CMOS Under Array Patents: Who Leads, Where the Gaps Are 2026
- 9.5% concentration at the top. The five leading assignees hold 15,565 of 164,392 records in scope — a real edge, but not a lock on the field.
- Filing has cooled from its 2021 peak. Volume ran from 834 records in 2017 to a peak of 972 in 2021, then eased toward 764 at the 2022 midpoint, with the most recent year understated by publication lag.
- The most-cited prior art predates the current wave. The highest-cited records in this corpus are FET-array sensing patents from the 2009–2010 window, still shaping how examiners read array claims today.
Filing growth compares 2021 (972 records) with 2024 (575) — a three-year span. 2024 is the most recent year we treat as complete: publication lags filing by roughly 18 months, so 2025 onwards are still filling in and any growth rate that ends there would understate the field. Top-5 share is the combined record count of the five largest assignees divided by all 164,392 records in scope (CR5), not by the ranked leaders only.
What this landscape covers
CMOS under array and bonded array architectures separate the memory or sensor array from its peripheral logic, letting each be built with a different process and thermal budget before the two are joined. This landscape tracks 164,392 published records filed against wafer-, bonding- and test-access-related claim language, spanning 2015 through the 2026 data cut-off. The scope includes both under-array CMOS integration, where logic sits beneath the array on a single substrate, and bonded-array approaches that fabricate array and logic separately and join them at the wafer or die level.
Filing is dominated by memory and logic manufacturers rather than research entities, and the technology composition skews toward core semiconductor device classes rather than narrow process sub-classes, which signals that most activity still sits in general integration claims rather than in specialised bonding mechanics.
Filing trends and technology composition
Two views of the same 164,392-record set: how filing volume has moved year over year, and which IPC subclasses carry the claim density.
Filing trend, 2017–2026
Records rose from 834 in 2017 to a peak of 972 in 2021, then declined toward 764 by the 2022 midpoint and down to 47 in the most recent, still-incomplete year. Publication lag of roughly 18 months means the last one to two years will always look thinner than they eventually turn out to be.
Technology composition by IPC subclass
H01L (semiconductor devices) carries the largest share at 6.7% of all records, well ahead of H10B (memory device manufacture) and G11C (static and digital memories), each at 1.6%. Because a single record can carry several IPC classes, these shares are read against the full 164,392-record base and sum to more than 100%.
Shares are the percentage of the 164,392 records in scope. A patent can carry several IPC classes, so the shares add up to more than 100%.
Go deeper on CMOS Under Array and Bonded Array Architectures with Eureka
This page is one run against one query. Ask Eureka your own question about cmos under array and bonded array architectures and every answer comes back with the patent numbers behind it.
Try EurekaFoundational and most-cited records
US5292681A — Method of processing a semiconductor wafer to form an array of nonvolatile memory devices with peripheral CMOS
Discloses fabricating a semiconductor wafer to form a memory array and peripheral area, the array comprising nonvolatile memory devices employing floating gate transistors and the peripheral area comprising CMOS transistors. A first layer of conductive material is applied atop insulating layers, with a dielectric layer applied atop it for use in the array's floating gate transistors, then etched from the peripheral area before a second conductive layer covers the peripheral area and remaining dielectric.Filed 1994-03-08, now assigned to Round Rock Research, LLC.


| # | Publication no. | Patent title | Citations |
|---|---|---|---|
| 1 | US20090026082A1 | Methods and apparatus for measuring analytes using large scale FET arrays | 3,188 |
| 2 | US20090127589A1 | Methods and apparatus for measuring analytes using large scale FET arrays | 2,419 |
| 3 | US6034882A | Vertically stacked field programmable nonvolatile memory and method of fabrication | 2,123 |
| 4 | US5550677A | Automatic rearview mirror system using a photosensor array | 2,033 |
| 5 | US20100282617A1 | Methods and apparatus for detecting molecular interactions using FET arrays | 1,928 |
| 6 | US5877897A | Automatic rearview mirror, vehicle lighting control and vehicle interior monitoring system using a photosenso… | 1,566 |
| 7 | US7351258B2 | Apparatus and method for fixation of vascular grafts | 1,479 |
| 8 | US5661053A | Method of making dense flash EEPROM cell array and peripheral supporting circuits formed in deposited field o… | 1,222 |
| 9 | US5760962A | Automatic rearview mirror system using a photosensor array | 1,100 |
| 10 | US6185122B1 | Vertically stacked field programmable nonvolatile memory and method of fabrication | 1,055 |
Citation counts favour older filings that have had more time to accumulate citations within this searched corpus; treat them as a measure of influence on later filers, not of current commercial relevance.
Each row carries its publication number; clicking a row searches Eureka by that number.
Put your own technology through the same analysis
Eureka on the web
When you want the answer in the next five minutes.
The agent works the prompt against patents and technical literature, citing every source.
Run your analysis now →MCP server & REST API
When it has to run inside your own pipeline.
Patent search, landscape analysis and assignee resolution as MCP tools. Drop them into any agent framework, or call REST directly.
Browse MCP servers →What the numbers mean for filing strategy
Three signals from the dataset that matter more than the raw record count.
Leadership is real but not dominant
The five leading assignees combined hold 15,565 records, 9.5% of all 164,392 in scope, and the top 10 combined reach 15.7%. That leaves the large majority of the field to a long tail of filers, which means freedom-to-operate analysis cannot rely on clearing just the household names.
Filing has already crested
Volume climbed from 834 records in 2017 to a peak of 972 in 2021, then fell back toward 764 at the 2022 midpoint. Recent-year momentum data shows steep year-over-year declines across the leading assignees, though the most recent year is still understated by publication lag.
Claims cluster in general semiconductor device classes
H01L accounts for 6.7% of all 164,392 records, several times the share of any single narrower subclass such as H10B or G11C at 1.6% each. That skew toward broad device claims over specific bonding or test-access mechanics is where the open ground sits.
Eureka can read the same corpus for gaps instead of for coverage: under-claimed branches adjacent to cmos under array and bonded array architectures, with the prior art for and against each one.
Who is filing, and where the gaps sit
Filing is led by large memory and foundry manufacturers, with co-assignment patterns showing tight corporate-family filing rather than broad industry collaboration.
A single leader well ahead of the field
The top-ranked assignee holds 5,558 records against a fifth-place figure of 2,242 and a tenth-place figure of 1,910, a steep drop-off that marks out a clear leader rather than a tightly bunched top group.
Co-filing concentrates inside single corporate groups
The strongest co-assignee pairing in the dataset links a parent group with one of its subsidiaries at 145 joint filings, with two further intra-group pairings at 54 and 48. This is internal corporate-structure filing, not cross-company collaboration.
Even active filers are pulling back sharply
Every assignee tracked for recent-year momentum shows a steep year-over-year decline, from roughly -79% to -93%, though several still filed in the single digits in the latest year. This looks like a maturing claim space rather than an exit, and the newest year's counts will rise somewhat as publications catch up.
| Assignee | Recent year | YoY |
|---|---|---|
| Samsung Electronics Co., Ltd. | 9 | -86% |
| Yangtze Memory Technologies Co., Ltd. | 8 | -79% |
| BOE Technology Group Co., Ltd. | 2 | -80% |
| Taiwan Semiconductor Manufacturing Co., Ltd. | 2 | -86% |
| Micron Technology, Inc. | 1 | -93% |
| SK Hynix Inc. | 1 | -92% |
| Disco Corporation | 1 | -88% |
| International Business Machines Corporation | 0 | — |
Where to take this analysis
The dataset points to a field that has passed its filing peak but is still contested among a long tail of filers below the leaders.
Map claim scope against the leader's portfolio
With one assignee at 5,558 records against a steep drop to the tenth-ranked filer, a claim chart against the leader's active family clarifies what freedom to operate actually looks like before committing to a filing direction.
Explore assignee portfolios in EurekaWatch the under-claimed bonding and test-access branches
Narrower subclasses such as bonding pad pitch and post-bond test access sit well below the H01L share of 6.7%, suggesting room for claims that do not compete head-on with the densest prior art.
Run a white space search in EurekaCommon questions on CMOS under array patents
CMOS under array builds the peripheral logic directly beneath the memory or sensor array on a single substrate, sharing one process flow and thermal budget. Bonded array architecture fabricates the array and the logic on separate wafers or dies, each optimised for its own process, and then joins them, typically through wafer-to-wafer or die-to-wafer bonding. The bonded approach lets a manufacturer use a more advanced logic node without forcing the array to tolerate that node's thermal or process constraints, which is why bonding-specific claims around pad pitch and alignment have become a distinct filing area.
The leading assignee in this dataset holds 5,558 records, well ahead of the fifth-ranked filer at 2,242 and the tenth-ranked filer at 1,910. The five leading assignees together account for 15,565 records, 9.5% of the 164,392 records in scope, and the top ten reach 15.7%. That leaves the large majority of filings spread across a long tail of other companies, so leadership at the top does not mean the field is closed to new entrants.
No, filing volume has already passed its peak. Records rose from 834 in 2017 to a peak of 972 in 2021, then eased back toward 764 at the 2022 midpoint, and recent-year momentum figures show steep year-over-year declines across the leading assignees. The most recent year's count will understate true activity because publication typically lags filing by around 18 months, but the broader trend across the period is flat to declining rather than accelerating.
Foundational claims tied to separating array and peripheral fabrication steps, such as those in US5292681A, describe the basic pattern of building a memory array with one process sequence and CMOS peripheral transistors with another on the same wafer. Because this pattern underlies most later under-array and bonded-array implementations, later filers generally design around it through the specific bonding, alignment or test-access mechanics rather than the core separation concept itself, since that concept sits in older, broadly cited prior art rather than in a single currently enforceable position for most jurisdictions.
Narrower IPC subclasses such as H10P, H10D and H10W each account for under 1.5% of the 164,392 records in scope, well below the 6.7% held by the general H01L semiconductor device class. That gap suggests claim space specific to bonding pad pitch scaling, post-bond test access and thermal budget separation between tiers is comparatively lightly claimed relative to broad device-level integration claims, making these narrower mechanics a more promising area for a first-mover filing strategy than the crowded general device classes.
Research CMOS Under Array and Bonded Array Architectures in depth with Eureka
Go past this page: query the whole cmos under array and bonded array architectures corpus yourself, in your own scope.
Every answer comes back with patent numbers you can open.
Disclaimer. This page is generated from Patsnap Eureka data drawn from a limited snapshot of global patent and scientific-literature records, and is provided for general information and reference only.
Patent data carries inherent limitations: recent filings (typically the most recent 18–24 months) are under-counted due to standard publication lag; counts may be reported at either a patent-family or a patent-record basis and are not always directly comparable; classification, applicant-name, and citation data may contain errors, duplicates, or omissions; and the underlying search query defines and constrains the scope shown. As a result, the analysis may be incomplete or inaccurate and may not reflect the full technology landscape.
Nothing on this page constitutes an exhaustive prior-art, novelty, freedom-to-operate, or validity search, nor does it constitute legal, financial, investment, or professional advice, and it should not be relied upon as such. Any patent, commercial, or strategic decision should be verified independently and reviewed with qualified patent, legal, and domain professionals. Patsnap makes no warranties, express or implied, as to the accuracy, completeness, or fitness for any particular purpose of the information presented.
Machine translation. Assignee and organisation names originally recorded in Chinese, Japanese or Korean have been rendered into English by an AI translation step so that the tables stay readable. These renderings are best-effort and may not match a company’s registered English name; the original name is what the underlying patent record carries, and it is what any Eureka query launched from this page uses.