Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=titr20

Download by: [Gazi University] Date: 15 January 2016, At: 08:24

IETE Technical Review

ISSN: 0256-4602 (Print) 0974-5971 (Online) Journal homepage: http://www.tandfonline.com/loi/titr20

Design Issues of Piezoresistive MEMSAccelerometer for an Application Specific MedicalDiagnostic System

Sonali Biswas & Anup Kumar Gogoi

To cite this article: Sonali Biswas & Anup Kumar Gogoi (2016) Design Issues of PiezoresistiveMEMS Accelerometer for an Application Specific Medical Diagnostic System, IETE TechnicalReview, 33:1, 11-16, DOI: 10.1080/02564602.2015.1065713

To link to this article: http://dx.doi.org/10.1080/02564602.2015.1065713

Published online: 20 Jul 2015.

Submit your article to this journal

Article views: 23

View related articles

View Crossmark data

Design Issues of Piezoresistive MEMS Accelerometer for an Application SpecificMedical Diagnostic System

Sonali Biswas and Anup Kumar Gogoi

Department of Electronics and Electrical Engineering, Indian Institute of Technology, Guwahati, India

ABSTRACTThis paper reports the design and simulation of finite element method (FEM)-based micro-electro-mechanical-systems (MEMS) piezoresistive accelerometer used in a strapdown medical diagnosticsystem. A square proofmass with four flexures is used in this design. Here, p-doped single-crystalsilicon are used as eight piezoresistors implanted on the flexures for sensing the maximum stress;and they are connected in the form of a wheatstone bridge with each arm containing twopiezoresistors. Basically, here two different geometric configuration but of same material have beendesigned and simulated using FEM-based tool for dynamic range (§)2g. To assure betterperformance characteristics, different analysis for both the configurations are done. Configuration 1 isfound to have better performance with respect to load withstanding capability, eigen value analysis,better response characteristic, and low cross-axis sensitivity. Furthermore, the measurement schemeadopted is the wheatstone bridge, which gives better performance as offset is almost nullified.

KEYWORDSCross-axis sensitivity; Eigenvalue; MEMS piezoresistiveaccelerometer

1. Introduction

Micromachined MEMS accelerometers find its applica-tion from a simple vibration monitoring system to acomplex inertial navigation system. Demand for low-cost, high-resolution, and high-sensitivity microma-chined accelerometers have been increasing to cater tothe need in various fields. The performance factors ofMEMS accelerometers depend on several aspects of geo-metric design, device sensing mechanism, scaling limita-tions, associated electronic circuits, and fabricationaspects. There are different transduction mechanismslike capacitive, piezoelectric, and piezoresistive [1].Stress-based measurement generally use piezoresistiveand piezoelectric whereas displacement-based measure-ments usually use capacitive. Among the many techno-logical alternatives available, the advantages ofpiezoresistive types are that they have a DC response,they have simple read-out circuits and are capable ofhigh sensitivity and reliability. They utilize simple prin-ciple of transduction and microfabrication as comparedto other types. This paper focuses on the modelling andsimulation of two different types of piezoresisitive accel-erometer configurations. A relative comparison of boththe structures are done and the simulated results of eigenmode analysis, time-dependent analysis, and sensitivityanalysis are given. This comparison would help in opti-mizing the design structure and solve various issuesrelated to stress, doping concentration of the piezoresis-tors, reduction in cross-axis sensitivity, etc. Hence, with

enhanced sensitivity and performance of the MEMSaccelerometer, it can be made suitable to be used in astrapdown diagnostic system for the diagnosis of variousneurodegenerative disorders. Neurodegenerative essen-tially means progressive loss of structure or function ofneurons [2]. The main symptoms are tremor of bodyparts, rigidity, bradykinesia, postural instability, etc.None of the pathologies that cause tremor are fullyunderstood, and epidemiological and neurophysiologicalstudies are hampered by the lack of diagnostic methodsother than purely clinical. Certain literature reveal theloss of dopamine to be one of the main causes of tremor.Different types of neurodegenerative diseases wheretremor occurs are Parkinson, essential tremor, epilepsy,stroke, etc. There are different pathological tremor likerest tremor (3�6 Hz), kinetic tremor (4�12 Hz), andpostural tremor (2�7 Hz) [2]. Tremor is a roughly sinu-soidal oscillatory movement. The non-stationery fea-tures of tremor demands sensitive, reliable, low-cost andstable sensor mandatory. Among the different MEMSsensors available, accelerometer possess desirable fea-tures suitable for emerging applications in biomedicalapplications. MEMS accelerometers respond to fre-quency and intensity, measure tilt and body motion, andminiaturization has made batch fabrication and low-costpossible. As tremor carries social and psychological bur-den, hence an attempt has been made to design and sim-ulate an optimized MEMS accelerometer having abandwidth ranging from 0.1 to 25 Hz with a maximum

© 2015 IETE

IETE TECHNICAL REVIEW, 2016VOL. 33, NO. 1, 11�16http://dx.doi.org/10.1080/02564602.2015.1065713

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016

acceleration amplitude of 2 g. Hence, a relative compari-son of two different configuration of piezoresistiveMEMS accelerometer has been done in order to developa final optimized structure for housing it in the strap-down tremor diagnostic system.

2. Theoretical background on piezoresistiveeffect

The piezoresistive effect is a change in the electricalresistivity of a semiconductor or metal when mechanicalstrain is applied. The resistance value of a resistor isdetermined both by the bulk resistivity and the dimen-sion as given by R = rl/A, where R is the resistivity, r isthe bulk resistivity, and l and A stands for length andarea, respectively. The resistivity changes are seen to bemore dominant in comparison to dimensional changesin semiconductor materials and this have led to its usagein applications involving stress sensitivity. The piezore-sistivity coefficients of a doped single-crystal silicon pie-zoresistor are influenced by its relative orientation tocrystallographic directions. For a rectangular coordinatesystem, there exist a symmetric resistivity matrix relatingthe electric field components Ei and current densitycomponent ii as given by [3]

ExEyEz

24

35 ¼

r1 r6 r5r6 r2 r4r5 r4 r3

24

35£

ixiyiz

24

35:

Thus , matrix E D ranisotropic £ current density. Here,ranisotropic is the resistivity of the semiconducting materi-als. Also, applied stress on such anisotropically affectedmaterials may be expressed as a matrix with six indepen-dent stress components corresponding to its cause andeffect direction as given by

s ¼s1 s6 s5

s6 s2 s4

s5 s4 s3

24

35:

The relative change in resistivity can be expressed as theproduct of piezoresistive coefficient matrix and the stressmatrix. The piezoresistive coefficients of single-crystalsilicon are not constants but are influenced by the dop-ing concentration, type of dopant, and temperature ofthe substrate. Different piezoresistive coefficients getaffected differently by temperature and doping concen-tration. The piezoresistive coefficient for the piezoresis-tors along the (110) direction is given by 1/2[p11+p12+p44] [3]. For single-crystal silicon, differentpiezoresistivity components exist for different dopingconcentrations [4].

3. Proposed pathological tremor diagnosticsystem

The proposed tremor diagnostic system consists of thehigh-performance triaxial MEMS accelerometer, signalconditioning unit and data acquisition unit all housedinside the mobile unit, and the hospital unit consistingof the presentation analysis and storage unit. The blockdiagram of the pathological tremor diagnostic system isshown in Figure 1.

4. Design configuration of MEMS accelerometerand working

The MEMS accelerometer design chosen is a simplemass spring device. Here, two configurations aredesigned and simulated using COMSOL 4.3. Designconfiguration 1 consist of a proofmass, four flexures oneither side of the proofmass aligned with its edge andframe on single-crystal silicon. Here, p-type single-crys-tal silicon (110) piezoresistors are implanted at two dif-ferent locations, one near the proofmass and the othernear the frame. Piezoresistors are positioned keeping inmind that it senses the maximum stress and gives themaximum deflection in the desired direction. Designconfiguration 2 is chosen to be of same geometric sizebut the flexures are placed in position non-aligned withthe proofmass edge. Both the accelerometer design con-figuration structures with all its parts have been simu-lated and the 3D view is shown in Figures 2 and 3 below.

The MEMS accelerometers have piezoresistorsimplanted on their beams. Whenever there is proofmassdisplacement, the suspension beams deflect and thiscauses strain on the piezoresistors. This results in changein resistivity as explained in Section 2 and thus, the sig-nal transduction occurs which can be picked up by anyelectronic circuit. Each suspension beam contains twopiezoresistors therefore out of total eight piezoresistorsfour suffer from tension when the other four suffer fromcompression. It is seen that compressive stress increasesthe resistance value and due to tensile stress the resis-tance value increases. The piezoresistors are connectedin the form of a wheatstone bridge as this type of

Figure 1: Block diagram of the proposed pathological tremordiagnostic system.

12 S. BISWAS AND A.K. GOGOI

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016

connectivity gives us a reduced cross-axis sensitivity [5].The change in the output voltage of the wheatstonebridge is proportional to the applied acceleration.

Eigen value analysis is very important as it is the basis ofmany dynamic response analysis. Here modal analysishas been done for an overall understanding of the naturalfrequencies and the modal shapes. Here, the first six eigenfrequencies are simulated for both the structures of designconfigurations 1 and 2 as shown in Figures 4�9. Theeigen frequency value are found to be higher in case ofconfiguration 1 as compared to configuration 2 as shownin Table 1. This shows that structure 1 has safer zone andbetter performance even if the frequency range of opera-tion of the device is increased to several orders. Also,higher the natural frequency the more the linear is theoutput-to-input characteristic and higher the frequency itcan measure. Here, the eigen modes 1, 3, and 5 for bothconfigurations 1 and 2 are plotted.

5. Time-dependent analysis

The MEMS accelerometer has been simulated for variousdisplacement 2 g. Sinusoidal input is given and point

evaluation at centre of the mass as well as various otherpoints of the structure has been observed for both config-urations 1 and 2. The structures are taken and point eval-uation of displacement, velocity, and acceleration for bothhave been simulated. Also stress analysis have been donefor both the structure. The simulated results of the dis-placement characteristic and the von Mises stress for boththe structures are shown in Figures 10�13. It has beenobserved that the total displacement is higher in configu-ration 1 than in configuration 2. Also, the von Mises stressshow that configuration 1 can withstand better loadingcondition though both are of the same material.

Figure 2: Accelerometer structure1 simulated using COMSOL 4.3.

Figure 3: Accelerometer structure 2 simulated using COMSOL 4.3.

Figure 4: Sensor deflection at eigen mode 1 for configuration 1.

Figure 5: Sensor deflection at eigen mode 3 for configuration 1.

Figure 6: Sensor deflection at eigen mode 5 for configuration 1.

IETE TECHNICAL REVIEW 13

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016

6. Sensitivity analysis

Sensitivity is determined by the relative change in outputper unit voltage per applied differential acceleration. Formaximum sensitivity, the accelerometer must have max-imum prime axis sensitivity and small cross-axis

sensitivity [5]. The cross-axis sensitivity shows howmuch perpendicular acceleration or inclination is cou-pled to the signal. Some of the factors which affect the

Figure 7: Sensor deflection at eigen mode 1 for configuration 2.

Figure 8: Sensor deflection at eigen mode 3 for configuration 2.

Figure 9: Sensor deflection at eigen mode 5 for configuration 2.

Table 1: Eigen frequencies for six different modes for configurations 1 and 2

Modes 1 2 3 4 5 6

Config1, eigen freq(Hz) 2412.68 3922.97 5446.59 25,636.50 1.119e5 1.23e5Config2, eigen freq(Hz) 2374.64 2427.18 5467.72 26,228.96 83,805.51 1.15e5

Figure 10: Total displacement in configuration 1.

Figure 11: Total displacement in configuration 2.

Figure 12: von Mises stress observed in configuration 1.

14 S. BISWAS AND A.K. GOGOI

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016

sensitivity of an accelerometer include the geometricconfiguration of the device mainly the positioning of thepiezoresistors along with its length, cross-sectional area,and the doping concentration. Also, the measurementscheme associated for the signal pick-up must be able tonullify the offset and hence enhance the overall perfor-mance of the device. In this case p-type-doped piezore-sistors is taken and the flexure thickness is minimum soas to get the largest resistance change, hence the largestsensitivity. The measurement scheme chosen here is thewheatstone bridge circuit. To achieve low cross-axis sen-sitivity along a particular direction, stiffness should behigh along that direction. Stiffness depends upon a num-ber of flexures, geometric structure, and position of flex-ures [6]. In this case, the stiffness is high along the x-axis, and as the proofmass is parallel to the x-axis, thecross-axis sensitivity is low. Also, by placing the flexuresin-line with the proof mass edges, the y-cross-axis sensi-tivity can be improved. In configuration 2, it is seen thatthe distance between the flexure is less compared to con-figuration 1. When the distance between the flexures isincreased, the stiffness along the y-axis is found toimprove which in turn reduces the cross-axis sensitivity.Thus, configuration1 has better performance in case ofcross-axis sensitivity, giving maximum output voltagefor the z-axis acceleration.

7. Conclusion

The piezoresistive MEMS accelerometer of configura-tions 1 and 2 have been modelled and simulated for the

triaxial sensing of tremor used for medical diagnosticsystem. A relative study of both the structures have beendone and the results relating to eigen values, time-dependent analysis, stress analysis, and sensitivity-basedstudies have been done and simulated. Also, study hasbeen done to get low off-axis sensitivity and wheatstonebridge measurement scheme to enhance the perfor-mance of the device has been chosen. The stress distribu-tion is not same everywhere and maximum stress occursat the edges which gives an idea where to place the pie-zoresistors for maximum sensitivity. Thus, the study andsimulation results show that configuration 1 where theflexures are aligned with the proof mass edge is a betterdesign structure in terms of performance as comparedto configuration 2. Also, a similar type of structure as in[5] has a sensitivity of 6.5 mili volts/g as compared tothe given configuration 1 here which has a calculatedsensitivity of 10.5 mv/V/g. The idea is to fabricatethe three individual single accelerometer and finally bycomplex packaging technique giving a triaxialaccelerometer.

Acknowledgements

The authors would like to thank Indian Institute of Technol-ogy, Guwahati for carrying out the work.

References

1. N. Yazdi, F. Ayazi, and K. Najafi, “Micromachined inertialsensors,” Proc. IEEE, Vol. 86, pp. 1640�59, Aug. 1998.

2. L. J. Findley, and W. C. Koller, Handbook of Tremor Disor-ders. New York, NY: M. Dekker, 1995.

3. C. Liu, Foundation of MEMS. Upper Saddle River, NJ:Prentice Hall, 2011.

4. Y. Kanda, “Piezoresistance effect of silicon,” Sensors Actua-tors, Vol. A28, pp. 83�91, July 1991.

5. R. Mukhiya, A. Adami, A. Bagolini, M. Zen, and S. Kal,“FEM based design and simulation of bulk micromachinedMEMS accelerometers with low cross axis sensitivity,” inProceedings of the Seventh International Conference onThermal, Mechanical and Multiphysics Simulation andExperiments in Micro-Electronics and Micro-Systems,Como, Italy, pp. 1�5, Apr. 2006.

6. A. R. Sankar, J. G. Jency, J. Ashwini, and S. Das, “Realiza-tion of silicon piezoresistor accelerometer with proof massedge aligned flexures using wet anisotropic etching,” IETMicro Nano Lett., Vol. 7, no. 2, pp. 118�21, Feb. 2012.

Figure 13: von Mises stress observed in configuration 2.

IETE TECHNICAL REVIEW 15

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016

Authors

Sonali Biswas has obtained her BEdegree in instrumentation engineeringfrom Dibrugarh University, India andMTech in electronics design and tech-nology from Tezpur University, India.She is working as an assistant professorin Jorhat Engineering College, DibrugarhUniversity. She is currently pursuingPhD in the Department of Electronics

and Electrical Engineering, Indian Institute of Technology,Guwahati. Her research interest include instrumentation andmeasurements, MEMS devices, digital circuit design etc.

E-mail: sonali.jec@gmail.com.

Prof. Anup Kumar Gogoi has joinedIndian Institute of Technology in theyear 1996. He is currently working as aprofessor in the Department of Electron-ics and Electrical Engineering, IndianInstitute of Technology, Guwahati. He isworking in varied areas of electronicsand electrical engineering. His researchinterest areas mainly include electro

magnetics, microwave engineering, RF circuits, system designetc.

E-mail: akg@iitg.ernet.in.

16 S. BISWAS AND A.K. GOGOI

Dow

nloa

ded

by [

Gaz

i Uni

vers

ity]

at 0

8:24

15

Janu

ary

2016