Seeing the Light Through the Dark J

No. 103 – March 2001 Seeing the Light Through the Dark J. ALVES1, C. LADA2 and E. LADA3

1European Southern Observatory, Garching, Germany 2Harvard-Smithsonian Center for Astrophysics, Cambridge MA, USA; 3University of Florida, Gainsville FL, USA

1. Introduction: The Search for years of study, little is understood about posed of molecular hydrogen, which is the Initial Conditions for Star the internal structure of these clouds virtually inaccessible to direct observa- Formation and consequently the initial conditions tion. Because of its symmetric struc- that give rise to star and planet forma- ture, the hydrogen molecule possesses Stars and planets form within dark tion. This is largely due to the fact that molecular clouds. However, despite 30 molecular clouds are primarily com- (Continued on page 15)

Figure 1: Visible and near-infrared images of Barnard 68. The images are a B, V, and I-band composite (left) and a B, I, Ks-band composite (right). At visual wavelengths the cloud is completely opaque owing to extinction of background starlight caused by small (~ 0.1µm) interstel - lar dust particles that permeate the cloud. The red stars detected at 2 µm through the visually opaque regions of the cloud (right) are the stars that will provide direct measurements of dust extinction through the cloud.

1 T E L E S C O P E S A N D I N S T R U M E N TAT I O N La Silla Telescope Status, a Great Achievement on Image Quality Performances A. GILLIOTTE, ESO

Nowadays almost all La Silla tele- cell was designed for such a process, by the 3.6-m was found insufficient be- scopes deliver very good image quality, then on the 3.6-m and is soon to be car- cause the improvements in the per- routinely achieving sub-arcsec images. ried out on the 2.2-m as well. formance and dynamic range of the de- In some cases, the theoretically pre- The optimal frequency of the CO2 tectors showed more accurately the im- dicted performances of the telescope is cleaning, resulting in clean mirrors age quality achieved by the telescope. matched, or the limitations are at least without too much time spent on the op- With the availability of wavefront sen- understood. eration, was found to be one week. The sors – the first one available was a The image quality, however, is not NTT was the test bench for the process Shack-Hartmann type called Antares – only a function of the telescope optical with the mirror surface maintaining the optical quality of the telescope was set-up, it is also highly dependent on good reflectivity for a three-year period measured more often, and more pre- the cleanliness of the optical surfaces. with the regular CO2 cleaning and four cise aberration information was ob- “in-situ” washings (Fig.1). tained. On several occasions, large Main Mirror Maintenance Aphotograph of the washing operation aberrations were identified. The spher- Five years ago, systematic cleaning at the NTTis also presented in Figure 2. ical aberration, often mistaken for see- of the primary mirrors using CO2 was All the above-mentioned telescopes ing quality, appeared clearly and only implemented on the major telescopes: were realuminised during the year 2000 the decentring coma was adjusted after the NTT, 3.6-m, 2.2-m and Danish 1.54- and the reflectivity is still above 88% at measurements performed only on m. Main mirror cleaning is now carried 670 nm with low roughness. The 2.2-m zenith. It was found that almost all third- out by the telescope team as part of the main mirror suffered some water con- order aberrations were affecting the tel- routine maintenance operation. T h e tamination during the bad winter period escope. The most critical issue con- only restriction with this process is that and a first washing is scheduled for the cerned the instability of the aberrations the humidity cannot be too high. near future. Recently, the 3.6-m main at different telescope orientations as Indeed, high humidity causes the con- mirror also suffered water contamina- well as with changes in temperature. A densed water surrounding each cold tion from cooling liquid, and the wash- dedicated study with eventual periods CO2 particle to stick the dust to the mir- ing is already scheduled end of March. of closed telescope was really compul- ror surface, resulting in an accumula- The status of the mirror is verified on sory. In 1994, a complete study of the tion of dust on the mirror. a monthly basis with a measure of the 3.6-m with a contracted optical engi- A mirror water washing operation, R% and roughness. A database of the neer on La Silla was initiated to deter- performed by the opticians, was also mirror status for the main telescopes mine the status of the telescope and to implemented about three years ago. will soon be available on the La Silla find a way of improving it. The best period to wash the mirrors is web site. During this process, we gathered con- the end of the summer, when the high siderable experience in solving problems humidity is decreasing and hence the Telescope Image Quality with the telescope, but also a deeper frequency of the CO2 cleaning was dis- About nine years ago, the awareness knowledge of the limitations of the tele- rupted. The mirror washing was first im- of the observer towards the image qual- scope was reached. The 2.2-m, Danish plemented on the NTT, whose mirror ity increased and the quality delivered 1.54-m and the NTT have all benefited from this exercise. Of course, the active optics concept, greatly demonstrated with the NTT, allowed a better understanding of the local thermal contribution. The contributions of dome and mirror seeing were identified and a solution to decrease both has been found. Thanks to all the La Silla staff for their ongoing interest and success in improving the image quality at the telescopes on site. The NTT Status This telescope delivers the best image quality on La Silla with the active optics system fulfilling all expectations. Initially, the limiting factor for this telescope was found to be the image quality of the La Silla site. Now, after a long period of operation, more image quality limitations have appeared. To monitor the image quality, the NTT has a distinct advantage over the other tele- Figure 1: The variation of the reflectivity (R%) and roughness of the NTT main mirror during scopes as repeated image analysis is the period from August 1996 to August 2000. made during the observations. All re-

2 en to reach the same operational mode as used on the VLT. A complete maintenance of the astatic levers is underway to recalibrate and readjust each lever. For perfect imaging quality at the zenith, each lever must be adjusted to the medium range of the counterweight move, and the spring compensation tuned to compensate the spherical aberration constant term. This time-consuming operation will be performed during 2001. The possible upgrade of the active optics will also include a standardisa- tion of the present software to comply with the latest version at the VLT.

The 3.6-m Status

Excellent work at the 3.6-m has re- sulted in improved image quality due to Figure 2: The NTT main mirror being rinsed with distilled water. The last washing step before the minimisation of the spherical aber- drying the mirror surface. ration, triangular and astigmatism contributions, even when the telescope is sults are logged in a database, and sta- – Install a Nasmyth shutter to close inclined. In addition, the dome and mir- tistics can be performed with respect to the optical path not in operation. ror seeing contributions are now almost external meteorological conditions or The new baffle and fans have been negligible. The main limitations affect- other parameters. The image quality installed and the new results already ing image quality are the coma instabil- generally becomes poorer when the show a substantial improvement in the ity when moving the telescope far from wind speed is below 5 m/s or if the tem- image quality (cf. Fig. 3b). the zenith. Astigmatism is still slightly perature is dropping rapidly as demon- The air-conditioning of the instrument high, with a contribution coming from strated by Figure 3a. rooms was slightly improved but an op- the axial fixed points. A new design of Several recommendations were timal solution could not be implemented the axial fixed point contact on the mir- made to the NTT team to improve the because of high cost. ror back side is compulsory to reduce a image quality: The timing of a complete image lateral stress applied on the mirror. – Install high flux fans to produce ar- analysis and mirror correction cycle is A new support for the M2 unit is cur- tificial wind to reduce mirror seeing. unfortunately long in comparison with rently under study. It will mainly be a – Install a new M2 baffle with open the VLT active optics system. On some copy of the present NTT M2 adjusting shade to avoid warm bubble forma- occasions, more frequently than be- support. The improvement to image tion. fore, the active optics is not converging quality will be significant, with the coma – Improve the air conditioning in the properly. Hence the telescope does not variation kept under control at a low instrument room to avoid a positive gra- always operate in closed-loop correc- value. We are still awaiting the green light dient. tion. Some efforts should be undertak- to build and implement this new unit.

Figure 3a represents the “fitting” residual variations (non-modelised Figure 3b corresponds to the same layout as Figure 3a, after instal - 7 first aberrations) of the image quality with NTT dome position rela - lation of the new baffle and fans. tive to the wind direction as well as the wind speed. Colour code of wind speed is at top of figure in m/s.

3 Figure 4 shows that coma terms are almost stable and be- the image quality is, low 0.2 arcsec. on several The telescope focus has been shift- occasions, better ed to compensate the high constant than that measured term of the spherical aberration. How- by the Dimm moni - tor. Local site ever, the variability of this aberration effects and the mir - was not confirmed at the time when the ror fans are most amount by which to shift the focus was likely the reason for decided. Although the right correction the better seeing may not have been applied, observers measured at are not reporting poor image quality. EFOSC2 on these The main limitation of image quality at occasions. Image the Danish 1.54-m is still the instrument quality can be as detector sampling. Even after the re- good as 0.6 arcsec and ranges up to cent detector changes (October 2000), 1.5 arcsec, with the the sampling of 0.81 arcsec (2 pixels of coma variation be - 0.015 mm with a scale of 27 arcsec/ ing the main reason mm) is often reached when the night for the measured seeing conditions are around or below spread in EFOSC2 0.6 arcsec. (The previous CCD suf- seeing quality. fered from diffusion transfer increasing in fact the pixel size.)

The present status of the image qual- Observers often report the image ity at the 3.6-m, despite the two re- quality as very good, and sub-arcsec Pending Issues maining limitations, is already within the images are often obtained when the objectives fixed at the beginning of the outside seeing is good. The limiting res- A further step on improving the image study. We have achieved the goal of 0.9 olution of the Wide Field Imager pixel quality of the telescopes will be consid- arcsec within 60 deg Zenithal Distance size is often reached when the seeing ered this year. A large part of the stray and image quality as good as 0.6 arc- is below 0.5 arcsec and the mirror cold- light level affecting the depth of the im- sec has been measured with EFOSC2. er than the outside air. age, is related to the telescope baffling The EFOSC2 sampling limit of two pix- Image quality limitations still exist but quality as well as the cleanliness of the els is now often attained. are more related to thermal contribu- optical surface. At least three cases For the last few months, the EFOSC2 tions rather than opto-mechanical fea- must be considered, the first case is image quality is logged together with tures. when there is a high density of bright the external seeing. Figure 4 speaks for stars in the field, the second when there itself about the present quality of im- The Danish 1.54-m Status are very bright objects close to the tar- ages. The EFOSC2 image quality is al- get and the last case when there is a ready acknowledged by the ESO com- The ongoing problem of the spherical bright star almost alone in the field. The munity. aberration varying with the temperature improvement of the telescope baffling The Infrared mode of the telescope has been studied again. Experience will reduce at least the second case. with the F/35 chopping M2 mirror has gained on the 3.6-m, NTT and 2.2-m The others depend primarily on the light also been improved. The image quality presented a possible explanation of this diffusion by the optical surfaces. Of obtained at 10 and 20 microns with the challenging feature. Again, the opto- course, the mirror polishing quality im- new TIMMI2 is diffraction limited. mechanical mirror support seems to be proved dramatically between the con- guilty and the behaviour of the axial struction of the 3.6-m and the VLT, but The 2.2-m Status fixed points with temperature variations this benefit could be easily lost if the is the most probable explanation. A cleanliness of the mirror surface is not During 2000, the telescope has been bending effect at the mirror edge, ensured. greatly improved with the large de- where the spherical aberration is more The baffling status of the 2.2-m, the crease in the contribution from astig- sensitive, could produce the 0.3 arcsec NTT and the 3.6-m telescopes must be matism. A separate article describes variable term of this aberration. The de- verified and recommendations will be these activities. focus, astigmatism, quadratic and issued.

Image Quality Improvement of the 2.2-m A. GILLIOTTE, ESO

Historical Overview imaging instrument called EFOSC2 Hartmann called Antares. The optical was installed at the telescope. Ob- quality of the EFOSC instrument was On the very first period of operation servers soon began seeing variable im- strongly dependent on the precision of the 2.2-m telescope, a direct CCD age elongations across the full field, with which focus had been achieved, camera was offered to the community. which were later identified as coming and subsequent variations with temper- With a pixel size of 0.35 arcsec and a from the instrument and not the tele- ature. The EFOSC camera focus did small field of 3 arcmin, the telescope scope. Meanwhile, the optical quality of not include temperature compensation image quality was never reported as the telescope was measured to be as as was the case with EFOSC1 on the being bad or showing asymmetric, el- good as 0.35 arcsec d80% close to 3.6-m. The focus degradation intro- liptical images. In the mid-1990s, a new zenith, using our portable Shack- duced field curvature and increasing

4 astigmatism as one moved off-axis. The astigmatism aberration was con- tions. This problem is now fixed. Note Optical quality tests with EFOSC after firmed during all optical tests. It is al- that this “astigmatism” differs from that refocusing the camera and performing ways a challenge to identify the origin caused by elastic deformation of the a careful thorough focus sequence re- of the astigmatism, but unfortunately mirror. Field astigmatism varies within established the instrument and tele- this can take a long time. Any kind of the field whereas the “stress” astigma- scope quality within the resolution de- stress or buckling on the telescope mir- tism is constant. Of course, in some livered by the two pixels sampling 0.7 ror will trigger the first elastic deforma- cases a combination of both effects arcsec. tion mode, and this corresponds to the could have been present. optical astigmatism term. Image Quality Degradation From the outset, all supporting ele- The Opto-Mechanical ments of both mirrors should be sus- Contribution After the installation of the Wide- pected. A complete check of all fixed Field Imager (WFI), several observers points as well as astatic levers (radial All the tests were performed with the reported bad, elongated images. The and axial) needs to be performed. participation of the mechanic team and defect was identified as astigmatism Because astigmatism was almost sym- long, fruitful discussions took place with which was found to increase for large metric over such a large field, the aber- the team members. A first check was North and East telescope orientations. ration could not have been produced by conducted on the lateral astatic pads On several occasions, even observa- a misalignment of the mirrors. In this after dismounting the mirror cell from tions at zenith showed the same fea- case, the origin points more to how the the telescope. The mirror is kept later- tures. Only a telescope “shake-up” mirrors are held. The defect appears at ally in position by a reference sphere in could remove the astigmatism, and large inclination and sometimes re- contact with the Cassegrain hole. even then, only temporarily. mains at zenith. In a first pass, the main Radial astatic levers maintain the mirror Optical tests were performed on mirror support system should be stud- laterally by pushing or pulling, and no several occasions with the new Cur- ied followed by the secondary mirror if force is applied when the mirror is vature Sensing Method (CSM) and nothing is found. horizontal. All levers were found to be these confirmed the astigmatism varia- Of course, the tests were conducted moving freely without mechanical tion for large zenithal distances. Earlier during short test periods within long stress. However, the reference sphere tests only performed at zenith showed stretches of observing time. This meant was found to be dirty on the northern the correct quality. Observations with that it was mandatory to keep the tele- side. The cause of this is thought to be direct CCD and EFOSC2 in the past scope in a stable condition for the ob- cleaning of the mirror when it is in- were not of sufficiently high spatial servers; which means that the tests clined, allowing some of the Carbon resolution to detect the problem. could only be done in small conserva- Dioxide snow to fall into the Cassegrain Therefore, the defect could have been tive steps. hole, taking dust from the mirror sur- present all along since the early days face with it. and gone unnoticed until the arrival of A Contribution from the The design of the mirror cell includes the WFI. Instrument Operation the facility to keep the mirror in a “park” position. In this case the mirror rests on Optical Tests An additional problem was discov- three supports without the control of ered which almost certainly contributed astatic levers. Three axial fixed points The tests could not reproduce the de- to the early reports of image problems. define the mirror orientation while the fect at zenith. Therefore, it seems that In the early days, the focus offsets pro- mirror is supported over the astatic there was some dependence on the duced by changes in temperature, were levers under the control of pneumatic history of telescope movements. As had only applied to the nominal reference pressure. When the air pressure is re- previously been done at the 3.6-m, the filter, and not automatically corrected moved, the mirror moves down by optical tests were performed by systema- when a different filter was used. For around 2 mm and the three axial fixed tically moving the telescope along the large focus offsets, the instrument was points move down within a spring ten- same sequence each time. First, a blank shifted out of focus and this produced sion device. The springs are used to sequence was followed to set the tele- the field astigmatism image elonga- keep the fixed points in contact with the scope properly, followed by South- North and East-West series. The CSM uses an ST8 SBIG CCD mounted on a translation stage to obtain the two defocused intra and extra focal images. Thirty-second exposures were performed to ensure that variations in the seeing were averaged out. This CCD is good in terms of sampling, with 9 mm pixels and a correct linearity to restore the beam intensity variance. The beam heterogeneity between both extrafocal images is produced by the optical aberrations. The WFI image quality is very good when correct focus is achieved. The small difference in the filter optical thickness must be properly compensat- ed by telescope focus offset. Te m- perature changes also introduce variation in the telescope focus. For conditions of very good seeing, the focus must be performed as carefully as possible. The WFI pixel size is only 0.24 Figure 1: Behaviour of the axial fixed points with changes in telescope position. The meas - arcsec, and with this, the sensitivity to urements shown were made during three intervals: before the April tests (short dashed lines), telescope imaging defects is increased. during the June tests (long dashed lines) and after all tests (solid line).

5 ror when the cosine astatic forces de- creased. Therefore, the pad pushing against the mirror introduced astigmatism and a small triangular deformation.

June 2000 Period

After a complete overhaul of the three fixed points a new round of tests was performed. Now the northern fixed point showed a new pattern with a slight decrease of the applied force in inclined telescope positions. The spring tension had to be increased to obtain the correct force, both at zenith and at large inclination. The last measurement of the axial fixed points showed the correct pattern, with a slight decrease for each in- Figure 2: Variation of the astigmatism over the same range of telescope positions as Figure clined position, due to the decrease in 1. The large deviation of the astigmatism for the east and west part of the cycle is clearly vis - the cosine component of the mirror ible. Part of the last July sequence was not performed because of bad weather. The larger weight. variations found during the last test night are due partially to a local thermal effect. July 2000 Period reference position with the correct ap- A Summary of the Tests A new check of the aberrations in the plied force on zenith. The astatic levers telescope was undertaken. The fixed are distributed around two rings. Two At each telescope position during the points again showed the correct be- manual air pressure controllers distrib- measurement sequence, a check of haviour for the sequence of measure- ute the pressure equally on all levers. both mirror position and the amount of ments. However, the spherical aberra- Each unit delivers air pressure to one aberration was made. tion term as well as those of the astig- ring only. At the beginning of the obser- matism increased “abnormally” at some vation, the telescope start-up operation April 2000 Period telescope positions. includes the air pressure distribution on Figure 1 shows the range of move- the mirror cell. The mirror raises until During this test period, the behaviour ment in the axial fixed points over the the mirror weight is in equilibrium with of the western fixed point was not cor- full range of telescope position at differ- the astatic levels, assuming the correct rect for large inclinations towards the ent times. Figure 2 and 3 show the be- positioning of the fixed points in contact north and east. A correlation with an in- haviour of the astigmatism and the tri- with their respective references. Due to crease of astigmatism and triangular angular aberration during the different the lack of load cell, the force applied aberration terms was also identified. test periods. on the mirror is not known (it is a non- The three fixed points were checked active optics). Departure on force distri- during the re-aluminisation of the 2.2-m The thermal contribution bution cannot be verified directly. The that took place on April 20. Effectively, only part allowing access to the mirror the west point was found to be tilted “Unforeseen” aberrations appeared position check are the three axial with improper mechanical contact in July, and only temporarily. They are points, where linear gauges have been which increased the force of the related to local thermal activity that acts installed. springs. The effect was to raise the mir- to produce a wavefront deformation. In this case, “thermal aberration” would be a more correct term to describe what takes place. On this test night the cold temperature of the outside air was producing large local effects. This effect is a good demonstration of the limitations that can exist in telescope imaging quality, not just from misalignment of the telescope (with mirror deformation in an improper cell design) but also from thermal convection. The thermal convection can be separated in different components, such as the dome/tube seeing, dome/slit seeing, mirror seeing and local perturbations produced by extra warm sources. The thermal conditions of each test night were different, with almost all except the last having the mirror colder than ambient air (by between zero and Figure 3: Variation of the triangular aberration over the same range of telescope positions. 2 degrees). Even so, the thermal ef- Again, the variations are large for the west and east part of the cycle before the April test. fects were still negligible compared to The thermal conditions of the last night (with a strong temperature decrease) produced a very the contribution of the axial fixed points unfavourable local convection effect on the mirror edge, thereby affecting the wavefront. to image degradation.

6 • Almost good images with 0.5 arcsec seeing; Zenith telescope position. tortion of the defocused images due to local thermal perturbations. S Extra-focal S Intra-focal Conclusion The arrows show a Following the interventions per- E beam asymmetry pro- E duced by convection formed up to and including July 2000, on the East side the telescope image quality remains good with several reports by observers a b as being very good. Under good external seeing conditions, it is possible to achieve sub-arcsec images, sometime • Inclined telescope with 1.1 arcsec seeing; at 60 ZD West. East side of the mirror lower. approaching 0.5 arcsec. The image quality achieved with the WFI is as ex- S Extra-focal pected over a large part of the sky. Intra-focal Improvements could still be made to minimise the thermal contribution when Large deformation on colder temperatures are experienced. E East side with large E The best method (which is already in astigmatism (0.33²) use on the 3.6-m and the NTT) would c d be to ventilate the main mirror with high-flux fans. A reduction in the mirror seeing, as well as local perturbations, could be achieved by blowing air •Two extra-focal images obtained within the same telescope position at 2 minutes interval! across the mirror, from north to south. The installation of load cells on the with 0.22² astigmatism three axial fixed points will be also a for- at the left ward step on the telescope improvement. The force delivered by each of with only 0.09² astig- the axial fixed points could be fine E matism at the right E tuned, to reach a new minimising of the residual low astigmatism. The limit of A clear crossing warm Bubble the image quality will be then defined e f by the pixel size of the WFI.

Acknowledgements

Figure 4: Thermal contribution to the image quality as seen in defocused images. We are grateful to the mechanics team for the discussions and participation during all the steps of this study. On the last night, however, the mirror causing a larger gradient with the re- The assistance of the Medium Size was warmer by 1.5 degrees, with a cor- maining warmth of the telescope delta Telescope Team, in allowing regular responding degradation in the mirror pads. Figure 4 clearly shows how this test periods and operational help, was seeing. The outside air temperature local effect is visible in the defocused also essential to the success of the im- reached a low value of 4 degrees, images. The figure also shows the dis- age quality improvement.

NEWS from the NTT O. HAINAUT and the NTT Team

A Motor-related Disaster motors have been removed, our (sin- the motor. This task is not as simple as gle) spare installed, and the electronics it sounds: the motors are actually com- At the time this is being written, the started to perform a complete check of plete servo-drive units, including in a NTT is running very nicely. While this is all the system driving the motors. In that very compact design the motor itself, its how the NTT is supposed to behave, it process, it was discovered that a third water/glycol cooling system, the has not been the case during the last motor presented some minor signs of tachometer and the brake, the com- month. Indeed, on January 16, it was damages. plete unit weighing 550 kg. Moreover, detected that one of the 4 main azimuth The next step was to open one of the some special tools had to be manufac- motors had died. The team immediate- faulty motors to diagnose and, ideally, tured; indeed, when the motors were ly started to reconfigure the drive sys- to repair it. For this purpose, we invad- delivered, it was not foreseen that they tem to operate without that motor (in ed the dome of the Schmidt telescope, would ever have to be re-opened, and case of emergency, we can run on 2 which was de-commissioned some the assembly tools were left at the fac- motors only). During that process, a time ago. This large room has plenty of tory. To make the situation even more second motor died! We decided to stop space, is very clean and equipped with challenging, it occurred in January, the operation and shut down the tele- a crane, making it the ideal place for which is right in the Chilean vacation scope to investigate; indeed, while we disassembling the motors. The La Silla period. La Silla was operated with a re- can survive with 2 motors, we cannot Mechanics, Electronics and NTT Team duced staff. Eventually, the main motor afford to kill one per day. The two faulty worked almost round-the-clock to open coil was accessed, and we realised that

7 Instruments

Since the previous report from the NTT, SOFI still had a few hick-ups related to its cryo-mechanics. T h e wheels, which suffer from the same problems as ISAAC on the VLT, will eventually be replaced by equivalent ones made of different material, hope- fully solving the problem in a definitive way. In the mean time, the instrument control software has been modified to avoid the problem. Asymmetric spectral lines have been reported by some of our observers using EMMI in Blue Medium Dispersion mode. This was tracked to a misalignment of the blue arm that has been corrected: the BLMD mode now provides perfectly symmetric lines. We still have to re-align the instrument for the dichroic mode. At the same time, the calibration of the long-slit unit used in medium dispersion spectroscopy has been checked and found quite accu- Figure 1: Dead motor in the Schmidt dome, ready for disassembly. rate. SUSI2’s detector head has been replaced by a new model that will not cause any contamination of the chips, a it was severely damaged: water had careful examination of the upper part of problem that plagued SUSI2’s early leaked on that coil over a long period of the motor chassis revealed little specks days. The detectors themselves have time, slowly building up a mixture of sul- of oxidation right were the cooling tubes been baked and cleaned, and SUSI2 is fate and carbon that eventually short- are located, so we suspect that the wa- now as good as new. circuited the coil with its steel chassis. ter diffused through micro-porosity of Unfortunately, even after cleaning this the steel. The next step is now to main- Improvements gunk out and restoring the connectors, tain and repair the remaining motors, the coils were still short-circuited, indi- and to fix them so that this problem In addition to the upgrade of the con- cating some internal damages that does not repeat itself in the future. trol system to the 2000 release of the could not be repaired on the spot. As By re-arranging the schedule and us- VLT software (which is extremely stable we still had a motor missing for normal ing a 4-night technical run combined and reliable), new workstations have operation, we decided to open the with a 2-night service observing run, we been installed. Visitors will appreciate spare altitude motor; while the mechan- could perform the priority observations the power of the HP Visualise B2000 ic parts are different, the coils are simi- of all but one of the observation runs and J2240 that are put at their dis- lar. The team, becoming expert in dis- that had been lost. posal. assembling these motors, removed the coil from the altitude motor and placed it in the body of the azimuth motor in 24 hours only. The final rush was to re-as- semble the azimuth motor and test it. I won’t go into details in describing the atmosphere in the Schmidt dome when the motor started to spin; enough said that the satisfaction was palpable. The mechanics immediately re-installed the repaired motor in the telescope and adjusted it (a task that lasted till the early hours of the next day), so that the NTT electronics could immediately start the adjustments and tests. The following night, we were on the sky for a quick, but thorough test and commissioning of the new system. The NTT was back to life, after 12 nights lost for observations. This was the longest down-time ever for the NTT. The cause of the short-circuit was obviously water. The main question is then to know where this water came from. The presence of glycol indicates that it was from a leak of the cooling system, not from condensation. The O- rings sealing the cooling pipes were found in perfect condition, so it seems Figure 2: The faulty coil removed from the motor; the brush on the right gives the scale of this they are not at the origin of the leak. A 60 kg piece.

8 Figure 3: Final avoids the formation of a hot-air bub- moments of the ble that would introduce some aber- re-assembly of the motor. rations. Another article in this issue of The Messenger describes these improvements and their results in more details.

Staff

Since our last report, Ismo Kastinen the hot air cells and Hernan Nuñez (Telescope and that could form Instrument Operators) have left the over the mirror. team; Ismo is now a Software Engineer This strongly con- at La Silla, and Hernan a TIO at the tributes to im- VLT. Gabriel Martin (TIO) is about to prove the image leave the NTT to works at the Magellan quality when there Observatory on Las Campanas as is no wind. T h e Instrument Specialist. Monica Castillo baffle of the sec- and Duncan Castex have joined the ondary mirror has Team as TIO. There are also many been changed for changes in the astronomical staff : a l i g h t - w e i g h t, Vanessa Doublier (fellow) has left the Kevlar and Car- Team, and Stephane Brillant (fellow) is bon fibre structure currently on his last shift at the NTT; that is optically both of them have been hired as staff equivalent to the astronomers at the VLT. Obviously, the former baffle, but NTT is an excellent training camp for We also installed a set of high-speed whose upper part is completely trans- Paranal! Malvina Billeres and Merieme fans around the main mirror, to produce parent to the wind. This provides an Chadid have joined the NTT Team as a constant artificial wind which disrupts air-flow over the secondary mirror, and fellows.

part of the Phase-2 preparations for New CCD on the Danish 1.54-m 2p2 Team News your programme. If you have an upcoming Vi s i t o r A new CCD was commissioned at H. JONES Mode WFI run, then you will need the Danish 1.54-m in September by a to familiarise yourself with the P2PP team from the Copenhagen University software. Specifically, you should know Observatory. The new EEV/MAT CCD Personnel Movements how to create and edit OBs in this (2048 ´ 4096 pixels) replaces the old environment, as this is what is used at Loral 2048 ´ 2048 detector to bring In December we welcomed Emanuel the telescope. You should also be about improvements on two fronts. Galliano to our team. Emanuel is a aware of the different types of WFI-spe- First, the new device does not suffer French student at ESO Chile who is al- cific templates available to build OBs, from the same charge diffusion prob- ready familiar with La Silla, through his and plan your observations according- lem as the old CCD. This problem was previous work with the DENIS group. ly. Whether or not you choose to pre- thought to have been responsible for He will be working primarily on opera- pare your OBs in advance of coming to the consistently poorer seeing meas- tions at the 2.2-m. observe is up to you. If you already ured at this telescope compared to In February, however, we bade have experience with P2PP, you may others. Second, the EEV CCD has farewell to Emanuela Pompei after wish to install the software at your half the read-out noise (3 e– rms) of nearly two years with the team. home institute and create your OBs the old device, and a much larger full- Although Emanuela is leaving La Silla, ahead of time. Otherwise it is better if well. she will remain with ESO in Chile, com- you can create your OBs on the moun- However, the optics of DFOSC do mencing work as a Staff Astronomer on tain. In this case, it is highly desirable not allow the full area of this large Paranal in March. We wish her all the that you arrive the day before your first format device to be used. The region best in her move north. night if this is at all possible. Allowing used suffers from some defects such ample time for preparation plays a ma- as bad columns and charge traps, P2PP and BOB Now on jor part in the efficiency of the observ- in a similar way to the Loral chip. The the 2.2-m ing run. quantum efficiency of the EEV de- Further information can be found at vice is slightly less, peaking around 450 In December, the final commission- our 2.2-m P2PP/BOB web page, at nm and declining steadily to the red. ing phase of the new operating soft- ht t p : / / w w w. l s . e s o . o r g / l a s i l l a / Te l e s c o p e s / The parallel charge transfer efficiency ware at the 2.2-m took place. This 2p2T/E2p2M/WFI/P2PP_BOB/ . It con- shows no losses although there is a means that Wide Field Imager (WFI) tains links to the P2PPHome Page and small but non-negligible loss in the se- observing programmes at the 2.2-m the new WFI Templates Manual, which rial direction due to trap in the serial telescope are now performed using describes observing templates avail- register. VLT Observing Software, with all point- able for OB creation. The page also has A full report on the characteristics of ing and exposure acquisitions con- instructions for installing and running the device (by Anton Norup Sorensen trolled through Observation Blocks the software at your home institute, in- of the Copenhagen University Obser- (OBs). cluding use of the WFI Instrument vatory), is available from the 2p2 Team If you have upcoming Service Mode Package. As always, questions or com- Web Page at h t t p : / / w w w. l s . e s o . o r g / observations, then you will be contact- ments can be directed to the 2p2 Team l a s i l l a / Te l e s c o p e s / 2 p 2 T / D 1 p 5 M / m i s c / ed directly about the creation of OBs as at any time ([email protected]). ringo.ps .

9 Tunable Filters and Large Telescopes H. JONES (ESO Chile), A. RENZINI (ESO Garching), P. ROSATI (ESO Garching) and W. SEIFERT (Landessternwarte, Heidelberg)

Introduction Kudritzki et al. (2000) used FORS1 emission-line galaxies. Figure 2 shows at the VLT for the spectroscopic follow- example scans and scanning narrow- Traditionally, astronomy has relied up to a sample of emission-line objects, band “spectra” obtained with the tun- upon filters with a fixed bandpass to se- identified by narrow-band imaging in able filter for some of the same objects. lect the wavelengths of the light allowed the field of the Virgo cluster. The ex- Understandably, there is growing in- to reach the detector, thus allowing the pectation was to confirm them as intra- terest in the role that tunable filters can astronomer to derive some colour infor- cluster planetary nebulae, given their play in the instruments currently under mation about the objects under study. detection with a [OII] (lambda = 5007 Å) design and construction for the new In the optical, these filters are most of- filter. As it turned out, however, the nar- generation of large telescopes. These ten classical broadband UBVRI, or narrow passband was equally good at re- include tunable filters in instruments for row passbands centred at the wave- vealing Ly-alpha-emitting objects at z ~ the GranTeCan (OSIRIS: Cepa et al. lengths of the common emission-line 3.1, and nine were found. 2000) and SOAR telescopes (Cecil features, either at rest-frame or red- In some of the above examples, a fil- 2000), among others under considera- shifted wavelengths. ter with the desired passband luckily tion. The technique is all the more pow- Examples of the latter are becoming matched the project requirements; in erful when the focal reducing instru- numerous, especially on the 8–10-m- others, one had to be designed with a ments in which they are placed have class telescopes that make it possible specific target in mind. However, stud- the capability for both tunable imaging to detect very faint, distant emission- ies of this kind (as well as many others and multi-object spectroscopy, since line objects, even through narrow pass- in the local universe), would clearly the two modes are complementary. In bands. In this vein, Kurk et al. (2000) benefit given the use of a passband this article we review tunable imaging used FORS1 at the VLT with a 65-Å- that can be easily tuned both in its width and the future role it could potentially wide filter at 3814 Å to image a z = 2.2 and central wavelength, over the full play at the VLT. We also briefly mention radio galaxy, searching for nearby Ly- optical range. The use of (Wide-band) the wide range of science, both alpha detections at the same redshift. Tunable Filter (WTF) instruments at the Galactic and extragalactic, that could They detected around 50 such objects, Anglo-Australian and William Herschel be undertaken with a tunable filter on collectively suggestive of strong clus- Telescopes (Bland-Hawthorn & Jones an 8-m-class telescope. tering around the dominant radio 1998a,b) in the past five years has in- galaxy. Moreover, they also found ex- deed seen a very broad range of astro- Making a Filter Tunable tended Ly-alpha emission (~ 100 kpc in physical applications. At low redshifts, extent) centred on the galaxy, adding science undertaken with these instru- There are many ways of making a fil- further evidence to the possible sce- ments includes studies of brown dwarf ter with tuning capability, and conse- nario of protocluster formation. atmospheric variability (Tinney & Tolley quently, many types of tunable filter. Steidel et al. (2000) used an 80-Å- 1999), and the identification of optical These include those using birefringent wide filter on Keck to search for Ly- counterparts to Galactic X-ray sources materials (such as the Lyot, Solc and alpha emitters at z = 3.09, the redshift (Deutsch, Margon & Bland-Hawthorn acousto-optic tunable filters), more tra- of a prominent peak in the redshift dis- 1998). High-redshift science has in- ditional interferometers such as the tribution of their original sample of cluded estimates of the cosmic star- Michelson and Fabry-Perot, and even broad-band selected Ly m a n - b r e a k formation history (Jones & Bland- liquid crystal tunable filters. We will not galaxies. This took the number of Hawthorn 2001), identification of galaxy describe details of each technology galaxies associated with the peak from clustering around high-redshift QSOs here, but instead refer the interested 24 to 162, a gain of almost a factor of 7, (Baker et al. 2001), deep imaging of jet- reader to Bland-Hawthorn (2000), who thereby demonstrating the power of cloud interactions in powerful radio discusses the merits of each for tunable narrow-band observations in the de- galaxies (Tadhunter et al. 2000), and imaging. While all have advantages tailed mapping of large-scale structures the detection of a large ionised nebula and limitations, in the end it is the strin- at high redshift. They also found ex- around a nearby QSO (Shopbell et al. gent demands of night-time astronomy tended (~ 100 kpc) Ly-alpha emitting 2000). Figure 1 shows a section of field that dictate which are feasible. For as- “blobs’’, that again may point to in- from a tunable filter survey on the tronomical applications, the ideal filter cipient cluster formation at these red- Anglo-Australian Telescope (Jones & should have high peak transmission, a shifts. Bland-Hawthorn 2001), for distant broad (rectangular) profile, and should be large enough to admit a generous beam size. It should also be of good imaging quality, produce a stable and reproducible passband, and cover a large range of wavelengths, to name just the major requirements. Of all the possibilities, it is the Fabry- Perot interferometer that has been the popular choice of astronomers for three-dimensional spectral imaging. This is because they are readily available on a commercial basis and make use of well-established technology. Astronomical applications of these in- Figure 1: A section of field (approximately 6 2 arcmin with north up, east left) from the emis - struments have included studies of ex- sion-line galaxy survey of Jones & Bland-Hawthorn (2001). The numbers next to each emis - tended diffuse nebulae (e.g. Haffner, sion-line candidate are object identifications. Reynolds & Tufte 1999) and obtaining

10 kinematic information (such as line widths and radial velocities) on nearby disk galaxies (e.g. Amram and Östlin in this issue, p. 31; Laval et al. 1987; Cecil 1989; Veilleux, Bland-Hawthorn & Cecil 1997; Shopbell & Bland-Hawthorn 1998). The Fabry-Perot systems em- ployed have traditionally used narrow wavelength coverage and high spectral resolutions (resolving powers R 1500). There are two problems to be over- come in the adaptation of a Fabry-Perot interferometer to tunable imaging. First, it must work at sufficiently low spectral resolution (in other words, narrow plate spacing) that scanning in spectral steps over larger ranges of wavelength is feasible. Second, the filter coatings must be optimised over a large range of wavelengths. Tradition- ally, Fabry-Perots have had neither the wavelength coverage nor the ability to work at such small plate spacing.

Fabry-Perot Tunable Filters

It is a little more than one hundred years since Charles Fabry and Alfred Perot first highlighted the potential of an interference device producing fringes from two parallel silvered plates (Perot & Fabry 1899; Fabry & Perot 1901). Modern Fabry-Perot interferometers consist of two parallel glass plates held a small distance apart, such that con- structive interference of light between the plates causes only specific wavelengths to be transmitted. However, with typical plate spacings in the range 20 to 500 microns, Fabry-Perots for astronomical work have been confined to high orders of interference (50 to 2000), thereby giving rise to the high resolving powers mentioned earlier. Tunable filters differ from convention- Figure 2: (Top) Individual object scans for some of the same candidates as in Figure 1. al Fabry-Perot devices in two novel but Individual images are 9 arcsec on a side with north at top, east to the left. Circles denote aper - important ways. First, the plates are op- ture size. (Bottom) Spectral flux measurement for the same galaxies. Both preliminary (dotted line) and erated at much smaller plate spacings final (solid line) continuum fits are shown. Numbers shown on the right are flux ( 10–16 than the Fabry-Perot instruments so far ergs/s/cm2, a star-galaxy classification parameter and deviation of the line detection in sig - used for astronomy. The effect of this is ma. Deviant points (excluded from the final continuum fit) are indicated by circles. The zero to widen the central interference region flux level is shown by the horizontal tickmarks (where present) and non-detections are rep - of the chosen wavelength. A conven- resented on this level by crosses. Galaxy 214.14 has independently been found to have tional Fabry-Perot, with a plate spacing emission in [OII] by Ellis et al. (1996); the emission we see here is H-alpha and [NII]. of many tens or even hundreds of microns, presents an interference region as a narrow ring on the sky, with very ter, aiming to access as broad a tunable or more deep frames at the same small area (Fig. 3a). A tunable filter, range as is possible, needs to access a etalon spacing, Fig 3ii), and, with a plate spacing of no more than a much wider range of plate settings. (iii) tunable filters can obtain a se- few microns, aims to provide a broad- This is made possible by having a stack quence of monochromatic images with- ened central interference region, of piezo-electric transducers (PZTs) to in a field defined by the Jacquinot spot, known as the Jacquinot spot (Fig. 3b). control plate spacing, instead of the (Fig. 3iii). The latter is more useful for survey usual single-layer. These structural dif- Atherton & Reay (1981) were the first work, where one seeks a common ferences between a tunable filter and a to suggest the possibilities of a Fabry- wavelength transmitted across the full conventional Fabry-Perot contribute to Perot as a tunable imager. However, field and where lower spectral resolu- the different types of data that are ob- the technology available at the time tions are desired. tained with each instrument: was not sufficient for precise control of The second way in which a tunable (i) conventional Fabry-Perots can be the plates over a such wide range of filter differs from a conventional Fabry- used to obtain a high-resolution nar- spacings, and suitable coatings were Perot is in its ability to access a much row-range spectrum at each pixel posi- not very good by the standards of today wider range of plate spacings. Con- tion over a wide field (Fig. 3i), (see Pietraszewski 2000 for descrip- ventional devices are most commonly (ii) conventional Fabry-Perots can tions of the current state of the art in used to scan through a relatively small also be used to obtain a single spec- Fabry-Perot technology). In the mid- range of wavelengths around a single trum of a diffuse source which fills a 1990s, J. Bland-Hawthorn (AAO) re- spectral feature. However, a tunable fil- large fraction of the aperture (from one visited the tunable filter concept by

11 ered by a tunable filter are not strictly monochromatic, but shift slightly to the blue as one moves from the location of the optical axis on the image, to the edge. This phase effect is a natural consequence of interference between two surfaces and easily characterised through the cosine of the off-axis angle (at the etalon). The red TTF shifts about 18 Å at a distance of about 5 arcmin from the optical axis, as measured on the sky. The phase effect is not normally a problem for most applications, especially if the objects of interest are compact sources such as distant galaxies and stars.

A Tunable Filter versus Many Narrow-band Filters

Is a tunable filter any better than simply having a large set of filters? With a Fabry-Perot tunable filter it is possible to control both the width and placement of the bandpass. The only restriction on placement is that it must lie within one of the order-sorting filters. For example, the red TTF on the AAT can select a bandpass of between 6 to 60 Å in any of the 7 order-sorting filters that collectively cover 2300 Å in the 6500 Å to 1 micron range of the device. With a set of fixed filters one is stuck with a predetermined width and placement for the bandpass. Tilt-tuning might be considered an option but this quickly broadens and degrades the passband, in the sense that peak transmission is lowered and the profile Figure 3: Different modes of Fabry-Perot use in the case of both (a) conventional instruments, skewed. Furthermore, such tilt tuning and (b) tunable filters. only allows adjustment to the blue. The inability to set the bandpass means one cannot optimise spectral resolu- having an old, disused conventional ure 4 shows, the answer lies in the tion, nor background level, nor central Fabry-Perot, refitted with stacked PZTs, nature of the transmission profile of the wavelength nor sampling (in the then repolished and recoated for the Fabry-Perot, and the different effect of case of scans). The ability to tune and range 6500 Å to 1 micron. The result making small and large adjustments. optimise is critical to projects with was the first TAURUS Tunable Filter Figure 4a shows the set of blocking fil- (i) objects at arbitrary redshift, (ii) back- (TTF; Bland-Hawthorn & Jones ters used with the red TTF. These are ground-limited observations, and (iii) 1998a,b) implemented at the Anglo- necessary to block the light of unwant- objects with a specific spectral fea- Australian Telescope (AAT). Two years ed orders and make excellent interme- ture in mind. These make up the major- later, an instrument coated for the blue diate-band filters in their own right, giv- ity of front-line narrow-band observing, (3700 to 6500 Å) was commissioned. en their placement between the bright- since much has already been done in These instruments are operated at the est parts of the night-sky background. the way of imaging bright objects at Cassegrain focus of the AAT, and to- Suppose we wanted to scan around H- common spectral features. Further- gether have been used extensively. alpha at 6563 Å. First we would put the more, the exact optical characteristics The instruments can also be used in R0 blocking filter in place (Fig. 4a, solid of individual filters will vary slightly conjunction with a CCD charge-shuf- line). Then we would set the spacing of from filter to filter, implying inhomo- fling mode, that allows repeated the plates according to the desired geneities which will need to be dealt multi-band imaging on different parts width of our passband. Figure 4b with on a filter-by-filter basis. This limits of the CCD frame, before it is read shows that if we set the plates to 8 or the ability to do precise differential im- out (Bland-Hawthorn & Jones 1998a, 10 microns, we get a very narrow pro- aging between two bands – a common b). More details of the different fea- file (at orders 24 or 30 respectively); if application in both stellar and extra- tures of TTF can be found at http:// we set the plates to just 2 or 4 microns galactic work. w w w. a a o . g o v. a u / l o c a l / w w w / j b h / t t f / . we get a much wider band (orders 6 or The advantages of using a Fabry- 12). If we then wanted to scan the pass- Reducing Tunable Filter Data Perot for tunable imaging has also band, we would adjust the spacing be- been recognised by other groups (e.g. tween the plates by small amounts, There is often the perception that Thimm et al 1994, Meisenheimer et al. thereby shifting the chosen order one Fabry-Perot interferometers produce 1997). way or the other in wavelength. The data sets that are difficult for first- How does one set about tuning a dotted profiles in the lower panel of time users to understand and reduce. passband to a specific width and place- Figure 4b show the effect of changing 2 This is not true in the special case of ment, when all one is changing is the microns slightly to 1.98, 1.96 and 1.94 tunable filters, as the variety of pub- spacing between the plates? As Fig- microns. Note that the images deliv- lished results from the AAT and WHT

12 Figure 4: (a) Wavelength region covered by the red TTF at the AAT, showing the location of the interme - diate blocking filters with respect to bands of OH night-sky emission. (b) Transmission profile of the tunable filter plot - ted on the same scale, as a function of chang - ing plate separation (left). Even-numbered orders are indicated. The dotted lines in the lower panel show the effect of changing the plate spacing to 1.98, 1.96 and 1.94 µm.

show. Conventional Fabry-Perots work tasks (TFred) offering a range of tools about 135 mm in height (as measured at much higher resolving powers and to treat tunable filter data in IRAF. A along the optical axis). This space so when they are used to map kine- more comprehensive treatment is given could be accommodated in the colli- matics in nearby galaxies, it requires in Jones, Shopbell & Bland-Hawthorn mated beam of FORS2 if the upper of precise mapping of the wavelength (2001), where approaches to the re- the two grism wheels were dismounted. change across the field, so that a 3D- duction of Fabry-Perot tunable filter Nevertheless, the full spectroscopic ca- spectral cube can be constructed and photometry are described. pability of FORS2 is preserved, albeit subsequently transformed into a 2D with more frequent grism exchanges. map of velocities. However, tunable Tunable Filters and the VLT The FORS2 echelle mode would be filter data are typically concerned only lost, although it would be possible to re- with scanning surveys of small point- There is currently no tunable filter install the corresponding grisms in like sources such as stars or distant capability on the VLT, and (in the short- FORS1. Most importantly, no major galaxies. This requires nothing more term at least), neither on 8–10-m-class mechanical hardware modifications than the detection, matching and pho- telescopes elsewhere. It is therefore would be necessary to effect the im- tometry of each object on each frame – worthwhile to contemplate if and how it plementation. The control electronics routine steps in any imaging survey, might be possible to implement a tun- needed to operate the tunable filter are with or without a tunable filter. There able filter at the VLT. While a detailed delivered by industry. However, to allow is still a change in wavelength across technical, cost and manpower study re- for efficient use of the filter by the ob- the field of the tunable filter, but the mains to be done, we have investigat- server, a full integration into the GUI of lower resolving power makes this a ed the main parameters of a possible the instrument would be needed. less dramatic effect in terms of the incorporation of a tunable filter into the As the free aperture of the tunable fil- broader width of the transmitting band- FORS2 focal reducer. ters is slightly smaller than that needed pass. The largest commercially available to cover the full FORS field of view, a vi- Tunable filter data from the AAT have tunable filter which could be fitted into gnetting of 7.5 % occurs at the corners been reduced with scripts utilising both either of the FORS instruments has a of the detector, preserving an unvi- the FOCAS (Valdes 1993) and SEx- free aperture of 116 mm. The mechan- gnetted central field of 4.8 arcmin di- tractor (Bertin & Arnouts 1996) pack- ical size of such a unit (sealed to min- ameter. The blue shift of the filter trans- ages for the object detection. One of us imise thermal/environmental influ- mission at the edge of the unvignetted (Jones) has written a collection of IRAF ences) is 200 mm in diameter and field is only a fraction of the transmis-

13 sion profile width and therefore not a members being identified relatively Universe in Three Dimensions”, A S P problem for most applications. For the easily through association with the ‘red Conf. Series 195, eds. W. van Breugel lowest resolution, ordinary broad-band sequence’ of passively evolving ellipti- and J. Bland-Hawthorn, 34. filters could be used as blocking filters. cals. With such methods, however, spi- Cecil, G. 1989, in “Extranuclear Activity in Additional filters needed to work at rals and star-forming galaxies that may Galaxies”, eds. E. J. A. Meurs and R. A. E. Fosbury, (ESO: Garching bei Mün- higher orders could be placed in the also be cluster members are much chen, Germany). two interference filter wheels in front of more difficult to identify, given their Cecil, G. 2000, in “Optical and IR Telescope the detector. The optimum passband broad range of colours, which can Instrumentation and Detectors”, Proc for such filters is slightly less than the sometimes act to make them indistin- SPIE 4008, eds. M. Iye and A. F. M. Moor- free spectral range at the highest re- guishable from foreground or back- wood, 83. solving powers envisaged for use. ground galaxies. However, tuning the Cepa, J., A g u i a r, M., Escalera, V. G., The useful wavelength range of the tunable filter to a suitable emission Gonzalez-Serrano, I., Joven, E., Peraza, tunable filter is largely arbitrary but line at the cluster redshift easily per- L., Rasilla, J. L., Rodriguez, L. F., Gon- needs to be decided at the time of man- mits identification of these late-type zalez, J. J., Cobos, F. J., Sanchez, B., ufacture, as it is governed by the design galaxies. Tejada, C., Bland-Hawthorn, J., Militello, C. , Rosa, F., 2000 in “Optical and IR (and resulting performance) of the Mapping the large-scale structure out Telescope Instrumentation and Detec- etalon coating. For example, the two to z ~ 5 is one of the main goals within tors”, Proc SPIE 4008, eds. M. Iye and A. tunable filters in use at the AAT individ- reach of the current generation of large F. M. Moorwood, 623. ually cover 3700–6500 Å and 6500 Å–1 telescopes and their instruments. The Deutsch, E. W., Margon, B., Bland-Haw- micron. With a device in FORS, spec- pilot experiment by Steidel et al. (2000) thorn, J. 1998, PASP 110, 912. tral resolutions achievable at 370 nm clearly demonstrates the advantages of Ellis, R. S., Colless, M., Broadhurst, T., Heyl, for example, would range from 330 to narrow-band detection for this kind of J., Glazebrook, K. 1996, MNRAS 280, 1350, while at 650 nm would encom- work. Indeed, tuning the filter to the 235. pass 180 to 720. Such resolutions as- redshift peaks found through future Fabry, C., Perot, A. 1901, ApJ 13, 265. sume a variation of the spacing be- Lyman-break galaxy surveys will ex- Haffner, L. M., Reynolds, R. J., Tufte, S. L. 1999, ApJ, 523, 223. tween the etalon plates from 2 to 8 mi- pand the number of galaxies associat- Jones, D. H., Bland-Hawthorn, J. 2001, ApJ, cron, which can be achieved through ed with these large-scale features in press. stacked piezo-electric transducers. The many times over. Multi-object spec- Jones, D. H., Shopbell, P. L., Bland-Haw- tunable filter has a high efficiency com- troscopy will further complement this, thorn, J. 2001, MNRAS, submitted (see parable or even superior to common by determining the dynamics of sheets, http://sc6.sc.eso.org/~hjones/WWWpubl/ narrow-band filters. filaments, and proto-clusters. detect_meas.ps.gz). The approach of using FORS2 to fur- As the new VLT instrument VIMOS Kudritzki, R.-P., Mendez, R. H., Feldmeier, J. nish the VLT with a tunable filter carries comes into operation, it becomes nec- J., Ciardullo, R., Jacoby, G. H., Freeman, both pros and cons. On the one hand essary to re-assess the role of the two K. C., Arnaboldi, M., Capaccioli, M., there is the relatively small effort com- FORS instruments, since the former Gerhard, O., Ford, H. C. 2000, ApJ, 536, 19. pared to that of building an entirely new will outperfom the FORSes in many of Kurk, J. D., Rottgering, H. J. A., Pentericci, instrument, and the extension of the their current applications. An upgrade L., Miley, G. K., van Breugel, W., Carilli, C. scientific uses of FORS2 following the plan for the two FORS instruments is L., Ford, H., Heckman, T., McCarthy, P., commencement of VIMOS. On the oth- therefore under study at ESO, including Moorwood, A. 2000, A&A, 358, L1. er there is the need to remove one of the red-optimisation of the CCDs on Laval, A., Boulesteix, J., Georgelin, Y. P., the grism wheels to make the space. FORS2, which is planned for later in the Georgelin, Y. M., Marcelin, M. 1987, A&A, Eliminating one grism wheel will make y e a r. Installing a tunable filter on 175, 199. for increased manual intervention in the FORS2 is another possibility for con- Meisenheimer K., Beckwith, S., Focken- exchange of grisms if one wants to pre- sideration, and in this article we brock, R., Fried, J., Hippelein, H., Hopp, serve all the filter/grism combinations have illustrated some of the scientific U., Leinert, Ch., Roeser, H.-J., Thommes, presently available. Clearly, the relative advantages and a possible technical E., Wolf, C. 1997, in “The Early Universe with the VLT”, (Springer: Berlin), 165. weight of these different arguments implementation. The VLT also currently Perot, A. and Fabry, C. 1899, ApJ 9, 87. needs to be evaluated before deciding lacks an efficient UV imager, and in- Pietraszewski, K. A. R. B. 2000, in “Imaging how to proceed. deed another upgrade under consider- the Universe in Three Dimensions”, ASP ation concerns the UV optimisation of Conf. Series 195, eds. W. van Breugel Potential Science on 8-m-class FORS1, all the way to the atmospheric and J. Bland-Hawthorn, 591. Telescopes cutoff. Together, these upgrades would Shopbell, P. L., Bland-Hawthorn, J. 1998, restitute new scientific utility to the ApJ 493, 129. Scientific applications of a tunable FORS instruments, while significantly Shopbell, P. L., Veilleux, S., Bland-Haw- filter at an 8-m telescope span an ex- expanding the capabilities of the VLT thorn, J. 2000, in “Imaging the Universe in tremely wide range, potentially satisfy- overall. Three Dimensions”, ASP Conf .Series 195, eds. W. van Breugel and J. Bland- ing the needs of what is a very diverse Hawthorn, 404. user community. Several such applica- References Steidel, C.C., A d e l b e r g e r, K.L., Shapley, tions were described at the beginning A.E., Pettini, M., Dickinson, M., Giava- of this article. Here, we mention a few Amram P. and Östlin G., 2001, T h e lisco, M. 2000, ApJ, 532, 170. more possibilities. At the high limit of Messenger 103, p. 31. Tadhunter, C., Villar-Martin, M., Morganti, R. spectral resolution (R ~ 1500) it may be Atherton, P. D., Reay, N. K. 1981, MNRAS 2000, in “Imaging theUniverse in Three possible to probe the internal dynamics 197, 507. Dimensions”, ASP Conf. Series 195, eds. of most kinds of emission-line nebulae Baker, J. C., Hunstead, R. W., Bremer, M. W. van Breugel and J. Bland-Hawthorn, and relatively nearby galaxies, along N., Bland-Hawthorn, J., Athreya, R. A., 284. with that of QSO and radio galaxy envi- Barr, J. 2001, AJ, in press. Thimm, G. J., Roser, H.-J., Hippelein, H., Meisenheimer, K. 1994, A&A 285, 785. ronments. At lower resolutions (R ~ Bertin, E., Arnouts, S. 1996, A&A 117, 393. Bland-Hawthorn, J., Jones, D. H. 1998a, Tinney, C. G., Tolley, A. J. 1999, MNRAS 150), most (but not all) applications will PASA, 15, 44. 304, 119. concern the distant universe. For ex- Bland-Hawthorn, J., Jones, D. H. 1998b, in Valdes, F. 1993, “FOCAS User’s Guide”, ample, high-redshift clusters of galax- “Optical Astronomical Instrumentation”, NOAO. ies are usually found either in deep X- Proc SPIE 3355, ed. S. D’Odorico, 855. Veilleux, S., Bland-Hawthorn, J., Cecil, G. ray or infrared surveys, with cluster Bland-Hawthorn, J. 2000, in “Imaging the 1999, AJ 118, 2108.

14 R E P O RT S F RO M O B SE RV E R S

Seeing the Light Through the Dark (Continued from page 1) no dipole moment and cannot produce this need by enabling the direct meas- 5146 (Lada et al. 1994, Lada et al. a readily detectable signal under the urement of the dust extinction toward 1999), L 977 (Alves et al. 1998), and conditions that characterise cold, dark thousands of individual background Barnard 68 (Alves, Lada, & Lada 2001), clouds. The traditional methods used to stars observed through a molecular where we detected nearly 7000 infrared derive the basic physical properties of cloud. Such measurements are free sources background to these clouds such molecular clouds therefore make from the complications that plague mo- and produced detailed maps of the ex- use of observations of trace H2 surro- lecular-line or dust-emission data and tinction across the cloud to optical gates, namely those rare molecules enable detailed maps of cloud structure depths and spatial resolution an order with sufficient dipole moments to be to be constructed. of magnitude higher than previously easily detected by radio spectroscopic possible (AV ~ 40 magnitudes, spatial techniques (Lada 1996), and interstel- 2. The Method and Results resolution ~ 10 arcsec). lar dust, whose thermal emission can So Far We have used our extinction obser- be detected by radio continuum tech- vations to measure the masses, densi- niques (e.g., André et al. 2000). The most straightforward and reliable ty structure, extinction laws, and dis- However, the interpretation of results way to measure molecular cloud struc- tances to these objects. We found the derived from these methods is not al- ture is to measure dust extinction of radial density profiles of filamentary ways straightforward (e.g., Alves, Lada, background starlight. We have devel- clouds (IC 5146 and L 977) to be well & Lada 1999; Chandler & Richer 2000). oped a new powerful technique for behaved and smoothly falling with a Several poorly constrained eff e c t s measuring and mapping the distribution power-law index of a = –2, significantly inherent in these techniques (e.g., devi- of dust through a molecular cloud using shallower than predicted by early theo- ations from local thermodynamic equi- data obtained in large-scale, multi- retical calculations of Ostriker (1964) (a librium, opacity variations, chemical wavelength, infrared imaging surveys. = –4). Moreover, because we are using evolution, small-scale structure, deple- This method combines measurements pencil beam measurements of dust col- tion of molecules, unknown emissivity of near-infrared colour excess to di- umn density along the line of sight to properties of the dust, unknown dust rectly measure extinctions and map background stars, we were able to temperature) make the construction of the dust column density distribution demonstrate that the small-scale struc- an unambiguous picture of the physical through a cloud (see Fig. 2). It is the ture of the clouds is surprisingly smooth structure of these objects a very difficult most straightforward and unambiguous with random density fluctuations (dAV / task. There is then a need for a less way of determining the density struc - AV) present at very small levels (< 3%!). complicated and more robust tracer of ture in dark molecular clouds. More- This result is in very good agreement H2 to access not only the physical over, the measurements can be made with optical studies of the small-scale structure of these objects but also to at significantly higher angular resolu- structure of the diffuse Interstellar accurately calibrate molecular abun- tions and substantially greater optical Medium (ISM) (Thoraval, Boissé, & dances and dust emissivity inside these depths than previously thought possi- Duvert 1999). clouds. The deployment of sensitive, ble. We have conclusively demonstrat- When convolved to the appropriate large-format infrared array cameras on ed the efficacy of this technique with spatial resolution, our maps showed large telescopes, however, has fulfilled our study of the dark cloud complex IC structure in the dust distribution which was strikingly well correlated with millimetre wave CO and CS emission maps of the cloud (see Fig. 3), although showing crucial differences at high optical depths where these other tracers of column density become unreliable (see Fig. 4). These comparisons en- abled us, for the first time, to directly + derive CO, CS, and N2H abundances, and variations of, over an extinction range of 1–30 magnitudes, a range nearly an order of magnitude greater than achieved previously with optical star-counting techniques. In a recent experiment we were able to make a direct measurement of molecular depletion in a cold cloud core (Kramer et al. 1999). Finally, a comparison between our extinction data and millimetre continuum emission data allowed a most accurate measurement of the ratio of dust absorption coefficients at millimetre and near-infrared wavelengths (Kramer et al. 1998).

3. Barnard 68 as a Stellar Seed Figure 2: Illustration of the Near-Infrared Colour Excess (NICE) method used to derive and map dust column density in molecular clouds. Because the amount of reddening is directly proportional to the total extinction, we can determine the line-of-sight extinction to each star Recently we have been concentrat- seen through a molecular cloud (at near-infrared wavelengths, NIR) using a reddening law ing efforts on mapping the densest re- for interstellar dust and knowledge of the star’s intrinsic NIR colour. gions of the ISM that are likely places of

15 Large telescope (VLT) on Cerro Paranal, fitted with FORS1 CCD camera (Appenzeller et al. 1998), during one night of March 1999. The results of the optical and near-infrared imaging are displayed in Figure 1 (see page 1). At optical wavelengths obscuring dust within the cloud renders it opaque and completely void of stars (left). However, due to the wavelength dependence of dust extinction (i.e., opacity), the cloud is essentially transparent at infrared wavelengths enabling otherwise invisi- ble stars behind the cloud to be imaged (right). We detected 3708 stars simulta- neously in the deep H and K band images out of which ~ 1000 stars, lying behind the cloud, are not visible at optical wavelengths. Because dust opacity decreases sharply with wavelength (Fig. 6), the colours of stars that are detected through a dust screen appear reddened compared to their intrinsic colours. We have accurately sampled the Figure 3: Dust extinction map and C18O molecular line map of L977. The structure in the dust dust extinction and column density dis- distribution correlates strikingly well with the millimetre-wave C18O emission map of the cloud, tribution through the Barnard 68 cloud although showing crucial differences at high optical depths (see next figure) where the CO at more than a thousand positions with tracer becomes unreliable (from Alves et al. 1999). extraordinary (pencil beam) angular resolution. Although the individual measurements are characterised by future star formation. We performed Third, the cloud is sufficiently nearby high angular resolution, our mapping of very sensitive near-infrared imaging that foreground star contamination is the dust column density in the cloud is observations to map the structure of a negligible. highly undersampled. Consequently, type of dark cloud known as a Bok glob- We used the SOFI (Moorwood, Cuby, we smoothed these data to construct ule (Bok & Reilly 1947), one of the least Lidman 1998) near-infrared camera on the first 10 arsec resolution map of dust complicated configurations of molecu- the European Southern Observatory’s extinction of a cold dark cloud (Fig. 7) lar gas known to form stars (Fig. 5). The New Technology Telescope (NTT) to and an azimuthally averaged radial ex- target cloud for our study, Barnard 68 obtain deep infrared J band (1.25 mm), tinction (dust column density) profile of (Figs. 1 and 5), is itself one of the finest H band (1.65 mm) and Ks band (2.16 the cloud (Fig. 8). This is the most fine- examples of a Bok globule, and was se- mm) images of the cloud over two nights ly sampled and highest signal-to-noise lected because it is a nearby, relatively in March 1999. Complementary optical radial column density profile ever ob- isolated and morphologically simple data were obtained with ESO’s Very tained for a dense and cold molecular molecular cloud with distinct bound- aries, a known distance (125 pc; Launhardt & Henning 1997), and temperature (16 K; Bourke et al. 1995). It was first discovered by E. E. Barnard (Barnard 1919) and was the target of several optical dust extinction studies by Bart Bok and co-workers (Bok & Reilly 1974; Bok 1977). Although a very dense cloud, Barnard 68 does not present any of the signatures of ongoing star formation, such as IRAS sources, outflows, or mm continuum sources (Avery et al. 1987; Reipurth, Nyman, Chini 1996). Barnard 68 lies in the direction of the centre of the Galaxy but above the galactic plane where it is projected against the rich star field of the galactic bulge. This makes Barnard 68 a particularly ideal candidate for an infrared extinction study for the following reasons. First, the background bulge stars are primarily late-type (giant) stars whose intrinsic infrared colours span a narrow range and can be accurately deter- 18 mined from observations of nearby Figure 4: Relation between N(C O)LTE and visual extinction AV for molecular cloud L 977. The solid straight red line represents the result of a linear least-squares fit, with errors in both control fields. Second, the background coordinates, over the entire data set. There is a clear deviation from the linear relation at ex - star field is sufficiently rich to permit a 18 tinctions ͧ10 magnitudes above which C O becomes a very poor tracer of H2 (from Alves detailed sampling of the extinction et al. 1999). Follow-up molecular line study of this cloud (Tafalla et al. 2001) suggests that, across the entire extent of the cloud. as in the IC 5146 cloud, depletion of CO might be occurring at high optical depths.

16 cloud. For the first time, the internal structure of a dark cloud has been specified with a detail only exceeded by that characterising a stellar interior.

3.1 Bonnor-Ebert Spheres

The extinction profile in Figure 8 is the projection of the cloud volume density profile function, and therefore provides an exquisite view of the internal structure of this dense dark cloud. As early as 1948 Bart Bok pointed out that roughly spherical homogenous looking clouds, such as Barnard 68, resemble single dynamical units much like the polytropic models of Lane and Emden used to describe stellar structure (Lane 1870; Emden 1907). Can Barnard 68 be described as a self-gravitating, polytropic sphere of molecular gas? To investigate the physical structure of the cloud, we begin with the assumptions of an isothermal equation of state and spherical symmetry. The fluid equation that describes a s e l f-gravitating, isothermal sphere in hydrostatic equilibrium is the following well-known variant of the Lane- Emden equation (Lane 1870, Emden 1907):

Figure 5: Palomar Digitized Sky Survey image of the neighbourhood of Barnard 68. Complexes of globules like the ones in this image (Barnard 68 to Barnard 72) may be the (1) precursors of small young stellar groups, like the well-known TW Hya.

Figure 6: Deep BVIJHK imaging of dark molecular cloud Barnard 68 done with FORS1 at the VLT and SOFI at the NTT. The wavelength de - pendence of interstellar dust extinction in Barnard 68 is clearly depicted in these images. The analysis of the near-infrared colours of the stars seen through the dark cloud allow the construction of the first 10 resolution map of mass as traced by dust extinction, and the most finely sampled and higher S/N density profile ever obtained for a cold dark cloud (from Alves et al. 2001b).

17 deed an isothermal, pressure confined, Figure 7: 10 dust extinction map of and self-gravitating cloud. It is also like- Barnard 68. The ly to be in a state near hydrostatic equi- contours start at AV librium with thermal pressure primarily = 4 mag and supporting the cloud against gravita- increase in steps of tional contraction. For Barnard 68, xmax 2 mags. The peak is very near and slightly in excess of the extinction measured critical radial parameter and the cloud through the very may be only marginally stable and on centre of the cloud is the verge of collapse. If this is the case, 33 magnitudes of extinction (from we should expect molecular radio- Alves et al. 2001a). spectroscopy of this cloud to reveal a quiet, non-turbulent cloud with narrow molecular emission lines. Indeed, preliminary results from our radio-spectroscopy observing campaign of Bar- nard 68 (with the IRAM 30-m Radio Telescope at Pico Veleta, Granada) reveal that the line width of the C18O line –1 in this cloud is Du ~ 0.18 kms , one of the narrowest lines ever observed in molecular clouds, in perfect agreement with the Bonnor-Ebert sphere nature of Barnard 68.

3.2 Reverse Engineering: Distance where x is the non-dimensional radius, ROSAT satellite (Breitschwerdt et al. and Gas to Dust Ratio to Few 2000). Percent The close correspondence of the (2) observed extinction profile with that The exact physical state of the Bar- predicted for a Bonnor-Ebert sphere nard 68 cloud is further constrained by while y(x) equals, strongly suggests that Barnard 68 is in- the fact that xmax, which is solely de-

(3) for which r is the distance from the centre of the sphere, is the isothermal sound speed inside the gas cloud ( = ₂A (KT/ m ) ), is the density, and c is the central density. For an isothermal sphere bounded by a fixed external pressure there is a family of solutions characterised by a single parameter (Ebert 1955, Bonnor 1956):

(4)

Here xmax is the value of x at the outer boundary, R. Each of these solutions corresponds to a unique cloud mass density profile. Bonnor demonstrated that for xmax > 6.5 such a gaseous con- figuration would be unstable to gravitational collapse (Bonnor 1956). The high quality of our extinction data permits a detailed comparison with the Bonnor- Ebert predictions and we find that there is a particular solution, (xmax = 6.9 ± 0.2, that fits the data extraordinarily well as seen in Figure 8. For the known distance (125 pc), and temperature (16 K), Barnard 68 has a physical radius of 12 ,500 AU, a mass of 2.1 solar masses, and a pressure at its boundary of P = 2.5 ´ 10–12 Pa. This surface pressure is an order of magnitude higher than that of the general ISM (McKee 1999) but it is in rough agreement with the pressure Figure 8: Azimuthally averaged radial dust column density profile of Barnard 68. The red cir - inferred for the Loop I superbubble, cles show the data points for the averaged profile of a subsample of the data that do not in - where Barnard 68 is embedded, de- clude the cloud’s south-east prominence, seen in Figure 6. The solid black line represents rived from X-ray observations with the the best fit of a theoretical Bonnor-Ebert sphere to the data.

18 Figure 9: Dark cloud BHR 71 caught in the act of forming a stellar binary. This cloud will be a primary target for deep extinction mapping with the VLT. The nebula seen against the dark cloud is a cavity carved out by a spectacular molecular outflow (Bourke et al. 1997; Garay et al. 1998 – see also Garay et al. in The Messenger No. 83). Combining data from ISO and SEST, Bourke (2001) has shown that BHR71 proba - bly contains a binary system in the making, with each protostar driving its own outflow. This colour composite was constructed by R. Fosbury and R. Hook with data taken during Science Verification of FORS2.

rived from the shape of the observed cloud, our measurement of its extinc- certainty (3%) arises solely from the un- column density distribution, uniquely tion and angular size (combined with certainty in xmax. Accounting for the un- specifies the combination of central the constraint that xmax = 6.9 ± 0.2) in- certainty in the temperature measure- density, sound speed and physical size dependently gives the distance to the ment, the overall uncertainty increases that characterises the cloud (i.e., cloud, provided the gas-to-dust ratio is only to 8%, or ± 9 pc. From its associa- Equation 4). Independent knowledge of assumed. The temperature of the motion with the Ophiuchus complex, the any two of these parameters directly lecular gas in Barnard 68 has been pre- distance to the cloud has been estimat- determines the value of the third. For viously measured using observations of ed to be 125 ± 25 pc (de Geus et al. example, in Barnard 68 we independ- emission from the (1,1) and (2,2) 1989), which within the uncertainties ently measure the dust extinction metastable transitions of the ammonia agrees with our derivation. This, in turn, (which is directly related to the mass molecule and found to be 16 ± 1.5 K implies that the gas-to-dust ratio in this column density, via the gas-to-dust ra- (Bourke et al. 1995). For a canonical dense cloud must be close to the tio in the cloud) and the angular size of gas-to-dust ratio (1.9 ´ 1021 protons/ canonical interstellar value. Indeed, if the cloud (which is directly related to its magnitude; Bohlin, Savage, & Drake we independently know the distance to physical size via the cloud’s distance). 1978), we derive a distance to Barnard the cloud, our modelling directly yields Thus, if we know the temperature of the 68 of 112 ± 3 pc. Here the quoted un- the gas-t o-dust ratio in the cloud.

19 Assuming a distance of 125 pc, our tain newly-formed stars (Launhardt & (Tucson: University of Arizona Press), measurements allow a high-precision Henning 1997) and thus it is likely that eds. Mannings, V., Boss, A.P., Russell, determination (2%) for the gas-to-dust our observations of the starless Bar- S. S. ratio in this cloud of 1.73 ± 0.04 ´ 1021 nard 68 cloud provide the first de- A p p e n z e l l e r, I., Fricke, K., Fürtig, W. , protons/magnitude. If we also account tailed description of the initial condi- Gässler, W., Häfner, R., Harke, R., Hess, H., Hummel, W., Jürgens, P., Kudritzki, for the uncertainties in distance and tions prior to the collapse of dark glob- R., Mantel, K., Meisl, W., Szeifert, T., & temperature, the overall uncertainty ules and the formation of isolated, low- Tarantik, K. Successful Commissioning of (accuracy) of our determination in- mass stars. FORS1 – the First Optical Instrument on creases to ± 0.4 ´ 1021 protons/magni- the VLT, The Messenger 94, 1–6 tude, or 23%. Within the overall error, (1998). this ratio is the same as the long ac- 4. The Future Bok, B. & Reilly, E. 1947, ApJ, 105, 255. cepted value characterising low-density Bok, B. 1948, Centennial Symposia (Har- interstellar gas and is the first inde- Significant progress in extinction vard Observatory Monographs No. 7); Har- pendent and relatively accurate deter- mapping studies will result when the vard-College Observatory, Cambridge. mination of this important astrophysical DENIS and 2MASS all-sky near-infra- Bourke, T., Hyland, A., Robinson, G., James, S. & Wright, C. 1995, MNRAS, 276, 1067. parameter in a dense molecular cloud red imaging surveys are completed and Bourke, T., Garay, G., Lehtinen, K., Koeh- core. released. These surveys will be suffi- nenkamp, I., Launhardt, R., Nyman, L., ciently sensitive to produce moderate- May, J., Robinson, G., Hyland, A. 1997, 3.3 On the Origin of Small Stellar depth extinction maps (i.e., AV £ 25 ApJ, 476, 781. Groups mags) of many nearby dark clouds in Bourke, T. 2001, ApJ, submitted. those directions of the Galaxy where Breitschwerdt, D., Freyberg, M. & Egger, R. Recently, there has been special at- field stars suffer little extraneous ex- 2000, A&A, 361, 303. tinction. In the immediate future, large- Chandler, C. & Richer, J. 2000, ApJ, 530, tention for isolated and sparsely popu- 851. lated associations of young low-mass aperture telescopes, such as the VLT outfitted with ISAAC and NAOS/CONI- de Geus, E., de Zeenw, P. & Lub, J. 1989, stars similar to the recently identified A&A, 216, 44. TW Hydra association (Rucinsky & CA, will provide the additional capabili- Emden, R. 1907, Gaskugeln, (Leipzig: Krautter 1983). The TW Hydra associa- ty to perform deeper surveys of the Teubner). tion is a stellar group near the solar sys- higher extinction regions (25 < AV < 60 G a r a y, G., Köhnenkamp, I., Bourke, T. , tem consisting of a handful of young mags) in these clouds (as the star form- Rodriguez, L., Lehtinen, K. 1998, ApJ, low-mass, sunlike stars, which has ing globule BHR 71 in Figure 9). Finally, 509, 768. been a primary target of recent substel- space-based infrared telescopes, such Garay, G., Köhnenkamp, I., Rodriguez, L., lar objects search (e.g. Neuhäuser et as the NGST, should enable the re- 1966, The Messenger 83, 31. Kramer, C., Alves, J., Lada, C., Lada, E., gions of deepest extinction (AV > 60 al. 2000a, Neuhäuser et al. 2000b). Sievers, A., Ungerechts, H., Walmsley, M. The existence of such a young stellar mags) to be probed. Together, such observations promise to render a very 1997, A&A, 329, 33. group presents an interesting problem Kramer, C., Alves, J., Lada, C.J., Lada, E.A., complete understanding of the phys- to astronomers because its origin is dif- Sievers, A., Ungerechts, H., Walmsley, ficult to explain given its youth and rel- ical and chemical structure of mo- C.M. 1999, A&A, 342, 257. atively large distance from known sites lecular clouds and of the general ini- Lada, C.J., Lada, E.A., Clemens, D.P., & of star formation. Bok globules such as tial conditions to the star-formation Bally, J. 1994, ApJ, 429, 694. those in the Barnard 68 group are process. Lada, C.J. 1996, Proceedings of IAU thought to be remnant dense cores Symposium CO: Twenty-Five Years of produced as a result of the interaction Millimeter-Wave Spectroscopy, eds. W.B. Latter et al. (Kluwer-Dordrecht), p. 387. of massive O stars and molecular 5. Acknowledgements Lada, C.J., Alves, J., Lada, E.A. 1999, ApJ, clouds (Reipurth 1983). Over their short 512, 250. lifetimes, such massive stars, through The authors acknowledge Marco Lombardi for fruitful discussions and Lane, J. 1870, American Journal of Science ionisation, stellar winds and ultimately and Arts, Series 2, 4, 57. supernova explosions, very effectively assistance, the Paranal Science Launhardt, R. & Henning, T. 1997, A&A 326, disrupt the molecular clouds from which Operations team for observing Barnard 329. they formed. In the process, large 68 with FORS1 on VLT Antu, Richard Mckee, C. 1999, in The Origin of Stars and shells of expanding gas are created. West and Edmund Janssen, Richard Planetary Systems, eds. C. Lada & N. When surrounding clouds are disrupted Hook and Robert Fosbury for compos- Kylafis (Dordrecht: Kluwer). by the passage of these shells, a few of ing Figure 1. The authors are thankful Moorwood, A., Cuby, J. & Lidman, C. SOFI their most resilient dense cores will be to Monika Petr for helpful discussions Sees First Light at the NTT, The Messenger 91, 9–13 (1998). left behind, embedded within the shell’s during the preparation of the VLT observations. Neuhäuser, R., Brandner, W., Eckart, A., hot interior. Remnant cores with just the Guenther, E., Alves, J., Ott, T., Huélamo, right mass can establish pressure equi- N., Fernandez, M. 2000a, A&A, 354, 9. librium with the hot gas within the shell N e u h ä u s e r, R., Guenther, E., Petr, M., and survive to become Bok globules. References B r a n d n e r, W., Huélamo, N., Alves, J. Eventually, as a result of processes 2000b, A&A, 360, 39. described above, these clouds will Alves, J., Lada, C.J., Lada, E.A., Kenyon, S., Ostriker, J. 1964, ApJ, 140, 1056. evolve to form low mass, sun-like stars & Phelps, R. 1998, ApJ, 506, 292. Reipurth, B. 1983, A&A, 117, 183. which are relatively isolated and far Alves, J., Lada, C.J., & Lada, E.A. 1999, Reipurth, B., Nyman, L. & Chini, R. 1996, from the original birthplace of the O ApJ, 515, 265. A&A, 314, 258. Alves, J., Lada, C.J., & Lada, E.A. 2001a, Rucinski, S. & Krautter, J. 1983, A&A, 121, stars. We suggest that Barnard 68, and Nature, 409, 159. 217. its neighboring globules B69, B70 and Alves, J., Lada, C.J., & Lada, E.A. 2001b, Tafalla, M., Alves, J., & Lada, C. 2001, in B72 (Figure 5) may be the precursors ApJ, submitted. prep. of a small stellar group, like TW Hya. André, P., Ward-Thompson, D., & Barsony, Thoraval, S., Boissé, P., Duvert, G. 1999, Up to 35% of all Bok globules con- M. 2000, Protostars and Planets I V, A&A, 351, 1051.

20 Strong Accretion and Mass Loss Near the Substellar Limit F. COMERÓN, ESO, Garching, Germany M. FERNÁNDEZ, Instituto de Astrofísica de Andalucía, Granada, Spain

Accretion and outflows of mass are their millimetre-wave emission. Even Young brown dwarfs are currently among the most distinctive phenomena stars with only a few tenths of the mass known to share many characteristics associated to star formation. Their ob- of the Sun display in their earliest with the more massive T Tauri stars. servational manifestations cover a stages spectacular signatures of inter- The similarities include mid-i n f r a r e d broad range of appearances and wave- action with the circumstellar environ- emission, revealed by ISOCAM (e.g. lengths, from the large X-ray emitting ment, such as the strong emission lines Comerón et al. 1998, Persi et al. 2000) bubbles caused by stellar winds mov- seen in T Tauri stars or the fast-moving from warm dust in circumstellar disks or ing at several thousands of kilometres jets that produce Herbig-Haro objects. envelopes that provide large reservoirs per second, to the cold dust shells Can strong accretion and mass loss of mass for accretion. The spectra of around low-mass stars detected by take place even at substellar masses? very young brown dwarfs often display

Figure 1: A B, V,RC image of the field around LS-RCrA1 obtained with the Wide Field Imager at the ESO-MPG 2.2-m telescope. LS-RCrA 1 is the faint object at the centre and marked with an arrow. The bright nebulosity at the upper left corner contains the T Tauri stars R and T CrA. The comma-shaped nebula to the bottom right of that nebula is the Herbig-Haro object HH 100, and the red compact nebula at the bot - tom centre of the image is HH 101. The bright star near the centre of the image, to the left and below LS-RCrA1, is V709 CrA. The field is approximately 10 10 in size.

21 coexist. This is the bidden lines due to [OI], [OII], [NII], and latest-type object [SII] are also clearly identified, as well for which such an as the Ca ll triplet near 8550 Å. The ap- intense emission- pearance and intensity of these emis- line spectrum has sion lines is not unprecedented, and been observed so the line ratios are similar to those of f a r. The late-t y p e Krautter’s star (Th 28; Graham and spectrum and the Heyer 1988), a G8-K2 star that powers faintness of the a string of Herbig-Haro objects. How- underlying object e v e r, the underlying spectrum of suggest that it is LS-RCrA1 is much later than that of Th near or below the 28 and other T Tauri stars with strong borderline sepa- emission, and can be reliably classified rating stars from as M6-M7. At the age of the R Coronae brown dwarfs, Australis, such a late spectral type im- showing that such plies a mass below 0.1 solar masses, spectacular spec- and probably substellar. tral signatures can The ISAAC K-band spectrum also be present even at presented in Figure 2 is remarkably masses of a few featureless. At the resolution and sig- per cent of a solar nal-to-noise of these observations, a mass. The details late M-type spectrum should display of this work are de- atomic features in that region due to scribed in extent in Na I, Ca I, Mg I, as well as prominent a separate paper CO bandheads starting at 2.29 µm. (Fernández & Co- Only the latter are clearly identified in merón 2001). the spectrum on LS-RCrA 1, but with equivalent widths of less than half the Observations typical value for its spectral type. Finally, the H2 line at 2.12 µm is clearly The special char- seen in emission. acteristics of the A final surprise from our observations object that we pre- of LS-RCrA1 comes from its faintness. sent here, hereafter The measured brightness from our WFI referred to as LS- images is B = 23.1, V = 21.3, RC = 19.8, RCrA 1 (where LS IC = 18.0, and from our SOFI images J stands for La Silla) = 15.3, H = 14.5, K = 13.9. Since its were first revealed moderately blue colours imply a slight in a slitless spec- extinction at most and the R Coronae troscopy survey of Australis clouds are only 150 ± 20 pc the core of the R away, this places LS-RCrA1 among the Figure 2: Visible (top) and near-infrared (bottom) spectra of LS- RCrA 1, obtained respectively with FORS2 and ISAAC in March Coronae A u s t r a l i s intrinsically faintest members known of 2000. The most prominent emission lines are marked in the visible star-forming re- a star-forming region. spectrum. In the infrared spectrum we mark the position of the de - gion, carried out tected lines due to H2 and CO, as well as the expected positions of with DFOSC at the Interpretation some atomic features that should be detectable in this region for a 1.5-m Danish tele- normal M6 star. scope at La Silla in The numerous forbidden emission April 1999. T h e lines in the spectrum of LS-RCrA 1 Ha emission that is commonly associ- slitless spectroscopy frames showed strongly resemble those seen in ated to accretion, as confirmed from only a dot at the expected position of l o w-mass stars undergoing strong high-resolution spectroscopy (Muze- Ha for an otherwise inconspicuous mass loss, such as those powering rolle et al. 2000). Like T Tauri stars, object of RC ȃ 19.5, with a continuum Herbig-Haro objects (e.g. Reipurth et young brown dwarfs also have been too faint to be seen in those ob- al. 1986). On the other hand, the large found to possess X-ray emission servations. Near-infrared JHK photom- strength of the Ha line as compared to (Neuhäuser & Comerón 1998; Come- etry was obtained with SOFI at the [SII] suggests that most of the Ha emis- rón, Neuhäuser, and Kaas 1998) NTT in February 2000. Further spec- sion is actually due to accretion, rather caused by the magnetic fields that play troscopic observations were carried out than to mass loss. The similar equiva- a fundamental role in regulating the at the VLT, both with FORS2 in the lent widths of the Ca II lines near flow of mass from the accretion disk visible and ISAAC in the K band, in 8500 Å, together with the absence of onto the surface (Hartmann 1998). In March 2000. Finally, short exposures in forbidden-line emission of [CaII], also view of these similarities, one may won- BVRC IC were obtained with the WFI hint that CaII emission arises from a der if the spectral signposts of intense at the 2.2-m telescope in August, to dense, optically thick region. It seems accretion sometimes displayed by clas- check for possible photometric vari- therefore that LS-RCrA 1 is simulta- sical TTauri stars, and very often found ability. A colour composite of a part of neously undergoing strong accretion to be correlated with strong mass loss, the WFI image is shown in Figure 1, and mass loss, just like its higher-mass may also be found near the substellar centred on LS-R C r A 1 (see also TTauri counterparts that lie at the origin limit or even below. http://www.eso.org/outreach/press-rel/ of Herbig-Haro flows. Here we present our observations of pr-2000/phot-25-00.html). The K-band spectrum presented also a very late-type faint member of the R Figure 2 shows both the FORS2 and shows some characteristics commonly Coronae Australis star-forming cloud ISAAC spectra of LS-RCrA1. The most found among very young objects en- that displays an unusually rich emis- outstanding feature in the visible is the shrouded by circumstellar material, sion-line spectrum, similar to that of abundance of strong emission lines, such as H2 emission and the absence more massive counterparts, in which dominated by Ha with an equivalent of any detectable absorption features both accretion and outflow signatures width of approximately 330 Å. For- apart from much weakened CO band-

22 heads. The lack of absorption lines is Figure 3: Position of commonly interpreted as due to strong LS-RCrA 1 (filled veiling of the photospheric spectrum by circle with error emission from warm circumstellar dust, bars) in a tempera - that contributes most of the flux at 2 mm ture-luminosity dia - gram, with theoreti - and beyond, and is usually correlated cal isochrones and with the appearance of H2 emission evolutionary tracks (Greene & Lada 1996). In that respect, from Baraffe et al. the K-band spectrum of LSRCrA 1 is (1998) plotted for “Class I-like”, following the widely-used comparison. Also classification of young stellar objects shown are the posi - (e.g. Shu, Adams, and Lizano 1987). tions of other late- However, the shape of the continuum at type objects in star- 2 µm is remarkably flat in that region, forming regions, whose spectral while the infrared JHK colours show no types and infrared appreciable sign of the circumstellar photometry are tak - excess emission that would be needed en from the to dilute the photospheric features be- literature and trans - yond detectability in our spectra. In other formed into temper - words, dust does not seem to signifi- ature and luminosity cantly contribute to the flux of LS-RCrA using the same 1 in the 2 micron region, which in that method as for respect is “Class III-like”. Such a co- LS-RCrA 1 (see Fernández and existence of Class I-like and Class III- Comerón 2001 for like features in the K-band spectrum of details on the trans - LS-RCrA 1 is not found among the formation). The oth - higher-mass objects that have been er sources are from studied so far, and leads us to consider Chamaeleon I (open circles; Comerón, Neuhäuser, and Kaas 2000) and IC 348 (open other possible explanations to the lack squares; Luhman 1999). Also plotted are V410 X-ray 3 (open triangle; Luhman 2000) and of atomic features and the weakness of Oph 162349.8-242601 (filled square; Luhman, Liebert, and Rieke 1997). As explained in the CO bands. An interesting possibility the text, the error bars come primarily from the uncertainty in the transformation from spec - in this respect is that the photospheric tral type and photometry to temperature and luminosity, and affect in a similar way the posi - tions of all the sources. Therefore, although the precise luminosity and temperature of spectral features in the 2 µm region LS-RCrA1 are uncertain by the amount given by the error bars, the offset relative to the may be largely filled by emission pro- other sources plotted is a well established feature. duced in the heated infalling material near the surface of the star, without being accompanied by continuum emis- amount. Therefore, although the tem- time. The impact of both strong accre- sion (Martin 1996). perature and luminosity of LS-RCrA 1 tion and mass loss on the structure of What is the mass and age of LS- are determined with only a rather limit- the atmosphere of LS-RCrA1 may thus R C r A 1? Its membership in the R ed accuracy, its large offset with re- be sufficient to invalidate a direct com- Coronae Australis star-forming region spect to other known late-type young parison between its observational char- and the vigorous accretion and mass- objects is well established. acteristics and the predictions of theo- loss activity, found only at the earliest If taken at face value, the position of retical models that do not take those stages of stellar evolution, both suggest LS-RCrA 1 seems to imply that its age factors into account. Indeed, calcula- an age of only a few million years. At is of the order of several times 107 tions performed by Hartmann, Cassen, this age, LS -RCrA 1 should be early in years, about one order of magnitude and Kenyon (1997) have found that ac- its contraction track and have a rela- older than the age inferred from the rest cretion increases both temperature and tively large radius, and therefore bright- of the members of the R Coronae luminosity with respect to the predic- ness. However, as mentioned earlier, Australis region (Wilking et al. 1997), tions of accretionless models that as- LS-RCrA 1 is surprisingly faint as com- and also much older than other stars sume the same mass and age of the pared to objects of similar spectral type displaying such strong signs of accre- central object. The net result is to make in star-forming regions. Figure 3 illus- tion. The rather implausible age and the the object appear hotter, and somewhat trates this: we have plotted in it the offset with respect to other young ob- older, than an object of equal mass and position of LS-RCrA1 in a temperature- jects of similar underlying spectral char- age but without accretion. The calcula- luminosity diagram, where these quan- acteristics leads us to look for alterna- tions of those authors use only moder- tities are inferred from its spectral type tive interpretations to the unexpected ate accretion rates on central objects of and available photometry. Also shown position of LS-RCrA 1 in the tempera- larger mass than the one presented for comparison are pre-main sequence ture-luminosity diagram. here, and can therefore not be directly evolutionary tracks and isochrones The most obvious peculiarity that extrapolated to LS-RCrA 1. However, from Baraffe et al. (1998), and the posi- separates LS-RCrA 1 from the other they do suggest that LS-RCrA1 may be tions of other very low mass stars and very low mass objects plotted in Figure actually younger than the 5 ´ 107 years, brown dwarfs identified in other star- 3 is the signs of strong accretion and and less massive than the ~ 0.08 solar forming regions. The main contribution mass loss on an object of such a low masses, implied by the direct compari- to the error bars is due to uncer- temperature and luminosity, and this son to pre-main-sequence tracks. tainties in translating spectral types may be the reason why LS-RCrA 1 Since 0.08 solar masses is very close and magnitudes into temperatures and looks so old and so different from those to the borderline separating low-mass luminosities. Since the position of the other objects. Modellers of low-mass stars from brown dwarfs, the possibility other objects plotted in Figure 3 was pre-main sequence evolution in the last that accretion is actually making the computed in the same way as that of decade have stressed the great impor- spectral type appear earlier than it LS-RCrA1, any systematic errors in the tance of an appropriate, realistic treat- would be without accretion suggests estimate of temperature and luminosity ment of the boundary condition repre- that LS-RCrA 1 may have a mass well of the latter should move both LS- sented by the atmosphere for correctly below the brown dwarf limit. RCrA 1 and the other sources in the reproducing the evolution of tempera- In any case, regardless of whether same direction and by a similar ture and luminosity as a function of LS-RCrA 1 is stellar or substellar, it is

23 clearly a so far unique object that and distribution of dark and hot spots cution of our service mode observa- poses an interesting case study in on its surface? Can we trace the reser - tions with FORS2 at Kueyen. several respects. It demonstrates that voir of cold circumstellar gas around the spectacular emission-line systems L S-R C r A 1 through its mid-, far-i n- References found in classical T Tauri stars can be frared, and radio emission? How com- Baraffe, I., Chabrier, G., Allard, F., Hau- present also at much lower masses, mon are objects like LS-RCrA 1 in schildt, P. H., 1998, A&A, 337, 403. and suggests that intense accretion star-forming regions? Is the rich emis- Comerón, F., Neuhäuser, R., Kaas, A. A., and mass loss can dramatically alter sion-line spectrum of LS-RCrA1 a rari- 1998, The Messenger, 94, 38. the spectroscopic and photometric ty among very low mass objects, or Comerón, F., Neuhäuser, R., Kaas, A. A., properties of the underlying object. It does it rather represent a short-lived 2000, A&A, 359, 269. stresses the need to complement exist- evolutionary phase? Is the simultane- Comerón, F., Rieke, G. H., Claes, P., Torra, ing models for the early evolution of ous appearance of Class I and Class III J., Laureijs, R. J., 1998, A&A, 335, 522. low-mass objects with significant accre- characteristics a part of the peculiarities Fernández, M., Comerón, F., 2001, submit- tion and mass loss rates. A conse- of LS-RCrA 1, or is it common among ted to A&A. Graham, J. A., Heyer, M. H., 1988, PASP, quence of this is the intriguing pos- young very low mass stars and brown 100, 1529. sibility of biases in current studies of dwarfs? Are there other signs of activi- Greene, T. P., Lada, C. J., 1996, AJ , 112 , 2184. the mass and age distributions of ty, such as X-ray emission, associated Hartmann, L., 1998, “Accretion processes in young stellar aggregates if a significant to LSRCrA 1? Answering these ques- star formation”, Cambridge Univ. Press. fraction of their members undergo actions and understanding objects like Hartmann, L., Cassen, P., Kenyon, S. J., cretion and mass loss phases like LS-RCrA 1 from a theoretical, quantita- 1997, ApJ, 475, 770. LSRCrA 1, due to the reliance of such tive point of view, is essential for plac- Luhman, K. L., 1999, ApJ, 525, 466. studies on pre-main sequence evolu- ing LS-RCrA 1 in the context of what is Luhman, K. L., 2000, ApJ, 544, 1044. tionary tracks. already known about the early stages Luhman, K.L., Liebert, J., Rieke, G. H., 1997, ApJ, 489, L165. Of course, LS-RCrA 1 also suggests of the lives of stars at different masses, Martin, S. C., 1996, ApJ, 470, 537. a variety of follow-up of observational extending such knowledge beyond the Muzerolle, J., Briceno, C., Calvet, N., studies: what information can we obtain substellar edge. Hartmann, L., Hillenbrand, L., Gullbring, from high-resolution spectra of the E., 2000, ApJ, 545, L141. emission lines? Does the mass loss of Neuhäuser R., Comerón F., 1998, Science, LS-RCrA1 cause any visible impact (as Acknowledgements 282, 83. yet undetected in our images) in the Persi, P., et al., 2000, A&A, 357, 219. surrounding interstellar medium, like We are pleased to thank the staffs of Reipurth, B., Bally, J., Graham, J. A., Lane, the Herbig-Haro objects generated by both La Silla and Paranal Obser- A. P., Zealey, W. J., 1986, A&A, 164, 51. Shu, F. H., Adams, F. C., Lizano, S., 1987, more massive stars? How do the sig- vatories for excellent support during our ARA&A, 25, 23. natures of accretion and mass loss vary observations, as well as to Paranal Wilking, B. A., McCaughrean, M. J., Burton, with time? What do possible photomet- Science Operations and the User M. G., Giblin, T., Rayner, J. T., Zinnecker, ric variations tell us about the existence Support Group for the successful exe- H., 1997, AJ, 114, 2029.

Star Formation at z = 2–4: Going Below the Spectroscopic Limit with FORS1 J. U. FYNBO1, P. MØLLER1, B. THOMSEN2

1European Southern Observatory, Garching, Germany 2Institute of Physics and Astronomy, University of Århus, Århus, Denmark

Introduction galaxies fainter than the current spec- 2000b). In March 2000 we carried out troscopic limit, one has to rely on other f o l l o w-up Multi-Object Spectroscopy The population of bright galaxies at z selection criteria than the Ly m a n - with FORS1 on the 8.2-m Antu tele- = 2–4 has been studied intensively us- Break. Two promising possibilities are scope (UT1). We also obtained deeper ing the Lyman-Break technique (Steidel (i) to select galaxies with Ly-a emission broad-band B and I imaging reaching 5 et al. 1996; Cristiani et al. 2000). lines, and (ii) to study the host galaxies (2)s detection limits in 1 arcsec2 circu- Currently, redshifts can be determined of Gamma-Ray Bursts (GRBs). lar apertures of 25.9 (26.9) in the I-band from absorption features of galaxies se- and 26.7 (27.7) in the B-band (both on lected in this way down to R Ȃ 25.5 Ly- Selected Galaxies the AB system). The results of these (e.g. Steidel et al. 2000), which is com- observations are presented in Fynbo et monly referred to as the spectroscopic Ly-a selection of high-redshift galax- al. (2001a), and summarised here. limit. Currently, very little is known ies has been attempted for many years, about the galaxy population below the but only recently with significant suc- Imaging spectroscopic limit. This is an unfortu- cess (Møller and Warren 1993; Francis nate situation since all the information et al. 1995; Cowie and Hu 1998; Pas- In Figure 1 we show image sections on the chemical enrichment of young carelle al. 1998; Kudritzki et al. 2000; with Ly-a (top), VLT B-band (middle) galaxies (Damped Ly-a Absorbers) ac- Fynbo et al. 2000a; Steidel et al. 2000; and VLT I-band (bottom) for S7–S12. cessible through QSO absorption lines Kurk et al. 2000). In 1998 we detected As seen, despite the faint detection lim- seems to be valid mainly for galaxies 6 candidate Ly -a emitting galaxies its, only for S7 and S9 is there a corre- significantly fainter than R = 25.5 (called S7–S12) in the field of the QSO sponding source detected in the broad (Fynbo et al. 1999; Haehnelt et al. Q1205-30 at z = 3.036 with deep NTT bands (B(AB) = 25.6 and 25.4 respec- 2000). In order to select and study n a r r o w-band imaging (Fynbo et al. tively). For the sources S8 and

24 Figure 1: Image sections (12 12 arcsec2) from the NTT narrow-band (top), VLT B-band (middle) and VLT I-band (bottom) for each of the six candidate emission-line galaxies S7–S12 (Fynbo et al. 2000a). East is to the left and north up.

S10–S12 no convincing broad-band 4600 Å–5270 Å around the emission Ly-a emitters and not foreground emis- counterpart is detected. line (left) and in the spectral region sion-line galaxies. In the upper panel For the brightest source, S9, there is 6050Å–6720Å (right). As seen, there is we show the spectrum of Q1205-30 in an offset of 0.6 ± 0.2 arcsec between one strong emission line detected in the the same spectral regions as for the Ly-a centroid and then centroid of blue part of the spectrum. This line S7–S12 to illustrate, in the same way, the continuum source. The same offset could be due to a foreground emission- that there is no CIV emission from is also seen in the 2-dimensional spec- line galaxy at z = 0.313 with [OII]l 3727 S7–S12. Hence, the Ly-a emission is trum of the source (Fig. 2, the slit was in the narrow filter, or Ly-a at z = 3.036. most likely powered by star-formation. nearly aligned along the direction of the To discriminate between the two possi- These results show that Ly-a selec- offset). To assess whether a likely ex- bilities we look for the presence of oth- tion allows the study of high-redshift planation for this is that we see two er lines. In the lower panel we show the galaxies that are currently not accessi- sources superposed, we calculate the spectrum of a z = 0.224 (B(AB) = 24.2) ble with the Lyman-Break technique. probability for a chance alignment emission-line galaxy detected in one of The inferred space density of S7–S12 along the line of sight. The surface den- the redundant slits. This spectrum has is about 10 times higher than that sity of galaxies down to the 5s detec- been redshifted to z = 0.313 so that the of R < 25.5 Ly m a n-Break galaxies tion limit in the VLT B-band image is 50 [OII]l 3727 emission line falls at the (Fynbo et al. 2000a). Only ~ 20% of the per arcmin2. The probability to find a same wavelength as the observed Lyman-Break galaxies show Ly-a in galaxy by chance within 0.6 arcsec emission lines of S7–S12. Wi t h emission (Steidel et al. 2000). This may from a given position on the sky is dashed, vertical lines we indicate the be an underestimate if Ly-a and con- therefore roughly p ´ 0.62 ´ 50/3600 = positions of the [NeIII], Hb and [OIII] tinuum emission often have different 1.6%. The probability of such a chance emission lines seen in the spectrum of spatial distributions. However, if this alignment in one of six cases is hence the z = 0.313 galaxy. As seen, there is fraction is valid further down the lumi- roughly 10%, which is small but not no hint of these lines in the composite nosity function, then there could be 50 negligible. Therefore, with the present spectrum of S7–S12, which confirms times more galaxies like S7–S12 than data we cannot conclude whether S9 is that S7–S12 indeed are high redshift R < 25.5 Lyman-Break galaxies. Of a single object with strong Ly-a emission centred 0.6 arcsec from the continuum emission, or a chance alignment of two objects. Note here that evidence that the spatial distribution of Ly-a emission of high-redshift galaxies can be different from that of the continuum emission of the same galaxy, was reported by Møller and Warren (1998) and Roche et al. (2000).

Spectroscopy

The individual spectra of S7–S12 are shown in Fynbo et al. (2001a). All 6 sources have confirmed emission lines. In the middle panel of Figure 3, we show the composite spectrum of all Figure 2: The 2-dimensional spectrum of the brightest Ly- source, S9, showing the offset 6 sources in the spectral region between the Ly- emission and the continuum emission.

25 Figure 3: The middle level at z = 2. Otherwise it would be un- panel shows the likely to detect R = 28 galaxies as GRB composite spectrum of hosts. S7–S12. The lower pan - els show the spectrum of an emission-line Conclusion: Faint Galaxies at galaxy redshifted to z = High Redshifts 0.313 so that the [OII] line falls at the The study of Damped Ly-a wavelength of the Absorbers (Fynbo et al. 1999; Haehnelt observed emission line et al. 2000), Ly-a selected galaxies and of S7–S12. As seen, the GRB host galaxies independently sug- composite spectrum of gest that there are 1–2 orders of mag- S7–S12 shows none of nitude more galaxies fainter than R Ȃ the lines expected if 25.5 than brighter than this limit at z = S7–S12 had been fore - ground emission-line 2–4. The properties of galaxies at this galaxies. The upper faint end of the luminosity function are panels show the currently very uncertain. Since the pop- spectrum of Q1205-30 ulation of bright, Lyman-Break selected to indicate the position galaxies is already (observationally) of the CIV emission ex - very well studied and characterised, pected for AGNs. The progress in the understanding of the absence of CIV high-redshift galaxy population will emission from S7–S12 most likely come from the study of shows that the Ly- emission is powered by galaxies below the spectroscopic limit. star-formation and not by a central AGN. References

Andersen M. I., Hjorth J., Pedersen H., et al., 2000, A&A 364, L54. Cowie L. L., Hu E. M., 1998, AJ 115, 1319. Cristiani S., Appenzeller I., Arnout S., et al., 2000, A&A 359, 489. Fynbo J. U., Møller P., Warren S. J., 1999, MNRAS 305, 849. Fynbo J. U., Thomsen B., Møller P., 2000a, A&A 353, 457. Fynbo J. U., Thomsen B. Møller P., 2000b, The Messenger 95, 32. Fynbo J. U., Gorosabel J., Dall T. H., et al., 2001a, submitted to A&A. Fynbo J. U., Møller P., Thomsen B., 2001a, submitted to A&A. Francis P. J., Woodgate B. E., Warren S. J., et al., 1995, ApJ 457, 490. Fruchter A., Thorsett S. E., Metzger M. R., et al., 1999, ApJ 519, L13. Fruchter A., Hook R., Pian E., 2000a, GCN 757. Fruchter A., Metzger M., Petro L., 2000b, course we need to measure the space the host galaxy of the former (R = 24.6) GCN 701. density of faint Ly-a emitters to similar is more than 20 times brighter than the Haehnelt M. G., Steinmetz M., Rauch M., depths as in the field of Q1205-30 in latter (R = 28) (Holland and Hjorth ApJ 534, 594. several (blank) fields before this con- 1999; Fruchter et al. 1999, 2000a). In Holland S., Hjorth J., 1999, A&A 344, L67. clusion can be drawn. the same way, GRB 000301C and GRB Jensen B. L., Fynbo J. U., Gorosabel J., et 000926 occurred at z = 2.0404 and z = al., 2001, A & A submitted (astro-p h / GRB Host Galaxies 2.0375, but have very different host gal- 0005609). axies. The host galaxy of GRB 000301C Kudritzki R.-P., Méndez R.H., Feldmeier In some widely accepted scenarios, remains undetected down to a detection J.J., et al., 2000, ApJ 536, 19. GRBs are related to the deaths of very limit of R = 28.5 (Fruchter et al. 2000b; Kulkarni S. R., Djorgovski S. G., Rama- prakesh A. N., et al., 1998, Nat 393, 35. massive, short-lived stars and further- Smette et al. 2001; Jensen et al. 2001), Kurk J. D., Röttgering H. J. A., Pentericci L., more gamma-rays are not obscured by whereas the host galaxy of GRB 000926 et al., 2000, A&A 358, L1. dust. Hence, a sample of GRB host is relatively bright (R = 24, Fynbo et al. Roche N., Lowenthal J., Woodgate B., 2000, galaxies may be considered star-for- 2001b). Finally, no host galaxy has MNRAS 317, 937. m a t i o n-selected independent of the been detected for GRB 000131, that Møller P., Warren S. J., 1993, A&A 270, 43. amount of extinction of the rest-frame occurred at z = 4.50, down to a limit of Møller P., Warren S. J., 1998, MNRAS 299, UV and optical emission. R = 25.7 (Andersen et al. 2000). 661. Several high-redshift galaxies have If GRBs indeed trace star-formation, Odewahn S. C., Djorgovski S. G., Kulkarni been localised as host galaxies of these observations indicate that at S. R., et al., 1998, ApJ 509, L5. GRBs. A Lyman-Break type galaxy at these redshifts galaxies covering a Pascarelle S. M., Windhorst R. A., Keel W. re d s hift = 3.42 with an R-band magni- broad range of luminosities contribute C., 1998, AJ 116, 2659. z Smette A., Fruchter A. S., Gull T. R., et al., tude of R = 25.6 (and a prominent Ly-a significantly to the overall density of 2001, ApJ in press (astro-ph/0007202). emission line) was found to be the host star formation. Furthermore, as the ob- Steidel C. C., Giavalisco M., Pettini M., galaxy of GRB 971214 (Kulkarni et al. served R-band flux is proportional to Dickinson M., Adelberger K., 1996, ApJ 1998; Odewahn et al. 1998). GRB the star-formation rate, there must be 462, L17. 990123 and GRB 990510 occurred at 1–2 orders of magnitude more galaxies Steidel C. C., Adelberger, Shapley A. E., et nearly identical redshifts (z ~ 1.6), but at the R = 28 level than at the R = 24 al., 2000, ApJ 532, 170.

26 Discovery of a Bow-Shock Nebula Around the Pulsar B0740-28 H. JONES (ESO Chile); B. STAPPERS (University of Amsterdam); B. GAENSLER (MIT)

Bow-shock nebulae around high- arcsec to the left of the head of the velocity pulsars provide our primary bow-shock), and the measured prop- insight into the interaction between a er motion for the pulsar of 29 mas/yr pulsar and its surrounding environ- westwards (Bailes et al. 1990), con- ment. Specifically, optical observa- firm the bow-shock interpretation for tions of such nebulae allow us to de- the nebula. rive full three-dimensional pulsar ve- All of the pulsars with known bow- locities which are extremely impor- shock nebulae in the optical have tant for the birth rates and evolution high energy loss due to spin-down of pulsars. They can also provide im- and/or high velocities (including the portant information on the density, recent addition of PSR B0740-28). temperature and composition of the However, at the same time, they ex- surrounding ambient medium. Unfor- hibit a diverse range of spin periods, tunately, only a few bow-shock nebu- ages and magnetic field strengths, lae have been discovered, despite highlighting the variety of pulsar there being nearly a thousand pul- winds which can be probed by these sars known from radio surveys. We sources. The recently-discovered have therefore commenced a search a nebula associated with a so-called for pulsar bow-shocks, using the re- “radio-quiet” neutron star further ex- sults to characterise the properties of emplifies the variety of neutron star the associated pulsars, pulsar winds forms known to power nebulae (van and ambient environments. Kerkwijk 2000, ESO press release During the first two nights of this 19/00). programme (January 4 and 5, 2001), Detailed studies of nebular emis- we discovered an optical bow-shock sion, such as that carried out on nebula around the radio pulsar B1957+20 by Aldcroft et al. (1992), B0740-28 using SUSI2 at the NTT. enable a determination of not only Prior to this, only four pulsars were the nature of the shock and the prop- known to power optical bow-shock erties of the ambient ISM, but also of nebulae (see Cordes 1996 and refer- the kinematic distance to the pulsar. ences therein), and only one of these Such a distance determination is es- at southerly declinations (J0437-4715; sential of course in improving our un- Mann, Romani & Fruchter 1999). derstanding of the individual system, Figure 1 shows Ha images of a but also allows an independent test newly-discovered nebula associated of the pulsar distance scale as deter- with PSR B0740-28 (Fig. 1a, top), mined by its dispersion measure. We and of the (previously known) bow- b are currently in the process of doing shock associated with PSR J0437- this for B0740-28, and in doing so, 4715 (Fig. 1b, bottom). Each was taken Figure 1: Two pulsar-powered bow-shock solve another small piece of the puzzle. through the 656/7 filter of SUSI2 and is nebulae imaged in H on 4–5 January 2001 However, the first step towards a 1 arcmin on a side with north to the top at the NTT. (a) The newly-discovered nebu - general understanding of these fasci- and east to left. The B0740-28 image is la associated with PSR B0740-28 (the pul - nating objects is to build a sample of the result of nearly 2 hours of integra- sar is located about 1 arcsec to the left of the them sufficiently large that it canvasses tion while that for J0437-4715 was ob- head of the bow shock) and (b) the bow- the full range of conditions encoun- shock associated with PSR J0437-4715. tered. This is the main aim of our con- tained in a little more than 30 minutes. The star directly behind the shock front is a The faint star directly behind the shock companion to the pulsar which itself is not tinuing survey. front of J0437-4715 is a white dwarf detected in the optical. Images are 1 arcmin We would like to thank the NTT Team companion to the pulsar (which is not on a side with north up, east left. for their assistance with various as- seen in the optical). Both images were pects of the run, particularly Support treated using standard techniques for Astronomers Stephane Brillant and bias and flat-field signature removal, continuum component was not sepa- Pierre Leisy, and telescope operator image registration and then co-addi- rately observed and so has not been Duncan Castex. tion. CCD defects and cosmic rays subtracted in either case. were removed using cross-pixel inter- The images in Figure 1 show the References polation on the final frame rather than nebula associated with PSR B0740-28 filtering through the stack, in order to to have a closed cometary morphology, Aldcroft et al., 1992, ApJ, 400, 638. preserve as much of the faint nebula more like the nebula around PSR Bailes et al., 1990, MNRAS, 247, 322. signal as possible. A residual back- B2224+65 (the “Guitar” nebula; Cordes, Romani & Lundgren, 1993, Nature, ground was subtracted separately from Cordes, Romani & Lundgren 1993) 362, 133. the B0740-28 image, presumably aris- than the open bow-shock of PSR Cordes, 1996, in IAU Colloq .160, p. 393. ing from scattered moonlight on one of J0437-4715. Like the bow-shock asso- Kennel & Coroniti, 1984, ApJ, 283, 710. the nights. Finally, the nebula counts ciated with PSR B2224+65, the B0740- Kulkarni et al., 1992, Nature, 359, 300. were top-hat filtered out and smoothed 28 nebula has a distinct key-hole Mann, Romani & Fruchter, 1999, BAAS, before being recombined with the origi- shape, with the spherical head of the 195, 41.01. nal, to increase the contrast of the ob- shock-front protruding from the fanned Michel, 1982, Rev Mod Phys, 54, 1. ject against the sky background. The tail. The position of the pulsar (about 1 Rees & Gunn, 1974, MNRAS, 167, 1.

27 Young Stellar Clusters in the Vela D Molecular Cloud L. TESTI1, L. VANZI2, F. MASSI 3

1Osservatorio Astrofisico di Arcetri, Florence, Italy; 2European Southern Observatory, Santiago, Chile 3Osservatorio Astronomico di Teramo, Teramo, Italy

It is now well established by means tween these two modes of formation with membership size distribution of the of direct and indirect observations that should occur in the intermediate-mass form g(N) ~ N–1.7 picking stars from a most, if not all, stars are formed in regime, namely 2 <~ M/Mᓃ <~ 15. standard IMP (Bonnell & Clarke 1999). groups rather than in isolation (Clarke, In order to probe this transition, Testi In this view, since massive stars are Bonnell & Hillenbrand 2000). An impor- et al. (1999) recently completed an ex- rare objects compared to low-mass tant result that strongly constrains tensive near infrared (NIR) survey for stars, they will be observed only as theories of massive star and stellar young clusters around optically visible members of large ensembles of stars cluster formation is that the stellar den- intermediate-mass stars (Herbig Ae/Be (clusters), while the detection of an iso- sity of young stellar clusters seems to stars) in the northern hemisphere. The lated high-mass star would have a rel- depend on the mass of the most mas- main result of this survey is that there is atively low probability. As discussed in sive star in the cluster. Low-mass stars a strong correlation between the spec- Testi et al. (2001), there are two obser- are usually found to form in loose tral type of the Herbig Ae/Be stars and vational strategies to provide additional groups with typical densities of a few the membership number of the stellar constraints on which of the two models stars per cubic parsec (Gomez et al. groups around them. Furthermore, is the most appropriate: to expand the 1993), while high-mass stars are found there is compelling evidence that the sample of optically revealed young O within dense stellar clusters of up to 104 most massive stars in their sample are and B stars to increase the statistics, stars per cubic parsec (e.g. the Orion surrounded by denser, not simply more and to search for clusters in complete Nebula Cluster, Hillenbrand & Hart- populous clusters. These findings are samples of luminous embedded mann 1998). To explain this different in qualitative agreement with models sources in giant molecular clouds. The behaviour, it has been proposed that that suggest a causal relationship be- young high-mass isolated objects pre- massive stars may form with a process tween the birth of a massive star and dicted by the random model should be that is drastically different from the stan- the presence of rich stellar clusters. detected in such surveys. However, to dard accretion picture, e.g. by coales- The observed correlation and scatter, properly compare with the models, it is cence of lower mass seeds in a dense however, could also be explained in essential to carry out observations cluster environment. The transition be- terms of random assembling clusters around the target luminous objects

Figure 1: NTT/SOFI near infrared “true-colour” images (J blue, H green, Ks red) of the fields centred on the luminous IRAS sources in the Vela D molecular cloud. The sources are ordered by increasing far infrared (IRAS) luminosity from left to right and top to bottom.

28 ters. We selected the most luminous 3 (Lbol > 10 Lᓃ) IRAS sources in the subsample of Massi et al. (1999, 2000), namely IRS 17, 18, 19, 20 and 21 (following the classification of Liseau et al. 1992), adding IRS 16, a source not included by Liseau et al. (1992) in their final list of protostellar objects associated with the VMR possibly because of its failure in fulfilling some of the rather conservative selection criteria chosen, although lying toward an HII region. Our observations were designed to reach at Ks band a completeness magnitude high enough to be sensitive to all stars more massive than 0.1 Mᓃ in all the clusters. From the observations of Massi et al. (2000), the age of all known young clusters in our sample is less than 1 Myr and the visual extinction much less than 30 mag, as derived from the brightest members of the clusters. Using the methods described in Figure 2: Ks luminosity functions for the observed sample. For each field in the top panel we Testi et al. (1998) and the PMS evolu- show the observed K sLF at the image centre (black histogram) and at the edges (red shad - tionary tracks from Palla & Stahler ed histogram), normalised to the same area. The estimated K LF of each cluster, obtained s (1999), these constraints translate in a by subtracting the “edge” from the “central” K sLFs, are shown in the bottom panels as green histograms. The dotted vertical lines mark the K completeness magnitude. required Ks completeness magnitude of s ~ 17, this figure having been used to design the observing strategy and inte- complete down to at least 0.1–0.2 Mᓃ On this basis, they found no O-type gration times. The expected cluster over a field of view large enough to stars recently having been formed in sizes were estimated from the Testi et cover the expected cluster size. The the VMR, although birth of intermediate- al. (1999) and Massi et al. (2000) sur- NTT/SOFI combination, with the pro- mass stars is in progress. Massi et al. veys, which found an average cluster vided field of view and sensitivity is an (1999, 2000) examined in detail NIR radius of 0.2 pc, corresponding to ~ 1 ideal asset to collect the required data images of the subsample of IRAS pro- arcmin at the distance of the Vela D and settle this fundamental issue. tostellar sources (12) belonging to the cloud. Moreover, due to the reduced extinction D cloud, concluding that most of their The clusters were observed with compared to the optical, the near in- bolometric luminosity arises from single SOFI at the NTT through the J, H and frared bands (especially Ks) are the young stellar objects (or close pairs of Ks broad-band filters and with the large- most effective for this type of studies, in young stellar objects) of intermediate field objective offering an instanta- fact the young clusters are expected to mass embedded in young stellar clus- neous field coverage of ~ 5 arcmin with be at least partially embedded within their parent molecular cloud core. In this paper we report on the results of a study of a complete sample of luminous IRAS sources in the Vela-D giant molecular cloud.

1. The Vela-D Luminous IRAS Sources

Low-resolution observations in the CO(1–0) mm-line of the region of the galactic plane defined by 255° ͨ I ͨ 275°, –5° ͨ b < +5° carried out by Murphy & May (1991) uncovered the existence of an emission ridge in the –1 range 0 ͨ LSR ͨ 15 km s which was promptly dubbed “Vela Molecular Ridge” (VMR). These authors found the molecular gas complex to be made out of four main molecular clouds that they indicated as A, B, C and D, ~ 105 Mᓃ each. Liseau et al. (1992) studied the association of luminous IRAS sources with the VMR and selected among them a complete sample of protostellar objects based on IRAS colours, their Figure 3: (J–H) vs. (H–Ks) colour-colour diagrams for the six observed regions. Sources with - spectral slopes from the near- to the in one arcminute from the field centres are shown in green, sources further away are shown f a r-infrared and the velocity of the in blue. The red line marks the location of main sequence, non reddened stars. The redden - parental molecular gas. They also dis- ing vectors are shown as dashed lines with a red cross every 5 magnitudes. Sources in the cussed the distance of the VMR, con- inner regions show on average redder colours, some of them exhibit an infrared excess, typ - cluding that clouds ACD are likely to be ical of young stellar objects. Only very few sources are affected by extinction exceeding 25 located ~ 700 ± 200 pc from the Sun. mags in the visual.

29 the relative number of sub-stellar objects in the Vela D cluster is smaller than in other cluster forming regions (such as the Orion Nebula Cluster, Hillenbrand & Carpenter 2000), or that most of the sources within the clusters are affected by a visual extinction much lower than the Av = 30 mag value that we have assumed to compute the sensitivity estimates. This latter explanation is in agreement with the near infrared colour-colour and colour-magnitude diagrams shown in Figures 3 and 4, where sources within 1 arcmin from the image centres are shown in green. On average, they show redder colours than sources further away from the centre and a fraction of them display a clear infrared excess, typical of young stellar objects. In the figures, the reddening vectors and the main sequence loci are also displayed, only very few sources are consistent with an extinction greater than 20 mag in the visual. To obtain a quantitative estimate of Figure 4: J vs. (H–Ks) colour-magnitude diagrams for the six observed regions. Sources with - in one arcminute from the field centres are shown in green, sources further away are shown the richness of the detected young stel- in blue. The red line marks the location of main sequence, non reddened stars. The arrows lar clusters, we followed the method de- show the direction of the reddening vectors and its length correspond to a visual extinction of scribed in Testi et al. (1999) and de- 20 mag. rived the richness indicator IC for all six sources. IC gives the “effective” number of cluster members, since it is defined a pixel scale of 0.292 arcsec. We inte- functions of the central regions with as the integral of the radial Ks stellar grated for ~ 15 minutes per filter and those of the image edges. In Figure 2, surface density profile centred on the cluster. Object and sky were alterna- we show such comparison for all fields peak stellar surface density and cor- tively observed to have a good sam- on our sample. In all cases, the pres- rected for the foreground/background pling of the background variations. The ence of a cluster near the centre is stellar density as computed on the images were reduced following the clearly indicated by the excess of lumi- edge of the images. In Figure 5, we standard procedure and the “Special nous sources with respect to the edge show I C as a function of the maximum FlatField” technique. After combining regions. In Figure 2 we also show the stellar mass in each cluster for the high- the dithered frames, the final image Ks completeness magnitudes for each luminosity sources in the Vela D cloud quality is ~ 0.85 arcsec for all fields but field. As expected, our observations are compared with the results of Testi et al. IRS 16, which was observed under deep enough to obtain a complete cen- (1999) toward Herbig Ae/Be systems worse seeing conditions (1.1 arcsec). sus of whole relevant young stellar pop- and the low luminosity sources in the ulation in all clusters. Additionally, we Vela D cloud (Massi et al. 2000). The 2. Results note that the derived cluster KsLFs are mass sensitivity of our NTT survey is all sharply declining above our com- similar to the estimated mass sensitivi- In Figure 1 we show a subsection of pleteness limits, suggesting that either ty of the Herbig Ae/Be survey, while the the NTT/SOFI near infrared “true- colour” images of the fields surrounding the six IRAS sources that we surveyed. Figure 5: IC versus From the images it is immediately clear maximum stellar that we detected groups of very red mass for the lumi - sources in every field, and an increase nous young clusters of the stellar surface density toward the in the Vela D centre, where the luminous IRAS molecular cloud sources are located, is also evident. All (red open circles), clusters are embedded within a diffuse compared with the nebulosity, likely due to cluster mem- results toward low luminosity sources bers light scattered toward the line of in the same cloud sight by the dust in which the younger (cyan open circles) sources are still embedded. In a few and the Herbig cases, IRS 17 and IRS 20 are the most Ae/Be sample (blue evident, we clearly detect in the Ks filled circles). The broad-band filter the line emission from prediction of the collimated jets (see also Massi et al. “random” model 1997 for a narrow- band survey of the (see text) are region) emerging from the inner regions shown in green: median results (sol - of the clusters, confirming the youth of id line), 50% of the the objects. realisations More quantitatively the presence of (dashed lines), an excess of (relatively) bright sources and 98% of the toward the map centres can be made realisations (dotted clear by comparing Ks-band luminosity lines).

30 Massi et al. survey of low-luminosity does not necessarily imply that mas- Hillenbrand L. A. & Hartmann L. W., 1998, sources in the Vela D cloud has a sive stars are formed by coalescence in ApJ, 492, 540. slightly lower sensitivity. The maximum (proto-)cluster environments, but sug- Hillenbrand L. A. & Carpenter J. M., 2000, stellar masses in the Vela clusters have gests that the conditions to form a mas- ApJ, 540, 236. been computed assuming that a frac- sive star are such that this process is Liseau R., Lorenzetti D., Nisini B., Spinoglio L., Moneti A., 1992, A&A 265, 577. tion of the total luminosity ranging from associated with the formation of a Massi F., Lorenzetti D., Vitali F.: “Near 30% to 100% is emitted by the most cluster of (lower-mass) objects. The clus- Infrared H2 imaging of YSOs in Vela massive object. The young IRAS ter could be either the catalyst of high- Molecular Clouds” in Malbet F., Castets A. sources in the Vela molecular cloud mass star formation or a by-product of it. (eds.), Low Mass Star Formation from show the same trend as the Herbig These conclusions should and will be Infall to Outflow – Poster Proceedings of Ae/Be stars: more massive stars are made more firm by combining larger the IAU Symposium n. 182 on surrounded by rich clusters, while low- samples from various surveys toward Herbig-Haro Flows and the Birth of Low mass stars are found in relative iso- different regions. Mass Stars, Observatoire de Grenoble lation. 1997. Massi F., Giannini T., Lorenzetti D. et al., In the same plot, the result of the var- 1999, A&AS 136, 471. ious surveys are compared with the References Massi F., Lorenzetti D., Giannini T., Vitali F., predictions of a “random sampling 2000, A&A 353, 598. model” (as described in Testi et al. Bonnell I. A., & Clarke C. J., 1999, MNRAS, Murphy D. C., May J., 1991, A&A 247, 202. 2001). Our results clearly deviate from 309, 461. Palla F. & Stahler S.W., 1999, ApJ, 525, 772. the prediction of the model, since no Clarke C. J., Bonnell I. A., & Hillenbrand L. Testi L., Palla F., Natta A., 1998, A&AS, 133, A., 2000, in Protostars and Planets IV, massive object is found in isolation, and 81. eds. V. Mannings, A. Boss & S. S. Russell Testi L., Palla F., Natta A., 1999, A&A, 342, all lie above the median predictions of (Tucson: University of Arizona press), p. the model. These results suggest that 515. 151. Testi L, Palla F., Natta A., 2001, in “From there is a physical connection between Gomez M., Hartmann L., Kenyon S. J., Darkness to Light”, eds. T. Montmerle and clusters and high-mass stars. T h i s Hewett R., 1993, AJ, 105, 1927. Ph. André, ASP Conf. Series, in press.

Building Luminous Blue Compact Galaxies by Merging P. AMRAM1 AND G. ÖSTLIN 2

1Observatoire de Marseille, France; 2Stockholm Observatory, Sweden

ABSTRACT Observations of six luminous blue compact galaxies (LBCGs) and two star-forming companion galaxies were carried out with the CIGALE scanning Fabry-Perot interferometer attached to the ESO 3.6-m telescope, tar - geting the H emission line. The gaseous velocity field presents large-scale peculiarities, strong deviations to pure circular motions and sometimes, secondary dynamical components. In about half the cases, the observed rotational velocities are too small to allow for pure rotational support. If the gas and stars are dynamically cou - pled, a possible explanation is either that velocity dispersion dominates the gravitational support or the galax - ies are not in dynamical equilibrium, because they are involved in mergers, explaining the peculiar kinematics. In two cases, we find evidence for the presence of dark matter within the extent of the H rotation curves and in two other cases we find marginal evidence. For most of the galaxies of the present sample, the observed peculiarities have probably as origin merging processes; in five cases, the merger hypothesis is the best way to explain the ignition of the starbursts. This is the most extensive study as yet of optical velocity fields of lu - minous blue compact galaxies.

1. Introduction are terminated by SN winds, but when ing us to achieve two-dimensional ve- gas later accretes back, a new star- locity fields with both high spatial and A Blue Compact Galaxy (BCG) is burst may be ignited; (2) galaxy inter- spectral (velocity) resolutions. BCGs characterised by blue optical colours, actions and (3) collapse of protocloud if are obviously the galaxies for which 2- –21 < MB < –12, an HII-region-like BCGs are genuinely young galaxies. D data are absolutely requested due to emission-line spectrum, a compact ap- Most arguments have been based on the non-axisymmetry of the velocity pearance on photographic sky-survey photometry alone. On the other hand, field around the centre of mass. plates, small to intermediate sizes, high the dynamics of these systems are not The selected BCGs are among the star-formation rates per unit luminosity well explored, still the creation of an en- more luminous ones known in the near- and low chemical abundances (e.g. ergetic event like a sudden burst of star by universe. The galaxies were ob- Searle and Sergent, 1972). Moreover, formation is likely to have dynamical served at the Ha -emission line with the most BCGs are rich in neutral hydro- causes and impacts, complicating the ESO 3.6-m telescope on La Silla. The gen. There is no consensus on the interpretation. exposure times ranged between 24 process(es) that trigger the bursts of To improve our understanding of the minutes and 160 minutes. In Östlin et star formation. Three main scenarios dynamics and the triggering mecha- al. (1999), we presented and described have been proposed to explain it: (1) nisms behind the starburst activity, we the data: Ha images, velocity fields, cyclic infall of cooled gas: Starbursts have obtained Fabry-Perot data allow- continuum maps and rotation curves. In

31 Figure 1: Comparison of the performances, expressed in terms of signal-to-noise ratio vs. the flux of the extended source for two CCDs (with different quantum efficiency) and an AsGa IPCS in a narrow-band imaging application (left) and in a multiplex application like a scanning Fabry-Perot interferometer (right). See text for further details.

Östlin et al. (2001), we will discuss the The instrument CIGALE based at La w w w. o b s . c n r s-m r s . f r / i n t e r f e r o m e t r i e / interpretation of these observations Silla is used both on the ESO 3.6-m instrumentation.html#AsGa). T h e and their implications on the dynamics, and on the Marseille 36-cm telescope. IPCS, with a time resolution of 1/50 the mass distribution and the triggering On the Marseille telescope, CIGALE second and zero readout noise makes mechanism behind the starbursts. has a wide field of view (40 arcmin it possible to scan the interferometer In this paper, we give a flavour of the square) and is used to make a kine- rapidly, avoiding problems with varying observations and provide some de- matical deep Ha survey of gaseous sky transparency, airmass and seeing scriptions of two galaxies: ESO 350- emission regions in the Milky Way and during long exposures; and thus has IG38 and ESO 338-IG04. Then, we in the Magellanic Clouds (see for in- several advantages over a CCD for this point out the hot spots of the sample stance Georgelin et al., 2000). On the application. Figure 1 compares the total and suggest an interpretation. 3.6-m, CIGALE has a one hundred efficiency of a CCD and an IPCS used times smaller field (~ 4 arcmin square, in narrow-band imaging (left) and in 2. A Scanning Fabry-Perot Inter- providing a pixel size of 0.91 arcsec). multiplex application (right). It clearly ferometer on the ESO 3.6-m The spectral scanning step depends on shows that at low intensity level, the the interferometer used, it was of 4.8 signal-to-noise ratio is much higher with The instrument CIGALE attached to km s–1 for the present study. A full de- an IPCS than with a CCD in the case of the Cassegrain focus of the ESO 3.6-m scription of the instrument CIGALE is multiplex observations as for instance was used for the observations. CIGALE given in a previous paper in T h e scanning Fabry-Perot interferometry. is basically composed of a focal reducer, Messenger (Amram et al., 1991). By Figure 1 has been plotted from a simu- a scanning Fabry-Perot interferometer, the way, a previous discussion in The lation of one hour exposure time on an an interference filter and an IPCS (Im- Messenger on BCGs (Infants of the 8-metre telescope, for a pixel size of age Photon Compting System). CI- Universe?) including ESO 338-I G 0 4 0.25 arcsec square, 48 scanning chan- GALE (for Cinématique des GALaxiEs) can be found in Bergvall & Olofsson nels, transmissions of: 80% for the at- is a visiting instrument belonging to (1984). Data reduction was performed mosphere; 64% for the 2 mirrors of the Marseille Observatory, mounted for the using the http available ADHOCw soft- telescope; 70% for the interference fil- first time on the ESO 3.6-m in 1991 and ware (http://www-ob s . c n r s -mrs.friadhoc/ ter; 80% for the optics and 90% for the commissioned on the average for one adhoc.html); a complete data-reduction Fabry-Perot (Gach et al., 2001). The run a year since that time. (In fact, a procedure is given for instance in data presented here were obtained in Fabry-Perot etalon plus an image tube Amram et al. (1991, 1996). In August August 1995 with the old system; new has been mounted on the ESO 3.6-m 1999 and in September 2000, two ma- observations were performed for anoth- by the Marseille team for the first time jor upgrades were realised on CIGALE: er 15 BCGs, extending down to fainter in 1979, see Marcelin et al., 1982). With the data-acquisition system and the re- luminosities in 1999 and 2000 with the the efficient cooperation of the 3.6-m ceptor were removed. (1) Two hundred new system, this will allow us to obtain ESO team and of the astro-workshop, kg of electronics were substituted by a a more comprehensive picture on the using the facilities of the telescope, the Matrox board allowing the acquisition of evolution of BCGs. instrument is easily mounted during the frames 1024 ´ 1024 px2 at high fre- daytime of the first observing night. quency and the computation in real 3. Results: Description of Two Several replica of this instrument exist time of the events on each frame. BGCs throughout the world, one of them is (2) The 20-year old Thomson IPCS was presently extensively used on the OHP replaced by a new AsGa IPCS. The 3.1 ESO 350-IG38 (Haro 11) 1.93-m telescope (Observatoire de s e m i-conductor photocathode A s G a Haute-Provence, France) to provide a (built by Hamamatsu) of the new de- The morphology of this object is com- sample of about 200 velocity fields of tector offers a d.q.e. five times higher plex. The three bright starburst nuclei nearby galaxies (the GHASP’s project (25% instead of 5% for the old are composed of numerous individual – Gassendi HAlpha SPiral galaxies sur- Thompson IPCS). The output of the super star clusters (MV ͧ –15, Östlin, vey – see h t t p : / / w w w-o b s . c u r s m r s . f r / AsGa tube is coupled by optical 2000) as it can be seen for an i n t e r f e r o m e t r i e / G H A S P / g h a s p . h t m l ). fibers to a 1024 ´ 1024 CCD. (http:// HST/WFPC2 image (Fig. 2 – left,

32 Figure 2: ESO 350-IG38. North is up, East is left. (Left): HST/WFPC2 V+R (F606W) broad-band image. The scale of the box is 24 arcsec Ȃ square. (Right): R-band CCD image. Note the irregular morphology at all isophotal levels. The faintest visible structures are mR 26 mag/arc - sec2. The size of the field is 1 by 1 arcminute, corresponding to 23 23 kpc.

8 Malkan et al. 1998). The properties of void of cool gas (MHI < 10 Mᓃ = de- extension in the south-east is evident. the central regions bear a close resem- tection limit) despite a very high star- This may be a tidal tail in development blance with the classical colliding formation rate (Bergvall et al., 2000). or most likely the remnants of such. galaxies NGC 4038/4039 as seen in This suggests that either the starburst The Ha line profiles of this galaxy are HST data (Whitmore and Schweizer is about to run out of fuel or the gas is broad, up to FWHM = 270 km/s, and 1995). This galaxy is the most massive to a large extent in molecular and have non-gaussian shape. This sug- in the whole sample and also has the ionised form as suggested by calcula- gests that two or more non-virialised highest star-formation rate (~ 20 tion. The outer isocontours are distort- components may be present. In the Mᓃ/yr). The inferred supernova (SN ed out to very faint isophotal levels (mR centre, double peaked lines are present type II) frequency is one every 7 years. Ȃ 26 mag/arcsec 2) on broad-band im- consistent with the presence of a Surprisingly, the galaxy is very lumi- ages (Fig. 2 – right). This is still visible counter rotating disk or high velocity nous in IR but seems to be rather de- but less extended in the Ha line demon- blobs. This galaxy presents a strong strating that the central velocity gradient and large- light originates scale distortions (see Fig. 4 – up). If we mainly in stars. do not decompose the velocity field in Hence, the large- several components, the rapid increase scale distribution of the rotation curve is followed by a of stars in this keplerian decrease. Alternatively, if we galaxy is highly disentangle two possible components, asymmetric. At all the main one (in terms of intensity and isophotal levels an extension) provides a low-velocity amplitude flat rotation curve while the second one gives a high-velocity amplitude but only in the central parts of the Figure 3: Ve l o c i t y galaxy (see Fig. 5 – left). These prop- fields for ESO 350- erties indicate that the centre is not dy- IG38 (up) and ESO namically relaxed, while the outer ve- 3 3 8-IG04 (bottom). locity field shows a very slow rotation. The two galaxies The estimated stellar mass density have the same pixel exceeds by far what can be supported size (0.91 arcsec/ px) and scale on by the observed amount of rotation the image; the hori - (G 30 km/s). Thus the galaxy is either zontal size of the not in equilibrium, or it is not primarily box is 42 arcsec. supported by rotation. These proper- ESO 350-IG38: The ties strongly suggest that the starburst velocity amplitude has been triggered by a merger ranges linearly from process. violet (6220 km s–1) –1 to white (6300 kms ). 3.2 ESO 338-IG04 E S O 3 3 8-IG04: The velocity amplitude (Tololo 1924-416) ranges linearly from violet (2750 km s–1) HST observations of this well-known to white (2850 km starburst (see Fig. 4 – left, Östlin et al., s–1). 1998) have revealed that in addition to

33 Figure 4: ESO 338-IGO4. North is up, East is left. (Left): HST/WFPC2 true composite colour BVI-image. The scale of the box is 28 arcsec square. (Right): R-band CCD image. Note the irregular morphology at all isophotal levels. The faintest visible structures are R Ȃ 26 mag/arc - sec2. The size of the field is 60 arcsec 53 arcsec, corresponding to 11 9.5 kpc. many young compact star clusters, it side in the tail? The colour of the tail vealing large-scale asymmetries in the contains a system of intermediate-age suggests that it has a distinctly different distribution of stars. In most cases we (~ 2 Gyr) globular clusters, a fossil of a stellar population from the rest of the see clear signatures of merging/inter- previous dramatic starburst event. This galaxy. The most likely is that the tail is action. The young burst population galaxy has a companion at a projected a remnant of a merger and that projec- dominates the integrated optical lumi- distance of 70 kpc to the south-west. tion effects prevent us from seeing the nosities, while only contributing 1 to 5% There is a 5 kpc long tail towards east, true velocity amplitude. The part of the of the total stellar mass, which ranges and large-scale isophotal asymmetries galaxy on the eastern side which is not from a few times 108 to more than 1010 2 down to the R = 26 mag/arcsec level in the tail, does not emit strongly in Ha , Mᓃ. The mass is dominated by an old- (Fig. 4 – right). At fainter levels, the hence we have no information on its er underlying population and the inte- morphology becomes more regular, but kinematics. Where the tail meets the grated (burst + old population) mass to a boxy shape remains on the western starburst region, we see an increased light ratios are M/LV ~ 1. side. The tail has much bluer colours Ha velocity dispersion, perhaps due to The velocity fields, in spite of the high than the rest of the galaxy outside the a shock. This coincides with the loca- signal-to-noise ratio of the emission starburst region, signifying a younger tion of the perpendicular dynamical lines, appear irregular and distorted, stellar population. It contains stellar component discussed above. The com- except for the two companion galaxies clusters, and cannot be explained by a panion is probably too far away for tidal included in the sample. As the S/N is purely gaseous tail. forces to influence the internal gas high, these irregular isovelocities This galaxy has an almost chaotic kinematics to such an extent that it may should be considered as real. Some ro- velocity field (see Fig. 3 – bottom), with have caused the starburst and the pe- tation curves appear strange or strong gradients and an extended tail culiar velocity field. non-uniform; this indicates that the sim- with little internal velocity structure. Radio interferometric observations ple assumptions of a regular warp-free Approximately 10 arcsec east of the (Östlin et al., in preparation) reveal that disk and pure circular motion around centre, just at the border of the central the galaxy is embedded in a very large the centre are not valid in general. The star-forming region, there is a velocity HI cloud, more than 7 arcminutes estimated dynamical masses ranges component whose kinematical axis is across (corresponding to 80 kpc at the from a few times 108 to a few times 109 perpendicular to the photometric major distance of ESO 338-IG04). The HI Mᓃ. Secondary components have axis of the galaxy (Figs. 5 and 6 in cloud has irregular morphology with been detected in several cases. Östlin et al., 1999). The radial light dis- two main components and no single In about half the cases, the observed tribution indicates that the observed axis of rotation. ESO 338-IG04 appears rotational velocities are too small to al- mean rotational velocity (if a such is at to be located in the eastern HI cloud. low for pure rotational support. A possi- all meaningful to define in view of the ir- The companion is detected in HI but ble explanation is that velocity disper- regular velocity field) cannot support lies further away. Although we cannot sion dominates the gravitational sup- the system gravitationally. The tail has exclude that the starburst in ESO port. This is consistent with the ob- almost no velocity gradient with respect 3 3 8-IG04 is triggered by interaction served line widths (DHa = 35 to 80 to the centre (Du ͨ 10 km/s). On the with the companion, a merger appears km/s), but does not explain the strange other hand, the western half of the more likely in view of the complex ve- shape of many of the rotation curves galaxy has a strong gradient and an im- locity field and the non-rotating arm. (see Fig. 5). Another possibility is plied rotational velocity of 80 km/s at a that the galaxies are not in dynamical distance of 3 kpc from the centre. This 4. Discussion on the Whole equilibrium, e.g. because they are in- is identical to the rotational velocity in Sample volved in mergers, explaining the pe- the companion ESO 338-IG04B that culiar kinematics. It is also possible that has an equal photometric mass. 4.1 Photometric and kinematical gas and stars are dynamically decou- Hence, the western part of the galaxy analysis pled and the Ha velocity field does not shows about the expected level of ve- trace the gravitational potential. A way locity difference with respect to the cen- The morphologies of the BCGs pre- to distinguish between these alterna- tre for rotational support to be possible. sent strong large-scale asymmetries tives would be to obtain the rotation But what is happening at the eastern down to the faintest isophotal levels, re- curve and velocity dispersion for the

34 Figure 5: Mass model for ESO 350-IG38 (left) and ESO 338-IG04 (right). The solid line is the photometric rotation curve for the disk compo - nent; and the green shaded area is the allowed range based on the uncertainty in M/L. The filled circles show both sides’averaged observed rotation curves, the error bars represent mainly the deviation to pure circular motions. ESO 350-IG38 (left): the blue open symbols show the rotation curve based on the decomposed velocity field. The blue open circles correspond to the main component; the secondary component, which is counter rotating is shown by the blue open squares. The maximum disk model (in order to provide an upper limit of the (M/L)disk) has been computed on the main component. The dash-dot red line is the stellar disk component from the mass model and the crosses is the halo component (red crosses). As the halo component is very faint, the total model (disk + halo) matches the stellar component. ESO338- IG04 (right): The blue open symbols show both sides of the rotation curve. The blue open circles correspond to the receding side while the approaching one is shown by the blue open squares. Due to the disagreement between the approaching and the receding sides, the aver - aged rotation curves do not represent the axisymmetric potential well of the galaxy, the instrumental accuracy of the velocity being 2–3 km/s at high signal-to-noise ratio. Furthermore, no dynamical model can be plotted for this galaxy. stellar component. In two cases, we and star-formation histories (Östlin et sions in the gaseous component of the find evidence for the presence of dark al., 2001). The photometric total mass- galaxies. Furthermore, the gas be- matter within the extent of the Ha ro t a - es of the disk and of the burst have comes a bad tracer of the potential well tion curves, and in two other cases we been obtained by integrating the lumi- but a good tracer of starburst activity. find marginal evidence. Indeed, even nosities with the M/L values in the al- Stars should be a better tracer of the with a maximum disk model, the rota- lowed range. The dynamical M/L of the potential well but their observations tion curves computed from the surface disk have been computed using a present difficulties due to the weakness brightness profiles cannot fit the ob- t w o-component (exponential disk + of the continuum emission and absorp- served rotation curve without an addi- dark halo) best-fit mass model or max- tion lines in this class of galaxies. tional dark halo component. imum disk model (Carignan, 1985). In When considering the kinematics most cases, the M/L obtained from the and morphologies of this sample of lu- 4.2 Spectral Evolutionary Models dynamical mass models are lower than minous BCGs, we are led to the con- and Dynamical Analysis the ones obtained from the spectral clusion that dwarf galaxy mergers are evolutionary synthesis models. T h i s the favoured explanation for the star- Spectral evolutionary synthesis mod- means that the dynamical mass is un- bursts. The dynamics of the studied els in combination with colour profiles in der-evaluated, the velocity fields being galaxies fall into two broad classes: one the optical and near infrared are used disturbed by non-circular motions with well behaving rotation curves at to estimate the mass to light ratios or/and that the galaxies are not rota- large radii and one with very perturbed (M/L) of the galaxies (the models are tionally supported. dynamics. This may indicate a distinc- described in Bergvall and Rönnback, tion of the fate of these galaxies, once 1995). Photometric mass distributions 4.3 Merging in Process the starbursts fade. Alternatively, de- were derived by integrating the lumi- pending on the state in which we see nosity profiles for the disk and burst For this sample of LBCGs, two kinds the interaction/merger, we will detect components and using their corre- of perturbation are probably in compe- more or less chaotic velocity maps. sponding M/L values. The photometric tition: (1) merging processes distorting mass distributions are further com- the velocity fields at large scale and (2) 5. Perspectives pared to dynamical mass models (see winds driven by starbursts strongly dis- Fig. 5). From B-band surface bright- turbing the circular motion of the Dwarf galaxies are very common ob- ness photometry, an exponential disk ionised gas at smaller scales. Merging, jects in the universe; compact galaxies has been separated from the burst interaction and wind driven by super- (CGs) are frequent and blue compact by extrapolation of the profile of the novae can induce non-circular motions galaxies (BCGs) are not rare. At high outer regions where the contribution to of high amplitude to the gas. Then, due redshift, gas-rich dwarf galaxies and the burst is marginal. The photometric to dissipative collisions, the gas is prob- perhaps also CGs even can be the M/L of the disks and of the bursts have ably decoupled from the rotation of the building blocks of the larger galaxies, been evaluated by matching the ob- stars which constitute the dominant merge and form stars during episodic served colours to extended sets of mass of the disk. Stars and gas are processes in their evolution. The lumi- models with different IMF, mass limit then decoupled due to dissipative colli- nous blue compact galaxies (LBCGs)

35 ESO VACANCY

Applications are invited for the position of Head of the Office for Science Assignment: A strong research-oriented scientific staff with diverse expertise is essential to fulfil ESO’s mandate of providing and maintaining international competitive observatories for its community of users. In this context the Head of the Office for Science will be responsible to the Director General for the science policies, and the main tasks will be

• to foster science across the entire Organisation and increase the exchange between ESO Chile and Garching; • to support the efforts of ESO staff to conduct science at a high level; • to provide a reference point for young scientists; • to pursue actively the Fellowship, Studentship, Visitor programmes and carry out scientific exchanges with the European community through a strong and attractive Workshop Programme; • to be the ESO representative at the International Research School recently created in the Munich area.

Experience and knowledge: Candidates must have an outstanding record of achievement in one or more areas of modern astrophysics and preferably be widely experienced in the use of large ground-based optical and/or radio telescopes. Excellent communication and management skills and a strong sense of team spirit are essential. Duty station: Garching near Munich, Germany, with regular trips to Chile. Starting date: 1 September 2001 or later. We offer an attractive remuneration package including a competitive salary (tax free), comprehensive social ben- efits and financial help in relocating your family. The initial contract is for a period of three years with the possibility of a fixed-term extension. Serious consideration will be given to outstanding candidates willing to be seconded to ESO on extended leaves from their home institutions. Either the title or the grade may be subject to change according to education and the number of years of experience. If you are interested in working in areas of front-line science and technology in a stimulating international research environment, please send us your CV (in English language) and the ESO Application Form (to be obtained from the ESO Home Page at http://www.eso.org/) and four letters of reference before 15 May 2001. For further information, please consult the ESO Home Page.

have luminosities and properties similar Acknowledgements Bergvall N., Rönnback J., 1995, MNRAS to galaxies seen at intermediate red- 273, 603. shifts (e.g. Guzman et al., 1997). We would like to thank our collabora- Carignan C., 1985, ApJ 299, 59. Moreover, they are among the least tors Nils Bergvall, Josefa Masegosa, Östlin G., in “Massive Stellar Clusters”, A. massive/luminous galaxies that are Jacques Boulesteix and Isabel Lancon and C. Boily Eds., 2000, ASP Conf. Ser. 211, p. 63. possible probes also at high redshifts. Marquez for their contributions to this Östlin G., Amram P., Masegosa J., Bergvall The sample of LBCGs presented here work. Jean-Luc Gach from Marseille N., Boulesteix J., 1999, A&AS, 137, 419. is not representative of classical BCGs Observatory is thanked for providing Östlin G., Amram P., Masegosa J., Bergvall but can be regarded as a reference the computations for Figure 1. Nils N., Boulesteix J. and Marquez I. 2001, to sample for high redshift LBCGs, for Bergvall is thanked for his comments be submitted. which, moreover, observations present and his careful reading of this paper. Östlin G., Bergvall N., Rönnback J., 1998, a serious bias. Even if there are evolu- The 3.6-m ESO and the astroworkshop A&A 335, 85. tionary processes that make distant teams are thanked for providing effi- Gach J.L. et al., 2001, in preparation for galaxies unique, the evolutionary histo- cient support during the set-up of the in- ApJS. ry of a galaxy is not only a time-de- strument. Guzman R., Gallego J., Koo D.V. et al., pendent parameter but it may also 1997, ApJ. 489, 559. strongly depend on the environment. Georgelin Y.M., Russeil D., Amram P. et al., Furthermore, LBCGs can be thought of References 2000, A& A 357, 308. as nearby sites which mimic galaxy in- Malkan, Gorjian and Tam, 1998, ApJS 117, teraction, merging and the star trigger- Amram P., Boulesteix J., Georgelin Y.M. et 25. Marcelin, M.; Boulesteix, J.; Courtes, G.; ing in higher-density environments of al., 1991, The Messenger, 64, 44. Amram P., Balkowski C., Boulesteix J. et al., Milliard, B. Nature, vol. 297, May 6, 1982, the young universe. 1996, A& A, 310, 737. p. 38–42. These local dwarf galaxy mergers Bergvall N. and Olofsson K., 1984, The Searle L. and Sargent W.L.W., 1972, ApJ. may be the best analogues of hierar- Messenger 38, 2. 173, 25. chical build-up of more massive galax- Bergvall N., Masegosa J., Östlin G. and Whitmore B., Schweizer F., 1995, AJ 109, ies at high redshifts. Cernicharo J., 2000, A&A 359, 41. 960.

36 O T H E R A S T R O N O M I C A L N E W S EIS Data on the Chandra Deep Field South Released The purpose of this note is to an- image ever taken with a total integration measured within a 2 ´ FWHM aperture. nounce that the ESO Imaging Survey time of one million seconds. The data Given the general interest on this programme has released a full set of obtained by EIS include J and Ks infra- field, fully calibrated images and asso- optical/infrared data covering the so- red observations of an area of 0.1 ciated weight maps as well as source called Chandra Deep Field South square degree nearly matching the lists have been made available world- (CDF-S) rapidly becoming a favoured Chandra image down to JAB ~ 23.4 and wide on February 16, 2001 and March target for cosmological studies in the KAB ~ 22.6 and UU’BVRI optical obser- 5, 2001. The data can be requested southern hemisphere. The field was vations over 0.25 square degree, through the URL “http://www.eso.org/ originally selected for deep X-ray ob- matching the XMM field of view, reach- eis”. A first impression of the data is giv- servations with Chandra and XMM. The ing 5 s limiting magnitudes of U’AB = en by the colour composite image former have already been completed pro- 26.0, UAB = 25.7, BAB = 26.4, VAB = shown below which combines U, R, ducing the deepest high-resolution X-ray 25.4, RAB = 25.5 and IAB = 24.7 mag, as and J +K infrared images. A N N O U N C E M E N T S

ESO Studentship Programme 2001

The European Southern Observatory research student programme aims at providing the opportunities and the facilities to en- hance the post-graduate programmes of ESO member-state universities by bringing young scientists into close contact with the instruments, activities, and people at one of the world’s foremost observatories. For more information about ESO’s astronomical research activities please consult Research Projects and Activities (http://www.eso.org/science/ or http://www.eso.org/projects/).

Students in the programme work on an advanced research degree under the formal tutelage of their home university and de- partment, but come to either Garching or Vitacura-Santiago for a stay of normally up to two years to conduct part of their studies under the supervision of an ESO staff astronomer. Candidates and their national supervisors should agree on a research project together with the potential ESO local supervisor. This research programme should be described in the application and the name of the ESO local supervisor should also be mentioned. It is highly recommended that the applicants start their Ph.D. studies at their home institute before continuing their Ph.D. work and developing observational expertise at ESO.

The ESO studentship programme comprises about 14 positions, so that each year a total of up to 7 new studentships are available either at the ESO Headquarters in Garching or in Chile at the Vitacura Quarters. These positions are open to students en- rolled in a Ph.D. programme in the ESO member states and exceptionally at a university outside the ESO member states.

The closing date for applications is June 15, 2001.

Please apply by using the ESO Studentship application form available on-line (http://www.eso.org/gen-fac/adm/pers/forms/).

European Southern Observatory Studentship Programme Karl-Schwarzschild-Str. 2 85748 Garching bei München Germany [email protected]

P E R S O N N E L M O V E M E N T S

SLUSE, Dominique (B), Student International Staff WITTKOWSKI, Markus (D), VLTI Astronomer WOODS, Paul (UK), Student (January – March 2001) DEPARTURES ARRIVALS EUROPE EUROPE TORWIE-SCHMER, Claudia (D), Accounting Assistant BRANDNER, Wolfgang (D), Adaptive Optics Instrument Scientist CHATZIMINAOGLOU, Evanthia (GR), Associate EIS CONAN, Rodolphe (F), Associate Local Staff CONDORELLI, Livio (I), VLT Software System Manager (December 2000 – March 2001) for UNIX computers DEMARCO, Ricardo (RCH), Student EKVALL, Jenny (S), Associate ARRIVALS HACKENBERG, Wolfgang (D), Laser Specialist ALARCON, Hector, Telescope Instruments Operator, KORNMESSER, Martin (D), Graphics Designer Paranal LAMPERSBERGER, Brigitte (D), Associate GODOY, Angelina, Bilingual Secretary, Santiago MARCHETTI, Enrico (I), Associate HERRERA, Cristian, Telescope Instruments Operator, NELSON, Martine (UK), Division Secretary/Assistant Paranal TRIPICCHIO, Alfredo (I), Associate MORALES, Alex, Maintenance Engineer, Paranal VAN BOEKEL, Roy (NL), Student ROJAS, Manuel, Telescope Instruments Operator, Paranal VANZELLA, Eros (I), Student DEPARTURES CHILE ABUTER, Roberto, Software Engineer, Paranal BAUVIR, Bertrand (B), Software Engineer ALVAREZ, Roberto, Bilingual Secretary, Paranal HOUSEN, Nico (B), Software Engineer BROWNING, Victoria, Paranal Administrator, Paranal HUBRIG, Swetlana (D), Operations Staff Astronomer FIGUEROA, Edgardo, Precision Mechanics, La Silla MOREL, Sébastien (F), VLTI Instrument Operator MARTIN, Gabriel, Telescope Instruments Operator, La Silla SCHÜTZ, Oliver (D), Student RAHMER, Gustavo, Optical Detector Engineer, Paranal

ESO, the European Southern Observa- SCIENTIFIC PREPRINTS tory, was created in 1962 to “… establish and operate an astronomical observatory (January – March 2001) in the southern hemisphere, equipped with powerful instruments, with the aim of 1408. E. Scannapieco and T. Broadhurst: stellar Activity of the Be Star furthering and organising collaboration in Linking the Metallicity Distribution of µCentauri. III. Multiline Nonradial astronomy …” It is supported by eight Galactic Halo Stars to the Enrichment Pulsation Modeling. A&A. countries: Belgium, Denmark, France, History of the Universe. ApJ Letters. 1411. B. Paltrinieri, F.R. Ferraro, F. Paresce Germany, Italy, the Netherlands, Sweden 1409. M. Rejkuba: Deep VLT Search for and G. De Marchi: VLT Observations and Switzerland. ESO operates at two Globular Clusters in NGC 5128: of the Peculiar Globular Cluster NGC sites. It operates the La Silla observatory Color-Magnitude Diagrams and 6712. III. The Evolved Stellar Popu- in the Atacama desert, 600 km north of Globular Cluster Luminosity Func- lation. AJ. Santiago de Chile, at 2,400 m altitude, tion. A&A. 1412. B. L. Jensen et al.: The Afterglow where several optical telescopes with di- 1 4 1 0 . Th. Rivinius, D. Baade, S. Št e f l , of the Short/Intermediate-Duration ameters up to 3.6 m and a 15-m submil- R.H.D. Townsend, O. Stahl, B. Wolf Gamma-Ray Burst GRB 000301C: A limetre radio telescope (SEST) are now in and A. Kaufer: Stellar and Circum- Jet at z = 2.04. A&A. operation. In addition, ESO is in the process of building the Very Large Telescope (VLT) on Paranal, a 2,600 m high mountain approximately 130 km south of Antofagasta, in the driest part of the Atacama desert. The VLT consists of four 8.2-metre and three 1.8-metre telescopes. These telescopes can also be Contents used in combination as a giant interferometer (VLTI). In April 1999, the first 8.2-me- J. Alves, C. Lada and E. Lada: Seeing the Light Through the Dark . . 1 tre telescope (called ANTU) started regular operation, and also the three other ones (KUEYEN, MELIPAL and YEPUN) TELESCOPES AND INSTRUMENTATION have already delivered pictures of excellent quality. Over 1200 proposals are made each year for the use of the ESO A. Gilliotte: La Silla Telescope Status, a Great Achievement on telescopes. The ESO Headquarters are Image Quality Performances ...... 2 located in Garching, near Munich, Ger- many. This is the scientific, technical and A. Gilliotte: Image Quality Improvement of the 2.2-m ...... 4 administrative centre of ESO where tech- O. Hainaut and the NTT Team: NEWS from the NTT ...... 7 nical development programmes are carried out to provide the La Silla and Paranal H. Jones: 2p2 Team News ...... 9 observatories with the most advanced in- H. Jones, A. Renzini, P. Rosati and W. Seifert: Tunable Filters and struments. There are also extensive astronomical data facilities. In Europe ESO Large Telescopes ...... 10 employs about 200 international staff members, Fellows and Associates; in Chile about 70 and, in addition, about 130 REPORTS FROM OBSERVERS local staff members. J. Alves, C. Lada and E. Lada: Seeing the Light Through the Dark The ESO MESSENGER is published four times a year: normally in March, June, (Continued from page 1) ...... 15 September and December. ESO also pub- F. Comerón and M. Fernández: Strong Accretion and Mass Loss lishes Conference Proceedings, Pre- prints, Technical Notes and other material Near the Substellar Limit ...... 21 connected to its activities. Press Releases J.U. Fynbo, P. Møller, B. Thomsen: Star Formation at z = 2–4: inform the media about particular events. For further information, contact the ESO Going Below the Spectroscopic Limit with FORS1 ...... 24 Education and Public Relations Depart- H. Jones, B. Stappers, B. Gaensler: Discovery of a Bow-Shock ment at the following address: Nebula Around the Pulsar B0740-28 ...... 27 EUROPEAN L. Testi, L. Vanzi, F. Massi: Young Stellar Clusters in the Vela D SOUTHERN OBSERVATORY Karl-Schwarzschild-Str. 2 Molecular Cloud ...... 28 D-85748 Garching bei München P. Amram and G. Östlin: Building Luminous Blue Compact Galaxies Germany Tel. (089) 320 06-0 by Merging ...... 31 Telefax (089) 3202362 [email protected] (internet) URL: http://www.eso.org OTHER ASTRONOMICAL NEWS http://www.eso.org/gen-fac/pubs/ messenger/ EIS Data on the Chandra Deep Field South Released ...... 37 The ESO Messenger: Editor: Marie-Hélène Demoulin Technical editor: Kurt Kjär ANNOUNCEMENTS

Printed by ESO Vacancy. Head of the Office for Science ...... 36 J. Gotteswinter GmbH Buch- und Offsetdruck ESO Studentship Programme 2001 ...... 38 Joseph-Dollinger-Bogen 22 Personnel Movements ...... 38 D-80807 München Germany ESO Vacancies. A Challenge for Astronomers, Engineers in the Field of Software, Electronics and/or Mechanics ...... 39 ISSN 0722-6691 Scientific Preprints (January – March 2001) ...... 40