive multiwavelength activity including X-ray, optical, and radio observations with detected low-frequency QPOs. Distance is 2.96 ± 0.33 kpc. NICER analyses during the early decay suggested a stable broad Fe K profile indicating a disk extending close to the black hole (∼2 Rg) and contracting thermal reverberation lags, implying evolution of the corona rather than changes in the inner disk radius, i.e., a contracting corona. Because reflection features are strongest at 20–50 keV and electron temperatures are best constrained by high-energy cutoffs, Insight-HXMT’s 1–250 keV coverage is suited to probe coronal and disk evolution. The study aims to determine how the corona+disk geometry and dynamics evolve during the hard state, focusing on the first outburst and especially the decay phase where high-energy coronal emission dominates.

Prior work established state transitions and coronal Comptonization as key in BHXRB spectra. NICER found a nearly unchanged broad Fe K line, inner disk close to ISCO, and evolving soft lags in MAXI J1820+070, suggesting a contracting corona. NuSTAR and other missions observed the source and reported timing evolution. Theory predicts reflection fraction depends on coronal geometry and dynamics; lamppost models link reflection fraction to source height under static assumptions, while outflowing corona models (e.g., pair-dominated flares ejecta) predict beaming that reduces disk illumination and reflection. Previous studies have invoked jet-base or standing-shock interpretations of the corona and found supersolar Fe abundances often arise due to reflection model density limitations. NuSTAR analyses with two lamppost sources found changing heights and a generally stable reflection spectrum; however, a moving-corona framework reconciles timing evidence of contraction with spectral evidence of decreasing reflection.

Data: Insight-HXMT observations from 2018-03-14 (MJD 58191) to 2018-06-17 (MJD 58286) during the first outburst were analyzed using LE (2–10 keV), ME (10–30 keV), and HE (28–200 keV) data with 1.5% systematics. Spectral features between 21–24 keV in ME (Ag K-shell fluorescence) were excluded. Standard screening criteria in HXMTDAS v2.0 were applied. HE spectra from small/large FoV detectors were combined to improve S/N. Spectra were analyzed in XSPEC v12.10.0. Modeling: The broadband continuum is thermal Comptonization with a high-energy cutoff. Reflection features (broad Fe K line at 3–10 keV and Compton hump at >20 keV) motivated a relativistic reflection model. The adopted baseline model was tbabs(diskbb + relxillCp + xillverCp)*constant, where relxillCp captures relativistic disk reflection plus coronal nthcomp continuum; xillverCp accounts for distant, non-relativistic reflection producing a narrow Fe Kα component; diskbb models the disk thermal component; tbabs models interstellar absorption (NH ≈ 1.5×10^21 cm^-2). Emissivity profile used a broken power law with q2=3 outside Rbr; inner radius fixed at Rin=R_ISCO; Rout=1000 Rg; spin a=0.998; inclination i=63°. Seed photon temperature in relxillCp fixed at 0.05 keV (tested also at 0.5 keV). Fitting: 70 epochs were fitted, employing MCMC to explore parameter posteriors and derive medians and 68% confidence intervals. Reflection fraction Rf is defined as the ratio of coronal intensity illuminating the disk to that reaching the observer within relxill normalization conventions. Parameter evolutions (Rf, emissivity q1, electron temperature Te, photon index Γ, ionization ξ, Fe abundance AFe, cross-calibration constants) were tracked. Outflow modeling: To assess effects of coronal bulk motion, reflection fractions were computed using relxilllpionCp (v1.4.0) which includes lamppost height H and outflow velocity β=v/c with GR light bending. Grids of Rr(H,β) were generated. Direct fits fixing H (e.g., H=7 Rg) allowed estimation of β evolution, acknowledging degeneracy with H. Simulated NuSTAR spectra were generated for various H and β to examine EW behavior. Robustness checks: alternative spin (a=0.5) and free inclination fits to representative epochs; two-lamppost (relxilllpCp) comparisons to NuSTAR; tests with fixed iron abundance (AFe=5) to probe correlations induced by assumed disk density in reflection model.

- Broadband Insight-HXMT spectra across 70 epochs are well fitted by tbabs(diskbb + relxillCp + xillverCp)*constant with reduced χ2 < 1.0. - Electron temperature: Te decreases from ~230 keV to ~50 keV during the rise, then slowly increases to ≤100 keV in the decay, consistent with MAXI/GSC + Swift/BAT results. - Reflection fraction: Rf steeply increases to ~0.5 in the rise, then decreases to ~0.1 during the decay, indicating a decreasing fraction of coronal photons illuminating the disk in the decay. - Geometry vs dynamics: NICER timing indicates contracting corona (decreasing height) during decay, which in a static lamppost would increase reflection. However, the observed decrease in Rf implies significant bulk outflow of the corona; beaming reduces disk illumination. relxilllpionCp calculations show R decreases with increasing β at fixed height and increases with decreasing H at fixed β. - Bulk velocity estimation (illustrative): Fixing H=7 Rg and fitting decay spectra yields an increasing outflow speed from ~0 c to ~0.8 c over time, reproducing the declining reflection fraction. At a later epoch (MJD 58271), consistent solutions include moderate β~0.5 with Rf~0.6; at peak (MJD 58204), Rf~1.2 with a stationary corona. - Emissivity and geometry: Inner emissivity index q1 is flattened (<1) with break radius Rbr ~ 20–60 R_ISCO, suggesting a spatially extended corona (tens of Rg) rather than a point source; fixing Rin=20 Rg forces steeper q1~6–8, illustrating degeneracy. - Iron line and hump: Broad Fe K line profile remains roughly stable; narrow 6.4 keV core and Compton hump normalization decline through decay. Estimated Fe line EW stays ~200 eV during early-to-mid decay; EW is relatively insensitive to Rf in the relxill normalization scheme. - Abundances and ionization: Fits prefer supersolar iron abundance AFe ≥ 5, with an apparent increase during decay and correlation with Rf (p≈7.5×10^-21). High ionization parameters (log ξ ≥ 3.8) are found. Authors attribute supersolar AFe trends to limitations of low-density (n=10^15 cm^-3) reflection models; increasing disk density during decay could mimic elevated AFe and require higher ξ to match soft X-rays. - Disk evolution: diskbb Tin increases from ~0.4 to 0.6 keV in decay; disk flux (0.01–100 keV) increases until ~MJD 58250 then decreases. Inner radius estimates from diskbb suggest proximity to ISCO when considering only direct disk flux, but are uncertain due to coupling with Comptonized disk fraction and spectral degeneracies. - Photon index Γ increases (softens) during decay while Rf decreases; this can arise if increased intercepted disk flux (decreasing geometric factor μs as the corona approaches the disk) outweighs hardening due to increased outflow velocity. - Robustness: Lower spin (a=0.5) or freeing inclination still yield Rf<1 and decreasing with time, supporting the outflowing corona interpretation. NuSTAR spectra re-fitted with relxillCp+xillverCp show Rf decreasing from ~0.8 to ~0.3 over decay, consistent with Insight-HXMT trends. - Comparative modeling: Two-lamppost static fits can match spectra but imply high Rf at low H, inconsistent with the decreasing Rf trend unless bulk outflow is included. Degeneracy between H and β prevents unique separation without external constraints.

The results address the central question of how the corona and disk co-evolve during an outburst’s hard-state decay. Despite timing evidence for a contracting corona (decreasing height), the reflection fraction declines, which is incompatible with a static lamppost geometry. Incorporating coronal bulk motion resolves this: as the corona contracts toward the black hole, its material outflows faster, beaming emission away from the disk and reducing reflection. This jet-base or standing-shock picture aligns with independent timing detections of a compact X-ray jet and provides a dynamical framework wherein both the coronal position (from lags) and bulk velocity (from reflection fraction) can evolve. The approximately constant broad Fe K equivalent width and diminishing narrow core reflect the interplay of relativistic disk reflection and a fading distant reflector. Flattened emissivity profiles imply an extended coronal region, consistent with a jet-base rather than a point source. The softening of the Comptonized continuum while reflection weakens can be understood if increased disk photon interception (due to shrinking corona-disk separation) reduces the Compton amplification factor even as the outflow speed increases. Overall, the findings showcase reflection fraction as a diagnostic of coronal dynamics and support a unified disk–jet coupling where acceleration occurs closer to the black hole.

Insight-HXMT broadband spectroscopy of MAXI J1820+070 reveals that during the hard-state decay the reflection fraction decreases while timing indicates a contracting corona. Modeling with a moving lamppost shows that an outflowing, jet-like corona whose bulk speed increases as it approaches the black hole can reconcile these observations. Additional findings include decreasing electron temperatures in the rise then modest increase in decay, flattened emissivity profiles suggesting an extended corona, and a stable broad Fe K line with a fading narrow component. The work advances a dynamical corona framework, interpreting the corona as a jet base or standing shock with evolving position and velocity. Future work should (i) obtain simultaneous high-throughput timing and broadband spectroscopy to break the height–velocity degeneracy, (ii) develop and apply high-density reflection models to derive physical abundances and ionization, and (iii) test the contracting-and-accelerating corona scenario across other BHXRBs and AGN to assess its universality.

- Degeneracy between lamppost height and outflow velocity in spectral fits prevents unique determination of both without independent constraints (e.g., precise reverberation distances). - Reflection models assume low, fixed disk density (n=10^15 cm^-3), likely biasing iron abundance and ionization parameter estimates; supersolar AFe trends may be artifacts. - Cross-instrument limitations: Insight-HXMT LE effective area is smaller than NICER, limiting precision of soft-band lags; converting lags to absolute distances requires detailed modeling (geometry, inclination, GR delays). - Assumptions in emissivity profile and fixed Rin=R_ISCO influence inferences about coronal extent; q1–Rin degeneracies remain. - The xillverCp distant reflector normalization evolution is empirical; physical origin and geometry of the distant reflector are not uniquely constrained. - Simulations and fixed-height fits suggesting β up to ~0.8c are model-dependent and illustrative rather than definitive.