Nuclear Instruments and Methods in Physics Research B 164±165 (2000) 879±885 www.elsevier.nl/locate/nimb Probing inelastic interactions of ions moving in solids by electron spectroscopy R.A. Baragiola b a,* , S.M. Ritzau a, R.C. Monreal b a Laboratory for Atomic and Surface Physics, Engineering Physics, University of Virginia, Charlottesville, VA 22901, USA Universidad Aut onoma de Madrid, Departmento Fõsica Te orica de la Materia Condensada, C-V, Cantoblanco, 28049 Madrid, Spain Abstract We have measured energy distributions of electrons emitted from clean aluminum surfaces by 20 keV/amu H , He and He0 projectiles. Ratios of energy distributions obtained from two types of projectiles carry information on differences in the electron excitation collisions. We discuss these dierences in terms of screening of the projectile by valence electrons, Auger capture and projectile ionization, with the aid of linear response theory. We ®nd the shapes of secondary electron energy spectra are dominated by the energy dependence of the electron escape depth and of the binary ion±electron interaction. The latter must be determined by considering the screened potential of the projectile and its changing charge state as it penetrates the solid. Ó 2000 Elsevier Science B.V. All rights reserved. PACS: 79.20.-m; 34.50.Dy; 34.70.+e; 79.20.Ap Keywords: Electron emission; Electron capture; Electron loss; Charge fractions; Eective charge 1. Introduction The study of collision processes of atomic particles in condensed matter is hindered by the diculty of extracting information about individual events from observations made outside the solid. The usual method of probing inelastic interactions of ions moving in solids is by measuring an integral quantity, the energy loss of projectiles traversing a thin section of material, from which the stopping power is derived. Another view of * Corresponding author. Tel.: +1-804-982-2907; fax: +1-804924-1353. E-mail address: [email protected] (R.A. Baragiola). electronic excitations can be achieved by studying the ejection of electrons from the solid. This approach has the added complexity that other properties such as the surface potential barrier and electron escape depths intervene, but useful information can be derived from the dependence of the electron yield c on the type and energy of the projectile, the type of material, surface properties, and other variables [1±5]. Two energy transfer mechanisms can energize electrons so they can be ejected from the solid. In potential electron emission, the excitation energy is provided by the potential energy of the projectile that is released, for instance, upon the capture of electrons from the solid. A dierent process, dominant at high-projectile velocities, is kinetic electron emission, which 0168-583X/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 5 8 3 X ( 9 9 ) 0 1 0 1 3 - 7 880 R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 results from energy transfers enabled by the momentum of the projectile. The underlying mechanisms for kinetic emission have been the subject of several recent theoretical studies [6±12]. Compared with energy loss measurements, the study of electron emission has the advantage of being capable of providing information on dierential energy transfers above a few electron volts by measuring the energy distribution of emitted electrons. This method has been applied to the study of collisions in the gas phase for decades [13], which bene®t from the simplicity of not having to consider (usually) multiple collision events. In dense gases and solids, especially at high-projectile velocities, one also has to consider the cascade of electron collisions initiated by energetic electrons. These types of studies have not been common; the vast majority of previous experimental and theoretical research have focused on describing total electron yields. Here, we explore the use of energy distributions of electrons emitted from solids by energetic ions to obtain information about electronic excitation mechanisms. Our study involves experiments and a theory that considers multiple charge-exchange collisions by the projectiles and the transport of excited electrons in the solid. We focus this paper on the comparison of electron emission by H and He (the so-called Z1 dependence) and by He and He0 (the charge state dependence). 2. Experiments The experiments were performed in an ultrahigh vacuum chamber attached to a 120 kV ion accelerator (Fig. 1), using procedures described previously [12]. Projectiles were incident at 60° to the sample normal and electrons ejected in the direction perpendicular to the surface were energy analyzed with a hemispherical sector electrostatic spectrometer, operating at a resolution of 0.2 eV. Clean, polycrystalline surfaces were produced by in situ vapor deposition of 99.999% pure aluminum, and the cleanliness was monitored by Auger electron spectroscopy. For the results reported in this work, we used 20 keV H and 80 keV He and He0 to compare projectiles at the same velocity (20 keV/ Fig. 1. Schematic drawing of the experimental setup. amu or m 0:89 a.u., close to the Fermi velocity of Al). The ion current was measured with a Faraday cup. The energy distributions for R ion impact are normalized so that the total area N E dE equals the known electron yields (c 2:0 for H and 3.85 for He at 60° incidence [14,15]). In the case of neutrals, where the current could not be measured, we normalized the energy distributions to the case of ion impact, so that c0 =c 0:9 0:1, consistent with previous measurements of these ratios in carbon [16] and gold [17] targets. Fig. 2 compares electron energy spectra N(E) obtained with 20 keV/amu He and H projectiles. It is seen that He excites more electrons than H over the whole electron energy range measured. Although He gives much more potential emission than H at very low impact velocities because of its larger potential energy, we do not expect this mechanism to be important at 20 keV/amu. This is because the Auger capture cross -section for 80 keV He is 2:5 10ÿ17 cm2 [18], which implies at 60° incidence, a mean depth for capture of 34 A much larger than typical escape depths of lowenergy electrons. On the other hand, enhanced kinetic excitation by He compared with H is expected since a larger nuclear charge produces larger ion±electron scattering cross sections. The shape of N(E) for kinetic emission is determined mainly by three factors. (1) The energy distribution of electrons from excitation collisions in the bulk decreases with E, a property of the screened Coulomb interaction. (2) The escape depth of electrons, determined by electron±electron and R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 Fig. 2. Electron emission from H and He impact on Al at the same velocity corresponding to 20 keV/amu. Top: energy distributions N(E). Bottom: ratio of energy distributions, R E N E He =N EH . The points are the experimental data. The lines are the results of theory for the following cases: (±±±±) charge of the ion frozen to singly ionized and no elastic scattering of electrons, (- - -) charge of the projectile evolves with depth, no elastic scattering of electrons. (±±+±±+) charge of the projectile evolves with depth, elastic scattering included. elastic electron±atom collisions decreases with E. (3) The probability that an excited electron escapes from the surface increases from zero at E 0 and tends asymptotically to 1 at high E. Also shown in Fig. 2 is the ratio E=NH E. It reveals more clearly the R E NHe dierences between the two projectiles and is more accurate than the individual N(E), because the ratio removes the energy dependence of the eciency of the spectrometer. The ratio R(E) increases with energy. In general terms, this was expected because higher electron energies require collisions closer to the nucleus of the projectile (for unscreened charges, the Rutherford cross-section is proportional to the square of the charge, and the ratio should be 4). 881 Fig. 3. Electron emission from He and He0 impact on Al. Top: experimental energy distributions normalized so that the ratio of the total electron yield is c0 =c 0:9. Bottom: ratio of energy distributions, R E N E0He =N E He . The points are the experimental data. The line is the result of theory, considering the evolution of projectile charge with depth and elastic scattering of electrons. Fig. 3 compares energy distributions of helium ions and neutrals. The ratio R E NHe E= NHe E is seen to fall monotonically with energy. At high E, R(E) is below 1, as expected from the larger screening in He than in He . However, the ratio increases at low E, which is not consistent with a purely screening mechanism in projectileelectron scattering. In fact, at low-electron energies, neutral atoms eject more electrons than ions. 3. Theory As the projectile penetrates the metal, electron transfer processes become operative and the charge state of the projectile evolves in time until equilibrium is reached deep in the solid [18] 882 R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 Fig. 4. Top: attenuation function for electrons of 2 and 20 eV, as a function of depth z. Middle: charge fractions for initial He beams as a function of depth. Bottom: charge fractions for an initial H beam as a function of depth. The graphs illustrate the change in the weighting of dierent charge states with electron energy, due to the energy dependence of the attenuation length. (Fig. 4). There are several mechanisms by which the particle can excite electrons. (1) Binary collisions between the projectile, either in a neutral or charged state, and conduction electrons. (2) Charge transfer processes such as Auger capture and loss and resonant electron loss accompanied by excitation of the metal and/or projectile electrons. (3) Close collisions between the projectile and core electrons of the target can lead, by electron promotion, to the formation of core-holes that will be ®lled with the emission of Auger electrons. In the list of mechanisms we do not include direct Coulomb excitation of plasmons since the projectile velocity in this work is below the threshold for this process [12]. Electrons from these types of excitation events can escape from the metal if they have enough energy to surmount the surface barrier and are excited at depths comparable or smaller than the mean escape depth. These electrons can in turn excite other metal valence electrons or plasmons. These processes give rise to a cascade of secondary electrons and electrons resulting from plasmon decay which tend to pile-up at low-electron energies. Excitation of cascade electrons above the vacuum level is only possible when the energy of the electron from the primary excitation is E > u (energy larger than 2u with respect to the Fermi level), where u is the work function of the surface ( 4:3 eV for Al). In this work, we will be concerned with the excitation of primary electrons by incident He , He0 and H on Al; analysis of the electron collision cascade will be the subject of a future paper. Its contribution is important at the lowest electron energies, where they cannot be disentangled from other processes such as Auger capture and electron loss from the projectile. As we will discuss, we do not ®nd that the eect of cascades is important in our analysis. The possible charge states of He considered are neutral (0), singly ionized (1) and double ionized (2). For H on Al, the negative ion ()1) is possible together with the neutral and the bare proton. The time evolution of each charge state is obtained from a set of coupled rate equations which, for the case of He, reads d/2 C12 /1 t ÿ C21 /2 t; dt d/0 C10 /1 t ÿ C01 /0 t; dt d/1 C21 /2 t C01 /0 t ÿ C10 C12 /1 t: dt 1 Here Cif denotes the rates for electron transfer from the initial charge state i to ®nal charge state f, and /i the fraction of He in charge state i (0, 1, 2). These are standard expressions, except that they do not include, for clarity, the unlikely doubleelectron transfer processes. Eq. (1) is solved with the initial conditions /1 t 0 1, for incident ions and /0 t 0 1 for incident neutrals. Similar equations are solved for H on Al. The solutions of Eq. (1) are of the form /i t /1 i Ai t ; 2 R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 where /1 i is the stationary (equilibrium) solution and Ai (t) is the transient solution (see Fig. 4). The probability per unit time that a particle in a charge state i excites electrons of energy E in an energy interval dE, within linear response theory and the Random Phase Approximation, is easily derived from Section 7 of [18]. It reads Z Z X 4p 2 q q 2 dPi 1 p i 2E dX dx dE 2p2 q2 eL q; x k 0 <kF ÿ d E ÿ Ek0 0 ÿ x d x ÿ qm; 3 with the restriction of momentum conservation k ÿ k 0 q. In Eq. (3), where atomic units are used, k 0 and k are the initial and ®nal electron momentum. The electron excited with energy transfer x is described by its energy E and its angle X. qi (q) is the Fourier transform of the electric charge density given by the projectile nucleus and the electronic charge distribution of the projectile in charge state i, and eL (x, q) is the Lindhard dielectric function. The spatial distribution of excited electrons is, in general, anisotropic but elastic scattering randomizes this distribution eciently, since mean for free paths for elastic collisions are less than 6 A E < 30 eV [1]. Then, for the number of electrons emitted in process 1 above, we write XZ 1 dN dPi 1 ÿz=L E T E e dt/i t ; dE 2 dE 0 i 4 where z mt cos h; m being the projectile velocity and h its incidence angle. In Eq. (4), T(E) is the probability of transmission of an electron over the surface barrier potential, and L(E) is the electron attenuation length. The factor of 1/2 takes care of the fact that, on the average, only one half of the excited electrons travel towards the surface. The attenuation factor exp()z/L(E)) is the probability that an electron of energy E, excited at a depth z, reaches the surface without energy degradation, an average over initial emission directions and paths to the surface. The concept of attenuation is valid in the case of low-energy electrons since the probability that they lose only a small fraction of their energy in an inelastic collision is small. One can see that L(E) aects the shape of not only N(E) 883 but also of R(E), because it determines how much ``memory'' of the initial charge state of the projectile is carried by the emitted electron. Eq. (1) was solved for He and H, using theoretical cross-sections given in [18]. The time scale over which the transient solution Ai (t) of Eq. (2) dies out can be obtained approximately by neglecting /2 for the case of He, and /ÿ1 for the case of H. This is a good approximation for the velocities of the present work since the equilibrium fractions of He2 and Hÿ are both only about 10%. Calling C the sum of the electron capture and loss rates we obtain Ai t Ai 0 eÿCt . Then, charge equilibrium will be reached with a characteristic depth scale Only those electrons excited within D mC 8 A. a depth of the order of L(E) will be detected outside. Therefore, very low-energy electrons, for which L E > D, may originate deep in the solid under conditions close to equilibrium while high-energy electrons, for which L(E) [ D retain some memory of the initial charge of the projectile. This eect is illustrated in Fig. 4, where we plot the charge states for incident H and He , and the attenuation factor exp ÿz=L E) for E 2 and 20 eV. 4. Discussion The ratio R(E) for He /H increases with electron energy. This is expected since larger energy transfers imply closer ion±electron collisions so that the eect of screening is lower. At our velocities, the limit of R 4 for Rutherford scattering, which is the ratio of the square of the nuclear charge, is not achieved since even the head-on collisions are aected by screening. The ratio of the total electron yields (ratio of areas under the N(E) curves) is 1.93 and close to the ratio of electronic stopping powers [19]. Our theory, which considers only electrons excited in binary collisions, gives already the overall shape and magnitude of the experimental results. The importance of attenuation length and the depth evolution of the projectile charge are shown in Fig. 2, where we give the results of calculations performed using the inelastic mean free path for L(E) (dashed line) and an improved description of electron transport by adding elastic scattering as 884 R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 well (crosses). In the calculation where the projectile charge is kept frozen at the incident value, the eect of attenuation lengths is cancelled out when taking the ratios. In this case, R(E) gives directly the ratio of probabilities of producing an electron of energy E by He and by H . However, in the case where the projectile charge is allowed to evolve with depth, the theoretical R(E) changes its behavior at low E while retaining its shape in the highenergy region. This is because, as stated above, high-energy electrons are emitted near the surface and therefore retain memory of the charge of the projectile, while low-energy electrons are emitted from depths where charge equilibrium is attained. These low-energy electrons are therefore more sensitive to the values of the attenuation lengths than high-energy electrons, as can be seen by comparing the results with both choices of L(E). Improvements in theory may be attempted using a non-linear approach, since recent results for the stopping power of protons on Al [20] using a Density Functional approach show that the nonlinear stopping power is two times larger than the linear stopping power at m 0:9 a:u. For balance, this may also require treating electron transfer processes and electron attenuation using a nonlinear theory. One can notice in Fig. 2 that the ratios grow fast with E J 20 eV, unlike the theoretical results. The reason is that, at these energies, the electron spectra contain Al-2p Auger electrons degraded in energy, which are excited more eciently by He than by H projectiles at 20 keV/amu [21,22]. Al-2p Auger electrons and other high-energy electrons excited in close collisions can have a high probability of producing secondary electrons of lower energies, which we have not included in the calculation. However, those cascade electrons will be produced with high ratios R (those relevant to the high energy of the electrons that can initiate the cascade), whereas low values of R are found at the low-energies characteristic of cascade electrons. This shows that the cascade produced by the Al-2p Auger and other high-energy electrons is relatively unimportant in R(E). On the other hand, fast electrons can eciently excite plasmons, whose decay contributes to the electron spectrum below 11 eV [12]. Since R is large for fast elec- trons, the plasmon contribution will also have a high R value compared with electrons at the same energy originating directly from ion±electron scattering. At low-electron energies, the calculated ratios are higher than experiment. A possible reason is the contribution of electrons excited in the charge exchange processes (2) above. These electrons are not accounted for in this work; they are fewer than the electrons excited in binary collisions, as found in [7]. Auger capture, which will be more important for H than for He projectiles, produces electrons in the low-energy region. Another contribution to the low-energy part of N(E) is electron loss (projectile ionization) from the neutralized projectiles which is more likely for H than for He . In gasphase collisions, most projectile ionization produces electrons with near-zero energy in the projectile frame. In the laboratory frame, they appear as a broad peak centered at the velocity of the projectile. In the case of ions moving inside solids we need to take into account that the ®nal state of the electron needs to be above the Fermi level. An electron cannot move inside the solid with the velocity of the projectiles, because it is below mFermi , in the occupied part of the band. We thus expect that most of the electrons from projectile ionization will remain inside the solid near the Fermi level, with a tail extending above the vacuum level populating low E. These electrons are possibly the ones that give the enhancement of the ratio R above the theoretical values for binary ion±electron collisions. Suppression of electron loss by the unavailability of ®nal electron states explains the drop of the electron loss cross sections in solids compared with gases at low velocities reported in [18]. In comparison between He0 and He projectiles, we note that binary collisions calculated by theory can account for the reduced N(E) at intermediate and high-electron energies (Fig. 3). However, R(E) is again above theory for low-electron energies, which could also be explained by electron loss. 5. Conclusions Measurements of energy distributions of electrons emitted in ion±solid interactions carry in- R.A. Baragiola et al. / Nucl. Instr. and Meth. in Phys. Res. B 164±165 (2000) 879±885 formation about dierent electronic excitation processes. We have identi®ed several factors that are important in understanding energy distributions. Ratios of energy distributions for dierent projectiles at the same velocity give information of the screening of the projectile nuclear potential by target electrons. The screened potential determines the strength of the binary ion±electron interaction, which is the main mechanism of electron excitation for light ions at our impact velocity. Another important factor is the evolution of the charge state distribution and screening with depth in the solid, due to charge-changing collisions. Here, the important parameter is the characteristic distance for charge equilibrium D compared with the electron escape depth L. At high-electron energies (e.g., 30 eV), L is smaller than D, so the energy distribution is strongly aected by the initial charge state. For electron energies of a few eV, L is of the order or larger than D and therefore we expected that incident neutrals and ions would give similar energy distributions. However, we see a strong enhancement in the emission of low-energy electrons for He0 compared to He that we hypothesize is due to projectile ionization. A similar argument is applied to explain the enhancement in the electron yield for H with respect to He at low E. The comparison of distributions of high-energy electrons needs to consider the emission of Auger electrons from the decay of core holes excited in close collisions. Finally, we point to research avenues that can be taken to improve theory to achieve a better understanding of these phenomena. These include the consideration of non-linear eects in ion±electron interactions and the inclusion of the electronic cascade in the solid. Acknowledgements This work was partially funded by the Southwest Research Institute, the National Science 885 Foundation, and the Spanish Comisi on Interministerial de Ciencia y Tecnologõa (contract PB970044). References [1] M. R osler, W. Brauer, J. De Vooght, J.-C. Dehaes, A. Dubus, M. Cailler, J.-P. Ganachaud, Particle Induced Electron Emission I, Springer, Berlin, 1991. [2] D. Hasselkamp, H. Rothard, K.-O. Groeneveld, J. Kemmler, P. Varga, H. Winter, Particle Induced Electron Emission II, Springer, Berlin, 1992. [3] R.A. Baragiola, Ionization of Solids by Heavy Particles, Plenum Press, New York, 1993. [4] R.A. Baragiola, Nucl. Instr. and Meth. B 78 (1993) 223. [5] R.A. Baragiola, in: J.W. Rabalais (Ed.), Low Energy Ion± Surface Interactions, Wiley, New York, 1994, Chapter 4. [6] M. R osler, Nucl. Instr. and Meth. B 115 (1996) 278. [7] M. R osler, F.J. Garcõa de Abajo, Phys. Rev. B 54 (1996) 17158. [8] A. Dubus, M. R osler, Nucl. Instr. and Meth. B 235 (1997) 18. [9] J.I. Juaristi, M. R osler, F.J. Garcõa de Abajo, Phys. Rev. B 58 (1998) 15838. [10] S. Bartholome, A. Dubus, J. Devooght, J.P. Ganachaud, C. Attard, Nucl. Instr. and Meth. B 125 (1997) 13. [11] J. Kawata, K. Ohya, Nucl. Instr. and Meth. B 153 (1999) 42. [12] S.M. Ritzau, R.A. Baragiola, R.C. Monreal, Phys. Rev. B 59 (1999) 15506. [13] N. Stolterfoht, R.D. DuBois, R.D. Rivarola, Electron Emission in Heavy Ion±Atom Collisions, Springer, Berlin, 1997. [14] R.A. Baragiola, E.V. Alonso, A. Oliva-Florio, Phys. Rev. B 19 (1979) 121. [15] B. Svensson, G. Holmen, J. Appl. Phys. 52 (1981) 6928. [16] S.M. Ritzau, R. Baragiola, Phys. Rev. B 58 (1998) 2529. [17] G. Lakits, A. Arnau, H. Winter, Phys. Rev. B 42 (1990) 15. [18] P.M. Echenique, F. Flores, R.H. Ritchie, in: H. Ehrenreich, D. Turnbull (Eds.), Solid State Physics: Advances in Research and Applications, Academic Press, New York, 1990, Vol. 43, p. 229. [19] S. Kreussler, C. Varelas, R. Sizmann, Phys. Rev. B 26 (1982) 6099. [20] A. Salin, A. Arnau, P.M. Echenique, E. Zaremba, Phys. Rev. B 59 (1999) 2537. [21] C. Benazeth, N. Benazeth, L. Viel, Surf. Sci. 65 (1977) 165. [22] C. Benazeth, M. Hou, N. Benazeth, C. Mayoral, Nucl. Instr. and Meth. B 30 (1988) 514.
© Copyright 2026 Paperzz