semi合集-English.pdf - 第7388页

SEMI MF1526-95 © SEMI 2004 4 10. Test Specimen Meas urement 10.1 Load the test specimen into t he TXRF instrument in a contam ination free e nvironm ent of Class 100 or better. 10.2 Measure the TXRF spectrum for the test…

SEMI MF1526-95 © SEMI 2004 3

6. Interferences

6.1 The interferences in conventional X-ray fluorescence spectroscopy are common to TXRF also. These include, but are

not limited to: overlap of fluorescence lines, escape peak and sum peak overlap, energy gain calibration drift, X-ray source

stability, and instrumental background peaks. However, no X-ray fluorescence corrections for secondary fluorescence or for

matrix absorption are required for TXRF. Interferences common to software procedures and calculations can be evaluated by

comparing data sets; see Annex A1.

6.2 In addition to conventional interferences, there are interferences that are unique to TXRF as follows:

6.2.1 If the glancing angle calibration is not reproducible, variability is introduced to the measurement,

6.2.2 If the glancing angle calibration is inaccurate, bias is introduced to the measurement,

6.2.3 If the anglescan of the known elemental impurity on the calibration specimen is different from the elemental impurity

anglescan on the test specimen, this may introduce a bias to the quantification. An example can be the measurement of

particulate metal contamination on a test specimen while using a calibration specimen which has the calibration metal

localized within the native oxide,

6.2.4 Mechanical vibration may degrade the detector energy resolution, and this may degrade detection limits,

6.2.5 If the specimen is not chem-mechanically polished, there will be a loss in detection capability, a bias in quantification,

and an increase in variability of the measurement. There is some possibility that the surface roughness and waviness

differences from different cleaning processes may cause these interferences also. The semiquantitative effect of surface

roughness upon the TXRF measurement is under study.

6.2.6 Bias in the assigned elemental areal density of the calibration specimen can introduce bias into the TXRF measured

areal densities,

6.2.7 Surface contamination introduced during handling of the test specimens or during the measurement itself will

introduce a bias to the measurement, if this surface contamination contains the element(s) to be measured,

6.2.8 Bias in the RSF of a fluorescence line can introduce bias to the measurement,

6.2.9 Nonlinearity of detected fluorescence signal versus impurity areal density may occur due to high deadtime of some

detectors under the condition of high-total signal count rates, and

6.2.10 Fluorescence curve smoothing may affect the quantification accuracy.

6.2.11 Instrumental peaks may be generated by the incident X-ray beam diffracting from the silicon crystal and the

diffracted beam entering the detector to excite metals in the detector window or detector housing. This effect can be tested for

by appropriate experiments.

7. Apparatus

7.1 TXRF Instrument, equipped with a monochromatic X-ray source, test specimen handling equipment, a method for

glancing angle calibration, an energy-dispersive spectrometer X-ray detector, software for background subtraction, peak

integration, analysis and RSFs, and an analysis ambient without argon (for example, vacuum of 10

2

torr, or helium gas). The

methods for glancing angle calibration are presently proprietary for each TXRF instrument manufacturer. A TXRF

manufacturer may include a subtraction routine for escape peaks and these signals may already be removed.

7.2 Reference Wafer— The suitability of the apparatus shall be determined with the use of a reference wafer and its

associated data set in accordance with the procedures of Annex A2, or by performance of a statistically-based instrument

repeatability study to ascertain whether the equipment is operating within the manufacturer's stated specification for

repeatability.

7.3 Class 100 Air Environment—The area for sample transfer to the instrument measurement stage must be enclosed in

this.

8. Sampling

8.1 A sampling procedure must be used to evaluate the characteristics of a group of silicon wafers. No general sampling

procedure is included as part of this test method, because the most suitable sampling plan will vary considerably depending

upon individual conditions. For referee purposes, a sampling plan shall be agreed upon before conducting the test. See

Practice E 122 for suggested choices of sampling plans.

9. Specimen Requirements

9.1 Test specimens must be chem-mechanical polished on the side used for analysis.

7 Yakushiji, K., Ohkawa, S., Yoshinga, A., and Harada, J.,“ Origins of Spurious Peaks of Total Reflection X-Ray Fluorescence Analysis of Si Wafers

Excited by Monochromatic X-Ray Beam W1-beta,” Japanese Journal of Applied Physics, Vol 33, 1994, pp. 1130—1135.

SEMI MF1526-95 © SEMI 2004 4

10. Test Specimen Measurement

10.1 Load the test specimen into the TXRF instrument in a contamination free environment of Class 100 or better.

10.2 Measure the TXRF spectrum for the test specimen.

10.3 Calculate the net integrated counts/second for detected elemental peaks.

10.4 Using the calibration wafer data for the calibration element and the RSFs for the other elements, calculate the

elemental areal densities for each detected element on the test specimen according to Eq. 1.

um

()

1/F

sm

()

CPS

um

()

CPS

1

()

(1)

where:

su,m

= the areal density in atoms/cm

of element “m” on the

test specimen “u”,

s,m

= the RSF of element “m” with respect to the

calibration element “s”,

CPS

u,m

= the integrated counts/second of element “m”

detected on test specimen “u”,

CPS

= the integrated counts/second of the calibration

element on the calibration standard, and

= the assigned areal density in atoms/cm

of the

calibration element.

11. Calibration Standard

11.1 The calibration standard and the method of its preparation must be agreed upon between the parties.

NOTE 1—Several methods for preparing calibration standards are in practice.

The present state-of-the-art calibration standard is a chem-mechanical polished silicon substrate with a surface metal areal density of between

1 by 10

and 1 by 10

atoms/cm

located in the analysis area. The k-alpha fluorescence signal from the calibration standard metal is free

of interferences, that is, escape peaks, sum peaks, and fluorescence peaks from other contamination, and free from external contamination

sources. The preferred element should not be easily added as a contaminated element (for example, iron) or diffuse away from the analyte

depth over time (for example, gold or copper). The preferred element is nickel or vanadium. Calibration can also be done using a set of

wafers with different known areal densities of a particular element.

The calibration standard should have been measured by an appropriate analytical method to determine the metal areal density of the

calibration metal. Some analytical methods which have been used for assigning quantitative values to calibration standards include:

(a) Nitrogen-Beam Rutherford Backscattering Spectrometry (N-RBS)—If this method is used, there should not be any other metals present at

levels greater than 1 % of the calibration metal areal density within a mass range of ±5 atomic mass unit. The N-RBS measurement is

absolute and must be made within the analysis area of the TXRF. Other backscattering methods are also used to calibrate elemental areal

densities. These include forward scattering Rutherford backscattering (F-RBS) and heavy ion backscattering spectrometry (HIBS).

(b) Vapor Phase Decomposition—This followed by atomic absorption spectrometry (VPD/AAS) is used to calibrate a spin coating

contamination process for making calibration standards. The VPD/AAS measurement is destructive. The contamination process should also

be shown to be uniform across the wafer by using TXRF or SIMS mapping. The VPD/AAS absolute assignment is based upon Atomic

Absorption Spectroscopy standards. An error in the elemental recovery rate of VPD is a source of error in accuracy.

(c) Implant—Ion implant the reference element into a pre-amorphized silicon surface and using solid phase epitaxy to regrow the amorphous

silicon into crystal silicon.

This procedure sweeps the ion implanted reference element to the sample surface if the reference element is

much more soluble in the amorphous silicon than in the crystal silicon. The assignment of the quantitative level in atoms/cm

is done using

the ion implant current to determine the implant dose.

(d) Diluted Atomic Absorption Standard Solution—Deposit onto a polished silicon substrate a diluted atomic absorption standard solution of

a metal onto a localized spot much less than the size of the TXRF analysis area. This approach assumes none of the calibration metal is lost

8 Hockett, R. S., “TXRF Reference Standards: A Discussion,” Contamination Control and Defect Reduction in Semiconductor Manufacturing III, Vol 94-9,

edited by Dennis N. Schmidt, The Electrochemical Society, Pennington, NJ, 1994, pp. 323–334.

9 Hockett, R. S., and Jacobson, D. C., “A New Approach to TXRF Standards,” Extended Abstracts, Vol 93-1, The Electrochemical Society, Inc., Pennington,

NJ, 1993, p. 1289; Jacobson, D. C., Poate, J. M., Higashi, G. S., Boone, T., Eaglesham, D. J., and Hockett, R. S., “Ion Implanted Calibration Standards for Si

Surface Contamination Detection by TXRF,” Materials Research Society, Proceedings of the Spring '93 MRS Meeting, San Francisco, April 26–29, 1993.

during the drying of the solution. The deposition spot must be easily found by the TXRF instrument in order to analyze the entire deposited

dried solution. The absolute assignment is based upon the AAS standards. The deposited solution must produce one of two TXRF anglescan

forms:

(1) A metal fluorescence count rate versus angle that is independent of glancing angle for angles 80 % below the critical angle (see Curve (a)

in Fig. 2

(2) A metal fluorescence count rate versus angle that is characteristic of metal contamination localized within a 3-nm thickness of the surface

(see Curve (b) in Fig. 2).

11.2 An anglescan of the calibration standard must be made to verify if the anglescan of the calibration element is similar to

Curve (a) (residue) or Curve (b) (in the film) in Fig. 2.

11.3 An appropriate non-TXRF measurement of the calibration standard must be agreed upon between parties in order to

assign the elemental areal density to the calibration standard element.

11.4 The calibration of the other elements detectable by the TXRF instrument must be completed using relative sensitivity

factor (RSFs) developed by the instrument manufacturer and stored in the instrument computer program. These RSFs are a

function of the X-ray source energy, the atomic number of the fluorescing element, and the fluorescence energy level, so if the

X-ray source energy is changed, a different set of RSFs must be used.

12. Procedure

12.1 Turn on the instrument in accordance with the manufacturer's instructions.

12.2 Analytical Conditions:

12.2.1 Choose and record the analytical conditions for the test specimen measurement. This includes:

12.2.1.1 Voltage to the X-ray source,

12.2.1.2 Current to the X-ray source,

12.2.1.3 Ambient in the analysis chamber,

12.2.1.4 Glancing angle,

12.2.1.5 Integration time,

12.2.1.6 X-ray source energy, and

12.2.1.7 Analysis location on the test specimen.

13. Calibration

13.1 Load the calibration specimen into the TXRF instrument in a contamination free environment defined by Class 100 or

better.

13.2 Measure the TXRF spectrum of the calibration specimen under the same operating conditions of current and voltage

applied to the X-ray source and of glancing angle to be used for the measurement of the test specimen. The measurement time

and the ambient (for example, vacuum, air, nitrogen, helium, etc) in the analysis chamber may be different between the

calibration and test specimen measurements. If the calibration anglescan is similar to Curve (b) of Fig. 2, use a glancing angle

which is 70 to 80 % of the critical angle. If the calibration anglescan is similar to Curve (a) of Fig. 2, use any glancing angle

that is below 85 % of the critical angle.

13.3 A polished silicon substrate with no detectable elemental contamination (that is, a BLANK), or a series of polished

silicon substrates each of which shows no detectable metal contamination for a selected element of interest (that is, a set of

element-specific BLANKs) and which collectively represent a blank for all elements of interest, must be measured by the

TXRF measurement under the condition of the test specimen to verify three are no instrumental background signals. An

example BLANK for all elements except sulfur is shown in Fig. 3.

13.4 Have the TXRF instrument integrate the calibration elemental fluorescence signal and subtract the background to

obtain a net integrated counts/second that corresponds to the assigned elemental areal density. The background may be

subtracted by a deconvolution routine or by a linear fit routine.

14. Report

14.1 Record the following information (see also 12.2.1):

14.1.1 Test specimen identification,

14.1.2 Calibration specimen identification,

10 Eichinger, P., Rath, H. J., and Schwenke, H., “Application of Total Reflection X-Ray Fluorescence Analysis for Metallic Trace Impurities on Silicon

Wafer Surfaces,” Semiconductor Fabrication: Technology and Metrology, ASTM STP 990, ASTM, 1989, pp. 305–313.