Identification of the Top TESS Objects of Interest for Atmospheric Characterization of Transiting Exoplanets with JWST

Since the first exoplanets were discovered by Wolszczan & Frail (1992) and Mayor & Queloz (1995), over 5000 exoplanets have been confirmed, opening up a wide array of planets of varying sizes, temperatures, and masses for study. The rate of exoplanet discovery has notably accelerated over time, originating with serendipitous or targeted observations and culminating in the concerted efforts of ground-based surveys such as the Wide Angle Search for Planets (WASP; Pollacco et al. 2006), the Hungarian-made Automated Telescope Network (HATNet; Bakos et al. 2004), and HATSouth (Bakos et al. 2013), and space-based observatories such as the COnvection, ROtation and planetary Transits satellite (CoRoT; Auvergne et al. 2009; Moutou et al. 2013), Kepler (Borucki et al. 2010), K2 (Howell et al. 2014), and the Transiting Exoplanet Survey Satellite (TESS; Ricker et al. 2015).

Although the exoplanet discovery process can reveal important properties of planets like mass and radius, further observations and analysis are required to understand the conditions on the planets themselves and examine the planet's atmospheric composition and dynamics. The first observation of an exoplanetary atmosphere was conducted by Charbonneau et al. (2002), and since then, in a parallel to the diversity of the types of exoplanets, spectroscopic characterization has revealed a wide variety of atmospheric compositions and aerosol properties as well (e.g., Sing et al. 2016; Welbanks et al. 2019; Mansfield et al. 2021; Changeat et al. 2022; August et al. 2023).

Transmission and emission spectroscopy have proven to be the workhorses of exoplanetary atmospheric characterization. These methods utilize the absorption of stellar flux transmitted through the exoplanetary atmosphere and the thermal emission from the exoplanet to probe the atmospheric characteristics of the planet. Exoplanet atmospheric characterization and spectral modeling have greatly expanded our understanding of the formation and evolution of planets, the physical and chemical processes that shape planetary atmospheres, and atmospheric aerosol properties (e.g., Madhusudhan 2019; Mollière et al. 2022; Wordsworth & Kreidberg 2022) as well as the range of diverse conditions within each of these individual topics. As the outermost layer of a planet, the atmosphere is the easiest component of an exoplanet to probe in detail and can be used to infer other planetary properties.

Although space- and ground-based resources for atmospheric characterization have become more abundant since the first transmission spectrum was taken, these resources remain in high demand. The premier atmospheric characterization tools have largely been the Hubble Space Telescope (HST) and, until its retirement in 2020, the Spitzer Space Telescope, both of which have historically been heavily oversubscribed. High-resolution spectrographs on ground-based telescopes have become increasingly important in the study of exoplanet atmospheres, but these are often limited by what is visible in the night sky and signal-to-noise ratios (S/Ns).

The highly anticipated JWST launched in 2021 (Gardner et al. 2006, 2023) with promises of greatly improved capabilities for transit and eclipse exoplanet atmospheric characterization (e.g., Deming et al. 2009; Greene et al. 2016; Stevenson et al. 2016) owing to its large aperture and infrared (IR) instrument complement. Although still early in its mission, JWST has already begun delivering on these promises with its first year of exoplanet results (e.g., Ahrer et al. 2023; Greene et al. 2023; Kempton et al. 2023; Tsai et al. 2023). This is not even to mention JWST's capabilities for the spectroscopy of directly imaged exoplanets (e.g., Miles et al. 2023), which is impressive but outside the scope of this work. But time on JWST is in high demand, and this, coupled with the review process for general observer programs, has resulted so far in a patchwork of exoplanet atmospheric observations.

When it comes to identifying targets for atmospheric characterization observations, there is a critical synergy between JWST and TESS. Touted from the very beginning as a "finder scope for JWST," the almost-all-sky survey strategy of TESS was intended to find a myriad of new planets around bright, nearby stars that would be amenable to atmospheric characterization with JWST (Deming et al. 2009), in contrast to the dimmer, more distant host stars of Kepler planetary systems. So far, TESS has discovered more than 300 confirmed planets, with more than 4000 planet candidates classified as unconfirmed TESS Objects of Interests (TOIs) without either a false positive or confirmed planet disposition. ¹²³ There is currently no published false-positive rate for TESS, although recent work estimates that it could be somewhere between 15% and 47% depending on the mass of the planet and host star (Zhou et al. 2019; Kunimoto et al. 2022). Therefore, it is probable that many of these 4000 TOIs are false positives. However, if even a fraction of them are true planets, this would dramatically grow the sample of planets whose atmospheres may be well suited to observe and characterize with JWST.

In fact, some of the highest-quality (i.e., highest S/N) atmospheric characterization exoplanet targets likely still lie among the unconfirmed TOIs list, since TESS has unique capabilities for finding small planets orbiting bright stars in particular. The JWST-TESS synergy is demonstrated especially by the fact that ∼37% of JWST Cycle 1 and ∼56% of JWST Cycle 2 exoplanet targets are TESS discoveries. This high proportion of TESS-discovered JWST targets is displayed in Figure 1. With JWST already flying, it is of the utmost importance to systematically and expeditiously identify the best JWST targets to provide a uniform coverage of parameter space.

Zoom In Zoom Out Reset image size

In an effort to better streamline use of JWST for atmospheric characterization and identify which targets are likely to exhibit the most clearly detectable features in their atmospheric spectra, we present a set of "best-in-class" targets for transmission and emission spectroscopy. Our aim is to produce a sample of planets (or likely planets) well suited for JWST atmospheric characterization to serve as a community resource for upcoming JWST proposal cycles and future observing programs aimed at regions of planetary parameter space where the highest S/N targets have yet to be identified. Under mass assumptions that we describe in Section 3, these targets are expected to be well suited for JWST. In order to increase the usefulness of our prioritization scheme, we also perform vetting and validation analyses on each member of the best-in-class sample, since our initial rankings make no distinction between confirmed and candidate planets. We first describe our analysis procedure for our ranking and validation analyses in Section 2.

Our analysis procedure falls into three main steps: (i) generation of the best-in-class sample and ranking of confirmed and candidate planets within each parameter space bin, (ii) vetting of each unconfirmed candidate planet that appears in the best-in-class sample using a variety of ground-based observations and results from previously published vetting software, and (iii) validation analysis of each planet candidate using multiple statistical-validation software packages. A graphical summary of this procedure can be found in Figure 2.

Zoom In Zoom Out Reset image size

Our best-in-class sample consists of the targets ranked in the top five according to the transmission spectroscopy metric (TSM) and emission spectroscopy metric (ESM) from Kempton et al. (2018) within each cell of a grid spanning the planetary radius and equilibrium temperature (R_p–T_eq) space, which is described in Section 3. R_p–T_eq axes were chosen since planetary radius (R_p) is expected to be a proxy for metallicity (Baraffe et al. 1998; Fortney et al. 2013) while temperature correlates to chemistry and aerosol formation (Gao et al. 2020), and both parameters are easy to estimate for transiting exoplanets. Metallicity and atmospheric chemistry can both provide insights to the formation, physical processes, and composition of a planet's atmosphere and are important to probe. We account for the technical capabilities of JWST's instruments through the inclusion and calculation of various additional metrics (e.g., stellar host magnitude, expected atmospheric signal size, and observability metrics benchmarked against JWST's instrumental capabilities) for each target and further incorporate these values into our rankings, thus tuning our best-in-class sample to JWST specifically. A further description of our methods in obtaining the sample of planets and planet candidates, generating atmospheric characterization and observability metrics and ultimately ranking the targets into our best-in-class framework, can be found in Section 3.

In our rankings, we initially make no distinction between confirmed planets and unconfirmed TOIs in order to assess how the TESS planet candidates fit in with the overall sample and to identify which TOIs might displace known planets as best-in-class atmospheric characterization targets. For each unconfirmed TOI on our best-in-class list, we perform cursory vetting and statistical validation to determine which targets are likely false positives and which are worthy of additional follow-up prior to future atmospheric characterization observations with JWST. We note that while we only statistically "validate" planets rather than label them as "confirmed," we consider them to be planets for the purposes of our best-in-class sample (Torres et al. 2004, 2011). For our vetting analysis, we utilize a host of ground-based reconnaissance photometric, spectroscopic, and high-resolution imaging observations to inform our understanding of each unconfirmed TOI and help make determinations on the planetary status of each. Information on these ground-based observations can be found in Section 4. The ground-based observations are used in conjunction with results from automated vetting software that have already been published for a large fraction of our sample. This overall vetting process is described in Section 5. Finally, each unconfirmed planet candidate is run through statistical validation software in order to obtain a false-positive probability (FPP) that can be used to quantify the likelihood that each unconfirmed planet candidate is a true planet and further prioritize targets within the best-in-class sample. This analysis also utilizes our ground-based follow-up. Further description of this analysis can be found in Section 6.

We present the final results of our ranking, vetting, and validation analyses in Section 7. This section contains the final ranked best-in-class grids and our assessed disposition for each target (e.g., statistically validated, likely false positive, etc.).

Identifying targets across R_p–T_eq space that are well suited to atmospheric characterization with JWST is critical to our understanding of exoplanet atmospheres. By sampling across this parameter space, we expect to cover a range of metallicities as well as atmospheric chemistry and aerosol regimes that would allow us to tease out trends and test models on the population level. This could include a mass–metallicity relation, an aerosol–T_eq relation, or a transition between planets that have CO versus CH₄ in their atmospheres as the dominant carbon carrier. To accomplish this, we have divided up the R_p–T_eq parameter space into a grid, sorted each planet and planet candidate into cells within this grid, and ranked each target according to its expected S/N approximated via its TSM or ESM.

3.1. Provenance of Sample Parameters and Transmission Spectroscopy Metric and Emission Spectroscopy Metric Calculation

In order to obtain a standardized list of planets and planet candidates to consider when determining which are the best-in-class for atmospheric characterization with JWST, we relied on the data tables maintained by the NASA Exoplanet Archive and the parameter values contained therein. ¹²⁴ The Exoplanet Archive collates parameter sets for confirmed and unconfirmed planets and acts as a single repository for published parameter values for each target. For the confirmed planets, we downloaded the Planetary Systems table, which contains every planet that has a published validation or confirmation, and the accompanying set of parameter values, with a single parameter set labeled as the default by the archive staff for each planet. For the unconfirmed TOIs, we downloaded the TESS Candidates table from the Exoplanet Archive, which updates directly from the TESS TOI Catalog (Guerrero et al. 2021) with new targets and refined parameter values from the TESS mission. These two tables were both downloaded on 2022 November 3. The highest TOI number alerted at this time was TOI-5863.

We elected to use the parameter set denoted as the default set of values for each of the planets in the Planetary Systems table throughout our analysis. In the case that the default parameter set was incomplete and missing values for critical parameters necessary to our analysis, values were pulled from other, nondefault parameter sets for each planet, if they existed. Critical values were planetary radius (R_p), stellar radius (R_*), stellar effective temperature (T_*), semimajor axis (a), J magnitude, and K magnitude. Values with lower uncertainties from other parameter sets were given priority for inclusion in the final parameter set. This list of critical parameters represents the bare minimum set of quantities necessary to perform our best-in-class ranking procedure. The only other quantity that is used, if available, is planet mass (M_p) and its associated uncertainties for planets that are confirmed. Taken together, the critical quantities listed above and M_p are all of the parameters used by our ranking framework.

We calculated the TSM and ESM for each planet according to the prescription outlined in Kempton et al. (2018), specifically their Equations (1) and (4). The calculation of TSM and ESM assumes cloud-free atmospheres, solar composition for planets larger than 1.5 R_⊕, and a pure H₂O steam atmosphere for planets smaller than 1.5 R_⊕. These two values represent analytical metrics that quantify the expected S/N in transmission and thermal emission spectroscopy for a given planet and can be used to identify which planets are best suited for atmospheric characterization with JWST relative to one another. We maintained two separate samples for best-in-class targets: one for transmission spectroscopy driven by TSM and the other for emission spectroscopy driven by ESM. Both of these initially started with the same overall sample of planets and planet candidates downloaded from the Exoplanet Archive, and were each shaped by the observational constraints unique to each respective sample. Figure 3 illustrates the parameter space coverage of our combined best-in-class samples.

Zoom In Zoom Out Reset image size

Even after pulling values from other parameter sets, some targets did not contain finite values for all of the parameters necessary to calculate the spectroscopy metrics and the observability criteria with which we defined and ranked our sample. For targets without a value for the ratio between semimajor axis and stellar radius, a/R_*, we converted both the semimajor axis and the stellar radius to units of meters and took the ratio of the two. In the case that a was missing but a/R_* was a finite value, a/R_* was multiplied by R_* to calculate a. A similar procedure was performed for the ratio between the planet and stellar radii, R_p/R_*. We preferred to use the reported ratios if they existed to reduce the propogation of potential errors in generating these ratios from the reported values of their individual components. Reported mass and equilibrium temperature (T_eq) values were used when reported, but were calculated later in the procedure if unavailable. All targets that still lacked full parameter sets to perform the necessary calculations were removed from the sample. We checked each parameter set to ensure that R_p/R_* < 1, and targets with values that did not conform to this criterion were replaced with a value from another parameter set, if available.

For planets from the Planetary Systems table and candidates from the TOI list that did not have published masses, we calculated masses using a mass–radius distribution adapted from the mean of the Chen & Kipping (2017) mass–radius distribution. Specifically, we altered the value of the S³ coefficient in the power-law function that describes the mass–radius relation of the Jovian planets, R ∝ ${M}^{{S}^{3}}$. We set this value to be 0.01 rather than −0.044, to ensure that each radius value corresponded to a unique mass, while minimally affecting the shape of the curve as presented in Chen & Kipping (2017). We used this distribution up to planetary radii of 15 R_⊕, fixing the mass of planets larger than this threshold to 1 M_Jup. Above this radius, the scatter of the mass–radius distribution is large and results in a mean that is nearly constant in mass across radius. This is the same procedure that is used by the Exoplanet Archive to calculate expected masses.

We divided the sample into three categories: confirmed planets with >5σ mass measurements, planets marked as confirmed on the Exoplanet Archive with <5σ mass measurements, and unconfirmed planet candidates without any mass measurement. Batalha et al. (2019) showed that different mass confidence levels result in different precision with which an exoplanet's atmosphere can be characterized. A stratification of these targets based on mass measurement will also allow the community to better prioritize follow-up resources for the best-in-class targets, and allowed us to identify which targets are unconfirmed and in need of statistical validation.

Additionally, we calculated the mass of the host star for each TOI based on the star's log g and radius because stellar mass is not included in the Exoplanet Archive's TOI table. Using the host star's reported effective temperature, we also assigned each host star an approximate stellar type for reference. We then calculated the T_eq of each planet—both TOI and confirmed—according to Equation (3) of Kempton et al. (2018). This was done to ensure a uniform data set for T_eq since the definition of equilibrium temperature varies with each data set on the Exoplanet Archive, with different assumptions regarding surface albedo and atmospheric heat distribution serving as variables with no set standard. Since T_eq is integral to our determination of the best targets for transmission and emission spectroscopy, we elected to calculate the value for each planet and planet candidate to ensure a uniform comparison. Our calculation of T_eq assumes zero albedo and full day–night heat redistribution.

3.2. Observability Cuts

While useful for relative comparisons between targets, the TSM and ESM only predict the S/N but do not account for other observability considerations such as the absolute signal size relative to the instrumental noise floor or the target being within an instrument's brightness limits. To incorporate the observability of our sample with JWST into our best-in-class rankings, we also calculated the expected sizes of transmission spectral features and secondary-eclipse depth for transmission and emission spectroscopy, respectively.

3.2.1. Observability in Transmission

We again follow the prescription outlined in Kempton et al. (2018), expressing the size of expected spectral features at one scale height as

$$\begin{eqnarray}&&\displaystyle \frac{2{R}_{{\rm{p}}}}{{R}_{* }^{2}}\times \displaystyle \frac{{{kT}}_{\mathrm{eq}}}{\mu g},\end{eqnarray} \tag{ 1 }$$

where k is the Boltzmann constant, μ is the mean molecular weight of the atmosphere, and g is the surface gravity of the planet. For planets with R_p > 1.5 R_⊕, we assume μ = 2.3 (in units of proton mass, m_p ) while for planets with R_p < 1.5 R_⊕, we assume μ = 18 proton masses, following the assumption that all planets in a given radius bin have the same atmospheric composition as made by Louie et al. (2018). We calculated g using the expression g = GM_p/${R}_{{\rm{p}}}^{2}$, where G is the gravitational constant and M_p and R_p are the mass and radius of the planet, respectively. The second term of Equation (1) represents the scale height of the planetary atmosphere, H. This is used as a proxy for spectral feature size as it represents the depth into that atmosphere that is probed at a specific wavelength, which in turn determines the measured wavelength-dependent differences in transit depth.

We assumed a depth of 2H when calculating expected spectral feature size based off of the spread in the sizes of H₂O features observed using HST's near-IR (NIR) WFC3 instrument (Stevenson 2016). The average size of these features was reported to be ∼1.5H, but at longer wavelengths such as those probed by JWST the size of spectral features for molecules such as H₂O increases (e.g., Coulombe et al. 2023), so we elected to assume a depth slightly above the average reported by Stevenson (2016). Assuming a larger expected spectral feature also allows for us to capture more planets for comparison within our sample as well as to account for differences in cloud cover or the mean molecular weight of exoplanet atmospheres.

In fact, for all constraints applied to our sample, we chose liberal thresholds in order to allow for more targets to appear in our best-in-class sample, especially in parameter spaces where there otherwise would be no promising targets. This was done not only for illustrative purposes but also to attempt to account for some of the variance in parameters governing exoplanet atmospheres and potentially improved observational capabilities going forward.

To ensure that all best-in-class targets would be observable with JWST, we imposed a requirement for a 2σ spectral signal size assuming a noise floor of 10 ppm for the NIRCam, NIRISS, and NIRSpec instruments on JWST. These instruments are all ideal for transmission spectroscopy since their wavelength coverage includes prominent transmission spectral features. We note the TSM was benchmarked for use with NIRISS (Kempton et al. 2018).

3.2.2. Observability in Emission

We perform a similar procedure for the secondary-eclipse depth in order to determine which targets are amenable for emission spectroscopy with the MIRI instrument on board JWST. The expected secondary-eclipse depth can be estimated using the expression

$$\begin{eqnarray}&&\displaystyle \frac{{B}_{7.5}({T}_{\mathrm{day}})}{{B}_{7.5}({T}_{* })}\times {\left(\displaystyle \frac{{R}_{{\rm{p}}}}{{R}_{* }}\right)}^{2},\end{eqnarray} \tag{ 2 }$$

where B_7.5 is the Planck function evaluated for a given temperature at a representative wavelength of 7.5 μm, T_day is the dayside temperature of the planet as calculated by 1.1 × T_eq, and R_p/R_* is the ratio of the planetary and stellar radii. We calculate the dayside temperature as 1.1 × T_eq to account for the dayside hotspot on the planet, following the analysis by Kempton et al. (2018) that tuned this relation according to a suite of global circulation and one-dimensional atmospheric models. The 7.5 μm was chosen as the representative wavelength since it is the center of the "conservative" MIRI Low Resolution Spectroscopy (LRS) bandpass on JWST as data beyond 10 μm are often unreliable (Bell et al. 2023; Kempton et al. 2023) and 7.5 μm is still near the peak of the MIRI LRS response function (Kendrew et al. 2015; Rieke et al. 2015). We imposed a requirement that the secondary-eclipse depth be measurable to the 3σ level assuming a noise floor of 20 ppm for the MIRI instrument on JWST. There were more small planets contained within the emission spectroscopy sample, and so we were able to adopt a more conservative 3σ threshold rather than the 2σ threshold applied to the transmission spectroscopy sample. We also imposed an ESM > 3 requirement on our emission spectroscopy sample to remove targets that would produce small secondary eclipses even under ideal observing conditions with JWST. Like TSM with NIRISS, ESM was benchmarked for use with MIRI, which is ideal for emission spectroscopy among JWST's instruments thanks to its longer wavelength coverage that maximizes the ratio between the flux of the planet and that of the host star.

3.3. Additional Cuts and Organizing the Sample

3.4. Description of Best-in-class Grids

The planets contained within each bin in radius and temperature space were then sorted and ranked by TSM and ESM for the transmission spectroscopy and emission spectroscopy samples, respectively. This ranking was agnostic to confirmation status and the existence of a well-constrained mass, resulting in a combination of confirmed planets and unconfirmed planet candidates within each grid cell. The top five targets in each bin are considered the best-in-class for that portion of parameter space. Our rankings of the transmission and emission spectroscopy targets are contained within the grids shown in Figure 4.

Zoom In Zoom Out Reset image size

Almost every bin for both the transmission and emission target samples has at least one unconfirmed planet candidate, with most bins dominated by unconfirmed candidates. While certainly not all of the planet candidates are true planets, if even a fraction of the them are these rankings indicate that there is a large number of TESS planet candidates that are both (i) among the best currently known targets for atmospheric characterization with JWST from a S/N perspective, and (ii) required to provide a uniform coverage of the R_p–T_eq space.

In order to determine which of the TESS-discovered planet candidates in our best-in-class samples are true planets, we first collated all of the follow-up observations for each target. These follow-up observations provided valuable, independent information on the validity of each planet candidate as true planets. We worked closely with the TESS Follow-up Observation Program (TFOP) subgroups (SGs) to compile available photometric, spectroscopic, and imaging follow-up observations for each target. ¹²⁵ These observations were used in initial vetting to determine whether each target was a likely false positive or if it could proceed to more in-depth vetting and validation. TFOP follow-up observations and the constraints that they impose on the system were incorporated into our vetting and statistical-validation procedures where possible (see Sections 5 and 6). The follow-up resources used in vetting and validating the best-in-class planet candidates are summarized here, with a representative sample of the specific observations used for individual targets detailed in Table 5 located in Appendix B and a full, machine-readable version available from the online version of this article. An outline of where follow-up observations were used in our vetting procedures can be found in the middle panel of Figure 2.

4.1. Ground-based Photometry

4.2. Reconnaissance Spectroscopy

4.3. High-resolution Imaging

In order to determine the planetary nature of each target, we performed a uniform vetting procedure on each of the unconfirmed candidates. This included utilizing a mix of publicly available resources and follow-up observations obtained by TFOP. We outline our overall procedure in schematic form in the middle panel of Figure 2. We ran each target through as many steps of our vetting procedure as possible given the availability of resources at the time of analysis in early 2023 since not all targets had the resources to complete each step in our procedure.

Although our vetting procedure checked for a number of false-positive indicators, we refrained from classifying a target as a likely false positive unless multiple false-positive indicators suggested that the origin of the transit signal could not have been a planet. Our conservative approach to vetting passed most targets on to statistical validation and provided invaluable information to be used in conjunction with the results from our validation analysis to make a final determination, such as if the signal is on target and if there were any potentially contaminating sources contained in the light curve's extraction aperture. In this way, vetting served as a complement to a more holistic determination of the planetary nature that accounts for a larger number of factors than any individual analysis alone could provide.

For all of our vetting, we used the orbital and planetary parameter values posted on ExoFOP unless follow-up observations revealed more accurate or precise values for a given parameter, in which case the parameters obtained from follow-up were used. ¹²⁶ There were seven targets with ambiguous periods contained in our best-in-class samples from the query of the Exoplanet Archive (TOIs 706.01, 1856.01, 1895.01, 2299.01, 4317.01, 5575.01, and 5746.01). These targets were single transits that transited again in one or more later TESS sectors, but without measuring two or more consecutive transits the period could not be confidently determined. The orbital period of TOI-5575.01 was uniquely determined to be 32.07 days through follow-up observations over the course of our analysis. ¹²⁷ We propagated this updated period throughout our analysis and report the planet's updated parameters in our final best-in-class sample. Since we are unable to obtain the true periods of these remaining six targets without a concerted observing campaign, we analyzed them with the reported ExoFOP periods that represent upper limits to the true periods. Shorter periods would likely result in higher equilibrium temperatures, which, although this would boost the TSM, could place these targets in a different temperature bin where they may not rank in the top five targets in their R_p bin. We recognize that the periods and therefore amenability to atmospheric characterization with JWST may change for these targets, but we include them in our best-in-class samples to emphasize their potential as prime JWST targets and encourage their further study.

5.1. TESS SPOC Data Validation Report

5.2. DAVE Vetting from Cacciapuoti et al. (2022)

5.3. Reconnaissance Photometry

5.4. Reconnaissance Spectroscopy

5.5. Imaging Constraints

While vetting is an integral step in determining whether a periodic signal is indeed due to the presence of a planet, it cannot alone demonstrate that a signal is not a false positive. The preferred method for determining whether a signal is a planet is a mass measurement through RV observations, however these oftentimes require a significant commitment of resources and time on targets that may not prove to be planets.

In lieu of a mass measurement, statistics can be used to validate the target rather than confirm it. Statistical validation of a target often only requires photometric and imaging observations as well as planetary and orbital parameters input into one or multiple statistical-validation software packages. Targets that are validated to a greater than 99% confidence threshold are considered planets despite not having a mass measurement (Morton 2012; Giacalone et al. 2021). Since the time and observational resources required to validate a planet are far less than required to obtain a mass measurement, statistical validation serves as an excellent intermediate step to weed out targets that are very likely not planets in order to better streamline and prioritize the RV observations required to confirm a target as a bona fide planet.

In the case of our best-in-class samples, since there are undoubtedly false positives among the unconfirmed planet candidates, we performed statistical validation on all candidate planets to not only determine which targets are most likely to be true planets but which merit follow-up with RV observations. To do this, we run the statistical-validation software vespa (Morton 2012, 2015) and TRICERATOPS (Giacalone & Dressing 2020) on each of our unconfirmed targets in both the transmission and emission spectroscopy samples.

For all of our targets, we use the orbital and planetary parameters from ExoFOP unless the follow-up observations reported refined parameters (Section 4), in which case the refined parameters were used. For vespa, this also included stellar parameters. The refined parameters from follow-up observations were used to recalculate the TSM and ESM values for each target. The refined parameter values were almost always similar to the ExoFOP values, and so incorporating these refinements from follow-up did not produce changes in the best-in-class rankings. TESS photometry was used to produce the phase-folded transits used in both vespa and TRICERATOPS. When possible, we favored light curves produced by the TESS SPOC at the shortest cadence available since shorter-cadence TESS data have been shown to be more photometrically precise when binned than data taken at the binned cadence itself (Huber et al. 2022). A small subset of targets did not have SPOC PDCSAP light curves, in which case we used light curves produced by MIT's QLP.

6.1.  vespa

6.2.  TRICERATOPS

Similar to vespa, TRICERATOPS (Giacalone & Dressing 2020) compares the user-provided phase-folded transit, orbital, and stellar parameters against a set of astrophysical false-positive scenarios to rule out portions of parameter space in which the false-positive scenarios can remain viable. The methodology of TRICERATOPS is identical to vespa in many respects, however, in contrast to vespa, TRICERATOPS was developed specifically for TESS and accounts for the real sky background of each target out to 25 as well as the TESS point-spread function and aperture used to extract the photometric light curve in each sector of TESS data. An example of what TRICERATOPS considers in this portion of its analysis is seen in Figure 5.

Zoom In Zoom Out Reset image size

For each target, we used the extraction apertures produced by the TESS SPOC contained within the headers of the SPOC PDCSAP light curves queried by lightkurve (Lightkurve Collaboration et al. 2018) on a sector-by-sector basis. For the targets missing SPOC PDCSAP light curves from some or all TESS sectors they were observed in, we used a standard aperture of 5 × 5 TESS pixels. This is larger than any of the PDCSAP apertures and is the TRICERATOPS default for sectors without provided apertures.

When accounting for nearby background stars for each target, TRICERATOPS queries the TICv8 for the stellar parameters of each star. The TIC is based heavily on Gaia Data Release 2, which has since been updated by Gaia DR3. Therefore, in our analysis, we queried the R.A., decl., mass, effective temperature, parallax, and Gaia G magnitude of the host star Gaia DR3 Catalog for use in our analysis in lieu of using the values provided by the TIC. To convert the Gaia magnitude to TESS magnitude, we used Equation (1) from Stassun et al. (2019), which is valid for dwarfs, subgiants, and giants of any metallicity. We then cross-referenced each Gaia target with the Two Micron All Sky Survey catalog (Skrutskie et al. 2006) to obtain J, H, and K magnitudes where available as these magnitudes are used by TRICERATOPS in its estimation of FPP.

Additionally, we included follow-up constraints into our analysis with TRICERATOPS. When available, we included a contrast curve from high-resolution imaging to constrain the existence of additional stellar-mass companions in the system. Unlike vespa, TRICERATOPS accepts only a single contrast curve per target, so in the case a target possessed multiple contrast curves from follow-up observations, we included only the contrast curve that provided the greatest imaging contrast magnitude agnostic of bandpass to most stringently constrain possible companions in the system. Furthermore, our photometric follow-up allowed us to clear individual nearby stars of potentially harboring EBs that would cause the observed transit signal on-target. Background stars that were definitively determined to not be EBs at the target period or have an eclipse depth that could cause the observed transit on-target were discarded from consideration as potential sources of a false positive. Targets whose transits were observed on-target had all background stars removed from false-positive consideration. As recommended by Giacalone et al. (2021), we ran multiple trials of the TRICERATOPS FPP calculation for each target with a minimum of 10 trials per target, and report the mean of these FPPs.

TRICERATOPS provides not only a final FPP value but also a nearby false-positive probability (NFPP) value that encapsulates the probability that the signal originates from a star other than the target. Giacalone et al. (2021) defines validated planets as signals with FPP < 0.015 and NFPP < 10⁻³ and outlines a separate category for marginal validations when FPP < 0.5 and NFPP < 10⁻³. We adopt these categories in our determination of the planetary nature for our best-in-class samples. We extend the marginal validations category to vespa, which does not explicitly have such a distinction. In the case of vespa, we conservatively set the marginal validation threshold to FPP < 0.25, lower than that of TRICERATOPS.

Morton et al. (2023) recommends the use of TRICERATOPS in favor of vespa since the latter is no longer maintained and has not been updated to account for the modern astronomy landscape. We present validation using both software packages as an independent check on one another in order to be as conservative as possible in any claim of validation made for a given target. Since both tools can be used in evaluating the validity of a target, that enables a more granular target prioritization within our best-in-class sample. Additionally, while TRICERATOPS was developed specifically for TESS and accounts for the actual background star field, it is only able to ingest a single contrast curve as an observational constraint whereas vespa is able to utilize multiple ground-based data sets to constrain the FPP. That said, we emphasize the results of TRICERATOPS over those of vespa in cases where their FPP values may disagree. This means that many of the targets in our best-in-class sample that are classified as "likely planets" may actually fall within the realm of true statistical validation when considering only the results from TRICERATOPS.

We also note that our statistical-validation analysis cannot rule out the scenario in which validated planets with R_p > 9 R_⊕ are actually brown dwarfs. A measured mass is required to disentangle the brown dwarf and planet scenarios, and we encourage follow-up on all validated planets to this effect.

Of the 103 unconfirmed TESS planet candidates contained in our best-in-class samples, 19 passed vetting and were calculated to have FPP values firmly meeting the threshold for statistical validation from both vespa and TRICERATOPS. Additionally, 11 of the original 103 unconfirmed planet candidates reside in potential multiplanet systems (TOI-1468.01, TOI-1468.02, TOI-1798.02, TOI-1806.01, TOI-2134.01, TOI-3353.01, TOI-406.01, TOI-4443.01, TOI-4495.01, TOI-836.02, and TOI-880.02). Three of these have already been confirmed by independent teams (TOI-1468.01 and .02 and TOI-836.02). The remaining eight TOIs are able to take advantage of a "multiplicity boost" to drive their FPP values lower. It has been shown that transit-like signals in systems with multiple transit-like signals are more likely to be true planets, assuming false positives are uniformly distributed throughout the sky (Lissauer et al. 2012). This results in a decreased FPP value of up to 54× depending on the size of the planets, how crowded the field is for signals detected with TESS, and the pipeline with which they were detected (Guerrero et al. 2021). Additionally, these potential multiplanet systems represent an excellent opportunity to perform comparative planetology with the other planets in their system using JWST.

We applied this multiplicity boost to each of the eight candidates listed above, resulting in FPP values below the validation threshold for each of them. Four of these eight already possessed FPP values from vespa and TRICERATOPS that were low enough to be statistically validated, but the FPP values of the other four targets (TOI-880.02, TOI-1798.02, TOI-1806.01, and TOI-4443.01) moved from the "marginal validation" range into the "validated planet" range. These targets are shown in Table 3. We strongly recommend these targets for additional, in-depth study and confirmation to measure their masses and model their orbits and atmospheres in preparation for potential observation with JWST.

A total of 29 targets were deemed "likely false positives" (LFPs). These targets all exhibited clear signs of a false positive in the vetting stage and/or produced FPP values from both statistical-validation software packages. A target was deemed a likely false positive if the FPP from both vespa and TRICERATOPS did not meet either the validation or marginal validation thresholds. For one of these likely false-positive targets, we were unable to locate the transit-like event that was flagged by the TESS SPOC during our manual inspection of the phase-folded light curve and we deemed it a false alarm (TOI-1022.01). Most of these 29 likely false-positive targets exhibited obvious V-shaped transits indicative of an EB, and a subset of them were revealed by TFOP follow-up to have a nearby (≤2'') companion star that served as the likely cause of the signal.

There was a subset of targets with high FPP values that could be large grazing planets or systems with a high planetary-to-stellar-radius ratio (R_p/R_*) rather than their current LFP classification. Grazing transits or high- R_p/R_* systems often produce transits that look somewhat V-shaped and can masquerade as a stellar eclipse rather than a planet transit. These scenarios are limiting cases for the validation software since the analyses rely so heavily on transit shape. Therefore, targets with high FPP values that could potentially fall under these categories warrant further follow-up. For our purposes, we keep these targets classified as LFPs not only for the sake of a uniform analysis but also because grazing transits are nonideal candidates for transit and eclipse spectroscopy. However, we flag them here for future study and as examples of the limitations of statistical validation.

A third category of validation emerged for targets with FPP values that did not quite meet the threshold for validation but also were not clear false positives. These 31 targets were deemed to be marginal validations and had at least one or both FPP values from vespa and TRICERATOPS that met the marginal validation criteria described in Section 6. This category was further subdivided into "likely planets" (LPs) and "potential false positives" (pFPs). LPs were targets with either both FPP values residing in the marginal validation zone or one FPP in the marginal validation zone and the other meeting the threshold for validation. pFPs were targets with one marginal validation FPP and one FPP that indicates a false positive.

The results of our vetting analysis agree with these distinctions based on FPP. Almost all of the targets in the pFP category had at least one vetting factor that could indicate a false-positive origin (e.g., V-shaped transit, possible odd–even transit depth differences, etc.) but are not definitive enough to warrant labeling the target a likely false positive. There were a total of 13 targets in the LP category and 18 in the pFP category of marginal validations. We encourage future study and follow-up of these targets to ascertain their true nature as they could potentially be prime targets for atmospheric characterization with JWST. Examples of transits from each disposition category are shown in Figure 6.

Zoom In Zoom Out Reset image size

The remaining four targets produced inconclusive vetting and validation results. This category is distinct from marginal validations in that in most of these inconclusive cases vespa and TRICERATOPS disagree significantly on the status of each target or there are additional factors precluding an adequate vetting or validation analysis. The targets TOI-1355.01, TOI-1954.01, and TOI-4552.01 were validated by one statistical-validation software while the other software produced an FPP that did not meet the threshold for even a marginal validation.

In the case of TOI-1355.01, the discrepant FPPs may be due to overly constraining photometric follow-up observations. The transit shape is slightly V-shaped, and our follow-up observations rule out a large portion of parameter EB and BEB parameter space, but those are the models that fit the phase-folded transit the best (resulting in FPP values with large uncertainties from TRICERATOPS). This target may be a grazing planet, which would explain the V-shape as well as the small parameter space for EBs and BEBs.

In the case of TOI-1954.01, very little follow-up exists and the target is in a crowded field, both of which likely combine to cause the discrepancy between vespa and TRICERATOPS. For TOI-4552.01, the signal is shallow and the light curve exhibits some variability, which is likely causing variability in the FPP values calculated by the different validation software packages.

The final inconclusive case is TOI-4597.01. This target was statistically validated by vespa but TRICERATOPS was unable to run on it. This is likely due to the short periodic oscillations that appear in the light curve as a result of stellar activity or variability. A clear transit exists, but we are unable to complete our vetting and validation analysis without properly modeling the variability in the light curve to produce a clean transit. This is beyond the scope of this work as it would require a physically motivated model to subtract from the light curve that our vetting and validation procedure is incapable of. We encourage follow-up analysis of these four inconclusive targets to determine their true nature.

There were an additional 21 targets from our samples that were confirmed by independent teams over the course of our analysis. These targets are TOI-179 b (Desidera et al. 2023; de Leon et al. 2023), TOI-238 b (Mascareño et al. 2024), TOI-332 b (Osborn et al. 2023), TOI-622 b (Psaridi et al. 2023), TOI-836 b (Hawthorn et al. 2023), TOI-969 b (Lillo-Box et al. 2023), TOI-1099 b (Barros et al. 2023), TOI-1194 b (Wang et al. 2023), TOI-1347 b (Rubenzahl et al. 2024), TOI-1468 b & c (Chaturvedi et al. 2022), TOI-1853 b (Naponiello et al. 2023), TOI-2134 b (Rescigno et al. 2024), TOI-3235 b (Hobson et al. 2023), TOI-3884 b (Almenara et al. 2022), TOI-4463 A b (Yee et al. 2023), GJ 806 b (TOI-4481b; Palle et al. 2023), HD 20329 b (TOI-4524b; Murgas et al. 2022), TOI-4641 b (Bieryla et al. 2024), TOI-4860 b (Almenara et al. 2024; Triaud et al. 2023), and Wolf 327 b (TOI-5727 b; Murgas et al. 2024). This high number of targets in our best-in-class sample being confirmed in such a short period of time is very positive for the prospects for atmospheric characterization with JWST. Indeed, the goal of our synthesis of the best-in-class sample is to highlight and elevate targets potentially well suited to such observations for follow-up to measure their masses and confirm their planetary nature.

For the targets that had masses measured independently over the course of our analysis, we recalculated their TSM and ESM values according to their updated planet parameters and reranked them within their respective bins. In the cases of TOI-179 b, TOI-238 b, TOI-332 b (Osborn et al. 2023), TOI-836 b, TOI-969 b, TOI-1099 b, TOI-1194 b, TOI-1347 b, TOI-1853 b, TOI-2134 b, TOI-3235 b, TOI-4463 A b, TOI-4641 b, and TOI-4860 b, the updated parameters differed significantly from those originally listed on the NASA Exoplanet Archive, resulting in either TSM and/or ESM values much lower than originally calculated or placement into a different parameter space bin where they no longer ranked within the top five targets. In either case, this resulted in removal from either the best-in-class transmission or emission spectroscopy samples. In the cases of TOI-238 b, TOI-836 b, TOI-969 b, TOI-1099 b, TOI-1347 b, TOI-1853 b, TOI-2134 b, TOI-4463 A b, and TOI-4641 b, this amounted to removal from the entire best-in-class sample.

Our best-in-class sample also includes seven targets with ambiguous periods: TOI-706.01, TOI-1856.01, TOI-1895.01, TOI-2299.01, TOI-4317.01, TOI-5575.01, and TOI-5746.01. These targets were originally discovered as single transits before transiting again in later sectors of TESS data. The periods reported on ExoFOP represent the upper limit on their periods since additional transits of these targets could have fallen in gaps in the TESS data and their true periods may be shorter. We performed our vetting and validation using the stated periods but knowing that future observations could reveal shorter periods that would alter the TSM and ESM values as well as their observability with JWST. We choose not to discard these targets from our best-in-class samples to emphasize their potential as ideal JWST targets and emphasize the need for additional follow-up on them. Only TOI-4317.01 had low enough FPP values to be considered statistically validated, but due to its ambiguous period we place it in the LP category, and we caution that a deeper analysis is required for this target to identify the true period and therefore its true planetary status.

We note that the planets with long or ambiguous periods in our sample should have their orbital periods further scrutinized as the TESS observing strategy makes it difficult to determine such long orbital periods and they may change depending on individual circumstances. Should these targets be shown to be true planets on shorter periods, their status in the best-in-class sample is not likely to change. This is because they are all members of the transmission spectroscopy best-in-class sample and a shorter period will increase their T_eq, therefore increasing their TSM value.

The final best-in-class sample is displayed in Figure 7, which mirrors Figure 4 but now includes updated dispositions for all targets in our samples, both confirmed and unconfirmed. ¹²⁸ Additional information on each target can be found in Table 4 in Appendix A. An extended machine-readable version of this table is also available in the online version of this article. We also include an updated version of Figure 4 reflecting the new dispositions of the sample. This can be found in Figure 8.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

There were a number of targets in our best-in-class sample that are also in the process of being validated by independent teams. These include TOI-4226.01 (M. Timmermans 2024, in preparation) and TOI-4317.01 (H. Osborn et al. 2024, in preparation). We direct the reader to these upcoming publications for a more in-depth analysis and exploration than is available here and to treat such in-depth analyses as the definitive discovery papers for these individual targets. C. Dressing et al. (2024, in preparation) is also conducting a parallel large-scale validation effort on TOIs 261.01, 4317.01, 4527.01, 4602.01, and 5082.01, as is Mistry et al. (2023) for TOI-771.01, and we direct the reader to this upcoming publication for an additional, independent vetting and validation of these targets. Additionally, independent teams are conducting confirmation and characterization of TOI-1410.01 (J. Livingston 2024, in preparation; A. S. Polanski 2024, in preparation) and TOI-880.02 (L. D. Nielsen 2024, in preparation), and we direct the reader to these papers for an in-depth analysis of these targets.

In this paper, we present a set of best-in-class planets for atmospheric characterization with JWST through both transmission and emission spectroscopy. Our vetting, validation, and results are summarized here:

• 1.  
  We queried the NASA Exoplanet Archive for all transiting confirmed planets and unconfirmed TESS candidates and calculated their TSM, ESM, and observability with JWST.
• 2.  
  We divided all planets into grids with bins in equilibrium temperature from 100 to 3000 K and planetary radius from 0.3 to 25.0 R_⊕, and the top five planets and candidates were ranked by spectroscopy metric in each bin to create a best-in-class sample for each spectroscopy method.
• 3.  
  The 103 unconfirmed TESS-discovered candidates from the transmission and emission spectroscopy grids were vetted using a combination of follow-up observations collected by TFOP and independent analyses such as the SPOC DV reports.
• 4.  
  We used vespa and TRICERATOPS to calculate the FPPs and determine a final disposition for each target.
• 5.  
  Our analysis resulted in 18 validated targets, 29 likely false positives, 31 targets that were marginally validated, and four inconclusive validations. Of our original targets, 21 were independently confirmed over the course of our analysis.
• 6.  
  This final sample represents the best-in-class targets for atmospheric characterization with JWST, and deeper analysis on each target is highly encouraged.

The best-in-class sample presented in this paper is meant to represent an initial look at many of the targets with the potential to yield high-quality spectra from JWST. We hope that this work paves the way for future studies of a similar sort. We highly encourage independent analysis of each target presented here to discern the true nature of each and build a catalog of planets that can reliably provide exquisite atmospheric data from JWST.

We recognize that this best-in-class sample will undoubtedly change over time as new targets supplant previous ones in the JWST observability rankings, as targets are shown to be false positives, or as the orbital and planetary parameters of targets are refined with further observation. Since the date on which we queried the NASA Exoplanet Archive to generate our sample, ∼750 new TESS candidates have been discovered. It is possible that anywhere from five to 10 of these new discoveries could possess TSM or ESM values that place them within the best-in-class sample. These targets primarily appear in the bins containing the largest planets and the hottest planets. As TESS probes fainter stars, new detections are even more biased toward large, hot planets as they possess a sufficient S/N to be detected around faint stars.

This sample may also change based on the assumptions used to generate it. Our analysis calculated the ESM value for planets in all portions of parameter space even though it was originally developed by Kempton et al. (2018) for terrestrial planets. Parameter values baked into the ESM quantity such as the day–night heat redistribution on a planet may be different from what is assumed by our calculations. However, since our rankings of best-in-class targets are relative to other planets and candidates of similar radius and equilibrium temperature, this factor can likely be ignored. Furthermore, the discrete boundaries of our bins may bias our best-in-class sample toward targets at the hot and large edges of their bins, so different binning schemes may change the specific targets that are contained within the best-in-class sample.

Additionally, it is possible that the thermal emission of planets hotter than ∼800 K can be observed with NIR instruments rather than with MIRI, as assumed by our analysis. This would open up access to brighter stars due to the favorable ratio between the flux of the planet's thermal emission and the flux of the star, and would allow for study of a different set of spectral features compared to those available to MIRI. The sample presented here makes parameter cuts for emission spectroscopy based on the performance of MIRI, but a blend of instruments would open up the pool of potential best-in-class targets for the hottest portions of parameter space.

This best-in-class sample may also prove useful for future missions that will study exoplanet atmospheres, such as the upcoming Ariel mission (Tinetti et al. 2018), which will conduct a survey of around one thousand exoplanetary atmospheres. A total of 69 of our best-in-class targets are contained within the Ariel target list described by Edwards & Tinetti (2022). This overlap may grow as both our best-in-class sample and the Ariel target list are updated.

To a first-order approximation, out of 103 total targets originally unconfirmed in our best-in-class sample, 52 of them were either statistically validated, marginally validated and ruled LPs, or were confirmed independently. This suggests that at least ∼50% of the TESS candidates analyzed are true planets, although this value may be higher if any of the targets deemed "potential false positive" or "likely false positive" are actually planets.

This sample also demonstrates the power of TESS to discover planets amenable for atmospheric characterization from which we can learn a great deal about their atmospheric structure and composition. Approximately 57% of the targets in the final best-in-class sample (excluding likely false positives) are TESS discoveries. However, TESS has surprisingly missed the detection of some small planets orbiting small stars, so planet searches beyond TESS are also required (Brady & Bean 2022). It is therefore important to continue searching for planet candidates that could turn out to be excellent targets for atmospheric study since, as shown here, many of the best planets for study with JWST are still being revealed.

We thank T. Komacek, D. Richardson, A. Boss, and C. Hartzell for their helpful discussion of this work.

Funding for the TESS mission is provided by NASA's Science Mission Directorate.

This research has made use of the Exoplanet Follow-up Observation Program (ExoFOP) website (NExScI 2022), which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program.

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.

This paper includes data collected by the TESS mission, which are publicly available from the Mikulski Archive for Space Telescopes (MAST) and produced by the Science Processing Operations Center (SPOC) at NASA Ames Research Center. This research effort made use of systematic error-corrected (PDCSAP) photometry. Funding for the TESS mission is provided by NASA's Science Mission directorate.

Resources supporting this work were provided by the NASA High-End Computing (HEC) Program through the NASA Advanced Supercomputing (NAS) Division at Ames Research Center for the production of the SPOC data products.

Some/all of the data presented in this paper were obtained from the Mikulski Archive for Space Telescopes (MAST) at the Space Telescope Science Institute. The specific observations analyzed can be accessed via 10.17909/rd5r-m387.

This work makes use of observations from the LCOGT network. Part of the LCOGT telescope time was granted by NOIRLab through the Mid-Scale Innovations Program (MSIP). MSIP is funded by NSF.

This paper is based on observations made with the MuSCAT3 instrument, developed by the Astrobiology Center and under financial support by JSPS KAKENHI (grant No. JP18H05439) and JST PRESTO (grant No. JPMJPR1775), at Faulkes Telescope North on Maui, HI, operated by the Las Cumbres Observatory.

This paper makes use of observations made with the MuSCAT2 instrument, developed by the Astrobiology Center, at TCS operated on the island of Tenerife by the IAC in the Spanish Observatorio del Teide.

This paper makes use of data from the MEarth Project, which is a collaboration between Harvard University and the Smithsonian Astrophysical Observatory. The MEarth Project acknowledges funding from the David and Lucile Packard Fellowship for Science and Engineering, the National Science Foundation under grant Nos. AST-0807690, AST-1109468, AST-1616624 and AST-1004488 (Alan T. Waterman Award), the National Aeronautics and Space Administration under grant No. 80NSSC18K0476 issued through the XRP Program, and the John Templeton Foundation.

C.M. would like to gratefully acknowledge the entire Dragonfly Telephoto Array team, and Bob Abraham in particular, for allowing their telescope bright time to be put to use observing exoplanets.

B.J.H. acknowledges support from the Future Investigators in NASA Earth and Space Science and Technology (FINESST) program (grant No. 80NSSC20K1551) and support by NASA under grant No. 80GSFC21M0002.

K.A.C. and C.N.W. acknowledge support from the TESS mission via subaward s3449 from MIT.

This research made use of Lightkurve, a Python package for Kepler and TESS data analysis (Lightkurve Collaboration et al. 2018).

Some of the data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California, and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain.

D.R.C. and C.A.C. acknowledge support from NASA through the XRP grant No. 18-2XRP18_2-0007. C.A.C. acknowledges that this research was carried out at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0004).

This research was carried out in part at the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National Aeronautics and Space Administration (80NM0018D0004).

S.Z. and A.B. acknowledge support from the Israel Ministry of Science and Technology (grant No. 3-18143).

The research leading to these results has received funding from the ARC grant for Concerted Research Actions, financed by the Wallonia-Brussels Federation. TRAPPIST is funded by the Belgian Fund for Scientific Research (Fond National de la Recherche Scientifique, FNRS) under the grant No. PDR T.0120.21. TRAPPIST-North is a project funded by the University of Liege (Belgium), in collaboration with Cadi Ayyad University of Marrakech (Morocco). M.G. is F.R.S.-FNRS Research Director and E.J. is F.R.S.-FNRS Senior Research Associate. The postdoctoral fellowship of K.B. is funded by F.R.S.-FNRS grant No. T.0109.20 and by the Francqui Foundation.

H.P.O.'s contribution has been carried out within the framework of the NCCR PlanetS supported by the Swiss National Science Foundation under grant Nos. 51NF40_182901 and 51NF40_205606.

F.J.P. acknowledges financial support from the grant No. CEX2021-001131-S funded by MCIN/AEI/ 10.13039/501100011033.

A.J. acknowledges support from ANID—Millennium Science Initiative—ICN12_009 and from FONDECYT project 1210718.

Z.L.D. acknowledges the MIT Presidential Fellowship and that this material is based upon work supported by the National Science Foundation Graduate Research Fellowship under grant No. 1745302.

P.R. acknowledges support from the National Science Foundation grant No. 1952545.

Some of the observations in this paper made use of the High-Resolution Imaging instruments 'Alopeke and Zorro, and were obtained under Gemini LLP Proposal Number GN/S-2021A-LP-105. 'Alopeke/Zorro were funded by the NASA Exoplanet Exploration Program and built at the NASA Ames Research Center by Steve B. Howell, Nic Scott, Elliott P. Horch, and Emmett Quigley. Alopeke/Zorro was mounted on the Gemini North/South 8 m telescopes of the international Gemini Observatory, a program of NSF's NOIRLab, which is managed by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the National Science Foundation, on behalf of the Gemini partnership: the National Science Foundation (United States), National Research Council (Canada), Agencia Nacional de Investigación y Desarrollo (Chile), Ministerio de Ciencia, Tecnología e Innovación (Argentina), Ministério da Ciência, Tecnologia, Inovações e Comunicações (Brazil), and Korea Astronomy and Space Science Institute (Republic of Korea).

This work is partly supported by JSPS KAKENHI grant Nos. JP17H04574, JP18H05439, JP21K20376; JST CREST grant No. JPMJCR1761; and Astrobiology Center SATELLITE Research project AB022006.

This article is based on observations made with the MuSCAT2 instrument, developed by ABC, at Telescopio Carlos Sánchez operated on the island of Tenerife by the IAC in the Spanish Observatorio del Teide. This paper is based on observations made with the MuSCAT3 instrument, developed by the Astrobiology Center and under financial supports by JSPS KAKENHI (grant No. JP18H05439) and JST PRESTO (grant No. JPMJPR1775), at Faulkes Telescope North on Maui, HI, operated by the Las Cumbres Observatory.

This publication benefits from the support of the French Community of Belgium in the context of the FRIA Doctoral Grant awarded to M.T.

D.D. acknowledges support from TESS Guest Investigator Program grant Nos. 80NSSC22K1353, 80NSSC22K0185, and 80NSSC23K0769.

A.B. acknowledges the support of M.V. Lomonosov Moscow State University Program of Development.

T.D. was supported in part by the McDonnell Center for the Space Sciences.

V.K. acknowledges support from the youth scientific laboratory project, topic FEUZ-2020-0038.

Facilities: Adams Observatory, ASP, Brierfield Private Observatory, Campo Catino Astronomical Observatory, Catania Astrophysical Observatory, Caucasian Mountain Observatory, CHAT, CROW Observatory, Deep Sky West, Dragonfly, El Sauce, ExTrA, FEROS, Fred L. Whipple Observatory, Gaia, Gemini ('Alopeke, Zorro), George Mason University, HATSouth, Hazelwood Observatory, Keck, Carlson R. Chambliss Astronomical Observatory (CRCAO) at Kutztown University, LCOGT, Lewin Observatory, Lick Observatory, Lookout Observatory, MASTER-Ural, MEarth-S, Mt. Stuart Observatory, MuSCAT, MuSCAT2, MuSCAT3, Observatori Astronòmic de la Universitat de València, Observatori Astronòmic Albanyà Observatorio del Roque de los Muchachos, Observatory de Ca l'Ou, Palomar Observatory, PEST, Privat Observatory Herges-Hallenberg, PvDKO, RCO, SAINT-EX, Salerno University Observatory, Solaris SLR2, SPECULOOS, SUTO-Otivar, TESS, TRAPPIST, University of Louisville telescopes at the University of Southern Queensland's Mt. Kent Observatory and at Mt. Lemmon Observatory, Union College Observatory, Villa '39, VLT, WASP, WCO, Wellesley College Whitin Observatory, WIYN.

Software: AstroImageJ, Astropy (Astropy Collaboration et al. 2013, 2018, 2022), astroquery (Ginsburg et al. 2019), BANZAI (McCully et al. 2018), DEATHSTAR (Ross et al. 2024), Jupyter (Kluyver et al. 2016), Lightkurve (Lightkurve Collaboration et al. 2018), matplotlib (Hunter 2007), numpy (Van Der Walt et al. 2011), pandas (McKinney 2010), TESS Transit Finder (Jensen 2013), Tesscut (Brasseur et al. 2019), TRICERATOPS (Giacalone & Dressing 2020; Giacalone et al. 2021), VESPA (Morton 2012, 2015).

Table 4 shows our final set of best-in-class targets and values for a selection of planetary and stellar parameters as well as their disposition within our sample. Targets that were removed over the course of our analysis are not included.

Table 5 contains information on all of the publicly available TFOP observations that were used in the synthesis of the reconnaissance photometry and spectroscopy dispositions incorporated into our vetting and validation analysis. We note that no photometry or spectroscopy data were used directly, only the synthesized dispositions created for each target by TFOP's SG1 and SG2. Also contained in Table 5 is information on the high-resolution imaging that was used as observational constraints in our statistical-validation analysis. A further overview of how these dispositions and follow-up observations were used can be found in Sections 5 and 6.

It is important to note that not included in this table are observations that are not publicly available on ExoFOP (see footnote 127). This may be because the observations area currently within a proprietary period, contained within a private archive (e.g., an archive accessible only to members of a specific collaboration), or the observing team decided not to post their observations publicly. However, there was a subset of observations which fall under this category that were communicated to the TFOP subgroup leads for use in the synthesis of dispositions and were therefore indirectly utilized by our analysis, but cannot be listed here as they were not made public on ExoFOP. These include observations from HARPS-N (TOI-261.01, TOI-1683.01, TOI-5082.01), PFS (TOI-261.01), MINERVA-Australis (TOI-261.01), CHIRON (TOI-1895.01), the Network of Robotic Echelle Spectrographs (NRES; TOI-3353.01, TOI-5082.01), CORALIE (TOI-3353.01), and Keck/HIRES (TOI-1683.01). We direct the reader to these teams for further information on the observations obtained by these instruments that are not publicly available on ExoFOP.

Furthermore, not every observation that is currently posted on ExoFOP was utilized in the synthesis of TFOP dispositions for each target. This is because an observer needs to submit their observations to the respective subgroup in an opt-in fashion for the observations to be used in the synthesis of TFOP dispositions.

In this section, we include longer descriptions of some of the observatories used to obtain follow-up observations that were incorporated into our analysis. For descriptions on how the follow-up observations were used in our vetting and validation analysis of the unconfirmed TOIs in our sample, see Sections 5 and 6.

C.1. Ground-based Photometry

C.2. Reconnaissance Spectroscopy