Secondary ion mass spectrometric ion yields and detection limits of

detection limits for secondary Ion mass spectrometric analysis of electrically important Impurities in InP. To evaluate the ion yields and the detecti...
0 downloads 0 Views 522KB Size
sa

Anal. Chem. 1988, 60,58-61

Secondary Ion Mass Spectrometric Ion Yields and Detection Limits of Impurities in Indium Phosphide Tohru Tanaka,* Yoshikazu Homma, and Satoru Kurosawa NTT Applied Electronics Laboratories, Musashino-shi, Tokyo 180, Japan

This paper presents a detaHed look at the ion yields and the detection limits for secondary Ion mass spectrometrlc analysis of elecbtcayy hnportant ImpurMes in InP. To evaluate the ion yleMs and the detection limits, highly doped InP crystals have been prepared and analyzed by chemical analysis as standard samples for secondary ion mass spectrometric analysis. Detection of molecular ions M e a d of atomic ions was found to be useful for sensitlve quantitative analysis of impurltles In InP. I n particular, for Ge, the negative compound ion combind wlth P (MP-) has an lon yield about 50 thes larger and a detection ihnR 50 times lower than Rs atomic ions. Also, for other elements, these compound Ions were found to have rather large lon ylekls. The secondary ion mass spectrometric - ~ detection limits are lowered to 1012-1015atoms ~ m levels.

Table 1. Chemical Analysis" of Impurities in Doped InP Crystals detection limit, atoms dopant

result, atoms cm-3

Asb

4.5 x 1020 5.5 x 10'6 2.7 X 10l8 2.9 X J0l8 9.3 x 10'8 3.5 x 10'8 1.6 x 1019 2.1 x 1019 1.1 x 10'8 1.3 X 10ls

Fe Ga

Geb S=

Sb Seb Si Snb Te

cmd 3.7 x 8.3 x 1.4 X 6.5 X 2.0 x 3.8 X

7.3 x 1.1 x 6.3 x 2.2 x

1015 1015 10l7 10l6 10'7

10l6 1014 1015 1015 1017

OThese are analyzed by ICP analysis. bHydride formation technique is used. CThisis analyzed by colorimetry. The quality of an indium phosphide (InP) undoped crystal must be closely examined when using InP single crystals for such InP-based devices as InGaAsP-InP lasers or InP field effect transisters (FET). During the past decade the net carrier concentration of an undoped InP single crystal has seen a steady reduction from 10'' to 1015 cm-3 (1). Although secondary ion mass spectrometry (SIMS) can successfully provide analytical data from the lox5atoms cm-3 level, SIMS cannot yield quantitative data without employing standard samples. To achieve SIMS quantitative analysis of impurities in semiconductors, ion-implanted samples (2,3!or heavily impurity-doped samples whose impurity concentrations are determined by chemical analysis ( 4 ) have been used as calibration standards. Heavily doped GaAs crystals have been prepared as standard samples for SIMS quantitative analysis of impurities in previous work ( 4 , 5). Leta and Morrison have extensively reported the ion yields and the detection limits for impurities in semiconductors by using the ion-implanted standards (6). However, their work lacks data of electrically important elements and isoelectronic elements. In InP, residual impurities such as Fe, Ge, S, Se, Si, Sn, and Te are thought to play an electrical role in semiinsulation, and isoelectrical impurities such as As, Ga, and Sb are thought to be effective for reducing dislocation density. The ion yield data and the detection limit data for these impurity elements are needed to examine the quality of undoped InP crystals. The elements such as Ge, which have relatively high ionization potentials and low electron affinities, do not easily form positive and negative atomic ions. In the impurity analysis for GaAs, the ion yields have been investigated in detail for molecular as well as atomic ions of impurities (7). These molecular ions of the elements in groups I11 and IV (groups 13 and 14 in 1985 notation) in the periodic table were found to have relatively high ion yields. Also, in the impurity analysis of InP, use of the molecular ion is expected to reduce the detection limits. In this work, we studied a highly sensitive method of using SIMS for analyzing Fe, Ge, S, Se, Si, Sn, Te, Ga, As, and Sb 0003-2700/88/0360-0058$01.50/0

in InP. The quantitation of SIMS was established by using highly doped InP crystals as standard samples analyzed mainly by inductively coupled plasma emission spectrometry (ICP-ES). The accuracy of the SIMS analysis was carefully investigated and the detection limits were lowered by choosing the optimum ion detection conditions. In particular, the ion yields of molecular ions were investigated for each impurity.

EXPERIMENTAL SECTION Standard Samples. All highly doped InP crystals were grown by the liquid encapsulated Czochralski (LEC) technique (8). As, Fe, Ga, Ge, S, Sb, Se, Si, Sn, and Te were separately doped to more than 10l8 atoms cm-3 levels. The doping level for Fe was 10l6atoms ~ m - These ~ . doped crystals were cut into two l-mmthick pieces having a 10 X 10 mm area for chemical and SIMS analyses. Quantitative analyses of these highly doped samples except for the S-doped sample were performed by ICP-ES with an SPS-1100 spectrometer. The hydride formation technique (9) was used for ICP analysis of As, Ge, Sb, and Se. The typical quantitative analysis procedure employed is as follows: 0.1 g of InP crystal was dissolved with 2 mL of an acid mixture (HC1/ HN03/H20 = 6/1/6), diluted with ultrapure water to a 1% solution, and then analyzed by ICP-ES. S analysis was carried out by colorimetry with the methylene blue method (10). The matrix elements were intentionally added to all standard solutions to control the analytical condition for sensitivity. For each impurity element, at least five standard solutions were prepared, which contained 1%InP and the known concentrations of the elements to be analyzed. The impurity concentrations in highly doped crystals obtained by chemical analysis are summarized in Table I. These concentration values were utilized in the calculations of ion yields and detection limits in SIMS analysis. In this work, the calculation of the SIMS detection limit was based on estimation of background levels in measuring the SIMS intensity in the highly pure crystals. For this purpose, undoped InP crystals were also prepared by the LEC method and were cut into chips. SIMS Analysis. SIMS analyses were performed by using a of 10.5 keV having a beam current of CAMECA IMS-3F. 02+ 1.5 pA was used as a primary ion during positive ion detection. Cs+ of 14.5 keV having a beam current of 0.2 pA was also utilized 1987 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 00, NO. 1, JANUARY 1, 1988

Table 111. Relative Ion Yields of Negative Secondary Ions in Indium Phosphide (IW/CM x Cln/Iln-or IMx-/CM x

Table 11. Relative Ion Yields of Positive Ions in Indium Phosphide (I,+/CM X Cp/Ip+ or Im+/Cm X Cp/Ip+)(l element (M) As

Fe Ga Ge S

Sb Se Si

Sn Te a

M+ 1.6 2.8 X 1.4 X 1.4 X 1.7 1.5 1.6 5.1 X 7.2 X 4.7

lo2 lo3 10

10 10

MO+

MP+

4.2 X lo-' 3.0 X 10 4.8 X lo-' 7.1 X lo-' 1.4 1.6 X 10-1

6.8 X lo-' 1.0 X 10 3.3 5.4 X lo-' 4.9 3.0 X lo-' 1.7 3.0 X lo-' 5.7 1.9

2.6 1.1

MIn+ 7.2 3.3 X 1.7 X 1.1 X 1.2 x 7.5 3.5 3.0 2.2 x

59

CIn/IIn-)a

element (M)

lo2

4.3 x 1.0 x 3.0 6.6 X 4.1 x 5.0 X 2.3 x 1.2 x 6.1 X 2.9 x

As

Fe

lo2 10 10

Ga Ge S

Sb Se Si

10

Sn Te

CM is corrected for natural abundance.

M-

MP5.8 X 1.2 X 3.4 X 2.8 X 1.7 x 7.8 x 2.6 x 4.1 x 1.8 X

10 10

10 103 10 103 io2

10 103

MIn-

10

lo2

lo2 lo3 103 10 102

103 lo3

8.0 X 6.9 X 4.6 8.1 4.0 x 1.0 x

10 10

103 102 1.1 x 103 7.3 1.4 X 10 1.3 x 103

CM and CI,are corrected for natural abundances. for detecting negative ions. The typical beam diameters were 70 I.tm in both cases. The primary ion beam was rastered over an area of 120 X 120 km and the secondary ions emitted were detected from the 60-wm-diameter area in the center of the rastered area.

RESULTS AND DISCUSSION Ion Yields. Relative ion yields are useful for understanding the ionization mechanism of each element because other factors such as the instrument constant and the sputtering yield are already canceled. The relative ion yields were determined by using the ratio of secondary ion intensity to the concentration of the doped InP samples. The ion intensities were corrected for natural isotopic abundances in this procedure. The relative positive ion yields normalized to the P+yield are summarized in Table 11. An almost linear relationship can be obtained between atomic ion (M+)yields and ionization potentials. The larger the M+ ion yields, the smaller the ionization potentials. Ga has the highest M+ ion yield of all elements and the smallest ionization potential of all. For molecular ions, that is, oxide ions (MO') and compound ions containing one of the matrix elements, the ion yields are calculated by the same treatment and are given in the table. Although the MO+ of Fe has a rather large ion yield, the MO+ ion yields of other elements are small. Some examples of oxide ions with higher ion yields can be seen in the analyses of impurities and constituents in steels and Si (11,12). However, it seems difficult for the elements analyzed here to form oxides. Also, the ion yields of positive compound ions with P are very small. The positive compound ions did not exhibit any great advantage in all cases of positive analyses. This leads to the general conclusion that the positive molecular ions studied here did not give higher SIMS sensitivity than the atomic ions. However, among these ions, the positive compound ions with In are found to have large ion yields, which are comparable to those of their positive atomic ions for most elements. For As, Fe, S, Sb, and Se, their ion yields slightly exceed those of positive atomic ions. The relative negative ion yields are given in Table I11 together with those of the compound ions composed of one of the matrix elements, using the In- yield for normalization. The measurements were carried out under Cs+ ion bombardment. Negative atomic ion yields are enhanced by Cs (13). Large negative atomic ion yields are observed for S, Se, and Te. Their large electron affinities (ca. 2.0 eV) exert a great influence upon ion stabilities. On the whole, negative compound ions have larger ion yields than negative atomic ions. Compound ions are expected to give highly sensitive analysis. In particular, the compound ions that combine with P (MP-) often have larger ion yields. The elements in groups IV, V, and VI (groups 14,15, and 16) studied here can be regarded as having covalent bonds with P. These suggest that the covalent bonds may contribute

M- v M P - mMill 10'. 0

v

-I

w*

v

.*

v

m.

10'.

0 w

v

I

2-I toz.

Z.

w

a 10 .

.

Go Ill

Figure 1.

... .=

v

m 9

Si GeSn IV

AsSb

v ELEMENT

S SeTe VI

Relative negative ion yields for impurities in InP crystals.

to the stable compound ion formation. With regard to impurity analysis in G&, the MAS- ion yields for B, C, and Si are 1-3 orders of magnitude larger than those of atomic ions (7). These are thought to form covalent bonds with As. A similar circumstance also exists for the InP results. In Figure 1, the negative ion yields are plotted for group 111, IV, V, and VI (group 13, 14,15, and 16) elements in the periodic table. The M P ion yields are large in even-numbered groups and small in odd-numbered groups. The MP- ion yields in group IV (group 14) are at least 1order of magnitude larger than atomic ion yields. The elements in group VI (group 16) also have large ion yields. The ion yield difference between groups can be explained with the valence numbers of elements (14). Because P has stable valence numbers of r 3 and r 5 , it is reasonable that group IV (group 14) elements with stable valence numbers of f 4 have stable singly charged negative ions with P. For group VI (group 16) elements having stable valence numbers of f 2 , f4, and f6, singly charged negative compound ions combined with P are thought to be stable. The M p ion yields are thought to have a relation with the valence of elements. The compound ions containing In (MIn-) render a simple relationship between ion yields and element group, where the larger the group number is, the larger the MIn- ion yields are. This indicates that electronegativities of elements seem to be the key of ion stabilities. The MP- ion yields for the elements in the group IV (group 14) elements in InP were found to be extremely large in the same way as MAS- ion yields for several elements in GaAs. This fact suggests that the covalent bond is important for understanding of molecular ion formation in InP and GaAs (7).

80

ANALYTICAL CHEMISTRY, VOL. 60, NO. 1, JANUARY 1, 1988

Table IV. Detection Limits of Positive Analysis Impurities in Indium Phosphide

element

ion

mass

sensitivibackgroundD ty factor, intensity atoms ~rn-~ ratio

As

As As0

75 91 106 190 56 72 87 171 69 85 100 184 70 86 101 185 34 50 65 149 121 137 152 236 82 98 111 197 28 44 59 143 120 136 151 235 130 142 161 245

1.2 x 4.7 x 2.9 X 2.8 X 7.7 x 1.3 X 4.7 x 2.1 x 1.5 x 6.8 X 1.1x 1.9 x 6.9 X 1.4 x 1.8 x 9.0 x 2.8 x 3.3 x 9.5 x 3.7 x 2.3 X 2.1 x 1.2 x 4.6 X 1.4 X 1.5 x 2.3 X 6.1 X 4.2 X 8.2 X 7.1 X 7.1 X 1.1 x 5.9 x 1.4 X 3.7 x 1.2 x 9.7 x 3.0 X 5.0 x

Fe

Ga

ASP AsIn Fe FeO FeP FeIn Ga GaO GaP GaIn

Ge

Ge

S

GeO GeP GeIn S

so SP

Sb

Se

Si

Sn

Te

SIn Sb SbO SbP SbIn Se SeO SeP SeIn Si Si0 Sip SiIn Sn SnO SnP SnIn Te TeO TeP TeIn

102 1022

loz2 loz1 1019 loz2 1021 1020

1019 loz2 1022 1020 loz1 1023 1023 1021 1023 1023 1022 10'8 loz2 1023 1023 loz1 loz2 1024

loz2 loz2 lo2" loz1

loz2 loz1 1021 1020 IOz2 102' 1022 1022

loz2 1017

6.7 x 1.6 x 2.8 x 2.0 x 3.9 x 3.7 x 1.7 X 1.8 x 4.9 x 1.6 X 1.6 X 5.7 x 7.1 X 4.1 X 1.9 x 9.4 x 3.6 x 1.7 x 2.4 x 2.8 4.8 X 3.9 x 5.2 X 3.7 x 5.2 X 5.2 X 3.1 x 3.2 x 9.4 x 2.0 x 2.6 X 1.9 x 9.7 x 2.6 x 1.1 x 3.5 x 3.6 x 1.7 x 2.4 x 2.8

10-5 10-5 10-5 10-4 104 10" lo4 10-5 10-4 lo4 10" 10-5 lo4 10" 10+ 104 10-5 10-5 10-5

lo4 104 10" 10"

lo4

10" 10-5 10-5 10-5 10-5 10" 10-5 10-5 10-5 10-4 10-5 10-5 10-5 10-5

detection limit, atoms 8x 8x 8X 6X 3x 5x

Table V. Detection Limits in Negative Analysis of Impurities in Indium Phosphide

element

8x 4x 8X 1x 2x 1x 5x 6X 3x 9x 1x 6X 2x 1x 1x

10'1 10'1 10'l 10'l 1014 1016 1015 1015 1015 10'7 1016 1016 1016 1017

As

As

Fe

ASP AsIn Fe FeP FeIn

Ga

Ga

GaP GaIn Ge

These background intensity ratios are twice the intensity ratios of impurity ions to 31P+in the undoped sample. a

Detection Limit. The detection limit was calculated as a product of the sensitivity factor and the background intensity ratio by modifying Clegg's method (15). The background intensity ratio replaces twice the intensity ratio measured in the undoped sample. The intensity ratio was obtained by dividing the intensity at the mass attributed to the impurity ion by the ion intensity of the matrix ions such as P+ and Il3In-. The sensitivity factor, which is the reciprocal of the sensitivity, was the intensity ratio in the standard sample divided by the impurity concentration. A discussion will be given in the next section concerning the reproducibility of the sensitivity factor. Detection limit measurement involved rastering of the primary ion beam to an area of 120 x 120 pm, which is a small area compared with usual rastering area of 250 X 250 pm. Reducing the rastering area leads to an increase in current density of primary ions. As the ion current density becomes larger, the secondary ion intensity of the impurity becomes larger if the sample contains some impurities. However, background intensity is independent of the increase in primary current density, that is, reduction of the rastered area. The background intensity ratio of the elements was obtained by adopting the minimum intensity ratio measured for several undoped InP crystals, where the background intensities

Ge

GeP GeIn

S

1017

10'6 1019 10l8 10'8 1019 1017 8 x 1017 6 X loll 2 x 10'6 7 x 10'7 8 x 1018 7 x 10'7 2 x 10'8 4 x 1016 2 x 1017 2 x 10'1 1 x 1017 1 x 1017 2 x 1016 2 x 1018 1 x 1017 4 x 1017 2 x 1018 7 x 1017 1 x 1018

ion

Sb Se Si Sn

Te

a

S

SP SIn Sb SbP SbIn Se SeP SeIn Si Sip SiIn Sn SnP SnIn Te TeP TeIn

mass

sensitivity factor, atoms cm-3

75 106 190 56 87 171 69 100 184 70 101 185 34 65 149 121 152 236 82 111 197 28 59 143 120 151 235 130 161 245

2.1 x 1.4 x 9.1 x 7.7 x 7.4 x 1.2 x 5.1 X 4.2 X 2.6 X 6.2 X 1.3 X 1.9 x 4.8 X 1.1 x 4.2 X 3.3 x 1.4 x 1.3 x 4.0 X 6.5 X 8.1 X 7.9 x 6.3 X 1.3 X 4.2 x 2.1 x 3.1 X 7.7 x 1.4 X 2.0 x

backgroundn intensity ratio

1.9 x 2.2 x 1.1 x 2.6 x 2.7 x 1.7 x 7.8 x 1.1 x 3.9 x 3.0 x 1OI8 4.6 x 1021 4.8 x 10l8 4.2 x 1019 9.3 x 1Ol8 9.6 x 1019 1.2 x 1019 2.1 x 1019 7.8 x 10l8 1.3 x 10l8 3.1 x 10l8 3.7 x 1018 1.1 x 10l8 1.6 x lozo 4.6 x 1019 1.7 x 1018 6.2 x lozo 1.3 x 1017 8.6 X loz1 2.2 x 10'8 2.0 x

1019 1019 10'8 1019 10'8 1019 1020 10l8 lozo lom

10-3 10-4 10-3 10-4 10-4 10-3 10-5 10-4 10-5 10-5 10-5 10-5 10-5

lo4

10-4 10-4 10-5 10-4 10-5 10-4 10-4 10-5 10-3 10-4 10-4 10-4 10-4 10-4 10-4

detection limit, atoms cm-3 4 x 3x 1x 2 x 2 x 2x 4x 5x 1x 2x 6X 9x 2x 1x 4x 4x 3x 1x 5x 2 x 3x 9x 1x 6X 7x 1x 4x 7x 3x 4x

10'6 1015 1016 10'6 1015 10'6 10'6 1014 1016

1015 1013 10'6 1015 10'6 1016 1015 1014 1016 1013 1015 1015 1013

1016 10l6 1015 1015 10'6 1012 1017 1014

These background intensity ratios are twice the intensity ratios

of impurity ions to ll3In- in the undoped sample.

do not vary in accordance with the change in the rastered area. The detection limits in positive analysis are summarized in Table N. The sensitivity fador corresponds to a reciprocal of ion yield since ion yield was calculated from the impurity ion intensities divided by the impurity concentrations. The smaller the sensitivity factor is, the larger the ion yield. For Fe and Ga having large positive ion yields, the detection limits are expected to be low because the sensitivity factors are small. The detection limit of 3 X l O I 4 atoms cm-3 is obtained by using the positive atomic ion of Fe. Though the positive atomic ion of Ga was indicated as being sensitive, the lowest detection limit was actually obtained in the negative MP- ion because of its lower ion background ratio (Table V). The other elements exhibit rather high detection limits both in positive atomic ions and in positive compound ions. This is thought to be attributed to their low sensitivity corresponding to difficulty in forming positive ions. The detection limits of negative analysis are also calculated in the same way as the positive analysis and summarized in Table V. S, Se, and Te were found to exhibit relatively low detection limits of their negative atomic ions (M-). This corresponds to the ease in forming negative ions due to their high electron affinities. Te, which did not suffer contamination, exhibited extremely low detection limits of around lOI3 atoms ~ m - This ~. level seems to be the lower limit of the mass spectrometry detection system. Se is also not affected by contamination, but negative atomic ions of 80Seand '%e having large natural abusdances showed relatively high detection limits. These ions were thought to be superimposed by some phosphide ions, for examples, P,OH,- (80) and PzO- (78) in mass spectra. Se shows the lowest detection limit by detecting @%e-. Because the 32S-intensity depended not merely on the contamination

ANALYTICAL CHEMISTRY, VOL. 60, NO. 1, JANUARY 1, 1988

Table VI. Analysis of As-Doped InP" run

result,

1 2 3

lomatoms

4.4 4.5 4.1

error, %

2.2 0 4.4

OMean value is 4.5X1020 atoms ~ r n - ~RSD . is 3.5% a n d m a x i mum error is 4.4%.

level present in the SIMS instrument but also on the spectrometrical interference of 02(32), detecting instead of 32S-achieved a detection limit of 2 X 1015 atoms ~ m - ~ . Si, Ge, and Sn do not easily form positive and negative atomic ions due to their relatively high ionization potentials and low electron affinities. Since M p ions of Ge and Sn were found to have large ion yields as shown before, low detection limits were obtained for their ions. Negative atomic ions of 74Gewith the largest natural abundance had a somewhat high background level owing to some phosphide compound ions, perhaps P2C- (74). The lowest background level of the MPreduces the analysis level of Ge to 6 X lOI3 atoms ~ m - ~Sn's . detection limit also can be lowered by using MP- instead of M-. However, Si in the same group has the lowest detection limit of negative atomic ion detection. Also for As and Sb, the MP- ions were found to exhibit lower detection limits than other ions. The detection limits of the MP- are 1 order of magnitude lower than those of the atomic ions in As and Sb analyses. SIMS Analysis Accuracy. The SIMS analysis accuracy is symbiotic with the accuracy of the sensitivity factor because the fluctuation in the SIMS intensity is very small, ca. +2% in coefficient of variation. The accuracy of the sensitivity factor was mainly affected by the accuracy of the impurity concentration in standard samples obtained by chemical analyses. In this work, the accuracy of the impurity concentration in the standard samples has been studies in detail. The accuracy of the standard samples must be estimated from the accuracy of chemical analysis in interlaboratory as well as in intralaboratory experiments. The intralaboratory accuracy of the standard samples was examined through three analysis repetitions and was expressed by using standard deviations. Generally, the accuracy becomes low near the detection limit levels. Table I shows the detection limits of chemical analysis used in this work as well as the concentrations for all elements. The concentration obtained here is at least 1 order of magnitude larger than their detection limits of chemical analysis for each element. Accordingly, the intralaboratory accuracy of chemical analysis is rather high

01

in these analytical region. One analysis example of As is given in Table VI. The standard deviation and maximum deviations from the mean value were only 3.5% and 4.4%, respectively. The intralaboratory accuracy of within *5% was verified in the analyses of the other impurity elements. A round-robin study on the analyses of impurities in GaAs (5) was used to estimate the interlaboratory accuracy of chemical analysis. The participating laboratories studied the chemical analyses and obtained a conclusion that the interlaboratory accuracies of chemical analyses are within *30%. The relatively high values seem to be due to systematic errors in preparing standard solutions for quantitative analyses. Judging from the results of both intralaboratory and interlaboratory experiments, it is acceptable that the accuracy of InP standard samples is at worst within *30%. Accordingly, the SIMS analysis accuracy can be determined by the accuracy of the standard sample and is estimated to be from k5% to &30%.

ACKNOWLEDGMENT The authors thank Y. Ishii and N. Honma for their helpful discussion. We also thank E. Kubota for the samples provided. Registry No. As, 7440-38-2;Fe, 7439-89-6;Ga, 7440-55-3;Ge, 7440-56-4; S, 7704-34-9;Sb, 7440-36-0; Se, 7782-49-2;Si, 7440-21-3; Sn, 7440-31-5; Te, 13494-80-9; InP, 22398-80-7.

LITERATURE CITED Cockayne, B.; Macewan, W. R.; Brown, G. T. J. Mater. Sci. 1980, 15, 2785. Hornma, Y.; Ishii. Y. J . Appl. fhys. 1985, 5 7 , 2931. Ramseyer, 0. 0.; Colton, R. J. J. Vac. Sci. Technol., A 1985, A 3 , 1356. Kurosawa, S.; Homrna, Y.; Tanaka, T.; Yamawaki, M. I n Proceedings of the Fourth International Conference on Secondary Ion Mass Spectrometty S I M S - I V ; Benninghoven, A., Okano, J., Shimizu. R., Werner, H. W., Eds.; Springer: Berlin, 1984; pp 107-109. Homrna, Y.; Kurosawa, S.; Yoshoka, Y.; Shibata, M.; Nomura, K.; Nakamura, Y. Anal. Chem. 1985, 5 7 , 2928. Leta, D. P.; Morrison, G. H. Anal. Chem. 1980, 52, 514. Homrna, Y.; Tanaka, T. Anal. Chem. 1988, 5 8 , 1108. Bachmann, K. J.; Buehkr, E.; Miller, B. I.; McFee, J. H.; Thiei, F. A. J. Cryst. Growth 1977, 3 9 , 137. Thompson, M.; Pahiavanpour, B.; Walton, S. J. Analyst, (London) 1978, 103, 568. Yokosuka. S.; Shirakawa, E. Bunseki Kagaku 1958, 7 , 368. Morgan, A. E.; Werner, H. W. Anal. Chem. 1976, 48, 699. Wittmaack, K. I n Inelastic Ion-Surface Collisions; Tolk, N. H., Tuliy, J. C.. Heiland, W., White, C. W., Eds.; Academic: New York, 1977; pp 153- 199. Storms, H. A.; Brown, K. F.; Stein, J. D. Anal. Chem. 1977, 4 9 , 2023. Piog, C.; Wiedrnann, L.; Bennlnghoven, A. Surf. Sci. 1977, 6 7 , 565. Clegg, J. B. S I A , Surf. Interface Anal. 1980, 2 , 91.

RECEIVED for review December 15,1986. Accepted September 4, 1987.