Today, the majority of the cosmic baryons in the Universe are not observed directly, leading to an issue of ‘missing baryons’ at low redshift. Cosmological hydrodynamical simulations have indicated that a significant portion of them will be converted into the so-called warm–hot intergalactic medium (WHIM), with gas temperature ranging between 10^5 and 10^7 K. While the cooler phase of this gas has been observed using O vi and Ne viii absorbers at ultraviolet (UV) wavelengths, the hotter fraction detection relies mostly on observations of O vii and O viii at X-ray wavelengths. Here, we target the forbidden line of [Fe xxi] λ 1354 Å which traces 10^7 K gas at UV wavelengths, using more than 100 high-spectral resolution (⁠|$R\sim 49\, 000$|⁠) and high signal to noise VLT/UVES quasar spectra, corresponding to over 600 h of VLT time observations. A stack of these at the position of known Ly α absorbers lead to a 5σ limit of |$\log [N\mathrm{([Fe\,{\small XXI}])]\lt }$|17.4 (EW_rest < 22 mÅ), three orders of magnitude higher than the expected column density of the WHIM |$\log [N\mathrm{([Fe\,{\small XXI}])]\lt }$|14.5. This work proposes an alternative to X-ray detected 10^7 K WHIM tracers, by targeting faint lines at UV wavelengths from the ground benefiting from higher instrumental throughput, enhanced spectral resolution, longer exposure times, and increased number of targets. The number of quasar spectra required to reach this theoretical column density with future facilities including 4MOST, ELT/HIRES, MSE, and the Spectroscopic Telescope appears challenging at present. Probing the missing baryons is essential to constrain the accretion and feedback processes that are fundamental to galaxy formation.