© The Authors 2024

1. Introduction

Since the advent of the ROSAT All-Sky Survey (RASS; Snowden & Schmitt 1990), the X-ray sky is known to include different extended sources of diffuse X-ray emission, of which the North Polar Spur (NPS), the Eridanus-Orion superbubble, and the Monogem, Antlia, and Vela supernova remnants (SNRs) have the largest angular size (Zheng et al. 2024a). Although these different sources encompass a moderate range in surface brightness from faint (e.g., Monogem and Antlia) to very bright (e.g., NPS), they are easiest to recognize and distinguish from the background emission by a simple visual inspection of any X-ray soft band map.

The recent advent of the first eROSITA All-Sky Survey (eRASS1; Merloni et al. 2024) improved on this view through the higher signal-to-noise ratio (S/N) it provided. The quality of the eRASS data has been assessed through a detailed comparison with the RASS data, and their consistency has been validated (Zheng et al. 2024a). The smaller number of systematic sources of noise, for example, recently allowed for the recognition of two very large and extended sources of diffuse emission that are symmetrically displaced around the Galactic center and Galactic plane, known as the eROSITA bubbles (Predehl et al. 2020). In a similar fashion, we study here an additional extended feature found in the eRASS1 data that was not recognized in the ROSAT maps because stronger systematics close to the South Ecliptic Pole (SEP) prevented this. Around the SEP, the sensitivity of both the RASS and eRASS1 data is enhanced by the frequent passage of the telescope field of view (FoV) because the survey strategy of the two instruments is similar. Although frequent observations have the advantage of a higher sensitivity, they can carry a higher bias in case of a background noise that is extended in time. Examples of this are the long-term enhancement experienced by ROSAT that is attributed to a higher solar activity at the time of the RASS, as well as to the passage of ROSAT through the South Atlantic Anomaly. This region has a highly increased particle background (Freyberg 1998; Robertson & Cravens 2003). Therefore, the new eRASS1 data potentially allow us to detect extended diffuse emission in excess of the background in all the regions that are affected by systematics in RASS, as happened with the eROSITA bubbles (Zheng et al. 2024a). The new coherent patch of diffuse X-ray emission presented in this work remarkably extends over ∼1000 deg² and encompasses several constellations (e.g., Dorado, Mensa, Hydra, Reticulus, Volans, Pictor, and Chamaleon). One of its brightest features is an arc-shaped brightening spanning ≥10 deg and resembling a goat’s horn. To easily refer to the extended coherent patch of diffuse emission studied in this work (which is difficult to associate with a particular constellation), we call it the Goat Horn complex.

In general, multiwavelength observations of diffuse sources and possible correlations between bands are able to provide insight into the nature of the diffuse source. In particular, the codetection of diffuse X-ray and radio emission usually shows shocks (traced by the synchrotron emission that they induce) that heat plasma to very high temperatures (traced by the X-ray thermal emission, with kT ∼ 0.1 − 1 keV; e.g., Ponti et al. 2019, 2021). Synchrotron emission is often observed by radio surveys at gigaherz frequencies as a major diffuse component (e.g., Carretti et al. 2019). At the largest angular scales, the synchrotron emission is usually associated with a Galactic origin, arising by the combination of a Galactic cosmic-ray spectrum and a Galactic magnetic field. Polarized intensity in particular highlights the nonthermal (i.e., synchrotron) nature of the emission and has been used to detect large-scale (i.e., 10–100 deg) circular rings and arcs, which are referred to as radio loops (Vidal et al. 2015, and references therein; the loops are numbered and named after their projected length). These loops are usually thought to be linked to local Galactic phenomena in general, such as nearby supernova remnants and Galactic outflows, or they are associated with OB star complexes (Bracco et al. 2023).

In Sect. 2 we present the different datasets used in this work. In Sect. 3 we present the observational evidence that an additional source of extended emission is present that was never recognized before. In Sect. 4 we test different hypotheses based on the observational evidence about the nature of the new extended source. In Sect. 5 we conclude and list open questions.

2. Data

2.1. X-ray intensity

The X-ray observations were carried out by the extended ROentgen Survey with an Imaging Telescope Array (eROSITA; Predehl et al. 2021) on board the Spektrum-Roentgen-Gamma (SRG) observatory (Sunyaev et al. 2021), during eRASS1. The data selection, analysis, and imaging are the same as described in (Zheng et al. 2024a, we refer the reader there for further details) and in Zheng et al. (2024b). In Fig. 1 we show narrow-band images around the bright soft X-ray O VII (0.543–0.613 keV) and O VIII lines (0.614–0.694 keV) in a region around the polarized radio loop XII (Vidal et al. 2015) and centered on the position of the Large Magellanic Cloud (LMC). We note that for displaying purposes only, in Fig. 1 we filtered out all bright discrete sources by selecting pixel values five times higher than the median within a circular region with a radius of 2 deg centered on the same pixel. The pixel value was then substituted with the median value. This procedure allowed us to highlight the diffuse emission on scales larger than 2 deg in the maps.

Fig. 1. O VII and O VIII narrow-band eRASS1 maps. The images are centered at the LMC coordinates (l, b) = (280, −33.7) deg. All the maps presented in this work use the same projection. The solid white line represents the great circles b = 0 deg and l = 0 deg. The dotted white lines are separated by Δl = 30 deg in longitude and Δb = 30 deg in latitude. Left column: O VII (top) and O VIII (bottom) eRASS1 maps (Locatelli et al. 2024; Zheng et al. 2024a). Right column: deabsorbed intensity of the warm-hot CGM component of the Milky Way, extracted from the O VII (top) and O VIII (bottom) eRASS1 maps (i.e., left column). We refer to Locatelli et al. (2024) for details of the CGM intensity retrieval and of the deabsorption method.

By selecting photons in narrow energy bands centered on the energy of bright emission lines, Zheng et al. (2024b) have built maps of two high-ionization states of the oxygen (i.e., O VII, O VIII), which are thought to trace the warm-hot phase (T ∼ 10⁶ K) of the Milky Way circumgalactic medium (e.g., Locatelli et al. 2024). To detect diffuse emission in excess of the background, we defined the stripe encompassing the longitude range 250 < l < 220 deg as the background region. This stripe encompasses all latitudes b ∈ [ − 90; 90] deg. The longitude range was selected by the absence of evident structures (see Fig. 1) in the X-ray background and is considered as representative of the emission produced by the warm-hot phase diffuse plasma of the Milky Way circumgalactic medium (Locatelli et al. 2024). At fixed latitude b within the background region, we computed the signal-to-noise ratio (S/N) of the deabsorbed intensity¹ by first subtracting the estimated circumgalactic medium (CGM) emission at each latitude b,

$$\begin{array}{c}\hfill \stackrel{\sim }{I}=I_{\mathrm{X}}(l,b)-⟨I_{\mathrm{cgm}}(b)⟩,\end{array}$$(1)

where the notation _cgm(b) indicates a quantity sampled in a region limited by 220 < l < 250 deg and b ± 5 deg at a given b. The significance map S/N is then obtained as

$$\begin{array}{c}\hfill \frac{S}{N}=\frac{\stackrel{\sim }{I}(l,b)}{\mathrm{std}({\stackrel{\sim }{I}}_{\mathrm{cgm}}(b))}.\end{array}$$(2)

Figure 2 shows the derived S/N map. While known bright and extended soft X-ray sources are evidently visible as coherent dark features in Fig. 2 (e.g., SNRs, the Eridanus-Orion superbubble, and the eROSITA bubbles), we found an extended patch of diffuse emission without a known counterpart around the LMC in pojection. We defined the Goat Horn complex as the coherent patch bounded by S/N ≳ 2 in the LMC region, as shown by the blue circle in Fig. 2. The region is approximately limited by the edge of the eROSITA bubble on the east side, while the northern boundary extends all the way up to b = −10 deg, which is close to the Galactic disk (where the model of the absorption layer may introduce biases due to high values of the column density).

Fig. 2. S/N map of the O VIII deabsorbed image. S/N is the RMS value of the deabsorbed O VIII map at 220 < l < 250 deg and fixed b (±5 deg). The longitude range was selected to represent the intensity of the soft X-ray background. In order to enhance the diffuse emission, the map was smoothed with a Gaussian kernel with a full width at half maximum of 1 deg, and bright sources on scales smaller than 1 deg were removed. Coherent patches at S/N ≳ 2 are associated with discrete extended sources, as labeled. The new Goat Horn complex is defined by the coherent S/N ≳ 2 patch within the dashed blue circle.

2.2. X-ray line intensity ratio: Temperature

The ratio of the O VIII to O VII CGM brightness was converted into a proxy for the plasma temperature, and a first temperature map of the soft X-ray sky was provided (Zheng et al. 2024b). In Fig. 3, we show the line-ratio map obtained after corrections: The background and foreground emission components were subtracted from the narrow-band maps of O VII and O VIII. The result was then deabsorbed, as described in Locatelli et al. (2024) and (Zheng et al. 2024b, we refer to this publication for details of the narrow-band data preparation). We note that the conversion of the line ratio into a temperature associated with a plasma is biased when the assumption is broken that a single foreground absorption layer accounts for the total observed N_H. For instance, this might occur in a circle with a radius of ∼2 deg from the LMC and the Small Magellanic Cloud (SMC) centers and across the Magellanic Bridge. In these regions, a high HI column density N_HI is observed and is physically associated with the Magellanic Clouds, that is, behind local and Galactic X-ray emission, at a distance of ∼50 kpc. These high N_H regions within the Magellanic Clouds are currently not excluded from the computation of the foreground optical depth. Therefore, we note that the X-ray intensity and line ratio measured in the inner regions of the Magellanic Clouds and Bridge was considered as biased and was excluded from our analysis here. However, the excluded regions cover only a few percent of the overall area covered by the Goat Horn complex and do not affect the rest of our analysis, unless stated otherwise.

Fig. 3. O VIII/O VII temperature proxy maps. Contributions from the local hot bubble foreground, the instrumental background, the cosmic X-ray background, and the foreground absorption were removed from the maps before we computed the ratio (see the text and Zheng et al. 2024a for further details). Lower panel: ratio of the narrow-band images centered on the O VIII and O VII emission lines. Upper panel: corresponding temperature under the assumption that the narrow-band energy ranges are dominated by O VII and O VIII lines, produced by collisionally ionized plasma.

In addition, the O VIII to O VII line ratio cannot be converted into a meaningful temperature value whenever the two different line intensities involved in the ratio are produced by different sources or spectral components. For the reasons above and for completeness, we provide and refer to the line ratio value and to the temperature value corresponding to the ratio under the assumption of a single plasma temperature (see Fig. 3, referred to as “pseudo-temperature” 𝒯 in Zheng et al. 2024b), when potentially useful. We caution that rather than an absolute calibration of the plasma temperature, the temperature map is mostly meant to show the approximate location and amplitude of potential temperature transitions, as the absolute normalization of the temperature relies on the details of the back- and foreground subtraction performed on the narrow-band images and is thus model dependent.

2.3. Optical reddening

Lallement et al. (2019) studied optical extinction in the Gaia data and produced a 3D map of the dust surrounding the Sun up to 600 pc in height and 3 kpc in distance along the plane parallel to the Milky Way stellar disk, centered on the position of the Sun. We converted the optical extinction into an equivalent hydrogen column density following the method proposed by Willingale et al. (2013). The precise method used on our data is also reported in Appendix A of Locatelli et al. (2024). We refer to this publication for further details. After converting the extinctions into N_H in all the data cube, we reprojected the cube onto HEALPix grids (Górski et al. 2005) at increasing distances from the Sun. At each distance step, the HEALPix angular grid resolution was chosen to be the one closest to the angular size of a voxel in the original data cube by Lallement et al. (2019) as seen from the Sun. Therefore, we were able to easily query the obtained Heapix maps iteratively in a given direction and reconstructed the N_H differential (and cumulative) profile along the line of sight (LoS).

2.4. Radio emission

Radio emission at gigaherz frequencies mostly arises from synchrotron emission by gigaelectron volt electrons that are thought to be provided in the ISM by Fermi-like acceleration processes at the shock fronts of SNRs. In addition, the same shocks compress the magnetic field lines and in turn amplify the magnetic field strength. The enhancement obtained at the SNR boundaries is easily and commonly observed on scales of arcminutes for discrete young and far SNRs, while it is observed as coherent patches of brighter emission diffused along arcs of 10–100 deg, which are also known as radio loops (e.g., Vidal et al. 2015).

We used the S-Band Polarization All Sky Survey (S-PASS; Carretti et al. 2019). This survey, run at 2.3 GHz by the Parkes telescope, has the advantage of combining a high sensitivity (0.81 mK beam⁻¹, 8.9′ resolution) with a wide coverage of the southern ecliptic hemisphere (Dec < 1 deg; this largely overlaps with the western Galactic hemisphere). It therefore covers our field entirely. The S-PASS provides all Stokes products (I, Q, U, and V). The polarized radio emission (i.e., the linear combination of the Q and U Stokes $P=\sqrt{Q^2+U^2}$) is also provided. The S-PASS polarized intensity P and polarization fraction p = P/I are shown in Fig. 4 upper and lower panels, respectively.

Fig. 4. S-PASS 2.3 GHz radio polarization maps (Carretti et al. 2019). The dashed white line represents the location of polarized radio loop XII as originally defined by Vidal et al. (2015). Upper panel: linear polarization P. Lower panel: fractional polarization p ≡ P/I.

In addition, we considered the all-sky full-Stokes synchrotron maps taken by the Planck satellite at 30 GHz (Planck Collaboration IV 2020). A linear integral convolution (LIC) of the Q and U Stokes parameters allowed us to show the direction of the magnetic field lines in terms of a color gradient (Fig. 5). In principle, the LIC could also be performed on the S-PASS Stokes products. However, the S-PASS frequency of 2.3 GHz can be affected by Faraday rotation, an effect proportional to ν⁻². The ten times higher frequency of Planck is instead negligibly affected by Galactic Faraday rotation (e.g., Vidal et al. 2015), and we consider it a more reliable proxy of the Galactic magnetic field direction. The LIC intensity gradients are perpendicular to the magnetic field direction. For example, the magnetic field is aligned along the extent of radio loop XII, and is coincident with a higher polarization intensity and fraction in Fig. 4 upper and lower panels, respectively.

Fig. 5. Linear integral convolution of the 30 GHz Planck Stokes parameter Q and U (Planck Collaboration IV 2020). The Q and U Stokes parameters were rotated in order to produce angles parallel to the magnetic field lines.

2.5. Absorption toward X-ray background sources

Additional constraints on the physical properties of the hot plasma that causes the X-ray emission in the Goat Horn complex come from the study of high-ionization lines (e.g., O VI, O VII), when detected as absorption features in the UV and X-ray spectra of bright background sources. Previous studies of the black hole binary system LMC X-3 detected the O VII absorption line at redshift zero in that direction (Wang et al. 2005; Bregman & Lloyd-Davies 2007). LMC X-3 is located 6.9 deg away (i.e., 6.7 kpc) from the center of the LMC. PKS 0558-504 ((RA, Dec) = (05h 59m 47.38s, − 50d 26m 52.4s), (l,b) = (257.96, − 28.57) deg), another source close to the Goat Horn complex, but located in projection outside its boundaries, shows an O VII absorption line with a similar equivalent width as LMC X-3 (Bregman & Lloyd-Davies 2007).

With the aim of further testing the results obtained by Wang et al. (2005) and Bregman & Lloyd-Davies (2007) in additional and independent directions, we searched the XMM-Newton Science Archive² for other bright soft X-ray sources found in projection within the boundaries of the Goat Horn complex. We found and analyzed all the observations with available data from the reflection grating spectrometer (RGS) on board XMM-Newton of the Seyfert 1 galaxy 1H 0419-577, collected between 2000–2018. The source is located at coordinates (RA,Dec) = (04h 26m 00.80s, − 57d 12m 01.0s) or (l,b) = (266.987, − 41.996) deg. The total available exposure covering the O VII z = 0 wavelength (λ_O VII = 21.60 Å) is 369 ks. To the best of our knowledge, the X-ray spectrum of 1H 0419-577 has never been used to study Galactic O VII absorption, as most of the observations have been carried after the thorough study of Galactic absorption by Bregman & Lloyd-Davies (2007). In this case, 1H 0419-577 is also serendipitously located in projection very close to the region in which we studied the X-ray spectrum of the excess emission, along the bright arc. The sky positions of 1H 0419-577, LMC X-3, and PKS 0558-504 are shown in Fig. 6 (cyan stars in the lower right panel, from the lowest latitude upward). We present the results of this analysis in the following section (Sect. 3.8).

Fig. 6. Chart map of the sources and features used and discussed in this work. The map is the same as Fig. 1. The dotted lines perpendicular to the Galactic disk and centered at b = 0 show directions tangential to the closest spiral arms of the Milky Way. By increasing l (i.e., right to left), they represent the Carina (green dots, l = 284 deg), Norma (orange dots, l = 306 deg), Centaurus (yellow dots, l = 312 deg), and Perseus (purple dots, l = 229 deg) arms. The large white dots represent the positions of the SMC and LMC. The empty white squares show the LoS characterized by high absorption, shown in Fig. 8. The cyan stars show the positions of the three bright X-ray sources with a detected O VII absorption line at z = 0: PKS 0558-504, LMC X-3, and 1H 0419-577, listed by decreasing latitude. The dashed white line shows the position of radio loop XII. The rainbow lines show the data we used to build the profiles and correlations in Figs. 9 and 11. The orange triangles show the positions of the on-source (upward) and background (downward) diffuse emission spectra shown in Fig. 10.

2.6. Hα emission

The Winsconsin Hα Mapper Northern Sky Survey³ (WHAM, Haffner et al. 2003) provides the Hα calibrated intensity over the full sky with an angular resolution of 1 deg and a 12 km s⁻¹ spectral resolution, down to a 3σ sensitivity of 0.15 R (Rayleigh). The WHAM map shown in Fig. 7 shows emission from the warm ionized medium (WIM) within ∼ ± 100 km s⁻¹ of the local standard of rest.

Fig. 7. WHAM map of Hα emission from the Milky Way (Haffner et al. 2003).

3. Results

In this section, we present different aspects of the observations and focus to some extent on possible interpretations of their origin (these are explored in Sect. 4). We list the different important findings in a series of independent paragraphs in order to better distinguish between them.

3.1. X-ray morphology

The soft X-ray intensity maps (see Fig. 1) show an arc-shaped brightening close to the position of radio loop XII. The center of this arc in projection is located close to the LMC (compare with Fig. 6), while the western end seems to point to the position of the SMC. These morphological features motivated us to investigate whether the Goat Horn complex is physically connected to the LMC. We discuss this hypothesis in Sect. 4. We also note an additional arc-shaped enhancement just south of the brightest arc, tracing ∼S/N ≥ 2. However, this second arc is much fainter. The bright X-ray arc seems to merge with the Goat Horn complex edge (at S/N ≃ 2) on the eastern side.

3.2. Distance

Some absorbing clouds can be distinguished as a visual dimming of the X-ray intensity. These clouds can be used to place useful constraints (i.e., lower limits) on the distance of the Goat Horn complex. From the dust extinction 3D data converted into N_H estimates as explained above (see Sect. 2.3), we selected LoS that pierced some of the most evident absorbing clouds, as shown in Fig. 8. The peaks in the upper panel show the likely position of the X-ray absorbing clouds observed in front of the Goat Horn complex. The distance of these clouds is as far as ∼500 pc from the Sun. Therefore, we place a lower limit d > 500 pc on the distance of the Goat Horn complex from the position of the Sun.

Fig. 8. Column density NH of the X-ray absorbing material as a function of distance from the Sun in the solar neighborhood. Top panel: differential NH. Bottom panel: cumulative NH for the same measurements as in the top panel (solid lines). The dashed lines and hatched regions around them show the NH value with its uncertainty, as retrieved from the HI4PI survey in the same directions as indicated by the labeled colors in the top panel (see also the empty white squares in the chart map of Fig. 6).

3.3. X-ray jump

Driven by the morphology of the soft X-ray map, we investigated the jump in the X-ray brightness across the bright arc. We extracted a surface brightness profile from the paths that is shown in Fig. 6 in two radial directions that were chosen to be perpendicular to the arc curvature in the south (S) and southwest (SW) directions with respect to the arc centroid (i.e., close to the south ecliptic pole and the LMC). The resulting profiles are shown in the top panels of Fig. 9. The intensity values and their uncertainty are defined as the mean and the standard error of the mean inside each bin. A jump in the X-ray surface brightness at Δθ ∼ 15 deg shows the position of the bright arc in Fig. 1. A second intensity jump (also briefly mentioned above in Sect. 3.1 seems to be present at Δθ ∼ 18 deg in both directions. We note that the detection of the brightness jumps does not depend on the choice of the intensity map (i.e., deabsorbed or not; right and left panels of Fig. 1, respectively). This is due to the relatively low N_H ≤ 5 × 10²⁰ cm⁻² value across the southern half of the Goat Horn complex (see Locatelli et al. 2024, Fig. A.1). Therefore, the intensity discontinuity is not generated by absorption.

Fig. 9. Radial profiles across the bright arc found within the Goat Horn complex. The sky paths from which the profiles were extracted are shown in Fig. 6. The rainbow colors match the data in this figure. Δθ ≡ 0 is defined at coordinates (l, b) = (280, −34). The hatched area shows the region dominated by a sharp intensity transition in the X-ray data. Top panel: eROSITA soft X-ray intensity (0.5–1.0 keV). Central panel: temperature proxy (right vertical axis), as derived from the ratio of eROSITA narrow-band images centered around the O VII and O VIII emission lines (left vertical axis). Bottom panel: polarization intensity P along the same paths (S-PASS). The broad central peak matches the position of loop XII along the paths.

From similar paths across the jump, indicated by the rainbow paths in Fig. 6, we built O VIII/O VII line ratio profiles (Fig. 3) as a proxy for temperature variations. Background and foreground components that were unrelated to the hot CGM of the Milky Way were modeled and subtracted from the O VII and O VIII maps (Locatelli et al. 2024; Zheng et al. 2024b) before we took the maps ratio. We show the resulting profiles in the central panels of Fig. 9. Trends of increasing temperatures are observed at the location of the intensity jump (hatched region). The line ratio increase along the paths and across the hatched region amounts to ∼25% and maps into a temperature increase of ∼10%. The line ratio (i.e., temperature) increase appears to be higher in the southern than in the southwest direction because scatter of the data in the latter direction is stronger. However, a clear difference appears when the values before and after the hatched region in Fig. 9 are compared that encompasses radio loop XII.

Arc-shaped X-ray intensity jumps usually trace the positions of shocks or cold fronts (hereafter, we refer to the set of all the mentioned classes as fronts). From an observational point of view, shocks compress (i.e., they increase the density) and heat (i.e., they increase the temperature) the medium downstream of the front (i.e., where the shock has already passed). Cold fronts, while they also compress the downstream gas, are observed to produce a temperature jump in the opposite direction, with the upstream gas holding the higher temperature. In other words, while shocks produce a pressure jump, cold fronts maintain the pressure (∝nT) about constant across the front. Therefore, in order to distinguish between them, the temperature profile is required. If the source of the X-ray emission is a hot thermal plasma with temperature kT of about 0.1–1 keV, a proxy of the temperature can be obtained by taking the ratio of the O VIII and O VII line intensities, as discussed above.

3.4. Thermal plasma

We tested the assumption of a thermal plasma by extracting soft-band spectra from a region in coincidence of the bright arc emission (upward orange triangle in Fig. 6, bottom right panel) and from outside the Goat Horn complex (downward orange triangle). The latter was considered as a background reference and was subtracted from the former. We exploited the formal division of the eROSITA data into sky squared tiles of 3 by 3 deg. The spectra from the eROSITA sky tiles IDs 061147 and 044141 were extracted, and their difference was taken.

The resulting spectrum in the soft X-ray range 0.3–1.2 keV, associated with the excess emission, is shown in Fig. 10. The two bumps at around 0.57 and 0.65 keV are well fit by O VII and O VIII line emission convolved with the eROSITA response matrix. This shows that a hot plasma model fully accounts for the excess emission and thus confirms its thermal nature. eROSITA has in fact already proved to be sensitive to diffuse hot plasma emission (Ponti et al. 2023; Zheng et al. 2024a; Locatelli et al. 2024). We fit the spectrum of the excess emission using a collisionally ionized plasma model⁴. We set Z/Z_⊙ to either 0.1 (Ponti et al. 2023) or 1.0. We tested different assumptions on the metal abundance of the hot plasma (Z = 1.0,  0.1 Z_⊙). The best-fit temperatures are 0.21 ± 0.01 and 0.19 ± 0.01 keV (2.4 ± 0.1 × 10⁶ K and 2.2 ± 0.1 × 10⁶ K), respectively. The faint nature of the emission in the Goat Horn complex, combined with the spectral energy resolution of eROSITA, unfortunately prevents us from constraining the metal abundance of the plasma. The fit does not strongly favor one of the assumed abundances (or temperatures). The metal abundance and the spectrum normalization are degenerate. The degeneracy is expected because both parameters depend on the amount of underlying continuum emission, which is largely unconstrained by CCD instruments.

Fig. 10. 0.3–1.2 keV spectrum of the excess emission at the position of the bright arc. The spectrum from a background region from outside the Goat Horn complex was subtracted from the data. The on- and off-source (i.e., background) regions are indicated by the upward and downward orange triangles in Fig. 6, respectively. The black and red lines show the best fits for different assumptions on the metal abundance of the hot plasma (Z = Z⊙ and Z = 0.1 Z⊙, respectively).

In addition, we tried a multitemperature APEC model (GADEM). GADEM assumes a Gaussian distribution of temperatures. The best-fit temperature distribution peaks at the same temperature as for the simple APEC fit (with the corresponding metal abundance). In addition, the dispersion σ_T = 0.01 keV of the temperature distribution is consistent with zero. The goodness of fit (both χ² and c-statistics) does not improve with respect to APEC, even though the number of free parameters is higher. We thus concluded that a single temperature model describes the excess emission reasonably well, as observed by eROSITA, in the Goat Horn complex.

Regardless of the details of the fit of the excess emission, the evident O VIII and O VII features suffice to confirm that a thermal plasma is the main source of the extended diffuse emission of the Goat Horn complex. We note that the thermal nature does not contradict a possible shock and/or cold front. Fronts can even help to enhance the X-ray emission by increasing the density and/or temperature on one side of the front (i.e., increasing the number of collisions between the atoms), while the production of atomic transitions bearing X-ray photons due to atoms collision within the hot plasma is still the thermal mechanism at work.

3.5. X-ray intensity is anticorrelated with the oxygen line ratio

In general, the X-ray intensity can be interpreted as a proxy for n²L, where n is the (average) density of the thermal plasma, and L is the length through the source along the LoS. By interpreting the line ratio as a proxy for the temperature, the anticorrelation between X-ray intensity (i.e., density) and the line ratio (i.e., temperature) then traces a contact discontinuity or cold front. Cold fronts are surfaces of constant pressure where a density jump is anticorrelated with a temperature difference (i.e., hotter on the low-density side of the front). On one hand, the anticorrelation is evident in the bottom left panel of Fig. 11. While we plot all the data points, we highlight those located across the bright X-ray jump and loop XII using the same colors and shapes as in Fig. 9. The anticorrelation is also followed by the green to blue color gradient, indicating LoS piercing the front by increasing the distance to the origin. On the other hand, while the X-ray brightness in Fig. 9 shows a rather sharp transition, the temperature trend is milder and is harder to interpret as a neat jump. Therefore, while the anticorrelation between intensity and line ratio is evident, it remains unclear whether the transition traces an actual cold front. The association with a cold front is therefore uncertain, but possible.

Fig. 11. Scatter plots of the quantities presented in Fig. 9. The symbols and colors of all the points correspond to those of Fig. 9. Only the points included in the region of loop XII (see the hatched region in Fig. 9) are plotted with color in order to highlight their trends. The sets of points are fit with a linear relation. The corresponding Pearson correlation coefficients (R) are reported in the legend. The dashed and dotted lines fit the south (squares) and southwest (circles) points, respectively. The gray and red lines show the fit for all and for the loop XII region points, respectively.

3.6. Jump as X-ray counterpart of radio loop XII

One of the previously known radio loops, radio loop XII, is now found to match the southern boundary of the Goat Horn complex. To the best of our knowledge, this is the first time that this radio loop is connected to any other multiwavelength feature and is studied. In particular, the northern edge of loop XII closely matches the bright X-ray arc in the south(-west) quadrant for most of its length, while the southern edge of loop XII seems to match the S/N ≥ 2 contour defining the Goat Horn complex. A quantitative and complementary analysis of the polarized intensity (Fig. 9, bottom panel) reveals an anticorrelated jump with the X-ray brightness across the same front as presented above (Fig. 11, top left panel). The polarization jump is evident in both directions (south and southwest). An enhanced radio polarized signal anticorrelated with the X-ray intensity jump is in principle consistent with the presence of a cold front (Markevitch & Vikhlinin 2007; Kotarba et al. 2011; Geng et al. 2012; Donnert et al. 2018).

In addition, we note that from the S-PASS map that radio loop XII seems to continue and to form a closed loop around the Goat Horn complex, with some of the intensity seemingly elongated in the azimuthal direction around it. Due to the fainter nature of this circular feature, future inspections of the radio data will be crucial to probe a physical connection with the loop XII and/or the Goat Horn complex.

Due to the anticorrelation between the line ratio and the X-ray intensity (Fig. 11, bottom panel), the anticorrelation between the radio polarization and X-ray intensity (upper left panel) may be recast in terms of a correlation between the radio polarization intensity and the temperature (top right panel). The potential correlation is expected in a scenario where a shock or a cold front compress and align the magnetic field lines at the front location, thus boosting the polarized radio emission. We note that the data close to the front are dominated by large scatter (greenish points in Fig. 11). We show the Pearson correlation coefficient (R) in the legend subpanels in Fig. 11. The R-coefficient absolute value ∈[0, 1] shows the strength of a putative linear correlation between two quantities. We note that by considering all points in the extracted profiles, the strength of the correlation does not always seem to increase between two quantities and for different directions. In the profiles in Fig. 9, the line ratio and the linear polarization (trivial) both show bumps at the location of loop XII. This effect is more evident in the southern direction, but is possibly also seen in the southwest, although with larger scatter. Outside the region of loop XII, the two quantities do not show similar trends, with the line ratio gently increasing outward, while the behavior of the linear polarization depends on the direction considered. The latter behavior may be explained to mean that the X-ray emission upstream of the front is not dominated by a medium that is spatially close to the Goat Horn complex, and is thus unrelated. A comparison of Figs. 1 and 4 shows that while the former shows a transition in the Goat Horn complex at around l ∼ 270 deg, the latter does not. This also shows that the correlation is mainly driven by the points extracted from the profile toward the south. A projection effect may play a role in introducing scatter to the relation toward the west side of the Goat Horn complex.

3.7. Magnetic fields

In addition to the enhanced polarized emission, we observe a magnetic direction that follows the length of the radio loop, that is, parallel to the front length, as already pointed out by Vidal et al. (2015). The vector direction is shown by a modulation of the intensity map given by the LIC of Stokes Q and U of the 30 GHz Planck data (see Fig. 5). In principle, we do not know the environment that is characterized by the detected magnetic field lines, while a Galactic origin is usually assumed (Berkhuijsen et al. 1971; Planck Collaboration XXV 2016) given the coherent and very broad angular extent of the patterns. A Galactic origin of the polarized synchrotron emission was assumed and corroborated in the studies of all the other known radio loops, however. The magnetic field direction in the region of the Goat Horn complex other than radio loop XII follows the souteast-northwest direction to some extent and is clearly displaced along the length of a dust-absorbing cloud located in projection at about 10 deg north of the LMC. The cloud north of the LMC can be easily spotted as the brightest central feature in the fractional polarization map p = P/I of Fig. 4 (lower panel). Outside the Goat Horn complex, the magnetic field morphology looks more disturbed and/or tangled compared to the regions within the boundaries of the Goat Horn complex, strengthening the case for a coherent and distinct nature of the Goat Horn complex.

The polarized intensity in the S-PASS data shows a coherent decrease (consistent with 0) over a broad circular patch in the central regions of the Goat Horn complex (see the dark region in Fig. 4 upper and lower panels. We searched for the same feature in the WMAP 23 GHz 9-year (Bennett et al. 2013) and the Planck 30 GHz product release 3 (Planck Collaboration IV 2020; Planck Collaboration XXV 2016) polarized intensity maps, and found that the inner region of the Goat Horn complex is characterized by low values at both frequencies. This disfavors the interpretation that the inner regions of the Goat Horn complex are due to depolarization. Instead, by considering the circular boundary observed at 2.3 GHz, indications of it could already be found in the 23–30 GHz polarized maps, but the S/N was too low for the circle to be fully retrieved and/or recognized.

3.8. O VII absorption

From the study of high-ionization z  =  0 lines in the UV and X-ray spectra of the bright source LMC X-3 (Bregman & Lloyd-Davies 2007), which is behind the Goat Horn complex, the observed wavelength of the O VII absorption line at rest frame 21.592 ± 0.016 Å was found to be consistent with a Galactic origin (Bregman & Lloyd-Davies 2007). Instead, this value is marginally inconsistent with the radial velocity of the LMC (expected at 21.623 Å) at the 2σ level. Similar results were previously obtained by Wang et al. (2005). We note that the RGS nominal sensitivity in the O VII band is 5 m Å⁵, and it is thus largely sufficient for probing a potential shift. The oxygen column density in the LMC X-3 direction was measured as EW_O VII = 21.0 ± 5.0 m Å. In addition, according to Wang et al. (2005), the nonthermal broadening $\xi ={70}_{-20}^{+60}\mathrm{km}{\mathrm{s}}^{-1}$ that was found to contribute to the EW could be due to a vertical velocity gradient of the hot gas close to the Milky Way disk, as expected in a supernova-driven Galactic fountain (Shapiro & Field 1976; Bregman 1980). The same scenario also explains the observed trend of a decreasing scale height with respect to the lower ionization state when comparing O VII (and now also O VIII, Locatelli et al. 2024) with O VI, N V, C IV, and Si IV (Savage et al. 2000). In this picture, supernova shocks heat the hotter gas that expands and cools on its way to greater heights from the disk (Shapiro & Field 1976; Bregman 1980). As already pointed out, PKS 0558-504, another source close to the Goat Horn complex, but located in projection outside its boundaries, shows an O VII absorption line of EW_O VII = 21.7 ± 7.8 m Å, similar to that of LMC X-3.

We added the constraints on the O VII column density obtained in the direction of the bright source 1H 0419-577. By detecting the line and measuring an equivalent width of EW_O VII = 42 ± 31 m Å, we derived an oxygen column density of N_O = (8.4 ± 6.2) × 10¹⁵ cm⁻² (converted to N_H = (1.7 ± 1.3) × 10¹⁹ cm⁻² assuming an O/H ratio of 4.9 × 10⁻⁴, Lodders 2003). We note that despite the separation of 12–16 deg between the lines of sight toward 1H 0419-577, LMC X-3 and PKS 0558-504, the values of the column density N_O are consistent with each other. However, as our measurement is still characterized by large uncertainties, a moderate amount of matter could still be hidden below our uncertainty values.

3.9. Upper limit on Hα emission

The Goat Horn complex is not clearly detected by the WHAM survey (Haffner et al. 2003). Even though diffuse emission is present within the Goat Horn complex boundaries in Fig. 7, its morphology resembles the dust distribution obtained from the dust extinction measurements E_B − V (Planck Collaboration Int. XLVIII 2016). An evident correlation between these quantities is shown in Fig. 12, where each point is extracted in a ∼0.25 deg² region within the Goat Horn complex boundaries. Therefore, we conservatively consider the detected Hα emission as an upper limit to that of the Goat Horn complex in the same region. It is thus possible to derive an upper limit for the Hα to synchrotron ratio, where we considered the S-PASS total intensity I Stokes parameter as a measurement of the synchrotron emission at 2.3 GHz from the Goat Horn complex. We constrain the Hα to synchrotron ratio to ≤35(ν_GHz/2.3)^−sR/K (95% range) in the Goat Horn complex, where s is the spectral index of the synchrotron radio emission. Considering s = 1, the limit translates into ≤58R/K at 1.4 GHz.

Fig. 12. Hα intensity vs. optical reddening from dust in the region of the Goat Horn complex.

4. Discussion

4.1. Faint and old supernova remnant (alone) cannot explain the X-ray intensity and extent

The Goat Horn complex is not unique in terms of extension and brightness when observing the X-ray sky. Other noticeable and similar regions are, for instance, the Eridanus-Orion superbubble (Joubaud et al. 2019) and the Antlia and the Monogem Ring supernova remnants (SNRs; Knies 2022). In addition, the position of the Antlia SNR, for example, shares a similar longitude range and (absolute) latitude with the Goat Horn complex. Therefore, we test in this section the hypothesis that a single SNR can explain the emission of the Goat Horn complex.

Above, we set a stringent lower limit on the distance of the hot plasma in the Goat Horn complex, of d > 500 pc. A physical size of R ≥ 131 pc (Θ/15 deg)(d/500 pc) was thus obtained. This value lies at about the upper bound of typical SNR sizes (10–100 pc, Vink 2012), but is still consistent with the possible range. In the SNR scenario, we assume the bright arc to be the SNR shock surface because its shape is roughly circular. Despite having ruled out a shock nature for the front before, we applied a Sedov-Taylor formalism for shocks in SNRs as a test. The Sedov-Taylor equation links the density of the medium downstream of the front with the age and the shock distance to the center of the SNR.

The observed X-ray excess emission in the O VIII narrow band is

$$\begin{array}{c}\hfill \langle I_{\mathrm{O}viii}^{\mathrm{cgm}}\rangle =A_{\mathrm{eff}}^{\mathrm{eRO}}\frac{ϵ\mathrm{EM}}{4\pi \mathrm{\Omega }}{|}_{\mathrm{O}viii}\sim 1.0\mathrm{ph}{\mathrm{s}}^{-1}{\mathrm{deg}}^{-2},\end{array}$$(3)

where EM ≡ ∫_srcn² ds ≃ ⟨n² ⟩R is the source emission measure, Ω ≃ 700 deg² is the solid angle within 15 deg from the center of the Goat Horn complex, and $A_{\mathrm{eff}}^{\mathrm{eRO}}\simeq 1000{\mathrm{cm}}^2$ and ϵ1(kT = 0.15 keV) = 1.65 × 10⁻¹⁵(Z/Z_⊙) ph s⁻¹ cm³ are the eROSITA effective area and the hot plasma emissivity as evaluated in the O VIII band, respectively.

Based on the putative SNR radius R ≃ 130 pc as defined above and solving the equation for the density, we obtain an average value within the Goat Horn complex of n ≃ 0.4 − 3.8 × 10⁻² cm⁻³, where the quoted range includes the scatter on the observed surface brightness in the region downstream of the front, as well as the metal abundance range of Z = (0.1 − 1)Z_⊙. By assuming E = 10⁵¹ erg as the total energy of the SNR, the age estimate derived by the Sedov-Taylor equation is Δt = 1.7 − 5.3 × 10⁷ yr. Now, the typical age up to which SNRs are X-ray bright is Δt < 10⁵ yr, which is incompatible with our estimate. A complementary approach instead is to compute the expected density using the upper limit on the age of Δt ∼ 10⁵ yr (Vink 2012). This keeps the timescale consistent with that of a faint and extended SNR located in the Milky Way halo (i.e., at high Galactic latitudes |b|). Under this assumption, we obtain n ∼ 1 × 10³ cm⁻³ (R/131 pc)⁵(Δt/10⁵ yr)⁻², which is unrealistically large for an old SNR.

The nondetection of Hα emission is also in contrast with the SNR scenario. SNRs typically show Hα emission from a thin shell of colder gas that is swept up and accumulates at the boundary of the remnant during its expansion into the ISM. The cooling of the cold and dense shell is a runaway process that causes the gas layer to shine bright in Hα emission. The ratio of Hα to diffuse synchrotron radio emission can also be used to explore the nature of a diffuse source (e.g., Sofue et al. 2023).

We stress that the results presented in this section are obtained for the closest possible distance to the SNR. In addition, d = 500 pc for the SNR geometrical center is already at odds with the total observed N_H. The hypothesis that all the detected excess is produced by an old and faint SNR (located in the Milky Way halo) is thus unlikely. However, it is still possible that a single and very extended SNR may contribute to part of the observed emission, and this may be explored in the future with multiwavelength data, based on which, we can search for other signatures typical of SNRs. The connection between X-ray and polarized radio intensity was initially discovered by following this approach. We can so far only exclude that a single SNR can explain all of the properties of the Goat Horn complex. In Sect. 4.3 we explore whether multiple supernova remnants and the local star formation in general are able to sustain the observed X-ray luminosity of the Goat Horn complex.

4.2. Hot atmosphere around the Magellanic Clouds (alone) cannot explain the X-ray intensity and extent

As we noted above, the LMC is found (at least in projection) at the center of the Goat Horn complex. The curvature of the bright X-ray arc and the loop XII also point toward a centroid close to (or overlapping with) the inner regions of the LMC itself. Cold, warm, and warm-hot gas phases associated with the MCs have been detected as far as 45 deg from the LMC (McClure-Griffiths et al. 2009; Fox et al. 2014; Barger et al. 2017; Krishnarao et al. 2022; Smart et al. 2023). These associations were usually based on the match of the radial velocity of emission lines with that of the LMC (v_LMC 321 ± 24 km s⁻¹, Kallivayalil et al. 2013), although the chances of a coincidence match are not negligible (Richter et al. 2015). The arguments above pose the question whether the Goat Horn complex origin is connected to the Magellanic system (as a whole or in part).

On one hand, the detection of a warm-hot gas phase associated with the LMC that extends up to 45 deg from its center (Lucchini et al. 2020; Krishnarao et al. 2022; Smart et al. 2023) sets the conditions for the presence of hotter gas in the case of a relatively broad distribution of temperatures (e.g., Gulick et al. 2021). In addition, discontinuities in the gas properties produced by the interaction of the Magellanic Clouds with the CGM of the Milky Way, as well as the mutual interaction between the Small and Large Magellanic Clouds, are plausible (Besla et al. 2010; Setton et al. 2023) and observed (e.g., D’Onghia & Fox 2016; Conroy et al. 2021). In this scenario, the front detected in the Goat Horn complex could then be caused at the interface between the Magellanic hot corona and the CGM of the Milky Way due to three-body interaction. Despite its position at the opposite side of the direction of the LMC motion with respect to the Milky Way, a correlation is not expected in general (Ascasibar & Markevitch 2006), and a different front may still be located in the leading direction while remaining hidden behind the thick dust clouds observed north of the Goat Horn complex. We also note that shocks and cold fronts are known to compress and align magnetic field lines, thus providing a possible explanation for the loop XII radio emission and its anticorrelation with the X-ray intensity. Although this is interesting, we note, however, that the radio emission does not allow us to distinguish between this particular scenario and a front with similar observed properties but produced by a smaller-scale (i.e., Galactic) phenomenon.

On the other hand, as mentioned above, studies of the high-ionization lines (O VI, O VII, O VIII and Ne IX) detected in absorption are inconsistent with the LMC properties in terms of redshift and dispersion (Wang et al. 2005; Bregman & Lloyd-Davies 2007). The new LoS presented in this work toward the quasar 1H 0419-577 shows an O VII absorption line with an equivalent width of about EW = 42 ± 31 m Å, consistent with the values observed toward LMC X-3 and PKS 0558-504. The lines of sight toward these different objects are separated by 12–16 deg. The similarity of the measured equivalent widths (and in turn of the column density) can be explained if the main contribution to the z = 0 O VII absorption line in the spectrum of the background sources comes from hot gas of Galactic origin, spread over a broad region. This suggests that the bulk of the observed X-ray emission of the Goat Horn complex is associated with a Galactic origin. We note, however, that a small contribution (N_O < 10¹⁵ cm⁻²) from a hot Magellanic corona may be still possible.

Overall, hot gas in the Magellanic system, although it cannot provide a viable explanation for the Goat Horn complex as a whole, may be still connected to some of its features. For instance, it could account for part of the observed emission or be connected to the front, as suggested by the intensity morphology. Complementarily, the active star-forming region 30 Dor, located within the LMC may have provided a large energy input in the past million years to heat and pressurize the gas that eventually outflows from the LMC. The total supernovae (core-collapse + Ia) formation rate of the Magellanic Clouds is estimated to be 2.5 − 4.6 × 10⁻³ SNe yr⁻¹ (Maoz & Badenes 2010). At an energy input of ∼ 10⁵¹ erg SN⁻¹, the rate provides a luminosity of L_SNe, LMC = 0.8 − 1.5 × 10⁴¹ erg s⁻¹. In the Goat Horn complex region, we observe an X-ray intensity of I_X = (1.7 ± 0.8) cts s⁻¹ deg⁻² (10–90% of the distribution, the LMC disk is masked) spread over Ω_GH ∼ 1400 deg². The total luminosity is thus

$$\begin{array}{c}\hfill L_{\mathrm{X}}=\frac{E_{\mathrm{O}viii}{I}_{\mathrm{X}}{\mathrm{\Omega }}_{\mathrm{GH}}}{A_{\mathrm{eff},\mathrm{O}viii}}·4\pi d_{\mathrm{LMC}}^2=(3.6\pm 1.7)\times {10}^{38}{\mathrm{erg}\mathrm{s}}^{-1},\end{array}$$(4)

where EW_O VIII = 1.05 × 10⁻⁹ erg is the energy of a O VIII photon, $A_{\mathrm{eff}}^{\mathrm{eRO}}=1043$ cm² is the eROSITA effective area in the O VIII band, and d_LMC = 50 kpc is the distance to the LMC. Therefore, we obtain L_X = (0.1 − 0.7)% L_SNe, LMC.

While this scenario is thus consistent with energy budget estimates, no clear morphological feature relates the emission to 30 Dor or the LMC disk in our maps except for the central position of the LMC. In addition, WIM gas seen in Hα emission associated with the LMC is found to fill only the closest ∼10 deg (Smart et al. 2023), leaving a connection with the X-rays detected in this work possible but uncertain. Unfortunately, concerning the soft X-ray emission detected in this work and characterized by prominent O VII and O VIII lines, the energy resolution of eROSITA prevents us from detecting a line shift or broadening of the diffuse emission. Recent and future instruments probing the soft X-ray line width with a resolution of electronvolts (e.g., XRISM, Athena, and LEM) will be crucial to further test this scenario.

4.3. Possible connection to star formation in the disk

In regions of similar brightness and/or extent (i.e., large angular scales), the faint X-ray emission has been explained by the overlap of several SNRs and/or winds from hot stars. One notable example in addition to single SNRs is the Eridanus-Orion superbubble. Superbubbles are connected to SNRs and hot OB star associations that are usually found in giant molecular clouds (Nishimura et al. 2015; Abdullah & Tielens 2020) and have already been associated with radio loops (Bracco et al. 2023). A superbubble is thus a reasonable candidate for an explanation of the Goat Horn complex. Therefore, we checked for the location of HII star-forming nearby regions. HII regions are usually found in the densest parts of the Galactic disk (i.e., at low Galactic absolute latitudes). A recent survey of these regions provides information on their distribution along the Galactic plane (Anderson et al. 2014). HII regions are not detected nearby the center of the Goat Horn complex (i.e., nearby the LMC). However, they may be located at lower latitudes in projection. A sharp decrease at around l = 300 − 330 deg along Galactic longitudes is observed in the distribution of the HII regions. A recent catalog of OB associations found these structures to be distributed throughout the Galactic disk in projection. In particular, the longitude range l = 280 − 290 deg exhibits a small concentration of OB associations, which might be connected to the X-ray emission in the Goat Horn complex. In general, by assuming a spherical geometry, the Goat Horn complex physical diameter is equal to its distance from us given its apparent angular size Δθ ∼ 45 deg. The OB star associations at similar longitude are located in a range of 2–3 kpc from the Sun. This would imply a physical radius of 1–1.5 kpc. We note that this size differs from (i.e., is much larger than) the size of typical superbubbles of ∼10² pc. Another tracer for warm ionized gas produced in superbubbles is the Hα line emission. The Eri-Ori superbubble is clearly detected in large angular scale maps of this tracer (Gaustad et al. 2001; Haffner et al. 2003). In contrast, neither the eROSITA bubbles nor the Goat Horn complex are detected in the same maps. However, this may be explained by the larger distance (i.e., fainter emission) of the Goat Horn complex from us compared to the Eri-Ori superbubble.

Other than superbubbles, regions of enhanced star formation activity in the disk of the Milky Way have been mapped through a cross match between massive young stellar objects, HII regions and methanol masers (Urquhart et al. 2014), producing a map of the clumps of massive stars in the disk of the Milky Way. Clumps of massive stars are found at Galactic longitudes as low as l = 280 deg, similarly to the OB associations. Unfortunately, the eastern bound of the Goat Horn complex is unconstrained because of the brighter emission produced by the western edge of the southern eROSITA bubble at around l = 290 deg. Therefore, we can only place an upper limit on the Galactic longitude reached by the Goat Horn complex of l < 290 deg. This limit is close to the edge of the longitude range of active star-forming regions, but still consistent with it.

An additional possibility lies in the connection with gas that is heated and expelled by the stellar formation and evolution activity into the spiral arms of the Milky Way. A recent map of the closest arms (Xu et al. 2023) provides the direction where the LoS is tangential to the arms. We show their positions in the chart map of Fig. 6. The Carina spiral arm seems to match the longitude range of the Goat Horn. If related to it, the relative distance of 3–4 kpc to this arm would also imply a similar size of the Goat Horn complex, thus reaching heights smaller than but still comparable to those of the eROSITA bubble. In this scenario, the apparent (i.e., projected) connection of radio loop XII with the central regions of the Milky Way could be explained by the actual proximity of these regions.

The question now is whether the local star formation toward the Carina arm can explain the observed X-ray luminosity, We computed an X-ray flux attributed to the Goat Horn complex of $F_{\mathrm{O}\mathrm{V}\mathrm{I}\mathrm{I}\mathrm{I}}^{\mathrm{GH}}=(7.8\pm 0.3)\times {10}^{-10}\mathrm{erg}{\mathrm{s}}^{-1}{\mathrm{cm}}^{-2}$ by integrating the observed O VIII intensity within a 15 deg radius from its center, located at about the LMC position. When the Goat Horn complex is placed at the distance of the Carina arm tangential point of d_Car ≃ 6.3 kpc (Xu et al. 2023), the implied X-ray luminosity is $L_{\mathrm{O}\mathrm{V}\mathrm{I}\mathrm{I}\mathrm{I}}^{\mathrm{GH}}=4\pi d_{\mathrm{Car}}^2·F_{\mathrm{O}\mathrm{V}\mathrm{I}\mathrm{I}\mathrm{I}}^{\mathrm{GH}}=3.7\times {10}^{36}\mathrm{erg}{\mathrm{s}}^{-1}$. An average star formation surface density of (2.2 ± 0.8) × 10⁻³ M_⊙ yr⁻¹ kpc⁻² has been inferred for the solar neighborhood (Spilker et al. 2021). If the Goat Horn complex lies within the Milky Way, given its longitude range close to −90 deg, its distance from the Galactic center has to be similar or equal to ours. The star formation rate in the disk may thus be of similar magnitude than the one computed in the solar neighborhood. The rate of supernovae explosions per stellar mass in Milky Way-like galaxies is estimated to be $f_{\text{SN}}=8.8\times {10}^{-3\hspace{0.17em}\text{SN}\hspace{0.17em}{M}_{\odot }^{-1}}$ (Horiuchi et al. 2011; Adams et al. 2013). Therefore, by assuming a supernova energy output of ∼10⁵¹ erg, the supernovae luminosity per unit area (in the Milky Way stellar disk) is about Σ_L ≃ 6 × 10³⁸ erg s⁻¹ kpc⁻². An area with a linear size of 0.1 kpc placed at about the position of the Carina arm is thus sufficient to produce the observed soft X-ray flux of the Goat Horn complex, thus providing a good match between theoretical values and observations. Although the connections with star-forming regions in the disk of the Milky Way seem promising, a strong physical link with the emission of the Goat Horn complex can neither be confirmed nor ruled out at the moment.

5. Conclusion

We summarize the new observations and results discussed in this work below.

• A large and coherent patch of soft X-ray emission exceeding the background, extending up to ∼1000 deg², is found around the Dorado region. This feature (called the Goat Horn complex) has not previously been recognized as a coherent and separate structure of the sky background.

• We observe foreground dust clouds absorbing the soft X-ray emission. These clouds are found as far as 500 pc from the Sun in the available 3D extinction data.

• An additional brightening of the X-ray emission, characterized by an arc-shaped morphology centered at about the LMC position, is blended with the emission from the Goat Horn complex, at least in projection.

• We find two X-ray intensity jumps across the boundary of the enhanced X-ray feature (i.e., a front).

• The spectrum of the excess emission at the front is consistent with the spectrum produced by a collisionally ionized hot plasma.

• The region delimiting the front (north) and the boundary of the Goat Horn complex (south) coincide with an arc-shaped radio polarization enhancement previously known as radio loop XII.

• We find a temperature gradient increase across the front that is anticorrelated with the X-ray intensity.

• An anticorrelation between the X-ray brightness and the polarized radio intensity is also detected across loop XII. The two detected anticorrelations are characteristic of cold fronts.

• The lower limit to the distance prevents the Goat Horn complex from being associated with a single supernova remnant.

• The morphology of the Galactic magnetic field lines is aligned with radio loop XII (i.e., upstream of the front) and along a foreground dust cloud. The lines are less ordered in the region outside the boundaries of the Goat Horn complex. The inner regions of the Goat Horn complex are characterized by a low fractional polarization at both 2.3 GHz and at 23–30 GHz, disfavoring depolarization (physical or instrumental) as the source of the low values.

• Absorption of X-ray photons by bright quasar due to $O_6^{+}$ ions is small. These measurements translate into tight upper limits on the O VII column density, leaving little room for the presence of an extended hot atmosphere around the LMC to explain the observed X-ray emission from the Goat Horn complex.

Our work discussed different possible interpretations of the Goat Horn complex. The emission of the Goat Horn complex might still be explained by combinations of the models presented above. Future measurements of the width of the high-ionization lines emitting X-rays by upcoming missions (e.g., XRISM, LEM, and Athena) will be crucial in order to distingush between different emission components and to better constrain the source distance.

Acknowledgments

This work is based on data from eROSITA, the soft X-ray instrument aboard SRG, a joint Russian-German science mission supported by the Russian Space Agency (Roskosmos), in the interests of the Russian Academy of Sciences represented by its Space Research Institute (IKI), and the Deutsches Zentrum für Luft- und Raumfahrt (DLR). The SRG spacecraft was built by Lavochkin Association (NPOL) and its subcontractors, and is operated by NPOL with support from the Max Planck Institute for Extraterrestrial Physics (MPE). The development and construction of the eROSITA X-ray instrument was led by MPE, with contributions from the Dr. Karl Remeis Observatory Bamberg & ECAP (FAU Erlangen-Nuernberg), the University of Hamburg Observatory, the Leibniz Institute for Astrophysics Potsdam (AIP), and the Institute for Astronomy and Astrophysics of the University of Tübingen, with the support of DLR and the Max Planck Society. The Argelander Institute for Astronomy of the University of Bonn and the Ludwig Maximilians Universität Munich also participated in the science preparation for eROSITA. The eROSITA data shown here were processed using the eSASS software system developed by the German eROSITA consortium. NL, GP and XZ acknowledge financial support from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program HotMilk (grant agreement No. [865637]). GP also aknowledges support from Bando per il Finanziamento della Ricerca Fondamentale 2022 dell’Istituto Nazionale di Astrofisica (INAF): GO Large program and from the Framework per l’Attrazione e il Rafforzamento delle Eccellenze (FARE) per la ricerca in Italia (R20L5S39T9).

References

All Figures

Fig. 1. O VII and O VIII narrow-band eRASS1 maps. The images are centered at the LMC coordinates (l, b) = (280, −33.7) deg. All the maps presented in this work use the same projection. The solid white line represents the great circles b = 0 deg and l = 0 deg. The dotted white lines are separated by Δl = 30 deg in longitude and Δb = 30 deg in latitude. Left column: O VII (top) and O VIII (bottom) eRASS1 maps (Locatelli et al. 2024; Zheng et al. 2024a). Right column: deabsorbed intensity of the warm-hot CGM component of the Milky Way, extracted from the O VII (top) and O VIII (bottom) eRASS1 maps (i.e., left column). We refer to Locatelli et al. (2024) for details of the CGM intensity retrieval and of the deabsorption method.

In the text

Fig. 2. S/N map of the O VIII deabsorbed image. S/N is the RMS value of the deabsorbed O VIII map at 220 < l < 250 deg and fixed b (±5 deg). The longitude range was selected to represent the intensity of the soft X-ray background. In order to enhance the diffuse emission, the map was smoothed with a Gaussian kernel with a full width at half maximum of 1 deg, and bright sources on scales smaller than 1 deg were removed. Coherent patches at S/N ≳ 2 are associated with discrete extended sources, as labeled. The new Goat Horn complex is defined by the coherent S/N ≳ 2 patch within the dashed blue circle.

In the text

Fig. 3. O VIII/O VII temperature proxy maps. Contributions from the local hot bubble foreground, the instrumental background, the cosmic X-ray background, and the foreground absorption were removed from the maps before we computed the ratio (see the text and Zheng et al. 2024a for further details). Lower panel: ratio of the narrow-band images centered on the O VIII and O VII emission lines. Upper panel: corresponding temperature under the assumption that the narrow-band energy ranges are dominated by O VII and O VIII lines, produced by collisionally ionized plasma.

In the text

	Fig. 4. S-PASS 2.3 GHz radio polarization maps (Carretti et al. 2019). The dashed white line represents the location of polarized radio loop XII as originally defined by Vidal et al. (2015). Upper panel: linear polarization P. Lower panel: fractional polarization p ≡ P/I.
In the text

	Fig. 5. Linear integral convolution of the 30 GHz Planck Stokes parameter Q and U (Planck Collaboration IV 2020). The Q and U Stokes parameters were rotated in order to produce angles parallel to the magnetic field lines.
In the text

Fig. 6. Chart map of the sources and features used and discussed in this work. The map is the same as Fig. 1. The dotted lines perpendicular to the Galactic disk and centered at b = 0 show directions tangential to the closest spiral arms of the Milky Way. By increasing l (i.e., right to left), they represent the Carina (green dots, l = 284 deg), Norma (orange dots, l = 306 deg), Centaurus (yellow dots, l = 312 deg), and Perseus (purple dots, l = 229 deg) arms. The large white dots represent the positions of the SMC and LMC. The empty white squares show the LoS characterized by high absorption, shown in Fig. 8. The cyan stars show the positions of the three bright X-ray sources with a detected O VII absorption line at z = 0: PKS 0558-504, LMC X-3, and 1H 0419-577, listed by decreasing latitude. The dashed white line shows the position of radio loop XII. The rainbow lines show the data we used to build the profiles and correlations in Figs. 9 and 11. The orange triangles show the positions of the on-source (upward) and background (downward) diffuse emission spectra shown in Fig. 10.

In the text

	Fig. 7. WHAM map of Hα emission from the Milky Way (Haffner et al. 2003).
In the text

Fig. 8. Column density NH of the X-ray absorbing material as a function of distance from the Sun in the solar neighborhood. Top panel: differential NH. Bottom panel: cumulative NH for the same measurements as in the top panel (solid lines). The dashed lines and hatched regions around them show the NH value with its uncertainty, as retrieved from the HI4PI survey in the same directions as indicated by the labeled colors in the top panel (see also the empty white squares in the chart map of Fig. 6).

In the text

Fig. 9. Radial profiles across the bright arc found within the Goat Horn complex. The sky paths from which the profiles were extracted are shown in Fig. 6. The rainbow colors match the data in this figure. Δθ ≡ 0 is defined at coordinates (l, b) = (280, −34). The hatched area shows the region dominated by a sharp intensity transition in the X-ray data. Top panel: eROSITA soft X-ray intensity (0.5–1.0 keV). Central panel: temperature proxy (right vertical axis), as derived from the ratio of eROSITA narrow-band images centered around the O VII and O VIII emission lines (left vertical axis). Bottom panel: polarization intensity P along the same paths (S-PASS). The broad central peak matches the position of loop XII along the paths.

In the text

Fig. 10. 0.3–1.2 keV spectrum of the excess emission at the position of the bright arc. The spectrum from a background region from outside the Goat Horn complex was subtracted from the data. The on- and off-source (i.e., background) regions are indicated by the upward and downward orange triangles in Fig. 6, respectively. The black and red lines show the best fits for different assumptions on the metal abundance of the hot plasma (Z = Z⊙ and Z = 0.1 Z⊙, respectively).

In the text

Fig. 11. Scatter plots of the quantities presented in Fig. 9. The symbols and colors of all the points correspond to those of Fig. 9. Only the points included in the region of loop XII (see the hatched region in Fig. 9) are plotted with color in order to highlight their trends. The sets of points are fit with a linear relation. The corresponding Pearson correlation coefficients (R) are reported in the legend. The dashed and dotted lines fit the south (squares) and southwest (circles) points, respectively. The gray and red lines show the fit for all and for the loop XII region points, respectively.

In the text

	Fig. 12. Hα intensity vs. optical reddening from dust in the region of the Goat Horn complex.
In the text