1112
Anal. Chem. 1988, 58,1112-1115
Therefore, it can be expected that the calculated ion yields will accurately represent the acutal ion yields for many of the elements listed in the table. It should be emphasized, however, that the LTE calculation gives impractical yields for some elements. For example, it provides too low yields for low ionization potential elements such as F and C1, and for Sn it produces a yield 2 orders of magnitude higher than that obtained by Leta and Morrison ( I ) . These apparently incorrect data have been excluded in Table VI. Other incorrect cases are naturally expected. In general, the accuracy of the LTE is considered to be within a factor of 2 (6, 7,19). However, the relative ion yields can vary depending on the instrument used and the tuning condition of the instrument transmission. Thus, the ion yield set presented in this work cannot by itself provide sufficiently accurate quantitative values. It can, however, provide foreknowledge of the ion intensities in GaAs or the semiquantitative values for elements without calibration standards. In this context, LTE calculation can be thought of as being effective for practical use.
ACKNOWLEDGMENT We are indebted to S. Hattori and S. Kurosawa for their valuable discussions apd encouragement in this work. Registry No. Li, 7439-93-2; Be, 7440-41-7;B, 7440-42-8; C, 7440-44-0; N2, 7727-37-9; Na, 7440-23-5; Mg, 7439-95-4; Al, 7429-90-5; Si, 7440-21-3;P, 7723-14-0;S, 7704-34-9;K, 7440-09-7; Ca, 7440-70-2; SC, 7440-20-2; Ti, 7440-32-6; V, 7440-62-2; Cr, 7440-47-3; Mn, 7439-96-5; Fe, 7439-89-6; Co, 7440-48-4; Ni, 7440-02-0; Cu, 7440-50-8; Zn, 7440-66-6; Ge, 7440-56-4; Se, 7782-49-2; Bra, 7726-95-6; Nb, 7440-03-1; Mo, 7439-98-7; Ag, 7440-22-4; Rb, 7440-17-7; Cd, 7440-43-9; In, 7440-74-6; Sb,
7440-36-0; Te, 13494-80-9; Cs, 7440-46-2; Ba, 7440-39-3; Ta, 7440-25-7; W, 7440-33-7; Pt, 7440-06-4; Au, 7440-57-5; Hg, 77827439-97-6;T1, 7440-28-0; Pb, 7439-92-1; Bi, 7440-69-9; 02, 44-7; GaAs, 1303-00-0.
LITERATURE CITED Leta, D. P.; Morrison, G. H. Anal. Chem. 1980, 52,514. Homma, Y.; Ishii, Y.; Kobayashi, T.; Osaka, J. J. Appl. Phys. 1985, 57,2931. Kurosawa, S.;Homma, Y.; Tanaka, T.; Yamawaki, M. I n "Proceedings of the Fourth International Conference on Secondary Ion Mass Spectrometry SIMS-IV"; Benninghoven, A., Okano, J., Shimizu, R., Werner, H. W., Eds.; Springer: Berlin, 1964; pp 107-109. Homma, Y.; Kurosawa, S.;Yoshioka, Y.; Shibata, M.; Nomura, K.; Nakamura Y. Anal. Chem. 1985, 57,2928. Storms, H. A.; Brown, K. F.;Stein, J. D. Anal. Chem. 1977, 49, 2023. Morgan, A. E.;Werner, H. W. Anal. Chem. 1976, 48, 699. Morgan, A. E.; Werner H. W. Anal. Chem. 1977, 49, 927. Andersen, C. A.; Hinthorne, J. R. Science (Washington, D.C.) 1972, 775, 853. Andersen, C. A.; Hinthorne, J. R. Anal. Chem. 1973, 45, 1421. Ehlert, T. C. J. Phys. €1970, 3, 237. Wittmaack, K. I n "Inelastic Ion-Surface Collisions"; Tolk, N. H., Tully, J. C., Heiland, W., White, C. W., Eds.; Academic Press: New York, 1977; pp 153-199. Homma, Y.; Tanaka, H.; Ishii, Y. I n "Proceedings of the Fourth International Conference on Secondary Ion Mass Spectrometry SIMS-IV"; Benninghoven, A., Okano, J., Shimizu, R., Werner, H. W., Eds.; Springer: Berlin, 1984; pp 98-100. Pretzer, D. D.; Hagstrum, H. D. Surf. Sci. 1986, 4, 265. Ibach, H.; Horn, K.; Dorn, R.; Luth, H. Surf. Sci. 1973, 38, 433. Shimizu, R.; Ishitani, T.; Ueshima, Y. Jpn. J. Appl. Phys. 1974, 73, 249. de Galan, L.; Smith, R.; Winefordner, J. D. Spectrochim. Acta, Part 6 1968, 23.521. Liebl, H. J. Vac. Sci. Technol. 1975, 72, 385. Morgan, A. E.; Werner, H. W. Surf. Sci. 1977, 65,687. Ramseyar, G. 0.; Morrison, G. H. Anal. Chem. 1984, 55,1963.
RECEIVED for review August 8,1985. Accepted December 9, 1985.
Emission Spectroscopic Studies of Sputtering on Silver-Copper Alloy Surfaces Kazuaki Wagatsuma and Kichinosuke Hirokawa* The Research Institute for Iron, Steel, and Other Metals, Tohoku University, Sendai 980, Japan
Glow discharge Sputtering of Ag-Cu alloy surfaces was studied by optical emission spectrometry. The change in the intensity ratio between Ag and Cu emission llnes was monltored durlng the course of sputterlng. The metallurgical structures (fine partlcles of the eutectoid and coarse grains of the primary crystallltes) played an Important role In determining the length of transftion perlods before steady-state sputtering. A reduction in the overall sputtering yields wlth sputter time, probably due to the coating of Cu atoms onto the Ag grdns, was also observed.
Glow discharge emission spectrometry (GDS) using a Grimm-type light source ( I , 2 )is recognized as a useful and powerful technique for qualitative and/or quantitative surface analysis (3-6). Cathode sputtering in the discharge tube is basic for understanding how sample atoms are introduced into the glow discharge plasma. If an alloy surface is subjected to ion bombardment, the different components are sputtered
at different rates. This effect is called preferential sputtering (7). Preferential sputtering leads to a change in the surface composition and the atomic density of the sputtered elements in the plasma. Sputtering of many binary alloy systems has been studied, mainly with Auger electron spectroscopy (7-9)) and these investigations experimentally show the altered layers on the alloy surfaces. These surface modifications would also influence analytical results obtained with GDS. Therefore, observation of emission intensities from the different components in multiphase systems provides information about glow discharge sputtering mechanisms. Ion bombardment of multiphase targets shows much more complicated behavior than occurs with single-phase materials. The crystallites materials are heterogeneous. The crystallities in such structures represent different compositions, different metallurgical phases, and various sizes. While sputtering of homogeneous materials proceeds rapidly under steady-state conditions and altered layers are formed on the surfaces, sputtering of multiphase alloys is marked by a long transition time before steady-state conditions are reached ( I O ) , and such
0003-2700/86/0358~1112$01.50/0 0 1986 American Chemlcal Society
sputtered surfaces are modified by the progress of surface microrelief (11,12)and coating by the low-yield element (13, 14). The prolonged transition period as well as the development of the surfaces topography is closely related not only to the sputtering yields of the constituents but also to grain size and orientation (14, 15). We have previously reported surface investigation of several binary and ternary alloys using a low-power GDS (16,17).It has been concluded that the mild sputtering conditions that are realized in the low-power glow discharge are suitable for the study of surfaces. In this paper, sputtering phenomena of silver-copper alloys are investigated in detail, and variation in the emission intensities with the course of the cathode sputtering is discussed. In Ag-Cu alloys at room temperatures, the solid solubility of one component in the other is negligible; that is, pure Ag graihs coexist with pure Cu grains. This two-phase alloy has a eutectic point a t 39.9 atom % Cu (18). The eutectic phase is composed of a mixture of very fine particles of Ag with those of Cu (18). On the other hand, away from the eutectic point, the coarse grains of primary crystal (Cu for hypereutectic and Ag for hypoeutectic) remain. The grain size of the primary crystal is much larger than that of the eutectoid. Some authors have measured the sputtering yield of pure elemebts and their alloys (19,20), and the sputtering yield of Ag is estimated to be somewhat higher than that of Cu.
a
%?ai0.9
RESULTS AND DISCUSSION Four resonance lines (Cu 1324.7 nm, Cu 1327.4 nm, Ag I 328.1 nm, and Ag 1338.2 nm), which have nearly the same upper energy level values (24, 251, are suitable for use as analytical emission lines. If the intensity ratios are strongly modified by self-absorption effects, the signal intensity ratio of the emission lines cannot correspond linearly to the atomic density ratio of the analyte atoms. Because the transition probability (gA value) of Cu 1324.7 is about twice as large as that of Cu I 327.4 (25),it is assumed that the degree of self-absorption effects, if any, is different between these Cu
I
r
I
t
I
Ag-68.4at%Cu hyper-eutectic
-
Plasma Gas : Argon : 6.7X &a Discharge : 0.5mA.310 v
Ag4u
b
I
'
I
I
I
I
I
I
Ag-68.4at%Cu
0 . q
-
hyper-eutectic
0.5 Plasma Gas : Argon : 6.7X102Pa Discharge : 1. OmA ,380V
EXPERIMENTAL SECTION Silver-copper alloys containing 40.1 and 68.4 atom % Cu were prepared with 99.99% Cu and Ag ingots in an induction furnace with an argon atmosphere. The samples were rolled to about 1.0 mm thickness and annealed at 600 "C for 10-12 h at 10-2-10-3 Pa. With an optical microscope, it was observed that the sample at the eutectic point (40.1 atom % Cu) consisted of only the eutectic-phase grains and that the alloy at 68.4 atom % Cu consisted of the coarse particles (primary crystal) and the eutectoid. The average grain size of the primary crystals can be estimated to be 50-100 pm and that of the eutectoid to be less than 1.0 ,urn. The surfaces were polished with waterproof emery papers and then finished to mirror faces with A1203abrasive materials. Schematic diagrams of our glow discharge lamp (21)and our apparatus (22)are given elsewhere. The structure of the lamp was similar to the original model by Grimm (I). The inner diameter of the hollow anode was 4.0 mm, and the distance between the anode and cathode was adjusted to be 0.4-0.6 mm. The lamp was evacuated to 4.0-1.3 Pa and then filled with argon gas (99.9995% purity) to 6.7 X lo2 Pa. A decrease in emission intensities, which is caused by a drop in the excitation temperature and a decrease in the amount of sputtered atoms, may occur with the low-power GDS. An applied voltage modulation technique (AVM) was employed to detect emission signals as effectively and reliably as possible (22,23). This method is based on a phase-sensitive detection technique with a lock-in amplifier. The radiation emitted from the plasma can be periodically modulated by a cyclic variation in input voltage. Only the modulated components are selectively detected against predominant noise sources. Compared to a conventional dc amplification method, the signal-to-noise ratio has been improved by a factor of 20-50 (22).
I
I
'
0
20
60
40
80
rnin .
SPUTTER TIME
c
'
I
I
I
I
I
I
I
Ag - 68 4at%Cu hyper-eutectic
0 .a
I
I Plasma Gas : Argon : 6.7XidPo Discharge : i.5mA. 395 V
1 0
I
1
20
I
I
I
40
SPUTTER TIME
t
60
1
I
80
min.
Flgure 1. Sputter time dependence of the relative intensity between AQ 1328.1 and Cu I 327.4 nm for Ag - 68.4 atom% Cu (hypereutectic). Ar' bombardment Is carried out under the following conditions: Ar pressure of 6.7 X lo2 Pa and discharge powers of (a, top) 0.5 mA/310 V, (b, middle) 1.0 mA/380 V, and (c, bottom) 1.5 mA/395 V.
I lines. Therefore, it is useful to examine the intensity ratio between two resonance lines of Cu. If the ratio of these intensities does not vary with the input power, self-absorption effects are not significant. For Ag, similar considerations apply. Intensities of these resonance emission lines are recorded in an input voltage range of 310-420 v. Even though the emission intensities were greatly altered by changing the input voltage, the intensity ratios, (Cu 1324.71Cu I 327.4) and (Ag I 328.1/Ag I 338.2), were calculated to be 1.9-2.0 and 2.0-2.1, respectively, and independent of the input voltages. Therefore, when these emission lines are employed in the voltage range described above, the contribution of self-absorption effects can be neglected.
1114
ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986 Ag- 68.4 ot XCU
&mr- eutectic 0.5mA. X'OV
5;'IO*
19'30"
0 5 m a . 310 V I
0
PO
40
I
I
60
SPUTTER T I M E
I
80
I
I
100
I
I
I20
min.
Figure 1 indicates the change in the intensity ratio (Ag I 328.1f (Ag I 328.1 + Cu I 327.4)) as a function of sputtering time in Ag-68.4 atom % Cu alloy (hypereutectic). Stronger input power resulted in shorter transition periods before steady-state conditions were reached. The fact that Ag relative intensities are higher than those of Cu during the transient stage is due to the higher sputtering yield of Ag compared to the yield of Cu. By use of gravimetric analysis, the erosion rate can be estimated to be 0.5-2.0 pm/min when the lamp is operated at 1.5 mA/395 V. Thus, in such conditions, a sample thickness of 5-20 pm would be removed from the target surface before a steady-state sputter. Because the grain size of the eutectic phase is very fine (less than 1.0 pm), it seems that the erosion which occurred in the transient stage was related to the particle size of the primary crystals rather than the eutectoid. Some studies have indicated that steady-state conditions are reached after sputter ejection of a surface layer of the order of a grain diameter in multiphase allby systems (13,15).However, if the formation of surface microrelief such as cones, pyramids, and etch pits is the dominant process with the development of sputtering, steady-state conditions are obtained more rapidly (15,26). In our experiments, cone structures were also observed on the sputtered surfaces. Therefore, the results in Figure 1 can be interpreted from both the difference of sputter yield between Ag and Cu and the formation of surface topography. As shown in Figure 1, the intensity ratio for steady-state sputtering gradually decreases with an increase in the input voltage and current. This effect shows that the surface composition depends on the sputtering conditions even in steady-state sputtering. The progressive alteration of surface microrelief leads not only to changes in microscopic aspects of the surface but also to changes in the surface composition due to coating of the low-yield element onto the high-yield grains in a multiphase system (13). Furthermore, a great surface segregation of Ag atoms has been reported in Ag-Cu alloys subjected to ion bombardment at room temperatures (27). If the erosion rate is so fast, the segregation process is expected not to be dominant, and surface coatings will occur mostly due to rapid collapses of the low-yield grains. Therefore, our observations can be explained from these surface modification processes. The emission intensity of argon atoms and ions provides useful information on populated Ar atom and ion states. Variation in the excitation temperature of the plasma is reflected by a change in the emission intensities (28). Figure 2 shows the relation between the intensities of Ar emission lines and the sputter time at an input power of 0.5 mA/310 V. For three different emission lines (Ar I1 413.2 nm, Ar I1
i
*-
,
I
r
I
I
I
Ag-684of%Cu hyper- eutectic
0.9
0
5Z2O"
J
Flgure 2. Relation between the intensltles of three different argon emission lines and sputter time at a discharge power of 0.5 mA1310 V. The Ar line intensities are plotted as follows: ( 0 )Ar I415.8, (0) Ar I I 413.2, and (e)Ar I 1 410.4 nm.
5
10
I5
20.0
SPUTTER
5 TIME
IO
15
20
min.
Flgure 4. Sputtered surface, which was treated at an input power of 1.O mAI380 V for 20 min, subsequently subjected to Ar' bombardment under the same sputtering conditions. Plots of the relative intensity are shown for both the as-polished and sputtered surface.
410.4 nm, and Ar 1415.8 nm), these intensities indicate no time dependence. It is considered that plasma conditions, i.e., excitation temperature, remained constant through the measurement. Therefore, the emission intensity from sputtered elements is determined mainly by the amount of ejected atoms in the plasma. Figure 3 shows changes in the intensities of Cu I and Ag I lines with increased sputtering time at an input power of 0.5 mA/310 V. A gradual increase in the emission intensities is observed for ca. 5 min after breakdowns, indicating that apparent sputtering yields are suppressed by surface contaminants such as a hydrocarbons. After the removal of the initial surface layer, it is significant to note a decrease in the emission intensity of Ag I 328.1 nm. In contrast, the emission intensities of Cu 1327.4 nm and 324.7 nm hardly change with the sputter time. Because the Cu grains, which are left with the development of preferential sputtering, will collapse onto the Ag grains, these phenomena are caused by coating of Cu atoms onto the Ag grains. Therefore, the overall
ANALYTICAL CHEMISTRY, VOL. 58, NO. 6, MAY 1986
sputter erosion of a layer in the order of a few grain diameters (15). As described above, the grain size of the eutectoid is very fine; therefore, the transition to the steady-state conditions is assumed to be more rapid than that of the hypereutectoid. Furthermore, the structure of surface microrelief in the eutectic phase sample is quite different from that in the hypereutectoid due to the absence of coarse crystallites. A steady-state surface topography may be easily attained when the particle mixture of multiphase systems is fine. It is interesting to note that the response of emission signals with the course of sputtering strongly depends upon the metallurgical structure and the composition of the sample in a multiphase alloy system. Particularly, the prolonged transition periods in the hypereutectoid should be marked when multiphase alloys are employed as a sputter-target material for the production of thin films. In single-phase alloy systems, as far as the authors know, there are no reports of such phenomena when the samples are sputtered at ambient temperatures. Registry NO.Ag, 7440-22-4;CU,7440-50-8; Ag-Cu, 11144-43-7.
c: 0.8 I
Plosmo Gas : Argon : 6.7XiO'Po Discharge : 0 . 5 m A . 3 3 0 V
,
20
0
Ag- 40. lat%Cu eutectic
40
80
60
SPUTTER TIME
b
'
I
I
,
min. I
I
CutAg
0.8
?
t
1
Plosma Gas
:
Arm
:6 . 7 x i 0 ~ ~ 0
Dischorge
0
20
: 1.0mA , 3 9 0 V
40
Ag- 4 0 . / d % C u eutectic
60
BO
1
min . Flgure 5. Sputter time dependence of the relative Ag emission intensity for Ag - 40.1 atom % Cu (eutectic). Ar+ bombardment is carried out under the following conditions: Ar pressure of 6.7 X 10' Pa and discharge power of (a, top) 0.5 mA1330 V and (b, bottom) 1.0 mA1390 V.
SPUTTER
1115
TIME
sputtering yield will decrease with the sputter time. A deformation layer, whose composition is different from the bulk composition, will form on the sputtered surface. As shown in Figure 4, after an as-polished sample was subjected to sputtering at 1.0 mA/380 V for 20 min, the surface was again sputtered at 1.0 mA/380 V. Compared to the as-polished surface, the transient period was shortened in the curve of the sputtered surface. This effect implies that Cu-rich altered layers are formed on the sputtered sample by the first sputtering, which is not the case with the as-polished sample. Figure 5 shows plots of the intensity ratio as a function of sputtering time in an Ag-40.1 atom % Cu alloy (eutectic). This sample containing only eutectic-phase grains gives very short transition periods until steady-state conditions are reached. Steady-state conditions would be reached after
LITERATURE CITED Grlmm, W. Naturwissenschaften 1967, 5 4 , 588. Grlmm, W. Spectrochim. Acta, Part 8 1968, 238, 443. Berneron, R.; Charbonnier, J. C. SIA, Surf. Interface Anal. 1981, 3, 134. Waitlevertch, M. E.; Hurwltz, J. D. Appl. Spectrosc. 1976, 30, 510. Belle, C. J.; Johnson, J. D. Appi. Spectrosc. 1973, 2 7 , 118. Takadoum, J.; Pivin, J. C.; Pons-Carbeau, J.; Berneron, R.; Charbonnier, J. C. SIA, Surf. Interface Anal. 1984, 6, 175. Betz, G. Surf. Sci. 1980, 9 2 , 283. Shimizu, R.; Saeki, N. Surf. Scl. 1977, 62, 751. Andersen, H. H.; Bay, H. L. Radiat. Eff. 1973, 19, 63. Anderson, 0. S. J. Appl. Phys. 1989, 40, 2884. Wehner, G. K.; Hajlcek, D. J. J . Appl. Phys. 1971, 42, 1145. Tarng, M. L.; Wehner, G. K. J . Appi. Phys. 1972, 43, 2268. Dahlgrem, S. D.; McClanahan, E. D. J . Appl. Phys. 1972, 43, 1514. Wilson, I. H. Radiat. f f f . 1973, 18, 95. Betz, G.; Wehner, G. K. "Sputtering by Particle Bombardment 11"; Springer-Verlag: Berlin, 1982; Chapter 2. Wagatsuma, K.; Hlrokawa, K. Anal. Chem. 1984, 56, 412. Wagatsuma, K.; Hirokawa, K. Anal. Chem. 1984, 5 6 , 908. Hansen, M. "Constltutlon of Blnary Alloys"; McGraw-Hill: New York, 1958. Laegreid, N.; Wehner, 0. K. J. Appl. Phys. 1961, 32, 365. Oechsner, H. Z . Phys. 1973, 261, 37. Wagatsuma, K.; Hlrokawa, K. SIA , Surf. Interface Anal. 1984, 6, 167. Wagatsuma, K.; Hirokawa, K. Anal. Chem. 1984, 5 6 , 2732. Wagatsuma, K.; Hirokawa, K. Bunko Kenkyu 1984, 3 3 , 320. Boumans, P. W. J. M. "Theory of Spectrochemical Excitation"; Hilger and Watts, Ltd.: London, 1966. Corliss, C. H.; Bozman, W. R. "Experimental Transition Probabilities for Seventy Elements"; Natlonal Bureau of Standard: Washington, DC, 1962, NBS Monograph No. 53. Greene, J. E.; Natarajan, B. R.; Sequeda-Osorio, F. J. Appl. Phys. 1978. 49, 417. Betz, G.; Braun, P.; Farber, W. J. Appl. Phys. 1977, 4 8 , 1404. Wagatsuma, K.; Hlrokawa, K. Anal. Chem. 1984, 5 6 , 2024.
RECEIVED for review October 14,1985. Accepted December 9, 1985.
