Accretion Rate Changes Detected in a Polluted White Dwarf

TL;DR

This study reports accretion-rate changes in the polluted white dwarf WD 0106-328 by tracking Ca II K and Mg II 4482 Å across six epochs over 25 years with data from VLT, Keck, and HST. Multi-wavelength modeling ties the line variations to evolving accretion from a circumstellar disk, providing empirical support for rapid diffusion in warm DAZ atmospheres and suggesting disk processing as a key driver of atmospheric metal abundances, including a potential exponential decay on decadal to centennial timescales. The ultraviolet and optical abundances reveal an Fe-dominated, metal-rich accreted material consistent with a differentiated parent body, likely composed of core and crust material with limited mantle contribution, and hint at complex planetary disruption histories. The inferred disk-processing timescale of roughly 78–96 years aligns with viscous spreading in a metal-dominated α-disk ($\alpha\approx0.1$), underscoring the value of long-term monitoring to interpret exoplanetary bulk compositions from white-dwarf pollution and to constrain debris-disk evolution.

Abstract

This letter reports statistically significant changes in the equivalent widths of MgII and CaII lines in the dusty and polluted white dwarf WD 0106-328, based on six epochs of spectroscopy using the VLT and Keck spanning 25 yr. Furthermore, the ratio of these two equivalent widths may also vary, with a 7% probability of being constant. Between 2000 and 2025, both Mg and Ca have experienced decreases in accretion rates, of approximately 20 and 60%, respectively, but with individual variation during the interim. These metal abundance decreases are the first empirical corroboration of diffusion theory in white dwarfs, which predict sinking timescales on the order of days for this star. However, the persistent atmospheric metals require a more gradual, circumstellar process, where one possibility is viscous spreading in an ionized disk of metals, consistent with $α\approx0.1$ within that formalism. The combination of optical and ultraviolet spectroscopy with the Hubble Space Telescope detects all the major rock-forming elements (O, Mg, Si, Fe), and demonstrates that Fe dominates the accreted material by mass, and that it is delivered mostly as pure metal from within a differentiated parent body. This inference is consistent with the possibility that chemically-segregated accretion may result from a combination of planetary assembly, fragmentation, disk evolution, and be observed on relatively short timescales.

Figures (5)

• Figure 1: Equivalent width time-series measurements for Caii K and Mgii 4482Å in 0106$-$328. The top panel shows the measurements for each spectral line at each epoch, and the middle panel plots their ratio. The bottom panel shows examples of variation in each feature, seen independently in two distinct instruments and observatories.
• Figure 2: Two segments of the COS spectrum shown in grey, with the adopted model overplotted in red. Absorption features are labelled by the corresponding element, and (fitted) interstellar lines are marked with dashed vertical grey lines.
• Figure 3: Abundance ratios Z/Si in the accreted material as compared to those of the bulk Earth (BE), CI chondrites, and the Sun mcdonough2000lodders2003, with representative error bars of 0.1 dex shown.
• Figure 4: The infrared light curves of the debris disk orbiting 0106$-$328 as measured by WISE in channels 1 and 2. In these plots, the starlight has been subtracted similar to published time-series dusty white dwarf studies using Spitzerswan2020noor2025. The circles are individual measurements by the spacecraft and the data points with error bars are the weighted mean and standard error of the mean for each epoch of measurements. The number of reliable detections in channel 2 and the corresponding S/N are both lower. The green dashed vertical lines correspond to the dates of the three most recent spectroscopic epochs in Table \ref{['obs']}.
• Figure 5: Same as the top panel for Figure \ref{['ews']}, but now only for Ca abundances plotted as instantaneous accretion rates. The decreasing values are fitted with a straight line whose slope is determined via $\upchi^2$ minimization, where the dashed line accounts for the individual errors, and the dotted line gives equal weight to all the data. The slopes yield exponential decay timescales of 78 (unweighted) and 96 yr (weighted), which compare favorably to viscous spreading in a metal-dominated, $\upalpha\approx0.1$ disk metzger2012.