JPH03238348A - Method and device for measuring fine defect of material in non-destuctive manner - Google Patents

Abstract

PURPOSE: To measure the density and directions of fine crystals on the surface of a material capable of the penetration of electromagnetic radiation, and on crystals just under it, and other fine crystals in a nondestructive way by measuring electromagnetic radiation diffused in a defective part. CONSTITUTION: An electromagnetic radiation beam 16 generated by a laser 32 is directed to a part 10 to be tested by a specified fixed incident angle, and is condensed by a beam expander 36. And the detecting line 26 of a detector 34 is directed to the part 10, and diffusion from a minute fixed angle near the line 26 is inputted to the detector 34 for detection. Here, the extent of electromagnetic radiation diffused is limited, and its one part is detected, and the radiation is converted into an electric signal proportional to the detected intensity. In contrast to the detector 34 fixed, the part 10 moves concerning axes X, Y, rotates around axis A, and the position of rotation of maximum diffusion intensity are selected and mapped. These are proportional to dimensions and the number of defects respectively or relate to the directional characteristics of the defects, and the distribution of fine defects in a specified region is measured by these.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a technique for non-destructively measuring the density and direction of fine defects on the surface of a material, on crystals just below the surface, and on other microscopic defects.

BACKGROUND OF THE INVENTION Many materials used in semiconductor manufacturing, optics, and various applications require the highest quality to meet future expected performance. This is particularly important where surface quality and cleanliness are concerned, and where crystal structure, material defects and impurity concentrations are concerned.

For example, in the case of silicon and gallium arsenide semiconductors,
Crystal defects, or impurities at or near the surface of a material, can significantly degrade the performance of electrical components and integrated circuits made from the material, and can render them inoperable. be. Defects in optical materials such as semiconductors, glasses, and metals, when used with high power lasers,
Or if secondary optical effects are used, it may have fatal consequences. This situation has been recognized for some time, and various devices have been disclosed or developed to measure the surface properties of these specific materials. For example, US Patent No.
No. 4,314.7H No. 2 Defect Detection System disclosed one of several techniques for measuring surface defects and contamination on semiconductors. Measurements were even more difficult, for example in US Patent No. 4.391,5, which is similar to the aforementioned patent.
24 entitled "Method for Inspecting the Quality of Light Diffusing Materials" disclosed one technique developed for that purpose. The second method is disclosed in U.S. Pat. a method and apparatus for determining;
"and U.S. Patent No. 4,314.017 entitled "Semiconductor Surface Quality Determination Apparatus." All of these Mj standard methods, material improvements, and the most important subsurface crystallization in using these materials. It has important limitations when determining a1 damage and other micro-defects at the surface and subsurface.

The term “defect” as used in this document refers to defects formed when external materials are introduced into the crystal structure, such as contents, precipitates, impurity films, and other impurities related to defects, as well as defects caused by slippage. , positional errors, stacking errors, and defects found near the surface of large materials such as embedded scratch marks, or various structural crystals that grow within layers or thin films or that occur secondarily during processing. It also includes embedded scratches and holes that occur in amorphous and polycrystalline materials, and non-crystalline structural defects at interfaces between layers that are purposefully processed or formed during processing. The term "defect" is also used to describe holes, scratches, damage, pinholes, and treated impurity films, contents, and air bubbles, as well as chips and other surface contamination.

Basically, there are three ways in which defects occur in the target material. First, defects are created within the material when it is manufactured into large chunks. For example, when a single crystal of silicon or gallium arsenide is grown, positional errors or pools ( 1) oule) may be formed. Polycrystalline and amorphous materials! Regarding 4, impurities can also easily come from the starting materials, the equipment that comes into contact with the materials, and the surrounding gases when the materials are manufactured.

Bubbles and contents are also formed during the melting and cooling process.

Second, after the material is manufactured, it must be cut into useful sizes and the surface must be ground and polished for further processing. The cutting, grinding, and polishing steps also cause not only surface defects such as holes and scratches, but also sub-surface slippage, positional errors, and other impurities in the crystal structure. Polycrystalline and amorphous materials have the same problem with surface defects. These materials have surface defects caused by the high pressures generated during grinding and polishing. These buried defects form part of the surface layer of the material and recrystallize under pressure, or become amorphous and agglomerate of the material. Impurities are formed in materials by action by diffusion or other mechanisms. For semiconductors, the second class of defects created during this process is typically 1,000 times smaller than the defects grown in the original pool of material.
It becomes a number from 0 to 1.000,000 times. Defects that are not only large in number but also grow across the width of the material are distributed throughout the material, while defects such as those described above are all located near the surface.

Third, defects such as stack-up defects, deposits, positional error lines, and defect-induced ion implants can be caused by various processing processes commonly used to process semiconductor wafers. Similarly, for optical and other materials, not only crystal defects (but also pre-treatments can introduce buried defects in amorphous and polycrystalline materials, coatings, etching, ion implantation, etc.) Processing processes such as cleaning and cleaning can create surface and subsurface defects. These defects can affect the reflection from or the transmission path of light through the optical material, and can affect the properties of the electrical material. As can be easily understood,
Another effect is the coupling between all types of subsurface defects and coating defects appearing on the surface. It has long been known that defects in the surface of a substrate result in defects in the thin film coating adjacent to that surface. Subsurface defects are complex in that they are very difficult to detect non-destructively, but can be similar to surface defects in the occurrence of defects within coatings. Such effects are extremely important because optical, electrical, and many other applications allow for a wide range of uses for coatings.

For example, epitaxial layers grown on semiconductor wafers may have stacking faults created during the manufacturing process.
These can be related to defects already present in the underlying wafer.

One current technique for measuring crystal damage is described in U.S. Patent No. 4,352.01f3 and no.

No. 4.352,017. In this method, the reflection of ultraviolet rays from the surface of a semiconductor sea urchin or shell is determined using two wavelengths. This technique is known as a method that does not affect damage at any depth because it uses ultraviolet light that penetrates shallowly into the semiconductor material. The second factor that greatly limits sensitivity is the reflectance measurement itself. Such measurements are very difficult and result in looking for a small number of species out of a large number, which is one of the reasons why this technique requires measurements at two wavelengths. Practical applications of this reflectance measurement technique demonstrate these deficiencies.

The second method is described in U.S. Patent No. 4,391.5
24. This method can measure light diffused from the surface and subsurface regions, but due to the measurement geometry, important data is missing.

There are three factors that cause this result that are independent of the wavelength of light chosen for the detection beam. First, the angle of incidence of the detection beam is O''. This eliminates any possibility of determining the directional properties of the defect or the use of polarized light to help distinguish between surface and subsurface defects. Second, since the detector faces a large fixed angle, it integrates the diffusion from all directions, again making directional defect determination impossible and diluting the features indicative of the defect being measured at the same time. Finally, the line of sight detection line is also set to 0° or 0
It is around °. This is an important magnitude of the measured signal of surface diffusion, which under these conditions
It is almost impossible to separate from subsurface diffusion. Subtle changes in surface diffusion can mask diffusion from below the surface, which is the objective of the measurement. As a result, the measurement is not sensitive to directional defects, which are most subsurface defects, and is also insensitive to many small surface defects that have no directionality.

Other techniques aimed at measuring surface and/or subsurface defects using diffusion measurement techniques all measure the total integrated diffusion from the surface of the test section.

This technique, known as TIS, integrates diffusion from the surface and subsurface as well as from all directions. As a result, it is insensitive to the Mj constant of surface roughness, which is the original objective of this technique. The surface diffusion component of the total diffusion from the material is so large that it overwhelms the subsurface components if the diffuse light is concentrated in some area near the specially reflected beam. Many microdefects, such as holes, debris, are carved such that diffusion is generally large in one direction and absent in all other directions. For large holes and debris, this accumulation does not dilute the signal, but for very small holes, debris, the accumulation signal from the TIS may not show significant changes.

SUMMARY OF THE INVENTION All diffusion #J techniques used for non-destructive measurement of surface defects are limited to defects of approximately 1 micron or less when the defects are uniformly formed and are not spherical or hemispherical. cannot be detected accurately. Actual defects, especially fragments, are not shaped like--but instead have rectangular or unusual shapes. These irregularly shaped defects do not spread light uniformly and cannot be accurately detected by these techniques. In fact, only a small portion of such defects can be detected by these techniques.

(Means and actions for solving 1ffi) The object of the present invention is to improve the ability to measure crystal surface and subsurface densities and other micro-defects in semiconductors, optical materials and other specialty materials, and coatings. The objective is to improve the situation. The purpose of this invention is to nondestructively measure the density and direction of micro defects on the surface of a material, on crystals just below the surface, and other microscopic defects through which electromagnetic radiation can penetrate, and to measure the radiation diffused at defective areas. This can be done by

According to the invention, when subsurface measurements are desired, the M
The material specified must have low surface roughness and be clean. The material must be illuminated by a suitable beam of electromagnetic radiation from a laser or other high power single wavelength source. Properly selected polarization and wavelength are required;
The penetration depth of the radiation is thereby controlled. Measurement geometry is particularly important for optimal results. For materials with high optical transmission, the natural diffusion of the material should not significantly exceed the diffusion from defects. The thickness of the part must be thick enough to separate the front and back side diffusions of the surface, otherwise both surfaces are important for the particular application. This separation depends on the field of measurement with respect to the detector. For thin transparent sections that are polished on both sides, the diffusion on the front and back sides of the surface overlap and cannot be easily separated.

In this case, the overall diffusion provides an overall estimate of the subsurface portion and bulk diffusion associated with each side of both surfaces. When bulk diffusion is small, this form of evaluation for surface and combined subsurface is of great benefit. The intensity of the beam must provide a sufficiently large spread to allow subsurface measurements from a given depth. However, the luminous density of the beave must not be so strong as to cause damage to the material being measured.

Multiple selected wavelengths can also be used to simultaneously detect defects at different depths. Depth gradient defect detection can be performed by separating the diffusion signals for each wavelength and subtracting the shallow penetration wavelength signal from the deep penetration wavelength signal. As a result, it is possible to detect defective regions at a certain depth while eliminating the effect of defects near the surface. This can also be done by varying the laser intensity, since higher intensity gives more force to deeper parts, allowing detection of defects in deeper parts.

The angle of incidence of the beam, the angle of detection of the detector, the polarization of the beam, and the polarization of the detected radiation are determined by the diffusion features from defects and at the same time from unwanted sources in and on the material.
s) is used to minimize diffusion from Relative rotation of the material with respect to the detection beam is used to determine the direction of the defect and enhance the properties of diffusion from the directional defect.

This relative rotation can be used to enhance diffusion from oriented or non-oriented defects in different directions and to minimize diffusion from known defect directions in or on the material. . For example, by choosing a rotating part of the diamond, a direction parallel to the scratch marks caused by the diamond rotation, one can remove the high dispersion associated with these marks and detect other very small defects below or on the surface. .

The measurement technique of the present invention is very useful for detecting processing-induced defects on or within the surface of single-crystal semiconductor materials. Single crystal semiconductor materials include silicon, gallium arsenide, indium phosphide, mercury cadnium telluride, as well as generally optically transparent materials such as quartz and sapphire. Defects on or within polished metal surfaces can also be detected, but the depth of defects is limited for such metals.

This technique can remove actual debris, holes, and other particles smaller than a micron on semiconductors, optical devices, and other highly polished surfaces such as plate magnets used for computer data storage, and optical disks. is effective in detecting surface defects.

To use the data generated by such a system constructed in accordance with the present invention, it is desirable to display and plot the data in the form of a map of defects. This can be done by transposing color changes to diffusion intensity changes and plotting these color changes on a map about the X and Y axes of the diffusion measurement. By adjusting the color distribution on the map, certain features can be found or enhanced. For example, separate impurity films just below the surface are each indicated as high diffusion points, and the diffusion level of these diffusion points is often higher than the back diffusion level.

These diffuse points can be separated as a single color of high contrast and their distribution shown.

Many of these indications are possible as specific defects are related to specific diffusion characteristics. Black and white maps can also be created by using shades of gray as an indication of diffuse intensity. mapping
is defined as storing data on a computer for later analysis. The result is a number or string of numbers,
Or it may be an electrical signal for the operation of another computer, other device, or person.

EXAMPLE Figures 1 and 1a show a material, namely a test section 10, which includes a beam of electromagnetic radiation 16 incident at an angle of incidence A.
an optical material having an illuminated surface 15, or a semiconductor wafer. Key material or part surface to be measured 15
should have only microroughness as shown in enlarged view 12 of FIG. 1a. For example, for a wavelength of 632.8 nanometers, with a root mean scare (rms) microroughness of 0.005xL, or about 30 angstroms or less, the surface diffusion 17 will overwhelm the near subsurface diffusion 18. Never. Also, the surface 15 must be clean so that diffusion from surface dirt does not overwhelm the diffusion 18 from surface defects. Otherwise, the desired result is a measurement of surface 19 staining. Similar surface roughness criteria apply when detecting isolated surface defects. In most cases, the general surface roughness should be so low that backside diffusion is not a significant part of the signal.

For measurements of superficial defects, the incident beam 16 is set to optimize wavelength, polarization, intensity, and angle of incidence for transmission to a selected depth into the material 10. This increases the diffusion 18 from below the surface, while the diffusion 18 from the surface
Decrease 7. The wavelength is selected by examining the material properties, the extinction coefficient k, and the penetration depth of the radiation L/(2pi
k) (where pi is a well-known constant C3,
14159)) must be greater than the defect depth in the subsurface defect region 20 and shallow enough that the radiation does not penetrate the thin part of the back surface. For example, defects created during processing in silicon wafers are only a few microns to a few tens of microns, whereas the thickness of an evaporator is 350 to 400 microns. At the same time, a beam with a wavelength of 632.8 nanometers has a penetration depth of approximately 3
．． It is 7 microns. For thick sections, the material can be transparent to the wavelengths used and k can be a very small value. The limit in this case is the level of natural bulk diffusion of the material. If the level of bulk diffusion is significantly higher than the diffusion from surface and/or subsurface defects, accurate measurements are very difficult.

Penetration depth gives a relative indication of how far below the surface the detection beam will penetrate before it is completely extinguished.

The actual detection depth, ie the depth of meaningful detection as determined by these diffusion features, is approximately proportional to the penetration depth.

The detection depth for a given wavelength is strongly related to the energy density (power/area) of the detection beam at the surface, and this detection depth also depends on the sensitivity of the diffuse photodetector system, as well as the polarization, angle of incidence, detection It is also related to other physical parameters such as the solid angle detected, and the system signal/noise ratio, which is defined in this document as the diffuse signal that is detected separated from the overall system noise.

The polarization for surface defect diffusion is selected for the minimum intensity of the reflected beam 22 and the maximum intensity of the transmitted beam 24. In all cases, the maximum transmitted power is output by p-polarized light. p in this case refers to the state in which the electric vector 25 of the incident radiation 16 is parallel to the plane of incidence. This maximizes the intensity of the transmitted beam 24 and therefore reduces the spread from the subsurface defects.
8 is the maximum.

Other polarizations can be used to achieve various results. For example, if S deflection is used (in this case S
(for the state in which the electric vector is perpendicular to the plane of incidence), the reflected intensity is maximum and the transmitted intensity is minimum. Therefore, the surface diffusion 17 is enhanced relative to the subsurface diffusion 18. This polarized light is used to detect surface defects.

The intensity of the incident beam 16 must be sufficient to provide a sufficiently large surface spread intensity 18 for measurements at the depth of the object. The intensity decreases with depth, so by adjusting the intensity the detection depth can be controlled. However, the beam energy intensity at the surface (
power/area) should not approach the damage threshold of the material being tested. Even a low-power beam, when focused into a tiny spot, can store enough energy to damage the crystalline structure within a material, causing defects when measured with this beam.

The angle of incidence A can vary between 06 and 906. Generally, large angles of incidence produce the best results. This is because the penetrating electromagnetic radiation interacts with subsurface linear defects, and the surface defects act like a lattice, diffusing light through the surface and back. The enlarged section 12 in FIG. 1a shows what the actual surface and subsurface are like and shows the density of defects increasing with depth.

FIG. 2 is a schematic representation of the surface 15 of the part 10 subjected to testing. The X1YSZ axis is shown with the Z axis perpendicular to surface 15. What is the angle of incidence A of the incident beam 16? The line of sight 26 of the detector (not shown) is determined by the corner. Incident beam 16, reflected beam 22, detector line of sight 2
6, and the Z axis are all on the same plane. The angle formed by plane 28 and the X axis is angle R. Z axis, entrance beam 16
, and the detector line of sight 26 intersect at the same point on the surface of the test part, ie the point being tested.

In subsurface measurements, the angle of incidence A of the beam 16 is adjusted to the Brewster angle in order to maximize the energy transferred into the material and at the same time minimize the energy diffused from the surface. angle)(P
It is better to be as close as possible to the minimum reflection angle in deflection (sometimes called the polarization angle). For example, the Brewster angle for silicon is approximately 75°.

Generally, the angle of incidence is greater than 55'', which is problematic because the point of impact of the incident beam 16 begins to spread through the surface.

The reflectance of P-polarized light at a wavelength of 632.8 nanometers and an angle of incidence of 55'' is approximately 15%, i.e., 8596 of the light penetrates the material. In practice, this percentage of light is 30
Energy density in watts/am2, diameter (1,25 m
With a suitable detector using m-beam, defects in silicon that are 4 to 5 microns below the surface can be detected. With the main axis of the ellipse perpendicular to the plane of incidence, the beam cross section can be shaped from a circle to an ellipse, thereby allowing even more circular cross sections to be projected onto the surface at large angles of incidence. Other angles less than 55° can be used, but the signal-to-noise ratio is inferior and this signal-to-noise ratio is inferior.
The noise ratio is used in this document as the ratio of the detected subsurface diffusive power to the total detected power.

For detection of surface defects using S-polarized light, the angle of incidence of beam 16 should maximize the amount of surface diffusion and minimize subsurface diffusion. This is necessary because for any given material and given wavelength of the S-polarized incident beam, the greater the angle of incidence, the greater the reflectivity.
Since Brewster's stagger does not exist for S-polarized light, the largest angle can be selected. For convenience in design and equipment construction, Brewster's angle can be used, depending on the type of defect measured, surface or subsurface. Depending, the polarization switches from S to P.

The corner of the detection line 26 of the detector should be located in the opposite direction of the reflected beam 22. It is desirable that the angular difference between ADs is small. less than 30" for large values of A, and greater than 30" for small values of A. Energy diffused from below the surface must traverse some thickness of the material (19 in Figure 1A), and from high index The above is correct because it appears at a low index.
This causes severe refraction of the diffused light towards the surface 15 of the material 10 being tested, especially for high index materials such as semiconductors. Other angles and A can be used, but will result in a reduced signal-to-noise ratio. Values of D that are spectrally close to the reflected beam 22 increase the surface diffuse component of the detected signal to the extent that subsurface measurements cannot be made very accurately.

The detection line 26 of the detector is very important with respect to this invention. The reason is that the fixed angle 27 blocked by the detector is relatively small compared to other measurement techniques.

In most diffusion II constant techniques, the diffusion is collected from a large fixed angle. As an effect, this averaging method is
We eliminate one type of important information in surface and subsurface measurements, and that is R-angle position information. The above is an important factor in enhancing the diffusion characteristics since the defect portion acts like a lattice and directs the diffusion in a specific R-angle direction while eliminating diffusion from virtually all directions. A large fixed angle 27 aggregates and dilutes this angular information. The invention described herein is based on the fixed angle or zero angle intercepted by the detector.
．． It works best when it is less than 1, preferably between 0.001 and 0.01 steradians.

For highly permeable materials, the range of the measurement area is limited, natural bulk diffusion and extraneous diffusion from other surfaces are not detected, and
distort the results. A common method is to limit the detection range of a detector by setting a range stop on the detector optics, but other methods can also be used to solve this problem.

In some situations, diffusion from the back side of the component is also important, such as transparent membranes that are polished on both sides. In this case, a wide field of view is required for detection of surface and subsurface diffusion.

The effects of the natural bulk diffusion of the material must be considered for this condition.

The angle R of the plane of incidence 28 is very important in determining the orientation of subsurface defects within a particular material. The nature of surface defects in many materials is in the form of lines, which are formed during cutting, grinding, polishing, and even cleaning processes.

These lines are a type of defect area where length is more important than width.

The effect is these lines from a precision grid. Diffusion from these features is therefore strongly directed and the plane of incidence 2
8 or perpendicular to them, the diffusion is very strong in the plane of incidence. At all other angles, the diffusion is less than or non-existent in the plane of incidence. This means that the angle R is directly related to the direction of the subsurface defect.

This directionality is directly related to the process used to form the surface. Therefore, the effect of the type of treatment can be directly observed. Ideally, the angle R and the maximum diffusion with respect to this corner are determined separately for each direction of the material.

Other positions for the angle R are also important from a crystallographic point of view. For example, certain defects, such as deposition errors,
Oriented in a certain crystallographic direction.

Therefore, if this particular type of defect is important, select an angle R that coincides with this known particular direction to detect deposition errors while minimizing the effects of other defects. There are situations where there are known directions that must be avoided because the signal from that direction may overwhelm the signal from the defect of interest.

This is the case when the measuring diamond rotates.

The scratch marks left by the diamond rotation on the surface and below the surface are large and the signals from them make smaller defects undetectable. By avoiding directions perpendicular to the scratch marks and concentrating on directions parallel to the scratch marks, small surface and subsurface defects can be detected.

Other types of using the R-angle are possible, such as when determining a '2nd maximum or quadratic effect that can be detected using only the maximum R-angle position.

The R angle is also very important when looking for isolated surface defects. Many surface defects, especially debris that are smaller in size than the wavelength of the detection beam and have unusual shapes, scatter a significant amount of energy in a single, narrowly defined R-angle direction. By examining the change in spread vs. R angle at a given point, these debris or other defects can be detected even if they are much smaller than the wavelength of the detection beam.

Detection of diffusion is possible with fixed state devices or photomultiplier type devices. Detection of diffuse signals can be enhanced in all cases by chopping or pulsing the beam, a technique well known in the art. This is necessary when detecting high quality single crystal materials and achieving the highest possible signal-to-noise ratio.

Regarding each aspect of the invention described above, the angle of incidence A,
The viewing angle detection line, the Z-axis angle R, and the fixed angle intercepted by the detector 27 are beneficial for improving the diffusion properties from subsurface defects.

These four aspects of the invention can be combined to significantly improve the sensitivity of the measurement device to the most difficult types of subsurface defects. The embodiments of the invention described below have been specifically selected to enhance the ability of the apparatus to best utilize these elements of the invention.

The following drawings are main embodiments of the invention. A first embodiment is shown in FIG. In this embodiment, the source of electromagnetic radiation, laser 32, and detector 34 are stationary, and test station 10 moves about the X and Y axes and rotates about the Z axis.

Detector 34 is a photomultiplier tube or solid state detector. The various operations of the testing department are
This is accomplished by computer-controlled, motor-driven micropositioning stages 39 and 37 for movement in the X-axis, Y-axis, and stage 41 for rotational movement about the Z-axis.

The three stages are configured such that the Z axis always captures the test surface at the intersection of the X-Y coordinates being measured.

At each X and Y position, the angle R is from 0'' to 360
The maximum value is read by the detector 34 and received by the computer. A map of maximum diffuse intensity versus position is generated. This is just one of many possible datasets. Maps consisting of directions rather than maxima, or multiple directions, can also be created. These maps are the output of the measuring device of this invention.

A special filter beam expander 36 imparts a Gaussian cross-section to the beam, focuses the beam to a predetermined diameter, and impinges on the test surface, increasing the energy density to a level suitable for measurement. Other beam cross-sections are possible, some scanning like a tophat with constant intensity through the beam diameter, but a Gaussian cross-section is acceptable and the easiest to produce. It is. Polarizer 38 ensures the correct polarization P or S depending on whether surface or subsurface defects are being measured. S and P
Combination of polarizations is possible, but in this case the detector would have to select the polarization of interest, or the result would be a combination of surface and subsurface effects. A beam spectrometer 30 splits the beam so that a small portion of the energy is directed towards a detector 31 whose output is fed to a computer.

This information is used to calibrate the output of detector 34 to changes in the input from laser 32.

The cylindrical lens 33 reshapes the beam cross-section into an ellipse, making the footprint on the test section surface circular. The reflected beam 22 is
It is blocked by the absorber 40 to eliminate false reflections.

In FIG. 4, the detector 34 is fixed and the test section 10 makes only rotational movement about the Z-axis, which is also fixed with respect to the test section 10. In FIG. The detection beam 16 is scanned in the X and Y directions using movable mirrors 42,44. The movable mirrors 42, 44 oscillate in planes that are 90" from each other. A lens 46 calibrates the angle of beam change caused by the scan. The beam stop 40 is large enough to accommodate the scanned beam. , prevents erroneous readings due to the beam being reflected back to the detector or diffused. With this configuration, scanning can be performed over a wide range in a short time. Using the micro position setting section 41, By rotating a material, the angle R must still change, but the material is fixed in the X and Y directions.

According to the embodiment, it is not possible to detect the X-Y coordinates, and it is not possible to rotate around the Z-axis through a specific point of the test part. The R angle must be fixed,
The entire surface of the test section is scanned in the X and Y directions to create one complete map.

This requires considerable processing to obtain a map of the final configuration, or is used when the direction of the defect is known from previous measurements. In this configuration, the detection range of the field of view must encompass the entire area scanned to obtain accurate results.

In FIG. 5, the test section 10 is moved in the X and Y directions by three minute position setting sections and 37, and the beam is rotated by a stage 41 around the Z axis.

This separation of the shafts can simplify the design in many ways, especially since there are no electrical connections to the rotating parts of the device. The beam is generated by laser 32 and processed as described above, except that element 38 converts the incoming radiation to circularly polarized light. The beam is then moved through the center of the annular mirror 50 by the directional scanning mirror 48 and through the rotation stage 41
Determine the direction through the center of. At this point, beam 16
coincides with the Z axis. The beam is directed by mirror 56 through a polarizing filter 62 and a cylindrical focusing lens 64, which converts the incoming radiation to the correct polarization.

Finally, the beam 16 is incident on a mirror 60 which provides the correct angle of incidence for the beam to enter the test section 10. The reflected beam 22 is intercepted by a beam absorber 40. The diffused electromagnetic radiation is collected by mirror 58 , reflected by mirror 56 , passes through a hole in stage 41 to mirror 50 , and travels via path 26 to detector 34 .
reach. All optical components mounted on the optical henna 52 rotate about the Z axis of the movable part of the stage 41.

FIG. 6 shows an embodiment of the invention, a high speed defect detector on semiconductor wafers or planar surface materials.

In this case, the test section 10 is moved in the X and Y directions by means of the micropositioning section 39.37, and the beam is rotated about the Z axis by means of the air hair ring support optical element 65. The air bearing 66 allows the optical element to rotate very quickly, so that measurements can be made quickly. Detection beam 16 is generated by laser 32, and the beam is filtered and shaped as described above via optical element 36. Optical element 38 converts the beam to circularly polarized light. The beam must align with the Z axis, which is the axis of rotation of the air bearing 66. The beam passes through rotating optical element 65. The rotating optical element 65 consists of two prisms 67 and 68, which are made of a material 69 that does not transmit light.
separated by. Prism 67 reflects the beam 90 degrees about the Z-axis through the face of prism 67, which has a polarizing film 70 on its surface. This converts the circularly polarized beam 16 to a P polarization state for subsurface measurements. The beam is reflected from the annular mirror 71 to provide the correct angle of incidence when entering the test section 10, and the reflected beam 22 then returns to the mirror 71 and passes through the prism 68, which has a polarizing film 72 that passes only the S-polarized light. Reach Fes. As a result, the intensity of the reflected beam 22 is reduced by an important factor, converting the remaining light into S-polarized light. The beam then reflects off the face of angled prism 68, which has a polarizing film 73 that passes only P-polarized light, reducing the intensity of the beam again through the bottom face of prism 68. The remaining light is
Impinging on the angled interior surface of the rotating element 65 , the rotating element 65 has a backside absorbent coating 74 . The purpose of this portion of optical element 65, and the circuit path for reflected beam 22, is to eliminate any reflections of the detection beam back into the optical system that could interfere with the measurements. For surface measurement, polarizing film 70.72
and 73 can be set to s, p, and s polarization, respectively. Light scattered from surface defects in test section 10 is collected by annular mirror 75 , reflected and returned to rotating optical element 65 and annular mirror 75 via path 26 . Condensing optical system 76
The light reaches the detector 34 via the . The vertical micro-positioning section 77 can adjust the height of the rotating element 65 above the test section 10 and adapt to any thickness of different test sections. Computer C analyzes the data and provides mapping functionality. The mapping function converts the data to color according to the current scale and also
The process is to map the color according to its coordinates.

FIG. 7 is an embodiment of the invention, similar to the example shown in FIG. 6, but this example includes three lasers with different wavelengths and three detectors for those wavelengths. Embodiments of the invention are specifically aimed at subsurface measurements.

Each of the three lasers 78.79.8n operates at a different wavelength. Each wavelength is chosen to obtain distinctly different penetration depths for the material being measured. Each laser has a spatial filter/beam expander 36 to focus and filter the output, with a portion of the beam in each case being directed to mirror 3.
separated by 0 and captured by detector 31,
Beam power is monitored.

The mirrors 81, 82, 83 are each designed to reflect well the corresponding wavelength, while transmitting other wavelengths. As a result, single beam 16 contains three wavelengths and is circularly polarized by optical element 38. This single component beam is reflected by the rotating element 65 and reaches the annular mirror 71, which directs it towards the test station 10. The light is focused by lens 76, referenced by optical element 90, and directed to transfer mirror 87.
．． 88.89 and reaches the detector 84.85.86. These transfer mirrors also strongly reflect selected wavelengths and pass other wavelengths. Each mirror therefore reflects only a single wavelength to its corresponding detector, which tunes and filters that wavelength.

For example, mirror 89 reflects a single wavelength to detector 86 while transmitting two other wavelengths to the next two mirrors.

The effect of using two, three, or more wavelengths in the M1 constant device described above is to separate the wavelengths, penetrate them to different depths, and detect them individually.

Each detector receives a different signal, which includes signals at wavelengths that have subsurface defect diffusion characteristics and are capable of penetrating to any depth within the material being tested. The wavelength that penetrates the smallest depth contains the least amount of information, while the next wavelength that penetrates deeper contains all the previous information plus newer information for one of the deeper defects.

If shallow information is obtained by subtracting from deeper information, the result is information about deeper defects. When these depths are known, defective areas of known depths are recognized. A similar process is performed to obtain a third depth region with the next wavelength, the first depth region being obtained only with the shallowest penetration wavelength, and the second depth region being obtained with the shallowest penetration wavelength only.
It is obtained by subtracting the first depth region from the deepest penetration wavelength. More or fewer wavelengths can be used depending on the amount of information desired and the complexity of the device. Since the information subtraction varies with the directional depth of the defect, it must be performed with respect to the rotational direction of the defect. The best method is to take the rotational data for each wavelength at a given point and subtract it to obtain both the magnitude and direction for that point in the desired depth region.

An advantage of embodiments according to the invention is that the depth of defects can be accurately determined in a non-destructive manner.

This depth information is not available in previous embodiments of this invention.

In the embodiment shown in FIG. 8, the defect depth is determined by a different method. This embodiment is very similar to the embodiment shown in FIG. 6 and can be specifically designed for measuring subsurface defects. The difference is that now the laser 32 has a power modulator 91 controlled by the computer C. This allows the power of the laser to be varied and controlled and different levels for the input beam 16 to be used.

The wavelength and power of the detection beam are selected to achieve a predetermined detection depth, and since the detection beam power decreases exponentially once it enters the test material, a decrease in power decreases the detection depth. do. This is true. This is because the diffusion from surface defects is very low, and there is an absolute power level that is below the noise level of the system.

This detection depth with power can be used in the same way as the change in wavelength, as shown in FIG. 7, and is used to determine the depth of the defect area. However, the application in this case is slightly different.

One way in which the second technique can be applied is to actually create several paths, each with a different power level, across the test section and subtract the appropriate data to obtain the depth information. Another method is to modulate the power level on the fly so that two or more levels are available at each rotational position. The data is then subtracted appropriately to obtain depth information.

This method has several advantages over the multiple wavelength scheme shown in FIG. The resulting device has a simple construction and is therefore not expensive. It is possible to control the desired depth by computer, and this control allows for delicate control over small changes in the detection depth.The main problem is that the detection depth by increasing the power The increase in testing department,
In particular, power levels can easily be reached that can damage semiconductors. This is due to the non-linearity of light absorption in the test situation.

Another possible method is a combination of the embodiments shown in FIGS.
A modulator 91 is added to 0.

Thereby, multiple wavelengths and multiple power levels are combined to optimize the operation of the device for the material being tested and the desired detection depth.

Any embodiment of this invention should be separated from the situation in two important ways. This device should be isolated from vibrations, eliminating the effects of subtle movements between the detection beam and the test part.

Vibrations can cause large errors since adjacent parts of the test section can have widely different diffusion properties. The device should also be isolated in a clean environment, eliminating problems with contamination of the measured surfaces. This is necessary because the sensitivity of the device with respect to all types of contamination, especially to contamination of individual particles, is very high. This last effect is beneficial when using the device for debris detection and location detection.

While the measurement methods and apparatus described in this document constitute preferred embodiments of the invention, the invention is not limited to the methods and apparatus described herein.

(Effects of the Invention) The present invention effectively improves the ability to nondestructively measure crystal surface and subsurface densities and other microdefects in semiconductors, optical materials and other specialty materials, and coatings. can. Therefore, defects of approximately 1 micron or less, which are not shaped like a spherical or hemispherical surface, cannot be accurately detected by conventional techniques, but the present invention can be detected accurately by

[Brief explanation of drawings]

1 is an enlarged cross-sectional view of a material with surface and subsurface defects illuminated by a beam of electromagnetic radiation according to the invention; FIG. 1a is an enlarged view of the intersection of the material and the beam; and FIG. 3 is a perspective view and a schematic diagram showing the detection line of the detector and the direction of the beam with respect to the material surface; FIG. 3 is an embodiment of the invention in which the material moves below a fixed beam and a fixed detector; FIG. 4; Figure 5 shows an embodiment of the invention where the beam scans in the literal X and Y directions, while the material rotates about the X axis and the detector is fixed. FIG. 6 shows an embodiment of the present invention where the beam of electromagnetic radiation rotates about the X-axis, and FIG. FIG. 7 shows a case where three wavelengths are used simultaneously for the detection beam in the embodiment shown in FIG. 6, and FIG. 6 shows the case where the detection beam is varied in the embodiment shown in FIG.

Claims (1)

Claims: 1. (a) generating a beam of electromagnetic radiation; (b) directing the beam at a predetermined fixed angle of incidence onto the surface of a material to be measured so that a minute portion of the material is exposed to the electromagnetic radiation; focusing the beam to receive radiation; (c) directing a detection line of a detector toward the surface of the material until the beam captures the surface; and (d) microfixing near the detection line. input to said detector to detect the diffusion from the corners, limit the extent of the diffused electromagnetic radiation, detect a portion of the diffused electromagnetic radiation, and convert this radiation into an electrical signal proportional to the detected intensity; and (e) generating a relative rotation of the beam and the material about an axis substantially orthogonal to the surface and about the portion where the beam captures the surface, and further comprising: (f) generating a relative lateral movement between the beam and the material to expose a portion of the adjacent material to electromagnetic radiation; and (g) step (e) and (f) is repeated for each portion of the material exposed to electromagnetic radiation until the predetermined area is covered in an approximating manner; and (h) the selected maximum diffusion intensity for said predetermined area is mapping the coordinate position of each measurement point, the diffusion intensity being proportional to the size and number of defects, and the rotational position of maximum diffusion being related to the directional characteristics of the defects, comprising: A measurement method that measures the distribution of micro defects in a predetermined area. 2. The method of claim 1, wherein the electromagnetic radiation is P-polarized for subsurface defect detection. 3. Method according to claim 1, characterized in that the electromagnetic radiation is S-polarized for surface defect detection. 4. Method according to claim 1, characterized in that the wavelength of the electromagnetic radiation is selected according to the optical properties of the material and penetrates only to a depth necessary for the detection of defects of interest. 5. The method of claim 1, wherein the angle of incidence of the beam of electromagnetic radiation is generally close to Brewster's angle. 6. The method of claim 1, wherein the detection line of the detector relative to the surface is at a sharp angle with respect to the incident beam and in the same direction as the incident beam. 7. The extent of the diffuse electromagnetic radiation received by the detector is 0.001 steradians and 0.01 steradians.
2. The measuring method according to claim 1, wherein the measuring method is between steradians. 8. Multiple beams of electromagnetic radiation polarized in P have predetermined wavelengths and can penetrate to different material depths, and these multiple beams are simultaneously generated and directed to the surface for detection by multiple detectors. the detection line, one of each wavelength is synchronized and directed toward the surface of the material until the point where the multiple beams capture the surface, in order to detect defects at different depths of the material. 2. The measuring method according to claim 1, further comprising manipulating a line. 9. Method according to claim 1, characterized in that a beam of electromagnetic radiation polarized in P is generated and the power level of said radiation is controlled and varied in order to detect defects at different depths in said material. . 10. (a) means for generating a beam of electromagnetic radiation; (b) means for directing said beam at a predetermined fixed angle of incidence onto the surface of a material to be measured so that a minute portion of said material receives said electromagnetic radiation; (c) means for directing a detection line of a detector towards the surface of the material until said beam captures said surface; and (d) detecting dispersion from a small fixed angle in the vicinity of said detection line. means for detecting a portion of the diffused electromagnetic radiation and converting this radiation into an electrical signal proportional to the detected intensity; e) an axis substantially perpendicular to said surface and a selected rotational position of maximum divergence with respect to the portion where said beam captures the surface, generating a relative rotation of the beam and said material about the portion where said beam captures the surface; (f) means for generating a relative lateral movement between the beam and the material, exposing a portion of the adjacent material to electromagnetic radiation; and (g) means for determining said means (e) and (f) to the electromagnetic radiation. repeating for each portion of the material exposed to , until the predetermined area is covered in an approximating manner; and (h) the selected maximum diffusion intensity for said predetermined area versus the coordinates of each said measurement point. means for mapping the position, the diffusion intensity being proportional to the size and number of defects, and the rotational position of maximum diffusion being related to the directional characteristics of the defects, comprising: A measurement device that measures the distribution of defects.