A Systematic Search for New δ Scuti and γ Doradus Stars Using TESS Data

Abstract

Focusing on the discovery of new $\delta$ Scuti and $\gamma$ Doradus stars, we analyzed the Transiting Exoplanet Survey Satellite (TESS) light curves for 193,940 A-F stars selected from four legacy catalogs—the Henry Draper Catalogue (HD), the Smithsonian Astrophysical Observatory (SAO) Star Catalog, the Positions and Proper Motions Catalog (PPM), and the Bonner Durchmusterung (BD, including its extensions). Through visual inspection of light curve morphologies and periodograms, combined with evaluation of stellar parameters, we identified over 51,850 previously unreported variable stars. These include 15,380 $\delta$ Scuti, 18,560 $\gamma$ Doradus, 28 RR Lyrae stars, 260 heartbeat candidates, and 2645 eclipsing binaries, along with thousands of other variable types. Notably, over 4145 variables exhibit hybrid $\delta$ Scuti-$\gamma$ Doradus pulsations, and more than 380 eclipsing binaries feature pulsating primary components. This study reveals a substantial population of bright, previously undetected variables, providing a valuable resource for ensemble asteroseismology, binary evolution studies, and Galactic structure research. Our results also highlight the surprising richness in variability still hidden within well-known stellar catalogs and the continued importance of high-precision, time-domain surveys such as TESS.

1. Introduction

Variable stars of spectral types A and F (hereafter “A-F” stars) are of significant astrophysical interest. Most A-F stars lie on the main-sequence (MS) evolutionary stage, and the detection and classification of pulsating A-F stars are crucial for advancing our understanding of stellar evolution and Galactic structure. These stars occupy key regions in the Hertzsprung–Russell (H–R) diagram, representing multiple classes of pulsating variables. A high percentage of A-F stars show periodic or quasi-periodic brightness variations, predominantly due to stellar rotation—a common phenomenon among MS stars, as well as pulsations.

The region of A-F variables on the H–R diagram, including MS, pre-MS, and post-MS stars with masses between 1.2 and 2.5 $M_{\odot }$, hosts $\delta$ Scuti ($\delta$ Sct, DSCT), $\gamma$ Doradus ($\gamma$ Dor, GDOR) pulsators, and hybrid groups. This is the location where energy transfer via radiation and convection can be observed  [1]. A central open question in seismic studies of A-F stars concerns the excitation and mode selection mechanism of pressure (p) and gravity (g) modes [2]. Hybrid pulsators, offering new insights into the physics of stellar oscillations, may be common among A-F stars [3,4]. From a pulsational point of view, $\delta$ Sct and $\gamma$ Dor stars are clearly distinguished, as they pulsate in two different modes—conventionally the opacity $\kappa$-mechanism and the convective blocking mechanism, respectively. Earlier studies show that the location of $\gamma$ Dor and $\delta$ Sct classes in the $T_{\mathrm{eff}}$–$logg$ diagram has been extended: Kepler $\delta$ Sct stars exist beyond the red edge of the observational instability strip; Kepler $\gamma$ Dor pulsations appear in both hotter and cooler stars than observed before; and Kepler hybrid stars occupy the entire region between the blue edge of the $\delta$ Sct instability strip and the red edge of the $\gamma$ Dor instability strip, and even beyond [2]. Moreover, a recent study of pulsating A-F stars in the Kepler field [5] concluded that $\gamma$ Dor variables do not occupy a separate instability strip but lie entirely within the $\delta$ Sct instability region. The frequencies in $\gamma$ Dor stars may therefore be an effect of mode selection rather than a consequence of distinct driving and damping mechanisms. Balona [5] thus proposed that $\gamma$ Dor stars should not be considered a separate class of variable, since no two classes of pulsating star are known to share the same instability region. On the theoretical side, Xiong et al. [4] pointed out that both $\delta$ Sct and $\gamma$ Dor stars may be regarded as a single class of pulsating variable, with their pulsations driven by a combination of the $\kappa$-mechanism and the coupling between convection and oscillations. Most pulsators within the $\delta$ Sct -$\gamma$ Dor instability strips are expected to be hybrids, exhibiting both pressure and gravity modes simultaneously. To some extent, current pulsation theories need to be revised to allow the driving of both p- and g-modes in A-F type stars across a broad temperature range.

Asteroseismology probes the internal structures of stars by comparing their identified sequences of observed pulsation frequencies with pulsation modes calculated from theoretical models. Traditionally, the identification of pulsation modes has relied on high-resolution spectroscopy or multi-color time series. However, information on the pulsation modes can also be extracted directly from frequency-spacing1 patterns  [6], although stellar rotation may distort regular frequency and period patterns [7,8]. In A-F stars, oscillation modes do not always produce evident frequency patterns in their spectra [2]. This is mainly due to the limited precision and time baselines of ground-based photometry, as well as the typically low amplitude of the variations.

The detection of brightness variations in stars benefits greatly from high-precision photometry. Recent advances in space photometry have brought about major progress in asteroseismology of A-F variables. The Transiting ETESSxoplanet Survey Satellite (TESS, Ricker et al. [9]) offers an unprecedented opportunity to examine stellar light variations with high precision. TESS typically achieves a photometric precision of 3% to 60 parts per million (ppm) for stars brighter than TESS magnitude ${16}_{·}^{\mathrm{m}}0$  [10], or 200–300 ppm for stars of ${10}_{·}^{\mathrm{m}}0$ in the form of Combined Differential Photometric Precision (CDPP) measured for all 2 min cadence targets light curves at 1 h integration time. Detailed photometric precisions at different magnitudes refer to Figure 1 of Twicken et al. [11] and TESS Instrument Handbook [12]. The CDPP can be converted (by a factor of $\sqrt{60/c}$, c is cadence in seconds) to per-point precisions of 1090–1650 and 280–430 ppm, for 2 min and 30 min cadences, respectively. Simply, our practices show ∼20 ppm in the Fourier amplitude spectrum, and the light curve scatter is generally ≤0.005 mag for stars brighter than ${10}_{·}^{\mathrm{m}}5$2.

Numerous discoveries based on TESS data have been reported (e.g., [13,14,15,16,17,18,19]). For example, of the 14,042 $\delta$ Scuti stars identified by Balona [17], 9091 and 706 have HD and BD identifiers, respectively—indicating that many bright variable stars remain undiscovered. This may result from CCD saturation in bright-star surveys or limited sky coverage from ground-based observatories. For instance, surveys like All Sky Automated Survey [for Supernovae] (ASAS, ASAS-SN) designed to monitor the sky for transient phenomena [20,21], or the Zwicky Transient Facility (ZTF, [22,23,24]), may overlook bright stars due to such limitations.

Recently, 960 regular period-spacing patterns were identified in 611 Kepler $\gamma$ Dor stars, providing valuable probes of stellar interior [25]. In that study, periods, period spacings, and the gradients of the period spacings were used to guide mode identifications and to estimate the rotation rate of the near-core region via the Traditional Approximation of Rotation (TAR), since the slope of the spacing pattern constrains rotation [8,26]. Many of these stars are hotter and show longer period-spacing patterns than predicted by theory, and they also excite more radial orders than expected. This discrepancy points to simplified physics in the models, such as assumptions about mixing-length parameter, chemical abundances, angular-momentum transport, differential rotation, magnetism, and the limits of TAR in rapid rotators. In addition, the observed fraction of slow rotators is larger than predicted by angular-momentum transport models [27], and the near-core regions of $\gamma$ Dor stars rotate more slowly than theory suggests [27,28]. Results from a large sample of 2085 A-F stars observed over four years by Kepler [25] further indicate that most stars in the $\gamma$ Dor instability strip do not display period-spacing patterns. Using TESS data, Garcia et al. [29] recently reported a catalog of 140 gravity-mode period-spacing patterns detected in 106 $\gamma$ Dor stars and two slowly pulsating B-type (SPB) stars.

For $\delta$ Sct stars, one of the main breakthroughs has been the detection of regular frequency spacings that resemble the large separations known from solar-like oscillators, which are excited stochastically by convection. In particular, Bedding et al. [30] reported very regular sequences of high-frequency pulsation modes in 60 intermediate-mass MS stars (57 from TESS and 3 from Kepler, among a sample of 1330 $\delta$ Sct variables), with a measurable large separation $\Delta \nu$ that scales with mean density. They focused on identifying $\delta$ Sct stars that pulsate at high frequencies (above 30 ${\mathrm{d}}^{-1}$ ). The $\delta$ Sct stars with regular frequency spacings tend to lie near the zero-age main-sequence (ZAMS), and are therefore relatively young, with masses between 1.5 and 1.8 $M_{\odot }$. Stars showing regular patterns can serve as benchmarks for mode identification in the much larger population of $\delta$ Sct stars whose pulsation spectra are less regular. Period-spacing patterns enable definitive mode identification by allowing estimates of mean stellar density through the scaling of large separations and by constraining the radial orders of p-modes, which is crucial for mode identification. They have also opened the way to grid-based modeling of stellar mass, radius, and evolutionary stage, in close analogy to solar-like stars. Thus, period-spacing patterns have opened the way for ensemble asteroseismology of pulsating stars. Searching for new $\delta$ Sct and $\gamma$ Dor in TESS data is therefore of great importance. TESS observations are expected to reveal many more $\delta$ Sct stars with high-frequency overtones. For the potential usefulness of these stars, we have intentionally marked the newly identified $\delta$ Sct variables with pulsation frequencies above 30 ${\mathrm{d}}^{-1}$ as “DSCT-U” in our catalog.

At the same time, challenges remain: mode selection is still poorly understood, nonlinear effects complicate the oscillation spectra, and theoretical models must reconcile the observed regularities with the expected complex mode interactions in $\delta$ Sct stars. The rapidly growing number of $\delta$ Sct stars showing regular patterns among their pulsation frequencies necessitates improved modeling tools to interpret the observations. Building on these discoveries, modeling frameworks have advanced. For example, a recent study constructed $2\times {10}^5$ models of young $\delta$ Sct stars to interpret the observed $\Delta \nu$ patterns [31]. These authors found that $\Delta \nu$ for MS $\delta$ Sct stars deviates from the solar scaling relation by about 13%, and that the lowest radial order is often poorly reproduced.

In addition, one of the intriguing classes of A-F type variables first discovered in Kepler data is the heartbeat stars (HBSs) (KOI-54, [32]). These are close binary star systems in highly eccentric orbits, with pulsations driven by tidal forces or tidally excited oscillations (TEOs) [17,32,33,34,35,36,37]. Kepler and TESS have revealed many systems showing TEOs (at harmonics of the orbital frequency), making them powerful probes of tidal dissipation and stellar interiors. Kirk et al. [38] reported 173 HBSs from Kepler, while 991 HBSs are identified in OGLE data [39], and more than 300 have since been discovered using TESS observations [40,41,42,43], including several dozen from this project. Using Kepler photometry and Keck spectroscopy,  Guo et al. [44] revealed $\delta$ Sct -$\gamma$ Dor hybrid pulsations in the evolved A-type eccentric eclipsing binary system KIC 4142768, where some of the g modes are exact orbital harmonics and likely tidally excited. HD 74423 was the first ellipsoidal variable (ELL) discovered with tidally trapped pulsations in a close binary system [45] based on TESS data. More recently,  Jennings et al. [46] used TESS light curves to identify tidal effects perturbing several p- and g-modes among 133 significant pulsation frequencies detected in the double-lined detached eclipsing binary KIC 9851944, in which two F-type $\delta$ Sct and $\gamma$ Dor subgiants exhibit pressure- and gravity-mode pulsations, respectively.

TESS was designed to monitor bright stars, targeting nearby and luminous stars in search of transiting exoplanets. As a result, it has evolved into an all-sky variability survey with optimal photometric performance for stars with TESS magnitudes ($T_{\mathrm{mag}}$) between 7 and 13.5 mag. This range aligns well with many historical star catalogs based on photographic plates, which typically include stars brighter than $V\sim {12}_{·}^{\mathrm{m}}0$. The variability in such stars is well-suited to follow-up spectroscopy using medium-sized ground-based telescopes. Another reason for selecting target stars from these catalogs is that bright stars are generally less affected by common photometric issues in TESS data—such as blending or contamination from nearby sources—than fainter stars.

Accordingly, we initiated a project to search for previously unrecognized bright variable stars of A-F spectral types in four legacy catalogs HD, BD, SAO, and PPM utilizing public TESS [9] and Gaia [47] data. Below is a brief overview of these sampling catalogs:

• The Henry Draper Catalogue (HD): Originally published in 1918, HD provides spectroscopic classifications for initial 225,300 stars. Together with the Henry Draper Extension (1925–1936, 46,850 stars) and the Henry Draper Extension Charts (1937–1949, 86,933 stars), the HD series includes 359,083 stars in total (see [48]).
• The Bonner Durchmusterung (BD, [49]) catalog was published between 1852 and 1859. BD covers 325,037 stars in the northern sky and portions of the south (from BD+89 38 through BD−01 4530). BD was extended by the Southern Durchmusterung (SD, Schoenfeld 1886; 134,834 stars, declinations −1° and −23° degrees) and further supplemented by the Córdoba Durchmusterung (CD, Thome [50,51], 613,959 stars, −22 to −90°). The Cape Photographic Durchmusterung (CPD, ∼450,000 stars) also covers the southern sky (−18 to −90) which was appended to BD. However, CPD was not sampled in this work.
• The Smithsonian Astrophysical Observatory (SAO) Star Catalog: Includes positions and proper motions for 258,997 stars (mostly complete to V = $9_{·}^{\mathrm{m}}0$), plus 4503 stars fainter than V = ${10}_{·}^{\mathrm{m}}0$ [52].
• The PPM (Positions and Proper Motions) Catalog: A more comprehensive successor to SAO, PPM contains positions, proper motions, spectral types, and photometry for 468,586 stars. It is nearly complete to V = $9_{·}^{\mathrm{m}}5$, with 102,672 stars fainter than V = ${10}_{·}^{\mathrm{m}}0$ and 22,395 fainter than V = ${11}_{·}^{\mathrm{m}}0$. [53,54,55,56,57].

The four catalogs collectively contain a total of 2,160,496 entries (summarized in

Table 1. Objects in the legacy catalogs and scheduled screening phases (I–V).

Catalog	Objects	Notes
PPM	468,586	I–III
HD	359,083	IV
SAO	258,997	IV
BD	325,037	V
SD	134,834	V
CD	613,959	V
CPD	450,000	not sampled
Total	2,160,496	∼609,009 unique

), including numerous duplicate objects across the catalogs. According to VizieR [58], 441,601 BD+CD stars, 220 582 HD stars, 257 075 SAO stars are cross-matched with PPM numbers. Accounting for these repetitions, the four catalogs actually have approximately 609 009 stars. Astronomers preferentially use the HD designation as the main identifier for an astronomical object3.

This survey project—briefly outlined in Zhou [59,60]—uses high-precision TESS [9] and Gaia [47] data to systematically screen A-F stars selected from the aforementioned four legacy catalogs, aiming to identifying previously unknown bright pulsating variables. The primary objective is to discover several interesting types of pulsating A-F variable stars with a focus on $\delta$ Scuti, $\gamma$ Doradus, located within the lower instability strip of the Hertzsprung–Russell diagram, along with RR Lyrae (RR Lyr) stars and eclipsing binary systems (EBs). The project proceeded through five phases (Table 1):

• Phases I and II: Scanned 160,670 objects from the PPM catalog.
• Phase III: Added 308,192 stars, completing the scan of all PPM entries (468,862 objects in all the four parts of the PPM catalog) [53,54,55,57].
• Phase IV: Extended the search to PPM-unmatched stars in the HD and SAO catalogs [59,60,61,62].
• Phase V: Included A-F stars from BD, SD, and CD catalogs [63].

In total, TESS light curves of approximately 193,940 A-F type stars (31.8%) with effective temperature ($T_{\mathrm{eff}}$) between 6100 and 9200 K selected from 2,160,496 catalog entries in the PPM, HD, SAO, BD (plus SD and CD) catalogs are analyzed. A substantial number of more than 51,800 new variables were identified. Preliminary results from each phase were released instantly in the Research Notes of the American Astronomical Society (RNAAS) [59,60,61,62,63]. The purpose of promptly releasing these newly classified variable stars is to facilitate community awareness, engagement, and follow-up. By making this catalog publicly available, other researchers can identify and select interesting targets for further investigation. At the same time, this helps ensure that these already identified stars are properly acknowledged in future variable star searches, reducing the likelihood of redundant efforts or the erroneous reporting of known variables as new discoveries.

Dozens of discoveries have since been independently confirmed and published, such as 28 new heartbeat stars reported by Li et al. [42,43] and a new eclipsing binary with a $\gamma$ Dor primary presented by (TIC 140736015, [64]). A separate catalog of 42 new heartbeat stars was also compiled by [65]. This paper presents the final results and conclusions of the project. Section 2 describes the datasets used, Section 3 details the classification methodology, and Section 4 summarizes the results along with discussion and concluding remarks.

2. The Data and Custom Processing

This study is based on data from two main sources: TESS and Gaia. The primary dataset analyzed consists of brightness measurements obtained by the TESS space telescope. We initially used the TESS Input Catalog (TIC, v8.2, [66]) for cross-identifying stars and retrieving preliminary stellar parameters. Cross-identifications were performed using Simbad and MAST (Mikulski Archive for Space Telescopes) catalogs, matching stars with TIC v8.2 [66] and Gaia DR3 identifiers [67]. Most objects were successfully matched within a 6″ radius of the source coordinates. Since the TIC itself is not directly cross-identified with the legacy catalog designations, we relied on Simbad for most cross-identifications. Accordingly, cross-matching was carried out by querying Simbad for both TIC and GDR3 identifiers of each source. Nevertheless, some legacy catalog stars cannot be found in Simbad and matched to TIC or GDR3 directly. In particular, when matching stars from the BD and CD catalogs with the TIC v8.2 catalog, a significant number of objects could not be identified, likely due to the limited astrometry precision of the older catalogs. In such cases, TIC was queried directly using the source’s coordinates: the nearest TIC source within 6″ (nearly a third of TESS pixel scale) and brighter than ${13}_{·}^{\mathrm{m}}5$ (the best TESS photometry limit magnitude) was selected, and the corresponding TYC identifiers from TIC v8.2 were adopted as source identifiers. However, CD catalog designations were retained whenever they served as the primary identifiers in Simbad. The resulting matched sample of stars was then subjected to further analysis, including the extraction of stellar parameters (e.g., effective temperature, luminosity), checking TESS data availability, visual inspection of light curves and periodograms, and classification of variability types. In total, TESS light curves of approximately 193,940 A-F type stars were analyzed.

2.1. TESS Data

The TESS archive hosted at the MAST Portal4 was accessed programmatically to check data availability for each star in the candidate sample. We downloaded the data products generated by NASA’s Science Processing Operations Centre (SPOC), including both light curve files (*-s_lc.fits) and Target Pixel Files (TPF, *-s_tp.fits). The light curve files were processed using the astropy and lightkurve Python packages (versions 4.1 and 2.0.11) [68,69], following guidelines provided in the TESS Archive Manual5.

Light curves often exhibit various forms of variability, including instrumental systematics, intrinsic stellar variability such as pulsations, and external flux changes caused by transiting exoplanets or eclipsing binaries. Any variability not originating from the star itself must be carefully removed prior to scientific analysis. Two types of light curves are extracted by the SPOC pipeline: (1) Simple Aperture Photometry (SAP) flux, which sums the calibrated pixel values within the TESS optimal photometric aperture, and (2) Pre-search Data Conditioned Simple Aperture Photometry (PDCSAP) flux is the SAP flux with long-term trends removed using Co-trending Basis Vectors (CBVs), and is nominally corrected for instrumental effects and scattered light. PDCSAP flux typically presents a cleaner dataset than SAP flux and is less prone to systematic trends. In some cases, however, the SAP flux may be mistakenly interpreted as intrinsic stellar variability due to uncorrected instrumental effects (refers to Figure 1).

As a result, PDCSAP flux is commonly used in final analyses without further detrending (e.g., Dumusque et al. [70], Astudillo-Defru et al. [71], Demory et al. [72], Battley et al. [73]). However, PDCSAP flux has known shortcomings in preserving astrophysical variability when de-trending systematics, as it may inadvertently suppress or eliminate genuine astrophysical signals, such as long-term variations or transient bursts, as discussed by Hill et al. [74], Littlefield et al. [75], and may even introduce spurious variability, as shown in Figure 2.

To address these concerns, some authors (e.g., Hon et al. [14], Prša et al. [15], von Essen et al. [76], Steindl et al. [77], Southworth and Van Reeth [78]) prefer working directly with SAP flux, applying customized processing and detrending techniques adapted to their specific scientific goals. Similarly, in this work, we selectively use SAP flux in cases where instrumental systematics are negligible or absent, allowing for reliable analysis (see Figure 3).

In addition to SPOC light curves, we also make use of TESS High Level Science Products (HLSP) generated by the Quick-Look Pipeline (QLP, Huang et al. [79]) developed by the TESS Science Office at MIT, and light curves produced by the TESS Asteroseismic Science Operations Center (TASOC). The QLP processes data from TESS full-frame images (FFIs), extracting light curves via multi-aperture photometry for all stars brighter than TESS magnitude ${13}_{·}^{\mathrm{m}}5$. The HLSP-QLP light curves contain uncorrected SAP flux. Both QLP and TASOC light curves may exhibit residual spacecraft systematics, which can impact the classification of variability types (see Figure 1). Special care was taken to avoid misclassifying instrumental artifacts as genuine stellar variability. In our experience, periodograms based on detrended data suggest that some low-amplitude $\delta$ Sct and $\gamma$ Dor pulsations may be masked by these systematics.

2.2. Gaia DR2/DR3 and TIC v8.2 Data

This study utilizes data from two major releases of the Gaia mission [47]: Gaia Data Release 2 (GDR2,  Gaia Collaboration et al. [80]) and Gaia Data Release 3 (GDR3, Gaia Collaboration et al. [67]). Stellar parameters such as effective temperature, luminosity, mass, and surface gravity are initially adopted from TIC v8.2 [66] and subsequently updated whenever corresponding values from GDR2 or GDR3 are available. Parallax, distance, Gaia photometric magnitudes ($G,BP,RP$), and color indices ($BP-RP$, $G-RP$) are extracted from GDR3. Since GDR3 does not provide stellar luminosities or radii, these two parameters are obtained from GDR2. The absolute magnitude in the Gaia G-band is computed directly using standard relation without correcting for interstellar extinction on a per-star basis: $M_G=G+5.0-5log\left(d\right)$, where G is the apparent G-band magnitude and d is the distance in parsecs. As reported in [81], average extinction in the Gaia G-band is $A_G\sim$0.46 mag, which can be used for uniform correction by $M_G=G-A_G+5.0-5log\left(d\right)$. We chose not to apply such corrections because (i) extinction values are highly uncertain and non-uniform across the sky, and (ii) the majority of our targets are relatively nearby A-F type stars, for which extinction is typically small at distances below a few hundred parsecs. We therefore expect our results—particularly the classification of variable types and their distribution in the H–R diagram—to be only minimally affected by neglecting extinction.

Spectral types reported in source catalogs are systematically reviewed and re-determined using updated parameters from TIC v8.2 and Gaia catalogs. It is important to note that effective temperatures provided in GDR2 and GDR3 can differ significantly; in all such cases, we adopt the values from the more recent GDR3. Discrepancies among TIC, GDR2, and GDR3 values are acknowledged and carefully considered during data validation.

In short, effective temperature ($T_{\mathrm{eff}}$) in GDR2 were derived from broad-band colors, which are more prone to degeneracies, while GDR3 estimates use the full Gaia $BP/RP$ spectra with Bayesian inference—generally more reliable, though still model-dependent. TIC prioritizes external spectroscopy measurements, and otherwise adopts photometric color–temperature relations with extinction corrections. Discrepancies in $T_{\mathrm{eff}}$ for the same star across these catalogs are common and expected, as the values depend on methodology. Variations arise from the choice of technique (e.g., photometric calibrations vs. spectral fitting), stellar atmosphere models, data sources (e.g., photometric filters, spectral resolution), and the treatment of systematics such as interstellar extinction. Consequently, $T_{\mathrm{eff}}$ values may legitimately disagree among the three sources for some stars.

In the TIC, $T_{\mathrm{eff}}$ values are derived independently, prioritizing spectroscopic measurements from surveys such as APOGEE and LAMOST when available, and otherwise relying on photometric relations from broad-band photometry (e.g., $UBVRI$, 2MASS $JHK$) with empirical color–temperature calibrations and extinction corrections. Since TIC v8.0 [82] incorporated Gaia DR2 data [80], the parameters in TIC v8.2 [66] generally agree with those from GDR2 for most stars.

Gaia-derived effective temperatures are known to carry substantial systematic uncertainties. Gaia DR2 and DR3 can give noticeably different $T_{\mathrm{eff}}$ values for the same star, reflecting the different methods used in each release.

Gaia DR2 used data from three broad photometric bands (G, and the integrated Gaia prism spectra $BP$ and $RP$ of 1.38 billion sources) to infer stellar effective temperatures for some 161 million sources brighter than G = 17 mag with $T_{\mathrm{eff}}$ in the range of 3000–10,000 K. No explicit extinction treatment was applied, leading to strong color–temperature–reddening degeneracies and catalog-dependent systematics. Typical accuracies are of order 324 K in $T_{\mathrm{eff}}$ and 15% in luminosity [81].  Balona [5] noted a typical uncertainty of $\approx 260\mathrm{K}$, when classifying A-F type pulsating stars using GDR2 data, while Balona et al. [83] adopted much larger uncertainty (up to ∼1000 K) for early-type B stars. These examples highlight that photometric survey-based estimates (Gaia DR2/DR3, TIC, etc.) can differ substantially depending on methodology and stellar type. They also underscore the limitations in accuracy and the resulting imprecision in defining the instability strip boundaries, reinforcing the expectation of several-hundred-Kelvin uncertainties in practice.  Balona [5] further emphasized that such uncertainties hinder accurate placement of A-F type pulsators within the instability strip, effectively limiting the reliability of detailed population studies.

Astrophysical parameters are a major component of Gaia DR3, produced by the Astrophysical Parameters Inference System (Apsis, [84]). By contrast, GDR3 employed a more advanced methodology that incorporates $BP/RP$ spectra, Radial Velocity Spectrometer (RVS) data, and explicit extinction treatment. In particular, GDR3 re-computed stellar parameters using the more sophisticated GSP-Phot pipeline (the General Stellar Parametriser from Photometry, [85]) which fits the low-resolution $BP/RP$ spectra together with photometry and parallax. For bright stars, spectroscopic parameters are also provided by the GSP-Spec module (the General Stellar Parametriser from spectroscopy, [86]). Extinction is explicitly modeled in both approaches. The GSP-Phot results for 471 million sources show typical deviations of only 110 K in $T_{\mathrm{eff}}$ relative to literature values, with improved accuracy for stars within 2 kpc, while an additional 5.6 million stars have derived from GSP-Spec.

It is therefore clear why effective temperatures homogeneously derived in Gaia DR2/DR3 can differ significantly from those in the TIC, which compiled values from heterogeneous photometric and spectroscopic calibrations. Consequently, discrepancies in stellar parameter persist between TIC v8.2, Gaia DR2, and DR3, particularly for $T_{\mathrm{eff}}$ and luminosity (see examples in Table 3 of Zhou [87]). These systematic differences and their underlying uncertainties remain unresolved and are beyond the scope of this work. For further discussion of $T_{\mathrm{eff}}$ accuracy, see Balona and Ozuyar [88]. Nonetheless, Gaia DR3 provides the most extensive homogeneous database of astrophysical parameters to date, derived solely from Gaia data [84]. Accordingly, Gaia DR3 values are preferred whenever available.

2.3. Data Reduction

Two primary tasks were carried out during the data screening process:

(1)

Light Curve Retrieval and Inspection: For each candidate star, available light curves were retrieved from the MAST archive. One TESS sector was selected for each object—typically the most recent sector with the shortest cadence data (Sectors 1–71). Priority was given to light curves produced by the SPOC, TESS-SPOC, QLP, and TASOC pipelines. Various diagnostic plots were generated, either interactively displayed or saved as PDF files for later visual inspection. These included: SAP and PDCSAP light curves, binned light curves (with 10 min and 30 min bins), periodograms with estimated noise-levels and signal-to-noise ratios (SNRs) of detected peaks. Noise was estimated as the mean amplitude within the Fourier spectrum. To better assess frequency-dependent behavior, noise and SNRs were calculated separately in two frequency ranges: 0–2 ${\mathrm{d}}^{-1}$ and 2–60 ${\mathrm{d}}^{-1}$ . In the 2–60 ${\mathrm{d}}^{-1}$ range, the SNR of 4.5 is taken for significant peaks (see Figure 4). This SNR matches the significance threshold of 4.7 times the mean noise level in the periodogram reported by Bell et al. [89], who analyzed TESS data with the same Lomb–Scargle implementation in Lightkurve [69] which builds on Astropy [68] to compute periodograms as used here. In the low-frequency range (0–2 ${\mathrm{d}}^{-1}$ ), an initial SNR threshold of 5.5 was adopted to reduce false detections caused by red noise. Baran and Koen [90] suggested that space-based photometry may require substantially higher SNR values than ground-based data in order to avoid spurious detections, since they found that SNR thresholds between 4.6 and 5.7 are needed to reach a false alarm probability (FAP) of 0.1%, depending on cadence and coverage. However, we later found this criterion to be too conservative for the high-precision TESS data: a threshold of SNR = 4.5 proved sufficiently robust, especially for TESS light curves. Charpinet et al. [91] similarly showed that a SNR of 4.4 is well above the $4\sigma$ threshold—four times the median noise level (a widely adopted value first proposed for ground-based multisite campaigns by Breger et al. [92]), while the probability of a true detection exceeds 99.99% for SNR ≳ 5.1. As shown in Figure 1 of Bognár et al. [93], an SNR = 4.54 corresponds to a detection limit at 0.1% FAP (or 99.9% reality). A detection threshold of SNR = 4.5, defined as 4.5 times the median noise level, is also adopted by Baran et al. [94,95] for analyzing TESS data regardless of data coverage.

(2)

Assessment of Stellar Parameters and H–R Diagram Position: Fundamental stellar parameters were obtained from the CDS databases and further supplemented with values from the TESS Input Catalog  (TIC v8.2, [66]) and Gaia DR3 [47,67]. Each star’s position on the H–R diagram was evaluated to help constrain its variability classification.

The overall data reduction pipeline included the following steps: (1) Querying and retrieving data from Simbad, the International Variable Star Index (VSX-AAVSO, [96]), TIC v8.2, TESS (at MAST), and Gaia; (2) Light curves preprocessing: detrending, outlier removal, binning, flattening, and plotting; (3) Periodogram analysis to identify significant frequencies, estimate noise levels, and compute signal-to-noise ratios (as illustrated in Figure 4); (4) Counting significant peaks and performing preliminary variability classification based on light curve features and frequency content.

3. Methodology and Identification

The survey proceeded in three main steps after selecting A-F stars from legacy catalogs. First, known variable stars were excluded from the sample. Next, data reduction, analysis (Section 2.3), and classification were carried out. Finally, potential blending or contamination was carefully examined for stars exhibiting complex or ambiguous variability.

3.1. Exclusion of Known Variables

In principle, previously known variable stars, as well as stars outside the A-F spectral range, were excluded. Only stars with effective temperatures between 10,000 and 6000 K—generally corresponding to spectral types A and F—and not classified as variables in Simbad or the AAVSO-VSX [96], were selected for inspection using TESS light curves. The survey was carried out non-interactively using a Python program. Prior to light curve retrieval, each candidate was first checked locally against an extensive collection of published variable star catalogs to ensure exclusion of previously identified objects. These reference catalogs include, but are not limited to: 70,552 DSCT and 8080 GDOR stars compiled by Zhou [97]; Gaia DR3 Part.4 Variability catalog (9,976,881 objects; Gaia Collaboration et al. [67]); Gaia DR2 variability results (363,969 records, [80]); OGLE-IV variables [98], 378,861 variables in the ASAS-SN Catalog of Variable Stars X [20]; ZTF variables [24], 123,841+ TESS variables [17,18]; and variable stars archived in VSX-AAVSO database, including: 6294 Cepheid variables (CEP), 144,760 RR Lyr stars, 271,251 rotating stars (ROT), 26,308 rotating ellipsoidal variables (ELL), 13,652 YSO (Young Stellar Objects, pre-MS stars), 983,461 EBs. As the project progressed over 2.5 years, newly published catalogs reporting $\delta$ Sct and $\gamma$ Dor stars [19,99,100,101,102] were monitored and incorporated into the exclusion process. Methodologies were consistent with those described in Section 2 of Zhou [103]. In particular, candidate stars from [19] were further assessed, and some were reclassified following in-depth analysis. Each star was also individually checked online against Simbad and the AAVSO VSX for its most recent variability classification status to ensure up-to-date validation.

3.2. Classification

The classification process involved visual inspection of the morphology of light curves and the distribution of peaks in periodograms, combined with the evaluation of each star’s astrophysical parameters. Specifically, variability types were determined based on the following four criteria:

(1)

Light Curve Morphology—Visual assessment of the shape and structure of the light curves.

(2)

Periodogram Analysis—Fourier transforms were applied to single-sector TESS data to resolve the periodic components of each star’s brightness variations.

(3)

Stellar Atmospheric and Astrophysical Parameters—Effective temperature, spectral type, luminosity, surface gravity, and mass for each star were first retrieved from TIC [66] and then updated using the GDR2 [80] and GDR3 [67] catalogs during data reduction. These parameters were subsequently considered to evaluate each star’s position on the H–R diagram, especially in the context of the known classes of pulsating variables (refers to Figure 3 of [104]).

(4)

Contamination and Blending Check—Applied particularly to stars exhibiting complex or ambiguous variability features, such as eclipses overlapping with pulsations, hybrid pulsators, or RR Lyr stars situated in crowded fields or globular clusters.

To manage ambiguous cases and minimize false identifications, the following classification conventions were adopted:

• Classifications follow the definitions and notations used in GCVS, or Simbad’s object types;
• The presence of multiple variability types or hybrid pulsation is indicated with a plus sign (“+”), e.g., EA+GDOR, DSCT+GDOR;
• If more than one classification is possible, multiple types are listed using a forward slash (“/”);
• $\gamma$ Dor stars are gravity-mode pulsators with $T_{\mathrm{eff}}$ below 8500 K, a threshold used to separate $\gamma$ Dor stars from hotter gravity-mode pulsators, which include Maia and slowly pulsating B stars following Aerts et al. [105] and pulsation frequencies lower than 5 ${\mathrm{d}}^{-1}$ (mainly in 0.3–3 ${\mathrm{d}}^{-1}$ ) following Balona et al. [106].
• Maia variables: Stars exhibiting DSCT, GDOR, or hybrid pulsation signatures with effective temperatures above 9200 K are classified as Maia variables; This group encompasses hot and anomalous $\gamma$ Dor stars, hot hybrid $\delta$ Sct-$\gamma$ Dor stars, and cool B-type stars with high-frequency pulsations. This classification is consistent with the observed characteristics of Maia stars as hotter extensions of $\delta$ Sct stars suggested by Balona and Ozuyar [88], Balona [107].
• Rotating Stars: The label “ROT” denotes general rotating variable stars, including those dominated by rotational modulation or starspots-induced variability (refers to Figure 4 of [104]). This label encompasses subtypes such as ELL, ${\alpha }^2$ Canum Venaticorum variables (ACVs), Chemically Peculiar stars (CP), Magnetic Chemically Peculiar stars (MCP), Rapidly Oscillating Ap stars (roAp), and solar-like oscillators. In most cases, detailed subtype classification is not pursued. Particularly, ROT group may merge non-eclipsing RS Canum Venaticorum variables (RS CVn) featured by traveling spots-induced variations or similar to rotational modulations6.
• Candidate Classification: Tentative variability types are marked with a question mark (“?”).
• Unclassified Variables: When a confident classification cannot be made, the general term “variable” is used.
• Uncertain and Non-Variable Stars: If no convincing variability is detected from the selected single-sector TESS light curves, the star is marked as “uncertain”. Stars that show neither significant periodic variations nor prominent peaks in the periodogram are labeled “constant”. Both groups are ultimately excluded from the final catalog of new variable stars.

The identification methodology, briefly outlined in earlier research notes [59,60], has been consistently applied and progressively refined over the course of this two-and-a-half-year project, spanning five phases. During the final preparation of this manuscript, a comprehensive re-evaluation was conducted for stars with ambiguous or unusual features, revisiting their light curves and periodograms. This process benefited from the author’s cumulative experience gained through the visual inspection of over half a million stars. As a result, hundreds of classifications were reassigned or corrected, further enhancing the reliability and accuracy of the final catalog.

3.3. Blending and Contamination Issue

Due to the TESS telescope’s low spatial resolution and relatively large CCD pixel size (21″ per pixel on sky), blending and contamination from nearby sources pose significant challenges in TESS photometry [103]. To account for this drawback, a weighted Photometric Purity Index (PPI) was constructed by combining the CROWDSAP and FLFRCSAP parameters available in TESS-SPOC light curve files:

$$\mathrm{PPI}={\mathrm{CROWDSAP}}^{0.6}\times {\mathrm{FLFRCSAP}}^{0.4}·$$

Here, CROWDSAP represents the ratio of the target flux to the total flux within the optimal photometric aperture, while FLFRCSAP measures the fraction of total target flux captured within that aperture. This index prioritizes a balance between minimizing contamination and maximizing captured signal, with respective weights of 0.6 and 0.4. A PPI value greater than 0.9 indicates well-isolated photometry with minimal contamination [62].

Following the discussion in Zhou [103], a dedicated Python tool was developed to analyze TESS Target Pixel File and determine whether the observed variability originates from the intended target or from contaminating neighboring stars within the photometric aperture. This tool proved particularly valuable in cases involving hybrid pulsators or eclipsing systems with overlapping intrinsic pulsations, where source blending could lead to misclassification. A demonstration is provided in Figure 5, with six additional representative examples shown in Figure 6, Figure 7, Figure 8, Figure 9, Figure 10 and Figure 11. Methodological details are further illustrated in Zhou [87]. These figures might have been further improved had we been aware of the identification charts tool tpfi by Xing et al. [108] earlier. Our independent approach shares some similarities with theirs and will be revised for a better presentation of contamination checks.

3.4. Classification Examples

Below we present seven classification examples to illustrate our methodology and implementation.

3.4.1. AG+05 807: Blended with a Known ACV

AG+05 807 (also known as TIC 234838664, spectral type A5) was initially classified as a ROT/SXARI candidate based on its high effective temperature ($T_{\mathrm{eff}}$= 9693 K from GDR3). However, the light curve shape and presence of multiple harmonics resembled that of a fundamental mode RR Lyr (RRab) star. Despite this similarity, the light curve profile was nearly symmetrical—unlike the asymmetrical steep rise and slow decline characteristic of RRab variables. Upon re-examination for blending effects, the observed variability was reclassified as originating from an ACV variable. The star is severely blended with TIC 234838865 (HD 46105, spectral type A1p C), a known magnetic chemically peculiar ACV-type star. The two stars have similar R-band magnitudes of 8$8_{·}^{\mathrm{m}}98$ and $8_{·}^{\mathrm{m}}16$, respectively.

Figure 5 shows the Digitized Sky Survey (DSS) image of the star field along with known variables from Simbad. The overlaid 3×3 grid simulates the approximate TESS photometric aperture, with each box scaled to 21″ × 21″ on sky. The $T_{\mathrm{eff}}$ value from GDR3 likely corresponds to this ACV star, while TIC 234838664 better matches its assigned A5 spectral type. TESS observed these stars in Sectors 6, 33, and 87, with cadences of 1800, 600, and 200 s, respectively. Fourier analysis of the Sector 87 data revealed a dominant rotating frequency of $f_0$= 1.260737 ${\mathrm{d}}^{-1}$, along with its harmonics at $2f_0,3f_0$, and $4f_0$. The corresponding rotational period of 0.793186 days is consistent with the value of 0.79326 days reported by Hümmerich et al. [109]. The light curves from Sectors 33 and 87 are shown in Figure 5, where the phase-folded light curves exhibit clear rotational modulation. The classification is further confirmed by the consistent variability pattern observed in multiple TESS sectors for both stars. Notably, HD 46105 is a double star system (HD 46105A,B; see Hümmerich et al. [109]).

3.4.2. HD 55823: A $\delta$ Sct in an Eclipsing Binary

The target HD 55823 (also known as TIC 59488906) is not surrounded by any known variable stars within a 2.5′ radius, according to the Simbad database. Although the TESS photometry aperture could enclose four nearby objects (see Figure 6), their contribution to the total flux is minimal. Parameters from the TESS Target Pixel File (TPF) indicate CROWDSAP = 0.9890 and FLFRCSAP = 0.9304, yielding a weighted Photometric Purity Index (PPI) of 0.9651. This suggests negligible blending or contamination. In Figure 6, the TPF image (typically 11×11 pixels, $\approx 3·{85}^{\prime }$ across) shows the photometric aperture masks (striped red squares) used in the pipeline extraction. The accompanying DSS image ($5^{\prime }\times 5^{\prime }$ field of view) displays a stimulated aperture of comparable size, overlaid with positions of nearby known stars to assess potential blending or contamination. Consequently, this object is identified as a new eclipsing binary system with a $\delta$ Sct-type pulsating primary component.

3.4.3. HD 201218: A $\delta$ Sct in an Eclipsing Binary

The target HD 201218 (also known as TIC 283648911) is not surrounded by any known variable stars within a 2.5′ radius, according to the Simbad database. Although the TESS photometry aperture enclosed several nearby objects (see Figure 7), their contribution to the total flux is minimal—TPF indicates CRODWSAP = 0.9453 and FLFRCSAP = 0.9499, and a corresponding PPI of 0.9472—suggesting negligible blending or contamination. Consequently, this object is identified as a new eclipsing binary system with a $\delta$ Sct-type pulsating primary component.

3.4.4. HD 290580: A $\gamma$ Dor in an Eclipsing Binary

The field around HD 290580 (also known as TIC 50789543) is relatively sparse, with no nearby contaminating stars or known variables within the vicinity, according to the Simbad database (see Figure 8). The TESS photometric aperture encompasses only the target itself. Furthermore, the TPF file reports parameters CRODWSAP = 0.9890 and FLFRCSAP = 0.8977, resulting in a PPI of 0.9514—indicating minimal blending or contamination. Therefore, the observed $\gamma$ Dor-type pulsations superimposed with eclipses are intrinsic to the source. The object is newly identified as an eclipsing binary system with a $\gamma$ Dor primary component.

3.4.5. BD+49 2033: A $\gamma$ Dor in an Eclipsing Binary

The field around BD+49 2033 (also known as TIC 287298034) is clean, with no nearby known variable stars according to the Simbad database (see Figure 9). The TESS photometric aperture exclusively encloses the target itself. In addition, the TPF file provides parameters CRODWSAP = 0.9961 and FLFRCSAP = 0.9449, resulting in a PPI of 0.9753—indicating negligible blending or contamination. BD+49 2033 was previously listed simply as “variable” in both Simbad and VSX. The observed $\gamma$ Dor-type pulsations superimposed with eclipses are thus intrinsic, confirming that this star is a newly identified eclipsing binary system with a $\gamma$ Dor primary component.

3.4.6. BD+47 3958: A $\gamma$ Dor in an Eclipsing Binary

The target BD+47 3958 (also known as TIC 67309744) is mildly affected by nearby objects, although no known variables are present in the surrounding field according to the Simbad database (see Figure 10). The TESS photometric aperture enclosed several neighboring fainter stars, which may contribute to the observed light curve. The TPF file provides parameters CRODWSAP = 0.9515 and FLFRCSAP = 0.9314, yielding a PPI of 0.9434. This value indicates minimal blending or contamination. Therefore, the observed $\gamma$ Dor pulsations superimposed with eclipses are intrinsic to the object, confirming it as a newly identified eclipsing binary system with a $\gamma$ Dor primary component. The light curves of both BD+47 3958 and HD 290580 (in Section 3.4.4) are similar to that of V2077 Cyg, a known eclipsing binary with $\gamma$ Dor pulsating component (last panel in Figure 1 [110]).

3.4.7. HD 40656: A Rotating Star with Eclipses

HD 40656 (also known as TIC 282502866, $T_{\mathrm{eff}}$= 6063 K) is newly classified as an EA+ROT system—an eclipsing binary with a rotating primary component. A blending check confirmed that the observed eclipsing variability originates from the target itself. Figure 11 presents the DSS image with the simulated TESS aperture overlaid, along with nearby stars known in Simbad.

3.5. Classification Accuracy and Reliability

Since the classification of variable stars in this work is performed visually in the traditional manner—based on light curve morphology, periodogram structure, and astrophysical parameters such as effective temperature, luminosity, and surface gravity—it is essential to assess the reliability and potential limitations of this approach. Although no formal machine learning or statistical classification model is employed, the methodology has been consistently applied across the entire dataset, with accumulated experience playing a key role in decision-making. Continuous refinements to the Python script and improvements in data processing have further enhanced classification accuracy and confidence.

Visual classification based on a single-sector light curve is generally straightforward for certain types, including eclipsing binaries (e.g., EA–Algol-type and EB–$\beta$ Lyrae-type), ELL, RRab stars, typical $\gamma$ Dor stars with distinctive asymmetric light curves (see top three panels in Figure 12; Figure 1 in Balona et al. [106]; Figure 1 in Zhou [60,61]; and additional examples in Figures 2 and 7–9 of Zhou [111]), as well as some rotational variables with prominent spots-induced modulations. When combined with Fourier spectra and effective temperature, multi-periodic $\delta$ Sct and $\gamma$ Dor stars and their hybrids can be reliably identified (as in the cases shown in Figure 4 and Figure 13). However, certain rotational variables exhibit low-frequency variability in their light curves that closely resemble that of $\gamma$ Dor stars, which can lead to confusion and misclassification. These similarities make $\gamma$ Dor stars more difficult to distinguish from other variables also exhibiting low-frequency behavior. See comments in Section 4.3.2.

In fact, to evaluate the credibility and accuracy of this visual classification method, the author first reviewed several dozens of $\gamma$ Dor stars selected from the earlier groups in history [112,113,114,115,116] with TESS light curves; then, the beginning 10,000 sample stars were deliberately not filtered to exclude known variable stars. This allowed direct comparison between the author’s classifications and those reported in the literature. The results show a match in over 95% of cases, demonstrating a high level of reliability. Most discrepancies occurred in stars whose revised stellar parameters (from TIC v8.2 and Gaia DR3) placed them outside the typical $\delta$ Sct or $\gamma$ Dor instability domains, despite originating from A-F type samples in the legacy catalogs.

For $\delta$ Sct stars, classification accuracy exceeds 98%, with only rare cases of ambiguity, typically due to contamination from nearby sources or misclassified W Ursae Majoris-type eclipsing binary (EW). For $\gamma$ Dor stars, the accuracy is estimated to be approximately 95%. The slightly lower figure arises from potential confusion with other low-frequency variables, especially ROT, or rarely SPB/Maia at higher $T_{\mathrm{eff}}$. If cases like those in Figure 8 of Balona et al. [106] are interpreted as rotational modulation from migrating starspots, the effective accuracy for identifying $\gamma$ Dor stars could drop to around 90%. Rotation-induced variability is further discussed in Section 4.3.2.

Ambiguities often arise when different types of variable stars exhibit similar light curve morphologies. In particular, $\gamma$ Dor stars, non-eclipsing RS CVn-type binaries, rotational variables or solar-like oscillators, EW-type eclipsing binaries, and first overtone RR Lyr (RRc) stars can all display low-frequency, similar sinusoidal light curve patterns, complicating classification. On the other hand, for instance, some stars showing $\gamma$ Dor-like variability but with unusually high temperature or luminosity are more likely to be misclassified and could belong to other variability types (e.g., $\alpha$ Cygni-type (ACYG), $\gamma$ Cassiopeiae-type (GCAS)), or subclasses of rotational variables (such as SX Arietis-type, ACV). ACYG are pulsating luminous supergiant variables, spectral type ranged from late B to early F, but mostly A-type. GCAS are eruptive B-type variables, a subset of classical emission-line B stars (Be variables, B0e–B9e). SX Arietis-type (SXARI) are rotating MS stars with strong magnetic fields and chemical peculiarities (B0p – B9p-type). Effective temperature thus becomes a key discriminant, especially for separating $\gamma$ Dor, $\delta$ Sct, and Maia candidates, as well as identifying SXARI and ACV. While luminosity is the key discriminant in classifying ACYG and GCAS along with $T_{\mathrm{eff}}$. In our practice, only very few stars are classified as ACYG, GCAS, or SXARI.

RRc and EW binaries also pose a challenge due to their similarly shaped light curves with comparable timescales. Harmonic structures in the light curves can further blur distinctions—especially between ACV stars and RRab variables (e.g., the case in Figure 5). Similarly, RRc and ROT can occasionally be misclassified for common presence of harmonics. When classification is uncertain or outside the scope of $\delta$ Sct and $\gamma$ Dor pulsators, we assign placeholder labels such as ‘ROT’ or ‘variable’ to maintain focus on the primary pulsating star population.

To summarize, typical misclassification can occur in a couple of cases and the classification accuracy is affected by several factors listed below:

• Data Issue: A classification may be biased due to presence of residual instrumental systematics in the light curves from HLSP-QLP and TASOC, or wrongly de-trended PDCSAP (e.g., Figure 2).
• Blending and Contamination: Stars may be affected by unresolved contamination from nearby sources, which can distort the observed variability pattern. The blending and contamination metrics—CROWDSAP and FLFRCSAP—are provided only for light curves produced by the TESS-SPOC pipeline. Light curves from other sources (e.g., HLSP-QLP) do not include these parameters and therefore cannot be evaluated for blending issue during data processing.
• Parameter Uncertainty: Inaccuracies in stellar parameters (e.g., effective temperature and luminosity) can shift a star’s position on the H–R diagram and influence type assignment, particularly among Maia, $\delta$ Sct, and $\gamma$ Dor.
• Potential similar light curves between ACV and RRab, ROT and GDOR, ROT and RS CVn, RRc and EW, RRc and ROT, RRc and DSCT, RRc and GDOR, EB and EW, etc.
• Ambiguities arose in classifying stars with $T_{\mathrm{eff}}$ > 9000 K, particularly among SPB, Be (B-type stars with emission lines), ACYG, GCAS, SXARI, ACV, where the classifications remain marginal and subject to high uncertainty).
• A few ambiguities likely occurred among EB, ROT, and RS CVn.
• Misinterpretation of rotational modulation or spots-induced variability as pulsations.

While this method relies on visual inspection and human pattern recognition, future improvements could include the development of quantitative confidence metrics, implementation of multithreaded data downloading and processing to improve speed, and, where feasible, semi-automated classification support using machine learning or template matching techniques in combination with human oversight.

4. Results and Discussions

4.1. Examples of Classifications

We present several illustrative examples of newly identified variable stars, including $\delta$ Sct, $\gamma$ Dor, RR Lyr stars, ellipsoidal variables, rotating stars, eclipsing binaries, and heartbeat stars. Figure 12 illustrates 15 representative cases that highlight the characteristic light curve morphology of each variability type. In addition, Figure 14 provides seven more examples, each accompanied by a periodogram to support the classification. Finally, Figure 15 features a selection of unusual cases, which may be of particular interest for future follow-up studies.

4.2. Summary

This project systematically screened four major legacy catalogs—PPM, HD, SAO, and BD (+SD+CD)—to identify new variable stars using high-precision photometric data from TESS. Among the 2,157,772 entries from the combined catalogs, 609,009 unique objects were identified after cross-matching for duplicates. A subset of 193,940 A-F stars (31.8% of the unique objects) was selected for detailed variability analysis. As a result, we have successfully identified over 51,850 new variable stars including: over 15,380 new $\delta$ Sct stars; more than 18,560 new $\gamma$ Dor stars; approximately 4145 hybrid $\delta$ Sct-$\gamma$ Dor pulsators; 2645 eclipsing binary systems, including more than 380 candidates with pulsating components; 28 new RR Lyrae stars; 260 heartbeat candidates; 16,058 rotating variables and solar-like oscillators (with 1148 ellipsoidal variables); along with several other types. An overview of the classifications and counts is provided in

Table 2. Outcomes of variability identifications of AF stars in the legacy catalogs.

Type	New	Reclassified *	Notes
$\delta$ Sct	15,380	1723
$\gamma$ Dor	18,560	1080
($\delta$ Sct +$\gamma$ Dor )	4145	364 †	subset
Maia	2879	289
RRLyr	28	1
EA/EB/EW	2645	298
(EA+ $\delta$ Sct /$\gamma$ Dor )	370	–	candidates
(EA+Maia)	39	–	candidates
(EA+ELL)	3	–	candidates
(EA+ROT)	150	–	candidates
Heartbeat stars	260	10	candidates
ROT	16,058	802
(ACV)	1346	–	candidates
(ELL)	1148	–
(ELL+$\delta$ Sct )	18	–	candidates
(SXARI)	122	–	candidates
GCAS	151	7	candidates
ACYG	53	10	candidates
flare/Emline	520	6	candidates ‡
unclassified	1306	12

^* Reclassified column are subsets of New column, these are reclassification of stars previously assigned in literature as Variable, PulsV (Simbad); msosc, short_ts, solar_rm (Gaia); VAR/MISC (VSX); etc. See text for details. ^† Values for indent items in parentheses are subsets of the preceding totals. ^‡ Solar system object (SSO) encounter events may be included; see [108].

. A comprehensive catalog of all identified variables is available online via Zenodo, DOI 10.5281/zenodo.10636928.

The present new variables included the reclassification of previously reported variable stars. Specifically, the following cases were re-evaluated: (1) Sources previously designated as “msosc” (main-sequence oscillators) or “short_ts” (short timescale variables) in the Gaia database. Gaia “msosc” objects are incorporated into Simbad with the object type “PulsV” (pulsating variable stars), but both “msosc/PulsV” and “short_ts” categories include a number of non-oscillating stars. These objects have been reclassified into more specific types such as $\delta$ Sct, $\gamma$ Dor, EB, ROT, or other variable classes, based on their TESS light curves. Gaia’s “solar_rm” (solar-like with rotational modulation) objects are excluded from analysis. (2) Suspected variables listed as “VAR” or “MISC” in the VSX database. (3) Unclassified “TESS variables” identified by Fetherolf et al. [18], as well as candidates labeled as “zero variability” by Gootkin et al. [19]. The latter study computed periodograms only within the 5–24 ${\mathrm{d}}^{-1}$ frequency range, thereby excluding $\gamma$ Dor stars, which typically show variability at lower frequencies. All such cases are annotated in Table 2 and included in the final catalog with updated classifications.

4.3. Discussions

4.3.1. H–R and CMD Diagrams

Using the up-to-date $\delta$ Sct and $\gamma$ Dor catalogs compiled by Zhou [97,117], we compared the spatial distribution of the newly discovered $\delta$ Sct and $\gamma$ Dor stars with that of previously known populations in Figure 16. The results show that most of the new variables are concentrated in the Galactic disk, with a distribution consistent with that expected for young and intermediate-age populations.

In addition, the H–R diagram and color-magnitude diagram (CMD) for the newly classified variables further confirm that the $\delta$ Sct and $\gamma$ Dor stars lie within their classical instability strips, which significantly overlap (shown in Figure 17) and primarily on their main-sequence evolutionary stages (see Figure 18). The Gaia field stars shown in the figure are reproduced from Figure 8 of Wang et al. [118], which itself is a version of the local Gaia H–R diagram shown in Figure 6c of Gaia Collaboration et al. [119] (for solar neighborhood within about 100 pc).

This survey demonstrates that significant numbers of variable stars remain undiscovered even within well-known legacy catalogs of bright stars. The combination of precise TESS photometry and the target selection from historical catalogs proves highly effective for expanding the census of stellar variability in the galaxy.

4.3.2. About Rotational Modulation and Classifying $\gamma$ Dor  Stars

While visual inspection remains a powerful tool for identifying characteristic variability patterns, it is inherently subjective and may be affected by noise, instrumental effects, blending, or similarities in the light curves morphology. We continue the discussion from Section 3.5 to address potential issues in the present classifications.

$\delta$ Sct stars are relatively easy to recognize, as their periodograms exhibit high-frequency peaks (typically 5–50 ${\mathrm{d}}^{-1}$). However, certain variable types, such as $\gamma$ Dor and rotational variables, may display similar morphologies, leading to possible misclassifications. Some $\gamma$ Dor-like stars show photometric properties more consistent with spotted rotational variables rather than genuine pulsators. In such cases, the periodogram alone is often insufficient to decisively distinguish pulsation from rotation.

Many stars cooler than about 7000 K host starspots, leading to rotational modulation of their light curves. This modulation is typically interpreted as brightness variation due to surface temperature changes caused by large-scale magnetic fields. The issue of confusion between $\gamma$ Dor pulsation and rotational modulation has been discussed in the literature (e.g., [25,120]). Rotational modulation can mimic the light curves of $\gamma$ Dor stars, and their variability was initially attributed to rotation. However, some light curves are too complex for this explanation, and in most cases, distinguishing pulsation from rotation remains challenging (Section 3, [106,121]). Simultaneous occurrence of rotational and pulsational frequencies is not uncommon in space-based photometry, although confirmed examples are still rare (e.g., [120]).

For these reasons, the $\gamma$ Dor stars are among the most difficult to classify, as their light curves often closely resemble those of spotted rotating stars. Balona et al. [121] emphasized the difficulty distinguishing $\gamma$ Dor from rotational variables. We follow Balona’s suggestion that "if harmonics of the principle frequency are present, then the variation is more likely to be rotational modulation” (Section 3, [121]), since harmonics are more typically associated with rotating starspots than with the low-amplitude, nonlinear pulsations of $\gamma$ Dor stars.

Balona further noted that $\gamma$ Dor classification is more reliable in two cases: (1) stars with clearly asymmetric light curves (e.g., Figure 8, Figure 9 and Figure 10, and the top three panels in Figure 12), and (2) stars whose multiple-frequency spectra are too complex to be explained by rotational modulation alone (e.g., Figure 13). He also advised that stars with only one dominant low frequency are probably spotted stars. “It is very probable that some of the single-frequency ROT stars are actually $\gamma$ Dor stars”. The real challenge in distinguishing $\gamma$ Dor from ROT lies in cases with a single frequency.

Accordingly, stars showing spots-traveling signatures and obvious starspots activity (as seen in Figures 8 and 9 of Balona [122]) are classified as “ROT”. Those with light curves resembling Figure 6 of Balona [122] (beating patterns), or Figure 1 (asymmetric light curves), Figure 2 (symmetric light curves), Figure 3 (multiple modes) of Balona et al. [106] are classified as $\gamma$ Dor. For stars exhibiting a single low-frequency peak, their light curves are evaluated to determine whether they are rotational variables or $\gamma$ Dor pulsators.

Certainly, visual classification is subject to bias, influenced by the classifier’s prior knowledge, experience, and pattern recognition ability. This may lead to a tendency to assign ambiguous cases to familiar or preferred categories, and inconsistent criteria may be applied across different cases. These limitations underscore the need for machine learning or artificial intelligence-based methods in modern variable star classification. However, such automated approaches are not immune to error; indeed, numerous misclassifications produced by machine learning algorithms were encountered during the course of this survey. A hybrid approach—combining automated classification with human verification—may offer a more robust and effective solution (e.g., [123]).

5. Conclusions

This study represents one of the most comprehensive systematic searches for $\delta$ Sct, $\gamma$ Dor, and other A-F type variable stars in four major legacy catalogs (HD, SAO, PPM, and BD, including CD and SD extensions). Leveraging the high-precision photometry of TESS, we analyzed 193,940 stars and identified over 51,850 previously unrecognized or unclassified variables, including pulsators ($\delta$ Sct, $\gamma$ Dor, RR Lyr), eclipsing binaries, heartbeat stars, and rotational variables.

Our discovery of 33,940 $\delta$ Sct and $\gamma$ Dor stars—including some 4145 hybrid pulsators—significantly expands the known population of short-period pulsating stars. Together with the identification of 2645 eclipsing binaries, including more than 380 candidates with pulsating components, these results provide rich opportunities for ensemble asteroseismology [124,125,126,127] and for testing stellar evolution models through combined photometric and dynamical analysis.

The spatial distribution of these newly identified variables is consistent with the expected populations of young to intermediate-age stars in the Galactic disk, as supported by their positions in the Hertzsprung–Russell diagram and Gaia CMD. The discovery of over 51,800 previously unknown variable stars from the legacy catalogs demonstrates the power of combining TESS data with historical stellar archives to reveal hidden variability.

By focusing on bright, historically cataloged stars with established identifiers, this work fills a critical gap in the census of nearby variable stars. The resulting sample represents a rich population of prime targets suited for spectroscopic follow-up and further investigation. These findings advance our understanding of stellar interiors, binarity, and the interplay between pulsation and orbital dynamics.

This survey yields the largest homogeneous catalog of $\delta$ Sct, $\gamma$ Dor, and hybrid pulsators selected from legacy catalogs, offering a solid foundation for future studies of stellar interiors, binary evolution, and Galactic structure. It also underscores that even well-explored stellar catalogs still contain numerous undiscovered variables, emphasizing the importance of continued high-precision time-domain surveys.

Key highlights of the work include:

• Discovery of 51,850 new variable stars, including 15,380 $\delta$ Sct, 18,560 $\gamma$ Dor, and 4145 hybrid $\delta$ Sct–$\gamma$ Dor pulsators, 260 heartbeat candidates, 28 RR Lyr stars, and over 16,000 newly identified rotational variables, along with several dozen other variable stars.
• Identification of 2645 eclipsing binaries, including 370 candidates with pulsating primaries of $\delta$ Sct or $\gamma$ Dor, offering a critical sample to study pulsation–binary tidal interactions.
• Spatial and Gaia CMD analysis confirms that these new variables predominantly trace young to intermediate-age stellar populations in the Galactic disk.
• Demonstration that TESS data, combined with legacy catalogs, is an efficient method for revealing overlooked variables—even at bright magnitudes.
• A valuable catalog for follow-up spectroscopy and time-domain studies with both ground- and space-based platforms.
• The largest homogeneous catalog of bright $\delta$ Sct/$\gamma$ Dor stars to date, enabling ensemble asteroseismic investigations into pulsation physics [124,125].