VOLUME 45, NUMBER 25 PHYSICAL RKVIKW LKTTKRS 22 DECEMBER 1980 Observation of — 100'/o Spin-Polarized Photoelectrons from a Transversely Magnetized Ni(110) Single Crystal E. Kisker, W. Gudat, E. Kuhlmann, R. Cj.auberg, and M. Campagna Institut fu r Festko vPer'forschung der Kernforschungsanlage Julich, D 5I 70-Ju'lich, West Germany (Received 4 August 1980) We have measured for the first time the electron spin polarization P of photoelectrons emitted from a transversely magnetized sample in the absence of an externally applied magnetic field. With this photoemission geometry, use has been made of optical selec- tion rules for selectively exciting spin-polarized electron states of different symmetry as illustrated for emission from the Z3 and Z4 bands of Ni(110), where values of P up to — 10010 have been observed. PACS numbers: 79.60.Cn, 71.25.Pi, 75.50.Cc In this Letter we demonstrate for the first time that spin-polarized photoemission experiments from ferromagnets can successfully be performed with a transversely magnetized sample (i.e. , transverse geometry) without applying an exter- nal magnetizing field H, during the measurements. The long-standing belief that a longitudinal geom- etry (i.e. , H, parallel to the electron optical ax- is) is a necessary condition in spin-polarized pho- toemission studies does therefore not hold true any more. ' As a further consequence, many spin-polarized photoemission experiments (es- pecially energy-resolved spin-polarized work) will be much more easily performed. We use this new method for investigating Ni(110) and report for the first time effects on the spin- polarization spectra of a ferromagnetic metal by selective excitation of electronic energy bands on the same single-crystal surface. Conventional angle-resolved photoemission spectroscopy (ARPES)" and spin-polarized pho- toemission studies' ' of single-crystal Ni recent- ly yielded definite experimental evidence that the exchange splitting b, and the Stoner gap 5 (dis- tance of the top of the majority-spin d band from the Fermi energy ZF) are only about half of the value or smaller than that predicted by self-con- sistent band calculations. The observed negative spin polarization at photothreshold is already pre- dicted by the simple Stoner-Wohlfahrt-Slater the- ory of ferromagnetism, " but the value of the crossover energy E, from negative to positive spin polarization is a more difficult problem and has been the subject of several theoretical inves- tigations. ' " Recently, the first energy-resolved spin-po- larization experiments on Ni(111) with synchro- tron radiation from the ACO storage ring at Laboratoire pour 1'Utilisation de Rayonnement Electromagndtique (LURE), Orsay, up to 40 eV" showed that it is possible to detect spin-polarized bands in ferromagnetic metals at binding ener- gies beyond 0.5 eV below the Fermi energy, a range where ARPES can no longer distinguish spin up and spin down, primarily because of life- time effects. After the failure of the first and other spin-po- larized photoemission experiments performed with the so-called "transverse" geometry, ' ' ' all the subsequent successful studies were per- formed with the so-called "longitudinal" geom- etry, as was the first successful one. " In all these studies the direction of emission of the pho- toelectrons (i.e. , the angle information) is com- pletely lost in the strong magnetic field' and a (differential) energy analysis is made very troub- lesome because of the occurrence of transverse velocity components when extracting the photo- electron beam out of the magnetic field. "" Only by taking advantage of the fact that the electrons emitted with maximum kinetic energy are leaving the crystal along the surface normal, angle and energy analysis has been performed recently with use of the retarding-field energy analyzer. " With the transverse geometry (magnetization perpen- dicular to the electron optical axis) the above- mentioned electron-optical problems, which are specific for the longitudinal geometry, are not present. In this Letter we show that, in contrast to com- mon belief and to the findings of earlier experi- ments, the transverse geometry can be used suc- cessfully in spin-polarized photoemission experi- ments, " for electron kinetic energies down to the millielectrovolt region. The apparatus used is a modification of the one described by Kisker et al. ' A new electron-lens system, schematically shown in Fig. 1, was con- structed in order to collimate the photoelectrons effectively since the (initially) focusing longitudi- 1980 The American Physical Society 2053