Free Access
Issue
A&A
Volume 586, February 2016
Article Number A58
Number of page(s) 7
Section Stellar structure and evolution
DOI https://doi.org/10.1051/0004-6361/201527239
Published online 26 January 2016

© ESO, 2016

1. Introduction

V404 Cygni also known as GS 2023+338 is a low-mass X-ray binary (LMXB) originally discovered during its 1989 outburst by the Japanese GINGA satellite (Kitamoto et al. 1989). Intensive photometric and spectroscopic follow-up of this historical event yielded an orbital period of about 6.5 d and remarkably concluded with the presence of a massive stellar black hole in the system (Casares et al. 1992; Wagner et al. 1992). The companion star of V404 Cygni has been proposed to be an object of a late-type K0(±1) III-V spectral type (Casares et al. 1993). Radio detections using Very Long Baseline Interferometry (VLBI) enabled an accurate parallax measurement that places V404 Cygni at a distance of 2.39 ± 0.14 kpc (Miller-Jones et al. 2009). These VLBI observations also constrained the size of any quiescent jets to less than 1.4 AU. The quiescent X-ray spectrum has a power-law photon index Γ ≃ 2.0 seen through a a total column density of NH = (1.0 ± 0.1) × 1022 cm-2 (see e.g. Reynolds et al. 2014, and references therein). The unabsorbed 0.3–10 keV luminosity approaches several times 1032 erg s-1, thus making V404 Cygni the brightest black hole LMXB in quiescence.

On 16 June 2015, V404 Cygni was reported to be in outburst by the Swift Burst Alert Telescope (Barthelmy et al. 2015), and soon confirmed by the Monitor of All-sky X-ray Image on board the International Space Station (Negoro et al. 2015). After these early warnings, an intensive observational effort was deployed by many observers that was quickly reflected in an intense flow of tens of related electronic telegrams. Soon, it became evident that this was an extraordinary outburst event after decades of quiescence. Optical, infrared, radio, X-ray and gamma-ray telescopes have so far collected an impressive amount of data that will emerge in the scientific literature in the upcoming months. But the extraordinary thing was that even small optical telescopes could join this endeavor since V404 Cygni became a bright and highly variable source at visible wavelengths (Hynes et al. 2015; Hardy et al. 2015; Wiersema 2015; Scarpaci & Maitra 2015).

In this work, we present the optical data acquired using the educational astronomical facilities available at the University of Jaén (UJA). Our aim is to contribute to the wealth of public astronomical data about this transient event while, at the same time, attempting to better constrain the nature of this phenomenon.

2. Observations

The observations were carried out on 26 June 2015 from the UJA Astronomical Observatory. The observatory is located in an urban area inside the Campus of Las Lagunillas, and hosts an automated 41 cm Schmidt-Cassgrain with f/8 focal ratio. The UJA telescope, heareafter UJT, operates using a ST10-XME commercial CCD camera with 2184 × 1472 pixels of 6.8 μm size. The pixel scale is 0.′′42 pixel-1. The camera is equipped with a wheel of UBVRcIc Johnson-Cousins filters (Johnson & Morgan 1953; Cousins 1974a,b) manufactured according to the Bessell filter prescription (Bessell 1979). The seeing average was typically of about 2′′, and therefore 2 × 2 binning was used.

Differential VRcIc photometry was performed on V404 Cygni during 3 h with exposure times of 60 s in each filter. The total number of measurements acquired per filter was Nobs = 51. Four comparison stars in the field were used whose photometric behaviour, within 0.01–0.02 mag, was found to be very stable in all bands. Their VRcIc magnitudes were retrieved from the AAVSO database1, and are given in Table A.1. Photometry in U and B-bands was not acquired because of very low source counts. For each observed photometric band, here generically indicated by an effective wavelength λ, the magnitudes of the variable target were derived with respect to the comparison stars according to: (1)where is the magnitude of the comparison star being used, and Δmλ,ins and ΔCλ,ins are the instrumental differences in magnitude and colour between the target and comparison star, respectively. The factor Tλ,Cλ is the colour transformation coefficient that was separately determined using standards in clusters and other fields. The colour (VRc) was used in Eq. (1) for V- and Rc-band observations, while the colour (VIc) was preferred for Ic-band observations instead. The system photometric properties are approximately stable from night to night and we typically obtain | Tλ,Cλ | < 0.1. Finally, the different measurements of were weighted according to their respective uncertainty and then averaged. The corresponding results are presented in Table A.2 and Figs. 1 and 2. The errors quoted here are representative of the differential photometry process, and do not include the uncertainty in the absolute calibration of the comparison stars (~0.02 mag). Main flaring events in Fig. 1 are labelled from 1 to 6 for later discussion.

thumbnail Fig. 1

Top: light curves of V404 Cygni in outburst as observed with the UJT on 26 June 2015 in the V, R and I-bands. Flares are labelled for clarity using vertical arrows. Bottom: behaviour of the four comparison stars, plotted at the same scale, which remained constant in brightness within 0.01–0.02 mag.

Open with DEXTER

thumbnail Fig. 2

Top: variability of colour indices VR, VI and RI of V404 Cygni in outburst as observed with the UJT on 26 June 2015. They are shown plotted in black, red, and magenta colours, respectively. Bottom: the same kind of plot for the four comparison stars plotted at the same scale. Their colours remained constant within 0.02–0.03 mag.

Open with DEXTER

3. Discussion

The light curves in Figs. 1 and 2 clearly show how V404 Cygni varied and flared intensively during the observation. The amplitude of variation was as large as one magnitude and several tenths of a magnitude in brightness and colour, respectively. These variations occurred on timescales as short as 10 min. The significance and intrinsic origin of these flares is ensured given that the comparison stars remained practically flat within a few hundreds of a magnitude. This agrees well with the behaviour reported by different observers at other times during the intensive coverage of the outburst (e.g. Hynes et al. 2015). From causality arguments, our data puts a coarse upper limit of ~1 AU to the region from where the optical emission arises in the vicinity of the accretion disk around the V404 Cygni black hole.

It is also very interesting that the observed variability apparently followed a clear pattern in colour–colour diagrams. This is illustrated in Fig. 3 where the pattern becomes more evident as the source evolved in a very restricted region of the diagram. A simple regression fit yields a correlation coefficient as high as 0.89, thus suggesting that a strong connection between the source colours existed during the flaring events. The modelling of such behaviour is beyond the observational scope of this paper, and we limit ourselves here to state this observational fact. Nevertheless, we speculate that it could be intimately linked to the source spectral evolution discussed below.

thumbnail Fig. 3

Evolution of V404 Cygni in the colour–colour plane during a few hours on 26 June 2015 as observer with the UJT. The dotted line is a linear regression fit guiding the eye to better appreciate the colour–colour correlation suggested in the text.

Open with DEXTER

thumbnail Fig. 4

Dereddened light curves of V404 Cygni in outburst as observed with the UJT on 26 June 2015 in the V, Rc and Ic-bands assuming an interstellar extinction of AV = 4.0 mag. The brightness level has been converted from magnitudes to a flux density scale, in mJy units. Flares are labelled as in Fig. 1.

Open with DEXTER

The intrinsic amplitudes of the observed flares become more obvious when de-reddening the photometric observations, and expressing them in terms of flux density. The interstellar extinction law in Mathis (2000) was used for de-reddening purposes. This is presented in Fig. 4, where two isolated flares (#3 and #4) are easily visible with their peaks separated by nearly a half hour. Other flaring events both at the beginning (#1 and #2) and the end (#5 and #6) of the observation are also present within comparable time intervals. However, their individual evolution cannot be so easily disentangled because they are close in time or not well sampled. By analogy with other similar flaring behaviours, such as in the LMXB GRS 1915+105 at radio and near-infrared wavelengths (Mirabel et al. 1998), one is tempted to interpret this recursive flare pattern as episodes of replenishment and emptying of the inner accretion disk. These kind of events come accompanied by the ejection of plasmons along collimated jets as a result of accretion disk instabilities. These plasma clouds typically have a characteristic non-thermal, synchrotron emission mechanism. Their spectrum spans from radio to much shorter wavelengths depending on the energy cut-off of relativistic electrons and the strength of magnetic field. The fact that strong radio emission from V404 Cygni was detected in 1.4 GHz observations carried out just six hours before ours, peaking at 0.364 ± 0.030 Jy (Tsubono et al. 2015), would be naturally understood in this context. The similarity between V404 Cygni and GRS 1915+105 has already been noted by Mooley et al. (2015).

The observed optical flares appeared superimposed onto an average pedestal emission level likely due to hot accretion disk in V404 Cygni. From Fig. 4, the average de-reddened flux densities were found to be 2.27 ± 0.16, 2.02 ± 0.11, and 1.10 ± 0.04 Jy in the V, Rc and Ic-bands, respectively. By fitting a simple power law, the average spectral energy distribution (SED) of V404 Cygni depends on frequency roughly as ν2.1 ± 0.3. No correction for the contribution of the Hα emission line to Rc-band data has been attempted here. The resulting power law index is consistent with the UJT data sampling the Raleigh Jeans ν2 part of the accretion disk spectrum. Extrapolation of this power law to the radio band falls orders of magnitude below the reported radio flux densities. Therefore, an additional emission component must exist to explain the Tsubono et al. (2015) radio detection. As stated before, this component is most naturally interpreted as originating in plasma ejection events.

In the scenario outlined here, one would expect the peak time of an individual flare to be dependent on the wavelength of observation. According to the simple van der Laan (1966) model of a synchrotron-emitting expanding plasmon, the light curve observed at a wavelength λ reaches its maximum after a time given by (2)where tm,λ0 is the time of maximum at a reference wavelength λ0, and p is the power-law index of the energy distribution of relativistic electrons.

In the absence of contemporaneous high time resolution radio photometry, where a λ-dependent delay would be easier to measure, we searched our optical data for possible time lags between the different light curves at V, Rc and Ic-bands. Taking, for instance, the isolated flare #3, which peaks around MJD 57 199.965 in Figs. 1 and 4, the rising time from the pedestal emission level to the flare peak in the V-band is tm,V ≃ 0.01 d. Assuming p ≃ 1.0, and given that the central effective wavelengths of our photometric filer set (λV = 0.545 μm, λRc = 0.641 μm, and λIc = 0.798μm), we can estimate the expected time delays in the UJT observations. Using Eq. (2), and taking the V-band as reference, the predicted delays are 1.2 min and 3.0 min for the Rc and Ic-band, respectively. Similarly, a 1.8 min delay would be expected between Ic and Rc data. Assuming other reasonable values of p does not change significantly these numbers. For instance, taking p = 2.5 one gets delays of 1.0 min, 2.4 min, and 1.4 min, respectively.

What happens with real data? To answer this question, we performed a cross-correlation function (CCF) exercise between the different UJT light curves. A problem arises here because the data are not evenly spaced in time. The CCF calculation was then carried out in two different ways. The first calculation was specially designed for such irregular time series and is based on averaging data products with similar time lags within a user defined binning (Edelson & Krolik 1988). The second calculation simply makes a linear interpolation to obtain a regular sampling of the data from which the traditional estimator of the CCF is computed. In our case, this second approach is justified because our observations were nearly evenly spaced. When applying the first method, the resulting CCFs do not have enough resolution to clearly establish a non-zero time lag although hints of delay between the longer and shorter wavelength filters were obtained. In contrast, the interpolation method provided well-defined CCF maxima shifted from zero (Figs. A.1 and A.2). Here, the key fact is that all maxima occur with a clearly negative lag, i.e. consistent with the longer wavelength light curves that are delayed with respect to the shorter wavelength curves, as expected. Interestingly, similar negative time lags have also been observed in some of the V404 Cygni flares seen by the INTEGRAL satellite (Rodriguez et al. 2015), where the optical V-band emission was delayed with respect to hard X-rays and soft γ-rays from 1.5 to 20–30 min. However, these lags correspond to extremely different energy bands in contrast to our results here reported within the optical domain.

The measured offset of the maxima with respect to the origin (in Figs. A.1 and A.2) indicates that the Ic-band light curve is delayed with respect to V-band by 108 ± 1 s, while the Rc-band lags it by 34 ± 1 s. Concerning the two reddest filters, Ic-band also lags the Rc-band by 34 ± 1 s. The uncertainties in the estimates of the CCF peak positions were derived using the White & Peterson (1994) formula (3)where WCCF is the full width at half maximum of the CCF peak with rmax amplitude, and Ntotal = (Nobs − 1)N the full number of available points, where N is the amount of sampling points. In our case, WCCF ≃ 1000 s, rmax ≃ 1, and we interpolated up to N = 104 points.

Although these numbers are not identical to the values anticipated using Eq. (2), all of them appear to be in the expected sense and with the expected order of magnitude. Such agreement is remarkable given the simplicity of the Van der Laan model and the limitation of our data.

In order to assess the robustness of our finding, we performed a sensitivity analysis of how interpolated sampling affects the results. As seen in Fig. A.3, the derived lags always reach a constant level after a sufficiently large number of interpolated points. Computing the rms dispersion of the latest 1000 points provides values of ~1 s comparable to the uncertainty estimates derived with Eq. (3). Figure A.3 also shows that, even with coarse interpolation, the V vs. Ic time lag still remains significant although at a lower level. The V vs. Ic time lag amounts to 86 ± 16 s for N = 10 interpolated sampling points. The other filter combinations, V vs. Rc and R vs. Ic, which are closer in wavelength, remain consistent with zero both amounting to 22 ± 16 s.

Moreover, to further ensure the confidence on the lag estimates with interpolated light curves, we conducted a series of simulations by cross-correlating the observed light curves with themselves after introducing different artificial time lags (see Table A.3). Our procedure was able to confidently recover these time lags, within a few seconds, but always with a systematic negative bias that results as an intrinsic effect of the interpolation method used. In our case this is not expected to introduce a severe effect because of the lack of skewness in the time sampling distribution (Rehfeld et al. 2011). Therefore, the measured lag values quoted above should be actually interpreted as lower limits. These two consistency checks reinforce our confidence in the lag detection provided that interpolation is actually a good approximation.

Another test that could be performed is the dependence of the maximum flux densities Sm,λ as a function of wavelength. Following van der Laan (1966), the expected dependence is: (4)considering the same isolated flare as above, and removing the pedestal flux density with a simple linear fit, we estimate the V-band maximum as Sm,V ≃ 1480 mJy. For p = 1, one would then predict Sm,Rc ≃ 1260 mJy and Sm,Ic ≃ 1010 mJy. Our Rc and Ic light curves in Fig. 4 were not well sampled to appropriately catch their maxima. With the pedestal flux density removed they provide Sm,Rc ≳ 1020 mJy and Sm,V ≳ 650 mJy, respectively. While the expected values are not exactly reproduced, the results are still compatible.

All together, we consider that this finding is supportive of our tentative interpretation of the optical flares as due to non-thermal plasmon ejection events. Similar flares in the prototypical case of GRS 1915+105 could not be observed in the optical because of the higher interstellar extinction.

In addition, the CCFs in Fig. A.1 display clear secondary maxima corresponding to time lags of ~1500 s. This agrees with the idea, already expressed before, that the optical light curve of V404 Cygni in outburst during our observation consists of the superposition of flares that appear every half hour on average. Flaring behaviour with a similar timescale is also present in optical light curves reported by other observers just before and after our UJT observations (Scarpaci & Maitra 2015; Scarpaci et al. 2015).

The energy involved in each ejection event can also be derived based on simple equipartition arguments using the Pacholczyk (1970) formulation. Looking at Fig. 4, the flare incremental flux density with respect to the pedestal value is in the range ~0.5–1.0 Jy, and this occurs typically ~103 s after the estimated onset of the flares. Such flux density increments are comparable to the 1.4 GHz emission levels reported near the time of our observations within a factor of a few. This renders the possibility of a nearly flat synchrotron component from radio to optical wavelengths conceivable. Assuming an expansion velocity of ~0.5c, the plasmon size at the time of maximum would be ~1.5 × 1013 cm. This is equivalent to 0.4 milli-arcsec at the V404 Cygni distance and consistent with previous size upper limits. Using this dimension, the radio/optical flux densities quoted before, and assuming relativistic electrons with Lorentz factor γ ≤ 104, the total energy content is estimated to be 2 × 1040 erg, with a magnetic field of 12 G. The source brightness temperature does not exceed a few 1012 K. Under all such assumptions, the corresponding non-thermal luminosity, from the radio to optical domain, amounts to 2.2 × 1036 erg s-1, which implies a synchrotron life-time of about 2 h for individual flaring events. This is only four times their observed recurrence interval, thus flare superposition should become negligible after a few flaring events. Finally, the plasmon mass involved in the ejection is found to be ~2 × 1020 g provided that there is one proton per relativistic electron. These numbers do not differ by more than one order of magnitude when compared with the reference case of GRS 1915+105.

4. Conclusions

We have reported new optical photometric data for the LMXB V404 Cygni during its 2015 June outburst obtained with a small educational telescope. The time interval covered by our observations complements and contributes to the multi-wavelength campaign carried out as a world-wide effort by many observers. From the analysis of our data alone, some interesting findings can already be advanced and we summarize them as follows.

The UJT optical light curves consisted of consecutive and partially overlapping flaring events, appearing about every half hour and with very large amplitudes (~1 mag). Timescales of variability as short as 10 min could be clearly detected both in brightness and colour. The optical variability observed appears to display a correlation pattern in the colour-colour plane, whose interpretation will deserve future theoretical work.

The de-reddened optical continuum had an average positive spectral index close the + 2 value. Thus, for most of the time the observed photometric bands (VRcIc) were mainly sampling the Rayleigh Jeans part of the accretion disk spectrum. The V404 Cygni flares, superposed on this continuum, are tentatively interpreted as non-thermal flaring events. With all caution, we suggest that the flares are due to relativistic plasmons ejected following successive replenishing and emptying episodes of the inner accretion disk. The fact that non-thermal, synchrotron emission appears to contribute to optical wavelengths renders V404 Cygni a very interesting system to better study the energetics of black hole LMXBs.

From a CCF analysis, we have found a systematic time lag between the UJT optical light curves acquired with the different filters during our observations. These lags are such that emission at shorter wavelengths precedes emission at longer wavelengths by ~1 min. This finding was most evident when light curves were resampled and evenly interpolated, which we believe is a reasonable approach when dealing with nearly evenly spaced data. This delay is in qualitative agreement with a simple Van der Laan model for the expansion of a synchrotron emitting plasmon, and the relative amplitude of flare maxima is also consistent. An estimate of the plasmon physical parameters required for the non-thermal spectrum to extend from radio to

optical wavelengths appears to be similar to other well-known systems, such as GRS 1915+105.


1

American Association of Variable Stars Observers, http://www.aavso.org and references therein.

Acknowledgments

This work was supported by grant AYA2013-47447-C3-3-P from the Spanish Ministerio de Economía y Competitividad (MINECO), and by the Consejería de Economía, Innovación, Ciencia y Empleo of Junta de Andalucía under research group FQM-322, as well as FEDER funds.

References

Appendix A: Additional material

Table A.1

AAVSO comparison stars used in this work.

Table A.2

Optical photometry of V404 Cygni wiht the UJT telescope.

Table A.3

Recovered artificial time lags with interpolated sampling CCF for V-band data.

thumbnail Fig. A.1

CCFs of the optical light curves V vs. Rc (black), V vs. Ic (blue) and Rc vs. Ic (red). The vertical dashed line corresponds to zero time lag. The V vs. Rc CCF is normalized to unity while the other two have been slightly scaled above and below for easier display.

Open with DEXTER

thumbnail Fig. A.2

Zoomed view of the central maxima of the CCFs where a clear asymmetry is visible. Colours and normalization are as in Fig. A.1.

Open with DEXTER

thumbnail Fig. A.3

Dependence of the derived time lags using the cross-correlation technique on the amount of interpolated sampling of the data. Colours are as in Figs. A.1 and A.2.

Open with DEXTER

All Tables

Table A.1

AAVSO comparison stars used in this work.

Table A.2

Optical photometry of V404 Cygni wiht the UJT telescope.

Table A.3

Recovered artificial time lags with interpolated sampling CCF for V-band data.

All Figures

thumbnail Fig. 1

Top: light curves of V404 Cygni in outburst as observed with the UJT on 26 June 2015 in the V, R and I-bands. Flares are labelled for clarity using vertical arrows. Bottom: behaviour of the four comparison stars, plotted at the same scale, which remained constant in brightness within 0.01–0.02 mag.

Open with DEXTER
In the text
thumbnail Fig. 2

Top: variability of colour indices VR, VI and RI of V404 Cygni in outburst as observed with the UJT on 26 June 2015. They are shown plotted in black, red, and magenta colours, respectively. Bottom: the same kind of plot for the four comparison stars plotted at the same scale. Their colours remained constant within 0.02–0.03 mag.

Open with DEXTER
In the text
thumbnail Fig. 3

Evolution of V404 Cygni in the colour–colour plane during a few hours on 26 June 2015 as observer with the UJT. The dotted line is a linear regression fit guiding the eye to better appreciate the colour–colour correlation suggested in the text.

Open with DEXTER
In the text
thumbnail Fig. 4

Dereddened light curves of V404 Cygni in outburst as observed with the UJT on 26 June 2015 in the V, Rc and Ic-bands assuming an interstellar extinction of AV = 4.0 mag. The brightness level has been converted from magnitudes to a flux density scale, in mJy units. Flares are labelled as in Fig. 1.

Open with DEXTER
In the text
thumbnail Fig. A.1

CCFs of the optical light curves V vs. Rc (black), V vs. Ic (blue) and Rc vs. Ic (red). The vertical dashed line corresponds to zero time lag. The V vs. Rc CCF is normalized to unity while the other two have been slightly scaled above and below for easier display.

Open with DEXTER
In the text
thumbnail Fig. A.2

Zoomed view of the central maxima of the CCFs where a clear asymmetry is visible. Colours and normalization are as in Fig. A.1.

Open with DEXTER
In the text
thumbnail Fig. A.3

Dependence of the derived time lags using the cross-correlation technique on the amount of interpolated sampling of the data. Colours are as in Figs. A.1 and A.2.

Open with DEXTER
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.