Sensor Glass Removal by Application: Why It Matters — and Where It Makes All the Difference

Hyperspectral cameras don't just capture images — they capture a full spectral signature at every pixel, generating what researchers call a "data cube." Each layer of that cube corresponds to a narrow wavelength band, and together they can identify materials, detect chemical composition, and reveal what the human eye — or a conventional camera — simply cannot see.

The challenge is that standard cover glass is designed for visible light. It introduces absorption and reflectance characteristics that corrupt the very data hyperspectral systems are built to collect. In VNIR (visible to near-infrared) and SWIR (short-wave infrared) systems, even marginal interference at critical wavelength bands can throw off material classification models and spectral analysis algorithms.

Removing the cover glass — and in many cases, replacing it with a custom optical substrate optimized for the target wavelength range — gives hyperspectral sensors clean, unobstructed access to the full spectral domain they were designed to capture. This is especially important in fields like agricultural monitoring, where the NIR reflectance signature of crops indicates water stress, chlorophyll content, and disease onset; or in mineralogy, where spectral fingerprints distinguish ore compositions in drill core samples.

Laser-based imaging systems are uniquely sensitive to the reflective properties of any optical element in the beam path — and cover glass is no exception. When a coherent light source interacts with the flat, parallel surfaces of a sensor cover glass, it can produce etalon effects: interference fringes created by light bouncing between the inner and outer surfaces of the glass. In beam profiling and laser diagnostic applications, these fringes are not a nuisance — they are a measurement error.

Additionally, some laser wavelengths are partially absorbed by standard borosilicate or soda-lime cover glass, reducing the signal reaching the sensor and distorting intensity measurements. For applications where the exact power distribution, beam shape, or temporal profile of a laser pulse needs to be characterized precisely, any optical element that modifies that distribution is a problem.

Removing the cover glass eliminates the glass-air interfaces that cause interference and allows the sensor to see the beam without mediation. When a replacement window is needed, UV-grade fused silica or other laser-grade substrates with tight flatness tolerances can be installed in its place.

Short-wave infrared imaging occupies the 0.9–1.7 µm range (and up to 2.5 µm in extended SWIR systems). In this band, objects that look identical under visible light often have dramatically different reflectance — making SWIR invaluable for sorting, inspection, and detection tasks where the naked eye or a standard camera falls short.

Standard cover glass used in many industrial cameras is not designed to be transparent across the SWIR spectrum. At wavelengths beyond approximately 1.0 µm, borosilicate and similar glass types begin absorbing light at an increasing rate, cutting off exactly the region that InGaAs and similar sensors are built to exploit. The result is a sensor that's spectrally capable but optically constrained by its own packaging.

Removing the original cover glass — or replacing it with a SWIR-transmissive window such as fused silica or sapphire — restores the sensor's designed sensitivity across its full operating range. Applications that benefit include electronic board inspection (where silicon is transparent at SWIR wavelengths, revealing subsurface features), solar cell inspection, produce quality grading, pharmaceutical tablet inspection, and nighttime surveillance using ambient SWIR illumination.

Most standard camera sensors have their cover glass specifically chosen to block ultraviolet light. This is by design — UV radiation can, over time, degrade silicon sensor performance, and for general-purpose imaging applications, this is a reasonable trade-off. But for applications that depend on UV sensitivity, this blocking is exactly the wrong characteristic.

Standard borosilicate glass begins to cut off UV transmission below approximately 350 nm, with significant absorption even in the near-UV range (350–400 nm). Deep-UV work — semiconductor photolithography inspection, plasma diagnostics, or certain fluorescence assays — requires working with wavelengths in the 200–350 nm range, where standard glass is essentially opaque.

Removing the original cover glass is the first step. The second is either leaving the sensor unprotected in a controlled cleanroom environment, or replacing the glass with UV-grade fused silica or quartz — materials that maintain high transmission across the near- and deep-UV spectrum. This combination enables applications such as detecting micro-defects in semiconductor wafer patterns, inspecting resin coating uniformity in electronics manufacturing, UV-based plastic sorting at recycling facilities, and fluorescence imaging in biomedical research.

Machine vision systems for inline manufacturing inspection often depend on precise spectral control — illuminating a scene with a narrow wavelength band and capturing only that reflected light, filtering out ambient interference and maximizing contrast for defect detection algorithms.

Achieving this with a bandpass filter mounted externally — in front of the lens — works, but introduces additional glass-air interfaces, potential for focus shift, and added optical path length that can compromise tight depth-of-field requirements. A far more elegant solution is to mount the filter directly on the sensor, in place of the cover glass.

This approach requires removing the original cover glass first. Once removed, Wilco's engineers can install and align a customer-specified bandpass, narrowband, or notch filter to 1-micron alignment accuracy — placing the optical element as close to the sensor surface as physically possible. The result is a more compact, more robust optical assembly with fewer air-gap interfaces, better stray-light rejection, and no effective increase in optical path length.

This technique is valuable across a range of machine vision applications: detecting surface defects on metallic components, inspecting adhesive or coating uniformity, identifying contamination in food processing lines, and verifying print or label quality at line speed.

In laboratory and clinical imaging environments, optical performance is never a secondary concern — it is the product. Whether a camera is being used in a fluorescence microscope, a confocal system, a flow cytometer, or a whole-slide scanner, every photon counts, and any unnecessary loss in the optical path represents degraded sensitivity, reduced signal-to-noise ratio, and potentially missed biological events.

Standard cover glass in industrial cameras is not AR-coated for the specific excitation and emission wavelengths used in fluorescence microscopy. Even modest reflection losses across multiple optical surfaces can significantly reduce the brightness of fluorescent signal — particularly for dim, low-abundance targets. Removing the original glass and either leaving the sensor bare (in clean instrument environments) or installing a custom substrate with appropriate AR coatings enables researchers to recover that lost signal.

For research cameras used in demanding spectroscopy or imaging flow cytometry applications, the ability to install a custom optical substrate directly on the sensor — with coatings tuned to the instrument's specific wavelength requirements — represents a meaningful performance advantage over any off-the-shelf configuration.