Fizika.hfd.hr

PHOTON-STIMULATED DESORPTION OF HYDROGEN IONS FROM SEMICONDUCTOR SURFACES: EVIDENCE FOR DIRECT AND INDIRECT Ca,1, A. HOFFMANb, G. COMTETc, L. HELLNERd and G.
aDepartment of Electronic Materials Engineering, Research School of Physical Sciences and Engineering, The Australian National University, Canberra ACT 0200, Australia bChemistry Department and The Solid State Institute, Technion, Haifa 32000, Israel cLaboratoire pour l’Utilisation du Rayonnement Electromagn´etique (LURE), e de Paris-Sud, 91405 Orsay Cedex, France dLaboratoire de Photophysique Mol´eculaire, CNRS, Bˆatiment 213, e de Paris-Sud, 91405 Orsay Cedex, France Received 22 November 1999; revised manuscript received 21 February 2000 Photon-stimulated desorption of positive hydrogen ions from hydrogenated dia-mond and GaAs surfaces have been studied for incident photon energies aroundcore-level binding energies of substrate atoms. In the case of diamond surfaces,the comparison between the H+ yield and the near edge X-ray absorption finestructure (NEXAFS) for electrons of selected kinetic energies reveals two differentprocesses leading to photodesorption: an indirect process involving secondary elec-trons from the bulk and a direct process involving core-level excitations of surfacecarbon atoms bonded to hydrogen. The comparison of H+ photodesorption andelectron photoemission as the function of photon energy from polar and non-polarGaAs surfaces provides clear evidence for direct desorption processes initiated byionisation of corresponding core levels of bonding atoms.
Keywords: desorption of positive hydrogen ions, photon-stimulated, hydrogenated dia-mond surface, GaAs surface, photon energies around core-level, direct and indirect pro-cesses c et al.: photon-stimulated desorption of hydrogen ions from . . .
The bombardment of a sample surface by low-energy photons can induce des- orption of neutral and ionic species from the surface, particularly if the surfacecontains an adsorbed layer [1]. This effect, known as photon-stimulated desorption(PSD), is a consequence of primary electronic excitations whose relaxation causesbond breaking and desorption. Several different models have been proposed for ex-planation of desorption from ionic and covalent surfaces. Although these differ indetails, all models consider valence or core electronic excitations, followed by rapidelectronic rearrangement to repulsive states, as the driving force for desorption [1].
It now appears that, in many cases, the valence mechanism plays only a minorrole in positive-ion desorption, as a growing number of experiments show dominantonsets in desorption yield of positive ions at energies related to the excitation ofcore levels [2,3].
Desorption induced by core-level excitations may involve direct and indirect processes. A direct process is initiated by the Auger decay of a core hole, created byphoton bombardment, which may lead to localised repulsive states in the bondingorbitals [4]. On the other hand, an indirect desorption process is initiated by thesecondary electrons (released during an Auger process) that induce the valenceband excitations followed by the desorption of positive ions [5]. The contributionof indirect processes is believed to be small, but may dominate the total yield inthose cases when the other desorption mechanisms are suppressed [6].
The mechanism of stimulated desorption and the contribution of direct and indirect processes may be studied in some detail by comparing the positive ion yield,as the function of photon energy, and the near edge X-ray adsorption fine structure(NEXAFS) recorded by measuring the partial electron yield (PEY) for selectedelectron-kinetic energies. We employed this method to study desorption of positivehydrogen ions from different hydrogenated semiconductor surfaces. For some ofthem, however, the NEXAFS measurements may be hindered around a particularcore-level by overlapping with the valence band photoemission. In order to overcomethis problem in the case of GaAs, we have carried out the PSD measurements ondifferently terminated GaAs surfaces.
This research has also been driven by the great technological importance of hy- drogenated semiconductor surfaces: understanding the chemistry and bonding ofhydrogen on semiconductor surfaces is crucial for both wafer cleaning and polish-ing and for epitaxial growth where hydrogen is thought to inhibit epitaxy [7]. Wealso point out here that the characterisation of hydrogen on semiconductor surfacesrepresents a very difficult experimental problem. Commonly used surface sensitivetechniques, such as Auger electron spectroscopy (AES) and X-ray photoelectronspectroscopy (XPS), are not sensitive to hydrogen. Some other techniques, such asinfrared spectroscopy, that are sensitive to hydrogen bonding, lack surface sensitiv-ity. On the other hand, it is well known that the ion photodesorption represents avery sensitive surface probe for the investigation of the local electron structure andbonding of adsorbed species [1]. From the analytical point of view, it is importantto explore the desorption mechanism of positive ions under photon bombardmentand the sensitivity of PSD for hydrogen adsorbed on semiconductor surfaces.
c et al.: photon-stimulated desorption of hydrogen ions from . . .
Polycrystalline 10 µm thick diamond films were deposited on silicon substrates by the hot filament chemical vapour deposition (CVD) method, using a systemdescribed previously [8]. The Raman spectrum measured for these films shows onlythe characteristic diamond line at 1333 cm−1. No additional lines associated withamorphous carbon or graphite were measured in the Raman spectrum. Scanning-electron-microscopy (SEM) examination of the films indicated that they were con-tinuos and composed of crystallites of 2 –3 µm in size. Auger analysis of the filmsshows that their surface was free of oxygen and other impurities (< 0.05 at.%).
The surface of the as deposited films was terminated by atomic hydrogen. This wasdetermined ex-situ by temperature programmed desorption (TPD) measurementsof the as deposited films.
The GaAs samples were grown by molecular beam epitaxy (MBE) using semi- insulating GaAs (110) and (100) substrates. Following thermal oxide removal, anundoped GaAs buffer layer approximately 0.25 µm thick was grown on each sub-strate, having either the non-polar (110) surface with equal number of Ga and Asatoms or the polar, As terminated (100) surface. The samples were then allowed tocool under an As2 flux while a liquid nitrogen cooled finger contacted the substratemounting block. Over a period of some hours, an amorphous arsenic layer cappedthe surface, thus protecting it against degradation.
All measurements were performed in an ultra high vacuum (UHV) chamber con- nected to either the synchrotron beam-line SA72 or SA23 of Super-Aco at LURE,which delivers photons in the 150 – 600 eV or 35 – 125 eV range, respectively.
The UHV chamber is equipped with a hemispherical electron analyser (CLAM)for photoemission studies and a high-sensitivity quadrupole mass analyser (RiberMIQ 156) for ion detection. A gas manifold and an activation set-up was added foradsorption measurements.
The arsenic cap on GaAs samples was removed in the UHV chamber by heating each sample to about 400◦ C. This procedure is known to produce a surface ofquality similar to an as-grown MBE surface, with possibly a small amount of Asfrom the cap remaining on the surface [9]. The clean surfaces were exposed to3 × 104 L (langmuir) of molecular hydrogen (1 L = 10−6 torr of H2 × 1 s, 1 torr= 133 Pa). As H2 does not adsorb on the surface, atomic hydrogen was producedby dissociation of H2 at a hot filament placed about 5 cm away from the samplesurface. The hydrogen exposure used in this study was below the level causing ahighly disordered surface [10].
We start with the photodesorption from diamond surfaces, taken from Ref. [16], to illustrate the standard procedure in the determination of desorption channels.
PSD of positive hydrogen ions and the NEXAFS spectrum from the hydrogenated c et al.: photon-stimulated desorption of hydrogen ions from . . .
diamond surface, measured by photons in the 280 – 340 eV energy range, are shownin Fig. 1. The NEXAFS spectrum was taken for secondary electrons of kineticenergy of 8 eV, that are more bulk sensitive [12], and it basically represents thetotal electron yield (TEY).
Fig. 1. a) Desorption yield of H+ from the diamond film as a function of incidentphoton energy and b) the NEXAFS spectrum taken for 8 eV secondary electronsfrom the same surface. The inset shows the NEXAFS spectra of 8 eV and 35 eVelectrons.
The NEXAFS spectrum in Fig. 1 displays a threshold at 289 eV followed by a very sharp peak at 289.2 eV and broader peaks at higher photon energies. The dipat 302.4 eV is associated with the absolute second band gap of diamond and thesharp peak at 289.2 eV with a core exciton [13].
The H+ desorption yield displays features similar to those of the NEXAFS with an additional sharp structure at 287.5 eV, 2 eV below the threshold of theNEXAFS spectrum (see Fig. 1a). The comparison between the PSD and NEXAFSdata indicates that the H+ yield consists of a signal proportional to the TEY, witha threshold at about 289 eV, superimposed on a resonance at 287.5 eV.
c et al.: photon-stimulated desorption of hydrogen ions from . . .
The PSD of H+ from GaAs surfaces was measured using photons in the 35 – 120 eV energy range, which covers As 3d and Ga 3p core-level binding energies. ThePSD spectrum of H+ from hydrogenated GaAs (100) surface is shown in Fig. 2aas a function of the photon-beam energy. From this polar (As-terminated) surface,we detected two well resolved resonant-like peaks in PSD of H+ at 43 eV and 43.7 GaAs (100)
YIELD (arb.units)
PHOTON ENERGY (eV)
GaAs (100)
h =90.7eV
INTENSITY (arb.units)
BINDING ENERGY (eV)
GaAs(100)
YIELD (arb.units)
PHOTON ENERGY (eV)
Fig. 2. a) Normalised desorption yield of H+ from a hydrogenated GaAs (100)surface as a function of photon energy around As 3d core-level binding energy. b)As 3d photoemission spectrum from the same surface. Spin-orbit splitting of the3d level is indicated by arrows. c) Desorption yield of H+ from the same surfacearound Ga 3p core-level binding energy.
c et al.: photon-stimulated desorption of hydrogen ions from . . .
eV, corresponding to the As 3d core-level binding energies (indicated by arrowsin Fig. 2a) [11]. No thresholds or peaks were observed from this surface at higherenergies around the binding energy of the Ga 3p level, as shown in Fig. 2c. InFig. 2b, we show the photoelectron spectrum around the As 3d level, obtainedby synchrotron radiation spectroscopy. This spectrum is shown with reference tothe conduction-band minimum (CBM) to be directly compared to the PSD ofH+ (the peaks in these curves correspond to the 3d → CBM transitions). Thephotoemission spectrum in Fig. 2b clearly shows the fine structure caused by thespin-orbit (SO) splitting of corresponding core levels indicated by arrows. There isa striking similarity between the shape of the H+ PSD curve and correspondingphotoelectron spectrum. Even the fine details in the photoelectron spectra (i.e. theSO-splitting of corresponding core levels and a kink at 44.5 eV) are reflected in thePSD of H+.
3.3. Direct and indirect processes in PSD of H+ The results shown in Figs. 1 and 2 reveal two distinct processes in PSD of H+ which can account for the photodesorption of positive ions closely following theionisation of core levels. The first is an indirect desorption process that involvesvalence excitations at the surface induced by secondary photoelectrons. The secondmechanism is a direct process, in which a core-hole formation initiates an Augerdecay followed by desorption of positive ions.
The desorption of H+ from diamond surfaces (Fig. 1), taken from Ref. [16], represents a case showing both direct and indirect desorption channels. The directprocess is characterised by the resonance at 287.5 eV and does not show the featuresof the TEY (see Fig. 1). The weak structure at 287.5 eV has been also observedin NEXAFS taken for 35 eV electrons (see inset to Fig. 1), which are more surfacesensitive [12], and it has been previously assigned to a C(1s) → σ∗ (C-H) resonance[14]. The appearance of the 287.5 eV peak in the surface sensitive NEXAFS modeis, therefore, most probably associated with the presence of C-H species on thediamond surface. Consequently, the resonance at 287.5 eV in the PSD of H+ maybe also associated to the core-level excitations of surface carbon atoms bonded tohydrogen atoms. In Ref. 16, we suggested the following mechanism for this process.
The ionisation of the 1s core-level of carbon atom bonded to hydrogen initiatesan Auger transition leading to the emission of a C(KLL) Auger electron with thekinetic energy of about 270 keV and formation of two localised valence-band holes[4]. These localised two-hole states may cause the bond breaking and the emissionof H+ through the charge separation as the most likely way to relax the largehole-hole repulsive interaction.
On the other hand, the indirect process in desorption of H+ from diamond surfaces is generated by the bulk excitations, i.e. by the large flow of secondaryelectrons which are released in the decay process of C(1s) core holes created byphoton bombardment. The minimum energy required to desorb H+ from the di-amond surface (consisting of the sum of the C-H bond energy and the ionisationenergy of H, reduced by the electron affinity of the hydrogenated surface) is at least c et al.: photon-stimulated desorption of hydrogen ions from . . .
18 eV and some multielectron valence excitations may account for desorption of H+[15,16].
Turning now to the desorption of H+ from GaAs surfaces (Fig. 2), we note once again that, in general, an indirect mechanism can be identified by measuring theenergy dependence of the total absorption coefficient, represented by the TEY (asshown in Fig. 1 for the desorption from hydrogenated diamond films), while thefurther information can be obtained from the shift of PSD threshold caused bythe chemical shift of core levels due to the surface bonding [17]. Neither of thesemethods works on hydrogenated GaAs (100) surface around the As 3d edge. TheTEY measurements are hindered around the As 3d edge by overlapping with thevalence band photoemission. On the other hand, there is no (or very little) changein As core-level binding energies due to the hydrogen chemisorption on the surface[18]. In order to identify the desorption mechanism of H+ from GaAs surfaces, weapplied a new approach based on comparison of desorption yield from differentlyterminated surfaces.
A compound matrix, such as GaAs, exhibits different surface termination for different crystal orientation, and the PSD from different surfaces may provide theclue for the desorption mechanism. Therefore, we carried out the PSD measure-ments on both the polar (As terminated) GaAs (100) surface (Fig. 2) and thenon-polar GaAs (110) surface with equal number of Ga and As atoms (Fig. 3). Incontrast to the GaAs (100) surface, the resonant-like desorption of H+ from theGaAs (110) surface was detected at both the Ga 3p edge (at 105 eV and 108 eV,see Fig. 3a) and the As 3d edge. In Fig. 3b, we show the photoemission spectrumfrom the GaAs (110) surface around the Ga 3p edge which exhibits the character-istic SO splitting [11]. Comparison of the PSD yield of H+ with the photoemissionspectrum confirms, once again, that the structure in the H+ yield at the Ga 3pedge is related to the SO-splitting of that level.
The results shown in Figs. 2 and 3 provide a clear evidence for a direct process in desorption of H+ from GaAs surfaces. Namely, if an indirect process accounts forthe desorption of hydrogen, the PSD yield should exhibit a change at both As andGa edges on both surfaces, as photons of 40 – 120 eV ionise both As and Ga corelevels, thus producing secondary electrons of sufficient energy to induce an electron-stimulated desorption process. For the low hydrogen coverage, however, hydrogendesorbs only from As atoms on a polar, As-terminated GaAs surface, but from bothGa and As atoms on a non-polar GaAs surface with equal numbers of Ga and Asatoms. The absence of the Ga edge in the H+ yield in Fig. 2c clearly indicates thatthe photoelectrons formed on Ga sites are not effective in H+ desorption. Further,the fine structure in PSD of H+ has not been found in our photoabsorption mea-surements around the Ga 3p edge (represented by the TEY curve in Fig. 3a). Thus,it is reasonable to conclude that the indirect mechanism does not play a significantrole in the resonant-like desorption of H+ from hydrogenated GaAs surfaces. Themain contribution comes from a direct desorption process initiated by ionisation ofAs-H and/or Ga-H surface complexes. The core-hole formation in the M-shell of As(3d level) and/or Ga (3p level) initiates an interatomic Auger decay process whichmay produce a localised two-hole final state in the bonding orbital. The lifetime of c et al.: photon-stimulated desorption of hydrogen ions from . . .
that state is sufficiently long (about 10−13 s [5]) and desorption of H+ may occurvia unscreened hole-hole (Coulombic) repulsion.
GaAs (110)
INTENSITY (arb.units) 0.595
PHOTON ENERGY (eV)
GaAs (110)
INTENSITY (arb.units)
BINDING ENERGY (eV)
Fig. 3. a) Normalised desorption yield of H+ from a hydrogenated GaAs (110)surface as a function of photon energy around the Ga 3p core-level binding energy.
Total electron yield (TEY) curve for 40 eV photoelectrons is also shown for the sameenergy range. b) Ga 3p photoelectron spectrum showing the spin-orbit splitting.
We have shown that the PSD of positive hydrogen ions from hydrogenated semi- conductor surfaces may proceed via two distinctive processes: the first one resultsfrom bulk excitations and involves the secondary electrons, while the second oneresults from surface excitations and involves the Auger decay of core holes. Ourresults demonstrate that the mechanism of the desorption process can be revealed c et al.: photon-stimulated desorption of hydrogen ions from . . .
by the comparison of the PSD-threshold measurements with the photoemission andNEXAFS measurements around the core-level binding energies or by the compari-son of photodesorption from differently terminated surfaces.
This work was supported by the Australian Government’s grant under the ”Ac- cess to Large Facilities” sheme and the European Community’s ”Access to Large-Scale Facilities” programme.
1) M. L. Knotek, Rep. Prog. Phys. 47 (1984) 1499;2) R. A. Baragiola and T. E. Madey, in Interaction of Charged Particles with Solids and Syrfaces, Vol. 271 of NATO Advanced Study Institute, Series B: Physics, edited by A.
Gras-Marti, Plenum, New York (1991) p.313; 4) T. E. Madey, D. E. Ramaker and R. Stockbauer, Ann. Rev. Phys. Chem. 35 (1984) 5) R. D. Ramsier and J. T. Yates, Jr., Surf. Sci. Rep. 12 (1991) 243;6) D. E.Ramaker, T. E. Madey, R. L. Kurtz and H. Sambe, Phys. Rev. B 38 (1988) 2099.
7) S. J. Pearton, J. W. Corbett and T. S. Shi, Appl. Phys. A 43 (1987) 153;8) A. Hoffman, A. Fayer, A. Laikhtman and R. Brener, J. Appl. Phys. 7 (1995) 3126;9) R. W. Bernstein, A. Borg, H. Husby, B.-O. Fimland and J. K. Grepstad, Appl. Surf.
10) M. Kubal, D. T. Wang, N. Esser, M. Cardona, J. Zegenhagen and B. O. Fimland, 11) C. D. Wagner, W. M. Riggs, L. E. Davis, J. F. Moulder and C. E. Mullenberg, Handbook of X-Ray Photoelectron Spectroscopy, Perkin-Elmer/Physical Electronic Division, EdenPrairie, MN, (1979); 12) J. F. Mornar, F. J. Himpsel, G. Hollinger, G. Hughes and F. R. McFeely, Phys. Rev.
13) J. F. Mornar, F. J. Himpsel, G. Hollinger, G. Hughes and J. L. Jordan, Phys. Rev.
14) Y. Takata, K. Edamatsu, T. Yokoyama, K. Seki and M. Tohnan, Jpn. J. Appl. Phys.
15) L. Hellner, L. Philippe, G. Dujardin, M-J. Ramage, M. Rose, P. Cirkel and P. Dumas, c, G. Comtet, A. Heurtel, L. Hellner and G. Dujardin, Phys.
17) J. A. Yarmoff, A. Taleb-Ibrahimi, F. R. McFeely and Ph. Avouris, Phys. Rev. Lett. 60 18) P. K. Larsen and J. Pollmann, Solid State Commun. 53 (1985) 277.
c et al.: photon-stimulated desorption of hydrogen ions from . . .
FOTONIMA STIMULIRANA DESORPCIJA VODIKOVIH IONA IZ SINA: DOKAZI IZRAVNIH I POSREDNIH PROCESA cavali smo fotonima stimuliranu desorpciju pozitivnih iona vodika iz hidro- sina dijamanta i GaAs, za fotone energije oko energija vezanja unutarnjih elektrona atoma podloge. U sluˇ nosa H+ i fine strukture blizu-rubne apsorpcije X-zraˇ pciju: posredan proces uz sudjelovanje sekundarnih elektrona iz osnovnog materi-jala, i izravan proces uzrokovan uzbudom unutarnjih elektrona povrˇ ugljika vezanih na vodik. Usporedba fotodesorpcije H+ i emisije elektrona u ovis-nosti o energiji fotona iz polarnih i nepolarnih povrˇ izravne procese desorpcije uzrokovane ionizacijom odgovaraju´

Source: http://fizika.hfd.hr/fizika_a/av99/a8p275.pdf