采购指南 | LiiFoo 采购指南 – 第 5 页 – LiiFoo
Get Quote

标签： 采购指南

  • GaN 快充与射频器件采购实战：GaN-on-Si 与 GaN-on-SiC 路线选择、动态 Rds(on) 锁定与到岸成本（2026）

    先给结论：2026 年买 GaN，已经不是”买一种新材料”，而是”买一颗带材料级风险的器件”。决定项目成败的只有三件事：一、你选了哪条衬底路线；二、你有没有在合同里锁定动态 RDS(on)，而不是只看数据手册首页那个数；三、你有没有意识到 GaN 没有行业统一封装引脚定义——所以”第二供应商”是在 RFQ 阶段做的设计决策，不是量产后再做的采购决策。剩下的价格、运费、关税，都只是算术题。

    一、你买的到底是什么器件

    GaN 功率器件是横向 HEMT：电流沿 AlGaN/GaN 界面形成的二维电子气（2DEG）横向流动。这和硅 MOSFET、SiC MOSFET 的纵向结构有本质区别，而这些区别直接决定了采购要谈什么：

    • 没有体二极管。反向恢复电荷 Qrr 基本为零。这是 GaN 在硬开关桥式拓扑里压倒硅器件的核心原因，也是它能做到 MHz 级开关频率的前提。
    • 输出电荷 QOSS 和栅极电荷 QG 极低。开关损耗大幅下降，磁性元件缩小，功率密度上升——这就是 140 W 快充能做到掌心大小的物理原因。
    • 横向结构 = 对表面态敏感。高压开关应力下，表面和缓冲层的电荷陷阱会让通态电阻瞬态升高。硅 MOSFET 的数据手册里没有这一项，而绝大多数”GaN 用了不如预期”的案例都出在这里。
    • 栅极窗口极窄（p-GaN 栅型）。VGS 绝对最大值通常只有 6–7 V，而硅器件是 20 V。栅极驱动余量非常小，布局稍差不会在实验室炸，会在客户现场炸。

    二、衬底路线：决定成本下限和性能上限

    路线 典型电压/场景 选它的理由 必须注意
    GaN-on-Si 100 V / 200 V / 650 V——快充、适配器、服务器电源、微逆变、48 V DC-DC 每安培成本最低；可跑 6 英寸和 8 英寸硅产线，产能是真实可扩的 硅衬底导热约 150 W/m·K，限制连续功率；缓冲层工艺水平在不同厂商间差距极大
    GaN-on-SiC 射频功放、5G/6G 基站、雷达、卫通 SiC 衬底导热约 370–490 W/m·K，约为硅的 3 倍；功率密度与射频线性度最高 衬底成本主导 BOM；出口管制敞口最大；交期长
    GaN-on-sapphire 低功率消费类充电器 衬底最便宜，低占空比适配器够用 导热约 35 W/m·K，散热通道差。凡是持续负载或小体积密闭结构，不要接受这条路线
    GaN-on-GaN（同质衬底） 纵向器件、激光器、小众大电流 晶体质量最好，纵向导流 衬底供应薄、价格高。2026 年按研发料对待，不要按量产料排产

    实操判断：3 kW 以下的 AC-DC 变换，指定 GaN-on-Si，谈价格；射频应用，你只能买 GaN-on-SiC，谈的是交期和许可证文件，不是价格。

    三、器件架构：Cascode 还是真增强型

    两者都叫”常关型 GaN”，但行为不同，栅极驱动不可互换。

    • Cascode 级联——耗尽型 GaN HEMT 与低压硅 MOSFET 同封装串联。驱动方式与硅 MOSFET 相同（0–10/12 V），是存量设计升级风险最低的方案。代价：硅管带来一点非零 Qrr、额外封装寄生、以及略高的 RDS(on) 下限。
    • p-GaN 栅（e-mode）——真正的单芯片增强型。优值系数最好、可达最高开关频率，但栅驱动必须守住约 5–6 V 标称 / 6–7 V 绝对最大的窗口，而且要管的是栅极电流，不只是电压。
    • GaN 集成功率 IC——驱动、电平位移、保护（有的还带检测）与 HEMT 集成。物料更少、布局出错概率大幅下降、上市更快。代价是被单一厂商锁定，且没有任何 pin-to-pin 替代。当进度风险大于供应风险时才选它。

    四、RFQ 必须附上的规格清单

    只写电压、电流、封装的 GaN 询价单，拿回来的报价根本没法横向比较。以下字段必须锁死：

    1. VDS 额定值与瞬态上限。650 V 器件通常标有更高的瞬态 VDS，要供应商给出瞬态数值和允许持续时间，不要接受宣传口径。
    2. 25 °C 和 TJ=125 °C 两个温度点的 RDS(on)。GaN 从 25 °C 到 125 °C 的温度系数通常在 1.7–2.3 倍。只报 25 °C 的话，你的热设计建立在一个实际工作中永远不会出现的数上。
    3. 动态 RDS(on) 数据，且按你的开关条件测。要求按 JEDEC JEP180（GaN 功率器件动态通态电阻测量的行业指南）在额定 VDS、125 °C、你的开关频率下表征，并给出 RDS(on),dyn/RDS(on),static 比值。比值在 1.1–1.3 附近属于成熟工艺；如果供应商根本给不出这份数据，这个信息本身就已经很说明问题了。
    4. QG、QOSS、EOSS、COSS(tr)、CISS——QOSS 与 EOSS 决定软开关损耗，是最常被漏标的参数。
    5. 栅极：VGS 标称值、绝对最大值、阈值 VTH 及其分布。要跨批次的 VTH 分布带，不要只要典型值。
    6. 热：Rth(j-c)、Rth(j-a)，以及散热面朝向（顶部散热 vs 底部散热）。顶部散热封装会改变整块 PCB 和散热器设计——必须在布局前定。
    7. 封装与绝缘——PQFN、DFN、TOLL、嵌入式/晶圆级。横向 GaN 要按终端产品安规（如 IT/AV 设备的 IEC 62368-1）核对漏极—源极焊盘间的电气间隙与爬电距离。
    8. 可靠性认证证据——HTRB、HTGB、H3TRB、温度循环、IOL；车规还要 AEC-Q101 加 PPAP。特别要问开关应力下的可靠性（JEDEC JEP173 覆盖这一类测试）；对硬开关变换器而言，只有直流应力的认证数据是不够的。
    9. MSL 等级、干燥包装状态、回流焊曲线、编带盘规格、MOQ。
    10. 批次追溯与 COA 内容——晶圆批号、封装批号、日期码，以及成测环节实际测了哪些电参数。

    五、从中国采购 GaN 特有的三个风险

    5.1 镓相关物项的出口许可

    自 2023 年 8 月起，中国对镓相关物项实施出口许可管理，氮化镓属于管制类别之列。实践中，已封装成品器件/模块与镓金属、GaN 衬底、外延片的处理方式差异很大。如果你买的是成品晶体管或功率 IC，通常不会被卡；如果你买的是 GaN 外延片、衬底或裸芯，就要默认存在许可环节，并在排产计划里预留数周时间。下单前请书面向出口方确认四件事：申报用哪个 HS 编码、是否需要许可、许可是否已获批、当前审批周期多长。管制清单会调整——请以主管部门当期口径为准，不要以任何文章（包括本文）为依据。

    5.2 关税与归类敞口

    分立 GaN 晶体管与 GaN 集成功率 IC 的归类不同（晶体管税目 vs 集成电路税目），由此带来的关税与贸易救济待遇可能差别显著——对美国进口商尤其如此，近几轮调整中半导体相关税目税率有所上调。请在承诺量产价格之前，让报关行基于一份样品发票确认归类，并确保供应商申报的 HS 编码与之一致。建立在错误税目上的到岸成本模型不是模型，是意外。

    5.3 没有统一引脚定义——第二供应商是设计任务

    与硅 MOSFET 不同，不同厂商的 GaN 器件很少 pin 兼容，集成 GaN 功率 IC 则完全不兼容。这是 GaN 采购里最常见的结构性错误：研发按某家器件做完设计，量产爬坡时才发现无替代料。缓解手段按有效性排序：性能允许时优先选择产业界广泛支持的封装（TOLL、标准 PQFN 尺寸）；在开发阶段就在同一封装上认证两家供应商，而不是事后补；如果必须用单源集成 IC，就在谈价格的同时把安全库存或 last-time-buy 条款一起谈掉。

    六、到岸成本模型

    按以下顺序搭模型，第 4–7 项不能省——GaN 项目的成本优势通常就是在这几行里悄悄流失的：

    1. 年用量对应的 FOB 单价（要阶梯价表：1 万 / 10 万 / 50 万 / 100 万颗）
    2. 运费——半导体走空运，重量小但单票成本不低，尽量拼单
    3. 关税与贸易救济，按已确认的 HS 编码计算
    4. 报关费、ESD/MSL 合规包装、干燥包装处理
    5. 来料检验成本（见第七节）与预估不良率
    6. 因单源风险被迫持有的安全库存的资金占用成本
    7. 系统级对冲收益——GaN 真正的成本逻辑在系统层面，不在器件层面：磁性元件更小、散热器更小、电容更少、能效等级更高。要把因此删掉的 BOM 量化出来。如果只拿器件单价和硅器件比，GaN 永远输——那说明你测错了指标。

    2026 年价格方向：8 英寸 GaN-on-Si 已进入真实量产规模，650 V 芯片成本被压缩，竞争焦点已从”能不能供”转向”能不能证明动态 RDS(on) 和可靠性”。任何显著低于市场带的报价，应触发一轮数据索取，而不是被当成砍价成功。

    七、真正能拦住 GaN 问题的来料检验

    • 曲线仪/参数测试：指定 VGS 下的 RDS(on)、VTH、IDSS 与 IGSS 漏电。逐批跟踪 VTH 分布——漂移是”供应商没告知的工艺变更”的最早信号。
    • 双脉冲测试（DPT）：在你自己的拓扑里验证开关能量和动态通态电阻行为，这是唯一现实可行的方法。把 DPT 结果写进首件（FAI）验收条件。
    • 满载热成像——在额定功率、在你的实际外壳内测，不是在开放台面上测。
    • 封装/板级：抽样做 X-ray 或超声扫描（CSAM）查散热焊盘下的空洞；确认运输途中 MSL 车间寿命没有被破坏。

    八、必须写进规格书的设计红线

    下面这些失效模式会产生现场退货，然后被算到供应商头上：

    • 回路电感。高频 GaN 的功率回路通常要控制在几 nH 以内。回路一长，快速 dv/dt 就会变成超过瞬态额定值的 VDS 过冲。
    • 开尔文源极连接，把栅驱动与共源电感解耦。
    • 栅极回路保护——绝对最大值只有约 6 V，硅设计能容忍的振铃在这里是致命的。串联栅电阻、紧凑回流路径、必要处加钳位。
    • 死区时间管理——没有体二极管就没有宽容的反向导通通路，直通余量必须实测验证，不能靠假设。

    九、可落地的 2026 推进节奏

    1. 第 1–2 周：锁定应用、拓扑、频率与热预算。定衬底路线（第二节）与器件架构（第三节）。
    2. 第 2–4 周：带第四节完整规格清单发 RFQ。缺动态 RDS(on) 和 125 °C 数据的报价直接淘汰。
    3. 第 4–8 周：至少两家供应商、同一封装取样。做 DPT 并在真实外壳里做热验证。
    4. 第 8–12 周：书面确认 HS 编码与许可状态；量产前先走一票小额付费试运，验证单证、MSL 包装与报关行操作。
    5. 持续：逐批参数趋势跟踪；合同里写入变更通知条款，覆盖晶圆厂、外延供应商、缓冲层结构与封装厂。在 GaN 上，所谓”小的外延变更”从来不小。

    本文引用的标准与管制依据（JEP180、JEP173、AEC-Q101、IEC 62368-1、镓出口管制、关税归类）均作为采购检查点列出。签约前请向标准发布机构、报关行与供应商核实当期版本与现行法律状态。

  • 碳化硅（SiC）衬底采购与应用指南：功率电子用晶圆的2026选型、外延工艺与供应商筛选

    碳化硅（SiC）早已从实验室里的奇珍，变成了新一代功率电子项目的默认衬底选择——电动汽车牵引逆变器、光伏组串逆变器、超快充电桩，以及AI数据中心内部密集的功率级。如果你的2026年采购计划涉及上述任何一个领域，搞懂功率电子用碳化硅衬底已不再是可选项。本实战指南将带采购与工程买家厘清：这种晶圆到底是什么、哪些等级真正影响价格与良率、外延层如何改写规格表，以及SiC相比它正在快速替代的硅IGBT究竟如何。

    碳化硅衬底到底是什么

    这里的”衬底”指由4H-SiC多型体生长而成的抛光单晶SiC晶圆——正是其宽禁带与高电子迁移率，使高压开关得以实现。买家必须严格区分裸衬底与可量产的外延晶圆。几乎没有功率器件厂商会直接在衬底上加工电路，而是在其表面外延生长一层掺杂的碳化硅薄膜；真正进入晶圆厂的，正是这层6英寸碳化硅外延片——以及日益增多的8英寸规格。明确供应商所报的是哪一种形态，能避免”采购下单”与”产线可用”之间的昂贵错配。

    买家会遇到的等级与规格

    SiC衬底围绕一整套物理与电学指标报价。真正左右价格、良率与器件表现的是以下几项：

    • 多型体与晶向：4H-SiC沿<11-20>方向偏轴4°–8°切割，是MOSFET与肖特基二极管量产的标准。
    • 直径：6英寸（150 mm）是2026年的量产主力；合格的8英寸产线正在爬坡，但溢价更高、分配更紧。
    • 导电类型：半绝缘级用于射频，而低阻N型几乎是无例外的功率买家之选。
    • 表面与结构质量：微管密度（MPD）、基平面与螺纹位错密度（BPD/TSD）、总厚度偏差（TTV）、翘曲、弯曲与表面粗糙度（Ra）。近零MPD与低百位数的BPD，定义了高良率衬底。

    当报价低得可疑时，几乎总是这些结构指标在悄悄放松。在下单比对供应商前，请索取完整的计量报告——而不只是直径与多型体。

    为什么外延层会改写规格表

    如果你采购的是6英寸碳化硅外延片，议题就从衬底缺陷转向外延参数：漂移层厚度、掺杂浓度与外延缺陷密度。这些数值直接决定器件的耐压与导通电阻。1200 V MOSFET与650 V MOSFET需要完全不同的外延配方，因此”一片SiC晶圆”从来不是一个单一SKU，而是一个配置化构建。请明确目标电压等级，让供应商确认外延堆叠，而不是接受含糊的”标准外延”标签。

    SiC vs 硅IGBT：成本、效率与可靠性

    坦诚的采购问题是：更高的晶圆成本能否在应用中收回？在大多数高频、高温或空间受限的设计中，答案是肯定的。

    维度 硅IGBT 基于SiC衬底的SiC MOSFET
    禁带宽度 1.1 eV 3.3 eV
    开关损耗 高 最高降低约70%
    工作温度 约125–150°C 175–200°C及以上
    系统散热与体积 更大散热器 更小、更易被动散热
    晶圆与器件成本 更低 更高，2026年持续收窄
    严苛工况可靠性 良好 优异

    SiC与硅IGBT的对比很少只是芯片层面的事。SiC让设计者得以缩小无源元件、散热与磁件——在系统层面常常抵消掉器件级的价差，尤其是在电动汽车动力总成与直流快充中。

    2026采购核对清单

    1. 界定形态：裸衬底还是外延片，以及外延的目标电压等级。
    2. 锁定直径：量产以6英寸为准；8英寸留给合格且成本容忍度高的项目。
    3. 索要计量：MPD、BPD、TSD、TTV、翘曲、电阻率，以及外延厚度与掺杂。
    4. 梳理供应版图：中国、日本、美国与欧洲均有合格厂商——权衡交期、最小起订量与地缘风险。
    5. 先认证再放量：在承诺量产之前，对前三批做批次验收测试（LAT）。
    6. 跟踪产能爬坡：8英寸产能与中国本土良率，是2026年价格波动最大的两个变量。

    2026年的交期与价格

    标准等级的6英寸衬底交期通常为数周，紧规格或外延构建会更长。随着中国本土产能放量、8英寸产线合格化，价格正在走软，但最高质量、最低缺陷的晶圆仍存在分配风险。关键项目应预留缓冲，并尽早认证第二货源。

    需要避开的坑

    产线需要外延片时，却只指定了裸衬底；不问计量报告就追最低价；以为”SiC”是可互换的单一大宗商品——请把它当作一个经过配置、有完整溯源的可认证元件来对待。

    结论

    对2026年的功率电子买家而言，功率电子用碳化硅衬底是构建更快、更冷、更小系统的基础。吃透等级、坚持外延规格、执行严谨的认证流程——SiC与IGBT的账，终将站在你这边。

  • Power Electronics Buyers’ 2026 Field Guide to Silicon Carbide (SiC) Substrates: Grades, Epitaxy and a Sourcing Checklist

    Silicon carbide (SiC) has moved from a laboratory curiosity to the default substrate behind a wave of next-generation power-electronics programs — electric-vehicle traction inverters, solar string inverters, ultra-fast charging piles, and the dense power stages inside AI data centers. If your 2026 sourcing plan touches any of these segments, understanding the silicon carbide substrate for power electronics is no longer optional. This field guide walks procurement and engineering buyers through what the wafer actually is, which grades move price and yield, how the epitaxial layer rewrites the spec sheet, and how SiC compares with the silicon IGBT it is rapidly displacing.

    Three forces are pulling SiC into mainstream bills of materials in 2026. The shift to 800 V battery architectures in electric vehicles demands inverters that waste less energy during fast switching. Higher-power-density solar string inverters need the same efficiency at a lower cooling cost. And AI data centers, starved for power and rack space, reward any component that cuts loss and heat. In each case the substrate is where those gains begin.

    What a Silicon Carbide Substrate Actually Is

    In this context a “substrate” is a polished, monocrystalline SiC wafer grown from the 4H-SiC polytype — the crystal form whose wide bandgap and high electron mobility make high-voltage switching practical. Buyers should keep a sharp distinction between a bare substrate and a device-ready epitaxial wafer. Almost no power-device maker processes circuits directly on the substrate; instead a thin, doped silicon-carbide epitaxial film is grown on top, and it is that 6 inch SiC epitaxial wafer — and increasingly the 8-inch format — that actually enters the fab. Knowing precisely which form a supplier is quoting prevents the expensive mismatch between what procurement ordered and what the production line can use.

    Grades and Specifications Buyers Meet

    SiC substrates are quoted against a stack of physical and electrical specifications. The parameters that genuinely move price, yield, and device behavior are:

    • Polytype and orientation: 4H-SiC cut 4° to 8° off-axis toward the <11-20> direction is the production standard for MOSFETs and Schottky diodes.
    • Diameter: the 6-inch (150 mm) wafer is the 2026 volume workhorse; qualified 8-inch lines are ramping but command a premium and tighter allocation.
    • Conductivity type: semi-insulating grades serve RF, while low-resistivity N-type is what power buyers need almost without exception.
    • Surface and structural quality: micropipe density (MPD), basal-plane and threading dislocation densities (BPD/TSD), total thickness variation (TTV), warp, bow, and surface roughness (Ra). Near-zero MPD and BPD in the low hundreds define a high-yield substrate.

    When a quotation looks suspiciously cheap, it is almost always one of these structural numbers quietly relaxing. Request the complete metrology sheet — not just diameter and polytype — before you compare suppliers.

    Why the Epitaxial Layer Changes the Spec Sheet

    If you are purchasing the 6 inch SiC epitaxial wafer, the discussion shifts from substrate defects to epitaxy parameters: drift-layer thickness, doping concentration, and epi-defect density. These values set the device’s blocking voltage and on-resistance directly. A 1200 V MOSFET and a 650 V MOSFET require entirely different epi recipes, so “a SiC wafer” is never a single SKU — it is a configured build. State the target voltage class and let the supplier confirm the epi stack, rather than accepting a vague “standard epitaxy” label.

    SiC vs Silicon IGBT: Cost, Efficiency and Reliability

    The candid procurement question is whether the higher wafer cost pays back in the application. In most high-frequency, high-temperature, or space-constrained designs, it does.

    Dimension Silicon IGBT SiC MOSFET on SiC substrate
    Bandgap 1.1 eV 3.3 eV
    Switching loss High Up to ~70% lower
    Operating temperature ~125–150°C 175–200°C and beyond
    System cooling and size Larger heatsink Smaller, passive-friendly
    Wafer and device cost Lower Higher, narrowing through 2026
    Reliability in harsh duty Good Excellent

    The SiC vs silicon IGBT decision is rarely about the chip alone. SiC lets designers shrink the passives, cooling, and magnetics — frequently erasing the device-level price gap at the system level, particularly in EV powertrains and DC fast chargers.

    Which Applications Justify the SiC Premium

    Not every design needs SiC, but several clearly do. The clearest winners are:

    • EV traction inverters and on-board chargers, where efficiency extends range and shrinks the cooling package.
    • DC fast chargers, where SiC cuts the size and loss of the power stage at high switching frequency.
    • Solar and storage inverters, where system-level savings on magnetics and heatsinks outweigh the wafer premium.
    • Motor drives and industrial power supplies running at high frequency or in hot environments.
    • Rail and grid equipment, where reliability under thermal cycling is decisive.

    If your product switches at high frequency, runs hot, or must stay compact, the substrate investment usually pays back quickly.

    2026 Sourcing Checklist

    1. Define the form: bare substrate or epitaxial wafer, and the exact target voltage class for the epi.
    2. Pin the diameter: standardize on 6-inch for volume; reserve 8-inch for qualified, cost-tolerant programs.
    3. Demand the metrology: MPD, BPD, TSD, TTV, warp, resistivity, and epi thickness and doping.
    4. Map the supply base: qualified houses exist across China, Japan, the United States, and Europe — balance lead time, minimum order quantity, and geopolitical exposure.
    5. Qualify before scaling: run a lot-acceptance test on the first three lots before committing production volume.
    6. Track the ramp: 8-inch capacity and Chinese domestic yield are the two variables that will move 2026 pricing most.

    Documents and Certificates to Request

    Beyond the metrology numbers, ask each supplier for a certificate of conformance (CoC), X-ray orientation confirmation, resistivity mapping, a defect-map summary, and current RoHS and REACH declarations. For automotive or aerospace programs, evidence of IATF 16949 or AS9100-aligned quality systems should sit inside the qualification packet. Treat these documents as part of the product, not as paperwork.

    Lead Times and Pricing in 2026

    Expect 6-inch substrate lead times of several weeks for standard grades, lengthening for tight-spec or epitaxial builds. Pricing is softening as Chinese domestic capacity scales and 8-inch lines qualify, but allocation risk remains for the highest-quality, lowest-defect wafers. Build buffer into critical programs and qualify a second source early.

    Pitfalls to Avoid

    Do not specify a bare substrate when the line needs an epitaxial wafer. Do not chase the lowest price without the metrology sheet. And do not assume “SiC” is one interchangeable commodity — treat the substrate as a configured, qualified component with a documented pedigree.

    Bottom Line

    For power-electronics buyers in 2026, the silicon carbide substrate for power electronics is the foundation of a faster, cooler, and smaller system. Learn the grades, insist on the epitaxial specification, and run a disciplined qualification process — and the SiC-versus-IGBT math will work firmly in your favor.

  • Silicon Nitride Bearing Balls Demystified: Why Si3N4 Outperforms at High RPM and How to Specify Them in 2026

    Why Silicon Nitride Balls Suit High-Speed Bearings

    Silicon nitride (Si3N4) bearing balls combine high hardness, low density (about 40% of bearing steel), high elastic modulus, corrosion resistance, electrical insulation and a low expansion coefficient, making them ideal rolling elements for high-speed, high-precision bearings. In motor spindles, turbochargers, wind-gearbox and EV-drive bearings, ceramic balls cut centrifugal force and heat, enabling higher speed and longer life.

    Core Performance Parameters

    Parameter Typical Meaning
    Density 3.2 g/cm³ Lower inertia
    Hardness ≥90 HRA Wear / pitting resistance
    Thermal exp. ~3.2×10⁻⁶/K Match with steel rings
    Max temp ≥800℃ (inert) High-temp duty
    Electrical Insulating Prevents E-corrosion

    Grade & Procurement Essentials

    • Grade: G3–G10 (G3 highest); select by speed and noise target.
    • Roundness & surface finish: drive running smoothness and acoustic noise.
    • Batch consistency: ball-to-ball deviation within one bearing must be minimal to avoid load skew.
    • Density & microstructure: require porosity-free, uniform grains to avoid early fatigue.

    How to Specify Them

    The spec sheet should state nominal diameter, grade (e.g., G5), material standard (ASTM F2094 / GB type), surface quality and packaging. Hybrid ceramic bearings (steel rings + ceramic balls) are the mainstream, balancing stiffness and cost; full-ceramic suits aggressive corrosion or magnetic fields. Require third-party test reports and accept on bench-life validation.

    2026 Supply & Cautions

    Overseas suppliers include Toshiba, CoorsTek, NSK and Koyo; domestic players include Sinoma and Shandong Industrial Ceramics. Price rises sharply with grade and diameter. Caution: ceramic balls are impact-sensitive in assembly—use dedicated press and matching tools; mixing balls from different makers causes property scatter, so avoid it.

  • 氮化硅陶瓷球采购与应用指南：高转速轴承球的2026选型要点

    氮化硅陶瓷球为什么适合高转速轴承

    氮化硅（Si3N4）陶瓷球凭借高硬度、低密度（约为轴承钢的 40%）、高弹性模量、耐腐蚀、电绝缘与低膨胀系数，成为高转速、高精密轴承的理想滚动体。在电主轴、涡轮增压、风电齿轮箱与电动汽车电驱轴承中，陶瓷球可显著降低离心力与发热，支持更高转速与更长寿命。

    核心性能参数

    参数 典型值 意义
    密度 3.2 g/cm³ 降低转动惯量
    洛氏硬度 ≥90 HRA 耐磨抗点蚀
    热膨胀系数 ~3.2×10⁻⁶/K 与钢套圈匹配性
    最高使用温度 ≥800℃（惰性） 高温工况
    电压等级 绝缘 防电蚀

    精度等级与采购要点

    • 精度等级：G3–G10（G3 最高），按转速与噪音要求选定。
    • 圆度与表面粗糙度：决定运转平稳性与异音。
    • 批间一致性：同一轴承内球批差需极小，避免载荷偏载。
    • 密度与显微组织：要求无孔隙、晶粒均匀，防止早期疲劳。

    选型与规格写法

    规格书应明确公称直径、等级（如 G5）、材料标准（如 ASTM F2094 / GB 型）、表面质量与包装方式。混合陶瓷球轴承（钢套圈+陶瓷球）是主流方案，兼顾刚度与成本；全陶瓷轴承用于强腐蚀或强磁场场景。采购时建议索要第三方检测报告，并以台架寿命验证作为最终验收。

    2026 供应与注意事项

    海外供应商包括 Toshiba、CoorsTek、NSK、Koyo 等，国内亦有中材高新、山东工业陶瓷等。价格随等级与直径上升显著。注意：陶瓷球脆性对装配冲击敏感，压装与合套需专用工装；混批使用不同厂家球会导致性能离散，应避免。

  • Carbon Fiber Prepreg Unpacked: Resin Systems, Cure Kinetics and the Use Cases That Justify the Premium

    What Carbon Fiber Prepreg Is

    Carbon fiber prepreg is a thin, semi-cured intermediate material made by impregnating carbon fiber (unidirectional tape or fabric) with a matrix resin such as epoxy, bismaleimide (BMI) or phenolic under controlled temperature. By locking the fiber-to-resin ratio at the factory, prepreg delivers highly predictable laminate properties and low void content, making it the structural backbone of aerospace, wind-turbine blades and high-end sporting goods.

    Grade Systems & Key Specs

    Dimension Typical Values Procurement Meaning
    Fiber grade T300 / T700 / T800 / T1100 Strength & modulus class
    Resin system Epoxy / BMI / Phenolic Temperature & toughness
    Areal weight 100–600 g/m² Ply thickness design
    Resin content 33%–42% Mechanics & process window
    Cure temperature RT / 120℃ / 180℃ Tooling & energy cost

    Resin Systems & Cure Kinetics

    Epoxy is mature and tough, covering most aerospace and industrial parts. BMI withstands higher temperatures (long-term >180℃) for engine-adjacent and supersonic structures; phenolic targets flame retardancy and ablation. The cure window (gel time, ramp profile) sets the working time and part yield—request DSC curves and gel-time data at sourcing.

    Cold Chain & Storage

    Prepreg is temperature-sensitive and typically stored at -18℃, with a 6–12 month shelf life; after out-loading it must be laid up within a defined open time. Shipments require unbroken cold chain—verify the data-logger and batch labels on arrival to avoid premature activation and tack loss.

    Use Cases That Justify the Premium

    When a part demands high specific strength, fatigue reliability or designed lay-up (wind-turbine spar caps, aerospace secondary structures, race monocoques, satellite frames), prepreg beats hand-layup and liquid molding on control and material savings. For low-cost, high-volume, simple geometries, weigh its cold-chain and equipment overhead. Sourcing tip: pick the grade and supplier against the end-part certification standard (e.g., aerospace NADCAP) and require batch-consistency reports.

  • 碳纤维预浸料选型指南：牌号体系、固化工艺与风电航天应用场景（2026）

    碳纤维预浸料是什么

    碳纤维预浸料（Prepreg）是将碳纤维（单向带或织物）预先浸渍环氧树脂、双马（BMI）或酚醛等基体树脂，并在控制温度下半固化制成的薄片状中间材料。它把”纤维+树脂”的配比在工厂级完成，使复合材料构件性能高度可控、孔隙率更低，是航空航天、风电叶片与高端体育器材的结构基石。

    牌号体系与关键规格

    维度 常见取值 采购含义
    纤维牌号 T300 / T700 / T800 / T1100 强度与模量等级
    树脂体系 环氧 / BMI / 酚醛 耐温与韧性
    面密度 100–600 g/m² 铺层厚度设计
    树脂含量 33%–42% 力学与工艺窗口
    固化温度 室温 / 120℃ / 180℃ 模具与能耗成本

    树脂体系与固化行为

    环氧树脂工艺成熟、韧性好，覆盖绝大多数航空与工业件；BMI 耐温更高（长期 >180℃），用于发动机周边与超声速结构；酚醛侧重阻燃与烧蚀。固化窗口（凝胶时间、升温曲线）直接决定可操作时间与构件良率，采购时应索取 DSC 曲线与凝胶时间数据。

    冷链与贮存

    预浸料对温度敏感，通常需 -18℃ 冷库贮存，保质期 6–12 个月；出库后需在规定开窗期内完成铺贴。运输必须全程冷链，到货应核对温度记录仪与批次标签，避免提前活化导致粘性失效。

    值得付溢价的应用场景

    当构件需要满足高比强度、疲劳可靠性或可设计铺层（如风电主梁、航空次承力件、赛车单体壳、卫星结构）时，预浸料相比手糊/液体成型更可控、更省料。对低成本、大批量、外形简单的部件，则需权衡其冷链与设备投入。采购建议：按最终构件认证标准（如航空 NADCAP）反向选定牌号与供应商，并要求批次一致性报告。

  • 硅碳复合负极材料选型与应用指南：高能量密度锂电负极的2026采购要点

    为什么硅碳复合负极成为提能量密度的主流路径

    硅碳复合负极材料通过在石墨基体中引入纳米硅或氧化亚硅（SiOx），将负极可逆比容量从人造石墨的约 372 mAh/g 提升至 450–650 mAh/g 区间，同时为电芯带来 8%–20% 的能量密度增益。硅理论比容量高达 4200 mAh/g，但嵌锂体积膨胀超过 300%，必须依靠碳基体缓冲与结构封装才能实用化。2026 年，动力与高端储能电池普遍采用硅碳复合负极掺混方案提升续航与快充表现。

    三条主流技术路线对比

    路线 首效 膨胀控制 成本 适用场景
    纳米硅/石墨复合 中 中 中 高容量动力电池
    硅氧碳 SiOx/C 高 好 偏高 长循环储能/高端动力
    多孔碳包覆硅 中高 优 高 超高能量密度电芯

    采购必看的 6 项指标

    • 首次库伦效率（ICE）：决定成片可发挥容量，动力级通常要求 ≥86%。
    • 可逆比容量：按掺硅比例在 450–650 mAh/g 区间评估。
    • 振实/压实密度：影响极片面密度与电芯体积能量。
    • 中位粒径 D50 与粒径分布：影响涂布均匀性与倍率性能。
    • 500 周容量保持率：长循环场景建议 ≥80%。
    • 磁性异物（Fe/Cr/Ni）：动力级要求 ppb 级管控，防止自放电。

    选型与应用建议

    与三元体系掺混比例多为 3%–10%（质量比），与磷酸铁锂体系掺混需更关注 ICE 与循环；高镍三元更看重硅碳对能量密度的边际贡献。采购时应要求供应商提供扣电实测数据与极片膨胀率报告，并以批量一致性（批次 D50、ICE 波动）作为准入门槛。

    2026 供应与风险提示

    国内主要供应商包括贝特瑞、杉杉、璞泰来、凯金等，海外有 POSCO、Group14 等。价格随硅含量与包覆工艺波动。重点风险：硅含量越高循环衰减越快，需配套补锂工艺；低温与快充工况下硅负极极化更明显，应在电芯层级验证。

  • Advanced Materials Price Trend Daily, 24 Aug 2026 | Brent Breaks $94 and Reignites Costs, PI Film Lands a Second 20%+ Hike, PTFE Grades Split Further

    Price Trend Daily Report – 2026-08-24

    Key takeaway: Brent crude surged 6.4% in a single week to break $94/bbl, lifting the cost floor across the entire chain. PI film’s second round of 20%+ hikes has landed in July–August, making it the only genuine seller’s market this period. PTFE grade divergence is widening — commodity powder under pressure while electronic grades hold. High-end and low-end zirconia have fully decoupled.

    Price Overview

    Material Current Price Range WoW Trend
    PTFE resin (suspension medium granule) RMB 44,000–46,000/t (Shandong low end 31,800) -1% to 0% Weak, choppy ↘
    PTFE resin (suspension fine powder) RMB 47,000–50,000/t 0% Relatively resilient →
    PTFE dispersion resin / dispersion emulsion RMB 44,000–46,000 / 28,000–30,000/t 0% Stable →
    PEEK resin (domestic virgin) RMB 300,000–400,000/t (imported 550,000–1,000,000) 0% Stable to slightly weak →
    Carbon fibre T300-12K (Jilin) RMB 100/kg 0% (MoM +5.3%) Bottoming and recovering ↗
    Carbon fibre T700-12K RMB 105–135/kg 0% Range-bound at the bottom →
    PI film (electrical grade, 25μm) Uniaxial 110–170 / biaxial 170–220 RMB/kg +1% to +2% Confirmed uptrend ↑
    PI film (electronic grade) RMB 200–500/kg; high-end MPI approx. RMB 2.5m/t +2% to +3% Seller’s market ↑↑
    Fused zirconia RMB 33,250/t 0% Stalemate at highs →
    Zircon sand 65% / zirconium oxychloride RMB 11,300 / 19,000 per t Zircon sand easing Divergent ↘→
    Alumina (metallurgical grade) RMB 2,692.9/t -0.5% Weak and declining ↘
    [Cost driver] Brent crude USD 94.39/bbl +6.4% Sharp rally ↑↑
    [Cost driver] Anhydrous HF RMB 14,700–16,500/t 0% (+40% YTD) Firm at highs ↑
    [Cost driver] Yttria (China domestic) RMB 60.7/kg 0% (+21% YTD) Firm at highs ↑

    Notable Moves

    • Brent crude: +6.4% week on week. Brent settled at USD 94.39/bbl on 21 August and WTI at USD 87.06/bbl, a sixth consecutive session of gains and a three-week high. Crude transiting the Strait of Hormuz has fallen from a normal ~21.6 million bpd to 4.9 million bpd, with only 73 vessel transits in the week (versus 91 the prior week). The IEA now puts the Q3 global deficit at 1.8 million bpd, and observed global inventories fell 69 million barrels in July to below 7.9 billion barrels — the first time since April 2025. The US Treasury Secretary has said the “largest coordinated economic isolation in history” against Iran will be detailed on 24 August. This is the biggest variable this period: cost floors for acrylonitrile, dianhydrides/diamines and the fluorochemical chain will move up systemically, with pass-through expected within two to four weeks.
    • PI film: second round of 20%+ hikes confirmed. After a 20%–30% rise in Q2, prices went up more than 20% again in July–August — two rounds stacked. UBE has 80% of its electronic-grade capacity locked by downstream long-term contracts, and Kaneka, having raised global prices 20% on 16 April, is preparing another increase. The gap is structural: industry demand of 28,000 t against supply of 18,000 t. PSPI (DRAM/HBM passivation) demand is roughly 400 t in 2026, doubling to 800 t in 2027; MPI (memory packaging and optical modules) goes from 3,000–4,000 t to 7,000 t. Global supply sits with five or six offshore producers with no new capacity planned before 2030 — only incremental debottlenecking — and they prioritise Samsung, SK Hynix and TSMC. Shortage is expected to persist beyond 2028, with mainland China the tightest region.
    • PTFE: structural divergence widening. Mainstream suspension medium granule is under pressure at RMB 44,000–46,000/t while Shandong low-end quotes remain at RMB 31,800/t — an intra-month spread above 40%. Upstream fluorspar and anhydrous HF stay firm and R22 quota controls keep supply tight, so the cost floor holds, but pass-through remains blocked. Leading producers have softened list prices modestly, mid-size and small plants face growing destocking pressure and are discounting more actively. Commodity powder is falling hardest; high-end electronic and lithium-battery fine powders are relatively resilient.
    • Zirconia: high and low end fully decoupled. Zircon sand 65% has eased to about RMB 11,300/t and zirconium oxychloride holds at RMB 19,000/t, yet high-end yttria-stabilised zirconia is on an independent uptrend. Japan’s Tosoh has suspended dental-grade powder supply after losing yttria feedstock, creating a shortfall of roughly 6,000 t/year at the high end, and Sinocera raised zirconia powder prices 10%–40% effective 27 July. Year to date, zircon sand is up 17%, zirconium oxychloride 36% and yttria 21%.
    • Carbon fibre: flat on the week, recovering on the month. Jilin T300/12K stands at RMB 100/kg (+5.3% MoM) and T300/25K at RMB 90/kg (+5.9%); Jiangsu T700/12K is flat at RMB 105/kg while Guotai Dacheng offers T700/12K at RMB 135/kg. Industry cost is RMB 114,145/t (+3.2% MoM) with gross margin still negative at RMB -10,002/t. Inventory has eased slightly to 13,230 t, July output reached 11,295 t (+9.5% MoM) and utilisation was 69.8%. The rebound is cost-driven repair, not demand-led.
    • PEEK: flat. Domestic virgin resin is RMB 300,000–400,000/t and imported material RMB 550,000–1,000,000/t. The humanoid-robot narrative has cooled, and continued domestic capacity ramp-up at Zhongyan and Jida plus import substitution has rebalanced supply and demand.

    Impact Analysis

    Procurement cost. The oil rally will lift acrylonitrile, BPDA/PMDA dianhydrides and diamines within two to four weeks, feeding directly into PI film and carbon fibre precursor costs. PI film is the largest cost item this period and electronic grades have entered a “price without volume” phase — packaging, optical-module and AI-server BOMs must be repriced immediately (PI content is worth roughly RMB 20 per 800G optical module and RMB 32.7 per 1.6T unit). Commodity PTFE and metallurgical alumina remain low and are the only categories where cost can still be squeezed this period. High-end zirconia, by contrast, directly raises costs for dental, MLCC and PCB grinding-media applications.

    Supply chain. PI film — especially PSPI and MPI — and high-end yttria-stabilised zirconia are seller’s markets with lengthening lead times, and offshore capacity is reserved for offshore majors, so Chinese buyers must secure volume via long-term contracts and prepayment. Buyers retain leverage in PTFE, but the same strategy cannot be applied to commodity and electronic grades alike. Carbon fibre has ample domestic capacity and the best supply elasticity, with part of Zhongfu Shenying’s 30,000 t Lianyungang project already onstream. Tanker risk premiums will simultaneously raise landed costs and insurance on imported material.

    Actionable Recommendations

    Materials to lock in now

    • PI film (electronic grade / MPI / PSPI) — sign three- to six-month contracts or an annual framework. Securing volume matters more than securing price right now; a 10%–15% premium for allocation is acceptable. Start qualifying a domestic second source in parallel (Rui Hua Tai’s MPI has passed certification at two memory makers and already ships into optical modules).
    • High-end zirconia powder (yttria-stabilised / dental grade) — Tosoh’s 6,000 t/year gap remains unfilled; lock Q4 allocation with Sinocera or another domestic leader as soon as possible.
    • Carbon fibre T700-12K — cost is up 3.2% MoM with industry margins negative, and RMB 105/kg sits close to cash cost. Lock three to six months forward.
    • PTFE dispersion resin and high-end electronic fine powder — producer inventories are low and a catch-up move with HF is likely. Build two to four weeks of cover early.

    Materials to monitor

    • PTFE commodity suspension medium granule — the weak, choppy phase has not ended and the supply-demand balance remains soft. Wait two to four weeks for confirmed stabilisation and build inventory in tranches rather than all at once.
    • Alumina (metallurgical grade) — supply strong, demand weak. September is expected to range-trade low at RMB 2,580–2,780/t (monthly average near 2,680). Two to three weeks of safety stock is sufficient.
    • PEEK standard industrial grade — domestic capacity keeps expanding and the price centre is drifting lower, so buy to need. Only aerospace- and medical-certified grades warrant annual framework agreements.

    Signals to watch closely: daily vessel transits through the Strait of Hormuz and Iranian export loadings; the US sanctions detail due 24 August; China’s fuel price adjustment window at midnight on 28 August (institutions estimate an increase of about RMB 370/t, equal to RMB 0.27–0.30 per litre of gasoline); further developments in yttria export controls; and the timing of Kaneka’s next PI price increase letter.

    This report is compiled from publicly available market data for reference only. Procurement decisions should factor in your own inventory position, payment terms and supplier relationships.

    Market Intelligence Officer | New Materials Price Trend Monitoring | 2026-08-24

  • 2026-08-24 新材料价格趋势日报｜布油破94美元成本重启，PI薄膜第二轮提价20%+，PTFE牌号分化加剧

    2026-08-24 价格趋势日报

    核心结论：布伦特原油单周暴涨6.4%突破94美元/桶，全链条成本底系统性上移；PI薄膜7-8月第二轮提价20%+落地，是本期唯一真正的卖方市场；PTFE牌号分化加剧，通用粉料承压、电子级抗跌；氧化锆高低端彻底脱钩。

    价格概览表

    材料 当前价格区间 周环比 趋势
    PTFE树脂（悬浮中粒） 4.4-4.6万元/吨（山东低端3.18万） -1%~0% 弱势震荡 ↘
    PTFE树脂（悬浮细粉） 4.7-5.0万元/吨 0% 相对抗跌 →
    PTFE分散树脂 / 分散乳液 4.4-4.6万 / 2.8-3.0万元/吨 0% 稳定 →
    PEEK树脂（国产纯树脂） 30-40万元/吨（进口55-100万） 0% 稳中偏弱 →
    碳纤维 T300-12K（吉林） 100元/kg 0%（月环比+5.3%） 触底回升 ↗
    碳纤维 T700-12K 105-135元/kg 0% 底部盘整 →
    PI薄膜（电工级25μm） 单拉110-170 / 双拉170-220元/kg +1%~+2% 确认上行 ↑
    PI薄膜（电子级） 200-500元/kg；高端MPI约250万元/吨 +2%~+3% 卖方市场 ↑↑
    电熔氧化锆 3.325万元/吨 0% 高位僵持 →
    65%锆英砂 / 氧氯化锆 1.13万 / 1.9万元/吨 锆英砂回落 分化 ↘→
    氧化铝（冶金级） 2692.9元/吨 -0.5% 弱势下行 ↘
    ［成本项］布伦特原油 94.39美元/桶 +6.4% 急涨 ↑↑
    ［成本项］无水氢氟酸 1.47-1.65万元/吨 0%（年内+40%） 高位坚挺 ↑
    ［成本项］氧化钇（国内） 60.7元/kg 0%（年内+21%） 高位坚挺 ↑

    重点变动

    • 布伦特原油：+6.4%（周） —— 8月21日收94.39美元/桶，WTI 87.06美元/桶，连涨六个交易日、创三周新高。霍尔木兹海峡原油通行量已从常态约2160万桶/日降至490万桶/日，周船次仅73艘（前周91艘）；IEA测算Q3全球缺口扩至180万桶/日，7月全球可观测库存单月降6900万桶、跌破79亿桶（2025年4月以来首次）。美财长称8月24日将公布”史上规模最大”对伊经济孤立方案。这是本期最大变量：丙烯腈、二酐/二胺、氟化工链的成本底将系统性上移，预计2-4周向下游传导。
    • PI薄膜：第二轮提价20%+确认 —— Q2已涨20%-30%，7-8月再度提价20%以上，两轮叠加。宇部兴产80%电子级产能被下游长协锁定；钟渊化学4月16日全球提价20%后正酝酿新一轮上调。缺口是结构性的：全行业需求2.8万吨对供给1.8万吨；PSPI（DRAM/HBM钝化层）2026年需求约400吨、2027年翻倍至800吨，MPI（存储封装/光模块）3000-4000吨→7000吨；全球供给集中于海外5-6家厂商且2030年前无新增产能计划，仅靠技改小幅补量，优先保障三星、海力士、台积电。缺货预计持续至2028年之后，中国大陆货源最紧。
    • PTFE：结构性分化加剧 —— 主流悬浮中粒4.4-4.6万元/吨承压下跌，山东低端报价仍在3.18万元/吨，月内高低价差超40%。上游萤石、无水氢氟酸坚挺，叠加R22配额管控致货源偏紧，成本底稳固但传导持续受阻；头部挂牌价小幅松动，中小厂去库压力加大、让利出货增多，通用粉料跌幅明显，高端电子级与锂电细粉相对抗跌。
    • 氧化锆：高低端彻底脱钩 —— 65%锆英砂回落至约1.13万元/吨，氧氯化锆稳于1.9万元/吨，但高端钇稳定氧化锆走独立上涨行情：日本东曹因氧化钇断供暂停齿科粉体供应，形成年约6000吨高端缺口；国瓷材料自7月27日起氧化锆粉体涨价10%-40%。上游锆英砂年内+17%、氧氯化锆+36%、氧化钇+21%。
    • 碳纤维：周内持稳、月度回升 —— 吉林T300/12K 100元/kg（月环比+5.3%）、T300/25K 90元/kg（+5.9%），江苏T700/12K 105元/kg持平，国泰大成T700/12K报盘135元/kg。行业成本11.41万元/吨（月环比+3.2%），毛利仍为-10002元/吨；库存1.32万吨小幅去化，7月产量1.13万吨（环比+9.5%）、开工率69.8%。反弹属成本修复性，非需求拉动。
    • PEEK：持平 —— 国产纯树脂30-40万元/吨，进口55-100万元/吨；人形机器人题材热度回落，中研、吉大等国产产能持续爬坡叠加进口替代，供需重新平衡。

    影响分析

    对采购成本的影响：油价急涨将在2-4周内推高丙烯腈、BPDA/PMDA二酐、二胺等中间体，直接抬升PI薄膜与碳纤维原丝成本；PI薄膜已是本期最大成本项，电子级进入”有价无量”阶段，封装、光模块、AI服务器相关BOM必须立刻重估报价（800G光模块单只PI价值量约20元、1.6T约32.7元）。PTFE通用料与冶金级氧化铝仍处低位，是本期唯一可压成本的品类。高端氧化锆则直接推高齿科、MLCC、PCB研磨介质成本。

    对供应链的影响：PI薄膜（尤其PSPI/MPI）与高端钇稳定氧化锆已是卖方市场，交期拉长且海外产能优先保障境外大客户，国内买方必须以长协+预付锁量；PTFE买方议价权仍在，但通用料与电子级不可套用同一策略；碳纤维本土产能充裕、供应弹性最好，中复神鹰连云港3万吨项目部分产线已投产；油运风险溢价将同步推高进口料到岸成本与保险费用。

    行动建议

    建议锁定价格的材料

    • PI薄膜（电子级/MPI/PSPI） —— 签3-6个月长协乃至年度框架，此时锁量优先于锁价，可接受10%-15%溢价换配额；同步启动国产二供验证（瑞华泰MPI已通过两家存储厂认证、光模块已实际供货）。
    • 高端氧化锆粉体（钇稳定/齿科级） —— 东曹6000吨/年缺口未补，建议尽快锁定国瓷等国内龙头Q4配额。
    • 碳纤维T700-12K —— 成本月环比+3.2%、行业毛利为负，105元/kg接近现金成本底，建议锁3-6个月远期。
    • PTFE分散树脂与高端电子细粉 —— 厂家库存低位，随氢氟酸补涨概率大，可提前备2-4周量。

    建议观望的材料

    • PTFE通用悬浮中粒 —— 弱势震荡未止、供需格局偏弱，等2-4周确认企稳，逢低分批建库而非一次性备货。
    • 氧化铝（冶金级） —— 供强需弱，9月预计2580-2780元/吨低位盘整（月均约2680），维持2-3周安全库存即可。
    • PEEK标准工业级 —— 国产扩产持续、价格中枢下移，按需采购；仅航空、医疗认证牌号需签年度框架协议。

    需重点盯防的信号：霍尔木兹海峡日通行船次与伊朗出口装载量；8月24日美方对伊制裁细则；8月28日24时国内成品油调价窗口（机构预估上调约370元/吨，折合汽油+0.27~0.30元/升）；氧化钇出口管制后续动态；钟渊化学新一轮PI提价函落地时点。

    本报告基于公开市场数据整理，仅供参考。采购决策请结合自身库存水位、账期与供应商关系综合判断。

    市场情报官 | 新材料价格趋势监控 | 2026-08-24