Method for improving uniformity of optical properties of thin film devices

Abstract

The invention relates to a method for improving wafer level optical property uniformity of a thin film device. The method includes forming a first layer of dielectric overlying a first wafer and a second layer of dielectric overlying a second wafer. The method also includes measuring a refractive index profile of the second layer and measuring a first thickness profile of the first layer. The method also includes determining a second thickness profile of the first layer based on the refractive index profile and the first thickness profile. The method further includes non-uniformly and selectively removing material from the first layer based on the second thickness profile, resulting in a third layer in the second thickness profile, the third layer characterized by a spectral response having a characteristic wavelength uniformity superior to + / -2.5 nm across the first wafer.

Description

Technical Field

[0001] The present invention relates to a semiconductor device and a manufacturing method. Background Art

[0002] Optical Mux / Demux (Multiplexer / Demultiplexer) components are bulky and expensive and need to be implemented as separate components in the optical network. However, with the development of thin film integrated Mux / Demux components, it is now possible to integrate these components onto a single chip or substrate, which has significantly reduced the cost and size of wavelength division multiplexing (WDM) optical links.

[0003] Thin film integrated Mux / Demux components typically use waveguide technology to separate or combine light of different wavelengths. It can be implemented using a variety of technologies, such as silica on silicon, silicon on insulator, or polymer waveguides. The choice of technology will depend on factors such as the desired wavelength range, the desired level of integration, and cost constraints. Passive Mux / Demux based on SiN dielectric films has been widely studied as a potential solution. However, the yield of such devices is limited by manufacturing, resulting in non-established central wavelength shifts, which hinders their widespread adoption. Therefore, a new technology is proposed to improve the uniformity of optical properties at the wafer level of thin film optical circuits. Summary of the invention

[0004] In one aspect, the present disclosure relates to a method for processing a semiconductor device, the method comprising: providing a first wafer; providing a second wafer, the second wafer comprising a silicon material, wherein a size difference between the first wafer and the second wafer is less than 0.1%; forming a first layer of a dielectric on the first wafer and forming a second layer of the dielectric on the second wafer; measuring a refractive index profile of the second layer on the second wafer; measuring a first thickness profile of the first layer on the first wafer; calculating a second thickness profile of the first layer based on the refractive index profile and the first thickness profile; and removing material from the first layer based on a thickness difference from the first thickness profile to the second thickness profile, thereby obtaining a third layer having the second thickness profile, the third layer being characterized by a spectral response having a characteristic wavelength variation of less than 5 nm across the first wafer.

[0005] In another aspect, the present disclosure relates to a circuit comprising: a first layer of dielectric material on a portion of a first wafer, the first layer being characterized by a spectral response uniformity having a characteristic wavelength within + / - 2.5 nm of a target wavelength across the first wafer; wherein the first layer is made by the following steps: depositing the dielectric material on the first wafer in a first process; measuring a first thickness distribution across the first wafer; determining a second thickness distribution across the first wafer based on a refractive index distribution and the first thickness distribution, the second thickness distribution corresponding to the target wavelength; selectively removing the dielectric material based on the second thickness distribution to form the first layer; and patterning the first layer to define a plurality of dies, each die comprising an optical circuit based on the first layer.

[0006] In yet another aspect, the present disclosure relates to a method for processing a semiconductor device, the method comprising: providing a plurality of wafers; forming a first layer of a dielectric on each of the plurality of wafers in a first process; measuring a thickness distribution of the first layer across a first wafer selected from the plurality of wafers; patterning the first layer into a first plurality of optical circuit dies on the first wafer in a second process; measuring a spectral response of the first layer to obtain a wavelength distribution across the first wafer, the spectral response of the first layer being characterized by a wavelength of each of the first plurality of optical circuit dies; determining a correlation between a wavelength change and a thickness difference across the first wafer based on the wavelength distribution and the thickness distribution; determining a target thickness distribution of the first layer based on the correlation and a target wavelength; modifying the first layer on each of the plurality of wafers to obtain a second layer having the target thickness distribution; patterning the second layer into a second plurality of optical circuit dies on each of the plurality of wafers in the second process; and singulating each of the plurality of wafers to obtain the second plurality of optical circuit dies, each optical circuit being based on the second layer characterized by a spectral response having a center wavelength within + / - 2.5 nm of the target wavelength. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] A further understanding of the nature and advantages of particular embodiments may be achieved by reference to the remainder of the specification and to the drawings, in which like reference numerals are used to refer to similar components. In some instances, a sub-label is associated with a reference numeral to denote one of a plurality of similar components. When reference is made to a reference numeral without specifying an existing sub-label, it is intended to refer to all such plurality of similar components.

[0008] Figure 1A is an exemplary graph of the thickness distribution of the deposited dielectric layer on an 8-inch wafer.

[0009] Figure 1Bis an exemplary diagram of the refractive index distribution of a dielectric layer on an 8-inch wafer.

[0010] Figure 2 is an exemplary graph comparing the central wavelength variation of a dielectric layer-based demultiplexer before and after homogenization processing for each of two wafers in the same batch.

[0011] Figure 3A is an exemplary diagram of a wafer-level refractive index profile of a dielectric layer trimmed to achieve uniformity in optical properties in accordance with an embodiment of the present technology.

[0012] Figure 3B is an exemplary diagram of wafer-level thickness distribution of a dielectric layer trimmed to achieve uniformity of optical properties in accordance with an embodiment of the present technology.

[0013] Figure 4A is an exemplary graph of a first thickness profile of a deposited dielectric layer on a product wafer in accordance with an embodiment of the present technology.

[0014] Figure 4B is an exemplary graph of a second thickness profile of a dielectric layer modified for uniformity of optical properties according to an embodiment of the present technology.

[0015] Figure 5 is a flow chart of a method for improving wafer-level optical characteristic uniformity according to some embodiments of the present technology.

[0016] Figure 6 is a flow chart of a method for improving wafer-level optical characteristic uniformity according to some embodiments of the present technology.

[0017] Fig. 7A is an exemplary graph of output center wavelengths of thin film based in-wafer demultiplexers for all wafers before modifying film thickness in accordance with an embodiment of the present technology.

[0018] Figure 7B is an exemplary graph of output center wavelengths of thin film based in-wafer demultiplexers for all wafers after modification of dielectric layers in accordance with an embodiment of the present technology. DETAILED DESCRIPTION

[0019] The present technology relates to semiconductor devices and manufacturing techniques. In a specific embodiment, a method for improving wafer-level optical property uniformity is provided. The method includes measuring a thickness distribution and a refractive index distribution of a thin film of an optically transparent material formed on a wafer. A target thickness distribution can be determined based on the measured thickness distribution and the refractive index distribution, aiming to compensate for refractive index variations so as to achieve improved wafer-level optical property uniformity for all wafers in the same manufacturing process. Other embodiments exist.

[0020] In the following description, for the purpose of explanation, numerous details are set forth to provide a thorough understanding of the described embodiments. However, it should be apparent to those skilled in the art that other embodiments may be practiced without some of these details. Several embodiments are described herein, and although various features are attributed to different embodiments, it should be understood that features described with respect to one embodiment may also be combined with other embodiments. However, for the same reason, a single feature or several features of any described embodiment should not be considered essential to every embodiment of the present invention, as other embodiments of the present invention may omit such features.

[0021] For example, the term "circuit board", also referred to as a printed circuit board (PCB), including references to "primary circuit board" and "sub-dimension circuit board", refers to a flat, rigid board made of an insulating material, usually fiberglass or plastic, that contains a complex network of metallic paths or "traces" that form the circuitry of various electronic devices. Components such as resistors, capacitors, and integrated circuits are then mounted to the board and their leads are soldered to corresponding traces to form a functioning electronic circuit. "Primary" or "sub-dimension" refers primarily to the lateral dimensions of the circuit board, which is one of the focal features of the present disclosure.

[0022] When an element is referred to herein as being "connected" or "coupled" to another element, it is understood that the element may be directly connected to the other element or there may be intervening elements between the elements. In contrast, when an element is referred to as being "directly connected" or "directly coupled" to another element, it is understood that there may be no intervening elements in the "direct" connection between the elements. However, the presence of a direct connection does not exclude other connections in which there may be intervening elements. When an element is referred to herein as being "electrically coupled" to another element, it is understood that the element may be directly connected to the other element via an electrical conductor.

[0023] When an element is referred to herein as being "disposed" in a certain manner relative to another element (e.g., disposed on, between, below, adjacent, or in some other relative manner), it is understood that the element may be directly disposed relative to the other element (e.g., directly disposed on) or with intervening elements between the elements. In contrast, when an element is referred to as being "directly disposed" relative to another element, it is understood that there are no intervening elements in the "direct" instance. However, the presence of a direct disposition does not exclude other instances in which there may be intervening elements.

[0024] Similarly, when an element is referred to herein as being "joined" to another element, it is understood that the element may be joined directly to the other element (without any intervening elements) or with intervening elements between the joined elements. In contrast, when an element is referred to as being "directly joined" to another element, it is understood that there are no intervening elements in the "direct" joining between the elements. However, the presence of direct joining does not exclude other forms of joining in which intervening elements may be present.

[0025] Likewise, when an element is referred to herein as a "layer," it is understood that the layer may be a single layer or may include multiple layers. For example, a conductive layer may include a plurality of different conductive materials or multiple layers of different conductive materials, and a dielectric layer may include a plurality of dielectric materials or multiple layers of dielectric materials. When a layer is described as being coupled or connected to another layer, it is understood that the coupled or connected layer may include intervening elements that exist between the coupled or connected layers. In contrast, when a layer is referred to as being "directly" connected or coupled to another layer, it is understood that there are no intervening elements between the layers. However, the presence of directly coupled or connected layers does not exclude other connections in which intervening elements may exist.

[0026] Furthermore, the terms left, right, front, back, top, bottom, forward, reverse, clockwise, and counterclockwise are used for explanation purposes only and are not limited to any fixed direction or orientation. Rather, they are only used to indicate the relative position and / or direction between various components of an object and / or assembly.

[0027] In addition, the methods and processes described herein may be described in a particular order for ease of explanation. However, it should be understood that unless the context dictates otherwise, intervening processes may occur before and / or after any portion of the described process, and that various other procedures may be reordered, added, and / or omitted according to various embodiments.

[0028] Unless otherwise indicated, all numbers used herein to express quantities, sizes, etc. should be understood to be modified by the term "about" in all instances. In this application, unless otherwise specifically stated, the use of the singular includes the plural, and unless otherwise indicated, the use of the terms "and" and "or" means "and / or". In addition, the use of the terms "including" and "having" and other forms such as "include", "included", "has", "have" and "had" should not be considered non-exclusive. And, unless otherwise specifically stated, terms such as "element" or "component" cover both elements and components including one unit and elements and components including more than one unit.

[0029] As used herein, the phrase "at least one of" following a list of items (with the terms "and" or "or" separating any of the items) modifies the list as a whole, rather than each member of the list (i.e., each item). The phrase "at least one of" does not require selection of at least one of each of the listed items; rather, the phrase allows for a meaning that includes at least one of any of the items and / or at least one of any combination of the items. By way of example, the phrase "at least one of A, B, and C" or "at least one of A, B, or C" each refers to only A, only B, or only C; and / or any combination of A, B, and C. In instances where a selection of "at least one of each of A, B, and C" or, alternatively, "at least one of A, at least one of B, and at least one of C" is intended, it will be explicitly described as such.

[0030] In the context of thin film multiplexer / demultiplexer (mux / demux) devices, dielectric layers are components that facilitate the function of channeling multiple input signals into a single line (multiplexing) or separating a composite signal into multiple output lines (demultiplexing). The optical properties of the dielectric layers play a role in maintaining signal integrity, preventing leakage, and reducing crosstalk between different signal lines. In certain broadband applications such as optical communications, these dielectric layers can be used to construct waveguides and optical filters, enabling mux / demux devices to route signals based on their frequency or wavelength. Additionally, the dielectric constant of the layer can affect the performance of the device, such as signal speed, loss rate, and power handling capacity.

[0031] The specific composition, structure of the dielectric layers and their deployment within a mux / demux device will vary widely depending on the specific application, operating frequency and performance requirements. Accordingly, ongoing research and development continues to explore new materials and configurations for these dielectric layers to enhance the performance of mux / demux devices.

[0032] In an embodiment, the present disclosure provides a method for improving the uniformity of optical properties at a wafer level. The method includes providing a first wafer. The method also includes providing a second wafer, the second wafer may include a silicon material, and the size difference between the first wafer and the second wafer is less than 0.1%. The method also includes forming a first layer of a dielectric on the first wafer and forming a second layer of the dielectric on the second wafer. The method also includes measuring a refractive index distribution of the second layer on the second wafer. The method also includes measuring a first thickness distribution of the first layer on the first wafer. The method also includes calculating a second thickness distribution of the first layer based on the refractive index distribution and the first thickness distribution. The method also includes removing material from the first layer based on a thickness difference from the first thickness distribution to the second thickness distribution, thereby obtaining a third layer with the second thickness distribution, the third layer being characterized by a spectral response having a characteristic wavelength variation of less than 5 nm across the first wafer. As an example, the thickness difference between the first thickness and the second thickness may involve a gradual or non-gradual change in thickness.

[0033] Embodiments may include one or more of the following features. The method, wherein the first wafer may include a silicon-on-insulator wafer. The method, wherein a deposition process is selected from one or more methods including chemical vapor deposition (CVD), plasma enhanced chemical vapor deposition (PECVD), low pressure chemical vapor deposition (LPCVD), and atomic layer deposition (ALD). The first layer of dielectric and the second layer of dielectric may include one of non-stoichiometric dielectrics selected from SiNx, SiONx, SiOx, and amorphous Si, and the first layer and the second layer are formed by performing a deposition process.

[0034] The method may include reducing thickness variation of the second layer prior to measuring the refractive index profile. Measuring the refractive index profile may include using in-line ellipsometry. Measuring the first thickness profile may include using ellipsometry, reflectometry, or spectroscopy. Removing material may include using a focused positive ion beam to scan across the first layer on the first wafer with a vertical resolution of less than 1 nm and a lateral resolution of less than 1 um.

[0035] Ellipsometry is a powerful optical technique for characterizing the complex refractive index or dielectric function of thin layers. It measures the change in the polarization state of light after reflection or transmission and can be used to determine layer thickness, optical constants and other material properties. Ellipsometry is a non-destructive and highly sensitive method, and it has been widely used in many fields such as semiconductor technology, materials science and photovoltaics. Online ellipsometry provides the advantages of real-time monitoring during the manufacturing process, providing instant feedback that can be crucial for quality control and process optimization. Because online ellipsometry is a non-contact and non-destructive method, it does not interfere with the production process or damage the sample, which makes it a preferred choice for many industrial applications. Reflection measurement involves measuring the intensity of light reflected from a surface to determine the properties of the surface. It can be used to analyze surface roughness, layer thickness, refractive index and other optical properties of materials. Reflection measurement can be performed using various types of light including visible, infrared and ultraviolet, and it can be used in a wide range of applications. Spectroscopic measurement refers to a wide range of techniques for measuring and analyzing the spectrum of light or other types of electromagnetic radiation emitted or absorbed by a sample. Spectroscopic analysis can reveal a wealth of information about a sample, including its chemical composition, molecular structure, and physical properties. There are many different types of spectroscopic measurements, including mass spectrometry, atomic absorption spectrometry, and Raman spectrometry, each with its own set of applications and capabilities. Spectroscopic measurements are used in a variety of fields such as chemistry, physics, astronomy, environmental science, and medicine.

[0036] In addition, the method may include dicing the first wafer into first dies, the first die may include an optical circuit. The optical circuit may include a thin film filter. Calculating the second thickness profile may include calculating a first regional thickness profile corresponding to a first target wavelength on a first region of the first wafer and a second regional thickness profile corresponding to a second target wavelength on a second region of the first wafer. The method may include removing material from the first layer based on the first regional thickness profile on the first region and removing material from the first layer based on the second regional thickness profile on the second region, thereby obtaining a fourth layer on the first region and a fifth layer on the second region, the fourth layer being characterized by a first spectral response having a characteristic wavelength within + / -2.5 nm around the first target wavelength, and the fifth layer being characterized by a second spectral response having a characteristic wavelength within + / -2.5 nm around the second target wavelength. The method may include singulating the first wafer into a first plurality of dies from the first region and into a second plurality of dies from the second region, each of the first plurality of dies may include an optical circuit based on a first modified layer having the characteristic wavelength within + / - 2.5 nm around the first target wavelength, and each of the second plurality of dies may include an optical circuit based on a second modified layer having the characteristic wavelength within + / - 2.5 nm around the second target wavelength.

[0037] According to another embodiment, the present disclosure provides a wafer-level optical circuit. The circuit also includes a first layer of dielectric material on a portion of a first wafer, the first layer being characterized by a spectral response uniformity having a characteristic wavelength within + / - 2.5 nm of a target wavelength across the first wafer. The circuit also includes, wherein the first layer is made by the following steps: depositing the dielectric material on the first wafer in a first process; measuring a first thickness distribution across the first wafer; determining a second thickness distribution across the first wafer based on a refractive index distribution and the first thickness distribution, the second thickness distribution corresponding to the target wavelength; selectively removing the dielectric material based on the second thickness distribution to form the first layer; and patterning the first layer to define a plurality of dies, each of which may include an optical circuit based on the first layer.

[0038] Embodiments may include one or more of the following features. The circuit, wherein the first layer may include one of a non-stoichiometric dielectric selected from SiNx, SiONx, SiOx, and amorphous Si, and the first wafer may include an SOI wafer. The refractive index profile is determined by the following steps: depositing the dielectric material on a second wafer in a second process arranged with the same settings as the first process, the second wafer being a blanket silicon wafer; reducing a thickness variation of a second layer on the second wafer; and measuring the refractive index profile of the second layer. The optical circuit may include an optical multiplexer or demultiplexer for a coarse wavelength division multiplexing (CWDM) optical link or for a dense wavelength division multiplexing (DWDM) optical link.

[0039] According to yet another embodiment, the present disclosure provides a method for processing a semiconductor device. The method also includes providing a plurality of wafers. The method also includes forming a first layer of a dielectric on each of the plurality of wafers in a first process. The method also includes measuring a thickness distribution of the first layer across a first wafer selected from the plurality of wafers. The method also includes patterning the first layer into a first plurality of optical circuit dies on the first wafer in a second process. The method also includes measuring a spectral response of the first layer to obtain a wavelength distribution across the first wafer, the spectral response of the first layer being characterized by a wavelength of each of the first plurality of optical circuit dies. The method also includes determining a correlation between a wavelength difference and a thickness difference across the first wafer based on the wavelength distribution and the thickness distribution. The method also includes determining a target thickness distribution of the first layer based on the correlation and a target wavelength. The method also includes modifying the first layer on each of the plurality of wafers to obtain a second layer having the target thickness distribution. The method also includes patterning the second layer into a second plurality of optical circuit dies on each of the plurality of wafers in the second process. The method also includes singulating each of the plurality of wafers to obtain the second plurality of optical circuit dies, each optical circuit being based on the second layer characterized by a spectral response having a center wavelength within + / - 2.5 nm of the target wavelength.

[0040] The method, wherein each of the plurality of wafers may include a silicon-on-insulator (SOI) wafer. The first layer of dielectric may include one of a non-stoichiometric dielectric, including SiNx, SiONx, SiOx, amorphous Si. The process for forming thin film integrated optical circuits, such as Mux / Demux components, etc., begins with preparing a suitable substrate (e.g., silicon wafer, silica, glass, and more recently, silicon-on-insulator (SOI) wafer) for CMOS compatible processes. Then, a thin layer of optical core material is deposited onto the substrate using techniques such as sputtering, chemical vapor deposition (CVD), spin coating, etc. The optical core material is typically a high refractive index non-stoichiometric dielectric material, such as silicon nitride (SiNx), silicon oxynitride (SiONx), or doped silica, which transmits light within a thin film waveguide structure or selectively reflects light out, like a demultiplexer.

[0041] For many applications such as wavelength division multiplexing (WDM) optical circuits, SiNx based thin film passive Mux / Demux has been widely considered. Due to the film deposition tool configuration, the thickness and refractive index of SiNx films have a wafer process distribution that can cause large wavelength variations, resulting in central wavelength errors and limiting the yield of such devices. For example, Figure 1A An exemplary graph showing the thickness distribution of a SiNx film on an 8-inch wafer. Thickness distribution refers to the variation in thickness of a dielectric layer (i.e., SiNx film) from point to point across the wafer surface, as seen up to 32 nm in size. This property is one of many factors that directly impacts the performance, yield, and reliability of devices fabricated on the wafer. Improving the uniformity of dielectric layer thickness across the wafer is an ongoing challenge in semiconductor manufacturing. Figure 1B An exemplary diagram showing a refractive index profile of a SiNx layer on an 8 inch wafer with a range between 1.980 and 1.956. The refractive index profile refers to the variation in the refractive index of a dielectric layer from point to point across the wafer surface. This is a critical parameter in many optical and electronic applications because the refractive index affects the behavior of light passing through the dielectric layer. This variation can lead to performance inconsistencies in devices or systems that utilize dielectric layers. Because SiNx layers are typically non-stoichiometric, having varying compositions, defects, and stresses due to variations in energy, temperature, and chemical reactions that occur during the CVD deposition process, non-uniformity in the refractive index profile at the wafer level exists even when the layer can be obtained to a highly uniform thickness in any existing thin film homogenization process.

[0042] When light of a particular wavelength (λ) passes through a layer, if the thickness of the layer (d) is an integer multiple of half the wavelength of the light (considering the wavelength within the material, λ / n), then constructive interference occurs and the light is transmitted at its maximum intensity. This can be expressed as d=m*λ / (2n), where n is the refractive index of the dielectric layer and m is an integer representing the interference order. The layer will transmit light most efficiently at a wavelength λ, where d is an integer multiple of λ / (2n). This wavelength is the channel center wavelength. Therefore, if the thin film formation process for the dielectric layer results in non-uniformity in the thickness d of the dielectric layer or its refractive index n across the wafer, then this non-uniformity will affect the spectral response of the dielectric layer, for example, the channel center wavelength of a dielectric layer to be used as an optical filter.

[0043] Figure 2 is an exemplary diagram showing a comparison of the center wavelength variation of a SiNx layer-based demultiplexer before and after a homogenization process is performed on each of two wafers in the same batch. Before the thickness homogenization process, the layer thickness cannot be uniform in any way, and the center wavelength can vary from wafer to wafer by as much as 8nm to 10nm. However, after the thickness homogenization process, although the layer thickness is now uniform, the center wavelength variation is still close to 10nm due to large contributing factors from material properties (e.g., refractive index variation of the layer), which is unacceptable for integrated WDM applications.

[0044] To achieve wavelength uniformity of the spectral response of a thin layer, it ultimately depends on how to manage the phase of the light along its path through the layer. The phase can be affected by many factors such as waveguide width, sidewall angle, layer thickness, refractive index, process or temperature induced stress, etc. In some embodiments, the present disclosure provides a technique for improving the uniformity of the optical properties of a thin layer by engineering the layer thickness to reduce or compensate for all forms of phase errors due to refractive index, width, sidewall angle, and process related factors. Specifically, the layer thickness distribution of a dielectric layer on a wafer is locally modified to obtain a non-uniform target thickness distribution in order to compensate for the known non-uniform refractive index distribution across the wafer to achieve a wafer-level uniform wavelength target. In general, a dielectric thin layer on a wafer can be used to form many types of integrated planar optical circuits on the wafer, including WDM Mux / DeMux, splitters, combiners, filters, modulators, and detectors. The present disclosure uses only WDM Mux / DeMux as a demonstration example of the application of the proposed technology to improve the wafer-level uniformity of its characteristic center wavelength. The SiNx layer is also only an example in the present disclosure, as this technique should be generally applicable to all non-stoichiometric dielectric layers used to form integrated optical circuits on a wafer, such as SiONx, SiOx, or amorphous Si, etc. In general, the layers or thin films used to achieve multiple various channel wavelengths for Mux / DeMux can be a single layer or multiple layers with different optical properties. Considering that the thickness scale of the trimmed layer can mostly be around the range of 10nm or less and the film material is non-stoichiometric in nature, an ion beam trimming tool compatible with CMOS technology can be used in-line to trim the layer or mainly a single layer without significantly increasing the manufacturing cycle time.

[0045] In an embodiment, the phase error caused by the local refractive index variation of the non-stoichiometric dielectric material layer on the wafer can be compensated by trimming the layer thickness from the selected local area to achieve uniform optical properties across the wafer. The wafer-level refractive index distribution of the dielectric layer can be measured using an industry standard ellipsometer. The wafer-level thickness distribution of the layer can also be measured using online ellipsometric measurement or reflectometry or spectroscopy. Since the non-uniformity of the refractive index is mainly due to the composition of the dielectric layer applied by the deposition process of the dielectric material and is not sensitive to thickness, the refractive index distribution measured before trimming the layer thickness will be substantially the same as the distribution after trimming. In a specific embodiment, the real part n and the imaginary part k of the refractive index (n, k) are accurately measured for both the layer or film deposited on the product wafer and the reference wafer (usually a simple silicon wafer is used). Compare and ensure the correlation between the two refractive index distributions. Then, the target thickness distribution can be calculated based on the measured refractive index distribution on the reference wafer before trimming and the thickness distribution on the product wafer in order to trim the dielectric layer. Therefore, the required thickness difference across the local area of ​​the product wafer can then be determined.

[0046] Figure 3Aand 3B Both the wafer-level refractive index profile and the thickness profile of a trimmed or modified dielectric layer according to the desired thickness difference across the wafer are shown. Both profiles are normalized to the respective averages for an 8-inch (or 12-inch) diameter wafer. Only the real part n of the refractive index (n, k) is included in the profile map. Figure 3A and Figure 3B A comparison of clearly shows the complementary nature between the two profiles, for example, the higher refractive index values ​​near the lower right region of the wafer are compensated by relatively smaller thickness values ​​in the same region. This results in an improved homogenization of the phase, resulting in an improved uniformity of optical properties at the wafer level.

[0047] Figure 4A An exemplary diagram showing a first thickness distribution of a formed dielectric layer on a product wafer according to an embodiment of the present technology is shown. The thickness shown in the figure is normalized to an average value across the product wafer. In an example, the dielectric layer is a SiNx film intended to form a thin film optical filter of a WDM demultiplexer. The product wafer is an SOI wafer. A thin film optical filter is a device that selects or blocks light of a specific wavelength (or wavelength range) while allowing other light to pass. It does this by utilizing the interference of light waves reflected from the interface of thin layers or films of materials with different refractive indices. In an embodiment, based on the first thickness distribution and the refractive index distribution of the dielectric layer, a second thickness distribution that can compensate for the refractive index in all local areas can be calculated. Optionally, a compensation mask can be made based on the difference between the second thickness and the first thickness and the compensation mask can be applied to overlie the layer to perform non-uniform thickness trimming in order to obtain enhanced uniformity of optical properties. In some embodiments, an ion beam trimming (IBT) tool can be implemented without the need for mask mapping to scan across the wafer using a focused positive ion beam with controlled beam energy. Higher beam energy plus longer beam exposure time will result in a larger amount of dielectric material being removed to achieve thickness adjustment. For comparison, Figure 4B An exemplary diagram showing a second thickness distribution of a layer modified for uniformity of optical properties according to an embodiment of the present technology is shown. The second thickness distribution is also normalized to the average value of the first thickness before modification. The above layer modification process is essentially a trimming process, which results in a substantially smaller second thickness. Figure 4B As seen in Figure 1, about 8% or less of the layer thickness is trimmed. Using an ion beam trimming tool, the layer thickness can be very accurately controlled with a vertical thickness resolution of less than 1 nm and a lateral resolution of less than 1 um across an 8-inch wafer. Figure 3A and 3B , Figure 4B The trimmed thickness distribution shown in can properly compensate for the corresponding refractive index distribution of the dielectric layer formed on the same type of product wafer in the same deposition process to achieve wafer-level uniformity of optical properties of the trimmed layer.

[0048] Figure 5 1 is a flow chart of a method for improving wafer-level optical property uniformity according to an embodiment of the present technology. This figure is only an example and should not unduly limit the scope of the claims. A person of ordinary skill in the art will recognize many variations, substitutions, and modifications. Figure 5 Only a method or process for improving wafer-level thin film optical property uniformity by non-uniformly engineering layer thicknesses to compensate for corresponding refractive index non-uniformity is summarized.

[0049] Method 500 includes step 502 of providing a first wafer, which is a product wafer for making integrated optical circuits based on dielectric layers or thin films formed in a predetermined process. In an exemplary embodiment, the product wafer is a silicon-on-insulator (SOI) wafer. An SOI wafer is an advanced form of silicon wafer with a unique structure consisting of three layers: a thin surface silicon layer; an intermediate insulating material layer, which is typically silicon dioxide; and a thicker silicon substrate at the bottom, which is typically used to make thin-film integrated optical circuits based on a CMOS technology platform. Of course, other types of wafers may be used. The claimed technology is not limited to the described types of product wafers.

[0050] The method 500 includes step 504 of providing a second wafer. The second wafer serves as a reference wafer or test wafer having the same diameter as the first wafer. In general, typical dimensional differences in wafer diameter may be limited to 0.1% or less. In some embodiments, the second wafer is a simpler wafer than the first wafer, for example, it is a blanket silicon wafer. A silicon wafer in its simplest form is a thin slice of crystalline silicon having a well-polished surface and used as a base substrate for semiconductor devices such as integrated circuits (ICs). Using a silicon wafer instead of an SOI product wafer may reduce some errors or noise caused by multilayer scattering when measuring the refractive index profile of a dielectric layer on the wafer and reduce costs.

[0051] At step 506, method 500 includes forming a first layer of dielectric overlying the first wafer and a second layer of dielectric overlying the second wafer. In an embodiment, the first wafer and the second wafer are placed in the same manufacturing process environment (e.g., the same batch) in a deposition chamber. The first layer and the second layer include the same type of dielectric material selected for forming an integrated optical circuit, for example, a non-stoichiometric dielectric, including SiNx, SiONx, SiOx, and amorphous Si or other. In a specific embodiment for depositing the non-stoichiometric dielectric material, one of the following deposition methods is used, including chemical vapor deposition (CVD), plasma enhanced chemical vapor deposition (PECVD), low pressure chemical vapor deposition (LPCVD), and atomic layer deposition (ALD), etc. In an example for making a WDM Mux / DeMux optical circuit, a SiNx layer is formed.

[0052] At step 508, method 500 includes measuring the refractive index distribution of the second layer on the second wafer. Use ellipsometry to perform measurements that can be performed online after the layer is formed through a deposition process. Online ellipsometry is a powerful non-destructive optical characterization technique that provides real-time in-situ measurements of thin film properties (e.g., thickness, refractive index) by detecting polarization changes of a light beam reflected from a layer. Optionally, in order to eliminate the influence of thickness factors on the results, a thin layer homogenization process is first performed on the layer overlying the second wafer to make the layer thickness substantially uniform across the second wafer. Optionally, a focused ion beam with controlled beam energy is used in the thin layer homogenization process, which has a final vertical resolution of less than 1nm to ensure the uniformity of the layer. Using a blanket silicon wafer for the second wafer also eliminates complex multiple reflections from other layers on a typical SOI wafer. Therefore, the refractive index distribution of the first layer is measured. The data of the refractive index distribution is saved.

[0053] At step 510, method 500 also includes measuring a first thickness distribution of a first layer of a dielectric on a first wafer. The layer thickness can be measured by ellipsometry, reflectometry, and spectroscopy, all of which are non-destructive to the layer and performed in situ. The first thickness distribution is typically non-uniform. Figure 4A An example of a first thickness profile of this dielectric layer overlying an 8 inch SOI wafer is shown. As a result of this step, a first thickness profile is measured and the data is also saved.

[0054] The method 500 further includes a step 512 for calculating a second thickness profile of the first layer based on the refractive index profile and the first thickness profile. Since the first layer and the second layer are the result of the same batch on the same size wafer, the data of the refractive index profile of the second layer can be the same as the first layer, and the error can be ignored. Combining the data of the refractive index profile and the first thickness profile, the second thickness profile can be calculated so that the second thickness profile of the modified layer can effectively compensate for the non-uniform refractive index profile to achieve uniformity of optical properties across the wafer. In a specific embodiment, the optical property is a spectral response wavelength, for example, a center wavelength of an output of a WDM demultiplexer based on the modified layer. In some embodiments, a desired thickness difference from the first thickness profile to the second thickness profile across the first wafer is obtained, and the desired thickness difference can be used to design a refractive index (RI) compensation lookup table to guide the modification of the first layer. For example, the desired thickness difference across the first wafer is used as input data for programming a trimming tool to trim the first layer accordingly.

[0055] At step 514, method 500 includes selectively removing material from the first layer based on the thickness difference from the first thickness distribution to the second thickness distribution to obtain a modified layer with uniformity of optical properties. This is a step for engineering the layer thickness to improve the uniformity of optical properties. Optionally, after step 512, an RI compensation lookup table is applied to perform non-uniform thickness trimming on the first layer to obtain a modified layer with relatively improved wafer-level uniformity of optical properties. In an embodiment, an ion beam trimming tool is used to trim the first layer. Ion beam trimming (IBT) is based on ion beam etching of materials from a wafer by ion bombardment using a small positively charged ion beam. Other variants of IBT, including reactive IBT or chemically assisted IBT, can be used to selectively enhance throughput or material etching. During the trimming process, a focused ion beam is scanned across the wafer. By controlling the beam energy and adjusting the local beam dwell time, material thickness and device properties, such as Mux and Demux wavelengths, can be precisely modified. Under beam energy control and scanning speed (i.e., exposure time) control, vertical or thickness variations can be controlled with sub-nanometer resolution. The modified layer can be formed by using IBT to selectively remove dielectric material from the first layer to achieve a second thickness profile designed to target a characteristic response wavelength, such as a center wavelength of a demultiplexer based on the modified layer. Since optical characteristic uniformity is achieved at the wafer level, the modified layer on the first wafer also becomes a natural wafer-level optical circuit characterized by characteristic wavelength uniformity.

[0056] In certain embodiments, some additional downstream steps ( Figure 5The first wafer overlaid with the modified layer is further processed in a process (not shown) including patterning the modified layer to define a plurality of optical circuit dies on the first wafer and singulating the first wafer to obtain individual optical circuits. In an example, each optical circuit includes a demultiplexer or multiplexer with one or more wavelength channels (each having a center wavelength separated by 20 nm) for a coarse wavelength division multiplexing (CWDM) optical link. Or each optical circuit may include a demultiplexer or multiplexer with one or more wavelength channels (having more narrow wavelength bands) for a dense wavelength division multiplexing (DWDM) optical link. In some embodiments, the uniformity of the optical characteristics of these demultiplexers or multiplexers based on the modified layer across the wafer may be indicated by the degree to which their center wavelengths are targeted at a specified CWDM (or DWDM) channel wavelength. For example, applying method 500 to trim SiNx films to a target thickness profile across a wafer results in a plurality of demultiplexers based on trimmed SiNx films characterized by corresponding center wavelength variations that differ by less than 5 nm (or in one example, + / - 2.5 nm) around a specified CWDM channel wavelength.

[0057] In some embodiments, the method 500 disclosed above may have other variations. For example, one or more steps may be added, removed, repeated, rearranged, modified, replaced, and overlapped, and this should not limit the scope of the claims. In an embodiment, step 512 for calculating the second thickness distribution may include calculating the first region of the first wafer (represented by Figure 4A The first region thickness distribution corresponding to the first target wavelength on the first region surrounded by the dotted line shown in FIG. 1 and also shown in FIG. Figure 4A The second region of the first wafer (denoted by Figure 4A Then, the method may include the following steps: selectively removing material from the first layer based on the first regional thickness distribution on the first region and selectively removing material from the first layer based on the second regional thickness distribution on the second region. This step results in the first region (represented by Figure 4B The first modified layer having the first region thickness distribution on the first region (the first region surrounded by the dotted line in FIG. 1 ) and the second region (represented by Figure 4BThe method further comprises the steps of singulating the first wafer into a first plurality of dies from the first region and singulating the first wafer into a second plurality of dies from the second region. Each of the first plurality of dies comprises an optical circuit based on the first modified layer having a characteristic wavelength within a range of 5 nm around the first target wavelength, and each of the second plurality of dies comprises an optical circuit based on the second modified layer having a characteristic wavelength within a range of 5 nm around the second target wavelength.

[0058] In an alternative embodiment, the present disclosure provides a technique for performing non-uniform thickness trimming of dielectric layers or films based on characteristic wavelengths to improve wafer-level optical property uniformity. In this method, a target thickness distribution of a dielectric layer is determined by compensating for process-induced wavelength variations rather than compensating for refractive index variations. Due to tool limitations, obtaining accurate refractive index and thickness across a wafer may not always be feasible. Measuring the optical center wavelength of a Mux / Demux circuit (based on a layer) can provide another feedback for performing phase trimming to improve optical property uniformity, because the wavelength distribution of a wafer batch processed in the same time window is mostly consistent. Specifically, this method first measures the wavelength distribution of the final Mux / Demux circuit based on the first layer of a pre-delivered wafer selected from a wafer batch. Then, the correlation between the measured wavelength distribution and the corresponding thickness distribution can be determined to generate a target thickness distribution of the layer to be trimmed in all subsequent product wafers of the same wafer batch. Then, the layer on the subsequent product wafer is non-uniformly trimmed based on the target thickness distribution to compensate for the wavelength variation, for example, to allow the final Mux / Demux circuit to achieve a center wavelength target better than + / -2.5nm. This leads to an alternative approach for achieving wafer-level optical property uniformity by non-uniform thickness trimming, which will be presented in the next section.

[0059] Figure 6 1 is a flow chart of a method for improving wafer-level optical property uniformity according to some other embodiments of the present technology. This figure is only an example and should not unduly limit the scope of the claims. A person of ordinary skill in the art will recognize many variations, substitutions, and modifications. Figure 6Only the key process of using Mux / Demux wavelength as thickness compensation feedback to improve the uniformity of thin film optical properties at the wafer level is outlined. The mux / Demux wavelength can be affected by stress changes related to the process flow and thermal budget and the bending in the wafer. At least for wafer batches in a specific time window, if the film deposition process and the downstream integration process are well maintained, the wavelength changes caused by these factors are generally consistent. Wavelength feedback can certainly be applied to pre-compensate for changes in optical properties due to wafer bending and stress. In addition, stress, temperature and humidity factors during the downstream integration process can also change the optical properties and cause wavelength shifts. Again, the method proposed here can be applied to pre-compensate for changes in optical properties due to downstream integration processes.

[0060] Specifically, the method 600 includes a step 602 for providing a plurality of wafers. These wafers may be product wafers of the same wafer batch to be used in the actual manufacturing process of the thin film integrated optical circuit, for example, SOI wafers.

[0061] The method 600 also includes a step 604 for forming a first layer of dielectric overlying each of the plurality of wafers in a first process. Here, as an example, the first layer may be a SiNx film for making thin film WDM Mux / Demux circuits. More generally, the first layer is a non-stoichiometric dielectric material layer. As an example, the first process is a deposition process using one of the following methods: including chemical vapor deposition (CVD), plasma enhanced chemical vapor deposition (PECVD), low pressure chemical vapor deposition (LPCVD), and atomic layer deposition (ALD). Here, all wafers may experience various effects from factors such as layer composition, wafer bow, stress, temperature, or humidity in the first process. If the same set of process parameters of the first process is fixed, then these effects can be controlled to the same level for multiple wafers.

[0062] The method 600 also includes a step 606 for measuring a thickness distribution of a first layer across a first wafer selected from a plurality of wafers. The first wafer is selected as a reference wafer for both wafer-level thickness and optical characterization (which can be performed in situ). As an example, Figure 4A A normalized 2D thickness wafer map based on the measured thickness distribution is shown in FIG. Optical characterization (eg, characteristic wavelength measurement) will be performed after a series of downstream processes.

[0063] At step 608, method 600 also includes patterning the first layer into a first plurality of optical circuit dies on the first wafer in a second process. For example, each optical circuit die may be a demultiplexer (Demux). Here, the second process refers to a series of downstream integration processes used to define all the first plurality of optical circuit dies on the first wafer, including lithography, etching, cleaning, etc. Patterning without trimming the layer serves as a common prerequisite for wavelength feedback for all optical circuit dies. The second process will be fixed to the same settings for the first wafer (a test wafer that is patterned but the first layer is not trimmed) and the remaining product wafers (the first layer on the product wafer will be trimmed according to the to-be-determined wavelength feedback results from the sample wafer). Similarly, if the same set of process parameters is fixed for the second process, the first wafer is therefore used as a wafer that arrives in advance, and its characterization data is applicable to the remaining wafers in the plurality of wafers.

[0064] At step 610, method 600 further includes measuring a spectral response of the first layer characterized by a wavelength of each of the first plurality of optical circuit dies to obtain a wavelength distribution across the first wafer. This wavelength distribution corresponds to pre-trim values ​​of the first layer patterned without thickness trimming.

[0065] At step 612, method 600 further includes determining a correlation between wavelength changes and thickness differences across the first wafer based on the wavelength distribution obtained at step 610 and the thickness distribution obtained at step 606. The correlation between the wavelength changes of the first layer on the first wafer and the thickness differences of the first layer on the first wafer should be one-to-one with respect to their locations on the wafer because measurements of both distributions are made on each of the corresponding optical circuit dies. The correlation may also be compared to a historical correlation model between wavelength changes and thickness changes, the historical correlation model being based on previously established data collected for a process that produces a desired product having a particular target wavelength.

[0066] At step 614, method 600 also includes determining a target thickness distribution for the first layer based on the correlation and the target wavelength. Following the previous step 612, by comparing the correlation of the current process performed on the first wafer with the historical correlation model, a target thickness distribution or a map of the wafer to be trimmed can be determined based on the target wavelength using the measured wavelength distribution (for the first layer on the first wafer) as a feedback parameter. In a specific embodiment, the target thickness distribution can be converted into a lookup table for thickness trimming fed into a trimming tool, based on which each product wafer is trimmed to achieve a corresponding target thickness at different locations across the wafer. By design, the determined target thickness distribution will cause the characteristic wavelength of all optical circuit dies to tend towards the target wavelength for each of the wafers across the same batch. For example, the optical circuit die is a CWDM demultiplexer, the characteristic wavelength is the center wavelength of the channel output of the demultiplexer, and the target wavelength is a specified channel wavelength, for example, 1300nm. In another example, the optical circuit die is a DWDM demultiplexer, the characteristic wavelength is a center wavelength of a channel output of the demultiplexer, and the target wavelength is a designated DWDM channel wavelength.

[0067] At step 616, method 600 further includes modifying the first layer on each of the plurality of wafers to obtain a second layer having a target thickness distribution. Specifically, the first layer on each of the remaining wafers in the plurality of wafers (of the same batch) is modified or trimmed according to the wafer map to be trimmed determined in step 614. An in-situ ion beam trimming tool can accomplish this process in the same chamber where thin film deposition is performed. Ion beam trimming is a technique used to fine tune the properties of dielectric materials and devices (such as resistors or waveguides) after they are initially made. By directing a focused ion beam onto a device (e.g., the first layer), a small amount of material can be removed to adjust the device performance to a desired specification. The target thickness distribution represents the properties of the modified layer, which is designed to compensate for wavelength variations and produce a new wavelength distribution with reduced variations around the target wavelength across each wafer. In other words, the feedback effect based on wavelength measurement can produce an improved uniformity of optical properties.

[0068] At step 618, method 600 further includes patterning a second layer into a second plurality of optical circuit dies on each of the plurality of wafers in a second process. Again, patterning the second layer is performed in the same second process (i.e., a series of downstream integration processes including lithography, etching, cleaning, etc.) as used for the first wafer.

[0069] At step 620, method 600 finally includes singulating each of the plurality of wafers to obtain a second plurality of optical circuit dies. Thus, each of these optical dies is characterized by a characteristic wavelength that is well targeted at a target wavelength. For example, each die is a CWDM demultiplexer corresponding to a center wavelength that is well targeted (within + / - 2.5 nm) at a specified CWDM channel wavelength (e.g., 1300 nm). In another example, each die may be a DWDM demultiplexer having a center wavelength that is targeted within 1 nm of a specified DWDM channel wavelength.

[0070] In another embodiment, the method 600 presented above can be used to improve batch-to-batch optical property uniformity. For example, after applying the method 600, the corresponding center wavelength can be corrected to be within + / - 2.5 nm of the desired target wavelength. However, across many wafer batches, the corresponding center wavelength can drift away from the target wavelength. The batch-to-batch correlation between wavelength drift and thickness correction can be derived and used to generate a new lookup table to perform additional thickness trimming on wafers in upcoming batches. This will result in improved batch-to-batch optical property uniformity.

[0071] Fig. 7A is an exemplary graph of measured center wavelength variation of thin film-based demultiplexers in wafers for all wafers before modifying layer thicknesses in accordance with an embodiment of the present technology. As shown in this example, the center wavelength of layer-based demultiplexers is depicted without thickness trimming of each wafer from two wafer batches. Around the target center wavelength (denoted as C. Wavelength), the error can be as high as + / - 5 nm.

[0072] Figure 7B is an exemplary diagram of the measured center wavelength of a thin film based demultiplexer in a wafer after modifying the layer according to an embodiment of the present technology. As shown in this example, the demultiplexer was made by the same process from layers with trimmed thickness on wafers from two wafer batches, and the center wavelength error has been reduced from + / -5nm to only + / -1nm. This shows a clear advantage of wavelength feedback based non-uniform phase trimming, which produces a much tighter distribution for all wafers in the same production cycle.

[0073] The proposed non-uniform trimming technique for improving uniformity of optical properties is fully compatible with CMOS processes and in-line trimming tools and measurement equipment. For wafer-level processes, the proposed method can be implemented at high throughput and without significantly increasing the manufacturing cycle time. The method can generally be used to compensate for all other predictable phase errors caused by different reasons, including non-stoichiometric dielectric layers (e.g., SiNx films), coating layers, or stress introduced in downstream integration processes. This technique can also perform more complex phase engineering designs as required by many applications, for example, targeting different target wavelengths at different areas of the wafer. In some embodiments, both the feed-forward mechanism in method 500 (based on blanket wafer refractive index) and the feedback mechanism in method 600 (based on optical measurement correction) can be used together to apply a non-uniform trimming process to layers on product wafers. It is important to note that for all methods disclosed in this application, the feedback and feed-forward wafers are deposited at the same time as the product wafer to minimize the natural variations of the deposition tools and chambers.

[0074] Although the above is a complete description of specific embodiments, various modifications, alternative constructions and equivalents may be used. Therefore, the above description and illustrations should not be considered as limiting the scope of the present invention technology defined by the appended claims.

Claims

1. A method for processing a semiconductor device, the method comprising: providing a first wafer; providing a second wafer, the second wafer comprising a silicon material, wherein a size difference between the first wafer and the second wafer is less than 0.1%; forming a first layer of a dielectric on the first wafer and forming a second layer of the dielectric on the second wafer; measuring a refractive index profile of the second layer on the second wafer; measuring a first thickness profile of the first layer on the first wafer; calculating a second thickness distribution of the first layer based on the refractive index distribution and the first thickness distribution; as well as Material is removed from the first layer based on a thickness difference from the first thickness profile to the second thickness profile, thereby obtaining a third layer having the second thickness profile, the third layer being characterized by a spectral response having a characteristic wavelength variation of less than 5 nm across the first wafer. The method of claim 1 , wherein the first wafer comprises a silicon-on-insulator wafer.

3. The method of claim 1, wherein the first layer of dielectric and the second layer of dielectric comprise one of non-stoichiometric dielectrics selected from SiNx, SiONx, SiOx, and amorphous Si, and the first layer and the second layer are formed by performing a deposition process.

4. The method according to claim 3, wherein the deposition process is selected from one or more methods including chemical vapor deposition (CVD), plasma enhanced chemical vapor deposition (PECVD), low pressure chemical vapor deposition (LPCVD) and atomic layer deposition (ALD). The method of claim 1 , further comprising reducing thickness variation of the second layer before measuring the refractive index profile. The method of claim 1 , wherein measuring the refractive index profile comprises using in-line ellipsometry.

7. The method of claim 1, wherein measuring the first thickness profile comprises using ellipsometry, reflectometry, or spectroscopy.

8. The method of claim 1, wherein removing material comprises using a focused positive ion beam to scan across the first layer on the first wafer with a vertical resolution of less than 1 nm and a lateral resolution of less than 1 um.

9. The method of claim 1, further comprising dicing the first wafer into first dies, the first dies comprising optical circuits.

10. The method of claim 9, wherein the optical circuit comprises a thin film optical filter.

11. The method of claim 1, wherein calculating the second thickness distribution comprises calculating a first regional thickness distribution corresponding to a first target wavelength on a first region of the first wafer and a second regional thickness distribution corresponding to a second target wavelength on a second region of the first wafer.

12. The method according to claim 11, further comprising: Material is removed from the first layer on the first region based on the first region thickness distribution and material is removed from the first layer on the second region based on the second region thickness distribution, thereby obtaining a fourth layer on the first region and a fifth layer on the second region, the fourth layer being characterized by a first spectral response having a characteristic wavelength within + / - 2.5 nm around the first target wavelength, and the fifth layer being characterized by a second spectral response having a characteristic wavelength within + / - 2.5 nm around the second target wavelength.

13. The method according to claim 12, further comprising: The first wafer is singulated from the first region into a first plurality of dies and from the second region into a second plurality of dies, each of the first plurality of dies comprising an optical circuit based on a first modified layer having the characteristic wavelength within + / - 2.5 nm around the first target wavelength, and each of the second plurality of dies comprising an optical circuit based on a second modified layer having the characteristic wavelength within + / - 2.5 nm around the second target wavelength.

14. A circuit comprising: a first layer of a dielectric material on a portion of a first wafer, the first layer being characterized by a spectral response uniformity across the first wafer having a characteristic wavelength within + / - 2.5 nm of a target wavelength; The first layer is made by the following steps: depositing the dielectric material on the first wafer in a first process; measuring a first thickness profile across the first wafer; determining a second thickness profile across the first wafer based on the refractive index profile and the first thickness profile, the second thickness profile corresponding to the target wavelength; selectively removing the dielectric material based on the second thickness profile to form the first layer; and The first layer is patterned to define a plurality of dies, each die comprising an optical circuit based on the first layer.

15. The circuit of claim 14, wherein the first layer comprises one of a non-stoichiometric dielectric selected from SiNx, SiONx, SiOx, and amorphous Si, and the first wafer comprises a SOI wafer.

16. The circuit of claim 14, wherein the refractive index profile is determined by: depositing the dielectric material on a second wafer in a second process arranged with the same settings as the first process; reducing a thickness variation of a second layer on the second wafer; and measuring the refractive index profile of the second layer.

17. The circuit of claim 14, wherein the optical circuit comprises an optical multiplexer or demultiplexer for a coarse wavelength division multiplexing (CWDM) optical link or for a dense wavelength division multiplexing (DWDM) optical link.

18. A method for processing a semiconductor device, the method comprising: providing a plurality of wafers; forming a first layer of a dielectric on each of the plurality of wafers in a first process; measuring a thickness profile of the first layer across a first wafer selected from the plurality of wafers; patterning the first layer into a first plurality of optical circuit dies on the first wafer in a second process; measuring a spectral response of the first layer to obtain a wavelength distribution across the first wafer, the spectral response of the first layer characterized by a wavelength of each of the first plurality of optical circuit dies; determining a correlation between wavelength variation and thickness difference across the first wafer based on the wavelength distribution and the thickness distribution; determining a target thickness distribution of the first layer based on the correlation and the target wavelength; modifying the first layer on each of the plurality of wafers to obtain a second layer having the target thickness distribution; patterning the second layer into a second plurality of optical circuit dies on each of the plurality of wafers in the second process; and Each of the plurality of wafers is singulated to obtain the second plurality of optical circuit dies, each optical circuit being based on the second layer characterized by a spectral response having a center wavelength within + / - 2.5 nm of the target wavelength.

19. The method of claim 18, wherein each of the plurality of wafers comprises a silicon-on-insulator (SOI) wafer.

20. The method of claim 18, wherein the first layer of dielectric comprises one of a non-stoichiometric dielectric comprising SiNx, SiONx, SiOx, amorphous Si.

Images