Four microlensing giant planets detected through signals produced by minor-image perturbations

In contrast to the simplified representation of a planetary microlensing signal as a brief, discontinuous deviation in the smooth lensing light curve caused by the host star of the planet, the manifestations of planets exhibit a high degree of variability. These signals arise when the source crosses or approaches the caustic generated by the presence of a planet (Mao & Paczyński, 1991; Gould & Loeb, 1992). Caustics in microlensing represent the positions at which the magnification of a point source becomes infinitely large. Caustics induced by planets form single or multiple closed curves. The characteristics of these caustic curves, including their number, location, and size, vary depending on the separation and mass ratio between the planet and its host star. Combined with the varied trajectories of source stars, planetary signals exhibit diverse forms in terms of their location, duration, and shape. Consequently, simply depicting planetary signals as brief deviation often fails to capture their true complexity.

Identifying microlensing planets involves a meticulous procedure that entails considerable time and effort. Initially, anomalous events are sought by scrutinizing the light curves of lensing events. Subsequently, the planetary nature of these anomalies is discerned through rigorous analyses of the observed light curves. Current lensing surveys annually detect more than 3000 events, with roughly 10% exhibiting anomalies attributed to various causes. While some anomalies can be readily attributed to planetary presence, discerning the nature of others from their appearance alone is very challenging. To firmly identify anomalies, a detailed analysis involving complex procedures and extensive computations is necessary. Morphological studies play a crucial role by categorizing anomalies with similar characteristics and investigating the origins of each class. This approach not only aids in accurately characterizing lensing anomalies for future events with similar structures but also facilitates the early diagnosis of anomalies before conducting in-depth analyses.

As the count of microlensing planets rises, planets with signals exhibiting similar anomaly patterns are grouped and announced collectively. Han et al. (2017) and Poleski et al. (2017) provided illustrative instances of planetary signals emerging via a recurrent channel, as demonstrated in their analysis of the microlensing planets OGLE-2016-BLG-0263Lb and MOA-2012-BLG-006Lb, respectively. Jung et al. (2021) presented planetary signals observed in the lensing events OGLE-2018-BLG-0567 and OGLE-2018-BLG-0962, for which the planetary signals appeared on the sides of the lensing light curves due to the source stars’ crossings over caustics situated away from the planet hosts. Additionally, Han et al. (2024) introduced three microlensing planets – MOA-2022-BLG-563Lb, KMT-2023-BLG-0469Lb, and KMT-2023-BLG-0735Lb – whose signals exhibit consistent short-term dip features surrounded by weak bumps on both sides of the dip. Han et al. (2023) and Han et al. (2021a) exemplified weak short-term planetary signals generated without caustic crossings for the events KMT-2022-BLG-0475, KMT-2022-BLG-1480, KMT-2018-BLG-1976, KMT-2018-BLG-1996, and OGLE-2019-BLG-0954.

In this study, we present analyses of four planetary lensing events: KMT-2020-BLG-0757, KMT-2022-BLG-0732, KMT-2022-BLG-1787, and KMT-2022-BLG-1852. These events exhibit planetary signals with a common characteristic, originating from the source crossing over the ”planetary” caustic induced by ”close” planets, followed by source passage through the region of the ”minor-image” perturbation situated between the pair of planetary caustics. We elucidate the technical terms ”close,” ”minor-image,” ”planetary,” and ”central” caustics in the subsequent section.

The analyses of the planetary events are presented according to the following organization. In Sect. 2, we provide an overview of the observations conducted for the lensing events, including the instrumentation utilized for observations as well as the procedures implemented for data reduction and adjustment of error bars. In Sect. 3, we provide a brief overview on the fundamental principles of planetary microlensing and discuss the modeling process used to analyze the observed light curves of the lensing events. Subsequent subsections offer comprehensive analyses of individual events and their results. In Sect. 4, we examine the source stars associated with the events and estimate the angular Einstein radii of the events. In Sect. 5, we depict the Bayesian analyses conducted for each event and present estimates for the physical parameters of the planetary systems derived from these analyses. Finally, in Sect. 6, we summarize our findings and draw conclusions based on the results obtained.

All analyzed lensing events in this work were discovered from the microlensing survey conducted toward the Galactic bulge field by the Korea Microlensing Telescope Network (KMTNet: Kim et al., 2016). In Table 1, we present the equatorial and Galactic coordinates of the events, along with their baseline magnitudes ($I_{\rm base}$) and the $I$-band extinction ($A_{I}$) toward the fields. We investigated the availability of additional data from other lensing surveys and found that KMT-2020-BLG-0757 was also observed by the Microlensing Observations in Astrophysics (MOA: Bond et al., 2001) group, who designated the event as MOA-2020-BLG-249. For the analysis of this event, we utilized the combined data from both the KMTNet and MOA surveys. After analyzing the KMT-2020-BLG-0757 event, it was discovered that additional data had been obtained by the OMEGA group, who conducted follow-up of microlensing events in the entire sky. Our analysis incorporates these newly acquired data.

The KMTNet group utilizes a network of three telescopes that are strategically distributed in three locations of the Southern Hemisphere: at the Cerro Tololo Inter-American Observatory in Chile (KMTC), the South African Astronomical Observatory in South Africa (KMTS), and the Siding Spring Observatory in Australia (KMTA). These telescopes have identical specifications, featuring a 1.6 m aperture and are each equipped with a camera providing a field of view of 4 square degrees. The MOA group conducts its survey using the 1.8 m telescope lying at the Mt. John Observatory located in New Zealand. The MOA telescope is mounted by a camera providing 2.2 square degrees of field of view. Observations by the KMTNet and MOA surveys were mainly done in the $I$ and the customized MOA-$R$ bands, respectively. For both surveys, a fraction of images were taken in the $V$ band for the source color measurement. Observational cadences vary depending on the events and we will mention the cadence when we detail the analysis of each event. The supplementary observations of KMT-2020-BLG-0757 by the OMEGA group were conducted using the Las Cumbres Observatory Global Telescope (LCOGT) Network, which comprises multiple 1-meter telescopes (Brown et al., 2013).

Image reduction and photometry for the lensing events employed automated pipelines tailored to each survey. These pipelines utilized code developed by Albrow et al. (2017) for the KMTNet survey and by Bond et al. (2001) for the MOA survey. To ensure optimal data usage in the analysis, we performed an additional reduction of the KMTNet data employing the photometry code developed by Yang et al. (2024). The OMEGA data were processed using LCO’s BANZAI pipeline (McCully et al., 2018). Following the refinement of the data sets for each event, we adjusted the data error bars. This adjustment aimed not only to maintain consistency with the data scatter but also to ensure that the $\chi^{2}$ value per degree of freedom (dof) was set to unity for each dataset. The normalization process followed the procedure described in Yee et al. (2012).

Caustics induced by planets are classified into two types based on whether the planet is located within or outside the Einstein ring of the host star (Dominik, 1999). We use the notation $s$ to denote the normalized projected separation between the planet and its host, measured in units of the Einstein radius. Both close ($s<1$) and wide ($s>1$) planets generate two sets of caustics: one set near the host star (central caustic) and the other set located away from the host at approximately ${\bf s}-1/{\bf s}$ (planetary caustic). In cases where the planet-to-host mass ratio $q$ is very small and the planetary separation $s$ deviates from unity, the central caustics induced by close and wide planets closely resemble each other both in shape and size (Griest & Safizadeh, 1998). Conversely, the planetary caustics induced by these two types of planets differ from each other in various aspects: a wide planet generates a single four-cusp caustic on the planet side, while a close planet produces two sets of three-cusp caustics on the opposite side of the planet. As a result, the planetary signals arising from the planetary caustics of these two lens populations exhibit characteristics that can nearly always be distinguished. For an in-depth discussion on the properties of central and planetary caustics, refer to Chung et al. (2005) and Han (2006), respectively.

The planets discussed in this study share common characteristics stemming from perturbations of the minor image caused by their presence. When a source is gravitationally lensed by a single mass, it produces two images: the brighter one, known as the ”major image,” appears outside the Einstein ring, while the fainter image, referred to as the ”minor image,” lies within the Einstein ring (Gaudi & Gould, 1997). Perturbations to the primary image by a planet result in additional magnification due to the planet’s influence, leading to a positive deviation in the anomaly. Conversely, perturbations to the minor image by the planet lead to demagnification, resulting in negative deviations in the anomaly.

In Figure 1, we illustrate the magnification pattern of a lens system to show the emergence of caustics and the de-magnification region due to the presence of a close planet. The planet parameters of the lens system corresponding to the presented configuration are $(s,q)=(0.8,3\times 10^{-3})$, where $q$ denotes the mass ratio between the planet and its host. Noteworthy is the expansive span of the de-magnification region, which extends between the central and planetary caustics. This implies that the perturbation resulting from the traversal of a source through this area may persist for an extended period.

Assuming a rectilinear relative motion between the lens and the source, a planet-induced anomaly in a lensing light curve is characterized by seven fundamental lensing parameters. Among these, the first three parameters $(t_{0},u_{0},t_{\rm E})$ describe the approach of the lens and the source. Specifically, they represent the time of the closest lens-source approach, the separation (scaled to the angular Einstein radius $\theta_{\rm E}$) at $t_{0}$, and the event timescale, respectively. The subsequent two parameters $(s,q)$ define the planetary lens system, indicating the normalized separation and mass ratio between the planet and host. The next parameter $\alpha$ represents the incidence angle of the source relative to the planet-host axis. As the source crosses the caustic induced by the planet, the lensing magnifications are affected by finite-source effects. To incorporate these effects, an additional parameter, $\rho$, is necessary. This parameter is defined as the angular source radius normalized to $\theta_{\rm E}$ (normalized source radius), that is, $\rho=\theta_{*}/\theta_{\rm E}$.

For each lensing event, we undertake an analysis to determine a lensing solution, which comprises the set of lensing parameters that best characterize the observed anomaly. Initially, we search for the planet parameters $(s,q)$ using a grid approach with multiple initial values of $\alpha$, followed by finding the remaining parameters using a downhill approach based on the Markov chain Monte Carlo (MCMC) method. Subsequently, we construct a $\Delta\chi^{2}$ map on the parameter plane of $\log s$–$\log q$ to identify local solutions. In the subsequent stage, we refine the local solutions, and then establish a global solution by comparing the $\chi^{2}$ values of the local solutions. If the fits of the local solutions are comparable, we present all degenerate solutions and investigate the origin of the degeneracy. In the subsequent subsections, we provide detailed descriptions of the analyses conducted for the individual events.

In this section, we provide details regarding the source stars involved in the individual lensing events. With the specified source type, we estimated the angular Einstein radius and relative lens-proper motion using the relations:

where $\theta_{*}$ represents the angular radius of the source. The normalized source radius $\rho$ was derived by examining the section of the light curve corresponding to caustic crossings, whereas the angular source radius $\theta_{*}$ was inferred from the type of the source. Given that all events exhibited anomalies involving caustic crossings, we were able to constrain the normalized source radii and subsequently the Einstein radii.

We determined the source type employing the methodology outlined in Yoo et al. (2004). According to this method, we first measured the instrumental color and magnitude, $(V-I,I)_{\rm S}$, of the source, and placed the source in the color-magnitude diagram (CMD) of stars lying near the source. The $I$- and $V$-band source magnitudes were determined by fitting the light curves of the corresponding bands to the model. This process utilized the data processed with the pyDIA code (Albrow et al., 2017). We then calibrated the color and magnitude using the centroid of the red giant clump (RGC) in the CMD as a reference, that is,

Here $(V-I,I)_{{\rm S},0}$ and $(V-I,I)_{{\rm RGC},0}$ respectively represent the reddening and extinction-corrected colors and magnitudes of the source and RGC centroid, and the term $\Delta(V-I,I)=(V-I,I)_{\rm S}-(V-I)_{\rm RGC}$ indicates the offsets in color and magnitude between the source and RGC centroid. The RGC centroid can be used as a reference for calibration because its de-reddened color and magnitude are known from Bensby et al. (2013) and Nataf et al. (2013), respectively.

Figure 10 shows the source stars and RGC centroids in the CMDs for the lensing events. Table 6 presents the values of $(V-I,I)_{\rm S}$, $(V-I,I)_{\rm RGC}$, $(V-I,I)_{{\rm RGC},0}$, and $(V-I,I)_{{\rm S},0}$ for the events, along with the corresponding spectral types of the source stars. The spectral types inferred from the measured colors and magnitudes are F9V for KMT-2020-BLG-0757, K3.5III for KMT-2022-BLG-0732 and KMT-2022-BLG-1787, and K2.5 subgiant for KMT-2022-BLG-1852.

The angular source radius was determined based on the measured source color and magnitude. For this determination, we initially converted the $V-I$ color to $V-K$ color using the color-color relation provided by Bessell & Brett (1988). Subsequently, we utilized the $(V-K,I)$–$\theta_{*}$ relationship established by Kervella et al. (2004) to infer the angular source radius. With the determined angular source radius, the Einstein radius and proper motion were calculated using the relations in Eq. (1). We present the estimated values of $\theta_{*}$, $\theta_{\rm E}$ and $\mu$ in Table 6. We further checked the information on the source stars in the Gaia catalog (Gaia Collaboration et al., 2018). For KMT-2022-BLG-0732 and KMT-2022-BLG-1787, we identified the source stars, while those of KMT-2020-BLG-0757 and KMT-2022-BLG-1852 were not registered in the catalog. Even for the events with identified source stars, only the $G$-band magnitudes were available, with no information specifying the spectral types of the source stars.

In this section, we derive estimates for the mass $M$ and distance $D_{\rm L}$ of the discovered planetary systems. These physical lens parameters were derived from the lensing observables, namely the event timescale and the angular Einstein radius. These observables can provide constraints on the physical parameters because they are connected to $M$ and $D_{\rm L}$ through the relationships represented as:

Here $\kappa=4G/(c^{2}{\rm AU})\simeq 8.14\leavevmode\nobreak\ {\rm mas}/M_{\odot}$, $\pi_{\rm rel}=\pi_{\rm L}-\pi_{\rm S}={\rm AU}(1/D_{\rm L}-1/D_{\rm S})$ represents the relative lens-source parallax, and $D_{\rm S}$ denotes the source distance. The lens mass and distance can be uniquely determined by measuring an additional lensing observable of the microlens parallax $\pi_{\rm E}$ through the relations given by Gould (2000) as:

The microlens-parallax vector is defined as $\mbox{\boldmath$\pi$}_{\rm E}=(\pi_{\rm rel}/\theta_{\rm E})(\mbox{\boldmath$\mu$}/\mu)$, and its value can be determined by observing the subtle deformation of a lensing light curve, resulting from the deviation of the relative lens-source motion from rectilinear, caused by the orbital motion of Earth around the Sun: microlens-parallax effects (Gould, 1992). Because none of the events had a measured microlens parallax, we employed Bayesian analysis to estimate the lens parameters. This analysis incorporates the constraints provided by the measured lensing observables $t_{\rm E}$ and $\theta_{\rm E}$ together with priors of a Galaxy model and a mass function of objects within the Galaxy.

We carried out the Bayesian analysis according to the following procedure. In the initial stage, we conducted a Monte Carlo simulation to generate a large number of synthetic lensing events. Within this simulation, we extracted the distances to the lens and source, along with their relative proper motion from a Galactic model. Additionally, we derived the lens mass from a mass function model. Specifically, we employed the Galactic model outlined in Jung et al. (2021) to derive $D_{\rm L}$, $D_{\rm S}$, and $\mu$, and adopted the mass function described in Jung et al. (2022). In the subsequent stage, we computed the lensing observables corresponding to the lens and source parameters utilizing the relationships in Eq. (4). Subsequently, we formed posterior distributions for $M$ and $D_{\rm L}$ by assigning a weight to each synthetic event. This weight was determined by the expression:

where $(t_{{\rm E},i},\theta_{{\rm E},i})$ represent the time scale and Einstein radius of each synthetic event, $(t_{\rm E},\theta_{\rm E})$ denote the measured values, and $[\sigma(t_{\rm E}),\sigma(\theta_{\rm E})]$ are their uncertainties.

Figures 11 and 12 depict the posterior probability distributions for the lens masses and distances associated with the events. Table 7 provides a summary of the estimated masses of the host ($M_{\rm host}$) and planet ($M_{\rm planet}$), along with the distance and projected separation ($a_{\perp}$) between the planet and its host. For each parameter, the median value is presented as the central representative value, with the lower and upper limits determined as the 16% and 84% of the posterior distribution, respectively.

The identified planetary systems exhibit several common traits. Firstly, all planets orbit low-mass host stars, which are notably less massive than our Sun. The range of host star masses falls between $0.32$ and $0.58$ times the mass of the Sun. Secondly, the planets themselves are giants, exceeding the mass of Jupiter in our solar system within a range of $1.1$ to $10.7$ times the mass of Jupiter. Finally, all these planets are situated well beyond the ice line of their host stars, classifying them as ice giants.

Each panel of the posterior distributions separates the probability contributions from the disk and bulge lens populations using blue and red curves, respectively. The combined contribution is shown by the black curve. Table 7 details the probabilities for the disk ($p_{\rm disk}$) and bulge ($p_{\rm bulge}$) populations. Notably, for KMT-2022-BLG-0732, the lens is more likely situated within the disk with a probability $p_{\rm disk}\sim 69\%$. Conversely, the remaining events exhibit a higher probability of residing in the bulge with probabilities $p_{\rm bulge}\gtrsim 64\%$.

We conducted analyses of four microlensing events KMT-2020-BLG-0757, KMT-2022-BLG-0732, KMT-2022-BLG-1787, and KMT-2022-BLG-1852, for which the light curves commonly exhibit positive deviations and subsequent negative deviations. Unlike the usual brief anomalies observed in typical planetary microlensing events, the deviations in these events extend over a significant portion of the light curves. This prolonged deviation poses challenges in promptly identifying the presence of planets from the anomalies.

Our analysis revealed that each event’s anomaly was caused by a planetary companion situated within the Einstein ring of the primary star. The positive deviation of the anomaly was generated as the source traversed one of the planetary caustics induced by a close planet, while the negative deviation occurred as the source passed through the extended region of minor-image perturbations.

Upon estimating the physical parameters using the measured lensing observables of the events, we found several common features among the identified planetary systems. First, all the planets orbit low-mass host stars, significantly less massive than our Sun, ranging from 0.32 to 0.58 solar masses. Second, these planets themselves are classified as giants, exceeding the mass of Jupiter in our solar system, with masses ranging from 1.1 to 10.7 times Jupiter’s mass. Finally, all the planets reside well beyond the ice line of their host stars, making them ice giants.