JPS6042905B2 - Method for measuring impurity concentration in semiconductors - Google Patents

Description

[Detailed description of the invention]

The present invention relates to a method for measuring the impurity concentration of a semiconductor, and particularly to a method for measuring the average value of the impurity concentration in a silicon wafer using light in the infrared wavelength region. As integrated circuits become denser and larger-scale integrated, crystal defects in silicon substrates are having an increasingly large effect on device characteristics and yield. Oxygen is the impurity that causes these crystal defects.
Examples include carbon and heavy metals, and oxygen is said to have a particularly large effect. *It■IeXP(-KdO
RiC(1-R a+-4β2sinθ[1-Rexp(-
Kd(x))]2+4Rexp(-Kd(x))s2π
nd(x) where β=n=01112...λ
d(x): Thickness of sample at position x on sample Spectrometry is a simple method in which the infrared absorption spectrum is observed and the impurity concentration can be calculated from the absorption intensity. Conventionally, infrared spectrometry is a method that uses a resistance heating element such as silicon carbide as an infrared light source, and uses a diffraction grating or prism to separate the light emitted from it and convert it into a beam of a specific wavelength using a slit. determination has been made. When trying to measure impurities in a silicon crystal using this method, changing the incident wavelength causes periodic changes in the transmitted light intensity due to interference due to multiple reflections between the front and outer surfaces. To avoid this difficulty, as shown in FIG. 1, a sample 1 used for these absorption measurements is usually mirror-polished so that both the front and back surfaces form a wedge shape with a small angle φ. As shown in Fig. 2, a light beam with an intensity of 10 wavelengths λ that is incident on a sample is composed of the following light beams:
The light incident on a portion having a thickness that satisfies the interference condition strengthens or weakens each other, and is observed as a luminous flux with a transmission intensity It that satisfies equation (1). (Figure 3).
・・・・・・・・・Ta'(θ+β) R: Silicon surface reflectance tanθ: n2fk2-1 (absorption coefficient) ~ 0 for silicon, n (refractive index) ~ 3K: Impurity to be quantified In an ideal case, the absorption coefficient of , that is, the transmitted light beam will have an intensity distribution within the cross section of the light beam, as shown in FIG. In Fig. 3, the horizontal axis X indicates the position on the wafer, the interval a between the broken lines is the diameter of the luminous flux, and the vertical axis 1t is the transmitted intensity. Interference is observed. The light intensity that is actually detected is the average value of the intensity (1) in the cross-sectional area of the luminous flux. Therefore, it is. If the sample is ideally polished into a wedge shape with a small angle φ, then (2) for a ray bundle with a sufficiently large area with respect to the wavelength, the ray bundle is as shown in Figure 4. It is assumed that it is a rectangle with a length of 1 cm and a width of 17 m. Since there is symmetry in the width direction of the rectangle, it can be converted into an integral in the length direction. l is the length of the ray bundle. Also, since the angle φ is a very small angle, and the thickness hardly changes and can be replaced with an average value, since θ is almost 0, Sinθ is also almost 0, and the variables in equation (3) above are due to interference. It becomes only the term, and becomes . Here, i is the thickness of the silicon at the portion through which the center of the beam of light passes, that is, the average thickness of the portion through which the beam of light passes. Therefore, ideally, if we could measure the average thickness of the area through which the light beam passes in the case of oblique polishing at only a small angle, then (4
) can be used to calculate the impurity concentration from K. However, it is extremely difficult to polish with high precision at ideally small angles. Therefore, it is extremely difficult to accurately measure the thickness of a sample through which light passes using an ordinary instrument such as a micrometer. By the way, as shown in Figure 5,
If the mirror surface is distorted at the part where the beam of light is incident, the beam of light S=S1+. 52 holds, then Fd
s=Fdsl+FdS2 (5) holds true. If all of these regions of SlS2 are large compared to the wavelength, it means that even if the surface on which the light beam is incident is not ideally obliquely polished, it is formed from several regions that are large compared to the wavelength. If so, equation (4) holds true. This formula means that if the incident light beam is large compared to the wavelength, it does not depend on the wavelength of the incident light. By using this to observe a spectrum having a known absorption coefficient, such as a phonon spectrum, it is possible to calculate the average thickness of the portion where the light beam is incident. The method of the present invention is a method of infrared absorption measurement described above, in which the wavelength region is changed while canceling interference caused by multiple reflections from both surfaces of a silicon wafer, which is a measurement sample.
A wavelength λ1 different from the absorption wavelength λ by the target impurity
This method calculates the average thickness of the part where the infrared beam light is incident from the absorption by phonons having , and accurately determines the target impurity concentration. This will be explained in detail below using examples. CZ (100) boron doped specific resistance 5~10~
A flat silicon wafer is used. This sample has a rectangular shape of 1C1ri×2G and has been obliquely polished by about 0.3 degrees. Two examples of the results of infrared absorption measurements of this sample are shown in FIGS. 6 and 7. Here, in each figure, the peak where the wavelength difference is 0.9 μm is absorption due to interstitial oxygen, and the peak seen at 16.0 pm is TO(L) + TA(L) of the silicon crystal.
This is the absorption of phonons by the niphonone process. T.O.(L
)+TA(L) phonon is the sum of optical phonon and acoustic phonon absorption. This absorption by phonons changes significantly depending on the thickness, as seen in FIGS. 6 and 7. The correlation between the absorption coefficient of this phonon and the thickness as shown in FIG. 8 is obtained. Therefore, Figures 6 and 7
As shown in the figure, when absorption is measured in the wavelength range of 7 μm to 20 pm and the thickness is calculated from the peak value of the phonon at 16 μm, this is immediately the average value of the thickness of the part through which the light beam passes. Using this value, the oxygen concentration can be calculated from the absorption spectrum of oxygen at 9 μm. From the 4th formula ■-0283 press? So, this is transparent -1
-R2e''' rate. In the normal infrared wavelength region, the reflectance of Sl is constant at R = 0.3, and the denominator of this equation is approximately 1, so
(1-R)2e? Kd~ can be approximated. 610 (-1) of the absorption spectrum of the samples in Figures 6 and 7
Focusing on the absorption of TO+TA(L) phonons in
The transmittance at 61h-1 is 2S9% and (9)%, respectively. Substituting this into the above formula, K phonon (phonon absorption coefficient) d = 0.64 and 0.49, respectively, which becomes the value of absorption coefficient x thickness. From this, using Figure 8, The thickness of each sample was 800 pm. 600pm
It can be calculated as follows. In this way, the thickness of each sample can be determined by focusing on the absorption of 6-1 phonons.
Focusing on the absorption of O, the transmittance is 45.4% and 45.4%, respectively.
40%, so by substituting this into the above equation and using the sample thickness i determined above, KSl2O is 0.
954.3.38, oxygen concentration

[0] is

[0] =
Since Ksip×2.8X1I7(]-3, Fig. 6,
The concentration of the sample in Figure 7 is 2.7X1017a11! -3
, 9.5X1017C1111L-3, respectively. By measuring the absorption coefficient of TO + TA (L) phonons in this way, it is possible to determine the average thickness of the part where absorption due to the impurity to be measured is observed, and the absorption coefficient due to the impurity can be determined with high accuracy. be able to. As an explanation of the present invention above, TO+TA(L)
Although the phonon absorption coefficient is used, it goes without saying that the present invention is not limited thereto.

[Brief explanation of the drawing]

Figure 1 is a sample used for absorption measurement, Figure 2 is a diagram showing the state when light is incident on the sample, Figure 3 is a diagram showing transmitted intensity 1t, Figure 4 is a diagram showing ray flux, Figure 5 is a diagram showing the state when there is distortion in the mirror surface at the part where the beam of light is incident.
6 and 7 are sample data for explaining the method of the present invention, and FIG. 8 is a diagram showing the relationship between absorption coefficient and thickness. In the figure, 1 is the silicon wafer sample, φ is the tilt angle, It is the transmission intensity, and x is the position.

Claims (1)

[Claims] 1. A silicon wafer whose front and back surfaces have been mirror-polished to form a wedge shape is irradiated with infrared rays, and the average impurity concentration in the silicon wafer in the thickness direction is measured from the absorption wavelength and absorption intensity of this infrared ray. In this method, the absorption intensity of a wavelength other than the absorption wavelength of the impurity to be measured is measured at the same time as the absorption measurement by the impurity, and the average thickness of the infrared passing portion of the silicon wafer is calculated from the absorption intensity of the impurity other than the impurity to be measured. A method for measuring the impurity concentration of a semiconductor, characterized in that the absorption coefficient due to the impurity is determined using the obtained value.