Skip to content
SASemiAgoraSemiconductor knowledge & engineeringOpen lesson

SIMULATION / SA-PROC-LER-LWR-VARIABILITY-059

LER/LWR to SRAM variability chain
LER/LWR Variability

Two edges can move together, apart, or independently. Inspect how that relationship changes linewidth and local-CD variation before any bounded device or SRAM teaching proxy.

sectionExploresectionRun CardsectionExportsectionMethodssectionLimitssectionEvidence

EXPLORE

Start with the observable.

Use this page when the simulation is the primary artifact. The linked lesson explains the concept; the linked lab keeps the visual replay surface available.

guided exploration
  • Hold edge roughness constant and move left-right correlation from +1 through 0 to -1.
  • Compare the locked conditional LWR/LER relation with the exact width-variance identity.
  • Change correlation length and local-CD aperture, then inspect the bounded device, delay, SNM, and write sensitivity ratios.

BROWSER-LOCAL ROUGHNESS LAB / V127

Two edges can move together, apart, or independently.

Seeded synthetic traces make correlation visible before a bounded process-to-SRAM sensitivity bridge.

Current tab onlyillustrative-original-surrogateNo fab data, storage, upload, solver service, or network writeLWR is not universally 2 x LER. Extraction bandwidth and Measured replacement slots remain explicit.
Pooled LER / 3 sigma3.600 nm
LWR / 3 sigma4.105 nm
Local-CD / 3 sigma3.697 nm
Realized correlation0.35
EDGE POSITION

Same LER, different width response

Left and right traces are detrended separately. The gap between them is linewidth.

Left edgeRight edge256 samples / 1.004 nm pitch
Raw widthLocal CD9-point circular aperture

RUN CARD

Conditions before conclusions.

The browser generates a seeded synthetic AR(1) teaching trace in the current tab. It does not load fab data, fit a process, recommend a recipe, predict yield, or provide signoff.

current-tab deterministic synthesis

Primary knobs

  • Seed and nominal CD
  • edge sigma and left-right correlation
  • correlation length and measurement length
  • local-CD aperture
outputs

Saved outputs

  • paired edge and local-CD traces
  • LER, LWR, and extraction bandwidth
  • 16-bin local-CD histogram
  • bounded normalized bridge summaries
  • deterministic JSON and CSV

EXPORT

Download the public artifacts.

These exports are the supported public files for review. PDK files and restricted third-party material are not redistributed.

download

V127 public model contract

/data/semiagora-ler-lwr-variability-v1.json

Open file ->

METHODS

How the number was made.

Methods are written for citation discipline: the extraction rule matters as much as the plotted value.

method

Each edge is linearly detrended and normalized with the descriptive sample standard deviation using an N-1 divisor.

Keep this method attached when reusing the figure or metric.

method

A Gram-Schmidt companion constructs exact requested cross-correlation before edge sigma is applied.

Keep this method attached when reusing the figure or metric.

method

Local CD is a circular odd-point boxcar average; the wrap is a synthetic boundary, not a physical wafer edge.

Keep this method attached when reusing the figure or metric.

LIMITS

What this page does not claim.

The MVP is useful because the limits are visible. These statements prevent a teaching simulation from being cited as silicon evidence.

non-claim

All traces and distributions are deterministic synthetic teaching outputs, not fab, wafer, resist, mask, etch, device, circuit, or production data.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The AR(1) process is a convenient stationary correlation model, not a universal physical PSD or stochastic lithography model.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The reported descriptive 3-sigma values depend on the locked detrend, N-1 divisor, measurement length, sampling pitch, aperture, circular boundary, and absence of measurement-noise correction.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The requested correlation length is a synthesis input, not a fitted or uncertainty-qualified metrology result.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The equal-variance LER/LWR relation is conditional; unequal edge variance, different conventions, filtering, and real metrology can change the relation.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The DIBL, Vt, drive-current, inverter-delay, SNM, and write-margin values are original normalized proxies with arbitrary visible coefficients and no calibration.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

The product cannot establish physical causality, statistical representativeness, measurement uncertainty, model validity, process capability, recipe transfer, yield, signoff, or production readiness.

Carry this boundary into any derivative note, paper draft, or slide.

non-claim

Researcher-facing businesses remain deferred to a separately verified V200 contract.

Carry this boundary into any derivative note, paper draft, or slide.

EVIDENCE

Trace the claim back to artifacts.

The evidence route remains the catalog-level browser; this page is the simulation-level reading card.

evidence note

The LER/LWR relation is conditional on equal edge variance and this exact extraction convention.

SA-PROC-LER-LWR-VARIABILITY-059

evidence note

Every bridge coefficient and clamp is visible; none is calibrated to a technology or SRAM architecture.

SA-PROC-LER-LWR-VARIABILITY-059

evidence note

Four measured-replacement slots state what is required before a physical or production claim.

SA-PROC-LER-LWR-VARIABILITY-059

Public execution boundary

No server-side live ngspice execution, uploads, accounts, payments, comments, newsletter signup, or job/event submissions are active on this page.