In-vehicle wireless driver breath alcohol detection system using a microheater integrated gas sensor based on Sn-doped CuO nanostructures

In this paper, we have developed an in-vehicle wireless driver breath alcohol detection (IDBAD) system based on Sn-doped CuO nanostructures. When the proposed system detects the ethanol trace in the driver`s exhaled breath, it can alarm and then prevents the car to be started and also sends the location of the car to the mobile phone. The sensor used in this system is a two-sided micro-heater integrated resistive ethanol gas sensor fabricated based on Sn-doped CuO nanostructures. Pristine and Sn-doped CuO nanostructures were synthesized as the sensing materials. The micro-heater is calibrated to provide the desired temperature by applying voltage. The results showed that by Sn-doping in CuO nanostructures, the sensor performance can be significantly improved. The proposed gas sensor has a fast response, good repeatability along with good selectivity that makes it suitable for being used in practical applications such as the proposed system.


Results and discussion
Characterizations of sensing materials. Figure  www.nature.com/scientificreports/ Sn-doped sample, no additional peaks related to other phases was observed, demonstrating the incorporation of Sn into CuO lattice 54,55 . EDS elemental analysis results for pristine and Sn-doped CuO nanostructures are presented in Fig. 3b and c, respectively. For pristine sample, the peaks related to Cu and O elements were detected. However, after Sn-doping an additional peak related to Sn was observed, confirming doping of Sn into CuO lattice.
Fourier-transform infrared spectroscopy (FTIR) analysis results for pristine and Sn-doped CuO nanostructures are presented in Fig. 3d Raman spectra of pristine and Sn-doped CuO nanostructures are presented in Fig. 3e and f, respectively. Both Raman spectra exhibited two distinct peaks at around 294.7 and 630 cm −1 , which are respectively assigned to the A g and B g vibration modes of CuO 57 .    www.nature.com/scientificreports/ Gas sensing measurement. The static method was used to measure the response of the gas sensors. The microsyringe injected a desirable amount of target gas into the gas chamber, which was mixed with the air. For liquids such as VOCs, the liquid was heated on a heating plate until completely evaporated and mixed. Following formula was used to calculate the concentration of VOCs in gas chamber: where V (μL) is the volume of VOC, V (L) is the volume of the test chamber, C (ppm) is the concentration of gas, M is the molar weight of VOC, Tstan (K) is the temperature of standard condition, Tsh (K) is the ambient temperature, ρ (g cm −3 ) is the density of VOC, and 22.4 is the molar volume of the standard gas. For gases such as H 2 S, NH 3 and CO 2 , dry air was used as a balance gas. Dry air-balanced target gases from cylinders and dry air without humidity were introduced into the gas chamber using mass flow controllers (MFCs). The response was defined as V g /V a where V a and V g are the voltages in air and in the presence of target gas, respectively. The response time and recovery time, were defined as the period of time needed for the sensor voltage to return to 90% of its final stable value 58 . The schematic diagram of the gas sensing procedure is shown in Fig. 4.
Gas sensing studies. Figure S1a shows the variations of the baseline resistance of pristine and Sn-doped CuO gas sensors as a function of temperature. In both cases, the resistance decreases upon increase of the temperature, showing semiconducting behavior of the gas sensors 59 . Also, at all temperatures, the baseline resistance of Sn-doped CuO gas sensor is higher than that of pristine CuO gas sensor. This increase in the resistance is due to the presence of Sn ions in the CuO lattice, which play the role of electron donor. As a result, the density of holes will decrease and then the baseline resistance will increase. Besides, adding Sn to the p-type CuO will result in formation of a depletion area and then a potential barrier, which can be another reason for increasing the baseline resistance. Good resistance stability of Sn-doped gas sensor at different temperatures is shown in Fig. S1b. It is shown that the values of the resistance of the gas sensor upon increasing and decreasing of the sensing temperature are almost the same, revealing the good resistance stability of the gas sensor. Also, Fig. S1c shows good stability of the sensor resistance at a fixed temperature for a long time of more than 1600 s. Overall, the Sn-doped gas sensor shows a stable resistance that is beneficial for sensing studies.
To determine the optimal working temperature of the gas sensors, they were exposed to 100 ppm ethanol gas at different temperatures (Fig. 5). In both sensors, initially the response increased with temperature, then reached to a maximum value and finally decreased. The reason for this behavior is that at low temperatures there is no enough energy for gas to overcome the absorption barrier energy and at high temperatures the desorption rate is higher than absorption rate. At optimal sensing temperature, the absorption rate is equal to desorption rate and the maximum response is observed. The maximum response of pristine gas sensor to 100 ppm ethanol is 5.1 at 175 °C and it is increased to 48 for Sn-doped CuO gas sensor at 200 °C, demonstrating the promising effect of Sn-doping in CuO for ethanol gas sensing. We also exposed both gas sensors to different concentrations of ethanol at their optimal sensing temperature. Figure 6a and b show the dynamic voltage changes of pristine and Sn-doped CuO gas sensors to 25-200 ppm ethanol gas at 175 and 200 °C, respectively. In both cases, the sensors showed p-type behavior, resulting from intrinsic p-type nature of CuO. Corresponding calibration curves of both gas sensors are presented in Fig. 6c. Obviously for all tested ethanol concentrations (25-200 ppm), the response of Sn-doped gas sensor is higher than that of pristine gas sensor. Figure 7a and b show the dynamic voltage curves of pristine and Sn-doped gas sensors to 100 ppm ethanol at their optimal sensing temperatures, respectively. Based on these curves, the response time and recovery time for pristine gas sensor were calculated to be 76 and 78 s, respectively. Also, the response time and recovery time of Sn-doped gas sensor were 14 and 21 s, respectively. These values show the faster dynamics of Sn-doped gas sensor. Also, Fig www.nature.com/scientificreports/ that the Sn-doped gas sensor is superior to pristine gas sensor not only due to its higher response, but also due to its faster dynamics. Doping of CuO provides a space charge area that helps in increased modulation of the resistance in the presence of gas. Sn-doping in CuO also leads to more porosity and an increase in the surface to volume ratio. More porosity in the gas-sensitive layer causes more gas penetration in the material structure, which will result in more interaction of Sn-doped CuO nanostructures with ethanol gas that eventually improves the sensor response. Selectivity is one of the most important features of a gas sensor. In fact, weak selectivity leads to false alarm which means that the sensor is unable to detect the target gas. To explore the selectivity of both gas sensors, they were exposed to various concentrations of different gases at their optimal working temperatures (Fig. 8a). Obviously, the Sn-doped gas sensor shows a very high response to ethanol gas and much lower response to other gases such as acetone, isopropanol, methanol, H 2 S, toluene, CO 2 , and NH 3 . Even though the pristine gas sensor also shows its highest response to ethanol gas, its response to other gases is close to that of ethanol gas, showing the poor selectivity of this gas sensor. Interestingly, the Sn-doped gas sensor also showed a low response to water vapor, which is present in exhaled breath of drunk car drivers. One of the important requirements for gas sensors to be used in a smart system is repeatability. The repeatability of Sn-doped gas sensor was studied by www.nature.com/scientificreports/ exposing of this gas sensor to 100 ppm ethanol gas during ten sequential cycles (Fig. 8b). There is a negligible difference between the sensing behaviors in different cycles, demonstrating the excellent repeatability of the Sn-doped gas sensor. We also studied the sensing behavior of the Sn-doped gas sensor in the presence of 20-80% relative humidity (RH), measured by a humidity sensor. Figure  www.nature.com/scientificreports/ of Fig. 9a, the response of the sensor in the presence of 20, 30, 45, 70 and 80% RH is 48, 36, 35.2, 31.1 and 28.8, respectively. In general, with increasing of the RH, more water molecules become adsorbed on the surface of sensor, resulting in decrease of the available adsorption sites. Hence, less amounts of ethanol molecules can be adsorbed on the surface of the gas senor, resulting in a decrease of the response in the presence of humidity. However, it should be noted that still in the presence of 80% RH, the sensor has a high response to ethanol gas. The stability of the Sn-doped CuO gas sensor was examined during one month with 5-days intervals (Fig. 9b). There were almost no variations of the response and even after one month the response was 97% of its fresh state. This confirms the high stability of the sensor. The ethanol sensing performance of the Sn-doped CuO gas sensor in this research is compared with those of other gas sensors in Table 1. Obviously, it shows much higher response compared with other gas sensors. Furthermore, its sensing temperature is also relatively low. Accordingly, it can be considered as a very promising sensor for detection of ethanol gas.
In gas sensors based on metal oxides, changes in resistance, current or voltage can be tracked to monitor the concentration of a target gas. In this paper, we converted the resistance change to the voltage change and tracked the voltage change as the sensor output. Changes in voltage values are directly dependent on the concentration of target gas in surrounding atmosphere. In air, the oxygen molecules interact with the surface of the gas-sensitive layer and form O − 2 and O − species as follows: As a result of the absorption of electrons by oxygen ions, the concentration of holes on the surface of Sn-doped CuO increases and a hole accumulation layer (HAL) is formed ( Fig. 10; left), leading to a decrease of resistance Table 1. Comparison of the ethanol gas sensing properties of the Sn-doped CuO gas sensor with those of other gas sensors.  www.nature.com/scientificreports/ in air relative to vacuum condition. When the sensor is exposed to ethanol gas, it reacts with already adsorbed oxygen species and electrons will be released:

Co-doped
Upon release of electrons, the concentration of holes and the width of the HAL decrease ( Fig. 10; right) which results in a sharp increase in voltage.
Furthermore, upon doping of Sn into CuO lattice, some structural defects can be created which eventually provide the preferential adsorption sites for ethanol gas molecules and contribute to response enhancement in Sn-doped gas sensor 42 . Figure 11a shows a real photograph of the fabricated sensing device. The developed system consists of three boards that are placed sequentially on top of each other. A quad-band GSM/ GPRS module (SIM808) was used to collect and send the location data that can be seen on the first floor of the system in Fig. 11b. A GPS antenna with a frequency of 1575.42 MHz and a supply voltage of 3-5 V was used to send the location. Also, a GSM antenna with 2 dB antenna gain was used to send the SMS. A step-down (buck) switching regulator (LM 2596) was used to reduce the 12 V battery voltage to 9 V for powering the GSM/GPRS board. Figure 11c and d illustrate the designed and fabricated printed circuit board (placed on the second floor of the developed system) which includes a Wheatstone bridge circuit, an IoT platform (Node-MCU module), and a circuit that interfaces the Wheatstone bridge to the Node-MCU. The Node-MCU provides a wireless communication link between the gas sensor and the smart phones or monitoring computer connected to the internet. A Wheatstone bridge circuit including 1 kΩ, 2.2 MΩ and 2.2 MΩ resistors was used to convert the gas sensor resistance change to the voltage change. The output voltage is generated between points B and D of the Wheatstone bridge (Fig. S3a) to subtract the voltages from each other (point B and D), an interface circuit was designed based on two operational amplifiers (Fig. 11c).

Details of designed system.
Generally, it is the resistance variations that are tracked in metal oxide gas sensors 71 . However, by using a Wheatstone bridge, we converted the resistance change to the voltage change so that it can be used for next step processes. This conversion also reduces the errors caused by the battery and the next stage circuit [72][73][74] . That is  www.nature.com/scientificreports/ because the sensor resistance value in this design, has no dependence on the internal resistance of the battery or the interface circuit which may reduce the accuracy of the measurements. IDBAD system is powered by connecting to the vehicle battery. Since a typical car battery can provide 12 V voltage and up to 60 A current, there will be no problem in supplying the system, when the vehicle is working. Given that the power consumption of the sensor is 1.6 Wh, when the vehicle is switched off, it can remain in standby mode for about 19 days and there is no need to insert a battery to store the energy in the proposed system. Thus, the proposed system can recognize the ethanol gas for a long-time without external power sources and manage appropriate safety measures. The integrated micro-heater design in the proposed gas sensor has a lower energy consumption in comparison to other heater types used in similar gas sensors. This is very important since in general the working temperatures of metal oxide gas sensors are high, leading to high energy consumption and need for external heater on the back side of the gas sensor 75 . The Wheatstone bridge was designed to work in four different operating modes. Each operating mode can be selected by placing the jumpers in the appropriate position (Fig. S3b). This capability was added to the circuit to improve the ADC resolution. In this way, different resistance ranges are supported without losing the ADC accuracy. For this purpose, four Wheatstone bridges were designed using 2.2 MΩ, 1 MΩ, 100 kΩ and 10 kΩ resistors (two of each resistor in each bridge). Two other sides of the bridges had the variable resistance of the sensor and a 1 kΩ resistor. Hence, testing of a wide range of alcohol concentrations is possible.
Finally, the analog voltage at the output of the interface circuit was converted to digital voltage by the ADC of Node-MCU, which can then wirelessly send the data to the Android app installed in the phone (Fig. 11d), by using the Wi-Fi microchip (ESP-8266). The voltage required to power the Node-MCU was 5 V, which was supplied through the micro-USB port, and its current consumption is only 80 mA. Whenever the output voltage of the detection system reached to the voltage value corresponding to a certain concentration of ethanol gas (already programmed in the microcontroller), the location of the vehicle was automatically sent by SMS to the destination phone number (already programmed).
The upper floor of the IDBAD system ( Fig. 11e) was related to supplying the entire system. This board was consisted of two step-down (buck) switching regulators (LM 2596) and a power supply circuit including a L7805 regulator to supply Node-MCU, a heat sink, a capacitor (470 µF), a resistor (2.2 kΩ) and a green LED. Two LM 2596 modules were set at 9 and 4.1 V voltages and were used to power the SIM808 and supply the micro-heater that controlled the operating temperature of the sensor, respectively. On the power supply board input terminal was connected to the vehicle battery and there was a push button next to the terminal to on/off the system. The block diagram shown in Fig. 11f exhibits the working mechanism of the IDBAD system. The details of the designed Wheatstone bridge circuit and the interface circuit are shown in Figs. S3c-S3f.
As presented in Fig. 12a-g, in practice, the sensor can rapidly identify drunk drivers, generating a signal, sending the vehicle location to the smartphone and prevent the vehicle from starting. www.nature.com/scientificreports/

Conclusions
In this paper, we introduced a location-sender IDBAD gas sensing system based on Sn-doped CuO nanostructures. The proposed sensor was highly sensitive to alcohol exposure. The fabricated gas sensor was easily incorporated into a vehicle for alcohol detection. When the alcohol was sensed by sensor, the communication system wirelessly sent the vehicle location and stopped the vehicle. The developed IDBAD system showed very fast response and recovery time along with good selectivity to ethanol gas. In developed system, power can be supplied to the gas sensor from vehicle battery and also due to relatively low working temperature of gas sensor there is no concern about lack of power supply to the developed IDBAD system. We successfully used our proposed system in a real application and therefore we believe that it can be used for practical applications in different vehicles easily.

Methods
Starting materials. Analytical  Characterizations. The morphological analyses of sensing material were performed using field emission scanning electron microscopy (FE-SEM, MIRA3-TESCAN-XMU, Czech Republic). X-ray energy dispersive spectrometry (EDS) analysis was used to study chemical composition of samples. The phase and crystal structure studies were carried out using X-ray diffraction (XRD, Bruker D8-Advanced X-ray diffractometer, Germany) with Cuk α radiation (λ = 1.5409 Å). Fourier-transform infrared spectroscopy (FTIR, Bruker-Tensor II, Germany) was used for identification of chemical compounds in the products. Raman spectra were recorded using Horiba, XploRA PLUS, France, with a laser excitation wavelength of 532 nm. Fig. 13a, an alumina (Al 2 O 3 ) substrate (6 × 3 × 0.5 mm 3 ) was equipped with Pt interdigitated electrodes (pitch of 150 µm) on the front side. Its backside was used for integration of a Pt micro-heater. The synthesized sensing materials were ultrasonicated with DI water to create a homogeneous solution. Sensing material was coated over the substrate via drop casting technique (3 µl) and then it was heated at 50 °C to dry completely. Finally, it was annealed at 400 °C for 30 min to remove water and dry completely. Since the sensor pads were very small and thin, a sample holder (Fig. 13b) was designed to provide promising electrical connections to the sensor. It was comprised of two alumina plates, gas sensor, four wires and two mica sheets. The silver connection path to the sensor pads was over the alumina plates that were used as large conductor substrates. Silver pads with thicknesses of 500 nm were deposited on the alumina plates by employing DCspattering and a shadow mask. The silver pads were connected to small sensor pads from one end and the copper wires from the other end. By screwing two mica sheets, the silver pads of the two alumina plates were connected to the sensor pads and to the heater pads and the copper wires were hold tight. By applying an external voltage to Pt micro-heater and because of the Joule effect the heater generated heat, leading to increase of sensing device.