Discovery of a Probable SX Phoenicis Star in M107 (NGC 6171)

0.26 MB·English

by  Thayne A. McCombs

Journalof the Southeastern Association forResearch in Astronomy,7,51-55,December1
(cid:13)c 2012. SoutheasternAssociationforResearchinAstronomy. Allrightsreserved.
DISCOVERY OF A PROBABLE SX PHOENICIS STAR IN M107 (NGC 6171)1
Thayne A. McCombs2
DepartmentofPhysicsandAstronomy,BrighamYoungUniversity,Provo,Utah,84602
Erik D. Reinhart2
DepartmentofPhysics,WillametteUniversity,Salem,Oregon,97301
and
3
1 Andrew N. Darragh, Elliott W. Johnson, and Brian W. Murphy
0 DepartmentofPhysicsandAstronomy,ButlerUniversity,Indianapolis,IN,46208
2
ABSTRACT
n
a Using V images taken in May and June 2012 with the SARA Consortium’s 0.9-meter telescope
J located at Kitt Peak National Observatory, we searched for variable stars in the globular cluster

4 M107(NGC 6171). The searchwasaccomplishedusing the ISISv2.2imagesubtractionsoftware. We
2 refinedthe positionsofthe previouslyknownvariablesandconfirmedthe 21RRLyraevariablesfrom
Clement’s Catalog of Variable Stars in Globular Clusters (Clement et al. 2001). We also discovered

] a previously unknown variable which is likely an SX Phoenicis star. For this SX Phoenicis star we
R
measuredafundamentalpulsationfrequency19.0122day−1(P =0.05257days)andameanamplitude
S of 0.046 magnitudes in the V band. This variable had an averageV-band magnitude of 17.72,nearly
.
h 2 magnitudes dimmer than the horizontal branch of M107, typical of SX Phoenicis stars and blue
p stragglers lying just beyond the main sequence turn-off in globular clusters.
- Subject headings: stars: variables: general–Galaxy: globular clusters: individual: NGC 6171
o
r
t
s 1. INTRODUCTION with 108 of them being newly discovered (Conroy et al.
a
2012; Toddy et al. 2012; Darragh & Murphy 2012). Of
[ Given the similar age, distance, and composition of
the total we identified we classified 111 RR0, 79 RR1, 3
theirstars,globularclustersprovideanaturallaboratory
1 RRd/e, 8 SX Phoenicis, 7 eclipsing, and 22 long period
for studying stellar evolution. Since globular clusters
v variables. SX Phoenicis stars are of particular interest.
are an older population of stars they have a large num-
0 They arestarssimilarto aδ-scuticlass variablestarbut
0 ber of post-main sequence stars, including many that lie
withlowermetallicity(PopulationII),anomalouslylarge
8 in the instability strip, particularly RR Lyrae variables.
masses and younger ages (Rodr´ıguez & Lo´pez-Gonza´lez
5 These unstable pulsating variable stars give us insights
2000). Compared to RR Lyrae stars they have a much
. into the ways that stars evolve after leaving the main
1 higherpulsationfrequencyduetotheirpositionnearthe
sequence. In addition by studying pulsating variables
0 mainsequence. Asmentionedby Mateoetal.(1990)SX
starsandFourierdecompositionoftheirlightcurvesitis
3 Phe stars are blue stragglers, likely the results of stellar
also possible to determine their composition, mass, and
1
mergerslyingjustabovethemain-sequenceturnoff,thus
: absolute luminosities (Simon & Clement 1993).
v M107 (NGC 6171) is one such globular cluster con- givingpast clues to the dynamicalhistory ofthe cluster.
i Using an image subtraction method developed by
X taining a number of variablestars. Clement’s Catalogof Alard(2000)wesearchedthecentral13′×13′ofM107for
Variable Stars in Globular Clusters lists 22 RR Lyrae
r variable stars using observations during May and June
a stars (15 RR0,7 RR1), a Mira variable, an LB vari-
2012. Usingthis method weconfirmedallthe previously
able, and an unclassified variable in M107 (Clement et
known variables in our field, as well as discovered the
al. 2001). Clement & Shelton (1997) found the periods
existence of an SX Phe star in M107.
andlight-curveFouriercoefficientsofsixteenoftheseRR
Lyrae stars. Given the relatively small number of vari- 2. OBSERVATIONSANDREDUCTION
ables previouslyfoundin M107andits locationnear the
Imageswere obtained using the SoutheasternAssocia-
celestialequatorwhereitisobservableinboththenorth-
tionforResearchinAstronomy(SARA)North0.9mtele-
ern and southern hemisphere, we chose M107 to search
scope located at Kitt Peak National Observatory. The
for undiscovered pulsating variable stars.
system uses an Apogee Alta U42 CCD detector with
In previous studies by this groupwe have searchedfor
2048×2048Kodake2VCC42-40with a gainof1.2elec-
variables in 3 other globular clusters; identifying, classi-
trons per count and a read noise of 6.3 electrons cooled
fying,andproducingdetailedlightcurvesof229variables
to approximately -30◦C. Using 2×2 binning this gave a
1Based on observations obtained with the SARA Observatory plate scale of 0.82′′/px and a 13.6′×13.6′ field of view.
0.9mtelescopeatKittPeaknationalObservatory,whichisowned Thesetupisidealforfindingvariablesinglobularclusters
andoperatedbytheSoutheastern AssociationforResearchinAs- because the aperture is large enough to find faint vari-
tronomy(http://www.saraobservatory.org).
2Southeastern Associationfor Research in Astronomy(SARA) ablesbutsmallenoughtousereasonableexposuretimes.
In addition we had regular access to the telescope, and
NSF-REUSummerIntern
Electronicaddress: [email protected], [email protected] could therefore obtain enough data to analyze the light
52 McCombs et al.
does not produce accurate results. In this equation F
ref
17.85 is the reference flux, ∆F is the differential flux, and C
0
isaconstant. Toremedythisweuseamethoddescribed
by Baldacci et al (2005) to calibrate the differential flux
17.80
from ISIS with instrumental magnitudes from DAOPHOT.
The calibration was performed using PSF fitting pho-
e
d17.75 tometry on the night of 20 June 2012 using the stan-
nitu 1d9a8lo7n,e19D9A4O)P.HIOnTtahnedDAALOLPFHROATMpEapckaacgkeagwese(Sutseetdsotnh,ePF.INBD.
g
a17.70 procedure to find stars above a threshold of 3.5σ where
M
σ is the standard deviation of the sky background. We
17.65 then used the PHOT procedure which performed initial
aperture photometry on the stars found. However since
the field is crowded the results of aperture photometry
17.600.05 0.10 0.15 0.20 0.25 0.30 0.35 arenotacceptable,thereforeweusedPSFfittingformore
JD +2.4560986×106 accurate photometry. This was done using PICK to se-
lect the 25 brightest stars which were not overexposed
Fig. 1.— For the night of June 20, 2012, our longest night, we
and did not overlap with any brighter stars, followed
plotthemeasuredmagnitudesoftheSXPhestaragainsttheJD.
curves and determine periods of the variables. by PSF to construct a semi-empirical model of the PSF
Observations were made using a Bessel V filter on 23 for each frame. DAOMATCH was then used to produce a
May 2012 and 7, 8, 9, and 20 June 2012. Typical seeing rough transformation between the frames, followed by
was 2.5′′ and generally between 2′′ and 3.5′′. Exposures DAOMASTER to increase the accuracy of the transforma-
on 23 May were 120 seconds long, and all exposures in tion. After obtaining transformations we created a me-
Junewere90seconds. Figure1showstheunphasedlight dian image using MONTAGE2. Following this step we ran
curve of a low amplitude SX Phoenicis star on 20 June, FIND,PHOT,PICK,andPSFonthemedianimageandthen
our longest night. ALLSTAR to perform PSF fitting photometry on the me-
We performed standard image reduction using Maxim dianimage. FollowingthesubtractionofthePSFsforall
DL to apply bias, dark, and flat frames. In order to found stars from the original image we again ran FIND
reduce the number of false positives in ISIS we ran the andPHOTonthesubtractedimage,andaddthenewstars
frames through Maxim DLs hot and bad pixel removal to the star list, andperformALLSTARonthe originalim-
kernels,andusedthecraveragecosmicrayremovaltask age with the revised star list. This process is repeated
inIRAF.Wealsorejectedimageframescontainingsatel- until all stars have been added. We then ran ALLSTAR
lite trails, star trails from poor guiding, or seeing worse on all of the individual frames. Finally ALLFRAME was
than 3.5′′. applied to the transformation list to improve the PSF
fits for each frame by looking at all of the frames simul-
3. ANALYSIS taneously. DAOPHOT does not take into account the JD,
3.1. Image Subtraction airmass,or exposure time, so scripts were written to ex-
tract this information from the headers.
Image subtraction was done using the ISIS v2.2 soft-
After obtaining the instrumental magnitudes for the
ware package (Alard & Lupton 1998; Alard 2000) using
night of 20 June 2012we found the instrumental magni-
the same procedure used by Conroy et al. (2012). This
tude for the SX Phe star from the reference frame used
packageusesanumberofprogramsandc-shellscriptsto
in ISIS. This was accomplishedusing the procedure out-
align many frames together, perform image subtraction,
lined above, however since we only had a single frame
findobjectswithvariability,andproducelightcurvesfor
we did not perform the steps for DAOMATCH, DAOMASTER,
those objects.
MONTAGE2, or ALLFRAME, and instead used only ALLSTAR
ISIScompensatesforvariableseeingconditionsbycon-
to perform the photometry. However multiple runs were
volvinga highqualityreferenceframe witheachindivid-
usedtofindallstars,aswasdonewiththemedianimage.
ual frame, and then performing the subtraction. The
Next we attempted to fit the differential flux to the
SIGTHRESH parameter controls the threshold for find-
instrumental magnitude for this night as described by
ing variable objects. We modified this parameter until
Baldacci et al (2005). Since DAOPHOT uses a zero point
roughly a thousand objects were found, then manually
of 25 for its photometry we calculated the reference flux
examined the light curves to extract true variables from
for the SX Phe star from the reference frame using
the more numerous false positives.
Our reference frame was produced by combining the 7 F =10(25−minst)/2.5 (2)
best images from the night of 6 June 2012,because that
was the night with the best seeing. where m is the instrumental magnitude from the ref-
inst
erence frame. After extracting the instrumental magni-
3.2. Magnitude Calibration
tudes from the output of ALLFRAME, we used a program
ISIS outputs its photometry in the form of differential writteninC++whichattemptedseveralfitsofthediffer-
flux (the difference in flux between the reference frame entialfluxtoinstrumentalmagnitudeusingthe equation
and the image frame). Unfortunately, as Baldacci et al given by Baldacci et al (2005):
(2005) pointed out, a simple conversion from the differ-
ential flux to magnitude by mi =−2.5log[Fi,ref −A∆Fi]+C (3)
m=−2.5log[F −∆F]+C (1) varying A between 0.5 and 2, and C between 24 and 26.
ref 0
SX Phe in M107 53
TABLE 1
Amplitude variationsforthe SXPhe star
NightofObservation(JD) Amplitude(VMagnitude)
2456070 0.054
2456085 0.052
2456086 0.053
2456087 0.032
2456098 0.046
We optimized A and C to minimize the test statistic
(m −m )2
i,ALLF i,ISIS
X n (4)
i
where n is the number of acceptable data points. Data
points are acceptable if the residual for the current fit
(m −m )islessthan0.1magnitudes. Further-
i,ALLF i,ISIS
moreanyfitwheremorethan25%ofthedatapointsare
not acceptable is rejected.
Once the values of A and C were obtained, we con-
verted all of the ISIS data from fluxes to magnitudes
using equation 3. Although this method worked well
for the previously discovered variables, the SX Phe was Fig. 2.—PositionsofvariablesinM107
two faint for DAOPHOT to find on the individual frames, 0.00176 day−1. Two outliers were removed that obvi-
although we could perform photometry for it on the ref- ously were spurious points in the light curve. The av-
erence frame. Therefore we used the median values of A erage magnitude of the SX Phe star was also obtained
and C from the values for the other variables. This is fromPeriod04. Theresultsforotherharmonicswereun-
not a bad estimate since the values for A and C do not reliableduetothefaintnessoftheobjectandthequality
vary significantly in the field of view, and in this case at of the data.
least,themedian,modeandmeanareidenticaltowithin
0.01. We obtained the values A = 1.09 and C = 25.19, 3.4. Astrometry
both of which are close to the expected values of 1 and
AfteridentifyingvariablestarsusingISIStheircoordi-
25 respectively.
nates were in units of pixels. To convertto conventional
Finally to transform from instrumental magnitude to
celestial coordinates we used the ccmap and ccsetwcs
apparent magnitude we use
tasks in IRAF to apply a world coordinate system to
V =V −25+2.512log(t )+22.633 (5) our alignment image. We used a set of nine stars from
inst exp
the NOMAD catalog to match the physical coordinates
where V is the apparent magnitude, Vinst is the instru- of our frame with the fk5 celestial coordinate system.
mentalmagnitude,texpistheexposuretimeoftheframes We then used the center task to ensure that the pixel
usedforthe calibrationwith DAOPHOT,whichwas90sec- coordinates for the variable stars were centered on the
onds. The 22.633 term is the zeropoint of 25 used by centroid. Finally we converted the pixel coordinate sys-
DAOPHOT minus a measured offset for our system. This tem to celestial coordinates. Figures 2, 3, and 4 show
givesusanapproximateapparentmagnitudeof17.72for our field of view and subsets of our field of view, along
the SX Phe star. with the positions of known variables. The SX Phe star
For the RR Lyrae variables that were in our field we is indicated as number 26.
appliedthesameprocedure. WefoundamedianV mag-
nitude for these variables of 15.72, two full magnitudes 4. RESULTS
brighter than our suspected SX Phe star. Given that Usingthe imagesubtractionpackageISISwefoundall
RR Lyrae stars are horizontal branch stars this would of the variables documented by Clement et al. (2001)
also be the typical magnitude of the horizontal branch in our field and verified their classifications. We also
for M107. Typically the main-sequence turnoffwouldbe discovered an SX Phe star located at 16h32m32.69s,
3.5 magnitudes dimmer than the horizontal branch, or −13◦02′58.92′′. It was measuredto have a pulsationfre-
roughly 19.2. quency of 19.0122 day−1 (0.05257 days) with an ampli-
tude of 0.046 magnitudes, roughly 0.1 magnitudes from
3.3. Frequency Analysis
minimum to maximum. The average V magnitude was
Once our differential fluxes were converted into mag- 17.72. Using Period04’s Monte Carlo simulation we es-
nitudes via the procedure described in section 3.2 they timated the errors for the frequency and magnitude as
wereanalyzedusingthesoftwarepackagePeriod04(Lenz 0.0011 and 0.0031, respectively. The Monte Carlo er-
&Breger2005). Period04wasusedtodeterminethefre- rorfor frequency is less than our exposure time, and the
quency and amplitude of the fundamental frequency of error for magnitude is less than our error for the trans-
the SX Phe star. We searched for the fundamental fre- formation,whichsuggestsourresultsarelimitedprimar-
quency in the range 0 to 50 day−1 with a step size of ily by observational error. The frequency, V magnitude,
54 McCombs et al.
TABLE 2
Astrometry andclassification of variablesin M107
Variablea RA(2000)b Dec(2000) Type
V1c 16h 32m 24.340s -13◦ 11′ 55.60′′ M
V2c 16h 32m 41.780s -13◦ 09′ 42.40′′ RR0
V3 16h 32m 16.656s -13◦ 06′ 23.60′′ RR0
V4 16h 32m 25.237s -13◦ 05′ 55.52′′ RR1
V5 16h 32m 47.887s -13◦ 05′ 57.10′′ RR0
V6 16h 32m 31.290s -13◦ 04′ 25.60′′ RR1
V7 16h 32m 34.906s -13◦ 04′ 18.67′′ RR0
V8 16h 32m 32.800s -13◦ 03′ 59.93′′ RR0
V9 16h 32m 30.181s -13◦ 03′ 37.84′′ RR1
V10 16h 32m 28.054s -13◦ 03′ 10.32′′ RR0
V11 16h 32m 32.574s -13◦ 02′ 44.98′′ RR0
Fig. 3.— Positions of variables in the core of M107, star 26 is V12 16h 32m 35.984s -13◦ 02′ 16.57′′ RR0
thenewSXPhestar V13 16h 32m 30.095s -13◦ 02′ 06.09′′ RR0
V14 16h 32m 33.118s -13◦ 01′ 55.53′′ RR0
V15 16h 32m 33.182s -13◦ 01′ 17.45′′ RR1
V16 16h 32m 27.326s -13◦ 01′ 24.86′′ RR0
V17 16h 32m 25.146s -13◦ 02′ 07.07′′ RR0
V18 16h 32m 37.163s -12◦ 59′ 41.75′′ RR0
V19 16h 32m 47.839s -13◦ 00′ 32.52′′ RR1
V20 16h 32m 34.060s -13◦ 02′ 26.23′′ RR0
V21 16h 32m 37.579s -13◦ 05′ 42.22′′ RR1
V23 16h 32m 13.917s -13◦ 03′ 01.46′′ RR1
V24 16h 32m 31.958s -13◦ 03′ 09.73′′ RR0
V25d ··· ··· Lb
V26e 16h 32m 32.660s -13◦ 03′ 00.38′′ SXPhe
a
We usedthe same variablenumbersas Clementet al. (2001), with
Fig. 4.—AfinderchartzoomedinonthenewSXPhestar(26) theexceptionofV26whichwasnotinClement’scatalog.
b
Our coordinates differed by a few arcseconds from Clement et al.
(2001),likelyduetotheirusingphotographicplates.
c
V1 and V2 are outside of our field of view, the coordinates and
variabletypearetakenfromClementetal.(2001),andthecoordinates
maybeinaccurate.
d
V25wasdiscoveredbyLloydEvans&Menzies(1973),butnocoor-
dinateswereavailable,andwedidnotdetectthisstar,possiblydueto
17.60 aelongperiod.
V26wasdiscoveredinthisstudy.
stragglers are found.
17.65 Figure 1 shows our observations for June 20 (our
longest night) and Figure 5 shows all of the data com-
binedandphased. As canbeseenfromtheFigure1and
e
d17.70 Figure5, althoughthe frequency appearsconsistent, the
u
nit amplitude may be modulating. Using Period04 we de-
g termined amplitudes for each night separately, with the
a
M17.75 results listed in Table 1.
When performing astrometry we discovered that the
coordinates given by Clement et al. (2001), which orig-
17.80 inated from Oosterhoff (1938) were inaccurate by a few
arcseconds. Given that these coordinates were obtained
in 1938 with photographic plates this inaccuracy is not
17.85 surprising. Here we present revised coordinates in Table
2. Two of the variables, V1 and V2, were outside of our
0.0 0.2 0.4 0.6 0.8 1.0 field of view. Their coordinates are taken from Clement
Phase
et al. (2001).
Fig. 5.— The data from all nights was phased using Period04 5. CONCLUSION
with the estimated period of 0.05257 days and plotted together.
We have verified and refined the positions of 21
Differentnightsarerepresenteddifferentbycolors.
previously known variables in M107 as documented
and amplitude are consistent with the variable being an by Clement et al. (2001). We have also dis-
SX Phe star in a globular clusters. It is 2 magnitudes covered a probable new SX Phe star located at
dimmerthanthehorizontalbranchofM107. Thiswould 16h32m32.69s, −13◦02′58.92′′ with fundamental fre-
placeitroughly1.5magnitudesabovethemainsequence quency of 19.01221 day−1 (0.05257 days) with a mean
turn-off which is typically where SX Phe stars and blue V magnitude and amplitude of 17.72 and 0.046 magni-
SX Phe in M107 55
tudes, respectively. The discovery of this faint variable ableandPStetsonforusefuladviseinusingDAOPHOT
shows the power of using ISIS to look for variable stars. and ALLFRAME. This project was funded in part by
Inadditionwehaveimprovedtheastrometryforthepre- the National Science Foundation Research Experiences
viously discoveredvariables in M107. for Undergraduates (REU) program through grant NSF
Unfortunately the SX Phe star was too faint for us to AST-1004872. AdditionallyA.Darragh,E.Johnson,and
obtain a high signal-to-noise with our observations, and B.Murphy werepartially funded by the Butler Institute
therefore we were limited in the amount of frequency forResearchandScholarship. TheauthorsalsothankF.
analysis we could perform. Levinsonfora generousgiftenablingButler University’s
membership in the SARA consortium.
We thank C. Alard for making ISIS 2.2 publicly avail-
REFERENCES
Alard,C.,&Lupton,R.H.1998,ApJ,503,325 Mateo,M.,Harris,H.C.,Nemec,J.,&Olszewski,E.W.1990,AJ,
Alard,C.2000,A&AS,144,363 100,469
Baldacci,L.,Rizzi,L.,Clementini,G.,&Held,E.V.2005, A&A, Oosterhoff,P.T.1938,Bull.Astron.Inst.Netherlands,8,273
431,1189 Rodr´ıguez,E.,&L´opez-Gonza´lez, M.J.2000, A&A,359,597
Clement,C.M.,Muzzin,A.,Dufton,Q.,etal.2001,AJ,122,2587 Simon,N.R.,&Clement,C.M.1993,ApJ,410,526
Clement,C.M.,&Shelton,I.1997,AJ,113,1711 Stetson,P.B.1987,PASP,99,191
Conroy,K.E.,Darragh,A.N.,Liu,Z.J.,&Murphy,B.W.2012, Stetson,P.B.1994,PASP,106,250
JSARA,5,34 Toddy, J. M.,Johnson, E. W.,Darragh, A.N.,& Murphy, B. W.
Darragh,A.N.,&Murphy,B.W.2012,JSARA,6,72 2012,JSARA,6,63
LenzP.,BregerM.2005, CoAst,146,53
LloydEvans,T.,&Menzies,J.W.1973,IAUColloq.21:Variable
StarsinGlobularClustersandinRelatedSystems,36,151