Application of secondary neutral mass spectrometry in low-energy sputtering yield measurements

s

Nuclear Instruments

and Methods in Physics Research B 129 (I 997) 123- 129

__ __ Beam Interactions with Materials 6 Atoms

Ii!!

ELSEVIER

Application of secondary neutral mass spectrometry in low-energy sputtering yield measurements S. Bhattacharjee

*,

a, J. Zhang a, V. Shutthanandan a, PK. Ray a* N.R. Shivaparan R.J. Smith b

b,

a Mechanical Engineering Department, Tuskegee Uniuersity, Tuskegee, AL 36088, USA b Physics Department, Montana State University. Bozeman, MT 59717, USA Received

12 November

1996; revised form received 24 February 1997

Abstract An experimental study was initiated to measure low-energy (1.50 to 600 eV) sputtering yields of molybdenum with xenon ions using a Secondary Neutral Mass Spectrometer (SNMS). An ion gun was used to generate the ion beam. The ion current The SNMS spectra obtained at 50” incident angle were density at the target surface was approximately 30 p,A/cm’. converted to sputtering yields for perpendicular incidence by normalizing SNMS spectral data at 500 eV with the yield measured by Rutherford backscattering spectrometry. Sputtering yields as well as the shape of the yield-energy curve obtained in this manner are in reasonable agreement with those measured by other researchers using different techniques. Sputtering yields calculated by using two semi-empirical formulations agree reasonably well with the measured data.

1. Introduction The primary and auxiliary propulsion requirements for many future space missions can be met with ion thrusters [I]. In particular, the xenon ion thruster has been chosen as one of the new breed of propulsion systems for NASA’s New Millennium program [2]. However, a major technology issue of these thrusters is the sputter erosion of components with deposition of sputtered material onto the adjacent structures and subsequent formation of flakes when the sputtered material of sufficient thickness peels off from these structures [3,4]. In these thrusters, the ions are generated in a discharge chamber with energies less than 100 eV. In the accelerator grid region, the sputter erosion is caused by ions having energies less than 300 eV. However, the sputtering yield data at low energies have been found to exhibit considerable variation [5]. In view of this, a systematic investigation has been initiated at Tuskegee University to measure low-energy sputtering yields of various materials using an ion gun. In the first phase of this study, a radioactive tracer method was used to determine sputtering yields [6]. Chromium and cobalt were observed to sputter with argon

and xenon ions having energies as low as 20 and 10 eV respectively [7]. The disadvantage of the radioactive tracer method is that only a few suitable radioisotopes are available for use as tracers. For example, accelerator grids are made of molybdenum which does not have suitable radioisotopes. This prompted us to use a Secondary Neutral Mass Spectrometer (SNMS) as the detector of sputtered materials in the second phase of this study. Since SNMS spectra provide information on differential yields only, they were subsequently converted to total sputtering yields by normalizing the count rate at 500 eV with yields measured by Rutherford Backscattering Spectrometry (RBS). The ion gun is capable of producing ions at energies ranging from 10 eV to 3 keV. However, in the present geometrical arrangement, no data of good statistical quality could be collected below 150 eV. Sputtering yields of molybdenum with xenon ions in the 150 to 600 eV energy range are reported in this paper.

2. Experiment 2.1. Measurement

* Corresponding [email protected]

author.

Fax:

+ l-334-727-8090;

0168-583X/97/$17.00 0 1997 Published Pff SO168-583X(97)00145-6

email:

of SNMS spectra

The experimental set-up is shown schematically in Fig. 1. The experiments were performed in a 22.5 cm diameter spherical vacuum chamber. A 170 l/s turbomolecular pump was used to maintain the vacuum in the chamber.

by Elsevier Science B.V. All rights reserved

S. Bhnttacharjee

124

et ul./Nucl.

Instr. and Meth. in Phys. Res. B 129 (1997) 123-129

UH? Chamber

Gate Valve

Fig. 1. Schematic diagram of the experimental set-up. The base pressure of the chamber was about 2 X 10m9 Torr. A gas flow system maintained a stable xenon ion beam. Xenon gas of 99.999% purity was used in these studies. The mass spectrometer (SPECS Model SSM 2001 can be operated in both SNMS and Secondary Ion Mass Spectrometry (SIMS) modes as well as in the Residual Gas Analysis (RGA) mode. In the SNMS mode, the sputtered neutrals are ionized by electron impact inside the ionizer of the spectrometer. An electric retarding field prevents the low-energy residual gas ions from entering the energy filter. The postionized neutrals are mass analyzed by a quadrupole mass spectrometer and detected by a single channel electron multiplier. During the operation of the spectrometer in the SNMS mode, both positive and negative secondary ions are rejected by the application of appropriate electric fields. The aperture of the mass spectrometer intercepts only a small amount of particles sputtered from the sample. The mass spectrometer was positioned such that the axis of its entrance aperture was perpendicular to that of the ion gun. The center of the target was located 10 mm below the spectrometer aperture. In this geometric arrangement, the aperture of the spectrometer subtended a solid angle of 0.03 sr at the center of the target. The target was 6 mm thick. It was cut from a 12.5 mm diameter rod of 99.95% purity. It was screwed onto a sample holder which was mounted on a XYZB manipulator for precise positioning within the vacuum chamber. During sputtering, the target was placed at a distance of 20 mm from the exit plane of the ion gun. At this position, the ion beam could be focused to an area approximately 1 mm in diameter. It should be noted that the incident beam is not massanalyzed. The operational pressure in the chamber was main-

tained at 1 X 10M6 Torr. At this pressure, the beam current remained essentially constant from 150 eV to 600 eV at 0.23 PA. Based on the I mm spot diameter, the ion current density at the target was approximately 30 PA/cm’. During SNMS measurements, the target was sputter cleaned for about 30 min with a rastered ion beam at 2.5 keV. The ratio of ion current density to ambient gas particle density incident on the target during sputtering was about 500. The SNMS spectra were measured at an incidence angle of 50”. However, the sputtering yields reported by other researchers were measured at normal incidence. To facilitate comparison of data, the area under the peaks measured at 50” incidence angle in our set-up were converted to sputtering yields at normal incidence. It can be shown that the shape of the yield-energy curve at any angle of incidence is essentially similar to that of the peak area versus energy curve of the SNMS spectra measured at any other angle. Let A(@, E) be the area under the elemental peaks of the SNMS spectra at ion energy E and incidence angle B. Let the corresponding sputtering yield be Y(0, E). Then, Y(o,E)

=g(B,E)A(B,E),

(1)

where g is a function whose value depends on 0 and E. Similarly, the sputtering yield at normal incidence Y(O”, El is related to the sputtering yield at any other incidence angle by: Y(@,E)

=f(e,qqe,q.

(2)

When a surface is bombarded with low-energy ions, the angular distribution of emitted particles change slightly with energy. It is assumed in this analysis that in the limited energy range studied (150 to 600 eV), the change in angular distribution of the sputtered materials is not significant. Hence, the functions f and g are assumed to be dependent only on the angle of incidence and not on energy, i.e..

f(e,.q

=j,(e>

Combining

and

g(e,E)

=g,(e).

(3)

Eqs. (I) and (2) with Eq. (3), one gets:

Y(O”,E) =f,(e)g,(e>A(e,E>.

(4)

For SNMS spectra collected at a given angle of incidence, (0 = 50” in our case), f,(50”)g,(50°) = K = constant. Hence, Y(O”.E) = KA(B,E).

(5)

Hence, the total sputtering yield at normal incidence is proportional to number of particles sputtered into a small solid angle leading to the entrance of the SNMS aperture. The value of the constant K was determined by normalizing the peak area with the sputtering yield measured by RBS method at 500 eV.

S. Bhattucharjee et ul./Nucl. Instr. und Meth. in Phys. Res. B I29 (1997) 123-129 to retain the cosine

character

125

of the angular distribution,

the function:

f( 0) = A,cosO + A,cos28 + A,cos36’ + A,cos% has been used to fit our data points. The total sputtering yield was obtained fle) with respect to the solid angle:

(7)

by integrating

Y = jg’2f(H)2rrsinHde 0 =

.r target holder

i

target

2?r

(

A,

A2

A3

A4

2+3+4+7.

(8)

1

Fig. 2. Collector-target assembly for RBS measurement. 3. Results and discussion

3.1. SNMS spectra 2.2. Measurement

of sputtering yield using RBS method

For RBS measurements, the sputtered material was collected on a thin, 12.5 mm wide Al strip. It was mounted on a collector plate which formed a semi-circle of 15 mm radius around the position where the ion beam was focused on the target surface (Fig. 2). A 5 mm diameter hole in the center of the collector plate and the Al foil allowed the passage of the ion beam. The target was bombarded with 500 eV xenon ions for 30 h (over a 5 day period) at normal incidence to deposit enough sputtered material on the Al foil. After sputtering, the Al foil was removed and taken to Montana State University to determine the amount of MO deposited on it. It was analyzed by Rutherford backscattering with 1 MeV He+-ions. The backscattered He+-ions were detected with a silicon surface-barrier detector at a scattering angle of 1.55”. The He+-ion beam was approximately 2 mm in diameter and hence, the points along the Al strip where the measurements were made, were 2 mm apart. The differential sputtering yield Y(f3) was calculated from the following equation:

Each spectrometer scan was performed over a 15 s period. At each energy, data were collected using a sweep rate of 2 amu/s and were accumulated over 10 scans at 150 to 600 eV. It is estimated that the maximum target thickness that was sputtered away during SNMS experiments was less than 2 nm. A full range mass scan revealed the presence of the residual gases as well as small amount of impurities in the target such as Si, Ca, Ti and Zn. The amount of xenon was small indicating effective suppression of residual gas particles by the mass spectrometer energy filters. In spite of raster cleaning the target surface before acquiring the SNMS data, some MOO was always present in the mass spectra. Typical mass spectra at 150,200 and 500 eV are shown in Fig. 3 in the 90 to 120 amu mass range. The MO spectra at 150 and 200 eV are also shown in the inset in this figure

250

R2Nt( 0)q Y(e)

=

IT

’

where R is the radius of the collector strip, Aft(O) is the number density of sputtered atoms per unit area (determined from RBS measurements), q is the charge of an electron, I is the xenon beam current, and T is the total sputter time. In the case of high-energy sputtering (ion energy in the keV region), the sputtered atoms are observed to follow a cosine angular distribution [8]. Thus, the function Acos’W is generally used to fit the high-energy differential sputtering yield data. However, in low-energy sputtering, the angular distribution of the sputtered particles have been observed to be under-cosine 191. The same type of distribution was observed from our RBS measurements. In order

0

90

95

100

105

110

115

120

Mass (amu) Fig. 3. SNMS spectra of MO by 150, 200 and 500 eV xenon ions.

126

S. Bhuttacharjee

et crl./Nucl.

Instr. and Meth. in Phys. Res. B 129 (1997) 123-129

0.05 Ddferential

Fig. 5. Differential 2,111

,110

6W

0.10

Sputtering

sputtering

0 15

Yield (Atoms/lon’Sterad)

yield of MO by 500 eV xenon ions.

8””

Energy (keV) Fig. 4. Typical RBS spectrum of MO deposited on AI foil.

coefficient reality, the than 1 and little higher

of sputtered MO atoms on the Al foil is 1. In sticking coefficient is probably somewhat less hence, the actual sputtering yield should be a than 0.73 atom/ion as reported here.

in the 90 to 102 amu mass range. All seven isotopes can be

3.3.

identified in the MO spectra. Isotopic peaks in these spectra are not well resolved because data were acquired at a relatively low mass resolution to obtain high signal intensities. The intensity of MOO is less than 2.5% of the total intensity at 600 eV and increases monotonically to 6.5% of the total intensity at 150 eV. Even though nearly 6 nm of the target surface was removed during raster cleaning of the target, apparently some MOO either remained on the target or was continuously formed on the surface from the bombardment of residual oxygen atoms inside the vacuum chamber. The increase in the percentage of MOO with decreasing ion energy is probably due to the slight displacement of the focus of the ion beam at lower ion energies.

In the RBS measurements, both MO and MOO were counted, as MOO could not be distinguished from MO by He+-ion backscattering. Hence, areas under both MO and MOO peaks were used in converting the SNMS spectra to sputtering yields. The sputtering yield-energy curve for MO at normal incidence obtained using the procedure outlined in the previous sections is shown in Fig. 6. For comparison, xenon ion yields of MO reported by Rosenberg and Wehner [lo] and by Weijsenfeld et al. [l l] are also plotted in the same figure. In both of these measurements, spherical targets were immersed in a low-pressure, high-density

Sputtering

yield

I

3.2. RBS

.

I

8

I

,

,

spectra I

The RBS measurements indicated that measurable amount of sputtered material was deposited on the Al foil. The number of monolayers deposited along the Al strip range from 3 to 12. Fig. 4 shows a typical RBS spectrum taken at 75” target ejection angle with respect to the direction of the xenon ion beam. At this position the thickness of the sputtered MO was in excess of 6 monolayers. The differential sputtering yield obtained from RBS measurements is presented in Fig. 5. It is clear from this figure that the angular distribution of ejected particles is under-cosine. The maximum of the differential sputtering yield occurs at 45”. The total sputtering yield has been found to be 0.73 atom/ion by integrating the differential sputtering yield over all solid angles. In this calculation, it has been assumed that the sticking

31

-*-Present --0

---o--.Rosenberg

d

I

100

Work

Weqsenfeld

200

300

I

400

er al and Wehner

1

I

500

600

101

Energy (eV) Fig. 6. Sputtering

yield of MO as a function of xenon ion energy.

S. Bhattacharjee

et ul./Nucl.

Instr. and Meth. in Phys. Res. B 129 (1997) 123-129

plasma and the sputtering yields were determined from the weight loss of the target. It can be seen that the yields obtained by SNMS measurements fall in the range of yields measured by others. However, the shape of the yield-energy curve is observed to be slightly different. Particularly, beyond 300 eV ion energy, the slope of the yield-energy curve is found to be smaller. In a more recent experiment, thin films of molybdenum were sputtered by ions of xenon and other noble gases [12]. In the low-energy end, with xenon ions, sputtering yields were found to be 0.8 at 200 eV and 1.6 at 500 eV. Since these values are considerably higher than those obtained by other researchers, these values are not shown in Fig. 6.

4. Comparison

with theoretical

predictions

The most applicable theory of sputtering has been developed by Sigmund [ 13,141. This theory predicts good agreement with measured sputtering yields in medium to high energy sputtering cases. It is based on a nuclear energy loss mechanism in which the energy loss is shared among the large number of atoms which define the collision cascade. According to this theory, the sputtering yield Y by ions of energy E at normal incidence is given by the following equation: Y(E)

0.042 = raS,(E),

(9)

0

where U. is the binding energy of the target atoms in eV (for metals the sublimation energy is used), LYis a function which depends only on the target atom to ion mass ratio, M/M,, and S,(E)is the energy dependent nuclear stopping cross section. The value of (Y is approximated as [ 121: cy = 0.15 + O.l3M,/M,,

comparing the theory with sputtering yield data even though it predicted higher sputtering yields [14]. A more recent analytical expression for S,(E) was developed by Matsunami et al, [ 161: 3.44l&ln(e+2.718) %(‘)

= 1 + 6.35&

(14) where R,/R depends only on the mass ratio M,/M, and Et,, is the threshold energy of the specific ion-target combination. These two functions are expressed as:

R 1 .-I= R 0.4(M,/M,) + 1

(zf”

and

(16)

+

z;/‘)“’

Here S,(E) is the reduced elastic cross-section based on reduced energy E, and Z, and Z, are atomic numbers of the ion and target respectively. The value of l is given by [13,15]: 0.03255 =

v3 +zy -v*(z1

factor proposed by YMI is given by [ 181:

(12)

Here Q is a constant which depends only on the target material and Et,, is the threshold energy which is written as [16]: .,,=,[,.9+3.8(~)~i+0.134(~)‘~z1].

(18)

The reduced Lindhard inelastic stopping cross-section is given by [ 19,201:

s,( e>

1

There are several equations that one can use in describing the function S,(E). Sigmund used an expression based on the Thomas-Fermi model of atomic interaction in

(17)

.

Q[

(11)

l

(15)

analytical

z,z2

84.78

(13)

’

Sigmund’s theory can not be directly applied to calculate sputtering yields at low ion energies. In this energy range sputtering takes place by the knock-on process rather than through production of a collision cascade. However, we have calculated sputtering yields at low energies using Sigmund’s formula to compare with yields predicted by theoretical formulas provided by other researchers. For this calculation, we have used values of sn( E) expressed by Eq. (13). To determine sputtering yields more accurately at low ion energies, both Bodhansky [17] and Yamamura et al. (henceforth referred to as YMI) [18] proposed semi-empirical correction factors to Sigmund’s formula given in Eq. (9). The correction factor suggested by Bodhansky is [17]:

The correction

=

+ c( - 1.708 + 6.8826)

(10)

whereas S,(E) is described by the following function developed by Lindhard et al. [13,15]: S,(E)

127

z2/3z1/2

O.O79(M, + M2)3'2 Se(E)

=

,,,f3/2M;/2

( zy

,

z

+

z;/‘)3v4

E 1’2.

(19)

S. Bhattuchurjee

128

et ul./Nucl.

Instr.

and Mcth.

in Phys. Res. B 129 (1997)

123-129

1

0

0011

I 100

I

200

’ 300

’

400

I 500

’ 600

I

001

Energy (eV)

1,

I

100

200

Fig. 7. Comparison of theoretical sputtering yields with experimental data.

1

I.

300

I

400

’

I-

500

I

600

Energy (eV)

Fig. 8. Normalized theoretical yields using Bodhansky and YMI formulations. Bodhansky recommended scribe (Y [ 171:

the following

formula

(Y= 0.3( M*/MJ’? whereas the expression al. [20] is:

to de-

(20) proposed for (Y by Matsunami

u = 0.08 + 0.164( M2/M,)o’J

et 5. Conclusions

+ 0.0145(M,/M,)“29. (21)

The sputtering yields of molybdenum by xenon ions calculated by using the three formulations are shown in Fig. 7. The values of the parameters used in these calculations are given in Table 1. The values of (Y are nearly identical in all three formulations. Hence, the changes in the shape of the theoretical yield-energy curves are due to Eqs. ( 14) and ( 17) only. As expected, the yields obtained using Sigmund’s formula are high. Both Bodhansky’s and YMI’s formulations predict the sputtering data reasonably well given the uncertainty in the sputtering data themselves. Between 200 and 600 eV, sputtering yields predicted by Bodhansky is about 55% higher than those predicted by YMI, whereas at 100 eV the two values are nearly same. However, both theories provide the same

Table I Values of parameters used in the theoretical calculations Parameter

Bodhansky’s formulation

YMI’s formulations

0.24

0.23 0.84 [201

(R,/R)

0.77

(Jo

6.82

‘%h

61.8

eV eV

shape in the yield-energy curves as can be seen from Fig. 8. where the two curves are normalized at 600 eV.

6.82 48.9

eV eV

[21]

Sputtering yields of MO by xenon ions in the 150 to 600 eV energy range were measured using a SNMS. The SNMS spectra were obtained at an angle of incidence of SO”. They were converted to sputtering yields for perpendicular incidence by normalizing SNMS spectral data at 500 eV with the yield measured by Rutherford backscattering spectrometry. The values of sputtering yields obtained in this manner are in the range of those measured by other researchers. The shape of the yield-energy curve is observed to have a smaller slope at energies above 300 eV than reported previously. Sputtering yields were calculated by using formulations provided by Sigmund, Bodhansky and YMI. The two later formulations are applicable at low ion energies. They are semi-empirical in nature and extensions of the method outlined by Sigmund. The yields calculated from these two methods agree well with the measured values whereas those calculated from Sigmund’s equation provide higher yields.

Acknowledgements This research was supported through a NASA research grant NAG3-1388. The work at Montana State University was supported through NASA EPSCOR Grant NCCW0058.

S. Bhurtacharjre

ef al./Nucl.

Instr. and Meth. in Phys. Res. B 129 (19971 123-129

References [I] L. Carvey and R. Vondra. Paper No. AIAA-90-2621 (1990). [2] S. Ashley, Mechanic. Eng. I I7 (1995) 61. [I] J.R. Beattie and J.N. Matossian, Report No. NASACR 174974 (1989). [J] V.K. Rawlin, Paper No. AIAA-88-2912 (1988). [5] H.H. Andersen and H.L. Bay, in: Sputtering by Particle Bombardment I, ed. R. Behrisch (Springer, Berlin, 1981) p. 193-206. [h] A.K. Handoo, P.K. Ray, Appl. Phys. A 54 (1992) 92. [7] A.K. Handoo, P.K. Ray, Can. J. Phys. 71 (1993) 155. [Y] T. Okutani. M. Shikata, S. Ichimura. R. Shimizu, J. Appl. Phys. 5 I (1980) 2884. [9] G.K. Wehner. D. Rosenberg. J. Appl. Phys. 3 I (1960) 177. [lo] D. Rosenberg, G.K. Wehner, J. Appl. Phys. 33 (1962) 1842. [I I] C.H. Weijscnfeld. A. Hoogendoom, M. Koedam, Physica 27 (1961) 763. [I?] P.C. Zalm, J. Appl. Phys. 54 (1983) 2660.

129

[I31 P. Sigmund, Phys. Rev. 184 (1969) 383. [14] P. Sigmund, in: Sputtering by Particle Bombardment 1, ed. R. Behrisch (Springer, Berlin, 1981), chap. 2. [I51 J. Lindhard, V. Nielsen, M. Scharff, Dan. Vid. Selsk. Mat. Medd. 36 ( 1968) 10. [I61 N. Matsunami, Y. Yamamuura, Y. Itakawa, N. Iroh, Y. Kazumata, S. Miyagawa, K. Motita, R. Shimizu. Radiat. Eff. Lett. 57 (1980) 15. [I71 J. Bodhansky, Nucl. Instr. and Meth. B 2 (1984) 587. [I81 Y. Yamamuura. N. Matsunami, N. Itoh. Radiat. Eff. 71 (1983) 65. [I91 J. Lindhard, M. Scharff, H.E. Schiott, Mat.-Fys. Medd. Dank. Vid. Selsk. 33 (1963) 14. [20] N. Matsunami, Y. Yamamuura, Y. Itakawa, N. Itoh, Y. Kazumata. S. Miyagawa, K. Morita, R. Shim& and H. Tawara, AI. Data Nucl. Data Tab. 31 (1984) I. [21] C. Kittel, Introduction to Solid State Physics (Wiley. New York, 1976) p. 74.