Titan's atmosphere is optically thick and hides the surface and the lower layers from the view at almost all wavelengths. However, because gaseous absorptions are spectrally selective, some narrow spectral intervals are relatively transparent and allow to probe the surface. To use these intervals (called windows) a good knowledge of atmospheric absorption is necessary. Once gas spectroscopic linelists are well established, the absorption inside windows depends on the way the far wings of the methane absorption lines are cut-off. We know that the intensity in all the windows can be explained with the same cut-off parameters, except for the window at 2 μm. This discrepancy is generally treated with a workaround which consists in using a different cut-off description for this specific window. This window is relatively transparent and surface may have specific spectral signatures that could be detected. Thus, a good knowledge of atmosphere opacities is essential and our scope is to better understand what causes the difference between the 2 μm window and the other windows. In this work, we used scattered light at the limb and transmissions in occultation observed with VIMS (Visual Infrared Mapping Spectrometer) onboard Cassini, around the 2 μm window. Data shows an absorption feature that participates to the shape of this window. Our atmospheric model fits well the VIMS data at 2 μm with the same cut-off than for the other windows, provided an additional absorption is introduced in the middle of the window around ≃ 2.065 μm. It explains well the discrepency between the cut-off used at 2 μm, and we show that a gas with a fairly constant mixing ratio, possibly ethane, may be the cause of this absorption. Finally, we studied the impact of this absorption on the retrieval of the surface reflectivity and found that it is significant.