EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 654 (October 2021) A&A, 654 (2021) A44 Full HTML
Free Access
Issue
A&A
Volume 654, October 2021
Article Number A44
Number of page(s) 8
Section Stellar atmospheres
DOI https://doi.org/10.1051/0004-6361/202140451
Published online 08 October 2021

© ESO 2021

1 Introduction

Binary systems are natural environments to study shocks under variable, periodical conditions. An emerging class of these systems is colliding-wind binaries (CWBs), which are massive stellar systems with powerful stellar winds (De Becker & Raucq 2013). Such strong winds eventually interact, forming a bow-shocked wind collision region (WCR) delimited by two separated shock fronts surrounding the star with a weaker wind (Eichler & Usov 1993). Diffusive shock acceleration (DSA) can occur under these circumstances, accelerating particles up to high energies and leading to the emission of non-thermal radiation (Benaglia & Romero 2003; Reimer et al. 2006; De Becker 2007; Reitberger et al. 2014; Grimaldo et al. 2019; Pittard et al. 2020).

One of the most luminous and intriguing known Galactic sources is η Carinae, which has been monitored at different wavelengths for decades. It is a CWB with a primary luminous blue variable (LBV) star with M⊙ (Hillier et al. 2001). Its companion has not been directly observed, but inferred from its orbital variability to be an O or Wolf Rayet (WR) star with M⊙ (Hillier et al. 2001; Mehner et al. 2010). η Car is located in the Carina arm at a distance of 2350 ± 50 pc (Smith 2006), surrounded by the Homunculus nebula. For a detailed review on the history and characteristics of this system and its surroundings, interested readers can refer to Davidson & Humphreys (2012). The central binary system has a period of P ~ 2024 days (periastron at T0 = 50 799.3 MJD, Corcoran 2005), in a highly eccentric orbit of e ~ 0.9 (Nielsen et al. 2007). The powerful winds of both components have large mass-loss rates of M⊙ yr−1 and M⊙ yr−1, with terminal velocities of 500 km s−1 and 3000 km s−1, respectively (Pittard & Corcoran 2002).

Although no radio synchrotron emission has been detected from η Carinae (Duncan & White 2003), it is a non-thermal X-ray (Leyder et al. 2008; Sekiguchi et al. 2009; Hamaguchi et al. 2018) and γ-ray (Tavani et al. 2009; Abdo et al. 2010; Reitberger et al. 2015; H.E.S.S. Collaboration 2020) emitter. Its spectrum has been widely studied above 100 MeV, where two distinct components are detected above (high energy; HE) and below (low energy; LE) 10 GeV. Despite the consensus that the HE component has a hadronic origin, at LE the situation is unclear: Both leptonic (Farnier et al. 2011; Gupta & Razzaque 2017) and hadronic (Ohm et al. 2015; White et al. 2020) scenarios are still plausible.

Unfortunately, only one other CWB has been detected in γ-rays: γ2 Velorum, also known as WR 11 (Pshirkov 2016; Martí-Devesa et al. 2020). Despite numerous efforts, CWBs with significant synchrotron radio emission have been elusive to detection both with orbital and ground-based γ-ray observatories (Romero et al. 1999; Aliu et al. 2008; Werner et al. 2013; del Palacio et al. 2020). In particular, the upper limits reported in observations on WR 140, WR 146, or WR 147 (Werner et al. 2013) imply very low efficiencies for γ-ray emission viathe inverse Compton (IC) process in classical CWBs. Therefore, the best target source to understand CWBs is the bright η Carinae itself.

In this work, we perform a detailed study on the two full orbits of η Carinae observed by the Fermi-LAT in 12 yr of operations, including three periastron passages. Additionally, we take advantage of this data selection to search for new nearby γ-ray source candidates, with special interest in other CWBs (Fig. 1). Section 2 presents the analysis performed, and results are shown in Sect. 3. Then, those are discussed in Sect. 4. Finally, the conclusions of our study are summarised in Sect. 5.

thumbnail Fig. 1

Left: Carina Nebula as seen in the Second Generation Digitalized Sky Survey (DSS2; red filter), which contains the system WR 22 (orange). In red, zoomed region. Right: core of the Carina Nebula centred on η Carinae as seen in DSS2 (IR filter). The region shown contains η Carinae (blue), WR 25 (yellow), HD 93129A (green), and HD 93250 (pink), while WR 21a and WR 39 lie ~ 3° away from η Carinae, outside the nebula.

2 Observations and analysis

The Large Area Telescope (LAT) is the main instrument on board the Fermi Gamma-ray Space Telescope (Atwood et al. 2009), covering the energy range between 30 MeV to more than 100 GeV. Its energy-dependent point-spread function (PSF) goes from more than 5° below 100 MeV to less than 0.1° above 10 GeV at 68% containment. In this paper, observations from 2008 August 4 to 2020 May 29 are included. The analysis was performed using Fermitools-1.2.231 on P8R3 data (Atwood et al. 2013; Bruel et al. 2018). Fluxes are obtained performing a binned maximum likelihood fit (Mattox et al. 1996) using fermipy 0.192 (Wood et al. 2017). We used the 4FGL DR2 catalogue (gll_psc_v23) for our source model (Abdollahi et al. 2020; Ballet et al. 2020), while the diffuse emission was assessed using ‘gll_ iem_v07.fits’ and ‘iso_ P8R3_SOURCE_V2_v1.txt’ for the Galactic and isotropic components, respectively3. To evaluate the significance of the detection of any source, we used the test statistic , where Lmax,0 is log-likelihood value for the null hypothesis and Lmax,1 the log-likelihood for the complete model. The larger the value of the TS, the less likely Lmax,0 is. The square root of the TS is approximately equal to the detection significance of a given source.

In the 4FGL DR2 catalogue, the spectrum from η Carinae is described with a log parabola (LP), defined as: (1)

where N0 is the normalisation flux, Eb is the pivot energy, α the spectral index at Eb and β is a curvature parameter.

2.1 Datasets and their analysis

Since η Carinae shows two distinct components, we have used different datasets to optimise our analyses. We differentiate between LE and HE Fermi-LAT analyses below and above 10 GeV, using the FRONT+BACK event type (evtype = 3). We refer to these hereafter as the LE and HE datasets. But since the large PSF at lower energies could result in source confusion in the Galactic plane and the contamination of our target, we produced a comparison dataset where we only selected the ensemble quartile with the highest quality in the reconstructed direction (evtype = 324). Hereunder, we refer to it as the PSF3 dataset, which improves the PSF at 68% from the aforementioned ~5° to ~ 3° at 100 MeV. The full description of the cuts and details of each dataset is provided in Appendix A. Besides, the model described above has been modified in each particular analysis to assess new background sources (see Sect. 2.2) and use the appropriate isotropic background according to the instrument response functions (IRFs).

The analysis performed on the LE and PSF3 datasets was similar, fitting the spectrum of the binary with a LP while keeping free the normalisation for all sources within 5° (i.e. 39 sources) from η Carinae present in the 4FGL DR2, together with all parameters for sources within 1° of the binary (i.e. 3 sources). On the other hand, we used a power law (PL) spectrum for the HE dataset, and those distances were reduced to 3° (i.e. 18 sources) and 0.5° (i.e. 1 source), respectively. In all datasets, the normalisation of the diffuse components and the spectral index of the Galactic one are also freeparameters.

2.2 Background model extension

The Carina region is a densely populated, challenging field to characterise with γ-rays. Multiple sources and strong diffuse emission may distort our fits, and therefore our background model has to be properly updated for our region of interest (ROI). A recent study by White et al. (2020) used a CO template of the region to take into account the excesses seen in the residuals. However, we took a different approach. Using the updated 4FGL DR2 already includes a new source within 3° of η Carinae (4FGL J1054.0-5938), but in order to assess the possible excesses present in the residuals, we extended the background model with new sources in an iterative way using the method find_sources from fermipy. This method has to be used with caution in order to prevent false positives, especially in the Galactic plane. Therefore we employed our different selection datasets to take advantage of their particularities.

The HE is indeed the cleanest dataset in terms of Galactic diffuse contamination, and therefore the most suitable one for searching new (albeit preferentially hard) sources. After our initial fit, we removed all sources detected with less than 2σ (TS = 4) in our ROI. This procedure purposefully reduces the degrees of freedom in the fit at the expense of neglecting sub-threshold soft sources.Afterwards, we performed a search for sources on this dataset. We detected two new signals above 4.5σ, which were added to the model and subsequently re-fitted. In a second iteration, we performed a new search on the PSF3 dataset, using as prior model the new sources in addition to 4FGL DR2. In that case we made our requirements slightly more stringent, demanding at least 5σ for a detection. We found 7 source candidates, which were then added to the model and re-fitted. In the last step, all these sources were included in the LE analysis model. Following this procedure, we do not find further significant emission in our residuals coincident with high values in the CO template close to η Carinae as in White et al. (2020), since one of the sources is partially coincident with a high CO grammage in the line of sight (see Sect. 3.2 for a description of the sources and Fig. A.1 for the residuals).

3 Results

Like previous studies (Reitberger et al. 2015; Balbo & Walter 2017; White et al. 2020), we detect η Carinae with high significance: 10773 TS, 1017 TS, and 7714 TS for the LE, HE, and PSF3 datasets, respectively. For the PSF3 dataset, we obtain an energy flux of MeV cm−2 s−1 with α = 2.32 ± 0.02 and β = 0.17 ± 0.01. For the LE we obtain an energy flux of MeV cm−2 s−1 with α = 2.24 ± 0.03 and β = 0.15 ± 0.03. Finally, at HE we find MeV cm−2 s−1 with a spectral index Γ = 2.55 ± 0.10 (Table 1). The spectral energy distribution (SED) obtained combining both bands is fully compatible with previous studies, with only minor deviations (Fig. 2). The differences might arise from the different cuts employed, the extended dataset or the further updated analysis software.

Similarly to what was done by Reitberger et al. (2015), we also fitted the spectrum obtained in the PSF3 dataset with a smoothly-broken power law (SBPL): (2)

where E0 is the energy normalisation (set to 1 GeV), δ is a curvature parameter and Γ1 and Γ2 are the spectral indexes above and below the pivot energy Eb, respectively. A SBPL model implies Γ1 = 2.64 ± 0.09, Γ2 = 1.19 ± 0.06, and δ = 1.05 ± 0.22 for the PSF3 dataset (Table 1). Since for this comparison both hypotheses (LP and SBPL) share a common background model which remains fixed, these are nested (see for example Algeri et al. 2016) and we can compare their likelihood values. We find that the SBPL model is preferred over a LP with , probably driven from the average emission above 10 GeV.

thumbnail Fig. 2

Spectral energy distribution for η Carinae using the LE (violet), HE (green) and PSF3 (black) datasets. The black line represents the SPBL fit, with the 1σ uncertainty shown in grey.

3.1 Temporal results

In order to study similarities between both orbits, we produced a light curve from the HE and LE datasets described in Sect. 2 using the lightcurve function from fermipy. In each bin we free the parameters from Eta Carinae and the normalisation from background sources within 3°, including the diffuse components. We divided each orbit in equally time-spaced bins (Figs. 3 and 4 for the LE and HE bands, respectively). We observe the η Carinae peak around periastron in both bands, and we can distinguish between the first (54 848 MJD), the second (56 872 MJD), and the third (58 896 MJD) periastron as derived from P = 2024 days. We refer tothem as P2009, P2014, and P2020, respectively. We employed P = 2024 for consistency with earlier works by Reitberger et al. (2015) and Balbo & Walter (2017); using P = 2022.7 ± 1.3 days as obtained by Damineli et al. (2008) does not change the results notably. While at LE the trend is similar for both orbits, the HE component has a different behaviour. The light curve above 10 GeV shows a similar peak for the first and third periastrons, but we observe thatthe second one occurs several months before periastron. Trying different re-binnings – by shifting the initial bin or adding morebins – still produces the peak, and only with a large binning the peak is averaged out (for example, using six bins per orbit). Furthermore, a light curve of the least-distant source (Fermi J1042.9-5938) does not show any correlation with the period of η Carinae – particularly not around the peak flux. We note that such an effect would have an impact regardless of the source’s spectral index.

We now study if the spectrum at HE of the peaks observed at periastron varies by producing a SED for each periastron using data from 200 days around them (i.e. ± 100 days). The resulting spectra can be found in Fig. 5. We note that this selection is done to ensure sufficient statistics but, given the orbital parameters of η Carinae, the distance between both stars does change appreciably during periastron. The results mimic the behaviour observed in the light curve, providing integrated fluxes between 10 GeV and 500 GeV of (8.12 ± 2.13) × 10−10 photons cm−2 s−1 with TS = 71.5, (3.69 ± 1.54) × 10−10 photons cm−2 s−1 with TS = 21.5, and (10.34 ± 2.08) × 10−10 photons cm−2 s−1 with TS = 121.9 for P2009, P2014, and P2020, respectively.

Table 1

Spectral results for η Carinae.

thumbnail Fig. 3

Light curve of η Carinae using the LE dataset (100 MeV–10 GeV), with 12 bins per orbit (168.67 days). In red, P2009, P2014, and P2020 periastron passages. At least 5σ are required per detection in each bin. We do not observe variations of the spectral parameters beyond statistical uncertainties.
thumbnail Fig. 4

Light curve of η Carinae using the HE dataset (10–500 GeV), with 8 bins per orbit (253.0 days). In red, P2009, P2014, and P2020 periastron passages. At least 5σ are required per detection in each bin, while a 2σ upper limit is shown otherwise. We do not observe variations of the spectral index beyond statistical uncertainties.

3.2 Search for nearby sources

The η Carinae system is in a very densely populated region; thus we studied the possibility of finding new sources nearby the binary and, additionally, searched for γ-ray emission from nearby CWB systems.

3.2.1 Fermi J1042.9-5938 and the nova ASASSN-18fv

In our analysis of the ROI at HE, we found two new sources: Fermi J1036.1-5934 (TS= 48.16; l = 286.554 ± 0.017, b = −1.102 ± 0.020) and Fermi J1042.9-5938 (TS = 35.37; l = 287.345 ± 0.026, b = −0.713 ± 0.026).

Fermi J1036.1-5934 is spatially coincident with the nova ASASSN-18fv (or V906 Carinae), which occurred in March 2018 (Jean et al. 2018). A light curve of the source confirms that it is only detected during the reported outburst. We modelled it with a PL with an integrated energy flux of (1.51 ± 0.16) × 10−5 MeV cm−2 s−1 and a spectral index of Γ = 2.19 ± 0.05. The extended observation time and the more restrictive event cuts are unsuited for the detection of any cut-off as seen by Aydi et al. (2020). On the other hand, no clear counterpart is found for Fermi J1042.9-5938. It partially overlaps with the CO line-of-sight column density as mentioned by White et al. (2020), and it could be related to the source 3FGL J1043.6-5930, not present in the 4FGL DR2. The source is modelled with a PL with an integrated energy flux of (1.33 ± 0.21) × 10−5 MeV cm−2 s−1 and a spectral index of Γ = 2.19 ± 0.06. It does not show variability on annual scales, nor any sign of spectral curvature or extension. Both sources were added in the LE analysis model seen in Sect. 2.

Our search for new sources on the PSF3 dataset also obtained 7 significant detections. However, the detection of these new γ-ray sources has to be taken with caution at lower energies. Unlike the HE dataset, the PSF3 search is more sensitive to provide excesses caused by the differences between our analysis andthe weighted analyses from the original characterisation of the 4FGL (Abdollahi et al. 2020) or source confusion at low energies. Therefore we consider those excesses only as candidate sources, and we do not explore their nature – see Baldini et al. (2021) for searches of transient sources beyond the 4FGL catalogue.

thumbnail Fig. 5

HE SEDs for the three periastrons – P2009 (blue), P2014 (violet), and P2020 (orange) – observed with Fermi-LAT – together with the best-fit PL and its 1σ uncertainty. Each SED is obtained using ± 100 days aroundeach one. For comparison, the overall SED above 10 GeV is shown (grey).
thumbnail Fig. 6

Light curve of HD 93129A with 1 yr long bins integrated between 300 MeV and 10 GeV. Upper limits are shown at 2σ confidence level. The dashed red line (57 973 MJD) indicates the periastron passage according to Maíz Apellániz et al. (2017), while its updated value from del Palacio et al. (2020) is marked with a solid red line (58 374 MJD).

3.2.2 Nearby CWBs

The Carina region contains numerous massive stars, some of them in binary systems known to display strong, powerful shocks. A few CWBs from De Becker & Raucq (2013) can be found in the vicinity of η Carinae: HD 93129A, HD 93250, WR 39, and WR 21a. Besides, two other CWBs with a WR component (WR 22 and WR 25) are additionally studied given their similarities with γ2 Velorum in terms of stellar components and orbital characteristics (Williams et al. 1994; Schweickhardt et al. 1999). The latter systems are particularly close to Fermi J1042.9-5938, but its localization at HE (where the PSF is smaller than 0.1°) does not favour a tentative association with them.

To evaluate their possible detection, we added a test source using a PL spectrum with Γ =2 at the position of each binary, modelling the ROI according to the result from our LE dataset main analysis (see Sect. 2). We obtained only upper limits, which are summarised in Table 2. Given that the periastron from HD 93129A occurred in late 2017 or early 2018 (Maíz Apellániz et al. 2017), we produced a light curve for this particular binary. However, no detection above 5σ is found (see Fig. 6).

4 Discussion

These new Fermi-LAT results have multiple implications, both from the variability and spectral point of view, and should be compared with previous studies and multi-wavelength data.

4.1 Orbit-to-orbit variability

While the variability of the flux at LE does not differ comparing the first and the second orbits, η Carinae seems more puzzling above 10 GeV. In the analysis presented by Balbo & Walter (2017), the flux at HE during the P2014 periastron did not increase. Our analysis supports that result with a caveat: the flux of η Carinae during the P2014 periastron was indeed less bright than the other two – the flux values around P2014 and P2020 differ at 2.6σ confidence level – but the flux was indeed increasing before and its actual peak occurred earlier. Contrarily, the P2020 periastron has a slightly larger flux than P2009. We suggest that such variations might by caused by turbulences and changes in the WCR structure from orbit to orbit, and may indicate that both populations of particles which produce the LE and HE components are accelerated in different regions of the WCR – for example either both sides of the contact discontinuity or at different distances from the apex of the shock.

Especially interesting is the comparison of the present data with the multi-wavelength reports of the P2020 periastron. NuSTAR observationsof η Carinae in the pre-periastron phase reported similar fluxes of non-thermal X-rays compared with previous orbits (Hamaguchi et al. 2019). However, the emission was twice as large in the post-periastron phase (Hamaguchi et al. 2020). At higher energies, AGILE reported a substantially larger γ-ray flux before the periastron at 4σ above 100 MeV (Piano et al. 2019), increasing by an order of magnitude with respect to the flux reported in the second AGILE-GRID catalogue (2AGL) of photons cm−2 s−1 (Bulgarelli et al. 2019). Contrarily, a Fermi-LAT analysis above 100 MeV during the same dates provides a flux of photons cm−2 s−1 (TS = 2.4), thus consistent with the 2AGL result. Besides, a light curve around those dates does not seem to support such rise of the γ-ray flux (see Fig. 7). This analysis only differs with the LE analysis described in Sect. 2 in the energy range, dates and background model (only includes 4FGL DR2). We also explored the possible impact of the different observing modes of Fermi during those dates5 on the exposure. Indeed, the observing mode of Fermi varies for that time interval. However, we find that the exposure varies less than a factor 2 between the different bins shown in Fig. 7 – thus being an unlikely origin for the discrepancy, even if it was not properly assessed in our likelihood analysis.

Table 2

Upper limits at 95% confidence level obtained other CWBs close to η Carinae using the LE dataset.

thumbnail Fig. 7

Light curve of η Carinae from 15 November to 5 December in 2019. In black, dedicated analysis between 100 MeV and 500 GeV with Fermi-LAT (2-day bins, requiring at least 2σ for detection). The corresponding average flux is represented with a grey line. The γ-ray flux enhancement reported by AGILE is shown in green (Piano et al. 2019), which overlaps with the X-ray light curve peak on 28 November 2019, indicated by a violet dashed-line (Corcoran et al. 2019).

4.2 On the origin of the LE component

Although the HE component is commonly believed to have a hadronic origin, the LE one remains the major focus of discussion around η Carinae. Both leptonic (Farnier et al. 2011; Balbo & Walter 2017) and hadronic (Ohm et al. 2015) models have been proposed to explain its origin with inconclusive results. To distinguish between both scenarios, White et al. (2020) used the low energy regime of Fermi-LAT to search for a signature of the π0 -bump. According to their results, the spectrum could not be reproduced with a leptonic model. However, the same authors performed multiple analyses and found that a single PL connecting both hard X-rays and γ-rays could not be discardedat 68% confidence level. The spectral shape at lower energies (especially below 100 MeV) is significantly affected by source confusion due to the large PSF and strong Earth limb contamination, impacting the results in typical analyses with Fermi-LAT and therefore requiring alternative analyses (Principe et al. 2018). Our analysis with more stringent cuts on the PSF reconstruction provides a more conservative reference and does not confirm a significant sudden drop in the spectrum at 100 MeV. Using an SBPL model, the preferred spectral index is harder than for a single PL X-ray/γ-ray connection with Γ ≈ 1.65 (Hamaguchi et al. 2018) suggesting that a single PL scenario is disfavoured. But, incidentally, the indication of lower flux observed at 80–100 MeV may also have its origin in an over-subtraction of events due to a large normalisation of the diffuse components. Furthermore, we note that the 3FGL all-sky analysis demonstrated how, on average, the systematic uncertainty on the flux due to the Galactic diffuse emission resembles the statistical error (Acero et al. 2015).

4.3 The population of CWBs at high energies

The non-detections of CWBs in HE γ-rays (Werner et al. 2013; Pshirkov 2016) lead to strict constraints of their capabilities as particle accelerators. The number of such systems detected at γ-rays is still scarce, with the only confirmed cases η Carinae itself and γ2 Velorum (Pshirkov 2016; Martí-Devesa et al. 2020)6.

Both confirmed cases behave differently, showing that the γ-ray emissivity is highly dependent on the particularities of each system. The case of γ2 Velorum showed HE emission during apastron, with no detection of radio synchrotron (Benaglia et al. 2019; Martí-Devesa et al. 2020). This is unexpected because classical models estimate Lγ ~ 1∕d, where d is the distancebetween both components of a binary. This result invites us to reconsider emission coming from other systems, like WR 22, with a period of 80 days and all the particularities of other CWBs, but not detected at non-thermal radio frequencies (Parkin & Gosset 2011; De Becker & Raucq 2013). Although it is similar to γ2 Velorum, WR 22 is located at a distance of ~2.7 kpc (Gosset et al. 2009). Therefore a comparable γ-ray luminosity to γ2 Velorum would provide fluxes below the sensitivity limit of Fermi-LAT. Another example of a CWB without clear non-thermalradio emission in the Carina region with similar characteristics is WR 25 (Arora et al. 2019), which also remains undetected in our study. Since both binaries were not included in previous searches, we constrain the γ-ray emission of these CWBs with first upper limits.

On the other side, HD 93129A is a more classical CWB which has been studied in detail recently. Its WCR was resolved by Benaglia et al. (2015) at radio frequencies, showing that it was indeed a CWB system with synchrotron emission. This binary is composed by the earliest-type O stars in the CWB catalogue by De Becker & Raucq (2013), in a very wide orbit with eccentricity e > 0.95 and period of ~120 yr (Maíz Apellániz et al. 2017). Its periastron passage was estimated to occur around the early months of 2018, giving a unique opportunity to study very close periastron conditions in CWBs. Non-thermal emission was predicted during its periastron passage at high energies (del Palacio et al. 2016), but no clear X-ray non-thermal component nor γ-rays have been detected using NuSTAR nor AGILE (del Palacio et al. 2020). Our upper limits further constrain its putative emission between 300 MeV and 10 GeV, with an upper limit during the periastron passage of 5.49 × 10−9 photons cm−2 s−1 (Fig. 6). Although this limit is almost two orders of magnitude lower than the previous constraints from AGILE, it does not conflict with the updated emission model by del Palacio et al. (2020) considering its non-detection at hard X-rays.

5 Conclusion

In this work, we presented a comprehensive study on the second full orbit of η Carinae and its third periastron as seen by the Fermi-LAT. Our results also hint that the recent periastron passage in February 2020 was the brightest observed with γ-rays. We found evidence of orbit to orbit variability in this system above 10 GeV, suggesting that the transport of particles in the WCR might be affected by different turbulences in each orbit perturbing such structure. To study the origin of the LE component, we used a stringent cut on the reconstruction quality of the PSF, but we cannot confirm a significant π0 -bump. Complementarily, we searched for new sources in the Carina region not included in the 4FGL catalogue and found two sources above 10 GeV (one associated with the nova V906 Carinae) and seven candidate sources at lower energies. Unfortunately, no emission has been found coincident with other massive CWBs present in the Carina region.

In short, the complicated behaviour observed prevents any simplistic considerations of particle acceleration in η Carinae and other CWBs. A future work will explore the modelling of η Carinae with magnetohydrodynamical simulations to explain the observed phenomena reported in this study.

Acknowledgements

The Fermi-LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat à l’Energie Atomique and the Centre National de la Recherche Scientifique/Institut National de Physique Nucléaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), High Energy Accelerator Research Organization (KEK) and Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden. Additional support for science analysis during the operations phase from the following agencies is also gratefully acknowledged: the Istituto Nazionale di Astrofisica in Italy and the Centre National d’Etudes Spatiales in France. This work performed in part under DOE Contract DE-AC02-76SF00515. This work uses DSS2 images accessed via http://archive.eso.org/dss/dss. Southern hemisphere DSS2 data is based on photographic data obtained using The UK Schmidt Telescope. The UK Schmidt Telescope was operated by the Royal Observatory Edinburgh, with funding from the UK Science and Engineering Research Council, until 1988 June, and thereafter by the Anglo-Australian Observatory. Original plate material is copyright (c) of the Royal Observatory Edinburgh and the Anglo-Australian Observatory. The plates were processed into the present compressed digital form with their permission. The Digitized Sky Survey was produced at the Space Telescope Science Institute under US Government grant NAG W-2166.

Appendix A Fermi-LAT analysis

Table A.1

Description of the LE, HE and PSF3 datasets

thumbnail Fig. A.1

TS map of the ROI using the PSF3 dataset, after removing all residual sources as described in Section 2.2.

References

  1. Abdo, A. A., Ackermann, M., Ajello, M., et al. 2010, ApJ, 723, 649 [NASA ADS] [CrossRef] [Google Scholar]
  2. Abdollahi, S., Acero, F., Ackermann, M., et al. 2020, ApJS, 247, 33 [NASA ADS] [CrossRef] [Google Scholar]
  3. Acero, F., Ackermann, M., Ajello, M., et al. 2015, ApJS, 218, 23 [Google Scholar]
  4. Algeri, S., Conrad, J., & van Dyk, D. A. 2016, MNRAS, 458, L84 [NASA ADS] [CrossRef] [Google Scholar]
  5. Aliu, E., Anderhub, H., Antonelli, L. A., et al. 2008, ApJ, 685, L71 [NASA ADS] [CrossRef] [Google Scholar]
  6. Arora, B., Pandey, J. C., & De Becker, M. 2019, MNRAS, 487, 2624 [Google Scholar]
  7. Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071 [NASA ADS] [CrossRef] [Google Scholar]
  8. Atwood, W., Albert, A., Baldini, L., et al. 2013, ArXiv e-prints [arXiv:1303.3514] [Google Scholar]
  9. Aydi, E., Sokolovsky, K. V., Chomiuk, L., et al. 2020, Nat. Astron., 4, 776 [CrossRef] [Google Scholar]
  10. Balbo, M., & Walter, R. 2017, A&A, 603, A111 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  11. Baldini, L., Ballet, J., Bastieri, D., et al. 2021, ApJS, 256, 13 [NASA ADS] [CrossRef] [Google Scholar]
  12. Ballet, J., Burnett, T. H., Digel, S. W., & Lott, B. 2020, ArXiv e-prints [arXiv:2005.11208] [Google Scholar]
  13. Benaglia, P., & Romero, G. E. 2003, A&A, 399, 1121 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  14. Benaglia, P., Marcote, B., Moldón, J., et al. 2015, A&A, 579, A99 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  15. Benaglia, P., del Palacio, S., Ishwara-Chandra, C. H., et al. 2019, A&A, 625, A99 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  16. Bruel, P., Burnett, T. H., Digel, S. W., et al. 2018, ArXiv e-prints [arXiv:1810.11394] [Google Scholar]
  17. Bulgarelli, A., Fioretti, V., Parmiggiani, N., et al. 2019, A&A, 627, A13 [CrossRef] [EDP Sciences] [Google Scholar]
  18. Corcoran, M. F. 2005, AJ, 129, 2018 [NASA ADS] [CrossRef] [Google Scholar]
  19. Corcoran, M. F., Hamaguchi, K., Gendreau, K., et al. 2019, ATel, 13327, 1 [NASA ADS] [Google Scholar]
  20. Damineli, A., Hillier, D. J., Corcoran, M. F., et al. 2008, MNRAS, 384, 1649 [NASA ADS] [CrossRef] [Google Scholar]
  21. Davidson, K., & Humphreys, R. M. 2012, Eta Carinae and the Supernova Impostors (Berlin: Springer), 384 [Google Scholar]
  22. De Becker, M. 2007, A&ARv, 14, 171 [NASA ADS] [CrossRef] [Google Scholar]
  23. De Becker, M., & Raucq, F. 2013, A&A, 558, A28 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  24. del Palacio, S., Bosch-Ramon, V., Romero, G. E., & Benaglia, P. 2016, A&A, 591, A139 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  25. del Palacio, S., García, F., Altamirano, D., et al. 2020, MNRAS, 494, 6043 [NASA ADS] [CrossRef] [Google Scholar]
  26. Duncan, R. A., & White, S. M. 2003, MNRAS, 338, 425 [NASA ADS] [CrossRef] [Google Scholar]
  27. Eichler, D., & Usov, V. 1993, ApJ, 402, 271 [Google Scholar]
  28. Farnier, C., Walter, R., & Leyder, J. C. 2011, A&A, 526, A57 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  29. Gosset, E., Nazé, Y., Sana, H., Rauw, G., & Vreux, J. M. 2009, A&A, 508, 805 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  30. Grimaldo, E., Reimer, A., Kissmann, R., Niederwanger, F., & Reitberger, K. 2019, ApJ, 871, 55 [NASA ADS] [CrossRef] [Google Scholar]
  31. Gupta, N., & Razzaque, S. 2017, Phys. Rev. D, 96, 123017 [NASA ADS] [CrossRef] [Google Scholar]
  32. Hamaguchi, K., Corcoran, M. F., Pittard, J. M., et al. 2018, Nat. Astron., 2, 731 [NASA ADS] [CrossRef] [Google Scholar]
  33. Hamaguchi, K., Corcoran, M. F., & Ferrigno, C. 2019, ATel, 13368, 1 [NASA ADS] [Google Scholar]
  34. Hamaguchi, K., Espinoza-Galeas, D. A., Corcoran, M. F., et al. 2020, ATel, 13738, 1 [NASA ADS] [Google Scholar]
  35. H.E.S.S. Collaboration (Abdalla, H., et al.) 2020, A&A, 635, A167 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  36. Hillier, D. J., Davidson, K., Ishibashi, K., & Gull, T. 2001, ApJ, 553, 837 [NASA ADS] [CrossRef] [Google Scholar]
  37. Jean, P., Cheung, C. C., Ojha, R., van Zyl, P., & Angioni, R. 2018, ATel, 11546, 1 [NASA ADS] [Google Scholar]
  38. Leyder, J. C., Walter, R., & Rauw, G. 2008, A&A, 477, L29 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  39. Maíz Apellániz, J., Sana, H., Barbá, R. H., Le Bouquin, J. B., & Gamen, R. C. 2017, MNRAS, 464, 3561 [NASA ADS] [CrossRef] [Google Scholar]
  40. Martí-Devesa, G., Reimer, O., Li, J., & Torres, D. F. 2020, A&A, 635, A141 [Google Scholar]
  41. Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396 [NASA ADS] [CrossRef] [Google Scholar]
  42. Mehner, A., Davidson, K., Ferland, G. J., & Humphreys, R. M. 2010, ApJ, 710, 729 [NASA ADS] [CrossRef] [Google Scholar]
  43. Nielsen, K. E., Corcoran, M. F., Gull, T. R., et al. 2007, ApJ, 660, 669 [NASA ADS] [CrossRef] [Google Scholar]
  44. Ohm, S., Zabalza, V., Hinton, J. A., & Parkin, E. R. 2015, MNRAS, 449, L132 [NASA ADS] [CrossRef] [Google Scholar]
  45. Parkin, E. R.,& Gosset, E. 2011, A&A, 530, A119 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  46. Piano, G., Tavani, M., Verrecchia, F., et al. 2019, ATel, 13329, 1 [NASA ADS] [Google Scholar]
  47. Pittard, J. M., & Corcoran, M. F. 2002, A&A, 383, 636 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  48. Pittard, J. M., Vila, G. S., & Romero, G. E. 2020, MNRAS, 495, 2205 [NASA ADS] [CrossRef] [Google Scholar]
  49. Principe, G., Malyshev, D., Ballet, J., & Funk, S. 2018, A&A, 618, A22 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  50. Pshirkov, M. S. 2016, MNRAS, 457, L99 [NASA ADS] [CrossRef] [Google Scholar]
  51. Reimer, A., Pohl, M., & Reimer, O. 2006, ApJ, 644, 1118 [NASA ADS] [CrossRef] [Google Scholar]
  52. Reitberger, K., Kissmann, R., Reimer, A., & Reimer, O. 2014, ApJ, 789, 87 [NASA ADS] [CrossRef] [Google Scholar]
  53. Reitberger, K., Reimer, A., Reimer, O., & Takahashi, H. 2015, A&A, 577, A100 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  54. Rodríguez, L. F., González, R. F., Montes, G., et al. 2012, ApJ, 755, 152 [NASA ADS] [CrossRef] [Google Scholar]
  55. Romero, G. E., Benaglia, P., & Torres, D. F. 1999, A&A, 348, 868 [NASA ADS] [Google Scholar]
  56. Schweickhardt, J., Schmutz, W., Stahl, O., Szeifert, T., & Wolf, B. 1999, A&A, 347, 127 [NASA ADS] [Google Scholar]
  57. Sekiguchi, A., Tsujimoto, M., Kitamoto, S., et al. 2009, PASJ, 61, 629 [NASA ADS] [Google Scholar]
  58. Smith, N. 2006, ApJ, 644, 1151 [NASA ADS] [CrossRef] [Google Scholar]
  59. Tavani, M., Sabatini, S., Pian, E., et al. 2009, ApJ, 698, L142 [NASA ADS] [CrossRef] [Google Scholar]
  60. Werner, M., Reimer, O., Reimer, A., & Egberts, K. 2013, A&A, 555, A102 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  61. White, R., Breuhaus, M., Konno, R., et al. 2020, A&A, 635, A144 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  62. Williams, P. M., van der Hucht, K. A., Kidger, M. R., Geballe, T. R., & Bouchet, P. 1994, MNRAS, 266, 247 [NASA ADS] [Google Scholar]
  63. Wood, M., Caputo, R., Charles, E., et al. 2017, Int. Cosmic Ray Conf., 301, 824 [Google Scholar]
1

This is the nomenclature for the Fermi Science Tools released through Conda. See https://github.com/fermi-lat/Fermitools-conda/wiki

2

Python package for the Fermitools. See https://fermipy.readthedocs.io/en/latest/

3

The latest background models are provided in https://fermi.gsfc.nasa.gov/ssc/data/access/lat/BackgroundModels.html

6

We also note the tentative association of a new CWB in the 4FGL DR2 (Ballet et al. 2020). The γ-ray source 4FGL J1820.4-1609c is found to be compatible with CEN 1 (also known as Kleinmann’s star), a trapezium system formed by two O+O binaries (CEN 1a and CEN 1b) showing variable non-thermal radio emission from one of its components (Rodríguez et al. 2012). However, studying this system in detail is beyond the scope of this paper.

All Tables

Table 1

Spectral results for η Carinae.

In the text
Table 2

Upper limits at 95% confidence level obtained other CWBs close to η Carinae using the LE dataset.

In the text
Table A.1

Description of the LE, HE and PSF3 datasets

In the text

All Figures

thumbnail Fig. 1

Left: Carina Nebula as seen in the Second Generation Digitalized Sky Survey (DSS2; red filter), which contains the system WR 22 (orange). In red, zoomed region. Right: core of the Carina Nebula centred on η Carinae as seen in DSS2 (IR filter). The region shown contains η Carinae (blue), WR 25 (yellow), HD 93129A (green), and HD 93250 (pink), while WR 21a and WR 39 lie ~ 3° away from η Carinae, outside the nebula.
In the text
thumbnail Fig. 2

Spectral energy distribution for η Carinae using the LE (violet), HE (green) and PSF3 (black) datasets. The black line represents the SPBL fit, with the 1σ uncertainty shown in grey.
In the text
thumbnail Fig. 3

Light curve of η Carinae using the LE dataset (100 MeV–10 GeV), with 12 bins per orbit (168.67 days). In red, P2009, P2014, and P2020 periastron passages. At least 5σ are required per detection in each bin. We do not observe variations of the spectral parameters beyond statistical uncertainties.
In the text
thumbnail Fig. 4

Light curve of η Carinae using the HE dataset (10–500 GeV), with 8 bins per orbit (253.0 days). In red, P2009, P2014, and P2020 periastron passages. At least 5σ are required per detection in each bin, while a 2σ upper limit is shown otherwise. We do not observe variations of the spectral index beyond statistical uncertainties.
In the text
thumbnail Fig. 5

HE SEDs for the three periastrons – P2009 (blue), P2014 (violet), and P2020 (orange) – observed with Fermi-LAT – together with the best-fit PL and its 1σ uncertainty. Each SED is obtained using ± 100 days aroundeach one. For comparison, the overall SED above 10 GeV is shown (grey).
In the text
thumbnail Fig. 6

Light curve of HD 93129A with 1 yr long bins integrated between 300 MeV and 10 GeV. Upper limits are shown at 2σ confidence level. The dashed red line (57 973 MJD) indicates the periastron passage according to Maíz Apellániz et al. (2017), while its updated value from del Palacio et al. (2020) is marked with a solid red line (58 374 MJD).
In the text
thumbnail Fig. 7

Light curve of η Carinae from 15 November to 5 December in 2019. In black, dedicated analysis between 100 MeV and 500 GeV with Fermi-LAT (2-day bins, requiring at least 2σ for detection). The corresponding average flux is represented with a grey line. The γ-ray flux enhancement reported by AGILE is shown in green (Piano et al. 2019), which overlaps with the X-ray light curve peak on 28 November 2019, indicated by a violet dashed-line (Corcoran et al. 2019).
In the text
thumbnail Fig. A.1

TS map of the ROI using the PSF3 dataset, after removing all residual sources as described in Section 2.2.
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.