Detection of Pb II in the Ultraviolet Spectra of Three Metal-poor Stars^*

There are four stable isotopes of Pb: ²⁰⁴Pb (1.4% in solar system material), ²⁰⁶Pb (24.1%), ²⁰⁷Pb (22.1%), and ²⁰⁸Pb (52.4%). Of these isotopes, ²⁰⁷Pb has nonzero nuclear spin, I = 1/2, and thus has hyperfine splitting (HFS) structure. The field shift, which results from the volume difference between nuclei with the same number of protons but different numbers of neutrons, also creates isotope shifts (ISs). We adopt the ground level and excited level HFS A values and the IS measurements from Bouazza et al. (1986) for the Pb ii line at 2203 Å. We adopt the atomic transition probability from Quinet et al. (2007), $\mathrm{log}({gf})$ = −0.24 ± 0.07, because it is normalized to a radiative lifetime measurement from laser induced fluorescence. For comparison, the National Institutes of Standards and Technology (NIST) Atomic Spectral Database (ASD; Kramida et al. 2019) recommends $\mathrm{log}({gf})$ = −0.14, based on theoretical calculations referenced there, which is in fair agreement with our adopted value. We present the line component pattern in Table 2.

We adopt the model parameters (effective temperature, ${T}_{\mathrm{eff}}$; log of the surface gravity, log g; microturbulent velocity, ${v}_{{\rm{t}}}$; model metallicity, [M/H]) derived previously for these three stars, for consistency. HD 94028 is a main-sequence dwarf, while HD 196944 and HD 222925 are red horizontal branch stars. We interpolate model atmospheres from the 1D, α-enhanced ATLAS9 grid of models (Castelli & Kurucz 2004). Our synthetic spectra also include a macroturbulent velocity (V_t) component, which improves the fits to the high-resolution E230H spectra. We derive V_t by fitting the observed profiles of isolated lines of Fe-group elements. These values are listed in Table 1.

Figure 1 illustrates the Pb ii line in the spectrum of each of the three stars. Continuum regions around this line are easily identified, and they are matched by the synthetic spectra. We are confident that the absorption at 2203.534 Å is due to Pb ii for several reasons. The line strength varies with the expected heavy-element abundances in these stars, not the abundances of iron-group elements that are responsible for most UV absorption lines. This line is—by many orders of magnitude—the strongest Pb ii line with λ > 2000 Å, so there is no expectation that other Pb ii lines could be detectable. Furthermore, no other plausible species are found at this wavelength in the NIST ASD or the Kurucz (2011) line lists. Unidentified lines at 2203.427 and 2203.645 Å could be explained by Co ii and V i transitions, respectively, only if the $\mathrm{log}({gf})$ values recommended by the NIST ASD or the Kurucz (2011) line lists are underestimated by several dex. We treat their strengths as free parameters in our analysis, and this choice does not influence the derived Pb abundances.

Zoom In Zoom Out Reset image size

The Pb ii line is on the linear part of the curve-of-growth in HD 94028 and HD 222925, but it is saturated in HD 196944. We derive abundances using the 2017 version of LTE line analysis software MOOG (Sneden 1973; Sobeck et al. 2011). We adopt an s-process mix of Pb isotopes (Sneden et al. 2008) for HD 94028 and HD 196944 and an r-process mix for HD 222925. We generate the line list based on Kurucz (2011), Peterson et al. (2017), and the NIST ASD. We match the synthetic spectra to the observed spectra following the general methods described by Roederer et al. (2012).

Table 1 lists the derived abundances. The Pb abundance is defined as $\mathrm{log}\varepsilon$(Pb) $\equiv \,{\mathrm{log}}_{10}({N}_{\mathrm{Pb}}/{N}_{{\rm{H}}})+12.0$. The abundance ratio of elements Pb and Fe relative to the Solar ratio is defined as [Pb/Fe] $\equiv \,{\mathrm{log}}_{10}({N}_{\mathrm{Pb}}/{N}_{\mathrm{Fe}})-{\mathrm{log}}_{10}{\left({N}_{\mathrm{Pb}}/{N}_{\mathrm{Fe}}\right)}_{\odot }$, where $\mathrm{log}\varepsilon$(Pb)_⊙ = 2.04 and $\mathrm{log}\varepsilon$(Fe)_⊙ = 7.50. Following Roederer et al. (2018a), we compute 1σ uncertainties by drawing 10³ resamples of the stellar parameters, $\mathrm{log}({gf})$ values, and equivalent widths approximated from the abundance derived via synthesis using a reverse curve-of-growth method.

The Pb isotope mix has not been assessed previously in any metal-poor star. The HFS of the ²⁰⁷Pb isotope, particularly the upper level of the line at 2203 Å, and the IS of the four Pb isotopes are wide compared to the width of the stellar line profiles shown in Figure 2. As the isotope mix shifts from the r-process, where the ²⁰⁶Pb isotope (35.9%) and wide HFS of the ²⁰⁷Pb isotope (45.1%) are expected to dominate, to an s-process mix, where the ²⁰⁸Pb isotope is expected to dominate (69.5%), the absorption line profile narrows and shifts to shorter wavelengths.

Zoom In Zoom Out Reset image size

The observed Pb ii line profiles in HD 94028 and HD 196944 both favor a narrower profile, suggesting that an even-A isotope dominates. We evaluate which isotope it might be by comparing the line centroid to the isotope wavelengths. We set the local wavelength zero-point of the observed spectrum relative to the synthetic spectra using three relatively unblended Fe ii lines, with wavelengths known to better than 0.0011 Å (Nave & Johansson 2013), located in the same echelle order that contributes most of the signal to the Pb ii line. We also account for uncertainty in the center-of-gravity wavelength of the Pb ii line (0.0007 Å; Wood et al. 1974) and measurement uncertainties in the Fe ii and Pb ii line centroids. As shown in Figure 2, the centroid of the observed Pb ii line in HD 94028 and HD 196944 favors absorption by the isotopes situated farthest to the blue, ²⁰⁷Pb and ²⁰⁸Pb. Our method of setting the wavelength zero-point using Fe ii lines appears to have produced a small mismatch between the observed and synthetic line profiles in HD 94028 and HD 222925. Manual adjustment (≤1σ) of the observed spectrum in each case so that it matches the blue side of the line, which is relatively insensitive to the Pb isotope mix, still favors the ²⁰⁷Pb and ²⁰⁸Pb isotopes.

We conclude from these two tests—the narrow line profiles and the positions of the line centroids—that the ²⁰⁸Pb isotope is dominant in HD 94028 and HD 196944. The S/N is too low to draw any conclusions about HD 222925.

The abundances derived from the Pb ii line are higher than the abundances or upper limits derived from the Pb i lines in all three stars: [Pb ii/Pb i] = +0.36 ± 0.34, +0.49 ± 0.28, and >+0.04 in HD 94028, HD 196944, and HD 222925, respectively. Mashonkina et al. (2012) computed non-LTE corrections to the LTE abundances for several metal-poor atmospheres. That study found that the lower levels of the Pb i lines at 2833 and 4057 Å experience similar deviations from LTE. The ground state of singly ionized Pb is well described by LTE in their calculations. We assume that the excited 6s²6p level that gives rise to the Pb ii line at 2203 Å is also well described by LTE. The Mashonkina et al. non-LTE corrections for red giants (dwarfs) range from +0.26 to +0.62 dex (+0.22 to +0.32 dex) when using a Drawin scaling factor, S_H, of 0.1, which relates to the strength of inelastic collisions with neutral hydrogen. No models in their grid exactly match the stars in our sample, but the closest models predict non-LTE corrections ≈+0.27 and +0.52 dex for HD 94028 and HD 196944, respectively, which match the offsets we derive in LTE. We support the conclusion of Mashonkina et al. that departures from LTE impact Pb abundances derived from Pb i lines.

Pb is often used to constrain the s-process or i-process models used to explain nucleosynthesis patterns in stars (e.g., Hampel et al. 2019). A change in [Pb/Fe] by +0.4 dex is significant and could affect the final neutron exposure inferred from models, which sets, for example, the estimated timescale for an i-process event. Future work should incorporate non-LTE corrections to abundances derived from Pb i lines or derive Pb abundances in LTE directly from the UV Pb ii line.

Our LTE results confirm the enhanced Pb abundances in the s-process-enhanced stars HD 94028 and HD 196944. We derive [Pb/Ba] = +1.05 ± 0.35 in HD 196944. This is in good agreement with the AGB s-process model of Bisterzo et al. (2010) discussed at length in Placco et al. (2015; see also Abate et al. 2015). We derive [Pb/Ba] = +0.42 ± 0.17 in HD 94028, which supports the interpretation of Roederer et al. (2016) that the Ba and Pb in HD 94028 originated mainly via the s-process. Our isotopic analysis reaffirms theoretical predictions that the large Pb overabundances in low-metallicity s-process environments are dominated by ²⁰⁸Pb.

The $\mathrm{log}\varepsilon$(Pb/Eu) ratio in HD 222925, 0.76 ± 0.14, matches the solar system r-process ratio, $\mathrm{log}\varepsilon$(Pb/Eu) = 0.76 ± 0.10 (Sneden et al. 2008; Bisterzo et al. 2014; Prantzos et al. 2020). This result indicates that the r-process residuals for the Pb isotopes are, in aggregate, correct when the effects of low-metallicity AGB stars (i.e., the so-called "strong component") are included. Clayton & Rassbach (1967) argued that the dominant r-process isotopes of Pb must be ²⁰⁶Pb and ²⁰⁷Pb. The Pb ii line centroid in HD 222925 is not in conflict with this reasoning, although the S/N is too low in our spectrum to support a more definitive statement.

The close coupling between Th and Pb enables the use of Th/Pb as a chronometer pair that is relatively insensitive to the details of the r-process model used to calculate the initial production ratio. HD 222925 does not exhibit a prominent actinide boost (Roederer et al. 2018a), and its $\mathrm{log}\varepsilon$(Th/Pb) ratio is −1.20 ± 0.14. The $\mathrm{log}\varepsilon$(Th/Pb) ratio in HD 222925 corresponds to an age of 8.2 ± 5.8 Gyr using the production ratios from Roederer et al. (2009). The permitted age range is large, but improving the S/N at the Pb ii line in future observations would improve the age precision. Pb lines are easier to detect in metal-poor stars than the U ii line at 3859 Å, and the Th/Pb chronometer offers an alternative model-insensitive age indicator to the U/Th chronometer in r-process-enhanced stars.