Redefining Standards in White Light Interferometry: Bringing Data Quality to Light

In white light interferometry (WLI), also known as coherence scanning interferometry (CSI), the objective is moved vertically through the surface's focal region. Within this range, interference signals are generated and captured for each camera pixel in the form of a correlogram. Traditionally, the surface height is determined directly from this signal.

With smartWLI, the correlogram is also used to assess the quality of each measurement point. Using FFT-based analysis, the signal is evaluated in the frequency domain to determine whether a measurement point contains a valid interference signal or is affected by local disturbances, artifacts, or outliers.

A key advantage of this approach is its point-by-point evaluation. Each measurement point is assigned a reliability indicator, making it possible to identify critical regions without relying on global assumptions or post-processing estimates.

A representative example is provided by measurements performed on a Chirp Standard C5. In this specimen, structural features progressively decrease in size while local surface slopes become steeper. It is precisely in these regions that the limitations of signal evaluation become apparent.

The visualization combines a 3D dataset with an overlaid quality map, along with a magnified view of a critical surface region. Areas exhibiting reduced data quality are primarily located along steep flanks, where pronounced local slopes make CSI/WLI signal interpretation more challenging.

Despite this, the results indicate that more than 99% of the measurement points maintain very high data quality, and can therefore be used with strong confidence in further analysis. Points with lower reliability can either be excluded or selectively processed through additional filtering and evaluation steps that also consider neighboring data points.

From a metrological perspective, this distinction is crucial. Isolated missing points typically have minimal impact on the overall result, whereas outliers can significantly distort it. The introduction of a quality layer enables precise identification of these critical points, ultimately supporting more reliable and robust measurement outcomes

Measurements on Black Silicon present an even greater challenge. This type of surface is characterized by extremely fine structures, steep slopes, and complex signal behavior, making it a demanding test case for many optical 3D measurement systems.

Without FFT-based quality evaluation, these features can easily produce noisy or misleading height data. This difficulty arises not only from the surface morphology itself, but also from the complexity of interpreting the local interference signals.

The smartWLI nanoscan, equipped with a blue light source and an Olympus interference objective with a numerical aperture of 0.8, is particularly well suited for capturing such fine and high-aspect-ratio structures. Through FFT-based analysis, relevant signal components can be isolated, allowing reliable measurement points to be distinguished from artifacts and outliers. AQC then leverages this quality information to ensure that only trustworthy data contributes to the final result.

This approach enables robust and reliable 3D measurements, even on surfaces that approach the fundamental limits of optical metrology.

The integration of modern NVIDIA GPUs fundamentally transforms the practical use of CSI/WLI systems. Their massive parallel processing capabilities allow for real-time handling of large data volumes, rapid FFT-based signal analysis, and immediate quality assessment without the need for extended post-acquisition processing.

As a consequence, the application scope of white light interferometry is significantly broadened. Complex surfaces—featuring steep slopes, fine microstructures, or challenging material properties—can be evaluated with greater reliability compared to conventional systems lacking GPU-accelerated quality analysis.

The notion of CSI/WLI as a purely height measurement technique is becoming outdated. Today's optical 3D metrology not only relies on high-performance optics and precise mechanical design, but also on advanced, fast, and quality-aware signal processing to ensure accurate and dependable results.