JP3432739B2 - Pattern inspection device and inspection method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an inspection apparatus and inspection method for fine patterns, and more particularly, it compares pattern positions of fine circuit patterns formed on a photomask, a reticle, a wafer, etc. with design data. The present invention relates to an inspection device and an inspection method for evaluating accuracy.

[0002]

2. Description of the Related Art A semiconductor integrated circuit has a desired size of 4,
A mask pattern formed 5 times, 10 times, etc. is formed on the reticle, and the circuit pattern on this reticle is reduced and projected to 1/4, 1/5, 1/10 etc. by ultraviolet rays or far ultraviolet rays, It is generally manufactured by repeating the process of baking on the photoresist applied above and performing selective etching or selective doping using the pattern of this photoresist as a mask together with the thin film forming process. is there. The high performance of DRAMs and MPUs in recent years is largely due to the microfabrication technology centered on the photolithography technology of such semiconductor integrated circuits.

However, in order to meet the demand for higher performance of semiconductor integrated circuits, innovation in fine processing technology is required. For example, in terms of reticle quality, which is at the core of photolithography technology, the requirements for position accuracy, dimensional accuracy, and defect-freeness are becoming more stringent than before. With the improvement of reticle manufacturing technology such as circuit pattern drawing technology and mask process technology, it is essential to improve the inspection technology such as pattern inspection equipment and inspection method in order to finally guarantee the quality of the manufactured reticle. .

As a conventional method for measuring the positional accuracy and dimensional accuracy of a pattern, there is a method of optically detecting scattered light at a pattern edge on a reticle. In this method, a finely focused laser beam is scanned in the vicinity of a pattern edge of a sample held on a traveling stage, scattered light of the laser beam is detected, and the stage position at that time is read.

[0005]

However, the conventional method for measuring the positional accuracy and dimensional accuracy of the pattern has the following problems.

The scattered light intensity changes depending on the pattern size, and the pattern edge position changes.

If the number of repetitions is increased in order to improve the accuracy, the time required for measurement may increase, which may cause a temporal drift of the detection signal.

Calibration is required for correspondence with the actual pattern.

On the other hand, the LSI pattern has reached the deep submicron to nanometa level, and tends to be further miniaturized in the future. As a matter of course, it is expected that stricter positional accuracy will be required for the reticle on which these increasingly finer circuit patterns are formed. In view of the above problems, the problem that conventional pattern inspection devices and inspection method methods are not sufficient to evaluate strict position accuracy with high throughput, which will be required in the future, is gradually highlighted. Is coming.

In view of the above problems, it is an object of the present invention to provide a pattern inspection apparatus capable of evaluating pattern position accuracy with sufficient throughput and having high resolution.

Another object of the present invention is to provide a pattern inspection method method capable of evaluating pattern position accuracy with sufficient throughput and having high resolution.

[0012]

In order to solve the above-mentioned problems, a first feature of the present invention is to provide a sensor data generating section for generating sensor data, a reference data generating section for generating reference data, and a sensor data. The pattern inspection apparatus includes at least a multiplier that multiplies reference data by each other. Here, the sensor data generation unit photoelectrically converts the optical image of the pattern on the measured sample having lines / spaces of the pitch p in a certain direction to generate sensor data, and the reference data generation unit measures the measured sample . The held position coordinates of the stage and the design data of the two-dimensional graphic data are input, and the position coordinates or the two-dimensional graphic data are converted to generate reference data having a line / space of pitch q. That is, sensor data having a constant pitch p,
The first is that it is a pattern inspection device that detects a phase by synchronously detecting reference data having a pitch q different from this.
It is a feature of.

In the first feature of the present invention, the reference data generating section is an input device for inputting position coordinates of a stage holding a sample to be measured and design data of two-dimensional figure data, and the position coordinates and two-dimensional figure. It is preferable to have a reference data generation circuit that converts data to generate reference data and a first subtraction circuit that subtracts the first offset value from the reference data. In addition, the sensor data generation unit includes a photoelectric converter that converts an optical signal from a pattern on the measured sample into an electric signal, and an A / A connected to the photoelectric converter.
It is preferable to have a D converter and a second subtraction circuit that subtracts the second offset value from the output of the A / D converter.

According to the first feature of the present invention, it is possible to provide a high resolution pattern inspection apparatus capable of evaluating pattern position accuracy with sufficient throughput.

A second feature of the present invention is that the step of photoelectrically converting an optical image of a pattern on a sample to be measured having a line / space with a pitch p in a fixed direction to generate sensor data, and holding the sample to be measured. The position coordinates of the stage and the design data of the two-dimensional figure data are input, the position coordinates or the two-dimensional figure data are converted, and the line of pitch q /
It is a pattern inspection method including at least a step of generating reference data having a space and a step of multiplying the reference data and the sensor data with each other.
That is, multiply the reference data and the sensor data by each other,
A second feature is a pattern inspection method in which sensor data having a constant pitch and reference data having a pitch different from this are synchronously detected to detect a phase.

In the second aspect of the present invention, it is preferable that after the step of multiplying, the method further comprises the steps of passing the multiplication result through a low-pass filter and detecting the zero point of the output signal of the low-pass filter.

According to the second feature of the present invention, it is possible to provide a high-resolution pattern inspection method capable of evaluating pattern position accuracy with sufficient throughput.

[0018]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of a pattern inspection apparatus according to an embodiment of the present invention. As shown in FIG. 1, the pattern inspection apparatus includes a reference data generation unit 31 that generates reference data, a sensor data generation unit 32 that generates sensor data, and a multiplier 21 that multiplies the reference data and the sensor data by each other. At least.
Here, the reference data generation unit 31 inputs the position coordinates of the stage holding the sample 3 to be measured and the design data of the two-dimensional graphic data, and the reference data by converting the position coordinates and the two-dimensional graphic data. It has a reference data generation circuit 13 for generating data and a first subtraction circuit 14 for subtracting the first offset value from this reference data. The sensor data generation unit 32 also includes a photoelectric converter 15 that converts an optical signal from a pattern on the measured sample 3 into an electric signal, and an A / D converter 1 connected to the photoelectric converter 15.
6 and a second subtraction circuit 17 that subtracts the second offset value from the output of the A / D converter 16. In Figure 1, as the measured sample 3, but illustrates the reticle, the measured sample 3 is not limited to the reticle as a matter of course. Various photomasks such as a photomask for contact exposure and a photomask for proximity exposure may be used, and patterns such as an actual wiring layer provided on a semiconductor wafer may also be inspected.

Further, as shown in FIG. 1, a low-pass filter (LPF) 22 is connected to the multiplier 21, and a zero-point detection circuit 23 for detecting the zero point of the output signal of the low-pass filter 22 is connected to the low-pass filter (LPF) 22. Are connected. A pattern edge measuring circuit 24 is connected to the zero point detecting circuit 23.

The optical signal from the pattern on the measured sample 3, the light from the light source 1 of a halogen lamp or a mercury lamp or the like in the measured sample 3 through the objective lens 2 and irradiated to the measured sample 3 to be measured It is obtained by condensing the light transmitted through the sample 3 with the condensing lens 4. Although not shown, it goes without saying that the measured sample 3 is placed on a predetermined optical stage.
The optical signal from the pattern on the DUT 3 is a MOS
Photoelectric conversion is performed by a photoelectric converter 15 such as an image sensor or a CCD image sensor, and further A / D converted by an A / D converter 16.

An optical signal from the pattern on the non-measurement sample 3 is photoelectrically converted by a photoelectric converter 15 such as a MOS image sensor or a CCD image sensor, and further the A / D converter 1 is used.
A / D conversion is performed at 6.

FIG. 2 specifically shows the pattern on the reticle 3. As shown in FIG. 2A, a device pattern section 50 is provided in the center of the reticle 3, and LSI patterns and the like are arranged therein. Marks for measuring pattern position deviation, that is, alignment marks 51, 52, 53, 54 are provided on the periphery of the device pattern portion 50. An alignment mark 51 that constitutes sensor data,
Reference numerals 52, 53 and 54 are, for example, line / space patterns as shown in FIG. 2B, that is, binary patterns in which lines and spaces are alternately repeated at a cycle p in the X direction or the Y direction. For example, it may be considered that the line is a light shielding film pattern and the space is a glass pattern. Only the X direction will be described below, but it will be easily understood that the same applies to the Y direction.

The pattern on the reticle 3 does not follow the design data due to the influence of the circuit pattern drawing technology and the mask process technology. The reticle may be distorted due to the thermal effect in the mask process, and this distortion is not always isotropic. Further, in actual measurement, various miscellaneous situations occur such that the reticle 3 is arranged on the optical stage of the pattern inspection device by slightly rotating the XY plane. Therefore, at the nanometer level, the measured
The position of the pattern edge of the pattern on the constant sample 3 is usually different from the design data. The problem is whether or not this positional deviation falls within an allowable range in the semiconductor manufacturing process.

Next, a pattern inspection method for evaluating the pattern position accuracy by comparing the position of the pattern edge on the reticle with the design data according to the present invention will be described.

(A) First, the optical image of the pattern 51 on the sample 3 to be measured having the lines / spaces with the pitch p in a certain direction is photoelectrically converted by the photoelectric converter 15, and further A / D is converted by the A / D converter 16. D conversion is performed to generate sensor data. Figure 3
In FIG. 4A and FIG. 4A, the photoelectric converter 15 and A /
The sensor output which is the output of the D converter 16 is shown. This sensor output is the line of the alignment mark 51 in the X direction /
It is a photoelectric conversion of the optical image from the space pattern. However, in the present invention, the alignment marks 51, 5
Not only the line / space pattern on the lines 2, 53, 54 but also various patterns in the device pattern portion 50 may be applied. The sensor output shown in FIG. 3A and FIG. 4A is the A / D converter 16 expressing the light and darkness of the optical image from the alignment mark 51 by a digital numerical value. have p.

[0027] (b) Next, via the input device 11, the
The position coordinates of the stage holding the measurement sample 3 and the design data of the two-dimensional graphic data are input. The position coordinates of the stage and the design data of the two-dimensional graphic data are input to the reference data generation circuit 13.

(C) In the reference data generating circuit 13, x '= p. (X-x ₀ ) / q + x ₀ (1) is converted to the input stage position coordinate x'. Here, the position coordinate of interest is set to x = x ₀ . Alternatively, for the position coordinates of trapezoidal or polygonal graphic data that describes the line / space pattern,
Obtain the relative coordinates from the position coordinate x = x _{0 of} interest,
Scaling is performed by multiplying by a constant magnification q / p. As a result, the reference data generation circuit 13 detects the sensor data pitch p.
On the other hand, a line / space pattern having a pitch of q can be generated as reference data. For example, q = (n + 1) · p / n (2) or q = (n−1) · p / n (3). Here, n is an integer. Further, the position coordinate of interest is set to x = x ₀ . The value of n will be described later.
The larger the value, the higher the accuracy. Figure 3 (b)
4B shows reference data corresponding to the cross section of the mask pattern in the X direction.

FIG. 3 shows a case where the positional deviation ε between the sensor pattern and the reference pattern at x = x ₀ is 0, that is, “no positional deviation”. As shown in FIG. 3, when the expression (2) can be used,
Sensor pattern (a) of the (n + 1) th cycle at the pitch p
And the reference pattern (b) of the nth cycle at the pitch q is x
= Again at x ₁ . FIG. 4 shows a case where the positional deviation ε between the sensor pattern and the reference pattern at x = x ₀ is positive, that is, “with positional deviation”. It is x = x that the pattern edges of the sensor pattern (a) and the reference pattern (b) match again.
You can see that it has changed to the _second place. Further, as described later, x ₂ −x ₁ = − (n + 1) ε (4). That is, the position where the sensor pattern and the reference pattern at x = x ₀ coincide again changes in proportion to the positional shift ε.

(D) Next, using the first subtraction circuit 14, the average value level of the amplitude is subtracted from the reference data as an offset. Similarly, the second subtraction circuit 17 is used to subtract the average value level from the sensor data as an offset.

(E) Thereafter, the multiplier 21 is used to multiply the offset-subtracted sensor data and the reference data with each other. That is, the sensor data of the pitch p and the reference data of the pitch q are synchronously detected to detect the phase.

(F) Further, the multiplication result is passed through a low pass filter (LPF) 22. FIG. 5A shows a signal corresponding to the cross section in the X direction when there is no displacement. The solid line is the reference signal and the broken line is the sensor signal. Further, FIG. 5B shows a signal corresponding to a cross section in the X direction when there is a position shift. The solid line is the reference signal and the broken line is the sensor signal.

(G) As is apparent from FIG. 5, if the zero point of the output of the low-pass filter 22 is obtained, the positional deviation amount of the pattern edge can be measured.

The above description is summarized by mathematical expressions.
If the x-coordinate with the target position x = x ₀ as the origin is expressed, the coordinate x
Sensor data S (x) and reference data R in
(X) is expressed by the following equations (5) and (6).

[0035]

[Equation 1] Here, the pitch of the sensor data in the x direction is p, and the pitch of the reference data in the x direction is q = (n + 1) · p / n. Further, the pattern position shift between the sensor data and the reference data at the target position x = x ₀ is ε.

FIG. 6A shows the result of synchronous detection in the X direction when there is no displacement. The result is simply the solid line multiplied,
The broken line is the result of applying the low-pass filter 22. FIG. 6B shows the result of synchronous detection in the X direction when there is a position shift. The solid line shows the result of multiplication, and the broken line shows the result of further applying the low-pass filter 22. The zero point of the output of the low-pass filter 22 can be obtained to measure the positional deviation amount of the pattern edge.

As shown in FIG. 6, the synchronous detection result M (x) has a pitch of (n + 1) · p in the x direction, which is a pitch longer than the line / space pitch. That is, the synchronous detection result M (x) is expressed by the following equation (7).

[0038]

[Equation 2] As is clear from the equation (7), M (x) becomes a sine wave having a low frequency compared with the frequency of the line / space. That is, the low-pass filter (LPF) 22 may be any one as long as it cuts components of frequency 1 / p or higher and passes frequency 1 / ((n + 1) · p).

From the equation (7), x ₀ giving the “zero point of the synchronous detection result M (x)” where M (x) = 0 is given by the following (8)
It becomes like a formula.

[0040]

[Equation 3] Here, k is an arbitrary integer. As can be seen from the equation (8), the zero point of M (x) is enlarged by (n + 1) times the pattern position shift ε between the sensor data and the reference data. Therefore, ε can be obtained with high accuracy of (n + 1) th of the accuracy of obtaining the zero point position. Therefore n
The larger the value of, the higher the accuracy.

[0041]

According to the present invention, it is possible to provide a high-resolution pattern inspection apparatus capable of evaluating pattern position accuracy with sufficient throughput.

Further, according to the present invention, it is possible to provide a high-resolution pattern inspection method capable of evaluating the pattern position accuracy with sufficient throughput.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a pattern inspection device according to an embodiment of the present invention.

FIG. 2 is a plan view illustrating an outline of a plane pattern of a reticle used for a pattern inspection according to the embodiment of the present invention.

FIG. 3 is a diagram showing signal intensities in the X direction of a reference pattern and a sensor pattern in the case where there is no displacement (ε = 0) at x = x ₀ .

FIG. 4 is a diagram showing signal intensities in the X direction of a reference pattern and a sensor pattern in the case where there is a positional deviation (ε> o) at x = x ₀ .

5A is a signal level corresponding to a cross section in the X direction of reference data and sensor data when there is no displacement, FIG.
FIG. 5B shows signal levels corresponding to the cross section in the X direction of the reference data and the sensor data when there is a position shift.

6A is a synchronous detection result in the X direction when there is no displacement, and FIG. 6B is a synchronous detection result in the X direction when there is a displacement. Here, the solid line is the result of multiplication, and the broken line is the result of further low-pass filtering.

[Explanation of symbols]

1 light source 2,4 lens 3 reticle 11 Input device 13 Reference data generation circuit 14 First subtraction circuit 15 Photoelectric converter 16 A / D converter 17 Second subtraction circuit 21 Multiplier 22 Low-pass filter (LPF) 23 Zero detection circuit 24 pattern edge measuring circuit 31 Reference Data Generation Unit 32 Sensor data generator 50 device pattern section 51, 52, 53, 54 Alignment mark

─────────────────────────────────────────────────── ─── Continuation of the front page (56) Reference JP-A-10-32161 (JP, A) JP-A-7-243982 (JP, A) JP-A-7-128248 (JP, A) JP-A-6- 349715 (JP, A) JP-A-6-175353 (JP, A) JP-A-4-261538 (JP, A) JP-A-4-100044 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G03F 1/08 H01L 21/027

Claims (6)

(57) [Claims] 1. An optical image of a pattern on a sample to be measured having a line / space with a pitch p in a fixed direction is photoelectrically converted,
A sensor data generation unit for generating sensor data, position coordinates of the stage holding the sample to be measured and design data of two-dimensional figure data are input, and the position coordinates or the two-dimensional figure data are converted to obtain a pitch q. A pattern inspection apparatus comprising at least a reference data generation unit that generates reference data having a line / space, and a multiplier that multiplies the sensor data and the reference data by each other. 2. The reference data generation unit, an input device for inputting the position coordinates and the two-dimensional graphic data, and a reference data generation circuit for converting the position coordinates and the two-dimensional graphic data to generate reference data. And a first subtraction circuit that subtracts a first offset value from the reference data, wherein the sensor data generation unit includes a photoelectric converter that converts the optical signal into an electric signal, and the photoelectric converter The pattern inspection apparatus according to claim 1, further comprising a connected A / D converter and a second subtraction circuit that subtracts the second offset value from the output of the A / D converter. 3. The pattern on the sample to be measured is a pattern having a line / space with a pitch p in a fixed direction, and the position coordinate of interest is x = x ₀ , and the reference data generation circuit inputs X ′ = p · (x−x
₀ ) / q + x ₀ to generate the reference data by performing coordinate conversion to the position coordinate x ′, or the position coordinate x for the two-dimensional graphic data.
3. The relative coordinates from = x ₀ are obtained, scaling is performed by applying a constant magnification q / p, and the two-dimensional graphic data is expanded or reduced to generate the reference data. Pattern inspection device. 4. The pattern inspection apparatus according to claim 1, further comprising a low-pass filter connected to the multiplier, and a zero-point detection circuit that detects a zero point of an output signal of the low-pass filter. 5. An optical image of a pattern on a sample to be measured having a line / space with a pitch p in a fixed direction is photoelectrically converted,
Generating sensor data, inputting position coordinates of the stage holding the sample to be measured and design data of two-dimensional figure data, converting the position coordinates or two-dimensional figure data, and a line / space having a pitch q. And at least multiplying the reference data and the sensor data by each other. 6. The pattern according to claim 5, further comprising a step of passing the multiplication result through a low-pass filter after the step of multiplying, and a step of detecting a zero point of an output signal of the low-pass filter. Inspection method.