Epitaxial Gd(0001) has been investigated with spin-polarized low-energy-electron diffraction and the magneto-optic Kerr effect. The ferromagnetic critical temperature of the surface layer is found to lie up to 22 K above the bulk Curie point, demonstrating the existence of surface-enhanced magnetic order. Furthermore, spin-resolved photoemission spectroscopy reveals that the 4/ spins of the surface are not ferromagnetically coupled to the bulk moments.The concept of universality in the field of phase transitions has stimulated renewed interest for this mature field of solid state physics, because a variety of apparently very different phenomena can be described by the same theoretical principles. In this sense, for example, presetting criticality and pure surface-enhanced transitions are essentially the same phenomenon. 1 On the other hand, advances in experimental techniques today allow the crosschecking of such far-reaching predictions. Magnetic systems as model systems have always been appealing to experimentalists and theorists. Recently systems with free surfaces have been investigated extensively both analytically 2 " 4 and by Monte Carlo simulations. 5 For systems exhibiting a continuous bulk phase transition the critical behavior is found to be related to the ratio of J\/Jb of the parallel coupling Ji between surface spins in the top layer and that J b between bulk spins. For values of J x below a certain critical value J Xc the surface and the bulk will have the same critical temperature, while for J x > J Xc a pure "surface transition" has been predicted to occur. In such a case the surface spins undergo a critical ordering transition in the presence of disordered bulk spins. 5 One purpose of this Letter is to report the first direct observation of such a critical surface transition on epitaxial Gd(0001) by means of the surface-sensitive technique of spin-polarized low-energy-electron diffraction (SPLEED). For an in situ comparison with the corresponding bulk transition the magneto-optical Kerr effect was used. An interaction which is not included in the above model is that case when the perpendicular coupling between the topmost surface layer and the layer(s) below is different from J b . It is a further goal of this paper to investigate this question by means of spin-polarized photoemission. In fact, we find for the first time that the Af spins of the surface atoms of Gd(0001) undergo a so-called "magnetic surface reconstruction." 6 This observation has become possible by our making use of the surface-induced binding-energy shift of the 4/levels. 7 " 9The SPLEED technique as well as the spin-,

Temperature-induced changes in the electronic structure of Fe(100) have been investigated by spin-and angle-resolved photoemission for temperatures between room temperature and the Curie temperature T c . States nearly stationary in energy (T^r^) have been observed for photon energy hv = 60 eV. However, from a strong increase in minority-spin intensity for hv = 3l and 21 eV, a downwards shift of the AJ band is inferred to occur upon heating towards T c for large k vectors. PACS numbers: 75.50.Bb, 75.10.Lp, 79.60.Cn The electronic structure at finite temperatures of the 3d-transition metals Fe, Co, and Ni is currently a matter of strong theoretical interest. Spinpolarized band theory based on the self-consistent local-density-functional description gives an adequate account of the ferromagnetic ground state (e.g., cohesive energy, nonintegral moments). 1 However, controversial attempts have been made recently to describe transition-metal magnetism at finite temperatures. 2 "" 6 The basic common idea is to try to incorporate into the theory the existence of local magnetic moments even above T c . The ferromagnetic-to-paramagnetic phase transition is then governed by thermal disordering of the moments, requiring much less energy than singleparticle spin flips which would involve energy changes as large as the exchange splitting. The controversy is over the spatial extent of correlation among the magnetic moments, which is connected intimately to the present debate on the existence of spin waves above TQ?The ferromagnetic-to-paramagnetic phase transition of Fe has been studied by spin-unresolved, angle-resolved photoemission. 8 However, only by measuring the electron spin explicitly can exchange-split bands be identified unambiguously and the band dispersions be detected, as will be shown below. Furthermore, the spin dynamics at elevated temperatures, as spin rotations around the spontaneous magnetization direction or flips of local magnetic moments which currently are considered to be the driving force for the phase transition, can be observed only by this method. We have therefore, for the first time, performed a spin-, angle-, and energy-resolved photoemission experiment on temperature-induced changes in the electronic structure of Fe. Because of a predicted wave-vector (k) dependence of the temperature dependence of the exchange splitting, 5,9 we employed monochromatized tunable synchrotron radiation from the German storage ring BESSY, allowing selection of initial states with different k.The experiment is similar to a recent one on Ni(110) 10 using a resonance lamp. Total energy resolution, including the linewidth of light, was 0.4 eV at Ai/ = 60 eV and about 0.3 eV at Ai/ = 31 eV. The angular resolution was about ±3° at hv = 60 eV decreasing to about ±4° at hv = 31 eV, resulting in k resolution of about j-of the Brillouin zone. The sample was cleaned in situ by standard surfaceanalysis techniques and its surface conditions were monitored by low-energy electron diffraction and photoelectron spectroscopy. 11 T...

By spin-and angle-resolved photoemission with synchrotron radiation the electronic structure of Fe(100) has been tested between room temperature and the Curie temperature T, for photon energies in the range XI-70 eV. The spinresolved energy distribution curves (SREDCs) reflect the dispersions of the Ai'symmetry initial state bands. This manifests in an abrupt change in spin character of the peak near EF from predominantly minority spin to majority spin when tuning the photon energy across 33 eV. The non-spin-resolved EDCs thereby remain nearly unchanged. Upon heating to 0.85 T/T,, depending on photon energy, qualitative different changes in the SREDCs are observed: At hv = 60 eV, I'!& is found to be stationary in energy upon heating, and the spin-summed intensity decreases less than 5%. I?~~ becomes strongly broadened in energy and wave vector, resulting in a strong loss of intensity. Contrary, at hv = 31 and 21 eV, an increase in minority-spin (and total) photocurrent upon heating is observed. This is interpreted as resulting from a decrease of the exchange splitting with temperature near H.

No abstract

No abstract