A 1 µm Spot Is a Different Measurement, Not a Smaller Version of the Same One

Most thin-film metrology is selected by thickness range and accuracy. On patterned wafers, that ordering gets the problem backwards. Before accuracy matters, there is a geometric question: what is inside the spot? The SE200 publishes a spot size of 1 µm (100X) to 100 µm (5X) with three aperture options, ≤0.02 nm repeatability, and a thickness range of 4 nm–450 µm. Read those as one argument: the aperture choice, not the repeatability figure, decides what the number means.

A 100 µm spot and a 1 µm spot do not measure the same film

Illuminated area scales with the square of the diameter. A 100 µm spot covers 10,000× the area of a 1 µm spot.

On a blanket film, that is only throughput. On a patterned die it is a different measurement entirely. A 100 µm spot on a metallised test structure integrates metal, dielectric, via fill and etched edge into one reflected spectrum — and the instrument still returns a confident thickness, just a thickness of an average that exists nowhere on the wafer.

Symptom that you have this problem: a spectroscopic reflectometer reading that disagrees consistently with cross-section or ellipsometry on patterned areas, but agrees perfectly on blanket test structures. The tool is fine. The spot was too big.

Three apertures, because smallest is not automatically best

There are costs to shrinking the spot, and they land in a predictable order:

Positioning tolerance. At 1 µm, aiming is a real operation. The SE200 publishes CCD visual positioning for the micro-area spot — that is what makes sub-5 µm targets repeatable rather than lucky.
Focus and standoff sensitivity. High magnification narrows the depth over which the measurement is valid, so wafer warp and chuck flatness become error sources.
Signal and fitting stability on thin layers. Very thin films need measurable spectral structure; over a small spot you have less energy, and the model can be underdetermined before the photon budget is.
Time per point on large mapping jobs, where the grid is set by the number of sites, not the spot.

The practical reading: choose the largest spot that still isolates the feature, then let repeatability do its job. Three aperture options is a workflow, not a spec-sheet decoration.

When ≤0.02 nm is no longer the binding constraint

Sub-0.02 nm repeatability is easy to quote and easy to misuse. Repeatability holds everything else constant. On a real wafer the limiting term is usually elsewhere:

Model error. Fitting thickness and n/k from one reflectance spectrum requires a dispersion model. For a 4 nm film on silicon, a wrong k or a wrong interface layer moves the reported thickness more than photon noise does.
Substrate bias. Reflectance spectrum from a doped or oxidised wafer is part of your signal. Reference it on the same substrate family.
Drift between shifts. This is why the published better-than-0.05 nm long-term stability spec matters more for production use than the repeatability figure: it is what lets Tuesday's numbers be compared to Friday's, and what makes two tools correlated at all.

For SPC on a fab floor, order your concerns: stability (shift-to-shift) → reproducibility (tool-to-tool) → repeatability (same spot, again). Most thin-film spec sheets argue about the third.

Spectral range is a thin-film problem, not a thick-film one

Within the published 380–1000 nm, the near-infrared gives you fringes for optically thick layers; ultraviolet features give you contrast on films only a few nanometres across, where absorption structure is strongest. So the expandable 200–1700 nm option is a question about which end of your distribution you are trying to control — the sub-10 nm gate-dielectric end or the multi-micron passivation end — not a bigger-number upgrade.

Confirm the configuration against your own film stack; the range and the model have to be chosen together.

Where a C-mount retrofit changes the buying decision

The SE200 publishes a standard C-mount interface for existing microscopes, USB 2.0 to a PC, and non-contact measurement. On a line that already has die-level optical inspection, that turns a metrology purchase into a metrology channel: the site selection workflow, the operator and the optics are already there, and you are adding a measurement, not a station.

The useful pattern is not "lab tool versus line tool" but the same film measured at both scales — a full-field offline station for range and rate, a micro-area channel for the patterned die.

What this tool is not

It is not an inline production platform: see SE600X inline metrology.
It is not a wafer-substrate thickness gauge. It measures film on a wafer — see wafer thickness measurement.
Naming: SE200 (this offline micro-area instrument) is not SE200x (inline PL mapping and AOI) and not SE20 (offline defect mapping). Three different tools, close names.

Where a 1 µm Spot Sits in the Published Market

Spot size is a declared spec across this instrument class, so it can be compared directly. Published figures from vendor and facility sources, as retrieved:

Published smallest spot size by source
Source	Published smallest spot
KLA thin-film reflectometers	spots as small as 1 µm
rtec Instruments TFT-1000	4 µm – 2 mm, by objective
Filmetrics F40 (Oregon State University technique note)	as small as 10 µm, for patterned surfaces
UCSB Nanofab facility wiki	down to approximately 10 µm
TDM SE200	1 µm (100X) – 100 µm (5X), three apertures

Two readings. First, a 1 µm floor is not exotic — it is where the premium end of this class already sits, so a claim of sub-micron isolation needs to be checked for positioning and focus margin, not just for the diameter. Second, the 10 µm band is far more common, which is why "largest spot that still isolates the feature" beats "smallest spot available."

Facility pages (UCSB) and independent technique notes describe the same failure mode this page does: on patterned areas the spot averages whatever it contains. That is a property of the measurement, not of any one instrument.

Frequently Asked Questions

How can I measure the thickness of a thin film?

Optically, by fitting a measured spectrum to a layer model: a reflectometer records intensity versus wavelength and solves for thickness (4 nm–450 µm on the published SE200 range, 5–1200 nm on SE100); an ellipsometer records polarisation change and additionally resolves n and k. On patterned die the first question is not which technique but what is inside the spot — see the section above on illuminated area scaling.

How do you measure film thickness?

Three decisions in order: measurement stage (offline bench, inline station, in-situ at the coating head), spot size, then the optical model. Spot size comes before accuracy because a 100 µm spot on a metallised test structure returns a confident thickness of an average that exists nowhere on the wafer. Published spot options here are selectable 1–100 µm (three apertures, 1 µm at 100X and 100 µm at 5X on the SE200).

What is the thickness of a thin film?

The published TDM instrument ranges span 4 nm–450 µm (SE200) and 5–1200 nm (SE100, SE600X), so "thin film" in this catalogue covers single-molecule-coating territory up to multi-micron passivation. The useful dividing line is not thickness alone but whether the film produces measurable spectral structure across the chosen band — sub-10 nm layers need UV contrast, multi-micron layers need near-infrared fringes. The full model matrix is on the reflectometer page.

What is the 80/20 rule for DFT measurement?

A dry-film acceptance convention from coating inspection — most readings must meet the specified DFT, the rest may fall short only within a stated margin, and the threshold comes from the standard or contract in force, not from a vendor article. For optical metrology its relevance is statistical: an acceptance rule expressed over many points requires a tool that repeats point to point and shift to shift, which is why the published better-than-0.05 nm long-term stability figure matters more to that use case than the 0.02 nm repeatability one.

Related guides and products

Request a measurement configuration