JP3141499B2 - Ellipsometer parameter measurement method and ellipsometer - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an ellipsometric parameter measuring method for measuring an ellipsometric parameter used for accurately measuring a thin film thickness and an ellipsometer using the method. The present invention relates to an ellipsometer parameter measuring method and an ellipsometer for calculating a highly accurate ellipsometer parameter using a statistical method.

[0002]

2. Description of the Related Art An ellipsometry technique is used as a technique for measuring the thickness of a thin film. This method involves changing the polarization state when light is reflected from a sample surface such as a thin film, that is, the reflectance Rp of a component (P component) parallel to the plane of incidence of the electric field vector.
And the ratio ρ of the reflectance Rs of the vertical component (S component) to (5)
The film thickness d is determined in accordance with a predetermined relationship between the polarization reflectance ratio ρ and the film thickness d that has been established by the equation. ρ = Rp / Rs = tanψexp [jΔ] (5)

[0003] Here, the polarization reflectance ratio ρ is generally a complex number as shown in equation (5), so that two ellipso parameters, that is, an amplitude ratio tanψ and Δ, 決定 determined by a phase difference Δ, are given by: Need to ask.

Conventionally, as a method of obtaining the ellipsometric parameters ψ and Δ, there is a method called a rotation analyzer method. In this method, for example, light polarized at a predetermined angle from a light source is incident on a measurement target, and elliptically polarized reflected light from the measurement target is transmitted through a rotating analyzer and guided to a light receiver. Then, the ellipsometric parameter is calculated from a light intensity signal waveform obtained by the light receiver at that time.

However, when performing one measurement, it is necessary to rotate the analyzer by one rotation to observe the light intensity signal, so that a certain time or more is required for the rotation. Therefore, it is impossible to measure the film thickness of the measurement object moving at high speed. In addition, the presence of mechanically movable parts makes the apparatus itself large, and cannot be installed on a production line or the like in a factory to measure, for example, a continuously supplied measurement object online.

In order to eliminate such inconvenience, FIG.
As shown in FIG. 6, a three-channel ellipsometer in which a movable portion is removed has been developed (JP-A-63-3610).
No. 5, JP-A-1-28509).

For example, light having a single wavelength output from a laser light source 1 is converted into linearly polarized light by a polarizer 2 and is incident on a sample surface 3 to be measured at a predetermined angle φ. In addition,
On the sample surface 3, the incident surface is parallel to the paper surface, and as shown in the drawing, the direction parallel to the incident surface is defined as P direction, and the direction orthogonal to the incident surface is defined as S direction. The reflected light from the sample surface 3 is split into three beams by three non-polarized beam splitters 4a, 4b, and 4c.

Then, the two beam splitters 4a and 4
The light transmitted through b is incident on the light receiver 7a via the analyzer 5a and the condenser lens 6a. The light receiver 7a detects the light intensity I1. Similarly, the light transmitted through the beam splitter 4a and reflected by the next beam splitter 4b enters the light receiver 7b via the analyzer 5b and the condenser lens 6b. The light receiver 7b detects the light intensity I2. Further, the light reflected by the beam splitter 4a and transmitted through the next beam splitter 4c is transmitted to the analyzer 5c and the condenser lens 6c.
The light is incident on the light receiver 7c via c. The light receiver 7c detects the light intensity I3.

Here, the polarization direction of the analyzer 5a is set to the reference direction (azimuth 0 °), and the polarization direction of the analyzer 5b is set to be inclined by + 45 ° with respect to the reference direction.
The polarization direction of c is set to be inclined at -45 ° with respect to the reference direction. The reference direction is a direction in which a direction (P direction) parallel to the plane of incidence of light on the sample surface 3 is an azimuth of 0 °, as indicated by an arrow a direction in the drawing when viewed from the light receiver 7a side. . The angle is set counterclockwise from the reference direction when viewed from the light receiver 7a side.

Therefore, when the light reflected on the sample surface 3 is elliptically polarized as shown in FIG. 14, the light intensity I1 obtained by the photodetector 7a is the horizontal axis of the elliptically polarized light shown in FIG. The amplitude of the orthogonal projection to the (0 ° direction) is shown. The light intensity I2 obtained by the photodetector 7b indicates the amplitude of the orthogonal projection of the elliptical polarized light onto a line inclined at + 45 °. Further, the light intensity I3 obtained by the light receiver 7c indicates the amplitude of the orthogonal projection of the elliptically polarized light on the line inclined at -45 °.

Now, the output waveform of the light intensity obtained by the photodetector that detects the light in the polarization direction at the angle Ai (i = 1, 2, 3) is calculated by using the Jones vector except for a constant multiple. 6) It is shown by the equation.

[0012]

(Equation 2)

[0013] indicates incident polarized light having an amplitude ratio χ of P-polarized light and S-polarized light and a phase difference φ ₀ . j is an imaginary unit.
The first component of the equation (7) indicates a P-polarized component, and the second component indicates an S-polarized component. Also,

[0014]

(Equation 3) Is a matrix showing analyzers placed at an angle Ai. Then, in equation (6), A1 = 0 °, σ1 = σT ² _{, Φ 1 = 0 A2 = +} 45 °, σ2 = σTσR, φ 2 = 0 A3 = -45 °, σ3 = σTσR, φ 3 = 0 ... When (11), from (6), obtained in the respective channels The light intensity Ii (i = 1, 2, 3) becomes as shown in equation (12) if the gain of the light receiver is appropriately selected. I1 = I ₀ tan ² _{ψ I2 = [I 0/2} ] × [tan 2 ψ + (σT ・ σR ・ χ) ² + 2σT · σR · χ · tanψ · cos (Δ-φ 0)] I3 = [I 0/2] × [tan 2 ψ + (σT ・ σR ・ χ) ² −2σT · σR · χ · tanψ · cos (Δ−φ ₀ )] (12) where I ₀ is a constant that depends on the intensity of the incident light and the reflectance of the object to be measured.

Therefore, by solving the simultaneous equations consisting of the three equations shown in equation (12), the phase difference Δ and the amplitude ratio tanψ of the ellipsometric parameters can be calculated for each light intensity I 1,
It can be calculated from I2 and I3 by the following equations (13a) and (13b). cos (Δ−φ ₀ ) = [(I 2 −I 3) / 2I 1] × [I 1 / (I 2 + I 3 −I 1)] ^1/2 … (13a) tanψ = | σT · σR · χ | [I1 / (I2 + I3-I1)] ^1/2 … (13b)

[0016]

However, FIG.
The conventional ellipsometer described in (1) has the following problems to be improved.

That is, the shape of the elliptically polarized light is calculated using the three light intensities I1 to I3 detected by the light receivers 7a to 7c. Accordingly, if any one of the three light intensities has a large error, the error rate of the calculated ellipsometric parameters Δ and ψ is approximately equal to the error rate of the light error with the largest error. Therefore, the measurement accuracy of the ellipsometer parameter measured by the ellipsometer decreases.

For example, if the elliptical shape shown in FIG. 17 becomes more flat, only the third light intensity I3 of the three light intensities I1, I2, I3 is extremely higher than the other light intensities I1, I2. Is small.

On the other hand, the above-mentioned ellipsometric parameters ψ,
In order to calculate Δ by, for example, a computer, it is necessary to convert each of the analog light intensities I1 to I3 into a digital value by an A / D converter. Therefore, if only one light intensity has a small value, the number of significant digits of the A / D converted value decreases,
Each light intensity will contain a large error. As a result, the accuracy of the calculated ellipsometric parameters ψ and Δ decreases, and the measurement accuracy of the finally obtained film thickness d decreases.

As described above, the three-channel ellipsometer shown in FIG. 16 has no movable part, and thus is a very useful device capable of measuring the film thickness of the object to be measured at high speed. Depending on the object, there is a concern that the measurement accuracy is inferior to the ellipsometer using the rotary analyzer described above.

Note that if the number of measurements at the same measurement point is increased and the average value of the measurement results is taken, the error is reduced to some extent, but if the same measurement point is repeatedly measured, not only does the entire measurement time become longer, For example, it cannot be applied to a measurement object moving at a high speed, such as a film thickness inspection in a factory production line.

The present invention has been made in view of such circumstances, and separates reflected light having elliptically polarized light reflected by a measurement object into four or more polarization components having different polarization directions from each other. By detecting the light intensities of the components and statistically processing all the light intensities, the error rate of the calculated ellipsometric parameters can be greatly reduced, and the film thickness measurement accuracy can be improved while maintaining a high measurement speed. It is an object of the present invention to provide an ellipsometric parameter measurement method and an ellipsometer that can be significantly improved.

[0023]

In order to solve the above-mentioned problems, in the ellipsometric parameter measuring method according to the present invention, polarized light is made incident on a measuring object at a predetermined angle, and reflected lights of the measuring objects are mutually reflected. The light is separated into a plurality of different four or more polarization components, and from the light intensities of the separated plural polarization components, the phase difference Δ and the amplitude ratio tan に お け る in the reflected light are determined using a statistical method such as a least square method. The ellipsometric parameters Δ and ψ to be determined are obtained.

Further, the ellipsometer of the present invention comprises a light source section for causing polarized light to enter a measurement object at a predetermined angle, and an optical element for separating reflected light reflected by the measurement object into polarization components having four or more different polarization directions. System, and a plurality of light receivers for detecting the light intensity of each polarization component separated by this optical system,
An arithmetic unit for obtaining ellipso parameters Δ, さ れ る determined from the phase difference Δ and the amplitude ratio tan に お け る in the reflected light using a statistical method from the plurality of light intensities detected by the plurality of light receivers. I have. Further, in the ellipsometer of another invention, the calculation unit calculates the ellipsometric parameters Δ, ψ by using a statistical method represented by the following equations (1), (2), (3), and (4). tanψ = (A / C) ^1/2 … (1) cos (Δ−φ ₀ −φ _B ) = (B / C) (C / A) ^1/2 … (2) where A, B, and C are

[0025]

(Equation 4) Where x _i = cos ² _{Ai y i = 2σi · χ ·} cosAi · sinAi z i = Ii u i = - (σi · χ) 2 sin ² Ai… (4)

## EQU1 ## Further, Ai is the azimuth angle of the polarization of the reference direction and a direction parallel to the plane of incidence of the incident light to the measurement target, σi (i = 1,2,3, ...... ), φ B is the reflected light a plurality of It is a constant determined by the optical system that separates the polarization components.
φ ₀ is a constant determined by the polarization state of the incident light, and I i (i =
..) Are light intensities detected by the light receiver, and w _i (i = 1, 2, 3,...) Is a weight function. Each weighting function w _i (i = 1, 2, 3,...) Is preferably a function of the light intensity Ii of the corresponding channel i.

Further, in the ellipsometer of another invention, the optical system for separating the reflected light into polarized light components having four or more different polarization directions from each other splits the reflected light reflected from the object to be measured into light in a plurality of different directions. It is composed of a non-polarization beam splitter and a plurality of polarization beam splitters that decompose each light split by the non-polarization beam splitter into polarization components in mutually different directions.

Further, in the ellipsometer according to another invention, the optical system includes a plurality of non-polarizing beam splitters for splitting the reflected light from the object to be measured into light beams in four or more different directions. It is composed of a plurality of analyzers that transmit polarized light components in different directions in each light split by the splitter.

[0029]

First, the principle of operation of the ellipsometric parameter measuring method thus configured will be described.

As described above, when the polarized light from the light source unit enters the measurement object at a predetermined angle φ, the reflected light reflected by the measurement object becomes elliptically polarized light having a fixed shape determined by the film thickness of the measurement object. Become. And the ellipsometric parameters ψ, Δ
Is the amplitude ratio t between the P and S components of the reflected wave from the object to be measured
Since it is a value determined from anψ and the phase difference Δ, it is determined from the elliptical shape and the degree of inclination of the ellipse from the reference line. Therefore, as shown in FIG. 17, if at least three light intensities obtained by projecting an ellipse in each direction are obtained, the ellipse is uniquely determined.

Therefore, as shown in FIG. 2, it is possible to obtain projections from four or more directions with respect to the reference direction, and obtain the shape of the ellipse by using a statistical method. Note that the statistical method in this case is a method of obtaining a value that is considered to have a statistically small error from a large number of data such as a least square method, a minimum absolute deviation estimating method, an orthogonal regression method, and a simple averaging method. is there. Therefore, in this case, it is possible to obtain an elliptical shape that can be regarded as having the smallest error for all the light receivers. Therefore, the accuracy of the calculated ellipsometric parameters ψ and Δ is improved. Hereinafter, the procedure for obtaining the above equations will be described. For example, the incident light Ein

[0032]

(Equation 5) The electric field after passing through an analyzer or a polarizing beam splitter that detects light in the polarization direction with the azimuth angle Ai (i = 1, 2, 3, 4,..., N) is Ki R (Ai) PR (-Ai) BSEin (i = 1, 2,..., N). Here, Ki is a transmission coefficient of each optical path.

R (Ai) is a rotation matrix, P is a matrix representing an analyzer or a polarizing beam splitter, B is a matrix showing the influence of an element that divides the reflected light into a plurality, and S is a matrix showing the influence of an object to be measured. It is. Each of these matrices is defined as shown below.

[0034]

(Equation 6) Therefore, the light intensity Ii obtained by the light receiver is

[0035]

(Equation 7) Becomes Here, Gi is the gain of the electric system, and Kai is Kai = Ki Kp KbiRs (i = 1, 2,..., N). Here, when the above-mentioned gain Gi is determined by an appropriate method, the light intensity Ii shown in the equation (14) becomes the following equation. Ii = I ₀ [tan ² ψ ・ COS ² Ai + (σi χ) ² sin ² Ai + 2σi · χ · tanψ · cos (Δ-φ 0 -φ i) · cosAi · sinAi]

Here, I ₀ is a constant that depends on the incident light intensity, the reflectance of the object to be measured, and the like. Further, assuming that φ _i is a constant value φ _B independent of i, the light intensity I ₀ becomes
It can be shown as in equation (15). Ii = I ₀ [tan ² ψ ・ COS ² Ai + (σi χ) ² sin ² Ai + 2σi · χ · tanψ · cos (Δ−φ ₀ −φ _B ) · cosAi · sinAi] (i = 1,2,…, n)… (15) where a = tan ² ψ b = tanψ · cos (Δ -φ 0 -φ B) c = -1 / I 0 x i = cos 2 _{Ai y i = 2σi · χ ·} cosAi · sinAi z i = Ii u i = - (σi · χ) 2 sin ² Ai (i = 1,2, ..., n) and put _{a, ax i + by i + cz} i = u i (i = 1,2, ..., n) ... (16)

Is as follows. Therefore, (x _i, y _i,
When there are four or more pairs of z _i , u _i ), a, b, and c can be estimated using a statistical method. For example, 3
The error can be made smaller than that of a conventional ellipsometer which determines a, b, and c from a set of (x _i , y _i , z _i , u _i ). Next, among the statistical methods, for example, estimation by the least square method will be considered. For example, each observation data (x
_i , y _i , z _i , u _i ) and the unknown regression line u = ax + by + cz, the square error which is the sum of the squares in the u direction is represented by an error function U
(A, b, c) and this error function U (a, b, c) = {(ax _i + by _i + cz _i -u _i ) ²

The estimated values of a, b, and c are determined so that is minimized. Here, Σ is an operator for calculating the sum from i = 1 to i = n. Incidentally, as shown in equation (17), the error function U (a, b, c) as it is also possible to use a square error multiplied by the weight function w _i. U (a, b, c) = Σw i (ax i + by i + cz i -u i) 2 ... (1
7)

[0039] It should be noted that the above-mentioned _{w i} may be zero. As described above, when it is clear that the error of a certain light intensity Ii is large, it is possible to remove the light intensity Ii whose error is large from a calculation in advance by setting w _i = 0. In addition, 2 shown in equation (17)
When the square error is used, a, b, and c at which the square error U (a, b, c) is minimized are

[0040]

(Equation 8) Therefore, this equation may be solved. That is,

[0041]

(Equation 9)To obtain the following equation. a = tan^Two ψ = A / C b = tanψ · cos (Δ−φ₀−φ_B) = B / C Therefore, the above-described equations (1) and (2) are obtained. tanψ = (A / C)^1/2  … (1) cos (Δ−φ₀−φ_B) = (B / C) (C / A)^1/2  (2) where A, B, and C are expressed by the determinant (3) described above.
I get it.

[0042]

(Equation 10) Naturally, the calculation results of A, B, and C differ depending on the method of selecting the weight function w _i (i = 1, 2, 3,..., N). But,
This weighting function w _i is, for example, w _i = I i (i = 1, 2, 3,..., N)

It is conceivable that the function w _i = F (Ii) of Ii. Thus, by the function of the light intensity Ii measured the weight function w _i, by increasing the contribution ratio of the light intensity of the light intensity Ii is large channel, conversely, the light intensity Ii is smaller channel light It is possible to set the contribution ratio of the intensity to be small. As a result, the light intensity Ii
Is small, the influence of the error at the time of A / D conversion, which becomes a problem when A. B. It is possible to reduce the degree of C given to the calculation result. Therefore, finally the ellipsometric parameters Δ, 的 に
Measurement accuracy is improved. Note that the error function U (a,
It is also possible to define b, c) using an error other than the square error. Next, an example of a gain adjustment method will be described. First, the known polarization

[0044]

[Equation 11]

The light intensity output from the light receiver when the reflected light is directly incident on the optical system on which the reflected light is incident is calculated. In this case, the light intensity Ixi can be expressed in the same manner as in equation (14).

[0046]

(Equation 12) Where Li = ² ・ COS ² Ai + (σi η) ² sin ² Ai + 2 (ξ · η · σi) cos (φ _i + Φ) · cosAi · sinAi (i = 1,2,…, n) (19) In equation (19), σi, φ _i , Ai, ξ, η , Φ are all known. Therefore, Li is also known.

Next, all light intensities Ix1, Ix2,.
i, a certain value ... Ixn as I _0a, respectively, I _{0a
L1, I 0a L2, ...,} I 0a Li, ..., the gain to be equal to _{I 0a Ln G1, G2, ...} , Gi, ..., decide Gn. That is, Gi = I _0a / | Kia | ² (i = 1, 2,..., n)... (20), the aforementioned equation (14) can be expressed as equation (21). Ii = I ₀ [tan ² ψ ・ COS ² Ai + (σi χ) ² sin ² Ai -2σi · χ · tanψ · cos (Δ−φ ₀ −φ _B ) · cosAi · sinAi] (i = 1,2,..., N) (21) where I ₀ = I _0a | Ep | ² Defined. Therefore,
Equation (15) described above is obtained.

Thus, if the gain Gi (i = 1, 2,..., N) of the electric system is determined by the above-described method, the gain of the light receiver is determined without depending on the transmittance of the entire optical system. The characteristics of each light receiver can be made uniform.

The gain adjustment process is generally performed by adjusting the actual gain of an amplifier that amplifies the output signal of each photodetector. However, for example, in a data processing step of the data value of the light intensity taken into the arithmetic unit from each light receiver, the gain adjustment processing may be performed by the program processing means. Also, the known polarized light beam used for gain adjustment, unless the output of each receiver is zero,
In principle, any polarized light beam may be used.

As described above, using the four or more light intensities I 1, I 2,..., In measured at the same time, the ellipsometric parameters Δ, さ れ る estimated to have the smallest error are calculated by the statistical method. are doing. Therefore, the present invention has a feature that the measurement accuracy can be greatly improved in addition to the advantages of the conventional apparatus of FIG. 13 which can execute the film thickness measurement and the in-plane measurement of the film thickness of the moving body.

[0051]

An embodiment of the present invention will be described below with reference to the drawings.

FIG. 3 is a block diagram showing the entire ellipsometer employing the ellipsometric parameter measuring method of the embodiment. In the ellipsometer of this embodiment, the reflected light from the sample surface is separated into four polarized components (n =
4).

In the figure, reference numeral 10 denotes an ellipsometer main body housed in a case made of a light metal material. Each light intensity I1, I2, I output from the ellipsometer body 10
3 and I4 are converted into digital values by an A / D converter 11 and then input to a personal computer 12 as an arithmetic unit. The personal computer 12 calculates ellipsometric parameters ψ and Δ by the least squares method using the inputted four light intensities I 1, I 2, I 3 and I 4.
Further, using the calculated ellipsometric parameters と し て and Δ, the film thickness d of the sample surface 13 to be measured is calculated using a predetermined arithmetic expression.

Here, the A / D converter 11 time-divisionally converts the light intensities I 1, I 2, I 3, I 4 and sequentially performs A / D conversion. The conversion time of one light intensity is about 10 μse
c. Therefore, including the calculation time in the personal computer 12, the measurement time of the ellipsometric parameters ψ and Δ and the film thickness d at one sampled measurement point on the sample surface 13 is about 100 μsec. Since the light intensities I1, I2, I3, and I4 are simultaneously measured and held by the voltage holding circuit, even if the sample surface 13 moves at a high speed, it is possible to cope with the problem. FIG. 1 is an internal configuration diagram of the ellipsometer main body 10.

For example, a laser beam having a single wavelength output from the semiconductor laser light source 14 is converted into linearly polarized light by the polarizer 15. Therefore, the semiconductor laser light source 14 and the polarizer 15 constitute a light source unit 16. The incident light 17 converted into the linearly polarized light has an angle φ from the light source 16 to the sample surface 13.
Incident. The reflected light 18 reflected by the sample surface 13 changes from linearly polarized light to elliptically polarized light shown in FIG.
9 is incident.

The non-polarizing beam splitter 19 is made of, for example, a non-polarizing glass plate. Then, the incident reflected light 18 is split into two lights 20a and 20b while maintaining the elliptically polarized state without being polarized at all. The reflected light 20a is incident on one polarization beam splitter 21. Further, the transmitted light 20b that has passed enters the other polarization beam splitter 22.

Each of the polarization beam splitters 21 and 22 has the same configuration, and is composed of, for example, a Glan-Thompson prism, a Glan-Taylor prism, etc., and separates the incident elliptically polarized light into two orthogonal polarization components. And output as transmitted light and reflected light, respectively. It should be noted that a Wollaston prism or the like that separates transmitted light into two components at an angle may be used.

The polarization beam splitter 21 is arranged such that the direction of polarization of the transmitted light 21 a of the polarization beam splitter 21 is parallel to the plane of incidence of light on the sample surface 13 with respect to the above-described reference direction in which the azimuth is 0 °. It is positioned so as to be + 90 ° counterclockwise when viewed from the light receiver 23a side. Then, the transmitted light 21a having the polarization direction of + 90 ° output from the polarization beam splitter 21 is incident on the light receiver 23a. The reflected light 21b output from the polarization beam splitter 21 and having the polarization direction of 0 ° is received by the light receiver 23b.
Incident on

Further, the other polarization beam splitter 22
Are positioned such that the polarization direction of the transmitted light 22a of the polarization beam splitter 22 is + 45 ° with respect to the reference direction. Then, the polarization beam splitter 22
The transmitted light 22a having a polarization direction of + 45 ° output from the optical disc is incident on the light receiver 23c. The reflected light 2 having a polarization direction of −45 ° output from the polarization beam splitter 22.
2b enters the light receiver 23d.

Accordingly, the light intensity I1 (one channel) projected on the vertical axis of the elliptically polarized light shown in FIG. 2 of the reflected light 18 by the transmitted light 21a incident on the light receiver 23a is obtained. Also, the light intensity I2 (two channels) projected on the horizontal axis of the elliptically polarized light is obtained by the reflected light 21b incident on the light receiver 23b. Further, the light intensity I3 (3 channels) projected on a line inclined by + 45 ° with respect to the horizontal axis of the elliptically polarized light by the transmitted light 22a incident on the light receiver 23c.
Is obtained. Then, a light intensity I4 (4 channels) projected on a line inclined at -45 DEG with respect to the horizontal axis of the elliptically polarized light is obtained by the reflected light 22b incident on the light receiver 23d.

That is, the reflected light 18 from the sample surface 13 has the respective light intensities I 1, I 2, I 3 and I 4 respectively.
The light is separated into polarization components in four directions of °, 0 °, + 45 °, and -45 °.

Then, as described above, using these four light intensities I 1 to I 4, the reflected light 1 is obtained by the least squares method.
Ellipsometric parameters ψ and Δ that specify the eight elliptically polarized lights are calculated.

In the ellipsometer of this embodiment,
Enters the circularly polarized test light into the unpolarized beam splitter 19.
To adjust the gain of the electrical system. Sand
That is, in the aforementioned equation (19), | ξ | = | η | = 1, Φ
= 90 °, the ellipsometer of this embodiment
Are: φi = 0, σ1 = σ2 = σR, σ3 = σ4 = σT
Therefore, L1, L2, L3, and L4 are given by L1 = σR^Two _{L2 = 1 L3 = (1 + σT)} Two ) / 2 L4 = (1 + σT)^Two ) / 2. Therefore, each of the light receivers 23a, 23b, 23
c, 23d, the light intensities I1, I2, I3,
I4 is represented by the following equation. I1 = I₀・ ΣR^Two χ^Two  I2 = I₀・ Tan^Two ψ I3 = I₀[Tan^Two ψ + (σT ・ χ)^Two  ＋ 2σT ・ χ ・ tan co ・ cos (Δ−φ₀−φ_T)] I4 = I₀[Tan^Two ψ + (σT ・ χ)^Two  −2σT · χ · tanψ · cos (Δ−φ₀−φ_T)] …(twenty two)

Here, I ₀ is a constant that depends on the intensity of the incident light and the reflectance of the object to be measured. The phase difference φ ₀ and the amplitude ratio χ are parameters determined by the polarization state of the incident light with respect to the measurement object. In general, in order to facilitate the calculation, the phase difference φ ₀ = 0 ° The amplitude ratio χ = 1

Is set. ΣR is the amplitude reflectance ratio of the non-polarizing beam splitter 19, and σT is the amplitude transmittance ratio. Further, φ _T = 0. Then, the amplitude reflectance ratio σR and the amplitude transmittance ratio σT are expressed by Expression (23). Amplitude reflectance ratio σR = r _s / r _P amplitude transmittance ratio σT = (1-r _s ² ) / (1-r _P ² ) …(twenty three)

Here, r _P and r _s are P-polarized light and S, respectively.
These are the Fresnel reflection coefficients for polarized light, which are determined by the refractive index and the incident angle of the non-polarizing beam splitter 19.

If the refractive index of air is N ₀ , the refractive index of the non-polarizing beam splitter 19 is n _1, and the incident angle and the refracting angle to the non-polarizing beam splitter 10 are θ ₀ and θ ₁ , respectively, r _P , r _s is given by (24) (25). r _P = (n ₁ cos θ ₀ −N ₀ cos θ ₁ ) / (n ₁ cos θ ₀ + N ₀ cos θ ₁ ) (24) r _s = (N ₀ cos θ ₀ −n ₁ cos θ ₁ ) / (N ₀ cos θ ₀ + n ₁ cosθ ₁ )… (25)

Each of these constants is such that test light having a known linearly polarized light or elliptically polarized light is incident on the non-polarizing beam splitter 19, and the true ellipsometric parameters ψ, Δ
Is calculated in advance from the amount of deviation from. If the refractive index n _{1 of} the non-polarizing beam splitter 19 is a real number, φ
Note that _T = 0.

Personal computer 1 as arithmetic unit
Numeral 2 denotes a digital 4 which is input from the ellipsometer main body 10 via the A / D converter 11 in accordance with the flowchart of FIG.
From the respective light intensities I1 to I4, the film thickness d on the sample surface 13 is obtained.
Is calculated.

That is, when the flow chart is started, the four light intensities I1 to I4 are read. Next, the ellipsometric parameters ψ and Δ are calculated by substituting the four light intensities into the aforementioned equations (1) and (2). However, the determinant of equation (3) is expanded in advance.
When the ellipsometric parameters ψ and Δ are obtained, the film thickness d on the sample surface 13 is calculated using a separate calculation formula.

In the ellipsometer configured as described above, even if the measured four light intensities I1 to I4 include light intensities including a large error, the error is statistically calculated using all the light intensities. Ellipso parameters ψ and Δ which can be regarded as the smallest are calculated. Therefore, the accuracy of the calculated ellipsometric parameters ψ and Δ is improved.

As described above, since the ellipsometric parameters are calculated using the statistical method, the ellipsometric parameters ψ, Δ
As compared with a conventional three-channel ellipsometer which uses three light intensities I1 to I3 fixed in advance for the calculation of the above, the measurement accuracy can always be maintained at a higher value than a certain level. That is, the fluctuation of the measurement accuracy due to the measurement object and the measurement conditions is small, and the stable and high measurement accuracy can be always maintained.

Each optical component shown in FIG. 1 is fixed to a substrate, for example, and has no movable parts. Therefore, the measurement time required for one measurement point can be regarded as only the conversion time of the A / D converter 11 and the calculation processing time in the personal computer 12, so that about 1
00 μsec, which can be measured almost in real time. Therefore, even if the object to be measured is moving at high speed, the film thickness d can be measured correctly. FIG. 5 is a view showing a state in which the ellipsometer of the present invention is incorporated in an apparatus for measuring the distribution of oxide film thickness of a silicon wafer.

A moving table 32 is provided on a base 31, and a rotating support 33 is mounted on the moving table 32. Then, a silicon wafer 35 to be measured is mounted on the rotary support 33 by, for example, a suction mechanism. Therefore, silicon wafer 3
5 moves linearly in the direction of the arrow while rotating. Base 31
An existing thickness measuring device 36 for measuring the thickness of the entire silicon wafer 35 is provided thereon, and an ellipsometer main body 37 is fixed by a support member 38 at a position facing the thickness measuring device 36.

The thickness measuring device 36 and the ellipsometer main body 37 are connected to the moving table 32 and the rotary support 3.
In step 3, the entire thickness and the thickness d of the oxide film at the measurement positions (R, θ) of the silicon wafer 35 spirally moving are measured.

FIG. 6 is a flowchart showing a measurement process performed by the personal computer 12 incorporated in the ellipsometer. When the flowchart starts, the silicon wafer 3
5 is initialized. Next, each light intensity I1 to I4 at the corresponding measurement position is read. The four light intensities thus read are substituted into the above-mentioned equations (1) and (2) to calculate ellipsometric parameters ψ and Δ.

When the ellipsometric parameters ψ and Δ are obtained,
The film thickness d and the refractive index at the measurement position (R, θ) on the silicon wafer 35 are calculated using a separate calculation formula.
When the measurement of the film thickness d and the refractive index at one measurement position is completed, the measurement position (R, θ) is moved and the measurement is performed again. When the measurement processing at all the measurement positions is completed, the measurement of one silicon wafer 35 is completed.

FIGS. 7A and 8 show the phase difference Δ obtained by the rotating analyzer and the calculated phase difference when the film thickness is measured for a large number of silicon wafers 35 having various film thicknesses d. It is a figure showing relation with Δ. FIG. 7 (a)
8 shows the experimental results using the example ellipsometer, and FIG.
Shows the experimental results using a conventional ellipsometer.

As shown in the figure, in the conventional ellipsometer shown in FIG. 8, when the phase difference Δ at which one of the three light intensities I 1 to I 3 becomes extremely small is near 0 ° and 180 °. , An error reaching 10 ° to 12 ° occurs.

On the other hand, in the ellipsometer of the embodiment shown in FIG. 7A, an ellipso parameter which can be regarded as having a statistically small error is calculated using all the light intensities I1 to I4. Is 0 ° and 180 °
Even in the vicinity, the absolute error could be reduced to 5 ° or less.

As described above, the ellipsometer is downsized while maintaining high speed, and the measurement accuracy is increased.
It can be additionally installed on the existing thickness measuring device 36.
In a semiconductor process line, in addition to the above-mentioned silicon wafer, application to online measurement of a nitride film, a polysilicon film, a transparent electrode material and the like is possible.

FIG. 9 is a diagram showing a schematic configuration of an ellipsometer according to another embodiment of the present invention. 1 are given the same reference numerals. Therefore, the detailed description of the overlapping part is omitted.

In this embodiment, the reflected light 20a of the non-polarizing beam splitter 19 is
+ 45 ° and -4 with respect to the reference direction described above.
The transmitted light 20b of the non-polarizing beam splitter 19 is separated by the polarizing beam splitter 22 into 90 ° and 0 ° directions of polarization with respect to the reference direction. In this embodiment, the light intensities of the light receivers 23c and 23d are I1 and I2, and the light intensities of the light receivers 23a and 23b are I3 and I4. The constants .sigma.1 and .sigma.2 in equation (4) are the amplitude transmittance ratios of the P-polarized light and the S-polarized light of the non-polarizing beam splitter 19, and the constants .sigma.3 and .sigma.4 are the amplitude reflectance ratios. further,
φ ₁ , φ ₂ , φ ₃ , φ ₄ are all 0.

Even with such a configuration, as in FIG. 1, the respective light intensities of the four polarized light components in each direction obtained by shifting the reflected light 18 from the sample surface 13 by 45 ° with respect to the reference direction can be obtained. Therefore, substantially the same effects as those of the above-described embodiment can be obtained.

In the embodiment shown in FIG. 10, the light source unit 16 in the ellipsometer of FIG.
A quarter-wave plate 40 is inserted in the optical path of the incident light 17 with respect to. By inserting the quarter-wave plate 40 in this manner, it is possible to convert the incident light 17 entering the sample surface 13 from linearly polarized light to circularly polarized light. Therefore, the measurement range of the film thickness d can be shifted as compared with the ellipsometer of FIG.

FIG. 11 is a schematic diagram showing a schematic configuration of an ellipsometer according to still another embodiment of the present invention. FIG.
The same reference numerals are given to the same parts as those of the embodiment. Therefore, the detailed description of the overlapping part is omitted.

In this embodiment, the optical system that separates the reflected light 18 from the sample surface 13 into four different polarized light components comprises three non-polarized beam splitters 19, 41, and 42.
And four analyzers 43a, 43b, 43c, 43d.

That is, the reflected light 18 from the sample surface 13 is split by the non-polarizing beam splitter 19 into reflected light 20a and transmitted light 20b. The reflected light 20a is split into reflected light and transmitted light by another non-polarizing beam splitter 41. The reflected light is incident on the analyzer 43a, and the transmitted light is incident on the analyzer 43b. Further, the transmitted light 20b of the non-polarization beam splitter 19 is converted into another non-polarization beam splitter 42.
Is split into reflected light and transmitted light. The reflected light is incident on the analyzer 43c, and the transmitted light is incident on the analyzer 43d.

The four analyzers 43a to 43d are respectively + 90 ° with respect to the aforementioned reference direction of the incident light, and
The polarized light components in the directions of 0 °, + 45 °, and −45 ° are transmitted. As a result, the respective light intensities I1, I of the four polarization components having different polarization directions from the respective light receivers 23a, 23b, 23c, 23d disposed behind the analyzers 43a to 43d.
2, I3 and I4 are output. Therefore, it is possible to obtain substantially the same effects as those of the above-described embodiments.

FIG. 12 is a schematic diagram showing a schematic configuration of an ellipsometer according to still another embodiment of the present invention. FIG.
The same reference numerals are given to the same parts as those of the embodiment. Therefore, the detailed description of the overlapping part is omitted.

In this embodiment, the reflected light 18 from the sample surface 13 is separated into five polarization components having different polarization directions. That is, the reflected light 18 on the sample surface 13 is split into three lights 20a, 20c, and 20d by the two non-polarizing beam splitters 19 and 19a. The reflected light 20a of the non-polarizing beam splitter 19 is incident on the analyzer 43e. The analyzer 43e transmits the polarized light component of the incident light 20a in the direction of 90 ° with respect to the reference direction and transmits the polarized light component to the light receiver 23e.
To be incident.

The reflected light 2 of the non-polarized beam splitter 19a
0c enters the polarization beam splitter 45. The polarization beam splitter 45 extracts the polarization components of the incident light 20c in the directions of 30 ° and 120 ° with respect to the reference direction, and makes the light components enter the light receivers 23a and 23b, respectively. The transmitted light 20d of the non-polarization beam splitter 19a enters the polarization beam splitter 46. The polarization beam splitter 46 extracts each polarization component of the incident light 20d in the directions of 60 ° and 150 ° with respect to the reference direction, and makes the light components 23c and 23d respectively.

As a result, each of the light receivers 23a, 23b, 23
Light intensities I1, I2, I3, I4 and I5 of five polarization components having different polarization directions are output from c, 23d and 23e. In this case, the light intensities I1, I2, I3, I4,
I5 is given by the following equation (26), similarly to equation (14). I1 = G1 | Kb1 · Ep | ² [(Σ1R ・ χ) ² sin ² 90] I2 = G2 | Kb2 · Ep | ² × [tan ² ψ ・ cos ² 30+ (σ1T ・ σ2R ・ χ) ² sin ² 30 + 2σ1T ・ σ2R ・ χ ・ tan ・ cosΔ ・ cos30 ・ cos30] I3 = G3 | Kb3 ・ Ep | ² × [tan ² ψ ・ cos ² 120 + (σ1T ・ σ2R ・ χ) ² sin ² 120 + 2σ1T ・ σ2R ・ χ ・ tanψ ・ cosΔ ・ cos120 ・ cos120] I4 = G4 | Kb4 ・ Ep | ² × [tan ² ψ ・ cos ² 60+ (σ1T ・ σ2T ・ χ) ² sin ² 60 + 2σ1T ・ σ2T ・ χ ・ tanψ ・ cosΔ ・ cos60 ・ cos60] I5 = G5 | Kb5 ・ Ep | ² × [tan ² ψ ・ cos ² 150 + (σ1T ・ σ2T ・ χ) ² sin ² 150 + 2σ1T ・ σ2T ・ χ ・ tanψ ・ cosΔ ・ cos150 ・ cos150]… (26)

Here, σ1R and σ1R are beam beams, respectively.
These are the amplitude reflectance ratio and the amplitude transmittance ratio of the splitter 19. Ma
Σ2R and σ2T are the vibrations of the beam splitter 19a, respectively.
These are the width reflectance ratio and the amplitude transmittance ratio. For example, this is
Outputs I1x, I2x, I3x, I4 when light is directly incident
x and I5x are expressed by equation (27). I1x = G1 | Kb1 |^Two [(Σ1R)^Two  sin^Two 90] I2x = G2 | Kb2 |^Two [Cos^Two 30+ (σ1T ・ σ2R)^Two  sin^Two 30] I3x = G3 | Kb3 |^Two [Cos^Two 120 + (σ1T ・ σ2R)^Two  sin^Two 120] I4x = G4 | Kb4 |^Two [Cos^Two 60+ (σ1T ・ σ2T)^Two  sin^Two 60] I5x = G5 | Kb5 |^Two [Cos^Two 150 + (σ1T ・ σ2T)^Two  sin^Two 150] (27) Therefore, each light intensity I1x, I2x, I3x, I4x, I5x is
Each constant value I_0bWhere I1x = I_0b・ L1 I2x = I_0b・ L2 I3x = I_0b・ L3 I4x = I_0b・ L4 I5x = I_0bG5, G2, G3, G4, G5 are determined so that L5 ... (28).
Where L1 = (σ1R)^Two  sin^Two 90 L2 = cos^Two 30+ (σ1T ・ σ2R)^Two  sin^Two 30 L3 = cos^Two 120 + (σ1T ・ σ2R)^Two  sin^Two 120 L4 = cos^Two 60+ (σ1T ・ σ2T)^Two  sin^Two 60 L5 = cos^Two 150 + (σ1T ・ σ2T)^Two  sin^Two 150… (29). Thus, G1, G2, G3, G4, G5 are
If determined, each light intensity I1, I2, I3, I4, I5 is
Equation (30) respectively. I1 = I₀[(Σ1R ・ χ)^Two  sin^Two 90] I2 = I₀[Tan^Two ψ ・ cos^Two 30+ (σ1T ・ σ2R ・ χ)^Two  sin^Two 30 + 2σ1T ・ σ2R ・ χ ・ tanψ ・ cosΔ ・ cos30 ・ cos30] I3 = I₀[Tan^Two ψ ・ cos^Two 120 + (σ1T ・ σ2R ・ χ)^Two  sin^Two 120 + 2σ1T ・ σ2R ・ χ ・ tanψ ・ cosΔ ・ cos120 ・ cos120] I4 = I₀[Tan^Two ψ ・ cos^Two 60+ (σ1T ・ σ2T ・ χ)^Two  sin^Two 60 + 2σ1T ・ σ2T ・ χ ・ tanψ ・ cosΔ ・ cos60 ・ cos60] I5 = I₀[Tan^Two ψ ・ cos^Two 150 + (σ1T ・ σ2T ・ χ)²  sin² 1
50 + 2σ1T ・ σ2T ・ χ ・ tanψ ・ cosΔ ・ cos150 ・ cos150]… (30)

Here, it should be noted that the directly incident light does not need to be circularly polarized. For example, it may be a known polarization such as S-polarization. However, if P-polarized light or the like is used, the output of I1 becomes 0, and the gain of I1 cannot be determined.
Cannot be used to adjust the gain.

The gain may be determined by an electronic circuit, or may be processed by a computer after fetching data. Here, the output of each light receiver is made constant by directly entering the circularly polarized light and changing the variable resistance of the amplifier attached to the light receiver. Then, when actually measuring, when inputting each light intensity into the computer and calculating, the input I1, I2, I3, I4 and I5 are respectively L1 and I5.
Each of the light intensities I1, I2, I3, I4, and I5 is newly set by multiplying L2, L3, L4, and L5. Therefore, the ellipso parameters Δ and さ れ る, which are estimated to have the smallest error statistically, can be calculated from the five light intensities I1 to I5 by using the equations (1) and (2). In this case, the weight function w _i =
Ii (i = 1, 2,..., 5).

FIG. 7B shows a phase difference obtained by a rotary analyzer when a film thickness is measured on a large number of silicon wafers having various film thicknesses d using the ellipsometer of the five-channel embodiment. FIG. 7 is a diagram illustrating a relationship between Δ and a phase difference Δ calculated by an ellipsometer of the example. It can be understood that very good results can be obtained as in the case of the ellipsometer of the four channels shown in FIG. 7A.

FIG. 13 is a schematic diagram showing a schematic configuration of an ellipsometer according to still another embodiment of the present invention. FIG.
The same parts as those of the second embodiment are denoted by the same reference numerals. Therefore, the detailed description of the overlapping part is omitted.

In this embodiment, the reflected light 18 from the sample surface 13 is separated into six polarization components having different polarization directions. That is, the reflected light 18 on the sample surface 13 is split into three lights 20a, 20c, and 20d by the two non-polarizing beam splitters 19 and 19a. And each light 2
The light beams 0a, 20c, and 20d enter the polarization beam splitters 44, 47, and 48, and are separated into polarization components in different directions.

The polarization beam splitter 44 receives the incident light 2
The polarization components in the directions of 0 ° and 90 ° with respect to the reference direction of 0a are extracted and made to enter the photodetectors 23e and 23f, respectively. Further, the polarization beam splitter 47 has an angle of 30 ° with respect to the reference direction of the incident light 20c.
Each polarization component in the 0 ° direction is extracted, and the
a, 23b. Further, the polarizing beam splitter 48 is positioned at 6 degrees with respect to the reference direction of the incident light 20d.
The polarization components in the 0 ° and 150 ° directions are extracted and made incident on the photodetectors 23c and 23d, respectively.

As a result, each of the light receivers 23a, 23b, 23
The light intensities I1, I2, I3, I4, I of six polarization components having different polarization directions from c, 23d, 23e, 23f.
5 and I6 are output. Therefore, the ellipsometric parameters Δ and さ れ る, which are estimated to have the least error statistically, can be calculated from all six light intensities I1 to I6 by using equations (1) and (2). In this case, gain adjustment can also be performed using circularly polarized light. Therefore, substantially the same effects as those of the above-described embodiments can be obtained.

Also, by not using light intensities less than a predetermined value that can be regarded as including an error among the light intensities I1 to I6, the ellipsometric parameters Δ, で き る, which can be estimated to have a statistically small error, are used. Can be calculated. In this case, the weight function w _i corresponding to the light intensity I _i not used for the calculation
May be set to 0. By doing this,
The accuracy is slightly improved compared to the case where the ellipsometric parameters Δ and ψ are calculated by using all the light intensities I1 to I6 described above.

As described above, in the present invention, the sample surface 1
An optical system that separates the reflected light 18 from 3 into four or more polarization components can be realized by combining various components.

The present invention is not limited to the above embodiments. In the embodiment, the least-squares method is used as a statistical method for calculating the most accurate ellipso parameters Δ and か ら from four or more detected light intensities, but the method is particularly limited to the least-squares method. Not a thing,
Other statistical techniques for reducing errors can also be used.

Further, the gain adjustment method used in the present invention is as follows.
Generally, it can be applied to a multi-channel ellipsometer. In the embodiment, the case of four channels and five channels has been specifically described, and it has been described that the present invention is applicable to the case of six channels. However, it is needless to say that the present invention can be applied to a conventional three-channel ellipsometer. For example, in the ellipsometer already shown in FIG. 16, the light intensities I1, I2, I3 of the respective channels can be expressed by equation (31). I1 = G1 | Kc1 · Ep | ² [Tan ² ψ] I2 = G2 | Kc2 · Ep | ² × [tan ² ψ ・ cos ² 45+ (σT ・ σR ・ χ) ² sin ² 45 + 2σT · σR · χ · tanψ · cos (Δ-φ 0) · cos45 · cos45] I3 = G3 | Kc3 · Ep | 2 × [tan ² ψ ・ cos ² (-45) + (σT ・ σR ・ χ) ² sin ² (-45) + 2σT · σR · χ · tanψ · cos (Δ-φ 0) · cos (-45) · cos (-45)] ... (31)

Here, σR and σT are the amplitude reflectance ratio and the amplitude transmittance ratio of the non-polarizing beam splitter. Here, for example, when the P-polarized light is directly incident, the outputs I1x, I2x,
I3x is, I1x = G1 | Kc1 | ² I2x = G2 | Kc2 | ² / 2 I3x = G3 | Kc3 | ² /2...(32), each output I1x, I2x, I3x is a certain value I
Against _{0c, I1x = I 0c I2x =} I 0c / 2 I3x = I 0c / 2 ... and so as to gain G1 (33). Be determined to _{G2, G3, I1 = I 0} tan 2 Ｉ I2 = I ₀ [tan ² ψ ・ cos ² 45+ (σT ・ σR ・ χ) ² sin ² 45 + 2σT · σR · χ · tanψ · cos (Δ-φ 0) · cos45 · cos45] I3 = I 0 [tan 2 ψ ・ cos ² (-45) + (σT ・ σR ・ χ) ² sin ² (−45) + 2σT · σR · χ · tanψ · cos (Δ−φ ₀ ) · cos (-45) · cos (-45)] (34) Therefore, by solving the equation (34), the above-described equations (13a) and (13b) are obtained.

The three-channel ellipsometer includes:
In addition to the one using only the non-polarizing beam splitter shown in FIG. 16, an ellipsometer using a polarizing beam splitter as shown in FIG. 14 is also conceivable. That is, the laser light output from the light source unit 16 enters the data surface 13 via the quarter-wave plate 49. Light 1 reflected from document surface 13
8 is split by a non-polarizing beam splitter 19 into light in two directions. One of the split lights enters the light receiver 23a via the analyzer 50 that allows only the polarization component in the 0 ° direction to pass. The other split light is further separated by the polarization beam splitter 22 into polarization components in + 45 ° and −45 ° directions, and is incident on the light receivers 23b and 23c, respectively.
Each light intensity I1, I2, I3 obtained by each of the light receivers 23a, 23b, 23c can be expressed by the following equation (35). I1 = G1 | Kd1 · Ep | ² [Tan ² ψ] I2 = G2 | Kd2 · Ep | ² × [tan ² ψ ・ cos ² 45+ (σTχ) ² sin ² 45 + 2σT · χ · tanψ · cos (Δ-φ 0) · cos45 · cos45] I3 = G3 | Kd3 · Ep | 2 × [tan ² ψ ・ cos ² (-45) + (σT ・ χ) ² sin ² (−45) + 2σT · χ · tanψ · cos (Δ−φ ₀ ) · cos (−45) · cos (−45)] (35) Here, for example, the receiver output I1x when P-polarized light is directly incident. , I2x, I3x are given by: I1x = G1 | Kd1 | ² I2x = G2 | Kd2 | ² / 2 I3x = G3 | Kd3 | ² /2...(36) Since each output I1x, I2x, I3x is a certain value I
Against _{0d, I1x = I 0d I2x =} I 0d / 2 I3x = I 0d / 2 ... (37) and so as to gain G1. It is determined to G2, G3, as in the case described above, I1 = I ₀ tan ² Ｉ I2 = I ₀ [tan ² ψ ・ cos ² 45+ (σTχ) ² sin ² 45 + 2σT · χ · tanψ · cos (Δ-φ 0) · cos45 · cos45] I3 = I 0 [tan 2 ψ ・ cos ² (-45) + (σT ・ χ) ² sin ² (−45) + 2σT · χ · tanψ · cos (Δ−φ ₀ ) · cos (−45) · cos (−45)] (38) Therefore, by solving this equation (38), (3
Equation 9) is obtained. cos (Δ−φ ₀ ) = [(I 2 −I 3) / 2I 1] × [I 1 / (I 2 + I 3 −I 1)] ^1/2 tanψ = | σT · χ | [I1 / (I2 + I3-I1)] ^1/2 (39) In addition, when circularly polarized light is used as a light beam used for gain adjustment, the receiver outputs I1x, I2x, and I3x are I1x = G1 | Kd1 | ² I2x = G2 | Kd2 | ² (1 + σT ² ) / 2 I3x = G3 | Kd3 | ² (1 + σT ² ) / 2 (40), the outputs I1x, I2x, and I3x have a constant value I
Against _{0d, I1x = I 0d I2x =} I 0d (1 + σT 2 _{) / 2 I3x = I 0d (} 1 + σT 2 ) / 2 (41) so that the gains G1. If G2 and G3 are determined, the aforementioned equation (39) is obtained.

Further, as the three-channel ellipsometer, in addition to the structure shown in FIGS. 16 and 14, an ellipsometer using a composite polarization beam splitter 51 as shown in FIG. 15 is also conceivable. Composite polarized beam splitter 51
As shown in the figure, a non-polarizing glass 51a and a polarizing beam splitter 51b are integrally formed.

That is, the laser light output from the light source unit 16 enters the document surface 13. The reflected light 18 from the data surface 13 is incident on the non-polarizing glass 51a constituting the composite polarizing beam splitter 50 at a Brewster angle θ.
The light incident at the Bleester angle θ is separated into reflected light reflected on the incident surface and transmitted light entering the inside. The reflected light is only a polarization component polarized in a direction parallel to the incident surface (reflection surface). Therefore, the reflected light becomes only the polarization component in the reference direction (0 ° direction) and is incident on the light receiver 23a.

On the other hand, the transmission into the non-polarizing glass 51a
The light is transmitted to the substrate by the polarization beam splitter 51b.
Two deviations of + 45 ° and -45 ° with respect to the sub-direction
The light components are separated into light components 23b and 23c, respectively.
Incident. Obtained with each of the photodetectors 23a, 23b, 23c
Each of the light intensities I1, I2, and I3 can be expressed by equation (42).
Noh. I1 = G1 | Ke1 · Ep |^Two (ΣRχ)^Two  I2 = G2 | Ke2 · Ep |^Two  × [tan^Two ψ ・ cos^Two 45+ (σT ・ χ)^Two  sin^Two 45 ＋ 2σT ・ χ ・ tanψ ・ cos (Δ−φ₀) ・ Cos45 ・ cos45] I3 = G3 │Ke3 ・ Ep │^Two  × [tan^Two ψ ・ cos^Two (-45) + (σT ・ χ)^Two  sin^Two (-45) + 2σT ・ χ ・ tanψ ・ cos (Δ−φ₀) · Cos (−45) · cos (−45)] (42) Here, circularly polarized light was used as a light beam used for gain adjustment.
In the case, the outputs I1x, I2x and I3x of the light receiver are I1x = G1 | Ke1 |^Two ・ ΣR^Two  I2x = G2 | Ke2 |^Two (1 + σT^Two ) / 2 I3x = G3 | Ke3 |^Two (1 + σT^Two ) / 2 (43), the outputs I1x, I2x, and I3x have a constant value I
_0eWhere I1x = I_0e・ ΣR^Two  I2x = I_0e(1 + σT^Two ) / 2 I3x = I_0e(1 + σT^Two ) / 2 (44) so that the gains G1. If G2 and G3 are determined,
Equation (42) can be expressed as in equation (45). I1 = I₀(ΣRχ)^Two  I2 = I₀[Tan^Two ψ ・ cos^Two 45+ (σT ・ χ)^Two  sin^Two 45 ＋ 2σT ・ χ ・ tanψ ・ cos (Δ−φ₀) ・ Cos45 ・ cos45] I3 = I₀[Tan^Two ψ ・ cos^Two (-45) + (σT ・ χ)^Two  sin^Two (-45) + 2σT ・ χ ・ tanψ ・ cos (Δ−φ₀) · Cos (-45) · cos (-45)]… (45) Therefore, by solving this equation (45), equation (46) is obtained.
You. cos (Δ-φ₀) = [ΣR^Two (I2 -I3) / (2.sigma.T.I1)]

× [I 1 / ｛σR^Two (I2 + I3)-? T^Two ・ I1｝]^1/2 _{tanψ = | χ | [｛σR} Two (I2 + I3)-? T^Two ・ I1｝ / I1]^1/2  … (46) Thus, the present invention is suitable for ellipsometers of various structures.
It is possible to use

[0112]

As described above, according to the ellipsometric parameter measuring method and the ellipsometer of the present invention, the reflected light having the elliptically polarized light reflected by the object to be measured is separated into four or more polarized components having different polarization directions. Then, the light intensity of each polarized light component is detected, and by using all the detected light intensities,
The ellipsometric parameters Δ and せ る, which can be regarded as having the highest statistical accuracy, are calculated. Therefore, it is possible to always obtain a high film thickness measurement accuracy higher than a certain level while maintaining a high measurement speed.

[Brief description of the drawings]

FIG. 1 is a schematic diagram showing a structure inside a main body of an ellipsometer according to one embodiment of the present invention;

FIG. 2 is a diagram showing elliptically polarized light of reflected light in an example ellipsometer;

FIG. 3 is a schematic configuration diagram of the entire ellipsometer of the embodiment;

FIG. 4 is a flowchart showing the operation of the ellipsometer of the embodiment;

FIG. 5 is a schematic configuration diagram of an apparatus for measuring an oxide film thickness of a silicon wafer using an example ellipsometer,

FIG. 6 is a flowchart showing the operation of the oxide film thickness measuring apparatus;

FIG. 7 is a diagram showing a relationship between the phase difference Δ obtained by the ellipsometer of the embodiment and the phase difference Δ obtained by the rotation analyzer method.

FIG. 8 is a diagram showing a relationship between a phase difference Δ obtained by a conventional ellipsometer and a phase difference Δ obtained by a rotation analyzer method.

FIG. 9 is a diagram showing a structure inside a main body of an ellipsometer according to another embodiment of the present invention;

FIG. 10 is a view showing the structure of an ellipsometer according to another embodiment;

FIG. 11 is a view showing a structure of an ellipsometer according to still another embodiment;

FIG. 12 is a view showing a structure of an ellipsometer according to still another embodiment;

FIG. 13 is a view showing a structure of an ellipsometer according to still another embodiment;

FIG. 14 is a view showing a structure of an ellipsometer according to still another embodiment;

FIG. 15 is a view showing a structure of an ellipsometer according to still another embodiment;

FIG. 16 is a schematic diagram showing a schematic configuration of a conventional ellipsometer.

FIG. 17 is a view showing elliptically polarized light of general reflected light.

[Explanation of symbols]

10: Ellipsometer body, 11: A / D converter,
12 personal computer, 13 sample surface, 14
Semiconductor laser light source, 15: polarizer, 16: light source unit, 17
... incident light, 18 ... reflected light, 19, 19a ... non-polarizing beam splitter, 21, 22, 44, 45, 46, 47, 4
8 ... polarization beam splitter, 23a to 23f ... light receiver,
35: silicon wafer, 40, 49: quarter wave plate, 43a to 43d, 50: analyzer, 51: composite polarizing beam splitter.

────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tomoyuki Kaneko 1-2-1, Marunouchi, Chiyoda-ku, Tokyo Inside Nippon Kokan Co., Ltd. (56) References JP-A-62-251629 (JP, A) JP-A-2 -62930 (JP, A) JP-A-3-160331 (JP, A) JP-A-5-71923 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01B 11/06 G01J 4/04

Claims (7)

(57) [Claims] 1. A polarized light is made incident on a measurement object at a predetermined angle, and reflected light of the measurement object is separated into four or more different polarization components, respectively, different from each other. Ellipsometric parameters Δ, エ リ determined from the phase difference Δ and the amplitude ratio tan に お け る in the reflected light using a statistical method from the light intensity of the ellipsometric parameters. 2. The method according to claim 1, wherein the statistical method is a least squares method. 3. A light source unit for causing polarized light to enter a measurement object at a predetermined angle, an optical system for separating reflected light reflected by the measurement object into polarization components having four or more polarization directions different from each other, and the optical system. A plurality of light receivers for detecting the light intensity of each polarized light component separated by the system, and a phase difference Δ in the reflected light using a statistical method from the plurality of light intensities detected by the plurality of light receivers. And an arithmetic unit for calculating ellipsometric parameters Δ and 決定 determined from the amplitude ratio tanψ. 4. A light source unit for causing polarized light to enter a measurement target at a predetermined angle, an optical system for separating reflected light reflected by the measurement target into polarization components having four or more polarization directions different from each other, and this optical system A plurality of light receivers that detect the light intensity of each polarized light component separated by the system, and a plurality of light intensities detected by the plurality of light receivers, the following equations (1) (2) (3) (4) And an arithmetic unit for obtaining ellipsometric parameters Δ, さ れ る determined from the phase difference Δ and the amplitude ratio tan に お け る in the reflected light using the following. tanψ = (A / C) ^1/2 … (1) cos (Δ−φ ₀ −φ _B ) = (B / C) (C / A) ^1/2 (2) where A, B, and C are given by: Where x _i = cos ² _{Ai y i = 2σi · χ ·} cosAi · sinAi z i = Ii u i = - (σi · χ) 2 sin ² Ai ... (4). Ai is the azimuthal angle of polarized light with the direction parallel to the plane of incidence of the incident light on the measurement object as the reference direction, and σi (i =
1,2,3, ......), φ _B is a constant determined by the optical system for separating the reflected light into a plurality of polarized light components, χ, φ ₀ is a constant determined by the polarization state of the incident light, Ii (i = 1,2,3, ……)
Is the light intensity detected by the light receiver, w _i (i = 1,2,3,
……) is a weight function. 5. The ellipsometer according to claim 4, wherein each of the weighting functions w _i (i = 1, 2, 3,...) Is a function of the light intensity Ii of the corresponding channel i. 6. The non-polarizing beam splitter, which splits the light reflected by the measurement target into light in a plurality of directions different from each other, and separates each of the lights split by the non-polarizing beam splitter. 5. The ellipsometer according to claim 4, comprising a plurality of polarization beam splitters for decomposing into polarization components of different directions. 7. The optical system includes a plurality of non-polarizing beam splitters for splitting reflected light reflected by the measurement target into lights in four or more different directions, and splitting by each of the non-polarizing beam splitters. 5. The ellipsometer according to claim 4, comprising a plurality of analyzers for transmitting polarized light components of different directions in each light.