**1. Introduction**

Over the last decade, the public interest in air pollution measurement has gradually increased, giving rise to high demand for air quality sensors [1,2]. There are various different approaches to classify the air quality based on the concentration measurement of air pollutants including either a specific or a mix of volatile organic compounds (VOC), total suspended particles (TSP), or relative humidity (RH) [3–6]. Moreover, the concentration of carbon dioxide (CO2) has been shown to be one of the most applicable indicators for indoor air quality. CO2 is a major indoor pollutant which, even at slightly elevated gas concentrations, leads to declining work performance and diminishing focus capacity in the human organism [7]. Monitoring CO2 levels indoors is, therefore, essential to achieve an optimal balance between maximizing human performance and minimizing the need for energy-expensive ventilation of indoor spaces. In automotive environments, CO2 can reach increased concentration levels with undesirable physiological effects in a very short period of time, due to exhaust gases entering the vehicle cabin or due to metabolically induced CO2 emissions by the passengers. Typical adverse effects are fatigue, drowsiness, and lethargy which can lead to a higher risk of traffic accidents [8].

Although the CO2 concentration limits inducing adverse effects in humans are not well defined, the standardly used value dates back to Max von Pettenkofer in 1858, who defined

**Citation:** Gassner, S.; Schaller, R.; Eberl, M.; von Koblinski, C.; Essing, S.; Ghaderi, M.; Schmitt, K.; Wöllenstein, J. Anodically Bonded Photoacoustic Transducer: An Approach towards Wafer-Level Optical Gas Sensors. *Sensors* **2022**, *22*, 685. https://doi.org/10.3390/ s22020685

Academic Editor: Krzysztof M. Abramski and Piotr Jaworski

Received: 28 December 2021 Accepted: 14 January 2022 Published: 17 January 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

a hygienic limit of 1000 ppm CO2 [9]. The consensus is that monitoring and keeping the CO2 concentration as low as possible is beneficial for indoor air quality.

To date, various different gas sensing technologies are available to determine the CO2 content in the air, including, but not limited to, non-dispersive infrared (NDIR) spectrometry, photoacoustic spectroscopy (PAS), quartz crystal microbalance (QCM) sensors, and several technologies based on chemical interactions of gas with a sensing material [10–15]. Most sensors currently available on the market for consumer electronics are based on either NDIR or PAS, as they are the only mature techniques with high selectivity to CO2, which are still cheap and durable enough to be relevant. Both are based on the absorption of electromagnetic waves with a wavelength tailored to match the specific CO2 absorption wavelengths.

NDIR sensors directly measure gas quantities by detecting slight changes in the transmittance of infrared light in an absorption path between a source and a detector. For selectivity towards a target analyte, optical filters are used. However, these filters used in NDIR sensors can only be designed for optical bandwidths rather than for the absorption spectrum of a molecule. Thus, the sensor may not only be selective towards its target gas, but also towards interfering gases, such as humidity. Furthermore, these filters are prone to temperature variations as the optical properties of such filters change with temperature. The change of intensity is described by the Beer-Lambert-Law and is strongly dependent on the concentration of absorbing molecules in the path and the traveled distance of the light. NDIR sensors typically use pyroelectric or bolometer infrared detectors, which naturally have a relatively low signal-to-noise ratio (SNR). In order to get significant signal levels, these sensors use long absorption paths achieved by multiple reflections inside the housing and narrow optical filtering. Although there were successful optimizations of reflector geometries for the design of compact and still optically efficient NDIR sensors, these reflectors are rather costly to manufacture [16].

PAS sensors, on the other hand, make use of the photoacoustic effect, discovered by Bell and Roentgen in 1881, and sense CO2 indirectly by measuring sound [17,18]. These sensors detect gas concentrations by introducing periodically modulated light of a specific wavelength into a detection chamber in which a target gas absorbs precisely that wavelength. The molecules of the target gas get excited to higher energy levels by the absorption and generate thermal energy by colliding with molecules of a carrier gas. The thermal energy change in the closed chamber is translated into a pressure change, which is sensed by a microphone. The signal level of photoacoustic sensors correlates with the concentration of CO2 inside the sensor. However, as these sensors use microphones to detect the photoacoustic signal, optimized gas port geometries need to be used in order to prevent acoustical interference such as banging doors or loud conversations. These gas inlets cause an increase in the response time of PAS CO2 sensors [19,20].

Here, we investigate a new sensing concept which combines elements of both, NDIR and PAS sensors and includes a manufacturing approach for a detector module, which offers potential for miniaturized and very selective gas measurements. In contrast to NDIR detectors, the presented photoacoustic transducer is highly selective towards CO2 as only its CO2 filling absorbs infrared light. The detectors are simple to manufacture and therefore offer great potential for fast and low-cost production of miniaturized CO2 sensors. In addition, the approach also resolves the issue of acoustic interference by concept, as the microphone is embedded within a closed cavity, without increasing the response time of the sensor.
