Light Detection and Ranging (LIDAR) is arguably the most versatile active remote sensing technique, employed across various disciplines and platforms. LIDAR is one of the most important technologies in atmospheric and Earth sciences. Currently, the technology is being used in autonomous vehicles, security, urban planning, infrastructure development, and other applications. This flood of unique applications has brought with it a renewed interest, and an entry of technological advancements, in LIDAR sensors. This has made the technology more accessible and reduced the cost.
In order to progress with the technology, the LIDAR interference filters should be designed to improve signal-to-noise ratios by isolating the target LIDAR return signal. Recent technological developments in thin-film, ultra-narrow bandpass interference filters have enabled: less than 0.1 nm bandwidths, more than 95% transmission, a square spectral shape, steep edges, wide range blocking measured at more than OD 8 (-80 dB), minimal thermal dependence, and uniform coatings. High performance LIDAR interference filters have reduced the requirement for multiple filtering techniques, leading to an improvement in the signal-to-noise ratios.
Principles of LIDAR
Most LIDAR sensors, such as laser altimeters, employ a 'time of flight' technology. A pulsed laser is scanned across the environment, and the return time of the reflected signals is determined. The return time is calculated based on the precise position and orientation of the sensor when the signals are emitted and received. This is shown in Figure 1. In order to achieve this, a LIDAR system must have five basic components: a laser, a receiver or photodetector, a mechanical or software-based scanning system, a high-precision clock, and a GPS unit. To find out the orientation, aerial LIDAR systems also need to have an inertial measurement unit (IMU).
Figure 1. An aerial laser altimeter system used for mapping topography and canopy cover. Image credit: Alluxa
The distance between the sensor and the object is determined by the basic equation:
where R is the range in meters, na is the index of refraction of air, c is the speed of light, tp is the time when the pulse is emitted, and t is the time when the signal returns.
The result, as shown in Figure 2, is a data point cloud. High-resolution digital elevation models (DEMs) or 3D images of features in the surrounding environment can be created using this data point cloud. A single LIDAR pulse can encounter several objects, resulting in several reflected signals. Based on the associated software, LIDAR systems will either display the data as a waveform showing each return as a function of time, or record these returns as discrete points (Figure 1).
Figure 2. LIDAR data point cloud showing a detailed 3D street view. Image credit: Oregon State University
A filtering technology must be incorporated into the LIDAR receiver in order to isolate the return signal. This is needed for operation during daylight hours or in the presence of stray light. Common filtering systems include Fabry-Pérot interferometers, interference filters, Fabry-Pérot etalons, atomic line filters, spectrometers, and filtering algorithms. A combination of two or more of these filtering methods is generally needed in LIDAR systems that require a high level of precision.
The choice of filter depends on the exact system requirements and LIDAR application. Generally, LIDAR systems use hard-coated, thin-film interference filters due to their inherent durability, and maintenance-free and calibration-free operation . This is important as LIDAR sensors can be mounted on airplanes, UAVs, satellites, and autonomous vehicles. Sensors on these platforms must be able to function under extreme environmental conditions with little to no maintenance.
LIDAR Interference Filters
Thin-film interference filters are produced by depositing alternating layers of materials onto a substrate, where the materials have contrasting indices of refraction. When light passes through the filter, a part of the light is reflected at each layer, which results in internal interference. The net result, based on the configuration and thickness of the layers, is that certain wavelengths of light are transmitted through the filter, while others are reflected off of it or absorbed by it.
Figure 3. A hard-coated, flat-top, 532 nm ultra-narrow bandpass interference filter with > 95% transmission, steep edges measured to OD 7 (-70 dB), and wide-range OD 7 blocking.
The most common type of interference filter employed to isolate LIDAR return signals is the ultra-narrow bandpass filter (shown in Figure 3). This filter can have narrow bandwidths of even 0.1 nm, without compromising transmission, which is usually more than 90%.
The design is based on Fabry-Pérot resonant cavities, in which two dielectric reflectors having pairs of low and high index layers, each with an optical thickness of ¼ wavelength, are symmetrically separated by a spacer consisting of one or more ½ wavelength thick layers. This combination of mirror pairs with symmetrical separation by a spacer allows the formation of a single cavity. When a single cavity is used in the design, it results in a filter with a peaked spectral shape.
However, when more cavities are added to the design, the passband yields a transmission spectrum with a square spectral shape, flat top, steeper edges, and more attenuation just outside the passband (Figures 4 and 5). Therefore, multi-cavity designs enable higher LIDAR signal-to-noise ratios and more precise transmission of the target signal.
Figure 4 & 5. The effect of cavity count on filter shape and out-of-band blocking. Higher cavity counts result in steeper edges, deeper blocking, and a square spectral shape.
By controlling layer thickness during the manufacturing process, flat-topped ultra-narrow filters with high-transmission and low passband ripple that consistently match theory are able to be produced. Alluxa is ble to achieve this by employing an advanced software-controlled system, that tracks and optimizes the optical thicknesses of all layers in real-time.
LIDAR interference filters should be specified with broad range out-of-band blocking in porder to isolate the target signal from sunlight and other extraneous light sources. Dielectric reflectors are employed in the design of the blockers. In the reflectors, the mirror pairs are stacked and manipulated in a specific way, so that the filter attenuates light across broad wavelength regions. Specific blocking levels and blocking ranges are application-dependent and are discussed later in this article.
Interference Filters for Specific LIDAR Applications
Laser altimeters are the most common type of LIDAR sensors. They detect the laser pulse echo when it is reflected off of an object, the water, or the ground. Topographic LIDAR sensors generally emit an eye-safe 1064-nm pulsed Nd:YAG laser that reflects at the same wavelength. Both the bare ground and the canopy can be mapped by the resultant point cloud, because of multiple returns.
Bathymetric LIDAR ensors use a 532-nm frequency-doubled pulsed Nd:YAG laser, which can penetrate the surface of water. This laser is employed alone or combined with a 1064-nm laser that reflects off of the surface. Monitoring of glaciers and ice sheets over time can also be achieved by employing the 532-nm laser, as seen with the Advanced Topographic Laser Altimeter System (ATLAS) on board NASA's soon-to-be-launched ICESat-2 mission.
For laser altimeters, ultra-narrow interference filters must have a center wavelength (CWL) at the laser line and a full width half maximum (FWHM), or a bandwidth between the points where edges measure at 50% of peak transmission, that is 1.5 nm or narrower. The specific instrument and the laser wavelength dictate the out-of-band blocking requirements. Greater than OD 6 (-60 dB) blocking from 300 nm to 1300 nm is commonly found in many sensors while > OD 5 over a narrower wavelength range is enough in others. While single-cavity filters are a cost-saving option, multi-cavity designs are well-suited for these systems (Figures 4 and 5).
LIDAR for Autonomous Vehicles
The LIDAR sensors incorporated into autonomous vehicles stem from the basic laser altimeter technology (Figure 6). These sensors generally make use of a 905-nm or 1550-nm pulsed laser together with a photodetector, such as a silicon avalanche photodiode. The sensors have a mechanical scanning system for capturing a 360° field of view (FOV) or they can be fully solid state. Solid state LIDAR units are inexpensive and small, and do not have moving parts, since they are a software-based scanning system. Resolution and accuracy are comparable to mechanical systems, but the major trade off is reduced FOV. This issue can be resolved by mounting more than one unit onto a single vehicle.
Figure 6. Diagram showing a LIDAR system for an autonomous vehicle. Image credit: Alluxa
The filter specifications for mechanical LIDAR systems and laser altimeters are similar. However, fully solid state systems usually rely on a filtering algorithm for isolating the reflected signal. In some instances, the algorithm can be combined with a wideband interference filter having a > OD 4 wide-range out-of-band blocking.
Fluorescence LIDAR systems detect the emission of various naturally occurring molecules, such as carotenoids, chlorophylls, phycobilins, or other photosynthetic pigments, when laser-excited. In water, this enables the detection of toxic red tide events or increased concentrations of cyanobacteria and algae, indicating hypoxia. On land, fluorescence LIDAR finds use in vegetation studies and in the evaluation of the conservation status of cultural monuments.
In atmospheric research, fluorescence LIDAR is used for the study of airborne pathogens, complex organic aerosols, and atmospheric gases. Moreover, NASA has recently proposed a fluorescence LIDAR technology, called the Bio-Indicator LIDAR Instrument (BILI), to detect organic molecules and bio-signatures on other planets.
In these units, the bandpass CWL and FWHM are placed in accordance with the fluorescence emission spectrum of target molecule. For a high signal-to-noise ratio, the laser line and extraneous light should be attenuated to a level of ≥ OD 6 for land or aquatic studies, but for atmospheric systems, deeper blocking levels are necessary.
Mie, Rayleigh, Raman, and Doppler Atmospheric LIDAR Systems
Air quality monitoring is carried out by Mie and Rayleigh LIDAR instruments. These instruments are ground-based or forward-facing aerial systems, detecting the elastic Mie or Rayleigh backscatter of aerosols and homonuclear diatomic gases at laser wavelengths of 355 nm, 532 nm, and 1064 nm. Doppler LIDAR is sometimes incorporated into these systems and used to determine wind velocity. This is done by detecting the Doppler shift of the elastic backscatter from gases or aerosols.
Raman LIDAR instruments detect the inelastic rotational and vibrational Raman scattering that occurs when molecules of interest are excited by a laser. Pure rotational Raman signals can also be isolated by using interference filters, thus enabling particle extinction, air temperature, and other properties to be determined [5,7].
Since atmospheric LIDAR systems are dependent on relatively weak backscatter signals, it is important to maximize the signal-to-noise ratio. The LIDAR equation for atmospheric systems in its basic form indirectly highlights the importance of employing high-performance interference filters:
P (R) = K * G (R) * β (R) * T(R)
where P(R) is the signal power, K is the system efficiency, G(R) is the measurement geometry, β(R) is the backscatter coefficient, and T(R) is the transmission term. Other than K, all others are a function of range. Modifications to the LIDAR system can only improve K and G(R), because the other two terms describe inherent atmospheric optical properties . Since system efficiency depends on receiver efficiency, signal power can be improved by using high-performance interference filters.
While for some Mie and Rayleigh LIDAR sensors OD 6 or 7 blocking can be enough, Raman LIDAR filters should preferably have blocking exceeding OD 7 or 8 [1,5]. This is particularly needed at the laser wavelength to allow the blocking of the stronger elastic backscatter signals. However, the tight transition from >90% transmission at the Stokes or Anti-Stokes branch of the Raman signal to > OD 7 blocking at the laser line also requires a high-cavity-count design in order to ensure adequate steepness.
Recently, Veselovskii et al. employed an Alluxa-manufactured ultra-steep Raman LIDAR interference filter to isolate a relatively temperature-insensitive portion of the Anti-Stokes pure rotational Raman spectrum for O2 and N2 following a 532.12-nm laser excitation. This filter was designed with a slope of < 0.1% of edge wavelength from 90% transmission to OD 4. In this specific case, more than 95% transmission in the passband enabled two identical interference filters to be used in series in order to achieve OD 8 rejection of the elastic signal with a < 1 nm transition to the passband .
Differential Absorption LIDAR
Differential Absorption LIDAR (DIAL) units make use of the absorption spectra of atmospheric gases and water vapor to determine their concentrations. DIAL systems use two different lasers. While the first is tuned to an on-resonance, or high-absorption, wavelength for the target molecule, the second is tuned to an off-resonance wavelength. The on and off-resonance wavelengths are usually less than 1 nm apart, thus enabling researchers to assume that any measured backscatter differences are caused by absorption by the species of interest, allowing a concentration profile to be determined from the backscatter ratio as a function of range.
Since the near and far channels of DIAL systems are very close in wavelength, these systems usually require multiple filtering techniques. Ultra-narrow interference filters are used for sunlight attenuation, while Fabry-Pérot etalons are employed to separate the channels . A FWHM of < 0.5 nm and > OD 6 wide range out-of-band blocking are commonly seen in these ultra-narrow interference filters. Multi-cavity designs are necessary to achieve high sunlight attenuation at wavelengths close to the passband.
Measuring Ultra-Narrow Bandpass Filters
The spectral resolution and light sources of many spectrophotometers, and other grating-based measurement systems, are not adequate to measure and evaluate steep edges, blocking beyond ~ OD 6, or ultra-narrow bandpass filters. For instance, when the filter edge transition from high transmission to deep blocking is steep enough to rival the spectral resolution of the spectrophotometer, the filter edge will fail to resolve properly. Therefore, the actual steepness will fail to be reflected in the measurement. This also holds true for filters with less than ~ 1 nm FWHM. In such cases, the effect of edge smearing is compounded as the bandwidth is also narrow enough to rival the spectral resolution of the instrument. This yields an unresolved passband, with the final measurement exhibiting relatively low transmission and a rounded off peak.
By introducing narrow apertures and reducing the spectral beam width, spectral resolution can be enhanced in standard spectrophotometers. However, such strategies also decrease the light intensity, which results in a reduced signal-to-noise ratio and a higher measurement noise floor. This makes it more difficult to measure blocking. Removing the apertures, increasing the spectral beam width and attenuating the reference beam can lower the noise floor of many of these instruments. This technique is useful to measure wide blocking regions specified up to ~ OD 6 or 7. However, it must not be used to assess the filter edges or passband, because the coarse spectral resolution will result in significant edge smearing and an inaccurate bandwidth.
More advanced measurement technologies are needed to measure high-performance and ultra-narrow filters. One common method used to measure ultra-narrow bandpass filters involves a tunable laser, where filter transmission is precisely measured at each wavelength within the range of the laser. However, several lasers are needed to measure the full transmission and blocking ranges of a single filter, as many tunable lasers have a narrow wavelength range. For this reason, this method is not suitable for high-volume manufacturing scenarios.
Recently, Alluxa has developed a new system to accurately measure its highest-performance and narrowest bandpass filters, as shown in Figure 7. The system is also robust enough for use in a high-volume manufacturing setting. The HELIX Spectral Analysis System can resolve bandwidths of less than 0.1 nm, blocking ranges of > OD 8, and transitions as steep as 0.4% of the edge wavelength from 90% transmission to OD 7. Alluxa’s new system is particularly useful for evaluating the true performance of ultra-narrow filters for Raman LIDAR and other demanding applications, requiring very steep edges and > OD 8 blocking.
Figure 7. A 607.4-nm LIDAR interference filter for an N2 Raman channel measured with both a standard spectrophotometer and Alluxa’s HELIX Spectral Analysis System. The HELIX System is able to resolve filter edges all the way to OD 7 (-70 dB).
Additional Filter Considerations
The chamber geometry and deposition process should be precisely controlled to maximize coating uniformity. If uniformity is left uncontrolled, the surface of the filter will have variable layer thicknesses, resulting in a location-dependent wavelength shift of the filter spectrum across the clear aperture of the part. It is especially important to eliminate this wavelength shift for narrowband LIDAR interference filters because of the precise nature of the LIDAR return signals.
Accurate control of uniformity is a multivariable problem. The usual techniques include adding a physical mask into the coating chamber, or changing the coating chamber geometry, such as the angle or distance between the source material and the substrate. Additional changes to uniformity can be carried out by changing the coating process variables, such as deposition rate and temperature. However, such changes can lead to other complications in the deposition process, adversely affecting the final product.
Alluxa's advanced uniformity control system decreases the location-dependent wavelength shift at the time of coating, so as to achieve precision uniformity over a large area. As shown in Figure 8, such a strategy yields a filter with minimal spectral variation across the clear aperture (CA), even for large-format parts.
Figure 8. Measured results of a 72-mm diameter ultra-narrow filter manufactured using Alluxa’s advanced uniformity control system demonstrating < 0.035% variation in CWL over the clear aperture.
Ground-based and aerial LIDAR systems are operated at temperatures ranging from -40°C to +105°C. Satellite LIDAR operating ranges depend on the thermal control system and the orbit of the satellite. Interference filters incorporated into systems that work at extreme temperatures must be designed to maximize temperature stability.
Extreme temperatures can result in the contraction or expansion of the thin-film layers in standard interference filters, leading to a shift in the wavelength of the passband. This shift can be significant if the filter has not been specifically designed to operate in extreme environments.
As shown in Figure 9, Alluxa's hard-coated ultra-narrowband interference filters show a very low temperature-dependent wavelength shift, varying between 2pm/°C and 5pm/°C within the operating range of many LIDAR instruments . Thermal stability can be enhanced even more by selecting a substrate material with a higher coefficient of thermal expansion (CTE) than the CTE of the coating. It can also be done by modifying the material ratios and other design properties. Post-deposition annealing can also help with chemical and thermal stability. In this process, the thin-film layers undergo thermal expansion while the materials are further oxidized, reducing the coating stress.
Figure 9. Measured performance of a narrowband filter when heated from room temperature to 105°C.
Other Thin-Film Optical Components for LIDAR
Other than bandpass interference filters, LIDAR systems also make use of dichroic filters, mirrors, and other high-quality thin-film optical components. High-reflectivity dielectric scanning mirrors with > 99.5% reflection are employed in many LIDAR sensors, while ultra-light beryllium mirrors are suitable for fast scanning systems. Dichroic filters with regions of high reflectivity and high transmission, separated by a steep edge transition, allow the return signal to be directed to the appropriate receiver channel. When designing a LIDAR sensor, the performance of these optical components should also be considered in order to optimize system efficiency.
Thanks to several new applications, LIDAR is becoming more affordable, efficient and accessible. Recent technological improvements have enabled increased resolution and accuracy, even in small systems that can be discretely hidden within the side mirrors of self-driving cars. Thin-film interference filters have progressed with the technology through modern, sophisticated design and coating techniques, tight uniformity control, advanced measurement systems and minimal thermal dependence. Alluxa's high-performance interference filters will maximize system performance and signal-to-noise ratio for any LIDAR instrument.
 Hauchecorne, A., Keckhut, P., Mariscal, J., C., d'Aameida, E., Dahoo, P., and J. Porteneuve. (2016). An innovative rotational Raman LIDAR to measure the temperature profile from the surface to 30 km altitude. EPJ Web of Conferences, 119, 06008. DOI: 10.1051/epjconf/201611906008.
 Kovalev, V. A. and W. E. Eichinger (2004). Elastic LIDAR: Theory, practice, and analysis methods. John Wiley & Sons, Inc. Hoboken, NJ.
 Scobey, M., Egerton, P., Fortenberry, R., and A. Czajkowski (2013). Ultra-narrowband optical bandpass filters with large format and improved temperature stability. Alluxa White Paper Series. http://www.alluxa.com/learning-center.
 Spuler, S. M., Repasky, K. S., Morley, B., Moen, D., Hayman, M., and A. R. Nehrir. (2015). Field-deployable diode-laser-based differential absorption LIDAR (DIAL) for profiling water vapor. Atmos. Meas. Tech. 8: 1073-1087.
 Veselovskii, I., Whiteman, D. N., Korenskiy, M., Suvorina, A., and D. Perez-Ramirez. (2016). Implementation of rotational Raman channel in multiwavelength aerosol LIDAR to improve measurements of particle extinction and backscattering at 532 nm. EPJ Web of Conferences, 119, 17002. DOI: 10.1051/epjconf/201611917002.
 Wandinger, U. (2005). Chapter 1: Introduction to LIDAR. LIDAR: Range-resolved optical remote sensing of the atmosphere. Edited by Weitkam, C. Springer Science+Business Media, Inc. New York, NY.
 Wandinger, U. (2005). Chapter 9: Raman LIDAR. LIDAR: Range-resolved optical remote sensing of the atmosphere. Edited by Weitkam, C. Springer Science+Business Media, Inc. New York, NY.
This information has been sourced, reviewed and adapted from materials provided by Alluxa.
For more information on this source, please visit Alluxa.