NuSTAR detection of 4s Hard X-ray Lags from the Accreting Pulsar GS 0834-430

. The NuSTAR hard X-ray telescope observed the transient Be / X-ray binary GS 0834 − 430 during its 2012 outburst. The source is detected between 3 – 79keV with high statistical signiﬁcance, and we were able to perform very accurate spectral and timing analysis. The phase-averaged spectrum is consistent with that observed in many other magnetized accreting pulsars. We fail to detect cyclotron resonance scattering features in either phase-averaged nor phase-resolved spectra that would allow us to constrain the pulsar’s magnetic ﬁeld. We detect a pulse period of ∼ 12 . 29s in all energy bands. The pulse proﬁle can be modeled with a double Gaussian and shows a strong and smooth hard lag of up to 0.3 cycles in phase, or about 4s between the pulse at ∼ 3 and (cid:38) 30keV. This is the ﬁrst report of such a strong lag in high-mass X-ray binary (HMXB) pulsars. Previ-ously reported lags have been signiﬁcantly smaller in phase and restricted to low-energies (E < 10keV). We investigate the possible mechanisms that might produce such lags. We ﬁnd the most likely explanation for this e ﬀ ect to be a complex beam geometry.

Abstract. The NuSTAR hard X-ray telescope observed the transient Be/X-ray binary GS 0834−430 during its 2012 outburst. The source is detected between 3 -79 keV with high statistical significance, and we were able to perform very accurate spectral and timing analysis. The phase-averaged spectrum is consistent with that observed in many other magnetized accreting pulsars. We fail to detect cyclotron resonance scattering features in either phase-averaged nor phase-resolved spectra that would allow us to constrain the pulsar's magnetic field. We detect a pulse period of ∼ 12.29 s in all energy bands. The pulse profile can be modeled with a double Gaussian and shows a strong and smooth hard lag of up to 0.3 cycles in phase, or about 4s between the pulse at ∼ 3 and 30 keV. This is the first report of such a strong lag in high-mass X-ray binary (HMXB) pulsars. Previously reported lags have been significantly smaller in phase and restricted to low-energies (E<10 keV). We investigate the possible mechanisms that might produce such lags. We find the most likely explanation for this effect to be a complex beam geometry.

NuSTAR observation of GS 0834−430
GS 0834−430 is an accreting Pulsar discovered in 1990 [1]. It is an example of Be/X-ray binary. It has a spin period of ∼ 13 s, an orbital period of 108d with a wide orbit (a x sin i ∼ 128 l − s) [2] around a B0-2 V-IIIe companion [3]. At the peak of its latest outburst [4], on UT 2012 July 11, the source was observed by NuSTAR [5] for about 31ks. In these pointings, at 10 keV the source is 10 3 times brighter than the background, and at 79 keV the source is still 10 times above the background. We could perform very high quality timing and spectral analysis up to 79 keV.
The pulse profile is detected significantly up to ∼ 60 keV. The energy-resolved pulse profile shows clear phase lags between the pulse profiles at different energies. The shift is about 0.3 cycles in phase, or 4 s in time, between 3 and 30 keV. Fitting the profile with a double Gaussian, the two components seem to move together, almost linearly at first, then saturating above 20 keV (Fig. 1). The primary peak dominates at low energy, then the second becomes the stronger one above 30 keV.
The source spectrum varies considerably with pulse phase. We use an NPEX model to characterize the spectrum in 10 phase bins (Fig. 1) of similar hardness ratio. The model fits adequately the spectrum for all phase bins, and there are no signatures that suggest the presence of a cyclotron line. A weak (EW∼ 40eV) iron line is present, mildly anti-correlated with flux. We find no evidence of cyclotron resonance scattering features (CRSF), often observed in accreting pulsars and yielding an approximate measure of the magnetic field in these sources [see 7, for a review]. 06011-p.2

Discussion
Time lags are observed in many accreting sources [see 8, for a review]. They can be produced through delays due, e.g., to Comptonization of the signal [e.g. 9], reflection from a medium far from the source [e.g. 10,11], relativistic effects [e.g. 12], or propagation of signals in a disk [e.g. 13]. But 4s time lags are very difficult to reconcile with the most common geometries proposed for lag production (see the full discussion in [6]).
A CRSF can also produce a similar effect as the one we observe, producing an apparent delay in the spectral emission due to the different contributions of the feature in the various phases of the pulse profile [e.g. 14], but we do not find evidence for CRSFs in the spectrum in the relevant bands.
Similarly, it is possible that the effect is produced by the different emission components coming from the different regions of the accretion column. If the source has a luminosity between ∼ 10 37 erg s −1 , as is the case for GS 0834−430 during this outburst, calculations show that a Compton shock is formed in the accretion column, but that the luminosity is not sufficient to produce a radiation dominated shock, which would decelerate the infalling material high above the neutron star surface [15]. In this intermediate regime, a complex beam profile is therefore produced: radiation, after being upscattered to high energies by inverse Compton scattering inside the accretion column, emerges through the walls of the accretion column (fan beam), as well as along the magnetic field lines (pencil beam). The fan and pencil beam components can in principle have different spectral properties. Given the right geometry, this could create the phase-shift effect seen in the NuSTAR data through a gradual mixing of a softer and harder continuum. These processes are complicated and strongly dependent on the single source physical structure, and this might explain why only this source shows such strong lags.
If this is the case, our observation represents a very interesting step in the understanding of these processes. More details can be found in [6].