Open Access
Add to BibSonomy
Abstract
Miniaturized, energy-efficient and application-specific spectral sensing systems promise to be a highly sought-after technology in the coming years, with potential applications in areas such as: distributed sensor systems, IoT devices, mobile autonomous platforms, and many others. We present in this work the design, construction and measurement results of a compact, mid-infrared spectrometer working in the 3 - 4 µm spectral region, attractive for applications requiring the identification of polymer materials. The spectrometer is based on linear-variable filters (LVF) combined with an uncooled HgCdTe linear-detector array (LDA). The design and architecture of the device is described and discussed in the context of miniaturization challenges and constraints. Measured spectra of thin polyimide and polystyrene foils are presented to prove the applicability of the developed device to polymer materials detection and identification.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Methods based on quantitative measurements of light-matter interactions lie at the heart of many modern optical analytic techniques. Spectral analysis of light that has interacted with a sample enables the detection and identification of various substances. Spectrometers of that kind are often designed with a specific attention given to interactions that occur in the mid-infrared (MWIR or mid-IR) spectral range, which is defined in the standardization documents as the range from 3 to 50 $\mathrm {\mu }$m [1].
Photons from this spectral region can strongly interact with vibrational states of chemical bonds in various molecules [2], revealing well-defined and most importantly – unique spectral characteristics. These interactions occur especially for wavelengths in the 7 – 25 $\mathrm {\mu }$m wavelength region, also called the "fingerprint region" [3].
Devices that can reliably analyze optical spectra, especially in the mid-IR, are however usually bulky and have to be operated by trained personnel in specialized laboratories. Measurements of samples collected in the field are thus generally not performed in situ and samples have to be transported to off-site laboratories. One of the possible solutions to this issue is a radical miniaturization and portabilization of spectrometers while maintaining parameters that are at least sufficient for detection and/or identification and/or analysis of samples measured in the field, depending on the application. Designing such application-specific devices and optimizing certain aspects of their functionality (e.g. low spectral resolution and very fast operation, high resolution and a very narrow spectral range) can therefore be the key to developing devices that enable reliable spectrometry in on-site and field applications.
Potential applications of miniaturized spectrometers include, but are not limited to: drones and other autonomous systems for wide-area monitoring [4,5], sensor systems and methods applicable to smart agriculture [6] or gas leakage monitoring [7], handheld devices for food quality analysis and safety monitoring [8,9]. A significant advantage provided by this level of miniaturization is the related decrease in power consumption, especially important for devices that have to function using limited internal power supplies. Micro- and nanoscale spectrometers may also find use in more exotic applications such as microfluidic chips [10] or wearable biomedical devices [11]. Moreover, miniature spectrometers can be successfully employed in multi- and hyperspectral imaging [12,13], especially when their architecture enables simple integration with focal-plane arrays that don’t require the use of scanning systems [14].
Miniature spectrometer designs and their applications have been developed for over 30 years, yielding a wide range of system architectures and approaches: from miniaturized and modified classical monochromator architectures (Czerny-Turner [15], or Littrow [16]), through MEMS-based systems (planar/lab-on-chip Michelson [17,18] and Mach-Zehnder [19] interferometers for Fourier transform spectroscopy, tunable Fabry-Pérot cavities [20]), up to detector arrays featuring an engineered spectral response [21].
Modern miniature spectrometers can be divided into four categories according to their operating principles, as presented in [22]:
- 1. Dispersive/diffractive optics (e.g. using transmission or reflection diffraction gratings, prisms or waveguide dispersion based methods)
- 2. Narrowband filters (using non-dispersive methods such as optical thin-film bandpass layers, in scanned or static architectures)
- 3. Fourier transform (designs using dynamic/scanned or static interferogram acquisition methods and subsequent harmonic analysis for spectrum retrieval)
- 4. Reconstructive techniques (precalibrated, numerical reconstruction of spectral characteristics from data acquired using scattering media, detector and/or filters with a set of random, quasi-orthogonal and/or otherwise tailored responses)
In this paper we describe the development, manufacturing and testing of a miniature spectrometer module that can be classified under the 2nd category, operating in the 3 – 4 $\mathrm {\mu }$m wavelength range.
To the best of our knowledge this device is a novel design for the mid-infared region, combining uncooled, HgCdTe linear detector arrays and linear-variable filters in this spectral range. The spectral selectivity of the combined elements, coupled with a broadband thermal light source, allows reliable qualitative detection and identification of thin polymer foils (e.g., polyimide, polystyrene) by analyzing their spectral absorption characteristics. Results provided by the constructed demonstrator serve as a proof-of-concept for further miniaturization of the device and work towards compact, integrated spectrometer systems.
2. Design goals and theoretical parameters
Our device was designed with three main goals in mind:
- a. Achieving an application-adequate spectral resolution in the selected spectral range
- b. Featuring no moving/actuating parts for spectral scanning
- c. Being as compact as possible, with an initial design goal of a volume below 3 cm3
The first design goal (adequate resolution in a spectral range) was defined based on the spectral range intended to be covered by potential applications of the constructed device, such as industrial monitoring and detection of polymers (e.g. on production lines) and water quality monitoring (e.g. detection of microplastics). At the same time, achieving the second goal (no moving parts) which was dictated as part of the effort to simplify the design as much as possible, meant that there was a relatively limited choice of spectrometer architectures, having to relinquish scan-based acquisition of spectral characteristics in the device.
The spectrometer design is based on two core components - a linear-variable filter (LVF) and a linear detector array (LDA).
Linear-variable filters are optical elements that transmit different wavelengths in different points along their length, usually by means of deposited optical thin-films [23] or by using a tilted-mirror optical resonator cavity [24].
The LVF is placed in close proximity to the LDA, effectively restricting the spectral response of each pixel on the LDA’s to a narrow wavelength band and providing sufficient spectral selectivity of the system. Similar designs have been proposed in e.g. [25–29], however to the best of our knowledge, our design implements an uncooled HgCdTe (MCT) linear photodiode array for the first time in a device of this type.
To test the feasibility of the LVF+LDA design we calculated several key parameters of the system.
2.1 Optical power and SNR
Optical power in the system is attenuated primarily due to the transmission characteristic of the LVF. Around 50% of light passing through it is attenuated – additionally, each pixel of the LDA receives only a small amount of power from a narrow wavelength range, due to the spectral transmission characteristics of the LVF.
A secondary cause of losses is the LDA’s geometry. The effective photosensitive area of the detector is a fraction of its total size due to distances between pixels. For a 32-element 125$\times$1000 $\mathrm {\mu }$m array with 25 $\mathrm {\mu }$m of space between each pixel, the effective optical area can be calculated from the total LDA area A and is equal to $A_{eff} = 0.83 A$.
We assume a spectrally flat, broadband source emitting in the range of 3 – 4 $\mathrm {\mu }$m, resulting in a spectral width of $W_{\lambda }$ = 1000 nm. Assuming an average FWHM = 70 nm at each point of the LVF, after losses related to the LVF’s transmission T = 0.5 and LDA surface area Aeff, each of the N = 32 pixels will be receiving the power Popt, calculated in Eq. (1).
2.2 Spectral resolution and resolving power
The theoretically achievable resolution of our module is influenced by two main factors – the bandwidth of the LVF’s transmission in each point of its length and the density of pixels on the LDA. We assume, using a simplified resolution criterion [31], that at least three pixels are needed to sample a spectral feature.
The theoretical achievable resolution in our setup can be calculated as shown in Eq. (4).
The resolving power of our module can be then calculated as shown in Eq. (5).3. Design, construction and initial measurements of the spectrometer
To test the idea experimentally we designed a spectrometer for the 3 - 4 $\mathrm {\mu }$m spectral range. This range of wavelengths was chosen due to the presence of characteristic absorption peaks of potential polymer analytes, such as polystyrene (PS), low- and high-density polyethylene (LDPE, HDPE) and polyimide (PI) to enumerate a few.
For the spectrometers LDA we used a custom-designed VIGO Photonics 32-element photodiode array [32], consisting of 32 HgCdTe (MCT) photodiodes, each featuring a 125 $\mathrm {\mu }$m × 1000 $\mathrm {\mu }$m photosensitive area.
The detector array’s spectral response was designed to accommodate the LVF’s spectral transmission range, providing a relatively high spectral responsivity (>1.5 A/W) in the 3 – 4 $\mathrm {\mu }$m wavelength range. The theoretical specific detectivity D* of each pixel was 3.9 $\cdot$ 1010 cm $\cdot$ Hz1/2 $\cdot$ W−1 in a temperature of 300 K.
The theoretical spectral responsivity of the LDA’s photodiodes was calculated for the 2 - 6 $\mathrm {\mu }$m spectral range. In order to confirm their proper operation in the desired range, spectral responses of the fabricated photodiodes were measured. Both characteristics are shown in Fig. 1.
Fig. 1. Theoretical spectral responsivity (blue) and the measured spectral response (green) of the detector array in the 2 – 6 $\mathrm {\mu }$m wavelength range
To measure the LDA’s responses we used a modified PerkinElmer Frontier FTIR spectrometer in which the stock photosensitive element was replaced with the measured detector array in an external optical system, coupled with a single-channel, ultra low noise amplifier circuit, connected to each of the LDA’s elements separately. The resulting amplified signal was then coupled back into the FTIR and processed, yielding the spectral response of the measured channel.
The measured spectral response correlates with the theoretical responsivity shape, apart from the clearly visible absorption peaks at approx. 2.7 $\mathrm {\mu }$m, 3.4 $\mathrm {\mu }$m and 4.3 $\mathrm {\mu }$m which are related to respectively: water vapor and CO2 absorption, absorption of the silicate underfill medium inside the LDA, CO2 absorption [33]. Additionally, a fringing pattern is visible between 3.5 $\mathrm {\mu }$m and 4.5 $\mathrm {\mu }$m, stemming from the photodiode GaAs substrate which forms a low-Q resonance cavity.
The detector’s pixels were 1 mm long in order to increase their photosensitive area and in effect improve the SNR of the system. This had the drawback of decreasing the shunt resistance of the photodiodes and potentially increasing noise currents. Measurements of the fabricated arrays indicate resistance values of approx. 70 - 90 $\mathrm {\Omega}$.
The array was placed on an intermediate Al2O3 substrate with etched gold signal traces which was then mounted on top of a three-stage thermoelectric cooler (TEC), inside a Kovar, butterfly-type package. When the package is hermetically sealed, the LDA contained within can be cooled down to 230 K, increasing the achievable detectivity by up to two orders of magnitude. However during further measurements we didn’t use the cooler, in order to both reduce the energy footprint of the device and confirm its correct operation in 300 K. Pads on the intermediate Al2O3 substrate are wirebonded to a PCB placed inside the package, around the array. Figure 2 shows the LDA’s geometry and package.
The LVF we used in the spectrometer design was a commercially available filter from Vortex Optical Coatings [34], operating in the 2.5 - 5 $\mathrm {\mu }$m spectral range. The spectral resolution calculated in Eq. (4) is lower than the average FWHM of the used LVF (50 - 80 nm, increasing for longer wavelengths), and thus won’t be limited by it. The chosen LVF operates using thin-film optical filters deposited on one of its faces, which vary the transmitted central wavelength along its length linearly. In order to block unwanted transmission due to higher order interferences, the LVF’s backside is covered with a broadband optical filter.
Experimental work was initiated by measuring the spectral response of the selected commercial LVF using a single-element photodiode, which had a similar spectral characteristic to the LDA’s pixels. The measurement scheme is shown in Fig. 3.
The LVF was placed in front of the exit aperture of a FTIR spectrometer (Perkin Elmer Frontier FTIR [35]) with an external measurement path. A single-element photovoltaic detector that had a similar spectral response to the detector array (Fig. 1), was mounted on an X-Y translation stage and moved to one edge of the LVF. Signals from the detector were amplified in an external two-stage circuit. Spectral characteristics in the range of 2 – 6 $\mathrm {\mu }$m were acquired by scanning along the length of the LVF, one measurement every 150 $\mathrm {\mu }$m, approximating the geometry of our LDA. Results of the single-element measurements are shown in Fig. 4.
Fig. 4. Results of spectral transmission measurements of the commercial LVF performed with a single-element detector.
Each position along the LVF had a shifted spectral response in the range between 3.1 and 4.2 $\mathrm {\mu }$m. In the spectral range around 2.2 $\mathrm {\mu }$m secondary transmission peaks of the filter can be seen, however this did not influence the measurements.
After this measurement, a LVF was cut down on each side in order to work in the narrower 3 - 4 $\mathrm {\mu }$m spectral range. The modified filter was attached to the LDA’s package using a resin-printed mount which ensured a precise geometric orientation between those two components. The spectral response of each of the 32 channels was then acquired using the measurement setup shown in Fig. 5.
In this setup the measured LVF and LDA remained static and only the output channels were changed. Once again an external, two-stage amplifier circuit was used. The results of the 32-channel measurements are shown in Fig. 6.
Fig. 6. Spectral transmission characteristic of the LVF measured with an LDA. Each of the displayed peaks corresponds to a single channel.
The measured spectral responses of the LVF+LDA setup clearly show a cutoff at approx. 3.75 $\mathrm {\mu }$m, different from the anticipated 4 $\mathrm {\mu }$m. This is presumably the result of an unintentional position mismatch between the LVF and the LDA, where the LVF’s long-wave edge is partially outside of the detector array. Another possible factor that reduces the long-wave range of the device is a non-parallelity between the LVF and LDA, which could change the responses seen on the LDA. Both factors are currently being tested in a new version of the system which will optimize the LVF’s mount.
After initial measurements which confirmed that the setup is spectrally selective, we started work on a compact electronic amplification circuit that could be integrated with the LVF+LDA module. The designed circuit consists of two stages: first a transimpedance amplifier (TIA) that converts the current signal to a voltage signal and a secondary operational amplifier (op-amp) which provides additional gain. Each of the LDA’s 32 channels is amplified separately, an approach chosen to reduce potential cross-talk and increase the systems acquisition speed. Moreover, this allows us to control and mitigate the differences between responses of the LDA’s channels by modifying individual amplification circuits. A significant drawback is the increased complexity of the circuit and its high cost due to multiple amplifier ICs.
After optimizing its behavior in a laboratory setting, the setup was subsequently miniaturized and implemented in the form of a compact demonstrator device. Figure 7 shows the demonstrator’s structural schematic.
As shown in the schematic, the system is divided into three distinct parts: optical, electrical (subdivided into analog and digital electronics) and mechanical. Figure 8 shows the hardware implementation.
Fig. 8. Internal structure of the designed spectrometer. S - light source; M - OAP mirror; L - cylindrical lens; LVF - linear-variable filter; LDA - linear detector array.
3.1 Optical system
The optical system was first developed as a part of the laboratory setup and then subsequently miniaturized. Figure 9 shows the spectrometer’s optical system.
Fig. 9. Optical system of the designed 3 - 4 $\mathrm {\mu }$m spectrometer. S - miniature black-body source; OAP - off-axis parabolic mirror; L - cylindrical lens; LVF - linear-variable filter; LDA - linear detector array.
Broadband light is emitted by the miniature black-body source [36] used in the previously described laboratory setup. A miniaturized chopper wheel is placed in front of the source, providing amplitude modulation to the optical signal. The light is collimated using the same OAP [37] as previously.
The collimated light is then focused in one axis (shown as the solid lines in Fig. 9) using a 1” CaF2 cylindrical lens (L) [38] and passes through the LVF where it is selectively filtered. Focusing the collimated beam in only one axis with L allowed us to increase the signal returned by the setup while keeping the spectral resolution unchanged. Figure 10 shows the optical module of the spectrometer.
Fig. 10. Light source (S), chopper, collimator (M/OAP) and focusing lens (L) which form the spectrometer’s optical module.
The filtered light then falls onto the linear detector array (LDA) and is amplified, as described before.
As shown in Fig. 7, spectral characteristics of samples are measured in a transmission configuration. The system was initially designed to analyze samples placed in a collimated light beam - due to miniaturization constraints in the spectrometer’s design we moved the sample insertion port between the cylindrical lens and the LVF+LDA module. During measurements performed on thin polymer foils (described in Section 4) no noticeable differences between results from the two configurations were observed.
3.2 Electronic and software design
The device is equipped with an internal power supply circuit that provides power to the light source, signal amplifiers, and chopper motor. This allows operation of the spectrometer with a single 12V DC power supply.
Amplified analog signals from the device are acquired using two external Advantech USB-4716 [39] data acquisition cards (DAQs). Each of the DAQs serves 16 separate channels with a 16-bit precision and 200 kS/s. The cards are controlled over USB using a python script executed on an accompanying PC.
In order to increase the effective signal-to-noise ratio during spectrum acquisitions, the voltage values on each channel were averaged over 50 measurements, resulting in an acquisition and processing time of approx. 0.5 s.
4. Measurements of thin polymer foils in the 3 - 4 $\mathrm {\mu }$m wavelength range
To test the functionality of the developed spectrometer demonstrator, the transmission characteristics of thin polymer foils made of polyimide (PI) and polystyrene (PS) were measured. Samples of the foils were approximately 17 $\mathrm {\mu }$m thick.
Each measurement consisted of first acquiring a background spectrum and then a sample spectrum. Spectral transmission characteristics were calculated by dividing the acquired sample spectra by the background. We performed reference transmission measurements of the same samples using a PerkinElmer Frontier FTIR spectrometer [35] with a spectral resolution of 16 cm−1.
Figure 11 shows the spectral transmission of PI measured using the developed spectrometer, overlaid on the referential FTIR measurement.
Fig. 11. Spectral transmission characteristic of a thin polyimide foil in the 3 - 4 $\mathrm {\mu }$m wavelength range, measured using the LVF-32E spectrometer (PI_LVF) and a PerkinElmer Frontier FTIR spectrometer (PI_FTIR).
Figure 12 shows the spectral transmission of PS measured with the constructed spectrometer, overlaid on the referential FTIR measurement.
Fig. 12. Spectral transmission characteristic of a thin polystyrene foil in the 3 - 4 $\mathrm {\mu }$m wavelength range, measured using the LVF-32E spectrometer (PS_LVF) and a PerkinElmer Frontier FTIR spectrometer (PS_FTIR).
Initial tests with thicker polymer materials have been performed but due to increasing light losses have been unsuccessful, with optical signal values below of the detector’s detection limit for materials thicker than approx. 30 $\mathrm {\mu }$m.
The presented results allow for repeatable, qualitative determination of the polymer material under test, regardless of the low spectral resolution in our system.
5. Conclusions and perspectives for further development
Results presented in Figs. 11 and 12 confirm the possibility of qualitative identification of material (in this case polymer) types using the described spectrometer system, irrespective of the relatively low spectral resolution.
The spectrometer design can be potentially extended, using longer and/or denser linear detector arrays and linear variable filters with narrower spectral transmission characteristics, the conjunction of which would yield an improved resolution.
An advantage of the proposed architecture also lies in the possibility of tailoring spectral responses of each pixel towards application-specific sensing, e.g. measuring only selected spectral features (necessary for detection and identification) with variable spectral resolution. This would require a change in the filter in order to form a discrete, spectrally selective surface above each detector channel.
The currently constructed system exhibits several issues to be solved, of which the most significant is the apparent shift towards shorter wavelengths with regard to the reference FTIR measurement. This is caused by lack of proper length positioning of the filter (e.g. making sure where certain wavelengths lie with regard to the filter’s mechanical dimensions) and calibration. Because of this, since the spectrometer’s resolution is low and measurements count only 32 data points, the chance of missing narrow spectral lines is high.
A discrepancy between the referential and measured absorption peak amplitudes is also observable, as seen in the case of the 3.4 $\mathrm {\mu }$m peak, present in both PI and PS samples. The measured characteristics are significantly less pronounced, suggesting an issue with the response uniformity of the device.
A method to mitigate this issue could be a calibration procedure, ideally performed actively during assembly using a referential narrowband (e.g. laser or monochromator) source, in order to match the position of the LVF with the position of each respective LDA channel. Another approach we consider is the direct integration of filters on the LDA’s backside, which potentially offers improved SNR and an avenue for further miniaturization.
Scalability of the device (e.g. increasing the number of spectral channels) in its current form is strongly dependent on improving the readout method and introducing signal multiplexing before the first amplification stage, in order to reduce both the electronic system’s complexity and its cost.
Further work will focus on modifying the designed device’s LVF to operate in a slightly shifted spectral range of 3.2 - 4.4 $\mathrm {\mu }$m which should improve the SNR for shorter wavelengths. Additionally, a set of different detector arrays based on InAsSb superlattices [40] and filters designed for the 8 - 12 $\mathrm {\mu }$m wavelength range will be tested, as it promises access to more potential analytes with well-defined spectral features. Moreover, changing the detector’s composition from RoHS prohibited substances (mercury and cadmium) to safer III-V materials promises a wider range of applications, e.g. detection of polymer contaminants (micro- and nanoplastics) in water resources or in food production lines.
While the designed device promises potential for further miniaturization, e.g. by direct integration of LVF’s and redesign of the electronics, it is still constrained to a relatively big volume by the nature of its optical system. One of our goals is to make use of the mid-infrared photonic integrated circuits (PICs) platform developed as part of the MIRPIC project [41], integrating photodiodes developed for the LVF-32E with application-specific PICs, enabling a radical reduction of the device footprint.
Funding
Narodowe Centrum Badań i Rozwoju (MAZOWSZE/0090/19-00, TECHMATSTRATEG‑III/0026/2019).
Disclosures
FŁ, JK, RP: VIGO Photonics S.A. (F, E)
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. “Optics and photonics Spectral bands,” Standard ISO 20473:2007, International Organization for Standardization, Geneva, CH (2007).
2. B. Schrader, Infrared and Raman spectroscopy: methods and applications (John Wiley & Sons, 2008).
3. J. G. Smith, “Mass spectrometry and infrared spectroscopy,” Organic chemistry pp. 463–488 (2011).
4. S. Natesan, C. Armenakis, G. Benari, and R. Lee, “Use of UAV-borne spectrometer for land cover classification,” Drones 2(2), 16 (2018). [CrossRef]
5. B. Tuzson, M. Graf, J. Ravelid, P. Scheidegger, A. Kupferschmid, H. Looser, R. P. Morales, and L. Emmenegger, “A compact QCL spectrometer for mobile, high-precision methane sensing aboard drones,” Atmos. Meas. Tech. 13(9), 4715–4726 (2020). [CrossRef]
6. A. M. Cavaco, A. B. Utkin, J. Marques da Silva, and R. Guerra, “Making Sense of Light: The Use of Optical Spectroscopy Techniques in Plant Sciences and Agriculture,” Appl. Sci. 12(3), 997 (2022). [CrossRef]
7. Q. Tan, X. Pei, S. Zhu, D. Sun, J. Liu, C. Xue, T. Liang, W. Zhang, and J. Xiong, “Development of an optical gas leak sensor for detecting ethylene, dimethyl ether and methane,” Sensors 13(4), 4157–4169 (2013). [CrossRef]
8. C. A. T. Dos Santos, M. Lopo, R. N. Páscoa, and J. A. Lopes, “A review on the applications of portable near-infrared spectrometers in the agro-food industry,” Appl. Spectrosc. 67(11), 1215–1233 (2013). [CrossRef]
9. A. J. Das, A. Wahi, I. Kothari, and R. Raskar, “Ultra-portable, wireless smartphone spectrometer for rapid, non-destructive testing of fruit ripeness,” Sci. Rep. 6(1), 32504 (2016). [CrossRef]
10. I. Rodríguez-Ruiz, T. N. Ackermann, X. Mu noz-Berbel, and A. Llobera, “Photonic lab-on-a-chip: Integration of optical spectroscopy in microfluidic systems,” (2016).
11. M. Lacerenza, M. Buttafava, M. Renna, A. Dalla Mora, L. Spinelli, F. Zappa, A. Pifferi, A. Torricelli, A. Tosi, and D. Contini, “Wearable and wireless time-domain near-infrared spectroscopy system for brain and muscle hemodynamic monitoring,” Biomed. Opt. Express 11(10), 5934–5949 (2020). [CrossRef]
12. Y. Song, A. Konar, R. Sechrist, V. P. Roy, R. Duan, J. Dziurgot, V. Policht, Y. A. Matutes, K. J. Kubarych, and J. P. Ogilvie, “Multispectral multidimensional spectrometer spanning the ultraviolet to the mid-infrared,” Rev. Sci. Instrum. 90(1), 1 (2019). [CrossRef]
13. A. Bodkin, A. Sheinis, A. Norton, J. Daly, S. Beaven, and J. Weinheimer, “Snapshot hyperspectral imaging: the hyperpixel array camera,” in Algorithms and Technologies for Multispectral, Hyperspectral, and Ultraspectral Imagery XV, vol. 7334 (SPIE, 2009), pp. 164–174.
14. A. Toulouse, J. Drozella, S. Thiele, H. Giessen, and A. Herkommer, “3D-printed miniature spectrometer for the visible range with a 100× 100 μm 2 footprint,” Light: Adv. Manuf. 2(1), 20–30 (2021). [CrossRef]
15. A. Yan, W. Zhenye, Z. Tao, D. Keyan, and L. Xinhang, “Development status and aberration overview of micro spectrometer with Czerny-Turner structure,” in 2016 IEEE Optoelectronics Global Conference (OGC), (IEEE, 2016), pp. 1–3.
16. P. Mouroulis and R. O. Green, “Review of high fidelity imaging spectrometer design for remote sensing,” Opt. Eng. 57(04), 1 (2018). [CrossRef]
17. E. R. Deutsch, D. Reyes, E. R. Schildkraut, and J. Kim, “High-resolution miniature FTIR spectrometer enabled by a large linear travel MEMS pop-up mirror,” in Next-Generation Spectroscopic Technologies II, vol. 7319 (SPIE, 2009), pp. 157–164.
18. Y. Warashina, T. Suzuki, K. Kasamori, R. Okumura, Y. Matsuo, and M. Takemura, “MEMS based miniature FT-IR engine with built-in photodetector,” in MOEMS and Miniaturized Systems XIII, vol. 8977 (SPIE, 2014), pp. 93–102.
19. D. Khalil, H. Omran, M. Medhat, and B. Saadany, “Miniaturized tunable integrated Mach-Zehnder MEMS interferometer for spectrometer applications,” in MOEMS and Miniaturized Systems IX, vol. 7594 (SPIE, 2010), pp. 251–263.
20. M. Ebermann, N. Neumann, K. Hiller, M. Seifert, M. Meinig, and S. Kurth, “Tunable MEMS Fabry-Pérot filters for infrared microspectrometers: a review,” in MOEMS and miniaturized systems XV, vol. 9760 (SPIE, 2016), pp. 64–83.
21. K. D. Hakkel, M. Petruzzella, F. Ou, A. van Klinken, F. Pagliano, T. Liu, R. P. van Veldhoven, and A. Fiore, “Integrated near-infrared spectral sensing,” Nat. Commun. 13(1), 103 (2022). [CrossRef]
22. Z. Yang, T. Albrow-Owen, W. Cai, and T. Hasan, “Miniaturization of optical spectrometers,” Science 371(6528), eabe0722 (2021). [CrossRef]
23. A. Piegari, A. K. Sytchkova, J. Bulir, B. Harnisch, and A. Wuttig, “Thin-film filters for a high resolution miniaturized spectrometer,” in Advances in Optical Thin Films III, vol. 7101 (SPIE, 2008), pp. 397–404.
24. N. P. Ayerden and R. F. Wolffenbuttel, “How accurate is the fabry-perot approximation in high-finesse linear variable optical filters for gas absorption spectroscopy?” in 2017 19th International Conference on Solid-State Sensors, Actuators and Microsystems (TRANSDUCERS), (IEEE, 2017), pp. 2111–2114.
25. N. P. Ayerden and R. F. Wolffenbuttel, “The miniaturization of an optical absorption spectrometer for smart sensing of natural gas,” IEEE Trans. Ind. Electron. 64(12), 9666–9674 (2017). [CrossRef]
26. A. Emadi, S. Grabarnik, H. Wu, G. de Graaf, and R. Wolffenbuttel, “Spectral measurement with a linear variable filter using a LMS algorithm,” Procedia Eng. 5, 504–507 (2010). [CrossRef]
27. S. C. K. Goh, L. L. Shiau, M. S. B. Mohamad, C. Lee, and C. S. Tan, “Highly compact linear variable filter in the mid infrared region for acetone level monitoring,” IEEE Sens. J. 20(8), 4171–4178 (2020). [CrossRef]
28. S. Zhang, W. Bin, B. Xu, X. Zheng, B. Chen, X. Lv, H. San, and W. Hofmann, “Mixed-gas CH4/CO2/CO detection based on linear variable optical filter and thermopile detector array,” Nanoscale Res. Lett. 14(1), 348 (2019). [CrossRef]
29. B. Wiesent, D. Dorigo, and A. Koch, “3.3-A miniaturized MID-IR-Spectrometer based on a linear variable filter and pyroelectric line array–Monitoring oil condition,” Proceedings IRS 2013 pp. 59–64 (2013).
30. R. C. Jones, “Quantum efficiency of photoconductors,” in Proc. IRIS, vol. 2 (1957).
31. R. Pini, “Resolving resolution,” https://spie.org/news/resolving-resolution?SSO=1. Accessed: 2023-08-10.
32. “HgCdTe (MCT) 32-Channel IR Detection Module,” https://vigophotonics.com/product/32em-5-01/. Accessed: 2023-08-10.
33. P.-S. Wei, H.-H. Chiu, Y.-C. Hsieh, D.-L. Yen, C. Lee, Y.-C. Tsai, and T.-C. Ting, “Absorption coefficient of water vapor across atmospheric troposphere layer,” Heliyon 5(1), e01145 (2019). [CrossRef]
34. “LVF-2.5-5.0-3.5-15-0.5-2%,” https://vortexopticalcoatings.co.uk/product/lvf-2-5-5-0-3-5-15-0-5-2/. Accessed: 2023-08-10.
35. “PerkinElmer Frontier FT-IR, NIR and FIR Spectroscopy Brochure,” https://resources.perkinelmer.com/lab-solutions/resources/docs/bro_frontierftir.pdf. Accessed: 2022-08-10.
36. “INFRASOLID - Thermal Infrared Emitters - HIS2000R-0WC,” https://www.infrasolid.com/en/product/his2000r-0wc. Accessed: 2023-08-10.
37. “Thorlabs MPD254254-90-P01 - Ø1“ 90° Off-Axis Parabolic Mirror, Prot. Silver, RFL = 2”,” https://www.thorlabs.com/thorproduct.cfm?partnumber=MPD254254-90-P01. Accessed: 2023-08-10.
38. “J5195RM-E - Ø1“ Mounted Plano-Convex CaF2 Cylindrical Lens, f = 50.0 mm, ARC: 2 - 5 µm,” https://www.thorlabs.com/thorproduct.cfm?partnumber=LJ5195RM-E. Accessed: 2023-08-10.
39. “USB-4716 200 kS/s, 16-bit, 16-ch Multifunction USB Module,” https://www.thorlabs.com/thorproduct.cfm?partnumber=LJ5195RM-E. Accessed: 2023-08-10.
40. J. Jureńczyk, Ł. Kubiszyn, K. Michalczewski, and J. Piotrowski, “Commercialization readiness of HOT LWIR detectors based on InAs/InAs1-xSbx T2SL at VIGO System SA,” in Physics and Simulation of Optoelectronic Devices XXIX, vol. 11680 (SPIE, 2021), pp. 202–210.
41. “Photonic Integrated Circuits technologies for MIDIR (MIRPIC),” https://vigophotonics.com/about-us/rd-projects/techmatstrateg/. Accessed: 2023-08-10.
