Thermal oxidation and diffusion budget
Compare dry and wet oxidation growth, silicon consumption, and diffusion broadening before treating thermal steps as reusable evidence.
43 nm / 174 nm / sigma = 86 nm
Open simulation ->PROCESS REPLAY LAB / SA-PROC-002
Process fundamentals.
Compare the evidence cards.
This lab collects five precomputed process simulations for paper-first review: oxidation/diffusion, lithography, etch, CMP, and implant/anneal.
RUN ONE QUESTION
Pick the process claim first.
Every card keeps a condition, output, method, source, and limitation attached. The point is evidence discipline, not facility operation.
Photolithography exposure and focus window
Replay a dose/focus matrix to see why a printed CD is only useful when the process window and variation boundary are visible.
18 / 25 / 0 nm / 5 x 5
Open simulation ->Etch selectivity and profile
Compare isotropic, directional, and low-selectivity etch cases to keep mask budget, CD loss, and sidewall profile attached.
45 nm / 48 nm / 3:1 to 30:1
Open simulation ->CMP removal and planarization
Replay Preston-style removal, pattern-density warnings, radial nonuniformity, and endpoint margin before citing a CMP result.
120 nm/min / 80 nm/min / +/-2.56%
Open simulation ->Ion implantation and anneal activation
Replay projected range, straggle, activation, and diffusion broadening to keep doping evidence separate from a recipe.
Rp = 80 nm / 72% / 28 nm
Open simulation ->Lithography focus-exposure process window
Replay focus, exposure, CD-error proxy, overlay proxy, and pass/fail window to explain process windows without scanner recipes.
31 / 81 / -80 to 80 nm / -8% to 8%
Open simulation ->Plasma etch UQ profile window
Replay aspect ratio, bias class, and wafer location to compare anisotropy, selectivity, CD bias, and pass-probability proxy.
0.7728 / AR 1 / mid / center / 27 cases
Open simulation ->CMP pattern-density dishing and erosion replay
Replay pattern density, feature width, overpolish proxy, and pad age to connect layout density with dishing, erosion, nonuniformity, and planarization evidence.
63.168 / 55.524 nm / 54 cases
Open simulation ->Etch CD bias to DIBL and delay interactive lab
Filter chemistry, pressure, RF power, and overetch to inspect etch-rate, CD-bias, LER, DIBL, and inverter-delay teaching proxies.
108 cases / 16.29-25.76 ps / tradeoff scatter
Open simulation ->MODEL BOUNDARY
Process intuition is useful only when the boundary is loud.
The public replay is for teaching, figures, and early paper planning. It is not a process recipe, tool instruction, wafer qualification, or signoff result.
Browse saved datasets
Open thermal, lithography, etch, CMP, and implant/anneal data with their assumptions visible.
Download model cards
Use the public source notes and limitations when making slides, drafts, or teaching figures.
No recipe workflow
No process upload, tool account, facility credential, live execution, endpoint control, or qualified recipe claim is present.