Anti - Sound Wave Interference Thermal MEMS Motion Sensors

Xiang Zheng Tu

  

It has been reported that a research team of University of Michigan used a $5 speaker and precisely tuned acoustic tones to deceive 15 different models of accelerometers into registering movement that never occurred. The approach served as a backdoor into the devices - enabling the team to control other aspects of the system. This research calls into question the longstanding computer science belief that software can automatically trust hardware sensors, which feed autonomous systems with fundamental data they need to make decisions.

In this research the accelerometers are capacitive MEMS devices which are typically structured with a diaphragm acting as a mass that undergoes flexure in the presence of acceleration. As shown in the above figure two fixed plates sandwich the diaphragm, creating two capacitors, each with an individual fixed plate and each sharing the diaphragm as a movable plate. The flexure causes a capacitance shift by altering the distance between two parallel plates, the diaphragm itself being one of the plates. Under zero net force the two capacitors are equal but a change in force will cause the moveable plate to shift closer to one of the fixed plates, increasing the capacitance, and further away from the other fixed reducing that capacitance. This difference in capacitance is detected and amplified to produce a voltage proportional to the acceleration. The dimensions of the structure are of the order of microns.

It is not surprise that the diaphragm of the accelerometer is also sensitive to acoustic pressure and works like a capacitive MEMS microphone. A capacitive microphone is commonly formed by a movable membrane and a rigid back plate, forming a structure with a plate capacitor. The movable membrane responds and changes its position when the acoustic pressure hit its surface, producing a capacitance variation between the back plate and the membrane, which in turns produces a current flow proportional to the distance variation between the membrane and the back plate.

There is no essence difference between these two capacitive MEMS devices. It is true that said Kevin Fu, U-M associate professor of computer science and engineering, the fundamental physics of the hardware allowed us to trick capacitive accelerometers into delivering a false reality to the microprocessor. And their findings resonantly upend widely held assumptions about the security of the underlying hardware.

However POSIFA’s thermal MEMS motion sensors can make these things total different. Using the POSIFA’s thermal MEMS motion sensors sound waves are no longer allowed to hack everything from phones to fitness trackers. Reference to the above figure a POSIFA’s thermal motion sensor comprises a thermal isolated plate created in a silicon substrate, a resistive heater, and two thermopiles both are formed on the surface of the plate. The laws of physics teaches that the temperature field generated by a moving heat source is asymmetry and able to be measured. In steady state, the vertical cross-sectional temperature field is a sequence of symmetry concentric circles each representing an isotherm on the lateral plane. When the heat source moves the vertical cross-sectional temperature field will be skewed towards down motion direction. The skewed lateral cross-sectional temperature field consists of a contracted half plane and an expended half plane both are divided by a line perpendicular to the motion direction. Since two thermopiles sensors are placed on the plane around the heat source, all isotherms can be reconstructed. A lot of useful information including the direction and velocity of the moving heat source can be extracted form the reconstructed plane isotherms.

Acceleration is used to measure the change in velocity, or speed divided by time. For example, a car accelerating from a standstill to 60 mph in six seconds is determined to have an acceleration of 10 mph per second (60 divided by 6). So with several accelerometers on your smart phone, you can determine if the smart phone is moving uphill, whether it will fall over if it tilts any more, or whether it’s flying horizontally or angling downward. And you know how to tilt your smart phone it can rotate their display between portrait and landscape mode accordingly.

The thermopile flow sensors can replace capacitive accelerometers for measuring the speeds of any moving objects including smart phones. The working principle is based on the fact that a moving object experiences an apparent wind that is the wind in relation to the moving object. Suppose the object is a riding bicycle on a day when there is no wind. Although the wind speed is zero, the rider will feel a breeze on the bicycle due to the bicycle is moving through the air. This is the apparent wind. On the windless day, the measured apparent wind will always be directly in front and equal in speed to the speed of the bicycle. It is very clear that it is impossible for the thermal motion sensors to response sound wave because there is no sound wave sensing mechanism to take place. 

Thermopile Temperature Sensors

Xiang Zheng Tu

POSIFA’s thermopile sensors are based on the technologies of silicon micromachining and CMOS manufacturing. In the thermal sensor fabrication a part of silicon substrate is removed by a porous silicon etching process, leaving on top only a thin sandwich layer membrane of PECVD SiO₂/Si₃N₄, which has low thermal conductivity. On the membrane a resistor and two thermopiles are formed there which are used as the main device elements of the sensor. The resistor is located along the central line of the membrane and made of a deposited thin polysilicon film. Two thermopiles are located on the two opposite sides of the resistor respectively. The thermopiles have alternate hot junctions disposed near the resistor and alternate cold junctions expending out of the membrane and ended on the bulk part of the silicon substrate. The junctions are formed by both a deposited thin polysilicon film and a deposited thin aluminum film.

The thermopile sensors can be used as a lot of different thermal sensors with a little modification which include fluid flow sensors, gas thermal conductivity sensors, vacuum sensors and temperature sensors. In working of the temperature sensors the polysilicon resistor is used as a heater and one thermopile is used for temperature different sensing.

Since the membrane is heated by the heater the temperature different generates between the membrane and the bulk part of the silicon substrate. It is need to know that the low thermal conductivity of the membrane is beneficial to the temperature different maintain.

The power dissipated in the polysilicon resistor causes its temperature to rise above a room temperature by:

T_PR-T_room=(V²/R_PR)θ                                       (1)

Where

T_PR=the raised temperature of polysilicon resistor due to internal power dissipation

T_room  =the reference temperature of polysilicon resistor

V=the driving voltage of polysilicon resistor in V

R_PR=the value of the polysilicon resistor in ohms at T_PR; and

θ=the self-heating polysilicon resistor ting effect in °C/mW.

For an ideal thermocouple, the open-circuit voltage obtained is proportional to the temperature difference between the hot junctions and cold junctions which are constructed of polysilicon film and aluminum film,

  

∆V=S(T_PR-T_room)                                                (2)

where S is the relative Seebeck coefficient, expressed in µV/K. The relative Seebeck coefficient of a junction can be calculated as the absolute value of the each Seebeck coefficient of polysilicon and aluminum; that is,  

S=S_Poly−S_Al                                                                      (3)

Because a voltage is produced when a temperature difference exists between the two junctions of the thermocouple junction pair, the thermocouple can be used as a detector of incident radiation. In open-circuit operation the output voltage produced is usually low, on the order of a tenth of a microvolt per degree celsius of temperature difference for a single junction pair. In order to increase the output voltage, several junction pairs may be connected in series. The responsivity is then increased by n if n thermocouple junction pairs are placed in series; that is,

∆V=n(S_Poly−S_Al) (T_PR-T_room)                                  (4)

Such a device is called a thermopile.

Combining equation (1) and (5) results the equation as:

∆V=n(S_Poly−S_Al ) (V²/R_PR)θ                                     (5)

The resistance of a polysilicon resistor is specified at a room temperature, any other resistance at another temperature is determined by:

R_PR=R_room[1+α(T_PR-T_room)]                                     (6)

Where
R_room=the resistance at a room temperature in ohm
α=the temperature coefficient of the resistance

After replacing equation (6) into equation (5) an equation is obtained as

∆V=n(S_Poly−S_Al ) V²θ / R_room[1+α(T_PR-T_room)]        (7)

From the equation (7) it can be seen that the output voltage ∆V of the thermopile is inversely proportional to the resistance R_room at a room temperature.

An example of a resistance and temperature coefficient of polysilicon resistors with temperature are shown in the following figure.

 

These polysilicon resistors were made of a 700 nm thick polysilicon films that were grown at 580^◦C in furnace by low-pressure chemical vapor deposition (LPCVD). The polysilicon films were partially implanted by boron (p-type) and phosphorus (n-type) from 4×10¹⁵ to 10×10¹⁵ at cm⁻² doses. The implant energy for each group doses was 70-80kev. Afterward, the doped polysilicon films were annealed in furnace at 1000^◦C for 30 min to activate dopants and obtain a uniform doping profile through the whole thickness and repair the defects in the crystalline structure.

As shown in the above figure the resistance of the polysilicon resistor varies with temperature. If the room temperature T_room rises from 15⁰C to 30⁰C the resistance will change about 200Ω for a polysilicon resistor with 15800Ω at 20⁰C. Since the output voltage ΔV of the thermopile is inversely proportional to the resistance of the polysilicon resistor (as a heater) a specified room temperature can be determined from the measured output voltage of the thermopile.

Off cause polysilicon resistors without thermopiles can also be used as temperature sensors. But common types of resistor temperature sensors are made from platinum instead of polysilicon. Platinum has temperature coefficient α = 0.003925 Ω/(Ω·°C) and polysilicon with sheet resistance ranging from 25 to 150Ω/□ has temperature coefficient α =1x10^-3 Ω/(Ω·°C). So the sensitivity of the platinum resistor temperature sensors is much higher than the polysilicon resistor temperature sensors.

Thanks to CMOS manufacturing technology it allows POSIFA to integrate up to 40 pair of thermocouples in each thermopile. That means the responsivity of the thermopile temperature sensors is increased by 40 which is comparable with platinum resistor temperature sensors.

There are three additional advantages that justify the use of MEMS and CMOS manufacturing technologies.

1. CMOS offer the thermopile structure materials with higher Seebeck coefficient.

2. CMOS beneficial to tune the main characteristics of the thermopiles by doping.

3. MEMS allow the thermal capacity of the thermopile to be reduced effectively by miniaturization.

Thus POSIFA’s thermopile temperature sensors have the best potential to satisfy the demands on miniaturization and mass production.