White Light Interferometry: How It Works, and What Its Numbers Actually Mean
White light interferometry (WLI) is a non-contact optical method that maps surface height at nanometre resolution by scanning the sample through the coherence envelope of broadband light. Each camera pixel finds its own absolute focus position, so steps, pits and printed lines map without the phase ambiguity that limits monochromatic interferometry. TDM's SE320 publishes RMS repeatability ≤0.1 nm in phase-shift mode, with vertical resolution up to 0.01 nm.
What WLI actually measures
It measures optical path difference, converted into a height map. The outputs that follow from that are surface height, step height, 3D roughness statistics, and the shape terms derived from the map — bow, warp, curvature, coating beads and edge thinning. Everything else a vendor lists is a downstream calculation on this same map.
Two properties of the technique decide most buying and reporting mistakes:
- Vertical and lateral resolution are decoupled. Vertical detail comes from the coherence length and the phase measurement, not from pixel size. A 5× objective can still resolve sub-nanometre height. Contact profilometry cannot do this — there, tip radius limits both directions.
- It is an area technique, not a point technique. One scan returns a full field of pixels, each independently located in Z. That is what makes areal roughness statistics and mapping practical.
Why white light, and not a laser
A laser is nearly monochromatic: its interference fringes repeat every half wavelength, so a fringe alone tells you *where you are modulo λ/2*, not where you absolutely are. That is the 2π ambiguity — the central problem of monochromatic interferometry.
Broadband white light fixes it. Because different wavelengths only reinforce each other over a very short path-length difference, fringes appear only when the two interferometer arms are nearly matched — within roughly a micron of the zero-path-difference position. Sweep the sample through Z and every pixel shows a burst of fringes whose peak contrast (the coherence envelope) marks its absolute height. The thicker the film of coherence, the sharper the localization; the envelope position is an absolute measurement, which is why coherence scanning handles tall steps that phase-shifting cannot.
TDM's published SE320 implements both mechanisms on one bench: coherence scanning for absolute height across the full Z range, phase shifting for sub-nanometre detail on smooth surfaces.
Coherence scanning vs phase shifting
| Coherence scanning (CSI / WLI) | Phase shifting interferometry (PSI) | |
|---|---|---|
| What it reads | Position of maximum fringe contrast during a Z scan | Fractional phase of a monochromatic fringe at one position |
| Height result | Absolute | Ambiguous modulo λ/2 (about 275 nm at a 550 nm centre wavelength — a general optical quantity, not a TDM specification) |
| Vertical capability (TDM published, SE320) | Locates coarse heights across a 100 mm Z-axis travel | RMS repeatability ≤0.1 nm, vertical resolution up to 0.01 nm |
| Surface it needs | Works on rougher, steeper, lower-contrast surfaces | Needs a smooth, phase-continuous surface |
| Tall steps | Handles them — no unwrapping needed | Needs phase unwrapping, or it reports a wrong height |
| Environment | More tolerant of vibration | Sensitive: vibration and air turbulence corrupt the phase |
| Speed per field | Slower (the scan) | Faster (few positions) |
| Typical role on a real sample | First pass: find the absolute topography | Second pass: refine the flat, smooth areas |
Both answers come from the same instrument in practice; the software picks the envelope where contrast is broken and the phase where the surface is smooth.
The number most often misquoted: 0.01 nm is the vertical resolution WLI/PSI reaches on a surface height or a step. It is not a film-thickness range. An instrument described as measuring "0.01 nm films" is being described as a different technique — thickness on that scale is reflectometry or ellipsometry, which report a value inside a layer stack, not a height map.
Both use interference — they solve different unknowns
This is where buyers mix up two instrument classes, because the physics looks the same on a slide.
| White light interferometry | Spectroscopic reflectometry | |
|---|---|---|
| What is swept | Sample position along Z | Wavelength |
| What the fringes encode | Where the surface is | How thick the film is |
| Unknown solved | Height (topography) | Thickness (and, with constraints, optical constants) |
| Needs a step or edge? | Yes, for step height; no for roughness | No |
| Sees a buried interface? | No | Yes, through a transparent layer |
| Needs refractive index known? | No | The fit depends on it |
| TDM instrument | SE320 (WLI + PSI) | SE200 (4 nm–450 µm, repeatability ≤0.02 nm, 1–100 µm spot) |
If your question is "how thick", WLI is usually the wrong request unless there is a physical step to measure. If your question is "how tall, how rough, how flat", reflectometry has no answer to give.
Seven places WLI fails, or tells a convincing lie
- Steep slopes. Objective depth of focus and fringe period limit usable slope. Past that, contrast is averaged away inside a pixel and the map has holes. This is why light-section triangulation keeps working on machined walls where interferometry gives up.
- Thin transparent films on the measured surface. Reflections from the film's top and bottom interfaces add secondary envelope peaks (multiple-reflection artefacts) and shift the phase. The height you report is then the film's influence, not the substrate's step.
- Dark, absorbing or scattering surfaces. Too little returned intensity, or too little contrast, and the envelope is never found.
- Rough surfaces beyond the method. PSI loses phase continuity; coherence scanning degrades to gaps in the map.
- Stitching and flattening. Large-area maps are assembled from several fields plus a plane fit. Unqualified stitching leaves artificial steps that look exactly like real ones.
- Vibration and air currents. Worst for PSI, since it reads phase directly.
- Areal roughness numbers without filter settings. This is the one that quietly ruins comparisons. The objective's numerical aperture acts as a spatial filter: short-wavelength components are cut by the pixel and optics, long-wavelength components by whatever flattening you applied. Two Sa or Sz values measured through different cut-off filters are not the same number, however careful both labs were. Always record the filter settings alongside the value.
TDM publishes a mechanism-level capability (0.1 µm light-section on SE310; ≤0.1 nm RMS repeatability in PSI mode on SE320). The areal parameters our software reports, and the exact ISO 25178 filter settings applied to them, are confirmed per configuration in writing — we do not quote roughness statistics we have not defined the filters for.
Reading a WLI spec sheet: five lines that decide the rest
- What the vertical figure is. Resolution, repeatability and accuracy are three different claims on three different timescales. Ask which one is printed.
- Z travel and the scan range at each objective. A 100 mm Z axis (SE320) is what lets a high step and a shallow terrace live in one setup.
- XY stage and mapping. SE320 publishes a 200 × 200 mm fully motorized stage with closed-loop grating encoders; that sets whether your map is one field or a stitched mosaic.
- Objective set and NA. Five objectives with automatic switching (5× / 10× / 20× / 50× / 100×) plus a 5-megapixel camera (2592 × 1944) on SE320 define the lateral side of the trade-off — and the filter cut-offs behind every roughness number.
- Filter conventions for the reported roughness. If a datasheet is silent on this, treat the Sa/Sz figures as vendor-specific, not comparable.
Which TDM instrument does this
- WLI + PSI topography and sub-nanometre step measurement, large-area mapping → SE320 3D Optical Profiler with Scan Platform
- Steep walls, machined edges, dark or shiny surfaces, tall coatings — non-interferometric 3D → SE310 3D Optical Profiler (light-section, 0.1 µm, 1 scan/second, 300 × 150 × 70 mm)
- Thickness of transparent and multi-layer films as an imaging map → SE8000 Imaging 3D Thin-Film Thickness Measuring Instrument (imaging 3D thickness and multi-layer capability published; no numerical range published)
- Film thickness where there is no edge to step over → SE200 or SE100
Continue here: Optical Profilometer — What It Measures and When Not to Use One · Stylus Profilometer vs Optical · Spectroscopic Reflectometer: Principle and Range · How to Measure Thin Film Thickness · Wafer Thickness Measurement
Send us the surface, not the spec
One sample is enough. We run it on the WLI/PSI bench and the light-section bench, send back both 3D maps with the filter settings recorded, and tell you which mechanism resolved your feature — including the honest answer "neither; this is a thickness question".
Written by TDM Metrology Team · published 2026-11-13
Frequently Asked Questions
What is white light interferometry?
A non-contact optical technique that maps surface height by sweeping a sample through the coherence envelope of broadband light. Each camera pixel independently locates the position of maximum fringe contrast, which gives that pixel's absolute height — so WLI produces a full 3D topography map rather than one profile line.
What is the difference between WLI and phase shifting interferometry (PSI)?
WLI (coherence scanning) reads where the fringes appear and disappears, which gives an absolute height and handles tall steps. PSI reads the fractional phase of a near-monochromatic fringe, which is far more precise but repeats every half wavelength, so it needs a smooth, phase-continuous surface and cannot resolve a step larger than its ambiguity range on its own. A capable bench uses both: coherence scanning for absolute topography, PSI to refine smooth areas.
How accurate is white light interferometry?
In the vertical direction it resolves far below what any lateral pixel size suggests, because vertical detail comes from coherence length and phase, not from magnification. TDM's SE320 publishes RMS repeatability ≤0.1 nm in PSI mode with vertical resolution up to 0.01 nm — figures for surface height and step measurement. Real-world accuracy is governed by vibration, thermal drift, surface contrast and the filter settings applied to roughness statistics.
Can white light interferometry measure thin-film thickness?
Only where there is a physical step to measure — a scratched edge, a patterned terrace, an etched pit. A film lying flat under a transparent top layer has no edge for an interferometer to locate, and multiple reflections inside a thin transparent film distort height readings. Buried-film thickness belongs to spectral reflectometry (TDM SE200 publishes 4 nm–450 µm, repeatability ≤0.02 nm) or to fitting optical constants (SE100 publishes 5–1200 nm with n/k).
When should you not use white light interferometry?
Steep sidewalls, dark or strongly scattering surfaces, and rough topography that breaks phase continuity — light-section triangulation keeps working where WLI loses contrast. Avoid it for buried layers and for wet films measured at the coating head, and never compare areal roughness values between instruments without recording both sets of filter cut-offs.
Related guides and products
- SE320 3D Optical Profiler with Scan Platform
- Optical Profilometer Guide
- Stylus Profilometer vs Optical
- Spectroscopic Reflectometer Guide
- How to Measure Thin Film Thickness
- Wafer Thickness Measurement
