The Preexplosion Environments and the Progenitor of SN 2023ixf from the Hobby–Eberly Telescope Dark Energy Experiment (HETDEX)

Supernova (SN) 2023ixf was discovered on 2023 May 19. The host galaxy, M101, was observed by the Hobby–Eberly Telescope Dark Energy Experiment collaboration over the period 2020 April 30–2020 July 10, using the Visible Integral-field Replicable Unit Spectrograph (3470 ≲ λ ≲ 5540 Å) on the 10 m Hobby–Eberly Telescope. The fiber filling factor within ±30″ of SN 2023ixf is 80% with a spatial resolution of 1″. The r < 5.″5 surroundings are 100% covered. This allows us to analyze the spatially resolved preexplosion local environments of SN 2023ixf with nebular emission lines. The two-dimensional maps of the extinction and the star formation rate (SFR) surface density (ΣSFR) show weak increasing trends in the radial distributions within the r < 5.″5 regions, suggesting lower values of extinction and SFR in the vicinity of the progenitor of SN 2023ixf. The median extinction and that of the surface density of SFR within r < 3″ are E(B − V) = 0.06 ± 0.14, and ΣSFR=10−5.44±0.66M☉yr−1arcsec−2. There is no significant change in extinction before and after the explosion. The gas metallicity does not change significantly with the separation from SN 2023ixf. The metal-rich branch of the R 23 calculations indicates that the gas metallicity around SN 2023ixf is similar to the solar metallicity (∼Z ☉). The archival deep images from the Canada–France–Hawaii Telescope Legacy Survey (CFHTLS) show a clear detection of the progenitor of SN 2023ixf in the z band at 22.778 ± 0.063 mag, but nondetections in the remaining four bands of CFHTLS (u, g, r, i). The results suggest a massive progenitor of ≈22 M ☉.


INTRODUCTION
The life of a massive star ends with a dramatic and energetic explosion known as a core-collapse supernova (CCSN; Branch & Wheeler 2017).These impressive events present a unique opportunity to delve into the cxliu@ynu.edu.cnphysics of massive star evolution and the creation of heavy elements through nucleosynthesis.It has been discovered that the progenitor stars of CCSNe experienced eruptive mass loss in the years prior to the core-collapse (e.g.Smith 2014).Pre-SN outbursts are often observed especially in Type IIn that display narrow emission lines for extended times after the explosion (e.g.Ofek et al. 2014) as a consequence of the shock interaction between the slow-moving circumstellar medium (CSM) and the Liu et al.
ejected material of the explosion (Smith 2014).However, the underlying physical mechanism that triggers those outbursts and the mass loss rates involved remain elusive (e.g.Beasor et al. 2020;Humphreys et al. 2020;Gofman et al. 2020;Ou et al. 2023;Hosseinzadeh et al. 2022).Direct identification of progenitors and the environments of stars via pre-explosion imaging becomes crucial in understanding these unsolved questions.Nonetheless, resolving individual stars in external galaxies at large distances is challenging, making the detection and study of SN events in nearby galaxies particularly important.
M101 is an Scd-type galaxy with active star formation (Lin et al. 2013).Several SNe events have been identified in the past century, e.g. the extensively studied SN 2011fe.Detailed comparison of the environments before and after the SN explosion may provide a direct constraint to the nature of SN feedback in the study of galaxy formation.Here we report the detection of the candidate progenitor star in the archival data from the Canada-France-Hawaii Telescope Legacy Survey (CFHTLS).We analyze the pre-explosion environments of the progenitor using integral field spectroscopic observation from the Hobby Eberly Telescope Dark Energy Experiment (HETDEX).In this work, we use the distance estimate of the host galaxy 6.85 ± 0.15 Mpc reported by Riess et al. (2022) and the redshift estimate of z = 0.000804 (Perley et al. 2023).The coordinates of SN 2023ixf are (RA= 210.910674637,Dec= +54.3116510708,J2000) 1 .
A typical HETDEX observation uses a 3-point dither pattern to fill in the gaps between fibers, with each dithered exposure being 6 minutes in length.The IFUs themselves are distributed in a grid, with each IFU separated from its nearest neighbor by 50 ′′ (except in the very center of the array, where other HET instruments are located).The result is that each HETDEX observation has a fiber-fill factor of 1 in 4.5.M101 lies within the footprint of the HETDEX survey, but to facilitate studies of the galaxy, additional pointings were defined to give a near complete fill factor of 1.The data of the full M101 observations were collected over the period April 30, 2020 -July 10, 2020, during a time when ∼ 70 of the planned 78 VIRUS IFUs were operational.
In total, there are 21 HETDEX observations covering the M101 field.Unlike the main HETDEX survey design, the footprint was chosen to provide as complete sky coverage as possible with tiling designed to fill in the gaps between the IFU array of the VIRUS spectrograph.We generate a datacube mosaic from these set of observations using a different reduction method than that described in Gebhardt et al. (2021).The HET-DEX observations of M101 were processed by Remedy 2 .Here we briefly summarize the data reduction and calibration of the M101 observations but further details are found in Zeimann et al. (in preparation).Astrometry is determined by matching wavelength collapsed images produced from the VIRUS spectra with stars within the Pan-STARRS (Chambers et al. 2016) Data Release 2 catalog.The root mean-square (rms) of the stars positions about the solution is ≲ 0.25 ′′ and the rms of the solution is typically less than ≲ 0.05 ′′ .
In typical HETDEX observations, as described in Gebhardt et al. (2021), flux calibration is determined using field stars of the Sloan Digital Sky Survey (SDSS; York et al. 2000).However, in the case of M101, where there are no SDSS fields available and we simply adopt an average throughput curve from all HETDEX observations, adjusted for the conditions of that observation.We normalize our response curves to the monochromatic fluxes obtained from the narrow-band [O III] λ5007 images of Herrmann et al. (in preparation).To do this, we apply the narrow-band filter's transmission curve 3 to VIRUS spectra to create a [O III] λ5007 synthetic image.We then smooth both the synthetic image and Mosaic image to account for differential seeing, and use the biweight of the ratio of these two images to normalize each observation's response curve.
We construct a single data cube for all HETDEX observations in the area of M101, centered at RA= 210 • .800 and DEC= 54 • .333 with spectral pixels of 1 ′′ × 1 ′′ × 2 Å.To go from the non-uniform sky positions of the fibers to the uniform grid of the data cube, we placed each fiber's flux into the nearest pixel at every wavelength.If multiple fibers from multiple observations contributed to the same pixel, we calculated the median value.All pixels without a flux contribution from a fiber were masked.We then performed a Gaussian convolution using a seeing of 1. ′′ 8 to reconstruct the image at each wavelength.The final data cube is 1201 × 1201 × 1036 pixels.

CFHTLS
The Canada-France-Hawaii Telescope Legacy Survey (CFHTLS) is a deep sub-arcsecond (0.8 ′′ ) wide-field (157 deg 2 total) optical survey with u, g, r, i, z bands.The 7 th and final release of CFHTLS (CFHTLS-T0007; Cuillandre et al. 2012) is produced by TERAPIX based on a data set collected with MagCam with 36 chargecoupled devices (CCDs) on the CFHT.M101 lies within the footprint of the CFHTLS-T0007.It was observed in i-and z-bands on May 11, 2005, and June 7, 2006, respectively.These observations included six consecutive images with 615-second exposures in i-band and 600second exposures in z-band.

PRE-EXPLOSION ENVIRONMENTS
Figure 1 shows the HETDEX image of SN 2023ixf created from the integrated flux within [3550,5450] Å.The bluest wavelengths and the reddest wavelengths are not used to avoid potential edge effects from CCDs.The white gaps are the pixels with no fiber coverage.This 60 ′′ × 60 ′′ cutout has a fiber coverage of 80%.The red circle marks the r < 5.5 ′′ region, within which the fiber coverage is 100%.This region does not include the fibers on the edges of IFUs.The flux of edge fibers can be potentially overestimated due to large aperture corrections.Therefore, the r < 5.5 ′′ region is selected for the study of the pre-explosion environments.
Figure 2 shows the spectrum of the HETDEX pixel closest to SN 2023ixf.The center of this pixel is 0.54 ′′ away from SN 2023ixf.As shown by the six panels in the bottom two rows, the nebular emissions of [O II] λ3727, Hγ, Hβ, [O III] λ4959, and [O III] λ5007 are significantly detected within our wavelength range.The Hδ emission can be marginally recovered after stellar subtraction using the templates of Bruzual & Charlot (2003), hereafter BC03.These nebular lines allow the estimates of extinction (Section 3.1), gas metallicity (Section 3.2), and the star formation rates (Section 3.3).It is worth noting that no significant variations of the line widths or redshifts/blueshifts of the emission lines are detected as a function of the distance to the SN center with our 2 Å resolution.However, a velocity below our resolution could not be ruled out.

Extinction
The gas extinction E(B − V ) gas can be derived from the Balmer decrement.In this paper, due to the limitation of the wavelength coverage of HETDEX, we use the Hγ/Hβ ratio instead of the more commonly used Hβ/Hα ratio to calculate the gas extinction.We assume an Galactic extinction curve (R V = A V /E(B − V ) = 3.1) (Cardelli et al. 1989)   0.47 (T = 15, 000 K, case B recombination; Storey & Hummer 1995).The reddening suffered by the stellar continuum from the inter-stellar medium (ISM) E(B − V ) ISM can then be roughly estimated as 0.44 times the reddening suffered by the ionized gas E(B − V ) gas (Calzetti 2001).
Figure 3 shows the extinction map in the upper left panel and its radial variations in the upper right panel.
Only pixels with both the Hγ and Hβ emission lines strong enough, equivalent widths (EWs) greater than 3 Å, are calculated with their extinctions and shown.There is a weak increment of the extinction as a function of the separations from SN 2023ixf.The median values of E(B − V ) ISM of all pixels within r <1 ′′ , 3 ′′ , and 4.5 ′′ are 0.03 ± 0.06 mag, 0.06 ± 0.14 mag, and 0.13 ± 0.18 mag, respectively.For M101 exhibiting very

Liu et al.
high star formation rates, the mean Galactic extinction curve may not be suitable, and the value of R V can be much larger (e.g.∼ 4.05 ;Calzetti 1997).This will give slightly higher measurements of the extinction (∆E(B − V ) ISM ≲ 0.01 mag), but the weak increasing trend will remain.
Two high-resolution spectra on SN2033ixf was taken with the TK2 cross-dispersed echelle spectrograph (resolving power R = 60, 000) on the 2.7-m Harlan J. Smith  2023), and it is not significantly different from our pre-explosion measurement of E(B − V ) = 0.03 ± 0.06 in the regions of r < 1 ′′ .Niu et al. ( 2023) measures the pre-explosion extinction of the environment within r < 4.5 ′′ as E(B − V ) = 0.15 mag using the resolved stars in the HST images.This is consistent with our median E(B − V ) = 0.13 ± 0.18 mag in the regions of r < 4.5 ′′ .

Gas Metallicity
We use the R 23 strong-line method to estimate the gas metallicity (Z), where R 23 is a line ratio defined as Hβ (e.g.Tremonti et al. 2004).Line fluxes are dereddened using the gas extinction derived in Section 3.1 and the extinction law of Cardelli et al. (1989).For a given R 23 , there are two associated metallicity values, one is in the metal-poor branch, and the other is in the metal-rich branch (e.g.Kobulnicky et al. 1999).SN 2023ixf is ∼ 4.5 kpc away from the center of M101, which is within the break radius in the abundance gradient at 15.4 kpc (Garner et al. 2022).This suggests that the surroundings of SN 2023ixf would probably favor the metal-rich branch.However, the possibility of the metal-poor branch can not be fully ruled out.
Figure 3 shows the metallicity map derived from the metal-rich branch in the middle left panel.In the middle right panel, both the radial variations of the metallicity from the metal-rich branch (solid upper triangles) and those of the metallicity from the metal-poor branch (open lower triangles) are shown.We only calculate the metallicity for pixels that satisfy the following criteria: EW [O II]λ3727 > 3 Å, EW [O III]λ5007 > 3 Å, and EW Hβ > 3 Å.Assuming all pixels around SN 2023ixf following the metal-rich branch, the median O abundances of all pixels within r <1 ′′ , 3 ′′ , and 4.5 ′′ are 12 + log(O/H) upper = 8.66 ± 0.05, 8.65 ± 0.11, and 8.61 ± 0.11, respectively.Considering the scattering, the gas metallicitly does not change significantly with their separations from SN 2023ixf.Assuming a solar O abundance of 12 + log(O/H) ⊙ = 8.69 and a solar metallicity of Z ⊙ = 0.013 (Asplund et al. 2021), the metal-rich branch gives a metallicity of Z ∼ 0.013 ± 0.002 (∼ Z ⊙ ), and the metal-poor branch gives Z ∼ 0.003 ± 0.001 (12 + log(O/H) lower = 8.04 ± 0.09 dex, ∼ 0.25Z ⊙ ).We note that considering a more recent solar O abundance estimation of 12 + log(O/H) ⊙ = 8.77 and a solar metallicity of Z ⊙ = 0.0225 from Magg et al. (2022), the metalrich branch would give a metallicity of Z ∼ 0.018±0.002,and the metal-poor branch gives Z ∼ 0.004 ± 0.001.
Pledger & Shara (2023) reported the pre-explosion O abundances of two near the SN site HII regions published in Kennicutt & Garnett (1996) (region 1098 and 1086 in their Figure 2(b)) as 12 + log(O/H) upper = 8.63 and 8.59, which are close to our pre-explosion on-site results observed in the year of 2020.Van Dyk et al. (2023) analyzed the after-explosion on-site O abundances to be 8.43 ≲ 12 + log(O/H) ≲ 8.86 from Gemini Spectroscopy on June 3, 2023.The metallicity does not change significantly after the SN 2023ixf event.

Surface Density of Star Formation Rates
The commonly used emission-line indicator of the star formation rate (SFR), the Hα emission, is out of the wavelength coverage of HETDEX.SFR is therefore estimated using the correlation between the luminosity of [O II] λ3727 and SFR (Equation 4 in Kewley et al. 2004).For our case, the calculated SFR of each pixel (1 ′′ × 1 ′′ ) is the surface density of SFR (Σ SFR ) in units of M ⊙ yr −1 arcsec 2 .
Figure 3 shows the Σ SFR map in the bottom left panel and its radial variations in the bottom right panel.Again, only pixels with EW [O II]λ3727 > 3 Å are calculated along with their SFRs.There is an increasing trend of the surface density of SFR as a function of the separations from SN 2023ixf.The median log Σ SFR of all pixels within r <1 ′′ , 3 ′′ , and 4.5 ′′ are −5.59 ± 0.28, −5.44 ± 0.66, and −5.04 ± 0.84, respectively.

MASS OF THE PROGENITOR
We reprocessed the images of the progenitor from CFHTLS to obtain the magnitudes of the progenitor in the single-exposure images and in the stacked image.Before stacking the images using swarp 4 , we remove cos-mic rays using the python package ccdproc5 .We then recalibrate the astrometry using scamp6 , with reference stars from Gaia DR3.SExtractor7 and PSFEx8 are used to do the photometry of the progenitor.The photometric zero point is taken directly from the header of the archival fits files.We validate this value using the SDSS survey and the Pan-STARRS survey.z 0 recorded in the header of each individual z-band image is 24.754 mag.We cross match each individual image with the SDSS survey and the Pan-STARRS survey.The zeropoint magnitude is then calculated using the common stars recorded in the SDSS survey and the Pan-STARRS survey separately.Take the z-band single exposure 851171p as an example: the derived zeropoint from SDSS is z 0 = 24.761± 0.172 mag and that derived from Pan-STARRS is z 0 = 24.753± 0.176 mag.Both surveys confirm the photometric zeropoint z 0 = 24.754mag recorded in the header.The progenitor is significantly detected in z-band at 22.778 ± 0.063 mag in the stacked images (the bottom left panel and the bottom middle panel of Figure 4).All six individual z-band images show clear detections at the position of SN 2023ixf (top two rows of Figure 4).It is worth noting that the progenitor is also significantly detected in the z-band observation of the Dark Energy Camera Legacy Survey (DE-CaLS9 ) at 23.30±1.58mag, which confirms our measurements from CFHTLS images.The progenitor is not detected in u-, g-, r-, and i-bands.The i-band detection limit is 24.57mag (5σ).
Three sources of extinction have been taken into account: the Milky Way (E(B − V ) MW = 0.008 mag; Schlegel et al. 1998;Schlafly & Finkbeiner 2011), the ISM of the host galaxy (E(B − V ) ISM ∼ 0.03 mag; this work), and the circum-stellar medium (CSM) of the progenitor.Here we assume R V = 3.1 for all three sources of extinction.We note that R V can be slightly higher for ISM in the star-forming regions, which will lead to a slightly brighter and redder measurement.Considering the low E(B − V ) ISM ∼ 0.03 mag, this wouldn't affect the color and magnitude measurements significantly.Niu et al. (2023) fits the spectral energy distribution (SED) with the C-rich dust models, and obtains E(B − V ) CSM = 1.64 ± 0.2 mag.We also note that the C-rich dust model derived extinction should be applied with cautions, as detailed in Section 9.2 of Van Dyk et al. (2023).For the CFHTLS i-and zbands, R i =1.799 and R z =1.299 are adopted (Zhang & Yuan 2023).These together give A i ∼ 3.02 mag and A z ∼ 2.18 mag.The distance modulus of the host galaxy is 29.18 (Riess et al. 2022), which gives the absolute magnitudes of the progenitor as M i ≳ −7.63 mag (5σ detection limit), M z = −8.58mag, and a color limit of M i − M z ≳ 0.95.
We use the PARSEC10 stellar evolutionary isochrones (Bressan et al. 2012) to estimate the mass of the progenitor (Figure 5).The isochrones that match the CFHTLS colour and magnitude best have an age of ∼ 8-9 Myr, an initial mass between ∼ 20−22 M ⊙ , and a solar metallicity ∼ Z ⊙ .This metallicity of the progenitor is close to the pre-explosion gas metallicity derived from the metalrich branch in Section 3.2.We also present the evolution of stars with 0.25Z ⊙ (the pre-explosion gas metallicity derived from the metal-poor branch in Section 3.2) by the dashed lines for comparison reason, although it does not match our color measurement.We note that models with different input evolutionary tracks may result in different initial mass estimates.The widely used PARSEC isochrones do not fit our (i−z) color limit very well.The solar metallicity isochrones stop at (i − z) 0 ∼ 0.9, while our color limit of (i − z) ≳ 0.95 is beyond that.The BPASS (Eldridge et al. 2017) single-star models indicate a lower initial mass range of ∼ 15−17 M ⊙ .We also note that our initial mass estimate is close to the maximum possible mass for SN IIP progenitors ∼ 20 M ⊙ (Davies & Beasor 2020).The z-band magnitude combined with an (i − z) color limit may not constrain the evolution track to the best.Jencson et al. (2023) fit the Hubble Space Telescope (HST) photometry and they also concluded with a massive initial mass of 17 ± 4 M ⊙ for SN 2023ixf.Van Dyk et al. (2023) analyzed the multi-band data of SN 2023ixf from HST, Spitzer, Herschel, and Wide-Field IR Survey Explorer (WISE), and their SED fitting indicates a less massive initial mass of ∼ 12 − 15 M ⊙ .

SUMMARY AND DISCUSSIONS
In this paper, we study the pre-explosion environments of SN 2023ixf with the HETDEX IFU observations taken in 2020.There are no significant variations of the line widths or redshifts/blueshifts of the emission lines detected as a function of the distance to the SN center with our 2 Å, or 133 km/s in velocity space, resolution.We find that SN 2023ixf exploded in a region having low extinction and low SFR locally.This low SFR, however, might be somewhat different from that PARSEC stellar evolutionary isochrones along with our progenitor candidate.The cyan point is the colour-magnitude position measured from CFHTLS images.The solid lines present the isochrones of solar metallicity Z⊙ (the gas metallicity derived from the metal-rich branch using the R23 method).
The dashed lines present those of 0.25Z⊙ (the gas metallicity derived from the metal-poor branch using the R23 method).
of the progenitor's birthplace if the progenitor were a runaway star, like e.g.Betelgeuse which has a low velocity of ∼ 30 km/s through ISM.Our pre-explosion measurement of E(B − V ) = 0.03 ± 0.06 mag in the regions of r < 1 ′′ based on Hγ/Hβ line ratio is not significantly different from our after-explosion measurement of E(B − V ) = 0.033 ± 0.03 mag from the resolved NaD doublet.The gas metallicity does not vary significantly with their separations from SN 2023ixf.The metal-rich branch of the R 23 method gives a near solar gas metallicity ∼ Z ⊙ (12 + log(O/H) = 8.66 ± 0.05 dex), and the metal-poor branch gives a gas metallicity of ∼ 0.25Z ⊙ (12 + log(O/H) = 8.04 ± 0.09 dex).If there are future follow-up observations, especially spatially resolved spectroscopic ones, we would learn more about the SN 2023ixf feedback to the host galaxy by comparing with our pre-explosion measurements of the properties of the surrounding ISM.We measure the magnitudes of the progenitor of SN 2023ixf in the CFHTLS stack images observed prior to the year of 2010.The progenitor is significantly detected in the z-band images at 22.778 ± 0.063 mag, but not detected in u-, g-, r-, i-bands.A comparatively massive progenitor with an initial mass ∼ 22 M ⊙ and a solar metallicity ∼ Z ⊙ is suggested by comparing the extinction-corrected magnitudes with isochrones.If the local gas environments of SN 2023ixf do follow the metal-rich branch of the R 23 method, this suggests that the metallicity of the progenitor is comparable with that of the surrounding gas.

3Figure 1 .
Figure 1.The ±30 ′′ HETDEX image cutout centered on SN 2023ixf taken in the year of 2020.The image is created from the integrated flux within the wavelength range of [3550, 5450] Å.The pixel size is 1 ′′ × 1 ′′ .The red circle shows the region of r < 5.5 ′′ , which is selected for the study of the local environments.

Figure 2 .
Figure 2. The spectrum of the HETDEX pixel closest to SN 2023ixf.The top panel shows the spectrum in the full HETDEX wavelength range with the black data points.The red line is our best-fit spectrum consisting of the best-fit stellar continuum using BC03 templates and the best-fit emission lines.The blue dashed lines mark the emission lines.The six panels in the 2nd and 3rd rows show the sub-regions of the six emission lines: [O II] λ3727, Hδ, Hγ, Hβ, [O III] λ4959, and [O III] λ5007.The green lines are the best-fit stellar continuum in all seven panels.The grey curves in all panels of the bottom two rows are the HETDEX spectrum in each wavelength range.The black data points in the bottom two rows show the continuum subtracted spectra.The magenta lines are the best-fit emission line profiles.

Figure 3 .
Figure 3.The 2-D maps of the diffuse ISM extinction (E(B − V )ISM, top left panel), O abundance derived from the metal-rich branch of the R23 method (middle left panel), and the surface density of the star formation rate (log ΣSFR, bottom left panel) centered on SN 2023ixf.Pixels with fiber coverage but not shown are because their nebular emission lines are not strong enough for reliable measurements of the extinction, O abundance, and the star formation rates, see contexts for details.Their variations as a function of the separations from SN 2023ixf are shown in the right panels.Data points in the right panels are associated with the individual pixels with measurements within the red circles in the left panels.The blue, orange, and green dashed lines in the right panels marks the median values of E(B − V )ISM, 12 + log(O/H)upper, and log ΣSFR of pixels within r < 1 ′′ , 3 ′′ , and 4.5 ′′ , respectively.The median values are also labeled in the upper left corners.The red solid lines in the right panels are simple linear fits to all pixels within r < 5.5 ′′ (the red circles in the left panels) given to guide the eye to radial trends.In the middle right panel, both the O abundances derived from the the metal-rich branch of the R23 method (the solid upper triangles), and those derived from the metal-poor branch (open lower triangles) are shown.

Figure 4 .Figure 5 .
Figure 4.The CFHTLS image cutouts centered on SN 2023ixf.The six panels in the top two rows show the six individual z-band images.The bottom panels show the z-band co-added, the z-band background subtracted co-added, and the i-band background subtracted co-added images from left to right.The z-band observations were taken on June 7, 2006.The i-band observations were taken on May 11, 2005.The green and red circles mark the r < 1.5 ′′ and r < 5.5 ′′ regions of SN 2023ixf, respectively.
and an intrinsic Hγ/Hβ ratio of