6. Comparisons with observations and inferences on the LIC electron density
One of the main results of the present model is the very small rejection of neutral oxygen outside the heliosphere. As a matter of fact, for the relatively high plasma density used for the computation, about 70% of the neutral O succeeds in penetrating closer than 50 A.U. from the Sun on the upwind side. For a given number density ratio beetwen the heliosphere and the ISM, such an easy entrance allows for higher interstellar plasma densities.
At present the most accurate comparison between interstellar and heliospheric oxygen makes use of Hubble Space Telescope (HST) nearby stars observations on the one hand, and Ulysses SWICS pick-up data on the other. The most appropriate HST data in the solar environment are those of Linsky et al (1993). These authors have measured both neutral oxygen and neutral hydrogen columns along the line-of-sight towards the nearby star Capella. Towards this target located at 12 pc, only one velocity cloud component produces interstellar absorption lines, the one corresponding to our Local Cloud (Lallement and Bertin, 1992, Lallement et al, 1995). As a consequence, relative abundances deduced from column-densities calculations towards this star directly apply to the LIC (and not to companion clouds) and are the most representative of the extra-heliospheric gas. The Linsky et al data imply a neutral oxygen to neutral hydrogen ratio OI /HI of 4.8 with a rather small uncertainty of the order of 10%. From the charge-exchange equilibrium described in Sect. 3, the corresponding O ions density outside of the heliosphere is 4.2 cm-3, and is independent of the neutral H density. For a plasma density of 0.1 , and using the Fig. 6, the ion density at 5 A.U. is found to be 0.015 x 4.2 = 6.4 cm3. The corresponding ion flux assuming all ions are moving at the solar wind speed choosen for the model (450 km ) is found to be of the order of 30 ions cm2 s-1. This number is dependent on our assumptions on the local proton (= electron) and neutral H densities and on the (assumed constant) solar wind characteristics.
The theoretical and measured values then agree within a factor of 1.5. If taken strictly, the SWICS measurements imply a larger filtration and a plasma density higher than the model value used here, i.e. 0.1 cm-3, or alternately a smaller oxygen density. However a smaller oxygen density implies a smaller neutral H density, and the value we have used here is very close to the lower limit for the neutral hydrogen outside the heliosphere (see Quémerais & al, 1994.) However, in our opinion, the model assumptions on the one hand, in particular the absence of magnetic field, and the constancy of the solar wind, the uncertainties in the fluxes measurements on the other hand (e.g. ions at velocities of the order of 800 km s-1 are not detected, and the ion flux close to the Sun is directly related to the instantaneous solar wind), preclude such a firm conclusion. More data are needed for averaging over spatial and temporal variations of the solar wind and the photoionization. Nevertheless, this rough agreement is promising. It also certainly shows that there is not too much oxygen in the heliosphere for precluding a filtering. On the contrary, the comparison with the interstellar data suggests a substantial filtering, possibly am already decrease included in the present model, plus the suggested additional decrease by 1.5 quoted above).
The neutral oxygen abundance relatively to helium has been deduced by Geiss et al (1994) from oxygen ions fluxes by comparison with pick-up helium fluxes, in the context of classical models, i.e. the resulting number refers to the inner heliosphere, outside the region of significant interaction with the Sun. A He/O ratio of 290 (+ 190, -100) has been inferred from zero temperature models for O and He. Although the comparison is more indirect than the previous one, it is in with the astronomical observations of Linsky et al (1993), which provide HI/OI= 2,100. However, it requires an assumption about the HI/HeI ratio. The Extreme UltraViolet Explorer measurements have provided a series of determination of this ratio towards nearby white dwarfs, with the conclusion that HI/HeI is comprised in the range 9.3- 20.0 depending on the targets, with a mean value of 14.5, implying that in the Sun environnement helium is always at least as ionized as hydrogen (Dupuis et al, 1995). We will make use of these results. If the helium and hydrogen were equally ionized (HI/HeI = 10, i.e. close to the lower value) then the Linsky LIC results give HeI/OI= 210. If HI/HeI has the mean value 14.5 then the Linsky data imply HeI/OI= 145. If one compares with the total range of values derived by Geiss & al (190-480), this implies a large range of possibilities, including no filtration on one hand, or a very strong decrease by a factor of 3 on the other. As a conclusion, the oxygen/helium ratio is also compatible nt filtration of the oxygen atoms. However, the Geiss & al (1994) O and He ions relative fluxes are in good agreement with their models if the oxygen ionization rate is of the order of 3 s-1 or less. This is in one sense satisfying because it is a reasonable value. But, as already noticed by Fahr & al, 1995, the use of such a reasonable ionization rate, results in a very high HeI/OI ratio of about 400 or more, much larger than the solar system value of 114 suggesting a very strong filtration.
Here we compare with what has been shown above to be the most likely interstellar ratio (HeI/O I= 145-210), i.e. a more appropriate quantity, somewhat larger than the solar system value, but it still implies a rather strong filtration (a decrease by a factor of at l east 2). A filtration as large as by a factor of 2 is certainly precluded because it would imply a too large plasma density and then a too small heliosphere. Further work is needed to explain these contradictions, in terms of differences between classical and interface models, or the neglect of one or more processes. A first attempt here is a comparison between the distribution of O atoms at small heliocentric distances as it comes out from the interface model, with a density distribution issued from classical hot models without interface (only supersonic solar wind).
In Fig. 7 are plotted oxygen densities calculated in the frame of a "classical" model for the same velocity and temperature of the inflowing oxygen as those prevailing outside the heliosphere in the interface model. In the classical case the photo-ionization and the charge-exchange are represented by a unique term for loss processes calculated for the mean solar wind velocity, and varying as r-2. This is at variance with the full model for which all charge-e tive velocity of the ion and the atom. The total ionization rate at 1 A.U. is 5.79 10 -7 s -1 (lifetime against ionization 1.73 10 6 s). The comparison is done for upwind, sidewind and downwind lines-of-sight. To allow an easier comparison, classical densities have been divided by 1.2, which corresponds to the filtration factor on the upwind side. Although our statistics are too poor to quantify in a precise way the changes in the filling very close to the Sun, the densities issued from the two models are not extremely different, except at large distances (the oxygen wall) and possibly very close to the Sun, and no firm conclusions can be derived from this first attempt.
We now turn our attention to the cosmic ray measurements. The ACR are observed in the outer heliosphere by the Voyager cosmic ray instruments since 1977. In addition to oxygen, He, Ne, C, N and Ar are also detected, allowing for the first time relative abundances determinations of neutral species in the interstellar extra-heliospheric gas, through modelling of the ionization, pick-up and acceleration processes (e.g. Cummings et al, 1984, Jokipii, 1986, Cummings and Stone, 1990). It was recently suggested by Frisch (1994) to make use of the C and O abundances derived from the ACR, to infer the electron density in the local ISM. When applying this method, Frisch (1994,1995) deduces an electron density range of 0.22-0.63 cm-3, largely above previous estimates, but in good agreement with independent estimates from the interstellar magnesium equilibrium, as observed from the absorption lines in stellar spectra (Lallement et al, 1995), also providing a surprisingly large electron density. In principle, these estimates from ACR data should include a correction for the neutral oxygen filtration, because only pick-up ions created in the supersonic solar wind have a peculiar enough velocity distribution making them the seed particles in the ACR production. A very precise correcting factor requires a full analysis combining ACR convection models and O-ion source functions as presented here, which is well beyond the scope of this work. However, a simple look at Fig. 3, 5 and 6 shows that for the two-shocks model for =0.1 , the total number of neutral oxygen atoms which penetrates the supersonic wind region, is smaller by at most 30% as compared with the "no interface" case. As a consequence, on this basis it does not appear necessary to apply a large correction to the Frisch (1994) results. A rough estimate shows that the inferred plasma density decreases by about the same amount. It remains that carbon filterin d also be computed, which is a difficult task due to uncertainties in the charge-exchange cross-sections.
© European Southern Observatory (ESO) 1997