Mems Sensors for Search-and-rescue Robots

Mems Sensors for Search-and-rescue Robots

Initial Report

Introduction

During the last few decades, the death toll from natural disasters was more than 700,000 worldwide ^[1]. Detection of survivors under the rubble within a short amount of time, which is 72 hours for healthy adult with a supply of fresh air ^[2], will permit a successful rescue. Search and rescue team often deal with highly complex and dangerous environments. These sites are a great threat to rescue workers and survivors. In such case, robots equipped with advanced sensors are being identified as a great help for human rescuers where they are replaceable, can reach spaces humans can’t, operate continuously in harsh environments, and collect important information such as detecting human presence. This is where MEMS can be used. In this project, our team will design 2 MEMS sensors, IR temperature sensor and a MEMS microphone, for a small search-and-rescue robot to detect body heat and human breath.

1)   IR thermoelectrical sensor

Introduction

Scientists discovered that all objects whose temperature is more than zero Kelvin emits an infrared radiation at some wavelength, with the wavelength being directly proportional to the temperature and IR radiation of the object ^[3]. This radiation is an electromagnetic energy that is invisible to the human eye but can be modelled and detected using thermal electronic devices. By taking advantage of sensing the infrared fluctuations, we can design a passive IR sensor, does not emit any energy of any type but merely sits ‘passive’ accepting infrared energy, that detect human presence.

Sensor elements:

A. Thermopile

Thermopile is a device that converts thermal energy into electrical energy. It works based on the Seebeck effect where two types of materials, usually metals, are connected together in series, a current will be generated due to the temperature difference after creating hot junctions on highly heat-resistant conductive membranes, and cold junctions on highly thermal isolated material. This current called a thermal current, and the corresponding electromotive force is called the thermoelectric potential ( $V_{\mathit{out}}$

) and its direction depends on the direction of the temperature gradient ^[3]. See figure [1].

B. Wavelength filter

In general, sensor elements are sensitive to radiation of wide range of wavelengths but due to the use of a filter window that limits the sensitiveness to the range that is most suitable to human body radiation. Several factors such as surface temperature, surface shape and object type will affect the wavelength range and energy density (i.e., intensity of radiation). It was found that the temperature of the skin of a typical human being is 37 degree, and the lowest and highest values are 14 and 46 degree respectively ^[5].  We can calculate the wavelength of the radiation emitted by a human after converting this temperature to Kelvin by using the relationship between wavelength and temperature called Wein’s displacement law:

IR detectors that are sensitive in the 8 um -11 um range would thus be able to detect humans.

Sensor working principle:

The working principle behind the thermoelectric micro-sensor based on MEMS depends on the IR radiation form of heat source. A high-performance lens gathers radiated IR rays within the wavelength specified limits on the absorber. the absorber will convert the electromagnetic energy into heat energy. The hot junction of the thermopile is attached to the absorber. When the absorber is heated, there will be a temperature difference between the cold and hot junction of the thermopile. This will generate a voltage that can be further analyzed through IC to convert sensor signal to digital temperature output. See figure [2].

Thermocouple produce a thermal potential due to the temperature difference as follow:

$V_{\mathit{out}}=\left({\alpha }_A–{\alpha }_B\right)∆T$

Where ${\alpha }_A\mathrm{ }\mathit{and}\mathrm{ }{\alpha }_B$

are the Seebeck coefficients for thermoelectric materials A and B respectively. $∆T$

is the temperature difference between the two junctions. Since a thermopile is a series-connected array of thermocouples, output voltage generated from each thermocouple is directly proportional to the number of thermocouples, N is given by:

$V_{\mathit{out}}=N\left({\alpha }_A–{\alpha }_B\right)∆T$

In order to design a thermoelectric sensor with the highest output voltage, it is important to select two materials with the greatest difference in Seebeck coefficients.  Another design consideration is the heat losses between hot junction and the thermocouple surroundings in MEMS-based Thermoelectric infrared sensors. It can be divided into three types: Solid thermal conduction, gas convection and thermal radiation see figure [3]. Thus, the total thermal conductance can be determined by the equations below.

$G_{\mathit{th}}=G_{\mathit{solid}}+G_{\mathit{gas}}+G_{\mathit{rad}}$

Solid thermal conduction, due to heat dissipation through solid suspended thermopile structure onto a substrate; gas convection, due to the heat lost by the molecular movement in air; and thermal radiation, due to the energy radiated from the heat sources in the form of electromagnetic waves ^[6].In our project, we are planning to apply a vacuum packaging, therefore, gas convection and thermal radiation can be neglected. 

Solid conduction can be calculated by the equation below:

$G_{\mathit{solid}}=N\frac{k_1{t}_1{w}_1}{l_1}+N\frac{k_2{t}_2{w}_2}{l_2}+\frac{k_3{t}_3{w}_3}{l_3}$

Where $k_n,t_n,w_n$

and $l_n$

represent material conductivity, layer thickness, total width of the layer and length of the layer, respectably. n=1,2 is for selected thermocouple materials, n=3 is for isolated membrane.  The output efficiency of the thermoelectric sensor (responsivity) can be determined by:

$R=\frac{V_{\mathit{out}}}{P_{\mathit{absorb}}}=\frac{N∆T({\alpha }_2–{\alpha }_1)}{\mathit{\eta A\sigma }\left(T_s^4–T_0^4\right){\mathit{sin}}^2\theta }$

where $P_{\mathit{absorb}}$

is the infrared power that is absorbed by the thermopile, A is the absorber area $\eta$

is the absorption rate of the absorber layer, σ is the Stefan–Boltzmann constant, Ts is the temperature of source, $T_0$

is ambient temperature, and  $\theta$

is the field view angle.

Noise of the sensor can by idntified using voltage electrical Johnson noise:

$v_n=\sqrt{4k_B\mathit{TR\Delta f}}$

where $k_B$

is the Boltzmann constant, R is the electrical resistance of the detector, $\mathit{\Delta f}\mathrm{ }$

is the measurement frequency bandwidth, and T is the absolute environment temperature. One of the challenges facing many research institutions is to improve the applicability of thermoelectric materials.

Initial selection:

Due to the advantages of the CMOS-compatible process, we will select the following:

1. heavily doped silicon slab is used as the absorber because of its high absorption ^[7].
2. Thermocouple materials: p doped Silicon and n doped polysilicon are used ^[8].

Table 2(Parameters of thermal couple materials ^[7])

2)   MEMS microphone

Introduction

MEMS microphones can be used to research human by detecting the sound they make such as their breathing frequency because it can convert sound into an electrical signal. MEMS microphones is consisted of MEMS micro-capacitance sensor, micro-integrated conversion circuit (amplifier), acoustic cavity and anti-noise circuit. With its structure, MEMS microphone completes the conversion of “sound electricity”.

Structure of MEMS microphone

Several layers of different materials are deposited on the wafer, then the unwanted material is etched to form a chamber on the base wafer, which is covered with a movable diaphragm and a fixed back plate. See figure [4]. The sensor backplate has a stiff perforated structure that allows air to move through it easily, while the membrane is a thin solid structure which will flexes in response to the changes in air pressure caused by sound wave. And the movement of the membrane creates a change in the amount of capacitance between the membrane and the backplate, which is translated into an electrical signal by the ASIC.

The general process flow is as follows. More details will be considered into the second report.

1) Start with a silicon wafer. 2) Deposit a layer of silicon nitride and a photoresist. Pattern and dry etch. Ash resist. 3) Deposit a layer of SiO2 and a layer of photoresist. Pattern and dry etch. Ash resist. 4) Deposit a thin layer of conductive material. 5) Deposit a layer of silicon Deposit a layer of silicon nitride and a photoresist Pattern and dry etch. Ash resist. 6) Deposit a layer of SiO2 and a layer of photoresist. Pattern and dry etch. Ash resist. 7) Deposit a thick layer of conductive material. Pattern and dry etch the side first. 8) Use lift-off to deposit the electrodes. 9) Deposit a layer of photoresist. Pattern and dry etch the acoustic holes. 10) Use DRIE to wash the cavity. 11) Use BOE to etch SiO2.

Specifications of MEMS microphone

Breath Frequency fb: Breathe frequency is a low parameter, about 16 ~ 20 times/min on average. That is fb=0.27 ~ 0.33 Hz. This parameter helps us design an appropriate filter to reduce noise.

Breath pressure range P: A significant parameter used to calculate how much force is acting on the membrane.

Membrane displacement: When the sound wave comes, it will cause air pressure change which will vibrates the membrane

Bias Voltage ${\mathit{V}}_{\mathit{bias}}:$

Bias voltage applies a stable voltage to the sensor so that its slight voltage variation could be detect precisely.

Amplifier magnification: The voltage variation caused by breathing is in mV, which is too small for us to analyze. Therefore, at least one amplifier and its magnification are necessary.

Sensitivity: Sensitivity indicates how the microphone converts the acoustic pressure into an output voltage. High-sensitivity microphones generate more voltage, so less amplification is required in the mixer or recording device.

Self-noise: Self-noise is the level of sound that produces the same output voltage as a microphone does without sound. This represents the lowest point of the dynamic range of the microphone

Sensor Working Principle:

The flow chart is shown in figure [5].

Figure 5

The operation of the sensor is based on the capacitive principle. Sound wave generates pressure, which vibrates the membrane see figure [7]. As we have the range of the breath frequency, we can know the pressure of breath through the equation.

$P^2 =\mathit{ I\rho c}$

Where P is the sound pressure, $\mathit{I }$

is sound intensity, ρ is medium density, and c is the sound velocity. The magnitude of sound intensity is directly proportional to the speed of sound, the square of the frequency of the sound wave and the square of the amplitude.  Sound pressure changes the distance between the plates. Since force is applied at tip, if we find maximum tip displacement, the ratio of displacement to force is the spring constant.

$\mathrm{F}=k{\delta }_{\mathit{max}}$

$\mathrm{k}=\frac{32E{a}^3}{5b{l}^3}$

Where a is the height of the membrane, b is the width of the membrane, $l$

is the length of the membrane. If we Considered the simple mechanical formula, with F the force on the mass proof (membrane) see figure [8], then:

$F\left(t\right)=\mathrm{ }P\left(t\right)\mathit{bl}$

Thus, we get the equation motion of the membrane

$P\left(t\right)\mathit{bl}=\frac{32E{a}^3}{5b{l}^3}x$

Therefore, the capacitance changes, which is translated into an electrical signal. The sensor is sealed by a back chamber so that sound can only reach one side of the membrane. The variation of the gap leads to the change of capacitance. The equation of the capacitance is:

$\mathrm{C}=\frac{\mathrm{\epsilon }\mathrm{A}}{\mathrm{d}}$

$\mathrm{x}=d_0–d$

$A=\mathit{bl}$

Where C is the Capacitance, ε is Dielectric constant; A is membrane area; $d_0$

is Initial Distance; d is the gap distance, $\mathrm{ }x$

is the fluctuating Distance. Consider electrostatic force, the equation is

$F_e=\frac{\mathrm{\epsilon }\mathrm{A}{V}^2}{2d^2}$

$F_{\mathit{net}}=F_m–F_e$

$\mathit{keff}=k_m–k_e=–k+\frac{\epsilon A{V}^2}{d^3}$

Thus, we get the equation motion of the membrane

$p\left(t\right)\mathit{bl}=(\frac{32E{a}^3}{5b{l}^3}–\frac{\epsilon \mathit{bl}{V}^2}{{(d_0–x)}^3})x$

Since $\mathrm{C}=\frac{Q}{V}$

, Q must be a constant so that V could be detected precisely. Therefore, a high bias voltage (around 10~12V implemented with a charge pump) should be applied to the circuit. Once Q decreases, the voltage makes up for the loss in a second, which makes V be detected easily. The output circuit has to contain filters, amplifiers, and buffers. Filters reduce noise, not only the noise of breath signal, but the components. Amplifiers proportional increase the signal, making it more convenient for people to analyze. When it comes to buffer, it makes sure that the output voltage would not be influenced no matter how many components are connected.

For plates separating, the Sensitivity $S_C$

is:

$S_c=\frac{\partial C}{\partial d}=\frac{\mathit{\epsilon A}}{{(d_0–x)}^2}$

So, $d_0$

should be small enough to meet the requirement of capacity sensitivity

For the microphone sensor, the Sensitivity Sis:

$\mathrm{S}=\frac{\partial V}{\partial P}$

So

$\mathrm{V}=\frac{\mathrm{Q}(\mathrm{d}0–\mathrm{x})}{\mathrm{\epsilon A}}$

Since we get the equation of V and P, S can be calculated.

According to the equation, when $d_0$

becomes higher, S decreases, which means that the sensor is more sensitive. The S should be small enough to promise the microphone can detect the human’s breath. The general circuit is showed on figure [9]. More details will be considered on the second report.

More problems and possible solutions:

The general structure and signal processing circuit are figured out, but there are still many details must be considered. Here we list all the difficulties we could think of that are waiting to be solved. We also propose solutions about some of them.

1. Frequency and output voltage: The natural frequency of the membrane influences the gap too, which also makes a difference to the output voltage.

Solution: First determine the natural frequency range of the membrane, then compare it with the frequency of human breathe. If the natural frequency is apparently higher of lower than the respiratory rate, it can be easily filtered in the signal processing circuit, otherwise, we have to think of a way to minimize it.

2. Brownian motion: The air molecules in the acoustic holes aren’t perfectly evenly distributed, which build up a force from fluctuations. Moreover, the movement of the robot causes air to enter the acoustic holes to create noise.

Solution: Keep the sensor in a relatively low temperature when packaging to reduce air molecule activity. In addition, only keep the acoustic holes part outside, minimize the exposure area.

3. Sources of Noise: In the circuit part, components like resistances and amplifiers create noise.

Solution: First we must calculate the values of the noise. Then compare them with the respiratory signal. If components noise is much lower, it can be neglected, or we have to filter it also. The specific filtering method has not been determined yet. One possible solution is to add a MOEFET as a pre-amplifier.

Future work

the mechanical structure is not the final version. The material, size, and more details need to be revised after everything is considered.

Once all the frequency and value of noise are determined, the circuit can be improved, and the specific value of the components can be decided. Moreover, the mechanical structure is not the final version. The material, size, and more details need to be revised after everything is considered.

References

[1] United Nations Office for Disaster Risk Reduction (UNISDR).

[2] Zhang, D., Sessa, S., Kasai, R., Cosentino, S., Giacomo, C., Mochida, Y., . . . Takanishi, A. (2018). Evaluation of a Sensor System for Detecting Humans Trapped under Rubble: A Pilot Study. Sensors,18(3), 852. doi:10.3390/s18030852

[3] Lian-min, C., Ya-zhu, Z., Shi-jiao, S. et al. Microsyst Technol (2018) 24: 2463. https://doi-org.ezproxy.neu.edu/10.1007/s00542-018-3809-2

[4]  http://www.kippzonen.com/News/572/The-Working-Principle-of-a-Thermopile-Pyranometer#.W8y6Ii2ZPVo]

[5] Wikipedia contributors. (2018, October 18). Human body temperature. In Wikipedia, The Free Encyclopedia. Retrieved 16:22, October 23, 2018, from https://en.wikipedia.org/w/index.php?title=Human_body_temperature&oldid=86463576

[6]  Xu, Dehui & Wang, Yuelin & Xiong, Bin & Li, Tie. (2017). MEMS-based thermoelectric infrared sensors: A review. Frontiers of Mechanical Engineering. 1-10. 10.1007/s11465-017-0441-2

[7] Fonseca L, Santander J, Rubio R, et al. Use of boron heavily doped silicon slabs for gas sensors based on free-standing membranes[J]. Sensors and Actuators B: Chemical, 2008, 130(1): 538-545.)

[8] Müller, M., et al. “A thermoelectric infrared radiation sensor with monolithically integrated amplifier stage and temperature sensor.” Sensors and Actuators A: Physical 54.1-3 (1996): 601-605.)