In an earlier communication1 we reported the observation of electron spin resonance (ESR) in spinach ferredoxin, a nonheme iron protein isolated from spinach. An ESR signal centered on g = 1.94 was observed after reducing the protein with sodium dithionite; this signal was just visible as a broad line at 150'K, while on cooling to 90'K the line intensified and structure appeared on the wings. We now wish to report the behavior of this spectrum at temperatures down to 20K, to discuss the model recently put forward by Brintzinger, Palmer, and Sands,2 and to suggest an alternative explanation of the spectrum.The experimental procedures used were essentially those described by Hall, Gibson, and Whatley,' who followed the method of Hill and Bendall to prepare the spinach (Spinacea oleracea) ferredoxin. For ESR experiments at liquid hydrogen and liquid helium temperatures, the samples were placed in silica tubes 2 cm long with narrow neck and containing a little sodium dithionite to reduce the ferredoxin. The tubes were quickly plugged with silicone grease, shaken, and then frozen. The very low temperature experiments were performed on a transmission X-band spectrometer of conventional design.Our low temperature data show that the spectra previously observed at 900K1 and 40'K2 remain essentially the same down to 20K, and can be attributed to a ground state doublet with g9 = 1.88, g, = 1.94, and g, = 2.04. We have observed that at the lowest temperatures the spectrum is easily saturated, implying that some strongly temperature-dependent relaxation mechanism operates. In this connection, it would be most interesting to measure the temperature dependence of the relaxation process, for, if it were due predominantly to the so-called twophonon resonant mechanism,3 valuable information about the energy levels of the complex might be obtained. We shall come back to this point later, as the relaxation behavior supports our model. Brintzinger et al.2 have recently postulated that the iron in reduced spinach ferredoxin is low-spin ferric in a tetrahedral environment. However, their model has some features that are not too satisfactory. First, a very large tetrahedral ligand field splitting (of order at least 20,000 cm-') is required to cause spinpairing, and also to give the positive g,-shift observed; this is much larger than is observed in a wide variety4 of complexes (2000-5000 cm-'). Further, such a strong tetrahedral field with spin-pairing might be expected to cause a much more intense optical spectrum than is in fact observed, as both spin-allowed and Laporteallowed transitions are likely to occur.We wish to suggest that the two iron atoms in the spinach ferredoxin molecule strongly interact with one another through one or more ligands, which in this case are likely to be sulfur. If the reduced complex is of the schematic type Fe3+(d5, S = 5/2) -sulfur ligand(s) -Fe2+(d6, S = 2), then an antiferromagnetic exchange interaction between the two spins will couple 987
