Skip to content
SASemiAgoraSemiconductor knowledge & engineeringOpen lesson

SIMULATION / SA-DEV-2D-NDR-005

2D heterojunction NDR mechanism and evidence map
NDR evidence, not curve shape

Separate rectification, consecutive negative slope, peak/valley prominence, overlap-window shape, broadening, series/contact burden, sweep history, trap-memory contradiction, mechanism ambiguity, and owned replacement evidence without fitting or reproducing a published device.

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 alignment, broadening, background, contact/series burden, sweep direction, and trap-memory state fixed; move gate state and inspect how the synthetic overlap window shifts.
  • Read current, local slope, consecutive-negative count, peak/valley prominence, and NDR state together. One declining segment does not pass the stable-NDR rule.
  • Compare equal-magnitude positive and negative currents for rectification, then keep that result separate from the NDR classification.
  • Switch forward/reverse sweep and memory state to expose classification, peak-location, and curve-shape contradictions before discussing a transport hypothesis.
  • Use the decision gate to name the owned measurement, contact, uncertainty, and independently reproduced transport evidence required to replace the proxy.

BROWSER NDR EVIDENCE LAB

Require consecutive negative slopes. Compare the return sweep.

The SVG redraws one saved 16-point curve and its matched sweep. No physical transport solver or fitted device model runs in the browser.

Loading curveSweep classification alignedRectification stays separate
signed normalized I-V

Current shape and negative-slope evidence

selected sweepmatched sweepnegative segment
-1.120+1.12

Red segments satisfy the local slope threshold. The curve passes only when at least two are consecutive and the saved peak/valley prominence also clears the declared gate.

BiasCurrentLocal slopeConsecutiveRectificationNDR state

RUN CARD

Conditions before conclusions.

The browser filters a saved CSV and redraws a local SVG. It does not run DOS integration, DFT, NEGF, TCAD, compact models, SPICE, trap kinetics, measurement fitting, or optimization, and it cannot identify a mechanism or recommend a material, bias, device, circuit, or process.

precomputed-browser-csv-replay

Primary knobs

  • 16 signed normalized-bias samples
  • four gate-control states and three alignment-window hypotheses
  • narrow or broad window and low or elevated monotonic background
  • bounded or high series/contact burden
  • forward or reverse sweep with zero-memory reference or finite trap-memory proxy
outputs

Saved outputs

  • dimensionless current, local slope, consecutive-negative count, peak/valley prominence, and rectification state
  • matched-sweep peak and curve deltas, hysteresis contradiction, NDR state, and mechanism ambiguity
  • contradiction flags, bounded evidence-readiness score, decision gate, and next owned evidence check

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

2D NDR evidence CSV

/data/2d-heterojunction-ndr-mechanism-evidence-map-v104.csv

Open file ->
download

Typed web JSON

/data/2d-heterojunction-ndr-mechanism-evidence-map-web-v1.json

Open file ->
download

Model card

/data/2d-heterojunction-ndr-mechanism-evidence-map-model-card-v1.md

Open file ->
download

Metric contract

/data/2d-heterojunction-ndr-mechanism-evidence-map-metric-contract-v1.json

Open file ->
download

Source spine

/data/2d-heterojunction-ndr-mechanism-evidence-map-source-spine-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

Generate the complete 16 x 4 x 3 x 2 x 2 x 2 x 2 x 2 Cartesian sweep with no randomness, fitting, optimization, hidden state, source calibration, or external device value.

Keep this method attached when reusing the figure or metric.

method

Build an original dimensionless monotonic background plus a gate- and alignment-shifted overlap-window envelope, then apply a declared series/contact compression transform to both bias polarities.

Keep this method attached when reusing the figure or metric.

method

Require two consecutive positive-branch finite-difference slopes below the declared threshold and a minimum peak/valley prominence before labeling a stable NDR proxy; one falling segment never passes.

Keep this method attached when reusing the figure or metric.

method

Compare matched forward and reverse curves before classification, preserve rectification as a separate equal-magnitude branch ratio, and route every shape to an owned replacement-evidence gate.

Keep this method attached when reusing the figure or metric.

method

Keep nine topic-specific claim maps and five primary/open source anchors separate from all generated coefficients and rows.

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

Zero source numeric values, copied results, copied trace points, dimensions, material parameters, process conditions, contact values, calibrated models, live TCAD/NEGF/SPICE runs, measured rows, mechanism proofs, circuit advantages, rankings, recipes, optimization recommendations, signoff actions, or live server runs enter the dataset.

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

non-claim

No physical gate, bias, temperature, material, stack, contact, geometry, linewidth, current, ratio, density, mobility, time, frequency, energy, power, area, or application value is predicted or recommended.

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

non-claim

No neuromorphic, multivalued-logic, oscillator, low-power, product, vendor, material, architecture, or performance advantage is claimed.

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

non-claim

A real claim requires owned forward/reverse I-V sweeps across declared gate and temperature conditions, contact/series evidence, repeatability, uncertainty, model lineage, independently reproduced transport calculations, mechanism discrimination, and accountable review.

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

6144 rows across 384 curves and 192 matched forward/reverse pairs with no random sampling, fitting, optimization, or source calibration.

SA-DEV-2D-NDR-005

evidence note

9 topic-specific claim maps and 5 primary/open source anchors kept separate from every generated coefficient and row.

SA-DEV-2D-NDR-005

evidence note

Stable NDR requires 2 consecutive negative-slope segments and 8% prominence; rectification and sweep-history consistency remain separate gates.

SA-DEV-2D-NDR-005

evidence note

Independent Cartesian, equation, slope, peak/valley, rectification, sweep-pair, contradiction, decision-gate, zero-boundary, byte-identity, source-lineage, and SHA-256 verification is required in the canonical experiment folder.

SA-DEV-2D-NDR-005

Public execution boundary

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