VICTOR BYCHKOV Special Astrophysical Observatory, Russian Academy of Sciences, Nizhnij Arkhyz, Karachai-Cherkessia , Russia; vbych=sao.

Transcription

1 THE ASTRONOMICAL JOURNAL, 119:1542È1561, 2000 April ( The American Astronomical Society. All rights reserved. Printed in U.S.A. RAPID POLARIZATION VARIABILITY IN THE BL LACERTAE OBJECT S5 0716]714 CHRIS D. IMPEY Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721; cimpey=as.arizona.edu VICTOR BYCHKOV Special Astrophysical Observatory, Russian Academy of Sciences, Nizhnij Arkhyz, Karachai-Cherkessia , Russia; vbych=sao.ru SANTIAGO TAPIA MIT Lincoln Laboratory, 244 Wood Street, Lexington, MA 02420; stapia=gis.net YURI GNEDIN Pulkovo Observatory, M-140, St. Petersburg , Russia; gnedin=pulkovo.ru AND SIMON PUSTILNIK Special Astrophysical Observatory, Russian Academy of Sciences, Nizhnij Arkhyz, Karachai-Cherkessia , Russia; sap=sao.ru Received 1999 July 13; accepted 2000 January 4 ABSTRACT We present optical polarimetry of the violently variable BL Lacertae object S5 0716]714, obtained over a three year period using the 6 m telescope at the Special Astrophysical Observatory in Russia. The degree of polarization is high and variable throughout the observations. With a minimum time resolution of 1 minute, signiðcant variations on a timescale of 10È15 minutes are observed. The variations are complex and not fully sampled by this data, but they are characterized by large amplitude changes on a timescale of about 1 day superposed on smooth variations with a timescale of about 1 week. Possible periods of 12.5, 2.5, and 0.14 days have been discovered. S5 0716]714 is a highly compact radio source with nonthermal emission observed across the electromagnetic spectrum out to gamma rays. The optical polarization variations are consistent with relativistically beamed synchrotron emission viewed at a very small angle to the line of sight. Key words: BL Lacertae objects: individual (0716]714) È galaxies: nuclei È polarization È quasars: general È radio emission lines 1. INTRODUCTION The main elements of the standard paradigm of active galactic nuclei (AGNs) have been in place for nearly 35 years. The power source is presumed to be a gravitational engine in the form of a supermassive black hole (Zeldovich & Novikov 1964; Salpeter 1964). Fuelling of the central engine takes place via an accretion disk (Lynden-Bell 1969). Nonthermal radiation is emitted over a wide range of frequencies (Hoyle, Burbidge, & Sargent 1966). The plasma may be in bulk relativistic motion, and this anisotropic emission leads to a di erence between the observed and intrinsic properties of AGNs (Rees 1966; Woltjer 1966). Finally, the variable light output of quasars was used soon after their discovery to argue that the emission regions are small, based on light-travel arguments (Mathews & Sandage 1963). A subset of AGNs are not only variable but also show linear polarization at a level higher than is expected for interstellar processes. High and variable polarization is seen almost exclusively in strong radio sources, and it is ascribed to optically thin synchrotron emission from regions with organized magnetic Ðelds (Angel & Stockman 1980; Saikia & Salter 1988). AGNs with compact radio emission, rapid variability, and high optical polarization are collectively called blazars. ÏÏ Variability is seen across the electromagnetic spectrum, but it tends to be most rapid at short wavelengths (Wagner & Witzel 1995). Observation of signiðcant variations on timescales shorter than 1 day leads to constraints on source regions smaller than the solar system. In fact, the regions that give rise to variable and polarized optical emission are probably shocks within the nonuniform Ñow of a relativistic jet (Blandford & Ko nigl 1979; Marscher 1992). A summary of the properties of the variable BL Lacertae object S5 0716]714 (hereafter referred to as 0716]714 ÏÏ) is presented in 2. This is followed by a description of rapidly time-sampled polarimetry obtained over a period of three years in 3. In 4, we present di erent ways of characterizing the variable polarization, and we relate the optical behavior to variations at other wavelengths. Conclusions ( 6) follow a brief discussion of the implications of these observations for models of the AGN power source and the emission mechanism in THE BL LAC OBJECT 0716]714 S5 0716]714 is contained in the catalog of radio sources stronger than 1 Jy at 5 GHz (Ku hr et al. 1981). It has a Ñat radio spectrum and a core-halo structure on arcsecond scales (Antonucci et al. 1986), consistent with the morphology of a moderately powerful radio galaxy seen end-on. The overall Ñat radio spectrum seems to be the result of a superposition of a very compact core and several extended components (Eckart et al. 1986). VLBI observations show a dominant core and a one-sided jet (Eckart et al. 1987; Witzel et al. 1988). More recent VLBI data reveal proper motion in one jet component, which would only correspond to superluminal motion if z [ 0.4 for the source (Gabuzda et al. 1998). Polarimetry extracted from these maps reveals a high jet polarization of D50%, aligned with the jet axis, suggesting a transverse shock. The optical counterpart of 0716]714 is an unresolved object with a featureless spectrum, which earns it the classi- 1542

2 RAPID POLARIZATION VARIABILITY IN S5 0716] Ðcation as a BL Lac object. Unless the host galaxy is unusually subluminous, the redshift of the source must be z [ 0.3 (Wagner et al. 1996). No emission or absorption lines are seen over the wavelength range 3400È9500 A (Stickel, Fried, & Ku hr 1993). The BL Lac object 0716]714 has been detected at far-infrared (Impey & Neugebauer 1988), ultraviolet (Ghisellini et al. 1997), and X-ray wavelengths (Biermann et al. 1981, 1992). The source is also one of several dozen blazars that has been detected in gamma rays by the EGRET experiment on the Compton Gamma Ray Observatory (Lin et al. 1995). In fact, 0716]714 is a bright and popular target at most wavelengths; the NASA/ IPAC Extragalactic Database (NED) contains over 150 bibliographic references. With archetypal behavior for a blazar, 0716]714 is variable at all wavelengths across the electromagnetic spectrum. Wagner and collaborators have included this source in a number of monitoring campaigns; the source is almost always active, with variations on a timescale less than 1 day. On occasion, optical variations can be faster than 1 hr (Sagar et al. 1999). The most intensive observations revealed correlated optical and radio variations, which is evidence that the radio variations are intrinsic and not caused by interstellar scintillation (Wagner et al. 1996). This in turn implies an extremely high brightness temperature of 1018 K for the radio source (Heeschen et al. 1987; Quirrenbach et al. 1991), an estimate that uses the lack of visibility of a host galaxy to place a lower bound on the source redshift. The variability behavior provides independent support for the idea that the nonthermal emission is relativistically boosted. The BL Lac object 0716]714 has one of the highest duty cycles of variability of any blazar. Less is known about the polarization properties of 0716]714. The radio core polarization is 2%È4%, and the polarization of the VLBI jet is close to 50% (Gabuzda et al. 1998). No optical polarimetry was published until 1994 (Takalo, Sillanpa a, & Nilsson 1994), and the two measurements showed a polarization change of 3.5% in 1 day. The degree of polarization increases toward shorter optical wavelengths. Polarization measurements were also reported by Wagner (1998). This current paper provides an enormous increase in the number of published observations and allows us to characterize the polarization variations for the Ðrst time. 3. POLARIZATION OBSERVATIONS The observations presented here were made using the MINIPOL polarimeter on the 6 m telescope of the Special Astrophysical Observatory in the Stavropol territory of the Caucusus, in Russia. Polarimetry of 0716]714 was gathered during three observing runs in 1991, 1993, and 1994 SeptemberÈOctober. Observing conditions varied from clear and photometric to thin cirrus. MINIPOL was mounted at the prime focus, and the measurements were made through a circular aperture of diameter 4A. MINIPOL uses a Wollaston prism and two GaAs phototubes for high throughput. A superachromatic half-wave plate rotates through a full modulation cycle in 12 ms, allowing precise polarimetry in nonphotometric conditions. The characteristics of the instrument are described by Dolan & Tapia (1986). The sensitivity of MINIPOL in photon Ñux units has been calibrated over the entire GaAs wavelength range (3200È8800 A ). For the energy distribution of 0716]714, which is typical of strong, compact radio sources, the e ective wavelength of the polarization measurement is D5500 A, corresponding roughly to the V band. We did not perform photometric calibration, and no photometry is reported in this paper. In practice, the requirements for accurate photometry are far more severe than the requirements for accurate polarimetry. With over 80 modulation cycles per second, the di erential measurement of polarization using MINIPOL is immune from transparency variations due to thin cirrus. Moreover, since almost no instrumental polarization is imprinted by scattered light from the edge of the aperture, polarimetry using MINIPOL is invariant with the exact percentage of source light collected by the aperture. By contrast, photoelectric photometry is highly sensitive to seeing and transparency variations. The observing sequence consisted of multiple 1 minute object integrations followed by a single 1 minute integration on a blank sky patch located 20AÈ30A away. To reduce the statistical error of the sky observation, the number of object integrations, M, per sky integration was chosen such that 1/M was less than the square root of the ratio of sky counts to object counts. We conðrmed that the polarization error reduced according to photon statistics. In other words, the observed from the Ðt to the modulated signal agreed with the predicted polarization error, 100(2/N)1@2, where N is the number of counts. A detailed discussion of the precision and repeatability of MINIPOL observations is given by Impey, Malkan, & Tapia (1989). We checked for instrumental polarization using the unpolarized standard star BD ] from the list of Turnshek et al. (1990). The instrumental polarization was too low to be detected, with a measurement of 0.07% ^ 0.14%. The same standard star was used in conjunction with a Glan-Thompson prism to measure the modulation efficiency of MINIPOL. The modulation efficiency was D98%; we did not correct the observed polarizations for this small e ect. In principle, since the degree of polarization is a positive-only quantity, it must be corrected for low signal-to-noise bias when p \ 3% and p/p(p) \ 3 (Simmons & Stewart 1985). However, 0716]714 had high polarization, which was detected at a high degree of signiðcance during our observations. In this paper, the distinction between observed and unbiased polarization is smaller than the error in an individual measurement. The6mtelescope has an altitude-azimuth mounting, and MINIPOL did not rotate during the observations to preserve a Ðxed instrumental polarization. As a result, we must correct all the position angle measurements for the rotation of the sky. The position angle of each observation was h \ h [ h [ q, where q is the parallactic angle, h is n the polarization 1 0 position angle of a standard star, and h 0 is the relative telescope angle. We solved for h, using the 1 mid- point of each 1 minute integration, and found 1 it to be constant for all of the observation runs. In the north we found h \ 52.08^ 0.15 for 73 observations of BD ]64 106, and in 1 the south we found h \ 46.76^ 0.24 for 55 obser- vations of HD Each 1 of these stars was bright enough to cause double counting by MINIPOL, so a neutral density Ðlter of 4 mag was used for the standard observations. Since sky and object polarizations are not measured simultaneously, we used an additional Ðgure of merit for each observation. For each 1 minute integration, it was

3 TABLE 1 POLARIZATION OBSERVATIONS OF S5 0716]714 p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448,

4 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, [1.348 [ ,448, [1.273 [ ,448, [1.250 [ ,448, [1.258 [ ,448, [1.082 [ ,448, [0.871 [ ,448, [0.872 [ ,448, [1.002 [ ,448, [0.841 [ ,448, [0.384 [ ,448, [0.362 [ ,448, [0.474 [ ,448, [0.599 [ ,448, [0.285 [ ,448, [0.142 [ ,448, [0.238 [ ,448, [0.386 [ ,448, [0.307 [ ,448, [0.379 [ ,448, [0.348 [ ,448, [0.214 [ ,448, [0.256 [ ,448, [0.204 [ ,448, [0.249 [ ,448, [0.130 [ ,448, [0.201 [ ,448, [0.195 [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [

5 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448,

6 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [

7 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [ ,448, [3.607 [ ,448, [4.205 [ ,448, [4.128 [ ,448, [4.089 [ ,448, [4.237 [ ,448, [4.129 [ ,448, [4.090 [

8 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,448, [4.470 [ ,448, [3.873 [ ,448, [3.938 [ ,448, [3.359 [ ,448, [4.012 [ ,448, [3.447 [ ,448, [3.284 [ ,448, [3.676 [ ,448, [3.801 [ ,448, [3.386 [ ,448, [2.809 [ ,448, [3.253 [ ,448, [3.039 [ ,448, [3.172 [ ,448, [3.361 [ ,448, [2.713 [ ,448, [2.640 [ ,448, [2.765 [ ,448, [2.935 [ ,448, [2.838 [ ,448, [2.210 [ ,448, [2.894 [ ,448, [2.606 [ ,448, [2.660 [ ,448, [2.270 [ ,448, [2.469 [ ,448, [2.294 [ ,448, [2.347 [ ,448, [2.049 [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [

9 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, [ ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449,

10 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449, ,449,

11 1552 IMPEY ET AL. Vol. 119 TABLE 1ÈContinued p p(p) h p(h) Q/I U/I JD (2) (3) (4) (5) (6) (7) (1) (%) (%) (deg) (deg) (%) (%) 2,449, ,449, ,449, ,449, ,449, required that the error in the polarized sky counts be less than the error in the polarized object counts, N p(p ) \ N p(p ). Observations that did not satisfy this sky condition sky were obj rejected obj (less than 10% of the data was a ected, and most of that represented times when observations continued into a time of encroaching twilight). In this way, we ensured that the measured object polarization was not biased by a varying or imperfectly measured sky polarization. The polarimetric observations of 0716]714 are listed in Table 1. Column (1) gives the Julian Date of each 1 minute observation. Columns (2) and (3) give the degree of polarization in percent and its associated error. Columns (4) and (5) give the position angle in degrees and its associated error; position angle is measured from north through east in the equatorial system. Columns (6) and (7) give the normalized Stokes parameters Q/I and U/I in percent. A total of 531 measurements were obtained on 12 separate nights over a period of three years. The typical time span of monitoring on any particular night was 1È2.5 hr. 4. THE VARIATIONS OF 0716] Polarization and Position Angle The polarimetric variations during the 1991 campaign are summarized in Figure 1, which shows the degree of polarization and the position angle. Observations were FIG. 1.ÈDegree of polarization and position angle plotted for observations of S5 0716]714 during the 1991 observing season. made on seven nights during an 11 night run. A closer look at the data for each night is given in Figures 2aÈ2g. There are variations in both degree of polarization and position angle on all timescales sampled by the observations down to a fraction of an hour. The visual impression of Figure 1 is of rapid and irregular variations on a timescale of several hours, superposed on a smooth variation with a timescale of about 1 week. The most striking feature in Figure 1 is the rapid position angle rotation of 40 in only 2 hr on the third night of observation. The individual time sequences in Figure 2 conðrm that while there is jitter ÏÏ in the polarization and position angle from one minute to the next, these variations are not signiðcant. In particular, since the error in the polarized sky counts is often a signiðcant fraction of the error in the polarized counts from the BL Lac object, the formal polarization error of 100(2/N)1@2 can be a slight underestimate of the true error. Therefore, adjacent points in Figure 2 would have to di er by more than 3 times the error bars in each measurement to be considered a real variation. By contrast, the major variations on a timescale of 1 hr or more are highly signiðcant. It is noticeable that the polarization variation on a 1 hr timescale is similar to the amount of the interday polarization variation. This suggests that the variability on a typical timescale of 1 hr is well sampled by the 1 day spacing of the observations. The slowest variations appear to be inadequately sampled because they extend over the full span of the observations. The single night of observation in 1993 is considered in conjunction with the 1994 observations. Figure 3 shows four nights of data from 1994 obtained during a Ðve night run. An expanded view of the individual observations, preceded by the 1993 data, is shown in Figures 4aÈ4e. The overall degree of polarization is higher than in the 1991 observations. As with the earlier data, there are signiðcant polarization and position angle variations on a timescale of under 1 hr. Taken as a whole, 0716]714 spans a large polarization range, 2%È17%, and a broad range of position angle, 10 È120. While there is not a particular preferred position angle, two-thirds of the measurements fall in the relatively narrow range of 20 È60. These properties place 0716]714 in a bracket with the most violently variable BL Lac objects known, such as BL Lac itself and OJ 287 (Angel & Stockman 1980). Our polarimetry uses di erential photometry on a rapid timescale, so Stokes parameters can be calculated with a precision that is mostly governed by photon counting. However, the lack of an accurate photometric zero point limits the analysis of variations in total and polarized Ñux. This in turn places limitations on the analysis and interpretation of variations. For example, it is possible to imagine a polarizing screen that varied the polarized Ñux without a ecting total Ñux. Alternatively, if the emissivity of the

12 No. 4, 2000 RAPID POLARIZATION VARIABILITY IN S5 0716] FIG. 2.ÈExpanded view of the 1991 polarization and position angle variations, measured on the (a) Ðrst, (b) second, (c) fourth, (d) Ðfth, (e) ninth, ( f ) tenth, and (g) eleventh nights of the observing run. source changed but the geometry did not, then the total Ñux could change while the degree of polarization remained constant. In what follows, we restrict the analysis to variability in the degree of polarization Stokes Parameters An alternative way of presenting the data is in terms of Stokes parameters. Polarization has the disadvantage that it is a positive-only quantity with errors that are not normally distributed. The normalized Stokes parameters Q/I and U/I are dimensionless, and their variations can be considered independently. Figures 5È8 are analogous to Figures 1È4, except that all consecutive 1 minute integrations are connected by solid lines and a typical error bar is shown for each nightïs data (each separate night of data forms a very obvious clump in Figs. 5 and 7). An inspection of Figure 5 shows no obvious pattern to the variations of 0716]714 in the Q/I versus U/I plane. The best description of overall pattern is a random walk. At the level of intranight observations, the variations have the character of random variations from one minute to the next superposed on synoptic trends. On some nights the random jitter ÏÏ is dominant (Figs. 6a, 6b, and 6e), and on other nights a systematic trend is seen (Figs. 6c, 6d, 6f, and 6g). The 1994 observations summarized in Figure 7 show a similar behavior, with a mixture of random variations (as in Fig. 8a) and steady movement in the Q/U versus U/I plane (as in Fig. 8d). This level of accuracy and time resolution of