Flashing fast: characterising the 2025 outburst of MAXI J1957+032

TL;DR

MAXI J1957+032, an accreting millisecond X-ray pulsar in an ultra-compact orbit, was studied during the 2025 outburst with XMM-Newton, Swift, NuSTAR, and optical photometry to compare with the 2022 event. The study robustly detects coherent pulsations at ν ≈ 313.6437 Hz, finds no measurable spin derivative during the XMM-Newton observation, and reports a long-term spin-down between outbursts of ∼$-2\times10^{-14}$ Hz s$^{-1}$, compatible with magnetic-dipole braking; the pulse is almost sinusoidal with soft lags and a spectrum described by TBabs*(thcomp*bbodyrad) with a cool hotspot, and no reflection is detected. Magnetic-field constraints yield $B_p \lesssim 10^9$ G from spin-down and $(0.5-3)\times10^8$ G from accretion-torque considerations for $d=5\pm2$ kpc and $\xi=0.3-0.5$, suggesting a mildly leaky propeller may operate. The optical emission indicates disc irradiation with an underlying outer-disc component, and a possible early jet signature, with a delayed optical peak relative to X-rays that points to rapid disc evolution in these brief outbursts. Together, these results illuminate the disc–magnetosphere coupling and torque variability in ultracompact AMXPs and provide a benchmark for interpreting short, recurrent outbursts in this class.

Abstract

MAXI J1957+032 is an accreting millisecond X-ray pulsar that shows brief, recurrent outbursts in an ultra-compact ~1 h orbit. We characterise the 2025 outburst using X-ray timing and spectroscopy from XMM-Newton and Swift (and a late-time NuSTAR observation), together with contemporaneous optical photometry from LCO, and compare the spin frequency with the 2022 outburst. Timing searches detect coherent pulsations at ~313.6 Hz with no measurable frequency derivative during the XMM-Newton exposure. Relative to its 2022 outburst, we measure a long-term spin-down of ~-2x10^-14 Hz s^-1, consistent with magnetic-dipole braking in quiescence. The pulse profile is nearly sinusoidal, with significant power at the fundamental, second, and fifth harmonics; the fractional amplitude decreases with increasing flux and shows soft lags up to a few keV. The 0.5-10 keV spectrum is well described by absorbed thermal Comptonisation (photon index ~2.4) plus a cool blackbody (kT ~0.23 keV) consistent with emission from a surface hotspot; no reflection or Fe-line features are detected. Requiring R_m \leq R_co implies B_s ~(0.5-3)x10^8 G for d=(5 \pm 2) kpc and ξ=0.3-0.5, below the upper limit from the secular spin-down (B_p \leq 10^9 G), possibly indicating a mildly leaky propeller. The optical emission lies on the neutron-star branch of the L_OIR-L_X relation, consistent with reprocessing in a compact disc. The optical SEDs are broadly flat, while an early red excess suggests a transient jet contribution during the initial hard X-ray phase; an optical peak delayed relative to the X-rays may trace an outward-propagating heating front and rapid disc evolution in these short-lived outbursts.

Figures (9)

• Figure 1: Multi-wavelength light curves of MAXI J1957+032 during the 2025 outburst. The upper panel shows the 0.5--10 keV unabsorbed flux measured with Swift/XRT (blue points) and the XMM-Newton/EPIC-pn observation (pink point). The lower panel shows the optical $g'$, $r'$, and $i'$ magnitudes from LCO (see Sect. \ref{['sec:optical_data']}). An additional observation per- formed on May 13, 2025 (MJD 60808), corresponding to the blend of MAXI J1957+032 with a $g'\sim20$ nearby star, is shown for comparison. Error bars represent 1$\sigma$ uncertainties.
• Figure 2: Summary of the timing analysis for MAXI J1957+032 from the XMM-Newton/EPIC-pn observation. Top: 0.5--10.0 keV background-subtracted count rate versus time for each time interval adopted to estimate a significant pulse profile. Middle: fractional amplitude of the fundamental harmonic in each H-test-selected segment; $1\sigma$ errors from photon bootstrap. Bottom: pulse phase residuals relative to the best-fitting coherent timing model (Table \ref{['table:timing']}); the grey dashed line marks zero residual. Time is measured from $T_{0}=60805.40289977$ MJD (TDB), and the horizontal axis along the top shows the corresponding orbital phase over the observation .
• Figure 3: Pulse profile of MAXI J1957+032 in the 0.50--6.5 keV band. Points (with $1\sigma$ errors) show the mean-normalised, 64-bin profile plotted over two cycles. The solid line is the weighted least-squares fit at the H-test-selected order ($m_{\rm opt}=5$); dotted curves display the individually significant harmonic components (H1, H2, H5).
• Figure 4: Energy dependence of the pulsations in MAXI J1957+032. The XMM-Newton/EPIC-pn dataset is split into 16 significance-optimised bands spanning 0.5--10.0 keV. In each band, the signal is modelled with a pure sinusoidal function (fundamental harmonic) at the NS spin period. Panel (a): relative phase of the fundamental as a function of energy. Panel (b): fractional amplitude of the fundamental as a function of energy.
• Figure 5: Unfolded average spectrum of the persistent emission from MAXI J1957+032. The blue curves represent Swift/XRT spectra, from the earliest observation (darkest blue) to the latest (lightest blue), fitted with the TBabs*powerlaw model (see Sect. \ref{['sec:Swift_spectroscopy_analysis']}). In the legend, Swift/XRT observations are labelled as "SW" followed by their ObsID (see Table \ref{['tab:swift_obs']}). The red curve shows the XMM-Newton/EPIC-pn spectrum, fitted with TBabs*(thComp*bbodyrad) (see Sect. \ref{['sec:XMM_spectrum']}). The bottom panel displays the residuals with respect to the adopted models.