of the Less-Common
FROM RARE EARTH
G. LOUBRIEL Sandia National Laboratories,
NM 87185 (U.S.A.)
C. C. PARKS Lawrence Berkeley Laboratory, Berkeley,
(Received February 28,1983)
Summary Resonances in the photon-stimulated desorption (PSD) spectra of praseodymium, samarium, erbium, thulium and ytterbium oxides are reported at photon energies near their 4d edges. These resonances, which are also seen in soft X-ray absorption (SXA), arise from excitations of 4d electrons to the 4f shell. Comparisons of SXA and PSD show how PSD can be used to determine surface valency.
1. Introduction The electronic structure of rare earth elements and their compounds has recently been studied with renewed vigor [l-5]. Interest in these compounds stems from their use as hydrides and oxygen storage devices and from their mixed valency properties . The majority of recent studies of rare earth materials has focused on resonant processes seen in various spectroscopies such as photoemission [3, 41, appearance potential  and soft X-ray absorption (SXA) [S, 71. Our goal in this paper is to show how these resonances appear in photon-stimulated desorption (PSD) experiments [B, 91. We conclude that resonant PSD is an effective monitor of surface 4f level occupancy: it not only shows that the desorbed ion (in our case H+) bonds to a given rare earth atom but also gives the atom’s valency. For the problems of cerium and erbium oxidation we have previously shown that the erbium oxidation proceeds with a 3+ valency surface and bulk [l], while cerium may have (under certain oxidation conditions) a 3 + surface layer with subsurface layers of 4 + valency [Z]. *Paper presented at the Sixteenth Rare Earth Research Conference, The Florida State University, Tallahassee, FL, U.S.A., April l&%21,1983. 0022.5088/83/$3.00
0 Elsevier Sequoia/Printed
in The Netherlands
2. Experiments The PSD experiments were carried out using beam line I-l (4”) at the Stanford Synchrotron Radiation Laboratory which covers photon energies of 64-800 eV. The PSD ion measurement apparatus has been described elsewhere [S, 91. The total photoelectron yield (PEY) was measured using a spiraltron whose front end was biased positive. This PEY is used because it should be proportional to the bulk SXA. Both PSD and PEY measurements were normalized to the incident photon flux which was continuously monitored. For this study we prepared oxide samples starting from the powder oxides of praseodymium, samarium, erbium, thulium and ytterbium. These were mixed in a methanol slurry and deposited onto a tantalum film. For samarium we used a foil which was exposed to air to produce a thick oxide layer. The possibility that photoemitted electrons (mainly composed of inelastic electrons) themselves cause desorption processes as they cross the surface cannot be ruled out a priori. This mechanism would cause desorption yields of all ions to be identical and the desorption spectrum merely to reflect the absorption of photons deep inside the bulk of the material studied. We tested this possibility experimentally as follows. First we measured the total current escaping from the sample when no field was present. This current turned out to be (given reasonable cross section estimates) insufficient to cause the desorption signals observed unless the negative bias voltage (usually about - 700 V) used for measuring the positive ions deflected the electrons and caused a multiplicative effect on the current (by a factor of about 1000). If such were the case, lowering the bias voltage to about - 50 V would allow most of the electrons to escape and would reduce the ion yields. We observed an ion yield at about - 50 V which was within 10% of that at higher voltages.
3. Results of photon-stimulated
The SXAs of various lanthanides were measured for photon energies around the 4d threshold. The peaks observed are due to transitions of the type (4d)‘O(4f)“%
The multiplicity of peaks is the result of large spin-orbit splittings due to the exchange interactions between the 4f electrons themselves and between the 4f electrons and the 4d hole [lo]. The peak energies and intensities were calculated for many of the lanthanides using intermediate coupling and show that the number of peaks and their splittings are a strong function of the assumed 4f shell occupation (and perhaps valency) [lo]. The best example of this change in absorption with valency is afforded by the SXA of cerium [ll] which can exist as CeO, ((4f)‘) or as cerium metal and Ce,O, ((4f)‘). A large change in the SXA spectrum on going from CeO, to cerium metal is evident both in the number of peaks and in their energy splittings. The SXA of CeO, ((4f)O) differs less from that of La,O, ((4f)‘) than from that of cerium metal. All rare earths with the
same 4f occupation show similar SXA spectra [11-131. Thus SXA is an efficient technique for monitoring bulk rare earth chemistry. PSD spectra show the same resonances for praseodymium, samarium, erbium, thulium and ytterbium oxides, and therefore PSD probes rare earth surface chemistry in much the same way as SXA probes bulk rare earth chemistry. Previous work  has shown that the PSD H+ ion yield spectra from cerium oxides have features reminiscent of the SXA spectrum of cerium (in the loo-113 eV photon energy range) where H+ appears on the surface via segregation from the bulk. The PSD H+ ion yield spectrum after exposure to 750 langmuirs of oxygen at 300 K and subsequent heating to approximately 475 K is virtually identical with the SXA spectrum for cerium metal ((4f)‘) [ll]. If the surface is oxidized without heating, a new peak shows up in the PSD H+ ion yield in the region of the dominant SXA peak in CeO, ((4f)O) at 108.7 eV. The conclusion of Koel et al.  is that oxygen is being transported to the metaloxide interface resulting in a reduction of the surface to predominantly Ce,O,. Even at room temperature the surface has a mixture of 3 + and 4 + cerium atoms. Such a strong surface effect was not observed for erbium oxidation [l], presumably because erbium is not an efficient ionic conductor . Figure 1 shows the SXA of oxidized thulium  and the PEY and PSD of H+ ions. All three curves exhibit three peaks of similar intensity ratios and energy spacings. This agreement shows that the initial process which leads to H’ desorption at the surface and electron emission from the bulk is photon absorption by thulium atoms in a 3 + state. Specifically, H+ ions at the surface are bonded to thulium atoms in a (4f)12 configuration. Small disagreements in the peak intensities and peak energies can probably be traced to variations in the photon energy resolution and to problems in monochromator calibration which in our case only amount to an error of 0.5 eV. A third possible cause for
Fig. 1. Comparison of the PSD H + ion yield, the PEY from Tm,O, and the SXA of thulium oxide from ref. 13. These are typical absorption spectra of (4f)” compounds.
shifts in the energy is shifts in the surface core level. Indeed, our peak energy of 171.1 eV for thulium differs by at most 0.5 eV from those reported using SXA spectroscopy (171.6,174.5 and 178.5 eV) . Comparison of Tm,O, spectra with those of Yb,O, and Er,O, (Fig. 2) shows that neither a (4f)’ ’ nor a (4f)’ 3 configuration is possible for the thulium surface. As in the case of thulium, the spectra for erbium and ytterbium show
PHOTON ENEROY teV) .. :
.* . ..-.. *...*
PHOTON ENEROY (ev)
110 PNOTON ENEROY (eV)
.** . ..
. *. ..
Fig. 2. The PEY and PSD H’ ion spectra of (a) Er,O, and (b) Yb,O,. Fig. 3. The PEY and PSD H’ ion spectra of (a) Pr,O,, PEY of Pr,O, and Sm,O,.
and (c) pre-edge structures in the
excellent agreement between PSD and PEY which implies the same valency in the surface as in the bulk ((4f)” for erbium and (4f)13 for ytterbium). The peak energies for erbium occur at 163.0,164.8,167.0,170.7 and 174.2 eV and are only slightly shifted from those reported using SXA. For erbium we also observe a sharp edge (0.5 eV total width) at 162.8 eV. The ytterbium spectrum shows a main peak at 180.6 eV with a small peak at 172.4 eV. The SXA spectrum of ytterbium shows peaks at 179.6 eV and a peak slightly larger than ours at 171.3 eV. Figure 3 shows the PEY and PSD of H ’ ions from praseodymium and samarium oxides. These spectra differ from those previously shown in that small pre-edge resonances can be observed. These resonances are expanded in Fig. 3(c) and occur at photon energies of 147-132 eV in samarium and 106-120 eV in praseodymium. The pre-edge structure is an excellent indicator of the valency of the rare earth atom being probed. The low ion signal in our PSD experiments did not allow comparison of the pre-edge structure with that observed in PEY, although the PEY agrees well with SXA experiments [ll, 131. For praseodymium we find pre-edge structures at 106.2,107.2,108.1,108.4,109.7,110.2,110.6, 111.1, 112.4, 113.6, 114.9, 115.3, 116.1, 116.7, 116.9, 117.5 and 118.5 eV and large peaks at 123.6 and 131.0 eV. In view of our good photon energy resolution (about 0.12 eV) we are the first to show some of these structures (see Fig. 3(c)). For samarium we find pre-edge structures at 126.1,127.1,127.6,128.6,129.9,130.5 and 131.7 eV and large peaks at 136.7, 140.0 and 147.9 eV which have shoulders at 133.0,133.4,135.0 and 136.7 eV. Again, agreement between PSD and PEY show similar oxidation states in the surface and in the bulk for samarium and praseodymium oxides.
4. Conclusion We have shown the usefulness of comparisons between PSD, PEY and SXA spectra to gain chemical insight into specific local hydrogen bonds. This study demonstrates that, in contrast with cerium oxides, the oxides of praseodymium, samarium, erbium, thulium and ytterbium have the same surface and bulk valency. We observe significant H+ ion yields from the surface whose spectral dependence gives the 4f occupancy of the rare earth atom to which they were bonded.
Acknowledgments We wish to acknowledge useful discussions with B. E. Koel and M. L. Knotek and the technical assistance of J. M. Borders. Experiments were conducted at the Stanford Synchrotron Radiation Laboratory which is supported by NSF Grant DMR-77-27489. This work was supported by the U.S. Department of Energy at Sandia National Laboratories under Contract DEAC04-76DP-00789.
M. L. Knotek, R. H. Stulen, B. E. Koel and C. C. Parks, J. VW. Sci. 1145. B. E. Koel, G. M. Loubriel, M. L. Knotek, R. H. Stulen, R. A. Rosenberg and C. C. Parks, Phys. Rev. B, 25(1982) 5551. A. Zangwill and P. Soven, Phys. Rev. Lett., 45 (1980) 204. J. W. Allen, L. I. Johansson, I. Lindau and S. B. Hagstrom, Phys. Rev. B, 21(1980) 1335. D. Chopra, G. Martin, H. Naraghi and L. Martinez, J. Vuc. Sci. Technol., 18(1981) 44. W. Gudat and C. Kunz, Phys. Rev. Lett., 29 (1972) 169. H. W. Wolff, R. Bruhn, K. Radler and B. Sonntag, Phys. Lett. A, 59 (1976) 67. M. L. Knotek, V. 0. Jones and V. Rehn, Phys. Rev. Lett., 43(1979) 300; Surf. Sci., 102 (1981) 566. M. L. Knotek, Surf. Sci., 91(1980) L17; lOl(l980) 334. M. L. Knotek and P. J. Feibelman, Phys. Rev. Lett., 40 (1978) 964. R. Haensel, P. Rasbe and B. Sonntag, Solid State Commun., 8 (1970) 1845. J. Sugar, Phys. Rev. B, 5(1972) 1785; Phys. Rev. A, 6(1972) 1764. E. A. Stewardson and J. E. Wilson, Proc. Phys. Sot., London, Sect. A, 69 (1956) 7. V. A. Fomichev, T. M. Zimkina, S. A. Bribovskii and I. I. Zhukova, Sov. Phys.-Solid State, 9(1967) 1163. G. D. Mahan and W. L. Roth (eds.), Superionic Conductors, Plenum, New York, 1976. G. Loubriel,
Technol. A, l(l983)
2 3 4
9 10 11 12 13 14