Thin Solid Films 433 (2003) 326331

Monte Carlo simulation of electron interaction with a thin film

Vladimr Stary*
Faculty of Mechanical Engineering, Czech Technical University in Prague, Karlovo nam. 13, 121 35 Prague 2, Czech Republic

Abstract A Monte Carlo simulation (MC) of medium energy electron transport in solids is usually checked, where possible, by comparing calculated values with experimental ones. Unfortunately, some quantities are very difficult to measure, even though it is important to know them for proper analytical measurement by electron probe microanalysis. One of the most important parameters is the dimensions of the volume at which the interaction of electrons with matter produces X-ray radiation. In our previous works, reasonably good agreement was found between calculations and experiments for the electron backscattering coefficient, and also for X-ray emission, especially k-ratios (the ratio of intensities from sample and standard). By our MC code, we calculated these quantities for Au thin films on a Si substrate in dependence on the electron energy (in the range 530 keV, i.e. in the range used in SEM) and film thickness (in the range 0.050.5 mm). In this paper, we try to define the dimensions of the interaction volume for a thin film on a substrate of other material. According to this definition, and for the samples defined above, we present the values of the maximum efficient depth of X-ray production and those of the diameter of the interaction volume, calculated by our code. In addition to the produced intensities of characteristic X-ray radiation, we calculate both the radial and the depth profiles of X-ray production in the sample. 2003 Elsevier Science B.V. All rights reserved.
Keywords: Electron-matter interaction; Monte Carlo simulation; Electron probe microanalysis

1. Introduction A Monte Carlo simulation (MC) of medium energy electron transport in solids is very often used in electron microscopy, spectroscopy and microanalysis w1,2x. The reliability of physical models incorporated in a Monte Carlo simulation was checked many times by comparing calculated values with experimental ones; unfortunately, some quantities can be measured only indirectly or with some experimental problems. Among the most important parameters are the dimensions of the interaction volume, i.e. the dimensions of the volume at which the interaction of electrons with matter produces X-ray radiation. This quantity is important mainly when evaluating the composition of non-homogeneous samples, e.g. thin films on a substrate or particles in a matrix. Usually, the diameter of interaction volume D90 is defined as the diameter containing 90% of the produced (emitted) Xray intensity. Simultaneously, the depth containing 90% of the produced (emitted) X-ray intensity can be defined
*Tel.: q420-2-22435-7273; fax q420-2-2491-1406. E-mail address: stary@fsik.cvut.cz (V. Stary). 0040-6090/03/$ - see front matter PII: S 0 0 4 0 - 6 0 9 0 0 3 . 0 0 3 3 6 - 5

as the information depth. Another definition of information depth and, simultaneously, a possible way of calculating it, is given by the decay of electron energy under the lowest energy necessary for excitation of a given X-ray radiation. This depth can be measured or estimated for example using the Bethe formula for stopping power. The actual diameter of the interaction volume can be experimentally measured by moving an electron beam across the sharp boundary of two materials of different composition and simultaneously detecting the signal from element contained only in one material. However, the interface sharpness must be guaranteed and interdiffusion must be avoided, which is not too simple. MC simulation can directly provide the distribution of X-ray production in matter w3,4x, and in this way can establish some limits of the dimensions of the interaction volume. The purpose of this work is to define the dimensions of the interaction volume not only in homogeneous bulk samples, but also in case of film on the substrate and to estimate its dimensions, as well as to find their dependences on the electron energy and the thickness of the film.

2003 Elsevier Science B.V. All rights reserved.

 V. Stary / Thin Solid Films 433 (2003) 326331

327

2. Interaction of a medium energy electron with matter Elastic and inelastic collisions are usually considered as the basic interactions of primary electrons with atoms. The interaction efficiencies of elastic scattering are defined by differential cross-sections (DCS). For elastic scattering, the DCSs are calculated using quantum physics methods (e.g. w5x). These calculations are provided by using the static field approximation using the ThomasFermiDirac (TFD) potential w5x or the Dirac HartreeFockSlater (DHFS) potential for free atoms, or the HartreeFockWignerSeitz potential for atoms in solids (muffin-tin) with relativistic partial wave analysis w6x; these cross-sections can also be calculated by the PWADIR program. In single scattering calculations, we need to know the inelastic cross-sections, so the simple model of continuous slowing down cannot be employed directly. The calculation of inelastic differential cross-sections can be simplified by integrating inelastic DCS over the scattering angle. Then, inelastic scattering is described by energy loss distribution dsin(W)ydW (W is energy loss), which can be approximated by simple hyperbolic dependence w7x. This simple model of inelastic scattering can be reasonably appropriate for MC simulation of electron scattering by matter w8x. The value of the minimal energy loss Wmin is connected with the inelastic mean free path lin by the relation w7x linsWm ySB, where SB is the Bethe stopping power and Wm is the mean energy loss given as:

ps

Pn s expyPsE,u.. n!

(3)

3. Monte Carlo simulation In our MC code, we employed the single scattering model, where all these interactions are simulated. We used the PWADIR code with a muffin-tin model of atomic potential for calculating DCS of elastic collisions. For inelastic collisions, a model with hyperbolic distribution of electron energy losses was used, with Wmins 18.93 eV for Au and Wmins9.52 eV for Si. These are the optimal values of Wmin for the elements studied w8x. At low energy, the Bethe stopping power was corrected by a suitable correction w13x. The cut-off limit for electron energy was 0.1 keV. The plasmon excitation at input and output of the electron in the sample and energy losses at elastic collisions according to w14x were taken into account. We simulated the process for a thin Au film on a thick Si substrate (i.e. not transmittable for electrons of the energies used). The electron beam energy was in the range 530 keV, the film thickness was in the range 0.051.0 mm. The electrons reached the surface perpendicularly. The number of primary electrons was 2000, and the number of inelastic collisions was 105 106. As the backscattering coefficient had a relative standard deviation of approximately 2%, we suppose the number of trajectories is reasonable for relatively reliable results. In the calculations, we copied the conditions of measurement, especially the take-off angle of the detector (TOAs408). The code is written in PASCAL language. In the present code, MC simulation of the electron trajectory in a stratified sample uses the algorithm for transmission of electrons into the next layer with different composition w15x, with some corrections. In this model, the accumulated probability of collision increases during the electron path, and after transmission through the interface an electron can employ only the remnant of the accumulated probability; also, after transmission the electron keeps its direction. 4. X-Ray emission models X-Ray intensities were calculated from the probability of photon emission at an inelastic collision. For characteristic photon excitation, the Powell cross-sections w16x with the Schwaab coefficients w17x were used, sc,Psso bjzj B E E lnCcj F EEc D Ec G (4)

Wms

WfW.dW(WminlnC

Wmin

E E F D Wmin G

(1)

f(W) being the normalized distribution function. Values of Wmin were optimized by comparing MC data of the several elements with experimental data of electron elastic reflection from w9x; the obtained values of lin agree well with theoretical values given in w10x. The next type of inelastic scattering, where the electrons crossing the surface lose energy by excitation of surface plasmons, was added to the model according to w11x. The probability of escaping without surface plasmon generation is proportional to exp(yPs(E,u)); u is the input or emission electron angle. In w12x, the following formula was suggested for Ps (surface excitation parameter): 1 Pss 4y2E cosu p (2)

Due to the Poisson distribution of plasmon excitations, the probability of exciting n plasmons is:

where E is the energy of an electron; Ec is the critical energy for ionization of the shell in question; zj is the number of electrons in the jth shell; and the constant sos6.51=10y20 keV2 cm2.

328

V. Stary / Thin Solid Films 433 (2003) 326331

Fig. 1. The spectacular shape of the interaction volume, and definitions of the diameter of interaction volume D90 and the bottom depth of interaction volume H90.

In the present MC calculations, we have one element samples, the element type is selected according to the position of electron (being either in film or in substrate). The ionization probability is calculated at each inelastic scattering collision and at a given depth of the actual collision in sample z with actual electron energy. The intensity Ic of the element in question (i.e. the total number of photons of characteristic radiation of the element, detected in a solid angle DV of a hypothetical detector, and generated by the given number of electrons) is then calculated as: Ics 1 N1scvqt 4p (5)

posed of several elements are calculated. If the maximum production depth is lower than the whole sample thickness, its value is divided into 20 depth divisions. Otherwise, the sample thickness is divided in the same way. Next, radial division into 20 divisions of the same size is prepared. The X-ray produced radiation is placed into these cells, using the position of the actual inelastic collision; the intensities are calculated from the probabilities of X-ray radiation excitations. Into the radial divisions, the intensities are summed from the whole volume in a hollow cylinder of given dimensions, i.e. the signal is integrated over all azimuthal angles. 6. Results and discussion Using our MC code, we simulated the dependences of the dimensions of the interaction volume on the electron energy and film thickness. The spectacular shape of the interaction volume and the definitions of the diameter of interaction volume D90 and the bottom depth of interaction volume H90 are shown in Fig. 1. This figure was drawn according to a qualitative image of the interaction volume, using the simple MC code from w20x. The interaction volume in the substrate in the vertical direction starts at the filmysubstrate interface and continues down into the substrate. Table 1a shows the calculated shapes of the interaction volume for a homogeneous sample (Table 1a), and also for a film on the substrate and the substrate (Tables 1b and 1c, respectively). In our code, the cell dimensions are defined in mass thickness units (gycm2)in these units, the size for materials of different density is the same; the real size (e.g. in nanometers) is different. In our Table 1b and Table 1c the thickness scales are therefore not uniform, and the radius scales differ. Both radial and depth profiles of X-ray production in the samples were also calculated. Fig. 2 shows the dependences of the depth and radius of the interaction volume on film thickness. Fig. 2a shows that the depth

where scssc,P is defined above, N1sNAr yA is the number of atoms of given element per volume unit, v is the fluorescence yield calculated according to w18x, q is the ratio of the intensity of a measured line to the intensity of the whole X-ray line family (e.g. IKa y(IKaq IKb), and t is the actual trajectory before collision. Ic is summed during the whole run of program. The intensity of the emitted radiation is calculated from the produced intensity, supposing the exponential decay of this intensity given by the mass absorption coefficients of the film and substrate w19x and taking into account the actual take-off angle of the detector. 5. Cells for calculating the distribution of X-ray production in a sample In the code, the volume distribution of the produced X-ray intensities, and also both the radial and the depth profiles of X-ray production are calculated. The Bethe range is the maximum supposed depth of electron interactions in the sample, and the maximum depth of X-ray production is limited by the energy necessary for X-ray excitation. Thus, firstly, the Bethe range and the maximum production depth in a layered structure com-

 V. Stary / Thin Solid Films 433 (2003) 326331

329

Table 1a The truncated relative intensities of the produced characteristic X-ray radiation of pure Au, giving the shape of the interaction volume; at primary energy Es20 keV; 0 means relative intensity between 0 and 1, free space or means zero intensity in the cell Layer Depth (nm) 56.5 113 169 226 282 339 395 452 508 565 621 678 734 791 847 Outer cylinder radius (nm) 56 9 6 2 1 1 0 0 0 0 0 0 0 113 3 5 5 3 2 1 1 0 0 0 0 0 0 169 3 4 4 4 2 2 1 0 0 0 0 0 0 226 2 3 3 3 2 1 1 0 0 0 0 0 0 282 1 2 2 2 2 1 1 0 0 0 0 0 339 1 1 1 1 1 1 0 0 0 0 0 0 395 0 0 1 1 1 0 0 0 0 0 0 0 452 0 0 0 0 0 0 0 0 0 0 0 508 0 0 0 0 0 0 0 0 0 0 0 0 0 565 0 0 0 0 0 0 0 0 0 0 0 0 0 621 0 0 0 0 0 0 0 0 0 0 0 0 0 678 734 791 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 847 0 0 904 0 960

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

0 0 0

0 0

Layer 1 is the Au film; layer 2 is the Si substrate.

Table 1b The truncated relative intensities of the produced characteristic X-ray radiation of the Si substrate with a thin film of Au; primary energy Es20 keV, film thickness 259.1 nm; 0, free space and have the same meaning as in Table 1a Layer Depth (nm) Outer cylinder radius (gycm2)=1000, the 2nd row in nm 0.069 0.137 0.206 0.275 0.343 0.412 0.481 0.550 0.618 0.687 0.756 0.824 0.893 0.962 1.03 1.10 1.17 1.24 1.31 1.37 295 588 884 1180 1472 1768 2064 2361 2652 2948 3245 3536 3833 4129 4421 4721 5021 5322 5622 5880 3 2 1 0 0 0 0 0 0 0 6 7 3 1 0 0 0 0 0 0 8 8 5 3 1 1 0 0 0 8 7 6 4 2 1 0 0 0 9 8 5 3 2 0 0 0 0 7 8 5 3 1 0 0 0 6 5 4 3 1 0 0 0 6 4 2 1 0 0 0 0 4 4 2 1 1 0 0 2 2 1 1 0 0 0 2 2 0 0 1 0 0 1 1 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2

35.6 71.2 106.8 142.4 178.0 213.5 249.1 471.6 766.4 1061.2 1356.0 1650.8 1945.6 2240.4 2535.2 2830.1 3124.9 3419.7

Layer 1 is the Au film, layer 2 is the Si substrate.

Table 1c The truncated relative intensities of the produced characteristic X-ray radiation of the Au thin film on the Si substrate; primary energy Es20 keV, film thickness 259.1 nm; 0, free space and have the same meaning as in Table 1a Layer Depth (nm) Outer cylinder radius wgycm2x=1000, the 2nd row in nm 0.069 36 9 7 4 1 1 0 0 0 0.137 71 2 4 4 3 2 1 1 0 0.206 107 1 3 3 3 2 2 1 0 0.275 142 1 2 3 3 2 2 1 0 0.343 178 1 2 2 2 2 2 1 0 0.412 213 1 1 2 2 1 2 1 0 0.481 249 1 1 1 1 1 1 1 0 0.550 285 0 1 1 1 1 1 0 0 0.618 320 0 0 0 1 0 0 0 0 0.687 356 0 0 0 0 0 0 0 0 0.756 392 0 0 0 0 0 0 0 0 0.824 427 0 0 0 0 0 0 0 0 0.893 463 0 0 0 0 0 0 0 0 0.962 498 0 0 0 0 0 0 0 0 1.03 534 0 0 0 0 0 0 0 0 1.10 570 0 0 0 0 0 0 0 1.17 606 0 0 0 0 0 0 0 1.24 642 0 0 0 0 0 0 1.31 679 1.37 710

1 1 1 1 1 1 1 2 2

35.6 71.2 106.8 142.4 178.0 213.5 249.1 471.6 766.4

Layer 1 is the Au film, layer 2 is the Si substrate.

330

V. Stary / Thin Solid Films 433 (2003) 326331

Fig. 2. Dependences of the depth and radius of the interaction volume on electron energy and film thickness: (a) depth of the interaction volume in the Au film (H90Au); (b) diameter of the interaction volume in the Au film (D90Au); (c) depth of the interaction volume in the Si substrate (H90Si); (d) diameter of the interaction volume in substrate (D90Si).

of the interaction volume in the film material (H90Au) is limited by the film thickness, and increases with electron energy up to the value in pure Au (which is indicated at the thickness of Au film dAus1200 nm). The higher values of H90Au at the lowest thickness of Au film are given by the cell size value, because automatic division of interaction volume cannot position a cell exactly on the filmysubstrate interface. This also caused the non-zero intensity of the Au radiation in the last but one row of Table 1c, where the filmysubstrate interface crosses the cell (the thickness of the film was 259.1 nm, so the thin layer of the film is contained in these cells). The depth of the interaction volume in the Si substrate (H90Si) decreases from the value in pure Si due to the scattering in the Au film, which increases with Au film thickness. It is clear that at some thickness of the film no excitations proceed in the substrate; in this case H90Si reaches the zero value (though in the non-zero case the film thickness is added to H90Si see the definition in Fig. 1). The diameter of the interaction volume in the Au film (D90Au) (Fig. 2b) starts to increase relatively quickly with film thickness, and at a certain value of film

thickness (depending on the electron energy) the rate of growth is suppressed and the D90Au remains constant (and approx. equal to the bulk Au value). The diameter of the interaction volume in substrate D90Si also increases with film thickness in the observed thickness range, due to the stronger elastic scattering of electrons in Au than the scattering in Si. The sudden fall to the final zero value appears at a thickness equal to the maximum depth of penetration of electrons at a given primary electron energy (see the case at electron energy 5 and 10 keV). At this thickness the Si X-ray produced intensity also falls to zero. The dimensions of the interaction volume in the substrate do not reflect in absolute scale the change in the excited X-ray intensity, which decreases substantially as the film thickness grows. The dimensions of the interaction volume should be compared with experimental estimations, but these values can be measured only indirectly and with relatively low precision; also, the more or less precise semiempirical formulas that are usually used provide only a rough estimation.

 V. Stary / Thin Solid Films 433 (2003) 326331

331

7. Conclusion In our previous works, the MC data of electron backscattering and also of the k-ratios for thin films on substrate, in dependence on the energy of the electrons and on the film thickness, were compared with the experimental values, with reasonably good agreement w8,15x. Thus, we suppose it provides the required assessment of our MC code. Using this code, we have obtained reasonable estimations of the radial and vertical dimensions of the interaction volume and their dependence on the electron energy and on the film thickness. These findings are important for Electron Probe Microanalysis of non-homogeneous samples, e.g. for the analysis of films on substrates. Acknowledgments For the English language review, we are grateful to Mr Robin Healey. This work was supported by GA-CR grant No. 202y02y0237. References
w1x L. Reimer, Microchimica Acta 13 (1996) 1, wsupplx. w2x R. Shimizu, Z.-J. Ding, Rep. Progr. Phys 55 (1992) 487.

w3x D.F. Kyser, in: J.J. Hren, J.I. Goldstein, D.C. Joy (Eds.), Introduction to Analytical Electron Microscopy, Plenum, New York, 1989, p. 199. w4x K. Murata, H. Kawata, K. Nagami, Scanning Micr. Suppl. 1 (1987) 83. w5x A. Jablonski, Phys. Rev. B 13 (1991) 7546. w6x F. Salvat, R. Mayol, Comput. Phys. Commun. 74 (1993) 358. w7x D. Liljequist, J. Phys. D Appl. Phys. 11 (1978) 839. w8x V. Stary, J. Phys. D Appl. Phys. 32 (1999) 1811. w9x R. Schmid, H.K. Gaukler, H. Seiler, Scan. Electr. Micr. V.II (1983) 501. w10x S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interf. Anal. 17 (1991) 911. w11x Y.F. Chen, J. Vac. Sci. Technol. A 13 (1995) 2665. w12x Oswald R., Ph.D Thesis, Eberhard-Karls-Universitat, Tuebingen, 1992. w13x D.C. Joy, S. Luo, Scanning 11 (1989) 176. w14x V. Stary, in: A. Kling, F. Barao, M. Nakagawa, L. Tavora, P. Vaz (Eds.), Proc. of MC-Advanced Monte Carlo for Radiation Physics, Particle Transport Simulation and Applications, Lisbon, 2326 October 2000, Springer, Berlin, 2000, p. 369. w15x V. Stary, Microchimica Acta 139 (2002) 179. w16x C.J. Powell, Ultramicroscopy 28 (1989) 24. w17x P. Schwaab, Scanning 9 (1987) 1. w18x B.H.S. Burhop, J. Phys. Radium 16 (1955) 625. w19x J.I. Goldstein, D.E. Newbury, P. Echlin, D.C. Joy, C. Fiori, E. Lifshin, Scanning Electron Microscopy and X-Ray Microanalysis, Plenum Press, New York, 1981, p. 620. w20x T.D. Ly, D.G. Howitt, Scanning 14 (1992) 11.