Department of Physics and Astronomy, University College London, U.K.

As very massive, luminous stars evolve from the Main Sequence to become hydrogen-depleted Wolf Rayet stars they shed large quanties of matter over a very short period of time. The period of enhanced mass loss is associated with the Luminous Blue Variable (LBV) phase, which is characterised by the presence of spectroscopic and photometric variability on timescales from months to years. Very few LBVs are known within the Milky Way. Consequently, it is believed that the LBV phase is very short, possibly only of the order of a few times 10^4 yrs. However, the lack of an accurate galactic census prevents a refinement of this estimate and consequently much effort has been applied to the identification of new LBV candidates.

One such candidate is the highly reddened early type star AFGL 2298 [1], which is associated with a massive (~10 M-solar) dusty nebula [2]. AFGL 2298 is photometrically variable (Fig. 1 [1,2,3]), so we obtained two CGS4 K band spectra of AFGL 2298, separated by about a year, to investigate the physical properties of the star; these are presented in Fig.2.

FIGURE 1: IR lightcurve for AFG~2298 between 1999-2002. Data from [1] (green), [2] (red) and [3] (blue); where errorbars are not visible they are smaller than the symbol size. The two arrows in the bottom panel mark the dates of the spectroscopic observations presented in Fig. 2.	FIGURE 2: K band Spectra of AFGL 2298 from 2001-2002 (blue). Note the dramatic change in line strengths between the 2 epochs. Best fit synthetic spectra, produced with the NLTE stellar atmosphere code of [4] are overplotted in red.

Clearly, we have witnessed significant spectral variability between the 2 observations. Utilising the NLTE model atmosphere code CMFGEN [4] we have been able to determine the physical properties of the star at each epoch. As expected for an LBV, AFGL 2298 is a highly luminous 10^6.1 L-solar mid-B supergiant; indeed it appears to be one of the most luminous stars yet discovered! The variability in the spectra was found to be due to a combination of changes in both temperature (T_eff = 12.5 -> 15 K) and mass loss rate (0.5 -> 1.2 x 10^-4 M-solar/yr). Since the bolometric magnitude of AFGL 2298 remained constant between the two observations, the increase in temperature corresponds to a change in radius from ~205 -> ~166 R-solar in only one year, implying that the stellar surface contracted at an average rate of -2.3 km/s during this time!

Given the spectacular variation in physical properties over the course of a single year, further monitoring of this star is clearly warranted - a task clearly suited to UKIRT and service observations. Indeed, despite the mass loss rate of AFGL 2298 being amongst the highest ever measured for ANY hot star (rivalled only by other monsters such as Eta Carinae) the huge nebular mass implies that it must have been at least an order of magnitude higher in the past. We may only speculate as to whether the present increase will be maintained over the coming months and years. With a combination of temperature and luminosity that places it at the very edge of stability at the peak of the HR diagram, AFGL 2298 promises to provide unique insights into the lives - and deaths - of the most massive stars known.

References:

[1] Ueta et al., 2001, ApJ, 548, 1020
[2] Pasquali & Comeron, 2002, A&A, 382, 1005
[3] Clark et al., 2003, A&A, submitted
[4] Hillier & Miller, 1998, ApJ, 496, 407

Joint Astronomy Centre, Hilo, Hawai
Institute for Astronomy, University of Edinburgh, U.K.
Physical Research Laboratory, Ahmedabad, India

Wolf-Rayet (WR) stars represent the late stages in the evolution of massive stars, undergoing large scale mass loss ~ 10^-5 M_sun year^-1 via accelerated stellar winds with terminal velocities ranging from 750--5000 km/s. They are characterised by broad emission line spectra, originating in their fast stellar winds. Based on the emission lines seen, they are classified as WN and WC types, WC types showing lines of He, C and O in their spectra.

When the WR star is in a binary system with a massive O-type star, which also has a fast stellar wind (usually less dense than that of the WR star), the winds of the two stars collide. The stellar winds are compressed by shocks in the collision region and, in some cases, carbon dust forms. This process occurs only in a few systems (only WC stars, whose winds are carbon-rich) and appears to be very sensitive to the separation of the stars. So, in such a binary system, during epochs close to periastron passage, we see a rise in the IR flux due to colliding-wind dust formation. Such binaries are called "episodic colliding-wind WR dust making binaries", of which WR140 is the prototype (Williams 1995).

FIGURE 1: Sketch of the colliding-wind region between the WC and O-type stars in WR140. The lightly shaded area represents compressed WC stellar wind material flowing along the contact discontinuity between the stars. At this phase, most of the material (the fraction depending on the unknown orbital inclination) is flowing towards the observer.	FIGURE 2: Relative orbit of the WR140 binary showing the shock cones at various orbital phases. The CGS4 high resolution spectra of the HeI 1.083 micron line on epochs around the periastron passage are superposed.

WR140 (HD 193793) comprises WC7 and O4-5 stars in a highly eccentric orbit (e=0.84) with a period of 7.94 yrs (Williams et al. 1990). The star shows periodic brightening in the IR photometric bands, which is interpreted as being due to dust formation during periastron passage (Williams et al.). The star also shows variable X-ray and non-thermal radio emission arising from dissipation of the kinetic energy of the winds in the collision zone. Dust formation occurs for only a tiny fraction (3-4%) of the period around periastron passage. The latest periastron passage of WR140 (February 2001) gave us a unique opportunity to study the colliding-wind phenomenon in WR binaries in detail. This star has long been observed by UKIRT (see the UKIRT Newsletter #5, Sept. 1999), and we have since conducted spectroscopic monitoring of it in the near IR JHK bands using UKIRT (with CGS4) and the Mt. Abu 1.2m IR telescope in India.

In WR stars, the emission lines of different ionization species form at different regions of the wind. The HeI line at 1.083 micron forms in the asymptotic region of the WR stellar wind, giving it a characteristic flat-topped profile and thus making it a good diagnostic of excess emission attributable to other processes in the system. During epochs close to periastron passage, when changes in the line profiles might be expected, the spectrum of WR140 was observed with UKIRT+CGS4 at a resolution of ~5000 around the HeI line. The experiment was very successful: the line profile showed spectacular variations during this series of observations. A narrow sub-peak appeared on the flat top of the profile and it moved as the relative positions of the stars moved in their orbit. The sub-peak is believed to form in the dense wind material flowing along the wind interaction region. The sub-peak gets Doppler shifted when the shock cone moves around during periastron passage -- blue shifted before periastron passage, when the O star is closer to the observer, and red-shifted after the periastron passage, when the O star is `behind' the WR star. Fig. 1 shows the configuration in 2000 October, when the system was approaching periastron passage. The wind-collision region forms a `cone' around the O-type star and the compressed WC wind flows along this surface. At this phase, much of the dense flow is towards the observer, accounting for the blue-shifted peak. The appearance of this cooling flow and the fact that the periastron passage occurred close to what was predicted from previous studies (2001.14) points to an important fact: contrary to what was expected theoretically (e.g. Zhekov & Skinner 2000), the colliding-wind region of WR140 is not totally adiabatic during periastron passage, so the theoretical 1/r dependence of X-ray flux does not hold.

Fig. 2 shows a schematic diagram of the relative orbit of WR140, with the WR star at the focus, showing the evolution of the shock cone with varying positions of the component stars. CGS4 high-resolution spectra at the wavelength of the 1.083 micron HeI WR emission line taken around the periastron passage are superposed on it.

FIGURE 3: Observed J spectra of WR140. Epoch of observation and orbital phase are shown. The bottom spectrum is from the Mt. Abu telescope before the periastron passage and the rest are CGS4 spectra after the periastron passage. Note the sudden change in continuum after the periastron passage, when the dust formed.		FIGURE 4: Image of WR140, 19 months after the periastron passage, taken using UKIRT and UIST with the 0.06"/pix camera and the 3.99 micron narrow band filter. The peak thermal emission from the dust is shifted to mid-IR wavelengths at this time and it is clearly resolved at the high resolution of UIST on UKIRT. The image shown is deconvolved using the PSF derived from a nearby PSF star.

At later epochs, CGS4 observations were made at lower resolution, using the 40 l/mm grating in the JHK bands. Fig. 3 shows the evolution of the J-band spectra. A sudden change in the continuum and line equivalent widths are evident between 2001 January and March, when dust formed subsequent to periastron passage. Similar effects are seen in the H and K-band spectra (not shown here).

Presently, the dust formed during periastron passage has expanded and cooled and is suitable for imaging using UKIRT at thermal wavelengths. We observed WR140 at 3.99 microns using UIST on 2002 November 20. The expanding dust is well resolved and is seen as a clumpy arc. The UIST image, deconvolved using the PSF derived from a nearby star, is shown in Fig. 4. Prominent clumps resolved by Monnier et al. (2002) from the Keck interferometric images at 2.2 micron are seen to have expanded and cooled in our 3.99 micron images. We continue to study the evolution of the dust shell of WR140 using UKIRT and UIST, in service mode, and hope to present our results in a future papers.

References

Monnier, J. D., Tuthill, P. G., Danchi, W. D. 2002, ApJ, 567, L137
Williams, P. M. 1995, Proc. IAU Symp. No. 163, 335
Williams, P. M., van der Hucht, K. A., Pollock, A. M. T., et. al. 1990, MNRAS, 243, 662
Zhekov, S. A., Skinner, S. L. 2000, ApJ, 538, 808

¹Joint Astronomy Centre, Hilo, Hawai
²University of Helsimki, Finland
³Rensselaer Polytechnic Institute

Introduction

The extinction and reddening effects of the diffuse interstellar medium (DISM) play a major role in constraining the properties of modern dust models. Carbonaceous materials are a major component of many such models. The physical form and relative contributions of these components to the dust have recently been constrained by tightening of the cosmic abundance constraints (see Snow & Witt 1996 for a review), which appears to favour some forms of these materials over others (e.g. Mathis 1996; Li & Greenberg 1997). Of all the elements concerned, carbon imposes some of the strongest restrictions on current dust models, and yet the physical form of solid-phase and macromolecular carbon in the ISM remains a subject of debate.

FIGURE 1: DSS R, 2Mass J and 2Mass K images of StRS136 (5 arcminute field). This star has the photometric colours of an M star but is in fact a blue supergiant with spectral type between B8 and A9. The Galactic context is given by the IRAS 12-micron image (7 degree field).	FIGURE 2: Optical (INT) Spectra of StRS344 and StRS217

Solid-phase aliphatic hydrocarbons, Gas-phase aromatic hydrocarbons, Gas-phase aliphatic hydrocarbon chains, embedded aromatic molecules or their cations, fullerenes & related species have all been suggested as possible sinks for carbon atoms, some have been definitely detected and others remain debatable. We have worked to find a link between one of the better-understood tracers, the aliphatic C-H stretch vibration in hydrocarbon solids, and the diffuse interstellar bands (variously linked with any of the above list of carriers). The abundance requirements of carbon in the solid and gas phases appear to be broadly similar. Therefore, an observable anti-correlation of the DIBs with the solid-phase hydrocarbon bending and stretching features might be expected if UIB and DIB carriers may be liberated from grain surfaces into the gas phase by interstellar shocks (Duley & Jones 1990, Duley & Williams 1988, Galazutdinov et al. 1998). To measure such an effect is observationally challenging, due to the difficulty of observing visual absorption bands in objects that are reddened sufficiently to produce measurable 3.4 micron absorption. We have approached it using the catalogue of heavily-extinguished early-type stars derived by the authors (Rawlings, Adamson & Whittet 2000) from the Stephenson (1992) catalogue of red objects in the Galactic plane.

FIGURE 3: Removal of residual atmospheric cancellation problems. Note that the quality of cancellation in the 3.4 micron region is much better than it is shortward of 3.3 microns.

Survey Objects

The Stephenson objective-prism survey contained more than 400 sources claimed to exhibit extremely red colours but no photospheric TiO bands, suggestive of heav ily reddened early-type stars. Subsequent observations of small subsets of the c atalogue (Creese et al. 1995, Imanishi et al. 1996, Goto et al. 1997) suggested that a majority of the sources were in fact K- or M-type stars with only moderat e reddening. The first large-scale spectroscopic study of the S92 catalogue (Raw lings, Adamson & Whittet 2000) included 61006900Å spectroscopy of 191 stars, and confirmed that 87 per cent of that subset are of spectral type KM, (and a furthe r 3 per cent are S stars). Only 15 sources were found to be of spectral type G o r earlier, with extinctions between A(V) of 616. Figure 1 shows DSS R and 2Mass J and K images of StRS136, one of the objects in a line of sight toward the gala ctic centre region. While the Stephenson "reddened stars" survey as a whole has turned out to be a disappointment, our exploitation of it has, given the paucity of such high-extinction sightlines outwith the Galactic Centre, in fact more th an doubled the number of objects available for dust studies at high A(V). Figure 2 shows the visible-wave spectrum of StRS354, and compares it with that of the heavily-reddened supergiant Cyg OB2 no. 12; clearly the latter star, which has f or many years been the classic high-extinction early-type object (A(V)>10), has good company in the Rawlings et al. subset of the Stephenson survey.

Observations at UKIRT

UKIRT observations were taken with CGS4, working in the 3.2-4.0 micron region which contains the aliphatic C-H stretch (as well as a lot of telluric water and methane absorption - roll on flexible scheduling). Once one has dealt with the terrestrial absorption (a process described in detail in Rawlings et al. 2003, and shown in Figure 3) the hydrocarbon absorption is clearly present in all but one of the objects studied.

Some of the resulting spectra are shown in Figure 4 - in total, 12 stars were observed with UKIRT and 11 show the hydrocarbon band (the exception, StRS432, was not observed for long enough to detect this weak feature).

Measuring optical depths from these spectra provides solid hydrocarbon measurements for eleven stars. The combination of these data with our existing measurements of the Diffuse Interstellar Bands in these same lines of sight provides the unique comparison between gas-phase and solid-phase features which this sample of objects makes possible for the first time. It also enables a variety of tests for galactic variations (since the sample is in two broad groups - those in the Cygnus region and those more toward the Galactic Centre). Further, estimates of the visual extinction taken from the visual spectroscopy allow new determination of the variation of the C-H stretch with visual extinction. The results are discussed in full in a paper in press to MNRAS (Rawlings, Adamson & Whittet 2003; preprint on Astro-PH/0302060). As a taster of the content of this analysis, we show in Figure 5 the variation of C-H stretch absorption as a function of visual extinction: the results reinforce the known difference between the central cluster in the Galactic Centre and essentially every other sightline ever studied. Secondly, from these eleven sightlines, there is no evidence for significant trading of carriers from the solid phase to the gas phase.

FIGURE 4: A selection of optical depth spectra. Note the presence of Humphreys series H lines in StRS164.	FIGURE 5: Variation of hydrocarbon optical depth with visual extinction between 4m and 35m. The new data (best fit: solid line) fall somewhat below previous determinations (dotted line). The three points to the right are from the Galactic-centre cluster.

Further work It is clearly important in interpreting these results to quantify the contribution of dense-cloud material to the extinction in these lines of sight. We have therefore carried out observations of the short end of the 3-micron window, in a search for ice absorption. To increase the coverage of the optical DIBs and thus address the relationship between DIB families, galactic environment and the CH stretch is being done with shorter-wavelength spectroscopy obtained at the INT.

References
Creese et al. 1995 AJ, 110, 268
Duley W. W. & Jones, A. P., 1990, ApJ, 351, L49
Duley W. W. & Williams D. A., 1988, MNRAS, 230, 1P
Galazutdinov G. A. et al. 1998, MNRAS, 295, 437
Goto et al. 1997 PASJ, 49, 485
Imanishi et al. 1996 AJ, 112, 235
Li A. & Greenberg J. M., 1997, A&A, 323, 566
Mathis J. S., 1996, ApJ, 472, 643
Rawlings M. G., Adamson A. J., Whittet D. C. B., 2000, ApJS, 131, 531
Rawlings, Adamson & Whittet 2003 in press MNRAS.
Snow T. P., Witt A. N., 1996, ApJ, 468, L65
Stephenson,C.B. 1992 AJ, 103, 263

Acknowledgments
This work was supported by PPARC, the Academy of Finland, University of Central Lancashire, PPARC and NASA grants NAG5-7598 and NAG5-7884. Figure 1 was produced using data from the NASA Skyview facility. We gratefully acknowledge NASA and the contributors of the data (DSS, 2Mass).

¹Centro de Astrofisica da Universidade do Porto, Portugal.
²Joint Astronomy Center, Hawaii, USA.

The majority of (if not all) stars are born in clusters or associations. This must therefore be the predominant mode of star formation that yields the Galactic-field population. To understand how such aggregates form and evolve in a single stellar nursery remains an important problem in astrophysics. Recently, UFTI has been used to examine a number of young clusters. The resulting images have provided some intreguing observational data to help us better understand the birth of star clusters.

In the last few years there has been increasing progress in observational and theoretical studies of cluster formation (e.g. Clarke, Bonnell, Hillenbrand, 2000, PPIV conf. proc., p.151; Klessen & Burkert, 2000, ApJS, 128, 287; 2001, ApJ, 549, 386). Recent simulations of the fragmentation of magnetically subcritical clouds have resulted in predictions of cavities, rather than density enhancements, at the centres of cluster forming cores (Li & Nakamura, 2002, ApJ, 578, 256). These are thought to be caused by "magnetic tension" which helps to prevent a central density singularity.

FIGURE 1: (left) a "true-colour JHK UFTI image of the proto-cluster IRAS 22134+5834 (Kumar, Bachiller & Davis 2002). The excellent image quality at UKIRT and the fine pixel scale of UFTI allowed us to resolve the fine structure of the proto cluster. Note the faint red stars that delineate the boundary of the dark cavity at near-infrared wavelengths.	FIGURE 1: (right) a cartoon showing the predicted surface density distribution of matter during the fragmentation of a flattened SF cloud (from Li & Nakamura 2002).

Most of the star clusters that we know of are fairly circularly symmetric, with the centers of the clusters having more stars and therefore denser populations than the edges. For example, in the Orion Nebula Cluster, the most massive stars (the trapezium stars) are situated at the heart of the cluster. Yet the theoretical studies mentioned above predict that, while the clusters are being formed, they should have a very different spatial distribution of stars, i.e. cavities should appear at the centres of the clusters, with the stars distributed in rings. As the clusters grow older, the heavier stars sink to the center of the cluster and the lighter stars move towards the periphery, a process known as "dynamical relaxation". This relaxation process leads to rapid re-adjustment of the stars in each cluster, even before the regions shed their molecular cocoon and becomes visible at near-infrared and optical wavelengths.

A key part of observational studies of cluster formation therefore involves catching these embedded clusters before they significantly rearrange themselves. In June 2002 using UFTI on UKIRT we observed a carefully selected sample of cluster forming cores. The sample included several young star clusters yet to be dynamically relaxed. Figure 1 shows a three color image of one such (proto)cluster, obtained with UFTI in the JHK bands. The adjacent cartoon is a numerical simulation from Li & Nakamura (2002, ApJ, 578, 256) showing the surface density distribution of matter (perturbation mode m=5) predicted by their models during the fragmentation of a flattenned molecular cloud. The JHK color image shows striking similarity to the cartoon, clearly displaying the central cavity and demonstrating the flattenned nature of the cluster. If the star cluster were to be spherically symmetric rather than flattenned, the central dark cavity would have been covered by the outlying stars in this image.

Although these new images, published recently in ApJ, (Kumar, Bachiller & Davis 2002, ApJ, 576, 313), offer considerable support to current cluster formation models, additional observations of other young, pre-relaxation clusters are clearly needed if we are to fully understand the early stages of cluster formation.

¹Herzberg Institute of Astrophysics, NRC, Victoria, Canada
²Institute for Astronomy, Edinburgh, U.K.
³Dept. Astrophysics, Oxford, U.K.

Shortly after commissioning, UIST was used to obtain simultaneous H and K band spectra of the most distant quasar known, SDSS J1148+5251. The spectrum shows MgII redshifted from 0.28 microns right out to the K band. A precise redshift of z = 6.41 +/- 0.01 was also measured (a value more accurate than could be obtained from the optical spectrum). The Lyman alpha lines of the highest redshift quasars have lower equivalent widths than quasars at lower redshifts. We find the equivalent width of the MgII emission line is similar to that of lower redshift quasars, suggesting that the UV continuum is not dominated by a beamed component and that the low Lyman alpha equivalent widths are due to absorption of Lyman alpha photons.

FIGURE 1: Upper panel: UIST HK-Grism spectrum of the z=6.41 quasar SDSS J1148+5251 (red). The broad MgII emission line is marked. The blue curve shows the best fit model comprising a power law continuum (spectral index, alpha=0.2), a broadened FeII template and a broad MgII line. Lower panel: The residual spectrum after subtraction of the power law continuum and FeII template (red). The best fit to the broad MgII doublet (FWHM=6000 km/s) is also shown here (blue). The cyan region shows the 1-sigma noise.

The width of the MgII emission line can also be used to estimate the black hole masses of high redshift quasars by comparison with detailed observations of low redshift quasars. With a MgII FWHM of 6000 km/s, we estimate the black hole mass of 3 billion M-solar. The very high luminosity of the quasar shows that it is accreting at the maximal allowable rate for a black hole of this size (the Eddington limit). This result agrees with theoretical predictions that it is possible to create such large black holes so early in the Universe, but they will be rare and will be surrounded by a reservoir of fuel that allows them to accrete right at the Eddington limit. These z = 6 quasars pinpoint the first massive structures to have formed.