PHYSICAL REVIEW B VOLUME 42, NUMBER 18 15 DECEMBER 1990-II Spin-polarized photoelectron difFraction and valence-band photoemission from Mno(001) B. Hermsmeier, * J. Osterwalder, D. J. Friedman, ~ B. Sinkovic, ~ T. Tran, and C. S. Fadley Department of Chemistry, University of Hawaii, Honolulu, Hawaii 96822 (Received 11 June 1990) Spin-polarized photoelectron diffraction (SPPD) has previously been proposed as a technique for studying short-range magnetic order in magnetic materials, and the first experimental study of this kind was performed on the ionic antiferromagnetic KMnF3 [B. Sinkovic, B. Hermsmeier, and C. S. Fadley, Phys. Rev. Lett. 55, 1227 (1985)]. We present here a much more detailed study of SPPD for the antiferromagnetic transition-metal oxide MnO with a (001) surface orientation. The Mn 3s and Mn 3p multiplets have been studied using both low-energy (192. 6 eV) and high-energy (1486. 7 eV) x-ray excitation and the intensity ratios I('S( f ))/I('S($)) and I('P( f ))/I('P($)) have been mea- sured as a function of both direction and temperature. Data obtained with the lower excitation en- ergy and resulting in kinetic energies of 50-100 eV show an abrupt change or step in both the I('S( f ))/I( S($)) and I('P( t))/I('P($)) intensity ratios at =530+20 K or =4. 5 times the Neel temperature T&. This change is interpreted to be a new type of short-range-order transition occur- ring at what is denoted TsR. Also, these same quintet or septet intensity ratios show a weak peak at T&, suggesting for the first time that SPPD has sensitivity to long-range magnetic order. Data ob- tained for the I('S( t'))/I{ S($)) intensity ratio with the higher excitation energy show no such effects, a result consistent with the much weaker exchange scattering expected at such energies. Ad- ditional x-ray photoelectron spectroscopy spectra and azimuthal scans of Mn and 0 core-level in- tensities are considered and these establish that (i) the sample surface had good stoichiometry and was very clean and well ordered, (ii) the SPPD effects observed at TSR are not due to any surface structural change, and (iii) a single-scattering cluster (SSC) theoretical model is at least a qualitative- ly reasonable starting point for describing such effects. We also compare experimental results for the magnitudes of these steps with calculations based upon exchange scattering in the spin-polarized SSC model [B. Sinkovic and C. S. Fadley, Phys. Rev. B 31, 4665 (1985)], and conclude that there is at least qualitative agreement. A final aspect of our data concerns the temperature dependence of the Mn3d-dominated valence-band spectra: These spectra are found to show no measurable change in crossing TsR, but by contrast exhibit a large 0. 4-eV increase in width in going below T&, which is in contradiction to recent theoretical predictions. I. INTRODUCTION The use of spin-polarized photoelectron diffraction (SPPD) as a new technique for studying surface and near-surface short-range spin order in magnetic materials has previously been discussed from both theoretical' and e;xperimental points of view. A principal positive feature of this technique in comparison to other surface- sensitive probes of magnetic order, such as the spin- polarized versions of low-energy electron diffraction, Auger spectroscopy, photoemission, inverse photoemis- sion, or electron microscopy, is that in SPPD the origin of the spin-resolved electrons is internal to the sample, and thus the electron spins are referenced directly to the local magnetic lattice around a given emitter. This yields two important consequences: First, SPPD studies do not require an external spin detector or an external spin- polarized source, thereby greatly reducing the difficulty of the experiment. Second, because of this internal re- ferencing, SPPD is capable of investigating magnetic sys- tems more complex than simple ferromagnets; that is, there is no requirement that the sample have a macro- scopic magnetic moment (net magnetization). Thus, SPPD is particularly well suited for studying systems with antiferromagnetic order, and the first observations of such effects were in fact for the antiferromagnet KMnF3. The time scale of photoemission is also very short (approximately 10 ' — 10 ' sec), thus making it considerably faster than that of some other techniques currently used for studying short-range magnetic order such as neutron scattering, but comparable in time scale to that of the very surface-sensitive two-electron capture by deuterons. ' The spin-polarized electrons in SPPD arise from an exchange-split outer core-level multiplet in a high-spin ion. For all studies to date, the Mn + ion in a high-spin state of [3s 3p 3d ] S has been used. Emission of a 3s electron from this ion produces an easily resolved doublet with a separation of about 6 eV. Such multiplet splittings in ionic solids have been known for some time" ' and arise primarily via intra-atomic final-state I. -S term split- tings. In s emission this gives rise to a peak separation proportional to [2S+1]K(3s-3d), where S is the initial- state spin and I(: is the 3s-3d exchange integral, a result that is derived from the Van Vleck theorem. "' ' For high-spin Mn +, S = — ', and the final ionic states with a 3s hole are [3s'3p 3d ] coupled to S and S; the predicted splitting in simple one-electron Hartree-Fock theory is 11 895 1990 The American Physical Society