Precision Filters Hold the Key to Measuring How We See
Ion-beam sputtering, adaptive process control, and rigorous validation are a dynamic combination in the quest to transform human color perception.
By Peter Karp
Whether a smartphone screen, automotive dashboard, or surgical monitor, the first thing that one notices upon viewing a scene or object is the color. While this has been true throughout human history, today’s technology is consistently displaying higher saturated colors than ever before. Many displays cover the DCI-P3 gamut, defined 20 years ago. Others are now approaching the ITU-R Recommendation BT.2020 gamut, which spans a wider color range than the DCI-P3 space (see Sidebar).
Courtesy of Admesy.
Modern displays also now offer an increased dynamic range, compared with early system designs, spanning low luminance values ~0.001 cd/m2 for the darkest black to 1000 cd/m2 for the brightest white.
To accurately produce these displays, reliable measurements are crucial. Precision measurements translate the subjective impression of red or white into repeatable numbers that engineers can measure, compare, and, as necessary, correct. In the display and consumer electronics factories in which these modern devices are fabricated, obtaining measurements quickly is imperative for efficient economical production.
The challenge facing these device fabricators arises from the fact that color is a human perception — meaning that it is not a quality or metric that can be measured directly. In fact, a system that enables both measurement and comparison of color, in an objective manner, is the result only of intuitive physio-physical experiments1.
Nevertheless, the use of colorimetric measurement enables the fastest device
production times. By definition, a colorimeter outputs three values, which correlate to the sensitivity of the human eye — specifically, the three cone types in the retina.
Naturally, the quality of the colorimetric filters has a direct bearing on the accuracy of the color measurement. Both spot- and imaging-type colorimeters require the highest possible filter performance. Optimizing this parameter begins in the design stage and extends to process control and validation.
From human vision to tristimulus numbers
Three classes of cone cell types in the human eye — long-wave, medium-wave, and short-wave — mediate color vision. The signals that these cone cells generate are processed in our brains, which enables us to perceive color.
The ion-beam sputtering (IBS) method ensures a high degree of process control. This method can be used to create homogeneous layers that can be accurately modeled — crucial for stable and accurate process control and for shape-matching target filter curves.
Courtesy of Admesy.
That the visible spectrum is weighed in this way — by only three sensor types — offers an explanation as to why just three numbers are sufficient to describe a color. In 1931, the International Commission on Illumination published the color-matching functions (CMFs) to provide a standardized way to describe the color perception
of an average observer. Integrating any spectral power distribution with the CMFs yields the tristimulus values X, Y, and Z, which ultimately provides a concise numerical fingerprint of perceived color2. This integration can be performed mathematically from spectral measurements.
Alternatively, this integration can be performed directly using physical filters, modeling the standard CMFs. A colorimeter enables a user to realize this physical approach.
Measuring color in practice
The original International Commission on Illumination 1931 observer, which has become and remains the de facto industry standard, covers a 2° field, which is essentially — in the context of the human eye — the fovea. A supplementary 10° observer, published in 1964, accounts for peripheral cones and describes the color vision for a larger viewing field. More recent cone-based observers, for example, CIE 170-2:2015, also model age and physiological variation in the population (in some cases, these more recent standards may only differ on the 2° field versus the 10°). For practical display measurement, relevant industry standards such as IDMS, from the Society for Information Display and the International Committee for Display Metrology, as well as the Black Mura Standard, from the German Flat Panel Display Forum DFF, specify the use of the 1931 2° set for comparable measurement1.
In practice, there are two ways to effectively measure color: with spectroradiometers and colorimeters. Spectroradiometers sample the full spectrum in the visible and then calculate X, Y, and Z for any CMF observer. This flexibility, however, comes at the expense of speed, since low luminance values require a longer integration time.
Colorimeters, in contrast, directly measure the three X, Y, and Z values. Light passes three precision filters that are shaped to mimic the standard CMFs. The sensor currents are then proportional to X, Y, and Z, thereby enabling measurement rates of up to 60 kHz for flicker evaluation and fast measurement times for low luminance.
The accuracy of a colorimeter is largely defined with the spectral fidelity and uniformity of the three precision filters; a mere 0.2-nm shift in a Y-filter, for example, results in a spectral-mismatch index (f1′) of ~1% and a so-called “color point” offset of ~1(Δx,y ≈ 0.001)3. At 0.5 nm, the resulting chromaticity error reaches up to three “color points” (Δx,y ≈ 0.003), easily visible on neighboring displays (see Sidebar).
Yet even perfectly shaped filters are not enough. The response of the filter(s) must stay in tight tolerances across the wafer, and imaging colorimeters require lateral uniformity for f1′ of >0.4% to keep accuracy within two “color points” across the field.
These precision requirements cannot be enforced at the time of performance, or in application; the functionality of these components must be ensured during fabrication. Ion-beam sputtering (IBS) is a strong option to achieve such perfection.
Building filters atom by atom
In IBS, a broad, focused beam of energetic ions, typically comprised of argon, is directed onto a metal or silicon target. Ejected atoms combine with oxygen and condense on the rotating substrate as an amorphous layer. A single CMF filter requires 40 to 70 alternating layers of high- and low-refractive index oxides, typically with thicknesses between 6 and 200 nm.
IBS enables fabricators to achieve a high level of process control in part because it can be used to create homogeneous layers that can be accurately modeled. This is vital for stable, accurate process control and for shape-matching target filter curves, such as the International Commission on Illumination 1931 CMFs. Another advantage of IBS to produce precision filters is that the layers consist of dense, pore-free stacks. They exhibit virtually no spectral drift with humidity or temperature. This is important to ensure stable measurement results over time and under demanding practical conditions.
Broadband optical monitoring
Despite its many advantages, there are drawbacks to the use of IBS: The method is inherently slow, in terms of its deposition rate times. Coating a 70-layer stack, for example, may take a full day. During this time, even minute drifts in ion current or substrate temperature would accumulate into unacceptable spectral errors.
Broadband optical monitoring (BBOM) offers a solution. In the BBOM method, a light beam is measured twice. First, a reference measurement without the filter in place is obtained. The following
measurement of the light passing through the growing layer stack gives the transmission value. The transmission measurements enable the current layer thickness to be calculated.
An ion beam, typically composed of argon ions, is directed onto a silicon or metal target in the IBS process. Ejected atoms combine with oxygen and condense on the rotating substrate as an amorphous layer. Courtesy of Admesy.
It is essential that the spectroradiometer used to make this second measurement has excellent linearity. This ensures accurate readings with the decreasing transmission during the coating process.
Putting it all together
Combining IBS with adaptive monitoring and tight statistical post-screening yields filters that simultaneously maximize absolute accuracy and spatial uniformity. To achieve such adaptivity in the monitoring stage, it is essential that the spectroradiometer selected for the BBOM setup “compare” the spectrum with the design, layer by layer, so that the deposited layer thickness can be calculated. The systems in Admesy’s Neo series, for example, optimize the remaining layers after this sequence, to compensate for any variations. Here, the adaptive process allows the optimal match.
The combination of IBS and adaptive monitoring can offer value(s) of up to f1′ ≤1.5% for all three channels, which equates to approximately half the tolerance specified in IDMS for reference colorimeters, and provides uniformity of ≤0.1% layer thickness across a 40-mm wafer. This value is 5 to 10× better than magnetron-sputtered equivalents, bringing XY chromaticity differences in imaging systems down to <0.0024.
Importantly, these values are real-world metrics — not laboratory curiosities.
Equipment considerations
In addition to technical methodology, there are important equipment considerations to ensure the best possible filters. As it relates to spot colorimeters, sub-2% f1′ enables direct measurement of display white point without the urgent need for matrix correction, which can still be applied to match a reference spectroradiometer, although it is not a requirement. Measurement speeds of up to 60 kHz can describe the time-dependent behavior of a display at high resolution. This need is becoming especially relevant for gaming monitors and smartphones, since these types of systems use variable refresh rates.
A 40-mm-diameter wafer is expressed as nonuniformity index f1′. The f1′ value represents the spectral mismatch error of a colorimeter’s spectral responsivity. Courtesy of Admesy.
In an imaging colorimeter, it is vital to ensure a highly accurate measurement for each camera pixel. An automatically revolving XYZ filter wheel realizes the measurement of the three tristimulus color values and assures the capture of the most accurate luminance and chromaticity maps. Due to the effective filter uniformity, the edge pixels’ color values are as reliable as in the center.
Further, the same optical-monitoring know-how feeds back into spectroradiometer calibration. A patent-pending calibration lamp from Admesy, for example, uses three multichannel bandpass sensors to monitor and account for the spectral tilt of its halogen source, in real time, shrinking radiance calibration uncertainty from 2.5% to <1%.
The entire value chain
Turning the complexity of human vision into three electrical signals hinges on one deceptively simple component: the tristimulus filter. Achieving perfection in this component demands a marriage of advanced thin-film physics, real-time
spectral metrology, and meticulous statistical process control. IBS provides the materials science foundation; adaptive BBOM closes the loop; and high-linearity spectroradiometers validate every atomic layer.
The result — color filters with sub-2% f1′ values and wafer-wide uniformity of <0.4% for f1′ — empower a generation of colorimeters and imaging systems that, finally, “measure how we see.” These instruments enable us to achieve these critical measurements with fidelity that is indistinguishable from the human eye itself.
Meet the author
Peter Karp is senior sales and support manager at Admesy BV. He is passionate about all technologies related to human vision. With experience in cameras, photography, and lighting technologies, he specializes in electronic displays, printing, and color measurement; email: sales@admesy.com.
Acknowledgments
Admesy wishes to acknowledge and cite the contributions and collective expertise of its
engineering team for the information contained in this article.
References
1. P. Karp and R. Bouten (2017). Measure
how we see — the way to perfect display colors, Admesy white paper.
2. HunterLab 2022. CIE Standard Observers and calculation of CIE X, Y, Z color values — AN-1002b. HunterLab application note. https://support.hunterlab.com/hc/en-us/articles/203420099-CIE-Standard-Observers-and-calculation-of-cie-x-y-z-color-values-an-1002b.
3. W. Weltjens (April 2024). The Importance of Filter Accuracy (and Uniformity) in Colorimetry. Presented at the Electronics Displays Conference 2024 in Nuremberg, Germany.
4. R. Bouten (2023). Improving calibration accuracy for high-end spectroradiometers, Admesy white paper.
Color Spaces
As a specific, defined system for grouping colors, color spaces are widely used
both in industry and, more broadly, in society. The most famous example is the Pantone collection, which is used for graphic design and printing.
As it relates to color measurement, color spaces function as a system to show the
colors and/or luminance that a system, such as a display, can produce. This method represents all colors so that users can objectively compare device outputs and ensure the reproducibility of certain colors across multiple devices. The CIE color spaces, for example, represent a map-like approach, where a color is expressed as coordinates or points, which are referred to as “color points.” Precision can then be expressed as the amount of coordinate difference between two points — just as on a map.
/Buyers-Guide/Admesy-BV/c17329