# US 2004/0080761 A1 · Method and apparatus for thickness decomposition of complicated layer structures

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2004/0080761 A1
- Document ID: US20040080761A1
- Status: Published application
- Kind: A1
- Application number: 10/281,207
- Filed: 2002-10-28
- Published: 2004-04-29
- Earliest claimed: 2002-10-28
- Assignees: Tevet Process Control Technologies Ltd.
- Inventors: Ofer Du-Nour; Vladimir Rubinstein
- Web page: https://patentgrep.com/patent/US20040080761A1

## CPC

- G01B 11/0625: {with measurement of absorption or reflection}
- G01B: MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS

## Abstract

Thickness measurement apparatus for measuring layer thicknesses on patterned areas of a semiconductor wafer, comprises: a spectrum analyzer for obtaining reflection data taken from a patterned area and obtaining therefrom a frequency spectrum, a peak detector for searching the spectrum to find peak frequencies within said spectrum, the search being restricted to regions corresponding to peak frequencies found in an earlier learning stage, a frequency filter, associated with the peak detector, for filtering the spectrum about said peak frequencies, and a maximum likelihood fitter for using parameters obtained in the learning stage to carry out maximum likelihood fitting of said filtered spectrum to obtain the desired layer thicknesses. By carrying out maximum likelihood fitting using parameters obtained beforehand in a high resolution non-real time learning stage, it is possible to provide high resolution results in real time.

## Claims

### Claim 1

Thickness measurement apparatus for measuring layer thicknesses on patterned areas of a semiconductor wafer, the apparatus comprising:
a spectrum analyzer for obtaining reflection data taken from a patterned area and obtaining therefrom a frequency spectrum,
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum, said peak detector being operable to restrict said search to regions corresponding to peak frequencies found in a learning stage,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
and a maximum likelihood fitter for using parameters obtained in said learning stage to carry out maximum likelihood fitting of said filtered spectrum to obtain at least said layer thicknesses

### Claim 2 (depends on 1)

The apparatus of claim 1, wherein said reflection data is obtained from polychromatic light irradiation of said patterned area.

### Claim 3 (depends on 1)

The apparatus of claim 1, wherein said spectrum analyzer comprises a spectrometer, a dispersion corrector, a wave number transformer and a Fourier transformer.

### Claim 4 (depends on 1)

The apparatus of claim 1, wherein said peak frequencies found in said learning stage correspond to layer thicknesses of an initial sample.

### Claim 5 (depends on 4)

The apparatus of claim 4, wherein said layer thicknesses of said initial sample are determined in said learning stage by using spectral analysis to construct a spectrum at a higher resolution than that obtained by said spectrum analyzer.

### Claim 6 (depends on 1)

The apparatus of claim 1, wherein said frequency filter is operable to find minima on either side of each peak and to carry out filtering for each peak by filtering a range defined by said minima.

### Claim 7

Thickness measurement apparatus for measuring layer thicknesses on patterned areas of a semiconductor wafer, the apparatus comprising:
a) an input spectrum analyzer for obtaining a reflection spectrum from respective patterned areas of semiconductor wafers,
b) a learn mode unit comprising:
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum, said peak detector being operable to restrict said search to regions corresponding to expected thicknesses of layers in said patterned areas,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
a high resolution spectrum analyzer for obtaining parameters from said filtered spectrum for use in maximum likelihood fitting, and
a maximum likelihood fitter for using said parameters to carry out maximum likelihood fitting of said filtered spectrum to obtain at least learn mode layer thicknesses
and c) a run mode unit comprising:
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum, said peak detector being operable to restrict said search to regions corresponding to peak frequencies found by said learn mode unit,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
and a maximum likelihood fitter for using said parameters obtained by said learn mode unit to carry out maximum likelihood fitting of said filtered spectrum to obtain said layer thicknesses.

### Claim 8 (depends on 7)

The apparatus of claim 7, wherein said high resolution spectrum analyzer is operable to obtain said parameters by constructing a new spectrum at a resolution higher than a resolution obtained by said input spectrum analyzer.

### Claim 9

A method for measuring layer thicknesses on patterned areas of a semiconductor wafer, the method comprising:
obtaining reflection data taken from a patterned area,
obtaining therefrom a frequency spectrum,
searching said spectrum to find peak frequencies within said spectrum said search being restricted to regions corresponding to peak frequencies found in a learning stage,
filtering said spectrum about said peak frequencies,
and using parameters obtained in said learning stage to carry out maximum likelihood fitting of said filtered spectrum to obtain said layer thicknesses.

### Claim 10 (depends on 9)

The method of claim 9, wherein said reflection data is obtained from polychromatic light irradiation of said patterned area.

### Claim 11 (depends on 9)

The method of claim 9, wherein obtaining said frequency spectrum comprises measuring the spectrum of said reflection data, correcting for dispersion, transforming for wave number and carrying out Fourier transforming.

### Claim 12 (depends on 9)

The method of claim 9, wherein said peak frequencies found in said learning stage correspond to layer thicknesses of an initial sample.

### Claim 13 (depends on 12)

The method of claim 12, wherein said layer thicknesses of said initial sample are determined in said learning stage by using spectral analysis to construct a spectrum at a higher resolution than that obtained by said spectrum analyzer.

### Claim 14 (depends on 9)

The method of claim 9, wherein said filtering said spectrum comprises finding minima on either side of each peak and carrying out filtering for each peak over a range defined by said minima.

### Claim 15

Method for measuring layer thicknesses on patterned areas of a semiconductor wafer, comprising:
a) a stage of obtaining a reflection spectrum from respective patterned areas of semiconductor wafers,
b) a learning stage comprising:
searching said spectrum to find peak frequencies within said spectrum, whilst restricting said search to regions corresponding to expected thicknesses of layers in said patterned areas,
filtering said spectrum about said peak frequencies,
obtaining parameters from said filtered spectrum for use in maximum likelihood fitting, and
using said parameters to carry out maximum likelihood fitting of said filtered spectrum to obtain learn mode layer thicknesses; and
c) a run stage comprising:
searching said spectrum to find peak frequencies within said spectrum, whilst restricting said search to regions corresponding to peak frequencies found in said learning stage,
filtering said spectrum about said peak frequencies, and
using said parameters obtained in said learning stage to carry out maximum likelihood fitting of said filtered spectrum to obtain said layer thicknesses.

### Claim 16 (depends on 15)

The method of claim 15, wherein said stage of obtaining parameters from said filtered spectrum for use in maximum likelihood fitting comprises constructing a higher resolution version of said obtained reflection spectrum.

### Claim 17

Apparatus for controlling a semiconductor wafer production line, said production line having a plurality of stations, successive stations being for carrying out successive processes on wafers to add features to said wafers, at least one of said stations having a measuring unit for providing measurements of layers on a patterned surface part of a respective wafer, the measuring unit comprising:
a spectrum analyzer for obtaining reflection data taken from a patterned area and obtaining therefrom a frequency spectrum,
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum, whilst restricting said search to regions corresponding to peak frequencies found in a learning stage,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
and a maximum likelihood fitter for using parameters obtained in said learning stage to carry out maximum likelihood fitting of said filtered spectrum to obtain said layer thicknesses, said layer thicknesses being usable as control signals for said wafer production line.

### Claim 18 (depends on 17)

The apparatus of claim 17, wherein said reflection data is obtained from polychromatic light irradiation of said patterned area.

### Claim 19 (depends on 17)

The apparatus of claim 17, wherein said spectrum analyzer comprises a spectrometer, a dispersion corrector, a wave number transformer and a Fourier transformer.

### Claim 20 (depends on 17)

The apparatus of claim 17, wherein said peak frequencies found in said learning stage correspond to layer thicknesses of an initial sample.

### Claim 21 (depends on 17)

The apparatus of claim 17, wherein said layer thicknesses of said initial sample are determined in said learning stage by using spectral analysis to construct a spectrum at a higher resolution than that obtained by said spectrum analyzer.

### Claim 22 (depends on 17)

The apparatus of claim 17, wherein said frequency filter is operable to find minima on either side of each peak and to carry out filtering for each peak by filtering a range defined by said minima.

### Claim 23 (depends on 17)

The apparatus of claim 17, wherein said measuring units are located to take thickness measurements before, during and after processing at a given station.

### Claim 24

Apparatus for controlling a semiconductor wafer production line, said production line having a plurality of stations, successive stations being for carrying out successive processes on wafers to add features to said wafers, at least one of said stations having a measuring unit for providing measurements of layers on a patterned surface part of a respective wafer, the measuring unit comprising:
a) an input spectrum analyzer for obtaining a reflection spectrum from respective patterned areas of semiconductor wafers,
b) a learn mode unit comprising:
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum, whilst restricting said search to regions corresponding to expected thicknesses of layers in said patterned areas,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
a high resolution spectrum analyzer for obtaining parameters from said filtered spectrum for use in maximum likelihood fitting, and
a maximum likelihood fitter for using said parameters to carry out maximum likelihood fitting of said filtered spectrum to obtain learn mode layer thicknesses;
and
c) a run mode unit comprising:
a peak detector, associated with said spectrum analyzer, for searching said spectrum to find peak frequencies within said spectrum whilst restricting said search to regions corresponding to peak frequencies found by said learn mode unit,
a frequency filter, associated with said peak detector, for filtering said spectrum about said peak frequencies,
and a maximum likelihood fitter for using said parameters obtained by said learn mode unit to carry out maximum likelihood fitting of said filtered spectrum to obtain said layer thicknesses said layer thicknesses being usable as control signals for said wafer production line.

### Claim 25 (depends on 24)

The apparatus of claim 24, wherein said high resolution spectrum analyzer is operable to obtain said parameters by constructing a new spectrum at a resolution higher than a resolution obtained by said input spectrum analyzer.

### Claim 26 (depends on 24)

The apparatus of claim 24, wherein said measuring units are located to take thickness measurements before, during and after processing at a given station.
