- Email: [email protected]
Pd bilayers
Nuclear Instruments and Methods in Physics Research B52 (1990) 19-24 North-Holland
19
Ion beam mixing variance profiles of Al and Pd in Al/Pd bilayers J.H. Song, SO. Kim, K.H. Chae, J.J. Woo * and C.N. Whang Department of Physics, Yonsei University, Seoul 120-749, Korea
R.J. Smith Department of Physics, Montana State University, Bozeman, MT 59717, USA
Received 18 January 1990 and in revised form 6 July 1990
Evaporated Al/Pd bilayered thin films have been irradiated using 80 keV Ar+ with doses in the range of 1 x 10” to 1 X 1016 at room temperature. Auger electron spectroscopy depth profiles show that intermixing has occurred through the Al/Pd interface as a result of Arf bombardment. The mixing variances of Al and Pd are far from being constants, and the mixing efficiency of Al is higher than that of Pd. A strong dependence of the mixing variance on the nuclear energy deposition was observed. The experimental results are discussed in terms of the damage controlled atomic motion.
Ar+/cm’
1. Introduction Irradiation of a solid with energetic ions results not only in the production of vacancies and interstitial atoms, but also in the related phenomenon of ion beam mixing. This ion beam mixing process turns out to be a very suitable technique for producing either crystalline or amorphous surface alloys [l]. However the basic mechanisms of ion induced mixing are not yet very well understood; the rate of mixing of a binary system under heavy ion bombardment cannot be explained by only the cascade mixing mechanism. In some cases, thermal spike and/or radiation enhanced diffusion effects are believed to play an important role. Thus more than one mechanism may be operating in a particular system. The ion-induced mixing processes such as cascade, thermal spike and radiation enhanced diffusion have a very close relationship with the damage density produced by ion irradiation. The damage density profile due to ion bombardment in a binary system reveals [2] that the damage density in the heavy element side is higher than that in the light element side. This difference in the damage density between the light element and heavy element side might affect the atomic transport of the constituents in a binary system during ion beam mixing. Rauschenbach et al. [3] reported that in the Al/Fe bilayer system, the mixing efficiency of Al (light element) is significantly larger than that of Fe (heavy element) because of a strong dependence of * Permanent address: Department of Physics, Cheonnam National University, Kwangju 500-757, Korea.
diffusion coefficients on the damage profiles. Tao et al. [4] noted that in met&Si systems the atoms from the low damage density region (Si layer) move preferentially into the high damage density region (the metal layer), which is called the damage controlled atomic motion [4], and the light element (Si atom) plays the role of moving species during ion induced mixing due to this damage controlled atomic motion. The rate of intermixing at the interface in bilayered structure is represented by the mixing variance defined as a2 = 2Dt, where D is the diffusion coefficient and t is the diffusion time, respectively. In this study, we have investigated the depth profiles of mixing variance of Al and Pd atoms in the Al/Pd bilayered system using the Auger electron spectroscopy (AES) depth profiling technique. The experimental results are interpreted in terms of damage controlled atomic motion associated with the nuclear energy deposition profiles.
2. Experimental procedures A cleaned glass slide was placed in a specially designed chamber for in situ evaporation and ion beam mixing. A 70 nm thick Al layer (top layer) and a 20 nm thick Pd layer (bottom layer) were deposited on the glass at a rate of 0.1 rim/s and then bombarded with 80 keV Ar+ ions to cause mixing which was carried out at a base pressure of 2 x lo-’ Torr at room temperature. To avoid sample heating due to ion bombardment, the substrate was glued to a large copper block, ion current was maintained at an average current density of 0.5
0168-583X/90/$03.50 0 1990 - Elsevier Science Publishers B.V. (North-Holland)
20
J. H. Song et al. / Ion beam mixing profiles in AI/ Pd bilayers
pA/cm2, and the ion dose ranged between 1 X 10” and 1 x 1016 Arf/cm2. The Al film thickness of 70 nm was chosen to match with the mean projected range of the 80 keV Ar+ in the Al layer (69 f 21 nm). The mixing variance of Al and Pd were investigated using the AES depth profiling technique. The scanning Auger microprobe in the Montana State University Advanced Materials Center was operated with a primary electron beam of 3.0 keV. AES depth profiles were obtained using 1 keV Ar+ ion sputter etching with simultaneous monitoring of the Al LW (67 eV) and Pd MNN (330 eV) Auger transition lines. Intensities of the Auger peaks were determined as the peak-to-peak heights in the derivative spectra (dN(E)/dE mode). Prior the AES investigation, the reproducibility of the interfacial broadening due to Ar+ bombardment was checked using 2.5 MeV He’+ backscattering with 4 to 5 samples at each ion dose. The reproducibility was satisfactory within 10% error. In this case, the sample was tilted to an angle of 60° from the incident He’+ beam direction to give an enhancement of the depth resolution.
3. Results and discussion Fig. 1 shows typical AES depth profiles of Al (a) and Pd (b) after mixing using 1 X 1015, 5 X 1015 and 1 X 1016 Ar’/cm’ along with that of the as-deposited sample. The AES depth profile of the as-deposited sample exhibits a rather sharp transition region at the interface. With increasing Ar+ ion dose, the profiles tend to become broader across the interface region and the tendency toward a smearing out of the original concentration distribution at the interface is increased. This result clearly indicates that intermixing has occurred through the Al/Pd interface as a result of Ar+ bombardment. Also the peak position of the Pd signal
moves progressively towards the surface side with increasing ion dose because of the sputtering removal of the Al atoms during ion beam mixing. The amount of Al removed at a dose of 1 X 1016 Ar+/cm2 is 10 nm, as determined from the Rutherford backscattering (RBS) area analysis. The AES depth profiling method has the advantage that in the case of a constant sputtering rate, excellent depth resolution can be obtained, limited only by the escape depth of the Auger electrons (I 1 nm). However a severe problem arises in determining the sputtered depth since the sputtering rate is generally matrix dependent. In this case, a depth scale with high accuracy can be obtained by independently measuring the thicknesses of the evaporated Al and Pd layers using RBS spectra for the as-deposited sample [5], obtained while tilting the sample normal to an angle of 60 o or 78O from the incident He+ beam. The RUMP program [6] for the backscattering spectra is employed to evaluate the thicknesses of the evaporated layers. By assuming that the AES intensity distribution at the interface in the as-deposited sample is a steplike function, the sputtering time corresponding to the evaporated film thickness can be defined here as the full width half miximum (FWHM) of the AES signal. Comparing the RBS spectra with the AES depth profile for the as-deposited sample, the ion etching rates for the Al and Pd layers for 1 keV Ar+ during AES depth profiling are found to be 5.15 nm/min and 5.02 nm/min, respectively. In the case of the ion beam mixing experiment, Fick’s law is generally used under the assumption that the diffusion coefficient, D, is a constant, i.e., independent of atomic concentration, which results in a Gaussian error function type of depth profile in the interface region. This assumption is never strictly true. It can be clearly seen in fig. 1 that the shape of the AES depth profile for Al in the interface region at a dose of
(b)
SPUTTERING
TIME
(min.)
SPUTTERING
TIME
( min. 1
and for ion beam Fig. 1. AES depth profiles of Al (a) and Pd (b) in Al/Pd bilayered films for the as-depositedsample ( -), mixed sampleswith a dose of 1 x 1015(. . . . . .), 5 x 1Ol5(- - -), and 1 X 1Ol6(. -. - .) Ar+/cm2.
J.H. Song et al. / Ion beam mixing profiles in Al/ Pd bilayers
1 x lOi Ar’/cm’ deviates significantly from the Gaussian error function. So, we evaluated the variances, a2 = 2Dt, of Al and Pd as a function of depth from the AES depth profiles using the Boltzmann-Matano method [7]. In a real experiment, the diffusion coefficients can vary with concentration gradient. In this case, Fick’s second law must be written as
where c is the atomic concentration and x is the depth position in the sample, x = 0 at the interface. We assume that the initial conditions for c(x, t) are c(x < 0, 0) = 1 and c(x > 0,O) = 0 for the Al layer, and vice versa for the Pd layer. For these initial conditions, the partial differential eq. (1) can be transformed into an ordinary homogeneous differential equation for c(y), using the substitution y = x/h. Using this definition of y and eq. (l), we obtain the relation 02=2D(c’)t=
-(g)=,
/6,x&,
where c’ is any concentration (0 < c’ < 1). Then the variance for a given concentration c’ can be calculated from the slope of dx/dc at c = c’ and the area /,“xdc. The true mixing variance, CT&, is expressed as u,$, = u2(+) - u2(0), where u2(+) and u*(O) are the variances for a dose of I$ and for the as-deposited state, respecdepth profile tively. To calculate u,&, the measured data were fitted to a smooth cubic spline function, and then the differentiation and integration performed. Fig. 2(a) shows the calculated mixing variances of Al and Pd for the sample mixed with the dose of 1 x 1016 Ar+/cm2. These results show clearly that the mixing variances are far from being constant. The maxima of the mixing variances for Al and Pd are found near the Al/Pd interface, and that for Al is located behind the interface while that for Pd is located in front of the interface. And the maxima value of u& for Al is significantly higher than that for Pd by a factor of 3. This phenomenon of Al diffusing faster than Pd might be correlated with collisional events such as cascades, thermal spike and/or radiation enhanced diffusion effects, and presumably different atomic mixing mechanisms between the light element (Al) and heavy element (Pd). The ion induced mixing process is commonly divided into three phases [8,9], that is, the cascade phase, the thermal spike phase, and the radiation enhanced diffusion phase. All the phases are very closely related [lo] to the nuclear energy deposition distribution or the damage density profile produced by ion irradiation. Guinan and Kinney [ll] have observed, using molecular dynamics computer simulations for high energy cascades, that almost all defects are produced in the cascade phase,
21
and a large number of defects annihilate during the thermal spike phase to result in atomic transport that is more than an order of magnitude greater than that predicted by the equilibrium transport theory. Biersack [12] has demonstrated theoretically that in a collision cascade the lighter target atoms always have a larger displacement as compared to that of the heavier atoms in a binary system. We suggest that these observations might be the theoretical support for our experimental observations. Fig. 2(b) shows the nuclear energy deposition (S,) profile for 80 keV Ar+ ions in an Al(60 nm)/Pd(20 nm)/SiO, system obtained using the TRIM MonteCarlo simulation program [13] with 1000 incident Ar+ ions. The mean projected range (RP = 65 + 18 nnm) is marked with an arrow. The thickness of 60 nm for the Al layer was chosen taking into consideration the sputtered Al thickness of 10 nm at the dose of 1 X lO“j Ar’/cm’. The S, profile is very similar to the depth dependence of the mixing variences; near the interface, the energy deposition is highest and S,, on the Pd side is higher than S, on the Al side, just as the maxima of the mixing variances are near the interface and the mixing variance of Al is higher than that of Pd. This result reveals that the phenomenon of Al diffusing faster than Pd and presumably the related mixing mechanisms depend strongly on the nuclear energy deposition profile. Therefore we suggest that the higher energy deposition on the Pd side induces a pronounced vacancy concentration, which causes a pronounced increase in diffusion of Al by radiation enhanced diffusion via vacancy migration and/or thermal spike induced mixing. Moreover Al has the lowest vacancy migration energy among metals [14] and high solubility in Pd (about 18 at. 5%[15]) compared to the negligible value of Pd and Al and a Al/Pd binary system has a relatively large negative heat of mixing energy (-104 k_I/g at. [15]). As a consequence of these physical and chemical characterisitics, radiation enhanced diffusion and/or thermal spike induced mixing would be more pronounced for Al than for Pd. In order to clarify the relation between the mixing variance and the nuclear energy deposition in more detail, we calculated the mixing variances as a function of depth at various ion doses. Fig. 3 shows the mixing variances of Al (a) and Pd (b) as a function of depth at doses of 1 x 1015, 5 x 1015 and 1 x 1Ol6 Ar’/cm’. The mixing variance peak for Al is higher than that of Pd. The approximate interface positions are marked with arrows. The interface position moves progressively towards the surface side with increasing ion dose due to sputtering of the Al layer during ion beam mixing. The remaining thicknesses of the Al layer after sputtering are found to be 60 nm and 67 nm from the RBS spectra at the dose of 1 X 1016 and 5 x 1015 Ar+/cm2, respectively. The mixing variances at the dose of 1 X 1015
22
J. H. Song et al. / Ion beam mixing profiles in AI/ Pd bilayers 15,
I
I (a)
3 ‘;
lo-
; : B s 4 R s
5-
o5
I 10
I 15
SPUTTERING
2500 I
Al
TIME (min.)
-
50
65 DEPTH
( nm )
Fig. 2. The calculated mixing variances of Al () and Pd (- - -) at the dose of 1 X 1016Ar+/cm* (a) along with the depth profile of nuclear energy deposition yield by 80 keV Ar+ ions in the Al(60 nm)/Pd(ZO nm)/SiO, system using the TRIM CODE (b) for comparison. The arrow in (a) represents the approximate interface position and the arrow in (b) represents the mean projected range of 80 keV Ar+.
Ar’/cm’ have constant values in the interface region, while those at higher dose show sharp peak shapes near the interface. The difference between the maximum variances of Al and Pd at each ion dose increases with increasing ion dose. These phenomena may also relate closely to the variation in the nuclear energy deposition profile due to sputtering of the Al layer during ion beam mixing. Fig. 4 shows S,, profiles for Al(70 nm)/Pd(20 nm), Al(65 nm)/Pd(20 nm), and Al(60 mn)/Pd(20 nm). The figure illustrates the effect of the Al layer thickness on
the energy deposition profiles. The changes in the S,, value at the interface and the maximum value of S,, in the Pd side as the thickness of the Al layer is reduced show the same trend as the change in difference in mixing variance between Al and Pd and the change in mixing variance for Al with increasing ion dose as shown in fig. 3. In fig. l(a), the thickness of the Al layer at the dose of the 1 x 10” Ar’/cm’ is slightly thicker than that of the as-deposited one, the thickness of Al layer is found to be approximately 75 nm from the RBS spectrum. For a thicker Al layer of 80 nm, the dif-
J.H. Song ei al. / ion beam mixing profiles in Al/ Pd bilavers 57
‘o-
5-
06 0
5
15
10
SPUTTERING
LU SPTTERING
TIME (min.)
TIMEfmin.)
), 5X10’s (---) Fig. 3. The mixing variances of Al (a) and Pd (b) as a function of the sputter time at doses of 1 x 10” (and 1 x 1Or6 (. - .- .) to clarify the relation with the nuclear energy deposition. The arrows indicate the approximate interface positions.
ference in S,, at the interface is found to be very small or hardly verifiable, which may be the reason for the constant values of the mixing variance for Al and Pd in the interface region at the dose of 1 x lo’* Ar’/cm’. For better characterization of the mixing processes, we have calculated the average mixing variances over the whole depth. The average mixing variance is used for calculation of the mixing efficiency, defined as n = 4Dt/@ or ~&,/c#J. Fig. 5 shows the average mixing variances for Al and Pd at various ion doses; appearently the mixing variances increase monoto~cally with ion dose. The mixing efficiences of Al and Pd can
0
150
DEPTH
5
fnm)
Fig. 4. The depth profiles of .S, for Al{70 nm)/Pd(20 nm) (a), Al(65 nm)/Pd(M nm) (b) and Al(60 nm)/Pd(20 nm) (c) yield by 80 keV Ar+ ions using by TRIM CODE.
Af
ION DOSE ( X*0”
10
d
f
Fig. 5. The calculated average values of mixing variances for Al and Pd (- - -) at various ion doses. ( -)
24
J.H. Song et al. / Ion beam mixing profiles in Al/ Pd bilayers
be evaluated from the slopes of the linear fit. The mixing efficiences of Al and Pd are found to be 18 run4 and 6 mn4, respectively. The value of n of Al is higher than that of Pd by a factor of 3. This result shows also that Al atom diffuses faster than Pd due to the damage controlled atomic motion.
Acknowledgement This study was supported in part by the Korea Science and Engineering Foundation (KOSEF). The authors are grateful to Dr. R.Y. Lee of Danguk University of his useful comments and discussions. References
4. Conclusions Ar+ ion induced mixing in the Al(70 nm)/Pd(20 nm) bilayered system has been studied using the AES depth profiling technique over an ion dose range between 1 x 1015 and 1 x 1016 Ar’/cm’ at room temperature. The AES intensity profile at the Al/Pd interface region at higher ion dose deviated significantly from the Gaussian error function. Therefore the muting variances of Al and Pd are evaluated as a function of depth by the Boltzmamr-Matano method. The mixing variances are found to be far from being constants and show maxima near the Al/Pd interface, and the maximum value of the mixing variance for Al is significantly higher than that of Pd. The mixing efficiencies of Al and Pd are found to be 18 nm4 and 6 nm4, respectively. The profile of the mixing variance is very similar to the nuclear energy deposition profile, which suggests strongly that the phenomenon of Al diffusing faster than Pd relates closely to the nuclear energy deposition and its related mixing mechanism; that is, radiation enhanced diffusion and/or thermal spike induced mixing would be more pronounced for Al than for Pd as a consequence of the physical and chemical properties of constituents such as the nuclear energy deposition, vacancy migration energy, solid solubility, and heat of mixing energy.
[l] B.X. Liu, Phys. Status Solidi (a) 94 (1986) 11. [2] W. Xia, M. Femandes, C.A. Hewett, S.S. Lau, D.B. Poker, and J.P. Biersack, Nucl. Instr. and Meth. B37/38 (1989) 408. [3] B. Rauschenbach, M. Possert, and R. KuchIer, Appl. Phys. A48 (1989) 347. [4] K. Tao, C.A. Hewett, S.S. Lau, Ch. BuchaI, and D.B. Poker, Appl. Phys. Lett. 50 (1987) 1343. [5] R.Y. Lee, C.N. Whang, T.K. Kim, S.O. Kim, and R.J. Smith, Nucl. Instr. and Meth. B39 (1989) 114. [6] L.R. Doolittle, Nucl. Instr. and Meth. B15 (1986) 227. [7] P.G. Shewon, in Diffusion in Solids (McGraw-Hill, New York, 1963), chap. 1. [8] D. Peak and R.S. Aveback, Nucl. Instr. and Meth. B7/8 (1985) 561. [9] B.M. Paine and R.S. Averback, Nucl. Instr. and Meth. B7/8 (1985) 666. [lo] S.M. Myers, Nucl. Instr. and Meth. 168 (1980) 265. [ll] M.W. Guinan and J.H. Kinney, J. Nucl. Mater. 103/104 (1981) 1319. [12] J.P. Biersack, Nucl. Instr. and Meth. B19/20 (1987) 32. [13] J.F. Ziegler, J.P. Biersack, and U. Littmark, in: The Stopping and Range of Ions in Solids (Pergamon, New York, 1985). [14] A.D. Marwich, in: Surface Modification and Alloying, eds. J.M. Poate, G. Foti and D.C. Jacobson (Plenum Press, New York, 1983) p. 199. [15] R. Hultgren, in: Selected Values of the Thermodynamic Properties of Binary Alloys (American Society for Metals, Ohio, 1973) p. 199.
19
Ion beam mixing variance profiles of Al and Pd in Al/Pd bilayers J.H. Song, SO. Kim, K.H. Chae, J.J. Woo * and C.N. Whang Department of Physics, Yonsei University, Seoul 120-749, Korea
R.J. Smith Department of Physics, Montana State University, Bozeman, MT 59717, USA
Received 18 January 1990 and in revised form 6 July 1990
Evaporated Al/Pd bilayered thin films have been irradiated using 80 keV Ar+ with doses in the range of 1 x 10” to 1 X 1016 at room temperature. Auger electron spectroscopy depth profiles show that intermixing has occurred through the Al/Pd interface as a result of Arf bombardment. The mixing variances of Al and Pd are far from being constants, and the mixing efficiency of Al is higher than that of Pd. A strong dependence of the mixing variance on the nuclear energy deposition was observed. The experimental results are discussed in terms of the damage controlled atomic motion.
Ar+/cm’
1. Introduction Irradiation of a solid with energetic ions results not only in the production of vacancies and interstitial atoms, but also in the related phenomenon of ion beam mixing. This ion beam mixing process turns out to be a very suitable technique for producing either crystalline or amorphous surface alloys [l]. However the basic mechanisms of ion induced mixing are not yet very well understood; the rate of mixing of a binary system under heavy ion bombardment cannot be explained by only the cascade mixing mechanism. In some cases, thermal spike and/or radiation enhanced diffusion effects are believed to play an important role. Thus more than one mechanism may be operating in a particular system. The ion-induced mixing processes such as cascade, thermal spike and radiation enhanced diffusion have a very close relationship with the damage density produced by ion irradiation. The damage density profile due to ion bombardment in a binary system reveals [2] that the damage density in the heavy element side is higher than that in the light element side. This difference in the damage density between the light element and heavy element side might affect the atomic transport of the constituents in a binary system during ion beam mixing. Rauschenbach et al. [3] reported that in the Al/Fe bilayer system, the mixing efficiency of Al (light element) is significantly larger than that of Fe (heavy element) because of a strong dependence of * Permanent address: Department of Physics, Cheonnam National University, Kwangju 500-757, Korea.
diffusion coefficients on the damage profiles. Tao et al. [4] noted that in met&Si systems the atoms from the low damage density region (Si layer) move preferentially into the high damage density region (the metal layer), which is called the damage controlled atomic motion [4], and the light element (Si atom) plays the role of moving species during ion induced mixing due to this damage controlled atomic motion. The rate of intermixing at the interface in bilayered structure is represented by the mixing variance defined as a2 = 2Dt, where D is the diffusion coefficient and t is the diffusion time, respectively. In this study, we have investigated the depth profiles of mixing variance of Al and Pd atoms in the Al/Pd bilayered system using the Auger electron spectroscopy (AES) depth profiling technique. The experimental results are interpreted in terms of damage controlled atomic motion associated with the nuclear energy deposition profiles.
2. Experimental procedures A cleaned glass slide was placed in a specially designed chamber for in situ evaporation and ion beam mixing. A 70 nm thick Al layer (top layer) and a 20 nm thick Pd layer (bottom layer) were deposited on the glass at a rate of 0.1 rim/s and then bombarded with 80 keV Ar+ ions to cause mixing which was carried out at a base pressure of 2 x lo-’ Torr at room temperature. To avoid sample heating due to ion bombardment, the substrate was glued to a large copper block, ion current was maintained at an average current density of 0.5
0168-583X/90/$03.50 0 1990 - Elsevier Science Publishers B.V. (North-Holland)
20
J. H. Song et al. / Ion beam mixing profiles in AI/ Pd bilayers
pA/cm2, and the ion dose ranged between 1 X 10” and 1 x 1016 Arf/cm2. The Al film thickness of 70 nm was chosen to match with the mean projected range of the 80 keV Ar+ in the Al layer (69 f 21 nm). The mixing variance of Al and Pd were investigated using the AES depth profiling technique. The scanning Auger microprobe in the Montana State University Advanced Materials Center was operated with a primary electron beam of 3.0 keV. AES depth profiles were obtained using 1 keV Ar+ ion sputter etching with simultaneous monitoring of the Al LW (67 eV) and Pd MNN (330 eV) Auger transition lines. Intensities of the Auger peaks were determined as the peak-to-peak heights in the derivative spectra (dN(E)/dE mode). Prior the AES investigation, the reproducibility of the interfacial broadening due to Ar+ bombardment was checked using 2.5 MeV He’+ backscattering with 4 to 5 samples at each ion dose. The reproducibility was satisfactory within 10% error. In this case, the sample was tilted to an angle of 60° from the incident He’+ beam direction to give an enhancement of the depth resolution.
3. Results and discussion Fig. 1 shows typical AES depth profiles of Al (a) and Pd (b) after mixing using 1 X 1015, 5 X 1015 and 1 X 1016 Ar’/cm’ along with that of the as-deposited sample. The AES depth profile of the as-deposited sample exhibits a rather sharp transition region at the interface. With increasing Ar+ ion dose, the profiles tend to become broader across the interface region and the tendency toward a smearing out of the original concentration distribution at the interface is increased. This result clearly indicates that intermixing has occurred through the Al/Pd interface as a result of Ar+ bombardment. Also the peak position of the Pd signal
moves progressively towards the surface side with increasing ion dose because of the sputtering removal of the Al atoms during ion beam mixing. The amount of Al removed at a dose of 1 X 1016 Ar+/cm2 is 10 nm, as determined from the Rutherford backscattering (RBS) area analysis. The AES depth profiling method has the advantage that in the case of a constant sputtering rate, excellent depth resolution can be obtained, limited only by the escape depth of the Auger electrons (I 1 nm). However a severe problem arises in determining the sputtered depth since the sputtering rate is generally matrix dependent. In this case, a depth scale with high accuracy can be obtained by independently measuring the thicknesses of the evaporated Al and Pd layers using RBS spectra for the as-deposited sample [5], obtained while tilting the sample normal to an angle of 60 o or 78O from the incident He+ beam. The RUMP program [6] for the backscattering spectra is employed to evaluate the thicknesses of the evaporated layers. By assuming that the AES intensity distribution at the interface in the as-deposited sample is a steplike function, the sputtering time corresponding to the evaporated film thickness can be defined here as the full width half miximum (FWHM) of the AES signal. Comparing the RBS spectra with the AES depth profile for the as-deposited sample, the ion etching rates for the Al and Pd layers for 1 keV Ar+ during AES depth profiling are found to be 5.15 nm/min and 5.02 nm/min, respectively. In the case of the ion beam mixing experiment, Fick’s law is generally used under the assumption that the diffusion coefficient, D, is a constant, i.e., independent of atomic concentration, which results in a Gaussian error function type of depth profile in the interface region. This assumption is never strictly true. It can be clearly seen in fig. 1 that the shape of the AES depth profile for Al in the interface region at a dose of
(b)
SPUTTERING
TIME
(min.)
SPUTTERING
TIME
( min. 1
and for ion beam Fig. 1. AES depth profiles of Al (a) and Pd (b) in Al/Pd bilayered films for the as-depositedsample ( -), mixed sampleswith a dose of 1 x 1015(. . . . . .), 5 x 1Ol5(- - -), and 1 X 1Ol6(. -. - .) Ar+/cm2.
J.H. Song et al. / Ion beam mixing profiles in Al/ Pd bilayers
1 x lOi Ar’/cm’ deviates significantly from the Gaussian error function. So, we evaluated the variances, a2 = 2Dt, of Al and Pd as a function of depth from the AES depth profiles using the Boltzmann-Matano method [7]. In a real experiment, the diffusion coefficients can vary with concentration gradient. In this case, Fick’s second law must be written as
where c is the atomic concentration and x is the depth position in the sample, x = 0 at the interface. We assume that the initial conditions for c(x, t) are c(x < 0, 0) = 1 and c(x > 0,O) = 0 for the Al layer, and vice versa for the Pd layer. For these initial conditions, the partial differential eq. (1) can be transformed into an ordinary homogeneous differential equation for c(y), using the substitution y = x/h. Using this definition of y and eq. (l), we obtain the relation 02=2D(c’)t=
-(g)=,
/6,x&,
where c’ is any concentration (0 < c’ < 1). Then the variance for a given concentration c’ can be calculated from the slope of dx/dc at c = c’ and the area /,“xdc. The true mixing variance, CT&, is expressed as u,$, = u2(+) - u2(0), where u2(+) and u*(O) are the variances for a dose of I$ and for the as-deposited state, respecdepth profile tively. To calculate u,&, the measured data were fitted to a smooth cubic spline function, and then the differentiation and integration performed. Fig. 2(a) shows the calculated mixing variances of Al and Pd for the sample mixed with the dose of 1 x 1016 Ar+/cm2. These results show clearly that the mixing variances are far from being constant. The maxima of the mixing variances for Al and Pd are found near the Al/Pd interface, and that for Al is located behind the interface while that for Pd is located in front of the interface. And the maxima value of u& for Al is significantly higher than that for Pd by a factor of 3. This phenomenon of Al diffusing faster than Pd might be correlated with collisional events such as cascades, thermal spike and/or radiation enhanced diffusion effects, and presumably different atomic mixing mechanisms between the light element (Al) and heavy element (Pd). The ion induced mixing process is commonly divided into three phases [8,9], that is, the cascade phase, the thermal spike phase, and the radiation enhanced diffusion phase. All the phases are very closely related [lo] to the nuclear energy deposition distribution or the damage density profile produced by ion irradiation. Guinan and Kinney [ll] have observed, using molecular dynamics computer simulations for high energy cascades, that almost all defects are produced in the cascade phase,
21
and a large number of defects annihilate during the thermal spike phase to result in atomic transport that is more than an order of magnitude greater than that predicted by the equilibrium transport theory. Biersack [12] has demonstrated theoretically that in a collision cascade the lighter target atoms always have a larger displacement as compared to that of the heavier atoms in a binary system. We suggest that these observations might be the theoretical support for our experimental observations. Fig. 2(b) shows the nuclear energy deposition (S,) profile for 80 keV Ar+ ions in an Al(60 nm)/Pd(20 nm)/SiO, system obtained using the TRIM MonteCarlo simulation program [13] with 1000 incident Ar+ ions. The mean projected range (RP = 65 + 18 nnm) is marked with an arrow. The thickness of 60 nm for the Al layer was chosen taking into consideration the sputtered Al thickness of 10 nm at the dose of 1 X lO“j Ar’/cm’. The S, profile is very similar to the depth dependence of the mixing variences; near the interface, the energy deposition is highest and S,, on the Pd side is higher than S, on the Al side, just as the maxima of the mixing variances are near the interface and the mixing variance of Al is higher than that of Pd. This result reveals that the phenomenon of Al diffusing faster than Pd and presumably the related mixing mechanisms depend strongly on the nuclear energy deposition profile. Therefore we suggest that the higher energy deposition on the Pd side induces a pronounced vacancy concentration, which causes a pronounced increase in diffusion of Al by radiation enhanced diffusion via vacancy migration and/or thermal spike induced mixing. Moreover Al has the lowest vacancy migration energy among metals [14] and high solubility in Pd (about 18 at. 5%[15]) compared to the negligible value of Pd and Al and a Al/Pd binary system has a relatively large negative heat of mixing energy (-104 k_I/g at. [15]). As a consequence of these physical and chemical characterisitics, radiation enhanced diffusion and/or thermal spike induced mixing would be more pronounced for Al than for Pd. In order to clarify the relation between the mixing variance and the nuclear energy deposition in more detail, we calculated the mixing variances as a function of depth at various ion doses. Fig. 3 shows the mixing variances of Al (a) and Pd (b) as a function of depth at doses of 1 x 1015, 5 x 1015 and 1 x 1Ol6 Ar’/cm’. The mixing variance peak for Al is higher than that of Pd. The approximate interface positions are marked with arrows. The interface position moves progressively towards the surface side with increasing ion dose due to sputtering of the Al layer during ion beam mixing. The remaining thicknesses of the Al layer after sputtering are found to be 60 nm and 67 nm from the RBS spectra at the dose of 1 X 1016 and 5 x 1015 Ar+/cm2, respectively. The mixing variances at the dose of 1 X 1015
22
J. H. Song et al. / Ion beam mixing profiles in AI/ Pd bilayers 15,
I
I (a)
3 ‘;
lo-
; : B s 4 R s
5-
o5
I 10
I 15
SPUTTERING
2500 I
Al
TIME (min.)
-
50
65 DEPTH
( nm )
Fig. 2. The calculated mixing variances of Al () and Pd (- - -) at the dose of 1 X 1016Ar+/cm* (a) along with the depth profile of nuclear energy deposition yield by 80 keV Ar+ ions in the Al(60 nm)/Pd(ZO nm)/SiO, system using the TRIM CODE (b) for comparison. The arrow in (a) represents the approximate interface position and the arrow in (b) represents the mean projected range of 80 keV Ar+.
Ar’/cm’ have constant values in the interface region, while those at higher dose show sharp peak shapes near the interface. The difference between the maximum variances of Al and Pd at each ion dose increases with increasing ion dose. These phenomena may also relate closely to the variation in the nuclear energy deposition profile due to sputtering of the Al layer during ion beam mixing. Fig. 4 shows S,, profiles for Al(70 nm)/Pd(20 nm), Al(65 nm)/Pd(20 nm), and Al(60 mn)/Pd(20 nm). The figure illustrates the effect of the Al layer thickness on
the energy deposition profiles. The changes in the S,, value at the interface and the maximum value of S,, in the Pd side as the thickness of the Al layer is reduced show the same trend as the change in difference in mixing variance between Al and Pd and the change in mixing variance for Al with increasing ion dose as shown in fig. 3. In fig. l(a), the thickness of the Al layer at the dose of the 1 x 10” Ar’/cm’ is slightly thicker than that of the as-deposited one, the thickness of Al layer is found to be approximately 75 nm from the RBS spectrum. For a thicker Al layer of 80 nm, the dif-
J.H. Song ei al. / ion beam mixing profiles in Al/ Pd bilavers 57
‘o-
5-
06 0
5
15
10
SPUTTERING
LU SPTTERING
TIME (min.)
TIMEfmin.)
), 5X10’s (---) Fig. 3. The mixing variances of Al (a) and Pd (b) as a function of the sputter time at doses of 1 x 10” (and 1 x 1Or6 (. - .- .) to clarify the relation with the nuclear energy deposition. The arrows indicate the approximate interface positions.
ference in S,, at the interface is found to be very small or hardly verifiable, which may be the reason for the constant values of the mixing variance for Al and Pd in the interface region at the dose of 1 x lo’* Ar’/cm’. For better characterization of the mixing processes, we have calculated the average mixing variances over the whole depth. The average mixing variance is used for calculation of the mixing efficiency, defined as n = 4Dt/@ or ~&,/c#J. Fig. 5 shows the average mixing variances for Al and Pd at various ion doses; appearently the mixing variances increase monoto~cally with ion dose. The mixing efficiences of Al and Pd can
0
150
DEPTH
5
fnm)
Fig. 4. The depth profiles of .S, for Al{70 nm)/Pd(20 nm) (a), Al(65 nm)/Pd(M nm) (b) and Al(60 nm)/Pd(20 nm) (c) yield by 80 keV Ar+ ions using by TRIM CODE.
Af
ION DOSE ( X*0”
10
d
f
Fig. 5. The calculated average values of mixing variances for Al and Pd (- - -) at various ion doses. ( -)
24
J.H. Song et al. / Ion beam mixing profiles in Al/ Pd bilayers
be evaluated from the slopes of the linear fit. The mixing efficiences of Al and Pd are found to be 18 run4 and 6 mn4, respectively. The value of n of Al is higher than that of Pd by a factor of 3. This result shows also that Al atom diffuses faster than Pd due to the damage controlled atomic motion.
Acknowledgement This study was supported in part by the Korea Science and Engineering Foundation (KOSEF). The authors are grateful to Dr. R.Y. Lee of Danguk University of his useful comments and discussions. References
4. Conclusions Ar+ ion induced mixing in the Al(70 nm)/Pd(20 nm) bilayered system has been studied using the AES depth profiling technique over an ion dose range between 1 x 1015 and 1 x 1016 Ar’/cm’ at room temperature. The AES intensity profile at the Al/Pd interface region at higher ion dose deviated significantly from the Gaussian error function. Therefore the muting variances of Al and Pd are evaluated as a function of depth by the Boltzmamr-Matano method. The mixing variances are found to be far from being constants and show maxima near the Al/Pd interface, and the maximum value of the mixing variance for Al is significantly higher than that of Pd. The mixing efficiencies of Al and Pd are found to be 18 nm4 and 6 nm4, respectively. The profile of the mixing variance is very similar to the nuclear energy deposition profile, which suggests strongly that the phenomenon of Al diffusing faster than Pd relates closely to the nuclear energy deposition and its related mixing mechanism; that is, radiation enhanced diffusion and/or thermal spike induced mixing would be more pronounced for Al than for Pd as a consequence of the physical and chemical properties of constituents such as the nuclear energy deposition, vacancy migration energy, solid solubility, and heat of mixing energy.
[l] B.X. Liu, Phys. Status Solidi (a) 94 (1986) 11. [2] W. Xia, M. Femandes, C.A. Hewett, S.S. Lau, D.B. Poker, and J.P. Biersack, Nucl. Instr. and Meth. B37/38 (1989) 408. [3] B. Rauschenbach, M. Possert, and R. KuchIer, Appl. Phys. A48 (1989) 347. [4] K. Tao, C.A. Hewett, S.S. Lau, Ch. BuchaI, and D.B. Poker, Appl. Phys. Lett. 50 (1987) 1343. [5] R.Y. Lee, C.N. Whang, T.K. Kim, S.O. Kim, and R.J. Smith, Nucl. Instr. and Meth. B39 (1989) 114. [6] L.R. Doolittle, Nucl. Instr. and Meth. B15 (1986) 227. [7] P.G. Shewon, in Diffusion in Solids (McGraw-Hill, New York, 1963), chap. 1. [8] D. Peak and R.S. Aveback, Nucl. Instr. and Meth. B7/8 (1985) 561. [9] B.M. Paine and R.S. Averback, Nucl. Instr. and Meth. B7/8 (1985) 666. [lo] S.M. Myers, Nucl. Instr. and Meth. 168 (1980) 265. [ll] M.W. Guinan and J.H. Kinney, J. Nucl. Mater. 103/104 (1981) 1319. [12] J.P. Biersack, Nucl. Instr. and Meth. B19/20 (1987) 32. [13] J.F. Ziegler, J.P. Biersack, and U. Littmark, in: The Stopping and Range of Ions in Solids (Pergamon, New York, 1985). [14] A.D. Marwich, in: Surface Modification and Alloying, eds. J.M. Poate, G. Foti and D.C. Jacobson (Plenum Press, New York, 1983) p. 199. [15] R. Hultgren, in: Selected Values of the Thermodynamic Properties of Binary Alloys (American Society for Metals, Ohio, 1973) p. 199.