Hubble Space Telescope Observations of Mira Variables in the Type Ia Supernova Host NGC 1559: An Alternative Candle to Measure the Hubble Constant

TL;DR

The paper addresses the $H_0$ tension by introducing Mira variables as an independent intermediate-distance rung to calibrate Type Ia supernovae. Using year-long near-IR HST/WFC3-IR ($F160W$) time series, the authors identify 115 O-rich Miras in NGC 1559 and calibrate the SN Ia luminosity of SN 2005df by anchoring to the NGC 4258 maser distance, with careful treatment of crowding, incompleteness, and C-rich contamination. They derive a Mira-based distance modulus $\mu_{1559}=31.41\pm0.060$ mag and a fiducial SN Ia absolute magnitude $M_B^0=-19.27\pm0.13$ mag, yielding $H_0=72.7\pm4.6$ km s$^{-1}$ Mpc$^{-1}$ (and $H_0=73.3\pm4.0$ km s$^{-1}$ Mpc$^{-1}$ when including the LMC anchor), consistent with Cepheid-based measurements within uncertainties. The work demonstrates the viability of Miras as SN Ia calibrators, suggests that extending this program to additional local SN Ia hosts could reduce the overall $H_0$ uncertainty to ~3%, and highlights the potential of Miras to cross-check Cepheid distances or enable distance measurements in halo and early-type hosts, with JWST offering avenues for greater reach.

Abstract

We present year-long, near-infrared Hubble Space Telescope WFC3 observations used to search for Mira variables in NGC 1559, the host galaxy of the Type Ia supernova (SN Ia) 2005df. This is the first dedicated search for Miras, highly-evolved low-mass stars, in a SN Ia host and subsequently the first calibration of the SN Ia luminosity using Miras in a role historically played by Cepheids. We identify a sample of 115 O-rich Miras with P < 400 days based on their light curve properties. We find that the scatter in the Mira Period-Luminosity Relation (PLR) is comparable to Cepheid PLRs seen in SN Ia supernova host galaxies. Using a sample of O-rich Miras discovered in NGC 4258 with HST F160W and its maser distance, we measure a distance modulus for NGC 1559 of mu1559 = 31.41 +/- 0.050 (statistical) +/- 0.060 (systematic) mag. Based on the light curve of the normal, well-observed, low-reddening SN 2005df, we obtain a measurement of the fiducial SN Ia absolute magnitude of MB0 = -19.27 +/- 0.13 mag. With the Hubble diagram of SNe Ia we find H0 = 72.7 +/- 4.6 kms-1 Mpc-1. Combining the calibration from the NGC 4258 megamaser and the Large Magellanic Cloud detached eclipsing binaries gives a best value of H0 = 73.3 +/- 4.0 km s-1 Mpc-1. This result is within 1-sigma of the Hubble constant derived using Cepheids and multiple calibrating SNe Ia. This is the first of four expected calibrations of the SN Ia luminosity from Miras which should reduce the error in H0 via Miras to ~3%. In light of the present Hubble tension and JWST, Miras have utility in the extragalactic distance scale to check Cepheid distances or calibrate nearby SNe in early-type host galaxies that would be unlikely targets for Cepheid searches.

Figures (11)

• Figure 1: The distribution of the Stetson $L$ index for all of the objects detected in the master image in ten epochs of observations. We began the variability search by examining objects with $L \geq 0.75$, indicated in the plot by the black dashed line. There were $\sim 49,000$ objects in the master list and $\sim 25,000$ with 10 epochs of observations.
• Figure 2: Top: An example of a variable that passed the $F$-statistic cut. Bottom: An example of a variable that failed the reduced $F$-statistic cut.
• Figure 3: The fraction of simulated Miras recovered in our final sample as a function of period and smoothed into 25-day bins. We considered a period recovered if it was within 15% of the true period. The simulated Miras that passed our variability criterion were considered to be recovered as variables. The dashed line shows the 90% completeness limit. The black line shows the adopted 240-day lower period limit.
• Figure 4: The input and recovered periods in our simulations. Red points are considered "recovered" (within 15% of the true input period). Black lines indicate the region of recovered periods. Gray points show periods that were not recovered by the simulation. The blue line shows $x = y$, where the true and recovered periods match exactly.
• Figure 5: The zeropoint as a function of the starting period of each bin, using a boxcar fit with a width of 75 days. At $\sim 240$ days (black dashed line) the zeropoint starts to converge and oscillate about the true value. We chose 240 days as the minimum period based on this result. The black line shows the zeropoint using the F160W slope. The gray lines denote the uncertainty in the zeropoint at each point.