Telescopes and Instrumentation

ALMA Extends to 15-kilometre Baselines: Submillimetre Science down to 20-Milliarcsecond Resolution

Catherine Vlahakis1,2 operate the world’s largest millimetre/ antenna stations on the Chajnantor site Leonardo Testi2 submillimetre array, located on the that have been built for this purpose, Paola Andreani2,3 Chajnantor Plateau in the north of Chile stretching out to almost ten kilometres at an elevation of 5000 metres above sea from the centre of the array. The 2014 level. ALMA, inaugurated in 2013, has a ALMA Long Baseline Campaign (LBC) 1 Joint ALMA Observatory, Santiago, Chile full complement of 66 antennas. The Call carried out the first ever tests using these 2 ESO for Proposals for Early Science­ Cycle 3 distant antenna stations; it was the first 3 ESO ALMA Regional Centre has recently closed, with a record num- time the whole observatory infrastructure ber of proposals received. had been used for antenna separations above about three kilometres. The LBC From September to late November A key new capability offered in Cycle 3 tests were carried out from September to 2014, the Atacama Large Millimeter/ is observing at high angular resolution November 2014 and consisted of exten- submillimeter Array (ALMA) carried out using baselines of up to ten kilometres. sive testing with 23 antennas on long a Long Baseline Campaign to test and This mode was tested and verified during baselines of up to 15 kilometres, together verify baselines up to ~ 15 kilometres a dedicated campaign held in the last with additional antennas on short (about for the first time. The tests were suc- quarter of 2014. A summary of this cam- 300 metres) baselines in order to provide cessful, demonstrating the power of paign — the 2014 ALMA Long Baseline the short spacings that are necessary ALMA to produce submillimetre-wave Campaign — is presented here. Activities for imaging extended objects. A total of imaging at resolutions of tens of milli- continue to extend and upgrade the capa- 22–36 antennas were employed for the arcseconds. Science Verification obser- bilities, including new receivers, digital LBC tests, with the number depending vations of five targets, including an electronics, data handling and analysis on observing band; most tests were car- asteroid, a protoplanetary disc, an software, and new observing modes ried out in Band 3 (100 GHz). A paper evolved star and a gravitationally lensed (such as millimetre-wave very long base- (ALMA Partnership, Fomalont et al., 2015; high-redshift galaxy, were obtained. line interferometry), with the goal of main- hereafter ALMA I) provides details of the The results demonstrate that the scien- taining ALMA at the forefront of astro- technical and calibration aspects of the tific leap for which ALMA was planned physical research for many decades to LBC, and a brief summary is given here. has been made. A brief overview of the come. 2014 Long Baseline Campaign and its ALMA calibration normally uses the phase scientific results are presented. referencing observing mode, whereby The 2014 ALMA Long Baseline Campaign short scans of the science target and a nearby quasar calibrator target are alter- ALMA is a global project with partners Imaging at resolutions of tens of milliarc- nated over the duration of the observation, in Europe, North America (USA and seconds (mas) is a major goal of ALMA. such that the phase measured on the ­Canada) and East Asia (Japan, Taiwan This requires baselines of up to almost calibrator can be transferred to the tar- and the Republic of Korea), to build and 16 kilometres, achieved using the distant get. The accuracy with which this can be )/ESO wingsforscience.com Clem & Adri Bacri-Normier ( Bacri-Normier Adri & Clem

Figure 1. The ALMA Chajnantor site from the air in early 2013.

2 The Messenger 160 – June 2015 achieved was a critical factor for the out- term phase root mean square (rms) is the peak/image rms of a factor of two come of the long baseline observations, < 30 degrees. The cycle time needed for to six compared to imaging with phase and was addressed with a test strategy phase referencing calibration to produce referencing alone. For HL Tau, the that included: a) establishing the phase good image quality was found to be improvement was up to a factor of two. coherence of the long baseline array by 60–90 seconds in almost all cases. The For SDP.81, self calibration was not pos- carrying out test observations of quasars; most reliable method of determining sible at all. Conversely, for the quasar b) determining the observing strategy whether phase conditions are suitable for 3C 138, which has emission that is domi- needed to achieve good imaging results; carrying out a given observation of a nated by a strong compact core and other and c) carrying out Science Verification ­science target is with short observations relatively simple structures, the improve- (SV) observations to demonstrate the of a strong quasar (the “Go/noGo” pro­ ment ranged from a factor of 15 to 30. end-to-end process from observation to cedure), rather than predictions. The See ALMA I for further details. Thus, calibration to imaging. In order to address water vapour radiometer (WVR) correc- more sophisticated methods of self cali- the former two points, the LBC tests tion, which takes into account estimated bration may need to be implemented in included: variations in the amount of precipitable future for some extended sources. 1) determination of the statistics of the water vapour along the line of sight to Another finding from the SV imaging was temporal phase variation as a function each antenna (see Nikolic et al., 2011; that in many cases the integration time of baseline length using observations of 2013), typically improves the phase noise on source may be driven not by the need a single bright quasar; by a factor of about two for clear skies, to reach a given signal rms, but by the 2) determination of the accuracy of the and further phase fluctuations are need to obtain sufficient coverage of the phase transfer using alternating obser- thought to be due to variations in the dry uv (Fourier domain) plane. vations of two quasars that are close atmosphere. together on the sky; 3) development of a quasi real-time The lack of an accurate dry atmospheric Science Verification observations at long method to estimate the feasibility of delay model, due to pressure variation baselines given long baseline observations; across the array over distances of several 4) determination of the optimal time inter- kilometres and the current single pres- Continuum and spectral line observations val (cycle time) between calibrator sure measurement at the centre of the of five SV targets were carried out to scans; array, is the dominant cause of system- demonstrate the long baseline potential 5) determination of antenna positions and atic phase differences (longer-term varia- of ALMA, chosen from a broad range of delay model errors using observations tions, i.e., over timescales of minutes science topics. The targets were the of many quasars distributed over the to hours) between observations of the asteroid Juno, the asymptotic giant branch sky; phase calibrator and the science target. (AGB) star Mira, the circumstellar disc 6) flux density measurement for candidate The addition of further weather stations around the young star HL Tau, the quasar weak phase calibrators; in the coming months, providing several 3C 138, and the gravitationally lensed 7) investigation of potential significant pressure measurements across the submillimetre galaxy SDP.81. The base- structure in calibrator sources when array, will allow improvement of the delay lines typically ranged from 15 metres observed at long baselines; and model and will be the main focus of the to ~15 kilometres, and observations were 8) measurement of the positional accuracy further long baseline testing later in 2015. taken in two or three bands (selected at long baselines, by phase referencing It is also expected that the addition of from Bands 3, 4, 6 and 7). The fully veri- among many quasars that are close ­further pressure sensors will improve the fied SV datasets are publicly available on together on the sky. astrometric accuracy. the ALMA Science Portal1. Details of these experiments, the analysis and conclusions are described in ALMA I. The survey for weak calibrators will The results of the LBC are presented ­continue in order to increase the number in a collection of four ApJ Letters by the An enormous amount was learned about of sources in the ALMA calibrator cata- ALMA Partnership. A summary of the ALMA observing at long baselines during logue (Fomalont et al., 2014). However, in main test results and SV observations is the highly successful LBC and some of order to find the faintest usable calibra- presented in ALMA I. Initial results on the the main results are briefly summarised tors, and thereby ensure that a suitable SV targets Juno, HL Tau and SDP.81 are here (see ALMA I for details). Further calibrator can be found close enough presented in ALMA Partnership, Hunter investigations are also ongoing, and addi- to a science target, it is likely that a differ- et al. (2015; hereafter ALMA II), ALMA tional long baseline testing will take place ent observing strategy may need to be Partnership, Brogan et al. (2015; hereafter in the last quarter of 2015 in order to fur- adopted in future. ALMA III) and ALMA Partnership, Vlahakis­ ther improve the observing mode and et al. (2015; hereafter ALMA IV), respec- test long baseline imaging at higher fre- For some of the SV targets (see the next tively. quencies. section for details), the emission was ­sufficiently strong that it was possible to Juno It was found that calibration via phase use the self-calibration technique to ALMA’s combination of excellent contin- referencing should, for most sources, improve the image quality. For Juno, self- uum brightness sensitivity at high angular only be carried out when the short- calibration provided an improvement in resolution and well-matched wavelength

The Messenger 160 – June 2015 3 Telescopes and Instrumentation Vlahakis C. et al., ALMA Extends to 15-kilometre Baselines

ø = 0.33 ø = 0.37 ø = 0.45 ø = 0.49 ø = 0.58 Figure 2. The sequence of ten ALMA LBC 1.3 milli- 1 1 1 1 1 metre continuum images of asteroid 3 Juno obtained 0 1234over 60 % of a 7.2-hour rotation, labelled by the rota- tion phase. The cross marks the position of peak intensity. From ALMA Partnership, Hunter et al. (2015). ) to Sun will also be able to provide improved long-term modelling of asteroid orbits, ø = 0.62 ø = 0.72 ø = 0.77 ø = 0.85 ø = 0.89 1 1 1 1 1 enabling improved prediction from the milliarcsec

( 56789models, by delivering very accurate 100

Δδ astrometry. 0 The HL Tau protoplanetary disc –100 The 1.3-millimetre ALMA image of HL Tau — a heavily embedded young stellar 100 0 –100 object surrounded by a protoplanetary ∆α (milliarcsec) 024 6810 12 14 16 –1 disc — could be said to be synonymous Iv (mJy beam ) with the success of the LBC, having coverage will enable mapping of the 2.67 astronomical units (au) and during been the first image to emerge from the shape and surface temperature distribu- the SV observations the illumination was early part of the campaign and the sub- tion for large numbers of main belt aster- 94 %. Figure 2 shows the resulting ten ject of a high-impact press release2. The oids and Jupiter Trojans when using the 1.3-millimetre continuum images (two per image was featured on the front cover full high angular resolution capability. observing block; labelled by rotational of Messenger 158. However, the LBC SV With physical temperatures of 100–200 K, phase). The peak brightness temperature for this source extended beyond this it should be possible to image these types varies from 207 K to 222 K over the time i­nitial result, with data also observed at of bodies at high signal-to-noise (S/N) sequence, with a median value of 215 K. 2.9 mm (Band 3) and 0.87 mm (Band 7), at resolutions as fine as ~10 kilometres. Comparison with models of the size achieving even higher angular resolution The LBC SV observations of the asteroid and shape of the asteroid from the Data- at 0.87 mm (25 mas, or 3.5 au at the 3 Juno, obtained in October 2014 in five base of Asteroid Models from Inversion ­distance of HL Tau). blocks covering four hours, provided the Techniques (DAMIT; Durech et al., 2010) first ground-based images to significantly shows good agreement (ALMA II). HL Tau is located within a ~ 0.05 pc resolve the surface of an asteroid at molecular ridge in the Taurus star-forming ­millimetre wavelengths (1.3 mm, 233 GHz). While the LBC images of Juno provide a region, which is at a distance of 140 pc. Full details can be found in ALMA II. very good example of what ALMA can While only observed as a conical reflec- achieve for this asteroid at 233 GHz, tion nebula at optical wavelengths, due to Juno is a member of the S-class of observations of Juno and other main belt high extinction, HL Tau has however been ­asteroids (stony composite of metallic asteroids at shorter wavelengths would well observed in the near- and mid-­ iron and iron-bearing silicates). Its appar- provide higher spatial resolution, e.g., ent mean diameter (from occultation ~ 20 kilometres at 345 GHz (Band 7) with studies) is 267 kilometres, although it is a similar configuration when the asteroid Figure 3. ALMA continuum images of HL Tau at 102, 233 and 344 GHz (Bands 3, 6 and 7). The beam size a triaxial ellipsoid and its rotation period is at a more favourable opposition (at a is shown for each image. From ALMA Partnership, is 7.21 hours. Its mean orbital radius is distance from Earth of ~1 au). ALMA Brogan et al. (2015).

0.0054 2.9 mm (B3) 1.3 mm (B6) 0.87 mm (B7) 58.0 0.0048 0.0042

) 0.0036

00 0.0030 20 57.5 0.0024 –1 (J

0.0018 ion beam

at 0.0012 57.0 Jy 0.0006 Declin

+18:13:56.5

38.48 38.4438.40 4:31:38.36 Right ascension (J2000)

4 The Messenger 160 – June 2015 Figure 4. The depro- HCO+(1-0), HCN, 12CO(1-0) and CN. At 100 B7 jected 1.0 mm (287 GHz) this resolution the HL Tau region has D7 ALMA image of the cir- cumstellar disc around complex outflow emission (see ALMA III). + HL Tau at an angular The HCO (1-0) emission was also imaged resolution of 39 × 19 at 0.25-arcsecond resolution: most of the B4 mas. From ALMA Part- outflow emission is resolved out, allowing 50 D4 B3 nership, Brogan et al. D3 (2015). the morphology of the HL Tau molecular gas disc to be spatially resolved for the first time (ALMA III). u) B6 B5 B2 B1 (a

0 The gravitationally lensed z ~ 3 submilli- metre galaxy SDP.81

D6 D5 D2 D1 SDP.81 (HATLAS J090311.6+003906) is Distance a z = 3.042 submillimetre galaxy gravi­ tationally lensed by an intervening ellipti- –50 cal galaxy at z = 0.299. SDP.81 was first detected in theHerschel ATLAS (H-ATLAS) survey (Negrello et al., 2010) and subsequently resolved in the milli­

metre continuum as well as CO and H2O –100 molecular line emission at 0.6–3-arcsec- ond resolution with the Submillimeter –100 –50050 100 Array (SMA) and the Plateau de Bure Distance (au) Interferometer (PdBI); see Bussmann et al. (2013) and Omont et al. (2013). As infrared; it is found to be a Class I–II proto­ a higher spectral index compared to part of SV for the ALMA LBC, the target star of spectral type around K5 with a the bright rings; rather than being entirely was observed in Bands 4, 6 and 7: collimated outflow. Since the HL Tau disc devoid of emission, the dark rings are Band 4 included the CO J=5-4 line, Band is one of the brightest at millimetre and likely to be optically thin. The centres of 6 the CO J=8-7 and the low excitation submillimetre wavelengths, it has been a most of the rings are also found to be water line H2O (202 – 111) and Band 7 well-studied source since the early 1990s. ­offset from the emission peak, making included the CO J=10-9 line. Full details The highest resolution observations prior the rings non-circular, with HL Tau at one are presented in ALMA IV. to the ALMA LBC were obtained with the focus; the size of the offsets increases Combined Array for Research in Millimeter- with the ring radius. Furthermore, there Figure 5 shows the high resolution wave Astronomy (CARMA) at a resolution are apparent resonances between the ­continuum images at Bands 4, 6 and 7, of 0.13 arcseconds (18 au at 140 pc), radii of some of the rings. reaching angular resolutions as fine as which was sufficient to resolve the disc 23 mas. This resolution is better than the (Kwon et al., 2011). The striking conclusion from ALMA III is Hubble Space Telescope images of this that all these pieces of evidence, taken source, and corresponds to an unlensed ALMA observed HL Tau as part of the together, are highly suggestive of planet spatial scale of ~180 pc in the lensed LBC in October–November 2014, using formation within the disc. Alternative inter- galaxy, or a few tens of pc in the source Bands 3, 6 and 7 (2.9, 1.3 and 0.87 mm, pretations have been discussed in the plane considering a magnification factor or ~100, 240 and 345 GHz, respectively); literature and involve grain growth promo- of ~11–20. Viewed at this level of detail, details of the observations and initial tion and trapping at specific locations in strong detections of thermal dust emis- analysis are presented in ALMA III. Fig- the disc, either due to instabilities or sion in two incredibly thin arcs, which are ure 3 shows the continuum images; the chemical/phase transitions in the gaseous the main components of an Einstein ring, size of the synthesised beam for the component of the disc (Zhang et al., 2015). were achieved. In addition to the eastern 0.87 mm Band 7 image is an impressive Further modelling and observations will and western arcs, a continuum source 30 x 19 mas. Figure 4 shows the com- elucidate the initial results from the LBC. was detected in all three bands at the bined Band 6 and 7 (1.0 mm) image, de- The combination of high angular resolu- central position of the lens (the early-type projected by the line of sight inclination tion and high fidelity imaging has already galaxy SDSS J090311.57+003906.5). of the disc of 46.72 degrees (derived from revealed a wealth of detail in the circum- This central continuum source is likely due fitting each ring assuming circular orbits) stellar disc, demonstrating the power of to a weak, previously undetected, active with the bright and dark rings labelled ALMA to deliver transformational science. galactic nucleus (AGN) in the foreground (as B1-7 and D1-6, respectively). The elliptical galaxy (see ALMA IV) and there spectral index (derived from the flux vari- The HL Tau SV observations did not just is no evidence of a central lensed image ation with wavelength across the three include continuum emission. Four spec- in the molecular line data (Wong et al., bands) is found to vary as a function of tral lines were also imaged in Band 3, at 2015). radial distance, with the dark rings having 1.1-arcsecond angular resolution (~ 25 au):

The Messenger 160 – June 2015 5 Telescopes and Instrumentation Vlahakis C. et al., ALMA Extends to 15-kilometre Baselines

09.0 2.0 mm 1.3 mm 1.0 mm

) 08.0 000 2 (J

07.0 ion 06.0 Declin at 05.0

+0:39:04.0 11.6811.60 11.529:03:11.44 Right ascension (J2000)

Figure 5. (Above) ALMA ­continuum images of the Figure 6. (Below) ALMA CO integrated intensity gravitationally lensed z ~ 3 galaxy SDP.81 in Bands 4 images of the gravitationally lensed z ~ 3 galaxy (2.0 mm), 6 (1.3 mm) and 7 (1.0 mm), left to right. SDP.81 in Band 4 (CO J=5-4), Band 6 (CO J=8-7) The highest angular resolution, at 1.0 mm (290 GHz) and Band 7 (CO J=10-9), from left to right. The angu- in Band 7, is 31 × 23 mas (right panel). From ALMA lar resolution is ~170 mas. From ALMA Partnership, Partnership, Vlahakis et al. (2015). Vlahakis et al. (2015).

09.0 CO J=5-4CO J=8-7CO J=10-9

) 08.0 000 2 (J 07.0 ion at 06.0 Declin 05.0

+0:39:04.0 11.6811.60 11.529:03:11.44 Right ascension (J2000)

–1 The three transitions of CO (J=5-4, J=8-7 baselines were used, giving an angular mass-to-light ratio of ~ 2 MA LA , or a 8 and J=10-8) were also imaged (see Fig- reso­lution of ~ 0.9 arcseconds (ALMA IV). > 3 × 10 MA black hole if an AGN is ure 6). In order to achieve good S/N, the Nonetheless, with this detection at ­present (Tamura et al., 2015; Wong et al. spectral lines were imaged at somewhat 0.9-arcsecond resolution, ALMA has 2015). Dye et al. (2015), Rybak et al. coarser resolution (~170 mas). The CO achieved the highest resolution detection (2015a) and Swinbank et al. (2015) have morphology is broadly similar to that of thermal water emission in an extra­ modelled the background starburst seen in the continuum, tracing two main galactic source to date. The water emis- ­galaxy and find a highly non-uniform dis- gravitational arcs; the CO emission in sion is confined to only one of the velocity tribution of dust clumps. Swinbank et al. each arc, however, is clumpier. The spa- components, in agreement with previous (2015), for example, find a relatively tially integrated CO line profiles show observations with the PdBI (Omont et al., smooth CO velocity field resembling disc- ­evidence of two components separated 2013). like dynamics and infer a rotating disc by ~ 300 km s–1, with the lower velocity with a rotation velocity of 320 km s–1; they emission occurring predominantly in the There have already been several addi- suggest a disc that is in a state of col- western arc. Thermal H2O line emission tional studies based on the ALMA LBC lapse, and, from comparison with HST, was clearly detected in the eastern arc observations, including one on mass the data suggest a scenario where two and at low S/N in the western arc, but to modelling of the lensing elliptical at separate systems are merging. Alternative achieve these detections only the shorter z = 0.299 that suggests a stellar core of conclusions have been presented by

6 The Messenger 160 – June 2015 –3 Figure 7. ALMA Band 3 × 10 image of Mira AB showing 2 the binary (left) and the 0.4 Band 6 image of Mira B with the point source subtracted, showing the partially ion-

) 0.2 1.5 ised circumstellar shell 0 (right). From Vlemmings et

) al. (2015). as m seconds 0 (m

rc 1 et /bea (a fs Jy et

A Of fs –0.2 B –100 Of 0.5

–0.4

0 0.6 0.40.2 05–0.2 00 400 East offset (arcseconds)East offset (mas) other authors (Rybak et al., 2015b), and Bands 3 and 6, with a ­hotspot of References further analysis of the rich datasets will ~10 000 K brightness temperature on ALMA Partnership, Brogan, C. L. et al. 2015, ApJL, undoubtedly shed further light on the the stellar disc; it is suggested that the arXiv:1503.02649 (ALMA III) nature of this intriguing system. hotspot may be related to magnetic activ- ALMA Partnership, Fomalont, E. B. et al. 2015, ApJL, ity. A partially ionised region of ~ 2.4 au arXiv:1504.04877 (ALMA I) The evolved star Mira diameter (26 mas) around Mira B was ALMA Partnership, Hunter, T. R. et al. 2015, ApJL, arXiv:1503.02650 (ALMA II) The archetypical long-period variable star also resolved (Figure 7, right), and the ALMA Partnership, Vlahakis, C. et al. 2015, ApJL, Mira A (o Ceti) was observed in contin- emission is found to be consistent with arXiv:1503.02652 (ALMA IV) uum and several spectral lines in Bands 3 material close to the Mira B accretion Durech, J. et al. 2011, Icarus, 117, 313 and 6 as part of the LBC SV. Mira A, a disc that has been gravitationally cap- Dye, S. et al. 2015, MNRAS, submitted, arXiv:1503.08720 mass-losing AGB star, is the primary in a tured from the AGB wind (see Vlemmings Fomalont, E. et al. 2014, The Messenger, 155, 19 binary system with a companion, Mira B et al. [2015] for details). Kwon, W. et al. 2011, ApJ, 741, 3 (VZ Ceti; thought to be a white dwarf). At Negrello, M. et al. 2010, Science, 440, 1999 a distance of 92 pc (van Leeuwen, 2007), Nikolic, B. et al. 2011, The Messenger, 143, 11 Nikolic, B. et al. 2013, A&A, 552, A104 it is the closest such binary. The Mira AB Prospects Omont, A. et al. 2013, A&A, 551, A115 system has already been observed in Ramstedt, S. et al. 2014, A&A, 570, L14 the CO J=3-2 line with ALMA in Cycle 1 The ALMA LBC observations have Rybak, M. et al. 2015a, MNRAS, in press, (Ramstedt et al., 2014) and showed indi- been spectacularly successful, both in arXiv:1503.02025 Rybak, M. et al. 2105b, MNRAS, submitted, cations of a complex circumstellar enve- terms of reaching the high angular arXiv:1506.01425 lope, but the binary pair was only margin- ­resolutions possible at these baselines Swinbank, M. et al. 2015, ApJL, in press, ally resolved. The LBC observations have and addressing potential atmospheric arXiV:1505.05148 an angular resolution as fine as ~ 25 mas and instrumental limitations, and in terms Tamura, Y. et al. 2015, PASJ, in press, arXiv:1503.07605 and thus easily resolve the binary pair of the quality and scientific value of the van Leeuwen, F. 2007, A&A, 474, 653 (see Figure 7). SV observations. A number of analyses Vlemmings, W. H. T. et al. 2015, A&A, 577, 4 have already been published based on Wong, K. C. 2015, ApJL, submitted, Vlemmings et al. (2015) presented an the SV observations and more can be arXiv:1503.05558 Zhang, K., Blake, G. A. & Bergin, E. A. 2015, analysis of the LBC SV continuum data, expected given the quality and scientific arXiv:1505.00882 determining the size, shape, flux density novelty of these data. As a result of the and spectral index of both sources in the success of the LBC, the Call for Proposals AB system. For the first time in the sub- for Cycle 3 included allocations for long Links millimetre, the extended atmosphere of a baselines, up to ten kilometres for Bands 1 ALMA Science Archive LBC data: star was resolved — Mira A was resolved 3, 4 and 6, and showed a very strong https://almascience.eso.org/news/release-of-­ into an elliptical disc with a major axis of demand. A wealth of further exciting sci- science-verification-data-from-the-alma-long- 42 mas (~ 3.8 au) at 94 GHz (Band 3) and ence can certainly be expected in the baseline-campaign-2 2 LBC HL Tau Release: http://www.eso.org/public/ 43 mas (4.0 au) at 229 GHz (Band 6). years to come as ALMA achieves its full news/eso1436/ Brightness temperatures were found to potential for observing at very high reso- be ~ 5300 K and 2500 K, respectively in lution.

The Messenger 160 – June 2015 7