Prior investigations have attempted to characterize the longitudinal variability of the column number density ratio of atomic oxygen to molecular nitrogen (ΣO/N2) in the context of non-migrating tides. The retrieval of thermospheric ΣO/N2 from far ultra-violet (FUV) emissions assumes production is due to photoelectron impact excitation on O and N2. Consequently, efforts to characterize the tidal variability in O/N2 have been limited by ionospheric contamination from O+ radiative recombination at afternoon local times (LT) around the equatorial ionization anomaly. The retrieval of ΣO/N2 from FUV observations by the Ionospheric Connection Explorer (ICON) provides an opportunity to address this limitation. In this work, we derive modified ΣO/N2 datasets to delineate the response of thermospheric composition to non-migrating tides as a function of LT in the absence of ionospheric contamination. We assess estimates of the ionospheric contribution to 135.6 nm emission intensities based on either Global Ionospheric Specification (GIS) electron density, International Reference Ionosphere (IRI) model output, or observations from the Extreme Ultra-Violet imager (EUV) onboard ICON during March and September equinox conditions in 2020. Our approach accounts for any biases between the ionospheric and airglow datasets. We found that the ICON-FUV dataset, corrected for ionospheric contamination based on GIS, uncovered a previously obscured diurnal eastward wavenumber 2 tide in a longitudinal wavenumber 3 pattern at March equinox in 2020. This finding demonstrates not only the necessity of correcting for ionospheric contamination of the FUV signals but also the utility of using GIS for the correction.