Measurement of High-Performance Thin-Film Optical Filters

The HELIX Spectral Analysis System has redefined measurement abilities of high performance thin-film optical filters. HELIX was developed by Alluxa Engineering staff to solve the limitations of most commercially available spectrophotometers. The system’s abilities are four-fold: it is able to measure filter edges to OD7 (-70 dB), assess blocking to OD9 (-90 dB), resolve passbands that are as narrow as 0.1 nm at full width half maximum (FWHM), and resolve edges as steep as 0.4% in relation to edge wavelength from 90% transmission to OD7.

Figure 1. Transmission spectra of a high-cavity-count ultra-narrow bandpass filter measured with the HELIX System and compared to theory. The HELIX System is able to measure blocking up to a level of OD9 (-90 dB).

Overview

An increasing demand for high-performance optical filters with edge transitions steeper than 0.5% of edge wavelength from 90% transmission to OD6 (-60 dB), ultra-narrow filters with sub-nanometer bandwidths, and wide-range blocking beyond OD6 has driven the latest advancements in thin-film design and coating technologies. However, measuring the actual spectral performance of these filters is very challenging because of the intrinsic limitations of standard spectrophotometers and other grating-based measurement systems. These limitations cause the instrument to be incapable of resolving ultra-steep edges or very narrow passbands, resulting in measured transmission spectra that show inaccurate bandwidths and rounded peaks of ultra-narrow filters, noise floors that range between OD2 (-20 dB) and OD7, and smeared and rounded edges. This can be vexing for technologists who need measured data that shows the real performance of optical filters before the qualification phase of system development. Also for coating manufacturers, who have to make filters that meet the challenging requirements even though performance is almost impossible to measure in a majority of cases.

Therefore, a number of thin-film manufacturers have begun utilizing tunable lasers, custom-made spectrophotometers, and other advanced techniques to attain more accurate measurements. Unfortunately, a lot of these solutions can be sensitive to misalignment, cost prohibitive, or have a limited wavelength range, and so are not suitable for use in a high-volume factory. Fortunately, Alluxa has built the HELIX Spectral Analysis System, a custom solution that can measure blocking to OD9 (Figure 1), trace edges to OD7, and accurately resolve both sub-nanometer passbands and edge transitions that are as steep as 0.4% of edge wavelength from 90% transmission to OD7.

Traditional Measurement Techniques and Limitations

Spectrophotometers and Grating-Based Measurement Systems

In the thin-film industry, spectral measurements of optical filters are most frequently acquired through the usage of a double-monochromator-based spectrophotometer. These and other grating-based measurement systems determine optical filter transmission by means of a broadband light source used along with monochromators that are made up of a diffraction grating and a series of mirrors and slits. When broadband light hits the first diffraction grating, it is dispersed into its component wavelengths, with each reflecting off of the grating at a marginally different angle. By rotating the grating, the monochromator is able to sequentially project each wavelength, or band of wavelengths, through a slit. The beam then moves to a second monochromator or grating to minimize the stray light in the system before it finally exits through another adjustable slit that further narrows the spectral bandwidth (SBW) (Figure 2). By modifying the slit-width, both spectral resolution and light intensity can be manipulated.

Figure 2. Simplified diagram of a dual-monochomator-based spectrophotometer.

In thin-film metrology, high spectral resolution is vital because a number of optical filters are designed with steep transitions from high transmission to deep blocking across a short wavelength range. If the spectral resolution is coarse, the measurement will produce a smeared and rounded edge that is not illustrative of the true steepness of the filter. This inability to resolve transitions across a short wavelength range is also true for filters with ultra-narrow bandwidths. For these, the effect of edge smearing is additionally compounded as the bandwidth is close in value to the spectral resolution of the instrument. This results in an unresolved passband, with the final measurement displaying a rounded peak and comparatively low transmission.

The spectral resolution of a spectrophotometer can be enhanced by narrowing the exit slit, thus decreasing the SBW of the light that passes through. However, even if the SBW is reduced, some edge smearing will still be seen because of the relatively large cone angle (small f-number) and spot size of the incident beam. Spot size is a crucial consideration when measuring thin-film filters as the spatial uniformity gradient of the coating results in a moderate shift of the filter edges and bands according to the location of the incident beam. If the spot size is large, the measurement will reveal edge smearing that results from an averaging effect across all filter locations that fall into the beam.

Similarly, the f-number of the incident beam is an important consideration in thin-film metrology because the spectral response of optical filters is based on angle. At increasing angles of incidence (AOI), optical filter spectra shift down in wavelength at a rate that is determined by the effective refractive index of the filter. Furthermore, polarization splitting can result in a distorted spectral shape when AOI is large. Since light from a non-collimated incident beam hits the filter at all angles in the range of the cone, an averaging effect ensues that results in smeared edges and rounded passbands.

Thus, besides narrowing the exit slit, edge smearing can be reduced by adding a small aperture to decrease the beam size and cone angle of the incident light. However, as light intensity is diminished by both reducing the slit-width and adding apertures, the signal-to-noise ratio is also reduced, which finally results in a noise floor that can be elevated by several orders of magnitude. Therefore, these methods are beneficial for resolving marginally steep optical filter edges to a level of approximately OD2 or OD3, but cannot be used to assess attenuation deeper than that level (Figures 3a and 3b).

Conversely, broadening the slit and totally removing any apertures will boost the light intensity and lower the noise floor. Attenuating the reference beam, a method referred to as Rear Beam Attenuation (RBA), will help to additionally lower the noise floor by biasing the dynamic range of the detector, thus increasing the signal-to-noise ratio. These methods are beneficial for assessing blocking to a level of OD6 or OD7, which is the maximum noise floor of several spectrophotometers because of the scattered light usually present in the system. However, they also result in a trade-off because the improved light intensity is accomplished by increasing the SBW, which brings about a coarser spectral resolution. Even for standard-performance optical filters, a measurement using these methods will display inaccurate bandwidths and distorted edges because the edge transitions from high transmission to deep blocking rival the spectral resolution of the measurement (Figures 3a and 3b).

Figure 3a. and 3b. Comparison of two different techniques used to measure the same ultra-narrow bandpass filter: (1) narrow slit with a small aperture introduced and (2) wide slit with the aperture removed and a rear beam attenuator introduced.

For a number of standard filters, a measurement with minimal edge distortion and a noise floor of OD4 or OD5 can be realized with a moderate SBW setting and by adding a relatively broad aperture. However, all methods mentioned are inadequate for measuring and assessing steep edges, blocking beyond OD6, or ultra-narrow bandpass filters.

Advanced and Custom Measurement Options

Many off-the-shelf spectrophotometers come with integrated light sources and diffraction gratings that match the UV, visible, and /or IR measurement range of interest. The instrument automatically switches gratings and light sources at the suitable time, creating a seamless dataset. However, these components cannot be easily switched out for higher-powered light sources that would increase signal-to-noise ratios, or diffraction gratings with greater line densities that would optimize spectral resolution. Thus, custom-made grating-based measurement systems are required to enhance measurement capability. This can be realized by building a custom spectrophotometer in an optics lab. This option allows for the incorporation of numerous customizable monochromators into the system, which have the potential to lower the noise floor to OD8 (-80 dB) for a double monochromator, or OD12 (-120 dB) for a triple monochromator. However, these systems can be vulnerable to background light, misalignment, and other environmental conditions unless they are totally contained in a custom-built enclosure. Another option would be to customize a currently available off-the-shelf system, although most of these are designed without the intent to modify, so as to avoid problems pertaining to warranty and repair.

Many highly accurate and precise measurements of optical filters are performed by using tunable lasers rather than grating-based systems. As a result of the collimated beam, the increased power output, and well-defined wavelength, tunable lasers are able to create accurate, high OD measurements with very fine spectral resolution at each wavelength within their range. However, since the wavelength range of most tunable lasers is comparatively narrow, multiple lasers would need to be used to measure the complete transmission and blocking ranges of a single filter. Tunable lasers are also expensive to buy and maintain, hence they are not perfect for use in high-volume manufacturing scenarios.

HELIX Spectral Analysis System

System Overview

The HELIX Spectral Analysis System is a custom solution that reduces the trade-off between spectral resolution and light intensity, while attaining a large f-number and small spot size. It was built by combining a high-powered laser source into a double-monochromator system. Through the high intensity collimated light of this custom configuration, high spectral resolution is able to be realized without raising the noise floor. The outcome is a system with a wavelength accuracy of up to ± 0.05 nm and a spectral resolution of up to 0.01 nm that is capable of accurately and precisely measuring the highest performance optical filters.

Figure 4. Measurement comparison of a Raman LIDAR filter designed with a slope of < 0.1% of edge wavelength from 90% transmission to OD4 (-40 dB). Two identical filters were designed to be used in series to provide OD8 (-80 dB) blocking of the 532 nm Rayleigh signal. The HELIX System resolved edges to OD 7 (-70 dB).

Resolving Steep Edges

Because the HELIX System is capable of obtaining high spectral resolution measurements without raising the noise floor, it can accurately resolve steep edges all the way to OD7 as illustrated in Figure 4. This is particularly useful for assessing filters used in applications such as Raman LIDAR and Raman spectroscopy. In these applications, optical filters are built to provide more than 90% transmission of the Raman signal while attenuating the laser line or related Rayleigh signal to a level of OD6 or greater. Since the separation between target signal and laser line can be 1 nm or less in certain cases [1], many optical filters are designed to be steeper than 0.5% of edge wavelength from 90% transmission to OD6.

Since optical filters with this level of edge steepness are hard to measure using typical techniques, many thin-film manufacturers assess edges by using the corresponding theory trace to extrapolate edge locations outside the noise floor of the spectrophotometer. However, this technique only offers a rough estimate of the blocking level at the specified wavelength and does not consider influencers that could impact absolute achievable blocking, which could differ from run to run or from filter to filter. For optical filters with very tight edge transitions, establishing the exact attenuation level at a specified wavelength along the edge is vital so as to ensure that the filter will perform as intended. Therefore, the HELIX System is instrumental in establishing the true performance of these filters.

Measuring Wide-Range Blocking

Many applications, such as fluorescence imaging, flow cytometry, and Mie, Rayleigh, and Raman LIDAR, need filters with broad-range blocking at a level beyond OD7. As the noise floor of most spectrophotometers cannot be extended much beyond OD6, many thin-film manufacturers need to stipulate blocking levels that are OD7 or greater as “by design”. This means that the filter design will surpass the specified blocking level, however the measurement may not. Unfortunately, even if the filter is designed with blocking well beyond the specified level, errors during the coating process, pinholes and surface imperfections, and/or other defects that create light scattering in the filter assembly can decrease the attenuation level of the finished product. Since the HELIX System has the ability to make wide-range high OD measurements as shown in Figure 5, it is capable of identifying filters that do not meet the required attenuation level, ensuring ideal system performance.

Figure 5. Measurement comparison of a high-performance fluorescence filter designed with steep edges and greater than OD8 (-80 dB) blocking. Data from the HELIX System shows filter edges resolved to OD7 (-70 dB) and wide-range blocking at a level of OD8.

Measuring Ultra-Narrow Bandpass Filters

As stated, ultra-narrow bandpass filters are hard to measure because the sub-nanometer bandwidth rivals the spectral resolution of most spectrophotometers, which causes a measurement with a rounded peak, smeared edges, low transmission, and an inaccurate bandwidth.

Figure 6. Measurement comparison of a 1083 nm high-cavity-count ultra-narrow bandpass filter. The HELIX System completely resolved the square passband and measured the steep edges to OD7 (-70 dB).

This effect can be observed in ultra-narrow filters with standard edge steepness and bandwidths that are less than 1 nm at FWHM. However, the most intense distortions are witnessed in high-performance, ultra-narrow filters that are built for use in solar imaging, LIDAR, and laser applications. These filters can be designed with bandwidths that are as narrow as 0.1 nm at FWHM, and a number of them are designed with a high-cavity count, providing the filter with a flat top, a square spectral shape and steep edges [2]. However, since both the bandwidth and the edge steepness of these filters require a lot higher spectral resolution than what is available through most spectrophotometers, the square shape is rarely represented in standard measurements (Figure 6), especially when the FWHM is less than 0.5 nm.

Figure 7. Measurement comparison of a 0.2 nm FWHM ultra-narrow bandpass filter. Data from the HELIX System shows a completely resolved passband.

Since the peak transmission of ultra-narrow bandpass filters is hard to measure using standard techniques, it becomes necessary for manufacturers to specify high transmission requirements as “by design” for many filters with sub-nanometer bandwidths. However, when high-performance, ultra-narrow bandpass filters are measured with the HELIX System, even the narrowest and steepest passbands are able to be fully resolved (Figure 7) and the edges traced all the way to OD7. This allows for the assessment of the true peak transmission, edge steepness, spectral shape, and FWHM of the filter.

Measuring Notch Filters

Notch filters pose a unique metrological challenge to thin-film manufacturers because they are designed to attenuate a relatively small wavelength range while transmitting the adjacent spectrum on either side. Since so much light is transmitted through the filter during the measurement process, there is an increased quantity of stray light that ends up reaching the detector. When that occurs, the signal-to-noise ratio declines, and the noise floor of the measurement is raised. Although stray light is normally present in the spectrophotometer, it presents less of a problem for filters that attenuate in the working range of the detector, like bandpass filters. Even though stray light cannot be eliminated, the HELIX System is still capable of realizing measurements that extend beyond OD7 for narrow notch filters as illustrated in Figure 8, and beyond OD8 for wider notches as illustrated in Figure 9.

Figure 8. Measurement comparison of a narrow notch filter. Data from the HELIX System shows blocking measured to OD7 (-70 dB).

Figure 9. Measurement comparison of a wide notch filter. Data from the HELIX System shows blocking measured to OD8 (-80 dB).

Summary

Alluxa has created a custom spectral analysis system that can accurately measure the highest-performance optical filters, and at the same time is sufficiently robust for use in a high-volume manufacturing setting. The HELIX System is a vital tool for assessing true optical filter performance, ensuring that Alluxa's filters will meet and exceed the demanding requirements of even the most advanced systems.

Literature Cited

[1] Alannah Johansen, Amber Czajkowski, Mike Scobey, Peter Egerton, and Rance Fortenberry, PhD (2017). Thin-Film Interference Filters for LIDAR. Alluxa White Paper Series. http://www.alluxa.com/learning-center.

[2] Alluxa engineering staff (2012). Flat Top, Ultra-Narrow Bandpass Optical Filters Using Plasma Deposited Hard Oxide Coatings. Alluxa White Paper Series. http://www.alluxa.com/learning-center.

This information has been sourced, reviewed and adapted from materials provided by Alluxa.

For more information on this source, please visit Alluxa.