Smart devices enable us to take and share pictures, gather songs and sounds, and measure temperatures. We take such features for granted now, but they have triggered a dramatic change in our habits. The evolution of smart devices has mainly been driven by a generation of novel miniature sensors and electronics exhibiting higher performance at affordable prices. For example, cellphones first began to embed cameras nearly 20 years ago. The cameras offered poor picture quality, and storage capacity was limited to a few dozen photos. Today’s smartphones have revolutionized the camera function. They not only take much sharper pictures anytime and anywhere, but phones also employ their cameras for many other applications. Embedded imagers now enable us to leverage facial recognition functions, evaluate food quality, or use our smartphones as a mirror. This revolution has been driven by outstanding improvements in imaging technology that combine photonics, integrated circuits, and adaptive optics at the chip or packaging level. From the technical point of view, the key to success for embedded cameras has been a big reduction in size and cost — enabled, in part, by MEMS technology, which has allowed miniaturization, high-volume fabrication, and high performance at an affordable price for the consumer market. Sensing scent Similar trends have allowed electronic components to mimic several human senses. But one capability that we still lack is a digital equivalent for sensing odors. Detection of odors, and chemicals in general, is a very challenging task. Sensing the fragrance of a perfume requires evaluating the composition of complex molecule patterns, while monitoring air quality requires tracking concentrations of harmful chemicals down to mere traces. Few existing solutions for chemical sensing fulfill these requirements well. Some, such as gas chromatography, perform well but are costly and often require bulky housing. Others can be more compact and affordable, but show significant drift and poor selectivity, limiting their effectiveness. Optical sensing and spectroscopy techniques stand out for their high sensitivity and intrinsic selectivity due to the unique spectral absorption patterns of many molecules of interest. With the commercial emergence of solid-state, quantum cascade lasers emitting in the 3- to 12-µm range, spectroscopic sensing has been increasingly used in many applications, including industrial control, emission monitoring, and biomedical analyses. Within the category of optical methods, photoacoustic (PA) detection has become a powerful technique to study concentrations of gases down to the parts-per-billion level or lower. First discovered by Alexander Graham Bell late in the 19th century, this spectroscopic technique relies on the absorption of electromagnetic radiation by molecules in a gaseous, liquid, or solid state. The adsorbed light is then converted into kinetic energy via a nonradiative relaxation process, leading to a local increase in pressure and temperature. By modulating the light source at an acoustic frequency, these periodic variations of pressure and temperature can be detected as a sound by a sensitive microphone. PA detectors still rely on the same principles as Bell’s original apparatus. But to increase sensitivity, current sensors integrate an acoustic resonator generally associated with a highly monochromatic, collimated laser whose emission wavelength matches the absorption band of the molecule of interest. With this design, the output signal is proportional to the concentration of molecules and to the power of the source, and inversely proportional to the volume of the acoustic resonator1. Further, this configuration offers the potential for miniaturization: When the overall volume of the acoustic resonator scales down from several cubic centimeters to a few cubic millimeters, the acoustic signal increases accordingly. A mock-up of an on-chip photoacoustic sensor assembling an array of quantum cascade lasers, a mid-infrared circuit, and a photoacoustic detector, all fabricated on silicon. Courtesy of CEA-Leti. For a long time, this technology was restricted to university and R&D lab development. It was only with the advent of MEMS microphones that photoacoustic detection systems became commercially available. Among them, Gasera’s photoacoustic sensing systems stand out for remarkable performance. The sensors’ ability to achieve detection limits below the parts-per-billion threshold is enabled by a patented cantilever pressure sensor that is over 100× more sensitive than conventional capacitive MEMS microphones2. Sensors combining solid-state lasers and miniaturized PA detectors are excellent candidates for more widespread use of such technology in portable or wearable devices. However, such broader adoption requires a clear path toward volume fabrication. MEMS-based systems offer a fabrication path similar to miniaturized CMOS imagers and may also allow high performance and smaller footprints with lower manufacturing costs. Infineon leveraged MEMS technology to achieve some of these goals in a new product recently released for air quality monitoring3,4. The company’s solution may be the first MEMS sensor based on photoacoustic detection. The preliminary datasheet does not disclose the working principle, but it is reasonable to assume that the device is similar to the sensor described in a recent paper by Infineon5, which presents a nonresonant, closed PA architecture. The sensor described consists of a sealed PA cell filled with CO2, a thermal source modulated at a few hertz, and an open adsorption path. The light from the source crosses the adsorption path before entering the PA sensor. The presence of CO2 along the path modulates the signal measured by the microphone in the PA cell. The signal’s modulation amplitude is proportional to the concentration of CO2 along the path. Next-gen sensors For the past 10 years, CEA-Leti has worked on developing the main building blocks required for the next generation of chemical and biochemical sensors. Our approach consists of consolidating all of the manufacturing steps onto a platform that is compatible with silicon technology. On the detector side, we developed various types of PA cells, progressively reducing their size with each iteration and exploiting the favorable scaling with miniaturization. Our most recent micro-PA detector technology combines a fully integrated MEMS microphone and mid-infrared photonics into a single silicon chip. The specific design of the resulting PA cell increases immunity to external noise and other variations in measurement conditions. With this PA sensor, we demonstrated detection of several chemicals of interest below the parts-per-billion range6. This result stems from CEA-Leti’s proprietary technology allowing the fabrication of the MEMS’ mechanical diaphragm, nanometric piezoresistive gauges, and low-loss waveguides in an integrated package. On the source side, we transferred the fabrication of quantum cascade lasers (QCLs) to silicon substrates by bonding the III-V stack directly on the wafers7. By modifying and adapting the process flow to the processing capabilities of a 200-mm CMOS/MEMS pilot line, we increased yield to 98%, with performance comparable to QCLs fabricated on indium phosphide substrates. In parallel, we developed wafer-level testing capabilities on probe stations, allowing automatic characterization and statistical tests. Proving the full manufacturability of QCL-based sensors in a CMOS/MEMS pilot line will greatly accelerate the commercial takeoff of PA detector technology by decreasing fabrication costs and increasing reliability. Further, integrating the sensors on a common technological platform implemented on silicon substrates may allow faster penetration of these devices into a number of new markets beyond gas sensing. Meet the author Sergio Nicoletti, Ph.D., is business development manager at CEA-Leti. He joined the company in 2006 to develop mid-IR optical gas sensors. Nicoletti has since served as chief researcher and project leader, and he currently oversees development of industrial partnerships. He earned a doctorate in physics from Université Joseph Fourier in Grenoble, France; email: sergio.nicoletti@cea.fr. References 1. A Miklós et al. (2001). Application of acoustic resonators in photoacoustic trace gas analysis and metrology. Rev Sci Instrum, Vol. 72, pp. 1937-1955, www.doi.org/10.1063/1.1353198. 2. C.B. Hirschmann et al. (2013). Sub-ppb detection of formaldehyde with cantilever enhanced photoacoustic spectroscopy using quantum cascade laser source. Appl Phys B, Vol. 111, pp. 603-610, www.doi.org/10.1007/s00340-013-5379-4. 3. H. Riffi (Nov. 11, 2020). CO2 Sensor helps to reduce the risk of Covid-19 transmission indoors, www.eetimes.com/co2-sensor-helps-to-reduce-the-risk-of-covid-19-transmission-indoors. 4. Infineon Technologies AG (2020). XENSIV PAS CO2, www.infineon.com/dgdl/Infineon-eval_pasco2_sensor-datasheet-v01_00-en.pdf?fileId=5546d462758f5bd10175934ec4215c6a. 5. M. Eberl et al. (2019). Miniaturized photoacoustic CO2 gas sensors — a new approach for the automotive sector. AmE — Automotive meets Electronics, 10th GMM-Symposium, pp. 1-5, Dortmund, Germany. 6. A. Glière et al. (2020). Downsizing and silicon integration of photoacoustic gas cells. Int J Thermophys, Vol. 41, No. 16, www.doi.org/10.1007/s10765-019-2580-7. 7. J.G. Coutard et al. (2020). Volume fabrication of quantum cascade lasers on 200 mm-CMOS pilot line. Sci Rep, Vol. 10, No. 6185, www.doi.org/10.1038/s41598-020-63106-4.