Air Temperature Measurement Using CMOS-SOI-MEMS Sensor Dubbed Digital TMOS ^†

Abstract

Air temperature is an important meteorological parameter and is used for numerous purposes. Air temperature is usually observed using a radiation shield with ventilation, to obtain proper measurements by providing shade from direct solar radiation and increasing the heat exchange between the sensor and the atmosphere. In rural areas, such auxiliary equipment is not available and it is still a challenge to obtain the air temperature accurately without aspiration. In this study, we describe a novel, qualified, complementary metal-oxide-semiconductor-Microelectromechanical systems (CMOS-MEMS) low-cost sensor, dubbed Digital Thermal-MOS (TMOS), for remote temperature sensing of air temperature. The novel key ideas of this study are (i) the use of the Digital TMOS, (ii) a narrow optical bandpass filter (4.26 um +/− 90 nm) corresponding to the CO₂ carbon dioxide absorption band; (iii) simultaneously measuring the weather parameters.

1. Introduction

Air temperature sensors and methods of measurement have been reported in several recent articles. All the papers describe the importance and challenges of air temperature measurement [1,2,3,4].

Air temperature is an important meteorological factor, which has a wide range of applications in fields such as human health, virus propagation, growth and reproduction of plants, climate change, and hydrology. The WEB offers many types of thermometers for air temperature measurement, see for example [3].

What this report forgot to explain is that such thermometers cannot measure air temperature accurately since the measurement of air temperature is strongly influenced by environmental factors such as humidity, solar radiation/duration, solar altitude angle, wind speed/direction, rainfall, atmospheric pressure, and other objects within the field of view. These factors make air temperature measurement very challenging. These issues become more pronounced in low-cost air temperature sensors, which lack a radiation shield or a forced aspiration system, exposing them to direct sunlight and condensation.

Air temperature is often observed using a radiation shield with ventilation to obtain proper measurements by providing shade from direct solar radiation and increasing the heat exchange between the sensor and the atmosphere, thereby reducing error due to solar radiation incident on the sensor surface. Although a power-saving type shield that is driven by solar panels has been developed [4], many products still require mains power. However, mains power is not always available in rural areas, such as areas of farmland and forest. Meanwhile, recent studies have found strong contributions of the local background climate to the air temperatures observed at urban weather stations and these temperatures are sometimes different from those of nearby rural areas. It is thus important to observe the air temperature more easily in rural areas, and it has long been a challenge to obtain the air temperature accurately without aspiration.

In this study, we describe air temperature measurement, using a novel qualified CMOS-MEMS low-cost sensor, dubbed Digital TMOS, for remote temperature sensing of air temperature. The novel key idea of this study is the use of a narrow optical bandpass filter (4.26 um +/− 90 nm) corresponding to the CO₂ carbon dioxide absorption band, on top of the Digital TMOS.

2. The Digital Thermal Device for Remote Temperature Sensing—Dubbed TMOS

The Digital TMOS has been described in several recent publications [5,6,7,8,9]. It is also reported in the STMicroelectronics catalog [10]. The TMOS, short for Thermal-MOS, is a thermal sensor based on the thermally insulated MOSFET transistor. The TMOS is achieved by using CMOS-SOI and MEMS process.

The Digital TMOS is described in Figure 1 below:

The sample tailored to air temperature measurements contains two Digital TMOS units; each TMOS is covered with an optical filter of 4.26 µm +/− 90 nm (see Figure 2).

The reason we are using two Digital TMOSs is to prove the following: The readings of the sensors are changing very rapidly. We assume that if the variations are due to the environmental changes, the readings of the sensors will be in correlation. In contrast, if the changes are due to sensor instability, there will be no correlation. This measurement is required to validate the results.

3. The Role of the Narrow Bandpass Optical Filter around 4.26 µm

It is well established that remote sensing of temperature requires that the measured object will behave like a blackbody with known emissivity. The air emissivity is reported in [11]. The air is transparent to the visible and Near-Infrared (NIR) radiation. CO₂ is always present in the atmosphere as a greenhouse gas, with at least 400 PPM concentration. The CO₂ in the atmosphere absorbs the 4.26 um radiation within an optical path of ~20 m. Hence, at this wavelength, the air may be treated like a blackbody with emissivity close to 1.

The Digital TMOS with the CO₂ filter is first calibrated with a blackbody [12] in the range of 0–100 °C and then it is ready to measure air temperature.

According to Planck’s law, for a certain wavelength λ, at temperature T, the radiation power is as follows:

$$W\left(\lambda ,T\right)=\frac{2\pi h{c}^2}{{\lambda }^5}\times \frac{1}{\exp \left(\frac{hc}{\lambda kT}\right)-1} \left[\frac{\mathrm{Watt}}{{\mathrm{cm}}^2\mathsf{\mu }\mathrm{m}}\right]$$

(1)

where c is the velocity of light, h is Planck’s constant, and k is Boltzmann’s constant.

Since we assume that the air temperature T_air is equivalent to the blackbody temperature T_BB, at 4.26 um, we can rewrite the equation for a spectral measurement as

$$W_{{\lambda }_1-{\lambda }_2}\left(T_{BB}\right)=W_{{\lambda }_1-{\lambda }_2}\left(T_{air}\right)={{\displaystyle \int }}_{{\lambda }_1}^{{\lambda }_2}{W}_{\lambda }\left(\lambda ,T_{air}\right)d\lambda$$

(2)

where in our case the wavelength spectrum [λ₁, λ₂] is the CO₂ filter’s optical bandpass 4.26 µm ± 0.09 µm.

The radiation, measured by the Digital TMOS, is given by

$$\begin{gathered} P_{\lambda 1-\lambda 2}\left(T_{air}\right)={\epsilon }_{air}\times A_D\times W_{\lambda 1-\lambda 2}\left(T_{air}\right)\times {\mathit{sin}}^2\left(\frac{\theta }{2}\right)\times t_{filter}\times \eta \\ ={\epsilon }_{air}\times W_{\lambda 1-\lambda 2}\left(T_{air}\right)\times \left[\underset{PTF}{\underbrace{A_D\times {\mathit{sin}}^2\left(\frac{\theta }{2}\right)\times t_{filter}\times \eta }}\right] \end{gathered}$$

(3)

In the case of a conic view of the detector, the solid angle will be summed across the field of view:

$$Power Transfer Function=PTF=A_D\times {\mathit{sin}}^2\left(\frac{\theta }{2}\right)\times t_{filter}\times \eta$$

(4)

where A_D is the detector area, t_filter is the transmission factor of the filter, $\theta$ is the field of view (FOV) angle, and η is the absorption coefficient.

The total absorbed radiation by the digital sensor is

$$P_{\lambda 1-\lambda 2}\left(\epsilon , T_{air},T_{amb}\right)=PTF\times {\epsilon }_{air}{W}_{\lambda 1-\lambda 2}\left(T_{air}\right)+PTF\times \left(1-{\epsilon }_{air}\right)\times W_{\lambda 1-\lambda 2}\left(T_{amb}\right)$$

(5)

The second term on the right side is the reflected power that the detector senses and it is determined by the objects around the sensor, such as the earth. Obviously, with air emissivity close to 1, this term becomes less significant.

The Digital TMOS measures the incident radiation (Equation (5)) and by applying the calibration algorithm of the sensor, the air temperature is obtained.

4. Measurement Results

Our measurements are classified into two types of measurements:

• Measurements at nighttime at different view angles, made with a Digital TMOS tailored to the application;
• Measurements at daytime with different scenarios.

4.1. Nighttime Measurements

The measurements were taken from a window view, around 6–8 m above ground, in three different angles. (i) A low angle with a ground view where the sensor is looking downwards. (ii) A mid angle with a horizontal view; the sensor is looking forwards. (iii) A high angle with a sky view; the sensor is looking upwards. The setup view and the results for nighttime measurements are shown in Figure 3.

From the results, we can learn two interesting behaviors of the air temperature. First, the temperature changes rapidly. Second, the temperature changes with the view for obvious reasons, as explained below.

The heat capacity of the ground is larger than that of the air. Therefore, the ground is hotter than the air at nighttime. At the higher angle view, looking towards the sky, we measured the colder air temperature. At the lower angle view, the ground is viewed by the sensor and the temperature is warmer. The horizontal view corresponds to the air temperature.

4.2. Daytime Measurements

We measured the air temperature with 3 detectors:

• Digital TMOS tailored to the application;
• Sparkfun weather shield [13] with Si-7021 temperature/humidity detector, manufactured by Silicon Lab (Austin, TX, USA) [14];
• RTD PT 100 manufactured by Dracal Technology (Quebec, Canada) connected to a USB adapter [15].

We took additional measurements of the relative humidity with the Si-7021 detector, the ambient light with an ALS-PT19 Sensor manufactured by EVERLIGHT (New Taipei City, Taiwan) [16], and the barometric pressure with an MPL3115A2 pressure sensor manufacture by Freescale Semiconductor (Austin, TX, USA) [17]. The measurements were taken in three different views: a corridor view, a sunny view, and a shadow view. The temperature results are as shown in Figure 4.

Figure 4a examines the results in a corridor indoor view. The Digital TMOS and the Si-7021 are standing together at the end of the corridor, viewing its full view. The RTD is standing 8 m inside the corridor, nearer to the entrance, measuring the temperature locally.

We can see in the corridor measurement that the Digital TMOS shows the coolest temperature. Since there is no sun in the corridor, we assume that the main reason for the differences is air tunnels in the entrance of the corridor that lower the temperature in the far view from the Digital TMOS location. Because of that we can also see strong fluctuations in the TMOS measurements.

Figure 4b examines the results in a sunny view, next to a rocky ground, where the Digital TMOS is collecting data while standing in the shadow. The Si-7021 and the RTD PT100 are collecting data in the sun.

Figure 4c examines the results in a shadowy view, next to a rocky ground, where the Digital TMOS is collecting data standing in the sun, and looking on a shadowed view. The Si-7021 is measuring next to it, and the RTD PT100 is collecting data in the shadowed view. All measurements were taken in a two-minute time.

As for the sunny view measurements, we can see that all the temperatures have risen in all the detectors. However, this time the Digital TMOS has shown the highest temperature, due to the rocky ground that was warmer than the air temperature. We can also see strong fluctuations that we will try to explain later. As for the other detectors, the Si-7021 has a linear change in the temperature due to heating in the sun, and the RTD PT100 is showing a stable temperature at the shadowed view.

In the shadow view measurement, the RTD PT-100 is exposed to the sun and the Si-7021 changed location to the shadowed view. However, it took it much time to cool off. Compared to it, we can see that the Digital TMOS measurements are two degrees higher but still with a strong fluctuation during the measurement.

Figure 5 exhibits the inverse correlation between the Digital TMOS and the measurement of the relative humidity. Assuming that the absolute humidity does not change during the short measurement time, it is obvious that if the temperature increases, the relative humidity is reduced and vice versa.

The measurement results are summarized below:

• There is a strong and rapid connection between air temperature measurements derived from radiance power and the view we are measuring. We assume that it is due to rapid changes in the absolute humidity;
• Landscape and earthly view have strong influence on the measurements. In order to receive most accurate measurements, the field of view should be narrow and the sensor should see just air.

5. Summary

Accurate air temperature measurement remains challenging, despite decades of research and development to improve instruments and methods.

Air temperature measurements are an essential component of weather monitoring and climate research worldwide and will continue to be challenging given the tradeoffs between accuracy, power consumption, and costs of the instrument options. With the advent of sensor technology and machine learning techniques, the performance of air temperature sensors will keep improving.