New Oe stars in LAMOST DR5

Stars of spectral type Oe are very rare. To date, only 13 Oe stars have been identified within our Galaxy. In this paper, we present six new Oe stars and four new B0e stars found in LAMOST DR5. Repeated spectral observations of the same Oe stars show some emission line variability. The H$\beta$ emission of TYC 4801-17-1 shows rapid V/R variation. Phase lags in the V/R ratio of TYC 4801-17-1 spectra are also seen. We found the unusual O4.5 star RL 128 is an Oe star with variable H$\alpha$ intensity and its Ca II triplet emission appears when H$\alpha$ emission reaches maximum intensity. These newly identified early type Oe and B0e stars significantly increase the known sample.


INTRODUCTION
The classic Oe spectral stellar type was defined by Conti & Leep (1974) as O type spectra showing emission in the hydrogen lines, but without N III λ4634-4640-4642 or He II λ4686 emission features. Some Oe stars additionally show emission in their He I lines. Stars of type Oe are often characterized by rapid rotation and double-peaked Balmer emission lines. Differences between the origins of P Cyg profile (from strong stellar winds, not a characteristic of Oe stars) and the double-peaked emission profile characteristic of the Oe spectral type was clearly noted in Fig. 11 of Conti & Leep (1974).
Stars of type Oe are very rare (Negueruela et al. 2004). To date, there are only 13 Oe stars known in the Galaxy, reproduced here in Table 1. The Galactic O Star Spectroscopic Survey (GOSSS) (Sota et al. 2011(Sota et al. , 2014Golden-Marx et al. 2016) confirms an Oe/O ratio of 0.03 ± 0.01. For comparison, Zorec& Briot (1997) presents a mean Be/B ratio of at least 17%, and a B1e/B1 ratio is even 34%. These statistics are consistent with Oe stars representing the hot end of the Be temperature distribution, with the mechanism, a rotating disk, being common to Be and Oe stars. The rarity of Oe stars can be attributed to the hot strong O star stellar winds blowing away the envelope around the star and in most cases preventing a formation a rotating disk, where characteristic (Oe and Be) double-peaked emission lines originate. However, Vink et al. (2009) suggested Oe stars earlier than O9.5 have a different origin from classic Be stars.
Metallicity can affect the stellar wind, as stars with lower metallicity tend to have weaker winds. A stellar wind can take away angular momentum, so more metal-poor star with less wind rotates more rapidly. Thus, a metal-poor star can form a rotating disk more easily, which can be retained even by the much hotter star. Golden-Marx et al. (2016) presented about 30 classical Oe stars in Small Magellanic Cloud (SMC). The Oe/O frequency in SMC is 0.26 ± 0.04, much higher than that in the Galaxy. There are also 4 known LMC Oe stars with types as hot as O6 and O7. One of them, star 77616 even shows double-peaked He II λ4686 emission (Golden-Marx et al. 2016). Similar to classical Be stars, Oe stars often show significant emission line variability: Rauw (2015) found that Hα, Hβ, and He I λ5876 of two original Oe stars HD 45 314 and HD 60 848 showed strong variations. For HD 60 848, the variations of the equivalent widths of these lines are obviously asynchronous, while for HD 45 314, the emission lines are highly asymmetric and display strong line profile variations, and in 2013, these lines even changed from double-peaked to single-peaked. HD 120 678 also shows complex both light and spectral variations (Gamen et al. 2012). The origins of these variations are still unknown. A summary of long-term V/R (the ratio of violet peak strength to the red peak strength) variations can be found in Rivinius et al. (2013) and references therein.
Earlier stellar types have stronger stellar winds, so most Oe stars are later O type stars. The spectral classification of an O star is based on the ratio of the equivalent width of He II λ4542 to He I λ4471 where later O types have a small ratio. This standard, however, is not reliable for the Oe class. Frost & Conti (1976) noticed that He I λ4471 of HD 39 680 and HD 60 848 may suffer from emission infilling, which result in higher EW(He II λ4542) / EW(He I λ4471), thus mimicking earlier spectral types. All the spectral types given in Table 1 are from the standard classification methodology -some may be cooler than their spectral type suggests.
Details of the formation of the rotating "decretion" (outwardly moving) disks in Oe (and Be) stars remain complex and puzzling (Struve 1931;Zorec et al. 2016Zorec et al. , 2017Lee et al. 1991;Osaki 1986;Rivinius et al. 2001). Undoubtedly, a larger Oe sample can help understand the nature of this type of emission-line star.
Further identifying characteristics of Oe (and related Be) spectra to be kept in mind as we search the LAMOST DR5 sample for new members of the class: Porter & Rivinius (2003) summarized: (1) Classical Be stars are non-supergiant (i.e. luminosity class V to III) B stars with line emission in Balmer lines (and possibly other lines). The Be phenomenon is complex: not all B stars with emission lines are classical Be stars (see their Table 1). The definition of a classical Be star as a non-supergiant also extends to Oe stars. (2) The emission lines in classical Be stars originate in a decretion disk due to the high rotational speed of the star. (3) Be stars have different types of variability associated with the star and the disk. More specifically, the disk can disappear from time to time. That is why Be stars as usually defined as objects that "have, or had at one time ... Balmer lines in emission" (Collins 1987). One differentiates between a "Be spectrum" and a "Be star", as a Be star can have a Be spectrum permanently or only on occasions. Table 1 has a heterogeneous origin, where two or more spectral types are assigned to one star which imply that possible disappearance of the disk, possible errors in the classification or/and different part of a spectrum used. For example, ζ Oph only shows emission of Hα line in Conti & Leep (1974) and does not appear as an Oe star in Sota et al. (2014), where the spectrum does not cover the wavelength of Hα line. However, even in Negueruela et al. (2004), ζ Oph is still not an Oe star. In fact, its quiescence time is much longer than emission episode (Kambe et al. 1993). HD 60 848 is an O9.5 star in Negueruela et al. (2004) compared to O8 in other two references, which indicate that He I λ4471 emission from disk affect the spectral classification.
Regarding O stars, there are several categories that include emission lines without being Oe stars. Indeed, Oe stars are a minority among O stars with emission lines. Sota et al. (2011Sota et al. ( , 2014 list the other types of O stars with emission lines: Ofc, Onfp, Of?p, early Of/WN, and O Iafpe. We also note that, as the resolution of the LAMOST spectrographs are typically R∼1,800, when classifying Oe stars we should distinguish between observed/strict/uncorrected spectral types and those corrected for the infilling of He I 4471, something that can be done only with good-quality high-resolution spectroscopy (see e.g. the paragraph on V441 Pup in (Maíz Apellániz et al. 2016). In spite of these complexities, we believe uncovering more samples of the Oe phenomenon is important toward further progress in understanding their origins.
The paper is organized as follows: In Section 2, we introduce LAMOST and the selection method of the sample; new Oe and B0e stars found in LAMOST DR5 are presented in Section 3 and Section 4 respectively; Finally, conclusions are given in Section 5.

INTRODUCTION TO LAMOST AND SELECTION METHOD OF SAMPLE
The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST, also called Guoshoujing Telescope) (Cui et al. 2012;Wang et al. 1996;Su & Cui 2004) is a special reflecting Schmidt telescope with 5 • field of view and effective aperture 3.6m -4.9 m. It accommodates 4,000 fibers on its focal plane and can obtain nearly 4,000 spectra during one exposure. Its wavelength coverage is 3650Å-9000Åwith R ∼ 1,800. Each of 16 spectrographs records images of 250 fibers on two 4K × 4K CCDs. As of the end of July 2017, more than 9 million spectra have been obtained (see http://dr5.lamost.org/).
We select candidate O type stars in LAMOST DR5 by using the O type spectral star criteria given by Maíz Apellániz et al. (2016), then the Oe candidates stars are further selected by eye. We checked all 2D and 1D LAMOST spectra of individual exposures of Oe stars and rejected bad pixels or pixels contaminated by cosmic rays or poorly subtracted sky lines. All candidate Oe stars were previously cataloged as being of early type (usually late O or early B), though some of the previous observations date back decades. Some, but not all had emission noted, but none of our Oe (B0e) identified objects had previously been classified of the Oe (B0e) type.
We note that the wavelengths in LAMOST spectra are all in vacuum, but to be coherent with the literature, lines are still named by their air wavelengths. The spectral classification methodology and luminosity criterion used in this paper are from Sota et al. (2011), Sota et al. (2014) and Maíz Apellániz et al. (2016.

NEW OE STARS IN LAMOST DR5
Details of spectra of the six stars that we newly identify from LAMOST spectra as meriting the Oe spectral class as given here.
EM* GGR 149 In SIMBAD, it was assigned B0 (Brodskaya 1953;Gonzalez & Gonzalez 1956) or OB (Kohoutek & Wehmeyer 1999;Wackerling 1970). We found two spectra of this star in LAMOST DR5. In Fig. 1, the ratios of He II λ4542 / Si III λ4552 indicate its spectral type should be O9.7. The ratios of He II λ4686 / He I λ4713 indicate it is a dwarf, but the the Si IV λ4089 and λ4116 lines imply its luminosity class IV. We notice that He I λ5048 has emission wings, so He I λ4713 may suffer from infilling. The round tips of He II λ4686 indicate it is a rapid rotator. Thus, its spectral type should be O9.7 IVn.
Emission from Balmer series lines through Hǫ and Fe II λ5018 and λ5169 can be seen with intensity variations between two epochs, as shown in Fig.s 1 and 2. He I λ5048 shows double emission wings. Fig. 3 illustrates its Paschen series, O I λ8446 and N I emission lines.   TYC 4801-17-1 = EM* RJHA 83 It Hα emission was firstly reported by Robertson & Jordan (1989).We have three spectra in LAMOST DR5. In these three spectra, He II λ4542 / He I λ4388 and He II λ4200 / He I λ4144 indicate its spectral type O9.5. We notice that He I λ4713 and λ5016 are double-peaked emission lines, but weak Si IV λ4089 and λ4116 lines imply it is a dwarf. The broad He lines indicate it is a rapid rotator. Thus, we assign it a spectral type O9.5 Vn.
In Fig. 1, the most outstanding difference between these three spectra is the variation of the single-peaked [O III] λλ5007 emission, while its neighbor, double-peaked He I λ5016, remains unchanged.
The double-peaked Hα and Hβ emission lines can be seen in Fig. 2 and 1, respectively, with V/Rs ascending from MJD57795 to MJD57797. Its double-peaked Paschen series emissions are shown in Fig. 3. The V/Rs of Hα and Hβ are obviously greater than 1, while V/Rs of Paschen lines are all around 1. The N I λ8680/3/6 emission are obvious, while N I λ8629, λ8712 and λ8719 emission are weak. The double-peaked O I λ8446 emission can be inferred from V/R < 1 of Pa18 and the weak red component of O I λ8446 emission, while the V/R > 1 of Pa13 indicates weak Ca II triplet emission.
Long-term V/R variations of Be stars are well-known since McLaughlin (1961) with a mean period of about 7 years. TYC 4801-17-1 shows clearly rapid V/R variations over three consecutive days. The dramatically ascending V/Rs of Hα and Hβ from MJD57795 to MJD57797 imply that TYC 4801-17-1 may have the rapidest V/R variation in known Oe class to date.
The V/R phase lag is defined as different V/Rs between different emission lines in a spectrum. Wisniewski et al. (2007) reported V/R phase lags between Hα with V < R and other emission lines with V > R in ζ Tau. They supposed that there is a one-armed density wave with a significantly different average azimuthal morphology in the circumstellar disk, such that the different emission lines formed in different radii show different V/R phases, which is confirmed by Stefl et al. (2009) and explained by a global disk oscillation model (Carciofi et al. 2009). Chojnowski et al. (2015) also presented the phase lags of Be stars in APOGGE spectra. TYC 4801-17-1 shows obvious V/R phase lags as have been mentioned above. Specifically, the V/R ∼ 1 at Paschen series and He I λ5016, while the V/R > 1 at Hα and Hβ.
HD 255 055 In SIMBAD, it was assigned O (Wackerling 1970), O9.0Vp(e?)(Cruz-Gonz lez 1974), O9V:p (Hiltner 1956;Morgan et al. 1955), or O9V:pe: (Kohoutek & Wehmeyer 1999). There are five spectra in LAMOST DR5. Only two spectra with Hβ at maximum and minimum phases respectively are shown in Fig. 1. The star is a typical O9.5 star based on the ratios of He II λ4542/He I λ4388 and He II λ4200/He I λ4144 from the spectrum of MJD56968. The He II λ4686/He I λ4713 indicates luminosity class III. Thus, we assign it a spectral type O9.5 III.
Only the Balmer series lines show emission with variation -shown in Fig. 1. Its Hα emission line is saturated in the spectrum of MJD56968, and not given in Fig. 2. The weak emission in the Paschen lines Pa11 and Pa12 can be seen in the spectrum of MJD56968 in Fig. 3, while other Paschen lines are infilled.
It is also cataloged as a Kepler K2 star (EPIC 202060093), showing photometric variation at low frequency without any cause given (Buysschaert et al. 2015).
V* KM Cas It was classified as O9.5V((f)) by Gonzalez & Gonzalez (1956) and Massey et al. (1995). There are two spectra in LAMOST DR5. In Fig. 1, the ratios of He II λ4542/ He I λ4388 and He II λ4200/He I λ4144 are around 1, which indicate its spectral type O9, while its high ratio of He II λ4686/He I λ4713 indicates it should be a dwarf. The broad He lines indicate it is also rapid rotator. However, there is no sign of N III λ4634/40/41 emission. Thus, we assign it a spectral type O9 Vn. In SIMBAD, it is an eruptive variable star, which implies that our spectra might miss the high phase of N III λ4634/40/41 emission. Its Hα emission line shows variation between 2 epochs, as shown in Fig. 2.
LS II +23 14 It was assigned to a spectral type B0Vn by Reed (2003). The spectrum in Fig.1 indicates a spectral type O9.7 from the ratio of He II λ4542 / Si III λ4552, while the intensities of Si IV λ4089 and λ4116 lines indicate its luminosity III. The broad He I lines imply it is also a rapid rotator. Thus, it is an O9.7 IIIn star. Its Hα in Fig. 2 shows that it is an Oe shell star.
RL 128 = ALS 19 265 The unusual properties of this star was noted by Chromey (1979). It is very faint (V = 15.11), and very far from sun (65kpc) and was classified as O7. Recently, Maíz Apellániz et al. (2016) assigned it a spectral type O4.5V((c))z, based on their high quality spectrum. However, they also suspected that it may be a sdO. We found four spectra of RL 128 in LAMOST DR5. All our spectra confirm the spectral type O4.5 Vz, though the SNRs are all very low. RL 128 is also a Kepler K2 star (EPIC 202060098) with light curve period 5.03 days (Buysschaert et al. 2015), which was interpreted as rotation, but Balona (2016) suggest a binary. The maximum velocity difference is about 20 km s −1 between four LAMOST spectra, but it is still in question whether it is a binary or not because of low SNRs and resolutions of our spectra.
For clarity, we only illustrate two spectra in Fig. 4, with Hα reaching its maximum and minimum phases, respectively, among these four spectra. The Ca II triplet also show emission, though weak, when Hα reaches its maximum intensity. The Oe character is based only on Hα, as no Hβ emission is seen.

NEW B0E STARS IN LAMOST DR5
We find 4 new B0e stars in LAMOST DR5. The blue spectra of these B0e stars are shown in Fig. 5, while their Hα lines are shown in Fig. 6.
2MSS J03354517+5141074 We cannot find any information about this star in SIMBAD. In Fig. 5, its Balmer series until Hδ are all shown emissions in their centers. Moreover, Hβ emission is asymmetric. Besides, its He I λ5016, as well as λ5876, λ6678, λ7065, and λ7281 lines also show emissions. Its weak He II λ4542 indicates it is a B0 star, and its strong Si IV λ4089 and λ4116 lines indicate its luminosity III. Though suffering from infillings, its He I lines are all very broad, thus, we assign it a spectral type B0 IIIn.
BD+62 441 There are two spectra of different epochs for this star in LAMOST DR5, which are overplotted in Fig.  5 with the black spectrum from MJD57382 and the red spectrum from MJD57707. It was assigned B2III (Rydstrom 1978), B0 (Heckmann & Dieckvoss 1975) or B0I (Voroshilov et al. 1985). The Weak Si IV λ4089 and λ4116 lines indicate its a B0 V star. Its Hβ of MJD57707 is more emissive than that of MJD57382, though Hβ infilling is weak in either epoch.
HD 265 134 It was assigned B0 by Nesterov et al. (1995) or O9.5III by Bisiacchi et al. (1982). In Fig. 5, its weak He II λ4542 indicates it is a B0 star. The broad He I lines indicate it is a rapid rotator. Considering the shallow Si IV λ4089 and λ4116 are formed from the rotational broadening of intrinsically narrow deep lines, we assign it a spectral type B0 IIIn. Its Hα shown in Fig. 6 indicates it is a Be shell star. No infilling is seen in H β and the "e" suffix is based on Hα only.
BD+51 921 There are three spectra from different epochs for this star in LAMOST DR5. Two of them are overplotted in Fig. 5, with the black spectrum from MJD57798 and the red spectrum from MJD57058. It was assigned to B0II by Hiltner (1956), but in Fig. 5, the similar intensity of Si IV λ4116 to that of He I λ4121 indicates its luminosity class III, thus it is a B0 III star. The variation of Hα profile between three epochs is obvious shown in Fig.  6. No infilling is seen in H β and the "e" suffix is based on Hα only.

CONCLUSIONS
In this paper, we present 6 new Oe stars found in LAMOST DR5 data, increasing the numbers of known Oe stars by nearly 50%. We also list 4 new B0e stars found in LAMOST DR5. TYC 4801-17-1 shows rapid V/R variations at Hβ and a V/R phase lag. Moreover, we find the unusual O4.5 star RL 128 is also an Oe star with variable Hα intensityrare Ca II triplet emission appearsat the maximum phase of Hα emission.
All new Oe and B0e stars with their relevant spectral information from LAMOST DR5 are listed in Table 2 and 3, respectively. The parallaxes and their errors are from Gaia DR2(see https://gea.esac.esa.int/archive/).