# Detection of the Galactic warm neutral medium in H i 21-cm absorption

Detection of the Galactic warm neutral medium in H i 21-cm absorption ABSTRACT We report a deep Giant Metrewave Radio Telescope search for Galactic H i 21-cm absorption towards the quasar B0438−436, yielding the detection of wide, weak H i 21-cm absorption, with a velocity-integrated H i 21-cm optical depth of 0.0188 ± 0.0036 km s−1. Comparing this with the H i column density measured in the Parkes Galactic All-Sky Survey gives a column density-weighted harmonic mean spin temperature of 3760 ± 365 K, one of the highest measured in the Galaxy. This is consistent with most of the H i along the sightline arising in the stable warm neutral medium. The low-peak H i 21-cm optical depth towards B0438−436 implies negligible self-absorption, allowing a multi-Gaussian joint decomposition of the H i 21-cm absorption and emission spectra. This yields a gas kinetic temperature of $$T_{\rm k} \le (4910 \pm \lt t\gt 1900)$$ K, and a spin temperature of $$T_{\rm s} = (1000 \pm 345)$$ K for the gas that gives rise to the H i 21-cm absorption. Our data are consistent with the H i 21-cm absorption arising from either the stable WNM, with $$T_{\rm s} \ll T_{\rm k}$$, $$\rm T_k \approx 5000$$ K, and little penetration of the background Lyman-α radiation field into the neutral hydrogen, or the unstable neutral medium, with $$T_{\rm s} \approx T_{\rm k} \approx 1000\,{\rm K}$$. ISM: atoms, ISM: general, radio lines: ISM 1 INTRODUCTION Neutral hydrogen (H  i) is a key constituent of the interstellar medium (ISM) of galaxies, making up almost half the ISM in the Milky Way. The earliest H  i 21-cm studies (e.g. Clark 1965) found evidence that H  i exists in two phases, the cold neutral medium (CNM), which gives rise to the narrow, deep components seen in H  i 21cm absorption spectra towards background radio sources, and the warm neutral medium (WNM), which contributes to the relatively smooth and broad H  i 21-cm emission profiles but has very weak H  i 21-cm absorption. Similarly, theoretical models indicate that Galactic H  i exists in one of the above two stable phases, with low kinetic temperatures (Tk ≈ 40–300 K) and high number densities (≈10–100 cm−3) in the CNM, and high temperatures (Tk ≈ 5000–8000 K) and low number densities (≈0.1–1 cm−3) in the WNM (e.g. Field, Goldsmith & Habing 1969; McKee & Ostriker 1977; Wolfire et al. 1995, 2003). H  i at intermediate temperatures (≈500–5000 K) is expected to be unstable and to quickly move into one of the two stable phases. Over the last few decades, H  i 21-cm absorption studies have made much progress in characterizing conditions in the CNM in the Galaxy (e.g. Radhakrishnan et al. 1972; Dickey, Terzian & Salpeter 1978; Payne, Salpeter & Terzian 1983; Heiles & Troland 2003; Roy et al. 2013), finding that the CNM indeed has Tk ≈ 20–300 K, consistent with theoretical expectations. However, the complexity of the H  i 21-cm emission profiles and the difficulty of detecting the WNM in H  i 21-cm absorption have made it difficult to verify the prediction of a stable warm phase with Tk ≈ 5000–8000 K. Indeed, attempts at measuring the WNM temperature by fitting multi-Gaussian models to either a combination of H  i 21-cm emission and absorption spectra (Heiles & Troland 2003a); Heiles & Troland 2003b) or interferometric H  i 21-cm absorption spectra (Kanekar et al. 2003; Roy, Kanekar & Chengalur 2013; Murray et al. 2015) have found significant fractions of the H  i to be in the thermally unstable phase, with Tk ≈ 1000 K. Very few of the above Gaussian components show temperatures within or larger than the stable WNM range, ≈5000–8000 K. However, most of the sightlines in the above studies have complex H  i 21-cm absorption profiles, implying that one is typically searching for WNM absorption in the presence of far stronger, multicomponent CNM absorption. The best sightlines for a reliable detection of the WNM in H  i 21-cm absorption, and for an accurate estimate of the WNM kinetic temperature, are those with the least complexity in the H  i 21-cm absorption profile. Our earlier interferometric H  i 21-cm absorption survey of compact radio sources achieved root-mean-square (RMS) H  i 21-cm optical depth noise values of ≈0.001 per ≈1 km s−1 channel and yielded detections of H  i 21-cm absorption in 33 of 34 sightlines. The sole sightline without a detection of absorption, towards the quasar B0438−436, has one of the lowest H  i column densities of the sample ($$\rm {\it N}_{HI}= 1.29 \times 10^{20}$$ cm−2; Kalberla & Haud 2015) and is at a high Galactic latitude, implying low CNM contamination. The absence of CNM absorption on this sightline, despite the high sensitivity of our survey, suggests that B0438−436 is a good target for a clean search for the WNM in absorption. We hence used the Giant Metrewave Radio Telescope (GMRT) to carry out a deep search for H  i 21-cm absorption towards B0438−436, the results of which are described in this Letter. 2 OBSERVATIONS, DATA ANALYSIS, AND RESULTS Our GMRT search for Galactic H  i 21-cm absorption towards B0438−436 was carried out over 2011 May 6–15, using the L-band receivers, with a total observing time of ≈30 h over six observing sessions. The GMRT Software Backend (GSB) was used as the correlator, with a bandwidth of ≈1.067 MHz subdivided into 512 channels, yielding a total velocity coverage of ≈225  km s−1 and a velocity resolution of ≈0.43  km s−1. We used frequency-switching at the first GMRT local oscillator, on B0438−436 itself, to calibrate the system passband; the switching was carried out every 5 min with a throw of 5 MHz. Observations of 3C 48 and 3C 147 at the start and end of each run were used to calibrate the flux density scale. Since B0438−436 is a phase calibrator for the GMRT, no additional phase calibration was necessary. The use of the GSB for these observations implied a far superior data quality and a larger total velocity coverage than those of our earlier GMRT observations of B0438−436 (Roy et al. 2013). Allied with our higher sensitivity, this significantly improved our ability to detect wide H  i 21-cm absorption lines. All data were analysed in the Astronomical Image Processing System (Greisen 2003), following standard data editing, calibration, self-calibration, imaging, and continuum subtraction procedures (e.g. Roy et al. 2013). The data from the observing run on May 6 were found to be severely affected by radio frequency interference (RFI) and were hence excluded from the later analysis. The task cvel was finally used to shift the residual visibilities to the local standard of rest (LSR) velocity frame. These visibilities were then imaged to produce a spectral cube, using natural weighting and excluding baselines shorter than 1 kλ to reduce contamination from H  i 21-cm emission within the primary beam. The H  i 21-cm spectrum was obtained by taking a cut through the spectral cube at the location of the quasar; a second-order baseline was then fitted to line-free regions, and subtracted out, to obtain the final H  i 21-cm absorption spectrum. We measured a flux density of 4.2 Jy for B0438−436 from the final GMRT continuum image; the error on this value is dominated by uncertainties in the GMRT flux density scale, which we estimate to be ≈15  per  cent. The flux density is in reasonable agreement with the value of 5.0 Jy listed in the Very Large Array Calibrator Manual. For Galactic H  i 21-cm absorption studies, the spectral RMS noise depends on the observing frequency due to the contribution from the brightness temperature of the H  i emission in the beam (Roy et al. 2013). We followed the procedure of Roy et al. (2013) to combine the brightness temperature measured in the Parkes Galactic All-Sky Sky Survey (GASS; McClure-Griffiths et al. 2009; Kalberla & Haud 2015) H  i 21-cm emission spectrum with the GMRT system temperature (≈73 K at 1420 MHz) to determine the RMS noise spectrum. Note that this correction is very small for B0438−436 as the peak H  i 21-cm brightness temperature in the GASS spectrum is only ≈2 K, far lower than the GMRT system temperature at the observing frequency. The final optical depth RMS noise on the GMRT spectrum is ≈1.0 × 10−3 per 0.43  km s−1 channel. Fig. 1(A) shows our final Galactic H  i 21-cm absorption spectrum towards B0438−436, with H  i 21-cm optical depth plotted versus LSR velocity, after Hanning-smoothing and re-sampling the spectrum, at a velocity resolution of ≈0.86  km s−1 (with an optical depth RMS noise of ≈0.72 × 10−3 per 0.86  km s−1 channel). Weak absorption can be seen close to zero LSR velocity. Fig. 1(B) shows the spectrum after additional smoothing and re-sampling at a velocity resolution of ≈6.0  km s−1; the RMS optical depth noise on this spectrum is ≈2.4 × 10−4 per ≈6.0  km s−1. The H  i 21-cm absorption is now clearly visible, detected at ≈5.3σ significance: the velocity-integrated H  i 21-cm optical depth (over line channels) is ∫τdV = (0.0188 ± 0.0036) km s−1. The rest of the spectrum, away from the absorption feature, shows no evidence for any structure in the baseline. The Kolmogorov–Smirnov rank-1 and Anderson–Darling tests find that the offline channels are consistent with being drawn from a Gaussian distribution. Figure 1. View largeDownload slide The final GMRT H  i 21-cm absorption spectrum towards B0438−436, at (A) the Hanning-smoothed and re-sampled velocity resolution of ≈0.86 km s−1 (top panel), and (B) smoothed to, and re-sampled at, a resolution of ≈6.0  km s−1 (bottom panel). Figure 1. View largeDownload slide The final GMRT H  i 21-cm absorption spectrum towards B0438−436, at (A) the Hanning-smoothed and re-sampled velocity resolution of ≈0.86 km s−1 (top panel), and (B) smoothed to, and re-sampled at, a resolution of ≈6.0  km s−1 (bottom panel). 3 DISCUSSION Neutral hydrogen is usually characterized by two ‘temperatures’, the kinetic temperature, Tk, and the H  i 21-cm line excitation temperature, the spin temperature, Ts. In the case of the high-density CNM, the H  i hyperfine levels are expected to be thermalized by a combination of collisions and Lyman-α scattering, with the spin temperature approximately equal to the kinetic temperature (e.g. Wouthuysen 1952; Field 1958; Liszt 2001). Observationally, low spin temperatures (≈100 K) have indeed been found to be associated with the narrow H  i 21-cm absorption features that are expected to arise in the CNM (e.g. Dickey et al. 1978; Heiles & Troland 2003; Roy et al. 2013). However, in the case of the WNM, the low number density implies that collisions are not very effective in thermalizing the H  i 21-cm transition. Resonant scattering of Lyman-α photons is hence expected to be the main process in driving Ts towards Tk (e.g. Field 1965; Deguchi & Watson 1985; Liszt 2001). It has long been unclear whether the fraction of the Galactic Lyman-α background radiation threading into the WNM is sufficient to drive the WNM spin temperature to the kinetic temperature (e.g. Deguchi & Watson 1985; Liszt 2001). Theoretical studies suggest that Ts < Tk in the WNM due to both the low WNM number density and the expected low penetration of background Lyman-α photons into the WNM (e.g. Liszt 2001). A few early observational studies did obtain high spin temperature estimates, ≳5000 K (e.g. Kalberla, Mebold & Reich 1980; Payne, Salpeter & Terzian 1983; Kulkarni, Heiles & Dickey 1985), albeit usually on sightlines with complex H  i 21-cm absorption and with concerns about stray radiation affecting the H  i 21-cm emission spectra. Further, while Ts values ≳1000 K have recently been found at specific velocity ranges along multiple Galactic sightlines (e.g. Carilli, Dwarakanath & Goss 1998; Dwarakanath, Carilli & Goss 2002; Roy et al. 2013), it is not usually straightforward to estimate Tk at the same velocity ranges, making it difficult to test whether the H  i 21-cm hyperfine excitation is indeed sub-thermal in the WNM. Next, the H  i column density along a sightline is related to the velocity-integrated H  i 21-cm optical depth by the equation \begin{equation*}\rm \rm {\it N}_{HI}= 1.823 \times 10^{18} \langle {\it T}_s \rangle \int \tau _{21cm} d{\it V} \ ,, \end{equation*} (1) where $$\rm \langle {\it T}_s \rangle$$ is the column density-weighted harmonic mean spin temperature along the sightline. Note that $$\rm \langle {\it T}_s \rangle$$ is biased towards low CNM temperatures: for example, if the H  i along the sightline is equally divided between phases with $$\rm {\it T}_s \approx 100$$ K and $$\rm {\it T}_s \approx 8000$$ K, one would obtain $$\rm \langle {\it T}_s \rangle \approx 200$$ K. Even if 90  per  cent of the H  i along the sightline has $$\rm {\it T}_s \approx 8000$$ K, and 10  per cent has $$\rm {\it T}_s \approx 100$$ K, one would measure $$\rm \langle {\it T}_s \rangle \approx 900$$ K. Exacerbating this issue, spin temperatures in the WNM are expected to be lower than the kinetic temperature and are hence likely to be significantly lower than the assumed 8000 K (e.g. Liszt 2001). As such, high values of $$\rm \langle {\it T}_s \rangle$$, ≳1000 K, can only be obtained for sightlines with almost all the gas in the WNM. In the case of the sightline towards B0438−436, the GASS H  i emission spectrum yields an H  i column density of 1.29 × 1020 cm−2. Combining this with our measured integrated H  i 21-cm optical depth of ∫τ21 cmdV = (0.018 ± 0.036) km s−1 then yields $$\rm \langle {\it T}_s \rangle = 3760 \pm 765$$ K. This is one of the highest column density-weighted harmonic mean spin temperatures ever measured along a sightline in the Milky Way, comparable to values seen in low-metallicity damped Lyman-α absorbers at high redshifts (e.g. Kanekar & Chengalur 2003; Kanekar et al. 2014). The sightline towards B0438−436 is clearly dominated by the WNM, with almost no cold gas present towards the quasar. The lack of CNM along this low-$$\rm {\it N}_{HI}$$ sightline is consistent with the H  i column density threshold of ≈2 × 1020 cm−2 that has been suggested for the formation of significant amounts of cold atomic gas in the Milky Way (Kanekar, Braun & Roy 2011). Figs 2(A) and (B) compare the GMRT H  i 21-cm absorption spectrum towards B0438−436 with the GASS H  i 21-cm emission spectrum at a neighbouring location, with both spectra plotted versus LSR velocity, after smoothing to, and re-sampling at, a velocity resolution of 6.0  km s−1. It is clear that the peak H  i 21-cm absorption occurs at a slightly lower velocity (≈0  km s−1) than the peak of the H  i 21-cm emission (≈ +8  km s−1). Indeed, the bulk of gas detected in H  i 21-cm emission appears to not be detected in H  i 21-cm absorption. Figure 2. View largeDownload slide The three panels of the figure show (A) the H i 21-cm emission spectrum from the GASS survey (Kalberla & Haud 2015), (B) the GMRT H i 21-cm optical depth spectrum, and (C) the spin temperature Ts, for the sightline towards B0438−436, plotted versus LSR velocity, in km s−1. The spectra have been smoothed to, and re-sampled at, a velocity resolution of ≈6.0  km s−1. See main text for discussion. Figure 2. View largeDownload slide The three panels of the figure show (A) the H i 21-cm emission spectrum from the GASS survey (Kalberla & Haud 2015), (B) the GMRT H i 21-cm optical depth spectrum, and (C) the spin temperature Ts, for the sightline towards B0438−436, plotted versus LSR velocity, in km s−1. The spectra have been smoothed to, and re-sampled at, a velocity resolution of ≈6.0  km s−1. See main text for discussion. Fig. 2(C) shows the spin temperature, $$\rm {\it T}_s = {\it T}_B/(1 - {\it e}^{-\tau })$$, plotted versus LSR velocity, at velocities where the H  i 21-cm emission is detected at ≥5σ significance. Velocity channels with detections of H  i 21-cm absorption at ≥3σ significance are shown as filled circles with error bars, while channels with non-detections of H  i 21-cm absorption (i.e. with <3σ significance) are shown as 3σ lower limits to the spin temperature. While the two velocity channels with ≥3σ detections of absorption have $$\rm {\it T}_s \approx 1750$$ K, the two higher-velocity channels have Ts(3σ) ≳ 2500 K. These $$\rm {\it T}_s$$ values lie in the expected range for WNM spin temperatures (e.g. Liszt 2001). The above results demonstrate that most or all of the gas towards B0438−436 is warm, with high spin temperatures, ≳1750 K, consistent with an origin in the stable WNM, at all channels with significant H  i 21-cm emission. We also carried out a Gaussian decomposition of the H  i 21-cm absorption profile of Fig. 1(A) at a resolution of 0.86  km s−1; a single Gaussian, shown as the solid curves in Figs 1(A) and (B), provides an excellent fit to the spectrum; this has a full width at half maximum (FWHM) of 12.0 ± 3.1 km s−1, corresponding to a temperature of 3150 ± 1635 K. We emphasize that this estimate corresponds to the maximum allowed value of the kinetic temperature, sometimes referred to as the ‘Doppler temperature’ ($$\rm {\it T}_D$$; e.g. Roy et al. 2013), as non-thermal motions may contribute to the line broadening. This estimate of the Doppler temperature is very similar to the column density-weighted harmonic mean spin temperature estimate of $$\rm \langle {\it T}_s \rangle = 3760 \pm 765$$ K and may suggest that the gas towards B0438−436 is in the unstable phase, with Ts ≈ Tk. However, we emphasize that the above kinetic temperature estimate is consistent (within ≈1σ significance) with the standard stable WNM kinetic temperature range (≈5000–8000 K). This conclusion is rendered even more unlikely because Ts ≈ 1750 K at the velocity channels with the strongest H  i 21-cm absorption. Deeper H  i 21-cm absorption studies are needed to accurately estimate Tk via the Gaussian decomposition approach. Attempts have been made in the literature to carry out a joint multi-Gaussian decomposition of the H  i 21-cm emission and the H  i 21-cm absorption spectra (e.g. Heiles & Troland 2003; Murray et al. 2014). These have usually encountered difficulties in handling H  i 21-cm ‘self-absorption’ along the sightline by foreground CNM clouds of their own, and of background CNM and WNM, emission. The present sightline towards B0438−436 is very interesting in this context because the low peak H  i 21-cm optical depth (≈0.001) means that self-absorption is not an issue. The sightline thus allows the possibility of a joint Gaussian decomposition of the H  i 21-cm absorption and emission spectra without any effects of self-absorption. We carried out a joint multi-Gaussian fit to the H  i 21-cm emission and H  i 21-cm absorption spectra, the former at the original velocity resolution of ≈0.82 km s−1 and the latter after Hanning-smoothing and resampling, at a resolution of ≈0.86 km s−1. We allowed for H  i 21-cm absorption by a single Gaussian component and H  i 21-cm emission by multiple Gaussian components. In the fit, the position and FWHM of one of the emission components were constrained to be the same as that of the absorption component (using multiple absorption components did not yield a better fit, in terms of an improved reduced chi-square value). The best-fit model, with a reduced chi-square $$\chi ^2_N = 1.00$$, contains 5 emission components and one absorption component. The model is plotted in Figs 3(A) and (B), with the residuals from the fit displayed in the lower two panels; the residuals are seen to be consistent with noise. The properties of the different Gaussian components are summarized in Table 1, whose columns contain (1) the component number, (2) the peak brightness temperature of the component, in K, (3) the central LSR velocity, in km s−1, (4) the component FWHM, in km s−1, (5) the peak H  i 21-cm optical depth, or the 3σ upper limit to the peak optical depth (estimated at a velocity resolution equal to the component FWHM), (6) the kinetic temperature of the component, $$\rm {\it T}_k$$, and (7) the component spin temperature $$\rm {\it T}_s$$, or the 3σ lower limit to the spin temperature. Figure 3. View largeDownload slide Results of the multi-Gaussian joint decomposition of the (A) H i 21-cm emission and (B) H i 21-cm absorption spectra; the former spectrum is at a resolution of ≈0.82  km s−1, while the latter is at a resolution of ≈0.86  km s−1. The top panels show the best-fit model (solid curve) overlaid on the two spectra, while the bottom two panels show the residuals from the fit. Figure 3. View largeDownload slide Results of the multi-Gaussian joint decomposition of the (A) H i 21-cm emission and (B) H i 21-cm absorption spectra; the former spectrum is at a resolution of ≈0.82  km s−1, while the latter is at a resolution of ≈0.86  km s−1. The top panels show the best-fit model (solid curve) overlaid on the two spectra, while the bottom two panels show the residuals from the fit. Table 1. Gaussian joint decomposition of the H  i 21-cm emission and absorption spectra. Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Notes. For each component, the columns are: aThe peak brightness temperature; bThe central LSR velocity; cThe peak H  i 21-cm optical depth, or, for non-detections, the 3σ upper limit on the peak H i 21-cm optical depth at a velocity resolution equal to the line FWHM; dThe kinetic temperature inferred from the FWHM; eThe spin temperature or, for components 2–5, the 3σ lower limit to the spin temperature. View Large Table 1. Gaussian joint decomposition of the H  i 21-cm emission and absorption spectra. Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Notes. For each component, the columns are: aThe peak brightness temperature; bThe central LSR velocity; cThe peak H  i 21-cm optical depth, or, for non-detections, the 3σ upper limit on the peak H i 21-cm optical depth at a velocity resolution equal to the line FWHM; dThe kinetic temperature inferred from the FWHM; eThe spin temperature or, for components 2–5, the 3σ lower limit to the spin temperature. View Large Initially, we note that the third component in Table 1 is extremely wide, with an FWHM of ≈57 km s−1. Given the relatively high Galactic latitude of B0438−436 (b ≈ −41°), it appears unlikely that this large broadening arises from either Galactic rotation or turbulence. We suspect that it may be due to a residual spectral baseline in the GASS spectrum, although we note that a similar wide component is also visible in the H  i 21-cm spectrum from the Leiden–Argentine–Bonn survey (Kalberla et al. 2005). For now, the origin of the broad component remains unclear. The component with detected H  i 21-cm absorption has an FWHM of 15.0 ± 2.9  km s−1, in good agreement with that obtained from the fit to the H  i 21-cm absorption alone (12.0 ± 3.1 km s−1). The inferred upper limit on the kinetic temperature is then 4910 ± 1900 K in the stable WNM temperature range. We emphasize, as above, that this is formally an estimate of the Doppler temperature TD and hence yields only an upper limit to Tk due to the possibility of non-thermal broadening. We consider the two extreme possibilities for line broadening: (1) weak non-thermal motions, for which the kinetic temperature is approximately equal to the Doppler temperature, i.e. Tk ≈ TD = (4910 ± 1900) K, and (2) strong non-thermal broadening, in which case Tk ≪ TD. In the first case, the detected H  i 21-cm absorption arises from the stable WNM. The spin temperature of this component is 1000 ± 345 K, significantly lower than the kinetic temperature. Figs 2 and 5 of Liszt (2001) suggest that this would require the gas to be at low pressures, P/k ≲ 1000 cm−3 K, and with almost no Galactic Lyman-α background penetrating into the H  i. The data would then support a picture of little coupling between the WNM and the background Lyman-α radiation field. In the second case, of strong non-thermal broadening, the measured Doppler temperature yields only an upper limit to Tk. However, the measured spin temperature gives the constraint Tk ≥ 1000, since Tk ≥ Ts in the WNM. This implies the range $$\rm 1000 \, K \, \le {\it T}_{\rm k}\le 4910\, K$$ for the component detected in absorption. For strong non-thermal broadening, Tk ≪ TD, the detected H  i 21-cm absorption would then still arise in warm gas, albeit from the thermally unstable phase, with Tk ≈ Ts, i.e. with strong coupling between the unstable neutral medium and the background Lyman-α radiation field. We cannot rule out this possibility with the present data. Finally, the Doppler temperature of component 2 of Table 1 is marginally consistent with its origin in the stable WNM, albeit with large errors. It is interesting that this component has a relatively high lower limit on its spin temperature, Ts > 2730 K, which suggests a stronger coupling with the background Lyman-α radiation. The Doppler temperatures of components 4 and 5 appear consistent with an origin in the unstable phase, although again with large errors. Again, it is interesting that the lower limit on the spin temperature of component 5 is relatively high, approaching the Doppler temperature. We emphasize, however, that the errors in the Doppler temperature estimates are still quite large. Deeper observations of the H  i 21-cm absorption towards B0438−36 are needed to confirm the above results, both the absolute values of the Doppler temperatures and, hence, the relation between the spin and the kinetic temperatures in the stable WNM and the unstable neutral medium. 4 SUMMARY iWe have used the GMRT to detect wide, weak Galactic H  i 21-cm absorption towards the quasar B0438−436. The velocity-integrated H  i 21-cm optical depth is very low, ∫τdV = (0.0188 ± 0.0036) km s−1. Combining this with the H  i column density ($$\rm {\it N}_{HI}= 1.29 \times 10^{20}$$ cm−2) measured from the GASS H  i 21-cm emission spectrum on a neighbouring sightline yields a column density-weighted harmonic mean spin temperature of $$\rm \langle {\it T}_s \rangle = (3760 \pm 765)$$ K, far higher than typical in the Galaxy. The high $$\rm \langle {\it T}_s \rangle$$ value indicates that there is little or no CNM along the sightline towards B0438−436. A detailed comparison between the H  i 21-cm emission and absorption spectra indicates that Ts ≥ 1760 K over the range of velocities showing the strongest H  i 21-cm emission, consistent with the spin temperature range expected in the stable WNM. The extremely low H  i 21-cm optical depth further implies that H  i 21-cm self-absorption is negligible along this sightline. We carry out a joint multi-Gaussian decomposition of the H  i 21-cm emission and the H  i 21-cm absorption profiles to obtain Tk ≤ (4910 ± 1900) K for the component with detected H  i 21-cm absorption, consistent with the stable WNM temperature range. The spin temperature of this component is (1000 ± 345) K, significantly lower than the kinetic temperature. This suggests that the H   21-cm line excitation is sub-thermal in the WNM along this sightline, possibly due to low gas pressure or low threading of the Galactic Lyman-α background into the WNM. However, the present data cannot rule out the possibility that the detected H  i 21-cm absorption might arise from warm gas in the thermally unstable phase, with significant non-thermal line broadening, and the gas spin temperature comparable to the kinetic temperature. ACKNOWLEDGEMENTS We thank the GMRT staff who have made these observations possible. The GMRT is run by the National Centre for Radio Astrophysics of the Tata Institute of Fundamental Research. NK acknowledges support from the Department of Science and Technology via a Swarnajayanti Fellowship (DST/SJF/PSA-01/2012-13). NR acknowledges support from the Infosys Foundation through the Infosys Young Investigator grant. REFERENCES Carilli C. L. , Dwarakanath K. S. , Goss W. M. , 1998 , ApJL , 502 , L79 CrossRef Search ADS Clark B. G. , 1965 , ApJ , 142 , 1398 CrossRef Search ADS Deguchi S. , Watson W. D. , 1985 , ApJ , 290 , 578 CrossRef Search ADS Dickey J. M. , Terzian Y. , Salpeter E. E. , 1978 , ApJS , 36 , 77 CrossRef Search ADS Dwarakanath K. S. , Carilli C. L. , Goss W. M. , 2002 , ApJ , 567 , 940 CrossRef Search ADS Field G. B. , 1958 , Proc. I. R. E. , 46 , 240 CrossRef Search ADS Field G. B. , 1965 , ApJ , 142 , 531 CrossRef Search ADS Field G. B. , Goldsmith D. W. , Habing H. J. , 1969 , ApJ , 155 , L149 CrossRef Search ADS Greisen E. W. , 2003 , in Heck A. , ed., Astrophysics and Space Science Library, Vol. 285, Information Handling in Astronomy – Historical Vistas , p. 109 , Kluwer Academic Publishers , Dordrecht Google Scholar CrossRef Search ADS Heiles C. , Troland T. H. , 2003a , ApJS , 145 , 329 CrossRef Search ADS Heiles C. , Troland T. H. , 2003b , ApJ , 586 , 1067 CrossRef Search ADS Kalberla P. M. W. , Haud U. , 2015 , A&A , 578 , A78 CrossRef Search ADS Kalberla P. M. W. , Mebold U. , Reich W. , 1980 , A&A , 82 , 275 Kalberla P. M. W. , Burton W. B. , Hartmann D. , Arnal E. M. , Bajaja E. , Morras R. , Pöppel W. G. L. , 2005 , A&A , 440 , 775 CrossRef Search ADS Kanekar N. et al. , 2014 , MNRAS , 438 , 2131 CrossRef Search ADS Kanekar N. , Chengalur J. N. , 2003 , A&A , 399 , 857 CrossRef Search ADS Kanekar N. , Subrahmanyan R. , Chengalur J. N. , Safouris V. , 2003 , MNRAS , 346 , L57 CrossRef Search ADS Kanekar N. , Braun R. , Roy N. , 2011 , ApJ , 737 , L33 CrossRef Search ADS Kulkarni S. R. , Heiles C. , Dickey J. M. , 1985 , ApJ , 291 , 716 CrossRef Search ADS Liszt H. , 2001 , A&A , 371 , 698 CrossRef Search ADS McClure-Griffiths N. M. et al. , 2009 , ApJS , 181 , 398 CrossRef Search ADS McKee C. F. , Ostriker J. P. , 1977 , ApJ , 218 , 148 CrossRef Search ADS Murray C. E. et al. , 2014 , ApJ , 781 , L41 CrossRef Search ADS Murray C. E. et al. , 2015 , ApJ , 804 , 89 CrossRef Search ADS Payne H. E. , Salpeter E. E. , Terzian Y. , 1983 , ApJ , 272 , 540 CrossRef Search ADS Radhakrishnan V. , Murray J. D. , Lockhart P. , Whittle R. P. J. , 1972 , ApJS , 24 , 15 CrossRef Search ADS Roy N. , Kanekar N. , Chengalur J. N. , 2013 , MNRAS , 436 , 2366 CrossRef Search ADS Roy N. , Kanekar N. , Braun R. , Chengalur J. N. , 2013 , MNRAS , 436 , 2352 CrossRef Search ADS Wolfire M. G. , Hollenbach D. , McKee C. F. , Tielens A. G. G. M. , Bakes E. L. O. , 1995 , ApJ , 443 , 152 CrossRef Search ADS Wolfire M. G. , McKee C. F. , Hollenbach D. , Tielens A. G. G. M. , 2003 , ApJ , 587 , 278 CrossRef Search ADS Wouthuysen S. A. , 1952 , AJ , 57 , 31 CrossRef Search ADS © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Monthly Notices of the Royal Astronomical Society: Letters Oxford University Press

# Detection of the Galactic warm neutral medium in H i 21-cm absorption

, Volume Advance Article (1) – May 17, 2018
6 pages

/lp/ou_press/detection-of-the-galactic-warm-neutral-medium-in-hi-21cm-absorption-uanWdRYiJK
Publisher
Oxford University Press
ISSN
1745-3925
eISSN
1745-3933
D.O.I.
10.1093/mnrasl/sly087
Publisher site
See Article on Publisher Site

### Abstract

ABSTRACT We report a deep Giant Metrewave Radio Telescope search for Galactic H i 21-cm absorption towards the quasar B0438−436, yielding the detection of wide, weak H i 21-cm absorption, with a velocity-integrated H i 21-cm optical depth of 0.0188 ± 0.0036 km s−1. Comparing this with the H i column density measured in the Parkes Galactic All-Sky Survey gives a column density-weighted harmonic mean spin temperature of 3760 ± 365 K, one of the highest measured in the Galaxy. This is consistent with most of the H i along the sightline arising in the stable warm neutral medium. The low-peak H i 21-cm optical depth towards B0438−436 implies negligible self-absorption, allowing a multi-Gaussian joint decomposition of the H i 21-cm absorption and emission spectra. This yields a gas kinetic temperature of $$T_{\rm k} \le (4910 \pm \lt t\gt 1900)$$ K, and a spin temperature of $$T_{\rm s} = (1000 \pm 345)$$ K for the gas that gives rise to the H i 21-cm absorption. Our data are consistent with the H i 21-cm absorption arising from either the stable WNM, with $$T_{\rm s} \ll T_{\rm k}$$, $$\rm T_k \approx 5000$$ K, and little penetration of the background Lyman-α radiation field into the neutral hydrogen, or the unstable neutral medium, with $$T_{\rm s} \approx T_{\rm k} \approx 1000\,{\rm K}$$. ISM: atoms, ISM: general, radio lines: ISM 1 INTRODUCTION Neutral hydrogen (H  i) is a key constituent of the interstellar medium (ISM) of galaxies, making up almost half the ISM in the Milky Way. The earliest H  i 21-cm studies (e.g. Clark 1965) found evidence that H  i exists in two phases, the cold neutral medium (CNM), which gives rise to the narrow, deep components seen in H  i 21cm absorption spectra towards background radio sources, and the warm neutral medium (WNM), which contributes to the relatively smooth and broad H  i 21-cm emission profiles but has very weak H  i 21-cm absorption. Similarly, theoretical models indicate that Galactic H  i exists in one of the above two stable phases, with low kinetic temperatures (Tk ≈ 40–300 K) and high number densities (≈10–100 cm−3) in the CNM, and high temperatures (Tk ≈ 5000–8000 K) and low number densities (≈0.1–1 cm−3) in the WNM (e.g. Field, Goldsmith & Habing 1969; McKee & Ostriker 1977; Wolfire et al. 1995, 2003). H  i at intermediate temperatures (≈500–5000 K) is expected to be unstable and to quickly move into one of the two stable phases. Over the last few decades, H  i 21-cm absorption studies have made much progress in characterizing conditions in the CNM in the Galaxy (e.g. Radhakrishnan et al. 1972; Dickey, Terzian & Salpeter 1978; Payne, Salpeter & Terzian 1983; Heiles & Troland 2003; Roy et al. 2013), finding that the CNM indeed has Tk ≈ 20–300 K, consistent with theoretical expectations. However, the complexity of the H  i 21-cm emission profiles and the difficulty of detecting the WNM in H  i 21-cm absorption have made it difficult to verify the prediction of a stable warm phase with Tk ≈ 5000–8000 K. Indeed, attempts at measuring the WNM temperature by fitting multi-Gaussian models to either a combination of H  i 21-cm emission and absorption spectra (Heiles & Troland 2003a); Heiles & Troland 2003b) or interferometric H  i 21-cm absorption spectra (Kanekar et al. 2003; Roy, Kanekar & Chengalur 2013; Murray et al. 2015) have found significant fractions of the H  i to be in the thermally unstable phase, with Tk ≈ 1000 K. Very few of the above Gaussian components show temperatures within or larger than the stable WNM range, ≈5000–8000 K. However, most of the sightlines in the above studies have complex H  i 21-cm absorption profiles, implying that one is typically searching for WNM absorption in the presence of far stronger, multicomponent CNM absorption. The best sightlines for a reliable detection of the WNM in H  i 21-cm absorption, and for an accurate estimate of the WNM kinetic temperature, are those with the least complexity in the H  i 21-cm absorption profile. Our earlier interferometric H  i 21-cm absorption survey of compact radio sources achieved root-mean-square (RMS) H  i 21-cm optical depth noise values of ≈0.001 per ≈1 km s−1 channel and yielded detections of H  i 21-cm absorption in 33 of 34 sightlines. The sole sightline without a detection of absorption, towards the quasar B0438−436, has one of the lowest H  i column densities of the sample ($$\rm {\it N}_{HI}= 1.29 \times 10^{20}$$ cm−2; Kalberla & Haud 2015) and is at a high Galactic latitude, implying low CNM contamination. The absence of CNM absorption on this sightline, despite the high sensitivity of our survey, suggests that B0438−436 is a good target for a clean search for the WNM in absorption. We hence used the Giant Metrewave Radio Telescope (GMRT) to carry out a deep search for H  i 21-cm absorption towards B0438−436, the results of which are described in this Letter. 2 OBSERVATIONS, DATA ANALYSIS, AND RESULTS Our GMRT search for Galactic H  i 21-cm absorption towards B0438−436 was carried out over 2011 May 6–15, using the L-band receivers, with a total observing time of ≈30 h over six observing sessions. The GMRT Software Backend (GSB) was used as the correlator, with a bandwidth of ≈1.067 MHz subdivided into 512 channels, yielding a total velocity coverage of ≈225  km s−1 and a velocity resolution of ≈0.43  km s−1. We used frequency-switching at the first GMRT local oscillator, on B0438−436 itself, to calibrate the system passband; the switching was carried out every 5 min with a throw of 5 MHz. Observations of 3C 48 and 3C 147 at the start and end of each run were used to calibrate the flux density scale. Since B0438−436 is a phase calibrator for the GMRT, no additional phase calibration was necessary. The use of the GSB for these observations implied a far superior data quality and a larger total velocity coverage than those of our earlier GMRT observations of B0438−436 (Roy et al. 2013). Allied with our higher sensitivity, this significantly improved our ability to detect wide H  i 21-cm absorption lines. All data were analysed in the Astronomical Image Processing System (Greisen 2003), following standard data editing, calibration, self-calibration, imaging, and continuum subtraction procedures (e.g. Roy et al. 2013). The data from the observing run on May 6 were found to be severely affected by radio frequency interference (RFI) and were hence excluded from the later analysis. The task cvel was finally used to shift the residual visibilities to the local standard of rest (LSR) velocity frame. These visibilities were then imaged to produce a spectral cube, using natural weighting and excluding baselines shorter than 1 kλ to reduce contamination from H  i 21-cm emission within the primary beam. The H  i 21-cm spectrum was obtained by taking a cut through the spectral cube at the location of the quasar; a second-order baseline was then fitted to line-free regions, and subtracted out, to obtain the final H  i 21-cm absorption spectrum. We measured a flux density of 4.2 Jy for B0438−436 from the final GMRT continuum image; the error on this value is dominated by uncertainties in the GMRT flux density scale, which we estimate to be ≈15  per  cent. The flux density is in reasonable agreement with the value of 5.0 Jy listed in the Very Large Array Calibrator Manual. For Galactic H  i 21-cm absorption studies, the spectral RMS noise depends on the observing frequency due to the contribution from the brightness temperature of the H  i emission in the beam (Roy et al. 2013). We followed the procedure of Roy et al. (2013) to combine the brightness temperature measured in the Parkes Galactic All-Sky Sky Survey (GASS; McClure-Griffiths et al. 2009; Kalberla & Haud 2015) H  i 21-cm emission spectrum with the GMRT system temperature (≈73 K at 1420 MHz) to determine the RMS noise spectrum. Note that this correction is very small for B0438−436 as the peak H  i 21-cm brightness temperature in the GASS spectrum is only ≈2 K, far lower than the GMRT system temperature at the observing frequency. The final optical depth RMS noise on the GMRT spectrum is ≈1.0 × 10−3 per 0.43  km s−1 channel. Fig. 1(A) shows our final Galactic H  i 21-cm absorption spectrum towards B0438−436, with H  i 21-cm optical depth plotted versus LSR velocity, after Hanning-smoothing and re-sampling the spectrum, at a velocity resolution of ≈0.86  km s−1 (with an optical depth RMS noise of ≈0.72 × 10−3 per 0.86  km s−1 channel). Weak absorption can be seen close to zero LSR velocity. Fig. 1(B) shows the spectrum after additional smoothing and re-sampling at a velocity resolution of ≈6.0  km s−1; the RMS optical depth noise on this spectrum is ≈2.4 × 10−4 per ≈6.0  km s−1. The H  i 21-cm absorption is now clearly visible, detected at ≈5.3σ significance: the velocity-integrated H  i 21-cm optical depth (over line channels) is ∫τdV = (0.0188 ± 0.0036) km s−1. The rest of the spectrum, away from the absorption feature, shows no evidence for any structure in the baseline. The Kolmogorov–Smirnov rank-1 and Anderson–Darling tests find that the offline channels are consistent with being drawn from a Gaussian distribution. Figure 1. View largeDownload slide The final GMRT H  i 21-cm absorption spectrum towards B0438−436, at (A) the Hanning-smoothed and re-sampled velocity resolution of ≈0.86 km s−1 (top panel), and (B) smoothed to, and re-sampled at, a resolution of ≈6.0  km s−1 (bottom panel). Figure 1. View largeDownload slide The final GMRT H  i 21-cm absorption spectrum towards B0438−436, at (A) the Hanning-smoothed and re-sampled velocity resolution of ≈0.86 km s−1 (top panel), and (B) smoothed to, and re-sampled at, a resolution of ≈6.0  km s−1 (bottom panel). 3 DISCUSSION Neutral hydrogen is usually characterized by two ‘temperatures’, the kinetic temperature, Tk, and the H  i 21-cm line excitation temperature, the spin temperature, Ts. In the case of the high-density CNM, the H  i hyperfine levels are expected to be thermalized by a combination of collisions and Lyman-α scattering, with the spin temperature approximately equal to the kinetic temperature (e.g. Wouthuysen 1952; Field 1958; Liszt 2001). Observationally, low spin temperatures (≈100 K) have indeed been found to be associated with the narrow H  i 21-cm absorption features that are expected to arise in the CNM (e.g. Dickey et al. 1978; Heiles & Troland 2003; Roy et al. 2013). However, in the case of the WNM, the low number density implies that collisions are not very effective in thermalizing the H  i 21-cm transition. Resonant scattering of Lyman-α photons is hence expected to be the main process in driving Ts towards Tk (e.g. Field 1965; Deguchi & Watson 1985; Liszt 2001). It has long been unclear whether the fraction of the Galactic Lyman-α background radiation threading into the WNM is sufficient to drive the WNM spin temperature to the kinetic temperature (e.g. Deguchi & Watson 1985; Liszt 2001). Theoretical studies suggest that Ts < Tk in the WNM due to both the low WNM number density and the expected low penetration of background Lyman-α photons into the WNM (e.g. Liszt 2001). A few early observational studies did obtain high spin temperature estimates, ≳5000 K (e.g. Kalberla, Mebold & Reich 1980; Payne, Salpeter & Terzian 1983; Kulkarni, Heiles & Dickey 1985), albeit usually on sightlines with complex H  i 21-cm absorption and with concerns about stray radiation affecting the H  i 21-cm emission spectra. Further, while Ts values ≳1000 K have recently been found at specific velocity ranges along multiple Galactic sightlines (e.g. Carilli, Dwarakanath & Goss 1998; Dwarakanath, Carilli & Goss 2002; Roy et al. 2013), it is not usually straightforward to estimate Tk at the same velocity ranges, making it difficult to test whether the H  i 21-cm hyperfine excitation is indeed sub-thermal in the WNM. Next, the H  i column density along a sightline is related to the velocity-integrated H  i 21-cm optical depth by the equation \begin{equation*}\rm \rm {\it N}_{HI}= 1.823 \times 10^{18} \langle {\it T}_s \rangle \int \tau _{21cm} d{\it V} \ ,, \end{equation*} (1) where $$\rm \langle {\it T}_s \rangle$$ is the column density-weighted harmonic mean spin temperature along the sightline. Note that $$\rm \langle {\it T}_s \rangle$$ is biased towards low CNM temperatures: for example, if the H  i along the sightline is equally divided between phases with $$\rm {\it T}_s \approx 100$$ K and $$\rm {\it T}_s \approx 8000$$ K, one would obtain $$\rm \langle {\it T}_s \rangle \approx 200$$ K. Even if 90  per  cent of the H  i along the sightline has $$\rm {\it T}_s \approx 8000$$ K, and 10  per cent has $$\rm {\it T}_s \approx 100$$ K, one would measure $$\rm \langle {\it T}_s \rangle \approx 900$$ K. Exacerbating this issue, spin temperatures in the WNM are expected to be lower than the kinetic temperature and are hence likely to be significantly lower than the assumed 8000 K (e.g. Liszt 2001). As such, high values of $$\rm \langle {\it T}_s \rangle$$, ≳1000 K, can only be obtained for sightlines with almost all the gas in the WNM. In the case of the sightline towards B0438−436, the GASS H  i emission spectrum yields an H  i column density of 1.29 × 1020 cm−2. Combining this with our measured integrated H  i 21-cm optical depth of ∫τ21 cmdV = (0.018 ± 0.036) km s−1 then yields $$\rm \langle {\it T}_s \rangle = 3760 \pm 765$$ K. This is one of the highest column density-weighted harmonic mean spin temperatures ever measured along a sightline in the Milky Way, comparable to values seen in low-metallicity damped Lyman-α absorbers at high redshifts (e.g. Kanekar & Chengalur 2003; Kanekar et al. 2014). The sightline towards B0438−436 is clearly dominated by the WNM, with almost no cold gas present towards the quasar. The lack of CNM along this low-$$\rm {\it N}_{HI}$$ sightline is consistent with the H  i column density threshold of ≈2 × 1020 cm−2 that has been suggested for the formation of significant amounts of cold atomic gas in the Milky Way (Kanekar, Braun & Roy 2011). Figs 2(A) and (B) compare the GMRT H  i 21-cm absorption spectrum towards B0438−436 with the GASS H  i 21-cm emission spectrum at a neighbouring location, with both spectra plotted versus LSR velocity, after smoothing to, and re-sampling at, a velocity resolution of 6.0  km s−1. It is clear that the peak H  i 21-cm absorption occurs at a slightly lower velocity (≈0  km s−1) than the peak of the H  i 21-cm emission (≈ +8  km s−1). Indeed, the bulk of gas detected in H  i 21-cm emission appears to not be detected in H  i 21-cm absorption. Figure 2. View largeDownload slide The three panels of the figure show (A) the H i 21-cm emission spectrum from the GASS survey (Kalberla & Haud 2015), (B) the GMRT H i 21-cm optical depth spectrum, and (C) the spin temperature Ts, for the sightline towards B0438−436, plotted versus LSR velocity, in km s−1. The spectra have been smoothed to, and re-sampled at, a velocity resolution of ≈6.0  km s−1. See main text for discussion. Figure 2. View largeDownload slide The three panels of the figure show (A) the H i 21-cm emission spectrum from the GASS survey (Kalberla & Haud 2015), (B) the GMRT H i 21-cm optical depth spectrum, and (C) the spin temperature Ts, for the sightline towards B0438−436, plotted versus LSR velocity, in km s−1. The spectra have been smoothed to, and re-sampled at, a velocity resolution of ≈6.0  km s−1. See main text for discussion. Fig. 2(C) shows the spin temperature, $$\rm {\it T}_s = {\it T}_B/(1 - {\it e}^{-\tau })$$, plotted versus LSR velocity, at velocities where the H  i 21-cm emission is detected at ≥5σ significance. Velocity channels with detections of H  i 21-cm absorption at ≥3σ significance are shown as filled circles with error bars, while channels with non-detections of H  i 21-cm absorption (i.e. with <3σ significance) are shown as 3σ lower limits to the spin temperature. While the two velocity channels with ≥3σ detections of absorption have $$\rm {\it T}_s \approx 1750$$ K, the two higher-velocity channels have Ts(3σ) ≳ 2500 K. These $$\rm {\it T}_s$$ values lie in the expected range for WNM spin temperatures (e.g. Liszt 2001). The above results demonstrate that most or all of the gas towards B0438−436 is warm, with high spin temperatures, ≳1750 K, consistent with an origin in the stable WNM, at all channels with significant H  i 21-cm emission. We also carried out a Gaussian decomposition of the H  i 21-cm absorption profile of Fig. 1(A) at a resolution of 0.86  km s−1; a single Gaussian, shown as the solid curves in Figs 1(A) and (B), provides an excellent fit to the spectrum; this has a full width at half maximum (FWHM) of 12.0 ± 3.1 km s−1, corresponding to a temperature of 3150 ± 1635 K. We emphasize that this estimate corresponds to the maximum allowed value of the kinetic temperature, sometimes referred to as the ‘Doppler temperature’ ($$\rm {\it T}_D$$; e.g. Roy et al. 2013), as non-thermal motions may contribute to the line broadening. This estimate of the Doppler temperature is very similar to the column density-weighted harmonic mean spin temperature estimate of $$\rm \langle {\it T}_s \rangle = 3760 \pm 765$$ K and may suggest that the gas towards B0438−436 is in the unstable phase, with Ts ≈ Tk. However, we emphasize that the above kinetic temperature estimate is consistent (within ≈1σ significance) with the standard stable WNM kinetic temperature range (≈5000–8000 K). This conclusion is rendered even more unlikely because Ts ≈ 1750 K at the velocity channels with the strongest H  i 21-cm absorption. Deeper H  i 21-cm absorption studies are needed to accurately estimate Tk via the Gaussian decomposition approach. Attempts have been made in the literature to carry out a joint multi-Gaussian decomposition of the H  i 21-cm emission and the H  i 21-cm absorption spectra (e.g. Heiles & Troland 2003; Murray et al. 2014). These have usually encountered difficulties in handling H  i 21-cm ‘self-absorption’ along the sightline by foreground CNM clouds of their own, and of background CNM and WNM, emission. The present sightline towards B0438−436 is very interesting in this context because the low peak H  i 21-cm optical depth (≈0.001) means that self-absorption is not an issue. The sightline thus allows the possibility of a joint Gaussian decomposition of the H  i 21-cm absorption and emission spectra without any effects of self-absorption. We carried out a joint multi-Gaussian fit to the H  i 21-cm emission and H  i 21-cm absorption spectra, the former at the original velocity resolution of ≈0.82 km s−1 and the latter after Hanning-smoothing and resampling, at a resolution of ≈0.86 km s−1. We allowed for H  i 21-cm absorption by a single Gaussian component and H  i 21-cm emission by multiple Gaussian components. In the fit, the position and FWHM of one of the emission components were constrained to be the same as that of the absorption component (using multiple absorption components did not yield a better fit, in terms of an improved reduced chi-square value). The best-fit model, with a reduced chi-square $$\chi ^2_N = 1.00$$, contains 5 emission components and one absorption component. The model is plotted in Figs 3(A) and (B), with the residuals from the fit displayed in the lower two panels; the residuals are seen to be consistent with noise. The properties of the different Gaussian components are summarized in Table 1, whose columns contain (1) the component number, (2) the peak brightness temperature of the component, in K, (3) the central LSR velocity, in km s−1, (4) the component FWHM, in km s−1, (5) the peak H  i 21-cm optical depth, or the 3σ upper limit to the peak optical depth (estimated at a velocity resolution equal to the component FWHM), (6) the kinetic temperature of the component, $$\rm {\it T}_k$$, and (7) the component spin temperature $$\rm {\it T}_s$$, or the 3σ lower limit to the spin temperature. Figure 3. View largeDownload slide Results of the multi-Gaussian joint decomposition of the (A) H i 21-cm emission and (B) H i 21-cm absorption spectra; the former spectrum is at a resolution of ≈0.82  km s−1, while the latter is at a resolution of ≈0.86  km s−1. The top panels show the best-fit model (solid curve) overlaid on the two spectra, while the bottom two panels show the residuals from the fit. Figure 3. View largeDownload slide Results of the multi-Gaussian joint decomposition of the (A) H i 21-cm emission and (B) H i 21-cm absorption spectra; the former spectrum is at a resolution of ≈0.82  km s−1, while the latter is at a resolution of ≈0.86  km s−1. The top panels show the best-fit model (solid curve) overlaid on the two spectra, while the bottom two panels show the residuals from the fit. Table 1. Gaussian joint decomposition of the H  i 21-cm emission and absorption spectra. Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Notes. For each component, the columns are: aThe peak brightness temperature; bThe central LSR velocity; cThe peak H  i 21-cm optical depth, or, for non-detections, the 3σ upper limit on the peak H i 21-cm optical depth at a velocity resolution equal to the line FWHM; dThe kinetic temperature inferred from the FWHM; eThe spin temperature or, for components 2–5, the 3σ lower limit to the spin temperature. View Large Table 1. Gaussian joint decomposition of the H  i 21-cm emission and absorption spectra. Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Component $$\rm {\it T}_B{}^a$$ V0b FWHM τpkc Tkd Tse K km s−1 km s−1 ×10−3 K K 1 1.08 ± 0.30 0.63 ± 1.57 15.0 ± 2.9 1.08 ± 0.22 4910 ± 1900 1000 ± 345 2 1.31 ± 0.35 10.9 ± 1.9 13.6 ± 4.8 <0.48 4045 ± 2850 >2730 3 0.358 ± 0.067 0.90 ± 1.46 56.9 ± 5.0 <0.23 70615 ± 12410 >1555 4 0.26 ± 0.10 −12.1 ± 1.1 8.4 ± 2.2 <0.61 1540 ± 805 >425 5 0.81 ± 0.52 20.8 ± 2.4 11.4 ± 2.1 <0.52 2835 ± 1045 >1560 Notes. For each component, the columns are: aThe peak brightness temperature; bThe central LSR velocity; cThe peak H  i 21-cm optical depth, or, for non-detections, the 3σ upper limit on the peak H i 21-cm optical depth at a velocity resolution equal to the line FWHM; dThe kinetic temperature inferred from the FWHM; eThe spin temperature or, for components 2–5, the 3σ lower limit to the spin temperature. View Large Initially, we note that the third component in Table 1 is extremely wide, with an FWHM of ≈57 km s−1. Given the relatively high Galactic latitude of B0438−436 (b ≈ −41°), it appears unlikely that this large broadening arises from either Galactic rotation or turbulence. We suspect that it may be due to a residual spectral baseline in the GASS spectrum, although we note that a similar wide component is also visible in the H  i 21-cm spectrum from the Leiden–Argentine–Bonn survey (Kalberla et al. 2005). For now, the origin of the broad component remains unclear. The component with detected H  i 21-cm absorption has an FWHM of 15.0 ± 2.9  km s−1, in good agreement with that obtained from the fit to the H  i 21-cm absorption alone (12.0 ± 3.1 km s−1). The inferred upper limit on the kinetic temperature is then 4910 ± 1900 K in the stable WNM temperature range. We emphasize, as above, that this is formally an estimate of the Doppler temperature TD and hence yields only an upper limit to Tk due to the possibility of non-thermal broadening. We consider the two extreme possibilities for line broadening: (1) weak non-thermal motions, for which the kinetic temperature is approximately equal to the Doppler temperature, i.e. Tk ≈ TD = (4910 ± 1900) K, and (2) strong non-thermal broadening, in which case Tk ≪ TD. In the first case, the detected H  i 21-cm absorption arises from the stable WNM. The spin temperature of this component is 1000 ± 345 K, significantly lower than the kinetic temperature. Figs 2 and 5 of Liszt (2001) suggest that this would require the gas to be at low pressures, P/k ≲ 1000 cm−3 K, and with almost no Galactic Lyman-α background penetrating into the H  i. The data would then support a picture of little coupling between the WNM and the background Lyman-α radiation field. In the second case, of strong non-thermal broadening, the measured Doppler temperature yields only an upper limit to Tk. However, the measured spin temperature gives the constraint Tk ≥ 1000, since Tk ≥ Ts in the WNM. This implies the range $$\rm 1000 \, K \, \le {\it T}_{\rm k}\le 4910\, K$$ for the component detected in absorption. For strong non-thermal broadening, Tk ≪ TD, the detected H  i 21-cm absorption would then still arise in warm gas, albeit from the thermally unstable phase, with Tk ≈ Ts, i.e. with strong coupling between the unstable neutral medium and the background Lyman-α radiation field. We cannot rule out this possibility with the present data. Finally, the Doppler temperature of component 2 of Table 1 is marginally consistent with its origin in the stable WNM, albeit with large errors. It is interesting that this component has a relatively high lower limit on its spin temperature, Ts > 2730 K, which suggests a stronger coupling with the background Lyman-α radiation. The Doppler temperatures of components 4 and 5 appear consistent with an origin in the unstable phase, although again with large errors. Again, it is interesting that the lower limit on the spin temperature of component 5 is relatively high, approaching the Doppler temperature. We emphasize, however, that the errors in the Doppler temperature estimates are still quite large. Deeper observations of the H  i 21-cm absorption towards B0438−36 are needed to confirm the above results, both the absolute values of the Doppler temperatures and, hence, the relation between the spin and the kinetic temperatures in the stable WNM and the unstable neutral medium. 4 SUMMARY iWe have used the GMRT to detect wide, weak Galactic H  i 21-cm absorption towards the quasar B0438−436. The velocity-integrated H  i 21-cm optical depth is very low, ∫τdV = (0.0188 ± 0.0036) km s−1. Combining this with the H  i column density ($$\rm {\it N}_{HI}= 1.29 \times 10^{20}$$ cm−2) measured from the GASS H  i 21-cm emission spectrum on a neighbouring sightline yields a column density-weighted harmonic mean spin temperature of $$\rm \langle {\it T}_s \rangle = (3760 \pm 765)$$ K, far higher than typical in the Galaxy. The high $$\rm \langle {\it T}_s \rangle$$ value indicates that there is little or no CNM along the sightline towards B0438−436. A detailed comparison between the H  i 21-cm emission and absorption spectra indicates that Ts ≥ 1760 K over the range of velocities showing the strongest H  i 21-cm emission, consistent with the spin temperature range expected in the stable WNM. The extremely low H  i 21-cm optical depth further implies that H  i 21-cm self-absorption is negligible along this sightline. We carry out a joint multi-Gaussian decomposition of the H  i 21-cm emission and the H  i 21-cm absorption profiles to obtain Tk ≤ (4910 ± 1900) K for the component with detected H  i 21-cm absorption, consistent with the stable WNM temperature range. The spin temperature of this component is (1000 ± 345) K, significantly lower than the kinetic temperature. This suggests that the H   21-cm line excitation is sub-thermal in the WNM along this sightline, possibly due to low gas pressure or low threading of the Galactic Lyman-α background into the WNM. However, the present data cannot rule out the possibility that the detected H  i 21-cm absorption might arise from warm gas in the thermally unstable phase, with significant non-thermal line broadening, and the gas spin temperature comparable to the kinetic temperature. ACKNOWLEDGEMENTS We thank the GMRT staff who have made these observations possible. The GMRT is run by the National Centre for Radio Astrophysics of the Tata Institute of Fundamental Research. NK acknowledges support from the Department of Science and Technology via a Swarnajayanti Fellowship (DST/SJF/PSA-01/2012-13). NR acknowledges support from the Infosys Foundation through the Infosys Young Investigator grant. REFERENCES Carilli C. L. , Dwarakanath K. S. , Goss W. M. , 1998 , ApJL , 502 , L79 CrossRef Search ADS Clark B. G. , 1965 , ApJ , 142 , 1398 CrossRef Search ADS Deguchi S. , Watson W. D. , 1985 , ApJ , 290 , 578 CrossRef Search ADS Dickey J. M. , Terzian Y. , Salpeter E. E. , 1978 , ApJS , 36 , 77 CrossRef Search ADS Dwarakanath K. S. , Carilli C. L. , Goss W. M. , 2002 , ApJ , 567 , 940 CrossRef Search ADS Field G. B. , 1958 , Proc. I. R. E. , 46 , 240 CrossRef Search ADS Field G. B. , 1965 , ApJ , 142 , 531 CrossRef Search ADS Field G. B. , Goldsmith D. W. , Habing H. J. , 1969 , ApJ , 155 , L149 CrossRef Search ADS Greisen E. W. , 2003 , in Heck A. , ed., Astrophysics and Space Science Library, Vol. 285, Information Handling in Astronomy – Historical Vistas , p. 109 , Kluwer Academic Publishers , Dordrecht Google Scholar CrossRef Search ADS Heiles C. , Troland T. H. , 2003a , ApJS , 145 , 329 CrossRef Search ADS Heiles C. , Troland T. H. , 2003b , ApJ , 586 , 1067 CrossRef Search ADS Kalberla P. M. W. , Haud U. , 2015 , A&A , 578 , A78 CrossRef Search ADS Kalberla P. M. W. , Mebold U. , Reich W. , 1980 , A&A , 82 , 275 Kalberla P. M. W. , Burton W. B. , Hartmann D. , Arnal E. M. , Bajaja E. , Morras R. , Pöppel W. G. L. , 2005 , A&A , 440 , 775 CrossRef Search ADS Kanekar N. et al. , 2014 , MNRAS , 438 , 2131 CrossRef Search ADS Kanekar N. , Chengalur J. N. , 2003 , A&A , 399 , 857 CrossRef Search ADS Kanekar N. , Subrahmanyan R. , Chengalur J. N. , Safouris V. , 2003 , MNRAS , 346 , L57 CrossRef Search ADS Kanekar N. , Braun R. , Roy N. , 2011 , ApJ , 737 , L33 CrossRef Search ADS Kulkarni S. R. , Heiles C. , Dickey J. M. , 1985 , ApJ , 291 , 716 CrossRef Search ADS Liszt H. , 2001 , A&A , 371 , 698 CrossRef Search ADS McClure-Griffiths N. M. et al. , 2009 , ApJS , 181 , 398 CrossRef Search ADS McKee C. F. , Ostriker J. P. , 1977 , ApJ , 218 , 148 CrossRef Search ADS Murray C. E. et al. , 2014 , ApJ , 781 , L41 CrossRef Search ADS Murray C. E. et al. , 2015 , ApJ , 804 , 89 CrossRef Search ADS Payne H. E. , Salpeter E. E. , Terzian Y. , 1983 , ApJ , 272 , 540 CrossRef Search ADS Radhakrishnan V. , Murray J. D. , Lockhart P. , Whittle R. P. J. , 1972 , ApJS , 24 , 15 CrossRef Search ADS Roy N. , Kanekar N. , Chengalur J. N. , 2013 , MNRAS , 436 , 2366 CrossRef Search ADS Roy N. , Kanekar N. , Braun R. , Chengalur J. N. , 2013 , MNRAS , 436 , 2352 CrossRef Search ADS Wolfire M. G. , Hollenbach D. , McKee C. F. , Tielens A. G. G. M. , Bakes E. L. O. , 1995 , ApJ , 443 , 152 CrossRef Search ADS Wolfire M. G. , McKee C. F. , Hollenbach D. , Tielens A. G. G. M. , 2003 , ApJ , 587 , 278 CrossRef Search ADS Wouthuysen S. A. , 1952 , AJ , 57 , 31 CrossRef Search ADS © 2018 The Author(s) Published by Oxford University Press on behalf of the Royal Astronomical Society This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/about_us/legal/notices)

### Journal

Monthly Notices of the Royal Astronomical Society: LettersOxford University Press

Published: May 17, 2018

## You’re reading a free preview. Subscribe to read the entire article.

### DeepDyve is your personal research library

It’s your single place to instantly
that matters to you.

over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month ### Explore the DeepDyve Library ### Search Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly ### Organize Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place. ### Access Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals. ### Your journals are on DeepDyve Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more. All the latest content is available, no embargo periods. DeepDyve ### Freelancer DeepDyve ### Pro Price FREE$49/month
\$360/year

Save searches from
PubMed

Create lists to

Export lists, citations