A new search for carbon monoxide absorption in the transmission spectrum of the extrasolar planet HD 209458b

We have revisited the search for carbon monoxide absorption features in transmission during the transit of the extrasolar planet HD 209458b. In 2002 August-September we acquired a total of 1077 high-resolution spectra (lambda/delta lambda similar to 25,000) in the K-band (2 mu m) wavelength region using NIRSPEC on the Keck II telescope during three transits. These data are more numerous and of better quality than the data analyzed in an initial search by Brown et al. Our analysis achieves a sensitivity sufficient to test the degree of CO absorption in the first-overtone bands during transit on the basis of plausible models of the planetary atmosphere. We analyze our observations by comparison with theoretical tangent geometry absorption spectra, computed by adding height-invariant ad hoc temperature perturbations to the model atmosphere of Sudarsky et al. and by treating cloud height as an adjustable parameter. We do not detect CO absorption. The strong 2-0 R-branch lines between 4320 and 4330 cm(-1) have depths during transit less than 1.6 parts in 10(4) in units of the stellar continuum (3 sigma limit) at a spectral resolving power of 25,000. Our analysis indicates a weakening similar to that found in the case of sodium, suggesting that a general masking mechanism is at work in the planetary atmosphere. Under the interpretation that this masking is provided by high clouds, our analysis defines the maximum cloud-top pressure (i.e., minimum height) as a function of the model atmospheric temperature. For the relatively hot model used by Charbonneau et al. to interpret their sodium detection, our CO limit requires cloud tops at or above 3.3 mbar, and these clouds must be opaque at a wavelength of 2 mu m. High clouds comprised of submicron-sized particles are already present in some models but may not provide sufficient opacity to account for our CO result. Cooler model atmospheres, having smaller atmospheric scale heights and lower CO mixing ratios, may alleviate this problem to some extent. However, even models 500 K cooler than the Sudarsky et al. model require clouds above the 100 mbar level to be consistent with our observations. Our null result therefore requires clouds to exist at an observable level in the atmosphere of HD 209458b, unless this planet is dramatically colder than current belief.