Free Access
Issue
A&A
Volume 640, August 2020
Article Number A40
Number of page(s) 23
Section Catalogs and data
DOI https://doi.org/10.1051/0004-6361/202037750
Published online 10 August 2020

© ESO 2020

1. Introduction

The chemically peculiar (CP) stars of the upper main sequence (spectral types early B to early F) are traditionally characterised by the presence of certain absorption lines of abnormal strength or weakness that indicate peculiar surface abundances (Preston 1974). For most groups of CP stars, current theories ascribe the observed chemical peculiarities to the interplay between radiative levitation and gravitational settling (atomic diffusion) (Michaud 1970; Richer et al. 2000): whereas most elements sink under the force of gravity, those with numerous absorption lines near the local flux maximum are radiatively accelerated towards the surface. Because CP stars are generally slow rotators and boast calm radiative atmospheres, atomic diffusion processes are able to significantly influence the chemical composition of the outer stellar layers.

Following Preston (1974), CP stars are traditionally divided into the following four main groups: CP1 stars (the metallic-line or Am/Fm stars), CP2 stars (the magnetic Bp/Ap stars), CP3 stars (the Mercury-Manganese or HgMn stars), and CP4 stars (the He-weak stars). Although the chemical composition within a group may vary considerably, each group is characterised by a distinct set of peculiarities. The CP1 stars show underabundances of Ca and Sc and overabundances of the iron-peak and heavier elements. CP2 stars exhibit excesses of elements such as Si, Sr, Eu, or the rare-earth elements. The CP3 stars are characterised by enhanced lines of Hg and Mn and other heavy elements, whereas the main characteristic of the CP4 stars is anomalously weak He lines. Further classes of CP stars have been described, such as the He strong stars – early B stars with anomalously strong He lines –, the λ Bootis stars (Murphy & Paunzen 2017), which boast unusually low surface abundances of iron-peak elements, or the barium stars, which are characterised by enhancements of the s-process elements Ba, Sr, Y, and C (Bidelman & Keenan 1951). Generally, with regard to the strength of chemical peculiarities, a continuous transition from normal to peculiar stars is observed (Loden & Sundman 1987).

Most of the CP2 and He-peculiar stars possess stable and globally organised magnetic fields with strengths of up to several tens of kG (Babcock 1947; Aurière et al. 2007), the origin of which is still a matter of some controversy (Moss 2004). However, a body of evidence has been built up that strongly favours the fossil field theory, which states that the magnetic field is a relic of the “frozen-in” interstellar magnetic field (e.g. Braithwaite & Spruit 2004). These stars are often referred to as magnetic chemically peculiar (mCP) stars in the literature – a convention which we will adhere to throughout this paper. The magnetic field affects the diffusion processes in such a way that mCP stars show a non-uniform distribution of chemical elements (chemical spots or belts) on their surfaces, which can be studied in detail via the technique of Doppler imaging (Kochukhov 2017). As the magnetic axis is oblique to the rotation axis (oblique rotator model; Stibbs 1950), mCP stars show strictly periodic light, spectral, and magnetic variations with the rotation period. The photometric variability arises because flux is redistributed in the abundance patches (e.g. Wolff & Wolff 1971; Molnar 1973; Krtička et al. 2013).

The mCP stars show vastly differing abundance patterns. Some of the most peculiar objects belong to this group, such as the extreme lanthanide star HD 51418 (Jones et al. 1974) or Przybylski’s star HD 101065 (Przybylski 1966), which is widely regarded as the most peculiar star known. Excesses of several orders of magnitude are commonly observed in these objects. Morgan (1933) already recognised a relationship between a CP2 star’s temperature and the predominant spectral peculiarities and showed that the CP2 stars can thus be sorted into subgroups. Since then, many authors have proposed corresponding classification schemes with varying levels of detail (cf. the discussions in Wolff 1983; Gray & Corbally 2009). It is generally useful to at least differentiate between the “cool” CP2 stars mostly characterised by Sr, Cr and Eu peculiarities and the “hot” CP2 stars that generally show Si overabundances, although considerable overlap exists.

The mCP stars are important to astrophysics in several respects. Their complex atmospheres lend themselves perfectly to the investigation of such diverse phenomena as atomic diffusion, magnetic fields, stellar rotation and their interplay. They furthermore provide important testing grounds for the evaluation of model atmospheres (Krtička et al. 2009) and, through their characteristic light variability, allow the derivation of rotational periods with great accuracy and comparatively little effort.

The most up-to-date collection of CP stars – the most recent version of the General Catalogue of CP Stars – was published a decade ago (Renson & Manfroid 2009). It contains about 3500 mCP stars or candidates (∼2000 confirmed mCP stars and ∼1500 candidate mCP stars). Since then, several studies have identified new mCP stars on a minor scale (e.g. Hümmerich et al. 2018; Scholz et al. 2019; Sikora et al. 2019) but no large scale spectroscopic surveys have been conducted during the past recent decades that aim specifically at the identification of new mCP stars “en masse”.

The works of Hou et al. (2015) and Qin et al. (2019) warrant special mention as they exploited spectra collected by the Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) of the Chinese Academy of Science. Hou et al. (2015) presented a list of 3537 candidate CP1 stars from LAMOST Data Release (DR) 1. Building on this work, Smalley et al. (2017) investigated pulsational properties versus metallicism in this subgroup of CP stars. Qin et al. (2019) searched for CP1 stars in the low-resolution spectra of LAMOST DR5 and compiled a catalogue of 9372 CP1 stars. They identified CP2 stars as a contaminant and searched for corresponding candidates in their sample of CP1 candidates, identifying 1131 candidate CP2 stars in this process.

Here we present our efforts aimed at identifying new mCP stars using spectra from the publicly available LAMOST DR4, which have led to the discovery of 1002 mCP stars. With this work, we significantly increase the sample size of known Galactic mCP stars, paving the way for future in-depth statistical studies. Spectroscopic data and target selection process are discussed in Sect. 2. Spectral classification workflow and results are detailed in Sect. 3 and discussed, together with other interesting information on our sample of stars, in Sect. 4. We conclude our findings in Sect. 5.

2. Spectroscopic data and target selection

This section contains a description of the employed spectral archive and the MKCLASS code and details the process of target selection.

2.1. The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST)

The LAMOST telescope (Zhao et al. 2012; Cui et al. 2012), also called the Guo Shou Jing1 Telescope, is a reflecting Schmidt telescope located at the Xinglong Observatory in Beijing, China. It boasts an effective aperture of 3.6−4.9 m and a field of view of 5°. Thanks to its unique design, LAMOST is able to take 4000 spectra in a single exposure with spectral resolution R ∼ 1800, limiting magnitude r ∼ 19 mag and wavelength coverage from 3700 to 9000 Å. LAMOST is therefore particularly suited to survey large portions of the sky and is dedicated to a spectral survey of the entire available northern sky. LAMOST data products are released to the public in consecutive data releases and can be accessed via the LAMOST spectral archive2. With about 10 million stellar spectra contained in DR6, the LAMOST archive constitutes a real treasure trove for researchers.

2.2. The MKCLASS code

MKCLASS is a computer program written to classify stellar spectra on the Morgan-Keenan-Kellman (MKK) system (Gray & Corbally 2014). It has been designed to emulate the process of classifying by a human classifier. First, a rough spectral type is assigned, which is then refined by direct comparison with spectra from standard star libraries.

Currently, MKCLASS is able to classify spectra in the violet-green region (3800−5600 Å) in either rectified or flux-calibrated format. Several studies (e.g. Gray & Corbally 2014; Gray et al. 2016; Hümmerich et al. 2018) have shown that, providing input spectra of sufficient signal-to-noise (S/N), the results of MKCLASS compare well with the results of manual classification (precision of 0.6 spectral subclass and half a luminosity class according to Gray & Corbally 2014).

MKCLASS comes with two libraries of MKK standard spectra, which have been acquired with the Gray/Miller (GM) spectrograph on the 0.8 m reflector of the Dark Sky Observatory in North Carolina, USA. libr18 contains rectified spectra in the spectral range from 3800−4600 Å and a resolution of 1.8 Å that were obtained with a 1200 g mm−1 grating. libnor36 boasts flux-calibrated and normalised spectra in the spectral range from 3800−5600 Å and a resolution of 3.6 Å obtained with a 600 g mm−1 grating. MKCLASS allows for the use of additional spectral libraries tailored to the specific needs of the researcher.

An interesting feature of the MKCLASS code is its ability to identify a set of spectral peculiarities, such as found in CP1 and CP2 stars, barium stars, carbon-rich giants etc. For more information on the MKCLASS code, we refer the reader to Gray & Corbally (2014) and the corresponding website3.

2.3. Target selection criteria

To select suitable mCP star candidates, we specifically searched for the presence of the tell-tale 5200 Å depression in the LAMOST DR4 spectra of early-type stars. In the following, we provide background information and detail our selection criteria and the methods employed in the construction of the present sample of stars.

2.3.1. The flux depressions in mCP stars

The first to notice significant flux depressions at 4100 Å, 5200 Å, and 6300 Å in the spectrum of the mCP star HD 221568 was Kodaira (1969). Similar features in the ultraviolet region at 1400 Å, 1750 Å, and 2750 Å were later identified and investigated (Jamar 1977, 1978). It was found that these spectral features solely occur in mCP stars. Khan & Shulyak (2007) showed that Fe is the principal contributor to the 5200 Å depression for the whole range of Teff of mCP stars, while Cr and Si play a role primarily in the low Teff region. Figure 1 shows the 4800 Å to 5700 Å region of the spectra of a non-CP star, a corresponding synthetic spectrum and the newly-identified mCP star LAMOST J025951.09+540337.5 (#784; TYC 3701-157-1), illustrating the 5200 Å depression in the latter object.

thumbnail Fig. 1.

4800 Å to 5700 Å region of (from top to bottom) the non-CP A0 V star LAMOST J194655.00+402559.5 (HD 225785), a synthetic spectrum with Teff = 9750 K, log g = 4.0, [M/H] = 0.0 and a microturbulent velocity of 2 km s−1, and the newly-identified Si-strong mCP star LAMOST J025951.09+540337.5 (#78; TYC 3701-157-1). The position of the characteristic 5200 Å depression and the Si II lines at 5041 Å and 5055/56 Å are indicated. LAMOST spectra have been taken from DR4.

To investigate the flux depression at 5200 Å, Maitzen (1976) introduced the narrow-band three-filter Δa system, which samples the depth of this depression by comparing the flux at the center (5220 Å, g2) with the adjacent regions (5000 Å, g1 and 5500 Å, y) using a band-width of 130 Å for g1 and g2 and 230 Å for the Strömgren y filter. The index was introduced as:

This quantity is slightly dependent on temperature in the sense that it increases towards lower temperatures. Therefore, the intrinsic peculiarity index had to be defined as:

that is, the difference between the individual a value and the a value of non-peculiar stars of the same colour. The locus of the a0 values for non-peculiar objects was termed the normality line. Virtually all mCP stars have positive Δa values up to +75 mmag (Paunzen et al. 2005). Only extreme cases of CP1 and CP3 stars can exhibit marginally positive Δa values. Be/Ae, B[e] and λ Bootis stars exhibit significant negative values. In summary, it has been shown that the Δa system is an efficient and reliable means of identifying mCP stars.

2.3.2. Sample selection

In the present study, we concentrated on the publicly-available spectra from LAMOST DR4 (Zhao et al. 2012; Luo et al. 2018). As first step, the complete catalogue was cross-matched with the Gaia DR2 catalogue (Gaia Collaboration 2018). To identify suitable targets, we exploited the G versus (BP − RP) diagram to set a corresponding limit on the investigated spectral type range (hotter than mid F, i.e. (BP − RP) < 0.45 mag). From the remaining objects, apparent supergiants were excluded. As this approach is bound to miss highly-reddened hot objects, we searched the spectral types listed in the DR4 VizieR online catalogue5 (Luo et al. 2018) for additional early-type (B-, A-, and F-type) targets, which were also included into the analysis.

Only spectra with a S/N of more than 50 in the Sloan g band were considered for further analysis. This cut was deemed necessary because a lower S/N renders the detection of mCP star features difficult (Paunzen et al. 2011). If more than one spectrum was available for a single object, only the spectrum with the highest Sloan g band S/N was included into the analysis.

From the remaining spectra, suitable candidates were selected by the presence of the tell-tale 5200 Å depression. To calculate synthetic Δa values, all spectra were normalised to the flux at 4030 Å. This guarantees that the large absolute flux differences introduced by the apparent visual magnitude do not cause any numerical biases in the final magnitudes. The filter curves of g1, g2, and y as defined in Kupka et al. (2003) were then folded with the spectra and the corresponding magnitudes calculated. All objects with a significant positive Δa index were visually inspected for the presence of a 5200 Å depression in order to sort out glitches in the spectra or contamination by other objects such as cool stars with strong features in the 5200 Å range.

Figure 2 illustrates this process by providing sample LAMOST DR4 spectra of mCP stars showing 5200 Å flux depressions of various strengths. Also shown is the “impostor” LAMOST J001159.88+435908.5 (GSC 02794−00977). This object is actually a mid to late K star whose 5200 Å region is dominated by absorption lines of the Mg I triplet at 5167 Å, 5173 Å, and 5184 Å, which leads to a significantly positive Δa value and highlights the need for setting a limit on the investigated spectral type range via the above described colour-colour cut.

thumbnail Fig. 2.

4700 Å to 5700 Å region of (from top to bottom) the LAMOST DR4 spectra of the mCP stars LAMOST J034458.31+464848.7 (#139; TYC 3313-1279-1), LAMOST J040642.34+454640.8 (#180; HD 25706), LAMOST J072118.92+223422.7 (#792; TYC 1909-1687-1), and the late-type “impostor” LAMOST J001159.88+435908.5 (GSC 02794−00977). The position of the characteristic 5200 Å depression is indicated.

In this way, a list of 1002 mCP star candidates was compiled. This collection of stars is referred to in the following as the “final sample”. We here emphasise that our sample is obviously biased towards mCP stars with conspicuous flux depressions at 5200 Å. However, not all mCP stars show significant 5200 Å depressions, in particular in low-resolution spectra, and such objects will have been missed by the imposed selection criteria. On the other hand, early-type stars with significant 5200 Å depressions are nearly always mCP stars; therefore, the chosen approach should be well suited to collecting a pure sample of mCP stars.

3. Spectral classification

Spectral classification is an important tool in astrophysics, which allows for the easy identification of astrophysically interesting objects. Furthermore, it enables to place stars in the Hertzsprung-Russell diagram, thus enabling the derivation of physical parameters. However, in the era of large survey projects such as LAMOST, RAVE, or SDSS/SEGUE, which produce a multitude of stellar spectra, human manual classification is no longer able to cope with the amount of data and the need for automatic classification has arisen. For the present study, we chose to employ a modified version of the MKCLASS code (cf. Sect. 2.2) that has been specifically tailored to the needs of our project. More details are provided in this section, which details the spectral classification workflow.

3.1. Spectral classification with the MKCLASS code

As deduced from an investigation of the program code, the current version of the MKCLASS code (v1.07) is able to identify the following spectral features, which are important in the detection of CP2 stars: the blend at 4077 Å (which may contain contributions from Si II 4076 Å, Sr II 4077 Å, and Cr II 4077 Å), the blend at 4130 Å (due to enhanced Si II 4128/30 Å and/or Eu II 4130 Å), and the Eu II 4205 Å line. On a significant detection of these features, the following output is created: “Sr” (4077 Å), “Si” (4130 Å), “Eu” (4205 Å). As other elements besides Sr and Si contribute to the blends at 4077 Å and 4130 Å, the output may be misleading in some cases. Nevertheless, this allows a robust detection of mCP stars (e.g. Hümmerich et al. 2018) and is a good starting point for further investigations.

To suit the special needs of our project, which is solely concerned with the identification and classification of mCP stars among a sample of early-type candidate stars, we opted to refine the MKCLASS peculiarity identification routine. The code was therefore altered to probe several additional lines, with the advantage that the new version is now able to more robustly identify traditional mCP star peculiarities. In addition, we enabled the identification of Cr peculiarities and, to some extent, He peculiarities, which are relevant to the classification of mCP stars. The choice of lines was dictated by the resolution and quality of the input material (i.e. LAMOST DR4 spectra), in particular concerning the availability of neighbouring continuum flux to probe a certain line and the numerous line blends due to the low resolution. We therefore stress that the resulting version of the MKCLASS code, which is referred to hereafter as MKCLASS_mCP, has been created specifically for identifying and classifying mCP stars in LAMOST low-resolution spectra. Applying the code to spectra of other resolutions will require a corresponding update of the peculiarity classification routine and, perhaps, an update and enlargement of the employed standard star libraries (see below). Table 1 lists the lines and blends identified by MKCLASS_mCP, as well as the spectral range in which the corresponding features are searched for. We note that at the resolution of the employed LAMOST spectra, all these lines are, to some extent, blended with other absorption lines. Nevertheless, the listed ions generally constitute the main contributors to these blends in mCP stars.

Table 1.

Absorption lines and blends identified by the modified version of the MKCLASS code (MKCLASS_mCP) and used in the identification and classification of mCP stars in the present study.

A sample output of MKCLASS_mCP is provided in column five of Table 2. Further information on the interpretation of this output is provided below; a discussion of the accuracy of the derived classifications is provided in Sect. 3.2.

Table 2.

Spectral classifications derived by manual classification and the MKCLASS_mCP code.

We note that the Si II line at 5041 Å increases significantly with temperature type; therefore, different detection limits were applied depending on the investigated temperature range. Furthermore, the Si II 6347/71 Å lines were found to show a significant scatter in MKK standard stars. However, at the resolution of the LAMOST spectra, the red Si II lines are a readily detectable and outstanding feature of strong Si stars and contribute significantly to an unambiguous detection of Si peculiarity, in particular as the corresponding lines in the blue-violet region are difficult to detect because of continuum flux issues (3856/62 Å) and blending issues (4076 Å and 4128/30 Å).

As has been desribed in Sect. 2.2, MKCLASS is a computer program that emulates the workflow of a human classifier in the traditional MKK spectral classification process, which involves comparing the input spectrum to a set of standard star spectra. It is therefore imperative to carefully select standard star libraries that match the input spectra in resolution and calibrationwise. MKCLASS comes with the two standard libraries libr18 and libnor36 (cf. Sect. 2.2), which, unfortunately, do not match the spectral resolution of the LAMOST low-resolution spectra. Furthermore, as far as we are aware of, a standard library based on LAMOST spectra does not exist.

In the framework of the LAMOST-Kepler project, Gray et al. (2016) presented MKK spectral classifications of more than 100 000 LAMOST spectra of about 80 000 objects situated in the Kepler field. The authors solved the above-mentioned issue by degrading the flux-calibrated LAMOST spectra to a resolution of R ∼ 1100 and truncating them to the 3800−5600 Å region in order to enable the use of MKCLASS with the flux-calibrated libnor36 library (cf. also the MKCLASS documentation). Here we follow their approach, but we also searched for alternative methods, as degrading spectra obviously results in loss of information. This, however, is detrimental to the identification of the often subtle chemical peculiarities of our sample stars.

We synthesised a library of spectra using the program SPECTRUM6 (Gray & Corbally 1994) and ATLAS9 model atmospheres (Castelli & Kurucz 2003), which is referred to hereafter as the libsynth library. Only dwarf spectra (luminosity class V) were synthesised because no models were available to reproduce the subtle differences in surface gravity among early-type giant stars. Furthermore, we collected a set of LAMOST standard star spectra (the liblamost library) by carefully choosing a grid of suitable high S/N spectra from the list presented by Gray et al. (2016). Only dwarf and giant spectra were chosen, as not enough suitable spectra of higher luminosity class objects were available to build up a corresponding grid. We note, however, that it has been well confirmed that mCP stars are generally main-sequence objects (cf. Netopil et al. 2017, and references therein, and Sects. 1, 3.2, and 4.2); therefore, we do not expect that the absence of spectra of higher luminosity class objects in these two libraries significantly affects our results – in particular as the libr18 and libnor36 libraries boast corresponding spectra of all luminosity classes.

The stars and corresponding LAMOST spectrum identifiers of the liblamost library are given in Table C.1. Although it contains a very suitable grid of dwarf spectra, we note that the liblamost library is far from being a perfect set of standard star spectra (a corresponding quality flag that estimates the suitability of a spectrum as a standard is also provided in Table C.1). It contains a fast rotator and some spectra show “impurities” not expected in MKK standards. These shortcomings will lead to increased uncertainties in the derivation of the temperature and luminosity classes. However, the library consists of spectra obtained with the same instrument – and hence, importantly, of the same resolution as our input spectra – and was extremely valuable in the identification of chemical peculiarities. We nevertheless explicitly caution against using the liblamost library as a standard star library out of the context of the present investigation. We also emphasise the need for a standard star library based on LAMOST low-resolution spectra, which will greatly facilitate further research based on this excellent data source. The liblamost library may very well serve as a starting point; this, however, is beyond the scope of the present investigation.

In the following, an overview over the employed spectral libraries is presented. As an example, Fig. 3 illustrates the F0V standard spectra from the corresponding libraries. We note that the libsynth and liblamost libraries only contain spectra in the approximate spectral type range of our sample stars.

thumbnail Fig. 3.

Blue-violet region of (from top to bottom) the F0 V standard spectra from the liblamost, libsynth, libnor36, and libr18 standard libraries.

  • libr18: spectral range from 3800–4600 Å, resolution of 1.8 Å (R ∼ 2200), normalised and rectified spectra; all luminosity classes (Ia-V)

  • libnor36: spectral range from 3800–5600 Å, resolution of 3.6 Å (R ∼ 1100), flux-calibrated and normalised spectra; all luminosity classes (Ia-V)

  • libsynth: spectral range from 3800–4600 Å, smoothed to a resolution of 3.0 Å and an output spacing of 0.5 Å, flux-calibrated and normalised synthetic spectra; only dwarf spectra (luminosity class V); spectral types B5 to F5

  • liblamost: spectral range from 3800–5600 Å, resolution R ∼ 1800, flux-calibrated and normalised spectra; only dwarf and giant spectra (luminosity classes V and III); spectral types B3 to G0

mCP stars may exhibit peculiar Ca II K profiles and line strengths (Faraggiana 1987; Gray & Corbally 2009; Ghazaryan et al. 2018) as well as generally enhanced metal-lines. Several authors have therefore adopted a notation that indicates separate spectral types as derived from the Ca II K line (the k-type), the hydrogen lines (the h-type), and the general strength of the metal-lines (the m-type), in the same way as is usually done for CP1 stars. The MKCLASS code also assigns k/h/m-types in cases where discrepant spectral types are derived from the corresponding features. As mCP stars are prone to exhibiting marked Ca and He deficiencies (e.g. Gray & Corbally 2009; Ghazaryan et al. 2018) and often enhanced metal-lines, the hydrogen-line profile is a better indicator of the actual effective temperature (Gray & Corbally 2009). Where they have been derived by the code (or by manual classification), k/h/m types are listed in the present study.

For most stars, only minor differences in temperature and luminosity types were found between the results from the different spectral libraries. In cases where the same spectral type was derived more than once, the most common spectral type was adopted (cf. Table 2). If no common classifications existed, spectral types were favoured in the order liblamost > libsynth > libnor36 > libr18. In the case of strong differences between the derived classifications, the corresponding spectra were visually inspected and the best fitting type was chosen.

To determine significant chemical peculiarities from the “raw” MKCLASS_mCP output, the number of detections Ndet of the peculiar strength of a given line with the different standard star libraries (0 ≤ Ndet ≤ 4) was counted, which provides an estimation of significance. Obviously, Ndet = 4 is a very robust detection; Ndet < 2 detections, on the other hand, have to be viewed with caution. Furthermore, we required that the identification of overabundances cannot be based on a single strong line (with the exception of the Cr II 4172 Å line in the identification of a Cr peculiarity; see below).

To come up with an approach that forms a compromise between spurious detections and overly high thresholds required a good amount of experimentation and experience in comparing the results of manual and automatic classification. Table 3 lists the conditions found to work best with the input material and our methodological approach. Ndet(λ) is the number of detections of a peculiarly strong line at the specified wavelength (Å). For instance, a Cr peculiarity was flagged when (a) a strong Cr II 4172 Å line was detected with a least two different standard star libraries or (b) a strong Cr II 3866 Å line and a strong Cr II 4172 Å line were detected at least once or (c) a strong blend at 4077 Å was detected at least twice and a strong Cr II 4172 Å line was detected at least once.

Table 3.

Conditions employed to flag the presence of an overabundance from the “raw” MKCLASS_mCP output.

Following the conventions of the MKK system, the peculiarity types “Si”, “Cr”, “Sr”, and “Eu” were flagged according to the conditions given in Table 3 and attached to the temperature and luminosity types in the final spectral classification. Several stars in our sample that were not assigned any of the above mentioned peculiarity types nevertheless show strong blends at 4077 Å and/or 4130 Å. In these cases, we decided to add the non-standard suffixes “bl4077” and “bl4130” to the derived spectral types (e.g. “B8 IV bl4130”) if the corresponding blends had been detected at least twice. In these objects, apart from the strong blends, the peculiarities are either too subtle to have passed our significance criteria, no other significant features are present, or the code failed to identify them for some reason. Manual classification is necessary to throw more light on this matter (cf. Sect. 4.4).

In addition, we opted to probe the He I lines at 4009 Å, 4026 Å, 4144 Å, and 4387 Å to identify CP2 stars with weak He I lines and He-peculiar objects. The corresponding detection thresholds for all four standard star libraries were determined using the He lines of 626 apparently chemically-normal B stars with spectra boasting S/N > 100. The number of detections as “weak” or “strong” of the He lines with the different standard star libraries was counted and the results across all lines and libraries were added up to yield Ndet(He-wk) and Ndet(He-st). He-weakness and He-overabundance were assumed when Ndet(He-wk) > 2 and Ndet(He-st) > 2, respectively. In this way, we identified 55 mCP stars with weak He I lines and three mCP stars with apparently strong He I lines. As expected, these are mostly B7−B9 Si CP2 stars, which are notorious for their weak He lines, and mid-B type stars (likely He-peculiar objects). Interestingly, in three mid-B type stars, both weak and strong He I lines were identified, which strongly suggests He peculiarity. The corresponding suffixes “He-wk” and “He-st” were added to the derived spectral types. Several He-peculiar objects are discussed in Sect. 4.6.

In this way, peculiarities were identified in all but 36 stars from our sample, which highlights the efficiency of the chosen approach. The (mostly low S/N) spectra of the remaining objects were investigated manually and searched for the presence of chemical peculiarities. Most of these objects show subtle or complicated peculiarities that failed to meet the imposed significance criteria. Corresponding peculiarity types were manually added to the final spectral types. The remaining objects are a “mixed bag” containing stars with enhanced metal-lines and strong flux depressions that nevertheless lack the traditional Si, Cr, Sr, Eu peculiarities and several He-peculiar objects. Appropriate comments were added to the tables, in which manually-derived peculiarity types and all further additions that are not directly based on the MKCLASS_mCP output are highlighted.

3.2. Evaluation

As a test of the validity of our approach, we manually classified a sample of ten randomly chosen stars and compared our results with the final spectral types derived in the described manner from the MKCLASS_mCP output, which, for convenience, are termed hereafter the “MKCLASS final types”. In addition, we visually inspected the spectra of about 100 further stars to check for the presence of peculiarly strong lines and evaluate the reliability of the classifications. The results from the manual classification are shown in Table 2 and highlight the good agreement between the manually- and automatically-derived (hydrogen-line) temperature types. In general, we estimate the uncertainty of the derived temperature types to be ±1 subclass. This, however, increases to about ±2 subclasses towards later and more peculiar mCP stars for which the classification is notoriously difficult and, for the most extreme objects, unreliable. Figures 4 and 5 showcase, respectively, three “hot” mCP stars and three “cool” mCP stars, which have been newly identified as such in the present study. MKCLASS final types and, where available, manual types from Table 2 are indicated.

thumbnail Fig. 4.

Showcase of three newly identified “hot” mCP stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J035046.03+363648.2 (#151; Gaia DR2 220081859486642816), LAMOST J195631.74+253407.8 (#929; Gaia DR2 2026771741029840128), and LAMOST J062529.84−032411.9 (#576; TYC 4789-2924-1). MKCLASS final types and, where available, manual types derived in the present study are indicated. Some prominent lines of interest are identified. The asterisk marks the position of a “glitch” in the spectrum of LAMOST J062529.84−032411.9.

thumbnail Fig. 5.

Showcase of three newly identified “cool” mCP stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J034854.70+521413.1 (#150; Gaia DR2 251609324623302400), LAMOST J052816.11−063820.1 (#344; TYC 4765-708-1), and LAMOST J062221.82+595613.0 (#561; TYC 3776-269-1). MKCLASS final types and, where available, manual types derived in the present study are indicated. Some prominent lines of interest are identified.

There is also a generally good agreement in regard to the derived peculiarity types. However, with our approach, we obviously missed the presence of peculiarities in several objects (Table 2). A further investigation of these stars shows that this has been mostly due to either the imposed significance criteria, weak or complicated peculiarities, or the absence of continuum flux to probe certain lines (or a combination thereof). A good case in point is LAMOST J065647.94+242958.8 (#732; TYC 1898-1408-1), whose blue-violet spectrum is shown in the upper panel of Fig. 6. It shows enhanced Si II lines at 3856/62 Å, 4128/31 Å, and 4200 Å. Furthermore, strong Cr II lines are present at 3866 Å, 4111 Å, 4172 Å, 4559 Å, and 4588 Å. The blend at around 4077 Å is notoriously difficult to interpret and can contain contributions from Si II, Cr II, and Sr II. The strong Sr II line at 4216 Å, however, indicates the presence of a Sr peculiarity. Consequently, we have classified this star as A0 V SiSrCr. The MKCLASS_mCP code missed the rather subtle Si and Cr peculiarities (MKCLASS final type: B9.5 V Sr).

thumbnail Fig. 6.

Upper panel: comparison of the blue-violet spectra of the mCP star LAMOST J065647.94+242958.8 (#732; TYC 1898-1408-1; manual type: A0 V SiSrCr; MKCLASS final type: B9.5 V Sr) to the liblamost A0 V standard spectrum. Lower panel: comparison of the blue-violet spectra of the mCP star LAMOST J052816.11−063820.1 (#344; TYC 4765-708-1; manual type: kA1hA9mA9 SrCrEu; MKCLASS final type: kA1hA8mA9 SrEu) to the liblamost F0 V standard spectrum. Some prominent lines of interest are indicated. We note the weak Ca II K line with a peculiar profile and the unusual profile of the Hϵ line in LAMOST J052816.11−063820.1.

The bottom panel of Fig. 6 illustrates the case of LAMOST J052816.11−063820.1 (#344; TYC 4765-708-1; also shown in Fig. 5), a cool CP2 star that may serve as a warning and an example illustrating the difficulty of classifying the more extreme mCP stars, which may possess peculiarities that render their spectra difficult to match to any standard star (Gray & Corbally 2009). The star exhibits a weak Ca II K line with a peculiar profile. While the hydrogen-line profile points to a late A-type star, the broad K line is that of an early A-type star and reasonably matched by that of an A1 V standard. As CP2 stars are prone to exhibiting marked Ca deficiencies (e.g. Gray & Corbally 2009; Ghazaryan et al. 2018), the hydrogen-line profile is a better indicator of the actual effective temperature than the K line strength (Gray & Corbally 2009), hence LAMOST J052816.11−063820.1 is obviously a late-type mCP star. However, while the Hγ, Hδ, H8, and H9 lines are reasonably well matched by that of an A9 V star, the Hϵ line exhibits an unusual profile, which makes it difficult to match the hydrogen-line profile to that of any standard star. We also note the unusually weak Mg II line at 4481 Å. The MKCLASS_mCP code duly assigned final k/h/m-types of A1, A8, and A9; we prefer k/h/m-types of A1, A9, and A9.

The spectrum of LAMOST J052816.11−063820.1 shows strong blends at 4077 Å and 4130 Å, which – judging from the strong lines at 4216 Å (Sr II), 4205 Å (Eu II), and 4435 Å (Eu II) – are mainly caused by overabundances of Sr and Eu. The strong Cr II lines at 3866 Å and 4111 Å (the bump in the red wing of Hδ) indicate a Cr overabundance; a corresponding enhanced line at 4172 Å, however, is notably absent. We have classified this star as kA1hA9mA9 V SrCrEu. The MKCLASS_mCP code duly identified the main peculiarities (MKCLASS final type: kA1hA8mA9 SrEu).

The examples show that the peculiarity types derived in the present investigation are in many cases not exhaustive but rather denote the main peculiarities present. They are still very useful for first orientation and an excellent starting point for more detailed investigations; they are furthermore suited to statistical studies (cf. Sect. 4.4).

Some cautionary words are necessary in regard to luminosity classification. It is well known that problems with the luminosity classification may arise by the confusion of luminosity criteria and mCP star characteristica or peculiarities. In the early A-type stars, luminosity classification is primarily based on hydrogen-line profiles. The mCP stars, in general, are slow rotators that display narrow lines and hydrogen-line profiles that are easily misinterpreted as belonging to stars of higher luminosity. Indeed, the hydrogen-line profiles of many late B- and early A-type Si stars of our sample are best matched by standards of luminosity class III although there is no further indication that these star are in fact giant stars. Additional confusion may arise due to peculiarly strong lines that are also used in luminosity classification. Si II lines, for example, are enhanced in giants and supergiants as well as in several types of mCP stars, which might lead to corresponding misclassifications (Loden & Sundman 1989). This holds especially true for classifications based on photographic plates or low S/N spectra. In regard to CP1 stars, the term “anomalous luminosity effect” has been coined, which describes the perplexing situation that luminosity criteria from different regions of the spectrum indicate different luminosities. This also applies at least partly to mCP stars, which may show strong general enhancements of metal-lines in their spectra that are reminiscent of much cooler stars. Furthermore, while obtaining the hydrogen-line type is fairly straightforward for most mCP stars, the more peculiar objects show distorted atmospheres and unusual and peculiar hydrogen-line profiles that may not match any standard star (Gray & Corbally 2009), which may result in classification errors. Abnormal hydrogen-line profiles are especially observed in cool mCP stars (Kochukhov et al. 2002).

These issues also impact automatic classification routines such as the MKCLASS code and will surely be at the root of the high luminosity classification of many of our sample stars, which should be regarded with caution. It has been well confirmed that mCP stars are generally main-sequence objects (e.g. Netopil et al. 2017 and references therein) and the results from the colour-magnitude diagram (CMD) (Sect. 4.2) fully support this finding. A detailed analysis of this issue is necessary but beyond the scope of the present investigation.

At this point, we would like to also recall that at least half of the mCP stars are spectroscopic variables, that is, the observed line strengths may vary considerably over the rotation period (e.g. Gray & Corbally 2009), which should always be kept in mind when working with mCP star spectra.

Table A.1 presents the MKCLASS final types along with essential data for our sample stars.

4. Discussion

This section discusses properties of our final sample such as magnitude distribution, distances from the Sun, evolutionary status, and their distribution in space, and discusses interesting objects such as the eclipsing binary system LAMOST J034306.74+495240.7.

4.1. Magnitudes and distances from the Sun

In Fig. 7, we present the histograms of the G magnitudes and the distances from the Sun of our sample stars. The magnitude distribution peaks between 11th and 12th magnitude, which corresponds to a distance of about 1 kpc for the investigated range of spectral types. While our sample contains only a few new mCP stars within 500 pc around the Sun, there is a significant number of objects beyond 2 kpc, which might help to shed more light on the Galactic radial metallicity gradient (Netopil et al. 2016) and its influence on the formation and evolution of CP stars.

thumbnail Fig. 7.

Histograms of the G magnitudes (upper panel) and distances from the Sun (lower panel). For the construction of the latter, only stars with absolute parallax errors less than 25% were employed.

In summary, our sample is a perfect extension to the mCP stars listed in the catalogue of Renson & Manfroid (2009), which are on the average closer and brighter, peaking at 9th magnitude.

4.2. Evolutionary status

In the following sections, we investigate the evolutionary status of our sample stars in the (BP − RP)0 versus MG, 0 and mass versus fractional age on the main sequence spaces. We caution that, due to the imposed selection criteria (cf. Sect. 2.3.2), our sample is biased towards stars showing a conspicuous 5200 Å flux depression in the low-resolution LAMOST spectra. Therefore, our results, while being based on a statistically significant sample size, have to be viewed with caution and their general validity needs to be tested by a more extended sample selected via different methodological approaches.

4.2.1. Colour-magnitude diagram

To investigate the astrophysical properties of our sample stars in a CMD, we employed the homogeneous Gaia DR2 photometry from Arenou et al. (2018). Most of our sample stars are situated within the Galactic disk farther than 500 pc from the Sun; therefore, interstellar reddening (absorption) cannot be neglected. Because hardly any objects have Strömgren-Crawford indices available (Paunzen 2015), we relied on the published reddening map by Green et al. (2018). To interpolate within this map, parallaxes were directly converted to distances. To limit the error of the absorption values to 0.1 mag, only objects with relative parallax errors of at most 25% (942 in total) were used. The transformation of the reddening values was performed using the relations:

(1)

These relations already take into account the conversion to extinction in different bands using the coefficients as listed in Green et al. (2018).

In Fig. 8, we present the CMD of our sample stars together with PARSEC isochrones (Bressan et al. 2012) for solar metallicity [Z] = 0.020. We favour this value because it has been shown to be consistent with recent results of Helioseismology (Vagnozzi 2019). Also included is the reddening vector according to an uncertainty of 0.1 mag for E(B − V). About 20 stars are situated below the zero-age main sequence (ZAMS) to such an extent that it cannot be explained by errors in the reddening estimation. Inconsistent photometry or binarity might possibly have led to the observed positions but, with the available data, we are unable to shed more light on this matter.

thumbnail Fig. 8.

(BP − RP)0 versus MG, 0 diagram of our sample stars, together with PARSEC isochrones for solar metallicity [Z] = 0.020 (listed are the logarithmic ages). The arrow indicates the reddening vector for the maximum expected error due to the employed reddening map and the parallax error.

The stars LAMOST J061609.42+265703.2 (#537; MG, 0 = +7.78 mag, (BP–RP)0 = +0.840 mag), LAMOST J064757.48 + 105648.2 (#678; MG, 0 = +7.75 mag, (BP − RP)0 = +1.721 mag), and LAMOST J202943.73+384756.6 (#943; MG, 0 = +3.61 mag, (BP − RP)0 = +1.180 mag) are not plotted in the CMD because they lie outside the chosen boundaries. The available spectra clearly confirm that they are mCP stars. We double-checked the identifications in the Gaia DR2 and LAMOST catalogues and searched for nearby objects on the sky that might have influenced the photometry, albeit with negative results. In the case of LAMOST J202943.73+384756.6, we strongly suspect that binarity might be at the root of the observed outlying position. Its spectrum has an unusual black-body curve, and its spectral energy distribution (SED) looks like the superposition of two objects, with a clearly visible infrared excess. However, we are unable to explain the reasons behind the observed inconsistent photometry for the other two stars.

From the distribution of stars in Fig. 8, we conclude that the majority of our sample stars is between 100 Myr and 1 Gyr old. There are only a few very young stars and an accumulation of objects older than 300 Myr.

The here employed isochrones have been calculated for stars of standard composition, that is, chemically normal stars, assuming solar metallicity [Z] = 0.020. As the chemical bulk composition of mCP stars is unknown, however, the choice of the right chemical composition for the theoretical tracks remains an open question (cf. e.g. the discussion in Bagnulo et al. 2006). The main question is whether the apparent overabundances encountered at the surface are representative for the whole stellar interior. If diffusion is assumed as the main mechanism (in line with most theoretical studies), the overall abundance will be close to solar because corresponding underabundances are expected in the stellar interior. All current isochrone calculations are based on assuming a [Z] value for the whole star; it is currently not possible to consider different [Z] values for different layers of the stellar atmosphere. A detailed discussion of the influence of [Z] on the determination of age on the main sequence is provided in the following section.

4.2.2. Mass versus age on the main sequence

To examine the evolutionary status of our sample stars in more detail, we investigated the distribution of mass (M) versus age on the main sequence. Age on the main sequence (τ) is here defined as the fraction of life spent on the main sequence, with the ZAMS corresponding to τ = 0% and the terminal-age main sequence (TAMS) to τ = 100%. Only objects with absolute parallax errors less than 25% were considered in this process. We again used PARSEC isochrones (Bressan et al. 2012) for solar metallicity [Z] = 0.020 and 7.0 < log t < 10.0 (step size: 0.025). Bearing in mind the observational errors, the chosen step size is sufficient not to run into numerical inaccuracies due to the grid being too coarse. We did not interpolate within the grid but always selected the point with minimal Euclidean distance to the observed value in the (BP − RP)0 versus MG, 0 space. Only grid points representing the main sequence (flagged “1” in the corresponding isochrones) were used. As next step, we discarded all data points with a distance of more than 0.05 mag between observed value and theoretical grid point, which guarantees the exclusion of points below the ZAMS and above the TAMS. In this way, masses and ages were derived for 903 sample stars fulfilling the imposed accuracy criteria.

As final step, the lifetime on the main sequence was calculated for a given mass using the upper envelope of the isochrone grid. With this parameter, the fractional lifetime of a star on the main sequence can be easily calculated. To compute upper and lower limits for τ and M, the full error ellipse was taken into account. This procedure is described in more detail in Kochukhov & Bagnulo (2006).

Table B.1 lists the derived masses and fractional ages on the main sequence for solar metallicity [Z] = 0.020, which are graphically represented in Fig. 9. The density plot in the upper panel clearly shows that there are only very few stars in our sample younger than τ < 20%. Most stars have a relative age of τ ≥ 60%.

thumbnail Fig. 9.

Mass versus fractional age on the main sequence (τ) distribution assuming solar metallicity [Z] = 0.020 for the 903 sample stars fulfilling the imposed accuracy criteria. Upper panel: density plot for masses up to 4 M. The position of the spectral types has been based on the information given in Pecaut & Mamajek (2013).

To check the reliability of our results, we have investigated the error distribution of the age and mass estimates in detail. This analysis was done for solar metallicity. We have investigated the influence of the assumed metallicity in a second step. For this, masses were binned in 0.2 M and ages in 10% intervals. Sizes have been chosen to guarantee the availability of a significant number of data points in each bin. Figure 10 illustrates the corresponding histograms. The absolute errors of the masses increase linearly until M = 3.4 M and then flatten out, which means that the relative error stays constant over the whole investigated mass range. The situation is different for the derived ages; up to τ ≤ 90%, absolute errors remain almost constant. Relative errors obtained for younger stars, therefore, are significantly larger than for old ones. This, however, does not impact our conclusions (see below). The significant increase of the errors for the last age bin is due to the higher density of isochrones in this region; furthermore, taking the error ellipse into account, some stars may be located above the TAMS.

thumbnail Fig. 10.

Distributions of errors for the derived fractional ages on the main sequence (τ; upper panel) and masses (lower panel) assuming solar metallicity [Z] = 0.020.

In order to evaluate the effect of the chosen metallicity on our results, we have investigated the [Z] distribution for CP2 stars from the Ghazaryan et al. (2018) catalogue with a least three measurements of C, N, O, and S. These light elements were chosen because they contribute the most to the derived [Z] values. They appear significantly underabundant in most CP2 stars, which therefore show lower [Z] values than chemically normal stars (Fig. 11). For most objects, we find [Z] values in the range from about 0.008 to 0.060. From the reference source, we were not able to estimate the errors of the derived [Z] values, which mainly depend on the errors of the individual abundance determinations; these, however, are mostly not available.

thumbnail Fig. 11.

Distribution of [Z] values for the CP2 stars from the Ghazaryan et al. (2018) catalogue with a least three measurements of C, N, O and S.

We emphasise that all available isochrones use scaled abundances according to the abundance pattern of the Sun. Whether this approximation can also be applied to CP stars remains at the present time unknown (cf. Fig. 1 of Ghazaryan et al. 2018). On the basis of open cluster members, Bagnulo et al. (2006) investigated the influence of the overall metallicity on the error of the age determination for CP2 stars, using metallicities of [Z] = 0.020 (solar) and [Z] = 0.008, as derived from the corresponding host clusters. As clusters with [Z] > 0.020 are very rare in the Milky Way (Netopil et al. 2016), no isochrones for metallicities exceeding solar metallicity were considered in their study.

When estimating τ, we have to consider two effects, which are illustrated in Fig. 12. As is well known, lines of constant τ are not distributed equally across the main sequence (Fig. 12, upper panel). For a constant value of (BP − RP)0, they are much tighter in terms of MG, 0 for the first ∼70% of the main-sequence lifetime. The total interval of MG, 0 from ZAMS to TAMS spans about 2.6 mag, whereas the intervals covered to τ = 25% and τ = 50% amount to only 0.3 mag and 0.7 mag, respectively.

thumbnail Fig. 12.

Lines of constant fractional ages on the main sequence (τ) for solar metallicity [Z] = 0.020 (upper panel). Lower panel: positions of the ZAMS and TAMS for isochrones with [Z] = 0.008, 0.020, and 0.060. Values have been chosen to cover the main range of [Z] values found for CP2 stars (Fig. 11).

The lower panel of Fig. 12 explores the impact of isochrones of different metallicity on the positions of the ZAMS and TAMS. The magnitude differences between the positions of the ZAMS are nearly constant and amount to 0.4 mag, which corresponds to an age difference of about 30% at the ZAMS. This means that a ZAMS star of solar metallicity ([Z] = 0.020) would have already spent 30% of its main-sequence lifetime for [Z] = 0.008 but would be situated 0.4 mag below the ZAMS for [Z] = 0.060. This illustrates the dilemma ellicited by the lack of knowledge of the overall metallicity of mCP stars and the resulting loss of accuracy, as clearly demonstrated by Bagnulo et al. (2006). The uncertainty is largest for stars near the ZAMS and has to be considered together with the distribution of errors for a given distinct metallicity (Fig. 10). However, because individual [Z] values and their errors are unavailable for our sample stars, we are not able to provide reliable estimations of the contribution to the computed errors.

Figure 13 illustrates the mass versus τ distributions for all three investigated isochrones. The majority of our sample stars is situated in the rather narrow spectral range from B8 to A0 (cf. Sect. 4.4). However, there is also a lesser but significant amount of stars with spectral types between A5 and F0. Any calibrated mass distribution should represent these results to some extent. The lower-metallicity isochrone ([Z] = 0.008) yields mainly old (τ > 75%) and cooler (later than spectral type A0) stars; no young stars are present. On the other hand, adopting the isochrone for [Z] = 0.060 results in a quite homogeneous age and mass distribution. However, there is a lack of stars cooler than A5, in conflict with the observations. Overall, as expected, none of the employed isochrones is suitable to reproduce the observed distribution of spectral types. Nevertheless, assuming solar metallicity offers the best compromise, with most stars situated in the late B-type realm and a tail of objects extending down to spectral type F0.

thumbnail Fig. 13.

Mass versus fractional age on the main sequence (τ) distributions for isochrones with [Z] = 0.008, 0.020, and 0.060, illustrating the differences in the derived mass and age distributions.

To further tackle this important problem, a modern and detailed abundance analysis of the light elements is needed. The current available data are rare and mainly based on the assumption of local thermodynamical equilibrium (Roby & Lambert 1990). For the relevant spectral type domain, almost all suitable spectral lines (i.e. lines of sufficient strength) are situated in the spectral region redwards of 6000 Å. Unfortunately, the medium-resolution spectra of the LAMOST survey, which should be sufficient in terms of resolution, do not cover a significant amount of the specified spectral region (Zhang et al. 2020).

Finally, diffusion calculations for light elements are needed to estimate the influence of the magnetic field and to what extent the observed surface abundances are representative of the stellar composition. Until now, however, because of the lack of corresponding observations, such calculations are not available (Stift & Alecian 2012). Therefore, in the following, we have adopted the results for solar metallicity ([Z] = 0.020) as best approximation. Assuming isochrones for lower [Z] values would lead to the derivation of older fractional ages (cf. Fig. 13).

Hubrig et al. (2000) put forth the hypotheses that mCP stars with masses M < 3 M are concentrated towards the centre of the main-sequence band and that magnetic fields only appear in stars that have completed about 30% of their lifetime on the main sequence. In their investigation of the evolutionary status of mCP stars, Kochukhov & Bagnulo (2006) demonstrated that mCP stars with M > 3 M are distributed homogeneously among the main sequence. They further identified 22 young (τ < 30%) mCP stars among their sample stars with M ≤ 3 M, thereby rejecting the proposal of Hubrig et al. (2000) that all observably magnetic low-mass CP stars have completed a significant fraction of their main-sequence lifetime. That very young (ZAMS to 25% on the main sequence) mCP stars do exist has been unambiguously demonstrated by several studies on the basis of members of open clusters (cf. e.g. Bagnulo et al. 2003; Pöhnl et al. 2003; Landstreet et al. 2007, 2008).

Nevertheless, Kochukhov & Bagnulo (2006) also find an uneven distribution for mCP stars with masses of M < 3 M, in particular for stars with M ≤ 2 M, which tend to cluster in the centre of the main-sequence band. Confirming their previous results, Hubrig et al. (2007) again found that mCP stars with M < 3 M are concentrated towards the centre of the main-sequence band.

As is obvious from Fig. 9, most of our sample stars are situated in the 2 ≤ M ≤ 3 bin. In agreement with the results of the aformentioned studies, we also find an uneven distribution of the fractional lifetime; however, our sample stars boast a mean fractional age of τ = 63% (standard deviation of 23%; cf. Fig. 14). Young mCP stars, while undoubtedly present, are conspicuously underrepresented in our sample.

thumbnail Fig. 14.

Distribution of fractional ages on the main sequence (τ) among the 903 sample stars fulfilling our accuracy criteria.

In summary, our results strongly suggest an inhomogeneous age distribution among low-mass (M < 3 M) mCP stars as hinted at by previous studies. However, we stress that our sample is biased towards mCP stars showing a conspicuous 5200 Å flux depression in the low-resolution LAMOST spectra and has not been selected on the basis of an unbiased, direct detection of a magnetic field. Therefore, our results have to be viewed with caution and their general validity needs to be tested by a more extended sample selected via different methodological approaches. It remains to be sorted out in what way the occurrence of the 5200 Å depression is connected with this result, in particular why this feature is apparently much more prominent in older stars. Several studies have shown that the 5200 Å depression increases with magnetic field strength and the atmospheric metal content (e.g. Kochukhov et al. 2005; Khan & Shulyak 2006).

Our analysis has been based on a statistically significant sample of mCP stars. Furthermore, due to the applied methods, it is not impacted by potential error sources that have been proposed to influence the results of former studies, such as a displacement of stars from the ZAMS by the application of negative Lutz-Kelker corrections or incorrect Teff values caused by the anomalous flux distributions of mCP stars (cf. e.g. the discussions in Kochukhov & Bagnulo 2006; Netopil et al. 2008). Even if the here derived error margins had been significantly underestimated, which we see no reason to believe, the general conclusion would hold. However, we caution that individual [Z] values and their errors are not available for our sample stars and that the influence of [Z] on the derived fractional ages is large.

4.3. Space distribution

To investigate the location of our sample stars in the Galactic [XYZ] plane, the corresponding coordinates were obtained from a conversion of spherical Galactic coordinates (latitude and longitude) to Cartesian coordinates using the distance d from Bailer-Jones et al. (2018). In this work, the positive X-axis points towards the Galactic centre, the Y-axis is positive in the direction of Galactic rotation and the positive Z-axis points towards the north Galactic pole. Only objects with absolute parallax errors less than 25% were considered in this process. 942 stars satisfied this criterion.

We divided our sample in candidate members of the thin disk (scale height of 350 pc) and the thick disk (1200 pc). The scale heights were taken from Ojha (2001) and Aumer & Binney (2017). Our results are shown in Fig. 15. From the sample of 942 stars, 797 objects likely belong to the thin disk and 135 objects to the thick disk. The remaining ten stars qualify for being members of the halo population and are thus worth a closer look.

thumbnail Fig. 15.

Distribution of the 942 stars with absolute parallax errors less than 25% in the [XY] plane. Stars were divided in probable members of the thin and thick disk according to the scale heights given in Ojha (2001), Aumer & Binney (2017). Ten stars have Z values larger than 1200 pc and might be Halo objects.

As a first step, we checked the spectra of the halo star candidates and confirmed that all objects exhibit the typical spectral features of mCP stars and a clearly visible flux depression. For the calculation of the total spatial velocity vtot, the radial velocity (RV) is needed. Data from the LAMOST survey include automatically measured RV information of the spectra (Anguiano et al. 2018). We checked the reliability of these values for mCP stars, that is, the spectral range from late B- to early F-type objects, by searching for common entries with the RV catalogue of Kharchenko et al. (2007). In total, 11 stars were found that are common to both our sample and this catalogue. Some of these objects boast more than one spectrum in LAMOST DR4; in these cases, mean RV values were calculated. Comparing the RV values from both sources, we find a mean difference of +2.4 km s−1, which lends confidence that the LAMOST RVs are useful in a statistical sense. However, we caution that an external uncertainty we cannot account for is introduced by the spot-induced RV variations of mCP stars that can reach up to ±50 km s−1 (Polosukhina et al. 1999).

The space velocities were calculated following the formulae of Johnson & Soderblom (1987). The final values are listed in Table 4. Stars of the halo population show vtot > 180 km s−1 as compared to the local standard of rest (Venn et al. 2004). We therefore conclude that the stars LAMOST J122746.05+113635.3 (#876; Gaia DR2 3907547639444408064) and LAMOST J150331.87+093125.4 (#880; Gaia DR2 1167894108493926016) are kinematically true halo objects, which is of considerable interest as no halo CP2 stars have been discovered so far. Considering the error, the star LAMOST J140422.54+044357.9 (#879) does not satisfy this criterion.

Table 4.

Kinematic and astrometric data for the ten stars of our sample with a height larger than 1200 pc above the Galactic plane.

Beers et al. (1996) identified LAMOST J122746.05 + 113635.3 as candidate field horizontal-branch star. Its spectrum, however, is that of a Si CP2 star (Fig. 16). There are very strong Si II lines at 3856/62 Å, 4128/31 Å, 4200 Å, 5041/56 Å, and 6347/71 Å. In addition, the He I lines are weak, which is commonly observed in CP2 stars. It has consequently been classified as B8 IV Si (He-wk) by the MKCLASS_mCP code. Almost all blue horizontal-branch stars, on the other hand, are metal-weak and their spectra rather resemble that of λ Bootis stars (Gray & Corbally 2009). We feel therefore safe in rejecting the proposed horizontal-branch classification.

thumbnail Fig. 16.

Blue-violet spectra of the proposed halo stars LAMOST J122746.05+113635.3 (#876; MKCLASS final type B8 IV Si (He-wk); upper spectrum) and LAMOST J150331.87+093125.4 (#880; MKCLASS final type A8 V SrCrEu; lower spectrum). Some prominent lines of interest are indicated.

In summary, according to the available evidence, LAMOST J122746.05 + 113635.3 and LAMOST J150331.87 + 093125.4 are bona-fide CP2 stars whose distances and kinematical properties are in agreement with halo stars. If confirmed, they would be the first CP2 halo objects known and therefore of great interest.

LAMOST J155549.85+401144.4 (#881; Gaia DR2 1382933122321062912) is another interesting object because it is listed in the catalogue of hot subdwarfs by Geier et al. (2017). The location in the (BP − RP)0 versus MG, 0 diagram (Fig. 8), however, does not support this classification. The same is true for the spectrum, which is that of a classical Si CP2 star (MKCLASS final type: B8 IV Si). We note that the occurrence of abundance anomalies in hot subdwarfs has been well established; for example, Wild & Jeffery (2018) identified two hot subdwarfs with effective temperatures of about 37 000 K and enrichments of 1.5 to 3 dex in heavy metals. This, however, is very different from what we see in LAMOST J155549.85+401144.4, which is significantly cooler than that (∼14 000 K) and shows the abundance pattern of a CP2 star. The current evidence, therefore, points to it being no subdwarf but a Si CP2 star.

Although the LAMOST survey is avoiding dense regions such as star clusters, we searched for possible cluster members among our sample stars. To this end, the positions, diameter, proper motions, distances and their errors of star clusters from Kharchenko et al. (2013) and Cantat-Gaudin et al. (2018) were employed and we searched for matches within 3σ of these parameters. In total, seven matches in six open clusters were found, which are listed in Table 5. Judging from a comparison of the ages derived from Fig. 8 and the cluster ages, all objects seem to be true cluster members.

Table 5.

Open cluster members among our sample stars.

4.4. Peculiarity type distribution

Figure 17 explores the distribution of Si, Cr, Sr, and Eu peculiarities versus hydrogen-line spectral type for the 876 stars of our sample with unambiguous identifications (i.e. without the stars in which only strong 4077 Å and/or 4130 Å blends or no traditional peculiarities were identified). Stars with hydrogen-line types of B9.5 are not listed separately but included under spectral type B9. The number of stars per spectral type bin varies considerably, with the B8−A0 stars forming the vast majority of our sample. Nevertheless, some tentative trends can be identified, although the interpretation towards the low temperature end is severely hampered by the small number statistics for objects later than A9:

thumbnail Fig. 17.

Fractional distribution of chemical peculiarities versus hydrogen-line spectral type for the 876 stars with unambiguous peculiarity type identifications. The numbers above the bars indicate the number of objects in the corresponding spectral type bin. Because a single object may have multiple peculiarities, fractions may exceed 1.

  • Si peculiarities are present between spectral types B4 and F0. They play a dominant role in stars with spectral types earlier than B9, strongly decreasing in importance in later-type objects. Except for He peculiarites (which are not shown in the plot), Si peculiarities are the only chemical peculiarities identified in objects earlier than B8.

  • Cr peculiarities set in at spectral type B8 and form an important part of the peculiarity mix between spectral types B9 and A9.

  • Sr and Eu peculiarities set in at spectral type B8 and increase in strength towards later types.

This is in good agreement with the expectations and the literature. It is well known that Si peculiarities are present throughout a wide range of effective temperatures in mCP stars (e.g. Renson & Manfroid 2009). The hottest stars with Eu peculiarities from the Renson & Manfroid (2009) catalogue are of spectral type B8, which holds true also for the vast majority of stars with Cr and Sr peculiarities, with the exception of only three objects (HD 35502, spectral type B6 Sr Cr Si; HD 167288, spectral type B7 Si Cr; HD 213918, spectral type B7 Si Sr). Likewise, the work of Ghazaryan et al. (2018) contains atomic data for Eu, Cr, and Sr from effective temperatures of, respectively, 12 900 K (∼B8), 14 700 K (∼B6), and 13 300 K (∼B8) downwards. The good agreement of the peculiarity type “blue borders” between the present work and the literature provides independent proof of the reliability of the here derived spectral types.

Figure 18 illustrates the distribution of stars in which only strong blends at 4077 Å and/or 4130 Å were identified. Again, stars with hydrogen-line types of B9.5 are included under spectral type B9. These stars were not assigned Si, Cr, Sr, and Eu types with the here employed workflow because, apart from the strong blends, the peculiarities are either too subtle to have passed our significance criteria, no other significant features are present or the code failed to identify them for some reason (cf. Sect. 3.1).

thumbnail Fig. 18.

Distribution of stars in which only strong blends at 4077 Å and/or 4130 Å were identified.

Manual classification is necessary to throw more light on what elements contribute to the observed blends. Nevertheless, the distribution of the 4130 Å blend identifications, in particular for objects earlier than B9, is in general agreement with the distribution of Si peculiarities. We therefore expect that most of the “bl4130” stars in our sample will turn out to be Si stars. No similar predictions can be made for the “bl4077” stars from the available data.

4.5. Comparison with samples from the literature

The following sections compare our results with the works of Renson & Manfroid (2009), Skiff (2014), and Qin et al. (2019). We further note that 22 of our sample stars are contained in the sample of strongly magnetic Ap stars of Scholz et al. (2019). As the authors do not list spectral types, a direct comparison of results was not possible. The stars common to both samples are identified in Table A.1.

4.5.1. Comparison with the compilations of Renson & Manfroid (2009) and Skiff (2014)

Our final sample contains 59 mCP stars or candidates that are also included in the catalogue of Renson & Manfroid (2009). This low level of coincidence (6.65%) is expected because the Renson & Manfroid (2009) sample mostly consists of bright stars (peaking at around 9th magnitude) for which there are no LAMOST spectra available.

Table 6 gives a comparison of the spectral types from the present study, the RM09 catalogue, and the compilation of Skiff (2014). For most stars, there is a good general agreement between the different sources. For example, for the 46 stars that have at least one other detailed literature classification listing the temperature subtype, the determined hydrogen-line types agree within ±2 subclasses, which seems reasonable considering the inhomogeneous source material behind the literature classifications and the inherent difficulties in classifying mCP stars. For several stars, our results provide a first detailed classification; furthermore, we confirm several doubtful objects as mCP stars and show that some suspected CP1 stars are in fact mCP stars. A more detailed investigation into this matter will be the topic of an upcoming study that will be concerned with a new classification of stars in the RM09 catalogue based on homogeneous spectroscopic material.

Table 6.

Comparison of the spectral types derived in this study, the RM09 catalogue, and the compilation of Skiff (2014).

4.5.2. Comparison with the sample of Qin et al. (2019)

Qin et al. (2019) searched for CP1 stars in low resolution spectra of early-type stars from LAMOST DR5 and compiled a catalogue of 9372 CP1 stars. Because cooler CP2 stars may exhibit similar spectral features (Ca deficiency, overabundance of Fe-group elements), the authors expect a contamination of their sample by these objects. To identify potential CP2 stars among the CP1 star candidates, they used the 4077 Å blend as reference line, which may contain contributions from Si II, Cr II, and Sr II. Synthetic spectra with overabundances of Sr, Cr, Eu, and Si of 2.0 dex were computed for different effective temperatures and the equivalent widths of the 4077 Å blend were calculated and compared for both the templates and the observed spectra. If the equivalent width of the observed 4077 Å feature (EW4077_obs) exceeded that of the corresponding templates (EW4077_temp), a star was flagged as a CP2 star candidate. In this way, Qin et al. (2019) flagged 1131 stars within their sample of CP1 star candidates as CP2 star candidates. From a cursory investigation of about 20 randomly chosen objects, several bona-fide CP2 stars have indeed been found among the objects with high values of (EW4077_obs − EW4077_temp), in line with the expectations of Qin et al. (2019). However, the incidence of CP2 stars seems to drop rather sharply towards lower values of (EW4077_obs − EW4077_temp). We assume that this is because a strong 4077 Å feature alone, while often helpful, is an insufficient criterion for identifying CP2 stars.

Only 45 objects are common to both our sample and the Qin et al. (2019) catalogue. Of these stars, about 70% (31 objects) were flagged as CP2 star candidates. Table 7 compares the Qin et al. (2019) k/h/m spectral types to the final types derived in the present study for all objects common to both lists. In general, the agreement between the derived spectral types is poor. With the here employed methodology, different k/h/m-types were assigned to only 21 of these stars. A detailed investigation of the stars common to both lists, in particular a comparison of the automatically derived classifications to manually derived spectral types and an investigation of the source(s) of the observed discrepancies, is beyond the scope of the present paper but might be beneficial to both studies and help with the refinement of the employed algorithms.

Table 7.

Comparison of the spectral types derived in this study to the catalogue of Qin et al. (2019).

On first glance, the low level of coincidence between the Qin et al. (2019) catalogue and the here presented sample of mCP seems surprising. We have identified several reasons for this, which are related to the different approaches and goals of both studies.

Qin et al. (2019) explicitly searched for CP1 stars. CP2 stars were identified as a contaminant and corresponding candidates were subsequently identified. Because they searched for CP2 star candidates within their sample of CP1 star candidates, that is, among objects with pronounced differences between k and h spectral types, their sample will not contain any CP2 stars that do not share these characteristics. However, early-type CP2 stars generally do not show significant (if any) differences between k and h types; this phenomenon is mostly restricted to late-type CP2 stars. Thus, we assume that the Qin et al. (2019) subsample of CP2 star candidates is biased towards late-type CP2 stars7. This is further corroborated by the fact that, in agreement with the expectations for identifying CP1 stars, Qin et al. (2019) constrained their search to objects with 6500 K < Teff < 11 000 K. In fact, their final catalogue contains only 11 objects with Teff > 10 000 K. Thus, only very few stars hotter than spectral type A0 are present in their sample.

Our study follows a very different approach and is concerned with the identification of mCP stars that were selected among early-type targets by the presence of a significant 5200 Å flux depression. The spectra of CP1 stars, on the other hand, generally do not show this feature (Paunzen et al. 2005). However, not all mCP stars distinctly show a 5200 Å depression either, in particular in low-resolution spectra. Therefore, we will have missed any such mCP stars, which might have found their way into the Qin et al. (2019) candidate sample. Most important, however, is that the majority of our sample stars is situated in the spectral range of B8 to A0 (9700 K < Teff < 12 500 K), which renders them mostly incompatible with the candidate sample of Qin et al. (2019).

In summary, the above mentioned issues, combined with the fact that the incidence of CP2 stars seems to drop rather sharply towards lower values of (EW4077_obs − EW4077_temp) in the CP2 star candidate subsample of Qin et al. (2019) and the different source material (∼9.0 million spectra in DR5 vs. ∼7.6 million spectra in DR4), illustrate that no significant overlap between both samples is to be expected. We would like to again stress that it never was the intention of Qin et al. (2019) to collect a pure sample of CP2 stars. Their subsample of candidates, however, is a valuable starting point for further investigations. As a spin-off of the present study, we intend to investigate the Qin et al. (2019) CP2 star candidates to confirm or reject their status as mCP stars. It is clear that investigations based on a broader analysis of LAMOST early-type spectra (i.e. also including stars without conspicuous 5200 Å flux depressions) will lead to the discovery of many more mCP stars in the future.

4.6. The mid-B type mCP stars – He-peculiar objects?

23 stars of our sample have MKCLASS final types earlier than B7. More than half of these objects were classified as showing peculiarly weak He I lines (“He-wk”). Interestingly, three stars were identified as showing both weak and strong (“He-st”) He I lines, which strongly suggests He peculiarity. Besides that, only Si overabundances and strong 4130 Å blends were identified in several of these objects. As He-rich stars are generally hotter than spectral type B4 (Gray & Corbally 2009) and therefore not expected to contribute to our sample, we consider these objects good candidates for He-weak (CP4) stars.

Figure 19 showcases the spectra of five mid-B type stars. The hydrogen-line profiles and the prominent C II 4267 Å lines corroborate the classifications, although we manually derived slightly different temperature types in two cases. LAMOST J014940.99+534134.2 (#37; TYC 3684-1139-1) and LAMOST J062348.46+034201.1 (#567; HD 256582) are He-weak stars with Si overabundances. LAMOST J062307.91+264642.0 (#565; Gaia DR2 3432273606513132544) is also certainly He-weak but does not fit any of the standard subclasses of the He-weak stars (the “hot” Si stars, i.e. Si stars with hotter temperatures than the classical Si CP2 stars; the P Ga stars; the Sr Ti stars; cf. Gray & Corbally 2009). It is here classified as B4 V HeB7 (R. O. Gray, priv. comm.).

thumbnail Fig. 19.

Showcase of five newly identified peculiar mid-B type stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J014940.99+534134.2 (#37; TYC 3684-1139-1), LAMOST J052118.97+320805.7 (#318; HD 242764), LAMOST J055023.89+261330.2 (#421; TYC 1866-861-1), LAMOST J062307.91+264642.0 (#565; Gaia DR2 3432273606513132544), and LAMOST J062348.46+034201.1 (#567; HD 256582). MKCLASS final types and manual types derived in the present study are indicated. Some prominent lines of interest are identified. The asterisk marks the position of a “glitch” in the spectrum of LAMOST J014940.99+534134.2.

LAMOST J055023.89+261330.2 (#421; TYC 1866-861-1) boasts a rather noisy spectrum (g band S/N of 79) that is, apart from the hydrogen lines, basically a “smattering” of metal-lines, without any particularly outstanding features – except for the strong C II 4267 Å and the weak Mg II 4481 Å lines that support its classification as a mid B-type object. This is also supported by the colour index (BP − RP)0 = −0.182 mag. The He I lines, then, seem to be curiously absent from its spectrum, so the star may be related to the CP4 stars, although the metal-lines seem way too strong to support this interpretation. We here tentatively classify it as B4: V HeB9. The star shows a strong flux depression at 5200 Å, and, according to data from the SuperWASP archive (Butters et al. 2010), is a periodic photometric variable with a period of about 11.5 d. It certainly merits a closer look – this, however, is beyond the scope of the present investigation.

The He lines of LAMOST J052118.97+320805.7 (#318; HD 242764) do not look weak for its temperature type but have broad profiles suggesting rapid rotation. This, however, is not supported by the hydrogen-line profile, which almost exactly matches that of the B4 V standard. The presence of a number of unidentified metal-lines (which do show evidence for rotational broadening, but not to the extent the He I lines imply) and the conspicuous 5200 Å depression suggest that this star is indeed chemically peculiar, although it also does not fit into any of the standard mid-B peculiarity subtypes. It has been classified as B4 Vpn in the present study (R. O. Gray, priv. comm.).

Additional proof that this star is indeed chemically peculiar comes from its periodic photometric variability with a period of about 5.1 d in SuperWASP data and the conspicuous 5200 Å flux depression in its spectrum. Incidentally, the star is listed with a spectral type of B8 Si Sr in the Renson & Manfroid (2009) catalogue, which is not supported by the available LAMOST spectrum. We were unable to get at the root of this classification; however, the star is listed as B9 in the Henry Draper extension (Cannon 1931) and was classified as B5p by Chargeishvili (1988). Further He-peculiar objects and candidates can be gleaned from Table A.1.

4.7. The eclipsing binary system LAMOST J034306.74+495240.7

The star LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1) was identified as an eclipsing binary system in ASAS-SN data (Jayasinghe et al. 2019). It is listed in the International Variable Star Index of the AAVSO (VSX; Watson 2006) under the designation ASASSN-V J034306.74+495240.8 and with a period of 5.1431 d. We have analyzed the available ASAS-SN data for this star and derive a period of 5.1435 ± 0.0012 d and an epoch of primary minimum at HJD 2457715.846 ± 0.002. The light curve is shown in Fig. 20 and illustrates that the orbit is slightly eccentric, the secondary minimum occurs at an orbital phase of φ = 0.46. In addition, there is evidence for out-of-eclipse variability in agreement with rotational modulation on one component of the system.

thumbnail Fig. 20.

ASAS-SN light curve of the eclipsing binary system LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1). The data have been folded with the orbital period of Porb = 5.1435 ± 0.0012 d.

Two spectra are available for this star in LAMOST DR4. The first spectrum (“spectrum1”) was obtained on 23 October 2015 (MJD 57318; observation median UTC 17:33:00; g band S/N: 258), which corresponds to an orbital phase of φ = 0.890. The second spectrum (“spectrum2”) was taken on 19 January 2016 (MJD 57406; observation median UTC 11:46:00; g band S/N: 207), which corresponds to an orbital phase of φ = 0.952. Therefore, both spectra were taking during maximum light, and we find no significant difference between them. Both show a strong flux depression at 5200 Å and enhanced Si II lines at 3856/62 Å, 4128/31 Å, 4200 Å, 5041/56 Å, and 6347/71 Å. We have analyzed the spectrum with the highest S/N (spectrum1) and derive a spectral type of B9 III Si. Figure 21 compares the blue-violet part of both spectra to the liblamost B9 III standard, whose hydrogen-line profile provides a good fit to the observed ones.

thumbnail Fig. 21.

Comparison of the blue-violet spectra of the eclipsing binary system LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1) to the liblamost B9 III standard spectrum (upper spectrum). Some prominent lines of interest are indicated.

In summary, we conclude that at least one component of the LAMOST J034306.74+495240.7 system is a Si CP2 star. It is, therefore, of great interest because mCP stars in eclipsing binaries are exceedingly rare (Renson & Manfroid 2009; Niemczura et al. 2017; Kochukhov et al. 2018; Skarka et al. 2019) and accurate parameters for the components can be derived via an orbital solution of the system. We strongly encourage further studies of this interesting object.

4.8. The SB2 system LAMOST J050146.85+383500.8

Figure 22 illustrates the peculiar spectrum of LAMOST J050146.85+383500.8 (#272; HD 280281), which we suspect to be a blend of two different stars. This becomes especially obvious in the profile of the Hγ line (Fig. 23).

thumbnail Fig. 22.

Comparison of the blue-violet spectra of the propsed SB2 system LAMOST J050146.85+383500.8 (#272; HD 280281; MKCLASS final type B8 V Si) to the libsynth B8 V standard spectrum (upper spectrum). Some prominent lines of interest are indicated.

thumbnail Fig. 23.

Close-up view of the Hγ region of the proposed SB2 system LAMOST J050146.85+383500.8 (blue spectrum) and the libsynth B8 V standard (black spectrum), illustrating the peculiar profile of the Hγ line indicative of binarity.

To further investigate this matter, we employed the VO Sed Analyzer tool VOSA8 v6.0 (Bayo et al. 2008) to fit the SED to the available photometry. For comparison, we used a Kurucz ODFNEW/NOVER model (Castelli et al. 1997) with Teff = 12 500 K, which corresponds to a spectral type of B8. We emphasise that a change of Teff of about 2000 K in either direction will not impact our conclusions. Figure 24 illustrates the results of the fitting process. The flux model was fitted to either match the ultraviolet or the optical wavelength region. In any case, the discrepancies are readily visible and it is obvious that the observed flux distribution of LAMOST J050146.85+383500.8 cannot be fitted with a single star flux model. We note that it is well known that CP2 stars show a “blueing” effect (Maitzen 1980), which leads to observed flux discrepancies due to stronger absorption in the ultraviolet than in chemically normal stars. However, a slight shift in the ultraviolet region will not alter our conclusions.

thumbnail Fig. 24.

Comparison of the SED of LAMOST J050146.85+383500.8 (red dots) to a Kurucz ODFNEW/NOVER model with Teff = 12 500 K (black squares). The model was forced to either fit the ultraviolet (lower model) or optical flux (upper model). The discrepancies are clearly visible, the star’s SED cannot be fitted with a single star flux model.

Because the features of the companion star are readily visible in the available LAMOST spectrum, we conclude that its absolute magnitude must be similar to the B-type main-sequence component. We therefore assume that it is a supergiant star with a progenitor of higher mass. Such a combination of components is quite unusual among mCP stars; in order to put further constraints on this spectroscopic binary system, orbital elements or the analysis of light-travel time effects are needed. LAMOST J050146.85+383500.8, therefore, is an interesting target for follow-up studies.

5. Conclusions

We carried out a search for mCP stars in the publicly available spectra of LAMOST DR4. Suitable candidates were selected by searching for the presence of the characteristic 5200 Å flux depression. In consequence, our sample is biased towards mCP stars with conspicuous flux depressions at 5200 Å. Spectral classification was carried out with a modified version of the MKCLASS code (MKCLASS_mCP) and, for a subsample of stars, by manual classification. We evaluated our results by spot-checking with manually derived spectral types and comparison to samples from the literature.

The main findings of the present investigation are summarised in the following:

  • We identified 1002 mCP stars, most of which are new discoveries. There are only 59 common entries with the catalogue of Renson & Manfroid (2009). With our work, we significantly increase the sample size of known Galactic mCP stars, paving the way for future in-depth statistical studies.

  • To suit the special needs of our project, we updated the current version (v1.07) of the MKCLASS code to probe several additional lines, with the advantage that the new version (here termed MKCLASS_mCP) is now able to more robustly identify traditional mCP star peculiarities, including Cr peculiarities and, to some extent, He peculiarities.

  • mCP star peculiarities (Si, Cr, Sr, Eu, strong blends at 4077 Å and/or 4130 Å) were identified in all but 36 stars of our sample, highlighting the efficiency of the chosen approach and the peculiarity identification routine. The remaining objects (mostly mCP stars with weak or complicated peculiarities and He-peculiar objects) were manually searched to locate the presence of peculiarities.

  • Comparisons between manually derived spectral types and the MKCLASS_mCP final types indicate a good agreement between the derived temperature and peculiarity types. This is further corroborated by a comparison with spectral types from the Renson & Manfroid (2009) and Skiff (2014) catalogues and the good agreement of the peculiarity type versus spectral type distribution between this study and the literature. However, with our approach, we missed the presence of certain peculiarities in several objects. The peculiarity types presented here are therefore not exhaustive. They nevertheless form a sound basis for statistical and further studies.

  • Our sample stars are between 100 Myr and 1 Gyr old, with the majority having masses between 2 M and 3 M. We investigated the evolutionary status of 903 mCP stars, deriving a mean fractional age on the main sequence of τ = 63% (standard deviation of 23%). Young mCP stars, while undoubtedly present, are conspicuously underrepresented in our sample. Our results could be considered as strong evidence for an inhomogeneous age distribution among low-mass (M < 3 M) mCP stars, as hinted at by previous studies. However, we caution that our sample has not been selected on the basis of an unbiased, direct detection of a magnetic field. Therefore, our results have to be viewed with caution and their general validity needs to be tested by a more extended sample selected via different methodological approaches.

  • The mCP stars LAMOST J122746.05+113635.3 (#876) and LAMOST J150331.87+093125.4 (#880) boast distances and kinematical properties in agreement with halo stars. If confirmed, they would be the first CP2 halo objects known and therefore of special interest.

  • We identified LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1) as an eclipsing binary system (Porb = 5.1435 ± 0.0012 d) hosting a Si CP2 star component (spectral type B9 III Si). This is of great interest because mCP stars in eclipsing binaries are exceedingly rare.

  • The star LAMOST J050146.85+383500.8 was identified as an SB2 system likely comprising of a Si CP2 star and a supergiant.

Future investigations will be concerned with an in-depth study of the new mCP stars identified in this work, particularly with regard to their photometric variability, along with further development and refinement of the approach for identifying and classifying mCP stars in large spectroscopic databases using the MKCLASS code.


1

Guo Shou Jing (1231–1316) was a Chinese astronomer, hydraulic engineer, mathematician, and politician of the Yuan Dynasty.

4

The numbers given behind the identifiers refer to the internal identification number and facilitate easy identification in the tables.

7

Although not statistically significant, it is interesting to note that all stars from the Qin et al. (2019) candidate sample we were able to confirm as CP2 stars were indeed late-type CP2 stars.

9

We note that, as in the Renson & Manfroid (2009) catalogue, the “p” denoting peculiarity was omitted from the spectral classifications.

Acknowledgments

We thank the referee for his thoughtful report that helped to significantly improve the paper. This work has been supported by the DAAD (project No. 57442043). The Guo Shou Jing Telescope (the Large Sky Area Multi-Object Fiber Spectroscopic Telescope, LAMOST) is a National Major Scientific Project built by the Chinese Academy of Sciences. Funding for the project has been provided by the National Development and Reform Commission. LAMOST is operated and managed by National Astronomical Observatories, Chinese Academy of Sciences. This work presents results from the European Space Agency (ESA) space mission Gaia. Gaia data are being processed by the Gaia Data Processing and Analysis Consortium (DPAC). Funding for the DPAC is provided by national institutions, in particular the institutions participating in the Gaia MultiLateral Agreement (MLA). The Gaia mission website is https://www.cosmos.esa.int/gaia. The Gaia archive website is https://archives.esac.esa.int/gaia.

References

  1. Anguiano, B., Majewski, S. R., Allende-Prieto, C., et al. 2018, A&A, 620, A76 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  2. Arenou, F., Luri, X., Babusiaux, C., et al. 2018, A&A, 616, A17 [Google Scholar]
  3. Aumer, M., & Binney, J. 2017, MNRAS, 470, 2113 [NASA ADS] [CrossRef] [Google Scholar]
  4. Aurière, M., Wade, G. A., Silvester, J., et al. 2007, A&A, 475, 1053 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  5. Babcock, H. W. 1947, ApJ, 105, 105 [NASA ADS] [CrossRef] [Google Scholar]
  6. Bagnulo, S., Landstreet, J. D., Lo Curto, G., Szeifert, T., & Wade, G. A. 2003, A&A, 403, 645 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  7. Bagnulo, S., Landstreet, J. D., Mason, E., et al. 2006, A&A, 450, 777 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Bailer-Jones, C. A. L., Rybizki, J., Fouesneau, M., Mantelet, G., & Andrae, R. 2018, AJ, 156, 58 [NASA ADS] [CrossRef] [Google Scholar]
  9. Bayo, A., Rodrigo, C., Barrado Y Navascués, D., et al. 2008, A&A, 492, 277 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  10. Beers, T. C., Wilhelm, R., Doinidis, S. P., & Mattson, C. J. 1996, ApJS, 103, 433 [NASA ADS] [CrossRef] [Google Scholar]
  11. Bidelman, W. P. 1981, AJ, 86, 553 [NASA ADS] [CrossRef] [Google Scholar]
  12. Bidelman, W. P. 1983, AJ, 88, 1182 [NASA ADS] [CrossRef] [Google Scholar]
  13. Bidelman, W. P. 1985, AJ, 90, 341 [NASA ADS] [CrossRef] [Google Scholar]
  14. Bidelman, W. P. 1998, PASP, 110, 270 [CrossRef] [Google Scholar]
  15. Bidelman, W. P., & Keenan, P. C. 1951, ApJ, 114, 473 [NASA ADS] [CrossRef] [Google Scholar]
  16. Bidelman, W. P., & MacConnell, D. J. 1973, AJ, 78, 687 [NASA ADS] [CrossRef] [Google Scholar]
  17. Bond, H. E. 1972, PASP, 84, 446 [CrossRef] [Google Scholar]
  18. Braithwaite, J., & Spruit, H. C. 2004, Nature, 431, 819 [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
  19. Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127 [NASA ADS] [CrossRef] [Google Scholar]
  20. Butters, O. W., West, R. G., Anderson, D. R., et al. 2010, A&A, 520, L10 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  21. Cannon, A. J. 1931, Annal. Harv. Coll. Obs., 100, 61 [Google Scholar]
  22. Cantat-Gaudin, T., Jordi, C., Vallenari, A., et al. 2018, A&A, 618, A93 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  23. Castelli, F., & Kurucz, R. L. 2003, IAU Symp., 210, A20 [Google Scholar]
  24. Castelli, F., Gratton, R. G., & Kurucz, R. L. 1997, A&A, 318, 841 [NASA ADS] [Google Scholar]
  25. Chargeishvili, K. B. 1988, Abastumanskaia Astrofizicheskaia Observatoriia Byulleten, 65, 18 [Google Scholar]
  26. Clausen, J. V., & Jensen, K. S. 1979, IAU Colloq. 47: Spectral Classification of the Future, 9, 479 [Google Scholar]
  27. Cui, X.-Q., Zhao, Y.-H., Chu, Y.-Q., et al. 2012, Res. Astron. Astrophys., 12, 1197 [NASA ADS] [CrossRef] [Google Scholar]
  28. Drilling, J. S., & Pesch, P. 1973, AJ, 78, 48 [Google Scholar]
  29. Faraggiana, R. 1987, Ap&SS, 134, 381 [NASA ADS] [CrossRef] [Google Scholar]
  30. Floquet, M. 1975, A&AS, 21, 25 [NASA ADS] [Google Scholar]
  31. Gaia Collaboration (Brown, A. G. A., et al.) 2018, A&A, 616, A1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  32. Geier, S., Østensen, R. H., Nemeth, P., et al. 2017, A&A, 600, A50 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  33. Ghazaryan, S., Alecian, G., & Hakobyan, A. A. 2018, MNRAS, 480, 2953 [NASA ADS] [CrossRef] [Google Scholar]
  34. Gieseking, F. 1983, A&A, 118, 102 [NASA ADS] [Google Scholar]
  35. Gray, R. O., & Corbally, C. J. 1994, AJ, 107, 742 [NASA ADS] [CrossRef] [Google Scholar]
  36. Gray, R. O., & Corbally, C. J. 2009, Stellar Spectral Classification (Princeton, NJ: Princeton Univ. Press) [Google Scholar]
  37. Gray, R. O., & Corbally, C. J. 2014, AJ, 147, 80 [NASA ADS] [CrossRef] [Google Scholar]
  38. Gray, R. O., Corbally, C. J., De Cat, P., et al. 2016, AJ, 151, 13 [NASA ADS] [CrossRef] [Google Scholar]
  39. Green, G. M., Schlafly, E. F., Finkbeiner, D., et al. 2018, MNRAS, 478, 651 [NASA ADS] [CrossRef] [Google Scholar]
  40. Grenier, S., Baylac, M. O., Rolland, L., et al. 1999, A&AS, 137, 451 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  41. Guetter, H. H. 1977, AJ, 82, 598 [NASA ADS] [CrossRef] [Google Scholar]
  42. Hou, W., Luo, A., Yang, H., et al. 2015, MNRAS, 449, 1401 [NASA ADS] [CrossRef] [Google Scholar]
  43. Hubrig, S., North, P., & Mathys, G. 2000, ApJ, 539, 352 [NASA ADS] [CrossRef] [Google Scholar]
  44. Hubrig, S., North, P., & Schöller, M. 2007, Astron. Nachr., 328, 475 [NASA ADS] [CrossRef] [Google Scholar]
  45. Hümmerich, S., Mikulášek, Z., Paunzen, E., et al. 2018, A&A, 619, A98 [CrossRef] [EDP Sciences] [Google Scholar]
  46. Jamar, C. 1977, A&A, 56, 413 [NASA ADS] [Google Scholar]
  47. Jamar, C. 1978, A&A, 70, 379 [NASA ADS] [Google Scholar]
  48. Jayasinghe, T., Stanek, K. Z., Kochanek, C. S., et al. 2019, MNRAS, 486, 1907 [NASA ADS] [Google Scholar]
  49. Johnson, D. R. H., & Soderblom, D. R. 1987, AJ, 93, 864 [NASA ADS] [CrossRef] [Google Scholar]
  50. Jones, T. J., Wolff, S. C., & Bonsack, W. K. 1974, ApJ, 190, 579 [NASA ADS] [CrossRef] [Google Scholar]
  51. Kepler Mission Team 2009, VizieR Online Data Catalog: V/133 [Google Scholar]
  52. Khan, S. A., & Shulyak, D. V. 2006, A&A, 448, 1153 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  53. Khan, S. A., & Shulyak, D. V. 2007, A&A, 469, 1083 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  54. Kharchenko, N. V., Scholz, R. D., Piskunov, A. E., Röser, S., & Schilbach, E. 2007, Astron. Nachr., 328, 889 [NASA ADS] [CrossRef] [Google Scholar]
  55. Kharchenko, N. V., Piskunov, A. E., Schilbach, E., Röser, S., & Scholz, R. D. 2013, A&A, 558, A53 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  56. Kochukhov, O. 2017, A&A, 597, A58 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  57. Kochukhov, O., & Bagnulo, S. 2006, A&A, 450, 763 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  58. Kochukhov, O., Bagnulo, S., & Barklem, P. S. 2002, ApJ, 578, L75 [NASA ADS] [CrossRef] [Google Scholar]
  59. Kochukhov, O., Khan, S., & Shulyak, D. 2005, A&A, 433, 671 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  60. Kochukhov, O., Johnston, C., Alecian, E., & Wade, G. A. 2018, MNRAS, 478, 1749 [NASA ADS] [CrossRef] [Google Scholar]
  61. Kodaira, K. 1969, ApJ, 157, L59 [NASA ADS] [CrossRef] [Google Scholar]
  62. Krtička, J., Mikulášek, Z., Henry, G. W., et al. 2009, A&A, 499, 567 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  63. Krtička, J., Janík, J., Marková, H., et al. 2013, A&A, 556, A18 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  64. Kupka, F., Paunzen, E., & Maitzen, H. M. 2003, MNRAS, 341, 849 [NASA ADS] [CrossRef] [Google Scholar]
  65. Landstreet, J. D., Bagnulo, S., Andretta, V., et al. 2007, A&A, 470, 685 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  66. Landstreet, J. D., Silaj, J., Andretta, V., et al. 2008, A&A, 481, 465 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  67. Loden, L. O., & Sundman, A. 1987, J. Astrophys. Astron., 8, 351 [CrossRef] [Google Scholar]
  68. Loden, L. O., & Sundman, A. 1989, J. Astrophys. Astron., 10, 183 [CrossRef] [Google Scholar]
  69. Luo, A. L., Zhao, Y. H., Zhao, G., et al. 2018, VizieR Online Data Catalog: V/153 [Google Scholar]
  70. Maitzen, H. M. 1976, A&A, 51, 223 [NASA ADS] [Google Scholar]
  71. Maitzen, H. M. 1980, A&A, 89, 230 [NASA ADS] [Google Scholar]
  72. Michaud, G. 1970, ApJ, 160, 641 [Google Scholar]
  73. Molnar, M. R. 1973, ApJ, 179, 527 [NASA ADS] [CrossRef] [Google Scholar]
  74. Morgan, W. W. 1933, ApJ, 77, 330 [NASA ADS] [CrossRef] [Google Scholar]
  75. Moss, D. 2004, IAU Symp., 224, 245 [CrossRef] [Google Scholar]
  76. Murphy, S. J., & Paunzen, E. 2017, MNRAS, 466, 546 [NASA ADS] [CrossRef] [Google Scholar]
  77. Nassau, J. J., & Macrae, D. A. 1955, ApJ, 121, 32 [CrossRef] [Google Scholar]
  78. Netopil, M., Paunzen, E., Maitzen, H. M., North, P., & Hubrig, S. 2008, A&A, 491, 545 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  79. Netopil, M., Paunzen, E., Heiter, U., & Soubiran, C. 2016, A&A, 585, A150 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  80. Netopil, M., Paunzen, E., Hümmerich, S., & Bernhard, K. 2017, MNRAS, 468, 2745 [NASA ADS] [CrossRef] [Google Scholar]
  81. Niemczura, E., Hümmerich, S., Castelli, F., et al. 2017, Sci. Rep., 7, 5906 [NASA ADS] [CrossRef] [Google Scholar]
  82. Ojha, D. K. 2001, MNRAS, 322, 426 [NASA ADS] [CrossRef] [Google Scholar]
  83. Paunzen, E. 2015, A&A, 580, A23 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  84. Paunzen, E., Stütz, C., & Maitzen, H. M. 2005, A&A, 441, 631 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  85. Paunzen, E., Netopil, M., Pintado, O. I., & Rode-Paunzen, M. 2011, Astron. Nachr., 332, 77 [NASA ADS] [CrossRef] [Google Scholar]
  86. Pecaut, M. J., & Mamajek, E. E. 2013, ApJS, 208, 9 [Google Scholar]
  87. Pöhnl, H., Maitzen, H. M., & Paunzen, E. 2003, A&A, 402, 247 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  88. Polosukhina, N., Kurtz, D., Hack, M., et al. 1999, A&A, 351, 283 [NASA ADS] [Google Scholar]
  89. Preston, G. W. 1974, ARA&A, 12, 257 [NASA ADS] [CrossRef] [MathSciNet] [Google Scholar]
  90. Przybylski, A. 1966, Nature, 210, 20 [NASA ADS] [CrossRef] [Google Scholar]
  91. Qin, L., Luo, A. L., Hou, W., et al. 2019, ApJS, 242, 13 [CrossRef] [Google Scholar]
  92. Renson, P., & Manfroid, J. 2009, A&A, 498, 961 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  93. Richer, J., Michaud, G., & Turcotte, S. 2000, ApJ, 529, 338 [NASA ADS] [CrossRef] [Google Scholar]
  94. Roby, S. W., & Lambert, D. L. 1990, ApJS, 73, 67 [NASA ADS] [CrossRef] [Google Scholar]
  95. Sato, K., & Kuji, S. 1990, A&AS, 85, 1069 [NASA ADS] [Google Scholar]
  96. Scholz, R. D., Chojnowski, S. D., & Hubrig, S. 2019, A&A, 628, A81 [CrossRef] [EDP Sciences] [Google Scholar]
  97. Sikora, J., David-Uraz, A., Chowdhury, S., et al. 2019, MNRAS, 487, 4695 [CrossRef] [Google Scholar]
  98. Skarka, M., Kabáth, P., Paunzen, E., et al. 2019, MNRAS, 487, 4230 [CrossRef] [Google Scholar]
  99. Skiff, B. A. 2014, VizieR Online Data Catalog: B/mk [Google Scholar]
  100. Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163 [Google Scholar]
  101. Slettebak, A., & Stock, J. 1959, Astronomische Abhandlungen der Hamburger Sternwarte, 5, 105 [Google Scholar]
  102. Smalley, B., Antoci, V., Holdsworth, D. L., et al. 2017, MNRAS, 465, 2662 [CrossRef] [Google Scholar]
  103. Stibbs, D. W. N. 1950, MNRAS, 110, 395 [NASA ADS] [CrossRef] [Google Scholar]
  104. Stift, M. J., & Alecian, G. 2012, MNRAS, 425, 2715 [NASA ADS] [CrossRef] [Google Scholar]
  105. Vagnozzi, S. 2019, Atoms, 7, 41 [NASA ADS] [CrossRef] [Google Scholar]
  106. Venn, K. A., Irwin, M., Shetrone, M. D., et al. 2004, AJ, 128, 1177 [NASA ADS] [CrossRef] [Google Scholar]
  107. Watson, C. L. 2006, Soc. Astron. Sci. Annu. Symp., 25, 47 [Google Scholar]
  108. Wild, J. F., & Jeffery, C. S. 2018, MNRAS, 473, 4021 [NASA ADS] [CrossRef] [Google Scholar]
  109. Wolff, S. C. 1983, The A-stars: Problems and Perspectives (Washington: NASA) [Google Scholar]
  110. Wolff, S. C., & Wolff, R. J. 1971, AJ, 76, 422 [NASA ADS] [CrossRef] [Google Scholar]
  111. Zhang, B., Liu, C., Li, C.-Q., et al. 2020, Res. Astron. Astrophys., 20, 51 [CrossRef] [Google Scholar]
  112. Zhao, G., Zhao, Y.-H., Chu, Y.-Q., Jing, Y.-P., & Deng, L.-C. 2012, Res. Astron. Astrophys., 12, 723 [Google Scholar]

Appendix A: Essential data for our sample stars

Table A.1 lists essential data for our sample stars. It is organised as follows:

  • Column 1: Internal identification number.

  • Column 2: LAMOST identifier.

  • Column 3: Alternativ identifier (HD number, TYC identifier, or Gaia DR2 number).

  • Column 4: Right ascension (J2000). Positional information was taken from Gaia DR2 (Gaia Collaboration 2018; Arenou et al. 2018).

  • Column 5: Declination (J2000).

  • Column 6: MKCLASS final type, as derived in this study9. All further additions to the spectral type that are not directly based on the MKCLASS_mCP output are highlighted using italics. For an easy identification, manually altered spectral types are indicated by asterisks.

  • Column 7: Sloan g band S/N of the analysed spectrum.

  • Column 8: G mag (Gaia DR2).

  • Column 9: G mag error.

  • Column 10: Parallax (Gaia DR2).

  • Column 11: Parallax error.

  • Column 12: Dereddened colour index (BP − RP)0 (Gaia DR2).

  • Column 13: Colour index error.

  • Column 14: Absorption in the G band, AG.

  • Column 15: Intrinsic absolute magnitude in the G band, MG, 0.

  • Column 16: Absolute magnitude error.

All Tables

Table 1.

Absorption lines and blends identified by the modified version of the MKCLASS code (MKCLASS_mCP) and used in the identification and classification of mCP stars in the present study.

Table 2.

Spectral classifications derived by manual classification and the MKCLASS_mCP code.

Table 3.

Conditions employed to flag the presence of an overabundance from the “raw” MKCLASS_mCP output.

Table 4.

Kinematic and astrometric data for the ten stars of our sample with a height larger than 1200 pc above the Galactic plane.

Table 5.

Open cluster members among our sample stars.

Table 6.

Comparison of the spectral types derived in this study, the RM09 catalogue, and the compilation of Skiff (2014).

Table 7.

Comparison of the spectral types derived in this study to the catalogue of Qin et al. (2019).

All Figures

thumbnail Fig. 1.

4800 Å to 5700 Å region of (from top to bottom) the non-CP A0 V star LAMOST J194655.00+402559.5 (HD 225785), a synthetic spectrum with Teff = 9750 K, log g = 4.0, [M/H] = 0.0 and a microturbulent velocity of 2 km s−1, and the newly-identified Si-strong mCP star LAMOST J025951.09+540337.5 (#78; TYC 3701-157-1). The position of the characteristic 5200 Å depression and the Si II lines at 5041 Å and 5055/56 Å are indicated. LAMOST spectra have been taken from DR4.

In the text
thumbnail Fig. 2.

4700 Å to 5700 Å region of (from top to bottom) the LAMOST DR4 spectra of the mCP stars LAMOST J034458.31+464848.7 (#139; TYC 3313-1279-1), LAMOST J040642.34+454640.8 (#180; HD 25706), LAMOST J072118.92+223422.7 (#792; TYC 1909-1687-1), and the late-type “impostor” LAMOST J001159.88+435908.5 (GSC 02794−00977). The position of the characteristic 5200 Å depression is indicated.

In the text
thumbnail Fig. 3.

Blue-violet region of (from top to bottom) the F0 V standard spectra from the liblamost, libsynth, libnor36, and libr18 standard libraries.

In the text
thumbnail Fig. 4.

Showcase of three newly identified “hot” mCP stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J035046.03+363648.2 (#151; Gaia DR2 220081859486642816), LAMOST J195631.74+253407.8 (#929; Gaia DR2 2026771741029840128), and LAMOST J062529.84−032411.9 (#576; TYC 4789-2924-1). MKCLASS final types and, where available, manual types derived in the present study are indicated. Some prominent lines of interest are identified. The asterisk marks the position of a “glitch” in the spectrum of LAMOST J062529.84−032411.9.

In the text
thumbnail Fig. 5.

Showcase of three newly identified “cool” mCP stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J034854.70+521413.1 (#150; Gaia DR2 251609324623302400), LAMOST J052816.11−063820.1 (#344; TYC 4765-708-1), and LAMOST J062221.82+595613.0 (#561; TYC 3776-269-1). MKCLASS final types and, where available, manual types derived in the present study are indicated. Some prominent lines of interest are identified.

In the text
thumbnail Fig. 6.

Upper panel: comparison of the blue-violet spectra of the mCP star LAMOST J065647.94+242958.8 (#732; TYC 1898-1408-1; manual type: A0 V SiSrCr; MKCLASS final type: B9.5 V Sr) to the liblamost A0 V standard spectrum. Lower panel: comparison of the blue-violet spectra of the mCP star LAMOST J052816.11−063820.1 (#344; TYC 4765-708-1; manual type: kA1hA9mA9 SrCrEu; MKCLASS final type: kA1hA8mA9 SrEu) to the liblamost F0 V standard spectrum. Some prominent lines of interest are indicated. We note the weak Ca II K line with a peculiar profile and the unusual profile of the Hϵ line in LAMOST J052816.11−063820.1.

In the text
thumbnail Fig. 7.

Histograms of the G magnitudes (upper panel) and distances from the Sun (lower panel). For the construction of the latter, only stars with absolute parallax errors less than 25% were employed.

In the text
thumbnail Fig. 8.

(BP − RP)0 versus MG, 0 diagram of our sample stars, together with PARSEC isochrones for solar metallicity [Z] = 0.020 (listed are the logarithmic ages). The arrow indicates the reddening vector for the maximum expected error due to the employed reddening map and the parallax error.

In the text
thumbnail Fig. 9.

Mass versus fractional age on the main sequence (τ) distribution assuming solar metallicity [Z] = 0.020 for the 903 sample stars fulfilling the imposed accuracy criteria. Upper panel: density plot for masses up to 4 M. The position of the spectral types has been based on the information given in Pecaut & Mamajek (2013).

In the text
thumbnail Fig. 10.

Distributions of errors for the derived fractional ages on the main sequence (τ; upper panel) and masses (lower panel) assuming solar metallicity [Z] = 0.020.

In the text
thumbnail Fig. 11.

Distribution of [Z] values for the CP2 stars from the Ghazaryan et al. (2018) catalogue with a least three measurements of C, N, O and S.

In the text
thumbnail Fig. 12.

Lines of constant fractional ages on the main sequence (τ) for solar metallicity [Z] = 0.020 (upper panel). Lower panel: positions of the ZAMS and TAMS for isochrones with [Z] = 0.008, 0.020, and 0.060. Values have been chosen to cover the main range of [Z] values found for CP2 stars (Fig. 11).

In the text
thumbnail Fig. 13.

Mass versus fractional age on the main sequence (τ) distributions for isochrones with [Z] = 0.008, 0.020, and 0.060, illustrating the differences in the derived mass and age distributions.

In the text
thumbnail Fig. 14.

Distribution of fractional ages on the main sequence (τ) among the 903 sample stars fulfilling our accuracy criteria.

In the text
thumbnail Fig. 15.

Distribution of the 942 stars with absolute parallax errors less than 25% in the [XY] plane. Stars were divided in probable members of the thin and thick disk according to the scale heights given in Ojha (2001), Aumer & Binney (2017). Ten stars have Z values larger than 1200 pc and might be Halo objects.

In the text
thumbnail Fig. 16.

Blue-violet spectra of the proposed halo stars LAMOST J122746.05+113635.3 (#876; MKCLASS final type B8 IV Si (He-wk); upper spectrum) and LAMOST J150331.87+093125.4 (#880; MKCLASS final type A8 V SrCrEu; lower spectrum). Some prominent lines of interest are indicated.

In the text
thumbnail Fig. 17.

Fractional distribution of chemical peculiarities versus hydrogen-line spectral type for the 876 stars with unambiguous peculiarity type identifications. The numbers above the bars indicate the number of objects in the corresponding spectral type bin. Because a single object may have multiple peculiarities, fractions may exceed 1.

In the text
thumbnail Fig. 18.

Distribution of stars in which only strong blends at 4077 Å and/or 4130 Å were identified.

In the text
thumbnail Fig. 19.

Showcase of five newly identified peculiar mid-B type stars, illustrating the blue-violet region of the LAMOST DR4 spectra of (from top to bottom) LAMOST J014940.99+534134.2 (#37; TYC 3684-1139-1), LAMOST J052118.97+320805.7 (#318; HD 242764), LAMOST J055023.89+261330.2 (#421; TYC 1866-861-1), LAMOST J062307.91+264642.0 (#565; Gaia DR2 3432273606513132544), and LAMOST J062348.46+034201.1 (#567; HD 256582). MKCLASS final types and manual types derived in the present study are indicated. Some prominent lines of interest are identified. The asterisk marks the position of a “glitch” in the spectrum of LAMOST J014940.99+534134.2.

In the text
thumbnail Fig. 20.

ASAS-SN light curve of the eclipsing binary system LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1). The data have been folded with the orbital period of Porb = 5.1435 ± 0.0012 d.

In the text
thumbnail Fig. 21.

Comparison of the blue-violet spectra of the eclipsing binary system LAMOST J034306.74+495240.7 (#135; TYC 3321-881-1) to the liblamost B9 III standard spectrum (upper spectrum). Some prominent lines of interest are indicated.

In the text
thumbnail Fig. 22.

Comparison of the blue-violet spectra of the propsed SB2 system LAMOST J050146.85+383500.8 (#272; HD 280281; MKCLASS final type B8 V Si) to the libsynth B8 V standard spectrum (upper spectrum). Some prominent lines of interest are indicated.

In the text
thumbnail Fig. 23.

Close-up view of the Hγ region of the proposed SB2 system LAMOST J050146.85+383500.8 (blue spectrum) and the libsynth B8 V standard (black spectrum), illustrating the peculiar profile of the Hγ line indicative of binarity.

In the text
thumbnail Fig. 24.

Comparison of the SED of LAMOST J050146.85+383500.8 (red dots) to a Kurucz ODFNEW/NOVER model with Teff = 12 500 K (black squares). The model was forced to either fit the ultraviolet (lower model) or optical flux (upper model). The discrepancies are clearly visible, the star’s SED cannot be fitted with a single star flux model.

In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.