Abstract
EPR oximetry is used for in vivo and in vitro measurements of oxygen in biological systems, including experimental animals. The accuracy of oxygen measurements in T~2~ (line width) EPR oximetry is significantly improved if least‐squares simulation is used to extract the line width parameters. The oxygen effect on the EPR spectra of nitroxide solutions and aqueous suspensions of fusinite can be described as an additional homogeneous broadening that modifies the EPR spectrum of the oxygen‐free probe. This allows one to use a one‐parameter line width model in most cases. The simulations were carried out with the use of a fast‐convolution algorithm followed by Levenberg‐Marquardt optimization. The validity of error estimates provided by this method was tested on sets of experimental spectra taken under common conditions. It is shown that the accuracy of oxygen measurements in line width (T~2~) oximetry is determined not only by the probe sensitivity (rate of line width change versus oxygen concentration), but also by the signal‐to‐noise ratio, inhomogeneous contribution to the line shape (e.g., unresolved proton superhyperfine structure), and the spectral window. The accuracy of oxygen measurements is compared for aqueous solutions of two nitroxide radicals with different superhyperfine structure and for aqueous suspensions of fusinite.