JPH0246462A - Formation of pattern for measurement - Google Patents

Abstract

PURPOSE:To linearize the relation between the size of the wedge-shaped part of the pattern for measurement and the line width of resist patterns over a wide range of exposing conditions by forming the above-mentioned part to the width change rate smaller toward the front end part. CONSTITUTION:The wedge-shaped pattern TP is formed by determining the curves J1, J2 axisymmetrical with a central axis J parallel with, for example, an (x) axis and continuing the very small rectangular patterns changing in line width at 2.DELTAP each in the direction (y) along the curves J1, J2. The length in the direction (x) of the very small rectangular patterns Pa, Pb is gradually increased toward the front end side of the wedge. Step parts are approximated nearly as smoothly continuous curves by limitation of a projecting optical system and resist when such pattern TP is projected and exposed to a resist layer by using a stepper. The patterns having the edges along the curves J1, J2 are thus formed.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is used in manufacturing semiconductor devices, etc., and in photo processing, and also when a pattern formed on a mask or reticle is exposed to a sensitive substrate. The present invention also relates to a method for forming a measurement pattern suitable for measuring the line width of a pattern transferred to the substrate.

[Conventional technology]

Conventionally, in the lithography process in which circuit patterns are printed onto semiconductor wafers using masks and reticles, the line width of the circuit pattern is faithfully reproduced on the resist layer (1 to 5 μm thick) applied to the wafer. The question is how to form it.

In particular, with reduction projection exposure equipment (steppers, photorepeater, etc.) that has been widely used in recent years, the degree of integration of circuit patterns created on wafers has increased, and the line width of patterns to be transferred has also increased to submicron levels. It has become a territory. In the case of a stepper, the reticle pattern is projected on the wafer at a reduced size of 115 to 1/10 through a projection optical lens, but in recent years, the minimum line width (corresponding to resolution) of the projection lens required on the wafer has increased. is a value close to the limit of 0.8 to 0.6 μm. In order to maintain this performance sufficiently and stably, it is necessary to keep various conditions during exposure constant. In a stepper, the exposure amount condition and the focus condition are particularly important, and if either of these two conditions deviates from the set value, a problem arises in that a satisfactory line width cannot be obtained on the resist.

Therefore, by attaching a test reticle with a resolution chart etc. to the stepper, we performed trial printing of the chart pattern on the resist layer of the wafer (pair silicon) by varying the exposure conditions, and the line width of the developed resist pattern on the wafer A method is adopted in which the exposure conditions are determined to be the optimum exposure conditions when the desired 8191 width is obtained by actually measuring the width using an optical microscope or another dedicated measuring device (micro line width measuring device, etc.). In this case, line width measurement is a method of measuring the line width of a resist pattern (straight line) to determine how much line width the line pattern on the reticle, which was designed to have a certain line width, has as a resist pattern on the wafer. Reading from the edge interval.

By the way, another case where line width measurement is required is when inspecting the optical characteristics of a projection lens, particularly the image plane inclination and field curvature. In this case, set up a chart pattern for measurement at the center of the test reticle and many points around it, check the resolution (i.e. line width) at many points within the field of view of one exposure of the projection lens, and then What is necessary is to find the focus position that is estimated to have the highest resolution. Such inspections are carried out regularly on semiconductor device mass production lines, and serve as basic data for maintaining and managing the performance of steppers.

In addition, in the lithography process, there are many opportunities to measure the line width of patterns formed on the resist layer for various other purposes. Alternatively, it was measured directly using an electron microscope or the like.

In line width measurement as described above, attempts have been made to indirectly estimate the line width by measuring the entire length of the resist pattern (in a direction different from the line width direction). One example is the Siemens Star, which has a large number of radially extending wedge-shaped patterns arranged in a constant circle. This Siemens star is formed on the reticle so that the tip of each wedge-shaped pattern points toward the center of the circle, and when it is printed onto a wafer, the resolution of the tip of the wedge-shaped pattern on the resist image is determined. In other words, what is the diameter of the inner circle where each tip is connected in an annular shape?

[Problem that the invention seeks to solve]

Focusing on one of the wedge-shaped patterns of the Siemens Star mentioned above, the apex angle of the wedge-shaped pattern is determined to be about 2 degrees to several degrees, and the two sides (edges) that sandwich this apex angle are approximately It is formed by straight lines. That is, the rate of change in width of the wedge-shaped pattern in the longitudinal direction had a substantially linear relationship.

The inventors of the present invention exposed a resist layer to such a wedge-shaped pattern under various exposure conditions, and determined the length dimension of the resist image of the wedge-shaped pattern and the line width resolved at that time. As a result of investigating the relationship, it was confirmed that the relationship between the length dimension of the wedge-shaped resist image and the line width becomes non-linear, especially when the exposure dose conditions are changed.

i.e. a wedge pattern (or similar pattern)
In conventional measurement patterns, the linear relationship between the length of the wedge-shaped part and the line width is lost under certain exposure conditions, so it is difficult to accurately estimate the line width from the length of the wedge-shaped part. The disadvantage is that it becomes difficult to do so.

Therefore, the present invention aims at measuring the dimensions of a resist image obtained by exposing a measurement pattern including such a wedge-shaped portion, such that the measured dimension is a line resolved under a wide range of exposure conditions. The purpose is to provide a method that maintains a linear relationship with width.

[Means for solving problems]

Therefore, in the present invention, when a measurement pattern formed on a mask is transferred to a resist layer to form a resist pattern including a wedge-shaped portion in the resist layer, the wedge-shaped portion has a nonlinear width in the longitudinal direction. The shape of the measurement pattern on the mask was corrected so that the relationship of the rate of change was established.

In particular, by making the width change rate smaller as the wedge-shaped resist pattern approaches the tip, we can increase the range in which the relationship between the length of the wedge-shaped part and the line width can be kept well linear. Can be expanded.

[For production]

FIG. 1 schematically shows how the shape of a wedge-shaped pattern is corrected in the present invention, and FIG. 1(A) shows the shape of a single wedge-shaped pattern TP formed on a reticle. , FIGS. 1(B) and 1(C) show how two linear patterns TP, , TP are crossed at an acute angle and double exposed to obtain a wedge-shaped turn (moiré pattern).

When a wedge-shaped pattern is originally formed on the reticle (mask) as shown in FIG. 1(A), for example, a curve J , Jz (asymptotic slope) is determined, and this curve J
+, a minute rectangular (minute straight line) pattern whose line width changes by 2·ΔP in the X direction along J2 is continued. Therefore, the length of the minute rectangular patterns Pa and Pb in the X direction becomes gradually longer toward the tip of the wedge (pattern Pa). For example, the line width Db of the minute rectangular pattern Pb at the center of the wedge is thicker than the line width Da of the minute rectangular pattern Pa at the tip, but the length of the pattern Pb is shorter than the length of the pattern Pa. The reason why the two sides (edges) of the wedge are step-shaped along the curves JI and JZ is that the pattern TP is originally a microscopic shape, and that the electrons generated when forming the pattern TP on the reticle This is due to the fact that the line drawing device cannot draw edges smoothly along the curves JI and JZ. Here, to give an example, in terms of dimensions on the resist, 2·ΔP is about 0.05 to 0.02 μm, the shape Db of the tip of the minute rectangular pattern is 1.0 μm, and the width Da of the minute rectangular pattern Pa0 at the tip is 0.6 μm. , and the maximum width of the minute rectangular pattern at the end is 1.5 μm, and the total length is about 10 to 20 μm.

When such a wedge-shaped pattern TP is projected and exposed onto a resist layer of a photosensitive substrate using a stepper, fine steps cannot be transferred due to limitations in the projection optical system and the resolution of the resist. The stepped portion of ΔP is approximated as a substantially smooth continuous curve, and a resist pattern having edges along the curves J1 and J2 is formed. The inside of this pattern TP may be either a light transmitting part or a light blocking part, and the resist layer on the photosensitive substrate may be either a positive resist or a negative resist.

In this method, as shown in FIG. 1(D), in a state of underexposure, the pattern tip shape of the wedge-shaped resist pattern ■R becomes blunt, and the intersection point CC+ is found by extrapolating the two edges of the wedge. and actual wedge pattern tip CC
This is to correct the phenomenon that the distance d from 2 to 2 becomes large.

In other words, in the case of insufficient exposure, the part on the reticle that is closer to the tip (Pa) of the rust pattern TPO becomes the resist pattern tip CC2, so the length of the minute rectangular pattern is increased by increasing the distance d from the part closer to the pattern tip. This increases the strength of the situation.

Moreover, FIG. 1(B) shows two straight line patterns T with a line width l.
P + , T P z intersect angle θ. The figure shows the appearance of double exposure with a positive type resist layer and a straight line pattern TP.
If I and TP2 are both light-shielding parts (chromium layers), the diamond-shaped part indicated by the hatched area in FIG. . A diamond-shaped part like this (
left and right apex corners X, ,X
The spacing in the X direction of 2 is determined by the two patterns TP, , TP,
Even if the overlapping position of is shifted in parallel in the x and X directions, the angle θ is Ideally, it will always have constant dimensions as long as is constant. However, due to a slight change in the line width of the two linear patterns TP, TP2, and TP2 printed on the resist layer, the distance between XI and Xi, that is, the length dimension of the moire-like resist pattern, changes greatly.

The line width change amount is Δ! with respect to the ideal line width l! Then, X
Assuming the amount of dimensional change between I and X2 as ΔLa, t'a n
θ. /2 If it is several orders of magnitude or more (for example, 3 degrees), then tanθ. /2 is 1
The line width change amount Δl can be enlarged to a sufficiently large value (for example, approximately 40 times as θ=−3°).

In the case shown in Fig. 1 (B), a wedge-shaped pattern is formed with the apex corners X1 and X2 as the tips, and as shown in Fig. 1 (
Similar to A), it is necessary to correct the shape of the wedge portion.

FIG. 1(C) shows one example of such correction, in which two straight line patterns TP, , TP2 on the reticle are slightly bent at both ends thereof. The two straight line patterns TP1 and TP2 intersect at an angle θ near the center.

Although each edge intersects, the point X that is far left and right from the center
3. On the outside of each of X4, two straight line patterns TP,
, TPz each edge is at an angle θ. An angle θ smaller than
It has been bent and deformed so that it intersects at Therefore, the diamond-shaped ideal resist pattern has a wedge shape with an apex angle of θ1 on the outside from positions X3 and X4, and its tip end X' +
, X' t are each longer by ΔH than in the case of FIG. 1(B). In this case, the wedge-shaped portion has a width change rate of 2 in the longitudinal direction between positions X3 and Xs.
It is changing in stages.

Note that the width may be changed in three or more stages.

〔Example〕

Next, each embodiment of the present invention will be described, but before that, the configuration of a stepper suitable for carrying out the method of the present invention will be briefly described with reference to FIG. 2.

In FIG. 2, exposure light (g-line, i-line, etc.) from a lamp 1 such as a mercury discharge lamp is focused by an elliptical mirror 2, and then a shutter 3 that controls the exposure amount is used to make the illumination light uniform. The reticle R is illuminated via the optical integrator 4 and the main condenser lens CL. Reticle R is x, y, θ
It is held on a reticle stage R3 that moves slightly in the direction, and is positioned with respect to the apparatus by a reticle alignment system 5. The pattern of the reticle R is imaged and projected onto the wafer W by a projection lens PL that is telecentric on one or both sides. The wafer W is placed on a wafer stage ST that includes a zy stage that moves horizontally within the XY coordinate system, two stages that move vertically in the optical axis direction of the projection lens PL, a θ stage that rotates minutely in the horizontal plane, a leveling stage, etc. The stage controller 7 moves stepwise in response to commands from the stage controller 7 during step-and-repeat exposure. A known oblique incident light type focus detection system (hereinafter referred to as AP sensor) consisting of a light projector 14, a light projecting objective lens 15, a light receiving objective lens 16, a slit plate 17, and a light receiving element 18 is configured to detect the height of the surface of the wafer W. A positional change with respect to a reference plane in the horizontal direction is detected with high precision, and the positional change information is output to a focus control system (hereinafter referred to as an AF unit) 9. The AF unit 9 controls the two stages via the stage controller 7 based on the position change information, and focuses the wafer W on the projection lens PL.

Further, this stepper is provided with a TTL type wafer alignment system 11 for photoelectrically detecting an alignment mark (diffraction grating shape) on the wafer via a projection lens PL. The wafer alignment system 11 includes He
-Ne, He-Cd.

A lens system 1 l b includes a laser light source 11 a using Ar ions or the like as a light source, a cylindrical lens, and the like.

Beam splitter 1101 objective lens 10, lens system l
id, a spatial filter lie, and a light receiving element 11f are provided, and a photoelectric signal from the light receiving element 11f is sent to a signal processing system 1.
2 is output. Here, the beam from the laser light source 11a is converted into a beam having a slit-shaped cross section extending in one direction by the action of the lens system 11b, and the beam splitter 11
The light enters an off-axis position of the projection lens PL through the 01 objective lens 10, and is focused onto the wafer W as a slit-shaped spot light by the projection lens PL.

When the alignment mark on the wafer W overlaps with the spot light, scattered light and diffracted light are generated from the mark, and the scattered and diffracted light passes through the projection lens PL again and enters the objective lens 1.
It returns to 0, is reflected by the beam splitter llc, and reaches the spatial filter lie via the pupil relay system lid. The pupil relay system 11d connects the pupil of the projection lens PL and the spatial filter 1.
The spatial filter lie has an aperture that blocks specularly reflected light from the wafer surface and allows only diffracted light and scattered light to pass through.

The scattered and diffracted light that has passed through the spatial filter lie is condensed by a lens system and reaches the light receiving element 11f.

By the way, the stage controller 7 is equipped with a laser interferometer that measures the position of the wafer stage ST with respect to the two axes in the x and In response to the measurement pulse (up-down pulse), the signal processing system 12 digitally samples the waveform of the photoelectric signal from the light receiving element 12. Based on the sampled signal waveform, the signal processing system 12 uses a high-speed arithmetic processor to determine the position when the slit-shaped spot light matches the alignment mark on the wafer, and outputs the position information to the main controller 8. do.

In the above mark detection method, in order to extract the signal waveform, it is necessary to move the wafer W with respect to a slit-shaped spot light stationary within the field of view of the projection lens PL.

Now, the main control system 8 comprehensively controls the step-and-repeat exposure sequence and alignment sequence.
During exposure, the opening/closing timing of the shutter 3 and the length of the opening time are controlled via the shutter controller 6.

The stepper shown in FIG. 2 is equipped with a TTL type wafer alignment system 11 for detecting alignment marks on the wafer. It shall be shared to measure the dimensions of the resist pattern IR.

FIG. 3 shows the pattern arrangement of the test reticle R attached to the stepper as shown in FIG. Alignment marks RM, RM2, RM are formed. These marks RM, , RM, , RM3 are detected by the reticle alignment system 5 of the stepper, and are provided outside the pattern area PA. In the pattern area PA, for example, nine mark areas MA, , MAffi are arranged in a 3×3 matrix.

MA8, MAa, MAs, MAb, MA? ,M
A, , MA9 are formed, and the centers of the respective mark areas are separated from each other by Sx and Sy in the X direction and the X direction, and the mark area MA passes through the center of the pattern area PA, that is, the optical axis of the projection lens PL. It is formed near a point.

In each mark area MAn, a wedge-shaped pattern TP as shown in FIG. 1(A) is formed, and more specifically, it is arranged as shown in FIG. 4. As shown in Figure 4,
Wedge-shaped patterns TPa, TPb similar to FIG. 1(A)
, TPc are formed at a constant pitch in the X direction, and next to them are line-and-space linear patterns LPa, L extending in the X direction with a line width LD.
Pb and LPc are formed at a constant pitch in the X direction.

This straight line pattern LPa, 1. ．． Pb, 1, I-PC is only necessary when first specifying the relationship between the length of the resist image of the wedge pattern and the line width. The line width can be determined only by the length of the shaped pattern.

In addition, line and space-like straight line pattern group (L
A plurality of PaSLPb, LPc) with slightly different line widths may be provided in the mark area MAn.

Also, each pattern TPa, TPb, TPc, LPa, LP
b, LPc is formed in the transparent part as a light shielding amount,
The entire inside of the mark area MAn may be made into a light-shielding layer, and each of the above-mentioned patterns may be formed as a transparent portion within the light-shielding layer.

Furthermore, the wedge-shaped patterns TPa, TPb, and TPc are
The measuring pattern TP may be replaced with a measuring pattern TP having a symmetrical wedge portion as shown in FIG. 5(A).

Center on this pattern TP (center of the thickest line width)
When the pattern is divided into two in the X direction, one of them becomes the wedge-shaped pattern TP shown in FIG. 1(A).

In this case, resist pattern I formed on the resist layer
As shown in FIG. 5(B), R has an extremely flat diamond shape.

Therefore, this test reticle R is attached to a stepper, reticle alignment is performed, and a bare silicon wafer W coated with a positive resist is placed on a wafer stage ST. Based on the shot array data written in the main control system 8 of the stepper, a projected image of the pattern area PA of the test reticle R is sequentially exposed onto the wafer W in a step-and-repeat manner. At this time, for example, as shown in FIG. 6, the exposure conditions are changed slightly for each shot area. Figure 6 shows a total of 56 shot arrays, 7 rows horizontally and 8 columns vertically, and the exposure amount is changed little by little for each of the 7 shots in the horizontal (X) direction, and For each of the eight shots in the direction, the focus position is changed little by little. The exposure amount is changed by setting the opening time of the shutter 3 corresponding to the appropriate light amount preset in the shutter controller 6 as offset 0, and giving an offset of, for example, 10 m5ec before and after that. Therefore, in the case of Fig. 6, when the exposure amount offset is zero, for example, the exposure time (shutter 3
(opening time) is 250m5ec, then 56
Of these shots, the seven shots lined up in the X direction are 220.230.240.250, 260.
Printing is performed with an exposure time of 270.280 m5ec.

On the other hand, the focus position is changed in eight steps in the optical axis direction, for example, by 0.25 μm. In Figure 6, when the focus offset is zero, the focus signal from the stepper's AF sensor indicates focus, and autofocus is executed so that the wafer surface matches this focused point, and the focus offset is reduced by one step. (±1) from the main control system 8 to the AF unit 9.
When set to , the focus is set to be shifted from the original focused point by ±0.25 μm (here, the negative sign indicates the direction in which the wafer surface approaches the projection lens PL).

In addition, the code Lll written in each shot area in FIG.
,L12,...Le? represents the length of the resist pattern IR.

After the wafer W exposed by the above-described operation is subjected to predetermined development, it is again transported to the stepper and placed on the wafer stage ST in a pre-aligned manner. Then, a wedge-shaped resist pattern IR formed, for example, at the center of each shot area is relatively scanned in the longitudinal direction by a slit-shaped spot light from the wafer alignment system 11 of the stepper. This situation is schematically shown in FIG. FIG. 7(A) shows the case where a plurality of diamond-shaped patterns TP as shown in FIG. 5 are arranged at a predetermined pitch in the X direction, and a slit-shaped spot light SP
extends in the X direction and is relatively scanned in the X direction. The resist patterns IRa, IRb, IRc, and lRd are flat diamond shapes elongated in the X direction, and the light receiving element 11f of the wafer alignment system 11 is
The photoelectric signal has a waveform as shown in FIG. 7(B). The signal processing system 12 cuts this signal waveform at an appropriate slice level and measures the average length Lnm of the resist pattern IRa-IRd.

In this example, the change in length Lnm is
There is a linear relationship with the change in the line width LD of the straight line pattern (LPa, LPb, LPc, etc.) printed on the resist layer when the exposure amount condition is changed under the exposure conditions of the first region, especially the optimum focus condition. There is.

The line width of the resist pattern for the straight line patterns LPa, LPb, LPc was measured by 1/10 for each scimitar area using an EB length measuring machine using a separate scanning electron microscope (SEM), etc.
Wafer alignment system 1 of the actual measured line width LD and resist pattern IR with a resolution of about 0 μm.
The proportionality constant and the like of the linear relationship with the dimension measurement value Lnm using 1 are determined in advance.

Therefore, simply by measuring the length Lnm of the resist pattern IR, one can immediately know what the line width of the linear pattern formed on the resist layer will be under the exposure dose conditions (the focus conditions are constant) at that time. Conversely, if the length of the diamond pattern TPO formed on the test reticle R is known, the reduction magnification of the projection lens PL is 115 or 1/
Since it is a constant value of 10, the length Lnm of the resist pattern IR to be formed on the resist layer can also be uniquely estimated. Therefore, when a different length Lnm is measured from the estimated value, it is also possible to know the change in the exposure amount condition from the constant of the linear relationship.

Now, FIG. 8 is a diagram for explaining the above linear relationship, and FIG. 8(A) shows a normal wedge-shaped pattern T P o having an inclined taper along the straight line J3.
Figure (B) shows a wedge-shaped pattern TP with the same overall length and a non-linear line width change rate along the asymptote J, and Figure 8 (C) shows two wedge-shaped patterns TP, , TP. The relationship between the length Lnm of each resist pattern and the line width LD of a straight line pattern reproduced at that time is expressed as follows:
This is a graph obtained experimentally under certain focus conditions. The characteristic Va in FIG. 8(C) corresponds to the case in FIG. 8(A), and the linearity between the pattern length Lnm and the line width LD is lost when the pattern length Lnm is larger than 2. On the other hand, in the case of FIG. 8(B), linearity is well preserved over almost the entire range of pattern length Lnm to .about., as in the characteristic vb.

In this FIG. 8(C), the pattern length Lnm is 1
When , in any pattern TP, , TP,
Only the thickest part of the wedge remains, and the tip is not formed as a resist pattern at all. This occurs when the amount of exposure is too high.

In this state, the line width LD of the linear pattern is naturally
The line width becomes considerably thinner than the original value (optimal line width). Also, when the pattern length Lnm is 3, this corresponds to a case where the exposure amount is slightly less than the appropriate value, and the resist pattern is reproduced up to the parts near the tips of the wedge-shaped patterns TP, , , TP. However, as the exposure amount becomes smaller than the appropriate exposure amount, the characteristic Va appears as a thickening of the line width due to the insufficient exposure amount of the linear pattern. This is 2
In this case, the line widths of the minute rectangular patterns (Pa in Fig. 1(A)) with the smallest line width (for example, a value close to the resolution limit of the projection lens) at the tips of the two patterns TP, TP are made the same. However, this is due to the phenomenon that if the lengths differ, the amount of exposure required to reproduce the resist image of this minute rectangular pattern differs slightly.

As is clear from the above, as long as a normal wedge-shaped (or diamond-shaped) pattern such as pattern TP0 is projected exposed, the pattern length L near the optimum exposure amount
The relationship between nm and line width LD is non-linear, and in order to know the change in line width LD from the change in pattern length Lnm, it is necessary to find in advance a relationship that forms a curve like the characteristic Va. In this embodiment, the line width change rate of the wedge-shaped portion is made non-linear so that a non-linear relationship such as the characteristic Va is corrected to a linear relationship such as the characteristic vb.

Incidentally, the characteristics Va and vb in FIG. 8(C) are those when the wedge-shaped patterns TP, TP0, etc. are light-shielding parts and a positive resist is used, and the patterns TP, TP are transparent parts, When the resist becomes negative, the overall trend of properties is different.

However, in the case of pattern T P o, the pattern length is n
The point at which the change in m and line width LD becomes nonlinear occurs in exactly the same way. For example, if the pattern in Figure 8 (A) or Figure 5 (A) is made into a transparent part and exposed to positive resist, the wedge-shaped (diamond-shaped) exposed part will be removed during development and become a concave part, and the surrounding area will be The unexposed portion remains as a stepped edge of the resist layer. At this time, the length Lnm of the wedge-shaped pattern, which is a recess in the resist layer, is measured.
The relationship between the change in pattern length Lnm and the exposure amount has a tendency opposite to that in the previous case. It was also confirmed that even if the base of the resist layer was changed to aluminum, PSG, etc. and exposed in the same manner, the relationship between pattern length Lnm and line width LD maintained linearity under optimal focus conditions.

Furthermore, the characteristic vb in FIG. 8(C) has a tendency to uniformly shift in parallel in the direction in which the line width increases when the focus condition changes.

As described above, according to this embodiment, each of the nine mark areas MAn shown in FIG. 3 has a length L of the resist pattern IH.
By scanning and measuring nm with the spot light SP, it is possible to inspect the resolution at each of the nine locations within one shot area, and from this, the stepper can automatically measure the field curvature and field tilt. . In this case, it is necessary to examine the relationship between the focus offset amount and the change in the length Lnm of the resist pattern IR at each of the nine locations within the shot area under the optimum exposure amount.

Next, as explained in FIGS. 1(B) and 1(C), the case where a wedge-shaped (moiré-shaped) resist pattern IR is formed by the double exposure method will be described in detail.

Bent straight line pattern TP as shown in Figure 1(C)
When , , TP are actually created using an EB drawing exposure device, the peripheral portions are drawn with minute steps as shown in FIG. In FIG. 9, only the pattern TP is shown, and the line passing through the center of the line width is bent in an inverted S-shape at positions RXs and Xs. Here, the bending angle is determined to change from θo/2 to θ,/2, and the center of the pattern TP is at θ with respect to the X axis. /2 (for example, θ is about 3 degrees). When such a pattern TP (TP) is exposed to light on a resist layer, the minute step difference at the edge portion is below the resolution limit, so a linear latent image is formed that continues smoothly. The same goes for ,
The pattern T P + is formed into a line-symmetrical shape with respect to a line segment passing through the center point Oc and parallel to the X axis. These two patterns TPI and TP2 are arranged in each mark area MAn of the test reticle R as shown in FIG. 10, for example. In FIG. 10, the pattern TP is a pattern group in which a plurality of light shielding parts are arranged in the X direction, and the pattern TP is also a pattern group in which a plurality of light shielding parts are arranged in the X direction. Furthermore, a line-and-space linear pattern LP is also formed in the mark area MAn, similar to that shown in FIG. and two pattern groups TP
+ and TP are separated by XP in the X direction on the reticle.

Now, when using such a test reticle R, the first exposure is performed on each shot area on the wafer W using a step-and-repeat method, and then, during the second exposure, the wafer stage ST is adjusted for each shot area. Double exposure is sequentially performed in a step-and-repeat manner for each shot area under the same exposure conditions as the first exposure so that the stepping position is shifted by m-Xp (m is the reduction ratio of the projection lens) on the projection lens.

Alternatively, the position of the test reticle R during the first exposure is translated in parallel in the X direction by exactly XP during the second exposure, and the stepping position of the wafer stage ST is
Make it the same for each second exposure (.

After the above double exposure, when the wafer is developed, a plurality of resist patterns IR similar to those shown in FIG. 7(A) are formed.

Therefore, the length Lnm of the resist pattern IR can be measured in exactly the same way using the wafer alignment system 11 of the stepper.

Furthermore, in the double exposure method, pattern groups TP and TP may be provided separately on two test reticles (or device reticles), and the reticles may be replaced between the first and second exposures. .

When a wedge-shaped (or moiré-shaped) resist pattern IR is formed by double exposure as in this example, the apex portion of the tip can be resolved down to an extremely thin line width, and the value is determined by the projection lens. The resolution exceeds that of PL. This is because no matter how thin the line width of the pattern formed on the reticle becomes, the resolution limit is determined by the characteristics of the projection lens, whereas in the double exposure method, the wedge defined by two intersecting edges This is because the line width at the tip is determined only by the resolution characteristics of the resist layer (or development conditions). The resolution of the resist layer itself is extremely high, and if the development conditions are set to the best, it will be twice as high as the resolution of the projection lens (for example, 0.8 μm).
It is possible to easily secure more than double the thickness (for example, 0.4 μm).

In this example as well, even if the exposure amount was changed near the optimal focus condition, the length Lnm of the resist pattern IR and the line width LD maintained a linear relationship, and the same effect was obtained even if the base was changed.

Strategy 1: When forming a wedge-shaped (or moiré-shaped) resist pattern using the double exposure method described above, the line and space clarity of the linear pattern group TP, , TP, consisting of a plurality of linear patterns is If the line width ratio between the dark area and the dark area is set to a value other than 1:1, for example, the line width of the bright area is larger than the line width of the dark area, the length Lnm of the resist pattern IR and the line width LD will change near the optimum exposure condition. The rate of change can be kept almost constant. This will be explained with reference to FIG.

FIG. 11(A) shows a case where the line-and-space ratio of pattern groups TP, Only straight line patterns TP+ and TPz (line width 2+) are shown. FIG. 11(B) shows pattern groups TP, , TP,
The pitch between the bright and dark areas is kept the same as in FIG. 11(A), and the line width 12 of the dark area is set to about 1/2 of the line width of the bright area. In this case, the linear patterns TP, , TP, may not be corrected in shape as shown in FIG.
The resist pattern IR formed on the wafer by double exposure of the two pattern groups TP, , TP, has a line width on the reticle that is reduced from ffi+ to 12, so naturally its length Lnm is also geometrically It is calculated to be shorter by 2·ΔX.

Therefore, such a thin pattern group TP, , TP2 and a line-and-space pattern LP with a duty ratio of 1:1 are double exposed as shown in FIG. 6, and the resulting moire-like resist pattern IR is Measure the length Lnm using the stepper alignment system, and measure the length Lnm using the SEM method.
Line and space pattern (L/S) with B length measuring machine
Pattern) The line widths of line portions (convex portions formed by the resist layer) and space portions of the LP resist image and their ratios were determined.

At this time, if the line width ratio of the resist image of the L/S pattern LP measured by the EB length measuring machine is 1:1, the line width reproduction is optimal (optimal line width). The relationship between the length Lnm of the resist pattern IR and the line width LD is the first
The characteristic Vc is shown in FIG. 1(C). Characteristic V in the same figure
a shows a similar relationship in the case of FIG. 11(A),
The characteristic Vc is obtained by shifting the characteristic Va in parallel to the direction in which the line width value LD increases. Note that the nonlinearity explained in the previous embodiment is not corrected here.

Also, in the same way as in the case of Fig. 8 (C) above, Fig. 11 (C)
The line width LD in the double exposure pattern group TP+,
The line width of the TPz resist pattern is not limited to the line width of itself, but the line width of a resist image of a pattern with an arbitrary line width on a reticle is exactly the same. Furthermore, the optimum line width means that when an arbitrary line pattern (the line width is known by design) is exposed on the reticle,
It may be considered that the line width of the line pattern formed as a resist image is closest to the designed value.

Now, here, the line width ratio of the resist image of the L/S pattern LP is 1:1 (optimum line width) near the optimum focus condition.
In the vicinity of the exposure amount that resolves to
As shown in the figure, it can be seen that the linear relationship between the line width LD and the length Lnm is maintained over a wide range of pattern lengths Lnm before and after the pattern length Lnm. With the conventional characteristic Va, within the range before and after the optimum line width is obtained, particularly in the direction in which the line width LD becomes thicker than the optimum value, it immediately exceeds the linear range and enters the nonlinear range. Therefore, if the shape (duty) of the pattern groups TP, , TP2 is corrected so as to obtain the characteristic Vc, the line width L before and after the optimum line width is obtained.
A linear relationship between D and length Lnm can be established over a wide range. This holds true even if the base of the resist layer is changed.

In each of the above examples, the length measurement of the resist pattern IR is performed using the stepper's TTL method wafer alignment system 1.
Although this is carried out using the spot light SP from 1, it is also possible to use an off-axis wafer microscope and a television camera (CCD imaging device) installed close to the projection lens of the stepper. Alternatively, a TTR (through-the-reticle) type alignment system that detects marks in a local area on the wafer through a transparent window of the reticle R may be used.

Further, although the measurement pattern formed on the resist layer is etched through the development process, the latent image of the exposed pattern may be photoelectrically detected without development.

〔Effect of the invention〕

As described above, according to the present invention, the relationship between the length dimension Lnm of the wedge portion (or moiré shape) of the resist image of the measurement pattern and the line width LD of the linear pattern formed on the resist is as follows.
Since it is possible to maintain linearity over a wide range of exposure conditions, the relational expression between the length dimension Lnm and the line width LD can be easily created, which increases the measurement processing speed, and at the same time,
The effect is that errors caused by proximal areas can also be kept small.

Furthermore, since the linear relationship is maintained over a wide range, various accuracies of the exposure equipment and optical characteristics of the projection optical system can be easily determined using the alignment sensor included in the exposure equipment.
Furthermore, inspection can be performed with high precision, and it is also possible to perform automatic setup and self-check of the exposure apparatus unattended.

[Explanation of symbols of main parts]

TP,, TP, , TP2, TPa, TPb, TPc・
...Measurement pattern, R...Reticle, W...Wafer, IR, IRa, IRb, IRcS IRd...
Resist pattern, Lnm...Resist pattern length, θ. ,θ,
......Cross angle.

Claims (2)

[Claims] (1) A measurement pattern is provided on the mask, including a roughly wedge-shaped part that is elongated in a predetermined longitudinal direction and gradually narrows in width by a transmitting part or a shielding part, and the measurement pattern is attached to the resist layer of the photosensitive substrate. When identifying the line width of a mask pattern exposed on the resist layer by exposing under predetermined exposure conditions and measuring the dimensions of the measurement pattern formed on the resist layer, A method for forming a measurement pattern, characterized in that the substantially wedge-shaped portion has a shape having a nonlinear width change rate relationship with respect to the position in the longitudinal direction. (2) A first linear pattern and a second linear pattern formed on a mask by a transparent part or a shielding part,
The resist layer of the photosensitive substrate is double exposed under predetermined exposure conditions so as to intersect at a predetermined acute angle, and the dimensions of the measurement resist pattern including the approximately wedge-shaped portion formed in the resist layer are measured. , when specifying the line width of the mask pattern exposed to the resist layer, the ideal shape of the approximately wedge-shaped portion of the measurement resist pattern is determined by a nonlinear width change rate with respect to the longitudinal position of the wedge. A method for forming a measurement pattern, characterized in that a part of the first or second linear pattern is slightly deformed so as to have the same relationship.