Temporal Analysis and discussion

Applying the periodicity analysis techniques described in Section 2.3 leads to the detection, at high significance, of a well defined 8.447±0.002 day orbital period in the ∼12.8 Ms optimally filtered IBIS dataset of IGR J17354−3255. The LS

0 50 100 150 Period (Days) 0 20 40 60 80 Power 99.999% Confidence Level

Figure 5.9: Lomb-Scargle periodogram of the optimally filtered IGR J17354−3255 IBIS light curve showing a strong peak at 8.447±0.002 days. The 99.999% confi- dence level is also drawn at a power of 20.21 as derived by the method outlined in Section. 2.3.1.

providing a strong confirmation of the BAT detection reported by D’A`ı et al. (2011). The phase-folded light curve of IGR J17354−3255, computed using the determined orbital period and a zero phase ephemeris of MJD 52698.205, is shown in Fig. 5.10 and displays a smooth flux modulation that peaks at at an orbital phase of φ ∼ 0.1 whilst becoming consistent with zero at φ ∼ 0.6. The shape of this profile is likely generated by the orbital motion of a NS in an eccentric orbit with orbital phases of 0.1 and 0.6 relating to the passage of periastron and apastron respectively. The orbital phase locations of the IBIS outbursts of IGR J17354−3255 identified by Sguera et al. (2011) are indicated by the green points in Fig. 5.10 and are also seen to be detected at times consistent with the periastron passage of the compact object in the system. Loose constraints can be placed on the dynamically allowed orbital configurations through the L1 point separation as was performed for XTE J1739−302 (Section 5.1.4). In the case of IGR J17354−3255, however, there are no independent constraints on the stellar parameters of the supergiant

companion. Hence assuming, as for XTE J1739−302, a supergiant mass of 25 M

results in IGR J17354−3255 possessing an orbit with a semi-major axis of 52 R

and requiring e < 0.4 to allow for a supergiant radius of ≥ 17 R.

Given the smooth orbital emission profile and the well defined orbital period of IGR J17354−3255, a ‘Recurrence Analysis’, first described by Bird et al. (2009), can be performed. This process utilises the orbital parameters of a system to identify all of the periastron passages observed in the IGR J17354−3255 dataset and calculates the significance of the light curve over a certain duration to assess the regularity of emission production at this orbital phase. The same process is also performed on all

Figure 5.10: The orbital phase-folded light curve of IGR J17354−3255 using the best determined orbital period of 8.447 days and a zero phase ephemeris of MJD 52698.205. The green points illustrate the orbital phase location of the out- bursts of the source detected by IBIS (as defined by Sguera et al. 2011) and the red points the phase location of the flares of the spatially co-incident transient MeV source AGL J1734−3310 (Bulgarelli et al. in prep., private communication).

of the identified apastron passages to estimate the significance distribution generated by random noise in the data and the percentage of active periastron passages quantified. The periastron (red) and apastron (black) distributions derived from the IGR J17354−3255 dataset using a window of 3 days (periastron ±

1.5 days) are shown in Fig. 5.11. The apastron distribution is seen to peak at, and be approximately symmetric about, zero, with no detections above 3σ recorded during these times. The periastron distribution, however, displays an excess of detections above 3 sigma, rising to ∼9σ at its most significant, which corresponds to activity being detected on 26% of periastron passages. This is the second highest recurrence rate measured after that of SAX J1818.6−1703, which produces

detectable emission on ∼50% of periastron passages (Bird et al., 2009).

Taking into consideration the shape of the orbital phase-folded light curve, the clustering of hard X-ray outbursts around the periastron passage and the

recurrence analysis, it is suggested that IGR J17354−3255 is an intermediate SFXT system similar to SAX J1818.6−1703, but with a shorter orbit and a lower

eccentricity of ∼0.1−0.2 to account for the reduced recurrence rate. Compared to some other SFXTs the shape of IGR J17354−3255’s orbital emission profile is quite smooth. Given that the IBIS dataset used covered ∼ 400 orbital cycles of IGR J17354−3255 (at random sampling) and the recurrence analysis suggests emission detectable with IBIS is produced on approximately a quarter of orbital cycles, it is unfeasible that the 16 bright outbursts, which have luminosities in excess of 1036 erg s−1 and represent the most luminous events, are responsible for the smooth

!10 !5 0 5 10 Detection Significance 0 10 20 30 Frequency Periastron Apastron

Figure 5.11: The distribution of significances for IGR J17354−3255 generated at times of presumed periastron (red) and apastron (black) when summing a window of 3 days about the central periastron date. It is seen that the distribution of apastron passages is centred on zero and displays no detections above the 3σ whereas the distribution of periastron passages shows an excess above 3σ that relates to activity on 26% of orbits.

orbital profile observed. Instead the shape is attributed to a lower level of X-ray emission that is below the sensitivity of IBIS in an individual ScW but which sums to a significant level when the whole dataset is folded. This emission could either be smoothly varying, following a similar profile during each orbit that is slowly

accumulated to a detectable level, or the superposition of many low-intensity flares at X-ray luminosities of ∼1033_{− 10}34_{erg s}−1_{, where the probability of a flare being}

generated is a function of the orbital phase which, again, cumulatively generates the observed orbital profile. Unfortunately, due to the sensitivity limits of

INTEGRAL/IBIS on short timescales, it is not currently possible to determine which of these processes is generating the smooth orbital flux variations observed in IGR J17354−3255. In addition to the outbursts and lower level variations, the orbital phase of the quiescent flux upper limit detected with XMM-Newton (Bozzo et al., 2012) is calculated as φ ∼ 0.66 which is consistent with the apastron region of IGR J17354−3255’s orbit. However, the variation in the density of a smooth stellar wind as a function of orbital phase is again insufficient to generate the observed X-ray dynamic range in IGR J17354−3255. Assuming an orbital eccentricity of 0.4 (i.e. at the upper limit of the likely values) and using the same continuity

considerations as for XTE J1739−302 results in a maximum variation in smooth stellar wind density of only a factor of 10 between periastron and apastron in IGR J17354−3255. This variation is again incompatible with Bondi accretion of a smooth medium in this system, implying the influence of a highly structured wind to generate the ∼103 dynamic range observed in the X-ray band. Due to the lack of

dynamical constraints on the IGR J17354−3255 system, a deeper investigation into the cause of the X-ray variability observed and the possible action of accretion barriers is not possible. It is of note, however, that the presence of structure within the supergiant stellar wind has been shown to be required in systems that span over a decade of the orbital period parameter space of SFXTs.