Alkali migration in ion irradiated glasses

511 Nuclear Instruments and Methods in Physics Research Bl (1984) 511-515 North-Holland, Amsterdam ALKALI MIGRATION IN ION IRRADIATED GLASSES G. BATTAGLIN ‘, G. DELLA MEA ‘, G. 3E MARCH1 ‘, P. MAZZOLDI 2 and A. MIOTELLO 3 ’ Unitri GNSM-CNR, Dipartimento di Fish, Via Marzolo 8, 35100 Padovq Italy ’ Laboraiori Nazionali INFN Legnaro- Padova; Dipartimento di Fisk, ’ Istituto per la Ricerca Scientifica e Tecnologica, Povo (Trento), Italy Via Marzolo 8, 35100 Padova, Italy Ion implantation into alkali silicate glasses induces an alkali ion migration which modifies the glass composition in the implanted region. The alkali migration can be correlated to different mechanisms, clearly connected to the different stopping power regimes of incident particles. In the nuclear stopping power regime (heavy-ion irradiation) the observed alkali depletion at the surface has been interpreted on the basis of a phenomenological model, which takes into account a preferential ejection of alkali atoms from the surface and of an enhanced diffusion over the range of implanted ions. Glasses with different alkali contents have been irradiated by using Ar ions at different energies, currents and doses. Theoretical models are presented to explain the experimental results. The alkali depth profiles, before and after irradiation, have been determined by using Rutherford backscattering spectrometry and resonant nuclear reactions. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 1. Introduction 2. Experimental results Ion or electron irradiation of alkali silicate glasses induces an alkali ion migration which modifies the glass composition in the implanted region [l-6]. The alkali migration can be correlated to the different stopping power regimes of incident particles [7-91. In the electronic stopping power range (loss-mass particle irradiation) the alkali profile modification has been explained on the basis of the ordinary and field assisted diffusion processes [4,8]. However in the case of the nuclear stopping power regime (heavy ion bombardment) a phenomenological model, which takes into account a preferential sputtering mechanism of alkali atoms and an enhanced diffusion over the range of implanted ions has been introduced [9,10]. In this paper we present an improved study of the alkali (in particular sodium migration during Ar-irradiation with a quantitative analysis of the experimental results accomplished by numerical integration of the continuity equation for the enhanced alkali diffusion with a boundary condition for the surface ejection of alkali atoms. Glasses (of molar composition reported in table 1) have been irradiated with Ar ions at different doses and energies. The Ar+ implants were performed at the Laboratori Nazionali di Legnaro (LNL) in the dose range 10’5-10’7 ions/cm* at implantation energies between 50 and 100 keV. The current density varied between 0.5 and 3 PA/cm*. To minimize charge build-up during implantation, the samples were partially covered by a metal mask [Ill. The alkali profile modifications were obtained by using nuclear techniques (Rutherford backscattering of 1.8 MeV He4+ and the nuclear reaction 23Na(p, a)*‘Ne [12]) at low current to avoid alkali migration during the analysis. The effects of 50 keV-Ar+ irradiation as a function of dose, already reported in previous papers [3,9,10], are summarized in fig. 1 for the further discussion. We observed a Na depletion increasing with the irradiation dose. Steady-state profiles were obtained for implantation doses higher than 4 X 1016 ions/cm*. We extended the study by varying the current density between 0.5 and 2 PA/cm*. The Na profiles appeared to be independent of the Table 1 % molar composition of the glasses used in the irradiation experiments SO, W3 AW3 Na,O Glass 1 Glass 2 71.5 77.7 _ 4.8 0.8 2.6 13.8 _ Glass 3 77.4 4.8 2.6 0168-583X/84/$03.00 0 Elsevier Science Publishers (North-Holland Physics Publishing Division) I&O 14.9 Rb,O CaO MgO 8.7 5.2 _ 15.2 B.V. VIII. GLASSES G. Battaglin zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA et al. / Alkali migration in ion .irradiated ghses 512 I I 50 keV- Ar' r 1 . zyxwvutsrqponmlkjihgfedcbaZYXW lOOk &A,+ ’ ’ L----2.10%,m2 l ----4.10%m2 4----5.10Ycm2 .----1OTcd 07w DEPTH (nm) DEPTH [nml zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Fig. 1. Fxperimental (see refs. 3, 9 and 10) and theoretical sodium profiles (curves) after 50 keV-Ar irradiation at different doses. For the calculation, a D*r-value, 2 x lo- l4 cm2 s-l, greater than the D value evaluated for a temperature activated mechanism, has been assumed. implantation current. In fig. 2 we show the alkali depletion in glasses containing Rb,O or K,O, analyzed by using Rutherford backscattering after 50 keV-Ar’ irradiation at a dose of 4 x lOI ions/cm*. The Na profile in soda-lime glasses for the same irradiation conditions, is also reported to compare the alkali depleted layer thicknesses. The alkali depletion mechanism seems to be inda pendent of the particular alkali Na, Rb, K element but related only to incident ion energy. Fig. 3 shows the sodium ~st~bution after 100 keV-Ar+ i~adiation for different impl~~tion doses. The steady-state profile is reached for a dose of 10” ions/cm2. We analyzed the Na concentration modifications as a function of implantation energy between 50 and 100 keV, for a dose at which a steady-state profile was attained (10” ions/cm2). The results are reported in fig. 4. The Na depleted region presents a thickness increasing with the Fig. 3. Experimental and theoretical sodium profiles (curves) after 100 keV-Ar+ irradiation at different doses. The current density was 2 fi/cm’. The D** and D*, values used in the calculation are reported in tablt -2. implantation energy. In fig.5 we show the alkali depletion for 90 keV-Ar+ glasses at the three different compositions. The Ar dose was 4 X 1016 ions/cm2 and the current density 2 PA/cm*. The results confirm the independence of the alkali depleted region thickness on the particular alkali element. From the results reported in figs. l-5 we can underline the following-_points: (1) the thickness of the alkali depleted layer is a function of the implantation dose, reaching a steady-state value. (2) The alkali profile, for a fixed irradiation dose, is independent of the incident ion current density. (3) The depleted layer thickness increases with the implantation energy. (4) The alkali depletion mechanism seems to be independent (at least for near room temperature) of the alkali element present.in the glass, i.e., we obtained comparable results, in particular for the thickness of the depleted layer, for a fixed implantation dose and o Na OK . Rb po 0 I 100 I 200 -UNlMPLANTEO A---E==% keV a--E=60 keV O----E=80 keV l..-..E-1MkeV I 300 DEPTH hml Fig. 2. Experimental alkali depletion in three glasses containing Rb,O, KaO and NasO (see table 1) after 50 keV-Ar+ irradiation at a dose of 4x 1Ol6ions/cm*. The current density was 2 f&/cm’. DEPTH(nml Fig. 4. Experimental and theoretical sodium profiles (curves) after irradiation at different energies. The D*, and D*, values used in the calculation are reported in table 3. G. Battaglin et al. / Alkali migration in ion irradiated glasses I s I I tion is clearly evident with DT > 02 [9,16]. Whereas Df 90 kzyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA eV 4 * ldr A&m2 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA is related to enhanced diffusion mechanisms, 0; is mainly determined by the temperature increment during the implantation [9,19], at least for implantation energies at which extended defects are not produced [16] (see below). We account for the discontinuity in the diffusion coefficient value at a depth xo by the condition I o Na .K . Rb 90 0 I 100 energy, in glasses containing RbzO, K,O or Na,O. The reported experimental results may be qualitatively described as follows [9]: the alkali atom removal rate (larger than for other matrix components), preferential sputtering [13], and the subsequent migration towards th Al surface layer involved in the sputtering processes (Al < 10 A [14]), induce the observed alkali depletion up to a large depth. The alkali migration towards the Al surface layer may be described [9,15,16] from a mathematical point of view by the transport equation, in a one-dimensional form: ac(x,=D* a*+, t) at ax* a+, t) DfT=D*-,x=x a+, t) D’ zyxwvutsrqponmlkjihgfedcbaZYX I I In fig. 1 the calculated Na profiles (lines) are wm300 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA 200 OEPTH Lnm) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA pared to the experimental results. The Dj+ value is Fig. 5. Experimental alkali depletion in three glasses containing Rb,O, K,O and Na,O (see table 1) after 90 keV-Ar irradiation at a dose of 4~ 1016 ions/cm*. The current density used was 2 PA/cm’. + p(x9 4 ax where c(x, t) is the alkali concentration at a time t and a depth x and D* is the effective diffusion coefficient. D* accounts for the enhanced transport of alkali atoms in the matrix region affected by defect production, due to the local energy deposition in nuclear collision processes of incoming ions. The last term in eq. (1) takes into account a matrix erosion process, through matrix erosion speed, U. For the implantation energies used in the present experiments, the last term in eq. (1) can be neglected [17]. The use of the diffusion transport instead of the Boltzmann transport equation is justified under conditions of high energy deposition densities [18]. We account for the preferential sputtering of alkali atoms from the surface by means of the condition D* 513 ac(x, 9 -==Hc(x,t),X=O ax assuming that the alkali loss is proportional to the concentration itself at the surface [15,16]. Na experimental profiles suggest that a discontinuity in the diffusion coefficient D* must be assumed at a depth xo where a change in the profile concavity direc- 2 ax 2 X lo-l4 cm2/s, for an irradiation current density of 2 PA/cm’. For different irradiation currents Z(OS--2 pA/cmz) we observed a linear relation between Dt and Z values. The theoretical profiles were obtained by numerical integration [20] of eq. (1) with boundary conditions (2) and (3), assuming that all the sodium atoms which arrive at the surface layer AI are sputtered away. We observed that the H dependence on the incident ion energy, as reported in ref. 16, was not evident in the case of alkali sputtering in glasses, probably due to the weaker bond energy of alkali elements. In figs. 3 and 4 we report the Na distribution (lines) evaluated following th previous calculation model. The calculated Dr and 0; values are reported in tables 2 and 3. From table 2 we can observe that the alkali depletion can be calculated by using coefficient DT values independent of the irradiation dose, as obtained in a quite different system, Pt-Si [16]. Moreover, we confirm the linear relation between 0: and Z values (compare the values in line 5 with all the others in table 2). 0: is a function of irradiation energy (see table 3). Such a result can be explained by considering the increase of the total number of vacancies, created by incident ions, as a function of irradiation energy. The large increase in the 0: and D,* values for the highest implantation energies can be correlated to indepth extended defect production which may provide paths for rapid diffusion [21]. Such a suggestion is Table 2 Df and Dz values used for the calculation of sodium profiles reported in fig. 3. I is the current density, 9 the Ar+ dose, and R the percentage of sodium loss. The irradiation energy was 100 keV. The D* discontinuity depth was xr, = 1750 A $I (Ar+/cm*) I (PA/cm’) Df (cm2/s) D; (cm*/s) R 7x10’5 10’6 2x10’6 4x10’6 10” 1 1 1 1 2 4x 10-l’ 4x10-14 4x10-14 4x10-14 8x lo-r4 lo-r4 lo-‘4 lo-l4 lo-‘4 2x10-l4 0.9 0.87 0.81 0.77 0.61 VIII. GLASSES 514 G. Battaglin et al. / Alkufi ~igrution in ion irradiated ghses the depletion depths for different alkali elements at low temperatures, where the vacancy migration enthalpy and the vacancy-interstitial separation, at which spondose lOI Ar+/cm’ taneous recombination occurs, control the D* value. For high irradiation flux, vacancy agglomeration D2f (cm’/s) Df (cm’/s) i (keV) x,(A) R processes near the surface are expected due to lower 0.89 1100 2x10-‘4 4x 10-15 50 inter-vacancy distances. Such a mechanism could de0.80 1100 4 x lo-“+ 4x10-‘5 60 termine partial trapping of certain solutes [21] and 0.74 1450 4 x 1 0 -‘4 4x 1 0 -1 5 80 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA formation of blistering phenomena [25,26]. 0.61 1750 8~10-‘~ 2 x 1 0 -1 4 100 Fig. 6 shows the Na profile after 100 keV-Ar” irradiation at a dose of 4 X 1016 ions/cm2 for different current densities, Such trapping phenomena do not avoid the depletion in depth. We note that the thickness of the supported by the experimental observation of a swelling depleted region is, in this case too, independent of the process for high irradiation energies (221. current density. The mechanism by which ion irradiation influences The interaction between surface defects and alkali atomic transport have been discussed, including enelements can be dependent on the alkali type. Indeed in hanced diffusion via mobile point defects by Myers [21]. figs. 2 and 5 we observed a different near surface The increased vacancy concentration due to the irradiadistribution for K, Rb and Na elements, even when the tion causes a proportional increase in the diffusion of thickness of the depleted region was the same. atoms at low temperatures by the vacancy mechanism. Li surface accumulation has been observed by Arnold In addition substitutional atoms are ejected into intersand Borders in Li,O-2Si0, glasses after 250 keV-Xe titial sites, from which they diffuse rapidly. An analytiirradiation 1271.Work is in progress in order to clarify cal expression for I)*has been introduced by Myers this trapping mechanism for which agglomerates appear [21] on the basis of the formalism of Diemes and to dissolve after an annealing process 12-71. Damask [23]. Three regions are present: We are grateful to Dr F. Nicoletti of the Stazione (a) at high temperature the normal thermal diffusion Sperimentale de1 Vetro of Murano - Venezia, who dominates: supplied the glasses containing K and Rb oxides; Mr E. (b) at intermediate temperatures, D*, independent of Bolzan for technical assistance during implantations; temperature, is proportional to the rate at which Mr A. Rampazzo of the Sezione INFN of Padova, who vacancy-interstitial pairs are produced, and consemade the drawings for the figures; and Dr M. Prosperi quently to the ion flux; who revised the English language of the manuscript. (c) at still lower temperatures, D* depends on the This work has been partially supported by Ministero vacancy migration coefficient. Pubblica Istruzione. zyxwvutsrqponmlkjihgfedcbaZYXWV Our results confirm the linear dependence of D* on the incident ion current density. Similar results were obtained by Myers and Picraux [24] by a study of the enhanced diffusion of 2% in Al under high-flux heavy References ion irradiation. We expected to observe a difference in Table 3 0: and 0; values used for the calculation of sodium profiles reported in fig. 4. The current density was 2 PA/cm’ and the DEPTH(nm) Fig 6. Experimental Na profile after 100 keV-Ar+ irradiation at a dose of 4 x 1On’ions/cm’ for different current densities. (11 G.W. Arnold, Rad. Elf. 65 (1982) 17. [2] G.W. Arnold and P.S. Peercy, J. Non-Crystalline Solids 41 (1980) 359. (31 V. Chinellato, V. Gottardi, S. Lo Russo, P. Mazaoldi, F. Nicoletti and P. Polato, Rad. Eff. 65 (1982) 31. [4] G. Battaglin, G. Della Mea, G. De Marchi, P. Mazzoldi, A. Miotello and M. Gugliehni, J. Phys. C: 15 (1982) 5623. [S] G. Battaglin, G. Della Mea, G. De Marchi, P. MaazoIdi, 0. Pughsi, Rad Eff. 64 (1982) 99. [6] G. Battaghn, G. Della Mea, G. De Marchi, P. Mazzoldi and 0. Puglisi, J. Non-C~st~ne Solids 50 (1982) 119. [7] P. Mazzoldi, Proc. Int. Conf. IBMM. Grenoble 1982, Nucl. Instr. and Meth. 209/210 (1983) 1089. [S] A. Miotello and P. Mazzoldi, J. Phys. C: 15 (1982) 5615. [9] A. Miotello and P. Mazzoldi, J. Phys. C: 16 (1983) 221. lo] G. Battaglin, G. Della Mea, G. De Marchi, P. Mauoldi, A. Miotello nd M. Guglielmi, J. de Phys. C9 (1982) 645. [II] R.L. Hines and R. Amdt, Phys. Rev. 119 (1960) 623. G. Battaglin et al. / Alkali migration in ion irradiated glasses [12] A. Camera, G. Della Mea, A.V. Drigo, S. Lo Russo and P. Mazzoldi, J. Non-Crystalline Solids 23 (1977) 123. 1131 L. Holland, Brit. J. Appi. Phys. 9 (1958) 410. [14] R. Behrisch, ed., Sputtering by particle bombardment I, topics in applied physics (Springer, Berlin, Heidelberg, New York, 1981). [15] Z.L. Liau, J.W. Mayer, W.L. Brown and J.M. Poate, J. Appl. Phys. 49 (1978) 5295. [16] A. Miotello and P. Mazzoldi, J. Appl. Phys. 54 (1983) 4235. [17] R.L. Hines, J. Appl. Phys. 28 (1957) 587. (181 R. Collins and G. Carter, Rad. Eff. 54 (1981) 235. [19] G. Della Mea, G. De Marchi, E. Grinzato, A. Mazzoldi, P. Mazzoldi and A. Miotello, J. Phys. C, in press. [20] H.S. Carslaw and J.C. Jaeger, Conduction of heat in solids (Clarendon, Oxford, 1959) Ch. 18. 515 [21] S.M. Myers, Nucl. Instr. and Meth. 168 (1980) 265. [22] F. Geotti-Bianchini, P. Polato, S. Lo Russo and P. Mazzoldi, Int. Conf. on Glasses - Hamburg (1983). [23] G.J. Diemes and A.C. Damask, J. Appl. Phys. 29 (1958) 1713. [24] S.M. Myers and S.T. Picraux, J. Appl. Phys. 46 (1975) 4774. [25] C. Wang, Y. Tao and S. Wang, J. Non-Crystalline solids 52 (1982) 589. [26] B. Rauschenbach and W. Hinz, Silikattechnik 27 (1976) 406. [27] G.W. Arnold and J.A. Borders, J. Appl. Phys. 48 (1977) 1448. VIII. GLASSES