Bull. Astr. Soc.

India (2006) 34, 19

Broadband radio spectrum of SS433

Centre for Space Physics, Chalantika 43, Garia Station Rd., Garia, Kolkata 700084, India S.N. Bose National Centre for Basic Sciences, JD Block, Salt Lake, Sector III, Kolkata 700098, India 3 Max Plank Institute for Radio Astronomy, 53902 Bad Munstereifel-Eelsberg Germany Received 28 September 2005; accepted 8 December 2005

Abstract. We obtained a broadband radio spectrum of the galactic compact object SS433 through the multi-wavelength campaign using the Giant Meter Radio Telescope (GMRT) and the Eelsberg 100-m Radio Telescope. The observations took place from 2005 January 03 to 2005 February 08 using eight dierent frequencies. We observed a are which appeared to have started prior to January 18 and lasted till February 08. The light curves show a progressively larger time-delay as the radio frequency goes down. The peak of the 6 cm lightcurve is delayed by 3 days with respect to the peak of the 9 mm lightcurve. There is an indication of the attening of the spectrum and possibly a turnover at 1.5 GHz. If we interpret this to be due to synchrotron self-absorption, the required magnetic eld would become too large, however, free-free absorption by hot thermal surrounding medium formed due to stellar winds is still a possibility. All sky monitor (ASM) aboard Rossi X-ray Timing Experiment (RXTE) showed very high X-ray count on 2005 January 25 when the are was well underway. This may be due to slamming of the radio bullets with previously ejected, relatively slowly moving material. Keywords : Stars (individual): SS433 Radio Radio sources Radiofrequency spectra

1 e-mail: 2 e-mail:

space phys@vsnl.com chakraba@bose.res.in 3 e-mail: kraus@mpifr-bonn.mpg.de

S. Pal et al.

1.

Introduction

There have been quite a few multiwavelength campaigns in Radio frequencies in SS433 in the past (e.g., Seaquist et al. 1980, Seaquist et al. 1982, Vermeulen et al. 1993; Bursov and Trushkin, 1995) which studied SS433 when it was in quiet state as well as when it had ares. While some have claimed to obtain synchrotron turnover (Seaquist et al. 1980) at a low frequency when SS433 was in a quiet state and turnovers at very high frequencies (Seaquist et al. 1982), others have found clear evidences of increasing timedelay (on the order of a few days) as the observation frequencies go down (e.g., Bursov and Trushkin, 1995). It is generally found that at frequencies higher than the turnover frequency, the spectral slope is 0.6, i.e., I 0.6 . In the current Paper, we present one of our recent observations independently verifying most of these assertions using two dierent radio telescopes. The observations were carried out during 2005 January 03 to 2005 February 08. For observations at 9 mm, 1.3 cm, 2.0 cm, 2.8 cm and 6 cm, we used the 100 m Eelsberg Radio telescope, located at Eelsberg, Germany and for observations at 21.4 cm, 23.4 cm and 29.4 cm we used the Giant Meter Radio Telescope (GMRT) located near Pune in India. We obtained the lightcurves at these frequencies and observed a are. The spectrum of 2005 February 04 showed an indication of a turnover at 1.5 GHz. This could be due to synchrotron self-absorption or due to free-free absorption. We combine our results with the All Sky Monitor (ASM) data of RXTE satellite to show that SS433 was perhaps active in X-rays also in this period. In the next Section, we present the discussion on the observing techniques and the observational results. In Section 3, we present the discussion on the observations and nally, in Section 4, we present our results.

2.

Instruments used and data reduction

The multi-frequency radio observations in the wavelength range of 9 mm to 6 cm (frequency range of 4.85 GHz to 32 GHz) were carried out using the 100m Eelsberg radio telescope. Since the source was point-like (compared to the telescope beam) and quite > bright ( 0.3 Jy) in the range of frequencies observed, we were able to determine the ux densities with cross-scans: these consisted of 8 to 32 (depending on the observing frequency and the source strength) individual sub-scans through the source position. Half of these sub-scans (each with a length of about 4 to 5 telescope beams) were performed in azimuth and half in elevation. This enabled us to check the position osets in both the coordinates and correct for those. As the rst step of the data analysis, a Gaussian prole was tted to every sub-scan. The amplitude of these Gaussians (i.e., the result of the convolution of the point-like brightness distribution of the source with the antenna beam) was a measure of the ux density of the source. After applying a correction for small pointing osets, the amplitudes of all individual sub-scans in one cross-scan were averaged. At 2 cm, the sub-scans of

Broadband radio spectrum of SS433

one direction (azimuth or elevation) were averaged before tting the Gaussian in order to increase the SNR. Additionally, for the short wavelengths, the opacity of the atmosphere was determined and each measured ux density for the atmospheric attenuation was corrected. After correcting the measurements for the elevation-dependent sensitivity of the antenna, our observations were calibrated to the absolute ux density scale (Baars et al. 1977, Ott et al. 1994) by using the primary calibrators: 3C286, 3C48, and NGC7027. These observations could also be used to correct for a possible time-dependence of the telescopes sensitivity. The nal measurement errors were derived from the formal statistical errors and a contribution from the residual uctuations of the calibrators was obtained. The resulting measurement uncertainties usually lay in the range of 0.3 1 % at = 6, 2.8 cm; for the shorter wavelengths they were somewhat higher (1-5 % depending e.g., on weather conditions). The multi-frequency radio observations in the wavelength range of 21.4 cm to 29.4 cm (frequency range of 1020 M Hz to 1400 M Hz) were carried out using the Giant Meter Radio Telescope (GMRT) (see, Pal and Chakrabarti, 2004 for details). We used 16 MHz bandwidth. Each data was binned for 16 seconds. Bad data was agged. The data was background subtracted, calibrated, band-passed and channel averaged using appropriate tools of AIPS package. 3C48 and 3C286 were used as ux calibrator and 2011-067 was used as phase calibrator.

3.

Observational results

In Fig. 1, we show the multiwavelength lightcurves of SS433 from 2005 January 03 to 2005 February 08. The actual data points with error-bars are shown both for the Eelsberg and GMRT telescopes. The long & short-dashed, dotted, dashed, dash-dotted, long-dashdotted curves are respectively for 23.4 cm, 6 cm, 2.8 cm, 1.3 cm and 9 mm respectively. The observations at 21.4 cm (square) and 29.4 cm (cross) were made on JD 2453405+, i.e., on February 04 when observations were carried out simultaneously at Eelsberg. The light curves give indication of a are. This becomes prominent when we extrapolate on each days data and imagine the observed peak to be located at the intersection point (marked by solid squares which join nearby observed data points by solid lines). If we had taken continuous observations, we would expect the results at or near this point. If this linear decay of the source ux is assumed, then one could measure the delay as a function of frequency (see, Fig. 3). Another interesting property is that the low energy radiation decays faster than high energy emissions. For instance, at 6 cm, the source ux is reduced to half in just 4.5 days, while at 9 mm, it takes 9 days for the same fractional reduction. On the same plot, by a vertical arrow drawn at JD 2453395.12, we show the time-frame when a possible X-ray are occurred as per ASM/RXTE data. Fig. 2 shows the ASM plot of photon counts per second. SS433 being a very weak source, even negative counts have been reported on many days. However, on JD 2453395.12 and the

S. Pal et al.

Figure 1. Lightcurves of SS433 during the campaign at various frequencies marked. The solid pieces of the curves near the peak have been obtained by linear interpolation of the observed data to guess how the lightcurves would have looked like if the continuous observations were made. The solid squares are the interpolated peaks clearly showing several days of delay in between high and low frequency results.

day after, the count rates were very high. This could not have triggered the radio are detected by us since the are seemed to have started before hand. The other possibility is that the radio emitting ejecta or bullets (Chakrabarti et al. 2003) may have slammed into some previously ejected slowly moving component to produce this X-ray emission (e.g., Migliari 2003). On the other hand contamination from the sun cannot be ruled out as the ASM observes a wider section of the sky. In Fig. 3, we show the plot of the time delay (days) as a function of the frequency. The solid circles are the actual delay measured with respect to the peak emission at 9 mm. The superposed curve corresponds to t(days) t0 ( 0 )1/2 , where, t0 = 5.0d, 9 0 = 1.310 Hz. The delays of 6.0 cm, 2.8 cm and 1.3 cm with respect to 9 mm are 3.08 d, 2.0 d and 0.9 d respectively. Such delays of the order of a few days have been reported before (Bursov and Trushkin, 1995) as well. If t is the delay with respect to 9 mm are, it is easy to see that t , with 0.8. If we consider adiabatic expansion and employ the radio spectral index = 0.66 as observed in our data (see Fig. 4 below), then the particle spectrum (N (E) E p ) has the index p = 2 + 1 = 2.32. This can

Broadband radio spectrum of SS433

Figure 2. All sky monitor (ASM) results of the photon count rate during the campaign showing a signicant rise of X-ray counts in the midst of the radio are for a very short period denitely conned between JD2453395.1 and JD2453395.7.

be used to obtain the delay of the are peak at dierent frequencies from tm, = t0 4p+6 (van der Laan 1966). The proportionality constant t0 can be adjusted to get the peak at 9 mm in a way that the delays at other frequencies roughly match. We nd that t0 = 3.0 gives the delays at 6.0 cm, 2.8 cm and 1.3 cm with respect to 9 mm as 3.5 d, 1.8 d and 0.5 d respectively. Thus, according to van der Laan (1966) model, the are at 9 mm started about tm,0.9mm = 2.87 days prior to the estimated peak location at JD2453396.55, i.e., at JD2453393.7. This is not all that improbable as the beginning of the are could indeed be anywhere between JD2453386 and JD2453394 as the data is absent in between. In order to check if the time delay is due to adiabatic expansion, and the are is an isolated event, we can use a consideration similar to Ishwar-Chandra et al. (2002) applied to GRS 1915+105 and compute p and and the resulting time delays. These numbers seem to be o by a large margin. For instance, comes out to be 0.74 at 9 mm, but 2.54 at 6 cm. In Fig. 4, we present our multi-wavelength spectrum on three dierent days (i) on 2005 January 16 before the are began (ii) on 2005 January 30 when the are was on the way to decline and (iii) on 2005 February 04 when the are was over. The ux at high

p+4

S. Pal et al.

Figure 3. The delay in the peak intensities is plotted as a function of emitted frequency at solid circles. The solid curve is an analytical t of the form t(days) 5.0 ( [GHz] )1/2 . 1.3

frequencies on all the three days varies as 0.66 . There is an indication of a turn-over at around 1.5 GHz on February 04 observation. Given that the decay-time of ux is strongly frequency dependent, it is possible that the turnover is not real and in fact an artifact. It is to be noted that Seaquist et al., 1982, also reported turn-overs at around 3 GHz. This observation we have tted (solid line) with the following two-component model of synchrotron self-absorption of the form (e.g., Seaquist et al. 1982): S = S0 0 + S1 1 1 exp(1 1 2.5 ) 1 1 2.5 (1)

where, we nd the tted values to be S0 = 0.51, S1 = 1.1, 0 = 1 = 0.6 and 1 = 2.8. Here 1 is the optical depth at = 1GHz. The pair (S0 , 0 ) corresponds to the optically thin quiescent state, and (S1 , 1 , 1 ) characterizes the optically thick component modeled as a uniform slab which could be variable. These parameters can be correlated to 1/4 and B (see, Seaquist et al. 1982) and we obtain /B ) 0.34 102 (arcsecond Gauss1/4 ) which is comparable to what Seaquist et al. (1982) found. Since the radio emission from the jets of SS433 is mainly due to synchrotron radiation by the non-thermal electrons, the spectrum shows a power-law nature of spectral index = (p 1)/2, where p is power-law index of non-thermal electron distribution. The

Broadband radio spectrum of SS433

Figure 4. Spectra of SS433 in radio frequency before, during and after the aring event. While the results of 16th Jan. (JD 2453386.04) and 30th Jan., 2005 (JD 2453400.03) follow power law emission, that on the 4th of Feb. 2005 (JD 2453405.83) shows an indication of a turnover at around 1.5GHz.

emitting medium is optically thick for low frequency < s and remains optically thin for high frequency > s . The synchrotron self absorption frequency s is calculated by equating the observed ux with the source function for non-thermal emission. It gives the relation (see, Pacholczyk, 1970), 40
2 5/2 B 1/2 s , where, f (p) =

S = 66.2f (p)

1 1p

3p+19 12 3p+22 12

3p1 12 3p+2 12

3 4 5 4

. (2)

Here, S is the observed ux in Jansky, is the angular size of the emitting region in milliarcseconds (mas), B is the magnetic eld (averaged over the pitch angle) in Gauss and s is in GHz. Putting, 40mas and p = 1.65 as seen from the slope of the spectrum. We get a measure of the magnetic eld for s 1.5GHz to be more than 20000 Gauss which is quite impossible. On the other hand, if the turn-over is due to the two component free-free absorption,

8 the data could be tted with

S. Pal et al.

S = S0 0 + S1 1 exp[1 2.1 ],

(3)

with S0 = 0.39, S1 = 1.21, 0 = 1 = 0.6 and 1 = 0.91. In this case, the optical depth 1 at 1 GHz is related to the emission measure E and the electron temperature Te by, = 2.74 1020 ( Te 1.35 E ) ( 5 ). K cm (4)

For a standard stellar mass loss rate of M 105 M yr1 , the velocity of the wind of 1000 km s1 and the distance of the radio emission region of 1015 cm (Margon, 1984; Chakrabarti et al. 2005), the emission measure E becomes (Seaquist et al. 1982) E = 5 1025 cm. When we use this in Eq. (4) we nd the temperature of the emission region to be around 37, 700 K.

4.

Discussion and concluding remarks

In this Paper, we obtained a broadband radio spectrum of SS433 using the Giant Meter Radio Telescope (GMRT) and the Eelsberg 100-m Radio Telescope. We observed a moderate are. The light curves show a progressively increasing time-delay as the radio frequency goes down. The peak of the 6 cm lightcurve is delayed by 3 days with respect to the peak of the 9 mm lightcurve. If we t the van der Laan (1966) model of the expansion of the jets, we nd that the are may have started around JD2453393.7 i.e., around 2005 January 23. There is an indication of the attening of the spectrum and possibly a turnover at 1.5 GHz. If we interpret this to be due to synchrotron self-absorption, the required magnetic eld would become unreasonably large. However, the free-free absorption due to hot thermal winds is still a possibility for the observed turnover. On the other hand, given that the ares at lower frequencies decay faster, this turnover could be an artifact. All sky monitor (ASM) aboard Rossi X-ray Timing Experiment (RXTE) showed a very high X-ray count on 2005 January 25, when the are is well underway. From ASM data we deduce that the X-ray are, if any, must take place between JD2453395.1 and JD2453395.7, whereas the radio are started around or before JD2453393.7. We believe that this could be due to a glitch or some other inexplicable reason. This work is partly supported by a CSIR SRF (SP) grant and from ISRO RESPOND (SM). The authors also thank the RXTE team for the ASM light curve.

References
Baars, J. W. M., Genzel, R., Pauliny-Toth, I. I. K., and Witzel, A., 1977, ApJ, 61, 99. Bursov, N.N. and Trushkin, S.A. 1995, Astr. Lett., 21, 145. Chakrabarti, S.K., Pal, S., Nandi, A., Anandarao, B.G., and Mondal, S., 2003, ApJ, 595, L45.

Broadband radio spectrum of SS433

Chakrabarti, S.K., Anandarao, B.G., Pal, S., Mondal, S., Nandi, A., Bhattacharyya, A., Mandal, S., Sagar, R., Pandey, J.C., Pati, A. and Saha, S.K., 2005, MNRAS, 362, 957. Ishwara-Chandra, C.H., Yadav, J.S., and Pramesh Rao, A., 2002, A&A, 388, L33. Margon, B. 1984, ARA&A, 22, 507. Migliari, S., Fender, R. and Mndez, M., 2002, Science, 297, 1673. e Ott, M., Witzel, A., Quirrenbach, A., Krichbaum, T. P., Standke, K. J., Schalinski, C. J. and Hummel, C. A. 1994, A&A, 284, 331. Pacholczyk, A.G., 1970, Radio Astrophys. (San Francisco: Freeman). Pal, S. and Chakrabarti, S.K., 2004, A&A, 421, 13. Seaquist, E.R., Gilmore, W.S., Nelson, G.J., Payten, W.J. and Slee, O.B., 1980, ApJ, 241, L77. Seaquist, E.R., Gilmore, W.S., Johnston, K.J. and Grindlay, J.E., 1982, ApJ, 260, 220. van der Laan, H. 1996, Nature, 211, 1131. Vermeulen, R.C., McAdam, W.B., Trushkin, S.A., Facondi, S.R., Fiedler, R.L., Hjellming, R.M., Johnston, K.J. and Corbin, J., 1993, A&A, 270, 189.