JPH0612341B2 - Optical surface inspection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial application] The present invention scans the surface of a sample having a mirror-like surface with a light beam, and receives scattered light generated by unevenness of the sample surface or foreign matter adhering to the sample. The present invention also relates to an apparatus for optically inspecting the presence or absence of these irregularities or foreign matters and the degree of presence thereof, and particularly to an optical inspection apparatus for the surface of a semiconductor substrate in a semiconductor manufacturing process that requires high cleanliness.

[Conventional technology]

In the above-described optical inspection device, in order to perform inspection with high accuracy, it is required that the positioning of the sample for setting the inspection region and the reproducibility of the moving speed of the sample be as accurate as possible.

The scanning method of the light beam in this kind of optical surface inspection device, while moving the sample in the linear direction while rotating,
There are two methods: a method of scanning in a spiral with a fixed light beam, and a method of moving a sample in a linear direction and moving the light beam on a straight line perpendicular to the moving direction at a constant cycle for scanning.

FIG. 6 shows an example of the configuration of the main parts of a conventional optical inspection device that performs spiral scanning.

In FIG. 6 (a), the light projecting means 1 is provided with a light source 2 and a condenser lens 3, and the parallel light flux 4 projected from the light source 2 is perpendicular to the surface of the disk-shaped sample 5 by the condenser lens 3. And is projected as a light beam 6 that is condensed in a spot shape.
Abnormal place 7 such as unevenness or adhesion of fine particles on the surface of sample 5
When the light beam 6 irradiates this, pulsed scattered light 8 is generated. The scattered light 8 is received by a photodetector 9 arranged obliquely with respect to the surface of the sample 5, and a pulsed electric signal 10 corresponding to the amount of received light is output. When the abnormal portion 7 does not exist, the light beam 6 is specularly reflected, and since the direction of specularly reflected light is perpendicular to the surface of the sample 5, this specularly reflected light does not enter the photodetector 9.

The sample 5 is mounted on a rotary stage 11 driven by a drive device 18 having a drive speed control system 181 and a rotation speed control system 182. While rotating around a shaft 12 of the rotary stage 11, the sample 5 is further in a P direction indicated by an arrow. Can be moved in a straight line. That is, in this example, the rotary stage 11 serves both as a scanning unit and a relative moving unit. Since the light projection means 1 is fixed, this movement causes the light beam 6 to move the surface of the sample 5 to the sixth position.
As shown in FIG. 6B, the scanning is performed in a spiral shape, and the pulse-shaped component of the electric signal 10 is counted by the signal processing means 16 including the pulse counting circuit in the inspection area defined by the position of the axis 12, and the counting result 17 is output. And inspect.

FIG. 7 is a perspective view of a main part of an apparatus for scanning with a light beam moving on a straight line as a second conventional example.
In this example, the light projecting means 1 also serves as the scanning means, and the surface perpendicular to the sample 5 or having an angle is defined as the scanning surface 13, and the projection direction of the light beam 6 is cyclical within the scanning surface 13. Then, the surface of the sample 5 is inspected. The line of intersection between the surface of the sample 5 and the scanning surface 13 is the scanning line 14.

A light incident surface consisting of a cylindrical lens parallel to the scanning line 14 is provided in the vicinity of the sample 5, and a light condensing device 15 for converging the incident light by an optical fiber bundle is used for pulse-like scattering from the abnormal portion 7 on the surface of the sample 5. It is arranged at an angle for receiving only the light 8. The photodetector 9 is connected to the light collecting device 15, which outputs a pulsed electric signal 10 due to the scattered light 8, which is counted by the signal processing means 16 to obtain a counting result 17.

The sample 5 is linearly moved in the direction of the arrow Q intersecting the scanning line 14 by the support mechanism 20 driven by the driving device 19 having the driving speed control system 191, whereby the sample 5 is scanned with the light beam 6 over the entire surface. It

[Problems to be solved by the invention]

In the above-mentioned optical surface inspection apparatus, the normal light beam 6
In many cases, the number of times of scanning one abnormal point 7 is deterministically less than once or plural times, and the counting result of the pulse-shaped component of the electric signal 10 does not show the actual number of abnormal points. Therefore, the counting result is calibrated for the actual number of abnormal points and then the inspection is performed. Since the scanning cycle is constant, the number of times the entire inspection area on the surface of the sample 5 is scanned varies depending on the movement speed of the sample 5 and accordingly. Therefore, the number of scanning lines per unit length, that is, the scanning line density depends on the moving speed of the sample 5, and the number of times the same abnormal portion 7 is scanned, that is, the number of detected abnormal portions is the same as that of the sample 5.
Affected by the movement speed of. Therefore, it is necessary to maintain the same moving speed as the moving speed when calibrating the counting result of the pulsed component of the electric signal 10 for the actual number of abnormal points even in the subsequent inspection. In addition, the drive devices 18 and 19 must be provided with drive speed control systems 181, 191 and rotation speed control system 182 that keep the moving speed unchanged. Alternatively, calibration must be performed using a sample with a known number of abnormal points in each inspection.

In the conventional example shown in FIG. 6, the rotary stage 11 which is a sample moving mechanism moves by two kinds of motions, that is, a rotary motion and a linear motion. Therefore, it is necessary to keep the speed in each motion unchanged. Requires a drive speed control system 181 and a rotation speed control system 182.

The moving mechanism shown in the conventional example of FIG. 7 is a linear movement in one direction, which is significantly simpler than that of FIG. 6, but the drive speed control system 19 keeps the linear movement speed unchanged.
Can not miss one.

Therefore, in the optical surface inspection apparatus shown in these conventional examples, it is necessary to provide a sample moving mechanism having a precisely controlled moving speed unique to the apparatus as a part of the apparatus, and the optical system, the photodetector and its signal are required. It was extremely difficult to independently install only the processing system part and appropriately install it in a sample transfer system (for example, a belt type wafer transfer device) of an existing production line.

The present invention eliminates the need for a sample moving mechanism dedicated to the conventional optical surface inspection apparatus, simplifies the structure of the optical surface inspection apparatus, and improves the existing sample transfer apparatus with different feed rates. It is an object of the present invention to provide a device which is applicable and is not affected by the feed rate even when the feed rate is not constant.

[Means for solving problems]

According to the present invention, a proportional relationship is always established between the count value of the pulsed component of the electric signal output from the photodetector that receives the scattered light generated at the abnormal point on the sample surface and the scanning line density of the light beam. In addition, the number of abnormal points is accurately obtained by focusing on the fact that even if the moving speed of the sample is not constant, it can be regarded as constant within a short time.

That is, the signal processing means is provided with a scanning number counting means of the light beam, a moving distance measuring means of the sample, and an arithmetic means connected to a counting circuit for counting the pulse-like component of the electric signal, and the scanning of the light beam over the entire inspection region. The number of scans, the moving distance of the sample, and the number of pulsed components of the electric signal are measured for each of a plurality of times of scanning within a short time obtained by dividing the number of times, and the measured number of scans and the moving distance of the sample The scan line density in a plurality of scans is calculated. On the other hand, a scanning line density that gives a measured value of the number of pulse-like components equal to the known number of abnormal places is previously obtained as a reference scanning line density by a reference sample with a known number of abnormal places, and the scanning line density and the pulse-like component count value The measured pulse-like component count value is converted into a value for the reference scanning line density by using the proportional relationship of.
The number of abnormal points in the sample is calculated by summing these converted values obtained for each of the plurality of scans.

[Action]

If the time required for a plurality of scans in which the number of times the light beam is scanned over the entire sample is divided is short, the moving speed within the required time is constant even if the moving speed of the sample is not uniform. Can also be considered constant. In such a case, a proportional relationship is established between the scanning line density and the count value of the pulsed component of the electric signal. For this reason, the scanning line density is calculated by dividing the scanning number measurement value in the above-mentioned plural times of scanning by the moving distance measurement value of the sample, and the reference scanning line density is added to the pulse component count value measured in the plural times of scanning. If the ratio of the scanning line density is multiplied, the measured pulse-like component count value is converted into the pulse-like component count value when the sample is scanned at the reference scan line density, that is, the actual number of abnormal points. Calculated. The number of abnormal points in the sample can be obtained by calculating the sum of these converted values calculated for each of the plurality of scans.

Since both the scanning line density and the count value of the pulse-like component are proportional to the moving speed of the sample, if the moving speed is constant, the effect of the difference in the speed value is the influence on the scanning line density and the count value of the pulse-like component. Since they act in the same way, their effects are canceled in the conversion operation. Therefore, the calculated number of abnormal points is not affected even if the moving speed of the sample is different or not constant.

〔Example〕

FIG. 1 is a block diagram of an embodiment of the present invention. Light projection means
A light beam that is narrowly focused is projected from 21. This light beam consists of a beam expander 22, a rotary polygon mirror 23, and f · θ.
The surface of the sample 5 is scanned by the light beam 6 which is focused on the sample 5 through the scanning means 25 surrounded by the dotted line constituted by the lens 24. The beam expander 22 arranges the incident thin light beam into a parallel light beam 26 having a width of about several mm, and the rotating polygon mirror 23 reflects the parallel light beam 26 to form a light beam 27 which rotates around its axis. The f.theta. lens 24 is a lens in which the rotational angular velocity of the incident light beam 27 and the scanning velocity of the emitted light beam 6 are in a proportional relationship, and the light beam 6 is always focused on the surface of the sample 5 and linearly moved at a constant period. To scan its surface in a circular pattern. When there is an irregular portion 7 such as unevenness or adhesion of foreign matter on the sample 5, pulsed scattered light 8 is generated there. The scattered light 8 is detected by the light detecting means 28 surrounded by the alternate long and short dash line.

In the light detecting means 28, the scattered light 8 is received by an incident light condensing system 29 having a cylindrical lens on the light incident surface, and the received scattered light 8 is incident on the optical fiber bundle 30 and is emitted by the outgoing light condensing system 31. It is incident on the photodetector 32 via the light detector 32 and gives a pulsed component to the electric signal 10 output from the photodetector 32.

Since the surface of sample 5 is mirror-finished, sample 5
The light beam 6 irradiating the surface of is incident on the incident light condensing system 29 so as to be specularly reflected at an angle not incident. For this reason, only the extremely weak scattered light from the surface of the sample 5 and the scattered light 8 due to the abnormal portion 7 are incident on the incident light condensing system 29 and thus the photodetector 32.

The sample 5 is mounted on a sample table 34 which constitutes a relative moving means 33 surrounded by a two-dot chain line, and the sample table 34 is driven by a sample table driving device 35 in a direction perpendicular to the scanning direction of the light beam 6. As a result, the light beam 6 can scan the entire surface of the sample 5 for inspection.

The electric signal 10 output from the light detection means 28 is applied to the signal processing means 36 surrounded by a solid line. The signal processing means 36 includes an area setting means 37, a pulse counting means 38, a scanning number counting means 39, and a computing means 40 connected to the moving distance measuring means 71 as shown in the figure.

As will be described later in detail, the region setting means 37 outputs a rectangular wave pulse having a time width for scanning the sample 5 and the inspection region thereof as the signal 55a and the signal 59a each time the light beam 6 scans the sample 5. The signal 59a is the pulse counting means 38.
The pulsed component of the electric signal 10 is counted by opening the counting gate for the time corresponding to the width of the rectangular wave pulse. On the other hand, the signal 55a as a rectangular wave pulse is given to the scanning number counting means, and the number of scanning times that the light beam scans the sample 5 is counted. In this embodiment, this signal 55a
Is also given to the moving distance measuring means 71. The moving distance measuring means 71 in this embodiment is composed of a time measuring circuit for measuring the width of a rectangular wave pulse as the signal 55a, and counts clock pulses for measuring time to measure the scanning time of the surface of the circular sample 5. . Since this scanning time is proportional to the scanning line length as the chord length of the circle forming the sample 5, the distance from the center of the circle with respect to the moving direction of the sample 5 is calculated by the calculating means after this. To get the distance traveled.

When one scan is completed, the pulse setting scan end signal 70a is given from the area setting means 37 to the calculating means 40. Computing means
The scanning number counting means 39 receives the scanning end signal 70a.
Is the number of scans j up to that scan, and the moving distance measuring means 71
From the pulse count means 38, the count value Nj of the pulse-like component counted in the scan from the pulse counting means 38, respectively, as data 39a, 74a, 68a.
Read and memorize as the reset signal 74b and 68b
Are sent to the moving distance measuring means 71 and the pulse counting means 38, respectively, and the counting contents are cleared.

When all the scans for the sample 5 are completed, the area setting means 37
From this, a pulsed inspection end signal 62a is given to the arithmetic means 40. The calculation means 40 receives this signal and executes a calculation, and sends a reset signal 39b to the scanning number counting means 39 to clear the count content.

In the calculation means 40, the reference scanning density n ₀ obtained by calibration in advance, the distance d from the edge of the sample 5 to the inspection region, the diameter D of the sample, and the total number of scans S are equally divided into a plurality of number of scans. The number m, the period τ _C of the scanning time measuring clock pulse, and the scanning speed v are stored as constants.
The following calculation is performed using these constants, the total number of scans S, the number of scans j described above, the count value Xj of the clock pulse for measuring the scan time, and the count value Nj of the pulse component, and the number of abnormal points N _{0 in the} inspection area is calculated. To calculate.

1) The number of scans from the edge of the sample 5 to the inspection area S ₁ S ₁ = S (d / D) (1) 2) The number of scans S _C S _c = (S−2S ₁ ) / m (2) 3) The first number of scans S (i
1) and the number of scans at the end S (i2) 4) Count value Ni of the pulse-shaped component in the i-th multiple scans 5) Distances Y _(i1) and Y _{S (i2)} of the scanning line from the end of the disk-shaped sample 5 at the number of scans S (i1) and _{S (i2)} 6) Moving distance of the sample 5 in the i-th multiple times of scanning
l _i l _i = | Y _{S (i1)} −Y _(i2) (6) (i = 1, 2, ..., m) 7) Number of abnormal points in the inspection area N ₀ The abnormal point number N ₀ calculated by the equation (7) is output as the inspection result 46.

The grounds for establishing the above equations will be described later.

In the present invention, the calculation means 40 performs the calculation according to the equation (7) to eliminate the influence of the moving speed of the sample being different or changing during the inspection.

According to this equation (7), even if the moving speed of the sample 5 varies, the moving speed can be regarded as constant within a short time, and if the moving speed is constant, one abnormal portion 7 is scanned. The number of times is derived based on the idea that the scanning line density is proportional to the scanning line density and thus the count value of the pulse-like component is stochastically proportional to the scanning line density.

That is, the total number of scans of the inspection area is divided into m equal parts, and
The time required for Sc scans is short. Count value of pulse-like component in the i-th Sc times of scanning
Assuming that Ni and the moving distance l _{i of the} sample are obtained, the scanning line density ni is
Since ni = Sc / l _i , Ni is proportional to Ni and ni Given in. k is a constant. In equation (8), the scanning line density when the count value Ni of the pulsed component is equal to the actual number of abnormal points N _0i is the reference scanning line density n ₀ , so N _0i = kn ₀ ………………………… ………… (9). Eliminating k from both equations (8) and (9), N _0i = (n ₀ / n ₁ ) N _i = (n ₀ / S _c ) N _i l _i ………………………… (10 ) Get The equation (10) converts the count value Ni obtained in the case of the scanning line density n _i = S _c / l _i into the value in the case of the reference scanning line density n _i . Since Ni and n _i are in a proportional relationship and the moving speed of sample 5 acts on both in the same way, Ni and n _i have a numerator and denominator relationship, respectively. The value of N _0i that is canceled out is not affected by the moving speed of the sample. The relationship given by Eq. (7) is derived by taking the sum of i in N _0i in Eq. (10).

FIG. 2 shows that the proportional relationship shown in Eq. (8) holds.
It is the result confirmed experimentally using the apparatus shown in the figure.
Here detection rate is the ratio between the pulse-like component count N real anomaly number N ₀ this N as corresponding values to eta = N / N ₀
Is used. Sample 5 was a silicon wafer with a diameter of 10 cm, and 0.518 μm polystyrene latex spheres directly connected to the surface were uniformly dispersed to make it an abnormal point. The actual number N ₀ was visually counted with a microscope and shown in FIG. The detection rate η was obtained by dividing the count value N obtained by keeping the moving speed of the sample constant with the above apparatus by the above N ₀ . On the other hand, by varying the moving speed of the sample, the total number of scans S for the silicon wafer is obtained each time, the scanning line density n is derived from this, and the abscissa indicates n and S, and the ordinate indicates η to obtain the relationship between them. It was It has been shown that a linear relationship, that is, a proportional relationship, passing through the origin is clearly established between η and n or η and S. The regression line obtained for Fig. 2 is for 1000 or more scans Where η = 100%, the standard scan line density n ₀ is
₀ = 223.8 / cm or to obtain the S ₀ = 2238. This value is obtained by actually measuring the diameter r of the spot formed by the light beam 6 on the sample surface, and dividing the value by this r by the diameter D of the silicon wafer, that is, D / r, that is, the light beam does not overlap the sample surface,
In addition, the values are the same as those in the case of scanning without a gap.

FIG. 3 shows the surface of the sample 5 and one of the scanning lines by the light beam 6 for scanning the surface of the sample 5 and the photodetector 32 obtained by the scanning, as a premise for explaining the specific constitution of the measuring means 36 in this embodiment. 3 shows the waveform of the electric signal 10. FIG. 3 (a) is a plan view of the sample 5 on the sample table 34,
The light beam 6 is orthogonal to the moving direction Q of the sample 5 indicated by the arrow.
The scanning line 41 is formed by.

The peripheral region 42, which is shaded from the edge of the sample 5 to the distance d, is a region that the holder contacts when handling the sample 5, and is a region that is not an object of inspection because a large number of foreign substances are expected to adhere. Therefore, the area not shaded becomes the inspection area 43. The light beam 6 is projected onto a point O on the sample stage 34 to start scanning along the scanning line 41, scans the surface of the sample 5 and then irradiates the sample stage 34 again, and finishes scanning at point 0 ′. The fine particles 701, 702, and 703 attached to the surface of the sample 5 all correspond to the abnormal portion 7 in FIG. The distance ▲ ▼ 'between the point O and the point 0'is invariable in any scan.

In the waveform of the electric signal 10 shown in FIG. 3 (b), the scanning repetition cycle is a constant time T ₀ between time t ₀₀ and time t _01.
Therefore, the scanning of the light beam is performed at the constant time T between the times t ₀ and t ₆ . Since the scanning speed is constant, the elapsed time t in the horizontal axis direction corresponds to the scanning distance. Therefore, the position on the scanning line shown in FIG. 3 (a) is shown in parentheses for each time.

Between times t ₀₀ and t ₀ , the light beam 6 is not projected first, so that the level of the electrical signal 10 is at the noise level 44 determined by the noise signal of the photodetector 32 or its amplifier. Time t ₀
Then, the light beam 6 starts to be projected on the surface of the sample table 34. The surface of the sample stage 34 produces relatively strong scattered light, so that the level of the electrical signal 10 rises to the level V ₁ . When the light beam 6 comes to be projected on the surface of the sample 5 at time t ₁ , the scattered light sharply decreases and the level of the electric signal 10 lowers to the level V ₂ accordingly. Furthermore, at times t ₂ , t ₃ ,
When the light beam 6 irradiates the particles 701, 702, and 703 at t ₄ , pulse-like scattered light is generated each time, and pulses b ₁ , b ₂ , and b ₃ are generated in the electric signal 10, but the level of these scattered light is The level is weaker than the level of scattered light from the surface of the sample table 34, and therefore the level Vp of the peak value of the pulses b ₁ , b ₂ , and b _3.
Is less than V ₁ .

As described above, the electric signal 10 shows a decrease in level in the time zone T ₁ in which the light beam 6 scans the surface of the sample 5, and using this, the electric signal 10 is used at a predetermined level slightly lower than the level V _1.
V _s1 as 45 is determined in advance, and the scanning time T
At time t ₁ when the electrical signal 10 falls below the level V _s1 between the two edges of the sample and again from that state
The other edge is detected at time t ₅ above V _S1, and the surface of sample 5 is scanned at time T ₁ between times t ₁ and t ₅ . Further, when the time for scanning the distance d between P ₁ X ₁ or P ₂ X _{2 in} FIG. 3 (a) is τ, time t ₁ + τ and time t ₅ −τ =
T ₁ -2τ between t ₁ + T ₁ -τ is the scanning time of the inspection area.

On the other hand, with respect to the moving direction of the sample 5, since the number of scans and the moving distance of the sample are in a proportional relationship, diagonal lines are drawn from the edges P ₃ , P _{4 of} the sample 5 detected by the light beam 6 to the distance d respectively. Area Assuming the total number of scans S ₁ by the light beam 6 as S and the diameter of the sample 5 as D, the following re-expression (1) S ₁ = S (d / D) ……………… …………… given by (1). Therefore, the inspection area is scanned from the S ₁ + 1st time to the S−S _1st time, and the number of scans during that time is S−2S _1st time. By dividing this by m, Sc shown in equation (2) is determined.

In this embodiment, a silicon wafer which is usually circular is used as the sample 5. For silicon wafers, which are usually 5 to 10 cm in diameter, it is desirable to have a moving speed that results in a total number of scans in the range of 500 to 2000, and Sc is selected so that m is about 10.

The equation (3) showing the initial number of scans S (i1) and the last number of scans S (i2) of the i-th plural number of scans and the equation (4) showing the count value of the pulse-like component in the scan are It is self-explanatory and needless to say.

Since the silicon wafer as the sample 5 described above is an almost accurate circle, the length of the sample surface of the scanning line 41 in FIG. Is the chord length of a circle of diameter D forming sample 5, and the distance Y from the center C of this circle in the moving direction of sample 5 is Given in.

Therefore, the number of scans S (i1) and S (i2) of the first and last scans of the i-th multiple scans The length is measured to obtain Y _{s (i1)} and Y _{s (i2),} and the moving distance l _i of the sample 5 in the i-th plural number of scans is obtained from these differences as shown in the equation (6). You can

In this example Is determined by measuring the time T ₁ during which the electrical signal 10 shown in FIG. 3 is below a predetermined level V _s1 . If the count by the clock pulse of the period τ _c for time measurement is X and the scanning speed of the light beam is v And Y _{s (i1)} shown in _Eqs . (12) to (13) and (5)
Y _{s (i2)} is derived.

Note (5), the value of (6) in the calculation of X _{s (i2)} is successively compared with the value of X _{s (i2)} and _{D / (vτ c), X} s (i2) ≒ X
Y when _{s (i1)} or X _{s (i2)} ≒ D / (vτ _c )
_{s (i2)} is treated as a negative value.

FIG. 4 is a block diagram showing an example of the circuit configuration of the signal processing means 36. FIG. 5 is a waveform diagram of signals in each part of the circuit configuration of FIG. The operation of each part of FIG. 4 will be described below with reference to the waveform chart of FIG.

In FIG. 4, the electric signal 10 output from the photodetector 32 is the signal 51 from which white noise has been removed by the low-pass filter 51.
a, which is given to the area setting means 37 and the pulse counting means 38, which are surrounded by broken line frames. In the area setting means 37, first, by the system including the comparator 52, the monostable multivibrator 53, the inverter 54, and the AND gate 55, the time equal to the time zone T ₁ during which the light beam 6 scans the surface of the sample 5 A rectangular wave signal 55a having a width is obtained.

In this system, the level of the signal 51a is compared with a predetermined level V _s1 by the comparator 52, and when V> V _s1, a high level “H”, and when V <V _s1 , a low level “L” output signal. Get 52a. The monostable multivibrator 53 receives the signal 52a.
Is at a high level "H", it is triggered to output a rectangular wave having a time width equal to the scanning time T as a signal 53a. On the other hand, the inverter 54 inverts the polarity of the signal 52a into a signal 54a, supplies these signals 53a and 54a to the AND gate 55, and outputs a rectangular wave having a time width T ₁ as a signal 55a. If the light beam 6 does not scan the sample 5, T ₁ = 0 and the signal 55
a remains at low level "L".

Since the signal 55a is output as a rectangular wave each time of scanning, the scanning circuit counting means 39 which is a pulse counting circuit counts this as a logical pulse to obtain the number of scanning.

The signal 55a is also the AN of the moving distance measuring means 71 surrounded by a broken line.
It is provided as one input of D gate 73.

The moving distance measuring means 71 is configured as a time measuring circuit, and the clock pulse signal 72a generated by the oscillator 72 is given to the other input of the AND gate 73. Therefore, the pulse counting circuit 74 counts the clock pulse number X corresponding to the time corresponding to the width of the rectangular wave forming the signal 55a, that is, the time required for the light beam 6 to scan the surface of the sample 5. The count value is calculated as data 74a by the calculating means 40.
The moving distance of the sample is derived by the procedure described above and described above.

The signal 55a is also given to the inspection area setting system 56 and the sample detection system 57 surrounded by the alternate long and short dash line in the area setting means 37.

In the inspection area setting system 56, a rectangular wave signal 59a having a time width of the scanning time T ₁ -2τ of the inspection area 43 in the scanning direction of the rising light beam 6 is output each time 2τ after the time t _1. . The delay circuit 58 gives a signal 55a which is a rectangular wave having a time width T _{1 by} a predetermined time delay τ ₁ = 2τ, and outputs the signal 58a.
and a, the AND gate 59 given a signal 55a of the and the signal 58a outputs a rectangular wave of duration T ₁ -τ ₁ = T ₁ -2τ as signal 59a. When the light beam does not scan the sample 5 and the signal 55a is at the low level "L", the signal 59a also remains at the low level "L".

The signal 59a is supplied to the pulse counting means 38, and serves a gate signal for opening the T ₁ -2τ time counting gate inspection area 43 is scanned.

The signal 59a is also given to the inverter 69 to invert the polarity, and becomes a signal 69a which rises at time t ₅ when the light beam 6 finishes scanning the inspection area 43 of the sample 5 and triggers the monostable multivibrator 70. The pulse-shaped scanning end signal 70a to the arithmetic means 40.

The sample detection system 57 outputs a pulse signal 62a for starting the calculation of the calculation means 40 when the scanning of the sample 5 by the light beam 6 is completed.

The monostable multivibrator 60 is triggered by the signal 55a and outputs a rectangular wave having a time width T ₂ longer than the scanning period T ₀ , that is, T ₂ > T ₀ , as the signal 60a. Since this monostable multivibrator 60 is provided with a function capable of retriggering, when the light beam 6 repeatedly scans the sample 5, it is retriggered before the falling edge of the rectangular wave of the signal 60a each time scanning starts. Continuing, signal 60a while scanning is being performed
Keeps the high level "H". This signal 60a is converted into a signal 61a whose polarity is inverted via an inverter 61. Therefore, the signal 61a becomes a signal that maintains a low level "L" while the light beam 6 scans the sample 5 and shifts to a high level "H" when the scanning is completed. Therefore, when the scanning is completed, this signal 61a triggers the monostable multivibrator 62 and gives a pulse-shaped inspection completion signal 62a to the arithmetic means 40.

On the other hand, in the pulse counting means 38, first, the signal 51a and the signal 63a obtained by applying the signal 51a to the low-pass filter 63 and removing the pulse component are compared by the comparator 64 to obtain the signal 51a.
And a signal 63a is output as a signal 64a.

In the example shown in FIG. 5, the signal 64a has a pulse component b ₁ ,
This signal 64a including only b ₂ and b ₃ is further compared with a relatively low level threshold level V _{s2 by the} comparator 65, and if the _peak value V _p of the pulse component is V _p > V _s2 , it is rectangular. The wave pulse signal 65a is output.

The signal 65a is given a delay .tau. For a second predetermined time by the delay circuit 66, and enters the AND gate 67 as the signal 66a. AND
The gate 67 receives the signal 59a provided at the same time, so that the time t ₁ +
It operates as a counting gate opened for T ₁ -2τ time from 2τ to t ₁ + T ₁ and pulses a signal 67a corresponding to the portion of the signal 65a obtained by scanning in the inspection region 43 as described below. It is given to the counting circuit 68. As a result, only the pulse signals (b ₂ ) obtained by the fine particles 702 in the inspection area 43 are counted.

Here, the relationship between the signal 65a and the signal 67a will be described.
Letting t be the elapsed time from time t ₁ , and representing signals 65a and 66a as f (t) and g (t), respectively, g (t) = f (t−τ) ……………………………… (7). The signal 67a is brought by the AND gate 67 to the signal 66a, that is, between t = τ ₁ = 2τ and t = T _{1 of} g (t). According to equation (7), g (τ ₁ ) = f (τ ₁ −τ) = f (2τ−τ) = f (τ) g (T ₁ ) = f (T ₁ −τ) ………………… (8), the signal 67a is t = τ and t = T ₁ −τ of the signal 65a.
And a portion obtained from the inspection area 43.

The scanning number counting means 39, the counting circuit 68, and the data 39a, 68a, 74a and the reset signals 39b, 68b, 74b between the counting circuit 74 and the computing means 40 have already been described in FIG.

In this embodiment, the number of scans over the entire inspection region is equally divided into m to obtain a plurality of number of scans S _c , but it is not always necessary to equally divide the number of scans, and different number of scans Sci (i = 1). , 2, ..., M) may be simply divided into m parts.

Further, the moving distance l _{i of the} sample 5 is also an electric signal as in this embodiment.
Needless to say, it may be measured by a moving speed measuring means of a method of directly measuring the moving distance or moving speed of the sample 5 instead of the method of processing 10. In this case, a sensor for directly measuring the moving distance or moving speed of the sample 5 is provided as the moving distance measuring means in this inspection device, and those sensors are provided on the relative moving means side so that the output is connected to the calculating means. You may comprise. For example, bar codes at fixed intervals are attached to the surface of the sample table 34 shown in FIG. 1 along the moving direction of the sample 5, and the bar code, the optical sensor for reading the bar code, and the counting circuit for counting the output pulses can be simply used. A moving distance measuring means can be configured.

Such moving distance measuring means is required when the sample 5 has a shape in which it is difficult to calculate the moving distance from the length of the scanning line scanning the surface thereof.

〔The invention's effect〕

According to the present invention, the surface of the mirror-like sample is scanned with the light beam, the pulsed scattered light generated at the abnormal portion is received by the photodetector, and the pulse-shaped component of the output signal of the photodetector is processed. In a surface inspection apparatus for inspecting the number of abnormal points by counting with a pulse counting circuit in the means, the signal processing means includes a scanning number counting means, a sample moving distance measuring means, and a computing means for connecting the pulse counting circuit, The number of scans, the moving distance of the sample, and the number of pulsed components of the electric signal are measured for each of a plurality of scans obtained by dividing the number of scans over the entire region into several, and the measured number of scans and the moving distance of the sample The scanning line density in multiple scans is calculated, and by using the scanning line density value and the reference scanning line density value, the measured pulse component count value is converted into a value for the reference scanning line density. Times And to obtain calculated by the number of anomaly the sum of these converted values obtained for each scan. The reference scan line density is the scan line density when the count value of the above pulse-shaped component becomes equal to the actual number of abnormal points, and the time required for each multiple scans is short, and the moving speed of the sample during that period. Therefore, the scanning line density can be regarded as constant even when the moving speed of the sample is not uniform. Therefore, even if the moving speed of the sample is different and the number of abnormal points obtained by the above conversion and summation varies, it is possible to obtain the actual number of abnormal points regardless of these.

Therefore, it is not necessary to integrate the light projecting means, the scanning means, the light detecting means, and the signal processing means with the relative moving means in which the moving speed of the sample is controlled to be constant as in the conventional case, and the apparatus is downsized and the cost is reduced. Can be realized. Therefore, the moving speed of the sample is different from each other, and it can be properly installed in a necessary place even in an existing manufacturing line in which the moving speed is not uniform. For example, it is possible to apply a wide range of applications such as in-line inspection of the state of particle contamination of silicon wafers and monitoring of dust generation inside the semiconductor manufacturing equipment by installing it in the loader and unloader of the semiconductor manufacturing equipment to the semiconductor manufacturing line.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the present invention, and FIG. 2 shows the proportional relationship between the count value of abnormal points and the scanning line density as a detection rate which is the ratio between the count value of abnormal points and the actual number of abnormal points. Graph showing the results demonstrated by the proportional relationship with scanning line density, Fig. 3 (a), (b)
Is a plan view showing the sample and one scanning line and a waveform diagram of electric signals in the scanning, FIG. 4 is a block diagram showing a circuit configuration of the measuring means, and FIG. 5 is a signal in each part of the circuit configuration of FIG. FIG. 6, FIG. 6 and FIG. 7 are main-part configuration diagrams showing an example of an optical inspection device in the prior art. 1, 21 ... Light projection means, 5 ... Sample, 6 ... Light beam,
7 ... Abnormal part, 701, 702, 703 ... Fine particles, 8 ... Scattered light, 10 ... Electrical signal, 25 ... Scanning means, 33 ... Relative moving means, 34 ... Sample stage, 35 ... Sample stage drive, 16,
36 ... Signal processing means, 38 ... pulse counting means, 39 ... scanning number counting means, 40 ... computing means, 43 ... inspection area, 6
8,74 …… Pulse counting circuit, 71 …… Movement distance measuring means.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Keisuke Sugimoto 1-1, Tanabe Nitta, Kawasaki-ku, Kawasaki-shi, Kanagawa Fuji Electric Co., Ltd. (72) Inventor Yasuaki Namba, Tanabe Nitta, Kawasaki-ku, Kawasaki-shi, Kanagawa No. 1 inside Fuji Electric Co., Ltd.

Claims (4)

[Claims] 1. A light projecting means for projecting a light beam, a scanning means for linearly and repeatedly scanning a specular surface of a sample surface with the light beam, and a scanning means for scanning the sample and the sample. Relative moving means for relatively moving in a direction intersecting with the direction, light detecting means for detecting scattered light generated on the sample surface of the light beam and a region other than the sample surface, and output of the light detecting means A signal that processes the electrical signal and counts the pulsed component based on the scattered light generated in the abnormal area in the predetermined inspection area set on the sample surface in the pulse counting circuit and outputs the counting result In the above-mentioned apparatus for inspecting the surface of a sample provided with a processing means, the signal processing means connects the light beam scanning frequency counting means, the sample moving distance measuring means, and the pulse counting circuit. Means for measuring the number of scans, the moving distance of the sample and the pulse-like component count value of the electric signal for each of a plurality of scans obtained by dividing the number of scans of the light beam over the entire inspection region, and the measured number of scans and the sample The scanning line density is calculated based on the moving distance of the scanning line for multiple scans, and the reference scan that gives the measured value of the number of pulse-like components equal to the known number of abnormal points, which was previously obtained by the reference sample with the known number of abnormal points. The measured pulse-like component count value was converted to a value for the reference scanning line density from the linear density and the calculated scanning line density, and the number of abnormal points was calculated for each of a plurality of times of scanning. An optical surface inspection device characterized by being calculated as a sum of values. 2. The apparatus according to claim 1, wherein the scanning number counting means comprises a rectangular wave counting circuit forming a part of the signal processing means, and light detection is performed when the light beam scans the sample surface. An optical surface inspection apparatus characterized by counting a rectangular wave formed when an electric signal output from the means falls below a predetermined level. 3. An optical surface inspection apparatus according to claim 1 or 2, wherein the sample is disk-shaped. 4. The apparatus according to claim 3, wherein the sample moving distance measuring means comprises a time measuring circuit forming a part of the signal processing means and an arithmetic means connected to the time measuring circuit.
The time measuring circuit causes the electric signal output from the photodetector to fall below a predetermined level by scanning the surface of the sample with the light beam in each of the first scanning and the last scanning of the plural scannings of the light beam. The measuring means multiplies the measured time by the scanning speed of the light beam as a known amount to calculate the length of the scanning line on the sample surface in the first scanning and the final scanning. The distance between each of the two scanning lines and the center of the circle is calculated from the lengths of these two scanning lines and the diameter of the circle of the disk-shaped sample as a known amount, and these are calculated. An optical surface inspection apparatus characterized in that a moving distance of a sample is obtained as a difference between different distances.