TWI769172B - Methods of generating three-dimensional (3-d) information of a sample using an optical microscope - Google Patents

Abstract

A method of generating 3D information including: varying the distance between the sample and an objective lens of the optical microscope at pre-determined steps; capturing an image at each pre-determined step; determining a characteristic value of each pixel in each captured image; determining, for each captured image, the greatest characteristic value across a first portion of pixels in the captured image; comparing the greatest characteristic value for each captured image to determine if a surface of the sample is present at each pre- determined step; determining a first captured image that is focused on an apex of a bump of the sample; determining a second captured image that is focused on a first surface of the sample based on the characteristic value of each pixel in each captured image; and determining a first distance between the apex of the bump and the first surface.

Description

Method for generating three-dimensional (3-D) information of a sample using an optical microscope

The described embodiments are generally concerned with measuring three-dimensional information of a sample and more specifically about automatically measuring three-dimensional information in a fast and reliable manner.

Three-dimensional (3-D) measurements of various objects or samples are useful in many different applications. One such application is during wafer level packaging processing. Three-dimensional metrology information of a wafer during different steps of wafer-level manufacturing can provide insight into the presence of wafer handling defects that may exist on the wafer. Three-dimensional metrology information of wafers during wafer-level manufacturing can provide insight into the absence of defects before expending additional capital to continue wafer processing. Three-dimensional measurements of a sample are currently collected by human manipulation of a microscope. A human user focuses the microscope using their eyes to determine when the microscope is focused on one of the surfaces of the sample. An improved method of collecting 3D measurement information is needed.

In a first novel aspect, an optical microscope is used to generate three-dimensional (3-D) information of a sample by changing the distance between the sample and an objective lens of the optical microscope in predetermined steps; Capture an image at a predetermined step; determine a characteristic value of each pixel in each captured image; determine, for each captured image, a maximum characteristic value across all pixels in the captured image; compare each captured image The maximum characteristic value of the image is taken to determine whether a surface of the sample exists at each predetermined step; the focus is determined on the sample based on the characteristic value of each pixel in each captured image a first captured image on a first surface; determining a second captured image focused on a second surface of the sample based on the characteristic value of each pixel in each captured image; and A first distance between the first surface and the second surface is determined.

In a second novel aspect, a three-dimensional (3-D) measurement system includes: determining a thickness of a translucent layer of a sample; and determining a thickness of a metal layer of the sample, wherein the thickness of the metal layer is equal to the difference between the thickness of the translucent layer and the first distance, wherein the first surface is a top surface of a photoresist layer, and wherein the second surface is a top surface of a metal layer.

In a third novel aspect, an optical microscope is used to generate three-dimensional (3-D) information of a sample by changing the distance between the sample and an objective lens of the optical microscope in predetermined steps; capturing an image at a predetermined step; determining a characteristic value of each pixel in each captured image; determining a maximum characteristic value across a first portion of pixels in the captured image for each captured image; comparing the maximum characteristic value of each captured image to determine whether a surface of the sample exists at each predetermined step; determine a first captured image focused on a vertex of a bump of the sample; based on each captured image The characteristic value of each pixel in the captured image determines a second captured image focused on a first surface of the sample; and determines a first between the vertex of the bump and the first surface distance.

In a fourth novel aspect, a maximum characteristic value is determined for each x-y pixel location within a second portion of x-y pixel locations across all of the captured images, wherein the second portion of x-y pixel locations is included in each at least some of the x-y pixel locations included in the captured image; determining a subset of the captured image in which only the captured image that includes a maximum characteristic value for an x-y pixel location is included in the subset; and determining Among all the captured images within the subset of captured images, the first captured image is focused at a highest z position compared to all other captured images within the subset of captured images.

Further details and examples and techniques are described in the description below. SUMMARY OF THE INVENTION It does not define the invention. The present invention is defined by the scope of the invention application patent.

1: Semi-automatic 3D metrology system

2: Stage

3: Wafer

4: Computer

5: On/Off button

10: 3D Imaging Microscopy

11: Adjustable objective lens

12: Adjustable stage

20: 3D Metrology System

21: 3D Microscopy

22: Sample handler/stage

23: Computer

24: Processor

25: Storage device

26: Network Devices

27: Display

28: Input device

29: Internet

30: Silicon substrate

31: photoresist layer

40: Boundary of top surface opening

41: Best Fit Line

42: Boundary of bottom surface opening

43: Best Fit Line

200: Flowchart

201: Steps

202: Steps

203: Steps

204: Steps

205: Steps

300: Flowchart

301: Steps

302: Step

303: Step

304: Step

305: Steps

311: Steps

312: Steps

313: Steps

314: Steps

315: Steps

321: Steps

322: Steps

323: Steps

324: Steps

The accompanying drawings, wherein like numerals refer to like components, illustrate embodiments of the present invention.

FIG. 1 is a diagram of a semi-automatic three-dimensional metrology system 1 that performs automated three-dimensional measurement of a sample.

FIG. 2 is a diagram of a three-dimensional imaging microscope 10 including an adjustable objective lens 11 and an adjustable stage 12 .

FIG. 3 is a diagram of a three-dimensional metrology system 20 including a three-dimensional microscope, a sample handler, a computer, a display, and input devices.

FIG. 4 is a diagram illustrating one method of capturing images when changing the distance between the objective lens and the stage of the optical microscope.

5 is a graph showing the distance between the objective lens of an optical microscope and the sample surface, where each x-y coordinate has a maximum characteristic value.

FIG. 6 is a three-dimensional plot of an image rendered using the maximum characteristic value of each x-y coordinate shown in FIG. 5 .

7 is a diagram illustrating a peak mode operation using images captured at various distances.

8 is a graph showing peak mode operation using images captured at various distances when a photoresist opening is within the field of view of an optical microscope.

9 is a graph showing three-dimensional information from peak mode operation.

10 is a diagram illustrating a summation mode operation using images captured at various distances.

FIG. 11 is a diagram showing false surface detection when operating using summing mode.

FIG. 12 is a graph showing three-dimensional information from sum mode operation.

13 is a diagram illustrating a range mode operation using images captured at various distances.

14 is a graph showing three-dimensional information from range mode operation.

Figure 15 shows only a graph of pixel counts with a characteristic value within a first range.

Figure 16 shows only a graph of pixel counts with a characteristic value in a second range.

Figure 17 is a flow diagram showing one of the various steps involved in peak mode operation.

FIG. 18 is a flow diagram showing one of the various steps involved in range mode operation.

Figure 19 is a view of a captured image (including a single feature) focused on the top surface of a photoresist layer.

FIG. 20 is a diagram illustrating a first method of generating an intensity threshold.

Figure 21 is a diagram showing a second method of generating an intensity threshold.

FIG. 22 is a diagram illustrating a third method of generating an intensity threshold.

Figure 23 is a three-dimensional view of a photoresist opening in a sample.

24 is a two-dimensional view of the top surface opening of the photoresist shown in FIG. 23. FIG.

25 is a two-dimensional view of the bottom surface opening of the photoresist shown in FIG. 23. FIG.

Figure 26 is a captured image focused on a top surface of a photoresist layer.

FIG. 27 is a diagram showing detection of a boundary of the photoresist layer shown in FIG. 26. FIG.

Figure 28 focuses on a captured image on a bottom surface of a photoresist layer.

FIG. 29 is a diagram showing detection of a boundary of the photoresist layer shown in FIG. 28. FIG.

Figure 30 focuses on a captured image on a top surface of a photoresist layer in a trench structure.

FIG. 31 is a diagram showing detection of a boundary of the photoresist layer shown in FIG. 30. FIG.

Figure 32 is a three-dimensional view of a photoresist opening partially filled with metallization.

33 is a cross-sectional view of a photoresist opening partially filled with metallization.

Figure 34 is a three-dimensional view of a photoresist opening with metallization.

Figure 35 is a cross-sectional view of a photoresist opening with metallization.

Figure 36 is a three-dimensional view of a metal pillar above the passivation layer.

Figure 37 is a cross-sectional view of a metal post above the passivation layer.

Figure 38 is a three-dimensional view of the metal over the passivation layer.

Figure 39 is a cross-sectional view of the metal over the passivation layer.

Figure 40 is a cross-sectional view showing a measurement of a translucent material proximate to a metallized surface.

41 is a graph showing peak mode operation using images captured at various distances when a photoresist opening is within the field of view of an optical microscope.

FIG. 42 is a graph showing three-dimensional information derived from the peak mode operation shown in FIG. 41 .

43 is a view of a captured image, including an outline of a first analysis area A and a second analysis area B, focused on a top surface of a photoresist layer in a trench structure.

Figure 44 is a three-dimensional view of a bump above the passivation structure.

45 is a top view of the bump above the passivation structure, including an outline of a first analysis area A and a second analysis area B. FIG.

46 is a top view of the adjusted analysis area A and the analysis area B when the entire bump is not positioned in the original analysis area A. FIG.

Figure 47 is a cross-sectional view of the bump above the passivation structure.

Figure 48 is a graph showing peak mode operation using images captured at various distances when only one photoresist layer is within region B of the optical microscope's field of view.

FIG. 49 is a graph showing three-dimensional information derived from the peak mode operation of FIG. 48. FIG.

Cross-references to related applications

This application is a continuation-in-part of non-provisional U.S. Patent Application Serial No. 15/338,838, filed on October 31, 2016, entitled "OPTICAL MEASUREMENT OF OPENING DIMENSIONS IN A WAFER," and is claimed under 35 U.S.C. § 120 priority. The entire disclosure of this case is incorporated herein by reference. Application 15/338,838 is a continuation-in-part of non-provisional U.S. patent application Ser. priority. The entire disclosure of this case is incorporated herein by reference.

Reference will now be made in detail to background examples and some embodiments of the present invention, examples of which are illustrated in the accompanying drawings. In the following description and the scope of the patent application, relational terms such as "top", "below", "top", "bottom", "top", "bottom", "left" and "right" may be used to describe the described relative orientation between the different parts of the structure, and it should be understood that the overall structure described may be oriented in three-dimensional space in virtually any manner.

FIG. 1 is a diagram of a semi-automatic three-dimensional metrology system 1 . The semi-automatic three-dimensional metrology system 1 includes an optical microscope (not shown), an on/off button 5 , a computer 4 and a stage 2 . In operation, a wafer 3 is placed on the stage 2 . The function of the semi-automatic three-dimensional metrology system 1 is to capture a plurality of images of an object and automatically generate three-dimensional information describing various surfaces of the object. This is also known as a "scan" of an object. Wafer 3 is an example of an object analyzed by semi-automated three-dimensional metrology system 1 . An object can also be called a sample. In operation, the wafer 3 is placed on the stage 2 and the semi-automated three-dimensional metrology system 1 begins the process of automatically generating three-dimensional information describing the surface of the wafer 3 . In one example, the semi-automated three-dimensional metrology system 1 begins by pressing a connection to a computer 4 A designated key on a keyboard (not shown). In another example, the semi-automated three-dimensional metrology system 1 begins by sending an initial command to the computer 4 across a network (not shown). The semi-automated 3D metrology system 1 can also be configured to interface with a semi-automated wafer handling system (not shown) that removes a wafer after completing a scan of the wafer and inserts a new one The wafer is scanned.

A fully automated 3D metrology system (not shown) is similar to the semi-automated 3D metrology system of FIG. 1; however, a fully automated 3D metrology system also includes a robotic handler that automatically picks up a wafer without human intervention And place the wafer on the stage. In a similar fashion, a fully automated 3D metrology system can also use a robotic handler to automatically pick up a wafer from the stage and remove the wafer from the stage. A fully automated three-dimensional metrology system is desirable during the production of many wafers because it avoids possible contamination by a human operator and improves time efficiency and overall cost. Alternatively, a semi-automated three-dimensional metrology system 1 may be desired during research and development activities when only a small number of wafers need to be measured.

FIG. 2 is a diagram of a three-dimensional imaging microscope 10 including a plurality of adjustable objective lenses 11 and an adjustable stage 12 . The three-dimensional imaging microscope can be a confocal microscope, a structured illumination microscope, an interferometric microscope, or any other type of microscope well known in the art. A confocal microscope will measure the intensity. A structured illumination microscope will measure the contrast of a projected structure. An interferometer microscope will measure the fringe contrast.

In operation, a wafer is placed on the adjustable stage 12 and an objective lens is selected. The three-dimensional imaging microscope 10 captures multiple images of the wafer while adjusting the height of the stage on which the wafer rests. This results in multiple images of the wafer being captured when the wafer is positioned at various distances away from the selected lens. In an alternative example, the wafer is placed on a fixed stage and the position of the objective is adjusted, thereby changing the distance between the objective and the sample without moving the stage. in another real For example, the stage can be adjusted in the x-y direction and the objective lens can be adjusted in the z direction.

The captured images may be stored locally in a memory included in the three-dimensional imaging microscope 10 . Alternatively, the captured images may be stored in a data storage device included in a computer system, where the three-dimensional imaging microscope 10 communicates the captured images to the computer system across a data communication link. Examples of a data communication link include: a Universal Serial Bus (USB) interface, an Ethernet connection, a FireWire bus interface, a wireless network (such as WiFi).

FIG. 3 is a diagram of a three-dimensional metrology system 20 including a three-dimensional microscope 21 , a sample handler 22 , a computer 23 , a display 27 (optional) and an input device 28 . The three-dimensional metrology system 20 is an example of one of the systems included in the semi-automated three-dimensional metrology system 1 . The computer 23 includes a processor 24, a storage device 25 and a network device 26 (optional). The computer outputs information to a user via the display 27 . If the display 27 is a touch screen device, the display can also be used as an input device. The input device 28 may include a keyboard and a mouse. The computer 23 controls the operation of the three-dimensional microscope 21 and the sample handler/stage 22 . When a start scan command is received by computer 23, the computer sends one or more commands to configure the three-dimensional microscope for image capture ("microscope control data"). For example, the correct objective needs to be selected, the resolution of the image to be captured needs to be selected, and the mode in which the captured image is stored needs to be selected. When an initial scan command is received by computer 23, the computer sends one or more commands to configure sample handler/stage 22 ("handler control data"). For example, the correct height (z direction) adjustment needs to be selected and the correct horizontal (x-y direction) alignment needs to be selected.

During operation, computer 23 causes sample handler/stage 22 to be adjusted into position. Once the sample handler/stage 22 is properly positioned, the computer 23 will cause the three-dimensional microscope to focus on a focal plane and capture at least one image. Next, the computer 23 will cause the stage to move in the z-direction so that the distance between the sample and the objective of the optical microscope is changed. Once the stage is moved to the new position, the computer 23 will cause the optical microscope to capture a second image. This procedure continues until the light An image is captured at each desired distance between the objective lens of the microscope and the sample. The images captured at each distance are transferred from the three-dimensional microscope 21 to the computer 23 ("image data"). The captured images are stored in the storage device 25 included in the computer 23 . In one example, computer 23 analyzes the captured images and outputs three-dimensional information to display 27 . In another example, the computer 23 analyzes the captured images and outputs the 3D information to a remote device via the network 29 . In yet another example, the computer 23 does not analyze the captured images, but sends the captured images to another device via the network 29 for processing. The three-dimensional information may include rendering a three-dimensional image based on the captured image. The 3D information may not contain any images, but rather data based on various characteristics of each captured image.

FIG. 4 is a diagram illustrating one method of capturing an image while changing the distance between the objective lens of the optical microscope and the sample. In the embodiment shown in FIG. 4, each image includes 1000 by 1000 pixels. In other embodiments, the image may include various pixel configurations. In one example, the interval between successive distances is fixed to a predetermined amount. In another example, the interval between successive distances may not be fixed. This variable spacing between images in the z-direction can be beneficial if only a portion of the z-direction scan of the sample requires additional z-direction resolution. The z-direction resolution is based on the number of images captured per unit length in the z-direction, so capturing additional images per unit-length in the z-direction will increase the measured z-direction resolution. Conversely, capturing fewer images per unit length in the z-direction will reduce the measured z-direction resolution.

As discussed above, the optical microscope is first adjusted to focus on a focal plane positioned at a distance of 1 from one of the objective lenses of the optical microscope. Next, the optical microscope captures an image, which is stored in a storage device (ie, "memory"). Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is distance 2. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 3. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 4. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 5. Next, the optical microscope captures an image, and the image is stored in the storage device. The procedure continues for N different distances between the objective of the optical microscope and the sample. Information indicating which image is associated with each distance is also stored in the storage device for processing.

In an alternative embodiment, the distance between the objective of the optical microscope and the sample is fixed. In fact, the optical microscope includes a zoom lens, which allows the optical microscope to change the focal plane of the optical microscope. In this way, when the stage and the sample supported by the stage are fixed, the focal plane of the optical microscope varies across N different focal planes. An image is captured for each focal plane and stored in a storage device. Next, the captured images across all of the various focal planes are processed to determine three-dimensional information for the sample. This embodiment requires a zoom lens that provides sufficient resolution across all focal planes and introduces minimal image distortion. Additionally, calibration between zoom positions and the resulting focal length of the zoom lens is required.

FIG. 5 is a graph showing the distance between the objective lens of an optical microscope and the sample, where each x-y coordinate has a maximum characteristic value. Once images are captured and stored for each distance, the characteristics of each pixel of each image can be analyzed. For example, the light intensity of each pixel of each image can be analyzed. In another example, the contrast of each pixel of each image can be analyzed. In yet another example, the fringe contrast of each pixel of each image can be analyzed. The contrast of a pixel can be determined by comparing the intensity of a pixel with the intensities of a predetermined number of surrounding pixels. For an additional description of how contrast information is generated, see US Patent Application Serial No. 12/699,824, entitled "3-D Optical Microscope," filed February 3, 2010 by James Jianguo Xu et al. is incorporated herein by reference).

FIG. 6 is a 3D plot representing a 3D image using the maximum characteristic value of each x-y coordinate shown in FIG. 5 . All pixels with an X position between 1 and 19 have a maximum characteristic value at distance 7 in the z direction. All pixels with an X position between 20 and 29 have a maximum characteristic value at distance 2 in the z direction. All pixels with an X position between 30 and 49 have a maximum characteristic value at distance 7 in the z direction. All pixels with an X position between 50 and 59 have a maximum characteristic value at distance 2 in the z direction. All pixels with an X position between 60 and 79 have a maximum characteristic value at distance 7 in the z direction. In this way, the three-dimensional image depicted in FIG. 6 can be generated using the maximum characteristic value per x-y pixel across all captured images. In addition, when distance 2 is known and distance 7 is known, the well depth shown in FIG. 6 can be calculated by subtracting distance 7 from distance 2.

Peak Mode Operation

7 is a diagram illustrating a peak mode operation using images captured at various distances. As discussed above with respect to Figure 4, the optical microscope is first adjusted to focus on a plane positioned at a distance 1 from one of the objective lenses of the optical microscope. Next, the optical microscope captures an image, which is stored in a storage device (ie, "memory"). Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is distance 2. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 3. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 4. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 5. Next, the optical microscope captures an image, and the image is stored in the storage device. The procedure continues for N different distances between the objective of the optical microscope and the stage. Indicates which image and each distance The associated information is also stored in the storage device for processing.

Determine the maximum characteristic value of all x-y positions in a single captured image across a z distance in peak mode operation, rather than determining the maximum value of each x-y position across all captured images at various z distances characteristic value. In other words, for each captured image, the largest characteristic value across all pixels included in the captured image is selected. As shown in Figure 7, the pixel position with the largest characteristic value will likely vary between different captured images. Properties can be intensity, contrast, or fringe contrast.

8 is a graph showing peak mode operation using images captured at various distances when a photoresist (PR) opening is within the field of view of an optical microscope. The top view of the article shows the cross-sectional area of the PR opening in the x-y plane. The PR opening also has a depth of one of the specified depths in the z-direction. The top view in Figure 8 below shows images captured at various distances. At distance 1, the optical microscope is not focused on the top surface of the wafer or the bottom surface of the PR opening. At distance 2, the optical microscope was focused on the bottom surface of the PR opening, but not on the top surface of the wafer. This results in an increased characteristic value (intensity/contrast/fringe) for one of the pixels receiving light reflected from the bottom surface of the PR opening compared to the pixel receiving light reflected from the other surface (top surface of the wafer) that is out of focus contrast). At distance 3, the optical microscope is not focused on the top surface of the wafer or the bottom surface of the PR opening. Therefore, at distance 3, the maximum characteristic value will be substantially lower than the characteristic value measured at distance 2. At distance 4, the optical microscope was not focused on any surface of the sample; however, due to the difference between the refractive index of air and that of the photoresist layer, the maximum characteristic value (intensity/contrast/fringe contrast) was measured an increase. Figure 11 and accompanying text describe this phenomenon in more detail. At distance 6, the optical microscope was focused on the top surface of the wafer, but not on the bottom surface of the PR opening. This results in an increased characteristic value (intensity/contrast/fringe) for one of the pixels receiving light reflected from the top surface of the wafer compared to pixels receiving light reflected from the other surface that is out of focus (the bottom surface of the PR opening). contrast). one Once the maximum characteristic value from each captured image is determined, the results can be used to determine at which distances a surface of the wafer is located.

9 is a graph showing three-dimensional information from peak mode operation. As discussed with respect to FIG. 8 , the maximum characteristic values of the images captured at distances 1, 3, and 5 have a maximum characteristic value that is less than the maximum characteristic value of the images captured at distances 2, 4, and 6. The curve of the maximum characteristic value at various z-distances may contain noise due to environmental effects such as vibration. To minimize this noise, a standard smoothing method, such as Gaussian filtering with a certain kernel size, can be applied before further data analysis.

One method of comparing maximum characteristic values is performed by a peak finding algorithm. In one example, a derivative method is used to locate the zero crossings along the z-axis to determine the distance at which each "peak" exists. Next, the maximum characteristic value at each distance where a peak is found is compared to determine the distance at which the maximum characteristic value is measured. In the case of Figure 9, a peak will be found at distance 2, which serves as an indication that a surface of the wafer is located at distance 2.

Another method of comparing maximum characteristic values is performed by comparing each maximum characteristic value with a predetermined threshold value. Threshold values can be calculated based on wafer material, distance and optical microscope specifications. Alternatively, threshold values can be determined by empirical testing prior to automated processing. In either case, the maximum characteristic value of each captured image is compared to a threshold value. If the maximum characteristic value is greater than the threshold value, it is determined that the maximum characteristic value indicates the presence of a surface of the wafer. If the maximum characteristic value is not greater than the threshold value, it is determined that the maximum characteristic value does not indicate a surface of the wafer.

Summation mode operation

10 is a diagram illustrating a summation mode operation using images captured at various distances. As discussed above with respect to Figure 4, the optical microscope is first adjusted to focus on a plane positioned at a distance 1 from one of the objective lenses of the optical microscope. Next, an optical microscope captures an image, The image is stored in a storage device (ie, "memory"). Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is distance 2. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 3. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 4. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 5. Next, the optical microscope captures an image, and the image is stored in the storage device. The procedure continues for N different distances between the objective of the optical microscope and the sample. Information indicating which image is associated with each distance is also stored in the storage device for processing.

Instead of determining the maximum characteristic value for all x-y positions in a single captured image across a z-distance, the characteristic values for all x-y positions of each captured image are summed together. In other words, for each captured image, the characteristic values of all the pixels included in the captured image are summed together. Properties can be intensity, contrast, or fringe contrast. An aggregated property value that is substantially greater than the average aggregated property value of adjacent z-distances indicates the presence of a surface of the wafer at that distance. However, this approach can also lead to false positives as described in FIG. 11 .

FIG. 11 is a diagram showing false surface detection when operating using summing mode. The wafer shown in FIG. 11 includes a silicon substrate 30 and a photoresist layer 31 deposited on top of the silicon substrate 30 . The top surface of the silicon substrate 30 is positioned at a distance of 2. The top surface of photoresist layer 31 is positioned at distance 6. An image captured at distance 2 will result in a sum of characteristic values that is substantially greater than the other images captured at a distance where a surface of the wafer is not present. The image captured at distance 6 will result in a sum of characteristic values that is substantially greater than the other images captured at a distance where a surface of the wafer is not present. At this point, the summing mode operation appears to be a valid indicator that there is a surface on one of the wafers. However, in An image captured at a distance of 4 will result in a sum of characteristic values that is substantially greater than the other images captured at a distance where a surface of the wafer is not present. This is a problem because, as clearly shown in Figure 11, one of the surfaces of the wafer is not positioned at distance 4. In reality, the increase in the sum of the characteristic values at distance 4 is an artifact located on the surface at distances 2 and 6. A major portion of the light irradiating the photoresist layer is not reflected, but travels into the photoresist layer. The angle at which this light travels changes due to the difference in refractive index of air and photoresist. The new angle is closer to normal than the angle of light irradiating the top surface of the photoresist. The light travels to the top surface of the silicon substrate below the photoresist layer. Next, the light is reflected by the highly reflective silicon substrate layer. As the reflected light leaves the photoresist layer and enters the air, the angle of the reflected light changes again due to the difference in refractive index between the air and the photoresist layer. This first redirection, reflection and second redirection of the irradiated light causes an increase in one of the characteristic values (intensity/contrast/fringe contrast) at distance 4 observed by the optical microscope. This example shows that whenever a sample contains a transparent material, the sum mode operation will detect surfaces that are not present on the sample.

FIG. 12 is a graph showing three-dimensional information from sum mode operation. This graph shows the results of the phenomenon depicted in FIG. 11 . A large value of the summed property value at distance 4 erroneously indicates the presence of a surface at distance 4. There is a need for a method that does not result in a false positive indication of the presence of the surface of the wafer.

Range Mode Operation

13 is a diagram illustrating a range mode operation using images captured at various distances. As discussed above with respect to Figure 4, the optical microscope is first adjusted to focus on a plane positioned at a distance 1 from one of the objective lenses of the optical microscope. Next, the optical microscope captures an image, which is stored in a storage device (ie, "memory"). Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is distance 2. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 3. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 4. Next, the optical microscope captures an image, and the image is stored in the storage device. Next, adjust the stage so that the distance between the objective lens of the optical microscope and the sample is the distance 5. Next, the optical microscope captures an image, and the image is stored in the storage device. The procedure continues for N different distances between the objective of the optical microscope and the sample. Information indicating which image is associated with each distance is also stored in the storage device for processing.

Determining a count of pixels with a characteristic value within a specific range contained in a single captured image at a z-distance, rather than determining all characteristic values across all x-y positions in the single captured image the sum. In other words, for each captured image, a count of pixels having a characteristic value within a specific range is determined. Properties can be intensity, contrast, or fringe contrast. A pixel count at a particular z-distance that is substantially greater than the average pixel count at adjacent z-distances indicates that a surface of the wafer is present at that distance. This method reduces the false positives described in Figure 11.

14 is a graph showing three-dimensional information from range mode operation. With knowledge of the different material types present on the wafer and the optical microscope configuration, an expected range of property values can be determined for each material type. For example, the photoresist layer will reflect a relatively small amount of light (ie, 4%) that irradiates the top surface of the photoresist layer. The silicon layer will reflect light that irradiates the top surface of the silicon layer (ie, 37%). The re-directed reflection (ie, 21%) from the top surface of the photoresist layer observed at distance 4 will be substantially larger than the reflection observed at distance 6; however, the observed reflection at distance 4 from the silicon substrate The re-directed reflection of the top surface (ie, 21%) will be substantially smaller than the reflection observed at distance 2. Therefore, when looking for the top surface of the photoresist layer, a first range centered on the expected characteristic value of the photoresist can be used to filter out pixels with characteristic values outside the first range, thereby filtering out pixels having characteristic values that do not originate from Pixels with characteristic values of reflection from the top surface of the photoresist layer. 15 shows that by applying the first characteristic value range Count of pixels across all distances resulting from the perimeter. As shown in Figure 15, some but not necessarily all pixels from other distances (surfaces) are filtered out by applying the first range. This occurs when characteristic values measured at multiple distances fall within a first range. However, applying the first range before counting pixels still serves to make pixel counts at the desired surface more prominent than other pixel counts at other distances. This is illustrated in FIG. 15 . After applying the first range, the pixel count at distance 6 is greater than the pixel count at distances 2 and 4, while before applying the first range, the pixel count at distance 6 is smaller than the pixel count at distances 2 and 4 (as shown in Fig. 14).

In a similar manner, when looking for the top surface of the silicon substrate layer, a second range centered on the expected characteristic value of the silicon substrate layer can be used to filter out pixels with characteristic values outside the second range, thereby filtering out pixels with characteristic values outside the second range. Pixels with characteristic values that are not derived from reflections from the top surface of the silicon substrate layer. Pixel counts across all distances produced by applying the second characteristic value range are shown in FIG. 16 . This range application reduces false indications that a wafer surface is located at distance 4 by knowing the expected property values of all materials present on the scanned wafer. As discussed with respect to Figure 15, some but not necessarily all pixels from other distances (surfaces) are filtered out by applying a range. However, when the characteristic values measured at multiple distances do not fall within the same range, the result of applying the range will eliminate all pixel counts from other distances (surfaces). Figure 16 shows this case. In Figure 16, a second range is applied before generating pixel counts at each distance. The result of applying the second range is that only pixels at distance 2 are counted. The resulting surface of the silicon substrate is positioned at a distance of 2 as a very clear indication.

It should be noted that to reduce the effects caused by potential noise, such as ambient vibration, a standard smoothing operation (such as Gaussian filtering) can be applied to the total pixel counts along the z-distance before any peak-seeking operations are performed.

FIG. 17 shows a flowchart 200 of one of the various steps involved in peak mode operation. In step 201, the distance between the sample and the objective lens of an optical microscope is changed in predetermined steps. in step In step 202, an image is captured at each predetermined step. In step 203, a characteristic of each pixel in each captured image is determined. In step 204, for each captured image, the maximum characteristic across all pixels in the captured image is determined. In step 205, the maximum characteristic of each captured image is compared to determine whether a surface of the sample exists at each predetermined step.

FIG. 18 shows a flowchart 300 of one of the various steps involved in range mode operation. In step 301, the distance between the sample and the objective lens of an optical microscope is changed in predetermined steps. In step 302, an image is captured at each predetermined step. In step 303, a characteristic of each pixel in each captured image is determined. In step 304, for each captured image, a count of pixels having a characteristic value within a first range is determined. In step 305, it is determined whether a surface of the sample exists at each predetermined step based on the pixel count of each captured image.

Figure 19 is a diagram of a captured image including a single feature. An example of a feature is an opening in the photoresist layer in a circular shape. Another example of a feature is an opening in the photoresist layer in the shape of a trench (such as an unplated redistribution (RDL) structure). During wafer processing, it is advantageous to measure various characteristics of a photoresist opening in a wafer layer. Measurement of a photoresist opening provides detection of defects in the structure before metal is plated into the hole. For example, if a photoresist opening is not the correct size, the plated RDL width will be wrong. Detection of this type of defect can prevent further fabrication of a defective wafer. Preventing further fabrication of a defective wafer saves material and processing costs. Figure 19 shows that the measured intensity of light reflected from the top surface of the photoresist layer is greater than the measured intensity of light reflected from openings in the photoresist layer when the captured image is focused on the top surface of the photoresist layer strength. As discussed in more detail below, the information associated with each pixel in the captured image can be used to generate an intensity value for each pixel in the captured image. Then, the intensity value of each pixel can be compared to an intensity threshold to determine whether each pixel is associated with a first region of the captured image (such as the top surface of the photoresist layer) or with a first region of the captured image Two regions, such as photoresist opening regions, are associated. This can be done by the following steps Done: (i) first apply an intensity threshold to the measured intensity of each pixel in the captured image; (ii) classify all pixels with an intensity value below the intensity threshold as the same as the measured intensity associating with a first region of the captured image; (iii) classifying all pixels having an intensity value above an intensity threshold as being associated with a second region of the captured image; and (iv) associating a A feature is defined as a group of pixels within the same area that are adjacent to other pixels associated with the same area.

The captured image shown in Figure 19 may be a color image. Each pixel of the color image contains red, blue and green (RBG) channel values. Each of these color values can be combined to produce a single intensity value for each pixel. Various methods for converting the RBG value of each pixel to a single intensity value are described below.

A first method converts the three color channels into an intensity value using three weighting values. In other words, each color channel has its own weighting value or conversion factor. We can use a preset set of three conversion factors defined in a system recipe or modify the three conversion factors based on their sample measurement needs. A second method is to subtract the color channel of each pixel from a predetermined color channel value of each color channel, and then convert this result to an intensity value using the conversion factors discussed in the first method. A third method uses a "color difference" scheme to convert colors to intensity values. In a color-difference scheme, the resulting pixel intensity is defined by how close a pixel's color is to a predefined fixed red, green, and blue color value. An example of color difference is the weighted vector distance between a pixel's color value and a fixed color value. Yet another method of "chromatic aberration" is a chromatic aberration method with a fixed color value automatically derived from an image. In one example, the border region of one of the images is known to have a background color. The weighted average of the colors of the pixels in the border area can be used as a fixed color value for the color difference scheme.

Once the color image has been converted to an intensity image, an intensity threshold can be compared with the intensity of each pixel to determine the image region to which the pixel belongs. In other words, with a value above one of the intensity thresholds A pixel with an intensity value indicates that the pixel received light reflected from a first surface of the sample, and a pixel with an intensity value below the intensity threshold value indicates that the pixel did not receive light reflected from the first surface of the sample. Once each pixel in the image is mapped to an area, the approximate shape of the feature in focus in the image can be determined.

Figures 20, 21, and 22 illustrate three different methods of generating an intensity threshold that can be used to distinguish pixels measuring light reflected from the top surface of the photoresist layer from light not measuring light Pixels of light reflected from the top surface of the resistive layer.

Figure 20 illustrates a first method of generating an intensity threshold for analyzing captured images. In this first method, a pixel count is generated for each measured intensity value. This type of graph is also known as a histogram. Once the pixel counts per intensity value are generated, a determination can be made between the peak counts of pixels originating from the measured light reflected from the photoresist layer and the peak counts of pixels originating from the measured light not reflected from the photoresist layer strength range. Select one of the intensity values within this intensity range as the intensity threshold. In one example, the midpoint between the two peak counts is chosen as the threshold intensity. In other examples falling within the present disclosure, other intensity values between the two peak counts may be used.

Figure 21 generates a second method for analyzing an intensity threshold of the captured image. In step 311, a determination is made regarding a first percentage of the captured image representing the photoresist area. This determination can be made by physical measurement, optical inspection, or based on production specifications. In step 312, a determination is made regarding a second percentage of the captured image representing the photoresist open area. This determination can be made by physical measurement, optical inspection, or based on production specifications. In step 313, all pixels in the captured image are classified according to the intensity measured by each pixel. In step 314, all pixels having an intensity within the penultimate percentage of all pixel intensities are selected. In step 315, all selected pixels are analyzed.

Figure 22 illustrates a third method of determining an intensity threshold. In step 321, a predetermined intensity threshold value is stored in memory. In step 322, the intensity of each pixel is compared to a stored intensity threshold. In step 323, all pixels with an intensity value less than an intensity threshold value are selected. In step 324, the selected pixels are analyzed.

Regardless of how the intensity threshold value is generated, the threshold intensity value is used to roughly determine where the boundaries of the features in the captured image are located. Next, the approximate boundaries of the features will be used to determine a more precise measure of the boundaries of the features, as discussed below.

FIG. 23 is a three-dimensional view of a photoresist opening shown in FIG. 19. FIG. Various photoresist opening measurements such as the area of the top and bottom openings, the diameters of the top and bottom openings, the circumference of the top and bottom openings, the cross-sectional widths of the top and bottom openings, and the depth of the openings are of interest during the process. A first measurement is the top surface open area. Figure 8 (and accompanying text) describes how to selectively focus one image on the top surface of the photoresist opening and one image focused on the bottom surface of the photoresist opening from a plurality of images obtained at different distances from the sample. Once the image focused on the top surface is selected, the above top opening measurement can be determined using the image focused on the top surface of the photoresist opening. Likewise, once the image focused on the bottom surface of the photoresist opening is selected, the bottom opening measurement described above can be determined using the image focused on the bottom surface of the photoresist opening. As discussed above and in US Patent Application Serial No. 12/699,824, entitled "3-D Optical Microscope," by James Jianguo Xu et al., the subject matter of which is incorporated herein by reference, the When taking multiple images, a pattern or grid can be projected onto the surface of the sample. In one example, an image of the projected pattern or grid is included for determining photoresist opening measurements. In another example, a new image captured at the same z-distance that does not include a pattern or grid is used to determine photoresist opening measurements. In the latter example, the new image without a projected pattern or grid on the sample provides a "clearer" image that provides easier detection of the boundaries of the photoresist openings.

FIG. 24 is a two-dimensional view of the top surface opening shown in FIG. 23. FIG. The two-dimensional map clearly shows the boundary (solid line) 40 of the top surface opening. The boundaries are traced using a line of best fit (dashed line 41). Once the best fit line trace is generated, the diameter, area and circumference of the best fit line 41 can be generated.

FIG. 25 is a two-dimensional view of the bottom surface opening shown in FIG. 23 . The two-dimensional map clearly shows the boundaries of the bottom surface opening (solid line 42). The boundaries are traced using a line of best fit (dashed line 43). Once the line of best fit trace is generated, the bottom surface opening diameter, area and circumference of the line of best fit can be calculated.

In this example, the line of best fit is automatically generated by a computer system in communication with the optical microscope. A line of best fit can be generated by analyzing the transition between dark and light portions of a selected image, as discussed in more detail below.

Figure 26 is a 2D image of an opening in a photoresist layer. The image is focused on the top surface of the photoresist layer. In this example, the light reflected from the top surface of the photoresist layer is bright because the microscope is focused on the top surface of the photoresist layer. The light intensity measured from the photoresist opening is dark because there are no reflective surfaces in the photoresist opening. The intensity of each pixel is used to determine whether the pixel belongs to the top surface of the photoresist or an opening in the photoresist. Intensity changes from transitions between the top surface of the photoresist and the openings in the photoresist can span multiple pixels and multiple intensity levels. The image background intensity may also be non-uniform. Therefore, further analysis is required to determine the exact pixel location of the border of the photoresist. To determine the pixel location of a single surface transition point, an intensity average is obtained in an adjacent bright region outside the transition region, and an intensity average is obtained in an adjacent dark region outside the transition region. The median intensity value between the average of adjacent bright areas and the average of adjacent dark areas is used as the intensity threshold to distinguish whether a pixel belongs to the top surface of the photoresist or an opening in the photoresist. This intensity threshold may be different from the previously discussed intensity threshold for selecting features within a single captured image. Once the median intensity threshold is determined, the median intensity threshold is compared to all pixels to distinguish pixels belonging to the top surface of the photoresist or openings in the photoresist. If the pixel intensity is higher than the intensity threshold, the pixel is judged as a photoresist image white. If the pixel intensity is lower than the intensity threshold value, the pixel is determined as an open area pixel. Multiple boundary points can be determined in this manner and used to fit a shape. Next, the fit shape is used to calculate all the desired dimensions of the top opening of the photoresist. In one example, the fitted shape may be selected from the group of circles, squares, rectangles, triangles, ovals, hexagons, pentagons, and the like.

FIG. 27 depicts the variation in measured intensity across adjacent regions around the luminance transition of FIG. 26 . At the leftmost portion of the adjacent region, the measured intensity is higher because the microscope is focused on the top surface of the photoresist layer. The measured light intensity is reduced by luminance transitions in adjacent regions. The measured light intensity drops to a minimum range at the rightmost portion of the adjacent region because there is no top surface of the photoresist layer in the rightmost portion of the adjacent region. Figure 27 plots this variation in measured intensity across adjacent regions. Next, a boundary point indicating where the top surface of the photoresist layer ends can be determined by applying a threshold intensity. The boundary point where the top surface of the photoresist ends is located at the intersection of the measured intensity and the threshold intensity. This procedure is repeated at different adjacent regions located along the luminance transition. A boundary point is determined for each adjacent region. Next, the boundary points of each adjacent region are used to determine the size and shape of the top surface boundary.

Figure 28 is a 2D image of an opening in a photoresist layer. The image is focused on the bottom surface of the photoresist opening. In this example, the light reflected from the bottom surface of the photoresist opening area is bright because the microscope is focused on the bottom surface of the photoresist opening. The light reflected from the photoresist area is also relatively bright because the substrate is a silicon or metal seed layer with high reflectivity. The light reflected from the boundaries of the photoresist layer is darker due to light scattering caused by the boundaries of the photoresist layer. The measured intensity of each pixel is used to determine whether the pixel belongs to the bottom surface of the photoresist opening. Intensity changes from transitions between the bottom surface of the photoresist and the open regions of the photoresist can span multiple pixels and multiple intensity levels. The image background intensity may also be non-uniform. Therefore, further analysis is required to determine the exact pixel location of the photoresist opening. To determine the pixel location of a boundary point, the location of the pixel with the smallest intensity is determined within adjacent pixels. Multiple boundary points can be determined in this manner and used to fit a shape. Next, use the fitted shape to calculate the bottom opening The desired size of the mouth.

FIG. 29 shows the variation in measured intensity across adjacent regions around the luminance transition of FIG. 28. FIG. At the rightmost portion of the adjacent region, the measured intensity is higher because the microscope is focused on the bottom surface of the photoresist opening. The measured light intensity is reduced to a minimum intensity and then reduced by luminance transitions in adjacent regions. Due to light reflection from the substrate surface, the measured light intensity rises to a relatively high intensity range at the rightmost portion of the adjacent region. Figure 29 plots this variation in measured intensity across adjacent regions. Next, the boundary points indicating where the boundaries of the photoresist openings are located can be determined by finding the location of the minimum measured intensity. The boundary point is located where the minimum measured intensity is located. The procedure is repeated at different adjacent regions located along the luminance transition. A boundary point is determined for each adjacent region. Next, the boundary points of each adjacent region are used to determine the size and shape of the bottom surface boundary.

30 is a two-dimensional image of a trench structure in a photoresist layer, such as an unplated redistribution (RDL) structure. The image is focused on the top surface of the photoresist layer. In this example, the light reflected from the top surface of the photoresist layer is bright because the microscope is focused on the top surface of the photoresist layer. The light reflected from the openings in the photoresist layer is darker because less light is reflected from the open trench areas. The intensity of each pixel is used to determine whether the pixel belongs to the top surface of the photoresist or an open area in the photoresist. Intensity changes from transitions between the top surface of the photoresist and the open regions in the photoresist can span multiple pixels and multiple intensity levels. The image background intensity may also be non-uniform. Therefore, further analysis is required to determine the exact pixel location of the border of the photoresist. To determine the pixel location of a single surface transition point, an intensity average is obtained in an adjacent bright region outside the transition region, and an intensity average is obtained in an adjacent dark region outside the transition region. The median intensity value between the average value of adjacent bright areas and the average value of adjacent dark areas was used as the intensity threshold value for distinguishing top surface photoresist reflections from non-top surface photoresist reflections. Once the mid-intensity threshold is determined, the mid-intensity threshold is compared to all adjacent pixels to determine a boundary between the top surface pixel and the photoresist opening area. If the pixel intensity is above the intensity threshold, the pixel Determined to be a top surface photoresist pixel. If the pixel intensity is lower than the intensity threshold value, the pixel is determined as a photoresist opening area pixel. Multiple boundary points can be determined in this manner and used to fit a shape. Next, the fitted shape is used to calculate all desired dimensions of the photoresist opening of the trench, such as the trench width.

FIG. 31 depicts the variation in measured intensity across adjacent regions around the luminance transition of FIG. 30. FIG. At the leftmost portion of the adjacent region, the measured intensity is higher because the microscope is focused on the top surface of the photoresist layer. The measured light intensity is reduced by luminance transitions in adjacent regions. The measured light intensity drops to a minimum range at the rightmost portion of the adjacent region because there is no top surface of the photoresist layer in the rightmost portion of the adjacent region. Figure 31 plots this variation in measured intensity across adjacent regions. Next, a boundary point indicating the end of the top surface of the photoresist layer can be determined by applying a threshold intensity. The boundary point where the top surface of the photoresist ends is located at the intersection of the measured intensity and the threshold intensity. This procedure is repeated at different adjacent regions located along the luminance transition. A boundary point is determined for each adjacent region. Next, the boundary points of each adjacent region are used to determine the size and shape of the top surface boundary.

With respect to Figures 26-31, pixel intensities are only one example of pixel characteristics that can be used to distinguish pixels from different regions in an image. For example, the wavelength or color of each pixel can also be used to distinguish pixels from different regions in an image in a similar manner. Once the boundaries between regions are precisely defined, the boundaries are then used to determine the critical dimension (CD) of a PR opening, such as its diameter or width. Typically, the measured CD values are then compared to values measured on other types of tools, such as a critical dimension scanning electron microscope (CD-SEM). This type of cross-calibration is necessary to ensure measurement accuracy in production monitoring tools.

Figure 32 is a three-dimensional view of a photoresist opening partially filled with metallization. The openings in the photoresist layer are in the shape of trenches, such as a redistribution wiring (RDL) structure. During wafer processing, it is advantageous to measure various features of the metallization deposited into the photoresist openings while the photoresist is still intact. For example, if Jin The thickness is not thick enough, we can always plate additional metal as long as the photoresist has not been stripped. The ability to spot potential problems while the wafer is still in a workable stage prevents further fabrication of a defective wafer and saves material and handling costs.

33 is a cross-sectional view of a photoresist opening partially filled with metallization. Figure 33 clearly shows that the height of the top surface of the photoresist ("PR") region is greater than the height of the top surface of the metallization. The width of the metallized top surface is also depicted in FIG. 33 . Using the various methods described above, the z-position of the top surface of the photoresist region and the z-position of the top surface of the metallization can be determined. The distance between the top surface of the photoresist region and the top surface of the metallization (also referred to as the "step height") is equal to the difference between the height of the top surface of the photoresist region and the height of the top surface of the metallization. To determine the thickness of the metallization, another measurement of the thickness of the photoresist area is required. As discussed above with respect to FIG. 11, the photoresist regions are translucent and have an index of refraction different from that of the open air. Therefore, the focal plane of the captured image focused on the light reflected from the bottom surface of the photoresist area is not actually located at the bottom surface of the photoresist area. At this time, however, our goals were different. We don't want to filter out false surface measurements, but rather the thickness of the photoresist area is now required. 40 illustrates how a portion of incident light that is not reflected from the top surface of the photoresist region travels through the photoresist region at an angle different from the incident light due to the index of refraction of the photoresist material. If this error is not resolved, the measured thickness of the photoresist area is D' (the measured z-position of the captured image focused on light reflected from the top surface of the photoresist area minus the measured z-position focused on the self-resistance area The measured z-position of the captured image on the light reflected from the bottom surface of the area), the measured thickness D' clearly depicted in Figure 40 is not close to the actual thickness D of the photoresist area. However, the error introduced by the refractive index of the photoresist region can be removed by applying a correction calculation to the measured thickness of the photoresist region. A first correction calculation is shown in Figure 40, where the actual thickness (D) of the photoresist region is equal to the measured thickness (D') of the photoresist region multiplied by the refractive index of the photoresist region. A second correction calculation is shown in Figure 40, where the actual thickness (D) of the photoresist area is equal to the measured thickness (D') of the photoresist area times the photoresist area The refractive index plus an offset value. The second correction calculation is more general and takes into account the fact that the refractive index of the photoresist varies with wavelength and that spherical aberration of an objective can affect z-position measurements when imaging through a transparent medium. Therefore, the actual thickness of the photoresist area can be calculated using a z-position of the focal plane of the captured image focused on the light reflected from the bottom surface of the photoresist area, provided proper calibration procedures are followed.

Once the correction equation is applied to the measured thickness of the photoresist area, the true thickness of the photoresist area can be obtained. Referring again to Figure 33, the thickness of the metallization can now be calculated. The thickness of the metallization is equal to the thickness of the photoresist region minus the difference between the z-position of the top surface of the photoresist region and the z-position of the top surface of the metallization.

Figure 34 is a three-dimensional view of a circular photoresist opening with metallization. 35 is a cross-sectional view of the circular photoresist opening with the metallization shown in FIG. 34. FIG. The cross-sectional view of FIG. 35 is similar to the cross-sectional view of FIG. 33 . Figure 35 clearly shows that the height of the top surface of the photoresist ("PR") region is greater than the height of the top surface of the metallization. Using the various methods described above, the z-position of the top surface of the photoresist region and the z-position of the top surface of the metallization can be determined. The distance between the top surface of the photoresist region and the top surface of the metallization (also referred to as the "step height") is equal to the difference between the height of the top surface of the photoresist region and the height of the top surface of the metallization. To determine the thickness of the metallization, another measurement of the thickness of the photoresist area is required. As discussed above with respect to FIG. 11, the photoresist regions are translucent and have an index of refraction different from that of the open air. Therefore, the focal plane of the captured image focused on the light reflected from the bottom surface of the photoresist area is not actually located at the bottom surface of the photoresist area. At this time, however, our goals were different. The thickness of the photoresist region is now required. 40 illustrates how a portion of incident light that is not reflected from the top surface of the photoresist region travels through the photoresist region at an angle different from the incident light due to the index of refraction of the photoresist material. If this error is not resolved, the measured thickness of the photoresist area is D' (measured for the captured image focused on the light reflected from the top surface of the photoresist area z position minus the measured z position of the captured image focused on the light reflected from the bottom surface of the photoresist area), the measured thickness D' clearly depicted in Figure 40 is not close to the actual thickness of the photoresist area D. However, the error introduced by the refractive index of the photoresist region can be removed by applying a correction calculation to the measured thickness of the photoresist region. A first correction calculation is shown in Figure 40, where the actual thickness (D) of the photoresist region is equal to the measured thickness (D') of the photoresist region multiplied by the refractive index of the photoresist region. A second correction calculation is shown in Figure 40, where the actual thickness (D) of the photoresist region is equal to the measured thickness (D') of the photoresist region times the refractive index of the photoresist region plus an offset value. The second correction calculation is more general and takes into account the fact that the refractive index of the photoresist varies with wavelength and that spherical aberration of an objective can affect the z-position measurement when imaging through a transparent medium. Therefore, the actual thickness of the photoresist area can be calculated using a z-position of the focal plane of the captured image focused on the light reflected from the bottom surface of the photoresist area, provided proper calibration procedures are followed.

Once the correction equation is applied to the measured thickness of the photoresist area, the true thickness of the photoresist area can be obtained. Referring again to Figure 35, the thickness of the metallization can now be calculated. The thickness of the metallization is equal to the thickness of the photoresist region minus the difference between the z-position of the top surface of the photoresist region and the z-position of the top surface of the metallization.

Figure 36 is a three-dimensional view of a metal pillar above the passivation layer. 37 is a cross-sectional view of a metal pillar above the passivation layer shown in FIG. 36. FIG. Figure 37 clearly shows that the height of the top surface of the passivation layer is less than the height of the top surface of the metal layer. The diameter of the metallized top surface is also depicted in FIG. 37 . Using the various methods described above, the z-position of the top surface of the passivation layer and the z-position of the top surface of the metal layer can be determined. The distance between the top surface of the passivation layer and the top surface of the metal layer (also referred to as the "step height") is equal to the difference between the height of the top surface of the metal layer and the height of the top surface of the passivation layer. To determine the thickness of the metal layer, another measurement of the thickness of the passivation layer is required. As discussed above with respect to Figure 11, a translucent material, such as a photoresist region or a passivation layer, has a different refraction than the open air A rate of refraction. Therefore, the focal plane of the captured image focused on the light reflected from the bottom surface of the passivation layer is not actually located at the bottom surface of the passivation layer. At this time, however, our goals were different. The thickness of the passivation layer is now required. 47 illustrates how a portion of incident light that is not reflected from the top surface of the passivation layer travels through the passivation layer at an angle different from the incident light due to the index of refraction of the passivation material. If this error is not resolved, the measured thickness of the passivation layer is D' (the measured z-position of the captured image focused on light reflected from the top surface of the passivation region minus the bottom of the self-passivation region focused The measured z position of the captured image on the light reflected from the surface), the measured thickness D' clearly depicted in Figure 47 is not close to the actual thickness D of the passivation layer. However, errors introduced by the index of refraction of the passivation layer can be removed by applying a correction calculation to the measured thickness of the passivation layer. A first correction calculation is shown in Figure 47, where the actual thickness (D) of the passivation layer is equal to the measured thickness (D') of the passivation layer multiplied by the index of refraction of the passivation layer. A second correction calculation is shown in Figure 47, where the actual thickness (D) of the passivation layer is equal to the measured thickness (D') of the passivation layer times the index of refraction of the passivation layer plus an offset value. The second correction calculation is more general and takes into account the fact that the refractive index of the passivation layer varies depending on wavelength and that spherical aberration of an objective lens can affect z-position measurements when imaging through a transparent medium. Therefore, the actual thickness of the passivation layer is calculated using a z-position of the focal plane of the captured image focused on the light reflected from the bottom surface of the passivation layer, provided proper calibration procedures are followed.

Once the correction equation is applied to the measured thickness of the passivation layer, the true thickness of the passivation layer can be obtained. Referring again to Figure 37, the thickness of the metal layer can now be calculated. The thickness of the metal layer is equal to the sum of the thickness of the passivation layer and the difference between the z-position of the top surface of the passivation layer and the z-position of the top surface of the metal layer.

Figure 38 is a three-dimensional view of the metal over the passivation layer. In this particular case, the metal structure shown is redistribution (RDL). 39 is a cross-sectional view of the metal over the passivation layer shown in FIG. 38. FIG. Figure 39 clearly shows that the height of the top surface of the passivation layer is less than the height of the top surface of the metal layer Spend. Using the various methods described above, the z-position of the top surface of the passivation layer and the z-position of the top surface of the metal layer can be determined. The distance between the top surface of the passivation layer and the top surface of the metal layer (also referred to as the "step height") is equal to the difference between the height of the top surface of the metal layer and the height of the top surface of the passivation layer. To determine the thickness of the metal layer, another measurement of the thickness of the passivation layer is required. As discussed above with respect to FIG. 11, the translucent material, such as a photoresist region or a passivation layer, has an index of refraction that is different from that of the open air. Therefore, the focal plane of the captured image focused on the light reflected from the bottom surface of the passivation layer is not actually located at the bottom surface of the passivation layer. At this time, however, our goals were different. The thickness of the passivation layer is now required. 40 illustrates how a portion of incident light that is not reflected from the top surface of the passivation layer travels through the passivation layer at an angle different from the incident light due to the index of refraction of the passivation material. If this error is not resolved, the measured thickness of the passivation layer is D' (the measured z-position of the captured image focused on light reflected from the top surface of the passivation region minus the bottom of the self-passivation region focused The measured z-position of the captured image on the light reflected from the surface), the measured thickness D' clearly depicted in Figure 40 is not close to the actual thickness D of the passivation layer. However, errors introduced by the index of refraction of the passivation layer can be removed by applying a correction calculation to the measured thickness of the passivation layer. A first correction calculation is shown in Figure 40, where the actual thickness (D) of the passivation layer is equal to the measured thickness (D') of the passivation layer multiplied by the index of refraction of the passivation layer. A second correction calculation is shown in Figure 40, where the actual thickness (D) of the passivation layer is equal to the measured thickness (D') of the passivation layer times the index of refraction of the passivation layer plus an offset value. The second correction calculation is more general and takes into account the fact that the refractive index of the passivation layer varies depending on wavelength and that spherical aberration of an objective lens can affect z-position measurements when imaging through a transparent medium. Thus, the actual thickness of the passivation layer can be calculated using a z-position of the focal plane of the captured image focused on light reflected from the bottom surface of the passivation layer, provided proper calibration procedures are followed.

Once the correction equation is applied to the measured thickness of the passivation layer, the true thickness of the passivation layer can be obtained. Referring again to Figure 39, the thickness of the metal layer can now be calculated. The thickness of the metal layer is equal to the passivation The sum of the thickness of the layer and the difference between the z-position of the top surface of the passivation layer and the z-position of the top surface of the metal layer.

41 is a graph showing peak mode operation using images captured at various distances when a photoresist opening is within the field of view of an optical microscope. The captured image shown in FIG. 41 was obtained from a sample similar to the sample structure shown in FIG. 32 . This structure is a metal plated trench structure. The top view of the sample shows the area of the photoresist opening (a metallization) in the x-y plane. The PR openings also have a depth (higher than metallization) of a specific depth in the z-direction. The top view in Figure 41 below shows images captured at various distances. At distance 1, the optical microscope is not focused on the top surface of the photoresist region or the top surface of the metallization. At distance 2, the optical microscope was focused on the top surface of the metallization, but not on the top surface of the photoresist area. This results in an increased characteristic value (intensity/contrast/ stripe contrast). At distance 3, the optical microscope is not focused on the top surface of the photoresist region or the top surface of the metallization. Therefore, at distance 3, the maximum characteristic value will be substantially lower than the maximum characteristic value measured at distance 2. At distance 4, the optical microscope was not focused on any surface of the sample; however, due to the difference between the refractive index of air and that of the photoresist area, the maximum characteristic value (intensity/contrast/fringe contrast) was measured an increase. Figures 11, 40 and accompanying text describe this phenomenon in more detail. At distance 6, the optical microscope was focused on the top surface of the photoresist region, but not on the top surface of the metallization. This results in an increased characteristic value (intensity/contrast/ stripe contrast). Once the maximum characteristic value from each captured image is determined, the results can be used to determine at which distances the surfaces of the wafer are located.

FIG. 42 is a graph showing three-dimensional information derived from the peak mode operation shown in FIG. 41 surface. As discussed with respect to FIG. 41 , the maximum characteristic value of the images captured at distances 1, 3, and 5 has a maximum characteristic value that is lower than the maximum characteristic value of the images captured at distances 2, 4, and 6. The curve of the maximum characteristic value at various z-distances may contain noise due to environmental effects such as vibration. To minimize this noise, a standard smoothing method, such as Gaussian filtering with a certain kernel size, can be applied before further data analysis.

One method of comparing maximum characteristic values is performed by a peak finding algorithm. In one example, a derivative method is used to locate the zero crossings along the z-axis to determine the distance at which each "peak" exists. Next, the maximum characteristic value at each distance where a peak is found is compared to determine the distance at which the maximum characteristic value is measured. In the case shown in Figure 42, a peak would be found at distance 2, which serves as an indication that a surface of the sample is located at distance 2.

Another method of comparing maximum characteristic values is performed by comparing each maximum characteristic value with a predetermined threshold value. Threshold values can be calculated based on wafer material, distance and optical microscope specifications. Alternatively, threshold values can be determined by empirical testing prior to automated processing. In either case, the maximum characteristic value of each captured image is compared to a threshold value. If the maximum characteristic value is greater than the threshold value, it is determined that the maximum characteristic value indicates the presence of a surface of the wafer. If the maximum characteristic value is not greater than the threshold value, it is determined that the maximum characteristic value does not indicate a surface of the wafer.

An alternative use of the peak mode method described above, the range mode method described in Figure 13 and related text can be used to determine the z-position of different surfaces of a sample.

43 is a view of a captured image, including an outline of a first analysis area A and a second analysis area B, focused on a top surface of a photoresist layer in a trench structure. As discussed above, an entire field of view of each captured image can be used to generate three-dimensional information. However, it would be advantageous to have the option to generate three-dimensional information using only one selectable portion of the field of view (region A or region B). In one example, a user uses a mouse or Touch screen device selection area. Once selected, we can apply different thresholds to each region to more efficiently pick out a particular surface peak as shown in Figure 42. This case is depicted in FIG. 43 . When it is desired to obtain three-dimensional information about a metallized top surface, a selectable portion of the field of view (Area A) is set to include metallized regions, since characteristic values associated with a metal surface are typically greater than those associated with light Resistor-associated characteristic values, so a high threshold value can be applied to region A to filter out the photoresist-associated characteristic values to improve detection of metal surface peaks. Alternatively, when it is desired to obtain three-dimensional information about the top surface of a photoresist region, a selectable portion of the field of view (region B) is set to a cell located at the center of an image. Characteristic values associated with a photoresist surface are typically relatively weak compared to characteristic values associated with metal surfaces. The quality of the original signal used for the determination of the characteristic value calculation is optimal around the center of the field of view enclosed in the area B. By setting an appropriate threshold value for region B, we can more effectively detect a weak characteristic value peak on the photoresist surface. The user can set and adjust Regions A and B, as well as the threshold values used in each region, and save them in a recipe for automated measurement via a graphical interface that displays overhead images of the samples.

Figure 44 is a three-dimensional view of a bump above the passivation structure. 45 is a top view of the bump above the passivation structure shown in FIG. 44, including an outline of a first analysis area A and a second analysis area B. FIG. Area A can be set such that area A will always include the apex of the metal bump during an automated sequence of measurements. Region B does not enclose any part of the metal bump and only part of the passivation layer. Analyzing only area A of all captured images provides pixel filtering so that most of the pixels analyzed contain information about the metal bumps. Analyzing region B of all captured images provides pixel filtering so that all pixels analyzed contain information about the passivation layer. The application of the user-selectable analysis area provides pixel filtering based on location rather than pixel value. For example, when the location of the top surface of the passivation layer is required, Region B can be applied and can be immediately eliminated from the analysis by metal bumps All effects caused by the block. In another example, when the location of the apex of the metal bump is required, Region A can be applied and all effects caused by large passivation layer regions can be immediately eliminated from the analysis.

In some instances, it may also be useful to fix the spatial relationship between region A and region B. When measuring a metal bump of a known size (such as shown in FIGS. 44 and 45 ), it is useful to fix the spatial relationship between area A and area B to provide consistent measurement, since area A is always It is used to measure the 3D information of the metal bump and the area B is always used to measure the 3D information of the passivation layer. Furthermore, when area A and area B have a fixed spatial relationship, an adjustment in one area automatically causes an adjustment in the other area. This scenario is depicted in FIG. 46 . 46 is a top view of the adjusted analysis area A and the analysis area B when the entire bump is not positioned in the original analysis area A. FIG. This can occur for a variety of reasons, such as inaccurate placement of one of the samples by the handler or procedural variations during sample fabrication. Whatever the reason, region A needs to be adjusted to be properly centered on the apex of the metal bump. Area B also needs to be adjusted to ensure that area B does not contain any part of the metal bumps. When the spatial relationship between area A and area B is fixed, then adjustment of area A automatically causes realignment of area B.

47 is a cross-sectional view of the bump above the passivation structure depicted in FIG. 44. FIG. When the thickness of the passivation layer is substantially greater than the distance between predetermined steps of the optical microscope during image acquisition, the z-position of the top surface of the passivation layer can be easily detected as discussed above. However, when the thickness of the passivation layer is not substantially greater than the distance between the predetermined steps of the optical microscope (ie, the passivation layer is relatively thin), the z-position of the top surface of the passivation layer may not be easily detected and measured. The difficulty arises due to the small percentage of light reflected from the top surface of the passivation layer compared to the large percentage of light reflected from the bottom surface of the passivation layer. In other words, the characteristic value peak associated with the top surface of the passivation layer is very weak compared to the characteristic value peak associated with the bottom surface of the passivation layer. The captured image while focusing at a predetermined step on the high-intensity reflection from the bottom surface of the passivation layer and focusing on the When the captured images at a predetermined step of the low-intensity reflection on the top surface of the layer are less than a few predetermined steps apart, it is impossible to distinguish the reflection received from the bottom surface of the passivation layer from the reflection received from the top surface of the passivation layer . This problem can be solved by operating in different ways.

In a first approach, the total number of predetermined steps across the scan can be increased in order to provide additional resolution across the entire scan. For example, the number of predetermined steps across the same scan distance can be doubled, which will result in a doubling of the Z resolution of the scan. This approach will also result in doubling the amount of images captured during a single scan. The resolution of the scan can be increased until the characteristic peak measured from the top surface reflectance can be distinguished from the characteristic peak measured from the bottom surface reflectance. 49 illustrates a situation in which sufficient resolution is provided in the scan to distinguish reflections from the top and bottom surfaces of the passivation layer.

In a second method, the predetermined total number of steps is also increased, however, only a part of the steps is used to capture the image and the rest is skipped.

In a third method, the distance between the predetermined steps may be changed such that the distance between the steps is smaller near the passivation layer and the distance between the steps is larger outside the vicinity of the passivation layer. This method provides greater resolution near the passivation layer and lesser resolution outside the vicinity of the passivation layer. This approach eliminates the need to add additional predetermined steps to the scan, but redistributes the predetermined steps in a non-linear fashion to provide additional resolution where needed at the expense of lower resolution without requiring high resolution.

For additional description of how to improve scanning resolution, see US Patent Application No. 13, entitled "3D Microscope Including Insertable Components To Provide Multiple Imaging and Measurement Capabilities," filed December 21, 2011 by James Jianguo Xu et al. /333,938 (the subject matter of this case is incorporated herein by reference).

Using any of the methods discussed above, the z-position of the top surface of the passivation layer can be determined.

The height of the apex of the metal bump relative to the top surface of the passivation layer ("bumps above the passivation layer Height") is also a measurement of interest. The bump height above the passivation layer is equal to the z position of the apex of the bump minus the z position of the top surface of the passivation layer. The determination of the z-position of the top surface of the passivation layer is described above. The determination of the z-position of the vertices of the bump can be performed using different methods.

In a first method, the z-position of the apex of the bump is determined by determining the z-position of the peak characteristic value across each x-y pixel position of all the captured images. In other words, for each x-y pixel position, at each z position, the measured characteristic values are compared across all captured images and the z position containing the largest characteristic value is stored in an array. The result of performing this procedure across all x-y pixel locations is an array of all x-y pixel locations and the associated peak z location for each x-y pixel location. The maximum z position in the array is measured as the z position of the vertices of the bump. For additional description of how to generate three-dimensional information, see US Patent Application Serial No. 12/699,824 and US Patent No. 8,174,762, filed February 3, 2010, by James Jianguo Xu et al., entitled "3-D Optical Microscope" (the subject matter of these cases is incorporated herein by reference).

In a second method, the z-position of the apex of the bump is determined by generating a fit of a three-dimensional model of one of the bump's surfaces and then using the three-dimensional model to calculate the peak value of the bump's surface. In one example, this can be done by generating the same array as described above with respect to the first method, however, once the array is complete, the array is used to generate the three-dimensional model. Three-dimensional models can be generated using a second-order polynomial function fitted to the data. Once the three-dimensional model is generated, the derivative of the surface slope of the bump is determined. The apex of the bump is calculated to be located where the derivative of the surface slope of the bump is equal to zero.

Once the z-position of the apex of the bump is determined, the bump height above the passivation layer can be calculated by subtracting the z-position of the top surface of the passivation layer from the z-position of the apex of the bump.

Figure 48 is a graph showing peak mode operation using images captured at various distances when only one passivation layer is within region B of the optical microscope's field of view. By analyzing only the pixels within region B (shown in Figure 45), all pixel information about the metal bumps is excluded. Therefore, by analyzing The three-dimensional information generated by the pixels in region B will only be affected by the passivation layer present in region B. The captured image shown in FIG. 48 was obtained from a sample similar to the sample structure shown in FIG. 44 . This structure is a metal bump above the passivation structure. The top view of the sample shows the area of the passivation layer in the x-y plane. With only the pixels within region B selected, the metal bumps are not visible in the top view. The top view in Figure 48 below shows images captured at various distances. At distance 1, the optical microscope is not focused on or on the top surface of the passivation layer. At distance 2, the optical microscope was not focused on any surface of the sample; however, due to the difference between the refractive index of air and that of the passivation layer, one of the maximum specific values (intensity/contrast/fringe contrast) was measured increase. Figures 11, 40 and accompanying text describe this phenomenon in more detail. At distance 3, the optical microscope is not focused on the top surface of the passivation layer or the bottom surface of the passivation layer. Therefore, at distance 3, the maximum characteristic value will be substantially lower than the characteristic value measured at distance 2. At distance 4, the optical microscope is focused on the top surface of the passivation layer, which results in an increase in pixels receiving light reflected from the top surface of the passivation layer compared to pixels receiving light reflected from other surfaces that are out of focus Large characteristic value (intensity/contrast/fringe contrast). At distances 5, 6 and 7, the optical microscope was not focused on the top surface of the passivation layer or the bottom surface of the passivation layer. Therefore, at distances 5, 6 and 7, the maximum characteristic value will be substantially lower than the characteristic value measured at distances 2 and 4. Once the maximum characteristic value from each captured image is determined, the results can be used to determine at which distances each surface of the sample is located.

FIG. 49 is a graph showing three-dimensional information derived from the peak mode operation of FIG. 48. FIG. Due to the pixel filtering provided by analyzing only pixels within region B of all captured images, peak mode operation only provides an indication of one of the surfaces of the passivation layer at the two z positions (2 and 4). The top surface of the passivation layer is positioned at the higher of the two indicated z-position positions. The lowest of the two indicated z-position positions is a false "artifact surface" in which light reflected from the bottom surface of the passivation layer is measured due to the index of refraction of the passivation layer. The top surface of the passivation layer is measured using only the pixels positioned within region B The z-position simplifies peak mode operation and reduces the likelihood of erroneous measurements due to light reflection from metal bumps positioned on the same sample.

An alternative use of the peak mode method described above, the range mode method described in Figure 13 and related text can be used to determine the z-position of different surfaces of a sample.

Although certain specific embodiments are described above for instructional purposes, the teachings of this patent document have general applicability and are not limited to the specific embodiments described above. Accordingly, various modifications, adaptations and combinations of the various features of the described embodiments may be practiced without departing from the scope of the invention as set forth in the Claims of Invention.

Claims (22)

A method of generating three-dimensional (3-D) information of a sample using an optical microscope, the method comprising: changing the distance between the sample and an objective lens of the optical microscope in predetermined steps; capturing at each predetermined step an image in which a first surface of the sample and a second surface of the sample are within a field of view of each of the captured images; determining a characteristic value of each pixel in each of the captured images, wherein the characteristic value of each pixel is selected from the group consisting of intensity, contrast, and fringe contrast; determining, for each captured image, a maximum characteristic value across a first portion of pixels in the captured image; comparing each captured image the maximum characteristic value of the image to determine whether a surface of the sample exists at each predetermined step; determine a first captured image focused on a vertex of a bump of the sample; based on each captured image The characteristic value of each pixel of the sample determines a second captured image focused on a first surface of the sample; and determines a first distance between the vertex of the bump and the first surface, wherein the The bump is a metal bump. The method of claim 1, wherein the optical microscope comprises a stage, wherein the sample is supported by the stage, wherein the optical microscope is adapted to communicate with a computer system, wherein the computer system comprises a stage adapted to store each A memory device that captures an image, and wherein the optical microscope is selected from the group consisting of a confocal microscope, a structured illumination microscope, and an interferometer. The method of claim 1, wherein the determining of the first captured image further comprises: determining a maximum characteristic value of each x-y pixel location within a second portion of x-y pixel locations across all the captured images, wherein the second portion of x-y pixel locations includes at least some of the x-y pixel locations included in each captured image; determining a subset of the captured images that includes only one x-y pixel location maximum characteristic value The captured images of the captured images are included in the subset; and it is determined that among all captured images within the subset of captured images, the first captured image is compared to all captured images within the subset of captured images Other captured images are focused at an uppermost z position. The method of claim 1, wherein the first portion of pixels includes all pixels included in the captured image. The method of claim 1, wherein the first portion of pixels includes less than all of the pixels included in the captured image. The method of claim 3, wherein the second portion of pixels includes all pixels included in the captured image. The method of claim 3, wherein the second portion of pixels includes less than all of the pixels included in the captured image. The method of claim 1, wherein the first portion of the pixel does not receive light reflected from the metal bump. The method of claim 3, wherein the second portion of the pixel receives light reflected from the vertex of the metal bump. The method of claim 3, wherein the spatial relationship between the first portion of the pixel and the second portion of the pixel is fixed. The method of claim 3, wherein the second portion of pixels is contiguous and centered on the vertex of the bump. The method of claim 1, wherein the first surface is a top surface of a passivation layer. A method of generating three-dimensional (3-D) information of a sample using an optical microscope, the method comprising: changing the distance between the sample and an objective lens of the optical microscope in predetermined steps; capturing at each predetermined step an image in which a first surface of the sample and a second surface of the sample are within a field of view of each of the captured images; determining a characteristic value of each pixel in each of the captured images, wherein the characteristic value of each pixel is selected from the group consisting of intensity, contrast, and fringe contrast; a count of pixels having a characteristic value within a first range across a first portion of a pixel is determined for each captured image, which does not have one of the characteristic values within the first range All pixels of the sample are not included in the pixel count; determine whether a surface of the sample exists at each predetermined step based on the pixel count of each captured image; determine whether a surface is focused on a vertex of a bump of the sample a first captured image; determining a second captured image focused on a first surface of the sample based on the characteristic value of each pixel in each captured image; and determining that the vertex of the bump is the same as the A first distance between the first surfaces, wherein the bump is a metal bump. The method of claim 13, wherein the optical microscope comprises a stage, wherein the sample is supported by the stage, wherein the optical microscope is adapted to communicate with a computer system, wherein the computer system comprises a stage adapted to store each A memory device that captures an image, and wherein the optical microscope is selected from the group consisting of a confocal microscope, a structured illumination microscope, and an interferometer. The method of claim 13, wherein the determining of the first captured image further comprises: determining a maximum characteristic value of each x-y pixel location within a second portion of x-y pixel locations across all of the captured images, wherein the second portion of x-y pixel locations includes at least some of the x-y pixel locations included in each captured image; determining a subset of the captured images that includes only one x-y pixel location maximum characteristic value The captured images of the captured images are included in the subset; and it is determined that among all captured images within the subset of captured images, the first captured image is compared to all captured images within the subset of captured images Other captured images focus on the most at the high z position. The method of claim 13, wherein the first portion of pixels includes all pixels included in the captured image. The method of claim 13, wherein the first portion of pixels includes less than all of the pixels included in the captured image. The method of claim 15, wherein the second portion of pixels includes all pixels included in the captured image. The method of claim 15, wherein the spatial relationship between the first portion of the pixel and the second portion of the pixel is fixed. The method of claim 15, wherein the second portion of pixels is contiguous and centered on the vertex of the bump. A method of generating three-dimensional (3-D) information of a sample using an optical microscope, the method comprising: changing the distance between the sample and an objective lens of the optical microscope in predetermined steps; capturing at each predetermined step an image in which a first surface of the sample and a second surface of the sample are within a field of view of each of the captured images; determining a characteristic value of each pixel in each of the captured images, wherein the characteristic of each pixel value selected from the group consisting of intensity, contrast, and fringe contrast; determine a z-position of a vertex of a bump of the sample; determine focus on one of the samples based on the characteristic value of each pixel in each captured image a first captured image on a first surface; and determining a first distance between the vertex of the bump and the first surface, wherein the bump is a metal bump. The method of claim 21, wherein the determining of the z position of the vertex comprises: identifying a plurality of x,y,z pixel positions across all of the captured images, wherein the plurality of x,y,z pixel positions are the same as the A top surface of the bump is associated; a best fit algorithm is applied to generate a continuous three-dimensional estimate of the top surface of the bump; and a maximum height of the continuous three-dimensional estimate is determined.

Images