JPH03111713A - Interference type apparatus for detecting tilt or height, reduction stepper and method therefor - Google Patents

Abstract

PURPOSE:To enable highly-precise detection of the height and the tilt of the surface of an optical multilayer object by providing monochromatic light sources emitting beams of a plurality of wavelengths, beam splitters, an irradiating optical means, a detecting optical system, a reference light optical system, a processing circuit, etc. CONSTITUTION:Beams having different wavelengths and emitted from laser light sources 1 and 1' are split in two by beam splitters 18 and 18' and then adjusted so that the split beams pass on the same optical paths respectively. One 16 (16') of the beams made parallel through a prism 110 is reflected as a measuring light by a mirror 13 and then falls on a wafer 4 at an incident angle of about 88 deg.. The other one 17 (17') of the beams does not fall on the wafer and the two lights overlap each other as a reference light at a point A. The beams passing through the point A are made to reach a CCD sensor 3 by a mirror 23 and lenses 21 and 22 and the interference fringes of the measuring light reflected by the wafer and the reference light are detected by the sensor. Since a large number of elements are arranged separately in the CCD, discrete signals are subjected to A/D conversion 52 sequentially in a processing circuit 5 and a data processing means 54 detects a tilt and a height accurately. Based on informations on the tilt and the height thus obtained, a wafer stage 7 is controlled by fine adjustment.

Description

Detailed Description of the Invention [Field of Industrial Application] The present invention relates to an apparatus for optically detecting the inclination or height of an optical multilayer object such as a wafer, and a reduction projection exposure apparatus and method thereof. [Technology] A conventional inclination detection device for an optical multilayer structure object such as a semiconductor wafer detects inclination and focus, as shown in FIG. are doing. In this known example, focus detection is performed by irradiating focused light on the wafer, focusing the position of the reflected light on a position sensor using an imaging lens, and performing height detection (focus detection) from that position. . As for the inclination, parallel light is irradiated onto the lens, the reflected light is focused onto a position sensor through a condensing lens I, and the inclination is determined from the detected position. In any of these detections, it is difficult to obtain an incident angle of 2856 degrees or more to the sea urchin I. A large amount of light is refracted and incident on the film coated with the resist, making it difficult to truly detect the resist surface. Therefore, depending on the reflectance of the base and the thickness of the resist, the detection position and the position of the true resist surface (d) may be significantly different from each other. .

[Problem to be solved by the invention]

With the above conventional technology, it is difficult to maintain an angle of incidence on the wafer of 85° or more. This is because in the case of focus detection, the angle of convergence of the focused light on the wafer must be set to a certain degree, and the beam diameter at the converging position must not be made small. This is because sufficient detection sensitivity cannot be obtained. Also, in the case of tilt detection, the beam diameter of the parallel light must be made small (otherwise, it will not be possible to irradiate only the exposure area that you want to detect, and in this case too, if the beam diameter is too small, the condenser lens will cause the beam to fall onto the sensor). The narrowed beam spot diameter becomes large, and sufficient detection sensitivity cannot be obtained. Therefore, in order to obtain detection sensitivity, the incident angle has been set to about 80°. However, when the incident angle is about 80°, a large amount of the incident light is The source that is refracted into the resist and reflected by the wafer pattern below the resist contributes as detection light.
The detected focal position (height) and inclination will vary greatly depending on the reflectance of the underlying pattern and the resist thickness. For this reason, it was necessary to perform a trial exposure for each wafer exposure process, to determine the detected deviation from the resist surface as an offset value, and to make corrections. Furthermore, even with the same process wafer, when the resist thickness changes, the offset value fluctuates, creating problems that impede high-precision detection.

The purpose of the present invention is to solve the above-mentioned problems, and to enable accurate detection of the height and inclination of the uppermost surface of resist, etc. coated on a wafer (substrate) at all times, regardless of the wafer (substrate) for each process. An object of the present invention is to provide an interference type tilt or height detection device, an ambiguity, a reduction projection type exposure device, and a method thereof.

[Means to solve the problem]

In order to achieve the above object, in the present invention, as shown in Japanese Patent Application No. 1-100025, which has already been filed by the present inventor, a coherent parallel beam such as a laser is used to measure an optical multilayer object, etc. Irradiates the object at a large angle of incidence, and the reflected light and
Interference fringes are generated between the reference beam separated from the parallel beam, and the focus (height) is detected from the phase of the interference fringes, and the inclination is detected from the pitch. At this time, as described above, when the incident angle is increased, particularly to 85° or more, the reflected component on the resist surface becomes large, and highly accurate detection becomes possible. However, in this case, if the base has a very high reflectance like AJ, the detection error will be relatively large. The present invention further improves the precision of Japanese Patent Application No. 1-100025 and provides a means for accurately detecting the height and slope of the resist surface without being affected by the underlying material or resist thickness. be. For this purpose, in the present invention, monochromatic light having two or more different wavelengths is used as the wavelength to be detected, and these plural wavelengths are selected and used. That is, in the present invention, the error caused by the component reflected from the base and transmitted through the resist surface changes synchronously depending on the thickness of the resist, the reflection coefficient of the base, and the wavelength, as will be explained in detail later. Focusing on this, monochromatic light with a plurality of wavelengths is prepared, and the wavelength used for measurement is selected depending on the above-mentioned conditions of the object to be measured. In this wavelength selection method, if the thickness of the resist is known in advance, that data is used to determine the wavelength to be used based on a theoretical formula, etc., which will be described later. However, even if such resist thickness data is not available, it is possible to detect errors that should be adopted by measuring the amount of light that is incident at a large angle and reflected, that is, the reflectance, over multiple wavelengths. It becomes possible to select a wavelength that can be ignored.

[Function] The present invention will be explained in detail. FIG. 2 is a diagram showing the state of reflection and refraction at a boundary surface of light that is irradiating an optical multilayer object at an incident angle θ. Medium 1 is usually air and has a refractive index n
, is 1,0. In the case of a semiconductor wafer, the medium 2 is a photoresist and usually has a refractive index n.

is about 1.65. The medium 5 is the underlying pattern, which varies depending on the process and may have a multilayer structure.
Let nb be the refractive index seen from the interface with the medium 2. As shown in Figure 2, if we focus on the reflection and refraction at the boundary surface 0 point of linearly polarized light with amplitudes Ape (p polarized light) and Ape (s polarized light) incident on the resist surface, there are four components of light. I know that there is. That is, reflected light R,, Rs l refracted light Dxp
I I)1. and DIp incident from within the resist at the 0 point
It is ID II. It is widely known that p-polarized light has a larger refraction component than S-polarized light, so it is not very suitable for the purpose of detecting the surface of the present invention. So 5(Iiii
If l light is the incident light, the amplitude R1 of the reflected light is given by the angle of incidence θ, the angle of refraction ψ, and the amplitude A of the incident light, according to the conditions regarding the continuity of the electric and magnetic fields at the 0 point on the boundary surface in Figure 2. It can be expressed by the following formula.

nl Here, α is the reflection coefficient ra of the substrate (generally r4 is a complex number), the phase φ that changes during one pass through the photoresist, and the light absorption if there is a light absorbing material in the resist. It is given by the following equation using the coefficient β.

The first term in parentheses of equation (3) is the phase difference due to the optical path length difference between the light reflected at 0 and the light reflected at the base, and the second term is attenuation due to absorption. When the base is AI, rb is the largest, and at a detection wavelength in the visible range, rb is about 878 skin. When the base is AJ and the second term in parentheses is 00, that is, when the resist does not absorb the detection light, the error will be most influenced by the base. When the base is Al, standing waves are generated when exposing the pattern, so a light absorbing material may be added. However, is this light absorbing material g? 1M (456nm>, i-line (
365nm) or KuF excimer laser light (,2as
nm), but not necessarily the laser beam used for tilt and height detection. Therefore, in the worst case where the influence of the reflection from the substrate is the strongest and the error becomes large, even when AI is the substrate and the extinction coefficient β = 0, accurate detection is guaranteed, but in other cases there is no problem ( , detection can be performed with even higher accuracy.

The detailed description of the present invention will be continued assuming the above case.

If β=0 and substituted into equation (3) and equation (2), R1 becomes a complex number and can be expressed by the following equation.

Re = reIFAs
(4) Now, if the reflection of the base is 0, that is, α, = 0, (from the equation), and when compared with equation (4), v = 0. This is the boundary between the refractive indexes n and n. β in equation 0(5), which is the reflection equation of
= 0, so if α is illuminated by the following formula, α 2 rb
eiφ
(5) (1) , (4) , (
From equation 5), R, , F in equation (4) can be found as follows: 0(
However, rb is a real number for the time being) (8) Set the amplitude of the incident light @A to 1 and find E from equation (7).

(6), which changes with the thickness d, the reflectance rb of the AI base = 0.878, and the incident angle θ, (2
) and the resist refractive index n, = 1.63, and by substituting ψ found from θ into (8), θ=88°
86° to 80°, R1 can be illustrated on a compound plane, and Figures 3 to 5 can be determined. The value shown on the circumference of each graph is φ determined by equation (6), which changes with the resist thickness. This graph will be explained. The length of the line segment connecting one point on the curve and the coordinate origin is lR81, and represents the amplitude reflectance including the influence of the base. This line segment and real coordinates (horizontal axis)
The angle formed by this indicates the phase change of the reflected light. This phase shift is not affected by the substrate, and in the case of only surface reflection, φ = 0 as described above, so the phase shift due to the effect of the substrate,
In other words, the height detection error is due to the influence of the base. Let's examine how much this error will be in the interference detection method. The relationship between the interference fringes obtained by the detection optical system shown in FIGS. 1 and 9, which will be explained in detail later, and the wafer height Δ2 and inclination Δθ is given by the following equation.

Al (lal > lbl with a, b constants) Here, X is the coordinate in the fringe pitch direction at the position where the two elements overlap and interfere, θ0.0 is the angle of the reference beam and measurement beam with respect to the perpendicular line Δθ is the target chip on the wafer from the horizontal, Δ2 is the height change in the focus direction. Also, ψ0 is the phase constant 0(
m in equation 9) is 1. when the object to be measured is irradiated once as shown in Figure 1. In the two-time irradiation shown in FIG. 9, the amount of vertical movement Δ2 of the wafer required for the interference fringes expressed by the equation 0(9) to change by one pitch is 2. is the following formula.

m; For cases 1 and 2, λ, == 0.6528μ
FIG. 6 shows the results obtained for θ=800 to 89° with m. When calculating the height Δ21 from the intensity of equation (9) obtained from interferometric measurement, if a phase shift of V given by equation (8) occurs in the measurement light due to the influence of reflection from the base, the error ΔZe of the measurement result will be From equation (10), the following equation is obtained.

Figure 8 shows the values obtained for the Al base for θ=88.5°, 88°, 86°, and 80°. Note that @8 Figure (B) shows the value of V. The angle of incidence is 88
Diagram of R1 on the complex plane when the angle is 86°! As is clear from FIGS. 3 and 4, the maximum angle FmaX when viewing the circle 8 from the origin gives the maximum detection error by entering this value into 1 of equation (11). Therefore, when the circumference reaches the second and fifth quadrants and is 0≦85°, V is 0~
Since it changes up to 360°, the maximum detection error is Δzy
becomes equal to and cannot be measured. Therefore, in the case of an Al base, it is essential to set the incident angle to 85° or more.

The maximum value Δzem of the detection error ΔZe is also shown in Figure 8.
ax (maximum value of the graph) with respect to the reflectance γb of the base,
FIG. 7 shows various angles of incidence determined as parameters. The name of the material indicated by the arrow in the graph of Fig. 7 is the material that forms the base of the semiconductor wafer, and from this figure it can be seen that
Although it is not a problem for materials other than l,
It can be seen that when Aj is the base, the maximum detection error reaches about 0.6 μm at a specific resist thickness (specific φ) even if the incident angle is set to 85° or more. Phase φ and detection error Δ in Figure 8
Although the graph of Ze is drawn only up to 180°,
Regarding 180 to 360 degrees, it is obtained by rotating the curve of the graph of 0 to 180 degrees by 180 degrees around the point (180 degrees, 0 μm). Looking at this graph, φ=0~
The detection error is about 0.1 μm or less between 120° and 240° to 560°. That is, if it can be used within this range, high detection accuracy can be obtained.

This method will be explained below based on the relational expression (6) between φ, the resist thickness d, the wavelength λ of the measurement light, and the incident angle 0.

For example, two wavelengths λ at the same angle of incidence, = 0.652
When a laser beam of 8 μm and λ = 0.6119 μm is incident on the object to be measured and measurement is performed by interferometry, the phase value φ at which the measurement error increases to 0.1 μm or more as shown in Figure 8 (G) is 120. ~240° range. According to equation (6), this region exists at a constant resist thickness period. Figure 10 shows the area where the error at these two wavelengths becomes large using line segments. As is clear from this figure, when the resist thickness is in the range of approximately 1.2 to 2.4 μm, if either of the two laser beams is used, the error is sufficiently small and accurate measurement is possible.

〔Example〕

The present invention will be explained below using examples.

FIG. 1 is a diagram showing a practical example of the present invention, which is applied to a reduction projection type exposure apparatus. 81 is an exposure illumination system, and the exposure illumination light emitted from this illuminates the reticle 9, and the transmitted light is transmitted through the reduction lens 8 in two directions and Δθ(2
A wafer stage 7 having a fine adjustment mechanism with an inclination centered around an orthogonal axis - 41 A is 1 = 1 g - only the axis will be explained
On the upper wafer 4, a reduced image g1 of the reticle original is imaged and exposed. A large number of chips are lined up on the wafer 4, and one to several chips are printed in one exposure. Since the wafer of each chip is not completely flat, just before exposure, the height and inclination of the exposed area were determined by the method below, and corrected using the Unino stage 7, and the wafer surface was set in the state with the highest resolution. After that, focus the light. A photoresist is coated on the wafer interface to a thickness of about 1.5 μm. In order to accurately detect the height and inclination of the photoresist surface on this wafer, the laser light source 1 that performs the interferometric detection shown in the embodiment shown in Figure 1 is a He-Ne laser with a wavelength λ of 0.6528 μm. . Laser light source 1' has a wavelength λt of 0.6119μ
m) (6-Ne laser).

The beams emitted from each laser light source are turned on and off by optical shutters 111 and 111'. Each laser has a beam splitter 18, 18' consisting of a grating, for example.
! Therefore, after being split into two, the wavelength selection mirror 19 transmits the light with wavelength λ, reflects the light with wavelength λ, and adjusts the beams split into two at each wavelength so that they pass on the same optical path. Ru. The prism 110 functions so that the two divided beams of each wavelength become substantially parallel. One of the parallel beams L6 (16') is reflected by the mirror 13 as measurement light and then enters the wafer at an incident angle of 88°. The other beam 17 (17') does not enter the wafer, and both beams overlap at point A as a reference beam.

The beam passing through point A passes through mirror 25. Lens 21.22
This leads to the CCD sensor. The light receiving surface of the sensor is A
The interference fringes between the measurement light and reference light reflected by the wafer are detected. The interference fringe detection signal changes in intensity as shown in equation (9), but since the CCD has many elements arranged separately, each signal is processed by A in the processing circuit 5.
A/D converter 52 sequentially performs A/D conversion to generate a digital signal. Measurements for two wavelengths are performed using shutter 18.1.
This is done by sequentially opening and closing 8'. FIG. 11 is an embodiment showing the flow of data processing executed by the processing circuit 5 of FIG.

The interference fringe data Fi (Xj) (i is 1 for the wavelength λ, and 2 for λ) that has been A/D converted by the A/D converter 52 as shown at 60 is a fast Fourier Conversion (F
FT) circuit 56 performs Fourier transform as shown at 61 in a time of approximately 1m8. The spectrum data It(ωj) subjected to Fourier transformation by the fast Fourier transform circuit 53 has two peaks at ω=ωo (=0) and ω=ωi as shown in FIG. 11 (Q). ω is the DC bias. component, and ωi is a spectrum corresponding to the period of interference fringes. ω=
It has a local maximum at ω1, but since ωi is a discrete sample point, the true peak position exists near the sample point that gives this local maximum. The data processing means 54 calculates the spectrum ωi' giving the true maximum value by using various known methods as shown at 62, ωi and ωi+1.
．． In this way, the true peak value II obtained for the two wavelengths as shown at 62 by the data processing means 54
(ω,′) and It(ω,′) and ω. Spectral value at ■.

Comparing means 5 for each ratio of (ω0) and Is(ω0)
5 as shown in 63. That is, 11 = 1
．． (ω+') / L (ω0)r, = I, (ωt')
/ It (ω0) r, and r! Find the larger one. For example, if rl>γ, then 1o=2, and the night length of 1o is used to detect the inclination and height. Since FFT data has already been obtained at each wavelength, the data processing means 54 uses the data for 1=10 obtained in step 63 to calculate ω10' and ω.
Interpolated value l16(ωio') of FFT data at 10'
(complex number), and as shown in 64, this l16(ω
The phase Ll'y2 is determined from the imaginary/real numbers of io'). Then, as shown at 65, the data processing means 54 determines the pitch of the interference fringes from the true peak position ω10', that is, the height Δ2 of the wafer surface from the inclination Δθ1 and the phase 1 of the wafer surface. By feeding back the determined inclination Δθ and height Δ2 to the wafer stage 7, the partial inclination and height of the wafer 4 are controlled. Here, the reason why accurate measurement can be performed by using the wavelength λio of 1 (=io), which is the larger of rl and rl, will be explained using Figure @8. Figure 8 (
4) and (B) have already been explained.

Figure 8C) is a graph @Figure 5 showing the complex amplitude R1 of the reflected light on the complex plane when the base is Al, and the slope of Figure 4 and the amplitude r of the reflected light R1 were obtained. . Figure 5 shows the change in the complex reflection coefficient of Uninot coated with A14 resist due to resist thickness fluctuation (phase φ fluctuation).0 (Incidence angle θ = 88°, Al amplitude reflectance rb =
0.878). The graph of the prisoner in Figure 8 and the horizontal axis match. As described above, when φ reaches a resist thickness corresponding to 120° to 240°, the detection error ΔZe becomes large. Further, in this region φ, the amplitude r of the reflected light R1 becomes small as shown in FIG. 8(C). When the amplitude γ of R1 becomes smaller, the interference fringes change ′v! 4 degrees becomes smaller, and as a result, the peak value of the spectrum It(ω
1/ ) becomes smaller. In order to remove the influence of light intensity fluctuations of the light source, if Ii (ωt') is normalized by dividing it by the DC bias component It (ω0) as described above, this value (
The above-mentioned r,, rt) have a small value when φ is 120 to 240°. However, λ.

= 0.6528 μm and λ weight = 0.6119 μm, the photoresist thickness where φ is 120 to 240° is @1
This is the region shown by the line segment in Figure 0, but the resist thickness where the line segments at these two wavelengths overlap does not exist between 1.2 μm and 2.4 μm. Therefore, the data processing means 54 in the processing circuit 5 processes the above-mentioned r at two wavelengths as described above.
, and r1 and measure using light of the larger wavelength, the measurement error will not exceed 0.1 μm within this resist thickness range, and the data processing means 54 in the processing circuit 5
It becomes possible to accurately detect the inclination and height. 10th
The figure shows the resist thickness at which a detection error exceeding the allowable value occurs when the detection wavelength is changed (resist thickness corresponding to the solid line segment above)
, and the resist thickness region (lower line segment) where detection accuracy within the tolerance is obtained when wavelength selection is performed using λ and λ.
This shows that.

In this embodiment, the input polarization 51 to the processing circuit 5 in FIG.
0) For l, process as described above.

For samples other than Al with low base reflectance, λ.

Or λ! Precise measurement of inclination and height is possible even if the wavelength is fixed to one of the wavelengths. The wafer stage 7 is controlled to move slightly based on the information on the inclination and height obtained in this way.

No. 91. is an embodiment of the present invention. @1 Part numbers that are the same as those in the drawing represent the same parts. Light source 1 has a wavelength λ of 851 nm
The light source 1' is a semiconductor laser having a wavelength λ of 810 nm. The beam emitted from the two-party source passes through lens 1
1 and 11' become parallel beams, which are transmitted by the wavelength selection mirror 19, λ is transmitted, and λ is reflected and guided to the same optical path.0 cylinder redirection lenses 110 and 120 adjust the parallel beam of the semiconductor laser to a desired beam diameter. ing. A beam splitter 10 separates the beam into measurement light and reference light. The measurement light reflected by the half mirror 12 is reflected by the surface of the wafer 4, returned vertically by the folding mirror 14, and transmitted through the half mirror 12 after being reflected again by the surface of the wafer 4. On the other hand, the reference beam is reflected by the No-7 mirror 12 and directly passes through the reflection mirror 1.
4, it is returned vertically and passes through the half mirror 12. Both parties mirror 210. Lens 21. Mirror 220. lens 2
2 and the wavelength λ passes through the wavelength selection mirror 28, and the double glass 24 in the optical path of the 0 reference light that overlaps and forms interference fringes on the imaging surface of the imager 3'' by the lens 22' It is used to refract the reference light so that the imaging surface and the vicinity of the mirror surface of the reflection mirror 14 are in a post-reciprocal relationship, and both sides overlap on the imaging surface.Similarly, the light with the wavelength λ is The selection mirror 28 creates interference fringes of λ!/z2
The interference fringe data of both wavelengths are simultaneously detected and input to the processing port @5.The processing circuit 5 has input terminals such as key input terminals and magnetic cards. Information such as the underlying material of the wafer to be exposed and resist thickness information are inputted by the means 51'.These pieces of information have been measured in advance with a separate device, etc.The information regarding these wafers is inputted by the inputting means 51'. Then, Figure 7,
The software in the processing circuit 5 determines the wavelength at which the error occurs using the method explained in FIGS. 8 and 10, and the slope and height are detected based on the measurement data of that wavelength. FIG. 12 shows an embodiment of the present invention, and the same part numbers as in FIGS. 1 and 9 represent the same parts. Semiconductor laser 1, 1'
, 1" have wavelengths of 831 nm and 810 nm7, respectively.
50 nm, and each semiconductor laser has Peltier elements 120, 120', 120 for temperature control.
The temperature is kept constant and the oscillation wavelength is stabilized.The oscillation of each semiconductor laser is controlled by a control circuit 5. The beams emitted from each semiconductor laser are guided to the same optical path by collimating lenses 11.11', 11'' and beam splitters 18.18', 19.1?', generating a reference beam 17 and an object beam 16 on the same optical path. The object beam is incident on the wafer at an incident angle of 86°, and C
The CD sensor 3 detects the interference image. Semiconductor laser 1,1
′ is caused to oscillate in time series, and the interference pattern at each wavelength is sequentially extracted, and the spectral ratio γ is determined by the above-mentioned spectral ratio γ.

γ, and use the larger wavelength. This wavelength selection is achieved by inserting a shutter (not shown) behind the prism 110 in FIG.
3, and selection may be made from that level. Since the two wavelengths used for selection are relatively close to each other, a semiconductor laser of 750 nm, which is different from these wavelengths, is used to remove uncertainty in height detection. In particular, when a wafer is loaded from a cassette (not shown) onto a wafer stage 7, the height of the resist surface on the wafer varies by about 25 μm due to variations in the thickness of the wafer. However, at 86° incidence, λ = 810 μm, from equation (1o), Δ
z.

=5.8 μm (m=1). That is, for one wavelength, the height change becomes the same phase value for integral multiples of 5.8 μm, so the true height is not known when the wafer 4 is mounted on the wafer stage 7. Therefore, if we compare the phases at two wavelengths, 750 nm and 8511 m, the phase relationship at both wavelengths will be the same.The wafer height change ΔZllll is calculated by the following formula % formula % () Therefore, in the height range of 55.2 μm , the height can be determined accurately from the relationship between the phases of the two wavelengths. In the embodiment of FIG. 12,
As mentioned above, by using two wavelengths that are close to each other for highly accurate height detection, and by performing rough detection using the other wavelength and one of the wavelengths, it is possible to accurately detect height over a wide range with almost no influence from the background. You can find the slope and slope. λ1 and λ
By selecting as described above, φ is 0 to 120.
°, 240° to 360°, so the 8th
As shown in the figure, even with the Al base putter 7, the height detection accuracy is within 0.1 μm. However, it is also possible to perform even more precise measurements. If the thickness of the resist can be measured in advance, by inputting that value through the input terminal 51'' of the control circuit 5'', the error value can be determined as shown in FIG. 8 (4), and this value can be used as the correction value. By correcting the height control value, extremely accurate detection and control becomes possible.

In the embodiment shown in FIG. 12, λ is used only for rough detection, but
It may be used for precise detection, or may include a fourth wavelength. The present invention can also be realized using a wavelength tunable laser such as a dye laser. Further, collimated light may be generated not only from a laser beam but also from a point light source with a strong power such as a mercury lamp to obtain a narrow band spectrum with insulator fringes.

Although the embodiments described in FIGS. 1, 9, and 12 are all intended for exposure focusing of a reduction exposure device, the present invention is not limited to such applications.
Needless to say, it can be widely used to detect the height and inclination of a surface with a high precision of less than 100 cm, and to control the height and inclination of the detection surface based on the detection results.

〔Effect of the invention〕

As explained above, the present invention is capable of detecting the height and inclination of the surface with high accuracy within 0.1 μm even if the base of the object is made of a material with extremely high reflectivity such as AI. For example, 0.5μ in semiconductor exposure equipment.
When exposing a fine pattern of mL to S, it is possible to control the image plane and the resist surface to perfectly match each other, making it possible to form a pattern with almost no variation in line width. In addition, as a result, the yield of pattern path light is greatly improved,
It has a great economic effect.

[Brief explanation of drawings]

Figures 1, 9, and 12 are diagrams showing embodiments of the present invention, and Figure 2 is a diagram of reflection and refraction of light on a surface for detailed explanation of the present invention. The figures are complex imaging width diagrams of reflected light when AI is the base, Figure 6 is a diagram showing the incident angle and height change corresponding to the interference fringe pitch in the present invention, and Figure 7 is the reflectance of the base and maximum error. Figure @8 is a diagram showing the relationship between detection error 9 phase of reflected light 9 amplitude of reflected light due to phase change due to change in resist thickness, Figure 10 is tolerance when changing detection wavelength FIG. 11, which is a diagram showing the resist thickness exceeding the detection error, is a diagram showing an embodiment of the present invention showing a method of processing detected interference fringe information and wavelength selection. 1.1', 1"... light source for interference

Claims (1)

[Scope of Claims] 1. A monochromatic light source that emits beams of a plurality of different wavelengths, a collimator that converts the beam emitted from the monochromatic light source into substantially parallel light, and a beam splitter that splits the beam into a plurality of beams. , an irradiation optical means for irradiating an optical multilayer object with one of the parallel beams split by the beam splitter, and a detection optical system that guides the object light irradiated by the irradiation optical means and reflected by the optical multilayer object to a detector. and a reference light optical system that guides the other parallel beam split by the beam splitter to the detector and causes it to be superimposed on the detector with the object light obtained by the detection optical system, and is compatible with light of multiple wavelengths, An interference type inclination or height selecting means for selecting one of the information signals of a plurality of interference fringes detected by the detector and detecting the inclination or height of the optical multilayer object. Detection device. 2. The interferometric tilt or height detecting device according to claim 1, wherein the selection in the selection detecting means is made based on the optical characteristics of the optical multilayer object and information on the film structure. 3. The interferometric tilt or height detection device according to claim 1, wherein the selective detection means includes comparison means for comparing object light beams at a plurality of wavelengths. 4. The selection detection means performs Fourier transform on the interference fringe information, and calculates the frequency ω_0 (=0) of the Fourier transform spectrum.
The ratio I(ω')/I(ω
2. The interferometric tilt or height detection device according to claim 1, further comprising comparison means for comparing the values of _0) at each wavelength. 5. Claim 5, wherein the selection detection means is configured to correct the measurement result based on the selected wavelength using the optical characteristics of the optical multilayer object and information on the film structure.
or the interferometric tilt or height detection device according to 4. 6. The interferometric tilt or height detection device according to claim 1, wherein the incident angle of the irradiation optical means is set to 85° or more. 7. The interferometric tilt or height detection device according to claim 1, wherein the monochromatic light source is configured to emit S-polarized light. 8. A light source that emits beams of a plurality of different wavelengths in a reduction projection type exposure apparatus that step-and-repeat the substrate and perform reduction projection exposure of a circuit pattern formed on a mask using a reduction projection lens onto the substrate; a collimator that converts the beam emitted from the light source into substantially parallel light; a beam splitter that splits the beam into a plurality of beams; and a beam splitter that splits one of the parallel beams into a plurality of parallel beams that is split between the reduction projection lens and the substrate. an irradiation optical means for irradiating the substrate through the irradiation optical means; a detection optical system that guides the object light irradiated by the irradiation optical means and reflected on the substrate to a detector through a gap between the reduction projection lens and the substrate; and the beam splitter. A reference beam optical system guides the other parallel beam divided by 1. A reduction projection type exposure apparatus, comprising: selection detection means for selecting one of a plurality of information signals of interference fringes and detecting a partial tilt or height of the substrate. 9. In a reduction projection exposure method in which a circuit pattern formed on a mask is exposed by step-and-repeat on the substrate using a reduction projection lens, beams of multiple different wavelengths emitted from a light source are used. The beam is made almost parallel and divided into a plurality of beams, one of the divided parallel beams is irradiated onto the substrate through the gap between the reduction projection lens and the substrate, and the object light reflected on the substrate is transmitted between the reduction projection lens and the substrate. The other split parallel beam is guided to the detector through a gap between the beam and the object beam, and is superimposed on the object beam on the detector. 1. A reduction projection exposure method, characterized in that one of the striped information signals is selected and a partial tilt or height of the substrate is detected.