- Email: [email protected]
Surface contamination detection below the ppb range on silicon wafers
Journal of Crystal Growth 89 (1988) 39—42 North-Holland, Amsterdam
39
SURFACE CONTAMINATION DETECTION BELOW THE ppb RANGE ON SILICON WAFERS A. CORRADI DNS
—
“,
M. DOMENICI and A. GUAGLIO
Dynamit Nobel Silicon, Viale Gherzi 31, 1-28100 Novara, Italy
The atomic absorption technique has been used to detect trace impurity contamination in silicon wafers at different stages of processing. Suitable wafer etching by quantitative silicon removal followed by impurity concentration made it possible to reach significant detection limits for all the interesting contaminants, including Fe, Cr, Cu, Ni, K, Zn, Co, Mn, each present at levels below 2. The number of contaminant species can be further increased and the detection limits improved. The extent to which 1012 atoms/cm correlations are possible with other less direct methods is discussed.
1. Introduction In a previous paper a survey was conducted of the major analytical methods that are used in the silicon industry for bulk and surface contamination detection [1]. In this paper we address studies employing the Atomic Absorption method (AA) with the implementation of some improvement and refinement. The method is, in fact, relatively simple and suitable for routine use, and enables many determinations per unit time. It requires only some calibration with other methods to check for consistency and reproducibility. Other methods, mainly physical or indirect, such as the “drift method” and DLTS, are described in ref. [1].
2. Principle of the method A defined portion of the wafer surface is treated with a nitric acid solution at a stabilized temperature of typically 60°C.Under these conditions all metal ions present on the silicon surface are dissolved into the solution. The concentration of the contaminant in solution is determined by conven-
tional AA with the additional procedures as described below.
3. Expenmental techniques 3.1. Apparatus
(a) For sample preparation. A schematic illustration of the apparatus in which the metals’ stripping takes place is shown in fig. 1. (b) For analysis. The atomic absorption spectrophotometer is equipped with a graphite furnace and the capability for multiple sample injections in order to have pre-concentration of a suitable factor by automatic sampling before final atomization. The detectivity limits for single injection are shown in table 1. (c) Ambient. The content of particulates in the ambient must be not higher than class 100 (particularly for certain contaminants and/or impurity levels to be detected; for polished wafers class 100 is almost mandatory).
3.2. Reagents
*
Present address: SGS Microelectronics AGRATE Brianza, Milan, Italy.
The following reagents are employed to minimize extraneous contamination: DI water, metal “free” having a resistivity > 18 MQ cm, hyper-
0022-0248/88/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
40
A. Corradi et at
/ Surface contamination
detection below ppb range on Si wafers Table I Detection limits for single injection
PVOF
Element
Concentration Nitrogen
PVDF Wafer
~
~.iti
An~izedAl
Fig. 1. Schematic illustration of the apparatus.
pure nitric acid 65% (having a maximum metal content of 2 ppb), hydrofluoridric acid 40% (haying a maximum metal content of 5 ppb), and stripping solutions: nitric acid 0.2% or nitric + hydrofluoric acid (0.2% and 0.002% respectively). 3.3. Procedure
Metal stripping is accomplished as follows. After positioning of the wafer on the wafer holder, 20 cm3 of stripping solution are poured in. The metal dissolution takes place at a controlled temperature of 60°C for S mm. After this time the solution is directly transferred to the analytical equipment. To control the metal dissolution, further strip-
Ag Al As b) Au B
0.05 0.6 0.4 0.3
Be Ba Bi Ca Cd Co Cr Cs
0.05 2.5 0.35 0.07 0.15 0.25 0.15 0.4
Er Cu Dy Eu Fe Ga Hg In K Li Mg Mn Mo Na Ni p b) Pb Pd Pt Rb
4.3 0.15 1.8 1.0 0.1 1.5 19 2.3 0.035 0.18 0.013 0.05 0.9 0.01 0.5
—
—
0.15 0.6 5.5 0.13
a)
(ng/ml)
Argon
Weight (pg) Argon
0.05 0.35 0.4 0.2 50
1.0 7.0 8.0 4.0 1000
0.8 0.04 0.35 0.03 0.15 0.25 0.13 0.3
16 0.8 7.0 0.6 0.3 5.0 2.5 6.0
3.8 0.13 1.8 0.8 0.1 1.0 18 1.3 0.025 0.13 0.1 0.035 0.4 0.01 0.5 2000 0.15 0.6 4.5 0.08
75 352.5 15 2.0 20 350 25 0.5 2.5 0.2 0.7 8.0 0.2 10.0 40000 3.0 12 90 1.5
Sb 0.5 0.45 Sc 14 1.4 1.3 Si 1.4 1.3 Sn~ 1.2 1.1 Sr 0.3 0.15 Th 0.18 Te 1.0 1.0 Ti 3 2.3 Tl 1.5 1.3 V 1.5 1.4 Zn 0.0 0.01 a) Characteristic concentration is the quantity required to give an absorbance of 0.0044. ~ Using chemical modifier.
9.0 25 25 22 3.0 3.5 20 45 25 28 0.25 of the element
ping operations are carried out after the first, in order to verify that no residual metals are still present. The results are illustrated in fig. 2.
A. Corradi el a!. / Surface contamination detection below ppb range on Si wafers 1E 16
At the present state of this work, where we are continuously improving the method in all the analytical aspects, the current detection limits are: Fe: 1013 atoms/cm2, because of environmental reasons; Cu, Cr, Ni, Mn, Co, Mo, K, Zn: 1012 atoms/cm2. As far as the detection limits are concerned, a measurement is considered valid if the reference value, in the actual working condition, is at least 5 times lower than that of the actual analysis. I
I
\ 1E 15 ~ 1E 14
IE 13 1E12 I
I I
JLst 2nd 3rd Fig. 2. Check for quantitative stripping
41
I
4th Stripping of impurities.
3.4. Results The AA analysis method gives results which are concentrations in terms of ppb or less of the original solution. Such a concentration can be referred with a suitable computing program, to the actual silicon surface (stripped surface) involved in the analysis (atoms/cm2) or to a corresponding bulk silicon volume (atoms/cm3) [2].
4.3. Comparison with other methods In order to compare our current AA method with existing and already described methods [1], we qualified four cleaning processes and procedures indicated as A, B, C and D in table 2. D is a cleaning process set up as a result of many investigations making large use of the present analytical method; the improvement is substantial and the trend is confirmed by almost all the available indirect methods. One cannot expect a comparison in the absolute values because of the two following reasons: The AA technique is applied, at the present stage of its refinement, with enough sensitivity to lapped wafers (i.e. having a large effective surface) while DLTS and the drift test are be applied to the polished face of the wafers. The AA measures the total metal (e.g., Fe) content, while the Drift Test measures all the electrically active species regardless of their nature and DLTS all the unprecipitated (electrically active) atoms of an identified species. —
4. Discussion 4.1. Field of application One of the advantages of this method is that it can be applied on any kind of water surface and at any point throughout the entire wafer manufacturing process. It can also be applied to analyze any kind of metal, provided that: it is soluble in the stripping solution at the standard temperature; instrumental sensitivity is reasonable high; it is not commonly present as a contaminant in the solution, in the ambient or in the analytical equipment.
—
—
— —
4.2. Detection limits The detection limits strongly depend on the environment, the specific equipment, instrumental sensitivity and the punty of the reagents employed.
Table 2 Comparison of AA and other analytical techniques in monitoring different cleaning processes (atoms/cm3) Cleaning process
Drift (polished wafers)
DLTS (polished wafers)
AA (lapped wafers)
A B C D
1 xlO’3 1.1 X 1013 3.3x10’3 O.6x10’3
1.4x10’2 1.8 X 1012 1.8x1013 2.4x1012
4 X1O’3 6 X i0~3 3 x10~ 1.3x1013
_________________________________________
42
A. Corradi et al.
/ Surface contamination detection below ppb
range on Si wafers
5. Summary
References
The improved atomic absorption method described here is effective for detecting metal contamination in silicon wafers at ppb levels. It is applicable at various stages of wafer processing and gives the total metal contamination levels.
[1~ M. Domenici, P. Malinverni and M. Pedrotti, J. Crystal Growth 75 (1986) 80. [2] A. Corradi, M. Domenici and A. Guaglio, Dynamit Nobel Silicon, Novara, unpublished results, 1987.
39
SURFACE CONTAMINATION DETECTION BELOW THE ppb RANGE ON SILICON WAFERS A. CORRADI DNS
—
“,
M. DOMENICI and A. GUAGLIO
Dynamit Nobel Silicon, Viale Gherzi 31, 1-28100 Novara, Italy
The atomic absorption technique has been used to detect trace impurity contamination in silicon wafers at different stages of processing. Suitable wafer etching by quantitative silicon removal followed by impurity concentration made it possible to reach significant detection limits for all the interesting contaminants, including Fe, Cr, Cu, Ni, K, Zn, Co, Mn, each present at levels below 2. The number of contaminant species can be further increased and the detection limits improved. The extent to which 1012 atoms/cm correlations are possible with other less direct methods is discussed.
1. Introduction In a previous paper a survey was conducted of the major analytical methods that are used in the silicon industry for bulk and surface contamination detection [1]. In this paper we address studies employing the Atomic Absorption method (AA) with the implementation of some improvement and refinement. The method is, in fact, relatively simple and suitable for routine use, and enables many determinations per unit time. It requires only some calibration with other methods to check for consistency and reproducibility. Other methods, mainly physical or indirect, such as the “drift method” and DLTS, are described in ref. [1].
2. Principle of the method A defined portion of the wafer surface is treated with a nitric acid solution at a stabilized temperature of typically 60°C.Under these conditions all metal ions present on the silicon surface are dissolved into the solution. The concentration of the contaminant in solution is determined by conven-
tional AA with the additional procedures as described below.
3. Expenmental techniques 3.1. Apparatus
(a) For sample preparation. A schematic illustration of the apparatus in which the metals’ stripping takes place is shown in fig. 1. (b) For analysis. The atomic absorption spectrophotometer is equipped with a graphite furnace and the capability for multiple sample injections in order to have pre-concentration of a suitable factor by automatic sampling before final atomization. The detectivity limits for single injection are shown in table 1. (c) Ambient. The content of particulates in the ambient must be not higher than class 100 (particularly for certain contaminants and/or impurity levels to be detected; for polished wafers class 100 is almost mandatory).
3.2. Reagents
*
Present address: SGS Microelectronics AGRATE Brianza, Milan, Italy.
The following reagents are employed to minimize extraneous contamination: DI water, metal “free” having a resistivity > 18 MQ cm, hyper-
0022-0248/88/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
40
A. Corradi et at
/ Surface contamination
detection below ppb range on Si wafers Table I Detection limits for single injection
PVOF
Element
Concentration Nitrogen
PVDF Wafer
~
~.iti
An~izedAl
Fig. 1. Schematic illustration of the apparatus.
pure nitric acid 65% (having a maximum metal content of 2 ppb), hydrofluoridric acid 40% (haying a maximum metal content of 5 ppb), and stripping solutions: nitric acid 0.2% or nitric + hydrofluoric acid (0.2% and 0.002% respectively). 3.3. Procedure
Metal stripping is accomplished as follows. After positioning of the wafer on the wafer holder, 20 cm3 of stripping solution are poured in. The metal dissolution takes place at a controlled temperature of 60°C for S mm. After this time the solution is directly transferred to the analytical equipment. To control the metal dissolution, further strip-
Ag Al As b) Au B
0.05 0.6 0.4 0.3
Be Ba Bi Ca Cd Co Cr Cs
0.05 2.5 0.35 0.07 0.15 0.25 0.15 0.4
Er Cu Dy Eu Fe Ga Hg In K Li Mg Mn Mo Na Ni p b) Pb Pd Pt Rb
4.3 0.15 1.8 1.0 0.1 1.5 19 2.3 0.035 0.18 0.013 0.05 0.9 0.01 0.5
—
—
0.15 0.6 5.5 0.13
a)
(ng/ml)
Argon
Weight (pg) Argon
0.05 0.35 0.4 0.2 50
1.0 7.0 8.0 4.0 1000
0.8 0.04 0.35 0.03 0.15 0.25 0.13 0.3
16 0.8 7.0 0.6 0.3 5.0 2.5 6.0
3.8 0.13 1.8 0.8 0.1 1.0 18 1.3 0.025 0.13 0.1 0.035 0.4 0.01 0.5 2000 0.15 0.6 4.5 0.08
75 352.5 15 2.0 20 350 25 0.5 2.5 0.2 0.7 8.0 0.2 10.0 40000 3.0 12 90 1.5
Sb 0.5 0.45 Sc 14 1.4 1.3 Si 1.4 1.3 Sn~ 1.2 1.1 Sr 0.3 0.15 Th 0.18 Te 1.0 1.0 Ti 3 2.3 Tl 1.5 1.3 V 1.5 1.4 Zn 0.0 0.01 a) Characteristic concentration is the quantity required to give an absorbance of 0.0044. ~ Using chemical modifier.
9.0 25 25 22 3.0 3.5 20 45 25 28 0.25 of the element
ping operations are carried out after the first, in order to verify that no residual metals are still present. The results are illustrated in fig. 2.
A. Corradi el a!. / Surface contamination detection below ppb range on Si wafers 1E 16
At the present state of this work, where we are continuously improving the method in all the analytical aspects, the current detection limits are: Fe: 1013 atoms/cm2, because of environmental reasons; Cu, Cr, Ni, Mn, Co, Mo, K, Zn: 1012 atoms/cm2. As far as the detection limits are concerned, a measurement is considered valid if the reference value, in the actual working condition, is at least 5 times lower than that of the actual analysis. I
I
\ 1E 15 ~ 1E 14
IE 13 1E12 I
I I
JLst 2nd 3rd Fig. 2. Check for quantitative stripping
41
I
4th Stripping of impurities.
3.4. Results The AA analysis method gives results which are concentrations in terms of ppb or less of the original solution. Such a concentration can be referred with a suitable computing program, to the actual silicon surface (stripped surface) involved in the analysis (atoms/cm2) or to a corresponding bulk silicon volume (atoms/cm3) [2].
4.3. Comparison with other methods In order to compare our current AA method with existing and already described methods [1], we qualified four cleaning processes and procedures indicated as A, B, C and D in table 2. D is a cleaning process set up as a result of many investigations making large use of the present analytical method; the improvement is substantial and the trend is confirmed by almost all the available indirect methods. One cannot expect a comparison in the absolute values because of the two following reasons: The AA technique is applied, at the present stage of its refinement, with enough sensitivity to lapped wafers (i.e. having a large effective surface) while DLTS and the drift test are be applied to the polished face of the wafers. The AA measures the total metal (e.g., Fe) content, while the Drift Test measures all the electrically active species regardless of their nature and DLTS all the unprecipitated (electrically active) atoms of an identified species. —
4. Discussion 4.1. Field of application One of the advantages of this method is that it can be applied on any kind of water surface and at any point throughout the entire wafer manufacturing process. It can also be applied to analyze any kind of metal, provided that: it is soluble in the stripping solution at the standard temperature; instrumental sensitivity is reasonable high; it is not commonly present as a contaminant in the solution, in the ambient or in the analytical equipment.
—
—
— —
4.2. Detection limits The detection limits strongly depend on the environment, the specific equipment, instrumental sensitivity and the punty of the reagents employed.
Table 2 Comparison of AA and other analytical techniques in monitoring different cleaning processes (atoms/cm3) Cleaning process
Drift (polished wafers)
DLTS (polished wafers)
AA (lapped wafers)
A B C D
1 xlO’3 1.1 X 1013 3.3x10’3 O.6x10’3
1.4x10’2 1.8 X 1012 1.8x1013 2.4x1012
4 X1O’3 6 X i0~3 3 x10~ 1.3x1013
_________________________________________
42
A. Corradi et al.
/ Surface contamination detection below ppb
range on Si wafers
5. Summary
References
The improved atomic absorption method described here is effective for detecting metal contamination in silicon wafers at ppb levels. It is applicable at various stages of wafer processing and gives the total metal contamination levels.
[1~ M. Domenici, P. Malinverni and M. Pedrotti, J. Crystal Growth 75 (1986) 80. [2] A. Corradi, M. Domenici and A. Guaglio, Dynamit Nobel Silicon, Novara, unpublished results, 1987.