JPH08327566A - Method and device for quantitative determination in total reflection x-ray fluorescence analysis - Google Patents

Abstract

PURPOSE: To enable quantitative determinations to be stably made without using a standard sample and being affected by changes in intensity of an irradiation X-ray by quantitatively determining the concentration of an element to be measured, using the fluorescent X-ray relative sensitivity coefficient of the element to be measured and an element of known concentration and shape other than the element to be measured and using either X-ray penetration length or X-ray reflectance. CONSTITUTION: It is determined whether the distribution state of an element to be measured shows particle-like contamination, surface contamination, or internal contamination (exponentially diffusing contamination), from changes in X-ray fluorescence intensity when the incidence angle of an irradiation X-ray is varied, and a concentration determining method suited for the shape of that impurity is applied as follows, that is, using the fluorescent X-ray intensity of the element to be measured which emanates from a sample to be measured and using the fluorescent X-ray intensity of an element of known concentration and shape other than the element to be measured, the relative sensitivity coefficient of the element to be measured and the latter element and either X-ray penetration length or X-ray reflectance are calculated, and the use of these in determining the concentration of the element to be measured results in accurate analysis.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a quantitative method and apparatus for total reflection X-ray fluorescence analysis. In the field of semiconductor manufacturing technology in recent years, with the high integration of semiconductor devices, there is a demand for reduction of metal pollution that affects the reliability of the device and particle pollution that is greatly related to the yield in the manufacturing process. It is desired to develop analytical methods for assessing pollution.

In order to meet this demand, the concentration of contaminant elements on the surface of a semiconductor wafer or the like is non-destructively set to 10 ⁹ atoms /
For the purpose of evaluating the cleanliness of a semiconductor wafer by a total reflection fluorescence analysis method capable of measuring with high sensitivity of about cm ² when manufacturing a semiconductor wafer or when manufacturing a semiconductor device using the semiconductor wafer. Has been introduced. The fluorescent X-rays obtained by the total reflection fluorescent X-ray analysis method are suitable for the evaluation of trace elements because they have high intensity, good parallelism, and selectivity of the incident X-ray wavelength.

[0003]

2. Description of the Related Art In the conventional total reflection X-ray fluorescence analysis method, an X-ray is irradiated onto a sample at a small incident angle, and the fluorescent X-ray generated from the sample is detected. Although trace elements in a sample are detected by analyzing the X-ray, X-rays are incident at a minute incident angle ψ (0.05 to 0.1 °) at which the X-rays are totally reflected. The intensity of the X-ray standing wave on the sample and the intensity of the X-ray in the sample fluctuated due to an error in setting the incident angle of the rays, which resulted in an error in quantifying the contamination concentration.

Further, as described above, since the X-ray standing wave is generated on the sample, the intensity of the fluorescent X-ray excited by the shape of the contaminant (particle, surface contamination, internal contamination) contained in the sample is different. There was an error in the concentration determination.

In the quantification method using total reflection fluorescent X-rays which is generally used at present, the fluorescent X-ray intensity from the element to be measured is directly compared with the fluorescent X-ray intensity of a standard sample whose surface is uniformly polluted. However, it is difficult to make such a standard sample because the difference in the shape of the contaminants and the error in the setting of the incident angle of X-rays during measurement deteriorate the quantitative accuracy. In addition, there are many problems such as the need to determine the concentration of pollutants in this standard sample by chemical analysis.

[0006]

Therefore, when the conventional total reflection X-ray fluorescence analysis method is introduced in line control of the degree of contamination in the manufacturing process of semiconductor devices, the standard sample is contaminated during use and quantitative errors occur. It is difficult to compare the concentration of impurities at different manufacturing sites. The present invention, once the relative sensitivity coefficient of fluorescent X-rays from each element is obtained, establishes a stable quantitative method that is not affected by the intensity change of irradiated X-rays without using a standard sample for comparison. By doing so, it is an object of the present invention to provide a means by which the total reflection X-ray fluorescence analysis method can be popularized.

Further, according to the present invention, the shape of the element to be measured is classified into a uniform element, a particle, a surface element, and an internal diffusion element based on the incident angle dependence of the fluorescent X-ray intensity of the element to be measured from the sample. However, it is an object of the present invention to provide a quantitative method and a quantitative apparatus for total reflection X-ray fluorescence analysis, by which a more accurate measurement value can be obtained by applying a concentration quantitative method corresponding to the shape of the impurity.

[0008]

In order to realize the above-mentioned total reflection fluorescent X-ray analysis method, in the quantitative method of total reflection X-ray fluorescence analysis according to the present invention, fluorescence X of an element to be measured from a sample to be measured is measured. Using the line intensity and the fluorescent X-ray intensity of an element of which concentration and shape are known other than the element to be measured, the relative sensitivity coefficient of the element to be measured and the element other than the element to be measured, the X-ray penetration length, or A procedure for quantifying the concentration of the element to be measured using X-ray reflectance was adopted.

In this case, the fluorescent X-ray intensity of an element whose concentration and shape are known other than the element to be measured can be used as the fluorescent X-ray intensity of the sample substrate to be measured.

Further, in this case, the fluorescent X-ray intensity of an element having a known concentration and shape other than the element to be measured can be used as the fluorescent X-ray intensity of the standard sample.

In another quantitative method for total reflection X-ray fluorescence analysis according to the present invention, the shape of the element to be measured is determined from the dependence of the fluorescent X-ray intensity of the element to be measured from the sample to be measured on the incident angle. A procedure was adopted in which the element was classified into uniform element, particulate contamination, surface contamination, and internal contamination, and the concentration quantification method suitable for the shape of each element to be measured was applied.

Further, in the total reflection X-ray fluorescence analyzer according to the present invention, means for detecting the fluorescent X-ray intensity from the sample to be measured, penetration length of X-rays into the sample to be measured, X-ray reflectance or Means for calculating X-ray intensity distribution or obtaining from accumulated data, fluorescent X-ray intensity of element to be measured, penetration length of X-ray into the sample to be measured, X-ray reflectance or X-ray intensity distribution and element to be measured A structure having means for quantifying the concentration of the element to be measured is adopted from the relative sensitivity coefficient of.

[0013]

The problems of the conventional total reflection X-ray fluorescence detection method can be solved by the following method. The contaminated shape of the sample is evaluated from the change in the fluorescent X-ray intensity when the incident angle of the irradiated X-ray is changed, and a quantitative method suitable for the shape is used. The fluorescent X-ray intensity is standardized and evaluated by the reference element such as the substrate element of the sample so that it is not easily affected by the change of the incident angle of the irradiated X-ray.

Specifically, in the solution, the distribution shape of the pollutant is (a) a uniform element, (b) a particle, based on the incident angle dependence of the fluorescent X-ray intensity of the element to be measured from the sample.
It can be classified into either (c) surface element or (d) internal diffusion element. Further, in the solution, the ratio of the fluorescent X-ray intensity from the substrate element or the element uniformly distributed on the substrate and the fluorescent X-ray intensity from the element to be measured is calculated, and the quantitative determination is performed based on the value. At this time, the concentration of the element to be measured can be quantified using an element whose concentration is known, such as a substrate element.

The measurement principle of the quantitative method of total reflection X-ray fluorescence analysis according to the present invention will be described below. The reflection and transmission of X-rays incident on the surface of the sample can be described by Fresnel's law using the complex index n of the sample. The refractive index of the substance for the X-ray region is slightly smaller than 1, and n = 1-
It can be expressed as δ + iβ. Here, δ and β can be expressed as follows using the atomic scattering factors Z + f ′, f ″: δ = cλ ² ρ (Z + f ′) / A, β = cλ ² ρ (f ″ / A) , C = r _e N ₀ / 2π (1)

Where r _e is the classical radius of the electron, N ₀ is the Avogadro number, λ is the X-ray wavelength, ρ is the density, A is the atomic weight, and Z is Z.
Is an atomic number, f ′ is a dispersion correction term, and f ″ is an absorption term.
These atomic scattering factors are listed in the table by Henke (BL.
Henke, J .; C. Davis, E .; M. Gulli
kson and R.K. C. C. Perera, Law
lens Berkeley Lab. Rep. LB
L-26259 (1988), M.A. N. Thomas,
J. C. Davis, C.I. J. Jacobsen an
d R. C. C. Perera, Lawrence B
erkeley Lab. Rep. LBL-27668
(See (1989)) and a table by Sasaki (S. Sasaa).
ki: KEK Report 88-14, Febru
ary (1089)). From the above equation, the refractive index is proportional to the density of the sample and is not affected by the chemical bond state.

When X-rays are incident on a flat substrate, the incident X-rays and the reflected X-rays interfere with each other on the substrate to generate a standing wave. Since the refractive index is smaller than 1, the total reflection critical angle ψc (=
Below √ (2δ)), X-rays undergo total internal reflection, so that only evanescent waves that decay exponentially from the surface exist in the substrate. Therefore, the X-rays do not excite the substrate so much, and the elements existing near the surface are detected with high sensitivity.

When the unit incident X-ray flux is incident on the substrate at an incident angle (angle formed by the irradiation plane and the incident X-ray flux) ψ, the phase change of the in-plane X-rays is omitted and the incident X-rays are omitted.
Using the electric fields of the X-ray and the reflected X-ray as E ₀ , the reflection coefficient r of E _r , the phase difference φ at the surface, and the position z from the surface, E ₀ = exp
(−ik _z z), E _r = rexp (iφ) ezp (ik
_z z), the X-ray intensity F (ψ, z) above the substrate
Is F (ψ, z) = (E ₀ + E _r ) ² = 1 + R + 2√ (R) cos (φ−4πz / λ) (z ≧ 0) (2) Where k _z is the z component of the wave vector, R = r ²
Is the reflectance.

The X-ray intensity in the substrate can be expressed as F (ψ, z) = T (ψ) exp (z / z _x ) (z <0) (3). Where T (ψ) ≡F (ψ, 0)
Is the X-ray intensity on the substrate surface, and z _x (cm) is the penetration length of X-rays.

Further, R, φ, T, z_xEtc. are Fresnel's formula
(LG Parrat, Phys.
Rev. 95, 359 (1954)), on the sample surface
Given a roughness σ with a Gaussian distribution, the reflectance is
R '= Rexp (-q_zq ′ _zσ²). here
q_z, Q ′_zIs the momentum transfer vector on and in the substrate
Leq_z= (4π / λ) √ (n²-Cos²ψ)
(L. Nevotte and P. Croce, Rev.
Phys. Appl. 15, 761 (1980), S.M.
K. Shinha, E .; B. Sirota and
S. Garoff, Phys. Rev. B. 38,22
97 (1988)). Also, even in the case of a multilayer film sample, X
The line intensity can be calculated similarly (B. Vidal
 and P.D. Vincent, Appl. Optic
s, 23, 1794 (1984)).

FIG. 1 is an explanatory view of the X-ray intensity distribution above and inside the sample due to unit X-ray fluxes incident at different angles. In this figure, the horizontal axis represents the vertical distance of the sample, and the vertical axis represents the X-ray intensity. This figure is
Different angles (0.05 degree, 0.17 degree,
Z of the X-ray intensity (relative value) F (ψ, z) above and inside the sample due to the unit X-ray flux incident at 0.3 degrees)
The distribution is shown. Here, Si with surface roughness of 3Å
Calculated for the board.

When the vicinity of the surface is evaluated, the reabsorption of the fluorescence intensity can be ignored, so that the fluorescent X-ray yield I (ψ) is the element distribution Φ (z) (1 / cm ³ ) and the X-ray intensity I ₀ F. It can be expressed as an integral. That is, I (ψ) = ωεI ₀ ∫ (−∞ to ∞) F (ψ, z) Φ (z) ds (4) where ω is the fluorescence yield, ε is the detection efficiency, and I ₀ is the incident. X-ray flux. With synchrotron radiation, the divergence angle of incident X-rays is small and its effect can be ignored. From the dependence of the X-ray intensity on the incident angle in FIG. 1, it is clear that the distribution shape of the elements is reflected on the dependence of the fluorescence yield on the incident angle.

The measurement was carried out at a synchrotron radiation facility (KEK.PF beamline 17 of the Institute for High Energy Physics). Specifically, the emitted light is emitted from the Si wafer (11
1) Convert to a monochromatic color with a wavelength of 1.2Å using a double crystal monochromator
It was incident on a sample placed upright in a vacuum chamber. The semiconductor detector (SSD) was arranged in the horizontal direction with little scattered X-rays by using the linearly polarized light of the emitted light to improve the lower limit of detection. Incident X-ray size is 0.1mmW × 4mm with slit
The X-ray divergence angle when narrowed down to H is about 0.1 mrad, which is one digit smaller than the divergence angle of 1 mrad of the commercially available device, and the influence can be almost ignored.

To adjust the incident angle, the critical angle of total reflection of the substrate is calculated, and the measured value is corrected to determine the incident angle with an accuracy of about 2/1000 °. In this measurement, the detection sensitivity is good in the part where the incident angle of the background from the substrate is low is 0.1 ° or less, and the current concentration conversion coefficient of the surface contamination element is at the X-ray incident angle of 0.05 °. Ni
At 36 cps / 10 ¹² atoms / cm ² and a corresponding lower limit of detection of 2 × 10 ⁹ atoms / cm ² .

Fluorescence X generated when the surface of a sample made of a substrate such as Si having various pollutant elements is irradiated with X-rays.
The intensity distribution of the line will be described. (A) Substrate element In the case of a constituent element of the substrate such as Si or an element which is uniformly present in the substrate, Φ (z) = Φ ₀ (z ≦ 0) (ato
ms / cm ³⁾ a and, since the X-ray intensity in the substrate is (3), the fluorescence yield (3), from _{(4), I B (ψ} ) = (ωε) B I 0 T (ψ ) · Φ ₀ · z _x (ψ) (5), which is proportional to the X-ray intensity and the X-ray penetration length on the substrate surface.
Note that z _x (ψ) is the X-ray penetration length, and T (ψ) is the X-ray intensity on the sample surface, which can be calculated.

FIG. 2 shows the Si fluorescence yield and N from the Si substrate.
It is an explanatory view of incident angle dependence of i fluorescence yield. In this figure, the horizontal axis represents the X-ray incident angle, and the vertical axis represents the Ni fluorescence yield and the Si fluorescence yield. This figure shows the agreement between the incident angle dependence of the Si fluorescence yield from the Si substrate and the result calculated by (5).

(B) Surface contamination The distribution function of the surface element has a concentration of φs (atoms / cm).
² ) can be expressed as Φ (z) = φs · δ (z), the fluorescence yield can be calculated from (4) as I _S (ψ) = (ωε) _S I ₀ T (ψ) · φ _S ···・ It becomes (6) and it is proportional only to the X-ray intensity of the substrate surface. A standard surface contamination sample is obtained by treating a Si wafer with aqueous ammonia hydrogen peroxide containing metal impurities, and the metal element is present in a natural oxide film of about 1 nm. FIG. 2 shows the incident angle dependence of the Ni fluorescence yield from the standard sample using Ni as the additive metal and the result of the fit in the equation (6).

(C) Particle Contamination Fluorescence from particles is excited by the standing wave (2) on the substrate and has an angle dependence depending on the size. For rectangular particles with height z ₀ , Φ (z) = φ _P /
z ₀ (0 <z <z ₀ ) and the fluorescence yield is
From (2) and (4), I _P (ψ, z ₀ ) = (ωε) _P I ₀ Φ _P {1 + R− (√ (R) λ / 2πψz ₀ ) [sin (φ-4πψz ₀ / λ) − sin (φ)]} (7)

Here, the first two terms are fluorescent X-rays excited by incident X-rays, and the second term is intensity vibration due to a standing wave. The latter decreases in proportion to the size z ₀ ,
The fluorescence yield from particles larger than m is almost determined only by the preceding two terms, and becomes I _P (ψ) = (ωε) _P I ₀ (1 + R (ψ)) Φ _P (8) In this specification, a portion proportional to (1 + R) is referred to as an asymptotic form of particles. Note that R
(Ψ) is the reflectance and can be calculated.

FIG. 3 is an explanatory diagram of size dependency and incidence angle dependency of fluorescence yield of rectangular particulate contamination. In this figure, the horizontal axis represents the X-ray incident angle and the vertical axis represents the particle fluorescence yield. For hemispherical particles, the vibration is slightly weaker than this. Since there is a distribution in size in actual contamination, the sum of particles having an appropriate size width is fitted to the data as I _P (ψ) = ΣI _P (ψ, z _0i ).

FIG. 4 is an explanatory view of the dependence of the fluorescence yield of Ni and the fluorescence yield of Fe on the incident angle. In this figure, the horizontal axis is X
The line incident angle is shown, and the vertical axis shows the Ni fluorescence yield and the Si fluorescence yield. The white circles in this figure are the measurement results of a microdrop sample obtained by dropping a solution of a Ni standard stock solution diluted on a hydrofluoric acid-treated Si substrate and drying it. The solid line shows the fit results, but only the asymptotic shape of particles. I was able to reproduce the data. By observing with an optical microscope, it was confirmed that nitrate, which is an additive of the solution, was deposited and formed into a thick plate shape.

The black circles in this figure represent Fe generated during the process.
This is an example of pollution. The solid line indicates the particle size is 40n
As a result of fitting into the data by dividing it into m steps, the data could be reproduced in the asymptotic form of surface contamination and particles. This contamination was found to be stainless steel (SUS) particles because Ni and Cr have the same in-plane distribution in addition to Fe.

The oscillation was not observed in the incident angle dependence of the fluorescence yield because the particles were large, or even if small particles were present, the fluorescence yield was proportional to the volume thereof, resulting in a large size. It seems that the angle dependence is determined by the particles of. In many cases, the measured data can be reproduced in asymptotic form of surface contamination and particles (A. Range and
H. Schwenke, Advances in X
-Ray Analysis, V.I. 35,899 (19
92)). In the case of particle contamination, the fluorescence intensity is almost constant at a low incident angle, the intensity is higher than that of surface contamination, and the lower limit of detection and quantification accuracy of concentration are good.

FIG. 5 is an explanatory view of the incident angle dependence of the bulk ratio obtained by dividing the fluorescence yields from various forms of contaminant elements by the fluorescence yield of the substrate element. In this figure, the horizontal axis represents the X-ray incident angle, and the vertical axis represents the bulk ratio obtained by dividing the fluorescence yield from various forms of contaminant elements by the fluorescence yield of the substrate element. The fluorescence yield from particle contamination (Particulate) is 0.05 compared to surface contamination (Surface).
It becomes several times around °. In a recent report, similar data were obtained for the particle-like residue obtained by dissolving and drying the natural oxide film on the Si substrate with hydrofluoric acid vapor, and the detection sensitivity was improved (Kunihiro Miyazaki, Shimazaki Ayako, IEICE Technical Report, Vol.94, No.39, 7 (1992)).

(D) Internal Contamination It has been pointed out that the distribution shape in the depth direction can be evaluated from the dependency of the fluorescence yield on the X-ray incident angle. Attempts such as depth evaluation have been reported (A. Iida, K. Sakura).
i, A. Yoshinaga and Y. Gohsh
i, Nucl. Instrum. & Methods, A
246, 736 (1986)).

However, the disadvantage of this method is that the penetration depth of X-rays changes rapidly and non-linearly near the critical angle, so the depth distribution of the element is separated from the integral fluorescence yield (4) by inverse transformation. Requires a large amount of highly accurate data. Fluorescence intensity is weak with trace impurities such as pollutants, and it is difficult to obtain such data.

On the other hand, in the contamination evaluation, the strict depth shape is not always important, and it is often sufficient to know the average depth. The inventors of the present invention have investigated the possibility of estimating the average depth of contamination by assuming an exponential function as the depth distribution.

The depth distribution of elements is Φ (z) = φ _D / z _D
It is expressed as exp (z / z _D ) (z <0) and is characterized by the 1 / e depth z _D of the element. The fluorescence intensity is as follows from (3) and (4). I _D (ψ) = (ωε) _D I ₀ T (ψ) Φ _D · Z _X (ψ) / (Z _D + (Z _x (ψ))) ... (9) In the analysis, the fluorescence yield is By using the bulk ratio divided by the fluorescence yield of the substrate, various measurement errors can be reduced.

FIG. 5 above shows the calculated bulk ratio of internal contamination with average depths of 10 nm and 100 nm. The depth range that can be evaluated from the angle dependence of the fluorescence yield is about 1 μm, and the accuracy is 100 nm or less. Further, the calculation result of the fluorescence intensity distribution in the case of the Gaussian distribution or the uniform thickness as the depth distribution shows almost the same shape, so it is difficult to obtain the details of the depth shape of the element from the fluorescence yield.

In an experiment, the inventors of the present invention have shown that arsenic (A
s) Experiments were carried out on cross-contamination of phosphorus (P) during injection. In this experiment, As was injected using an ion implanter in which P was injected at a high concentration. Since the fluorescence energy of P is close to that of Si, the cross-contamination of P was found in the evaluation using a substrate coated with a resist for the purpose of suppressing the fluorescence of Si.

FIG. 6 is an explanatory view of the X-ray incident angle dependence of the bulk ratio obtained by dividing the fluorescence yields from phosphorus and arsenic by the fluorescence yield of sulfur uniformly contained in the resist. In this figure, the horizontal axis represents the X-ray incident angle, and the vertical axis represents the bulk ratio obtained by dividing the fluorescence yields of phosphorus and arsenic by the fluorescence yield of sulfur (S) uniformly contained in the resist.

The solid line is the result of fitting by the equations (9) / (5), and z _P = 14 nm and z _As = 4 as the average depth.
10 nm was obtained, and it was found that phosphorus has a depth distribution. This is P sputtered from the sample holder etc.
Is believed to be the result of being attached to the substrate surface and being struck by arsenic ions. As a result of analyzing the phosphorus of the silicon (Si) substrate by the high-resolution SIMS by the same process, the phosphorus concentration decreased almost exponentially from the surface, and the depth was about the same as the above value.

[0043]

EXAMPLES A quantitative method of total reflection X-ray fluorescence analysis according to an example of the present invention will be described below. In the quantitative method of total reflection X-ray fluorescence analysis of this example, the distribution form of the element to be measured was (a) uniform from the incident angle dependence of the fluorescent X-ray intensity of the element to be measured from the sample shown in FIG. Element, (b) particulate contamination,
(C) Surface contamination or (d) Internal contamination (exponential diffusion contamination) is determined, and an accurate analysis result is obtained by applying a concentration quantification method suitable for the shape of the impurity. Is. The concentration quantification method is the fluorescent X-ray intensity of the element to be measured generated from the sample to be measured, and the fluorescent X-ray intensity of an element other than the element to be measured generated from the sample to be measured whose concentration and shape are known. Is used to quantify the concentration of the element to be measured using the relative sensitivity coefficient of the element to be measured and the element other than the element to be measured, the X-ray penetration length, or the X-ray reflectance.

As described above, it is assumed that the internal contamination exhibits an exponential depth distribution, and the particle contamination uses the asymptotic form of the particles, which can be expressed as follows according to each distribution mode. . (A) Substrate element I _B (ψ) = (ωε) _B I ₀ T (ψ) · Φ ₀ · z _x (ψ) (5) (B) Surface contamination I _S (ψ) = (ωε) _{S I 0 T (ψ) ·} φ S ··· (6) (C) particle shape contaminated _{I P (ψ) = (ωε} ) P I 0 (1 + R (ψ)) Φ P ··· (8) (D ) Internal pollution I _D (ψ) = (ωε) _D I ₀ T (ψ) Φ _D · Z _X (ψ) / (Z _D + (Z _x (ψ))) ... (9) Below, each pollution The quantification method for the shape of will be described.

(A) Substrate element I_B(Ψ) = (ωε)_BI₀T (ψ) ・ Φ₀・ Z_xLooking at (ψ) ... (5), I_B(Ψ) measures the fluorescent X-ray intensity of the substrate
Can be obtained by_BIs determined by the element
Mari, ε_BIs determined by the X-ray detector used (ω
ε)_BAnd T (ψ) is Fresnel's equation
Can be obtained from Φ₀Is the element of the substrate or reference sample
(Si is 5 × 10 in the case of Si) ^{twenty two}cm³) Request
Can be z_x(Ψ) is the penetration length of X-rays
It can be calculated, so I₀= I_B(Ψ) / (ωε)_BT (ψ) ・ Φ₀・ Z_xIt can be obtained as (ψ) ... (9).

[0046] (B) surface contamination Substituting this I ₀ in Equation _{(6), I S (ψ} ) = (ωε) S I B (ψ) · φ S / (ωε) B · Φ 0 · z x ( ψ) ··· (10) Therefore, φ _S = ((ωε) _B / (ωε) _S ) Φ ₀ (I _S (ψ) / I _B (ψ)) · z _x (ψ) (11)

Therefore, (ωε) _B / (ωε) _S = S
_BS (determined by a relative sensitivity coefficient of the substrate element or the reference element and the measured element device) (the concentration of the substrate element or reference _{element) Φ 0 = Φ Bo, I} S (ψ) / I B (ψ) = ( I
An _S / I _B) (the ratio of the X-ray incident intensity and can easily be measured) _{to, φ S = S BS Φ Bo} (I S / I B) · z X (ψ) ··· ( 12) can be expressed as

[0048] (C) If the I ₀ particle-like contamination destination into equation _{(8), I P (ψ} ) = (ωε) P (1 + R (ψ)) φ P I B (ψ) / (ωε) B T (φ) · Φ ₀ · z _x (φ) (13) _{Therefore, φ P = ((ωε)} B / (ωε) P) Φ 0 (I P (ψ) / I B (ψ)) · z x (ψ) · T (ψ) · / (1 + R (ψ)) (14)

Therefore, (ωε) _B / (ωε) _P = S
_BP (determined by a relative sensitivity coefficient of the substrate element or the reference element and the measured element device) (the concentration of the substrate element or reference _{element) Φ 0 = Φ Bo, I} P (ψ) / I B (ψ) = ( I
A _P / I _B) (the ratio of the X-ray incident intensity can be easily measured) and _{when, φ P = S BP Φ Bo} (I P / I B) · z X (ψ) T (ψ) / (1 + R (ψ)) ··· (15)

(D) Internal contamination I₀Substituting into equation (9), I_D(Ψ) = (ωε)_DI_B(Ψ) φ_D/ (Ωε)_BΦ₀(Z_D+ Z_X(Ψ)) ... (16) Therefore, φ_D= I_D(Ψ) (ωε)_BΦ₀(Z_D+ Z_X(Ψ)) / I_B(Ψ) (ωε) _D (17)

Therefore, (ωε) _B / (ωε) _D = S
_BD (determined by the board element or the reference element is a relative sensitivity coefficient of the measuring element device) (the concentration of the substrate element or reference _{element) Φ 0 = Bo, I D} (ψ) / I B (ψ) = I D /
(The ratio of the X-ray incident intensity, easily can be measured) I _B when _{that, φ D = S BD Φ Bo} (I D / I B) · (z X (ψ) + Z D) ··· It can be expressed as (18).

In the above description, the substrate element was used as a reference, but it is of course possible to use another surface contamination sample whose concentration is known, for example. In that case, I ₀ from equation (6) can be used. Can be calculated and substituted into equations (5), (8) and (9).

As is clear from the above description, since the relative detection sensitivity ratio S _IJ of the element I and the element J is a value peculiar to the detector, once the relative detection sensitivity ratio S _IJ is obtained, the standard sample is not used. Can be quantified with.

In the conventional quantitative method of total reflection X-ray fluorescence analysis, the impurity concentration φ is determined as φ = k · I. Here, I is the fluorescent X-ray intensity (cps), and k is the concentration conversion coefficient, which is obtained by comparison with a standard sample whose concentration is known. In this method, the influence of the incident angle setting error or the intensity fluctuation of the X-ray source directly becomes the error of I,
gives the error of φ. Moreover, in this method, the incident angle of the X-ray ψ
When it changes, it is necessary to obtain the coefficient k (ψ) at the incident angle ψ, whereas in the quantitative method of this embodiment, it is possible to quantify only from the relative sensitivity coefficient at any angle.

Table 1 below shows the measurement results of Co concentration by the total reflection X-ray fluorescence analysis quantitative method according to one embodiment of the present invention.

[0056]

[Table 1]

In this measurement, the fluorescence intensity from a standard sample in which the surface of a silicon (Si) wafer was uniformly contaminated with cobalt (Co) element was measured at 9 different points on the Si wafer. is there.

The measurement points are (0, 0), (0, 30), (-30, 0), (30, from the center of the Si wafer.
0), (0, -20), (-20, 20), (20, 2)
0), (-20, -20), (20, -20) (unit is mm), the fluorescence intensity of Si and Co when measured with an X-ray incident angle of 0.005 °, and the conventional The concentration determined by the method and the present invention is shown. Although the incident angle of X-rays is set to 0.05 °, it is considered that the fluorescence intensity of Si varies considerably depending on the position on the wafer due to the error in adjusting the incident angle and the height of the sample. There is a clear correlation with the measured Co fluorescence intensity.

In the conventional method, the concentration conversion is φ _Co (atoms / cm ² ) = k · I _Co (cps), k
= 1.312 × 10 ¹¹ (atoms / cm ² ), and in the present patent, from formula (5), φ _Co (atoms / cm ² ) = S _Co · Z _X · Φ _Si · (I
_Co / I _Si ), S _Co = 0.00067, Z _X = 3.4 × 10 ⁻⁷ c
m, Φ _Si = 5 × 10 ²² (atoms / cm ³ ), φ _Co (atoms / cm ² ) = 1.14 × 10 ¹³ (I _Co
/ I _Si ), and as a result, in the conventional method, the average φ _Co = 142 × 10 ¹⁰ (atoms / cm ² ) error 3σ _ICo = 6.0 × 10 ¹⁰ (atoms / cm ² ). In the present invention, the average φ _Co = 143 × 10 ¹⁰ (atoms / cm ² ) error 3σ _ICo = 18.0 × 10 ¹⁰ (atoms / c)
m ² ), and the error 3σ decreased from 42% to 13%.

It can be considered that this difference of 29% is the effect of reducing the measurement error according to the present invention, and the remaining 13% includes the contribution of the actual concentration distribution.

FIG. 7 shows the total reflection fluorescence X of one embodiment of the present invention.
It is explanatory drawing of Co concentration measurement result by the quantitative method of line analysis.
This figure is a graph of the measurement results shown in Table 1 above, in which the horizontal axis represents the position within the wafer and the vertical axis represents the Co concentration. From this figure, the measurement results by the conventional method vary greatly from the average value, but the measurement results by the present invention are gathered near the average value, and the measurement accuracy by the present invention is more accurate than the measurement results by the conventional method. I understand.

[0062]

As described above, according to the present invention, metal contamination, which affects the reliability of the characteristics of the semiconductor device, and particle contamination, which is greatly related to the yield in the manufacturing process, can be accurately and stably evaluated. By establishing a quantitative method for quantitative analysis of total reflection X-ray fluorescence and a quantitative apparatus for achieving the above, it contributes greatly to the improvement of the yield of highly integrated circuit devices and the like.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of X-ray intensity distribution above and inside a sample due to unit X-ray flux incident at different angles.

FIG. 2 is an explanatory diagram of incident angle dependence of Si fluorescence yield and Ni fluorescence yield from a Si substrate.

FIG. 3 is an explanatory diagram of size dependency and incident angle dependency of fluorescence yield of rectangular particulate contamination.

FIG. 4 is an explanatory view of the dependence of the fluorescence yield of Ni and the fluorescence yield of Fe on the incident angle.

FIG. 5 is an explanatory view of the incident angle dependency of the bulk ratio obtained by dividing the fluorescence yields from various forms of contaminant elements by the fluorescence yields of the substrate elements.

FIG. 6 is an explanatory diagram of the X-ray incident angle dependence of the bulk ratio obtained by dividing the fluorescence yield from phosphorus and arsenic by the fluorescence yield of sulfur uniformly contained in the resist.

FIG. 7 is an explanatory diagram of Co concentration measurement results by a total reflection X-ray fluorescence analysis quantitative method according to an example of the present invention.

Claims (5)

[Claims] 1. Using the fluorescent X-ray intensity of an element to be measured from a sample to be measured and the fluorescent X-ray intensity of an element other than the element to be measured whose concentration and shape are known, A quantitative method for total internal reflection fluorescent X-ray analysis, which comprises quantifying the concentration of an element to be measured using the relative sensitivity coefficient of elements other than the element to be measured, the X-ray penetration length, or the X-ray reflectance. 2. The fluorescent X-ray intensity of an element having a known concentration and shape other than the element to be measured is the fluorescent X-ray intensity of the sample substrate to be measured. Quantitative method of total reflection X-ray fluorescence analysis. 3. The total internal reflection according to claim 1, wherein the fluorescent X-ray intensity of an element having a known concentration and shape other than the element to be measured is the fluorescent X-ray intensity of the standard sample. Quantitative method of fluorescent X-ray analysis. 4. The shape of the element to be measured is classified into any of uniform element, particulate contamination, surface contamination, and internal contamination based on the incident angle dependence of the fluorescent X-ray intensity of the element to be measured from the sample to be measured. A quantitative method of total reflection X-ray fluorescence analysis characterized by applying a concentration quantitative method suitable for the shape of each element to be measured. 5. A means for detecting the intensity of fluorescent X-rays from the sample to be measured, and a means for calculating the penetration length of X-rays into the sample to be measured, the X-ray reflectance or the X-ray intensity distribution, or obtaining it from accumulated data. Quantifying the concentration of the element to be measured from the fluorescent X-ray intensity of the element to be measured, the penetration length of X-rays into the sample to be measured, the X-ray reflectance or the X-ray intensity distribution, and the relative sensitivity coefficient of the element to be measured. An X-ray fluorescence analyzer for total internal reflection, which comprises: