Applied Surface Science 167 Ž2000. 12–17 www.elsevier.nlrlocaterapsusc Work function change caused by alkali ion sputtering of a sample surface Yuriy Kudriavtsev ) , Rene Asomoza Dep. Ingenieria Electrica-SEES, Centro de InÕestigacion y Estudios AÕanzados del IPN, AÕ. IPN a2508, Apdo Postal 14-740, Mexico 14 D.F. 07300, Mexico Received 20 January 2000; accepted 7 May 2000 Abstract The steady state ion sputtering of a solid surface was studied. Based on a simple model, the surface concentration of ions implanted during sputtering was calculated in the case of cesium ion sputtering of silicon. The corresponding work function shift was calculated using the model developed. q 2000 Published by Elsevier Science B.V. Keywords: Ion sputtering; Ion implantation; Work function 1. Introduction During ion sputtering, a part of the primary ions may be implanted into a near surface layer of the bombarded target. In the case of alkali ion sputtering, such implantation leads to a change of the work function of the target surface. A decrease of the work function leads to an increase of negative ion emission. This effect is used in SIMS as a standard method for the measurement of electronegative elements. According to a current theoretical model the ionization probability of sputtered particles depends on the work function F as follows PyA exp y Ž F y A . r´n , Ž 1. in the case of positive secondary ions. Here, IP is the ionization potential, A is the electron affinity, ´ n and ´ p are parameters which depend on the normal component of the ion emission velocity w1x. So, for interpretation of SIMS data, it is important to know the real value of the work function. It was experimentally demonstrated that the shift of the work function is proportional to the surface concentration of alkali atoms w2–4x. It is, however, a difficult task to measure the real surface concentration of alkali atoms during alkali ion sputtering of samples. This paper is devoted to a simple method for calculation of the surface concentration of ions implanted during sputtering as well as the corresponding change of the work function. in the case of negative, and PqA exp y Ž IP y F . r´ p , ) Ž 2. 2. Surface concentration of ions implanted into target during sputtering Corresponding author. Fax: q52-5747-7114. E-mail address: [email protected] ŽY. Kudriavtsev.. The first step of our study was the calculation of the surface concentration of the implanted projec- 0169-4332r00r$ - see front matter q 2000 Published by Elsevier Science B.V. PII: S 0 1 6 9 - 4 3 3 2 Ž 0 0 . 0 0 3 8 1 - 0 Y. KudriaÕtseÕ, R. Asomozar Applied Surface Science 167 (2000) 12–17 tiles. Below, we present shortly the simple model developed in Ref. w5x. Two assumptions form this model. Ž1. It is well known that the number of implanted Žduring sputtering. ions may reach values around 10% Žrelative concentration. and it looks reasonable to consider the near surface layer as a binary system consisting of target atoms with a concentration of Ct and implanted ions with a concentration of Ci . Hence, it follows that we can describe the sputtering of this layer as the sputtering of two component targets. The total sputtering yield, in this case, is given by Ytot s Yi q Yt , Ž 3. where Yi and Yt are the partial sputtering yield of implanted particles and target atoms, respectively. The thickness of this layer is the result of a competition between ion implantation and ion sputtering. Ž2. Under the steady state sputtering regime, the number of primary ions irradiating the surface Ž jo . becomes equal to the secondary flux of the same particles due to scattering-jo R and resputtering-jo Yi jo s jo R q jo Yi , Ž 4. here R is the reflection coefficient. For steady state sputtering, Patterson and Shirn w6x developed the following equation Yi s Yt Civ Ctv , Ž 5. where C stands for concentration and the superscription v means AvolumeB Žsee below.. A combination of the Eqs. Ž4. and Ž5. with the evident relation Civ q Ctv s 1 leads to the equation for the relative concentration of implanted particles Civ f b Yt q b , Ž 6. where b is the coefficient of accommodation: b s 1 y R. It is well known that the sputtering of binary systems is usually accompanied by the preferential sputtering of one of the elements. As a result, the binary layer, which we study, may be separated into two AsublayersB with different compositions: the AsurfaceB layer, which consist of approximately two monolayers on top, and an undersurface layer. The concentration of implanted projectiles in the under- 13 surface layer may be calculated using Eq. Ž6. ŽAvolumeB concentration.. The AsurfaceB composition we calculated using well-known Andersen– Sigmund equation w7x Yi s Yt Cis Mt 2m Cts Mi Ut 1y2 m , Ui Ž 7. here Uj and M j are the surface binding energy and mass of the components, respectively; m is a parameter laying in the range from 0 to 1, whose value depends on the primary ion energy. Combining Eqs. Ž7. and Ž4., and taking into account that Cis q Cts s 1, we can obtain the formula for the relative AsurfaceB concentration of implanted projectiles Cis f b Yt w MtrMi x 2m w UtrUi x 1y2 m q b . Ž 8. Note that we did not take into account the AsecondaryB effects like diffusion and segregation accompanying the sputtering. Now, concerning b , for Mi ) Mt its value is close to 1 w5x. We have used Eq. Ž8. for the calculation of the Cs concentration at the silicon surface, during Cs ion sputtering. Eq. Ž8. contains the sputtering yield of the target atoms Yt . So, the next step of our approach was the study of the energy dependence of the sputter yield Yt . We used the semi-empirical model of Yudin w8x. According to this model, the sputtering yield is inversely proportional to the mean path length of an ion in the substrate between displacement collisions. Yudin suggested an empirical formula for the stopping power as follows Sn f c'e Ž 9. dqe where e is the reduced energy: e s EF, E is the incident energy, c s 0.45 and d s 0.3 are empirical constants and the ratio F is obtained by the following equation F s 6.92 P 10 6 a tf Mt Zi Zt Ž Mi q Mt . , Ž 10 . where Z j and M j are the atomic number and mass of primary ions and target atoms, respectively; a tf is the Thomas–Fermi screening length: a tf s Y. KudriaÕtseÕ, R. Asomozar Applied Surface Science 167 (2000) 12–17 14 0.8853 A brŽ Zi2r3 q Zt2r3 .1r2 Žhere A b s 5.29 = 10y9 cm is the Born radius.. For normal ion incidence, Yudin derived the following formula for the sputtering yield Yt s 2Ymax (ErE max 1 q ErEmax , Ž 11 . where Emax s 0.3rF is the energy of the primary ions at which the sputtering yield reaches its maximum value Ymax s L Ž Zi ,Zt . Nps 2r Ž FUo . , Ž 12 . here, N is the atomic concentration of the target, s s p a 2tf is the screening cross-section, Uo is the surface binding energy of the target. The coefficient LŽ Zi ,Zt . has the following empirical dependence Fig. 2. The surface and undersurface ŽAvolumeB . concentration of cesium implanted in silicon as a function of the energy of primary ions ŽCsq .. L Ž Zi ,Zt . s L Ž Zt . y 4.65 P 10y1 2 Ž Zi y 18 . , Ž 13a . considered normal incident of the projectiles. In the case of other angles of incidence for our consideration Ž Mi 4 Mt ., the empirical formula derived in Ref. w9x may be used where L Ž Zt . s 1.3 P 10y1 0 Zt0.5 ž = 1 q 0.25sin 2p Zt q Z Zo q pr2 / Yt Ž u . . Ž 13b. For Zt - 18, Zo s 8 and Z s 0; and for Zt ) 18, Zo s 18 and Z s 2. Fig. 1 shows the energy dependence of the sputter yield of silicon under Csq ion bombardment. We Fig. 1. The sputtering yield of silicon as a function of the primary Csq ion energy, calculated using Yudin’s model for two angle of incidence of the primary ions: 08 and 428 off to the normal incidence. Yt Ž 0 . 1 f cos Ž u . , Ž 14 . where u is the angle of incidence of primary ions Žsee the second curve in Fig. 1.. The calculated energy dependence of the sputtering yield allows us to analyze the dependence of the surface and undersurface concentration of implanted projectiles on the primary ion energy. Fig. 2 demonstrates the energy dependence of the surface and undersurface concentrations of Cs in the case of the steady state sputtering of silicon by Csq ions. Following a common practice we used in our calculations ŽEq. Ž8.. the sublimation energy of element as its surface binding energy. The surface concentration of Cs is approximately two times less than undersurface concentration. The reason for this effect is a very low surface binding energy of Cs Ž0.812 eV. in comparison with Si Ž4.74 eV. w8x. 3. Change of work function due to ion implantation The change of the work function of different materials upon adsorption of alkali metals has been Y. KudriaÕtseÕ, R. Asomozar Applied Surface Science 167 (2000) 12–17 Fig. 3. Schematic illustration of cesium atoms deposed on Ža., and implanted into Žb. silicon target. studied for a long period. A depolarization model of Tooping w10x developed in 1927, based upon a depolarization field formed by the dipole moment of adparticles, has frequently been used. According to this model, the work function change DF depends on the number of adparticles per unit surface area Nad as follows w11x DF s " eNad po ´o Ž 1 q a 9Nad3r2 . , Ž 15 . where e is the elementary charge, ´o is the permittivity of vacuum, po is the dipole moment of an isolated adparticle, and a is the polarizability of the adparticle. It was demonstrated in Ref. w12x that the polarizability of the adparticles is a function of the density of adatoms Nad : for low coverage, the polarizability is constant and is approximately equal to the polarizability of the ion up to about 0.25 ML. Then the polarizability increases with coverage and reaches a maximum at 0.7 % 1.0 ML of coverage. The maximum value of a is between 50% and 75% of the gas 15 phase value. The dipole moment is extracted from the slope of the work function Žexperimental. as a function of coverage in the limit of zero coverage. Taking this into account, it is difficult to use Eq. Ž15. for any estimation. However, a more important question is the possibility to use the described model Ž15. in the case of ion bombardment. Indeed, the position of the implanted alkali particles differs radically from the position of adsorbed particles Žsee Fig. 3.. The ionization energy of Cs atom Ž3.89 eV. is lower than the work function of silicon Ž4.85 eV. Žas well as than the work function of other semiconductors and metals.. Thus, the cesium should be absorbed as a positive ion or, more exactly, as a strongly polarized atom ŽFig. 3a.. In this case, the layer of the adparticles produces an electrical field, which reduces the work function of the target. The obvious illustration for specific character of this effect was done in Ref. w13x: the Na deposition on metallic Cs leads to decrease of the work function of Cs. The work function change achieves y0.06 eV for small coverage of Na. Another world, the charge transfer occurs from Na to the more electropositive Cs Žthe work functions of Cs and Na are presented in Table 1.. In the case of ion bombardment, the alkali particles are incorporated into the lattice ŽFig. 3b.. The implanted particles are in close association with neighbor atoms and previously described effect caused by deposed dipoles is absent. Generally, the molecular bond between implanted Cs atoms and target atoms ŽSi. is partly ionic in contrast with the bond between Si atoms in pure silicon. So we can suggest that the change of the work function is caused by Cs implantation. Additionally, the work function can be modified due to amorphization of the near surface layer caused by ion bombardment. It was experimentally obtained w14x that the work function of samples covered by monolayers of alkali metal atoms is close to the work function of the Table 1 Work function of some elements w16x Element F ŽeV. Element F ŽeV. Si 4.85 4.91 Ž100. 4.60 Ž111. Ga Na Cs 4.2 2.5 2.14 Y. KudriaÕtseÕ, R. Asomozar Applied Surface Science 167 (2000) 12–17 16 Fig. 4. The shift of the work function of silicon as a function of the surface concentration of cesium implanted during sputtering. corresponding alkali metal. Some differences are associated with the crystalline structure of the target. There is a great difference in the surface states distribution, it depends on the crystalline orientation of the surface and on the target temperature w14,15x. The adsorption of alkali metals can result to the bond bending effect, which also modifies the work function w15x. It is clear that all AcrystallineB effects are insignificant in the case of amorphous target. So we can skip these effects in our future consideration of Cs sputtered surfaces. Everything discussed above allows us to suggest that the work function change of the sample under ion bombardment strongly correlates with the surface composition change. And we wrote the work function Fq of an ion irradiated target as AweightedB t average of the work functions of components Žtarget atoms and implanted atoms in this case. 1 Cis Cts f q , Ž 16 . Fq Fi Ft t where F j is the work function of pure target Žt. and the pure material used as the bombarding ions Ži., and C js is the surface concentration of the components. A simple transformation of Eq. Ž16., taking into account the evident relation: Fq t s F t y DF Žwhere DF is the change of work function due to the ion implantation., leads to an equation that correlates the change of the work function with the surface concentration of implanted ions Fi DF f F t 1 y s . Ž 17 . Ci Ž F t y F i . q F i ž / The work function varies between the limits of the work function of the pure matrix F t Žbefore ion sputtering. and the work function of ion pure material F i , in the case when the concentration of these ions on the sample surface achieves 100% Žthe physical limit.. The change of the work function of silicon as a function of cesium surface concentration calculated using Eq. Ž17. is presented in Fig. 4. In our calculations, we used the experimentally obtained work functions of pure materials w16x ŽTable 1.. Note that we have used the work function of polycrystalline silicon to take into account the amorphization of a near surface layer due to ion bombardment. For comparison, we present in Fig. 4 the curve calculated using Tooping’s model, Eq. Ž15.. We ˚ 3 w12x Žaphave used for cesium a equals to 38 A proximately 64% of gas phase value.. The dipole moment po was calculated from the AboundaryB conditions: DF is equal to the difference between the work function of pure silicon and silicon covered with 1 ML of Cs. Additionally, we calculated the change of the work function in the different regimes of sputtering of Cameca ims-4f% 6f ion microprobes Žsee Table 2.. The analytical parameters: the primary ion energy Ž E . and the angle of incidence Ž u ., as well as the sputtering yield Ž Yt ., the surface and undersurface concentration of Cs, calculated using ŽEqs. Ž6., Ž8., Ž11. and Ž14. are presented in the Table 2. It is seen that the work function decreases with decrease of the primary ion energy. According to Eq. Ž1., this leads Table 2 s . v . Sputtering yield Yt , surface Ž CCs and undersurface Ž CCs concentration of Cs, and corresponding change of the work function Ž DF ., calculated using Eqs. Ž6., Ž8., Ž11., Ž14. and Ž17.. Typical regimes of Cameca ims-4f%6f instruments were studied in the case of Csq ion sputtering of silicon No. E ŽkeV. u Ž8. Yt Žatomsr ion. v CCs Ž%. s CCs Ž%. DF ŽeV. 1 2 3 4 5 6 7 1 1.6 2 3 5.5 10 14.5 60 17.8 52.2 54.7 42 24.7 24 3.15 2.08 3.61 4.66 4.85 5.22 6.11 24.1 32.4 21.7 17.7 17.1 16.1 14.1 7.1 10.3 6.2 4.9 4.7 4.4 3.8 y0.398 y0.560 y0.353 y0.282 y0.272 y0.255 y0.220 Y. KudriaÕtseÕ, R. Asomozar Applied Surface Science 167 (2000) 12–17 to an increase in the ionization probability of the secondary negative ions. But simultaneously, the decrease of the primary ion energy causes a decrease in the sputtering yield. So the monitored SIMS signal of analyzed elements will be a result of a competition of two noted effects. We must note a lack of experimental results, which correlates the work function change with the concentration of implanted alkali atoms. A comparison of the calculated work function change and the measured one reported in Refs. w4,17x is difficult because of a conflicting character of the experimental data. In both cases, authors studied the Cs implanted silicon. The authors of Refs. w4,17x obtained the same shift of the work function of 0.55 eV. Note that authors of these works used different sputtering regimes: in Ref. w4x, the sputtering was performed by Gaq ions and in Ref. w17x, by Csq ions. However, the Cs concentration in the case of Ref. w17x was 30.5% and in Ref. w4x, it was equal to 3.3% Žthe bulk concentration in both cases.. Moreover, the authors of both works did not take into account the possible preferential sputtering of Cs and the change of the surface concentration of Cs in comparison with its bulk concentration during their measurements. Taking this into account, we recalculated the surface concentration of Cs for both studies, Refs. w4,17x, using Eq. Ž7.. The obtained surface concentration was equal to 9.5% and 0.86% for experiments of Refs. w17,4x, respectively. These data are presented in Fig. 4. Note that obtained difference can be partly explained by difference in primary ions used for sputtering. We can note that the data of Ref. w17x showed a good agreement with calculations based on the proposed model, Eq. Ž17.. The data of Ref. w4x are close to the prediction of Tooping’s model. Another result was reported in Ref. w18x. Its author measured the work function change for initial stage of cesium incorporation on the Csq irradiated silicon surface. The work function change of y1.3 eV was obtained for the steady state sputtering with energy of 5.5 keV and the angle of incidence of 428. Corresponding cesium concentration at the surface layer was calculated Žusing T-DYN code. as 12%. The result is presented in Fig. 4 and lies between two AtheoreticalB curves. In such manner, the experimental data are rather conflicting. In order to find a ArightB model, it is 17 necessary to obtain experimentally the work function change as a function of the surface cesium concentration. As a next step in our study, we are going to perform such experiments. 4. Conclusions A simple technique for calculating of the work function of targets, which are being under ion bombardment, was presented. Based on this model, the shift of the work function of silicon due to cesium ion sputtering was calculated for a number of experimental regimes. References w1x N.D. Lang, Phys. Rev. B 27 Ž1983. 2019. w2x T. Yasue, T. Ueyama, M. Takada, R.S. Li, T. Koshikawa, in: G. Gillen, R. Lareau, J. Bennet, F. Steive ŽEds.., Secondary Ion Mass Spectrometry SIMS-XI, Wiley, 1997, pp. 927–930. w3x H. Gnaser, in: G. Gillen, R. Lareau, J. Bennet, F. Steive ŽEds.., Secondary Ion Mass Spectrometry SIMS-XI, Wiley, 1997, pp. 891–894. w4x M. Ferring, Th. Mootz, F. Saldi, H.N. Migeon, in: G. Gillen, R. Lareau, J. Bennet, F. Steive ŽEds.., Secondary Ion Mass Spectrometry SIMS-XI, Wiley, 1997, pp. 911–914. w5x Yu. Kudriavtsev, Nucl. Instrum. Methods B 160 Ž2000. 307–310. w6x W.L. Patterson, G.A. Shirn, J. Vac. Sci. Technol. 4 Ž1967. 343–355. w7x N. Andersen, P. Sigmund, K. Dan, Vidensk. Selsk., Mat.-Fys. Medd. 39 ŽN3. Ž1974.. w8x V.V. Yudin, Elektron. Tekh., Ser. Poluprovodn. Prib. 172 Ž1984. 3–16, Žin Russian.. w9x H. Andersen, H. Bay, in: R. Behrisch ŽEd.., Sputtering by Particle Bombardment I, Springer-Verlag, 1981. w10x J. Tooping, Proc. R. Soc. London A114 Ž1927. 67. w11x E.V. Albano, Appl. Surf. Sci. 14 Ž1982–1983. 183–193. w12x R.W. Verhoef, M. Asscher, Surf. Sci. 391 Ž1997. 11–18. w13x F. Xu, G. Manico, F. Ascione, A. Bonanno, A. Oliva, Phys. Rev. B 54 Ž1996. 10401. w14x T. Aruga, Yo. Murata, Prog. Surf. Sci. 31 Ž1992. 61–130. w15x H.J. Clemens, J. Von Wienskowski, W. Monch, Surf. Sci. 78 Ž1978. 648–666. w16x R.C. Weast, M.J. Astle ŽEds.., Handbook of Chemistry and Physics, 61st edn., CRC Press, Boca Raton, FL, 1980–1981. w17x Th. Mootz, W. Bieck, H.N. Migeon, in: G. Gillen, R. Lareau, J. Bennet, F. Steive ŽEds.., Secondary Ion Mass Spectrometry SIMS-XI, Wiley, 1997, pp. 953–956. w18x H. Gnaser, Phys. Rev. B 54 ŽN23. Ž1996. 16456–16459.
© Copyright 2024