Semiconductor equipment industry popular science
Source: Content comes from UISEE Capital, thank you.
1. Industry trends: Semiconductor cycle turning point, domestic substitution emerges
1.1 Semiconductor industry: The big cycle lasts about ten years, and the core driver of demand comes from technological development
The global semiconductor industry market size will reach US$580.1 billion in 2022, reaching a record high, with a compound growth rate of 7.4% over the past ten years. By analyzing the year-on-year growth rate of global semiconductor sales in the past 20 years, it was found that the semiconductor industry has a large cycle of about 10 years, that is, an "M" shaped fluctuation every 10 years. The main reason is that on the one hand, it is affected by changes in global GDP growth rate; on the other hand, Mainly due to industry development driven by technology. In the first half of 2023, global semiconductors were in a downward cycle, but the new round of technological innovation brought by AIGC has triggered a significant increase in demand, and the industry is expected to usher in an upward cycle in 2024.
Figure: Global semiconductors show volatile growth, with a ten-year cycle
Source: Encyclopedia of Integrated Circuit Industry, WSTS, iFinD, Zheshang Securities Research Institute
Big cycles look at technology, small cycles look at demand. The small cycle is mainly affected by the fluctuation of the downstream demand cycle. From the perspective of global semiconductor sales year-on-year, the industry's small cycle is about three years. The previous cycle high was in August 2021. Global semiconductor sales in January 2023 were US$41.3 billion, a year-on-year decrease of 19%. Judging from the industrial cycle, the second half of 2023 is expected to usher in the turning point of the downward cycle. In 2024, on the one hand, traditional chips will enter an inventory inflection point; on the other hand, AIGC's sharp increase in demand for computing power will drive an explosion in demand for emerging chips and accelerate the arrival of the upward cycle.
Figure: The global semiconductor small cycle is about 3 years
Source: Semiconductor Industry Association, Zheshang Securities Research Institute
From 2012 to 2022, the compound annual growth rate of the global and Chinese semiconductor equipment markets reached 11% and 27% respectively, and the Chinese market grew faster than the global market. Mainland China is the world's largest semiconductor equipment market. The global semiconductor equipment market will be US$107.6 billion in 2022. Mainland China's semiconductor equipment sales account for 26% of global sales, reaching US$28.3 billion, surpassing Taiwan (25%), South Korea (20%), and North America (10%), continuously. It has become the world's largest semiconductor equipment market in three years.
Chart: China’s share of the global equipment market will increase from 6% in 2006 to 26% in 2022
Figure: Mainland China will become the world’s largest market for semiconductor equipment for three consecutive years in 2022
Source: SEMI, Japan Semiconductor Manufacturing Equipment Association, Zheshang Securities Research Institute
1.2 Semiconductor equipment: Chip manufacturing processes continue to shrink, driving equipment capital expenditures to increase
The shrinking of chip manufacturing processes has led to an increase in equipment capital expenditures and an increase in the scale of the semiconductor equipment market. Historically, chip development has followed Moore's Law . The core content of Moore's Law is that the number of transistors that can be accommodated on an integrated circuit will double approximately every 18 to 24 months. The core of Moore's Law is an economic law. As chip manufacturing processes continue to shrink, Moore's Law gradually expires. In 2018, the gap between actual chip performance and the requirements of Moore's Law expanded 15 times. With the failure of Moore's Law, the shrinkage of chip manufacturing processes has led to a rapid increase in construction costs, driving up equipment capital expenditures. The construction cost of a wafer factory for every 10,000 5nm chips is as high as US$5.4 billion, which is six times that of 28nm.
Chart: Construction cost of wafer fab per 10,000 wafers (USD 100 million)
Figure: Process development roadmap of leading logic chip manufacturers
Source: IBS, McKinsey, IC inghts, Zheshang Securities Research Institute
Global semiconductor capital expenditure: IC insights predicts that global semiconductor capital expenditure will be US$181.7 billion in 2022, a year-on-year increase of 19%. Amid the weakening memory market and U.S. sanctions against China, global semiconductor equipment capital expenditure is expected to be US$146.6 billion in 2023, a year-on-year decrease of 19%.
Cyclic analysis: Judging from the year-on-year growth rate of global semiconductor capital expenditures from 2000 to the present, global semiconductor capital expenditures cycle approximately every three years. 2023 will be at the bottom of the industry cycle, and capital expenditures are expected to reverse in 2024.
Chart: Global semiconductor industry capital expenditures
Figure: Global semiconductor capital expenditure cycles fluctuate, with a cycle of about three years.
Source: IC insights, McKinsey, Zheshang Securities Research Institute
1.3 Semiconductor equipment: The blockade of semiconductor equipment in the United States, Japan and the Netherlands has forced the localization rate to increase rapidly
Since 2018, the United States has continued to tighten controls on China's semiconductors, extending from downstream to upstream such as Huawei, ZTE, and SMIC. On October 7, 2022, the U.S. BIS implemented semiconductor controls on China, extending the scope to advanced chips, equipment, parts, personnel, etc. The scope of U.S. semiconductor equipment control: manufacturing equipment required for advanced logic process chips below 16/14nm, NAND flash memory chips with more than 128 layers, and DRAM memory chips with 18nm half-pitch or lower.
Figure: The scope of US semiconductor sanctions against China continues to expand, and advanced equipment is included in the prohibited export scope
Source: Xinmo Research, Zheshang Securities Research Institute
Anti-globalization in the semiconductor industry has become a trend. Since 2022, various countries have actively formulated support policies to support the development of the local semiconductor industry. On March 10, 2023, the First Session of the 14th National People's Congress decided to reorganize the Ministry of Science and Technology and establish the Central Science and Technology Commission. This move will help coordinate the efforts of all parties in scientific and technological innovation, promote and improve the new national system, optimize the management of the entire chain of scientific and technological innovation, and promote Transform scientific and technological achievements and promote the integration of science and technology with economic and social development. On April 6, 2023, the National Technical Committee for Standardization of Integrated Circuits was established, which plays an important role in promoting the high-quality development of the integrated circuit industry.
Figure: Semiconductor industry policies of various countries around the world in 2022
Source: Jiwei.com, Zheshang Securities Research Institute
2. Process control: Measurement and testing are important guarantees for semiconductor manufacturing yield.
Process control: After different processes in the semiconductor wafer manufacturing process, dimensional measurement, defect detection, etc. are often required for process control and yield management, which require speed and accuracy. Dimensional measurement, defect detection, etc. are applied after each manufacturing process. IC measurement equipment is used for process control and yield management, and the detection requirements are fast, accurate, and non-destructive. In the development process, IC measurement is faced with various technical difficulties in terms of size reduction, complex 3D, and new materials, and it is constantly upgrading in the face of various needs such as storage, CIS, compound semiconductor and other different semiconductor testing. The technical categories of IC measurement equipment include probe microscopes, scanning/transmission electron microscopes, optical microscopes, ellipsometers/scattering meters, etc. The technical development direction includes continuing the existing non-destructive measurement technology, promoting parallel electron beam technology in electron microscopy, and scattering meters. Extend to EUV and X-rays to reduce the wavelength, and combine multiple measurement methods and machine learning to achieve hybrid measurements.
Process control equipment includes measurement equipment (Metrology) and defect (including particles) inspection equipment (Inspection) used in the process. During the chip production process, online process inspection equipment must conduct non-destructive quantitative measurements and inspections of wafers after different processes to ensure the key physical parameters of the process (such as film thickness, line width, trench/hole depth, side wall angle, etc. ) meet the requirements, while discovering possible defects and classifying them, rejecting unqualified wafers to avoid waste in subsequent processes. Another role of process testing equipment is to assist in optimizing equipment operating parameters and photomask design during process development and trial production, optimizing the entire process flow, shortening development time, improving yield and achieving mass production.
Semiconductor measurement Metrology mainly includes:
1) Deviation measurement of overlay alignment;
2) Thickness measurement of film materials;
3) Critical dimension (CD) measurement of the wafer after photoresist exposure and development, etching and CMP process;
4) Others: such as wafer thickness, bending warp (Bow/Warp), 1D/2D stress, wafer topography, four-point probe resistance measurement RS, XPS measurement of implant content, etc., AFM (atomic force microscope)/Metal Plus (ultrasonic) measures step height (Step Height), etc.
Semiconductor inspection inspection mainly includes:
1) Non-graphic defect detection, including particles , residues, scratches, vigilance for primary pits (COP), etc.;
2) There is image defect detection, including break, line edge defect (bite), bridge, linear change (Deformation), etc.;
3) Mask defect detection, including particles, etc.;
4) Defect re-inspection, use an optical microscope or scanning electron microscope to confirm the existence of the defects detected (location, size, type).
Figure: Semiconductor measurement and inspection classification
Source: Zhongke Feichai Prospectus, Founder Securities Research Institute
Zhongdao Inspection is oriented to advanced packaging processes, mainly for flip-chip (Flip-Chip), wafer-level packaging (Wafer-LevelPackage) and through-silicon via (ThroughSilicon Via, TSV) and other advanced process requirements for bumps and through-holes. Conduct non-contact quantitative inspection and measurement of defects/foreign matter residues in holes (TSV), copper pillars (Copper-Pillar), etc. and their shape, spacing, and height consistency, as well as the redisriburion layer (RDL). Back-end testing mainly uses electricity to test the functions and electrical parameters of the chip, which mainly includes two links: wafer testing and finished product testing.
Figure: Process Control (Detection, Measurement) and ATE (Testing) Market Space in 2021
Source: Gartner, Huajing Industrial Research Institute, Founder Securities Research Institute
Figure: Detecting defects & measuring dimensions
Source: KLA, Founder Securities Research Institute
Depending on the different materials and structures used in the manufacturing process, process inspection equipment uses a variety of different technologies including broadband spectroscopy (ultraviolet to infrared), electron beams, lasers, and X-rays. In terms of performance indicators, as the process continues to develop towards finer line widths, the device morphology and structure are also changing from a two-dimensional planar structure to a three-dimensional structure. Therefore, there are higher requirements for the sensitivity, applicability, stability and throughput of the detection equipment. .
Figure: Comparison of characteristics of optical inspection technology, electron beam inspection technology and X-ray measurement technology
Source: Zhongke Feichai Prospectus, Founder Securities Research Institute
Equipment using optical detection technology can relatively achieve high-precision and high-speed balancing, and can meet functions that cannot be achieved by other technologies, such as three-dimensional topography measurement, photolithography overlay measurement, and multi-layer film thickness measurement. application, so adoption accounts for the majority. According to VLSI Research and QY Research, in the global semiconductor inspection and measurement equipment market in 2020, the equipment market shares of optical inspection technology, electron beam inspection technology and X-ray measurement technology accounted for 75.2%, 18.7% and 2.2% respectively.
Chart: Process control classification and market size in 2021 (USD billion)
Source: Gartner, Founder Securities Research Institute
2.1 Measurement
Overlay accuracy measurement equipment: used to measure the overlay error between layers, that is, the plane distance between the centers of the two-layer graphic structures. There are three main measurement systems, optical microscopy imaging (IBO) system, optical diffraction (DBO) ) system and scanning electron microscope (SEM-OL) system. Optical microscopy imaging systems are the most commonly used, calculating overlay errors through imaging; optical diffraction systems use non-imaging methods, measuring the intensity of the diffracted beam through a light intensity sensor to determine overlay errors. They use fewer optical elements and are often used in advanced In the photolithography process control; the scanning electron microscope system is mainly used for the final overlay error measurement after etching, and the measurement speed is slow.
The object of overlay accuracy measurement is the overlay target graphics. These graphics are usually made in the scribing groove area. The target graphics used in the imaging overlay measurement system usually include (a) block in block, (b) bar in bar and (c) ) target (AIM) graphics.
Figure: Overlay error measurement
Figure: Commonly used overlay error measurement target graphics
Source: Hitachi High-Tech official website, Founder Securities Research Institute, "Integrated Circuit Industry Complete Book", Qihong Semiconductor
Common optical overlay equipment are KLA’s Archer series and ASML’s YieldStar series. The Archer series uses IBO and DBO measurement technology and can measure a variety of overlay target patterns; YieldStar uses DBO measurement technology; Hitachi’s CD-SEM CV series uses high pressure. Accelerated scanning electron microscopy (SEM-OL).
Film thickness measurement: After multiple thin film depositions of various materials on the wafer, it is necessary to accurately judge the film thickness and other parameter properties to ensure that the product meets the performance design requirements. There are many ways to measure film thickness. The traditional method is to measure directly through a step meter. However, this method causes great damage to the film itself. At the same time, the measurement results are greatly affected by the accuracy of the instrument, so the cost of accurate measurement is high. Currently, the most commonly used methods include non-optical methods and optical methods. Non-optical methods can only be used to measure film thickness, including the four-probe method, eddy current method, capacitance method, electromagnetic method, etc. Among them, the four-probe method The method and eddy current method are the most widely used methods; the methods that use optical principles to measure film thickness mainly include prism coupling guided mode method, light section method, multi-beam interference method, spectrophotometry, ellipsometry, etc. Among them, ellipsometry is the most widely used and can simultaneously measure the optical parameters of films.
Figure: Thin film measurement method
Source: "Film Thickness Measurement Based on Elliptical Polarization Method", Founder Securities Research Institute
The four-probe method (4PP) refers to using four equidistant metal probes to contact the surface of the film material. The two outer probes are connected to a DC current, and the two middle probes are connected to a potentiometer to measure the voltage drop. Based on the measured voltage and The current flow gives the resistance at the specific location, and the film thickness is determined by dividing the resistivity of the film material by the resulting resistance, usually calculated by software. The eddy current method (EC) means that the time-varying current passing through the coil produces time-varying eddy currents in the conductive layer. These time-varying eddy currents in turn generate a magnetic field that changes the impedance of the drive coil, which is inversely proportional to the resistance of the film material. The film thickness parameter can also be obtained by conversion.
Figure: Four-probe method
Figure: Eddy current method
Source: KLA, Founder Securities Research Institute
The basic principle of ellipsometry is as follows: a beam of elliptically polarized light is irradiated onto the sample at a certain incident angle. The polarization state of the incident light is known, and the specific structure of the film material can be determined based on the change in polarization state, and its thickness and optical properties can be obtained. Parameter information (such as complex refractive index, thickness or complex permittivity, etc.).
Figure: Principle of ellipsometry
Source: Wikimedia, Founder Securities Research Institute
At present, the main non-optical film thickness measurement products include Kelei's Filmetrics R50 series, and the main optical film thickness measurement products include Kelei's Aleris series and SpectraFilm series. Domestic optical film thickness measurement products include the EFILM series of Shanghai Precision Measurement.
Chart 14: Filmetrics R50, Aleris, and SpectraFilm, in order
Source: KLA, Founder Securities Research Institute
Critical dimension measurement equipment: Mainly used for online measurement of critical dimensions (CD), height, and side wall angle in the chip production process and performance monitoring of key equipment (photolithography machines, glue coating and development equipment, etc.). Among them, critical dimension measurement is mainly used for post-development inspection (ADI: After Development Inspection) and post-etch inspection (AEI: After Etch Inspection). At present, the non-imaging measurement technology mainly relies on the optical scatterometer. The optical scatterometer is also called the optical critical dimension measuring instrument (OCD). Its principle is to reconstruct the three-dimensional shape of the sample to be measured by measuring the scattering information of the sample and solving the inverse scattering problem. Appearance.
Figure: Basic process of optical scattering measurement
Source: "Online Optical Measurement and Inspection Technology for Integrated Circuit Manufacturing: Current Situation, Challenges and Development Trends", Founder Securities Research Institute
The basic process of OCD measurement mainly includes two important steps: forward problem and inverse problem. The forward problem is to obtain the scattering information of the sample structure to be measured through a certain scattering measurement device. What needs to be solved is the instrument measurement problem. The inverse problem is to obtain from the measurement. To extract the three-dimensional morphological parameters of the sample to be measured from the scattering data, what needs to be solved is to construct a forward scattering model of the interaction between light and the structure of the sample to be measured and select an appropriate solution algorithm.
Measuring instruments: The currently used optical scattering devices can be divided into angle-resolved scatterometers and spectral scatterometers. The advantage of angle-resolved scatterometer is that due to the use of a single wavelength, there is no need to make assumptions about the dielectric function of the sample material during data analysis; in addition, it can be relatively easily extended to shorter wavelength ranges, such as EUV and X-rays. The disadvantage is that it contains moving components, which limits the speed of measurement. The advantage of a spectral scatterometer is that the measurement speed is very fast. Currently, a commonly used scatterometer based on the spectral ellipsometer (SE) can achieve very high vertical resolution. Compared with the angle-resolved scatterometer, it can obtain more measurement information. The disadvantage is that for accurate measurements, especially in ellipsometric scattering measurements, precise calibration is required and the optical constants of the sample material over a wide spectral range need to be determined in advance. In actual use, in order to improve measurement sensitivity, two scattering measurement methods are usually combined.
Figure: Schematic diagram of different optical scattering measurement devices Note: (a) (b) angle-resolved scatterometer; (c) (d) spectral scatterometer
Source: "Online Optical Measurement and Inspection Technology for Integrated Circuit Manufacturing: Current Situation, Challenges and Development Trends", Founder Securities Research Institute
Construct a forward scattering model and select a solution algorithm: In optical scattering measurement, extracting the three-dimensional morphology parameters of the sample to be measured from the scattering measurement data is essentially a process of solving the inverse scattering problem. Currently, there are two main solution methods, namely library matching method and nonlinear regression method. In the parameter extraction process of library matching, the scattering simulation database is established in advance, and then the measured data is compared with the simulation data to obtain the parameter values to be measured corresponding to the simulation data that best matches the test data. The nonlinear regression method continuously adjusts the input parameters to reduce the difference between the measured data and the simulation data calculated by the forward scattering model within the allowable range. The advantage of the library matching method is that it can quickly extract the parameters to be measured, but it requires the establishment and storage of a huge simulation database in advance, and the accuracy is limited by the grid spacing of the database. The advantage of the nonlinear regression method is that it does not require the establishment of a simulation database and can obtain relatively accurate results. However, since each iteration must call the forward scattering model to calculate the simulation data, it is very time-consuming.
In addition to OCD, atomic force microscopy (AFM) and scanning electron microscopy (SEM) are also the two most widely used types of micro-nanowire spacing size measurement instruments. The lateral resolution of the images obtained by the two is similar. AFM obtains the morphological structure image of the surface of the product being tested, which is a true three-dimensional image, and can measure the three-dimensional information of the sample. SEM can only provide two-dimensional images, but its images have a large depth of field and a large field of view. Critical dimension scanning electron microscope (CD-SEM) is a scanning electron microscope with the function of automatically positioning and measuring lines, and is widely used to monitor the line width of semiconductor production lines.
Electronic Industry Special Report 15 Please pay attention to the special statement and disclaimer at the end of the article. Scanning electron microscopes use the differences in surface characteristics of materials to produce different brightness differences through different areas of the sample under the action of electron beams, thereby obtaining a certain contrast. image. The imaging signal is secondary electrons, backscattered electrons or absorbed electrons, among which secondary electrons are the most important imaging signal. High-energy electron beams bombard the surface of the sample to stimulate various physical signals on the surface of the sample, and then use different signal detectors to receive the physical signals and convert them into image information.
Figure: Information generated by the interaction of electron beams with matter
Figure: Schematic diagram of scanning electron microscope
Source: "
Application Examples of SEM in Semiconductor Process Research", Founder Securities Research Institute Source: Shuosibai Testing, Founder Securities Research Institute
Algorithm: CD-SEM After obtaining the image of the measurement pattern. CD-SEM performs measurements and uploads data. The CD-SEM measurement algorithm needs to be continuously optimized and improved so that the measurement results truly and accurately reflect the performance of the sample. For example, new measurement methods are introduced, including Edge Roughness Gap, Wiggling, Overlay, Center Gravity, etc. In addition, measurement reliability needs to be continuously improved, and Sensitivity to product process fluctuations.
Figure: SEM image
Figure: Create a Line Profile from the SEM image to calculate the measured value
Source: Hitachi High-Tech official website, Founder Securities Research Institute
2.2 Detection
Graphic wafer defect detection: Optical graphics wafer defect detection equipment uses high-precision optical detection technology to detect and identify defects and contamination on the wafer, and provide the wafer factory with product quality issues of wafers in different production nodes. And confirm whether the operation of process equipment is normal, thereby improving yield and saving costs. Optical detection technology uses wide-spectrum illumination from deep ultraviolet to visible light bands or deep ultraviolet single-wavelength high-power laser illumination, and uses optical brightfield or darkfield imaging methods with high resolution and large imaging field to obtain information about the circuit on the wafer surface. Pattern image, real-time alignment, noise reduction and analysis of circuit patterns, as well as defect identification and classification, to achieve the capture of wafer surface pattern defects.
Optical pattern wafer defect detection equipment is divided into two categories: bright field and dark field. The main difference between the two is: bright field equipment collects vertically reflected light signals on the wafer surface to analyze defects, while dark field equipment collects wafer surface defects. The light signal scattered back from the surface is analyzed. If the wafer surface is flat and has no defects, the light reflected by the bright field device is relatively complete incident light, while the incident light of the dark field device is totally reflected, and it receives scattered light signals.
As the equipment industry continues to develop, the definitions of brightfield and darkfield are also changing. Nowadays, brightfield generally refers to the illumination light path and collection light path that share the same microscope objective in adjacent wafer sections, while darkfield refers to the illumination light path and collection light path. Completely separated in physical space. Because of the difference in vertical reflection and scattered light signals, the detection sensitivity of brightfield equipment is higher than that of darkfield equipment, but the scanning speed of brightfield equipment is also slower.
Figure 23: Brightfield and darkfield optical pattern wafer defect inspection equipment
Source: "28nm Key Process Defect Detection and Yield Improvement", Founder Securities Research Institute
The development trend of bright field optical pattern wafer defect detection equipment: brighter light source illumination, wider spectral range, higher rendering resolution, larger numerical aperture, larger imaging field of view, etc. Traditional light sources range from xenon lamps or mercury discharge lamps to current laser continuous discharge lamps. The light source wavelength range is 180~650nm, and the optical system is mainly lens. In order to achieve better optical resolution in a wide spectral wavelength range, multi-layer reflective lenses will be added to reduce chromatic aberration. For different types of wafers, brightfield optical pattern wafer defect detection can use different configurations, that is, different combinations of optical parameters and system parameters. The number of current equipment configurations exceeds 10,000. The current mainstream equipment on the market is KLA’s 39xx series, 29xx series and Applied Materials’ UVision series.
Electron beam pattern wafer defect inspection equipment. With the advancement of semiconductor technology and the reduction of defect tolerance in advanced processes, compared with ordinary optical brightfield and darkfield wafer defect inspection equipment, electron beam pattern wafer defect inspection equipment can detect physical defects of patterns (particles, protrusions, bridges, holes, etc.) with higher resolution and the unique ability to detect hidden defects through voltage contrast. Due to its advantages such as being able to detect defects invisible under an optical microscope through voltage contrast imaging, it has gradually played an increasingly important role and has become a powerful supplement to optical inspection equipment.
Electron beam pattern wafer defect inspection equipment is a process inspection equipment that uses scanning electron microscopes to directly detect defects in the etching patterns on integrated circuit wafers in the front-end process. Its core is a scanning electron microscope, which scans the wafer surface through a focused electron beam, receives the reflected secondary electrons and backscattered electrons, and converts them into a corresponding grayscale image of the wafer surface topography. By comparing images of the same location on different chips on the wafer, or by directly comparing images with chip design graphics, defects in etching or design can be found. Its performance emphasizes higher scanning and image acquisition rates, larger scanning fields, high-speed sample motion positioning capabilities, and image quality at low incident voltages.
Compared with optical defect inspection equipment, although electron beam inspection equipment has superior performance, its inspection speed is too slow due to its point-by-point scanning method. It cannot meet the throughput requirements of wafer factories and cannot replace optical equipment on a large scale. It undertakes online testing tasks and is currently mainly used for the development of advanced processes. The working mode is mainly sampling testing. The mainstream suppliers in the current market are ASML (acquired Hanmin Microtest Technology) and Applied Materials.
Figure: ASML HMI Series Electron Beam Patterned Wafer Defect Inspection System
Source: ASML official website, Founder Securities Research Institute
Unpatterned wafer defect detection equipment: Its function is to detect bare wafer defects and lay the foundation for subsequent patterned inspection. The types of defects that the patternless wafer surface inspection system can detect include particle contamination, pits, watermarks, scratches, shallow pits, epitaxial stacking, CMP protrusions, crystal pits, slip lines, etc. There are three main application areas:
(1) Chip manufacturing: mainly includes incoming material quality inspection, process control, wafer backside contamination detection, equipment cleanliness monitoring, etc.;
(2) Silicon wafer manufacturing: mainly includes defect detection during process development and final inspection process before silicon wafers leave the factory;
(3) Semiconductor equipment manufacturing: mainly includes defect detection in process research and development, process quality assessment of equipment (particles, metal contamination), etc.
Inspection process: Non-patterned wafer defect detection equipment can count defects on the surface of non-patterned wafers and identify the type and spatial distribution of defects. By illuminating a single-wavelength beam onto the wafer surface and using an optical system with a large collection angle, the scattered light signals of defects existing on the surface of the wafer that are moving at high speed are collected. Through multi-dimensional optical modes and multi-channel signal collection, wafer surface defects are identified in real time, the type of defects is determined, and the location of the defects is reported.
Figure: Patternless wafer defect detection process
Source: Hitachi High-Tech official website, Founder Securities Research Institute
Dark field patternless detection: In optical defect detection, dark field scattering technology is usually used for surface defect detection of precision components such as wafers. It has many advantages such as non-contact, non-destructive, high sensitivity and fast detection speed. At the same time, compared to bright field inspection, dark field inspection has faster detection speed, is more suitable for high-frequency three-dimensional topography, and can detect defects that are much smaller than the system resolution and optical size, so it is especially suitable for non-patterned wafer defects. detection.
Figure: Typical dark field scattering diagram and different defect scattering diagrams
Source: "Development of Patternless Wafer Surface Defect Detection System Based on Dark Field Scattering", Founder Securities Research Institute
Minimum sensitivity and throughput are key metrics. The minimum sensitivity indicates that the equipment can detect the diameter of the smallest particle defect on the wafer surface. The smaller the value of this indicator, the equipment can detect smaller-sized defects on the wafer surface; the throughput indicates the wafers that the equipment has completed detection per unit time. Quantity, the larger the value of this indicator, the faster the detection speed of the device.
3. Measurement and testing continues to be upgraded and is one of the main tracks for front-line equipment.
Global fab equipment spending is expected to recover to US$97 billion in 2024. According to SEMI's latest global semiconductor equipment forecast report, the global semiconductor equipment sales market size in 2023 is expected to drop 18.6% year-on-year from a record high of US$107.4 billion in 2022 to US$87.4 billion, and then recover to a market size of more than US$100 billion in 2024. . The decline in market size in 2023 is mainly due to weak chip demand and increased inventory of consumer and mobile terminal products. The recovery in market demand in 2024 is mainly due to the end of semiconductor inventory correction and the growth of semiconductor demand in high-performance computing (HPC) and automotive fields.
Front-end equipment remains the main driver of the industry’s rebound. In terms of equipment segment, the wafer fab equipment market size will decrease by 18.8% year-on-year to US$76.4 billion in 2023. It will also be the main driving force for the overall equipment market to return to US$100 billion in 2024. It is expected that the wafer fab equipment market will be The market size of factory equipment reached US$87.8 billion. In terms of the backend equipment market, due to challenges in the macroeconomic environment and weak overall demand in the semiconductor industry, the semiconductor test equipment market is expected to decline by 15% year-on-year to US$6.4 billion in 2023 (expected to increase by 7.9% year-on-year in 2024), and packaging equipment is expected to decline year-on-year. 20.5% to US$4.6 billion (expected to increase by 16.4% year-on-year in 2024).
The demand for advanced process equipment is relatively stable, while the storage equipment market fluctuates violently. Among wafer manufacturing equipment, in terms of application fields, the market for equipment used in OEM and logic factories is expected to decline by 6% year-on-year to US$50.1 billion in 2023. It is still the application field with the highest proportion in the semiconductor equipment industry. The demand for advanced process equipment in 2023 Remaining stable, equipment demand for mature nodes has declined slightly, and investment in this field is expected to increase by 3% in 2024. Due to weak consumer and enterprise market demand at the same time, the DRAM equipment market is expected to shrink by 28.8% to US$8.8 billion in 2023. However, as the market gradually recovers, SEMI predicts that this market will grow by 31% to US$11.6 billion in 2024. The NAND equipment market is expected to shrink significantly by 51% to US$8.4 billion in 2023, while also growing strongly by 59% year-on-year in 2024 to US$13.3 billion.
Chart : Global semiconductor equipment market size by segment (USD 100 million)
Figure: Global wafer manufacturing equipment market size by application (USD billion)
Source: SEMI, Founder Securities Research Institute
Mainland China leads the global semiconductor equipment market in 2024. In terms of regions, Mainland China, Taiwan Province of China, and South Korea dominate the global equipment market. Among them, SEMI predicts that mainland China will lead the global market size in 2024. At the same time, we have also seen that mainland China’s share of the global semiconductor equipment market has been on an upward trend in recent years, and the importance of the mainland equipment market is increasing.
Chart: The increasing importance of mainland China’s equipment market (left axis: billion US dollars)
Source: Wind, Founder Securities Research Institute
The semiconductor equipment industry shows obvious cyclicality, and the pace of capital expenditures of downstream manufacturers changes significantly. In 2017, storage manufacturers' substantial capital expenditures drove huge demand for semiconductor equipment, and this momentum continued into the first half of 2018. But then overcapacity caused storage prices to fall, causing DRAM and NAND manufacturers to postpone equipment orders. The overcapacity of storage continued until the first half of 2019. At the same time, the overall semiconductor industry was not prosperous in the first half of the year. Although in the second half of the year, as the industry prosperity recovered, wafer factories represented by TSMC successively increased capital expenditures and significantly expanded production. In 2019 Demand for semiconductor equipment throughout the year still fell by about 2% year-on-year. In 2020, various parts of the world have been affected by the epidemic. However, the storage industry's capital expenditure recovery, advanced process investment and digitalization, and the strong demand in various downstream fields brought by 5G have led to a year-on-year equipment market growth of 19%. As semiconductor manufacturers begin a new round of capital expenditures, the global equipment market will continue to grow significantly by 44% in 2021. Current overseas equipment leaders Applied Materials and Lam Group expect the global equipment market to further grow in 2022.
Chart: Semiconductor equipment market growth rate is cyclical
Source: Wind, Founder Securities Research Institute
The global process control semiconductor equipment market accounts for approximately 10.5%, and there is continued demand for upgrades. The global process control equipment market space in 2021 will be approximately US$10.4 billion, of which photolithography-related (overlay error measurement, mask measurement and inspection, etc.) related demand will be approximately US$2.8 billion, defect detection demand will be approximately US$5.8 billion, and film thickness measurement The demand is approximately US$1.7 billion. The process control market accounts for approximately 10.5% of the total global semiconductor equipment market (including wafer manufacturing and packaging and testing equipment), which is relatively stable. With the advancement of process shrinkage and 3D stacking, the demand for measurement and inspection in wafer manufacturing continues to increase. Accuracy requirements are also constantly increasing, and process control equipment continues to need to be upgraded.
Chart: Global process control market (USD 100 million)
Sources: VLSI, Zhongke Feichai Prospectus, Gartner, Wind, Founder Securities Research Institute
*Disclaimer: This article is original by the author. The content of the article is the personal opinion of the author. The reprinting by Semiconductor Industry Watch is only to convey a different point of view. It does not mean that Semiconductor Industry Watch agrees or supports the view. If you have any objections, please contact Semiconductor Industry Watch.
Today is the 3542nd issue of "Semiconductor Industry Observation" shared with you. Welcome to pay attention.
Recommended reading
★ Taiwan’s packaging will also be divided up
★ To break NVIDIA’s hegemony, Meta unleashes a big move!
★ ReRAM, designed to replace Nor Flash!
Semiconductor Industry Watch
" Semiconductor's First Vertical Media "
Real-time professional original depth
Identify the QR code , reply to the keywords below, and read more
Wafers | Integrated circuits | Equipment | Automotive chips | Storage | TSMC | AI | Packaging
Reply
Submit an article
and read "How to Become a Member of "Semiconductor Industry Watch""
Reply Search and you can easily find other articles you are interested in!