# Highly stable, nanotube-enhanced CMOS-MEMS thermal emitter for mid-infrared gas sensing | Scientific Reports

2021-11-25 10:30:27 By : Ms. Kathy Huang

Thank you for visiting Nature. The browser version you are using has limited support for CSS. For the best experience, we recommend that you use a newer version of the browser (or turn off the compatibility mode in Internet Explorer). At the same time, to ensure continued support, we will display sites without styles and JavaScript.

Driven by many socio-economic and industrial factors, the gas sensor market is growing rapidly. Mid-infrared (MIR) gas sensors provide outstanding performance for a growing number of sensing applications in the healthcare, smart home, and automotive industries. Obtaining a low-cost, miniaturized, and energy-saving light source is essential for the monolithic integration of MIR sensors. Here, we show an on-chip broadband thermal mid-infrared source manufactured by combining a complementary metal oxide semiconductor (CMOS) microthermal plate with a dielectric encapsulated carbon nanotube (CNT) black body layer. The micro-hot plate is used as a micro-reactor in the manufacturing process to promote the high temperature (>700 $$^{\circ}$$C) growth of the CNT layer, and is also used for thermal annealing after growth. We have demonstrated for the first time the stable extended operation of a device with a dielectric encapsulated CNT layer in the air at a heater temperature higher than 600 $$^{\circ }$$ C. The displayed device exhibits almost uniform emissivity in the entire MIR spectrum, providing an ideal solution for the low-cost, highly integrated MIR spectrum of the Internet of Things.

Driven by many scientific, industrial and commercial applications, gas sensors are at the center of ever-increasing research and development efforts. This includes monitoring environmental pollutants from deforestation2, vehicles and industry3, and air quality in buildings4. People are becoming more aware of the impact of air pollution on human health3, leading to increased demand for low-cost, accessible, compact, and easy-to-deploy air quality monitoring5. In order to maintain emerging global demand, gas sensors must find a suitable and challenging balance between performance and cost1. In addition to being economically feasible, more and more sensors have strict power and size constraints1, such as those deployed in the Internet of Things (IoT)6 and mobile platforms7. These requirements have prompted researchers to explore new materials, designs, and technologies to achieve component miniaturization, monolithic integration, low cost, low power consumption, and manufacturability1.

Among various sensing technologies, optical gas sensors have many advantages in terms of selectivity and long-term operational stability1. It is worth noting that non-dispersive infrared (NDIR) sensors currently dominate the carbon dioxide (CO$$_2$$) gas sensor market and also serve many other applications8. However, although NDIR gas sensors have inherent advantages (for example, for spectral sensing), they are currently mainly used to detect a single analyte or several substances at the same time. One limitation of wider adoption is the availability of low-cost and optically efficient small broadband mid-infrared light sources (arguably the core of optical gas sensors)1. Traditionally, light bulb-based heat sources have been used, but they are fragile, bulky, and limited in optical efficiency at wavelengths above 5 $$\upmu$$ m. Light-emitting diodes (LEDs) provide higher integration and reliability, but due to the use of professional III-V semiconductor technology, the manufacturing cost is higher9.

Using standard complementary metal oxide semiconductor (CMOS) processes is an attractive way to manufacture low-cost integrated thermal mid-infrared sources and detectors, and has led to many innovative devices based on microelectromechanical systems (MEMS) 1,10. Various techniques have been proposed to improve the emissivity/absorption rate of CMOS-MEMS thermal devices1, including the use of a carbon nanotube (CNT) adsorption layer to provide near-unit broadband emissivity 11, 12, or plasma for specific mid-infrared bands Body metamaterials 13,14. Multi-species spectroscopy detection requires the MIR source to operate under the set of target MIR bands, so that the overall CNT broadband emission enhancement 11,12 is attractive for spectroscopy. However, despite their blackbody-like advantages15,16, so far, most studies have observed such CNTs, and generally all graphite nanocarbon adsorption layers burn out in the air when operating at temperatures above 400 ℃ $$^{\circ }$$ C17,18. This imposes limitations (light emission and operational stability) for its integration into CMOS MEMS micro hot plate MIR sources (usually operating at these temperatures). Although inert gas can be used to prevent CNTs from burning out, this requires the use of special sealed ceramic or metal packaging, which will significantly affect the cost20,21.

Here, we propose a solid-state method based on a dielectric encapsulation method that can achieve long-term operational stability of CNT-coated thermal emitters. We show that aluminum oxide (Al$$_2$$ O$$_3$$) encapsulated carbon nanotubes grown on MEMS micro hot plates can withstand more than 800 when operating in air. $$^{\circ }$$ The temperature of C. The packaged CNT adlayer has a nearly uniform emissivity (reference standard MEMS device, $$\sim$$ increased by 8 times), and stable operation for 10 days under 600 $$^{\circ }$$ C. This work It paves the way for packaging technology to be more widely used in temperature and air-sensitive nanomaterials, enabling them to operate stably in the air, well above their normal temperature threshold.

Equipment manufacturing. (a) Cross section of micro hot plate using tungsten (W) heating element embedded in $$\sim$$ 5 $$\upmu$$ m thick silicon dioxide (SiO$$_2$$) (not to scale) Film formed by deep reactive ion etching. (b) The optical image of the micro hot plate, showing the heating element of the multi-ring design surrounded by the film. Chip size = 1.76 mm $$\times$$ 1.76 mm. (c) Micro heater temperature as a function of power consumption. (d) Active heating elements have three purposes: (i) Local in-situ heating during CNT growth (microreactor); (ii) Used as a thermal analyzer for self-annealing (adlayer flashing); (iii) Used in IR emission is generated during equipment operation (usually at a temperature of 300–600 $$^{\circ }$$ C). (e) Typical SEM image of spaghetti-like CNT structure grown by in-situ CVD process using acetylene (C$$_2$$ H$$_2$$) mixed with ammonia (NH$$_3$$) on iron (Fe) catalyst, and (f) SEM image of $$\sim$$ 50 nm alumina (Al$$_2$$ O$$_3$$) after packaging. (e) SEM images of denser and more aligned CNTs synthesized by mixing C$$_2$$ H$$_2$$ and hydrogen (H$$_2$$ ), and (h) SEM images after packaging .

For our experiment, we use an in-house designed micro electric furnace manufactured in a commercial foundry. The cross section of the micro hot plate is shown in Figure 1a. It consists of a silicon dioxide (SiO$$_2$$) film (1200 $$\upmu$$ embedded in $$\sim$$ 5 $$\upmu$$ m thick m diameter) to ensure low direct current (DC) power consumption19. Compared with doped polysilicon or aluminum, W was chosen as the heating element and interconnect metal because of its excellent electromigration resistance and higher glass transition temperature10,19. The micro heating plate adopts CMOS-SOI (Silicon-on-Insulator) technology, with a monolithic integrated monocrystalline silicon thermal diode, which can operate linearly with high precision at a recording temperature of up to 600 $$^{\circ }$$ C22. When the micro heating plate is used as an IR transmitter, the thermal diode can be used as an accurate temperature sensor, and the resolution is allowed to be less than 0.5 $$^{\circ }$$ C during calibration. For CNT growth and operation above 600 $$^{\circ }$$ However, it is better to use W heater as resistance temperature detector (RTD). W has a large and stable temperature coefficient of resistance (TCR) ($$\sim$$ 4.5 $$\times$$ 10$$^(-3)$$ K$$^(-1)$$) and has been Prove that it can be run with relatively high accuracy under 1000 $$^{\circ }$$ C $$\sim$$ 2 $$^{\circ }$$ C19. The micro hot plate can reach a temperature of more than 700 $$^(\circ )$$ C and has a fast thermal transient time> 4 $$\times$$ 10$$^4$$ $$^{\circ }\ ) C/s, to achieve voltage-controlled thermal ramp and stable MIR emission at a very low cost, with excellent reproducibility10,19. The optical image of the micro heating plate we made shows the heating element surrounded by the film, as shown in Figure 1b. We use in-situ chemical vapor deposition (CVD) process to integrate CNT adlayer11,12. The CVD process usually requires a substrate temperature of more than 400 \(^{\circ }$$ C23,24, which is incompatible with CMOS because temperature effects can damage integrated circuits (for example, due to accelerated alloying and atom migration)10. Due to the thermal isolation provided by the thin dielectric film, our micro hot plate can easily reach a temperature of more than 750 $$^{\circ }$$ C (Figure 1c) in the local "hot zone" without affecting the functional chip substrate The peripheral CMOS circuit on the board is only a few microns away from the hot zone. Therefore, our micro hot plate can be effectively used as an ideal CMOS compatible microreactor, allowing viable CNT-CMOS integration at the wafer level12. In addition, our design allows relatively low DC power consumption to achieve such a high temperature (for example, $$\sim$$ 100 mW at 500 $$^{\circ }$$ C, as shown in Figure 1c), which It can be further minimized by using a modulated drive signal (for example, 50% duty cycle in our case). In order to test the packaging efficiency, we used two commonly used process gases, ammonia (NH$$_3$$ )12 and hydrogen (H$$_2$$ )11, which are the same as the carbon-containing gas [acetylene (C$$_2$$ H$$_2$$ )], in the iron (Fe) catalyst process. As reported elsewhere in the CNT synthesis literature, the use of NH$$_3$$ tends to produce more spaghetti-like nanotube bundles 12, while the use of H$$_2$$ produces more spatially dense, vertical arrangements. The nanotube 11 thus provides us with a comprehensive test platform for our experiments (the influence of CNT crystallography and morphology). NH$$_3$$ :C$$_2$$ H$$_2$$ The optical image of the synthesized CNT sample is shown in Figure 1d. Scanning electron microscope (SEM) examination confirmed the successful growth of spaghetti-like (Figure 1e) and vertically aligned (Figure 1g) nanotube forests.

Characterization and stability testing in the air. (a) The infrared absorption spectrum of the uncoated micro heating plate (black line) compared to the CNT coated plate (red line) that absorbs almost all light. (b) Measure the emission spectra of the uncoated (black line), CNT-coated (blue line) and encapsulated (red line) micro hot plates at 600 $$^{\circ }$$ C. Compared with uncoated devices at the same temperature, CNT-coated devices show a $$\sim$$ 8-fold increase in emission. However, unencapsulated CNTs will burn within a few minutes, causing their The emission spectrum (blue line) drops from the initial value close to the packaged device (red line) to the uncoated device (black line). Optical inspection (inset) showed that the carbon nanotubes were almost completely burned. (c) The emission spectrum measured at 600 $$^{\circ }$$ C in 10 days, for the device packaged with NH$$_3$$: C$$_2$$ H$$_2$$ -synthetic CNT (as shown in Figure 1f). The optical images recorded at various times show that the packaged CNTs are not affected by high-temperature operation. After running at 850 $$^{\circ }$$ C for 4 hours at a higher temperature, the optical image of the test sample is displayed in the lower right corner, showing that the CNT is intact. (D) (bottom two lines) Raman spectra measured at 532 nm before and after the operational stability test shown in (c). For comparison, the two lines at the top show the same test (not shown) measurement performed on the CNT synthesized using H$$_2$$ :C$$_2$$ H$$_2$$ (shown in Figure 2) Raman spectroscopy. 1h) In contrast, it shows that both samples remain stable.

As we all know, carbon nanotubes burn in air at a temperature higher than 400 $$^{\circ }$$ C (usually within a few minutes), depending on their diameter, number of walls or number of defects17,18, Therefore, their application as thermal emitters is hindered. In order to isolate the CNTs exposed to the air from oxidation 17, 18 and create a thermally stable emitter interface, we encapsulated the grown CNTs in $$\sim$$ 50 nm Thick deposited atomic layer (ALD) Al$$_2$$ O$$_3$$ coating. The package based on Al$$_2$$ O$$_3$$ guarantees high thermomechanical stability, and has been proven to have the ability to operate at temperatures higher than 500 $$^{\circ }$$ C, and at the same time proved Good oxygen (O$$_2$$) and water (H$$_2$$ O) 25,26 barriers. Figure 1f,h shows the SEM images of CNT (spaghetti-like NH$$_3$$ :C$$_2$$ H$$_2$$ synthesis) and g (aligned) shown in Figure 1e after packaging. (_2\) :C$$_2$$ H$$_2$$-combine separately). In order to improve the thermal stability after packaging, we used the equipment self-annealing 26,27 process, in our case, it is activated by the micro heater itself. The thermal distribution is applied to the electric heating modulation in step C of $$\sim$$ 100 $$^{\circ }$$, and the temperature is as high as $$\sim$$ 800 $$^{\circ }$$ C; this makes the film The induced thermo-mechanical stress on the surface is consistent with the operating frequency, thereby avoiding membrane rupture.

The typical light absorption spectrum of our micro hot plate (see "Methods"), in the 2-14 $$\upmu$$ m band, as shown in Figure 2a (black line), the absorption peak is $$\ sim$$ 85% The characteristics of Si-O stretching vibration in the 8.5 $$\upmu$$ m; SiO$$_2$$ film28,29. In the same picture (red line), the micro heating plate with the in-situ grown NH$$_3$$ :C$$_2$$ H$$_2$$ CNT (Figure 1f) layer shows almost 100% absorption, This behavior is attributed to the black body-like nature of the CNT layer15,16. In order to study the emission characteristics of our uncoated, CNT coated and packaged devices, we implemented a temperature controller based on proportional integral derivative (PID), which can control the temperature of the self-heating micro hot plate at $$\sim$$ 0.5 \ Within (^{\circ }\) C resolution. The emission spectrum of the self-heating device was then measured by an MIR spectrometer (Bentham) (Figure 2b). For the packaged (NH$$_3$$ :C$$_2$$ H$$_2$$) sample, the representative spectrum recorded at 600 $$^{\circ }$$ C is shown in Figure 2b ( The red line), compared with the uncoated device (black line) operating at the same temperature, shows a $$\sim$$ 8 times emission enhancement. We also measured the emission spectra of the unencapsulated CNT-coated device. As expected, it exhibited a rapid ($$\sim$$ minute) emission reduction at elevated temperatures, which is the same as that of the CNT when exposed to air. In the middle, it will burn out unanimously 17. The sample spectrum is shown in Figure 2b (blue line), recorded at 600 $$^{\circ }$$ C after running for 10 minutes $$\sim$$. A visual comparison between the black and blue lines indicated that most of the CNTs had been burned, which was confirmed by visual inspection under a microscope (Figure 2b, inset).

In order to study the long-term operational stability of our packaged devices, we used normal indoor conditions [standard pressure (1010–1020 mbar), temperature (18–21 $$^{\circ }$$ C) and relative humidity (30–50%) )] At a temperature as high as 900 $$^{\circ }$$ C, it is significantly higher than the actual equipment operating temperature ($$\sim$$ 500–600 $$^{\circulation}$$ C). For the typical emission spectrum of the equipment running continuously for 10 days (regularly recorded every 24 hours) under 600 $$^{\circ }$$ C as shown in Figure 2c, where $$\sim$$ 3 \ (\times\ ) 10$$^{-5}$$ Variation in standard deviation, indicating excellent stability. An optical image recorded at a representative time during the test is also provided, which shows that the encapsulated CNT is physically unaffected by the high operating temperature. When up to 900 $$^{\ circ }$$ C. The optical image of the sample is displayed in the lower right corner of Figure 2c, and it is recorded after running in the air of 850 $$^{\circ }$$ C for 4 hours, and further shows The stability of physical carbon nanotubes at high temperatures is improved. It is worth noting that at such high temperatures, some films (not carbon nanotubes) will fail due to extreme thermally induced mechanical stress. The faulty device was analyzed through optical and SEM imaging, and the identified fault was attributed to film rupture, rather than burnt-out of the packaged CNT layer, and remained intact even under extreme temperatures. Similar film rupture at extreme temperatures was also observed using uncoated devices (no CNT layer). In order to check the quality of the encapsulated CNTs before and after the 10-day operation test, we used Raman spectroscopy to characterize them. Figure 2d (bottom two lines) plots the Raman spectrum of the above device. We did not observe any changes in the position, width and intensity of the D peak, indicating that long-term high temperature operation will not cause additional defects in the starting material30. As a comparison (see the first two lines), we also show the Raman spectrum of the same test run with CNT packaged with H$$_2$$ :C$$_2$$ H$$_2$$, showing good operation Stability of two samples, thus highlighting the wider application potential of our packaging technology.

Non-dispersive infrared (NDIR) experiment. (a) Schematic diagram of NDIR gas sensor setup. (b) The relative optical response recorded by the thermopile at 4.26 $$\upmu$$ m is used for uncoated (black line), CNT coated (blue dashed line) and encapsulated (red line) micro hot plates. At temperatures higher than 600 $$^{\circ }$$ C, the detected IR emission of the unpackaged device (blue dotted line) drops rapidly (minutes), consistent with the data provided in Figure 2b. (c) CO$$_2$$ The sensor response voltage has uncoated (black wire) and encapsulated (red wire) micro-heating plates respectively. For packaged CNT devices, it can be observed that $$\sim$$ relative voltage$$\Delta V=V_{0 ppm}-V_{lock-in}$$ has increased by 8 times.

In order to test the performance of our MIR transmitter in the application type setting, a custom NDIR gas sensor designed for CO$$_2$$ detection was used to benchmark samples against uncoated equipment. The schematic diagram of our experimental setup is shown in Figure 3a. We use our thermal emitter as the MIR light source, coupled to a single-channel thermopile detector (Heimann HMS-J21), integrated 4.26 $$\upmu$$ m filter tuned to the CO$$_2$$ absorption band. Both the emitter and the detector are installed in a custom-made gas chamber, and the length of the optical path between the emitter and the detector is 4 cm. The gas chamber is connected to the gas mixing system so that the CO$$_2$$ concentration ($$c_{CO_2}$$) can pass through the cylinder of dry air at 5$$\%$$. Our customized experimental device allows the CO$$_2$$ concentration to be controlled below 100 ppm, and the minimum controlled change in concentration is a few ppm. The relative optical signal recorded by the thermopile at room temperature (see "Methods") is shown in Figure 3b. CNT (NH$$_3$$ :C$$_2$$ H$$_2$$ synthesis) The response of the coated device (blue dotted line) and the packaged device (red line) to temperature rise is similar to $$\sim\ ) 600 \(^{\circ }$$ C, but the emission level began to decrease rapidly (within a few minutes) towards a typical uncoated device, confirming the results shown in Figure 2b. After several weeks of operation, we found that the response of the CNT-encapsulated device did not change significantly (Figure 3b, red line). Figure 3c shows the voltage response of a sensor with a encapsulated CNT device (red wire). Compared with an uncoated device (black wire), $$c_{CO_2}$$ ranges from 0 to $$\sim$$ 21,500 ppm . The response of the packaged CNT device at the maximum CO$$_2$$ concentration is $$\Delta V=V_{0 ppm}-V_{lock-in}$$ = 111 mV, while the response of the uncoated device is only 22 mV. The standard deviation of CO$$_2$$ measurement is $$\sim$$ 0.6 mV, which is much smaller than 200-300 mV after the amplitude recovery signal is amplified, mainly because the signal-to-noise ratio (SNR) is limited by thermal noise. Considering the relative sensitivity defined as $$S=\frac{d\Delta V}{dc_{CO_2}}|_{c_{CO_2}=0}$$ and $$\sim$$ 3 dB to measure SNR, we estimate The detection limit of $$\sim$$ 0.12 ppm for the packaged device is compared with the detection limit of $$\sim$$ 1 ppm for the uncoated device, which is consistent with the results in Figure 2b.

In summary, we show a highly efficient CMOS compatible CNT packaged micro hot plate-based MIR emitter, which is manufactured on a single SiO$$_2$$ dielectric film, with a near uniform emissivity and long-term stable operation sex. By adopting a standard CMOS tungsten heater as a microreactor for precise in-situ CNT growth, the processing of the chip is simplified, allowing scalable integration at the wave level. The same heating element can also be used as a contributing factor for the multi-functional self-annealing thermal curve, which can be easily optimized for various processes. We demonstrated the stable operation of alumina encapsulated CNTs at a record temperature of up to 900 $$^{\circ }$$ C. Using the transmitter in the proof-of-concept optical sensing demonstration, we measured that the relative sensitivity has increased by 8 times compared to using the traditional MEMS transmitter, CO$$_2$$. Our transmitter exhibits almost uniform emissivity across the entire MIR band, making it particularly attractive for various low-cost, low-power and large-capacity spectral applications in the MIR spectral region.

The micro hot plate is designed in Cadence$$\copyright$$ and is manufactured on a 6-inch silicon (Si) wafer using a commercial 1$$\upmu$$ m SOI-CMOS process. The film is formed by deep reactive ion etching (DRIE) on a 400 $$\upmu$$ m thick Si substrate, where the buried SiO$$_2$$ layer serves as an etch stop layer. The silicon nitride (Si$$_3$$ N$$_4$$) passivation layer protects the film from environmental factors, such as humidity 31.

Carbon nanotubes are grown on Fe catalyst by in-situ thermal CVD of C$$_2$$ H$$_2$$. Coat (ALD) the micro hot plate with $$\sim$$ 10 nm Al$$_2$$ O$$_3$$, and then sputter with 2-4 nm Fe catalyst. Then these devices [installed on a TO-type package connected to a power supply (Keithley 2400)], and then transferred to a custom CVD chamber to grow CNTs, and then evacuated to a base pressure of $$\sim$$ 0.5 millibar. The CNT growth process is optimized by $$\sim$$ 0.5 $$^{\circ }$$ C resolution based on PID temperature controller, realized in LabVIEW© software, set to 20 $$^{\circ }$$ C/s heating rate. When the micro heater reaches $$\sim$$ 500 $$^{\circ }$$ C, introduce high purity NH$$_3$$ or H$$_2$$ into the chamber, and then at 725 $$^ {\circ }$$ C continues for 60 seconds to form small catalyst Fe islands. Then introduce C$$_2$$ H$$_2$$ into 5% H$$_2$$ through a separate pipeline: C$$_2$$ H$$_2$$ or 25% NH$$_3$$) :C$$_2$$ H$$_2$$ The atmosphere was maintained at $$\sim$$ 4 mbar during the growth of $$\sim$$ for 10 minutes. Then the device was introduced into the ALD reactor (Cambridge NanoTech) to deposit $$\sim$$ 50 nm Al$$_2$$ O$$_3$$ [using trimethyl aluminum (TMA) and water (H$$_2$$ ) O) as a precursor is encapsulated in 200 $$^{\circ }$$ C], and then packaged in $$\sim$$ 400, 500, 600 and 700 $$^{\circ }$$ C, respectively 30 minutes.

In order to obtain the optical absorption (A) spectrum distribution diagram, in the wavelength range of 2-14 $$\upmu$$ m, the transmission (T) and reflection (R) FTIR measurements (normal incidence) are coupled to obtain A = 1- Rebroadcast. The optical aperture of the micro FTIR system (Agilent Cary 620 FTIR microscope) is set to image only the heater area of ​​the micro hot plate. The emission spectrum curve was measured by installing the equipment to a custom MIR spectrometer (Bentham) consisting of a monochromator (TMc300) connected to a cryogenically cooled mercury cadmium telluride detector (DH-MTC). Raman spectra were obtained by Renishaw's inVia Raman microscope under 532 nm excitation.

The NDIR sensor is connected to a National Instruments DAQ card (NI USB-6353) to allow automatic control and data acquisition through LabVIEW© software. The micro heater uses a custom amplifier for voltage modulation with a 2 Hz periodic square wave. A custom preamplifier (60 dB voltage gain) and a software-based lock-in amplifier (1 s integration time; $$\sim$$ 50 dB SNR) are used to recover the signal detected by the thermopile from the background noise. A complete 16-bit A/D conversion is used for measurement. The total flow rate used for CO$$_2$$ sensing is 200 sccm, which is realized by a combination of a computer-controlled mass flow controller (MKS).

Popa, D. and Udrea, F. are committed to integrating mid-infrared gas sensors. Sensor 19, 2076. https://doi.org/10.3390/s19092076 (2019).

Seymour, F. & Harris, NL Reduce tropical deforestation. Science 365, 756–757. https://doi.org/10.1126/science.aax8546 (2019).

Landrigan, PJ et al. The Lancet Pollution and Health Committee. The Lancet 391, 462–512. https://doi.org/10.1016/S0140-6736(17)32345-0 (2018).

Barnes, JH, Chatterton, TJ & Longhurst, JW Emissions and Exposure: Increasing injustices caused by air pollution related to road traffic in the UK. Part D of the transshipment reservoir: Transp. environment. 73, 56-66. https://doi.org/10.1016/j.trd.2019.05.012 (2019).

Morawska, L. et al. Application of low-cost sensing technologies in air quality monitoring and exposure assessment: how far have they gone? environment. internationality. 116, 286–299. https://doi.org/10.1016/j.envint.2018.04.018 (2018).

CAS Article PubMed PubMed Central Google Scholar

Cerchecci, M. etc. Low-power IoT sensor node architecture for waste management in smart city environments. Sensor 18, 1282. https://doi.org/10.3390/s18041282 (2018).

Sun, J. etc. Suspended AlGaN/GaN HEMT NO$$_2$$ gas sensor integrated with micro heater. J. Microelectromechanical. system. 28, 997–1004. https://doi.org/10.1109/JMEMS.2019.2943403 (2019).

Non-dispersive infrared (NDIR) market. technology. Representative, MarketsandMarkets (2020).

Fujita, H. et al. High luminous efficiency AlInSb mid-infrared LED for gas sensors. Physical statistics. Solidi (A) 215, 1700449. https://doi.org/10.1002/pssa.201700449 (2018).

Udrea, F. & Luca, AD CMOS technology platform for ubiquitous micro-sensors. 2017 International Conference on Semiconductors (CAS), 43-52. https://doi.org/10.1109/SMICND.2017.8101149 (2017).

DeLuca, A. et al. An enhanced spectral gas sensor using carbon nanotubes grown in situ. App Physics Wright. 106, 194101. https://doi.org/10.1063/1.4921170 (2015).

Santra, S. etc. Post-CMOS wafer-level growth of carbon nanotubes for low-cost microsensors-proof of concept. Nanotechnology 21, 485301. https://doi.org/10.1088/0957-4484/21/48/485301 (2010).

Pusch, A. etc. A high-efficiency CMOS nano-plasmonic crystal enhanced slow-wave thermal emitter improves infrared gas sensing equipment. science. Rep. 5, 17451. https://doi.org/10.1038/srep17451 (2015).

Lochbaum, A. etc. On-chip narrow-band heat emitter for mid-infrared optical gas sensing. ACS Photon. 4, 1371–1380. https://doi.org/10.1021/acsphotonics.6b01025 (2017).

Kaul, AB, Coles, JB, Eastwood, M., Green, RO & Bandaru, PR uses multi-walled carbon nanotubes to collect ultra-high light absorption efficiency from ultraviolet to infrared. Small 9, 1058-1065. https://doi.org/10.1002/smll.201202232 (2013).

Mizuno, K. et al. A black body absorber made of vertically aligned single-walled carbon nanotubes. The process of the national team Akkad. science. 106, 6044–6047. https://doi.org/10.1073/pnas.0900155106 (2009).

Bom, D. etc. Thermogravimetric analysis of oxidation of multi-walled carbon nanotubes: Evidence for the role of defect sites in carbon nanotube chemistry. Nanolet. 2, 615–619. https://doi.org/10.1021/nl020297u (2002).

Singh, DK, Iyer, PK & Giri, PK Diameter dependence of oxidation stability in multi-walled carbon nanotubes: the effect of defects and the effect of vacuum annealing. J. Application Physics 108, 084313. https://doi.org/10.1063/1.3491022 (2010).

Ali, SZ etc. Low-power, low-cost infrared transmitter using CMOS technology. IEEE Sens. J. 15, 6775–6782. https://doi.org/10.1109/JSEN.2015.2464693 (2015).

Reichl, H. & Grosser, V. Overview and development trends in the field of MEMS packaging. In the technical digest. MEMS 2001. The 14th IEEE International Conference on Micro-Electro-Mechanical Systems (catalog number: 01CH37090), 1-5. https://doi.org/10.1109/MEMSYS.2001.906464 (2001).

Esashi, M. Wafer-level packaging of MEMS. J. Micromachines. Micro engineering. 18, 073001. https://doi.org/10.1088/0960-1317/18/7/073001 (2008).

Santra, S., Guha, PK, Ali, SZ, Haneef, I. & Udrea, F. Silicon-on-insulator diode temperature sensor-detailed analysis of ultra-high temperature operation. IEEE Sens. J. 10, 997–1003. https://doi.org/10.1109/JSEN.2009.2037822 (2010).

Terranova, M., Sessa, V. and Rossi, M. The world of carbon nanotubes: an overview of CVD growth methods. Chemical vapor. Deposit 12, 315–325. https://doi.org/10.1002/cvde.200600030 (2006).

Wirth, CT, Zhang, C., Zhong, G., Hofmann, S. & Robertson, J. The diffusion and reaction of carbon nanotube forests restrict growth. ACS Nano 3, 3560–3566. https://doi.org/10.1021/nn900613e (2009).

Nasiri, A. & Ang, S. The application of alumina-based ceramic slurry in high-temperature electronic packaging. J. Electronics. Package. 143, 20902. https://doi.org/10.1115/1.4049292 (2021).

Golim, OP, etc. Synthesis of alumina ceramic package for thermal barrier coating self-repairing material. J. Phys: Conf. Sir. 985, 012036. https://doi.org/10.1088/1742-6596/985/1/012036 (2018).

Musil, J. et al. Thermal stability of alumina film containing $$\gamma$$-Al$$_2$$O$$_3$$ phase prepared by reactive magnetron sputtering. Application surfing. science. 257, 1058–1062. https://doi.org/10.1016/j.apsusc.2010.07.107 (2010).

Arun Kumar, D., Merline Shyla, J. & Xavier, FP Synthesis and characterization of TiO$$_2$$ /SiO$$_2$$ nanocomposites for solar cell applications. Applications Nanoscience. 2, 429–436. https://doi.org/10.1007/s13204-012-0060-5 (2012).

Popa, D., Ali, SZ, Hopper, R., Dai, Y. & Udrea, F. Smart CMOS mid-infrared sensor array. choose. Wright. 44, 4111. https://doi.org/10.1364/OL.44.004111 (2019).

Dresselhaus, M., Dresselhaus, G., Saito, R. & Jorio, A. Raman spectra of carbon nanotubes. Physics Rep. 409, 47-99. https://doi.org/10.1016/j.physrep.2004.10.006 (2005).

Brand, O. & Fedder, GK (eds) CMOS-MEMS (aaWILEY-VCH, 2005).

We thank EPSRC (EP/S031847/1, EP/S030247/1, EP/P005152/1) for funding. V.-PV-R. Recognizing the EPSRC Doctoral Training Award (EP/M508007/1) and the support of the NPL, JAW recognizes the support of his Dorothy Hodgkin Research Fellowship from the Royal Society.

Department of Engineering, University of Cambridge, Cambridge, CB3 0FA, UK

Daniel Popa, Richard Hopper, Ye Fan, Vlad-Petru Veigang-Radulescu, Jack Alexander-Webber, Stephan Hofmann and Florin Udrea

Flusso Limited, Cambridge, CB4 0DL, UK

Syed Zeeshan Ali & Andrea De Luca

Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, UK

Department of Physics, University of Cambridge, Cambridge, CB3 0HE, UK

School of Biomedical Physics, Physics and Astronomy, University of Exeter, Exeter, EX4 4QL, UK

Faculty of Engineering, University of Warwick, Coventry, CV4 7AL, UK

Xing Yuxin & Julian William Gardner

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

DP and FU conceived this idea. DP and RH designed the project. DP, RH, SZA and MTC manufactured these devices. DP, RH, YF, V.-PV-R., RC, and JN characterize these devices. DP and RH tested these devices. DP, RH, YX, JWG and FU analyzed the results. DP supervised the work and wrote the manuscript. All authors reviewed the manuscript.

The author declares no competing interests.

Springer Nature remains neutral on the jurisdiction claims in the published maps and agency affiliates.

Popa, D., Hopper, R., Ali, SZ, etc. A highly stable nanotube-enhanced CMOS-MEMS thermal transmitter for mid-infrared gas sensing. Scientific Report 11, 22915 (2021). https://doi.org/10.1038/s41598-021-02121-5

DOI: https://doi.org/10.1038/s41598-021-02121-5