JPS6049858B2 - Defect inspection equipment - Google Patents

Description

Detailed Description of the Invention The present invention is useful for inspecting defects such as short circuits, disconnections, roughness (roughness) on the edges of stripes, pinholes, etc. caused by semiconductor materials formed on circular glass plates, etc. It is related to monitors.

A typical defect inspection method conventionally used is a sensory test in which defects are visually inspected using an optical microscope within a field of view determined by its magnification.

This method has drawbacks such as (1) it takes a long time to examine the entire sample surface, and (2) oversights due to inspector fatigue are inevitable. An object of the present invention is to provide a monitoring-enabled defect inspection device that automatically inspects defects such as stripes regularly arranged on a glass face plate or the like.

FIGS. 1 to 3 schematically show enlarged examples of defects in the striped glass face plate.

FIG. 1 shows defects such as short circuits, FIG. 2 shows wire breaks, and FIG. 3 shows defects such as roughness (roughness) on stripe edges. As shown in FIG. 3, there are two types of roughness on the stripe edge: whisker-like roughness that protrudes outward from the edge, and roughness that is recessed inside the edge, and both are often less than 2 μm in size. In Figures 1 to 3, 1 is a glass face plate, 2 is a stripe made of Cr or semiconductor material formed on the glass surface, 3 is a short circuit caused by the same material as the stripe, and 4 is a short circuit made of the same material as the stripe.
5 is a break in the stripe, 5 and 6 are rough edges on the stripe edge, 5 is a whisker-like thing that protrudes outside the edge,
6 is a recess on the inside of the edge.

Conventionally, an abnormality detection device has been proposed that detects defects such as scratches, dust, stains, etc. similar to such defects by optical means in a non-contact manner.

Figure 4 shows its typical configuration.

The outline is explained below. The outline of the principle is as follows.

A thin parallel light beam is supplied to the laser 7, this parallel light beam is scanned in an arbitrary area, and the sample is irradiated with this, and the Fraunhoher diffraction that occurs in the abnormal area on the sample is Fourier transformed to obtain a diffraction pattern. , except for the diffracted light in the vicinity of the zeroth order, which is received by a photoelectric conversion element and converted into an electrical signal to detect an abnormal portion. The converted electrical signal is processed by a signal processing unit to obtain information on the size and number of abnormal parts. Scanning is performed by deflecting a narrow parallel beam from a laser 7 to a rotating mirror 8 (X-axis direction) and a rotating mirror 11 (Y-axis direction).

Lenses 9, 10 and 12 are used auxiliary.
The focal point of the lens 9 is at the center of the surface of the rotating mirror 8, and the focal point of the lens 10 is
The focal points of and 12 are located at the center of the rotating mirror 11. Due to this arrangement, the light beam always gathers at the center of the surface 11 of the rotating mirror 8 even when the rotating mirror 8 scans in the X-axis direction, and the reflected light from the rotating mirror 11 also scans in the X- and Y-axis directions. Regardless, the lens 12 makes it parallel to the optical axis of the lens 12. In this way, sample 13
This means that the entire surface can be scanned with a narrow parallel beam of light. Therefore, the zeroth order of the diffracted light from sample 13 is always in the vicinity of lens 1.
It will pass through the focal point of 4. The diffracted (Fraunhoher diffraction) light due to the abnormal part of this parallel light beam is Fourier transformed by a Fourier transform lens (condensing lens) 14,
A diffraction pattern is obtained at the focal plane of this lens. However,
The light intensity near the zeroth order of the diffracted light is extremely strong, and there is not much difference in the intensity of the diffracted light between the normal part and the abnormal part, so the S/N ratio is poor. To solve this problem, a partial shielding filter 15 is installed immediately before the photoelectric conversion element to remove near-zero-order diffracted light.
Place. That is, the photoelectric conversion element 16 receives high-order diffracted light and converts it into an electrical signal. Photoelectric conversion element 1
By counting the signal pulses obtained from 6, the number of defects will be known. The drawbacks of the above method are (1)
In order to be able to detect smaller defects, it is necessary to focus the light beam more narrowly, but there is a limit to this (2
) The spread of the diffraction pattern after Fourier transformation of the zero-order third refraction light depends on the size of the defect, so it is necessary to change the shape of the shielding part of the partial shielding filter each time, but this is impossible during inspection. There are times when it is.

3. FIG. 5 is a block diagram showing a defect detection device according to the present invention that eliminates the above-mentioned drawbacks. The light source section 18 supplies a parallel light beam with an arbitrary diameter, the scanning section 19 scans the parallel light beam within an arbitrary area, and irradiates the sample 4/surface 20 as a narrow parallel light beam, and the diffraction focusing optical system The detection unit 21 separates and focuses the normal diffracted light and the abnormal diffracted light from the defective part.
The detection section converts the defects into electrical signals, and the display section 22 monitors the size and number of defects to distinguish between good and defective products.

A control section 23 is provided to operate the scanning section and the display section synchronously. FIG. 6 shows a specific embodiment of the present invention based on the basic configuration shown in FIG.

In FIG. 6, the light source unit 18 is a laser 7 and a beam expander 24, and the scanning unit 19 is a rotating deflection mirror and its drive circuit 31, a lens 12, a sample scanning mechanism 30 and its drive circuit 32, and a diffracted light condensing optical system. The system 20 includes a glass face plate 13, a diffracted light condensing lens 14, the detection unit 21 includes a slit 28, a photoelectric conversion element 16, and a preamplifier 29. The display unit 22 includes an integrator 35, a changeover switch 36, and normal observation. The control section 23 is composed of an oscillator 33 and a frequency divider 34, respectively.

A beam expander 24 of the light source section 18 expands the diameter of the parallel light beam 25 from the laser 7 to a parallel light beam 26 to any desired thickness, and the lens 12 converts it into a thin parallel light beam of several tens of microns. The rotating deflection mirror 8 of the scanning unit 18 and the sample scanning mechanism 30 perform scanning in the X-axis and Y-axis directions, respectively. The center of the rotational axis of the rotating deflection mirror 8 is within its reflecting surface, and the first focal point of the lens 12 is located on the center of the rotational axis. Therefore, when the light beam 26 is emitted from the lens 12, the center of the light beam is parallel to the optical axis of the lens 12, and the sample surface is irradiated perpendicularly. Sample 13 is placed at the second focal plane of lens 12. At this time, the luminous flux diameter DB on the sample surface is given by:

Here, f is the focal length of the lens 12, λ is the wavelength of the laser used, and D8 is the diameter of the light beam 26 expanded by the beam expander 24. Therefore, in order to obtain the desired DB, a short focal length lens should be used as the lens 12.
This is done by increasing the diameter of the light beam incident on the lens 12, etc. Alternatively, a laser with a short wavelength input may be used. The purpose of using the beam expander 24 is to make the DB thicker. The method shown in Figure 4 does not have this feature, so it cannot be narrowed down. The sample is
The stripes are arranged so that the longitudinal direction thereof is perpendicular (Y-axis direction) to the deflection direction (X-axis direction) of the light beam.

Under this arrangement, diffraction from the normal stripe portion occurs regularly as a light source behind the sample 13. if,
If a stripe has a defect such as a break or a short, in addition to regular diffracted light from the normal stripe portion, irregular diffraction occurs at the defective portion. A typical example is shown in FIG. The sample 13 is illuminated by the light beam 26 from the front. The scanning direction is the X-axis direction. The longitudinal direction of the stripes is parallel to the y-axis. Now, if we consider that the light beam 26 is stationary at the center 0 of the sample, if there is a defect as described above near the center of the sample, the diffraction due to the stripe will be in the X-axis direction. Furthermore, the zero-order light is irregular in diffraction due to defects, and the higher-order light 27'' is mostly in the y-axis direction.
Therefore, if the diffracted light is focused in the θ direction from the y-axis within the YZ plane, only the diffracted light from the defective part can be focused without focusing the diffracted light from the stripes. Ru. This configuration needs to satisfy the following equation due to the arrangement of the O device.

It can be calculated with reference to FIG. 7 and using FIG. Here, I is 1/2 the size of the sample, d is 1/2 the size of the condenser lens, and f is the focal length of the condenser lens.

In this way, the diffracted lights in the ζ and θ directions are collected by the condenser lens 14 and converted into electrical signals by the photoelectric conversion element 16 and the preamplifier 29.

According to this arrangement, the photoelectric conversion element does not receive diffracted light due to regularly formed stripes, and a signal with a high S/N ratio can be obtained without requiring a partial shielding filter. Furthermore, since it does not contain a DC component, it has the advantage of being able to utilize a wide dynamic range of the amplifier system. Diffraction light condensing lens 14
The first focal point of is located at the center of the sample 13 and the second focal point is located at the center of the slit 28.

The slit 28 is a rectangular slit with a length that covers the scanning range of the sample, and its longitudinal direction is aligned with the scanning direction of the light beam. Note that this slit 28 is used to remove stray light from sources other than the sample surface. There are drive circuits 31 and 32 to drive the rotating deflection mirror 8 and the sample scanning mechanism 30, respectively, and the oscillation frequency is controlled directly from an oscillator 33 with a variable oscillation frequency or by a frequency divider 34 so that the scanning speed can be arbitrarily selected. The signals are supplied to respective drive circuits 31 and 32 via the respective drive circuits 31 and 32.

In this way, the X and Y scans are synchronized. Here, an example of a signal is shown. 9A shows a signal during line scanning due to a short circuit defect as shown in FIG. 1, and FIG. 9B shows a similar signal due to a disconnection defect as shown in FIG. The electrical signal from the preamplifier 29 whose defect was detected in this way is transferred to a cathode ray tube (or TV).
The signal from the oscillator 33 is branched to the horizontal scanning signal H of the cathode ray tube, and the output of the frequency divider 34 is branched to the vertical scanning signal (■) of the cathode ray tube.
The detected electrical signal is supplied as a brightness signal Z to a cathode ray tube to form a monitor, and information such as the size, number, and location of defects can be obtained at the same time. In other words, defects such as disconnections and short circuits are displayed as bright spots on the cathode ray tube, and the size of the defect is expressed as the size and brightness of the bright spot, the number of defects is expressed as the number of bright spots, and the location of the defect is expressed as the location of the bright spot. can be made to correspond to each. When monitoring defects as shown in FIGS. 1 and 2, the changeover switch 36 is connected to contact A.

Next, consider diffraction due to defects as shown in FIG. (
1) Irregular shape, especially the direction of the ridge of the whisker-like protrusion,
(2) The size can be expressed as undefined. Although it is not possible to mathematically solve diffraction due to such complex shape defects, the following analogy can be made based on the general properties of diffraction images in light diffraction theory. (a) The diffraction image has a striped film-like appearance, and this stripe has many stripes in different directions (point symmetry of the aperture and superimposition of their breaks, etc.). (b) The direction in which the diffraction image appears covers almost the entire circumference (irregularities in the orientation of the aperture, i.e., the points and
(Breaking from symmetry and line symmetry, and reciprocity of diffraction images). (c) Since the defect is smaller than defects such as stripe breaks and shorts, the intensity of the diffraction image is weak.
Therefore, for a defect like that shown in FIG. 3, with the configuration shown in FIG.
The total intensity of the light collected is weak, and the electrical signal converted by the photoelectric conversion element is also small and in the form of many pulses.

Therefore, in order to detect this defect, it is sufficient to obtain an envelope over one period of scanning using an integrator 35 with a low time constant, and determine whether or not there is a jagged defect based on the size of the envelope. FIG. 10 shows signal waveforms obtained in the form of envelopes depending on whether or not the stripe edges are rough. Figure a shows a case where the stripe edges are not rough, and figure b shows a case where the stripe edges have roughness. As you can see from this,
The waveform and magnitude of the signal obtained in the form of an envelope vary depending on whether or not the edges are rough. When monitoring this defect, the selector switch 36 is connected to contact B in FIG.

At this time, the presence or absence of a defect and the degree of roughness of the stripe edge can be obtained on the monitor as the size of the shining part and its shading. FIG. 11 shows another specific embodiment of the present invention, particularly showing the scanning section.

The omitted parts have the same structure as in Fig. 6, and the X and Y
Both axial scanning is performed by a rotating deflection mirror. The rotating deflection mirror 8 scans in the X-axis direction, and the rotating deflection mirror 11 scans in the Y-axis direction. In FIG. 11, the first focal point of the lens 9 is located on the center of the rotation axis of the rotary deflection mirror 8. In FIG.

Therefore, the light beam deflected by the rotating mirror 8 enters the lens 10 parallel to the optical axis of the lens 9. The second focal point of the lens 10 is located on the center of the rotation axis of the rotating deflection mirror 11. Therefore, the light flux emitted from the lens 10 is concentrated at the center of the rotating deflection mirror 11. In this way, scanning in the X-axis direction is performed.
Scanning in the Y-axis direction is performed in the same way, but the first
The focal point is on the center of the rotation axis of the rotating deflection mirror 11, and the second focal plane is on the surface of the sample 13. There are each above. The diffracted light condensing lens 14 is arranged with its optical axis aligned in the θ direction, similarly to FIG. 6. Figure 6 shows the optical system when the sample 13 is transparent, but the glass surface should be coated with a non-transparent coating!
After application, stripes may be formed.

In this way, even when the sample is non-transparent, the detection section configured as shown in Fig. 12 can detect defects in the same way as the embodiments shown in Figs. can be configured. This configuration can also be applied to reflective samples. As explained above, according to the present invention, by arranging the beam expander next to the laser used as a light source,
It is possible to narrow down the luminous flux from 0 to several tens of microns,
By moving the optical axis of the diffracted light condensing lens away from the optical axis of irradiation to the sample, the diffracted light from the defect can be focused without being affected by the diffracted light in the vicinity of zero order, and the defects inside and outside 5p can be focused. In addition, it is possible to detect the roughness of the stripe edge by integrating the signal obtained from the diffracted light due to the roughness of the stripe edge and processing the signal in the form of an envelope. By displaying the image on a television, it is possible to realize a glass face plate defect detection device that can monitor defect inspections and distinguish between good and defective products in a short time.

[Brief explanation of the drawing]

Figures 1, 2, and 3 are diagrams showing typical examples of defects occurring on the object to be inspected (glass face plate), Figure 4 is a diagram showing a typical example of the conventional defect inspection method, and Figure 5 is a diagram showing the defects according to the present invention. Fig. 6 is a block diagram showing an overview of the inspection device, Fig. 6 is a configuration diagram showing an embodiment of the present invention, and Fig. 7 explains the irradiation state to the sample, the diffraction pattern due to stripes and defects, the arrangement of the condenser lens, etc. Figure 8 is a geometric layout diagram for determining the value of O, Figure 9 is an example of signals due to short circuit and disconnection defects, and Figure 10 is the shape of the envelope curve depending on whether or not the stripe edge is rough. FIG. 11 is a partial configuration diagram of another embodiment of the present invention, and FIG. 12 is a diagram showing the configuration of an in-situ defect detection section in which the sample (object) is non-transparent.

Claims (1)

[Claims] 1. A laser as a light source, a means for scanning a light beam emitted from the laser onto a subject, and a photoelectric conversion light receiving section that receives diffracted light generated from the subject by scanning the light beam. defect inspection, wherein the photoelectric conversion light receiving section is located at a position offset from the optical axis of the light beam, and the photoelectric conversion light receiving section receives only diffracted light other than the zero order of the diffracted light. Device.