SkyMapper and the search for EMP stars. Mike Bessell The Australian Na/onal University

Transcription

1 SkyMapper and the search for EMP stars Mike Bessell The Australian Na/onal University

2

3 SkyMapper filters SkyMapper filters are large 310 x 310 x 15 mm. Left, our 3 layer all-glass g filter. Right, Hα filter coated on a single red glass substrate. In 2006 uniform interference filters could not be made that large. In 2014 Materion can now make them.

4 SkyMapper Shallow Sky Survey The Shallow Survey targets the en6re Southern hemisphere with short exposures, ensuring full hemisphere coverage early in the project and a bright satura6on limit. Several dithered visits aim at fully covering the target area despite gaps in the CCD mosaic, and repeat observa6ons to ascertain the sta6c or transient nature of any detected sources. During each visit a six-colour sequence is observed within less than five minutes. While the first year of SkyMapper survey opera6ons was dedicated mostly to the Shallow Survey, it now con6nues only in the brightest nights around full moon. The Shallow Survey will be complete beyond magnitude 17.5 in all six bands uvgriz and provide the calibra6on reference for the following Main Survey.

5 SkyMapper Main Survey Most of SkyMapper's observing 6me is now dedicated to the Main Survey. The Main Survey targets the same sky area as the Shallow Survey in the same six bands, but aims to be complete to g, r ~ 22. In combina6on with the Shallow Survey the en6re SkyMapper Southern Survey will provide calibrated uvgriz photometry from magnitude 9 to 22. The Main Survey includes two six-colour sequences, each taken within a 20-min interval, as well as addi6onal visits to collect pairs of gr images. It operates under dark and grey sky. During astronomical but not nau6cal twilight addi6onal pairs of iz images are collected.

6 SkyMapper Main Survey This has an expected depth (in AB mags) for a signal-tonoise ra6o of 5 in 2.5" seeing, of: The status of the surveys are con6nuously updated on the SkyMapper webpage. hqp://skymapper.anu.edu.au/surveys/

7 SkyMapper Main Survey SkyMapper s regular opera6ons model includes daily calibra6on procedures including bias frames, twilight flazields and several visits to dedicated SkyMapper standard fields during the night. Eight such standard fields have been defined, which include a Hubble Space Telescope (HST) spectrophotometric standard star. The standard field observa6ons have not been used for DR1 but will be used in the future to refine the photometric calibra6on.

8 The Transient Survey operates mostly in the 30- percen6le of worst seeing, but complements its 6me series with good-seeing images whenever it needs to ensure con6nuity in its cadence. It repeatedly observes a 2,000 deg 2 subset of the Southern sky in the gri bands aiming for a 4-day cadence. The main science drivers of the Transient Survey are crea6ng a sample of well-calibrated type-ia supernovae at z < 0.1 to improve the local anchor for cosmological studies The Alert Program operates in a trigger mode and overrides other telescope ac6vi6es on a 1-minute 6mescale. Currently, its major science targets are mergers of neutron stars and black holes detected as Gravita6onal Wave events by the Advanced LIGO collabora6on, as well as alerts from searches for Fast Radio Bursts at the Parkes telescope (SUPERB) and the recently refurbished Molonglo telescope (UTMOST). Finally, SkyMapper dedicates 20% of the available observing 6me in the five years from March 2014 to Third-Party Science via proposals reviewed by the ANUTAC. This category may not copy the public Southern Survey ac6vi6es and is intended for deeper surveys with SkyMapper s bespoke filters, 6me-cri6cal observa6ons of astronomical events or observa6ons with custom filters (at this point only a 657/12 Halpha-filter is available).

9

10 u v g and stellar spectra Wavelength Spectra of a [Fe/H]=-2.5 and 0.0 solar-type star with the u v g bands superimposed

11 The SAGE u and v bands are narrower with steep edges compared to the Skymapper glass filter defined bands.

12 The SkyMapper telescope is remotely controlled and programmed by a sophis6cated scheduler that monitors the weather, seeing, transparency, the posi6on of bright stars, planets and the moon, and interrupts from SN/ gamma ray burst alerts. The data is flat-fielded using twilight flats that are precisely scheduled (cloud cover permijng) evening and morning. The ZP magnitude for each amplifier (64) field is derived using APASS hqps:// stars and synthe6c rela6ons between u-g or u-v, v-g or v-v etc for FGK stars (unreddened colors between g -i ~0.5 and 0.9). The reddening is es6mated for each amplifier field from Schlegel maps and observed color-color loci.

13 %

14 New Releases DR1 May % Southern sky Lower noise BeQer calibra6on Variable data quality DR2 in square deg MS Priori6zed fields SDSS Stripe 82 2dFLenS, GAMA Shapley SC ATCA/ATLAS ASKAP Early Science K2 fields etc.

15 Skymapper Survey Data Access Updated 10 Jan The ARC LIEF grant which funds the opera6on of SkyMapper's Southern Sky Survey provides for priority data access to Australian users. For the purposes of data access, the defini6on of 'Australian users' is taken to mean persons residing within Australia. Members of approved Survey Science Projects who reside outside of Australia will also be provided with access to the data. Aqer a month proprietary period, each data release from DR1 onwards will become globally available. Details of how SkyMapper data is accessible from the Virtual Observatory can be found from the SkyMapper web page or from Murphy_SkyMapper_ASVO_3Apr14.pdf

16 hqp://skymapper.anu.edu.au/

17 For the last year we have been using data from the SkyMapper Early Data Release, the coverage of which is shown above. Much of the data was zero-point calibrated using 2MASS photometry and subsequently had some problems with reddening correc6on uncertain6es. Some fields also had only two observa6ons. However, we have been regularly taking spectroscopic follow-up R=3000 spectroscopic observa6ons with the 2.3m WiFeS spectrograph of EMP candidates and echelle observa6ons with Magellan and Keck of the most metal-poor objects.

18 We selected from the EDR database: class star > 0.9 (high confidence star) photometry flag < 3 (Source Extractor photometry OK) >1 measurement in all ugriz, >2 measures for v g_psf < 16 (allows high dispersion follow-up) tabulated photometry errors <0.03 g, i; <0.05 v E(B-V) < 0.20 (Schlegel et al scale) no other EDR or 2MASS detec6on with 5arcsec This selec6on yields 2.75 million objects.

19 Assessment of EDR zeropoint calibra6on and SkyMapper passbands have been made using synthe6c photometry from different spectrophotometric atlases. The Pickles ATLAS is par6cularly useful as it represents stars of different luminosity and temperature.

20 The EDR sample ploqed here are the approximately 1000 brightest stars with the lowest es6mated reddening.

21 The EDR stars ploqed here are several thousand of the reddest stars in order to establish the cool dwarf and giant sequences. The turnup redward of g-i=1.5 is due mainly to the red leak in the u filter.

22 Tracks from Vennes This is early data overlaid with white dwarf loci for different masses (gravi6es).

23

24

25

26

27

28 SkyMapper EDR photometry metallicity sensitive 2-colour diagram metal-poor giants? To inves6gate the usefulness of this diagram we crossmatched our EDR sample with a compila6on of high dispersion abundance measurements ([Fe/H] < -2.0). There were 111 stars in common, aqer minimizing the effects of photometry errors and reddening by selec6ng known abundance stars in the EDR to have E(B-V) < 0.1 and photometric uncertain6es < 0.02 mag for all of v, g and i.

29 SkyMapper EDR photometry metallicity sensitive 2-colour diagram Generally the loca6on of known metal-poor ([Fe/H] < -2 ) stars in the diagram seems (mostly) sensible. The ones that are very discrepant are generally very carbon-rich.

30 SkyMapper EDR photometry metallicity sensitive 2-colour diagram Shown in purple are a set of Dartmouth isochrones for age 12.5 Gyr and metallici6es between solar and The lower part of the main sequence is NOT ploqed i.e., biased to giants, subgiants and turnoff stars. A metallicity index offset has been applied to the isochrones. The isochrones and the loca6on of known metal-poor stars are (mostly) consistent.

31 Given that metallicity sensitive diagram seems to show the expected sensitivity, how should we go about selecting candidates for low resolution spectroscopy at the 2.3m? The previous plot suggests we should concentrate on giants because for the hotter stars the isochrones are relatively close together (i.e., metallicity sensitivity is reduced) and the known metal-poor stars are not as well separated from the large number of solar-type dwarfs. So we have chosen:! 0.4 < (g-i) 0 < 0.9! < m_index < 0.11 The upper m_index value allows for errors in the m_index (which are potentially as large as mag) while the lower cutoff is approximately the location of the -2.0 isochrone. This selection yields about ~8000 candidates for g < 16 and ~3000 candidates for g < 15.

32 SkyMapper EDR photometry metallicity sensitive 2-colour diagram The value of the lower cutoff is quite crucial as the number of candidates goes up very steeply as it s increased. On the other hand it would a pity to miss any prime candidates because their metallicity_index errors scaqered the star downwards.

33 What are the many objects that lie well above the theoretical EMP isochrones? Index < -0.2.

34 What are the many objects that lie well above the theoretical EMP isochrones? Index < The reddest stars (g-i > ~0.8) tended to be active K and M stars with Ca H&K in emission T Tauri stars. A few bluer stars were cool white dwarfs and/or cataclysmic variables. Many objects were QSO or AGN (in poor seeing). Amongst the bluer objects that were not QSOs, most were RR Lyrae variables, with measures taken at different phases. Others were very close pairs of stars not picked up as such in the photometry of poor seeing images. Some apparently single objects gave a combination of two different spectra. Only a few were bad photometry. Necessary to cross correlate with SIMBAD and NED Research School of Astronomy & Astrophysics

35 We have then taken the candidate list and proceeded to observe objects from the list with the 2.3m telescope and the WiFeS spectrograph (B3000 setup λ coverage ~ Å). The reduced 2.3m spectra are flux calibrated and the best fitting model from a grid of MARCS model atmosphere fluxes determined using the fitter code developed by Simon Murphy (described in Norris, Bessell, Yong et al 2013). The spectrophotometric model fits provide estimates of T eff, log g and [Fe/H] for each observed spectrum. Here we report the outcome of six 2.3m observing runs in 2016, during which spectra were obtained of 339 EMP candidates in the EDR selection box, plus standards, known abundance stars, etc. Note: because the 2.3m observing started before the final EDR release, we have observed ~40% more objects than the specific defined sample just discussed. However the on-going observations now use just the EDR photometry defined sample.

36 Example Spectrum 360sec exposure. Star has g = and m_index=0.005 and (g-i) 0 = 0.58

37 Example of fitter output (same star as previous slide)

38 WiFeS R=3000 spectra of EMP stars CD [Fe/H]= CD HE [Fe/H]= SM [Fe/H]< -7.1 Note the region around the CaII H&K lines and the CH G-band Clearly extremely low [Ca/H] and high [C/H] can be seen in WiFeS spectra. Relatiive Absolute Flux 2.0 HE SM Wavelength (nm) B3000 WiFeS spectra

39 Fitting WiFeS spectra

40

41 C-rich giant EMP star

42 One obvious question is how good are the fitter [Fe/H] values? Plot shows a well defined rela6on between the fiqer [Fe/H] values (quan6zed at the 0.25 dex level) and high dispersion spectroscopic values of [Fe/H] from the literature for 23 stars. Selec6ng stars with fiqer abundances at and below -3 is unlikely to significantly bias the actual metallicity distribu6on of the stars.

43 Metallicity Distribution Function The MDF for the 339 EDR selected stars observed at the 2.3m confirms that the selec6on process is basically working: we reproduce the standard simple model result that a 1 dex decrease in [Fe/H] gives a 1 dex reduc6on in the number of stars; the turnover in the metallici6es is at [Fe/H] = -2.75, which is where we d like it. The efficiency is excellent (30% at or below -3). There are 6 stars at or below -3.75, whereas extrapola6ng the line would suggest 14 expected (~1.8σ effect). Need more stars.

44 Conclusions Overall a useful set of candidate EMP stars can be reliably selected from the SkyMapper EDR dataset. 2.3m spectroscopic follow-up has produced a number of candidates for follow-up at high dispersion with Magellan. EMP stars are nevertheless rare and we will need to continue to pursue the 2.3m program in order to increase the number of candidates at the lowest metallicities. A sample 5-7x larger (~2000 stars) may well be required to find significant number (~dozen) of stars below -4.5.

45 Improvements The photometric quality of DR1 will be much improved likely to reduce contamination of the selection window by metal-richer objects. DR1 should also allow the search to be extended to lower galactic latitudes/higher reddening, a region less well explored compared to the halo. So far we have only used, the metallicity index vs g-i 2-colour diagram for selection. We plan to make use of the r, z photometry (and u for DR1) plus 2MASS, WISE to investigate improved selection processes (cross-matches to other surveys are already part of the SkyMapper EDR).

46 Issues: Star symbols are objects with [C/Fe] > +0.7; some are substan6ally more C-rich. It appears we are unlikely to find stars at [Fe/H] -3.0 and [C/Fe] ~1.5. Research School of Astronomy & Astrophysics

47 Issues: Clearly we would NOT select these very carbon-enhanced EMP stars with the current selection window. They lie low in the metallicity index plot because the additional CH-feature absorption in the v filter band pass causes them to mimic a more metal-rich object. The effect of this potential bias remains to be quantified, and it may be possible to isolate stars of this type using additional photometric colours. The bias will be a function of both [Fe/H] and [C/Fe]. We note that the bias may only apply to very strong carbon enhancements ([C/Fe] > +2 at [Fe/H] = -4?) because the Jacobson et al 2015 sample did not obviously lack CEMP stars, although it did not contain any stars with [C/Fe] > 1.5. The extremely iron-poor star SMSSJ was found and it is very C-rich.

48 WiFeS spectra of EMP stars CD [Fe/H]= CD HE [Fe/H]= SM [Fe/H]< -7.1 Note the region around the CaII H&K lines and the CH G-band Clearly extremely low [Ca/H] and high [C/H] can be seen in WiFeS spectra. Relatiive Absolute Flux 2.0 HE SM Wavelength (nm) B3000 WiFeS spectra

49 -0.10 blue -2.0 subgiant isochrone 12.5Gyrs red -3.5 subgiant isochrone 12.5 Gyrs dots C-rich with -5.5 metallicity log g=0.5 at 4250 to 4.0 at 6500 [C/Fe]=+3.6, [N/Fe]=+2.2, [O/Fe]=+2.4, [Fe/H]= v-g-1.5(g-i)* g-i* Research School of Astronomy & Astrophysics

50 The CH and CN bands can be very strong in very C-rich stars with [Fe/H] > -3. The CH and C2 bands affect the blanke6ng in the v and g bands, while the CN bands mainly affects the i and z bands. However, in such a C-rich star, r-i should be bluer and v-g and g-r redder than normal stars. So it is likely that we can devise some colors, including those involving IR bands that will enable us to separate out C-rich MP stars, for [Fe/H] > -3, and [C/Fe]>-2. HE1429 [C/Fe] =2.2, [Fe/H]=-2.5 HE2139 [C/Fe] =2.6, [Fe/H]=-4

51 Image Processing EDR / DR1 EDR DR1 Bias temporally unstable 10+ principal components, changing per night and CCD Also: CR removal (LACosmics) EDR DR1

52 Aqer many years and many set backs we are now looking forward to SkyMapper DR1 for Australian astronomers and collaborators in May 2017 followed by whole world release in will also see the release of DR2 incorpora6ng photometric flat calibra6on using HST spectrophotometric fields. In 2018 I also look forward to hearing about the comple6on of the calibrated SAGE Bok u, v survey closely followed by comple6on of the SAGE Nanshan gri survey and the discovery of many new northern EMP stars.