Strong X-ray flaring activity of the BL Lacertae source OJ 287 in 2016 October–2017 April Free

ABSTRACT

1 INTRODUCTION

BL Lacertae objects (BL Lacs) form an extreme subclass of blazars which are remarkable for the absence of emission lines, compact radio structure, high, and variable radio-optical polarization, apparent superluminal motion of some components, very broad continuum extending from the radio to the very high-energy (VHE) γ-rays (E > 100 GeV) and strong flux variability in all spectral bands (see Massaro, Paggi & Cavaliere 2011). According to the widely accepted scenario, a jet of magnetized plasma is launched with relativistic bulk velocity from the vicinity of a central supermassive black hole (SMBH), aligned almost along our line of sight, yielding a Doppler boosting of the observed multiwavelength (MWL) flux and decreasing the variability time-scale (Falomo, Pian & Treves 2014).

A broad-band spectral energy distribution (SED) of BL Lacs presents two different components in the log ν–log νF_ν plane. The lower energy ‘hump’ extends from the radio to the X-ray energy range, and its origin is widely accepted as synchrotron emission of a relativistic, magnetized plasma (Falomo et al. 2014). Based on the position of the synchrotron SED peak E_p, BL Lacs are divided into two subclasses (Padovani & Giommi 1995 and references therein): HE-peaked objects [HBLs, peaking at ultraviolet (UV)–X-ray frequencies], and low-peaked objects [LBLs, with E_p situated in the infrared (IR)–optical part of the spectrum]. Moreover, a third subclass of intermediate-energy-peaked BLLs with synchrotron peaks at optical–UV frequencies (IBLs) is also considered (Falomo et al. 2014).

However, there are various models for the origin of the SED higher energy component (extending from the X-ray to the VHE frequencies in LBLs and IBLs): (1) an inverse Compton (IC) scattering of synchrotron photons by their ‘parent’ electron–positron population (so-called synchrotron self-Compton model, SSC; Marscher & Gear 1985); (2) external Compton model (EC, Dermer,Schlickeiser & Mastichiadis 1992); and (3) hadronic models incorporating a production of γ-rays by relativistic protons, either directly (the proton synchrotron model; Mücke et al. 2003) or indirectly (e.g. synchrotron emission from a secondary electron population, produced by the interaction of HE protons with ambient photons; Mannheim 1993). A valid model can be selected through an intense MWL flux variability and interband cross-correlation study: the one-zone SSC model predicts nearly simultaneous variations in both the synchrotron and Compton components, while multizone SSC and hadronic models can explain more complicated MWL behaviour (Fossati et al. 2008).

The internal structure of BL Lacs is mostly unresolved via direct astronomical observations, and only the outer parts of relatively extended, misaligned jets are studied by the Very Long Baseline Array (VLBA; see, e.g. Rector, Gabuzda & Stocke 2003; Piner, Pant & Edwards 2010). An intensive study of MWL flux variability also is an efficient tool for evaluating the sizes of emission zones (based on the light-travel argument). A check of the MWL variability properties and interband cross-correlation aids in solving other fundamental problems, like jet launching and particle acceleration, the separation of the emission zone from the SMBH, jet matter composition, etc. Therefore, BL Lacs represent one of the favoured targets of MWL campaigns performed with different ground-based telescopes and space missions.

OJ 287 (⁠|$z$| = 0.306, Miller, French & Hawley 1978) is one of the best studied and extensively observed in the optical range, showing regular outbursts with a period of ∼12 yr (Silampää et al. 1988; Lehto & Valtonen 1996; Valtonen et al. 2006, 2016), attributed to the central binary SMBH system in which the orbit of the secondary BH is extremely eccentric, crossing the accretion disc of the primary one twice per each encounter in every 12 yr and triggering an enhanced accretion to the primary BH which can yield the propagation of relativistic shocks though the jet. As a result, the source shows a two-peak optical outburst during each impact (Valtonen et al. 2016). This model was corroborated by radio monitoring data (Valtaoya et al. 2000) which revealed differences between the first and second bursts, hinting at the possibility that the first burst is caused by a disc-crossing while the second burst is related to enhanced accretion causing a shock front in the jet. Recently, the predicted first peak of the outburst was observed in 2015 December (which was the brightest optical level in 30 yr), followed by the comparable second peak ∼3 months later (Gupta et al. 2017). The source showed another optical outburst (expected in the case of the repeated passage of the secondary BH through the accretion disc of the primary BH) during 2017 September–December (Kushwaha et al. 2018a). Additionally, a 60-yr variability was claimed by Valtonen et al. (2006), and possible shorter term periodic behaviour in the radio–optical range was suggested by Gupta et al. (2012) and Hughes et al. (1998). Based on 120 VLBA observations during 1995 April–2017 April, Britzen et al. (2018) found that the parsec-scale radio jet of OJ287 is possibly precessing on a time-scale of ∼22 yr. Half of the jet-precession time is of order the dominant optical periodicity time-scale. In addition, the 14.5 GHz single-dish data are consistent with a jet-axis rotation on a yearly time-scale.

Recently, OJ 287 was detected at VHE γ-rays at >5 standard deviations above background during the VERITAS observations performed in 2017 February (O’Brian et al. 2017).

In this paper, we present the results of a detailed X-ray timing and spectral analysis of OJ 287 focused on the epoch of its significantly enhanced X-ray flaring activity during 2016 October–2017 April. During this period, the source was monitored extensively with X-ray telescope (XRT, Burrows et al. 2005) onboard Neil Gehrels Swift Observatory (Gehrels et al. 2004). Although these data were included in the study of Kushwaha et al. (2018a), they were used in the search for multiband correlations, as well as for the construction and modelling of the broad-band SED. On the basis of these observations, we have studied X-ray flares on time-scales from several days to a few weeks in particular parts of the aforementioned period, and checked the contemporaneous MWL behaviour of the source using all the publicly available data obtained with (i) Ultraviolet–Optical Telescope (UVOT; Roming et al. 2005) onboard Swift; (ii) Burst Alert Telescope (BAT; Barthelmy et al. 2005) onboard Swift; (iii) Monitor of All Sky X-ray Image (MAXI; Matsuoka et al. 2009); (iv) Large Area Telescope (LAT) onboard Fermi (Atwood et al. 2009); (v) various ground-based optical telescopes; and (vi) the 40-m telescope of Owens Valley Radio Observatory (OVRO; Richards et al. 2011). For comparison, we also have included the XRT observations preceding (2016 April–June) and following (2017 May–June) the epoch of strong X-ray flaring activity. We have also performed an extensive study of the X-ray variability on intraday time-scales and present the 0.3–10 keV spectra which show an upward curvature.

The paper is organized as follows. Section 2 describes the data processing and analysing procedures. In Section 3, we provide the results of a timing analysis and those from the X-ray spectroscopy in Section 4. We discuss our results in Section 5, and provide our conclusions in Section 6.

2 OBSERVATIONS AND DATA REDUCTION

2.1 X-ray data

We retrieved the raw data obtained with the grazing incidence Wolter I telescope Swift-XRT (Burrows et al. 2005) from the publicly available archive, maintained by HEASARC.¹ The Level 1 unscreened XRT event files were processed with the xrtdas package developed at the ASI Science Data Center (ASDC) and distributed by HEASARC within the heasoft package (version 6.22.1). The information about each pointing and the measurement results are presented in Table 1.² The task XRTPIPELINE was launched with standard screening criteria using the XRT CALDB calibration files version 20171113.

During 2016 April 23–2017 June 13, the source was targeted 143 times with XRT, yielding a total good time interval of 137 ks after the standard screening. The majority of these pointings (78 per cent) were performed in the windowed timing (WT) mode characterized by compressing 10 rows of the X-ray CCD into a single one in the serial register and then reading out only the central 200 columns of the CCD.³ We selected the events with 0–2 grades. The source and background light curves and spectra were extracted with xselect using circular areas with radii of 12–25 pixels depending on the source brightness and exposure. In the case of several observations (e. g. ObsID 30901034, 2016 November 1), the image centre of OJ 287 was just at the edge the observational area and we excluded them from our study to avoid an incorrect reconstruction of the point spread function (PSF). The light curves were then corrected using XRTLCCORR for the resultant loss of effective area, bad/hot pixels, pile-up, and vignetting. The ancillary response files (ARFs) were generated using the XRTMKARF task, with corrections for PSF losses, different extraction regions, vignetting, and CCD defects.

The large field-of-view (FOV, 1.4 sr; Barthelmy et al. 2005) instrument Swift-BAT observed OJ 287 in the 15–150 keV energy range during our period of study in the framework of the Hard X-ray Transient Monitor program⁵ (Krimm et al. 2013). However, the source generally is very faint in this energy range and the daily-binned data do not yield detections with 5σ significance (the threshold generally applied to coded-mask devices). Using the tool REBINGAUSSLC from heasoft, we rebinned these data within the time intervals 1–4 weeks. However, OJ 287 was detected only few times with 5σ significance in those cases and, therefore, we have not included the BAT data in our study. A similar situation occurred with the 2–20 keV observations performed with X-ray slit cameras of the MAXI mission: the publically available, weekly binned data⁶ do not show a detection of OJ 287 with 5σ significance.⁷ Using the online tool MAXI ON-DEMAND PROCESS,⁸ we extracted the weekly binned light curve of OJ 287 in the 2–6 keV band which generally yields the highest signal-to-noise ratios from the MAXI data, although the source was not detectable with the aforementioned significance during 2016 October–2017 June also in that case.

2.2 UV, optical, and radio observations

The OVRO 40-m telescope uses off-axis dual-beam optics and a cryogenic high electron mobility transistor (Richards et al. 2011). The regular OVRO observations of OJ 287 at the frequency 15.0 GHz have been carried out since 2008 January 9. We retrieved the publicly available OVRO data of our target from the corresponding website.¹¹ The details of the data reduction and calibration procedure can be found in Richards et al. (2011).

We also have used the archival 15 GHz data obtained in the framework of Monitoring Of Jets in Active galactic nuclei with VLBA Experiments¹² (MOJAVE). The details of the data reduction and calibration procedure are provided in Lister et al. (2009). For the construction of long-term radio light curve, we also used the 14.5 GHz observations performed with the the 26-m telescope of University of Michigan Radio Astronomy Observatory¹³ (UMRAO; see Aller et al. 1985 for details).

2.3 γ-ray observations

Fermi-LAT generally monitors the entire sky in the energy range of 20 MeV–300 GeV every 3 h, and this instrument is characterized by the energy resolution better than 10 per cent and an FOV of ∼2.4 sr, with an angular resolution (68 per cent containment angle) better than 1 deg at the energies higher than 1 GeV (Atwood et al. 2009). We extracted the 0.1–300 GeV fluxes from the LAT observations selecting the events of the diffuse class from a region of interest (ROI) of radius 10 deg, centred on OJ 287, and processed with the Fermisciencetools package version v11r5p3. For this purpose, we used the instrument response function and P8R2_V6 and the unbinned maximum likelihood method GTLIKE.¹⁴ A cut on the zenith angle (>100 deg) was done to reduce contamination from the Earth-albedo γ-rays. The data taken when the rocking angle of the spacecraft is larger than 52 deg are discarded to avoid a contamination from the Earth’s limb photons.

A background model including all γ-ray sources from the Fermi-LAT 4-yr point source catalogue (3FGL, Acero et al. 2015) within 20 deg of OJ 287 was created. The spectral parameters of sources within the ROI were left free during the minimization process, while those outside of this range were fixed to the 3FGL catalogue values. The Galactic and extragalactic diffuse γ-ray emission as well as the residual instrumental background were included using the recommended model files gll_iem_v06.fits and iso_P8R2_SOURCE_V6_v06.txt. The normalizations of both components in the background model were allowed to vary freely during the spectral fitting. For the spectral modelling of our target, we adopted a log-parabola model, as done in the 3FGL catalogue. Its detection significance σ was then calculated as σ ≈ (TS)^1/2 (Abdo et al. 2009). When the source was not detectable at the 3σ significance, we calculated the upper limit to the photon flux using the CALCUPPER tool included in the user-contributed package latanalysisscripts (version 0.2.1).¹⁵

3 FLUX VARIABILITY

3.1 Long-term X-ray and MWL flux variability 2016 April–2017 June

OJ 287 showed a strong X-ray outburst during 2016 October–2017 April: the 0.3–10 keV count rate from the XRT observations showed an increase by a factor of ∼10 on 2016 October 19 (MJD 57681) compared to the quiescent level in the 2016 April–May period (Fig. 1a). Consequently, it became the sixth LBL/IBL source with the rate higher than 1 counts s⁻¹ (after S5 0716 + 714, BL Lacs, 3C 66A, VER J0521 + 212, and AO 0235 + 16; see Table 3).Afterwards, the source continued a strong X-ray flaring activity on weekly time-scales and attained its highest historical 0.3–10 keV brightness corresponding to 1.89 ± 0.03 counts s⁻¹ on 2017 February 2 (MJD 57786; see Fig. 2a). The mean count rate during 2016 October–2017 April |$\overline{\text{CR}}$| = 0.932 ± 0.002 counts s⁻¹, while this quantity was 4.5 times smaller during the previous 300 XRT observations of OJ 287 and the highest 0.3–10 keV rate was 0.38 ± 0.10 counts s⁻¹ in that period (recorded on 2009 March 5; MJD 54896). Along with the strong optical outbursts in 2005–2006 and 2015 December–May (attributed to the impact of the secondary BH to the accretion disc of the primary one; see Figs 1e–f), the source showed low X-ray activity.¹⁶ Unfortunately, the source was not observed with Swift for almost 4 months due to the Sun constraint during 2016 June14–October 11 when it showed high states in the optical R band, IRJ band, and at 15 GHz (see Figs 2f–h, as well as Kushwaha et al. 2018a, Valtonen et al. 2017 and the R-band light curve provided on the website of Tuorla Observatory¹⁷).

Figure 1.

Historical light curves of OJ 287 from the MWL observations in 2005–2017 with XRT (top panel), Fermi-LAT (panel b), UVOT bands UVW2, UVW1, and B (panels c–e), R band of the Johnson–Cousins system (various ground-based telescopes; panel f), IR J band with the SMARTS 1.3-m telescope (panel g), and radio observations with the UMRAO and OVRO telescopes (panel h). In the latter case, grey and black points are used for the UMRAO and OVRO data, respectively. For the LAT observations, we used two-weekly bins, while daily bins are adopted for those performed with other facilities. Grey triangles in panel (b) represent 2σ upper limits to the LAT flux when the source was detected below the 3σ significance. The 2016 April–2017 June period is indicated by vertical dashed lines.

Open in new tabDownload slide

Figure 2.

The same as Fig. 1 for the period 2016 April–2017 June. The triangles in panel (b) stand for 2σ upper limits to the LAT flux when the source was detected below the 3σ significance. The vertical dashed lines indicate the periods discussed in Section 3.2.

Open in new tabDownload slide

Similar to X-rays, the source showed a strong outburst by factors of 4.6–6.5 in the UVOT bands V–UVW2 (Figs 1c–e) in 2016 October and kept its high optical–UV states during the next months, although its flaring activity was not as strong as in the 0.3–10 keV energy range in this period (see Figs 2c–e). A similar behaviour was observed also in the optical R and IR J bands, although the flare in 2016 October was characterized by lower amplitudes (by factors of 2.7–3.5) compared to those at the higher frequencies (Figs 2f–g). Finally, a radio-band flare by a factor of 2.9 was observed along with the X-ray one and the source showed its 15 GHz peak on MJD 57813, delayed by 27 d from the highest historical 0.3–10 keV state (see Figs 1h and 2h). Moreover, OJ 287 exhibited its lower radio states during the UV–IR outburst in 2015 December–2016 May.

Since the start of Fermi operations (2008 August), OJ 287 mostly was a faint γ-ray source at HE γ-rays (E >1 MeV) and, therefore, we had to use two-weekly LAT binned data to construct the historical 0.1–300 GeV light curve (Fig. 1b). However, the source was not detectable with 3σ significance or the parameter N_ped (the predicted number of the photons) showed values less than 10 (yielding less credible values of the derived photon flux; see, e.g. Raiteri et al. 2013) on 22 occasions, and we plotted a 2σ upper limit in Figs 1(b) and 2(b). In the case of one-weekly binned LAT data, we applied 2σ upper limits to the 0.1–300 GeV flux in 43 per cent of the cases during 2016 April–2017 June (see Table 4). In contrast to other spectral bands, the source did not show an outburst in the LAT energy range during the here-presented period and underwent only a moderate 0.1–300 GeV flare in 2016 July–August when it was not observed with Swift. On average, the LAT-band flux in 2016 October–2017 April was one of the lowest during the entire 2008–2017 period, and the source showed its strongest γ-ray outburst during MJD 55800–56920 (2011 August–December): the 0.1–300 GeV flux exceeded a level of 5 × 10⁻⁷photons cm⁻² s⁻¹ (see Fig. 1b) and was brighter than 10⁻⁶photons cm⁻² s⁻¹ when using shorter time bins (Hodgson et al. 2017).

Figure 3.

The variability of the MWL flux in different intervals. Grey triangles in LAT-band panels stand for upper limits to the LAT flux when the source was detected below 3σ significance.

Open in new tabDownload slide

3.2 Shorter term flares

During interval 1 (see Table 5 for the corresponding time range), the source was observed only eight times with Swift, showing a typical 0.3–10 keV level corresponding to CR ∼0.15 counts s⁻¹ and a brightening by ∼50 per cent during the last two XRT pointings (Fig. 3a, top panel). A similar behaviour was observed also during the contemporaneous optical–UV observations, although a low-amplitude flare was observed in this energy range in the beginning of this period (see panels 4 and 5). Meanwhile, OJ 287 demonstrated lower LAT-band states and showed an HE peak 3 weeks after the last XRT pointing. During interval 1, the 15 GHz brightness underwent a long-term increase and exhibited its highest level with ∼1.5 months delay with respect to the LAT-band peak.

At the beginning interval 2, the source was found in a highly elevated X-ray state, superimposed by shorter term flares. Namely, it showed a flare by a factor of 2.3 in 5 d, followed by a drop to the initial level in the same time (Fig. 3b). A similar behaviour was observed also in the optical–UV energy range. Afterwards, the source underwent two subsequent lower amplitude X-ray flares along with long-term optical–UV decline. Two subsequent, double-peak flares by factors of 2.3–2.8 occurred in interval 3 (Fig. 3c). The source showed optical–UV flares along with X-ray ones, although the corresponding light curves do not show a double-peak structure. The 15 GHz flux showed an increase along with the second X-ray flare, while the LAT band did not exhibit any correlation with those in other energy ranges.

In interval 4, the 0.3–10 keV rate increased by a factor of 3.1 in 6 d to the highest historical level, and then showed a much longer decline to the initial level (lasting ∼40 d), superimposed by low-amplitude brightness fluctuations (Fig. 3d). The optical–UV light curves show a correlated behaviour, although with some time shifts. The highest LAT-band state was observed in the epoch of the highest UVOT-band states, although the brightness increase was considerably smaller (followed by uncorrelated variability). Two subsequent X-ray flares were observed in interval 5 and the second, considerably stronger one was not accompanied by enhanced activity in the optical–UV energy range: the source exhibited a long-term declining trend in that epoch (Fig. 3e).

In interval 6, the source did not show as high X-ray states as in the previous intervals, and its behaviour was characterized by brightness enhancements by only 40–80 per cent on weekly time-scales (Fig. 3f). The optical–UV fluxes showed little correlation with the 0.3–10 keV one and the source did not exhibit a significant variability (similarly to the previous time interval).

3.2.1 Intraday variability

We have also performed an intensive search for intraday X-ray variability (IDV, i.e. a flux change within a day; see Wagner & Witzel 1995) from the XRT observations performed during 2016 April–2017 June, using the χ²-statistics introduced by Kesteven, Bridle & Brandie (1976). For light-curve construction, we used time bins of 60–240 s (depending on the target’s brightness) and performed bin-per-bin subtraction of the background signal from the source one (by means of the xronos task LCURVE, included in heasoft).

The source showed 32 instances of 0.3–10 keV IDVs whose summary is provided in Table 6. The latter contains the values of reduced χ², fractional variability amplitude, the ranges of the spectral parameters a, b, E_p, and HR (defined in Section 4) for each instance, derived from separate orbits or segments of the corresponding XRT observation.

Figs 4(a)–(c) present the X-ray behaviour of the source during its most densely sampled observations performed during 2017 March 11–31, with a total exposure of 20.8 ks distributed over 22 XRT pointings. During this campaign, the source showed nine IDVs (see Table 6): a flux variability within 1 ks exposures was recorded during ObsIDs 34934022 and 34934027 (panels 1 and 6 of Fig. 4a, respectively); ObsIDs 34934030, 34934032, 34934033, and 34934035 (panels 2, 4, 5, and 7 of Fig. 4b, respectively);ObsIDs 34934039 and 34934041 (panels 4 and 6 of Fig. 4c, respectively). Moreover, two another IDVs occurred between two different orbits of ObsID 34934023 (panel 2 of Fig. 4a). These instances were characterized by F_var = 24.7(4.9)–59.7(5.8) per cent and ranges of the spectral parameters (see Section 4.1 for their definitions): Γ = 2.30(0.11)–2.85(0.15), and HR = 0.18(0.05)–0.56(0.10).

Figure 4.

The 0.3–10 keV light curves from the most densely sampled XRT observations of OJ 287 during 2016 April–2017 June.

Open in new tabDownload slide

The source showed seven 0.3–10 keV IDVs during the XRT campaign in 2017 February 15–27 (ObsIDs (3493400)2–8, panels 3–7 of Fig. 4(d) and panels 1 and 2 of Fig. 4e) which were observed within the exposures ≲1 ks. However, they showed lower fractional amplitudes than those recorded during the aforementioned campaign (F_var = 14.7(3.9)–23.9(4.8)) and narrower ranges of the spectral parameters (Γ = 2.30(0.08)–2.75(0.07), and HR = 0.22(0.03)–0.47(0.06)). Two IDVs occurring within 1 ks intervals were observed during the campaign of 2016 October 15–21 (ObsID 30901221 and the first orbit of ObsID 30901227, panels 2 and 7 of Fig. 4f, respectively).

Other interesting 0.3–10 keV IDVs, mostly occurring within 1 ks exposures are presented in Fig.  5. However, the instances, provided in the bottom row, were characterized by relatively large background which was comparable to the count rates corresponding to the minima of light curves (amounting to 60–80 per cent of the background-subtracted signal). Therefore, these IDVs should be treated with caution.

Figure 5.

Other very fast 0.3–10 keV IDVs in OJ 287 during 2016 April–2017 June.

Open in new tabDownload slide

OJ 287 showed eleven 0.3–0 keV IDVs at the 99.5 per cent confidence level (so-called possible variability, Andruchow, Romero & Cellone 2005; see Table 6). Note that the majority of these instances belong to the epochs of the intensive XRT campaigns performed during 2016 October 15–21 and 2016 March 11–31 (discussed above). Finally, nine instances of optical–UV IDVs were detected at the 99.5 and 99.9 per cent confidence levels with UVOT, characterized by F_var = 7.3(1.9)–16.6(1.6) per cent (see Table 7).

4 RESULTS FROM THE SPECTRAL ANALYSIS

We performed the 0.3–10 keV spectral analysis using the xspec package included in heasoft. For this purpose, the latest response matrix from the XRT calibration files from Swift CALDB and ARFs were adopted. The latter were generated by means of the XRTMKARF task to account for the PSF losses, different extraction regions, vignetting, and CCD defects. The instrumental channels were combined to include at least 20 photons per bin by means of the ftools task GRPPHA (enabling the use of χ²-statistics). However, in the case of the low X-ray state of the source and short exposure, the corresponding spectrum was very poor and we did not group channels in order to use Cash statistics.