JPH01145508A - Measuring apparatus - Google Patents

Abstract

PURPOSE:To enable highly accurate measurement of a depth of a recess, by obtaining spectral reflection characteristic of a part containing the recess of an object to be measured and a part containing no recess to correct a noise component by both the characteristics. CONSTITUTION:A luminous flux of white light source 2 is made to irradiate an object 1 to be measured held on a stage 100 through a condenser lens 3, a half mirror 5 and an objective lens 4. A part of reflected light from the object 1 being measured passes through a pinhole 7 by way of the lens 4, the half mirror 5 and an imaging lens 6 to detect 8 spectral reflection characteristic of the reflected light, which is initially stored 11 as such of a part with a trench through an A/D converter 9 and a gate 10. Then, the object 1 being measured is set to a part with no trench and a spectral reflection characteristic with no trench is stored 12 in the same procedure. Then, a signal processing for the removal of noises or the like is performed with a bandpass filter 14 using the spectral reflection characteristics. Thus, a depth of the recess can be detected at a high accuracy by further performing a computation with a trench depth calculating section 17.

Description

Detailed Description of the Invention (Industrial Field of Application) The present invention relates to an uneven depth measuring device for measuring the depth of grooves and holes in an object to be measured. The present invention relates to an uneven part depth measuring device suitable for measuring with high precision the depth of an uneven part of several microns such as a provided groove, a so-called trench, etc.

[Conventional technology]

2. Description of the Related Art Various measuring devices have been proposed for measuring with high accuracy the depth of irregularities of several microns in size provided on an object to be measured.

In recent years, in particular, as IC circuits have become highly integrated in semiconductor manufacturing, so-called trench capacitors, in which a capacitor is formed by digging a trench of several microns on a silicon substrate, have become promising. There is a need for a measuring device that can measure the depth of the trench at this time with high precision. A scanning electron microscope (SEM) is used to measure the depth of a trench by cutting the object to be measured and measuring its cross section.

Scanning electron m1crosc
opy ), or by making a beam of light incident on the object to be measured.
There is an apparatus that performs measurement using interference caused by reflected light from the object to be measured.

FIGS. 6A and 6B are schematic diagrams of a conventional measuring device that utilizes interference caused by reflected light from uneven portions.

In the figure, a light beam from a white light source 62 is irradiated onto an object to be measured 61 via a condenser lens 63, a half mirror 65, and an objective lens 64. The reflected light from the object to be measured l is
The light is guided to a pinhole 67 via an objective lens 64, a half mirror 65, and an imaging lens 66. Here, the pinhole 67 and the object to be measured 61 are conjugate, and the pinhole diameter is φ11. When the measurement area is φS and the imaging magnification from the object to be measured 61 to the surface of the pinhole 67 is β, φS=φP/β.

Further, the light beam passing through the pinhole 67 is separated into spectra by a spectroscope 68. Now, when the uneven structure of the object to be measured 61 has a so-called masking agent such as a multilayer film (including a single layer film) on the top surface of the trench 36 having a shape as shown in FIG. When the light beam is incident on the surface to be measured where a trench has been dug, the number of layers of masking agent is N, the depth of the trench is DT,
The refractive index is nT.

The film thickness of the first layer of the masking agent is D Mi , and the refractive index is n
Let Mi be the amplitude reflectance of the surface containing only the masking agent as ρe i
dM (this is generally a formula for multiple reflections), and the amplitude reflectance of the bottom of the trench is ρ, eIIIT, and the area ratio (aperture ratio) occupied by the trench within the measurement area φS is A.
If the efficiency of the light beam within one trench opening is K, interference occurs depending on the trench depth between the reflected light 32 from the mask surface and the reflected light 31 from the trench bottom, and the spectrometer 8 in FIG.
Spectral reflection characteristics (spectral reflectance R(λ
)) is expressed as the following formula.

λ is the wavelength, and θ is the incident angle, R(λ), = (1-A)・ρy” +A”K・ρt′
+2Jwa to E1-A) -K・ρ8・ρ□・COS (φ
8-φ□-δ)・・・・・・・・・・・・・・・・・・・＝
-(1) In the above equation, if ρ9=ρT and φM=φT, the reflectance will be only for the trench without the masking agent. Here φM.

Assuming that φ changes little with respect to wavelength λ, equation (1) is a periodic function whose amplitude is 2 FT (1=−K·gM·DT) from the third term, and its period is determined by the trench depth dT. Moreover, the first term is the film thickness of the masking agent.

It represents the characteristics of a masking agent determined by its refractive index, and for a single layer film,
It becomes a periodic function proportional to the film thickness, and multilayer films have more complex characteristics. As the sum of these characteristics, R(λ) has a shape such as 41 in FIG. 4, for example. Normally, the characteristic indicating the trench depth (the third term in equation (1)) has a higher frequency than the characteristic of the masking agent (the first term in equation (1)), so it is not detected by the spectrometer 8. The characteristics are quantized by an A/D converter 69 and then passed through a digital bypass filter 70.
, and its output is noise, that is, low frequency components. The second term is removed and only the high frequency component of the third term, which indicates the depth of the trench, is extracted.

Thereafter, the amplitude is extended to a certain value by the amplitude expansion unit 71 in order to cope with the case where the opening ratio A and the efficiency K become small and the signal indicating the trench depth is weak.
Maximum value (1/λ1) and minimum value (2/λ2) shown in Figure 42
seek. Here, in a simple case without a masking agent, at the maximum value 2nrdT-CO8θ=mλ1 (m is an integer) at the minimum value 2nTdT-CO5θ=(m!'
6) Since θ is known since it is A2, λ1 . The trench depth dT can be determined from λ2. Even when there is a masking agent, the trench depth dl can be determined by applying a coefficient for correcting the influence of the masking agent to the same calculation formula (2).

Further, as shown in FIG. 6B, after the data is quantized by the A/D converter 69, F
It is also possible to perform trench depth calculation 76 from the data converted and processed by the FT unit 75 to determine the trench depth.

[Problem that the invention is trying to solve] However,
The measurement method shown in FIG. 6 had the following problems.

1) When using a sample with a masking agent, the trench aperture ratio (A
) and efficiency (K) become small, and the signal indicating the trench depth becomes weak. High frequency signals remain due to interference between the light reflected from the upper surface of the masking agent and the light multiplexed between the upper and lower surfaces of the masking agent, and these signals become noise and are confused with the period indicating the trench depth, making it possible to extract them.

2) When the optical path length of the mask agent and the optical path length of the trench depth are close to each other, it is difficult to obtain only a signal indicating the trench depth using a bypass filter, and the above characteristics may be mixed, or each characteristic may be different from each other. As a result, only the characteristics indicating the trench depth cannot be extracted.

3) In the method of finding the signal of the trench depth by converting the obtained output into the frequency domain, that is, decomposing it into each frequency component using Fourier transform, especially when the signal becomes weak, it is difficult to find the period of the characteristic of the masking agent. It cannot be mixed and extracted with an integer multiple of the signal.

Even when the present invention has the above-mentioned problems, it is possible to more accurately extract only the signal indicating the trench depth, and it is possible to measure samples that were previously impossible to measure. The purpose of the present invention is to provide a suitable irregularity depth measuring device for trench depth measurement, which makes it possible to measure the depth of irregularities with higher accuracy even though the measurement was unstable.

[Means and effects for solving the problem] According to the present invention, the spectral reflection characteristics of an object to be measured are measured, and the spectral reflection characteristics of a portion of the object to be measured with an uneven portion and a portion without an uneven portion are determined. is measured, and signal processing is performed using the multi-spectral reflection characteristics to extract only the characteristics indicating the depth of the uneven portion.

〔Example〕

FIG. 1 is a schematic diagram of an optical system and a block diagram of a processing process according to an embodiment of the present invention. l may have a trench dug,
The object to be measured has a masking agent on the trench surface, 2 is a white light source, 3 is a condenser lens, 4 is an objective lens, 5 is a half mirror, 16 is an imaging lens, 7 is a pinhole, 8 is a spectrometer, and 9 is a spectrometer. A/D converter, 10 is a gate, 11. 12
13 is a memory for storing spectral characteristics, 11. 12 is an arithmetic processing unit that calculates the difference between data; 14 is a bandpass filter that transmits only the signal characteristics of the trench depth; 15 is a bandpass filter that transmits only the signal characteristics of the trench depth;
1 is an amplitude expansion section that expands the contrast of the trench depth signal, and 13 to 15 constitute a signal processing section that extracts only the trench depth signal. Furthermore, 1
6 is a detection unit for maximum and minimum peaks of trench depth characteristics;
17 is a trench depth calculation section; 16. 17 constitutes a calculation section. 18 is an output section. 100 is a stage that holds the object to be measured, and 19 is a stage drive unit that drives the stage.

The light beam from the white light source 2 is irradiated onto the object to be measured 1 via a condenser lens 3, a half mirror 5, and an objective lens 4. The reflected light from the object to be measured 1 passes through an objective lens 4, a half mirror 5, and an imaging lens 6, and a part of it passes through a pinhole 7, and is separated by a spectrometer in the wavelength range necessary for measurement. . Here, the object to be measured l and the pinhole 7 are in a conjugate relationship, and if the pinhole diameter is φP and the magnification from the object to be measured l to the surface of the pinhole 7 is β, the measurement area φS is φS=φP/β. Become.

In other words, only the reflected light within φS is measured.

Before measurement, the object to be measured 1 is moved by the stage drive unit 19 and set so that the measurement region φS is located in a portion of the trench. After the spectral characteristics of the reflected light are detected in the process described above, they are input to the A/D converter 9, quantized, and first stored in the memory 11 by the gate 10 as the spectral characteristics of a certain portion of the trench. Next, the stage drive section 19
The object to be measured 1 is set in a part without a trench,
In a similar process, the spectral characteristics are measured and reach the gate 10. Here, the gate 10 is switched and the spectral characteristics of the portion without the trench are stored in the memory 12.

Memory 11. Each spectral reflection characteristic of the memory 12 is expressed as RTM (λ
), RM(λ), the difference calculation unit 13 calculates ΔR
(λ)=RrM(λ)−K(λ)・RM(λ)...
...The calculation in (3) is performed. Here K(λ)
As shown in Fig. 4, is a correction coefficient for adjusting the difference in the absolute values of the spectral reflection characteristics RTM (λ) and RM (λ) of the memories 11 and 12, and is originally a wavelength-dependent function. However, a constant that does not depend on wavelength may be used as long as the value does not change significantly within the wavelength range used. In addition, the correction coefficient K
(λ) may be manually input by the operator or automatically calculated by calculation. As for the latter method using calculation processing, for example, taking the characteristics shown in FIG. 4 as an example (41 is RTM (λ).

42 corresponds to RM (λ)) (characteristics of masking agent), R
Using digital low-pass filters for TM(λ) and RM(λ), only the period seen in RM(λ) 42 is extracted as shown in FIG. The output of RTM(λ)41 is converted to HTMLO(
λ)51. The output of RM(λ)42 is converted to RMLO(λ)52
Then, K(λ) = R considering dispersion due to wavelength
TMLO(λ) / RMLO(λ)・・・・・・・・・
・・・・・・・・・・・・・・・・・・(4) Or, as a coefficient independent of wavelength, K(λ) = Awe (HTMLO(λ) / RM
LO(λ))・・・・・・・・・・・・・・・(5)(
Ave() indicates the average value of the coefficients for each wavelength).

Equation (4) is particularly effective when the trench depth signal becomes weak because there is a difference between the optical path length of the trench depth and the optical path length of the masking agent. Further, equation (5) is effective when the optical path length of the trench depth and the optical path length of the masking agent are close to each other and the trench depth signal is strong. Also, instead of formula (3), ΔR' (λ
)=RtM(λ)/(K(λ)·RM(λ)) may be used.

The measurement order of RTM (λ) and RM (λ) described above may be reversed.

ΔR(
λ) is filtered by a digital bandpass filter 14 that transmits only the characteristics indicating the trench depth, which removes the characteristics of the masking agent that could not be removed by the difference calculation 13, as well as noise. The contrast of the signal is magnified, making it easier to detect extreme values. The peak detector 16 detects all extreme values within the measurement wavelength range. Here, only one of the difference calculation 13 and the digital bandpass filter 14 may be processed.

The characteristic (third term) indicating the trench depth in equation (1) is expressed as R1・
(λ) and (that is, R・r (λ) = 2 F[(1−A) K・Opa・
or-cos (φ'M-φT-δ).・
・・・・・・・・・・・・・・・・・・・・・・・・
・・・・・・・・・－・・・・・・・・・ (6)(6)
The formula is maximum at φM-φd-δ=2mπ (m is an integer).

It takes the minimum value at φM-φd-6=(2m-1)π (m is an integer).

That is, from FIG. 4, λ1. If λ2 is detected, = 2m
π ・・・・・・・・・・・・・・・・・・・・・・・・
・・・・・・・・・・・・・・・・・・・・・・・・
From equations (7), (7), and (8), the trench depth dv is expressed as follows.

Here, φM(λ1), φM(λ2), φT(
λI).

φd (λ2) is the extreme wavelength λ1. If λ2 is detected and the composition of the mask agent (film thickness, refractive index) is known, the trench depth can be determined using Fresnel coefficients that can be calculated, and since nT and θ are known.

In reality, stability is improved by obtaining a large number of trench depths from all adjacent extreme values using equation (9) and averaging them.

Furthermore, in equation (9), especially in the case of a multilayer film, φ9(
8) changes in a complicated manner depending on the wavelength. Therefore, each extreme value interval is theoretically not equally spaced. For such a signal, if a conventional method such as FFT is used to convert the signal into a frequency domain, the detected peak will be rounded and the stability will be significantly lacking, making stable measurement impossible.

In addition, in a sample where the shape of the trench bottom is distorted, the efficiency K
decreases, and the amplitude of equation (6) attenuates. By detecting the magnitude of this amplitude or the enlargement rate at the amplitude enlargement section, it is also possible to apply this method to a method for determining the shape of the bottom surface of a trench.

FIG. 2 shows an optical layout diagram and a block diagram of a processing process in a second embodiment of the present invention.

21 is a white light source, 22 is a condenser lens, 23 is a half mirror, 124 is an objective lens, 25 is an imaging lens, 2
6 is a pinhole, and 27 is a spectrometer. The number is the same as in Figure 1.

In this embodiment, means for measuring the spectral reflection characteristics of a portion with a trench and a portion without a trench are provided separately. With the configuration consisting of 2 to 9, the spectral reflection characteristics of the part without the trench are measured, and with the same configuration consisting of 21 to 28, the spectral reflection characteristics of the part with the trench are measured,
Thereafter, the signals are processed in the same way to determine the trench depth. 2
Channel measurement means allow simultaneous measurements and reduce measurement time. Furthermore, the A/D converters 9 and 28 may be removed and differential calculations may be performed as analog quantities, and an analog band pass filter may be used instead of the digital band pass filter. In addition, if there is no masking agent or the trench depth signal is large, the signal of the spectral reflection characteristics of the uneven part is directly input to the bandpass filter without taking the difference, and the signal is processed to calculate the depth. It may also be possible to switch as shown in FIG. Simply passing through a bandpass filter has the effect of removing low frequency components such as waviness due to film thickness and higher frequency components such as waviness due to multiple reflections of light between the upper and lower surfaces of the film.

〔Effect of the invention〕

As mentioned above, by measuring each spectral reflection characteristic of a part with an uneven part and a part without an uneven part, and by prompting a specific signal processing for each spectral reflection characteristic, it is possible to measure the spectral reflection characteristics of a part with an uneven part and a part without an uneven part. This makes it possible to remove the signal indicating the depth of the uneven part and extract only the signal indicating the depth of the uneven part, making it possible to perform stable measurements even on objects to be measured that were impossible or unstable to measure with conventional technology. became.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of an optical layout for trench measurement and a block diagram of a processing process according to the present invention; FIG. 2 is a block diagram of an optical layout for trench measurement and a block diagram of a processing process according to a second embodiment of the present invention; The figure is a cross-sectional view of the trench sample, Figure 4 is a diagram showing the spectral reflection characteristics of the surface with the trench and the surface with only the masking agent, Figure 5 is a characteristic diagram with a low-pass filter applied to Figure 4, and Figure 6A , B is a schematic diagram of the prior art. In the figure,

Claims (2)

[Claims] (1) A device for determining the depth of a concave portion of an object to be measured, including means for holding the object to be measured, means for illuminating the object to be measured held by the holding means, and a device to be measured illuminated by the illuminating means. one or more measuring means for measuring spectral reflection characteristics by receiving light from an object, and the measuring means measures the spectral reflection characteristics of a portion including the recess and a portion not including the recess. A measuring device characterized in that the depth of the recess is measured by obtaining spectral reflection characteristics and correcting noise components using both characteristics. (2) The means for measuring the part without the recess and the means for measuring the part with the recess are the same means, and a means for moving the measurement area of the spectral reflection characteristics of the object to be measured is provided. A measuring device according to claim 1.