We report on the first quantitative determination of the electric quadrupolar contribution to the nuclear spin-lattice relaxation in a transition metal. For 186 Ir and 189 Ir in Fe we have determined the magnetic and the electric quadrupolar part of the relaxation for magnetic fields between 0.01 and 2 T. The quadrupolar part gives information on the role of the orbital motion of the electrons for the relaxation process. Our results prove that the unexpected high relaxation rates in Fe and their magnetic field dependence are due to a nonorbital relaxation mechanism. DOI: 10.1103/PhysRevLett.88.057601 PACS numbers: 76.60.Es, 75.50.Bb, 76.80. +y The nuclear spin-lattice relaxation in metals arises predominantly from the magnetic hyperfine interaction between the nuclear spins and the conduction electrons. It gives information on the density of states at the Fermi energy and the magnitude of low frequency spin fluctuations [1 -3].The interpretation of the relaxation is complicated in transition metals by the fact that several relaxation mechanisms can contribute, but only the total relaxation rate R is usually measured: According to the type of the responsible hyperfine interaction, R can be subdivided into an orbital (R o ), a Fermi-contact (R c ), a spin-dipolar (R sd ), a core-polarization (R cp ), and a quadrupolar (R q ) contribution [3,4]. The first four contributions represent together the magnetic relaxation rate R m . R q is due to the electric hyperfine interaction and arises from electric field gradient fluctuations at the nuclear site.The various contributions involve quite different electronic excitations: R c , for example, is connected to spin flips of s electrons; R o and R q are connected to orbital momentum changes of p or d electrons. A separate determination would, therefore, give valuable information on the relaxation process. Such information is especially desirable since experiment and theory are in serious disagreement for several transition metal systems [5,6].In this situation, one can make use of two special features of R q : R q can be determined separately and it is so closely related to R o that R o can be estimated from R q [4,7]. This offers the unique possibility to decompose experimentally the relaxation into an orbital and a nonorbital part.This possibility is used here for the first time. It is used to study the origin of two still unsolved problems in the nuclear spin-lattice relaxation in Fe: (i) The relaxation rates are usually larger than predicted by ab initio calculations. This effect is especially prominent and well documented for the 5d impurities in Fe [6]. (ii) For magnetic fields below 1 T there is a so far unexplained magnetic field dependence due to which the relaxation in zero field is typically 3 to 10 times faster than at high fields [8][9][10].This work is also the first experimental determination of R q in a transition metal, although the presence of this contribution to the relaxation in transition metals is meanwhile well established in the theoretical wor...