ABSTRACT We analyse the transmission spectra of KELT-20b/MASCARA-2b to search for possible thermal inversion agents. The data consist of three transits obtained using HARPSN and one using CARMENES. We removed stellar and telluric lines before cross-correlating the residuals with spectroscopic templates produced using a 1D plane-parallel model, assuming an isothermal atmosphere and chemical equilibrium at solar metallicity. Using a likelihood-mapping method, we detect Fe i at > 13σ, Ca ii H$\&$K at > 6σ and confirm the previous detections of Fe ii, Ca ii IR Triplet, and Na i D. The detected signal of Fe i is shifted by −3.4 ± 0.4 km s−1 from the planetary rest frame, which indicates a strong day–night wind. Our likelihood-mapping technique also reveals that the absorption features of the detected species extend to different altitudes in the planet’s atmosphere. Assuming that the line lists are accurate, we do not detect other potential thermal inversion agents (NaH, MgH, AlO, SH, CaO, VO, FeH, and TiO) suggesting that non-chemical equilibrium mechanisms (e.g. a cold-trap) might have removed Ti- and V-bearing species from the upper atmosphere. Our results, therefore, show that KELT-20b/MASCARA-2b cannot possess an inversion layer caused by a TiO/VO-related mechanism. The presence of an inversion layer would therefore likely be caused by metal atoms such as Fe i and Fe ii. Finally, we report a double-peak structure in the Fe i signal in all of our data sets that could be a signature of atmospheric dynamics. However, further investigation is needed to robustly determine the origin of the signal.