J. MATER. CHEM., 1991, 1(4), 489-502 FEATURE ARTICLE Photoemission in the Study of Oxide Superconductors Mark S. Golden," Russell G. Egdellb and Wendy R. Flavell*" a Centre for High Temperature Superconductivity, Department of Chemistry, Imperial College of Science, Technology and Medicine, South Kensington, London SW7 ZAY, UK Inorganic Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QR, UK " Department of Chemistry, UMIST, PO Box 88, Manchester M60 IQD, UK Photoemission has the potential to yield important information about the electronic structure of high-temperature superconducting oxides. However, the surface specificity of the technique, combined with the chemical complexity of these oxides can lead to considerable problems in data analysis and interpretation.Here we review photoemission results for c-axis-oriented thin films of Bi,Sr,CaCu,O,-based materials, where the surfaces are prepared for photoemission studies by in situ oxygen annealing. The current status of photoemission studies of high-T, oxides is surveyed, highlighting areas of controversy. Keywords: Oxide superconductor; Photoemission; Feature article 1. Introduction Following the discovery of high-temperature superconduc- tivity in a number of families of complex metal oxides, there have now been at least 500 investigations of these materials by electronic spectroscopies such as photoemission (XPS and UPS), inverse photoemission (IPES), X-ray absorption, Auger electron spectroscopy and electron energy loss.A great deal of attention has been focussed on photoemission studies in particular, as the technique has the potential to give detailed information about the filled band structure of these materials. Currently, there is considerable interest in detailed studies of the density of states (DOS) at the Fermi energy, both in terms of the chemical character of these states, and the way in which they evolve through the superconducting transition tempera- ture, T,. These studies are valuable as, at the very least, they place constraints on possible theories of superconductivity in these oxides. One important feature of photoemission is that it is intrin- sically surface sensitive. The photoelectron flux generated within the material is attenuated as it leaves the sample in accordance with a Beer-Lambert-type law, characterised by an electron mean-free-path length of only ca.10-15 A.' Thus the technique may be used powerfully to probe processes occurring at the surfaces of the new materials, for example reaction with atmospheric gases, or surface segregation of dopant or impurity atoms. For complex metal oxides of this type, it is common for the surface composition and electronic structure to be quite distinct from that of the bulk, owing to perturbations caused by the bulk termination created by the surface.2 This feature of the technique gives photoemission studies an added importance, since, in any working device, contacts and interfaces will have to be made to the supercon- ductor surface.It seems likely that the intrinsic surface proper- ties of the materials will ultimately determine the scale of their application. Unfortunately, the surface sensitivity of photoemission leads to a number of problems in data interpretation, and has led to a number of continuing controversies within the literature. In particular, it has now been evident for some time that all families of oxide superconductors are to some extent subject to atmospheric degradation reactions when in contact with air containing water ~apour.~-" Although the extent of the problem varies from material to material, the nett result is always the creation of a number of extraneous insulating phases at the surface. Given the extreme surface sensitivity of photoemission, this adventitious contamination must have a strong influence on the spectra obtained.It is clear that very careful attention must be paid to surface preparation and Cleaning before meaningful spectroscopic measurements can be undertaken. The surface preparation methods used to date fall into two categories. The first of these involves the creation of a new surface in ultra-high vacuum (UHV), using techniques such as cleavage, abrasion or ion milling. Alternatively, existing external surfaces may be studied, using in situ oxygen annealing to reverse surface degradation reactions. This is the procedure that we have favoured in most of our work in this area.4,11,12 The normal preparative conditions for ceramic, thin-film and single-crystal materials all involve extended oxygen annealing treatments, and the in situ cleaning procedure mimics the conditions of these.Whereas the electronic structure revealed by cleaved surfaces may be more representative of the bulk, annealed materials provide a more accurate model for the surfaces typically encountered in device manufacture. In a previous paper4 we reviewed our results using in situ oxygen annealing for a range of systems, and compared them to results reported elsewhere in the literature at the time. The aim of the present paper is to update this review, drawing attention to results obtained since publication of our earlier paper. Studies published elsewhere in this period will be surveyed, and areas of controversy highlighted. 2.Surface Ageing Reactions of SuperconductorSurfaces Like many other oxides, high-temperature superconductors are susceptible to reaction with atmospheric C02 and H20 to give surface carbonate and hydroxide Extraneous features arising from this atmospheric degradation are always evident in spectra taken from 'as-presented', un- cleaned oxide superconductor surfaces. It is therefore rather surprising that there have been so few studies dedicated to the investigation of these reactions by photoemission.3~13-19 The degradation reaction of YBa2Cu307 is perhaps the best stUdied,3-6.10,14,15,17-20 with the dominant reaction now established as 2YBa2Cu307+ 3C02 +Y,BaCuO, + 3BaC03+ 5CuO + at the synthesis temperature of 950 0C,496 possibly changing to 2YBa2Cu307 +4C0,-+4BaC03 +Y2Cu205 +4CuO+3O2 at lower temperatures.6 These reactions are strongly catalysed by the presence of water ~apour.~ The presence of degradation products at the surfaces of these oxides has a very strong effect on both core-level and valence-level photoemission ~pectra.~.~ In the case of core- level spectra, the effect is probably most dramatic on the shape of the 0 1s core-level peak; this has contributed con- siderably to controversy in the literature concerning the detailed interpretation of this peak shape, and is considered further in sections 3 and 4.In the valence band spectrum, contaminant-related features appearing at 9-10 eV below the Fermi energy EFhave also been the subject of heated debate; these are considered further below and in section 5.The build-up of the insulating contaminant phases may be rapid on the photoemission depth scale under ambient conditions; a contaminant layer of ca. 12 A thick develops on the surface of YBa2Cu307 ceramics after only ca. IOmin of air exp~sure.~ However, the degradation of some of the other oxide superconductors such as YBa2Cu40810 and Bi,Sr,CaCu208 (where the dominant products appear to be SrC03 and CaC0313*21 are apparently not so rapid. Superconductor/Ag composite materials, such as (YB~,CU~O~)~-xAg, exhibit enhanced resistance to degra- dation." The possibility of passivating the surfaces of these oxides using Ag, Cu or Au overlayers has now been investi- gated in a number of photoemission studies of precious-metal deposition on single-crystal surfaces.22-26 These studies indicate a contrast between the surface reactivity of Bi,Sr,CaCu208 (BSCCO) and related materials such as Bi2Sr2Cu06, and that of YBa2Cu307-, (YBCO) and its homologue EuBa2Cu307 -,.In the case of BSCCO materials, deposited Ag,,, Cu2, or Au23,24 causes only weak reaction with the Bi-0 (001) cleavage plane (the preferred cleavage plane for the BCSCO structure). However, similar deposition on YBCO materials causes more extensive disruption of the surface electronic stru~ture,~~,~~,~~ and the metallicity of EuBa,Cu307 -,in the surface region is essentially de~troyed.,~ As the degradation reactions are catalysed by water vapour, important information about the initial stages of the reaction may be gained by valence-band photoemission studies of water adsorption.Very little experimental work has been done in this area. Early studies of polycrystalline La, -,Sr,Cu04 indicated that water is dissociatively adsorbed at room temperature,I6 while studies of polycrystalline YBa2Cu307-, indicated that water is chemisorbed at low temperatures, becoming dissociated when the surface is warmed to 300 K.'4315These findings are in line with what might be expected from studies of other perovskites, such as SrTi03.27 One study of water adsorption on single-crystal Bi2Sr2CaCu208(001) has been carried out using synchrotron radiation (Fig. 1).Again, the reactivity of the Bi-0 planes making up the natural cleavage face of the (001) surface appears to be very low. At a temperature of 90K and low coverage, water remains essentially physisorbed (rather than chemisorbed), as evidenced by the absence of any bonding shift of the 3a, molecular orbital of the adsorbed water (Fig. I). (In cases of chemisorption of water on metal oxides, the 3a, level is shifted to higher binding energy relative to the Ib, and 1b2 levels by as much as 1.3 eV.13) Studies of adsorption on single-crystal YBa,Cu307 -,surfaces, and further studies of BSCCO materials at a range of temperatures J. MATER. CHEM., 1991, VOL. 1 12 8 4 EF binding energylev Fig. 1 Photoemission spectra of (a) water-dosed and (b) clean Bi,Sr2CaCu20, (001) at 90 K, recorded at near-normal emission using 33eV photon energy.(c) The difference spectrum (2 L H20-clean) was obtained by normalising the intensity away from the adsorbate-induced features. Vertical ionization energies for gas- phase water3' are also shown, aligned at the 1b2 peak positions. Note the absence of any bonding shift of the 3a, level away from its position relative to Ib, in gas-phase water (from Flavell et a1.I3) are necessary to investigate this apparent difference in re-activity further. However, there appear to be several plausible explanations for the difference. The (001) surface of Bi,Sr2CaCu208 cleaves easily, resulting in a surface termin- ated by relatively defect-free, planar Bi-0 1a~ers.I~This is in contrast to other less anisotropic high-T, materials such as YBa2Cu307-x, where stepped (and hence highly reactive) surfaces may be produced.28 By analogy with the low surface reactivity of the ruthenates of bismuth and lead (Bi,Ru207 and Pb2Ru207-J towards water,29 it may also be possible that the surface Bi"' cations possess stereochemically active sp-hybrid lone pairs of electrons protruding from the surface.This would make the surface particularly inert towards a species such as H20, which would tend to be adsorbed via its oxygen atom at a Lewis-acid site. 3. Surface Preparation for Photoemission Studies Although photoemission may be a useful tool in investigation of the surface reactions of superconducting oxides, it is clear that any adventitious degradation products must be removed from the surface before photoemission can be used to probe the intrinsic band structure of the materials.In our previous review, we surveyed the cleaning techniques which have been applied to these oxide surface^.^ These techniques fall into two categories. The first involves the creation of a new surface in UHV. One common UHV cleaning technique which might be used to achieve this is argon-ion bombardment. Unfortu- nately, as with all complex oxides, this has the effect of irreversibly changing the surface cation ratios, and in the case of copper oxide superconductors reduces surface copper cat- ions to CU'.~ Other techniques that have been used to create new surfaces include surface abrasion (of ceramic^^,^^-^^, single crystal^^'.^^ or thin films3,), fracture of ceramic bar^,^^.^^ and peeling or cleavage of single ~rystals.~,~~-~~ In some cases, the new surface created has been found to degrade rapidly, and to minimise this surface preparation has been carried out at low temperatures. The two most important instances where this has been necessary to date are studies of Nd2-,Ce,Cu04 and YBa2Cu307-,.In the case of Nd2 -,Ce,CuO,, several authors have now found it necessary to cool the samples to temperatures between 20 and 100 K J. MATER. CHEM., 1991, VOL. 1 before cleaving single crystals38 or scraping ceramic^.^^-^^ In the case of YBa,Cu307 --x, some authors have reported degra- dation of cleaved single-crystal surfaces even at quite low temperatures (ca.50 K).4 In some recent studies, YBa2Cu3O7--x crystals have been cleaved at temperatures as low as 8 K, in order to maintain a representative ~urface.~~~~~ 49 1 . .The exact cause of surface deterioration in this case remains rather uncertain, and will be discussed further in section 5. Interestingly, the same authors have reported the (001) surface of Bi2Sr2CaCu208 to be considerably more table:^^^^ in line with the observations of the previous section. We have previously discussed the advantages and disadvan- tages of methods of surface preparation which involve the creation of a new ~urface.~Perhaps the most significant disadvantage is that all these techniques are destructive, and are thus rather difficult to apply to plate-like single crystals, or thin films.Most of our own work has been conducted using an alternative preparation technique, designed to elimin- ate this problem. This involves reversing the effects of atmos- pheric degradation by in situ annealing under conditions chosen to mimic those used in the synthesis of the samples. Our experiments have been conducted in a two-chamber VG Escalab Mark I1 spectrometer. Samples are mounted on platinum stubs using platinum clasp wires and transferred to the spectrometer preparation chamber, which is filled to 1 bar with pure oxygen. Samples are then subject to an annealing and slow cooling cycle by r.f. induction, using a Radyne 400 kHz, 1.5 kW generator.The process has been described in more detail el~ewhere.~ The efficacy of this cleaning procedure may be gauged by monitoring the C 1s:O 1s intensity ratio in XPS. The efficiency of the procedure varies from substrate to substrate; in BSCCO- based materials, we have found that it is possible to reduce the C 1s signal below levels of dete~tability,~.'~,~~,~~whilst in the case of ceramic YBa,Cu,O,-,, it may only be possible to reduce the ratio to below 1:100.4," The technique has been particularly successful when applied to thin films of the Bi2.1Sr1.7Ca0.85-xYxcu208 (Y-BSCCO) and related techniques have been applied by other authors to thin 523 526 529 532 535 538 binding energylev Fig.2 The effect of sample cleaning by oxygen annealing on the 0 Is peak shape in Al-Kcr XPS (hv=1486.6eV).(a) Biz.lSr,.,Cao.425Yo.,25CU208+6thin film before cleaning by in situ oxygen annealing. (b) As in (a), after cleaning, revealing the intrinsic peakshape due to the superconductor (from Golden et ~1.~') contact must be made for electrical measurements. However, there may be significant differences in chemical composition and electronic structure between cleaved surfaces and post- annealed external surfaces. This may be due to intrinsic segregation of one cation to the surface (e.g. Sr in Laz -,S~,CUO~~) or purely due to the apparently very different films of YBa2Cu30,- +x52-55 and Nd, -,C~,CUO~.~~.~~ reactivities Contaminated surfaces display a strong, high-binding energy 0 1s component.This is chemically shifted by ca. 2-3 eV to higher binding energy relative to the 0 1s peak at 528-529 eV associated with the superconductor phase. For all the substrates which we have studied (including YBa,Cu307 -x,ll La, -,SrxCu04,4 and Bi,Sr,Ca,-'Cu,,04+2,,"), the high-binding-energy peak shows enhanced intensity in grazing-emission experiments, thus demonstrating its surface origin. This feature is also found to increase in intensity as a function of length of exposure of the sample to the ambient atm~sphere.~ As an example, the 0 1s peak shape for a sample of Y-BSCCO with x=O.425 is shown in Fig. 2, to illustrate the effect of the oxygen cleaning procedure. The contaminant- related high-binding-energy component of the peak shape is completely removed during oxygen annealing, leaving a single, slightly asymmetric peak.In the case of YBCO, a small shoulder remains in this region after cleaning, which may be intrinsic to the superconductor itself." The detailed structure of the 0 1s envelope in XPS in superconducting oxides continues to excite debate31,32,39,54*58-63 and will be discussed further in section4. However, we would maintain that the observation of appreciable structure beyond 531 eV is merely indicative of a significantly contaminated surface. Oxygen-annealed surfaces are of immediate interest in relation to technological application of oxide superconductors, as they represent 'intrinsic' free surfaces of samples after furnace preparation, and it is generally to these surfaces that of equilibrated annealed surfaces and freshly cleaved surface^.^ This effect has been noted previously for the perovskite SrTi03.27,64 In general, we have found surfaces prepared in this way to be stable in UHV for prolonged periods (in some cases, several days).4. Core-level Studies The major interest in core-level photoemission lies in the ability of the technique to distinguish different valence states and chemical environments of atoms, using the resulting chemical shift differences in binding energy. One of the most important questions is then how much core-level studies can tell us about the formal valency of copper in the Cu-0, planes of the copper oxide superconductors (or the valency of bismuth in the Bi-based materials).We will discuss this in section 4.2. First, however, we will consider studies of the 0 1s core levels, as here our conclusions are clearly influenced by comments in the previous section. 4.1 0 1s Core Levels In our earlier paper, we surveyed 0 Is core-level studies carried out up to late 1989.4 The discussion here will centre on work which has appeared since this date, although the controversy surrounding the shape of the 0 1s peak does not appear to have been in any way resolved. Many published spectra show two components of variable intensity ratio for 0 1s (Fig. 2),4l3l with some as-presented surfaces displaying more complex structure.54 It is widely accepted that structure at ca.528-529eV is intrinsic to the oxide superconductor, whereas much of the intensity at ca. 531 eV can be attributed to an extrinsic, surface component,23,2426,32.36.39,43,45$54,65-7 1 whose size is dependent on a variety of surface treatment^,^' and is enhanced in grazing emissi~n.~,~'*~~ The exact origins of this surface component have not been defined precisely; possible suggestions for the YBa2Cu307-x material are BaC0333491 formed as a result of atmospheric degradation, Ba(OH)2,39 Ba04*39 formed by the decomposition of BaC03 during surface treatment, B~CUO~,~~or a BaCu0,-like phase.54 In the case of BSCCO materials, the most likely candidates appear to be SrC03,4*21*71 CaC03,21 or SrO formed by decomposition of SrC03 during surface treat- ment4v7' However, the surface contaminant feature appears to be present in spectra of uncleaned surfaces of less well J.MATER. CHEM., 1991, VOL. 1 removed from the surface, leaving a single feature (e.g.Fig. 2), this peak is generally found to be somewhat asymmet-ric.4y16,24v39,43,45,49*66Calculations show that the strong Cu 3d-0 2p covalency in the Cu-02 planes may give rise to several possible final states (in the case of CuO correspond- ing to 'well screened' 101s' Cu 3d9 Lo> and 'poorly screened' 101s' Cu 3d" L1>(where L refers to ligand oxygen, and annderscore denotes holes), with the latter appearing at ca. 1.5 eV higher binding energy than the main Thus, weak high-binding-energy structure may be intrinsic to cuprate phases. 4.2 Cu 2p Core Levels By comparison, the interpretation of Cu 2p features from copper oxide superconductors is reasonably uncontroversial, studied materials; these include Nd2-xCexCu04,36*45*68,69and the lineshapes of these peaks may be used as an indication (Tlo~5Pbo~s)Sr2(Cal-xThx)Cu20,, 67 and Ba, -,RbxBi03.46 In these cases, the nature of the contamination has been less well identified.Most oxide superconductors contain oxygen in a number of different lattice sites. It might be expected that these would give rise to 0 1s signals at differing binding energies. However, calc~lation~~and O-Ka X-ray emission spectra73 have shown that distinct oxygen sites differ in binding energy by at most 1.5eV, so that in the normal XPS experiment, with resolution of ca.1 eV, these contributions to the intrinsic '528 eV' super- conductor peak are not distinguishable, and a single, rather broad peak is ~een.~',~~However, in recent studies by Parmigiani et ~l.,~'using a monochromated Al-Kcc source and low analyser pass energy to study oxygen-annealed Bi2Sr2CaCu208 single crystals, a resolution of 0.35 eV was obtained, enabling two components of the bulk superconduc- tor feature to be resolved. The studies indicate that oxygens close to the Bi-0 planes have a slightly higher binding energy than oxygen atoms in Sr-0 and Cu-0, layers,61 roughly in accord with calc~lation.~~Interestingly, when crystals are re-annealed in a high overpressure of oxygen (12 atmt rather than 1 atm), a new feature which is enhanced in grazing emission appears at 531.2 eV.The authors attribute this to surface oxygen. Specific surface oxygen species, such as superoxide and peroxide have been proposed previously to explain structure at higher binding energy than the bulk superconductor fea- t~re.~*~'Recent studies by Qiu et al. on the reaction of alkali metals with oxygen have indicated that assignment of this structure to peroxides may be possible, but that superoxide appears at substantially higher binding energy (ca. 535 eV).59*74 An alternative interpretation of 0 1s core-level structure is given by Balzarotti et a1.31,62,63He re, reduction in intensity of the 531 eV feature on vacuum annealing YBa2Cu30,-, is attributed to stepwise loss, first of contaminant oxygen, but predominantly of the chain oxygen atom^.^',^^ Re-annealing in oxygen is found to restore the 531 eV peak.In the case of Bi2Sr2CaCu208, these authors contribute a 53 1.5 eV peak to oxygen in the Bi-0 and Sr-0 planes, and a third compo- nent, centred at 533.5 eV, to adsorption of adventitious CO in UHV. Detailed measurements from single crystals as a function of photoemission take-off angle are necessary to resolve these controversies. However, we would point out that parallel measurements of the Cu 2p core levels from our own oxygen annealed surfaces (showing low intensity at 531 eV) indicate that these surfaces are not oxygen deficient (section 4.2, be lo^).^^^^^^' Even when extraneous surface phases or species can be 1 atmx101 325 Pa.of the formal oxidation state of the Cu-02 planes in both hole-doped4,23.24.26.31,33,35,39,50,51,62,6S,67,7S-78 and electron- doped materials.36,43,45,68*69As an example, in Fig. 3 we show the Cu 2p3,, multiplet for c-axis oriented thin films of Biz.lSrl.7Cao.8s-xYxcu208 (Y-BSCCO) for x=O (metallic,m) and x =0.85 (non-metallic, n) corn position^.^^^^^ The x= 0 material is a hole-doped superconductor, with a formal copper valency of 2.3, whereas the x=0.85 material has a corresponding Cu valency of 1.9,i.e. essentially a Cu" material. The major features of both spectra are similar to those of CUO,~~and show the familiar double-peaked structure com- mon to all high-TT, copper oxides.This arises from strong m n 929 933 937 941 945 949 binding energy/eV Fig.3 Al-Ka XPS profiles in the Cu 2p,,, region for Bi2,1Srl,7YxCal-xCu208+d thin films for metallic (m, x=O) and non- metallic (n, x=0.85) samples. The difference spectrum (d, m-n) is obtained after normalisation of the areas of the complete 2p,,, structure and alignment of Bi 4f or 0 1s core-level peaks. AI-Ka,., satellite structure has been subtracted. The horizontal lines indicate FWHM, and the vertical line indicates the peak maximum in the metallic (m) spectrum (from Golden et ~1.") J. MATER. CHEM., 1991, VOL. 1 Cu/O mixing in the initial state, allowing for valence electron transfer from the 0 2p to the Cu 3d levels during photoionis- ation.In CuO, the main peak at ca. 933 eV is due to the well screened 12p"d"L' > final state, and the satellite centred at 493 significant intensity shift to lower binding energy, with the difference spectrum of Fig. 3 (d; m-n) showing a peak at 945 eV. In the case of YBa2Cu307-,, which is prone to oxygen ca. 942 eV% due tothe unscreened 12p '3d9L0 > final ~tate.~,~~ loss, much lower satellite intensities may be obtained,35 for The satellite is subject to final-statemultiget splitting into a total of eight closely spaced states,80 giving a complex peak structure. The observation of more than one final state in XPS indicates that the core-valence Coulomb interaction in the final state is sufficiently large to localise the electrons on copper.This behaviour is characteristic of the oxides at the end of the first transition-metal series, and reflects the fact that electron correlation in the initial state is strong; CuO, NiO, COO, FeO and MnO are magnetic insulators. Obviously in the case of copper oxide superconductors, the initial state is itinerant, rather than localised, but core-level studies show that correlation forces are nevertheless large, and the oxides can generally be made to undergo a metal-to-non-metal transition by quite subtle changes in chemical doping." [Note that core-level studies of this type give us information primar- ily about the final-state configurations; the distribution of the itinerant electrons in the initial state (e.g.whether the holes in p-type materials are on copper or oxygen) cannot be probed; only aforrnal Cu oxidation state is obtained.] In the case of a Cu' material, such as Cu20, the only accessible final state is 12p13d"L0 >, and no satellite feature is observed.79 Converselzfor C$", e.g. in N~CUO~,~'-'~ and L~CUO,,'~" three states are accessible: 12p '3d"L2 >, )2p '3d9L1 > and )2p '3d8L0 >. XPS shows thacn thiscase, G"' is-manifest mainly as the 'two-ligand-hole' state, 12p '3d"L >, appearing at the high-binding-energy side of the main933 eV peak seen in the CuO spectrum. From Fig. 3, we can see that in going from the metallic (m) Bi2~lSrl,7Cao~8sCu208+ato the non-metallic (n), fully Y-sub- stituted material Bi2~lSrl~7Yo~85C~208+6, we see a transfer in spectral weight from the high-binding-energy side of the 12p '3d"L" > peak at 933 eV to the low-binding-energy side, together-with an overall narrowing of this peak.We have previously shown that this is consistent with the formal change in copper oxidation state, and hence the absence of the 12p'3d''L2> final state in the case of the non-metallic material?' Parallel effects are observed in valence-band photoemission (section 5). Similar trends have been found for YBa2Cu307-,, where the formal oxidation state is >2 in the superconducting state with x <0.5, and <2 in the non-metallic Itstate with x >0.5.4,26,39,5',65,75.77 now seems clear that core-level XPS is able to provide a measure of the valence- band hole concentration in hole-doped materials.Lindberg et have pointed out that the mean-free-path length of Cu 2p electrons is extremely small, (as low as 6 in Bi2Sr2CaCu208) owing to their low kinetic energy, so that the Cu 2p spectrum is a very sensitive test of surface stoichiometry. Low intensity in the satellite feature at 942eV may often be indicative of some surface reduction to as no satellite feature is expected for Cu20. In the case of hole-doped materials, which should give 'Cu"-like' and 'Cu"'-like' final states, this interpretation is not entirely without pitfalls. This is because Cu"' compounds, unlike Cu" materials, also appear to give very low satellite inten~ity.~~.~'~'~ Thus, increasing hole concentration may in some cases lead to decreasing satellite inten~ity.'~' Combined XES and XPS studies78 have indicated that the 'Cu"'-like' (2p '3d9L1 > contribution to the satellite feature is a low-intensity peak centred at 945 eV.Thus in the case of Y-BSCCO, where we have a nett Cu valency change from 2.3 to 1.9, we see a minimal change in satellite intensity (in fact a slight increase, as the non-metallic material is very close to a Cu" oxide)." However, there is a example by heating in uacu0,4*51i75indicative of significant reduction to Cu'. (On heating in uacuo, the composition obtained is essentially YBa2Cu306, corresponding to CU'.~~, ref. 51.) In a recent paper, Fowler et associated low satellite intensity with poor-quality YBCO surfaces showing large amounts of reduction, and Yeh et al.have in turn correlated this with poor superconducting transition tempera- tures in thin films.35 In principle, interpretation of satellite structure should be more straightforward in the case of electron-doped materials, such as Nd,-,Ce,CuO,, as the only formal oxidation states involved here are Cu' and Cu". The satellite intensity should decrease regularly with the amount of hole doping, i.e. with increasing Ce concentration. Studies of this material tend to indicate that the satellite intensity in Ce-doped compositions is indeed lower than that in Cu0.36,43,45,68369As-grown samples of this material do not superconduct, and must undergo a reducing annealing cycle before a superconducting transition is obtained.In general, as-grown, unreduced samples appear to show a satellite intensity which decreases with increasing Ce-doping leve1.36,43*45 However, the decrease is less marked than that which would be expected for the occupation of the Cu 3d orbitals by all doped electron^.^^,^^^^^ After reduction the satellite intensity is more in keeping with the model of one doped electron per introduced Ce,36,43,45 although the presence of extraneous Cu'-containing surface phases caused by the reduction cannot be ruled o~t.~~'~~ 5. Valence-level Studies Photoemission has been very widely used to study the elec- tronic structure of the filled density of states in oxide supercon- ductors in the valence-band region and close to the Fermi level.Particular interest has focussed on the nature of satellite structure in valence-region photoemission (which reflects elec- tron correlation), and on the electronic states close to the Fermi energy (as these are the states primarily involved in superconducting behaviour). Experiments have been carried out using both conventional, fixed-energy photon sources (noble-gas discharge lamps and soft X-ray guns), and tuneable synchrotron radiation. Changes in intensity of spectral features with changing photon energy reflect changes in atomic ionis- ation cross-sections and are of value in helping to identify the atomic nature of states responsible for the emission. Synchro- tron studies are particularly valuable here, as resonant enhancement of these cross-sections at core excitation thresh- olds may be exploited using tuneable energy radiation.How- ever, the superior resolution attainable with a noble-gas discharge lamp has meant that studies using conventional sources are very important in monitoring slight changes occurring at the Fermi energy when a sample is cooled through the superconducting transition temperature T,. Valence-region studies using either synchrotron radiation or a discharge lamp source generally involve low-energy electrons. The momentum resolution possible is then generally sufficient to locate a transition to within a small fraction of the typical dimension of a Brillouin zone. This opens up the possibility of mapping electronic structure using angle- resolved photoemission, in cases where high-quality single crystals are available.Measurement of the extent of dispersion along various zone directions allows for detailed comparison with theoretical models of electronic structure. In the case of the oxide superconductors, both angle-integrated and angle- resolved photoemission have been used extensively. 5.1 General Features of the Valence-band Region In order to illustrate the general features of the valence band structure of these materials, in Fig. 4 we show valence- region spectra for oxygen-annealed, c-axis-oriented Bi2~3Srl~3Cal~oC~2.008+b thin films (a single-phase, two-layer '(2212)' BSCCO-material with a T, of 85 Kg5). These are taken with three different conventional laboratory sources, He I (hv=21.2 eV), He I1 (hv=40.8 eV) and Al-Ka (hv= 1486.6 eV).The He I spectra are dominated by a steeply rising background of secondary electron emission beyond ca. 10 eV binding energy, whereas the higher-energy sources allow the obser- vation of shallow core-level structure (in this case Sr 4p at ca. 18 eV binding energy). The main valence band region, which is composed of 0 2p and Cu 3d levels (ca. 1.5-7 eV binding energy), changes very little in shape on raising the photon energy from the UV to the soft X-ray range. Over this photon energy range, the 0 2p:Cu 3d cross-section ratio decreases from ca. 0.35 to 0.08.4,11*75The Al-Ka spectrum thus reflects predominantly the Cu 3d density of states, and indicates that 15 1'0 5 6 binding energy/eV Fig.4 Valence-region photoemission spectra of c-axis-oriented thin- film Biz,3Srl &al,oCuz.oO, (001) ['(2212)'-phase material] excited with (a) Al-Kcr (hv= 1486.6 eV), (b) He I1 (hv=40.8 eV) and (c) He I (hv=21.2 eV) radiation. Note the increase in the intensity of the 3.4 eV feature with decreasing photon energy. The Fermi level was established from measurements on a clean Ni stub. All spectra are taken at room temperature and normal emission J. MATER. CHEM., 1991, VOL. 1 there is a considerable copper contribution across the whole width of the valence band. This is in contrast to the profiles from d" perovkites containing elements from earlier in the transition series: where it is possible to distinguish well separated 0 2p and M nd bands, which undergo dramatic intensity fluctuations in a comparable e~periment.~~~~ It is clear that in the cuprate superconductors, there is considerable covalent mixing between copper and oxygen.The only signifi- cant change that is discernible is the decrease in intensity of the feature at 3.4 eV as the photon energy is raised, suggesting that this feature has predominantly 0 2p character. This observation is consistent with our previous studies of -x4311975YBa2Cu307 and a range of BSCCO-based material^.^^'^^^^*^^ A small, but distinct, density of states at the Fermi energy is evident as a sharp cut-off in both the He I and the He I1 spectra of Fig. 4. This small feature can be seen more clearly in Fig.5, where we show expanded valence band regions for the same BSCCO thin film, recorded using He I (hv=21.2 eV) and Ne I (hv= 16.8 eV) radiation. With the use of these low- energy photon sources, four distinct valence band features are resolved at ca. 1.6, 3.4, 4.7, and 5.7 eV binding energy. These energies are in good agreement with those observed in studies of single-crystal Bi2Sr2CaCu208 (001),13*22,42*87-93indicating that our oxygen-annealed c-axis-oriented thin-film material shows essentially the same density of states in the valence- band region as a cleaved single crystal. The relative intensity of the valence-band shoulder at 1.6 eV binding energy decreases with decreasing photon energy. However, since the electron mean-free-path length in the solid changes fairly rapidly over this kinetic energy range,94 this effect may not be attributable solely to cross-section changes.The nature of the density of states at the Fermi energy is discussed in more detail in section 6. 86420-2 binding energy/eV Fig. 5 Valence-level spectra excited within (a) Ne I (hv = 16.8 eV) and (b) He I (hv= 21.2 eV) radiation. Note the clear DOS at EF and the decrease in the intensity of the ca. 1.6eV feature with decreasing photon energy (see text). Spectra are normalised to the height of the valence band maximum. The Fermi level was established from measurements on a cleaned Ni stub J. MATER. CHEM., 1991, VOL. 1 5.2 Resonant Photoemission Studies In principle, resonant photoemission using a synchrotron source may be used to investigate the atomic character of the main valence band more closely.The most important reson- ances explored here are that at the 02p+O 2s threshold (at ca. 18-20 eV), and the Cu 3p-C~ 3d transition at ca. 74 eV. In fact, the relative intensity changes in the features in this part of the spectrum as a function of photon energy, even around resonant thresholds, appear rather muted. This again points to a very strong mixing of 02p and Cu 3d character in the main valence band. However, resonant photoemission 495 summarise the current situation. In superconducting oxides, resonant Cu 3p enhancement of satellite intensity is observed at 12.3- 12.9 eV binding energy in La, -xSrxCu04,105 YBa2C~307-x,53,105-115 Bi2Sr2CaC~208,87.98*116-119 Bi2Sr2C~06,93 Nd2-x Ce Cu04,38,44,98~104~120 and x Pb2Sr2Y1 -,Ca,Cu, -xAgx03121 and is generally assumed to be analogous to the behaviour of the 12.9 eV satellite of CuO.The observation of further structure at ca. 9.5eV binding energy has been a contentious issue.4 This additional structure could simply be part of the Cu 3d8 multiplet, by analogy with cuo~4,99,100,103However, alternative assignments have been in the region of the 02s tends to reinforce our conclusions regarding the 3.4 eV feature of the valence band of fig. 4 and 5. We reviewed the use of resonance photoemission as applied to oxide superconductors in our earlier paper.4 Since this time, a number of groups have investigated the behaviour of valence-band photoemission around a wide range of resonant thresholds in addition to 0 2s and Cu 3p.383a395-98 In particu- lar, a number of rare-earth 4d+4f thresholds have been used to explore the rare-earth (RE) partial density of states con- tributing to the valence-band emission from Nd, -xCexCu04 and its homolog~es.~~*~*~~ These show quite a large contri- bution to the valence bands of these materials from hybridised RE 4f-0 2p states, in addition to structure corresponding to proposed, including 3d1'LL, where LL denotes two holes on core excitation thre~hold~~.~'.~~ different oxygen atoms.'ITResonantphotoemission experi- ments in this spectral range gave very conflicting results; in general very little res~nance,~'~,~~~,~~~ or no resonance at 10.111 was observed at the Cu 3p threshold, while in some cases very large resonances were seen at the 02s threshold123 (for a full discussion see ref.4). The observation of time- and temperature-dependent inten- sity changes in this part of the spectrum has led to the suggestion that a strong features at 9.5 eV is associated with some type of surface contamination. A feature has been observed in this region of the spectrum for the Bi-based superconductor BaPb, -xBix03; in this case the intensity of the feature increases with time after the initial surface prep- In the In experiments by Ark0 and co-workers on cleaved arati~n.~~specific multiplet lines of localised 4f electron~.~~*~'~~ EuBa2Cu307-x124-126 and YB~,CU~O,-~,~'~ the feature was case of Pr, .85Ceo.l,Cu04, the hybridised component is par- ticularly intense and very close to the top of the valence band.98 Studies of Pb2Sr2PrCu308 and PrBa,Cu,O, -x show a similar Pr 4f contribution to the valence band density of states, with hybridisation in the latter being particularly strong;96 it is suggested that this Pr-0 or Pr-Cu hybridisation may be important in the disruption of superconductivity in this material.96 One of the most widespread applications of resonance photoemission has been to the study of the 'satellite' structure appearing below the main valence band, at binding energy >8 eV.This is characteristic of transition-metal oxides toward the right-hand side of the Periodic Table, and arises in a similar way to the satellite structure that we saw in the Cu 2p core levels (section 4.2).In valence-band photoemission, new final states arise where the photohole is screened by transfer of electron density from surrounding atoms. Thus in CuO, the main valence-band emission in fact arises from states of the type 13d9L1 >,and a 'satellite' structure corresponding to 13d8> unscreened final states appears between 10.5 and 12.9 eV below EF.99*100The energy shift is due to strong on- site Coulomb repulsion. [Similarly in the case of NiO, ionis- ation of Ni 3d electrons gives an unscreened Ni 3d7 feature at ca. 10 eV binding energy, the screened 3d peak ( (3d8L1>) being centred at 2 eV."1.102] In the square-planar co-ordi- nation (D4h)symmetry (corresponding to the CuO, planes in the high-T, superconductors) the d8 states span singlet and triplet irreducible representations.The CuO satellites show found to appear as single crystals cleaved at 20 K warmed to room temperature. This led to the interesting suggestion that the 9.5 eV feature is associated with adsorbed or interstitial oxygen derived from the decomposing bulk crystal, as the crystal is warmed from 20 K.'06 However, this possibility has been refuted by other workers who find YBCO surfaces to be relatively stable to oxygen Perhaps a more obvious possibility is that a time-dependent increase in intensity at 9.5eV may be associated with adsorption of H20 or other contaminants from the residual vacuum.Studies of polycrys- talline YBa2Cu30714 and La, ,8Sro.2C~0416 have identified the presence of an OH derived peak at 9.0-9.4eV binding energy, following reaction with H20 (section 2). In the case of YB~,CU,O,-~, the authors note that water exposures as low as 0.1 Lt can modify the valence-region ~pectrum.'~ In Fig. 1 we showed results for the physisorption of H20 on Bi,Sr2CaCu208 (OOl), where it can be seen that water adsorp- tion does indeed modify the spectral features in this range.13 The overall effect is somewhat similar to that noted by Balzarotti et al. for Bi2Sr,CaCu208 (OOl), and attributed to adventitious CO ad~orption.~, The presence of spectral features at ca.9-10eV binding energy from Nd, -xCexCu0444,45*57and ErBa2CU40865 have also been associated with contaminated or degraded sample surfaces. The feature has been found to increase in intensity on mild sputtering of YBa2Cu307 -x ceramic surfaces,127 but as this will tend to increase the reactivity of the surface, it does not allow us to distinguish between oxygen loss/readsorption, and strong resonant enhancement at the Cu 3p threshold.99~100~103 extrinsic contamination. (In the model of Gunnarsson et this is ultimately related to their strong d8 character.) The triplet states are expected to resonate less strongly than the singlet states,1oo the nett effect being that the satellite structure at 12.5 eV in CuO shows the strongest resonant enhancement.99~100*103 The main Cu 3d valence band shows antiresonance intensity variations, centred around 3 eV binding energy.99 Provided that correlation effects in the superconductor materials are similar to those in CuO, we expect to see similar resonance behaviour in the high-T, materials. In our earlier paper,4 we discussed this issue at some length, and the controversies then surrounding it.Here we merely attempt to In summary, it now appears likely that although a weak resonantly enhanced Cu 3d-derived satellite is found at 10 eV binding energy,lo6 the majority of the intensity in this region is not intrinsic to the superconductor itself, but is derived from an oxygen-containing surface species. However, more exotic mechanisms, such as that recently proposed by Kasow- ski et al., involving two-electron processes128 cannot be ruled out until control over sample quality and surface stoichi- ometry is substantially improved.t I L=1.32~10-~mbars. 5.3 Effectsof Chemical Doping on Valence-band Photoemission One question which is currently attracting considerable atten- tion is the way in which the electronic structure of supeycon- ducting oxide changes with chemical doping level. In particular, the position of the valence band edge relative to the Fermi energy as a function of composition is of importance in allowing us to infer whether or not a rigid band model is applicable to these oxides. Fig. 6 shows the changes in the He I (hv=21.2 eV) valence band spectra which occur on crossing the metal-to-non-metal transition for c-axis-oriented thin films of the Bi2.1Sr,.,Cao.8s -xYxCU208+b (Y-BSCCO) ~ystem.~~,’~The corresponding changes in the Cu 2p core- level spectra were discussed in section 4.2.Spectra for x=O and 0.25 show a clear density of states at the Fermi energy. The metal-to-non-metal transition occurs at ca. x = 0.425,’l and the x=O.85 sample shows no appreciable intensity in this region. The onset of the main part of the valence band (between 1.0 and 1.5 eV below EF) moves away from the Fermi energy as the system becomes non-metallic. This is consistent with a simple interpretation of the electronic struc- ture of this p-type material in terms of the progressive filling of holes in the valence band to form a magnetic insulator.The progressive, rather than sudden, disappearance of the density of states at the Fermi energy is also consistent with this type of m0de1.l~’ In the simplest Mott-Hubbard model, the band concerned would be a lower Hubbard band of Cu 3d character. However, we have seen in section 5.1 that the 0 2p and Cu 3d levels overlap in energy very strongly, and it ...... ... .. EF 10 8 6 4 2 0 -2 binding energylev Fig. 6 He I (hv= 2 1.2 eV) valence-level photoemission of Biz,lSr,,,Y,Ca, -,CuzO8 thin films, for a series of compositions spanning the metal-to-non-metal transition in this system. Values of x: (a) 0,(b) 0.25,(c) 0.425, (d) 0.85. The Fermi level was established from measurements on a cleaned Ni stub (from Golden et ~1.’~) J.MATER. CHEM., 1991, VOL. 1 appears that the band gap in these materials may be better described as of the charge-transfer type; in other words, the doped holes go into 0 2p orbitals (see section 6). An alternative model begins with normal Fermi-liquid states consisting of 0 2p and Cu 3d energy bands. In this model, the density of states at EF arises on doping through a Kondo-type many-body resonance of Cu 3d character resulting in renormalised heavy-electron bands being formed in the vicinity of the Fermi level or through creation of new impurity states. The basic difference between the two approaches is that in the Kondo-resonance or impurity-state approach, the states filled by doping do not pre-exist in the insulator, and the Fermi level is pinned by creation of these states on doping.In the alternative, rigid-band approach, the doping creates holes in pre-existing states of predominant 0 2p character, which are split off from the top of the 0 2p valence band by hybridisation with Cu 3d8 states, as described in cluster calculation^.'^^^'^^ The movement of the valence band relative to the Fermi level should in principle allow for the two models to be distinguished. However, this issue is currently somewhat con- troversial. Our conclusion regarding the Y-BSCCO system discussed above (i.e.that hole doping is not in itself responsible for the appearance of structure above the main valence band edge) is reinforced by the experiments of Fujimori et ~l.,’~~ Itti et and Fukuda et ~2E.l~~However, no valence-band l~~shift was observed in the work of Matsuyama et ~1.These authors conclude that hole-doping produces new impurity states at the Fermi level, created by strong hybridisation between doped 0 2p hole orbitals and empty Cu 3d orbitals. These authors also point out that the dominant 0 2p character of the states at EF (see section 6) tends to preclude the possibility of these states arising owing to a Kondo resonance, as a Kondo-like state should have dominant Cu 3d ~haracter.’~~ The experimental situation is no less confusing in the case of other oxide superconductor systems. Some interesting results have been obtained for the Bi-based systems.Shifts of the valence-band edge to lower binding energy on doping have been reported for Ba, -xRbxBi03,66 Ba, -xKxBi03135 and BaPb, -,Bi,03.34 However, in the case of BaPb, -,Bi,03, this shift is rather small, and the authors argue that the Fermi level is pinned in new electronic states created at the Fermi level by doping.34 By contrast, the shift observed in the related Ba,-,M,BiO, systems is roughly in accord with that pre- dicted on the basis of a one-electron rigid-band m~del.~~,~~’ However, since this model fails to describe the electronic structure of the parent compound BaBi03 (which is a semicon- ductor),66 this conclusion must be treated with caution. It is tempting to suggest that comparison of the effect of doping on hole-and electron-doped superconductors should lead to some rationalisation of experimental data.However, here again, quite conflicting results have In thebeen ~btained.~~-~~@.~’ electron-doped material Nd2-,Ce,Cu04, a simple Mott-Hubbard approach would imply that the doped electrons occupy the upper Hubbard band of Cu 3d character. In this case, we would expect the valence-band edge-Fermi level separation to be greater than that found in the hole-doped material La2-,Sr,Cu04, and to increase with doping. Experimentally the situation is compli- cated by the sensitivity of surfaces of the Nd-compound to degradation, so that Fermi-level emission has not been observed in all ~tudies.~’ An increase in the separation of the valence-band edge from EFwith doping is seen in combined photoemission and inverse photoemission measurements by Reihl et aL3’ and interpreted in terms of rigid-band behaviour.However, this shift is not observed by Suzuki et who interpret their data in terms of the creation of a narrow band J. MATER. CHEM., 1991, VOL. 1 of impurity states in the band gap on doping. In a comparative study of Nd, -,Ce,CuO, and La, -,Sr,CuO,, Namatame et .~~1 observe a larger valence band-Fermi energy separation in the former, but argue that the separation is not large enough to be consistent with a model where the doped electrons enter the d" conduction band. They argue that the doped electrons are introduced into impurity levels within the band gap.In another comparison of these hole-and electron- .~~doped materials, Allen et ~1 report that the position of the Fermi energy in Nd,-,Ce,CuO, is virtually identical to that in La, -,Sr,CuO,, and that in both cases the doping produces new states filling the charge-transfer gap. This leads the authors to suggest that these gap-filling states of the doped materials obey a Luttinger-type sum rule at EF,central to the Fermi-liquid theory of metals. The controversy, particularly surrounding the electron- doped materials is unlikely to be resolved until sample quality and problems of surface degradation in these materials can be better controlled. 5.4 Angle-resolved Valence-band Studies Controversy surrounding the electronic structure of these materials has led to a number of detailed angle-resolved measurements, aimed at studying any dispersion of the states close to the Fermi level. These latter measurements are discussed in more detail in section 6.In addition, there have been a number of valence-band studies using angle-resolved photoemission, and the rather interesting electronic properties of these highly correlated materials have led to some re- investigation of the magnetic insulators CuO, NiO, COO and MnO by both angle-resolved and angle-integrated photo- emission.100,103,136-140 There have been two main approaches to understanding the electronic structure of magnetic insulators. In Hubbard- type models, we think of localised states of d" ions, with itinerant behaviour inhibited by electron-repulsion effects.On the other hand, it has been suggested that these materials can be understood in band-theory terms when the correct ground- state magnetic order is incl~ded.'~~~'~~~'~~ In these spin- dependent band models, the band gap arises essentially owing to exchange splitting of the d orbitals (giving an energy difference between spin-up and spin-down electrons on a given ion), combined with crystal-field effects. If these effects are not included, one-electron band theory calculations predict a metallic ground state for these oxides. However, the band gap predicted by spin-dependent calculations for MnO and NiO is generally almost an order of magnitude smaller than the 3-4eV found experimentally. The prediction of a gap at all relies on the metals possessing a half-filled d sub-band; this is not the case for COO, which is predicted to be a metal, even though the experimental gap is similar to that found in MnO and Ni0.129 The existence of the gap in these calcu- lations is also dependent on ground-state antiferromagnetic order; however, experiment shows that the gap does not disappear above the magnetic ordering temperature.A general feature of one-electron band-structure calculations is that the calculated DOS has to be rigidly shifted to higher binding energy for a comparison to be made with experiment, with this effect being generally attributed to ~orrelation.~ However, failings are also found in the localised approach using cluster calculations. In particular, these models are not capable of treating d band dispersion, although this effect has now been observed in ARPES of COO and Ni0.138-140 The observation of band dispersion has been widely cited as supporting the band-theory approach to the structure of these oxides and high-T, materials. However, the observed d-band dispersions are generally less than those predicted by band 497 calculations.140 This leaves us in a rather unfortunate situation where neither the 'localised' nor the 'itinerant' approach appears to describe the electronic structure of magnetic insu- lators satisfactorily. has suggested that the way forward may be to introduce the effect of a Hubbard 'U' into the band model via a Hartree-Fock calculation.In such a calcu- lation, the energy of the empty states include extra Coulomb terms. This type of approach is used by Anisimov et ~1.,'~' who introduce an unoccupied-states potential correction in the local spin-density functional formalism. This model repro- duces the experimental band gaps for NiO, MnO, FeO and COO, and produces an insulating ground state for COO and FeO. A number of ARPES studies of the high-T, materials have now been carried OU~.~~.~~,~~~,~~-~~,~~~ In general, similar features are observed to those for the magnetic insulators. Dispersion of valence band states is observed, particularly in the basal a-b planeg2 (for Bi,.oSrl.8Cao.8Lao.3C~2.108+~, dispersionless behaviour is found in the perpendicular direc- tiongo) but the experimental dispersion is generally less than that predicted by band the~ry.~',~~-~* Again, it is found to be necessary to introduce a rigid downward shift of the DOS calculated by band theory to make comparison with experi- ment.4,92*106*143*1uAs we have discussed above, various corre- lation-related satellite features also appear in the valence- band photoemission, which are not reproduced by a band- theory approach.Thus, the situation appears to be very similar to that found for the magnetic insulators, and further theoretical work is needed to produce a model which can reproduce both the observed band dispersions and the satellite structure due to electron correlation. 6. Fermi-level Studies The transition of cuprate materials from a normal metallic state into a superconducting state depends ultimately on the behaviour of electrons close to the Fermi energy in the normal state.Thus there has been a great deal of interest in Fermi- level structure in photoemission, and the way in which this evolves on going into a superconducting state. We begin by discussing the normal state properties of this structure. 6.1 The Appearance and Size of the DOS at EFin the Normal State In most photoemission studies from oxide superconductors, a clear density of states at EF is observed at room temperature, whether the material is in ceramic, thin-film, or single-crystal f~rm.~?'~'The exceptions to this appear to be andNd2 -,C~,CUO~~~ YBa,Cu,O, -x.4 We discussed the possible surface deterioration of both of these materials in section 2.The non-appearance of a DOS at EFin the case of many room-temperature studies of YBa2Cu307 -x has been a particularly controversial topic., A reproducible Fermi-level DOS has been obtained most successfully for YBCO and its homologues by cleaving single crystals of the material at low temperat~re.~~*~~*~~~-~~~,~~~,~~~This led to the controversial suggestion, which we discussed earlier, that the bulk crystal decomposes losing oxygen (and hence becoming non-metallic) on warming from 20 K; this oxygen then is readsorbed on the surface, giving rise to structure in valence-band photoemis- sion at 9.5 eV and in the 0 1s core-level peak. However, as we pointed out above, there are grounds for suspecting that both the last two effects may be attributable to extrinsic contamination, related to the obviously very high reactivity of this surface.The issue is further confused by the fact that a number of authors have now obtained a finite DOS at E, from YBa2Cu307 -,at room or above the superconducting transition temperature of ca. 91 K.147 Interestingly, the '124' system, YBa2Cu40y appears to be considerably more stable, and a DOS at EF is observed at room temperature. 149 The low stability of these surfaces has meant that the majority of studies of the states close to the Fermi level have been carried out on the considerably more stable BSCCO systems. Our own findings (for example, in fig.5) of a small, but well defined DOS at EF in spectra of these materials is in broad agreement with those of many other groups (see ref. 4). The size of the Fermi edge discontinuity has attracted some attention. Interpreting our own data'2*21v49-'1 in terms of a very simple model that ignores variation in ionisation matrix elements and densities of accessible final states across the occupied valence band,150 the height of the Fermi-edge discontinuity implies that the cut-off in the density of states corresponds to ca. 0.5 states per eV cell for the BSCCO thin films shown in fig. 4 and 5 (T,=85 K). We have previously shown that the size of the Fermi-energy discontinuity is correlated with the superconducting transition temperature as expected in BCS the~ry.~.~' Even taking into account the simplicity of the model used, this is rather lower than typical estimates of 2-3 states per eV cell from band structure calculations.15' (Note, though, that these calculations reflect hypothetical electronic structure at 0 K, where the disconti- nuity is a step function; in photoemission we measure the centroid of the Fermi-level cut-off, which corresponds to half this value, assuming a Fermi-liquid model is appropriate.) Our observations are consistent with those of other workers who have attempted to quantify their spectra in this way; for example, Ark0 et al.lo6 observe a DOS at EF from YBa2Cu306.9 cleaved at low temperature which is two to five times lower than that expected from band structure calculations.A more accurate comparison of the intensity of the observed Fermi level structure with theory is rather difficult. However, very recent ARPES measurements from YBa2C~306.9147 indi-cate that the density of states close to EF below the supercon- ducting transition temperature is consistent with that predicted by BCS theory, assuming a Fermi-liquid description for the normal state. These ARPES measurements are dis- cussed further below. 6.2 Resonant Photoemission of the Fermi-level States Resonant photoemission at the Cu 3p and 0 2s thresholds has been used to investigate the atomic character of the states close to the Fermi level. The data relating to the Cu 3p resonance is relatively uncontroversial. A number of groups have now observed very little or no enhancement of these states at the Cu 3p threshold in Bi2Sr2CaCu208, YBa2Cu,07--x and La, -xSrxCu04.87,106~116~118~119Both Shen et ~1."~and Takahashi et al."' argue that the absence of either resonance or antiresonance behaviour at the Cu 3p threshold indicates that the states at EF have low Cu 3d character.However, List and ~o-workers~~.~~~use cross-section arguments to reach the conclusion that these states have ca. 35% Cu3d character in Bi,Sr2CaCu208 and ca. 20% Cu 3d character in YB~&U,O~-~. The authors argue that the non-appearance of intensity fluctuations at the Cu 3p threshold is not unusual, as an equivalent resonance has not been observed in Ni metal.'" We would point out that this is not the conclusion of Thuler et al., who observe antireson- ance at the Ni3p threshold for the states at EF in both Ni metal and NiO.lS2 In Cu metal, CuO and Cu20, where the 3d band lies at higher binding energy, these authors observe clear antiresonance at the band maximum, ca.3 eV binding energy.99 We would therefore argue that the absence of any J. MATER. CHEM., 1991, VOL. 1 resonant or antiresonant behaviour at the Cu 3p threshold precludes the possibility of the states at EF having appreciable Cu 3d character. Resonance photoemission data at the 02s threshold (ca. 18-22 eV) has aroused considerably more argument. Early observation of resonant enhancement of two Fermi-level features near the 0 2s threshold"' in Bi2Sr2CaCu208 led to their assignment as states having high 0 2p character.This conclusion was disputed by other groups who argued that 18 eV enhancement does not represent the true 0 2s threshold resonance beha~iour."~.' ', This initial confusion seems to have been caused to some extent by the misassignment of the Sr 4p and 02s core levels in the original paper by Takahashi et al."' More recently, Wells et al. reported anomalous resonant enhancement of features at 0.3,0.65 and 1.6 eV below the Fermi energy near 18 eV photon energy." Photoemission final-state effects were proposed as a possible origin." How-ever, recent work has demonstrated unambiguously that a significant part of this structure is attributable to Sr 4p and 0 2s core level peaks excited by a small second-order compo- nent of the exciting radiation (i.e.with energy twice that of the nominal selected photon energy).ls4 In further preliminary work carried out using a monochromator having a very low second-order contribution, this anomalous enhancement appears to be ab~ent.'~' It is clear that the issue of oxygen resonance at EF still remains an open q~estion,''~ and much more experimental work is needed to clarify this problem. However, studies of the empty states above EF by inverse ph~toemission,"~~'~~soft X-ray ab~orption''~ and electron energy-loss spectroscopies'60*'61 tend to indicate that there is indeed significant 0 2p character around the Fermi level in these hole-doped oxides. 6.3 Angle-resolved Measurements of Fermi-level States At the date of our last review, there had been very few investigations by ARPES of the states close to the Fermi leve1.56,'7v92.117 However, a number of high-quality angle-resolved studies have now been reported for BiZSr2CaCu,08,41.47,48,161~162YBa2Cu307 and-x409146*147 Nd2 -xCe,Cu04.56*'7 In general, these studies have revealed dispersive behaviour for the states close to the Fermi energy, with Fermi-level crossings at k values that are in general agreement with the predictions of band the-ory.40,47,56,57,146,162,161However, dispersion of the experimen- tal bands is considerably smaller than the theoretical di~persion.~~,~~,~~,'~~The deviation from band-theory predic- tions away from the Fermi level is attributed to strong correlation effects.Although no detailed resonance photo- emission data are yet available for the Fermi-level states in Nd, -,CexCu04, ARPES measurements in this case indicate that a band which disperses through the Fermi level at the r point has predominant Cu 3d character (in contrast to the dominant 0 2p character indicated for the hole-doped mater- ial~),~~,'~opening up the possibility of a Kondo-type reson- ance state. ARPES measurements for YBa2C~306.940.'46 appear to show very little intensity at the r point. This may go some way toward explaining the anomalously low DOS at EF often observed in studies of this material. 6.4 Measurements of the Superconducting Gap The possibility of measuring the changes in structure which occur close to the Fermi energy (on an energy scale of order k,T,) has led to an enormous effort to improve the resolution attainable in valence-band photoemission, and has also meant that such changes have to date only been observed in oxides with relatively high transition temperatures, T, J.MATER. CHEM., 1991, VOL. 1 (Bi2Sr2CaCu208 and YBa2C~306.9). In these studies, the superior resolution attainable with a conventional gas-discharge lamp has been particularly important; a resolution of 13.5 meV has recently made possible a study of the super- conducting transition in the conventional A1 5 superconductor Nb,Al( T,= 18.6 K).63 Thus, there seems no intrinsic reason why this type of study should not ultimately be possible for all the oxide superconductors.On cooling a normal metal, one expects to see merely a sharpening of the Fermi edge, in accordance with the Fermi- Dirac distribution f~ncti0n.l~~ However, in a BCS-like super- conducting state an energy gap in the quasiparticle density of states opens around EF,and in photoemission one expects spectral weight to be pulled below the Fermi energy as shown schematically in Fig. 7. Spectral profiles which conform to this picture have now been observed by several groups.47,48,147,153,162,165,166The data may be interpreted consistently in terms of an energy gap A = 24-30 meV, corre- sponding to 2A(0)/kBT, z 6.8-8, larger by almost a factor of two than the weak coupling value of BCS theory.One crucial experiment involves angle-resolved photo- emission above and below the superconducting transition temperature, as measurements at different points on the Fermi surface should allow for the detection of any basal plane anisotropy in the size of the superconducting gap. This places constraints on theories of the superconducting pairing mech- I I I I I I I I I I I I A 0.2 0.1 0 -0.1 -0.2 binding energy/eV Fig. 7 Schematic band shapes in photoemission spectra: (a) normal metal at 300 K; (b) normal metal at 0 K; (c) superconductor at 0 K assuming a BCS-like gap with A = 30 meV. Broadening effects due to limited experimental resolution are not included. In (c) empty states which could in principle be seen in inverse photoemission (if resolution could be improved) are shown by the broken curve (from Egdell et a1.4) anism, as models based on d-wave pairing predict gap ani- sotropy in the a-b plane.Recent ARPES measurements for Bi2Sr2CaCu208 have revealed a gap of A =24 meV (& 5 meV) for four distinct points in the rXYZ plane, indicating no gap anisotropy in the a-b plane.48 Wells et al. have used Au deposition on the (001) surface of this oxide in combination with ARPES above and below the superconducting transition temperature to enable them to distinguish the atomic charac- ter of the dispersing bands at the Fermi energy.41 They assign the band-crossing along the TM direction to states having Bi-0 character (i.e.to the surface Bi-0 plane of this layer material). The Bi-0 planes are thus fully metallic (at least when the crystal is oxygen annealed), and contribute to the DOS at EF.This is in line with our own observation of a Doniach-Sunjic lineshape in the Bi 4f and Pb 4f core-level features from Bi2Sr2CaCu208 materials and Pb-stabilised Bi2Sr2Ca2Cu3010, implying a significant Bi/Pb 6p partial DOS at EF.12’49--51A band crossing along TX and TY arises from the Cu-0 states.41 These observations are in general agreement with band-structure calc~lations.~~ Rather different results are obtained for surfaces of crystals which are not properly oxygen annealed, implying that, in these cases, the surface Bi-0 plane is not metallic and superc~nducting.~~ Data for YBa2C~306.9 have been recently reported, and represent the first gap study on this mate~ia1.l~’ (In previous work, the DOS at EF tended to disappear on warming the crystal above the superconducting transition, as we discussed in section 6.1.) Here, the BCS gap is measured in two distinct Fermi surfaces, one originating from the Cu02 planes, and the other from the hybridisation of the planes and the chains of the YBCO structure.Identical gaps of A=25 k2.5 meV are found in both cases.147 In the light of our ideas about electron pairing, these results are perhaps rather unexpected, and pave the way for much further detailed investigation by high-resolution photo- emission. 7.Summary The measurements referred to in the last section show us that photoemission has the potential to yield important infor- mation about the electronic structure of high-temperature superconducting oxides.The fact that so much controversy surrounds the interpretation of many of the experimental data is intimately connected to the currently indifferent quality of the available samples and our rather poor control over the degradation of these samples on the photoemission depth scale, both in atmosphere and in UHV. The experimental resolution needed to study the changes occurring in the density of states at the Fermi level on passing through the superconducting transition push the technique to its current limit, whereas cooling below T, during the experiment may exacerbate problems of surface contamination in uacuo.How-ever, as the quality of single-crystal material, in particular, improves, and enhanced resolution becomes available (for example from third generation synchrotron sources), studies by photoemission should make an increasing contribution to our search for an explanation for superconductivity in these oxides. References 1 M. P. Seah and W. A. Dench, Surf: Interface Anal., 1979, 1, 2. 2 e.g. R. G. Egdell, W. R. Flavell and P. J. Tavener, J. Solid State Chem., 1984,51, 345; S. Campbell, L. L. Cao, R. G. Egdell, W. R. Flavell, K.F. Mok and W. C. Mackrodt, J. Phys. Condensed Matt., 1989, 1, SB237. 500 J. MATER. CHEM., 1991, VOL. 1 3 R. G. Egdell, W. R. Flavell and P. C. Hollamby, J.Solid State Chem., 1989, 79, 238. 40 J. C. Campuzano, G. Jennings, M. Faiz, L. Beauaigue, B. W. Veal J. Z. Liu, A. P. Paulikas, K. Vandervoort, H. Claus, R. S. 4 R. G. Egdell, W. R. Flavell and M. S. Golden, Supercond. Sci. List, A. J. Ark0 and R. J. Bartlett, Phys. Rev. Lett., 1990, 64, Technol., 1990, 3, 8. 2308. 5 Y. Gao, Y. Li, K. L. Merkle, J. N. Mundy, C. Zhang, U. 41 B. 0.Wells, Z-X. Shen, D. S. Dessau, W. E. Spicer, C. G.Olson, Balachandran and R. B. Poeppel, Muter. Lett., 1990, 9, 347. D. B. Mitzi, A. Kapitulnik, R. S. List and A. Arko, Phys. Rev. 6 Y. Gao, K. L. Merkle, C. Zhang, U. Balachandran and R. B. Lett., 1990, 65, 3056. 7 Poeppel, J. Muter. Res., 1990, 5, 1363. S-G. Jin, Z-Z. Zhu, L-M. Liu and Y-L. Huang, Solid State 42 R.Manzke, T. Buslaps, R. Claessen, G. Mante and Z. X. Zhao, Solid State Commun., 1989, 70, 67. Commun., 1990, 74, 1087. 43 H. Ishii, T. Koshizawa, H. Kataura, T. Hanyu, H. Takai, K. 8 P. Svoboda, P. Vasek, 0.Smrckova, D. Sykorova, V. Plechacek Mizoguchi, K. Kume, I. Shiozaki and S. Yamaguchi, Jpn. and M. Nevriva, Solid State Commun., 1990, 75, 331. J. Appl. Phys., 1989, 28, L1952. 9 W-K. Lee and A. S. Nowick, J. Muter. Res., 1990, 5, 1855. 44 H. Namatame, A. Fujimori, Y. Tokura, M. Nakamura, K. 10 T. R. Cummins, R. G. Egdell and G. C. Georgiadis, J. Less- Yamaguchi, A. Misu, H. Matsubara, S. Suga, H. Eisaki, T. Ito, 11 12 Common Met., 1990,164-165, 1149. R. G. Egdell and W. R. Flavell, Supercond. Sci. Technol., 1988, 1, 118.D. A. Geeson, M. S. Golden, S. J. Golden and W. R. Flavell, 45 46 H. Takagi and S. Uchida, Phys. Rev. B, 1990, 41, 7205. A. Fujimori, Y. Tokura, H. Eisaki, H. Takagi, S. Uchida and E. Takayama-Muromachi, Phys. Rev. B, 1990,42, 325. J. C. Campuzano, L. C. Smedskjaer, R. Benedek, G. Jennings 13 Supercond. Sci. Technol., 1989, 2, 279. W. R. Flavell, J. H. Laverty, D. S-L. Law, R. Lindsay, C. A. Muryn, C. F. J. Flipse, G. N. Raiker, P. L. Wincott and G. Thornton, Phys. Rev. B, 1990,41, 11623. 47 and A. Bansil, Phys. Rev. B, 1991,43, 2788. C. G. Olson, R. Liu, D. W. Lynch, R. S. List, A. J. Arko, B. W. Veal, Y. C. Chang, P. Z. Jiang and A. P. Paulikas, Phys. Rev. B, 1990, 42, 38 1. 14 S. L. Qiu, M. W. Ruckman, N. B. Brookes, P. D. Johnson, J. Chen, C.L. Lin, M. Strongin, B. Sinkovic, J. E. Crow and C-S. Jee, Phys. Rev. B, 1988, 37, 3747. 48 C. G. Olson, R. Liu, D. W. Lynch, R. S. List, A. J. Arko, B. W. Veal, Y. C. Chang, P. Z. Jiang and A. P. Paulikas, Solid State Commun., 1990, 76, 41 1. 15 S. L. Qiu, C. L. Lin, M. W. Ruckman, J. Chen, D. H. Chen, 49 M. S. Golden, D. A. Geeson, S. E. Male and W. R. Flavell, D. Nichols and J. E. Crow, AZP Con$ Proc., 1988, 182, 368. Supercond. Sci. Technol., 1989 2, 185. 16 R. L. Kurtz, R. Stockbauer, T. E. Madey, D. Mueller, A. Shih and L. Toth, Phys. Rev. B, 1988, 37, 7936. 50 M. S. Golden, S. J. Golden, R. G. Egdell and W. R. Flavell, J. Less-Common Met., 1990, 164-165, 1084. 17 D. Majumdar, D. Chatterjee and G. Pav-Pujalt, Physica C, 1989, 158,413.51 M. S. Golden, S. J. Golden, R. G. Egdell and W. R. Flavell, J. Muter. Chem., 1991, 1, 63. 18 19 20 R. P. Vasquez, M. C. Foote and B. D. Hunt, AZP Conf: Proc., 1988, 182, 376. R. P. Vasquez, B. D. Hunt and M. C. Foote, Appl. Phys. Lett., 1988, 53, 2692. S. Myhra, P. R. Chalker, P. T. Moseley and J. C. Rivikre, 52 53 Y. Sakisaka, T. Komeda, T. Maruyama, M. Onchi, H. Kato, Y. Ajura, H. Yanahima, T. Terashima, Y. Bando, K. Iijima, K. Yamamoto and K. Hirata, Phys. Rev. B, 1989, 39, 2304. Y. Sakisaka, T. Komeda, T. Maruyama, M. Onchi, H. Kato, Y. Ajura, H. Yanahima, T. Terashima, Y. Bando, K. Iijima, K. 21 Physica C, 1990, 165, 270. W. R. Flavell, J. A. Hampson, C. Neeson and R. J. D. Tilley, Supercond. Sci. Technol., 1989, 1, 221. 54 Yamamoto and K. Hirata, Phys.Rev. B, 1989, 39, 9080. G. Frank, Ch. Ziegler and W. Gopel, Phys. Rev. B, 1991, 43, 2828. 22 D. S. Dessau, Z-X. Shen, B. 0.Wells, W. E. Spicer, R. S. List, 55 J. Hockertz, J. Werno and H. J. Clemens, J. Less-Common Met., A. J. Arko, R. J. Bartlett, Z. Fisk. S-W. Cheong, D. B. Mitzi, 1990, 164-165, 1224. A. Kapitulnik and J. E. Schirber, Appl. Phys. Lett., 1990, 57, 307. 56 Y. Sakisaka, T. Maruyama, Y. Morikawa, H. Kato, K. Edamoto, M. Okusawa, Y. Aiura, H. Yanashima, T. Terashima, Y. Bando, 23 B. 0. Wells, P. A. P. Lindberg, Z-X. Shen, D. S. Dessau, I. Lindau, W. E. Spicer, D B. Mitzi and A. Kapitulnik, AZPIAVS K. Iijima, K. Yamamoto and K. Hirata, Solid State Commun., 1990, 74, 609. 24 Conf. Proc., 1989, 182, 391. P. A. P. Lindberg, Z-X.Shen, B. 0. Wells, D. B. Mitzi, I. 57 Y. Sakisaka, T. Maruyama, Y. Morikawa, H. Kato, K. Edamoto, M. Okusawa, Y. Aiura, H. Yanashima, T. Terashima, Y. Bando, Lindau, W. E. Spicer and A. Kapitulnik, Phys. Rev. B, 1989, 40,8769. K. Iijima, K. Yamamoto and K. Hirata, Phys. Rev. B, 1990, 42, 4 189. 25 D. M. Hill, H. M. Meyer 111, J. H. Weaver, C. F. Gallo and K. C. Goretta, Phys. Rev. B, 1988, 38, 11331. 58 C. L. Lin, S. L. Qiu, J. Chen, J. M. Tranquada and M. Strongin, Physica C, 1989, 162-164, 1325, and references therein. 26 27 Ph. Niedermann, A. P. Grande, J. K. Grepstad, J-M. Triscone, M. G. Karkut, 0. Brunner, L. Antonazza, W. Sadowski, H. J. Scheel and 0. Fischer, J. Appl. Phys., in the press. I. W. Owen, N. B. Brookes, C. H. Richardson, D.R. Warburton, 59 60 M. Strongin, S. L. Qiu, J. Chen, C. L. Lin and E. M. McCarron, Phys. Rev. B, 1990,41, 7238. N. D. Shinn, J. W. Rogers and J. E. Schirber, Phys. Rev. B, 1990, 41, 7241. F. M. Quinn, D. Norman and G. Thornton, Surf: Sci., 1986, 61 F. Parmigiani, Z. X. Shen, D. B. Mitzi, I. Lindau, W. E. Spicer 178, 897. and A. Kapitulnik, Phys. Rev. B, 1991, 43, 3085. 28 K. N. R. Taylor, personal communication, 1990. 62 A. Balzarotti, M. De Crescenzi, N. Motta, F. Patella and A. 29 R. G. Egdell, J. B. Goodenough, A. Hamnett and C. Naish, Sgarlata, Proc SATT3, ed. C. Ferdeghini and A. Siri, World J. Chem. SOC. Faraday Trans l., 1983,79, 893. Scientific, Singapore, in the press. 30 K. Siegbahn, J. Electron. Spectrosc. Relat. Phenom., 1974, 5, 1.63 A. Balzarotti, M. De Crescenzi, N. Motta, F. Patella, A. Sgarlata, 31 A. Balzarotti, F. Patella, M. De Crescenzi, N. Motta, A. Sgarlata P. Paroli, G. Balestrino and M. Marinelli, Phys. Rev. B, in the and F. Licci, Phys. Rev. B, 1991, 43, 351. press. 32 M. H. Frommer, Supercond. Sci. Technol., 1990, 3, 523. 64 R. G. Egdell and P. D. Naylor, Chem. Phys. Lett., 1982, 91, 33 H. Zhang, Q. R. Zhang, Z. P. Su and M. R. Ji, Phys. Rev. B, 200. 34 1990,42, 2248. H. Matsuyama, T. Takahashi, H. Katayama-Yoshida, Y. Okabe, 65 P. Adler, H. Buchkremer-Hermans and A. Simon, 2. Phys. B, 1990, 81, 355. 35 36 H. Takagi and S. Uchida, Phys. Rev. B, 1989, 40, 2658. J-J. Yeh, 1. Lindau, J-Z. Sun, K. Char, N. Missert, A. Kapitulnik, T.H. Geballe and M. R. Beasley, Phys. Rev. B, 1990,42, 8044. T. Suzuki, M. Nagoshi, Y. Fukuda, K. Oh-ishi. Y. Syono and 66 67 R. Itti, K. Tomeno, K. Ikeda, K. Tai, N. Koshizuka and S. Tanaka, Phys. Rev. B,1991, 43,435. W-M. Hurng, S. R. Horng, S. F. Wu, C. Y. Shei, W. H. Lee and P. T. Wu, Solid State Commun., 1991, 77, 269. M. Tachiki, Phys. Rev. B, 1990, 42, 4263. 68 M. K. Rajumon, D. D. Sarma, R. Vijayaraghavan and C. N. 37 B. Reihl, Y. Maeno, I. Mangelschots, K. 0.Magnusson and C. R. Rao, Solid State Commun., 1989, 70, 875. Rossel, Solid State Commun., 1990, 74, 31. 69 C. S. Ares Fang, Y. D. Yao and C. C. Wu, Physica C, 1990, 38 J. W. Allen, C. G. Olson, M. B. Maple, J-S. Kang, L. Z. Liu, 167, 89. J-H. Park, R. 0. Anderson, W. P Ellis, J. T.Markert, Y. 70 I-S. Yang, A. G. Schrott and C. C. Tsuei, Phys. Rev. B, 1990, Dalichaouch and R. Liu, Phys. Rev. Lett., 1990, 64, 595. 41, 8921. 39 D. E. Fowler, C. R. Brundle, J. Lerczak and F. Holtzberg, J. 71 L. Zhenxiang, X. Kan, W. Naijuan, S. Dainhong and L. Electron. Spectrosc. Relat. Phenom., 1990, 52, 323. Zhangda, Chin. Phys. Lett., 1989, 6, 261. J. MATER. CHEM., 1991, VOL. 1 501 72 73 74 75 76 77 78 79 80 81 82 83 84 85 S. Massidda,.J. Yu and A. J. Freeman, Physica C, 1988, 152, 251. F. Burgazy, C. Politis, P. Lamparter and S. Steeb, 2. Nutur-forsch., Teil A, 1989, 44,780. S. L. Qiu, C. L. Lin, J. Chen and M. Strongin, Phys. Rev. B, 1990,41, 7467. W. R. Flavell and R. G. Egdell, Phys. Rev. B, 1989,39, 231. Y.Shichi, Y. Inoue, F. Munakata and M. Y. Yamanaka, Phys. Reo. B, 1990, 42, 939. B. Dauth, P. Sen, B. Rupp, W. Gudat, Ch. Ziegler and W. Gopel, 2. Phys. B, 1990,79, 169. M. Fujinami, H. Hamada, Y. Hashiguchi and T. Ohtsubo, Jpn. J. Appl. Phys., 1989, 11, 1959. J. Ghijsen, L. H. Tjeng, J. van Elp, H. Eskes, J. Westerink, G. A. Sawatzky and M. T. Czczyk, Phys. Rev. B, 1988, 38, 11322, and references therein. G. van der Laan, C. Westra, C. Haas and G. A. Sawatzky, Phys. Rev. B, 1981,23,4369. P. Steiner, V. Kinsinger, I. Sander, B. Siegwart, S. Hufner, C. Politis, R. Hoppe and H. P. Muller, 2.Phys. B, 1987, 67, 497. W. Herzog, M. Schwartz, H. Six1 and R. Hoppe, 2. Phys. B, 1988, 71, 19. M. Ospelt, J. Henz, E. Kaldis and P. Wachter, Physicu C, 1988, (a)K. Allan, A. Campion, J.Zhou and J. B. Goodenough, Phys. Rev. B, 1990, 41, 11572. (b) P. A. P. Lindberg, 1. Lindau and W. E. Spicer, Phys. Rev. B, 1989, 40, 6822. (c)C. N. R. Rao, G. R. Rao, M. K. Rajumon and D. D. Sarma, Phys. Rev. B, 1990, 42, 1026. S. J. Golden, T. W. Bloomer, F. F. Lange, A. M. Segadaes, K. J. Vaidya and W. L. Olson, Proc. Muter. Res. SOC., 1989, 169, 603. 153-155, 159. 104 105 106 107 108 109 110 111 112 0. Gunnarsson, J. W. Allen, 0. Jepsen, T. Fujiwara, 0. K. Andersen, C. G. Olson, M. B. Maple, J-S. Kang, L. Z. Liu, J-H. Park, R. 0. Anderson, W. P. Ellis, R. Liu, J. T. Markert, Y. Dalichaouch, Z-X. Shen, P. A. P. Lindberg, B. 0. Wells, D. S. Dessau, A. Borg, I. Lindau and W. E. Spicer, Phys. Rev. B, 1990,41,4811. Z-X. Shen, J. W. Allen, J. J. Yeh, J-S.Kang, W. Ellis, W. E. Spicer, I. Lindau, M. B. Maple, Y. D. Dalichaouch, M. S. Torikachvili, J. Z. Sun and T. H. Geballe, Phys. Rev. B, 1987, 36, 8414. A. J. Arko, R. S. List, R. J. Bartlett, S-W. Cheong, Z. Fisk, J. D. Thompson, C. G. Olson, A-B. Yang, R. Liu, C. Gu, B. W. Veal, J. Z. Liu, A. P. Paulikas, K. Vandervoort, H. Claus, J. C. Campuzano, J. E. Schirber and N. D. Shinn, Phys. Rev. B, 1989, 40,2268. P. Thiry, G. Rossi, Y. Petroff, A. Revcolevski and J. Jegovdez, Europhys. Lett., 1988, 5 55. M. Tang, N. G. Stoffel, Q. B. Chen, D. LaGraffe, P. A. Morris, W. A. Bonner, G. Margaritondo and M. Onellion, Phys. Rev. B, 1988, 38, 897. N. G. Stoffel, J. M. Tarascon, Y. Chang, M. Onellion, D. W. Niles and G. Margaritondo, Phys. Rev. B, 1987, 36, 3986. M. Onellion, Y.Chang, D. W. Niles, R. Joynt, G. Margaritondo, N. G. Stoffel and J. M. Tarascon, Phys. Rev. B, 1987, 36, 819. T. Takahashi, F. Maeda, H. Arai, H. Katayama-Yoshida, Y. Okabe, T. Suzuki, Y. Takakuwa, S. Hosoya, A. Fujimori, T. Miyahara, T. Koide, T. Shirade, M. Sato, S. Shamoto and M. Onoda, Physicu C, 1987, 148B, 476. T. Takahashi, F. Maeda, H. Arai, H. Katayama-Yoshida, Y. Okabe, T. Suzuki, S. Hosoya, A. Fujimori, T. Shidara, T. Koide, T. Miyahara, M. Onoda, S. Shamoto and M. Sato, Phys. Rev. B, 1987,36, 5686. 86 87 88 89 90 91 92 93 94 95 96 97 98 99 see e.g. C. C. Naish, 3.Phil. Thesis, University of Oxford, 1985. R. S. List, A. J. Arko, R. J. Bartlett, C. G. Olson, A-B. Yang, R. Liu, C. Gu, B. W. Veal, Y. Chang, P. Z. Jiang, K. Vandervoort, A. P. Paulikas and J.C. Campuzano, Physicu C, 1989, 159, 439. C. G. Olson, R. Liu, D. W. Lynch, B. W. Veal, Y. C. Chang, P. Z. Jiang, J. Z. Liu, A. P. Paulikas, A. J. Ark0 and R. S. List, Physica C, 1989, 162-164, 1697. B. 0. Wells, P. A. P. Lindberg, Z-X. Shen, D. S. Dessau, W. E. Spicer, I. Lindau, D. B. Mitzi and A. Kapitulnik, Phys. Rev. B, 1989,40, 5259. P. A. P. Lindberg, Z-X. Shen, D. S. Dessau, B. 0. Wells, D. B. Mitzi, I. Lindau, W. E. Spicer and A. Kapitulnik, Phys. Rev. B, 1989,40, 5169. P. A. P. Lindberg, Z-X. Shen, D. S. Dessau, B. 0. Wells, A. Borg, D. B. Mitzi, I. Lindau, W. E. Spicer and A. Kapitulnik, T. Takahashi, H. Matsuyama, H. Katayama-Yoshida, Y. Okabe, S. Hosoya, K. Seki, H. Fujimoto, M. Sato and H. Inokuchi, Phys. Rev. B, 1989,39, 6636. H. Eisaki, H.Takagi, S. Uchida, H. Matsubara, S. Suga, M. Nakamura, K. Yamaguchi, A. Misu, H. Namatame and A. Fujimori, Phys. Rev. B, 1990, 41, 7188. D. A. Shirley, in Photoemission in Solids I General Principles, Topics in Applied Physics 26, ed. M. Cardona and L. Ley, Springer Verlag, Berlin, 1978. P. A. P. Lindberg, Z-X. Shen, B. 0. Wells, D. S. Dessau, W. P. Ellis, A. Borg, J-S. Kang, D. B. Mitzi, I. Lindau, W. E. Spicer and A. Kapitulnik, Phys. Rev. B, 1989, 40, 8840. D. S. Dessau, Z-X. Shen, P. A. P. Lindberg, B. 0. Wells, A. Borg, I. Lindau, W. E. Spicer, J. V. Waszczak and L. F. Schneemeyer, Phys. Rev. B, 1989, 40, 6726. Z-X. Shen, P. A. P. Lindberg, B. 0. Wells, D. S. Dessau, A. Borg, I. Lindau, W. E. Spicer, W. P. Ellis, G. H. Kwei, K. C. Ott, J-S. Kang and J. W.Allen, Phys. Rev. B, 1989, 40, 6912. A. Grassmann, J. Strobel, G. Saemann-Ischenko and R. L. Johnson, preprint. M. R. Thuler, R. L. Benbow and Z. Hurych, Phys. Rev. B, 1982, 26, 669. Physics C, 1989, 162-164, 1313. 113 114 115 116 117 118 119 120 121 122 123 124 125 126 R. L. Kurtz, R. L. Stockbauer, D. Mueller, A. Shih, L. E. Toth, M. Osofsky and S. A. Wolf, Phys. Rev. B, 1987, 35, 8818. A. J. Lowe, D. Greig, M. A. Howson, M. J. Walker, H. H. Boghosian, R. Stevens, Y. L. Chen, D. S-L. Law, D. Norman, F. M. Quinn and J. A. D. Matthew, J. Phys. C, 1988,21, L763. A. Samsavar, T. Miller, T-C. Chiang, B. G. Pazol, T. A. Friedmann and D. M. Ginsberg, Phys. Rev. B, 1988, 37, 5164. Z-X. Shen, P. A. P. Lindberg, P. Soukiassian, C. B. Eom, I. Lindau, W. E. Spicer and T.H. Geballe, Phys. Rev. B, 1989, 39, 823. F. Maeda, T. Kawamura, M. Oshima, Y. Hidaka and A. Yamaji, Jpn. J. Appl. Phys., 1989, 28, L361. T. Takahashi, H. Matsuyama, H. Katayama-Yoshida, Y. Okabe, S. Hosoya, K. Seki, H. Fujimoto, M. Sato and H. Inokuchi, Nature (London), 1988, 334, 691. W. R. Flavell, J. H. Laverty, D. S-L. Law, R. Lindsay, C. A. Muryn, C. F. J. Flipse, G. N. Raiker, P. L. Wincott and G. Thornton, unpublished work. Y. Hwu, M. Marsi, A. Terrasi, D. Rioux, Y. Chang, J. T. McKinley, M. Onellion, G, Margaritondo, M. Capozi, C. Qua-resima, A. Campo, C. Ottaviani, P. Perfetti, N. G. Stoffel and E. Wang, Phys. Rev. B, 1991,43, 3678. Y. Hwu, Y. Chang, M. Onellion and G. Margaritondo, Solid State Commun., 1989, 71, 415. D. E. Ramaker, Phys. Rev. B, 1988, 38, I18 16.R. L. Kurtz, S. W. Robey, R. L. Stockbauer, D. Muller, A. Shih and L. Toth, Phys. Rev. B, 1989, 39, 4768. A. J. Arko, R. S. List, Z. Fisk, S-W. Cheong, J. D. Thompson, J. A. ORourke, C. G. Olson, A-B. Yang, T-W. Pi, J. E. Schirber and N. D. Shinn, J. Mugn. Magn. Muter., 1988, 75, L1. R. S. List, A. J. Arko, Z. Fisk, S-W. Cheong, S. D. Conradson, J. D. Thompson, C. B. Pierce, D. E. Peterson, R. J. Bartlett, N. D. Shinn, J. E. Schirber, B. W. Veal, A. P. Paulikas and J. C. Campuzano, Phys. Rev. B, 1988, 38, 11966. R. S. List, A. J. Arko, Z. Fisk, S-W. Cheong, S. D. Conradson, J. D. Thompson, C. B. Pierce, D. E. Peterson, R. J. Barlett, J. A. O’Rourke, N. D. Shinn, J. E. Schirber, B. W. Veal, A. P. Paulikas and J. C. Campuzano, AIP Conk Proc., 1989, 182, 283.100 101 102 103 J. Ghijsen, L. H. Tjeng, H. Eskes, G. A. Sawatzky and R. L. Johnson, Phys. Rev. B, 1990,42, 2268. S. Hufner, F. Hulliger, J. Osterwalder and T. Riesterer, Solid State Commun., 1984, 50, 83. G. A. Sawatzky and J. W. Allen, Phys. Rev. Lett., 1984, 53, 2339. Z-X. Shen, R. S. List, D. S. Dessau, F. Parmigiani, A. J. Arko, R. Bartlett, B. 0.Wells, I. Lindau and W. E. Spicer, Phys. Rev. B, 1990, 42, 808 1. 127 128 129 130 F. Parmigiani, G. Samoggia, C. Calandra and F. Manghi, J. Appl. Phys., 1989, 66, 5958. R. V. Kasowski, P. Soukiassian and G. Collin, Solid State Commun., 1991, 77, 135. e.g. P. A. Cox, Transition Metal Oxides: An Introduction to their Electronic and Magnetic Properties, Oxford University Press, to be published. A. Fujimori, Y.Tokura, H. Eisaki, T. Takagi, S. Uchida and M. Sato, Phys. Rev. B, 1989, 40,7303. 502 J. MATER. CHEM., 1991, VOL. 1 131 132 133 134 135 136 137 138 139 140 141 A. Fujimori, Phys. Rev. B, 1989, 39, 793. R. Itti, F. Munakata, K. Ikeda, N. Koshizuka and S. Tanaka, Physica B, 1990, 165-166, 1223. Y. Fukuda, K. Terashima, Y. Nakanishi, T. Suzuki, M. Nagoshi, Y. Syono and M. Tachiki, Physica C, 1989, 162-164, 1315. H. Matsuyama, T. Takahashi, H. Katayama-Yoshida, T. Kashi- wakura, Y. Okabe, S. Sato, N. Kosugi, A. Yagishita, K. Tanaka, H. Fujimoto and H. Inokuchi, Physica C, 1989, 160, 567. M. S. Hedge, P. Barboux, C. C. Chang, J. M. Tarascon, T. Venkatesan, X. D. Wu and A. Inam, Phys. Rev. B, 1989, 39, 4752. B. Hermsmeier, J. Osterwalder, D. J. Friedman, B.Sinkovic, T. Ran and C. S. Fadley, Phys. Rev. B, 1990, 42, 11895. A. Fujimori, N. Kimizuka, T. Akahane, T. Chiba, S. Kimura, F. Minami, K. Siratori, M. Taniguchi, S. Ogawa and S. Suga, Phys. Rev. B, 1990,42, 7580. Z-X. Shen, J. W. Allen, P. A. P. Lindberg, D. S. Dessau, B. 0. Wells, A. Borg, W. Ellis, J. S. Kang, S-J. Oh, I. Lindau and W. E. Spicer, Phys. Rev. B, 1990,42, 1817. Z-X. Shen, P. A. P. Lindberg, C. K. Shih, W. E. Spicer and I. Lindau, Physica C, 1989, 162-164, 13 11. Z-X. Shen, C. K. Shih, 0.Jepson, W. E. Spicer, I. Lindau and J. W. Allen, Phys. Rev. Lett., 1990, 64, 2442. V. I. Anisimov, M. A. Korotkin and E. Z. Kurmaev, J. Phys. Condensed Matt., 1990, 2, 3973. 150 151 152 153 154 155 156 157 158 159 160 R. G. Egdell and M. D. Hill, J. Phys.C, 1983, 16,6221. C. Guillot, Y. Ballu, J. Paigne, J. Lecante, K. P. Jain, P. Thiry, R. Pinchaux, Y. Petroff and L. M. Falicov, Phys. Rev. Lett., 1977, 39, 1632. M. R. Thuler, R. L. Benbow and Z. Hurych, Phys. Rev. B, 1983, 27, 2082. R. Manzke, T. Buslaps, R. Claessen and J. Fink, Europhys. Lett., 1989, 9, 477. W. R. Flavell, J. H. Laverty, D. S-L. Law, R. Lindsay, C. A. Muryn, C. F. J. Flipse, G. N. Raiker, P. L. Wincott and G. Thornton, Phys. Rev. B, in the press. P. L. Wincott et al., unpublished work, to be presented at the 3rd International Conference on Materials and Mechanisms of Superconductivity, Kanazawa, Japan, July 1991. B. 0. Wells, Z-X. Shen, D. S. Dessau and W. E. Spicer, Phys. Rev. B, in the press. R. Claessen, R. Manzke, H. Cartensen, B. Burandt, T. Buslaps, M. Skibowski and J. Fink, Phys. Rev. B, 1989,39, 7316. T. J. Wagener, Y-J. Hu, M. B. Jost and J. H. Weaver, Phys. Rev. B, 1990,42, 1041. P. Kuiper, M. Grioni, G. A. Sawatzky, D. B. Mitzi, A. Kapitul- nik, A. Santaniello, P. de Padova and P. Thiry, Physica C, 1989, 157,260. N. Nucker, H. Romberg, X. X. Xi, J. Fink, B. Gegenheimer and Z. X. Zhao, Phys. Rev. B, 1989,39, 6619. 142 R. Bottner, N. Schroeder, E. Dietz, U. Gerhardt, W. Assmus 161 J. Fink, N. Nucker, H. Romberg, M. Alexander, P. Adelmann, and J. Kowalewsi, Phys. Rev. B, 1990, 41, 8679. J. Mante, R. Claessen, T. Buslaps, S. Harm, R. Manzke and M. 143 M. A. Korotkin, V. I. Anisimov, S. M. Butorin, V. R. Galakhov Skibowski, J. Less-Common Met., 1990, 164-165, 967. 144 145 146 and E. Z. Kurmaev, Mater. Lett., 1990, 10, 34. G. Drager, F. Werfel-and J. A. Leiro, Phys. Rev., 1990,41,4050. P. A. P. Lindberg, Z-X. Shen, W. E. Spicer and I. Lindau, Sur-Sci. Rep., 1990, 11, 1. J. C. Campuzano, L. C. Smedskjaer, R. Benedek, G. Jennings and A. Bansil, Phys. Rev. B, 1991, 43, 2788. 162 163 164 C. G. Olson, R. Liu, A-B. Yang, D. W. Lynch, A. J. Arko, R. S. List, B. W. Veal, Y. C. Change, P. Z. Jiang and A. P. Paulikas, Science, 1989, 245, 73 1. M. Grioni, D. Malterre, B. Dardel, J-M. Imer, Y. Baer, J. Muller, J. L. Jorda and Y. Petroff, Phys. Rev. B, 1991,43, 1216. G. K. L. Cranstoun, R. G. Egdell, M. D. Hill and R. Samson, 147 148 J. C. Campuzano, G. Jennings, N. River, A. J. Arko, R. S. List, B. W. Veal and A. P. Paulikas, Phys. Rev. Lett., submitted. P. D. Johnson, S. L. Qiu, L. Jiang, M. W. Ruckman, M. Strongin, S. L. Hulbert, R. F. Garrett, B. Sinkovic, N. V. Smith, R. J. Cava, C. S. Jee, D. Nichols, E. Kaczanoxicz, R. E. Salomon 165 166 J. Electron. Spectrosc. Relat. Phenom., 1984, 33, 23. Y. Chang, M. Tang, R. Zanoni, M. Onellion, R. Joynt, D. L. Huber, G. Margaritondo, P. A. Morris, W. A. Bonner, J. M. Tarascon and N. G. Stoffel, Phys. Rev. B, 1989, 39,4740. J-M. Imer, F. Patthey, B. Dardel, W.-D. Schneider, Y. Baer, Y. 149 and J. E. Crow, Phys. Rev. B, 1987, 35, 881 1. R. Itti, T. Miyatake, K. Ikeda, K. Yamaguchi, N. Koshizuka Petroff and A. Zettl, Phys. Rev. Lett., 1989, 62, 336. and S. Tanaka, Phys. Rev. B, 1990, 41, 9559. Paper 1/01714D; Received 9th April, 1991