軒

無聊看看

[駿HaYaO @QQ_Timmy]

電子級氫氟酸產業
1. 產業背景：電子級氫氟酸在晶圓製造中的角色
電子級氫氟酸是半導體晶圓製造過程中關鍵的濕式電子化學品，主要用於清洗與蝕刻兩大核心工序，是單位晶圓耗量最大的濕式電子化學品之一。

清洗環節：氫氟酸用於去除晶圓表面的自然氧化層、金屬離子污染及顆粒殘留。隨著製程從 28nm 向 7nm 及以下演進，清洗步驟��從約 100 步增加至 200 步以上。在 3nm 製程中，由於極紫外光刻（EUV）對光刻膠表面缺陷的敏感度更高，清洗步驟進一步增至 300 步以上。每一步清洗均需使用稀釋氫氟酸（DHF）或緩衝氧化蝕刻液（BOE）。單片 12 吋晶圓在先進製程廠的耗酸量較 28nm 廠高出約 2-3 倍。

蝕刻環節：氫氟酸用於二氧化矽及氮化矽的選擇性蝕刻。在 3D NAND 快閃記憶體中，隨著堆疊層數從 128 層向 176 層、232 層乃至 300 層以上邁進，高深寬比蝕刻（High Aspect Ratio etch，簡稱 HAR etch）的次數呈指數級增長。更高的堆疊層數意味著更深的溝槽與孔洞，對氫氟酸的純度、蝕刻速率及蝕刻選擇比提出更嚴格的要求。

AI 算力拉動方面：AI 訓練與推理晶片（如 H100、B200、MI300 系列）採用先進製程與大尺寸晶片設計，單片晶圓面積較通用邏輯晶片更大，等效耗酸量相應提升。隨著 AI 資料中心資本開支持續擴張，台積電、三星、英特爾等頭部晶圓廠的先進製程產能利用率維持高位，直接拉動濕式電子化學品的消耗量成長。
根據 SEMI 數據，2025 年全球半導體設備支出約 1255 億美元，中國區占比持續提升，晶圓廠擴產是濕式電子化學品需求成長的底層驅動。

2. 邊際變化：價格、供給與認證
2.1 價格端
根據百川盈孚數據，2026 年 6 月 29 日相較年初，UP 級（對應 G3-G4）電子級氫氟酸價格上漲約 19%，UPS 級（對應 G5）上漲約 17%。價格上漲來自兩端：成本端螢石、硫酸價格上行形成支撐；需求端晶圓廠擴產帶動耗酸量提升，而日系供應商擴產節奏緩慢，供需邊際趨���。

2.2 供給端事件
據韓媒 The Elec 報導（2026.5.14），韓國半導體材料企業所需無水氫氟酸約 90% 從中國進口，2026 年 5 月以來出現高價搶貨現象。該事件雖未構成「斷供」，但反映出日系產能瓶頸下，中國已通過認證的 G5 級供應商在供給端地位的提升。

2.3 認證進展
中國企業中，多氟多 G5 級電子級氫氟酸已向台積電、三星電子、華虹半導體、長鑫存儲等主流晶圓廠批量供貨。

3. 產業格局：日系主導與中國國內突破
日系主導格局：高端電子級氫氟酸（G5 級及以上）長期由日本企業主導，Stella Chemifa 是全球少數具備 G6 級電子級氫氟酸量產能力的企業，森田化學在 G5 級市場佔據主要份額。兩家日系企業均無在華規模化產能，中國下游廠商的電子級氫氟酸採購依賴進口或貿易商轉口。

中國突破難點體現在三個面向：
一、純度控制：根據 SEMI G5/G6 標準，G5 級要求金屬離子含量低於 10ppt，G6 級進一步要求低於 1ppt，對原料選型、設備材質、潔淨環境的要求極為嚴苛。

二、設備材質：電子級氫氟酸具有強腐蝕性，儲罐、管道、泵閥等接觸部件需採用 PFA（全氟烷氧基樹脂）等高純氟材料，以防止金屬離子析出造成二次污染。

三、批次一致性：半導體廠要求批次間的金屬離子含量、顆粒度、蝕刻速率等指標波動控制在極小範圍內，認證週期通常為 12-24 個月，新進入者從送樣到批量供貨需要較長時間的工藝磨合。
https://t.co/b2honkHLOF

[投研荟 @freearkshaw]

【电子特气与电子大宗气体全景分析报告|行业专题】-国信证券
核心要点：
【含氟电子特气横向比较】

1、技术壁垒与市场价值分层

第一梯队：六氟丁二烯、六氟化钨
第二梯队：八氟环丁烷、八氟丙烷
第三梯队：六氟乙烷、三氟化氮
第四梯队：四氟化碳、六氟化硫。

2. 国产替代进度梯队
已基本实现自给：四氟化碳、六氟化硫、三氟化氮
加速替代中：八氟环丁烷、六氟乙烷
初步突破、替代空间大：八氟丙烷、六氟化钨
单点突破、独家供给：六氟丁二烯

3. 未来需求增长弹性排序
六氟丁二烯(高端刻蚀) > 六氟化钨(沉积) > 八氟环丁烷(高端刻蚀) > 八氟丙烷(中端刻蚀) > 三氟化氮(清洗) > 六氟乙烷(基础刻蚀) > 四氟化碳 (基础刻蚀) > 六氟化硫(清洗)

4.产品价格排序
六氟丁二烯 > 六氟化钨> 八氟环丁烷> 八氟丙烷 >六氟乙烷> 三氟化氮 > 四氟化碳 > 六氟化硫

主要上市公司：【中船特气】【昊华科技】【中巨芯】【金宏气体】【华特气体】

[Elihu（半導体） @I5uHR1aNk9aocRn]

AIに作らせた半導体成膜装置カオスマップ（PVD、CVD、ALD…）

デバイスごとの区分はほとんど意味ないんだろうなぁ。膜種ごとでプラットフォーム名とプロセスモジュール名がわかると嬉しいんだが。

[pepper 花椒 (赚钱版) @off_thetarget]

康宁的GlassBridge技术，近期在光通信圈里引起了不少讨论，

简单来说，这是一套基于玻璃晶圆的光纤到光子芯片（PIC）���接方案，核心价值在于让CPO/NPO架构里最头疼的“光纤与芯片对接”问题，有了可量产、低成本的路径，

传统FAU（光纤阵列单元）依赖V型槽和手动对准，一根一根光纤调，单通道耗时30-60秒，良率只有60%-70%，成本高且难以规模化，

康宁的GlassBridge用离子交换（IOX）工艺，在玻璃内部直接“刻”出渐变折射率的掩埋波导，一端9μm匹配光纤，另一端数百纳米适配硅光芯片，模场天然匹配，无源对准就能耦合，大幅减少主动对准工序，

配合炬光科技解决PIC内部垂直出光问题的专利，两家公司实际上形成了CPO/NPO接口的完整闭环，这对整个产业链意味着什么？

1. 对主营FAU的企业（如天孚通信），技术路线有变化，但机会在于主动拥抱GlassBridge，开发新型组件，承接产业上量红利，

2. 对主营MT/MPO、光纤混洗的企业**（如仕佳光子、太辰光），技术变化不大，反而会受益于NPO/CPO渗��带来的连接器需求增长，

3. 对NPO/CPO光引擎企业（如中际旭创、东山精密、新易盛），GlassBridge的低成本、高通道能力，能直接改善光引擎的竞争力和交付效率，

[P Equity Research 📰 @pequityresearch]

CICC: SiC to the left, GaN to the right—third-generation semiconductors emerge as the inevitable solution for high-voltage architectures in data centers

> Our estimates suggest that by 2030, a single MW of data center deployment could require approximately 10,000 SiC devices and 21,000 GaN devices, corresponding to per-MW values of $220,000 and $49,000, respectively—indicating considerable market opportunities.

> Between 2026 and 2030, rack-side (white zone) and server-room-side (gray zone) systems are expected to be progressively upgraded to 800V/±400V DC configurations, thereby driving demand for third-generation compound semiconductors. In the short term (1–2 years), transitional architectures led by 800V sidecar units may offer limited upside for SiC demand. However, as the following developments materialize: (1) blade-level voltage step-down or even high-density 800V-to-6V power conversion at the rack side, (2) centralized rectification at the server room side, and (3) implementation of solid-state transformer (SST) solutions, we believe the long-term optionality of SiC/GaN-related companies warrants proper pricing.

> We believe Chinese companies have already established deep positions in the SiC/GaN space, with continuously strengthening competitiveness. Going forward, Chinese firms across the third-generation compound semiconductor value chain stand to benefit substantially from the adoption of high-voltage data center architectures.

> Legacy data center racks average around 7 kW. In comparison, NVIDIA’s Hopper architecture consumes ~40 kW, Blackwell GB300 NVL72 reaches 134–140 kW, Rubin will exceed 200 kW, and future architectures (Kyber in 2027 and Feynman in 2028) are projected to scale to 600 kW and over 1 MW per rack.

The industry is projected to move through four transitional stages to implement 800V DC setups:

Phase 1 (2026/2027): White Space Retrofit with "Sidecar"

Setup: Gray space (facility level) remains untouched. A custom 800V DC side-mounted power cabinet ("Sidecar") is added next to the IT rack to rectify AC to 800V DC locally.

Semiconductor Impact: SiC demand is concentrated in the sidecar's front-end rectification/PFC modules.

SiC Volume: ~1,594 units per MW.

Phase 2 (2027/2028): Native 800V DC Computing (The GaN Inflection Point)

Setup: Centralized low-voltage UPS systems are phased out in favor of distributed rack-level battery backup units (BBUs) and supercapacitors. The 800V DC bus connects directly to the compute blade.

Semiconductor Impact: GaN replaces a massive portion of SiC for the main onboard power step-down stage (Intermediate Bus Converters) to meet strict space and thermal restrictions near the GPU. However, SiC finds rigid demand in high-voltage hot-swap protection and Solid-State Circuit Breakers (SSCBs).

Volume: ~1,755 SiC units per MW; ~10,303 to 10,667 GaN units per MW.

Phase 3 (2028/2029): Centralized Rectifiers in the Gray Space

Setup: Power rectification moves entirely upstream into the gray space. Massive facility-level centralized rectifiers convert grid power to an 800V DC injection backbone distributed throughout the facility.

Navitas GaN Transformation (June 2026): Driven by Navitas' GaNFast integration into NVIDIA’s MGX ecosystem, setups will begin converting 800V directly to 6V, entirely eliminating the traditional 48V intermediate step-down bus.

Volume: SiC surges to ~6,948 units per MW (driven by massive industrial DC distribution, BBUs, and energy storage systems). GaN usage explodes to 20,800–21,600 units per MW.

Phase 4 (Post-2029): The Final Solid-State Transformer (SST) Architecture

Setup: Both the low-voltage transformer and low-voltage rectification stages are completely eliminated. Megawatt-scale Solid-State Transformers (SSTs) directly convert medium-voltage grid AC to 800V DC.

Semiconductor Impact: High-frequency SST operation relies heavily on high-voltage SiC. The gray zone infrastructure captures 54% of the total SiC value chain.

SiC Volume: Skyrockets to ~9,886 units per MW.

Market Value

> As setups transition from Phase 1 to Phase 4, the hardware value of SiC per MW jumps drastically from $10,000–$25,000 to approximately $270,000. This value growth is primarily driven by voltage rating upgrades (moving from 650V to 1200V and 3.3kV devices).

> GaN addressable market value grows from $33k–$33.8k/MW in Phase 2 to $46.5k–$49.1k/MW in Phases 3/4.

Silicon Carbide (SiC) Unit Costs

The unit cost for SiC scales aggressively as devices move from standard rack components to high-voltage, high-density infrastructure:

Non-SST / In-Rack Devices: $2.5 – $3.5 per unit.

UPS / PDU / BESS Peripherals: $3.5 – $5.0 per unit.

800V Transitional / Early SST Stages: $10.0 base price per unit (utilizing mass-production 1200V-class SiC MOSFETs).

Full SST Architectures:

1200V SiC Modules: $15.0 – $20.0 base price per unit.
3.3kV SiC Modules: $50.0 base price per unit (reflecting the severe technical premium for medium-voltage direct conversion).

Gallium Nitride (GaN) Unit Costs

GaN pricing is stratified by application scenario, specifically focusing on its proximity to the high-voltage bus and the GPU itself:

Phase 2 (On-Blade Intermediate Bus Converters / IBC):

High-Voltage IBC (800V --> 50V): $3.8 per unit
(utilizing 650V integrated GaN ICs).

Low-Voltage IBC (50V --> 12V): $2.2 per unit (utilizing
100V-class GaN chiplets).

Phases 3/4 (Centralized & GPU-Proximate):

Navitas PDB Board System: $3.2 per unit.

GPU-Proximate Embedded GaN: $2.2 per unit (scaled AI server order pricing).

The SiC Step-Function Explosion

SiC exhibits massive value elasticity, meaning the financial value grows significantly faster than the physical unit count due to the transition to higher-priced, higher-voltage (1200kV -->3.3kV) devices.

Volume Growth: Scalings jump from 1,594 units/MW (Phase 1) to 9,886 units/MW (Phase 4), representing a strong ~83.73% CAGR.

Value Growth: Financial capture skyrockets from $2,000 – $15,000/MW (Phase 1) to $220,000/MW (Phase 4). This represents a staggering value CAGR of 144.78% to 379.14%.

The GaN High-Volume, Steady-Value Curve

Conversely, GaN follows a high-volume, highly localized deployment strategy. Its volume scales rapidly, but because it operates at lower, more commoditized voltage classes closer to the compute blade, its dollar-value growth is more linear.

Volume Growth: Unit counts double from 10,303 – 10,667 units/MW (Phase 2) to 20,800 – 21,600 units/MW (Phases 3/4), a CAGR of 95% – 110%.

Value Growth: Financial capture shifts from $33,000 – $33,800/MW to $46,560 – $49,120/MW, representing a more modest value CAGR of 38% – 49%.

[Omer Cheema @Omercheema]

A nice flow chart for packaging technologies.

[法克魷吉米 Jimmy | Real TSMC Insight | 純愛を学び中 💜 @jimmy_yoasobi]

1/10

Everyone talks about CPO as if the hard part is only the ASIC, lasers, or silicon photonics.

But there is a very physical bottleneck in the middle:

fiber-to-PIC coupling.

How do you plug many optical fibers into a photonic chip with low loss, high density, and enough reliability for AI data centers?

That is what $GLW GlassBridge is trying to address.

[livermoer @livermoerR]

半导体气体高纯二氧化碳或陷入短缺

据The Elec消息，由于炼油及石化工厂开工率下降，导致二氧化碳产量大幅减少。通常情况下，半导体制造商与相关供应商各自都会储备两周二氧化碳用量，合计能维持一个月的库存量，# 但近期库存已经跌破一个月水平。

高纯度二氧化碳主要用于半导体超临界清洗工艺。根据科创板日报，三星电子每月的高纯度二氧化碳用量约1800至2000吨，SK海力士则需要600至700吨。这两家巨头的生产尚未受到实际影响，但库存余量持续减少，正全力加紧采购。#不过即便提高单价、也难以确保额外供应量。

液态二氧化碳价格已较年初上涨约20%，关注高纯二氧化碳价格变化。# 广钢气体、金宏气体

[P Equity Research 📰 @pequityresearch]

Nomura Securities: Global Memory

> Global Data Center (DC) capex is projected to skyrocket from $668 billion in 2024 to $6,127 billion ($6.13 trillion) by 2030.

> The "Big 8" technology firms continue to represent the largest chunk of this spending, starting at $282 billion (42%) in 2024 and expanding to $2,653 billion (43%) by 2030.

> AI DC capex is expected to grow from $206 billion in 2024 to $3,379 billion by 2030, taking a 55% mix within total DC capex by the end of the decade.

> Total global memory revenue is forecasted to experience a massive surge, climbing from $181 billion in 2024 to $2,625 billion by 2030.

> Data centers will become the dominant force in memory consumption. The share of memory revenue coming from data center capex rises from 33% in 2024 to a peak of 56% in 2029 (leveling off slightly to 53% in 2030).

> While AI memory demand grows substantially (from $20B to $517B), memory demand for general-purpose data centers is projected to spike even faster in terms of absolute revenue contribution, moving from $39 billion in 2024 to $881 billion by 2030.

Shift in Hardware Architecture Costs (Memory vs. Logic)

Blackwell: Memory/Logic ratio stands at 200%.

Blackwell Ultra: Rises to 337%.

Rubin: Jumps sharply to 677% as NAND requirements enter the equation at $49,371 million.

Rubin Ultra: Explodes to 1,190%, with total memory TAM reaching $189,977 million compared to a rapidly shrinking relative logic share (only 8%).

[法克魷吉米 Jimmy | Real TSMC Insight | 純愛を学び中 💜 @jimmy_yoasobi]

1/8

We previously separated two materials that are easy to confuse:

Glass core sits inside the package substrate.

Glass cloth is used inside CCL / PCBs.

But this leaves a more important question:

Why do AI servers care about the materials inside these substrates and boards?

Because before data becomes light,

it usually travels through a high-speed electrical path.

[軒 @Raitue]

@luuu_xixi 你好福～

[軒 @Raitue]

@Cindy06010601 可以聊天囉～

[駿HaYaO @QQ_Timmy]

光晶片&CW光源紀要
1. EML晶片技術與良率現狀
- 性能：初期與博通、Lumentum、住友在電吸收調製器消光比有差距，經優化後已大幅縮小，達到規模化量產穩定水準。
- 良率：光晶片製程超300道工序，業界合格50%以上、60%為高於平均、70%以上世界一流；Lumentum、住友多在60%多區間，公司目前已進入此梯隊。

2. 高功率CW Laser晶片進展
- 研發順利：300-350mW產品需新增光放大結構，難度大幅提升，目前已在客戶端認證 + 內部嚴苛可靠性測試。
- 應用與進度：主要用於CPO外部光源光引擎，已送樣NVIDIA，最快2026年Q2末期或Q3出初步認證結果。
- 價格參考：70mW≈5美元、100mW≈十幾美元、300mW以上≈30-40美元（功率越高溢價越顯著）。

3. 產能擴張策略
- 目前完全��賴現有產線優化：2吋晶圓升級至4吋 + 盤活閒置產能 + 良率提升。
- 新增產線：已啟動廠房與設備採購，但仍處設備調試階段，最快2027年Q1才能釋放新增產能。
- 資本開支：主要集中在廠房建設與設備採購，人員投入佔比低。

4. 客戶訂單與合作進展
- 訂單支撐明確：AWS已提前鎖定部分產能；Google200G EML + CW Laser總需求預估達2億顆，驗證結束進入訂單商談；Coherent因自身EML缺貨主動求供貨，並以旋光片業務互補。
- AWS合作：目前以光模組為主，但正積極將晶片需求轉化為更多模組訂單。
- Meta：1.6T光模組已通過認證，800G曾因單一批次良率觸發「灰階測試」暫停，已於2026年1月台灣工廠現場審核通過恢復正常。

5. 自身光模組業務與晶片分配策略
- 2026年自身模組業務預計消耗約6,000萬顆光晶片；2027年若1.6T需求翻倍，消耗量同步大幅上升。
- 分配原則：優先保障內部光模組需求（雖然模組毛利率較低，但整體產值較高，對營收拉動較顯著）。

6. 市場供需與價格趨勢
- 2026年全球光晶片缺口約1-1.5億顆，2027年缺口將進一步擴大。
- 價格趨勢：需求爆發 + 擴產週期長 + 原物料上漲，光晶片明確具漲價空間。
- 客戶鎖定產能時：均簽正式合約，一併鎖定價格、交期與數量。

7. 1.6T光模組最新動態
- 認證與出貨：Meta已通過，重點爭取Google（同步提供EML+矽光兩方案）與NVIDIA（CPO光引擎突破）。
- 發貨時程：原計劃2026年Q3批量發貨，因需求預計提前至Q2。
- 技術方案：EML（自研200G EML晶片） vs 矽光（INTEL代工 + 羲禾供應），EML單台成本較矽光高約200美元。

8. 競爭格局與中國內廠商比較
- 長光華芯：100G EML已推出，但短期無法進入北美市場，僅限國內競爭。
- 源傑：CW Laser進度較前，但EML晶片尚未規模量產，100G/200G認證未完成，整體落後。
- 公司優勢：已通過嚴苛客戶認證 + 產能鎖定，在北美高端市場領先。

9. 2027年收入展望
- 依市場預測：2027年1.6T光模組預估7,500萬隻（2026年2,000萬隻，成長超3倍）；800G約8,500萬隻。
- 公司整體營收規模將隨核心產品出貨量與定價大幅提升，2026-2027年充分受益AI光通訊超級週期。

[看不懂的SOL @DtDt666]

这堂1小时的彭博终端课程，能让兄弟们学到的交易市场知识，比在高盛或摩根大通实习2年还要多。

建议兄弟们收藏，

无论如何都要花一小时看完。