JPS6026174B2 - Defect detection device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an automatic detection device for defects on face plate-like samples such as silicon wafers and glass face plates.The various conventional methods of detecting defects on common face plate samples are outlined below. That's right.

The first method is to visually examine defects within a field of view determined by the magnification using an optical microscope. This method includes
'1) It takes a long time to examine the entire sample surface, '2) There are oversights and the inspection is not complete, and [3] Detection factors are necessary, and the reliability of the inspection is affected by the fatigue of these personnel. decline,
Such shortcomings are essentially subordinate. Second, there is a method of combining an optical microscope and an ITV camera to monitor the field of view determined by the magnification of the microscope using the output signal of the ITV camera. This method eliminates the need for human intervention in the first method. The advantage of this method is {1
1. Multiple people can observe at the same time, reducing inspection errors; 2. By processing the ITV output signal, it can be output in the form of a grain distribution; 3. Various images can be displayed on the monitor TV. For example, it is possible to display images that have been processed. However, since an optical microscope is used, drawback 1 of the first method is not solved. In this method, practical equipment that performs advanced signal processing is commercially available, but it is currently expensive. Thirdly, there is a method of detecting defects by irradiating a light beam onto the sample surface and capturing the light scattered by the defective portion.

There are several types of this method, and the principle block diagram of the first type is shown below. A laser is used as a light source (not shown), and the direction of the beam 1 is changed by a prism 2. This beam is
reach the top. Scattered light 4 from the defect is collected by a converging lens 6 via a double-sphere condenser lens 5 into a photoelectric conversion element 7 and converted into an electric signal. In this configuration, the sample is placed on the plane of the double-sphere condenser lens 5, and a field stop (integrated with the mount of the photoelectric conversion element) is placed just in front of the photoelectric conversion element. This is done by installing it on a rotating body 8 and rotating it, and by translating the whole of 2, 4 to 7. Figures 2 and 3 are examples of signals obtained by this method. Figure 2 shows the integration of the total scattered light during spiral scanning over the entire surface of the sample, when there are many defects, and Figure 3 when the sample is clean and non-defective. The structure and structure give clues as to the surface condition of the sample.As long as the content described above is maintained, this method can only determine the presence or absence of defects on the surface of the sample to be inspected by comparing it with a standard product. do not have.

Next, FIG. 4 is a diagram showing the principle configuration of a second type of method for capturing scattered light.

A laser 9 is used as a light source, the beam is expanded to a diameter of about one needle by a beam expander (laser body), this beam 10 is passed through a mirror 11 with a hole in the center, and a galvanometer mirror 12 is used to expand the beam at a right angle. It is bent vertically downward, and the convergent lens 13 focuses the irradiation onto the sample surface 3 to 15 peaks.
With this arrangement, if there is no dust or other defects on the sample surface, reflected light with an intensity determined by the reflectance on the sample surface will reverse the irradiated path and return in the m-za direction. Next, if there is dirt or other defects on the sample surface, the illuminating light will be scattered by these. This scattered light 14 enters the diameter of the converging lens occupied by the incident light and its surrounding area, follows the irradiation path in the opposite direction, is reflected by the mirror 11 with a hole in the center, is converged by the lens 15, and is transmitted to the photoelectric conversion element 1. Convert to electrical signal. As can be seen from the above explanation, the lens 15 receives and converges only the scattered light 14 due to dust and other defects. In this method, the scanning galvanometer driver 17 vibrates the galvanometer mirror to deflect the laser beam, and
This is done by moving the beam 8 perpendicular to the beam deflection direction.

The advantages of this method include being able to count the total number of tl-order defects, and being able to display the location of defects as a map on the 2' oscilloscope.

Further, the size of the smallest defect that can be displayed is one mountain surface. However, the current state of this method does not satisfy the need to know the particle size distribution of defects (see DR.
○svald, et al. ;;A bad serTec
hnique for Electronic Mat
erials SmfaceEvaluation, J
．． Electronics 3 [1], 225-242
(74). In addition to the prior art described above, techniques for detecting and measuring various defects on the surface of sheet specimens, such as rolled steel plates, have been developed.

Since each of these different inspection objects requires a unique measurement technique, the points raised in the prior art regarding sheet-like samples will be omitted here. An object of the present invention is to provide a defect detection device for solving the various problems pointed out above.

The present invention will be explained in detail below using examples.

FIG. 5 is a diagram explaining the principle of the present invention. The light source is Caesar (H
e-Ne)9. The laser beam 1 is focused by a lens 19 to a beam of several tens of inches at a time, passes through a rear prism 20, then passes through a mirror 21, 1/Jakeskin Nagasaka 24, and is focused by a rear lens 22 to a diameter of about 1 Buddha. The surface of the sample 3 is irradiated.The sample is held on the rotating mechanism 8 by a vacuum chuck system and rotated.Scanning of the sample is performed by this rotation and parallel movement of the laser beam spot on the sample surface. Therefore, the scanning state on the sample surface is spiral.
Under this situation, when the beam spot passes through the defective part, there will be a difference in the amount of light returning in the direction of the irradiation path due to phenomena such as diffraction, absorption, refraction, and scattering in the defective part. 1' There is always a component that traces in the opposite direction of the nail skin long plate 24 and the mirror 21. This retrograde light is Prism 2
The light is bent at a right angle by the river, reaches a photodetector 23, and is converted into an electrical signal. Naturally, there is a retrograde component that returns to the photodetector even while the beam spot passes through the defect-free area. The power at this time is P. , the beam passing through the defective portion is Pd, and the power of the beam spot on the sample surface is P. At this time, Ri.

=P. /P, [11rid=PrenoP[ {21
Defining this, the condition of the defect depends on d≦. becomes.
When there is an equal sign, it is not possible to determine the defect, but this rarely happens. If it's Rikuri. When , it's a sword. This becomes a negative signal with . And conversely, rid〉
the law of nature. When , it's a sword. A positive pulse signal is obtained with the reference level as . In the above configuration, the laser is linearly polarized, and the prism has a high transmittance (low reflectance) for P-polarized light centered on the cemented surface, and a high reflectance (low transmittance) for S-polarized light. using something.

1/The long plate was used so that P polarized light became S polarized on the left and right sides of the bonding surface.

Here, the differences in the optical systems in FIGS. 1, 4, and 5 will be compared to clarify the characteristics of FIG. 5.

[1) Laser beam: In Figure 1, the laser beam is not focused.

In FIGS. 4 and 5, the laser beam is narrowed down. The former is 15〃m, the latter is 1.6 m.'21 Irradiation power intensity; What if the beam diameter is 1 willow? If the same laser is used, the irradiation power intensity will be 1 in the case of FIG. 1, 4.4×1 in FIG. 4, and 3.9×1 in FIG.

'31 Effective convergence solid angle (c) of signal light; as can be seen by comparing Figures 1, 4, and 5, ■
Generally, c is largest in the case of FIG. 5, followed by FIG. 1 and then FIG. 4.

'4} S/N ratio: The S/N is in the order of FIG. 5, FIG. 4, and FIG. 1.

{51 Minimum detectable defect size; 10 peaks in FIG. 1, 1 peak in FIG. 4, and 0.5 r in FIG. 5.

Combining the above, it can be seen that the method shown in FIG. 5 is more advantageous than the methods shown in FIGS. 1 and 4.

Although the above explanation is sufficient in principle, there are problems associated with rotating the sample for scanning.

This is due to eccentric movement of the rotating body, waviness of the sample, etc.
At the beam spot irradiation position on the sample surface, the apparent
This means that vertical displacement is occurring. Although it is necessary to consider this vertical displacement in consideration of the depth of focus of the lens 22, a kind of general deviation occurs, and an error occurs in the electric signal generated by the photodetector 23. To solve this problem, it is necessary to always maintain the best focus state by moving the lens 22 up and down in response to slight vertical fluctuations of the sample. The part that achieves this purpose is called the automatic focus tracking part. In this part, a means for detecting the vertical displacement of the sample is important, and FIG. 6 shows the configuration and principle of the optical method. The laser beam is split by beam splitter 25 into beams 1-1 and 1-2, the former corresponding to beam 1 in FIG.

1-2 is a mirror 26, a lens 27, a prism 28, a mirror 29,
The light passes through the half mirror 30 and follows the same optical path as the optical path shown in FIG.

At this time, they are arranged offset from the optical axis of the lens 22. Therefore, the sample is obliquely irradiated by the lens 22. The reflected light from the sample surface also becomes obliquely incident on the lens 22, travels backward along the irradiation path, and is bent at a right angle by the prism 28. The number of beams at this time is 32. If the sample 3 is slightly displaced downward with this arrangement, the point of incidence on the lens 22 will be different, and this will move backwards, causing the prism 28
When the beam is reflected by the beam 33, it becomes a beam 33. Further, if the sample 3 is displaced upward, it becomes a beam 34. If a photoelectric conversion element 31 having characteristics as shown in FIG. 7 is placed at this position with respect to the irradiation position of the beam spot, it becomes possible to detect slight vertical displacement of the sample. This photoelectric conversion element is used as an optical position detector. FIG. 8 is a block diagram of an electric circuit for moving the lens 22 using the signal obtained by the optical position detector.

35 is a resistor for supplying a bias current to the optical position detector 31.

The output of 31 is amplified by a differential amplifier 36, and a main amplifier 37,
The signal is input to the booster 39 via the phase compensation circuit 38.

A coil integrated with the lens 22 is placed in a DC magnetic field (described later). By supplying current from the booster 39 to this coil, the lens 22 is moved up and down. At this time, the lens 22 is moved in the direction of displacement of the sample 3. Automatically combine Figures 6 and 8. It forms a focus following section. Details of the vertically movable lenses 22, including the coils, are shown in FIG.

This part is called a movable lens. The four coils are wound around a metal bobbin 41. A permanent magnet 42 is used for the DC magnetic field, and a coil 40 is placed in the gap between the yokes 43 and 44. The lens 22 is integrated with the bobbin 41, and the whole is suspended in the air by springs 45-1 and 45-2, and the lens 22 moves up and down depending on the direction of the current flowing through the coil. In order to hold these in their entirety, they are fixed with a podium 46. The rotation mechanism 8 uses an air bearing and is directly connected to the rotor of a motor as a drive source.

The reason why ordinary ball bearings are not used is to minimize eccentric rotation as much as possible. Furthermore, in order to move the beam spot, a mechanism is used in which the entire table incorporating the optical system is moved in parallel at a desired speed. For example, the first
As shown in FIG. 5, the irradiation optical system shown in FIG. 5 and the automatic focus tracking section 66 shown in FIG. This moves the beam spot on the sample surface. Next, the defect detection electric signal converted by the photodetector 23 is processed as follows.

Depending on the purpose of use of the defect detection device, the following cases can be considered. [1-Teach the total number of defects. ■Categorize defects by size. [3' Know the location of the defect. Processing methods for these cases will be described below. FIG. 10 is a block diagram of a signal processing circuit for determining the total number of defects.

The signal pulse detected by the photodetector 23 is transmitted to the front layer amplifier 4
7 is amplified. This signal is converted into a digital pulse by a pulse shaping circuit 48 and then inputted to a counter 49, and the output thereof is displayed on a digital display device 50. In addition, when distinguishing between whether to send the specimen to production or whether it is an apricot by the number of defects on the face plate, a division determination device 51 is provided that uses the output of a counter to determine whether the number is greater than or less than a specified number, and , a display unit 5 that displays whether or not the subject can be passed through the output.
2, and a terminal 53 for control signals to the production line. FIG. 11 is a block diagram of a signal processing circuit for distinguishing the size of a defect.

The photodetector 23 was omitted. Next to the front layer amplifier 47, a waveform shaping circuit 54 converts the pulse into a pulse having a width depending on the size of the defect. The output is converted by a pulse width peak value conversion circuit 55 into a pulse having a peak value depending on the size of the defect. FIG. 12 shows an example of a specific method of the pulse width peak value conversion circuit.

55-1 is an integrator with built-in set and reset switches.

Reference numeral 55-2 is an impedance converter for creating a signal for turning on and off the set switch of the integrator, and it can also serve as an inverter if necessary.

55-3 is a monostape multivibrator which is activated at the rising edge of the input signal pulse 55-5, and the pulse width formed has a predetermined Tp.

The output pulses 55-6 reset the integrator set switch. Therefore, the integrator time is the same as the input pulse width. 55-4 is a peak detector that detects the peak of the output of the integrator.

It has a built-in reset switch, which is reset by the output pulse of the monostabilized multipipulator.
Therefore, the retention time of the peak value in the peak detector is Tp
becomes. As is clear from the above description of the configuration, the width of the input pulse arriving at the integrator is integrated, and therefore the value after the integration is proportional to the width of the input pulse. This value is detected by the peak detector and simultaneously converted into a pulse voltage having a width of Tp, so that the initial target pulse width can be converted into a peak value. The signal pulses thus obtained are separated into desired channels and then digitally displayed, or the size of the defect is analyzed in detail by a pulse height analyzer 56.

Next, two methods will be shown below to find out the location of the defect.

FIG. 13 is a block diagram showing the first method. Illustration of the photodetector 23 is omitted. Since scanning is a spiral (polar coordinates), the idea is to convert this to rectangular coordinates and monitor it on TV. A signal generator 57 is provided to simulate the polar coordinate component r,8 which changes in synchronization with the rotation of the rotating mechanism 8. The output signals r and 8 are converted to polar coordinates and rectangular coordinates converter 5.
8 into Cartesian coordinate signals x and Y. These ×, Y and the output of the front layer amplifier are converted to a scan converter 59.
Enter. The output is monitored on TV60. 1st
FIG. 4 is a block diagram showing the second method. First, the polar coordinate components r, 8 are converted into rectangular coordinate components x, Y in the same manner as described above. Here, the x and Y components are separately digitized by A/D converters 61-1 and 61-2, and input into the address register 62. The output of the address register (address signal) is the memory 63
Enter. On the other hand, the output signal of the previous layer amplifier is pulsed by the comparator 64, and this is input into the memory 63 at the same time as the address signal. In order to return the output of the memory 63 to an analog signal, it is sent to the TV via the D/A converter 65.
input and monitor. Furthermore, it is also possible to adopt a configuration that includes all three types of signal processing methods described above. As described in detail above, the present invention uses optical means to detect dust and defects on the sample surface on the face plate, and also uses optical means to realize the best focus state of the beam spot. has an automatic tracking section, scans by rotating the sample and moving the beam spot in parallel, and
This is a defect detection device that has a signal processing section that is provided depending on the purpose of use of the device, and has the characteristics of being able to perform inspections in a shorter time than conventional methods and devices, and improving inspection accuracy.

[Brief explanation of the drawing]

FIG. 1 is an explanatory diagram of the main parts of the conventional device, FIGS. 2 and 3 are examples of signals obtained by the method of FIG. 1, FIG. 4 is an explanatory diagram of the second conventional method, and FIG. The figure is a diagram of the principle of the present invention, Figure 6 is an explanatory diagram of the optical system configuration and principle of the automatic tracking section of the lens of the present invention, and Figure 7 is the input/output characteristics of the optical position detector among the components of Figure 6. FIG. 8 is a block diagram of the electric circuit of the automatic tracking section, FIG. 9 is a sectional view of the movable lens, and FIGS. 10 and 11 are block diagrams of the signal processing circuit. FIG. 12 is a specific example of the pulse width peak value conversion circuit 55 in FIG. 11, and FIGS. 13 and 14 are block diagrams of other signal processing circuits. FIG. 15 is a diagram showing an example of a mechanism for moving the light beam on the sample surface in parallel. 1...beam, 3...sample, 8...rotation mechanism,
9... Laser, 20... Prism, 22... Lens, 2
3...Photodetector, 24...1/Long skin plate, 66...Automatic, focus follower, 67...
...Movable stand. Cluster / ゞ Zutsutomu / Figure many boxes Tsutomu Zumanzu Ku side hair Ku Figure 7 Figure multiple a next Tsutomu Te Zutsutomu /

Claims (1)

[Claims] 1. An apparatus for optically detecting various defects in a plate-shaped sample, which includes an optical system for irradiating a light beam from a light source onto the sample surface via a converging lens, and an optical system for optically detecting the vertical displacement of the sample. a detection means, a servo mechanism for vertically moving a converging lens of the optical system according to the vertical displacement of the sample detected by the detection means, a rotation mechanism for rotating the sample, and the sample surface. A mechanism for parallelly moving the light beam on the sample surface, and a secondary mechanism that returns from the sample surface after being disturbed by reflection, diffraction, scattering, absorption, etc. when the light beam on the sample surface passes through a defect location. A defect detection device comprising a photoelectric conversion means that receives light and converts it into an electrical signal, a signal processing section that receives this signal and processes it according to a purpose, and a display section corresponding thereto.