A multi-wavelength view of the open cluster NGC 2527: Discovery of active stars

We have analysed data from the XMM-Newton (Jansen, et al., 2001) observations (ObsId = 0057570101) of NGC 2527 centered on $\alpha=08^{h}04^{m}47.20^{s}$, $\delta$ = $-28\degr 07\arcmin 52.0\arcsec$ performed on March 18, 2002, and March 19, 2002 during orbit 432. During these observations, the PN camera (Strüder, et al., 2001) was active in extended full frame mode, and the MOS cameras (Turner, et al., 2001) were active in the full frame mode, using the medium filter. We have analysed only the deepest XMM-Newton EPIC (PN, MOS1, and MOS2) exposures of NGC 2527. Several simultaneous short exposure OM (Mason, et al., 2001) observations in imaging mode with UVM2 (2310 Å) and UVW2 (2120 Å) filters were also used. In the OM observations with UVM2 filter, one of the mosaic windows needed to cover the full field of view was not taken. The observation log of XMM-Newton observations is given in Table  1. The footprints of EPIC (PN and MOS) and OM (UVM2 and UVW2) XMM-Newton observations of NGC 2527 are overplotted on the Digitized Sky Survey (DSS) image of NGC 2527 in Figure 1. The details of EPIC and OM data processing and analysis are given in Section  3.1.

We obtained the Swift UVOT photometric data of NGC 2527 from Siegel, et al. (2019). These data had been processed using the reduction pipeline outlined in Siegel, et al. (2014). The footprints of multiple UVOT observations in UVW2 (1928 Å), UVM2 (2246 Å), and UVW1 (2600 Å) filters are overplotted on the DSS image of NGC 2527 (see Figure 1). Bright sources with high count rates that lead to significant coincidence losses were not included in the Swift UVOT catalogue. We converted the UVOT magnitudes to count rates using the eqn. 1 (Poole, et al., 2008).

where m and C_source is the magnitude and count rate of a source in the UVOT filters, and Z_pt is the zero point for each UVOT filter obtained from Breeveld, et al. (2011). The flux in each UVOT filter was calculated by converting the estimated count rates to fluxes using the conversion factors for Vega magnitude scale. The conversion factors were obtained from Brown, et al. (2016). The magnitudes and extinction corrected fluxes of cluster members in UVW1, UVW2 and UVM2 filters are given in online-only Table A.

Gaia DR2 offers a precise and accurate 5-parameter astrometric solution i.e. positions ($\alpha$ and $\delta$), proper motions in right ascension ($\mu_{\alpha}$) and declination ($\mu$ _δ), and parallaxes ($\varpi$), magnitudes in three photometric filters (G, $G_{BP}$, and $G_{RP}$) for more than 1.3 billion sources and radial velocities (RV) for more than 7 million stars (Gaia Collaboration, et al., 2018). Cantat-Gaudin, et al. (2018b) used precise and accurate astrometric (proper motion and parallax) measurements from Gaia DR2 catalogue to determine the cluster membership probability of stars for 1229 OCs in the Milky Way. They have excluded sources fainter than G = 18 mag because of high uncertainties in parallax and proper motion values for such sources. NGC 2527 is part of the Gaia DR2 open cluster population in Milky Way Catalogue (Cantat-Gaudin, et al., 2018b). The catalogue gives membership probabilites of 400 potential members stars that lie around the center of NGC 2527. Out of these 400 stars, only 356 stars have membership probability $>50\%$ that are considered as highly probable members of NGC 2527. As per Cantat-Gaudin, et al. (2018b), half of the cluster members are present within a a circle of radius $\approx 25\arcmin$ from the cluster centre as shown on the DSS image of NGC 2527 in Figure 1. The Gaia DR2 catalogue gives radial velocity of 43 bright highly probable members of NGC 2527. The mean radial velocity of these stars is 41 $\pm$ 9 km/s. It is consistent with the previous estimates of radial velocity (42.4 $\pm$ 1.3) by RAVE survey of OCs (Conrad, et al., 2014).

Raw data obtained from the XMM-Newton Science Archive were processed using the XMM-Newton Science Analysis Software (SAS 17.0.0). SAS tasks ’emchain’ and ’epchain’ were used to generate event files for the MOS and PN detectors, respectively. These tasks perform energy and astrometry calibration of events registered on each CCD chip and combine them into a single data file. The event lists of both MOS and PN were filtered using SAS task ’evselect’ in the 0.3-7.0 keV energy band and only the single, double, triple, and quadruple pixel events were retained as genuine X-ray events. This removes the contamination due to low pulse height events. Furthermore, the data from all three cameras were individually screened for soft proton background flaring, by applying the criterion of total count rate for single events with energy $>$ 10 keV that exceeded 0.4 counts $s^{-1}$ and 1.0 counts $s^{-1}$ for the MOS and PN detectors respectively. The time intervals affected by soft proton flares were thus excluded from the analysis. The observation log and useful exposure time for XMM-Newton observations are summarised in Table  1.

We produced images in two energy bands, a soft band (0.3-2.0 keV), and a hard band (2.0-7.0 keV) for all three EPIC detectors using the ’evselect’ and ’emosaic’ SAS tasks with the filtered MOS and PN event lists. During the imaging process, only the events with FLAG = 0, PATTERN$\leq$4 (single and double pixel events) in the hard band and PATTERN == 0 (single pixel events) in the soft band were selected. These events were selected to reject noise at the extremities of the PN detector along Y-direction. Default image binning of 600 $\times$ 600 pixels was chosen, similar to the one used in the 3XMM catalogue.

We used the SAS task ’edetect$\_$chain’ to search for sources present in both the soft and hard bands in the XMM-Newton EPIC (PN, MOS1 and MOS2). The ’edetect$\_$chain’ task is a script which runs a series of subtasks to detect point sources from the input filtered event files. It determines the source parameters such as coordinates, count, count rates, etc through simultaneous maximum likelihood PSF (point spread function) fitting to the source count distribution in the given energy bands of each EPIC instrument. A combined Maximum Likelihood (ML) value for all three instruments was taken to be greater than 10. We detect 64 sources in the soft band and 34 sources in the hard band. The merged PN and MOS soft band X-ray image of NGC 2527 smoothed with a 2D Gaussian $\sigma$ = 2 pixel is shown in Figure 2.

The OM observations taken in Imaging mode were processed using the SAS task ’omichain’ with default parameters. This task is a script which detects sources in the OM image and performs aperture photometry on the detected sources. It processes the data from multiple exposures simultaneously and returns various data products. We obtained images and source lists for individual exposures in each filter, mosaiced images, and merged source list from all the observations. The merged source list contains information of source position, count rates, fluxes, magnitudes, and quality flags. The SAS pipeline corrects the count rates and fluxes for coincidence losses, dead time and time dependent sensivity of the detector. The combined multiple exposures of OM images in UVW2 and UVM2 filters is shown in Figure 3. A total of 729 sources were thus detected in the merged OM catalogue. The extinction corrected fluxes and the quality flags of cluster members detected in OM are given in online-only Table A. The SAS pipeline provides a quality flag for each OM source. The sources with quality flag equal to zero are good detections. SAS task detects a significant number of sources which have quality flag $>$ 0 that lie on a readout streak, a bad pixel, near a bright source, near the edge or on a star-spike which are spurious detections and sources with high count rates (rate > 0.97 c/frame). Out of the 729 sources, only 316 detected in UVW2 filter and 217 sources in UVM2 are good detections and not artefacts. The OM sources with bad quality flags were excluded from the subsequent analysis for Spectral Energy Distributions.

In order to determine the nature of sources detected in Swift UVOT and XMM observations, we searched for their optical counterparts in the Gaia members catalogue of NGC 2527.

We determined the optimum cross-correlation separation between soft band X-ray catalogue and optical counterparts using the method outlined by Jeffries, et al. (1997). We find that the best cross-correlation radius of 3″ optimizes the number of real identifications while minimizing the number of spurious detections due to background sources. The details of the best fit cross-correlation function is given in Appendix A. Within a radius of 3″, 12 soft band X-ray sources possess a unique optical counterpart that is a cluster member. None of the hard band X-ray sources has an optical counterpart that is a cluster member within a radius of $\leq$ 30″ from the hard band X-ray source position, suggesting that the sources in the hard X-ray band are not members of the cluster. X-ray sources that have a unique optical counterpart that is a proper motion member of the cluster have been considered as real sources, and other sources are considered as spurious detections. To identify optical counterparts to the UV (OM/UVOT) sources, we assumed a cross-correlation radius of 3″. Within a 3″ radius, we find that 44 XMM-OM, and 54 UVOT sources have unique optical counterparts (see Table 2).

The total Galactic column density along the line of sight of NGC 2527, but extending to the edge of the Galaxy as given by the (HI4PI Collaboration, et al. 2016; see HEASARC $N_{H}$ column density tool¹¹1https://heasarc.gsfc.nasa.gov/cgi-bin/Tools/w3nh/w3nh.pl) is N_H = 44.2 $\times$ 10²⁰ cm^-2. To estimate the number of extragalactic sources in the field of view (fov) of XMM-Newton, we assumed a power-law model with a photon index $\Gamma$= 2, and the quoted value of N_H. An upper limit on the background count rate in the hard band for EPIC MOS1 observations was found to be $\sim 0.002$ counts/s (see also Sect. 3.6). Since the log N $-$ log S distribution derived from the ROSAT deep sky survey (Hasinger, et al., 1993) gives flux in the 0.5 $-$ 2.0 keV energy band, we determined the total flux for the power-law model in this energy band using the HEASARC WebSpec²²2https://heasarc.gsfc.nasa.gov/webspec/webspec.html tool. For a count rate of $\sim$0.002 counts/s, exposure time of 25300 sec, the $\Gamma$ = 2 power law with N_H = 44.2 $\times$ 10²⁰ cm^-2 and normalization = 1.63 $\times$ 10^-5 has a flux of 1.3 $\times$ 10^-14 ergs cm^-2 s^-1 and 3.1 $\times$ 10^-14 ergs cm^-2 s^-1 in 0.5-2.0 keV and 2.0-7.0 keV energy band respectively. A comparison of the background flux in the 0.5-2.0 keV energy band with the log N $-$ log S distribution suggests that approximately 30-35 extragalactic sources are likely to be detected above a maximum likelihood threshold of 10. The number of hard band X-ray sources detected in our XMM-Newton observations is consistent with the estimated number of background sources.

X-ray properties of cluster members are given in Table 3. Positions of X-ray, OM - UVW2, and UVW1 sources which have unique optical counterparts that are cluster members are shown in Figure 2 and Figure 3. In the OM images, sources with good and poor quality flags are marked with green and red colours respectively to distinguish between good and bad quality detection. A unique object id SwiftXMMOM (SXOM) has been assigned to the cluster members that are detected in the XMM and Swift UVOT observations. Sources that have been detected only in the XMM-OM and not in the Swift UVOT observations are considered as spurious detections. SXOM30 and SXOM46 are two such sources which were excluded from further analysis.

The Hertzsprung-Russell (H-R) Diagram or the Color-Magnitude Diagram (CMD) of OCs can be used to distinguish between stars in various stellar evolutionary stages and identify peculiar stars that deviate from standard evolutionary models. These are essential tools to determine the fundamental properties of a cluster like age and extinction. These parameters can be determined by fitting standard stellar evolutionary tracks or isochrones to the observed CMD.

The optical CMD of NGC 2527 based on the Gaia magnitudes (G, $G_{BP}-G_{RP}$) is shown in Figure 4. The main sequence, as well as the binary sequence, can be distinctly identified in the optical CMD. Previous ground-based optical photometric studies of NGC 2527 using CMD (Lindoff, 1973; Dodd, et al., 1977) were limited only to a few cluster members and the distance to NGC 2527 was not well known. Therefore, the fundamental parameters of NGC 2527 given in the literature have high uncertainity. The distance to NGC 2527 can now be calculated fairly accurately from the average of Gaia DR2 parallaxes of cluster members. We have corrected the Gaia DR2 parallaxes for bias by adding +0.029 mas to the parallax values as prescribed by Lindegren, et al. (2018) and inferred the mean distance to NGC 2527 to be $642\pm 30$ pc, which corresponds to a distance modulus of $G-M_{G}$ = 9.0 $\pm$ 2.4 mag. The PARSEC-COLIBRI isochrone (Marigo, et al., 2017) of log(age) = 8.8 or 630 million years with a reddening of E(B-V) = 0.13, fixed metallicity [Fe/H] = -0.1 (Reddy, et al., 2013) and distance modulus of $G-M_{G}$ = 9.0 mag is found to match the observed MS and main sequence turn off (MSTO) very well. The extinction coefficients used for the Gaia and UVOT filters were obtained from the VOSA Filter Profile Service (Rodrigo, et al., 2012). The PARSEC-Isochrone with these parameters along with an equal mass binary isochrone are superimposed over the observed optical Gaia CMD in Figure 4.

Of the 44 cluster members detected in the OM only 16 (and 25) have good quality flags in the UVM2 (in the UVW2) filter, as compared to 51 cluster members in each UVOT (UVM2, UVW2, and UVW1) filter. Since there are fewer sources in OM, it is difficult to distinctly visualize the main sequence in UV-optical CMDs with OM magnitudes. Therefore, we only use UVOT magnitudes for UV-optical CMDs. The UV-optical CMD based upon UVOT photometry in the UVW2, UVM2, and UVW1 filters is shown in Figures 5, 6 and 7. The PARSEC-COLIBRI isochrones with the same parameters as the optical isochrones are overlaid on UV-optical CMDs. The stars near the main sequence turn-off are not detected in the Swift UVOT because they are saturated or lie outside the fov of Swift UVOT. The PARSEC isochrones fit the MS very well in the UV-optical CMDs based on the UVW1-G and UVM2-G colours. The faint MS stars in the UVW2-G CMD have an excess blue colour as compared to the PARSEC isochrones. Siegel, et al. (2019) had also found an excess in the UVW2 - UVW1 colours of faint MS stars as compared to the isochrones.

We checked the position of each star in the Gaia optical CMD and UV-optical CMD, and have classified those which lie close to the binary isochrone as Binary (B). Similarly those lying close to the single star isochrone have been classified as Main Sequence (MS), the ones near the red giant phase of the isochrone as RGB, and the ones lying between single star and binary isochrone in the optical CMD but closer to the binary isochrone in the UV CMD as Potential Binaries (PBs) (see Table 3). A few stars that deviate from the standard evolutionary models are classified as red straggler (RS) or sub-subgiant (SSG) depending upon whether they lie in the RS or the SSG region. As per Geller, et al. (2017), SSG refer to stars that are fainter than the normal subgiants and redder than the normal MS star, whereas the RS stars are brighter than the normal subgiants and redder than the normal RGB. The fundamental parameters of the peculiar stars that lie outside the fov of XMM and Swift UVOT detectors are given in the appendix Table D. Below, we focus only on the stars which are detected in either XMM-OM, Swift UVOT, and XMM-EPIC observations.

A bright X-ray source SXOM61 is the only source that is found to have a significant number of counts ($>$ 100) for spectral analysis. We extracted PN and MOS spectra for the source and background regions. The source spectrum was extracted from a circular region of radius 30″centred on the X-ray source. For the MOS, the background spectrum was extracted from an annulus with inner radius of 40″and outer radius of 60″centered on the X-ray source. For the PN, such an annular region coincided with PN chip gap, so the background spectrum was extracted from a circle with a radius of 60″close to the source position. We used the SAS task ’especget’ to extract these spectra from the filtered PN and MOS event files. We grouped the spectral (PHA) channels using the ’specgroup’ task such that each grouped PHA channel had at least 16 counts, sufficient for the assumption of Gaussian distribution of uncertainties. This task also performs the subtraction of the background spectrum from the source spectrum.

We performed the spectral fitting using the XSPEC (Arnaud, 1996) software package. The PN, MOS1 and MOS2 background subtracted source spectra were loaded in XSPEC, and the spectral groups containing channels with energies $<$ 0.3 keV and $>$ 2.0 keV were ignored. The X-ray spectra of SXOM61 obtained from PN, MOS1, and MOS2 were fitted simultaneously using collisionally-ionised plasma (apec) model with one characteristic plasma temperature multiplied with an absorber model (tbabs). We estimated the hydrogen column density $N_{H}$ value from the standard relation for dust and extinction for our Milky Way Galaxy, $N_{H}/A_{V}$ = (2.08 $\pm$ 0.02) $\times$ 10²¹ H cm^-2 mag^-1 (Zhu, et al., 2017). For an E(B-V) = 0.13 along the line of sight of NGC 2527, we found that $N_{H}$ = (8.12 $\pm$ 0.08) $\times$ 10²⁰ cm^-2, and we fixed the column density to this value in the tbabs model. We used a common value for the metallicity of all the elements relative to the solar abundance values as given by Asplund, et al. (2009). We used the $\chi^{2}_{red}$ minimization method to determine the best fitting model. At first we determined the best fit parameters for the models using the elemental abundance values fixed to 1 relative to the solar. The observed X-ray spectra of SXOM61, along with the best fitting model are shown in Figure 8. The best fit model parameters are given in Table 4. We also tried a fixed value for the sub-solar metallicity, fixed to 0.3 solar usually seen in many X-ray active late type stars Briggs & Pye (2003) and performed the spectral fitting again and obtaining the best fit temperature of the plasma and its normalization. There was little change in the $\chi^{2}_{red}$ value in between the sub-solar and solar fits (see Table 4). We estimated the X-ray flux coming from SXOM61 in the 0.3-2 keV energy band for both the solar and sub-solar fits using the task ’flux’ in the XSPEC. The total flux in the 0.3-2 keV energy band of SXOM61 from the X-ray spectrum is (2.1 $\pm$ 0.2) $\times$ 10^-14. There is no significant difference in the estimated flux from a solar and sub-solar fits (see Table 4).

We obtained the energy conversion factor (ECF = Flux/Rate) using the estimated flux and MOS1 count rate of SXOM61 in the 0.3-2 keV energy band. The value of the ECF is found to be 1.73 $\times$ 10^-12 ergs cm^-2 counts^-1. We used this value of ECF to calculate the X-ray fluxes of the other 11 X-ray cluster members. The corresponding X-ray Luminosity was calculated using the formula L = 4$\pi$ $d^{2}$ $F_{X}$ where $F_{X}$ is the X-ray flux of sources and d is the distance of source obtained from Gaia. The estimated X-ray fluxes and luminosity are reported in Table 3. We used the ’esensmap’ task to generate a background sensitivity map of MOS1 soft and hard band image for an ML of 10 and took it’s median value to estimate the detection limit of rate (counts $s^{-1}$). For a given XMM EPIC image, the ’esensmap’ task (part of the ’emchain’ task) creates a sensitivity map giving point source detection upper limits for each image pixel. The median background count rate in the soft band and hard band in the XMM MOS1 image is 0.00187 counts $s^{-1}$. The detection limit for the X-ray luminosity of cluster members is thus found to lie between 1.35 $\times$ 10²⁹ $-$ 1.77 $\times$ 10²⁹ ergs s^-1 cm^-2.

The optical CMD given in Figure 4 shows that many sources detected in the UV and X-rays are located along single and binary isochrones, while a few other stars deviate from these isochrones. To understand the physical properties of these stars, it is essential to estimate their fundamental parameters that include luminosity ($L_{\odot}$), radius ($R_{\odot}$), mass ($M_{\odot}$), and effective temperature ($T_{eff}$). These parameters can be derived by fitting UV to IR Spectral Energy Distribution (SED) of these stars with Kurucz models (Castelli, et al., 1997). Kurucz model library consists of 3808 discrete stellar atmosphere models with a unique set of $T_{eff}$, metallicity and log $g$ parameters. The $T_{eff}$ of these models ranges from 3500 to 50000 K and log $g$ from 0.0 to 0.5 dex. The available Kurucz model metallicities are -2.5, -2.0, -1.5, -1.0, -0.5, 0.0, 0.2 and 0.5.

The SEDs of X-ray and UV bright stars were constructed by combining observed photometric data from UV (XMM OM - UVW2 and UVM2; UVOT - UVM2, UVW2, and UVW1), optical (Gaia - G, G_BP and G_RP; Paranal - u, g, r, Halpha, and i; MISC/APASS B and B; PAN-STARRS - g, r, z, y and i), near-infrared (2MASS - J, H, and K_s), and mid-infrared (WISE - W1, W2) observations. We used the Virtual Observatory SED Analyzer (VOSA) (Bayo et al., 2008) tool to obtain the fluxes in optical (Gaia, APASS, Paranal, and PAN-STARRS) and IR (2MASS and WISE) bands within a search radius of 5″. We downloaded the Kurucz model synthetic flux from VOSA for these bands. VOSA corrects the observed flux for extinction in their respective wavebands using the extinction law of Fitzpatrick (1999) improved by Indebetouw, et al. (2005) in the infrared. The dereddened fluxes in optical and infrared bands are given in online-only Tables B and C.

Since VOSA only performs single spectrum SED fitting which is not suited for potential binary stars, we developed a python code for SED analysis that can perform both single and composite spectral fitting. The code constructs the observed SED and compares the observed flux with the synthetic Kurucz model flux using $\chi^{2}_{red}$ statistical analysis. The synthetic Kurucz model fluxes are scaled by a variable scaling parameter $M_{d}$ = ($\frac{R}{D}$)², which is optimized using Nelder-Mead optimization technique to get the least $\chi^{2}_{red}$ for each model. For each Kurucz model, the code finds the optimum scaling parameter to minimize the $\chi^{2}_{red}$. It compares the optimum value of $\chi^{2}_{red}$ for each model and returns the best fit model parameters whose $\chi^{2}_{red}$ is closest to 1. The code also estimates the residual flux, where Residual = $\frac{ObservedFlux-ModelFlux}{ObservedFlux}$.

For a single spectrum SED fitting, the $\chi^{2}_{red}$ function is given by eqn. 2. In the case of composite spectrum fitting, the sum of fluxes of two Kurucz models is compared with the observed fluxes using eqn. 3. The composite spectrum obtained by the addition of two Kurucz model spectra is referred to as Kurucz and Kurucz model.

where N is the number of photometric data points, N_fit is the number of free parameters in the model, $F_{o,i}$ is the observed flux, $M_{d}F_{m,i}$ is the model flux of the star, and $\sigma_{o,i}$ is the error in the observed flux, $M_{d}$ = ($\frac{R}{D}$)² is the scaling parameter, R is the radius of the star, and D is the distance to the star.

where N is the number of photometric data points, N_fit is the number of free parameters in the model, $F_{o,i}$ is the observed flux, $F^{1}_{m,i}$ and $F^{2}_{m,i}$ is the model flux of first and second component respectively, $\sigma_{o,i}$ is error in the observed flux, $M_{d}^{1}$ and $M_{d}^{2}$ are the scaling parameters for the first and second component.

We obtained $T_{eff}$, log $g$, and metallicity from the best fitting Kurucz model parameters. We estimated the radius, R_⊙, from the scaling parameter $M_{d}$, and bolometric luminosity (L_⊙) of the star using eqn 4. We found the mass of the star by comparing the estimated luminosity and temperature with the best fitting PARSEC isochrone. The uncertainties in R_⊙ and L_⊙ were estimated using eqn. 5 and eqn. 6 respectively.

where R is the radius of star, $\sigma$ the Stefan-Boltzmann constant and T is the effective temperature of star estimated from SED fitting.

where $\bigtriangleup D$ is the uncertainty in distance for each star given in Table 3.

where $\bigtriangleup$T is the uncertainty in T.

We estimated the fundamental parameters of 53 cluster members that were detected by XMM-OM and Swift UVOT. The metallicity of the NGC2527 cluster is near solar ([Fe/H] = -0.1, Reddy, et al. 2013). The metallicities of Kurucz models that bracket this value are [Fe/H] = 0.0 and -0.5. In our analysis, we thus adopted the [Fe/H] = 0.0 models. Given the near-solar metallicity of the cluster, this simplification is unlikely to have a significant impact on the SED fits.

If a star shows residual flux in excess of more than 50% in UV band, it is classified as having UV excess. 23 out of 53 stars are found to show an excess in the UV band. 10 stars are found to show a prominent dip feature near 2250 Å. The fundamental parameters of these stars are given in Table 5. A closer inspection of PAN-STARRS, 2MASS, and WISE images shows that three stars SXOM12, SXOM17, and SXOM56 are multiple stars that are resolved in the PAN-STARRS image but are unresolved in the WISE and 2MASS images. So the results obtained from SED fitting of these stars are deemed spurious.

Of the remaining 20 stars that show UV excess, 14 are MS stars, 4 are binary stars, and 2 are RGB stars according to their position in the CMD. In order to understand the nature of excess UV residual, and determine the fundamental parameters of both the components in binaries, we performed composite spectrum SED fitting for 4 binaries and 2 RGB stars that show UV excess. In addition, we also performed composite spectrum SED fitting for 4 more binary stars which emit in X-rays. A comparison of single and composite spectrum SED fit of these stars is summarized in Table 6. We present the results below.

In the very early studies of NGC 2527 by Lindoff (1973) and Dodd, et al. (1977), they had estimated the age to be around 500 Myr and 1 Gyr, with reddening E(B-V) = 0.077 $\pm$ 0.016, and a distance modulus of 8.70 mags. Since these studies had limited coverage in area and photometric depth, the sample of cluster members is incomplete, and estimated parameters were not reliable. Cantat-Gaudin, et al. (2018a) determined the membership of cluster stars using Gaia-TYCHO astrometric solution and found the age of NGC 2527 to be around 1 Gyr, metallicity [Fe/H] = -0.26 $\pm$ 0.11, E(B$-$V) = 0.077 $\pm$ 0.016 and V$-$M_V = 8.37 $\pm$ 0.10 mag by fitting isochrones to CMD based on 2MASS magnitudes. They used an automated code BASE-9 to determine these parameters, but the estimated metallicity is relatively lower as compared to the average value of [Fe/H] = -0.1 (Reddy, et al., 2013) obtained from spectroscopic studies of RGB stars. Also, the distance modulus was lower than that obtained from Gaia DR2. Siegel, et al. (2019) fitted PARSEC isochrones to UV CMDs and found an age of 450 Myr, with a distance modulus of V$-$M_V = 8.7 and a reddening E(B$-$V) = 0.04 mag. They assumed a lower distance modulus obtained from early studies of the cluster by (Dodd, et al., 1977), which led them to estimate a younger age for the cluster.

In this study, we determined the fundamental parameters of NGC 2527 by fitting PARSEC isochrones to the CMD based on Gaia magnitudes. The unprecedented coverage and depth in the photometry offered by Gaia DR2, allows us to cover a complete sample of cluster members. We find that the age of NGC 2527 is $\sim$ 630 Myr, reddening as E(B$-$V) = 0.13 mag, and distance modulus as $G-M_{G}$ = 9.0 $\pm$ 2.4 mag from the isochrone fitting. The average radial velocity of cluster members from Gaia DR2 radial velocity measurements is 41 $\pm$ 9 km/s.

We detect 3 RGB stars in either X-ray or UV that lie in between single and binary isochrones in the optical CMD. SXOM61 emits in both X-ray and UV. SXOM42 lies in the field of view of XMM-EPIC detectors but is only detected in the UV. SXOM65 emits X-rays but is not detected in the UV observations because it lies outside the field of view of Swift UVOT and XMM-OM detectors. The RGB stars have high proper motion membership probability (given in Table 3), and their Gaia radial velocities are close to the mean radial velocity of the cluster. Therefore the RGB stars are kinematic members of the cluster.

SXOM61 and SXOM42 occupy a similar position in between the single and binary isochrone in UV CMDs, except in UVW2-G CMD. In UVW2-G CMD, they show an excess UVW2-G colour and lie close to the single star isochrone. The single spectrum SED fits to these 3 stars estimate R_⊙ $\approx$ 10 $R_{\odot}$ and T_eff $\approx$ 5500 K respectively, which corresponds to a spectral type G2 III. SED residual plots of SXOM42 and SXOM61 (see Figure 10) show a 75% excess in UVW2 flux as compared to the Kurucz model flux of a G2 III star. The origin of excess flux in the UVW2 band in these stars could be due to chromospheric activity, presence of a hotter companion, or red leakage in the UVW1 filter. The composite spectrum fitting using a combination of Kurucz and Kurucz models does not significantly decrease the excess flux, and the best fit temperature of individual components is exceptionally high, making the fit unreliable. The flux of these RGB stars in UV filters is at least $10^{4}$ times higher as compared to the detection limits suggests that the high flux is unlikely due to red leakage. Hence, the excess in UV is most likely due to high chromospheric activity or a hotter companion.

The X-ray spectrum of SXOM61 shown in Figure 8 is fitted by a single component plasma model that has a temperature of 0.62 $\pm$ 0.06 keV or (7.1 $\pm$ 0.7) $\times$ 10⁶ K. The high plasma temperature is suggestive of a highly active corona. The X-ray luminosity (Lx) of SXOM61 and SXOM65 is 10.2 $\times$ 10²⁹ ergs s^-1 and 3.8 $\times$ 10²⁹ ergs s^-1 suggecting that both the stars have an active corona. As stars evolve off the main sequence, they undergo rapid loss of angular momentum as a result of magnetic breaking through coronal winds. Therefore, the RGB stars are slowly rotating stars with relatively less active corona. It is highly unlikely for a RGB star to have an active corona, unless it is tidally locked in a binary system or a fast rotating merger remnant of a contact binary, like FK Comae. The high coronal activity and excess UV flux hints that these stars are potential FK Comae candidates. It is a rare class of rapidly rotating late-type giants or subgiants, which show signs of active chromospheres and X-ray emission (Bopp & Stencel, 1981).

We detect strong X-ray emission from 12 stars, out of which 2 are potential FK Comae candidates, 5 are RS CVn candidates, and 5 are MS stars which are coronally active. A comparison of X-ray luminosity function (XLF) of FK Comae and RS CVn stars in intermediate to old open clusters is shown in Figure 14. The star clusters used for comparison with NGC 2527 (0.63 Gyr) are : NGC 6633, IC 4756 (0.7 Gyr; Briggs, et al., 2000), M67 (4 Gyr; Mooley & Singh, 2015), and NGC 188 (7 Gyr; Belloni, et al., 1998; Gondoin, 2005). We have shown a comparison with the XLF of field RS CVn binaries (Singh, et al., 1996). NGC 6633 and IC 4756 do not have any significant detection of FK Comae type or contact binaries. The XLF suggest that the X-ray luminosity of both RS CVn and contact binaries increases with age i.e., more active binaries are present in older clusters as compared to younger clusters. Mooley & Singh (2015) have also compared the XLF of OCs of various ages, and they also find a similar trend.

Main sequence stars as they evolve off the main sequence undergo rapid loss of angular momentum as a result of magnetic braking through coronal winds. These massive RGB stars can emit X-rays only if they are fast rotating stars. Webbink (1976) modeled the formation of FK Comae stars by the coalescence of W UMa contact binaries. During the merger, the orbital angular momentum of the binary is converted to rotational momentum of the single star which spins it up. The high X-ray luminosity is most likely due to dynamo action induced by the rapid rotation. Recently Howell, et al. (2016) found 18 rapidly rotating FK Comae type of stars, in a sample of bright X-ray sources in the Kepler field. The detection of 2 potential FK Comae candidates suggests that there are many other hidden contact binaries or post-interaction mass transfer products lying along the main sequence in intermediate-age OCs.

We detect UV excess from several stars lying close to the main sequence and binary isochrone, or the red giant branch in the optical CMD. The UV excess in these stars could be due to high chromospheric activity, presence of a hot WD companion or red leakage in UVW2 and UVW1 filters. The absence of a FUV flux data makes it difficult to distinguish between chromospheric activity and the presence of a WD companion.

We find that cool subgiant stars (T $\leq$ 5000 K) whose flux is close to detection limits could suffer from significant red leakage. These stars are detected only in UVW1 and UVW2 filters that suffer from red leak at optical wavelengths, suggesting that the UV excess in these stars is an artefact. Though RGB stars have similar temperatures (T $\sim$ 5000 K), their intrinsic flux is $\sim$ 10³ larger than the cool subgiant stars. Therefore, the origin of UV excess in the RGB stars is a reliable feature.

In a recent study of UV population in M67 by Sindhu, et al. (2018), they detected 2 RGB stars which show excess in the GALEX NUV and FUV bands and the presence of Mg II h+k emission lines in the UV spectra suggestive of chromospheric activity. The models explaining the origin of chromospheric activity are not yet well-established (Pérez Martínez, et al., 2011). The chromospheric activity could be due to acoustic heating of the atmosphere (Schrijver, 1995) or magnetic activity. Since SXOM42 is not observed in X-ray, the origin of the excess UV flux can be attributed to acoustic heating of the atmosphere leading to high chromospheric activity.

Several cool stars deviate from the single star isochrones in the UVW1-UVW2 CMD as shown in the Figure 15. Siegel, et al. (2019) had also identified this strange feature in the UVW2-UVW1 CMDs of a few star clusters. They had suggested that this feature could be due to red leak in UVW1 and UVW2 filters or deficiencies in the underlying atmosphere models. According to the broad-band SED, most of these stars show a prominent saddle dip feature around 2250 Å indicating the presence of strong absorption lines. These UV absorption features are not well characterized by atmosphere models of cool stars. The UVBLUE project by Rodríguez-Merino, et al. (2005, 2009) and Chavez, et al. (2007) also highlights this problem of incomplete characterization of absorption features in cool stars. They had retained the models longward of 2200 Å for intermediate-type F, and G stars as the lower wavelengths do not properly fit the observed spectrum. In Figure 15 the stars have been differentially coloured based on their Teff obtained by comparing the optical magnitudes with the best fitting PARSEC isochrone. We notice that the deviation of these cool stars from isochrones begins around 6500 K, which is similar to that identified by Siegel, et al. (2019). These deviations from isochrones have also been noted in NUV photometry of cool stars by Barker & Paust (2018). The conspicuous dip feature in most of these stars suggests that the deviations are likely due to the inefficient characterization of absorption lines in atmospheric models of cool stars.