Long-term infrared light-curves of episodic dust-making WC7 stars.

Peredur Williams¹ & Karel van der Hucht²

¹Institute for Astronomy, University of Edinburgh, U.K.
²SRON Laboratory for Space Research, Utrecht, Netherlands.

Introduction

Earlier this year, we drew the line under — we cannot write "completed" as this looks like being a very long-term project — our long-running UKIRT Service programme 165PMW. This was a photometric monitoring programme directed at the initially unexpected phenomenon of episodic dust formation by a small sample of WC Wolf-Rayet stars. The time-scales on which these events occur are so long that their study requires either frequent access to a nearby telescope or a facility service such as the UKIRT Service Observing Programme.

The light curves

Many evolved stars of different types form dust particles in their outer regions. This is readily detected in the infrared, where the spectrum of the classical "infrared excess" is determined by the grain type and temperature.

Shown on the front cover of this Newsletter are 3.8-micron (L') light curves of the three stars observed in the programme (UKIRT data are marked with triangles). This is the best wavelength at which to track newly formed dust, which has a colour temperature ~1000K. The amplitude of variation in K is typically a magnitude smaller.

The 1977 dust formation episode by WR140 (HD193793) was discovered with the IRFC on Tenerife (the direct ancestor of UKIRT, using an ancestor of the photometer UKT1) and the fading of the dust emission was tracked with UKIRT, often in time generously contributed by visitors. Persistence was rewarded when another dust-formation episode was observed in 1985. The subsequent fading and the 1993 episode were observed in the UKIRT service programme, augmented with observations with other telescopes in an international campaign. The 7.94-year "period" inferred from the 1985 eruption and previous observations, and confimed in 1993, led to demonstration of a binary orbit and modelling of the radio and X-ray observations (e.g. Williams et al. 1990, Eichler & Usov 1993).

The rising infrared flux from WR137 (HD192641) was discovered during our infrared survey of late-subtype WR stars using UKIRT and ESO telescopes. Intensified observing identified the maximum in 1984.5 and subsequent observations over the next decade showed a decline slower than WR140's, because WR137 has a slower wind dispersing the dust, together with a number of secondary eruptions. After a further three years, the flux began to rise again to a major maximum in about 1997.7. Comparison with the earlier data suggests a period of ~13.2 years — but a conservative observer would want to see at least one more maximum (we expect the next in 2010.9) before being convinced of the periodicity!

The intervals between dust-formation episodes by WR125 appear to be even longer. In this case, the dust formation was discovered in Service Observing monitoring. We included WR125 in the programme when its radio emission faded like that of WR140 before its dust formation. We conjectured that the non-thermal radio emission arose between two stars in a binary system; that the radio fading was caused by the movement of that region deeper into the opaque Wolf-Rayet wind as the (conjectured) binary approached periastron passage and that dust formation might be triggered by movement of the wind collision system to a higher density part of the WR wind. The fourth year of monitoring showed the beginning of the dust formation episode, which continued until 1993; since then the IR flux has been fading as the dust has been dispersed by the wind. The H, K, L', M/nbM and [8.75]/N1 light curves have been modelled by Williams et al. (1994), who also found spectroscopic evidence that WR125 could be a binary. If it is a binary and the episodes are periodic, the period must be even longer than that of WR137. The light curve (front cover) shows that the L'/nbL flux had not yet faded to the wind level by 1999, almost nine years after it started to rise, and that there was no dust emission in 1982.5. This indicates that the previous outburst cannot have occurred later than 1973.5, giving a minimum period of 17 years.

FIGURE 1: Comparison of the density of the stellar wind going into the shock as a function of orbital phase (line) and observed K magnitude phased to the same elements (points). The delay in phase between the two maxima is attributed to the time the wind compressed material takes to reach the outer regions of the wind, where dust can form.

The Model

The significance of dust formation by Wolf-Rayet stars is that the conditions for this seem so unpromising: the WC stars are hot helium cores of initially massive stars which emit most of their radiation in the UV. The harsh radiation fields tend to evaporate any grains within ~80 AU of the star, restricting grain formation to the outer regions of the stellar winds where the densities (assuming spherical winds) are too low by 3 - 4 orders of magnitude to allow the chemical processes on the pathway to the formation of carbon dust in WC stars (C -> C2 -> carbon chains -> monocyclic rings -> PAC rings -> fullerenes) discussed by Cherchneff & Tielens (1995) to occur. Therefore, our observations of epsiodes of dust formation show that conditions in the stars change periodically so as to produce large-scale high-density structures in the outer regions of their flows.

The key observation is that dust-formation episodes by WR140 coincided with periastron passage in its binary orbit. The density enhancements required for dust formation are material compressed in shocks formed where the fast winds of the WC7 and O4-5 stars collide. Compression of the wind by a factor of ~10000 can occur within the shock if the wind cools sufficiently by radiation (Usov 1991). Of course, the WC7 and O4-5 stellar winds in WR140 collide and compress wind material all the time, so we then have to ask: what varies round the orbit so as to trigger dust formation for only a fraction of a period during periastron passage? The answer is the systematic variation in the pre-shock wind density between the stars. This region lies where the momenta of the WC7 and O4-5 winds balance and is much closer to the O4-5 star, whose mass-loss rate is ~ 60 times less than that of the WC7 star. Because the orbit is very eccentric (e=0.84), the separation of the stars and the distance of the interaction region from the WC7 star vary strongly around the orbit. This is especially so around the time of periastron passage: for a very short time, the density of the WC7 stellar wind going into the shock (and being compressed by it) is ~50 times greater than that during most of the orbit (Fig. 1). The consequent `spikes' in the pre-shock density appear to be the clock that triggers the dust condensation.

The processes of compression and cooling in the shocks sufficient to allow dust formation by WR140 have been modelled by Usov (1991). However, dust cannot condense until the compressed wind material has been carried far enough away from the stars so that the grains are not heated to sublimation by the stellar radiation field: a distance ~ 150 AU. Assuming the compressed material moves with the wind terminal velocity ~ 2900 km/s, this introduces a delay of ~ 90 days between the times of maximum pre-shock wind density (at periastron passage) and maximum dust formation — consistent with the observed phase difference (Df ~ 0.03P) between maximum pre-shock density and infrared (K band) maximum (Fig. 1).

The light curve of WR137 suggests that there were several minor outbursts of dust formation in the years after the 1984 event and that there are secondary structures in the wind. Marchenko et al. (1999) have imaged WR137 with NICMOS-2 on the HST in 1997.7 and 1998.4, finding very asymmetric dust formation and movement of dust clumps away from the star, carried by its wind.

In a parallel programme in the Southern hemisphere, Veen et al. (1998) have observed a second dust formation episode for the episodic dust maker WR19, enabling us to set a "period" of 10.1 years for this system. There is a whole class of variable star making dust at intervals of the order of decades and our observations from the UKIRT service observing programme have played a major role in their discovery.

References

Cherchneff, I. & Tielens, A.G.G.M. 1995, in Wolf-Rayet Stars: Binaries, Colliding Winds, Evolution, IAU Symposium 163, eds K.A. van der Hucht, P.M. Williams, (Kluwer, Dordrecht), p. 346

Eichler, D., & Usov, V., 1993, ApJ 402, 271

Marchenko, S.V., Moffat, A.F.J., Grosdidier, Y., 1999, 522, 433

Usov, V.V. 1991, MNRAS, 252, 49

Veen, P.M., van der Hucht, K.A., Williams, P.M., Catchpole, R.M., Duijsens, M.F.J., Glass, I.S., Setia Gunawan, D.Y.A., 1998, A&A, 339 L45

Williams, P.M., van der Hucht, K.A., Pollock, A.M.T., Florkowski, D.R., van der Woerd, H., Wamsteker, W.M. 1990, MNRAS, 243, 662

Williams, P.M., van der Hucht, K.A., Kidger, M.R., Geballe, T.R., Bouchet, P. 1994, MNRAS, 266, 247

A wide-field, high-resolution, k-band mosaic of Abell 851 using UFTI.

Ian Smail¹ & Rob Ivison²

¹Dept. of Physics, University of Durham, U.K.
²University College London, UK.

The most direct method for investigating the formation and evolution of galaxies is to study galaxies at great distances and thus earlier times. The drawback with this technique is the difficulty in identifying the progenitors and decendants of different classes of galaxies at different epochs. Such studies have indicated that the characteristics of galaxies within rich clusters have altered radically over the past 5 billion years. The centres of rich local clusters are dominated by a population of old, bright elliptical galaxies and these are also seen in similar regions of distant clusters. In contrast, blue, actively star-forming galaxies are rarely seen in the centres of rich clusters today - but appear frequently in this environment at higher redshifts. However, to reliably compare the properties of galaxies, both star-forming and passive, in distant and local clusters, and to quantitatively trace these changes we need to select samples of galaxies using a characteristic which is (relatively) insensitive to the recent star formation histories of galaxies, such as their luminosity in the near-infrared. Hence, by comparing the properties of near-infrared selected samples of galaxies in local clusters with similar samples from distant systems we can uncover the physical processes responsible for the evolution of galaxy populations in these environments and, in doing so, obtain profound insights into the rules governing the formation and evolution of galaxies.

FIGURE 1: A "true colour" image of the distant galaxy cluster Abell 851, created from high-resolution, ground-based I- and K-band imaging. The old, bright elliptical galaxies are clearly visible in the cluster centre, while the blue star-forming galaxies are spread more widely and probably represent an infalling population of star-forming galaxies. The subsequent evolution of this population will have a profound effect on the mix of galaxies found within the cluster core at later epochs. The region shown here is smaller than the 5.3' x 6.8' field covered for which complementary I-band imaging is available from Keck-II. Note that the stellar halos and bleeding only effect the Keck data (shown in blue).

FIGURE 2: A false-colour K-band image of a single pointing from the A851 UFTI mosaic. This frame is 92"x92" in size, has an exposure time of 2400s and seeing of ~0.45" (around 3 kpc at the cluster redshift of H_o/q_o = 0.5). Weak spiral arms are visible in the bright galaxy in the centre of the frame and several of the brighter galaxies can be easily morphologically classified as S0 or E.

FIGURE 3: A true colour image (R,I,K; constructed from WFPC2/HST, LRIS/Keck and UFTI/UKIRT) of a 60"x60" region (roughly 400 kpc across) from the A851 UFTI mosaic. The frame shows a compact clump of galaxies on the edge of the cluster which are possibly being accreted onto the central core. Dust is evident in the colours of several of the spiral galaxies in the centre of the field. The reddest of the three central galaxies is also detected in VLA data suggesting that it hosts an obscured starburst. A faint, extended K source visible to the SW of the field centre is undetected at R and I, indicating that it has I-K>6. It is thus an extremely red object .

In semester 99A we used the new UFTI camera on UKIRT to obtain high quality panoramic near-IR imaging of galaxies within the rich cluster Abell 851 (at z=0.41) which is the subject of an intensive observational campaign with HST and ground-based telescopes. The UFTI image (part of which is shown in Fig. 1) comprises a 5x5 mosaic of UFTI fields covering 7.5'x7.5' square (the size of the field covered in the HST WFPC2 mosaic) equivalent to 3 Mpc at the cluster redshift. Each pointing totals 2400s on-sky integration and reaches K~20. The median seeing for the observations was 0.5"; we were able to take full advantage of this with UFTI's 0.09"/pixel sampling (Fig. 2). Indeed, several of the 2400s stacked images have ~0.35" seeing measured from the final co-add, and periods of sub-0.25" seeing were experienced. Recent telescope upgrades mean that we can take better advantage of the periods of excellent seeing experienced on UKIRT.

These exquisite K-band images will be combined with morphological information from the HST R-band mosaic, as well as multi-colour ground-based imaging (Fig.3) and extensive spectroscopic observations. The full data base will be analysed to study the morphologies and properties of K-band-selected popuations of active star-forming, post-starburst and passive galaxies, as a function of radius within the cluster, to search for the mechanisms responsible for rapid evolution of the galaxy populations.

Credits: UFTI; Dr Ian Smail (Durham), Dr Rob Ivison (UCL), Dr Alan Dressler (OCIW) and Dr Bianca Poggianti (Padova). HST; MORPHS group. Keck image; in collaboration with Prof. Len Cowie, Dr Amy Barger (Hawaii).

Simultaneous IR and MM observations of disk ejection events from the black hole candidate GRS 1915+105.

Rob Fender¹ & Guy Pooley²

¹Astronomical Institute `Anton Pannekoek', Univ. of Amsterdam.
²Mullard Radio Astronomy Observatory, Univ. of Cambridge.

The `microquasar' GRS 1915+105 has been the focus of intense observational scrutiny since its identification as the first Galactic source of apparent superluminal motions in radio jets (Mirabel & Rodriguez 1994; see also Fender et al. 1999). The true nature of the source remains mysterious, in part due to heavy interstellar extinction (Av > 20 mag), and in part due to its extreme and unique behaviour in all energy regimes from radio through hard X-rays.

Among its unusual radio properties are strong quasi-periodic oscillations with periods in the range 20-120 min and amplitudes of more than 50% (Pooley & Fender 1997). The oscillations have infrared counterparts (Fender et al. 1999; Eikenberry et al. 1998, Mirabel et al. 1998; Fender & Pooley 1998) and are strongly coupled to dips observed in the X-ray light curve (Pooley & Fender 1997; Eikenberry et al. 1998; Mirabel et al. 1998). These dips have been interpreted as the disappearance of the inner few hundred km of the accretion disc around the black hole (Belloni et al. 1997). The radio — IR oscillation events are interpreted as arising from ejecta which may comprise in whole or part the inner disc material. For the first time we are getting a glimpse of the very rapid coupling between accretion and ejection events in the immediate vicinity of a black hole.

However, the emission mechanism of the oscillations remains uncertain. The extremely flat spectrum inferred from the radio-infrared oscillations is impossible to explain with simple optically-thin synchrotron models, and hard to explain even in terms of (partially) self-absorbed synchrotron models (the spectrum appears to be considerably flatter than even that of `flat-spectrum' AGN). A possible way out of this conundrum was a spectrum which peaked in the (sub)mm and coincidentally reached the same flux level at cm and micron wavelengths. Alternatively, the infrared oscillations may simply be reprocessed X-ray emission, again coincidentally reaching the same amplitude as the radio oscillations (although the detailed infrared : X-ray observations of Eikenberry et al. 1998 show this to be unlikely).

FIGURE 1: Simultaneous JCMT SCUBA 1.3 mm and UKIRT IRCAM3 K-band (2.2 mm) observations of the black hole candidate `microquasar' GRS 1915+105, obtained as target-of-opportunity observations on May 20,1999. The infrared data have been dereddened by A_K = 3.3 mag. Each oscillation is believed to correspond to a discrete ejection of material from the inner accretion disc around the black hole; however the mechanism which can produce such a flat spectrum across the radio-mm-infrared regime remains unclear.

On 1999 May 20, alerted by more than a week of radio oscillations observed with the Ryle Telescope (Cambridge, UK) we triggered a simultaneous override on UKIRT and JCMT to observe GRS 1915+105. We used IRCAM3 continuously in the K-band and SCUBA (almost) continuously with the 1.3 mm filter, in order to obtain the most complete light curves. Figure 1 shows the results (infrared data have been dereddened). We have been lucky enough to catch very large-amplitude oscillations simultaneously at mm and infrared wavelengths. These are the largest-amplitude oscillations ever observed from the source at infrared wavelengths, and possibly the most violent activity ever observed from ANY mm source. There can now be no doubt that the oscillations observed from GRS 1915+105 display an almost perfectly flat spectrum through radio—mm—infrared wavelengths. This puts further strain on synchrotron models. While four of the five oscillation events show an excellent correlation between the mm and infrared, it is clear that the second event (peak in infrared around MJD 50308.594) is not well correlated. We have no simple explanation for this, beyond the implication that the optical depth structure in this ejection was different to that of the other four events, presumably implying a difference in density, temperature and/or physical structure.

In addition, while an infrared—radio delay of at least 7 min has been observed (Mirabel et al. 1998; Fender & Pooley 1998), these data show that there is no detectable delay (<1 min) between the peaks of the infrared and mm emission. As this must reflect the optical depths at different frequencies at different times, it gives us vital clues about the structure and density of the ejecta. These observations were part of a larger campaign which included radio and X-ray observations and will be presented in more detail in a future paper, to be submitted to MNRAS.

Acknowledgements

We are indebted to many people who variously gave up sleep, their own observing time, and helped approve and coordinate these observations. In no particular order these include John Davies, Graeme Watt, Fred Baas, Ian Robson, Iain Coulson, Andy Adamson, Garrett Cotter, Will Grainger, Mark Lacey, Susan Ridgway, Tim Carroll and Thor Wold.

References

Belloni T. et al., 1997, ApJ, 488, L109

Eikenberry S.S. et al., 1998, ApJ, 494, L61

Fender R.P. et al., 1999, MNRAS, 304, 865

Fender R.P. & Pooley G.G., 1998, MNRAS, 300, 573

Mirabel I.F & Rodriguez L.F., 1994, Nature, 371, 46

Mirabel I.F. et al., 1998, A&A, 330, L9

Pooley G.G. & Fender R.P., 1997, MNRAS, 292, 925

A new solar neighbourhood population of Methane Dwarfs.

Sandy Leggett

UKIRT/Joint Astronomy Centre, Hilo, Hawaii, USA.

Brown dwarfs are objects with a mass below the minimum mass for stable hydrogen burning; they are less massive than about 0.08 solar masses and after a very brief period (<0.1 Gyr) of deuterium burning, they will fade. Their luminosity decreases by a factor of 100 in about 1 Gyr (e.g. Burrows et al 1997, ApJ, 491, 856). Astronomers have been searching for these faint objects for decades, in particular as an explanation of missing mass.

In 1995, just when the community was beginning to think that somehow the process of star formation did not allow brown dwarfs to be formed, the first non-controversial brown dwarf was discovered during a search for low-mass companions to young M-dwarfs (Nakajima et al 1995, Nature, 378, 463). The definitive infrared spectra of this object, Gliese 229B, were obtained by Tom Geballe and collaborators using CGS4 on UKIRT (Geballe et al 1996, ApJ, 467, L101). The spectra look planet-like, with strong absorption bands of methane and water (steam). The presence of methane implies an effective temperature cooler than around 1200K (e.g. Lodders 1999, ApJ, 519, 793), and this necessarily implies a sub-stellar mass. Recent data suggest Gliese 229B is aged 0.5-1.5 Gyr, has Teff~900K and mass~0.02-0.04 solar masses (or about 20-40 Jupiter masses), e.g. Leggett et al 1999, ApJ 517, L139.

Since 1995 the community have been searching very hard for another example of a methane-dwarf. However although objects with T(eff) ~ 2000K have been found, 1000K objects were elusive - until a few months ago. On April 21st 1999 astronomers from the Sloan Digital Sky Survey contacted myself and Tom Geballe (Gemini) and asked us to get infrared spectra of a candidate brown dwarf. The Sloan survey uses narrowband filters sampling 3500A to 9100A to map an area around the North Galactic Pole. This object was the reddest confirmed object in the Sloan preliminary survey area, and an optical spectrum had shown no flux blueward of 0.8mm, as well as water and CS absorption features. Initially I obtained an H-band spectrum with CGS4 and it was clear even in the raw image that we indeed had another methane dwarf - the image showed a narrow region of flux near the centre of the H-band and nothing else, the remainder being absorbed by water and methane. Despite being 3am at 14000ft this was a thrilling moment! Further spectroscopy and photometry obtained by myself and Tom showed this object, SDSS 1624+00, to be extremely similar to Gliese 229B (Strauss et al 1999, ApJL, 522, or astro-ph /9905391).

A few weeks later the Sloan group contacted me with a second candidate,and CGS4 spectra showed this to be another methane dwarf (Tsvetanov et al in preparation). Furthermore the then three known examples of this type of object had amazingly similar energy distributions, as shown in Figure 1. The Sloan objects do differ from Gliese 229B as they are isolated field objects. Although we do not yet know the distance to the Sloan dwarfs, and so cannot determine luminosity and constrain age, the presence of methane in the spectra implies an effective temperature of less than 1200K, and the presence of cesium and potassium implies an effective temperature greater than about 700K (below this the alkali elements become chlorides, Lodders 1999). Assuming an effective temperature similar to that for Gliese 229B, the Sloan dwarfs will have a mass between 15-60 Jupiters for an age 0.3-6 Gyr.

In June and July three papers became available via astro-ph announcing methane dwarf discoveries - the Sloan paper (9905391), as well as announcements by the 2MASS team (9907019) of four other objects, and by ESO (9907028) of another, fainter, example. The total number now known is eight - Gliese 229B, and seven isolated field objects. We can think of this as a population! There is some debate as to what letter of the alphabet to give methane dwarfs. The objects cooler than M with T_eff~2000K are being called L-dwarfs, and it has been proposed that the ~1000K objects are called either T-dwarfs (Kirkpatrick et al 1999, ApJ, 519, 802) or H-dwarfs (Martin, in preparation). Perhaps the best mnemonic ("Oh be a Fine Guy...") will win the day!