E-beam metrology and inspection is not a single-equipment niche, but a collection of "high-resolution process control" capabilities in the front-end of semiconductor manufacturing. It covers critical dimension (CD) metrology, defect review, and e-beam patterned wafer inspection. Its core value is not to replace all optical equipment, but to provide the information that determines yield and process convergence speed in areas where optical tools "cannot see clearly, accurately, or deeply," such as advanced nodes, 3D structures, EUV (Extreme Ultraviolet) / high-NA (Numerical Aperture) EUV, GAA (Gate-All-Around), and 3D NAND. (Applied Materials)
I. What Exactly Is This Market Segment
Strictly speaking, "e-beam metrology and inspection" encompasses at least three main categories, plus a mask-side branch. The first category is CD-SEM (Critical Dimension Scanning Electron Microscopy), which is critical dimension metrology, used to measure line width, hole diameter, edge roughness, bottom CD of deep structures, etc., essentially acting as the "ruler" of the wafer fab; the second category is DR-SEM (Defect Review Scanning Electron Microscopy) / eBeam Review, which is defect review and analysis, using high resolution to confirm what the defect exactly is, which layer it is on, its morphology, and even its composition; the third category is EBI (E-Beam Inspection) / e-beam patterned wafer inspection, which is e-beam patterned wafer inspection, directly finding critical defects on patterned wafers; the fourth category is photomask CD-SEM / mask review, serving not wafers but masks.
From an industry positioning perspective, CD-SEM leans more towards "metrology," DR-SEM leans more towards "analysis/review," and EBI leans more towards "inspection." All three use electron beams, but their technical challenges, throughput requirements, and customer purchasing logic are different. Applied Materials defines its CD-SEM on its webpage as the most precise "fab ruler," while KLA positions e-beam inspection as capable of finding critical defects that other inspection tools cannot detect. Both Applied Materials and Hitachi position review SEMs for complex 3D structure defect identification and high-volume manufacturing (HVM) yield improvement.
II. Product Classification and Respective Characteristics
- CD-SEM: The Most Mature, Most Essential, and Most "Infrastructure-Like" E-Beam Product. It is primarily deployed in key processes such as post-lithography and post-etch, used to measure critical dimensions and feed data back to process and lithography modules. In the era of EUV and high-NA EUV, photoresists are thinner and structures are finer. CD-SEM not only needs to see clearly but also maintain high resolution at low landing energies to avoid resist damage, resist shrinkage, and measurement distortion. Applied Materials' VeritySEM 10 explicitly takes "low-energy electron beam, sub-nanometer resolution, faster scan rates, and compatibility with GAA and 3D NAND deep structures" as its core selling points.
- DR-SEM / eBeam Review: The "High-Resolution Forensic Expert" in Yield Engineering. It typically receives the results from the previous defect inspection step, performing high-resolution confirmation, classification, and root cause analysis on candidate defects. Applied Materials' SEMVision G10/H20 emphasizes sub-nanometer resolution, visualization capabilities for surface and buried defects, automatic classification, and AI (Artificial Intelligence); Hitachi's defect review SEM emphasizes inline defect observation, ADC/ADR (Automatic Defect Classification / Automatic Defect Review), 3D observation, and elemental analysis. In other words, it may not be responsible for sweeping out all defects at the first instance, but it is responsible for telling you "whether this is truly a critical defect."
- EBI / e-beam inspection: The Highest Technical Barrier and Strategic Value. These tools perform e-beam inspection directly on patterned wafers, aiming to discover critical defects in advanced logic, DRAM, and 3D NAND that are difficult to find optically. Official materials for KLA's eSL10 emphasize several keywords: high resolution, high beam current density, high throughput, wide operating range, simultaneous collection of SE (Secondary Electrons) and BSE (Backscattered Electrons) in a single scan, and deep learning algorithms to distinguish real defects from noise. This shows that the essence of EBI is not "taking clearer pictures," but achieving a balance between sensitivity, throughput, false alarm rate, and complex structure adaptability that is viable for high-volume manufacturing.
- Mask-Side CD-SEM / Review: Smaller Volume Than Wafer-Side, But High Barrier. Both Advantest and Holon are very strong in photomask CD-SEM. They emphasize high-precision, high-stability, and high-throughput measurement for advanced EUV masks, masks for sub-10nm and even more advanced nodes, as well as low damage to photoresist patterns, charge suppression, AI focusing, and contour extraction capabilities. This branch is not as large as front-end wafer equipment, but the technical requirements are extremely high, and it is deeply tied to the lithography ecosystem.
III. What Are the Core Technologies in This Market Segment
The first is the electron optical system. Electron guns, lenses, aberration control, beam spot size, beam current density, and landing energy control determine whether you can achieve high resolution, low damage, and reasonable throughput simultaneously. KLA emphasizes "high beam current density under small beam spots" on the eSL10, Applied Materials emphasizes the second-generation CFE (Cold Field Emission) on the H20, and Holon directly emphasizes the new electron optical system and aberration correction.
The second is the detection and imaging system. Truly advanced tools do not rely on just one type of signal. SE is suitable for surface information, while BSE is more suitable for material contrast, deep structure, and HAR (High Aspect Ratio) structure imaging. KLA’s Simul‑6 enables simultaneous collection of SE and BSE in a single scan, and Applied Materials' VeritySEM 10 also emphasizes the importance of BSE for GAA and 3D NAND deep structure metrology.
The third is the stage, auto-focus, alignment, and tool-to-tool matching control. Whether e-beam equipment can be used in high-volume manufacturing depends not only on "the best single image" but also on whether the measurements are consistent across different tools and whether long-term drift is controllable. Hitachi directly lists tool-to-tool matching as a key performance requirement for CD-SEM on the GT2000, which is a very typical judgment.
The fourth is charging, contamination, and sample damage suppression. EUV photoresists are thinner, and certain structures are more prone to charging and damage, so the equipment must simultaneously address low landing energy, low vacuum, VUV assistance, charge suppression, and anti-contamination issues. Both Holon and Applied Materials include low damage, anti-charging, and low-energy imaging in their core selling points.
The fifth is algorithms and software. The competition in e-beam equipment is no longer "hardware vs. hardware," but "hardware + algorithms + process knowledge + data systems." KLA emphasizes AI deep learning to distinguish defects from noise, and Applied Materials emphasizes AI to automatically extract real defects and improve sorting accuracy; domestic manufacturers are also increasingly incorporating yield management software, automatic defect classification, lithography feedback modeling, and data platforms into their product matrices.
IV. What Exactly Are the Performance Requirements for This Type of Equipment
Truly production-ready e-beam equipment must meet at least six types of requirements simultaneously.
First, high resolution, but not at the expense of "slowness." Resolution is the ticket to entry, but what customers ultimately purchase is "resolution with a viable takt time." KLA's eSL10 emphasizes high resolution while maintaining high-speed scanning and high beam current density, and Applied Materials' H20 emphasizes improving processing speed and sample analysis capabilities while maintaining sensitivity and resolution.
Second, low energy and low damage. Especially in EUV and high-NA EUV scenarios, excessively high landing energy can cause resist shrinkage and measurement errors. Both VeritySEM 10 and Holon HSS-1000 take low energy and low damage as key capabilities.
Third, visualization capabilities for 3D / HAR structures. GAA, CFET (Complementary FET), 3D NAND, and deep holes/trenches make traditional top-down measurements increasingly insufficient. Hitachi's GT2000, Applied Materials' VeritySEM 10, and SEMVision G9/H20 all focus on upgrading characterization capabilities for high aspect ratios, buried defects, or deep structures.
Fourth, tool-to-tool matching control and long-term stability. Advanced fabs do not accept "a single tool looking great but drifting when switched to another." Hitachi's GT2000 explicitly lists multi-tool matching as a key requirement, and Advantest's mask CD-SEM also emphasizes long-term stable CD measurement and stable throughput.
Fifth, automation and high-volume manufacturing takt time. High-capacity wafer handling, high-speed stages, auto-focus, auto-addressing, ADC/ADR, and automatic recipe generation are what HVM lines truly care about. Holon, Advantest, and Shanghai Jingce Semiconductor all emphasize throughput, automation, and high-speed positioning capabilities in public materials. (Holon-ltd.co.jp)
Sixth, integration with fab data flows. Today, customers do not just buy a SEM; they buy its capability to integrate into the fab's ecosystem. In its annual report, Advantest has already listed automatic defect classification, yield management, virtual metrology, and lithography feedback modeling as system capabilities; SiEnvision also places OPC (Optical Proximity Correction), EBI, CD-SEM, DR-SEM, and YieldBook within a single yield management framework.
V. Major Domestic and International Players
The division of labor among overseas leaders is basically very clear.
Hitachi High-Tech is the absolute veteran leader in wafer-side CD-SEM. It officially disclosed that cumulative shipments of CD-SEM have exceeded 6,000 units, with a global market share of approximately 70%. At the same time, it continues to upgrade the GT2000 towards high-NA EUV, GAA, 3D memory, and tool-to-tool matching.
KLA is the strongest in the defect inspection and review system, particularly excelling in the "inspection + review + software" process control combination. The eSL10 represents high-end e-beam inspection capabilities, and KLA itself also has a complete optical inspection portfolio, making it more like a platform-based process control supplier in advanced production lines.
Applied Materials excels at both ends: one end is advanced CD-SEM like VeritySEM, and the other end is defect review/analysis like SEMVision H20/G10. It heavily emphasizes AI, CFE, deep defects in 3D structures, and advanced logic/memory nodes.
Advantest and Holon lean more towards mask-side e-beam metrology and review, and are important players in the mask e-beam niche market.
Domestic players are currently still in the stage of "catching up but starting to differentiate."
SiEnvision is one of the companies with the most complete platform narrative in public discourse. Its official website publicly displays OPC, EBI, CD-SEM, DR-SEM, rigorous lithography simulation PanGen Sim, and YieldBook, clearly indicating that it does not just make single-point equipment but is following the path of "equipment + software + lithography/yield data platform."
Shanghai Jingce Semiconductor also has a very complete e-beam product line. Its official website publicly lists products such as eMetric e-beam CD metrology, eView e-beam defect review, and eVC/eSpec e-beam defect inspection, along with a large amount of optical metrology and management software, indicating that it follows a comprehensive metrology and inspection route. Its eMetric page also emphasizes independent intellectual property rights for core components, hundreds of megahertz-level image acquisition, high-speed auto-focus, and high throughput.
Advantest's past strengths mainly lie in optical defect inspection, metrology, and yield software. Its 2024 annual report still focuses on disclosing optical inspection/metrology and software closed loops; it was not until the end of 2025 that the company officially released its first e-beam critical dimension metrology equipment, MAGNOLIAEBM-600, and incorporated it into a more complete front-end metrology solution. In other words, in the e-beam direction, it follows the path of "platform already exists, e-beam just completes the key puzzle piece."
While this is rarely stated explicitly, it is a fairly clear industry reality when you look at the product positioning of global incumbents, the release cadence of domestic suppliers, and their platform maturity: the progress of domestic manufacturers in CD-SEM and review/related e-beam products is significantly faster than the comprehensive maturity of truly high-end EBI.
VI. Future Industry Trends
- The More Advanced the Node, the Higher the Importance of E-Beam. Trends such as EUV, high-NA EUV, GAA, CFET, and 3D NAND have made "smaller, taller, deeper, and more complex structures" the norm. New products from Hitachi, Applied Materials, and KLA explicitly target these scenarios. E-beam is not optional; rather, it takes on more and more "observability for the last mile" in these scenarios.
- Optical and E-Beam Will Coexist Long-Term, with Tighter Synergy. Public product materials have already shown a consensus: advanced fabs do not choose "optical or e-beam," but rather "optical for broad coverage first, and e-beam for precise recognition and critical inspection/review." Applied Materials' H20 even directly mentions that optical inspection at advanced nodes will generate denser candidate defect maps, and the number of candidates handed over to e-beam systems for processing may increase by 10 to 100 times.
- AI Will Evolve from an Auxiliary Function to a Core Competitiveness. KLA writes AI into the core selling points of the eSL10, Applied Materials writes deep learning defect classification into the core value of the H20, and Holon also incorporates AI focus and positioning prediction into product capabilities. This shows that the next stage is not about who has AI, but who can truly embed AI into the inspection, review, ADC, recipe, and process feedback chain.
- Competition Will Shift from Single-Tool Parameters to Platform Capabilities. Domestic and international leaders are all integrating equipment, defect classification, yield management, lithography modeling, and data platforms. Advantest and SiEnvision have already laid out in this direction in public materials. In the future, the more valuable companies will not necessarily be those with the "highest single-tool parameters," but those that can "best embed themselves into the customer's process closed loop."
- Domestic Substitution Will First Take Root in Local Areas, Then Advance to High-End Core Segments. From public progress, domestic companies have gradually broken through in e-beam critical dimension metrology, review, and software collaboration, but in the most difficult high-end patterned wafer e-beam inspection, the system capabilities and high-volume manufacturing validation of international leaders are still stronger. In the next few years, the real watershed for domestic companies is not "whether they have released products," but whether they can cross the hurdles of advanced node validation, stable high-volume delivery, and multi-tool consistency.
E-beam metrology and inspection is a market segment characterized by "high value, high craftsmanship, and strong platform attributes" in semiconductor equipment. Among them, CD-SEM is more like the entry point for early landing and commercialization, DR-SEM is the grasp for yield analysis, and true EBI/e-beam inspection is the most difficult and the crown that best defines the platform height. The overseas landscape is already well-defined, and domestic opportunities are indeed opening up, but in the short term, the most worth paying attention to is not who shouts the most comprehensively, but who has truly linked the four things of "equipment performance, customer validation, software closed loop, and high-volume stability".
Welcome to join the industry communication group. Please note your position + company, and contact Tiger Shuo Xin (add WeChat: tigerchip)