R. Ferlet, M. Lemoine and A. Vidal-Madjar
Institut d'Astrophysique de Paris, CNRS, 98bis Bd. Arago, 75014 Paris,
France
Present address: Dept. of Astronomy & Astrophysics,
University of Chicago, 5640 South Ellis Ave., Chicago IL 60637
Keywords: Deuterium, Interstellar Abundance
It is now well accepted that deuterium and some other light elements are only produced in significant amount during primordial nucleosynthesis (BBN, see e.g., the proceedings of the ESO Symposium The Light Element Abundances, Crane Ed., Springer). On the contrary, D is destroyed all along the galactic evolution when processed in stellar interiors. Hence, deuterium appears as a key element for cosmology and subsequent evolution.
Notwithstanding the meteoritic/solar system abundance of deuterium, the first measurements of the interstellar D/H ratio were reported in the early seventies, and their number have then increased following the availability of the adequate instrumentation. For more than a decade, these interstellar abundances, which are representative of the present epoch, have been used to constrain BBN in a direct way. However, abundances measured at lower metallicities are less contaminated by the effect of galactic evolution. Underlying this simple statement is another trivial statement: interstellar abundances cannot be used, in any case, to constrain only BBN parameters, since the chemical evolution of the Galaxy always appears as a conditional assumption to these constraints. Simple as it is, it is our opinion that this has been overlooked in the past decades, and that this is still being overlooked.
Hence, to provide a bottom line to this review, the main significance of interstellar abundances is to constrain both BBN and chemical evolution models. Whenever other abundances are measured at a lower metallicity, they generally supersede the interstellar abundance in its significance to BBN. Conversely, the interstellar abundance is then used to constrain chemical evolution through the comparison to this latter abundance. In the following, we discuss the present status of the interstellar deuterium abundance and show that its ``true'' value may not be well-known.
There are several methods to measure the interstellar abundance of
deuterium (see Vidal-Madjar 1991, Ferlet 1992). One of them is to observe
deuterated molecules such as HD, DCN, etc... and to form the ratio of
the deuterated molecule column density to its non-deuterated
counterpart (H
, HCN, etc...). More than 20 different deuterated
species have been identified in the ISM, with abundances relative to the
non-deuterated counterpart ranging from 
to 10
. Conversely,
this means that fractionation effects are important and that, as a consequence,
this method cannot provide a precise estimate of the true interstellar D/H
ratio; rather, this method is used in conjunction with estimates of the
interstellar D/H ratio to gather information on the chemistry of the ISM.
Another way to derive the D/H ratio comes through radio observations of the
hyperfine line of DI at 92cm. The detection of this line is, however,
extremely difficult, and no firm detection has ever been reported. The
detection of this line would allow to probe more distant interstellar media
than the local medium discussed below; however, because a large column density
of D is necessary to provide even a weak spin-flip transition, these
observations aim at molecular complexes. As a result, the upper limit derived
toward Cas A (Heiles et al. 1993): D/H
may as well
result from a large fraction of D and H being in molecular form in these
clouds, as from the fact that one expects the D/H ratio to be lower nearer to
the galactic center (since D is destroyed in stellar processing).
Finally, the only way to derive a reliable estimate of the interstellar D/H ratio is to observe the atomic transitions of D and H of the Lyman series in the far-UV, in absorption in the local ISM against the background continuum of cool or hot stars. These observations have been performed using the Copernicus and the IUE satellites, and now the Hubble Space Telescope. Both types of target stars present pros and cons.
The main advantage of observing cool stars is that they can be selected in the
vicinity of the Sun. This results in low HI column densities, and trivial
to nearly trivial lines of sight. In effect, due to the low atomic weight of
HI and DI, to the DI-HI -82
isotopic shift, and
to the abundance of HI in the local medium, the DI line cannot be
detected at Lyman 
in the wing of the HI line for HI
column densities larger than 
cm
. Also, the presence of several
interstellar components with different b-values may imply a large error on
the HI column density if these components are unresolved. For this
reason, deriving the HI column density has always been the limiting
factor of accurate D/H ratios measurements. Note that the spectral resolutions
of Copernicus and IUE were, respectively, 15 and 30
and, as a consequence,
a non-trivial line of sight, even in the local ISM, would generally go
unresolved. Even though HST-GHRS now offers a spectral resolution of 3.5
,
the thermal width of the DI line in the local ISM is 
, so
that one has to observe lines of heavier species (thinner lines) to fully use
the resolving power of HST.
However, as a result of the HI chromospheric emission profile at Lyman
, the interstellar absorption of NI 1200Å (triplet) is not
available; yet, NI was shown to be an excellent tracer of HI in
the ISM (Ferlet 1981). Hence, one usually observes the strong lines of
MgII, FeII or similar elements, although these elements probe
mainly HII media and not HI media. Moreover, the chromospheric
emission line has to be modeled to set the continuum for the interstellar
absorption. Such a procedure necessarily introduces systematic errors.
Nevertheless, this method has provided the most precise measurement of the
local D/H ratio in the direction of Capella, using HST--GHRS:
(D/H)
(Linsky et al. 1993, 1995), assuming no systematics.
Hot stars are unfortunately located further away from the Sun, so that one
always has to face a high HI column density and often a non-trivial
line of sight structure. In these cases, DI could not be detected at
Ly
, and one has to observe higher order lines, e.g., Ly
,
Ly
, Ly
, hence these measurements have primarily come through
Copernicus observations. The stellar continuum is however smooth at the
location of the interstellar absorption, and, moreover, the NI triplet
as well as other NI lines are available to probe the velocity structure
of the line of sight.
All published D/H ratios are collected in Fig. 1, distinguishing hot
stars from cool stars observations (all references except Allen et al. 1992
can be found in Vidal-Madjar 1991, Ferlet 1992). The D/H ratios range from
to 
. A large scatter is clearly
detected in Fig. 1, and represents differences of the D/H ratio in the local
ISM, that may be as large as a factor 
over scales as small as a
few parsecs. The essential question is: do these variations really exist?
Figure: Measurements of the D/H ratio in the local ISM. The left hand-side
box collects data obtained toward hot stars, while the right hand-side one
shows cool stars observations. The x-axis has no physical significance, and
merely labels the different stars. Data points next to each other, within
less than 1 x-axis unit, correspond to the same target star. The open circle
represents the Linsky et al. (1995) measurement using HST toward Capella. All
other data points come from Copernicus or IUE observations.
Unfortunately, one cannot answer this question and at the same time be
perfectly objective. On one hand, note that the Linsky et al. (1995)
measurement does not agree with any of the previous D/H measurements toward
Capella. However, one cannot ignore this measurement of unprecedented quality.
On the other hand, one could use the observed scatter between different
measurements toward a same star to get an estimate of the systematics;
although this estimate is rough, it does not seem to be able to account for
the large scatter of Fig. 1. Finally, it could be tempting---but
also very arbitrary---to claim that the systematics associated with IUE and
Copernicus observations are large, that they account for the observed
discrepancies, and that only the Linsky et al. (1995) value should be kept.
Although undoubtedly of great quality, this measurement toward Capella is
however coming through the modelling of the chromospheric emission line of an
unresolved binary system and the detection of only one absorbing component on
the line of sight, the local cloud in which the Sun is imbedded. In fact, this
apparently trivial velocity structure was obtained through MgII and
FeII observations; hence, it could be that HI media, shifted from
the local cloud by a few 
went unnoticed in MgII and FeII
although they would play an important role in the HI saturated profile.
Such systematics were not considered by Linsky et al. (1993,1995).
To answer to the reliability of the observations shown in Fig. 1, one
has to re-analyze all these data in a consistent way, looking for possible
undetected systematics. One has to recall, for example, that time variations of
the D/H ratio have already been reported toward 
Per
(Gry et al. 1983), which were interpreted as due to the ejection of high
velocity hydrogen atoms from the star. Another (complementary) way would be to
proceed with further observations of the local D/H ratio using HST as well as
the Lyman-FUSE mission which should be launched in 1998. Finally, note that
various explanations to these possible fluctuations of the D/H ratio have been
put forward as early as Vidal-Madjar et al. (1978), Bruston et al. (1981):
this is a long-standing problem.
We have inaugurated a new promising observational strategy, which is to
observe nearby white dwarfs. Not only such targets may be selected near to
the Sun, circumventing the main disadvantage of hot stars, but they can also
be chosen in the high temperature range, so as to provide a smooth stellar
profile at Ly
. At the same time, the NI triplet at 1200Å would
be available, allowing thus an accurate sampling of the line of sight. Such
observations have now been conducted using HST toward two white dwarfs: Hz43
(see Landsman et al. these proceedings), and G191--B2B (Lemoine et al. 1995,
Vidal-Madjar et al. 1996).
In the case of Hz43, the structure of the line of sight seems, at a first glance, to be trivial, i.e., consisting of only the local cloud. The D/H ratio, as well as the HI column density, are consistent, in the single-cloud hypothesis, with those of the local cloud, obtained by Linsky et al. (1995). However, due to the relative faintness of this target, the NI triplet was observed at medium resolution only, and other interstellar components cannot be ruled out as of now. Obviously, this target looks very promising, providing first observations be complemented with higher resolution higher signal-to-noise ratio data.
In the case of G191-B2B, data were obtained in Cycles 1 and 5 at high
resolution (
90000) for Ly
(Fig. 2), NI 1200Å,
OI 1302Å, MgII and FeII, and at medium resolution
(
20000) for other species like CII, SiII and
SiIII. The line of sight velocity structure coherent in all these lines
(about 15) comprises two HI regions---one being the local cloud
observed toward Capella---together with one HII region, clearly seen
in SiIII but also detectable in very strong lines such as those of
HI Ly
and OI. The analysis in terms of column densities
is underway, but it seems already clear that the two HI components are
in different ratios according to the studied ions. In particular, if the D/H
ratio for the component common to the G191--B2B and Capella sight-lines is
forced to be that found by Linsky et al. (1995), then the D/H ratio for the
other HI component appears significantly lower.
Figure: The Lyman 
interstellar line toward G191--B2B recorded with
GHRS--HST during Cycle 5. The Geocoronal Ly
emission centered in the
core of the HI saturated absorption has been released. The HI
(dotted line) and DI (dashed line) Voigt profiles after convolution with
the instrumental function are also shown absorbing against the Lorentzian
stellar profile.
We have, therefore, put forward a contamination of the HI interstellar absorption toward G191--B2B by residuals atoms of neutral hydrogen from an HII region. It is not unlikely that this effect is also present in other lines of sight, in particular toward Capella. Such systematics were not considered by Linsky et al. (1993, 1995). The observation of the whole Lyman series, possible with FUSE, will allow to easily detect this effect. New HST observations toward other white dwarfs have been obtained.
If the variations of the D/H ratio in the local interstellar medium are
illusory, then one could quote as an average of the published values:
(D/H)
. The rather large error bar arises
from a subjective although conservative viewpoint. On the contrary, if the D/H
does vary in the ISM, one has to understand why; until then, no
measurement of the D/H ratio in the ISM or the IGM should be quoted as
reliable. Moreover, one should expect these variations to be larger in
reality than what is observed. The actual value might in fact be very
different from what is observed, if these variations are systematic,
i.e., act in one way only; this in turn would heavily bear on the
chemical evolution of deuterium. It also appears that the upper bound on
, the baryonic density parameter of the Universe, is obtained from
BBN predictions through the interstellar abundance of deuterium: this bound
would have to be removed until the variations and their cause are properly
understood.
There is all hope that the FUSE mission will solve these problems. It will
probe the ISM further than the local medium, up to extragalactic low-redshift
objects. It will look for gradients of the deuterium abundance with
galactocentric distance and with galactic height in the halo. These dedicated
studies with FUSE should greatly clarify the problem of the chemical evolution
of deuterium. In particular, the present standard models are unable to
reconcile the interstellar and solar D abundances with its very unexpected,
very high primordial one as recently measured in very metal poor absorbing
clouds toward quasars and as deduced from observations of primordial 
He.
The study of the interstellar D/H ratio toward white dwarves with HST has been initiated by two of the authors (AVM and RF). Subsequently, several French and US collaborators were included.
Allen, M., Jenkins, E.B., & Snow T.P. 1992, ApJS, 83, 261
Bruston, P., Audouze, J., Vidal-Madjar, A., & Laurent, C. 1981, ApJ, 243, 161
Ferlet, R. 1981, A&A, 98, L1
Ferlet, R. 1992, IAU Symposium 150, 85
Gry, C., Laurent, C., & Vidal-Madjar, A. 1983, A&A, 124, 99
Heiles, C., McCullough, P., & Glassgold, A. 1993, ApJS, 89, 271
Lemoine, M., Vidal-Madjar, A., Bertin, P., Ferlet, R., Gry, C., & Lallement, R. 1995, A&A, in press
Linsky, J. et al. 1993, ApJ, 402, 694
Linsky, J. et al. 1995, ApJ, 451, 335
Vidal-Madjar, A., Laurent, C., Bruston, P., & Audouze, J. 1978, ApJ, 223, 589
Vidal-Madjar, A. 1991, Adv. Space Res., 11, 97
Vidal-Madjar, A., Lemoine, M., Ferlet, R., & Gry, C. 1996, in preparation