Optical IFU observations of GOALS sample with KOOLS-IFU on Seimei Telescope: Initial results of nine U/LIRGs at z < 0.04

Abstract

1 Introduction

In the last two decades, galaxy mergers have been recognized as a key phenomenon for understanding galaxy evolution (e.g., Conselice 2014, and references therein). Star formation (SF) and/or active galactic nucleus (AGN) activity in galaxies are often triggered by galaxy mergers,¹ which enhance the inflow of material on a galactic scale into the close environments of the nuclear region. The galaxy merger also enhances infrared (IR) luminosity (L_IR) (e.g., Sanders et al. 1988; Veilleux et al. 2002; Imanishi et al. 2007; Imanishi 2009; Toba et al. 2015, 2017a; Toba & Nagao 2016).² These are observed as luminous IR galaxies (LIRGs) and ultraluminous IR galaxies (ULIRGs) with L_IR greater than 10¹¹ and 10¹² L_⊙, respectively (Sanders & Mirabel 1996). Recent works also suggested that the relative strength of SF and AGN activity in U/LIRGs may be varied in the course of galaxy mergers (e.g., Narayanan et al. 2010; Ricci et al. 2017; Blecha et al. 2018; Yamada et al. 2019, 2021). In particular, the radiation from an AGN is expected to interact with the interstellar medium (ISM) and lead to the ejection or heating of the gas, which could be tightly associated with the growth of the galaxy and its supermassive black hole (SMBH) (e.g., Cazzoli et al. 2016; Toba et al. 2017b; Harrison et al. 2018; Chen et al. 2019; Finnerty et al. 2020; Jun et al. 2020; Fluetsch et al. 2021). That AGN-driven outflow is often observed in U/LIRGs: recent observations reported that a dusty, IR luminous AGN often shows a strong ionized gas outflow, and its strength seems to be correlated with L_IR (Bischetti et al. 2017; Toba et al. 2017c; Chen et al. 2019; Jun et al. 2020). Therefore, U/LIRGs are a good laboratory for examining how an AGN-driven outflow could influence the host galaxy and its SMBH. However, those works are often limited to handling a whole galaxy as a system. To address, from a spatially resolved point of view, how the SF and AGN activity could be enhanced/quenched as a function of merger sequence, Integral Field Unit (IFU) observation is quite a powerful tool that could trace the gas kinematics and energetics (e.g., Bae et al. 2017; Shin et al. 2019; Pan et al. 2019).

In this work, we focus on the Great Observatories All-sky LIRG Survey (GOALS; Armus et al. 2009) that provides a complete, flux-limited (i.e., flux density at 60 μm > 5.24 Jy) U/LIRG sample in the local universe. In addition to their multi-wavelength follow-up observations, Stierwalt et al. (2013) divided the GOALS U/LIRGs into sub-samples regarding the merger stage, based on a visual inspection of the IRAC/Spitzer 3.6 μm images and/or higher resolution images taken by e.g., Hubble Space Telescope (see also Haan et al. 2011). Therefore, the GOALS sample is an ideal laboratory for investigating the role of galaxy mergers in the co-evolution of galaxies and SMBHs.

In this paper, we present the initial results of our follow-up campaign of the GOALS sample with an optical IFU, the Kyoto Okayama Optical Low-dispersion Spectrograph with optical-fiber IFU (KOOLS-IFU; Yoshida 2005; Matsubayashi et al. 2019) on the Okayama 3.8 m Seimei Telescope (Kurita et al. 2020). The structure of this paper is as follows. Section 2 describes the sample selection, observations, and the data reduction of observed targets. In section 3, we present the results of our optical IFU observations and discuss the dependence of ionized gas outflow on distance from the galaxy center and merger stage. We summarize the results of this work in section 4. Throughout this paper, the adopted cosmology is a flat universe with H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_M = 0.28, and Ω_Λ = 0.72, which are the same values as those adopted in a series of GOALS papers that are relevant to this work (e.g., Armus et al. 2009; Rich et al. 2015).

2 Data and analysis

2.1 Target selection

The observed targets were drawn from the GOALS sample, which is itself a subset of the IRAS Revised Bright Galaxy Sample (Sanders et al. 2003), providing a 60 μm flux-limited sample of U/LIRGs at z < 0.088. The GOALS sample has been extensively examined by multi-wavelength observations from X-ray (e.g., Iwasawa et al. 2011; Ricci et al. 2017, 2021; Torres-Albà et al. 2018; Yamada et al. 2020, 2021), ultraviolet (UV; e.g., Howell et al. 2010; Petty et al. 2014), optical–near-IR (NIR; e.g., Haan et al. 2011; Kim et al. 2013; Linden et al. 2017; Jin et al. 2019; Larson et al. 2020), mid–far-IR (MIR–FIR; e.g., Inami et al. 2013, 2018; Stierwalt et al. 2013, 2014; Chu et al. 2017; Díaz-Santos et al. 2017; Yamada et al. 2019; Armus et al. 2020) to radio (e.g., Privon et al. 2015; Barcos-Muñoz et al. 2017; Yamashita et al. 2017; Herrero-Illana et al. 2019; Linden et al. 2019; Condon et al. 2021) [see also Casey (2012) and U et al. (2012) for SED analysis]. We employed the merger classifications provided by Stierwalt et al. (2013), which have five categories from N (no signs of merger activity or massive neighbors), to stage A (galaxy pairs prior to a first encounter), B (post-first encounter with galaxy disks that are still symmetric and intact but with signs of tidal tails), C (showing amorphous disks, tidal tails, and other signs of merger activity), and D (two nuclei in a common envelope), as a sequence of merger stages [see subsection 2.5 in Stierwalt et al. (2013) for more detail, and also, e.g., figure 1 in Ricci et al. (2017) for an example image of each merger stage].

We selected nine objects as the target by taking into account the magnitude in optical bands (r mag < 16.0), merger stage, previous IFU observations, and visibility from the Seimei Telescope. The basic information of observed targets is summarized in table 1. Figure 1 shows distributions of redshift, IR luminosity, and merger stage of our sample. We also present those distributions for the complete GOALS sample (Armus et al. 2009) and sub-sample reported in Rich et al. (2015), who performed IFU observations of 27 southern U/LIRGs from the GOALS sample with the Wide Field Spectrograph (WiFeS) installed on a 2.3 m telescope that is similar in capability to KOOLS-IFU. We confirm in figure 1 that our sample is widely distributed in terms of IR luminosity and merger stage while our sample is biased toward low-z (z < 0.04). Figure 2 shows i-band images of our sample taken by Pan-STARRS (Chambers et al. 2016).³

Fig. 1.

Histograms of (a) redshift, (b) IR luminosity, and (c) merger stage of the GOALS sample (black/gray), GOALS sample with WiFeS provided in Rich et al. (2015) (blue), and GOALS sample with KOOLS-IFU, i.e., our sample (red).

Open in new tabDownload slide

Fig. 2.

Sample's i-band postage stamps (60″ × 60″) taken by Pan-STARRS. The footprint of the KOOLS-IFU with 127 fibers is overlaid as red circles. RA and Dec are relative coordinates with respect to those in table 1 in which (ΔRA, ΔDec) = (0, 0) also corresponds to the center of the field of view for KOOLS-IFU (i.e., fiber ID = 37, see figure 3). North is up and the east is left in all images.

Open in new tabDownload slide

We note that the WiFeS-GOALS IFU survey (Rich et al. 2011, 2012, 2015), the VLT-VIMOS IFU survey (Arribas et al. 2008), the PMAS IFU survey (Alonso-Herrero et al. 2009), the INTEGRAL IFU survey (García-Marín et al. 2009), the VLT-SINFONI IFU survey (Piqueras López et al. 2012), Physics of ULIRGs with MUSE and ALMA (PUMA; Perna et al. 2021), and the Keck OSIRIS AO LIRG Analysis (KOALA; U et al. 2019) observed (a part of) the GOALS sample (see also García-Marín et al. 2006; Rodríguez-Zaurín et al. 2011; Bellocchi et al. 2013; Arribas et al. 2014; Kakkad et al. 2022; Perna et al. 2022). In particular, the WiFeS-GOALS IFU survey provides specially resolved maps for emission-line ratios based on various emission lines, such as Hβ, [O iii]λ5007, Hα, [N ii]λλ6549, 6583, and [S ii]λλ6716, 6731 that are also targeted lines in this work (see subsection 3.2). The sample in this survey is observable from the southern hemisphere, and thus our sample (which is observable from the northern hemisphere) is complementary to those reported in Rich et al. (2015). The remaining projects basically focused on one emission line (e.g., Hα), and spatially resolved emission-line ratios were not well investigated.⁴

2.2 Observations and data reduction

The targets were observed with the KOOLS-IFU on the Seimei Telescope in the 2019B, 2020A, and 2020B semesters (PI: Y. Toba; proposal IDs = 19B-N-CN01, 19B-K-0001, 20A-N-CN01, and 20B-K-0004). The Seimei is a new 3.8 m diameter optical–IR alt-azimuth mount telescope located at Okayama Observatory, Kyoto University, Okayama prefecture in Japan, where the typical seeing on-site is |${1{^{\prime \prime}_{.}}2}$|–|${1{^{\prime \prime}_{.}}4}$|⁠. The science operation was started in 2019. The KOOLS-IFU consists of 127 fibers with a total field of view (FoV) of ∼15″ in diameter.⁵ Each fiber is assigned a unique fiber ID as shown in figure 3. The spatial sampling is |$\sim {1{^{\prime \prime}_{.}}2}$| per fiber, and the 10σ limiting magnitude is 17–18 AB mag given 10 min of exposure. We used the VPH 495 and VPH 683 grisms among four grisms equipped with the KOOLS-IFU.⁶ The wavelength coverage and the spectral resolution (R = λ/Δλ) of VPH 495 and VPH 683 are 4300–5900 Å and R ∼ 1500 and 5800–8000 Å and R ∼ 2000, respectively.⁷ The observational log is summarized in table 2.

Fig. 3.

Schematic view of KOOLS-IFU with fiber ID.

Open in new tabDownload slide

The data reduction was executed with dedicated software in which the Image Reduction and Analysis Facility (IRAF; Tody 1986, 1993) tasks are utilized through the Python package (PyRAF; Science Software Branch at STScI 2012).⁸ In particular, the IRAF imred.hydra package (Barden et al. 1994; Barden & Armandroff 1995) was mainly used for spectrum extraction, flat-fielding, and wavelength calibration in a standard manner. Following Matsubayashi et al. (2019), spectrum extraction was done by measuring the center position of a spectrum at each pixel based on dome flat frames in which we summed the counts within ±2.5 pixels of the center position. The wavelength was calibrated with arc (Hg and Ne) lamp frames. The flux calibration was carried out by using an observed spectrum of a standard star (see table 2). Since we utilized all fibers of KOOLS-IFU for targets, we took sky frames separately for sky subtractions.

2.3 Spectral fitting

In order to measure the properties of emission lines such as line flux and line width for our GOALS sample, we conducted a spectral fitting to the reduced spectra using the Quasar Spectral Fitting package (QSFit v1.3.0; Calderone et al. 2017).⁹ Following Toba et al. (2021b), we fit KOOLS-IFU spectra with the following five components; (i) an AGN continuum with a single power law, (ii) Balmer continuum modeled by Grandi (1982) and Dietrich et al. (2002), (iii) host galaxy component taken from an empirical SED template in which a Spiral galaxy template (Polletta et al. 2007) is employed,¹⁰ (iv) iron blended emission lines with UV–optical templates (Vestergaard & Wilkes 2001; Véron-Cetty et al. 2004), and (v) emission lines with Gaussian components. QSFit allows us to fit all the components simultaneously based on a Levenberg–Marquardt least-squares minimization algorithm with MPFIT (Markwardt 2009) procedure.

Since QSFit is optimized to optical spectra taken by the SDSS (e.g., Calderone et al. 2017; Toba et al. 2021a, 2021b), we modified the code for general purpose, which allows us to do the spectral fitting to KOOLS-IFU data. We fit the Balmer lines with narrow and broad components in which the FWHM of narrow and broad components is constrained in the ranges of 100–1000 km s⁻¹ and 900–15000 km s⁻¹, respectively, to allow a good decomposition of the line profile, in the same manner as Toba et al. (2021b). Galactic extinction is corrected according to Schlafly and Finkbeiner (2011). Since one of the purposes of this work is to measure the kinematics of an ionized gas outflow, we also account for a blue-wing component of [O iii]λλ4959, 5007 lines. Following Calderone et al. (2017), the FWHM of the broad blue-wing component is constrained to be larger than the narrow component in the range of 100–1000 km s⁻¹ while the broad component is allowed to be blueshifted up to 2000 km s⁻¹. The velocity offsets of [O iii]λ4959 and [O iii]λ5007 are tied.¹¹ The measurement errors are estimated from the Monte Carlo resampling method in which we adopted the 1σ dispersion of each value by measuring them 100 times for spectra while randomly adding the noise in the same manner as in Toba et al. (2017b) [see appendix B in Calderone et al. (2017) for a full explanation of this procedure].

3 Results and discussion

3.1 Result of the spectral fitting

Figure 4 shows examples of the optical spectral fitting with QSFit. We confirm that about |$72\%$| of the fibers are successfully fitted, of which |$\sim\!\! 90\%$| have reduced χ² < 1.2. This indicates that the spectra of our sample taken by KOOLS-IFU are well fitted with QSFit. The failure of the fitting is mainly due to the low signal-to-noise ratio (SN) of spectra.

Fig. 4.

Examples of spectral fitting for our objects in fiber with VPH 495 (top) and VPH 683 (bottom). The black lines are observed data. The blue and yellow curves show the narrow and broad components of emission lines, respectively. The best-fitting spectrum is shown with red lines. The vertical green dashed lines correspond to the rest-frame wavelengths for Hβ, [O iii]λλ4959,5007, [N ii]λλ6549,6583, Hα, and [S ii]λλ6716,6731 lines.

Open in new tabDownload slide

3.2 Line luminosity maps and line ratio diagnostics

Figure 5 shows maps for line luminosity in units of erg s⁻¹. Luminosity is the sum of the broad and narrow emission line components as long as a broad component is detected. Otherwise, a narrow component of the emission line is used for estimating the line luminosity. Since each spectrum in a fiber is fitted 100 times via the Monte Carlo realization (see subsection 2.3), their weighted mean is used. Line luminosity is successfully estimated for over |$95\%$| of fibers for most objects. We note that we did not correct for internal dust extinction because it is quite hard to estimate the amount of attenuation through the Balmer decrement in all fibers due to poor SN of Hα and/or Hβ line in some fibers (see figure 5). However, this does not affect estimating the outflow properties that are relevant to the conclusion of this work (see subsection 3.3).

Figure 6 shows a spatially resolved Baldwin, Phillips, and Terlevich (BPT) diagram (Baldwin et al. 1981) in which each fiber position is classified as Seyfert 2 (Sy2), low-ionization nuclear emission-line region (LINER), composite, star-forming region (SF), and Unknown. We employed diagnostics with line ratios of [N ii]λ6583|$/$|Hα and [O iii]λ5007|$/$|Hβ provided by Kauffmann et al. (2003) and Kewley et al. (2006) (see also Toba et al. 2013, 2014), in which the narrow component of each emission line is used. If we define a fiber with Sy2 or composite classification as AGN-dominant, NGC 1614, CGCG 468-002W, Mrk 273, and NGC 7674 are AGN-dominated with |$50\%$|–|$70\%$| of fibers being classified as Sy2/Composite. For NGC 3690 West and East (which is classified as merger stage C), about |$50\%$| and |$30\%$| of fibers are classified as AGNs, respectively, which suggests that NGC 3690 is basically SF-dominated, but the AGN contribution to the western part is larger than that of the eastern part for NGC 3690. Indeed, Alonso-Herrero et al. (2000) reported on an interaction-induced star formation for this system. Hard X-ray observations also supported the presence of AGNs in NGC 3690 West and the absence of (or very faint) AGNs in NGC 3690 East (Yamada et al. 2021). A caution here is that an obscured AGN often lacks the signature of AGN emission from narrow-line regions (NLRs) in optical spectra because NLRs are not well developed, and hence those are even outside of the AGN region in the BPT diagram (Hickox & Alexander 2018, and references therein), which makes the diagnostics with line ratios difficult.

Fig. 5.

Maps for several line luminosities (Hβ, [O iii]λ5007, Hα, [N ii]λλ6549,6583, and [S ii]λλ6716,6731, from left to right) for our sample. Gray circles denote fibers for which line luminosity could not be estimated due to poor SN. North is up and the east is left in all images.

Open in new tabDownload slide

Fig. 6.

Spatially resolved BPT diagram for our sample. Red, green, yellow, and blue circles correspond to Sy2, LINER, composite, and SF, respectively. Gray circles denote “Unknown” fibers for which line ratio diagnostics could not be executed because either or both lines for line ratio were undetected. North is up and the east is left in all images.

Open in new tabDownload slide

3.3 Outflow properties as a function of distance from the galaxy center

We then investigate how σ₀ could depend on the distance from the galaxy center (i.e., IFU center). In order to obtain reliable σ₀ with a relatively small uncertainty, we estimate σ₀ in the following three annular areas with an inner–outer radius of (i) 0–1 kpc, (ii) 1–2 kpc, and (iii) 2–5 kpc. Figure 7 shows the resultant σ₀ as a function of galaxy center for various merger stages. The only object classified as merger stage “B,” CGCG 468−002W, is shown only in the innermost area because σ₀ could not be measured in the outer regions due to poor SN.¹² We find a negative correlation between distance from the center and σ₀ regardless of the merger stage, with a coefficient of r ∼ −0.4 in which uncertainties of σ₀ are taken into account (see, e.g., Kelly 2007; Toba et al. 2019). This result supports an AGN-driven outflow: an ionized gas outflow is launched from the galactic nucleus. Focusing on the galactic center, σ₀ in merger stage D is the strongest that is comparable to dust-obscured AGN (Toba et al. 2017b; see also subsection 3.4).

Fig. 7.

Outflow strength (σ₀) as a function of the galaxy (IFU) center. Blue, green, yellow, and red data points represent the merger stages of A, B, C, and D, respectively.

Open in new tabDownload slide

It should be noted that a starburst-driven outflow is also reported in some galaxies and is reproduced by numerical simulations (e.g., Jogee et al. 1998; Leon et al. 2007; Schneider et al. 2020). Hence σ₀ may not be tracing purely AGN-driven outflow. To test this possibility, we utilize SDSS spectra of nearby starburst galaxies and execute the spectral analysis. We first select 430 objects from the SDSS DR17 that satisfy the following criteria: (i) 0.01 < z < 0.04, (ii) r mag < 15, and (iii) SUBCLASS = “STARBURST.”¹³ The redshift and magnitude ranges are similar to those in our GOALS sample. The fiber diameter of the SDSS/BOSS Sspectrograph is 2″–3″, which corresponds to the central two–three fibers in KOOLS-IFU. We then perform the spectral fitting with QSFit in the same manner as is described in subsection 2.3 and estimate σ₀. As a result, the weighted mean of σ₀ for the starburst galaxy sample is log  σ₀ ∼ 2.7. This value is smaller than those in the inner part of our GOALS sample, which means that it could be difficult for starburst galaxies to produce such a strong outflow with log  σ₀ > 2.8 as we observed for our sample (figure 7). This suggests that high σ₀ basically originates from an AGN-driven outflow.

In order to see how the galaxy merger could affect the gas density from the spatially resolved view, we also estimate electron density (n_e) of ionized gas from the line flux ratio of [S ii]λλ6716,6731 that is known as a good tracer of n_e and widely used for AGN¹⁴ (e.g., Osterbrock & Ferland 2006; Kawasaki et al. 2017; Kawaguchi et al. 2018; Joh et al. 2021), which tells us how n_e could be different among the inner and outer regions of our GOALS sample. Here, a conversion formula is employed to estimate n_e from the [S ii] ratio (see, e.g., Kakkad et al. 2022, and references therein). Albeit with a large uncertainty (⁠|$\sim\!\! 30\%$|⁠), we find a spatial gradient of n_e, i.e., high-density gas with n_e ∼ 10³ cm⁻³ is located in the central 0.5 kpc while low-density gas with n_e < 100 cm⁻³ is distributed in the outer region.

3.4 Outflow properties as a function of merger stage

Finally, we investigate how σ₀ could be associated with the merger stages as shown in figure 8. We find that ionized gas outflow is more powerful as a sequence of merger stages; the outflow strength in the late-stage (stage D) mergers is about 1.5 times stronger than that in the early-state (stage B) mergers. We also check the correlation between σ₀ and D₁₂ in which D₁₂ is the separation between the two nuclei in units of kpc (see Yamada et al. 2021). We confirm a negative correlation between two quantities for samples with merger stages from B to D. We further find that n_e estimated from the [S ii] doublet (see subsection 3.3) seems larger in the late-stage merger. Recently, Yamada et al. (2021) reported that the outflow velocity of molecular gas of X-ray AGN in merger stage D correlates with Eddington ratio.¹⁵ Molecular outflows are also correlated with the ionized gas outflow regarding the mass outflow rate (Fiore et al. 2017). These results suggest that the galaxy merger could induce a dense ionized gas outflow driven by AGN, particularly in the late-stage merger. At the same time, an important caveat to be kept in mind is that those results are based only on nine objects. We cannot rule out the possibility that the observed trends are affected by a selection effect (see below). Further observations for the GOALS sample with a wider range of redshift, brightness, and dynamical state are needed to corroborate our conclusion.

Fig. 8.

Outflow strength (σ₀) as a function of the merger stage. Since especially objects classified as stage A might be affected by the selection effect or even classified as stage D, our conclusions are based primarily on results from stages B to D (see the text for details).

Open in new tabDownload slide

We also note that σ₀ for merger stage A is comparably large to stage D. One possible reason for this is that merger stage A is too early a phase (i.e., two galaxies are too far from each other) to affect the strength of the ionized gas outflow, and thus σ₀ could depend on the nature of the individual object rather than the contribution of the merger process. Indeed, NGC 6921, NGC 7674, and NGC 7679 (which are classified as merger stage A) are all reported as strong AGNs with X-ray detections¹⁶ (see Yamada et al. 2021 for details). In particular, NGC 7674 is expected to harbor dual SMBHs [Kharb et al. (2017), but see also Breiding et al. (2022) for counterargument], which could have quite a strong AGN outflow.¹⁷ A full investigation of this object will be reported (S.Yamada et al. in preparation). Another possibility is that those objects are indeed at the very last stage of former merging events so that no tidal features are visible (which means they are stage D). On the other hand, if the trend in figure 8 is robust without being affected by the selection effect, this result could suggest that a galaxy merger is not the only path to ignite AGNs as introduced in section 1.

Arribas et al. (2014) reported that SF-driven outflow strength traced by the Hα line in SFGs is correlated with dynamical phases [which correspond to the merger stage; see Rodríguez-Zaurín et al. (2011) for more detail] along the merging process. Rich et al. (2015) also reported from an IFU observation that the low-velocity shocks become increasingly prominent as a merger progresses (see also Rich et al. 2011). In addition, the outflow strength (σ₀) seems to be correlated with the Eddington ratio (e.g., Bae & Woo 2014; Toba et al. 2017b). Taking account of those findings, our observations support that galaxy mergers enhance both SF and AGN activity and play an important role in the co-evolution of galaxies and SMBHs.

4 Summary

In order to reveal how a galaxy merger could affect the strength of the ionized gas outflow from a spatially-resolved point of view, we analyze nine local U/LIRGs at z < 0.04 through the IFU observations with KOOLS-IFU on the Seimei Telescope. The observed targets are selected from a 60 μm flux-limited sample in the GOALS project, covering a wide range of merger states from an early-stage (A) to late-stage (D). We successfully detect various emission lines of Hβ, [O iii]λλ4959,5007, Hα, [N ii]λλ6549,6583, and [S ii]λλ6716,6731 over |$70\%$| of fibers. This work mainly focuses on the strength of the ionized gas outflow (σ₀) by using a combination of velocity shift and dispersion of the [O iii]λ5007 line. With all the caveats discussed in subsections 3.3 and 3.4 in mind, we find that σ₀ shows a negative correlation with the distance from the galaxy center and a positive correlation with a sequence of merger stages. This indicates that galaxy mergers may induce AGN-driven outflow and maximize the late-stage merger, which is in good agreement with previous works reporting that the Eddington ratio of AGN increases along with the merger stage.

Since our follow-up observation campaign with KOOLS-IFU is ongoing, and about 130 GOALS samples are observable from the Seimei Telescope, this work provides a benchmark for revealing the spatially resolved outflow properties from a statical point-of-view in the future.

Acknowledgements

We gratefully thank the anonymous referee for a careful reading of the manuscript and very helpful comments. We acknowledge Dr. Kyuseok Oh for providing a finding chart for our observations in 2019. This work is supported by JSPS KAKENHI Grant numbers 18J01050, 19K14759, and 22H01266 (YT), 19J22216 (SY), 20H01946 (YU), 20K04027 (NO), and 21J13894 (SO). SY is grateful for support from RIKEN Special Postdoctoral Researcher Program.

Footnotes

References

Author notes