WorksPhysics and Astronomy
Near-Ultraviolet and Optical Observations of the White Dwarf ZTF J03281219 Reveal Variable Periodicity in its Transiting Debris System
Abstract
We observed the transiting debris disk system around the white dwarf ZTF J03281219 using the Hubble Space Telescope (HST) in 2025 October, obtaining near-ultraviolet (NUV) spectroscopy with STIS over five consecutive HST orbits. Our abundance analysis supports the accretion of a chondritic body. We complemented our HST program with 16 days of optical time-series photometry beginning in 2025 October, which reveal variable periodicity and rapidly evolving transit morphology. Previous monitoring by TESS during Sectors 4 and 31 confirmed a predominant, stable transit orbital period, hr, along with a second, transient period, hr. Our 2025 autumn light curves show transits are dominated by hr with initially appearing as a weaker, secondary feature and eventually returning to predominance. We also identify a candidate transit period of 7.64 hr during the final days of the 2025 autumn campaign, although its significance and persistence remain uncertain. We add simultaneous light curves collected with ULTRACAM in 2024 November that are dominated by and show no significant color term. We further compare the transit depths in optical photometry concurrent with our NUV observations, again measuring a lack of transit-dependent extinction except during the fifth HST orbit where a reddening term may arise. The observed variable periodicity and potentially changing color behavior illuminate ongoing dynamical processes, namely collisions, and demonstrate the opportunity for studying dust physics in real time at white dwarf debris disks.
Email: [email protected]
Introduction
Both bookends of stellar evolution are influenced by circumstellar debris. Protostars are born embedded in thick, flared disks that nurse planetary formation (see [149, 150 and the references therein]). Spectroscopy of white dwarf stars indicates some of this material that coagulated into planets and planetesimals is recycled back into circumstellar debris disks many eons later. More than 40% of white dwarf spectra show photospheric absorption of heavy metals, indicating the recent accretion of planetary material ([160, 161, 87, 113]).
The debris white dwarfs accrete arrives at the circumstellar environment following the dynamical scattering of a minor planetary body by at least one larger, intact planet that survived post-main sequence evolution. Upon entering the tidal disruption radius of the star on a likely extremely eccentric orbit, the material is disrupted and ground down into an eccentric debris disk that circularizes and migrates inwards towards the star ([36, 78]). Eventually, sufficiently fine dust is sublimated and then dragged onto the stellar photosphere, manifesting the observed metal pollution (see reviews by [46, 143], and the references therein).
White dwarfs are observed to accrete from their remnant planetary systems across the single-star evolution cooling sequence ([17]). Accretion is most sensitively detected in the ultraviolet, where recent accretion can be assessed through the relative abundance ratios of silicon and carbon ([87, 113]). X-ray emission from G29–38 directly verifies the bombardment accretion of planetary debris ([30]).
Derived relative abundances from UV and optical spectra are generally consistent with bulk Earth or chondritic natures ([52, 19] (Xu et al. 2014; Xu et al. 2019a; Xu & Bonsor 2021)), and on rare occasion are consistent with water-rich parent bodies ([47, 132]). UV spectroscopy is often required to obtain the most-accurate census of pollutants, even for cooler white dwarfs ([74]), though systematic discrepancies between the derived abundances from the optical and ultraviolet persist ([52, 90] (Xu et al. 2019a)). Mid-infrared spectroscopy illuminates the mineralogy of the debris, where 10-m silicate emission from the disk dominates ([49, 121]).
White dwarf debris disks are ubiquitously dynamic ([109]), having their emitting areas and particle size distributions regularly reshaped by collisional cascades (Kenyon & Bromley 2017a,b; Swan et al. 2021). Like their protoplanetary counterparts, debris disks around white dwarfs are likely flared and fall along random inclinations ([79, 8, 12]). The systems with the most edge-on inclinations should cast dimmings as the debris transits the stellar cross section. More than 20 transiting debris systems are now known ([133, 112]), with most hosting the debris well beyond the nominal Roche limit ([144]). All but five transiting debris systems appear metal polluted in low-resolution spectroscopy—all of five which likely travel extremely eccentric, long-duration orbits ([66, 133])—and at least two show infrared excesses: WD 1145+017 (Xu et al. 2018) and SBSS 1232+563 ([35, 129]).
Transit recurrence timescales indicate a distribution of orbits. There are at least two cases of debris orbiting near the tidal disruption radius of the host star, WD 1145+017 at 4.49 hr ([137]) and ZTF J1944+4557 at 4.9704 hr ([67]), next to the 5.910 hr period that briefly emerged during a deep transit of ZTF J0436+5728 ([133]). Three others show orbital periods ranging from about half a day to one day: ZTF J0328−1219 (9.937 hr and 11.2 hr; [136]), SBSS 1232+563 (14.842 hr; [72]), and WD 1054−226 (25.02 hr; [50]). ZTF J0139+5245 enters deep transits roughly every 107 days ([135]). The probability distribution of orbital periods for:VEVENT tidally disrupted asteroids around white dwarfs matches this spread, roughly covering 4 hr d and peaking over 12–16 hr (Li et al. 2025).
Recurrence timescales are unmeasured at all other transiting systems. Several systems have shown only a single transit in their ZTF light curve across over eight years of monitoring, including ZTF J0347−1802 [66], ZTF J1013−0427 [13], and five systems published in van Roestel et al. [133]. SBSS 1232+563 enters deep transits sporadically every 6-8 years, in addition to the 14.8 hr periodicity observed by TESS following the 2023 deep transit [72]. ZTF J0923+4236 undergoes deep transits every few weeks without clear periodicity [66, 6]. Several systems that show transits on short timescales have yet to be verified to show periodic transits [66, 13].
ZTF J0328−1219 (Gaia DR3 5161807767825277184) was confirmed as a spectral type DZ white dwarf host to transiting debris by Guidry et al. [66]. Periodic transits repeating at hr were observed in follow-up light curves collected during 2020 December and 2021 February and corroborated by photometry from TESS Sectors 4 and 31 [136]. A second period, hr, was detected exclusively during the first half of TESS Sector 31 in 2020 October. High-resolution spectroscopy revealed circumstellar gas absorbing in the optical at the sodium D doublet and in the near-infrared at the central line of the calcium II triplet ( Å; [136]). Evidence for circumstellar dust around ZTF J0328−1219 in the near-infrared is lacking. Photometry from un-WISE of the system are consistent with the white dwarf model photosphere, although there is potentially significant excess emission at the J-band [136]. The system is not observed to vary in its un-WISE light curve collected by NEOWISE [65].1 Optical survey photometry records evidence for transits occurring since at least 2006, including a dimming of roughly 0.3 mag beginning in 2012 and lasting through 2014 [6]. The transits were consistent with being colorless across both - and -band ZTF DR5 photometry [136] and follow-up - and -band light curves [56].
In this manuscript we publish follow-up, time-series photometry and spectroscopy of ZTF J0328−1219. These include Hubble Space Telescope (HST) spectroscopy and time-series photometry in the near-ultraviolet (NUV), and also ground-based, optical observations. Our work tightens constraints on the relative abundances of the accreted debris, confirms consistent transit depths across simultaneous -, -, and -band photometry in 2024 November and between near-concurrent NUV and optical photometry in 2025 October, and demonstrates variable periodicity in the system with the predominant orbital period changing from in 2024 November to in 2025 October. In § 2 we recount our various observing campaigns. We demonstrate our periodicity analysis in § 3. § 4 provides our tests for transit color terms. We update the model spectrum of the system in § 5. We contextualize these results and offer explanations for the observed dynamical activity in § 6 and recapitulate our key findings in § 7.
Observations of ZTFJ0328-1219
Optical, Ground-Based, Time-Series Photometry
We coordinated several follow-up campaigns for optical, ground-based, high-speed, time-series photometry of ZTF J0328−1219, especially to complement our HST program. The methods by which we reduced these photometric data and extracted light curves varied depending on the facility. We describe these campaigns and associated reductions chronologically by the observing date. Our observing log of these photometry is available in Table A1. We plot these light curves in both Figure 1 (2024 November ULTRACAM) and Figure 2 (2025 autumn light curves).
Figure 1. 2024 November 2-5 ULTRACAM -band (black) and (gray) light curves phase folded on hr (left) and the conditional entropy periodogram of combined, averaged & light curve (right). A prominent transit centered around Phase persists across the four consecutive nights. We annotate on the periodogram and as red and light blue triangles, respectively. Only the A period is significant to the nominal threshold (see § 3.1), delineated by the dash-dotted green line. The disappearance of features in the flat light curve indicates achromatic transits (§ 4.1).

Figure 2. 2025 October and November light curves of ZTF J0328−1219 phase-folded on hr, the best measured period for this epoch (§ 3). Light curves are color-coded by bandpass (: black, & : green, : red, NUV: light blue, and unfiltered: white) and each observatory is assigned a unique marker shape. Repeating patterns are detectable early on, such as at Phase during Cycles 1, 4, and 6. The transit morphologies evolve much more rapidly at later cycles and no longer show obvious repeating patterns at this period. This dynamical activity is corroborated by the changing periodograms in the bottom panel of Figure 3. The offset between Cycles 16 and 17 is boosted to account for the enhanced total flux excursion observed by HST and LAST. Some light curves are binned by a factor of two or three for improved visibility.
| Date Obs (UTC) | Telescope | Filter | (s) | Duration (hr) | FWHM () | P2P (%) | |
| 2024-11-02T02:57:33.104 | ULTRACAM | 6252 | 3 | 6.02 | - | 1.7 | |
| 2024-11-02T02:57:33.104 | ULTRACAM | 6252 | 3 | 6.02 | - | 2.1 | |
| 2024-11-02T02:57:36.651 | ULTRACAM | 2084 | 10 | 6.02 | - | 4.8 | |
| 2024-11-03T01:45:02.239 | ULTRACAM | 7531 | 3 | 7.25 | - | 1.7 | |
| 2024-11-03T01:45:02.239 | ULTRACAM | 7531 | 3 | 7.25 | - | 2.1 | |
| 2024-11-03T01:45:05.786 | ULTRACAM | 2511 | 10 | 7.25 | - | 3.8 | |
| 2024-11-04T01:56:51.445 | ULTRACAM | 7320 | 3 | 7.05 | - | 2.0 | |
| 2024-11-04T01:56:51.445 | ULTRACAM | 7320 | 3 | 7.05 | - | 2.4 | |
| 2024-11-04T01:56:54.992 | ULTRACAM | 2440 | 10 | 7.05 | - | 3.9 | |
| 2024-11-05T02:54:50.387 | ULTRACAM | 6297 | 3 | 6.06 | - | 1.5 | |
| 2024-11-05T02:54:50.387 | ULTRACAM | 6297 | 3 | 6.06 | - | 1.8 | |
| 2024-11-05T02:54:53.935 | ULTRACAM | 2099 | 10 | 6.06 | - | 3.7 | |
| 2025-10-18T08:51:38.00 | OAN-SPM | 171 | 60 | 3.82 | - | 1.5 | |
| 2025-10-18T18:17:26.83 | PAT | 117 | 45 | 1.82 | - | 7.2 | |
| 2025-10-18T23:57:16.36 | Warw | 1722 | 10 | 5.88 | - | 6.8 | |
| 2025-10-19T06:39:56.01 | McD | 1297 | 10 | 3.61 | - | 4.4 | |
| 2025-10-20T08:42:50.00 | OAN-SPM | 330 | 30 | 4.06 | - | 1.3 | |
| 2025-10-21T00:00:25.00 | Stew | 146 | 50 | 2.47 | - | 1.3 | |
| 2025-10-21T02:50:31.11 | Warw | 115 | 60 | 1.96 | - | 1.5 | |
| 2025-10-21T06:03:51.01 | PTO | 900 | 20 | 6.25 | 2.50 | 1.7 | |
| 2025-10-21T08:35:42.00 | OAN-SPM | 161 | 30 | 1.85 | - | 1.5 | |
| 2025-10-22T05:37:08.22 | PTO | 186 | 20 | 1.31 | 2.41 | 2.3 | |
| 2025-10-22T09:03:10.01 | PTO | 455 | 20 | 3.29 | 3.59 | 1.6 | |
| 2025-10-23T01:37:31.19 | Warw | 648 | 20 | 3.94 | - | 2.1 | |
| 2025-10-23T05:27:56.02 | PTO | 525 | 20 | 3.65 | 3.61 | 1.5 | |
| 2025-10-23T09:45:20.19 | PTO | 369 | 20 | 2.56 | 3.04 | 1.6 | |
| 2025-10-24T05:27:11.07 | PTO | 983 | 20 | 6.83 | 2.87 | 1.5 | |
| 2025-10-25T01:28:30.01 | Warw | 651 | 20 | 3.95 | - | 2.3 | |
| 2025-10-25T09:16:11.41 | LDT | 550 | 15 | 2.98 | 0.99 | 1.1 | |
| 2025-10-25T19:19:30.543 | HST | NUV | 192 | 60 | 6.73 | - | 4.7 |
| 2025-10-25T20:04:41.808 | LAST | Open | 540 | 30 | 6.30 | - | 6.8 |
| 2025-10-26T04:59:15.8 | HAO | Open | 96 | 121 | 3.45 | - | 3.7 |
| 2025-10-26T05:18:24.5 | RVO | Open | 350 | 67 | 6.76 | - | 2.3 |
| 2025-10-26T05:21:13.00 | LDT | 183 | 15 | 1.10 | 1.36 | 1.1 | |
| 2025-10-26T06:29:04.57 | LDT | 332 | 15 | 1.85 | 1.36 | 0.9 | |
| 2025-10-26T08:34:39.45 | LDT | 24 | 15 | 0.13 | 1.06 | 0.7 | |
| 2025-10-26T09:07:38.47 | LDT | 565 | 15 | 3.18 | 1.20 | 1.6 | |
| 2025-10-27T00:56:47.70 | Warw | 202 | 60 | 3.63 | - | 2.1 | |
| 2025-10-27T04:56:45.3 | HAO | Open | 80 | 111 | 2.84 | - | 4.1 |
| 2025-10-27T05:25:20.8 | RVO | Open | 343 | 60 | 6.59 | - | 2.1 |
| 2025-10-28T04:32:07.5 | HAO | Open | 72 | 111 | 2.57 | - | 4.6 |
| 2025-10-28T05:35:46.9 | RVO | Open | 334 | 60 | 6.36 | - | 2.1 |
| 2025-10-29T04:39:26.9 | HAO | Open | 55 | 111 | 2.01 | - | 3.3 |
| 2025-10-29T05:22:13.4 | RVO | Open | 343 | 60 | 6.52 | - | 1.9 |
| 2025-10-30T05:02:10.6 | HAO | Open | 77 | 111 | 2.59 | - | 5.3 |
| 2025-10-30T06:22:14.1 | RVO | Open | 194 | 60 | 3.70 | - | 1.7 |
| 2025-10-31T05:07:39.5 | RVO | Open | 292 | 60 | 5.59 | - | 1.9 |
| 2025-10-31T05:45:58.7 | HAO | Open | 30 | 111 | 1.33 | - | 5.0 |
| 2025-11-01T05:05:23.1 | RVO | Open | 292 | 60 | 5.56 | - | 1.9 |
| 2025-11-02T05:21:06.5 | RVO | Open | 286 | 60 | 5.72 | - | 2.1 |
| 2025-11-03T05:22:55.2 | RVO | Open | 533 | 20 | 5.65 | - | 2.7 |
Table A1. Table A1. Journal of Time-Series Photometry Observations of ZTF J0328−1219. Note—Telescope key names are the same defined in § 2.1. We approximate the average seeing using FWHM measurements taken from the hipercam pipeline outputs. P2P is the average point-to-point scatter in % used to adjust the normalization of our light curves (§ 2.1.1).
While reduction methods varied from telescope to telescope, we did not normalize any of our light curves to a low-order (linear or quadratic) polynomial because of our inherent uncertainty of the light curve morphology—we never definitively observed the stable, out-of-transit flux of ZTF J0328−1219. Low-order polynomial division corrections are usually performed to correct for differential extinction introduced by the airmass of the target varying over the observing baseline (see, e.g., [116]).
ULTRACAM — We observed ZTF J0328−1219 for a total 26.39 hr across the four consecutive nights of 2024 November 2-5 using the ULTRACAM [38] photometer mounted onto the 3.58-m New Technology Telescope at La Silla Observatory in Chile. This instrument is a triple-beam, frame-transfer CCD imager. During each night the target and two or more field stars were observed simultaneously in bandpasses similar to the SDSS filters but with higher throughput. Individual frame times were 3.47 s in all three bands, but in the -band three frames were co-added to improve signal-to-noise (effective exposure times of 10.40 s). We show both the -band and residual light curves in Figure 1. For our analysis of the color term to these transits (Section 4.1), we binned the - and -band light curves by a factor of three so we can compare to the -band photometry with consistent exposure times. Following Farihi et al. [48], we processed the raw images using the reduction software hipercam2 (Dhillon et al. [39]). Each frame was corrected for bias and flat fielded using sky flats taken each night, with typical read noise in the range 2–3 ADU in each of the three bias frames. We carried out differential photometry on the science target and two field stars for reference. The reduction package tracks the mean full width at half maximum for all three stars, and scales photometric apertures to twice this size, with sky annuli that are typically four to eight times larger. Each light curve was constructed from the science target flux divided by the sum of these two reference stars, normalized to a median of one. We applied barycentric corrections to the timestamps using the methods of Eastman et al. [43].
Whole Earth Telescope (WET) — We utilized the WET network during the XCOV31 campaign3 to observe ZTF J0328−1219 as a secondary target at several sites across the globe in late 2025 October. These included the Ruca imaging system aboard the 1.5-m Harold Johnson Telescope at Observatorio Astronómico Nacional at San Pedro Mártir (OAN-SPM; Zazueta et al. [158]; October 18, 20, 21), the 0.43-m Photometric Auxiliary Telescope at Xingming Observatory using the QHYCCD Cameras2 Capture instrument (PAT; October 18), the Warwick 1m Telescope at Roque de los Muchachos Observatory with the Andor iKon-L CCD camera (Warw; Gänsicke et al. [53]; October 18, 21, 23, 25, 27), McDonald Observatory 2.1-m Otto Struve Telescope with the ProEM photometer (McD; October 19), the 1.54-m Kuiper Telescope at the Catalina Station of Steward Observatory with the Mont4K CCD imager (Stew; October 21). Exposure times varied from 10–60 s across these sites, aggregating to a total non-overlapping baseline of 35.05 hr. The OAN-SPM, Warw, and McD observations used a filter, while the PAT and Stew observations were collected through a - and -band filter, respectively.
We reduced the WET photometry identically, regardless of facility, following the procedure developed by Provencal et al. (2012). We used Maestro (Dalessio [31]) to perform aperture photometry with circular annuli ranging between 2 and 20 pixels from the centroid of the target, as well as for up to 29 field stars in the images to use as comparisons. We then used phot21c (Vanderbosch [134]) to create the final divided light curves by removing outliers, optimizing comparison star selections, and applying barycentric corrections to the GPS timestamps using Astropy (Astropy Collaboration et al. [3, 4, 5]).
Perkins Telescope Observatory (PTO) — At the 1.8-m PTO we observed ZTF J0328−1219 using the PRISM instrument (Janes et al. [77]) over the four consecutive nights of 2025 October 21-24 for a total of 23.90 hr. All our photometry were collected using the blue-transmitting, red-cutoff filter with integration times of 20 sec with windowed chips to 410×380 pixels at a plate scale of 0.39″ per pixel, reducing the overhead from readouts to typically 5 s.
We used custom-built Python-based routines to calibrate our raw science images by subtracting CCD biases and dome flats. We then reduced our images using the hipercam pipeline, doing PSF photometry with variable apertures and PSF profiles, all conducted in the optimal extraction mode (Naylor [106]). To account for sky backgrounds we drew circular annuli 20–30 pixels from the centroids of our target and chosen comparison stars. To optimize our photometry we modified particular parameters, mainly search_half_width, if the sky conditions reduced the S/N enough to hamper centroid retrieval. We generated divided light curves
using the phot2lc, clipping outliers and applying barycentric corrections to the timestamps.
Lowell Discovery Telescope (LDT) — We immediately followed our PTO campaign with two more consecutive nights of high-speed, time-series photometry on 2025 October 25-26 using the Large Monolithic Imager Massey et al. [101] 4.3-m LDT. We observed for a total of 9.23 hr over the two nights with 15 s exposures in the -band. We windowed the CCD to 500500 pixel arrays with 33 binning yielding a plate scale of 0.36″ per pixel. Readout times averaged to roughly 4.55 s. Our observations on October 26 were heavily impacted by intermittent clouds.
We reduced our LDT/LMI photometry very similarly to the PTO/PRISM photometry, using hipercam in the same manner with the sole change of using the normal extraction mode instead of optimal. We used phot2lc to generate optimized light curves with barycentric-corrected timestamps.
Large Array Survey Telescope (LAST) — We used three 28-cm telescopes within the LAST array Ofek et al. [110] to observe ZTF J0328−1219 on 2025 October 25, providing us with 6.30 hr of unfiltered optical photometry concurrent with our HST program (§ 2.2).
We reduced the images and performed PSF photometry using the standard LAST pipeline Ofek et al. [111] one telescope at a time and then coadded the resultant photometry. We carried out absolute photometric calibration following the method of Garrappa et al. [54], which models the instrumental response and atmospheric transmission for each image. We applied an additional relative photometric calibration of our target using the comparison star Gaia DR3 5161911602954639232, which has no Gaia variability flags nor is observed to vary in ZTF, as a comparison star to calibrate our PSF photometry. We applied barycentric corrections to the LAST timestamps using Astropy.
Hereford Arizona Observatory (HAO) — We observed ZTF J0328−1219 with the HAO 11-in Celestron telescope based in Hereford, Arizona, the nights of October 26, 27, 28, 30, and 31, exposing for 111 s without a filter. Readouts were typically about 10 s. We used MaxIm DL to calibrate our science images and a custom Excel created by author B.L.G. to optimize the MaxIm DL photometry. We used Astropy to apply barycentric corrections.
Raemor Vista Observatory (RVO) — We observed ZTF J0328−1219 with the RVO 1.1-m telescope based in Sierra Vista, Arizona, the nights of October 26-31 and November 1-3, exposing for 60 s on all nights except November 3 where we took 20 s integrations, with typical overheads of roughly 7 s. Every light curve was collected without a filter. We used AstroImageJ to calibrate and analyze our science images and applied barycentric corrections to the timestamps using Astropy.
Light Curve Normalization
The true, out-of-transit flux baseline of ZTF J0328−1219 is unconstrained because we continuously observed it to be transited and we do not know the optical depth of the transiting debris. The phot2lc software outputs mean-normalized light curves in units of relative flux, which we notate here as . We work in this relative flux space for our period search analysis (§ 3).
For our color analysis and light curve presentation, we sought to define a zero point for each individual light curve. We re-normalize the relative fluxes to a factor we define :
Here, is equal to the median of the relative flux values greater than the percentile corresponding to , where is the average point-to-point scatter of the light curve in percent. For example, if a light curve has an average point-to-point scatter of 5%, then is the median of all fluxes brighter than the 95th percentile of the light curve. The typical point-to-point scatter levels of our light curves range from about 1 − 5% (see the observing log in Table A1). We plot light curves in values as “Local Normalized Flux.”
Hubble Space Telescope/STIS NUV Spectroscopy
We targeted ZTF J0328−1219 with HST in Cycle 32 program GO #17789 Guidry et al. [64], obtaining five near-ultraviolet (NUV) spectra over five consecutive HST orbits beginning 2025 October 25 18:54:27 and lasting through 2025 October 26 02:09:02 (all times UTC). We designed our experiment to utilize the NUV MAMA detector in the Space Telescope Imaging Spectrograph (STIS) with the G230L grating and the 52×0.5″ aperture in TIME-TAG mode centered on Å with an average dispersion of Å/pixel (). Pointing was nominal for each orbit, with RMS jitter <0.003″ and no pointing excursions >0.006″ from the target, allowing us to confidently infer slit losses were minimized. The respective integration times, discounting overheads, for each orbit were 1940, 2505, 2482, 2482, and 2482 s.
For our spectroscopic analysis (§ 5), we utilize the Hubble Advanced Spectral Product (HASP; Debes et al. [34]) co-added and absolute flux-calibrated spectrum available online at MAST.
NUV Light Curve Extraction
We generated an NUV light curve from our time-tagged STIS spectroscopy by performing aperture photometry on 2D spectra images generated with 60-s effective integrations. Relying on the guidelines provided by the STIS Data reduction guide Rickman & Brown [120] and STIS Time-Tag Analysis guide Dashevsky et al. [32], we first partitioned the spectra into 60-s effective exposures using lightcurve (Ely et al. [45]; Sandhaus et al. [124]). We applied barycentric corrections to the timestamps using the IRAF-based routine ODELAYTIME as implemented in PyRAF (Science Software Branch at STScI [126]). We generated accum images with 60-s exposures using calstis as implemented by stistools.4
We performed aperture photometry on the background-subtracted, flat-fielded flt images outputted by calstis using photutils (Bradley et al. [21]). We drew rectangular apertures to sample the spectrum from 1800–3100 Å, yielding dimensions of 838 pixels wide by 10 pixels in height. The target spectrum in the calibrated 2D spectra tends to be Gaussian with a typical standard deviation of 1.5–2.0 pixels, hence our choice for a height of 10 pixels to sample roughly ±5σ about the center of the spectrum in each flt image. The pixel locations of the target spectra in these images are consistent, allowing us to use the same aperture for all images except for the fifth orbit, which we shifted down by a single pixel.
The validation of our aperture photometry was meticulous. We estimated flux uncertainties using the error images packaged within the flt FITS image arrays by passing them into the photutils aperture photometry routine. We divided the summed flux in counts and their associated uncertainties by the mean flux of the entire five-orbit light curve. Orbit by orbit, we compared the mean relative flux of the aperture photometry light curve to a light curve we generated from integrating the flux calibrated, time-tagged 1D spectra from 1800–3100 Å with the same 60-s effective exposure times and excluding pixels with problematic data quality flags. We adjusted the mean relative flux of each orbit to be in unity with that of the integrated 1D spectra light curve, the differences of which were 1–3%, to calibrate our aperture photometry. We kept the aperture photometry light curve for our analysis because its point-to-point scatter was 4.7%, compared to 6.2% for the integrated 1D spectrum light curve. For a second validation, we compared our aperture photometry light curve to the light curve we generated using the software developed by the Rocky Worlds DDT program5 (Redfield et al. [119]). The Rocky Worlds light curve had a higher average point-to-point scatter at 7.1%, motivating our selection of the aperture photometry light curve for our analysis. Finally, we normalized the aperture photometry light curve in the same fashion as our ground-based light curves (§ 2.1.1).
SOAR/Goodman Optical Spectroscopy
We obtained a new optical spectrum of ZTF J0328–1219 on 2025 October 29 using the Goodman High Throughput Spectrograph (Clemens et al. [25]) with the 4.1-m Southern Astrophysical Research Telescope (SOAR) at Cerro Pachón, Chile, facilitated through NOIRLab AEON Program 2025B-134660. We took four 300-s exposures using the red camera with the 400 line/mm M1 grating and the 1.0″ slit, yielding a resolving power of . We reduced these spectra and produced a final 1-D coadd using PyPeIt (Prochaska et al. [115]). We flux calibrated the spectrum using a sensitivity function in PyPeIt generated from the ESO standard star HR 7596, which we observed with two 0.4-s exposures on the same night with the same setup used for our target.
Long-Term and Rapid Variable Periodicity in the Transits
Our ground-based, follow-up, time-series photometry campaign was designed to begin several days in advance of our HST program so that we could readily identify and track repeating features, such as those centered around Phase in Figure 1. Our light curves quickly revealed that the system entered a new state: transits that recurred at period were no longer apparent. Transit morphologies instead aligned at , a period previously exclusively detected in TESS Sector 31 observations of ZTF J0328–1219 (Vanderbosch et al. [136]), and quickly lost coherence over consecutive orbits.
Here we detail our period search analysis of our ground-based, follow-up, optical photometry collected during both epochs of 2024 November and 2025 autumn. We keep only the highest fidelity light curve when overlapping coverage is present. Our period search draws on previous work that combines conditional entropy and Lomb-Scargle periodograms to resolve periodicity in transit debris (Guidry et al. [67]). We did not find improvement using the Fast Periodicity Weighting algorithm (FPW; Finkbeiner et al. [51]), as the forest of similarly prominent peaks in the and regions made period identification non-trivial at bin choices of 5, 10, and 20. FPW may struggle at reliable period detection for embedded transiting debris systems whose light curve morphologies can be rapidly dynamic. The best period values from both algorithms are reported in Table 1.
| Method | |||
| hr | hr | hr | |
| 2024 Nov | |||
| CE | - | - | |
| 2025 Oct/Nov | |||
| CE | † | ||
| LS | |||
| Pyriod | - |
Table 1. Period Determinations from Conditional Entropy and Lomb-Scargle Periodograms
Conditional Entropy
2025 October
Conditional entropy as a period-finding algorithm is optimized for non-sinusoidally varying, irregularly-sampled time-series, being inherently agnostic to the intrinsic model driving observed variability (Graham et al. [61]), making it the ideal tool for measuring orbital periods of transiting debris (Guidry et al. [67]).
We use GPU-accelerated Conditional Entropy (gce6; Katz et al. [80]) to period search our optical light curves from 2025 October. Our search design follows precedents established by El-Badry et al. [44]; Guidry et al. [67]. We first search the entire dataset across a frequency grid with resolution $1/T$, where is the total observing baseline. We oversample the grid by a factor of 100 and truncate the grid at 5000 Hz; no variability has been detected in the system more rapid than 1000 Hz (Vanderbosch et al. [136]). Our search evaluates over a grid of 30 phase bins and 10 flux bins. We normalize the resultant periodogram by subtracting a sliding median filter with a width of 7.20 Hz (equivalent to 1001 resolution elements), yielding the periodogram in the middle panel of Figure 3. The minimum in the periodogram is located at 24.8747 Hz$, or equivalently 11.17 hr. We attempt to optimize this measurement by fitting a Gaussian to the core of the minimum, however there is a secondary feature at around 11.32 hr that skews the fit. Instead, we only record the HWHM as an estimate of the uncertainty to our measurement, yielding hr. The region around is even more definitively non-Gaussian, requiring us to again simply record the local minimum of the periodogram and we adopt a 0.2 Hz uncertainty: hr. Analogous to Figure 2, we illustrate our 2025 autumn light curves folded over in Figure B.1. We annotate the location of and and their harmonics in Figure 3, several of which are significant to the nominal detection threshold (, for mean and standard deviation ; Katz et al. [80]). The typical 0.5:1:2 harmonic pattern (Graham et al. [61]) is especially notable for .

Figure 3. Lomb-Scargle (top) and conditional entropy (middle and bottom) periodograms of our 2025 autumn optical light curves of ZTF J0328−1219. The top and middle panels incorporate the entire dataset, while the bottom panel is partitioned into three windows of 4 − 7 nights. Each panel shares the same frequency axes. Multiple harmonics of (red triangles) and (light blue triangles) are significant beyond the bootstrapped 0.1% false-alarm probability and 10 thresholds for the Lomb-Scargle and conditional entropy periodograms, respectively, each delineated as dash-dotted green lines. In the top panel we inset the spectral window to clarify our resolution. We inset in the middle panel our Gaussian fit to the core of . The bottom panel highlights rapid variations in the periodicity of the transits, where over the course of the campaign the light curves shift from being dominated by to . A new period emerges during the final window, (yellow triangle). Aliases are likely present, namely . The rapidly varying periodicity suggests ongoing dynamical activity within the system that is bringing material from the various orbits into and out of view, matching the morphological complexity within the light curves of Figure 2.
To test the stability of these periods across our campaign, we partition the photometry into three windows, each a few days in length. We define trial frequency grids and phase-flux grids identically to our global search, and normalize the calculated periodograms using a sliding median filter with a width of 1001 resolution elements (or widths of 16.9, 37.2, and 22.0 Hz, respectively). These periodograms are depicted in the bottom panel of Figure 3. The periodogram of the 2025 October 18-24 data indicates signals from and and their second harmonics without higher-order structure. The periodogram takes an entirely different form over 2025 October 25-28, becoming dominated by higher-order harmonics of both periods, whose emergence suggests a more complicated transit morphology, in turn helping explain the absence of repeating patterns in Figure 2 during this window. The third window of 2025 October 29-November 3 shows greatly diminished signal at and a preference for . Like the first window, the transits are simpler with no power existing above the fourth harmonic. A new “C” period notably emerges within this window, hr, measured again from the local minimum and assuming 0.2 Hz uncertainty. Its second harmonic is identifiable and marked. We defer further analysis of this new period to § 3.3.
2024 November
We also period search our 2024 November ULTRACAM photometry using gce, adhering to the same method as with our 2025 autumn campaign. This conditional entropy periodogram is shown as the right panel of Figure 1. We measure the minimum within the region at hr after selecting a 0.2 Hz uncertainty, agreeing with the original measurement by Vanderbosch et al. [136]. We recover the best alignment when folding the light curves over 9.937 hr as found in Figure 1. and are not significant during this epoch.
Lomb-Scargle Periodogram (2025 Autumn)
We also compute a Lomb-Scargle periodogram (Lomb [94]; Scargle [125]) over the same trial frequency grid used in our conditional entropy search. We employ the Astropy implementation using PSD normalization with nifty-ls (Garrison et al. [55]) as the evaluating method. We convert the output power, , into amplitude: , for many points in the time series. This periodogram is plotted in the top panel of Figure 3, where a quick inspection reveals a complex structure with significant power from and and several of their harmonics, along with possible aliases. We bootstrap a 0.1% false-alarm probability significance threshold by simulating 10,000 periodograms with replacement (VanderPlas [138]). The maximum frequency within the region equates to hr, where we assume an uncertainty of 0.2 Hz, and resonates out to at least its eighth harmonic. We detect at a much weaker, albeit still significant amplitude where hr. Several harmonics of this period also appear, with the second harmonic appearing to be merged with a higher-amplitude, unidentified peak.
We attempt to refine these periodicity measurements using Pyriod (Bell [11]), where we fit sinusoids of frequencies and to the light curves. Because linear least squares fits tend to underestimate uncertainties on fits to time series, we calculate uncertainties in the frequency, amplitude, and phase following Montgomery & O’Donoghue [105]. We find best-fit values of hr and hr, both in agreement with our conditional entropy measurements to at least . The less accurate measurement may be a consequence of the nearby beating or aliasing and relatively weaker signal.
7.64-hr Drifter in 2025 October & November
We mark in the bottom panel of Figure 3 a potential new period that emerges during the final few days of
**Note—**For each period, conditional entropy (CE) measurements are taken directly from the minima of the cores of each period. The Lomb-Scargle (LS) measurements are likewise read directly from the local maxima. We assume uncertainty, except for the CE measurement of during 2025 October and November, which we were able to fit with a Gaussian and record the HWHM as a conservative error. We also show our measurements derived using Pyriod with uncertainties derived using the methods from Montgomery & O’Donoghue [105]. Hyphens denote non-detections. †: We refine this measurement by eye, observing the transits best phase at hr (Figure B.2).
our 2025 autumn observing season, hr. We are able to map this period to a recurring transit captured by our RVO light curves during the consecutive nights of 2025 October 29-November 1. Folding these light curves over this period, however, does not align the deep transit in phase. Instead, we find hr best reproduces the periodicity, which we plot in Figure B.2. We formally adopt hr as the measurement of this new period. It is unclear how stable this period is and its formal significance in the periodogram is less straightforward to assess. Our November 2-3 RVO light curves lack the phase coverage needed to observe the transit recur. The emergence of this new transit period evokes the “drifters” at WD 1054−226, which were repeating transits found on orbits clearly distinct from the 25.02-hr fundamental orbital period (Farihi et al. [50]).
An Alias of PA at Approximately 7hr
We briefly note that during both the 2024 November and 2025 autumn observing seasons we detect a significant period of about 7 hr in our periodograms. Conditional entropy indicates a period of 6.96 hr within our 2024 November ULTRACAM light curves. We recover power sustained near this period in both our Lomb-Scargle and conditional entropy periodograms of the 2025 autumn light curves at 6.92 hr. A sharp minimum emerges at 7.07 hr during the third window of our observations (2025 October 29-November 3), traced by the pale blue conditional entropy periodogram in the bottom panel of Figure 3.
We infer this period is an alias of . Our ULTRACAM light curves sample seven cycles of . If this period is the second harmonic of a 13.92 hr modulation, our light curves sample five cycles. When folded over 13.92 hr, we see the deep transit align in phase at cycles 0 and 5. Indeed, . We read the presence of this alias within our 2025 autumn light curves as additional evidence for the persistence of . We mark on Figure 3 the locations of this alias at , which appears distinct and unrelated to the nearby .
Other aliases are likely present in our periodograms. For example, peaks are present at both roughly 33.3 and 61.2 (equivalently 8.35 and 4.54 hr) appear significant, yet we observe no discernible repeating patterns in the light curves folded over these periods. The same is true for the peak at roughly 44 seen in the October 29-November 3 window. We do not exhaustively test for variability at every nominally significant period, only the most prominent.
Transit Color Term Analysis
Previous attempts to measure a color term to the transits at ZTF J0328−1219 found the - and -band transit depths consistent with each other, both using the public ZTF DR5 light curve (Vanderbosch et al. [136]) and follow-up photometry from 2023 December–2024 January (Gary & Kaye [56]). We update this exploration using our simultaneous multi-color ULTRACAM photometry from 2024 November, when transits recurred at , and also with our near-concurrent NUV and optical light curves from 2025 October, when was predominant along the line of sight. We explore linear transit color terms of the form , where the slope corresponds for given blue () and red () bandpasses.
Colorless Transits along PA in 2024 November
We measure the transit depths within our simultaneous ULTRACAM photometry to be slightly inconsistent across the wavelengths sampled. We emulate previous experiments using our simultaneous ULTRACAM photometry to search for linear trends in the , , and color distributions (Farihi et al. [50]; Guidry et al. [67]). We use orthogonal distance regression (Boggs & Donaldson [18]) as implemented by scipy (Virtanen et al. [145]) to determine the extinction laws between these various blue and red bands. The best-fit color terms are as follows: , and . Since only the law is consistent with no color term, we evaluate the significance of the possible bluing from to and the possible bluing from to by running an MCMC simulation developed by Bailey [7], where we recover the same color terms with uncertainties that agree out to three decimal places. We further compare each distribution against the null hypothesis of colorless transits, modeled by a horizontal line with a fixed slope of zero, finding each distribution is consistent with a zero-slope trend. The residual variances of the best-fit and zero-slope lines are both 0.81 for the vs. law, 0.86 and 0.90, respectively, for the vs. law, and both 0.85 for the vs. law. Despite the marginal preference for the bluing from to , we cannot claim a detection of any color term over colorless transits with rigid statistical significance.
Without constraints on the mineralogy we can only make tenuous estimates on the particle size distributions of the transiting debris from these ULTRACAM benchmarks. Dust grains that extinguish starlight at equal efficiencies with characteristic sizes satisfy (Xu et al. 2018). The limits from our ULTRACAM photometry then require µm. The distribution of local ISM dust grains peaks at an effective size of µm [102, 85, 84], and characteristically extincts starlight by and [146, 50]. Dust-less transits by macroscopic debris can be produced via the YORP effect [140], or tidally disrupted rubble pile debris that has yet to be collisionally ground-down into smaller grains [103, 127]. Alternatively, the colorless transits we observe may simply reflect the optical depth of the transiting debris. Irrespective of the grain size distribution, disrupted debris that is optically thick will cast transits with no chromaticity [76].
Possibly Variable Color Terms between the NUV and Optical in 2025 October
Inspired by a previous experiment on WD 1145+017 (Xu et al. 2019b), in Figure 4 we compare the difference in the transit depths between our HST/STIS light curve in the NUV to the near-concurrent optical follow-up obtained with LAST. For this measurement we consider only observations occurring during the second, third, and fourth HST orbits (no LAST data overlap with the first HST orbit); observations during the first HST orbit are the most normalization and noise dominated. We define NUV and OPTICAL bands by convolving our HST/STIS spectrum and the LAST QE profile through our model spectrum (§ 5), yielding effective wavelengths of Å and Å.

Figure 4. Comparison between the transit depths as seen in the NUV with HST/STIS to the optical as observed by LAST (binned from 30-s to 60-s exposures to match the effective HST sampling). We calculate the transit color term in the right panel using only data separated by no more than 60 s. We use orthogonal distance regression to measure the reddening law . The residual variance of the best-fit line (4.06; solid gold line) is preferred to the null hypothesis, the fixed horizontal line (5.84; dashed gray line). This color term is intrinsic to the data during fifth HST orbit; we measure considering only the second through fourth HST orbits. The particle size distribution could significantly vary dip-to-dip and over consecutive orbital cycles as collisions rearrange the transiting debris, however, the fifth orbit data are the most normalization-dominated and their transits the shallowest. Over HST orbits 2-4, NUV starlight is constantly preferentially extincted given by the best-fit y-axis intercept of the linear fit being mag.
We obtain the NUV color term measurement using the same orthogonal distance regression method we applied to our ULTRACAM photometry, pairing the nearest HST and LAST observations taken within 60 s, equivalently one exposure time (we binned the LAST light curve by a factor of two from 30-s exposures into 60-s effective exposures). A total of 79 HST epochs are considered, separated on average 17 s from the nearest LAST timestamp. There are no repeat observations within the 79 epochs, which span an observing baseline of 3.8 hr. We detect no significant color term between the NUV and optical transits: . This is corroborated by the residual variance of the best-fit line (2.98) being consistent with that of the zero-slope, horizontal line (2.96). We validated the uncertainty on this result by performing the same MCMC analysis as we did with our ULTRACAM color terms, finding no difference. While there is no significant trend in color with transit depth, the best-fit y-intercept over HST orbits 2-4 indicates a constant color offset of mag, quantifying the relative extinction of NUV starlight during this epoch.
A significant reddening term emerges if we include the LAST photometry overlapping with the fifth HST orbit: . With this addition of 28 epochs, the total baseline spanned lengthens to roughly 5.4 hr. This significant reddening is confidently preferred to the null hypothesis of a horizontal, zero-slope distribution, with respective residual variances of 4.06 and 5.84. At the same time, we raise the caveat that both the NUV and optical transits during the fifth HST orbit are the shallowest and normalization dominated, and thus the most noise dominated. The pointing of HST was reliable for each of the five orbits and without anomalies (§ 2.2). Notwithstanding this potential limitation, it is still plausible that the transits in view during the fifth orbit are drawn from a different distribution of particle sizes that do not extinct NUV light as efficiently relative to the transits of the second, third, and fourth orbits.
While deep, long-period transits have reddened white dwarfs ([13, 133]), similar experiments measured the transits at WD 1145+017 ([1, 159, 28] (Xu et al. 2018)), WD 1054−226 ([50]), and ZTF J1944+4557 ([67]) all to be wavelength-independent in the optical7.
Updated NUV-Optical Model Spectrum
We update the model spectrum of ZTF J0328−1219 published by Vanderbosch et al. [136] through joining our HST/STIS NUV spectrum with a new, nearly concurrent SOAR/Goodman optical spectrum (Figure 5). We imitate their process by using the same

Figure 5. Model spectrum of ZTF J0328−1219 (red) superimposed over our HST/STIS spectrum in the NUV (top, gray) and optical SOAR/Goodman spectrum (middle, gray). We highlight several prominent metal transitions seen as absorption and the Lyman wing in the NUV, along with insetting a magnification of the optical Ca-II H and K transitions. Our SOAR spectrum exhibits absorption at the sodium D doublet ( Å) with no detectable absorption at H. The bottom panel compares the SED photometry to the model from the NUV to the NIR. Red circles correspond to the model fluxes and the photometry is shown as the black error bars without markers. The silver error bars represent the SDSS u and GALEX NUV photometry and are only shown for reference.
Coutu et al. [26] fitting procedure with the Blouin et al. [15] DA model atmospheres. Given flux excursions of are consistently observed toward ZTF J0328−1219 (Figure 1), we suspect the catalog photometry of ZTF J0328−1219 is not reliable for flux calibration. Instead of anchoring our model to the published Pan-STARRS DR2 photometry ( mag; Magnier et al. [95] (2025)), we generate synthetic Pan-STARRS photometry by convolving the Gaia XP spectrum (Carrasco et al. [23]; De Angeli et al. [33]) through the published response curves to calibrate our estimate of the photosphere (see Arseneau et al. [2]). Gaia XP spectra (which span 3300 − 10,500 Å with ) are the average of every BP and RP transit (De Angeli et al. [33]); ZTF J0328−1219 has at least 36 in each channel in Gaia DR3. The resultant synthetic photometry are inherently more uncertain but are likely more representative of the out-of-transit flux baseline for the white dwarf. Unlike Vanderbosch et al. [136], we elect to incorporate the near-infrared and VHS photometry into our fitting process to replace our highly uncertain synthetic -band benchmark. We exclude the unWISE photometry from our fit, despite being consistent with the white dwarf photosphere, since their inclusion worsens the fit to the STIS spectrum and does not significantly affect the derived atmospheric parameters ( K, for example). We report our best-determined atmospheric parameters in Table 2.
| Parameter | Value | Uncertainty |
| (K) | 7310 | 100 |
| (cgs) | 8.14 | 0.03 |
| () | 0.662 | 0.020 |
| () | 0.0107 | 0.0002 |
| [H/He] | −4.20 | 0.20 |
| [C/He] | −8.70 | 0.30 |
| [O/He] | < −5 | |
| [Mg/He] | −8.70 | 0.12 |
| [Si/He] | −9.01 | 0.14 |
| [S/He] | < −5 | |
| [Ca/He] | −9.89 | 0.12 |
| [Cr/He] | −10.31 | 0.27 |
| [Fe/He] | −8.56 | 0.13 |
| [Ni/He] | −9.61 | 0.30 |
| [Mg/Si] | 0.31 | 0.18 |
| (Gyr) | 1.77 | |
| (Gyr) | 3.20 |
Table 2. Model NUV-Optical Spectrum Atmospheric Parameters of ZTF J0328−1219
We detect several transitions from multiple elements, mostly from Fe II and Mg II. The C I line at 1930 Å stands out in particular. We estimate [C/Ca] = 1.19 ±
Note— and are estimated by wdwarfdate (Kiman et al. [86]) using a non-DA model IFMR (Cummings et al. [29]).
0.32, consistent with a chondritic-like body ([C/Ca] = 1.14 for CI chondrites Lodders et al. [93]). The evidence for a chondritic composition is illustrated in Figure 6, where we observe the abundance ratios relative to the Solar metallicity of the accreted debris to be consistent with chondrites to within at all elements except silicon using models from Dufour et al. [41] (see Rogers et al. [121]). Both build-up (BU) and steady state (SS) accretion scenarios can explain the derived abundances ratios; declining phase (DP) accretion is much less likely. While the nickel, magnesium, iron, silicon, and chromium abundance ratios are also consistent with that of bulk Earth, our carbon detection is appreciably discrepant. Convective dredge-up should uplift some carbon to the photosphere of ZTF J0328−1219 (Bédard et al. [10]; Blouin et al. [16]). A K and hydrogen-deficient white dwarf is expected to show [C/He] = −10.4 ± 1.2 due to dredge up in the PG 1159 progenitor (Blouin et al. [14]). Our detection of [C/He] = −8.7 ± 0.3 suggests a significant fraction of the observed carbon is accreted onto ZTF J0328−1219. We do not detect any other volatile species besides carbon, although we estimate upper limits of [O/He] < −5 and [S/He] < −5 from our model.

Figure 6. Photospheric abundance ratios of select elements with respect to calcium derived from our model spectrum of ZTF J0328−1219, all normalized to the Solar metallicity (gray dashed line). We consider build-up (charcoal squares), steady state (blue circles), and declining phase (orange triangles) accretion scenarios. The debris is consistent to within with a composition akin to CI chondrites (yellow dashed–dotted line) and delivered via build-up accretion, except for silicon. Our carbon detection is about 2 dex enriched relative to the expectation for bulk Earth material (green dashed–dotted line).
At [Mg/Si] = 0.31 ± 0.18, our detection appears to be the highest yet observed for a polluted white dwarf with characterized disk compositions, though higher [Mg/Si] values are known in the broader population (Williams et al. [151]). A high [Mg/Si] ratio may indicate disk silicates dominated by olivine (MgSiO) rather than pyroxene (MgSiO) (Rogers et al. [121]), assuming olivine is in the Mg-rich form, forsterite. Although overlapping accretion from compositionally distinct parent bodies can produce a photospheric [Mg/Si] ratio that does not reflect any single parent body, the high value observed here makes this less likely.
The red Lyman- wing dominates the bluest portion of our NUV spectrum, granting us the ability to estimate the photospheric hydrogen-to-helium abundance ratio, and in turn grounding all other abundance ratios we report. Our derived [H/He] = −4.2 ± 0.2 is consistent with what is observed in field DZs (Hollands et al. [73]; Coutu et al. [26]), and can be explained by the convective mixing of a superficial hydrogen layer with the helium envelope underneath. At K, based on Table 3 of Rolland et al. [123] and assuming a 0.6 white dwarf, mixing would have occurred around − 10,000 K, and when the spectral type would have been DA(Z) and the hydrogen layer mass would be of order − . At this current epoch, our [H/He] = −4.2 detection supports (Rolland et al. [123]).
ZTF J0328−1219 has accreted a fraction of calcium ([Ca/He] = −9.89 ± 0.12) roughly 1 dex lower than DZs with similar effective temperature (Dufour et al. [42]; Hollands et al. [73]). Extrapolating from our hydrogen layer mass benchmark , we esti mate g. We cannot confidently rule what amount of the trace hydrogen in the photosphere is accreted. With this much calcium deposited onto the photosphere, ZTF J0328−1219 does not appear to be an extreme outlier like some metal-polluted white dwarfs that appear to have excess hydrogen relative to the calcium they have accreted [57], which is a signpost for the possible accretion of a water-enriched body.
All detected metal lines in our HST/STIS spectrum appear consistent with a photospheric origin at this resolution. Our model extends into the optical in the middle panel of Figure 5, where the Na-I D doublet ( Å) detected in our SOAR/Goodman is notably not reproduced. The high-resolution MIKE spectrum () collected by Vanderbosch et al. [136] captured these sodium transitions as circumstellar gas in absorption, given their distinct velocities relative to the weak H- line they detected (we do not find absorption at H- in our SOAR/Goodman spectrum). While our model appears to account for all detected transitions in the NUV assuming only photospheric absorption, it is possible some of the absorption we observe may be due to unresolved circumstellar gas. If present, unaccounted gas would skew the accuracy of our derived abundances [90].
The atmospheric parameters of our updated joint NUV-optical model spectrum are largely consistent with what was previously estimated by Vanderbosch et al. [136]. Notably, the white dwarf mass decreases by about . Using Gaia photometry alone, Gentile Fusillo et al. [58] estimate assuming a helium-dominated atmosphere. The inclusion of our HST/STIS spectrum tightens our constraint on what the true, out-of-transit, photospheric flux is, compared to the underluminous, transit-dominated Gaia estimate. Our model also agrees with the atmospheric parameters derived by Kilic et al. [83] from their fit of the publicly cataloged SED photometry and the DESI spectrum of ZTF J0328−1219 (, , and ) and with the bespoke DZ modeling done by Ould Rouis et al. [112] (, , and ).
Discussion
Our campaigns of follow-up, time-series photometry and spectroscopy of ZTF J0328−1219 establish a few key results:
hr has remained stable through our 2024 November and 2025 autumn observing campaigns, consistent with the original discovery light curves collected in 2020 December and 2021 February and TESS Sectors 4 and 31 light curves analyzed by Vanderbosch et al. [136].
Our light curves in 2025 October and November show rapidly variable periodicity; the line of sight evolves from being dominated by transits recurring on hr to -dominant. In particular, the prominence and harmonic structure of these periods in the conditional entropy periodogram varied over the roughly two weeks we observed the system, correlating with the heightened morphological complexity we observe in our light curves. Additionally, a new candidate period emerges within October 29-November 1 RVO light curves at hr.
The color of the transits does not vary with transit depth. The color terms we measure using our simultaneous ULTRACAM photometry from 2024 November are consistent with being wavelength independent. Near-concurrent NUV and optical time-series photometry from 2025 October when the white dwarf was most extremely occulted demonstrates a preferential extinction of NUV flux that is constant with transit depth. A statistically significant reddening term emerges during the fifth HST orbit, but arrives with the caveat that these particular data are the most normalization dominated and include the shallowest transits.
Our joint NUV-optical spectrum collected in late 2025 October updates the census of accreted elements in the photosphere, which includes carbon. Our abundance analysis supports that the white dwarf is accreting debris consistent with originating from a chondritic-like parent body.
Collisions Can Explain Variable Periodicity
Never before has a change in the predominant orbital period of a transiting debris system been so clearly observed. Orbital periods have been unreliable for predicting transits, namely at ZTF J0139+5245 [135] and SBSS 1232+563 [72], and periodic transits have vanished at WD 1145+017 [6] and ZTF J1944+4557 [67], eventually resuming at the latter. at ZTF J0328−1219 was previously only known to be transient, appearing exclusively during the first half of TESS Sector 31 [136].
Transience in white dwarf debris disks is most easily observed as photometric variability in the infrared. The collisional cascade in the disk surrounding WD 0145+234 inflated the emitting area enough to brighten the system by a factor of about 2.5 at both 3.4 & 4.6 [147, 130]. The system returned to its pre-collisional cascade state after about 3 yr [128]. Collisions at WD J210034.65+212256.89 generated a significant brightening too, especially in the band, with correlated intensification of the gaseous emission at the calcium triplet in particular [122]. Gas emission correlates with dramatic near-infrared variability across the class of dusty white dwarfs [129, 65, 109, 108], underscoring collisions as a critical production site for observed circumstellar gas.
Minor collisions are likely regularly occurring in the disk around ZTF J0328−1219, given the continuously evolving transit morphologies observed in Figure 1 and Figure 2, and also light curves of the system published in the literature [136, 6, 88]. The circumstellar gas in absorption observed by Vanderbosch et al. [136] may further lend credence to these ongoing collisions. We postulate that transits associated with could have become the predominant pattern along the line of sight if a collisional cascade occurred within this orbit, boosting its occulting fraction enough to overtake that of the debris transiting on the orbit, either through elevating the optical depth or increasing the ring’s scale height.
Collisions are likely necessary to grind down disrupted fragments into the observed dust in white dwarf debris disks [152, 129, 22] (Li et al. 2021). For example, rubble piles orbiting white dwarfs at periods hr cannot be ground down into particles smaller than m without collisions [127]. Fragments along eccentric orbits, like those associated with and , inevitably collide and eventually grind down into sub-micron grains (Kenyon & Bromley 2017a,b; Malamud & Perets 2020; Malamud et al. 2021). While it appears the transit depths in 2025 October were most likely colorless as a function of transit depth, the possible in-transit reddening during the fifth orbit of our HST program commensurate with the transits beginning to recur at could be evidence for recent dust production by collisions within the B ring. However, we are unable to determine whether the material that cast this possible preferential extinction of NUV light is associated with the A or B orbit as the transits superimpose over each other.
Disks are predicted [40, 81 b] and observed [128] to return to pre-cascade states following collisions. has been the predominant if not only period observed within the system during all previous epochs except 2025 autumn. If a collisional cascade occurred within the B orbit, the system should eventually return to being dominated by transits repeating at ; we may have observed this settling as returned as the predominant period during the third and final window of 2025 autumn season (Figure 3). Direct evidence for a collisional cascade would be found in an infrared brightening circa 2025 October, similar to what was observed at WD 0145+234 and WD J210034.65+212256.89.
Instead of collisions there could be a cyclical or periodic process that brings the B ring into and out of view independent of collisions. Gas disks around white dwarfs are expected to general relativistically precess over year-long timescales [104, 60], and the radial velocities of gas disk emission profiles are observed to vary over these timescales [99, 100]—except for the especially rapid variability at HE 1349−2305 [37]—as does the gas disk viewed in absorption around WD 1145+017 [24]. Precession was posited as a possible explanation to the approximately 2 yr-transit of ZTF 1013−0427 [13], another white dwarf with a gas disk in emission, which significantly reddened the star as observed by ZTF in the and bands. The transits at ZTF J1944+4557 appear to oscillate from being in view and out of view every 1.5–2 yr [67]. Long-term, multi-year monitoring will be necessary to rule out the possibility that a cyclical or periodic process is responsible for the emergence of transits repeating at by instead observing these transits arise sporadically, possibly akin to SBSS 1232+563 [72]. Including our observations, transits repeating at have never been observed to persist beyond two weeks, although the time of their emergence for both epochs is unconstrained.
Another possibility may be that various independent orbits destabilized one another. Transit light curves of WD 1145+017 suggest several co-orbital rings of debris at distinct periods [53], a model reproduced by -body simulations [142]. Co-orbital debris rings are subject to dynamical instabilities on timescales of yr if the orbits are not nearly circular ([68]). Since each of , , require eccentric orbits to disrupt an asteroid analog ([118, 141, 139, 136]), it is plausible the erratic, evolving transit morphologies and the alteration of the predominant period may reflect a destabilization of the system. The bottom panel of Figure 3 spotlights the incoherence of the periods in the system over our observing baseline in 2025 autumn, unlike in 2024 November where the transit morphologies of the A period appeared to be more consistent orbit-to-orbit (Figure 1). Previous simulations demonstrated interactions with an external body of mass are capable of perturbing the A and B orbits at ZTF J0328−1219 ([136]).
An Emerging Dichotomy of Transiting Debris Light Curve Morphology
Transit light curve morphologies thus far appear categorizable into two modes: (1) “discrete” and (2) “relentless” or “embedded.” Systems such as ZTF J1944+4557, on the short-period end, and ZTF J0923+4236 and ZTF J0139+5245, at the long-period end, have shown phases of clearly defined, discrete phases of starlight un-occulted by transiting debris. On the other hand, ZTF J0328−1219, WD 1054−226 ([50]), and ZTF J0923+7326 ([13]) are always observed to be in transit. These stars appear embedded behind their disks which continuously block flux as they revolve, and are thus constantly underluminous. It is possible that embedded transiting systems either host disks with the largest vertical heights and/or are the most edge-on among the class; the degree to which they are slightly inclined off of 90 deg should have observable consequences ([12]).
Transit morphologies, irrespective of their discrete or embedded nature, are dynamic. The activity levels of WD 1145+017 have been regularly and densely monitored since 2015, showing phases of heightened and diminished activity along with varying from discrete and embedded morphologies; its transits recently vanished altogether ([6]). ZTF J1944+4557 likewise shows variable transit activity, with transits fully ceasing and returning roughly every 2 years ([67]). When transits are observable at ZTF J1944+4557, only the ingresses and egresses align at the 4.9704 hr orbital period, with the in-transit morphology always changing from one cycle to the next. The periodic patterns of the relentless transits at ZTF J0328−1219 and WD 1054−226 change on a timescale of weeks-to-months ([136, 50]); our observations show variations at ZTF J0328−1219 can be as rapid as one orbital cycle. All of this likely suggests some combination of dynamical processes at play, namely collisions, and possibly even differential fragmentation and streaming or interactions from an external, shepherding body.
Conclusions and Outlook
We collected over 176 hr of follow-up, high-speed, time-series photometry in 2024 November and 2025 autumn that demonstrate ZTF J0328−1219 is still being continuously transited by planetary debris. Transits recurred exclusively at hr during 2024 November, matching the period measured by TESS during Sectors 4 and 31. The previously transient period hr began to dominate the line of sight in 2025 October; by 2025 November was again predominant. Our conditional entropy periodogram detects a weak local minimum at 7.69 hr during the final six days of our 2025 autumn observing season, which we connect to a potentially repeating transit at a new period of hr captured by our RVO light curves from October 29–November 1. Altogether, our light curves from both observing seasons illustrate the stability of and also rapidly variable periodicity and transit morphology in this system, which we attribute to collisions.
We obtained an NUV spectrum with STIS over five consecutive HST orbits in 2025 October. We paired this spectrum with a new optical spectrum also obtained in 2025 October using SOAR/Goodman to update the model spectrum of ZTF J0328−1219. Our NUV spectrum illuminates several metal lines from accreted material, including carbon, whose relative abundance ratios support the accretion of a chondritic object.
We generated an NUV light curve from our time-tagged HST/STIS spectrum to compare against near-simultaneous optical, ground-based photometry, measuring transits constant in color: . This result matches our measurement of likely colorless transits across our simultaneous photometry with ULTRACAM in 2024 November. Only during the fifth and final HST orbit do we measure a significant reddening term . However, the transits during this fifth orbit are the most normalization and noise dominated among the dataset.
Our work like others before it (e.g., [6, 89]) encapsulates the importance of long-term, dedicated monitoring of transiting debris at white dwarfs. We also affirm the power of panchromatic astronomy. Measuring critical dust properties within white dwarf debris disks requires broad-band sampling of the SED (e.g., [49, 122] (Xu et al. 2019b)). These results will be complemented by a forthcoming tandem study analyzing near- and mid-infrared observations of ZTF J0328−1219 collected in 2026 February by JWST DDT Program #12507.
Acknowledgments
We are grateful for the constructive feedback offered by the anonymous reviewer that improved the communication of our analysis and findings. Discussions during the weekly Boston University White Dwarf Group meetings were critical to synthesizing these results and brainstorming how to most effectively communicate them. JAG acknowledges Ciarra Coston and Reuben Dsouza for their roles in these meetings. Colleagues at the 24th European Workshop on White Dwarfs shared similarly beneficial insight. The authors are indebted to all the observatory staff and telescope operators who made the collection of these light curves possible. We particularly thank Colt Pauley (PTO), Beck Maier, Ishara Nisley, Olivia Weiss (LDT), and César Briceño (SOAR).
This material is based upon work supported by the National Aeronautics and Space Administration under Grant No. 80NSSC23K1068 issued through the Science Mission Directorate.
This research is based on observations made with the NASA/ESA Hubble Space Telescope obtained from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 526555. These observations are associated with program GO #17789. The data described here may be obtained from the MAST archive at doi:10.17909/rye3-py98.
This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (Grant agreement No. 101020057).
These results made use of the Lowell Discovery Telescope (LDT) at Lowell Observatory. Lowell is a private, non-profit institution dedicated to astrophysical research and public appreciation of astronomy and operates the LDT in partnership with Boston University, the University of Maryland, the University of Toledo, Northern Arizona University and Yale University.
This research is based on observations made with the Perkins Telescope Observatory, an astronomical observatory located at Anderson Mesa Station, Arizona, and is owned and operated by Boston University.
Based upon observations carried out at the Observatorio Astronómico Nacional on the Sierra San Pedro Mártir (OAN-SPM), Baja California, México. We thank the daytime and night support staff at the OAN-SPM for facilitating and helping obtain our observations.
This work makes use of data from the Warwick 1m telescope operated on the island of La Palma by the University of Warwick at the Observatorio del Roque de los Muchachos of the Instituto de Astrofísica de Canarias.
The contributions of C.-H. B., X.-J. J., H.-J. W., J.-Q. W., and T.-H. Z. were supported by the Urumqi Nanshan Astronomy and Deep Space Exploration Observation and Research Station of Xinjiang (XJYWZZ2023) and the Strategic Priority Research Program of the Chinese Academy of Sciences, Grant No. XDB0550100.
The SOAR/Goodman observations were conducted as part of program 2025B-136460, obtained through the Astronomical Event Observatory Network (AEON), a joint endeavor of the Las Cumbres Observatory and of NSF NOIRLab, which is managed by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the U.S. National Science Foundation.
This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.
J.A.G. attests generative artificial intelligence and large language models were not employed for any component of this analysis nor for the preparation of this manuscript.
Facilities: Gaia, HST (STIS), LDT (LMI), McD:Struve (ProEM), NTT (ULTRACAM), OANSPM:HJТ, Perkins (PRISM), SO:Kuiper, SOAR (Goodman)
Software: Astropy ([3, 4, 5]), astroquery ([59]), ccdproc ([27]), extinction ([9]), GCE ([80]), hypercam ([39]), IRAF ([131]), lmfit ([107]), matplotlib ([75]), nifty-ls ([55]), numpy ([71]), pandas (Wes McKinney 2010; pandas development team 2024), phot2lc ([134]), photutils ([20]), pypeit ([115]), PyRAF ([126]), Pyriod ([11]), scipy ([145]), stistools ([69]), stenv (https://stenv.readthedocs.io/, [86])
Observing Log of High-Speed, Time-Series, Follow-up Photometry
Table A1 contains our observing log for the light curves described in § 2 and shown in Figure 1 and Figure 2. Not included here is our attempt to obtain an -band light curve the night of 2025 October 25, which would have been concurrent with our HST observations, using the Mookodi instrument on the 1-m Lesedi telescope at the South African Astronomical Observatory. Inclement weather prevented us from collecting photometry.
Supplemental 2025 October Light Curve Representations
We supplement Figure 2 by folding all our 2025 autumn light curves on in Figure B.1 and the RVO and HAO light curves from October 29–November 3 on in Figure B.2. We plot our light curves unfolded in Figure B.3 to illustrate transit activity over the course of our campaign.

Figure B.1. Light curves folded on hr. It appears possible that Cycles 14 and 16 and Cycles 19 and 22 show evidence for being influenced by transits from this orbit. All light curves are plotted in the same fashion as that of Figure 2.

Figure B.2. RVO and HAO light curves from 2025 October 29-November 3, the same grouping of light curves used for the third window in the bottom panel of Figure 3, folded on hr. A transit recurs through Cycles 0-9 at phase , despite conditional entropy indicating 7.69 hr is the best period. Our RVO light curves during the nights of November 2 and 3 lacked the phase coverage to resolve the transit further.

Figure B.3. Our entire collection of 2025 October-November follow-up light curves (top) and the maximum total flux excursions computed over 1-hr bins of each light curve (bottom). Our HST observations notably occurred during the point of greatest activity we sampled. All light curves are plotted in the same fashion as that of Figure 2.
2024 November ULTRACAM Light Curve Color Term Analysis
We show our multi-color ULTRACAM light curves collected in 2024 November in Figure C.1, along with the color term distributions we analyze in § 4.1. None of the color terms are definitively statistically significant when compared against the null hypotheses of horizontal, zero-slope distributions.

Figure C.1. ULTRACAM light curves from 2024 November (top panels) and associated color term distributions. Despite the apparently significant color terms from vs. and vs. , the residual variances of the linear ODR fits (salmon-colored lines) are indistinguishable from the null hypothesis of a horizontal line (i.e., colorless transits). See § 4.1.
References
- [1]Alonso, R., Rappaport, S., Deeg, H. J., & Palle, E. 2016, A&A, 589, L6, doi: 10.1051/0004-6361/201628511DOI
- [2]Arseneau, S. M., Hermes, J. J., Camisassa, M. E., Raddi, R., & Bauer, E. B. 2026, ApJ, 1000, 297, doi: 10.3847/1538-4357/ae4c4bDOI
- [3]Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068
- [4]Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f
- [5]Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167, doi: 10.3847/1538-4357/ac7c74DOI
- [6]Aungwerojwit, A., Gänsicke, B. T., Dhillon, V. S., et al. 2024, MNRAS, 530, 117, doi: 10.1093/mnras/stae750DOI
- [7]Bailey, D. 2018, ODR Fitter, v0.1, David C. Bailey. https://www.physics.utoronto.ca/apl/python/ ODR-Fitter Description.pdf
- [8]Ballering, N. P., Levens, C. I., Su, K. Y. L., & Cleeves, L. I. 2022, ApJ, 939, 108, doi: 10.3847/1538-4357/ac9a4aDOI
- [9]Barbary, K. 2016, extinction v0.3.0, Zenodo, doi: 10.5281/zenodo.804967
- [10]Bédard, A., Bergeron, P., & Brassard, P. 2022, ApJ, 930, 8, doi: 10.3847/1538-4357/ac609dDOI
- [11]Bell, K. 2022, Pyriod: Period detection and fitting routines, Astrophysics Source Code Library, record ascl:2207.007. http://ascl.net/2207.007
- [12]Bhattacharjee, S. 2025, PASP, 137, 114204, doi: 10.1088/1538-3873/ae0fe3DOI
- [13]Bhattacharjee, S., Vanderbosch, Z. P., Hollands, M. A., et al. 2025, PASP, 137, 074202, doi: 10.1088/1538-3873/ade0eaDOI
- [14]Blouin, S., Bédard, A., & Tremblay, P.-E. 2023a, MNRAS, 523, 3363, doi: 10.1093/mnras/stad1574DOI
- [15]Blouin, S., Dufour, P., & Allard, N. F. 2018, ApJ, 863, 184, doi: 10.3847/1538-4357/aad4a9DOI
- [16]Blouin, S., Kilic, M., Bédard, A., & Tremblay, P.-E. 2023b, MNRAS, 525, L112, doi: 10.1093/mnras/slad105
- [17]Blouin, S., & Xu, S. 2022, MNRAS, 510, 1059, doi: 10.1093/mnras/stab3446DOI
- [18]Boggs, P. T., & Donaldson, J. R. 1989, Orthogonal distance regression: , National Institute of Standards and Technology, Gaithersburg, MD, doi: https://doi.org/10.6028/NIST.IR.89-4197
- [19]Bonsor, A. 2026, in Encyclopedia of Astrophysics, Volume 1, Vol. 1, 564–572, doi: 10.1016/B978-0-443-21439-4.00020-1
- [20]Bradley, L., Sipócz, B., Robitaille, T., et al. 2024, astropy/photutils: 2.0.2, 2.0.2, Zenodo, doi: 10.5281/zenodo.13989456
- [21]Bradley, L., Sipcz, B. M., Robitaille, T. P., et al. 2026, Photutils, 3.0.0, Zenodo, doi: 10.5281/zenodo.19636730
- [22]Brouwers, M. G., Bonsor, A., & Malamud, U. 2022, MNRAS, 509, 2404, doi: 10.1093/mnras/stab3009DOI
- [23]Carrasco, J. M., Weiler, M., Jordi, C., et al. 2021, A&A, 652, A86, doi: 10.1051/0004-6361/202141249DOI
- [24]Cauley, P. W., Farihi, J., Redfield, S., et al. 2018, ApJL, 852, L22, doi: 10.3847/2041-8213/aaa3d9DOI
- [25]Clemens, J. C., Crain, J. A., & Anderson, R. 2004, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 5492, Ground-based Instrumentation for Astronomy, ed. A. F. M. Moorwood & M. Iye, 331–340, doi: 10.1117/12.550069
- [26]Coutu, S., Dufour, P., Bergeron, P., et al. 2019, ApJ, 885, 74, doi: 10.3847/1538-4357/ab46b9DOI
- [27]Craig, M., Crawford, S., Seifert, M., et al. 2017, astropy/ccdproc: v1.3.0.post1, doi: 10.5281/zenodo.1069648
- [28]Croll, B., Dalba, P. A., Vanderburg, A., et al. 2017, ApJ, 836, 82, doi: 10.3847/1538-4357/836/1/82DOI
- [29]Cummings, J. D., Kalirai, J. S., Tremblay, P.-E., Ramirez-Ruiz, E., & Choi, J. 2018, ApJ, 866, 21, doi: 10.3847/1538-4357/aadfd6DOI
- [30]Cunningham, T., Wheatley, P. J., Tremblay, P.-E., et al. 2022, Nature, 602, 219, doi: 10.1038/s41586-021-04300-wDOI
- [31]Dalessio, J. 2010, in American Astronomical Society Meeting Abstracts, Vol. 215, American Astronomical Society Meeting Abstracts #215, 452.09
- [32]Dashevsky, I., Sahu, K., & Smith, E. 2000, STIS Time-Tag Analysis Guide, Instrument Science Report STIS 2000-02, 28 pages
- [33]De Angeli, F., Weiler, M., Montegriffo, P., et al. 2023, A&A, 674, A2, doi: 10.1051/0004-6361/202243680DOI
- [34]Debes, J., Sankrit, R., Fischer, T., et al. 2024, The Hubble Advanced Spectral Product (HASP) Program, Instrument Science Report COS 2024-01, 31 pages
- [35]Debes, J. H., Hoard, D. W., Wachter, S., Leisawitz, D. T., & Cohen, M. 2011, ApJS, 197, 38, doi: 10.1088/0067-0049/197/2/38DOI
- [36]Debes, J. H., & Sigurdsson, S. 2002, ApJ, 572, 556, doi: 10.1086/340291DOI
- [37]Dennihy, E., Clemens, J. C., Dunlap, B. H., et al. 2018, ApJ, 854, 40, doi: 10.3847/1538-4357/aaa89bDOI
- [38]Dhillon, V. S., Marsh, T. R., Stevenson, M. J., et al. 2007, MNRAS, 378, 825, doi: 10.1111/j.1365-2966.2007.11881.xDOI
- [39]Dhillon, V. S., Bezawada, N., Black, M., et al. 2021, MNRAS, 507, 350, doi: 10.1093/mnras/stab2130DOI
- [40]Dominik, C., & Deeg, H. 2003, ApJ, 598, 626, doi: 10.1086/379169DOI
- [41]Dufour, P., Kilic, M., Fontaine, G., et al. 2012, ApJ, 749, 6, doi: 10.1088/0004-637X/749/1/6DOI
- [42]Dufour, P., Bergeron, P., Liebert, J., et al. 2007, ApJ, 663, 1291, doi: 10.1086/518468DOI
- [43]Eastman, J., Siverd, R., & Gaudi, B. S. 2010, PASP, 122, 935, doi: 10.1086/655938DOI
- [44]El-Badry, K., Conroy, C., Fuller, J., et al. 2022, MNRAS, 517, 4916, doi: 10.1093/mnras/stac2945
- [45]Ely, J., Bourque, M., Debes, J., Kriss, G., & McCullough, P. R. 2015, in IAU General Assembly, Vol. 29, 2258383
- [46]Farihi, J. 2016, NewAR, 71, 9, doi: 10.1016/j.newar.2016.03.001DOI
- [47]Farihi, J., Gänsicke, B. T., & Koester, D. 2013, Science, 342, 218, doi: 10.1126/science.1239447DOI
- [48]Farihi, J., Robert, A., & Walters, N. 2024, MNRAS, 529, L164, doi: 10.1093/mnrasl/slae014DOI
- [49]Farihi, J., Su, K. Y. L., Melis, C., et al. 2025, ApJL, 981, L5, doi: 10.3847/2041-8213/adae88DOI
- [50]Farihi, J., Hermes, J. J., Marsh, T. R., et al. 2022, MNRAS, 511, 1647, doi: 10.1093/mnras/stab3475DOI
- [51]Finkbeiner, D. P., Prince, T. A., & Whitebook, S. E. 2025, PASP, 137, 054504, doi: 10.1088/1538-3873/add20cDOI
- [52]Gänsicke, B. T., Koester, D., Farihi, J., et al. 2012, MNRAS, 424, 333, doi: 10.1111/j.1365-2966.2012.21201.xDOI
- [53]Gänsicke, B. T., Aungwerojwit, A., Marsh, T. R., et al. 2016, ApJL, 818, L7, doi: 10.3847/2041-8205/818/1/L7
- [54]Garrappa, S., Ofek, E. O., Ben-Ami, S., et al. 2025, A&A, 699, A50, doi: 10.1051/0004-6361/202453440DOI
- [55]Garrison, L. H., Foreman-Mackey, D., Shih, Y.-h., & Barnett, A. 2024, nifty-ls: Fast Lomb-Scargle periodogram, Astrophysics Source Code Library, record ascl:2412.008. http://ascl.net/2412.008
- [56]Gary, B. L., & Kaye, T. G. 2024, Research Notes of the American Astronomical Society, 8, 173, doi: 10.3847/2515-5172/ad5d71
- [57]Gentile Fusillo, N. P., Gänsicke, B. T., Farihi, J., et al. 2017, MNRAS, 468, 971, doi: 10.1093/mnras/stx468DOI
- [58]Gentile Fusillo, N. P., Tremblay, P.-E., Cukanovaite, E., et al. 2021, MNRAS, 508, 3877, doi: 10.1093/mnras/stab2672DOI
- [59]Ginsburg, A., Sipőcz, B. M., Brasseur, C. E., et al. 2019, AJ, 157, 98, doi: 10.3847/1538-3881/aafc33DOI
- [60]Goksu, O. A., Kutra, T., & Wu, Y. 2024, AJ, 167, 236, doi: 10.3847/1538-3881/ad3216DOI
- [61]Graham, M. J., Drake, A. J., Djorgovski, S. G., Mahabal, A. A., & Donalek, C. 2013a, MNRAS, 434, 2629, doi: 10.1093/mnras/stt1206DOI
- [64]Guidry, J. A., Gaensicke, B. T., Hermes, J. J., & Vanderbosch, Z. 2024a, NUV observations of transiting exoplanetary debris in a 10-hr orbit around a white dwarf, HST Proposal. Cycle 32, ID. #17789
- [65]Guidry, J. A., Hermes, J. J., De, K., et al. 2024b, ApJ, 972, 126, doi: 10.3847/1538-4357/ad5be7DOI
- [66]Guidry, J. A., Vanderbosch, Z. P., Hermes, J. J., et al. 2021, ApJ, 912, 125, doi: 10.3847/1538-4357/abee68DOI
- [67]Guidry, J. A. 2025, ApJ, 992, 167, doi: 10.3847/1538-4357/adfecbDOI
- [68]Gurri, P., Veras, D., & Gänsicke, B. T. 2017, MNRAS, 464, 321, doi: 10.1093/mnras/stw2293DOI
- [69]Hack, W., Dencheva, N., Sontag, C., Megan, S., & Droettboom, M. 2025, stsci.tools v0.1, STScI. https://stistools.readthedocs.io/en/latest/index.html
- [70]Hallakoun, N., Xu, S., Maoz, D., et al. 2017, MNRAS, 469, 3213, doi: 10.1093/mnras/stx924DOI
- [71]Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2DOI
- [72]Hermes, J. J., Guidry, J. A., Vanderbosch, Z. P., et al. 2025, ApJ, 980, 56, doi: 10.3847/1538-4357/ada5fdDOI
- [73]Hollands, M. A., Koester, D., Alekseev, V., Herbert, E. L., & Gänsicke, B. T. 2017, MNRAS, 467, 4970, doi: 10.1093/mnras/stx250DOI
- [74]Hollands, M. A., Tremblay, P.-E., Gänsicke, B. T., & Koester, D. 2022, MNRAS, 511, 71, doi: 10.1093/mnras/stab3696DOI
- [75]Hunter, J. D. 2007, Computing in Science & Engineering, 9, 90, doi: 10.1109/MCSE.2007.55
- [76]Izquierdo, P., Rodríguez-Gil, P., Gänsicke, B. T., et al. 2018, MNRAS, 481, 703, doi: 10.1093/mnras/sty2315DOI
- [77]Janes, K. A., Clemens, D. P., Hayes-Gehrke, M. N., et al. 2004, in American Astronomical Society Meeting Abstracts, Vol. 204, American Astronomical Society Meeting Abstracts #204, 10.01
- [78]Jura, M. 2003, ApJL, 584, L91, doi: 10.1086/374036DOI
- [79]Jura, M., Farihi, J., Zuckerman, B., & Becklin, E. E. 2007, AJ, 133, 1927, doi: 10.1086/512734DOI
- [80]Katz, M. L., Cooper, O. R., Coughlin, M. W., et al. 2021, MNRAS, 503, 2665, doi: 10.1093/mnras/stab504DOI
- [81]Kenyon, S. J., & Bromley, B. C. 2017a, ApJ, 844, 116, doi: 10.3847/1538-4357/aa7b85DOI
- [83]Kilic, M., Bergeron, P., Blouin, S., et al. 2026, ApJ, 1006, 248, doi: 10.3847/1538-4357/ae825fDOI
- [84]Kim, S.-H., & Martin, P. G. 1995, ApJ, 444, 293, doi: 10.1086/175604
- [85]Kim, S.-H., Martin, P. G., & Hendry, P. D. 1994, ApJ, 422, 164, doi: 10.1086/173714
- [86]Kiman, R., Xu, S., Faherty, J. K., et al. 2022, AJ, 164, 62, doi: 10.3847/1538-3881/ac7788DOI
- [87]Koester, D. B., Gänsicke, B. T., & Farihi, J. 2014, A&A, 566, A34, doi: 10.1051/0004-6361/201423691DOI
- [88]Korth, J., Mustill, A. J., & Parviainen, H. 2026a, Research Notes of the American Astronomical Society, 10, 250, doi: 10.3847/2515-5172/ae9e29DOI
- [89]Korth, J., Mustill, A. J., Parviainen, H., et al. 2026b, A&A, 709, A27, doi: 10.1051/0004-6361/202558102DOI
- [90]Le Bourdais, É., Dufour, P., & Xu, S. 2024, ApJ, 977, 93, doi: 10.3847/1538-4357/ad90b7DOI
- [93]Lodders, K., Bergemann, M., & Palme, H. 2025, SSRv, 221, 23, doi: 10.1007/s11214-025-01146-wDOI
- [94]Lomb, N. R. 1976, Ap&SS, 39, 447, doi: 10.1007/BF00648343DOI
- [95]Magnier, E. A., Schlafly, E. F., Finkbeiner, D. P., et al. 2020, ApJS, 251, 6, doi: 10.3847/1538-4365/abb82aDOI
- [99]Manser, C. J., Gänsicke, B. T., Marsh, T. R., et al. 2016, MNRAS, 455, 4467, doi: 10.1093/mnras/stv2603DOI
- [100]Manser, C. J., Dennihy, E., Gänsicke, B. T., et al. 2021, MNRAS, 508, 5657, doi: 10.1093/mnras/stab2948DOI
- [101]Massey, P., Dunham, E. W., Bida, T. A., et al. 2013, in American Astronomical Society Meeting Abstracts, Vol. 221, American Astronomical Society Meeting Abstracts #221, 345.02
- [102]Mathis, J. S., Rumpl, W., & Nordsieck, K. H. 1977, ApJ, 217, 425, doi: 10.1086/155591DOI
- [103]McDonald, C. H., & Veras, D. 2021, MNRAS, 506, 4031, doi: 10.1093/mnras/stab1906DOI
- [104]Miranda, R., & Rafikov, R. R. 2018, ApJ, 857, 135, doi: 10.3847/1538-4357/aab9a2DOI
- [105]Montgomery, M. H., & O’Donoghue, D. 1999, Delta Scuti Star Newsletter, 13, 28
- [106]Naylor, T. 1998, MNRAS, 296, 339, doi: 10.1046/j.1365-8711.1998.01314.xDOI
- [107]Newville, M., Stensitzki, T., Allen, D. B., et al. 2016, Lmfit: Non-Linear Least-Square Minimization and Curve-Fitting for Python, Astrophysics Source Code Library, record ascl:1606.014. http://ascl.net/1606.014
- [108]Noor, H. T., Farihi, J., & Kenyon, S. J. 2026, arXiv e-prints, arXiv:2609.34868, doi: 10.48550/arXiv.2609.34868DOI
- [109]Noor, H. T., Farihi, J., Kenyon, S. J., et al. 2025, MNRAS, 543, 1602, doi: 10.1093/mnras/staf1380DOI
- [110]Ofek, E. O., Ben-Ami, S., Polishook, D., et al. 2023a, PASP, 135, 065001, doi: 10.1088/1538-3873/acd8f0DOI
- [111]Ofek, E. O., Shvartzvald, Y., Sharon, A., et al. 2023b, PASP, 135, 124502, doi: 10.1088/1538-3873/ad0977DOI
- [112]Ould Rouis, L. B., Hermes, J. J., Bergeron, P., et al. 2026, ApJ, 1011, 236, doi: 10.3847/1538-4357/aeaa35
- [113]Ould Rouis, L. B., Hermes, J. J., Gänsicke, B. T., et al. 2024, ApJ, 976, 156, doi: 10.3847/1538-4357/ad86bbDOI
- [115]Prochaska, J., Hennawi, J., Westfall, K., et al. 2020, The Journal of Open Source Software, 5, 2308, doi: 10.21105/joss.02308
- [116]Provencal, J. L., Montgomery, M. H., Kanaan, A., et al. 2009, ApJ, 693, 564, doi: 10.1088/0004-637X/693/1/564DOI
- [118]Rappaport, S., Sanchis-Ojeda, R., Rogers, L. A., Levine, A., & Winn, J. N. 2013, ApJL, 773, L15, doi: 10.1088/2041-8205/773/1/L15DOI
- [119]Redfield, S., Batalha, N., Benneke, B., et al. 2024, arXiv e-prints, arXiv:2404.02932, doi: 10.48550/arXiv.2404.02932DOI
- [120]Rickman, E., & Brown, J. 2024, in Space Telescope Imaging Spectrograph Data Handbook v8.0, 8
- [121]Rogers, L. K., Bonsor, A., Le Bourdais, É., et al. 2025a, MNRAS, 542, 293, doi: 10.1093/mnras/staf1221DOI
- [122]Rogers, L. K., Manser, C. J., Bonsor, A., et al. 2025b, MNRAS, 537, L72, doi: 10.1093/mnras/slae117
- [123]Rolland, B., Bergeron, P., & Fontaine, G. 2018, ApJ, 857, 56, doi: 10.3847/1538-4357/aab713DOI
- [124]Sandhaus, P. H., Debes, J. H., Ely, J., Hines, D. C., & Bourque, M. 2016, ApJ, 823, 49, doi: 10.3847/0004-637X/823/1/49DOI
- [125]Scargle, J. D. 1982, ApJ, 263, 835, doi: 10.1086/160554DOI
- [126]Science Software Branch at STScI. 2012, PyRAF: Python alternative for IRAF, Astrophysics Source Code Library, record ascl:1207.011. http://ascl.net/1207.011
- [127]Steckloff, J. K., Veras, D., & Volk, K. 2026, ApJL, 1004, L27, doi: 10.3847/2041-8213/ae7571DOI
- [128]Swan, A., Farihi, J., Su, K. Y. L., & Desch, S. J. 2024, MNRAS, 529, L41, doi: 10.1093/mnras/slad198
- [129]Swan, A., Farihi, J., Wilson, T. G., & Parsons, S. G. 2020, MNRAS, 496, 5233, doi: 10.1093/mnras/staa1688DOI
- [130]Swan, A., Kenyon, S. J., Farihi, J., et al. 2021, MNRAS, 506, 432, doi: 10.1093/mnras/stab1738DOI
- [131]Tody, D. 1986, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 627, Instrumentation in astronomy VI, ed. D. L. Crawford, 733, doi: 10.1117/12.968154DOI
- [132]Trierweiler, I. L., Melis, C., Le Bour d ais, É., et al. 2025, ApJ, 993, 53, doi: 10.3847/1538-4357/ae003dDOI
- [133]van Roestel, J., Bhattacharjee, S., Guidry, J. A., et al. 2026, arXiv e-prints, arXiv:2609.08923, doi: 10.48550/arXiv.2609.08923DOI
- [134]Vanderbosch, Z. 2023, zvanderbosch/phot21c: phot21c v1.7.8 release, v1.7.8, Zenodo, doi: 10.5281/zenodo.8169807
- [135]Vanderbosch, Z., Hermes, J. J., Dennihy, E., et al. 2020, ApJ, 897, 171, doi: 10.3847/1538-4357/ab9649DOI
- [136]Vanderbosch, Z. P., Rappaport, S., Guidry, J. A., et al. 2021, ApJ, 917, 41, doi: 10.3847/1538-4357/ac0822DOI
- [137]Vanderburg, A., Johnson, J. A., Rappaport, S., et al. 2015, Nature, 526, 546, doi: 10.1038/nature15527DOI
- [138]VanderPlas, J. T. 2018, ApJS, 236, 16, doi: 10.3847/1538-4365/aab766DOI
- [139]Veras, D., Carter, P. J., Leinhardt, Z. M., & Gänsicke, B. T. 2017, MNRAS, 465, 1008, doi: 10.1093/mnras/stw2748DOI
- [140]Veras, D., & Heng, K. 2020, MNRAS, 496, 2292, doi: 10.1093/mnras/staa1632DOI
- [141]Veras, D., Leinhardt, Z. M., Bonsor, A., & Gänsicke, B. T. 2014, MNRAS, 445, 2244, doi: 10.1093/mnras/stu1871DOI
- [142]Veras, D., Marsh, T. R., & Gänsicke, B. T. 2016, MNRAS, 461, 1413, doi: 10.1093/mnras/stw1324DOI
- [143]Veras, D., Mustill, A. J., & Bonsor, A. 2024, Reviews in Mineralogy and Geochemistry, 90, 141, doi: 10.2138/rmg.2024.90.05DOI
- [144]Veras, D., Steckloff, J. K., & Volk, K. 2026, MNRAS, 549, stag978, doi: 10.1093/mnras/stag978DOI
- [145]Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2DOI
- [146]Wang, S., & Chen, X. 2019, ApJ, 877, 116, doi: 10.3847/1538-4357/ab1c61DOI
- [147]Wang, T.-g., Jiang, N., Ge, J., et al. 2019, ApJL, 886, L5, doi: 10.3847/2041-8213/ab53edDOI
- [149]White, R. J., Greene, T. P., Doppmann, G. W., Covey, K. R., & Hillenbrand, L. A. 2007, in Protostars and Planets V, ed. B. Reipurth, D. Jewitt, & K. Keil, 117, doi: 10.48550/arXiv.astro-ph/0604081DOI
- [150]Williams, J. P., & Cieza, L. A. 2011, ARA&A, 49, 67, doi: 10.1146/annurev-astro-081710-102548DOI
- [151]Williams, J. T., Gänsicke, B. T., Swan, A., et al. 2024, A&A, 691, A352, doi: 10.1051/0004-6361/202450509DOI
- [152]Wyatt, M. C., Clarke, C. J., & Booth, M. 2011, Celestial Mechanics and Dynamical Astronomy, 111, 1, doi: 10.1007/s10569-011-9345-3DOI
- [158]Zazueta, S., Pedrayes, M., Cobos, F., et al. 2000, RMxAA, 36, 141
- [159]Zhou, G., Kedziora-Chudczer, L., Bailey, J., et al. 2016, MNRAS, 463, 4422, doi: 10.1093/mnras/stw2286DOI
- [160]Zuckerman, B., Koester, D., Reid, I. N., & Hünsch, M. 2003, ApJ, 596, 477, doi: 10.1086/377492DOI
- [161]Zuckerman, B., Melis, C., Klein, B., Koester, D., & Jura, M. 2010, ApJ, 722, 725, doi: 10.1088/0004-637X/722/1/725DOI
