Scientific contract and independent review

This is a generic educational demonstrator, not reproduction of a reported transistor or calibrated predictive device simulator.

Architecture

Single-layer 1H-MoS2 on SiO2, on heavily doped Si global back gate; Au top source/drain. The source is grounded, drain has VDS >= 0, gate has VGS. L is the inner contact-edge gap. Monolayer thickness and all exploded gaps are visually exaggerated. Pure Au contacts in this architecture class are reported by Ahn et al., 2017: https://pmc.ncbi.nlm.nih.gov/articles/PMC5481332/ . Their oxide is 285nm; this demonstrator's 300nm default is independently chosen, not attributed to that experiment.

Current model

Ideal charge-sheet/long-channel gradual-channel approximation: Cox=eps0 epsr/tox, beta=mu Cox W/L. Vov=VGS-Vth. Id=0 when Vov<=0; Id=beta(Vov VDS - VDS²/2) for 0<=VDS<Vov; Id=beta Vov²/2 for VDS>=Vov. This is a generic illustrative model applied to a 2D accumulation channel, not the p-body inversion structure of a silicon MOSFET. Default mobility 10cm²/Vs and threshold 1V are demonstration parameters, not verified material constants.

University derivations: https://lampz.tugraz.at/~hadley/psd/L10/gradualchannelapprox.php and https://ocw.mit.edu/courses/6-012-microelectronic-devices-and-circuits-fall-2009/resources/mit6_012f09_lec11_gradual/ . The application does not implement contact resistance/Schottky barriers, traps, quantum capacitance, subthreshold/leakage, velocity saturation, self-heating or a Poisson/quantum transport solver. Real contact limitations are documented in Liu et al., ACS Nano 2015 DOI10.1021/nn506512j: https://www.media.mit.edu/publications/impact-of-contact-on-the-operation-and-performance-of-back-gated-monolayer-mos2-field-effect-transistors/ . All displayed characteristics are model-generated on a linear current axis.

Atomic sheet

Ideal 1H trigonal-prismatic S–Mo–S. Triangular a1=(a,0), a2=(a/2,sqrt(3)a/2); Mo at 0 and S at (a1+a2)/3 in both upper and lower planes. a=3.16Å and Mo–S vertical h=1.565Å are approximate illustrative geometry, not one measured parameter set. Each interior Mo has three S neighbors above and three below. References: https://pmc.ncbi.nlm.nih.gov/articles/PMC4945536/ and https://qpeng.org/publication/pdf/Peng_2013_pccp_MoS2.pdf . Atomic view is a separate scale, not all atoms of the micron-scale device.

Animation and audit

Electron markers illustrate source-to-drain net drift for positive VDS; conventional current points drain-to-source. Markers remain in the MoS2 channel display plane, never in the oxide, and net motion stops at zero VDS or cutoff. Markers do not represent individual trajectories, actual carrier count or physical speed. No field arrows or electronic band diagram are claimed as computed outputs.

An independent AI reviewer checked these scientific assumptions, supplied numeric regression values, and reviewed model.mjs, model.test.mjs, device.mjs, index.html and this document. Four geometry/numerical tests also passed independently. The reviewer found two visual inconsistencies: spherical direction markers partly intersected the oxide, and longitudinal/transverse display scale differed. Corrections use zero-thickness planar direction markers within the channel and equal x/z scale factors; their full rendered bounds and aspect ratio are checked by the browser verification. Transient texture disposal was also repaired. This is an AI-assisted review, not certification by an experimental device specialist.

Regression: Cox=1.151044415664e-4 F/m² (11.51044nF/cm²); for VGS=5V and VDS=1V, Id=0.805731090965µA; for VGS=5V and VDS>=4V, Id=1.8416710650624µA.

Review closure

The independent AI reviewer rechecked the saved corrections and closed both required findings, with no remaining material blocker. Equal x/z device scale and planar markers were confirmed in the final code; texture/instance cleanup was also confirmed. Browser checks remain separate evidence, recorded in VALIDATION.md when complete.

Presentation refinement 0.4

The physical model source remains SHA256 152DB807FC683B8FE9B6F8707B2CB6C87835060C7C20CDB318C43E0592418AD8. Materials have illustrative procedural roughness/bump shading; these are not measured optical properties or surface topography. Contact bevels preserve the original external contact faces and source–drain gap (independently checked with the vendored geometry library).

Staggered display offsets separate contacts before channel and oxide and reassemble in reverse. The independent reviewer and new tests checked layer order across intermediate positions. Carrier heads and fading tails stay planar and confined; tails are removed at the source reset rather than drawn across the device. Any requested or actual display separation hides markers/arrows and pauses their phase. Curves continue to refer to the assembled ideal device, explicitly labeled in the view. The independent atomic inset has a scale-change transition rather than claiming a physical continuous device-to-atom simulation.

Independent AI review found no material blocker in these changes and independently passed all seven numerical, geometry and presentation tests. Browser checks are recorded separately.

二维器件实验室：科学模型与独立 AI 审核

这是通用单层 1H-MoS₂ 背栅晶体管的教学演示，不复现特定论文器件，也未经实验标定。

结构、坐标与接线

Au 顶部源漏接触单层 MoS₂；300 nm SiO₂ 将沟道与重掺杂 Si 全局背栅隔开。源极接地，VDS ≥ 0，栅极施加 VGS。L 是接触内缘间距。器件纵向厚度、分层间距为显示夸张；横向 x/z 使用一致缩放。Ahn 等报道了此类 Au 接触架构，但论文氧化层为 285 nm，本演示独立选取 300 nm。

电流模型与单位

采用理想电荷薄层、长沟道渐变沟道近似：Cox = ε0 εr / tox，β = μ Cox W/L，Vov = VGS − Vth。Vov ≤ 0 时 Id = 0；0 ≤ VDS < Vov 时 Id = β(Vov VDS − VDS²/2)；VDS ≥ Vov > 0 时 Id = β Vov²/2。内部使用 SI 单位，界面显示 µA、µm 和 nF/cm²。默认 μ = 10 cm²/(V·s)、Vth = 1 V 为演示参数，不是已核实的材料常数。

这是二维积累沟道，不采用硅 MOSFET 的 p 型体反型叙述。没有求解接触电阻、肖特基势垒、陷阱、量子电容、亚阈值和漏电、速度饱和、自热或 Poisson/量子输运。曲线为线性电流坐标上的模型计算结果。

独立原子尺度

原子视图是理想 1H 三棱柱配位 S—Mo—S 单层。a1=(a,0)，a2=(a/2,√3a/2)，Mo 位于原点，上下 S 的面内位置均为 (a1+a2)/3。a≈3.16 Å、S 高度 ±1.565 Å 为近似示意参数。内部 Mo 分别与上、下各三个 S 配位。原子视图采用独立 Å 尺度，不是微米器件的全原子模型，也未经 DFT 弛豫。

动画约定

正 VDS 时，电子净漂移从源到漏，常规电流相反。平面标记始终位于 MoS₂ 沟道显示平面，不进入氧化层。零漏压或截止时停止净漂移；只要请求或实际存在分层，标记和箭头隐藏，运动暂停。曲线仍对应装配状态模型。标记数量、速度、材质、纹理不代表真实载流子数、物理时间、光学常数或实测表面形貌。

审核与数值证据

独立 AI 审核复核了公式、晶格几何、尺度和动画位置。曾发现球形标记与氧化层相交、横纵显示比例不一致；已改为沟道内零厚度标记和一致 x/z 缩放。后续审核检查分层顺序、归位输运恢复、轨迹尾部重置和资源释放，没有剩余实质阻塞。审核不等同于实验器件专家认证。

基准：Cox = 1.151044415664×10⁻⁴ F/m² = 11.51044 nF/cm²。VGS=5 V、VDS=1 V 时 Id=0.805731090965 µA；同一栅压下 VDS≥4 V 时 Id=1.8416710650624 µA。数值、几何、展示测试与浏览器验证分开记录。本轮中英文适配只更改展示层，科学模型保持 SHA256 152db807fc683b8fe9b6f8707b2cb6c87835060c7c20cdb318c43e0592418ad8。