In the Trenches of the Solar-stellar Connection. V. High-resolution Ultraviolet and X-Ray Observations of Sun-like Stars: The Curious Case of Procyon

Table 1 summarizes the fundamental properties of the three STIS stars. The values for α Cen AB were copied from Paper I, and are mainly drawn from Kervella et al. (2017). The two dwarfs closely bracket the Sun in mass, radius, and luminosity. Alpha Cen A is similar in temperature to the Sun, while B is several hundred K cooler, befitting its later spectral type. The α Cen system is slightly metal-rich compared to the Sun, by about 0.2 dex. Even though, in principle, a binary system with accurately known masses, luminosities, and composition should tightly constrain the age through evolutionary considerations, estimates for AB span a broad range. Consensus values suggest that the system is roughly 1 Gyr older than the Sun.

Procyon is the nearest of the F-type stars and is one of brightest. Technically, it is solar-like, with M ∼ 1.5 M_⊙ and a temperature only 800 K warmer than the Sun's. However, unlike the Sun, Procyon is a subgiant, on a post-main-sequence track heading toward the red giants. Like AB, Procyon is in a visual binary, although the companion is a faint, hydrogen-deficient, metal-polluted white dwarf, which indicates that the system has witnessed significant evolution over its lifetime and that the current primary was once the secondary. Procyon's apparent orbit on the sky is compact (semimajor axis ∼43) and the period is a relatively short 41 yr. Orbital elements were challenging to determine historically, given the difficulty to recover the very faint secondary in the glare of the close-by bright primary. In the spacecraft era, Bond et al. (2015) used high-precision HST imaging to track about half the orbital path and, combined with the historical astrometry, significantly refined the parameters. The inclination of the system is low, about 31°, and the eccentricity is high, e ∼ 0.40. The inferred mass of Procyon is 1.48 ± 0.01 M_⊙.

As of 2021 June, SIMBAD listed nearly a hundred T_eff entries for the subgiant, although some are duplicates. A consensus 6560 ± 90 K encompasses the recently proposed values. Most of the metallicities cited in SIMBAD are indistinguishable from solar, and much of the evolutionary modeling cited below has assumed solar composition (although, to be sure, there still are heated debates as to what constitutes the true solar scale, as noted below).

A detailed study by Aufdenberg et al. (2005) had superseded an earlier effort by Kervella et al. (2004), who measured the interferometric radius of Procyon with the VLT. Aufdenberg et al. derived fundamental parameters for the subgiant utilizing 3D convection models to establish the critical limb-darkening factors needed to interpret the interferometric measurements. The derived radius is 2.03 ± 0.01 R_⊙, and the effective temperature estimate is 6540 ± 80 K.

Age is a key factor in coronal activity, but is challenging to determine for arbitrary Field stars. As mentioned earlier, binary systems with known masses and compositions offer additional constraints on the stellar age through the joint evolution. Nevertheless, these efforts have led to a disappointingly wide range of estimates for Procyon, partly because the current secondary has essentially finished its life and thus offers less of a lever arm than would a currently evolving Main sequence companion. Bond et al. (2015) cite ages of 1.7–2.7 Gyr, which they obtained with different assumptions concerning core convective overshooting, and note that the asteroseismic age is on the higher side. An earlier evolutionary study by Liebert et al. (2013) proposed a tighter range, 1.9 ± 0.1 Gyr, using the Asplund et al. (2005) solar abundance scale (Z = 0.015); and a slightly higher value, 2.0 Gyr, with the Grevesse & Sauval (1998) composition (Z = 0.019), although the authors reported that the latter model fits were poorer. According to these authors, part of the difficulty in pinning down the age of mid-F stars such as Procyon is the importance of overshooting, given the small convective cores and thin convective envelopes. More recently, Sahlholdt et al. (2019), in a study of Gaia Benchmark stars (Procyon is one), proposed 1.5–2.5 Gyr for the F subgaint, with 2 Gyr the preferred age, again based on model tracks and seismology. It seems likely that Procyon is substantially younger than the Sun or α Cen, by perhaps 2–3 Gyr, yet is more advanced in its evolution. Thus, it is a unique comparison star as far as the state of its magnetic activity is concerned, especially in terms of its core and envelope structure at the edge of convection.

Another key dynamo parameter is the rotation rate. An attempt was made by Ayres (1991), alluded to earlier, to record rotational modulations of bright FUV lines of Procyon, including H i 1215 Å Lyα, using high-S/N trailed exposures with the low-dispersion SWP mode of IUE, but without success. Allende Prieto et al. (2002) reported $\upsilon \sin i\sim 3.2\pm 0.5$ km s⁻¹ based on high-quality optical absorption spectra of Procyon from the McDonald Observatory (∼1.5 km s⁻¹ resolution; S/N up to 2000), which was interpreted with the help of 3D radiation-hydrodynamic convection models. The authors cautioned that the projected velocity might be overestimated by as much as 0.5 km s⁻¹ owing to subtleties associated with the finite numerical resolution of the simulations. Earlier, Gray (1981) had obtained 2.8 ± 0.3 km s⁻¹ for Procyon, also using high-resolution McDonald spectra, but with a novel Fourier analysis that was popular at the time. Taking a compromise $\upsilon \sin i\sim 3$ km s⁻¹, and assuming that the spin axis of the star is aligned with the orbit, implies a rotational velocity of 6 km s⁻¹ and a period of about 17 days, which is somewhat shorter than the midlatitude solar value of 25 days.

The last important parameter is the bolometric flux, f_BOL (erg cm⁻² s⁻¹ at Earth), which will later be used to normalize the UV fluxes to allow a fairer comparison among the three stars of the study, which differ significantly in sizes and distances. The usual approach for arbitrary late-type stars is to apply a formula based on, say, the Gaia broadband G magnitude and the Gaia (bp − rp) color (see, e.g., Paper III, Appendix A). Alternatively, nearby bright stars such as Procyon often have detailed spectral irradiance coverage, extending from the FUV up into the mid-infrared. The broad spectral energy distributions can be integrated directly to yield a higher quality estimate of f_BOL than could be obtained from, say, the G-based formula, especially given that nearby stars such as Procyon (and α Cen AB) are too bright for Gaia, so G magnitudes and colors are lacking in the first place. Aufdenberg et al. (2005) carried out such a SED integration for Procyon, and found f_BOL ∼ 1.79 ±0.09 × 10⁻⁵ erg cm⁻² s⁻¹, which is adopted here.

The High-Resolution Camera (HRC-I) of the Chandra X-ray Observatory has been especially valuable in the study of nearby optically bright late-type stars that also have high count rates in the 0.2–2 keV soft X-ray band. HRC-I is immune to "optical loading," the impact of low-energy photons on CCD-type X-ray detectors, which requires either blocking filters or short frame times to mitigate; as well as "pile-up," the collision of multiple X-ray photons in a single cell of a CCD detector during the integration time. HRC-I also retains excellent sensitivity to the softer X-ray emission typical of low-activity coronae such as the Sun's, whereas the CCD cameras of Chandra have lost significant low-energy response owing to layers of molecular contamination that have built up over the years. The one downside of HRC-I is its lack of energy resolution compared to CCD cameras, which typically achieve modest but useful E/ΔE ∼ 50. However, the poor spectral response of HRC-I is countered by its relatively constant energy conversion factor (ECF: erg cm⁻² counts⁻¹) over the 1–10 MK coronal temperature range favored by nearby low- and moderate-activity stars. Furthermore, several of the nearby bright stars have been subjected to detailed high-resolution X-ray spectroscopy with Chandra's Low-Energy Transmission Grating Spectrometer (LETGS: E/ΔE ∼ 500–1000). Models of the distribution of coronal plasma with temperature, and associated coronal abundances, can be derived from these high-quality X-ray spectra (e.g., Wood et al. 2018, and references to previous work therein); in fortuitous cases, at extremes of activity cycles (Ayres 2015b for AB; LETGS pointings on Procyon were also included in that study).

Long-term X-ray imaging programs have been carried out on α Cen AB with HRC-I since late-2005. The subarcsecond spatial resolution of Chandra has been vital given that the AB orbital separation was shrinking over that period, bottoming out at 4'' in 2016, and will continue to be below 10'' (approximate resolution of XMM-Newton) until the latter part of the current decade. Since 2010, the Chandra program has been conducted jointly with HST/STIS, which will be covered in more detail later. The long-term Chandra HRC-I program on α Cen AB was described in Paper I. Table 2 provides an additional Chandra pointing carried out recently, after publication of Paper I.

Table 2. Chandra HRC-I Pointings: α Cen AB (New: post-2020.1)

ObsID	UTmid	${t}_{\exp }$	(CR)A	(CR)B	${({L}_{{\rm{X}}})}_{{\rm{A}}}$	${({L}_{{\rm{X}}})}_{{\rm{B}}}$
	(yr)	(ks)	(counts s−1)		(1027 erg s−1)
1	2	3	4	5	6	7
21575	2020.428	5.11	0.49 ± 0.06	3.90 ± 0.44	0.29	3.87

Note. Col. 3 exposure time includes dead-time correction. Cols. 4 (A) and 5 (B) count rates were time-filtered to remove flare enhancements, if any; and were corrected for the 95% encircled energy of the r = 15 detect cell. Cited uncertainties reflect standard deviations of time-binned count rates, including the influence of flares, with respect to reported flare-filtered averages. Cols. 6 (A) and 7 (B) X-ray luminosities (0.2–2 keV) were calculated from the count rates using CR-dependent ECFs derived for A and B separately (see Ayres 2014 for further details), and d = 1.338 pc (e.g., Kervella et al. 2017). No correction was made for the time-dependent sensitivity decline of HRC-I (about 2% per year since 2010). ${({L}_{{\rm{X}}})}_{\odot }\sim 0.15\mbox{--}1.2$ in same energy band and luminosity units, over recent (weak) sunspot Cycle 24.

Download table as:  ASCIITypeset image

Procyon has been observed by Chandra sporadically since early in the mission, mainly with LETGS (+HRC-S camera) prior to 2016, although with one High-Energy Transmission Grating Spectrometer (HETGS+ACIS-S) pointing (2014) and a single on-star HRC-I exposure (2008). In 2016, a dedicated joint Chandra/HST project (HRC-I + STIS) was approved, which has continued to the present. Initially, the program featured semiannual 10 ks exposures of the subgiant, which were scaled back to 5 ks integrations roughly annually in recent years (for reasons that will become obvious shortly). Procyon is a bright, ∼1 count s⁻¹ X-ray source, which can yield a high-S/N detection in 1 ks. However, the longer duration exposures guarded against short-term flare events, which had previously been seen in α Cen B pointings, and can skew the measurement away from the desired quiescent level in that epoch (Ayres 2014). However, no such flaring was detected in any of the Procyon light curves.

Table 3 lists all of the HRC-I pointings on Procyon through 2021 April. On the one hand, the zeroth-order images of the LETGS (or HETGS) spectra could have been measured to provide additional points on the X-ray time history of Procyon (and those of AB as well). On the other hand, the HRC-I exposures alone are sufficient for the purpose, and avoid the uncertain scaling between the attenuated zeroth-order image and a direct HRC-I exposure.

The Chandra HRC-I event lists were processed as described in Ayres (2014), incorporating count-rate-dependent ECFs developed specifically for α Cen AB and Procyon based on emission-measure modeling of the respective LETGS spectra. Fortuitously, the three available epochs of LETGS spectra of AB, taken over about 11 yr, happened to capture coronal low and high states of both stars, so the hardening of each SED toward the cycle MAX could be tracked. One further note: the HRC-I has declined in sensitivity for soft coronal sources by about 20%–30% since 2010, according to WebPIMMS. ¹ The sensitivity decline was not compensated for the X-ray luminosities in Tables 2, but was taken into account in subsequent figures involving the coronal emissions (as described below).

2.2.1. Solar X-Rays: Revised FISM2

2.2.2. X-Ray Time Histories of the Sun-like Stars

Figure 1 depicts time histories of bolometrically normalized coronal emissions of the Sun, α Cen AB, and Procyon, from about 2005 to the present. The Chandra points were corrected for a somewhat arbitrary 2% per year HRC-I sensitivity decline since 2010. This particular slope was chosen because it flattens the recent Procyon HRC-I time series, which would be consistent with the stable long-term behavior shown by the earlier LETGS spectra (and the one HRC-I from 2008). The lower panel highlights epochs when HST/STIS pointings on the three stars were carried out. An upward pointing triangle indicates that an FUV spectrum was obtained; a diamond indicates FUV + NUV.

Zoom In Zoom Out Reset image size

The Sun displays one complete 11 yr cycle (Cycle 24) and the tail of preceding Cycle 23. Alpha Cen B shows one 8 yr cycle, and much of a second currently in progress. Alpha Cen A has most of a 19 yr cycle in the Chandra epochs, including a long minimum between 2005 and 2010; α Cen A had an earlier peak in the late 1990s bridging the end of the ROSAT mission and beginning of Chandra's (see Ayres 2014). The Procyon time series is flat, somewhat by design given the assumed sensitivity decline of HRC-I. However, if the true decline is less, or more, then it would only tilt the Procyon time series up or down by a small amount, preserving the long-term trend of relative constancy, certainly when compared to the rather larger excursions exhibited by the other stars. It is possible, in fact, that Procyon fortuitously is an excellent calibration star to document HRC-I's soft response over time.

The medium-resolution (λ/Δλ ∼ 45,000) and high-resolution (λ/Δλ ∼ 110,000) UV echelle modes of STIS are well-suited for nearby bright late-type stars, in terms of sensitivity, photometric stability, and wavelength precision (as validated by frequent calibration exposures of onboard hollow-cathode lamps). The long-term UV campaigns on α Cen AB were described in Paper I and also by Ayres (2015b). A newer set of observations of AB occurred in early-2021, after publication of Paper I. The STIS material for Procyon comes from two efforts. The first was the Advanced Spectral Library Project (ASTRAL ³ ), which obtained full UV coverage of Procyon—in medium- and high-resolution in the FUV, and high-resolution in the NUV—over seven separate HST visits between 2011 April and October. The second effort was a recent joint Chandra/HST campaign, which has been collecting more focused FUV and NUV spectra (the latter capturing the important Mg ii doublet at 2800 Å) since 2016, initially semiannually but more recently annually.

Table 4 is a catalog of the STIS echelle observations utilized here, of the three reference stars. A few blank exposures are excluded from the inventory owing to various types of failures, especially associated with Guide-Star reacquisitions. Usually, the failed observations were repeated within a few months. All of the FUV pointings on Procyon are included but only the NUV exposures that captured the Mg ii hk doublet, the main spectral target of interest here at the longer wavelengths. The table contains additional information, beyond the exposure specifications and epochs, namely empirical radial velocities and NUV flux scale factors.

Table 4. HST/ STIS Pointings: α Cen AB and Procyon

ObsNUM	ObsID	UTmid	Setting	Aperture	Range	${t}_{\exp }$	S/N	RVF	RVN	${{ \mathcal F }}_{\mathrm{NUV}}$
		(yr)		('')×('')	(Å)	(s)		(km s−1)
1	2	3	4	5	6	7	8	9	10	11
α Centauri A
1A	ob8w01010	2010.065	E140M-1425	0.2 × 0.2	1140–1729	3800	13	−22.1	⋯	⋯
2A	oblh01020	2011.108	E140M-1425	0.2 × 0.2	1140–1729	4950	15	−21.7	⋯	⋯
3A	oblh02020	2011.411	E140M-1425	0.2 × 0.2	1140–1729	4950	14	−22.0	⋯	⋯
4A	obua01010	2012.165	E140M-1425	0.2 × 0.2	1140–1729	4275	14	−21.4	⋯	⋯
5A	obua02010	2012.766	E140M-1425	0.2 × 0.2	1140–1729	4275	14	−22.2	⋯	⋯
6A	oc1i10010	2013.222	E140M-1425	0.2 × 0.2	1140–1729	4275	14	−22.9	⋯	⋯
7A	oc1i11010	2013.669	E140M-1425	0.2 × 0.2	1140–1729	4275	14	−22.5	⋯	⋯
8A	oc7w10010	2014.118	E140M-1425	0.2 × 0.2	1140–1729	4275	13	−20.3	⋯	⋯
9A	oc7w11010	2014.562	E140M-1425	0.2 × 0.2	1140–1729	4275	14	−21.7	⋯	⋯
10A	ocre10010	2015.018	E140M-1425	0.2 × 0.2	1140–1729	1500	8	−20.6	−19.5	1.255
	ocre10020	2015.018	E230H-2812	31 × 0.05NDC	2667–2931	500	59
11A	ocre11010	2015.613	E140M-1425	0.2 × 0.2	1140–1729	1500	8	−21.0	−19.9	1.269
	ocre11020	2015.613	E230H-2812	31 × 0.05NDC	2667–2931	500	58
12A	octr10010	2016.064	E140M-1425	0.2 × 0.2	1140–1729	1500	8	−19.9	−19.6	1.335
	octr10020	2016.064	E230H-2812	31 × 0.05NDC	2667–2931	500	57
13A	octr12010	2016.797	E140M-1425	0.2 × 0.2	1140–1729	1500	8	−21.8	⋯	⋯
14A	od5c10010	2017.107	E140M-1425	0.2 × 0.2	1140–1729	1499	9	−21.8	−19.6	1.125
	od5c10020	2017.107	E230H-2812	31 × 0.05NDC	2667–2931	500	62
15A	od5c11010	2017.708	E140M-1425	0.2 × 0.2	1140–1729	1499	8	−22.5	−19.4	1.139
	od5c11020	2017.708	E230H-2812	31 × 0.05NDC	2667–2931	500	61
16A	oduz10010	2019.494	E140M-1425	0.2 × 0.2	1140–1729	1750	9	−19.9	−19.1	1.056
	oduz10020	2019.494	E230H-2713	31 × 0.05NDC	2758–2835	500	53
17A	odzy10010	2020.392	E140M-1425	0.2 × 0.2	1140–1729	1750	9	−19.5	−18.4	1.000
	odzy10020	2020.392	E230H-2713	31 × 0.05NDC	2758–2835	500	54
18A	oeae10010	2021.137	E140M-1425	0.2 × 0.2	1140–1729	3800	13	−20.3	−18.4	1.023
	oeae10020	2021.137	E230H-2713	31 × 0.05NDC	2758–2835	500	53
α Centauri B
1B	ob8w02010	2010.497	E140M-1425	0.2 × 0.2	1140–1729	3600	9	−20.9	−21.9	14.417
	ob8w02040	2010.497	E230H-2713	0.1 × 0.03	2758–2835	884	18
2B	oblh01050	2011.108	E140M-1425	0.2 × 0.2	1140–1729	4950	10	−23.5	⋯	⋯
3B	oblh02050	2011.412	E140M-1425	0.2 × 0.2	1140–1729	4950	10	−22.3	⋯	⋯
4B	obua01020	2012.165	E140M-1425	0.2 × 0.2	1140–1729	4275	10	−21.6	⋯	⋯
5B	obua02020	2012.766	E140M-1425	0.2 × 0.2	1140–1729	4275	9	−26.0	⋯	⋯
6B	oc1i10020	2013.222	E140M-1425	0.2 × 0.2	1140–1729	4275	10	−21.1	⋯	⋯
7B	oc1i11020	2013.669	E140M-1425	0.2 × 0.2	1140–1729	4275	9	−25.6	⋯	⋯
8B	oc7w10020	2014.118	E140M-1425	0.2 × 0.2	1140–1729	4275	8	−21.1	⋯	⋯
9B	oc7w11020	2014.562	E140M-1425	0.2 × 0.2	1140–1729	4275	9	−25.4	⋯	⋯
10B	ocre10030	2015.018	E140H-1307	0.2 × 0.2	1199–1397	1500	5	−23.5	−23.2	1.160
	ocre10040	2015.018	E140H-1489	0.2 × 0.2	1390–1586	1500	2
	ocre10050	2015.018	E230H-2713	0.2 × 0.09	2758–2835	750	83
	ocre10060	2015.018	E230H-2862	31 × 0.05NDA	2723–2988	750	40
11B	ocre11030	2015.614	E140H-1307	0.2 × 0.2	1199–1397	1500	5	−23.3	−23.1	1.307
	ocre11040	2015.614	E140H-1489	0.2 × 0.2	1390–1586	1500	2
	ocre11050	2015.614	E230H-2713	0.2 × 0.09	2758–2835	750	78
	ocre11060	2015.614	E230H-2862	31 × 0.05NDA	2723–2988	750	53
12B	octr10030	2016.064	E140H-1307	0.2 × 0.2	1199–1397	1500	4	−24.5	−23.4	1.000
	octr10040	2016.064	E140H-1489	0.2 × 0.2	1390–1586	1500	2
	octr10050	2016.064	E230H-2713	0.2 × 0.09	2758–2835	745	71
	octr10060	2016.064	E230H-2862	31 × 0.05NDA	2723–2988	745	62
13B	octr12040	2016.797	E140H-1307	0.2 × 0.2	1199–1397	2000	6	−24.1	⋯	⋯
	octr12020	2016.797	E140H-1489	0.2 × 0.2	1390–1586	450	1
	octr12030	2016.797	E140H-1489	0.2 × 0.2	1390–1586	862	2
14B	od5c10030	2017.107	E140H-1307	0.2 × 0.2	1199–1397	1500	6	−25.3	−24.2	1.192
	od5c10040	2017.107	E140H-1489	0.2 × 0.2	1390–1586	1500	2
	od5c10050	2017.107	E230H-2713	0.2 × 0.09	2758–2835	745	81
	od5c10060	2017.107	E230H-2862	31 × 0.05NDA	2723–2988	745	25
15B	od5c11030	2017.708	E140H-1307	0.2 × 0.2	1199–1397	1500	5	−24.7	⋯	⋯
	od5c11040	2017.708	E140H-1489	0.2 × 0.2	1390–1586	1500	2
16B	oduz10030	2019.494	E140H-1307	0.2 × 0.2	1199–1397	2500	8	−24.1	−24.8	1.189
	oduz10040	2019.494	E140H-1489	0.2 × 0.2	1390–1586	1900	3
	oduz10050	2019.494	E230H-2713	0.2 × 0.09	2758–2835	500	67
17B	odzy10030	2020.392	E140H-1307	0.2 × 0.2	1199–1397	2500	8	−25.3	−24.7	1.034
	odzy10040	2020.392	E140H-1489	0.2 × 0.2	1390–1586	1900	3
	odzy10050	2020.392	E230H-2713	0.2 × 0.09	2758–2835	500	71
18B	oeae10040	2021.137	E140M-1425	0.2 × 0.2	1140–1729	2560	8	−24.6	−24.4	1.124
	oeae10030	2021.137	E230H-2713	0.2 × 0.09	2758–2835	500	68
Procyon (α Canis Minoris A)
1P	obkk13040	2011.259	E140H-1271	0.2 × 0.09	1163–1356	2850	10	−5.0	⋯	⋯
	obkk13030	2011.258	E140M-1425	0.2 × 0.06	1140–1729	2850	22
2P	obkk12040	2011.269	E140H-1271	0.2 × 0.09	1163–1356	1425	7	−4.9	−4.0	1.801
	obkk12030	2011.269	E140M-1425	0.2 × 0.06	1140–1729	1425	16
	obkk12010	2011.269	E230H-2762	0.2 × 0.05ND2	2621–2887	1824	54
3P	obkk11040	2011.307	E140H-1271	0.2 × 0.09	1163–1356	2850	10	−4.5	−4.1	1.497
	obkk11030	2011.307	E140M-1425	0.2 × 0.06	1140–1729	2850	23
	obkk11020	2011.307	E230H-2762	0.2 × 0.05ND2	2621–2887	2820	74
4P	obkk14040	2011.371	E140H-1271	0.2 × 0.09	1163–1356	2850	10	−3.9	⋯	⋯
5P	obkk16030	2011.399	E140H-1271	0.2 × 0.09	1163–1356	2850	9	−4.2	−4.2	5.514
	obkk16020	2011.398	E140M-1425	0.2 × 0.06	1140–1729	2850	22
	obkk16010	2011.398	E230H-2762	0.2 × 0.05ND2	2621–2887	1824	31
6P	obkk17010	2011.830	E140M-1425	0.2 × 0.2	1140–1729	1882	24	−4.7	⋯	⋯
7P	od0610010	2016.251	E140H-1307	0.2 × 0.2	1199–1397	2342	11	−5.3	−4.3	1.000
	od0610020	2016.252	E140H-1489	0.2 × 0.2	1390–1586	1393	12
	od0610030	2016.252	E230H-2762	0.2 × 0.05ND2	2621–2887	500	38
8P	od0611010	2016.777	E140H-1307	0.2 × 0.2	1199–1397	2342	12	−5.0	−4.0	1.371
	od0611020	2016.777	E140H-1489	0.2 × 0.2	1390–1586	1393	13
	od0611030	2016.777	E230H-2762	0.2 × 0.05ND2	2621–2887	500	32
9P	od5510010	2017.265	E140H-1307	0.2 × 0.2	1199–1397	2423	12	−4.5	−4.0	1.922
	od5510020	2017.265	E140H-1489	0.2 × 0.2	1390–1586	1521	12
	od5510030	2017.265	E230H-2762	0.2 × 0.05ND2	2621–2887	500	27
10P	od5511010	2017.823	E140H-1307	0.2 × 0.2	1199–1397	2423	12	−5.3	−4.2	1.141
	od5511020	2017.823	E140H-1489	0.2 × 0.2	1390–1586	1521	13
	od5511030	2017.823	E230H-2762	0.2 × 0.05ND2	2621–2887	500	35
11P	odg710010	2018.182	E140H-1307	0.2 × 0.2	1199–1397	2415	12	−4.7	−3.9	1.446
	odg710020	2018.182	E140H-1489	0.2 × 0.2	1390–1586	1526	13
	odg710030	2018.182	E230H-2762	0.2 × 0.05ND2	2621–2887	500	31
12P	odg751010	2018.762	E140H-1307	0.2 × 0.2	1199–1397	2355	11	−5.0	−4.1	1.335
	odg751020	2018.762	E140H-1489	0.2 × 0.2	1390–1586	1466	13
	odg751030	2018.762	E230H-2762	0.2 × 0.05ND2	2621–2887	500	33
13P	oduz11010	2019.817	E140H-1307	0.2 × 0.2	1199–1397	2355	12	−4.8	−3.5	1.352
	oduz11020	2019.818	E140H-1489	0.2 × 0.2	1390–1586	1386	12
	oduz11030	2019.818	E230H-2713	0.2 × 0.05ND2	2758–2835	500	31
14P	odzy11010	2020.258	E140H-1307	0.2 × 0.2	1199–1397	2355	12	−4.9	−3.9	1.093
	odzy11020	2020.258	E140H-1489	0.2 × 0.2	1390–1586	1386	13
	odzy11030	2020.258	E230H-2713	0.2 × 0.05ND2	2758–2835	500	34
15P	oeae11010	2021.257	E140M-1425	0.2 × 0.2	1140–1729	2355	26	−4.8	−3.5	1.113
	oeae11030	2021.257	E230H-2713	0.2 × 0.05ND2	2758–2835	500	33

Note. The exposures are grouped according to the epoch of observation. Col. 3 is the UT of midexposure. Decomposition of the the Col. 4 echelle settings is as follows: "140" are FUV, "230" are NUV; "M" is medium resolution, "H" is high; trailing 4-digit CENWAVE is in angstroms. Col. 5: apertures ending in ND# are neutral-density filtered slits (at 2800 Å, the effective NDs are: NDA ∼ 0.6, NDB ∼ 1.1 (not used here), NDC ∼ 1.3, and ND2 ∼ 2.1); 0.2 × 0.2 is the "photometric" slot; 0.2 × 0.06 and 0.2 × 0.09 are default "spectroscopic" slits for M and H, respectively. Col. 6 is the approximate wavelength coverage of the setting. Integrations longer than 2500 s (Col. 7) were normally split into two, or more, equal length sub-exposures. Col. 8 is the average signal-to-noise per resolution element (resel: 2 pixels), which is more meaningful for the continuum-dominated NUV but less so for the emission-line dominated FUV. Cols. 9 and 10 are the derived radial velocities of the FUV and NUV spectra, respectively, based on narrow low-excitation chromospheric emissions (FUV) or photospheric absorptions (NUV). Col. 11 is a flux scale factor applied to the NUV spectrum to counteract variable slit throughputs.

Download table as:  ASCIITypeset images: 1 2 3

The latter were compensation for the heavy use of neutral-density (ND) filtered slits owing to the high continuum intensities in the NUV regions of especially α Cen A and Procyon, which normally would violate MAMA camera bright limits. All of the ND-filtered slits are very narrow (005), so are subject to variable throughputs caused by thermally driven "breathing" of HST's telescope assembly. Furthermore, the 31''-tall NDA and NDC slits that were used for some of the AB NUV exposures required an ORIENT constraint to avoid having both stars on the long slit at the same time, which would not only create a very messy echellegram but would also cause a serious NUV-MAMA global bright-limit violation.

There was one related case where the tiny 0.1 × 0.03 ⁴ "Jenkins" ultra-high spectral resolution aperture was employed, again in an effort to cut down the otherwise troublesome NUV light levels. The throughput of this aperture can especially be affected by telescope breathing, as was the case for the leading α Cen B NUV observation ("1B" in the table). Subsequently, the 31 × 0.05NDA (ND ∼ 0.6) long slit was used for α Cen B with setting E230H-2812, with much better results. Later, it was recognized that setting E230H-2713 could be paired with the standard clear spectroscopic slit, 0.2 × 0.09, without violating the bright limits, despite predictions of the (overly conservative) Exposure Time Calculator to the contrary. Thus, the clear slit has been featured in the more recent α Cen B NUV exposures. Unfortunately, the enhanced photospheric NUV continua of hotter α Cen A and Procyon still required the ND slits, including the very attenuated 0.2 × 0.05ND2 aperture for Procyon.

The variable effective transmission of the ND-filtered slits was countered by determining the effective throughput (integrated counts per second) in a pair of 8 Å NUV continuum bands at 2780 Å and 2820 Å, flanking the Mg ii doublet, for each star and then adjusting the other flux scales to the one with the highest value. This has the undesirable side effect of eliminating any variability in the NUV continuum, but that was considered a minor annoyance given that the observed solar MIN to MAX contrast at 2820 Å was a mere 0.9% (${f}_{\max }/{f}_{\min }-1$: Table 7, Paper I).

In the FUV region, even the superbright H i 1215 Å Lyα emissions of all three reference stars are well below the detector safety thresholds. Whenever possible, the 0.2 × 0.2 "photometric" slot was used for the FUV exposures to preserve photometric accuracy. However, the ASTRAL sequence on Procyon, circa 2011, employed the FUV spectroscopic slits (0.2 × 0.06 for M; 0.2 × 0.09 for H) for almost all of the program to achieve optimum spectral purity for the Library Project. However, one E140M-1425 exposure was taken with the high-throughput 0.2 × 0.2 slot to serve, if necessary, as a photometric scaffold for the other FUV-M and FUV-H spectra.

High-southern-declination Alpha Cen could be observed in HST's Continuous Viewing Zone (CVZ), which allowed both stars to be captured in a single visit of two orbits. Procyon, a low-declination target outside the CVZ, also required two orbits per visit to achieve high-S/N in both the FUV and NUV, owing to the lower efficiency of the Earth-occulted passes.

The typical observing sequence for α Cen AB involved an initial slew to the primary, based on a predicted location in that epoch taking into account the large proper motion of the system and the non-negligible orbital motion of A; an ND-filtered CCD acquisition of the brighter primary in visible light; an FUV medium-resolution observation of A through the 0.2 × 0.2 slot (the CCD centering was accurate enough for the purpose); then a peak-up with an ND-filtered slit; and lastly a high-resolution NUV exposure of the Mg ii region using the same slit.

Following the A sequence, a short offset slew was performed to the predicted location of the secondary star according to the orbital ephemeris. In earlier years of the program, when the AB separation was less than about 6'', a peak-up was performed with one of the low-resolution visible-light gratings and the 0.3 × 0.05ND3 slit to precisely center the secondary star following the offset maneuver. In recent years, the AB separation has increased to the point where a simpler CCD acquisition could be used to capture B following the short offset slew, without the possibility that A might accidentally graze the 5'' × 5'' "postage stamp" of the sub-array field of view.

After centering B, either a single E140M (possibly split into sub-exposures) or a pair of overlapping E140H's (1307 Å and 1489 Å) were taken, again through the 0.2 × 0.2 aperture to ensure good photometry. The emission lines of B are somewhat narrower than those of A, and the thought was they would benefit from the significantly higher (roughly 2.5 times) resolution achieved with the E140H settings. However, the higher resolution was at the expense of lower S/N, given that the available time had to be split between a pair of H exposures to cover the same spectral territory as a single E140M. Recent versions of the program have returned to FUV medium resolution for B to achieve higher signal at the fainter features, especially a pair of Fe xii coronal forbidden lines. Following the FUV sequence, an exposure of B's Mg ii region was obtained, initially using the mildest of the ND-filtered long slits but more recently with the 0.2 × 0.09 clear aperture (for reasons described earlier). In contemporary incarnations of the long-term program, deeper-than-normal wavecal exposures have countered the fading brightness of the lamps over many years of use.

The (post-ASTRAL) observing sequence for Procyon was analogous to that of α Cen B. Following the initial CCD acquisition, two overlapping E140H exposures were taken to cover the FUV region. The FUV lines of Procyon are broader than those of the α Cen stars but Procyon is also brighter in the ultraviolet, so the reduced FUV-H sensitivity did not affect the S/N so much and provided exceptional resolution at the several important interstellar absorptions. The very bright Mg ii region required the heavily attenuated ND2 slit and a dispersed-light peak-up prior to the NUV exposure. Similar to α Cen B, recent versions of the Procyon program have replaced the pair of E140H exposures with the single setting E140M-1425, again to boost S/N at the faint Fe xii features.

2.3.1. STIS Spectral Processing

The STIS echellegrams were processed through a series of protocols that were developed for the ASTRAL Project mentioned earlier. One major departure from ASTRAL was the use of the standard CALSTIS pipeline "x1d" data product, a set of wavelength-calibrated, extracted spectra (and photometric errors) together with data quality flags, tabulated for each of the several dozen echelle orders of a setting. ASTRAL also utilized the x1d file, but from a custom version of the pipeline with specially crafted reference files to accommodate upgraded wavelength and photometric calibrations developed by the author for the large-scale Library Project—which, at present, contains (mostly) full-coverage FUV + NUV spectra of about 40 stars, representing early and late types. The main feature of the advanced processing was a detailed correction to the pipeline-assigned wavelength scales to account for subtle, but systematic, distortions of the camera images of higher spatial frequency than was accounted in the relatively low-order polynomial dispersion relations used by CALSTIS (an earlier version of the approach was described by Ayres 2010). However, it was judged that the rather complex wavelength correction utilized in the ASTRAL protocols was excessive for the vast majority of practical applications. Furthermore, given the evolving nature of the CALSTIS reference files, especially with regard to photometric calibrations, it would be better to design a simpler correction that could be applied directly to the CALSTIS x1d files, as disgorged from the on-the-fly (OTF) calibration system in the MAST archive. ⁵ The new wavelength correction strategy is briefly outlined in Appendix B.

CALSTIS x1d files for the three STIS stars were copied from the MAST archive as of 2021 late April, as OTF-processed with the most up-to-date reference files available at the time. These order-separated spectra were then merged into coherent 1D tracings using one of the ASTRAL procedures (called "UNPACK"). In many cases, especially the E140M exposures, a long observation had been split into two or three equal length sub-exposures. These were combined using ASTRAL procedure "ZERO," which registers the wavelength scales by cross-correlation prior to co-adding the sequential sub-exposures. In the specific case of Procyon, there were visits in which an initial E140M was taken in the leading orbit and then another E140M was taken at the beginning of the second orbit, usually with a different duration to allow for the subsequent NUV peak-up and exposure to fill out the orbit. These unequal exposures, but from the same setting and epoch, were combined with another ASTRAL procedure, "ONETWO," using the same registration strategy as ZERO but now weighting the flux densities according to the total net counts in each exposure. In a number of cases, adjacent overlapping spectra were taken in the FUV or NUV (or both) in a given epoch. These pairs were spliced together, utilizing the ASTRAL procedure "THREE." The spectra were aligned in wavelength by cross-correlating a prominent sharp feature (emission or absorption) in the overlap zone between the settings and the intensity scales were adjusted, if necessary, according to the ratio of the integrated flux densities in that overlap zone.

Finally, the ONETWO and THREE procedures were used sequentially to construct epoch-average FUV and NUV spectra in the distinct resolution modes: FUV-M for α Cen A; FUV-M and FUV-H for α Cen B and Procyon; and NUV-H for all three targets. The results are shown in Figure 2 for FUV-M and NUV-H: α Cen A (dark curves), α Cen B (red), and Procyon (blue). The upper panel is FUV-M, while the lower panel is a subsection of NUV-H. The tracings are presented in observed flux densities, which provided a better separation among the stars.

Zoom In Zoom Out Reset image size

In the upper panel (on a logarithmic flux scale), the downward sloping FUV continuua of the three stars dominate the appearance at the long wavelength end, while a progression of sharp emission lines becomes more prominent at the short end. The strong feature near 1215 Å is H i Lyα, which is by far the most intense emission of the FUV region of late-type stars. Note the prominent jump (toward shorter wavelengths) at the Si i photoionization edge near 1520 Å in α Cen B, but apparent absorption dip (albeit weak) at the edge in Procyon.

In the lower panel for the NUV, now on a linear scale, the depressed photospheric continuum of cooler α Cen B is conspicuous compared to bright, warmer Procyon; yet the chromospheric Mg ii hk emission peaks, which are barely visible at the bottoms of the deep photospheric Mg ii absorptions bracketing 2800 Å, are roughly the same strength in all three stars (in observed flux). The other deep absorption feature in this interval, near 2850 Å, is the Mg i 2852 Å resonance line. The Mg ii emissions appear rather uninspiring relative to their surroundings compared with, say, FUV Lyα, but in reality the combined hk core flux is several times that of the hydrogen line in all three stars.

Before describing the detailed measurements of the UV emission lines of the three STIS stars, it is instructive to show, in Figure 3, the epoch-average profiles of representative features—C ii 1334 Å, C ii 1335 Å, Si iv 1393 Å, and Mg ii 2796 Å (k)—and the solar counterparts, at similar spectral resolution, from the IRIS full-disk mosaics described in Paper IV. The upper darker curves for the Sun are averages from near the Cycle 24 peak. The lower red curves represent an average over very quiet conditions. The darker curves for α Cen B and Procyon are from epoch-average high-resolution STIS spectra. The red curves for all three stars (including α Cen A) are from epoch-average medium-resolution FUV spectra. (The NUV Mg ii k line was exclusively recorded in high resolution.) Deep interstellar absorptions appear in the low-ionization ground-state transitions C ii 1334 Å and Mg ii 2796 Å, which are sharper in high resolution than medium resolution. Of course, the solar profiles are not affected. The ISM features are more centered in the α Cen stars but significantly redshifted in Procyon, due to the differences between the stellar systemic velocities and those of the local interstellar clouds.

Zoom In Zoom Out Reset image size

Additional small differences in the ISM absorptions between α Cen A and B accrue from the relative orbital motion (most obvious in high-resolution Mg ii). Also note the tiny redshifted divot at the peak of Procyon's excited-state transition C ii 1335 Å (whose lower level is 63 cm⁻¹ above ground). This notch must be interstellar, but the absorption is severely reduced compared to the ground-state resonance line owing to the radiative depletion of the C⁺ excited fine-structure population under the extremely tenuous conditions in the local interstellar gas. The reduced ISM absorption in C ii 1335 Å has revealed clear central reversals in the stars owing to high opacity, mimicking the slanted-topped peak of the solar full-disk profile (high-spatial resolution tracings of solar 1335 Å often show more obvious central dips: see Paper IV). The same story is repeated in Mg ii 2796 Å, as follows: strong central reversals in the solar profiles; somewhat obscured reversals in the α Cen stars; and a cleaner feature in Procyon, thanks to the redshifted ISM absorption. Superficially, the α Cen A and solar profiles are very similar in shape and strength; the α Cen B counterparts are narrower and stronger than solar (with the f_λ /f_BOL normalization); and the Procyon features are wider and stronger, most conspicuously chromospheric C ii and Mg ii. Aside from the sharp interstellar features, differences between the medium-resolution and high-resolution FUV line shapes of α Cen B and Procyon are comparatively minor.

As described in the Introduction, Paper IV introduced a pseudo-Gaussian fitting model,

$$\begin{eqnarray}&&\phi (\lambda )={e}^{-{\left(| \lambda -{\lambda }_{0}| /{\rm{\Delta }}{\lambda }_{{\rm{D}}}\right)}^{a}},\end{eqnarray} \tag{ 1 }$$

in which the normal pure-Gaussian exponent (a = 2) was allowed to take on arbitrary values. Here, λ₀ is the line-center wavelength and Δλ_D is the characteristic e-folding line width. A pure Gaussian, as has been noted in prior solar and stellar work, tends to have a shallower, broader peak, slightly bowed sides, and narrower wings than typical empirical TZ line shapes. In some previous work, the mismatch was accommodated by introducing a bimodal Gaussian model in which a broad component provided the extended line wings, while a narrow component contributed the sharper central peak. However, Paper IV demonstrated that the simple pseudo-Gaussian model could replicate a diversity of solar line shapes, including the narrow-core, broad-wing profiles of optically thin TZ lines, as well as the boxier shapes of the optically thick chromospheric features (albeit ignoring the central reversals if too prominent, as in the Mg ii features).

Figure 4 illustrates two of the main optically thin TZ lines, N v 1238 Å and Si iv 1393 Å, from the epoch-average medium-resolution spectra of the three STIS stars. These features are less affected by extraneous blends than other TZ emissions (although note the faint Ni ii feature in the blue wing of the Si iv line of Procyon). Each emission profile was fitted by a least-squares Gaussian (red) and a pseudo-Gaussian model (green; derived a exponent listed for each line; a = 2 is the pure Gaussian). Dark dots represent the observed STIS spectra. Points within the intervals marked by the blue inverted-L symbols were included in the modeling for each approximation. Horizontal blue-dashed lines depict long-range continuum fits, which were subtracted from the target flux densities prior to the profile modeling. Each observed feature was normalized to its peak intensity for display purposes: note the y-axis scales at the left and right.

Zoom In Zoom Out Reset image size

The pseudo-Gaussian profiles match the observed line shapes better than the pure Gaussians, which is unsurprising given the additional degree of freedom available. Perhaps more intriguing is that the derived pseudo-Gaussian exponents are similar for the two different TZ lines in each star, despite the obvious diversity in line widths; as well as similar among the stars, with values ∼1.5, significantly less than the pure-Gaussian case. (Uncertainties in the modeling were assessed by a Monte Carlo approach, in which the best-fit profile was Gaussian-perturbed point-by-point according to the local photometric errors, then refitted, in numerous independent trials. The standard deviations of the derived parameters, especially the a exponent, over the many trials was a gauge of the uncertainties.)

Paper IV further showed that the straightforward interpretation of the double-Gaussian alternative—that each component was contributed by a separate, distinct class of surface structures with different internal kinematics, which became blended in the disk average—was too simplistic because the same pseudo-Gaussian line shape of Si iv 1393 Å was seen not only in the full-disk average but also down to the finest spatial scales of the IRIS mosaics. This suggested that some widespread process was shaping the TZ lines at very fine spatial scales. One possibility mentioned was the so-called "κ-mechanism" (Dudík et al. 2017) owing to the similarity between the κ profile and the pseudo-Gaussian model for a < 2. There might be other, alternative non-Maxwellian processes that could produce the characteristic TZ line shape as well. In Paper IV, the pseudo-Gaussian modeling was preferred over the κ profile from the pragmatic point of view that the former can accommodate both the sharply peaked a < 2 TZ profiles, as well as the more rounded a > 2 line shapes of the optically thick chromospheric features; whereas, the κ profile has the limiting form of a pure Gaussian for large κ, and thus cannot access the realm inhabited by the boxier shapes of, say, the C ii and Mg ii emission cores. (The more rectangular a > 2 profiles also are inaccessible to the double-Gaussian modeling.) For that reason of generality, the pseudo-Gaussian line shape was preferred here to model the UV profiles of the three STIS stars.

The pseudo-Gaussian was implemented in an algorithm driven by parameters that were customized for each target spectral feature and each star. The emission lines were affected to varying degrees by accidental blends, whose influence varied among the stars (e.g., the Ni ii blend in Si iv in Figure 4 and the interstellar absorptions seen in Figure 3). In addition, the shapes of the target emission lines also varied among the stars (specifically the line widths). Two separate line lists were used to drive the algorithm.

The first list (Table 5) was designed to validate the velocity scales of the individual (or average) spectra, and consisted of a set of narrow low-excitation neutral chromospheric emissions for the FUV, and photospheric absorptions (in the Mg ii hk wings) for the NUV. Although careful corrections to the STIS (relative) wavelength scales were made during the initial processing steps, the absolute zero-point of an observation can be affected by the accuracy of the target centering, or by subsequent drifts and telescope breathing. Beyond that, all three STIS stars are in binary systems, so there are additional orbital velocity shifts, which vary over time. The strategy was to assume that low-excitation FUV emission lines forming in the deep chromosphere, or absorption lines in the Mg ii wings arising in the high photosphere, would have instantaneous velocities close to that of the photosphere, and thus could serve as zero-point calibrators. As a first step in the analysis, all the individual spectra, and the various M and H epoch-average versions, were adjusted in radial velocity according to the average of the respective sets of FUV or NUV reference lines. The derived values of the apparent radial velocities for the individual STIS spectra were noted previously in Table 4.

In the Table 5 velocity calibrator line list, there are two sets of wavelength offsets, one for the AB lines, the other for Procyon. The offsets—which are sometimes symmetrical, but sometimes not—indicate the wavelength range relative to the laboratory value, λ_LAB, over which the particular profile was fitted (also using the pseudo-Gaussian model). Asymmetric offsets accommodated situations such as O i 1355 Å, where another feature (in this case a weak C i transition to the red) was close enough to affect one of the line wings.

The second line list (Table 6) was for the target emission lines of the FUV and NUV regions, including several broad continuum bands. In the NUV, as noted earlier, the continuum bands were employed to adjust the flux density scales to counter narrow-slit variable transmissions. Parameters for the three stars are tabulated in separate rows organized by line transition or continuum band. The first pair of entries in each row are wavelength bounds for an integrated flux measurement, which are asymmetric in some cases for the same reasons as given earlier for the velocity reference lines. The middle pair denotes the outer bounds of the pseudo-Gaussian fitting, as in Table 5 for the velocity calibrators. The final set of entries in the row, if present, indicates an inner wavelength interval that was ignored because of an optically thick central reversal or a sharp interstellar absorption. In what follows, the quoted fluxes of the lines will be the integrated values, encompassing a central reversal or ISM absorption if present; rather than, say, the value derived from the pseudo-Gaussian fit itself. This was done to ensure that the time histories of the fluxes were not affected by artifacts associated with the fitting process, or the over-prediction of the line intensities in the cases where central reversals were ignored.

For practical numerical reasons, the least-squares pseudo-Gaussian fitting algorithm restricted the exponent a to the range 1–5, which accommodated the vast majority of the STIS profile shapes. Only the broad Lyα feature, which is severely truncated in its core by interstellar absorption, encroached on the boundaries of the restricted a range, in that case on the low side (a = 1 is a pure exponential).

Figure 5 illustrates the profile fitting approach applied to the epoch-average STIS spectrum of α Cen A (additional examples for α Cen B and Procyon can be found in Appendix C). Each panel depicts one of the 24 emission lines or broad continuum bands from Table 6. For all the features, flux densities were numerically integrated over the wavelength bandpasses marked in green. Blue-dashed lines under the emission features are long-range continuum fits ⁶ that were subtracted prior to the fitting procedure (not applied to the continuum bands). In the cases of O v 1218 Å and coronal forbidden line Fe xii 1241 Å, the continuum fits were allowed to be tilted, or curved, to account for the wings of nearby stronger emissions: Lyα and N v 1242 Å, respectively. (Here, the outcome is more noticeable for O v.) Most of the emission lines then were subjected to the pseudo-Gaussian modeling, taking target flux densities between the outer blue inverted-L symbols, and, in cases of strong central reversals or ISM absorptions, ignoring the points between the inner inverted-Ls. For α Cen A, here, and Procyon later (in Appendix C), both of the faint Fe xii coronal forbidden lines were too noisy to be reliably fitted by a pseudo-Gaussian. In contrast, the Fe xii features of α Cen B were much stronger thanks to the higher intensity corona of the K dwarf, and thus were able to be successfully modeled.

Zoom In Zoom Out Reset image size

The Mg ii hk lines were a special case. A minimum wing intensity (highlighted by a yellow dot) was determined in a slightly smoothed version of the local spectrum in a narrow band (marked by vertical orange tics) on whichever side of the k or h line that was least affected by blends. This baseline flux density was subtracted from the k or h profile prior to fitting, in the same way as the background continuum was compensated in the more general case. The difference here is that some fraction of the Mg ii flux under the baseline represents legitimate chromospheric emission, because without a chromosphere the line cores would be deep, broad U-shaped absorption features (see Linsky & Ayres 1978). These absorption cores then are filled-in to varying degrees by chromospheric emission, depending on the activity level. Nevertheless, the adopted background-subtraction approach for Mg ii was expedient, avoiding overestimating the hk flux if instead the flux integration were done without the subtraction (i.e., integrating all the flux densities above zero between the local minimum features); and further avoiding a challenging correction to establish the exact proportion of the "block flux" under the Mg ii minima that truly is chromospheric. The same scheme was applied to Mg ii in the Sun and α Cen AB in Paper I, with good results. Namely, the background-subtracted hk fluxes displayed tight correlations with other chromospheric tracers, such as the O i 1305 Å triplet and the C ii 1335 Å multiplet. At the same time, the k and h block fluxes (c2796 and c2803, respectively), which were retained separately, also showed good correlations with legitimate chromospheric diagnostics but with much shallower slopes than the corresponding Mg ii core fluxes, as anticipated if a chromospheric contribution is present but is relatively minor.

Table 7 summarizes the line shape parameters and bolometrically normalized fluxes measured in the FUV-M and NUV-H epoch-average spectra of the three stars, again presented in separate rows. Also included are MAX/MIN flux contrasts (defined as ${f}_{\max }/{f}_{\min }-1$ in percents based on the time-resolved spectra). The leading uncertainties provided on the profile shape parameters are 1σ measurement errors (related to the S/N), while the following parenthetical values are 1σ standard deviations over the available epochs, excluding the two most extreme outliers in each series, probing variability of the respective quantities over the stellar activity cycles (or a more quiescent period, as was the case for Procyon).

Figure 6 is a super-summary of the integrated UV flux measurements of the three STIS stars and the Sun, also including the broadband X-rays. Error bars on AB's X-rays were determined by 0.3 ks parsings of the Chandra event lists (∼1.5–3 hr total duration each) including any flare excursions (so far only clearly recognized in α Cen B), whereas the flares were intentionally filtered out for the mean values. Error bars for ostensibly constant Procyon were simple Poisson $\sqrt{N}$. Each time series was normalized to an average flux, f_AVE, as determined from the time interval outlined by vertical dotted–dashed lines in each panel (possibly different for the X-rays, and Mg ii of α Cen A, compared with the FUV). The selected intervals were intended to encompass a local cycle minimum of the time series, or as close to a minimum as practical. The normalization was according to f_L/f_AVE − 1. In the Chandra X-ray series, a 2% per year sensitivity decline of the HRC-I camera since 2010 was compensated. That decrease is roughly compatible with predictions by the exposure time calculator WebPIMMS, and entirely flattens the Procyon coronal time series.

Zoom In Zoom Out Reset image size

The other emissions of Procyon that were recorded independently by STIS, with its own sensitivity variation model, show very stable long-term behavior; aside from perhaps the Fe xii coronal forbidden line, ⁷ which is weak and has lower S/N than the more prominent subcoronal emissions. The three Sun-like stars, excluding Procyon, display excursions of their integrated fluxes that generally follow the (much larger) changes in coronal X-rays, and the amplitudes generally increase with formation temperature (from bottom to top in each panel). Figure 6 further emphasizes the remarkable long-term stability of subgiant Procyon across the broad temperature range spanned by its corona and subcoronal atmosphere, despite its rather respectable super-solar L_X/L_BOL.

The following diagrams compare different aspects of the emission-line profile shapes, separated according to low-temperature chromospheric species (T < 2 × 10⁴ K) and high-temperature TZ species (T > 5 × 10⁴ K). The representatives of the chromosphere are: O i 1306 Å ("O1306" in the figure legends); Cl i 1351 Å (Cl1351); C ii 1335 Å (C1335); and Mg ii 2796 Å (Mg2796). The TZ representatives are: Si iii 1206 Å (Si1206); N v 1238 Å (N1238); Si iv 1393 Å (Si1393); and C iv 1548 Å (C1548).

Figures 7(a) and (b) pit the pseudo-Gaussian exponent a against the e-folding line width, Δλ_D as in Equation (1), for the two temperature classes, and include all the available measurements of the three STIS stars regardless of FUV resolution mode, M or H (all the NUV spectra were H). The e-folding line width appears here instead of the more common FWHM because there is already a built-in correlation between FWHM and the pseudo-Gaussian a factor:

$$\begin{eqnarray}&&\mathrm{FWHM}=2{(\mathrm{ln}2)}^{(1/a)}\,{\rm{\Delta }}{\lambda }_{{\rm{D}}},\end{eqnarray} \tag{ 2 }$$

so the e-folding width is, in some sense, more fundamental (although both, of course, are interchangeable).

Zoom In Zoom Out Reset image size

Figure 7(a) is for the low-temperature group. The individual species are color-coded as in the upper left-hand legend. The STIS stars are represented by different symbols according to the legend at the lower right. Error bars on the individual epoch-resolved measurements are ±1σ. The diagram illustrates systematic correlations among the line shape parameters with increasing opacity: Cl1351 → O1306 → C1335 → Mg2796. Namely, the line widths increase, and the shapes evolve from sub-Gaussian (sharper cores) for the lower opacity species to super-Gaussian (boxy cores) for the optically thickest. Also note the two distinct clusters of Cl1351 points for α Cen B and Procyon, which is an influence of the M and H resolution difference and is most conspicuous in the thinnest line (both in optical depth and e-folding width). (Recall that α Cen A was observed exclusively in medium resolution, so there is no double clustering of its Cl1351 points.) A similar separation is seen in the O1306 lines of α Cen B and Procyon, and in C1335 of α Cen B (less obvious for Procyon). Furthermore, several of the species display tilted correlations between a and Δλ_D, suggesting that the profiles become more rounded as they become broader. Finally, there is a general increase in the line widths α Cen B → α Cen A → Procyon, as noted earlier in Figure 3.

Figure 7(b) is for the high-temperature group, again with color-coding for the emission species and symbol-coding for the STIS stars. In contrast to the previous figure, here almost all the line shapes are sub-Gaussian, with only α Cen A's Si1206 encroaching slightly into super-Gaussian territory. Among the species illustrated, Si1206 is most likely to be optically thick—all of the others are more likely to be thin. There is not as clear a separation among the species of a given star in terms of line width but rather the strong separation by star is still maintained, with α Cen B displaying the narrowest lines and Procyon the widest. Furthermore, most of the species show relatively constant widths over a small range of measured a values, although the Si1206 trend for Procyon is noticeably tilted in the same sense as C1335 previously in Figure 7(a)—namely, the profiles become more rounded as their widths increase.

Figures 8(a) and (b) are similar to Figure 7 but now for the FWHM versus normalized line strength f_L/f_BOL (which is a proxy for cycle phase, at least for α Cen AB). Figure 8(a) is for the low-temperature species (same legends as Figure 7(a)). Here, the distinct lines are arrayed in flux in the same opacity pattern as previously: Cl1351 the thinnest and lowest in flux; Mg2796 the thickest and highest in flux. Across all of the stars, the average line width of a species generally increases with increasing average line strength. However, within the time series of a given species, there is a suggestion that the widths decrease with increasing line strength, especially for α Cen B; although the series for Procyon displays an insignificant variation in line fluxes, but a modest dispersion in measured line widths (the dispersion might reflect mainly measurement errors). The same M versus H separation in the FUV line widths for α Cen B is present here (and for Procyon as well), which might be biasing the suggested negative trend. However, the FUV-M and FUV-H spectra were fairly evenly distributed across the recent peaks and valleys of B's cycle.

Zoom In Zoom Out Reset image size

Figure 8(b) is for the hotter lines. Here, the widths are strongly clustered according to the star: α Cen B has the narrowest lines; Procyon the widest; with α Cen A in between. Now, the trend of decreasing line width with increasing line strength is more pronounced, at least for α Cen B, which has a much larger contrast in the line intensities over its cycle than α Cen A. The dichotomy between M and H spectra in the previous figure is less apparent because the TZ lines tend to be broader than their chromospheric cousins, at least when compared to Cl1351 and O1306. Again, the Procyon widths display a modest variation, which is more or less consistent with the measurement errors, while the line fluxes are essentially constant.

The next set of diagrams, Figure 9, depict line shifts versus normalized fluxes, again separated into chromospheric and TZ species. Figure 9(a) is for the low-temperature lines. Cl1351 clusters around zero for all of the stars, which was expected because it was one of the four narrow chromospheric emissions that established the radial-velocity normalization for each FUV spectrum. The O1306 feature is also close to zero for the α Cen stars, but moves into slightly positive territory for Procyon. C1335 displays a small redshift in all of the stars, while Mg2796 is slightly blueshifted, more so for Procyon but less so for α Cen B. Significantly, the same opposing behavior of redshifted C ii and blueshifted Mg ii was seen in spatially resolved full-disk images of the quiet Sun, which were derived from one of the IRIS mosaics in Paper IV (see Figures B1(a) and (b) therein).

Zoom In Zoom Out Reset image size

Figure 9(b) is the analogous comparison for the hot lines. The appearance is entirely different than Figure 9(a), with all of the species displaying significant redshifts; less so for α Cen B, and more so for Procyon, in order of decreasing surface gravity. Within a given species, there is little apparent dependence of the shifts on the line intensities, especially in the α Cen B time series, which display the largest line-to-line flux contrasts from cycle MIN to MAX. Again, Procyon exhibits a dispersion of centroid values, which is more or less consistent with the measurement errors, over a narrow range of normalized fluxes.

Figure 10 summarizes the Doppler shift behavior of the low-temperature and high-temperature species together, arrayed as a function of formation temperature. Since there was little dependence of the line shifts on the fluxes, the highest S/N epoch-average spectra of each STIS star were appropriated for the purpose (FUV-M and NUV-H). Furthermore, many of the species illustrated in the previous figures were part of a multiplet. Now, both major members are shown (ordered in increasing wavelength from left to right in each shaded block). In addition, several new species appear, such as O i 1355 Å (which is another of the narrow chromospheric RV calibrators) and O v 1218 Å (which is the hottest of the high-S/N TZ lines, at least in CIE).

Zoom In Zoom Out Reset image size

The two purple shaded bars at the left-hand side of the diagram are for Cl i 1351 Å and O i 1355 Å. The O i features illustrated just to the right, in the gray-shaded band, are 1304 Å and 1306 Å, which are the two members of the oxygen triplet least affected by ISM absorption. The other species in the diagram are self-explanatory. As described previously, Mg ii displays a small blueshift in all the stars, now in both spectral components, and C ii a small redshift, also in both members. The hotter TZ species mostly show a smooth change in their redshifts from Si iii temperatures (5 × 10⁴ K) up to O v (∼2 × 10⁵ K). An exception is Procyon's C iv 1550 Å, which displays a dramatically smaller redshift than the stronger member of the doublet. Examination of the epoch-average FUV-M and FUV-H spectra of Procyon (Figures C2(a) and (b), in the Appendix) finds an oddly enhanced blue wing on the 1550 Å component, which is probably caused by an unrecognized blend, and apparently pulls the line centroid slightly toward the blue. A similar distortion is not obvious in the 1550 Å features of either α Cen A or B. The Procyon spectrum is characterized by a more dominant FUV continuum than the other stars and a different set of prominent low-temperature species (e.g., numerous Ni ii transitions). Ignoring the minor distraction of 1550 Å, it is clear from Figure 10 that all three STIS stars display similar velocity profiles through the TZ, ostensibly peaking near $\mathrm{log}T\sim 5.1$, but could equally be constant through that region given the uncertainties in the laboratory wavelengths of several of the lines—especially O iv, which is an intersystem transition, and O v, which is an intercombination line; both of which are difficult to measure in the laboratory.

There have been numerous reports of Doppler shifts of FUV (and EUV) TZ, and cooler, lines in the Sun. Among the studies most relevant to the present work, Hassler et al. (1991) found that the downflows in C iv 1548 Å—measured by a sounding rocket that obtained center-to-limb spectra wavelength-calibrated in flight by an onboard Pt lamp—were dominantly radial and had a peak velocity of nearly +8 km s⁻¹ at disk center. Furthermore, C iv showed the prominent emission ring at the extreme limb described in Paper IV for similar temperature Si iv. Cooler chromospheric lines in the sounding-rocket spectra, such as Fe ii and Si ii, displayed velocities closer to zero, as did hotter Ne viii 770 Å (T ∼ 6 × 10⁵ K), the latter observed in the second order of the grating together with first-order C iv. A second pertinent study was conducted by Brekke (1999), who summarized several previous investigations. He proposed that the solar redshifts at disk center increased smoothly from near zero at 10⁴ K in the upper chromosphere, to a peak of around +10 km s⁻¹ at 2 × 10⁵ K, and then perhaps declined toward higher (e.g., Ne viii) temperatures (although extension to the hotter, near-coronal, regime was hindered by the lower-resolution spectrometers used in the EUV, lack of internal calibration lamps, and less well-known laboratory wavelengths). A qualitatively, if not also quantitatively, similar trend is seen among the STIS stars, noting that if the stellar downdrafts are primarily also radial, then the disk-average redshifts (as measured for the stars) would be about half that of a disk-center perspective. The stellar measurements, here, benefited from a higher-resolution spectrometer than is typically available for solar studies, together with decades of experience calibrating the STIS instrument using its high-quality internal hollow-cathode wavelength lamps. The agreement between the stellar full-disk and solar disk-center trends is encouraging and suggests that similar physical processes are operating in the respective outer atmospheres.

To reiterate: all of the line shape diagrams have shown differences—in several cases systematic—that exist among the three STIS stars in their UV spectra. Yet, the most striking finding is how similar—in pseudo-Gaussian exponents, widths, and Doppler shifts—Procyon is to the α Cen stars (and the Sun), despite the rather different physical properties of the subgiant (especially convection zone depth and evolutionary status).

The final comparisons of the present study are a series of flux–flux diagrams, which illustrate the dependence of the normalized intensities of one species against those of another (e.g., coronal X-rays versus chromospheric Mg ii). These comparisons are based on integrated fluxes of the stellar emission lines, rather than the detailed profile line shapes, so the corresponding low spectral resolution, but full-disk, solar irradiance results from Paper I can be incorporated (utilizing 81 day averages, as in Figure 6, over the period 2010–2020, corresponding to Cycle 24). Figures 11(a)–(d) display several sets of these flux–flux correlations: low- and moderate-temperature features versus mid-chromosphere Mg ii 2800 Å (hk combined: Figure 11(a)); high-temperature species versus Mg ii (Figure 11(b)); mixed-temperature lines versus high-chromosphere C ii 1335 Å (Figure 11(c)); and high-temperature species versus low-TZ Si iii 1206 Å (Figure 11(d)). These flux–flux comparisons parallel those of Paper I but with the latest STIS flux calibrations; one new epoch of STIS spectra for AB; and, of course, the addition of an entirely new star, Procyon. (Note: a detailed discussion of the correlations is deferred until Section 4.)

Zoom In Zoom Out Reset image size

Figure 11(a) depicts several species, from the middle chromosphere up to the low-TZ, vs. the combined Mg ii hk flux. The α Cen stars and the Sun (where measurements are available) follow similar power-law trends, but the Procyon points cluster near the top of each panel, except for H i 1215 Å Lyα. (The large downward displacement of the stars from the solar Lyα trajectory is caused by ISM absorption.)

Figure 11(b) shows various high-temperature species vs. chromospheric Mg ii. Similar to the previous figure, the TZ lines of Procyon display elevated fluxes compared to the chromospheric anchor, Mg ii. Curiously, however, the enhancement apparently ends at coronal temperatures because Procyon's trends for Fe xii and soft X-rays fall closely in line with the other solar-stellar correlations.

Figure 11(c) compares various ionization stages to upper-chromosphere C ii, which, according to previous Figure 11(a), appears to share in the flux enhancement of the TZ-temperature lines. Now, Procyon fits in better with the extensions of the α Cen AB and solar power laws for the TZ species, although displays a clear deficit with respect to X-rays. Furthermore, the STIS stars show a range of behavior relative to Lyα, but all the slopes vs. C ii are close to unity. Lastly, the chlorine line, which is fluoresced by C ii (Shine 1983), displays some odd behaviors, which nevertheless are similar to those seen previously in Paper III (e.g., Figure 7(a) therein), where the F stars displayed enhanced Cl i emissions relative to the cooler dwarfs.

The final diagram of this series, Figure 11(d), depicts correlations between various high-temperature species vs. low-TZ Si iii. In Paper I, Si iii (and to some extent also Si iv) showed steeper than expected power laws compared with chromospheric and other TZ emissions. Part of the exaggerated response was speculated to be due to a differential enhancement of the TZ silicon abundance with increasing activity, which is somewhat analogous to the first-ionization potential (FIP) boost in low-FIP Fe-group coronal abundances (e.g., Laming et al. 1995). Although the FIP abundance anomalies were initially recognized in coronal species, the electro-dynamic chemical fractionation must have its roots in the lower atmosphere where high-FIP species like CNO are mostly neutral and low-FIP elements are mostly singly ionized. In Figure 11(d), Procyon's TZ lines are more closely aligned with the trends displayed by the α Cen stars and the Sun but, conversely, Procyon's coronal signatures show strong deficits.

Careful examination of these flux–flux correlations reveals a number of subtle differences among α Cen AB and the Sun, as well as the more blatant deviations by Procyon noted along the way. These systematic behaviors would be a ripe target for numerical modeling, such as stellar extensions of multi-D radiation-magnetohydrodynamic simulations currently in vogue for the Sun (e.g., Martínez-Sykora et al. 2017; Bjørgen et al. 2018).