Appl. Phys. A 35, 51-59 (1984) Applied ,,,,,,, Physics A Surfaces "n" 9 Springer-Verlag 1984 Monte Carlo Calculations of keV Electron and Positron Slowing Down in Solids. II S. Valkealahti and R. M. Nieminen Department of Physics, University of Jyv/iskyl/i, SF-40100 Jyv/iskyl/i 10, Finland Received 17 January 1984/Accepted 17 March 1984 Abstract. An improved Monte-Carlo simulation technique has been used to investigate positron and electron slowing down in solid matter. Elastic scattering is based on exact cross sections of effective crystalline potentials and inelastic processes are described by Gryzinski's semiempirical expression for each core and valence electron excitation. Calculations with normal and oblique angles of incidence have been made for positrons and electrons impinging on semi-infinite aluminium, copper, tungsten, and gold. Interesting differences have been found between positron and electron penetration and backscattering features. PACS: 61.80, 78.70, 29.70 In a recent publication [-1] we have presented results of Monte Carlo (MC) simulations of electron and positron implantation in solid targets in the keV energy range. The results are useful for a variety of purposes, e.g. slow positron beam studies [2], x-ray and Auger microprobes, electron microscopy, etc. The implantation properties obtained in [1] are equal for positrons and electrons. This is based on the assumption that in the relevant energy range both the elastic and inelastic cross sections are very similar for the two particles. While this is a plausible first approximation, it is of interest to explore the electronpositron differences in the slowing down process. Furthermore, the calculations reported in [1] used the screened Rutherford cross section for elastic scattering. It is known that this cross section is not accurate for scattering off atoms at low energies, and it is therefore desirable to use more realistic cross sections based on crystalline atom potentials. In this paper we report results of extended and improved MC simulations of slowing down and backscattering for 1-10keV positrons and electrons incident on solid targets. The differences with and extensions from [1] are as follows: (i)the elastic scattering cross sections are calculated exactly using single-particle crystalline atom potentials for both electrons and positrons, (ii) the small but nonvanishing differences between the inelastic cross sections for electrons and positrons are accounted for, and (iii) the angle of beam incidence on the surface is varied. The simulation sequences are used to extract the statistical distributions of overall and projected implantation ranges, slowing down times, backscattering energy and angular distributions and total backscattering yields, for incident energies between 1 and 10keV. Special attention is paid to mapping out the electron-positron differences. Comparison with experiment, where possible, shows generally good agreement. In particular, the backscattering yields obtained are much closer to experimental estimations than those reported in [-1]. Analytic fits are presented for the stopping profiles and their Laplace transforms. These should prove useful in analyzing, e.g., re-emission data of implanted, thermalized particles. 1. Calculation Procedure The main ideas of the MC simulation have been outlined in [1], and more practical details have been presented in, e.g., [3]. Improving on [1], the elastic scattering processes are here calculated using accurate atomic cross sections and the inelastic processes are 52 S. Valkealahti a n d R. M. Nieminen calculated separately for each energy level using the Gryzinski's excitation function [4]. The effect of lattice structure is approximately included in the elastic collisions via modification of atomic cross sections, but channelling processes are still excluded. 1.l. Elastic Scattering The differential scattering cross section for a central potential can be calculated from the partial wave sum as [5] da I dr2 - k 2 l=o k [(2l+1) eia' sin31Pl(cos0)] 2, CD (1) 0.0 5 0.2 Ener-'g~ where k is the wave number, Pg is the Legendre function, and 6~is the phase shift of the/th partial wave. The total elastic scattering cross section is obtained by integrating over angles as & O-el= j' ~ dO. (2) The Legendre functions can be effectively generated using their recurrence relation [6]. The phase shifts are obtained by numerically solving the radial Schr6dinger equation [ d2 2/~ -~r~r~+~V(r)-~ I(1+1) r2 ] k 2 u,,k=0 (3) and matching the solution to the asymptotic form sin(kr- 17c/2+ (~l)" V(r) in (3) is the effective single-atom scattering potential in condensed matter. We evaluate it as follows. Firstly, free atom potentials and electron densities are calculated selfconsistently by the densityfunctional method [7, 8]. The exchange and correlation between the electrons are included by using a local-density-dependent exchange-correlation potential [8]. Secondly, an approximation for the crystalline charge density and Coulomb potential is built up by superimposing atomic charge densities in a lattice. The spherical average of the charge density and Coulomb potential around an atom are then evaluated. The exchange-correlation potential for electrons is added in the local density approximation. For positrons, only the spherically averaged Coulomb potential (with the reversed sign) is included. The scattering potentials are normalized to zero at the Wigner-Seitz radius. Equation (3) is then solved using the effective central crystalline potential for each partial wave by numerical integration. Inserting the calculated phase shifts in (1, 2) the differential and the total elastic scattering cross section can be obtained. The da(E,O)/dO-data, calculated separately for positrons and electrons, is evaluated for values of 10 [ keY ) Fig. 1. Total elastic scattering cross section for copper crystalatom. - ..... positrons; electrons; screened Rutherford energy and angle from 20eV to 10keV and from 0 to 180 degrees, respectively. In the simulations, the differential cross section for an arbitrary value of incident particle energy is calculated by interpolation in the calculated data set. The scattering angle 0 is determined by selecting a uniform random number 0<RI_-< 1 and finding the value of 0 from the cross section data which satisfies Rl=2rcid~ , o ~ sin 0,dO/O'el (4) The total elastic scattering cross section of positrons and electrons in metallic Cu as a function of energy is shown in Fig. 1. Electrons generally have a larger elastic scattering cross section than positrons. The difference between electrons and positrons decreases as a function of energy and increases as a function of atomic number. Larger elastic scattering cross sections for electrons result from two different phenomena. Firstly, electrons encounter an attractive potential, which makes large angle scattering more probable, whereas positrons sense a repulsive potential. Secondly, the exchange-correlation potential for electrons makes the atomic potential larger in amplitude in the outer region, which increases the small angle scattering. At low energies, the electron cross section may show rapidly varying (resonant) behavior and may actually be smaller than the positron cross section (Figs. 1-3). For comparison, the screened Rutherford cross section (used in El]) is also shown in Fig. 1. As can be seen, the Rutherford formula overestimates the elastic scattering for small energies, more so for heavier elements. Note, however, that the overestimation mainly occurs at small angles and thus the effect on slowing down is not as dramatic. The use Monte Carlo Calculations ofkeV Electron and Positron Slowing Down in Solids. II e§ beam e--beam 53 e§ beam e-- beam da ~ sr -1) ~, do" (/],2sr -1) d _ _ q ~ ~/ (~2sr-I) ,1' dO (/~2sr -1) -aft- Fig. 2. Polar plot of the differential elastic scattering cross section of 100 eV positrons and electrons off a Cu crystal-atom Fig. 3. Polar plot ofthe differentialelasticscatteringcross section of 1keV positrons and electronsoff a Cu crystal-aton:t of the screened Rutherford differential scattering cross section has also been discussed in [9, 10]. Another important difference between electrons and positrons appears in the differential cross sections. Electrons have a greater probability to scatter into large angles than positrons, especially at small energies (Fig. 2). Electrons generally also have a larger probability to scatter into small angles (Fig. 3). This is, again, fairly unimportant for slowing down, because elastic forward scattering does not affect the penetration properties. At low energies large-angle scattering causes the differences between electron and positron backscattering yields while at larger energies the more frequent small angle scatterings lead to smaller differences between the overall electron and positron backscattering probabilities. we have used the approximation 1.2. Inelastic Scatterin9 All contributing core and valence electron excitations have been described in terms of Gryzinski's [4] excitation function. For a target electron shell denoted by i, dai(AE) roe4 EB~( E "]a/z d( E) - ( d E ) 3 E 9 ~ffff~B(1-En]E)+~ln 2 . 7 + \ EB, ] ,(5) where AE, EB,, and E are the energy loss, the mean electron binding energy, and the primary projectile energy, respectively. Equation (5) is basically valid only for ionization processes of bound electrons. Therefore dai(AE) dai(AE) A~=E,, d(AE) A~<=e,,~- d(AE) (6) for processes where the energy loss is smaller than the binding energy of outermost shell electrons. The total inelastic scattering cross section for shell i is O'inel'i= ET~da,(AE') 0 d(AE') d(AU), (7) where the maximum energy loss Emax is for positrons the particle energy E before collision and for electrons (E+E~)/2. The lower cutoff for electrons appears because the most energetic particle is followed after the collision of the indistinguishable particles [11]. This effect increases the total path length of electrons by up to 10% and the penetration depth by less than 5%. Another phase-space effect occurs in inelastic scattering off conduction electron gas, where the Pauli principle restricts the final electron states to be above the Fermi surface. This also weakens the inelastic scattering of electrons as compared to positrons. In the present case, the total inelastic cross section is obtained by summing over shells i, O'inel= ~ O'inel,i 9 i=1 (8) At each inelastic scattering event, the energy loss is calculated by selecting a uniform random number R2, and then finding a value of AE which satisfies ~ax dai(AE,) Re= JA~ d(~E~ d(AE')/ai'e1'~' (9) 54 S. Valkealahti and R. M. Nieminen Table 1. Basic data for the calculations. The electron configuration is the one included in inelastic processes; for the lattice potential calculation the full atomic potential is used Material A1 Cu W Au Atomic no. 13 29 74 79 Density Lattice type Electron configuration [g/cm3] Lattice constant [A] 2.70 8.96 19.3 19.3 4.05 3.61 3.16 4.08 ~c fcc bcc fcc 2s22p63sZ3p63dl~ 1 3sZ3p6...5p65d46s 2 3sZ3p6...5p65dl~ ls22s22p63sZ3pl where the energy loss AE is between 0 < A E < Emax. The scattering angle co after an inelastic collision is obtained from the binary collision model [-4] from (1 O) sin co = (A E/E) 1/2 . 1.3. The Simulation Procedure The mean free path of a penetrating particle is ,~- A N AOa' (11) where A, NA, 0, and a are the atomic mass, the Avogadro constant, the mass density, and the collision cross section, respectively. The inverse of the total mean free path 2r is a sum over the inverses of the different mean flee paths, 1/2T= 1//~el-I- ~ i=1 1/2i, (12) where 2el is the elastic mean free path and 2~ is the inelastic mean free path of a specific core (valence) excitation, where i runs over the contributing electron states. The distance travelled between collisions is obtained from S= -2rlnR3, 2. Results We present the most interesting results of our calculations for aluminum, copper, tungsten, and gold. Comparison with experimental results has been done wherever accurate enough experimental values are available. We concentrate here in the features of penetrating positrons and in the differences between positrons and electrons. Attention has also been paid to the effects of changing the angle of incidence. (13) where R3 is a uniform random number. Another random number R4 is used to determine whether an individual scattering event is elastic or one of the inelastic processes. The event is of type i (i = 1.... , m + 1) provided: O< L - , f~ = 1/2r < R4 < = 1/2r < - 1, Rutherford cross section drastically overestimates the forward elastic scattering cross section for small particle energies. The termination energy of 20eV is used [1]. The slowing down time is of the order of 10-14 s. The computing time for one "experiment" is typically around one hour of CPU time in a UNIVAC 1000/60 computer, depending naturally on the incident particle energy and the material in question. No appreciable difference in computing time is observed between positrons and electrons. The input parameters [-12-14] for the calculations are collected in Table 1. (14) where f o = 0 , fl=l/~-el, f 2 = l / 2 e ~ + l / 2 1 , ..., fm+l = 1 for the contributing scattering processes. 2000 particle histories are used in each of the normal incidence calculations and 1000 particle histories for other incidence angles. One particle history typically contains a few hundred scattering events, which is much less than in the calculations using the Rutherford cross section. The reason for this is that the screened 2.1. Normal Incidence Positron and electron penetration is simulated in semiinfinite aluminum, copper, tungsten, and gold for normal incidence at energies between 1-10 keV. An example of the distribution of the trajectory endpoints is shown in Fig.4. The first difference between positrons and electrons is in the penetration depth (Figs. 5 and 6), with positrons going deeper than electrons. In heavy elements positrons penetrate up to 30% deeper, whereas the difference in light elements is only a few percent. This result agrees with earlier theoretical and experimental results for high energies [11, 15] and with qualitative expectations. The main reason is in the different elastic scattering cross sections of positrons and electrons (Figs. 2 and 3). Electrons have a distinctly larger probability to backscatter from Monte Carlo Calculations of keV Electron and Positron Slowing Down in Solids. II 55 CO ,,.,< AI, e § C9 CD O E = 5 keV, r ~ Au --- 200-=< CO .J X xl-- ++ +++ o<( Do + ++ + + + -r+ o + 0._ Ld I N Z 9 ~< lO0+ + + .~ + -2 -~ T rY FuJ z [lJ .~+ 2 X--axug ( <- 'lOOO ~ Z < ] Fig. 4. The distribution of the trajectory endpoints, projected onto y - z plane, for 5 keV positrons at normal incidence on aluminum. The arrow denotes the entrance position I I I l I I --L I I (,,o) ,.d~ ; / ' ][ ' 0 I 0 5 ENERGY r T ' I '0 I0 (keY) Fig. 6. Mean penetration depth vs. initial beam energy for semiinfinite gold zxMCe+; oMCe-; x [ 1 8 ] e - , thin film data; + [19]e , thin film data i AL c~E 200- --<. <z 21_ o4: o O o 212 l.Ck Ld NO I ~: ~'~z~ ~ Cu, e§ t\ o CD Z O 100- g 02 I-Ld Z Ld [1_ 2 Z <CD Ld 2 O- 0 r- '~ ' ' l * 5 ENERGY ' I I l .0 10 (keY) D PENETRATION DEPTH ( t000 A ] Fig. 5. Mean penetration depth vs. initial beam energy for semiinfinite aluminum, zxMCe+; o M C e - ; x [16]e +, thin film data; + [17] e-, thin film data Fig. 7. Stopping profiles for 3 and 5 keV positrons in semi-intinite copper. The full curves denote fits of (16) to the Monte Carlo simulation data and the dotted lines the thin film data for 3.1 and 5.0 keV positrons from [16] atoms, which arises from the attractive force between electron and atom, and from the lack of exchange effects between positron and core electrons. On the other hand, the indistinguishability of electrons increases the electron penetration depth by up to ~ 5% and thus decreases the difference between electrons and positrons. The mean penetration depth in a semi-infinite target can be described rather well by the formula ~= c~E", where E is the incident energy and c~ and n are parameters, which are for electrons weakly dependent on target material. For electrons rough estimates are c(_~4 gg/cm z and n ~_ 1.5, respectively. For positrons, the dependence on target material is somewhat stronger (Figs. 5 and 6). Calculated mean penetration depths fit rather well with the experimental results. The mean positron penetration depths of Mills and Wilson [16] for A1 and Cu, estimated from thin film data, fit well with the MC values. Experimental mean penetration depths for electrons have been estimated from the film transmission data [17-19] but are unfortunately quite inaccurate. The simulated positron stopping profiles are compared with the experimentally estimated stopping profiles [16] for Cu in Fig. 7. The maxima of the 56 S. Valkealahti and R. M. Nieminen I I I L I I i I , I J ~ , I , r ~ , I 2>- AI m n 0,5• CO < CO o Au b-- X rn < rn X x ++ 0,2- ~- ~0 • orY x + "--0....8++ 9 X • O-- 9 Z .------'-- 0 ~ ~ X 0.4- 0./0-I- • EL Z / o ~ O x / O/ 0.3- Ld Ld I-b< U t.q ../ I A - < zx / U 0.1- Zxl j A j U < CO U < CO 0 I o I I I I I I 5 ENERGY (keY) I i ] IO 0.2- 0.1- 0 ' ~ ' o ~ l I , 5 ENERGY (keY) , ' I IO Fig. 8. Backscattering probability as a function of incident energy from semi-infinite aluminum, zxMCe+; oMCe-; x [20]e-; + [17]e- Fig. 9. Backscattering probability as a function of incident energy from semi-infinite gold. zxMCe+; 9 MCe-; x [2lie- simulated distributions are higher than those of the experimental estimates and the MC values near the surface are lower, respectively. These differences are due to the ignoring in the experimental analysis of the increase of the backscattering probability as a function of film thickness, which is important especially for thin films. If we take the backscattering effect into account, the experimental curves would be much lower near the surface and thereby the maximum values would increase due to normalization. A conclusion can be drawn that the agreement is quite satisfactory. Difficulties in estimating positron and electron implantation profiles from thin film transmission measurements have been discussed in more detail in For light elements the backscattering probability of electrons decreases as a function of energy whereas for heavy elements it increases. The reason is that light elements have a relatively larger scattering probability to large angles (see the lobes in Figs. 2 and 3). These decrease in magnitude with increasing energy. For heavy elements the probability distribution of scattering to different angles remains very similar at these energies (the backscattering lobes are negligible). At energies above 10 keV the relative backscattering probability begins to decrease also for heavy elements. Comparisons with experimental electron backscattering coefficients [17, 20, 21] have been made and fair agreement is found. The difference between the experimental and the MC coefficients is probably due to the energetic secondary electrons. Their contribution to the experimentally observed total backscattering can be of the order of 10% [22]. For positrons no experimental data is so far available. [i]. A most interesting difference between positrons and electrons is in the backscattering probability. Positrons have a much smaller probability to scatter back from semi-infinite targets than electrons (Figs. 8 and 9). Electrons at 1 keV have more than twice as large a probability to scatter back as positrons. The difference is largest for small energies and decreases as a function of energy. The reason lies again in the elastic scattering cross section of crystal atoms for positrons and electrons. Electrons have a relatively larger probability to undergo large-angle scattering events, whereas positrons almost always scatter by less than 30 ~ (Fig. 2). Positrons also have a smaller total elastic scattering cross section than electrons (Fig. 1), which of course decreases the overall positron backscattering probability. 2.2. Oblique Incidence Effects of changing the angle of incidence are studied for A1 and W at incident energies of 1 and 5 keV. Positrons behave similarly to electrons as a function of incidence angle. The effect of changing the angle is a little larger for positrons than for electrons, for example, in the penetration depth and in the backscattering probability (Figs. 11 and 13). Generally the electron-positron differences decrease with Monte Carlo Calculations of keV Elec~:ron and Positron Slowing Down in Solids. II 00 i c~ i (.13 33 i AI, e* E=5 keV, r 57 L _ _ A( ~o J J = ~ (D C) 0 13_ + r~ .g X O + + ++ _/ +++ _~+ + N ~ 0 0 C) d X--OXLS ( "1OOO Fig. 10. The distribution of the trajectory endpoints, projected onto the y - z plane, for 5 keV positrons at 80 ~ angle of incidence on aluminum. The arrow denotes the entrance position AI E=5 keY O IN A~'~-~___ F~ O~ O~ O ~A~A o ~o ~o so ANGLE qb ( o ) Fig. 11. Mean penetration depth as a function of incidence angle for 5 keV positrons and electrons impinging on semi-infinite A1. Ae+~ Oe- i i i AI, e § ~< o ;o AngLe /I,) GD C) O C O.5 90 ( ~ Fig. 13. Dependence of the backscattering probability on angle of incidence for aluminum at 5 keV. A e +, O e . The lines are fits of (15) to the MC points increasing angle. The distribution of the trajectory endpoints for 5 keV positrons at the angle of incidence of 80 ~ impinging on aluminum is shown in Fig. 10. Between 0 ~ (normal incidence) and 30 ~ the change of incidence angle does not have much effect on the penetration properties (Fig. i1). Only with larger angles do the results start to differ from those of normal incidence. The decrease of the mean penetration depth is about 25% for electrons and 35% for positrons when the angle of incidence is changed from 0 ~ to 80 ~ This decrease of the penetration depth naturally moves the implantation profiles nearer to the entrance surface (Fig. 12). The difference in penetration depth between 0 ~ and 80 ~ angle of incidence decreases with energy. The most distinct effect of changing the angle of incidence is on backscattering probability (Fig. 13). Backscattering is almost constant between 0 ~ and 30 ~ but increases rapidly at larger angles. Only about half of the particles absorbed in the material on normal incidence are absorbed when the angle of incidence is 80 ~. The backscattering probability B(q~) for a positron or electron beam with the angle of incidence q~ (with respect to the surface normal) can be presented as [23] .J 'S B(r = b J & 0 _+J o~ 0 . 0 - - - o 4 Penetrotion ~ depth 4 ( '1ooo A) Fig. 12. Fits of(16) to the stopping profiles (normalized to unit area) of 5 keV positrons on aluminum for 0 ~ 40 ~ and 80 ~ incidence angles , (15) where B• is the backscattering probability at normal incidence and b is an experimentally determined constant. Fits of (15) to the MC data give a value of 0.79 _+0.01 for b. The value of b is very slightly (=< 0.01) larger for positrons, and it also increases by 0.02 for both particles when the incident energy is changed from 1 to 5 keV. Similar behaviour of backscattering as a function entrance angle has been found for 58 S. Valkealahti and R. M. Nieminen 10-100 keV electrons [23, 24]. Kalef-Ezra et al. [23] obtained for b the value of 0.89 at the 10-100 keV region. In our calculations the b value increases as a function of energy and by extrapolation agrees with [23]. The reason for the increase of b with energy is that at high energies (above 10 keV) forward scattering becomes relatively more dominating. The backscattering energy distribution of electrons and positrons also changes with the angle of incidence. At normal incidence the maximum is in the region of intermediate energies (0.4Eo < E <0.7Eo, where Eo is the initial energy) and moves to higher values with increasing angle of incidence. This is due to the increase of the backscattering probability for the first few elastic collisions. The mean backscattering energy also increases slightly as a function of incident energy E 0 and target atomic number Z. The reason is that the elastic collision probability increases with respect to the inelastic one as a function of energy and atomic number. The mean value of the backscattering energy is about 10% higher for electrons than for positrons. The reason is again that electrons have a larger probability to scatter into large angles than positrons. Thus the backscattered electrons have more probably suffered a large angle scattering event, whereas positrons have encountered several small angle scatterings and have lost more energy during the longer traversed path in the material. 2.3. Parametrization of the Implantation Profile In [1] we discussed the parametrization of the range distributions and presented an approximation for calculating the re-emission probability of positrons. The formula used for the stopping profile is [25, 26] mzm- 1 P(z) = - - zorn exp [ - (Z/Zo)"] 9 (16) where m and Zo are parameters. The shape parameter m was found to be nearly constant (~1.9), slightly decreasing as a function of atomic number. The penetration parameter zo is a function of incident energy as z0 = ~E", and is proportional to the mean penetration depth i (Zo = 1.1M). These results are confirmed by the present calculations at normal incidence. The value of m for positrons is little larger (by ~0.1) than that of electrons. (The statistical uncertainty of m is typically less than 0.1.) Thus the conclusions and estimates made in [1] remain qualitatively valid. Especially useful is the formula used for the overall re-emission probability of positrons from the entrance surface [27] f=v (1 (26)1/2 xI, (17) where I = O(_dzp(z)exp - ~=~) zj. (18) 0 Above, D is the diffusion constant, 2 the depletion (annihilation) rate in the medium, v the rate of emission from the surface, and P(z) the implantation profile (normalized to unit area). Using (16), one finds for m = 1 t= - 1+ 1 (19) SZo and for m = 2 I = 1 - ~ - SZo e (sz~ erfc(szo/2). (20) In (19) and (20), s = ()~/D) 1/2. Changing the angle of incidence affects the shape of the stopping profiles (Fig. 12). This is also seen as a decrease of m as a function of incidence angle. Positrons have larger values of m than electrons at small angles, but the m value decreases much more rapidly for positrons than for electrons with increasing angle. The decrease of m as a function of incidence angle decreases with increasing energy both for electrons and positrons. For example, when the angle of incidence on A1 changes from 0 ~ to 80 ~ m decreases from 2.0 to 1.4 and from 1.9 to 1.6 for 1 keV positrons and electrons, respectively. The decrease for 5 keV positrons and electrons is from 2.0 tO 1.5 and from 1.9 to 1.7, respectively. The positron re-emission parameter I does not change much below the incidence angle of 30 ~. At larger angles the decrease of the mean penetration depth as a function of angle increases I, if P(z) is normalized to unit area. If, however, the increase of backscattering as a function of angle is taken into account, the reemission probability will decrease as a function of angle. 3. D i s c u s s i o n Electrons typically have more than one lobe in the differential elastic scattering cross section (Fig. 2). This feature is, of course, strongest at low energies. The number of these lobes and their position depends on the material and the energy of the incident electrons. They arise from the attractive potential of an atom for electrons, whereas the repulsive potential for positrons only causes the forward peak. While the scattering lobes have no drastic effects on the stopping properties, some of their consequences should be seen, e.g., in the backscattering energy or angular distributions with good statistics and Monte Carlo Calculations of keV Electron and Positron Slowing Down in Solids. II accuracy. On the contrary, in low energy electron diffraction (LEED), the effect of the peaks and zeros is most important, and complicates the multiple scattering calculation of the L E E D intensity curves. On the other hand, in L E P D the analysis is much simpler due to the much smoother angular variation of the cross sections. At energies around 1 keV the absorption probability of positrons in solid targets is around 85-90%, which is much larger than was expected on the basis of the k n o w n electron absorption probabilities, The small backscattering probability of positrons at keV energies makes the use of low energy monoenergetic positron beams more effective. This is beneficial for the use of keV positrons as a probe to give information of the surface or near-surface region. Acknowledgements. This research has been in part supported by the Academy of Finland. Useful discussions with Dr. H. E. Hansen are gratefully acknowledged, References 1. S. Valkealahti, R.M. Nieminen: Appl. Phys. A32, 95 (1983) 2. A.P. Mills, Jr.: In Proc. L X X X I I I Intern. School of Physics 'Enrico Fermi', ed. by W. Brandt, A. Dupasquier (Academic Press, New York 1983) 3. K.F.J. Heinrich, D.E. Newbury, H. Yakowitz (eds.): Use of Monte Carlo Calculations in Electron Probe Microanalysis and Scanning Electron Microscope (US Govt. Print. Off., Washington 1976) 4. M. Gryzinski: Phys. Rev. A138, 305, 322, and 336 (1965) 5. E. Merzbacher : Quantum Mechanics (Wiley, New York 1963) 59 6. I.S. Gradshteyn, I.M. Ryhzik: Table of Integrals, Series and Products (Academic Press, New York 1965) 7. P. Hohenberg, W. Kohn: Phys. Rev. B 136, 864 (1964) W. Kohn, L.J. Sham: Phys. Rev. A140, 1133 (1965) 8. O. Gunnarsson, B.I. Lundqvist: Phys. Rev. B13, 4274 (1976) 9. M. Kotera, K. Murata, K. Nagami: J. Appl. Phys. 52, 997, 7403 (1981) 10. S. Ichimura, K. Aratama, R. Shimizu: J. Appl. Phys. 51, 2853 (1980) 11. F. Rohrlich, B.C. Carlson: Phys. Rev. 93, 38 (1954) 12. S. Fraga, K.M.S. Saxena, J. Karwowski: Handbook of atomic Data (Elsevier, New York 1976) 13. C. Nordling: Fysik Handbok (Almqvist Wiksell, Stockholm 1973) 14. R.C. Weast (ed.): Handbook of Chemistry and Physics (CRC Press, Ohio 1974-75) 15. H.H. Seliger: Phys. Rev. 100, 1029 (1955) 16. A.P. Mills, R.J. Wilson: Phys. Rev. A26, 490 (1982) 17. A.Ya. Vyatskin, V.Yu. Khramov: Radio Eng. Electron. Phys. 21, 107 (1976) 18. A.Ya. Vyatskin, V.V. Trunev: Radio Eng. Electron. Phys. 9, 1526 (1967) 19. I.R. Kanicheva, V.A. Burtsev: Fizika Tverdogo Tela 1, 1250 (1959) 20. H.J. Fitting: Phys. Status. Solidi 26, 525 (1974) 21. J. Schou, H. Sorensen: J. Appl. Phys. 49, 816 (1978) 22. R. Shimizu, S. Ichimura: Surf. Sci. 133, 250 (1983) 23. J. Kalef-Ezra, Y.S. Horowitz, J.M. Mack: Nucl. Instrum. Methods 195, 587 (1982) 24. H. Niedrig: J. Appl. Phys. 53, R15 (1982) 25. A.F. Makhov: Sov. Phys. Solid State 2, 1934, 1942, and 1945 (1960) 26. H.E. Hansen, U. Ingerslev-Jensen: J. Phys. D 16, 1353 (1983) 27. R.M. Nieminen, J. Oliva: Phys. Rev. B22, 2226 (1980)