Fundamental parameters of 8 Am stars: comparing observations with theory

In this paper we present a detailed analysis of a sample of eight Am stars, four of them are in the {\it Kepler} field of view. We derive fundamental parameters for all observed stars, effective temperature, gravity, rotational and radial velocities, and chemical abundances by spectral synthesis method. Further, to place these stars in the HR diagram, we computed their luminosity. Two objects among our sample, namely HD\,114839 and HD\,179458 do not present the typical characteristic of Am stars, while for the others six we confirm their nature. The behavior of lithium abundance as a function of the temperature with respect the normal A-type stars has been also investigated, we do not find any difference between metallic and normal A stars. All the pulsating Am stars present in our sample (five out of eight) lies in the $\delta$~Sct instability strip, close to the red edge.


INTRODUCTION
Among main sequence, A-type stars show a large variety of chemical peculiarities. They are driven by several physical processes, such as diffusion and/or magnetic field, just to quote some of them. All these processes have the same factor in common, i. e. the very stable radiative atmosphere which is the principal condition needed for peculiarities to arise.
The metallic or Am stars are those whose Caii K-line types appear too early for their hydrogen line types, and metallic-lines types appear too late, such that the spectral types inferred from the Caii K-and metal-lines differ by five or more spectral subclasses. The marginal Am stars are those whose difference between Caii K-and metal-lines is less than five subclasses. The commonly used classification for this class of objects include three spectral types prefixed with k, h, and m, corresponding to the K-line, hydrogen-lines and metallic lines, respectively. The typical abundances pattern show underabundances of C, N, O, Ca, and Sc and over-⋆ Based on observations made with the Italian Telescopio Nazionale Galileo (TNG) operated on the island of La Palma by the Fundación Galileo Galilei of the INAF (Istituto Nazionale di Astrofisica) at the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias † E-mail: gca@oact.inaf.it abundances of the Fe-peak elements, Y, Ba and of the rare earths elements (Adelman et al. 1997;Fossati et al. 2007). The presence of magnetic field has also been investigated but with null result by Fossati et al. (2007). The abundance of lithium in Am stars compared to that observed in normal A-type stars, has been discussed in the literature since the work of Burkhart & Coupry (1991). They found that in general Li abundance in Am stars is close to the cosmic value or even lower in some case. Richer, Michaud, & Turcotte (2000) developed models of the structure and evolution of Am stars in order to reproduce the observed chemical pattern of 28 elements. The most important improvement of these models has been the introduction of turbulence as the hydrodynamical process competing with atomic diffusion, in such a way that the resulting mixing reduces the large abundance anomalies predicted by previous models, leading to abundances which closely resemble those observed in Am stars.
Another open question in the framework of Am stars concerns the pulsations in these objects. For many years it was thought that Am stars did not pulsate, in accordance with the expectation that diffusion depletes helium from the driving zone. Recently, intensive groundbased (Smalley et al. 2011, SuperWASP survey) and spacebased , Kepler mission) observations have shown that many Am/Fm stars do pulsate. Smalley et al. Table 1. Physical parameters estimated from photometry and parallaxes. The different columns show: (1) and (2) the HD number and an alternative name (if any) for the target star; (3) and (4) the B and V magnitudes adopted (σ B ,σ V ∼0.020,0.015 mag, respectively); (5) the Ks photometry from 2MASS (σ Ks ∼0.015 mag); (6) the b − y colour (σ b−y ∼ 0.01 mag); (7) the parallax (after van Leeuwen 2007); (8) the E(B − V ) values (the uncertainty is 0.01 mag for the first four stars, 0.02 mag for the ramaining four objects; (9) the bolometric correction in the V band (after Bessell, Castelli & Plez 1998); (10) and (11) the T eff estimated from (V − Ks) and uvbyβ photometry, respectively; (12) and (13) the log g estimated from uvbyβ photometry and Eq. 1, respectively. (2011), for example, found that about 169, 30 and 28 Am stars out of a total of 1600 show δ Sct, γ Dor or Hybrid pulsations (see Grigahcène et al. 2010, for a definition of these classes). These authors found also that the positions in the Hertzsprung-Russel (HR) diagram of Am stars pulsating as δ Sct are confined between the red and blue radial fundamental edges, in agreement with Balona et al. (2011) and Catanzaro & Balona (2012).
In this study we continue a programme devoted to determining photospheric abundance pattern in Am stars by means of high resolution spectra. Three Am stars have already been analyzed by us, namely: HD 178327 (KIC 11445913) and HD 183489 (KIC 11402951) in Balona et al. (2011), andHD 71297 in Catanzaro, Ripepi &Bruntt (2013), for which fundamental astrophysical quantities, such as effective temperatures, gravities and metallicities have been derived. The addition of these three stars does not alter the homogeneity of our sample, since all of them have been observed with the same instrumentation and the spectra were reduced and analyzed with the same procedure that we will describe in Sect. 2. Such kind of studies are crucial in order i) to put constraints on the processes occurring at the base of the convection zone in non-magnetic stars and ii) to try to define the locus on the HR diagram occupied by pulsating Am stars.
With these goals in mind, we present a complete analysis of other eight stars previously classified as Am stars. Four of them belong to the sample observed by the Kepler satellite ) and other four are Am stars discovered to be pulsating from ground-based observations. For our purposes high-resolution spectroscopy is the best tool principally for two reasons, i) the blanketing due to the chemical peculiarities in the atmospheres of Am stars alters photometric colors and then fundamental stellar parameters based on them may not be accurate (see Catanzaro & Balona 2012, and Sect. 3.3 for details) and ii) the abnormal abundances coupled with rotational velocity result in a severe line blending which makes difficult the separation of the individual lines. Both problems could be overcome only by matching synthetic and observed spectra.
For the confirmed Am stars we will compare the observed abundance with the predictions of the models and we will place them on the HR diagram by evaluating their luminosities.

OBSERVATION AND DATA REDUCTION
Spectroscopic observations of our sample of Am stars (see Tab. 1 for the list of targets) were carried out with the SARG spectrograph, which is installed at the Telescopio Nazionale Galileo, located in La Palma (Canarias Islands, Spain). SARG is a high-resolution cross-dispersed echelle spectrograph (Gratton et al. 2001) that operates in both single-object and longslit observing modes and covers a spectral wavelength range from 370 nm up to about 1000 nm, with a resolution ranging from R = 29 000 to 164 000.
Our spectra were obtained in service mode in 2011, between February 21 th and June 12 th , at R = 57 000 using two grisms (blue and yellow) and two filters (blue and yellow). These were used in order to obtain a continuous spectrum from 3600Å to 7900Å with significant overlap in the wavelength range between 4620Å and 5140Å. We acquired the spectra for the stars with a signal-to-noise ratio S/N of at least 100 in the continuum.
The reduction of all spectra, which included the subtraction of the bias frame, trimming, correcting for the flat-field and the scattered light, the extraction for the orders, and the wavelength calibration, was done using the NOAO/IRAF packages 1 . The IRAF package rvcorrect was used to make the velocity corrections due to Earth's motion Table 2. Results obtained from the spectroscopic analysis of the sample of Am stars presented in this work. The different columns show: (1) identification; (2) effective temperatures; (3) gravity (log g); (4) microturbolent velocity (ξ); (5) rotational velocity (v sin i), (6) Heliocentric Julian Day of observation; (7) radial velocity (V rad ); (8) indication of binarity (Y=binary; N=not binary; U=data insufficient to reach a conclusion); (9) indication of belonging to the Am star class (Y=Am; N=not Am); (10) indication of presence of pulsation (Y=pulsating; N=not pulsating; after Balona et al. 2011, and references therein).

PHYSICAL PARAMETERS
Temperatures and gravities for our sample stars have been derived by spectral synthesis, as described in Sect. 3.2. In order to speed up the iterative calculations, we needed starting values for both parameters, that have been estimated from photometric calibrations, as described in the following Sect. 3.1. In the same section we estimate the infrared excess and the bolometric corrections, needed to compute stellar luminosities of our stars (see Sect.5).
3.1 Parameters from photometry: T eff and log g For five out of eight stars in our sample (HD 104513, HD 113878, HD 114839, HD 118660, and HD 179458) complete Strömgren-Crawford uvbyβ photometry is available (Hauck & Mermillod 1998). For the remaining 3 objects (HD 176843, HD 187254, and HD 190165), only Johnson photometry is available, mainly in BV filters. For these stars we derived the Johnson B, V magnitudes from Tycho (BT , VT ) photometry adopting the transformations into the standard system provided by Bessell (2000). The same procedure was applied to all the other stars for homogeneity. The resulting B, V magnitudes are listed in Tab. 1 (column 2 and 3). In the near-infrared, JHKs photometry of good quality is present in the 2MASS catalogue (Skrutskie et al. 1996) for all the targets. We adopted an updated version of the TempLogG 2 software (Rogers 1995) to estimate T eff and log g by using the calibrations present in the package, namely Balona (1984); Moon (1985); Moon & Dworetsky (1985); Napiwotzki et al.
2 available through http://www.univie.ac.at/asap/manuals/tipstricks/templogg.localaccess.html (1993); Ribas et al. (1997). In addition, we considered the results by Smalley & Kupka (1997) and Heiter et al. (2002) who provided uvby grids based on the Kurucz model atmospheres but with different treatment of the convection. In particular, we used Smalley & Kupka (1997) grids built using Canuto & Mazzitelli (1991) convection treatment and two choices for the grids 3 by Heiter et al. (2002): i) standard mixing-length theory (MLT) 4 ; ii) the Canuto, Goldman & Mazzitelli (1996) treatment of the convection. For each star, the different determinations T eff and log g were comparable with each other and we decided to simply average them. The result is shown in Table 1 (columns 9 and 10).
As for the reddening estimate, we have adopted different methods, depending on the data available.
• For the five stars possessing uvbyβ photometry, we used TempLogG to estimate the values of E(b − y), that were converted into E(B − V ) using the transformation E(B − V ) = 1.36 E(b − y) (Cardelli, Clayton & Mathis 1989).
• We inspected the spectra af all our targets looking for the presence of the interstellar lines Nai 5890.0Å(D1) and Ki 7699Å. The equivalent widths (EWs) of these lines can be converted into E(B − V ) according to e.g. Munari & Zwitter (1997). As a result of this procedure, the only measurable lines were Nai in HD 187254 (EW∼140 mÅ) and Ki in HD 179458 (EW∼15 mÅ), corresponding to E(B − V )=0.04±0.02 mag for both stars. For the remaining objects the interstellar lines were not measurable because they were too small (compatible with the almost zero absorption in the direction of HD 104513, HD 113878, HD 114839, and HD 118660 as derived from uvbyβ photometry) or completely embedded into the photospheric line. Is it worth noticing that for HD 179458 the uvbyβ photometry provided a different reddening estimate than that estimated from Ki, and precisely E(B − V )=0.01±0.01 mag. Since we 3 These grids are available on the NEMO site www.univie.ac.at/nemo/gci-bin/dive.cgi judge that the Munari & Zwitter (1997) calibration are reliable, for HD 179458 we decided to adopt the reddening evaluated from the interstellar lines.
• For the two remaining stars devoid of reddening estimate through the aforementioned methods (namely, HD 176843, HD 190165), we adopted the tables by Schmidt-Kaler (1982) in conjunction with the spectroscopic T eff and log g (see next section) to estimate their instrinsic color (B − V )0. A simple comparison with the observed ones gives an estimate of the reddening for these stars.
The adopted reddening estimated are reported in Table 1 (column 6).
To estimate a star's fundamental parameters from photometry and parallax, we need to evaluate first the visual bolometric correction BCV . To this aim we adopted the models by Bessell, Castelli & Plez (1998) where it is assumed that M bol,⊙ = 4.74 mag. We interpolated their model grids adopting the correct metal abundance that we derived in Sect. 4 as well as the values of T eff and log g derived spectroscopically (see next section). The result of this procedure is reported in Table 1 (column 7).
An additional photometric estimate of T eff can be derived for all the targets using the calibration T eff =T eff ((V − Ks)0,log g and [F e/H]) published by e.g. Masana, Joedi & Ribas (2006) or Casagrande et al. (2010). Both works give similar results and we decided to use Masana, Joedi & Ribas (2006)'s calibration for homogeneity with our previous papers (e.g. Catanzaro et al. 2011). As quoted above, the photometry in V and Ks is available from Tycho and 2MASS, respectively. As for log g and [F e/H] we used the values from our spectroscopy. To deredden the observed (V − Ks) colours we adopted the reddening reported in Table 1, (column 4) using the relation (Cardelli, Clayton & Mathis 1989). The resulting T eff and the relative errors are reported in Table 1 (column 8).
Concerning log g, it is possible to estimate with good accuracy this quantity independently from both spectroscopy and Strömgren photometry if the parallax is known with sufficient precision (i.e. 10%). As shown in Tab. 1 (column 5), this is the case for three stars in our list, namely HD 104513, HD 114839, and HD 118660, whereas for HD 113878 the error on the parallax is of the order of 30%. To estimate log g we used the following expression: where the different terms of the above relationship have the usual meaning and M/M⊙ is the mass of the star in solar unit. Before using Eq. 1, we have to evaluate the mass of the three stars. This can be done by adopting the calibration mass-MV by Malkov (2007) that was derived on the basis of a large sample of eclipsing binaries stars. Hence, by using our MV estimate discussed in Sect. 5, we evaluated log(M/M⊙)=0.20, 0.24, and 0.15 dex with a common error of 0.05 dex (dominated by the dispersion of the mass-MV relation) for HD 104513, HD 114839, and HD 118660, respectively. For HD 113878 we obtained log(M/M⊙)=0.50±0.11 dex, being the error dominated by the large uncertainty on the parallax. Finally, the log g resulting from the above procedure are listed in column (11) of Table 1.

Atmospheric parameters from spectroscopy
In this section we present the spectroscopic analysis of our sample of Am stars, in order to derive fundamental astrophysical quantities, such as: effective temperatures, surface gravities, rotational velocities and chemical abundances. The approach used in this paper has been succesfully used in other papers devoted to this topics, see for instance Catanzaro et al. (2011);Catanzaro & Balona (2012); Catanzaro, Ripepi & Bruntt (2013). In practice, the procedure used for our targets was to minimize the difference among observed and synthetic spectrum, using as goodnessof-fit parameter the χ 2 defined as where N is the total number of points, I obs and I th are the intensities of the observed and computed profiles, respectively, and δI obs is the photon noise. Synthetic spectra were generated in three steps. First, we computed LTE atmospheric models using the ATLAS9 code (Kurucz 1993a,b). Second, the stellar spectra were then synthesized using SYNTHE (Kurucz & Avrett 1981). Third, the spectra were convolved for the instrumental and rotational broadenings.
We computed the v sin i of our targets by matching synthetic lines profiles from SYNTHE to a number of metallic lines. The Mgi triplet at λλ5167-5183Å was particularly useful for this purpose. The results of these calculations are reported in Tab. 2.
To determine stellar parameters as consistently as possible with the actual structure of the atmosphere, we performed the abundances analyses by the following iterative procedure: (i) T eff was estimated by computing the ATLAS9 model atmosphere which gave the best match between the observed H β and H δ lines profile and those computed with SYNTHE. The models were computed using solar opacity distribution functions (ODF) and microturbulence velocities according to the calibration ξ = ξ(T eff , log g) published by Allende Prieto et al. (2004). For what concerns the treatment of convection, models cooler than 8000 K were computed using the classical MLT with fixed α = 1.25 (Castelli, Gratton & Kurucz 1997). The effects of different convection treatment on the Balmer lines profiles has already been investigated in Catanzaro, Ripepi & Bruntt (2013), for the specific case study of HD 71297. In that paper we concluded that theoretical profiles change according to the convection treatment, in the sense that the separation between the two profiles increases from the line core towards the wings. However, the maximum difference is very low, of the order of 1.5 %, really indistinguishable at our level of S/N and for our resolving power. Since the star analyzed in that paper share the same classification (Am) of the targets presented here, and it has been observed with the same equipment (SARG@TNG) and in the same observing run, we are confident that the conclusions obtained in Catanzaro, Ripepi & Bruntt (2013) continue to apply also here. These two Balmer lines are located far from the echelle orders edges so that it was possible to safely recover the whole profiles. The simultaneous fitting of two lines led to a final solution as the intersection of the two χ 2 iso-surfaces. An important source of uncertainties arised from the difficulties in normalization as is always challenging for Balmer lines in echelle spectra. We quantified the error introduced by the normalization to be at least 100 K, that we summed in quadrature with the errors obtained by the fitting procedure. The final results for effective temperatures and their errors are reported in Tab. 2. The surface gravity was estimated accordingly to the effective temperature of the star: for HD 179458 and HD 187254, i.e. the only stars of our sample hotter than 8000 K, we used the wings of Balmer lines as a diagnostic tool, while for the others, we derived log g from fitting the wings of broad lines of Mgi triplet at λλ 5167, 5172, and 5183Å, which are very sensitive to log g variations. As an example, we show in Fig. 1 the fit for three stars of our sample, with different rotational velocities. In practice, we have first derived the magnesium abundances through the narrow Mgi lines at λλ 4571, 4703, 5528, 5711Å (not sensitive to log g), and then we fitted the wings of the triplet lines by fine tuning the log g value. To accomplish this task is mandatory to take into account very accurate measurements of the atomic parameters of the transitions, i.e. log gf and the radiative, Stark and Van der Waals damping constants. Regarding log gf we used the values of Aldenius et al. (1997), Van der Waals damping constant is that calculated by Barklem , and the radiative damping constant is from NIST database (log γ rad = 7.99).
The values of log g, derived with this methods, have been checked by the ionization equilibrium between Fei lines (not sensisitive to gravity change) and Feii (very sensisitive to log g). This procedure results in the final values reported in Tab. 2. Uncertainties in T eff , log g, and v sin i were estimated by the change in parameter values which leads to an increases of χ 2 by unity (Lampton, Margon & Bowyer 1976).
(ii) As a second step we determine the stellar abundances by spectral synthesis. Therefore, we divide each of our spectra into several intervals, 50Å wide each, and derived the abundances in each interval by performing a χ 2 minimization of the difference between the observed and synthetic spectrum. The minimization algorithm has been written in IDL language, using the amoeba routine. We adopted lists of spectral lines and atomic parameters from Castelli & Hubrig (2004), who updated the parameters listed originally by Kurucz & Bell (1995).
For each element, we calculated the uncertainty in the abundance to be the standard deviation of the mean obtained from individual determinations in each interval of the analyzed spectrum. For elements whose lines occurred in one or two intervals only, the error in the abundance was evaluated by varying the effective temperature and gravity within their uncertainties given in Table 2, [T eff ± δT eff ] and [log g ± δ log g], and computing the abundance for T eff and log g values in these ranges. We found a variation of ∼0.1 dex due to temperature variation, while we did not find any significant abundance change by varying log g. The uncertainty in the temperature is the main error source in our analyses.

Comparison between astrophysical parameters derived by different methods
It is useful to compare the values of T eff and log g derived spectroscopically (see Table 2) with those obtained via photometric methods (see Table 1). Quantitatively, a weighted mean of the differences gives: From an analysis of these results it appears that the T Spec eff are in good agreement within the errors with the T eff estimated from (V − Ks) colour, whereas they are colder than T uvbyβ eff by about 150 K, even if the significance of this value is only marginal (∼ 1 σ). Similarly, the log g Spec seems to be systematically smaller than log g uvbyβ and, to a smaller extent, than log g HIP . In the first case the discrepancy is not significant at 1σ level. In the second case, with the exception of HD 104513, there is agreement within the errors.
The above results for uvbyβ photometry are in agreement with those by Catanzaro & Balona (2012) who showed how the Strömgren indices are correlated with effective temperature and log g and how they are affected by blanketing in Am stars. These authors concluded that effective temperature can be reliably derived by Strömgren photometry, but because the sensitivity of (b-y) to abundances, it is in general higher of about 200 K. The situation is worst for the gravities. Indeed, given the strong effect of blanketing on the c1 index, the gravities, and, in turn, the luminosities, are completely unreliable.

CHEMICAL ABUNDANCES
In this section we present the results of the abundance analysis obtained for each star in our sample. The derived abundances and the estimated uncertainties, expressed as log N el N T ot , are reported in Tab. 3. The abundance patterns for each star, expressed in terms of solar values (Grevesse et al. 2010), are shown in Fig. 2. We also searched for binarity among our sample, combining our own measurements of radial velocity (reported in Tab. 2) with those found in literature, when available.
At the end of this section, we will discuss separately lithium abundance in Am stars with respect the normal Atype stars.

HD 104513
This star is known to be a metallic enhanced star since the pioneering work of Morgan (1932), who has noticed a strong Europium line at λ4129Å. Cowley et al. (1969), by using metal spectral lines classified this star as A7 marginal metallic star, that is in agreement with that found later by Hauck (1973). This author, in his "Catalogue of Am stars with known spectral types", reported HD 104513 to be an A7 from the Caii K line. Abt (1975) found vsin i = 65 ± 10 km s −1 .
Radial velocity measurements have been found in the literature, Abt & Levy (1985) published 23 velocities that are in agreement with the one measured by us and reported in Tab. 2. These velocities suggest a possible orbital motion, but since the amplitude is too low (≈ 5 km s −1 ) compared to errors on each measurements, we cannot conclude anything on the binarity of this object.
HD 104513 was the first marginal Am stars discovered to pulsate (Kurtz 1978). He found indication of multiple periodicities in the δ Scuti regime, with periods ranging from 0.81 hr to 1.90 hr.
To our knowledge, this is the first extensive abundances analysis so far published in the literature for HD 104513. We estimated T eff = 7100 ± 200 K and log g = 3.6 ± 0.1 dex, that are typical for an F0/1 star, and a vsin i = 72 ± 7 km s −1 totally consistent with that published by Abt (1975). Moreover, we found moderate overabundances of about 1 dex for P, Sr, Y, and Ba, a slight overabundance of iron and iron peak elements and moderate underabundances of Ca and Sc, about 0.2 dex and 1 dex, respectively. Thus we confirm the classification of a marginal Am star, but from the Balmer Figure 2. Chemical pattern for our targets, ordered by increasing effective temperature, from the coolest (top) to the hottest (bottom). Horizontal dashed line corresponds to solar abundance (Grevesse et al. 2010) and metallic lines we suggest it could be a star with a spectral type of F0/1.

HD 113878
HD 113878 was firstly classified as Am by Olsen (1980), who estimated its spectrum peculiarity on the basis of Strömgren photometric indices. Later on, this classification has been confirmed spectroscopically by Abt (1984), who define it as a kF1hF3VmF3 marginal Am star, because of its strong Srii lines and weak Cai λ4026Å line.
From the pulsational point of view this star has been intensitively studied in a series of paper by Joshi and collaborators. Joshi (2005), in his photometric search for variability in Ap and Am stars, discovered this star to pulsate with a period of about 2.3 hours, which is typical of δ Scuti stars. This period has been refined later by Joshi et al. (2006), who found P = 2.31 hr. Further observations carried out by Joshi et al. (2009) led the authors to conclude that HD 113878 is an evolved star.
From our analysis, we found T eff = 6900 ± 200 K and log g = 3.4 ± 0.1 dex, that are typical for an F1 evolved star, confirming both the results obtained by Joshi et al. (2009) and those from Casagrande et al. (2011). Regarding the abundance pattern, we found a slight underabundance of scandium of ≈ 0.5 dex, and a moderate overabundance of manganese, cobalt, germanium, strontium, yttrium, zirconium and barium, all ranging from 0.4 to ≈ 1 dex. A strong overabundance of copper, ≈ 1.8 dex, has also been observed. This pattern confirms the classification of this star as a marginal Am star.

HD 114839
Following Hill et al. (1976), this object is reported in the "General Catalogue of Ap and Am stars" (Renson, Gerbaldi & Catalano 1991) as an uncertain Am star. Pribulla et al. (2009) carried out medium resolution (R = 12000) spectroscopic observations at the David Dunlop Observatory, centered on the Mgi triplet at λλ5167-5184Å, from which they measured vsin i = 70 km s −1 and they concluded that it is a metallic line star of spectral type F4/5. Balona et al. (2011) reported a spectral type of kA5hF0mF3.
Only one measurement of radial velocity is reported in Gouthcharov (2006): −5.60 ± 1.40 km s −1 . This value is in agreement with our own reported in Tab. 2, at least within the experimental errors.
HD 114839 has been discovered as hybrid pulsator by King et al. (2006) by using space-based data carried out with the MOST satellite. They identify 15 frequencies, of which 4 are in the range between 1 and 2.5 c/d, consistent with γ Dor g-modes pulsations, while the remaining are between 6.5 and 22 c/d, typical for δ Sct p-modes.
For this star, we derived T eff = 7100 ± 200 K, log g = 3.8 ± 0.1 dex, and vsin i = 70 ± 7 km s −1 . These parameters led to a moderate (∼0.5 dex) overabundances of Na, Mg, S, Co, and Sr and only a strong (∼1.8 dex) overabundance of Ba. For what concerns the characteristic elements of the Am classification, we found only a moderate underabundance of scandium, while other light and iron-peak elements are almost solar. Thus, in conclusion we cannot confirm the Am peculiarity for this star.
A similar conclusion has been reached by Hareter et al. (2011). They performed an extensive spectroscopic study of HD 114839 with the aim to search for a link between the Am phenomenon and hybrid pulsators. Their effective temperature, surface gravity and rotational velocity are consistent with those derived in this study. Barry (1970) was the first who noted marginal characteristic of Am phenomenology in the spectrum of HD 118660. Later on, Cowley & Bidelman (1979) gave the first spectral classification relying on their Hγ spectrograms, denoting the star as a marginal A5m.

HD 118660
Two measurements of radial velocity have been reported in literature for HD 118660, Gouthcharov (2006) (−1.7 ± 2.9 km s −1 ) and Wilson (1953) (−1.7 km s −1 ). Those values are in perfect agreement with our measured velocity, so we can confirm the absence of variability. Joshi et al. (2006) discovered δ Scuti-like pulsations in this star, with a dominant period of about 1 hr and another prominent period of about 2.52 hr.
To our knowledge, this is the first detailed abundance analysis performed for HD 118660. Atmospheric parameters are: T eff = 7200 ± 200 K and log g = 3.9 ± 0.1 dex, and vsin i = 100 ± 10 km s −1 . Rotational velocity is consistent with the value reported by Royer et al. (2002) of 94 km s −1 . By using these values in our synthetic analysis, the most overabundance inferred was that of phosphorus of ∼1.5 dex. Moderate overabundances in the range 0.2 -0.6 dex have been found for S, Sc, iron-peak elements, Sr, Y, Zr and Ba. Solar to about −0.2 dex have been derived for other elements, including calcium and scandium. This result led us to conclude that HD 118660 is a marginal Fm star. This conclusion is corroborated by the work of Charbonneau & Michaud (1991), who established a rotational velocity limit of 90 km s −1 above which diffusion processes cannot cause Am peculiarities.

HD 176843
HD 176843 has been classified as kA3mF0, that is a marginal Am star, by Floquet (1975), but no studies are present in the recent literature regarding its astrophysical parameters.
Observed by the Kepler satellite, its periodogram has been presented firstly by Balona et al. (2011), who discovered excess power at two frequencies in the δ Sct domain, about at 34.4 c/d and 37.7 c/d. Uytterhoeven, Moya, & Grigahcene (2011) classify this object as a binary star with a δ Sct component. Unfortunately, we did not find any other measurements of radial velocity in literature, so we can not verify the possible binarity.
Even for this star, our study is the first ever reported in literature. Using the parameters we found, i.e. T eff = 7600 ± 150 K, log g = 3.8 ± 0.1 dex and vsin i = 27 ± 3 km s −1 , we found slight underabundances of Ca and Sc, normal values for C, Mg, Si, and Ti, and overabundances for the heavier elements of about 0.5 ÷ 1 dex. Strong overabundance of Ba (∼2 dex) have been observed, as well.
In conclusion this star shows the typical pattern of Am stars.

HD 179458
The nature of this star has been debated in the past years, but in spite of this discussion, its classification is still doubtful. MacRae (1952) noted its possible peculiar spectrum, but he did not give any details. Then the star was observed by Floquet (1970), which classified it as a normal A7 star. The uncertain nature is reported also in the "General Catalogue of Ap and Am stars" (Renson, Gerbaldi & Catalano 1991). No measurements of radial velocity are present in literature.
Observed by Kepler, its periodogram does not show any sign of variability  Our study shows that HD 179458 is an A4 main sequence star, with T eff = 8400 ± 200 K, log g = 4.1 ± 0.1 dex and vsin i = 75 ± 7 km s −1 . The most part of chemical elements observed in this star show overabundances, if compared with the respective solar values, from about 0.2 dex to about 1.5 dex. Besides its chemical pattern is far from the solar one, it is not typical for Am stars, so we can conclude that HD 179458 is not belonging to this class of peculiarity.

HD 187254
Reported as a metallic star by Mendoza (1974), HD 187254 has been then classified as kA2mF0 by Floquet (1975).
Seven radial velocities have been reported by Fehrenbach et al. (1997). Our measurement of radial velocity is compatible with those data, so that we confirm the presence of an orbital motion since the amplitude is ≈ 36 km s −1 , but we can not attempt for a search of orbital parameters due to the lack of a sufficient number of data.
From the pulsational point of view, this star has been studied by Balona et al. (2011) who analyzed the periodogram obtained with photometric data taken by the Kepler satellite. They concluded that it does not show any significant power excess in the δ Sct or γ Dor range, though clear low-frequency variability is present. Some of this lowfrequency variability may be of instrumental origin as longterm trends in Kepler data are not fully corrected. However, intrinsic variability could arise as a result of rotational modulation, for example. While no Am star is known to vary in this way from ground-based observations, it cannot be ruled out in Kepler photometry due to the extraordinary high precision.
Our study is the first ever detailed spectroscopic study, at least to our knowledge. From our spectrum we obtained: T eff = 8000 ± 150 K, log g = 4.1 ± 0.1 dex and vsin i = 15 ± 2 km s −1 . The only elements that appear to be solar are carbon and scandium, while a slight underabundance of ≈ 0.2 dex has been observed for calcium. Iron and iron-peak elements are slightly overabundant, as well light elements are. Strong overabundances have been observed for Cu, Sr, Y, and Zr almost 1 dex, and for Ba, about 2.4 dex. Therefore no doubt that it is an Am star.

HD 190165
This star is known to belong to the Am group since the work of Mendoza (1974), who carried out multicolor photometry for a sample of metallic stars. One year later, it was classified as kA2mF2 by Floquet (1975). Despite the fact that its nature has been known for a long time, both a detailed spectroscopic studies aimed at computing its chemical pattern and a measurement of the rotational velocity for HD 190165 are missing.
Regarding binarity, besides the two radial velocities reported in the literature are in agreement each other, v rad = −16.90 km s −1 (Gouthcharov 2006;Wilson 1953), we found a discrepant value of v rad = −7.45 ± 0.45 km s −1 . In any case we can not make any conclusion about its variability.
Kepler observations have been analyzed by Balona et al. (2011) and, like the case of HD 187254, they found only low-frequency variability.
From our spectrum we obtained T eff = 7400 ± 150 K and log g = 4.1 ± 0.1 dex, and vsin i = 58 ± 6 km s −1 . The chemical pattern computed by using these parameters showed underabundances of about 0.5 dex for calcium and scandium, while heavy elements are all overabundant, from 0.4 dex for iron-peak elements to about 1.4 dex for barium.
In conclusion the Am nature of HD 190165 is confirmed.

Lithium abundance
The lithium abundance in Am stars is a topic that has been discussed in several papers in the recent literature. Burkhart & Coupry (1991) and then Burkhart et al. (2005) concluded that, in general, lithium in Am stars is close to the cosmic value of log NLi/NT ot ≈ −9.04 dex, although a small fraction of them are Li underabundant. Fossati et al. (2007) analysed a sample of eight Am stars, belonging to the Praesepe cluster, in the range of temperature between 7000 K and 8500 K. By using the Lii 6707Å line, they were able to compute abundances that appears to be higher than the cosmic value. Catanzaro & Balona (2012) computed the abundance of lithium in the Am star HD 27411, deriving a value of log NLi/NT ot = −8.42 ± 0.10, in agreement with the values reported by Fossati et al. (2007). In this study we derived the lithium abundances for our Am stars (when possible) and we compared them with those reported in various literature sources for normal Atype stars.
To estimate the lithium abundance we applied the spectral synthesis method to the Lii 6707Å line, taking into account the hyperfine structure (Andersen, Gustafson, & Lambert 1984), as well. Due to the high rotational velocity of some stars, we detected the line and than we were able to compute the relative abundance for only five stars: HD 113878, HD 176843, HD 187254, HD 190165 (see Tab. 3), and HD 71297 (log NLi/NT ot = −8.78 ± 0.11). Lithium abundances for these objects are shown (red filled circles) in Fig. 3 as a function of the effective temperature. For comparison purposes we plotted in the same figures the lithium abundances for various samples of Am stars. In particular we show with cyan filled triangles the results by Burkhart & Coupry (1991) and Burkhart et al. (2005) and with blue filled Figure 3. Lithium abundances plotted as a function of effective temperature. Filled symbols refer to Am stars, in particular circles (red) represent our data, triangles (cyan) are from Burkhart & Coupry (1991) and Burkhart et al. (2005), squared (blue) are from Fossati et al. (2007), and asterisk (magenta) are from Catanzaro & Balona (2012). Opend circles refer to normal A-type stars taken from varius literature sources as outlined in the text. Typical errors are indicated in the bottom right corner of the plot.
squares the data for Am belonging to Praesepe cluster (Fossati et al. 2007).
With the aim of comparing the Lithium abundances in Am and normal A stars, we computed the abundances for a sample of these latter objects in two ways: i) converting the equivalent width of the Lii 6707Å line taken from various sources: Coupry & Burkhart (1992), Glaspey, Pritchet & Stetson (1994), Balachandran, Mallik & Lambert (2011) or ii) measured by us in spectra available on the Elodie archive (Observatoire de Haute Provence). For homogeneity purpose, all the computations have been performed for all the stars by using WIDTH9 (Kurucz & Avrett 1981) applied to AT-LAS9 models 5 (Kurucz 1993a,b). These stars are listed in Tab. 4, together with their effective temperatures, derived by using Strömgren photometry as we described in Sect. 3.1, equivalent widths and Li abundances. The normal A-type stars are shown in Figure 3 with empty circles.
An inspection of Fig. 3 allows us to make some reflections. First, the lithium abundance estimated in our sample of Am stars is on average lower by ≈ 0.2 dex with respect to that measured in the Am stars belonging to Praesepe cluster (Fossati et al. 2007). Second, albeit our targets fall  Glaspey, Pritchet & Stetson (1994) c from Elodie archive in the range of effective temperatures of the so-called Li dip, a region of the diagram T eff − log NLi/NT ot between the temperatures 6600 K to 7600 K, where lithium shows a sudden drop of about 1.6 -1.8 dex (Boesgaard & Tripicco 1986), none of them present abundances lower than the cosmic value (see dotted line in Fig. 3). Third, it appears clear that there is no difference between the peculiar and normal A-type stars. Even the Li dip is present both in Am and in normal stars.

POSITION IN THE HR DIAGRAM
In principle, the stellar parameters log g and log T eff determined in the previous section allow us to estimate the luminosity of the investigated objects. In a previous paper  we accomplished this task by interpolating the tables by Schmidt-Kaler (1982). However, it can be noticed that the space of parameters log g, log L/L⊙, log T eff (spectral type) adopted by Schmidt-Kaler (1982) is rather poorly sampled. To improve this situation we decided to use a different calibration log L/L⊙=log L/L⊙(log g, log T eff ) whose derivation is described in a different pa-  (3) −(0.913 ± 0.014) log g rms = 0.093 dex.
Hence, we estimated the values of log L/L⊙ for the eight stars studied in this paper by inserting in the above equation the spectroscopically derived values for log g, log T eff as reported in Tab. 2. The result of this procedure is listed in Tab. 5 where we report in columns (2) to (4) the log L/L⊙, the distance and the M 0 V , respectively. The last two quantities were derived from the estimate of log L/L⊙ by means of simple algebric passages and the information included in Tab. 1.
As a check for these estimates, we derived the same quantities directly form the parallax measured by Hipparcos for four stars in our sample (see Tab. 1). The results are shown in column (5) to (7) of Tab. 5. A comparison between column (2) and (5) reveals that the two independent log L/L⊙ estimates are in good agreeement within the errors, with the exception of HD 104513, which appears to be too bright if luminosity is estimated by means of Eq. 4. We have already discussed in Sect. 3.3 the possible origin of this discrepancy. In any case, in the following we adopted the parallax-based log L/L⊙ for HD 104513, HD 114839 HD 118660 and HD 190165, whereas for HD 113878 we preferred to adopt the estimate from Eq. 4, given the large error on parallax.
One of the aims of this paper is to try to constrain the locus occupied by the pulsating Am star in the HR diagram. This is done in Fig. 4 where we plotted the eight stars analysed in this paper (log T eff from column (2) of Table 2; log L/L⊙ from columns (2) or (5) of Table 5). In the same figure we added the three pulsating Am stars analysed in our previous works, namely: HD 71297 (after Catanzaro, Ripepi & Bruntt 2013), HD 178327 and HD 183489 (after Balona et al. 2011). Note that for the latter two stars, the value of log L/L⊙ was recalculated. In particular, for HD 183489, we used the Hipparcos parallax value (π=5.91±0.63; van Leeuwen 2007) to estimate its luminosity, obtaining log L/L⊙=1.11±0.09 dex. Unfortunately, Hipparcos did not observe HD 178327. However, this star appears to be a twin of HD 183489, showing exactly the same log g or log T eff and chemical abundance (within the errors, see Catanzaro, Ripepi & Bruntt 2013) of this object. Hence, we decided to assign to HD 178327 the same luminosity of HD 183489, but increasing the error by 50% to Table 5. Luminosities, distances and absolute visual magnitudes obtained from Eq. 4 (columns 2-4) and from Hipparcos parallaxes (columns 5-7). See text for details. allow for the uncertainties in the stellar parameters (i.e. log L/L⊙=1.11±0.14 dex).
To have an idea about the masses and ages of the investigated objects, Fig. 4 shows the evolutionary tracks (solid lines) and the isochrones (dotted lines) for 0.5, 0.7 and 1.0 Gyrs, respectively (the models, calculated for Y=0.273, Z=0.0198, were taken from the BaSTI database 6 ). We also show in the figure the comparison with the edges of the δ Sct (after Breger & Pamyatnykh 1998) and γ Dor (after Warner et al. 2003) instability strips, respectively.
An analysis of the figure shows that only the cooler part of the δ Sct instability strip is occupied by the pulsating Am stars investigated here, whereas no object falls in the region where only γ Dor pulsation is allowed. Only HD 104513 (among the pulsating Am stars) lie in the region where both δ Sct and γ Dor variability are excited. Moreover, all the stars have an age between 0.5, 0.7 and 1.0 Gyrs.
For comparison purposes, Fig. 4 shows with small yellow filled circles the location in the HR diagram of the pulsating Am stars found by the SuperWASP survey (Smalley et al. 2011). An inspection of the figure reveals that our results are in perfect agreement with those obtained by Smalley et al. (2011) on the basis of a larger sample: hot Am stars do not pulsate. This results is also valid for the object observed with very high precision by the Kepler satellite (see Balona et al. 2011). For the physical implication of this finding we refer the reader to the quoted papers.

DISCUSSION AND CONCLUSION
In this work we presented a spectroscopic analysis of a sample of 8 stars classified in literature as to belong to the class of the metallic Am stars. The analysis is based on high resolution spectra obtained at the Telescopio Nazionale Galileo with the SARG spectrograph. For each spectra we obtained fundamental parameters such as effective temperatures, gravities, rotational and radial velocities, and we performed a detailed computation of the chemical pattern, as well. To overcome the problem arising from blending of 6 http://albione.oa-teramo.inaf.it/  Balona et al. 2011). Note that the value of log L/L ⊙ of HD 178327 was artificially increased by 0.02 dex to avoid a complete overlap with HD 183489. Filled and empty circles show pulsating and non-pulsating Am stars, respectively. The empty pentagon refers to a star that is neither Am nor pulsating, whereas the filled triangles represent objects which are pulsating but not Am. Small yellow filled circles show the pulsating Am stars from the SuperWASP survey (Smalley et al. 2011) The red dashed lines shows the δ Sct instability strip by Breger & Pamyatnykh (1998); the blue dotted-dashed lines shows the theoretical edges of the γ Dor instability strip by Warner et al. (2003). The evolutionary tracks (thin solid lines) for the labelled masses as well as the ZAMS (thick solid line), and the isochrones for 0.5, 0.7 and 1.0 Gyrs (dotted lines) are from the BaSTI database.
spectral lines, we applyed the synthesis method by using SYNTHE (Kurucz & Avrett 1981) and ATLAS9 (Kurucz 1993a) codes. The typical errors was about 200 K for T eff , 0.1 dex for log g, and a few km s −1 for the v sin i.
The values of T eff and log g derived here have been used to determine the luminosity of the stars and to place them on the HR diagram.
According to our analysis, we ruled out two stars from the group of the Am stars, namely: HD 114839 and HD 179458. The reasons are different, HD 114839 showed abundances almost solar in conten, while HD 179458 has a chemical pattern far from the solar one, but nevertheless its peculiarity is not the one typical for Am stars.
All the observed stars lie in the δ Sct instability strip next to the red edge, in agreement with Smalley et al. (2011) and Catanzaro & Balona (2012).
In the scenario described by the diffusion models developed by Richer, Michaud, & Turcotte (2000), stars in the range of temperature and age compatible with those of our sample should have underabundances of about 0.1 to 0.3 dex for elements such as C, N, O, Na, Mg, K, and Ca, normal abundance for Si and S, while Al, Ti, Cr, Mn, Fe, and Ni resulted overabundant of about 0.1 to 0.8 dex.
For what that concern lithium, Richer, Michaud, & Turcotte (2000) models predict anomalies of ≈ −0.2 dex with respect the cosmic value. For our stars, in general we obtained abundances almost 0.2 dex over the cosmic value, a result in agreement with the abundances found in the Am star HD 27411 (Catanzaro & Balona 2012) and in the Praesepe cluster (Fossati et al. 2007). In conclusion we measured more lithium than that predict by theory.
Recently, Vick et al. (2010), in the context of the project to explore various macroscopic processes which compete with atomic diffusion in Am/Fm stars, computed a grid of models in which mass-loss has been used instead of turbulence. Those models predict at the side of Li dip, where our objects lie, a smaller anomaly but still not sufficient to explain our observations. As the authors suggested, it is likely that more than one mechanism compete to diffusion, i. e. mass-loss in combination with turbulence, but at the moment is not possible to conclude about one of this possibility.
In any case, our detailed abundance analysis can help theorist in setting more constraints in their diffusion models.