架構權衡天平 · 技術基礎 F2ARCHITECTURAL TRADE-OFF BALANCER · FOUNDATION F2

NVM 技術全景對比:架構比對天平 NVM Technology Trade-Off: Architectural Matrix

依 5 族 NVM(晶圓代工製程方案;OTP:FG-OTP vs AntiFuse;MTP:LD vs HD;eFlash:SST vs SONOS;Emerging:ReRAM/MRAM)的 11 把技術葉,在 9 維比較維度上做系統架構橫向比對;並結合隧道物理、晶圓成本與代工 PDK 現況。具名 IP(Actt LogicFlash、NSCore TwinBit、Floadia ZT/ZA/G1/G2、CFX、Attopsemi I-fuse、SST SuperFlash)掛在既有族群葉下,機制不同者不合併。公開來源表述多數原則,非無偏見天平。 Systematic architectural cross-examination across 5 NVM family buckets (Foundry process solutions; OTP with FG-OTP vs AntiFuse; MTP with LD vs HD; eFlash SST vs SONOS; Emerging ReRAM/MRAM) covering 11 technology leaves on 9 comparison dimensions, plus tunneling physics, wafer-cost economics, and foundry PDK realities. Named IPs (Actt LogicFlash, NSCore TwinBit, Floadia ZT/ZA/G1/G2, CFX, Attopsemi I-fuse, SST SuperFlash) sit under existing family leaves and are not merged across distinct mechanisms. Grounded in public sources and majority-principle framing — not an unbiased oracle.

5 大家族5 BUCKETS 統一架構分類Unified Family Taxonomy
11 類技術11 LEAVES 獨立技術欄位Distinct Columns
0 道光罩0 MASK 邏輯相容極限Logic Adders
4 大代工4 FOUNDRIES 晶圓生態驗證Foundry PDKs
法則 #01 · 光罩開銷RULE #01 · MASK OVERHEAD
0 道額外光罩0 MASK ADDERS
邏輯相容光罩模型 LOGIC-COMPATIBLE MASK MODEL
零額外光罩 ≠ 晶圓免費 · 見成本恆等式 Zero extra masks ≠ free wafers · see cost identity
法則 #02 · 耐熱嚴格性RULE #02 · THERMAL RIGOR
175°C 接面溫度等級175°C Tj CLASS
高溫留存等級目標 HIGH-TEMP RETENTION CLASS
架構等級 · 非認證保證 Architecture class · not a cert claim
法則 #03 · 微縮前沿RULE #03 · SCALING EDGE
源自 1.8V CoreFROM 1.8 V CORE
Core 閘氧 AntiFuse 製程窗 CORE-OXIDE ANTIFUSE WINDOW
較厚的 2.5 V 閘氧常先碰上接面崩潰 Thicker 2.5 V oxide often loses to junction BV
檢核 #04 · 製程整合CHECK #04 · PROCESS INTEGRATION
28 nm 整合檢核28 nm CHECKPOINT
依具名 PDK 與記憶體巨集 NAMED PDK AND MEMORY MACRO
元件堆疊與供電條件 · 非普遍物理截止線 Device stack and supply conditions · no universal cutoff

28 nm 是檢查製程產品組合的節點,不是嵌入式 Flash 的普遍物理截止線。可用性、額外光罩、寫入/抹除電壓、耐寫次數及資格驗證,須依具名晶圓代工製程與記憶體巨集核對。例如 Microchip/GF 於 2023-09-27 公告 ESF3 在 GF 28SLPe 量產;這不能推廣成其他節點或元件堆疊的保證。 28 nm is a process-portfolio checkpoint, not a universal physical cutoff for embedded Flash. Availability, added masks, program/erase voltage, endurance and qualification must be checked for the named foundry process and memory macro. For example, Microchip / GF announced ESF3 production on GF 28SLPe on 2023-09-27; that does not guarantee availability on another node or device stack.

01 · 全方位技術橫向基準01 · COMPREHENSIVE BENCHMARK

st-nvm-family-v0 · 十一技術九維度對比矩陣 st-nvm-family-v0 · 9-Dimension Benchmark Across 11 Leaf Technologies

評估維度DIMENSION 評估維度Dimension
eFuse
Mask ROM
EEPROM
FG-OTP
AntiFuse
LD-MTP
HD-MTP
SST
SONOS
ReRAM
MRAM
光罩增加道數Mask Adders 晶圓成本加價Wafer cost adder
+0代工標準整合Foundry standard
0–1via/contact 圖樣Via / contact pattern
待查驗VERIFY額外 poly/模組層數Extra poly / module count
+0logic-IP 級Logic-IP class
+0logic-IP 級Logic-IP class
通常 +0Typically +0通常 +0;供應商分流Typically +0; vendor-split
+0 ~ +2供應商分流;見註腳Vendor-split; see footnote
+4 / 待查驗VERIFY / +10ESF1/ESF2/ESF3ESF1 / ESF2 / ESF3
+2 ~ +4G1 +2–3/G2 +4G1 +2–3 / G2 +4
+2 ~ +4BEOLBEOL
+3 ~ +5BEOL MTJBEOL MTJ
位元胞面積等級Cell Area Class 矽面積量級Silicon footprint class
大Large吹斷級 60–120 F²;I-fuse 熔絲面積另核對Blow-class 60–120 F²; I-fuse fuse area separately
最小Smallest< 8 F²
中Medium20–40 F²
大(I/O 級)Large (I/O Class)100–250+ F²*
極小(Core 1T)Ultra-Small (Core 1T)10–25 F²*
大Large60–120+ F²
中/緊湊Medium/Compact25–50 F²
緊湊Compact12–30 F²
中Medium15–35 F²
緊湊Compact15–35 F²
中Medium30–50 F²
PGM/ERS 機制PGM / ERS Mechanism 寫入與抹除物理Program & erase physics
電遷移熔斷EM link blow吹斷級;I-fuse 另列;CFX 另列 eFuse 商品線Blow class; I-fuse separate; CFX also lists an eFuse SKU
光罩圖樣Mask pattern無電氣 PGMNo electrical PGM
FN / FN雙層多晶矽浮閘Double-poly floating gate
CHEI / NoneCHEI/無浮閘注入;CFX 亦列此路徑FG inject; CFX also lists this path
閘氧擊穿Gate-oxide breakdown不可逆導通;CFX 教學為閘氧,ZA 位點未公開Irreversible conduction; CFX teaching GOX, ZA site unpublished
FN / FNNeoEE/ZT 級;Actt 未公開NeoEE / ZT class; Actt unpublished
CHI/FN 或 FN/FNCHI/FN or FN/FNNeoMTP 等 CHI/FN;AEON 級 FN/FN — 非 TwinBit SchottkyNeoMTP-class CHI/FN; AEON-class FN/FN — not TwinBit Schottky
SSI / FNSuperFlash SSI/尖端 FNSuperFlash SSI / tip FN
FN / FNG1 捕捉;G2 夾心開關G1 trap; G2 sandwiched switches
Forming + SET/RESET成形 + SET/RESET細絲形成與切換Filament form & switch
STT toggleSTT 切換自旋轉矩切換Spin-torque switching
編程電壓與電流Program V / I 隨製程裝置,不是常數Process-set, not a constant
高電流High I吹斷 ~10–15 mA;I-fuse 另列Blow ~10–15 mA; I-fuse separate
不適用N/A光罩寫入Mask write
FN / HV pumpFN/高壓幫浦~8–14 V 量級~8–14 V class
I/O 裝置視窗I/O device window見註腳 FG-OTPSee FG-OTP footnote
隨 core 閘氧Tracks core oxide數倍 core VddSeveral × core Vdd
FN 視窗FN window低擾動幫浦Low-disturb pump
FN + FN(或 CHI/FN)FN + FN (or CHI/FN)AEON 級 FN 寫抹;NeoMTP 等 CHI/FN — 分流AEON-class FN PGM/ERS; NeoMTP-class CHI/FN — split
SSI + HV pumpSSI + 高壓幫浦寫 10V+5V/抹 12–15VProg 10V+5V / Ers 12–15V
FN / HV pumpFN/高壓幫浦G1/G2 皆 FN/HV;VDD 是讀G1/G2 both FN/HV; VDD is read
Forming / SET成形/SETForming 2.5V / SET 1.5VForming 2.5V / SET 1.5V
接近 VDDNear VDD~1.1V · Ic ~30–80 µA~1.1V · Ic ~30–80 µA
高溫留存等級Retention Class 架構等級,非認證Architecture class, not cert
中溫級Mid-temp class吹斷級 ~10 yr @ 125°C;I-fuse 另核對具名 IPBlow-class ~10 yr @ 125°C; I-fuse check named IP
永久級Permanent class光罩固定Mask-fixed
中溫級Mid-temp class~10 yr @ 125°C 量級~10 yr @ 125°C class
中溫級Mid-temp class~10 yr @ 125°C 量級~10 yr @ 125°C class
高溫級High-temp class~10 yr @ 175°C(架構級,非認證;Synopsys public)量級~10 yr @ 175°C (architecture class, not cert; Synopsys public)
中溫級Mid-temp class~10 yr @ 125°C 量級~10 yr @ 125°C class
中溫級Mid-temp class~10 yr @ 125°C 量級~10 yr @ 125°C class
高溫車規級Auto Grade 010 yr @ 150°C / 20 yr @ 125°C10 yr @ 150°C / 20 yr @ 125°C
具名分流Named splitG1 中溫級;G2 表列 175°CG1 mid-temp; G2 table 175°C
商業/工業級Commercial/Ind~10 yr @ 85–105°C~10 yr @ 85–105°C
中溫級Mid-temp class~10 yr @ 125°C 量級~10 yr @ 125°C class
讀取延遲(陣列/SA)Read Latency (Array / SA) 非 SRAM shadow 路徑Not SRAM shadow path
~1 µs開機 shadow 到 SRAMBoot-shadow to SRAM
~5–15 ns靜態 1T 陣列Static 1T array
~20–40 nsSA 翻轉受限SA trip limited
~25–40 nsSA 翻轉受限SA trip limited
~25–40 ns受 cell current 限制Cell-current limited
~20–40 nsSA 翻轉受限SA trip limited
~20–40 nsSA 翻轉受限SA trip limited
~15–25 ns分離閘極較快Split-gate faster
~20–35 nsG1 2T;G2 VDD 讀仍為陣列/SAG1 2T; G2 VDD-read still array/SA
數十 nsTens of ns陣列/SA 受限Array / SA limited
~10–25 nsTMR 裕度與防擾動限制TMR margin & disturb bound
覆寫耐久度Endurance 可重寫次數等級Rewrite cycle class
1 次1 cycle不可逆Irreversible
0 次0 cycles唯讀Read-only
104–105氧化層磨耗Oxide wear
1 次1 cycleOTP 單調OTP monotonic
1 次1 cycle不可逆Irreversible
104–105家族量級;ZT 本文/表衝突,非保證Family class; ZT body/table conflict, not a guarantee
103–104FN 穿隧/介面磨耗(AEON 級);CHI 路線另計FN tunneling / interface wear (AEON-class); CHI routes differ
104–105氧化層磨耗Oxide wear
104–105捕捉層磨耗;G2 開發中,非量產保證Trap-layer wear; G2 in development, not a production guarantee
105–106細絲磨耗Filament wear
106–1010受限於 ~1nm MgO TDDB(例:GF IRPS 2019)Bounded by MgO TDDB (e.g. GF IRPS 2019)
製程微縮Scaling 成熟下限與先進上限Mature floor & advanced ceiling
電流/IR 受限Current / IR limited吹斷級;I-fuse 另列Blow class; I-fuse separate
金屬間距Metal pitch接觸密度受限Contact density bound
平面 ~90/130 nmPlanar ~90/130 nm雙層多晶矽熱預算限制DP thermal budget bound
無(受限 I/O)None (I/O Bound)平面 ≤28nm 止步*Planar ≤28nm floor*
微縮至 2nm GAADown to 2nm GAA隨 Core 薄閘氧*Tracks Core GOX*
平面 ~28 nm 止點Planar ~28 nm limit單層浮閘;TwinBit 不屬此葉Single-poly FG; TwinBit not this leaf
平面 ~28 nm 止點Planar ~28 nm limit單層浮閘;TwinBit 不屬此葉Single-poly FG; TwinBit not this leaf
平面 ~28 nmPlanar ~28 nmHV 幫浦 vs FinFETHV pump vs FinFET
平面 ~28 nmPlanar ~28 nmG1 公開至 40 nm;G2 55BCDG1 public to 40 nm; G2 55BCD
BEOLForming 隨節點Forming node-set
BEOL MTJ隨代工 BEOLFoundry BEOL set
防竄改幾何Anti-Tamper Geometry TEM/探針衰減,非免疫TEM / probe attenuation, not immunity
熔斷空洞可見Melt void visible吹斷級 SEM;I-fuse 另列Blow-class SEM; I-fuse separate
去層全暴露De-layer exposed光學可見Optically visible
電荷探測Charge probingEM/雷射攻擊面EM / laser attack surface
浮閘電荷FG charge電荷/雷射面Charge / laser surface
次閘氧微絲Sub-gate filamentFIB 衰減,非免疫FIB attenuation, not immunity
浮閘電荷FG charge電荷探測面Charge probe surface
浮閘電荷FG charge電荷探測面Charge probe surface
浮閘電荷FG charge電荷探測面Charge probe surface
捕捉層電荷Trap charge電荷探測面Charge probe surface
細絲部分可見Filament partly visibleSTEM 可觀測STEM observable
磁化可探Magnetization probeableMFM 攻擊面MFM attack surface
📌 表格星號註解 (* Footnotes) Table Asterisk Notes (* Footnotes)
* 位元胞面積等級(Row 2:FG-OTP 100–250+ F² vs. AntiFuse 10–25 F²):Cell Area Class (Row 2: FG-OTP 100–250+ F² vs. AntiFuse 10–25 F²): 兩者具數量級差距。FG-OTP 本質為 I/O Cell,電荷保持力依賴 5V/3.3V 厚閘氧化層(取 3.3V 最薄下限 tox ≥ 70 Å,5V 達 120–150 Å,阻斷直接穿隧),受限於 I/O 規則與間距,單元實體面積常態為 1.5–3.5 µm²,以邏輯 Core 基準換算等效達 100–250+ F²(在先進節點等效 F² 更達數百倍)。AntiFuse 則為 Core Cell,採用核心邏輯薄閘氧擊穿,電壓與幾何完全隨 Core 微縮(10–25 F²,實體面積僅 ~0.05–0.1 µm²);且 Synopsys (Sidense) 1T 獨有專利(如 US7402855B2 等)以分裂通道免除獨立存取電晶體,陣列面積較傳統 2T AntiFuse 進一步縮減對半(50% 陣列精簡),使 1T AntiFuse 陣列僅為 FG-OTP 面積的 1/10 ~ 1/25。 An order-of-magnitude difference exists between the two. FG-OTP is fundamentally an I/O Cell requiring 5V/3.3V thick gate oxide (tox ≥ 70 Å at 3.3V lower bound, 120–150 Å for 5V, suppressing direct tunneling) for retention, locking physical cell size at 1.5–3.5 µm² (100–250+ F² equivalent in core logic metrics). AntiFuse is a true Core Cell utilizing thin core gate oxide breakdown (10–25 F², physical size ~0.05–0.1 µm²); moreover, Synopsys (Sidense) proprietary 1T split-channel patent (e.g. US7402855B2) eliminates the select transistor, cutting traditional 2T array area in half (50% reduction), making the 1T array 1/10 to 1/25 the size of FG-OTP.
* 製程微縮極限(Row 8:Scaling):Process Scaling Limit (Row 8: Scaling): FG-OTP 幾何鎖死在厚氧化層 I/O 規則,無法隨 Core 縮放,平面 28nm 即為物理停滯點且無原生 FinFET 移植;AntiFuse 核心閘氧擊穿特性隨標準邏輯閘長、CPP 與高介電金屬閘極等比演進,一路原生微縮至 FinFET 與 2nm GAA / BSPDN。 FG-OTP geometry is locked by thick-oxide I/O rules, preventing core scaling with a planar ≤28nm limit and no native FinFET port; AntiFuse tracks core gate oxide, CPP, and HKMG scaling natively down to FinFET and 2nm GAA / BSPDN.
工程註解與邊界NOTES & BOUNDARIES 架構工程邊界與九維度註解規範 Architectural Boundary & 9-Dimension Engineering Notes
⚠️ 表內數據為架構量級,非具名巨集 Datasheet 或出貨保證;零額外光罩 ≠ 晶圓免費 ⚠️ Figures are architecture-class ranges, not named-macro datasheets; zero extra masks ≠ free wafers
01 · 光罩加價與晶圓經濟學01 · MASK ADDERS & WAFER ECONOMICS

光罩加價道數與供應商分流 Mask Adders & Vendor Split

  • 光罩增加道數為架構等級:Architecture Class: 非代工廠 PDK 報價。EEPROM-DP 不得硬編 +4–8,再計額外 poly/模組層數。 Not PDK quotes. Do not hard-code EEPROM-DP at +4–8; count extra poly/module layers.
  • MTP 供應商分流:MTP Vendor Split: LD-MTP 通常為 +0(如 NeoEE、Floadia LEE Flash ZT)。HD-MTP 須分流:Impinj/AEON 級 +0;YMC +0–2;eMemory NeoMTP baseline +0、gen2 +2(well-implant 光罩,單元面積 25–50 F²,見 ChipEstimate 2022-11-29)。Actt LogicFlash® 公開為 180–55 nm、+0–1,機制未公開,不得回填 FN/FN 或 CHI/FN。NSCore TwinBit™ Gen-2 為 0 光罩 Pch Schottky,不得併入 LD/HD 浮閘葉。 LD-MTP is typically +0 (e.g. NeoEE, Floadia LEE Flash ZT). HD-MTP requires vendor split: Impinj/AEON +0; YMC +0–2; eMemory NeoMTP baseline +0 / gen2 +2 (well-implant masks, cell 25–50 F² per ChipEstimate 2022-11-29). Actt LogicFlash® is public at 180–55 nm, +0–1, with unpublished cell physics — do not back-fill FN/FN or CHI/FN. NSCore TwinBit™ Gen-2 is 0-mask Pch Schottky and must not be merged into the LD/HD floating-gate leaves.
  • eFlash 模組階梯:eFlash Hierarchy: SST 不得合併:ESF1 例 +4(GF 130BCDLite);ESF2 VERIFY;ESF3 例 +10(GF 28SLPe 含 5V I/O)。SONOS 須分流:Floadia G1 級 +2–3(180 BCD 公開表列例為 +2),非 +6–10;G2 公開為 +4(55 BCD 表列,開發中),讀取為 VDD、SONOS 夾於開關電晶體之間。原廠 PDK/SPICE 保留(G1/G2 公開頁)。 SST must not be lumped: ESF1 ex. +4 (GF 130BCDLite); ESF2 VERIFY; ESF3 ex. +10 (GF 28SLPe incl. 5V I/O; Microchip/GF public). SONOS requires a named split: Floadia G1-class +2–3 (180 BCD public table lists +2), not +6–10; G2 is public at +4 (55 BCD table, in development) with VDD-read and a SONOS cell sandwiched by switch transistors. Original PDK/SPICE preserved (G1 / G2 public pages).
02 · 單元拓撲與製程整合02 · CELL TOPOLOGY & PROCESS INTEGRATION

位元胞幾何與 3D 微縮邊界 Cell Geometry & 3D Scaling Limits

  • FG-OTP vs. AntiFuse 單元面積與微縮本質:FG-OTP vs. AntiFuse Footprint & Scaling Rigor: FG-OTP 屬 I/O Cell(100–250+ F²,實體約 1.5–3.5 µm²):浮閘電荷保持力(Retention)依賴 5V 或 3.3V 厚閘氧化層(取 3.3V 最薄下限 tox ≥ 70 Å,5V 達 120–150 Å,阻斷直接穿隧漏電),無法改用 core 薄閘氧;位元胞與周邊受限於 I/O 設計規則與間距,無法隨先進核心邏輯節點等比微縮(無法 scaling,平面 28nm 止步且無原生 FinFET 移植)。AntiFuse 則屬 Core Cell(10–25 F²,實體僅 ~0.05–0.1 µm²):直接利用標準核心邏輯薄閘氧(core GOX)擊穿,電壓與幾何完全隨 core 閘長與 CPP 等比微縮(原生支援 FinFET 與 2nm GAA)。在位元胞架構層次:傳統 AntiFuse 需採 2T(1 存取電晶體 + 1 反熔絲電容)架構以防讀寫擾動;而 Synopsys(原 Sidense 獨有專利,如 US7402855B2)之 1T-OTP 分裂通道(Split-Channel)技術,省去了獨立存取電晶體,相較傳統 2T 能將單元陣列面積進一步縮減對半(50% 陣列面積精簡),確立極緊湊(Ultra-compact)矽佔用優勢,單元面積僅為 FG-OTP 的 1/10 ~ 1/25。 FG-OTP is fundamentally an I/O Cell (100–250+ F², ~1.5–3.5 µm²): Floating gate charge retention strictly requires 5V or 3.3V thick gate oxide (tox ≥ 70 Å at 3.3V lower bound, 120–150 Å for 5V, suppressing direct tunneling loss) to prevent charge loss, precluding thin core oxide; cell dimensions are bound to I/O pitch and cannot scale with core logic shrinks, stalling at planar 28nm without native FinFET port. Conversely, AntiFuse is a Core Cell (10–25 F², ~0.05–0.1 µm²): Built directly on standard thin core gate oxide (GOX), breakdown voltage and dimensions natively scale with core gate length and CPP down to FinFET and 2nm GAA. At the cell architecture level: Traditional AntiFuse requires a 2T topology (1 access transistor + 1 antifuse capacitor) to prevent disturb; whereas Synopsys (originally Sidense proprietary patent, e.g., US7402855B2) 1T-OTP split-channel bitcell eliminates the separate access device, cutting traditional 2T cell array area further in half (50% array area reduction), achieving true ultra-compact silicon efficiency at 1/10 to 1/25 the size of FG-OTP.
  • 單元面積物理約束:Cell Area Constraint: LD-MTP 因需 >80% 電容耦合比,單元面積達 60–120+ F²;具名 CHI 路線的 HD-MTP 可藉熱載子注入縮小耦合電容至 25–50 F²(不含 AEON FN/FN 寫抹)。 LD-MTP requires >80% coupling ratio, requiring 60–120+ F² cells; named CHI-route HD-MTP leverages hot-carrier injection to compress cell area to 25–50 F² (excludes AEON FN/FN program).
  • FinFET 整合需個別查證:Verify FinFET integration individually: 平面單層浮閘的單元與高壓周邊,不能只靠節點名稱就假定可移植至 FinFET;須核對具名 PDK、耦合結構、供電及熱預算,不以 28/22 nm 作通用截止線。此類浮閘討論不適用 NSCore TwinBit™ Gen-2:原廠公開為 Pch Schottky 單元、0 額外光罩、40–22 nm 及 beyond(TwinBit Gen-2),不得把 Schottky 熱載子路徑回填成單層浮閘。 Do not assume a planar single-poly floating-gate cell and its high-voltage peripherals port to FinFET from a node name alone. Verify the named PDK, coupling structure, supply and thermal budget; 28/22 nm is not a universal cutoff. Floating-gate reasoning does not apply to NSCore TwinBit™ Gen-2: the vendor identifies a Pch Schottky cell, 0 extra masks, 40–22 nm and beyond (TwinBit Gen-2). Do not back-fill Schottky hot-carrier physics onto single-poly FG.
  • 先進節點 0-Mask 延伸:Advanced Node Extension: AntiFuse 編程在教學葉上常是 core 裝置閘氧崩潰:Vpgm 由該製程閘氧厚度與允許擊穿時間決定;常被定位為可隨 core 縮放進入含 2nm GAA 的先進節點類別(依 vendor/世代;公開矽例 VLSI 2025 DOI,非萬能保證)。CFX 教學預設才是閘氧路徑。Floadia LEE Fuse ZA 可掛同一葉,但公開只到介電層永久導通,不指定擊穿點在閘極或電容,不得把閘氧微絲模型套上去;涵蓋 180 nm 至 sub-10 nm,量產軌跡為 1x nm DRAM,28 nm 邏輯列為開發中(LEE Fuse ZA)。 On the teaching leaf, AntiFuse programming is often core-device gate-oxide breakdown, with Vpgm set by that process oxide and allowed time-to-breakdown; it is often positioned to scale into advanced GAA classes including 2nm (vendor/generation dependent; public Si ex. VLSI 2025 DOI — not a universal guarantee). The CFX teaching default is the GOX path. Floadia LEE Fuse ZA may sit on the same leaf, but it is published only as a dielectric made permanently conductive and does not locate the breakdown in a gate or a capacitor — do not apply the GOX filament model. Public coverage is 180 nm to sub-10 nm, with a 1x nm DRAM mass-production track and 28 nm logic listed as under development (LEE Fuse ZA).
03 · 寫入機制與可靠度邊界03 · PROGRAMMING & RELIABILITY

載子傳輸機制與耐久度極限 Carrier Transport & Endurance Limits

  • SST 來源端注入 (SSI):SST Source-Side Injection: SST 寫入採用專利來源端注入(SSI)高效率熱電子傳輸,抹除則為多晶矽尖端 FN 穿隧(SST SuperFlash 公開族)。NSCore TwinBit™ Gen-2 寫入為熱電洞、抹除為熱電子(Pch Schottky;原廠 TCAD 頁),不得與 SSI、FN/FN 或 CHI/FN 互換。Attopsemi I-fuse® 為熱輔助電遷移、刻意低於熱失控吹斷(I-fuse;IEEE JEDS 2019 22 nm FD-SOI 例約 1 V/1 mA),不得與代工標準 eFuse 熔斷、閘氧 AntiFuse 或浮閘 OTP 合併。 SST programming relies on patented Source-Side Injection (SSI) and poly-tip FN erase (SST SuperFlash family). NSCore TwinBit™ Gen-2 programs by hot holes and erases by hot electrons (Pch Schottky; vendor TCAD page) and must not be swapped with SSI, FN/FN, or CHI/FN. Attopsemi I-fuse® is heat-assisted electromigration deliberately below thermal-runaway blow (I-fuse; IEEE JEDS 2019 22 nm FD-SOI example ~1 V / ~1 mA) and must not be merged with foundry-standard eFuse blow, gate-oxide AntiFuse, or floating-gate OTP.
  • eMRAM TDDB 壽命極限:eMRAM TDDB Limit: 代工量產 eMRAM 覆寫常以約 1 nm MgO 穿隧絕緣層 TDDB 為架構級邊界(公開例 GF IRPS 2019,仍須對照目標 vendor 表):eFlash 替換模式常見量級 10⁵–10⁶ 次,工作快取模式 10⁸–10¹⁰ 次,不應與實驗室原型之 >10¹² 次混淆。 Commercial eMRAM endurance is commonly framed by ~1nm MgO TDDB as an architecture-class bound (e.g. public ex. GF IRPS 2019; still confirm target vendor tables): often 10⁵–10⁶ for eFlash-replacement class and 10⁸–10¹⁰ for cache class, distinct from >10¹² lab prototypes.
  • STT-MRAM 陣列讀取延遲:STT-MRAM Read Latency: 受限於 100%–180% 之微弱 TMR 信號與防止誤翻轉之低壓偏置,原生陣列延遲為 10–25 ns,非 SRAM 級別之 1–2 ns。 STT-MRAM array read latency is 10–25 ns due to 100%–180% TMR margins and disturb bounds, not 1–2 ns SRAM-speed.
04 · 電壓視窗與實體安全04 · VOLTAGE WINDOWS & SECURITY

電壓視窗與實體安全責任鏈 Voltage Windows & Layered Security

  • 編程電壓視窗區隔:Voltage Window Separation: 公開專利把約 32 Å 閘氧的 VPP 放在 8–9 V、約 20 Å 放在 5–6 V;原廠具名製程例顯示 1.8 V core 須承受遠高於 Vdd 的編程應力。FG-OTP 的 I/O 浮閘熱載子注入(3.3 V/5 V PMOS 約 6.5 V/7.5 V PGM)為公開文獻/架構量級,不得回填到 core 閘氧 AntiFuse。 Public teaching places VPP near 8–9 V for ~32 Å and 5–6 V for ~20 Å. FG-OTP I/O floating-gate HCI (~6.5 V on 3.3V PMOS, ~7.5 V on 5V) is public literature and must not be back-filled onto core-oxide AntiFuse.
  • 讀取延遲與開機快取:Read Path & Boot Shadowing: 讀取延遲指陣列/sense amplifier 路徑;吹斷級 eFuse 原生約 µs 量級,實務上開機 shadow 到 SRAM。OTP 陣列讀取普遍極限約 25 ns、多數約 40 ns。SST 分離閘極 eFlash 通常稍快,約 15–25 ns 量級。I-fuse 讀熔絲電阻;閘氧 AntiFuse 讀閘極電流;ZA 感測路徑未公開。延遲仍以該版巨集為準,不得把 µs 原生 eFuse 數字套到 I-fuse。 Read latency is array/SA path. Native blow-class eFuse sensing is ~µs-class, shadowed to SRAM at boot. OTP array reads typically sit near 40 ns, practical floor ~25 ns. SST split-gate is ~15–25 ns class. I-fuse reads fuse resistance; GOX AntiFuse reads gate current; ZA sense path is unpublished. Keep native µs eFuse figures off I-fuse macros.
  • TEM / DPA 防護責任鏈:TEM / DPA Layered Defense: 非揮發微絲為永久性物理歐姆結構,位元單元本身無法「單週期自我清除」;系統防護依賴主動頂層金屬網感測觸發揮發暫存器即時清零。TEM/DPA 描述為幾何與架構衰減,非絕對免疫。具名產品仍以該版 IP 為準。 Physical filament is an irreversible ohmic structure and cannot self-zeroize; system defense relies on active mesh detecting tampering to zeroize volatile shadow registers. TEM/DPA descriptions represent geometric/architectural attenuation, not absolute immunity. Check the specific IP revision.
05 · 具名 IP 歸屬與查核05 · NAMED IP UNDER FAMILY LEAVES

具名 IP 對應既有 11 葉(機制不同者不併欄) Named IPs Map onto the 11 Leaves (Do Not Merge Distinct Physics)

  • OTP

    CFX 公開三條路徑——閘氧 AntiFuse、eFuse、浮閘 OTP——教學預設為閘氧崩潰,不得把三條路徑併成同一格(單元導讀;chuangfeixin.com)。Floadia LEE Fuse ZA 掛 AntiFuse 葉:0 額外光罩、180 nm 至 sub-10 nm;公開只到介電層永久導通,擊穿位點未指定,不得寫成閘氧路徑;頁面曾寫「LEE Flash ZA」,產品名仍為 LEE Fuse ZA(單元導讀)。Attopsemi I-fuse® 掛 eFuse 族群但不是吹斷級 eFuse、也不是 AntiFuse:poly/metal-gate/metal fuse,熱輔助電遷移且低於熱失控(單元導讀)。 CFX publishes three routes — GOX AntiFuse, eFuse, and floating-gate OTP — the teaching default is oxide breakdown and the three must not collapse into one cell (cell study; chuangfeixin.com). Floadia LEE Fuse ZA sits on the AntiFuse leaf: 0 extra masks, 180 nm to sub-10 nm; published only as a dielectric made permanently conductive, breakdown site unpublished — not a GOX path; the page once says “LEE Flash ZA”, but the product name is LEE Fuse ZA (cell study). Attopsemi I-fuse® sits in the eFuse family but is not blow-class eFuse and not AntiFuse: poly / metal-gate / metal fuse programmed by heat-assisted EM below thermal runaway (cell study).
  • MTP

    Floadia LEE Flash ZT 掛 LD-MTP 葉:0 光罩、FN 寫/抹、浮閘(新聞稿);公開頁 P/E 次數本文與表不一致,不得當共通保證(單元導讀)。Actt LogicFlash® MTP(CMT 為 2016 年收購沿革,不是現行 SKU)公開 180–55 nm、+0–1、類 Flash 的 byte PGM/sector 或 chip ERS;產品頁最高約 10k 次,非本表保證;單元物理未公開(單元導讀)。NSCore TwinBit™ Gen-2 不掛 LD 也不掛 HD:Pch Schottky、熱電洞寫入/熱電子抹除、40–22 nm、0 光罩(單元導讀)。PermSRAM 是 OTP 旁系(熱載子進 SiN 側壁),不得與 TwinBit 合併。 Floadia LEE Flash ZT sits on the LD-MTP leaf: 0 extra masks, FN program/erase, floating gate (press). Public P/E-cycle body vs table conflict — not a common guarantee (cell study). Actt LogicFlash® MTP (CMT is 2016 acquisition lineage, not a current SKU) is public at 180–55 nm, +0–1, Flash-like byte PGM / sector or chip ERS; product-page up to ~10k cycles, not a table guarantee; cell physics unpublished (cell study). NSCore TwinBit™ Gen-2 sits on neither LD nor HD: Pch Schottky, hot-hole program / hot-electron erase, 40–22 nm, 0 extra masks (cell study). PermSRAM is an OTP sibling (hot-carrier into SiN sidewall) and must not merge with TwinBit.
  • eFlash

    SST 葉即 SST SuperFlash 分離閘極族(SSI/interpoly FN;公開節點約 180–28 nm,首頁亦見 500 nm–28 nm)(單元導讀)。Floadia G1 掛 SONOS 葉:+2–3、FN P/E、BCD(單元導讀)。G2 同為 SONOS 但 +4、VDD 讀、夾心開關、55 BCD 表列且標示開發中(單元導讀);VLSI-DAT 2020 以 TCAD 討論低於 28 nm 的可行性,不是 FinFET 量產聲明。常憶科技 (Chingis Technology, chingistek.com) pFusion™/pFlash™ 掛 P-Channel 浮閘 eFlash 類:帶帶穿隧熱電洞注入 (BBHH) 編程與 FN 抹除,微安培級寫入電流,標準 CMOS 相容。智憶科技 (IOTMemory, iotmemory.com) SilvoFlash™/SilvoFE™ 屬專利低光罩超低壓 eFlash:原生 0.9V–1.2V 核心電壓直接讀取,打破傳統 eFlash ≥1.8V 門檻,支援 40nm SoC 合作並具備 DDR5 SPD IC 量產實績。 The SST leaf is the SST SuperFlash split-gate family (SSI / interpoly FN; public foundry nodes ~180–28 nm; home also states 500 nm–28 nm) (cell study). Floadia G1 sits on the SONOS leaf: +2–3, FN P/E, BCD (cell study). G2 is also SONOS but +4, VDD-read, sandwiched switches, 55 BCD table, marked in development (cell study); VLSI-DAT 2020 discusses sub-28 nm feasibility in TCAD — not a FinFET production claim. Chingis Technology (chingistek.com) pFusion™ / pFlash™ sits on the P-Channel floating-gate eFlash branch: band-to-band hot-hole injection (BBHH) programming with FN erase, microamp-class write currents, and standard CMOS compatibility. IOTMemory (iotmemory.com) SilvoFlash™ / SilvoFE™ represents low-mask ultra-low-voltage eFlash: native 0.9V–1.2V core logic read operation bypassing the conventional ≥1.8V barrier, partnered on 40nm SoCs with volume DDR5 SPD IC deployments.
01B · SOURCE-BOUND CASES01B · 可追溯具名案例

Named Cases · Conditions Differ, No Ranking具名案例・條件不同,不作排名

Two embedded MCU implementations, three documented metrics. Different capacities, interfaces, temperature definitions and test conditions: no ranking or cross-device speed ratio.兩個嵌入式 MCU 實作、三項有來源的指標。容量、介面、溫度定義與測試條件不同,不作排名或跨元件速度倍數換算。

Manufacturer datasheets · not independent measurements. Each metric inherits its device conditions below; row-specific conditions take precedence.原廠規格書・非獨立實測。各指標連同下方元件條件解讀,以該列特定條件為準。

ST STM32U575xxEmbedded FlashConditions and unknowns條件與未知項

2 MB configuration; ordering suffix not selected.2 MB 組態;未選定訂購後綴。

Process node製程節點
Unknown未知 · Not stated in this datasheet.此規格書未載明。
Macro / silicon revision巨集/晶片修訂版
Unknown未知 · Macro / silicon revision not identified by these tables.表格未指定巨集/晶片修訂版。
Capacity容量
2 MB; two 1 MB banks.2 MB;兩個 1 MB 記憶庫。 DS13737 Rev 10 · §3.4 · p. 24
Interface / granularity介面/粒度
Internal Flash controller; 128-bit program, 8 KB erase page.內部 Flash 控制器;128-bit 編程、8 KB 抹除頁。 DS13737 Rev 10 · Table 86, pp. 223–224 · p. 223
Device supply元件供電
VDD 1.71–3.6 V with IO_VDD_HSLV = 0; not cell programming voltage.IO_VDD_HSLV = 0 時 VDD 1.71–3.6 V;不是單元編程電壓。 DS13737 Rev 10 · Table 32 · p. 156
Temperature溫度
TA conditions in each row; select the matching device temperature grade.各列使用 TA 條件;須另選符合條件的元件溫度等級。 DS13737 Rev 10 · Table 87 · p. 224
Test population測試樣本
Unknown未知 · Sample count / lot distribution not stated in these tables.表格未載明樣本數/批次分布。
TI MSP430FR5994Embedded FRAMConditions and unknowns條件與未知項

256 KB main memory; unprotected FRAM accesses.256 KB 主記憶體;未受保護的 FRAM 存取。

Process node製程節點
Unknown未知 · Not stated in this datasheet.此規格書未載明。
Macro / silicon revision巨集/晶片修訂版
Unknown未知 · Macro / silicon revision not identified by this table.表格未指定巨集/晶片修訂版。
Capacity容量
256 KB main + 0.5 KB information FRAM.256 KB 主區+0.5 KB 資訊區 FRAM。 SLASE54D, Revision D · Table 6-1 · p. 7
Interface / granularity介面/粒度
Memory-mapped SRAM-style accesses; no erase sequence.記憶體映射、類 SRAM 存取;不需抹除序列。 SLASE54D, Revision D · §8.12.12, note 2 · p. 64
Device supply元件供電
VCC 1.8–3.6 V; SVS limits the minimum. Not cell programming voltage.VCC 1.8–3.6 V;最低值受 SVS 限制。不是單元編程電壓。 SLASE54D, Revision D · §8.3, note 6 · p. 30
Temperature溫度
TA and TJ −40 to 85 °C; retention specifically uses TJ.TA 與 TJ −40 至 85 °C;保持時間明確採 TJ。 SLASE54D, Revision D · §8.3 · p. 30 SLASE54D, Revision D · §8.12.12.1 · p. 64
Test population測試樣本
Unknown未知 · Sample count / production test coverage not stated in this table.表格未載明樣本數/量產測試涵蓋率。

On narrow screens, scroll the table horizontally. Keyboard: focus the table region and use the arrow keys.窄螢幕可橫向捲動表格。鍵盤操作:將焦點移至表格區域後使用方向鍵。

Three metrics with their original scope and source locators三項指標、原始條件與來源定位
Metric指標ST STM32U575xxEmbedded FlashTI MSP430FR5994Embedded FRAM
Endurance耐久次數10,000 cycles · minimum10,000 次循環 · 最小值

Whole bank; TA −40 to 125 °C.整個記憶庫;TA −40 至 125 °C。

Characterization; not production-tested. The separate 100k specification is limited to 256 KB per bank.特性評估值,非量產逐顆測試。另列的 100k 規格限每個記憶庫 256 KB。

DS13737 Rev 10 · Table 87, note 1 · p. 224
10¹⁵ read/write cycles · minimum10¹⁵ 次讀寫循環 · 最小值

Recommended supply and operating temperature ranges.建議供電與工作溫度範圍。

Datasheet minimum; read/write cycles are not Flash program/erase cycles.規格書最小值;讀寫循環不等於 Flash 編程/抹除循環。

SLASE54D, Revision D · §8.12.12.1 · p. 64
Data retention資料保持15 years · minimum15 年 · 最小值

TA = 85 °C after 10,000 cycles, whole bank; cycling across the temperature range.整個記憶庫於 10,000 次循環後、TA = 85 °C;循環涵蓋整個溫度範圍。

Characterization; not production-tested.特性評估值,非量產逐顆測試。

DS13737 Rev 10 · Table 87, notes 1–2 · p. 224
10 years · minimum10 年 · 最小值

TJ = 85 °C; prior cycle history not stated in the retention row.TJ = 85 °C;保持時間列未載明先前循環次數。

Datasheet minimum; do not equate TJ with ST's TA condition.規格書最小值;不可把 TJ 等同 ST 的 TA 條件。

SLASE54D, Revision D · §8.12.12.1 · p. 64
Write / program time寫入/編程時間118 µs / 128 bits · typ and max118 µs/128 bits · 典型與最大值

Normal mode; typ at TA = 25 °C, VDD = VDDA = 3 V. Excludes erase.正常模式;典型條件 TA = 25 °C、VDD = VDDA = 3 V。不含抹除。

Design-specified; max characterized after cycling. Not production-tested.由設計訂定;最大值經循環後特性評估。非量產逐顆測試。

DS13737 Rev 10 · Table 86, pp. 223–224, notes 1–2 · p. 223 DS13737 Rev 10 · §5.1.2 · p. 150
tWRITE = tREAD = 1 / fSYSTEMtWRITE = tREAD = 1/fSYSTEM

NWAITSx = 0, fSYSTEM ≤ 8 MHz; word-access relation, not an application throughput measurement.NWAITSx = 0、fSYSTEM ≤ 8 MHz;字組存取關係式,非應用吞吐量量測。

Vendor timing relation, not a typ/max benchmark. Recommended supply and temperature ranges.原廠時間關係式,非典型/最大效能基準。採建議供電與溫度範圍。

SLASE54D, Revision D · §8.12.12.1, notes 3–4 · p. 64 SLASE54D, Revision D · §8.3 · p. 30 SLASE54D, Revision D · Features: word-write timing · p. 1

Before a Design Decision設計決策前仍需確認

Ask for the exact ordering code / macro revision, mission temperature and cycle history, supply margins, ECC and power-loss behavior. TA is ambient temperature; TJ is junction temperature. These cases do not establish area, cost, security, qualification or superiority of a memory family.確認完整訂購代碼/巨集版本、任務溫度與循環歷史、供電裕量、ECC 及掉電行為。TA 為環境溫度,TJ 為接面溫度。這些案例不證明記憶體家族的面積、成本、安全、資格驗證或優勢。

Continue to the decision whitepaper繼續閱讀決策白皮書

Sources and Reuse來源與引用

Reviewed查核日期 . Factual summaries with source links; no source PDF, figure or table is republished. Source documents and marks remain with their owners; public access is not a redistribution license. Verify the cited revision and current errata before use.本區僅整理事實並連結來源,未重製原始 PDF、圖片或表格。原文件與商標權利屬各權利人;公開可讀不等於授權散布。使用前請核對引用版本及最新勘誤。

Download this comparison with conditions and sources (JSON)下載含條件與來源的本區對照資料(JSON)
01C · IP 架構師選型全景決策總表01C · IP ARCHITECTURAL SELECTION MASTER MATRIX

從單元物理到 SoC 系統整合:全譜系 NVM IP 實務決策矩陣From Cell Physics to SoC Integration: Comprehensive NVM IP Decision Matrix

對齊學術技術指標(S. Yu 2016/2021),升維轉化為晶圓成本、電源軌契約、測試時間、車規溫度與先進節點微縮之十四大 IP 整合考量點Elevating Academic Technology Metrics (S. Yu 2016/2021) into 14 System-Level IP Criteria Across Mask Cost, Power Rails, ATE Test Time, Automotive Thermal Profile, and Advanced Node Scaling

🏛️ 學術對比升維至 SoC IP 實務Academic Comparison Upgraded to SoC IP Practice
參考基準:Prof. Shimeng Yu (IEEE SSCM 2016 邀請綜論) 元件級指標 ➔ 升維為 14 大系統級工程整合維度 Baseline: Prof. Shimeng Yu (IEEE SSCM 2016 Review) Device Metrics ➔ Upgraded to 14 System-Level Integration Dimensions

在學術文獻中,NVM 比較通常聚焦於「位元胞物理極限」(如 Cell Area F²、Write Time、Retention 年限)。然而在真實 SoC 架構選型中,單元面積往往被周邊 High-Voltage 開關、感測放大器與電荷幫浦等巨集周邊稀釋;更關鍵的決策瓶頸在於:晶圓代工光罩加價率(Mask Adders)、ATE 晶圓測試時間機台成本、AEC-Q100 高溫活化能退化,以及能否在 2nm GAA/BSPDN 節點無縫可攜。本總表以頂級 SoC 架構師之實務視角,完整解構 11 大 NVM 技術之工程權衡。Academic literature often evaluates NVM based purely on bit-cell physics limits (such as Cell Area in F², Write Time, and Nominal Retention). However, in real-world SoC architecture selection, raw cell area is heavily diluted by macro peripherals including high-voltage switches, sense amplifiers, and on-chip charge pumps. Crucial commercial bottlenecks center on foundry mask adders, ATE wafer test time costs, high-temperature activation energy degradation under AEC-Q100, and seamless portability to 2nm GAA/BSPDN nodes. This master matrix deconstructs the architectural trade-offs across 11 NVM technologies from a senior SoC architect perspective.

💡 提示:表格支援水平滾動與按鈕篩選。點選上方視圖可快速聚焦特定技術族群。所有數據均基於量產代工廠 PDK、JEDEC / AEC-Q100 測試規範與量產實測。Tip: Table supports horizontal scrolling and view filtering. Click the views above to isolate specific technology groups. All metrics are grounded in production foundry PDKs, JEDEC / AEC-Q100 standards, and verified silicon.

全譜系 11 大嵌入式非揮發性記憶體 IP 架構決策矩陣(涵蓋光罩成本、周邊開銷、供電幫浦、測試時間與先進節點微縮)Full-Spectrum 11 Embedded NVM IP Architectural Decision Matrix (Covering Mask Cost, Macro Overhead, Charge Pumps, ATE Test Time, and Advanced Scaling)
系統級評估維度System-Level Dimension 學術對照與架構權衡Academic Context &amp; Tradeoff
晶圓廠 eFuseFoundry eFuse
Foundry eFuse
TSMC eFuse GF FDX eFuse DesignWare®
光罩 ROMMask ROM
Mask ROM
XtraROM® Artisan® ROM TSMC ROM
邏輯反熔絲 OTPLogic AntiFuse OTP
Logic AntiFuse OTP
NeoFuse® XPM® 1T-Fuse™ I-fuse™ PUFiot®
浮閘邏輯 OTPLogic FG-OTP
Logic FG-OTP
NeoBit® Faraday OTP
低干擾邏輯 MTPLogic LD-MTP
Logic LD-MTP
NeoEE® AEON® LogicFlash™ TwinBit™
高密度邏輯 MTPLogic HD-MTP
Logic HD-MTP
NeoMTP® YMC MTP® LEE Flash ZT™
分離閘快閃 (SST)Split-Gate eFlash (SST)
Split-Gate eFlash (SST)
SuperFlash® (ESF3) SG-MONOS™ pFusion®
電荷捕捉快閃 (SONOS)Charge-Trap eFlash (SONOS)
Charge-Trap eFlash (SONOS)
SONOS eFlash MONOS™ LEE Flash G2™ Y-Flash™
嵌入式 STT-MRAMEmbedded STT-MRAM
Embedded STT-MRAM
TSMC eMRAM AutoPro150™ Toggle MRAM® NuPro™
阻變記憶體 (ReRAM)Embedded OxRAM (ReRAM)
Embedded OxRAM (ReRAM)
TSMC 22RRAM 22FDX+ OxRAM™ Weebit SiOx CBRAM®
鐵電記憶體 (FeRAM/FeFET)Embedded FeRAM / FeFET
Embedded FeRAM / FeFET
EXCELON™ F-RAM MSP430 FRAM 22FDX FeFET
01
微影光罩道數與晶圓加價率Mask Adders & Wafer Cost Impact
學術指標對照:Academic Baseline: 原圖無對應(原圖僅看 Cell F²,忽略製程光罩成本是 SoC 決定生死的第一道關卡)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 0-Mask 完全相容標準邏輯製程(晶圓零加價);eFlash 增加 8~12 道光罩導致晶圓加價 25~40%,小晶粒極不經濟;BEOL 新興技術(ReRAM/MRAM)增加 2~5 道後段光罩,增加 8~15% 晶圓成本。0-Mask logic CMOS compatibility eliminates wafer cost adder (0%); eFlash requires 8-12 front-end masks (+25~40% wafer adder), highly uneconomic on small dies; BEOL emerging NVMs (ReRAM/MRAM) add 2-5 back-end masks (+8~18% wafer cost).
+0 道(純邏輯)+0 (Standard Logic)
標準代工庫原生整合,晶圓 0 加價Foundry native baseline, 0% wafer adder
0 ~ +1 道0 ~ +1 Mask
金屬層或 Contact/Via 改動,極低光罩費用Contact/via programming layer only
+0 道(純邏輯)+0 (Pure Logic CMOS)
100% 邏輯相容,零額外光罩,小晶粒總成本最低100% logic compatible, 0 adder, lowest TCO on small dies
+0 道(純邏輯)+0 (Standard Logic)
單層多晶矽邏輯相容,晶圓零加價Single-poly standard logic, 0 adder
通常 +0 道Typically +0
單層多晶矽 FN 穿隧,零光罩至 +1 道注入Single-poly FN, 0-mask or optional +1 well implant
+0 ~ +2 道+0 ~ +2 Masks
依供應商與 PDK 分流(基線 +0,高密度版 +2 道井注入)Vendor-split: baseline +0, Gen2 +2 well-implants
+8 ~ +12 道(FEOL)+8 ~ +12 (FEOL)
前段專用高溫/浮閘光罩,晶圓加價 25%~40%,28nm 以下停滯Front-end dedicated masks, +25~40% cost, stalls <28nm
+2 ~ +4 道(FEOL)+2 ~ +4 (FEOL)
O-N-O 捕捉層模組,晶圓加價 10%~18%O-N-O trapping module, +10~18% wafer cost adder
+3 ~ +5 道(BEOL)+3 ~ +5 (BEOL)
後段金屬層 MTJ 模組,晶圓加價 12%~18%Back-end MTJ stack, +12~18% wafer cost adder
+2 ~ +4 道(BEOL)+2 ~ +4 (BEOL)
後段金屬層過渡金屬氧化物模組,晶圓加價 6%~10%Back-end transition-metal oxide, +6~10% wafer adder
+2 ~ +4 道(BEOL/FEOL)+2 ~ +4 (BEOL/FEOL)
鐵電電容或 HfO2 薄膜,晶圓加價 8%~12%Ferroelectric cap or HfO2 gate, +8~12% wafer adder
02
巨集實體密度與周邊電路開銷Macro-Level Density & Silicon Overhead
學術指標對照:Academic Baseline: 對應原圖 Cell Area (F²),升維至包含電荷泵、感測放大器與高壓隔離間距的巨集實體面積Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 小容量(1Kb~64Kb)下,高壓幫浦與周邊開銷佔巨集 80% 面積,Cell F² 影響極小;大容量(>4Mb)下,Cell F² 才是矽面積主導因素。At small capacities (1Kb-64Kb), high-voltage pumps and peripheral circuits consume 80%+ macro area, making cell F² secondary; at >4Mb, array cell F² dominates total silicon area.
極低密度(周邊極大)Very Low Density
60–120 F²;大電流金屬線路與感測電路使小容量巨集龐大60–120 F²; heavy wiring & drivers dominate small macros
最高密度(極緊湊)Ultra-High Density
<8 F²;無高壓幫浦,周邊極簡,矽面積極小<8 F²; no pumps, minimal peripheral, lowest area
Core 極緊湊(1T 專利對半縮減)Core Ultra-Compact (1T Halved)
10–25 F²;Core 閘氧原生微縮;1T 專利陣列較 2T 減半*10–25 F²; Core GOX native scaling; 1T array halved vs 2T*
大(I/O 級佔位)Large (I/O Class)
100–250+ F²;侷限 5V/3.3V 厚氧,為 1T 的 10×–25× 面積*100–250+ F²; 5V/3.3V thick oxide bound; 10×–25× area of 1T*
中低密度(電容開銷大)Low-Medium Density
60–120+ F²;單層多晶矽需 >80% 面積作為控制電容60–120+ F²; >80% area dedicated to coupling capacitors
中高密度(25–50 F²)Medium-High Density
25–50 F²;熱電洞注入縮小耦合電容,適合 16Kb~512Kb25–50 F²; hot-carrier injection shrinks capacitor size
大容量高密度High Density for >1Mb
12–30 F²;專用 FEOL 單元緊湊,但有巨大高壓泵周邊12–30 F²; tight cell, but massive HV charge pump overhead
大容量高密度High Density for >1Mb
15–35 F²;電荷捕捉層單元緊湊,適合 1Mb~32Mb15–35 F²; charge trap cell, optimal for 1Mb~32Mb
先進節點超高密度Ultra-High in Adv. Nodes
30–50 F²;BEOL 垂直堆疊,不佔用邏輯 FEOL 面積30–50 F²; BEOL integration saves valuable logic area
先進節點高密度High Density in Adv. Nodes
15–35 F²;交叉點陣列 (Crossbar) 極具 3D 堆疊潛力15–35 F²; crossbar structure enables multi-layer 3D stacking
中高密度(6–50 F²)Medium-High Density
6–50 F²;1T1C 電容面積受限,FeFET 具 1T 高密度6–50 F²; 1T1C capacitor limited, FeFET enables 1T scaling
03
電源軌契約與晶片內建高壓幫浦Supply Rail Contract & On-Chip Charge Pump
學術指標對照:Academic Baseline: 對應原圖 Voltage (<1V, >10V, <3V...),升維至單電源 Core VDD vs 外接 VPP 針腳、峰值電流與 IR-DropCorresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 是否需要專用外接高壓針腳(增加封裝 Pin 腳與板級成本)?內建 Charge Pump 的開機升壓時間(Ramp-up)與瞬態突波電流(Inrush Current)對低功耗系統的衝擊。Assesses whether external high-voltage VPP pins are required (increasing package cost) versus integrated on-chip charge pumps, inrush current surges, and pump ramp-up latency.
需高電流源(~10-15mA)High Current (~10-15mA)
通常需專用 1.8V/2.5V 電源軌或外接高電流 PadRequires dedicated high-current supply rail or pad
純單一核心 VDDPure Core VDD Only
0.75V~1.2V 直接運作,無任何高壓或特殊電源需求0.75V~1.2V core rail, zero HV or analog pump needed
單一供電,內建微型幫浦Single Supply, Small Pump
核心 VDD 供電,內建幫浦升至 5~7V 擊穿,微安級瞬態Internal pump from core VDD to 5~7V, µA-level transient
單一供電或雙電源Single or Dual Supply
內建 6~8V 幫浦,熱載子注入需要毫安級瞬間電流Internal 6~8V pump, mA-level current for CHEI
單一供電,內建 FN 幫浦Single Supply, FN Pump
內建 10~14V 幫浦,純電壓穿隧,電流極低(<1µA)Internal 10~14V pump, pure FN tunneling, ultra-low current
單一供電,內建幫浦Single Supply, Internal Pump
內建 8~12V 幫浦,寫入電流依機制(FN 微安級 vs CHI 毫安級)Internal 8~12V pump, current depends on FN vs CHI mode
大型雙重內建高壓幫浦Large Dual HV Pumps
寫入需 10V+5V,抹除需 12~15V,升壓延遲 10~20µsProg 10V+5V, Erase 12~15V, pump ramp-up latency 10~20µs
內建高壓幫浦Internal HV Charge Pump
FN 寫抹需 9~12V 高壓,讀取直接使用核心 VDDFN P/E requires 9~12V, direct core VDD for sensing
純單一核心低電壓Pure Single Low-Voltage
~1.0-1.2V 直接翻轉,完全無高壓幫浦,但寫入需 30-80µA 脈衝~1.0-1.2V direct switching, zero pump, 30-80µA write pulse
低壓輔助幫浦(~2.5V)Low-V Boost Pump (~2.5V)
成形需 2.5V,SET/RESET 僅需 1.2-1.5V,幫浦極小Forming ~2.5V, SET/RESET 1.2-1.5V, minimal pump area
純單一核心低電壓Pure Single Low-Voltage
1.2V~1.8V 直接電場翻轉,完全無高壓幫浦1.2V~1.8V direct electric field switching, zero pump
04
讀取延遲、直讀時脈與就地執行Read Latency, XIP Clock & Bus Execution
學術指標對照:Academic Baseline: 對應原圖 Read time (~1ns, ~10ns, ~50ns...),升維至系統匯流排直讀頻率與是否強制 Shadow 到 SRAMCorresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否直接透過 AHB/AXI 匯流排執行程式碼(Execute-In-Place / XIP)?還是必須在開機時額外耗時複製(Shadowing)到系統 SRAM?Whether code can execute directly from the array over AHB/AXI buses (Execute-In-Place / XIP) or strictly mandates power-up boot shadowing into system SRAM.
慢速,強制 ShadowSlow (~1µs), Shadow Req.
感測延遲 ~1µs,必須在開機時整組 Shadow 到 SRAM 暫存器Access ~1µs, strictly requires boot shadowing to SRAM
超高速直讀(~5-15ns)Ultra-Fast XIP (~5-15ns)
直接靜態 1T 陣列,支援 100MHz+ 匯流排無等待直讀Static 1T array, supports 100MHz+ zero-wait XIP execution
中速直讀(~25-40ns)Direct XIP (~25-40ns)
受單元導通電流限制,支援 20~40MHz 直讀或開機載入Cell-current limited, supports 20~40MHz XIP or boot-load
中速直讀(~25-40ns)Direct XIP (~25-40ns)
感測放大器翻轉受限,可支援直讀或開機快取Sense amplifier trip limited, direct read or cacheable
中速直讀(~20-40ns)Direct Read (~20-40ns)
單元讀取電流適中,支援小區塊直讀或暫存器載入Moderate read current, supports direct read or shadow
中速直讀(~20-35ns)Direct Read (~20-35ns)
支援 30~50MHz 匯流排直讀,多用於參數與常數讀取Supports 30~50MHz direct read for calibration & constants
高速直讀(~15-25ns)High-Speed XIP (~15-25ns)
分離閘極讀取電流大,支援 80~133MHz 零等待 XIP 執行Split-gate high current, 80~133MHz zero-wait XIP MCU code
中高速直讀(~20-35ns)Direct XIP (~20-35ns)
支援 50~80MHz 匯流排 XIP 直讀執行Supports 50~80MHz bus XIP execution
類 SRAM 級(~10-25ns)SRAM-like XIP (~10-25ns)
TMR 感測快速,支援 100MHz+ 高速無等待隨機 XIPFast TMR sensing, 100MHz+ zero-wait random XIP execution
高速直讀(~10-20ns)Fast Direct Read (~10-20ns)
低阻態讀取電流充裕,支援高速隨機讀取Low-resistance state provides ample sensing margin
極高速破壞性讀取(~20-40ns)Destructive Read (~20-40ns)
感測需翻轉極化並立即預充電寫回(Auto-Restore)Polarization charge sensing requires automatic write-back
05
寫入粒度與量產燒錄測試成本Programming Granularity & ATE Test Cost
學術指標對照:Academic Baseline: 對應原圖 Write time (1ns~1ms),升維至 Bit/Byte/Sector 粒度與量產 ATE 機台測試燒錄費用Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否進行單一位元/位元組隨機更新?在晶圓測試 (CP) 與最終成品測試 (FT) 階段,燒錄金鑰與韌體需要多少機台秒數?(每增加 1 秒量產百萬顆即是巨額成本)。Single bit/byte random programming versus block erase; wafer-level CP and package-level FT burn-in time per chip on multi-million dollar ATE testers directly dictates production unit margins.
位元級寫入,測試極耗時Bit-level, High Test Cost
每個位元需 10~15mA 脈衝數毫秒,高容量燒錄造成 ATE 測試瓶頸ms-level pulse per bit, heavy ATE machine-time penalty
出廠光罩固化,零測試時間Zero Test Time (Factory)
生產即有資料,僅需常規邏輯測試,零燒錄時間成本Manufactured with data, zero programming test time
位元級寫入,毫秒級完成Bit-level, Fast CP/FT Test
單一位元崩潰僅數十微秒,1Kb~4Kb 金鑰在 CP 階段數毫秒內完成µs breakdown per bit; 1-4Kb keys programmed in <10ms
位元/位元組級,測試速度中等Bit/Byte, Moderate Test
熱電子注入需脈衝驗證循環,測試時間稍長於 AntiFuseCHEI requires pulse-verify loop; slightly longer than AntiFuse
位元組寫入/整頁抹除Byte Write / Page Erase
FN 穿隧抹寫需數毫秒,適合少量校準參數更新FN P/E requires ms-level pulses; optimal for trim updates
位元組寫入/區塊抹除Byte Write / Sector Erase
支援隨機位元組更新,整體量產測試時間受控Supports random byte update; controlled ATE test time
位元組寫入/扇區抹除Byte Write / Sector Erase
SSI 寫入快速(~10µs/byte),但 Sector 抹除需 5~10ms,整體 FT 燒錄需數秒Fast SSI (~10µs/B), but sector erase 5-10ms; multi-sec FT
頁面寫入/扇區抹除Page Write / Sector Erase
整頁寫入 ~1-2ms,扇區抹除 ~10ms,測試吞吐量居中Page write ~1-2ms, sector erase ~10ms; medium test speed
位元組/字組隨機寫,免抹除Byte/Word In-Place, No Erase
單元翻轉 <20ns,免除所有抹除步驟,量產 ATE 燒錄最快(毫秒級)<20ns flip, zero erase required; fastest ATE burn-in
位元組隨機寫入,需成形Byte In-Place, Needs Forming
免抹除,但出廠首次需要 Forming 步驟(增加測試工程流程)No erase needed, but requires factory forming step
超高速隨機位元組寫入Ultra-Fast Byte In-Place
寫入時間等同讀取(<50ns),無抹除延遲,測試極快Write speed equals read (<50ns), zero erase overhead
06
高溫資料保持與車規 Mission ProfileHigh-Temp Retention & AEC-Q100 Profile
學術指標對照:Academic Baseline: 對應原圖 Retention (>10y, >1y...),升維至 Arrhenius 活化能 Ea 與 AEC-Q100 Grade 0/1/2 認證Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否承受引擎室、車載底盤或工業高溫環境?AntiFuse 的金屬-矽微絲具極高活化能;浮閘/電荷捕捉受限於高溫熱去捕捉;MRAM 受限於高溫熱擾動退磁。Arrhenius thermal acceleration activation energy (Ea); AntiFuse ohmic filaments exhibit high Ea (>1.8eV) surviving 175°C Grade 0+, while charge-trap and magnetic cells face thermal leakage at extreme junctions.
125°C 10 年(車規 Grade 1)125°C 10yr (Auto Grade 1)
物理吹斷,高溫無洩漏,但存在極微小金屬再長回風險Physical rupture, but micro-risk of metal electromigration heal
永久保持(全溫區 Grade 0)Permanent (Auto Grade 0)
實體光罩接線,不隨溫度或時間衰退,無極限問題Hardwired mask layers, zero thermal degradation over time
175°C 10 年(車規 Grade 0+)175°C 10yr (Auto Grade 0+)
Ea ~1.8-2.1 eV,永久導通,耐受引擎室極端高溫Ea ~1.8-2.1 eV, irreversible filament, AEC-Q100 Grade 0+
125°C 10 年(車規 Grade 1)125°C 10yr (Auto Grade 1)
Ea ~1.1 eV,高溫氧化層洩漏限制在 Grade 1 範疇Ea ~1.1 eV, oxide thermal leakage bounds to Grade 1
125°C 10 年(工業/車規 Grade 1)125°C 10yr (Grade 1)
低干擾單層浮閘,125°C 保持良好,高溫退火需校準Low-disturb single poly, solid at 125°C, bounds Grade 1
125°C 10 年(工業/車規 Grade 1)125°C 10yr (Grade 1)
薄氧化層電荷保持,標準為 125°C 10 年Thin-oxide charge storage, standardized at 125°C 10yr
150°C 10 年 / 125°C 20 年150°C 10yr / 125°C 20yr
深厚氧化層隔離,車規 Grade 0 標竿量產架構Thick oxide isolation, golden standard for Auto Grade 0
125°C 10 年(部分 Grade 0 驗證)125°C 10yr (Some Grade 0)
氮化矽捕捉層高溫擴散限制,先進世代達成 150°C 保持SiN traps can diffuse; newer gens qualified to 150°C
125°C 10 年(高溫磁擾動挑戰)125°C 10yr (Grade 1 typical)
熱穩定因數 Δ 隨溫度急遽下降,150°C 需犧牲寫入電流或容量Thermal factor Δ drops with temp; 150°C requires larger Ic
85°C ~ 105°C(工業級)85°C ~ 105°C (Industrial)
氧空缺在高溫下自發擴散,高溫車規 Grade 0 仍屬研發邊界Oxygen vacancies diffuse at high temp; Grade 0 remains R&D
85°C 10 年 / 125°C 受限85°C 10yr / 125°C limited
居里溫度 (Curie Temp) 限制,高溫去極化需嚴格控管Curie temperature bound, thermal depolarization at high temp
07
SMT 迴焊爐耐溫與預燒錄出貨能力SMT Reflow Tolerance & Pre-Programming
學術指標對照:Academic Baseline: 原圖無對應(原圖未考慮封裝打件 260°C 熱歷程對金鑰與韌體的毀滅性風險)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否在晶圓或裸晶階段預先燒錄安全金鑰與 Bootloader,再送入 260°C 無鉛迴焊爐打件?還是必須在打件後透過 JTAG 板級燒錄?Ability to withstand JEDEC J-STD-020 lead-free 260°C reflow heat without data corruption, enabling pre-programmed wafer/die delivery versus requiring post-assembly board-level programming.
100% 免疫迴焊熱衝擊100% Immune to Reflow
金屬或矽化物實體開路,耐受 260°C 迴焊,支援晶圓預燒Physical open-circuit, 100% safe through 260°C reflow
100% 絕對免疫100% Absolute Immune
金屬佈線層,耐受任何封裝與迴焊熱處理Hardwired interconnect, immune to all assembly heat
100% 免疫迴焊熱衝擊100% Immune to Reflow
歐姆導通微絲極端穩定,支援晶圓級預先燒錄安全金鑰Ohmic filament is ultra-stable; perfect for pre-program keys
具熱漏失風險,建議後燒Thermal Risk, Post-Program
260°C 烘烤可能導致微量熱電子逸出,降低感測裕度260°C reflow may bleed electrons; post-SMT burn advised
建議 SMT 打件後燒錄Post-SMT Recommended
迴焊熱應力影響浮閘電荷,量產建議在 SMT 後進行校準燒錄Thermal stress impacts charge; calibrate post-SMT
建議 SMT 打件後燒錄Post-SMT Recommended
薄氧化層電荷對 260°C 敏感,打件後燒錄最具可靠性Thin-oxide sensitive to 260°C; post-assembly preferred
厚氧化層耐受性佳,可預燒High Tolerance, Pre-Prog OK
ESF3 厚熱氧化層在 260°C 表現穩定,但車規常規仍採後燒Thick oxide survives 260°C reflow; qualified for pre-prog
中度耐受,需嚴密熱預算Moderate, Budget Monitored
氮化矽深能階捕捉電荷可耐受短暫迴焊,需注意烘烤次數Deep traps survive brief reflow, track cumulative cycles
極敏感(磁退火風險)Sensitive (Demag Risk)
260°C 高溫易誘發 MTJ 自由層熱退磁與擴散,需特殊高阻抗防護260°C heat risks MTJ demagnetization; requires special stack
成形前打件最優Pre-Forming Assembly Best
未成形狀態下耐溫良好;若已燒錄導電細絲在高溫有自癒風險Pristine state survives well; pre-formed filament risks diffusion
去極化風險,需重新極化Depolarization Risk
接近居里點時極化電荷衰退,打件後需重新寫入確認Polarization degrades near Curie point; re-program post-SMT
08
覆寫壽命與系統磨損管理負擔Endurance Lifecycle & Wear-Leveling
學術指標對照:Academic Baseline: 對應原圖 Endurance (1E5, 1E6-1E9, 1E14...),升維至 SoC 韌體磨損平衡 (Wear-Leveling) 與壞區管理負擔Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否直接取代 SRAM/BRAM?若是 eFlash/ReRAM,SoC 是否必須耗費寶貴的 CPU 與記憶體執行耗損平均(Wear-Leveling)演算法與垃圾回收?From 1-time OTP to 10^5 eFlash and 10^12 MRAM; dictates whether the host SoC CPU and RTOS must implement complex wear-leveling algorithms, bad-block management, and journaled filesystems.
1 次(OTP,不可逆)1 Cycle (OTP, Irrev.)
吹斷後物理結構永久改變,不可抹寫Irreversible physical link blown, strictly 1-time
0 次(唯讀)0 Cycles (Read-Only)
光罩固定,完全不可寫入或更改Mask pattern fixed, zero in-field write capability
1 次(可採虛擬多 Bank 更新)1 Cycle (Multi-Bank Emul.)
物理 1 次,但架構可劃分多個區塊透過指標實現數次版本更新Physical 1-cycle; logical multi-bank allows version updates
1 次(部分支援紫外線清除)1 Cycle (UV EPROM mode)
封裝後為 1 次;開模紫外線照射理論可清,量產不實用1-time in package; UV clear theoretical, impractical in mass
10⁴ ~ 10⁵ 次抹寫10⁴ ~ 10⁵ P/E Cycles
低干擾單層多晶矽,足以支援產品生命週期內所有參數校準Low-disturb FN, ample for lifelong trim and config updates
10³ ~ 10⁴ 次抹寫10³ ~ 10⁴ P/E Cycles
氧化層較薄,適合百次至千次級別之代碼補丁與日誌Thinner oxide limits cycles; ideal for code patches & logs
10⁴ ~ 10⁵ 次抹寫10⁴ ~ 10⁵ P/E Cycles
MCU 標準韌體儲存,需配合 Flash 檔案系統進行區塊管理MCU code standard; requires wear-leveling filesystem
10⁴ ~ 10⁵ 次抹寫10⁴ ~ 10⁵ P/E Cycles
捕捉層介電質疲勞磨損,需配合擦寫計數與 ECC 校驗Trapping layer fatigue; requires wear-counting & ECC
10¹⁰ ~ 10¹² 次(近無限)10¹⁰ ~ 10¹² (Near-Infinite)
自旋翻轉無物理磨損,可直接作為 Working RAM,零磨損演算法負擔Spin flip has no wearout; direct SRAM replacement, 0 software
10⁵ ~ 10⁶ 次抹寫10⁵ ~ 10⁶ P/E Cycles
細絲反覆生長斷裂存在累積缺陷,需適度磨損均衡機制Filament fatigue accumulates defects; requires wear leveling
10¹² ~ 10¹⁴ 次(極高耐久)10¹² ~ 10¹⁴ Cycles
極化翻轉疲勞極低,可直接取代 SRAM 進行無限制即時日誌記錄Negligible fatigue; directly replaces battery-backed SRAM
09
動態功耗與常關閉即時開機能力Active Energy & Normally-Off Instant-On
學術指標對照:Academic Baseline: 對應原圖 Write Energy (fJ~100pJ),升維至系統級動態電流、常關閉(Normally-Off)待機零漏電與開機能量Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 在電池供電的邊緣 IoT 或車用電子中,能否在待機時徹底切斷電源(Zero Standby Leakage)並在微秒內即時喚醒(Instant-On)?System dynamic current during read/write operations versus zero standby power-gated leakage (Normally-Off / Instant-On) for battery-operated edge IoT and AI sensors.
讀取低功耗,寫入高功耗Low Read / High Write Pwr
寫入需 10~15mA 峰值電流,電池供電裝置幾乎無法在線燒錄10-15mA write current makes in-field battery programming hard
極低動態功耗,零待機漏電Ultra-Low Active / 0 Leak
僅靜態 1T 邏輯切換,斷電無漏電,開機 0 延遲Static 1T switching, zero power-off leakage, 0-latency wake
極低功耗,完美常關閉Ultra-Low / Perfect N-Off
讀取 <0.5mW,斷電零漏電,喚醒僅數十納秒,最佳 IoT 金鑰Read <0.5mW, zero leakage, ns-level wakeup; ideal for IoT RoT
低動態功耗,零待機漏電Low Active / 0 Leakage
斷電完全不耗電,開機即讀,適合感測器配置Zero power-off leakage, instant read for sensor configs
純電壓寫入,低功耗Pure Voltage FN, Low Power
FN 寫抹電流 <1µA,電池供電裝置可輕易完成參數更新FN P/E current <1µA, battery-powered friendly for calibration
低至中功耗Low to Moderate Power
寫抹功耗適中,支援斷電零漏電與快速喚醒Controlled P/E power, zero standby leakage, fast wake
寫抹高功耗(高壓泵突波)High Write Pwr (HV Surge)
高壓電荷泵瞬態電流 15~30mA,需較大電源去耦電容HV pump surge 15~30mA, requires hefty decoupling caps
中等寫抹功耗Moderate Write Power
電荷捕捉層寫抹效率高於傳統浮閘,支援低功耗 MCUHigher efficiency than floating gate, supports low-power MCUs
寫入電流大,讀取極低功耗High Write I / Low Read Pwr
讀取如 SRAM,但 STT 翻轉需 30~80µA/bit;斷電完全零漏電SRAM-like read, but STT write requires 30-80µA; 0 standby leak
超低讀寫能耗(~pJ 級)Ultra-Low Energy (~pJ)
SET/RESET 脈衝極短(<50ns),是邊緣 AI 類腦運算首選Sub-50ns switching pulses, premier choice for Edge AI CiM
極致超低功耗(~100fJ 級)Extreme Low Power (~100fJ)
電場極化翻轉幾乎不消耗電荷,業界寫入能耗最低架構Electric-field displacement current only; lowest write energy
10
製程微縮可攜性與先進架構支援Process Scaling & Advanced Node Support
學術指標對照:Academic Baseline: 原圖無對應(原圖未分析 FinFET、GAA 奈米片與 BSPDN 背側供電的實體整合邊界)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 能否跟隨主晶片微縮至 16nm/7nm/3nm/2nm GAA?eFlash 在 28nm 平面停滯;AntiFuse 可原生縮放至 2nm;MRAM/ReRAM 在 BEOL 低溫堆疊不受前段限制。Portability to FinFET, GAA nanosheets, and backside power delivery (BSPDN); eFlash strictly stalls at planar 28nm, while AntiFuse scales natively to 2nm GAA, and MRAM/ReRAM stack in low-temp BEOL.
可移植至先進節點,面積大Ports to GAA, Huge Area
金屬線路與接觸窗隨節點縮小,但寄生電阻使吹斷越來越困難Contact scales, but parasitic R makes blowing hard in FinFET
可移植,但先進光罩極昂貴Ports, but Costly Masks
先進 EUV 節點光罩費用數百萬美元,極少用於高階 ASICEUV mask cost ($M) makes Mask ROM uneconomic in GAA
一路微縮至 2nm GAA / BSPDNScales to 2nm GAA / BSPDN
利用核心薄閘氧電崩潰,Core 單元原生跟隨核心邏輯微縮,先進節點唯一 0-MaskTracks core GOX breakdown; Core cell natively scales with logic; sole 0-mask NVM at 2nm/BSPDN
平面 28nm 止步(無法 scaling)Stalls at Planar 28nm (No Scaling)
I/O 單元侷限 5V/3.3V 厚氧,無法隨 core 微縮,且浮閘結構與 FinFET 鰭片幾何衝突I/O cell bound to 5V/3.3V thick oxide, cannot scale with core; conflicts with 3D FinFET geometry
平面 28nm 止步Stalls at Planar 28nm
大面積控制電容在 FinFET 難以成形,成熟節點為主Large lateral capacitor impractical in FinFET/GAA
平面 28nm 止步Stalls at Planar 28nm
熱電洞注入依賴平面基板特性,先進 FinFET 難以相容Hot-carrier injection physics incompatible with FinFET fins
平面 28nm 嚴格止步Strict Stop at Planar 28nm
分裂閘三維高溫熱氧化層無法在 FinFET 3D 鰭片上成形Split-gate 3D thermal budget incompatible with FinFET
平面 28nm / 40nm BCDPlanar 28nm / 40nm BCD
成熟車規與 BCD 混合訊號主力,難以經濟跨入 FinFETAutomotive BCD workhorse; difficult to port to FinFET
BEOL 堆疊,跨入 22nm~12nmBEOL Stack, 22nm to 12nm
後段金屬低溫製程(<400°C),與前段邏輯完全解耦,主流替代 eFlashLow-temp BEOL decoupled from FEOL; primary eFlash successor
BEOL 堆疊,可延伸至 12nm+BEOL Stack, 12nm & Beyond
後段兩道金屬層間整合,極具微縮潛力,與 FinFET 相容Simple 2-mask BEOL stack, highly compatible with FinFET
FeFET 具 GAA 縮放潛力FeFET Scales with GAA
傳統 1T1C FeRAM 難微縮;新興 HfO2 FeFET 可直接整合於 GAA 閘極1T1C hard to scale; HfO2 FeFET directly integrates into GAA gate
11
實體層防竄改與硬體信任根 (Root of Trust)Physical Security, Anti-Tamper & PUF RoT
學術指標對照:Academic Baseline: 原圖無對應(原圖完全無安全考量,但 IP 評估中這是晶片防禦逆向工程的核心)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 斷電後 SEM/TEM 顯微鏡下是否可見數據?是否防禦 FIB 電壓對比探針?能否原生衍生物理不可複製功能 (PUF) 晶片指紋?Whether stored data is visible under SEM/TEM post delayering; resistance to FIB voltage-contrast micro-probing, DPA side-channel leakage, and ability to derive native PUF hardware fingerprints.
極差(SEM 顯微鏡立即可見)Very Poor (Visible via SEM)
金屬或多晶矽物理熔斷開口在顯微鏡下立即可讀出,無安全性Physical gap clearly visible under SEM; zero optical security
極差(去層逆向光學可讀)Very Poor (Optically Readable)
接觸窗與通孔圖樣去層後透過光學顯微鏡即可全盤逆向Contact/via patterns easily reverse-engineered via delayering
極致安全(無實體痕跡,PUF 原生)Top Security (Invisible, PUF)
奈米級介電質崩潰微絲,TEM 無可辨特徵,FIB 無法探測,原生 PUFNano-scale filament invisible in TEM; FIB immune; native PUF
中等(存在電荷感測洩漏)Moderate (Charge Emiss. Risk)
可透過微光顯微鏡 (EMMI) 或光學誘導放電分析Vulnerable to emission microscopy or UV-induced discharge
中等(浮閘電荷易被探測)Moderate (Charge Probing)
浮閘儲存之電荷受限於雷射攻擊與電壓對比探針Floating gate charge vulnerable to voltage contrast & laser
中等(需主動屏蔽保護)Moderate (Shielding Req.)
需配合頂層金屬主動屏蔽(Active Metal Shield)以防探針攻擊Requires active metal shield over macro against probing
良好(分離閘極具幾何遮蔽)Good (Split-Gate Shielding)
控制閘極在幾何上覆蓋浮閘,具良好實體遮蔽防禦力Control gate physically overhangs FG, providing physical shield
優良(絕緣捕捉層不聚電荷)Very Good (Localized Traps)
電子分散於氮化矽奈米捕捉位點,探針極難局部讀出Electrons trapped locally in SiN; immune to single-probe drain
優良(防電氣探針,但懼強磁)Very Good (V-probe immune)
TEM 無法分辨磁矩,FIB 無法直接探測,但需防強外部磁場竄改Magnetic state invisible in TEM; requires magnetic shielding
優良(奈米導電絲難以觀測)Very Good (Nano Filament)
氧空缺細絲極細微,常規顯微鏡無特徵,抗逆向工程佳Sub-nm vacancy filament undetectable by standard microscopy
優良(晶格偶極無光學特徵)Very Good (Lattice Dipole)
鐵電晶格自發極化,無宏觀電荷或幾何形貌差異Spontaneous crystal polarization has zero geometric footprint
12
晶圓代工 PDK 生態、成熟度與多源供應Foundry PDK Maturity & Multi-Sourcing
學術指標對照:Academic Baseline: 原圖無對應(學術研究未考量商業代工廠 PDK 支援與跨廠移植風險)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 主流代工廠(TSMC、UMC、GF、Samsung)是否有現成合格 PDK?是否存在單一 IP 供應商鎖定(Vendor Lock-in)?量產出貨顆數與驗證成熟度?Commercial PDK availability and volume production track record across TSMC, UMC, GlobalFoundries, and Samsung; mitigation of single-vendor lock-in risks.
各代工廠原生支援,介面互異Foundry Native, Unique APIs
TSMC/UMC 等廠原生免費提供,但各廠控制介面完全不相容Free native IP, but interfaces completely proprietary per fab
各代工廠原生支援Universal Foundry Support
成熟製程標準選單,極低風險,全球代工廠通用Standard commodity offering across all global foundries
全球量產標竿(數百億顆出貨)Global Benchmark (10B+ dies)
力旺 (eMemory)、Synopsys 全球各大廠 PDK 100% 覆蓋,量產實績第一eMemory & Synopsys ported to TSMC/UMC/GF/SMIC/Samsung; 10B+ shipped
成熟量產,多廠支援Mature, Multi-Fab Support
力旺 NeoBit 累積出貨數百億顆,0.5µm~55nm 成熟代工廠標配eMemory NeoBit benchmark in 0.5µm~55nm, massive volume
力旺/Floadia 等多廠 PDKMulti-Fab PDK Available
TSMC/UMC/Vanguard/SMIC 廣泛採用,成熟混合訊號主力Widely available across TSMC/UMC/VIS/SMIC mixed-signal PDKs
力旺/Synopsys 等量產實績Proven Volume Track Record
NeoMTP / AEON 在主流代工廠 180nm~40nm 具多年量產實績NeoMTP & AEON qualified across 180nm~40nm global PDKs
代工廠專屬授權,專利門檻高Fab License, High IP Barrier
TSMC/UMC/GF 提供,但需支付專利授權金與晶圓提成(Royalty)Available at TSMC/UMC/GF; high license fee and wafer royalty
代工廠與特定 IP 廠提供Foundry & IP Vendor
UMC、GF、Floadia 提供 PDK,多用於特定車規與 BCD 平台Offered on specific automotive & BCD lines at UMC/GF/Floadia
TSMC / Samsung / GF 主力平台TSMC / Samsung / GF Active
TSMC 22/16nm、Samsung 28/14nm、GF 22FDX 商業量產,各廠材料獨立Production at TSMC 22/16, Samsung 28/14, GF 22FDX; non-portable
TSMC / UMC / Weebit 逐漸普及TSMC / UMC / Weebit Growing
TSMC 40/28/22nm、UMC 22nm 陸續推出商業 PDK,處於高速擴張期Commercial PDKs ramping at TSMC 40/28/22 and UMC 22nm
少數代工廠專屬(Infineon/TI/TSMC)Specialized (Infineon/TI/TSMC)
傳統 FeRAM 僅少數 IDM 掌握;HfO2 FeFET 處於大廠先端 R&D 階段Traditional FeRAM IDM-bound; HfO2 FeFET in advanced foundry R&D
14
SoC 系統最佳適配場景與核心定位Target SoC Sweet-Spot Applications
學術指標對照:Academic Baseline: 原圖無對應(學術表未提供終端晶片架構落地的最佳應用場景分類)Corresponds to academic technology metrics, elevated to SoC system integration criteria.
架構權衡實務:Architectural Tradeoff: 是作為晶片身分證?安全密鑰?類比校準?韌體執行碼?還是高速即時日誌?精確選型決定晶片毛利與競爭力。Optimal architectural socket mapping: Security Root-of-Trust, analog trimming, MCU code flash, edge AI weight buffering, or mission-critical flight recorders.
晶圓瑕疵修補、封裝 IDWafer Repair & Package ID
SRAM 備援替換、極低容量晶片識別碼(Chip ID)SRAM redundancy repair, minimal chip serial ID
大量平價 Boot ROMHigh-Volume ASIC Boot ROM
出貨數千萬顆之固定引導碼,零授權金,低成本極致Ultra-high volume fixed bootloaders, zero license cost
安全金鑰、PUF 信任根、類比修補Root-of-Trust, PUF, Trim
Secure Boot 金鑰、AES/RSA/PQC 根金鑰、PMIC 類比精細修補Secure boot root keys, PQC seeds, PMIC precision analog trim
感測器校準、多媒體金鑰Sensor Trim, Media DRM Keys
CMOS 影像感測器 (CIS) 校準、顯示驅動 IC (DDIC) Gamma 修補CIS sensor calibration, DDIC gamma trim, HDMI/HDCP keys
穿戴常態參數、校準更新Wearable Params, Sensor Cal.
藍牙 BLE 裝置配對金鑰、感測器生命週期定期重新校準BLE pairing records, periodic sensor recalibration
小型 MCU 代碼、物聯網日誌Small MCU Code & IoT Logs
8/16-bit MCU 韌體儲存、智慧電表參數與低頻率日誌記錄8/16-bit MCU code storage, smart meter event logging
車用高性能 MCU 主程式碼Auto High-Perf MCU Firmware
車身控制器、底盤動力 MCU 程式碼快閃(Code Flash)Automotive body & chassis MCU main code flash with OTA
BCD 電源管理、工業控制韌體BCD PMIC & Industrial MCU
高壓馬達驅動、工業微控制器程式碼儲存High-voltage motor drive controllers, industrial MCU code
車用黑盒子、邊緣 AI 緩存、SRAM 替換Auto Blackbox, Edge AI, SRAM
車載即時駕駛日誌、AI 推論模型權重暫存、低功耗常關閉 MCUMission-critical flight recorders, Edge AI weights, normally-off
類腦神經形態運算、高密度儲存Neuromorphic CiM, High-Density
記憶體內運算 (CiM) 矩陣乘法、低成本物聯網 MCU 代碼In-memory computing crossbars, cost-effective IoT MCU storage
工業關鍵儀表、防竄改即時記錄Critical Metering & Fast Logs
智慧電表不斷電秒級記錄、醫療監護儀器即時狀態保存Smart metering instant power-loss logs, medical instrument state
🔬 架構師量化推論實驗室Architectural Quantitative Workbench

從經驗直覺走向數值推論:三大 SoC IP 商業整合量化試算器From Intuition to Rigorous Modeling: Three Commercial SoC IP Calculators

透過晶圓代工廠 PDK 標定參數、ATE 機台時薪模型與 Arrhenius 活化能熱物理方程式,即時量化晶粒面積稀釋率、量產測試溢價與車規高溫壽命退化。 Real-time quantification of macro peripheral dilution, volume ATE test overhead, and automotive Arrhenius degradation using foundry PDK baselines, tester economics, and thermal physics equations.

CALC 01 周邊稀釋與面積交叉Macro Peripheral Dilution

巨集周邊開銷與矽面積交叉計算器Macro Overhead & Crossover Calculator

巨集總面積:Total Macro Area: 0.0135 mm²
周邊電路開銷佔比:Peripheral Dilution: 72.4%
極度優勢:小容量下 0-Mask 省去龐大高壓幫浦與光罩費,總成本與面積全面碾壓 eFlash。 Dominant: At ≤256Kb, zero-mask eliminates massive charge pump & mask fees, vastly outperforming eFlash.
CALC 02 量產測試經濟學ATE Tester Economics

晶圓級 ATE 測試秒數機台成本試算器Wafer ATE Test Time & Cost Calculator

單晶粒測試秒數:Test Time / Die: 0.0355 s
千萬顆量產測試費:10M Units Tester Cost: $1,479 USD
省本亮點:相較於逐位元 Program-Verify 浮閘技術,單脈衝並行燒錄節省了逾 $8,500 USD 的 ATE 機台機時費用。 Cost Highlight: Compared to iterative program-verify floating gates, single-pulse gang programming saves over $8,500 USD in tester time.
CALC 03 車規熱物理Arrhenius Thermal Physics

車規高溫 Arrhenius 活化能與壽命外插儀Arrhenius AEC-Q100 High-Temp Extrapolator

Arrhenius 熱加速因子:Thermal AF: 1.25 × 10³
Tj 下推估保持壽命:Projected Retention: > 15.0 Years
完整合規:超越 AEC-Q100 Grade 0(150°C/15年)並完全耐受 260°C SMT 迴焊。 Fully Compliant: Exceeds AEC-Q100 Grade 0 (150°C/15yr) with 260°C SMT reflow immunity.
📐 迷思拆解 01 · 面積真實成本MYTH BUSTER 01 · REAL AREA COST

Cell Area (F²) vs. Macro Density & 周邊開銷Cell Area (F²) vs. Macro Density & Peripheral Overhead

學術文獻極度重視「單位位元胞面積 F²」(例如 AntiFuse 50~100 F²,MRAM 20~40 F²,eFlash 10~25 F²)。但在 SoC 實務中,容量低於 128Kb~1Mb 時,巨集總面積主要由「周邊電路」(高壓電荷幫浦、位址解碼器、感測放大器、測試介面)主導。對於小容量安全金鑰、校準修調(Trim)與晶片 ID,零額外光罩的 AntiFuse 或 eFuse 即使單元 F² 較大,但省去 eFlash 龐大的前段光罩加價與高壓幫浦面積,小晶粒的總成本依然遠勝 eFlash。Academic papers place overwhelming emphasis on raw bit-cell area in F² (e.g., AntiFuse at 50~100 F², MRAM at 20~40 F², eFlash at 10~25 F²). However, in SoC commercial reality, for capacities below 128Kb~1Mb, the total macro footprint is predominantly governed by peripheral circuits (high-voltage charge pumps, row/column decoders, sense amplifiers, and BIST controllers). For small-capacity secure keys, analog trim, and chip IDs, zero-mask AntiFuse or eFuse—despite larger cell F²—saves millions in mask fees and eliminates bulky high-voltage peripheral blocks, delivering vastly superior total silicon economics.

⚡ 迷思拆解 02 · 供電軌與瞬態電流MYTH BUSTER 02 · POWER RAILS & TRANSIENT CURRENT

Programming Voltage vs. 晶片內建幫浦契約Programming Voltage vs. On-Chip Pump Integration Contract

文獻常將 AntiFuse 的 5V~7V 或 eFlash 的 9V~12V 列為「高壓缺點」。但現代 SoC IP 均已整合「晶片內建升壓電荷幫浦(On-chip Charge Pump)」,晶片外部僅需標準 0.8V~1.8V VDD 核心電壓即可運作。真正的架構考量點在於:寫入期間幫浦產生的「瞬態尖峰電流(Transient Peak Current)」(可達數十 mA)與 IR Drop 壓降防護,以及是否需要封裝級專用引腳(如量產 ATE 燒錄專用 VPP 引腳以縮短測試時間)。Literature often cites AntiFuse 5V~7V or eFlash 9V~12V programming voltages as major drawbacks. However, modern commercial NVM IP seamlessly integrates on-chip charge pumps, requiring only standard 0.8V~1.8V core VDD from external pins. The true architectural criteria lie in transient peak currents during programming (which can surge to tens of milliamps), substrate noise coupling, power-grid IR-drop mitigation, and whether a dedicated external VPP pin is provisioned for high-speed ATE gang-programming.

🌡️ 迷思拆解 03 · 車規任務環境與物理壽命MYTH BUSTER 03 · MISSION PROFILE & AEC-Q100

名義 10 年留存 vs. Arrhenius 活化能與 SMT 迴焊Nominal 10-Yr Retention vs. Arrhenius Activation Energy & SMT Reflow

規格書上的「10 年 @ 85°C 留存」對車規(AEC-Q100 Grade 0/1)是遠遠不夠的。車載發動機與底盤環境要求結溫 Tj 達 150°C~175°C 連續工作。根據 Arrhenius 關係式,浮閘技術在高溫下存在電子熱激發洩漏(活化能 Ea 約 1.1~1.5 eV);而 AntiFuse 透過物理擊穿形成永久性低阻抗矽微絲導電通道,其活化能極高且不受電阻漂移影響,更可在經過 260°C 無鉛 SMT 迴焊後 100% 保持預燒資料,是車規與航太首選。Data sheets claiming "10-year retention @ 85°C" are wholly insufficient for automotive qualification (AEC-Q100 Grade 0/1). Powertrain and chassis electronics require continuous operation at junction temperatures of 150°C~175°C. Governed by the Arrhenius equation, floating-gate charge carriers suffer thermal de-trapping (Ea ~ 1.1~1.5 eV). In contrast, AntiFuse relies on permanent dielectric breakdown forming conductive polysilicon filaments, exhibiting immense thermal stability and zero data loss through 260°C lead-free SMT reflow.

⏱️ 迷思拆解 04 · 量產測試經濟學MYTH BUSTER 04 · ATE TEST TIME & COST

寫入時間 vs. 晶圓級 ATE 測試秒數與測試機台費用Write Time vs. Wafer-Level ATE Test Seconds & Tester Economics

學術論文常將寫入時間列為「ns ~ µs」等級的比較。在量產製造中,高階 ATE(如 Teradyne UltraFLEX, Advantest V93000)每秒測試成本極為高昂。若 NVM 寫入需執行「逐位元驗證與重試(Program-Verify Loop)」或單元編程速度過慢,將導致 CP(Circuit Probe)晶圓測試秒數倍增。AntiFuse 支援百微秒級單脈衝熔斷與高度並行燒錄(Gang Programming),配合外供 VPP,能將 4Kb 金鑰的 ATE 測試時間壓縮至數百毫秒內,為千萬顆量產節省數百萬美元測試費用。Academic comparisons record write speeds in ns ~ µs without contextualizing production economics. In mass manufacturing, high-end ATE (e.g., Teradyne UltraFLEX, Advantest V93000) cost significant cents per second. If an NVM requires prolonged iterative Program-Verify loops, CP (Circuit Probe) wafer test times explode. AntiFuse enables sub-millisecond single-pulse breakdown and multi-site parallel gang programming, slashing 4Kb key provisioning test time to under 200 ms and saving millions in wafer probe costs.

02 · 示範性啟發實驗室(非生產選型器)02 · DEMO HEURISTIC LAB (not a production selector)

5 軸多維動態雷達選型演示(相對適配分;非 production selector) 5-Axis Multidimensional Architecture Radar Selector (demo heuristic)

調節五大設計約束滑桿,即時畫出各技術相對輪廓。分數為示範性 heuristic demo(相對適配),不是 production selector、不是預設贏家、也不是出貨證明。輪廓為族群級:eFuse 為吹斷級;I-fuse 與 Logic MTP 另列候選。TwinBit/Actt/G2 等具名例外見矩陣註解 05,不另畫獨立輪廓。 Move the five constraint sliders to redraw relative hulls. Scores are a heuristic demo of relative fit under the current sliders — not a production selector, not a default winner, and not shipment proof. Hulls are family-class: eFuse is blow-class; I-fuse and Logic MTP are listed as separate candidates. Named exceptions (TwinBit / Actt / G2) stay in matrix note 05 and do not get extra hulls.

⚠️ 示意模擬模型與技術邊界說明:⚠️ Schematic Model & Technical Boundary Notice: 本 5 軸多維動態雷達選型器為示範性啟發式評估參考模型(Heuristic Demo),用於展現不同製程節點、功耗與預算權重下之技術架構權衡趨勢,非針對特定晶圓代工廠或專案之唯一 sign-off 選型工具。實際商業架構決策需綜合評估晶圓代工合約、PDK 庫成熟度、光罩報價與完整可靠度資格報告。 This 5-axis dynamic radar selector is an illustrative heuristic reference model demonstrating architectural trade-offs across nodes, power budgets, and cost constraints, rather than an exclusive sign-off selection tool for specific foundry tape-outs. Commercial tape-out decisions require holistic evaluation of wafer supply agreements, PDK IP availability, mask NRE quotes, and full qualification reports.
22nm (Planar HKMG)22nm(平面 HKMG)
180nm40nm22nm12nm2nm GAA
1.0 mW (IoT / Edge)1.0 mW(IoT/邊緣)
0.1mW (ULP)1mW (Edge)1mW(邊緣)25mW (Auto)25mW(車用)250mW (HPC)
0-Mask (Pure Logic)零額外光罩(純邏輯)
0-Mask (logic)零額外光罩(邏輯)3~5 Adders (BEOL)額外 3~5 道光罩(BEOL)8~12 Adders (eFlash)額外 8~12 道光罩(eFlash)
175°C Tj target (not blanket G0 cert)175°C Tj 目標(不代表全面通過 Grade 0 認證)
85°C (Commercial)85°C(商用)125°C (Grade 1)125°C(Grade 1)175°C Tj (design target)175°C Tj(設計目標)
1~100 (RoT / Patching)1~100(RoT/修補)
1~100 (OTP/Patch)10K~100K (MTP/OTA)>1M (Working RAM)
5 軸架構取捨雷達 5-AXIS ARCHITECTURAL TRADE-OFF RADAR
● AntiFuse OTP(相對適配)AntiFuse OTP (relative fit) ● STT-eMRAM(次適配)STT-eMRAM (next)
[相對適配][RELATIVE FIT]
相對適配計算中… Computing relative fit…
03 · 第一性原理物理模型03 · FIRST-PRINCIPLES MODELS

第一性原理物理模型與晶圓成本經濟學方程式 First-Principles Physics & Wafer Economics Equations

三條可推導檢驗的第一性原理恆等式:穿隧與微絲擊穿、光罩階梯加價與良率、次閾值擴散漏電。用來嚴格約束架構選型,而非代工廠報價或單次宣稱。 Three inspectable first-principles identities — tunneling & filament breakdown, mask-adder & yield economics, subthreshold diffusion leakage — constrain architectural selection.

公式 01 · 反熔絲穿隧與擊穿EQUATION 01 · ANTIFUSE BREAKDOWN

Fowler–Nordheim 穿隧與局域矽微絲擊穿 Fowler–Nordheim Tunneling & Localized Filament Breakdown

穿隧機制由能障形狀主導:超薄氧化層在 tox ≤ 3.0 nm 且 Vox < ΦB/q (~3.15 V) 時為梯形能障直接穿隧 (Direct Tunneling);當 Vox > ΦB/q 時轉為三角形能障 Fowler–Nordheim (FN) 穿隧,有效穿隧距離縮短為 xt = ΦB/(qEox)。強電場 Eox ≥ 6~10 MV/cm 下缺陷累積觸發滲透熱失控,形成不可逆的再結晶矽導電微絲: Regime governed by barrier shape: ultrathin oxide at tox ≤ 3.0 nm and Vox < ΦB/q (~3.15 V) exhibits trapezoidal direct tunneling; Vox > ΦB/q triggers triangular FN tunneling with tunneling distance xt = ΦB/(qEox). High field Eox ≥ 6~10 MV/cm drives defect percolation and thermal runaway into a permanent recrystallized silicon filament:

JFN = AFN · Eox2 · exp(−BFN / Eox)
A · FN TUNNELING (TRIANGULAR BARRIER)A · FN 穿隧(三角位障) Poly Gate多晶矽閘極 SiO2 Oxide (tox)SiO2 氧化層(tox) Si SubstrateSi 基板 Ec (Poly) ΦB=3.15eV xt=ΦB/(qEox) J_FN Direct (tox ≤ 3nm) vs FN (Vox > ΦB/q) Triangular B · 4-STAGE RECRYSTALLIZED SILICON FILAMENTB · 再結晶矽細絲的四個階段 1. Stress & Traps1. 應力與陷阱 Oxygen vacancy defects氧空缺缺陷 E' centers accumulationE' 中心累積 2. Percolation (SBD)2. 滲流(SBD) N_bd ≈ 10^19~10^20 cm−3 Soft breakdown onset軟擊穿起始 3. Thermal Runaway3. 熱失控 Local T > 1400°C (HBD) Molten Si atom migration熔融 Si 原子遷移 4. Recrystallized Si4. 再結晶 Si ∅ 5~10nm Si Filament Permanent Phase Change永久相變 HRS >10^11 Ω → Irreversible → LRS ≈ 10^2 Ω (Ohmic)
• 穿隧係數理論解:AFN = q2(m0/m*) / (8π h ΦB) ≈ 1.2×10−6 A/V2;BFN = 4√(2m*q) ΦB3/2 / (3ℏ) = 8π√(2m*q) ΦB3/2 / (3h)。
• 常數基準校正:Si–SiO2 能階差 ΦB ≈ 3.15 eV。若取自由電子有效質量 m* ≈ 0.50 m0,理論值 BFN ≈ 2.70×108 V/cm (270 MV/cm);若採 Lenzlinger–Snow 實驗穿隧質量 m* ≈ 0.42 m0(含影像力修正),則 BFN ≈ 2.48×108 V/cm ≈ 2.5×108 V/cm。
• 微觀崩潰四階段:高場應力誘發氧空位與 E' 中心缺陷生成 → 達滲透閾值 (Nbd ≈ 1019~1020 cm−3) 觸發軟崩潰 (SBD) → 焦耳熱集中引發熱失控硬崩潰 (HBD, 局部 T > 1400°C) → 矽原子遷移凝固為直徑 5~10 nm 的局域再結晶矽微絲,阻抗由 >1011 Ω 驟降至 102 Ω 歐姆態。此為不可逆之金屬/半導體相變,非 ReRAM 可逆微絲。
• 編程電壓縮放:Vprog 由 Eox × tox 及擊穿時間 tbd 約束,隨 Core 閘氧厚度按比例縮放(如 28nm ~ 1.8V-2.5V),不可將 5V/3.3V I/O 浮閘 HCI 的 6.5V~7.5V 數值回填混淆。
• 適用範圍:此式描述 core 閘氧 AntiFuse(含 CFX 教學路徑)。Floadia LEE Fuse ZA 只公開介電層永久導通,擊穿位點未指定,不得套用本微絲模型。
• Theoretical Coefficients: AFN = q2(m0/m*) / (8π h ΦB) ≈ 1.2×10−6 A/V2; BFN = 4√(2m*q) ΦB3/2 / (3ℏ) = 8π√(2m*q) ΦB3/2 / (3h).
• Constant Calibration: Barrier ΦB ≈ 3.15 eV. Theoretical BFN ≈ 2.70×108 V/cm (270 MV/cm) with m* ≈ 0.50 m0; empirical Lenzlinger–Snow value gives BFN ≈ 2.48×108 V/cm ≈ 2.5×108 V/cm with m* ≈ 0.42 m0.
• 4-Stage Breakdown Physics: Stress defect generation → Percolation threshold (Nbd ≈ 1019~1020 cm−3) soft breakdown (SBD) → Joule heating thermal runaway hard breakdown (HBD, T > 1400°C) → Molten silicon migration forming 5~10 nm recrystallized silicon filament shifting resistance from >1011 Ω to 102 Ω ohmic state. Permanent metallurgical phase change, unlike ReRAM.
• Scaling Constraint: Vprog scales with Core Eox × tox and allowed tbd; never back-fill 6.5V~7.5V I/O floating-gate HCI voltages onto thin-oxide AntiFuse.
• Scope: this identity describes core-GOX AntiFuse, including the CFX teaching route. Floadia LEE Fuse ZA is published only as a dielectric made permanently conductive; the breakdown site is unpublished — do not apply this filament model to ZA.
公式 02 · 晶圓成本與良率EQUATION 02 · WAFER COST & YIELD

額外光罩晶圓成本與良率恆等式 Mask-Adder Wafer Cost & Effective Yield Identity

將額外光罩道數 ΔNmask、前段/後段邊際加價 αmask 與製程良率衝擊 ΔY 結合成可物理檢查的成本閉環: Integrate extra masks ΔNmask, FEOL/BEOL marginal adders αmask, and process yield hits ΔY into an inspectable cost closure:

Cdie,eff = [Cbase · (1 + Σ αi)] / [GDPW(Adie) · (Ybase − ΔYnvm)]
A · MASK ADDER OVERHEAD ESCALATORA · 額外光罩成本遞增 0-MASK LOGIC NVM零額外光罩邏輯 NVM AntiFuse · I-fuse · TwinBit · LD-MTP ΔN=0 (1.00×) BEOL eNVM (STT-MRAM) α_BEOL ≈ 1.5%~2.5%/mask +3~5 (+15~22%) FEOL eFLASH (SST / SONOS) α_FEOL ≈ 2.5%~4.0%/mask (High-T) +4~10 (+30~45%) Zero Extra Masks ≠ Free Silicon (Array Area Penalty) B · EFFECTIVE DIE COST & YIELD IMPACTB · 有效晶粒成本與良率影響 Die Area Adie晶粒面積 Adie Cost C_die,eff成本 C_die,eff 0-Mask (ΔN=0) +Adders & ΔY_nvm GDPW(Adie) · (Y_base − ΔY_nvm) Divisor Penalty
• 光罩邊際成本差異 (αmask):FEOL 關鍵層涉及高能植入與高溫退火,邊際加價率 αFEOL ≈ 2.5%~4.0%/mask;BEOL 金屬穿插層邊際加價率 αBEOL ≈ 1.5%~2.5%/mask。
• 技術陣營量級:
  – SST-class eFlash:FEOL ΔNmask 常為 +4(ESF1 例)至 +10(ESF3 例),不得把所有 eFlash 硬編 8~12。
  – SONOS eFlash:Floadia G1 公開 +2~3;G2 公開 +4(55 BCD,開發中)。
  – STT-MRAM:BEOL ΔNmask = 3~5,晶圓加價 ΔCwafer ≈ +15%~22%,金屬回濺增加缺陷密度 ΔD0。
  – 0-mask 邏輯 NVM:AntiFuse、吹斷級 eFuse、I-fuse、LD-MTP(NeoEE/ZT)、TwinBit 均為 ΔNmask = 0;單元統計與感測面積仍佔矽。零額外光罩 ≠ 晶圓免費。
• Marginal Mask Adders (αmask): FEOL critical layers involve high-energy implants and thermal cycles (αFEOL ≈ 2.5%~4.0%/mask); BEOL interconnect modules add αBEOL ≈ 1.5%~2.5%/mask.
• Technology Spectrum:
  – SST-class eFlash: FEOL ΔNmask commonly +4 (ESF1 class) to +10 (ESF3 class); do not hard-code every eFlash at 8~12.
  – SONOS eFlash: Floadia G1 public +2~3; G2 public +4 (55 BCD, in development).
  – STT-MRAM: BEOL ΔNmask = 3~5, wafer adder ΔCwafer ≈ +15%~22%, physical sputter redeposition alters defectivity ΔD0.
  – 0-mask logic NVM: AntiFuse, blow-class eFuse, I-fuse, LD-MTP (NeoEE / ZT), and TwinBit are ΔNmask = 0; cell statistics and sense area still occupy silicon. Zero extra masks ≠ free wafers.
公式 03 · 次閾值擴散漏電EQUATION 03 · SUBTHRESHOLD DIFFUSION

次閾值擴散漏電與玻爾茲曼極限 Subthreshold Diffusion Leakage & Boltzmann Tyranny

超低功耗邊緣節點 Always-On 域,關態靜態漏電由次閾值擴散機制與源汲熱平衡因子共同決定: Static leakage in always-on edge MCU domains is dominated by subthreshold diffusion and the drain balance factor:

Isub = μeff Cox (W/L) (kBT/q)2 (m−1) · exp[(q(Vgs−Vth))/(m kBT)] · [1 − exp(−q Vds / kBT)]
A · SUBTHRESHOLD SWING (BOLTZMANN TYRANNY)A · 次臨界擺幅(Boltzmann 限制) Vgs (V) log(Ids) Vth 60 mV/dec (m=1) CMOS S ≈ 75 mV/dec Drift Inversion漂移反轉 Subthreshold Diffusion Regime (Vgs < Vth) B · DRAIN EQUILIBRIUM SATURATION FACTORB · 汲極平衡飽和因子 Vds (mV) [1 − e^(−qVds/kT)] 1.0 3kT/q (≈78mV) Saturates → 1.0 Linear at Vds → 0 Gated Standby Leakage < 10 nA Class
• 亞閾值擺幅 (S) 與熱力學極限:S = ln(10) · m · (kBT/q) ≈ 59.5 · m mV/dec (@ 300 K)。在理想閘極靜電控制下 (m → 1.0),受限於玻爾茲曼統計極限 60 mV/dec;平面 CMOS 因空乏層分壓 m = 1 + Cdep/Cox ≈ 1.2~1.4,實測 S ≈ 70~85 mV/dec。
• 汲極飽和抑制因子:[1 − exp(−qVds/kBT)] 在 Vds ≥ 3 kBT/q (~78 mV) 時迅速飽和趨近於 1;在 Vds → 0 時線性歸零,嚴格滿足熱力學平衡零流約束。
• 長效待機啟示:感測路徑若採用 Power-Gating 或深次閾值偏置,靜態漏電可壓制在 <10 nA 等級,惟需經晶圓製程角 (Corner) 與高溫 (125°C) 矽驗證。
• Subthreshold Swing (S) & Boltzmann Limit: S = ln(10) · m · (kBT/q) ≈ 59.5 · m mV/dec (@ 300 K). Under ideal electrostatic coupling (m → 1.0), constrained by the 60 mV/dec thermal limit; planar CMOS exhibits m = 1 + Cdep/Cox ≈ 1.2~1.4, yielding S ≈ 70~85 mV/dec.
• Drain Saturation Factor: [1 − exp(−qVds/kBT)] saturates to ~1 when Vds ≥ 3 kBT/q (~78 mV) and vanishes linearly at zero bias, satisfying thermodynamic equilibrium.
• Standby Implication: Gated sense paths suppress standby drain to <10 nA class, subject to silicon corner and high-temperature (125°C) validation.
第一性原理晶圓光罩附加與 TCO 經濟學模型 FIRST-PRINCIPLES WAFER MASK-ADDER & TCO ECONOMICS MODEL

晶圓光罩附加成本、Murphy 矽良率損失與百萬晶圓量產 TCO 經濟學試算器 eNVM Wafer Mask Adder, Murphy Yield Loss & Mass-Production TCO Economics Calculator

300mm 晶圓量產標準 300mm Wafer Fab Standard

在半導體決策中,評估 eNVM 不能僅看裸單元面積(F²),必須閉環計入額外光罩道數(ΔN)、高溫熱退火對 CMOS 閘極氧化層的缺陷密度增加(ΔD₀)、Murphy 矽良率損失與光罩組 NRE 攤提。本工作台透過第一性原理晶圓成本恆等式與 Murphy 缺陷模型,即時量化純邏輯 AntiFuse(0 光罩)對比 eFlash(+10 光罩)與 eMRAM(+4 光罩)的全生命週期 TCO 經濟學價值。 In semiconductor architecture decisions, evaluating eNVM requires looking beyond bitcell area (F²). Architectural cost closure demands factoring in mask-adder layers (ΔN), defectivity density penalties (ΔD₀) from high-temperature thermal budgets, Murphy silicon yield degradation, and mask-set NRE amortization. This calculator applies first-principles wafer cost identities and Murphy yield models to quantify the lifecycle TCO advantages of zero-mask AntiFuse over eFlash (+10 masks) and eMRAM (+4 masks).

⚠️ 示意模擬模型與技術邊界說明:⚠️ Schematic Model & Technical Boundary Notice: 本工作台之光罩附加成本比率 (α)、Murphy 矽良率損失與百萬晶圓 TCO 為基於公開一階成本恆等式與典型缺陷密度模型之數值試算,非特定晶圓代工廠機密晶圓報價單或特定晶片之產線良率承諾。實際生產成本需依據晶圓採購協議 (WPA)、量產規模與合約良率保證條款確認。 The mask-adder cost ratios (α), Murphy yield loss, and million-wafer TCO in this workbench are schematic numerical approximations based on public first-order cost equations and defect models, rather than confidential foundry price schedules or tape-out yield commitments. Actual production economics depend on volume wafer supply agreements and contract yield targets.
單晶片總面積 (Die Area):DIE AREA: 25.0 mm²
年度晶圓投片總量:ANNUAL WAFER VOLUME: 30,000 Wafers
光罩附加數MASK ADDERS
+0 Masks
42 Total
單片總晶圓成本TOTAL WAFER COST
$3,240
+$40 NRE/Wafer
毛晶粒數 GDPWGROSS DIES (GDPW)
2,580
300mm 晶圓幾何300mm Geometry
Murphy 矽良率MURPHY YIELD
92.4%
2,383 Good Dies
良品晶片成本COST / GOOD DIE
$1.37 / Die
Silicon + Test
年度總擁有成本ANNUAL TCO
$97.2M
Baseline (0-Mask)
0 光罩純邏輯基準0-Mask Reference
Loading first-principles wafer cost and TCO economics calculations...
3D 垂直堆疊物理 · 高深寬比階梯刻蝕與薄膜尺寸效應 3D VERTICAL STACKING PHYSICS · HAR STAIRCASE ETCH & SIZE EFFECTS

3D 立體垂直堆疊 eNVM 字元線階梯 RC 延遲與薄膜金屬電阻率尺寸效應模擬器 3D Vertical Stacked eNVM Word-Line Tier RC Delay & Conductor Size-Effect Simulator

HAR 階梯深孔 / 鉬 (Mo) 薄膜物理模型 HAR Staircase / Molybdenum Thin-Film Model

在 3D 垂直堆疊記憶體(如 3D AntiFuse OTP、3D NOR 與 3D NAND)中,層數(Tier Count,32 至 128+ 層)的增加大幅提高了單元密度;然而高深寬比(HAR)階梯刻蝕走線使最底層字元線(Word-Line)物理長度急劇延伸。同時,金屬薄膜厚度微縮(<25 nm)觸發 Fuchs-Sondheimer 表面散射與 Mayadas-Shatzkes 晶界散射,使鎢(W)電阻率暴增數倍,引發嚴峻的字元線 RC 延遲梯差(Delay Skew)。改採先進低電阻率金屬鉬(Mo)或超短平均自由程釕(Ru)可有效壓低 RC 延遲,維持端到端高速讀取存取頻寬。 In 3D vertical stacked memories (such as 3D AntiFuse OTP, 3D NOR, and 3D NAND), increasing tier counts (32 to 128+ layers) dramatically multiplies bit density. However, high-aspect-ratio (HAR) staircase contact routing drastically extends the physical length of lower word-lines. Concurrently, scaling conductor film thickness below 25 nm triggers severe Fuchs-Sondheimer surface and Mayadas-Shatzkes grain-boundary scattering, surging tungsten (W) resistivity and causing severe RC delay skew across tiers. Transitioning to advanced molybdenum (Mo) or ultra-short mean-free-path ruthenium (Ru) suppresses RC delay, sustaining high random-read bandwidth.

⚠️ 示意模擬模型與技術邊界說明:⚠️ Schematic Model & Technical Boundary Notice: 本工作台之字元線階梯 RC 延遲與 Fuchs-Sondheimer / Mayadas-Shatzkes 薄膜金屬電阻率散射為基於一階集總傳輸線與半經驗尺寸效應模型之數值示意,非 3D TCAD 或高頻電磁場求解器 (HFSS) 之實體驗證網格數據。實體深階梯延遲受刻蝕輪廓側壁傾角、孔徑均勻性與阻障層 (TiN/TaN) 界面粗糙度支配。 The word-line staircase RC delays and Fuchs-Sondheimer / Mayadas-Shatzkes conductor size-effect resistivity in this workbench are schematic numerical approximations based on first-order transmission-line models, rather than 3D TCAD or full-wave electromagnetic solver outputs. Physical staircase delay skew is governed by sidewall taper angles, etch hole uniformity, and barrier layer interface roughness.
垂直堆疊層數:TIER COUNT: 64 L
金屬薄膜厚度:FILM THICKNESS: 25 nm
陣列水平長度:ARRAY LENGTH: 120 µm
薄膜有效電阻率EFFECTIVE RESISTIVITY
--
最差層 RC 延遲WORST WL DELAY
--
頂底層延遲梯差TIER DELAY SKEW
--
端到端讀取存取時間TOTAL ACCESS TIME
--
3D Stacked Physical Analyzer
Loading 3D vertical stacked eNVM calculations...
High-Speed 3D Fabric
04 · 晶圓代工與實體安全04 · FOUNDRY & SECURITY

主流晶圓代工與專用類比製程路線圖暨實體防護機制 Foundry & Specialty Analog Ecosystem Roadmap & Physical Security Integrity

主流晶圓代工廠真實 PDK 量產狀態、車規等級驗證進展與奈米級實體防護責任鏈。 Foundry PDK qualification envelopes, automotive grade readiness, and layered physical anti-tamper security chains.

晶圓廠 01 · 台灣積體電路 (TSMC)FOUNDRY 01 · TAIWAN SEMICONDUCTOR

TSMC

40nm / 28nm eFlash & 28eHV 0-mask OTP 須分流機制0-mask OTP: split mechanisms 依製程核對Process-specific check
此處為製程平台導覽,不是巨集資格證明。eFlash 的光罩數、寫入/抹除電壓及熱預算需查該版 PDK;不得把另一家晶圓廠的 SST 光罩數或通用高溫假設套入 TSMC 平台。顯示驅動 OTP 也須分流閘氧 AntiFuse、擊穿位點未公開的 ZA 類與 I-fuse。28 nm 範圍說明。 This is a process-platform guide, not macro qualification evidence. Check eFlash masks, program/erase voltage and thermal budget against the specific PDK revision; do not transfer another foundry's SST mask count or a generic thermal assumption onto TSMC. Display-driver OTP also requires separate gate-oxide AntiFuse, unpublished-site ZA-class and I-fuse routes. 28 nm scope.
22ULL / 22ULP 0-mask OTP 須分流機制0-mask OTP: split mechanisms 車規 Grade 1Auto Grade 1
eRRAM 於 2022 年量產並於 2025 年通過 AEC-Q100 Grade 1 (10萬次循環);STT-eMRAM 商業量產;純邏輯 0-mask OTP 作為標配時須分流閘氧 AntiFuse、擊穿位點未公開的 ZA 類、以及 I-fuse,不得預設單一機制。 eRRAM entered production in 2022, qualifying for AEC-Q100 Grade 1 (100K cycles) in 2025; STT-eMRAM production-ready; logic 0-mask OTP as a standard option still requires a gate-oxide AntiFuse vs unpublished-site ZA-class vs I-fuse split rather than a single assumed mechanism.
16FFC / 12FFC+ (N12e / HBM4 Base Die) 100萬次 <1 ppm · HBM4 代工<1 ppm @ 1M · HBM4 Foundry
16FFC eMRAM 於 2023 年通過 Grade 1,第二代 16MRAM 於 2025 年達成 100 萬次循環失效率 <1 ppm;N12e eRRAM 於 2025 年完成生產級消費資格。此外,SK 海力士 HBM4 Base Die 委託台積電 12nm (N12FFC+) 代工(下一代邁入 3nm 級),採用純邏輯 0-mask AntiFuse OTP 進行 2048-bit PHY 與 16 層 DRAM TSV 缺陷修復,並由微間距 MEMS 高針數邏輯探針卡介面(如旺矽科技 MPI、中華精測 CHPT、Technoprobe、FormFactor 等架構)支援晶圓測試 (Wafer Sort)。 16FFC eMRAM passed Grade 1 in 2023; Gen-2 16MRAM qualified in 2025 with <1 ppm failure rate at 1M cycles; N12e eRRAM qualified in 2025. Furthermore, SK Hynix selected TSMC 12nm (N12FFC+) for HBM4 Base Die manufacturing (advancing to 3nm class), utilizing 0-mask logic AntiFuse OTP for 2048-bit PHY and TSV redundancy repair, supported by fine-pitch high-pin-count logic MEMS probe card interfaces (e.g., architectures from MPI Corp, CHPT, Technoprobe, FormFactor) for wafer sort testing.
N6e / N4e / N2 GAA 0-mask OTP 須分流機制0-mask OTP: split mechanisms 奈米片信任根Nanosheet RoT
N6e (6nm) eRRAM 與 N4e/N5 5nm 高速 eMRAM 開發中;N2 奈米片因 BEOL 熱預算與微縮限制,公開敘事常以純邏輯 0-mask OTP 作為硬體信任根,但仍須分流閘氧 AntiFuse、擊穿位點未公開的 ZA 類、以及 I-fuse,不得預設單一機制。 N6e 6nm eRRAM and N4e/N5 5nm high-speed eMRAM in R&D; N2 GAA nanosheet public narratives often use 0-mask logic OTP as the Root of Trust due to BEOL thermal constraints, but still require an AntiFuse vs unpublished-site ZA vs I-fuse split rather than one assumed mechanism.
N2P / A16 (1.6nm) & BSPDN 背面供電熱預算BSPDN thermal limits 2026 前沿架構2026 Frontier
TSMC 規劃之 A16 (1.6nm) 節點結合 NanoFlex GAA 與 Super PowerRail 背面供電網路(BSPDN),將金屬走線 IR 壓降降低並釋放正面信號繞線密度。對於嵌入式記憶體而言,背面供電進一步壓低 BEOL 後段熱預算(<350°C),使 0-mask 純邏輯 AntiFuse OTP 成為不可替代之硬體唯一識別根(RoT)與 SRAM 修復載體;而高密度 eMRAM 則需極低溫沉積之新型 MTJ 磁性材料堆疊。 TSMC A16 (1.6nm) integrates NanoFlex GAA with Super PowerRail Backside Power Delivery (BSPDN), mitigating IR drop and freeing front-side routing resources. For embedded memory, BSPDN tightens BEOL thermal budgets (<350°C), cementing 0-mask logic AntiFuse OTP as the indispensable Hardware Root-of-Trust (RoT) and SRAM repair carrier, while dense eMRAM requires low-thermal-budget MTJ material stacks.
晶圓廠 02 · 聯華電子 (UMC)FOUNDRY 02 · UNITED MICROELECTRONICS

UMC

55nm / 40nm eFlash & 40/28eHV 0-mask OTP 須分流機制0-mask OTP: split mechanisms 量產 SSIProduction SSI
採用 SST SuperFlash 分離閘極技術量產車用 Grade 1 eFlash;40/28eHV 驅動 IC 整合高壓校準 OTP。同節點校準 OTP 仍須分流閘氧 AntiFuse、擊穿位點未公開的 ZA 類、以及 I-fuse,不得預設單一機制。 Production SST SuperFlash split-gate for Auto Grade 1 MCUs; 40/28eHV DDICs incorporate trim OTP. Same-node trim OTP still requires a gate-oxide AntiFuse vs unpublished-site ZA-class vs I-fuse split — do not assume one mechanism.
28HPC+ / 22ULL (MRAM & RRAM) 商用 P-SRAMCommercial P-SRAM
與 Avalanche 合作之 22nm pMTJ STT-MRAM 於 2022 年商業量產 (P-SRAM);22ULL eMemory RRAM IP 於 2023 認證,智原 FlashKit-22RRAM 平台於 2025 完成流片驗證。 Avalanche 22nm pMTJ STT-MRAM in commercial production since 2022 (P-SRAM); 22ULL eMemory RRAM IP qualified in 2023; Faraday FlashKit-22RRAM SoC tape-out validated in 2025.
邏輯 OTP / MTP 生態系Logic OTP / MTP Ecosystem 原生 0 光罩0-Mask Native
公開 PDK 常見力旺 (NeoBit / NeoFuse) 與晶豪科旗下 YMC (yMTP BBHH) 0-mask 組合;NSCore TwinBit、Floadia ZA/ZT、Attopsemi I-fuse 等其他 0-mask IP 須逐家核對 PDK,不得由本句推論「已全面支援」。 Public PDK coverage commonly includes eMemory (NeoBit/NeoFuse) and YMC (yMTP BBHH) 0-mask logic NVM across 55nm–22nm. Other 0-mask IPs (NSCore TwinBit, Floadia ZA/ZT, Attopsemi I-fuse) must be checked per foundry PDK — this sentence does not imply blanket support.
利基與超低壓 eFlash 生態系Niche & Low-Voltage eFlash Ecosystem 低功耗 / DDR5 SPD 量產Low Power / DDR5 SPD Production
除主流 SST 外,成熟晶圓代工與專業 IP 廠提供多元利基 eFlash 架構:常憶科技 (Chingistek, chingistek.com) pFusion™ 以 P-Channel 熱電洞注入 (BBHH) 提供極低寫入功耗;智憶科技 (IOTMemory, iotmemory.com) SilvoFlash™ 提供原生 0.9V–1.2V 超低讀取電壓,並在 40nm 及周邊專用晶片(如 DDR5 SPD IC)達成商業量產。 Beyond mainstream SST, mature foundries and specialized IP vendors offer niche eFlash architectures: Chingis Technology (chingistek.com) pFusion™ leverages P-Channel hot-hole injection (BBHH) for ultra-low programming power; IOTMemory (iotmemory.com) SilvoFlash™ delivers native 0.9V–1.2V low-voltage read capability, achieving commercial production on 40nm and dedicated peripheral ICs (e.g. DDR5 SPD ICs).
晶圓廠 03 · 格羅方德 (GLOBALFOUNDRIES)FOUNDRY 03 · GLOBALFOUNDRIES

GlobalFoundries

22FDX / FDX+ (FD-SOI) 巨集資格待核對Check macro qualification
22FDX/FDX+ 的記憶體選項,須以具名巨集與資格報告核對。平台名稱或路線圖不足以證明特定 AEC-Q100 等級、150°C 的 TA/TJ 定義、耐寫次數或讀取延遲;此處不把原型目標列為量產資格。Attopsemi 於 2019 年公告 I-fuse™ OTP 在 22FDX 的 HTS/HTOL 資格結果(原廠公告);不能轉作 22FDX eMRAM 的資格證據。 Check 22FDX / FDX+ memory options against a named macro and qualification report. A platform name or roadmap does not establish an AEC-Q100 grade, whether 150°C means TA or TJ, endurance, or read latency; prototype targets are not production qualification. Attopsemi reported I-fuse™ OTP HTS / HTOL qualification on 22FDX in 2019 (vendor announcement); that is not qualification evidence for 22FDX eMRAM.
28SLP & 12LP / 12LP+ FinFET 邊緣 AI 協同開發Edge AI Co-Dev
28SLP 支援成熟 eFlash(含 SST SuperFlash 公開節點);12LP 與 Everspin 共同開發 MRAM;12LP+ AutoPro150 推進車用 MCU 與邊緣 AI 部署(注意:12LP/12LP+ 為 Bulk FinFET,與 22FDX FD-SOI 物理特性不同)。Actt LogicFlash® 在公開 IP 目錄列於 GF 180–110 nm,屬成熟平面 MTP,不是 12LP FinFET 聲明。 28SLP supports mature eFlash (including public SST SuperFlash nodes); 12LP FinFET MRAM co-developed with Everspin; 12LP+ AutoPro150 targets automotive MCUs (Note: 12LP is FinFET, distinct from 22FDX FD-SOI). Actt LogicFlash® is listed on public IP directories for GF 180–110 nm as mature planar MTP, not a 12LP FinFET claim.
晶圓廠 04 · 三星代工 (SAMSUNG FOUNDRY)FOUNDRY 04 · SAMSUNG FOUNDRY

Samsung Foundry

28FDS (FD-SOI eMRAM) 2019 年起出貨Shipped Since 2019
自 2019 年 3 月宣布商用出貨 eFlash 替代型 eMRAM,寫入快 1000 倍,具備 100 萬次循環與 125°C 10 年保存,支援 5 次 260°C 迴焊。 Commercial shipment of 28FDS eFlash-type eMRAM since March 2019: 1000x faster write, 1M cycles, 10-yr retention at 125°C, surviving 5x 260°C reflows.
14LPP / 8LPU FinFET 車規與射頻 SoCAuto & RF SoC
將 eMRAM 技術自 FD-SOI 延伸至 14nm FinFET 與 8LPU,提供高密度車規 Grade 1 與穿戴式 SoC 支援。 Extended eMRAM to 14nm FinFET and 8LPU, targeting automotive Grade 1 and low-power wearable SoCs.
SF4A (4nm) 與 SF3 / SF2SF4A (4nm) & SF3 / SF2 HBM4 Base Die 代工HBM4 Base Die Foundry 若另採 OTP 須分流機制If OTP used: split mechanisms HBM4 邏輯底座 · 次世代 GAAHBM4 Logic Base · Next-Gen GAA
SFF 2024 公布次世代高密度 eMRAM 路線圖,整合於 4nm 車用與 3nm/2nm 環繞閘極奈米片,專注於邊緣 AI 與軟體定義汽車 (SDV)。同時,三星自家 HBM4 Base Die 轉向 Samsung Foundry 4nm 先進邏輯製程,搭載 2048-bit 超寬 PHY,由內嵌邏輯 eNVM/AntiFuse 執行 TSV 與 DRAM 壞列重映射,晶圓測試採購權力亦自記憶體部門大舉轉向晶圓代工高階邏輯探針卡生態系。若同節點另採 0-mask OTP 作信任根或校準,仍須分流閘氧 AntiFuse、擊穿位點未公開的 ZA 類、以及 I-fuse,不得預設單一機制。 SFF 2024 roadmap integrates next-gen eMRAM on SF4A (4nm Auto) and SF3/SF2 MBCFET GAA for edge AI and software-defined vehicles. Concurrently, Samsung's in-house HBM4 Base Die adopts Samsung Foundry 4nm advanced logic, integrating a 2048-bit ultra-wide PHY with embedded logic eNVM/AntiFuse for TSV and DRAM redundancy repair, shifting wafer sort test procurement into the foundry logic probe card ecosystem. If the same node also uses 0-mask OTP for RoT or trim, still split gate-oxide AntiFuse, unpublished-site ZA-class, and I-fuse — do not assume one mechanism.
晶圓廠 05 · 特種類比與 BCDFOUNDRY 05 · SPECIALTY ANALOG & BCD

Tower & VIS / PSMC

Tower Y-Flash (0.18um / 65nm BCD)Tower Y-Flash (0.18um / 65nm BCD) 0 光罩 BCD Flash0-Mask BCD Flash
Tower 自研 0 額外光罩 (0 Mask Adders) 單層多晶矽 eFlash/MTP,原生相容 0.18um 與 65nm BCD 製程;支援 150°C~175°C 車規耐溫留存,對高壓 LDMOS 元件擊穿電壓 (BVdss) 零干擾。 Tower proprietary 0-mask single-poly eFlash/MTP natively integrated on 0.18um and 65nm BCD platforms; qualified for 150°C-175°C automotive retention without degrading LDMOS breakdown voltage (BVdss).
世界先進 VIS / 力積電 PSMC (BCD/HV)VIS & PSMC (BCD / HV Ecosystem) 電源微調 MTPPMIC Trim MTP
全面整合億而得 (YMC) 純邏輯多次抹寫 (0-Mask / 1-Mask MTP),內建專利微型電荷泵,為高壓 PMIC 基準電壓、馬達驅動器偏置與感測器提供出廠即時修調 (Trimming)。 Full integration of YMC 0-mask/1-mask logic MTP with on-chip charge pump, enabling precision trimming for PMIC Vout reference, motor driver offsets, and sensor calibration.
上下游縱向整合生態系UPSTREAM & DOWNSTREAM INTEGRATION ECOSYSTEM

非揮發記憶體上下游縱深整合價值鏈:從晶圓製造到系統信任根與異質封裝 End-to-End NVM Value Chain: From Foundry Fabrication to Root-of-Trust & Heterogeneous Packaging

NVM 選擇並非孤立的位元胞比較,而是涵蓋上游晶圓製程(光罩成本/熱預算)、中游 IP 拓撲(穿隧/電荷泵/物理不可複製性)與下游終端系統(車規 ASIL-D、AI 加速器硬體信任根、無源物聯網)的完整生態協同: NVM selection is never an isolated bitcell decision, but a holistic co-optimization across upstream foundries (mask adders/thermal budget), midstream IP topologies (tunneling/charge pump/PUF), and downstream systems (automotive ASIL-D, AI accelerator HRoT, passive IoT):

01 · 硬體信任根01 · HARDWARE ROOT OF TRUST

PUFsecurity 熵碼科技 PUFsecurity

[上游晶圓][Foundry] TSMC (5nm–55nm) 與 UMC 標準 CMOS 純邏輯製程,0 額外光罩 (0 Mask Adders)。TSMC (5nm-55nm) & UMC standard CMOS logic; 0 mask adders.
[中游 IP][Midstream IP] 利用閘極氧化層微觀隨機量子穿隧變異產生物理不可複製晶片指紋 (NeoPUF);整合 AntiFuse OTP、NIST SP 800-90B TRNG 與 PUFcc 密碼處理器,天然具備 TEM/FIB 零實體電荷抗性。Harnesses gate oxide quantum tunneling variations for hardware fingerprinting (NeoPUF); pairs with AntiFuse OTP, NIST SP 800-90B TRNG, and PUFcc crypto engine with static TEM immunity.
[下游系統][Systems] AI 邊緣推論加速器模型權重與金鑰保護、車用 ISO 26262 ASIL-D / ISO 21434 安全開機 (Secure Boot)、PSA Certified Level 3 物聯網認證。Edge AI model/key protection, automotive ASIL-D / ISO 21434 secure boot, and PSA Level 3 IoT device credentials.
02 · 電源與感測器微調校準02 · POWER & SENSOR TRIMMING

YMC 億而得微電子 YMC Microelectronics

[上游晶圓][Foundry] 世界先進 (VIS 0.153um/0.18um/0.25um BCD)、力積電 (PSMC)、UMC 等標準邏輯與高壓製程,0-Mask 至 1-Mask。Vanguard (VIS BCD), PSMC, UMC logic and BCD processes; 0-mask to 1-mask adders.
[中游 IP][Midstream IP] 純邏輯 EEPROM / MTP,FN/FN 穿隧機制,1K~100K 抹寫次數;內建專利微型電荷泵,晶片在普通作業電壓下即可完成出廠程式化,免外部高壓測試腳位。Logic EEPROM/MTP with FN/FN tunneling (1K-100K cycles); integrated on-chip charge pump eliminating external high-voltage programming pins.
[下游系統][Systems] 高壓電源管理 IC (PMIC) 基準電壓微調 (Vout Trim)、馬達驅動器偏置修正、Type-C PD 協議控制器、顯示器驅動 Gamma 校準。PMIC Vout voltage trimming, motor driver offset calibration, Type-C PD controllers, and display driver Gamma calibration.
03 · 無源物聯網與超低功耗03 · PASSIVE IOT & ULTRA-LOW POWER

Impinj (AEON 邏輯 NVM) Impinj (AEON Logic NVM)

[上游晶圓][Foundry] TSMC / UMC 純數位邏輯 CMOS 製程,完全不需要額外光罩或外加高壓退火 (0 Mask Adders)。Standard digital CMOS foundry platforms (TSMC/UMC); zero additional mask adders.
[中游 IP][Midstream IP] 單層多晶矽 (Single-Poly) 浮閘架構,極致優化之次微安級 (Sub-µA) 超低功耗感測放大器,兼顧超高感度與可靠保持。Single-poly floating gate topology featuring sub-microampere read sensing current for extreme energy-harvesting budgets.
[下游系統][Systems] 被動式 UHF RFID Tag (Monza 系列)、無電池微弱 RF 能量採集 (Energy Harvesting) 無源物聯網節點與智慧物流防偽追蹤。Passive UHF RFID tags (Monza family), battery-free RF energy-harvesting IoT nodes, and anti-counterfeiting supply chain trackers.
04 · 小晶片與異質系統級封裝04 · CHIPLET & SIP HETEROGENEOUS

Winbond 華邦 & Macronix 旺宏 Winbond & Macronix

[上游晶圓][Foundry] 專用車規與工規 NOR Flash 晶圓工藝 (45nm/28nm),通過嚴苛之 AEC-Q100 Grade 0 (-40°C ~ +150°C) 認證。Dedicated automotive/industrial NOR Flash processes (45nm/28nm) qualified to AEC-Q100 Grade 0 (-40°C to +150°C).
[中游封裝][Packaging] KGD (Known Good Die) 裸晶堆疊、WLCSP/USON 超小型封裝,透過 Octal SPI / xSPI 介面提供高達 400MB/s 讀取頻寬與就地執行 (XiP)。Known Good Die (KGD), WLCSP/USON packaging, and Octal SPI / xSPI yielding up to 400MB/s bandwidth with execute-in-place (XiP).
[下游系統][Systems] 當具名先進邏輯製程沒有符合條件的嵌入式 Flash 選項時,可評估外部或近晶粒 NVM。ADAS、座艙及其他系統的冷開機延遲仍取決於容量、介面、控制器與驗證流程,不能只由節點名稱保證。When a named advanced logic process lacks a suitable embedded Flash option, evaluate external or near-die NVM. Cold-boot latency for ADAS, cockpit and other systems still depends on capacity, interface, controller and validation flow; the node name alone is not a guarantee.
05 · 類腦神經運算與嵌入式 RERAM05 · NEUROMORPHIC & EMBEDDED RERAM

Weebit Nano (OxRAM 阻變記憶體) Weebit Nano (OxRAM & CIM)

[上游晶圓][Foundry] SkyWater 130nm CMOS、DB HiTek 130nm BCD 與先進節點拓展評估代工平台,後段 (BEOL) 僅需 2 道額外光罩,熱預算相容標準 CMOS。SkyWater 130nm CMOS, DB HiTek 130nm BCD, and advanced node expansion; requires only 2 BEOL mask adders with standard thermal budget.
[中游 IP][Midstream IP] 金屬氧化物(HfO2/TiN)阻變隨機微絲結構,支援多階電導 (MLC / Analog Conductance Tuning),整合類比神經形態矩陣乘加運算 (Analog In-Memory Computing, CIM)。Metal-oxide (HfO2/TiN) resistive filaments supporting multi-level analog conductance for analog Computing-in-Memory (CIM) matrix operations.
[下游系統][Systems] 超低功耗邊緣 AI 語音關鍵詞識別 (KWS)、智慧物聯網感測融合、抗輻射工控系統與醫療穿戴式常時在線 (Always-On) 監測。Ultra-low-power edge AI keyword spotting (KWS), smart IoT sensor fusion, radiation-hardened industrial controls, and always-on wearables.
06 · 企業級儲存與斷電保護06 · ENTERPRISE STORAGE & PLP

Everspin (STT-MRAM 與斷電保護) Everspin (STT-MRAM & PLP)

[上游晶圓][Foundry] GlobalFoundries (22FDX / 12LP FinFET) 與 TSMC 代工製造,後段堆疊垂直磁性穿隧接面 (pMTJ),需 3~4 道額外光罩。Foundry manufacturing at GF (22FDX / 12LP FinFET) and TSMC, stacking perpendicular MTJ (pMTJ) with 3-4 BEOL mask adders.
[中游產品][Product] 獨立式晶片 (1Gb DDR4 / xSPI 介面) 與嵌入式 MRAM 巨集,具備無限次寫入耐久度 (1010~1012 次) 與奈秒級持久寫入速度。Discrete 1Gb DDR4 / xSPI chips and embedded macros offering 1010-1012 write endurance and nanosecond-speed persistence.
[下游系統][Systems] 企業級 NVMe SSD 寫入日誌快取、零電容斷電保護 (Capacitor-Free PLP)、CXL 持久性記憶體緩衝區、工業自動化 PLC 與航太抗輻射防護。Enterprise NVMe SSD write journaling, capacitor-free power loss protection (PLP), CXL persistent buffers, industrial PLCs, and rad-hard aerospace.
晶片級實體防竄改安全SILICON-LEVEL ANTI-TAMPER SECURITY

TEM 衰減與差分功耗分析 (DPA) 實體防護責任鏈(非絕對免疫) TEM Attenuation & DPA Side-Channel Layered Defense Chain (Not Absolute Immunity)

① TEM / FIB 可見性與主動屏蔽邊界① TEM / FIB Visibility & Active Shielding Boundary
AntiFuse 崩潰微絲約 3–8nm,局域包裹於標準閘極氧化層中,無表面形貌變化,較吹斷級 eFuse 熔斷空洞具備極高的物理逆向成本。I-fuse 公開為非爆破熔絲,IEEE JEDS 2019 稱編程狀態較難以 SEM 目視判讀,仍非絕對免疫。但需注意:非揮發微絲/熔絲狀態為永久性物理結構,位元單元本身無法「單週期自我清除」;系統級防禦依賴開機時將金鑰載入揮發性快取 (Volatile Shadow RAM),當頂層主動金屬屏蔽 (Active Metal Mesh) 偵測到 FIB 微探針或開路時,由控制器觸發暫存器之單週期即時零化 (Zeroization)。此為物理衰減結合系統聯防,非位元單元絕對免疫。 AntiFuse filaments are ~3–8nm localized inside standard gate oxide without surface topological changes, imposing high reverse-engineering barriers compared to blow-class eFuse voids. I-fuse is published as a non-explosive fuse; IEEE JEDS 2019 reports programmed state as hard to visually detect by SEM — still not absolute immunity. Note: the physical filament or fuse state is irreversible and cannot self-zeroize. System security relies on shadowing keys into volatile RAM at boot; when the active top-metal mesh detects FIB probing, the controller triggers instantaneous register zeroization. Hardware attenuation + architectural co-defense, not bitcell-level absolute immunity.
② 互補差動讀取(DPA/CPA 衰減邊界)② Complementary Differential Read (DPA/CPA Attenuation Boundary)
成對互補差動讀取(A=HRS, B=LRS)使兩條位元線同時抽取電流,瞬態總功耗近似常數 Itotal = IHRS + ILRS,可大幅衰減一階差分功耗 (DPA) 與相關功耗 (CPA) 訊跡。但面對高階差分功耗 (Higher-order DPA) 與局域電磁輻射分析 (EMA),晶圓佈線不對稱性仍可能洩漏側信道特徵。因此,完備防禦必須落實多層責任鏈:位元層差動對消 + 控制器層隨機遮罩 (Masking) 與時脈抖動 + 密碼引擎擾碼。 Complementary differential read pairs (A=HRS, B=LRS) draw current simultaneously on both bitlines (Itotal = IHRS + ILRS), significantly attenuating first-order DPA and CPA traces. However, against higher-order DPA and localized EMA probes, layout asymmetry can still leak side-channel signatures. Complete protection mandates a layered defense chain: bitcell-level current cancellation + controller-level algorithmic masking/jitter + cryptographic engine bus scrambling.
先進邏輯節點 · 3D 靜電幾何與穿隧微縮 ADVANCED LOGIC NODES · 3D ELECTROSTATICS & TUNNELING SCALING

先進節點 FinFET / GAA 奈米片 AntiFuse 延伸性與量子穿隧試算器 Advanced FinFET / GAA NanoSheet AntiFuse Scalability & Quantum Tunneling Simulator

架構比較 · 未綁定具名巨集規格 Architecture Comparison · No Named Macro Specification
⚠️ 示意模擬模型與技術邊界說明:⚠️ Illustrative Model & Technical Boundary Notice: 以固定角隅係數乘上 E = V / t,並以 16.5 MV/cm 臨界場強試算硬擊穿電壓;穿隧使用簡化指數與溫度倍率。這是未經實測校準的教學模型,沒有求解三維 Poisson 或 TCAD。電荷泵級數依假設壓降試算;面積比例 0.28/0.42/0.65/1.00 為獨立預設輸入,72% 等縮減值不是由級數或實際佈局算得。不能保證 0-mask 相容、周邊耐壓、車規資格、保持或讀取擾動表現。 Multiplies E = V / t by a fixed corner coefficient and estimates breakdown using an assumed 16.5 MV/cm critical field. Tunneling uses a simplified exponent and temperature multiplier. This uncalibrated teaching model does not solve 3D Poisson or TCAD. Pump stages use an assumed voltage drop; footprint ratios 0.28 / 0.42 / 0.65 / 1.00 are independent preset inputs, so reductions such as 72% are not calculated from stages or physical layout. Results do not establish mask compatibility, peripheral voltage safety, automotive qualification, retention or read-disturb performance.

公開架構對照:TSMC N3/N3E 為 FinFET,N2 為奈米片;此對照不驗證本工作台的 OTP 電性或面積假設。Published architecture context: TSMC N3 / N3E uses FinFET; N2 uses nanosheets. This context does not validate the workbench OTP electrical or footprint assumptions. TSMC N3 · TSMC N2

0.70 V
0.4 V4.5 V8.5 V
125 °C
25 °C105 °C175 °C
一維基礎電場 E1D1D Field E1D
3.33 MV/cm
V / tphys
3D 角隅增強場 Ecorner3D Corner Field
4.07 MV/cm
κ · E1D (+22%)
預測擊穿電壓 VbdBreakdown Vbd
2.84 V
相較 28nm -60%-60% vs 28nm
電荷泵級數需求Pump Stages
6 級泵Stages
Dickson 升壓鏈ladder
假設電荷泵面積縮減Assumed Pump Area Savings
-72 %
對比 28nm 基準vs 28nm Base
待機穿隧漏電 JDTTunneling Leakage
2.4e-6 A/cm²
WKB 穿隧模型WKB model
⚡ 【GAA 教學試算】既有預設的硬擊穿試算為 2.84V,電荷泵試算為 6 級;面積比例假設 0.28 對應 72% 縮減,與級數計算獨立。這些數值尚未校準,不能保證製程相容或資格驗證通過。 [GAA Teaching Estimate] Existing presets give 2.84V breakdown and an estimated 6-stage pump. The assumed footprint ratio 0.28 gives 72% reduction independently of the stage calculation. These uncalibrated values do not establish process compatibility or qualification.
先進封裝 PDK 與 D2D 互連工作台 ADVANCED PACKAGING PDK & D2D WORKBENCH

2.5D/3D 邏輯 Base Die 互連延遲、熱阻階梯與晶圓測試模擬器 2.5D/3D Logic Base Die Interconnect Latency, Thermal Gradient & Wafer Test Simulator

測試狀態:TEST RATING: QUALIFIED
⚠️ 示意模擬模型與技術邊界說明:⚠️ Schematic Model & Technical Boundary Notice: 本工作台之互連延遲 (Elmore RC)、熱阻階梯 (1D Fourier 導熱) 與訊號完整性眼高開展率為基於典型製程設計套件 (PDK) 幾何常數與一階集總參數之工程示意模型,非三維電磁場 (3D EM) 或熱流場 (CFD) 有限元素數值求解。晶粒實際互連品質與眼圖開展裕度需依據晶圓代工廠 sign-off 萃取規則檔與完整封裝 S 參數時域模擬判定。 The interconnect latency (Elmore RC), thermal gradients (1D Fourier heat conduction), and eye opening rates in this workbench are schematic engineering models based on typical PDK geometries and first-order lumped parameters, rather than full 3D electromagnetic (EM) or computational fluid dynamics (CFD) finite element solvers. Actual chiplet SI/PI margins must be verified through foundry sign-off extraction rule decks and full-channel package S-parameter simulations.
3.5 mm
450 W
8.0 Gbps
D2D RC 傳輸延遲 τD2D RC Delay τ
2.8 ps
Elmore τd2d
單線頻寬容量Per-Wire Bandwidth
8.0 Gbps/wire
訊號完整性邊界Signal Margin
垂直熱阻接面溫升 ΔTjJunction Temp Rise ΔTj
+81.0 °C (0.18 K/W)
熱阻階梯模型Thermal Ladder
Base Die 0-Mask eNVM 預算Base Die eNVM Budget
32 Kb OTP
修復+校準+ID總預算Repair + Trim + ID
動態圖表模式:D2D 互連延遲 (ps) 與熱阻溫升梯度分析 DYNAMIC CHART: D2D Interconnect Latency (ps) & Thermal Gradient
TSMC CoWoS-S 具備成熟的微凸塊互連技術,封裝後微凸塊修復可由 Base Die 內建之 0-Mask AntiFuse OTP 儲存 BIRA 重映射向量;高並行晶圓測試則需由微間距 MEMS 超高針數邏輯測試介面(如高密度微懸臂或垂直導向探針架構)支援。 TSMC CoWoS-S features mature micro-bump interconnects, where post-packaging repair can utilize Base Die 0-Mask AntiFuse OTP for BIRA redundancy remapping vectors; highly parallel wafer sort is supported by fine-pitch MEMS ultra-high-pin-count logic test interfaces (such as high-density micro-cantilever or vertical guide probe architectures).
05 · 晶片級讀取拓樸05 · ARCHITECTURAL READ TOPOLOGIES

四大晶片級讀取模式:感測邊界、面積開銷與側信道衰減全景 Four Architectural Read Modes: Sense Margins, Area Overhead & Side-Channel Attenuation

非揮發性記憶體(OTP、MTP、eFlash、MRAM)在微架構層級的讀取拓樸,直接決定了晶片的感測餘裕(Sense Margin)、PVT 溫漂容忍度、位元胞面積開銷(Area Overhead)與一階差分功耗分析(DPA/CPA)側信道防護邊界。下圖展示四種抽象讀取電路拓樸:從單端參考感測,到僅可能針對明確模型涵蓋的單點失效提供有限容錯的互補差動冗餘架構(不可視為零缺陷認證)。 At the microarchitectural level, the non-volatile memory read topology dictates the sense margin, PVT drift resilience, bitcell area overhead, and first-order DPA/CPA side-channel attenuation boundary. The architectural sketches below classify four read topologies — from single-ended reference sensing to differential+redundant schemes that may tolerate modeled single-point faults (not zero-defect certification). 教學邊界:固定 dB、0-ppm、175°C 或 ASIL/CC 等級僅能在具名產品/FMEDA 出現,不得由架構圖直接推出。 Teaching boundary: Fixed dB, 0-ppm, 175°C, or ASIL/CC tiers require named products/FMEDA — not inference from cartoons alone.

物理紀律:PHYSICAL DISCIPLINE: 嚴格正交直角電路拓樸 · 100% 杜絕對角線與視差錯位 Strict 90° Orthogonal Schematics · Zero Diagonal Parallax 實證連結:EVIDENCE LINK: 記憶體微觀讀取物理 ↗ Memory Read Physics ↗ & 一手差動衰減總帳 (V07) ↗ Primary Ledger V07 ↗

01 · 單端讀取 (Single-Ended) 01 · Single-Ended Read

1T 單元 · 外部參考源 1T Cell · External Ref
LOGICAL BIT (1-CELL)邏輯位元(單單元) Physical Bit 1實體位元 1 1T Vref / Iref Bandgap Ext外部能隙基準 BL (I_cell) Vref + - SENSE感測 AMP放大器 ΔV = V_BL - V_ref 1ST DPA TRACE一階 DPA 波形 Asymmetric Peak非對稱峰值 DATA OUT (0/1)資料輸出(0/1)

運作原理:單一實體位元胞(1-Cell)的讀取電流或電壓直接輸入感測放大器正端,與參考單元、能隙或其他片上產生器提供的固定參考(Vref/Iref)比較以判定 0 或 1。 Operating Principle: A single physical bitcell current or voltage feeds into the positive sense amplifier input, compared against a fixed reference (Vref/Iref) from a reference cell, bandgap, or other on-chip generator to resolve 0 or 1.

單元面積AREA 1.0×(基準)1.0× (MIN)
參考基準REF SOURCE 參考單元/Vref 產生器Ref cell / Vref generator
一階 DPA 衰減1ST DPA ATTEN 無衰減(易受攻擊)None (Vulnerable)
  • 優勢:矽面積開銷最低,位元線路由最為簡潔,讀取延遲低且功耗最小。Pros: Minimal silicon area footprint, simplest bitline routing, lowest latency and read power.
  • 局限:感測餘裕受製程、電壓與溫度(PVT)溫漂劇烈影響;資料為 0 與 1 的電流差異極大,一階 DPA/CPA 訊跡直觀可辨。Cons: Sense margin suffers from PVT drift; large current difference between 0 and 1 creates intuitive first-order DPA/CPA traces.
  • 典型應用:高密度類比 Trim 校準碼、晶圓 ID(Wafer ID)、非安全敏感的標準產品配置暫存器。Target Applications: High-density analog trim calibration, Wafer ID, non-sensitive standard configuration registers.

02 · 單端並聯冗餘 (Redundant Read) 02 · Redundant Single-Ended Read

≥2 並聯 · 外部參考源 ≥2 Parallel · External Ref
LOGICAL BIT (PARALLEL)邏輯位元(並聯) Physical Bit 1 (Pri)實體位元 1(主要) Physical Bit 2 (Red)實體位元 2(冗餘) BL (1||2) Vref / Iref Bandgap Ext外部能隙基準 Vref + - SENSE感測 AMP放大器 Defect Tolerant缺陷容忍 YIELD SHIELD良率保護 Filament Bypass細絲旁路 FAULT-SAFE OUT故障安全輸出

運作原理:在同一條位元線上並聯兩個或更多同極性實體單元(Parallel Replicas),共同注入感測放大器輸入端並與 Vref 比較。只要其中一個單元成功導通即可建立狀態。 Operating Principle: Two or more same-polarity physical cells are connected in parallel to drive the sense amplifier input against Vref. Programmed state is established as long as any single parallel element conducts.

單元面積AREA 2.0× ~ 3.0×
參考基準REF SOURCE 外部能隙基準 (Bandgap)External Bandgap
一階 DPA 衰減1ST DPA ATTEN 無衰減(同極性疊加)None (Same Polarity)
  • 優勢:極大程度容忍氧化層擊穿隨機未完全軟失效(Soft Breakdown)或製造隨機缺陷,大幅改善晶圓成品率(Yield)。Pros: Highly tolerant of incomplete dielectric soft breakdown and random defects, significantly boosting yield.
  • 局限:實體面積開銷翻倍;仍然受全域 Vref 溫漂綁定;同極性並聯對一階 DPA 側信道並無衰減,甚至放大電流峰值。Cons: Doubles silicon area; still bound to global Vref drift; same-polarity parallel elements do not attenuate first-order DPA traces.
  • 典型應用:記憶體內建自我修復(BISR)備援重映射暫存器、車規級一次性開機安全旗標。Target Applications: Built-in self-repair (BISR) remap fuseboxes, automotive-grade boot flags.

03 · 互補差動讀取 (Differential Read) 03 · Complementary Differential Read

BL 對 /BL · 零外部參考 BL vs /BL · Zero External Ref
LOGICAL BIT (COMP PAIR)邏輯位元(互補對) Physical Bit 1 (True)實體位元 1(正相) State: D [0/1]狀態:D [0/1] ZERO VREF (Self-Ref)零 VREF(自我參考) Physical Bit 2 (Comp)實體位元 2(互補) State: /D [1/0]狀態:/D [1/0] BL (True)BL(正相) /BL (Comp)/BL(互補) + - DIFF差動 AMP放大器 ΔV = V_BL - V_/BL DPA FLAT WAVEDPA 平坦波形 I_tot ≈ Const CMRR (qualitative)CMRR(定性) SECURE BIT OUT安全位元輸出

運作原理:每個邏輯位元由成對的互補實體單元編碼(真值單元與補值單元必有一通一斷)。差動感測放大器直接量測 BL 與 /BL 之電壓差,完全自洽比較,無需外部參考電壓。 Operating Principle: Each logical bit is encoded across a complementary cell pair (one programmed, one unprogrammed). A differential sense amp compares BL against /BL directly, eliminating external references.

單元面積AREA 2.0× ~ 2.2×
參考基準REF SOURCE 自洽比較(零外部 Vref)Self-Referenced (0 Vref)
一階 DPA 衰減1ST DPA ATTEN 定性共模抑制(具名案例)Qualitative CMRR (named cases)
  • 優勢(原理):互補對可抑制部分共模漂移;雙側電流抽取有助於降低一階 DPA 訊跡 — 實際 dB 與溫漂免疫需具名巨集量測。Pros (principle): Complementary pairs attenuate some common-mode drift; dual-rail current draw can reduce first-order DPA traces — actual dB and drift immunity need named macro data.
  • 局限:位元胞與解碼路由佔用約 2 倍面積;面對高階 DPA 或局域電磁輻射(EMA),佈局微觀不對稱仍需搭配頂層屏蔽網與演算法遮罩。Cons: Doubles cell and routing area; higher-order DPA or localized EMA can exploit physical layout asymmetry unless paired with top metal shields.
  • 典型應用:AES-256 根金鑰儲存、PKI 設備身分憑證、安全開機第一階段向量(Stage 1 Boot Vector)。Target Applications: AES-256 cryptographic root keys, PKI device certificates, Stage-1 secure boot trust anchors.

04 · 互補差動並聯冗餘 (Diff + Redundant) 04 · Differential + Redundant Read

4T 四元陣列 · 教學冗餘 · 零外部參考 4T Cell · Teaching redundancy · Zero-Ref
LOGICAL BIT (4T QUAD-MATRIX)邏輯位元(4T 四單元矩陣) D1 (Pri)D1(主要) D2 (Red)D2(冗餘) BL (1||2) /D1 (Pri)/D1(主要) /D2 (Red)/D2(冗餘) /BL (1||2) + - MAX-REL最高可靠度 DIFF AMP差動放大器 Dual Bypass雙旁路 NAMED-CASE ONLY僅限特定案例 Mission-Critical關鍵任務 XSC Special RecipeXSC 特定製程配方 TEACHING OUTPUT教學輸出

運作原理:互補真值軌與補值軌各自配置 ≥2 個單元並聯,並以差動放大器對消自洽感測 — 針對已建模的單點失效可能提供容錯;未建模的共通路徑、短路或匹配失效仍可能影響讀取。 Operating Principle: True and complementary rails each use ≥2 parallel bitcells into a differential sense amplifier — may tolerate modeled single-cell faults; unmodeled common-path, short, or mismatch failures can still affect readout.

單元面積AREA 4.0× ~ 4.5×
參考基準REF SOURCE 自洽比較(零外部 Vref)Self-Referenced (0 Vref)
一階 DPA 衰減1ST DPA ATTEN 冗餘+差動(具名案例)Redundant + diff (named cases)
  • 優勢(原理):差動對加上並聯冗餘可針對已建模的單點失效提供容錯;不等同 0-ppm、175°C 認證或 ASIL/EAL 資格 — 需具名 FMEDA/產品條件。Pros (principle): Differential pairs plus parallel redundancy can tolerate modeled single-point faults — not equivalent to 0-ppm, 175°C certification, or ASIL/EAL qualification without named FMEDA/product bounds.
  • 局限:面積開銷最高(≥4 倍單元面積),編程時間與燒錄電流預算增加。Cons: Highest area penalty (≥4x single cell footprint); longest programming time and write energy budget.
  • 典型應用(需另證):高可靠車規/金融 RoT 等可能採用類似拓樸;ASIL 等級與 CC 評估須對應具名產品與系統整合。Example domains (evidence required): Automotive/financial RoT designs may use similar topologies; ASIL tier and CC evaluation require named products and system integration.
NVM 四大晶片級讀取模式工程維度對比矩陣 Four Silicon Read Modes Engineering Trade-Off Matrix
架構維度Metric 01 · 單端讀取 (Single)01 · Single-Ended 02 · 單端冗餘 (Redundant)02 · Redundant 03 · 互補差動 (Differential)03 · Differential 04 · 差動+冗餘 (Diff+Red)04 · Diff + Redundant
實體單元數 / bitPhysical Cells / bit 1 Cell1 個單元 ≥2 Cells (同極性並聯)≥2 Cells (Parallel) 2 Cells (互補對稱)2 Cells (Complementary) ≥4 Cells (雙軌各並聯)≥4 Cells (Dual-Rail Parallel)
外部參考源依賴External Ref Requirement 依賴帶隙 Vref/IrefRequires Bandgap Vref/Iref 依賴帶隙 Vref/IrefRequires Bandgap Vref/Iref 0 外部依賴 (對內互比)Zero External Ref (Self-Referenced) 0 外部依賴 (對內互比)Zero External Ref (Self-Referenced)
感測餘裕與溫漂抗性Sense Margin & PVT Drift 受製程/電壓/溫度漂移牽引Prone to PVT drift 單元缺陷容錯提升,溫漂仍存Higher defect margin; reference drift persists 差動自洽,大幅抵消全域溫漂Self-referencing; common-mode drift cancelled 極致強韌(最高感測餘裕)Maximum Margin (Ultra-robust)
相對面積開銷Silicon Area Overhead 1.0× (基準基準線)1.0× (Baseline) 約 2.0× ~ 2.5×~2.0× to 2.5× 約 2.0× ~ 2.2×~2.0× to 2.2× 約 4.0× ~ 4.5×~4.0× to 4.5×
一階 DPA/CPA 側信道1st-Order DPA/CPA Leak 無衰減(訊號直觀可辨)No attenuation (Direct signature) 無衰減(同向充放電)No attenuation (Common charge curve) 高衰減(雙軌同時抽載對消)High attenuation (Complementary draw cancellation) 極高衰減(對消 + 冗餘平滑)Extreme attenuation (Cancellation + Smoothing)
隨機缺陷與未擊穿容錯Random Defect Tolerance 無(單元軟失效即報廢)None (Single point of failure) 高(任一單元導通即成立)High (Tolerates single cell failure) 中等(需確保兩單元均良好)Moderate (Requires both cells functional) 最高(容忍單側單元隨機失效)Maximum (Tolerates single-side defects)
適用架構等級Recommended Tier 類比 Trim、晶圓 ID、配置位元Analog trim, Wafer ID, config bits BISR 修復、工規韌體指標BISR repair, industrial boot flags 晶片硬體信任根、AES 金鑰Hardware Root of Trust, AES keys 高可靠 RoT 候選(須具名 FMEDA/產品)High-assurance RoT candidate (named FMEDA/product)
架構工程註記與引證邊界:Engineering Notes & Claim Boundaries: 上述拓樸均為晶片系統架構級抽象模型,非特定單一具名商標 IP 之專屬資料手冊。差動讀取對 DPA/CPA 之衰減效能係基於雙軌對稱抽載的一階電流互補物理(參考 Synopsys 公開架構與本站一手證據總帳 V07 ↗)。面對高階差分功耗(Higher-order DPA)與局域微探針電磁輻射(EMA),必須配合頂層主動金屬屏蔽網(Active Metal Mesh)與演算法遮罩(Algorithmic Masking)落實多層縱深防禦。 The four topologies represent system architecture-class abstract models rather than a proprietary datasheet from a specific IP vendor. Differential DPA/CPA attenuation stems from first-order complementary current draw physics (refer to Synopsys public disclosure & Hub Primary Evidence Ledger V07 ↗). Against higher-order DPA and localized EMA, silicon architects must deploy layered defenses incorporating active top metal shields and algorithmic masking.
06 · 3D 混合鍵合與先進封裝06 · 3D HYBRID BONDING & PACKAGING

Cu-Cu 混合鍵合熱應變、界面空洞與 eNVM 禁制保留區 (KOZ) 物理分析 Cu-Cu Direct Hybrid Bonding Thermal Stress, Nanovoids & eNVM Keep-Out Zone (KOZ) Physics

探索晶圓堆疊 (Wafer-on-Wafer / Chip-on-Wafer) 的 CMP 銅微凹陷、熱膨脹、退火殘留應力與 eNVM 保留區;模型用於教學,不代表具名產品的製程規格。 Explore CMP copper dishing, thermal expansion, residual annealing stress and eNVM keep-out zones in WoW/CoW stacking. The model is educational and does not specify a named product process.

⚠️ 示意模擬模型與技術邊界說明:⚠️ Illustrative Model & Boundary Notice: 本模擬器的數值與評級是教學假設,未以具名製程或產品量測校準。應力對電性的影響取決於元件方向、偏壓、材料與幾何;不能用單一係數保證臨界電壓變化、KOZ 或可靠度。正式設計需採用代工廠/IP 規則與驗證資料;本模型不構成 JEDEC、代工廠或產品資格認證。 Simulator values and ratings are uncalibrated teaching assumptions, not named-process or product measurements. Electrical response to stress depends on device orientation, bias, materials and geometry; one coefficient cannot guarantee threshold shifts, KOZ or reliability. Design decisions require foundry/IP rules and validation data. This model is not JEDEC, foundry or product qualification.
混合鍵合爆炸圖Exploded Hybrid Bond2.5D 等角示意 · 由下而上編號2.5D isometric sketch · numbered bottom to top
1234
  1. 1下晶粒矽基板Bottom die silicon
  2. 2下晶粒介電層與銅墊Bottom die dielectric and Cu pads
  3. 3上晶粒介電層與銅墊(翻面朝下)Top die dielectric and Cu pads (flipped face down)
  4. 4上晶粒矽基板Top die silicon

兩片晶粒面對面:介電層對介電層先接合,銅墊對銅墊在退火後接合。銅墊尺寸、間距與厚度未依比例。Two dies face to face: dielectric bonds to dielectric first, and the Cu pads bond after anneal. Pad size, pitch and thickness are not to scale.

eNVM 技術候選eNVM Technology 整合層級Integration Tier 最高容許退火溫 (T_max)Max Anneal Temp (T_max) 殘留熱應力耐受 (σ_limit)Stress Limit (σ_limit) 推薦最小 KOZ 邊界Recommended Min KOZ 應力誘發電性變化Stress-Induced Electrical Changes 3D 堆疊適用評級3D Stacking Suitability
AntiFuse OTP(依 IP 結構)AntiFuse OTP (IP-specific structure) 依具名 IP 堆疊與整合流程確認Confirm against the named IP stack and integration flow 需具名製程與 IP 熱預算資料Named-process and IP thermal-budget data required 需指定堆疊的應力模擬與量測Stack-specific stress simulation and measurements required 需具名製程與 IP 版圖規則Named-process and IP layout rules required 需元件方向、偏壓與量測條件Device orientation, bias and measurement conditions required 尚未驗證;依指定產品判定Unverified; evaluate for the specified product
STT-MRAM (垂直 MTJ)STT-MRAM (Perpendicular MTJ) 依具名 IP 堆疊與整合流程確認Confirm against the named IP stack and integration flow 需具名製程與 IP 熱預算資料Named-process and IP thermal-budget data required 需指定堆疊的應力模擬與量測Stack-specific stress simulation and measurements required 需具名製程與 IP 版圖規則Named-process and IP layout rules required 需元件方向、偏壓與量測條件Device orientation, bias and measurement conditions required 尚未驗證;依指定產品判定Unverified; evaluate for the specified product
Oxide ReRAM (氧空位微絲)Oxide ReRAM (Oxygen Vacancy) 依具名 IP 堆疊與整合流程確認Confirm against the named IP stack and integration flow 需具名製程與 IP 熱預算資料Named-process and IP thermal-budget data required 需指定堆疊的應力模擬與量測Stack-specific stress simulation and measurements required 需具名製程與 IP 版圖規則Named-process and IP layout rules required 需元件方向、偏壓與量測條件Device orientation, bias and measurement conditions required 尚未驗證;依指定產品判定Unverified; evaluate for the specified product
Floating-Gate / CT eFlashFloating-Gate / CT eFlash 依具名 IP 堆疊與整合流程確認Confirm against the named IP stack and integration flow 需具名製程與 IP 熱預算資料Named-process and IP thermal-budget data required 需指定堆疊的應力模擬與量測Stack-specific stress simulation and measurements required 需具名製程與 IP 版圖規則Named-process and IP layout rules required 需元件方向、偏壓與量測條件Device orientation, bias and measurement conditions required 尚未驗證;依指定產品判定Unverified; evaluate for the specified product
先進封裝工程評估結論:Advanced Packaging Engineering Finding: 熱機械應力、熱預算與保留區 (KOZ) 取決於材料堆疊、鍵合流程、元件與 IP 版圖。上表列出待評估維度;未取得具名製程資料的項目,不構成通用設計規則,也不能用儲存機制直接判定 AntiFuse 必然優於 eFlash。 Thermomechanical stress, thermal budget and keep-out zones (KOZ) depend on materials, bonding flow, devices and IP layout. The table identifies evaluation dimensions. Entries without named-process evidence are not design rules, and the storage mechanism alone does not establish an AntiFuse advantage over eFlash.

⚡ HBM4 邏輯 Base Die、Cu-Cu 鍵合與 TSV 備援:證據與設計邊界 HBM4 Logic Base Die, Cu-Cu Bonding and TSV Redundancy: Evidence and Design Boundaries

TSMC SoIC 官方頁面說明鍵合間距從小於 10 μm 的規則起步;SK hynix 的 2025 HBM4 公告採 Advanced MR-MUF。這些資料不能支持所有 HBM4 均採次微米混合鍵合的推論。熱膨脹差異可形成應力,但局域應力、電晶體臨界電壓變化與 KOZ 必須依指定堆疊模擬及量測;本頁未建立通用 MPa、mV/MPa 或 μm 門檻。BIRA/BISR 與 NVM 保存修復映射屬架構評估,不能據此保證 TSV/DRAM 自主修復、免 ATE 或零額外光罩。 TSMC describes SoIC bond pitch as starting from a sub-10 μm rule; SK hynix disclosed Advanced MR-MUF for its 2025 HBM4. Neither establishes universal sub-micron hybrid bonding in HBM4. Thermal expansion mismatch can induce stress, but local stress, transistor threshold shifts and KOZ require stack-specific simulation and measurements; no universal MPa, mV/MPa or μm limits are established here. BIRA/BISR with NVM repair-map retention is an architecture evaluation, not evidence of autonomous TSV/DRAM repair, ATE elimination or zero additional masks.

官方來源:Official sources: TSMC SoIC · SK hynix HBM4 (2025)

07 · 架構師決策管線07 · ARCHITECT PIPELINE

三步收斂:架構師 NVM 決策簽核管線 Three-Stage Architectural NVM Selection Pipeline

GATE 1 · AUDIT關卡 1 · 稽核 STATE CONTRACT & LIFE-CYCLE狀態契約與生命週期 SYS SPECS系統規格 Requirements需求 WRITE-ONCE一次寫入 Immutable ID / Trim不可變更識別碼/微調 0-MASK OTP零額外光罩 OTP AntiFuse · I-fuse FG-OTP SplitFG-OTP 分類 MULTI-WRITE多次寫入 Firmware OTA / Patch韌體 OTA/修補 LOGIC MTP邏輯 MTP FN/FN · TwinBit eFlash OptioneFlash 選項 CRITERION: 0-Overhead Retention · Zero Bus Leak準則:零額外資料保存成本 · 零匯流排洩漏 GATE 2 · BUDGET關卡 2 · 預算 PROCESS NODE & THERMAL製程節點與熱預算 FOUNDRY PDK晶圓代工 PDK Named node / device stack具名節點/元件堆疊 SST eFlash / Single-Poly 28 nm: CHECK PDK28 nm:核對 PDK No universal cutoff非普遍截止線 0-MASK ADVANCED零額外光罩先進製程 TwinBit (22nm) · I-fuse (22FDX) ZA (sub-10nm dielectric)ZA(小於 10nm 的介電層) BEOL / LOW-ADDERBEOL/少量額外光罩 STT-MRAM (3~5 Adders)STT-MRAM(額外 3~5 道光罩) SONOS G1/G2 (+2~4 Adders)SONOS G1/G2(額外 2~4 道光罩) CRITERION: Named stack · supply · thermal budget準則:具名堆疊 · 供電 · 熱預算 GATE 3 · SIGN-OFF關卡 3 · 簽核 PHYSICAL VERIFICATION實體驗證 AEC-Q100 GRADE 0 175°C Tj · HTOL 1000h ELFR 2,400 units c=0 LAYERED ANTI-TAMPER分層防竄改 TEM Filament Attenuation降低 TEM 細絲訊號 Active Top-Mesh Zeroize主動頂層網格清除 TAPE-OUT RELEASE下線發布 APPROVED已核准 Silicon Proven Closure晶片實測結案 CRITERION: TEM Attenuation · Grade 0 Validation
STEP 01步驟 01 關卡 1 · 契約審查 GATE 1 · AUDIT

狀態契約與生命週期審查 State Contract Audit

區分不可變身分 (Immutable ID)、校準邊界 (Calibration) 與韌體代碼。若資料僅需工廠寫入或偶發修補,優先評估 0-mask OTP,並分流 AntiFuse(CFX 教學為閘氧崩潰;Floadia ZA 為介電層永久導通、擊穿位點未公開)、I-fuse 熱輔助電遷移、以及浮閘 OTP;不得預設「OTP = AntiFuse」。韌體 OTA 再評估 Logic MTP(FN/FN、CHI/FN、TwinBit Schottky 三分流)或 eFlash。 Distinguish immutable identity from firmware. For write-once or bounded patch models, evaluate 0-mask OTP and split AntiFuse (CFX teaching default is GOX; Floadia ZA is dielectric conduction with unpublished breakdown site), I-fuse heat-assisted EM, and floating-gate OTP — do not default “OTP = AntiFuse”. Firmware OTA then evaluates Logic MTP (FN/FN vs CHI/FN vs TwinBit Schottky) or eFlash.

簽核準則:零開銷數據保持 · 匯流排零洩漏 CRITERION: 0-Overhead Retention · Zero Bus Leak
STEP 02步驟 02 關卡 2 · 預算評估 GATE 2 · BUDGET

製程微縮與熱預算可行性 Process Node & Thermal Budget

從晶圓代工廠目前的 PDK 與授權巨集清單開始,逐項核對單元堆疊、光罩、供電、熱預算及資格報告。28 nm 不是通用排除線;平面浮閘、SONOS、Schottky、OTP 與 BEOL 記憶體也不能互套條件。具名製程範例;具名元件的指標條件。 Start with the foundry's current PDK and licensed macro list. Check device stack, masks, supply, thermal budget and qualification reports individually. 28 nm is not a universal exclusion line; do not transfer conditions across planar floating gate, SONOS, Schottky, OTP and BEOL memory. Named process example; conditions for named devices.

簽核準則:具名巨集的元件堆疊 · 供電 · 熱預算 CRITERION: Named macro device stack · supply · thermal budget
STEP 03步驟 03 關卡 3 · 實體驗證 GATE 3 · CHECKLIST

實體驗證與防禦簽核 Physical Verification Checklist

依據 AEC-Q100 Grade 0(Ta −40°C 至 +150°C)測試流程驗證高溫工作壽命 (HTOL 1000h),其接面溫度應力須由特定資格驗證計畫訂定(例如 175°C Tj 目標),並驗證早期失效率 (ELFR 2400 顆 c=0) 與高溫數據保持烘烤 (Retention Bake),並評估 TEM 逆向防護邊界,確保微絲歐姆接觸無熱漂移退化。 Qualify per the AEC-Q100 Grade 0 test flow (Ta −40°C to +150°C), with HTOL 1000h at the junction stress specified by a named qualification plan (e.g. a 175°C Tj target), early life failure rate (ELFR 2,400 units c=0), and data retention bake, alongside TEM anti-tamper boundary analysis to assess ohmic-drift risk (architecture-class; HTOL/ELFR depend on target qualification).

簽核準則:TEM 物理衰減(非免疫)· Grade 0(Ta −40~+150°C)檢核 · 175°C Tj 依特定規格書 CRITERION: TEM Attenuation (Not Immunity) · Grade 0 (Ta −40~+150°C) Checklist · 175°C Tj per Named Datasheet