Photoetching alignment method for structured light illumination

Abstract

The invention belongs to the technical field of semiconductor processing, and discloses a structured light illumination photoetching alignment method, which comprises the following steps of: overlapping a silicon wafer and a mask etched with an alignment mark on the surface, and performing collimation illumination on the alignment mark by using structured light to form secondarily modulated moire fringes; performing image acquisition on the secondarily modulated moire fringes by using an image acquisition unit; inputting the acquired image into a phase analysis model, and calculating alignment offset information of each group of moire fringes; the alignment offset information is processed through a demodulation algorithm model, and the final alignment offset is obtained through calculation; and finishing alignment of the mask and the silicon wafer by using the alignment offset. According to the invention, the precision and sensitivity of the photoetching Moire alignment technology are improved by applying structured light illumination, and sub-nano photoetching alignment is realized.

Description

Technical Field

[0001] The invention belongs to the technical field of semiconductor processing, and in particular relates to a photolithography alignment method for structured light illumination. Background Art

[0002] The moiré fringe lithography alignment technology realizes alignment by real-time monitoring of the grating interference fringe image. However, in the process of realizing the present invention, the inventors found that there are at least the following problems in the prior art: the acquisition of the lithography alignment signal is limited by the optical diffraction limit of the imaging system, resulting in the inability to acquire high-frequency signals with high precision; the grating alignment signal formed by the small-period lithography alignment mark has insufficient amplification, resulting in unclear acquired signals, leading to large errors. Summary of the invention

[0003] The present invention aims to solve the above technical problems at least to a certain extent. To this end, the present invention aims to provide a lithography alignment method using structured light illumination.

[0004] The technical solution adopted by the present invention is:

[0005] A structured light illumination lithography alignment method, comprising the steps of:

[0006] S1, placing a silicon wafer and a mask with alignment marks etched on the surface overlapping each other, and using structured light to collimate the alignment marks to form secondary modulated moiré fringes;

[0007] S2, using an image acquisition unit to acquire an image of the secondary modulated moiré fringes;

[0008] S3, input the collected image into the phase analysis model to calculate the alignment offset information of each group of moiré fringes;

[0009] S4, processing the alignment offset information through a demodulation algorithm model to calculate and obtain a final alignment offset;

[0010] S5. Use the alignment offset to complete the alignment of the mask and the silicon wafer.

[0011] Preferably, the structured light is formed by a light beam emitted by a laser light source passing through a collimating lens and a structured light grating. Alternatively, the structured light is formed by a light beam emitted by an LED light source passing through a collimating lens, a structured light grating and a projection lens. Alternatively, the structured light is formed by superposition and interference of two coherent light beams.

[0012] Preferably, the phase resolution model is a phase resolution model based on traditional image processing. Step S3 comprises: transforming the image from the spatial domain to the frequency domain using Fourier transform, filtering out the signal at the frequency spectrum point corresponding to the moiré fringes using a windowing method, transforming it to the spatial domain using an inverse Fourier transform, and extracting the corresponding value using a phase extraction function.

[0013] Preferably, the phase analysis model is a phase analysis model based on a convolutional neural network. The step S3 comprises: inputting the image into a trained phase regression network, and implementing regression prediction of the phase using a forward operation of a multi-layer convolutional network structure.

[0014] Preferably, the step S4 comprises:

[0015] S41, the first demodulation: using a center frequency of f c2 mixer, multiply by cos(2πf c2 t), and remove the high-frequency components through a low-pass filter to obtain the signal

[0016] S42, second demodulation: using a center frequency of f c1 mixer, multiply by cos(2πf c1 t), and remove the high-frequency components through a low-pass filter to obtain the signal

[0017] S43, signal recovery: amplify s2(t) to the original signal strength and restore the original signal s(t).

[0018] The present invention uses structured light illumination to improve the accuracy and sensitivity of photolithography moiré alignment technology and achieve sub-nanometer photolithography alignment. The beneficial effects of the present invention are:

[0019] Structured light illumination can be applied to all lithography alignment systems and can improve the accuracy by an order of magnitude without changing the original measurement optical path.

[0020] This method uses global signals for alignment, has good robustness, and has no stringent requirements on beam quality and measurement environment.

[0021] It can be used in conjunction with the coarse alignment method, with fast alignment speed and high alignment efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 It is a schematic diagram of the optical path system of the present invention.

[0023] Figure 2 It is a schematic diagram of a method for generating structured light.

[0024] Figure 3It is a schematic diagram of another method of generating structured light.

[0025] Figure 4 This is a schematic diagram of another method for generating structured light. DETAILED DESCRIPTION

[0026] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] It should be understood that it should also be noted that in the embodiment, the functions / actions that appear may be different from the order in which the drawings appear. For example, depending on the functions / actions involved, they can actually be executed substantially concurrently, or two figures shown in succession can sometimes be executed in the reverse order.

[0028] like Figure 1 As shown, a structured light illumination lithography alignment method of this embodiment includes the steps of:

[0029] S1. Overlap a silicon wafer and a mask with alignment marks etched on the surface, and use linear or circular structured light to collimate the alignment marks to form secondary modulated moiré fringes.

[0030] S2. Use an image acquisition unit to acquire an image of the secondary modulated moiré fringes.

[0031] S3, the collected image is input into the phase analysis model, and the alignment offset information of each group of moiré fringes is calculated. In this embodiment, the phase analysis model in step S3 is a phase analysis model based on traditional image processing. Accordingly, step S3 includes: using Fourier transform to transform the image from the spatial domain to the frequency domain, and then using the windowing method to filter out the signal at the frequency spectrum point corresponding to the moiré fringes, and then using the inverse Fourier transform to transform it to the spatial domain, and finally using the phase extraction function to extract the corresponding value.

[0032] In another embodiment, the phase resolution model is a phase resolution model based on a convolutional neural network. Accordingly, step S3 includes: inputting the image into a trained phase regression network, and implementing regression prediction of the phase using a forward operation of a multi-layer convolutional network structure.

[0033] S4. Process the alignment offset information through a demodulation algorithm model to calculate and obtain a final alignment offset.

[0034] The demodulation algorithm model after the spectrum is modulated twice is mainly used to demodulate the signal after two modulations and restore the original signal. The following is a basic demodulation model to illustrate the demodulation process after the signal is modulated twice. Assume that the signal being processed is a linear modulation signal:

[0035] in,

[0036] Modulation signal: original signal s(t);

[0037] The first modulation carrier: f c 1 is the carrier frequency, and the modulated signal is s m 1(t);

[0038] Second modulation carrier: f c 2 is the second modulation carrier frequency, and the final modulated signal is s m 2(t);

[0039] The signal after two modulations can be expressed as:

[0040] s m 2(t)=s m 1(t)·cos(2πf c2 t);

[0041] in:

[0042] s m 1(t)=s(t)·cos(2πf c1 t);

[0043] The signal after secondary modulation can be written as:

[0044] s m 2(t)=s(t)·cos(2πf c1 t)·cos(2πf c2 t);

[0045] Using trigonometric identities:

[0046]

[0047] Expands to:

[0048]

[0049] Therefore, the modulated signal contains two frequency components: one is (f c1 -f c2 ), the other is (f c1 +f c2 ).

[0050] The demodulation step, i.e. step S4, comprises:

[0051] S41, first demodulation (for removing the second carrier): using a center frequency of f c2 mixer, multiply by cos(2πf c2 t), and remove the high frequency components through a low-pass filter.

[0052] s′(t)=s m 2(t)·cos(2πf c2 t);

[0053] Expanding via trigonometric identities:

[0054]

[0055] By trigonometric identities:

[0056]

[0057] After low-pass filtering, cos(2πf c1 t) Ingredients:

[0058]

[0059] S42, second demodulation (for removing the first carrier): using a center frequency of f c1 mixer, multiply by cos(2πf c1 t), and remove the high frequency components through a low-pass filter.

[0060] s2(t)=s1(t)·cos(2πf c1 t);

[0061] Expands to:

[0062]

[0063] After low-pass filtering, the remaining:

[0064]

[0065] S43, signal recovery: finally amplify s2(t) to the original signal strength and restore the original signal s(t).

[0066] Through the above process, the signal after secondary modulation can be restored to the original signal through two demodulations. The demodulation algorithm model can be further applied to complex modulation systems. The key steps include: 1. Using two-stage carrier demodulation. 2. Low-pass filtering to remove high-frequency components. 3. Capturing accurate signal characteristics through secondary spectrum information.

[0067] S5. Input the alignment offset into the control terminal, and the control terminal completes the alignment of the mask and the silicon wafer through the motion mechanism.

[0068] like Figure 2 As shown, structured light is formed by a light beam emitted by a laser light source passing through a collimating lens and a structured light grating.

[0069] like Figure 3 As shown, the light beam emitted by the LED light source is formed after passing through the collimating lens, the structured light grating and the projection lens.

[0070] like Figure 4 As shown, it is formed by the superposition and interference of two coherent light beams.

[0071] The structured light generated by the above three methods is applicable to the photolithography alignment method of this embodiment, which has a wide range of application prospects and flexible implementation methods.

[0072] The present invention is not limited to the above-mentioned optional implementation modes. Anyone can derive other various forms of products under the inspiration of the present invention. However, no matter what changes are made in the shape or structure, all technical solutions that fall within the scope defined by the claims of the present invention fall within the protection scope of the present invention.

Claims

1. A structured light illumination lithography alignment method, characterized in that: Includes steps: S1, placing a silicon wafer and a mask with alignment marks etched on the surface overlapping each other, and using structured light to collimate the alignment marks to form secondary modulated moiré fringes; S2, using an image acquisition unit to acquire an image of the secondary modulated moiré fringes; S3, input the collected image into the phase analysis model to calculate the alignment offset information of each group of moiré fringes; S4, processing the alignment offset information through a demodulation algorithm model to calculate and obtain a final alignment offset; S5. Use the alignment offset to complete the alignment of the mask and the silicon wafer.

2. The photolithography alignment method according to claim 1, characterized in that: The structured light is formed after a light beam emitted by a laser light source passes through a collimating lens and a structured light grating.

3. The photolithography alignment method according to claim 1, characterized in that: The structured light is formed after the light beam emitted by the LED light source passes through a collimating lens, a structured light grating and a projection lens.

4. The photolithography alignment method according to claim 1, characterized in that: The structured light is formed by superposition and interference of two coherent light beams.

5. The photolithography alignment method according to claim 1, characterized in that: The phase resolution model is a phase resolution model based on traditional image processing.

6. The photolithography alignment method according to claim 5, characterized in that: The step S3 includes: transforming the image from the spatial domain to the frequency domain by Fourier transform, filtering the signal by windowing at the frequency spectrum point corresponding to the moiré fringes, transforming it to the spatial domain by inverse Fourier transform, and extracting the corresponding value by phase extraction function.

7. The photolithography alignment method according to claim 1, characterized in that: The phase resolution model is a phase resolution model based on a convolutional neural network.

8. The photolithography alignment method according to claim 7, characterized in that: The step S3 includes: inputting the image into a trained phase regression network, and implementing regression prediction of the phase using a forward operation of a multi-layer convolutional network structure.

9. The photolithography alignment method according to claim 1, characterized in that: The step S4 comprises: S41, the first demodulation: using a center frequency of f c2 mixer, multiply by cos(2πf c2 t), and remove the high-frequency components through a low-pass filter to obtain the signal S42, second demodulation: using a center frequency of f c1 mixer, multiply by cos(2πf c1 t), and remove the high-frequency components through a low-pass filter to obtain the signal S43, signal recovery: amplify s2(t) to the original signal strength and restore the original signal s(t).

10. The photolithography alignment method according to claim 1, characterized in that: The structured light is linear or circular.

Images