The Active Algol Binary KZ Pavonis
E. Budding,
S. C. Marsden ,
O. B. Slee
, PASA, 18 (2), in press.

---

Next Section: Discussion
Title/Abstract Page: The Active Algol Binary
Previous Section: Introduction
Contents Page: Volume 18, Number 2

---

Subsections

• Radio data
• Flux density versus orbital phase
• Optical photometry
• Optical light curve analysis

---

Observations

We were granted 24hrs of continuous ATCA time in August 1998.¹ There are six 22-m dishes in the ATCA, accurately mounted on an east-west baseline. They can be arranged at appropriate separations to obtain specific angular resolutions at preassigned microwave observation frequencies (for further instrumental details cf. e.g. Manchester, 1991). A double C-band experiment was carried out, i.e. we employed two adjacent 128MHz bands at 4.848 and 4.976 GHz, which were co-added to improve the signal-to-noise properties of the observations. The full 6-km array with 6 antennas was used. High angular resolution was desirable for KZ Pav, both to get a good position and to rule out confusion over possible emission from close-by companions. The ATCA data covered slightly less than one complete orbital cycle in a continuous run. Combining such observations with temporally close optical data is effective for physical interpretation. Photometric data were arranged from the Mt Kent Observatory of the University of Southern Queensland, Toowoomba, and were gathered over the period Aug-Sep, 1998.

Radio data

ATCA observations started at 02h 12m UT on Aug 28, 1998 and continued until 23h 00m UT on Aug 29. The discrete source PKS B 0823 -500 provided primary calibration of the flux levels at 4.848-GHz and 4.976-GHz. The high declination source PKS B 2146 -783, conveniently placed for phase calibrations, was observed for 4-minute integrations, before 25-minute dwell times on KZ Pav. Excellent phase stability within 2 was maintained throughout the observations. The ATCA data were conveniently handled on-site using the MIRIAD data reduction package (Sault & Killeen, 1996). The full available instrumental bandwidth for these microwave ranges is 128 MHz. Our processing procedure made use of the `birdie' option in the MIRIAD routine ATLOD, which permits 13 x 8-MHz independent, interference-free, non-overlapping channels to be used for either band, summing to 104 MHz of total bandwidth. The flux levels used to construct the light curves discussed in what follows were read from uncleaned sky maps produced from the recorded visibility data via the MIRIAD data-reduction suite.²
Source No. 2 is KZ Pav, Aug 28, 1998. Position errors are 0.5 arcsec.
We concatenated the uv data from both IF bands and then produced a clean map of a large area around the position of KZ Pav. The rms variation over the cleaned map was close to the theoretical value of 22 Jy/beam. This map (Fig. 1) shows three clear radio sources including KZ Pav (source No 2). The measurements are summarized in Table 2. Each source was fitted with a 2-dimensional elliptical Gaussian superposed on a linear baseline, from which positions, flux densities (background subtracted) and angular sizes at the mean frequency of 4.912 GHz were determined. All three sources were unresolved. The position errors in the fitting are 0.05s in RA and 0

14 in dec., while flux densities have a formal error of 0.04 mJy. Our measured position for KZ Pav may be compared with that of the HIPPARCOS catalogue, i.e. (1998.75) RA 20h 58m 40.119s; dec -70 25 19.80

. An enlarged extract of the field about KZ Pav is shown in Fig. 2. The digital sky survey was searched for optical identifications for sources 1 and 3, but no coincident objects were found above the plate limit of m_J = 22.0. The formal error of 0.06 mJy in each of the two IF bands is too large to enable a useful measurement of the radio spectral index over the 128 MHz frequency separation of the two bands.

Flux density versus orbital phase

Microwave raw data on KZ Pav, covering the whole orbital phase range in the 4.8480 and 4.976-GHz bands, consisted of 41 individual MAXFIT (from MIRIAD) evaluations of the 25 m integrations. Flux densities, carefully derived from both `peak' and `profile' fitting techniques to point source images, were found to be not greater than 1 mJy in either frequency band during the run, and values of around 0.3 mJy, including background sky noise, were typically measured. Noise levels for the summed data sets are about 0.08 to 0.12 mJy in both bands, increasing with the atmospheric air mass. More information about this preliminary data was presented previously (Budding et al., 1998b). Signal levels in these original data sets were too weak to allow useful detailed analysis, however, and we concentrate on longer time interval integrations in what follows, (cf. Figure 3, which shows the co-added data and 1 error bars).

The combined data curve (Fig. 3), with 2-hour integrations, shows some enhancement centred around phase 0.7-0.8, a tendency to drop thereafter, and a further, briefer increase around phase 0.3. The same general trend is visible in the two separate contiguous C-band data sets, co-added to produce Fig. 3 (cf. Budding et al., 1998b). The co-added flux densities have 1- errors ranging between 70 and 100 Jy/beam. Fig. 3 shows noticeable enhancements at orbital phases 0.14-0.34 and 0.60-0.80. testing of the conformity to a uniform distribution in phase of the combined flux values, about their mean (Table 3), shows non-uniformity to be significant at about the 20% level, based on Poissonian error values. In other words, the evidence for phase dependence of the apparent variation is not very persuasive: such non-uniformity would occur about once in five trials if of purely random origin. If the variation with phase were real, there would be an argument for high temperature sources smaller in size than the scale of component radii. Such sources are frequently discussed for active stars. Postulating a source region of typical radius a third that of the secondary component, we have an emitting area of about

. At a distance of 99pc, a flux enhancement of 0.1 mJy corresponds to a full luminosity of about W/Hz at the source. Dividing this source emission by the foregoing area, we derive a mean brightness temperature of about K. Such a brightness temperature is typical of the emission generation scenarios of gyrosynchrotron radiation from mildly relativistic electrons in moderate magnetic fields (Dulk, 1985), such as are posited for RS CVn-like stars. Further details for this kind of source were given by Budding et al. (1999) for CF Tuc. In the case of CF Tuc, this emission scenario was supported by contemporaneous optical effects, indicating large-scale active regions on the subgiant.

Optical photometry

KZ Pav was observed at the University of Southern Queensland's Mount Kent Observatory, using the 40cm Webb telescope and SBIG ST6 CCD camera (cf. Waite, 1999) equipped with a standard V filter. Observations from 8 nights, between 29 Aug 1998 and 30 Sep 1998, were used. The comparison and check stars used were HD199190 and HD198971, respectively (cf. Walker & Budding, 1996). Walker & Budding's (1996) photometric calibrations of these stars were adopted for the present reductions. The light contribution of the companion star was retained in the tabulated magnitudes. Walker & Budding (1996) estimated the magnitude and colour of this companion as V = 8.144 and B - V = 0.513, and corrected their light and colour curves accordingly. The photometric data were analysed using the DAOPHOT routines of IRAF (cf. Stetson, 1998). No significant variations between the comparison and check stars were found for any of the nights. In order to determine the changes in photometric conditions during the course of observations, the comparison star's data were fitted with a cubic polynomial for each night. The standard deviation of the comparison star data to the fitted polynomial was less than

mag for data accepted as reliable. This polynomial was then subtracted from the KZ Pav magnitudes, to remove any spurious variations during the observations. If the comparison star data were found not to be monotonically increasing or decreasing during the course of a night, or if this data could not be satisfactorily fitted by the polynomial (due to variations from either dew or cloud), then the data were rejected as `non-photometric'. In this way, two complete nights of data and parts of another three nights were rejected. A total of 569 individual data points for KZ Pavonis were then left for analysis. In the IRAF reduction procedure observation times (accurate to 1 second) were converted to Heliocentric Julian Dates, and the phase-folded light curve (Figure 4) produced, using the ephemeris:

(cf. Mallama, 1982). The light curve shows the following general features: (a) two well-defined minima, (b) the light curve is not quite complete in a region just after primary eclipse, (c) the eclipse minima occur somewhat later than the ephemeris predicts, (d) the secondary minimum is slightly asymmetric, (e) there is rather a large scatter in the data just after secondary eclipse, and some suggestion of a depression before that eclipse. There is no convincing evidence of any large-scale maculation effect. The low excursion around phases 0.35-0.42, for example, has an amplitude of not more than 0.01 mag (cf. also Figs. 5,6 ), i.e. about the 1 level for the standard model curve-fitting (Table 4).

Optical light curve analysis

The light curve, prepared with phases and differential magnitudes corresponding to the foregoing information, was subsequently binned into 75 `normal' points for ease of analysis, without significant information loss. Curve fitting was carried out using the Information Limit Optimization Technique (ILOT) (cf. Banks & Budding, 1990; Budding, 1993). An adopted optimal curve-fit is shown in Figure 5. The solution was checked using the Binary Maker software of Bradstreet, which utilizes the numerical integration procedure of Wilson and Devinney (1972). The Binary Maker fit and corresponding `semi-detached' model are shown in Figure 6. The parameters used for this fit are based on the ILOT solution, with slight changes to some of the values. These changes are within the errors determined by the ILOT programme's evaluation of the error matrix in the vicinity of the optimum solution. The primary temperature from Table 1 was adopted from its Main Sequence spectral type, (cf. Walker & Budding, 1996). Finally adopted model parameters are listed in Table 4 (cf. also Table 1). The differences between observed points and the theoretical model are shown in Fig. 7. Any systematic trends in such differences are difficult to make out against what looks like scatter at the 0.01 mag level. If we suppose that this scatter contains effects such as maculation from e.g. low latitude spots, such variation would imply active regions of projected radius 6 if set against the whole system light. However, the secondary component, on which such active regions are expected to be located, accounts for only 14% of the system light. Spot radii of up to 16 could therefore exist on that star (i.e.

0.3 x its radius) without there being unambiguous evidence of their presence in the optical domain.

---

Next Section: Discussion
Title/Abstract Page: The Active Algol Binary
Previous Section: Introduction
Contents Page: Volume 18, Number 2

---

© Copyright Astronomical Society of Australia 1997