Вы находитесь на странице: 1из 40

NORTHEASTERN UNIVERSITY EECE7244/ME6260 INTRODUCTION TO MICROELECTROMECHANICAL SYSTEMS

MEMS Accelerometer Integration with a Football Helmet


Final Project Report
Caitlin Curtis, Alex Krull, Shailesh Tendulkar, Raul Vyas, Osman Yigid 12/12/2012

Our MEMS device is an open loop capacitive accelerometer, utilizing a differential capacitance measurement technique to detect acceleration in the range of 500gs with a resolution of 500mg and electrical sensitivity of 16.8mV/g. Each single fabrication is composed of two 2-axis accelerometers, which utilize four parallel guided beams to control the displacement of a collection of comb fingers, from which differential capacitance is measured. In order to achieve three dimensional measurement capabilities, two of these fabricated devices will be mounted orthogonally during the final packaging phase the device. In addition to these performance measures, the device has been optimized to achieve material strength and cost targets. Its max stress at worst case acceleration impact (200g) is 9.0 MPa, which is less than 1% of the yield strength of polysilicon, the material from which it is fabricated. With regard to device size, its envelope has been minimized (1218 m X 1218 m) in order to reduce cos ts and achieve improved economies of scale, while maintaining the key performance metrics required for operation.

Table of Contents
List of Figures ................................................................................................................................................ 2 List of Tables ................................................................................................................................................. 3 Introduction .................................................................................................................................................. 4 Market Value................................................................................................................................................. 4 Mechanical Design ........................................................................................................................................ 5 Mechanical Model .................................................................................................................................... 5 Electrical Model ........................................................................................................................................ 5 Geometric Design Constraints .................................................................................................................. 6 Stress Analysis ........................................................................................................................................... 8 Mechanical Specifications Summary ........................................................................................................ 8 Electrical Design ............................................................................................................................................ 9 Parallel Plate Capacitor ............................................................................................................................. 9 Transverse Comb Capacitance Configuration ........................................................................................... 9 Lateral Comb Configuration .................................................................................................................... 10 Capacitance Accelerometer Design ........................................................................................................ 11 Circuit Simulation Results ....................................................................................................................... 13 Sensitivity Analysis .................................................................................................................................. 16 Pull-in ...................................................................................................................................................... 17 Noise ........................................................................................................................................................... 18 Mechanical (Brownian) Noise ................................................................................................................. 19 Capacitive Noise ...................................................................................................................................... 19 Sensor Charging Reference Voltage (SCRV) Noise .................................................................................. 19 Flicker Noise ............................................................................................................................................ 20 Noise Calculations ................................................................................................................................... 20 Cantilever Design ................................................................................................................................ 20 Comb Finger Design ............................................................................................................................ 20 Simulation ................................................................................................................................................... 20 Simulation Results................................................................................................................................... 21 Fabrication .................................................................................................................................................. 25 Packaging .................................................................................................................................................... 29 System Dynamics ........................................................................................................................................ 31 Future Work ................................................................................................................................................ 34 Conclusions ................................................................................................................................................. 35 Works Cited ................................................................................................................................................. 36 Appendix A: Matlab m.file ........................................................................................................................ 38

List of Figures
Figure 1: Final 3D Model ............................................................................................................................... 4 Figure 2: Parallel Plate Configuration ........................................................................................................... 9 Figure 3: Detailed Comb Fingers ................................................................................................................... 9 Figure 4: Comb Fingers Overview ............................................................................................................... 10 Figure 5: Lateral Comb Configuration ......................................................................................................... 11 Figure 6: Mechanical Suspension................................................................................................................ 12 Figure 7: Sample Electrical Circuit............................................................................................................... 13 Figure 8: Accelerometer Circuit .................................................................................................................. 14 Figure 9: Acceleration=0g ........................................................................................................................... 14 Figure 10: Acceleration=75g ....................................................................................................................... 15 Figure 11: Acceleration=150g ..................................................................................................................... 15 Figure 12: Acceleration=-75g ...................................................................................................................... 15 Figure 13: Output Voltage vs. Acceleration ................................................................................................ 17 Figure 14: Lumped Model Accelerometer .................................................................................................. 18 Figure 15: Simplified SCRV Schematic......................................................................................................... 20 Figure 16: Total Cantilever Displacement for 100g .................................................................................... 21 Figure 17: Total Cantilever Displacement for 200g .................................................................................... 21 Figure 18: Stress on a Cantilever for 100g .................................................................................................. 22 Figure 19: Stress on a Cantilever for 200g .................................................................................................. 22 Figure 20: Final Simulation Model .............................................................................................................. 23 Figure 21: Final Displacement of the Model ............................................................................................... 23 Figure 22: Final Stress Values...................................................................................................................... 24 Figure 23: Mask 1 ........................................................................................................................................ 25 Figure 24: Mask 2 ........................................................................................................................................ 26 Figure 25: Final Model ................................................................................................................................ 27 Figure 26: Comb Finger Dimensions ........................................................................................................... 28 Figure 27: Spring Dimensions ..................................................................................................................... 28 Figure 28: Packaging Process ...................................................................................................................... 30 Figure 29: Cross Sectional View .................................................................................................................. 30 Figure 30: Mass Damper Spring System ..................................................................................................... 31 Figure 31: 200g Step Response ................................................................................................................... 32 Figure 32: 200g Impulse Response ............................................................................................................. 33 Figure 33: 1g Step Response ....................................................................................................................... 33 Figure 34: 1g Impulse Response ................................................................................................................. 34 Figure 35: Accelerometer Performance...................................................................................................... 35

List of Tables
Table 1: Component Length Constraints ...................................................................................................... 6 Table 2: Design Choices ................................................................................................................................ 7 Table 3: Bending Stress [2]............................................................................................................................ 8 Table 4: Mechanical Specifications ............................................................................................................... 8

Introduction
The MEMS device our team proposed was in the category of protective sports gear. Specifically, the group designed a modified (American) football helmet to improve player health and safety. The football helmet will be modified to include embedded MEMS devices, which will be used to measure the acceleration of the user and more specifically the impact of the user during contact with another player or object in the course of the game. The MEMS device will utilize an accelerometer to calculate the impact. The data generated will be wirelessly transmitted to a computer program that will calculate the blunt force trauma resulting from a given impact using an RF module transmitter. Through statistical methods, the data center processes the collected data to diagnose potential brain injuries in real-time as they occur and prevent additional injuries in the future. The final model can be seen in Figure 1.

Figure 1: Final 3D Model

Market Value
Advances in training have led to bigger, faster players who have made the high-impact sport more dangerous, particularly at the college and professional level. The football helmet has come under scrutiny as scientists learn more about the drastic effects concussions can have on the human brain. Every season, more than one-hundred NFL players suffer concussions, which doctors have linked to depression, early onset of Alzheimers and chronic traumatic encephalopathy, a degenerative brain disease. The rate of concussions among high-school and college players (where they go unreported) is probably much higher. Every year, millions of dollars are spent on the purchase of helmets, and even higher amounts are spent on the treatment of concussions and head injuries related to on-field impacts [1].

The data that we calculate from our device will aid doctors, officials, coaches, and possibly the players themselves in diagnosing concussions. It may also lead to a more clear understanding of what types of hits are causing these issues. As the demand for faster, more efficient plays is on the rise, so is the demand for protecting the players and preventing head-related injuries.

Mechanical Design
Mechanical Model
The device can be modeled as a proof mass attached to four spring-damper segments in parallel. The proof mass is the vehicle from which a series of comb fingers extend for capacitance measurement. Damping results due to the interaction of our device motion with the surrounding air (see System Dynamics section for details). The model can be seen in Figure 1. For simplicity and in order to model the displacement of the proof mass for a given force, the damping effect of the surrounding air has been ignored in the below set of calculations. The effective stiffness of the four parallel guided beam segments can then be modeled using the below equation [2]:

Where is the Youngs Modulus of polysilicon, is the thickness of each guided beam segment, is the width of each guided beam segment, is the number of series guided beams in given segment, and is the length of the guided beam. From Hookes Law ( ), the total displacement of the proof mass is:

Where represents the force applied to the beam, which is equivalent to the mass of the proof mass, , multiplied by the acceleration imparted on it due to a given impact. Substituting in and , the displacement becomes:

Electrical Model
Our device utilizes comb fingers in order to increase the magnitude of the capacitance a given acceleration generates. The capacitance of our device can be modeled as [3]: ( )

Where is the number of comb fingers extending from each side of the proof mass, is the thickness of the proof comb fingers (which is the same as the thickness for all other device features guided 5

beams, comb fingers), is the overlap distance between the static and dynamic comb fingers attached to the proof mass, and is the gap between the dynamic and static comb fingers. In order to increase the magnitude of the capacitance for a given acceleration to allow for the detection of accelerations in the range of 200g our device maximizes the number of comb fingers, their thickness, and the while minimizing the gap between each comb finger; all while working to maintain the maximum device envelop of 1300 m. Sensitivity should be optimized in order to ensure adequate acceleration resolution [3]. Sensitivity is the derivative of capacitance with respect to the distance between an accelerometers capacitive plates. In the case of our device, the sensitivity can be modeled as: ( )

Where is the starting gap between the static and dynamic comb finger plates, and is the displacement caused by a given input acceleration. In order to optimize our device sensitivity, we maximized the capacitance change per gap change in order to achieve our target of 500mg resolution.

Geometric Design Constraints


From a geometric perspective, we sought to reduce the cost of our device through minimizing its footprint; this more devices per silicon wafer could be fabricated, reducing the unit cost of our device. The expressions in Table 1 below represent the lengths of each component making up a single side of the device.
Table 1: Component Length Constraints

Feature Mid Anchors Guided Beam Proof Mass Center Anchor End Anchor

Expression

# Features Per Side Length 8 4 2 1 2

Expression for the total length of a given side:

The teams design choices can be seen in Table 2 below.


Table 2: Design Choices

Feature Device Thickness

Description Thickness Number Beam Segments

Symbol t

Length 15 m

20 m

Guided Beams

Width Gap Between Beam Segments Length Number Comb Fingers Width

2.25 m 2 m 190 m 10 m 2 m

Rationale t < 15 m for final packaging PSG application Reduce stiffness of guided beams to achieve max comb finger displacement without pull-in and For optimal fabrication methods for optimal fabrication methods Optimize capacitance to allow detection of 200g and resolution of 500mg for optimal fabrication methods Design choice: minimize go to reduce stress during displacement, maximize displacement to improve Sensitivity Design choice: Optimize capacitance to allow detection of 200g and resolution of 500mg Design choice: Optimize capacitance to allow detection of 200g and resolution of 500mg Design choice: ensure envelop < 1.3mm Design choice: ensure envelop < 1.3mm Design choice: ensure envelop < 1.3mm Design choice: (1) ensure envelop < 1.3mm; (2) ensure mass allows adequate comb finger displacement Design choice: (1) ensure envelop < 1.3mm; (2) ensure mass allows adequate comb finger displacement Design choice: ensure mass allows adequate comb finger displacement

Comb Finger Gap, Static Comb Fingers

6 m

Length

Lf

40 m

Overlap

Lo

35 m

Mid Anchor Device Anchors End Anchor Center Anchor

10 m 60 m 394 m

Length width Proof Mass

146 m

300 m mass 1.54 m

The final device geometry can be seen in Figure 25, with the comb fingers and springs detailed in Figure 26 and Figure 27, respectively.

Stress Analysis
The yield strength of a material is the point after which plastic, permanent deformation begins to occur in that material. Once a material reaches its yield point, the risk of material failure increases as does the risk of change to the devices structural properties (e.g. stiffness) and hence its performance. Our MEMS device is fabricated from polysilicon, which has a yield strength of 1.21 GPa [4]. The primary stressed features in our MEMS device are the guided beam segments; it is therefore critically important to the performance of our device that that these features do not reach the yield point. The stiffness and max bending stress of a single fixed guided beam and cantilever beam are given in the table below. The guided beam max stress expression, at 200g force input, was utilized to understand the worst case stress to which each individual guided beam would be exposed.
Table 3: Bending Stress [2]

Single Fixed Guided Beam Stiffness Max Bending Stress Max Bending Stress in 20 Series Guided Beams

Cantilever Beam

N/A

In the equation contained in Table 3 above, is the length of the guided beam (190 m), is the thickness of the guided beam in the direction of the force (2.25 m), is the width of the guided beam in the direction orthogonal to the force (15 m), and is the force applied to the tip of each guided beam (3 N, considering the mass of the proof mass and 200G acceleration). The resulting max bending stress in each guided beam in our device final configuration is 9.0 MPa, which is 0.7% of polysilicons yield strength. This safety factor is adequate to ensure low risk of material failure during normal device operation.

Mechanical Specifications Summary


The parameters are summarized in Table 4.
Table 4: Mechanical Specifications

Parameter Range Resolution Displacement at 200g Envelop Max Stress in Guided Beam

Spec 200g 500mg 4.0 m 1218 m 9.0 MPa

Criteria Minimum 150% of 130g Minimum 500mg Maximum 4.0 m Maximum 1.3mm

< 10%

Electrical Design
MEMS capacitance configurations can be characterized into three parts; parallel plate capacitor, transverse comb capacitance configuration and lateral comb configuration.

Parallel Plate Capacitor


Figure 2 shows the parallel plate configuration. In this configuration there are two plates that are parallel to each other and the capacitance formed between these plates changes when the distance between these plates changes. This configuration is generally used to sense in the z-axis sensing schemes. This configuration may result very high sensitivity, but the main disadvantage of this configuration is the non-linear force and sensitivity values, which occurs with the movement of one of the plates towards the other [5].

Figure 2: Parallel Plate Configuration

Transverse Comb Capacitance Configuration


In this configuration there is a movable electrode between two stationary electrodes. Due to this differential topology, this configuration shows an almost linear force-displacement and voltagedisplacement relationship. In addition to linearity, this configuration gives a high sensitivity due to varying distance topology. The nonlinearity comes from fringing field capacitances and this effect is reduced if high aspect ratio fingers are used. Examples can be seen in Figure 3 and Figure 4 [6].

Figure 3: Detailed Comb Fingers

Figure 4: Comb Fingers Overview

Lateral Comb Configuration


In this configuration the capacitance forms between two combs looking each other. The change of the capacitance is formed by the change of the distance of these combs. This configuration shows a constant force-displacement characteristic, hence does not form an electrostatic spring constant, but has a very poor sensitivity due to varying overlap area topology. Due to this linearity and poor sensitivity this configuration is generally used in the actuating parts of the sensors. This configuration can be viewed in Figure 5 [6].

10

Figure 5: Lateral Comb Configuration

Capacitance Accelerometer Design


Typical MEMS accelerometer is composed of movable proof mass with plates that is attached through a mechanical suspension system to a reference frame, as shown in Figure 6. Movable plates and fixed outer plates represent capacitors. The deflection of proof mass is measured using the capacitance difference. The free-space (air) capacitances between the movable plate and two stationary outer plates C1 and C2 are functions of the corresponding displacements x1 and x2: C1 = *A* C2 = *A* where d is the initial capacitance gap. If the acceleration is zero, the capacitances C1 and C2 are equal because x1 = x2. The proof mass displacement x results due to acceleration. In case of a non-zero displacement x, the capacitance difference is found to be C2 - C1 = 2*C = 2**A* Measuring C, we can find the displacement x by solving the non-linear algebraic equation, 11 = *A* = *A* = C0 - C1 = C0 + C2

C*x2 + *A*x C*d2 = 0 This equation can be simplified. For small displacements, the term C*x2 is negligible. Thus C*x2 can be omitted.

Figure 6: Mechanical Suspension

Hence from: x * C = d*

We can conclude that the displacement is approximately proportional to the capacitance difference C. As one can see in Figure 6, every sensor has a lot of capacitor sets. All upper capacitors are wired parallel for an overall capacitance C1 and likewise all lower ones for overall capacitance C2, otherwise capacitance difference would be negligible to detect. Equation 1 now doesnt hold true just for one pair of capacitors, but for whole system. Lets take a look at how does a simplified electric circuit that measures capacitance change, look like as shown in Figure 7. Sensors fixed plates are driven by 1 MHz square waves with voltage amplitude V0 coming out of oscillator. Phases of the square waves that drives upper and lower fixed plates differs for 1800.

12

Figure 7: Sample Electrical Circuit

First of all we are interested in voltage output Vx which is actually the voltage of the proof mass. It holds true that: (Vx + V0)*C1 + (Vx - V0)*C2 = 0 We get the voltage output: Vx = V0*

= * V0

Vx is square wave with the right amplitude proportional to acceleration. We also cant just simply use this output signal, because it is weak and noisy. When there is no acceleration (a1 = 0), the proof mass doesnt move, and therefore, the voltage output is zero. If we accelerate the sensor (a1 > 0), the voltage output Vy changes proportional to alternating voltage input V0. To avoid signal attenuation, we read Vx with voltage follower (buffer), therefore signal Vy is actually Vx multiplied by 1. If we inverse the acceleration (a1 < 0), signals Vx and Vy get negative sign. Demodulator then gives us the sign of the acceleration, because it multiplies the input signal Vy with the square waves V0 coming from oscillator. Now we finally have voltage output Vout with the right sign of acceleration and the right amplitude.

Circuit Simulation Results


Figure 8 shows the circuit used to find the output voltage for different values of acceleration.

13

Figure 8: Accelerometer Circuit

Figure 9, Figure 10, Figure 11and Figure 12 are the waveforms for some typical values of acceleration which are in agreement with the waveforms of Figure 7.

Figure 9: Acceleration=0g

14

Figure 10: Acceleration=75g

Figure 11: Acceleration=150g

Figure 12: Acceleration=-75g

15

Sensitivity Analysis
Let us first take a look at the displacement sensitivity, the displacement sensitivity Sd of the device along the sensitive direction can be expressed as:

Sd =

[ [ ]

m/g

The value of Sd for our analysis comes out to be 2.02 X E^-09 m/g. We draw the following conclusions from the above equation, 1. We can conclude that the sensitivity of the folded-beam comb accelerometer is inversely proportional to the third power of the beam width Wb , that is, Sd (1/Wb3) . The device sensitivity changes rapidly with the beam width. In other words, beam width of the folded beam accelerometer is a highly sensitive parameter to adjust the sensitivity of the device. A folded beam accelerometer of desired sensitivity can be designed by adjusting beam width. Theoretically we can narrow down the beam width Wb to achieve very high device sensitivity. However, there is always a bottom limit for the beam width set by the minimum line width in a fabrication process. If the beam width is too narrow (e.g. less than 2m), it will become very challenging to fabricate the beam because the beam is extremely fragile and can be easily broken. This will greatly reduce the fabrication yield. Thus there is a tradeoff between them and we cannot shrink the beam width unlimitedly. In order for high sensitivity, we may rely on adjusting the other design parameters (such as beam length, mass width) as well. Sensitivity of the folded beam accelerometer Sd is directly proportional to the third power of beam length. That is, Sd Lb3. However, increasing the beam length will also increase the overall device area. Sensitivity of the folded beam accelerometer Sd is directly proportional to the mass width and mass length. That is, Sd Wm and Sd Lm.

2.

3.

We can also define sensitivity of a sensor by the ratio of the output voltage over input acceleration. This can be expressed as:

V/g

Using this formula the sensitivity of the accelerometer is calculated to be around 16.8mV/g. Figure 13 shows the measure of the output voltage with respect to the output acceleration i.e. is a measure of sensitivity. The equation of the straight line obtained is given by y = 0.0168x + 3E-15. From this, we obtain the sensitivity as 16.8mV/g.

16

4.000 3.500 3.000 2.500 2.000 1.500 1.000 0.500 0.000 0 50 100 150 200 250 y = 0.0168x + 3E-15 Series1 Linear (Series1)

Figure 13: Output Voltage vs. Acceleration

Pull-in
Pull-in can be examined at two different levels i.e. static and dynamic and this concept is discussed at both the levels [7]. 1. Static Pull-in: This is caused while considering that pull-in is solely due to electrostatic action. Here the inertial and the damping effects are neglected and the variation in the voltage is considered slow enough so that the equilibrium is obtained anytime by the static components. 2. Dynamic Pull-in: This is a more accurate analysis as it takes into account the inertial and the damping effects and the additional effect of external acceleration. The basic phenomenon is the loss of stability of the equilibrium position. The device under analysis follows the equilibrium principle: Finertia +Fdamping +Felastic + Felectrostatic = 0 We perform the pull-in analysis using the lumped model of the accelerometer shown in Figure 14. Our accelerometer operates in the symmetric mode, i.e for Asymmetric mode (V1=V and V2=0) and for Symmetric mode (V1=V2=V).

17

Figure 14: Lumped Model Accelerometer

In our case V1 = V2 = V0: FElectrostatic 1 = FElectrostatic 2 = Using above two equations we get: Fnet = = Using stability criteria: < (0, V0) 2*C0* This equation gives us the value of the Pull-in voltage. k < 0 V < [ ] + + k*x k

Noise
Limitations in measurement can arise from interference from other signals or by random fluctuations, collectively referred to as noise. There are several noise sources affecting the overall system resolution of an accelerometer system. Since our device has an open-loop mode of operation, the noise sources can be classified into these main categories; mechanical (Brownian) noise, capacitive noise, sensor charging reference voltage (SCRV) noise and flicker noise [8, 9].

18

Mechanical (Brownian) Noise


For a damped harmonic oscillator that is not driven, any initial motion will be damped out. But this cannot be an accurate description of reality, because in reality, this damped motion will have a noise force. This noise force arises due to the damping of the proof mass, and is directly related to the sensing structure & environment. Its spectral density function is 4kBTb, similar to Johnson noise. This Brownian noise corresponds to an equivalent mean-square acceleration noise of: an, rms =

Where kB = Boltzmanns constant, T = temperature, b = damping = 2m0, 0 = natural frequency, = damping coefficient and m = mass.

Capacitive Noise
When a capacitor is connected to a resistor that is allowed to come into contact with a thermal reservoir, the mean-square noise on the capacitor is kBT/C. This also applies to a capacitive equivalent like the spring which is coupled to a thermal reservoir by a damping mechanism [10]. The mean-square voltage in the capacitor is: = kBT/C Similarly, the mean-square force on the spring is: = kBTk

Sensor Charging Reference Voltage (SCRV) Noise


Sensor readout is performed by charging the sense capacitance with a fixed reference voltage in each cycle and detecting this charge by the interface electronics. Therefore, any noise on this reference voltage directly contributes to the overall noise performance, which is known as sensor charging reference voltage noise. Large low-frequency components of this noise can easily dominate the system noise performance. This noise can be calculated by : Vout-SCRV = Where, fS = sampling frequency, Rswitch = switch resistance, Vn = reference voltage noise, Cs = sensing capacitance and Cint = integration capacitance. A simplified version of this can be seen in Figure 15.

19

Figure 15: Simplified SCRV Schematic

Flicker Noise
Noise with a 1/f spectral density shows up in many systems and is caused by a wide variety of phenomena like Diodes, earthquakes & biological systems. The amplifier in our project has a low corner frequency so that the flicker noise does not affect our system.

Noise Calculations
Cantilever Design The mean-square acceleration noise was calculated as 27.8g/ The capacitive Noise was calculated as 88V/ . . .

The Force Fluctuation was calculated as 2.5 10-12 N/

Comb Finger Design The mean-square acceleration noise was calculated as 60.7g/ The capacitive noise for C1 was calculated as 5.61V/ The capacitive noise for C2 was calculated as 2.77V/ The Force Fluctuation was calculated as 1.0806 10-12 N/ .

Typical Value SCRV equivalent noise was calculated as 0.89V/ and Cs and Cint as 10pF & 5 pF respectively.

for a sampling frequency of 100kHz,

Simulation
The accelerometer was simulated for both cantilever and comb-finger designs, to decide what design should be used for better device performance. The simulations were done using COMSOL MULTIPHYSICS 4.2. It was also mandatory to find out the amount of stress each component of the device undergoes, because basic mathematical calculations only give the stress values for the entire device. Also the displacement for different points on the device was analyzed for different values of the G-Force. The simulations were done using the electromechanics and the solid mechanics physics in the MEMS module [11, 12]. 20

Simulation Results
The displacement was plotted for the cantilever and the maximum displacement of the tip of the cantilever for 200G concussive force was out to be 0.8068 m, which resulted in a capacitance of 14.5fF, a more accurate value considering the effects of damping forces at work; as opposed to a theoretical displacement value of 2.16 m. The displacements for 100g and 200g are in Figure 16 and Figure 17, respectively.

Figure 16: Total Cantilever Displacement for 100g

Figure 17: Total Cantilever Displacement for 200g

21

As a result of the cantilever simulation, it was determined that much higher values of capacitances could be obtained that would capture the acceleration better if the capacitance generation method was changed to a comb-finger design. Similarly, the von Mises stress values for the cantilever were calculated and were determined to be of the order of 4.3194 106 N/m2 for a 200G-Force. The stress values for 100g and 200g are in Figure 18 and Figure 19, respectively.

Figure 18: Stress on a Cantilever for 100g

Figure 19: Stress on a Cantilever for 200g

It was also determined that the stress values could be lowered for a more robust design that could handle the high acceleration values of the concussive Forces. Hence, a simple comb-finger design was simulated on COMSOL to find out the stress and the displacement values for a 200 G Concussive Force. The values for displacement were 0.0376 m for each comb-finger that resulted in higher capacitances with C1 as 20.53 fF and C2 as 20.79 pF. Using this simulation as a base to improve the design by increasing the number of comb-fingers and springs keeping the size and dimension considerations in 22

mind, the eventual design was decided (seen in Figure 20). The final results can be seen in Figure 21 and Figure 22.

Figure 20: Final Simulation Model

Figure 21: Final Displacement of the Model

23

Figure 22: Final Stress Values

24

Fabrication
After careful consideration and discussion the group agreed on the steps necessary to fabricate the device. The fabrication steps are as follows [13, 14, 15]: 1. Start with a polished SOI wafer with a 20 m device layer, a 5 m oxide layer and a 20 m substrate layer. 2. Standard RCA clean. 3. Spin coat 3 m thick standard positive photoresist and pattern with Mask 1, as seen in Figure 23.

Figure 23: Mask 1

25

4. 5. 6. 7.

DRIE the exposed silicon down to the oxide. Strip the photoresist with oxygen plasma ash. Standard RCA clean without HF dip. Deposit 1 m on the anchor in the middle and the outside with Mask 2, as seen in Figure 24.

Figure 24: Mask 2

26

8. 9. 10. 11.

Perform photolithography. DRIE the exposed silicon down 5 m. Strip the photoresist with oxygen plasma ash. Standard RCA clean without HF dip.

The final model can be seen in Figure 25 with the comb fingers and springs detailed in Figure 26 and Figure 27, respectively.

Figure 25: Final Model

27

Figure 26: Comb Finger Dimensions

Figure 27: Spring Dimensions

28

Packaging
Packaging is very important for MEMS devices. Reliability, performance of a MEMS device largely depends on packaging. With contrast to IC packaging there is no standard way for MEMS packaging. There are some constraints that make it difficult to pack a MEMS accelerometer. Those are: Moving parts of accelerometer Stiction between movable parts Access of moisture

In order to eliminate those drawbacks, a package should: Provide protection and be robust enough to withstand its operating environment Be hermetic Dissipate generated heat Minimize stress from external loading Handle power from electrical connection leads without disruption Be cost effective

We choose wafer-level-thin-film packaging method to pack our device. This method is cost effective, simple, minimizes the device and protective for the moving parts. Device is fabricated, yet we have not released and die sawed it. Packaging process is carried out in vacuum environment. The packaging process is as follows [16]: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Deposit 20m PECVD thick oxide layer on top of the device. Deposit 2m LPCVD thin oxide layer to form the etch hole and bonding pad sides. The surface is not smooth, so polish the oxide layer to get very smooth, flat surface. Deposit 2m LPCVD silicon nitride. Coat 1m positive photoresist with photolithography mask and then pattern it. This will be the etch hole. Etch a hole through silicon nitride with DRIE. O2 plasma ashing the photoresist Release the oxide layer through the etch hole with HF dip. CO2 critical point drying Deposit 2m LPCVD silicon nitride over the silicon nitride layer to plug the etch hole, make it sturdy and hermetic. Coat 1m positive photoresist with photolithography mask and then pattern it. This will be the bonding pad areas. Dry etch through silicon nitride to expose bonding pads. O2 plasma ashing the photoresist Apply bumps of solder to bonding pads. 29

15. Separate wafer into dies. 16. Flip the package over onto the PCB, and solder by applying heat. The steps in this process can be seen in Figure 28 and the cross section of the final device can be seen in Figure 29.

Figure 28: Packaging Process

Figure 29: Cross Sectional View

30

System Dynamics
An accelerometer can be represented as mass-damper-spring system shown in Figure 30 below.

Figure 30: Mass Damper Spring System

From physics: We obtain the equation of motion for the system:


) and

We obtain the equation of motion of the system in steady-state (when there is no velocity ( acceleration ( )), equation of motion reduces to: After some steps:

Natural frequency is constant for the system. In that point we can define the range of displacement and acceleration. We obtain natural frequency of the system. This is the mechanical constraint of designing an accelerometer. Then we can define k and m parameter along with the other constraint (i.e. fabrication, electrical) to optimize the system performance. According to our research results a concussion occurs at 130g acceleration. For safety issues we would like the design our device to handle up to 200g acceleration. We also design our device such that it moves up and down. We decided maximum displacement to be 4 m. 31

To plot the system response, one way is to use Laplace transform to obtain transfer function between output of the system ( ) and input to the system( ). Assuming zero initial conditions:

We can now plot the linear time response of the system using available engineering software. We use Matlab to simulate the system response. The file can be viewed in Appendix A: Matlab m.file. As it is seen from the step response plot maximum displacement reaches 4m for the 200g acceleration along with the chosen and values. The results can be viewed in Figure 31, Figure 32, Figure 33 and Figure 34.

Figure 31: 200g Step Response

32

Figure 32: 200g Impulse Response

Figure 33: 1g Step Response

33

Figure 34: 1g Impulse Response

Future Work
With additional time and resources, our group would increase the breadth and depth of our mechanical, electrical, fabrication, and packaging optimization to improve on performance and costs in an effort to position our device for success on the open market. From a mechanical design perspective, additional work would be put in to ensure the device size has been fully minimized, while still allowing for the acceleration range/sensitivity measurement capabilities and strength properties of the existing device. This would ensure our device achieved the lowest possible cost for the performance required. From a fabrication and packaging perspective, additional work would be conducted to evaluate the effect of fabrication stresses and tolerances on the performance of our device. Our current device performance is defined by the stiffness of the guided beam segments, proof mass volume, and comb finger dimensions. Each of these critical design features will not end exactly as specified due to inherent variation in the fabrication and packaging processes. Linear dimension inaccuracies, thickness undercutting, and thermal stresses will result in differences in the actual dynamic behavior of the device compared to the model. This variation must be planned and corrected for through a calibration process following device packaging (e.g. electrical hardware trimming). This calibration/tuning process would be investigated further by our group to ensure the final performance of our device [17]. Finally, our group would investigate the implementation of a gyroscope feature as part of the center anchor of the current device in order to allow for angular acceleration detection. Although the three34

axis linear acceleration detection designed into our current device is adequate for the purpose of monitoring head trauma, a gyroscope feature would add a means for understanding a second form of motion and its impact on sports injury. The present device (a capacitive accelerometer) is more cost effective and accurate as compared to other devices. It also less prone to noise and variations with temperature, typically dissipates less power, and can have larger bandwidths due to internal feedback circuitry. Figure 35 shows that there are comparative advantages between capacitive accelerometers and piezoelectric accelerometers. The present device offers adequate range and sensitivity for the purposes of our application but the group would have liked to have the opportunity to explore piezoelectric accelerometers to see if another design could have been optimized.

Figure 35: Accelerometer Performance

Conclusions
The present device is more than adequate to estimate the concussive forces. With the integration of devices like this, the group hopes that there are continuing advances in concussion research. Measurement like this will prove invaluable to solving the puzzle that concussions present to the NFL, other sports organizations, and the public.

35

Works Cited
[1] R. C. Johnson, "High-G MEMS help detect concussions," EE Times, 24 May 2012. [Online]. Available: http://www.eetimes.com/electronics-news/4373812/High-G-MEMS-help-detect-concussions. [Accessed 11 December 2012]. [2] C. Livermore, Composer, ME6260/EECE7244, Lecture 4: Stress, Strain, and Structures. [Sound Recording]. Northeastern University. 2012. [3] C. Livermore, Composer, ME6260/EECE7244, Lecture 10: Introduction to Electrostatic Sensors and Actuators. [Sound Recording]. Northeastern University. 2012. [4] W. N. Sharpe, "New Test Structures and Techniques for Measurement of Mechanical Properties of MEMS Materials," SPIE, vol. 2880, p. 78, 1996. [5] K. Sharma, I. Macwan, L. Zhang, L. Hmurcik and X. Xiong, "Design Optimization of MEMS Comb Accelerometer," University of Bridgeport, Bridgeport, CT, 2009. [6] M. Andrejaic, "MEMS Accelerometers," Andrejaic, 2008. [7] A. Kannan, "Design and Modeling of a MEMS-Based Accelerometer with Pull In Analysis". [8] S. Senturia, Microsystem Design, Dordrecht: Kluwer Academic Publishers, 2009. [9] C. Livermore, Composer, ME6260/EECE7244, Lecture 21: Noise. [Sound Recording]. Northeastern University. 2012. [10] L. K. Baxter, Capacitive Sensors, Piscataway: IEEE Press, 1997. [11] A. Gupta , M. Alagappan, S. Vijayakumar and G. Vijila, "Design and Analysis of 3D Capacitive Accelerometer for Automotive Applications," PSG College of Technology, Tamil Nadu, India, 2011. [12] H. Kulah, J. Chae and K. Najafi, "Noise Analysis and Characterization of a Sigma-Delta Capacitive Microaccelerometer," University of Michigan, Ann Arbor, 2003. [13] Y. P. Wang, R. Q. Hsu and C. W. Wu, "Design and Test of a MEMS-Based High G Smart Sensor," IEEE Sensors Journal, vol. 11, no. 4, 2011. [14] M. Hrairi and B. H. Baharom, "Design and Modeling of Silicon MEMS accelerometer," in International Conference on Mechanical, Automotive and Aerospace Engineering, 2011. [15] J. Voldman, Composer, Case Study: A Capacitive Accelerometer. [Sound Recording]. Massachusetts

36

Institute of Technology. 2007. [16] D. F. Guillou, "Packaging MEMS: New Manufacturing Methodology Substantially Reduces Smart MEMS Costs," Sensors Mag, December 2003. [Online]. Available: http://archives.sensorsmag.com/articles/1203/20/main.shtml. [Accessed 11 December 2012]. [17] C. Livermore, Composer, ME6260/EECE7244, Lecture 22: Packaging for MEMS. [Sound Recording]. Northeastern University. 2012.

37

Appendix A: Matlab m.file


clear clc

gforce = 200; amax = gforce*9.8; %maximun acceleration xmax = 4e-6; %maximum displacement wn_sys = sqrt(amax/xmax) %Natural frequency of the system for 200g % acceleration and 4m displacement E = 150e9; %Elasticity modulus of PolySi Ns = 20; %Number of guided_spring Lb = 190e-6; %Length of each guided_spring tb = 15e-6; %Thickness of guided_spring wb = 2.25e-6; %Width of guided_spring k = 4*E*tb*wb^3/(Lb^3*Ns); %Spring constant of the system

Lm = 300e-6; %Length of proof mass wm = 146e-6; %Width of proof mass tm = 15e-6; %Thickness of proof mass rho = 2300; %Density of PolySi mt = rho*Lm*wm*tm %Mass of the proof mass me = 200*pi*(1.5e-6)^2*15e-6*rho;%Mass of the etch holes ma = rho*(20*40*2*15e-18+2*16*10*15e-18); %Mass of the anchors around mass m = mt-me+ma %Total mass

Wn = sqrt(k/m) %Natural frequency of the system 38

ksi = 1; %Damping ratio (critically damped system) b = 2*m*Wn*ksi; a = amax; num = 1; den = [1 b/m k/m]; sys = tf(num,den) %Transfer function of the system

figure(1) [x1,t1] = step(a*sys); step(a*sys) %Step response of the system text(max(t1)/2,max(x1)/2,strcat(num2str(gforce),'g acceleration'))

figure(2) [x2,t2] = impulse(1e-6*a*sys); impulse(1e-6*a*sys) %Impulse response of the system text(max(t2)/2,max(x2)/2,strcat(num2str(gforce),'g acceleration'))

39

Вам также может понравиться