Debris of Gaia-Sausage-Enceladus that made a H I hole in the Milky Way ≈ 20 million years ago

The Perseus arm is known as one of the two 1–3 or four 4,5 dominant spiral arms of the Milky Way. While there is a large number of Massive Young Stellar Objects in the outer portion of the arm, a lower density of those is found in the inner portion 6–8 . Inner Perseus arm shows a noncircular motion of > 70 km s − 1 at a Galactic longitude of ∼ 50 ◦ , and its origin remains unclear 9 . Here we report an analysis of the kinematics and spatial distribution of neutral hydrogen (H I ) gas, star-forming regions (SFRs) and stars, together with an analysis of the star’s chemical abundances. We discovered that H I gas with ∼ 10 6 solar mass was lacked in the inner Perseus arm, and a similar amount of H I gas was distributed above the Galactic plane. The extended H I gas is well followed by retrograde low-metallicity stars, which are likely fossil stars from Gaia − Sausage − Enceladus 10–13 . Orbit integration shows that the fossil stars crossed the inner Galactic disk about 20 million years ago. The lower star-formation detailed structure of the spiral arms in the disk. compare kinematics of the inner Perseus arm the velocity distribution of 35 SFRs at a Galactocentric speciﬁc 8 range less affected by the bulge (Galactic thus studying the effects of spiral arms. Inner Perseus-arm sources show slower velocities V compared other spiral-arm statistical peculiar (noncircular) motion of G049.41 much larger than would be expected given the gravitational potential of the spiral arm the origin of the peculiar (noncircular) motion the marginally signiﬁcant vertical motion ± the R

deed, the faint area (∼10 K; gray area) is more than two times as faint as the surrounding area (> 1 20 K). Physical size of the faint area scales as 1 Galactic longitude, and d is heliocentric distance. The distance of G049.41 is 6.6 +1.1 −0.4 kpc 14 . Fig. 3 2 indicates an existence of H I hole with a size of ∼1 kpc around G049. 41. H I mass in the figure 4 can be estimated with a general procedure 19  between the faint and surrounding areas is >2×10 6 M ⊙ at the distance of G049.41. A similar shape 8 (i.e., black polygon in Fig. 2), but with bright emissions, was discovered toward a high-velocity 9 gas in M101 20 . M101 is the nearly face-on spiral galaxy, and shows holes in H I distribution 21,22 . 10 The high-velocity gas is moving perpendicular to the disk of M101, and its origin is thought to be 11 recent collisions of extragalactic gas clouds with the disk of M101 20 . 12 To reveal the origin of the faint H I emissions, we integrated H I emissions over the velocity 13 range in the black polygon of Fig between the excess emissions above the plane and the faint emissions in the disk will be further 19 discussed below. 20 To estimate the distance of the excess H I emissions, we obtained the 6D phase space infor-1 mation for stars from the early installment of the Gaia's third data release (EDR3) [24][25][26] . Stars that 2 satisfied the LSR velocity range in the black polygon (Fig. 2) and a parallax accuracy of better than 3 20%, were selected (see Methods for details). The final sample was composed of 424,059 stars, 4 of which 47,695 stars had metallicity information (the common logarithm of the iron-to-hydrogen 5 ratio divided by the solar value; [Fe/H]). Stars with [Fe/H] < −1.0 dex (i.e., less than one tenth 6 of the solar metallicity) are defined as "low metallicity stars" in this paper (430 stars identified). 7 We found that the low metallicity stars were systematically distributed above the Galactic plane 8 with a median Galactic height (z) of 1.8 kpc, whereas stars with [Fe/H] ∼ 0 (i.e., solar metallicity) 9 were distributed more closely to the plane (Extended Data Fig. 1). We examined the kinematics of 10 the low metallicity stars, and found that retrograde low-metallicity stars (i.e., V φ < 0 km s −1 ) are 11 moving away from the Galactic plane with a median vertical velocity (V z ) of 68 km s −1 (Extended

12
Data Figures 2 and 3). The retrograde low-metallicity stars and G049.41 are superimposed on 13 l − b plots of H I emissions (Figures 3a and 3b). Surprisingly, the distribution of the retrograde 14 low-metallicity stars is well matched with those of H I emissions above the plane. Mass of H 15 I emissions scales as where b max − b min is a range of Galactic latitude, and the others were explained previously. The 17 median distance of the retrograde low-metallicity stars is 5.5 kpc. In Fig. 3b tion by low-metallicity thick-disk stars. The low-metallicity thick-disk stars are thought to be born 12 during or after the GSE merger 27 . 13 We checked to determine when retrograde low-metallicity stars with e >0.7 crossed the 14 Galactic disk, by orbit integration (see Extended Data Fig. 5  disk. Raw material for star formation in the inner Perseus arm could have been reduced by the disk 10 crossing, although relationship between the arm and the disk crossing should be further examined.

11
The parental cloud of G049.41 might be perturbed by shock wave induced by the disk crossing.

12
The above interpretation is schematically summarized in Fig Table 2, and those associated with the source are defined in Extended 7 Data Table 3. The parameters and the definitions are applied throughout the paper. Here, we only 8 describe details about the stellar sample because we applied general procedures for H I and VLBI 9 data analyses. We checked to determine each radial velocity as a function of Galactic longitude satisfied the 15 LSR velocity range in the black polygon (Fig. 2). Note that radial velocity in Gaia EDR3 is 16 calculated in the solar barycentric reference frame (∼heliocentric radial velocity V Helio ), and thus 17 we converted each radial velocity to LSR velocity (V LSR ) for the comparison. 18 Also, we added the restriction of a parallax accuracy better than 20% ( π δπ > 5). This is 19 because estimating distance by simply inverting the parallax can result in the Lutz-Kelker bias, 20 which becomes significant when the parallax error is large (e.g., π δπ ≤ 4) 36   stars that satisfy the LSR velocity range in the black polygon (Fig. 2)    sion. LAMOST is operated and managed by the National Astronomical Observatories, Chinese Academy 10 of Sciences.

14
The LAB H I data was analyzed by AIPS, Astronomical Image Processing System 46 . AIPS is produced and 15 maintained by the National Radio Astronomy Observatory, a facility of the National Science Foundation 16 operated under cooperative agreement by Associated Universities, Inc.

17
The orbit integration was performed with the galpy 39 (see HP:http://github.com/jobovy/galpy). 18 The software TOPCAT was used for making figures. TOPCAT was also used for the cross-matching be-19 tween Gaia EDR3 and metalicity data (i.e., APOGEE DR16 and LAMOST DR5). TOPCAT was initially 20 Bewilligungsnummer 05A08VHA), the European Space Agency, and the FP7 project GENIUS. All of this 4 support is gratefully acknowledged.

5
Author contributions All the authors contributed to the work. N.S. led the project and contributed to all 6 the aspects of the paper (i.e., data reduction; discussion; paper writing). H.N. provided initial idea of the 7 research, disk crossing by a stream. The idea allowed us to discover the stellar and gaseous streams. H.N. 8 and K.K. contributed to the writing and provided stimulated discussions, which improved the quality of the 9 paper.

10
Code Availability There is no custom code or mathematical algorithm that is deemed central to the con-11 clusions in this paper. 12 Competing Interests The authors declare that they have no competing financial interests. 13 Correspondence Correspondence and requests for materials should be addressed to Nobuyuki Sakai (email: 14 nsakai@kasi.re.kr). 15