Axion Constraints from White Dwarf Cooling in 47 Tucanae

TL;DR

This study uses HSTWD data from 47 Tuc to test for axion-induced extra cooling in white dwarfs via axion bremsstrahlung from electrons. By generating MESA-based WD cooling models that include axion energy loss and comparing them to unbinned, two-field CMD data with a rigorous likelihood framework, the authors constrain the axion-electron coupling, obtaining $g_{aee} \le 0.81 \times 10^{-13}$ (95% CL) and, for DFSZ, $m_a \sin^2\beta \le 2.85$ meV. The result strengthens previous bounds and disfavors the $m_a \sim 4-10$ meV region favored by Galactic WD cooling hints, while highlighting the role of envelope thickness and WD mass as degeneracy-breakers. The work demonstrates the power of joint WD population modeling in globular clusters to probe axion properties with different systematics than red-giant branch analyses, and it suggests applying the method to additional clusters to tighten the bounds further.

Abstract

We analyse the cooling of white dwarfs in the globular cluster 47 Tucanae to look for evidence of axion emission affecting the rate of white dwarf cooling. If axions exist and couple to electrons, then axions could be produced at an appreciable rate in the electron-degenerate core of a white dwarf through axion bremsstrahlung from electrons. The emission of these axions would provide an additional cooling mechanism for white dwarfs that would affect the cooling rate, and hints of axions have been suggested based on observations of anomalous cooling reported for white dwarfs in the Galactic disc and halo. We performed stellar evolution simulations of white dwarf cooling that accounted for the additional energy loss due to axion bremsstrahlung from electrons, producing a suite of white dwarf cooling models for different values of the axion-electron coupling constant, as well as the white dwarf mass and envelope thickness. These cooling models are compared to observations of white dwarfs in 47 Tucanae from the Hubble Space Telescope through an unbinned likelihood analysis. The optimal model found by this analysis corresponds to the case of no axion emission with a thick white dwarf envelope, and we find a new bound on the axion-electron coupling of $g_{aee} \leq 0.81 \times 10^{-13}$ at 95% confidence level. This improves upon the previous white dwarf cooling bound for this coupling and excludes the range of values favoured by the axion hints from the anomalous cooling of Galactic white dwarfs.

Figures (28)

• Figure 1: Theoretical cooling curves for various axion mass values (left; a) and envelope thickness (right; b) when all other model parameters are fixed. In both cases, the white dwarf mass is $M_\mathrm{wd} = 0.5388~M_\odot$ and diffusion is fully on. For the series of axion mass values shown in the left panel, the envelope thickness always has a fixed value of $q_H = 2.24 \times 10^{-4}$. For the series of $q_H$ values shown in the right panel, the axion mass always has a fixed value of $m_a = 4.0~\mathrm{meV}$.
• Figure 2: CMDs showing the data space selections used in the unbinned likelihood analysis for the WFC3/UVIS data (left; a) and ACS/WFC data (right; b). The plots are focused on the white dwarf cooling sequence, with the data shown as black points. The boundaries of the white dwarf data space selection for each data set are indicated by the red curves. The evolutionary track predicted by the model in each case before accounting for photometric errors is shown as an orange curve.
• Figure 3: Joint posterior probability density distribution after marginalising over the birthrates. Slices of the distribution as a function of axion mass ($m_a$) and envelope thickness ($q_{H}$) are shown for fixed values of white dwarf mass ($M_\mathrm{WD}$). The probability density ($p$) has been scaled by its maximum value ($\hat{p}$) so that the plotted quantity is $p / \hat{p}$. The filled contours are drawn at level values corresponding to the $\sigma$ levels indicated on the legend for a two-dimensional normal distribution. The lowest mass case ($M_\mathrm{WD} = 0.5240~M_\odot$) is not shown because it is excluded at $5~\sigma$.
• Figure 4: Two-dimensional joint credible regions of axion mass ($m_a$) and envelope thickness ($q_H$) after marginalising over the other parameters. The filled contours show regions of enclosed probability containing the same probability as enclosed by a two-dimensional normal distribution for the $\sigma$ levels indicated by the legend.
• Figure 5: One-dimensional posterior probability density distributions for each parameter after marginalising over all other model parameters.