C

resolution of 25 meV and at 300 K [17.26],

I " I > I ' 1 ' I ' I ' I ' I ' I ' I ' I 6876543210-1

Binding energy (eV)

ated with the lower crystal symmetry of the higher fullerenes. The higher fullerenes (e.g., C84) also show line broadening associated with the presence of isomers [17.27].

17.1.2. XPS Spectra for C60

X-ray photoelectron spectroscopy (XPS) refers to the PES process when the excitation photon is in the x-ray range and the electron excitation is from a core level. XPS spectra have been obtained for a variety of fullerenes. For example, Fig. 17.4 shows the XPS spectrum for C60. In this spectrum, we can see well-defined peaks which show an intense, narrow main line (peak #1 in Fig. 17.4) identified with the emission of a photoexcited electron from the carbon Is state, which has a binding energy of 285.0 eV and a very small linewidth of 0.65 eV at half-maximum intensity [17.28]. The sharpest side band feature in the downshifted XPS spectrum (labeled 2) is identified with an on-site molecular excitation across the HOMO-LUMO gap at 1.9 eV [17.28]. Features 3, 4 and 5 are the photoemission counterparts of dipole excitations seen in optical absorption, while features 6 and 10 represent on-molecule plasmon collective oscillations of the it and <t charge distributions, respectively. Plasma excitations are also prominently featured in core level electron energy loss spectra (EELS) discussed in §17.2.3.

XPS spectra for C60 have been obtained by a number of workers [17.29, 30] and have been compared with theory, as well as with XPS spectra for various aromatic compounds.

Relative Binding Energy (eV)

Relative Binding Energy (eV)

Fig. 17.4. X-ray photoelectron spectra for the carbon ls-derived satellite structures for Cw] [17.28]. The sharp feature 1 is the central XPS line, while the downshifted features 2-10 refer to CH) excitations (see text). The XPS data are shown on two energy scales to emphasize features near Ef (upper spectrum) and features farther away in energy (lower spectrum) [17.28].

17.1.3. PES and IPES Spectra for Doped Fullerenes

For the case of doped fullerenes, surface states and dangling bonds associated with nonstoichiometry are more important than for pristine C60 and the photoemission experiments may not reflect bulk properties as accurately as for their undoped counterparts. Nevertheless, photoemission and inverse photoemission data for doped fullerenes do provide convincing evidence for charge transfer and for band filling as x in KxC60 increases. Referring to Fig. 17.1(a), we can see that the density of states peak associated with the ilu-derived band just above the Fermi level EF in the C60 trace moves closer to EF as K is added to C60 [17.2], Upon further addition of potassium, the /lu-related peak crosses the Fermi level and eventually for 5.8 < x < 6.0, the -derived peak falls below Ef, indicating complete filling of the /^-derived level; the data for this composition further show a small band gap to the next higher lying /^-derived level (see Fig. 12.2). PES and IPES studies on doped fullerenes have been carried out by many groups, and more recent work has tended to be done under higher resolution and with more control in sample preparation [17.24,31-35]. Photoemission and inverse photoemission [17.8] studies have further confirmed strong similarities between the electronic structures of K,C60 and both RbxC60 and Cs,C60 [17.3,3638]. Weak features at 0.3 and 0.7 eV below EF in the photoemission spectra of K3C60 [17.39] have been identified with electrons coupling to vibrational modes (~0.3 eV) and to plasma excitations (~0.7 eV) [17.35,40]. Differences have also been clearly demonstrated between the density of states for KjQo and those for both Na^C60 and Ch^Qq, for which multiple metal ions can be accommodated in the octahedral sites, because of the small size of the Na and Ca ions (see §8.5).

Calculations to explain the photoemission and inverse photoemission spectra [17.3,13,28,41-44] have been based on band models for the electronic states, and good agreement between theory and experiment has been reported by many authors [17.4]. It is not believed that the agreement between the experimental and calculated densities of states favors any particular approach to the calculation of the electronic levels, whether by band calculation approaches or by many-body approaches. Because of the molecular nature of solid fullerene films, the electronic states up to -100 eV above the Fermi level which form the final states in the photoemission process relate to the free molecule to some degree. Photoemission experiments thus show that details in the density of states depend to some degree on the photon energy used for the photoemission probe. For this reason, the PES measurements of the occupied density of states are not expected to correspond in detail to generic calculations.

UV photoemission spectra taken on K3C60 [17.33,36,45] show that the spectral peaks are not sensitive to the electron mean free path. Furthermore, the Auger spectra and the photoelectron spectroscopy self-convolution for KgQo and K3C60 have common features and the analysis of these spectra yields the same values for the Hubbard U = 1.4 ± 0.2 eV for both K3C60 and K^Q,, [17.45]. A similar experiment for undoped C60 yielded U = 1.6 eV [17.9] (see §17.3).

Regarding other doped fullerene systems, photoemission studies have been carried out both on M^Qo and M,C70 systems for various metals M. For example, PES studies on K^C^ [17.22,46] have shown regions of stability for K,C70, K4C70, and K6C70 and have demonstrated band filling at x — 6, but no metallic phases for the composition range 0 < x < 6. Photoemission studies have also been carried out on RbxC70 [17.47], Regarding alkaline earth-doped fullerides, photoemission spectra for Ca5C60 [17.48] and BafiQo [17.33,49] have been reported showing partial filling of the tlg LUMO+1 band for these systems.

17.1.4. XPS Studies of Adsorbed Fullerenes on Substrates

XPS spectra of the carbon Is core level for C6(l on noble metal substrates (Ag, Cu, Au) show a downshift relative to bulk C60 in the Is binding energy, as illustrated in Fig. 17.5(a). This shift rapidly disappears with increasing C60 coverage, showing that about two monolayers (ML) are sufficient to yield bulk values [see Fig. 17.5(b)] [17.50]. This result is consistent with temperature-programmed desorption studies (see §17.5) and scanning tunneling microscopy (STM) studies (see §17.4), showing a different behavior for the first monolayer relative to subsequent C60 layers.

A more detailed study of C60 on Cu (100) surfaces [17.51] shows a downshift in the carbon Is XPS line between multilayer and monolayer Cgg coverage as shown in Fig. 17.5(c). This downshift (0.56 eV) is comparable to that for K3C60 films (0.66 eV) and is attributed to (1) charge transfer from the Cu substrate to the adsorbed C60 molecule, and (2) hybridization between the Cu (4s) and C (2p) states [17.51]. Improved calibration of the surface coverage of deposited C60 is achieved using single crystal substrates and flashing off all but the first monolayer [17.52], which adheres more strongly to the surface (see §17.4.1).

Photoemission has also been used to study the interaction between C60 and transition metals such as tungsten (100) [17.53] and Rh (111) [17.54], showing that the binding of C60 to these transition metal surfaces is stronger than the C60-C60 bond, just as for the noble metals discussed above. Measurement of the binding energy of the HOMO level shows charge transfer from the W (100) surface to C60, while for the Rh (111) surface it is claimed in one report [17.54] that electrons from C60 flow into the transition metal surface.

17.2. Electron Energy Loss Spectroscopy

Electron energy loss spectroscopy (EELS) is a complementary tool to optical probes for studying excitations in solids. The emphasis of EELS studies is in surface science because inelastic electron scattering experiments typically are carried out in an energy range where the electron penetration depth is only two or three nanometers. Because the interaction between the incoming electron and the fullerenes is much stronger than that of the photon, selection rules are less important in the EELS experiment. In this way the EELS technique is complementary to optical probes. Although most of the publications on fullerenes deal with the inelastic scattering of electrons, a few structural studies using elastic electron scattering have also been reported, including both electron diffraction and low-energy electron diffraction (LEED) to examine bulk and surface structural features

288 286 284 282 BINDING ENERGY (eV)

Fig. 17.5. (a) XPS spectra showing the XPS carbon Is peak (taken with Mg Ka radiation) of 0.5-ML Qo on Cu, Au, and Ag substrates compared to bulk CM. (b) Shift in the C Is peak with increasing C^ coverage, compared to bulk Cut. The points indicate an Ag substrate (o), Cu substrate (□), Au substrate (x). On all three substrates, bulk CM and six or more ML of Cm yield the same Is peak energy, as shown in the figure [17.50]. (c) X-ray photoemission spectra of the C (Is) core level for Cm films at coverages of 0.24, 0.52, 1.0, 1.7, 2.7, and 4 ML of C60 deposited on Cu (100). The curves for the various coverages are displaced vertically for clarity and the scale factors indicated for each curve compensate for the relative coverage change and attenuation [17.51].

288 286 284 282 BINDING ENERGY (eV)

Fig. 17.5. (a) XPS spectra showing the XPS carbon Is peak (taken with Mg Ka radiation) of 0.5-ML Qo on Cu, Au, and Ag substrates compared to bulk CM. (b) Shift in the C Is peak with increasing C^ coverage, compared to bulk Cut. The points indicate an Ag substrate (o), Cu substrate (□), Au substrate (x). On all three substrates, bulk CM and six or more ML of Cm yield the same Is peak energy, as shown in the figure [17.50]. (c) X-ray photoemission spectra of the C (Is) core level for Cm films at coverages of 0.24, 0.52, 1.0, 1.7, 2.7, and 4 ML of C60 deposited on Cu (100). The curves for the various coverages are displaced vertically for clarity and the scale factors indicated for each curve compensate for the relative coverage change and attenuation [17.51].

(§17.2.1). Electron diffraction studies are directed toward elucidating either the structure within the first few atomic layers from the surface or the structure of very small samples, utilizing the high interaction cross section of electron probes with matter. Regarding inelastic electron energy loss spectroscopy (EELS), most of the effort has been directed toward study of interband transitions within a few electron volts of the Fermi level or transitions from core levels to levels near the Fermi level. EELS also provides a powerful tool for the investigations of plasmons and their dispersion relations. In such measurements, the electron energy loss function Im[-l/e(w)] is determined because of the sensitivity of Im[—l/e(w)] to plasmons, since the plasma frequency is defined by the vanishing of e,(w). The function Im[—l/e(<o)] is determined either at fixed electron momentum (wave vector) transfer as a function of electron energy (hco) or at fixed energy as a function of momentum transfer. EELS thus provides information on the dielectric function over a range of q values, whereas optical probes are confined to q values very close to zero. Using high-resolution techniques, vibrational spectra can also be explored with the EELS technique in a complementary way to infrared and Raman scattering, allowing many modes that are silent using infrared or Raman spectroscopy to become observable, but the vibrational modes observed by the EELS technique have much lower energy resolution. Progress with the EELS techniques for the investigation of fullerenes and related materials is reviewed in the following subsections.

17.2.1. Elastic Electron Scattering and Low-Energy Electron Diffraction Studies

Because of the strong interaction between electron probes and matter, electron diffraction is a method of choice for the structural analysis of materials only available in very small quantities, such as higher-mass fullerenes, or for the structural analysis of only a few atomic layers on a surface. Electron diffraction data are now available for the structure of crystalline C60, C70, C76, and Cg4 (see Fig. 17.6) [17.23]. The results show that each of these molecules crystallizes in a face-centered cubic (fee) structure with a lattice constant afcc ~ ^/n^, where nc is the number of carbon atoms in the fullerene. The relation between a[cc and nc is consistent with a shell model for fullerenes and a diffraction lineshape determined by a simple spherical form factor sin(qr)/(qr) for these fullerenes, where r is the average radius of the fullerene and q is a distance in reciprocal space. The distance df between fullerenes in the fee lattice is about the same for all of these fullerenes (df ~ 2.9 Â) and is related to the lattice constant a{cc of the fee

Fig. 17.6. Electron diffraction profile of a thin film of C76 recorded in the EELS spectrometer with a momentum transfer resolution of 0.06 Â"1. The inset is a plot of the fee lattice constant atcc as determined from such measurements for various fullerenes with nc carbon atoms, all crystallizing in the fee structure [17.23],

Fig. 17.6. Electron diffraction profile of a thin film of C76 recorded in the EELS spectrometer with a momentum transfer resolution of 0.06 Â"1. The inset is a plot of the fee lattice constant atcc as determined from such measurements for various fullerenes with nc carbon atoms, all crystallizing in the fee structure [17.23],

MOMENTUM TRANSFER (Â"1)

MOMENTUM TRANSFER (Â"1)

lattice and to the fullerene radius r by

Good agreement is obtained between the EELS diffraction data and selected area diffraction studies done by high-resolution transmission electron microscopy (TEM) techniques [17.23] yielding values of a0 = 14.2, 15.0, 15.3, and 15.8 A for Qo> C70, C76, and C84, respectively, to an accuracy of ±0.1 A [17.26]. The deviation of the data point for C70 in the inset to Fig. 17.6 from the straight line may be associated with the nonspherical shape of the C70 molecule.

Low-energy electron diffraction has also been used to probe the surface structure of crystalline fullerene surfaces adsorbed to various substrates. LEED studies of C60 on unheated Cu surfaces showed poor LEED patterns for less than two or three monolayers of C60 on Cu (100), Cu (110), and Cu (111) surfaces. However, for Cu substrates heated to ~300°C, epitaxial, single-domain growth occurred for thicker C60 growth on Cu (111). Good epitaxy was also achieved for the heated Cu (100) and Cu (110) surfaces, but with a strongly preferred (111) orientation for C60 growth, since the Cu (111) surface has the lowest surface energy [17.51],

17.2.2. Electronic Transitions near the Fermi Level

Low-energy electronic transition studies by the EELS technique for primary electrons of 10 eV or less have provided complementary information to optical probes. EELS results have been especially important for understanding the electronic structure near the HOMO-LUMO gap. Because of differences in the interaction Hamiltonian for electron probes, spin flipping transitions can occur in the electronic excitation from the ground state (h"') to the lowest excited state {h9ut\u), allowing direct identification of the lowest triplet exciton level (denoted as 71,) at an excitation energy of ~1.55 eV [17.11] (see §13.1). The EELS spectrum in Fig. 17.7 for a C60 multilayer, in fact, shows two sharp lines (one at 1.55 eV and one at 1.84 eV). The onset of the more intense line at 1.84 eV is identified with the absorption edge to the singlet exciton manifold of the h9ut\u configuration, which peaks at 2.2 eV, the intensity at higher energies coming from larger momentum transfer, with the peak corresponding to a momentum transfer of

Was this article helpful?

0 0

Post a comment