JPH11325870A - Focus position measurement method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a focus position measurement method capable of eliminating the restriction of measuring a pattern after development or the like, eliminating the waste of time of graphing a length measured result or the like, and reducing dimension dispersion or the like between respective wafers or between lots. SOLUTION: The pattern P3 of a latent image capable of obtaining a best exposure condition is secured as a data base beforehand, the patterns P1 ', P2 ',... obtained by changing the focus position of respective devices or wafers and performing exposure are pattern-collated with the pattern P3 of the data base, and the focus position f2 ' corresponding to the latent image pattern P2 ' in the case that both patterns match is set as a best focus.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a focus position for performing pattern exposure in a manufacturing process of a semiconductor device or the like, particularly in a lithography process, by using ultraviolet light having a wavelength of 450 nm or less, X-rays, or a charged beam as an exposure energy source. It relates to a measuring method of the above.

[0002]

2. Description of the Related Art In recent years, the miniaturization of semiconductor elements has been further advanced, and the design rule at the development level is 0.25 to 0.18 μm.
It has become. For this reason, the specifications of the pattern dimensions are becoming stricter than in the past.

[0003] One of the major factors of the pattern size variation is the influence of the drift of the exposure apparatus, and the defocus caused by the variation or change of the substrate state.

Conventionally, the following measures have been taken to eliminate such defocus.

First, a resist film is formed on a flat substrate, such as a bare Si substrate, instead of an actual substrate, in order to determine a reference best focus position (hereinafter, referred to as a reference best focus) for an exposure apparatus. Using the formed sample, the resist film is selectively exposed while deliberately shifting the focal position, and the pattern after development is measured by using a length-measuring SEM or, more recently, provided on a stage of an exposure apparatus. The dimensions and shape of the pattern are measured using a laser interferometer.

[0006] Then, the focus position at which the difference between each measured value and the desired value is minimized is determined as the reference best focus.

The setting of the reference best focus is measured and managed daily as a daily management item.

On the other hand, in a semiconductor device, since each layer is usually formed on a substrate by lithography, the above-described setting of the reference best focus is not sufficient, and the optimum focus position for each layer (hereinafter referred to as the best focus position for each layer). Focus).

[0009] For this purpose, conventionally, first, samples in which layers having different thicknesses are formed on a substrate so as to sequentially deviate from the reference best focus are prepared, and these samples are used to form a resist film. Exposure is performed, and the dimensions and shape of the developed pattern are measured to determine the best focus for each layer.

Next, the amount of deviation between the best focus of each layer and the reference best focus is calculated, and these deviations are set as offset amounts with respect to the reference best focus.

Note that the measurement of the best focus of each layer is performed only at the development stage or at the time of starting up the introduction of the factory in order to suppress a decrease in the throughput of the product. At present, the offset amount is not confirmed or reviewed for each case.

[0012]

However, the above method has the following problems.

(1) First, in setting the reference best focus, since the pattern after development is measured,
The consumption of the developing solution and the downtime of other devices become longer.

In addition, when a conventional length measuring SEM is used,
Time is required for plotting and graphing the length measurement result, and the data processing time is wasted.
In the method using a laser interferometer, unless the measurement pattern to be formed on the semiconductor substrate is a predetermined pattern, accurate pattern dimensions and shapes cannot be measured, and the measurement pattern is limited.

(2) As for the offset amount for determining the best focus of each layer, since the offset amount has not been confirmed or reviewed for each lot or each wafer, a trouble occurs during the manufacturing process (an abnormality in the exposure apparatus or the substrate state). Even if the offset amount changes due to the occurrence of the error, the change cannot be detected, and as a result, the pattern becomes abnormal, and it becomes necessary to reproduce the entire lot, thereby lowering the yield. .

In view of the above problems, the present invention can eliminate the restriction of measuring a pattern after development,
Another object of the present invention is to provide a method capable of wasting time such as graphing a measurement result and reducing a dimensional variation between wafers or lots.

[0017]

Accordingly, the present invention provides a technique for obtaining information from only the surface of a latent image of a pattern with high accuracy and a technique for recognizing a pattern (including a signal and an image) which has been remarkably developed in recent years. The pattern recognition recognizes the high-precision measurement information of the latent image shape, the high-resolution observation information of the pattern after development, and the feature at each focal position obtained from the information. This is achieved by doing so.

In other words, in the focus position measuring method according to the first aspect, a latent image pattern obtained by changing the focal position of each device or wafer and exposing is previously secured as a database, and separately exposed and exposed. The obtained latent image pattern is subjected to pattern matching with the database, and when both patterns match, the focal position of the database is set as the focal position of the pattern obtained by the separate exposure.

In this method, the restriction of measuring the pattern after development can be eliminated, and the downtime can be reduced as compared with the conventional method.

Further, in the focus position measuring method according to the third aspect, a pattern after development obtained by changing the focal position of each device or wafer is previously secured as a database, and separately exposed and developed. The developed pattern obtained by performing pattern matching with the database, and when the two patterns match, the focal position of the database is set as the focal position of the pattern obtained by separately exposing and developing, I have.

According to this method, the pattern is collated after development, but time can be wasted such as graphing the length measurement result, and downtime can be reduced as compared with the conventional method.

This focus position measuring method is performed for each lot,
Alternatively, since it can be applied to each wafer, if the focus position measurement method is used for each wafer at the beginning of the lot, the amount of deviation from the best focus in that layer can be used as an offset amount and fed back during subsequent wafer exposure, and as a result Variations in lot-to-lot dimensions and shapes due to changes in the offset amount can be significantly reduced.

[0023]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, a focus position measuring method according to an embodiment of the present invention will be described. Here, the method of measuring the reference best focus will be described. However, as will be described later, the best focus setting for each layer can be similarly performed for each lot or each wafer by applying the present invention.

(Embodiment 1) First, a sample having a resist film formed on a semiconductor substrate is prepared in order to collect data of a latent image pattern serving as a reference for determining a reference best focus. After selectively exposing the resist film to the sample while sequentially shifting the focal position, a heat treatment (hereinafter referred to as PEB) is performed.

Next, the latent image formed on the resist surface was measured with an atomic force microscope (hereinafter, referred to as AFM), and FIG.
As shown in (a), data of latent image patterns P ₁ , P ₂ ,... for each focal position f ₁ , f ₂ ,. In this AFM, since the surface shape of a latent image can be measured three-dimensionally on the order of nm, information such as the depth and inclination angle of the unevenness can be obtained with high accuracy.

Subsequently, the sample after each exposure is developed to measure the pattern shape and dimensions, and the focus position at which the difference between the measured value and the desired value is minimized is set as the reference best focus. Then, data of the latent image pattern (shape signal waveform) when the reference best focus is obtained is registered as a database. For example, in FIG.
If the reference best focus f _3, since the latent image pattern P ₃ corresponding thereto, the data of the latent image pattern P ₃ gives the optimum exposure conditions, registered in the computer as a database.

Next, in order to set a reference best focus in the exposure apparatus B different from the above, for a sample having a resist film formed on another semiconductor substrate, the resist film is selectively shifted while sequentially shifting the focal position. Exposure to PEB
After the latent image was measured using the AFM, as shown in FIG. 1 (b), the focal position f ₁ ', f _2', the latent image pattern P ₁ for each ... ', P _2', Collect data for ...

Next, the measurement data P ₁ ′, P ₂ ′,... Are compared with a latent image pattern (here, data P _{3 in} FIG. 1A) registered in a computer as a database in advance. .

Then, for example, P ₂ ′ in FIG.
Is the same as the latent image pattern P _{3 in the} database, the focal position f ₂ ′ corresponding to the matched latent image pattern P ₂ ′ is set to the reference best focus f ₂ ′ in the exposure apparatus B used this time. To be determined.

In this way, even when the exposure apparatus A, B,... Has its reference best focus changed over time due to the influence of temperature drift or the like, pattern matching is performed with reference to a pre-registered database. Thus, the reference best focus can be adjusted to an optimum one in a short time.

Although the setting of the reference best focus has been described in the first embodiment, the setting of the best focus of each layer can be similarly performed.

That is, in each layer, a latent image pattern for each layer obtained at each focal position including the best focus is registered in advance as a database.
If the latent image pattern obtained for each layer for the first wafer in each lot is compared with the database for each lot, the amount of deviation from the reference best focus for each layer can be obtained as an offset amount, and the subsequent wafers can be obtained. The data can be fed back during exposure.

For this reason, the offset amount can be confirmed and reviewed for each wafer or for each lot. As a result, variation in pattern between lots due to a change in the offset amount during manufacturing can be greatly reduced. it can.

(Embodiment 2) First, a sample having a resist film formed on a semiconductor substrate is prepared in order to collect data of an exposure pattern serving as a reference for determining a reference best focus. After selectively exposing the resist film to the sample while sequentially shifting the focal position, PEB is performed, and subsequently, development is performed.

[0035] Then, the pattern shape of the sample after development was measured using a length measuring SEM, as shown in FIG. 1 (a), the focal position f _1, f _2, each of the ... latent image pattern P _1, Collect data of P ₂ , ...

Subsequently, the pattern shape and dimensions of the developed sample are measured, and the focus position at which the difference between the measured value and the desired value is minimized is determined as the reference best focus. For example, in FIG. 1 (a), if the reference best focus is at f ₁ in the current exposure apparatus A, the developed pattern because the P ₁ corresponding thereto, the data of the developed pattern P ₁ is the optimum exposure conditions To give it, register it in the computer as a database.

Next, in order to set a reference best focus in the exposure apparatus B different from the above, for a sample having a resist film formed on another semiconductor substrate, the resist film is selectively shifted while sequentially shifting the focal position. After exposure, P
After performing the EB, the sample is further developed.

Then, as shown in FIG. 1 (b), each of the focal positions f ₁ ′,
f ₂ ', ... each of the developed pattern P _1', P ₂ ', to collect ... data.

Next, the measurement data P ₁ ′, P ₂ ′,... Are compared with a development pattern (here, data P _{1 in} FIG. 1A) registered in a computer as a database in advance.

Then, for example, P ₄ ′ in FIG.
Developing the pattern, if consistent with the development pattern P ₁ in the database, a 'focus position f ₄ corresponding to' the matched developed pattern P _4, is determined as a reference best focus f ₄ 'in the exposure apparatus B used this time .

In this way, even when the exposure apparatus A, B,... Has its reference best focus changed over time due to the influence of temperature drift or the like, pattern matching is performed with reference to a pre-registered database. Thus, the reference best focus can be adjusted to an optimum one in a short time.

Although the setting of the reference best focus has been described in the second embodiment, the setting of the best focus of each layer can be performed similarly to the case of the first embodiment.
The same can be done.

For this reason, the offset amount can be confirmed and reviewed for each wafer or for each lot. As a result, variation in pattern between lots due to a change in the offset amount during manufacturing can be greatly reduced. it can.

[0044]

Next, an example corresponding to the first and second embodiments will be described.

(Example 1) A specific example of the focus position measuring method corresponding to the first embodiment will be described below with reference to FIG.

First, the following experiment was conducted to determine the correlation between the latent image pattern information and the focal position obtained after development.

First, in order to determine the correlation between the focal position and the latent image pattern, a KrF excimer laser resist is formed to a thickness of 0.73 μm on a silicon substrate, and a resist of 20 mJ / cm ² is formed using a KrF excimer laser stepper. Exposure was performed at a focus position of exposure energy of -1.0 to +1.0 µm and a step of 0.1 µm.

Thereafter, a heat treatment was performed at 105 ° C. for 90 seconds, and a 0.25 μm L / S pattern latent image pattern formed on the resist surface was measured using an AFM.

FIGS. 2 (a) to 2 (c) each show a focal position of -0.1.
FIG. 5 is a schematic view showing a cross-sectional shape by AFM at 5, -0.1, and +0.3 μm.

Next, this substrate was washed with a developing solution NMD-3 for 60 minutes.
Develop for only seconds. 0.25 μL /
SEM for dimension measurement and shape observation of S resist pattern
I went in.

As a result, the reference best focus at this time was -0.1 μm. The data of the latent image pattern at each focal position was input to a computer, and a program was designed to perform pattern recognition (matching).

Next, another silicon substrate was exposed using another exposure apparatus under the same exposure conditions.

After PEB, the latent image was measured and subjected to pattern matching. As a result, the best focus was +0.2 μm.

The wafer was developed under the same developing conditions, and the dimensions were measured and the shape was observed.
It was 0.2 μm.

As described above, according to the first embodiment, the reference best focus can be obtained by performing pattern matching by measuring only the surface pattern of the latent image with high accuracy using the AFM. As a result, downtime of the developer and the apparatus can be reduced.

Next, an experiment was conducted in which the focus position measuring method of the first embodiment was applied to each wafer, and the result was fed back to the next wafer.

First, the resist used was the same as before, with a film thickness of 0.73 μm and exposure conditions of 20 mJ / cm ² . The exposure was performed with the lens controller turned off intentionally.

The first sheet was exposed with the reference best focus at that time being 0.0 μm. After the PEB, the latent image was measured and the pattern was checked. As a result, the pattern coincided with the latent image pattern of -0.1 μm. 0.0 μm as the reference best focus
The latent image pattern is -0.1 μm
The reference best focus at this time is +0.1 μm, that is, the reference best focus position is shifted by 0.1 μm to the + side. The result was fed back to the second sheet, and exposure was performed at a focal position of +0.1 μm. The focal position obtained from the latent image was equivalent to that of -0.1 μm.

That is, the reference best focus position is further shifted by 0.1 μm to the + side. This was sequentially repeated to expose 25 sheets. As a result, good dimensions and shapes were obtained for all wafers.

As described above, according to the first embodiment, only the surface shape of the latent image is measured with high accuracy by the AFM and pattern matching is performed, whereby feedback to the exposure process can be performed with high accuracy. As a result, variations in dimensions and shapes between wafers can be extremely reduced.

(Example 2) A specific example of the focus position measuring method corresponding to the second embodiment will be described below with reference to FIG.

First, the following experiment was conducted in order to obtain the correlation between the pattern shape and the focal position. First, in order to determine the correlation between the focal position and the pattern shape, a resist for a KrF excimer laser is formed on a silicon substrate by 0.73.
μm and a thickness, KrF excimer laser stepper exposure energy 20 mJ / cm ² using, -1.0 to + 1.0 micron
Exposure was performed at a focal point of 0.1 μm.

Thereafter, heat treatment was carried out at 105 ° C. for 90 seconds, development was carried out with NMD-3 as a developing solution for 60 seconds, and the dimension measurement and shape observation of the formed resist pattern were carried out by length measurement SEM.

FIGS. 3 (a) to 3 (d) show the formed resist pattern and the pattern, respectively, at a focal position of -0.1.
It is a figure which shows the schematic of the cross-sectional shape observed by the length measurement SEM at 5, -0.1, +0.3 micrometer.

As a result, the reference best focus at this time was -0.1 μm. The pattern shape at each focus position is registered in the image recognition system of the length measurement SEM,
The corresponding pattern shape can be identified from the file name.

Next, another silicon substrate was exposed using another exposure apparatus under the same exposure conditions. After development, the pattern at each focal point was subjected to pattern matching. As a result, the reference best focus was +0.2 μm. As a result of dimension measurement of the wafer, the reference best focus was +0.2 μm.

As described above, according to this embodiment, the reference best focus can be obtained by performing highly accurate pattern shape observation after development and pattern matching by the image recognition system of the length measurement SEM. As a result, no time is wasted on plotting and graphing the measurement results.

Next, an experiment was conducted in which the focus position measuring method was applied to the wafer at the head of the lot and the result was fed back to subsequent wafers.

First, the resist used was the same as the previous one.
The exposure was performed with a film thickness of 0.73 μm and exposure energy of 20 mJ / cm ² . The exposure apparatus used was different from the one that collected the image recognition data.

On the first sheet, exposure was performed with the reference best focus of the exposure apparatus that collected the image recognition data set at 0.0 μm.
After development, the result of pattern matching was -0.1 μm
Pattern shape. That is, the reference best focus position of the exposure apparatus that has performed the exposure and the exposure apparatus that has collected the image recognition data are different by 0.1 μm to the + side.

The result was fed back to the second and subsequent sheets, and exposure was performed at a focal position of +0.1 μm. as a result,
Good dimensions and shapes were obtained for all wafers.

As described above, according to the second embodiment, high-precision pattern shape observation after development and pattern matching by the image recognition system of the length-measuring SEM are performed, so that feedback to the exposure step is increased. Accuracy can be achieved, and as a result, variations in dimensions and shapes between wafers can be extremely reduced.

In the first and second embodiments, a resist for a KrF excimer laser was used as the chemically amplified resist. However, g-line, i-line, other KrF excimer laser,
A resist corresponding to an ArF excimer laser, X-ray, or charged particle beam may be used. Exposure light may be appropriate for those resists.

Further, in Examples 1 and 2, NMD-3 was used as the developing solution, but other developing solutions may be used, and dry developing may be performed without being limited to wet developing. Furthermore, any length measuring machine can be used regardless of the type of the length measuring machine as long as it can perform pattern recognition.

[0075]

As described above, the present invention has the following effects.

(1) According to the first aspect of the present invention, only the surface shape of the latent image is measured with high accuracy by the AFM and pattern matching is performed, so that the best focus can be obtained without developing and the focus position can be determined. As a result, the downtime of the developer and the apparatus can be reduced.

Further, the feedback to the exposure step can be performed with high accuracy, and as a result, the size and shape variation between wafers can be extremely reduced.

(2) According to the second aspect of the present invention, the best focus can be obtained or the focus position can be determined by performing high-precision pattern shape observation after development and pattern matching by an image recognition system of a length measuring SEM. Can be estimated.

As a result, the time for plotting and graphing the length measurement result is not wasted.

(3) Further, according to the first and second aspects of the present invention,
The feedback to the exposure process can be performed with high precision, and as a result, variations in dimensions and shapes between wafers can be extremely reduced. The change in the offset amount can be detected, and the yield can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining a focus position measuring method according to the present invention;

FIG. 2 is a diagram showing an outline of a shape directly above an AFM at a focal position of -0.5, -0.1, and +0.3 μm in Example 1 of the present invention.

FIG. 3 is a diagram illustrating a resist pattern formed according to a second embodiment of the present invention.
Diagram showing the outline of the shape observed with a length measuring SEM at +0.3 μm

[Explanation of symbols]

 1 Exposed area 2 Unexposed area

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁶ Identification code FI H01L 21/027 H01L 21/30 502G

Claims (4)

[Claims] 1. A latent image pattern obtained by changing the focal position of each device or a wafer and exposing is previously secured as a database, and the latent image pattern obtained by exposing is separately stored in the database and the pattern. A method of measuring the focal position, wherein the collation is performed, and when the two patterns match, the focal position of the database is set as the focal position of the pattern obtained by the separate exposure. 2. The method according to claim 1, wherein the latent image pattern is an atomic force microscope.
2. The focus position measuring method according to claim 1, wherein the focus position is obtained by using (AFM). 3. A developed pattern obtained by exposing each device or wafer by changing the focal position of the wafer is previously secured as a database, and a developed pattern obtained by exposing and developing is separately stored in the database and the pattern. Collating, and when both patterns match, a focal position of the database is set as a focal position of a pattern obtained by the separate exposure and development. 4. The scanning electron microscope according to claim 1, wherein the developing pattern is a scanning electron microscope.
4. The focus position measuring method according to claim 3, wherein the focus position is obtained by using (SEM).