astra spectrophotometry: design and overview · astra spectrometry: design and overview 3 indices....

**FULL TITLE**ASP Conference Series, Vol. **VOLUME**, **YEAR OF PUBLICATION****NAMES OF EDITORS**

ASTRA Spectrophotometry: Design and Overview

Saul J. Adelman1, Austin F. Gulliver2, Barry Smalley3, John S. Pazder4,P. F. Younger5, Louis J. Boyd6, Donald Epand6, and Thomas Younger5

1Department of Physics, The Citadel, 171 Moultrie Street, Charleston,SC 29409, USA2Department of Physics & Astronomy, Brandon University, Brandon,MB R7A 6A9, Canada3Astrophysics Group, School of Chemistry and Physics, KeeleUniversity, Straffordshire ST5 5BG, United Kingdom4Dominion Astrophysical Observatory, Herzberg Institute forAstrophysics, National Research of Canada, 5071 W. Saanich Road,Victoria, BC V9E 2E7, Canada5Aurora Astronomical Services, 585 Aurora Way, Victoria, BC V8Z3J8, Canada6Fairborn Observatory, HCR5 2, Box 256, Patagonia, AZ, 85624,USA

Abstract. The ASTRA (Automated Spectrophotometric Telescope ResearchAssociates) Cassegrain Spectrophotometer and its automated 0.5-m f/16 tele-scope will soon be working together at the Fairborn Observatory near Nogales,Arizona. Scientific observations are expected to begin in 2007. We provide anoverview of this project and review the design of the system. A separate paper(Smalley et al. 2007) presents details of the data reduction and flux calibrations.The Nogales site averages 150 photometric nights per year. ASTRA should ob-serve stars whose declinations are in the range +80◦ to -35◦.

In an hour the system should obtain S/N = 200 observations of stars as faintas 9.5 mag after correction for instrumental errors. Vega will require about 25seconds for observation and CCD readout. Usually the telescope will find its nexttarget in less than a minute. A small CCD camera finds and centers the targetand a second then guides on the zeroth order spectrum. The spectrophotometeruses both a grating and a cross-dispersing prism to produce spectra from boththe first and the second orders simultaneously. The square 30 arc second skyfields for each order do not overlap. The resolution is 7 A in second and 14A in first order. The wavelength range is approximately λλ3300-9000. We areinitially using about 10 minutes/hour to observe Vega and secondary standardcandidates.

Our scientific CCD is electronically cooled to -50◦ C with a water recircula-tion system heat sink. The same 4◦ C recycling water system provides thermalstabilization of the instrument. Our flat fielding system uses a second 0.5-mtelescope to produce a collimated beam from a 100 µ pinhole illuminated by aquartz halogen lamp. When the two telescopes point at one another this ”arti-ficial star” is focused by the ASTRA telescope which is then rocked to exposethe image from the top to the bottom of the entrance aperture.

A LINUX HP server at The Citadel will have databases of ASTRA ob-servations. Each observing request has its own priority and observing window,ASTRA can observe standard stars at a regular rate throughout the night, anyaccessible target at a given time, and variable stars. ASTRA will produce con-

1

2 Adelman, Gulliver, Smalley, Pazder, Younger, Boyd, et al.

siderable high quality data. Since the authors can analyze only part of yearlydata output, we are looking for research collaborators.

1. Introduction

Almost all the rotating grating scanners used for spectrophotometry are retired.Their replacements were not intended for stellar observations and lacked therequired accuracy and precision. The stellar fluxes from these scanners maycontain systematic wavelength-dependent errors due to those in the absolutecalibration, extinction, bandpass centering, and scattered light in the instru-ment. Such data typically consists of 15 to 20 values covering λλ3400-7100with 20 to 50 A wide bandpasses usually of spectral regions with minimal lineblanketing. Due to time constraints the extinction often was based on mean ob-servatory values with errors rarely better than 1%. In particular Breger (1976a),Ardeberg & Virdefors (1980), and Adelman et al. (1989) compiled these obser-vations. There are also 5-m Hale telescope measurements, based largely onrotating grating scanner calibrations, with now retired instruments: the multi-channel spectrophotometer (Oke & Gunn 1983, Gunn & Stryker 1983) and theDouble Spectrograph (Oke 1990 who provided standards for the Hubble SpaceTelescope) and satellite measurements of the optical ultraviolet (e.g., Code &Meade 1979).

With optical region grating scanner (and ultraviolet flux) data and Balmerline profiles, astronomers derived reasonably good effective temperatures andsurface gravities of normal single slowly rotating B, A, and F stars. These havebeen expressed in terms of filter photometric indices of various systems. Forstars with significant non-solar compositions, such calibrations are not necessar-ily accurate as metallicity, microturbulence, macroturbulence, and/or magneticfields affect the stellar fluxes in subtle, yet measurable ways (Adelman & Rayle2000). For stellar astrophysics, at the heart of our understanding the history andevolution of galaxies, accurate spectrophotometry is critical for future advance-ments. Spectrophotometry can also be an important technique for the study ofsolar system objects, nebulae, star clusters, and galaxies. NEW DIRECTIONSIN SPECTROPHOTOMETRY (eds. Philip, Hayes & Adelman 1988) discussesadditional uses. Our ASTRA instrument has been optimized for differential ab-solute spectrophotometry. Obtaining better absolutely calibrated standard starvalues is a task more appropriate for metrologists to perform.

2. The ASTRA System

We built a simple, efficient, and elegantly designed spectrophotometer with aCCD detector for use on an automated 0.5-m telescope at the Fairborn Obser-vatory, Washington Camp, AZ, just north of the US-Mexico border about 30kilometers east of Nogales, AZ (N 31.◦5) This unique instrument should producehigh quality fluxes through at least the λλ3300-9000 region. With a resolutionof 14 A in first and 7 A in second order, one could synthesize indices for mostoptical region filter photometric systems and then perform studies using such

ASTRA Spectrometry: Design and Overview 3

indices. However, at a resolution comparable with classification spectroscopy,ASTRA will obtain considerable astrophysical information that is lost at filterphotometric resolutions.

Our 0.5-m telescope is a new design by Louis Boyd which incorporates manyfeatures used by the other small automated telescopes of Fairborn Observatoryincluding control by ATIS (Automated Telescope Instruction Set whose docu-mentation is available at www.fairobs.org), disk and roller drives on both axes,and a very short search and find interval of about 30 seconds between succes-sive program stars. Don Epand is implementing the ATIS commands needed forspectrophotometric observations. Testing indicates that telescope flexure shouldnot be a problem.

The spectrophotometer will be mounted at the f/16 Cassegrain focus. Thescience CCD is an Apogee Instruments Alta system with a system heat sink. Thespectrophotometer is in an insulated box, which will be kept at a temperatureof 5◦ C by water cooling the optical plate. The CCD is an E2V 30-11 with 1024x 256 square 26-micron pixels and an ultraviolet coating. Its temperature isexpected to be close to -60◦C. By maximizing the optical ultraviolet response,the total exposure time to reach a S/N ratio of 200 at all usable wavelengthswill be minimized.

Both the instrument and the telescope are in the telescope shed and arebeing tested together. This should take 3 to 6 months culminating with thestart of scientific observing. The spectrophotometer has been thoroughly tested.Our test observations are now being used to prepare for observing.

3. Design Criteria for the ASTRA Spectrophotometer

Our criteria for the design of the spectrophotometer were the result of our ex-perience as observers and as designers.

1. The average seeing is between 1.5 to 2.0 seconds of arc.2. Zeroth order light from the grating is used to guide the telescope during

exposures.3. The science CCD should have a high quantum efficiency λλ3200-9500

especially shortward of the Balmer jump.4. The total spectral range will be observable with a single exposure.5. Light loss and scattering should be minimized by the use of a small

number of optical surfaces and thorough baffling.6. By rocking the telescope to widen the spectrum to at least 5 pixels,

software can remove cosmic ray signatures.7. Two-pixel resolution should be better than 15 A in second order.8. Optical components should not move significantly and thus are cemented

to solid mounts.9. A constant spectrophotometer temperature is needed for thermal stabil-

ity.10. Accurate sky subtraction requires use of a square projected aperture.11. Minimizing moments on the telescope necessitates a compact spec-

trophotometer.12. To minimize costs simple optical and mechanical designs are needed.


13. There should be minimal flexure in the telescope and the spectropho-tometer.

4. The Spectrophotometer

The spectrophotometer’s insulated case is a box, rectangular in cross-sectionwhich may be opened to provide access to the instrument. The length overall isroughly 38 cm, greatest width 28 cm, and height 14 cm. The box front is themounting disc which couples wit the back of the telescope, centered on its opticalaxis. A hole in the disc center holds a baffle, the upper end of which is sealed witha 1-cm thick fused silica window. The optical plate is a 1.25-cm thick aluminum;the remainder of the case structure was made from 0.95-cm aluminum. The massis about 20.5 kg (45 pounds). Thermal stability is accomplished by using a 4◦ Cwater to cool the optical plate and covering the entire spectrophotometer with1.25-cm maritime insulation. The instrument is appropriately baffled with blackflocking to minimize scattering problems.

The optical parts are epoxyed into solid mounts. With our slitless spec-trophotometer, focusing is done with the telescope. The optics placement onthe optical plate used an illuminated end of a 100 µ optical fiber placed at thetheoretical focal position inside the box. The grating rotation was set duringthe alignment. Both a He/Ne laser and an illuminated 100-micron fiber wereused to align the optics and set the grating rotation.

A prismatic cross disperser provides sufficient order separation for the spec-trophotometer to cover λλ3000-10000 in a single exposure. The main dispersionelement is a 300 gr mm−1 grating with a λ8600 blaze. From diffraction grat-ing efficiency data, the optimal order coverage is λλ5500-10000 in the first andλλ3000-6000 in the second order. A 500 A overlap of the orders allows the dataquality to be checked.

A 1.0 arc second object is 2 pixels wide at the image, which is the Nyquistfrequency of the 26 square micron CCD pixels. The optical performance of thespectrophotometer at 80% encircled energy is better than 17 microns (50% in8 microns) over the whole spectral range for a point source object. A 1.0 arcsecond image of the star with a width of 30 microns for a perfect camera, willhave an image size, at worst, of 35 microns. As the smallest bandpass is twopixels wide, the resolution is 14 A in the first and 7 A in the second order.

To preserve the resolution set by the stellar image in this slitless design andto find cosmic ray hits, the spectrum is widened to 5 or 6 pixels by mechanicalrocking of the telescope. The separation of the two orders is sufficient so thatduring the rocking, the sky exposures of each order do not overlap. The 30arcsec x 30 arcsec projected hole in the stellar acquisition mirror used to acquirethe star acts as a field stop. As the CCD read noise is about 8 electrons perpixel, widening the spectrum does not significantly degrade the S/N ratio.

The guide and centering stellar acquisition camera optics are both standardachromatic doublets. For the guide camera the image scale is 2 pixels for a 1.0arc second disc, and for the stellar acquisition camera it is 3 pixels for a 1.0 arcsecond disc.


5. Instrumental Diagrams

As figures we have four photographs taken recently. In Figure 1, Frank Youngerholds the ASTRA Spectrophotometer. The snout will be replaced by one made ofpvc at whose end there will be an optical window. Figures 2 and 3 show the opticswith the top of the box off. They are views towards the front end and the backend, respectively. Subsequent to when the photographs were taken additionalparts such as the filter wheel were installed at the front of the instrumental box.Figure 4 shows the bottom of the box with the Apogee CCD and the plumbingfor the 4

◦C water. In Figures 2 and 4 one can see the red engine mount to which

the Spectrophotometer was attached for much of the assembly. This permittedthe instrument to be moved for the ease of installation and testing.

6. Observing

With Astra’s wavelength coverage and the telescope’s rapid setting ability, thetime required to make frequent observations of standards is minimal. To exposea 5th magnitude A0 V star and read the CCD should take about 25 secondsto obtain a S/N ≥ 200 at all useful wavelengths. This estimate is based on thetotal system efficiencies, the Hayes & Latham (1975) calibration of Vega, a 0.5m telescope, and atmospheric extinction. For the brightest stars, we introducea 5 mag. neutral density filter just before the field stop. The second order isread out first to be useful even when the first order is overexposed.

The spectrophotometer uses two small CCD cameras for acquiring starsand for guiding. The flat mirror with the projected small square hole is placedat a 45◦ angle to the optical axis at the telescope focus. The light reflectedfrom the mirror is refocused onto an aquisition camera. An autoguider movesthe telescope so that the starlight falls through the hole in the mirror and thusenters the spectrophotometer. The zeroth order light is focused onto a guideCCD camera. Another autoguider keeps the zeroth order image at a particularpixel. To obtain flat fields, the telescope points at a second 0.5-m telescopewhich produces a collimated beam of light from a 100-µ pinhole illuminated bya quartz-halogen lamp .

The data reduction process is described by Smalley et al. (2007).

7. Projects

We want to obtain closure on the secondary standards in a year of observatingand then concentrate on the projects to demonstrate the various astrophysicaluses of spectrophotometric data. During the first two years of observing, we willbegin two major projects:

1. Revision and Extension of the Secondary Standards: The ASTRA fluxesmust be reduced to a uniform system based on the absolute values of Vega. Notall possible secondary standards are equally good for calibrating all wavelengthregions. Taylor (1984) extended and made more nearly uniform the existingbright star standards. For ASTRA, they must be redetermined as its resolutionis greater than of order 25 A for the older instruments. We selected 250 candidate


Figure 1. Frank Younger holds the ASTRA Spectrophotometer. The snoutto your right is used for testing. For observing, it will be replaced by a shorterstructure made of pvc and will have an optical window on its end.


Figure 2. The ASTRA Spectrophotometer with the top off. The view isfrom the back towards the front of the instrument. In this photograph not allof the parts near the front end have been installed.

secondary stars mainly from the least variable B0 to F0 stars in Hipparcosphotometry (ESA 1997) to supplement those given by Taylor.

2. Sample Fluxes of Population I and II Stars: This longer-term project willenable population synthesis analyses which require high quality optical regionfluxes of all known types of stars. We will observe all single stars in the BrightStar Catalogue (Hoffleit 1982) and its supplement (Hoffleit et al. 1983), stars inclusters and associations that pass within 45◦ of our zenith, and nearby stars withwell-determined distances to empirically calibrate the HR diagram. The latitudeof Fairborn Observatory is approximately N31.5◦. Two important auxiliaryprojects are:

A. Comparisons with Model Atmospheres: Model atmospheres analyticallylink the physical properties of stars (M, R, L, and composition) and the observedflux distributions and spectral line profiles. By comparing predictions of modelatmospheres with spectrophotometric fluxes and Balmer line profiles, effectivetemperatures and surface gravities can be found for a wide variety of stars.Our data should be far superior to existing data for this purpose. Comparisonsbetween the best-fitting model atmospheres calculated with different codes forthe same star will provide insight into how well each code reproduces theseobservables. By also deriving the elemental abundances, consistency checks canbe made. Hill, Gulliver & Adelman (1996) developed a powerful fitting programSTELLAR to perform a simultaneous rms error fit to the observed metallicand hydrogen line spectra as well as the continuous flux distribution of stars.Our complementary project is obtaining Balmer profiles of spectrophotometric


Figure 3. Similar to figure 2, but facing the back of the ASTRA Spec-trophotometer. The camera mirror is seen facing the prism. To the right isthe lens for focusing the zeroth order light. The brass circle is at the entranceof the guiding CCD housing.


Figure 4. The optical plate of ASTRA Spectrophotometer is on top, ratherthan on the bottom. The Apogee CCD is towards the back of the instrument.Surrounding it and towards the front, the plumbing for the 4

◦C water are

visible. The brass housing for the small guiding CCD is on the back face.


targets with the Dominion Astrophysical Observatory 1.22-m telescope’s coudespectrograph.

We recognize the need to include the effects of sphericity, NLTE, and veloc-ity fields for some types of stars. Further we want to add data from the infraredand from the ultraviolet to improve the effective temperatures and surface grav-ities. Hence we plan to use the infrared flux method.

B. Synthetic Colors and Line Indices from Spectrophotometry: By synthe-sizing colors from the spectrophotometry astronomers can check for consistencywith photometry and/or provide missing photometric indices (Breger 1976b).Systems of particular interest include Johnson UBV, Cousins RI, Sloan, ∆a,Stromgren, Geneva, and Vilnius/Stromvil. That our instrument will producedata without major gaps in wavelength coverage is a significant advantage overmost previous scanner flux data for color synthesis.

The strongest metal lines can be seen in continuous wavelength spectropho-tometry at somewhat lower resolution (e. g. 20 A, Fay et al. 1973). Hence onewill be able to create many line strength measures to determine the metallicity ofmany stars. Further for cool stars, one could measure the dependence of strongspectral features as functions of surface gravity, and [Fe/H] as have Burstein etal. (1986) and Gorgas et al. (1993) and use the results to study Population IIobjects.

There are many possible projects, a fair number of which will involve thedeterminations of Teff and log g or searches for companions for a particular kindof star.

Examples of other projects followA-type supergiants: sphericity effects, and pulsational propertiesConvection: ASTRA data of Am, Fm, and normal A and F stars should be

useful in helping to determine the efficiency of convection.Am stars: searches for broad, continuum features and binary signaturesHgMn stars: searches for signatures of peculiar compositions and binarity,

and the study of broad, continuum featuresMagnetic CP stars: variability around rotational periods, searches for sec-

ondary periods, details of broad, continuum features, use with Doppler imagingstudies, long term changes in variability characteristics

RR Lyrae stars and Cepheids: pulsational properties and searches for hy-drodynamical signatures

Reddening: need candidates with good MK classificationsExtraction of Composite Spectra: We will want to find some fast method

of identifying those stars with composite spectra. 2 MASS measurements mighthelp as well as the IRFM . Some systems will be obvious; others will only showsubtle effects. Spotted stars will cause problems.

Letting ASTRA Measure Fluxes of Stars With Fainter Stars in the Aper-ture: We are not calibrating as secondary standards any stars with other starswithin 7.5 mag. of the candidate in the aperture. For other targets the observingrestriction is 5.0 mag.

Fundamental stars with angular diameters : can extend previous work withnew observations.

Cooperation with Interferometric ProgramsEclipsing Binaries


Variability and physical mechanisms

8. Opportunities for Collaboration

The Fairborn Observatory site usually has the equivalent of at least 150 pho-tometric nights for each observing season from the middle of September to thebeginning of July. With ASTRA’s declination range for useful operation ofN80◦ to S35◦, we hope to obtain several 10s of thousands of observations peryear. Even using automated fitting routines, it is beyond the combined abili-ties of Adelman, Gulliver, and Smalley, who will be planning the observing andreducing the data to a usable form, to successfully analyze more than a smallfraction of the potential observations. As they realize that they will need help tomake the best scientific uses of the ASTRA data, they are interested in findingcollaborators.

The two major projects discussed above are basic to other applications andare natural parts/products of the calibration effort. The auxiliary projects maybe done in collaboration with others. The comparison of observations with modelatmospheres requires the calculation of grids of model atmospheres as well asobservations of Balmer line profiles. Observations of Hα, Hβ, Hγ, and perhapsother hydrogen lines at high dispersion will help us understand the usefulnessof those at spectrophotometric resolution and vice versa. The calculation ofsynthetic colors and line indices for all useful systems will become part of ourdata reduction effort.

We are particularly interested in using ASTRA to determine stellar param-eters and to investigate physical processes in stellar atmospheres. Spectropho-tometry will be especially useful to study those stars whose energy distributionschange. For variable stars it is important to realize that the time betweensuccessive observations can be as short as about 30 seconds for stars with Vmagnitudes in the 3.5 to 5 mag. range. Experimental programs to investigateall kinds of stars are appropriate.

Many studies of variable stars will likely utilize local spectrophotometricstandards so we want to begin their calibration as part of the initial effort.Thus we are interested in making some arrangements for collaboration beforethe start of observations. We are also very interested in analysis tools thatwill be useful for many different projects. As we are using LINUX and UNIXplatforms to run the various applications for ASTRA, we would like such toolsto run on these platforms.

9. Information for Possible Collaborators

1. All collaboration teams will have Adelman, Gulliver, and/or Smalley as amember(s) and as a coauthor(s) on all resulting papers. Two or all of them mayparticipate in a given project if they are particularly interested and contributemore than data and an understanding of its properties.

2. The ASTRA data will be analyzed and submitted for publication in atimely manner. To keep ASTRA operations funded requires the publication ofscientifically important and useful results. We will endeavor to provide sufficientobservations for each accepted collaboration in the first 1.5 years of scientific


operations so that an assessment of the project can be performed and an initialpublication can be written and submitted.

3. We are willing to obtain simultaneous as well as phase-dependent obser-vations.

4. We are interested in long term partnerships.5. All papers will be part of an ASTRA paper series.6. The spectrophotometric values will appear in a public catalog after a

substantial use of the data has been made. This catalog, coauthored by Adel-man, Gulliver, and Smalley, will include references to papers in which the datawere used.

7. Observations may be used for more than one program.8. Normally page charges, if any, will be the responsibility of the other

collaborators.If you are interested, please contract one of us as soon as possible. Our

email addresses follow:Saul J. Adelman ([email protected]),Austin F. Gulliver ([email protected]), andBarry Smalley ([email protected])

10. Status of ASTRA

We have written this paper to reflect the situation expected two months fromSeptember 15, 2006 when the spectrophotometer and its telescope are joined.The spectrophotometer has been fully tested in a laboratory and found to meetall of the specifications in optical and mechanical performance which were madebefore construction began. With the instrument on the telescope, our best guessfor the start of scientific observations is between March 1 and June 1, 2007.

Acknowledgments. This work is supported by NSF grant AST-0115612to The Citadel. This paper is ASTRA Paper number 5.

References

Adelman, S. J., Pyper, D. M., Shore, S. N., White, W. H., Warren, W. H., Jr. 1989,A&AS 91, 221.

Adelman, S. J., Rayle, R. E. 2000, A&A 355, 308.Ardeberg, A., Virdefors, B. 1980, A&AS 40, 307.Breger, M. 1976a, ApJS 32, 7.Breger, M. 1976b, ApJS 32, 1.Burstein, D., Faber, S. M., Gonzalez, J. J. 1986, AJ 91, 1130.Code, A. D., Meade, M. R. 1979, ApJS 39, 195.ESA 1997, The Hipparcos and Tycho Catalogues, SP-1200.Fay, T., Honeycutt, R. K., Warren, Jr., W. H. 1973, AJ 78, 246.Gorgas, J., Faber, S. M., Burstein, D., et al. 1993, ApJS 86, 153.Gunn, J. E., Stryker, L. L. 1983, ApJS 52, 121.Hayes, D. S., Latham, D. W. 1975, ApJ 197, 593.Hill, G., Gulliver, A. F., Adelman, S. J. 1996, in 5th Vienna Workshop on Stellar

Atmospheres and Spectrum Synthesis, eds. S. J. Adelman, F. Kupka, W. W.Weiss (San Francisco, ASP), ASP Conf. Ser. 105, 184.

Hoffleit, D. 1982, The Bright Star Catalogue, 4th edition (New Haven, CT, Yale Uni-versity Observatory).


Hoffleit, D., Saladyga, M., Wlasuk, P. 1983, A Supplement to the Bright Star Catalogue(New Haven, CT, Yale University Observatory).

Kurucz R. L., Avrett, E. H. 1981, Smithsonian Astrophys. Obs. Spec. Rep. 391.Kurucz, R. L. 1993, in Peculiar Versus Normal Phenomena in A-Type and Related

Stars, eds. M. M. Dworetsky, F. Castelli, R. Faraggiana (San Francisco, ASP),ASP Conf. Ser. 44, 87.

Oke, J. B. 1990, AJ 99, 1621.Oke, J. B., Gunn, J. E. 1983, ApJ 266, 713.Philip, A. G. D., Hayes, D. S., Adelman, S. J. 1988, eds. NEW DIRECTIONS IN

SPECTROPHOTOMETRY (Schenectady, L. Davis Press)Smalley, B., Gulliver, A. F., Adelman S, J. 2007, in The Future of Photometric, Spec-

trophotometric, and Polarimetric Standardization, ed. C. Sterken (San Francisco,ASP), ASP Conf. Ser. 999, xx.

Taylor, B. J. 1984, ApJS 54, 259.

astra spectrophotometry: design and overview · astra spectrometry: design and overview 3 indices....

Documents