Caught in the Act: A Metal-rich High-velocity Cloud in the Inner Galaxy

HD 156359 (at l, b, d = 32868, −1452, 9 kpc) is one of the best-studied GC sight lines (Sembach et al. 1991, 1995). The sight line lies in the inner Galaxy at the boundary of the southern Fermi Bubble and within the eROSITA X-ray bubble (Predehl et al. 2020), a region of enhanced X-ray emission (see Figure 1). This sight line also passes close to a complex of small HVCs dubbed “Complex WE” by Wakker & van Woerden (1991), less than 1° from one of the densest cores. Sembach et al. (1991) classified HD 156359 as a O9.7 Ib–II star on the basis of stellar photospheric lines and wind profiles in its UV spectrum. The spectral type and apparent magnitude of the star yield a spectroscopic distance of 11.1 ± 2.8 kpc. However, recent Gaia Early Data Release 3 parallax measurements (Bailer-Jones et al. 2021) place HD 156359 at a distance of kpc, which implies a z-distance below the plane of 2.3 kpc. We adopt the Gaia distance in our analysis.

The sight line toward HD 156359 intersects at least three spiral arms, the Sagittarius, Scutum, and Norma arms. Spiral-arm models from Vallée (2017) give the expected velocities of the Sagittarius, Scutum, and Norma arms at approximately −10, −55, and −103 km s⁻¹, respectively. Thus, absorption components observed at these velocities can be attributed to gas in (or associated with) the spiral arms. However, the spiral arms cannot explain the HVC observed at v_LSR = 125 km s⁻¹.

High-ion absorption toward HD 156359 was first observed with the International Ultraviolet Explorer (Sembach et al. 1991) and later by the Goddard High Resolution Spectrograph (GHRS; Sembach et al. 1995). The high-ion information in the Far-Ultraviolet Spectroscopic Explorer (FUSE) and Hubble Space Telescope (HST) Space Telescope Imaging Spectrograph (STIS) spectra of this sight line has not been previously published; the sight line is not included in the FUSE O vi survey of Galactic disk sight lines by Bowen et al. (2008), as that survey was restricted to ∣b∣ < 10°. We also include analysis of a single archival optical spectrum of HD 156359 taken with the Fiber-fed Extended Range Optical Spectrograph (FEROS) at the European Southern Observatory (ESO) at La Silla. Details of all observations of spectra used in this paper are described below.

HD 156359 was observed by FUSE (Moos et al. 2000) under programs P101, S701, and U109 between 2000 April 12 and 2006 April 25 (PIs Sembach, Andersson, and Blair, respectively). The raw spectra were downloaded from the Mikulski Archive for Space Telescopes (MAST) FUSE archive, and the CalFUSE pipeline (v3.2.2; Dixon & Kruk 2009) was used to reduce and extract the spectra. Data from the SiC, LiF1, and LiF2 channels were used for the analysis. A detailed explanation of the data reduction as well as refinements to the CalFUSE data-reduction procedures can be found in Wakker et al. (2003) and Wakker (2006). The spectra have a signal-to-noise ratio (S/N) ∼ 12–26 per resolution element and a velocity resolution of 20 km s⁻¹ (FWHM). The data were binned by 3 pixels for the fitting analysis. The FUSE wavelength coverage is ∼912–1180 Å.

A single exposure of HD 156359 was obtained with HST/STIS on 2003 March 26 under program 9434 (PI: Lauroesch) using the E140M grating. The data were downloaded from the MAST HST archive and reduced using calstis (v.3.4.2; Dressel et al. 2007). The echelle orders were combined to create a single continuous spectrum, and in regions of order overlap spectral counts were combined to increase the S/N. The data have S/N ∼ 10–38 per resolution element and a FWHM velocity resolution of 6.5 km s⁻¹, i.e., spectral resolution (λ/Δλ) ∼ 45,800. The STIS E140M wavelength coverage is ∼1160–1725 Å. Wavelengths and velocities for the absorption-line features are given in the local standard of rest (LSR) reference frame, where the correction factor v_LSR − v_Helio = −0.35 km s⁻¹ for the direction toward HD 156359.

A spectrum of HD 156359 was captured using the FEROS at the ESO La Silla 2.2 m telescope on 2006 April 30 (PI: Bouret; PID: 077.D-0635). The retrieved archival product covers the wavelength range ∼3565–9214 Å with R = 48,000 and FWHM = 6.25 km s⁻¹. The primary data product was reduced automatically using the FEROS Data Reduction Software pipeline version fern/1.0 (Kaufer et al. 1999). In the automated reduction process, the bias is subtracted using overscan regions and bad columns are replaced by the mean values of the neighboring columns. The orders are rectified and then extracted using the standard method. Next, the extracted spectra are flat-fielded and wavelength calibrated, then rebinned to a constant dispersion of 0.03 Å. Finally, the individual orders are combined into a single one-dimensional spectrum. As the archival reduced data are not flux calibrated, the continuum of the stellar spectrum was normalized using the linetools software package (Prochaska et al. 2017).

Given the spectral type of HD 156359 (O9.7 Ib–II; Sembach et al. 1991), the stellar continuum placement requires careful consideration. In addition to estimating the stellar continuum through comparison with FUSE spectra of stars of similar spectral type, we also constructed a TLUSTY model (Lanz & Hubeny 2003) of this type of star as a continuum-placement guide. This was particularly useful in regions where the interstellar absorption was stronger.

The adopted TLUSTY model has T_eff = 26,000 K, log g = 3.0, v_turb = 10.0 km s⁻¹, and is rotationally broadened by 90 km s⁻¹ in order to match the Si iii 1300 Å triplets. We matched the observed UV wind signatures found in the observed FUSE and STIS spectra, where all observed spectra are adjusted for the stellar radial velocity v_rad = −82 km s⁻¹ (Gontcharov 2006). The model is reddened by an E(B − V) = 0.09, using the mean Fitzpatrick & Massa (2007) extinction curve (extrapolated to 912 Å), and Lyman series absorption for a foreground H i column of N(H i ) = 6.0 × 10²⁰ cm⁻² is also included (Sembach et al. 1991). The model is then scaled by 1.25 × 10⁻²⁰ in order to match observations to ∼10%. Finally, all spectra were binned to 0.1 Å, ∼15–30 km s⁻¹, depending on wavelength; see Figure 2. When consulting the model as a guide for interstellar line continua, additional tweaks of about 5% were needed to match local continua. Our subsequent column-density measurements account for the uncertainty inherent in the continuum-placement process.

Zoom In Zoom Out Reset image size

The FUSE spectrum of HD 156359 covers almost the entire H i Lyman series, from Lyman-β at 1025 Å down to the Lyman limit at 912 Å. However, we limited our fitting to H i lines in regions where the stellar continuum was better defined and showed the least contamination from stellar features and/or interstellar molecular lines, which are prolific in the FUSE bandpass.

We selected H i λλ923, 926, 930, and 937 (see Figure 3) to derive the H i column density. We fit a continuum to each of these lines locally. In order to account for sensitivity to continuum placement, we consider a high- and low-continuum placement for our column-density measurement across the wavelength range λ917–944. H i λ923 lies in a noisy region of the upper Lyman lines frequently associated with the Inglis–Teller effect (Inglis & Teller 1939), in which overlapping stellar absorption lines have merged to produce the appearance of a depressed continuum. We perform a simultaneous Voigt-profile fit to λλ926, 930, 937 with the positions of the low-velocity absorption initially based on weak interstellar medium (ISM) lines. We derive log N_{H i} = 15.54 ± 0.02 ± 0.05 in the HVC at +125.6 km s⁻¹, where the first error is the statistical error due to photon noise and the second is the systematic continuum-placement error. The result of these fits is shown Table 2 and in Figure 3, where we show that the simultaneous fit is also consistent with the noisier H i λ923 profile. We include a detailed explanation of our continuum-placement procedures and how we considered contamination from stellar features in Appendix A.

Zoom In Zoom Out Reset image size

An important step in measuring H i absorption in FUSE spectra is to decontaminate H₂ absorption. Fortunately, the FUSE LiF1 and LiF2 channels provide us with an opportunity to model isolated H₂ lines. This allows us to account for the MW’s molecular contribution to the interstellar absorption lines blended with H i in the SiC2A spectrum. We conducted an H₂ decontamination, and describe and illustrate this process in detail in Appendix B.

After placing a smooth continuum through the O i line at 1302 Å in the STIS E140M spectrum, we noticed a small absorption feature near +125 km s⁻¹ (see Figure 3). This feature spans 5 pixels and has a significance of 3.2σ (EW/σ_EW = 3.2). Since only one STIS exposure exists, we cannot confirm the O i detection in another data set, and this feature is not seen in the much weaker O i line at 1039 Å in the FUSE spectrum. To explore whether this feature is real, we compared the absorption profile to another low-ion line. Figure 4 shows an apparent column-density profile comparison of the O i 1302 Å line to the weak Si ii 1526 Å line. We chose λ1526 because it is unsaturated, unblended, and seen in a region with high S/N. Although noisier, the profile of O i λ1302 is very similar to Si ii λ1526 in the velocity region near +125 km s⁻¹, and the ratio of their apparent column densities is flat with velocity. To confirm this similarity, the inset plot in Figure 4 shows a linear fit to the ratio over 5 pixels (over two resolution elements) with a slope equal to zero within the margin of error, as expected for a genuine O i detection. Therefore, we proceed with treating the O i λ1302 feature as real on the basis that (1) the line is detected at 3.2σ significance, (2) there is close kinematic consistency with Si ii, and (3) the feature is centered at the same velocity as multiple other ions, e.g., H i, C ii, N ii, and Ca ii. After applying a low and high continuum to this region, we measure log N_{O i} = 12.43 ± 0.08 ± 0.07, which includes the statistical error and a continuum-placement (systematic) error. This measurement is the foundation of our metallicity measurement in the HVC, which we discuss in Section 4.

Zoom In Zoom Out Reset image size

We detected C ii λ1334, Si ii λλ1190, 1193, 1304, 1526, N ii λ1083, Si iii λ1206, and Fe iii λ1122 in the HVC near +125 km s⁻¹ in the FUSE and STIS spectra (see Figure 3). Our absorption-line measurements of these lines are given in Table 2. Achieving a simultaneous Voigt-profile fit using all Si lines is hindered by difficulties in continuum placement for the stronger lines since large spans of the variable stellar continuum are absorbed. Instead, we adopt the Voigt-profile measurement log N_{Si ii} = 12.94 ± 0.07 for the weakest unblended line available at 1526 Å. We show that it is a good fit for λλ1190, 1193, 1304 in Figure 3.

Although detected, N ii 1083 lies in a portion of the FUSE spectrum where the stellar continuum is highly variable over a short range in wavelength. Comparing the neighboring low-velocity H₂ J3 1084 Å line to other J3 lines in regions with smooth continua reveals that the continuum must drop significantly in this region, making it difficult to estimate log N_{N ii} . For this reason, our measurement for N ii is an upper limit. Fe ii 1144 has a low absorption profile in the vicinity of the HVC. We applied a high, middle, and low continuum across the absorbing region of λ1144 and calculate a significance of 2.2σ after including a continuum-placement error. We find a 3σ limiting column density of log N_{Fe ii} ≤ 12.99; however, given the Fe iii detection in the HVC, we use the Fe iii column density in metallicity calculations going forward. In addition to the high-velocity absorption we observe for Fe iii λ1122, we detect an intermediate-velocity cloud (IVC) centered near +86 km s⁻¹ with log N = 12.98 ± 0.13.

The archival ESO FEROS optical spectrum covers the Ca ii K and H and Na i D lines; see Figure 5. Telluric H₂O absorption lines are present in the velocity range of the HVC; however, there is a nondetection of high-velocity Na i, and we find a 3σ limiting column-density limit of log N_{Na I} < 9.45. We see absorption in Ca ii across 6 pixels at 123.7 km s⁻¹ and measure an apparent optical depth (AOD) column density of log N_{Ca II} = 10.68 ± 0.01 for Ca ii 3934. We also performed a simultaneous Voigt-profile fit to Ca ii λ3934, 3969 and find log N = 10.63 ± 0.10. The resulting b-value for this small component has a high error. However, since the log N value from Voigt-profile fitting agrees with the AOD measurement within the margin of error, we adopt its value.

Zoom In Zoom Out Reset image size

The C iv and Si iv profiles are complex in the STIS E140M spectrum, showing multiple distinct low- and intermediate-velocity components, in addition to the HVC. The high-ion absorption features for C iv, Si iv, and N v lie on well-defined continua due to their corresponding broad and smooth stellar P Cygni wind profiles, which gradually elevates their stellar continua in the region from −100 to +200 km s⁻¹ for this star, as shown in Figure 6. The high-ion continuum fitting was also guided by fits to the GHRS data previously published by Sembach et al. (1991, 1995).

Zoom In Zoom Out Reset image size

We detect weak, but significant, absorption in the HVC near +130 km s⁻¹ in C iv λ1548, Si iv λ1393, and N v λλ1238, 1242. We calculated the detection significance in each line by adding the continuum-placement error in quadrature with the statistical (photon-noise) error. We found a 3.6σ significance for the high-velocity C iv λ1548 feature. We observe a slightly weaker feature near +160 km s⁻¹ on the rising C iv λ1548 wind profile, which we attribute to either noise or a detector artifact, as it does not appear in any of the other high ions at the same velocity. The much weaker C iv λ1550 feature falls at the peak of the P Cygni stellar wind profile and we do not detect any significant absorption in the HVC in this line. We measure a 3.2σ significance for the stronger Si iv 1393 feature, but only measure 2.0σ for the weaker Si iv λ1402 line. For the N v λλ1238, 1242 lines, we measure detection significances of 3.0σ and 4.9σ, respectively, where the higher significance for 1242 reflects the high S/N near this line, even though it is intrinsically weaker.

We performed Voigt-profile fitting to C iv, Si iv, and N v high ions with the initial positions of the components and b-values determined from the C iv 1548 Å line. The position and b-values were allowed to vary freely and the results for the fits are given in Table 2. We note the detection of an IVC centered near +78 km s⁻¹ in both C iv and Si iv with log N =13.57 ± 0.08 and 13.09 ± 0.10, respectively.

We observe high-velocity absorption in O vi 1031 and 1037 in the LiF1 channel of the FUSE spectrum at v_LSR = 141 km s⁻¹. However, we only include λ1031 in our analysis, because λ1037 lies on the steep blueward side of the O vi P Cygni profile and suffers from significant blending with C ii* 1037 and H₂ J1 1038 Å. We measure the O vi λ1031 HVC feature at 7.5σ after including a continuum-placement error.

The wavelength region around O vi 1031 Å contains several absorption lines which can serve as contaminants, most notably HD 6–0 R(0) λ1031.915, Cl i λ1031.507, and several lines of molecular hydrogen including H₂ 6–0 P(3) λ1031.192 and H₂ 6–0 R(4) λ1032.351. To gauge the effect of contamination by HD 6–0 R(0) λ1031.915, we examined other HD lines of similar oscillator strength that are isolated from interstellar absorption, such as 5–0 R(0) λ1042.850, 7–0 R(0) λ1021.460, and 8–0 R(0) λ1011.461. We detect no HD molecular absorption in these lines and conclude that no subtraction of the HD 6–0 R(0) line at λ1031.915 from the O vi profile is necessary. For Cl i, a small amount of absorption in Cl i λ1347 is present near 0 km s⁻¹ in the STIS spectrum and we simultaneously fit Cl i λ1031 to derive its contribution to the O vi profile. For molecular hydrogen, H₂ 6–0 R(4) λ1032.351 is the most relevant potential contaminant as it overlaps with the high-velocity component in O vi λ 1031. We modeled other isolated H₂ J4 lines in the FUSE spectrum, e.g., 5–0 R(4) λ1044.543 and 4–0 R(4) λ1057.381 (see description in Appendix B). Through simultaneous fitting of those lines with H₂ 6–0 R(4) λ1032, we account for the small amount of H₂ present in the O vi high-velocity absorption. A similar procedure was followed to account for contamination from H₂ 6–0 P(3) λ1031.192 near v_LSR = −200 km s⁻¹ using the isolated H₂ J3 lines 6–0 R(3) λ1028.986 and 5–0 P(3) λ1043.503. The resulting Voigt-profile fit and O vi column density are shown in Figure 6 and Table 2.

To assess the role of photoionization as an ionization mechanism in the HVC, as well as characterize its physical conditions and elemental abundances, we incorporate Cloudy (v.17.02; Ferland et al. 2017) photoionization models into our analysis. We designed a multidimensional grid of models for the low and intermediate ions, assuming they arise in the same gas phase as the H i and are photoionized by the same incident radiation fields. The model assumes a plane-parallel slab of uniform density exposed to both the MW disk-halo radiation field model (Fox et al. 2005, 2014; Barger et al. 2013) and the extragalactic UV background radiation field (Khaire & Srianand 2019). We do not include the effects of ionizing radiation produced by any hot collisionally ionized plasma, which has been shown to also contribute to the photoionization of the gas (Wakker et al. 2012).

The MW disk-halo radiation field is a nonisotropic composite field consisting of hard and soft radiation fields from the escaping UV ionizing flux from O–B stars. The escape fraction for the hard component is derived from the intensity of photoionized Hα emission measured in HVCs at known distances (Bland-Hawthorn & Maloney 1999; Putman et al. 2003), whereas the escape fraction for the soft field is derived from the mean observed opacity of the observed spectrum in the solar neighborhood. At the position of HD 156359 a distance of ∼2.3 kpc below the GC, the magnitude of the escaping hydrogen ionizing flux (log Φ = 5.9) in the MW disk-halo model dominates the extragalactic background radiation by ∼2 orders of magnitude. We consider the degree to which the Cloudy results are sensitive to the magnitude of the ionizing flux by running models with the best-fit magnitude changed by ±0.5 dex (a factor of 3 times higher or lower); see Appendix C.

We ran our models for a grid of metallicity values with [Z/H] varying from −0.5 to +1.5 in steps of 0.1 dex, each over a range of hydrogen number densities log (n_H/cm⁻³) from −3 to 0, in order to explore possible metallicities and densities. We determined the best-fit log n_H for each metallicity model by matching the observed column-density ratio of Si iii/Si ii to the model value (see the top panel of Figure C2). The Si iii/Si ii ratio was chosen because both ions have unsaturated detections in the HVC, and using a ratio of adjacent ions from the same element (Si) minimizes metallicity or depletion effects from different metal ions. Using the pairs of metallicity and log n_H determined from the Si iii/Si ii ratio, we narrow the range of metallicities by identifying which models are consistent with the observed log N_{O i} = 12.43 ± 0.11, as illustrated in the bottom-left panel of Figure C2. We then run an even finer grid of metallicity values in increments of 0.02 dex over the narrowed metallicity and number density region to determine the range of metallicities and densities allowed by the data.

We find that metallicities in the range 0.18 ≤ [O/H] ≤ 0.51 and densities from −1.69 ≤ log (n_H/cm⁻³) ≤ −1.37 are consistent with the data, giving a model best fit of [O/H] = at log n_H = −1.53 ± −0.16. We repeated this procedure for C ii (see bottom-right panel of Figure C2), which is expected to be weakly depleted (Jenkins 2009), and find that the range 0.21 ≤ [C/H] ≤ 0.42 overlaps with the observed data, giving a model best fit of [C/H] = at log n_H = −1.49. The agreement between the oxygen-based metallicity and the carbon-based metallicity lends support to our methodology and to the robustness of the supersolar metallicity we infer.

We found that repeating this process for a hydrogen ionizing flux ±0.50 dex higher and lower gives a similar best fit of [O/H] = +0.36 and [O/H] = +0.30 ± 0.20, respectively. Thus, our main results are consistent for log Φ = 5.9 ± 0.5. Additionally, we tested whether our choice of radiation field could be influencing our results. To do this, we repeated our suite of models with Cloudy's built-in unextinguished local interstellar radiation field “Table ISM” (Black 1987), which is a commonly used physically motivated field available for modeling interstellar gas, and found [O/H] = +0.23 ± 0.13, again consistent with our main results; see Table C1 and Figure C1 in Appendix C for a detailed description of how we addressed uncertainties in the incident radiation fields by varying their intensity and shape.

The metallicity for this cloud ranks among the highest UV-based metallicities of any HVC observed thus far, along with the HVC at −125 km s⁻¹ toward M5-ZNG1 (l = 39, b =+477 at z = +5.3 kpc) with [O/H] = +0.22 ± 0.10 (Zech et al. 2008). M5-ZNG1 was also observed with FUSE and STIS E140M spectra. Their observed metallicity is not corrected for ionization effects, but is likely higher than reported, given the measured low log N_{H i} = 16.50 ± 0.06 in the cloud and that oxygen shows a positive ionization correction when log N(H i) < 18.5 (Bordoloi et al. 2017). The metallicity of the HD 156359 HVC is also on the high end of the range of <20% to 3 times solar reported by Ashley et al. (2022) in their survey of metallicities of Fermi Bubble HVCs. We note that a supersolar abundance in the inner Galaxy may not be unexpected, given the oxygen abundance gradients reported from emission-line measurements in H ii regions (see Wenger et al. 2019; Arellano-Córdova et al. 2021).

In ISM abundance work, [O i/H i] is often considered a robust indicator of elemental abundance [O/H], since (1) charge-exchange reactions closely couple the ionization state of hydrogen and oxygen together (Field & Steigman 1971), and (2) oxygen is only lightly depleted onto dust grains (Jenkins2009). However, the assumption that [O i/H i] = [O/H] breaks down when N(H i) is so low that the gas is optically thin or when the ionizing photon flux is extremely high (Viegas 1995). In the HVC at +125 km s⁻¹ toward HD 156359, log N(H i) is only 15.54 ± 0.05, so an ionization correction must be made to account for the column densities of all observed neutral and ion stages.

Our Cloudy models directly provide the ionization corrections for all our observed ion stages. We define the ionization correction as the difference between the model (true) elemental abundance and the observed ion abundance, i.e.,

The resulting ion abundances, ionization corrections, and ionization-corrected elemental abundances are given in Table 3. The relationships between the ionization corrections and the hydrogen number density for [Z/H] = 0.36 are shown in the top panel of Figure 7 and the ionization-corrected abundances are shown in the middle panel of Figure 7.

Zoom In Zoom Out Reset image size

The positive ionization correction of IC(O i) = from the photoionization modeling results in a high gas-phase oxygen abundance of +0.36. We calculate gas-phase abundances of [C/H] = 0.27 ± 0.13, [N/H] = ≤0.93, [Si/H] = 0.03 ± 0.09, [Fe/H] = 0.06 ± 0.17, and [Ca/H] = −0.20 ±0.11 (see Table 3).

Following convention, we define the depletion δ_O(X) of each refractory element X as the ionization-corrected abundances of that element compared to the ionization-corrected oxygen abundance, i.e.,

where oxygen represents an undepleted (or lightly depleted) volatile element. In this formalism, a negative value of δ_O(X) means that element X is depleted relative to oxygen. This method assumes that the total (gas+dust) abundances are solar, which is often assumed to apply to the Galactic ISM (Savage & Sembach 1996), though local ISM abundance variations cannot be ruled out in some sight lines (De Cia et al. 2021). We show δ_O(X) for the low ions in the bottom panel of Figure 7 compared to F_*, the line-of-sight depletion strength factor, for [X/O] determined for low-depletion (F_* = 0) and high-depletion (F_* = 1) gas from the comprehensive ISM gas-phase element depletions study of Jenkins (2009).

For the HVC toward HD 156359, we find a low value of δ_O(C) = −0.09 ± 0.17, i.e., carbon shows no significant depletion. This is consistent with the low values of [C/O] measured in Galactic ISM gas (Jenkins 2009). For the other low ions, we find a 3σ upper limit for the nitrogen depletion δ_O(N) ≤ +0.57 and low depletions for silicon, iron, and calcium of δ_O(Si) = −0.33 ± 0.14, δ_O(Fe) = −0.30 ± 0.20, and δ_O(Ca) = −0.56 ± 0.16, respectively. To summarize the final results of the CLOUDY modeling after the dust depletion levels have been derived, Figure 8 shows the model curves of the detected ions shifted by their respective depletions at [X/H] = +0.36 and log n_H = −1.53. The ability of this model to match the observed column densities shows that all low-ion measurements can be explained by photoionization once dust depletion effects are accounted for.

Zoom In Zoom Out Reset image size

A low level of dust depletion for the HVC is consistent with the well-known Routly–Spitzer (RS) effect, in which the amount of dust observed in HVCs decreases significantly at higher LSR velocity (see Routly & Spitzer 1952; Siluk & Silk 1974; Vallerga et al. 1993; Smoker et al. 2011; Ben Bekhti et al. 2012; Murga et al. 2015). The RS effect is typically traced by the N(Na i)/N(Ca ii) ratio, which has been found to significantly decrease with increasing LSR velocity. Although the RS effect was historically interpreted as a dust depletion effect, it may also be related to ionization effects, and likely differs depending on the environment of the cloud. We measure an 3σ upper limit of log N(Na i)/N(Ca ii) ≤ −1.2 in the +124 km s⁻¹ HVC toward HD 156359 from the optical FEROS data, which is on the low end of the observed ratios for HVCs (Smoker et al. 2011; Ben Bekhti et al. 2012; Murga et al. 2015), and is also lower than is typically observed at low velocities. In summary, the dust depletion pattern we derive in the HVC is consistent with other HVCs and with the RS effect, even though the cloud has high metallicity.

The observed high ions have column densities that are too high to be explained by the Cloudy photoionization models described in Section 4.1. Although these models do not include energetic photons from the radiation field intrinsic to the Fermi Bubbles or in situ ionizing photons from the hot gas itself, which could partially contribute to the photoionization of the C iv and Si iv gas, the models underproduce the observed Si iv and C iv column densities by ∼0.7 and 1.0 dex, respectively. The N v and O vi column densities are underproduced by orders of magnitudes (see Table C1). These discrepancies imply that a separate ionization mechanism is likely required for C iv and Si iv, and is definitely required for N v and O vi. Here, we explore collisional ionization in gas with temperatures of T ≳ 10⁵ K as the separate mechanism. We consult the collisional ionization models from Gnat & Sternberg (2007) to determine the exact range of temperatures allowed by the high-ion column densities and their ratios. While we considered the more recent models from Gnat (2017) that include photoionization effects from the extragalactic UV background, we do not adopt them because HD 156359 lies in close proximity to the radiation field of the Galactic plane, which is much stronger. We considered both the collisional ionization equilibrium (CIE) and nonequilibrium regimes.

A single-temperature solution explaining all four high ions (Si iv, C iv, N v, O vi) in the HVC is ruled out, as no such solution exists to the observed high-ion column densities. Instead, we find that a two-phase solution is needed, with one temperature explaining the N_{C iv} /N_{Si iv} ratio and another explaining N_{O vi} /N_{N v} . This two-phase model is consistent with the kinematic information in the UV spectra, where Si iv and C iv show very similar line profiles but O vi is offset in velocity centroid. The two-phase high-ion model is illustrated in Figure 10, which shows the observed high-ion ratios of N_{C iv} /N_{Si iv} and N_{O vi} /N_{N v} compared to model predictions from Gnat & Sternberg (2007) for solar ([Z/H] = 0) and supersolar ([Z/H] = +0.3) metallicities.

Zoom In Zoom Out Reset image size

For the C iv/Si iv phase, we find two possible solutions for the temperature. First, the nonequilibrium isochoric (constant volume) and isobaric (constant pressure) models give a low-temperature solution at T = 10^4.2−4.4 K, where the lower end corresponds to the solar-metallicity isochoric model and the higher end with the supersolar isobaric model. Second, the CIE model returns a higher temperature T = 10^4.9 K. We are inclined to adopt the nonequilibrium (lower-temperature) solution, as plasma near T = 10⁵ K is at the peak of the cooling curve, where oxygen dominates the radiative cooling and the gas can cool faster than it recombines, reaching a nonequilibrium state. For the O vi/N v phase, a single-temperature solution for the observed log N_{O vi} /N_{N v} = 1.01 is found at T = 10^5.4 K for all models (see bottom panel of Figure 10).

In conclusion, we can successfully model the high-ion plasma in the HVC toward HD 156359 as containing collisionally ionized gas at two temperatures: a cooler phase seen in C iv and Si iv at T = 10^4.2−4.4 K, and a hotter phase seen in N v and O vi at T = 10^5.4 K. We note that the velocity offset of ∼20 km s⁻¹ for O vi from the low ions is consistent with offsets on the order of tens of kilometers per second predicted by conductive interface models of evaporation between hot and cool gas (Bohringer & Hartquist 1987; Borkowski et al. 1990) or from models for dynamical cooling flows (Wakker et al. 2012). Given the small velocity centroid uncertainties from the Voigt-profile modeling (∼1–5 km s⁻¹; see Table 2), the ∼20 km s⁻¹ offset is significant and could indicate we are tracing hot gas in an interface layer moving at a higher velocity than the cooler gas inside the cloud.