The spin-rotation of a planet arises from the accretion of angular momentum during its Near-infrared high-dispersion spectroscopy has been used to characterize the atmospheres of hot Jupiters in close-in orbits [10][11] . Such observations utilize changes in the radial component of the orbital velocity of the planet (resulting in changes in Doppler shift) to filter out the quasi-stationary telluric and stellar contributions in the spectra. Here we make use of the spatial separation and the difference in spectral signature between the planet and star, which allow the starlight to be filtered out. A similar technique 12 has been applied very successfully 13 at a medium spectral dispersion to characterize the exoplanet HR8799c. We observed the β Pictoris system 7,8 (K=3.5) using the Cryogenic highResolution InfraRed Echelle Spectrograph CRIRES 14 located at the Nasmyth focus of UT1 of the Very Large Telescope (VLT) of the European Southern Observatory (ESO) at Cerro Paranal in Chile on the night of 17 December 2013, with the slit oriented in such way that it encompassed the planet and star.An important step in the data analysis is the optimal removal of the stellar contribution along the slit, which for this A-star consists mostly of a telluric absorption spectrum. The resulting spectra were cross-correlated with theoretical spectral templates constructed in a similar way as in our previous work on hot Jupiters 10-11 , varying the planet's atmospheric temperature pressure (T/p) profile, the carbon monoxide abundance and that of water vapor and methane, which can also show features in the observed wavelength range. Note that there is a strong degeneracy between the atmospheric T/p profile and the abundance of the molecular species, meaning that different combinations of these parameters result in nearly identical template spectra.At the expected planet position a broad and blue-shifted signal is apparent (see Figure 1), which is strongest when the cross-correlation is performed with a spectral template from an atmospheric model with deep carbon monoxide lines and a small contribution from water. We estimate the signal to have a SNR of 6.4 by cross-correlating the residual spectrum with a broadened model template, and compare the peak of the cross-correlation profile with the standard deviation. If we use the crosscorrelation profile as seen in Figure 1 to estimate the SNR, we need to take into account the width of the signal and the dependence of adjacent pixels in the profile. This results in an SNR of 7.8, but this latter method we found to be less accurate, since it does not properly include contributions from correlated noise structures on scales of the broad signal. Cross-correlation with the optimal spectrum of water vapor alone provides a marginal signal at SNR~2, which means we cannot claim a firm detection of water in the planet atmosphere. No signal is retrieved for methane models (see Extended Data Fig. 1).We fit the planet profile using a grid of artificial cross-correlation functions, produced by crossc...