litopc, in pictures.

A chip is printed with light. Here is how, one picture at a time.

Step 1 · The mask

A stencil and a lamp

A chip is printed with light. A stencil on top, a lamp above it, and the wafer below.

litopc 3D view: a mask plate with an L-shaped opening floating above a wafer, with the light pattern glowing on the wafer — Real litopc output. The glow on the wafer is where the light lands.

Step 2 · The print

What you draw vs what prints

The shapes are so small that the light blurs on the way down. The print is not the stencil.

Simulated aerial image of the same mask under DUV 193 nm

Left: the L you drew. Middle: the light that actually arrives. Right: the shape that prints (solid line) against what you wanted (dashed).

Step 3 · The miss

Where the print misses

Corners round off. Line ends pull back. The middle gets thin.

How far the print lands from the target is called edge placement error, or EPE. Each black tick is one measurement of it, from the dashed target to the printed line. litopc takes 140 of them around this shape; one per edge is drawn here.

Step 4 · The fix

The fix

Where the print came out too small, make the stencil bigger. Where it came out too big, make it smaller. Then check again.

Mask as drawn

Mask after correction

This is optical proximity correction, or opc. The little ears and notches are the nudges. The target never moved; only the stencil did, and now the print lands on it. 5 rounds took the average miss from 25.3 nm to 3.9 nm.

Recap

All four steps, in half a minute

The same run, played through. Click a step to jump; switch to 3D at the end.

Actual litopc output · Mask as drawn

Mask

MaskPrintTargetEPE, one tick per edge

The target is the shape we want on the wafer; the mask starts as a copy of it.

This L-shape is the design intent. OPC will reshape the mask, never the target.

Recorded from a real litopc run · DUV 193 nm · NA 0.93 · L-corner · threshold 0.3 · 1100 nm field, 512 grid · 80 nm segments · gain 0.5 · 7 samples per edge

Step 4, in 3D

In 3D

The same correction in the 3D view: the stencil above, the light on the wafer below.

litopc 3D view of the L-corner, before correction: the mask plate above, the light on the wafer below — litopc 3D view of the recorded L-corner run, before and after five iterations. The corrected mask grows small ears and notches, and the light on the wafer comes out fuller along the whole L.

litopc 3D view of the L-corner, after correction: the mask plate above, the light on the wafer below — litopc 3D view of the recorded L-corner run, before and after five iterations. The corrected mask grows small ears and notches, and the light on the wafer comes out fuller along the whole L.

Going further

Lines close together

Neighbors matter. Lines drawn side by side print thinner, and their ends fall short.

Four lines: drawn (blue), wanted (dashed), printed (solid)

litopc 3D view of four parallel mask lines above their light pattern on the wafer — The same kind of pattern in 3D

Going further

The brightness knob

The wafer coating only reacts where the light is bright enough. Move the cutoff and watch the printed width change.

Intensity threshold τ = 0.30

60 nm linedoes not print
90 nm line94 nm
140 nm line145 nm

Mask, correctedTargetPrint at τProfile row

Three lines, 60, 90, 140 nm wide, under the same light (DUV 193_DRY, τ 0.3 in the app). The 90 and 140 nm masks carry 3 rounds of correction; the 60 nm line stays as drawn because it does not print at τ 0.3. Top view and profile share one horizontal scale. Move τ: the thin line prints only while τ is low, and every printed width follows its two crossings.

Your turn

Draw your own

The built-in patterns are a start. Draw any mask from rectangles, set the target, and run the same simulation and correction on it.

Drawn, 2 rectangles and a cut

Printed

Corrected, 5 rounds

This one is two bars and a cut, drawn for the EUV tool: the cut splits the bars into two line ends facing each other across a gap. Simulated and corrected with the same code the simulator runs: the average miss went from 5.4 nm to 1.0 nm in 5 rounds. Free accounts draw up to 5 rectangles and run 3 rounds per correction; Pro, up to 48 rectangles and 5 rounds, then more. Your numbers can differ a little from the recording as the engine evolves.

Open this example in the simulator

Glossary

The words used on this site, in the order you meet them.

Term	Plain meaning	Where you see it
The setup
Lithography	Printing a chip's pattern onto silicon with light.	Everything on this site.
Wafer, chip	The round silicon plate chips are made on; a chip is one small square of it.	The dark square in the 3D view.
Mask, reticle	The stencil that carries the pattern. A DUV mask is a glass plate with chrome on it; an EUV reticle is a mirror with an absorber pattern on top.	The pale plate in the 3D view: lit from above for DUV, from below at an angle for EUV.
Target	The shape you want to end up with.	The dashed outline.
Field	The square of wafer the simulation covers.	1100 nm for DUV, 212 nm for EUV; shown as FOV in the simulator.
Nanometer (nm)	A billionth of a meter. A hair is about 80,000 nm wide.	Every size here.
The light
Wavelength	The size of one ripple of the light; smaller draws finer detail.	193 nm (DUV) or 13.5 nm (EUV).
DUV	Deep ultraviolet, 193 nm from an ArF laser; the light most chip factories use. It passes through glass, so the tool is a column of lenses.	One of the two tools.
EUV	Extreme ultraviolet, 13.5 nm from a tin plasma. Nothing transmits it, so the tool is all mirrors and works in vacuum.	The other tool.
NA	Numerical aperture: how wide a cone of light the lens collects; bigger sees finer detail.	0.93 for DUV, 0.33 for EUV.
Partial coherence (σ)	How large the light source looks from the lens. A wider source softens contrast but prints dense lines better.	σ 0.7 in every preset; Pro can shape the source.
k1	Feature size × NA ÷ wavelength: how hard a pattern is for its tool. Near 0.35 the print needs correction to come out right.	The built-in patterns sit around k1 0.35.
Aerial image	The light pattern that lands on the wafer after the lens blurs it.	The glowing heat map.
Proximity effect	Neighbours change how a shape prints: light from nearby openings spills onto its edges.	Why dense and isolated lines print at different widths.
Exposure and focus
Resist	The light-sensitive coating on the wafer.	Simplified to one brightness cutoff.
Threshold (τ)	The brightness that counts as exposed; it stands in for exposure dose.	The slider from 0 to 1; 0.30 prints the default patterns to size.
Dose to size	The threshold at which the print comes out exactly as wide as the design.	The 'Size to target' link under the threshold.
Defocus	How far the wafer sits from best focus; the image blurs the same way in either direction.	The Defocus slider, in nanometers.
Depth of focus	How far focus can drift before the print leaves spec.	The Focus latitude study; the Rayleigh scale is ±112 nm for DUV, ±62 nm for EUV.
Bossung curve	Printed width plotted through focus: flat near best focus, falling away on both sides.	The Focus latitude chart.
Process window	The range of dose and focus over which the print stays in spec, both at once.	The Process window study (Pro).
Measuring the print
Printed contour	The outline that would really appear on the wafer.	The solid outline.
CD	Critical dimension: the width of the printed feature, measured across the centre of the field.	One number per simulation.
EPE	Edge placement error: how far the print sits from the target, in nanometers, averaged along the outline.	The number after correction.
NILS	How steeply the light falls off at the printed edge, scaled by the feature size; steeper survives dose noise better.	Shown next to CD after each simulation.
Corner rounding, line-end shortening	What the blur does to a sharp shape: corners print round and the ends of lines pull back.	Any L-shape or tip-to-tip pattern before correction.
Pitch	Centre-to-centre distance of repeated lines or holes.	The dense-lines template; the Through pitch study.
Iso-dense bias	The width difference between an isolated feature and the same feature inside a dense group.	The Through pitch reading.
MEEF	Mask error enhancement factor: how much a 1 nm change on the mask changes the print. Above 1, mask errors are amplified.	The MEEF study: the slope of wafer CD against mask CD.
Correcting the mask
OPC (optical proximity correction)	Reshaping the stencil so the blurred print comes out right.	The Correct button.
Mask bias	Moving every edge of the mask in or out by the same amount: the simplest correction.	The Mask bias study.
Segment	A piece of a mask edge that the correction moves on its own.	80 nm long on DUV, 20 nm on EUV by default.
Iteration	One round of simulate, measure, nudge.	3 per Correct on Free, 5 on Pro; 5 in the replay on the home page.
Add, Cut	The two drawing tools: Add places a rectangle, Cut removes one from the shape.	Edit ▸ Target in the simulator.
Study	Repeating a run while one setting changes, with a one-line reading of the result.	The Studies tab beside Set-up: 5 points on Free, up to 120 on Pro.

litopc is for learning and exploring. The pictures are real simulation results, but the model is simplified and not calibrated to any factory. The actual equations are in the model guide; longer reads are on the blog.

Open litopc simulator How the print is simulated