NICER Discovery that SRGA J144459.2–604207 Is an Accreting Millisecond X-Ray Pulsar

We present the discovery, with the Neutron Star Interior Composition Explorer (NICER), that SRGA J144459.2−604207 is a 447.9 Hz accreting millisecond X-ray pulsar (AMXP), which underwent a 4 week long outburst starting on 2024 February 15. The AMXP resides in a 5.22 hr binary, orbiting a low-mass companion donor with M d > 0.1 M ⊙. We report on the temporal and spectral properties from NICER observations during the early days of the outburst, from 2024 February 21 through 2024 February 23, during which NICER also detected a type I X-ray burst that exhibited a plateau lasting ∼6 s. The spectra of the persistent emission were well described by an absorbed thermal blackbody and power-law model, with blackbody temperature kT ≈ 0.9 keV and power-law photon index Γ ≈ 1.9. Time-resolved burst spectroscopy confirmed the thermonuclear nature of the burst, where an additional blackbody component reached a maximum temperature of nearly kT ≈ 3 keV at the peak of the burst. We discuss the nature of the companion as well as the type I X-ray burst.

Accreting millisecond X-ray pulsars (AMXPs) are weakly magnetized (B ∼ 10 8 G) neutron stars with millisecond spin periods accreting from a low-mass companion donor, M d < 1M ⊙ (see Di Salvo & Sanna 2022;Di Salvo et al. 2023, for recent reviews).AMXPs are believed to have formed from the sustained spin-up torque provided by accreted material from a donor undergoing Roche lobe overflow over Gyr timescales; this is known as the recycling scenario (Alpar et al. 1992).AMXPs represent the most extreme spin periods for collapsed stellar objects, and thus are excellent laboratories for testing the limits of accretion physics (Chakrabarty et al. 2003;Bhattacharyya & Chakrabarty 2017).By comparing the expected accretion luminosity produced by different orbital angular momentum sinks (e.g., magnetic braking and/or gravitational radiation) and the observed accretion luminosity during outbursts, AMXPs can be used to test mass transfer scenarios in these extreme systems (e.g., Marino et al. 2019a;Ng et al. 2021;Bult et al. 2021a).For example, in the latest outburst of SAX J1808.4−3658, the observed orbital evolution was proposed to be due to ejected material having a specific angular momentum equal to or greater than that of the companion (Applegate & Shaham 1994;Illiano et al. 2023).We can further refine the physics governing these systems through the discovery of additional AMXPs with all-sky multi-wavelength monitoring of any transient outburst activity.These intermittent outbursts are likely the result of thermal instabilities in the accretion disk around the central neutron star (Lasota 2001).
The Mikhail Pavlinsky ART-XC telescope on the Spectrum-Roentgen-Gamma (SRG) observatory (Pavlinsky et al. 2021;Sunyaev et al. 2021) first reported the discovery of a bright X-ray transient, SRGA J144459.2−604207, on 2024 February 21 (Mereminskiy et al. 2024).Additional data from the MAXI/GSC X-ray sky monitor showed that the outburst actually began around February 15 (Mihara et al. 2024).This discovery was promptly followed up by an array of telescopes and instruments across many wavelengths.NICER discovered coherent X-ray pulsations at around 447.9 Hz and a type-I X-ray burst in its initial observations (Ng et al. 2024).MeerKAT reported a non-detection at GHz radio wavelengths (Mariani et al. 2024), but the X-ray source localization provided by Chandra (Illiano et al. 2024) led to the detection of a GHz radio counterpart by ATCA (Russell et al. 2024).Optical follow-up observations did not detect an optical counterpart (Sokolovsky et al. 2024;Cowie et al. 2024;Baglio et al. 2024;Saikia et al. 2024).A candidate near-infrared counterpart was identified with the PRIME telescope (Guiffreda et al. 2024), but it is formally incompatible with the Chandra localization.Further observations by INTEGRAL (Sanchez-Fernandez et al. 2024a,b) and NinjaSat (Takeda et al. 2024) detected type-I X-ray bursts with recurrence timescales ranging from 1.7-2.9h, with the period increasing as the persistent emission flux decreased.
We present the NICER discovery of 447.9 Hz pulsations of the AMXP SRGA J144459.2−604207 and provide a preliminary timing solution for the orbital modulation of the pulse frequency.We also report on the discovery and characterization of the type-I X-ray burst.It is the 27th known AMXP and the first AMXP outburst discovered by the ART-XC telescope.We note that after this manuscript was submitted, the ART-XC team reported on ART-XC observations of the source (Molkov et al. 2024).In §2, we present the NICER Xray observations undertaken and the associated analysis procedures.We present the timing and spectroscopic results in §3 and discuss the results in §4.

OBSERVATIONS AND DATA ANALYSIS
The Neutron Star Interior Composition Explorer (NICER) is an external payload on the International Space Station.It consists of 56 (52 operational) coaligned X-ray concentrator optics and silicon drift detectors in focal plane modules (FPMs).NICER provides fast-timing capabilities in the 0.2-12.0keV energy range, and the onboard global positioning system (GPS) receiver allows for 100 ns time-tagging accuracy (Gendreau et al. 2016;LaMarr et al. 2016;Prigozhin et al. 2016).
We report on public NICER observations conducted over 2024 February 21 through 2024 February 23 (MJD 60361-60363), with observation IDs (ObsIDs) 6204190101-6204190103.Additional NICER observations were conducted under the NICER Guest Observer program (PI: A. Papitto) and will be reported elsewhere.Our observations were reduced and processed with HEASoft version 6.33 and the NICER Data Analysis Software (NICERDAS) version 12 (2024-02-09 V012) using calibration version xti20240206.We imposed the following filtering criteria for the observations: angular offset for the source of ANG DIST < 54 ′′ ; NICER being outside of the South Atlantic Anomaly; an Earth limb elevation angle of ELV > 20 • ; a bright Earth limb angle of BR EARTH > 30 • ; an undershoot rate (per FPM; for dark current) of underonly range = 0-500, and an overshoot rate (per FPM; for charged particle saturation) of overonly range = 0-30.This resulted in good time intervals (GTIs) totaling 9.15 ks for scientific analysis.
We transformed the photon arrival times into the inertial frame of the Solar System by performing the Solar System barycenter corrections in the ICRS reference frame with source coordinates R.A. = 221.• 24558, Decl.= −60.• 69869 (Illiano et al. 2024) obtained by Chandra, using barycorr in FTOOLS with the JPL DE421 Solar System ephemeris (Folkner et al. 2009).We made use of XSPEC 12.14.0(Arnaud 1996) for our spectral analysis.The generation of spectral products was enabled by the nicerl3-spect spectral product pipeline, which allowed us to: group the spectra with the optimal binning; rebin the spectra so that each bin had a minimum of 25 counts (Kaastra & Bleeker 2016); generate background spectra using the nibackgen3C50 model (Remillard et al. 2022), and generate the associated response matrices.

RESULTS
We place the NICER observations in context of the overall outburst evolution from MAXI/GSC observations in Figures 1(a) and (b) -as reported in the figure zoom-in, the NICER observations analyzed were around the peak of the outburst.

X-ray Pulsation Discovery and Timing
To search for coherent periodicity, we employed the Fourier-domain acceleration search (FDAS) method implemented in version 4.0 of the open-source pulsar search and analysis software PRESTO (Ransom 2011).The algorithm accounts for possible pulsation frequency modulations due to the binary Doppler effect with orbital period P orb (Ransom et al. 2002).The FDAS were carried out on individual GTIs throughout the observation span of SRGA J144459.2−604207, with a minimum interval length of 100 s.We had 15 GTIs overall, with a median segment length of 613 s (spanning 183 s to 911 s).We adopted this approach as the sensitivity of acceleration searches is optimized for segment lengths T such that T ≲ P orb /10, as the pulsar acceleration is approximately constant and the spin frequency is expected to linearly evolve within the segment (Ransom et al. 2002).We searched over the 1-1000 Hz frequency range, and in two energy bands: 0.3-2.0keV and 2.0-10.0keV.We found significant candidates in several, but not all, GTIs within the 2.0-10.0keV band.Specifically, we found ten candidates around 447.9 Hz with single-trial significance over 2.5-5.5σ.Given the overall pulsation search parameter space, the pulsation signal corresponds to a trials-adjusted significance of 11.3σ (5.4 × 10 10 trials) across all of the independent data segments, frequency bins, and Fourier bins searched.
With the secure detection of coherent X-ray pulsations, we proceeded to characterize the pulse.First, we derived a provisional circular orbit model by fitting an ellipse in the period-acceleration plane with the ten significant candidates.Next, we performed an epoch folding search in approximately 1 ks intervals and modeled the residual frequency variations, resulting in revised orbital parameters from the initial estimate.We also maximized the profile variance (χ 2 ) from exploring a grid of values defined by the spin frequency (ν 0 ) and epoch of ascending node passage (T asc ).Next, we optimized the pulse significance by calculating the H-statistic (de Jager et al. 1989) with two harmonics (m = 2) for (2) where t j are the photon arrival times corrected for Doppler modulation described by the provisional orbit model (j ∈ {1, . . ., N } for N photons), over a grid of energy values (Buccheri et al. 1983).We found that the H-statistic was maximized for the energy range 1.03-11.97keV.However, the H-statistic exhibited very little variance (maximum of ∆H = +2) throughout the energy grid search.We adopted an energy range of 1.0-10.0keV for the timing analysis as the background (see below) dominates below and above this energy range.
With this initial model, we generated 19 pulse times of arrival (TOAs) with the photon toa.py tool in the NICERsoft data analysis package1 , where each TOA had an integration time of 300 s (with minimum exposure time of 200 s).The TOAs were then correlated with a pulse template, using all of the data, comprising the sum of three Gaussians.Finally, we fit the pulse TOAs to the ELL1 binary orbit model (Lange et al. 2001) available within PINT, an open-source pulsar timing Python package (Luo et al. 2021).The best-fit orbital ephemeris is given in Table 1, and the corresponding pulse arrival time delay (with respect to a constant spin frequency model) illustrating the best-fit orbit solution is shown in Figure 1(c).We note that the final orbital solution presented here is significantly different than the preliminary solutions previously reported both with the same NICER data (Ray et al. 2024) and from Insight-HXMT (Li et al. 2024).However, of these three solutions, the final solution we present here has a significantly larger maximum H-statistic value of 867.61,  (Mihara et al. 2024).The shaded gray region corresponds to the NICER observation interval.b) NICER 0.3-12.0keV light curve with 4 s bins.The burst emission was removed, but instead the midpoint of the burst interval (including the pre-burst) is shown with the red arrow.c) The pulse arrival time delay as a function of the orbital phase, with the best-fit orbit shown with the dashed lines, and the pulse timing residuals shown in squares (with scale on the right y-axis).d) Folded 1.0-10.0keV pulse profile with the NICER observations.The profile also shows a significant second harmonic.There are 20 pulse phase bins and two cycles are plotted for clarity.compared to 149.78 and 60.43, respectively.The final folded pulse profile is shown in Figure 1(d), where fitting a two-component sinusoid yielded fractional root-meansquared amplitudes of 1.79 ± 0.04% and 0.35 ± 0.04%, for the fundamental and second harmonic, respectively.The evolution of the pulse profile throughout the outburst and its energy dependence is outside the scope of this Letter, and will be reported elsewhere.

Type-I X-ray Burst
During the NICER observations, we detected a sharp, short-lived increase in the X-ray flux, which the light curve evolution suggested was a type-I X-ray burst.We first characterized the light curve (f (t) as a function of time, t) by fitting with a hybrid fast-rise exponentialdecay (FRED) model with a plateau (Barrière et al. 2015), described by (3) where C is the persistent flux level (c/s), A is the burst amplitude (c/s), t 0 is the burst onset time (in seconds), τ R and τ D are the rise and decay timescales (in seconds), respectively, t pe = t 0 + √ τ R τ D is the time at the peak (in seconds), and t pl is the interval length of the plateau (in seconds).The corresponding best fit parameters are given in Table 2.We show the results of the fit in Figure 2, where we also show residuals from fitting a simple FRED model (Figure 2b).While both models provide a reasonable description of the burst, the hybrid FRED and plateau model significantly improves the residuals around the burst peak.The simple FRED model can also be rejected with probability 1.4 × 10 −11 compared   to the hybrid FRED and plateau model according to the Akaike information criterion (Akaike 1974;Liddle 2007). 3.3.Spectroscopy

Persistent Emission
First, we defined several intervals for spectral extraction.The pre-burst interval was between MJDs 60362.35006 to 60362.35047(TT units at NICER; spanning about 35 s), 5 s before the burst onset.The burst interval was defined to be between MJDs 60362.35047 to 60362.35146(spanning about 85 s), 80 s after the burst onset.We also present fits to the individual ObsIDs, but for ObsID 6204190102 which contains the type-I X-ray burst, we excised the pre-burst and burst intervals.
For the spectral fits, we adopted a line of sight hydrogen column density of n H = 2.9 × 10 22 cm −2 .This value was determined from a joint fit of the spectra from the three ObsIDs (not shown; pre-burst and burst intervals removed), where we employed an absorbed thermal blackbody and power-law model and untied all spectral parameters except for n H , and we found n H = 2.90 ± 0.03 × 10 22 cm −2 .All subsequent spectral fitting was restricted to 1.0-10.0keV as the soft Xrays were absorbed below around 1.0 keV because of the high n H , and the background dominated above 10 keV.The wilm elemental abundance model was adopted for the spectral fits (Wilms et al. 2000).During the course of the three reported observations, the blackbody nor-malization remained constant within uncertainties, with an average value of norm BB = 36.2+7.4 −5.8 (R km /D 10 ) 2 , where R km is the source region radius in km, and D 10 is the source distance in units of 10 kpc.The blackbody temperature marginally decreased from kT = 0.99 ± 0.08 keV to kT = 0.85 +0.07 −0.06 keV, the photon index increased from Γ = 1.81 ± 0.03 to Γ = 1.90 ± 0.02, and the power-law normalization marginally decreased from norm PL = 0.76 ± 0.03 photons/s/cm 2 /keV to norm PL = 0.69 ± 0.03 photons/s/cm 2 /keV.The results are reported in Table 3.All uncertainties are reported at 90% confidence levels.The detailed spectroscopic evolution of the persistent emission during the entire outburst will be presented in a separate publication.Note-The uncertainties are given to 90% confidence level.The parameter normBB is scaled by (R km /D10) 2 and normPL has units of photons/keV/cm 2 /s.The absorbed 1.0-10.0keV flux, F 1−10 keV is expressed in units of 10 −9 erg s −1 cm −2 .
Next, we analyzed the pre-burst and burst intervals.

Time-Resolved Burst Spectroscopy
To further characterize the burst, we performed timeresolved spectroscopy starting from 5 s before burst onset, to 40 s after burst onset.We generated dynamically binned spectra, ensuring that each spectrum contained at least 2000 photons, and used nicerl3-spect to generate the associated responses.We used the nibackgen3C50 background spectrum corresponding to ObsID 6204190102, as the background is not expected to vary significantly over the burst.We also regrouped the spectra such that each spectral bin had a minimum of 25 counts.Given that the background dominated above 6.5 keV, the spectral fits were restricted to 1.0-6.5 keV.We fixed the parameters of the pre-burst model, and added a second thermal blackbody model to account for the burst evolution (the full model is tbabs(bbodyrad+powerlaw+bbodyrad)), which is shown in Figure 3.The burst decay was welldescribed by a cooling thermal blackbody component, which supports the thermonuclear origin for the X-ray burst.Motivated by bursters which have exhibited excess emission above the burst, we tried to fit the spectra by including a scaling factor to the underlying preburst emission (Worpel et al. 2013(Worpel et al. , 2015;;Güver et al. 2022a,b), but it did not statistically improve the spectral fits.This is similar to what has been seen in sources with relatively higher hydrogen column density values of n H > 10 22 cm −2 (Bult et al. 2021b;Güver et al. 2021;Bostancı et al. 2023).The bolometric unabsorbed flux shown in Figure 3(a) was derived using the convolutional cflux model, and we extrapolated the model out to 0.1-100 keV by using the energies command in XSPEC.

DISCUSSION
We have presented the NICER discovery of 447.9 Hz pulsations of the accreting millisecond X-ray pulsar SRGA J144459.2−604207, which resides in a 5.22 hr binary.We reported on initial NICER observations of the source around the peak of the outburst, from 2024 February 21 to 2024 February 23.The observations also revealed a type-I X-ray burst, and time-resolved burst spectroscopy confirmed the thermonuclear nature of the burst.
The determination of the binary orbital parameters of the SRGA J144459.2−604207system allows us to constrain the properties of the companion/mass donor.We can employ the binary mass function, where M d and M ns are the donor and neutron star masses, respectively, and G is the gravitational constant.Assuming M ns = 1.4M ⊙ , we plot the donor mass-radius curve from Equation 4 assuming a Roche lobe-filling donor, which is shown in Figure 4. We also plot mass-radius curves for several companion types, including white dwarfs (WDs; Deloye & Bildsten 2003) and zero-age main sequence stars (Tout et al. 1996).We found that for all plausible cold (core temperature, Figure 3. Evolution of the second blackbody component during the type-I X-ray burst from time-resolved spectroscopy.a) bolometric (0.1-100.0 keV) unabsorbed flux in units of 10 −8 erg s −1 cm −2 ; b) blackbody temperature, kT (keV); c) blackbody normalization, norm bb scaled by (R km /D10) 2 ; d) reduced χ 2 for the fit.The cooling blackbody, the key signature of the thermonuclear nature of the burst, is evident.The offset, ti, corresponds to MJD 60362.35047(TT units at NICER).
T c < 10 6 K) and hot (T c > 10 6 K) WDs, the WDs are too small to fill their Roche lobes to be viable companions for SRGA J144459.2−604207.
On the other hand, main sequence (H-rich) companions are much more likely.For semi-detached binaries such as the Roche lobe-filling donor and AMXP here, the period-density relation ρ ≃ 107(P orb /1 hr) −2 g cm −3 (Knigge et al. 2011) estimates a stellar mean density, ρ of around ρ ≈ 4.0 g cm −3 , which suggests an early Mtype main sequence donor (Drilling & Landolt 2000).While the mass-radius curve presented here is for solar metallicity (Z = Z ⊙ = 0.02), for donor masses below 1 M ⊙ , the mass-radius curves are weakly sensitive to the metallicity (Tout et al. 1996).The mass-radius curves for SRGA J144459.2−604207 and the main sequence companion intersect at around 0.65 M ⊙ , which corresponds to an a priori probability over binary inclination of P (i ≤ i 0.65 M⊙ ) ≈ 11% assuming an isotropic population of binary inclinations.Consequently, most of the probability space, after considering binary inclination, lies above the main sequence in the mass-radius plane, indicating stars larger than a main sequence star at the same mass.The main sequence companion could have a bloated H atmosphere due to the X-ray irradiation by the neutron star, thus allowing for less massive (i.e., more probable) donors.Additionally, since we do not have full orbital coverage with this current data set, and given the non-contiguous nature of NICER observations, we cannot provide further inclination constraints through searches for dips or eclipses.
It is also much more plausible that the companion star is very old (∼Gyr) given the recycled nature of the system (e.g., Altamirano et al. 2008;Sanna et al. 2018), and the stellar isochrones for these old companions suggest a lower (more plausible) mass (Girardi et al. 2000).The spin and orbital properties of SRGA J144459.2−604207,along with the derived companion properties, fit well within the population of known AMXPs with main sequence companions (e.g., Papitto et al. 2007;Riggio et al. 2011;Altamirano et al. 2011;Papitto et al. 2011Papitto et al. , 2013;;Patruno & Watts 2021).While we used M ns = 1.4M ⊙ , we note that the mass-radius relation for SRGA J144459.2−604207depends weakly on the NS mass.However, recycled pulsars which experienced low levels of sustained accretion over long timescales (∼Gyr) should have higher masses (Romani 1990;Romani et al. 2022).
NICER has detected a type-I X-ray burst within the observation interval covered in this Letter.The burst light curve profile was best fit with a phenomenological hybrid fast-rise exponential decay and plateau model (see Table 2).The burst duration of < 100 s suggests low accretion rates of mixed H/He material (Fujimoto et al. 1987;in't Zand et al. 2005).Time-resolved burst spectroscopy as shown in Figure 3 revealed a constant blackbody normalization during the burst, suggesting that a photospheric radius expansion (PRE) event did not take place.However, we can derive an upper limit on the source distance assuming that the burst peak was at Eddington luminosity.The peak bolometric (0.1-100.0 keV) unabsorbed flux was around 2.8 × 10 −8 erg s −1 cm −2 , and given the empirical critical luminosity of 3.8 × 10 38 erg s −1 at the peak of type-I X-ray bursts during PRE events, this implies an upper limit on the source distance of d < 10.6 kpc (Kuulkers et al. 2003).
Subsequent NICER observations, as well as those by other instruments such as Swift/XRT (Mariani et al. 2024), INTEGRAL (Sanchez-Fernandez et al. 2024a,b), and NinjaSat (Takeda et al. 2024) have detected more type-I X-ray bursts from SRGA J144459.2−604207.In fact, INTEGRAL reported the "quasi-periodic" nature of the type-I X-ray burst trains, with a burst recurrence had regular burst recurrence timescales between 3.6-5.7 hr since its outburst onset in 1988 (Ubertini et al. 1999;Chenevez et al. 2016), though its bursting activity has dramatically reduced upon its transition into the spectrally softer "banana" state in 2016 (Chenevez et al. 2016;Yun et al. 2023).The other known clocked burster, 1RXS J180408.9−342058, was found to display clocked bursting behavior in its intermediate spectral state, with a recurrence timescales of around 0.9 h (Marino et al. 2019b).The 11 Hz burster IGR J17480−2446 also exhibited clock-like bursting activity (Chakraborty et al. 2011).Such behavior is thought to be related to the near-solar composition of the accreted material (Galloway et al. 2004;Heger et al. 2007;Lampe et al. 2016;Meisel 2018).We note that this is the second clocked burster for which we know the spin period and orbital properties, the other being IGR J17480−2446 (Strohmayer et al. 2010;Motta et al. 2011).A detailed study of the broadband timing and spectral properties of the type-I X-ray bursts from the outburst of SRGA J144459.2−604207 is outside of the scope of this Letter and will be reported in a future publication.
Assuming an averaged unabsorbed bolometric (0.1-100.0 keV) flux across the three observations of F = 7 × 10 −9 erg s −1 cm −2 , then for d < 10.6 kpc, the source luminosity is estimated to be L < 9.5 × 10 37 erg s −1 .If we assume spin equilibrium, then the dipolar NS magnetic field is approximately B = 4.2ζ (5) where ζ is the order-unity ratio of the magnetospheric and Alfvén radii (Ghosh & Lamb 1979), P = 1/ν is the pulsar spin period, and Ṁ is the mass transfer rate from the donor onto the NS.For a NS with radius R = 10 km and M ns = 1.4M ⊙ , we find for d < 10.6 kpc that Ṁ < 8.11 × 10 −9 (d/10.6 kpc) 2 M ⊙ /yr.Thus for d < 10.6 kpc, B < (6.6 − 96.6) × 10 8 G, which is within expectations (albeit a large range) from the distribution of estimated magnetic field strengths of AMXPs (Mukherjee et al. 2015).
A more accurate determination of the magnetic field strength will come from observing more outbursts from SRGA J144459.2−604207 to measure a long-term spin frequency derivative, ν.In fact, MAXI/GSC found signs of past weak activity in 2022 January and 2023 December (Negoro et al. 2024), and INTEGRAL reported significant hard X-ray activity in 2023 December (Sguera & Sidoli 2024), but there were no dense monitoring observations available for those periods.Further X-ray monitoring observations will be crucial in detecting the rising outbursts of SRGA J144459.2−604207 to further constrain the source properties, such as the evolution of the type-I X-ray burst morphology.Further all-sky highcadence multiwavelength surveys will be instrumental in discovering more AMXPs to understand the population and diversity of phenomena they exhibit.

Figure 1
Figure 1.a) SRGA J144459.2−604207monitoring 2-10 keV light curve by MAXI/GSC(Mihara et al. 2024).The shaded gray region corresponds to the NICER observation interval.b) NICER 0.3-12.0keV light curve with 4 s bins.The burst emission was removed, but instead the midpoint of the burst interval (including the pre-burst) is shown with the red arrow.c) The pulse arrival time delay as a function of the orbital phase, with the best-fit orbit shown with the dashed lines, and the pulse timing residuals shown in squares (with scale on the right y-axis).d) Folded 1.0-10.0keV pulse profile with the NICER observations.The profile also shows a significant second harmonic.There are 20 pulse phase bins and two cycles are plotted for clarity.

Figure 2 .
Figure 2. Phenomenological fit to the light curve of the type-I X-ray burst in the NICER ObsID 6204190102 with a simple FRED model and a model with a FRED component and a plateau.a) Light curve of the type-I X-ray burst, with 0.5 s bins, with the FRED + plateau model shown in the solid red line; b) Residuals from fitting the burst light curve to a simple FRED model; c) Residuals from a FRED + plateau model.While the fit with the FRED model is formally acceptable (χ 2 /d.o.f.= 1.08; 318 d.o.f.), the fit with the hybrid FRED and plateau model is significantly better at the 1.4×10 −11 level, and the systematic residuals around the plateau are minimized.The elapsed time of 0 corresponds to MJD 60362.35007(TT units at NICER).

Table 1 .
(Illiano et al. 2024A J144459.2−604207 from the 2024 February outburst.Note-The Solar System barycenter corrections were performed using the source coordinates determined with Chandra High Resolution Camera observations of the source region(Illiano et al. 2024).

Table 2 .
Best-fit parameters from fitting a FRED or a FRED + plateau model to the type I X-ray burst light curve with a reference time of MJD 60362.350531(2)(TT units at NICER).