结合这两篇文献中器件的制备制备过程,我想要制备顶栅器件,图片是我要汇报的内容,要求完善衬底制备和功能...

创建于:2025年10月11日

使用 GPT-5 Thinking 回答 Chat01

问题

Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf
2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf
image.pngimage.png

结合这两篇文献中器件的制备制备过程,我想要制备顶栅器件,图片是我要汇报的内容,要求完善衬底制备和功能层沉积两个环节的工艺流程

思考

Creating detailed process flow

I need to create a process flow for two key steps: substrate preparation and functional layer deposition for a top-gate device. The materials involved include NbO2 and MoS2, based on two papers. The image adds to the complexity with mentions of W, Pt, and TiN electrodes, indicating that the goal is a planar NbOx device.

Outlining process for top-gate device

I need to combine steps for substrate prep and functional layer deposition. For substrate prep, I’ll use a single-crystal wafer with thermal oxidation to form 100 nm SiO2. For the functional layer, I'll define a deposition window using lithography, deposit NbOx via PVD, and address details like cleaning, resist types, deposition conditions, thickness, and annealing. For the top-gate, I'll reference a Gate-last process, using materials like ALD HfO2 and gate metallization options like Au or Al.

Combining deposition techniques for top-gate

The NbO2 paper uses sputtering at 750°C on r-plane sapphire, followed by optical lithography and CF4/Ar dry etching before contacting with Pd/Al. Although it's a two-terminal device, I can adapt those steps for our top-gate structure, possibly using NbOx as the channel and HfO2 for the gate dielectric as suggested.

Proposing process steps

To outline the steps, I'll suggest using either Si/SiO2 or sapphire substrates for preparation. For functional layer deposition, I could propose sputtering NbO2, followed by seeding and ALD HfO2 for the gate dielectric, potentially adding TiN or W for gates. I'll remember to define windows before deposition as well.

Defining deposition process

For substrate prep, I'll propose using Si/SiO2 or sapphire, followed by RCA cleaning and thermal oxidation to grow 100 nm SiO2. For functional layer deposition, I'll suggest photolithography to define windows and PVD sputtering of NbOx. We'll clarify deposition conditions, including temperature and annealing to ensure proper material formation.

Thought for 1m 17s

回答

下面给你把“衬底准备”和“功能层沉积”两个环节补全为可直接照着跑的工艺流程卡(顶栅结构、后续要接 ALD 高-k 与金属栅)。我参考并融合了你给的两篇文献中的做法:顶栅栈的“播种层+ALD-HfO₂(8 nm 最优)+金属栅(Au/Al)”来自 MoS₂ 顶栅工艺;NbO₂/NbOₓ 薄膜的制备与刻蚀参数来自 Mott 神经元工作。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf) (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)


衬底准备(Si(100) / 100 nm SiO₂)

  1. 来料与前烘

    • 4-inch 单晶 Si(100),电阻率任意;200 ℃,5 min 脱水烘。
  2. 有机去污 + 颗粒去除

    • Piranha:H₂SO₄:H₂O₂=3:1,120 ℃ 10 min → DI 冲洗。
    • RCA-SC1:NH₄OH:H₂O₂:DI=1:1:5,75–80 ℃ 10 min → 冲洗。
    • RCA-SC2:HCl:H₂O₂:DI=1:1:6,75–80 ℃ 10 min → 冲洗。
  3. 生长 100 nm SiO₂(热氧化)

    • 湿氧化 1000 ℃,~20–30 min(以椭偏实测为准,目标 100 ± 5 nm)。
    • 干氧化收尾 5–10 min 稍作致密化。
    • N₂ 900 ℃ 30 min 退火(降低界面态)。
  4. 对准标记(可选,便于后续顶栅对准)

    • HMDS 处理 → 正胶(如 S1813)旋涂 4000 rpm 60 s,软烘 100 ℃ 1 min。
    • 光刻显影后溅射 Ti/Au=5/50 nm → Lift-off。
  5. 表面预处理(为后续 ALD 做准备)

    • O₂ 等离子 30 s(50 W)去轻微有机残留。
    • 备用:若计划直接在功能层上 ALD 高-k,可保留此步到功能层完成后。

说明:顶栅电介质我们采用“播种层 + ALD HfO₂(5/8/12 nm 比较后选 8 nm)”的做法;这是文献中验证过的短沟道顶栅最优栈结构,后续你在“栅介质沉积”页里只需接上这套参数即可。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)


功能层沉积(NbOₓ / NbO₂ 通道/有源层)

**推荐路线:先沉积整片 NbO₂,再光刻+干法刻蚀定义窗口/条带。**这样避免高温沉积与 Lift-off 的工艺冲突,并与后续 ALD-HfO₂ 顶栅兼容。该路线与文献 NbO₂ 器件的一致。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

A. 薄膜沉积(磁控溅射 NbO₂)

  • 设备/靶材:AJA/UHV 磁控溅射;NbO₂(或 Nb 金属靶 + 反应气 O₂)。
  • 基底温度:~750 ℃(晶化 NbO₂,文献方法);若工具/版图限制高温,可室温沉积后 650–800 ℃ 气氛退火晶化,见“低温备选”方案。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • 气氛:Ar 主气;若用 Nb 金属靶,配 O₂(几 sccm 級,闭环控电阻或光谱做成分反馈)。
  • 压力:数 mTorr 量级;
  • 厚度:100–200 nm(文献范围;150 nm 常用)。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • 冷却:随炉降温,避免热冲击致裂。

B. 图形化窗口(定义功能层区域)

  • 光刻:正胶(1.3 µm 级别),常规曝光/显影;
  • 刻蚀配方(CF₄/Ar 干法刻蚀)
    • CF₄ 60 sccm / Ar 30 sccm,腔压 5 mTorr,ICP 500 W,偏置 100 W,速率≈90 nm/min;刻蚀至露出 SiO₂;去胶。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

C. 表面清洁与“播种层”以利后续 ALD 高-k

  • 轻度 O₂ 等离子 10–20 s 去残胶;
  • 播种层:电子束蒸发极薄 (~1–2 nm) 高-k 亲核化材料/超薄金属氧化物(与 MoS₂ 顶栅相同思路),用于提升随后 ALD-HfO₂ 的均匀成核。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)

到此,“功能层沉积”结束;下一页即可接:ALD HfO₂ 8 nm → 顶栅金属(Au 35 nm 或 Al(5 nm)/Au(30 nm) 等,二者可用于阈值工程)→ 后续刻蚀/开接触等。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)


低温备选(若不能高温溅射)

  • 室温反应溅射 NbOₓ(x≈2)100–150 nm后退火晶化/定相
    • 650–800 ℃,Ar/O₂(或真空)30–60 min,使薄膜向 NbO₂ 相转变并提高导电/阈值稳定性(依据片上四探针与拉曼确认)。该路径用于兼顾前道对准金属或光刻胶耐温限制。(思路来自 NbO₂ 文献,但那篇直接 750 ℃沉积;这里给出工程化折中以适配 Si/SiO₂ 平台。) (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

你汇报页可直接落的两条 bullet(精炼版)

  • 衬底准备:Si(100)→Piranha/RCA→热湿氧 1000 ℃ 生长 100 nm SiO₂→N₂ 900 ℃ 致密化→(可选)Ti/Au 对准标记→O₂ 等离子预处理。
  • 功能层沉积:UHV 溅射 NbO₂ 100–200 nm(750 ℃)→光刻→CF₄/Ar 干刻开窗口(90 nm/min)→去胶→O₂ 等离子轻清洗→播种层(e-beam 1–2 nm)以便后续 ALD-HfO₂ 8 nm 顶栅介质与金属栅沉积。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf) (Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)

这样写既对齐你图上的两大环节,又把关键“可抄的参数/做法”都给到了:NbO₂ 的沉积与刻蚀配方来自 Mott 神经元文献;顶栅栈(播种层-HfO₂(8 nm)-栅金属)的选择与厚度来自 MoS₂ 顶栅工作。后续若需要,我可以把“栅介质与金属栅”“开源漏接触/退火”也补成同样风格的一页流程卡。

问题

CF₄/Ar 干刻开窗口的作用是什么

思考

Explaining CF₄/Ar dry etch function

The question is about the role of CF₄/Ar RIE in opening a window in NbOx planar devices. It’s used to pattern NbO₂ or NbOx films, etching away material to expose SiO₂ underneath. The CF₄/Ar etch is selective, providing smooth sidewalls and anisotropy. While CF₄ traditionally etches SiO₂, it etches NbO₂ through reaction to form volatile NbF₅.

Understanding window creation

The purpose of using CF₄/Ar dry etching is to open a window in NbO₂ where it's not needed, allowing for the deposition of a functional material in the central active region. The user’s process flow suggests a lift-off style, where resist is patterned and functional material is deposited in the window. However, our earlier suggestion was to first deposit and then etch.

Clarifying CF₄/Ar etch purpose

The user’s question is about the role of CF₄/Ar dry etching to "open windows" in NbO₂. This step is used to pattern the functional layer with a vertical profile and residue-free etching, while ensuring selectivity to SiO₂. They might also be considering opening contact windows in HfO₂, though HfO₂ etching is tricky and typically needs different gases. This clarifies the purpose of the CF₄/Ar etch in defining areas for deposition.

Purpose of CF₄/Ar etch

The CF₄/Ar dry etching step is used to etch NbO₂, creating a trench or window that defines the active region of the device. This isolates devices, exposes SiO₂ for later steps, and creates clean vertical edges for precise dimension control. It ensures minimal undercut and prevents leakage by isolating the conduction path for subsequent deposition.

Thought for 1m 0s

回答

简短结论:
CF₄/Ar 干刻“开窗口”的目的,就是把不需要的 NbOₓ/NbO₂ 薄膜各向异性地刻掉、露出下方的 SiO₂(或其他底层)并用清晰的侧壁来定义器件的有效功能区/沟道几何尺寸,从而实现器件隔离与版图精确控制。在这一步里,CF₄ 产生的 F 自由基与 Nb 形成挥发性的 NbFₓ(助刻蚀),Ar 提供物理轰击(定向、提速),两者协同得到可控速率和接近垂直的侧壁。该做法正是文献中对 NbO₂ 薄膜通道图形化的标准流程;他们用 CF₄/Ar(60/30 sccm, 5 mTorr, 500 W ICP/100 W bias)把 NbO₂ 刻成条带后再做电极与测量(速率≈90 nm/min)。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf) 另外,在 MoS₂ 顶栅工艺里也同样通过 RIE 先把通道区域干法刻蚀出来,以保证短沟道尺寸的精确与一致性(虽未指定气体,但目的完全一致)。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)


这一步“开窗口”的具体作用点

  • 版图定义与隔离:决定沟道的 L、W,并把功能层从非器件区移除,避免旁路/泄漏。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • 露底层作后续工艺基准:刻透 NbOₓ/NbO₂ 后露出 SiO₂,便于后续 ALD 高-k 栅介质覆盖在目标区域以及进行顶栅对准。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • 侧壁质量与良率:CF₄ 的化学反应 + Ar 的物理去除 → 各向异性更好、残留更少,减小边缘毛刺带来的电场集中与击穿风险。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • 工艺可控:有明确的刻蚀速率(~90 nm/min)与稳健窗口,便于设定轻微过刻以确保全刻透而不严重损伤 SiO₂。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

小提示:如果你选“沉积后刻蚀”的路线,就需要这一步 CF₄/Ar 来开窗口/定图形;若改用“光刻开胶窗→PVD 沉积→lift-off”的前图形化(lift-off)路线,则不需要对 NbOₓ 再做这道干刻。两种路线目的相同,都是为了在指定区域只保留功能层,但干刻法在薄膜致密、厚度较大或后续需要垂直边界时更稳健。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf) (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

如果你要把这句加到汇报页里,可以用一句话概括:
CF₄/Ar RIE 开窗口:选择性去除 NbOₓ/NbO₂,露出 SiO₂ 并精确定义沟道(F 化学反应 + Ar 物理轰击,速率~90 nm/min),为后续 ALD 栅介质/金属栅提供洁净、各向异性的图形边界。” (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

问题

B. 图形化窗口(定义功能层区域)

光刻:正胶(1.3 µm 级别),常规曝光/显影;

详细解释这一步骤

问题

B. 图形化窗口(定义功能层区域)

光刻:正胶(1.3 µm 级别),常规曝光/显影;

详细解释这一步骤

思考

Explaining photolithography step

The user wants a detailed explanation of photolithography with positive resist (1.3 µm). Key steps include resist choice (like S1813, AZ1512), spin parameters, softbake, priming, exposure (i-line 365 nm, dose 80-120 mJ/cm²), and post-processing; developing with MF-319 or AZ MIF 726. Hardbake, O2 descum, and thickness checks follow. This process involves CF4/Ar etching in NbOx films.

Considering resist thickness and etching

The resist thickness (1.3 µm) must withstand reactive ion etching (RIE). I need to check NbO2 selectivity to resist in CF4/Ar. Rating etch rate and resist selectivity (1:1–1:2) helps adjust this. Hard bake and adhesion steps are key, plus plasma descum to avoid residues. The user requested optical photolithography, so I will focus on that and include relevant citations.

Citing relevant sources

We need to cite relevant points from the documents we searched. The MoS2 paper mentions e-beam lithography, but doesn't cover photoresist specifics. The NbO2 paper mentions optical lithography before etching, so I'll cite that. I'll stress that parameters like S1813 are standard defaults and emphasize safety. We'll provide step-by-step details and a simplified version for presentation.

Keeping the response concise yet detailed

To keep the response concise, I’ll structure the explanation with bulleted steps, giving detailed information without overwhelming. I'll cite the MoS2 document for e-beam lithography and channel etching, even though the user asked for standard RIE, which is still relevant for the process. The NbO2 paper gives CF4/Ar parameters, so I’ll reference those too.

Organizing the response with citations

I'll begin with the goal and list the photolithography steps, from dehydration bake to descum. I'll reference the MoS2 paper for the RIE channel etching step and the NbO2 paper for optical lithography before CF4/Ar dry etching. I'll then summarize with a succinct bullet point for a presentation slide. The citations will follow at the end.

Thought for 1m 12s

回答

下面把“B. 图形化窗口(光刻:正胶 1.3 µm,常规曝光/显影)”写成一张能照着跑的工艺流程卡,并在每步标注目的与关键参数。所述光刻用于在NbOₓ/NbO₂ 功能层上开窗口,随后进入 CF₄/Ar 干刻(文献中就是“光刻→CF₄/Ar 干刻→成形 NbO₂ 条带”这一路线)。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
(对照:MoS₂ 顶栅流程图中亦先做通道图形化/刻蚀再沉积栅叠层,见其 Fig.1 与“channel etching”步骤。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf))


光刻(正胶 1.3 µm)标准步骤与参数

  1. 衬底前处理(提高附着/洁净)

    • 200 ℃ 脱水烘 5–10 min → 冷却至 <60 ℃。
    • HMDS 吸附(蒸汽或滴涂),室温 2–5 min;氮气吹干。
      目的:增强正胶对 SiO₂/NbOₓ 的附着,避免显影后“起皮”。
  2. 旋涂正胶(目标厚度≈1.3 µm)

    • 典型胶:S1813 / AZ1512 / AR-P 3510(三者任选其一)。
    • 4000 rpm × 60 s(加速 1000 rpm/s)→ 厚度约 1.3 µm;边缘去胶(EBR)。
      目的:提供足够的刻蚀掩膜厚度;1.3 µm 对 NbO₂(100–200 nm)干刻有充裕余量。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  3. 软烘(Soft Bake)

    • 100–110 ℃ 热板 60–90 s(或 90 ℃ 2 min);取出冷却 1 min。
      目的:挥发溶剂,稳定胶膜,减少后续站波。
  4. 对准与曝光(i-line 365 nm,接触/近接触)

    • 对准:使用前道 Ti/Au 对准标记;将沟道 L/W 与版图重合。
    • 曝光模式:软接触/硬接触(>1 µm 线宽建议软接触即可)。
    • 曝光剂量(参考起点)
      • S1813:80–120 mJ/cm²;AZ1512:** 50–80 mJ/cm²**。
    • 焦距/剂量矩阵的首片验证(CD 目标=沟道窗口的版图尺寸)。
      目的:把需要开窗口的位置在正胶上显影出来,作为后续干刻的掩膜。
  5. (可选)后烘(PEB)

    • 110 ℃ 60 s(部分正胶可省略)。
      目的:减弱站波、改善边壁;保持最小收缩。
  6. 显影(0.26 N TMAH 金属离子免疫显影液)

    • MF-319 / AZ MIF 726,20–25 ℃,45–70 s;轻微摆动或超声 10% 功率。
    • DI 冲洗 ≥30 s → N₂ 吹干。
      正胶原理:曝光区溶解 → 形成开口/窗口;未曝光区保留掩膜。
  7. 检查与微调

    • 显微镜/测量:确认最小线宽、边壁与对准误差(目标<±0.2 µm)。
    • CD 偏差修正:若窗口偏小→增剂量/延时显影;偏大→减剂量/缩短显影。
    • 表面清洁:O₂ 等离子 10–20 s, 30–50 W 轻度 descum 去除残胶丝。
      目的:得到干净、垂直的窗口边界,避免干刻后残留。
  8. 硬烘(Hard Bake,提升刻蚀抗性)

    • 115–120 ℃ 2–3 min(避免 >130 ℃ 以免胶边流动)。
      目的:提高抗等离子体侵蚀能力,保证后续 CF₄/Ar 干刻窗口形貌。

到此,光刻完成:版图中“开口”处露出 NbOₓ/NbO₂ 薄膜,非开口处由 1.3 µm 正胶保护。下一步直接进入 CF₄/Ar RIE 刻蚀将开口处的 NbOₓ/NbO₂ 去除、露出 SiO₂,正如文献的“光刻→CF₄/Ar 干刻”工艺链。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
MoS₂ 顶栅流程同样在“channel etching”步完成几何定义(见其 Fig.1/流程图),与这里的目的完全一致。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)


现场要点与常见问题

  • 选择正胶而非负胶:开窗口面积通常大于保留区,正胶显影更稳健,且后续若转为 lift-off 路线也通用。
  • 抗刻蚀裕量:以 NbO₂ ~90 nm/min 的刻蚀速率估算,刻 2 min(含 20–30% 过刻)仍远低于 1.3 µm 胶膜厚度,安全。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • 站波/反射:若边壁“锯齿”,可加 PEB 或用 ARC(底部减反层)优化。
  • 缺胶/起皮:多由脱水/HMDS 不足或显影过冲,复查步骤 1 与 6 的时间。

汇报页可用一句话(精炼版)

“光刻(正胶 1.3 µm):HMDS 处理→旋涂/软烘→i-line 对准曝光(80–120 mJ/cm²)→TMAH 显影→O₂ 轻 descum→硬烘,得到干净、垂直的窗口以供后续 CF₄/Ar 干刻定界沟道/功能区。”(工艺链与文献的光刻→干刻一致。)(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf) (Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)

如果你提供所用机台型号/光强,我可以把剂量与时间再细化成本线专用配方表

问题

image.pngimage.png

结合文献内容,继续完善这张ppt

思考

Considering electrode materials

For the NbO2 paper, they used optical lithography and CF4/Ar etch for patterning, with Pd/Al contacts. The deposition process employed AJA UHV sputtering at 750°C on sapphire. We’re considering adapting this by using TiN/Pt layers for our electrodes, with TiN providing a diffusion barrier and Pt offering good contact to NbOx, particularly for Schottky barriers.

Focusing on provided PDFs

I’ll stick to the provided PDFs and avoid bringing in literature from other sources. For the electrode materials, we can propose TiN/Pt as a CMOS-compatible option, but I'll clearly note that this suggestion is not from the papers. For the process flow, I’ll present generic instructions for EBL patterning, deposition, and gating, without making unsupported claims.

Proposing modifications for PPT

I can suggest using 8 nm HfO2 based on MoS2 results, but if they prefer HfZrO2 10 nm, I’ll mention that MoS2 shows 8 nm HfO2 reduces leakage and that HfZrO2 might be more suited for NC-FETs, though not in the papers. I'll propose modifications to the PPT with proper references, including seeding layers, gate metals, and TiN gating.

Presenting options for gating materials

For the threshold tuning, we can offer Au or Al (low work function) as options, but the slide proposes a 50 nm TiN gate. While TiN may give a different Vth, the papers suggest Au/Al as gating materials. I’ll present these as choices, ensuring the style stays consistent and clear. We’ll also provide detailed, generic process parameters for EBL.

Refining PPT content

For electrode creation, we’ll include step-by-step instructions: EBL patterning with detailed resist system options and development, followed by metal deposition (EBE), offering TiN/Pt or Pt options, and liftoff procedures. The top-gate process will have HfO2 or HZO options, depending on their choice. We’ll also add gating metal selections, citing data on Vth modulation with Au and Al. Additionally, references to figures like "Fig.1" and "Fig.3" will be included with simplified citations. Finally, content will be formatted for easy slide inclusion.

Thought for 1m 21s

回答

太棒了,这页已经很清晰了。我根据两篇文献把**“电极制作 / 栅极制作”**补成能直接落地的参数版,并在每条后面标注依据。你可以直接把粗体小标题和要点拷到 PPT 里用。


电极制作(S/D)

目的: 定义亚微米~微米级电极间隙,获得低接触电阻与可复现的边界。

  • 版图与写胶
    • 建议:PMMA 950 k A4(≈200 nm)或 ZEP520A(≈300 nm),180 ℃×2 min 软烘。
    • 适用性:EBL 用于亚微米沟道/间隙定义(文献在顶栅短沟道器件中用 EBL 定子源/漏与沟道;其 SEM 展示 LCH=200 nm 的清晰边缘)。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • EBL 暴光/显影(参考起点)
    • 剂量:PMMA 250–350 µC cm⁻²;显影 MIBK:IPA=1:3,60 s → IPA 冲洗 30 s。
    • 显影后 O₂ 轻 descum(30–50 W,10–20 s)去残胶丝,保证后续金属连续性。
  • 金属沉积(EBE)
    • 你方案里写的是 TiN / Pt 各 30 nm。实操建议:TiN(10–20 nm) 作阻挡/黏附 + Pt(20–30 nm) 作导体;速率 0.5–1 Å s⁻¹,基压 <5×10⁻⁷ Torr。
    • 旁证参考:两篇文献均采用 EBE+lift-off 的电极工艺;MoS₂ 采用 Au 35 nm(EBE),NbO₂ 器件采用 Pd(100 nm)/Al(400 nm)(EBE)+光刻 lift-off。思路与设备路径一致。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf) (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • Lift-off
    • NMP 60 ℃ 30–60 min 或丙酮室温浸泡;轻微超声协助;DI 冲洗、N₂ 吹干。
  • (可选)接触退火
    • 若仅接触电阻偏大:N₂ 200–300 ℃、≤30 min 评估;不宜高于 ~350 ℃ 以避免 NbOₓ 氧化态变化(保守建议)。

说明:若你的电极间隙不需要亚微米,可改用光刻 + lift-off。NbO₂ 文献即用光刻 + CF₄/Ar 干刻做通道图形,再 EBE 蒸发 Pd/Al 接触,验证了这条工艺链的可行性。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)


栅极制作(顶栅,Gate-last)

目的: 在功能层上形成致密、低漏电的高-k 栅介质与可调阈值的金属栅。

  • 介质体系(强烈建议按文献优化替换)
    • 播种层(EBE,~1–2 nm)→ ALD HfO₂(8 nm):这是短沟道顶栅器件里兼顾低漏电栅控的最优厚度选择;文献比对了 5/8/12 nm,8 nm 显著抑制 IG 且保持高栅控与小 EOT。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
    • 如果你必须用 HfZrO₂ 10 nm(如考虑铁电/NC-FET):也可保留,但需单独评估漏电与可靠性(两篇参考文献未报告 HZO)。
  • ALD 条件(建议)
    • TMA/H₂O 或 TEMAH/H₂O,150–200 ℃,前置 播种层是为了在惰性表面上均匀成核(文献的做法)。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • 金属栅选择与阈值调控
    • Au 栅(35 nm):较高功函数,器件 VTH 偏正
    • Al 栅(5 nm Al / 30 nm Au):较低功函数,可将 VTH 负移。文献用 Au/Al 双方案 + W/L 调节 做到 E/D-NMOS 反相器配对。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
    • 你方案中的 TiN 50 nm 完全可行(CMOS 兼容、热稳定好),但阈值位置会和 Au/Al 不同;若目标是做增强/耗尽配对,建议保留 Au 与 Al 两种栅金属选项用于版图同片对比。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • 栅极版图/曝光与沉积
    • EBL(二次对准到 S/D),显影后 EBE 金属栅沉积 → lift-off。该“Gate-last”流程与文献 Fig.1 的顶栅工艺一致(先 S/D、刻沟道,再播种/ALD 高-k、最后金属栅)。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)

与功能层(NbOₓ/NbO₂)图形化的衔接要点

  • 若走“沉积整片 → 光刻 → CF₄/Ar 干刻开窗口”路线:在光刻后用 CF₄/Ar=60/30 sccm,5 mTorr,ICP 500 W/偏置 100 W,速率≈90 nm min⁻¹ 刻至露出 SiO₂,得到清晰侧壁,随后进入播种层/ALD 高-k。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)
  • NbO₂ 薄膜沉积范围 100–200 nm;文献采用 AJA 溅射,750 ℃ 生长并在光刻 + 干刻后制备电极与器件。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

可直接贴到 PPT 的精炼版(替换你图上的文案)

电极制作(EBL + EBE + Lift-off)

  • EBL 写入电极图形(PMMA/ZEP,亚微米间隙可达 200 nm 量级);显影后 O₂ 轻 descum。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • EBE 沉积 TiN(10–20 nm)/Pt(20–30 nm)(或参考文献的 Au 35 nm / Pd/Al 叠层);Lift-off 得到 S/D。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf) (2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

栅极制作(Gate-last)

  • 播种层 (~1–2 nm) → ALD HfO₂(8 nm):低漏电、强栅控(文献最优)。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
  • 金属栅(EBE):Au 35 nm(VTH 偏正)/ Al(5 nm)/Au(30 nm)(VTH 负移),可实现 E/D 配对;也可选 TiN 50 nm 做工艺对比。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)

备注(备查图):文献 Fig.1 给出了完整的顶栅 Gate-last 流程(含“播种层 + HfO₂ + 顶栅”),Fig.3 的 SEM 展示了 200 nm 沟道线边形貌;介质厚度优化与漏电对比见 Fig.4(8 nm HfO₂ 优于 5/12 nm);阈值调控与反相器结果见 Fig.8–10。(Optimization_of_Short-Channel_Top-Gate_MoS2_FETs_Via_a_Non-Transfer_Fabrication.pdf)
NbO₂ 薄膜的沉积温度、厚度CF₄/Ar 干刻参数详见文献正文与方法学部分,可直接复用到“功能层沉积/图形化”页。(2025-07-NE-An electro-optical Mott neuron based on niobium dioxide.pdf)

需要的话,我可以把这一页做成 对比版版式(你现有 TiN/HZO 流程 vs 文献 HfO₂/Au-Al 流程)并附一张“参数表(可编辑)”,方便你现场切换口径。

分享这个问答