JPH03270122A - Wafer alignment - Google Patents

Abstract

PURPOSE:To perform correct positioning in spite of asymmetrical adhesion of film by finding an intensity changing ratio of light K reflected from a second projection part of a film adhered a symmetrically to the central axis of a first projection part and finding length of the film corresponding to a fixed intensity changing ratio level about the edges of a pair of object positions and analyzing a deviation amount due to asymmetrical sticking of the film in order to perform correction of a detection position. CONSTITUTION:Variation of an optical signal is a variation ratio (y) of (variation ratio of the intensity of light (y) for an arbitrary unit length of a film y= I/ d) of light reflected from a second projection part 5 of a film 2 asymmetricaly stuck to the central axis of a first projection part 1a, and determined by momentarily operat ing and processing the intensity of an optical signal and a scanning distance while scanning a beam from a light source on the film 2. Using the thus determined variation rate 3 as an alignment signal 3 and finding length dK of the film corresponding to the level of a fixed variation rate yK for the edges at symmetrical positions of at least a pair of second projection parts as dK1, dK2, dK3,...dKn, and further analyzing from these measurement values a deviation amount R due to asymmetrical adhesion, detecting positions is corrected.

Description

[Detailed description of the invention]

[Summary l] Regarding wafer alignment in a reduction projection exposure system used in semiconductor manufacturing, the purpose is to provide an alignment algorithm that accurately aligns the film regardless of whether the film is one-sided. When a film to be a component of a semiconductor element is integrally deposited on the base substrate on and around the alignment mark consisting of one convex part, the film forms a second convex part on the alignment mark. The film is deposited as shown, and the second convex part is exposed to light.
In a wafer alignment method in which the position of the second convex part is optically detected and alignment for patterning the film is performed with respect to the detected position using a reduction projection exposure apparatus, the central axis of the first convex part is The intensity change rate y of the reflected light applied to the second convex portion of the film asymmetrically attached to the
(Light intensity change rate y=Δ for any unit length Δd of the film
I/Δd) is calculated, and the length dk of the film corresponding to the constant intensity change rate y and level is calculated as d l+l+ cl,=, dma, , for at least one pair of edges at the target position.
+, dmll, and the above dm++ dm*
+ dmll , , , , dmll to analyze the amount of deviation R due to asymmetric adhesion of the membrane, and use this amount of deviation R to correct the detected position.

[Industrial application field]

The present invention relates to wafer alignment in a reduction projection exposure apparatus for semiconductor manufacturing. [Prior Art] In the manufacturing process of semiconductor devices, the process of coating wiring, glabellar insulating film, etc. on the film, applying resist on the film, and patterning the resist to match a specific position on the substrate is performed many times. However, the relative positions of the pattern at the bottom of the film and the resist pattern must be precisely aligned. FIG. 2 shows the surface contour of the film 2 formed on the base substrate portion 1 with the projections 1a as alignment marks. Since the growth direction of the film 2 is not only perpendicular to the underlying substrate but also diagonal, the film 2 is applied non-uniformly. In this way, the non-uniform film formation is a later resist patterning process, and does not depend on the way the six films are applied, and it is required that the positional relationship between the pattern under the film and the resist pattern be accurate. Conventional film coating is the second step in subsequent wafer alignment.
The pattern of the film 2, which is non-uniformly deposited with respect to the pattern (1a) at the bottom of the film as shown in the figure, is used as an alignment mark as it is, and alignment is performed by beam scanning or image processing. In other words, in conventional wafer alignment, light is shined onto the pattern as shown in Figure 2, and an alignment signal (waveform) obtained from the light reflected from the pattern edge.
I wanted the center more than A and B. However, depending on how the film 2 is attached, the center line of the film may be biased to one side with respect to the pattern at the bottom of the film (referred to as "one-handedness" in this application). The positional relationship between the pattern under the film and the resist pattern becomes inaccurate. Therefore, it is an object of the present invention to provide an alignment algorithm that accurately aligns regardless of the unilaterality of the membrane.

[Means for Solving the Problems 1] The present invention analyzes the way the film is attached (the amount of one-sidedness) from the shape of the pattern formed on the underlying pattern, and then performs each misalignment parameter in wafer alignment. The gist is to determine the amount of correction (especially scaling). Specifically, the present invention is directed to integrating a film that becomes a component of a semiconductor element onto a base substrate on and around an alignment mark consisting of a first convex portion formed at a specific position on a base substrate portion of a wafer. At the time of deposition, the film is deposited so that the film forms a second convex portion on the alignment mark, and the second convex portion is illuminated and the position of the second convex portion is determined optically. The present invention is applied to a wafer alignment method in which a reduction projection exposure apparatus performs alignment for patterning the film with respect to the detected position. In such a method, the present invention provides an intensity change rate y of reflected light of light applied to the second convex portion of the film that is attached asymmetrically with respect to the central axis of the first convex portion (an arbitrary value of the film). Find the rate of change in light intensity y=ΔI/Δd) with respect to the unit length Δd, and calculate the length dk of the film corresponding to the constant rate of change in intensity yk for at least one pair of edges at the target position d m+,
d kl+d kl, , , , , dma, and the above d m lyd kll dll! ＋ −
, , -dkn, the amount of deviation R due to asymmetric adhesion of the film is analyzed, and the detected position is corrected based on this amount of deviation R. The present invention will be explained in detail below. The base substrate part means not only a silicon substrate but also a multilayer substrate on which an arbitrary n-th layer such as a surface oxide film of a silicon substrate, an insulating film between the eyebrows, and wiring is formed. The first convex portion is information used to align the resist pattern by indicating a specific position on the underlying substrate, and is usually formed in large numbers on one wafer for each exposure shot. The second protrusion is formed at the same location as the first protrusion, and has a size that is optically detectable so as to be used for actual alignment. When light is applied to the second convex portion, an optical signal with an intensity inversely proportional to the distance from the light source is obtained. Hereinafter, an algorithm for optical signal processing, which is a feature of the present invention, will be explained with reference to FIG. 1.2. [Function] In FIG. 2, l is the base substrate, 1a is the protrusion or first convex part, 2 is the film, 3 is the alignment signal, 4 is the first convex part, 5
is the second convex portion. The intensity of the optical signal is inversely proportional to the distance from the light source to the film 2, and the amount of change in the optical signal becomes zero at the flat part of the film 2 (see 3f), and changes greatly where the outline of the film changes significantly. . Here, the amount of change in the optical signal is the intensity change rate y of the reflected light of the light applied to the second convex portion 5.
(Light intensity change rate y for any unit length Δd of the film =
ΔI/Δd), which is obtained by scanning the film 2 with a beam from a light source and calculating the intensity of the optical signal and the scanning distance from time to time. This light intensity change rate is used as the alignment signal 3. By the way, in FIG. 2, since the membrane 2 is shifted to the right, the left edge 5a of the second convex portion 5 is narrower and steeper than the right edge 5b, and the -universal side edge 5 is narrower and steeper than the right edge 5b.
b is wide and gentle. As a result of such one-handedness, the light intensity change rate y=ΔI/Δd is narrow on 5a,
The width is wider on 5b. Using such a rate of change in light intensity as the alignment signal 3, the length dk of the film corresponding to the constant rate of change in intensity ym (Fig. 1) is determined by the edge 4a at the target position of the second convex portion.
, 4b, b'=dk, .) <
c' = d m tab+ next edge 4b
It can be seen that the film 2 is displaced in the direction of. Therefore, in the present invention, the one-handed amount with the membrane is analyzed based on the alignment signal shown in FIG. Unilateralism can be corrected using signal processing techniques (signal processing techniques). The method of correction will be explained below with reference to FIG. The distance between the center of the pattern edge and the center of the other pattern edge is determined, and the width ratio of each pattern edge is determined from each alignment signal. a is the finished width of the pattern at the bottom of the film. S=a+b/2+c/2 ---■ c/b =c'/b' ---■ Therefore, the correction amount R due to pattern one-handedness is as follows: R= (a+b/2+c/ 2)/2 (a/2+b/
2) 1c/4-b/4, and the value obtained by subtracting the correction amount R from the center in the X direction determined from the alignment signals A and B is the position of the pattern. Similarly, the measured value d□ at each position within the wafer. The above calculations are also performed for dk□, dk, . [Example] A film of aluminum having a thickness of 1 μm was formed by sputtering on a protrusion (first protrusion) having a height of LLm and a width of 4 μm when viewed in cross section. In the reduced pitching reactor reactor, the XY points at four points in the aluminum sputtered wafer are each measured using a normal alignment algorithm, and the obtained alignment signals are substituted into the alignment algorithm of the present invention, and the signals are deposited on the alignment marks. The film asymmetry was calculated and the wafer scaling correction value was determined using the above formulas (1) and (2). Note that S=5.2 μm and a=4 μm. b':c'=1:1.4. Exposure is performed by adding the correction value R to the stage control parameters of the reduction projection exposure apparatus. The effect is that when using only a normal alignment algorithm, a positional deviation of about 0.15 μm to 0.2 μm occurs between the pattern at the bottom of the film and the resist pattern, whereas using the above method, when R=O, When Ium correction is performed,
The amount of deviation was suppressed to 0°05 μm or less. [Effects of the Invention] As explained above, according to the present invention, it is possible to accurately match the positional relationship between the pattern at the bottom of the film and the resist pattern at the top of the film, regardless of the way the film is attached. This greatly contributes to improving manufacturing reliability.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram of the method of the present invention, and FIG. 2 is an explanatory diagram of a membrane with one-sidedness, alignment, and signals. l - base substrate, la - first protrusion, 2 - film, 3 - alignment signal, 4 - first protrusion, 5 - second protrusion

Claims (1)

[Claims] 1. A first
When a film that will become a component of a semiconductor element is integrally deposited on the base substrate on and around the alignment mark consisting of a convex part, the film forms a second convex part on the alignment mark. A film is deposited on the film, the second protrusion is irradiated with light, the position of the second protrusion is optically detected, and alignment for patterning the film is performed with respect to the detected position by reduction projection exposure. In a wafer alignment method performed using an apparatus, the intensity change rate y of the reflected light of light applied to the second convex part of the film that is attached asymmetrically with respect to the central axis of the first convex part is
(Light intensity change rate y=Δ for any unit length Δd of the film
d_k_1, d_k_2, d_k_3, .
．． ．． ．． d_k_m, and the above d_k_1, d_k
_2, d_k_3, . ．． ．． ．． A wafer alignment method characterized in that the amount of deviation R due to asymmetric adhesion of the film is analyzed from d_k_m, and the detected position is corrected based on this amount of deviation R.