JPS638510A - Pattern detecting apparatus - Google Patents

Abstract

PURPOSE:To identify automatically a pattern whether it is formed by an inductive material or not, by irradiating a first energy beam onto a specimen receiving an energy beam issued from the irradiated part and detecting the pattern of this part. CONSTITUTION:A laser beam from its source 10 is developed as a spot light on a wafer W set on a stage 34 through various lenses. The laser spot scans 2-dimensionally over the wafer W by scanning unit 16 and having glass 20. Scattering beams of light are observed from edges of the pattern on the wafer W. The scattering beams are collected by a photo-electric detecting apparatus 42 installed annually on the periphery of an objective lens 32. A reflected beam of light from the wafer W is detected by a photo-electric detecting apparatus 52 via an iris diaphragm 47 and a beam of fluorescent light from the pattern on the wafer W is detected by a photo-electric detecting apparatus. By these three kinds of information and the scanning informations, measurements of the pattern on the wafer W are made in a multi-angular manner.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a pattern detection device, and in particular detects LWS such as the edge position and width of a photoresist pattern used in the manufacturing process of semiconductor devices. 'll
The present invention relates to a pattern position detection device using an energy beam suitable for.

[Conventional technology]

Semiconductor wafers used in the manufacture of semiconductor devices such as ICs and LSIs undergo various treatments during the manufacturing process, and in order to ensure the desired characteristics, line widths and dimensions of circuit patterns, etc. during the process are particularly important. It is important to accurately manage such matters. For this purpose, a pattern detection device that detects patterns with high precision is required.

Conventionally, there is a device of this kind that scans a wafer with a laser beam as probe light and detects the scattered light generated when the laser beam scans the edge of a pattern to detect the position of the edge. .

[Problem that the invention seeks to solve]

However, when such a device detects a pattern made of photoresist, an accident may occur in which the photoresist portion is exposed to the detection laser beam and is damaged (changed in quality or destroyed so that it no longer functions as expected). . In addition, the measures adopted to improve the detection accuracy of the device are
This increased the possibility of this damage occurring.

In other words, in order to improve detection accuracy, there are measures to shorten the wavelength of the laser beam by using a HeCd laser, Ar laser, etc., and measures to make the beam diameter (probe diameter) minute by focusing the beam with an optical system. However, the former method allows the beam wavelength to fall within the 8-light wavelength range of the photoresist, and the latter method increases the energy density of the beam irradiated to the photoresist, which reduces the degree of deterioration of the photoresist due to exposure to light. This has led to an increase in the number of accidents resulting in damage.

An object of the present invention is to provide a detection device capable of highly accurate pattern detection without damaging a sensitive material such as a photoresist forming a pattern.

[Means for solving problems]

In order to achieve this object, the present invention has the following configuration.

That is, the first energy beam irradiates the object to be inspected.
A pattern detection device comprising: an irradiation means; a detection means for receiving a second energy beam emitted from an irradiated portion of the object to be irradiated with the first energy beam and detecting a pattern of the irradiated portion; a second irradiation means for irradiating the object with a third energy beam; and a second irradiation means for irradiating the object with a third energy beam; determining means for determining whether or not a portion to be irradiated with the beam is a portion sensitive to the primary energy beam; and upon receiving an output of the determining means, determining whether or not the portion to be irradiated with the beam is a portion sensitive to the first energy beam; The device is characterized by comprising a control means for controlling the first energy beam to prevent the sensitive portion from being damaged by the first energy beam.

[Effect]

With this configuration, sensitive materials (such as photoresists)
It automatically discriminates between patterns formed by sensitive materials and patterns not formed by sensitive materials, and prevents irradiation of energy beams (laser beams, etc.) to patterns formed by sensitive materials in a manner that would damage them. It is something.

[Embodiment] Hereinafter, an embodiment of the pattern position detection device according to the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows the arrangement of optical devices in one embodiment of the present invention. In this figure, laser beam source 1
0 generates a short wavelength laser beam (coherent light). The Lede beam passes through the shaft 11 only when it is open. The diameter of the laser beam is expanded by a lens 12.14 constituting a beam expander.

A mirror integral moving unit (scanning unit) 16 is provided between the lenses 12 and 14 to scan the laser beam one-dimensionally. The scanning unit 16 shifts the optical axis of the laser beam in parallel without changing the optical path length of the laser beam. Further, a scanning amount monitor 18 composed of a laser interferometer, a linear encoder, etc. reads the amount of movement of the scanning section 16.

Next, a burping glass 20 is inserted between the lenses 12 and 14. This burping glass 20 shifts the scanning position of the laser beam, and the angle with respect to the incident laser beam is changed by a drive unit 22 such as a pulse motor or a DC motor.

Note that the configuration is such that the degree of this change, that is, the amount of shift of the laser beam, can be detected by the potentiometer 24.

The laser beam that has passed through the lens 14 is passed through a half mirror (
After passing through the half prism) 26, the image rotates through the prism 2.
8, is reflected by a dichroic mirror 30, and enters an objective lens 32.

The dichroic mirror 30 has spectral characteristics such that it reflects the laser beam and transmits light with a longer wavelength. The laser beam incident on the objective lens 32 is condensed and imaged as a (fine) spot light onto the wafer (test sample) W placed on the stage 34 'R'. This laser spot light scans the wafer W one-dimensionally as the scanning unit 16 moves, and the scanning position is set on the burping glass 2.
Shift by driving 0. That is, the laser spot light scans the wafer W two-dimensionally by the scanning unit 16 and the burping glass 20.

On the wafer W, a minute unevenness level difference, a so-called pattern edge, is formed, and when the spot light crosses the edge, scattered light (or diffracted light due to the level difference) is emitted from the process/nozzle.
occurs. Most of the scattered light is generated at an angle equal to or larger than the solid angle determined by the numerical aperture (N A ) of the objective lens 32, so that almost no light returns to the objective lens 32. The scattered light is then focused on a photoelectric detector 42 provided in an annular manner around the objective lens 32.

When the laser beam is irradiated onto the wafer W, the reflected light from the wafer W passes through the objective lens 32, the dichroic mirror 30, the image rotation prism 28, and the half mirror 26.
After being reflected at a right angle, the light is reflected at a right angle by a half mirror 44 and enters a lens 46. The reflected light that has passed through the lens 46 forms an image at the center of the aperture of the diaphragm 47 . The aperture 47 is
It functions to block stray light other than reflected light. The reflected light that has passed through the aperture 47 is detected by a detector 52 such as a silicon photodiode (SPD) via an aperture 50.

This photoelectric detector 52 outputs a photoelectric signal according to the amount of reflected light. The aperture 50 is placed at a position conjugate with the pupil of the objective lens 32, so that only the zero-order photoacid component of the reflected light is detected by the reflected light detector 52. Further, the pupil position of the lens 32 is the center of deflection of the laser beam by the scanning unit 16.

When a pattern of a photoresist layer is formed on the wafer W, fluorescence (or phosphorescence) is generated from the pattern when excited by a short wavelength laser beam. The fluorescent light is usually visible light with a wavelength of 500 to 700 nm having a spectrum peak around 600 nm, which is longer than the wavelength of the laser beam.

Therefore, the fluorescent light from the pattern passes through the objective lens 32, passes through the dichroic ink mirror 30, passes through the half mirror (half prism) 56, lens 54, dichroic mirror 58, and mirror 60, and then passes through the wavelength range of the laser beam. The light passes through a filter 62 that cuts the light and reaches a photoelectric detector 64 such as a photomultiplier.

The dichroic mirror 58 separates fluorescent light from the wafer W and observation illumination light.

Therefore, the visible illumination light from the illumination system 66 is reflected by the dichroic mirror 58, passed through the lens 54, reflected by the half mirror 56, transmitted through the dichroic mirror 30, and then incident on the objective lens 32 for observation on the wafer W. Epi-illuminate the area. In this state, light from illumination system 66 is blocked by dichroic mirror 58 from directly entering photoelectric detector 64 .

In addition, the visible light returning from the wafer W is transmitted through the objective lens 32.
, a dichroic mirror 30, a lens 15, and a mirror 19 to reach an imaging device 68 consisting of an fTV or the like. The imaging device 68 displays the observation area on the wafer W on a CRT (not shown).

The reflected light that has passed through the half mirror 44 enters an automatic focus adjustment section 95. The adjustment unit 95 is a device that detects the focused state of the laser beam spot on the wafer W and moves the objective lens 32 or the stage 34 in the optical axis direction until the laser beam spot is focused.

A specific configuration of such a device is described in U.S. Pat. No. 45770.
No. 95, etc., so the explanation will be omitted.

As described above, in this example, various patterns formed on Ueno Three types of optical information,
The configuration is such that the photoelectric conversion means 42, 52, and 64 take out and detect each of them separately. For this reason, these three types of optical information and the scanning amount (position) of the spot light
Edge detection, pattern position detection, and measurement of line width and dimensions of various patterns such as photoresist patterns and polysilicon patterns are performed from multiple angles. As a laser beam source, a HeCd laser with a wavelength of 325 nm or 442 nm, an Ar laser with a wavelength of 488 nm, etc. can be used. In addition to photoresists, materials that exhibit fluorescent properties when irradiated with a laser beam include nitride films (SiN, S13Na), PSO (phosphorus glass), polyimide, etc. in the Uni/'T process. However, this embodiment deals with measurement of patterns using photoresist.

Next, the signal processing circuit of the above embodiment will be explained. An example of such a circuit is shown in FIG. Note that the same reference numerals are used for the same components as those of the apparatus shown in FIG.

In FIG. 2, the operation and signal processing of the entire device are controlled in a unified manner by a host CPU 70, which is comprised of a microcomputer, a minicomputer, or the like. The photoelectric signals from the three photoelectric detectors 42, 52, and 64 are sent to an analog multiplexer (hereinafter referred to as rMPXJ) that arbitrarily selects two of the signals in response to commands from the CPU 70.
It is input to circuit 72. The two selected photoelectric signals are
Each is input as a numerator to a divider (hereinafter referred to as rD I VJ) 74 and 76. Each denominator of DIV74.76 is a digital-to-analog converter (rDA hereinafter) that converts a digital command value from the CPU 70 into an analog signal.
An output signal of 78.80 (CJ) is applied. For this reason, each of the two photoelectric signals is gain-controlled to the optimum level for processing.

DfV? The photoelectric signals from 4.76 are respectively input to sample and hold circuits (hereinafter referred to as rSHJ) 82.84, and then further input to analog-to-digital transformers (hereinafter referred to as rADCJ) 86.88. These 5H82,84
The sampling operation of and the conversion operation of ADCs 86 and 88 are as follows.
This is performed in response to a time-series pulse signal SP output from the scanning amount monitor 18 for each unit scanning amount. That is, the unit movement amount of the spot light on the wafer W (for example, 0.0
The magnitude of the photoelectric signal is sampled every 1 μm) and converted into a digital value. The pulse signal SP is applied to the 5Hs 82 and 84 and the ADCs 86 and 88 via a gate circuit 90 that opens and closes in response to commands from the host CPU 70. The digital values of the photoelectric signals converted by the ADCs 86 and 88 are stored in random access memories (
RAM) 92.94 in address order. RAM92
．． The access address of RAM 92.94 is configured to be updated sequentially in response to the pulse signal SP.
As disclosed in, for example, Japanese Unexamined Patent Publication No. 59-187208, a waveform corresponding to a scanning position of a photoelectric signal is stored.

The waveform data of the photoelectric signal thus stored in the RAM 92.94 is read into the CPU 70 and subjected to various processing.
Pattern positions, edge positions, etc. are detected. One method for this detection is disclosed in detail in Japanese Patent Application Laid-Open No. 187208/1983, so the explanation thereof will be omitted here. The CPU 70 also includes a scanning unit 16, a burping glass 20,
A drive unit 22 and a potentiometer 24 are connected to each other, so that the drive control of the scanning section 16 and the drive unit 22 is performed, and the shift amount of the laser beam can be detected by the potentiometer 24.

The stage drive unit 340 includes a device that drives the stage;
and an encoder that outputs a signal corresponding to the amount and direction of movement of the stage. The output of that encoder is CP
The CPU 70 controls the drive of the stage drive section 340.

The automatic focus adjustment section 95 starts operation upon receiving a signal from the CPU 70, and stops the operation when the focus adjustment is completed, and sends a signal to that effect to the CPU 70.

MPX201.5H202, ADC203, slave C
The PU 204, the laser shutter drive unit 110, and the neutral density filter drive unit 130 determine whether the portion irradiated with the laser beam is photoresist or not, and if it is photoresist, the PU 204, the laser shutter drive unit 110, and the neutral density filter drive unit 130 determine whether or not the portion irradiated with the laser beam is photoresist. Controls the laser shutter 11 or the variable dark filter 13 disposed adjacent thereto. These will be explained in detail below.

The MPX 201 receives the outputs of the reflected light detector 52 and the fluorescent light detector 64, and selectively sends one of the outputs to the 5H 202 according to a selection signal from the slave CPU 204. 5H2
The sampling operation of 02 and the conversion operation of ADC 203 are as follows:
This is performed in synchronization with a time-series pulse signal from the CPU 204.

The slave CPU 204 receives the output of the ADC 203 corresponding to the output of the reflected light detector 52 or the fluorescence detector 64, and determines whether the portion currently irradiated with the laser beam is a photoresist. As described above, the photoresist portion can be identified based on the output of the fluorescence detector 64.

The reason why the photoresist portion can be identified by detecting reflected light is as follows. That is, a photosensitive material such as a photoresist absorbs a large amount of light, becomes exposed to light, and emits fluorescent light, so its reflectance is low. Therefore, when several types of patterns are mixed on the wafer W, the location where the output of the reflected light detector 52 is lowest can be determined to be the photoresist.

Furthermore, when a laser spot is placed on a photoresist, the resist changes in quality due to chemical changes over time and its refractive index changes, resulting in a change in reflectance.

Therefore, the output signal of the reflected light detector 52 changes over time as shown in FIG. 14. By paying attention to this fact, it is possible to determine whether the irradiated portion of the spot is photoresist or not, depending on whether the reflectance changes over time.

In this way, it can be determined whether or not the spot irradiated portion is photoresist based on one or both of the low reflectance and the change in reflectance over time.

When it is determined that the area irradiated with the laser beam is photoresist, a drive signal is sent to the shutter drive unit 110 to close the shutter 11 to block the laser beam, or a drive signal is sent to the filter drive unit 130. is sent to lower the transmittance of the variable neutral density filter 13 and attenuate its intensity. As the variable attenuation filter 13, for example, a configuration in which a plurality of ND filters are appropriately inserted and withdrawn within the laser beam can be mentioned.

3 to 6 are flowcharts showing the operation of this embodiment. Fig. 3 is a general flowchart showing the entire measurement operation, Fig. 4 is a flowchart showing the measurement algorithm, Fig. 5 is a flowchart showing the algorithm for determining whether or not edge measurement is performed normally. 1 and 2 each describe a flowchart showing a series of operations to prevent damage to the photoresist portion. The operation of this embodiment will be explained below according to these flowcharts.

When the measurement operation is started in step S31 in FIG.
The position of the stage 34 is adjusted so that the convex portion (convex) is located at the center of the field of view observed by the objective lens 32. Then, the rotational position of the image rotating prism 28 is adjusted so that the scanning trajectory of the spot light is orthogonal to the linear pattern.

Next, the CPU 70 moves to step S33, where the slave C
The shutter drive unit 110 is operated via the PU 204 to open the shutter 11.

When this is completed, the CPU 70 proceeds to step S34, starts the automatic focus adjustment section 95, and focuses the spot of the laser beam on the wafer W.

Subsequently, in step S35, it is checked whether or not an abnormality has occurred in a series of measurement operations including an automatic focus adjustment operation.

If a positive result is obtained in step S35, that is, if an abnormality has occurred in the operation of the apparatus, the process moves to step S42, where the measurement operation is immediately interrupted and a warning operation to that effect is performed.

If a negative result is obtained in step 335, that is, if the operation of the device is normal, the process moves to step S36, and the CPU
70 is the DI used for gain control in Figure 3.
Output a fixed value to V74.76. As a result, after the signal gain is set constant, laser spot scanning is performed only once, the signal of the measurement pattern is digitized, and the signal is stored in the RAM.
Store at 92.94. This process corresponds to steps S1 and S2 in FIG. 4. The signal level of the pattern is measured by monitoring the digital data (step 536).

Thereafter, the CPU 70 moves to step S37, and controls the transmittance of the neutral density filter 13 by operating the filter drive unit 130 via the slave CPU 204 according to the measurement result in step S36. This adjusts the intensity of the laser beam to a level suitable for pattern measurement. Since the intensity of the laser beam is controlled in this way, the above DIV
It is possible to bring about the same effect as automatic gain control using 74.76 and DAC78,80.

In step 338, a pattern measurement operation is performed, which will be described in detail below.

When the measurement operation is completed, the CPU 70 proceeds to step S39, and operates the shutter drive unit 110 via the slave CPU 204, closes the shutter 11, and performs an operation to cut off the laser beam.

Then, in step S40, it is determined whether all pattern measurements have been completed, and if a negative result is obtained, the process returns to step S32, and if a positive result is obtained, step S
The process proceeds to step 41 to end the measurement operation.

Next, the pattern measurement operation (step 838) will be explained.

The CPU 70 is equipped with a function for inputting information such as the material of the specimen from the outside (operator).
From 70 onwards, depending on the shape and material of the pattern to be measured,
The command for selecting which photoelectric signal to use is M P
It is output to the X circuit 72.

For example, if the material of the pattern is metal (aluminum, etc.), it is desirable to measure using scattered light or specularly reflected light, and in the case of photoresist, it is possible to measure using either method, but in particular fluorescent light and other Measurement accuracy is significantly improved when measuring using both scattered light and specularly reflected light. Therefore, in the following description, it is assumed that the width of a linear pattern made of photoresist is measured using specularly reflected light and fluorescent light. Therefore, the CPU 70 uses the photoelectric detectors 52 and 64.
MP so that each photoelectric signal is input to DIV74.76
Controls the X circuit 72.

Then, the CPU 70 opens the gate circuit 90 and the scanning unit 16 until the spot light crosses the linear pattern from the scanning start point.
drive. Note that the burping glass 20 is held at an appropriate position. As a result, waveform data as shown in FIGS. 7(A) and 7(B) are stored in each of the RAMs 92 and 94.

In this case, it is assumed that, in addition to the linear photoresist pattern PP to be measured, a polysilicon or metal pattern PM is formed on the scanning trajectory of the spot light, as shown in FIG. In this case, the only pattern on the wafer W that generates fluorescence is the photoresist (organic material) layer, and the linear pattern PP can be identified by examining the portions of high intensity in the waveform data of the fluorescence. can.

As shown in FIG. 7(A), when the spot light LS scans the patterns PM and PP in the X direction (see FIG. 8(A)), the photoelectric signal due to the specular reflection stored in the R'AM92 is In the waveform, peaks and bottoms appear according to changes at the edges of the patterns PM and PP.

On the other hand, as shown in FIG. 7(B), in the waveform of the photoelectric signal caused by fluorescence stored in the RAM 94, a level change with simple rises and falls due only to the photoresist pattern PP appears.

For photoresist patterns, the base of the pattern (
It is important to measure the line width near the bottom (bottom), but the signal waveform due to fluorescence faithfully reflects the cross-sectional structure of the photoresist pattern PP shown in FIG. As shown in Figure (B), if the slice level V is set slightly higher than the background noise of the signal waveform and sufficiently lower than the maximum value, the base part at the stepped edges at both ends of the pattern PP can be position
I and XZ can be found accurately.

Therefore, the CPU 70 selects the photoelectric signal as described above, starts laser beam scanning, and initializes parameters (for example, the number of laser beam scans N) (step S1 in FIG. 4). Loading signal waveform data as shown in figure (B) (step S2 in the figure),
A comparison is made with the slice level V, and the positions XI,X
Z is detected (step S3 in the figure). This position xI,
XZ is the difference in the number of addresses from the address AD o of the RAM 94 corresponding to the position x0 of the scanning start point to the addresses AD, , ADZ corresponding to the positions xl and XZ, respectively, and the unit scanning amount (
For example, it is calculated from 0.01 μm).

However, if it is sufficient to measure the line width using only fluorescent light, it is sufficient to calculate the product of the number of addresses between AD2 and AD and the unit scanning amount without determining the positions X, , Xz.

When positions xl, From the waveform, pattern P
Position of stepped edges or interval between edges (line width)
is detected (see step S4 in the figure).

Next, the fifth step is to determine whether the above detection was performed normally or not.
The determination is made based on the flowchart shown in the figure (step S5 in FIG. 4).

First, the data stored in the RAM 92 is sequentially processed to determine the rising start position X1 (step 551).
. For example, X, ++ Xz>α (α is a sufficiently large value)
Place the point X after X.

Next, the rising end position X is determined (step 55
2). For example, X. + -Xi <Q and l Xl-
xi,,,l <β (n is 5 or more, β is a sufficiently small value)
Let the point X be X.

Next, the peak point X immediately before the cough X is determined (see step S53). For example, X, -X,.

〉0,X,. , -X□ is 0, and let it be X.

An example of these X, , Xb, and Xp is shown in FIG.

Next, with (X, , Xp) as the processing interval, all data within this interval are numerically differentiated (step 554). Specifically, find X, -X, . Then, by determining whether all of this differential data is positive or not (step 55
5) Determine whether there is an error in edge detection.

For example, when there is no influence of grain, the signal waveform of the reflected light becomes as shown in FIG. 10(A).

Therefore, the differential data becomes positive as shown in the same figure (B),
Edge detection will be performed normally (step 3
56).

On the other hand, if there is an influence of grain and the signal waveform of the reflected light is as shown in Figure 11 (A), the differential data will not be all positive as shown in Figure 11 (B), making it difficult to detect edges. An error will occur (step 557).

In the case of a reflected light waveform that is not affected by grain, edge detection is performed normally and edge positions P1 and PP are calculated (step S6 in FIG. 4). Then, the number of successfully measured scans is counted as +1, and
C(PP-PI) is calculated for a predetermined coefficient Bc, and the line width Wk is calculated (step S7 in the figure).

If the reflected light is affected by grain, it is determined in step S5 that the edge is not detected normally (step S55.557 in FIG. 5).
), the processes of steps S6 and S7 described above are skipped.

Next, the drive unit 22 is driven by the CPU 70,
The burping glass 20 rotates by a predetermined amount.

Therefore, as shown by the arrow in Figure 12, the spot of the laser beam shifts in the Y direction perpendicular to the scanning direction in the X direction, and the spot is scanned at a different position (Step 4 in Figure 4). S8). At this time, RAM92
is cleared, and signal data is newly written by reflected light through shifted scanning.

Then, the process of detecting the edge again, calculating the line width, driving the burping glass 20, and shifting the scanning position (steps s2 to S8 in the figure) is repeated n times. When n scans are completed (step S9 in the figure), the average value W=Σ W i /K of the data of the normally measured data is calculated, and this is used as the line width dimension value. In this way, the influence of grains on the pattern can be eliminated, or in other words, edges can be detected only when the signal waveform of the reflected light is as shown in Figure 2 (A), and fine variations in edge position can be eliminated. Measurements are performed by averaging.

In this way, it is possible to avoid the influence of grains, and since the same part is not scanned more than once, deterioration of the photoresist can be prevented, and even if there are minute variations in edge position, By averaging this, it is possible to perform line width measurements with stable and high reproducibility.

In addition, when using scattered light instead of reflected light,
The waveform of the photoelectric signal from the photoelectric detector 42 may be stored in the RAM 92 by the MPX circuit 72. Further, although FIG. 5 shows the case where an error is detected for the rising portion of the No. 3 waveform, the same applies to the falling portion.

Next, referring to FIG. 6, an explanation will be given of an operation for preventing damage to the photoresist caused by a laser beam.

The slave CPU 204 constantly monitors whether the beam is on the photoresist by detecting reflected light or fluorescent light. If it is detected that it is on the photoresist, it sends an interrupt signal to that effect to the CPU 70 (step 561).

Upon receiving this signal, the CPU 70 learns that the photoresist is being irradiated with the beam, and first determines whether the laser spot is being scanned in the X direction (step 562), which corresponds to the measurement operation in step 338. This means whether or not laser scanning is being performed only once for pattern signal measurement in step 336.

If it is being scanned, the CPU 70 sends a command to the CPU 204, and controls the transmittance of the neutral density filter 13 by causing the CPU 204 to operate the filter drive section 130. This transmittance is set to a value that is likely to cause damage to the photoresist when the laser spot is stopped, but is not damaged when the laser spot is being scanned.

Further, the CPU 70 determines whether the laser spot is repeatedly scanning the same location on the photoresist in the X direction (step 564).

Whether or not to repeat scanning is determined by the user in advance by the CPU 7.
Input to 0. Which one to use is determined depending on whether emphasis is placed on improving the reproducibility of measured values through the averaging effect or on improving throughput.

If the photoresist has been repeatedly scanned, the CPU 70 determines whether the photoresist has been significantly damaged by this scan and its line width dimension has decreased (step 56).
5). This determination is made by the CPU 70 repeatedly measuring the line width dimension value of the photoresist in the above measurement operation and detecting changes therein.

If the dimension values have changed, the photoresist has already been damaged, so it is determined whether to change the measurement location or interrupt the measurement based on the information input in advance to the CPU 70 (step 566). ). In the case of a pattern such as that shown in FIG. 12, it is effective to shift the laser spot in the Y direction in order to average out variations due to edges or to avoid damaged areas for measurement. However, in measuring a pattern equivalent to a circle called a contact hole, it is necessary to measure the diameter of the circle, so it is not possible to shift it in the Y direction. Therefore, based on this, the user decides whether or not to shift in the Y direction.

When changing the measurement location, it is shifted in the Y direction (step 567), and the laser spot is scanned in the X direction to continue measurement (step 568). This step S67.368 is the same process as step S8 described in FIG. 4.

Then, it is determined whether the number of repeated measurements has reached a predetermined value input to the CPU 70 in advance (step 569).
). This determination corresponds to step S9 in FIG. 4.

If the line width value has not changed in step S65, it is determined that there is almost no damage to the photoresist, and the process jumps to step S69.

If the predetermined value has been reached, the average line width is calculated in the same manner as step S10 in FIG. 4 (step 570), the measurement operation is ended, and the process returns to pattern scanning in step S36 in FIG. 3, and in step S37. whether to control the filter 13;
Alternatively, the process returns to the pattern measurement process in step 338 and proceeds to step S39 (step 571).

If the number of measurements has not reached the predetermined value in step S69, the spot scan in the X direction is repeated and the measurement is continued (step 572), and the process returns to step S65.

If repeated scanning is not performed in step S64, or if the laser spot is not shifted in the Y direction in step 366, the process jumps to step S71 and the measurement operation is ended.

Next, in step S62, photoresist is detected,
If the laser spot is not being scanned, the laser beam is automatically focused (specifically in step S33.544).
) is performed (step 573).
.

If it is determined in step S73 that adjustment is not in progress,
This means that the laser spot is focused or close to being focused on the photoresist, and a spot with high energy density is focused on the resist, which is likely to damage the resist. Therefore, in this case, immediately
The CPU 70 controls the shutter drive unit 1 via the CPU 204.
10, the laser shafter 11 is activated during the laser beam.
is inserted (step 574).

Then, the CPU 70 sets an error flag in the built-in RAM, ends this process, and returns to step S34 in FIG.
It is determined that there is an error in step S35.

If it is determined in step S73 that automatic focus adjustment is in progress, C
The PU 70 temporarily suspends its focusing operation (step 876).

The CPU 70 causes the laser spot to scan in the X direction or the Y direction, and at the same time causes the CPU 204 to detect whether or not there is a spot on the photoresist.
order to send information to that effect. When the shutter 70 is informed of the discrimination information that the material is not a photoresist by the CPU 204 during the scanning operation, the shutter 70 stops the laser spot at that position. The CPU 70 itself may determine whether it is a photoresist by monitoring the output of the reflected light detector 52 or the fluorescent light detector 64 (step 878). This prevents damage to the photoresist due to spot irradiation during focus detection.

If a position other than the photoresist is detected as a result of this operation, the automatic focus adjustment operation shown in FIG. 3 is restarted (step 579), and after the adjustment operation is completed, the process is ended.

If no position other than photoresist is detected within the scanning range of the laser spot in step S77,
Jumping to step 374, inserting the shutter 11 into the laser beam, setting an error flag in step S75,
This process ends in step S75. Then, in step S35 of FIG. 3, it is determined that there is an error.

The embodiments in which the present invention is applied to a device for measuring the line width of a pattern have been described above, but the present invention is not limited to this and can be applied to any device that requires pattern detection. Available.

For example, it can be applied to detecting alignment marks in exposure equipment for manufacturing semiconductor devices.

Furthermore, instead of scanning the spot light as in this embodiment, the stage 34 is moved, and instead of the scanning amount monitor 18, a light wave interference length measuring device is provided to detect the position or the amount of movement of the stage 34. You can get exactly the same effect.

In addition, in this embodiment, it is used for pattern detection,
Although a laser beam is used as the beam that causes tR scratches on the detection target, the present invention is not limited to this, and other energy beams may be used as long as they can be used for pattern detection and damage the detection target. Examples include electron beams and ion beams.

Further, examples of the measurement target (sensitive material) that can be damaged by the energy beam for pattern detection include photosensitive materials other than photoresist, such as magneto-optical materials, nitride films, etc., and heat-sensitive materials are also considered.

In this example, the energy beam for measurement and the energy beam for identifying the sensitive substance are the same laser beam, which has the effect of simplifying the configuration. The present invention is not limited to this, but also includes one in which separate beams are used for measurement and sensitive substance discrimination. For example, when measuring or observing a photoresist using an electron beam, excessive exposure to the electron beam may cause damage. In this case, damage to the photoresist can be prevented by identifying the photoresist by irradiating it with a laser beam similar to this embodiment and weakening the irradiation of the electron beam in accordance with this determination information.

Furthermore, when measuring the dimensions of a photoresist pattern by repeatedly scanning an electron beam, step S in FIG.
Measurement can be performed without causing major damage to the photoresist by performing operations similar to those from 65 to S70.

Next, a second embodiment of the present invention will be described with reference to FIGS. 14 and 15. FIG. 14 shows the arrangement of the optical system of the fluorescence detection system, and FIG. 15 is a circuit block diagram of a signal processing system corresponding to the optical system of FIG. 14. The switching mirror 58 shown in FIG. 1 is replaced with the switching mirror 58 shown in FIG. 14.
8a, the fluorescent light from the mirror 60 is divided into two by a half mirror 100, and the transmitted fluorescent light is separated by an optical bandpass filter 102, and then passed through a photomultiplier (hereinafter referred to as a photomultiplier). 104
The light is received by the On the other hand, the fluorescent light reflected by the half mirror 100 is divided into two parts by the next half mirror 106, and the fluorescent light reflected here is separated by the band pass filter 108 and then received by the photomultiplier 110.

The fluorescent light transmitted through the half mirror 106 is reflected by a mirror 112, separated into spectra by a bandpass filter 114, and received by a photomultiplier 116. In addition, Me7-112
, the bandpass filter 114, and the photomultiplier 116 are
In some cases, it is not necessary. Here, the hand-pass filters 102, 108, and 114 have different wavelength bands and separate the fluorescent light into two or three bands. This is to separate the fluorescence generated on the wafer W into the fluorescence from the photoresist and the fluorescence from other substances, such as dust and defects, and detect them separately.

For example, when a He-Cd laser with a wavelength of 325 nm is used, fluorescence due to photoresist is detected by a band pass filter 102 with a wavelength of 550 to 650 nm, and fluorescence due to dust is detected with a band pass filter 102 of 550 to 650 nm.
It can be detected with a band pass filter 108 of 00 to 500 nm. In addition, when detecting all kinds of fluorescent light, the bandpass filter 114 is used in the wavelength range 400 to 750.
to nm. If there is another fluorescent substance, it is sufficient to perform spectroscopy according to its wavelength. In this way, only the fluorescent light from the photoresist can be reliably detected, and when measuring with scattered light or specularly reflected light, even if the scattered light intensity or specularly reflected light intensity changes due to dust, It becomes possible to distinguish and detect the pattern to be measured and dust. Therefore, if fluorescence from dust is detected during measurement, the measurement is invalidated and the stage 34 is moved or the laser beam spot is shifted in the direction intersecting the scanning line to find a place free of dust. You can perform measurements again using

If fluorescence from dust is detected, the measures to prevent damage to the photoresist (steps S63 and S67.577) shown in the flowchart of FIG. 6 are unnecessary and are therefore not carried out.

Photoelectric detector 52.42.104 as shown in FIG.
．． Each photoelectric signal from 110.116 is MP
Analog multiplexer (MPX) similar to X circuit 72
) The circuit 120 is manually powered. The MPX circuit 120
A signal wavelength detection unit CH that selects any three photoelectric signals from one photoelectric signal and each has a divider, a sample hold circuit, an analog-to-digital converter, and a RAM.
.

, CHz, CH,. The MPX circuit 120 is
A command from the CPU 70 determines which photoelectric signal to select. For example, a signal waveform due to specular reflection light is stored in Uniso 1-CH, a signal waveform due to scattered light is stored in unit CH2, and a signal waveform due to fluorescence (550 nm to 650 nm) from a photoresist is stored in unit CH3. . By examining the signal waveforms of units CH+ and CH2 using the signal waveform of unit CH:l as a reference, the position, line width, size, etc. of pattern edges can be detected from more angles. In addition, by storing the signal waveforms of the fluorescent lights separated into each unit CH1, CH2, and CH3, it is possible to detect patterns by photoresist, detect the presence or absence of adhesion of dust, etc., and detect the presence of other organic substances. becomes possible. Incidentally, the fluorescent light generated on the wafer W may be increased or decreased into one or four parts as necessary, instead of being divided into three parts. Further, the number of signal waveform detection units does not need to be three, and may be two.

〔Effect of the invention〕

With this configuration, patterns formed by the sensitive material and exposed patterns are automatically distinguished from patterns that are not, and the energy beam is irradiated to the pattern formed by the sensitive material in a manner that damages it. Since this is prevented, highly accurate pattern detection is possible without damaging sensitive materials such as photoresists.

[Brief explanation of drawings]

FIG. 1 is a layout configuration diagram showing an optical part of an embodiment of a pattern position detection device according to the present invention, FIG. 2 is a circuit block diagram showing an electrical component of the same embodiment, and FIG. Flowchart showing the overall process of the same example, No. 4
Figure 5 is a flowchart showing an example of an algorithm for pattern detection, Figure 5 is a flowchart showing an example of an algorithm for determining errors in edge detection, Figure 6 is a flowchart showing operations to avoid damage to photoresist, and Figure 7 is A diagram showing edge detection using reflected light and fluorescent light, FIG. 8 is an explanatory diagram showing an example of a pattern, FIG. 9 is an explanatory diagram of the error detection procedure, and FIGS. 10 and 11 are examples of error judgment. FIG. 12 is an explanatory diagram showing the effect of burping glass, FIG. 13 is a diagram showing changes in reflectance of photoresist over time, and FIGS. 14 and 15 are examples of ponds of the present invention. FIG. 2 is a block diagram showing the main parts of the system. Explanation of symbols of main parts 10...Laser beam source, 16...Scanning section, 18.
...Scanning amount monitor, 20...Burping glass, 2
2... Drive unit, 24... Potentiometer,
34... Stage, 42.52.64... Photoelectric detector, 70... CPU, 72... Multiplexer circuit, W... Wafer

Claims (2)

[Claims] (1) A first irradiation means for irradiating a first energy beam onto a test object; receiving a second energy beam emitted from an irradiated part of the test object that has been irradiated with the first energy beam; A pattern detection device comprising: a detection means for detecting a pattern of a portion; a second irradiation means for irradiating the object with a third energy beam; A determining means for receiving a fourth energy beam emitted from the irradiated portion and determining whether or not the portion to be irradiated with the third energy beam is a portion sensitive to the primary energy beam; and receiving the output of the determining means. and a control means for controlling the first energy beam when the first energy beam irradiates the sensitive portion to prevent the sensitive portion from being damaged by the first energy beam. pattern detection device. (2) The first irradiation means and the second irradiation means are used together, and the first energy beam and the third energy beam are the same beam. Pattern detection device.