An X-Ray and Radio View of the 2022 Reactivation of the Magnetar SGR J1935+2154

Recently, the Galactic magnetar SGR J1935+2154 has garnered attention due to its emission of an extremely luminous radio burst, reminiscent of fast radio bursts (FRBs). SGR J1935+2154 is one of the most active magnetars, displaying flaring events nearly every year, including outbursts as well as short and intermediate bursts. Here, we present our results on the properties of the persistent and bursting X-ray emission from SGR J1935+2154 during the initial weeks following its outburst on 2022 October 10. The source was observed with XMM-Newton and NuSTAR (quasi-)simultaneously during two epochs, separated by ∼5 days. The persistent emission spectrum is well described by an absorbed blackbody plus power-law model up to an energy of ∼25 keV. No significant changes were observed in the blackbody temperature (kT BB ∼ 0.4 keV) and emitting radius (R BB ∼ 1.9 km) between the two epochs. However, we observed a slight variation in the power-law parameters. Moreover, we detected X-ray pulsations in all the data sets and derived a spin-period derivative of Ṗ=5.52(5)×10−11 s s−1. This is 3.8 times larger than the value measured after the first recorded outburst in 2014. Additionally, we performed quasi-simultaneous radio observations using three 25–32 m class radio telescopes for a total of 92.5 hr to search for FRB-like radio bursts and pulsed emission. However, our analysis did not reveal any radio bursts or periodic emission.


INTRODUCTION
Magnetars are a sub-group of isolated neutron stars with ultra-high magnetic fields of B ≈ 10 14 − 10 15 G, whose decay and instability are believed to be the main energy source of their emission (Duncan & Thompson 1992).Magnetars have spin periods P that range between 0.3-12 s and large spin down rates between Ṗ ∼ 10 −13 −10 −11 s s −1 , although magnetar-like emission has also been detected from peculiar pulsars that may not necessarily have P and Ṗ falling within the aforementioned range (e.g., Rea et al. 2010Rea et al. , 2016;;Archibald et al. 2016).Magnetars are persistent X-ray sources with luminosities of L X ≈ 10 31 − 10 36 erg s −1 (for reviews see e.g., Turolla et al. 2015;Kaspi & Beloborodov 2017;Esposito et al. 2021).In addition, they are characterised by transient activities, which may affect the spectral and timing properties of the persistent emission.Based on their duration, these activities can be divided into shortand long-lived events.The former include bursts of tens/hundreds of milliseconds duration and giant flares lasting up to a few minutes, and reaching peak luminosities as high as 10 47 erg s −1 .The latter, known as outbursts, are sudden increases of the persistent X-ray flux by a factor of 10-1000, followed by a gradual decay over a period of months to years (see e.g., the Magnetar Outburst Online Catalog 1 , Coti Zelati et al. 2018).
On 2014 July 5, the Burst Alert Telescope (BAT) on board the Neil Gehrels Swift Observatory (Swift; Gehrels et al. 2004) detected a short burst, leading to the discovery of a new magnetar, SGR J1935+2154 (SGR J1935 in the following; Stamatikos et al. 2014).Follow-up observations enabled the measurement of the source spin period P ∼ 3.24 s and spin-down rate of Ṗ ∼ 1.43 × 10 −11 s s −1 .These values resulted in a surface dipolar magnetic field B ∼ 2.2×10 14 G at the equator, confirming the magnetar nature of the source (Israel et al. 2016).The distance to the magnetar has been the focus of various works.Some of these studies associate SGR J1935 with the supernova remnant G57.2+0.8, for which distances of 6.6±0.7 kpc (Zhou et al. 2020) and ≤ 10 kpc (Kozlova et al. 2016) have been derived.On the other hand, other studies reported a distance of 4.4 +2.8 −1.3 kpc, based on the analysis of an expanding dust-scattering ring associated with a bright X-ray burst (Mereghetti et al. 2020).
Since its discovery, SGR J1935 has been a very active source, experiencing multiple outbursts in 2015, 2016 (twice) and 2020 (see e.g., Younes et al. 2017;Borghese et al. 2020), as well as frequent bursting episodes (e.g., Lin et al. 2020).Additionally, one day after the 2020 reactivation, a short and very bright, double-peaked radio burst (known as FRB 200428) temporally coincident with a hard X-ray burst was observed (CHIME/FRB Collaboration et al. 2020;Bochenek et al. 2020;Mereghetti et al. 2020;Ridnaia et al. 2021;Tavani et al. 2021;Li et al. 2021).This was the first time SGR J1935 was detected in the radio band.The radio burst showed properties similar to those of Fast Radio Bursts (FRBs), providing strong evidence that magnetars may power at least a subgroup of FRBs.
On 2022 October 10-11, multiple short X-ray bursts were detected from SGR J1935 by INTEGRAL , Swift/BAT and other X-ray satellites indicating a reactivation of the source (e.g., Mereghetti et al. 2022;Palmer 2022;Ibrahim et al. 2022).Following this bursting activity, NICER began observing the source and measured a persistent X-ray flux that was about one order of magnitude higher than the quiescent level (Younes et al. 2022b).A new outburst had begun.Similarly to the 2020 outburst, radio bursts with X-ray counterparts were also observed during the initial stage of this outburst (e.g., Maan et al. 2022;Pearlman & Chime/Frb Collaboration 2022;Younes et al. 2022a), but none as bright as FRB 200428.
Here, we report on the X-ray persistent and bursting emission properties of SGR J1935 during the first weeks of the most recent active period, as well as on our searches for single pulses and pulsed emission in quasisimultaneous radio observations.We first summarise the X-ray data analysis procedure in Section 2. We then present the timing and spectral analysis, as well as a search for short bursts in Section 3. In Section 4, we describe our radio observations.Finally, Section 5 presents a discussion of our findings.

X-RAY OBSERVATIONS AND DATA REDUCTION
We report on nearly simultaneous XMM-Newton and NuSTAR observations, carried out between 2022 October 15 and 22. Data reduction was carried out using heasoft package (v6.31;NASA High Energy Astrophysics Science Archive Research Center (HEASARC) 2014) and the Science Analysis Software (SAS 2 , v.19.1.0Gabriel et al. 2004) with the latest calibration files.

XMM-Newton
XMM-Newton observed SGR J1935 twice with the European Photon Imaging Camera (EPIC), for an exposure time of ∼ 40 ks and ∼ 50 ks for the first (ID:0902334101, between 2022 October 15, 19:48:48 UTC, and October 16, 12:06:17 UTC) and the second (ID:0882184001, 2022 October 22 between 03:22:56 and 22:12:09 UTC) observation, respectively.For each observation, the EPIC-pn (Strüder et al. 2001) was set in Small Window mode (time resolution of 5.7 ms) while the EPIC-MOS1 and EPIC-MOS2 (Turner et al. 2001) were set in Full Window mode (time resolution of 2.6 s) and Timing mode (time resolution of 1.75 ms), respectively.Following standard procedures, we filtered the event files for periods of high background activity, resulting in a net exposure of 39 ks and 41 ks for the first and the second pointings.No pile-up was detected.The source counts were extracted from a circle of radius 30 arcsec centered on the source and the background level was estimated from a 60-arcsec-radius circle far from the source, on the same CCD.In this study, our primary focus was on data collected with the EPIC-pn, because of its higher counting statistics owing to its larger effective area compared to that of the two MOS.However, we verified that the MOS data yielded consistent results.A similar region centered on a position uncontaminated by the source emission was used for the extraction of the background events.The light curves, the spectra and the corresponding response files for the two focal 2 https://www.cosmos.esa.int/web/xmm-newton/sasplane detectors, referred to as FPMA and FPMB, were extracted using the nuproducts script.

INTEGRAL
We searched the INTEGRAL archive for data obtained simultaneously with XMM-Newton and NuS-TAR observations.This resulted in 23 pointings where SGR J1935 was in the field of view of the IBIS coded mask imaging instrument.These pointings cover about 60% of the first XMM-Newton observation (from October 15 at 18:51 to October 16 at 04:47 UTC) and 15% of the first NuSTAR observation (on October 19, from 14:43 to 17:45 UTC).We used data from the IBIS/ISGRI detector that operates in the nominal energy range 15-1000 keV providing photon-by-photon data with excellent time resolution of 73 µs.INTE-GRAL data were only examined for the presence of short bursts.

X-ray timing analysis
To perform the timing analysis of SGR J1935, we first filtered out the burst events from the dataset so that they do not affect the integrated pulse profile morphology.We then used the photonphase task of the PINT software (Luo et al. 2021) to assign a rotational phase to the barycentered events by extrapolating the ephemeris from Borghese et al. (2022).In order to use the same fiducial reference phase for the XMM-Newton and NuS-TAR dataset, thus enabling phase coherence across the observations, only photons with energies below 15 keV were analysed.We then combined those events into a stable template profile which we modeled with multiple Gaussian components.Using the photon toa.py tool of the NICERsoft package3 , we extracted barycentric pulse time of arrivals (TOAs) and proceeded to phase-connect the four dataset with the TEMPO timing software (Nice et al. 2015).We achieved coherence across the dataset using a simple model that only has the spin frequency ν and its first derivative ν as free parameter.We show the post-fit residuals in Figure 1 and provide our coherent solution in Table 1.
Using our timing model, we then computed the rotational phase associated with the (barycentric) XMM-Newton and NuSTAR burst epochs (Table A1). Figure 2 shows the burst phases against the integrated pulse profiles observed with both instruments.We find no evidence for a preferred burst rotational phase: the burst cumulative distribution in phase across a full rotation Post-fit residuals (ms) Figure 1.Post-fit residuals of our best-fit coherent timing solution for SGR J1935 (Table 1).
cycle is statistically consistent with a uniform distribution (we determined a p-value > 25% using both an Anderson-Darling and Kolmogorov-Smirnov test).Similarly, Younes et al. (2020) found no obvious clustering at any particular phase for the ∼220 bursts emitted from SGR J1935 during the 2020 reactivation.Figure 3 shows the background-subtracted light curves folded using the timing solution presented in Table 1 as a function of energy for the two epochs.We modelled all the pulse profiles with a combination of a constant plus two sinusoidal functions, with periods fixed to those of the fundamental and first harmonic components.The pulse profile exhibits a simple morphology below 3 keV that evolves to a double-peaked shape at higher energies.At both epochs, the second peak (at phase ∼0.7) becomes more prominent above 10 keV and dominates in the 25-79 keV energy interval.The separation between the two peaks increases with energy for both epochs from ∼0.3-0.35 in phase at soft X-rays (<10 keV) to ∼0.65-0.7 in phase at hard X-rays (>10 keV).Moreover, we detected a phase shift ∆ϕ between the soft (0.3-10 keV) and hard (10-25 keV) energy bands.For the first peak, ∆ϕ 0.3−10/10−25 is 0.13 ± 0.02 cycles during the first epoch, with the hard photons anticipating the soft ones, and it is not significant for the second epoch.While, for the second peak, we determined a shift of ∆ϕ 0.3−10/10−25 = 0.19 ± 0.01 and 0.22 ± 0.01 cycles for the first and second epoch, respectively, with the soft photons leading the hard ones.Finally, we studied the dependence of the pulsed fraction (PF) with the photon energy and its time evolution.The PF was computed by dividing the value of the semi-amplitude of the fundamental sinusoidal component describing the pulse profile by the average count rate.We did not detect any specific trend in the PF, apart from (i) an increase between the 10-25 keV and 25-79 keV bands for both epoch, and (ii) an increase of the 25-79 keV PF between the two epochs.

X-ray spectral analysis of the persistent emission and search for diffuse emission
The light curves of our observations exhibited several bursts, which will be properly investigated in Sec.3.4.In order to exclude the bursts, we filtered out all the events with a count-rate higher than the average countrate during the persistent state.We then used these filtered events to extract the spectra corresponding to the persistent emission only.
The spectral analysis was performed with Xspec (v12.12.0;Arnaud 1996).We used specgroup and grppha tools to group the spectra with a minimum of 50 counts per energy bin for XMM-Newton/EPIC-pn and NuSTAR/FPMA datasets so as to use the χ 2 statistics.In the following fits, we only used NuSTAR/FPMA spectra, but checked that NuSTAR/FPMB gave consistent results.The XMM-Newton spectra were fit in the 0.5-10 keV energy interval, while for the NuSTAR ones the analysis was limited to the 3-25 keV energy band owing to the low signal-to-noise ratio above 25 keV.We adopted the tbabs model with chemical abundances from Wilms et al. (2000) and photoionization crosssections from Verner et al. (1996) to describe the interstellar absorption.
We simultaneously fit the XMM-Newton and NuS-TAR spectra with an absorbed blackbody plus powerlaw model (BB+PL), including a constant to account for cross-calibration between the two instruments (see Figure 4).N H was tied up across all the four spectra, resulting in N H = (2.57± 0.05) × 10 22 cm −2 (reduced chi-square χ 2 ν =1.08 for 567 degrees of freedom (dof)).This value is compatible with those derived in previous studies of SGR J1935 (see e.g., Younes et al. 2017).For each epoch (2022 Oct 15-18 and 22), we linked all the BB+PL parameters across the XMM-Newton and NuSTAR spectra.However, we allow these parameters to vary between the two epochs.Our analysis showed that there were no significant variations for the blackbody parameters between the first and second epoch, with an emitting radius of R BB ∼1.9 km and temperature of kT BB ∼0.4 keV.On the other hand, the photon index slightly changed from Γ = 1.51±0.02 to 1.41±0.02and the PL normalisation decreased by a factor of ∼ 1.5.The 0.5-25 keV observed fluxes were (1.26±0.02)×10−11 and (1.04 ± 0.02) × 10 −11 erg cm −2 s −1 , giving luminosities of (9.17±0.07)×10 34and (7.48±0.07)×10 34erg s −1 .The PL component accounted for ∼ 93% and ∼ 89% of the total luminosity at the first and second epochs, respectively.
We also inspected the data taken from the EPIC-MOS1 detector for diffuse emission.For both epochs, we extracted radial profiles of the X-ray emission up to a distance of 100-150 arcsec from the magnetar, both from the images covering the entire observation duration, and from the images covering variable time intervals following the detection of the brightest X-ray bursts (see Sec. 3.4 for more details).This second type of analysis was aimed at detecting short episodes of diffuse emission possibly associated with scattering haloes produced by the bursts.In no case did we find evidence of emission in excess of that from the magnetar.

Phase-resolved spectroscopy
We performed a phase-resolved spectroscopy of the XMM-Newton and NuSTAR datasets of the magnetar persistent emission.Our aim is to investigate any changes with rotational phase (and time) of the parameters of the spectra corresponding to the two pulse profile peaks.Therefore, we extracted the 0.5-10 keV EPIC-pn and 3-25 keV FPMA spectra from the 0.0-0.5 (peak I) and 0.5-1.0(peak II) phase intervals (see Figure 3).
The phase-resolved spectra were fit simultaneously with the BB+PL model.The column density was held fixed at the phase-averaged value (N H =2.57×10 22 cm −2 ; see Sec. 3.2).The spectral fitting results, reported in Table 2, revealed variations along the spin phase, which can be primarily attributed to fluctuations in the PL photon index.During the first epoch, the variability was more pronounced with the index decreasing from 1.58±0.04for peak I to 1.36±0.04for peak II.In contrast, the second epoch displayed less variability with the index   slightly changing from 1.30 ± 0.04 (peak I) to 1.43 ± 0.04 (peak II).At a given epoch, the BB parameters are consistent with each other in the different phase ranges.

X-ray burst search and properties
We investigated the XMM-Newton and NuSTAR light curves of all observations for the presence of short bursts, applying the method described by Borghese et al. (2020) (see also, e.g., Gavriil et al. 2004).We extracted time series with three different time resolutions (1/16, 1/32 and 1/64 s) in order to identify events of different durations.We classified a time bin as a burst if it had a probability <10 −4 (N N trials ) −1 of being a Poissonian fluctuation of the average count rate, where N is the total number of time bins in a given light curve and N trials is the number of timing resolutions used in the search.We detected a total of 22 and 12 bursts in the XMM-Newton/EPICpn and merged NuSTAR/FPMA+FPMB light curves, respectively.The burst epochs referred to the Solar system barycenter, as well as the burst fluences and durations, are reported in Table A1 and Figure A1 shows the light curves for the two strongest bursts detected in XMM-Newton and NuSTAR data.
We extracted the spectra for those events with at least 25 net counts for XMM-Newton and for the event with the highest counting statistics for NuSTAR (i.e., the burst labelled 80702311002 #9 in Table A1 with 80 net counts).The background level was estimated from time intervals of the same duration in the persistent state.We employed a minimum number of counts to group the spectra that varies from burst to burst depending on the fluence of the burst itself.We applied the chi-square statistic for model fitting, except for the cases where the counting statistic was too low.In such cases, we adopted the W -statistic instead.The spectra were fitted with an absorbed blackbody model, fixing N H to the value obtained from the analysis of the phase-average broadband spectrum.The fit results are reported in Table A1.
Furthermore, for each observation, we extracted a stacked spectrum of all bursts and assigned the spectrum of the persistent-only emission as the background spectrum.We then fit the stacked spectra using the same model we adopted for the spectra of the single bursts (i.e., an absorbed blackbody with N H fixed at 2.57×10 22 cm −2 ).The XMM-Newton spectra were well described by a single blackbody with temperature of kT BB = 1.14 ± 0.06 keV and kT BB = 1.88 ± 0.08 keV for the first and second epochs, respectively.Using the assumed distance of SGR J1935, i.e 6.6 kpc, we obtained radii of R BB = 0.9 ± 0.1 km for the first epoch and R BB = 1.14 ± 0.07 km for the second one.However, this model was unsatisfactory for the NuSTAR spectra, and thus a second blackbody component was added.This resulted in temperatures of kT BB,cold = 0.5 ± 0.2 keV and kT BB,hot = 3.1 ± 0.3 keV for the cold and hot components, respectively, with radii of R BB,cold = 8 +39 −3 km and R BB,hot = 0.27 +0.06 −0.04 km for the first epoch.For the second epoch, the temperatures were kT BB,cold = 0.8 ± 0.3 keV and kT BB,hot = 4 +4 −1 keV with radii of R BB,cold = 1.7 +6.6 −0.5 km and R BB,hot = 0.09 ± 0.03 km.For the INTEGRAL data, the burst search was carried out in the 30-150 and 30-80 keV energy ranges, by examining light curves binned on seven timescales between 10 and 640 ms.Only the pixels that had more than 50% of their surface illuminated by the source were considered in our analysis.Potential bursts were identified as significant excesses above the expected background level derived from a running average.Once identified, these excesses were then examined through an imaging analysis to confirm their authenticity and positional association with the magnetar.This search resulted in the detection of only two bursts.
Among the three bursts seen with XMM-Newton during the INTEGRAL observations (i.e., the bursts labelled 0902334101 #1, #2 and #3 in Table A1), only the brightest one (#3) was detected by INTEGRAL as well.The burst had a fluence of 36.6 counts (30-150 keV) in ISGRI, over a duration of about 90 ms.The light curve is shown in Figure A1.We assume a spectrum described by thermal bremsstrahlung with a temperature of 30 keV, which is commonly used to describe spectrum of magnetar bursts (e.g.Borghese et al. 2019).The resulting average count rate of 406.6 counts s −1 corresponds to a flux of 2.04 × 10 −8 erg cm −2 s −1 .The two bursts detected by NuSTAR (8070231100 #7 and #8) were not visible in the INTEGRAL data.The second burst detected with ISGRI occurred on 2022 October 19 at 15:25:54.037(UTC), during a time gap in the NuSTAR data.Its fluence and duration were 49 counts (30-150 keV) over 200 ms.The rate of 245.0 counts s −1 corresponds to a flux of 1.23 × 10 −8 erg cm −2 s −1 .

QUASI-SIMULTANEOUS RADIO OBSERVATIONS
We observed SGR J1935 using three radio telescopes in Europe: the 25-m RT-1 telescope in Westerbork, the Netherlands (Wb), the 25-m telescope in Onsala, Sweden (O8) and the 32-m telescope in Toruń, Poland (Tr).Observations were carried out at 1.4 GHz, 1.6 GHz (L-band) and 330 MHz (P-band) (see Table 3 for the observational setup).The source was monitored between October 15 and 19, 2022 for a total of 92.5 hr.This number reduces to 60.4 hr when taking into account the overlap between observations at different telescopes.

Single pulse search
We searched the data for FRB-like emission applying the custom pipeline described by Kirsten et al. (2021Kirsten et al. ( , 2022)).
Data is recorded as "raw voltages", also known as baseband data, at each station in .vdifformat (Whitney et al. 2010).This format encapsulates dual circular polarization with 2-bit sampling.In order to search the data, we first create Stokes I (full intensity) filterbank files with 8-bit encoding using digifil which is part of DSPSR (van Straten & Bailes 2011).For observations at L-band, the frequency resolution is 125 KHz, and the time resolution of the filterbank is 64 µs, with the exception of Tr, which has a time resolution of 8 µs.For the P-band observation, these values are 512 µs and 7.8125 KHz, respectively.We mitigated radio frequency interference (RFI) by applying a static mask.This mask is manually determined for each station and observa-tional setup by identifying channels affected by RFI.We then searched the data for burst candidates using Heimdall4 , setting a signal-to-noise threshold of 7. We only searched for bursts within a dispersion measure (DM) range of ± 50 units, with the known DM of SGR J1935 being 332.7206 ± 0.0009 pc cm −3 (CHIME/FRB Collaboration et al. 2020).Burst candidates are subsequently classified using the machine learning classifier FETCH (Agarwal et al. 2020).We use models A & H and set a probability threshold of 50%.The produced burst candidates were then all manually inspected to determine if they are astrophysical or RFI.

Search for pulsed emission
In an effort to detect pulsed radio emission from SGR J1935, we folded our radio data using the ephemeris derived from the X-ray data (see Sec. 3.1).Additionally, we also folded individual scans which were coincident with an X-ray burst.Overall, we had six instances of overlap between X-ray burst detections and radio coverage.Four of these instances were covered by multiple radio telescopes simultaneously (see Table A2 for details).
The radio observations are divided into scans each lasting typically 900 s.We first identified the scan that encompassed an X-ray burst, as well as the scans immediately before and after it, totalling roughly 2700 s of data.We used DSPSR to fold the data based on the ephemeris.Folding was only possible due to the contemporaneous X-ray and radio observations.These folded scans were subsequently combined into a single file using psradd.We then created a diagnostic plot using psrplot to determine the presence of pulsed emission.We validated this method by applying it to observations of the pulsar J1935+1616.

Results
No FRB-like bursts were found in the radio observations.This allows us to calculate a completeness threshold.The completeness threshold is the upper limit on the fluence of a burst that falls below the sensitivity of our instruments and can be derived using the radiometer equation, where (S/N) is the signal-to-noise detection threshold value, T sys G is the System-Equivalent Flux Density (SEFD), W is the width of the burst, n pol is the number of recorded polarizations and ∆ν is the recorded bandwidth.Using Equation 1 and the properties of the radio telescopes listed in Table 3, and assuming a width of 1 ms and a 7σ detection threshold, we can find completeness thresholds of 5 Jy ms for Onsala, 4 Jy ms for Toruń, 7 Jy ms and 46 Jy ms for Westerbork L-and Pband, respectively.Moreover, we folded radio data at the times of overlap between X-ray detections of bursts and we folded all recorded L-band data spread over four days from Westerbork and Toruń, which corresponds to 45.5 hr and 21.9 hr of observations, respectfully.We found no evidence for pulsed radio emission from SGR J1935 using both approaches.We can therefore determine an upper limit on the typical minimum flux density using the following equation: G n pol t obs ∆ν where β is a factor accounting for quantization effects and is approximated to be 1.1 (see Lorimer & Kramer 2004, and references therein); P is the spin period of the source as quoted in Table 1; and W is the width of the folded profile which is assumed to be equal to 10% of the period.A complete overview of all derived upper limits can be found in Table A2.For the Westerbork P-band observation we find a mean flux density limit of 14.86 mJy, while for the L-band observations we find flux density limits between 0.23 − 2.1 mJy for the different telescopes, configurations and integration times.

DISCUSSION
On 2022 October 10-11, the magnetar SGR J1935 entered a new outburst, characterized by the emission of several short X-ray bursts and an increase of the persistent X-ray flux.Moreover, like the previous outburst in 2020, the source emitted a few radio bursts with Xray counterparts (e.g., Younes et al. 2022a).This event is the sixth detected outburst from SGR J1935, making this magnetar one of the most active known so far.
Here, we presented the properties of the X-ray persistent emission and bursts of SGR J1935 during the first weeks of its most recent outburst based on observations obtained with XMM-Newton and NuSTAR.Additionally, we performed searches for single pulses and pulsed emission through quasi-simultaneous radio observations without any successful results.

Flux and spectral decomposition:
The outburst onset was marked by the emission of several short X-ray bursts between 10 and 11 October 2022 (see e.g., Palmer 2022;Mereghetti et al. 2022).Our observations were carried out ∼6 and 12 days later.At both epochs, emission was detected up to 25 keV (see Fig. 4).Hard X-ray emission from SGR J1935 was also seen in a pointing performed ∼5 days after the 2015 outburst onset and was still observed 5 months after the 2020 reactivation (Younes et al. 2017;Borghese et al. 2022).The persistent X-ray spectra were well modeled by the combination of a thermal and non-thermal components.The thermal component was well described by a blackbody model.Its parameters remained stable over time, with a temperature of ∼0.4 keV and radius of ∼1.9 km.The non-thermal component had a power-law shape and its contribution to the total 0.5-25 keV luminosity decreased only marginally from ∼93% to ∼89% in about 5 days.
The quiescent level of SGR J1935 is not known yet.
Here, we adopt the quiescent observed flux derived by Borghese et al. (2022) using a XMM-Newton observation performed on 2014 October 4, i.e.
(8.7 ± 0.3) × 10 −13 erg cm −2 s −1 (0.3-10 keV).The ratio between the 0.3-10 keV observed flux measured during our first observation, (6.45±0.05)×10−12 erg cm −2 s −1 , and that in quiescence is R 2022 ∼ 7.4.Assuming the same quiescent flux and considering the peak fluxes of the previous outbursts measured by Younes et al. (2017) and Borghese et al. (2020), we calculated the same ratio.Upon comparison, we found that R 2022 was greater than the values from the 2014 and 2015 events, which were R 2014 ∼ 4.9 and R 2015 ∼ 5.4, respectively.However, it was lower than the ratios from the May and June 2016 outbursts, which were R 2016May ∼ 9.7 and R 2016June ∼ 16, respectively.Notably, the 2020 reactivation was the most powerful, with a ratio of R 2020 ∼ 49.

Spin-down rate and pulse profile:
We detected the spin period and the spin-down rate using XMM-Newton and NuSTAR datasets, covering the period of 15-22 October 2022.We were able to establish a phase-coherent timing solution (see Table 1).The spin-down rate we inferred was markedly different from those derived during previous outbursts.Specifically, our results indicated that the spin-down rate during the first weeks on the 2022 reactivation ( Ṗ ≃ 5.52(5) × 10 −11 s s −1 ) was a factor of 3.8 times larger than the value measured during the first four months of the 2014 outburst ( Ṗ ≃ 1.43 × 10 −11 s s −1 ; Israel et al. 2016), and 1.5 times larger than the spin-down rate during the 2020 outburst ( Ṗ ≃ 3.5 × 10 −11 s s −1 ; Borghese et al. 2022, see also Younes et al. 2020, Younes et al. 2023).The observed variations in the spin-down rate suggest a notable change in the factors affecting the spin-down, e.g. the magnetospheric geometry and/or the relativistic wind of SGR J1935 during different outbursts.Moreover, changes in the spin-down rate are common during outbursts, indicating changes in the magnetosphere caused by the rearrangement of magnetic fields.To determine the secular spin-down rate of SGR J1935, a targeted monitoring campaign during the quiescence state is needed.The evolution of the pulse profile during the 2022 reactivation of SGR J1935 displays some differences when compared to previous outbursts.The pulse profiles observed in both XMM-Newton and NuSTAR observations exhibits a distinctive double-peaked morphology (see Fig. 3).Notably, the second peak (at phase ∼0.7) becomes more prominent at energies above 10 keV for both epochs.The observed double-peaked structure contrasts with the quasi-sinusoidal shape showed during the 2014 outburst, as reported in XMM-Newton and Chandra observations (Israel et al. 2016).However, it closely resembles that extracted from NuSTAR and XMM-Newton observa-tions taken during the 2020 outburst (Borghese et al. 2020(Borghese et al. , 2022)).The change of the pulse profile from a single-peak shape in the 2014 outburst to a doublepeak shape during the 2022 reactivation may be related to the fact that different regions on the neutron star surface are heated during each outburst.Similarly to the 2014 outburst, we detected an energy-dependent pulse profile phase shift.Slight phase shifts between the peak emissions in the soft and hard X-ray pulse profiles have been observed in a number of magnetars (see e.g., XTE J1810−197 (Borghese et al. 2021), 1E 1547.0−5408 (Coti Zelati et al. 2020), and references therein).This phenomenology is consistent with the widely accepted scenario that magnetars non-thermal X-ray emission stems from resonant inverse Compton scattering of photons emitted from the star surface by charged particles moving along magnetic loops anchored to the crust and corotating with the star (Wadiasingh et al. 2018, and references therein).In this scenario, the hard, nonthermal X-ray emission is expected to be beamed along the loop and to be misaligned (in most cases) to some extent with respect to the soft, thermal X-ray emission pattern from the hot spots on the star surface.The PF increased when shifting from the 10-25 keV to 25-79 keV energy bands at each epochs.We also observed a time-dependent change in the PF for the 25-79 keV and 3-25 keV energy intervals with its value increasing between the two epochs.These results are inconsistent with the findings reported by Israel et al. (2016), where they reported a time independent PF in the 17-21% range.

Pulse profile modelling:
We determine the emission geometry of SGR J1935 by examining the orientation of the hot spot relative to the line of sight and the star's rotational axis.To achieve this, we compared the observed PF to a set of simulated PFs generated using the method outlined by Perna et al. (2001) and Gotthelf et al. (2010).
Our approach involved creating a temperature map on the surface of the star.This map included a uniform background temperature and a single hot spot characterized by a Gaussian temperature profile.The hot spot's orientation with respect to the star's rotational axis was defined as an angle χ, while we also specified the line of sight's orientation as an angle ψ relative to the rotational axis.We then computed the observed phase-resolved spectra by integrating the local blackbody emission from the visible part of the stellar surface.In this calculation, we considered the effects of gravitational light bending, approximating the ray-tracing function (Pechenick et al. 1983;Page 1995)  loborodov (2002).Additionally, we took into account absorption by the interstellar medium.Since our model includes thermal emission only, we restrict our analysis to the energy range 0.3-2 keV where the blackbody component dominates the emission.In this range, the PF is 10.8 ± 1.4 % in the first epoch, and 7.3 ± 1.1 % in the second one.The pulse profile can be modelled using a simple sinusoidal function with a single peak per rotational phase, so in our modelling we consider a temperature map with a single hot-spot.For the temperature and the radius of the hot-spot, we considered the values obtained from the phase-resolved spectral-fit of peak I reported in Table 2.The contribution from the rest of the stellar surface is neglected since it does not contribute significantly to the emission.
We report the results of our analysis in Figure 5.The color map on the χ − ψ plane represents the value of the PF obtained by our modelling using the input parameters from the first epoch.

APPENDIX
The 2022 reactivation of SGR J1935+2154 13 A. LOG OF SHORT X-RAY BURSTS Table A1 lists the epochs, fluence, durations, best-fit spectral parameters and unabsorbed fluxes for the bursts detected in our datasets.The fluence refers to the 3-79 keV and 0.2-12 keV ranges for NuSTAR and XMM-Newton bursts, respectively.The duration has to be considered as an approximate value.We estimated it by summing the 15.625-ms time bins showing enhanced emission for the structured bursts, and by setting it equal to the coarser time resolution at which the burst is detected in all the other cases.
Table A1.Log of X-ray bursts detected in all datasets and results of the spectral analysis for the brightest events.The NH has been fixed to the average value in the spectral fits.a The notation #N corresponds to the burst number in a given observation.
b The flux was estimated in the 0.5-10 keV range for XMM-Newton and NuSTAR.† These bursts were covered by radio observations (for details, see Table A2).Table A2.Limits on the mean flux density Smean after folding the radio data for the entire Westerbork and Toruń observations using the ephemeris as derived in the X-ray analysis.Additionally, we also fold and place upper limits on the flux density in the case of X-ray burst overlap instances.

Figure 3 .
Figure3.Background-subtracted, energy-resolved XMM-Newton/EPIC-pn (black) and NuSTAR/FPMA+FPMB (green) pulse profiles for the 2022 October 15-18 (left-hand panel) and October 22 (right-hand panel) datasets.The dashed line in each panel indicates the best fit for the profiles (for more details, see Sec. 3.1).The vertical grey lines in the last two panels denote the phase intervals adopted for the phase-resolved spectroscopy (for more details, see Sec.3.3).The corresponding pulsed fraction values are reported in each panel.Two cycles are shown for clarity and some pulse profiles have been arbitrarily shifted along the y-axis.

Figure 4 .
Figure4.Spectra of the persistent emission of SGR J1935.The 0.5-10 keV XMM-Newton/EPIC-pn (black) and the 3-25 keV NuSTAR/FPMA (green) spectra are jointly fit with an absorbed blackbody plus power-law model.For each plot: the top panel shows the counts spectra and the best-fitting model; the middle panel shows the E 2 f (E) unfolded spectra and the contribution of the single components (dotted lines); the bottom panel shows the post-fit residuals in units of standard deviations.
e 92.5/60.4 a Wb: Westerbork RT1 25-m, O8: Onsala 25-m, Tr: Toruń 32-m b Effective bandwidth accounting for RFI and band edges.c From the EVN status page.d Using Equation 1, assuming a 7σ detection threshold and a pulse width of 1 ms.e Total time on source accounts for overlap between the participating stations.
Figure 5. Constraints on the emission geometry of SGR J1935, based on the PF measured in the first epooch (15th October 2022).The color scale represents the 0.3-2 keV PF at different angles.The white lines represent the measured value (PF = 10.8 ± 1.4%), while the red lines represent the measured value at the second epoch (PF = 7.3 ± 1.1%).
The white and red contours represent the regions matching the observed PF in the first and second epoch, respectively.Continuous curves represent the central value of the PF, dashed curves represents the 1σ uncertainty regions.While the two regions do not overlap, they are consistent within 2σ.Our analysis suggests two preferable configurations: one where both angles have moderate values (e.g.(χ − ψ) ∼ (25 • − 25 • )) and another where the lineof-sight is near the rotational axis and the hot-spot is almost perpendicular to it.AYI's work has been carried out within the framework of the doctoral program in Physics of the Universitat Autònoma de Barcelona.AYI, FCZ, EP, AM and NR are supported by the H2020 ERC Consolidator Grant "MAGNESIA" under grant agreement No. 817661 (PI: Rea) and Catalan grant SGR-Cat 2021 (PI: Graber).AB acknowledges support from the Consejería de Economía, Conocimiento y Empleo del Gobierno de Canarias and the European Regional Development Fund (ERDF) under grant with reference ProID2021010132 ACCISI/FEDER, UE.FCZ is supported by a Ramon y Cajal fellowship.This work was also partially supported by the program Unidad de Excelencia María de Maeztu CEX2020-001058-M.Research by the As-troFlash group at University of Amsterdam, ASTRON and JIVE is supported in part by an NWO Vici grant (PI Hessels; VI.C.192.045).This work was supported by the NWO XS grant: WesterFlash (OCENW.XS22.1.053;PI: Kirsten).Part of this work has been funded using resources from the INAF Large Grant 2022 "GCjewels" (P.I.Andrea Possenti) approved with the Presidential Decree 30/2022.This work acknowledges support from Onsala Space Observatory for the provisioning of its facilities/observational support.The Onsala Space Observatory national research infrastructure is funded through Swedish Research Council grant No 2017-00648.This work makes use of data from the Westerbork Synthesis Radio Telescope owned by ASTRON.ASTRON, the Netherlands Institute for Radio Astronomy, is an institute of the Dutch Scientific Research Council NWO (Nederlandse Oranisatie voor Wetenschappelijk Onderzoek).We thank the Westerbork operators R.Blauw, J.J. Sluman and H. Mulders for scheduling observations.This work is based in part on observations carried out using the 32-m radio telescope operated by the Institute of Astronomy of the Nicolaus Copernicus University in Toruń (Poland) and supported by a Polish Ministry of Science and Higher Education SpUB grant.

Table 1 .
Coherent timing solution of SGR J1935 derived from the XMM-Newton and NuSTAR data.

Table 2 .
Results of the phase-resolved spectral analysis presented in Section 3.3.

Table 3 .
Observational setup of the radio telescopes.Station a Band Frequency Range Bandwidth b Bandwidth per SEFD c Completeness d Time observed X-ray Station Band Start time aStop time a #Scans Exposure time Smean b The time elapsed between start and stop times is not continuous due to ∼ 10-s gaps between scans.bUsingEquation 2, properties from Table3and assuming a 10σ detection and 10% duty cycle.