6 Discussion

6.1 Extension to other cases of terrestrial exoplanets

6.1.1 Influence of the spectral type of the star

What would be the impact of changing the stellar spectrum on the photochemical production of ${\rm O}_{2}$ and ${\rm O}_{3}$? The most important parameter is the ratio $R_{\rm uv}={F}_{\rm uv}/{F}_{\rm tot}$ where $F_{\rm uv}$ is the stellar flux in the UV range important for photochemistry ([100-400] nm) and $F_{\rm tot}$ the total integrated flux. We performed simulations with $R_{\rm uv} < {\rm solar} ~ R_{\rm uv}$(cooler stars); in all cases they gave a lower ${\rm O}_{2}$ and ${\rm O}_{3}$ content. We also performed simulations with $R_{\rm uv} > {\rm solar} ~ R_{\rm uv}$ (hotter or very young stars), but only for case G ( ${\rm O}_{2}$ production from ${\rm H}_{2}$O photolysis and H escape), without changing our conclusions.

For other cases with a high $R_{\rm uv}$, it is difficult to guess what would happen: more UV produces more atomic oxygen, but not necessary more ${\rm O}_{2}$ or ${\rm O}_{3}$, because these species are also unstable under high UV irradiation. We note however that:

• A possible consequence of higher $R_{\rm uv}$ would be to produce in case F an ${\rm O}_{3}$ signature with less ${\rm CO}_{2}$ ( $\mbox{$P_{{\rm CO}_{2}}$ }<50$ mbar). Then, there would be less ${\rm CO}_{2}$ masking and we would see an unambiguous ${\rm O}_{3}$ feature. But anyway, case F is dry and does not display the ${\rm H}_{2}$O component of the triple signature required to trace bio-photosynthesis;
• Hot stars like F-stars are rarer than solar or less luminous stars and have a shorter life. We showed elsewhere (Selsis 2000a) that the 9.6 $\mu$m ${\rm O}_{3}$ band is not a robust way to detect ${\rm O}_{2}$-rich (Earth-like) atmospheres around such hot stars. Indeed, the temperature contrast between the surface and the emitting ozone layer is too low to induce a signature detectable by a Darwin-like instrument.

   
6.1.2 The case of reducing early atmospheres

When exploring the past of the Earth's atmosphere, we concentrated on the scenario favoured nowadays for the primitive atmosphere, i.e. a ${\rm CO}_{2}$- ${\rm H}_{2}$O- ${\rm N}_{2}$ mixture. If one considers the option of a mildly reducing atmosphere (Pavlov et al. 2000; Galimov 2000), the possibilities of photochemically producing a substantial amount of ${\rm O}_{2}$ are much lower. Namely, reducing gases like ${\rm H}_{2}$, ${\rm CH}_{4}$ or ${\rm NH}_{3}$ would react with ${\rm O}_{2}$ and strongly limit its level (see e.g. Selsis 2002 about the ${\rm CH}_{4}$/ ${\rm O}_{2}$ competition). Therefore abiotic photochemical production of ${\rm O}_{2}$ in reducing atmospheres is inefficient and the concomitant detection of ${\rm O}_{2}$ or ${\rm O}_{3}$ with ${\rm NH}_{3}$ (9 and 11 $\mu$m in the Darwin window) or ${\rm CH}_{4}$ (7.5 $\mu$m) would strongly suggest biological activity.

6.2 Consequences for Darwin and TPF

   
6.2.1 Maximum O $\mathsf{_2}$ abiotic production

Looking back to Table 3, we note that ${\rm O}_{2}$ can become, through photochemical production in one case (case E, i.e. 4 bar dry ${\rm CO}_{2}$ atmosphere), a major component of the atmosphere, reaching $x_{\rm O2} \sim 5$%. This result is close to the $x_{\rm O2}= 3.9$% found by Nair et al. (1994) for a lower pressure dry ${\rm CO}_{2}$ atmosphere (dry Mars), found with an equilibrium model. Although this 5% level is somewhat below the 20% level of present terrestrial ${\rm O}_{2}$, this questions our ability to discriminate between abiotic and biologic ${\rm O}_{2}$ origin. This is a crucial point when one considers searching for ${\rm O}_{2}$ photosynthetic life by the direct and unique detection of ${\rm O}_{2}$. Note however that the corresponding atmospheres are extreme and may be unlikely cases: is it possible to build up a ${\rm CO}_{2}$ atmosphere completely devoid of ${\rm H}_{2}$O? Even Mars and Venus have traces of water vapor. And many other compounds found in planetary atmospheres (sulfur or chlorine compounds, for example) can catalyse, at trace levels, the reverse process, reforming ${\rm CO}_{2}$ from CO and  ${\rm O}_{2}$.

   
6.2.2 Is case G (permanent water supply) a realistic false positive case?

Coming back to case G, where, although marginal, a triple signature might appear, we stress the ad hoc nature of the situation considered: a constant influx of ${\rm H}_{2}$O-bearing small particles or larger bodies, the absence of an oxygen sink (no reducing species in the atmosphere, no recycling of the surface), a high UV irradiation and an efficient escape of H atoms. To obtain the upper limit of 1% of ${\rm O}_{2}$, the water influx was set to 10¹⁰ molecules cm^-2 s^-1. If delivered via hydrated dust similar to solar IDPs with an average water content of 4 wt% (Maurette et al. 2000), this flux implies the infall of $1.9\times 10^{9}$ tons yr^-1 on the whole Earth. This is $3\times 10^{4}$ higher than the current flux ( $40~000\pm 20~000$ tons yr^-1, Love & Brownlee 1993). Earth might have experienced such a high flux only during the Late Heavy Bombardment (Maurette et al. 2000; Love & Brownlee 1993). So high delivery presupposes a very dusty interplanetary environment which lasts less than 400 My around main-sequence stars (Habing et al. 1999). Darwin and TPF will be able to detect and study terrestrial planets in the habitable zone only in the case of systems with low zodiacal emission (less than about 10 times the Solar System zodiacal light, Beichman et al. 1999). Therefore, such a case proves to be restricted to very young systems and moreover, probably non-observable in the mid-infrared. Moreover, although we considered only a deposition of pure water vapor, delivering water this way is also a source of highly reducing species. Indeed, cometary matter contains nearly as much mass in refractory complex organics than in water ice (Greenberg 2000) and collected micrometeorites have an average content of organic carbon of 3 wt% (Maurette 1998). Such species, delivered at the same time as water, would prevent any ${\rm O}_{2}$ accumulation.

   
6.2.3 The CO $\mathsf{_2}$ masking of O $\mathsf{_3}$: Advantages and disadvantages

We have seen (cases B, C, D and E) that when the ${\rm CO}_{2}$ pressure is high enough to photochemically produce an IR-absorbing ozone layer ( $\mbox{$P_{{\rm CO}_{2}}$ }> 50$ mbar for a dry atmosphere, $\mbox{$P_{{\rm CO}_{2}}$ }> 1$ bar for a humid atmosphere), high ${\rm CO}_{2}$ pressure bands (centered at about 7.3, 7.9, 9.4 and 10.5 $\mu$m) affect the thermal spectrum. With Darwin resolution ($R\sim 20$) and signal-to-noise ratio (SNR< 10), the two ${\rm CO}_{2}$ features arising on both side of the 9.6 $\mu$m band hide the ${\rm O}_{3}$ signature. By masking the abiotic ozone it sustains, ${\rm CO}_{2}$ masks also the potential "false positive imposters'' that photochemistry could produce in the mid-infrared.

For higher spectral resolution than foreseen for Darwin spectra in Fig. 6 show that ${\rm O}_{3}$ and ${\rm CO}_{2}$ features can in principle be distinguished, even in the case of a humid atmosphere. However, the "triple signature'' criterion still allows in this case to trace their possible abiotic origin, when high ${\rm CO}_{2}$ pressure bands are present. On the other hand, the masking effect (at low resolution) as well as the rejection of spectra with high ${\rm CO}_{2}$ pressure (at high resolution) can produce "`false negatives'', as they lead one to reject truly photosynthetic ecosystems that would be detectable under a ${\rm CO}_{2}$-poor atmosphere (Selsis 2002). If one considers it more important to avoid false positive detections than to miss inhabited planets, the masking of abiotic ${\rm O}_{3}$ and then of false positive detection is clearly an advantage, as this eliminates several potentially ambiguous cases.

   
6.2.4 Is the Darwin strategy validated?

Within the framework of our approach, does the Darwin concept still appear to be validated with respect to the risk of false positive detection? The answer is clearly yes. Relying on our simulations and discussions above, it turns out that the simultaneous signature of ${\rm H}_{2}$O, ${\rm CO}_{2}$ and ${\rm O}_{3}$ within Darwin's spectral window cannot be due to abiotic photochemistry. With an additional criterion imposing the absence of high ${\rm CO}_{2}$ pressure bands, this triple biomarker can be extrapolated to future high performance instruments working in the mid-infrared range. Searching for the triple IR signature of ${\rm O}_{3}$, ${\rm CO}_{2}$ and ${\rm H}_{2}$O with a Darwin-like instrument appears more robust than a direct but unique detection of ${\rm O}_{2}$. Indeed, while ${\rm O}_{2}$ can become abiotically a major atmospheric component (up to few percents) ${\rm O}_{3}$ cannot be detected in such cases at the same time as ${\rm H}_{2}$O and ${\rm CO}_{2}$, due to the masking effect of ${\rm CO}_{2}$, and/or to the catalytic cycles destroying ${\rm O}_{3}$ and following ${\rm H}_{2}$O photolysis. Thus, through the triple signature, Darwin effectively filters out false positive detections in all the cases. Moreover,

-

as oxygenic photosynthesis extracts ${\rm O}_{2}$ from ${\rm H}_{2}$O and fix carbon from ${\rm CO}_{2}$;

-

as life appears to be, as far as we know it, indissociable from ${\rm H}_{2}$O;

-

as ${\rm CO}_{2}$ is a constituent of all the known terrestrial planet atmospheres and an expected constituent of habitable extrasolar planets (Kasting et al. 1993);

-

and as ${\rm O}_{3}$ is a powerful, logarithmic, tracer of ${\rm O}_{2}$ (Léger et al. 1993);

searching for this triple signature in quest of photosynthetic sources of ${\rm O}_{2}$ is not a restrictive strategy when compared to search for ${\rm O}_{2}$, or ${\rm O}_{3}$, alone.

Furthermore, the relevance of this mid-IR triple signature is strengthened when considering that, in our simulations, the abiotic production of ${\rm O}_{2}$ and ${\rm O}_{3}$ was optimized on purpose. Adding more realistic processes or compounds (weathering and chemical interaction with rocks, delivery of external matter, volcanic emissions) to these "clean'' atmospheres diminishes the possible amounts of these two molecules. This trend was noted when simulating early Earth cases (C, D) and it clearly makes mid-infrared false positive detection even more unlikely.