THE ASTROPHYSICAL JOURNAL, 477:L57–L60, 1997 March 1
©1997. The American Astronomical Society. All rights reserved. Printed in U.S.A.

The Abundance of Interstellar Krypton 1

JASON A. CARDELLI 2

Department of Astronomy and Astrophysics, Villanova University, Villanova, PA 19085

AND

DAVID M. MEYER

Department of Physics and Astronomy, Northwestern University, Evanston, IL 60208; meyer@elvis.astro.nwu.edu

Received 1996 October 10; accepted 1996 December 18


ABSTRACT

     We present high signal-to-noise ratio HST Goddard High Resolution Spectrograph (GHRS) observations of the weak interstellar Kr I &lgr;1236 absorption toward the stars &tgr; CMa, &kgr; Ori, &egr; Ori, &lgr; Ori, &dgr; Sco, &ohgr;1 Sco, &egr; Per, and &zgr; Per. In combination with previous GHRS measurements of Kr I in two other sight lines (&zgr; Oph and 1 Sco), these new observations yield a mean interstellar gas-phase krypton abundance (per 109 H atoms) of 109 Kr/H = 0.96 ± 0.05. There is no statistically significant variation from sight line to sight line in the measured Kr I abundance and, in particular, no evidence for any correlation with the fraction of hydrogen in molecular form. Since Kr, as a noble gas, is not expected to deplete into dust grains, its gas-phase abundance should reflect the total interstellar abundance. Consequently, the GHRS observations imply that the interstellar Kr abundance in the vicinity of the Sun is about 60% of the solar system value of 109 Kr/H = 1.70 ± 0.30. This interstellar abundance deficit is similar to that recently found for oxygen with GHRS.

Subject headings: ISM: abundances—ISM: atoms

FOOTNOTES

     1 Based on observations obtained with the NASA/ESA Hubble Space Telescope through the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NASA-26555.

     2 Deceased 1996 May 14.

§1. INTRODUCTION

     From the chemical composition of Galactic interstellar dust (Savage & Sembach 1996) to the metallicity of QSO absorption-line systems (Pettini et al. 1995), solar system elemental abundances have served as the standard against which such measures have been compared and interpreted. However, recent analyses of the interstellar abundance of oxygen (Meyer et al. 1994; Cardelli et al. 1996), particularly in low-density sight lines, have suggested that solar abundances may not be generally appropriate as a “cosmic” reference standard. Specifically, the ISM oxygen measurements suggest that the solar system may be overabundant in O by as much as a factor of 1.5. Similar conclusions can also be inferred from recent ISM abundance studies of the Fe-peak element zinc (Roth & Blades 1995; Sembach et al. 1995). However, a major difficulty in arriving at definitive abundance determinations in these analyses is the fact that both elements are depleted from the gas phase into dust grains to some extent.

     Among the heavier-than-zinc elements recently detected in the ISM with the Goddard High Resolution Spectrograph (GHRS) on board the Hubble Space Telescope (HST) (Cardelli 1994), krypton presents a special opportunity to establish a cosmic reference abundance. As a noble gas, krypton does not form chemical bonds in its dominant neutral form (Kr I has an ionization potential of 13.999 eV). Furthermore, since the outer filled shell (4p6) is spherically symmetric, the van der Waals force is small (the neutral atom produces no induced dipole moment), and thus Kr is not prone to form mechanical bonds. Therefore, given the physical conditions prevalent in diffuse interstellar clouds, Kr should not be depleted from the gas phase into dust grains. In other words, the interstellar Kr I absorption-line strengths should directly reflect the cosmic Kr/H abundance ratio.

     In the two sight lines toward &zgr; Oph (Cardelli, Savage, & Ebbets 1991b) and 1 Sco (Hobbs et al. 1993), where Kr I has been detected with GHRS, the derived interstellar Kr abundance in both cases is about 60% of the Anders & Grevesse (1989) solar value (per 109 H atoms) of 109 Kr/H = 1.70 ± 0.30. The solar Kr abundance is somewhat uncertain because it is based on a combination of the 84Kr meteoritic abundance and measured solar wind isotope ratios from which the adopted value is obtained by fitting these data to adjacent elements on the periodic table assuming s- and r-process systematics (Trimble 1991). Despite this uncertainty, it is very interesting that the first two Kr measurements in the ISM suggest a solar system “overabundance” similar to that inferred for O. In this Letter we present new GHRS observations of interstellar Kr in a wide variety of sight lines chosen to test its applicability as such a cosmic abundance standard.

§2. OBSERVATIONS

     GHRS observations of the interstellar Kr I &lgr;1235.838 absorption toward &zgr; Per, &egr; Per, &lgr; Ori, &kgr; Ori, &dgr; Sco, and &ohgr;1 Sco were obtained in 1994 April, May, and September and 1995 March using the 19 km s-1 resolution of the G160M grating and the 0&farcs;25 small science aperture (SSA). In addition, the Kr I lines toward &egr; Ori and &tgr; CMa were observed with GHRS in 1995 October and November using the 3.5 km s-1 resolution ECH-A grating and the SSA. All of these observations consist of multiple FP-Split exposures centered near 1236 Å. Each FP-Split exposure is broken into four subexposures taken at slightly different grating positions for the purpose of minimizing the effects of the GHRS Digicon detector's fixed pattern noise on the signal-to-noise ratio (S/N) of the data.

     The data were reduced in a manner designed to maximize S/N, as described in detail by Cardelli & Ebbets (1994). Briefly, this approach involved (1) merging the subexposures in each FP-Split exposure in diode space in order to create a fixed pattern noise template spectrum, (2) dividing each subexposure by this noise template, (3) aligning all of the rectified subexposures in each FP-Split in wavelength space using the interstellar lines as a guide, and (4) summing all of the rectified subexposures and FP-Splits. As illustrated for four of the stars in Figure 1, the resulting continuum-flattened spectra reveal convincing detections of the interstellar Kr I &lgr;1236 line in all eight sight lines. The measured S/N ratios of the 19 km s-1 resolution spectra are all in the 400–800 range, and those of the echelle data are about 200. In order to facilitate comparison, the spectra in Figure 1 have been velocity-shifted to align the interstellar Kr I features at the 1235.838 Å rest wavelength and are displayed in order of the sight lines' total hydrogen column density.

Fig. 1

     As listed in Table 1, the measured Kr I &lgr;1236 equivalent widths are no more than a few milliangstroms in all of our GHRS spectra. Consequently, the Kr I column densities can be derived under the assumption that the lines are optically thin. The column densities listed in Table 1 were calculated using an adopted f-value of 0.20 for the &lgr;1236 transition. This oscillator strength should be accurate to better than 10% and is based on a number of experimental determinations that range between 0.187 ± 0.006 (Griffen & Hutchenson 1969) and 0.214 ± 0.011 (Chan et al. 1992). Since Kr I (13.999 eV) and H I (13.598 eV) have similar ionization potentials, Kr I is the dominant ion of Kr in H I regions, and little Kr I should originate from H II regions. Consequently, the ratio of N(Kr I) to the total H column density [N(H) = 2N(H2) + N(H I)] should accurately reflect the interstellar gas-phase Kr/H abundance ratio. In Table 1 we have calculated N(H) toward each of our eight program stars by averaging the N(H I) measurements of Bohlin, Savage, & Drake (1978) and Diplas & Savage (1994) and adopting the Bohlin et al. N(H2) values. The resulting N(Kr)/N(H) ratios are remarkably consistent among the eight sight lines as well as with the earlier &zgr; Oph (Cardelli et al. 1991b) and 1 Sco (Hobbs et al. 1993) results. Taken together, these measurements yield a weighted mean gas-phase interstellar Kr abundance of 109 Kr/H = 0.96 ± 0.05.

§3. DISCUSSION

     The top panel of Figure 2 illustrates the interstellar krypton abundances as a function of the fraction of hydrogen in molecular form, f(H2), toward our observed sight lines (squares) as well as &zgr; Oph and 1 Sco (triangles). The f(H2) parameter serves as a measure of the physical conditions in the gas by representing the balance between H2 grain-surface formation and gas-phase chemical or photo-destruction (Cardelli 1994; Cardelli et al. 1996). As such, any increase in the observed gas-phase abundances with decreasing f(H2) would signal changes in the elemental dust abundance due to grain destruction. Such variations are observed even for weakly depleted elements like Ge (Cardelli 1994) and Zn (Roth & Blades 1995; Sembach et al. 1995). Figure 2 shows that the interstellar gas-phase Kr abundances measured with GHRS show no dependence on f(H2). Indeed, the mean abundance in the three sight lines with log f(H2) < -2.0 (109 Kr/H = 0.96 ± 0.10) is essentially identical to that in the seven sight lines with log f(H2) > -2.0 (109 Kr/H = 0.96 ± 0.06). The absence of Kr gas-phase abundance variations over such a large range of f(H2) is consistent with the expectation that Kr is not depleted into dust grains. Consequently, our GHRS observations imply that the total Kr abundance is quite uniform throughout the local interstellar medium at a level that is about 60% of the solar system value of 109 Kr/H = 1.70 ± 0.30.

Fig. 2

     The possibility that Kr is tracing an interstellar elemental abundance deficit relative to the Sun is fortified by recent GHRS results on the abundance of interstellar oxygen. In the bottom panel of Figure 2, we have plotted the O abundances measured by Cardelli et al. (1991a), Savage, Cardelli, & Sofia (1992), Meyer et al. (1994), and Cardelli et al. (1996) toward seven stars as a function of f(H2) in these sight lines (four of which also have measured Kr abundances). The mean interstellar gas-phase O abundance toward these stars is 106 O/H = 308 ± 18. The uniformity of the individual values as a function of f(H2) implies that there is negligible oxygen exchange between gas and dust in these diffuse sight lines and that the total (gas plus dust) abundance of O must not vary significantly in the local ISM. Assuming various dust mixtures of oxides and silicates such as olivine, the abundance of interstellar oxygen tied up in dust grains is unlikely to surpass 106 O/H &ap; 180 (Cardelli et al. 1996). Any further increase of this O fraction in grain cores is difficult because the requisite metals are far less abundant than oxygen. Therefore, the GHRS O observations indicate that the total (gas plus dust) abundance of interstellar oxygen in the vicinity of the Sun is about two-thirds of the solar value of 106 O/H = 741 ± 130 (Grevesse & Noels 1993).

     If Kr is indeed undepleted into interstellar grains and its solar abundance is accurate, it appears that our part of the Galaxy has chemically evolved to the point where both Kr and O have similar subsolar abundances. It is important to note that this interstellar deficit is also reflected in Galactic B star measurements of the current epoch abundances of some elements (Kilian 1992; Gies & Lambert 1992; Kilian, Montenbruck, & Nissen 1994; Cunha & Lambert 1994). These studies yield median B-star CNO abundances that are about two-thirds of the solar values. Since the elemental makeup of the solar system presumably reflects that of the ISM at the time of the Sun's formation 4.6 Gyr ago, these results are difficult to understand in the context of standard Galactic chemical evolution models where abundance ratios such as O/H and Kr/H should increase with time (Audouze & Tinsley 1976; Timmes, Woosley, & Weaver 1995). It is also interesting that elements with nucleosynthetic histories as different as O and Kr should exhibit similar current epoch abundance deficits relative to solar. The primary source of O in the ISM is helium burning in massive stars that explode as Type II supernovae (Wheeler, Sneden, & Truran 1989). Krypton is mainly produced by the “weak” s-process associated with hydrostatic burning in such stars along with a secondary contribution from low-mass stars on the asymptotic giant branch (Raiteri et al. 1993).

     One possible explanation for the apparent overabundance of Kr and O in the solar system is that the protosolar nebula was enriched by the explosion of a nearby supernova (Reeves 1978; Olive & Schramm 1982). As discussed by Cunha & Lambert (1994), such a process may not be uncommon in large molecular clouds where a first generation of massive stars can evolve quickly and enrich the gas with newly synthesized elements that are incorporated in later generations of stars formed in the clouds. Indeed, an analysis of Fe in stars formed around the same time as the Sun show a spread in abundance of 0.5 dex with the Sun near the top of the distribution (Edvardsson et al. 1993). Yet we see no evidence of such large abundance inhomogeneities in the local ISM today. A glance at Figure 2 shows that the abundance of interstellar Kr is remarkably constant in a variety of sight lines out to distances of 500 pc. A key challenge for the enrichment hypothesis is how to make stars with large abundance variations arise from such a well-mixed ISM. Another challenge is designing supernovae that could have yielded similar overabundances of oxygen, krypton, and perhaps other elements with different nucleosynthetic histories such as zinc in the early solar system.

     An alternative view of the solar abundance question is that the local ISM has become underabundant in elements such as Kr and O over the past 4.6 Gyr. Meyer et al. (1994) have suggested that the interstellar O deficit could be the result of a recent infall of metal-poor gas, perhaps from the Magellanic Stream, into the solar neighborhood with a consequent reduction in O/H. The infall of low-metallicity extragalactic material has long been recognized as a potential factor in the chemical evolution of the Galaxy (Audouze & Tinsley 1976; Roy & Kunth 1995). Comeron & Torra (1994) have proposed a model to explain the dynamics of the Gould Belt that involves the impact of a million-solar-mass extragalactic cloud with the local Milky Way about 108 years ago. A specific prediction of this model would be that the mixture of the infalling metal-poor gas with the local ISM would lower the heavy-element abundances below their solar values by a similar amount. In other words, such a model could explain the similar O and Kr underabundances in the local ISM. At the same time, the discovery of an element whose interstellar abundance does not fit this pattern would present a serious problem for an infall “dilution” model.

     In addition to issues concerning Galactic chemical evolution, the similar underabundances of interstellar Kr and O have implications concerning the composition of interstellar dust. The traditional interpretation of interstellar gas-phase abundances is that any difference with the solar abundances reflects the amount of a particular element tied up in grains (Savage & Sembach 1996). In making a strong case that the “cosmic” abundances appropriate for the local ISM are actually more like two-thirds solar, the GHRS observations of Kr and O also imply a reduction in the solid-state abundances of other elements. The implications for carbon are particularly interesting. As discussed by Cardelli et al. (1996), the revised abundance of solid carbon can still explain extinction features such as the 2175 Å bump through graphite and/or PAHs, but it severely restricts the availability of C in grains to explain the total optical/UV dust opacity.

     Finally, apart from any conclusions based on comparisons with the solar Kr abundance, it is important to recognize that the uniformity of the interstellar Kr abundance in a variety of sight lines is consistent with the expectation that it can serve as a cosmic reference standard for the current epoch of the local ISM. As a probe of galactic chemical evolution, it would certainly be interesting, but difficult, to compare this standard with measurements of the weak Kr I &lgr;1236 line arising from distant parts of the Galaxy or in QSO absorption-line systems. A more realistic application of a Kr standard in the short term might be as a substitute for hydrogen as a reference abundance in studies of interstellar elemental depletion variations in dense cloud sight lines where the total hydrogen column density is hard to come by. It will be possible to pursue such studies efficiently with the Space Telescope Imaging Spectrograph on board HST.

ACKNOWLEDGMENTS

     It is a pleasure to thank U. J. Sofia for assistance with the data and helpful discussions. This work was supported by STScI through grants to Villanova and Northwestern Universities.

REFERENCES

FIGURES


Full image (49kb) | Discussion in text
     FIG. 1.—HST GHRS G160M spectra of the interstellar Kr I &lgr;1235.838 absorption line toward &kgr; Ori, &lgr; Ori, &dgr; Sco, and &ohgr;1 Sco taken at a velocity resolution of 19 km s-1. The spectra have been velocity shifted to align the interstellar features at the Kr I rest wavelength and are displayed from top to bottom in order of increasing total hydrogen column density in the observed sight lines. The measured S/N values of these spectra are all in the 400–800 range. The measured equivalent widths of the Kr I lines are listed in Table 1.

Full image (44kb) | Discussion in text
     FIG. 2.—Interstellar krypton and oxygen abundances measured with GHRS as a function of the logarithmic fraction of hydrogen in molecular form, f(H2) = 2N(H2)/[2N(H2) + N(H I)], in the observed sight lines. The Kr abundances marked with squares are presented in this paper (Table 1), and those marked with triangles are updated values for &zgr; Oph and 1 Sco taken from Cardelli (1994) and Hobbs et al. (1993). The O abundances are from Cardelli et al. (1991a), Savage, Cardelli, & Sofia (1992), Meyer et al. (1994), and Cardelli et al. (1996). The dashed lines among the data points represent the weighted mean values of 109 Kr/H = 0.96 ± 0.05 and 106 O/H = 308 ± 18 for the interstellar gas-phase krypton and oxygen abundances. This Kr abundance is about 60% of the Anders & Grevesse (1989) solar value (109 Kr/H = 1.70 ± 0.30) denoted by the topmost line in the figure. The dashed line near 106 O/H &ap; 490 represents the total (gas plus dust) interstellar oxygen abundance after an allowance is made for the O tied up in dust grains. This value is about two-thirds of the Grevesse & Noels (1993) solar oxygen abundance (106 O/H = 741 ± 130).

TABLES

TABLE 1
INTERSTELLAR KRYPTON ABUNDANCES
Star log f(H2) a log N(H) b
(cm-2)		 W&lgr;(1236) c
(mÅ)		 log N(Kr) d
(cm-2)		 109 Kr/H e
&tgr; CMa... -4.97 20.75 1.55 ± 0.20 11.76 1.02 ± 0.15
&kgr; Ori... -4.58 20.56 0.90 ± 0.15 11.52 0.92 ± 0.18
&egr; Ori... -3.59 20.46 0.70 ± 0.15 11.41 0.90 ± 0.23
&lgr; Ori... -1.39 20.81 1.75 ± 0.20 11.81 1.00 ± 0.22
&dgr; Sco... -1.38 21.10 3.65 ± 0.25 12.13 1.07 ± 0.14
&ohgr;1 Sco... -0.89 21.24 4.75 ± 0.35 12.24 1.01 ± 0.18
&egr; Per... -0.70 20.52 0.80 ± 0.15 11.47 0.89 ± 0.21
&zgr; Per... -0.23 21.20 4.05 ± 0.30 12.18 0.95 ± 0.11

     a f(H2) = 2N(H2)/[2N(H2) + N(H I)] is the fractional H2 abundance.
     b N(H) = 2N(H2) + N(H I) is the total column density of H nuclei. These values reflect the mean of the data from Bohlin, Savage, & Drake 1978 and Diplas & Savage 1994.
     c The measured equivalent widths and 1 σ uncertainties for the Kr I 1235.838 Å absorption line.
     d These Kr column densities were calculated using an f-value of 0.20 for the Kr I &lgr;1236 line (Chan et al. 1992; Griffen & Hutchenson 1969).
     e Abundance of interstellar gas-phase Kr per 109 H atoms. The 1 σ uncertainties reflect the propagated N(Kr) and N(H) errors. The solar abundance of krypton is 109 Kr/H = 1.70 ± 0.30 (Anders & Grevesse 1989).

Image of typeset table | Discussion in text