- Email: [email protected]
Surface composition changes of Ni-Zr alloy under Ar+ ion bombardment by ISS, AES and XPS
....
““:‘:
v.... :.::::~::::~:~~:~.~~,::~:~.:::::~
‘.‘.‘. ...n
....
......
5;,:,,., ,_,,.(,_,, _.,, ...,..l(_ ,,:;, ,;:,,,, ,::::,
.,..,,
_,,,
‘.‘.‘.“” ” ‘... ...a. .:.: : : .. .,.....,.,.,:.,., :::::?.‘.‘::::::,:,~.~:~,:..::_: ~.~.:.:.::.‘,~.~::::.:.:.:;.‘. ” ‘.‘..... ‘V. . .
::::~>$;:::::.....’
.(.
..A
.,.,
px
,,,,
,,,,,(,,,,
‘.‘...,~.~
surface science
EL-SEWER
Surface
Science 302 (1994) 363-370
Surface composition changes of Ni-Zr alloy under Ar+ ion bombardment by ISS, AES and XPS H.J. Kang ‘, R.H. Roberts, S. He, D.J. O’Connor
*, R.J. MacDonald
Department of Physics, University of Newcastle, Callaghan, NSW 2308, Australia (Received
8 January
1993; accepted
for publication
20 September
1993)
Abstract The changes induced in the surface composition of NiZr (50 at%) by Ar+ bombardment have been studied with ion scattering spectroscopy (ISS), Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS). In the ISS analysis, a novel approach using survival probability calculated directly from the measured charge-exchange parameter has been applied to obtain the composition of the outermost atomic layer. The subsurface composition has been obtained using AES and XPS. The results suggest that under equilibrium bombardment conditions the Ar+ ion bombardment induces an altered layer at the surface in which the outermost atomic layer is Zr-rich, whereas the underlying atomic layers are depleted in Zr. This is explained in terms of preferential sputtering, radiation-induced segregation and diffusion.
1. Introduction
Ion-induced surface composition changes in binary alloys have been studied extensively as part of investigations of sputter depth profiling in surface analysis, ion beam mixing, ion beam lithography and surface modification [1,2]. Surface composition changes due to ion sputtering were first observed by Gilliam [3] in 1959 in a Cu,Au alloy using electron diffraction. More systematic studies have been carried out on CuNi alloys by Tarng and Wehner using Auger electron spectroscopy (AES) [41. Since then AES has been widely used to investigate this effect for
* Corresponding author. address: Department of Physics, Chungbuk tional University, Cheongju, 360-763, South Korea.
’ Permanent
0039-6028/94/$07.00 0 1994 Elsevier SSDI 0039-6028(93)E0564-B
Science
Na-
various alloys [1,5,61 and in most cases these results have been explained in terms of preferential sputtering. Preferential sputtering results in the enrichment of constituent atoms with the lower sputtering yield at the surface and the composition of the inner layers exponentially approach the bulk concentration by radiation-enhanced diffusion (RED). This is supported by theoretical studies [7] based on a phenomenological steady-state equation for the surface concentration in terms of bulk concentration and the sputtering parameters. However, with the application of ion scattering spectroscopy (ISS) [8-121 to the study of surface composition, it has become clear that AES is not surface sensitive enough to measure the surface composition changes. AES measures an average composition over a number of layers from the outermost atomic layer. In contrast, ISS has proved ex-
B.V. All rights reserved
364
H.J. Kang et al. /Surface
tremely valuable in examining ion-induced surface composition changes of binary alloys because it is very sensitive in the outermost atomic layer. The ISS results together with Auger results have led to a recognition of the importance of radiation-induced surface segregation (RIS) which controls the mass transport from the second atomic layer to the outermost layer. RED accounts for the relaxation of the composition dip generated by the RIS, expanding the region of the altered layer much deeper into the sample. Recently, several theoretical studies [13-151 have strongly supported the importance of RIS in ion sputtering of binary alloys. However, only a limited number and a restricted range of alloy types have been studied 1161 and, therefore, further experimental results of other alloys of various compositions are required to gain a better understanding of the mechanisms associated with ioninduced surface composition changes. In this study, the ion-induced surface composition changes of a Ni-Zr alloy are examined. This alloy has been extensively studied because of its importance and their interesting structural, electronic and magnetic properties [17]. One aspect of particular interest is the charge-exchange process when low-energy ions are scattered from the surface. The interest in this particular alloy arises because of the considerable difference between the charge-exchange rates of noble-gas ions scattered from pure Ni and pure Zr [18]. The charge-exchange rate for Ar+ scattered off pure Ni is approximately double that for Art scattered off pure Zr [18]. Therefore, it is of interest to examine the charge-exchange rates in the Ni-Zr alloy to compare them with those obtained for pure Ni and Zr. This may lead to a better understanding of the mech~ism of neutralisation for ions scattered from metai and alloy surfaces. The aim of the work presented in this paper is to examine whether Ar+ ion bambardment of the Ni-Zr alloy under equilibrium conditions, at room temperature, induces changes in the surface composition. In this study, the chemical composition of the outermost atomic layer has been measured by ISS and the surface region by AES and XPS. The latter techniques reflect an average composition of the surface region depending upon the
Science 302 (1994) 363-370
escape depth of the electrons employed in the analysis. In quantitative ISS analysis, the sensitivity factors were obtained directly from differential cross sections and survival probabilities, instead of from standard spectra.
2. Experimental A sample of the poly-crystalline Ni-Zr alloy with 50 at% Zr was prepared by melting the constituents in an argon arc, using high-purity Ar which was gettered with Zr. The sample was then polished with emery paper (no. 1200) followed by 3 and 1 ym diamond paste and finished with 0.05 pm alumina powder. The ISS measurements were carried out in an angular-resolved ISS, SIMS and TOF apparatus, which was especially designed to study the charge exchange between low-energy ions and solid surfaces. The ISS chamber was pumped, using a Turbo-pump, to a base pressure of lo-” Torr. In this analysis no assumption was made concerning the neutralisation rate of ions scattered off the alloy surface relative to the rates previously measured off the elemental surfaces. The neutralisation rates for ions scattered off the NiZr alloy were measured as part of the quantitative ISS analysis of this sample. The sample stage was arranged to allow the characteristic velocity, I(,, in Eq. (2) to be measured using the same scheme as outlined in a previous paper [191. By rotating a chamfered target about an axis parallel to the incident ion beam it was possible to alter the scattered ion-beam conditions (in particular the perpendicular component of the velocity) while maintaining the incident ion energy, the approach path and the total scattering angle constant. In all experiments, an incidence angle of 60” (measured from the surface normal) and a scattering angle of 90” were used. Measurements of the neutralisation rates and ion-bombardmentinduced surface composition were made independently at Ar+ ion energies of 2, 4 and 6 keV. These were the energies used for sputtering and ISS measurements to determine whether there was a significant dependence on the incident energy of the surface composition profile. The
H.J. Kang et al. /Surface Science 302 (1994) 363-370
sample ion-current density used in the ISS analysis was approximately 5 FAA/cm’. The AES and XPS measurements were carried out in a system ~mp~sing a Leybold EQ.51 electron gun, PHI 04-151 X-ray source and VSW HA45 hemispherical analyser placed 90” to both the electron gun and the X-ray source. The AES/XPS chamber was pumped with a turbo- and a Ti-sublimination pump to a base pressure of 1 X 10-l@ Ton. In making the AES and XPS measurements, particular attention was given to positioning the sample with respect to the analyser by maximising the count rate at a particular energy. All Auger and XPS spectra were collected in the constant pass energy (100 and 200 eV> mode with corresponding energy windows of 2.2 and 4.0 eV, respectively. The primary electron beam current was approximately 250 nA scanned over an area of approximately 200 x 200 pm during the collection of spectra. A primary beam energy of 5 keV was used for Auger spectra and a Mg Ka X-ray source was used for the XPS measurements. The dwell time per spectrum point was 50 ms and the energy increment between the spectrum points was 1 (AES) and 0.25 eV (XPS). In all cases five spectra were averaged in order to reduce noise. The error in the energy setting was approximately 0.25 eV. Spectra were smoothed and in some instances differentiated using the techniques described by Savitzky and Golay [201. Sputtering was carried out with an Ion Tech saddle-field ion gun with 2-4 keV Ar+ ions at a current density of 1 PA/cm*. Before recording spectra, the surface was sputtered clean until all impurities, such as C, 0, etc., were below the detection sensitivity of AES.
3. Results 3.1. ISS measurements A typical ISS spectrum of 4 keV Ar’ scattered from the Ni-Zr (50 at%) alloy (Fig. 1) shows two clear surface peaks sitting on a background. From the ISS spectra, the composition of the outermost
365
a SCATTERED
ION ENERGY
(KeV)
Fig. 1. I!% spectrum of Ni-Zr (50 at%) alloy with 4 keV Ar’ ions.
atomic layer can be estimated using the equation
ml
(1) S = (ONi/Uz~XPNi/Pzr), where I,i and Zz, are the ISS yields for scattering of Ar+ off Ni and Zr in the alloy, while S is the relative sensitivity factor given by the ratio of the differential cross sections ((~~1and the survival probabilities (Pi). Here oi can be calculated using the interatomic cannot be predicted with potential but P,JPz, any certainty. It has been necessary in the past for the sensitivity factors to be assessed from the ratio of spectra obtained from pure standards for the two components, as in AES or XPS quantification. However, there are two important assumptions with this method. One is that the neutralisation probabilities of the elemental standard samples are the same as for the alloy. The other is that the structure factors which include shadowing and blocking effects in the alloy are the same as for the pure standards. To reduce the problems introduced by these assumptions, in ISS quantification, a new approach has been adopted in which the sensitivity factors are obtained directly from the survival probabili~ and differential cross-section ratios. The survival probability P+ of a scattered ion escaping the surface as ion is given by P+= exp( --K/V,),
(2)
H.J. Kang et al /Surface
366
Science 302 (1994) 363-370
Table 1 The measured composition of the outermost atomic layer of Ni-Zr (50 at%) alloy using ISS following ArC ion bombardment
0.5
1 EXIT
1.5 ENERGY
2
Energy CkeV)
*Ni /-Z’z,
2 4 6
0.61 0.59 0.58
[Ni / iZr
pNi ipZr
Z%, 0.44 + 0.05 0.42 f 0.04 0.35 f 0.02
0.076 f 0.003 0.084 f 0.003 0.077 * 0.001
78+3 75rl_3 73+2
2.5
Ei (KeV)
Fig. 2. The value of the characteristic velocity (I(,) normalised to the exit velocity (Vi) as a function of the energy of the exit Ar+ ion.
where I/, is the perpendicular component of velocity of the ion and V, is the characteristic velocity for the interaction which is made up of a transition rate and a screening length. The values of V,/V, were measured for the Ni-Zr alloy by the method used previously [19]. The measurements [19] are presented in Fig. 2 as a function of the energy of the exciting Ar+ ion, with the value of V, normalised to the exit velocity Vi. While the measurements for V,/V, (and the uncertain~) are presented in Table 2, the uncertainty in the concentration is less. This is because the charge-
Table 2 The measured surface and subsurface com~sitions of the Ni-Zr alloy using various analysis techniques with different depth sensitivities after exposure to 2-6 keV Ar+ ions Technique
Ni (at%)
Zr (at%)
2415 35*5
76i5
CW, ISS AES
Ni(MW) Zr(MNN) NKLMM) Zr(MNN) Ni 2p,/, Zr3ds/a Ni 3p Zr3dsja
XPS
0~“““~.)‘~*‘~‘~“““~(‘~I~“~~“.‘~“’~””’..~1 266 366 566 20 ELECTRON
Mean free path [251
Peak (eV)
ENERGY
740
56 146 844 146
2 4.8 5.7 13 5.7 8 20 17 20
65+5 55*5 45+2 5552 45+2 so*2 50+2
926
SV
Fig. 3. N(E) Auger spectra from clean Ni (a) and Zr Cc)and the Ni-Zr (50 at%) alloy (b) after ion bombardment.
367
H.J.Kungetal./Surface Science302 (1994)363-370
exchange-corrected concentration is taken from the intercept of the semilog fit of the data as a function of the perpendicular component of the velocity. The relative uncertainty in the intercept is less than that for the charge-exchange parameter which represents the slope of the fitted line. The values of V, measured for the components of the alloy are different to those previously obtained for the pure elements and this effect is the subject of a separate intensive study. It is important to point out that in this approach the collisional component of the neutralisation has been assumed to be the same for both elements, as has the incident path component. The differential scattering cross section was obtained using both the ZBL-universal potential and the Moliere potential, but there is no significant difference in uNi/gZr obtained from these two potentials. From the sensitivity factor and the ISS spectra the composition of the outermost atomic layer was assessed and the results are presented in Table 1. All these results clearly suggest that the outermost atomic layer of the Ni-Zr alloy under Arf ion bombardment tends to be Zr-rich.
A comparison of the AES spectra measured from the Ni-Zr alloy sample and pure elemental
standards of Ni and Zr are shown in Fig. 3. To determine the elemental composition in the NiZr alloy the low-energy Auger peaks of Ni MVV (56 eV) and Zr MVV (146 eV1 and the high energy Auger peak of NiLMM (844 eV) were used. Fortunately, the low-energy Auger peaks of Ni and Zr are sufficiently separated from each other for them to be used in quantification. The high-energy Zr LMM (1936 eV) peak could not be used because its intensity was too low. The concentration of the alloy constituents, Ci, for both AES and XPS, can be estimated using the equation [221 C Zr c,i-
-F---
~zr/%r (3) Ihdrk
where Zi and Ii0 are the intensities of the element i in the alloy and pure standards, respectively. F denotes the matrix correction factor which includes the backscattering factor (in AES, not in XPS), the inelastic mean free path and the density correction. The AES quantification was carried out using the surface analysis programme developed by Walker et al. [23], whereas in the XPS analysis the F values given by Seah [22] were used. The peak intensities of the Auger spectra were assessed from both the peak heights in the direct N(E) spectra after background subtraction [24]
6F1”~~~‘~‘I*I,.I~1..‘......I~~‘~........”.”.....~ SSS 710 330 350 ELECTRON
Fig. 4.XPS
’
ENERGY
1878
1258
SV
spectraof clean Ni (a) and Zr (c) and the Ni-Zr (50 at%) alloy (b) after ion bombardment.
368
H.J. Kang et al. /Surface
and from the peak-to-peak height of dN(E)/dE spectra. The latter was obtained by the numerical differentiation of the spectra in Fig. 3. This procedure was particularly useful in reducing the difficulties with background subtraction at low energies. The errors in the N(E) data were estimated from the N(E) counts using the standard deviation /‘N(E), and for the differential spectra from the mean noise level. Errors arising from both the NiZr alloy and elemental standards were included. No allowance was made for possible errors in the matrix correction factor. Table 2 lists the results of the surface compositions obtained by the various techniques employed. The AFS result using the NiMVV (56 eV> and MNN (146 eV) Auger signals suggests that the surface composition of the Ni-Zr alloy is Zr-rich, comprising 65 f 5 at% Zr, whereas the composition obtained using NiLMM (844 eV> and Zr MNN (146 eV) Auger peaks gives a surface composition that is slightly Ni-rich. The differences in composition determined from the N(E) spectra and the dN(E)/dE spectra are small and within the error of the measurement. In order to measure the composition averaged over a depth of a few nanometers, XPS was used. The XPS spectra from the Ni-Zr alloy and pure standards are shown in Fig. 4. In using the XPS spectra for quantification the Ni2p,,,, Zr3d and Ni3p peaks were used and the peak intensities were estimated from the peak areas of N(E) spectra. As shown in Table 2, the imposition obtained from Ni2p,,, and Zr3d yieIds a Zr content of 4.5 it 2 at% which is in agreement with the composition obtained from NiLMM and Zr MNN Auger peaks. This agreement is reasonable because the average inelastic mean free paths of the electrons of energies corresponding to the XPS peaks are similar to those for the AES peaks. The composition obtained using Ni3p and Zr3d peaks gives a value which is approximately the same as the bulk composition. In both the AES and XPS spectra, it is of interest to note that the N(E) spectrum for the Ni-Zr alloy is located midway between the N(E) spectra of the pure Ni and pure Zr standards.
Science 302 (1994) 363-370
4. Discussion It is of interest to note that the surface composition assessed from ISS is in good agreement with the composition obtained from low-energy AES. It is well known that ISS is the most surface sensitive technique as it gives information about the outermost atomic layer when using noble-gas ions as the primary ions. AFS, on the other hand, yields an averaged composition up to several atomic layers in depth depending on the escape depth of the Auger electrons. The composition of Zr assessed from AES is, therefore, given as
ri=
i
Cj(Z)
exp[-(2--1)/A
COS 01,
Z=l Cj =
Ni, Zr,
where Cj(z> is the composition of component j at a depth z. A is the inelastic mean free path (IMFP) and 0 is the emission angle of Auger electrons with respect to the surface normal. For Ni MVV (56 eV> and Zr MNN (146 eV>, h = 0.5 nm 125,261; the emission angle is 45”. Therefore, the analysis of the Ni-Zr alloy, using the low-energy Auger transitions, reflects mainly the surface composition of the outermost atomic layer. Although the IMFP of the low-energy Auger electrons is u 0.5 nm the analysis using these electrons will include contributions from the second and possibIy third layers. In view of this, the surface composition obtained with the low-energy AFS is in agreement with the ISS result and supports the conclusion that the outermost layer is Zr-rich. This then indicates that the outermost layer becomes Zr-rich as a rest& of ion sputtering. On the other hand, the surface composition obtained from NiLMM (844 eV), A = 1.3 nm [25,26] and the Zr MNN transitions is slightly N&rich, which agrees with the XPS results obtained using the Ni2p,,, and Zr3d peaks. The surface composition obtained using the Ni 3p and Zr3d peaks in the XPS spectra, for which the h
H.J. Kang et al. /Surface
are approximately 2.0 nm [26], is comparable with the bulk composition. The experimental results obtained from ISS, AES and XPS indicate that ion sputtering of Ni-Zr alloys induces a surface composition change with depth. That is, under equilibrium Ar+ ion-bombardment conditions the preferential sputtering of Ni atoms leads to a Zr enrichment at the outermost atomic layer and a Zr depletion beneath the outermost atomic layer exponentially approaching the bulk composition. The formation of the altered layer in alloys by ion sputtering has been recently explained in terms of preferential sputtering, radiation-induced segregation and radiation-enhanced diffusion [12-151. In Ni-Zr alloys, according to Monte Carlo simulation [15], the ratio of partial sputtering yield YNi/Yz, = 1.8 + 0.1 and is independent of the primary ion energy in the range 2 to 6 keV. If only the difference between the partial sputtering yield is considered, the outermost atomic layer is assessed to be approximately 65 at% Zr, which is not sufficient to explain the observed ISS energy spectra. Therefore, surface segregation of Zr following the preferential sputtering is considered to be an important factor in the composition of the outermost atomic layer. Our thermally annealed experiment 1271confirms the Zr segregation. There are many theoretical approaches [28-301, giving sometimes different predictions on segregation. Among these, we found that the theory 1281based on surface bond breaking and bulk elastic string energy could explain the Zr segregation. Zr, having the lower surface enthalpy 1291, is likely to surface segregate, leading to a more stable equilibrium state. This segregation causes the depleted layer of Zr atoms beneath the outermost atomic layer, whereas RED accounts for the relaxation of the composition dip of Zr generated by RIS, exponentially expanding the region of the altered layer. Therefore, it would appear that surface segregation followed by RED provides an adequate explanation for the observed variation of composition with depth in the NiZr alloy. In conclusion it has been clearly demonstrated that when the Ni-Zr alloy is exposed to continuous Ar+ ion bombardment, the outermost atomic
Science 302 (1994) 363-370
369
layer is Zr-rich, whereas the underlying atomic layers are depleted in Zr. Although Ni is preferentially sputtered, the difference in sputtering yields is not sufficient to explain the observed composition. Therefore, it is necessary to consider other processes such as RIS to explain this result. Zr should surface segregate, which further supports the view that part of the Zr enrichment is due to RIS. In the surface quantification of the Ni-Zr alloy, using ISS, a new approach has been developed and used successfully to obtain the sensitivity factor. This approach involves the use of the differential cross-section and survival probabilities and eliminates any ambiguities introduced by using reference spectra from standard samples.
Acknowledgements The generous assistance of J.D. Browne in preparing the alloys is gratefully acknowledged. This project is supported by an Australian Research Commission grant.
References [l] G. Betz and G.K. Wehner, in: Sputtering by Particle Bombardment II, Ed. R. Behrish (Springer, Heidelberg, 1983) ch. 2. [2] R. Kelly, in: Materials Modification by High Fluence Ion Beam, Eds. R. Kelly and M. Fernanda de Silva (Khrwer, Dordrecht, 1989) p. 305. [3] E. Gilliam, J. Phys. Chem. Solids 11 (1959) 55. [4] M.L. Tarng and G.K. Wehner, J. Appl. Phys. 42 (1971) 2449. [5] H. Shimizu, M. Ono and K. Nakayama, Surf. Sci. 36 (1973) 817. [6] P.S. Ho, J.E. Lewis, H.S. Wildman and J.K. Howard, Surf. Sci. 57 (1976) 393. [7] P.S. Ho, Surf. Sci. 72 (1978) 253. [8] T. Okutani, M. Shikata and R. Shimizu, Surf. Sci. 99 (1980) L410. [9] D.G. Schwarzfiger, S.B. Ziemecki and M.J. Kelly, J. Vat. Sci. Technol. 19 (1981) 185. [lo] H.J. Kang, R. Shimizu and T. Okutani, Surf. Sci. 116 (1982) L173; H.J. Kang, E. Kawatoh and R. Shimizu, Surf. Sci. 144 (1984) 541. [ll] H.H. Brongersma, P.A.J. Ackermans and A.D. van Langeveld, Phys. Rev. B 34 (1986) 5974.
370
H.J. Kang et al. /Surface
[12] G.N. van Wyk, J. du Plessis and E. Taglauer, Surf. Sci.
254 (1991) 73. [13] N.Q. Lam and H. Wiedersich, Nucl. Instrum. Methods B 18 (1987) 471. [14] R. Kelly, Nucl. Instrum. Methods B 39 (1989) 43. 1151H.J. Kang, J.H. Kim, Y.S. Kim, D.W. Moon and R. Shim&u, Surf. Sci. 226 (1990) 93. [16] R. Shim& Nucl. Instrum. Methods B 18 (1987) 486. [17] I. Turek, J. Hafner and Ch. Hausleitner, J. Magn. Magn. Mater. 109 (1992) L145. [18] D.J. O’Connor, Y.G. Shen, J.M. Wilson and R.J. MacDonald, Surf. Sci. 197 (1988) 277. [19] R.F. Garrett, R.J. MacDonald and D.J. O’Connor, Nucl. Instrum. Methods 218 (1983) 333. [20] A. Savitzky and J.E. Golay, Anal. Chem. 36 (1964) 1625. [21] D.J. O’Connor, B.V. King, R.J. MacDonald, Y.G. Shen and X. Chen, Aust. J. Phys. 43 (1990) 601.
Science 302 (1994) 363-370
[22] M.P. Seah, Quantification of AES and XPS, in: Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, Eds. D. Briggs and M.P. Seah (Wiley, New York, 1983). [23] C.G.H. Walker, D.C. Peacock, M. Prutton and M.M. El Gomati, Surf. Interface Anal. 11 (1988) 266. [24] E.N. Sickafus, Phys. Rev. B 16 (1977) 1436. [25] D.R. Penn, Phys. Rev. B 35 (1987) 482. 1261 S. Tanuma, C.J. Powell and D.R. Penn, Surf. Interface Anal. 11 (1988) 577. [27] H.J. Kang, D.J. O’Connor, R.H. Roberts and R.J. MacDonald, to be published. [28] F.F. Abraham, Phys. Rev. Lett. 46 (1981) 546. 1291J.R. Chelikowsky, Surf. Sci. 139 (1984) L197. [30] P.M. Ossi, Surf. Sci. 201 (1988) L519.
““:‘:
v.... :.::::~::::~:~~:~.~~,::~:~.:::::~
‘.‘.‘. ...n
....
......
5;,:,,., ,_,,.(,_,, _.,, ...,..l(_ ,,:;, ,;:,,,, ,::::,
.,..,,
_,,,
‘.‘.‘.“” ” ‘... ...a. .:.: : : .. .,.....,.,.,:.,., :::::?.‘.‘::::::,:,~.~:~,:..::_: ~.~.:.:.::.‘,~.~::::.:.:.:;.‘. ” ‘.‘..... ‘V. . .
::::~>$;:::::.....’
.(.
..A
.,.,
px
,,,,
,,,,,(,,,,
‘.‘...,~.~
surface science
EL-SEWER
Surface
Science 302 (1994) 363-370
Surface composition changes of Ni-Zr alloy under Ar+ ion bombardment by ISS, AES and XPS H.J. Kang ‘, R.H. Roberts, S. He, D.J. O’Connor
*, R.J. MacDonald
Department of Physics, University of Newcastle, Callaghan, NSW 2308, Australia (Received
8 January
1993; accepted
for publication
20 September
1993)
Abstract The changes induced in the surface composition of NiZr (50 at%) by Ar+ bombardment have been studied with ion scattering spectroscopy (ISS), Auger electron spectroscopy (AES) and X-ray photoelectron spectroscopy (XPS). In the ISS analysis, a novel approach using survival probability calculated directly from the measured charge-exchange parameter has been applied to obtain the composition of the outermost atomic layer. The subsurface composition has been obtained using AES and XPS. The results suggest that under equilibrium bombardment conditions the Ar+ ion bombardment induces an altered layer at the surface in which the outermost atomic layer is Zr-rich, whereas the underlying atomic layers are depleted in Zr. This is explained in terms of preferential sputtering, radiation-induced segregation and diffusion.
1. Introduction
Ion-induced surface composition changes in binary alloys have been studied extensively as part of investigations of sputter depth profiling in surface analysis, ion beam mixing, ion beam lithography and surface modification [1,2]. Surface composition changes due to ion sputtering were first observed by Gilliam [3] in 1959 in a Cu,Au alloy using electron diffraction. More systematic studies have been carried out on CuNi alloys by Tarng and Wehner using Auger electron spectroscopy (AES) [41. Since then AES has been widely used to investigate this effect for
* Corresponding author. address: Department of Physics, Chungbuk tional University, Cheongju, 360-763, South Korea.
’ Permanent
0039-6028/94/$07.00 0 1994 Elsevier SSDI 0039-6028(93)E0564-B
Science
Na-
various alloys [1,5,61 and in most cases these results have been explained in terms of preferential sputtering. Preferential sputtering results in the enrichment of constituent atoms with the lower sputtering yield at the surface and the composition of the inner layers exponentially approach the bulk concentration by radiation-enhanced diffusion (RED). This is supported by theoretical studies [7] based on a phenomenological steady-state equation for the surface concentration in terms of bulk concentration and the sputtering parameters. However, with the application of ion scattering spectroscopy (ISS) [8-121 to the study of surface composition, it has become clear that AES is not surface sensitive enough to measure the surface composition changes. AES measures an average composition over a number of layers from the outermost atomic layer. In contrast, ISS has proved ex-
B.V. All rights reserved
364
H.J. Kang et al. /Surface
tremely valuable in examining ion-induced surface composition changes of binary alloys because it is very sensitive in the outermost atomic layer. The ISS results together with Auger results have led to a recognition of the importance of radiation-induced surface segregation (RIS) which controls the mass transport from the second atomic layer to the outermost layer. RED accounts for the relaxation of the composition dip generated by the RIS, expanding the region of the altered layer much deeper into the sample. Recently, several theoretical studies [13-151 have strongly supported the importance of RIS in ion sputtering of binary alloys. However, only a limited number and a restricted range of alloy types have been studied 1161 and, therefore, further experimental results of other alloys of various compositions are required to gain a better understanding of the mechanisms associated with ioninduced surface composition changes. In this study, the ion-induced surface composition changes of a Ni-Zr alloy are examined. This alloy has been extensively studied because of its importance and their interesting structural, electronic and magnetic properties [17]. One aspect of particular interest is the charge-exchange process when low-energy ions are scattered from the surface. The interest in this particular alloy arises because of the considerable difference between the charge-exchange rates of noble-gas ions scattered from pure Ni and pure Zr [18]. The charge-exchange rate for Ar+ scattered off pure Ni is approximately double that for Art scattered off pure Zr [18]. Therefore, it is of interest to examine the charge-exchange rates in the Ni-Zr alloy to compare them with those obtained for pure Ni and Zr. This may lead to a better understanding of the mech~ism of neutralisation for ions scattered from metai and alloy surfaces. The aim of the work presented in this paper is to examine whether Ar+ ion bambardment of the Ni-Zr alloy under equilibrium conditions, at room temperature, induces changes in the surface composition. In this study, the chemical composition of the outermost atomic layer has been measured by ISS and the surface region by AES and XPS. The latter techniques reflect an average composition of the surface region depending upon the
Science 302 (1994) 363-370
escape depth of the electrons employed in the analysis. In quantitative ISS analysis, the sensitivity factors were obtained directly from differential cross sections and survival probabilities, instead of from standard spectra.
2. Experimental A sample of the poly-crystalline Ni-Zr alloy with 50 at% Zr was prepared by melting the constituents in an argon arc, using high-purity Ar which was gettered with Zr. The sample was then polished with emery paper (no. 1200) followed by 3 and 1 ym diamond paste and finished with 0.05 pm alumina powder. The ISS measurements were carried out in an angular-resolved ISS, SIMS and TOF apparatus, which was especially designed to study the charge exchange between low-energy ions and solid surfaces. The ISS chamber was pumped, using a Turbo-pump, to a base pressure of lo-” Torr. In this analysis no assumption was made concerning the neutralisation rate of ions scattered off the alloy surface relative to the rates previously measured off the elemental surfaces. The neutralisation rates for ions scattered off the NiZr alloy were measured as part of the quantitative ISS analysis of this sample. The sample stage was arranged to allow the characteristic velocity, I(,, in Eq. (2) to be measured using the same scheme as outlined in a previous paper [191. By rotating a chamfered target about an axis parallel to the incident ion beam it was possible to alter the scattered ion-beam conditions (in particular the perpendicular component of the velocity) while maintaining the incident ion energy, the approach path and the total scattering angle constant. In all experiments, an incidence angle of 60” (measured from the surface normal) and a scattering angle of 90” were used. Measurements of the neutralisation rates and ion-bombardmentinduced surface composition were made independently at Ar+ ion energies of 2, 4 and 6 keV. These were the energies used for sputtering and ISS measurements to determine whether there was a significant dependence on the incident energy of the surface composition profile. The
H.J. Kang et al. /Surface Science 302 (1994) 363-370
sample ion-current density used in the ISS analysis was approximately 5 FAA/cm’. The AES and XPS measurements were carried out in a system ~mp~sing a Leybold EQ.51 electron gun, PHI 04-151 X-ray source and VSW HA45 hemispherical analyser placed 90” to both the electron gun and the X-ray source. The AES/XPS chamber was pumped with a turbo- and a Ti-sublimination pump to a base pressure of 1 X 10-l@ Ton. In making the AES and XPS measurements, particular attention was given to positioning the sample with respect to the analyser by maximising the count rate at a particular energy. All Auger and XPS spectra were collected in the constant pass energy (100 and 200 eV> mode with corresponding energy windows of 2.2 and 4.0 eV, respectively. The primary electron beam current was approximately 250 nA scanned over an area of approximately 200 x 200 pm during the collection of spectra. A primary beam energy of 5 keV was used for Auger spectra and a Mg Ka X-ray source was used for the XPS measurements. The dwell time per spectrum point was 50 ms and the energy increment between the spectrum points was 1 (AES) and 0.25 eV (XPS). In all cases five spectra were averaged in order to reduce noise. The error in the energy setting was approximately 0.25 eV. Spectra were smoothed and in some instances differentiated using the techniques described by Savitzky and Golay [201. Sputtering was carried out with an Ion Tech saddle-field ion gun with 2-4 keV Ar+ ions at a current density of 1 PA/cm*. Before recording spectra, the surface was sputtered clean until all impurities, such as C, 0, etc., were below the detection sensitivity of AES.
3. Results 3.1. ISS measurements A typical ISS spectrum of 4 keV Ar’ scattered from the Ni-Zr (50 at%) alloy (Fig. 1) shows two clear surface peaks sitting on a background. From the ISS spectra, the composition of the outermost
365
a SCATTERED
ION ENERGY
(KeV)
Fig. 1. I!% spectrum of Ni-Zr (50 at%) alloy with 4 keV Ar’ ions.
atomic layer can be estimated using the equation
ml
(1) S = (ONi/Uz~XPNi/Pzr), where I,i and Zz, are the ISS yields for scattering of Ar+ off Ni and Zr in the alloy, while S is the relative sensitivity factor given by the ratio of the differential cross sections ((~~1and the survival probabilities (Pi). Here oi can be calculated using the interatomic cannot be predicted with potential but P,JPz, any certainty. It has been necessary in the past for the sensitivity factors to be assessed from the ratio of spectra obtained from pure standards for the two components, as in AES or XPS quantification. However, there are two important assumptions with this method. One is that the neutralisation probabilities of the elemental standard samples are the same as for the alloy. The other is that the structure factors which include shadowing and blocking effects in the alloy are the same as for the pure standards. To reduce the problems introduced by these assumptions, in ISS quantification, a new approach has been adopted in which the sensitivity factors are obtained directly from the survival probabili~ and differential cross-section ratios. The survival probability P+ of a scattered ion escaping the surface as ion is given by P+= exp( --K/V,),
(2)
H.J. Kang et al /Surface
366
Science 302 (1994) 363-370
Table 1 The measured composition of the outermost atomic layer of Ni-Zr (50 at%) alloy using ISS following ArC ion bombardment
0.5
1 EXIT
1.5 ENERGY
2
Energy CkeV)
*Ni /-Z’z,
2 4 6
0.61 0.59 0.58
[Ni / iZr
pNi ipZr
Z%, 0.44 + 0.05 0.42 f 0.04 0.35 f 0.02
0.076 f 0.003 0.084 f 0.003 0.077 * 0.001
78+3 75rl_3 73+2
2.5
Ei (KeV)
Fig. 2. The value of the characteristic velocity (I(,) normalised to the exit velocity (Vi) as a function of the energy of the exit Ar+ ion.
where I/, is the perpendicular component of velocity of the ion and V, is the characteristic velocity for the interaction which is made up of a transition rate and a screening length. The values of V,/V, were measured for the Ni-Zr alloy by the method used previously [19]. The measurements [19] are presented in Fig. 2 as a function of the energy of the exciting Ar+ ion, with the value of V, normalised to the exit velocity Vi. While the measurements for V,/V, (and the uncertain~) are presented in Table 2, the uncertainty in the concentration is less. This is because the charge-
Table 2 The measured surface and subsurface com~sitions of the Ni-Zr alloy using various analysis techniques with different depth sensitivities after exposure to 2-6 keV Ar+ ions Technique
Ni (at%)
Zr (at%)
2415 35*5
76i5
CW, ISS AES
Ni(MW) Zr(MNN) NKLMM) Zr(MNN) Ni 2p,/, Zr3ds/a Ni 3p Zr3dsja
XPS
0~“““~.)‘~*‘~‘~“““~(‘~I~“~~“.‘~“’~””’..~1 266 366 566 20 ELECTRON
Mean free path [251
Peak (eV)
ENERGY
740
56 146 844 146
2 4.8 5.7 13 5.7 8 20 17 20
65+5 55*5 45+2 5552 45+2 so*2 50+2
926
SV
Fig. 3. N(E) Auger spectra from clean Ni (a) and Zr Cc)and the Ni-Zr (50 at%) alloy (b) after ion bombardment.
367
H.J.Kungetal./Surface Science302 (1994)363-370
exchange-corrected concentration is taken from the intercept of the semilog fit of the data as a function of the perpendicular component of the velocity. The relative uncertainty in the intercept is less than that for the charge-exchange parameter which represents the slope of the fitted line. The values of V, measured for the components of the alloy are different to those previously obtained for the pure elements and this effect is the subject of a separate intensive study. It is important to point out that in this approach the collisional component of the neutralisation has been assumed to be the same for both elements, as has the incident path component. The differential scattering cross section was obtained using both the ZBL-universal potential and the Moliere potential, but there is no significant difference in uNi/gZr obtained from these two potentials. From the sensitivity factor and the ISS spectra the composition of the outermost atomic layer was assessed and the results are presented in Table 1. All these results clearly suggest that the outermost atomic layer of the Ni-Zr alloy under Arf ion bombardment tends to be Zr-rich.
A comparison of the AES spectra measured from the Ni-Zr alloy sample and pure elemental
standards of Ni and Zr are shown in Fig. 3. To determine the elemental composition in the NiZr alloy the low-energy Auger peaks of Ni MVV (56 eV) and Zr MVV (146 eV1 and the high energy Auger peak of NiLMM (844 eV) were used. Fortunately, the low-energy Auger peaks of Ni and Zr are sufficiently separated from each other for them to be used in quantification. The high-energy Zr LMM (1936 eV) peak could not be used because its intensity was too low. The concentration of the alloy constituents, Ci, for both AES and XPS, can be estimated using the equation [221 C Zr c,i-
-F---
~zr/%r (3) Ihdrk
where Zi and Ii0 are the intensities of the element i in the alloy and pure standards, respectively. F denotes the matrix correction factor which includes the backscattering factor (in AES, not in XPS), the inelastic mean free path and the density correction. The AES quantification was carried out using the surface analysis programme developed by Walker et al. [23], whereas in the XPS analysis the F values given by Seah [22] were used. The peak intensities of the Auger spectra were assessed from both the peak heights in the direct N(E) spectra after background subtraction [24]
6F1”~~~‘~‘I*I,.I~1..‘......I~~‘~........”.”.....~ SSS 710 330 350 ELECTRON
Fig. 4.XPS
’
ENERGY
1878
1258
SV
spectraof clean Ni (a) and Zr (c) and the Ni-Zr (50 at%) alloy (b) after ion bombardment.
368
H.J. Kang et al. /Surface
and from the peak-to-peak height of dN(E)/dE spectra. The latter was obtained by the numerical differentiation of the spectra in Fig. 3. This procedure was particularly useful in reducing the difficulties with background subtraction at low energies. The errors in the N(E) data were estimated from the N(E) counts using the standard deviation /‘N(E), and for the differential spectra from the mean noise level. Errors arising from both the NiZr alloy and elemental standards were included. No allowance was made for possible errors in the matrix correction factor. Table 2 lists the results of the surface compositions obtained by the various techniques employed. The AFS result using the NiMVV (56 eV> and MNN (146 eV) Auger signals suggests that the surface composition of the Ni-Zr alloy is Zr-rich, comprising 65 f 5 at% Zr, whereas the composition obtained using NiLMM (844 eV> and Zr MNN (146 eV) Auger peaks gives a surface composition that is slightly Ni-rich. The differences in composition determined from the N(E) spectra and the dN(E)/dE spectra are small and within the error of the measurement. In order to measure the composition averaged over a depth of a few nanometers, XPS was used. The XPS spectra from the Ni-Zr alloy and pure standards are shown in Fig. 4. In using the XPS spectra for quantification the Ni2p,,,, Zr3d and Ni3p peaks were used and the peak intensities were estimated from the peak areas of N(E) spectra. As shown in Table 2, the imposition obtained from Ni2p,,, and Zr3d yieIds a Zr content of 4.5 it 2 at% which is in agreement with the composition obtained from NiLMM and Zr MNN Auger peaks. This agreement is reasonable because the average inelastic mean free paths of the electrons of energies corresponding to the XPS peaks are similar to those for the AES peaks. The composition obtained using Ni3p and Zr3d peaks gives a value which is approximately the same as the bulk composition. In both the AES and XPS spectra, it is of interest to note that the N(E) spectrum for the Ni-Zr alloy is located midway between the N(E) spectra of the pure Ni and pure Zr standards.
Science 302 (1994) 363-370
4. Discussion It is of interest to note that the surface composition assessed from ISS is in good agreement with the composition obtained from low-energy AES. It is well known that ISS is the most surface sensitive technique as it gives information about the outermost atomic layer when using noble-gas ions as the primary ions. AFS, on the other hand, yields an averaged composition up to several atomic layers in depth depending on the escape depth of the Auger electrons. The composition of Zr assessed from AES is, therefore, given as
ri=
i
Cj(Z)
exp[-(2--1)/A
COS 01,
Z=l Cj =
Ni, Zr,
where Cj(z> is the composition of component j at a depth z. A is the inelastic mean free path (IMFP) and 0 is the emission angle of Auger electrons with respect to the surface normal. For Ni MVV (56 eV> and Zr MNN (146 eV>, h = 0.5 nm 125,261; the emission angle is 45”. Therefore, the analysis of the Ni-Zr alloy, using the low-energy Auger transitions, reflects mainly the surface composition of the outermost atomic layer. Although the IMFP of the low-energy Auger electrons is u 0.5 nm the analysis using these electrons will include contributions from the second and possibIy third layers. In view of this, the surface composition obtained with the low-energy AFS is in agreement with the ISS result and supports the conclusion that the outermost layer is Zr-rich. This then indicates that the outermost layer becomes Zr-rich as a rest& of ion sputtering. On the other hand, the surface composition obtained from NiLMM (844 eV), A = 1.3 nm [25,26] and the Zr MNN transitions is slightly N&rich, which agrees with the XPS results obtained using the Ni2p,,, and Zr3d peaks. The surface composition obtained using the Ni 3p and Zr3d peaks in the XPS spectra, for which the h
H.J. Kang et al. /Surface
are approximately 2.0 nm [26], is comparable with the bulk composition. The experimental results obtained from ISS, AES and XPS indicate that ion sputtering of Ni-Zr alloys induces a surface composition change with depth. That is, under equilibrium Ar+ ion-bombardment conditions the preferential sputtering of Ni atoms leads to a Zr enrichment at the outermost atomic layer and a Zr depletion beneath the outermost atomic layer exponentially approaching the bulk composition. The formation of the altered layer in alloys by ion sputtering has been recently explained in terms of preferential sputtering, radiation-induced segregation and radiation-enhanced diffusion [12-151. In Ni-Zr alloys, according to Monte Carlo simulation [15], the ratio of partial sputtering yield YNi/Yz, = 1.8 + 0.1 and is independent of the primary ion energy in the range 2 to 6 keV. If only the difference between the partial sputtering yield is considered, the outermost atomic layer is assessed to be approximately 65 at% Zr, which is not sufficient to explain the observed ISS energy spectra. Therefore, surface segregation of Zr following the preferential sputtering is considered to be an important factor in the composition of the outermost atomic layer. Our thermally annealed experiment 1271confirms the Zr segregation. There are many theoretical approaches [28-301, giving sometimes different predictions on segregation. Among these, we found that the theory 1281based on surface bond breaking and bulk elastic string energy could explain the Zr segregation. Zr, having the lower surface enthalpy 1291, is likely to surface segregate, leading to a more stable equilibrium state. This segregation causes the depleted layer of Zr atoms beneath the outermost atomic layer, whereas RED accounts for the relaxation of the composition dip of Zr generated by RIS, exponentially expanding the region of the altered layer. Therefore, it would appear that surface segregation followed by RED provides an adequate explanation for the observed variation of composition with depth in the NiZr alloy. In conclusion it has been clearly demonstrated that when the Ni-Zr alloy is exposed to continuous Ar+ ion bombardment, the outermost atomic
Science 302 (1994) 363-370
369
layer is Zr-rich, whereas the underlying atomic layers are depleted in Zr. Although Ni is preferentially sputtered, the difference in sputtering yields is not sufficient to explain the observed composition. Therefore, it is necessary to consider other processes such as RIS to explain this result. Zr should surface segregate, which further supports the view that part of the Zr enrichment is due to RIS. In the surface quantification of the Ni-Zr alloy, using ISS, a new approach has been developed and used successfully to obtain the sensitivity factor. This approach involves the use of the differential cross-section and survival probabilities and eliminates any ambiguities introduced by using reference spectra from standard samples.
Acknowledgements The generous assistance of J.D. Browne in preparing the alloys is gratefully acknowledged. This project is supported by an Australian Research Commission grant.
References [l] G. Betz and G.K. Wehner, in: Sputtering by Particle Bombardment II, Ed. R. Behrish (Springer, Heidelberg, 1983) ch. 2. [2] R. Kelly, in: Materials Modification by High Fluence Ion Beam, Eds. R. Kelly and M. Fernanda de Silva (Khrwer, Dordrecht, 1989) p. 305. [3] E. Gilliam, J. Phys. Chem. Solids 11 (1959) 55. [4] M.L. Tarng and G.K. Wehner, J. Appl. Phys. 42 (1971) 2449. [5] H. Shimizu, M. Ono and K. Nakayama, Surf. Sci. 36 (1973) 817. [6] P.S. Ho, J.E. Lewis, H.S. Wildman and J.K. Howard, Surf. Sci. 57 (1976) 393. [7] P.S. Ho, Surf. Sci. 72 (1978) 253. [8] T. Okutani, M. Shikata and R. Shimizu, Surf. Sci. 99 (1980) L410. [9] D.G. Schwarzfiger, S.B. Ziemecki and M.J. Kelly, J. Vat. Sci. Technol. 19 (1981) 185. [lo] H.J. Kang, R. Shimizu and T. Okutani, Surf. Sci. 116 (1982) L173; H.J. Kang, E. Kawatoh and R. Shimizu, Surf. Sci. 144 (1984) 541. [ll] H.H. Brongersma, P.A.J. Ackermans and A.D. van Langeveld, Phys. Rev. B 34 (1986) 5974.
370
H.J. Kang et al. /Surface
[12] G.N. van Wyk, J. du Plessis and E. Taglauer, Surf. Sci.
254 (1991) 73. [13] N.Q. Lam and H. Wiedersich, Nucl. Instrum. Methods B 18 (1987) 471. [14] R. Kelly, Nucl. Instrum. Methods B 39 (1989) 43. 1151H.J. Kang, J.H. Kim, Y.S. Kim, D.W. Moon and R. Shim&u, Surf. Sci. 226 (1990) 93. [16] R. Shim& Nucl. Instrum. Methods B 18 (1987) 486. [17] I. Turek, J. Hafner and Ch. Hausleitner, J. Magn. Magn. Mater. 109 (1992) L145. [18] D.J. O’Connor, Y.G. Shen, J.M. Wilson and R.J. MacDonald, Surf. Sci. 197 (1988) 277. [19] R.F. Garrett, R.J. MacDonald and D.J. O’Connor, Nucl. Instrum. Methods 218 (1983) 333. [20] A. Savitzky and J.E. Golay, Anal. Chem. 36 (1964) 1625. [21] D.J. O’Connor, B.V. King, R.J. MacDonald, Y.G. Shen and X. Chen, Aust. J. Phys. 43 (1990) 601.
Science 302 (1994) 363-370
[22] M.P. Seah, Quantification of AES and XPS, in: Practical Surface Analysis by Auger and X-ray Photoelectron Spectroscopy, Eds. D. Briggs and M.P. Seah (Wiley, New York, 1983). [23] C.G.H. Walker, D.C. Peacock, M. Prutton and M.M. El Gomati, Surf. Interface Anal. 11 (1988) 266. [24] E.N. Sickafus, Phys. Rev. B 16 (1977) 1436. [25] D.R. Penn, Phys. Rev. B 35 (1987) 482. 1261 S. Tanuma, C.J. Powell and D.R. Penn, Surf. Interface Anal. 11 (1988) 577. [27] H.J. Kang, D.J. O’Connor, R.H. Roberts and R.J. MacDonald, to be published. [28] F.F. Abraham, Phys. Rev. Lett. 46 (1981) 546. 1291J.R. Chelikowsky, Surf. Sci. 139 (1984) L197. [30] P.M. Ossi, Surf. Sci. 201 (1988) L519.