THE STRUCTURE OF THE LOCAL INTERSTELLAR MEDIUM. VI. NEW Mg ii, Fe ii, AND Mn ii OBSERVATIONS TOWARD STARS WITHIN 100 pc

The STIS is a powerful multi-mode UV/optical spectrograph, whose NUV (1600–3100 Å) high-resolution echelle capability is well suited for detecting the narrow absorption lines of heavy ions in the warm, partially ionized LISM. For the present "SNAPshot" program—designed to provide short observations to fill gaps in the HST schedule—the E230H/2713 Å setting was used exclusively. The spectral resolving power is R ≡ (λ/Δλ)  ∼  114, 000 with a simultaneous range of ∼200 Å. This setting captures several lines of interest, namely, Mg ii (2796.3543 Å and 2803.5315 Å), Fe ii (2586.6500 Å and 2600.1729 Å), and Mn ii (2594.499 Å and 2606.462 Å).⁵

The LISM SNAP survey recorded spectra for 34 sight lines toward stars within ∼100 pc (Figure 1). While the original sample was restricted to targets ⩽100 pc, the recent reexamination of Hipparcos parallaxes by van Leeuwen (2007) resulted in a revision of distance beyond 100 pc for two of our targets (HD141569 and R Ara). High-resolution NUV observations of nearby stars are ideally suited for an HST SNAP program because suitable targets are broadly distributed across the sky and the limited partial orbit exposure time can still achieve a S/N >10. In contrast, similar observations in the FUV require significantly more time, and must often be obtained at lower spectral resolution. This SNAP survey sought to optimize the LISM sample by taking high-resolution NUV observations of targets that already had the more time-consuming FUV spectra in the archive (from the HST's GHRS or STIS, or the Far-Ultraviolet Spectroscopic Explorer (FUSE)). While the original FUV observations might not have been taken with LISM analysis in mind, they can be utilized in combination with the new NUV spectra to make a comprehensive analysis of the LISM absorption along the particular line of sight. The SNAP NUV spectra have average S/N of 15, 10, and 10 for the Mg ii, Fe ii, and Mn ii lines, respectively, over a 10 km s⁻¹ range adjacent to any LISM absorption feature. Table 1 provides a full list of the observed targets.

Zoom In Zoom Out Reset image size

Table 1. Parameters for Stars in the LISM SNAP Program^a

HD No.	Other Name	Spectral	mV	vR	l	b	Distance	S/Nb	S/Nb	S/Nb	Other
		Type	(mag)	(km s−1)	(deg)	(deg)	(pc)	(Mg ii)	(Fe ii)	(Mn ii)	Spectra
209100	Ind	K5V	4.833	−40.4	336.2	−48.0	3.62	29	6	5	GHRS/Ech-A (Lyα)
115617	61 Vir	G5V	4.74	−8.5	311.9	44.1	8.56	18	6	7	STIS/E140M E230M
114710	β Com	G0V	4.311	6.1	43.5	85.4	9.13	21	8	8	FUSE
	WD1620–391	DA	10.974	43.2c	341.5	7.3	13.2	6	7	7	GHRS/G160M, FUSE
72905	π1 UMa	G1.5V	5.706	−12.0	150.6	35.7	14.4	22	8	7	FUSE
217014	51 Peg	G5V	5.524	−31.2	90.1	−34.7	15.6	8	5	6	STIS/G140M (Lyα), FUSE
120136	τ Boo	F7V	4.541	−15.6	358.9	73.9	15.6	17	8	10	STIS/G140M (Lyα)
142373	χ Her	F9V	4.672	−55.4	67.7	50.3	15.9	8	8	16	STIS/E140M
220140	V368 Cep	G9V	7.622	−16.8	118.5	16.9	19.2	19	5	2	GHRS/G140M G160M G270M
97334	MN UMa	G0V	6.476	−2.6	184.3	67.3	21.9	16	5	3	STIS/E140M E230M
	WD1337+705	DA	12.8	26	117.2	46.3	26.1	2	4	4	STIS/G430M, FUSE
222107	λ And	G8III–IV	3.975	6.8	109.9	−14.5	26.4	54	11	6	GHRS/Ech-A (Lyα), FUSE
180711	δ Dra	G9III	3.188	24.8	98.7	23.0	29.9	20	5	5	FUSE
12230	47 Cas	F0V	5.26	−26	127.1	15.0	33.2	12	17	19	GHRS/G140M, FUSE
163588	ξ Dra	K2III	3.867	−26.4	85.2	30.2	34.5	23	4	3	FUSE
216228	ι Cep	K0III	3.621	−12.6	111.1	6.2	35.3	24	5	5	FUSE
93497	μ Vel	G5III	2.818	6.2	283.0	8.6	35.9	39	11	8	STIS/E140M, FUSE
149499	V841 Ara	K0V	8.737	−24.8	329.9	−7.0	36.4	10	2	0	STIS/E140M, FUSE
131873	β UMi	K4III	2.238	17.0	112.6	40.5	40.1	9	2	3	FUSE
210334	AR Lac	G2IV	6.203	−34.6	95.6	−8.3	42.8	13	5	5	GHRS/G160M G270M, FUSE
28911	HIP21267	F5V	6.619	35	183.4	−22.6	44.7	12	6	10	FUSE
28677	85 Tau	F4V	6.02	36	180.9	−21.4	45.2	11	13	16	FUSE
204188	IK Peg	A8	6.06	−11.4	70.4	−22.0	46.4	7	12	15	GHRS/G160M, FUSE
	WD0549+158	DA	13.06	12.0	192.0	−5.3	49d	4	5	5	STIS/G140M G230M, FUSE
	WD2004–605	DA	13.14	−26.5	336.6	−32.9	58d	3	4	5	FUSE
9672	49 Cet	A1V	5.62	12.1e	166.3	−74.8	59.4	24	37	37	FUSE
43940	HR2265	A2V	5.88	24.0f	244.6	−22.4	61.9	19	29	30	FUSE
137333	ρ Oct	A2V	5.57	−11	307.0	−23.0	66.1	14	25	31	FUSE
	WD1631+781	DA	13.03	⋅⋅⋅	111.3	33.6	67d	0	0	0	FUSE
3712	α Cas	K0II–III	2.377	−4.3	121.4	−6.3	70.0	21	6	5	FUSE
149382	HIP81145	B5	8.872	3	11.8	27.9	73.9	21	24	20	FUSEg
	WD0621–376	DA	11.99	40.5c	245.4	−21.4	78d	8	9	9	FUSE
75747	RS Cha	A7V	6.02	26.0	292.6	−21.6	92.9	9	17	18	STIS/E230M, FUSE
	IX Vel	O9	9.503	20	264.9	−7.9	96.7	9	10	10	STIS/E140M, FUSE
141569	HIP77542	B9	7.143	−7.6h	4.2	36.9	116	11	14	13	FUSE
149730	R Ara	B9IV/V	6.73	±100i	330.4	−6.8	124	14	20	21	FUSE

Notes. ^aAll stellar parameters taken from the SIMBAD database unless otherwise stated. ^bS/N calculated over 10 km s⁻¹ bins on either side of any LISM absorption. ^cHolberg et al. (1998). ^dVennes et al. (1997). ^eHughes et al. (2008). ^fGontcharov (2006). ^gPlanned FUSE observations that were not completed before the end of the mission. ^hDent et al. (2005). ⁱReed et al. (2010).

Download table as:  ASCIITypeset image

Two additional sight lines (also listed in Table 1) were observed but did not result in spectra with detectable LISM absorption. One of the targets, β UMi, shows a dramatic P Cygni wind profile which prevents a confident placement of the continuum and obscures ISM features. The other target, WD1631+781, was not analyzed further owing to anomalously low S/N.

Initially, we processed the SNAP NUV spectra through the most recent available version of the calstis pipeline. We then utilized a second pipeline to perform a series of finer calibrations, based on protocols developed for the "StarCAT" catalog of STIS echelle spectra of 545 stellar objects (Ayres 2010). The StarCAT pipeline corrects each exposure for wavelength-dependent distortions in the STIS dispersion relations, specialized for the particular echelle setting, concatenates the ∼40 echelle orders captured in the exposure into a coherent one-dimensional spectrum, coadds spectra of similar type (i.e., subexposures in a given setting), and splices together neighboring spectra into a uniform tracing covering the available spectral range.

Many of the longest sight lines in the SNAP sample have both saturated and blended components. In order to obtain an accurate fit to the LISM absorption profile, we often must assume in these cases that Mg ii, Fe ii, and Mn ii are well mixed and in thermodynamic equilibrium within a cloud. We then constrain the spacing of the components in the saturated line using the velocities measured in the unsaturated lines. Such sight lines proved challenging because they provided minimal constraints on the Doppler parameter or column density of the components. For example, six ISM components were detected toward HD149730, a sight line 124 pc long (Figure 2(q)). The velocity structure would be impossible to disentangle using Mg ii alone, because it is completely saturated and blended. Instead, the velocities determined from unsaturated Fe ii and Mn ii lines were used to fix the spacing of the six components in the Mg ii line, leaving the other parameters—absolute velocities, Doppler widths, and column densities—free in the modeling.

In a few cases, the simultaneous fits indicated an apparent systematic error in the radial velocity measurements between the doublet components of the same ion. An example is β Com (Figure 2(b)). Because such anomalies affect only a few of the spectra, the cause is unlikely to be due to the StarCAT wavelength distortion correction. It could be associated with the complexity of the continuum placement in certain cases, or perhaps partial saturation in the stronger member of the doublet. In any event, the discrepancy is only of order 0.4 km s⁻¹, comparable to the internal precision of the StarCAT processing in the most favorable (i.e., high S/N) cases (0.3 km s⁻¹; Ayres 2010), and inconsequential for the measured absorption parameters.

Tables 2–4 list the final parameters of the fitting procedure: heliocentric radial velocities (v), Doppler parameters (b), and log  column densities (log N). Each value is a weighted mean based on the individual and simultaneous fits. Components seen in Mg ii but not in Fe ii or Mn ii have 3σ upper limits for the latter listed in the column density columns.

The radial velocity distribution (Figure 3) contains the projected velocities of absorbing material in the heliocentric rest frame. Although the LISM clouds each have distinct motions, they move in the same general direction with similar velocities (Frisch et al. 2002). Therefore, most sight lines in this sample show clusters of components in velocity space rather than isolated components distributed at random velocities. Only in the longest sight lines do we detect a broader range of velocities. At these large distances, traversed clouds may include structures outside of the local interstellar environment, where the kinematics diverge from the general LISM flow vector.

The radial velocities in the distribution range from −32 to +25 km s⁻¹. This span is consistent the bulk velocity of the warm LISM clouds (28.1 ± 4.6 km s⁻¹; Frisch et al. 2011), where an all-sky survey would sample the full range from positive values at this magnitude in the downwind direction to negative values at this magnitude in the upwind direction. When transformed to the reference frame of the Local Standard of Rest (Dehnen & Binney 1998), the distribution is similar, although centered on 0 km s⁻¹, slightly narrower (±20 km s⁻¹), and more peaked. The similarity of the general shape of the Mg ii and Fe ii distributions suggests that the ions are present in the same clouds. (Note that for the full LISM sample, the observed sight lines differ between Mg ii and Fe ii. For example, Redfield & Linsky (2001) analyzed 18 sight lines toward the Hyades, which were observed in Mg ii and not in Fe ii, resulting in the noticeable spike in detections at v_R  ∼  21 km s⁻¹.) If Mg ii and Fe ii are tracing the same material, then the $(v_{{\rm Mg\,\scriptsize{II}}} - v_{{\rm Fe\,\scriptsize{II}}})$ distribution should peak at 0 km s⁻¹, as indeed Figure 4 shows (mean = 0.06 km s⁻¹; standard deviation = 1.09 km s⁻¹).

Zoom In Zoom Out Reset image size

Mean values for the Doppler parameters and log column densities are listed in Table 5. The well-known dependence of the Doppler parameter (b; km s⁻¹) on temperature (T) and turbulent velocity (ξ) is

$$\begin{equation} b^2= \frac{2kT}{m}+\xi ^2=0.016629\frac{T}{A}+\xi ^2, \end{equation} \tag{ 1 }$$

where A is the ion's atomic weight in atomic mass units, k is the Boltzmann constant, and m is the ion's mass. Mg ii experiences larger Doppler broadening on average because it is a lighter ion and more susceptible to the thermal contribution. Conversely, Mn ii and Fe ii should have roughly equivalent mean Doppler parameters because turbulence—the dominant broadening mechanism for heavier ions—is independent of atomic weight. The 〈b〉 value of Mn ii is 0.4 km s⁻¹ greater than that of Fe ii, but this discrepancy should be discounted due to the small number of Mn ii detections. Redfield & Linsky (2002) report 〈b(Fe ii)〉 ∼ 2.4 km s⁻¹ with $\sigma _{{\rm Fe\,\scriptsize{II}}}\,{\sim}\,1.0$ km s⁻¹ and 〈b(Mg ii)〉 ∼ 3.1 km s⁻¹ with $\sigma _{{\rm Mg\,\scriptsize{II}}}\,{\sim}\,0.8$ km s⁻¹. While both means are 0.3 km s⁻¹ lower than the values obtained here, the difference is not very significant.

Table 5. Mean Values for Doppler Parameter and Log Column Density

Ion	〈b〉	σb	〈log N〉	$\sigma _{{\rm log}{N}}$
	(km s−1)	(km s−1)	log(cm−2)	log(cm−2)
Mg ii	3.36	0.90	12.89	0.58
Fe ii	2.72	0.84	12.61	0.54
Mn ii	3.16	0.99	11.91	0.65

Download table as:  ASCIITypeset image

Figure 5 compares Doppler widths of Mg ii and Fe ii absorptions for individual components. There are 71 pairings based on agreement in velocity. The solid line marks the ratio $b_{{\rm Fe\,\scriptsize{II}}}/b_{{\rm Mg\,\scriptsize{II}}}$ = 1, which would be the case if there were no thermal contribution to the line widths. On the other hand, if the broadening were entirely thermal, a ratio of $b_{{\rm Fe\,\scriptsize{II}}}/b_{{\rm Mg\,\scriptsize{II}}}$ = 0.66 would be obtained. When both broadening mechanisms contribute, the corresponding line width ratio should fall between the two lines. Indeed, 67 of the 71 components (94%) fall within this zone allowing for the 1σ error bars. Three of the four remaining outliers have very weak Fe ii absorption that yield an artificially broad or narrow fit. Alternatively, these unphysical ratios might highlight components in Mg ii and Fe ii that do not originate from a common cloud, or might result from partial saturation of one of the species.

Zoom In Zoom Out Reset image size

The column density distribution is included in Figure 3, and the means and standard deviations for each ion are listed in Table 5. Mg ii and Fe ii show similar column densities, but Mn ii is approximately an order of magnitude smaller. While the Mn ii sample is small, this difference still holds if one considers the average $\log N_{{\rm Mn\,\scriptsize{II}}}$ upper limit of 11.8 log (cm⁻²). The similarity in Mg ii and Fe ii column densities is attributable to two factors: both elements have comparable cosmic abundances, and both ions are the dominant ionization stages in the LISM (Slavin & Frisch 2008). The lower Mn ii column densities are due in part to the much lower cosmic abundance of Mn (about two orders of magnitude lower than Mg and Fe). While Mn is not included in the ionization balance calculations of Slavin & Frisch (2008), its first ionization potential is very similar to Mg and Fe, and so Mn ii is also likely the dominant ionization stage in the LISM.

The sight lines in this sample contain anywhere from one to six components. As would be expected, the number of components roughly correlates with the sight line length. With observations of enough sight lines, the number of components, together with an accurate distance to the background star, can provide insight into the distribution of LISM clouds as a function of distance. Redfield & Linsky (2004a) examined sight lines in 10 pc bins to judge how the average number of components per sight line changes with distance. A uniform distribution of identically sized clouds should show a steady increase in the average cloud number per 10 pc increment. Instead, the distribution remains flat after 30 pc, suggesting that LISM clouds are concentrated close to the solar system. The Redfield & Linsky (2004a) sample, however, suffers from an under-sampling of sight lines approaching 100 pc. In addition, a comprehensive survey of distant clouds requires dense spatial sampling on the sky.

In an effort to improve this measurement, we combined the older sample with the new SNAP sight lines, which include five lines of sight longer than 70 pc as well as many more shorter ones. The new distribution (Figure 6) shows a slight positive trend within the first 50 pc, but the uncertainties are such that the rise probably is not significant. This consistency in the average number of absorbers indicates that most clouds located within 50 pc actually begin within 10 pc of the Sun. More measurements are needed to determine whether the slight rise is real. Between 50 and 70 pc, there is a jump in the average number of absorbers. This increase might be related to the onset of the closest edge the Local Bubble at ∼55–60 pc (Lallement et al. 2003; Welsh et al. 2010). The elevated level appears to continue out to 100 pc.

Zoom In Zoom Out Reset image size

One of the fundamental results of the kinematics of the LISM is that it can, to first order, be described by a single bulk flow vector. This flow was first identified as coming from the direction of the Scorpio–Centaurus association by Crutcher (1982) using high-resolution, ground-based Ti ii absorption spectra. Lallement et al. (1986) recognized that while the interstellar flow is coming from a single general direction, multiple velocity vectors are needed to fully characterize the observed interstellar absorption toward nearby stars. In a refinement of their analysis, Lallement et al. (1995) identified a nearby cloud, the G Cloud, in the Galactic center direction as distinct from the kinematics of the cloud directly surrounding the solar system, the LIC. Using 96 sight lines, combining optical and UV observations, Frisch et al. (2002) evaluated the single bulk flow, but noted that significant deviations from this flow were apparent and identified seven LISM clouds. As the number of high-resolution LISM UV observations increased, spatially coherent discrepancies from the single bulk interstellar flow became evident (Redfield 2009). Within 15 pc of the Sun, Redfield & Linsky (2008) identified 15 LISM clouds based on 157 sight lines, of which only 81 were optimal high-resolution NUV spectra. In order to solve for a single velocity vector of an LISM cloud or test whether indeed a single bulk flow vector is sufficient to characterize the kinematics, one needs to correlate a large number of radial velocity measurements across an area of sky.

Subsequent observations provide excellent tests to the predictions made by the kinematic model developed by Redfield & Linsky (2008). For example, Welsh & Lallement (2010) analyzed the absorption spectra of three B stars at distances of ∼70 pc, finding a total of 11 individual absorption components. They note the good agreement between the kinematic predictions based on the model by Redfield & Linsky (2008), and the observed velocities, where all five predicted clouds were detected in these three sight lines. In addition, a cloud predicted to be near one of the sight lines, was also detected. Of the remaining five detected components, the authors associated two with more distant interstellar structures (the Local Cavity, and the Loop I superbubble), and three components remained unidentified (a similar fraction remained unidentified in the Redfield & Linsky 2008 analysis and may be associated with small, distant LISM clouds). Wood et al. (2014a, 2014b) analyzed two new sight lines toward nearby stars and detected three absorption components in total. In both cases, the Redfield & Linsky (2008) kinematic model predicted absorption by the LIC, which was confirmed by the observations, and the third component remains unidentified. The SNAP sample adds 34 new NUV sight lines within 100 pc, further increasing the total LISM sample size and enabling an additional and substantial test of the predictions made by the LISM kinematic model by Redfield & Linsky (2008).

Accompanying their paper, Redfield & Linsky (2008) provided an online "Kinematic Calculator"⁶ that calculates the projections of all 15 cloud velocity vectors toward any direction in the sky. It also lists any clouds predicted to be along the line of sight or within ∼20°, a rough estimate of the typical error in projected cloud boundaries given the modest sampling of the initial data set. The first step in testing whether a velocity component is associated with any cloud is to compare that velocity to the radial velocities of all the clouds predicted along the line of sight. When the difference between the observed and predicted velocities is within 3σ of zero, we consider the cloud to be a match. When more than one component along a line of sight meets that criterion, we select the one that agrees best. The process continues until all the components have been associated with a known cloud, including those for which the sight line passes within 20° of the cloud boundary. When a component's radial velocity matches that of a nearby cloud, it presents an opportunity to revise the cloud's boundaries. The boundaries were originally constructed by drawing contours around sight lines that showed spatial and kinematic similarities. The addition of more sight lines increases the "resolution" of the cloud boundaries. In the event that a component is incompatible with the velocities of all nearby clouds, it likely is the signature of a new, unidentified cloud.

Table 6 lists every velocity component detected in this sample and the cloud with which it best agrees. If the component does not match any clouds, it is labeled "NEW." The label "disk" denotes possible absorption by a circumstellar disk rather than the ISM (see Section 5.5). The listed velocities are the weighted means of the velocities measured in Mg ii, Fe ii, and Mn ii, unless one or both of the latter were not detected. In some cases, the Mg ii or Fe ii velocity is not used in the calculation of the mean when the line is severely blended, saturated, or marginally detected.

Table 6. Comparison with the LISM Kinematic Model

HD Number	Other	Distance	Component	v	Cloud	σa
	Name	(pc)	Number	(km s−1)
209100	Ind	3.62	1	−10.83 ± 0.35	LIC	1.1
115617	61 Vir	8.56	1	−14.74 ± 0.42	NGP	1.6
114710	β Com	9.13	1	−6.00 ± 0.17	NGP	0.1
	WD1620–391	13.2	1	−25.25 ± 0.28	G	1.2
72905	π1 UMa	14.4	1	13.29 ± 0.24	LIC	1.0
217014	51 Peg	15.6	1	−1.94 ± 0.31	Eri	0.7
			2	5.01 ± 0.55	Hyadesb	1.9
120136	τ Boo	15.6	1	−11.64 ± 0.21	NGP	2.0
142373	χ Her	15.9	1	−12.70 ± 0.15	NGP	2.5
220140	V368 Cep	19.2	1	6.06 ± 0.22	LIC	0.4
97334	MN UMa	21.9	1	4.54 ± 0.19	LIC	0.5
	WD1337+705	26.1	1	1.67 ± 0.32	LIC	0.0
222107	λ And	26.4	1	0.24 ± 0.70	NEW
			2	4.81 ± 0.24	LIC	1.2
			3	10.27 ± 0.21	Hyadesb	1.1
180711	δ Dra	29.9	1	−1.78 ± 0.14	LIC	0.2
12230	47 Cas	33.2	1	10.076 ± 0.071	LIC	0.2
163588	ξ Dra	34.5	1	−13.72 ± 0.76	Micb	0.9
			2	−6.27 ± 0.42	LIC	0.4
216228	ι Cep	35.3	1	2.93 ± 0.18	LIC	1.6
93497	μ Vel	35.9	1	−5.61 ± 0.53	G	0.0
			2	0.2 ± 1.3	Cetb	1.8
149499	V841 Ara	36.4	1	−25.90 ± 0.63	Aqlb	1.7
			2	−19.56 ± 0.91	G	1.2
			3	−13.32 ± 0.68	NEW
210334	AR Lac	42.8	1	−13.32 ± 0.34	NEW
			2	−0.63 ± 0.33	LIC	0.6
28911	HIP21267	44.7	1	14.30 ± 0.26	Hyades	1.0
			2	20.3 ± 2.2	Aur	0.9
			3	23.83 ± 0.45	LIC	0.3
28677	85 Tau	45.2	1	13.81 ± 0.77	Hyades	0.4
			2	18.6 ± 1.1	Aur	2.3
			3	23.26 ± 0.36	LIC	0.3
204188	IK Peg	46.4	1	−11.6 ± 1.3	Eri	1.1
			2	−6.72 ± 0.24	LICb	0.6
	WD0549+158	49	1	22.58 ± 0.81	LIC	0.8
	WD2004-605	58	1	−18.73 ± 0.28	Vel	2.3
			2	−12.06 ± 0.79	LICb	1.2
9672	49 Cet	59.4	1	10.15 ± 0.70	LIC	0.6
			2	13.69 ± 1.4	disk
43940	HR2265	61.9	1	11.13 ± 0.23	Blue	0.1
			2	18.21 ± 0.69	Dorb	1.5
			3	22.28 ± 0.64	NEW
137333	ρ Oct	66.1	1	−8.82 ± 0.59	G	0.6
			2	−1.9 ± 1.0	Blueb	1.0
			3	3.2 ± 3.4	Aqlb	0.7
			4	9.61 ± 0.57	NEW
3712	α Cas	70.0	1	−6.49 ± 0.36	NEW
			2	−2.4 ± 0.44	NEW
			3	9.24 ± 0.30	LIC	0.7
149382	HIP81145	73.9	1	−32.73 ± 0.31	G	3.1
			2	−25.28 ± 0.31	Mic	0.1
			3	−16.4 ± 1.7	Leob	2.5
	WD0621-376	78	1	9.11 ± 0.31	Blue	1.0
			2	16.00 ± 0.28	Dorb	0.6
			3	22.41 ± 0.29	NEW
75747	RS Cha	92.9	1	−4.86 ± 0.12	G	0.6
			2	−1.40 ± 0.11	Vel	1.7
			3	10.64 ± 0.12	NEW
			4	17.27 ± 0.11	NEW
	IX Vel	96.7	1	2.268 ± 0.062	LICb	2.4
			2	4.91 ± 0.37	Gb	1.1
			3	16.44 ± 0.32	Velb	0.7
			4	20.80 ± 0.73	Cetb	2.2
141569	HIP77542	116	1	−31.88 ± 0.56	NEW
			2	−27.55 ± 0.15	G	0.7
			3	−21.21 ± 0.18	Leoc	0.4
			4	−12.89 ± 0.23	NEW
			5	−6.0 ± 1.1	disk
			6	2.84 ± 0.44	NEW
149730	R Ara	124	1	−24.15 ± 0.63	Aqlb	0.0
			2	−19.3 ± 1.8	G	1.0
			3	−14.28 ± 0.57	NEW
			4	−7.02 ± 0.98	Blueb	0.2
			5	−2.5 ± 1.8	NEW
			6	2.13 ± 0.47	NEW

Notes. ^a$\sigma = | v_{\rm obs} - v_{\rm pred} | / \sqrt{\sigma ^2_{\rm obs} + \sigma ^2_{\rm pred}}$, where the predicted cloud velocities and errors are taken from the velocity vectors calculated by Redfield & Linsky (2008). ^bThis sight line is slightly outside the nominal cloud boundary determined in Redfield & Linsky (2008). ^cThe Oph Cloud is an alternative cloud assignment for this sight line, with σ = 0.0.

Download table as:  ASCIITypeset images: 1 2

For the new sample of 34 sight lines, the kinematic model developed by Redfield & Linsky (2008) predicts 40 absorption components spatially and dynamically consistent with the 15 known LISM clouds. These absorbers have measured velocities within 3σ of the predicted cloud radial velocity 100% of the time (i.e., 40/40). The excellent agreement provides strong support for the existence of spatially distinct, kinematic groups that can be characterized by simple bulk flow velocity vectors. The SNAP sample is approximately randomly distributed, so the two nearest clouds that subtend the largest area on the sky (LIC and G) have the largest numbers of new detections, 19 and 9, respectively. These are substantial increases in the total number of sight lines detected through these clouds. Redfield & Linsky (2008) used 79 sight lines for the LIC (so the new data represent a 24% increase), and 21 sight lines for the G Cloud (a 43% increase). In addition to the 40 velocity components consistent with the boundaries of the 15 clouds, there are also 18 additional velocity components that were successfully predicted by the kinematic model to be within ≲20° of a cloud boundary. These 18 velocity components will lead to refinements and improved spatial resolution in the LISM cloud boundaries.

Further, there were 18 components that are unaffiliated with known clouds. Two are candidates for circumstellar absorption (see Section 5.5), leaving 16 new unaffiliated radial velocities, half of which are detected in the lines of sight toward stars at distances >80 pc. This subsample represents ∼21% of the total number of detected absorbers, which is similar to the 19% of the total absorbers that are unaffiliated in the original kinematic model by Redfield & Linsky (2008). These absorbers likely are associated with more distant LISM clouds that subtend smaller angles, and therefore have fewer measurements with which to determine a velocity vector. However, as the sample grows, it will become possible to characterize the dynamics of even these more distant LISM clouds.

The LISM is an inhomogeneous structure (Frisch & York 1991; Diamond et al. 1995; Redfield & Linsky 2008). Depending on the direction of observation, total column densities can vary by more than an order of magnitude over the same distance (Redfield & Linsky 2002). These changes are apparent over large angular scales, but to examine the finer structure requires sight lines with small angular separation. Variations in the general ISM on subparsec scales have been found in previous studies of wide binaries (Meyer & Blades 1996; Watson & Meyer 1996) and high proper motion pulsars (Frail et al. 1994; Stanimirović et al. 2010). However, given the small number of observations, there are not many opportunities to make measurements of small-scale structure in the LISM. In their study of Mg ii absorption lines in the spectra of 18 stars in the Hyades with angular separations from 06–33°, Redfield & Linsky (2001) found a smooth variation in column density, consistent with a large-scale homogeneous morphology of the LIC. In the present survey, three pairs of stars provide spatial information on a scale of <3°.

V841 Ara and R Ara are separated by only 05, but their distances from the Sun are 37.1 pc and 124 pc respectively. The R Ara sight line shows three extra components, suggesting that three clouds are located between 37.1–124 pc, including absorption from the Local Bubble boundary (Figure 7). It also is possible that the cloud boundary is located in the 05 between the sight lines, although the physical separation would then correspond to only 0.32 pc at the distance of V841 Ara (and only 0.03 pc or 6300 AU at a distance of 3.5 pc, the approximate distance of the Aql Cloud, which matches component 1 for both lines of sight). The pair shows nearly identical velocity structure for the three bluest components, two of which have been identified as the Aql Cloud and the G Cloud. The ∼ −14 km s⁻¹ component seen in both sight lines appears to be a new cloud located within 37.1 pc. The other fit parameters (i.e., b-value and column density) also match the Aql and G clouds quite well. For the Doppler parameter, all components for both Mg ii and Fe ii agree to within 3σ of the Aql and G cloud velocities, with an average discrepancy of only 0.7σ. Likewise for column density, all agree to within 3σ with an average discrepancy of only 1.5σ. Two of the three components seen only in R Ara likely belong to unidentified clouds located beyond 37.1 pc. The third component is identified with the Blue Cloud which appears to traverse the sight line of R Ara, but not that of V841 Ara.

Zoom In Zoom Out Reset image size

With a separation of 13, HD43940 (61.9 pc) and WD0621-376 (78 pc) show three similar absorption components that agree in velocity (Figure 7). The two bluest components match the model predictions for the Blue and Dor Clouds. The third component is likely an unidentified cloud seen in both sight lines. The physical separation of clouds along the lines of sight is 1.4 pc at the distance of HD43940, 0.3 pc at the approximate distance of the Dor Cloud (11.7 pc), and only 0.06 pc or 12,000 AU at the approximate distance of the Blue Cloud (2.6 pc). The other fit parameters are quite similar to the Dor and Blue clouds, except for the ∼22 km s⁻¹ component in Fe ii. This component is discrepant in column density by 3.5σ and in Doppler parameter by 2.7σ. The disagreement can be easily seen in Figure 7, which shows that the HD43940 sight line has a markedly broader and deeper absorption. This component was previously unidentified, which makes it likely that it is a distant cloud, and therefore possible that a significant variation is seen across a small angular distance. The Mg ii absorption for this component does not show the same variation, but the absorption is near saturation making it difficult to identify subtle variations. The other components all agree to within 3σ, with an average variation in Doppler parameter of only 0.7σ and in column density of only 1.2σ. Since no new component appears in the longer sight line, no new clouds with detectable column densities are located in the 16.1 pc span between the stars.

The pair of stars, HD28911 (44.7 pc) and 85 Tau (45.2 pc), separated by 26, are remarkable in that they are so close both in terms of their angular separation and distance. These two sight lines are the only ones for which the kinematic model of Redfield & Linsky (2008) predicts three different clouds along a line of sight. Both sight lines show absorption at the expected radial velocity for all three clouds, with no additional components detected, implying that there are no LISM clouds beyond ∼5 pc (the approximate distance of the Hyades Cloud and the most distant of the three clouds), along these lines of sight out to 44.7 pc. The fit parameters for all three components match the LIC, Hyades, and Aur clouds very well. The Doppler parameters also fit those three clouds with discrepancies within 0.7σ, and the average variation agrees within 0.4σ. For column density, all measurements agree to within 1.8σ, and the average variation agrees within 0.6σ. While the physical separation of the pair at the distance of the closer star, HD28911, is only 2.0 pc, all three components are identified with LISM clouds. At the approximate distance of the Hyades Cloud (5.0 pc), which matches component 1 for both sight lines, the physical separation of the lines of sight is 0.23 pc and for component 2, which matches with the Aur Cloud (3.5 pc), the physical separation is 0.16 pc. Component 3 is associated with the LIC, which directly surrounds the solar system. Based on the Redfield & Linsky (2000) morphological model of the LIC, and assuming an average H i density of 0.2 cm⁻³, the distance to the edge of the LIC is only 2.0 pc. Therefore, the maximum physical separation of the sight lines at this distance is only 0.09 pc or 19,000 AU.

Based on these three close pairs, we find little evidence for significant small-scale structure in the dominant ions of the warm partially ionized clouds of the LISM. The predicted kinematic variations along such small separations is negligible. We find that sight lines are identical in measured radial velocity, Doppler width, and column density down to scales of the order of 10,000 AU.

Combining absorption measurements of multiple ions makes it possible to separate the thermal and turbulent contributions to the observed line widths. With increasing ion mass, the contribution of thermal broadening to the Doppler parameter drops and the relative contribution of turbulence and unresolved cloud components increases. High spectral resolution observations are essential for the heavy ions (e.g., Fe ii and Mg ii). On the other hand, medium-resolution spectra are often suitable for the light ions (e.g., H i, D i, and C ii) located in the FUV. The new sample of Mg ii and Fe ii Doppler widths can be combined with archival FUV data along the same sight lines to separate the temperature and turbulent velocity contributions. With a similar data set, Redfield & Linsky (2004b) measured the temperature (T) and turbulent velocity (ξ) for 50 individual absorbers. Their measurements yielded a weighted mean LISM gas temperature (T) of 6680 K (σ = 1490 K) and weighted mean turbulent velocity (ξ) of 2.24 km s⁻¹ (σ = 1.03 km s⁻¹).

Here, we present a new calculation of T and ξ for the LISM absorption along the line of sight toward Ind (3.62 pc), see Figure 8. The single absorption component is kinematically associated with the LIC. The H i line width is taken from Wood et al. (1996), who assumed a single component LISM absorption profile. Our high-resolution NUV observations support that assumption and now enable the first measurement of T and ξ for the LISM along that sight line. We find $T = 8340^{+450}_{-440}$ K and $\xi = 1.97^{+0.08}_{-0.09}$ km s⁻¹, which agree well with the mean values for the LIC (T = 7500  ±  1300 K, ξ = 1.62  ±  0.75 km s⁻¹) obtained by Redfield & Linsky (2008) using the 19 sight lines originally discussed in Redfield & Linsky (2004b). This result is dominated by Mg ii with its high precision line width measurement, and the (broader) H i width. Together with measurements of D i or H i absorption by Wood et al. (2000, 2005), our new sample makes it possible to derive temperatures and turbulent velocities for V368 Cep, 61 Vir, χ Her, and μ Vel. In the case of μ Vel, the previous assumption of only a single LISM absorber used in the analysis of the medium-resolution data turns out not to be valid as this sight line requires at least two LISM components. All told, the present sample—with 76 total absorbers and archived FUV observations—has the potential to more than double the number of temperature and turbulent velocity measurements for the LISM and to continue developing a more refined inventory of the local interstellar environment.

Zoom In Zoom Out Reset image size

Knowledge of the LISM environment around a star is essential for a complete understanding of its astrosphere. An astrosphere is analogous to the Sun's heliosphere, and marks the interface between the outward flow of the stellar wind and the inward pressure of the surrounding ISM. It can expand or contract depending on the density of the ISM as well as the strength of the stellar wind. Wyman & Redfield (2013) observed the LISM in the direction of the Sun's historical line of motion in order to investigate how it might have impacted the heliosphere and the modulation of the galactic cosmic ray flux at 1 AU. The ISM environment through which a star is passing might achieve densities great enough to compress the astrosphere to within the orbit of planets, exposing their atmospheres directly to the ISM and the full brunt of the galactic cosmic ray flux.

An astrosphere is detected by the spectral signature of its "hydrogen wall." When cold ISM neutrals interact with protons from the hot solar wind, they charge exchange, producing an abundance of decelerated neutral hydrogen atoms (Wood et al. 2004). This heated gas builds up at the interface of the stellar wind and the ISM, producing a deep, broad Lyα absorption feature. The astrospheric feature is often highly saturated and difficult to differentiate from interstellar H absorption blue-shifted relative to the ISM absorption. Observing heavy ions (e.g., Fe ii and Mg ii) in the same direction provides important constraints on the analysis of heliospheric and astrospheric H i absorption by measuring the number LISM components and their radial velocities.

Observations of Lyα toward Ind and λ And led Wood et al. (1996) to conclude that, for both sight lines, an astrospheric H i absorption component was necessary to explain the width and velocity discrepancies between the H i and D i absorption lines. For each sight line, they identified one LISM component and an astrospheric component. Assuming ξ = 1.2 km s⁻¹, they measured a LISM temperature T = 8500 ± 500 K for Ind and T = 11,500 ± 500 K for λ And. The λ And LISM temperature is high, suggesting that the broad Lyα absorption feature might contain blends.

The original Ind and λ And analyses were performed without knowledge of the ISM velocity structure. However, we have now observed Mg ii and Fe ii absorption for these lines of sight and determined the velocity structure. This provides motivation to revisit the Lyα analyses. As for Ind, the sight line contains only one absorption component for both Mg ii and Fe ii, meaning that the analysis of Wood et al. (1996) assuming just one ISM component requires no revision. The Doppler parameters indicate $\xi =1.97^{+0.08}_{-0.09}$ km s⁻¹, a higher value than they assumed, although the temperature measurement remains consistent. The sight line toward 61 Vir, with a detection of both heliospheric and astrospheric absorption is another example where the assumption of a single LISM absorber was confirmed by the new high-resolution NUV observations. The heliospheric and astrospheric analysis performed by Wood et al. (2005) on that star is therefore strengthened.

However, our examination of the Mg ii and Fe ii absorption lines toward λ And reveals three LISM components, not one. Therefore, the one-component Doppler parameter determined solely through Lyα absorption is artificially enhanced, leading to an overestimation of the LISM T and ξ. Furthermore, it is possible that consideration of the complex velocity structure could obviate the need for an astrospheric absorption component in the Lyα line. In light of these new results, a reconsideration of the Lyα analysis of λ And was warranted.

Figure 9 shows the Lyα spectrum of λ And. The extremely broad H i absorption is centered at about 10 km s⁻¹, with narrower deuterium (D i) absorption at −75 km s⁻¹. The methodology for Lyα modeling has been described in detail by Wood et al. (2005). In short, the D i and H i absorption are fitted simultaneously, with the D i and H i velocities and Doppler parameters forced to be self-consistent. In addition, we assume a column density ratio of D/H = 1.56 × 10⁻⁵ for all components (the generally accepted ratio within the Local Bubble, but note that D/H was not fixed in this manner in the original analysis of Wood et al. 1996). We assume that the velocity spacing of the three ISM components of H i is the same as that of the Mg ii components. The background stellar Lyα profile is similar to that of Wood et al. (1996), but with some minor modifications to achieve better fits given the fixed D/H ratio noted above.

Zoom In Zoom Out Reset image size

We modeled the Lyα line profile with and without an astrospheric contribution. Figure 9(a) illustrates the best ISM-only fit, which assumes the same column density ratios for the three components as derived from Mg ii but varies the Doppler parameters. The reduced $\chi ^2_{\nu }=1.21$, but there is a noticeable systematic discrepancy with the observed profile in the [ − 60, −50] km s⁻¹ velocity range in Figure 9(a), and the Doppler parameter of the most blueshifted ISM velocity component (Component 1) required a very large ISM cloud temperature (T  ∼ 18,000 K). A central issue regarding the acceptability of this fit is the plausibility of such a high temperature for Component 1. This component seems to be detected in Ca ii for the very nearby target, κ And (Vallerga et al. 1993). An 18,000 K temperature would suggest a Ca ii width of ∼2.7 km s⁻¹, which is much higher than the 1.5 km s⁻¹ width reported by Vallerga et al. (1993), albeit with substantial uncertainty.

We conclude, therefore, that there is good reason to prefer a fit to the data with an astrospheric component. Figure 9(b) shows such a fit, with the Doppler parameters (i.e., temperatures) of the ISM components forced to be consistent with best estimates inferred from the Mg ii, Fe ii, and Ca ii lines, but with column densities allowed to vary. The resulting fit exhibits a slightly improved $\chi ^2_{\nu }=1.19$, and with substantial qualitative improvement in the [ − 60, −50] km s⁻¹ velocity range compared to the ISM-only fit. We conclude that an astrosphere detection for λ And is likely, consistent with the original assessment of Wood et al. (1996); but this exercise clearly demonstrates how knowledge of the ISM velocity structure can help illuminate the issues involved in making such an assessment.

Narrow absorption features in stellar spectra are usually signatures of foreground interstellar absorption, but under certain circumstances, they could also result from circumstellar material, such as in a red-giant wind or the disk of a young star. Circumstellar disks, in particular, evolve through phases classified by their gas-to-dust ratio. Protoplanetary disks generally exist around pre-main-sequence stars where accretion of material is ongoing, and the associated gas-rich disks are massive and optically thick. Transitional disks have optically thin inner regions and optically thick outer regions as revealed by mid- to far-IR excesses but little to no near-IR excess. Submillimeter CO emission indicates that these outer regions are gas rich (e.g., Qi et al. 2004). Approximately 10⁷ yr into the star's lifetime, the primordial material clears, and the now main sequence star is surrounded by a gas-poor debris disk. Mechanisms that remove gas from the system include accretion, formation of a gas giant planet, depletion onto dust grains, and a wind from the central star.

UV and optical spectroscopy has been used to detect small amounts of gas in debris disks (e.g., Lagrange et al. 1998; Chen & Jura 2003). This gas is not primordial, but rather results from collisions and evaporation of planetesimals (Roberge & Weinberger 2008). Detecting gas in the debris disk of a star is challenging because by their nature very little gas exists in such disks. Sensitive observations of nearby, edge-on systems offer the best prospects for detecting absorption from disk gas in the UV and optical. The A-type star β Pictoris, which satisfies these prerequisites, has become the iconic example of well-characterized gas absorption in a debris disk. UV and optical observations have indicated roughly solar abundances of gaseous elements with the exception of a large overabundance of carbon (Roberge et al. 2006). Similar characterizations of other circumstellar disk systems will enable better understanding of planet formation and composition.

5.5.1. Absorption Measurements toward Stars with Circumstellar Disks

5.5.2. Absorption Measurements along Sight Lines Near a Star with a Circumstellar Disk