White dwarf and hot sub dwarf binaries
as p ossible progenitors of
type I a Sup ernovae
Christian Karl
July
White dwarf and hot sub dwarf binaries
as p ossible progenitors of
type I a Sup ernovae
Den Naturwissenschaftlichen Fakult aten
der Friedrich Alexander Universit atErlangen N urnberg
zur
Erlangung des Doktorgrades
vorgelegt von
Christian Karl
aus Bamberg
Als Dissertation genehmigt von den Naturwissenschaftlichen Fakult aten
der Universit atErlangen N urnberg
Tag der m undlichen Pr ufung Aug
Vositzender der Promotionskommission Prof Dr L Dahlenburg
Erstb erichterstatter Prof Dr U Heb er
Zweitberichterstatter Prof Dr K Werner
Contents
The SPY pro ject
Selection of DB white dwarfs
Color criteria
Absorption line criteria
Summary of the DB selection
The UV Visual Echelle Sp ectrograph
Instrumental setup
UVES data reduction
ESO pip elin e vs semi automated pip eline
Derivation of system parameters
De nition of samples
Radial velocity curves
Follow up observations
Radial velocity measurements
Power sp ectra and RV curves
Gravitational redshift
Quantitative sp ectroscopic analysis
Stellar parameters of single line d systems
Stellar parameters of double line d systems
Pro jected rotational velocities
Metal abundances of sdB stars
Double lined white dwarfs
Systems analyzed from their radial velocity curves alone
WD
HE
WD
Systems that cannot b e solved from their RV curves alone
HE a system solved through an additional sp ectral analysis
WD a system in need of an additional sp ectral analysis i
ii CONTENTS
White dwarfs with M dwarf companions
WD alias RR Caeli
WD
HS
HE
Single lined systems
White dwarf primaries
WD
WD
HS
WD
Sub dwarf B primary
WD
HE
HE
HE
HE
HE
HE
Sub dwarf O primary
HE
HE
HE
Peculiar ob jects
PN G alias EGB a central star of a planetary nebula
WD a white dwarf with a p eculiar H pro le
Stars with small RV amplitudes
HE
WD
WD
WD
WD
WD
HS
Conclusion and outlo ok
Conclusion
Outlo ok
A Selected DB candidates from the Hamburg ESO Survey
CONTENTS iii
B Follow up campaigns
B Campaigns at the DSAZ m telescop e
B Campaigns at the ESO NTT
B Campaigns at the ESO VLT
B Campaigns at the WHT and INT
C Radial velocity measurements
D Equivalent widths measurements
E List of abbreviations
List of Figures
Sp ectrum of a type Ia Sup ernova
Lightcurve of a type Ia Sup ernova
Progenitors of type Ia Sup ernova
Sample UVES sp ectrum of a hydrogen rich DA white dwarf
White dwarf distribution
SPY detection probability as a function of orbital p erio d
DB selection Color criteria
DB selection Signi cance threshold
DB selection Cuto criteria
The UV Visual Echelle Sp ectrograph
Sample UVES CCD exp osure
UVES reduction pattern
Order de nition for UVES sp ectra
Overlapping unblazed echelle orders
The ripple problem
Reduced UVES sp ectrum ESO pip eline vs our own semi automated pip eline
RV determination by means of MIDAS ts
RV curves of a single and a double lined system
Mass radius relation for white dwarfs
Evolutionary tracks for DA white dwarfs
Evolutionary tracks for sdB stars
Sample mo del ts of double lined systems
Analysis of a double lined system
WD Observed line pro le
WD Best t RV curve and p ower sp ectrum
HE Best t RV curve and p ower sp ectrum
WD Best t RV curve and p ower sp ectrum
HE Line shift within one night
HE Best t RV curve and p ower sp ectrum
HE Sample mo del atmosphere ts
WD Observed line pro les
WD Best t RV curve and p ower sp ectrum iv
LIST OF FIGURES v
Sample UVES sp ectrum of a white dwarf with an M dwarf companion
WD Best t RV curve and p ower sp ectrum
WD Emission lines
WD Splitted Balmer lines
WD Best t RV curve and p ower sp ectrum
HS Best t RV curve and p ower sp ectrum
HS Observed line pro les
HE Best t RV curve and p ower sp ectrum
Analysis of single lined systems
WD Best t RV curve and p ower sp ectrum
WD Best t RV curve and p ower sp ectrum
HS Best t RV curve and p ower sp ectrum
WD Best t RV curve and p ower sp ectrum
Sample UVES sp ectrum of an sdB star
Sample mo del t of an sdB star
Determination of the microturbulence
Pro jected rotational velocity v sin i
rot
WD Best t RV curve and p ower sp ectrum
WD Metal abundance pattern
HE Best t RV curve and p ower sp ectrum
HE Metal abundance pattern
HE Post RGB evolutionary scenario
HE Best t RV curve and p ower sp ectrum
HE Metal abundance pattern
HE Best t RV curve and p ower sp ectrum
HE Best t RV curve and p ower sp ectrum
HE Metal abundance pattern
HE Best t RV curve and p ower sp ectrum
HE Metal abundance pattern
HE Best t RV curve and p ower sp ectrum
HE Metal abundance pattern
Sample UVES sp ectrum of an sdO star
HE Best t RV curve and p ower sp ectrum
HE Best t RV curve and p ower sp ectrum
HE Best t RV curve and p ower sp ectrum
PN G or EGB T log g diagram
e
PN G or EGB Best t RV curve and p ower sp ectrum
WD Observed H line pro les
WD Best t RV curve and p ower sp ectrum
HE Best t RV curve and p ower sp ectrum
Best t RV curves
o
P M Diagram for i
o
P M Diagram for i
List of Tables
Follow up observations
WD System comp osition
HE System comp osition
WD System comp osition
HE Results of the mo del atmosphere analysis with fitsb
WD System parameters and derived results
WD System comp osition
HS Scenarios
HE Scenarios
WD Scenarios
WD Scenarios
HS Scenarios
WD Scenarios
WD Metal abundances
WD Scenarios
HE Metal abundances
HE Scenarios
HE Metal abundances
HE Scenarios
HE Scenarios
HE Metal abundances
HE Scenarios
HE Metal abundances
HE Scenarios
HE Metal abundances
HE Scenarios
HE Scenarios
HE Scenarios
HE Scenarios
PN G or EGB Scenarios
WD Scenarios
HE Scenarios vi
LIST OF TABLES vii
WD Scenarios
WD Scenarios
WD Scenarios
WD Scenarios
WD Scenarios
HS Scenarios
A Selected DB candidates from the HES
C Radial velocity measurements
D Equivalent width measurements
E Abbreviations
viii LIST OF TABLES
Zusammenfassung
Die vorliegende Arb eit b esch aftigtsich haupts achlich mit der Analyse Wei er Zwerge und unter
leuchtkr aftigerB und O Sterne in Dopp elsternsystemen Dab ei soll anhand der Beobachtungs
daten vor allem gekl artwerden ob die untersuchten Systeme m ogliche Vorl aufervon Sup ernovae
des Typs Ia darstellen
Sup ernovae des Typs Ia SNe Ia spielen eine Schl usselrolle in vielen Bereichen der mo der
nen Astronomie Sie sind die wichtigsten Pro duzenten von Eisen im Universum und dienen
als Standardkerzen au erdem zur Bestimmung extragalaktischer Entfernungen In letzterem
Falle wird davon ausgegangen da alle SN Ia Explosionen ahnliche Lichtkurven insb esondere
gleiche absolute Maximalhelligkeiten aufweisen Aus der tats achlich auf der Erde gemessenen
scheinbaren Helligkeit kann so auf die Entfernung geschlossen werden in der die b eobachtete Ex
plosion stattfand Derartige Messungen erm oglichten es den Astronomen in den letzten Jahren
den Wert der Hubblekonstanten mit bisher nie erreichter Pr azision zu b estimmen Dar uber
hinaus ergab en sich die ersten Hinweise auf ein b eschleunigt expandierendes Universum
Trotz ihrer herausragenden Bedeutung ist no ch immer nicht zweifelsfrei gekl art wie Su
p ernovae vom Typ Ia entstehen Als sicher gilt da es sich b ei einem SN Ia Ereignis um
die Explosion eines Wei en Zwerges handelt der eine kritische Masse uberschreitet Diese ist
wahrscheinlich die sogenannte Chandrasekharmasse M welche die Maximalmasse eines
Chandra
Wei en Zwerges und den Ub ergang zum Neutronenstern charakterisiert Allerdings kann sich
ein Wei er Zwerg dieser Grenzmasse nur ann ahern wenn er als Komp onente eines Dopp elstern
systems Masse von einem Begleiter erh alt Die genaue Art dieses Dopp elsternsystems v a
die Natur des Begleiters ist allerdings no ch nicht eindeutig gekl artund Gegenstand aktueller
Forschung Bisher werden zwei Szenarien favorisiert
Im sogenannten Single Degenerate Szenario b esteht das Dopp elsternsystem aus einem
Wei en Zwerg und einem Hauptreihen o der Riesenstern Von Letzterem ndet ein kon
tinuierlicher Massen uberstrom auf den Wei en Zwerg statt bis dieser seine kritische Masse
erreicht und in einer Sup ernovaexplosion vergeht Dieses Szenario ist konform mit der An
nahme da jede SNe Ia Explosion eine genau de nierte absolute Maximalhelligkeit b esitzt
Im sogenannten Double Degenerate Szenario ist der SN Ia Vorl auferTeil eines aus zwei
Wei en Zwergen b estehenden engen Dopp elsternsystems Dieses kann durch Gravita
tionsstrahlung Drehimpuls abgeb en was schlie lich zu einem Verschmelzen b eider Sterne
f uhrt Ein derartiges System gilt als Sup ernovavorl aufer falls b eide Komp onenten in
nerhalb einer Hubblezeit T fusionieren und deren Gesmatmasse die Chandrasekharmasse
H
uberschreitet Nach diesem Szenario sind auch sehr viel massereichere Sup ernovae denkbar
die leuchtkr aftigerw arenals bisher angenommen
No ch ist nicht gekl art welches der b eiden Szenarien der Realit at am ehesten entspricht
Denkbar ist auch da b eide relevant sind Um ab er Typ Ia Sup ernovae auch im jungen Univer
sum als Standardkerzen nutzen zu k onnen m ussenEntwicklungse ekte b er ucksichtigt werden
Daher ist es unbedingt notwendig zu ermitteln wie wichtig der Beitrag des Double Degenerate
Szenarios f urdie Bildung von Sup ernovae des Typs Ia tats achlich ist
Sup ernova Ia Progenitor SurveY Um diese Problematik n aherzu analysieren wurde das
SPY Pro jekt ins Leb en gerufen siehe Napiwotzki et al a mit dem Ziel das Dou
ble Degenerate Szenario durch Beobachtungen zu uberpr ufen Eine gro e Anzahl m oglicher
SN Ia Kandidaten manche eigens f ur SPY aus dem Hamburg ESO Survey HES selek
Very Large Telescop e VLT der Europ aischen S udsternwarte Europ ean tiert wurde am
Southern Observatory ESO b eobachtet Der UV Visual Echelle Sp ectrograph UVES und
eine selbstentwickelte halbautomatische Reduktionsroutine lieferten ho chaufgel oste Sp ektren
anhand derer nach Wei en Zwergen mit variablen Radialgeschwindigkeiten gesucht wurde Bei
der untersuchten Wei en Zwerge wurden Geschwindigkeitsvariationen gefunden die auf
einen unsichtbaren Begleiter hindeuten der mit dem im Sp ektrum sichtbaren Hauptstern den
gemeinsamen Schwerpunkt uml auft der untersuchten Ob jekte waren sogar dopp ellinige
Systeme deren Sp ektren direkte Anzeichen eines Begleiter aufweisen Bis dato waren nur sechs
solcher Systeme b ekannt Insgesamt konnte mit den neuentdeckten Systemen die Zahl der
b ekannten Wei en Zwerge in Dopp elsternen mehr als versiebenfacht werden
Parallel zum Survey wurde b egonnen die gefundenen Systeme am m Teleskop des Deutsch
Spanisches Astronomisches Zentrum DSAZ am ESO VLT am New Technology Telecscop e
William Herschel Telescop e WHT nachzubeobachten und auf ihre NTT der ESO und am
Eignung als m ogliche Vorl aufereiner Typ Ia Sup ernova hin zu untersuchen
Im Rahmen der vorliegenden Arb eit wurden sowohl m ogliche SN Ia Kandidaten aus dem
HES selektiert Kapitel als auch eine neue und verbesserte Reduktionsroutine f ur UVES
Daten entwickelt Kapitel Hauptziel war es ab er Nachbeobachtungen am DSAZ und NTT
durchzuf uhren und zu analysieren Kapitel bis
Besonders sechs dopp ellini ge Systeme sind hervorzuheben deren Sp ektren b ereits R uck
schl usseauf die Bewegung b eider Komp onenten des Dopp elsterns zulassen Daher konnte durch
eine Analyse der Radialgeschwindigkeitskurven das Massenverh altnis M M b eider Kom
prim sec
p onenten b estimmt werden F urvier der Systeme konnten zus atzlich Unterschiede in den Gravi
tationsrotverschiebungen der Einzelsterne dazu genutzt werden deren absolute Massen und die
Zeit t bis zu deren Verschmelzung zu b estimmen Diese Metho de funktionierte allerdings nicht
m
f urdie verbliebenen dopp ellini gen Systeme da sich deren Rotverschiebungsdi erenzen als nicht
aussagekr aftigerwiesen Um die Zahl der freien Systemparameter weiter einzuschr anken wurde
deshalb eine quantitative Sp ektralanalyse versucht eine Untersuchungsmethode die bislang
nur auf Einzelsterne angewandt wurde Da es sich b ei Sp ektren dopp ellini ger Systeme um eine
Ub erlagerung zweier Einzelsternsp ektren handelt erforderte diese Analyse die Entwicklung eines
neuartigen Programms F urHE gelang mit dessen Hilfe schlie lich eine Sp ektralana
lyse allerdings nur da aufgrund der Ahnlichkeit b eider Komp onenten weitere freie Parameter
eliminiert werden konnten Das letzte dopp ellinige System WD bleibt dagegen no ch
ungel ost
Vor allem WD erwies sich als ausgezeichneter SN Ia Vorl auferkandidat Beide
Komp onenten des aus zwei Wei en Zwerge b estehenden Dopp elsternsystems werden inner
halb einer halb en Hubblezeit zu einem Ob jekt verschmelzen das die Chandrasekharmasse
uberschreitet Damit erf ullt WD als eines von nur drei Systemen alle Kriterien die
vom Double Degenerate Szenario an einen m oglichen SN Ia Vor aufergestellt werden Im Rah
men des SPY Pro jektes wurden dar uber hinaus no ch zwei weitere Systeme gefunden welche die
Quali kationskriterien nur knapp verfehlen Den Anteil der im Double Degenerate Szenario
erzeugten SNe Ia k onnenzwar erst Mo dellrechnungen exakt b estimmen ab er b ereits jetzt d urfte
aufgrund der gefundenen Beobachtungsergebnisse klar sein da dieses Szenario als m oglicher
wenn nicht gar entscheidender Weg zur Bildung von Typ Ia Sup ernovae angesehen werden mu
Zu den b ereits b eschriebenen dopp ellini gen Systemen wurden no ch weitere einlinige Sys
teme untersucht deren Sp ektren keine zweifelsfreie Analyse der Bewegung der zweiten Kom
p onente zulassen Denno ch ergab en Radialgeschwindigkeitsanalysen f ur dieser Systeme ein
deutige Ergebnisse f ur die Bahnparameter w ahrend f ur die verbleibenden neun Systeme auf
grund des vorliegenden Datenmaterials keine eindeutigen L osungen gefunden werden konnte
F ur Letztere sind deshalb no ch zus atzliche Beobachtungen n otig um eine Aussage uber ihre
Systembescha enheit tre en zu k onnen
Unter den eindeutig gel ostenSystemen stellten sich drei als Wei e Zwerge mit M Haupt
reihensternen als Begleiter heraus die somit nach dem Double Degenerate Szenario nicht
als Typ Ia Sup ernovavorl aufer in Frage kommen Die ubrigen Systeme konnten nur uber
ihre Massenfunktion weitergehend untersucht werden Diese Funktion verkn upft die Massen
M und M b eider Komp onenten eines Dopp elsternsystems mit dem Inklinationswinkel i
vis invis
W ahrend sich die Massen M der sichtbaren Komp onente entweder durch eine quantitative
vis
Sp ektralanalyse b estimmen o der aus Evolutionsrechnungen ableiten lassen bleibt die Inklina
tion i unbestimmt da ihre Festlegung au er f urBedeckungsver anderliche kaum m oglich ist Die
einzig weitere M oglichkeit der Inklinationsmessung bieten b esonders enge Systeme b ei denen
die Sternrotation durch Gezeitenkr afte wie b eim Erde Mond System mit der orbitalen Bahn
b ewegung synchronisiert ist Mit Hilfe des Dopplere ekts kann aus den Sp ektren die pro jizierte
Rotationsgeschwindigkeit b estimmt und durch Vergleich mit der Bahnbewegung schlie lich auf
den Inklinationswinkel geschlossen werden Diese Metho de wurde auf sechs unterleuchtkr aftige
B Sterne des Samples angewandt deren schmale Metallinien sich b esonders gut f ureine derartige
Analyse eignen Die Auswertung ergab allerdings nur f ur HE sinnvolle Werte Bei
den ubrigenOb jekten stellte sich entweder die Annahme gebundener Rotation als falsch heraus
o der es zeigte sich da die Messungen der Rotationsgeschwindigkeiten aufgrund der nicht genau
b ekannten sp ektralen Au osungder Beobachtungsdaten nicht zuverl assiggenug waren
Wegen der Unbestimmtheit der Inklination wurden die Dopp elsternsysteme zun achst unter
o
der Annahme maximaler Inklination i untersucht Da diese Annahme in den Mini
malmassen der unsichtbaren Komp onenten resultiert werden die wirklichen Massen stets unter
sch atzt In diesem Sinne ergibt diese Annahme das strikteste Auswahlkriterium f ur eventuelle
SN Ia Kandidaten Zwar werden p otentielle SN Ia Kandidaten m oglicherweise nicht erkannt
ab er auch keine ungeeigneten Systeme als solche fehlklassi ziert
o
Des Weiteren wurden die Systeme unter der Annahme eines Inklinationswinkels von i
untersucht Dieser Winkel ist der statistisch wahrscheinlichste Wert der Inklination und liefert
somit auch die wahrscheinlichste Begleitermasse
o
Wie sich zeigte konnte b ei den einlini gen Systemen weder unter der Annahme i no ch
o
i ein System gefunden werden das sich als SN Ia Kandidat eignet Vier der untersuchten
o
Dopp elsternsysteme k onnten allerdings f urWinkel kleiner als die Voraussetzungen erf ullen
und als SN Ia Vorl aufergelten Die Perioden der ubrigenzehn Systeme sind jedo ch zu gro als
da selbst der massereichste Wei e Zwerg M M no ch innerhalb einer Hubblezeit
invis Chandra
mit der sichtbaren Komp onente verschmelzen k onnten
Als Neb enpro dukt der Radialgeschwindigkeitsanalysen wurden chemische H au gkeitsmuster
der untersuchten Sterne b estimmt F urf unfder untersuchten Ob jekte decken sich die gefundenen
H au gkeiten mit denen normaler unterleuchtkr aftigerB Sterne Nur HE el durch
seine ungew ohnliche Titananreicherung auf Solche p ekuli aren H au gkeitsmuster wurden nur
b ei drei anderen Ob jekten b eobachtet verstanden sind sie allerdings no ch nicht
Bislang sind erst etwa aller von SPY gefundenen Dopp elsterne nachbeobachtet und
analysiert Allerdings zeigen b ereits die in dieser Arb eit erzielten Ergebnisse deutlich da das
Double Degenerate Szenario f ur die Entstehung von Typ Ia Sup ernovae eine wichtige wenn
nicht sogar entscheidende Rolle spielt Dar uberhinaus wird SPY nach Abschluss aller Nach
b eobachtungen ein statistisch vollst andiges u limitiertes Beobachtungssample liefern anhand
dessen wichtige Entwicklungsparameter der Dopp elsternentwicklung kalibriert werden k onnen
In diesem Sinne tragen alle gefundenen und analysierten Systeme gleich ob SN Ia Vorl aufero der
nicht zum b esseren Verst andnisenger Dopp elsterne b ei
Das SPY Pro jekt wird in Kapitel detailliert erl autert Kapitel b esch aftigt sich mit der
Selektion Wei er Zwerge aus dem Hamburg ESO Survey die im Rahmen von SPY b eobachtet
werden sollten Kapitel stellt den f urden Survey verwendeten Sp ektrographen UVES und
die daf ur neu entwickelte Reduktionsroutine vor Die Metho den zur Analyse der gewonnenen
Sp ektren werden in Kapitel n ahererl autert w ahrend die Ergebnisse in den Kapiteln bis
diskutiert werden Schlu bemerkungen und ein weiterf uhrenderAusblick nden sich im letzten
Kapitel
Abstract
Sup ernovae of type Ia SNe Ia play a prominent role in many astrophysical environments and
their use as cosmological to ols is b eing pushed to redshift and b eyond The nature of their pro
genitors however remains a mystery which represents a ma jor limitation b oth for understand
ing the astrophysical phenomena in which SNe Ia are involved and for their use as cosmological
prob es In one of the two p ossible scenarios the SN Ia results from the merging of a binary
comp osed of two white dwarfs exceeding a critical mass most likely the Chandrasekhar mass
M However previous searches for binary white dwarfs double degenerates have failed
Chandra
Sup ernova Ia to pro duce a statistically signi cant sample to test this scenario Therefore the
Progenitor SurveY SPY was prop osed at the Very Large Telescop e VLT of the Europ ean
Southern Observatory ESO This pro ject was a large sp ectroscopic survey of white dwarfs
with the UV Visual Echelle Sp ectrograph UVES The purp ose was to observe more than
white dwarfs which would provide a de nite test of the double degenerate or merger
scenario The target list was comp osed of already known ob jects selected from various catalogs
as well as newly identi ed white dwarfs e g helium rich DB white dwarfs from the Ham
burg ESO survey A self developed semi automated reduction pip eline provided us with high
quality sp ectra to search for radial velocity variations an unambiguous hint for binarity
Follow up observations were started parallel to the survey in order to obtain radial velocity
curves for the most promising merger candidates found by SPY These follow ups were done with
Deutsch Spanisches Astronomisches Zentrum DSAZ the ESO New the m telescop e of the
Technology Telecscop e NTT the ESO VLT and the William Herschel Telescop e WHT
This thesis addresses the target selection from the HES chapter and the reduction
pip eline for UVES data chapter The main part however deals with the results of the
follow up campaigns done at the DSAZ and the NTT chapters to In the course of these
observations a total of binary systems have b een observed and analyzed in order to test
whether they qualify as SN Ia progenitors i e whether the total mass is larger than M
Chandra
and they merge in less than the Hubble time T due to gravitational wave radiation
H
Six of these systems are double line d already revealing evidence for a companion in their
sp ectra Thus the radial velocity analyses provide information not only ab out the motion of the
primary but also of the secondary This o ers the p ossibility to derive all orbital parameters
as well as the masses of b oth comp onents The highlight is the detection of the double line d
system WD This white dwarf binary has a total mass ab ove M and will merge
Chandra
in less than a Hubble time Therefore WD ful lls all criteria de ned by the merger
scenario to qualify as a SN Ia progenitor Along with another SPY ob ject see Napiwotzki et
al a and the sub dwarf B sdB plus white dwarf binary KPD Maxted et
al c WD is the third system supp osed to pro duce a SN Ia event In addition
the SPY ob jects HE this thesis and Karl et al and HE Napiwotzki
et al are only b elow the limit The relevance of the merger scenario is therefore
strongly enforced all the more if one considers that only ab out of the radial velocity
variable ob jects found by SPY have b een analyzed so far
In addition to the double lined ob jects single lined systems are analyzed as well For
we derive very accurate orbital solutions whereas for the remaining systems the solutions
are still ambiguous and more observations are needed to solve them de nitely Moreover three
systems show evidence for a main sequence companion and are therefore not suited for a SN Ia
precursor according to the double degenerate scenario
For the remaining ob jects we determined the masses of the visible comp onents via quanti
tive sp ectral analyses and evolutionary calculations e g the mass radius relation resp ectively
whereas the masses of the unseen comp onents are estimated from the mass functions f To
m
solve this equation de nitely however the determination of the the inclination i is crucial
The inclination angle can b e determined most easily in the case of eclipsing binaries However
our systems are not known to b e eclipsing Nevertheless assuming tidally lo cked rotation we
compared the orbital motions of six sdB stars with their sp ectroscopically derived pro jected
rotational velocities v sini in order to estimate upp er limits for i at least Unfortunately this
rot
metho d yields plausible results for HE only We conclude that either the remaining
ob jects do not rotate tidally lo cked or the determination of v sini was not accurate due to
rot
insu cient information ab out the instrumental pro le
Since the inclination i is still indetermined we estimate the most likely system comp ositions
o
from their mass functions and the statistically most probable inclination i whereas the
o
companions minimum masses are derived for i The latter inclination results in the
systems minimum masses and is therefore the most conservative criterium to select probable
o o
SN Ia candidates Although the analysis yields neither for i nor for i a probable
o
SN Ia candidate four systems could pass the criteria if their inclination angles are b elow
For the remaining ten binaries however the merging time t exceeds the Hubble time even for
m
the most favorable scenario Thus these systems certainly do not qualify as SNe Ia precursors
Besides the search for SN Ia candidates the sample of white dwarf binaries discussed in this
thesis will improve our understanding of close binary evolution Simulations of stellar p opula
tions forming double degenerates need several input parameters which describ e the formation
of stars stellar evolution and the outcome of close binary interactions in particular common
envelope ejection Some of the crucial parameters are only weakly constrained by present the
oretical calculations and thus it is essential to calibrate them empirically Once the analysis of
all double degenerates found by SPY is completed our binary sample will provide a ux limited
nearly complete sample as an ideal lab oratory for this task
As a spin o we determined metal abundance patterns of six radial velocity variable sdB
stars The derived abundances for ve of the ob jects are considered to b e normal whereas for
one the sub dwarf HE an overabundance of titanium was measured HE
is only the fourth sdB star with a p eculiar abundance pattern that has b een discovered The
existence of these sup er metal rich sdBs are without an explanation by theory so far
Chapter
The SPY pro ject
SPY is the Sup ernova Ia Progenitor SurveY an ESO large program carried out at the VLT to
test the double degenerate DD evolutionary channel for the formation of Sup ernovae SNe of
type Ia
Sup ernovae
Sup ernovae mark the violent termination of a star s life in a thermonuclear explosion According
to the shap e of their lightcurves and their sp ectral app earance they can b e divided in two main
classes and a large number of sub classes
Sup ernovae of type I I
show evidence for hydrogen within their sp ectra
have highly diverse lightcurves
Sup ernovae of type I
show no evidence for hydrogen within their sp ectra cf gure
have almost uniform lightcurves cf gure
However due to their sp ectral app earance type I SNe are sub divided into Ia Ib and Ic and
even these subtypes show p eculiarities in their abundance patterns and lightcurve characteristics
e g SN G SN bg or SN T see Li et al and Branch et al
Ia
SNe Ia are b elieved to b e the prime pro ducers of iron in the universe Furthermore the absolute
magnitude of SNe Ia is well de ned at M Figure and Branch et al
B
They are therefore used as an imp ortant to ol for cosmological distance determination The
role of SNe Ia as cosmological standard candles are forti ed even more by Phillips when
he found an empirical relation b etween the decline rate of their lightcurves and their absolute
magnitudes
Distance measurements from SNe Ia allowed the Hubble constant H to b e determined with
high accuracy SN Ia studies have also provided the rst evidence for an accelerating expansion
CHAPTER THE SPY PROJECT
Figure Type Ia SN mo del sp ectral energy distribution for solar and solar metalicities
Sup erp osed are transmission functions for standard passbands from left to right is U B V and
R The gure is taken from H o ich et al
of the universe i e Riess This astonishing result was recently con rmed by
WMAP investigations of the cosmic microwave background Wright Moreover Riess et
al and Leibundgut set b etter constraints to the density parameter and the
cosmological constant by the means of SN Ia observations Yet the kind of stars that pro duce
SN Ia events remain a mystery e g Livio
Progenitor
Host galaxies for SNe of type I I Ib or Ic are spirals and irregulars containing young stellar
p opulation whereas SN Ia can b e observed in all types of galaxies even in ellipticals with old
stellar p opulation only
The rapid evolution of SN Ia lightcurves see gure gives a hint on the nature of SNe Ia
precursors since they are dominated by the decay of the radioactive material synthesized in the
d
explosion This is mainly Ni half life sitting on the top of a decay chain leading to Co
d
half life and nally to stable Fe Due to the fact that massive progenitors would b e
able to hold back the rays pro duced by the radioactive decay the precursors of SNe Ia must
b e low mass compact ob jects The only candidates ful lling the criteria therefore are white
dwarfs To explo de however a white dwarf has to b e forced into a density and temp erature
regime where carb on and oxygen burn explosively and disrupt the star This happ ens at masses
near or ab ove the Chandrasekhar limit of M where the electron degeneracy no longer
stabilizes the star against the gravity At this p oint the white dwarf has either to collapse into
a neutron star or explo de as a SN
Since no physical pro cess is known leading to such conditions in a single white dwarf the
SN Ia precursor must b e part of a multiple system with matter b eing transferred from a com
panion to the white dwarf until the Chandrasekhar mass limit is reached Although there is
little doubt with regard to the white dwarf the de nite nature of the progenitor system remains
unclear Two main scenarios exist that di er by the type of the donor star
In the so called single degenerate SD scenario the white dwarf is accompanied by a regular
star as mass donor either a main sequence star a sup er giant or a helium star Whelan Ib en
The system p ossibly app ears as a symbiotic binary Munari Renzini a sup ersoft
Figure Lightcurve of a typical SN Ia Branch et al The horizontal axis shows the
time given in days b efore and after maximum brightness
X ray source van den Heuvel et all or a recurrent nova Hachisu Kato Although
we know that white dwarfs accreting from a non degenerate companion do exist we don t know
whether such systems exist in su cient numbers to account for the observed SN Ia frequency
nor whether the white dwarf grows enough to reach the critical mass for ignition
Contrary to the SD scenario the white dwarf s companion is another white dwarf in the
so called double degenerate DD scenario Ib en and Tutukov Close double degenerates
radiate gravitational waves that cause the orbit to shrink due to the loss of energy and angular
momentum until the system eventually merges To qualify as a SN Ia progenitor however the
initial separation of the system has to b e close enough that the two white dwarfs will merge
within a Hubble time and the combined mass has to exceeds the critical mass Because we know
that most stars end up as white dwarf remnants and that a ma jor fraction of stars are in binary
systems double degenerates must b e common among white dwarfs Nevertheless we do not
know if there exist enough double degenerates able to merge in less than a Hubble time and
pro duce a SN Ia event
Figure illustrates b oth evolutionary channels for SN Ia formation
The solution of the SD vs DD dilemma is of great imp ortance for assessing the role of
SNe Ia in a variety of astrophysical situations and p erhaps even more imp ortantly their
e ectiveness as accurate cosmological prob es
Surve Y
Several systematic radial velocity RV searches for double degenerates have b een undertaken
starting in the mid s Among investigated white dwarfs however only double
degenerates were indeed discovered and their p erio ds determined Marsh but none of
them qualify as a SN Ia precursor The absence of a SN Ia precursor in the sample of double
degenerates is not surprising considering the fact that theoretical predictions assume only a
few p ercent of all double degenerates to b e SN Ia progenitors Ib en et al Nelemans et
al In order to provide a de nite test of the DD scenario we therefore have to overcome
CHAPTER THE SPY PROJECT
the lack of a statistically signi cant sample of white dwarfs b eing checked for radial velocity
variations
The purp ose of SPY was to check more than white dwarfs for radial velocity vari
ability by the means of high resolution sp ectra Therefore we prop osed to use the ESO VLT
UT telescop e Kueyen equipp ed with the Ultraviolet Visual Echelle Sp ectrograph UVES
cf section to do this Our program takes advantage of those observing conditions which are
not usable by most other programs like bright mo on bad seeing and clouds and therefore was
conducted in service mo de by ESO sta astronomers SPY kept the VLT busy even when other
programs could not b e carried out During the four years that SPY was running we used as
much as of the time at the UT every year
Figure shows a typical example of a hydrogen rich DA white dwarf s sp ectrum taken
with UVES A characteristic feature of DA white dwarfs are the broad sp ectral lines caused
by the high densities in their atmospheres Obviously these broad lines are very ill suited for
radial velocity measurements However deviations from the lo cal thermal equilibrium LTE
pro duce sharp non LTE cores of the Balmer lines esp ecially of H which allow very accurate
radial velocity determinations This feature is not present in white dwarfs of sp ectral types DB
and DO without hydrogen but helium in their atmospheres The use of several helium lines
however enables us to reach similar accuracy as for DA white dwarfs
To obtain a su ciently high signal to noise ratio we restricted our input sample to ob jects
o
brighter than B mag Moreover only ob jects at quali ed for SPY due to
the lo cation of the VLT on the southern hemisphere Suitable DA candidates have b een drawn
by Ralf Napiwotzki from an up to date version of the McCo ok Sion Catalogue the
Hamburg ESO survey Wisotzki et al Reimers Wisotzki Christlieb et al
the Hamburg Quasar survey Hagen et al the Montreal Cambridge Tololo survey Lam
ontagne et al and the Edinburgh Cape survey Stobie et al Kilkenny et al
In addition Norb ert Christlieb and I selected helium rich DB white dwarfs from the Hamburg
ESO survey cf chapter A map of all known white dwarfs observed and unobserved by
SPY ful lling the SPY criteria is shown in gure
In order to detect radial velocity variations among the input sample we to ok at least two
exp osures of each ob ject at random ep o chs separated by at least one night Radial velocity shifts
of these survey sp ectra were evaluated by the means of a cross correlation based on a test
see Napiwotzki et al a The great advantage of the pro cedure is its exibili ty which makes
it an easy task to apply it to measure radial velocity shifts in stars of di erent sp ectral types
Balmer lines for DA white dwarfs He i lines for DBs He i i and metal lines for DOs and so
on The accuracy of this pro cedure is ab out km s Since merger precursors typically have
orbital velocities of km s or higher we are running only a small risk of missing a SN Ia
progenitor Figure displays the detection probability on this basis as a function of orbital
p erio d
Promising candidates are observed within our follow up campaigns in order to derive the
system parameters de nitely chapter
Figure Possible evolutionary channels for SN Ia formation via the double degenerate DD
and the single degenerate SD scenarios Napiwotzki et al The evolution starts with
two main sequence MS stars The more massive star b ecomes a red giant and its envelope is
ejected in a common envelope event In the double degenerate scenario the second MS star also
evolves to a red giant with a second common envelope event The result is the formation of a
close binary of two white dwarfs If this double degenerate system is close and massive enough
it will merge within a Hubble time due to gravitational wave radiation and will pro duce a SN
Ia event In the two variants of the SD scenario the secondary lls its Ro che lob e while i close
to the main sequence or as a red giant RG or ii as a He star after another common envelope
phase Mass is b eing transferred to the white dwarf and increases its mass until the critical mass
limit is reaches At this p oint the white dwarf will explo de as a SN Ia The picture is taken
from Napiwotzki et al
CHAPTER THE SPY PROJECT
Figure Sample UVES sp ectrum of a hydrogen rich DA white dwarf The upp er three panels
corresp ond to the blue channel and to b oth parts of the red channel covered by di erent CCDs
The lower panels show the H and H line cores For more details on the UVES instrument and the data reduction we refer to section and the references therein
o
Figure Distribution of all known white dwarfs south of and brighter than V
Squares and triangles indicates SPY observations while dots are unobserved ob jects The yellow
band indicates the p osition of the galactic disk The gure was taken from Napiwotzki et al
Figure Detection probability as a function of p erio d for SPY ob jects black line A Monte
Carlo simulation of the detection probability as a function of orbital p erio d was p erformed
for a sample of ob jects gray line The simulation takes into account ep o ch di erences
b etween exp osures and measurement errors on a star by star basis The results for the sample
d
are coadded Dips of the e ciency curve at e g P corresp ond to typical time intervals
b etween exp osures gure by R Napiwotzki priv com
Chapter
Selection of DB white dwarfs
Apart from the hydrogen rich DA white dwarfs our survey intends to observe white dwarfs of
all sp ectral types However up to July only DAs were selected from the Hamburg ESO
survey HES Christlieb et al Thus Norb ert Christlieb and I p erformed a DB selection
in August by means of mining the HES data base of ob jective prism sp ectra
DB white dwarfs display strong He I absorption lines over a wide e ective temp erature
range which are readily detectable in the HES ob jective prism sp ectra The strongest line
is He I A which is lo cated in a sp ectral region where confusion with absorption lines of nor
mal stars is avoided We therefore implemented algorithms for detecting that line in HES sp ectra
section The sum of the equivalent widths of H was successfully used by Christlieb et
al for the selection of DAs and it can therefore also b e used to reject them from the sample
of DB candidates section However the rst step in the selection of DBs is a separation
from non degenerate stars by means of color criteria section
Color criteria
U B and B V colors as well as the Str omgrenmedium band color index c can b e determined
directly from HES sp ectra as has b een describ ed in Christlieb et al The average
measurement accuracies of these colors in the HES are mag mag
U B B V
and mag for a large variety of ob ject types covering the total HES magnitude range
c
1
DBs can therefore b e roughly separated from non degenerate stars like eld horizontal branch
A type FHB A stars or main sequence turno stars by means of color criteria We explored
the b ehavior of these stars in color color diagrams using a sample of ob jects of known
type describ ed in Christlieb et al and known DBs from McCo ok Sion and
Friedrich et al present in the data set presently used for the exploitation of the stellar
content of the HES that is a set of overlap free HES sp ectra with S N from
HES plates Hereafter we refer to this data set as the stellar HES
We have chosen conservative color cuto criteria such that almost all DBs of our test sample
are included see gure that is
U B
B V
c
COLOR CRITERIA
Figure DB pre selection in color color space and c versus Balmer line sum feature space
Solid lines adopted selection b ox
CHAPTER SELECTION OF DB WHITE DWARFS
We veri ed with U B B V colors and c indices derived from a grid of synthetic DB sp ectra
Ko ester priv com that no DB in the e ective temp erature range T K
e
is rejected by our criteria The lowest values measured on that grid are U B
B V and c That is a DB in the ab ove mentioned temp erature range will not
b e rejected by our color criteria unless one of its U B B V colors or its c color index as
measured in the HES sp ectrum are in error by more than and resp ectively
Absorption line criteria
As can b e seen from gure it is p ossible to separate DBs from the main sequence and
Horizontal Branch HB stars and form AGNs but not from DAs and sdBs by broad and
medium band colors alone Therefore additional sp ectral prop erties have to b e used that is
absorption lines
Detection of He i A
We used three variants of the line detection algorithm describ ed by Christlieb et al
and a line index for detecting He i A in HES sp ectra However the helium lines of DB
white dwarfs are weaker than the Balmer lines of the DAs which makes feature detection and
consequently candidate selection more complicated for DBs
We t the He i A line by a Gaussian simultaneously with the continuum which is
approximated by a straight line in a A wide region centered on the line It was found
that the stability of the t critically dep ends on how close the start value for the p osition of the
line is to the real p osition The wavelength calibration zero p oint of the HES sp ectra derived
from an astrometric transformation b etween DSS I direct plates and HES plates is accurate to
A at A allowing to predict the p osition of the He I line with this accuracy We
used two di erent metho ds to further improve the start value a the x p osition of the global
ux minimum in the A wide t region is used and b the x p osition of the global ux
minimum in a small region around the predicted line p osition is used
It is imp ortant for the candidate selection to distinguish real detections from fake detections
due to noise We apply two criteria to do this First the detected line is required to have a
full width at half maximum FWHM broader than the minimum FWHM of the cross disp ersion
pro le h as measured during the extraction of the HES sp ectra Second the p osition of the
line as determined in the t must not deviate more than from the predicted line p osition
where is the uncertainty in x direction caused by the astrometric transformation b etween
DSS I direct plates and HES plates
In a third variant of the t metho d we t the He i i A line simultaneously with He i A
which improves the stability of the t for hotter DBs The same rejection criteria as ab ove are
applied
Finally we compute a line index by comparing the integrated ux in two continuum bands
neighboring He i A A and A with the integrated ux in a band
centered on the line A False detection are rejected by means of signi cance
threshold as a function of a HES internal magnitude cf gure with the help of a brake
nding algorithm Wisotzki et al used this algorithm for the selection of UV excess ob jects
ABSORPTION LINE CRITERIA
Figure Signi cance threshold for the He I A line index as a function of a HES internal
magnitude on one HES plate The threshold is determined with a brake nding algorithm
Balmer lines
Figure Adjustment of the cuto criterion for the Balmer line equivalent sum
The algorithm for measuring the equivalent widths of stellar absorption lines is describ ed
in Christlieb et al The authors used balmsum the sum of equivalent widths of H H
successfully for selecting DA white dwarfs in the HES Here we use this feature for rejecting non
DBs from our sample As can b e seen in gure there is a balmsum region in which DBs as well
as DAs is present and it is therefore a priori unclear which cuto value of balmsum one should
choose However a quantitative evaluation of the exp ected completeness and contamination of
the sample in dep endence of balmsum can guide the choice
We de ne the quality Q of a selection as completeness rate minus contamination rate This
yields Q for a selection with completeness all DBs of the test sample selected
and contamination One can regard the selection with the highest Q as the optimum and
CHAPTER SELECTION OF DB WHITE DWARFS
we have indeed adopted the corresp onding balmsum cuto value of A see gure This
choice yields
completeness
contamination
The balmsum cuto was applied applying the three color criteria and the He i A
line detection criteria Note that the ab ove estimates for completeness and contamination are
only valid if the relative fractions of DBs and non DBs in this subsample of the test sample are
representative
Summary of the DB selection
A star enters the raw candidate sample if its HES sp ectrum ful ll all of the three color criteria
if the sum of equivalent widths of H H is less than A and if the He I A
line is detected by at least one of the four detection metho ds describ ed in section Detected
means in case of the three variants of the t algorithm that the line t has not b een rejected
b ecause of to o small FWHM or deviating line p osition and in case of the line index it means that
the line index value must b e larger than the signi cance threshold applicable for the sp ectrum
These metho ds worked very well for our test sample presented in section for which
out of the DBs were re discovered Applying the pro cedure to the complete HES a set of
raw candidates were selected By means of visual insp ection of the prop er ob jective prism
sp ectra ob jects were nally not rejected as false candidates and by means of an even
stricter by eye classi cation criterium out of the candidates were classi ed as very
likely DB white dwarfs These ob jects are displayed in table A in the app endix
In the course of our SPY pro ject we observed of the selected ob jects totally However
it had turned out that our criteria are also sensitive to sub dwarfs of the sp ectral types O and
B Thus from the high resolution UVES sp ectra cf section it b ecame obvious that of
the observed ob jects were in fact sdO and sdB stars showing weak Balmer lines only Lisker
classi ed these helium dominated ob ject as He sdBs and He sdOs
Chapter
The UV Visual Echelle Sp ectrograph
For our SPY pro ject more than high resolution sp ectra were taken over the course of
four years using the Ultraviolet Visual Echelle Sp ectrograph UVES at the VLT Figure
shows a picture of the UVES instrument mounted at the Nasmyth B fo cus of the UT Kueyen
telescop e of the VLT at Paranal Observatory Chile
Figure The Ultraviolet Visual Echelle Sp ectrograph UVES at the Nasmyth B fo
cus of the UT telescop e Kueyen of the VLT picture taken from the UVES web page
http www eso org instruments uves
CHAPTER THE UV VISUAL ECHELLE SPECTROGRAPH
Figure Sample raw CCD image of a typical UVES observation Displayed here is the lower
part of the red arm Within the individual echelle orders the bright sky background and its
typical absorption lines can b e seen b eneath the more intense star sp ectrum In the lower left
corner there are some bad rows on the CCD
Instrumental setup
The setup used for our SPY pro ject op erated the instrument in dichroic mo de using Dichroic
No at central wavelengths of A in the blue and A in the red Nearly complete
wavelength coverage from A to A with only two small A wide gaps at A and
A is achieved A x binning was used to reduce read out noise and a wide slit to
minimize slit losses The wide slit reduces the resolution to R A at H or
b etter if the seeing discs were smaller than the slit width Exp osure times were to minutes
dep ending on the brightness of the targets The S N p er binned pixel A of the extracted
sp ectrum is usually or b etter Figure shows an example for a raw CCD frame For more
details on the UVES instrument see D Odorico and the UVES manual D Odorico et
al
UVES data reduction
For most of the nights a complete set of calibration frames were taken for each arm in addition
to the science frames These calibrations include multiple bias and at eld exp osures as well
as an order de nition at eld a at eld exp osure taken with a very small slit and a ThAr
calibration frame
ESO provides a sp ecial UVES pip eline based on MIDAS pro cedures to carry out the data
reduction Although the outcome of this pip eline was of go o d quality in most cases some
sp ectra show artifacts of varying strength e g a quasip erio dic pattern in app earance similar to
UVES DATA REDUCTION
2D science frame
Bias correction Bad pixel correction Cosmic ray correction Interorder BG correction
2D corrected 2D FF science frame frame
Order extraction
star sky 1.5D FF frame
Shifting
1.5D sky−corrected 1.5D shifted science frame FF frame
Order merging WL calibration
1D merged & FF corrected spectrum
Polynomial fit
Test not minimized chi^2 test
Test minimized
Normalization by DC WD
Fully reduced
spectra
Figure Schematic diagram of the reduction pro cess according to our self developed semi
automated pip eline An explanation of the individual reduction steps in more detail is given in
the text
a fringing pattern see gure for example Nevertheless the pip eline reduction was extremely
useful for a fast selection of radial velocity variable ob jects for follow up observations see b elow
For a more accurate data reduction however we developed our own set of pro cedures This enabled us for example to determine stellar parameters by the means of sp ectroscopic analysis
CHAPTER THE UV VISUAL ECHELLE SPECTROGRAPH
Ko ester et al with high accuracy
Our self developed semi automated program is based on MIDAS routines already used in
the ESO pip eline However our pro cedure di ers from the pip eline sequence and it enables me
to intervene the reduction at crucial p oints to handle problems Figure gives a schematic
overview ab out the reduction pro cess in our self developed pip eline In more detail the individ
ual steps displayed in this diagram are
Bias correction
To correct the o set caused by dark current we computed a two dimensional master bias from
the bias frames taken in each night Subsequently this master bias was subtracted from each
science frame
Cosmic ray correction
To purge the two dimensional science frames from cosmics rays we compared each frame with
a smo othed frame of itself Therefore each row was ltered by the means of a three pixel wide
median to create the smo othed template After that we compared this template to the raw
science frames Pixels that di er by more than de ned by read out noise and gain of the
CCD from the smo othed frame are substituted by the median value Since the echelle orders
on the CCD are nearly horizontal the lter used for the template works in the same direction
as the disp ersion The sp ectrum itself is therefore una ected by the pro cedure
Bad pixel correction
As can b e seen from gure are the CCDs used in the UVES instrument not p erfect but show
several defect columns and rows To correct these bad pixel we therefore substituted their
values by values of neighboring pixels
Order de nition
To get an idea ab out the p osition of the echelle orders on the CCD a Hough transform was
p erformed for a sp ecial order de nition at eld taken in the same night as or close to the
science frames This order de nition was essential for the determination of the interorder back
ground from the science frames later on and for obtaining the disp ersion relation from the ThAr
calibration frame
Interorder background correction
The interorder background was computed from the echelle raw sp ectrum in the sampling space
pixel to pixel Background values were estimated at reference p ositions b etween the orders and
interpolated by a bivariate p olynomial Afterwards the correction was carried out by subtraction
of interorder background from the science frame
Wavelength calibration
The echelle orders of a ThAr calibration frame were extracted at the CCD p ositions sp eci ed
by the order de nition routine An automatic line identi cation was done and the disp ersion
UVES DATA REDUCTION
1000
800
600
400
200
420 440 460 480 500 520 540
Position [pixel]
Figure Vertical cut through a two dimensional UVES sp ectrum The order de nition
at eld is overlayed to the normal science frame Using plots like this we estimated the o set
relative to the order de nition frame and the window size for the extraction of the star and the
sky background
relation was determined for each order Line tables for two sky windows and for the ob ject
window of the detected orders result Each window has a width of pixels or
Sp ectrum extraction
Since we were interested in an optimized data reduction we extracted the sp ectral orders not
at the full slit window but at a smaller one to increase the S N ratio Window size and o set
relative to the order de nition frame caused by refraction were determined for each observa
tion individual ly by the means of a vertical cut through the two dimensional CCD frame cf
gure Since the o set is less than pixels the wavelength calibration done previously is
still su ciently accurate and can b e assigned to the extracted echelle orders later on
dimensional The extracted orders are stacked into a single data le which we call the
data set
Sky background correction
The sky background was extracted at a CCD p osition ab ove and b elow the star s sp ectrum
As for the extraction of the sp ectrum o set and window size had to b e determined individuall y
for each observation from vertical cuts similar to the one displayed in gure The extracted
sky backgrounds were merged and subsequently weighted relative to the prop er sp ectrum by
the means of the window sizes Finally the weighted sky background is subtracted from the
extracted star s sp ectrum
Flat eld correction and the ripple problem
The general purp ose of the at eld correction is to correct for pixel to pixel variations However
it was also used in order to correct the wavelength dep endent sensitivity of the instrument as
CHAPTER THE UV VISUAL ECHELLE SPECTROGRAPH
8000
6000
4000
2000
0
4960 4980 5000 5020 5040
Figure Three overlapping echelle orders extracted from the lower part of the red arm Since
they are not at eld corrected yet they still show a blaze e ciency distribution In order to
avoid edge e ects and to increase the S N ratio the dashed parts are cut of b efore we merged
the orders
well Dominated by the grating e ciency the instrument s sensitivity function is mainly a blaze
distribution which can b e describ ed by a sinc function The shap e of the blaze distribution can
b e seen from gure where three overlapping echelle orders are displayed b efore the at eld
correction was done
To do the blaze correction we merged all at eld frames to create a single master at eld
which was afterwards corrected for the master bias Out of this dimensional master at eld
we extracted echelle orders at the same p osition and with the same o set as we did for the
science frame Subsequently the echelle orders of the star s sp ectrum were divided by its prop er
at eld order
Even though this ought to correct for the instrument s sensitivity sometimes ripple features
remained in the wavelength calibrated and merged sp ectra This problem is very similar to the
ripples pro duced by the ESO pip elin e
A close insp ection of these features shows that each ripple corresp onds to a individual echelle
order The describ ed at eld correction will not rectify the blaze function very accurately The
reason for the remaining sp ectral trend even in the corrected orders seems to b e a di erence
in the gradient of the at eld and the sp ectrum cf upp er panel of gure To get rid of
the ripple problem we therefore tried to match b oth gradients by shifting the extracted at eld
relative to the extracted star s sp ectrum cf lower panel of gure
Even a small shift of ab out a few pixel will heavily reduce the ripples By iteration we
determined the optimum shift as can b e seen from gure The at eld was shifted within a
range of to pixels in steps of pixel The resulting sp ectra were subsequently checked
by tting a second order p olynomial function to a wavelength region where no sp ectral lines are
exp ected in DA white dwarfs Since this t was done by means of a test the gives an idea
ab out the t s quality Scanning the values for the full shift range of to pixels we
determined the b est shift
UVES DATA REDUCTION
Figure The cause of the ripple problem The o set b etween the extracted at eld and the
star is displayed schematically on the left hand side while the right hand side shows three fully
reduced and merged echelle orders as an example for the result of the at eld correction done
for the prop er gradients on the left Gray shaded wavelength regions mark the overlap b etween
two neighboring orders The upp er panels displays a large shift of several pixels that results in
a ripple problem like the one apparent in some of the ESO pip eline sp ectra The lower panel
displays a at eld that is shifted but now matched the star s sp ectrum In this case the ripple
problem is small
Calibrating the ob ject sp ectrum
After the b est at eld shift was found and the corrected for a wavelength calibration of the
dimensional science frame was p erformed This was carried out by the means of the disp ersion
relation found previously
Order merging
The merging pro cess was done by an algorithm that computes a weighted average in the over
lapping region of adjacent orders The normalized weight was a linear ramp b etween and in
the overlapping region
To avoid low S N and artifacts caused by the echelle blaze function at the CCDs edges
however the b oundary regions of the echelle orders were cut o b efore merging the orders cf
gure Since the overlap of the orders is large enough the loss of information was a small price payed for an increased quality of the merged data set
CHAPTER THE UV VISUAL ECHELLE SPECTROGRAPH
Normalization
Finally we normalized the resulting star s sp ectrum by means of a sp ectrum of a white dwarf of
sp ectral type DC Since this kind of star shows no lines at all in the optical wavelength range
the DC s sp ectrum was smo othed until no feature noise p eaks cosmics artifacts and so on
but the continuum remains The resulting smo othed sp ectrum was used as the ux distribution
by which the science sp ectrum was divided to b ecome normalized
ESO pip eline vs semi automated pip eline
Although the ESO pip eline provides a reduction quality su cient for a fast detection of radial
velocity variability among the survey ob jects the ripples that b ecome obvious for some sp ectra
are a handicap for a quantitative sp ectral analysis as p erformed by Ko ester et al
Our self developed pip eline on the other hand handle the ripple problem very well by
shifting the at eld relative to the science frame To do this the b est shift is determined by
iterative means A disadvantage compared to the ESO pip eline however is the semi automated
nature of our pip eline Therefore in preparation for the extraction pro cess the window size and
the o set relative to the order de nition at eld has to b e determined by eye cf gure
This has to b e done for the star s sp ectrum as well as for the sky background However this
disadvantage is counterbalanced by the great advancement in reduction quality as can b e seen from gure
2
1.5
1
0.5 ESO UVES pipeline
0 4800 5000 5200 5400 5600
2
1.5
1
0.5 self-developed pipeline
0
4800 5000 5200 5400 5600
Figure A comparison of the two reduction pip elines for UVES data The upp er panel of
the gure displays the result of the UVES pip eline provided by ESO Ripples pro duced by this
pip eline are obvious The result of our self developed semi automated pip eline is displayed in
the lower panel The improvement concerning the ripple problem is great although the at eld
was shifted by only two pixels relative to the ob ject
Chapter
Derivation of system parameters
Only double degenerate mergers with a total mass ab ove the Candrasekhar mass limit and within
a merging time less than the Hubble time qualify as SN Ia progenitors While the system s total
mass results straight from the masses of the individual comp onents the merging time deduced
form gravitational wave radiation is a function of the individual masses and the system s p erio d
Landau Lifshitz
M M
t P
M
M M
In this equation the merging time is given in years whereas the masses are in solar units and
the orbital p erio d in hours
To answer the question whether or not a double degenerate qualifys as a SN Ia precursor
we have determine the system parameters and in particular the orbital p erio d and masses of its
comp onents Therefore we used a variety of astronomical techniques that are presented within
the next sections in more detail
De nition of samples
From the full set of candidate systems discovered by the SPY pro ject we de ned two subsamples
Sample one was constructed to search for SN Ia progenitors and sample two for obtaining a
statistical complete ux limited sample in order to test binary evolution theory
The most probable SN Ia precursors are systems with short p erio ds and large radial velocity
RV amplitudes and unambiguous binaries Therefore we checked our survey for the most
promising merger candidates double line d systems b oth white dwarfs visible and double
degenerates with radial velocity variations of km s and more up to km s This
sample contains ob jects
An additional sample was a magnitude limited one comp osed of double degenerates with
V In order to test binary evolution theory it is imp ortant to avoid observational
biases i e by selecting against long p erio d and low amplitude systems as is done in
in the rst sample Thus the second sample will improve our understanding of double
degenerate p opulation dramatically Sample two contains stars totally
CHAPTER DERIVATION OF SYSTEM PARAMETERS
Ob jects of b oth samples were subsequently re observed in the course of follow up campaigns in
order to derive their p erio ds and RV curves de nitely For a single lined system this nally
yields the mass functions f If the system is double lined the mass ratio of the comp onents
m
can b e determined
Radial velocity curves
Follow up observations
The follow up campaigns were p erformed at the European Southern Observatory Chile ESO
either at Paranal using UVES at the VLT UT or at La Silla using the New Technology Telescope
NTT equipp ed with the ESO Multi Mode Instrument EMMI Additional observations were
done at the m telescop e of the Deutsch Spanisches Astronomisches Zentrum DSAZ at
Calar Alto Spain equipp ed with the medium resolution TWIN Sp ectrograph TWIN at the
Cassegrain fo cus Our data set was complemented by sp ectra taken with the IDS or ISIS
instrumentation mounted at the Wil liam Herschel Telescope WHT op erated by the Isaac
Newton Group of Telescopes ING Spain Table gives an overview of all follow up runs
For more details on individual observational sites the prop er instrumentation and the reduction
techniques we refer to section B in the app endix
Radial velocity measurements
Radial velocities of the individual observations are determined by calculating the shifts of the
measured wavelengths to lab oratory measurements Therefore we p erformed a simultaneous t
of a set of mathematical functions to the observed line pro les using the ESO MIDAS package
Unless otherwise noted we concentrated on the H line for the t pro cess b ecause of its sharp
and well de ned non LTE line core A linear function was used to repro duce the overall sp ectral
trend a Lorentzian to mo del the line wings and a Gaussian for the innermost line core The
central wavelengths of the Lorentzian were xed to that of the Gaussian for physical reasons
Although we were mainly interested in double degenerates some ob jects observed within
the follow up campaigns were sub dwarfs of the sdO and sdB subtype or white dwarfs with a
M dwarf companion These ob jects show not only hydrogen absorption lines but have a more
diverse sp ectral app earance than DA white dwarfs Thus we usually used the He lines namely
at A and at A for radial velocity determination as well as H whenever analyzing
He rich ob jects Due to the fact that He lines are much shallower and more narrow than the
lines from the Balmer series only a linear function and a single Gaussian were required for the
tting pro cess
In the case of a white dwarf system with an M dwarf companion the Balmer series of the
white dwarf is overlayed by strong emission features pro duced by the dM The innermost core
of the white dwarf s absorption line is hardly visible making a t di cult or even imp ossible
Thus we usually can not determine the white dwarfs motion However b oth comp onents are
clearly visible in some outstanding cases like WD cf section The same set
of functions used for single lined white dwarfs can easily repro duce the dM s emission feature
as well as far as the Gaussian is tted in emission and de coupled from the Lorentzian This
pro cedure nally yields very accurate radial velocity measurements for the M dwarf companion
Some examples for ts to the observed line pro les of di erent kind of systems can b e seen from gure
RADIAL VELOCITY CURVES
Table Follow up observations done of candidates identi ed by the SPY pro ject The rst
column lists the date of observation while in the second the used telescop e and instrumentation
is shown The last column displays the participating observers
Date Telescope Instrument Observer
March DSAZ m TWIN R Napiwotzki E M Pauli
July DSAZ m TWIN R Napiwotzki C Karl
Octob er ESO VLT UVES R Napiwotzki
Octob er ING INT IDS G Nelemans T Reerink
February DSAZ m TWIN C Karl T Lisker
June ESO VLT UVES R Napiwotzki
August DSAZ m TWIN C Karl S
September ESO NTT EMMI C Karl
Octob er DSAZ m TWIN C Karl S
November ESO VLT UVES R Napiwotzki
January ING WHT ISIS G Nelemans P Gro ot
February ESO NTT EMMI C Karl
March ESO VLT UVES R Napiwotzki
March ING WHT ISIS G Nelemans R Izzard
April DSAZ m TWIN C Karl S
May DSAZ m TWIN C Karl S
July DSAZ m TWIN C Karl S
August ING WHT ISIS G Nelemans M Montgomery
January ESO NTT EMMI C Karl
S Service mode carried out by a sta astronomer at Calar Alto and the author via internet
Gijs Nelemans from the University of Cambridge provided additional radial velocity mea
surements derived from IDS and ISIS sp ectra These measurements were done by means of the
MOLLY routine Marsh et al
Power sp ectra and RV curves
In preparation for further analysis all measured radial velocities as well as the times of obser
vation have b een barycentrically corrected As reference the time of the middle of exp osure is
used
The p erio d search was carried out by means of a p erio dogram based on the Singular Value
Decomposition SVD metho d A sine shap ed RV curve was tted to the observations for a
multitude of phases which were calculated as a function of p erio d see Napiwotzki et al b
The di erence b etween the observed radial velocities and the b est tting theoretical RV curve
for each phase set was evaluated in terms of the logarithm of the sum of the squared residuals
as a function of p erio d This metho d nally results in the data set s p ower sp ectrum which
allows to determine the most probable p erio d of variability see Lorenz et al
From the b est t RV curve corresp onding to the most probable p erio d the ephemeris the
system s velocity and the semi amplitude were derived Figure shows two examples for RV
CHAPTER DERIVATION OF SYSTEM PARAMETERS
Figure RV determination by means of MIDAS ts a Results for a single lined DA white
dwarf WD b Double lined white dwarf HE c The M dwarf plus white
dwarf system HE The white dwarf can not b e tted for this ob ject but it can for
WD d Panels e and f display the ts of the H and Hei A for the sdB star
HE Results for the sdO HE can b e seen from g and h Note that the central emission in panel g is caused by a non LTE e ect in the photosphere of the sdO star
GRAVITATIONAL REDSHIFT
Figure Best t RV curves In the upp er panel the RV curve of the single lined system
HE is displayed whereas the double lined system HE can b e seen in the
lower panel
curves The upp er panel displays the b est t RV curve of a single lined system from which the
mass function f can b e derived
m
M sin i PK
invis
f
m
M M G
invis vis
The lower panel however displays the RV curve of a double lined system The mass ratio of
b oth comp onents can b e computed from the ratio of the radial velocity semi amplitudes
K M
sec prim
M K
sec prim
Gravitational redshift
In the gravitational eld of a star with mass M and radius R a photon su ers an energy loss
which results in a redshift
p
v GM GM g
grav
z
c Rc c
Thus the measured radial velocities of white dwarfs are always a combination of the intrinsic
star s velocity and the gravitational redshift z due to the high surface gravities of white dwarfs To correct for this e ect we have to obtain the star s mass and radius from other metho ds
CHAPTER DERIVATION OF SYSTEM PARAMETERS
0.02
0.015
0.01
0.005
0 0.5 1 1.5
Figure Mass radius relation of white dwarfs according to Eggelton published in Truran
Livio
For double line systems however the gravitational redshift o ers the p ossibility to derive
the masses for b oth comp onents For some of these systems it b ecome obvious that the system
velocities of the individual comp onents do not match In other words due to
prim sec
di erent gravitational redshifts
M G M
sec prim
z
c c R R
prim sec
This equation enables us to estimate the redshift as a function of white dwarf mass Due to
the fact that the mass ratio is also given by the RV curve equation only one combination
of masses can ful ll b oth constraints for a given mass radius relation However if the surface
gravity is low e g for sdB and sdO stars or if the masses of the comp onents are equal z will
not b e measurable
The mass radius relation of white dwarfs
It has b een shown by Chandrasekhar that for white dwarfs a mass radius relation
holds Mo dern evolutionary calculations for white dwarfs result in accurate mass radius re
lations Bl ocker et al Woo d for di erent hydrogen envelope masses However to
utilize these tables the e ective temp eratures and the surface gravities of b oth comp onents have
to b e known Whereas it is straight forward to determine T and log g for single lined systems
e
via sp ectral analysis see e g Ko ester et al double lined systems are more di cult cf
section Therefore we had to use a general mass radius relation derived by Eggelton rst
published in Truran Livio for double lined systems The advantage of this relation is
that it do es not dep end on stellar parameters This relation is displayed in gure
In order to compare Eggelton s mass radius relations with relations derived by Bl ocker et
al and Woo d we compared the masses determined by means of di erent distri
butions with each other We found that the masses derived from the di erent relations agree to
QUANTITATIVE SPECTROSCOPIC ANALYSIS
Figure Mass determination of single and binary DA white dwarfs Temperature and
gravity determined from mo del atmosphere analysis of the UVES sp ectra Ko ester et al
and priv com are compared to co oling sequences of white dwarfs Bl ocker et al for a
range of masses The gure was taken from Napiwotzki et al a
each other within the error margins and thus we conclude that the discrepancies b etween them
are insigni cant which justi es our use of Eggelton s relation
Quantitative sp ectroscopic analysis
A direct approach to estimate the mass over radius ratio M R of the ob jects is by determining
their stellar parameters T log g and n by means of a quantitative sp ectral analysis Sub
e He
sequently the stellar masses can b e estimated by comparison of the sp ectroscopically derived
temp erature and gravity to evolutionary calculations like the tracks computed by Bl ocker et
al for white dwarfs gure
For sub dwarf B stars appropriate computations were done by Dorman et al cf
gure Their evolutionary tracks were evaluated for single star evolution However sdB
stars can also form in the course of evolution of close binary systems by mass transfer Han et
al calculated evolutionary sequences for sub dwarfs in binary systems for twelve parameter
sets The masses range from M to M but all distributions p eak near M
which corresp onds to the canonical mass of sdB stars Therefore we adopt M M
sd
throughout this study Nevertheless accurate stellar parameters were determined in order to
determine the pro jected rotational velocities and subsequently the systems inclinations angles
cf section
CHAPTER DERIVATION OF SYSTEM PARAMETERS
Stellar parameters of single lined systems
By means of sp ectral analyses Detlev Ko ester from the Christian Albrechts Universit atin Kiel
evaluated T and log g for a large sample of single lined DA white dwarfs from SPY Ko ester et
e
al and priv com Recently Ralf Napiwotzki priv com completed this sample for DA
white dwarfs brighter than mag The sub dwarf B stars were analyzed by Thorsten Lisker
Lisker et al and the sdOs by Alexander Str oer
Stellar parameters of double lined systems
The sp ectral analysis of a double lined system is more complex than for a single line d system
b ecause the sp ectrum of the former is always a sup erp osition of two individual sp ectra This
problem is sometimes circumvented by assuming that the results of a simple single line d mo del
t is representative for the average parameters One p ossibility would b e to estimate the mean
gravity and individual temp eratures of the comp onents of the analysis of individual sp ectra
taken close to conjunction and quadrature phases e g HE in Napiwotzki et al
However this approach can only b e applied if b oth comp onents have quite similar parameters
A direct approach would b e to disentangle the observed sp ectrum by means of deconvolution
techniques into the sp ectra of the comp onents Subsequently these sp ectra could b e analyzed
by tting synthetic sp ectra developed for single lined white dwarfs to the individual line pro
les Such a disentangling pro cedure has b een developed by Simon and Sturm and was
successfully applied to main sequence double lined binaries However it has not b een tested
4.5
5.0
5.5
6.0
6.5
45 40 35 30 25 20 15
Figure Mass determination of single sdB stars Temperature and gravity determined from
mo del atmosphere analysis of the UVES sp ectra compared to evolutionary tracks Dorman et
al for a range of masses Our own determinations are marked as black dots whereas
parameters derived by Likser et al are indicated as gray crosses
QUANTITATIVE SPECTROSCOPIC ANALYSIS
for white dwarfs for which the wavelength shifts caused by orbital motions are much smaller
than the line widths of the broad Balmer lines Therefore we chose a di erent approach for
our analyses Ralf Napiwotzki developed the program fitsb Napiwotzki et al b which
p erforms a sp ectral analysis of b oth comp onents of a double lined system fitsb is based on
a minimization technique using a simplex algorithm The t is p erformed on all available
sp ectra covering di erent sp ectral phases simultaneously Thus all available sp ectral informa
tion is combined into the parameter determination pro cedure An application of the program
fitsb to the known system HE Napiwotzki et al yielded encouraging results
Napiwotzki et al b In the following study fitsb was also applied to HE
section A detailed description of the program and an evaluation of its p erformance will
b e given by Napiwotzki in prep
Pro jected rotational velocities
The rotation of a star causes a broadening of the sp ectral lines due to the Doppler e ect
However the rotational axis is usually not p erp endicul ar to the line of sight and thus we can
only determine the pro jected rotational velocity v sin i by means of a sp ectral analysis In
rot
combination with the p erio d P and the stellar radius R however we can therefore estimate the
unknown inclination i if we assume b ound rotation caused by tidal lo cking
v P
rot
sini
R
Whereas the p erio d is usually well de ned by the p ower sp ectrum the radius has to b e estimated
from a mass radius relation by means of the stellar atmospheric parameters T and log g This
e
M R relation is based on evolutionary calculations like the Bl ocker tracks for white dwarfs
Bl ocker at el For sdB stars the canonical mass is adopted and the radius follows from
s
M G M G
g R
R g
White dwarfs can b e analyzed for v sin i from the NLTE core from the H line as has
rot
b een shown by various studies e g Ko ester et al Heb er et al Karl et al DA
white dwarfs display no other sp ectral lines b eside the Balmer lines Sub dwarf B stars however
show helium and metal lines in addition to the Balmer series
Metal abundances of sdB stars
The b est suited lines are narrow metal lines e g the Mg i i doublet at A and A or
the N i i lines b etween A and A Thus the metal rich sdB stars are the b est candidates
for a study on rotational velocities
In order to derive v sin i we therefore determined metal abundance patterns for six sdB
rot
stars and calculated synthetical line pro les for these ob jects Finally the pro jected rotational
velocity was estimated by comparison of the synthetic and the observed line pro les The
inclination of the sdB systems or lower limits were computed thereof see section
Chapter
Double lined white dwarfs
The sp ectra of the so called double lined systems reveal sp ectral features arising from a compan
ion This may either b e a displaced Balmer absorption line in the case of a DA DA system or an
emission line of a M dwarf companion in the case of a DA dM system In any case tting these
features provides additional information ab out the system Figure shows typical examples
for double lined systems In the upp er panel the DA DA white dwarf system HE
is displayed whereas the DA dM binary WD can b e seen in the lower panel In the
following chapter we concentrate on white dwarf systems DA DA while white dwarfs with M
dwarf companions will b e discussed in the next chapter
1.2
1
0.8
0.6
0.4
6540 6550 6560 6570 6580 6590
3
2.5
2
1.5
1
0.5
6540 6550 6560 6570 6580 6590
Figure Double lined systems The upp er panel displays the DA DA white dwarf binary
HE and the lower panel WD a DA dM system Note that the last system
displays two emission lines one from the M dwarf and the second one most probably from a
hot DA white dwarf
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
Obviously the p erio d has to b e the same for b oth comp onents of a binary system Therefore
once the p erio d is determined by any means we can compute RV curves getting the ephemeris
the system velocity and semi amplitude for the primary as well as for the secondary
In order to form a binary system the origin of the ephemeris for the two comp onents have
to di er in phase by In some cases the system velocities derived from the RV curves
are in uenced by mass dep endent gravitational redshift For a given mass radius relation the
redshift can b e computed as a function of white dwarf mass cf section and since the mass
ratio is also given from the RV curve only one combination of masses can ful ll b oth constraints
With information ab out the absolute masses of the comp onents we can compute their sep
aration from the third Kepler law The total separation b ecomes
r
GP M M
3
prim sec
A
tot
According to the individual masses the distances of the comp onents from the barycenter can
b e computed to b e
M
sec
A A
prim tot
M M
sec prim
and similarly for the secondary The orbital velocities result from the system s p erio d and the
distances from the barycenter according to
A
prim
v
prim
P
Finally the inclination angle can b e computed from a comparison of the measured semi amplitudes
derived from the RV curve and the orbital velocities calculated from equation to b e
K
prim sec
sin i
v
prim sec
Usually no sp ectral analysis was p erformed for double lined systems due to the di culties
mentioned in section The comp onents of HE however were found to b e quite
similar with no measurable gravitational redshift di erence This ob ject could b e
analyzed by means of the fitsb program package cf sections and
Systems analyzed from their radial velocity curves alone
The following systems are analyzed without having use of a quantitative sp ectroscopical analysis Figure displays the approach for analyzing these systems in a schematic way
CHAPTER DOUBLE LINED WHITE DWARFS
RV curve
Κ P Gravitational Redshift 1,2
∆γ = ο
M /R − M /R M /M M−R relation 1 1 2 2 1 2
Α M tot 1,2
Α 1,2
v 1,2
i
t Μ
M tot
Figure Analysis of a double lined system schematic
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
WD
V
History
The McCo ok Sion catalogue classi ed WD as a DA white dwarf Moreover
it b ecomes obvious from the survey sp ectra that this is a double lined system and therefore
WD b ecame a high priority ob ject for our follow up campaign Observations were
done during our September run at the NTT and again at the VLT in November of the
same year
Radial velocity curve
Both comp onents of WD lo ok su ciently di erent in terms of sp ectral app earance
as can b e seen from gure We assign the stronger H line pro le to the comp onent called
Figure Observed H line pro le and tted curve of WD
the primary henceforth while the weaker one b elongs to the secondary However the weaker
H line was hardly visible in some of the sp ectra For this reason we excluded ambiguous
ts to the observed line pro les from our analysis and concentrated on RV measurements from
d
the primary for the determination of the system s p erio d The resulting p erio d of
h m s
is unequivocal as can b e seen from the p ower sp ectrum displayed in the lower panel
of gure The upp er panel of gure shows the b est t RV curves and yield to the system s
velocities and the semi amplitudes For the primary we derive km s and a
prim
RV semi amplitude of K km s The ephemeris for the time T de ned as
prim
the conjunction time at which the star moves from the blue shifted part to the red shifted part
of the RV curve is
HJD T E
prim
The corresp onding results for the secondary are km s K
sec sec
km s and
HJD T E
sec
CHAPTER DOUBLE LINED WHITE DWARFS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
RVs of b oth comp onents of WD as a function of orbital phase and tted sine curves
RVs measurements b elonging to the primary are marked as lled symbols while op en symbols
indicate data b elonging to the secondary Squares are used for ESO VLT observations and
triangles for observations from the ESO NTT Lower panel Power sp ectrum of WD
Since tting the secondary was not unambiguous in all cases the displayed p ower sp ectrum was
calculated using only data b elonging to the primary However for physical reasons the p erio d
has to b e the same for the secondary The inset shows the region near the main p eak in more
detail
d
Both ephemerides di er by in terms of HJD T This corresp onds to
which is close to as exp ected for physical reasons
System comp osition
Since WD is a double lined system we can derive the mass ratio from the semi
amplitudes of the RV curve to b e M M The system velocities de
prim sec
rived for the comp onents di ers by km s due to the mass dep endent
gravitational redshift Combining it with the mass radius relation we compute the masses
to b e M M and M M resp ectively Both comp onents are separated by
prim sec
A R only thus WD obviously underwent phases of strong binary interac
tot
tion in its history The distance from the barycenter and the resulting orbital velocity for the
primary is A R and v km s Corresp onding results for the secondary are
prim prim
o
A R and v km s The orbital velocities yield an inclination of i as
sec sec
can b e seen from comparison with the radial velocity semi amplitudes The merging time of the
system can b e computed from the orbital p erio d and the comp onents masses to b e t Gyrs m
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
only
This system is a very exciting one as its total mass M is likely to exceed the Chan
drasekhar limit As it also will merge in less than a Hubble time WD is a very go o d
candidate for a SN Ia progenitor Note however that the RV curve of the secondary lacks some
crucial data p oints esp ecially at see gure to fortify the semi amplitude and the
system velocity Therefore the masses derived for b oth comp onents are somewhat uncertain
Although WD quali es as an SN Ia precursor within the error margins of M M
prim sec
and cf table additional observation will b e necessary to improve the RV curve and
con rm the very imp ortant result
Table System comp osition for WD
Determined from RV curve
km s
M M
prim sec
P d
Deduced parameters
M M
prim
M M
sec
A R
tot
i deg
M M
tot
t Gyrs
m
SN Ia precursor yes yes yes yes yes
CHAPTER DOUBLE LINED WHITE DWARFS
HE
V
History
HE was rst observed in the Hamburg ESO survey and classi ed to b e a hydrogen
rich DA white dwarf by Christlieb et al Furthermore from the high resolution survey
sp ectra taken in August two H line cores b ecome obvious As a result of this outstanding
sp ectral app earance we deduced HE to b e a double lined system comp osed most
probably of two white dwarfs Follow up observations were carried out in January and
November at the ESO VLT The campaign was complemented by observations done at the
ESO NTT during the September run
Radial velocity curve
Since b oth H line pro les lo ok very similar there is no way for HE to discern the
comp onents by visual insp ection only However to b e able to p erform a radial velocity analysis
we have to assign the observed lines to the primary or to the secondary
Therefore we computed the separation of the comp onents rst and tted sine curves for a
range of p erio ds to the measured RV di erences and determined the value for each p erio d
Doing this we pro duced a di erence power spectrum indicating the quality of our RV t as
a function of p erio d This way we got a rst estimate of the p erio d of the system and were
subsequently able to identify the individual comp onents unambiguously for all of our sp ectra
This enabled us to compute a power spectrum for each individual comp onent To assign the
observed absorption lines to the primary and secondary unambiguously however we computed
a di erence power spectrum rst not using the absolute radial velocities but the RV di erences
of the comp onent This way we estimated the system s half p erio d and derived nally the p erio d
itself
Subsequently indep endent analyses of b oth comp onents yield individual power spectra This
d d h m s
metho d nally results in a p erio d of for the primary and the secondary s
p ower sp ectrum results in the same p erio d as it should b e for a binary system We nally
coadded the values of these p ower sp ectra creating a combined p ower sp ectrum representing
the whole system As can b e seen from the lower panel of gure the p erio d is quite well
d
de ned Only one alias at still remains The upp er panel of gure shows the b est
t RV curves For the primary we derive km s and K
prim prim
km s The ephemeris is
HJD T E
prim
The corresp onding results for the secondary are km s K km s
sec sec
and
HJD T E
sec
d
Both ephemerides di ers by in terms of HJD T or by
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
RVs of b oth comp onents of HE as a function of orbital phase and tted sine curves
RVs measurements of to the primary are marked as lled symbols while op en symbols indicate
data of the secondary Squares are used for ESO VLT observations and triangles for observations
from the ESO NTT Lower panel Combined p ower sp ectrum of HE
Table System comp osition for HE
Determined from RV curve
km s
M M
prim sec
P d
Deduced parameters
M M
prim
M M
sec
A R
tot
i deg
M M
tot
t Gyrs
m
SN Ia precursor no no no no no
Equation runs to sin i
CHAPTER DOUBLE LINED WHITE DWARFS
System comp osition
From the semi amplitudes of the RV curves we calculate the mass ratio of HE
to b e M M Due to the fact that b oth RV curves also di ers by
prim sec
km s in terms of system velocity we can use the mass radius relation for white
dwarfs to derive the masses of the individual comp onents For the primary the analysis re
sults in M M while the secondary has M M The orbits are A R
prim sec prim
and A R resp ectively Therefore the white dwarfs in the HE system are
sec
separated by A R From the size of the orbits and the p erio d orbital velocities can
tot
b e computed v km s for the primaryand v km s for the secondary By
prim prim
o
comparison with the observed radial velocity amplitudes we derive an inclination of i
Although the total mass M is only b elow the Chandrasekhar limit the large
merging time of Gyrs rules it out as a SN Ia progenitor candidate
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
WD +
V
History
WD was selected from the the Mc Co ok Sion catalogue to b e a DA white
dwarf Thus the ob ject was observed twice in August in the course of our survey
From b oth high resolution sp ectra double absorption line cores b ecame obvious which were
clearly separated from each other by km s and km s resp ectively Being double lined
WD b ecame a high priority ob ject for our follow up campaigns and was re observed
in July and February at the DSAZ Additional data from January were obtained
at the WHT and provided by Gijs Nelemans
Radial velocity curve
The sp ectral sp ectra of b oth comp onents is very similar and thus we cannot distinguish them
by visual means only Therefore we analyze the system in the same way as we did it in the
case of HE see section First of all we computed a di erence power spectrum
from the radial velocity di erences derived from each observation By means of this we could
estimate the half p erio d of the system and subsequently assign the observed absorption lines
to the comp onents Although there is no reason to favor one comp onent in front of the other
we call one of them the primary and the other secondary further on just to b e consistent with
previous discussions
Analyzing the radial velocity data sets of primary and secondary indep endently we computed
individual power spectra However the most likely p erio ds derived from these p ower sp ectra
d d h m s
slightly di er For the primary we found the main p eak at P
prim
d d h m s
while the highest p eak of the secondary s p ower sp ectrum is at P
sec
Moreover b oth p ower sp ectra also show aliases at the p erio d favored by the other comp onent
In some cases a visual insp ection of the corresp onding RV curves could help to ruled out aliases
However for WD this metho d do es not work b ecause neither RV curve show clearly
outliers Thus we can explain the discrepancy b etween the p erio ds only by means of insu cient
data Esp ecially radial velocity measurements based on the medium resolution sp ectra taken at
the DSAZ have relatively large error margins which may b e a reason for the problem Better
observations are necessary in order to derive the system s p erio d b eyond doubt
The lower panel of gure displays the combined power spectrum created by coadding
the values of the individual power spectra Both p eaks discussed ab ove can b e repro duced
d d d
as well as three aliases at P and The strongest p eak however
d
is at P corresp onding to the primary s p erio d The upp er panel of gure
shows the b est t RV curves for this p erio d and result in the most likely system s veloci
ties and the semi amplitudes For the primary we derive km s and
prim
K km s Its ephemeris is
prim
HJD T E
prim
The corresp onding results for the secondary are km s K km s
sec sec
CHAPTER DOUBLE LINED WHITE DWARFS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
RVs of b oth comp onents of WD as a function of orbital phase and tted sine curves
RVs measurements of to the primary are marked as lled symbols while op en symbols indicate
data of the secondary Squares are used for ESO VLT observations and circles for observa
tions from the DSAZ Data provided by Gijs Nelemans are marked as diamonds Lower panel
Combined p ower sp ectrum of WD
and
HJD T E
sec
d
Both ephemerides di er by in terms of HJD T or by consistent with
exp ectation
System comp osition
From the semi amplitudes of the RV curves we calculate the mass ratio of WD to b e
M M Due to the fact that b oth RV curves also di ers by km s
prim sec
in terms of system velocity we can use the mass radius relation for white dwarfs to derive the
masses of the individual comp onents For the primary the analysis results in M M
prim
while the secondary has M M Therefore the primary is most probably a white dwarf
sec
with a C O while the secondary has a core made of helium The orbits are A R and
prim
A R resp ectively Therefore b oth white dwarfs are separated by A R in total
sec tot
From the size of the orbits and the p erio d we compute the orbital velocity of the primary to b e
v km s and v km s for the secondary By comparison with the observed
prim prim
o
RV amplitudes we derive the inclination to b e i
SYSTEMS ANALYZED FROM THEIR RADIAL VELOCITY CURVES ALONE
Table System comp osition for WD
Determined from RV curve
km s
M M
prim sec
P d
Deduced parameters
M M
prim
M M
sec
A R
tot
i deg
M M
tot
t Gyrs
m
SN Ia precursor no no no no no
Neither the system s merging time of ab out Gyrs nor its total mass of M M
tot
quali es WD for b eing a SN Ia precursor
CHAPTER DOUBLE LINED WHITE DWARFS
Systems that cannot b e solved from their RV curves alone
HE a system solved through an additional sp ectral analysis
V
History
HE was discovered by the Hamburg ESO survey as a p otential co ol white dwarf and
therefore was included in our SPY pro ject From the rst survey sp ectrum taken in December
we found HE to b e a double lined binary system consisting of two DA white
dwarfs The H line cores of b oth comp onents were separated by A in the discovery sp ectrum
corresp onding to a large radial velocity di erence of km s Therefore we observed the
system during our follow up observations for the SPY pro ject at the Calar Alto Observatory
th th
Spain From July to July a total of medium resolution sp ectra were obtained
with the TWIN sp ectrograph at the m telescop e Figure shows a sequence of six H
sp ectra taken at Calar Alto Observatory during two hours The observed sp ectra are plotted as
well as the t functions used for RV determination The rapid change of the sp ectral app earance
due to the orbital motion is obvious Our follow up campaign was supplemented by sp ectra taken
with the Isaac Newton Telescope INT and by the two SPY sp ectra taken in December
th
and Octob er The INT set of data used for our analysis was taken on Octob er to
th
Octob er on La Palma Eight sp ectra were taken with the Intermediate Disp ersion
Sp ectrograph IDS
Radial velocity curve
Both comp onents are very similar but the blue shifted one in gure is slightly deep er and
broader This comp onent is called primary further on However sometimes it was di cult to
do an unambiguous identi cation of the comp onents esp ecially near conjunction phases and in
some sp ectra with relatively low S N ratio and of lower resolution from Calar Alto and La
Palma Thus we computed a di erence power spectrum rst in order to estimate the system s
p erio d Thus we were able to allo cate the observed line pro les de nitely to the primary and
the secondary resp ectively cf section
Subsequently an indep endent analysis of b oth comp onents yield individual power spectra
d h m s
b oth showing outstanding p eaks at An insp ection of the phased RV curves
created using p erio ds corresp onding to other p eaks in the p ower sp ectra allowed us to rule out
aliases b ecause of one or more strongly deviating radial velocity measurements This way
the p erio d corresp onding to the main p eak of the p ower sp ectrum remained as an unambiguous
solution for the primary as well as for the secondary exactly as it should b e for a binary system
Finally adding the values of the two individual p ower sp ectra we pro duced a combined power
spectrum for the whole system gure
To derive accurate orbital parameters for HE we used in a nal step the program
fitsb developed by Ralf Napiwotzki cf section and Napiwotzki et al b In a fashion
similar to the metho d describ ed in Maxted et al a a simultaneous t of all sp ectra was
p erformed Two mo del pro les one for each star consisting of a combination of two Gaussians
SYSTEMS THAT CANNOT BE SOLVED FROM THEIR RV CURVES ALONE
Figure Sp ectra of HE taken at the m telescop e of the Calar Alto Observatory
during one night The observed sp ectra are as well plotted as the ts The JD is helio centric
corrected and computed for the mid of the exp osure
for the line core and a Lorentzian for the line wings were tted The p osition of each pro le is
determined from
T T
RV K sin
prim sec prim sec prim sec
P
Free parameters are the mean velocities including gravitational redshift of each comp onent
the pro jected orbital velocities K the zero p oint T and the p erio d P plus
prim sec prim sec
the parameters de ning the line pro les The nal ephemeris of the system for the time T
de ned as the conjunction time at which the primary moves from the blue side of the RV curve
to the red one i e the primary is closest to the observer is
HJD T E
prim
Also from the RV curves see Fig of the fitsb analysis we found the semi amplitude of
the primary to b e K km s km s whereas for the sec
prim prim
ondary the semi amplitude is K km s km s The mass
sec sec
CHAPTER DOUBLE LINED WHITE DWARFS
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
RVs of b oth comp onents of HE as a function of orbital phase and tted sine curves
RVs measurements of to the primary are marked as lled symbols while op en symbols indicate
data of the secondary Squares are used for ESO VLT observations circles for observations from
the DSAZ and diamonds for data obtain at the INT Lower panel Combined p ower sp ectrum
of HE
ratio can b e computed from the ratio of the semi amplitudes to b e M M i e
prim sec
the masses of b oth white dwarfs are identical within the error limits As in the cases of the former
double lined systems the di erence of gravitational redshifts can b e measured from
prim sec
If this di erence is non zero we obtain an additional constraint equation which allows to solve
for the individual masses However since the masses of b oth comp onents of HE are
very similar the redshift di erence is zero within error limits Hence we cannot solve for the
individual masses from the radial velocity curve alone The required additional information
however will b e obtained from a quantitative sp ectral analysis describ ed in the next section
Sp ectral analysis
Since in the case of HE mass estimates from the RV curves alone are imp ossible we
had to rely on a mo del atmosphere analysis Because this system is double lined the sp ectra
are a sup erp osition of b oth individual white dwarf sp ectra Thus we used the program fitsb
Napiwotzki et al b which p erforms a sp ectral analysis of b oth comp onents of double line d
systems see section
A large grid of synthetic sp ectra for DA white dwarfs computed from LTE mo del atmospheres
with a co de describ ed in Finley et al was used for the analysis A simultaneous t of
the Balmer lines H to H UVES sp ectra or H to H TWIN sp ectra was p erformed We
SYSTEMS THAT CANNOT BE SOLVED FROM THEIR RV CURVES ALONE
Table Results of the mo del atmosphere analysis with fitsb The error limits include the
uncertainties of orbital parameter determination
comp T K log g M M
e
prim
sec
did not include the INT sp ectra for the mo del atmosphere analysis b ecause these sp ectra cover
only the H range For details refer to Ko ester et al and references therein The mo del
sp ectra were convolved with Gaussians with FWHMs corresp onding to the resolution of the
observed sp ectra
The total number of t parameters stellar and orbital is high Therefore we xed as
many parameters as p ossible b efore p erforming the mo del atmosphere analysis We have kept
the radial velocities of the individual comp onents xed according to the radial velocity curve
presented in the previous subsection Since the mass ratio is already accurately determined
from the radial velocity curve we xed the gravity ratio The remaining t parameters are
the e ective temp eratures of b oth comp onents and the gravity of the primary The gravity
of the secondary is adjusted according to the primary value during the tting pro cedure The
surface gravities also determine the relative weight of the two mo del sp ectra from the radius
obtained from the mass radius relation of Benvenuto Althaus The ux ratio in the
V band is calculated from the actual parameters and the mo del uxes are scaled accordingly
The individual contributions are up dated consistently as part of the iteration pro cedure
Strong NLTE cores are present in H and H which cannot b e repro duced by our LTE mo del
sp ectra In principle these lines are imp ortant for the temp erature determination esp ecially of
the co oler and fainter secondary However due to the large NLTE e ects we had to exclude H
completely and the core of H A The nal t results are summarized in table and a
sample t is shown in Fig The t of the line cores is not p erfect even after excluding H
The parameter which is most sensitive to details of the tting pro cedure is as may b e exp ected
the temp erature of the fainter secondary If e g we include the wings of the H line we increase
T of the secondary by almost K while the temp erature of the primary and the gravity
e
are only mo di ed within the error limits Thus while the temp erature of the secondary is less
well determined than that of the primary the basic prop erties of the HE system
are reliable Both comp onents have identical masses M M M but di erent
prim sec
temp eratures K vs K White dwarf masses were computed from the mass radius
relations of Benvenuto Althaus with thick hydrogen envelopes M M
H WD
This result do es not dep end much on the choice of a particular mo del computation From the
thick envelope co oling sequences of Woo d we derived virtually identical results M
and M resp ectively However since the HE system is the result of a common
envelope evolution it is not clear which envelope layer is the correct one Using the thin envelope
mo dels M of Benvenuto Althaus as the other extreme would yield slightly lower
H
masses M and M resp The resulting sum of masses of the HE system
is M The error limit for the sum of masses is smaller than exp ected from a simple
combination of the individual errors b ecause the mass errors of the individual comp onents are partially anti correlated
CHAPTER DOUBLE LINED WHITE DWARFS
Figure Sample Mo del atmosphere ts of HE phase sp ectra taken at di erent
phases The discovery sp ectrum taken with the UVES sp ectrograph and two TWIN
sp ectra and
System comp osition
The separation b etween b oth white dwarfs is quite small only R Thus HE
obviously underwent phases of strong binary interaction in its history From the size of the orbits
and the p erio d the orbital velocity can b e computed to b e km s The comparison with the
o
observed RV amplitudes allows us to determine the inclination of this system as i This
system will merge within Gyrs due to gravitational wave radiation Although the merging time
of HE is less than a Hubble time the system s total mass of M will not qualify
it for b eing a SN Ia precursor Nevertheless the critical mass limit was missed by ab out
only and thus HE is close to the region where such progenitors are exp ected
For a more detailed discussion of HE we refer to the pap er of Karl et al
SYSTEMS THAT CANNOT BE SOLVED FROM THEIR RV CURVES ALONE
WD a system in need of an additional sp ectral analysis
V
History
Being classi ed in the Mc Co ok Sion catalogue to b e a DAB white dwarf WD
entered our SPY pro ject This classi cation implies that He i lines are present in the sp ectrum
in addition to the Balmer lines The survey sp ectra were taken in Octob er and December
reveal two H line cores as well as He i at A Moreover it b ecame obvious that the H
line cores changed p osition from one exp osure to the next cf gure For this reason we
6540 6560 6580
Figure Observed line pro les for WD and tted curves The change of the line
p ositions is obvious
classify the system to b e a binary comp osed of a DA white dwarf with a DB companion rather
than a DAB star a scenario already suggested by Bergeron et al However no p erio d
was known so far for WD and thus we carried out follow up observations in order
to solve the system s ephemeris In addition to the survey sp ectra three medium resolution
sp ectra were taken during the September run at the NTT and high resolution sp ectra
at the VLT in November
Radial velocity curve
In order to measure radial velocities we p erformed a simultaneous t of b oth H line pro les
within the ESO MIDAS context The He i line was excluded from the radial velocity determina
tion due to its broad line pro le Sample ts can b e seen from gure It b ecomes obvious
from the sp ectra that the H line pro les of b oth comp onents are very di erent in terms of
line depth and width Therefore a simple by eye classi cation was su cient to allo cate the
observed line pro les to the individual comp onents
As we did the cases of the former double lined systems we tted in the rst instance the
radial velocity sets of b oth comp onents indep endently The resulting p ower sp ectra are very
d h m s
similar b oth yielding the same p erio d of Combining these p ower sp ec
tra we obtained the combined power spectrum of gure representing the complete binary
d m
system Although an alias at that cannot ruled out which di ers by only from
CHAPTER DOUBLE LINED WHITE DWARFS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
RVs of b oth comp onents of WD as a function of orbital phase and tted sine curves
Data p oints b elonging to the DA white dwarf are marked as lled symbols while op en symbols
indicates DAB related data Squares are used for VLT observations and triangles for observa
tions done at the ESO NTT Lower panel Combined p ower sp ectrum of WD The
inset displays the aliases near the main p eak
the main p eak This however is less than the typical exp osure time even for VLT data and
is therefore insigni cant from the discussion that follows From the t RV curve based on the
DA s data set we derive a system s velocity of km s and a semi amplitude of
DA
K km s cf upp er panel of gure The ephemeris for the DA is
DA
HJD T E
DA
Corresp onding results for the DAB are km s and K km s
DAB DAB
resp ectively Its ephemeris is
HJD T E
DAB
d
Both ephemerides di er by corresp onding to which as in almost p erfect
agreement with exp ectations
System comp osition
The mass ratio computed from the the RV semi amplitudes is M M
DAB DA
Due to the fact that is equal to zero within the error margin km s
however an analysis by means of the RV curves only is not p ossible for WD Thus
SYSTEMS THAT CANNOT BE SOLVED FROM THEIR RV CURVES ALONE
we have to p erform a quantitative sp ectral analysis by means of the fitsb program in order to
determine masses by comparison of the derived stellar parameters T and log g with evolutionary
e
calculations cf section This however is much more di cult than for HE
section b ecause the comp onents are so dissimilar This is b eyond the scop e of this thesis
and has still to b e done
If we adopt km s the resulting masses are very low i e M and M
resp ectively and can b e ruled out b ecause the corresp onding sin i according to equation
would need to b e larger than unity Even if we assume the largest the resulting total mass
M is b elow the Chandrasekhar mass table Despite of the merging time b eing lower
than the Hubble the system do es not qualify as a SN Ia progenitor candidate
Therefore we conclude that the derived parameters in particular are not de ned accu
rately enough As in the case of HE we would need an additional sp ectral analysis
to solve the system de nitely
Table System parameters and derived results for WD The input parameters
mass ratio and are varied by multiples of their error margins The assumed combination
of the prop er parameters are displayed in the columns one and two whereas derivated output
parameters are shown in columns three to seven
Determined from RV curve
km s
M M
prim sec
P d
Deduced parameters
M M
DAB
M M
DA
A R
tot
i deg
M M
tot
t Gyrs
m
SN Ia precursor no no no no
Equation runs to sin i
Chapter
White dwarfs with M dwarf
companions
The main intention of SPY is the search for p otential SN Ia precursors by means of checking
double degenerates Obviously binaries comp osed of a white dwarf and a M dwarf are not
double degenerate and thus do not qualify as SN Ia progenitors Nevertheless these ob jects are
interesting with resp ect to the study of binary evolution Therefore we observed three WD dM
systems in the course of the follow up campaigns in order to create an unbiased sample of white
dwarfs in binary systems
A sample sp ectrum of a white dwarf plus M dwarf binary can b e seen from gure
Figure Sample UVES sp ectrum of the white dwarf plus M dwarf system WD
see section The weaker blue shifted emission that can b e seen in H and H is due to a non LTE e ect in the DA s photosphere
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
WD alias RR Caeli
V
History
Classi ed as a DA white dwarf in the McCo ok Sion Catalogue WD was
implemented within our SPY pro ject and observed twice in November in the course of
the SPY pro ject However molecular absorption bands and two emission lines in the Balmer
series b ecomes obvious from these survey sp ectra as can b e seen in gure According to
the sp ectral app earance of WD the system has to b e classi ed as a hot DA white
dwarf and a M dwarf companion This classi cation is consistent with a previous study done
by Bruch et al Follow up observations for WD were done in February
at the ESO NTT and in March at the ESO VLT
Radial velocity curve
Some sp ectra lack the weaker DA emission feature seen in gure b ecause of low S N ratios
and or low resolution of the sp ectra From the H emission lines however we were able to deter
mine radial velocities for b oth comp onents for all but one VLT and for three NTT observations
see gure
The analysis was carried out the same way as for HE cf section By tting
the radial velocities of the DA primary and the dM secondary indep endently we computed
a single p ower sp ectrum for each comp onent We nally merge b oth data sets by coadding
the values derived from the individual power spectra By means of the combined power
d
spectrum pro duced this way cf lower panel of gure we determined a p erio d of
h m s
The inset displays the region around the main p eak in more detail revealing some
aliases While we can not rule out any alias by visual insp ection of the prop er RV curve however
all of them are within a time interval of less than seconds For the white dwarf we nally
derive a system s velocity of km s and a radial velocity semi amplitude of
WD
K km s The ephemeris is
WD
HJD T E
WD
The corresp onding results for the dM are km s K km s and
dM dM
HJD T E
dM
d
Both ephemerides di er by corresp onding to which is in almost p erfect
conformance with theory
System comp osition
According to the RV curves we found the mass ratio to b e M M while the
WD dM
system velocities di ers by km s However the gravitational redshift of the
M dwarf companion can b e neglected compared to the redshift caused by the white dwarf
Therefore equation and the mass radius relation yield M M The companions
WD
mass of M M results from the mass ratio Thus the companion is most likely a M
dM
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities of b oth comp onents of WD as a function of orbital phase and tted
sine curves Data p oints b elonging to the M dwarf are marked as lled symbols while op en
symbols indicates DA related data Like within the previous gures squares are used for VLT
observations and triangles for observations from the ESO NTT Lower panel Combined p ower
sp ectrum of WD The inset displays some of the aliases near the main p eak
dwarf see Drilling et al Both comp onents of the WD system are separated
by A R which means the white dwarf is at A R from the barycenter and the
tot WD
M dwarf at A R From the orbital velocities of v km s and v km s
dM WD dM
o
we compute an inclination of i Due to the radiation of gravitational redshift b oth
Table System comp osition for WD
Determined from RV curve
km s
M M
DA dM
P d
Deduced parameters
M M
WD
M M
dM
A R
tot
i deg
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
comp onents will merge in t Gyrs but WD will not pro duce a SN Ia event
m
b ecause the the dM is not degenerate This would b e a single degenerate scenario
Comparison with previous results
d
The p erio d of estimated ab ove is in almost p erfect agreement with the p erio d of
d
derived by Bruch et al by means of a light curve analysis However their analy
sis yields M and M for the white dwarf and the M dwarf resp ectively M M
WD dM
which can de nitely b e ruled out by our own analysis
WD
V
History
The McCo ok Sion catalogue classi ed WD as a DAO white dwarf Vennes et
al however found the DAO b eing only one comp onent of a quadruple system comp osed
of a wide double binary In the course of the SPY pro ject the DAO was observed twice at
the VLT in Octob er and November From these high resolution sp ectra emission features
b ecome obvious in the Balmer series and for He i and C i i lines see gure The Balmer line
emission b elongs to the M dwarf companion as do the emission lines which b ecomes obvious from
radial velocity measurements We also found variation of the line strengths For instance the H
emission feature has W A at while at the line strength is W A only
In a system with gravitational b ound rotation this can b e understo o d if the dM s hemisphere
facing toward the white dwarf is heated by the hotter DAO companion We conclude that this
is the most probable scenario for WD
2.5
2
1.5
1
0.5
3600 3800 4000 4200 4400
Figure UVES sp ectrum of WD taken at The DAO absorption sp ectrum
is overlayed by the dM s emission lines
Vennes et al already found the H emission pro le to b e split We detect the same
feature for the entire Balmer series as well as for some of the stronger Hei lines like A
Figure shows H and H emission for example The physical reason is most probably a
self absorption e ect in the dM s heated atmosphere
WD was also observed during the January NTT run
Radial velocity curve
From the observed emission lines we estimated a set of radial velocities for the dM companion
d h m s
From the p ower sp ectrum calculated by means of this data set we obtain
for the most probable p erio d Though our result is already close to the p erio d given by Vennes
m s
et al P we can improve our analysis if we merge Vennes data set with our own
d h m s
This way the p erio d b ecomes more accurate or The ephemeris is
HJD T E
dM
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
6540 6560 6580 4840 4860 4880
Figure UVES sp ectrum of WD taken at The splitted emission feature
in H right and H left is most probably due to self absorption in the dM s atmosphere
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities of WD as a function of orbital phase and tted sine curves Squares
are used for VLT observations triangles for observations from the ESO NTT and diamonds for
data taken from Vennes et al Lower panel Power sp ectrum of WD
Over a course of years this ephemeris di ers from Vennes result by less than
P The systems velocity derived from the b est t RV curve upp er panel of gure is
km s and the radial velocity semi amplitude is K km s
dM dM
System comp osition
Vennes et al derived a white dwarf with a mass of M M M and a
DAO
o o
M red dwarf companion of M They found the system s inclination to b e i
For more details on the dM DAO binary we refer to Vennes et al They also give a
short discussion ab out the second binary of the WD quadruple system
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
HS +
V
History
This ob ject was classi ed as a white dwarf from the Hamburg Quasar survey Thus it entered
our SPY pro ject and was observed twice at the VLT From the UVES sp ectra taken in July
however strong emission lines sup erp osing the H and H lines b ecame evident Thus
we re classi ed HS to b e a white dwarf plus M dwarf binary Moreover a radial
velocity shift of ab out km s was measured from the survey sp ectra Thus the system was
re observed in the course of our follow up campaigns in August at the DSAZ and in March
at the VLT
Radial velocity curve
By means of tting the emission feature of the M dwarf we obtained very accurate radial
velocity measurements Figure shows the resulting p ower sp ectrum and the b est t RV
d d h m s
curve We derive a most probable p erio d of a system s velocity of
km s and a semi amplitude of K km s for the M dwarf Its
ephemeris b ecomes
HJD T E
System comp osition
Although the RV curve lo oks normal at rst glance we derive a very large mass function of
f M Following Drilling et al we assume a M dwarf with a canonical mass
m
of M for the visible comp onent Thus we derive the minimum mass of the companion for
o
an inclination of i to b e M M Even if we adopt the lowest p ossible mass of
comp
M for the M dwarf the companion s mass is still M A companion as massive as that
has to b e either a main sequence star a neutron star or a black hole These ob jects however
are inconsistent with the sp ectral app earance of HS indicating that the companion
is a white dwarf as can b e seen from gure
The inconsistency of mass function and sp ectral app earance is caused by the large value
of f We therefore conclude that either the p erio d or the radial velocity semi amplitude are
m
falsely determined There are still more observations necessary to solve HS
Figure Best t RV curve and p ower sp ectrum for HS Upp er part Measured RVs
as a function of orbital phase and tted sine curve Squares are used for VLT observations and
circles for observations done at the m telescop e of the DSAZ Lower part Power sp ectrum of HS
1.2
1
0.8
0.6
0.4 3800 4000 4200 4400
2
1.5
1
6400 6500 6600
Figure Observed line pro les of HS The displayed wavelength regions are
th
extracted from a UVES sp ectrum taken in March the The dM s emission line sup er
p osing the H pro le is obvious as well as the higher Balmer lines of the white dwarf companion
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
Table Scenarios for HS Note that the nature of the companion is inconsistent
with the sp ectral app earance of HS which suggested a white dwarf companion
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i M M no MS NS or BH
vis
most likely i M M no MS or BH
vis
max i M M no MS NS or BH
vis
most likely i M M no MS or BH
vis
HE
V
History
HE was selected from the Hamburg ESO survey by Norb ert Christlieb to b e a
white dwarf and thus entered our SPY pro ject The survey sp ectra of HE taken in
August however reveal not only the Balmer series but also emission features that are
characteristic for a M dwarf companion Not b eing a double degenerate system HE
was nevertheless re observed in the course of our follow up campaigns b ecause of a large radial
velocity shift of more than km s evaluated from the the survey sp ectra Since the strong
emission line of the M dwarf can b e easily tted within the MIDAS context the DSAZ data
taken in August result in very accurate radial velocities
Radial velocity curve
d d h m s
Our analysis yields an unequivocal p erio d of P or From the b est t
RV curve displayed in the upp er panel of gure a system s velocity of km s
and a semi amplitude of K km s can b e derived The system s ephemeris is
HJD T E
System comp osition
We have no information ab out the motion of the white dwarf but ab out the M dwarf companion
only To estimate the white dwarf s mass from the mass function f M we assume the
m
M dwarf to b e of sp ectral type M with a mass of M Drilling et al The maximum
o
inclination of i yield to the white dwarf s minimum mass of M M Orbits
WD
and orbital velocities can b e computed from equations and to b e A R and v
dM dM
K km s resp ectively Corresp onding results for the white dwarf are A R
WD
and v km s Therefore the total separation of the stars according to this scenario is
dM
A R
tot
o
For the most probable inclination of i however we derive M M for
WD
the white dwarf The orbits are A R and A R resp ectively Orbital ve
WD dM
lo cities of v km s and v km s Therefore the system s separation b ecomes
WD dM
A R
tot
We conclude that the white dwarf in the HE system is most probably a DA
white dwarf with a C O core Table
CHAPTER WHITE DWARFS WITH M DWARF COMPANIONS
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities of HE a function of orbital phase and tted sine curves Squares are
used for VLT observations and circles for data based on DSAZ observations Diamonds mark
radial velocity data provided by Gijs Nelemans Lower panel Power sp ectrum of HE
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no low mass WD
most likely i no C O core WD
M M no most massive WD
invis Chandra
Chapter
Single lined systems
Most of our ob jects are single lined rather than double lined systems Thus only the system s
mass function equation can b e derived This however is an equation of three unknown
parameters Without additional information to set further constrains on the systems we cannot
solve for them
Nevertheless by assuming that the observable comp onent has a canonical mass we can
decrease the number of free parameters by one Thus we estimate a white dwarf s mass by means
of comparing its stellar parameters T and log g with evolutionary calculations by Bl ocker et
e
al cf gure For sdBs and sdOs we assume canonical masses of M The last
free parameter can b e eliminated by means of xing the system s inclination The statistically
o
most probable inclination angle is i which yields the most probable mass for the invisible
comp onent A lower limit for the comp onent s mass however can b e derived by assuming the
o
maximum inclination value of i The complete sequence of analyzing a single lined system
is displayed in gure in a schematic way
In section we discuss radial velocity variable single lined systems with a white dwarf
primary Systems with an sdO primary are discussed in section whereas sdB systems are
discussed in section
White dwarf primaries
A typical UVES sp ectrum of a hydrogen rich DA white dwarf can b e seen from gure
Most of the DA white dwarfs presented in the following chapter are already analyzed by Detlev
Ko ester by means of a quantitative sp ectral analysis Ko ester et al and priv com of the
SPY sp ectra For these ob jects we estimated masses of the visible comp onents by comparison of
the determined stellar parameters T and log g with co oling sequences of white dwarfs Bl ocker
e
et al cf section
CHAPTER SINGLE LINED SYSTEMS
Spectral Analysis RV curve
log g T v sin i P Κ eff rot 1
f M
Ρ i M−R relation 1
Μ Μ 1 2
Μ P 1,2
A 1,2
v 1,2
M t
tot M
Figure Analysis of a single lined system schematic
WHITE DWARF PRIMARIES
WD
V
History
Being classi ed as a DA white dwarf in the McCo ok Sion catalogue WD
was observed twice in July at the VLT A shift of km s was found from the H line
pro les In order to solve the system we carried out follow up observations at the ESO VLT in
Octob er and November and at the ESO NTT in September
Radial velocity curve
The resulting p ower sp ectrum displayed in the lower panel of gure shows an unambiguous
d d h m s
p erio d of while the upp er panel of the gure shows the b est t RV
curve computed for this p erio d We derive a system s velocity of km s a radial
velocity semi amplitude of K km s and an ephemeris of
HJD T E
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve Squares are used for
ESO VLT observations and triangles for observations from the ESO NTT Lower panel Power
sp ectrum of WD
CHAPTER SINGLE LINED SYSTEMS
Sp ectral analysis and mass determination
WD is one of the ob jects already analyzed by Detlev Ko ester in order to derive stellar
parameters cf section By comparison of these parameters T K and log g
e
with evolutionary tracks by Bl ocker et al we estimated the white dwarf s mass to b e
M M
WD
System comp osition
o
Assuming an inclination of i the mass function of f M yields a minimum mass
m
of M M for the unseen comp onent The separation b ecomes A R Based
comp tot
on this scenario we computed the orbits and orbital velocities by means of equations and
For the visible comp onent we derive A R and v K km s whereas
WD WD
the corresp onding results for unseen comp onent are A R and v K km s
comp comp
resp ectively
o
However the system s most probably inclination is i which yields the compan
ion s most likely mass of M M The separation b ecomes A R and the
comp tot
white dwarf s orbit A R v km s For the unseen companions we compute
WD WD
A R v km s to b e most likely
comp comp
Although the unseen comp onent is most probably a white dwarf with a C O core neither
scenario discussed ab ove will quality for a SN Ia precursor In b oth cases the merging time as
well as the system s total mass are insu cient as can b e seen from table The table shows
in addition the inclination angles yielding a total mass equal the Chandrasekhar mass and the
most massive white dwarf companion of M resp ectively
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no C O core WD
tot Chandra
M M no most massive WD
invis Chandra
WHITE DWARF PRIMARIES
WD +
V
History
WD is classi ed as a DA white in the McCo ok Sion catalogue and was
analyzed for radial velocity variability by Maxted et al However they observed the
system twice and found a radial velocity shift of km s only RV within the error
limits therefore they concluded the system to b e non variable We re observed the system
in the course of our SPY program to check this result Two survey observations were done in
January at the VLT given a much larger radial velocity shift of km s Follow up
observation from the NTT run in April and May provided a data set accurate enough for
a radial velocity analysis
Radial velocity curve
The p ower sp ectrum displayed in gure is not unequivocal since a large number of aliases
can not b e ruled out see inset of gure These aliases are concentrated in a time interval
m
of centered on the main p eak However for our purp oses a p erfect determination of the
p erio d is desirable but not necessary Using the highest p eak of the p ower sp ectrum we derive
d h m
the most probable p erio d s The system parameters derived from the b est
t RV curve are km s for the system s velocity and K km s for
the radial velocity semi amplitude The ephemeris of WD is
HJD T E
Sp ectral analysis and mass determination
Ralf Napiwotzki priv com determined the stellar atmospheric parameters of WD
to b e T K and log g Thus we estimated a mass of M M by means
e WD
of the Bl ocker evolutionary tracks cf gure
System comp osition
The mass function computed from these parameters is f M Using the mass we
m
estimated ab ove for the visible companion we derive the minimum mass of the unseen companion
o
to b e M M for i The companion is therefore most likely a late type main
comp
sequence star The separation of the two comp onents b ecomes A R which yields the
tot
orbits of the comp onents For the white dwarf we compute A R and for the unseen
WD
comp onent A R The orbital velocities determined by means of equation are
comp
v K km s and v K km s resp ectively
WD comp
o
The most probable system comp osition however results for an inclination of i
The companion s mass b ecomes M M which makes it a C O core white dwarf
comp
The orbits are A R for the visible white dwarf and A R for the unseen
WD comp
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for
ESO VLT observations and circles for observations done at the DSAZ Diamonds mark data
taken from Maxted et al Lower panel Power sp ectrum of WD The inset
shows the region near the main p eak in more detail
companion Thus the system s total expansion b ecomes A R Orbital velocities of
tot
b oth comp onents can b e computed to b e v km s and v km s resp ectively
WD comp
As can b e seen from table WD will not qualify as a SN Ia precursor due to
its large merging time Even for the most massive white dwarf companion of ab out M the
merging time will still exceeds the Hubble time
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no C O core WD
tot Chandra
M M no most massive WD
invis Chandra
WHITE DWARF PRIMARIES
HS +
V
History
HS was selected from the Hamburg Quasar Survey to b e a DA white dwarf The rst
survey observation was done in June followed by another VLT observation in April
Though we found a radial velocity shift of only km s from these sp ectra the exp ected radial
velocity semi amplitude seemed to b e very small and therefore we exp ected HS not
to b e a SN Ia candidate Since it is su cient bright the system was re observed in July
at the DSAZ in order to create an unbiased sample of double degenerates for probing binary
evolution theory
Radial velocity curve
d h m s
The p ower sp ectrum yields a most probable p erio d of but the result is
d d d
not unambiguous There are some aliases at P and that can
not b e ruled out by visual insp ection of the corresp onding RV curves The p ower sp ectrum of
HS is displayed in the lower panel of gure while the upp er panel shows the b est
d
t RV curve computed for the most probable p erio d of The ephemeris is
HJD T E
The system velocity and semi amplitude derived from the RV curve are km s
and K km s resp ectively
Sp ectral analysis and mass determination
Like WD HS was sp ectroscopically analyzed by Detlev Ko ester By
means of this results T K and log g we estimated a white dwarf s mass of
e
M M from the tracks of Bl ocker et al
WD
System comp osition
o
Assuming an inclination of i we determined the companion s minimum mass to b e
M M Thus the white dwarf circles around the barycenter at a distance of
comp
A R and at a orbital velocity equal to the semi amplitude K The companion is
WD
at A R from the system s barycenter at an orbital velocity of v km s
comp comp
The total separation of b oth comp onents b ecomes A R
tot
o
For the most probable inclination of i however the companion has M M
comp
A R and v km s Orbital size and velocity for the white dwarf are
comp comp
A R and v km s yielding A R for the system s total extend
WD WD tot
The companion of the visible white dwarf is most probable a massive brown dwarf However
o
for a system s inclination b elow the companions mass exceeds M which is the minimum
mass for central hydrogen burning Thus a late type M star can not b e ruled out either Table summarizes di erent scenarios for HS
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for HS Upp er panel Measured
RVs of HS as a function of orbital phase and tted sine curve The data p oints are
d
phase according to the most probable p erio d of Squares are used for VLT observations
and squares for observations from the DSAZ Lower panel Power sp ectrum of HS
Table Scenarios for HS
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no typical C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
WHITE DWARF PRIMARIES
WD +
V
History
WD was rst detected to b e a radial velocity variable white dwarf by Sa er et
al However since it is found in the McCo ok and Sion catalogue this sys
tem was indep endently observed at the VLT in May in the course of the SPY pro ject
The rst survey sp ectra already reveal an unequivocal radial velocity shift of ab out km s
con rming the variability found by Sa er et al Additional medium resolution sp ectra
were taken with TWIN at the DSAZ in March and July and in February The data
set was complemented by ve high resolution sp ectra taken over the course of at the VLT
Radial velocity curve
The resulting p ower sp ectrum is displayed in the lower panel of gure The unambiguous
d d h m s
p erio d of P is obvious From the b est t RV curve we derive a system s
velocity of km s and a radial velocity semi amplitude of K km s
d
Maxted et al has already found a p erio d of and a semi amplitude of km s for
WD which is in almost p erfect agreement with our own analysis The ephemeris of
is
HJD T E
Sp ectral analysis and mass determination
The sp ectral analysis p erformed by means of our high resolution UVES sp ectra gives T K
e
and log g Ko ester priv com Therefore we estimated the white dwarf s mass to b e
M M Although these results di ers slightly form a previous study of low resolution
WD
sp ectra Bragaglia et al T K log g and M M we rely on our
e WD
own results for the ongoing discussion
System comp osition
o
For an inclination of i the mass function of f M results in the companion s
m
minimum mass of M M The separation of the two comp onents is A R
comp tot
Orbit and orbital velocity of the visible white dwarf can b e computed from equations and
to b e A R and v K km s Corresp onding results for the companion
WD WD
are A R and v km s
comp comp
o
For the most probable inclination of i however the mass function yields a companion
with a mass of M M and a orbit of A R v km s According
comp comp comp
to this scenario the distance of the white dwarf to the barycenter can b e computed to b e
A R v km s This results in a total separation of R R
WD WD tot
On the basis of these results we conclude that the unseen companion in the WD
system is most likely a C O core white dwarf Although the system s total mass reaches the
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for VLT
observations and circles for data based on DSAZ observations Lower panel Power sp ectrum of
WD
o
Chandrasekhar limit in the case of the i scenario its merging time is to o large to qualify
WD as a SN Ia precursor Gyrs and Gyrs resp ectively cf table
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no most massive WD
invis Chandra
SUBDWARF B PRIMARY
Sub dwarf B primary
Subluminous B stars are generally considered to b e core helium burning stars with thin hydrogen
envelopes of less than M and total masses of ab out M In the Hertzsprung Russell
diagram they p opulate a very narrow area which lies on the blue ward extension of the horizontal
branch the so called Extreme Horizontal Branch From there they evolve directly to the white
dwarf graveyard without reaching the Asymptotic Giant Branch previously b ecause their thin
envelope cannot sustain hydrogen shell burning According to standard stellar evolution theory
however sdB star should not exist Their detailed evolution is still unclear and several scenarios
are still under debate cf Han et al Figure shows a typical example for an sdB s
sp ectrum taken with UVES
Although not b eing degenerate sdB stars are immediate progenitors of white dwarfs and
thus promising ob jects with resp ect to our search for sup ernova Ia precursors The only likely
SN Ia progenitor according to the merging scenario known so far is not a double degenerate but
h m
the sdB plus white dwarf binary KPD Maxted et al c Its p erio d is
which yields a white dwarf companion of at least M which is almost certainly a white
dwarf Since the sdB will evolve directly to a white dwarf and merge in less than Gyrs with
the companion KPD is indeed a very go o d candidate for a SN Ia event However
this result is still discussed since evolutionary calculations show that KPD will form
a ONeMg white dwarf rather than a SN Ia precursor Ergma et al
Unlike white dwarfs sdB stars display a large variety of metal lines in their sp ectra These
lines can b e used in order to estimate the pro jected rotational velocities v sin i from quan
rot
titative sp ectral analyses of the ob jects cf section Finally the systems inclination i
themselves could b e derived from equation by means of P and R The analysis has b een
p erformed in three steps
Creating high S N templates
In preparation for further analysis a high S N template was created for each ob ject rst For that
purp ose all available UVES sp ectra were weighted by their S N ratio and shifted to lab oratory
wavelengths Subsequently they were coadded in order to increase the S N ratio of the template
Atmospheric parameters
Although Lisker et al already determined stellar atmospheric parameters for sdB stars
by means of the SPY sp ectra we p erformed an indep endent analysis by means of the high S N
coadded UVES templates E ective temp eratures T surface gravities log g and helium
e
abundances n log n n were determined by tting simultaneously each observed hy
He He H
drogen and helium line to a grid of synthetic mo del sp ectra Therefore a pro cedure developed
by Napiwotzki et al based on Sa er et al was used The mo del grid was comp osed
of metal line blanked LTE mo del atmospheres Heb er et al with solar metal abundances
Figure shows a typical t result
For each line the continuum level was determined and normalized to one in order to compare
it to the synthetic sp ectrum The H line was never used for the parameter determination
b ecause its line core is sensitive to non LTE e ects and thus a LTE mo del has to fail repro ducing
the line However H was kept in the plot for examining p ossible deviations from the mo del H line
CHAPTER SINGLE LINED SYSTEMS
Figure Sample UVES sp ectrum of an sub dwarf B star HE see section
Metal abundance patterns
The equivalent widths were measured by means of tting Gaussians to the high S N templates
within the MIDAS context see Herrmann The central intensity and the full width at
half maximum FWHM were adjustable parameters which were subsequently used to compute
equivalent widths Results can b e seen from table D in the app endix
Thereafter mo del atmospheres were generated with the ATLAS co de of Kurucz
using the atmospheric parameters derived ab ove and solar abundances These mo dels were used
to calculate curve of growth for the observed metal lines Blends from di erent analysis were
omitted from the analysis Finally the abundances were determined from a detailed sp ectrum
synthesis by means of LINFOR using all metal lines
The atomic data used for this analysis were taken from the lists of Wiese et al
Kurucz Ekb erg and Hirata Horaguchi
Closely connected to the determination of metal abundances is the measurement of micro
scopical turbulences in the star s atmosphere These turbulences broadens the observed sp ectral
1
LINFOR was originall y developed by Holweger Ste en and Steenbok at Kiel University It has
b een enhanced and maintained by Lemke and was recently mo di ed by Przybilla For a description see
http www sternwarte uni erlangen de ai linfit linfor html
SUBDWARF B PRIMARY
7 He I 4026 He I 4472 log g = 5.32 He I 4922 6 log (He/H) = -3.21 He I 5016
He I 5876 5
4
3
2
1
0
-40 -20 0 20 40
Figure Sample mo del t of an sub dwarf B star HE see section
lines due to the Doppler e ect However if a su cient number of lines of one ion over a wide
range of strength is measured the microturbulence can b e estimated Lines at the at part of
the curve of growth are very sensitive to whereas lines at the linear part are insensitive By
varying a di erent slop e of the linear regression tted to the determined abundances versus
the equivalent widths results Figure demonstrates this e ect on the example of the oxygen
abundances determined from the O i i lines of He A vanishing slop e of the linear
regression is the means to determine the value of
Pro jected rotational velocity and inclination
Using the LINFOR program package we computed synthetic sp ectra over a small wavelength
range In order to do this we had to consider the previously determined parameters T log g
e
n and These pro les were subsequently folded with an adopted rotational velocity and
He
tted to the observed line pro les given by the template The quality of the t was estimated
by means of a test This way the b est t and the prop er pro jected rotational velocity was
derived An example for the determination of v sin i is shown in gure for HE rot
CHAPTER SINGLE LINED SYSTEMS
8
7.8
7.6
7.4
7.2 0.5 1 1.5 2
8
7.8
7.6
7.4
7.2
0.5 1 1.5 2
Figure Determination of the microturbulence for HE by means of the N i i
results Values of km s upp er panel and km s are applied lower panel For km s
the slop e of the regression line vanishes as exp ected if the choice of is correct
5000 5000.5 5001 5001.5 5002 5002.5
Figure Determination of the pro jected rotational velocity v sin i for the sdB star
rot
HE The lines used therefore are the N i i lines at A and A re
sp ectively was adopted
SUBDWARF B PRIMARY
WD
V
History
Classi ed as a DA white dwarf in the McCo ok Sion catalogue high resolution UVES
sp ectra were taken of WD in July for the SPY pro ject From these sp ectra
however it b ecame obvious that WD is a sub dwarf B star rather than a white dwarf
Moreover a radial velocity shift of km s was measured although a previous study by Bra
gaglia et al derived no variation for this ob ject from low resolution sp ectra Since the star
is bright and varies at a su ciently large radial velocity amplitude we included WD
into our follow up campaign Numerous observations were done at the VLT Octob er and
November the DSAZ m telescop e August and Octob er and the NTT Septem
b er The data set was complemented by WHT observations done in Octob er and in
August done by Gijs Nelemans
Radial velocity curve
The lower panel of gure shows the resulting p ower sp ectrum with the main p eak at
d d h m s
The b est t RV curve for the main p erio d can b e seen from the upp er
panel of gure We derive a system s velocity of km s a radial velocity
semi amplitude of K km s and an ephemeris of
HJD T E
Sp ectral analysis
A total of nine high resolution UVES sp ectra allows a more detailed analysis of the visible
comp onents of WD By means of a sp ectral analysis an e ective temp erature of
T K a surface gravity of log g and a helium abundance of n was
e He
derived The helium abundance however is b elow the lower b oundary of the used mo del grid
and thus only an upp er limit can b e determined These results are in almost p erfect agreement
with the stellar parameters derived by Lisker et al from the survey sp ectra only
The coadded sp ectrum furthermore reveals lines of several ions N i i O i i Mg i i Al i i i Si i i i
and iv S i i i Ti i i i and Fe i i i The abundances derived for these ions are shown in table and
displayed in gure with resp ect to the solar value Regarding prior studies e g Edelmann
WD shows no p eculiar abundance patterns except for the unusual titanium
abundance The curve of growth of nitrogen oxygen and iron are consistent with
A detailed analysis in order to determine the pro jected rotational velocity by means of the
Mg i i doublet yields an upp er limit of km s for v sin i only This result is not astonishing
rot
b ecause former studies e g Edelmann already showed that single star sdBs are usually
slow rotators The same however can b e exp ected for sdB stars in wide binaries without tidal
lo cking like WD
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve Squares are used for
ESO VLT observations triangles for observations from the ESO NTT and observations done at
the DSAZ m telescop e are marked as circles Lower panel Power sp ectrum of WD
Table LTE metal abundances for WD The ion is given in the rst column n is
the number of lines used for the analysis and log is the derived abundance
ion n log
N i i
O i i
Mg i i
Al i i i
Si i i i
Si iv
S i i i
Ti i i i
Fe i i i
System comp osition
The mass function of f M results in a minimum mass of M M for the
m comp
unseen comp onent assuming a canonical mass of M for the visible comp onent and an
o
inclination of i The total separation b ecomes A R which means the vis
tot
ible sub dwarf is at A R from the barycenter whereas the unseen companion is at
sdB
A R The orbital velocities of b oth comp onents are v K km s and
comp sdB
SUBDWARF B PRIMARY
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for WD relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles from
double ionized ions as lled circles and from triple ionized ions as lled triangles
v km s resp ectively
comp
o
However the most probable value of the inclination is i which gives a companion
mass of M M and a system s total separation of A R Orbital size and
comp tot
orbital velocity of the sub dwarf b ecome A R and v km s resp ectively The
sdB sdB
corresp onding results for the unseen companion are A R and v km s
comp comp
Based on the results derived ab ove we nally conclude that the unseen comp onent is most
likely a white dwarf with a C O core comp osition The system however will not qualify as a
SN Ia precursor b ecause its total mass will neither exceeds the Chandrasekhar mass nor will it
merge within a Hubble time as can b e seen from table Even for the most massive M
white dwarf companion the system s merging time will not b e b elow a Hubble time
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
Being selected from the Hamburg ESO survey HE was observed twice for SPY at
the VLT in April These sp ectra revealed helium lines at A and A and therefore
the star seemed to b e a sub dwarf rather than a white dwarf A classi cation that was con rmed
later on by means of a quantitative sp ectral analysis Nevertheless a large radial velocity shift of
ab out km s b ecame obvious as well and therefore the star was a very promising candidate
for follow up observation However its lo cation in the southern hemisphere rules out most of the
observational sites used for SPY follow ups only the ESO telescop es qualify for re observation
In fact HE was re observed at the ESO VLT during our runs in Octob er in
November and in March
Radial velocity curve
The p ower sp ectrum derived for HE is unambiguously and yields a p erio d of
d h m s
P From the b est t RV curve that can b e seen in the upp er panel
of gure we estimated a system s velocity of km s and a semi amplitude of
K km s The ephemeris b ecomes
HJD T E
Power sp ectrum and b est t RV curve for HE can b e seen from gure
Sp ectral analysis
HE was observed times at the VLT providing a large number of high quality
sp ectra yielding a coadded template with an S N ratio of A quantitative sp ectral analysis
results in very accurate stellar parameters of T K log g and n which
e He
are consistent with the results of Lisker et al
We found sp ectral lines of C i i N i i O i i Mg i i Si i i i S i i and S i i i and Fe i i i The abundances
derived for these ions are shown in table and shown in gure with resp ect to the solar
value As can b e seen from this plot HE shows no unexp ected p eculiarities among
its abundance patterns but lo oks like an ordinary sub dwarf B type star The curve of growth
analyses of nitrogen oxygen and iron reveal no hint for microturbulence Thus we assume
further on
A detailed analysis in order to determine the pro jected rotational velocity by means of the
N i i lines b etween A and A results in v sin i km s
rot
System comp osition
Assuming a canonical mass of M for the sdB the radius of the star b ecame R cf
o
equation For b ounded rotation this nally yields a system s inclination of i by
means of comparison of the computed rotational velocity with the measured one However the
SUBDWARF B PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve All measurements are based
on ESO VLT observations Lower panel Power sp ectrum of HE
Table LTE metal abundances for HE The ion is given in the rst column n is
the number of lines used for the analysis and log is the derived abundance
ion n log
C i i
N i i
O i i
Mg i i
Si i i i
S i i
S i i i
Fe i i i
mass function of f M results in M M for the companion which therefore
m comp
has to b e a black hole The separation b etween the sdB and the black hole b ecome A R
tot
and the orbital size and velocity of the sdB are A R and v km s resp ectively
sdB sdB
Although a sdB plus black hole binary is not imp ossible it is still unlikely In order to discuss
the evolutionary status of HE in more detail we therefore adopted a p ost RGB evo
lutionary scenario for HE instead of the usual p ost EHB evolution cf the discussion
for HD in Heb er et al Thus we derive a mass of M by means of comparison
of the stars p osition in the T log g plane with evolutionary track computed for p ost RGB e
CHAPTER SINGLE LINED SYSTEMS
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for HE relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles from
double ionized ions as lled circles and from triple ionized ions as lled triangles
evolution This can b e seen graphically from gure Thus the star s radius and subse
o
quently the system s inclination change to R R and i The companion s mass
sdB
however is still M M which means the companion has to b e a neutron star therefore
comp
For the p ost RGB scenario the separation of the system b ecomes A R whereas the
tot
orbital size and velocity of the sdB are A R and v km s resp ectively
sdB sdB
Both scenarios exempli ed ab ove results in very massive companions In the following
we abandon the assumption of b ounded rotation and discuss the nature of the companion
for the maximum and the most probable inclination angle Assuming the sdB star having a
o
canonical mass of M the companion s minimum mass computed for i b ecomes
M M which is either a low mass He core white dwarf or a late type main sequence
comp
star The separation of the system can b e calculated to b e A R and the orbital pa
tot
rameters for the comp onents are A R and v K km s for the sdB and
sdB sdB
A R and v km s for the companion resp ectively
comp comp
o
The most probable inclination however is i which yields M M for the
comp
companion s mass Thus the companion is again either a low mass white dwarf or a late
type main sequence star The orbital parameters for the comp onents are A R and
sdB
A R for the orbital sizes and v km s and v km s for the orbital
comp sdB comp
velocities The total extend of the system is A R
tot
Table summarizes the results of di erent scenarios for HE
SUBDWARF B PRIMARY
4.5
5.0
5.5
6.0
6.5
45 40 35 30 25 20 15
Figure Position of HE in the T log g plane and comparison with the low
e
mass evolutionary tracks computed by Drieb e et al for p ost red giant branch p ost
RGB evolution These tracks represent the evolution of stars whose hydrogen rich envelope
was stripp ed away by a companion during the rst red giant branch phase The remaining mass
is to o low for the ignition of helium burning and the star will eventually b ecome a low mass
white dwarf with a helium core
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
b d rot p ost EHB no black hole
b d rot p ost RGB no neutron star
max i no Low mass WD or late type MS
most likely i no Low mass WD or late type MS
M M yes massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
HE was selected from the Hamburg ESO survey by Norb ert Christlieb priv com
to b e a hydrogen rich DA white dwarf The high resolution sp ectra taken in the course of
SPY however reveal that this ob ject is in fact a sub dwarf of sp ectral type B Although one of
the survey sp ectra was of low quality only a radial velocity shift of more than km s was
estimated Thus HE was re observed in January at the WHT in February at
the NTT and in March at the VLT
Radial velocity curve
d h m s
From the p ower sp ectrum a p erio d of P results whereas the sys
tem s velocity was determined to b e km s and the RV semi amplitude to b e
K km s The ephemeris of HE b ecomes
HJD T E
Power sp ectrum and b est t RV curve can b e seen from gure
Sp ectral analysis
HE was observed eight times totally at the VLT Even after discarding the low S N
survey observation seven high resolution UVES sp ectra remain to b e coadded in order to obtain
a high S N template By means of a sp ectral analysis we derive T K log g and
e
n which is in almost p erfect agreement with the results of Lisker et al based
He
on the survey sp ectra only
Furthermore we identi ed sp ectral lines of the C i i N i i O i i Ne i i Mg i i Al i i i Si i i i
S i i i and Fe i i i ions The abundances are shown in table in more detail In addition
gure displays the derived abundance patterns with resp ect to solar values The derived
neon abundance is regarded as an upp er limit b ecause the equivalent width measurements yield
values only slightly ab ove the detection limit However the two lines are among the strongest
lines exp ected in the Ne line sp ectrum The microturbulence was deduced from the curve of
growth of the N i i and O i i lines to b e km s
The pro jected rotational velocity was determined by means of tting synthetic line pro
les to the Ne i i lines A A A A and A Thus
v sin i km s was derived
rot
System comp osition
From the pro jected rotation velocity a stellar radius of R R was determined for an
sdb
assumed canonical sdB mass of M cf equation Thus the inclination b ecomes
o
i The mass function of f M yields a companion s mass of M M
m comp
Therefore the companion is a neutron star or black hole The separation of b oth comp onents
SUBDWARF B PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve Squares are used for
ESO VLT observations and triangles for observations from the ESO NTT Data taken at the
WHT are provided from Gijs Nelemans and are marked as diamonds Lower panel Power
sp ectrum of HE
Table LTE metal abundances for HE The ion is given in the rst column n is
the number of lines used for the analysis and log is the derived abundance
ion n log
C i i
N i i
O i i
Ne i i
Mg i i
Al i i i
Si i i i
S i i i
Ar i i
Fe i i i
is A R The orbital velocity of the sub dwarf is v km s and its orbital size
tot sdB
A R
sdB
Like in the case of HE we alternatively consider p ost RGB evolution for
HE By means of comparison of the determined stellar atmospheric parameters
CHAPTER SINGLE LINED SYSTEMS
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for HE relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles from
double ionized ions as lled circles and from triple ionized ions as lled triangles
with the evolutionary tracks in a T log g plane we derive M M This yields a
e sdB
o
stellar radius of R R and subsequently an inclination angle of i From the
sdb
mass function a companion s mass of M M can b e derived which indicates a high
comp
mass white dwarf companion The systems total expansion b ecomes A R whereas orbit
tot
and orbital velocity of the sub dwarf are A R and v km s Corresp onding
sdB sdB
results for the white dwarf are A R and v km s Although the system s
WD WD
merging time is close to the Chandrasekhar mass limit the system will not qualify for a SN Ia
precursor due to its large merging time
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
b d rot p ost EHB no neutron star
b d rot p ost RGB no massive WD
max i no low mass WD or late type MS
most likely i no C O core WD
M M yes most massive WD
tot Chandra
M M yes most massive WD
invis Chandra
SUBDWARF B PRIMARY
In addition we also want to discuss the situation if rotation is not tidally lo cked to the orbit
Assuming a canonical sdB mass for the visible comp onent of HE and an inclination
o
angle of i we derive a minimum mass of M M for the companion Thus it
comp
is either a late type main sequence star or a low mass white dwarf The systems separation
b ecomes A R Both comp onents are at A R and A R resp ectively
tot sdB comp
Their orbital velocities are v K km s and v R
sdB comp
o
For the most probable inclination of i however the invisible companion b ecomes
M M which is most likely a C O core white dwarf The separation of the system
comp
can b e computed to b e A R In this scenario the sdB circles around the barycenter at
tot
A R v km s whereas the companion is at A R v R
sdB sdB comp comp
An overview ab out di erent scenarios for HE is given in table
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
HE was miss classi ed to b e an DA white dwarf from the Hamburg ESO sur
vey Christlieb et al and thus entered our SPY pro ject These sp ectra did not only
show evidence for a large radial velocity shift ab out of km s but furthermore reveal that
HE is an sdB star rather than a white dwarf Nevertheless in order to prob e binary
evolution theory by means of an unbiased sample the star was included into our follow up cam
paigns Observations were carried out at the WHT in January by Gijs Nelemans and at
the VLT in March by Ralf Napiwotzki
Radial velocity curve
Radial velocity measurements are based on the H line since there is no evidence for helium
in HE Figure shows the resulting p ower sp ectrum and the b est t RV curve
d d h m s
A p erio d of P a system s velocity of km s and
a semi amplitude of K km s can b e derived As can b e seen from the inset
d
of gure an alias at P cannot b e ruled out This alias however is only at
m
P s to the main p eak corresp onding to a relative error for the p erio d of ab out
The ephemeris of HE is
HJD T E
Sp ectral analysis
We did not p erform a sp ectral analysis for HE b ecause no additional high resolution
sp ectra have b een taken and thus we refer to the results of Lisker et al This study yields
T K log g and n
e He
System comp osition
By means of the mass function we can estimate the system parameters for a given scenario
o
Assuming a canonical mass of M for the sdB and an inclination of i the mass
function f M results in M M for the minimum mass of the unseen
m comp
companion Sub dwarf and companion are at A R and A R resp ectively
sdB comp
from the barycenter The system s total extension b ecomes A R The orbital velocities
tot
of the comp onent is v km s while the sdB has a velocity equal to the radial velocity
comp
semi amplitude
o
The most probable comp osition for the system results from an inclination of i The
companion s mass b ecomes M M The separation of the comp onents increases to
comp
A R whereas the orbital velocities decrease to v km s at A R and
tot sdB sdB
v km s at A R resp ectively
comp comp
SUBDWARF B PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for VLT
observations and diamonds for data provided by Gijs Nelemans Lower panel Power sp ectrum
of HE
The results for the companion s mass yields the conclusion that it is either a late type main
sequence star or a very low mass He core white dwarf Table summarizes di erent scenarios
for HE
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no low mass WD or late type MS
most likely i no low mass WD or late type MS
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
This is another ob ject miss classi ed as a white dwarf from the Hamburg ESO survey The rst
survey sp ectra taken in July already reveal the true nature of the visible comp onent in the
HE system which is an sdB star Although not a white dwarf a radial shift of ab out
km s made HE a promising ob ject for follow up observations Thus the system
was re observed at the DSAZ in February and during all runs in April May and
July
Radial velocity curve
The high resolution UVES sp ectra reveal weak He i A and He i i A lines however
no helium lines can b e seen in sp ectra taken with the TWIN instrument due to low S N ratios
Thus the radial velocity measurements were done by means of the H line core
Figure shows the resulting p ower sp ectrum of HE The outstanding p eak
d d h m s
corresp onds to a p erio d of P The b est t RV curve displayed in
the upp er panel of the gure yields a system s velocity of km s and a semi
amplitude of K km s The ephemeris of HE b ecome
HJD T E
Sp ectral analysis
Our sp ectral analysis was based on a template created from three survey sp ectra and yields
T K log g and n This is in go o d agreement with the previous study
e He
of Lisker et al except for the helium abundance where a di erence of n
He
b ecomes obvious
Although the S N ratio of the template was su cient no sp ectral features except
for the Balmer series He i A He i i A Mg i i A and probably Si iv A
were detected Nevertheless we p erformed a metal abundance analysis in order to derive upp er
limits at least Therefore the line detection limit was estimated by means of measuring fake
absorption from the noise level The average noise level in terms of equivalent width was
subsequently assumed for those lines that are predicted to b e strongest in sdB stars As a
reference the study of Edelmann was used Table summarizes the results of our
study and gure displays them with resp ect to the solar value Due to the lack of suitable
metal lines neither the microturbulence nor the pro jected rotational velocity v sin i could
rot
b e derived
System comp osition
From the semi amplitude and the p erio d we compute a mass function of f M
m
Supp osing the sdB to have a canonical mass of M and the system s inclination to b e
SUBDWARF B PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for VLT
observations and circles for data based on DSAZ observations Lower panel Power sp ectrum of
HE
Table LTE metal abundances for HE The ion is given in the rst column n
is the number of lines used for the analysis and log is the derived abundance
ion n log
C i i
N i i
O i i
Mg i i
Al i i i
Si iv
S i i i
Ar i i
Fe i i i
o
i the minimum mass for the unseen companion b ecomes M M Both com
comp
p onents are separated by A R The orbits and the orbital velocities can also b e
tot
computed For the companion we found A R and v km s while the
comp comp
corresp onding results for the sdB are A R and v K km s
sdB sdB
o
The most probable inclination of i however yields a slightly di erent system com
p osition Now we compute the companion s mass to b e M M Whereas the or
comp
CHAPTER SINGLE LINED SYSTEMS
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for HE relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles and
from double ionized ions as lled circles
bit of the sdB increases to A R v km s the orbit of the companion
sdB sdB
shrinks to A R v km s Finally the system s total expansion b ecomes
comp comp
A R for this scenario
tot
According to the results for the companions mass we conclude that the unseen star in the
HE system is most probable a white dwarf with a C O core Di erent scenarios for
HE are summarized in table
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no most massive WD
tot Chandra
M M no most massive WD
invis Chandra
SUBDWARF B PRIMARY
HE
V
History
Like the other sdB stars discussed in this chapter HE was miss classi ed from the
Hamburg ESO survey as a DA white dwarf From the UVES sp ectra taken in August
however strong helium lines b ecome obvious and thus the star was re classi ed to b e a sub dwarf
B rather than a white dwarf Despite of this HE b ecame part of our follow up
campaigns b ecause of its large radial velocity shift of ab out km s found from the survey
sp ectra The observations were carried out at the VLT and the NTT in June resp ectively in
September
Radial velocity curve
Within the high resolution UVES sp ectra a large number of He i lines can b e seen e g at
A A A AA and A The observations done at the NTT however are
restricted to the wavelength region centered on H Thus no helium lines can b e seen in the
medium resolution sp ectra taken with EMMI
Figure lower panel displays the resulting p ower sp ectrum that reveal an outstanding
d h m s
p eak at P From the b est t RV curve shown in the upp er panel of the
gure a system s velocity of km s and a semi amplitude of K km s
could b e derived The system s ephemeris b ecomes
HJD T E
Sp ectral analysis
By means of high resolution UVES sp ectra a template with an S N ratio of ab out was
created The sp ectral analysis p erformed thereof results in very accurate stellar atmospheric
parameters T K log g and n A comparison with Lisker et al
e He
yields an almost p erfect agreement b etween the two studies
The sp ectrum reveals a large number of sp ecies N i i S i i S i i i Ar i i Ar i i i Ca i i i Ti i i i
and also two lines we allo cate to Zn i i i The abundances derived for individual ions are shown
in table Figure displays them with resp ect to solar values It b ecame obvious that
the pro cess elements Ar Ca Ti and Zn are enriched similarly to the p eculiar sdB to
UVO Edelmann The microturbulence was estimated from the curve of growth
of N i i lines Thus km s was determined
In order to derive the pro jected rotational velocity we tted synthetic line pro les to our
template Concentrating on the strong N i i lines b etween A and A the b est t results
in v sin i km s rot
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities of HE as a function of orbital phase and tted sine curves Squares
are used for VLT observations and triangles for data based on NTT observations Lower panel
Power sp ectrum of HE
Table LTE metal abundances for HE The ion is given in the rst column n
is the number of lines used for the analysis and log is the derived abundance
ion n log
N i i
S i i
S i i i
Ar i i
Ar i i i
Ca i i i
Ti i i i
Fe i i i
Zn i i i
System comp osition
Assuming the sdB to have the canonical mass of M we can estimate the stars radius to
o
b e R R For b ounded rotation this yields an inclination of i
sdb
From the mass function of f M we deduce a companion s mass of M M
m comp
which is most probably a massive C O core white dwarf The systems total extend can b e
computed to b e A R Orbit and orbital velocity for the sdB b ecomes A R
tot sdB
SUBDWARF B PRIMARY
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for HE relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles from
double ionized ions as lled circles and from triple ionized ions as lled triangles
and v km s resp ectively Corresp onding results for the companion are A R
sdB comp
and v km s Although the system s total mass will exceeds the Chandrasekhar limit
comp
it will not merge within a Hubble time and therefore not qualify as a SN Ia precursor
Like for the previous systems we want to discuss the case of unbound rotation as well The
o
maximum inclination of i yields the companions minimum mass of M M
comp
which indicates either a late type main sequence star or a low mass white dwarf The sep
aration of b oth comp onents b ecomes A R and the orbits are A R and
tot sdB
A R resp ectively The orbital velocities can b e computed to b e v K km s
comp sdB
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
b d rot p ost EHB no massive white dwarf
max i no low mass WD or late type MS
most likely i no C O core WD
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
and v km s
comp
o
For the most probable scenario however we assume an inclination of i This
results in M M for the unseen companion which is a hint for a white dwarf
comp
The orbits are A R and A R resp ectively and the orbital velocities are
sdB comp
v km s and v km s Thus the system s separation b ecomes
sdB comp
A R Di erent scenarios for HE are summarized in table
tot
SUBDWARF B PRIMARY
HE
V
History
Like the previous ob jects HE was selected from the Hamburg ESO survey as a DA
white dwarf and subsequently entered our survey By means of visual insp ection of the rst
UVES sp ectra taken in August however it b ecame evident that HE is a sdB
star rather than a white dwarf Nevertheless the ob ject was included in our follow up campaign
b ecause of the radial velocity shift of ab out km s that was determined from the survey
sp ectra Therefore additional medium resolution sp ectra were taken during our August
run at the DSAZ
Radial velocity curve
d d h m s
From the p ower sp ectrum of HE we derive a p erio d of P
m
as can b e seen from the lower panel of gure The inset reveal some aliases at P
which are nearly as likely as the p erio d de ned by the main p eak However the relative error
caused by these aliases is less than and will therefore not in uence the further discussion
From the b est t RV curve displayed in the upp er panel of gure a system s velocity of
km s and a semi amplitude of K km s was derived The
ephemeris of HE b ecomes
HJD T E
Sp ectral analysis
Due to the fact that HE was observed at the VLT only in the course of the survey
our template used for the following analysis relys on two high resolution UVES sp ectra only
A quantitative sp ectral analysis done by means of the coadded high resolution sp ectra results
in the stellar parameters T K log g and n These parameters are
e He
almost consistent with the results derived by Lisker et al who measured T K
e
log g and n
He
Despite a low S N ratio of of the coadded sp ectrum N i i S i i i and Ar i i lines b ecomes
obvious For these lines abundances were derived whereas for C i i O i i Mg i i Al i i i Si i i i and
Fe i i i upp er limits were determined by means of the noise level see HE section
The results are summarized in table and displayed with resp ect to solar values in gure
The microturbulence was estimated from the curve of growth of the N i i lines to b e
km s Moreover in order to obtain the pro jected rotational velocity we tted the
observed N i i lines in the sp ectral range from A to A to synthetic pro les From the
b est t we estimate v sin i km s rot
CHAPTER SINGLE LINED SYSTEMS
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for VLT
observations and circles for data based on DSAZ observations Lower panel Power sp ectrum of
HE
Table LTE metal abundances for HE The ion is given in the rst column n
is the number of lines used for the analysis and log is the derived abundance
ion n log
O i i
N i i
O i i
Mg i i
Al i i i
Si i i i
S i i i
Ar i i
Fe i i i
System comp osition
The radius computed for a canonical mass of M is R R However for b ounded
sdb
rotation this results in sin i equation unless the star has a mass of M for
which sin i b ecomes unity or more Therefore a tidally lo cked scenario can b e ruled out for
HE
o o
Nevertheless we want to discuss the case of non b ounded rotation for i and i
SUBDWARF B PRIMARY
2
0
-2
C N O Mg Al Si S Ar Ca Ti Fe Zn
Figure LTE abundances with error bars for HE relative to solar values dashed
horizontal line Abundances derived from single ionized ions are marked as op en circles from
double ionized ions as lled circles and from triple ionized ions as lled triangles
as we did for the previous systems The mass function b ecomes f M Assuming a
m
o
canonical sdB mass of M for the sdB and i we determine the minimum mass of
the companion to b e M M Because b oth comp onents have almost the same mass
comp
their orbits and orbital velocities are quite similar The orbits are A R whereas
sdB comp
the orbital velocities b ecomes v K R The total separation of the comp onents
sdB comp
is A R
tot
o
The most probable inclination of i yields M M for the unseen compan
comp
ion Orbits and orbital velocity of the visible sdB star are A R and v km s
sdB sdB
resp ectively For the unseen comp onent we estimate A R and v km s The
comp comp
system s total extension b ecomes A R
tot
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
Based on the masses determined ab ove for the unseen companion we conclude that it is most
likely a white dwarf with a C O core Table gives an overview ab out di erent scenarios for
HE
SUBDWARF O PRIMARY
Sub dwarf O primary
Sub dwarfs of the sp ectral type O are hot stars with e ective temp eratures that ranges from
ab out K up to K at logarithmic surface gravities of log g From
their sp ectra broad Balmer lines b ecome obvious accombined by He i i absorption Mo ehler et
al The helium abundances varies b etween n Str oer Moreover
He
He rich sdO stars He sdOs show a large variety of metal lines like carb on nitrogen oxygen
neon magnesium and silicon
Some sdO stars were classi ed as DA white dwarfs from the Hamburg ESO survey Christlieb
et al and thus entered our SPY pro ject The true nature of these ob jects however b ecame
obvious by means of visual insp ection of the high resolution UVES sp ectra taken at the VLT
cf gure Nevertheless the survey sp ectra also reveal some highly radial velocity variable
ob jects among the sdO stars
In the following section the result of our analysis for three sdO stars is presented
Figure Sample UVES sp ectrum of an sub dwarf O star HE see section
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
Selected to b e a DA white dwarf from the Hamburg ESO survey Christlieb et al
HE was observed ve times in the course of our SPY pro ject However from the
sp ectra not only the Balmer series but also a strong Hei i absorption line at A b ecome
evident Thus we classi ed HE to b e either a DAO white dwarf or a sub dwarf of
type O Since a large radial velocity shift of ab out km s was detected from the survey
sp ectra HE was re observed in the course of our follow up campaigns Observation
were carried out at the ESO VLT in June and March and at the NTT in February
Radial velocity curve
From the sharp well de ned Hei i line we made very accurate radial velocity measurements Our
d h m s
analysis results in an unequivocal p erio d of The p ower sp ectrum can b e
seen in the lower panel of gure whereas the b est t RV curve is displayed in the upp er
panel This RV curve gives a system s velocity of km s and a radial velocity
semi amplitude of K km s The ephemeris of HE is
HJD T E
Sp ectral analysis
Recently a mo del t to the observed sp ectra resulted in T K log g and
e
n Str oer The visible comp onent of HE has to b e a sdO rather
He
than a DAO white dwarf b ecause of the low surface gravity
System comp osition
The mass function of f M allows a more detailed insight in the HE system
m
Assuming the sdO to have a canonical mass of M we compute a minimum mass for the
o
unseen companion of M M for an inclination of i The system s total extend
comp
b ecomes A R Sub dwarf and companion are at A R and A R
tot sdO comp
resp ectively from the barycenter Whereas the orbital velocity of the sdO is equal to the radial
o
velocity semi amplitude for i the companion s velocity b ecomes v km s
comp
o
The most probable comp osition for the system results from an inclination of i The
companion s mass b ecomes M M The separation of the two comp onents increase to
comp
A R whereas the orbital velocities decrease to v km s at A R
tot sdO sdO
and v km s at A R resp ectively
comp comp
For b oth scenarios the companion could b e either a low mass white dwarf or a late type main
sequence star Infrared photometry could help to distinguish b etween these options Therefore
SUBDWARF O PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curves Squares are used for VLT
observations and triangles for observations done at the ESO NTT Lower panel Power sp ectrum
of HE
we checked the MASS database Unfortunately the listed error margins are to o large to b e
useful see Stark Wade
Although a late type main sequence star cannot b e ruled out the system companion is
probably a low mass white dwarf with a helium core This means HE is indeed a
double degenerate system but it will not qualify as a SN Ia precursor due to its low total mass
and its merging time exceeding the Hubble time cf table Even for a M white dwarf
companion the most massive white dwarf p ossible the merging time would still b e larger
than the current age of the universe
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no low mass WD or late type MS
most likely i no low mass WD or late type MS
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
HE
V
HE was observed for SPY in July at the ESO VLT As in the former case of
HE cf section the survey sp ectra show not only narrow Balmer lines up to
H but also the He i i line at A Therefore we classi ed HE to b e a sdO star
in agreement with the most common classi cation system describ ed by Mo ehler et al
This result is in go o d agreement with an older study by Kilkenny et al classifying
HE alias PG to b e a sdB O star
The survey sp ectra reveals a radial velocity shift of km s Nevertheless b eing brighter
than mag HE quali ed for our subsample two and thus entered the follow up
program Sp ectra were taken at the DSAZ in February and April at the ESO NTT
in February and at the ESO VLT in March The set of radial velocity measurements
was complemented by data provided by Gijs Nelemans These measurements are based on WHT
observation from January and March
Radial velocity curve
d d h m s
Our radial velocity analysis results in an unambiguous p erio d of as
can b e seen from the p ower sp ectrum in the lower panel of gure The upp er panel of
the gure displays the b est t RV curve A system s velocity of km s and a
semi amplitude of K km s were derived The ephemeris of the visible comp onent
of the HE system is
HJD T E
Sp ectral analysis
A sp ectral analysis p erformed by Str oer results in T K log g and
e
n
He
System comp osition
From the system s parameter we compute the mass function to b e f M Assuming a
m
o
canonical mass of M for the sdO and a inclination of i the minimum mass of the
unseen comp onent b ecomes M M Orbit extension and velocity for the companion
comp
are A R and v km s resp ectively The corresp onding results for the sdO
comp comp
are A R for an orbital velocity equal to the radial velocity semi amplitude The total
sdO
expansion of the system b ecomes A R
tot
o
For the most probable inclination of i however we derive M M
comp
A R and v km s The results for the sdO are A R and
comp comp sdO
v km s which means that b oth comp onents are separated by A R totally
sdO tot
From the companion s masses derived ab ove we nally conclude that it is most likely a white
dwarf with a C O core As in the case of HE the system s merging time is many
times larger than the Hubble time and thus HE will not qualify as a SN Ia progenitor
cf table
SUBDWARF O PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve Squares are used for VLT
observations triangles for observations taken at the NTT and squares for observations from the
DSAZ Data provided by Gijs Nelemans are marked as diamonds Lower panel Power sp ectrum
of HE
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no C O core WD
most likely i no C O core WD
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER SINGLE LINED SYSTEMS
HE
V
History
Selected from the Hamburg ESO Survey by Norb ert Christlieb to b e a DA white dwarf Christlieb
et al HE was observed in April and May and once more in June and
in April From the sp ectral app earance however it b ecame evident that HE
is not a white dwarf but a sub dwarf of type O cf gure panels g and h The sp ectra
reveal moreover radial velocity shift of km s which b ecomes obvious from measurements
done by means of the He i i lines at A Thus follow up observations were p erformed at the
ESO NTT February at the VLT March and at the DSAZ February and
May in order to solve the system s RV curve
Radial velocity curve
d h m s
The radial velocity analysis results in an unambiguous p erio d of or The
prop er p ower sp ectrum can b e seen in the lower panel of gure whereas the upp er panel dis
plays the b est t RV curve The system s velocity and its semi amplitude are km s
and K km s resp ectively The ephemeris of the visible comp onent is
HJD T E
Sp ectral analysis
A sp ectral analysis of the high resolution UVES sp ectra results in T K log g
e
and n Str oer
He
System comp osition
From the system s parameter we compute the mass function to b e f M Assuming a
m
o
the minimum mass of the canonical mass of M for the sdO and a inclination of i
unseen comp onent b ecomes M M Orbit extension and velocity for the companion
comp
are A R and v km s resp ectively The corresp onding results for the sdO
comp comp
are A R for an orbital velocity equal to the RV semi amplitude The total extend of
sdO
the system b ecomes A R
tot
o
For the most probable inclination of i however we derive M M
comp
A R and v km s The results for the sdO are A R and
comp comp sdO
v km s which means that b oth comp onents are separated by A R totally
sdO tot
The companion is most probably a late type main sequence star Thus HE is not
a double degenerate system and will therefore not qualify as a SN Ia precursor Nevertheless a
double degenerate scenario and even a SN Ia precursor can not b e ruled out de nitely b ecause
o
for a very low inclination of i the companion s mass is M and the system s merging
time b ecomes less than a Hubble time cf table
SUBDWARF O PRIMARY
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
radial velocities as a function of orbital phase and tted sine curve Squares are used for VLT
observations triangles for observations taken at the NTT and squares for observations from the
DSAZ Lower panel Power sp ectrum of HE
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no late type MS
most likely i no late type MS
M M no massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
Chapter
Peculiar ob jects
PN G + alias EGB a central star of a planetary nebula
V
History
Not a white dwarf but the central star of the planetary nebulae PN G EGB is an
immediate precursor of a white dwarf Mendez et al derived T K log g
e
and n the central star of the nebula CSPN Figure Napiwotzki displays
He
this ob ject in a T log g diagram The dashed lines mark p ost RGB evolutionary tracks com
e
puted by Drieb e et al cf gure Therefore EGB is the result of p ost RGB
evolution as b ecomes obvious from the diagram
Over the course of three years we obtained more than sp ectra of this ob ject taken at
the ESO VLT and ESO NTT as well as at the m telescop e of the DSAZ The sample was
complemented by data taken at the WHT in Octob er and in January
Radial velocity curve
The sp ectra reveal lines from the Balmer series as well as He i at A and the He i i line at
A No hint for any other element can b e found even not in the high resolution sp ectra
taken at the VLT using UVES Thus we tted all available helium lines as well as H in order
to determine the radial velocities From the p ower sp ectrum displayed in the lower panel of
d d h m s
gure we deduce a p erio d of either There are some aliases however
that we cannot rule out by visual insp ection of the prop er RV curve The most signi cant of
d d d
these aliases are at and From the b est t RV curve displayed
in the upp er panel of gure we derive a system s velocity of km s and a
radial velocity semi amplitude of K km s The ephemeris of EGB is
HJD T E
Figure T log g diagram of hydrogen rich DA and DAO white dwarfs squares and
e
hydrogen po or ob jects triangles like PG stars DO white dwarfs and O He p ost AGB
stars CSPNe are encircled Also plotted are evolutionary tracks of p ost AGB stars solid lines
p ost RGB stars dashed lines and p ost early AGB stars dotted lines The gure is taken from
Napiwotzki
System comp osition
From gure we can estimate the mass of the sdB binary to b e ab out M The
o
mass function of f M and an inclination of i yield the minimum mass of
m
the companion of M M The system s total extend computed by means of equa
comp
tion b ecomes A R Thus the planetary nebula s central star is at A R
tot CSPN
v K km s whereas the unseen companion is at A R
CSPN comp
v km s
comp
o
The most probable inclination of i however gives M M and A R
comp tot
Orbit and orbital velocity of the visible companion b ecome A R and v km s
CSPN CSPN
resp ectively Corresp onding results for the companion are A R and v km s
comp comp
We nally conclude that the companion of the central star is most probably a brown dwarf
o
If the system s inclination is b elow however also a low mass M late type main
sequence star would b e p ossible The system s comp osition is very similar to AA Dor a well
known sdOB plus brown dwarf system cf Rauch Rauch Werner but the p erio d
is much larger Table summarizes di erent scenarios for EGB
CHAPTER PECULIAR OBJECTS
Figure Best t RV curve and p ower sp ectrum for PN G or EGB Upp er panel
Measured RVs of PN G alias EGB as a function of orbital phase and tted sine
curve VLT observations are marked as squares circles indicate m DSAZ data and triangles
are used for data taken at the NTT Diamonds represent RV data provided by Gijs Nelemans
The observations were done at the WHT Lower panel Power sp ectrum of PN G The
inset shows the region near the main p eak in more detail
Table Scenarios for PN G alias EGB
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
WD + a white dwarf with a p eculiar H pro le
V
History
Because this ob ject is classi ed as a DA white dwarf in the actual version of the Mc Co ok
Sion Catalogue WD entered our SPY survey Four high resolution sp ectra
were taken at the VLT revealing a radial velocity shift of up to km s Thus WD
was re observed during our DSAZ runs in August and Octob er and in April and May
The data set was complemented by four sp ectra taken at the NTT in September
Variable H line pro le
th
Figure displays ve medium resolution TWIN sp ectra taken in August and their
b est mathematical t It seems that the shap e of the H line pro le is temp orally variable
Up to now it is not clear if this time dep endent sp ectral app earance is due to observational
constrains or based on a real physical pro cess Although an unresolved double line d system can
explain at line pro les like the top and b ottom ones of gure this scenario can b e ruled
out b ecause even the high resolution UVES sp ectra show no hint for a companion except the RV shift
0.497177
0..439934
0.396203
0.371278
0.339352
6540 6560 6580
Figure Observed H line pro les of WD and their b est mathematical t results
All displayed sp ectra were taken in the night th to th of August at the DSAZ m
telescop e equipp ed with the TWIN instrument The HJDs can b e seen on the left
hand
Radial velocity curve
If we include all available RV measurements into our analysis we derive a p ower sp ectrum that
d h m s
yield to This result however is not very convincing b ecause of the large
number of aliases in the p erio dogram Moreover a large number of observed data p oints match
CHAPTER PECULIAR OBJECTS
the b est t RV curve insu ciently Therefore we checked the quality of the mathematical t
used for the radial velocity determination and classi ed all sp ectra by this means This was
done by a visual insp ection In order to improve our analysis we discarded subsequently all data
p oints classi ed to b e insu cient This however includes also radial velocity measurements
based on p eculiar at H line pro les
The resulting p ower sp ectrum and RV curve are much more unambiguously than the previous
ones computed for all data p oints Moreover no outliers remains within the RV curve The
d h m s
system s most probably p erio d is or as can b e seen from the lower panel
of gure From the b est t RV curve upp er panel of gure we estimate the system s
velocity to b e km s and the semi amplitude to b e K km s The
ephemeris b ecomes
HJD T E
Figure Best t RV curve and p ower sp ectrum for WD Upp er panel Measured
RVs as a function of orbital phase and tted sine curves Squares are used for VLT observations
and circles for data based on DSAZ observations The triangle however marks an observation
done at the NTT Lower part Power sp ectrum of WD
Sp ectral analysis
Detlev Ko ester priv com already analyzed WD in order to obtain stellar atmo
spheric parameters His study results in T K and log g Therefore we estimate
e
a mass of M M for the white dwarf from the Bl ocker tracks cf gure
WD
System comp osition
o
The mass function of f M results for an inclination of i in the companion s
m
o
minimum mass of M M Its most probable mass is M M for i
comp comp
The orbits are quite similar for b oth scenarios A R and A R resp ectively
WD comp
The system s total separation b ecome A R Orbital velocities of the white dwarf
tot
o o
are v K km s i and v km s i resp ectively The
WD WD
companion s velocity is nearly constant for b oth scenarios at v km s
comp
Due to the masses found ab ove for the unseen companion we conclude that it is most
probable a brown dwarf Table gives an overview ab out the scenarios discussed ab ove
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M yes massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
Chapter
Stars with small RV amplitudes
Although it is unlikely to nd a merger candidate among stars that show small radial velocity
amplitudes only we nevertheless re observed these ob jects as backup targets in the course of
our follow up campaigns In addition they are useful in order to build an unbiased sample of
double degenerates for probing binary evolution theory e g Nelemans et al For this
purp ose it is imp ortant to avoid observational biases i e by selecting against long p erio ds and
low amplitude systems Our observations are a the rst step toward this sample
The analyses of ob jects with small radial velocity amplitudes is di cult and not straight
forward as it has b een shown for WD section The following section will
concentrate on these systems Therefore we will discuss HE as an example in more
detail whereas ob jects with similar characteristics are shown in an overview
HE +
V
History
Being classi ed as a white dwarf from the Hamburg ESO survey HE entered our
SPY pro ject and was observed at the VLT in September and at the VLT Although
the rst UVES sp ectrum was nearly useless due to a very low S N ratio it b ecame obvious
that the ob ject was miss classi ed since HE is not a white dwarf but a sdB star
Regardless of the bad quality of the rst UVES sp ectra a radial velocity shift of ab out km s
was determined by a rst analysis of the sp ectra Therefore HE was re observed
during our July run at the DSAZ The routine used to determine the radial velocity shifts
from the UVES sp ectra however was improved recently and thus the former shift was corrected
to km s
Radial velocity curve
In the course of our analysis we found that the determination of the system parameters was
not simply straight forward Some of the follow up observations yield sp ectra of very low S N
ratios This results nally in large errors of the radial velocity determination The resulting
p ower sp ectrum and RV curve are therefore equivocal with a large number of aliases and outliers
resp ectively We found that the semi amplitude is in the same order of magnitude than some
of the error margins Thus we excluded all data p oints with error limits as large or even larger
than the semi amplitude Data p oints lying clearly outside the radial velocity range covered by
the RV curve were discarded as fake measurements as well
Figure shows the b est results for HE derived this way For these p ower
sp ectrum and RV curve however we had to discard data p oints totally As can b e seen from
the lower panel of the gure even this is not an unambiguous result b ecause some aliases still
remain We can also not rule out that HE is not variable at all A straight line
instead of a time dep endent sine function ts the observed data p oints also very well as can b e
seen from the dashed line in the upp er panel of gure Results of all RV measurements
even the discarded ones are given in table C Although the variability is not given for sure
d h m s
the b est result derived by the metho d describ ed ab ove yield a p erio d of or
The system parameters obtained by means of the b est tting RV curve are km s
and K km s The most probable ephemeris is
HJD T E
Sp ectral analysis
By means of a quantitative sp ectroscopic analysis of the second high resolution sp ectrum taken
at the VLT Lisker et al determined the stellar parameters of the visible comp onent to
b e T K log g and n
e He
CHAPTER STARS WITH SMALL RV AMPLITUDES
Figure Best t RV curve and p ower sp ectrum for HE Upp er panel Measured
RVs as a function of orbital phase and tted sine curve solid line Squares are used for VLT
observations and circles for observations done at the DSAZ The dashed line marks the average
radial velocity of km s Note that this velocity di ers slightly from the system s velocity
derived by means of a sine t Lower panel Power sp ectrum of HE The inset
displays the region around the main p eak in more detail
System comp osition
Assuming a canonical sdB mass of M the mass function of f M yields
m
o
a companion s minimum mass of M M for an inclination of i The most
comp
o
probable mass for i is the same due to the small mass functions Deviations are placed
at the third decimal Thus a probable companion if it really exist is most likely a brown dwarf
Table Scenarios for HE
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M M yes massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
Thus the sdB is lo cated almost in the system s barycenter whereas the companion circle at an
orbit of A R Its orbital velocity is v km s
tot comp
Except for the unlikely case of p ole on view this system do es de nitely not qualify as a SN Ia
progenitor The maximum inclination for a M neutron star companion for example is
o
b elow Table summarizes the scenarios discussed ab ove for a probable binary
CHAPTER STARS WITH SMALL RV AMPLITUDES
WD +
V
History
Being classi ed as a DA white dwarf in the Mc Co ok Sion Catalogue WD
entered our SPY pro ject and was observed twice at the VLT with UVES In addition four
medium resolution sp ectra were taken at the NTT during our January run
Radial velocity curve
d
The p ower sp ectrum derived from these six data p oints formally results in a p erio d of or
h m s
The b est t RV curve gives a system s velocity of km s and a very
small semi amplitude of only K km s Its ephemeris b ecomes
HJD T E
However within the error margins almost all individual RV measurements are consistent with
each other as can b e seen from gure Therefore it is p ossible that WD is not
radial velocity variable
Sp ectral analysis and system comp osition
Detlev Ko ester determined T K and logg for WD Thus we estimate
e
a mass of M M by means of the B ocker tracks Table summarizes the most
WD
o o
imp ortant characteristics for WD assuming i i the minimum mass
main sequence star companion and the most massive white dwarf companion
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M no massive C O core WD
tot Chandra
M M no most massive WD
invis Chandra
WD
V
History
Also selected from the Mc Co ok Sion Catalogue the UVES survey sp ectra reveal a
radial velocity shift of ab out km s only Nevertheless the initial VLT observations were
complemented over the course of two years by additional observations taken at the NTT the
DSAZ and at the WHT
Radial velocity curve
However we found no unambiguous p erio d from the resulting p ower sp ectrum and therefore a
d
non variable scenario can not b e ruled out for WD The most probable p erio d is
h m s
or From the b est t RV curve given in gure we estimate a semi amplitude of
K km s and a system s velocity of km s The ephemeris b ecomes
HJD T E
However six data p oints with error margin up to three times larger than the derived semi
amplitude had b een discarded
Sp ectral analysis and system comp osition
Detlev Ko ester determined T K and logg Because of its high surface gravity
e
however the star lies outside the range of the tracks by Bl ocker at el cf gure
Thus we have to extrap olate in order to derive the mass of the white dwarf M M
WD
o
Table summarizes the most imp ortant characteristics for WD assuming i
o
i the minimum mass main sequence star companion and the most massive white dwarf
companion
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M no low mass WD
tot Chandra
M M no most massive WD
invis Chandra
CHAPTER STARS WITH SMALL RV AMPLITUDES
WD
V
History
Maxted et al a already found this system to b e RV variable Due to the fact that no
p erio d was known so far we to ok additional sp ectra in order to solve the system parameters
Two observations were done in the course of SPY at the VLT one at the NTT in February
and two more during our March run at the WHT
Radial velocity curve
The additional radial velocity measurements based on these observations did not result in
an unambiguous solution for the p erio d and the RV curve Figure displays the b est t
d h m s
RV curve derived for the most probable p erio d of A semi amplitude of
K km s and a system s velocity of km s can b e derived The ephemeris
is
HJD T E
The RV variability found by Maxted et al a seems to dep end heavily on the data p oint
at km s After discarding it however the radial velocities can also b e tted very well
by a straight line Thus the systems variability is not for sure at all
Sp ectral analysis and system comp osition
The sp ectra analysis p erformed by Detlev Ko ester yields T K and logg Thus
e
we determine the white dwarf s mass to b e M M Table summarizes the most
WD
o o
imp ortant characteristics for WD assuming i i the minimum mass
main sequence star companion and the most massive white dwarf companion
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M yes massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
WD +
V
History
For b eing classi ed as a DA white dwarf in the Mc Co ok Sion Catalogue WD
entered our survey and was observed twice at the VLT The maximum radial velocity shift found
was only ab out km s Nevertheless follow up observations were done at the DSAZ in April
and May
Radial velocity curve
d h m s
We found a p erio d of The b est t RV curve displayed in gure gives a
semi amplitude of K km s only Therefore we cannot rule out de nitely that the
system is not variable at all The system s velocity we derived by our analysis is km s
The ephemeris is
HJD T E
Sp ectral analysis and system comp osition
Detlev Ko ester derived T K and logg Thus we estimate the white dwarf s
e
mass to b e M M Table summarizes the most imp ortant characteristics for
WD
o o
WD assuming i i the minimum mass main sequence star companion
and the most massive white dwarf companion
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M yes massive C O core WD
tot Chandra
M M yes most massive WD
invis Chandra
CHAPTER STARS WITH SMALL RV AMPLITUDES
WD +
V
History
WD is classi ed as DA white dwarf in the actual version of the Mc Co ok Sion
Catalogue and thus entered our survey From the survey sp ectra taken at the VLT
a small radial velocity shift of ab out km s b ecame obvious Nevertheless the ob ject was
re observed as a backup ob ject during our follow up runs at the DSAZ in
Radial velocity curve
The p ower sp ectrum of WD reveals a large number of aliases that are almost as likely
d d h m s
than the main p eak at or Moreover these aliases are not concentrated but
rather scattered over the full range covered by the p erio dogram Thus the resulting p erio d is
equivocal and a non variable scenario cannot b e ruled out either Nevertheless the most likely
ephemeris
HJD T E
The b est t RV curve can b e seen from gure This yields a system s velocity of
km s and a semi amplitude of K km s Although the RV
curve lo oks not so bad seven RV measurements have b een discarded b ecause of large error
margins
Sp ectral analysis and system comp osition
By means of a quantitative sp ectral analysis Detlev Ko ester determined T K and
e
logg Thus we estimate the white dwarf s mass to b e M M Table sum
WD
o o
marizes the most imp ortant characteristics WD for assuming i i the
minimum mass main sequence star companion and the most massive white dwarf companion
Table Scenarios for WD
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no brown dwarf
most likely i no brown dwarf
M M no minimum mass MS
invis
M M no C O core WD
tot Chandra
M M no most massive WD
invis Chandra
HS +
V
History
One of the few ob jects selected from the Hamburg Quasar survey HS reveals a radial
velocity shift of ab out km s from the survey sp ectra For a de nite test of the apparent
variability we re observed this ob jects during our July run at the DSAZ
Radial velocity curve
Some sp ectra have a very low S N ratio Thus we classi ed all data bye eye prior to our
analysis and discarded those radial velocity measurements based on insu cient sp ectra The
d h m s
resulting p ower sp ectrum yields a most probable p erio d of or This result is
based on an outstanding p eak of the p ower sp ectrum No striking alias b ecomes evident from the
p erio dogram However sp ectra were discarded for the reasons mentioned ab ove and thus the
analysis is based on the remaining ones only From the b est t RV curve gure we derive
a system s velocity of km s and a semi amplitude of K km s
HJD T E
If we try to improve our rst bye eye classi cation go o d and p o or quality by means of
a third in b etween category we found that the results change dramatically Therefore we
conclude that we need more observations with a go o d S N level in order to test the systems
variability de nitely
Sp ectral analysis and system comp osition
The stellar parameters derived by Ko ester priv com are T K and logg This
e
yields a dwarf s mass of M M from the Bl ocker tracks cf gure Table
WD
o o
summarizes the most imp ortant characteristics HS for assuming i i
the minimum mass main sequence star companion and the most massive white dwarf companion
Table Scenarios for HS
i M M M t SN Ia
vis invis tot m
Scenario deg M M M Gyrs prec nature of companion
max i no low mass MS star
most likely i no low mass MS
M M yes low mass WD
tot Chandra
M M yes most massive WD
invis Chandra
CHAPTER STARS WITH SMALL RV AMPLITUDES
Phase
Figure Best t RV curves of WD WD WD WD
WD and HS Squares are used for VLT observations triangles for NTT
observations and circles for observations done at the m telescop e of the DSAZ Diamonds are
either data p oints provided by Gijs Nelemans priv com or taken from Maxted et al b
The dashed lines mark the average radial velocities Note that these di er slightly from the
system s velocities derived by means of a sine t
Chapter
Conclusion and outlo ok
Conclusion
In this thesis results are presented from follow up campaigns of radial velocity variable white
dwarfs and hot sub dwarfs discovered in the course of the SPY Sup ernova Ia Progenitor SurveY
pro ject
Sup ernovae of type Ia SN Ia play an outstanding role for our understanding of galactic
evolution and the determination of the extragalactic distance scale In one of two p ossible
scenarios SNe Ia result from the merging of two white dwarfs double degenerates in a close
binary system via gravitational wave radiation According to this scenario a system quali es
as a SN Ia precursor if its comp onents will merge in less than the Hubble time T T and
M H
their total mass exceeds a critical mass limit which is most likely the Chandrasekhar mass at
M M However theoretical mo del calculations predict that only white
Chandra
dwarf systems ful ll these criteria see e g Ib en et al Nelemans et al
The intention of SPY is to fortify the double degenerate or merger scenario for the evolution
of SN Ia events by observational means Thus more than white dwarfs were observed for
SPY at the VLT Very Large Telescop e and degenerate stars were identi ed to b e radial
velocity RV variable SPY is the largest pro ject searching for radial velocity variations among
white dwarfs The number of examined systems was quintupled and the number of known
double degenerates increased by a factor of seven Most imp ortant are so called double line d
systems already revealing evidence for a companion in their sp ectra Only six of these ob jects
were known b efore SPY started
In parallel with the survey follow up sp ectroscopic observations have b een started to obtain
radial velocity curves from which p erio ds P and RV semi amplitude K have b een determined
Both parameters are crucial for the determination of the merging time t and the total mass
m
M of a binary system which are the criteria to decide whether or not the system quali es as
tot
a SN Ia precursor First results were presented by Napiwotzki HE Napiwotzki et
al b and HE Napiwotzki et al In this thesis we discussed additional
radial velocity variable systems found by SPY In order to derive their radial velocity curves
these ob jects were re observed over the course of three years with to m telescop es
Six of the observed ob jects are double lined systems Thus radial velocity analyses provide
information not only ab out the motion of the primary but also of the secondary This results
in the mass ratio M M In addition we made use of the di erent gravitational redshift
prim sec
of the comp onents to determine the masses of the comp onents of four double lined binaries
CHAPTER CONCLUSION AND OUTLOOK
Figure System parameters p erio d P and masses M of the systems discussed in the
tot
chapters except for WD see text and The masses of the unseen companions of
the single lined systems op en symbols are estimated from the mass function for the exp ected
o
average inclination angle of i In addition KPD Maxted et al is
marked as an op en diamond HE and HE Napiwotzki et al b and
resp ectively two binaries discovered in the course of SPY are marked as well a newly
discovered SN Ia precursor candidate Napiwotzki et al a The shaded area in the upp er
left part of the plot marks the region where SN Ia precursors are exp ected
as well as their merging time The remaining systems were in need of additional sp ectral
analysis by means of a mo del t However this is usually complicated for double lined ob jects
b ecause their observed sp ectra are always a sup erp osition of two individual white dwarf sp ectra
Nevertheless a mo del t was p ossible for HE b ecause b oth comp onents are quite
similar Therefore the number of free t parameters could b e reduced For the last system
WD this turned out to b e imp ossible to do with the sp ectral analysis to ols at hand
In addition to the double line d systems we analyzed a sample of single lined ob jects
classi ed as radial velocity variable from SPY For we derived very accurate orbital solutions
whereas for the remaining systems the solutions are still ambiguous and more observations are
needed to solve them de nitely
Since these systems are single lined radial velocity analyses reveal less information than for
double lined systems Only the mass function could b e derived which is an equation of three
unknown parameters The masses M and M of the two comp onents and the system s
vis invis
inclination angle i In order to set further constrains to this equation we have to determine
the masses of the visible primary comp onents For the white dwarf primaries this can b e
done with the aid of the mass radius relation and a quantitative sp ectral analysis For sub dwarf
sd B and O primaries we assumed a canonical mass of M predicted by evolutionary
CONCLUSION
o
Figure The same as in gure but for an inclination of i for the single line d
systems
calculation Subsequently we derived the most likely system comp ositions out of the mass
o
functions and the statistically most probable inclination i Figure displays our
results for the double lined systems from section as well as for the white dwarf primaries from
section the sdB primaries from section and the sdO primaries from section The
shaded area in the upp er left corner marks the region where SN Ia progenitors are exp ected
according to the merger scenario Whereas the total masses for the double lined systems are
determined lled symbols results for the single line systems op en symbols are not due to
the uncertainty of the inclination At least the minimum masses of the unseen companions
o
and thus the minimum masses of the systems themselves can b e derived if we assume i
gure The displayed symbols are therefore lower limits only
We want to highlight WD a binary system comp osed of two DA white dwarfs
merging in less than a Hubble time Since the total mass of the system is sup er Chandrasekhar
within the error margins the system will most likely explo de in an SN Ia event Thus it is
one of the most promising SN Ia progenitor candidates known so far b esides a further binary
discovered in the course of SPY Napiwotzki et al a and KPD Maxted et
al c The latter system however is a sub dwarf plus white dwarf binary rather than a
real double degenerate and thus some doubts have b een raised if it really quali es as a SN Ia
precursor see Ergma et al For WD the error margin is still somewhat large
and thus more observations are necessary in order to improve our result
As can b e seen from the gure and three SN Ia precursor candidates are known
so far and two additional systems HE and HE are only b elow
the limit Therefore the relevance of the merger scenario is strongly enforced all the more if
one considers that only ab out of the radial velocity variable ob jects found by SPY have
CHAPTER CONCLUSION AND OUTLOOK
b een re observed and investigated as probable SNe Ia precursors up to now Mo del calculations
have to examine whether or not the fraction of sup er Chandrasekhar SN Ia events pro duced via
the merger channel is signi cant enough to challenge the role of SNe Ia as standard candles
for cosmological distance determination
Simulations of stellar p opulations forming double degenerates need several input parameters
which describ e the formation of stars stellar evolution and the outcome of close binary interac
tions One rather uncertain ingredient for the formation of double degenerates is the parameter
governing the outcome of the common envelope CE evolution This parameter is de ned
CE
as the fraction of orbital energy released in the orbital contraction which is used to eject the
envelope Since this parameter is only weakly constrained by present theoretical calculations
it is essential to calibrate empirically A precise knowledge of is essential for our un
CE CE
derstanding of the results of common envelope evolution namely cataclysmic binaries very low
mass and very high mass white dwarfs and close binary central stars of planetary nebulae Once
the analysis of all double degenerates found by SPY is completed our binary sample will provide
an ideal lab oratory for this task
Outlo ok
Our analysis for HE HE HE and HE did not allow
us to prove that the invisible companion is a white dwarf A late type main sequence star
cannot b e ruled out This uncertainty could b e solved by searching for p erio dic light variations
b ecause main sequence companions may give rise to re ection e ects whereas white dwarfs are
to o small to show a measurable re ection see e g Heb er et al and Maxted et al c
d
This metho d however will only work for close binaries P . at intermediate inclination
o
angles i & Thus HE is probably not suited for such an analysis b ecause of
its large p erio d Nevertheless photometric monitoring of the remaining ob jects is urgently need
to rule out main sequence companions
Since the rotation of close binary systems can b e assumed to b e tidally lo cked to the or
bital motion we investigated the pro jected rotational velocities for six sdB stars This o ered
the p ossibility to set constraints to the systems inclination angles and therefore via the mass
function to the masses of the unseen companions However it had turned out that only for
HE M yields a reasonable value For WD and HE this could
invis
b e explained by their rather large p erio ds and separations Their rotation is most likely decou
pled from their orbit therefore HE HE and HE however
d d
cannot b e explained this way since their p erio ds are in b etween P Thus
we nally conclude that our determinations of the pro jected rotational velocities are insu cient
b ecause the resolution of our sp ectra was not known for certain Observations with optimum sp ectral resolution are needed to improve this study
App endix A
Selected DB candidates from the
Hamburg ESO Survey
Table A Selected DB candidates from the HES The Name is given in the rst column
whereas the co ordinates are displayed in column two and three Columns four to seven list
the luminosity in the V band and the B V U B and the Str omgren c colour indices
Comments are given in the last column
Name V B V U B c comments
HE
HE
HE
HE
HE
HE
HE Friedrich et al
HE Sion et al
HE
HE
HE
HE Sion et al
HE
HE
HE
HE Friedrich et al
HE
HE
HE
HE Sion et al
HE
HE
APPENDIX A SELECTED DB CANDIDATES FROM THE HAMBURG ESO SURVEY
Name V B V U B c comments
HE
HE
HE
HE
HE
HE
HE
HE
HE
HE
HE
HE
HE Friedrich et al
HE
HE
HE Friedrich et al
HE
HE
HE Friedrich et al
HE
HE
HE
HE
HE
HE
HE
HE
HE
App endix B
Follow up campaigns
Table gives an overview ab out all follow up campaigns p erformed in the course of our SPY
pro ject so far In addition the telescop es instruments setups and techniques used in order to
obtain and reduce the data are describ ed b elow in more detail
B Campaigns at the DSAZ m telescop e
We p erformed follow up runs totally at the Deutsch Spanisches Astronomisches Zentrum
DASZ at Calar Alto Spain Over the course of nights medium resolution sp ectra were
taken with the TWIN sp ectrograph at the the m telescop e Since this instrument allows to
observe in two wavelength ranges a dicroic mirror divided the incomming light at A For
the blue arm we used grating T while T was used in the red Both gratings have
lines mm resulting in a disp ersion of A pix The wavelength coverage range from A
to A in the blue at a medium resolution of A In the red part the wavelength scale
ranges from A to A at a mean resolution of A
The data were reduced with a semi automated program based on a set of standard routines
for long slit sp ectra from the ESO MIDAS package A two dimensional bias correction was
done for the ob ject data as well as for the dome ats After at eld correction the sp ectra
were extracted from the two dimensional images with resp ect to the sky background Finally we
p erformed the wavelength calibration using ThAr calibration sp ectra taken at the same p osition
of the telescop e and extracted at the same pixel space on the CCD as the ob ject sp ectra For
the purp ose of normalization we observed at least one DC white dwarf each night Since these
sp etral type do es not show any lines in the observed wavelength range normalization could b e
done easily for the DC and thereafter assigned to the sp ectra of the remaining program stars
The exp osure time typically range from minutes for bright stars V mag up to
minutes or more for fainter ob jects V mag Ob jects with outstanding sp ectral app earance
like double lined DA s are given a longer exp osure while stars showing strong features i e
emission lines are less exp osed
B Campaigns at the ESO NTT
Over the course of three runs we had a total of nights of observation at the New Technnology
Telescope NTT with the ESO Multi Mo de Instrument EMMI Long slit medium resolution
APPENDIX B FOLLOW UP CAMPAIGNS
sp ectra were taken in REMD and BLMD mo de resp ectively
We used two di erent setup In the red grating No covers a wavelength range from A
to A at a disp ersion of A pix and at a resolution of R Two MIT LL CCDs of
x pixel totally and m pixel size were used as detectors This setup was used for
ob jects who ose most interesting feature is the H line i e DA white dwarfs or white dwarfs
with M dwarf companions
A second setup optimized for the blue region used grating No centered on A The
wavelength scale ranges from A to A at a disp ersion of A pix and a resolution
of R The detector was also a MIT LL CCD but of x pixel only This setup
was used for observing H as well as the He i and He i i lines at A and A resp ectively
i e for sdB and sdO stars and for DB and DO white dwarfs
Exp osure times varies b etween and minutes dep ending on luminosity and sp ectral
app earance of the target For data reduction purp oses the ESO MIDAS program already used
for DSAZ m data was mo di ed to match NTT conditiones For more details on the data
reduction we therefore refer to section B
B Campaigns at the ESO VLT
Between and we carried out a total of follow up runs for nights at Paranal
observatory Like for the survey observations UVES at the UT Kueyen of the Very Large
Telescope VLT was used to obtain high resolution sp ectra of adequate S N The instrumental
setup was nearly the same as for SPY except for a smaller slit width A to slit
optimized for seeing conditions was used instead of the larger slit used in the survey For
more details on the sp ectrograph and the data reduction we refer to section
B Campaigns at the WHT and INT
In Octob er we had nights at the m Isaac Newton Telescope INT equipp ed with the
Intermediate Disp ersion Sp ectrograph IDS We used the mm camera and R R grating
for a disp ersion of A pix from A to A at a resolution of A The sp ectra were
reduced using the overscan bias level and a tungsten at eld with standard IRAF tasks using
optimal extraction Horne To p erform the wavelength calibration CuAr and CuNe were
taken at the same p osition of the telescop e and extracted at the same place on the CCD as the
ob ject sp ectra This typically gives a rms scatter less than A
In the course of we also observed nights at the m William Herschel Telescope
WHT equipp ed with the double armed medium resolution sp ectrograph ISIS We used the
instrument with a A dichroic and a slit In the red channel grating R R was
centered on A giving the sp ectra a mean resolution of A The disp ersion on the
MARCONI CCD k k m was A pix In the blue arm grating R B was
centered on A yielding to a mean resolution of A and a disp ersion of A pix The
detector was a blue sensitive EEV CCD of k k m The ISIS data reduction was p erformed in the same way as for the IDS data
App endix C
Radial velocity measurements
Table C Radial velocity measurements for all program stars
Run mid HJD RV RV
primary secondary
km s km s
WD
b
SPY
NTT
a
a
a
b
b
VLT
WD
SPY
VLT
WHT
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
DSAZ
NTT
DSAZ
VLT
WHT
HE
SPY
VLT conjunction
conjunction
Run mid HJD RV RV
primary secondary
km s km s
NTT conjunction
conjunction
conjunction
VLT conjunction
conjunction
conjunction
conjunction
conjunction
WD
SPY
VLT
NTT
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
a
VLT
WD
SPY
hardly visible
NTT hardly visible
hardly visible
b
hardly visible
hardly visible
b
b
hardly visible
VLT
WD
SPY
conjunction
NTT
VLT
conjunction
Run mid HJD RV RV
primary secondary
km s km s
conjunction
HE
SPY
VLT
VLT
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
VLT
WD
SPY
NTT
PN alias EGB
SPY
WHT
DSAZ
Run mid HJD RV RV
primary secondary
km s km s
WHT
c
c
VLT
NTT
HE
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
SPY
WHT
NTT
VLT
WD
SPY
NTT
WD
SPY
Run mid HJD RV RV
primary secondary
km s km s
WHT
c
c
c
c
NTT
c
DSAZ
c
WHT
NTT
HE
SPY
VLT
NTT
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
VLT
NTT
HE
SPY
DSAZ
a
WHT
a
a
a
NTT
VLT
Run mid HJD RV RV
primary secondary
km s km s
WHT
DSAZ
a
DSAZ
WD
SPY
DSAZ
DSAZ
Maxted et al
WD
SPY
NTT
WHT
Maxted et al a
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
HE
SPY
DSAZ
NTT
VLT
DSAZ
HS
SPY
DSAZ
Run mid HJD RV RV
primary secondary
km s km s
HE
SPY
DSAZ
DSAZ
WHT
HE
SPY
WHT
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
VLT
WHT
HE
SPY
DSAZ
WHT
VLT
WHT
DSAZ
DSAZ
DSAZ
WD
SPY
Run mid HJD RV RV
primary secondary
km s km s
DSAZ
DSAZ
c
c
c
HS
SPY
DSAZ
VLT
WD
SPY
c
DSAZ
c
DSAZ
c
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
DSAZ
c
c
c
c
WD
SPY
DSAZ
DSAZ
DSAZ
Run mid HJD RV RV
primary secondary
km s km s
WD
d
SPY
d
DSAZ
d
d
d
d
d
NTT
d
d
d
DSAZ
DSAZ
d
d
d
d
DSAZ
d
d
d
d
d
d
d
HS
SPY
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
DSAZ
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
d
HE
SPY
VLT
Run mid HJD RV RV
primary secondary
km s km s
NTT
HE
SPY
DSAZ
HE
SPY
DSAZ
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
HE
SPY
DSAZ
a
a
HE
SPY
DSAZ
c
c
c
c
Run mid HJD RV RV
primary secondary
km s km s
a
c
a
a
c
c
c
c
c
a
HE
SPY
DSAZ
APPENDIX C RADIAL VELOCITY MEASUREMENTS
Run mid HJD RV RV
primary secondary
km s km s
VLT
WHT
conjunction Only one comp onent is visible in the doublelined sp ectra We therefore assume that this observation was b e
done at conjunction The measured RV were used for b oth comp onents
a This data p oint lies clearly outside the b est t RV curve and was therefore discarded as an outlier from the nal analysis
b The sp ectra has a low S N ratio which makes the comp onent hardly visible Thus the data p oint was discarded from
the nal analysis
c The measured RV errors are larger than the RV semi amplitude We therefore discarded this data p oint from the nal
analysis
d Visual insp ection of the MIDAS t reveals a p o or t quality We therefore discarded this data p oint from the nal analysis
App endix D
Equivalent widths measurements
Table D Equivalent width measurements for sdB stars WD
HE HE HE HE
HE
Ion W mA
Mult A log g f
C i i
C i i i
N i i
APPENDIX D EQUIVALENT WIDTHS MEASUREMENTS
Ion W mA
Mult A log g f
Ion W mA
Mult A log g f
O i i
APPENDIX D EQUIVALENT WIDTHS MEASUREMENTS
Ion W mA
Mult A log g f
O i i i
Ne i i
Mg i i
Al i i i
Ion W mA
Mult A log g f
Si i i
Si i i i
S i i
S i i i
APPENDIX D EQUIVALENT WIDTHS MEASUREMENTS
Ion W mA
Mult A log g f
Ar i i
Ion W mA
Mult A log g f
Ar i i i
Ca i i i
Ti i i i
APPENDIX D EQUIVALENT WIDTHS MEASUREMENTS
Ion W mA
Mult A log g f
Fe i i i
Ion W mA
Mult A log g f
APPENDIX D EQUIVALENT WIDTHS MEASUREMENTS
Ion W mA
Mult A log g f
Zn i i i
App endix E
List of abbreviations
Table E Abbreviations
Solar
Systemic velocity
Wavelength
Microturbulent velocity
Degree
Arc Minute
Arc Second
A Angstr om
A Distance from barycenter
A System s total separation
tot
AGB Asymptotic Giant Branch
AGN Active Galactic Nucleus
Al Aluminum
Apr April
Ar Argon
Aug August
b d rot Bounded rotation
BH Black hole
C Carb on
Ca Calcium
CA Calar Alto
CCD Charged Coupled Device
CE Common Envelope
CNO Carb on Nitrogen Oxygen
comp Companion
CSPN Central star of a planetary nebula
CV Cataclysmic Variable
d
Day
DA Hydrogen rich white dwarf Degenerate type A
APPENDIX E LIST OF ABBREVIATIONS
DAO Very hot hydrogen rich white dwarf Degenerate type AO
DAZ Co ol hydrogen rich white dwarf Degenerate type AZ
DB Helium rich white dwarf Degenerate type B
DBA Helium rich white dwarf with traces of hydrogen Degenerate type BA
DC Co ol white dwarf wit no sp ectral features Degenerate type C
Dec December
deg Degree
DFG Deutsche Forschungsgemeinschaft
dl Double lined system
dM M dwarf
DO Very hot helium rich white dwarf Degenerate type O
DSAZ Deutsch Spanisches Astronomisches Zentrum
EC Edinburgh Cape
EHB Extended Extreme Horizontal Branch
EMMI ESO Multi Mo de Instrument
ESO Europ ean Southern Observatory
f Mass Function
m
Fe Iron Ferrum
Feb February
FWHM Full Width Half Maximum
g Gravity
G Gravitational Constant G dyne cm g
Gyrs Giga years
h
Hour
H Hydrogen
He Helium
HE Hamburg ESO
HeMS Helium Main Sequen ce
He sdB Helium rich sub dwarf B
He sdO Helium rich sub dwarf O
HES Hamburg ESO Survey
HJD Helio centric corrected Julian date
HQS Hamburg Quasar Survey
HR Hertzsprung Russell
HRD Hertzsprung Russell Diagram
HS Hamburg Schmidt
i Inclination
IDS Intermediate Disp ersion Sp ectrograph
INT Isaac Newton Telescope
invis Invisible comp onent
Jan January
JD Julian Date
Jun June
Jul July
K Radial Velocity Semi Amplitude
K Kelvin
km Kilometer
KPD Kitt Peak Downes
LINFOR LINLTE FOR
LTE Lo cal ThermoEquilibrium
M Mass
M Chandrasekhar mass M
Chandra
M Total mass
tot
Mar March
max Maximum
Mg Magnesium
MIDAS Munich Image Data Analysis System
MS Main Sequence star
n By number
N Nitrogen
n Logarithmic helium abundance log n n
He He H
Ni Nickel
NLTE Non Lo cal ThermoEquilibrium
Nov November
NS Neutron star
NTT New Technology Telescope
Oct Octob er
O Oxygen
P Period
PG Palomar Green
Q Quality
prec Precursor
prim Primary
R Radius
RGB Red Giant Branch
RV Radial velocity
RVV Radial velocity variable
s
Second
S Sulfur
sl Single lin ed system
S N Signal to Noise
sdB Sub dwarf B
sdO Sub dwarf O
sdOB Sub dwarf OB
sec Secondary
Sep September
Si Silicon
SN Sup ernova
SPY SN Ia Progenitor Survey
SVD Singular Value Decomp osition
APPENDIX E LIST OF ABBREVIATIONS
T E ective temp erature
e
T Hubble time
H
t Merging time
m
TAEHB Terminal Age Extended Extreme Horizontal Branch
Ti Titanium
UV Ultra Violet
v Orbital velocity
v sin i Pro jected rotational velocity
rot
vis Visual comp onent
VLT Very Large Telescope
WD White Dwarf
WHT William Herschel Telescope
y Year
ZAEHB Zero Age Extreme Extreme Horizontal Branch
Zn Zinc
References
Benvenuto O G Althaus L G MNRAS
Bergeron P Lieb ert J ApJ
Bl ocker T Herwig F Drieb e T et al in White dwarfs eds J Isern M Hernanz E
Garcia Berro Kluwer Dordrecht p
Bragaglia A Greggio L Renzini A D Odorico S ApJ L
Bragaglia A Renzini A Bergeron P ApJ
Branch D Fisher A Nugent P AJ
Branch D Tammann G A ARA A
Bruch A Diaz M P AJ
Chandrasekhar S An Introduction to the Study of Stel lar Structure University of Chicago
Press
Christlieb N Wisotzki L Reimers D et al A A
D Odorico S ESO Msngr
D Odorico S Kap er L Kaufer A UVES user manual Do c No VLT MAN ESO
Dorman B Ro o d R T O Connell R W ApJ
Drieb e T Sch onberner D Bl ocker T Herwig F A A
Drilling J S Landolt A U in Al len s Astrophysical Quantities th edition ed Arthur
N Cox AIP Press Springer New York
Edelmann H PhD Thesis Friedrich Alexander Universit atErlangen N urnberg
Ekb erg J O A AS
Ergma E Fedorova A V Yungelson L R A A L
Finley D S Ko ester D Basri G ApJ
Friedrich S Ko ester D Christlieb N et al A A
REFERENCES
Hachisu I Kato M ApJ
Hagen H J Gro ote D Engels D Reimers D A AS
Han Z Podsiadlowski P Maxted P F L Marsh T R MNRAS
Heb er U Drechsel H stensen R Karl C et al A A
Heb er U Edelmann H Lisker T Napiwotzki R A A L
Heb er U Napiwotzki R Reid I N A A
Heb er U Reid I N Werner K A A
Herrmann M Zulassungsarbeit Friedrich Alexander Universit atErlangen N urnberg
Hilditch R W Kilkenny D Lynas Gray A E Hill G MNRAS
Hirata R Horaguchi T Atomic spectral line list Department of Astronomy Kyoto
University and National Science Museum
H o ich P Wheeler J C Thielemann F K ApJ
Horne K PASP
Ib en I Jr Tutukov A V ApJS
Ib en I Jr Tutukov A V Yungelson L R ApJ
Karl C Napiwotzki R Heb er U Dreizler S Ko ester D Reid I N A A submitted
Karl C Napiwotzki R Nelemans G Christlieb N Ko ester D Heb er U Reimers D
A A
Kilkenny D Heb er U Drilling J S SAAOC
Kilkenny D O Donoghue D Ko en C Stobie R S Chen A MNRAS
Ko ester D Dreizler S Weidemann V Allard N F A A
Ko ester D Napiwotzki R Christlieb N et al A A
th
Kurucz R L Proc of the Symposium of the IAU eds B Barbuy A Renzini Kluwer
Dordrecht p
Lamontagne R Demers S Wesemael F Fontaine G Irwin M J AJ
Landau L D Lifshitz E M The Classical Theory of Fields Pergamon
Leibundgut B ARA A
Li W Filipp enko A V Tre ers R R Riess A G Hu J Qiu Y ApJ
Lisker T Diploma Thesis Friedrich Alexander Universit atErlangen N urnberg
REFERENCES
Lisker T Heb er U Napiwotzki R Christlieb N Han Z Homeier D Reimers D
accepted
Livio M in Type Ia Supernovae Theory and Cosmology Cambridge Univ Press p
Lorenz L Mayer P Drechsel H A A
Marsh T NewAR
Marsh T R Dhillon V S Duck S R MNRAS
Maxted P F L Marsh T R MNRAS
Maxted P F L Marsh T R Moran C K J a MNRAS
Maxted P F L Marsh T R Moran C K J Han Z b MNRAS
Maxted P F L Marsh T R North R C c MNRAS L
Maxted P F L Marsh T R Moran C K J a MNRAS
Maxted P F L Marsh T R Heb er U et al c MNRAS
Maxted P F L Burleigh M R Marsh T R Bannister N P b MNRAS
McCo ok G P Sion E M ApJ
M endezR H Kudritzki R P Herrero A Husfeld D Groth H G A A
Mo ehler S Richter T de Bo er K S Dettmar R J Heb er U A AS
Munari U Renzini A ApJ L
Napiwotzki R A A
Napiwotzki R Green P J Sa er R A ApJ
Napiwotzki R Christlieb N Drechsel H et al a AN
Napiwotzki R Edelmann H Heb er U Karl C Drechsel H Pauli E M Christlieb N b
A A L
Napiwotzki R Ko ester D Nelemans G et al A A
Napiwotzki R Christlieb N Drechsel H et al a ESO Msngr
Napiwotzki R Christlieb N Drechsel H et al b in From Twilight to Highlight ed W
Hillebrandt B Leibundgut Springer Verlag Berlin Heidelb erg p
Nelemans G Yungelson L R Portegies Zwart S F Verbunt F A A
Pauli E M Napiwotzki R Altmann M heb er U Odenkirchen M Kerb er F A A
Phillips M M ApJL L
REFERENCES
Rauch T A A
Rauch T Werner K A A
Reimers D Wisotzki L ESO Msngr
Riess A G Filipp enko A V Challis P Clo cchiatti A et al AJ
Riess A G PASP
Sa er R A Bergeron P Ko ester D Lieb ert J ApJ
Sa er R A Livio M Yungelson L R ApJ
Simon K P Sturm E A A
Stark M A Wade R A Aj
Stobie R S Kilkenny D O Donaghue D et al MNRAS
Str oerA Diploma Thesis Friedrich Alexander Universit atErlangen N urnberg
Truran J W Livio M ApJ
van den Haeufel E P J Bahttacharya D Nomoto K Rappap ort S A A A
Vennes S Thorstensen J R Polomski E F ApJ
Whelan J Ib en I Jr ApJ
Wiese W L Fuhr J R Deters T M Atomic transition probabilities of carbon nitrogen
and oxygen a critical data compilation eds W L Wiese J R Fuhr T M Deters American
Chemical Sco ciety for the National Institute of Standards and Technology Washington DC
Wisotzki L Christlieb N bade N et al A A
Wisotzki L Ko ehler T Gro ote D Reimers D A AS
th
Woo d M A in White Dwarfs Proceedings of the European Workshop on White Dwarfs
Springer Verlag Berlin Heidelb erg New York p Wright E L New Astronomy Review
Acknowledgments
First of all I want to express my gratitude to my sup ervisors Ralf Napiwotzki and Uli Heb er for
setting the stage for this They o ered me the opp ortunity to do my PhD on a promising and
inspiring eld of astronomy by joining the SPY consortium Without their continuous supp ort
and encouragement their shared talent and knowledge and the vast amount of time sp ent on
me and even patience I guess the last years would b e more stress and less pleasure
I want to thank the SPY consortium as well Esp ecially Gijs Nelemans who generously
shared radial velocity data for several stars Norb ert Christlieb for four wonderful weeks of
collab oration in Hamburg and Detlev Ko ester for doing all the sp ectral analysis
Many thanks also to all my colleagues in the Dr Remeis Sternwarte who made the institute
more than a simple working place I want to thank Irmela Bues for her steady interest in my
work and for many useful comments made over the course of the years I appreciate the supp ort
of Horst Drechsel who never refused me when I was confused by my computer or needed advice
how to handle astronomical problems Thank you Rainer Sterzer for the help with all the
little computer problems and for the countless hours you sp ent on teaching me how simple
life could b e My work also b ene ted from Norb ert Przybilla s generous assistance who was
at hand whenever help was needed I am grateful to Siegfried Falter Markus Ramsp eck and
Heinz Edelmann for all the inspiring discussions ab out Astronomy and more Furthermore I am
thankful to Heinz for all the useful tips and tricks ab out how to handle computers I appreciate
our native sp eaker Simon O Toole who kindly agreed to read through my thesis and saved the
reader from my crimes against his mother tongue Thanks a lot to the rest of the Remeis
Team Alex Str oer Eva Maria Pauli Martin Altmann Michael Bauer Sabine M ohler Stefan
Nesslinger Thorsten Lisker and Zorica Salomon for contributing to the nice atmosphere in our
go o d old institute I also want to mention Edith Day for all the administration work she did
during the last years
I am grateful to all the astronomers I have met during the last years By means of countless
discussions on conferences workshops meetings and seminars I discovered how exceptional
astronomers can b e Sp ecial thanks also to Klaus Werner for b eing a referee for this thesis and
to the sta of the Calar Alto Observatory Spain and the la Silla Observatory Chile for the
help supp ort and valuable assistance during my visits
I made extensive use of NASAs Astrophysics Data System Abstract Service ADS the
SIMBAD database op erated at CDS Strasb ourg France the Digital Sky Survey DSS and
the LEO English German Dictionary http dict leo org
This work was supp orted by the Deutsche Forschungs Gemeinschaft DFG under grant
Na In addition I gratefully acknowledge several travel grants to the Calar Alto obser
vatory
Thanks go also to all my friends who backed me over the years Thank you for b eing friends
And last but not least I want to thank my family who never lose faith on me Thank
you Mum and Dad Brother and Sister and you to o little nephew for b eing my eye of the
Hurricane all the time This work is dedicated to you
Curriculum Vitae
Name Christian Anton Karl
Geburtsdatum
Geburtsort Bamberg
Staatsangeh origkeit Deutsch
Beruf Diplom Physiker Univ
Familienstand Ledig
Konfession R omisch Katholisch
Anschrift Mainstrasse
Hallstadt
Beru icher Wertegang
Grundschule D or eins
Dientzenhofer Gymnasium Bamberg
WS WS Studium der Physik an der Universit atBayreuth
SS SS Studium der Physik an der Friedrich Alexander Universit at
Erlangen N urnberg
Sptember Diplompr ufung in Physik
Seit Oktob er Besch aftigungals Wissenschaftlicher Mitarb eiter
und Anfertigung der Dissertation an der
Dr Karl Remeis Sternwarte Bamberg
dem Astronomischen Institut der
Friedrich Alexander Universit at Erlangen N urnberg