TWI735471B - Method and system for identifying localized crystallographic defects in a monocrystalline silicon in a wafer - Google Patents

Abstract

A method that includes: illuminating a wafer with excitation light having a wavelength and intensity sufficient to induce photoluminescence in the wafer; filtering photoluminescence emitted from a portion of the wafer in response to the illumination; directing the filtered photoluminescence onto a detector to image the portion of the wafer on the detector with a spatial resolution of 1 μm x 1 μm or smaller; and identifying one or more crystallographic defects in the wafer based on the detected filtered photoluminescence.

Description

System and method for identifying local crystal defects in single crystal silicon of wafer

The present invention relates to identifying defects in integrated circuit devices (such as complementary metal oxide semiconductor image sensors).

Semiconductor materials are widely used in the fields of electronics and optoelectronics. Crystalline semiconductor materials are prone to crystal defects, which may be detrimental to the performance of devices using the material. Crystal defects may cause related photoluminescence that can be used to identify defects.

Complementary metal oxide semiconductor (CMOS) image sensors (CIS) in integrated circuit devices are an example of optoelectronic devices using crystalline semiconductor materials. The CMOS image sensor is used to convert the light intensity pattern into an electronic digital signal. In some embodiments, the CMOS image sensor is a two-dimensional photodiode array with complementary metal oxide semiconductor logic for signal processing. Each independent photodiode system with processing complementary metal oxide semiconductor logic is called a pixel. In some embodiments, the CMOS image sensor has 1,000,000 or more pixels.

CMOS image sensors are usually manufactured on n/n++ or p/p++ wafers. In an example, in some embodiments, thin lightly doped n-type or p-type epitaxial layers (e.g., each having a ^{doping concentration of 1×10 14} to 1×10 ¹⁵ cm ^-3 and a thickness of The layer of 3~5μm) is grown on a highly doped n++ or p++ substrate (for example, a substrate with a doping concentration of ^{1×10 18} to 1×10 ²⁰ cm ^-3). The CMOS image sensor is formed on the epitaxial layer, which is usually called the device active area. The performance of the CMOS image sensor is at least partially affected by the nature of the active region.

Highly doped substrates (usually referred to as handles) provide mechanical support for the active area during CMOS image sensor manufacturing. In some embodiments, the substrate also reduces the occurrence of cross-talk in the CMOS image sensor. For example, the substrate can reduce the crosstalk caused by the generation of minority carriers under a pixel in response to the red light reaching the adjacent pixels of the CMOS image sensor.

The CMOS image sensor can be configured according to various configurations. For example, the CMOS image sensor can be configured as a front side illuminated (FSI) CMOS image sensor, or as a back side illuminated (BSI) CMOS image sensor. Here, the "front" side refers to the side of the wafer on which the IC pixel structure is fabricated. In some embodiments, in order to fabricate a backside-illuminated CMOS image sensor, the CMOS image sensor wafer first undergoes CMOS image sensor processing on its front side. Then, the CMOS image sensor wafer is bonded to the wafer carrier along its front side, and its back side is thinned (for example, reduced by a few μm) until all the n++ or p++ substrates are removed. Then passivate the surface of the CMOS image sensor wafer and cover it with an anti-reflection coating, and fabricate a color filter on its back side. During use, the optical image is projected on the back side of the CMOS image sensor wafer, and the CMOS image The image sensor converts the light image into an electronic digital signal.

The light from the image projected on the CMOS image sensor with photon energy larger than the silicon band gap is mainly absorbed in the active area of the CMOS image sensor. This absorption produces electron-hole pairs that cause photocurrent. These light-generated minority carriers are then collected by the p-n junction at that location. The number of minority carriers generated by light is proportional to the number of photons absorbed in the active region of the CMOS image sensor, and changes with the light intensity. Therefore, the intensity of the light incident on the active area of the CMOS image sensor can be calculated according to the magnitude of the generated photocurrent. In fact, it is generally expected that each pixel of a CMOS image sensor can generate the same or substantially similar photocurrent in response to uniform, low-level illumination. Otherwise, pixels with lower or higher photocurrent (such as "defective" pixels) may cause bright or dark spots in the resulting image (that is, higher than the image intensity of the defect-free area).

In some embodiments, local crystal defects and heavy metal contamination can increase or decrease the photocurrent from a given pixel, resulting in an image with bright or dark spots under low illumination levels. When present in the space charge region of the p-n junction, these defects are used as minority carrier generation centers. This will cause an increase in the dark current of these pixels, and if the defect is severe enough, it will become a white spot or bright spot in the generated image. When existing outside the space charge region of the p-n junction, these defects serve as the minority carrier recombination centers (recombination centers). This will result in a decrease in the photoelectric flux collected by the junction, and if the defect is severe enough, it will become a dark spot in the image generated at a low illumination level.

Local crystal defects or heavy metal contaminants may be introduced into any step during the manufacturing process of the CMOS image sensor. Therefore, in order to To improve and control the manufacturing process of CMOS image sensors, it is important to quickly identify the processing steps that introduce these defects.

The present invention describes systems and techniques for identifying defects in semiconductor materials (including defects found in integrated circuit devices). More specifically, the system and technology can be used to identify locally independent crystalline defects in semiconductor materials that emit photoluminescence, especially defect photoluminescence that is located at a different energy from band-to-band photoluminescence. Such crystal defects can be introduced into semiconductor materials at various stages of materials or manufacturing devices (such as the manufacturing of CMOS image sensors).

Generally speaking, in the first aspect, the present invention is characterized by a method comprising: irradiating a wafer with excitation light having a wavelength and intensity sufficient to induce photoluminescence in the wafer; and filtering a part of the wafer The emitted photoluminescence is in response to the above illumination; the filtered photoluminescence is directed to the detector, and the part of the wafer is imaged for detection with a spatial resolution of 1μm×1μm or less On the device; and according to the detected filtered photoluminescence to identify one or more crystal defects in the wafer.

The implementation of the method of the present invention may include one or more of the following features and/or features in other aspects. For example, the detected filtered photoluminescence can correspond to photoluminescence from crystal defects in the wafer. Filtering can basically remove the photoluminescence of the inter-band transition in the wafer from the filtered photoluminescence.

The detected filtered photoluminescence may include light having an energy in the range of about 0.7 eV to about 0.9 eV. The wafer may be a silicon wafer. Filtering can basically block the detected light with energy greater than about 1.0 eV.

In some embodiments, filtering may further include filtering the excitation light reflected from the portion of the wafer.

Some embodiments may further include detecting the excitation light reflected from a portion of the wafer. The method of the present invention may also include comparing the detected photoluminescence with the detected excitation light, and identifying one or more defects in the wafer based on the comparison result.

The wafer can be used for complementary metal oxide semiconductor (CMOS) image sensors (CMOS image sensors). This part of the wafer can correspond to one or more pixels of the CMOS image sensor. The step of identifying one or more defects in the wafer may include identifying one or more defective pixels of the CMOS image sensor. In some embodiments, the step of identifying one or more defects in the wafer includes identifying one or more defects having a size of 1 μm or less. It can also identify larger defects.

The wafer can be used in power management IC (power management integrated circuit, PMIC) devices. Wafers can be used in optoelectronic devices. Wafers can be used for photovoltaic devices (photovoltaic devices).

The method of the present invention can identify defects at any stage of wafer processing (including after wafer processing is completed).

The method of the present invention may further include: forming a photoluminescence intensity map of the part of the wafer based on the photoluminescence emitted from the part of the wafer; and forming a crystal according to the excitation light reflected from the part of the wafer The reflection intensity graph of that part of the circle. The step of comparing the detected photoluminescence from the part of the wafer with the detected reflected excitation light from the area of the wafer may include: determining whether the photoluminescence intensity map includes the first position of the wafer The first intensity change; when judging that the photoluminescence intensity map includes the first intensity change at the first position of the wafer, the judgment is reversed Whether the radiation intensity map includes the second intensity change at the first position of the wafer; and when it is determined that the reflection intensity map does not include the second intensity change at the first position of the wafer, determine whether the first position of the wafer is defective . The step of comparing the detected photoluminescence from the part of the wafer with the detected reflected excitation light from the area of the wafer may further include: judging that the reflection intensity map includes the first position at the first position of the wafer 2. When the intensity changes, it is judged that there is no defect in the first position of the wafer.

In some embodiments, the method of the present invention further includes adjusting the characteristics of the excitation light. The step of adjusting the characteristics of the excitation light may include adjusting the wavelength of the excitation light. The photoluminescence emitted from the second part of the wafer can be increased by adjusting the wavelength of the excitation light. The second part of the wafer can be located at a different depth in the wafer from the first part.

The excitation light may have a wavelength in the range from 200 nm to 1,100 nm.

The wafer can be a silicon wafer or a compound semiconductor wafer.

The method of the present invention may further include performing processing steps on the wafer. The treatment step can be selected from an ion implantation step, an annealing step, a layer deposition step, an oxidation step, and a polishing step. The processing steps can be performed after the crystal defects are identified. The method of the present invention may include irradiating the processed wafer with excitation light, and identifying one or more additional defects in the processed wafer based on photoluminescence from the processed wafer. The method of the present invention may include comparing the identified crystal defects with additional defects in the wafer.

Crystal defects correspond to bright parts in the image of a part of the wafer.

Generally speaking, in another aspect, a system including the features of the present invention includes: an illumination module for illuminating the wafer with excitation light having a wavelength and intensity sufficient to induce photoluminescence in the wafer; and a detection module , Used to detect the photoluminescence emitted from a part of the wafer in response to the above illumination; an imaging lens (imaging optic) used to 1μm×1μm or less spatial resolution Image a part of the wafer onto the detection module; an optical filter for filtering the photoluminescence emitted from the part of the wafer before the detection module performs detection; and the processing module, with Identify one or more crystal defects in the wafer based on the detected filtered photoluminescence.

Embodiments of the system may include one or more of the following features and/or other aspects of features. For example, the filter can emit light corresponding to photoluminescence from crystal defects in the wafer to the detection module. The optical filter can basically block the photoluminescence from the inter-band transition in the wafer of the detection module.

The filter can emit light with energy in the range of about 0.7 eV to about 1.0 eV to the detection module. The optical filter can basically block light with energy greater than about 1.0 eV from the detection module.

In some embodiments, the filter self-detection module blocks at least some of the excitation light reflected from the wafer to the detection module.

The excitation light may have a wavelength in the range of 200 nm to 1,100 nm.

The lighting assembly may be arranged to irradiate excitation light to the wafer along an optical axis that is not perpendicular to the irradiation surface of the wafer. The illumination lens may have an optical axis that is nominally perpendicular to the illuminated surface of the wafer.

Among other advantages, the embodiments of the present invention can be used in the manufacturing process (for example, during or between the manufacturing process of the COS image sensor) and/or after the completion of the manufacturing process to identify the COS image sensor. Local defects (for example, part of the inspection after the manufacturing process). In some cases, the embodiments of the present invention can be used to identify defects related to a single pixel of a CMOS image sensor device, so that one or more individual defective pixels can be used for CMOS imaging. The image sensor device is recognized. In some cases, the embodiments of the present invention can be used to reduce the amount of positive values that may be additionally caused during the defect detection process due to the presence of particulate matter on the CMOS image sensor device. In some embodiments, the photoluminescence generated from crystal defects in the CMOS image sensor device can be separated from other sources of photoluminescence. For example, optical filtering can be used to distinguish photoluminescence at different energies. When photoluminescence from different sources (for example, from crystal defects versus photoluminescence from between energy bands), optical filtering can be used to isolate the photoluminescence from the crystal defects and thereby locate individual crystal defects in the device.

Generally speaking, bright field or dark field imaging can be used.

The details of one or more embodiments will be described below in conjunction with the drawings and description. According to the description, drawings and the content of the patent application, other features and advantages will be obvious.

100: system

110: platform components

132a-b: light source

134a-b: Correction lens

136a, 136b: filter

136a-b: filter

138, 152, 156: dichroic beam splitter

140: Focusing lens

150: optical components

154: contact lens

158: filter

160a, 160b: field lens

170: Imaging Kit

172a, 172b: detector

174: Processing Module

190: CMOS image sensor sample

200: photoluminescence spectrum

210: photoluminescence intensity

300: System

310: Illumination optical axis

320: imaging optical axis

330: lighting components

340, 342: Focusing lens

Figure 1 shows an exemplary system for detecting defects in CMOS image sensor samples.

Figure 2 is a photoluminescence intensity diagram showing changes in the photoluminescence intensity characteristics of defects in the silicon substrate.

Figure 3 shows another exemplary system for detecting defects in samples.

The same reference numerals in the various drawings indicate the same elements.

The device (such as CMOS image sensor device) consists of semiconductor Defects formed by the material can be identified by inducing photoluminescence in the active area of the device and checking the local intensity changes of photoluminescence.

For example, light with photo energy greater than the energy gap of silicon (for example, greater than 1.1 eV) can be used to induce photoluminescence in a silicon wafer. When the light is absorbed in silicon, electron-hole pairs are generated in the silicon. Some photo-generated carriers will recombine through radiative recombination and release photons. This phenomenon is called photoluminescence.

Since the intensity of photoluminescence changes according to the composition of the wafer, local changes in the composition of the wafer (for example, due to material defects or contamination) will result in local changes in the induced photoluminescence. Therefore, for example, it is possible to identify defects in the CMOS image sensor by irradiating the CMOS image sensor with excitation light sufficient to induce photoluminescence and checking the local change in the intensity of the photoluminescence, for example. Although the following description relates to defect assessment of CMOS image sensors, it is understood that the technology of the present invention can be more widely applied to other devices that use crystalline semiconductor materials that exhibit defect photoluminescence.

Figure 1 shows an example system 100 for identifying defects in CMOS image sensors. The system 100 includes a stage assembly 110, an illumination assembly 130, an optical assembly 150 and an imaging assembly 170. In an exemplary application of the system 100, a CMOS image sensor sample 190 is placed on the platform assembly 110 and positioned for inspection. The lighting component 130 generates excitation light suitable for inducing photoluminescence in the CMOS image sensor sample 190. The optical component 150 directs the excitation light generated by the illumination component 130 to the CMOS image sensor sample 190, thereby inducing photoluminescence in the CMOS image sensor sample 190 and/or causing the excitation light to be absorbed by the CMOS image sensor sample 190 reverse shoot. The optical component guides the photoluminescence generated by the CMOS image sensor sample 190 and/or the light reflected by the CMOS image sensor sample 190 to the imaging component 170. The imaging component 170 detects photoluminescence and reflected excitation light, and recognizes defects in the CMOS image sensor sample 190 based on the detected light.

During the inspection through the system 100, the platform assembly 110 supports the CMOS image sensor sample 190. In some cases, the platform assembly 110 can move along one or more axes so that the CMOS image sensor sample 190 can move relative to the illumination assembly 130, the optical assembly 150, and/or the imaging assembly 170. For example, in some cases, the CMOS image sensor can move along the x, y, and z axes of the Cartesian coordinate system to move the CMOS image sensor sample 190 along any three-dimensional relative to other components of the system 100 move.

When the illumination element 130 is incident on the CMOS image sensor sample 190, it generates excitation light that induces photoluminescence in the CMOS image sensor sample 190. The lighting assembly 130 includes light sources 132a-b, collimating lenses 134a-b, filters 136a-b, dichroic beam splitter 138, and focusing lens 140.

The light sources 132a-b generate light with specific properties suitable for inducing photoluminescence in the CMOS image sensor sample 190. In some cases, the light sources 102a-b are laser light sources that generate light with a specific wavelength and intensity. In some cases, each light source 132a-b generates light having a different wavelength, so that the lighting assembly 130 can provide different types of light. For example, the light source 132a may generate light having a first wavelength (for example, 532 nm), and the light source 132b may generate light having a second wavelength (for example, 880 nm).

In another embodiment, any one or both of the light sources 132a-b It can generate light with a wavelength less than 532nm (for example, 300nm, 350nm, 400nm, 450nm, 500nm or any other wavelength in between). In another embodiment, either or both of the light sources 132a-b can generate wavelengths between 200nm and 1100nm (e.g., 200nm, 300nm, 400nm, 500nm, 600nm, 700nm, 800nm, 900nm, 1000nm, 1100nm). Or other light of any wavelength in between). Although example wavelengths are described above, they are for illustrative purposes only. During implementation, the light sources 132a-b can each generate light with any other wavelength according to different implementations.

The light sources 132a-b can operate independently of each other, so that each light having a different wavelength can be generated individually or simultaneously. The light sources 132a-b include any element capable of generating light of a specific wavelength. For example, in some cases, the light sources 132a-b may include one or more laser lights or light emitting diodes (LEDs).

The light generated by the light sources 132a-b may also vary in intensity according to the implementation. In one embodiment, in some cases, the light sources 132a-b can each generate light with a power between 0.02W and 20W. In some cases, the intensity of the light generated by the light sources 132a-b can also be adjusted during the use of the system 100. For example, in some cases, the light generated by the light sources 132a-b can be adjusted between 0.02W and 20W during the operation of the system 100. In another embodiment, in some cases, the light generated by the light sources 132a-b can be adjusted during the operation of the system 100 to generate light with a power of less than 0.02W (for example, 0.015W, 0.010W, or 0.005W). . Although the strength of the example is described above, it is for illustrative purposes only. During implementation, the light sources 132a-b can each generate light with other intensities according to different implementations.

The excitation light system generated by the light source 132a-b is respectively directed and corrected Lens 134a-b. The correcting lenses 134a-b narrow the passing light beams so that the light emitted from the correcting lenses 134a-b is aligned along the optical axis of the correcting lenses 134a-b, respectively.

The corrected excitation light system from the correction lenses 134a-b is directed to the filters 136a-b, respectively. The filters 136a-b filter the passing light so that only light having a specific wavelength (or corresponding energy) or wavelength range can be substantially transmitted through the filters 136a-b, respectively. The filters 136a-b can be used to "clean" the light generated by the light sources 132a-b. For example, if the light source 132a generates light with a first wavelength (for example, 532nm), the filter 136a may be a bandpass filter that transmits light with the first wavelength (for example, 522nm to 542nm), and has a Light with wavelengths outside this range is basically not transmitted. In another embodiment, if the light source 132b generates light having a first wavelength (for example, 880 nm), the filter 136b may be a band pass filter that transmits light having a second wavelength (for example, 870 nm to 890 nm). However, light having a wavelength outside this range is basically not transmitted. In some cases, for example, in an embodiment where the light sources 132a-b include one or more laser lights, the filters 136a-b may also include speckle-reducing elements (such as moving diffusers). Components) to reduce the influence of interference effects in the laser beam.

The filtered excitation light from the filters 136a-b is directed to the dichroic beam splitter 138. The dichroic beam splitter 138 reflects light and/or transmits light according to the wavelength of the light incident thereon. For example, if the light source 132a generates light with a first wavelength (for example, 532 nm) and the light source 132b generates light with a second wavelength (for example, 880 nm), the dichroic beam splitter 138 may transmit light with the first wavelength and Reflects light having the second wavelength. The result is that although the light generated by each light source 132a-b is initially directed to a different direction, the dichroic beam splitter 138 redirects the light in a substantially similar direction.

The excitation light from the dichroic beam splitter 138 is guided to the focusing lens 140. The focusing lens 140 focuses the light toward the optical assembly 150.

The optical component 150 guides the excitation light generated by the illumination component 130 to the CMOS image sensor sample 190, and combines the photoluminescence generated by the CMOS image sensor sample 190 and/or the photoluminescence generated by the CMOS image sensor sample 190 The light guides the imaging assembly 170. The optical assembly 150 includes dichroic beam splitters 152 and 156, an objective lens 154, a filter 158, and field lenses 160a-b.

The excitation light from the focusing lens 140 is guided to the dichroic beam splitter 152. The dichroic beam splitter 152 reflects light and/or transmits light according to the wavelength of the light incident thereon. For example, if the light source 130a generates excitation light with a first wavelength (for example, 532 nm), the light source 130b generates excitation light with a second wavelength (for example, 880 nm), and the photoluminescence induced in the CMOS image sensor sample 190 With a third wavelength (for example, 1100 nm), the dichroic beam splitter 152 can partially reflect and partially transmit each light having these wavelengths. Therefore, the dichroic beam splitter 152 redirects at least some of the excitation light received from the illumination assembly 130 to the objective lens 154, and at least some of the photoluminescence induced in the CMOS image sensor sample 190 is transmitted through the dichroic The beam splitter 152 transmits to the imaging device 170, and at least some of the excitation light reflected by the CMOS image sensor sample 190 is also transmitted toward the imaging device 170.

The excitation light from the dichroic beam splitter 152 is guided to the objective lens assembly 154. The objective lens assembly 154 directs the excitation light to the CMOS image sensor sample 190. In some embodiments, the objective lens assembly 154 can stimulate The light is directed to a specific area of the CMOS image sensor sample 190 (for example, the area of the CMOS image sensor sample 190 being inspected), so that the intensity of the excitation light incident on the area of the CMOS image sensor sample 190 is uniform Or basically uniform. This area can be, for example, the whole of the CMOS image sensor sample 190 or a part of the CMOS image sensor sample 190.

The excitation light incident on the CMOS image sensor sample 190 can induce photoluminescence in the CMOS image sensor sample 190. In some embodiments, the photoluminescence in the CMOS image sensor sample 190 may have a wavelength between 950 nm and 1800 nm (for example, between 1100 nm and 1550 nm).

The objective lens assembly 154 can focus on specific areas of the CMOS image sensor sample 190 to obtain photoluminescence from these areas. In some embodiments, the objective lens assembly 154 may focus on a region of the CMOS image sensor sample 190 including one or more pixels of the CMOS image sensor sample 190, and the objective lens assembly 154 may include The wide-angle and shallow depth-of-field lens elements make it possible to analyze the photoluminescence from each pixel in the area. In some embodiments, the objective lens assembly has a focal length between 0.5 mm and 550 mm and a lens element with a depth of field between 1 μm and 400 μm. In some embodiments, the objective lens assembly 154 can analyze the light with sufficient analysis to distinguish the photoluminescence from each pixel. For example, if the CMOS image sensor sample 190 includes pixels having a size along the surface of 1 μm×1 μm, the objective lens assembly 154 can analyze the photoluminescence with a spatial resolution of 1 μm×1 μm or more.

The excitation light incident on the CMOS image sensor sample 190 can also cause reflection of the excitation light from the CMOS image sensor sample 190. The objective lens component 154 can also focus on a specific area of the CMOS image sensor sample 190 to obtain the excitation light reflected from the aforementioned area. Through the aforementioned method, in some embodiments, the objective lens assembly 154 can focus on the area of the CMOS image sensor sample 190 including one or more pixels of the CMOS image sensor sample 190, and the objective lens assembly The 154 may include a lens element with a wide angle and a shallow depth of field, so that it can analyze the excitation light reflected from each pixel in the area. In a manner similar to the foregoing, in some embodiments, the objective lens assembly 154 can analyze light with sufficient resolution to distinguish the excitation light reflected from each pixel. For example, if the CMOS image sensor sample 190 has pixels with a size of 1 μm ×1 μm along the surface, the objective lens assembly 154 can be analyzed with a spatial resolution of 1 μm ×1 μm or more. reflected light.

In some embodiments, the objective lens assembly 154 can be refocused to analyze the light from different areas in the CMOS image sensor sample 190. For example, in some embodiments, the focal depth of the objective lens element 154 can be changed to check the depth change of the photoluminescence from the surface of the CMOS image sensor sample 190 (for example, from the CMOS image sensor sample 190). The back surface to the front surface of the CMOS image sensor sample 190).

In some embodiments, the magnification of the objective lens element 154 can also be changed to inspect a specific area of the CMOS image sensor sample 190 with larger or smaller details. In some embodiments, the objective lens can be changed by moving the lens element of the objective lens assembly 154 and the opposite one (such as a "zoom" lens) or by modifying the optical path of the light passing through the lens assembly 154 in other ways. The magnification of component 154.

The photoluminescence and reflected excitation light is guided through the objective lens assembly 154 to the dichroic beam splitter 152. As mentioned earlier, there are 152 dichroic beamsplitters According to the wavelength of the light incident on it to reflect light and/or transmit light. For example, if the light source 130a generates excitation light with a first wavelength (for example, 532 nm), the light source 130b generates excitation light with a second wavelength (for example, 880 nm), and the photoluminescence induced in the CMOS image sensor sample 190 With a third wavelength (for example, 1100 nm), the dichroic beam splitter 152 can partially reflect and partially transmit each light of the aforementioned wavelengths. Therefore, at least some of the photoluminescence and the reflected excitation light can pass through the dichroic beam splitter 152 and transmit toward the imaging component 170.

At least a part of the photoluminescence and reflected excitation light is guided to the dichroic beam splitter 156 through the dichroic beam splitter 152. The dichroic beam splitter 156 further reflects light and/or transmits light according to the wavelength of the light incident thereon. For example, if the light source 130a generates excitation light with a first wavelength (for example, 532 nm), the light source 130b generates excitation light with a second wavelength (for example, 880 nm), and the photoluminescence induced in the CMOS image sensor sample 190 With a third wavelength (for example, 1100 nm), the dichroic beam splitter 156 can reflect the excitation light having the first and second wavelengths, and transmit the photoluminescence having the third wavelength. The result is that the photoluminescence and the reflected excitation light are redirected along different optical paths.

The photoluminescence transmitted by the dichroic beam splitter 152 is guided through the filter 158. The filter 158 filters the passing light so that only light having a specific wavelength or light of a specific wavelength range can be substantially transmitted through the filter 158. In some embodiments, the filter 158 can be used to "remove" the output of the dichroic beam splitter 152. For example, if the photoluminescence from the CMOS image sensor sample 190 is expected to have a specific wavelength (e.g., 1100 nm), the filter 158 can be configured to transmit a wavelength range including the photoluminescence wavelength (e.g., 1000 nm to 1200 nm). Bandpass filter of light, and having a wavelength outside the above range Basically, light cannot be transmitted. As another example, in some embodiments, the filter 158 may be a high-pass filter that attenuates light having a relatively short wavelength while transmitting light having a relatively long wavelength. For example, it is helpful to filter out the reflected excitation light that has a shorter wavelength than the photoluminescence from the CMOS image sensor sample 190 in many embodiments. Then, the photoluminescence and the reflected excitation light are directed to the field lenses 160a-b, respectively. The field lenses 160a-b focus the photoluminescence and the reflected excitation light to the detectors 172a-b of the optical assembly 170, respectively.

The imaging component 170 detects photoluminescence and reflected excitation light, and recognizes defects in the CMOS image sensor sample 190 based on the detected light. The imaging component 170 includes detectors 172a-b and a processing module 174.

The detectors 172a-b respectively measure the photoluminescence from the dichroic beam splitter 152 and the reflected excitation light. In some embodiments, the detectors 172a-b measure the intensity of light with a sufficiently high spatial resolution to analyze the photoluminescence of a single pixel of the CMOS image sensor sample 190 and the reflected excitation light. For example, if the CMOS image sensor sample 190 includes pixels with a size of 1 μm×1 μm along the surface, the detectors 172a-b can analyze the photoluminescence with a spatial resolution of 1 μm×1 μm or finer, respectively. In some embodiments, the detectors 172a-b may include a single detecting element or some similar detecting elements for measuring the intensity of light incident thereon. For example, in some embodiments, the detectors 172a-b may include a line of detection elements (such as a "line" detector) or a two-dimensional array of detection elements. In some embodiments, the detectors 172a-b may include one or more InGasAs-line cameras or arrays, or Si-line cameras or arrays.

In some embodiments, the detectors 172a-b integrate over a period of time The intensity of the light received in the chamber is used to measure the photoluminescence and reflected excitation light respectively. The integration time can be based at least in part on the intensity of light projected to the CMOS image sensor sample 190. For example, in some embodiments, reducing the intensity of the light incident on the CMOS image sensor sample 190 by a factor of two can increase the integration time by a factor of two. The measured noise from the detector increases with the integration time. In some embodiments, the intensity of the light projected onto the CMOS image sensor sample 190 can be adjusted to reduce the noise measured by the detectors 172a-b. Information is limited to an appropriate level. In some embodiments, the detectors 172a-b can be cooled to further reduce measurement noise. For example, in some embodiments, either or both of the detectors 172a-b may be cooled (for example, through a Peltier Cooler) to a specific temperature (for example, 100K) to reduce the gain The amount of noise in the measurement.

The measurement results from the detectors 172a-b are sent to the processing module 174 for interpretation. In some embodiments, the processing module 174 can generate one or more multiply dimensional maps representing the photoluminescence of a specific part of the CMOS image sensor sample 190 and the intensity of the reflected excitation light. For example, in some embodiments, the detectors 172a-b can include a two-dimensional array of detection elements, each of which can measure the intensity of light incident on the detection element. By using this information, the processing module 174 can generate a spatial map representing the photoluminescence of a specific location on the CMOS image sensor sample 190 and the intensity of the reflected excitation light.

The processing module 174 can also identify defects in the CMOS image sensor sample 190 based on the measurement results from the detectors 172a-b. For example, the processing module 174 can identify the area (such as spot, blot, line, Curves or other areas with stronger or weaker photoluminescence than surrounding areas). The processing module 174 can identify the above-mentioned area of the CMOS image sensor sample 190 as defective. In some embodiments, the processing module 174 can identify one or more specific pixels of the CMOS image sensor sample 190 as defective (for example, pixels related to local changes in photoluminescence).

In some embodiments, the local change in photoluminescence may not be a defect in the CMOS image sensor sample, but may be particulate matter on the surface of the CMOS image sensor sample. Since particulate matter can block or otherwise attenuate light, the presence of these particles can locally affect the intensity of the excitation light incident on the CMOS image sensor sample, and can cause local changes in photoluminescence. In order to distinguish between the defects of the local variation of photoluminescence and the particulate matter in the CMOS image sensor sample, the processing module 174 can determine whether the area of the CMOS image sensor sample 190 identified as having the local variation of photoluminescence is reflective The excitation light also has corresponding local changes.

In one embodiment, if the area has a local change in photoluminescence and a corresponding local change in reflected light, the processing module 174 determines that the change in photoluminescence is caused by the particulate matter on the surface of the CMOS image sensor sample. It is not a defect or contamination in the CMOS image sensor sample. Therefore, the processing module 174 can determine that there is no defect in the area.

In another embodiment, if the area has a local change in photoluminescence but does not have a corresponding local change in the reflected light, the processing module 174 determines that the change in photoluminescence is not the CMOS image sensor sample surface Caused by particulate matter on the surface. Therefore, the processing module 174 can determine that there is a defect in the area.

In some embodiments, the processing module 174 can use digital Implementation of electronic circuits, or computer software, firmware, or hardware, or a combination of one or more of the foregoing. For example, in some embodiments, the processing module 174 may be at least partially implemented as one or more computer programs (for example, one of the computer program instructions coded on a computer storage medium to execute or control its operation through a data processing device) Or multiple modules). The computer storage medium may be or may be included in a computer readable storage device, a computer readable storage substrate, a random or serial access memory array or device, or a combination of one or more of the foregoing. The term "processing equipment" includes all types of equipment, devices, and machines for processing data, including, for example, programmable processors, computers, system-on-chips, or multiple or combinations of the foregoing. The device may include dedicated logic circuits, such as FPGA (field programmable gate array) or ASIC (application-specific integrated circuit). In addition to the hardware, the device can also include codes for establishing the execution environment of the computer program in question, such as codes that constitute processor firmware, protocol stack, database management system, operating system, and cross-platform runtime environment , Virtual machine, or a combination of one or more of the foregoing. The equipment and execution environment can implement various computing model infrastructures, such as web services, distributed computing, and grid computing infrastructures.

Although the example system 100 has been shown and described, it is for illustrative purposes only. During implementation, the system 100 may have other configurations according to the implementation.

The implementation of the system 100 can be used during the manufacturing process of the CMOS image sensor device (for example, before, during, or after any step during the manufacturing process) and/or after the manufacturing is completed (for example, after forming an integrated circuit on a wafer) ) Identify local defects in the CMOS image sensor device. For example That is, in some embodiments, the implementation of the system 100 can be used to monitor one or more intermediate steps in the manufacturing process of one or more CMOS image sensor devices, and/or to check one or more completed CMOS image sensors. Detector device.

In some embodiments, when a defect in a CMOS image sensor device is detected, information about the location and nature of the defect can be used to modify the manufacturing process to reduce serious defects introduced into the CMOS image sensor device in the future . For example, information about defect detection can be used to identify the specific manufacturing equipment or process that partially or completely caused the defect. This information can then be used to repair and/or replace the equipment, or modify the process to improve the manufacturing process. In some embodiments, information about the location and nature of defects can also be used to identify defective wafers or parts of wafers, so that these wafers or parts of wafers can be discarded or not used in the future In the process.

The implementation of the system 100 can be used to identify local defects in a CMOS image sensor device with a spatial resolution of at least a single pixel of the CMOS image sensor device. For example, in some embodiments, a CMOS image sensor device constructed with 32nm technology has pixels of approximately 0.9×0.9μm; the implementation of the system 100 can be used to identify with a spatial resolution of 0.9×0.9μm or finer Local defects in the CMOS image sensor device.

Generally speaking, the system 100 can generate various types of excitation light according to the application. For example, in some embodiments, the system 100 can change the wavelength of the excitation light generated by the illumination module 130 to detect different depths under the surface of the CMOS image sensor sample 190. As an example, in some embodiments, the lighting module 130 can generate green light (for example, with a wavelength of about 532 or 540 nm) to generate a surface close to the CMOS image sensor sample 190 (for example, 1/(absorption) Coefficient) = 1.5μm) minority carriers and photoluminescence. On the other hand, in some embodiments, the illumination module 130 can generate near-infrared illumination (for example, having a wavelength of about 880 nm) to generate minority carriers farther from the silicon surface and photoluminescence. As described above, the lighting module 130 may include multiple light sources, and each light source may be selectively activated to generate light with different wavelengths. For example, in the example system 100 shown in FIG. 1, the lighting module 130 may include two light sources 132a-b, each of which is used to generate light with different wavelengths. Therefore, the light sources 132a-b can be selectively turned on or off to detect different depths under the surface of the CMOS image sensor sample 190. Although two light sources 132a-b are shown in the figure, in reality, the system 100 may include any number of light sources according to the implementation.

In some embodiments, the system 100 can change the intensity of the excitation light generated by the illumination module 130. In some embodiments, the excitation light generated by the illumination module 130 may have a high enough intensity to induce the photoluminescence in the CMOS image sensor sample 190, and also low enough to make the CMOS image sensor sample 190 There is basically no Auger recombination in the irradiated part.

Generally speaking, for low injection levels (for example, when the minority carrier concentration is less than the majority carrier concentration), the intensity of interband photoluminescence is related to the minority carrier concentration at a specific position and the majority carrier concentration. The product of the carrier concentration is proportional. For example, it can be expressed as: PL = A * C _{minority carrier} * C _{majority carrier} , where PL is the intensity of photoluminescence induced by a specific location (expressed as the number of photons), and C _{minority carrier} is the minority carrier at the above location The carrier concentration, C _{majority carrier} is the majority carrier concentration at the above position, and A is a constant.

The minority carrier concentration C _{minority carrier} (called the injection level) and the effective lifetime and generation rate of the minority carrier (that is, when a minority carrier exists, the number of photons adsorbed in silicon is normalized to that of silicon). Volume) is proportional. For example, it can be expressed as: C _{minority carrier} = R _generation * t _{life,effective} , where R _generation is the generation rate at the above position, and t _{life,effective} is the effective life of the minority carrier at the above position.

The intensity of the photoluminescence induced by the PL at the specific position is the same as the photoluminescence absorbed in the silicon at the above position-the intensity of the induced light (expressed as the number of photons), the effective lifetime of the minority carrier at the above position, and the above position The doping concentration in silicon is directly proportional. For example, it can be expressed as: PL = A * I _absorbed * t _{life, effective} * C _{majority carrier} , where I _absorbed is the photoluminescence absorbed in the silicon at the above position-the intensity of the induced light (expressed as the number of photons) .

Effective life t _{life, effective} has contributions from various recombination channels, especially bulk recombination, recombination at the interface, and Ojie recombination. For example, it can be expressed as: 1/ t _{life,effective} =1/ t _{recombination,bulk} +1/ t _{recombination,interfaces} +1/ tr _{ecombination,Auger} , where t _{recombination,bulk} is the life of a large number of composites, t _{recombination, Interfaces} is the composite life of the interface, t _{recombination, Auger} is the life of Aujie composite.

Defects in the pixel interface (for example, on the front surface, back surface, or wall of deep trench insulation (DTI)) reduce the effective lifetime t _{life,effective} in a given pixel and will cause the photoluminescence intensity from the pixel to decrease. For example, it can be expressed as: 1/ t _{life,interfaces} =1/ t _{recombination,front} +1/ t _{recombination,back} +1/ t _{recombination,DTI} , where t _{recombination,front} is the life of the front composite, t _{recombination, back} is the life of the back composite, t _{recombination, DTI} is the life of the deep trench insulation (DTI) wall.

其中d _interfaces 為介於介面之間之距離，v_{recombination,interface}為位於介面之重組速度。 The life t _{life, interfaces} of the interface recombination is inversely proportional to the recombination speed of the surface (that is, the interface) located on the interface and the distance between the aforementioned interfaces. For example, it can be expressed as:

Where d _interfaces is the distance between interfaces, and v _{recombination, interface} is the recombination speed at the interfaces.

In some embodiments, for deep trench insulation interfaces, the distance d _interfaces between the interfaces may be about 1 μm (that is, for a pixel having a size of 1 μm). The interface recombination rate of a good passivation interface is in the range of 1~10cm/sec. Assuming that the interface recombination rate is 10cm/sec, the recombination life t _{recombination of the deep trench insulation interface can be expected, and the DTI} is about 5×10 ^-6 seconds. If there are no large number of defects and the surface is well passivated, this will control the effective life t _{life,effective} in the pixel.

The effective life t _{life,effective} depends on the incident level (for example, the intensity of the light incident on the active area of the CMOS image sensor). Therefore, the effective life t _{life,effective} can be controlled at least in part by adjusting the injection level (for example, by adjusting the intensity of the excitation light that illuminates the CMOS image sensor sample). The injection level can be adjusted according to one or more standards.

For example, in some embodiments, the injection level can be adjusted to reduce the contribution of Ojie composite to the effective life of the CMOS image sensor in the active area. Under high injection level, the effective life can be controlled by Ojie compound. For example, in some embodiments, for an ^{injection level of 1×10 17} cm ⁻³ , Oge Composite limits the effective life in p-type silicon to 1×10 ⁻⁴ seconds. In another example, in some embodiments, for an ^{injection level of 1×10 18} cm ⁻³ , OJ composite limits the effective life in p-type silicon to 1×10 ⁻⁶ seconds. Since the Ojie compound is not sensitive to defects, the Ojie compound in the active area of the CMOS image sensor can not control the compound process by adjusting the injection level.

In addition, OJ Composite controls the effective life of highly doped substrates. For example, for p++ and n++ substrates, Ojie composite can limit the effective life of all injection levels. For example, in some embodiments, in a ^{p++ substrate with a carrier concentration of 1×10 20} cm ⁻³ , the effective lifetime is limited to 1×10 ⁻⁹ seconds. The Ojie composite lifetime changes drastically with the doping concentration. For example, a 10-fold reduction of the doping concentration (e.g., from 1×10 ²⁰ to 1×10 ¹⁹ doping concentration) will increase the effective lifetime by a hundred times (e.g., increase to 1×10 ⁻⁷ seconds).

The effective lifetime in the active area of the CMOS image sensor is based on the injection level in the low-injection level state (for example, when the contribution of Ogee recombination is not considered). Therefore, it is important to monitor the photoluminescence intensity as a function of the illumination level (ie, the incident level), because the lifetime corresponding to the incident level may be different for various defects. For example, for p-type silicon, defects such as interstitial Fe will increase the effective life as the injection level increases. However, defects such as Fe-B pairs will decrease the effective life as the injection level increases. For the increase of composite defects at the interface (such as SiO ₂ interface or the sidewall of deep trench insulation), the effect of changing the injection level on the composite life of the interface will depend on the state of the space charge region at the interface. For the interface located in the reverse phase of the low injection level, the interface compound life will not change with the increase of the injection level. When the injection level becomes higher (for example, greater than the majority carrier concentration), the lifetime decreases as the injection level increases. For the interface in the consumption of low injection level, the composite life of the interface increases as the injection level increases. However, for high injection levels, the composite life of the interface does not change with the increase in injection levels. Therefore, for a given defective pixel, the dependence of the photoluminescence intensity on the incident level can provide important clues to the nature of the defect.

As mentioned above, in some embodiments, it is important to include photoluminescence in the active region of the CMOS image sensor and to appropriately use low-injection levels. In addition, it is important to minimize or otherwise appropriately reduce the amount of light absorbed in the highly doped substrate. The photoluminescence intensity is proportional to the majority carrier concentration. Therefore, for the same amount of photons absorbed in the substrate and the epitaxial layer (such as the active area of the CMOS image sensor), the photoluminescence from the substrate may be stronger than the photoluminescence from the active area of the CMOS image sensor . For example, in the example CMOS image sensor device, the active area of the CMOS image sensor has an average doping concentration of ^{1×10 16} cm ^-3 at the beginning of processing, and has an effective of ^{5×10 -6 seconds} life. The substrate in this embodiment has an average doping concentration of ^{1×10 19} cm ^{-3 and an effective life of 1×10 -7} (controlled by Oujie). In some embodiments, it can cause about 20 μm The diffusion length (for example, the distance that minority carriers will diffuse). Given a similar amount of absorbed photons in the active area of the CMOS image sensor and the highly doped substrate, the photoluminescence from the substrate will be five times stronger than the photoluminescence of the active area of the CMOS image sensor. In order to reduce the background photoluminescence from the substrate to less than 5% of the photoluminescence from the active area of the CMOS image sensor, the amount of photons absorbed in the substrate can be limited to be less than the amount of photons absorbed in the active area of the CMOS image sensor 1% of the photon. Therefore, the wavelength of the light (for example, absorption coefficient) generated by photoluminescence can be appropriately selected.

As mentioned above, in some embodiments, it is also important to use the injection level to induce photoluminescence that does not control the lifetime in the active area of the CMOS image sensor sample (for example, when CMOS image sensor The contribution of Ou Jie compound in the device can be ignored.) As mentioned above, for ^{the injection level of 1×10 17} cm ^-3 , Ojie’s life span is 1×10 ^-4 seconds. However, for ^{the injection level of 1x10 18} cm ^-3 , Ojie's life span is 1x10 ^-6 . As an example, in some embodiments, for a CMOS image sensor active area with an effective (i.e., bulk and interface composite) lifetime in the range of ^{5×10 -6, 1×} The injection level of 10 ¹⁸ cm ^-3 , because at this injection level, the sensitivity to body phase and interface recombination will be lost, and the contribution to the effective life of the measurement will be less than 20%.

In some embodiments, it is also important to use an injection level that can prevent out-diffusion of minority carriers from the active region of the CMOS image sensor to the highly doped substrate to induce photoluminescence. Important concerns about the injection level (for example, the concentration of minority carriers generated in the active region of a CMOS image sensor) are related to The minority carriers generated in the active region of the CMOS image sensor are related to the out-diffusion to the highly doped substrate. Due to the difference in doping concentration between the highly doped p++ substrate and the lightly doped p-type material in the active region of the CMOS image sensor, the electric field existing at the p/p++ interface at the low incidence level will prevent a few Carriers diffuse into the substrate from the active area of the CMOS image sensor. However, the difference between the active area of the CMOS image sensor and the minority carrier concentration in the substrate will generate a driving force to diffuse the minority carriers from the active area of the CMOS image sensor to the diffusion field in the substrate. . As the injection level increases, the driving force for this diffusion will also increase. As long as the diffusion force is less than electronic repulsion, minority carriers will be contained in the epitaxial layer. For high incidence levels, the diffusion gradient can overcome electrical repulsion and some minority carriers can enter the substrate. As a result, strong background photoluminescence will be generated from the substrate. This will reduce the sensitivity of the photoluminescence that detects changes in the active area of the CMOS image sensor.

Therefore, in order to enhance the detection sensitivity of the system 100 and minimize the generation of photoluminescence in the highly doped substrate of the CMOS image sensor sample, an appropriate wavelength and intensity are selected for the excitation light applied to the CMOS image sensor sample Is important.

In some embodiments, the wavelength and intensity of the excitation light can be experimentally evaluated for each application. For example, whether the wavelength and intensity of the excitation light will cause photoluminescence mainly from the active area of the CMOS image sensor sample or whether the resulting photoluminescence has a large effect on the substrate of the CMOS image sensor sample. Experimental judgment of contribution.

In an exemplary evaluation process, a CMOS image sensor sample is irradiated with excitation light having a specific wavelength and intensity, and the CMOS image sensor sample is detected on a relatively large portion (such as a "macro region"). Measure the photoluminescence obtained. In some embodiments, the portion of the CMOS image sensor sample inspected in this way may be larger than the portion of the CMOS image sensor sample measured by the detectors 172a-b as described above. In some embodiments, the "macro" area may have an area of about 1 cm ² or greater (for example, 1 cm ² , 2 cm ² , 3 cm ² , 4 cm ² or greater). In some embodiments, the "macro" area may include the entire CMOS image sensor sample.

In some embodiments, the photoluminescence of the "macro" area can be determined using a detector separate from the detectors 172a-b. For example, a separate detector can be pointed at the CMOS image sensor sample 190 to obtain photoluminescence measurements along the "large" area of the detectors 172a-b. In some embodiments, the photoluminescence of the "macro" area can be determined by one of the detectors 172a-b. For example, in some embodiments, when imaging a "giant" area, the optical path between the CMOS image sensor sample 190 and the detectors 172a-b can be changed. For example, when imaging the "giant" area, By using a different objective lens 154 or adjusting the optical properties of the objective lens 154.

Inspect the obtained photoluminescence spectrum to obtain the intensity variation of defects usually related to the silicon substrate of the CMOS image sensor device. For example, silicon wafers (ie, "CZ wafers") manufactured by the Czochralski process usually include photoluminescence curves when irradiated by excitation light. As an example, Figure 2 shows an exemplary photoluminescence spectrum 200 of a CMOS image sensor device. In this example, the photoluminescence spectrum 200 includes a complex change of the photoluminescence intensity 210, which is represented by a circular or curved band of intensity change. In some embodiments, the change similar to that shown in Figure 2 is through the Czochralski method The feature of the manufactured silicon wafer and the feature pattern do not exist in the epitaxial layer of the device (for example, in the active area of the CMOS image sensor).

The CMOS image sensor sample can be irradiated with long-wavelength radiation (such as near-infrared radiation with energy greater than about 1.1eV, energy gap of Si) to quantitatively calculate the photoluminescence from the substrate for all detected photoluminescence The contribution of luminescence causes most of the carriers (for example, substantially most or substantially all of the carriers) to be generated in the substrate. Due to the characteristic substrate defect, the change in photoluminescence intensity for short wavelengths can be used as a reference for calculating the contribution of substrate photoluminescence to all detected photoluminescence.

As an example, when the excitation light having a relatively long wavelength (for example, excitation light having a wavelength that causes substantially most or substantially all of the carriers to be generated in the substrate of the CMOS image sensor sample) is used to illuminate the CMOS image sensor When measuring the sample, the defect contrast is 50% (for example, the photoluminescence intensity map includes local changes in intensity that make the surrounding intensity differ by 50%). However, when the CMOS image sensor sample is irradiated with excitation light having a relatively short wavelength, the defect contrast is 5% (for example, the photoluminescence intensity map includes a local change in intensity that makes the surrounding intensity differ by 5%). Therefore, in this embodiment, it can be estimated that for a relatively short wavelength, the contribution of the substrate to all photoluminescence is about 10% (for example, 5% divided by 50%).

If the substrate contributes too much to the photoluminescence, the measurement conditions can be adjusted (for example, by reducing the wavelength or light intensity of the excitation light applied to the CMOS image sensor sample) until the defect pattern is reduced or eliminated.

In some embodiments, the maximum injection limit that can be used to detect defects in a CMOS image sensor sample can be determined, at least in part, based on the Ogee composite lifetime that decreases as the injection level increases. As an example, for an ^{injection level of 1×10 17} cm ^-3 , the life of Oge in the example device is 100×10 ^-6 seconds. For the active area of the CMOS image sensor with an effective complex life of ^{5×10 -6} seconds from the body and interface complex, the OJ complex will contribute about 5% of the effective life.

The sensitivity of the system 100 to defects in the CMOS image sensor sample depends at least in part on the bulk phase and the interface recombination lifetime that are shorter than the Ogee recombination lifetime. In some embodiments, the wavelength and intensity of the excitation light can be changed to obtain a specific percentage of the contribution of Ogee recombination to the effective lifetime. For example, in some embodiments, the wavelength and intensity of the excitation light can be changed so that the Oge life of the active area of the CMOS image sensor is less than or equal to the effective recombination of the bulk phase and interface recombination from the active area of the CMOS image sensor 5% of life. In some embodiments, the threshold corresponds to an injection level of ^{about 1×10 17} cm ^{-3 or less.} For a CMOS image sensor active area with a thickness of about 5 μm, in some embodiments, this corresponds to an ^{absorption of approximately 1×10 19} photons/cm ² sec in the CMOS image sensor active area (for example, corresponding to 100 mW /cm ² power). Although an exemplary threshold is described above, it is for illustrative purposes only. In some embodiments, the wavelength and intensity of the excitation light can be changed so that the Oge life of the active area of the CMOS image sensor is less than or equal to the effective recombination life of the bulk phase and interface recombination from the active area of the CMOS image sensor. Percentage (e.g. 1%, 5%, 10%, 15% or any other percentage).

In the system 100 (Figure 1), the sample is irradiated with light that is nominally perpendicular to the surface of the sample. Therefore, in this geometric shape, the light reflected from the sample surface is collected by the objective lens of the system and transmitted to the detector. However, more Generally, it can also be realized through other configurations. For example, as shown in Figure 3, in some embodiments, oblique (rather than vertical) illumination may be used. Here, the system 300 includes an illumination assembly 330 for illuminating the CMOS image sensor sample 190 with excitation light incident along the optical axis 310 at a non-perpendicular angle to the surface of the sample. For example, the optical axis 310 may have an angle of 45° or greater (for example, 60° or greater, 70° or greater) with respect to the surface normal. Generally speaking, the incident angle should be such that little or no light reflected from the sample surface is collected by the objective lens 154.

The lighting assembly 330 includes light sources 132a and 132b, filters 136a and 136b, and a dichroic beam splitter 138. In addition, the lighting assembly 330 includes focusing lenses 340 and 342 for focusing the light from the light source collected by the beam splitter onto the sample 190.

The system 300 also includes an objective lens 154, a filter 158, and a field lens 160a. The objective lens 154 and the field lens 160a image the surface of the sample 190 onto the detector 172a. The detector communicates with the processing module 174.

Similar to the system 100 shown in Figure 1, the objective lens 154 has an optical axis that is nominally aligned perpendicular to the surface of the sample 190.

Due to the relative orientation between the illumination optical axis 310 and the imaging optical axis 320, the excitation light reflected from the surface of the sample 190 will not be collected by the objective lens 154 and will not be transmitted to the detector 172a. Only the photoluminescence, scattered light, and stray light from the sample 190 are transmitted to the detector 172a. Therefore, the image formed at the detector 172a is a dark field image. Therefore, even when the filter 158 is not used, the photoluminescence in the wafer originates from the bright area in the image.

Generally speaking, although some groups of the system 300 are shown in Figure 3 Components, but can also include other components not shown in the figure. For example, the system may also include a detector 172b and/or other optical elements (such as lenses, filters, and stops) for transmitting or imaging light from the sample 190.

In some embodiments, photoluminescence from different processes in the wafer can be distinguished, where different processes will cause photoluminescence of different wavelengths. For example, optical filtering can be used to distinguish between photoluminescence from crystal defects and photoluminescence from other sources (such as from inter-band transitions in silicon). Photoluminescence from crystal defects in silicon wafers (such as crystal grains or sub-grain boundaries, dislocation clusters, dislocation rings or deposits and/or overlap defects) usually occurs In the energy range of about 0.7eV to about 0.9eV. Conversely, photoluminescence of inter-band transitions in silicon usually occurs at wavelengths greater than about 1 eV. Therefore, the photoluminescence from these two different sources can be distinguished by blocking the light from one of the wavelength ranges of the detector. For example, an optical filter that blocks light with energy exceeding about 1 eV can be used to detect photoluminescence from crystal defects only, because photoluminescence from interband transitions in silicon has a value of about 1.1 at room temperature. eV or higher energy, and will be blocked by the above-mentioned optical filter. More generally, optical filters can be designed to block photon energy corresponding to inter-band transitions in various materials other than silicon, allowing other materials to be studied similarly. Through the above method, the above technology can be used to identify the existence and location of each crystal defect in a wafer (such as a silicon wafer). The above-mentioned position is displayed as a bright spot in the image obtained by blocking the photoluminescence from the inter-band transition by using the filter.

As an example, the filter 158 may include a band that substantially transmits light having an energy of about 0.7 eV to about 1 eV but blocks (for example, reflects or absorbs) light with a wavelength lower than about 0.7 eV and/or higher than about 1.0 eV. Pass filter. be usable The dichroic filter achieves the above-mentioned purpose. Therefore, only light with photon energy of about 0.7 eV to about 1 eV reaches the detector and contributes to the detected image.

In the embodiment, the optical filter 158 is a reflective optical filter, and the optical filter can be oriented at an angle corresponding to an angle of the optical path, so that the reflected light will not be directed back to the wafer. The filter 158 can be removed from the light path, for example, can be removed manually or automatically through an actuation component.

According to the content of the embodiment, the pass band of the filter 158 can be changed. For example, the passband has a full-width half maximum of about 0.3 μm (for example, from about 0.7 μm to about 1.0 μm), or smaller (for example, about 0.25 μm, about 0.2 μm, about 0.15 μm, About 0.1 μm, about 0.05 μm). The passband can be selected to selectively block the photoluminescence from another light source when the photoluminescence is transmitted from one light source. For example, a narrower pass band may allow different types of crystal defects to be distinguished.

Although the filter 158 is one implementation, it can also be implemented in other ways. For example, the position of the filter is not limited to the position described in the present invention. Generally speaking, the filter can be located anywhere in the light path from the wafer to the detector. In some embodiments, the filter may be located at the detector (for example, in combination with the detector). In some embodiments, the filter may be located in the pupil plane of the imaging system.

Monitoring the crystal defects of the wafer during different steps in the wafer processing can identify crystal defects (such as stacking faults, dislocation loops, or precipitates) caused by the processing. The above-mentioned defects can be identified with micron resolution. The processing steps that can be monitored include ion implantation steps, annealing steps, layer deposition steps (e.g. epitaxial growth, atomic layer deposition (ALD)), selective epitaxy (e.g. Si _1-x Ge _x ), oxidation steps, etching Step (plasma etching), and/or polishing step (e.g. chemical mechanical polishing)). Generally speaking, the description of the defect can be performed before and/or after any of the above processing steps.

Although the present invention describes an implementation for detecting defects in a CMOS image sensor device, it is for illustrative purposes only. When implemented, the implementation can be used to detect the minority carriers generated and the photoluminescence is basically limited to other devices or defects in the circuit in the active region of the device. For example, in some embodiments, the implementation can be used to detect defects in CMOS circuits constructed using fully depleted silicon on insulator (SOI) technology. In some embodiments, a wafer other than a wafer formed of pure silicon, such as a SiGe layer (such as Si _1-x Ge _x , where 0<x<1) may be used. Compound semiconductor wafers can also be used. For example, wafers formed from III-V or II-VI compounds can be characterized by using the technology of the present invention. More generally, the above-mentioned technology can be applied to various materials with crystal defect photoluminescence, especially semiconductor materials with crystal defect photoluminescence and interband photoluminescence occurring at different wavelengths.

Although the description contains many details, it is not intended to limit the scope of protection of the present invention, but can only be interpreted as a description of specific features of specific embodiments. It may also be combined with certain features described in the context of separate implementations in the specification. Conversely, various features described in the context of separate implementation can also be implemented in multiple embodiments separately or in any suitable sub-combination.

Although several embodiments of the present invention have been disclosed above, some changes and modifications can be made without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention should be defined as the scope of the appended patent application. allow.

100: System

110: platform components

130: lighting components

132a, 132b: light source

134a, 134b: correction lens

136a, 136b: filter

138, 152, 156: dichroic beam splitter

140: Focusing lens

150: optical components

154: contact lens

158: filter

160a, 160b: field lens

170: Imaging Kit

172a, 172b: detector

174: Processing Module

190: CMOS image sensor sample

Claims (52)

A method for identifying local crystalline defects in a single crystal silicon of a wafer, the method comprising: having sufficient light in the wafer to induce transitions from the bands of the single crystal silicon on the wafer The photoluminescence and the photoluminescence from one or more local crystal defects in the single crystal silicon on the wafer irradiate the single crystal silicon on the wafer with one wavelength and one intensity of excitation light; The excitation light reflected by a part of the circle corresponds to the photoluminescence irradiated from a part of the wafer in response to the illumination step to provide filtered photoluminescence, wherein the filtering step is basically derived from the photoluminescence from the filtering The band-to-band transition of the luminescent single crystal silicon removes the photoluminescence; directs the filtered photoluminescence to a multi-element detector with a spatial resolution of 1μm×1μm or less in the multi-element The above part of the wafer is imaged on the detector, wherein the detected filtered photoluminescence corresponds to the photoluminescence from the one or more local crystal defects in the single crystal silicon of the wafer And according to the detected filtered photoluminescence, the presence and location of the one or more local crystalline defects in the single crystal silicon of the wafer are identified with a spatial resolution of 1μm×1μm or less, The above-mentioned one or more local crystal defects correspond to bright areas in an image of the above-mentioned part of the above-mentioned wafer. The method according to claim 1, wherein the detected filtered photoluminescence includes light having an energy in a range of about 0.7 eV to about 0.9 eV. The method described in item 2 of the scope of patent application, wherein the wafer includes a silicon germanium layer. The method described in item 3 of the scope of patent application, wherein the silicon germanium layer includes Si _1-x Ge _x , where 0<x<1. According to the method described in item 3 of the scope of patent application, after executing one or more processing procedures on the above-mentioned wafer, the above-mentioned one or more local crystalline defects of the above-mentioned wafer are identified. According to the method described in item 5 of the scope of patent application, after the formation of the integrated circuit on the above-mentioned wafer is completed, the above-mentioned one or more local crystal defects of the above-mentioned wafer are identified. According to the method described in claim 2, wherein the filtering step basically blocks the detected light with energy greater than about 1.0 eV. According to the method described in claim 1, wherein the filtering step further includes filtering the excitation light reflected from the part of the wafer. The method described in item 1 of the scope of the patent application further includes detecting the excitation light reflected from the above-mentioned part of the above-mentioned wafer. The method described in item 9 of the scope of patent application further includes comparing the detected photoluminescence and the detected excitation light, and identifying the one or more local crystals in the wafer based on the comparison result defect. In the method described in claim 1, wherein the above-mentioned wafer is a wafer used in a CMOS image sensor. According to the method described in item 11 of the scope of patent application, after the formation of the integrated circuit on the above-mentioned wafer is completed, the above-mentioned one or more local crystal defects of the above-mentioned wafer are identified. The method described in claim 12, wherein the integrated circuit formed on the wafer includes a complementary metal oxide semiconductor circuit. The method described in item 11 of the scope of patent application, wherein the above-mentioned part of the above-mentioned wafer corresponds to one or more pixels of the above-mentioned CMOS image sensor. The method described in claim 12, wherein the step of identifying the presence and location of the one or more local crystal defects in the single crystal silicon of the wafer includes identifying one or more of the CMOS image sensors Defective pixels. The method described in item 12 of the scope of patent application, wherein the step of identifying the presence and location of the one or more local crystalline defects in the single crystal silicon of the wafer includes identifying one with a size of 1 μm or less or Multiple defects. The method described in item 8 of the scope of patent application, wherein the step of identifying the presence and location of the one or more local crystal defects in the single crystal silicon of the wafer includes identifying one or more with a size greater than 1 μm defect. According to the method described in item 1 of the scope of patent application, the above-mentioned wafer is a wafer used for power management integrated circuit devices. The method described in item 1 of the scope of the patent application further includes: forming a photoluminescence intensity map of the above-mentioned part of the wafer based on the above-mentioned photoluminescence irradiated from the above-mentioned part of the wafer; The above-mentioned excitation light reflected by the above-mentioned part of the wafer forms a reflection intensity map of the above-mentioned part of the wafer. The method described in item 19 of the scope of patent application, wherein the comparison comes from the above crystal The step of the detected photoluminescence of the above-mentioned part of the circle and the detected reflected excitation light from the above-mentioned area of the wafer includes: determining whether the photoluminescence intensity map is on a first stage of the wafer A position includes a first intensity change; when determining whether the photoluminescence intensity map includes the first intensity change at the first position of the wafer, it is determined whether the reflection intensity map is at the first position of the wafer Including a second intensity change; when it is determined that the reflection intensity map does not include the second intensity change at the first position of the wafer, it is determined that there is a defect in the first position of the wafer. The method described in claim 20, wherein the step of comparing the above-mentioned detected photoluminescence from the above-mentioned part of the above-mentioned wafer with the above-mentioned detected reflected excitation light from the above-mentioned area of the above-mentioned wafer It further includes: when determining that the reflection intensity map includes the second intensity change at the first position of the wafer, determining that there is no defect at the first position of the wafer. The method described in item 1 of the scope of the patent application further includes adjusting one of the characteristics of the above-mentioned excitation light. The method according to item 22 of the scope of patent application, wherein the step of adjusting the above-mentioned characteristics of the above-mentioned excitation light includes adjusting a wavelength of the above-mentioned excitation light. The method described in claim 23, wherein the wavelength of the excitation light is adjusted to increase the photoluminescence irradiated from a different part of the wafer. The method described in item 24 of the scope of patent application, wherein the above-mentioned wafer of the above-mentioned The different parts have different depths from the part of the wafer that reflects the excitation light. The method described in item 1 of the scope of patent application, wherein the above-mentioned excitation light has a wavelength in a range of 200 nm to 1100 nm. The method described in item 1 of the scope of patent application further includes performing a processing step on the above-mentioned wafer. The method according to the claim 27, wherein the above processing steps are selected from an ion implantation step, an annealing step, a layer deposition step, an oxidation step, a plasma etching step, and a polishing step. The method described in item 28 of the scope of patent application, wherein the above-mentioned layer deposition step is an atomic layer deposition step. According to the method described in claim 29, wherein the above-mentioned layer deposition step is an epitaxial layer deposition step. The method described in item 30 of the scope of patent application, wherein the above-mentioned epitaxial layer deposition step is used to form a layer with silicon or Si _1-x Ge _x , where 0<x<1. The method described in item 29 of the scope of patent application, wherein the above-mentioned layer deposition step is a selective epitaxy step. The method described in item 32 of the scope of patent application, wherein the above-mentioned selective epitaxy is Si _1-x Ge _x used in a channel or source/drain region of a complementary metal oxide semiconductor device. The method described in claim 29, wherein the above-mentioned etching step is a plasma etching step. The method described in item 27 of the scope of patent application, wherein the above-mentioned processing steps are performed after the above-mentioned crystal defects are identified. The method described in item 35 of the scope of patent application further includes: irradiating the above-mentioned processed wafer with excitation light; and identifying one of the above-mentioned processed wafers based on the photoluminescence from the above-mentioned processed wafer or Multiple additional defects. The method described in item 36 of the scope of patent application further includes comparing the above-mentioned crystal defects identified in the above-mentioned wafer with the above-mentioned additional defects. The method according to the first item of the scope of patent application, wherein the one or more local crystal defects correspond to the bright part in an image of a part of the wafer. The method described in item 1 of the scope of patent application, wherein after one or more processing steps are performed on the wafer, the one or more local crystal defects of the wafer are identified. The method according to item 39 of the scope of patent application, wherein after the above-mentioned wafer processing is completed, the above-mentioned one or more local crystalline defects of the above-mentioned wafer are identified. In the method described in item 1 of the scope of the patent application, the intensity of the excitation light is high enough to induce detectable photoluminescence from each crystal defect of the single crystal silicon, and it is also low enough to cause the damage of the wafer Auger recombination does not occur in the irradiated part. According to the method described in claim 1, wherein the step of identifying the location of the one or more local crystal defects includes generating a space map of the wafer, and identifying the one or more of the wafers based on the space map The location of a local crystal defect. The method described in item 1 of the scope of patent application further includes forming an image sensor using the above-mentioned wafer, wherein the above-mentioned one or more local crystals are identified The location of the defect includes identifying a defect related to a single pixel of the image sensor. The method described in item 1 of the scope of patent application further includes identifying one or more defective parts of the wafer according to the location of the one or more local crystal defects. The method described in item 1 of the scope of the patent application, wherein the one or more local crystal defects include a single stacking fault, a single dislocation, or a single precipitate. A system for identifying local crystalline defects in a single crystal silicon of a wafer, comprising: an illumination module for having sufficient energy to induce in the wafer from the single crystal silicon on the wafer The photoluminescence of the inter-band transition and the photoluminescence from one or more local crystal defects in the above-mentioned single-crystal silicon of the above-mentioned wafer have a wavelength and an intensity of excitation light irradiating the above-mentioned single-crystal silicon of the wafer ; A detection module, including a multi-element detector, the multi-element detector is used to detect the photoluminescence irradiated from a part of the wafer in response to the irradiation; imaging lens, used to The above-mentioned part of the wafer is imaged on the above-mentioned multi-element detector with a spatial resolution of 1μm×1μm or less; an optical filter is used to block the reflection from the above-mentioned wafer from the above-mentioned detection module to The excitation light of the detection module filters the photoluminescence irradiated from the part of the wafer before the detection module performs the detection to provide filtered photoluminescence to the detection module, wherein the filter Steps basically The interband transition of the single crystal silicon from the filtered photoluminescence removes the photoluminescence, wherein the detected filtered photoluminescence corresponds to the one from the single crystal silicon on the wafer Or photoluminescence with multiple local crystal defects; and a processing module for identifying the single crystal on the wafer with a spatial resolution of 1μm×1μm or less based on the detected filtered photoluminescence. The presence and location of the one or more local crystalline defects in the crystalline silicon, wherein the one or more local crystalline defects correspond to the bright areas in the image of the portion of the wafer. As in the system described in item 46 of the scope of patent application, the optical filter transmits the photoluminescence light corresponding to the one or more local crystal defects in the wafer to the detection module. According to the defect identification system described in claim 46, wherein the optical filter transmits light having an energy in a range of about 0.7 eV to about 1.0 eV to the detection module. The system described in the 48th patent application, wherein the optical filter basically blocks light from the detection module and having an energy greater than about 1.0 eV. The system described in item 46 of the scope of patent application, wherein the excitation light has a wavelength in a range of 200 nm to 1100 nm. According to the system described in claim 46, the illumination composition is configured to irradiate the excitation light to the wafer along an optical axis that is not perpendicular to an irradiated surface of the wafer. The system described in item 51 of the scope of patent application, wherein the above-mentioned lighting lens has There is an optical axis perpendicular to the above-mentioned illuminated surface of the above-mentioned wafer.

Images