The Planetary Nebula in the 500 Myr old Open Cluster M37

The nebula was discovered and classified as a PN candidate by Sabin (2008) from IPHAS imagery. Here we present new, high resolution radial velocity data that confirm this rare OC-PN link. The PN is identified as a ‘True’, bipolar, likely Type-I PN in HASH (Parker et al., 2016) as IPHASX J055226.2+323724 (PNG 177.5+03; HASH ID 31188) where a clearer image than that presented in Núñez et al. (2017) shows its evolved, bipolar nature. It is of very low surface brightness at the $\sim$5 Rayleigh sensitivity limit of IPHAS (Drew et al., 2005) with a major axis of 445$\pm$10 arseconds. The PN presents enhancements along the southern edges and a patchy internal structure (see fig. 1). The emission line spectra seen in the cluster stellar fibre-spectra for five stars in Fig. 5 of (Núñez et al., 2017) show high [N II]/H$\alpha$ ratios, indicative of a likely Type-I (nitrogen enriched) chemistry (Kingsburgh & Barlow, 1994).

Based on Sloan Digital Sky Survey imagery (Gunn et al., 2006) we found the CSPN at RA: 05:52:26.18, DEC: 32:37:24.63 (J2000), almost exactly at the PN’s geometric centre ($<$10 arcsecond displacement) which is $\sim$7.2 arcmin across its major axis. This is also the only blue star within a 116 arcsecond radius from the PN’s center, making it the only plausible CSPN candidate. This CSPN was previously identified as a WD candidate in Gentile Fusillo et al. (2015). It is also reported as a CSPN by Chornay & Walton (2020) via Gaia searches for CSPN based on PN in the HASH catalog (Parker et al., 2016), though no association with the cluster was made.

Low dispersion spectroscopy of the CSPN is presented in Griggio et al. (2022) that show both He II and C IV absorption lines, confirming the CSPN is a hot, hydrogen deficient WD. It is reported as an intermediate type between the DO class and the PG1159 stars, e.g. Werner & Herwig (2006) and Reindl et al. (2014). They report a stellar $T_{eff}>60,000$K. (We estimate a value closer to 100,000 K in this paper). Hence, this CSPN is hot enough to ionise the observed PN. The Gaia EDR3 ID of this CSPN is 3451205783698632704.

The white dwarf has been classified as a rotating variable with a period of 0.445 days and amplitude of 0.074 mags by Chang et al. (2015) (V1975 in their sample). Rotational variability has previously been found in hot DQ white dwarfs (Lawrie et al., 2013; Williams et al., 2016) where it is related to strong magnetic fields. The current star is not known to be of DQ type but this raises the possibility that it could be a transition object between PG1159 stars and DQ white dwarfs. Alternatively, the variability could be due to a close binary, however it is not classified as such by Chang et al. (2015).

M37 (NGC 2099) is the brightest and richest Galactic open cluster in the constellation of Auriga with a stellar mass of $\sim$1500 M_⊙ based on concordant Gaia EDR3 (Gaia Collaboration et al., 2021) high probability proper motions and potentially with as many as 4500 stars. It is a well-studied, intermediate-age ($\sim$500 million year old) cluster first discovered by the Italian astronomer Giovanni Battista Hodierna before 1654 and catalogued by Messier in 1781. It is about 30 arcminutes across at full extent. Previous modern work includes Kalirai et al. (2001), Kalirai et al. (2005) and Ahumada & Lapasset (2007) who show the cluster contains three blue stragglers while its hottest surviving main sequence star is a B9V. More recent work on M 37 concentrates on the Gaia data, e.g. Griggio & Bedin (2022) and Griggio et al. (2022) who also list 7 cluster WD candidates including the CSPN labelled WD 1 in their list. The weighted average cluster physical parameters obtained from the literature have been previously estimated as: angular diameter = 31.15 arcminutes, age = 470 $\pm$ 50 Myrs, reddening $E(B-V)$ = 0.26$~{}\pm$ 0.04, distance = 1.44 $\pm$ 0.13 kpc and metallicity $[Fe/H]$ = 0.03 $\pm$ 0.28 (Griggio & Bedin (2022); refer table 6.6 in Fragkou (2019)). Using 1136 stars with a cluster membership probability $>$ 0.8, Cantat-Gaudin et al. (2018) yields a Gaia DR3 mean distance of 1485 pc with $\sigma$ = 110 pc, in good agreement with previous estimates. Since this is based on the latest Gaia DR3 data, we are using this cluster distance value throughout. The same stars present a Gaia mean pmRA$\,=1.88\pm 0.18$ mas/yr and pmDec$\,=-5.62\pm 0.16$ mas/yr. The Gaia DR3 CSPN proper motion is in tight agreement with the cluster average, strongly supporting cluster membership (Table. 1). The parallax has a larger uncertainty but is consistent within $1\sigma$. For the adopted cluster distance and angular diameter, the cluster’s physical radius is 6.73 pc.

The ugriz CSPN magnitudes have been measured from SDSS DR10 (An et al., 2007) as u = 18.702 $\pm$ 0.020 mag, g = 18.978 $\pm$ 0.010 mag, r = 19.300 $\pm$ 0.011 mag, i = 20.248 $\pm$ 0.063 mag and z= 19.840 $\pm$ 0.082 mag (Gentile Fusillo et al., 2015). Following Jester et al. (2005) these are transformed to the Johnson-Cousins system as B = 19.07 $\pm$ 0.02 and V = 19.16 $\pm$ 0.02. The errors have been computed with standard error propagation with the transformation rms residuals added quadratically. The CSPN is also included in the Pan-STARRS catalogue (Chambers et al., 2016) with magnitudes $g=19.06\pm 0.015$, $r=19.40\pm 0.01$, $i=19.69\pm 0.02$, $z=19.93\pm 0.03$ and $v=20.03\pm 0.04$.

For PN radial velocity measurement we obtained observations with the high spectral resolution MEGARA spectrograph (García-Vargas et al., 2020) on 12^th January 2022 (with full moon contamination) and 4^th March 2022 (dark sky). MEGARA is an optical (3650–-9750Å) medium-high spectral resolution fibre-fed spectroscopic instrument on the 10.4-m Gran Telescopio CANARIAS (GTC) in La Palma, Canary Islands. We obtained optical spectroscopy at 7 distinct positions across the PN (refer Fig.3a). Multiple 5-minute exposures were taken in IFU mode. The IFU is a bundle of 567 (0.62 arcsec diameter) fibres that subtend 12.5 $\times$ 11.3 arcseconds on the sky. One pointing was centered on the CSPN (pointing a), plus an offset sky exposure. The VPH grating set up was the ‘HR–R’ highest dispersion mode (R = 20,000) that covers 6400–6800Å  and gives a radial velocity precision of around 1 km/s, ideal for our purpose. Four areas (a, b, c and d) of the total of the seven pointings were observed in both January and March 2022. The earlier data was affected by the Moon causing a strong H$\alpha$ absorption feature but the [NII] and [SII] lines were unaffected. The pointing map and an example combined nebula spectrum from pointings a, b, c, d (for observations made on March 4^th 2022) are shown in Fig. 3a,b.

The robust MEGARA pipeline was employed to reduce the data in standard fashion for IFU mode. Flux calibration and accurate sky subtraction are not essential for the current purposes. The wavelength calibration appears robust. The radial velocities for each pointing are measured from the nebular emission lines in the integrated IFU spectrum of each pointing. After the standard reduction with the MEGARA pipeline (which also combined the exposures for each observing block), the integrated 1-D spectrum for each pointing was constructed using standard IRAF techniques. The S/N for one of our pointings (area f) from January 13^th 2022 was too low for useful measurements. Otherwise the nebular emission lines were used for radial velocity determinations via Gaussian fitting.

We computed the radial velocity average for all pointings with repeated observations (more than one observing block) and calculated the median heliocentric corrected radial velocity of all pointings as $v_{\rm rad}=10.6\pm 4.9$ km/s from 6 individual combined IFU pointings (4 were observed twice on different nights). The nebular spectra covered the [N II], H$\alpha$ and [S II] emission lines which were all detected. The weaker [N II] 6548Å  and [S II] doublet lines had too low S/N in individual pointings to give reliable radial velocities due to larger errors. The 6584Å  [N II] line (from all individual observing blocks) and the H$\alpha$ line (from observations acquired in March and not affected by the strong H$\alpha$ absorption feature) have both been employed for the calculation of the mean nebular radial velocity. This value is compatible with the cluster heliocentric corrected velocity of $v_{\rm rad}=8.32\pm 0.56$ km/s providing a tight constraint for cluster membership. In most pointings the [N II] 6584Å  line was split into a blue-shifted and red-shifted component, with various asymmetries. allowing a direct measurement of the nebula expansion velocity at those pointings. The pointing ‘a’ on the PN center provided the clearest splitting of the [N II] 6584Å  line. These provided an average expansion velocity of 20 km/s with a standard deviation of 6.2 km/s. This is typical for a PN and allows a kinematic age to be determined from the PN physical size from its angular extent and distance, of $t_{\rm kin}\sim 78\times 10^{3}\pm 25\times 10^{3}$ yrs. This kinematic age is at the extreme end for PNe as befits such an evolved, large and low-surface brightness example. It may in fact be the largest PN kinematic age ever determined, assuming invariant expansion velocity over time.

For the [S II] line electron density estimate we combined the a,b,c,d pointings from March 4^th 2022 to provide a higher S/N, 1-D PN continuum subtracted spectrum. A box-3 smooth was applied before Gaussian fits to the two well detected [S II] lines using a wide wavelength range around the lines for determining the best base level. As expected the [S II] lines are found to be in the low density limit with [S II]6717/6731$\sim$1.45$~{}\pm~{}$0.20 from repeat measures. As such an electron density of $N_{e}<5cm^{-3}$, was obtained using the IRAF nebular add-on package and assuming a $T_{e}=10,000$ K. We use this value, the average of the nebular minor and major physical radius, and an assumed filling factor of 0.3 (Pottasch, 1996) to calculate a maximum ionised PN gas mass of 0.32 M_☉ (Boffi & Stanghellini, 1994). From the same data we measured a [N II]/H$\alpha$=$3.58\pm 0.10$ which supports the Type I nature of the bipolar nebula.

O5 Main Sequence stars present a $B-V$ color of $-0.33$ (Cox, 2000) and since hotter stars are expected to have almost identical colors, we use this to estimate the CSPN reddening $E(B-V)=(B-V)-(B-V)_{0}$. Hence, the $B-V=-0.092$ CSPN color implies a reddening of $E(B-V)=0.24\pm 0.03$, in excellent agreement with the cluster weighted average from the available literature of $E(B-V)=0.26\pm 0.04$. The Gaia magnitudes are reported in Griggio et al. (2022). An interstellar extinction profile was derived from the 3-D galactic model of Vergely et al. (2022) retrieved through EXPLORE³³3https://explore-platform.eu. It is an approximately linear rising trend until about 500pc, has a shallower slope then to 1000pc with A_o of $\sim$0.5 and flattens thereafter, rising only to A_o of $\sim$0.57 by 2000 pc. The main interstellar clouds are at 200 pc, 400 pc and 1 kpc, with another cloud at 1.6 kpc just behind M37. There is no extinction associated with the cluster itself. This also supports interpretation of the nebula as a PN rather than ionized interstellar material.

From the adopted cluster parameters and considering the time for the star to leave the Main Sequence and pass through the RGB/AGB phase, the adopted isochrone predicts a PN progenitor mass of 2.78${}_{-0.1}^{+0.12}$ M_☉.

For the adopted CSPN $V$ magnitude, cluster reddening and distance, the CSPN absolute $V$ magnitude is $M_{V}=7.56\pm 0.19$. By plotting the CSPN $V$ magnitude and nebular kinematic age along with evolutionary tracks (see e.g. Weston et al. (2009), their Figure 1, but using the Miller Bertolami (2016) improved tracks), we estimate a CSPN final mass of 0.63${}_{-0.04}^{+0.03}$ M_⊙ which is well within the expected range for a WD descended from a progenitor star of $\sim$2.8 M_⊙.

The CSPN absolute $V$ magnitude and a series of possible CSPN effective temperature values (50 – 300 kK, in steps of 10 kK) were used to calculate a series of CSPN luminosities, which were plotted along with the corresponding evolutionary track (Miller Bertolami, 2016). From the intersection of our line with the evolutionary tracks we estimate a CSPN effective temperature of 100 $\pm$ 20 kK (assuming 20% error). This leads to a CSPN luminosity estimate of $logL/L_{\sun}$ = 1.49 $\pm$ 0.25.

The known sample of cluster WDs for the latest IFMR estimates and semi-empirical ‘PARSEC’ fit (Cummings et al., 2018) is presented in Fig.5. Our new estimate for the initial and final mass for our M37 cluster PN IPHAS J055226.2+323724 is plotted as a red circle. The other two colored points are for the other known Galactic OC PNe; PHR 1315-6555 Parker et al. (2011) and Fragkou et al. (2019a) plotted as a yellow circle and the very high mass progenitor of PN BMP J1613-5406, Fragkou et al. (2019c) plotted as a green circle. The errors on each point reflect errors in the adopted cluster parameters and the spread of the estimated CSPN magnitudes.

The OC PNe fall below the plotted dashed line trend from the cluster WDs, which barely overlap with the error bars. Recently Marigo et al. (2020) found an increase in final masses for initial masses in the range 1.5–2 M_⊙, based on white dwarfs in the clusters R 137 and NGC 7789. Beyond this excursion the final masses fall back to the previous relation. The putative ‘kink’ is proposed to be related to carbon star formation on the Asymptotic Giant Branch. The interpretation is complicated by having only a single WD star in the range 2–2.5 M_⊙ from cluster NGC 752. The location of the two lower initial mass OC CSPN, the cluster white dwarfs and the proposed relation of Marigo et al. (2020) from their Fig.1 are shown in our version as Fig. 5b. Their errors in M_f are $\sim$0.5 $M_{\sun}$ but only $\sim$0.05 $M_{\sun}$ in M_i. Both cluster CSPN trace the key mass range just above the kink and in fact accentuate it, with one in the middle of the empty mass range 2–2.7 M_⊙. These cluster CSPN are consistent with the proposed reduction in this mass range from Marigo et al. (2020).