VOLUME 80, NUMBER 8 PHYSICAL REVIEW LETTERS 23 FEBRUARY 1998 Observation of the Quantum Well Interference in Magnetic Nanostructures by Photoemission R. K. Kawakami, 1 E. Rotenberg, 2 Ernesto J. Escorcia-Aparicio, 1 Hyuk J. Choi, 1 T. R. Cummins, 3 J. G. Tobin, 3 N. V. Smith, 2 and Z. Q. Qiu 1 1 Department of Physics, University of California, Berkeley, California 94720 2 Advanced Light Source, Lawrence Berkeley National Laboratory, Berkeley, California 94720 3 Lawrence Livermore National Laboratory, Livermore, California 94550 (Received 4 November 1997) The CuCoNiCo(100) system was investigated by photoemission to study the interference between the Cu quantum well and the Ni layer. By varying their separation, we found that the density of states of the Cu quantum well states were biperiodically modulated. This result provides clear evidence for the quantum interference between two quantum wells in magnetic nanostructures. The biperiodicity was identified to correspond to the two Fermi vectors of the Co minority energy bands. [S0031-9007(98)05389-7] PACS numbers: 75.70.Ak, 75.30.Kz, 75.30.Pd, 75.50.Bb Investigations on layered magnetic nanostructures have developed rapidly in the last decade after the discoveries of the oscillatory magnetic coupling [1,2] and the giant magnetoresistance [3]. It is now generally believed that the essential physics in these layered magnetic nanostruc- tures is the formation of the spin-polarized quantum well states (QWS) due to the spin-dependent electron confine- ment [4–12]. QWS in metallic systems were first ob- served in the AgAu(111) system by photoemission [8]. Its connection to the magnetic interlayer coupling was later demonstrated in the CuCo(100) system [9]. While this type of single quantum well system has been at- tracting great interest [8–12], little is known about the interference between quantum wells in metallic systems. Although avoided-crossing behavior below the Fermi en- ergy has been observed in the nonmagnetic AgAuAg Au(111) double quantum well system [13], it is unclear how the quantum interference in magnetic systems will affect the magnetic coupling. Evidence of this kind of quantum well interference first came out from the mag- netic coupling measurements [14] which show that the antiferromagnetic coupling strength depends not only on the spacer layer thickness, but also on the ferromagnetic layer thickness. Based on this quantum interference ef- fect, Bruno developed an interlayer coupling theory [5] which shows that the magnetic coupling arises from the density of states (DOS) oscillations due to the electron waves reflected from all interfaces in the layered structure. In this quantum interference model, the observed oscilla- tions in the antiferromagnetic coupling strength as a func- tion of the ferromagnetic layer thickness should simply be a result of the DOS modulation due to the multiple electron reflections within the ferromagnetic layer. While this explanation is plausible, it relies on the DOS modu- lation as a function of ferromagnetic thickness, but a di- rect observation of this effect has not been made. In this Letter, we report a direct observation of the DOS modu- lation in the CuCoNiCo(100) system as a function of Co layer thickness. Using photoemission, we found that the DOS in the Cu layer is biperiodically modulated as a function of the Co layer thickness. This observation pro- vides strong support for the quantum well picture of the magnetic coupling. Photoemission currently provides the most direct ob- servation of the QWS below the Fermi level. The for- mation of QWS at discrete energy levels is manifested as peaks in the photoemission energy spectrum. To ob- serve the interference effect between two quantum wells, a careful thickness-dependent photoemission measurement with high signal-to-noise ratio is demanded because any slight variations could overwhelm this type of interfer- ence effect. We grew wedged samples to control the film thickness on the monolayer (ML) scale [15,16]. Wedged samples, however, are not commonly used in photoemis- sion experiments because of the difficulty to perform lo- cal measurements. The third generation Advanced Light Source (ALS) at the Lawrence Berkeley National Labo- ratory provides a special opportunity to overcome this dif- ficulty. The beam line 7.0.1.2 at the ALS can focus the photon beam down to a 50–100 mm spot size with a high enough photon flux (.10 12 photons per sec at resolving power of 10 000) to do the photoemission experiment. Thus, for a wedge of 10 MLmm slope, a scan of a 50 mm photon beam across the sample will provide a sys- tematic thickness-dependent measurement with 0.5 ML thickness resolution. The samples were epitaxially grown on a 10 mm diam Cu(100) single crystal substrate in ultrahigh vacuum (less than 4 3 10 210 torr). The substrate was prepared by mechanical polishing down to 0.25 mm diamond paste, chemical polishing (55 ml orthophosphoric acid, 20 ml nitric acid, 25 ml glacial acetic acid) [17], and in situ cleaning with cycles of 2–5 keV Ar 1 sputtering and an- nealing at 500 ± C. Photoemission measurements were performed using a hemispherical analyzer with normal photoemission at a photon energy of 83 eV. The total resolution (electron 1 photon) was better than 60 meV. The total angular acceptance was about 1.5 ± . Under these 1754 0031-9007 98 80(8) 1754(4)$15.00 © 1998 The American Physical Society