JP2008083059A - Measuring system and measuring device for wafer - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To precisely measure the thickness, degree of planarity, and trench depth provided to a wafer. <P>SOLUTION: The method and device which measure the thickness, the degree of the planarity and the trench depth etched into the wafer. The trench is measured actually as a projection part actually by seeing the trench from a back side of the wafer at the measurement. Measurement is performed in a front side and the back side by a non-contact type optical instrument. This non-contact type optical instrument carries out simultaneous measurement of the wavelengths of the front side and of the back side. The distance between the wavelengths is converted into a measured value of the thickness. A near infrared beam as a light source is used for the measurement of the thickness and the trench depth of silicon wafer. The thickness, the degree of the planarity and the local shape can be measured by a calibration method, using a pair of optical styli. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates generally to the field of measurement of silicon wafers used in semiconductor manufacturing, specifically thin wafer thickness, flatness, local shape, and trench depth etched into the wafer. Is related to the measurement of

  When manufacturing semiconductors, the manufacturer starts with a blank of the silicon wafer. After many steps have been performed on the blank, the semiconductor is complete, but many of these steps involve forming a photomask image on the wafer. The various images must overlap each other with great precision. Since the dimensions of the features on the wafer shrink, high accuracy is required to overlay each layer.

  In that case, the dimensions of the feature structures provided on the wafer are so fine that even the shape of the wafer often affects the quality of the photo process. Imagine a potato chip shaped wafer. Since the optical device cannot focus on the entire surface at once, the wafer will be warped so that multiple masks can never be aligned. This type of problem causes a drop in production, resulting in significant losses for semiconductor manufacturers. An instrument that accurately and reliably measures the flatness and thickness of a wafer will help manufacturers improve the manufacturing process and produce higher quality ICs with higher productivity.

  The most popular current technique for measuring wafer thickness is called capacitance testing. A wafer is placed between the two electrodes, causing a change in capacitance in the wafer material in between. This capacitance change is a direct measure of the amount and type of material between the electrodes.

  For many years, this technology has yielded reliable results. However, the limitations and drawbacks of this technique have become much larger only recently as the required measurement accuracy has increased and wafers have become significantly thinner than before.

  One disadvantage of this technique is that detailed information is required about the properties of the wafer material, such as the relative dielectric constant. This is often a problem if the wafer has a lot of material or has solder protrusions.

  Another drawback is that the number of locations where the thickness can be measured is small due to the relatively large dimensions of the capacitance sensor. Typically, the number of measurement points is about 10 for a 10.16 cm (4 inch) wafer and about 30 for a 30.48 cm (12 inch) wafer. The manufacturer would ideally want to know detailed height and thickness information for the entire wafer, not just for a few locations. In addition, measurement accuracy is a problem. This is because the measured value is effectively the average thickness over the measuring range of the capacitance sensor.

  The measurement range of a capacitance sensor is typically about 1/2 inch in diameter. More importantly, however, the resolution of the capacitance sensor is no longer accurate enough to satisfy the increasing demands of the manufacturer. The thickness of the latest generation wafer is 800-200 micrometers, and the wafer of the future generation is expected to be about 40 micrometers thick, so the measurement accuracy needs to be less than 0.1 micrometers. is there.

  In connection with the problem of measuring the thickness of a thin wafer, measuring the depth of a trench formed in the wafer becomes a problem. When a semiconductor wafer is processed into a device such as an integrated circuit, a micro electro mechanical system (MEMS), or an integrated optical communication device, a manufacturer performs many processing steps, and in these steps, a trench is etched in the wafer. Work to be included. In many of the devices, the depth of the etched trench is critical to the proper performance of the finished device, and manufacturers usually want to measure that depth. However, current trench measurement methods have significant limitations.

  MEMS products typically contain a three-dimensional structure with deep and narrow trench areas with nearly vertical sidewalls. A typical example is a trench etched to 5 micrometers wide and 100 micrometers deep, and MEMS devices having these characteristics include sensors, actuators, RF devices such as inductors and comb switches. . A feature of all these devices is that deep vertical trenches need to be etched to isolate moving mechanical parts, and finger-like features are quite common.

  MEMS device manufacturers do not currently have a precise and inexpensive method for non-destructively measuring the depth of etched high aspect ratio trenches. Manufacturers need to have precise control over the entire depth of etching when manufacturing a working device, and the measurement of the etching depth is critical to the progress and control of the process. However, current metrology techniques cannot quickly and accurately measure high aspect ratio trench depths. Therefore, if a non-contact measuring instrument that can measure the etching depth of a high-aspect ratio trench, for example, a narrow finger-shaped trench, quickly and precisely is developed, the MEMS manufacturer can improve the processing progress and control. There will be tremendous benefits.

  Integrated circuits often require deep trench etching in the wafer to electrically isolate adjacent circuit devices such as transistors. The space on the wafer is always an important consideration, but in the case of trenches it must be deep enough to obtain the necessary insulation. For this reason, the aspect ratio of the etched trenches increases as technology advances. Currently, these trenches are 1 micrometer wide and 6 micrometers deep. In addition, these trenches typically have a bottom with a rounded or rough surface that absorbs any incident light. Manufacturers need to measure these trench depths for process control and characterization. Currently, the only way to measure these trenches is the destructive method of cutting the wafer.

  Photonics integrated devices are typically fabricated on a material other than silicon, or fabricated with multiple material layers that “grow” on the silicon. Examples of the material include SiC, InP, GaAIAs, and silicon nitride. These devices are etching structures to form waveguides, lasers, and other photonics devices. Etch structure shape parameters are critical to the performance of photonics devices. The present invention is concerned with measuring deep trenches of a wide variety of materials, as in the examples above, as well as wafer thickness.

  Due to the extremely steep sidewalls typical of this type of trench structure, methods using a needle profiler or other contact-type methods are not applicable to aspect ratios or lateral dimensions of this nature. For example, an atomic force microscope (AFM) or a needle profiler is not suitable. This is because even though the tip of the needle may penetrate the trench, it cannot move along the side wall, and the tip will appear to break upon exiting the trench.

  Standard non-contact optical instruments for measuring surface height are confocal microscopes, white light interferometers, phase shift interferometers, and trigonometric instruments. All of these optical techniques include a form in which the trench is irradiated in some form and the reflected light is analyzed. However, the steep walls of the trench prevent most of the light from reaching the bottom of the trench. In addition, some etched trenches have a rounded or rough surface at the bottom. If light enters these trenches, they will be completely absorbed. If the light does not come back, the trench depth is probably not measurable by any analysis method. Apart from these basic problems, each of the above non-contact measurement methods has its own problems.

  The standard confocal microscope does not help because if the trench is too narrow, the signal from the top of the trench is confused with the signal from the bottom. When the trench width approximates the pinhole size of the light source and the focal point is at the top of the trench, more light than that at the bottom will be detected. For this reason, even if the trench bottom is away from the focal plane, a confused signal is generated. Also, the confocal microscope operates very slowly because it is necessary to scan the measurement sample axially in order to find the best focal plane.

  White light interferometers have similar difficulties in that they are slow and require scanning in the axial direction. In addition, the fringe signal is weak due to scattered light from the sidewalls and the top. Phase-shifting interferometers are completely useless because phase unwrapping does not help detect steep sidewalls. Finally, a triangulation instrument, which can only be used if the direction of the incident beam can be precisely controlled relative to the direction of the trench so that light can enter the trench from the side. This type of instrument is not suitable for use because of these limitations.

  Common to all of the prior art methods described above is the point at which the trench is to be measured from above, i.e. the approach to the trench from the side equal to the etched surface of the optical beam or mechanical needle.

  The preferred embodiment of the present invention teaches a method for measuring the trench depth of a wafer. This method includes the following steps: That is, a step of locating a trench provided on the front surface of a wafer having a front surface and a back surface by a microscope, and the trench is regarded as a protrusion provided on the second surface (back surface) from the non-contact optical instrument. And placing the non-contact optical instrument facing the back surface, the non-contact optical instrument using a light source that can transmit through the wafer and reflecting from the back and front surfaces of the wafer. By receiving light, the front surface indicates the inside of the trench, and the method further measures the height of the protrusion using a set of confocal color sensors in a wavelength range that can be transmitted through the wafer. And calibrating the protrusion height by converting the measured height difference between the measured reflection wavelengths at the front and back surfaces, thereby detecting the shape and contour of the wafer at the trench location. This There.

  The second embodiment relates to a method for measuring the local thickness of a thin wafer. In this method, a local thickness area of a trench provided in a wafer having a front surface and a back surface is determined by a microscope, and a confocal color sensor transmits reflected light from the front surface and the back surface to one side of the wafer. The local thickness of the wafer is adjusted by converting the measurement height difference between the measured reflection wavelengths of the front and back surfaces, using the wavelength range that can be transmitted through the wafer, using the wavelength range that can be transmitted through the wafer. Calibrating and thereby determining the wafer local thickness.

  The third embodiment is an apparatus for measuring the trench depth or local thickness of a wafer having a front surface and a back surface. This measuring apparatus includes a non-contact optical instrument for height measurement, a fixing means for positioning a wafer which is effectively disposed under the optical instrument, a set of light sources in a wavelength range capable of transmitting the wafer, And the non-contact optical instrument is disposed on one side of the wafer so as to simultaneously receive reflected light from the front and back surfaces of the wafer, and the measuring device further includes a height measuring device. Calibration means are included to convert the data collected by the contact optics into distances to the front and back.

In all of the above three embodiments, a silicon wafer can be used, and the light source is a near infrared light source and has a wavelength range of 900 nm to 1700 nm. GaAs, GaAIAs, InP, SiC, SiO _{2 and} others can also be used.

  In the above-described three embodiments, any one of the following can be used as a non-contact optical instrument. That is, a confocal color sensor, a white light interferometer, a phase shift interferometer, a confocal scanning microscope, and a laser triangulation instrument. The use of confocal color sensors in trench measurements is the subject of U.S. Patent Application No. 2006/0109483 by inventor Marks and Grant of the present invention, but that technique is incorporated herein by reference. To do. Measurement with the confocal color sensor is performed using axial dispersion so that each wavelength is focused at a different distance. The sensor then measures the distance to the reflecting surface by detecting the wavelength of the most reflected light.

  The embodiments described above can be further modified with reference to the definition that a non-contact optical instrument for height measurement uses calibration means. The optical instrument converts the collected data corresponding to the wavelength of the reflected light from the front and back surfaces of the wafer and the corresponding height difference into a measured thickness, which is measured by the trench depth or locality of the wafer. Determine the thickness.

  Each of the above embodiments can be further modified based on the definition that the optical instrument used mechanically scans the wafer laterally at a pre-specified sample speed and density.

  Another embodiment of the invention involves a method for measuring the thickness, flatness, and local shape of a thin wafer. This thin wafer has a front surface and a back surface. The method includes calibrating the distance of the wafer from the first sensor and the second sensor. This calibration operation further includes placing a gauge block of known thickness between the first and second sensors, also in a suitable holder, and including a plurality of parallel surfaces. Yes. The first and second sensors are then adjusted in the Z plane so that the detected surface is centered in the sensor measurement range.

  The height response from each height sensor is calibrated individually. The height obtained from the sensor is then calibrated by a first placement operation in which an angle gauge block of known angle is placed between the first and second sensors, also in a suitable holder. The flat surface of the angle gauge block is perpendicular to the first sensor, and a surface with a known angle relative to the first sensor is shown. The angle gauge block is then rotated 180 degrees, the slope of the angle gauge block is placed in the opposite direction to the first arrangement, the angle gauge block is perpendicular to the first sensor, and a surface at a known angle relative to the first sensor. Is shown.

  The height sensor calibration is then mathematically converted to calculate the slope of the angle gauge block. The angle gauge block is then rotated and directed toward the second sensor, and the steps as applied to the first sensor are repeated. The local thickness of the wafer is measured, and this measurement step further includes placing the wafer in a suitable holder so that the first and second sensors can receive responses from both sides of the wafer. It is.

  A wafer placed in a suitable holder is then moved a predetermined number of locations individually or sequentially. In that case, the height value is recorded at each location or continuously. The height value is then converted to a thickness value at each location or continuously. The shape and change of the wafer is calculated mathematically and the resulting value is displayed via a display device such as a computer screen.

  Another embodiment of the present invention is an apparatus for measuring the thickness, flatness, and local shape of a thin wafer. The wafer has a front surface and a back surface. The apparatus includes a first stage that calibrates the distance of the wafer from the first sensor and the second sensor. Distance calibration further includes placing a gauge block of known thickness between the first and second sensors, also in a suitable holder, and including multiple parallel surfaces. It is.

  By adjusting the first and second sensors in the Z plane, the surface to be detected is positioned at the center of the measurement range of the sensor. The apparatus includes a second stage for calibrating the height obtained by the sensor, which further includes a known angle between the first and second sensors, also in a suitable holder. A first placement operation for placing the angle gauge block is included, the flat surface of the angle gauge block being perpendicular to the first sensor, and a surface at a known angle relative to the first sensor is shown.

  The angle gauge block is then rotated 180 degrees, the slope of the angle gauge block is placed in the opposite direction to the first arrangement, the angle gauge block is perpendicular to the first sensor, and a surface at a known angle relative to the first sensor. Is shown. The height sensor calibration is mathematically assembled to calculate the tilt of the angle gauge block.

  The angle gauge block is then rotated and presented to the second sensor, and multiple steps as applied to the first sensor are repeated. The apparatus includes a third stage for measuring wafer local thickness. This thickness measurement further includes placing the wafer in a suitable holder so that the first and second sensors receive responses from both surfaces of the wafer.

  The wafer is then moved in the appropriate holder through a predetermined number of locations individually or along a continuous path. The height value is recorded at each location or continuously and is converted to a thickness value at each location or continuously. Next, the shape and shape change of the wafer are mathematically calculated, and the calculation result is displayed on a display device such as a computer screen.

  The present invention seeks to measure the thickness of a thin wafer, as well as the depth of a trench etched into the wafer. For the measurement, an optical needle that is considered to be a light needle is used. The needle has different color focusing points at different levels. Therefore, any part that reflects the light of the optical needle reflects only the color at the focal point. FIG. 8 shows an optical needle. This system therefore associates color with height. The optical needle thus scans a wafer with whatever density the user requires so that it can capture thousands and possibly tens of thousands of data points that can form a surface.

  A normal wafer, ie a wafer with sufficient mass to support its own weight, can be measured using two optical needles, one for the front and one for the back To do. The separate surfaces captured by the two optical needles are related to each other by a calibration operation. In this way, the system can detect the shape of the wafer and can calculate warpage, bends, and various other shape values of interest to the manufacturer. The manufacturer can now check whether the wafer has sufficient flatness to process.

  Thin wafers used in heat sensitive applications often cannot support their own weight when holding the sides. Such wafers are actually flexible, which means that they align with the surface on which the wafer is placed, assuming sufficient vacuum, but for thick wafers I don't think so. This is because the thick wafer maintains its shape.

  Thin wafers still need to be measured for proper thickness and shape. This is because the thickness is critical for heat treatment. In addition, the front and back surfaces must be parallel. Thus, for thin wafers, a flat vacuum chuck is needed to secure the wafer and the surface is measured only from above. The system measures the surface of the vacuum chuck by a calibration operation, and detects the surface shape, thickness, and parallelism of the wafer with reference to the surface of the vacuum chuck.

  As can be seen in FIG. 9, the steps involved in calibrating the distance from the sensor are the first to place a gauge block of known thickness and parallel surfaces between the sensors, also in a suitable holder. Includes placement work. Then, by adjusting the plurality of sensors in the Z plane, the detected surface is positioned at the center of the sensor measurement range.

  As seen in FIG. 10, the steps involved in calibrating the height obtained from the sensors include a first placement step in which an angle gauge block of known angle is placed between the sensors, also in a suitable holder. The flat surface is perpendicular to the sensor, and a surface at a known angle to the sensor is shown. The angle gauge block is then moved by a known distance, giving a known change in the distance to the sensor in the Z plane. This process is repeated over the entire measurement range of the sensor. The angle gauge block is then rotated 180 degrees, the ramp is placed in the reverse direction, and the process is repeated in the reverse direction. The data is fit mathematically accordingly and given a measurement resolution higher than the resolution obtained from the number of calibrated points, the tilt of the stage is calculated. Finally, the angle gauge block is rotated and presented to the second sensor, and this process is repeated.

  FIG. 11 shows the stage of measuring the thickness of the wafer. Initially, the wafer is placed in a suitable holder so that the sensor can receive responses from both surfaces of the wafer. The appropriate stage containing the wafer is then moved along a predetermined number of locations or along a predetermined continuous route. The height value from each sensor is recorded at each location or continuously. The height value at each location is converted to a wafer thickness value at each location. Next, calculation of necessary shapes and shape changes is performed, and the calculation results are displayed.

The thickness and flatness are measured using the apparatus shown in FIGS. The frame 18 holds the entire device. FIG. 15 shows each component in detail. The granite base 1 is the main support structure of the entire system. The vacuum chuck 2 is used to hold a wafer to be measured and fix the wafer in place.
The top height sensor 3 is an optical needle described in US Patent Application No. 006/0109483 by Grant and Marks. The sensor, together with the objective lens, measures the front surface of the wafer. The lower height sensor 4 is another optical needle sensor that measures the back surface of the wafer.

An objective lens 5 is provided for the upper height sensor, and an objective lens 6 is provided for the lower height sensor. The granite bridge 7 forms a granite support structure for holding the upper height sensor above the wafer. The bevel-shaped mounting portion 8 is used as a moving plate for mounting the upper height sensor to the power stage 10.
The clamp 9 clamps the drum-shaped attachment portion 8 and attaches it to the power stage 10. The power stage 10 is for making the upper height sensor 3 approach or separate from the wafer. The upper height sensor 3 is moved so that the front surface of the wafer is positioned within the measurement range of the sensor 3. After this movement, the distance between the upper and lower sensors is calibrated.

The granite columnar body 11 forms a support structure for the granite bridge 7. A 15.24 cm × 15.24 cm (6 ″ × 6 ″) powered stage 12 is used to move the wafer in a plane between the upper sensor 3 and the lower sensor 4. While the stage 12 translates the wafer, the thickness and flatness of the wafer are measured, and the measured values of the upper sensor 3 and the lower sensor 4 are recorded.
The stage bracket 13 is used as a support structure for the lower sensor 4. The XYZ stage 14 is for aligning the lower sensor 4. The microscope mounting bracket 15 moves a plate for mounting the lower sensor 4 to the XYZ stage 14. The fastening stud 16 fastens the bracket 17, and the bracket 17 holds the optical needle sensors 3 and 4, and the sensor is attached to the drum-shaped mounting portion 8 or the microscope mounting bracket 15. The frame 18 forms a support table for the granite base 1 and holds any necessary electronic accessories.

  Many optical systems that cannot measure the trench depth can still measure the protrusion height well. Unlike the case of the trench, the incident light to the protrusion is not partially blocked. FIG. 2 shows the difference in reflection method between when light is propagated into the trench and when it is propagated onto the protrusion. As can be seen from the figure, the reflected light from the protrusion has more data points than the reflected light from the trench, so that more accurate reading is possible. The reflected light from the top of the protrusion returns directly to the sensor without undergoing many reflections. In the case of the protrusion, the reflection characteristics are similar no matter how high, but in the case of the trench, unlike the case, if it is deeper, the more light is disturbed by the side walls, the less the reflected light.

  A rounded rough trench absorbs all incident light, whereas a rounded rough projection always reflects some light (see FIG. 6). In fact, in the case of protrusions, no matter how fine the vertices are, some light is always reflected from the top. All of these differences indicate that the protrusion measurement is easier than the trench measurement. When the etched trench is viewed from the back, it appears as a protrusion to the sensor, so its height (depth) can be measured much more easily.

  Most height measuring optical systems currently on the market usually work with visible light. However, silicon is opaque to visible light. FIG. 5 is a diagram showing the case where the present invention uses a visible spectrum for measuring the thickness of a glass slide. In the present invention, however, an infrared beam, particularly an infrared beam having a long wavelength of 950 nm or more that transmits silicon is used. Since silicon is the most common wafer substrate and is specifically described herein for silicon, the present invention can be used with wafers of many different materials, such as glass, SiC, InP, GaAs, and any other transparent material. In the following description, an infrared beam is assumed to be the light source, but any wavelength band that transmits the material of interest can be used in the present invention.

  Silicon is transparent to wavelengths exceeding 1.1 μm (see FIG. 4). Since the refractive index of silicon at 1.1 μm wavelength is about 3.5, the reflectivity of normal incident light at the air / silicon interface is about 31%. The remaining light is light that passes through the silicon. A percentage similar to the percentage of light transmitted is the reflectance at the back air / silicon interface. Thus, a single sensor located on one side of the silicon wafer can receive reflected light from both the front and back surfaces of the wafer. FIG. 3 shows the relative signal level received from the back of the silicon wafer after passing through the wafer twice (incident and reflected).

  As shown in FIG. 6, in this preferred embodiment, the wafer is irradiated from the opposite side (back side) of the etched trench side. Since the wafer is transparent, it is reflected back from the etched surface (front). Therefore, the non-contact type height sensor can measure the contour of the front surface provided with the trench using reflection from the front surface. When viewed from the back, the trench appears to be a protruding portion, so that there are no particular disadvantages when viewed as a trench, such as light absorption by the trench sidewalls.

  In the present invention, many different non-contact optical sensors can be used to measure protrusion height (trench depth), while the preferred embodiment uses a confocal color sensor as schematically shown in FIG. The Other sensors that can be used include an optical interferometer, a phase shift interferometer, a confocal scanning microscope, and a laser triangulation instrument. The preferred embodiment uses a confocal color sensor, particularly in the near infrared (NIR) region of the spectrum from 900 nm to 1700 nm, and measures the wafer thickness by simultaneously measuring the distance between the front and back surfaces. Designed to measure. A calibration method that relates the measurement interval between two reflected waves to the actual thickness of the wafer is not obvious and is an integral part of the present invention. As described above, even in a similar system, the trench depth can be measured by simultaneously measuring the distance between the front surface and the back surface.

Confocal color sensors are well known to those skilled in optical measurement technology, examples of which are given in US Patent Application No. 2006/0109483 by Marks and Grant, two of the inventors of the present invention. The technology of which is described and incorporated herein by reference. In this type of sensor, the focal points are dispersed along the axial direction according to the color or wavelength of the light (FIG. 7). Thus, each wavelength is focused at a different level, and if there is a single object, only one wavelength is focused on that object.
As a result, the system can correlate color with height if it can detect which wavelengths are focused on the object. This detection is performed using a spectrometer. A preferred embodiment of a complete confocal color system (FIG. 7) has a broadband and includes a white light source, a dispersion optics, a spectrometer, a fiber coupler / splitter for connection to other components, and a confocal source And a detector aperture. However, many other configurations are possible.

  The confocal color sensor can also be easily combined with a microscope imaging system that shares the objective lens with the confocal sensor. The image of the object is presented by the microscope, and the position of the confocal measurement point can be directly referred to. For example, the wafer thickness measurement after the solder protrusions are provided on the wafer can be performed by guiding measurement points around the solder protrusions using a microscope image. Another example is the measurement of a pressure sensor diaphragm. These diaphragms are etched from the back side of the wafer, and a silicon thin film is formed for each die. Measurement of these diaphragms is not possible with capacitance sensors due to the small sensor dimensions. However, a confocal color sensor combined with a microscope can easily locate the local spot of the diaphragm and measure the thickness at the local spot. The combined microscope is also important for trench measurements. For this application, it is necessary to locate the trench with a microscope so that the trench can be measured.

  If the object is transparent, as in the case of silicon for the NIR spectrum, the reflected light from both the front and back of the object is returned to the spectrometer. Since there is a difference in height from the front and back of the wafer, the reflected light from each side will have a different wavelength. FIG. 5 shows the relative intensity of the reflected light from the front and back of the glass slide in the case of the visible spectrum. Similar results have been obtained for silicon wafer measurements in the near infrared spectrum. In general, confocal color sensors are designed such that the shorter wavelength is focused before the longer wavelength, but it is also possible to design the sensor in the opposite relationship. In the preferred embodiment, the light focused on the front side of the wafer has a shorter wavelength than the light focused on the back side. When these reflected lights pass through the spectrometer, the spectrometer shows two wavelength peaks. The position of each peak indicates the position of each surface, and the peak difference indicates the thickness of the wafer.

  The sensor mechanically scans the wafer laterally at a sample rate and data density specified by the user. Thousands of data points, possibly tens of thousands, are captured. The sensor has a spot size of about 5 micrometers and makes highly localized surface and thickness measurements. As the sensor moves across the wafer, the position of the peak changes as a function of the wafer surface shape and the relative position of the peak changes as a function of thickness. Thus, the complete shape of the wafer is obtained and the local thickness of the wafer is obtained.

  Known confocal color sensors are not designed for NIR work. The design of a confocal color sensor requires the use of a dispersive material in order to disperse multiple wavelengths along the focal axis. Many different materials can be used to obtain sufficient dispersion in the visible spectrum, but it is more difficult to achieve the required dispersion when using NIR. However, the double lens can be designed to provide axial dispersion and correction of spherical aberration. In one example, S-TIH53 and N-SSK8 glass are used, and in another example, S-TIH53 and N-SK4 glass are used. Many other examples including a double lens using S-NPH1 instead of S-TIH53 can also be designed. Several such double lenses can also be used in a series to increase the amount of axial dispersion in the system. Another way to form an axial dispersive lens system is to use a diffractive lens in combination with a refractive lens.

Calibration is a critical task for the present invention. The complexity of the backside measurement is influenced by focusing light over the entire wafer substrate. One of those effects is the addition of spherical aberration. This spherical aberration effectively stretches the focal point on the back and reduces the axial resolution.
However, spherical aberration can be minimized by properly designing the focusing lens system. Another effect is that when light propagates through the medium, a difference between the physical length and the optical path length occurs. Although the optical path length is measured with a confocal color sensor, an important quantity here is the physical length or thickness of the wafer. The measured thickness is approximately equal to the physical thickness divided by the refractive index of the wafer. However, the refractive index varies with wavelength. In the case of silicon, the axial focusing position of each wavelength is approximately z (λ) = f (λ) n (λ), where F is the axial focusing position in air and n is the silicon focusing position. The refractive index, λ, indicates the dependence of each of these quantities on the wavelength.

FIG. 16 shows how the refractive index of the silicon medium affects the measurement calibration. In the confocal color sensor, since incident light is refracted on the first surface of silicon, the confocal color sensor is actually measured at one point inside the wafer. Both wafer thickness (thickness seen on the left side of FIG. 16) and trench depth (depth seen on the right side of FIG. 16) are affected.
FIG. 16 shows a trench measurement that further takes refraction into account. The “coarse” measurement is the apparent axial position of the confocal spot measured with a calibrated spectrometer, while the “target” measurement is the trench depth. Geometrical optical analysis using Snell's law shows that the “target” measurement is effectively equal to the “coarse” measurement determined by the refractive index. Therefore, when the spectrometer response is calibrated to the confocal response in air, measuring the trench depth of the substrate medium is straightforward. Alternatively, the spectrometer response can be calibrated directly using a step height gauge made of the same material as the substrate.

  The movement of the color focusing point in the axial direction generates a confocal signal, but this movement is affected by the refractive index of the medium. The standard method of calibrating a confocal color system in air is to scan with precise lateral motion using an angle gauge block. This calibration will result in a calibration table between object height and wavelength peak. However, this calibration table will be inaccurate in media other than air. A suitable calibration table can be estimated by multiplying the object height by the refractive index. However, the change in refractive index with wavelength must be taken into account. Alternatively, the calibration gauge can be made of silicon so that direct calibration can be performed.

  The drawings and examples described above are for illustrative purposes and are not intended to limit the scope of the claims. This disclosure is to be construed as illustrative of the principles of the invention and is not intended to limit the spirit and scope of the invention or to limit the claims to the illustrated embodiments. Those skilled in the art will be able to make modifications to the present invention for specific applications of the present invention.

The figure which shows the prior art for wafer thickness and trench depth measurement. a is a figure which shows the incident light and reflected light with respect to a trench, b is a figure which shows the incident light and reflected light with respect to a protrusion part. The figure which shows the reflection method of the light from both the front surface and back surface of a silicon wafer. The figure which shows the relationship between a wavelength and the intensity | strength loss of light. For each wavelength in the near infrared spectrum, the intensity of the reflected light from the back of the silicon wafer relative to the intensity of the incident light is shown. Note how silicon absorbs light with wavelengths below about 1.1 μm. The figure which shows how the axial space | interval of the front surface of a microscope glass slide and a back surface is measured simultaneously when a confocal color sensor works in a visible spectrum area | region at the time of a wafer thickness measurement experiment. The figure which shows the suitable Example of this invention. By seeing through the back of the wafer using an infrared beam, the trench depth can be effectively captured as a protrusion. 1 is a schematic diagram of a confocal color detection system for trench measurement. The figure which shows the optical needle | hook of this invention. The flowchart which shows the operation | work which calibrates the detection distance by a sensor at the time of the thickness measurement of a thin wafer. The flowchart which shows the operation | work which calibrates the detection height by a sensor at the time of the thickness measurement of a thin wafer. The flowchart which shows the wafer thickness measurement step at the time of the thickness measurement of a thin wafer. The side view of the apparatus which measures the thickness and flatness of a wafer. The front view of the apparatus which measures the thickness and flatness of a wafer. The elements on larger scale of the apparatus which measures the thickness and flatness of a wafer. Sectional drawing cut along the AA line of FIG. The figure which shows how the refractive index of a silicon medium influences measurement calibration.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Granite base 2 Vacuum chuck 3 Upper height sensor 4 Lower height sensor 5 Objective lens 6 Objective lens 7 Granite bridge 8 Bevel attachment part 9 Clamp 10 Power stage 11 Granite column 12 Power stage 13 Stage bracket 14 XYZ stage 15 Microscope mounting bracket 16 Tightening stud 17 Bracket 18 Frame

Claims (36)

In a method for measuring the trench depth of a wafer,
The operation includes locating the trench of the wafer with a microscope, the wafer has a front surface and a back surface, the trench is provided on the front surface, and
Including placing a non-contact optical instrument facing the back surface, whereby the trench can be viewed as a substantially protruding portion provided on the second surface from the non-contact optical instrument; The non-contact optical instrument uses a light source that can pass through the wafer and receives reflected light from the front and back surfaces of the wafer, so that the front surface indicates the inside of the trench. In addition,
Measuring the height of the protrusion using a set of light sources in a wavelength range that can be transmitted through the wafer;
Calibrating the protrusion height by conversion of the difference in measured height between the front and back surfaces, thereby detecting the shape and contour of the wafer at the trench location, A method of measuring trench depth. Wherein said wafer has the following groups: silicone configured, GaAs, GaAIAs, InP, Sic , from one of SiO _2, according to claim 1.   The method of claim 1, wherein the light source is in the near infrared region and has a wavelength range of 900 nm-1700 nm.   The method of claim 1, wherein the non-contact optical instrument is a confocal color sensor.   The confocal color sensor uses calibration means by which the collected data corresponding to the reflected light wavelength from the front surface and the back surface of the wafer and corresponding to the height difference is obtained from the wafer. 5. The method of claim 4, wherein the method is converted to a measured thickness that determines the depth of the trench or the thickness of the local area.   The method of claim 1, wherein a white light interferometer is used in the non-contact optical instrument.   The method of claim 1, wherein a phase shift interferometer is used in the non-contact optical instrument.   The method of claim 1, wherein a confocal scanning microscope is used for the non-contact optical instrument.   The method according to claim 1, wherein a laser triangulation instrument is used for the non-contact optical instrument.   The method of claim 1, wherein the non-contact optical instrument mechanically scans the wafer laterally at a pre-specified sample speed and density. In a method for measuring the local thickness of a thin wafer,
Including locating the local thickness region of the wafer with a microscope, the wafer having a front surface and a back surface;
An operation in which a non-contact optical instrument is disposed on one side of the wafer so as to simultaneously receive reflected light from the front surface and the back surface;
Measuring the thickness of the wafer using a set of light sources in a wavelength range that can be transmitted through the wafer;
The local wafer thickness is calibrated by conversion of the measured height difference between the front and back surfaces, whereby the local thickness of the wafer is detected. A method of measuring local thickness. The method of claim 11, wherein the wafer is composed of one of the following groups: silicon, GaAs, GaAIAs, InP, Sic, SiO ₂ .   The method of claim 11, wherein the light source is in the near infrared region and has a wavelength range of 900 nm-1700 nm.   The method of claim 11, wherein the non-contact optical instrument is a confocal color sensor.   The confocal color sensor uses calibration means by which the collected data corresponding to the reflected light wavelength from the front surface and the back surface of the wafer and corresponding to the height difference is stored in the wafer. 15. The method of claim 14, wherein the method is converted to a measured thickness that determines a trench depth or the thickness of the local area.   The method of claim 11, wherein a white light interferometer is used in the non-contact optical instrument.   The method of claim 11, wherein a phase-shifting interferometer is used in the non-contact optical instrument.   The method according to claim 11, wherein a confocal scanning microscope is used for the non-contact optical instrument.   The method according to claim 11, wherein a laser triangulation instrument is used for the non-contact optical instrument.   The method of claim 11, wherein the non-contact optical instrument mechanically scans the wafer laterally at a pre-specified sample speed and density. An apparatus for measuring a trench depth of a wafer or a local thickness of the wafer, wherein the wafer has a front surface and a back surface, wherein the device comprises:
A non-contact optical instrument for height measurement,
Fixing means for positioning the wafer, arranged substantially below the non-contact optical instrument;
A set of light sources in a wavelength range capable of transmitting the wafer,
The non-contact optical instrument is disposed on one side of the wafer to simultaneously receive reflected light from the front and back surfaces, and the apparatus further comprises:
An apparatus for measuring a trench depth of a wafer or a local thickness of the wafer, comprising data collection means for receiving data from a non-contact optical instrument for measuring the height. The wafer is next group, i.e. silicon, GaAs, GaAIAs, InP, Sic , constructed from one of SiO _2, An apparatus according to claim 21.   The apparatus of claim 21, wherein the light source is in the near infrared region and has a wavelength range of 900 nm-1700 nm.   The apparatus of claim 21, wherein the non-contact optical instrument is a confocal color sensor.   The confocal color sensor uses calibration means by which the collected data corresponding to the reflected light wavelength from the front surface and the back surface of the wafer and corresponding to the height difference is stored in the wafer. 25. The apparatus of claim 24, converted to a measured thickness that determines a trench depth or the thickness of the local area.   The apparatus of claim 21, wherein a white light interferometer is used in the non-contact optical instrument.   The apparatus of claim 21, wherein a phase-shifting interferometer is used in the non-contact optical instrument.   The apparatus according to claim 21, wherein a confocal scanning microscope is used in the non-contact optical instrument.   The apparatus according to claim 21, wherein a laser triangulation instrument is used for the non-contact optical instrument.   The method of claim 21, wherein the non-contact optical instrument mechanically scans the wafer laterally at a pre-specified sample speed and density. A method for measuring the thickness, flatness, and local shape of a thin wafer, wherein the wafer has a front surface and a back surface, the method comprising:
Calibrating the distance of the wafer from the first sensor and the second sensor, the calibrating distance further comprising:
Placing a gauge block of known thickness between the first and second sensors, also in a suitable holder, and including parallel surfaces;
Adjusting the first and second sensors in the Z plane so that the surface to be detected is located in the center of the sensor measurement range, and the method includes:
Including measuring the local thickness of the wafer, the measuring operation further comprising:
Placing the wafer in a suitable holder so that the first and second sensors can receive responses from both surfaces of the wafer;
Working the wafer to move a predetermined number of locations in the appropriate holder to each individual or in a continuous path;
Working to record the height value at each of the locations or continuously;
Converting the height value to a thickness value at each of the locations or continuously; and
A mathematical calculation of the wafer shape and shape change;
A method for measuring the thickness, flatness, and local shape of a thin wafer, the method including displaying the mathematical calculation of the shape and shape change via a display means. Further calibration work is added, the calibration work is further
Calibrating the height obtained by the sensor, the height calibration operation further comprising:
A first placement operation is included between the first and second sensors, which also places an angle gauge block of known angle in a suitable holder, the flat surface of the angle gauge block being in contact with the first sensor. Is shown at a right angle, a surface of a known angle with respect to the first sensor is shown, and
And rotating the angle gauge block by 180 degrees to dispose the inclined portion of the angle gauge block in a direction opposite to the first arrangement, the angle gauge block being perpendicular to the first sensor, and A surface with a known angle for one sensor is shown, and
Calculating the slope of the angle gauge block by mathematically transforming the calibration of the height collected by the sensor;
The method of claim 13, wherein the working steps are repeated as the angle gauge block is rotated and presented to the second sensor and applied to the first sensor.   32. The method of claim 31, wherein the sensor is a confocal color sensor. An apparatus for measuring the thickness, flatness, and local shape of a thin wafer, wherein the wafer has a front surface and a back surface.
A first stage for calibrating the distance of the wafer from the first sensor and the second sensor;
Placing a gauge block of known thickness between the first and second sensors, also in a suitable holder, and including parallel surfaces;
Adjusting the first and second sensors in the Z plane so that the surface to be detected is located in the center of the measurement range of the sensor, and the device also includes:
A second stage for measuring the local thickness of the wafer;
Placing the wafer in a suitable holder so that the first and second sensors can receive responses from both surfaces of the wafer;
An operation of moving the wafer individually or continuously in a predetermined number of locations in the appropriate holder;
The operation of recording the height value at each of the locations or continuously,
Converting the height value into a thickness value at each of the locations or continuously, and
Mathematically calculating the shape and shape change of the wafer;
An apparatus for measuring the thickness, flatness, and local shape of a thin wafer, including the operation of displaying the mathematical calculation of the shape and shape change of the wafer by display means. A third stage is provided for calibrating the height obtained by the sensor, the calibration of the height obtained by the sensor further comprises:
A second stage for calibrating the height obtained by the sensor is included, and the calibration of the height obtained by the sensor further comprises:
A first placement operation is included between the first and second sensors, which also places an angle gauge block of known angle, in a suitable holder, and the flat surface of the angle gauge block is in the first sensor. A surface at a known angle relative to the first sensor is shown,
Then, the angle gauge block is rotated by 180 degrees, and the angle gauge block is inclined in a direction opposite to the first arrangement, and the angle gauge block is perpendicular to the first sensor, A surface at a known angle to the first sensor is shown;
Calculating the slope of the angle gauge block by mathematically transforming the height calibration collected by the sensor;
35. The apparatus of claim 34, comprising rotating the angle gauge block to be presented to the second sensor and applying multiple work steps as applied to the first sensor.   The apparatus of claim 34, wherein the sensor is a confocal color sensor.