Emilios T. Harlaftis,5,6
Jonay I. Gonza´lez Herna´ndez,1,2,3 Rafael Rebolo,1,4 Garik Israelian,1 Alexei V. Filippenko,7 and Ryan Chornock7
Received 2006 April 3; accepted 2006 April 26; published 2006 May 25
(BEING CONTINUED FROM 21/10/2015)
3. CHEMICAL ANALYSIS
A comparison of the observed spectrum with the spectra of 10 template stars (K0 V–M2 V) obtained with the same instrument
allows us to classify the secondary as a mid to late K-type star. In Figure 1, we display two spectral regions containing
some relevant lines for our analysis as well as a spectrum of a template star of similar spectral type for comparison
with available detailed chemical analysis (Allende Prieto et al.2004). We also show synthetic line profiles for different stellar
abundances computed with the local thermodynamic equilibrium (LTE) code MOOG (Sneden 1973), adopting the atomic
line data from the Vienna Atomic Line Database (VALD; Piskunov et al. 1995) and using a grid of LTE model atmospheres
(Kurucz 1993).
Fig. 1.—Best synthetic spectral fits to the ESI spectrum of the secondary star in the XTE J1118480 system (second and bottom panels) and the same for a template star (properly broadened) taken from Allende Prieto et al. (2004) shown for comparison (top and third panels). Synthetic spectra are computed for typical abundances for a halo star ([Fe/H]p 1.2; dash-dotted line), solar abundances ([Fe/H]p 0; dashed line), and best-fit abundances (solid line). In addition, note that for solar and best-fit abundances we have applied to the synthetic spectra the corresponding values for the veiling according to the solution found with the fitting procedure. However, for the low-metallicity synthetic spectra, we have not assumed any veiling. [See the electronic edition of the Journal for a color version of this figure.]
In order to perform the chemical analysis of the secondary star, we used a technique that combines a grid of synthetic
spectra and a 2-minimization procedure that includes Monte Carlo simulations (Gonza´lez Herna´ndez et al. 2004, 2005).
First, we inspected the observed spectrum in order to select the most suitable features for a chemical abundance determination.
We identified nine spectral features containing in total 30 lines of Fe i and eight lines of Ca i with excitation potentials
between 1 and 5 eV. The oscillator strengths of the relevant spectral lines were checked via spectral synthesis against the
solar atlas (Kurucz et al. 1984). We then generated a grid of about one million synthetic spectra of each of these features,
varying as free parameters the star effective temperature (T eff) surface gravity (log g), and metallicity ([Fe/H]), together with
the veiling from the accretion disk, which was assumed to be a linear function of wavelength, and thus described by two
additional parameters. Iron abundances were varied in the range -1.5 < [Fe/H] < 1, whereas the Ca abundance was fixed, for
each given iron abundance, according to the Galactic trend of Ca (Bensby et al. 2005) for [Fe/H] <0, and fixed to [Ca/Fe]= 0 in the range 0 < [Fe/H]< 1. A rotational broadening of 100 km s-1 and a limb darkening e=0.8 were adopted. The micro turbulence (ξ) was computed using an experimental expression as a function of effective temperature and surface gravity (Allende Prieto et al. 2004).
We compared, using a 2-minimization procedure, this grid x with 1000 realizations of the observed spectrum. Using a bootstrap
Monte Carlo method, we found the most likely values 
0.2 +-0.2, and a disk veiling (defined as 
) of less than disk total 40% at 5000 and decreasing toward longer wavelengths. The °A(1 σ) uncertainty in the iron-abundance determination takes into account the uncertainties in the stellar and veiling parameters.
The effective temperature and surface gravity are consistent with previous spectral classifications and similarly the reported
veiling values (Torres et al. 2004). Using the derived stellar and veiling parameters, we analyzed several spectral regions
where we had identified various lines of Fe, Ca, Al, Mg, and Ni. Abundances of all the elements are listed in Table 1. The
1 σ uncertainty in the abundance determination takes into account uncertainties in the stellar and veiling parameters.
Remarkably, we find a metallicity higher than solar, which is extremely atypical of halo stars (Allende Prieto et al. 2006).
In Figure 1, we show the best-fit synthetic spectrum to various features in two different spectral regions, and for comparison,
a model with 25 times lower metal content. Notice the inadequacy of low-metallicity models to reproduce the observed
features, even in the extreme case that no veiling is considered (adding veiling would make the discrepancy much worse). An
iron abundance of [Fe/H] =-1.2 is more than 6σ away from the best-fit solution ([Fe/H]= 0.18) and hence very unlikely.
We have also found that abundances of Al, Ca, Mg, and Ni are higher than solar (see Table 1).
Fig. 2.—Abundance ratios of the secondary star in XTE J1118480 (cross) in comparison with the abundances of G and K metal-rich dwarf stars. Galactic trends were taken from Gilli et al. (2006). The size of the cross indicates the uncertainty. Filled and empty circles correspond to abundances for planet host stars and stars without known planet companions, respectively. The dash-dotted lines indicate solar abundance values.
In Figure 2, we show that the abundance ratios of these elements with respect to iron are consistent with those of stars in the solar neighborhood from Gilli et al. (2006). We have also determined an upper limit to the Li abundance 
1.86 using the Li 6708 A° line. This value seems to be lower than typical high Li abundances measured in other late-type secondary stars in soft X-ray transients whose origin is still an open question (Martı´n et al. 1994).
(TO BE CONTINUED)
