MULTIWAVELENGTH OBSERVATIONS OF LS I +61° 303 WITH VERITAS, SWIFT, AND RXTE

LS I +61° 303 is one of the most extensively studied binary star systems in the Milky Way and, although it has been the subject of many observational campaigns, the true nature (i.e., microquasar or binary pulsar) of the system remains unclear. The system can be classified as a high-mass X-ray binary (HMXB) located at a distance of ∼2 kpc; the components of the system consisting of a compact object in a 26.496 (±0.003) day orbit around a massive BO Ve main-sequence star (Hutchings & Crampton 1981; Casares et al. 2005). The motion of the compact object around its main-sequence companion is traditionally characterized by the orbital phase, ϕ, ranging from 0.0 to 1.0. ϕ = 0 is set at JD 2443366.775 (Gregory & Taylor 1978), with periastron passage believed to occur at ϕ = 0.23 ±  0.02 (Casares et al. 2005) or ϕ = 0.30 ± 0.01 (Grundstrom et al. 2007), and apastron passage between ϕ = 0.65 and ϕ = 0.85. Historically, LS I +61° 303 has been an object of interest due to its periodic outbursts at radio (Paredes et al. 1998; Gregory 2002) and X-ray energies (Leahy et al. 1997; Taylor et al. 1996; Greiner & Rau 2001; Harrison et al. 2000). The radio outbursts are well correlated with the orbital phase (Gregory 2002), although the phase of maximum emission can vary between ϕ = 0.45 and ϕ = 0.95. LS I +61° 303 was first identified at gamma-ray energies with the COS-B source 2CG 135 +01 (Hermsen et al. 1977) and has also been identified with the EGRET source 3EG J0241+6103 which also shows evidence for 26.5 day modulation in the GeV band (Massi 2004). More recently, LS I +61° 303 has been detected as a variable TeV gamma-ray source (Albert et al. 2006; Acciari et al. 2008) with maximum emission observed near apastron.

LS I +61° 303 is one of the only three reliably detected TeV binaries: the other two being LS 5039 (Aharonian et al. 2005a) and PSR B1259-63 (Aharonian et al. 2005b). PSR 1259-63 is a confirmed binary pulsar (Johnston et al. 1992a, 1992b) whereas the nature of both LS 5039 and LS I +61° 303 is still under debate. The two main competing scenarios which can explain these systems are microquasar (i.e., nonthermal emission powered by accretion and jet ejection) or binarypulsar (i.e., nonthermal emission powered by the interaction between the stellar and pulsar winds). The microquasar model (for example, Bosch-Ramon et al. 2006) used to describe LS I +61° 303 is supported by evidence for strong jet outflows (Massi et al. 2001). However, this model suffers from the failure to detect blackbody X-ray spectra expected in an accretion scenario. The microquasar scenario has not been ruled out and is still the subject of much theoretical work, for example, see Romero et al. (2007). The binary pulsar model (for example, Dubus 2006) is most strongly supported by VLBA data (Dhawan et al. 2006) which reveal a cometary radio structure around LS I +61° 303 that is interpreted as due to the interaction between the pulsar and Be star wind structures. However, there is currently no detection of pulsed radio or X-ray emission confirming the presence of a pulsar. Possible models for LS I +61° 303 will be discussed further in Section 4.

X-ray monitoring campaigns conducted with RXTE (Harrison et al. 2000; Greiner & Rau 2001), ROSAT (Taylor et al. 1996), Chandra (Paredes et al. 2007), Beppo-Sax, and XMM-Newton (Sidoli et al. 2006) show that LS I +61° 303 is a highly variable hard X-ray source with flux levels modulated with the 26.5 day orbital period, the highest flux usually appearing between orbital phases 0.4 and 0.9. The XMM-Newton observations also detail very fast changes of flux, with fluxes doubling over the span of 1000 s (Sidoli et al. 2006). This result of kilosecond scale variability in the X-ray band has also been shown in Esposito et al. (2007) where the authors analyze Swift observations of LS I +61° 303 taken in 2006. These 2006 Swift observations are reanalyzed and presented in this work.

Chandra observations (Paredes et al. 2007) detail fast variability of the flux levels, while also showing evidence for extended X-ray emission reaching between 5'' and 125 to the north of LS I +61° 303. This provides an indication that particle acceleration may be taking place as far away as 0.05–0.12 pc from LS I +61° 303. Recent RXTE observations (Smith et al. 2009), which cover a total of six orbital cycles, show no strong orbital modulation of the 2–10 keV X-ray flux, but a highly significant correlation between spectral index and flux levels. These observations (which are used in this work) show the presence of three large flares, the largest peaking at a flux value of 7.2 (^+0.1_−0.2) × 10⁻¹¹ erg s⁻¹ cm⁻². Closer examination of these flaring states shows that the X-ray flux from LS I +61° 303 doubles within timescales of <2 s, indicating that the X-ray emission region is less than 10¹¹ cm in extent.

The MAGIC collaboration first detected LS I +61° 303 as a variable TeV source above 200 GeV using observations made in 2005/2006 (Albert et al. 2006). This data set covered six orbital cycles in the phase range ϕ = 0.1–0.8 and a strong gamma-ray flux was detected during orbital phases ϕ = 0.4–0.7, with the observed flux peaking at 16% of the Crab Nebula flux at phase ϕ = 0.6. The source was not detected during other orbital phases (i.e., ϕ = 0.1–0.3 and ϕ = 0.7–0.8), which includes the periastron passage. The extracted photon spectrum from 0.2 to 4 TeV measured by MAGIC is well fit by a power law with differential spectral index α=2.6 ± 0.4_stat+sys.

The MAGIC detection was subsequently confirmed by the VERITAS collaboration which detected the source in >300 GeV gamma rays over five orbital phases (Acciari et al. 2008). Overall, the two published TeV detections on this source indicate that it is only active at TeV energies near the apastron passage of the compact object in its orbit around the Be star. Additional observations conducted by the MAGIC collaboration in 2006 (Albert et al. 2009) sampled a total of four orbital cycles, accruing data in all orbital phases. From these observations, a TeV period of 26.8 (±0.2) days is derived, consistent with the accepted orbital period of the binary.

Although Albert et al. (2008) uses VLBA, Swift, and MAGIC TeV data points to claim a weak correlation between TeV and X-ray points, there has not yet been shown to be any statistically significant correlation between the two bands. Most of the favored models predict TeV emission via the inverse-Compton mechanism, which would result in correlated emission in the X-ray band, so it is important to simultaneously measure the flux at TeV and X-ray energies. Dedicated studies at both X-ray and TeV energies are also necessary to understand the variability of this source across the electromagnetic spectrum.

2.1. VERITAS Observations

2.2. X-ray Observations with RXTE and Swift

3.1. VERITAS Results

The TeV data set used in this work covers two, three, and four-telescope observations made from 2006 September to 2008 February, and includes a total of nine 26.5 day orbital cycles of the binary system (see Tables 1 and 2). Figure 1 shows the TeV light curve from both years binned by orbital phase. In this figure, excesses with significance above 2σ are shown as points with error bars, with all other points being shown as 95% confidence level (Helene 1983) upper limits. The data from the 2006/2007 observing season (taken with both two and three-telescope arrays) comprised a total of 43.6 hr of observations (after data quality selection) during the orbital phases of ϕ = 0.2–0.9 with significant coverage of phases ϕ = 0.4–0.9 (see Acciari et al. 2008). For the entire 2006/2007 data set, the source was detected at the 8.8σ significance level (128 excess events) for emission above 500 GeV. The source was detected as an active TeV source only during apastron phases ϕ=0.5–0.9, with the largest observed fluxes between phases ϕ = 0.6 and ϕ = 0.8 (see Table 1). VERITAS observations made within the same phase range measured flux values ranging from 10% to 20% of the Crab Nebula flux (100% Crab Nebula flux measured as 5.8 × 10⁻¹¹ cm⁻² s⁻¹ above 500 GeV). These observed fluxes are similar to those measured by MAGIC between phases ϕ = 0.6 and ϕ = 0.8 (Albert et al. 2006). The differential photon spectrum extracted from the 2006/2007 observations (Acciari et al. 2008) is well fit by a power law described by dN_γ/dE = (2.39 ± 0.32_stat±0.6_sys)$\times \frac{E}{1 \,\rm{TeV}} ^{-2.4\pm 0.2_{\rm stat}\pm 0.2_{\rm sys}}\times$10⁻¹² cm⁻² s⁻¹ TeV⁻¹ for emission above 300 GeV, in agreement with the spectrum derived from MAGIC observations.

Zoom In Zoom Out Reset image size

Table 1. VERITAS TeV Observations of LS I +61° 303 2006/2007

Date Observed (MJD)	Orbital Phase (ϕ)	Significance (σ)	Flux (>500 GeV) (×10−11 cm−2 s−1)
53999.4	0.31	−0.5	<0.33
54001.3	0.382	0.2	<0.59
54002.4	0.424	0.0	<0.68
54003.4	0.477	1.04	<0.56
54004.3	0.5	0.74	<0.85
54005.4	0.537	−1.49	<0.13
54006.4	0.575	1.14	<0.63
54008.4	0.65	4.39	1.08 ± 0.32
54009.4	0.688	3.81	1.00 ± 0.35
54031.3	0.515	2.04	0.32 ± 0.22
54035.4	0.666	5.4	1.15 ± 0.29
54036.4	0.707	3.62	0.64 ± 0.24
54038.4	0.782	1.57	<0.81
54050.2	0.23	−0.32	<0.49
54054.3	0.383	0.44	<0.36
54055.3	0.42	1.14	<0.63
54056.3	0.458	1.17	<0.81
54059.3	0.57	1.86	<1.09
54060.3	0.61	2.86	0.95 ± 0.46
54061.3	0.65	1.21	<0.92
54062.3	0.685	0.84	<0.99
54064.3	0.756	0.31	<0.35
54066.3	0.836	1.99	<0.59
54067.3	0.87	1.36	<0.73
54108.2	0.42	−0.06	<0.22
54109.2	0.45	2.15	0.32 ± 0.29
54110.2	0.49	0.34	<0.29
54115.1	0.677	3.37	0.71 ± 0.28
54116.1	0.715	2.97	0.39 ± 0.16
54117.2	0.75	2.24	0.53 ± 0.3
54137.1	0.756	−0.84	<0.38
54138.1	0.793	0.59	<0.52
54139.1	0.58	0.46	<0.62
54140.1	0.62	0.46	<0.47
54144.1	0.77	2.24	0.71 ± 0.4
54147.1	0.89	0.27	<0.48

Download table as:  ASCIITypeset image

Table 2. VERITAS TeV Observations of LS I +61° 303 2007/2008

Dates Observed (MJD)	Orbital Phase (ϕ)	Significance (σ)	Flux (>500 GeV) (×10−11 cm−2 s−1)
54381.9	0.73	2.9	0.33 ± 0.11
54382.9	0.77	0.02	<0.25
54383.9	0.81	2.7	0.33 ± 0.12
54384.9	0.84	1.9	<0.77
54389.9	0.03	0.32	<0.42
54390.9	0.07	1.1	<0.56
54393.8	0.18	1.1	<0.58
54404.2	0.59	0.76	<0.64
54406.7	0.67	1.6	<0.32
54407.8	0.71	4.4	0.49 ± 0.11
54408.8	0.75	0.63	<0.27
54409.7	0.78	2	0.27 ± 0.11
54411.9	0.86	−0.06	<0.33
54413.8	0.93	0.0	<0.57
54423.8	0.31	0.11	<0.42
54448.6	0.25	2.5	0.19 ± 0.08
54449.7	0.29	2.9	0.26 ± 0.09
54450.7	0.32	1.9	<0.18
54477.6	0.34	2.1	0.28 ± 1.38
54479.6	0.42	0.52	<0.55

Download table as:  ASCIITypeset image

The 2007/2008 season data set is composed of a total of 20.7 hr of four-telescope observations taken between 2007 October and 2008 January, spanning five separate orbital cycles. From these observations the source was detected at a significance level of 7.3σ (71 excess events) for emission above 500 GeV. These observations cover all orbital phases, however, due to factors such as poor weather conditions not all phase bins were covered with equal exposure time (see Figure 1). The source was significantly detected during the orbital phases of ϕ = 0.7–0.9 (see Figure 1 and Table 2) with observations taken during this orbital phase range detecting 41 excess events in 436 minutes, corresponding to an average flux of 4.0%(±0.6%) of the Crab Nebula flux above 500 GeV at a 6.5σ statistical significance.

For the 2007/2008 observations, the differential photon spectrum extracted from phases ϕ = 0.5–0.8 (the same orbital phases from which the 2006/2007 spectrum was extracted in Acciari et al. 2008) is well fitted by a power law described by dN_γ/dE = (2.43 ±  0.78_stat ±  0.6_sys)$\,\times\, \frac{E}{1\,\rm{TeV}}^{-2.6\,{\pm} \,0.6_{\rm stat}\,{\pm}\, 0.2_{\rm sys}}\times$ 10⁻¹² cm⁻² s⁻¹ TeV⁻¹, in good agreement with the 2006/2007 spectrum (see Figure 2), although the precision on the spectrum is reduced due to more limited statistics.

Zoom In Zoom Out Reset image size

To address the question as to whether there might be some low-level TeV emission occurring outside the apastron passage phases in the 2007/2008 observations, we combined the data taken outside phases ϕ = 0.5–0.9. This data set shows an excess of 28 events in 693 minutes of observation time, which is consistent with an average flux value of 1.7%(±0.6%) of the Crab Nebula flux at a 3.6σ statistical significance. However, we find that this excess is principally due to data from phases ϕ = 0.2–0.3 which show an excess of 15 events in 230 minutes of observation time. This excess corresponds to a flux of 2.7%(±0.8%) of the Crab Nebula at a 3.3σ statistical significance (post-trial significance of 2.8σ for six trials). While we do not consider this result to be statistically significant evidence for TeV emission outside apastron passage, the excess indicates the possibility of a low-level flux which warrants further investigation.

LS I +61° 303 has been demonstrated by previous observations (Acciari et al. 2008; Albert et al. 2006, 2008, 2009) to be variable TeV source. To confirm this result we performed a test for the probability of the source having a constant flux over both years during which the data set presented here was accrued. We tested 200 individual fluxes between 1% and 20% Crab flux strength and computed the corresponding probabilities that these constant fluxes would provide a reasonable fit to the observed nightly flux upper limits and detections based on the observed excess events, lifetime, and calculated effective area of each night's data. We found no probable constant flux fit to the observed data, with the best-fit constant flux corresponding to a 6.3% Crab Nebula flux above 500 GeV. This constant flux value resulted in a reduced χ² value of 4.1 (for 55 degrees of freedom), corresponding to a probability of less than 10⁻¹⁶ that LS I +61° 303 presented a constant flux over the two years of data presented here.

3.2. RXTE and Swift Results

The X-ray flux in the 2006/2007 season, as seen by both instruments, is highly variable, ranging between approximately 0.5 ×10⁻¹¹ erg cm⁻² s⁻¹ and 3 ×10⁻¹¹ erg cm⁻² s⁻¹ over the 26.5 day orbit (see Figures 3 and 4). Similar variability is also present in the 2007/2008 data set which also shows the presence of three exceptionally large X-ray flares, reaching a peak flux on MJD 54356.96 of 7.2×10⁻¹¹ erg cm⁻² s⁻¹ during a ∼500 s integration window (see Figure 4, top panel). This flare is the largest X-ray flare detected from this source, a factor of 2–3 larger than any previously measured. There were several other powerful X-ray flares occurring shortly after this flare in the RXTE data. Further analysis of these flares (Smith et al. 2009) shows that the X-ray flux from LS I +61° 303 undergoes doubling over timescales as short as several seconds, as well as changing by up to a factor of 6 in several hundred seconds. This rapid variability likely explains the apparent disagreement between the RXTE and Swift data points in Figure 4. The fit spectra for both the RXTE and Swift nightly observations are also variable, with indices ranging from 1.4 to 2.6 over the span of the observations. A detailed analysis of the 2007/2008 RXTE data shows that there is a strong correlation between the spectral index and flux values observed from the system (Smith et al. 2009), with the spectrum hardening as luminosity increases.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

3.3. TeV and X-ray Combined

The data presented in this work show that the X-ray and TeV emission from LS I +61° 303 is highly variable over the entire accrued data set from 2006 to 2008. A gamma-ray signal was significantly detected over the 2006/2007 observing season during apastron passage at a peak flux level of 15%–20% Crab Nebula flux (above 500 GeV), and during the 2007/2008 observations at a <5% Crab Nebula flux during apastron passage (above 500 GeV). Although LS I +61° 303 has only been detected during orbital phases ϕ = 0.5–0.9 (including MAGIC observations), there is marginal evidence for emission outside apastron passage as suggested by both the 2007/2008 VERITAS observations as well as the observations conducted by MAGIC (Albert et al. 2009). Given the relatively sparse TeV data detailed in this work, we are unable to place a constraint on periodicity within the TeV signal as reported in Albert et al. (2008).

The X-ray flux from LS I +61° 303 is also variable, with strong outbursts occurring at multiple regions of the orbit. The ZDCF analysis of the quasi-contemporaneous X-ray and TeV data set does not show evidence for a correlation between the two bands. However, due to the lack of dense TeV coverage overlapping with X-ray observations, our sensitivity to such a correlation is inadequate. More specifically, observations conducted with RXTE, Swift, and XMM-Newton (Smith et al. 2009; Paredes et al. 2007; Sidoli et al. 2006) show that the X-ray flux from LS I +61° 303 can change significantly over short timescales (up to a factor of 6 over several hundred seconds). If the X-ray and TeV emission are indeed correlated on fast timescales, truly simultaneous coverage in both bands would be necessary to confirm this correlation.

Although the data presented here do not conclusively rule out or reinforce any of the proposed models (i.e., binary pulsar or microquasar), the derived TeV and X-ray spectra can be compared to recent model predictions from both scenarios. Since both TeV spectra presented here are composed of data taken over several orbital cycles, it is not possible to construct a truly simultaneous or contemporaneous spectral energy distribution (SED) for the data examined in this paper. Instead, the RXTE data which fell between orbital phases 0.5 and 0.8 (the phases from which the TeV spectrum was derived in both seasons) were integrated into a single spectrum and were fit by the same procedure as described above for the nightly RXTE points. This resulted in an X-ray spectrum from 3 to 10 keV, which is well fit by the power law 5.84 (±0.06)×10⁻³ × E^{−1.89±0.05} cm⁻² s⁻¹ keV⁻¹. This spectrum is plotted along with the EGRET spectrum from Hartmann et al. (1999), and the VERITAS TeV spectra from Acciari et al. (2008) in Figure 5.

Zoom In Zoom Out Reset image size

An example of a binary pulsar model with a broadband SED prediction is the compactified pulsar wind scenario of Zdziarski et al. (2008). In this model, the X-ray to TeV emission is powered by inverse Compton scattering off stellar UV photons by energetic electrons injected from the pulsar wind into the fast, clumpy polar wind of the Be star as well as by the more traditional binary pulsar emission mechanism of a shock front located at the interaction point of the Be equatorial and pulsar winds. This model also predicts observable variations in the high-energy emission along the orbit due to accompanying density variations in the Be equatorial wind, as well as inhomogeneities introduced by the mixing of the fast polar and pulsar winds. The model prediction plotted in Figure 5 is a three part broken power law which represents the high X-ray state of the system. The model performs well in predicting the observed SED, with the resulting TeV spectrum prediction of index −2.76 falling within the allowed error range of the VERITAS fit (−2.40 ± 0.39_stat+sys). We note that in binary pulsar models the dominant emission mechanism is dictated by the magnetic field strength at the shock, which is in turn dictated by the so-called "stand-off" distance, or the distance from the compact object to the shock front. When the stand-off distance is smallest (i.e., when the stellar wind strength is greatest) near periastron, the magnetic field strength is the greatest, allowing the synchrotron loss channel into hard X-rays to dominate, quenching the production of TeV gamma rays via the inverse-Compton process. The possible existence of TeV emission near periastron passage calls into question the validity of this prediction in the binary pulsar scenario. If confirmed with further observations, the appearance of periastron TeV emission would necessitate modifications to this model.

An example of a microquasar model which may offer an explanation for the observed emission is that of Gupta & Boettcher (2006). This scenario provides a time-dependent leptonic jet framework that models synchrotron, synchrotron self-Compton, and inverse-Compton (on stellar UV photons) losses resulting from an accretion powered jet within the system. While all mechanisms contribute in varying amounts, the synchrotron contribution dominates the X-ray emission, while the inverse-Compton contribution (both external Compton and synchrotron self-Compton) is only significant above MeV energies. (see the double humped structure in Figure 5). This model examines the hypothesis that the observed TeV variability may be able to be interpreted solely as a geometrical effect arising from absorption in the dense photon field of the star. The regions of highest TeV production, therefore, are those that have limited exposure to this dense absorption field such as the ϕ = 0.5 orbital region which is plotted in Figure 5. As can be seen, while the synchrotron contribution adequately reproduces the observed X-ray spectrum, the inverse-Compton contributions underproduce both the EGRET and VERITAS spectra, predicting a cutoff at a few TeV. This fit could most likely be improved, however, if the constraint of absorption being the only contributing factor to the TeV variability was removed. Given the possible existence of TeV emission outside apastron passage, this would seem to be a necessary modification.

Recently, Swift-BAT has reported the detection of a short (0.23 s), extremely powerful X-ray burst with a luminosity of 10³⁷ erg s⁻¹ in the 15–150 keV energy range within the 90% containment radius of LS I +61° 303 (Barthelmy et al. 2008). While Barthelmy et al. (2008) notes the possibility that this emission episode was due to an unrelated short gamma-ray burst, they claim that the evidence is in favor of activity from a source within LS I +61° 303 (Barthelmy et al. 2008). Further analysis of Barthelmy et al. (2008) by Dubus & Giebels (2008) interprets this burst as evidence for magnetar activity within LS I +61° 303, the first within a HMXB. If confirmed by subsequent observations, this type of extremely powerful bursting may help to resolve the question of the identification of LS I +61° 303.

In conclusion, the multiwavelength observations reported here demonstrate clear variability of LS I +61° 303 in the high-energy regime, however, the TeV data have insufficient sampling to constrain the correlation between the X-ray and TeV bands. Additionally, while neither conclusively ruling out nor confirming either the microquasar or binary pulsar scenarios, the observations reported here show marginal evidence for TeV emission near periastron passage, a feature which, if confirmed, may be necessary to incorporate into future models. Future simultaneous, multiwavelength observations with instruments such as VERITAS and MAGIC in the TeV regime, combined with GeV observations by the Fermi Gamma-ray Space Telescope, and X-ray monitoring by various instruments will aid in the deeper understanding of this unpredictable and exciting source.

This research is supported by grants from the U.S. Department of Energy, the U.S. National Science Foundation, the Smithsonian Institution, by NSERC in Canada, by Science Foundation Ireland, and by PPARC in the U.K. We acknowledge the excellent work of the technical support staff at the FLWO and the collaborating institutions in the construction and operation of the instrument.

The submitted manuscript has been created by UChicago Argonne, LLC, Operator of Argonne National Laboratory ("Argonne"). Argonne, a U.S. Department of Energy Office of Science laboratory, is operated under Contract No. DE-AC02-06CH11357. The U.S. Government retains for itself, and others acting on its behalf, a paid-up nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, by or on behalf of the Government.