- Email: [email protected]
Control of chemical reactions by core excitations
Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255–266 www.elsevier.nl / locate / elspec
Control of chemical reactions by core excitations a, a b a a Kenichiro Tanaka *, Erika O. Sako , E. Ikenaga , Koji Isari , Saydul A. Sardar , Shin-ichi Wada a , Tetsuji Sekitani a , Kazuhiko Mase c , Nobuo Ueno d a
b
Department of Physical Science, Hiroshima University, Higashi-Hiroshima 739 -8526, Japan Hiroshima Synchrotron Radiation Center, Hiroshima University, Higashi-Hiroshima 739 -8526, Japan c Institute of Materials Structure Science, Tsukuba 305 -0801, Japan d Department of Materials Technology, Chiba University, Inage-ku, Chiba 263 -8522, Japan Received 8 February 2000; received in revised form 5 March 2001; accepted 25 March 2001
Abstract Recent progress in synchrotron radiation techniques in a soft X-ray region has developed a new field in chemistry. Using tunable synchrotron radiation in a soft X-ray region, site selective core excitation can be achieved. The site-specificity of the consequent chemical reactions is one of the attractive issues and has been studied for various kinds of molecules. During these studies, a remarkable site-specific reaction has been found in the case of photon stimulated ion desorption (PSID) of polymethylmethacrylate (PMMA) thin films. The PSID of PMMA has been examined in order to elucidate the reaction mechanism using total electron yield, total and partial ion yield, Auger electron, and Auger electron–photoion coincidence spectroscopies. In this paper we discuss the possibility of controlling chemical bond scission by using the site-specific photochemical surface reactions mainly based on the recent experimental results. 2001 Elsevier Science B.V. All rights reserved. Keywords: Control of chemical reaction; Core excitation; Photon stimulated desorption (PSD); Auger electron spectroscopy (AES); Poly-methylmethacrylate (PMMA)
1. Introduction Active control of chemical reaction using electromagnetic radiation is one of the main issues of photochemistry. Recent publications in the literature have demonstrated that laser control of chemical reactions is promising approach to this goal [1]. Active control can be achieved by using quantum mechanical interference between the different pathways to determine the outcome, but is less well established. Site-specific photochemical (photo-frag*Corresponding author. Fax: 181-824-24-7401. E-mail address: [email protected] (K. Tanaka).
mentation) reactions by core excitation offer an alternative and complementary approach to the control of chemical reactions. Core electrons are generally localized at a specific atom in a molecule and the differences in their binding energies for different atoms are large enough to allow selective excitation of different atoms, or even the same atom in chemically inequivalent environments within the same molecule. By using tunable soft X-rays, siteselective resonant core excitation therefore becomes possible. After such core excitation, two or more holes are formed in valence orbitals via an Auger process and site-selective bond breaking is expected in the vicinity of the atom where the primary core
0368-2048 / 01 / $ – see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S0368-2048( 01 )00301-2
256
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
Fig. 1. The concept of a ‘Molecular Scalpel’ where a tunable soft X-ray can be used as a scalpel.
excitation takes place. This demonstrates, as illustrated in Fig. 1, the possibility that a chemical bond in a molecular system can be cut selectively, by using a tunable soft X-ray. This is termed ‘molecular scalpel’ and is a promising candidate for the control of chemical reactions. However, there is a serious question about the site-selectivity of the bond scission, since there is the possibility of losing the site-specific character of the core excitation through various competitive channels such as the de-localization of excess energy. Historically, in 1983 Eberhardt et al. [2] presented the first observation of site-specific photofragmentation for the acetone molecule in the gas phase. C 1 and O 1 ions were selectively observed following the carbon 1s (C=O) to p*(C=O) excitation. Eberhardt et al. also foresaw the use of tunable soft X-rays as scalpel-like tools for breaking large organic molecules at certain selectable atomic sites (C, N, or O). This work gave a start to a great many researchers in this field. Similar experiments have since been conducted for many gaseous molecules, details of most of which are listed in Table 3 of the review article by Hanson [3]. However, for many molecules
other than acetone, less prominent site-specificity has been observed. Photon-stimulated ion desorption (PSID) from surfaces has also been studied. There was found to be a difference in ion yield between the gas phase and the surface [3–5]. For example, for H 2 O [4] and F 3 SiCH 2 CH 2 Si(CH 3 ) 3 [5], many kinds of ion are observed following core excitation in the gas phase, whereas, in the case of surface molecule, some kinds of ion are suppressed and the desorbed ions show a strong dependence on the excitation energy. This is due to the interaction with neighboring molecules in the condensed phase. In order to explain the differences in ionic fragmentation reaction between in the gas-phase and at the surface, we proposed a hypothetical reaction scheme [6]. Briefly, as shown in Fig. 2, we considered a primary core excitation and Auger decay is very fast processes within 10 214 s, thus there are essentially no deference through an Auger final state between the gas-phase and the surface. Generally Auger final states have large excess energies of
Fig. 2. Scheme of reaction paths for ionic fragmentation following core excitation.
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
10–30 eV. Thus, we assumed that there are two reaction paths leading to fragmentation reactions. One is the statistical reaction path in which the excess energy delocalizes (diffuses over the molecule) and redistribution of this energy leads to variety of chemical reactions. The probability of the reactions in this reaction path can be determined by statistical considerations and the reactions lose the memory of the site of the primary excitation. The other is a non-statistical reaction path in which direct and site-specific reactions occur. In the case of isolated gaseous molecule, the excess energy that diffuses over the molecule can lead to statistical chemical reactions and such statistical reactions are dominant, thus the site-specific (non-statistical) reaction is hidden. On the other hand, in the case of solid surfaces, the excess energy that diffuses over the molecule rapidly flow into the solid or neighboring molecules, thus it does not contribute to chemical reactions. Statistical reactions are thus suppressed and site-specific reactions are emphasized at the solid surfaces. According to this hypothesis site-specific reactions may well be more prominent at the surface than in the gas phase. This would seem advantageous for the surface reaction from the viewpoint of controlling the scission of chemical bonds. Because of the above situation, we investigated the photon-stimulated ion desorption (PSID) induced by core excitation for adsorbates, thin films and molecules condensed on a substrate, from the viewpoint of the site-specificity of the consequent chemical reactions. During these researches, remarkably site-specific reactions have been found in the PSID of poly-methylmethacrylate (PMMA) thin films [7,8]. Several fragment ions were observed in the carbon and oxygen core excitation regions and showed characteristic dependence on the primary excitation. For example, CH 1 3 ions are efficiently observed at the resonant core excitations of C1s and O1s in the side chain –OCH 3 group to the s*(O– CH 3 ) unoccupied excited state (289.4, 535.6 eV), and CHO 1 ions are efficiently observed at the resonant core excitation of O1s in the side chain –OCH 3 group to s*(C–OCH 3 ) unoccupied excited state (539.3 eV). These findings may suggest that the O–CH 3 bond and the C–OCH 3 bond of PMMA are selectively broken if the exciting energy is tuned to
257
289.4 eV (or 535.6 eV) and 539.3 eV, respectively. This was ever the first observation of site-specific ion desorption for a solid surface. In this way, the use of resonant core excitation and the slight change of excitation energies by several eV might enable the active control of chemical bond scission. Recently the mechanism of site-specific PSID of PMMA has been studied in detail by using an Auger electron–photoion coincidence spectroscopy and ab initio MO calculations. In this paper, the significance and more important results of the investigation of PSID of PMMA are briefly reviewed, and the possibility of controlling the scission of chemical bonds is discussed.
2. Photon-stimulated ion desorption (PSID) The photon-stimulated ion desorption (PSID) by core excitation of various surfaces has been shown to be very useful for probes for studying surface atom– atom and adsorbate–substrate interactions. Several models were proposed to explain the mechanism of PSID. Knotek and Feibelman proposed a model (KF model) based on core–hole Auger decay to account for the desorption of positive ions from ionic compounds [9–12]. The basic idea is that intra- and interatomic Auger decays of a core hole remove electrons from anions in sufficient numbers to make them positive. The positive ions are then repelled by their positive neighbors strongly enough to cause dissociation of molecule and desorption from the surface. Later, a generalized Auger stimulated ion desorption (ASID) model was proposed by Ramaker et al. [13–15]. This model takes into account the effects of hole–hole repulsive interactions and can be applied to covalent, ionic and chemisorbed systems. In a covalent system the Auger process creates two holes localized in a bonding orbital. If the holes remain localized for a sufficient long time in the bonding orbital, the expulsion of a positive ion takes place as a result of the unshielded nuclear–nuclear repulsion (also called hole–hole repulsion). The socalled multi-excited two holes, one electron (2h–1e) states are fundamental in initiating dissociation. In these 2h–1e states, the expulsion of the ion usually results from the emptying of a bonding orbital (2h) together with the occupation of a strongly anti-
258
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
bonding orbital (1h). These 2h–1e states are usually created via spectator Auger decays of the resonant core excited state. The stability of a chemical bond can be evaluated by the electron density between the atoms which constitute the bond. Bonding orbital normally has a higher electron density between the nuclei than in separated atoms, whereas the anti-bonding orbitals have a lower electron density between the nuclei. Thus, the strength of a bond can be represented in a very simple way by the bond order. The bond order is defined as the number of pairs of electrons in bonding orbitals minus the number of pairs localized in anti-bonding orbitals. Bond orders of 1, 2 and 3 correspond to what are conventionally called single, double and triple bonds. Bonds with bond order inferior to 1 are weak and very likely to break. Within this bond order picture, it is clear that chemical bonds in the multi-excited, 2h–1e states can be easily broken, since they can correspond to bond orders of 21.5 when two electrons are removed from valence bonding orbitals and one electron fills an anti-bonding orbital. The photodesorption of ions is strongly influenced by quenching processes, i.e., by delocalization of the excitation energy and charge. The site selectivity character of ion dissociation reactions is directly related with the electronic and nuclear motion time scales (initial electronic excitation |10 216 s; Auger decay |10 215 –10 214 s; two hole decay |10 214 s; desorption |10 214 –10 213 s). The primary excitation, its decay, and the bond breaking events should not be considered as completely separated in time as in a Frank–Condon model. The requirement of long localization time of two holes for desorption to occur was recently questioned by Menzel [16]. The existence of steep repulsive potentials for dissociative excited states can lead to the ultrafast dissociation which can occur even before the core hole decay. Some evidence for the ultrafast dissociation was given for H 1 and D 1 desorption from surface adsorbed H 2 O [17–20], NH 3 [21–23] and benzene [23]. In the case of benzene the strongest selective dissociation channel was coupled to a decay autoionization spectrum which was very different from all the others and was not explainable by decay of an intact molecular benzene. This fact was taken as evidence that dramatic structural
changes have happened in the molecule before core decay [23]. The time scale competition between the core hole decay and dissociation is still involved in much controversy. Any analysis of site selective dissociation should be done with great care. Presently, intense research is being devoted to the dynamic effects of core-excitations concerning the core hole localization and desorption time scales.
3. Site-specific reactions Surface studies of PSD processes led to the concept of localized valence-hole states. The essential idea is that Auger decay of a core hole produces two valence holes associated with a single atom because the core hole was localized on that atom. This localization of two holes may lead to atomic site-specific reactions. It is known that the quenching of the slow ion desorption processes is more effective on surfaces than on isolated molecules. This quenching would lead to a selective enhancement of fast processes and consequently facilitate the observation of site-specific reactions on surfaces. The studies of PSD of molecules condensed (or adsorbed) on metallic surfaces have revealed the importance of ESD process caused by photoelectrons and secondary electrons created in the decay process after the X-ray radiation (XESD, X-ray-induced ESD). Depending on the character of the substance and experimental conditions, the investigation of site-specific direct photodesorption from surfaces is limited by the dominance of the indirect ion desorption. The photon-to-electron conversion step underlying the XESD process may transfer the excitation to an adsorbate atom with no relation to the originally photoionized. In fact, in some cases, the ion signal does not yield surface-specific information at all but is simply a replica of the total electron yield (TEY) spectrum originating from the adsorbate bulk (or substrate bulk). This effect is most important for covalently bonded surface systems and at high photon excitation energies since the production of secondary electrons increases with excitation energy. This mixing of PSD and XESD prevents a clear analysis of the surface due to the background produced by the bulk. In other words, it is difficult to
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
discern any structural changes of true PSD from the XESD background. The contribution of XESD has been described by Jaeger et al. [24] for the NH 3 / Ni system. From the comparison of the total electron yield and H 1 yield of multilayers of NH 3 with variable thickness on Ni(110), it was found that the ion desorption is dominated by secondary-electron-induced valence excitations in the outermost layer. The direct core hole excitation of a surface molecule can be observed by a fine structural feature at 400.5 eV which is only present in the H 1 yield and not in the TEY spectrum in the N K-edge. Using this feature the contribution of direct PSD and XESD could be estimated to be 40 and 60%, respectively. The investigation of PSD from condensed and adsorbed water on Ru(001) also showed the importance of the XESD contribution for H 1 desorption [17]. The Auger electron and H 1 yields were measured from multilayer, bilayer and monolayer ice samples in the near oxygen K-edge region. At 534 eV, the H 1 signal showed a sharp peak with no corresponding feature in the Auger absorption spectrum. At about 535–542 eV, the H 1 signal showed a broad peak with good correspondence in the Auger spectrum. These two features were, respectively, named threshold peak and main peak and showed different saturation and polarization behaviors. These differences can be understood if we assume that the threshold peak is a true photodesorption feature caused by photoabsorption of the topmost layer, while the main peak contains a large contribution from the XESD and is dominated by bulk adsorption. Since these bulk electrons have insufficient energy to cause core excitations of the topmost layer, multiple valence excitation must be the bond breaking process. It was concluded that in mono- and bi-layer range the observed structures are true primary PSD events. On the other hand, in the multilayer range, the observed threshold peak is a true PSD feature while approximately half of the main peak is due to XESD caused by the numerous electrons liberated at condensate excitations. In all previous examples, direct PSD features can still be observed despite the background contribution of XESD. For these direct PSD features, it was observed that the transitions of the condensate are very similar to those of the free molecule. The larger
259
the alteration in symmetry and in bond strength (due to close interactions with neighboring molecules), the larger the shifts in energy and broadening of the peaks. The selective character of dissociation is pronounced and reflects the large dissociative character of the excited state formed by core excitation. This character is enhanced by the atomic motion that occurs to a small extent before or during core hole decay.
4. Experimental
4.1. Ion yield measurements The time-of flight (TOF) mass spectrometric technique has been used for the PSID investigations with great success. The TOF mass spectrometer combined with the pulsed synchrotron radiation used in this work is similar to that described previously by Knotek et al. [25], and is described in detail elsewhere [26]. Briefly, it consists of an accelerating plate, a drift tube of 9 cm, and a microchannel plate detector. Due to its small size it can be installed very close to the samples. In our experimental settings, it was placed 3 cm away from the sample surface in its normal direction. During single bunch operation of the Photon Factory storage ring, soft X-ray pulses with a period of 624 ns and width of 100 ps are generated. The sample is irradiated by this pulsed soft X-ray beam through a photon-flux monitor of the Au grid. The incident flux spectrum is recorded simultaneously as the photocurrent at the Au grid. All spectra are normalized by this flux spectrum to correct for fluctuations in beam intensity. As a result of irradiation of pulsed soft X-ray beam, ionic fragments are desorbed from the sample. These ions are subjected to an accelerating potential, mass analyzed in the drift tube, and collected by a microchannel plate detector. In the TOF normal operation, TOF the ions are measured using a timeto-amplitude converter (TAC-EG&G Ortec 567). The flight time is measured between the SR pulse and the amplified ion signal. A 1 / 312 divider is used to take the SR pulse timing the 500-MHz frequency signal from the microwave cavity of the storage ring. The total spectra are read and accumulated by an
260
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
analog-to-digital converter and a multichannel analyzer (Laboratory Equipments MCA / PC98A). Due to the very good time reproducibility of the SR pulses, heavy ions which have flight times longer than 624 ns can also be measured. For these heavier ions the time-of-flight is given by its position in the spectrum plus an integer multiple of the SR pulse interval (624 ns) corresponding to the number of cycles that have passed from the production of the ion to its detection. The mass assignments can be easily done by comparative spectra at different accelerating potential. Partial ion yield spectra (PIY) are measured by windowing the channels corresponding to the ion to be analyzed while scanning the photon energy. Better counting rates can be obtained using the TOF in reverse mode, i.e., measuring the time-of-flight of the ions starting at the MCP ion signal and stopping at the SR pulse. Both total electron yield absorption spectra (TEY) and total ion yield spectra (TIY) can be measured using the TOF mass spectrometer using positive or negative accelerating potentials, respectively.
power of CMA is typically 1% and determined by changing the size of the CMA slit. The axis of TOF-MS as well as CMA is located normal to the surface and the angle between the surface normal and SR was 558. The sample surface is excited by SR, and the energy of the emitted electrons is analyzed with the CMA, while the desorbed ions are accelerated towards the TOF-MS. The ion signal in coincidence with the detected electron gives a coincidence signal at a specific TOF, while signals irrelevant to the electron increase the background level. Since the energy-selected electron corresponds to a particular Auger transition, the coincidence signal intensity offers the yield of the ion desorption induced only by the Auger process. The EICO spectrometer was attached to an UHV chamber at the soft X-ray beamline BL13 of Hiroshima Synchrotron Radiation Center (HiSOR) using a dragon type spherical grating monochromator. The CMA and TOF-MS were also used for partial electron yield spectroscopy and total ion yield spectroscopy, respectively.
4.2. Electron–ion coincidence measurements
5. Results and discussion
Details of the PSID mechanism have mainly been discussed on the basis of the partial ion yield (PIY), i.e., the incident photon energy dependence on the yield of particular ion. In these studies the electronic transitions that correspond to thresholds or characteristic peaks in the yield curve were inferred to induce the ion desorption. This approach, however, offers no information about the intermediate Auger transitions. The coincidence measurement of ions with Auger electrons is a powerful approach for the PSID study, because the Auger final state is the direct precursor of ion desorption in the ASID mechanism [13–15]. Mase et al. [27,28] developed an electron–ion coincidence (EICO) apparatus and applied to the PSID studies [18,19,21,22,27–30]. The apparatus used in this work is recently remodeled one to improve the collection efficiency of electrons and ions [31,32]. It consists of a cylindrical mirror electron energy analyzer (CMA), a coaxial time-offlight ion mass spectrometer (TOF-MS), and an electronic system for measurements. The resolving
5.1. Ion yield spectra of PMMA The first study of the PSID from site-specific point of view was performed for PMMA (polymethylmethacrylate) thin films [7]. PMMA is a very suitable sample to use for elucidating the mechanism of site-specific PSID because its monomer contains five carbon atoms and two oxygen atoms, all of which have different chemical environments. Fig. 3 shows structure of the monomer unit of PMMA. Carbon and oxygen atoms at different sites are numbered according to decreasing binding energies. The relationship between the ion desorption and the primary excitation has been investigated in detail [7]. Several fragment ions such as H 1 , CH 1 , CH 1 2 , 1 CH 1 and COOCH 1 3 , CHO 3 were observed in the carbon and oxygen core excitation regions. Fig. 4 shows the partial ion yield spectra of typical ions together with a total electron yield spectrum of PMMA in the carbon and oxygen core excitation region. Several NEXAFS structures are observed in these spectra. The energies and proposed assign-
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
261
Table 1 Photon energies and possible assignments for the peaks observed in the carbon core excitation of a PMMA thin film Peak
A B B9 C D
E F
Energy
Assignment
(eV)
C=O
–O–CH 3
–C–CH 2 –
287.5 288.7 289.4 290.2 291.1 292.3 292.7 294.9 296.8 304.0
– p*(C=O) – – – s*(C–H) – IP s*(C–OCH 3 ) s*(C=O)
– – s*(O–CH 3 ) s*(C–H) – – IP – – –
s*(C–C) – – s*(C–C) IP s*(C–C) – – s*(C=O) –
Fig. 3. Structure of the monomer unit of PMMA.
ments of these structures are listed in Tables 1 and 2. The partial ion yield seems to produce a variety of linewidths for the most intense resonance compare to electron yield. This is because the ion yield spectrum reflects the electronic structure and reaction efficiency of the outermost surface of PMMA films.
On the other hand, the electron yield spectrum reflects the electronic structure of not only surface but also bulk because of its longer escape depth compare to ions. A difference is observed between ions with respect to the NEXAFS structure, and the ion desorption has a strong dependence on the initial
Fig. 4. Typical partial ion yield spectra of PMMA in the carbon and oxygen core excitation regions together with a total electron yield spectra.
262
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
Table 2 Photon energies and possible assignments for the peaks observed in the oxygen core excitation of a PMMA thin film Peak
Energy
(eV) A B C D E
531.5 534.3 535.6 537.8 539.3 539.4 546.8
Assignment C=O
–O–CH 3
p*(C=O) s*(C–OCH 3 ) – IP – – s*(C=O)
– p*(C=O) s*(O–CH 3 ) – s*(C–OCH 3 ) IP –
excitation. In particular, in the carbon core region (Fig. 4a), CH 1 and CH 21 ions are observed selectively at 289.4 eV which is slightly higher than the strongest peak at 288.7 eV assigned as a resonant transition of C1s(C=O) to p*(C=O) orbital. The yield spectrum of CH 1 3 ion, dominant desorbed ion, also has a shoulder at 289.4 eV. As shown in Table 1, the resonant transition at 289.4 eV has recently been assigned as a C1s(–OCH 3 )→s*(O–CH 3 ) transition [32]. In the oxygen core region (Fig. 4b), the 1 efficient production of CH 1 2 and CH 3 ions is observed for the transition of O1s (OCH 3 ) to s*(O– CH 3 ) orbital at 535.6 eV and the efficient production of CHO 1 ions is observed for the O1s(OCH 3 )→s*(C–OCH 3 ) transition at 539.3 eV. From these findings, it can be considered that the resonant core excitation plays an important role to determine the position of bond scission as well as to weaken the bond strength due to the anti-bonding character of the resonant excited state. The above result indicates that the localized character of the initial core excited configuration can be maintained after a spectator Auger decay. Furthermore, this effect can be observed by the changes in the ion yield spectra. The localized character of the antibonding excited state after the site-specific excitation, and the expulsion of the bonding electron near the core–hole atom by Auger decay may possibly induce the site-specific scission of the chemical bond combining with this atom.
directly related to the ion desorption by core excitation. The measurement of Auger electron–photoion coincidence is useful for clarifying the relationship between the Auger final state and the ion desorption. Because of the above reasons, we investigated the PSID of PMMA thin films, using the Auger electron–photoion coincidence (AEPICO) spectroscopy to clarify the Auger process that follows the resonant core excitation and the ion desorption mechanism related to the Auger process [32,33]. Fig. 5 shows a series of typical AEPICO spectra measured with the photon energy set for the s*(O–CH 3 ) resonant excitation (289.4 eV) at which CH 1 , CH 21 and CH 31 ions are selectively observed in their yield spectra (Fig. 4a). The Auger electron energy was sampled in 2.5 eV steps. Two coincidence peaks are clearly observed at about 420 and 1400–1700 ns with a good S /N ratio in the TOF difference spectra. These peaks are assigned as H 1 and CH n1 (n51–3) ions, respectively. The change in appearance of CH 1 n (n5
5.2. Auger electron–photoion coincidence (AEPICO) spectra of PMMA As explained by the ASID model [13–15], the final state of Auger decay (Auger final state) is
Fig. 5. AEPICO spectra of PMMA at the s*(O–CH 3 ) resonant excitation (289.4 eV).
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
1–3) AEPICO signal between Auger electron energies of 265 and 270 eV is of particular interest. In contrast to the broad peak of CH 1 n (n51–3) ions which are observed around 1450–1720 ns at 265 eV, only a single sharp peak is observed at 270 eV. The sharp peak is assigned to the CH 1 3 ion which is predominantly desorbed. The broad peak could be 1 deconvoluted into CH 1 , CH 1 2 and CH 3 components by Gaussian curve fitting [32]. The AEPICO yield for H 1 and CH 1 n ions are obtained by integrating the AEPICO signal intensity of each ion peak above the background level and are plotted in Fig. 6 together with the resonant Auger spectrum (AES) as a function of Auger electron energy. The CH 1 n AEPICO yield spectrum is remarkably enhanced in the higher Auger electron energy region
Fig. 6. AEPICO yield spectra of PMMA at the s*(O–CH 3 ) resonant excitation together with the resonant Auger spectrum as a function of Auger electron energy.
263
of AES (260–275 eV), while the H 1 AEPICO yield shows little dependence on the Auger electron energy. The enhancement of the CH 1 n AEPICO yield in the high Auger electron region may suggest that the spectator Auger transition following the C 2 1s→s*(O–CH 3 ) excitation occurs and produces two holes in the outer valence orbitals which strongly overlap with C 2 1s core hole and have a bonding character with respect to the O–CH 3 bonding. At the nearby resonant excitations, such as p*(C=O) at 288.7 eV and s*(C–H) at 290.2 eV, the CH 1 n AEPICO yield spectrum shows similar dependence on the Auger electron energy to that observed at the s*(O–CH 3 ) excitation at 289.4 eV, but rate of AEPICO yield is relatively low. The results can be explained on the basis of overlapping of these resonant transitions with the s*(O–CH 3 ) transition at 289.4 eV. Thus, the AEPICO yield spectra at the p*(C=O) and s*(C–H) excitation contain contributions of the s*(O–CH 3 ) excitation. It should be noted that the AEPICO yield spectra observed at 320 eV excitation in the ionization continuum, where a normal Auger decay occurs, are obviously different from those observed at above resonant excitations and show relatively low AEPICO yields and less structures. The most characteristic site-specific ion desorption in the carbon core excitation region is the desorption of CH 1 n (n51–3) by the resonant excitation of C1s in the side chain methyl group to the anti-bonding s*(O–CH 3 ) orbital. As mentioned before, the resonant core excitation weakens the O–CH 3 bonding and the subsequent spectator Auger decay produces two holes in the outer valence orbitals that may have a bonding character with respect to the bonding. In other words, two-hole one-electron spectator Auger final states (2h–1e) in which one electron is in an anti-bonding O–CH 3 orbital and the two holes in the valence bonding orbitals may effectively promote the selective O–CH 3 bond scission. In order to elucidate the mechanism of site-specific ion desorption in detail, we calculated the Auger transition probability and compared it with the results of AEPICO experiments [34]. PMMA, however, is too complicated molecules to characterize the Auger final state, thus a methylisobutyrate h(CH 3 ) 2 CHCOOCH 3 j molecule was used as a model molecule of PMMA in the calculations. The calculation is based on the single configuration state func-
264
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
tion (CSF) and limited configuration interaction (CI) methods, and Auger transition probabilities are estimated by overlap between core and valence molecular orbitals. In general, the resonant (spectator) Auger spectrum for a specific core excitation is similar to the normal Auger spectrum of the specific atom in the molecule only with a blue shift in energy due to the screening effect of the spectator electron. Therefore the normal Auger spectrum from this calculation can be compared with the s*(O–CH 3 ) resonant Auger spectrum. Fig. 7 shows the calculated Auger spectrum of a model molecule at the methoxy carbon site together with the experimental resonant Auger spectrum and the AEPICO yield spectrum for CH 1 n ions observed at the resonant core excitation of C 1s (OCH 3 )→s*(O–CH 3 ). The calculated Auger spectrum is shifted to align the position of the highest peaks at 256 eV. The calculated Auger spectrum reproduces the feature of the experimental one. In
the higher Auger electron energy region around 265 eV, where the AEPICO yield spectrum for CH 1 n ions is observed remarkably enhanced, calculated and experimental Auger spectra have a broad peak and a shoulder peak, respectively. As mentioned before, the Auger decay in this region is expected to produce two holes in the outer valence orbitals that have a bonding character with respect to the O–CH 3 bonding. To investigate this subject more precisely, the partial Auger transition probability was extracted from the total Auger transition probability. This indicates the probability of production of at least one hole in a certain molecular orbital (MO) by the Auger decay. The model molecule has 28 occupied MOs including seven inner core orbitals and these MOs are numbered from the core to valence orbitals. MOs from 15 to 28 are involved in the first broad peak around 265 eV of the calculated Auger spectrum. By careful comparison of the partial Auger spectra obtained for these MOs with the AEPICO yield spectrum for CH 1 n ions, MOs 16, 17, 19, and 24 remain appreciably important. These four MOs have a bonding character with respect to the O–CH 3 bonding. Fig. 8 shows the partial Auger spectra for these MOs and the sum of these four spectra together with the AEPICO yield spectrum. As can be seen from Fig. 8, the AEPICO yield spectrum for CH 1 n ions is in good agreement with the sum of partial Auger spectra. From this result, it can be concluded that the efficient desorption of CH 1 n ions occurs definitely through the Auger final states having two holes in these bonding orbitals. This finding demonstrates that the resonant core excitation of the atom that is directly connected to the particular bond to the anti-bonding orbital is of vital importance. Such excitation leads to effective elimination of the bonding electron near the atom by the resonant Auger decay process, and leads to selective scission of the chemical bond.
6. Conclusion
Fig. 7. Calculated Auger spectrum of methyl isobutyrate at the methoxy carbon site compared with the experimental resonant Auger spectrum and AEPICO yield spectrum for CH 1 n (n51–3) obtained at the s*(O–CH 3 ) resonant excitation.
As described above, site-specific surface reactions by core excitation offer an alternative and complementary approach to the active control of chemical reactions. The resonant core excitation of either atom at both ends of the particular bonding to the
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
265
the goal, further investigations in various molecular systems at different environments are needed.
Acknowledgements The authors wish to thank the staff of the PF and HiSOR for stable operation of the SR ring. This work was supported by a Grant-in-Aid on Research for the Future ‘Photoscience’ (JSPS-RFTF-98P01202) from Japan Society for the Promotion of Science (JSPS).
References
Fig. 8. Partial Auger spectra for four MOs (16, 17, 19, 24) and sum of these four partial Auger spectra compared with the AEPICO yield spectrum.
unoccupied anti-bonding orbital brings about weakening the bonding power, the effective elimination of bonding electrons, and the selective scission of chemical bond in the vicinity of the primarily excited atom. In this way, the position of the molecular scission can be controlled. This scenario, however, requires the precise and systematic investigation on the following subjects; the evaluation of the nature of the unoccupied anti-bonding orbital, the probability of resonant core transition from each atom at both ends of the chemical bonding to the anti-bonding orbital, and the partial Auger intensity for each bonding orbital and for each core hole at both ends of the chemical bonding. After such examinations, tuning the soft X-ray energy to the most probable resonant transition leads to the sitespecific bond scission. Finally, the ultimate goal of this kind of studies, we named ‘molecular scalpel’, is now not a dream and one step closer to realization. In order to achieve
[1] R.N. Zare, Science 279 (1998) 1875. [2] W. Eberhardt, T.K. Sham, R. Carr, S. Krummacher, M. Strongin, S.L. Weng, D. Wesner, Phys. Rev. Lett. 50 (1983) 1038. [3] D.M. Hanson, Adv. Chem. Phys. 77 (1990) 1. [4] T. Sekiguchi, H. Ikeura, K. Tanaka, K. Obi, N. Ueno, K. Honma, J. Chem. Phys. 102 (1995) 1422. [5] S. Nagaoka, K. Mase, I. Koyano, Trends Chem. Phys. 6 (1997) 1. [6] T. Sekitani, E. Ikenaga, K. Fujii, K. Mase, N. Ueno, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 101–103 (1999) 135. [7] M.C.K. Tinone, K. Tanaka, J. Maruyama, N. Ueno, M. Imamura, N. Matsubayashi, J. Chem. Phys. 100 (1994) 5988. [8] N. Ueno, K. Tanaka, Jpn. J. Appl. Phys. 36 (1997) 7605. [9] M.L. Knotek, P.J. Feibelman, Phys. Rev. Lett. 40 (1978) 964. [10] P.J. Feibelman, M.L. Knotek, Phys. Rev. B 18 (1978) 6531. [11] M.L. Knotek, P.J. Feibelman, Surf. Sci. 90 (1979) 78. [12] P.J. Feibelman, Surf. Sci. 102 (1981) L51. [13] D.E. Ramaker, C.T. White, J.S. Murday, J. Vac. Sci. Technol. 18 (1981) 748. [14] D.E. Ramaker, C.T. White, J.S. Murday, Phys. Lett. A 84 (1982) 211. [15] D.E. Ramaker, in: N.H. Tolk, M.M. Traum, J.C. Tully, T.E. Madey (Eds.), Desorption Induced by Electronic Transitions, DIET-I, Springer Series in Chem. Phys, Vol. 24, Springer, Heidelberg, 1983, p. 70. [16] D. Menzel, in: A.R. Burns, E.B. Stechel, D.R. Jennison (Eds.), Desorption Induced by Electronic Transitions, DIETV, Springer Series in Surf. Sci, Vol. 31, Springer, Berlin, 1993, p. 101. ¨ ¨ [17] D. Coulman, A. Puschmann, U. Hofer, H.-P. Steinruck, W. Wurth, P. Feulner, D. Menzel, J. Chem. Phys. 93 (1990) 58. [18] K. Mase, M. Nagasono, S. Tanaka, T. Urisu, E. Ikenaga, T. Sekitani, K. Tanaka, Surf. Sci. 390 (1997) 97. [19] K. Mase, M. Nagasono, S. Tanaka, T. Urisu, E. Ikenaga, T. Sekitani, K. Tanaka, J. Chem. Phys. 108 (1998) 6550.
266
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
[20] R. Scheuerer, P. Feulner, P. Wiethoff, W. Wurth, D. Menzel, J. Electron Spectrosc. Relat. Phenom. 75 (1995) 16. [21] M. Nagasono, K. Mase, S. Tanaka, T. Urisu, Surf. Sci. 390 (1997) 102. [22] M. Nagasono, K. Mase, S. Tanaka, T. Urisu, Jpn. J. Appl. Phys. 38 (Suppl. 1) (1999) 325. [23] D. Menzel, G. Rocker, H.-P. Steinruck, D. Coulman, P.A. Heinmann, W. Huber, P. Zebisch, D.R. Lloyd, J. Chem. Phys. 96 (1992) 1724. [24] R. Jaeger, J. Sohr, T. Kendelewicz, Surf. Sci. 134 (1983) 547. [25] M.L. Knotek, V.O. Jones, V. Rehn, Phys. Rev. Lett. 43 (1979) 300. [26] K. Tanaka, M.C.K. Tinone, H. Ikeura, T. Sekiguchi, T. Sekitani, Rev. Sci. Instrum. 66 (1995) 1474. [27] K. Mase, M. Nagasono, T. Urisu, Y. Murata, Bull. Chem. Soc. Jpn. 69 (1996) 1829. [28] K. Mase, M. Nagasono, S. Tanaka, M. Kamada, T. Urisu, Y. Murata, Rev. Sci. Instrum. 68 (1997) 1703.
[29] T. Sekitani, E. Ikenaga, H. Matsuo, S. Tanaka, K. Mase, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 88 (1996) 731. [30] T. Sekitani, E. Ikenaga, K. Tanaka, K. Mase, M. Nagasono, S. Tanaka, T. Urisu, Surf. Sci. 390 (1997) 107. [31] K. Mase, S. Tanaka, S. Nagaoka, T. Urisu, Surf. Sci. 451 (2000) 143. [32] E. Ikenaga, K. Isari, K. Kudara, Y. Yasui, S.A. Sarder, S. Wada, T. Sekitani, K. Tanaka, K. Mase, S. Tanaka, J. Chem. Phys. 114 (2001) 2751. [33] E. Ikenaga, K. Kudara, K. Kusaba, K. Isari, S.A. Sarder, S. Wada, K. Mase, T. Sekitani, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 114–116 (2001) 585. [34] E.O. Sako, Y. Kanameda, E. Ikenaga, M. Mitani, O. Takahashi, K. Saito, S. Iwata, S. Wada, T. Sekitani, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 114–116 (2001) 591.
Control of chemical reactions by core excitations a, a b a a Kenichiro Tanaka *, Erika O. Sako , E. Ikenaga , Koji Isari , Saydul A. Sardar , Shin-ichi Wada a , Tetsuji Sekitani a , Kazuhiko Mase c , Nobuo Ueno d a
b
Department of Physical Science, Hiroshima University, Higashi-Hiroshima 739 -8526, Japan Hiroshima Synchrotron Radiation Center, Hiroshima University, Higashi-Hiroshima 739 -8526, Japan c Institute of Materials Structure Science, Tsukuba 305 -0801, Japan d Department of Materials Technology, Chiba University, Inage-ku, Chiba 263 -8522, Japan Received 8 February 2000; received in revised form 5 March 2001; accepted 25 March 2001
Abstract Recent progress in synchrotron radiation techniques in a soft X-ray region has developed a new field in chemistry. Using tunable synchrotron radiation in a soft X-ray region, site selective core excitation can be achieved. The site-specificity of the consequent chemical reactions is one of the attractive issues and has been studied for various kinds of molecules. During these studies, a remarkable site-specific reaction has been found in the case of photon stimulated ion desorption (PSID) of polymethylmethacrylate (PMMA) thin films. The PSID of PMMA has been examined in order to elucidate the reaction mechanism using total electron yield, total and partial ion yield, Auger electron, and Auger electron–photoion coincidence spectroscopies. In this paper we discuss the possibility of controlling chemical bond scission by using the site-specific photochemical surface reactions mainly based on the recent experimental results. 2001 Elsevier Science B.V. All rights reserved. Keywords: Control of chemical reaction; Core excitation; Photon stimulated desorption (PSD); Auger electron spectroscopy (AES); Poly-methylmethacrylate (PMMA)
1. Introduction Active control of chemical reaction using electromagnetic radiation is one of the main issues of photochemistry. Recent publications in the literature have demonstrated that laser control of chemical reactions is promising approach to this goal [1]. Active control can be achieved by using quantum mechanical interference between the different pathways to determine the outcome, but is less well established. Site-specific photochemical (photo-frag*Corresponding author. Fax: 181-824-24-7401. E-mail address: [email protected] (K. Tanaka).
mentation) reactions by core excitation offer an alternative and complementary approach to the control of chemical reactions. Core electrons are generally localized at a specific atom in a molecule and the differences in their binding energies for different atoms are large enough to allow selective excitation of different atoms, or even the same atom in chemically inequivalent environments within the same molecule. By using tunable soft X-rays, siteselective resonant core excitation therefore becomes possible. After such core excitation, two or more holes are formed in valence orbitals via an Auger process and site-selective bond breaking is expected in the vicinity of the atom where the primary core
0368-2048 / 01 / $ – see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S0368-2048( 01 )00301-2
256
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
Fig. 1. The concept of a ‘Molecular Scalpel’ where a tunable soft X-ray can be used as a scalpel.
excitation takes place. This demonstrates, as illustrated in Fig. 1, the possibility that a chemical bond in a molecular system can be cut selectively, by using a tunable soft X-ray. This is termed ‘molecular scalpel’ and is a promising candidate for the control of chemical reactions. However, there is a serious question about the site-selectivity of the bond scission, since there is the possibility of losing the site-specific character of the core excitation through various competitive channels such as the de-localization of excess energy. Historically, in 1983 Eberhardt et al. [2] presented the first observation of site-specific photofragmentation for the acetone molecule in the gas phase. C 1 and O 1 ions were selectively observed following the carbon 1s (C=O) to p*(C=O) excitation. Eberhardt et al. also foresaw the use of tunable soft X-rays as scalpel-like tools for breaking large organic molecules at certain selectable atomic sites (C, N, or O). This work gave a start to a great many researchers in this field. Similar experiments have since been conducted for many gaseous molecules, details of most of which are listed in Table 3 of the review article by Hanson [3]. However, for many molecules
other than acetone, less prominent site-specificity has been observed. Photon-stimulated ion desorption (PSID) from surfaces has also been studied. There was found to be a difference in ion yield between the gas phase and the surface [3–5]. For example, for H 2 O [4] and F 3 SiCH 2 CH 2 Si(CH 3 ) 3 [5], many kinds of ion are observed following core excitation in the gas phase, whereas, in the case of surface molecule, some kinds of ion are suppressed and the desorbed ions show a strong dependence on the excitation energy. This is due to the interaction with neighboring molecules in the condensed phase. In order to explain the differences in ionic fragmentation reaction between in the gas-phase and at the surface, we proposed a hypothetical reaction scheme [6]. Briefly, as shown in Fig. 2, we considered a primary core excitation and Auger decay is very fast processes within 10 214 s, thus there are essentially no deference through an Auger final state between the gas-phase and the surface. Generally Auger final states have large excess energies of
Fig. 2. Scheme of reaction paths for ionic fragmentation following core excitation.
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
10–30 eV. Thus, we assumed that there are two reaction paths leading to fragmentation reactions. One is the statistical reaction path in which the excess energy delocalizes (diffuses over the molecule) and redistribution of this energy leads to variety of chemical reactions. The probability of the reactions in this reaction path can be determined by statistical considerations and the reactions lose the memory of the site of the primary excitation. The other is a non-statistical reaction path in which direct and site-specific reactions occur. In the case of isolated gaseous molecule, the excess energy that diffuses over the molecule can lead to statistical chemical reactions and such statistical reactions are dominant, thus the site-specific (non-statistical) reaction is hidden. On the other hand, in the case of solid surfaces, the excess energy that diffuses over the molecule rapidly flow into the solid or neighboring molecules, thus it does not contribute to chemical reactions. Statistical reactions are thus suppressed and site-specific reactions are emphasized at the solid surfaces. According to this hypothesis site-specific reactions may well be more prominent at the surface than in the gas phase. This would seem advantageous for the surface reaction from the viewpoint of controlling the scission of chemical bonds. Because of the above situation, we investigated the photon-stimulated ion desorption (PSID) induced by core excitation for adsorbates, thin films and molecules condensed on a substrate, from the viewpoint of the site-specificity of the consequent chemical reactions. During these researches, remarkably site-specific reactions have been found in the PSID of poly-methylmethacrylate (PMMA) thin films [7,8]. Several fragment ions were observed in the carbon and oxygen core excitation regions and showed characteristic dependence on the primary excitation. For example, CH 1 3 ions are efficiently observed at the resonant core excitations of C1s and O1s in the side chain –OCH 3 group to the s*(O– CH 3 ) unoccupied excited state (289.4, 535.6 eV), and CHO 1 ions are efficiently observed at the resonant core excitation of O1s in the side chain –OCH 3 group to s*(C–OCH 3 ) unoccupied excited state (539.3 eV). These findings may suggest that the O–CH 3 bond and the C–OCH 3 bond of PMMA are selectively broken if the exciting energy is tuned to
257
289.4 eV (or 535.6 eV) and 539.3 eV, respectively. This was ever the first observation of site-specific ion desorption for a solid surface. In this way, the use of resonant core excitation and the slight change of excitation energies by several eV might enable the active control of chemical bond scission. Recently the mechanism of site-specific PSID of PMMA has been studied in detail by using an Auger electron–photoion coincidence spectroscopy and ab initio MO calculations. In this paper, the significance and more important results of the investigation of PSID of PMMA are briefly reviewed, and the possibility of controlling the scission of chemical bonds is discussed.
2. Photon-stimulated ion desorption (PSID) The photon-stimulated ion desorption (PSID) by core excitation of various surfaces has been shown to be very useful for probes for studying surface atom– atom and adsorbate–substrate interactions. Several models were proposed to explain the mechanism of PSID. Knotek and Feibelman proposed a model (KF model) based on core–hole Auger decay to account for the desorption of positive ions from ionic compounds [9–12]. The basic idea is that intra- and interatomic Auger decays of a core hole remove electrons from anions in sufficient numbers to make them positive. The positive ions are then repelled by their positive neighbors strongly enough to cause dissociation of molecule and desorption from the surface. Later, a generalized Auger stimulated ion desorption (ASID) model was proposed by Ramaker et al. [13–15]. This model takes into account the effects of hole–hole repulsive interactions and can be applied to covalent, ionic and chemisorbed systems. In a covalent system the Auger process creates two holes localized in a bonding orbital. If the holes remain localized for a sufficient long time in the bonding orbital, the expulsion of a positive ion takes place as a result of the unshielded nuclear–nuclear repulsion (also called hole–hole repulsion). The socalled multi-excited two holes, one electron (2h–1e) states are fundamental in initiating dissociation. In these 2h–1e states, the expulsion of the ion usually results from the emptying of a bonding orbital (2h) together with the occupation of a strongly anti-
258
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
bonding orbital (1h). These 2h–1e states are usually created via spectator Auger decays of the resonant core excited state. The stability of a chemical bond can be evaluated by the electron density between the atoms which constitute the bond. Bonding orbital normally has a higher electron density between the nuclei than in separated atoms, whereas the anti-bonding orbitals have a lower electron density between the nuclei. Thus, the strength of a bond can be represented in a very simple way by the bond order. The bond order is defined as the number of pairs of electrons in bonding orbitals minus the number of pairs localized in anti-bonding orbitals. Bond orders of 1, 2 and 3 correspond to what are conventionally called single, double and triple bonds. Bonds with bond order inferior to 1 are weak and very likely to break. Within this bond order picture, it is clear that chemical bonds in the multi-excited, 2h–1e states can be easily broken, since they can correspond to bond orders of 21.5 when two electrons are removed from valence bonding orbitals and one electron fills an anti-bonding orbital. The photodesorption of ions is strongly influenced by quenching processes, i.e., by delocalization of the excitation energy and charge. The site selectivity character of ion dissociation reactions is directly related with the electronic and nuclear motion time scales (initial electronic excitation |10 216 s; Auger decay |10 215 –10 214 s; two hole decay |10 214 s; desorption |10 214 –10 213 s). The primary excitation, its decay, and the bond breaking events should not be considered as completely separated in time as in a Frank–Condon model. The requirement of long localization time of two holes for desorption to occur was recently questioned by Menzel [16]. The existence of steep repulsive potentials for dissociative excited states can lead to the ultrafast dissociation which can occur even before the core hole decay. Some evidence for the ultrafast dissociation was given for H 1 and D 1 desorption from surface adsorbed H 2 O [17–20], NH 3 [21–23] and benzene [23]. In the case of benzene the strongest selective dissociation channel was coupled to a decay autoionization spectrum which was very different from all the others and was not explainable by decay of an intact molecular benzene. This fact was taken as evidence that dramatic structural
changes have happened in the molecule before core decay [23]. The time scale competition between the core hole decay and dissociation is still involved in much controversy. Any analysis of site selective dissociation should be done with great care. Presently, intense research is being devoted to the dynamic effects of core-excitations concerning the core hole localization and desorption time scales.
3. Site-specific reactions Surface studies of PSD processes led to the concept of localized valence-hole states. The essential idea is that Auger decay of a core hole produces two valence holes associated with a single atom because the core hole was localized on that atom. This localization of two holes may lead to atomic site-specific reactions. It is known that the quenching of the slow ion desorption processes is more effective on surfaces than on isolated molecules. This quenching would lead to a selective enhancement of fast processes and consequently facilitate the observation of site-specific reactions on surfaces. The studies of PSD of molecules condensed (or adsorbed) on metallic surfaces have revealed the importance of ESD process caused by photoelectrons and secondary electrons created in the decay process after the X-ray radiation (XESD, X-ray-induced ESD). Depending on the character of the substance and experimental conditions, the investigation of site-specific direct photodesorption from surfaces is limited by the dominance of the indirect ion desorption. The photon-to-electron conversion step underlying the XESD process may transfer the excitation to an adsorbate atom with no relation to the originally photoionized. In fact, in some cases, the ion signal does not yield surface-specific information at all but is simply a replica of the total electron yield (TEY) spectrum originating from the adsorbate bulk (or substrate bulk). This effect is most important for covalently bonded surface systems and at high photon excitation energies since the production of secondary electrons increases with excitation energy. This mixing of PSD and XESD prevents a clear analysis of the surface due to the background produced by the bulk. In other words, it is difficult to
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
discern any structural changes of true PSD from the XESD background. The contribution of XESD has been described by Jaeger et al. [24] for the NH 3 / Ni system. From the comparison of the total electron yield and H 1 yield of multilayers of NH 3 with variable thickness on Ni(110), it was found that the ion desorption is dominated by secondary-electron-induced valence excitations in the outermost layer. The direct core hole excitation of a surface molecule can be observed by a fine structural feature at 400.5 eV which is only present in the H 1 yield and not in the TEY spectrum in the N K-edge. Using this feature the contribution of direct PSD and XESD could be estimated to be 40 and 60%, respectively. The investigation of PSD from condensed and adsorbed water on Ru(001) also showed the importance of the XESD contribution for H 1 desorption [17]. The Auger electron and H 1 yields were measured from multilayer, bilayer and monolayer ice samples in the near oxygen K-edge region. At 534 eV, the H 1 signal showed a sharp peak with no corresponding feature in the Auger absorption spectrum. At about 535–542 eV, the H 1 signal showed a broad peak with good correspondence in the Auger spectrum. These two features were, respectively, named threshold peak and main peak and showed different saturation and polarization behaviors. These differences can be understood if we assume that the threshold peak is a true photodesorption feature caused by photoabsorption of the topmost layer, while the main peak contains a large contribution from the XESD and is dominated by bulk adsorption. Since these bulk electrons have insufficient energy to cause core excitations of the topmost layer, multiple valence excitation must be the bond breaking process. It was concluded that in mono- and bi-layer range the observed structures are true primary PSD events. On the other hand, in the multilayer range, the observed threshold peak is a true PSD feature while approximately half of the main peak is due to XESD caused by the numerous electrons liberated at condensate excitations. In all previous examples, direct PSD features can still be observed despite the background contribution of XESD. For these direct PSD features, it was observed that the transitions of the condensate are very similar to those of the free molecule. The larger
259
the alteration in symmetry and in bond strength (due to close interactions with neighboring molecules), the larger the shifts in energy and broadening of the peaks. The selective character of dissociation is pronounced and reflects the large dissociative character of the excited state formed by core excitation. This character is enhanced by the atomic motion that occurs to a small extent before or during core hole decay.
4. Experimental
4.1. Ion yield measurements The time-of flight (TOF) mass spectrometric technique has been used for the PSID investigations with great success. The TOF mass spectrometer combined with the pulsed synchrotron radiation used in this work is similar to that described previously by Knotek et al. [25], and is described in detail elsewhere [26]. Briefly, it consists of an accelerating plate, a drift tube of 9 cm, and a microchannel plate detector. Due to its small size it can be installed very close to the samples. In our experimental settings, it was placed 3 cm away from the sample surface in its normal direction. During single bunch operation of the Photon Factory storage ring, soft X-ray pulses with a period of 624 ns and width of 100 ps are generated. The sample is irradiated by this pulsed soft X-ray beam through a photon-flux monitor of the Au grid. The incident flux spectrum is recorded simultaneously as the photocurrent at the Au grid. All spectra are normalized by this flux spectrum to correct for fluctuations in beam intensity. As a result of irradiation of pulsed soft X-ray beam, ionic fragments are desorbed from the sample. These ions are subjected to an accelerating potential, mass analyzed in the drift tube, and collected by a microchannel plate detector. In the TOF normal operation, TOF the ions are measured using a timeto-amplitude converter (TAC-EG&G Ortec 567). The flight time is measured between the SR pulse and the amplified ion signal. A 1 / 312 divider is used to take the SR pulse timing the 500-MHz frequency signal from the microwave cavity of the storage ring. The total spectra are read and accumulated by an
260
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
analog-to-digital converter and a multichannel analyzer (Laboratory Equipments MCA / PC98A). Due to the very good time reproducibility of the SR pulses, heavy ions which have flight times longer than 624 ns can also be measured. For these heavier ions the time-of-flight is given by its position in the spectrum plus an integer multiple of the SR pulse interval (624 ns) corresponding to the number of cycles that have passed from the production of the ion to its detection. The mass assignments can be easily done by comparative spectra at different accelerating potential. Partial ion yield spectra (PIY) are measured by windowing the channels corresponding to the ion to be analyzed while scanning the photon energy. Better counting rates can be obtained using the TOF in reverse mode, i.e., measuring the time-of-flight of the ions starting at the MCP ion signal and stopping at the SR pulse. Both total electron yield absorption spectra (TEY) and total ion yield spectra (TIY) can be measured using the TOF mass spectrometer using positive or negative accelerating potentials, respectively.
power of CMA is typically 1% and determined by changing the size of the CMA slit. The axis of TOF-MS as well as CMA is located normal to the surface and the angle between the surface normal and SR was 558. The sample surface is excited by SR, and the energy of the emitted electrons is analyzed with the CMA, while the desorbed ions are accelerated towards the TOF-MS. The ion signal in coincidence with the detected electron gives a coincidence signal at a specific TOF, while signals irrelevant to the electron increase the background level. Since the energy-selected electron corresponds to a particular Auger transition, the coincidence signal intensity offers the yield of the ion desorption induced only by the Auger process. The EICO spectrometer was attached to an UHV chamber at the soft X-ray beamline BL13 of Hiroshima Synchrotron Radiation Center (HiSOR) using a dragon type spherical grating monochromator. The CMA and TOF-MS were also used for partial electron yield spectroscopy and total ion yield spectroscopy, respectively.
4.2. Electron–ion coincidence measurements
5. Results and discussion
Details of the PSID mechanism have mainly been discussed on the basis of the partial ion yield (PIY), i.e., the incident photon energy dependence on the yield of particular ion. In these studies the electronic transitions that correspond to thresholds or characteristic peaks in the yield curve were inferred to induce the ion desorption. This approach, however, offers no information about the intermediate Auger transitions. The coincidence measurement of ions with Auger electrons is a powerful approach for the PSID study, because the Auger final state is the direct precursor of ion desorption in the ASID mechanism [13–15]. Mase et al. [27,28] developed an electron–ion coincidence (EICO) apparatus and applied to the PSID studies [18,19,21,22,27–30]. The apparatus used in this work is recently remodeled one to improve the collection efficiency of electrons and ions [31,32]. It consists of a cylindrical mirror electron energy analyzer (CMA), a coaxial time-offlight ion mass spectrometer (TOF-MS), and an electronic system for measurements. The resolving
5.1. Ion yield spectra of PMMA The first study of the PSID from site-specific point of view was performed for PMMA (polymethylmethacrylate) thin films [7]. PMMA is a very suitable sample to use for elucidating the mechanism of site-specific PSID because its monomer contains five carbon atoms and two oxygen atoms, all of which have different chemical environments. Fig. 3 shows structure of the monomer unit of PMMA. Carbon and oxygen atoms at different sites are numbered according to decreasing binding energies. The relationship between the ion desorption and the primary excitation has been investigated in detail [7]. Several fragment ions such as H 1 , CH 1 , CH 1 2 , 1 CH 1 and COOCH 1 3 , CHO 3 were observed in the carbon and oxygen core excitation regions. Fig. 4 shows the partial ion yield spectra of typical ions together with a total electron yield spectrum of PMMA in the carbon and oxygen core excitation region. Several NEXAFS structures are observed in these spectra. The energies and proposed assign-
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
261
Table 1 Photon energies and possible assignments for the peaks observed in the carbon core excitation of a PMMA thin film Peak
A B B9 C D
E F
Energy
Assignment
(eV)
C=O
–O–CH 3
–C–CH 2 –
287.5 288.7 289.4 290.2 291.1 292.3 292.7 294.9 296.8 304.0
– p*(C=O) – – – s*(C–H) – IP s*(C–OCH 3 ) s*(C=O)
– – s*(O–CH 3 ) s*(C–H) – – IP – – –
s*(C–C) – – s*(C–C) IP s*(C–C) – – s*(C=O) –
Fig. 3. Structure of the monomer unit of PMMA.
ments of these structures are listed in Tables 1 and 2. The partial ion yield seems to produce a variety of linewidths for the most intense resonance compare to electron yield. This is because the ion yield spectrum reflects the electronic structure and reaction efficiency of the outermost surface of PMMA films.
On the other hand, the electron yield spectrum reflects the electronic structure of not only surface but also bulk because of its longer escape depth compare to ions. A difference is observed between ions with respect to the NEXAFS structure, and the ion desorption has a strong dependence on the initial
Fig. 4. Typical partial ion yield spectra of PMMA in the carbon and oxygen core excitation regions together with a total electron yield spectra.
262
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
Table 2 Photon energies and possible assignments for the peaks observed in the oxygen core excitation of a PMMA thin film Peak
Energy
(eV) A B C D E
531.5 534.3 535.6 537.8 539.3 539.4 546.8
Assignment C=O
–O–CH 3
p*(C=O) s*(C–OCH 3 ) – IP – – s*(C=O)
– p*(C=O) s*(O–CH 3 ) – s*(C–OCH 3 ) IP –
excitation. In particular, in the carbon core region (Fig. 4a), CH 1 and CH 21 ions are observed selectively at 289.4 eV which is slightly higher than the strongest peak at 288.7 eV assigned as a resonant transition of C1s(C=O) to p*(C=O) orbital. The yield spectrum of CH 1 3 ion, dominant desorbed ion, also has a shoulder at 289.4 eV. As shown in Table 1, the resonant transition at 289.4 eV has recently been assigned as a C1s(–OCH 3 )→s*(O–CH 3 ) transition [32]. In the oxygen core region (Fig. 4b), the 1 efficient production of CH 1 2 and CH 3 ions is observed for the transition of O1s (OCH 3 ) to s*(O– CH 3 ) orbital at 535.6 eV and the efficient production of CHO 1 ions is observed for the O1s(OCH 3 )→s*(C–OCH 3 ) transition at 539.3 eV. From these findings, it can be considered that the resonant core excitation plays an important role to determine the position of bond scission as well as to weaken the bond strength due to the anti-bonding character of the resonant excited state. The above result indicates that the localized character of the initial core excited configuration can be maintained after a spectator Auger decay. Furthermore, this effect can be observed by the changes in the ion yield spectra. The localized character of the antibonding excited state after the site-specific excitation, and the expulsion of the bonding electron near the core–hole atom by Auger decay may possibly induce the site-specific scission of the chemical bond combining with this atom.
directly related to the ion desorption by core excitation. The measurement of Auger electron–photoion coincidence is useful for clarifying the relationship between the Auger final state and the ion desorption. Because of the above reasons, we investigated the PSID of PMMA thin films, using the Auger electron–photoion coincidence (AEPICO) spectroscopy to clarify the Auger process that follows the resonant core excitation and the ion desorption mechanism related to the Auger process [32,33]. Fig. 5 shows a series of typical AEPICO spectra measured with the photon energy set for the s*(O–CH 3 ) resonant excitation (289.4 eV) at which CH 1 , CH 21 and CH 31 ions are selectively observed in their yield spectra (Fig. 4a). The Auger electron energy was sampled in 2.5 eV steps. Two coincidence peaks are clearly observed at about 420 and 1400–1700 ns with a good S /N ratio in the TOF difference spectra. These peaks are assigned as H 1 and CH n1 (n51–3) ions, respectively. The change in appearance of CH 1 n (n5
5.2. Auger electron–photoion coincidence (AEPICO) spectra of PMMA As explained by the ASID model [13–15], the final state of Auger decay (Auger final state) is
Fig. 5. AEPICO spectra of PMMA at the s*(O–CH 3 ) resonant excitation (289.4 eV).
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
1–3) AEPICO signal between Auger electron energies of 265 and 270 eV is of particular interest. In contrast to the broad peak of CH 1 n (n51–3) ions which are observed around 1450–1720 ns at 265 eV, only a single sharp peak is observed at 270 eV. The sharp peak is assigned to the CH 1 3 ion which is predominantly desorbed. The broad peak could be 1 deconvoluted into CH 1 , CH 1 2 and CH 3 components by Gaussian curve fitting [32]. The AEPICO yield for H 1 and CH 1 n ions are obtained by integrating the AEPICO signal intensity of each ion peak above the background level and are plotted in Fig. 6 together with the resonant Auger spectrum (AES) as a function of Auger electron energy. The CH 1 n AEPICO yield spectrum is remarkably enhanced in the higher Auger electron energy region
Fig. 6. AEPICO yield spectra of PMMA at the s*(O–CH 3 ) resonant excitation together with the resonant Auger spectrum as a function of Auger electron energy.
263
of AES (260–275 eV), while the H 1 AEPICO yield shows little dependence on the Auger electron energy. The enhancement of the CH 1 n AEPICO yield in the high Auger electron region may suggest that the spectator Auger transition following the C 2 1s→s*(O–CH 3 ) excitation occurs and produces two holes in the outer valence orbitals which strongly overlap with C 2 1s core hole and have a bonding character with respect to the O–CH 3 bonding. At the nearby resonant excitations, such as p*(C=O) at 288.7 eV and s*(C–H) at 290.2 eV, the CH 1 n AEPICO yield spectrum shows similar dependence on the Auger electron energy to that observed at the s*(O–CH 3 ) excitation at 289.4 eV, but rate of AEPICO yield is relatively low. The results can be explained on the basis of overlapping of these resonant transitions with the s*(O–CH 3 ) transition at 289.4 eV. Thus, the AEPICO yield spectra at the p*(C=O) and s*(C–H) excitation contain contributions of the s*(O–CH 3 ) excitation. It should be noted that the AEPICO yield spectra observed at 320 eV excitation in the ionization continuum, where a normal Auger decay occurs, are obviously different from those observed at above resonant excitations and show relatively low AEPICO yields and less structures. The most characteristic site-specific ion desorption in the carbon core excitation region is the desorption of CH 1 n (n51–3) by the resonant excitation of C1s in the side chain methyl group to the anti-bonding s*(O–CH 3 ) orbital. As mentioned before, the resonant core excitation weakens the O–CH 3 bonding and the subsequent spectator Auger decay produces two holes in the outer valence orbitals that may have a bonding character with respect to the bonding. In other words, two-hole one-electron spectator Auger final states (2h–1e) in which one electron is in an anti-bonding O–CH 3 orbital and the two holes in the valence bonding orbitals may effectively promote the selective O–CH 3 bond scission. In order to elucidate the mechanism of site-specific ion desorption in detail, we calculated the Auger transition probability and compared it with the results of AEPICO experiments [34]. PMMA, however, is too complicated molecules to characterize the Auger final state, thus a methylisobutyrate h(CH 3 ) 2 CHCOOCH 3 j molecule was used as a model molecule of PMMA in the calculations. The calculation is based on the single configuration state func-
264
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
tion (CSF) and limited configuration interaction (CI) methods, and Auger transition probabilities are estimated by overlap between core and valence molecular orbitals. In general, the resonant (spectator) Auger spectrum for a specific core excitation is similar to the normal Auger spectrum of the specific atom in the molecule only with a blue shift in energy due to the screening effect of the spectator electron. Therefore the normal Auger spectrum from this calculation can be compared with the s*(O–CH 3 ) resonant Auger spectrum. Fig. 7 shows the calculated Auger spectrum of a model molecule at the methoxy carbon site together with the experimental resonant Auger spectrum and the AEPICO yield spectrum for CH 1 n ions observed at the resonant core excitation of C 1s (OCH 3 )→s*(O–CH 3 ). The calculated Auger spectrum is shifted to align the position of the highest peaks at 256 eV. The calculated Auger spectrum reproduces the feature of the experimental one. In
the higher Auger electron energy region around 265 eV, where the AEPICO yield spectrum for CH 1 n ions is observed remarkably enhanced, calculated and experimental Auger spectra have a broad peak and a shoulder peak, respectively. As mentioned before, the Auger decay in this region is expected to produce two holes in the outer valence orbitals that have a bonding character with respect to the O–CH 3 bonding. To investigate this subject more precisely, the partial Auger transition probability was extracted from the total Auger transition probability. This indicates the probability of production of at least one hole in a certain molecular orbital (MO) by the Auger decay. The model molecule has 28 occupied MOs including seven inner core orbitals and these MOs are numbered from the core to valence orbitals. MOs from 15 to 28 are involved in the first broad peak around 265 eV of the calculated Auger spectrum. By careful comparison of the partial Auger spectra obtained for these MOs with the AEPICO yield spectrum for CH 1 n ions, MOs 16, 17, 19, and 24 remain appreciably important. These four MOs have a bonding character with respect to the O–CH 3 bonding. Fig. 8 shows the partial Auger spectra for these MOs and the sum of these four spectra together with the AEPICO yield spectrum. As can be seen from Fig. 8, the AEPICO yield spectrum for CH 1 n ions is in good agreement with the sum of partial Auger spectra. From this result, it can be concluded that the efficient desorption of CH 1 n ions occurs definitely through the Auger final states having two holes in these bonding orbitals. This finding demonstrates that the resonant core excitation of the atom that is directly connected to the particular bond to the anti-bonding orbital is of vital importance. Such excitation leads to effective elimination of the bonding electron near the atom by the resonant Auger decay process, and leads to selective scission of the chemical bond.
6. Conclusion
Fig. 7. Calculated Auger spectrum of methyl isobutyrate at the methoxy carbon site compared with the experimental resonant Auger spectrum and AEPICO yield spectrum for CH 1 n (n51–3) obtained at the s*(O–CH 3 ) resonant excitation.
As described above, site-specific surface reactions by core excitation offer an alternative and complementary approach to the active control of chemical reactions. The resonant core excitation of either atom at both ends of the particular bonding to the
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
265
the goal, further investigations in various molecular systems at different environments are needed.
Acknowledgements The authors wish to thank the staff of the PF and HiSOR for stable operation of the SR ring. This work was supported by a Grant-in-Aid on Research for the Future ‘Photoscience’ (JSPS-RFTF-98P01202) from Japan Society for the Promotion of Science (JSPS).
References
Fig. 8. Partial Auger spectra for four MOs (16, 17, 19, 24) and sum of these four partial Auger spectra compared with the AEPICO yield spectrum.
unoccupied anti-bonding orbital brings about weakening the bonding power, the effective elimination of bonding electrons, and the selective scission of chemical bond in the vicinity of the primarily excited atom. In this way, the position of the molecular scission can be controlled. This scenario, however, requires the precise and systematic investigation on the following subjects; the evaluation of the nature of the unoccupied anti-bonding orbital, the probability of resonant core transition from each atom at both ends of the chemical bonding to the anti-bonding orbital, and the partial Auger intensity for each bonding orbital and for each core hole at both ends of the chemical bonding. After such examinations, tuning the soft X-ray energy to the most probable resonant transition leads to the sitespecific bond scission. Finally, the ultimate goal of this kind of studies, we named ‘molecular scalpel’, is now not a dream and one step closer to realization. In order to achieve
[1] R.N. Zare, Science 279 (1998) 1875. [2] W. Eberhardt, T.K. Sham, R. Carr, S. Krummacher, M. Strongin, S.L. Weng, D. Wesner, Phys. Rev. Lett. 50 (1983) 1038. [3] D.M. Hanson, Adv. Chem. Phys. 77 (1990) 1. [4] T. Sekiguchi, H. Ikeura, K. Tanaka, K. Obi, N. Ueno, K. Honma, J. Chem. Phys. 102 (1995) 1422. [5] S. Nagaoka, K. Mase, I. Koyano, Trends Chem. Phys. 6 (1997) 1. [6] T. Sekitani, E. Ikenaga, K. Fujii, K. Mase, N. Ueno, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 101–103 (1999) 135. [7] M.C.K. Tinone, K. Tanaka, J. Maruyama, N. Ueno, M. Imamura, N. Matsubayashi, J. Chem. Phys. 100 (1994) 5988. [8] N. Ueno, K. Tanaka, Jpn. J. Appl. Phys. 36 (1997) 7605. [9] M.L. Knotek, P.J. Feibelman, Phys. Rev. Lett. 40 (1978) 964. [10] P.J. Feibelman, M.L. Knotek, Phys. Rev. B 18 (1978) 6531. [11] M.L. Knotek, P.J. Feibelman, Surf. Sci. 90 (1979) 78. [12] P.J. Feibelman, Surf. Sci. 102 (1981) L51. [13] D.E. Ramaker, C.T. White, J.S. Murday, J. Vac. Sci. Technol. 18 (1981) 748. [14] D.E. Ramaker, C.T. White, J.S. Murday, Phys. Lett. A 84 (1982) 211. [15] D.E. Ramaker, in: N.H. Tolk, M.M. Traum, J.C. Tully, T.E. Madey (Eds.), Desorption Induced by Electronic Transitions, DIET-I, Springer Series in Chem. Phys, Vol. 24, Springer, Heidelberg, 1983, p. 70. [16] D. Menzel, in: A.R. Burns, E.B. Stechel, D.R. Jennison (Eds.), Desorption Induced by Electronic Transitions, DIETV, Springer Series in Surf. Sci, Vol. 31, Springer, Berlin, 1993, p. 101. ¨ ¨ [17] D. Coulman, A. Puschmann, U. Hofer, H.-P. Steinruck, W. Wurth, P. Feulner, D. Menzel, J. Chem. Phys. 93 (1990) 58. [18] K. Mase, M. Nagasono, S. Tanaka, T. Urisu, E. Ikenaga, T. Sekitani, K. Tanaka, Surf. Sci. 390 (1997) 97. [19] K. Mase, M. Nagasono, S. Tanaka, T. Urisu, E. Ikenaga, T. Sekitani, K. Tanaka, J. Chem. Phys. 108 (1998) 6550.
266
K. Tanaka et al. / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 255 – 266
[20] R. Scheuerer, P. Feulner, P. Wiethoff, W. Wurth, D. Menzel, J. Electron Spectrosc. Relat. Phenom. 75 (1995) 16. [21] M. Nagasono, K. Mase, S. Tanaka, T. Urisu, Surf. Sci. 390 (1997) 102. [22] M. Nagasono, K. Mase, S. Tanaka, T. Urisu, Jpn. J. Appl. Phys. 38 (Suppl. 1) (1999) 325. [23] D. Menzel, G. Rocker, H.-P. Steinruck, D. Coulman, P.A. Heinmann, W. Huber, P. Zebisch, D.R. Lloyd, J. Chem. Phys. 96 (1992) 1724. [24] R. Jaeger, J. Sohr, T. Kendelewicz, Surf. Sci. 134 (1983) 547. [25] M.L. Knotek, V.O. Jones, V. Rehn, Phys. Rev. Lett. 43 (1979) 300. [26] K. Tanaka, M.C.K. Tinone, H. Ikeura, T. Sekiguchi, T. Sekitani, Rev. Sci. Instrum. 66 (1995) 1474. [27] K. Mase, M. Nagasono, T. Urisu, Y. Murata, Bull. Chem. Soc. Jpn. 69 (1996) 1829. [28] K. Mase, M. Nagasono, S. Tanaka, M. Kamada, T. Urisu, Y. Murata, Rev. Sci. Instrum. 68 (1997) 1703.
[29] T. Sekitani, E. Ikenaga, H. Matsuo, S. Tanaka, K. Mase, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 88 (1996) 731. [30] T. Sekitani, E. Ikenaga, K. Tanaka, K. Mase, M. Nagasono, S. Tanaka, T. Urisu, Surf. Sci. 390 (1997) 107. [31] K. Mase, S. Tanaka, S. Nagaoka, T. Urisu, Surf. Sci. 451 (2000) 143. [32] E. Ikenaga, K. Isari, K. Kudara, Y. Yasui, S.A. Sarder, S. Wada, T. Sekitani, K. Tanaka, K. Mase, S. Tanaka, J. Chem. Phys. 114 (2001) 2751. [33] E. Ikenaga, K. Kudara, K. Kusaba, K. Isari, S.A. Sarder, S. Wada, K. Mase, T. Sekitani, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 114–116 (2001) 585. [34] E.O. Sako, Y. Kanameda, E. Ikenaga, M. Mitani, O. Takahashi, K. Saito, S. Iwata, S. Wada, T. Sekitani, K. Tanaka, J. Electron Spectrosc. Relat. Phenom. 114–116 (2001) 591.