semi合集-English.pdf - 第7311页
SEMI MF1188-0305 © SEMI 2003, 2005 5 after transform ation, a spectrum obtained at 4 cm 1 resolution shall contain at least one data poin t every two wavenumbers. 7.2.3 Phase Correction — The phase correction routine u…
SEMI MF1188 © SEMI 2003, 2005 4
5.2.1 certified reference material (CRM), n — a reference material, one or more of whose property values are
certified by a technically valid procedure, accompanied by or traceable to a certificate or other documentation issued
by a certifying body.
5.2.2 dispersive infrared (DIR) spectrophotometer, n — a type of infrared spectrometer that uses at least one prism
or grating as the dispersing element, in which the data are obtained as an amplitude-wavenumber (or wavelength)
spectrum.
5.2.2.1 Discussion — Some dispersive infrared spectrometers are used in conjunction with a computer, which is
used to store data. The data are then accessible for manipulation or computation, as required. These spectrometers
are referred to as computer-assisted dispersive infrared spectrophotometers (CA-DIR). Dispersive infrared
spectrometers that are not computer-assisted are referred to, for convenience, as simple dispersive infrared
spectrometers (S-DIR).
5.2.3 Fourier transform infrared (FT-IR) spectrophotometer, n — a type of infrared spectrometer in which the data
are obtained as an interferogram.
5.2.3.1 Discussion — An interferogram is a record of the modulated component of the interference signal measured
by the detector as a function of retardation in the interferometer. This interferogram is then subjected to a Fourier
transformation to obtain an amplitude-wavenumber (or wavelength) spectrum. Due to the complexity of the Fourier
transformation, FT-IR instruments are always used in conjunction with a computer.
5.2.4 reference spectrum, n — the spectrum of the reference specimen.
5.2.4.1 Discussion — In true double-beam spectrometers, the reference spectrum may be obtained directly, with the
reference specimen in the sample beam, and the reference beam empty. In single-beam spectrometers, it can be
calculated from the ratio of a spectrum obtained with the reference specimen in the IR beam, to a background
spectrum.
5.2.5 sample spectrum, n — the spectrum of the test specimen.
5.2.5.1 Discussion — In true double-beam spectrometers, the sample spectrum may be obtained directly, with the
sample specimen in the sample beam, and the reference beam empty. In single-beam spectrometers, it can be
calculated from the ratio of a spectrum obtained with the test specimen in the IR beam, to a background spectrum.
6 Summary of Test Method
6.1 The relative infrared transmittance spectrum of an oxygen-containing silicon slice is obtained using a reference
method with a calibrated IR spectrophotometer, preferably one calibrated by means of a suitable set of reference
materials. The oxygen-free reference specimen is matched closely in thickness to the test specimen, so as to
eliminate the effects of absorption due to silicon lattice vibrations.
6.2 After conducting selected instrumental checks, the infrared transmittance spectrum is measured over the
wavenumber range from 900 cm
1
to 1300 cm
1
.
6.3 The measured absorption coefficient of the 1107 cm
1
oxygen-in-silicon band is then used to determine the
interstitial oxygen content of the silicon slice.
7 Apparatus
7.1 Infrared Spectrophotometer , either a DIR (S or CA), or FT-IR instrument, as described in ¶5.2.2 and ¶5.2.3,
respectively, may be used. It must be possible to set the resolution of the spectrophotometer to 4 cm
1
, or better, for
FT-IR spectrophotometers, and to 5 cm
1
, or better, for DIR spectrophotometers. Use of an FT-IR spectropho-
tometer is strongly recommended.
7.2 The three following paragraphs apply only to FT-IR spectrophotometers:
7.2.1 Zero Filling — When an FT-IR instrument collects an unsymmetrical interferogram, an additional set of
points whose values are all zero shall be added to the end of the collected interferogram such that the total number of
points for performing the Fourier transform is double the number of data points originally collected.
7.2.2 Undersampling — The data collection method shall produce interferograms which, when zero-filled and
Fourier transformed, produce a spectrum containing at least two data points per resolution increment. For example,
SEMI MF1188-0305 © SEMI 2003, 2005 5
after transformation, a spectrum obtained at 4 cm
1
resolution shall contain at least one data point every two
wavenumbers.
7.2.3 Phase Correction — The phase correction routine used during Fourier transformation shall use at least 200
points on both sides of the point of zero retardation in order to produce a phase array that can be used to eliminate
phase errors.
12
7.3 Specimen Holders of Appropriate Size — If the specimen being measured is small, it must be mounted in a
holder that has an opening small enough to prevent any of the infrared beam from bypassing the specimen. The
specimens shall be held normal, or nearly normal, to the axis of the incident infrared beam (see ¶8.3).
7.4 Equipment and Materials, for slicing and polishing crystals to a thickness similarity of 0.5% or less and a
surface flatness of 2.25 m or better (one fourth the wavelength at the maximum absorption of the interstitial
oxygen impurity band).
7.5 Acid Fume Hood and Suitable Protective Gear, for use when stripping oxide films from specimens for
measurement.
7.6 Micrometer Caliper, or other instrument suitable for the measurement of the thickness of the specimens to a
tolerance of ±0.2%.
7.7 Thermocouple-Millivolt Potentiometer, or other system suitable for measurements of the specimen temperature
during test.
7.8 Hydrofluoric Acid, 4.9% (10:1 v/v), in accordance with Grade 2 of SEMI C29.
NOTE 1: Warning — The acids used in this test method are hazardous. All precautions normally used with these chemicals
should be strictly observed. Obtain and read the Material Safety Data Sheet prior to the use of any chemical.
8 Testing of the Apparatus
8.1 Evaluate the performance of S-DIR spectrometers according to Instrument Operation and Nature of Test
Sections of ASTM Practice E 932. Follow the appropriate paragraphs of these sections to evaluate the performance
of CA-DIR instruments.
8.2 Verify a proper purge condition for the specimen chamber by monitoring water vapor or carbon dioxide
absorption bands. The water vapor line is monitored at 1521 cm
1
and the carbon dioxide line at 667 cm
1
. The
instrument shall be sufficiently well purged or evacuated that the transmittance at these locations is between 98 and
102%.
8.3 Under certain conditions, the spectrophotometer may have a nonlinear response, or be plagued by undesirable
extraneous reflections between the specimen surfaces and the spectrometer components. Place a flat, double-side
polished and high resistivity (greater than 5 ·cm) silicon slice in the instrument. The effective transmittance of the
silicon slice, due to reflective losses at the silicon surfaces, should be 53.8% ± 2% in the 1600 cm
1
to 2000 cm
1
region. In some instruments, this silicon slice may have to be placed at a small angle to the axis of the incoming IR
beam, in order to minimize undesirable reflections between the silicon surfaces, and the spectrometer components.
This angle may be determined by initially placing the silicon slice normal to the axis of the incoming beam and then
gradually tilting the sample while repeatedly obtaining the transmittance spectrum of the slice above 1600 cm
1
.
The optimum angle is reached when a flat baseline as close as possible to 53.8% ± 2% is obtained from 1600 cm
1
to
4000 cm
1
. This optimum angle is typically less than 10°.
9 Calibration Specimens
9.1 Obtain a set of certified reference materials (CRMs) for interstitial oxygen in silicon with the following
characteristics:
9.1.1 Each specimen in the set is double-side polished and approximately 2 mm thick,
9.1.2 The set covers the oxygen range of interest with three or more samples,
12 For a discussion of the phase correction computation, see Chase, D. B., Applied Spectroscopy 36, 240 (1982).
SEMI MF1188 © SEMI 2003, 2005 6
9.1.3 The stated oxygen content is referenced to the IOC-88 calibration factor,
1
and
9.1.4 The set includes an additional oxygen-free reference specimen.
9.2 To extend the test method to measurement of test specimens with thickness other than 2 mm or with only a
single polished surface, suitable sets of reference materials must be used to calibrate the instrument for use with
specimens of similar thickness and surface conditions.
9.2.1 Sufficient sets of reference materials must be utilized to provide traceability from the reference materials with
thickness and surface conditions of the specimens to be tested to the set of CRMs.
NOTE 2: A practice for calibrating infrared spectrophotometers for measuring oxygen in silicon using reference materials of
various thickness and surface condition is under development. In the interim, some, but not complete, guidance for this
calibration can be found in the NIST SRM 2551 Report.
5
10 Test Specimen
10.1 Choose test specimens that are as homogeneous as possible, so that the oxygen content measured is a fair
representation of the oxygen content of the entire specimen. This is particularly important for dispersive
spectrophotometers, since in many such instruments the illuminated area of the specimen varies during the scan. For
specimens with a large lateral inhomogeneity, this area variation may result in the appearance of undesirable
instrumental artifacts in the spectra.
10.1.1 Prepare a slice of the crystal to be tested so as to obtain two optically flat surfaces parallel to 5 min of arc or
less, as measured with a micrometer caliper or other suitable instrument. The surfaces of the specimen must be as
free as possible of surface films.
NOTE 3: When the specimen faces are parallel and well-polished, and the data are being obtained at a sufficiently high
resolution, interference may occur between light rays reflecting from the front and back surfaces of the specimen. The contrast of
the interference fringes depends upon the parallelism of the specimen surfaces, and the fringe spacing depends on the optical
thickness of the specimen. These fringes can obscure a weak spectral line and prevent accurate measurement of the baseline. To
prevent obscuration by these interference fringes, nonparallel specimen surfaces may be a necessity. However, the use of
specimens with nonparallel surfaces can also result in photometric errors. Because silicon has a high refractive index, any
nonparallelism of the specimen can displace the spectrometer beam relative to the active area of the detector. Thus, an apparent
lowering of the transmission occurs. Improper positioning or nonparallelism of the specimen can be checked by rotating the
specimen to determine whether the transmission level stays constant. Any variation is a possible indication of problems with the
specimen positioning or preparation.
11 Reference Specimens
11.1 Because a difference technique is used in this test method, prepare an oxygen-free reference specimen of the
same type of material as the sample.
11.1.1 Choose the oxygen-free reference specimen from slices taken from five to ten different silicon crystals that
are thought to be free of oxygen. Compare these slices with one another, and choose the specimen with the lowest
absorption as the oxygen-free reference specimen. If no absorption is seen for any of the specimens, then the
assumption can be made that all specimens contain less than the limit of detection of oxygen and any of the
specimens can be used as the oxygen-free reference specimen.
11.1.2 Prepare the oxygen-free reference specimen to the same tolerances as the test specimen with a thickness of
equal to that of the test specimen to within ±0.5%.
11.2 Prepare separate oxygen-free reference specimens for each type of material to be tested.
12 Procedure
12.1 Calibration of Spectrophotometer
12.1.1 Obtain a set of certified reference materials for oxygen in silicon (see §9) and, if necessary, suitable sets of
traceable reference materials with different thickness or surface condition, or both.
12.1.2 Use these reference materials to calibrate the spectrophotometer in accordance with manufacturer’s
instructions. The calibration results in a linear regression curve that relates the measured absorption coefficient to
the interstitial oxygen content of the measured specimen (Note 1).