Scholarly article on topic 'Evaluation and modeling of adhesion layer in shock-protection structure for MEMS accelerometer'

Evaluation and modeling of adhesion layer in shock-protection structure for MEMS accelerometer Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Microelectronics Reliability
OECD Field of science
Keywords
{Shock-protection / MEMS / Accelerometer / "Multi-layer metal" / "Multi-physics simulation"}

Abstract of research paper on Materials engineering, author of scientific article — Daisuke Yamane, Toshifumi Konishi, Teruaki Safu, Hiroshi Toshiyoshi, Masato Sone, et al.

Abstract This paper presents evaluation and modeling of adhesion layer in shock-protection structure for a MEMS accelerometer fabricated by the multi-layer metal technology. The shock-protection structure is used to limit the proof-mass displacement at the input of excess acceleration. To clarify the mechanical characteristics of the shock-protection structure, we investigate the relationship between the input acceleration and the adhesion force on the adhesion layer in the multi-layer metal structure. For the multi-physics simulation, we propose an equivalent circuit model for a shock-protection module combined with the acceleration module. The adhesion force on the adhesion layer is examined by tensile tests in order to obtain the simulation parameters. The multi-physics simulation results are consistent with the measured capacitance-acceleration characteristics. The transient and capacitance-acceleration simulations at the input of high acceleration are also demonstrated to show the non-linear behavior of the shock-protection structure. Therefore, it is confirmed that the proposed module is useful for the multi-physics simulation to analyze the shock tolerance.

Academic research paper on topic "Evaluation and modeling of adhesion layer in shock-protection structure for MEMS accelerometer"

ARTICLE IN PRESS

MR-12216; No of Pages 7

Microelectronics Reliability xxx (2016) xxx-xxx

Contents lists available at ScienceDirect

Microelectronics Reliability

journal homepage: www.elsevier.com/locate/microrel

Evaluation and modeling of adhesion layer in shock-protection structure for MEMS accelerometer

Daisuke Yamane a'*, Toshifumi Konishic, Teruaki Safuc, Hiroshi Toshiyoshic, Masato Sone a, Kazuya Masu a, Katsuyuki Machida a,b

a Tokyo Institute ofTechnology, Yokohama, Kanagawa 226-8503, Japan b NTT Advanced Technology Corporation, Atsugi, Kanagawa 243-0124, Japan c The University of Tokyo, Meguro, Tokyo 153-8904, Japan

ARTICLE INFO

ABSTRACT

Article history: Received 6 June 2016

Received in revised form 5 September 2016 Accepted 27 September 2016 Available online xxxx

Keywords:

Shock-protection

Accelerometer Multi-layer metal Multi-physics simulation

This paper presents evaluation and modeling of adhesion layer in shock-protection structure for a MEMS accelerometer fabricated by the multi-layer metal technology. The shock-protection structure is used to limit the proof-mass displacement at the input of excess acceleration. To clarify the mechanical characteristics of the shock-protection structure, we investigate the relationship between the input acceleration and the adhesion force on the adhesion layer in the multi-layer metal structure. For the multi-physics simulation, we propose an equivalent circuit model for a shock-protection module combined with the acceleration module. The adhesion force on the adhesion layer is examined by tensile tests in order to obtain the simulation parameters. The multi-physics simulation results are consistent with the measured capacitance-acceleration characteristics. The transient and capacitance-acceleration simulations at the input of high acceleration are also demonstrated to show the non-linear behavior of the shock-protection structure. Therefore, it is confirmed that the proposed module is useful for the multi-physics simulation to analyze the shock tolerance.

© 2016 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

MEMS (microelectromechanical systems) accelerometers [1] can be a key device for motion sensing in loT (Internet of Things) technology, and would realize a variety of future applications by increasing the sensing range and resolution [2]. Thus, we have proposed an integrated MEMS accelerometer as schematically shown in Fig. 1 [3]. High density of gold has substantially reduced the thermal-mechanical noise in the moving parts of mechanical structures [4]. Highly-sensitive proof masses with different sensing ranges could be implemented on a single chip [3]. We have also experimentally confirmed that such a highly-sensitive MEMS accelerometer fabricated by the multi-layer metal technology [5,6] showed the potential of shock tolerance of up to 20 G (1 G = 9.8 m/s2) without mechanical failure [3]. The target sensing range of the proposed accelerometer is from micro G (< 1 mG) to 20 G, where the maximum target acceleration meets the requirement of typical commercial MEMS accelerometer [7]; the minimum target acceleration covers the range where technology challenges still remain [2]. In practical use, it would be required to investigate the behavior of the MEMS structure when exposed to higher acceleration. The requirement of shock tolerance in MEMS accelerometers depends on the target

* Corresponding author. E-mail address: yamane.d.aa@m.titech.ac.jp (D. Yamane).

applications [8,9]; consumer electronics and automotive uses would need shock tolerance of around 1000 G-10,000 G [10-12], while harsh environment applications may require over 10,000 G [13-15]. So far, several analyses have been reported for the shock reliability of MEMS accelerometers [9,16-23]. Among various MEMS analysis methods, multi-physics simulation [24] on an electrical circuit simulator would be a promising way to seamlessly analyze mechanical and electrical components. The advantages of multi-physics simulation over other methods are as follows [24]: (i) mechanical and electrical behaviors of MEMS are analyzed by using a single simulation platform, (ii) the platform can be implemented on SPICE-based circuit simulators that enable us to design MEMS and LSI (large scale integrated circuits) simultaneously, (iii) photo-mask layout generation is possible for both MEMS and LSI, and (iv) linear and non-linear behaviors of MEMS can be analyzed. Nevertheless, shock-tolerance analysis of MEMS accelerometer has not been reported in the multi-physics simulation platform. Shock tolerance is one of the requirements for MEMS reliability evaluation [25,26]. In MEMS accelerometers, shock-protection structures are often employed to limit the proof-mass displacement at the input of excess acceleration, and hence its behavior is a primary concern for the shock tolerance of MEMS accelerometers. In our previous work, we have proposed the multi-layer metal shock-protection structures [3]. For the design of shock-tolerance characteristics, we need to develop the analytical model of the shock-protection structures.

http: //dx.doi.org/10.1016/j.microrel.2016.09.018

0026-2714/© 2016 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ARTICLE IN PRESS

D. Yamane et al. / Microelectronics Reliability xxx (2016) xxx-xxx

Fig. 1. Schematic image of integrated MEMS accelerometer.

In this work, to analytically simulate the behavior of the shock-protection structures, we propose an equivalent circuit model for a shock-protection module to be implemented in the multi-physics simulation platform. In this model, we consider that the direction of the input acceleration is orthogonal to the substrate of the MEMS. We assume that the adhesion force on the adhesion layer has a major role in the mechanical characteristics of the MEMS structure. Multi-physics simulation is then carried out to show the electrical capacitance change of the accelerometer as a function of input acceleration.

2. Shock-protection module for multi-physics simulation

2.1. Shock-protection structure

Fig. 2 shows the design concept of the MEMS capacitive accelerom-eter developed by using the multi-layer metal technology. Input acceleration is detected by the capacitance change between the proof mass and the fixed electrode. The displacement of the proof mass is limited by the shock-protection structures when excess acceleration is applied. The chip photo of the actual device is shown in Fig. 3(a). The shock-protection structures are implemented at the corners of the proof mass. Fig. 3(b) shows the close-up view of one of the shock-protection structures fabricated by the multi-layer metal technology, where adhesion force on the adhesion layer has a large influence on the mechanical characteristics. Accordingly, in this study, we focus on the adhesion force for the crucial simulation parameter of the shock-protection structure.

2.2. Equivalent circuit model

To analyze the mechanical tolerance of the shock-protection structures, we proposed an equivalent circuit model that handles the adhesion force on the multi-physics simulation as shown in Fig. 4. The

Fixed electrode

Fig. 2. Design concept of the MEMS capacitive accelerometer.

module includes the adhesion force (Fs) on the metal adhesion layer; Fs is defined by the product of the anchor area (Ss) and the adhesion strength (Ps) as follows:

Fs — Ps X Ss.

The multi-physics simulation is performed on a circuit simulation platform (Cadence Virtuoso) to handle both the MEMS accelerometer and the LSI. We used Verilog-A compatible HDL (hardware description language) to analyze the behavior of the equivalent circuit model of the MEMS device, as reported in our previous study [24,27]. In this work, the shock-protection module was newly implemented. Moreover, we have modified the EOM (equation of motion) module to send calculated acceleration and total input force to the shock-protection module, and to receive the velocity and displacement of the proof mass from the module. According to the circuit operation manner, these parameters are translated as voltage or current, even though the parameters have their own SI unit such as the acceleration m/s2, mass kg, velocity m/s, and displacement m.

2.3. Shock-protection module

The details of the Verilog-A expression of the shock-protection module is summarized in Appendix A. that uses seven formula sequences coded for a MEMS accelerometer. Fig. 5 shows a device schematic image with the parameters used in the module. In the first formula sequence (A.1) we initialize the mechanical and electrical parameters summarized in Table 1. After the initializing step, in the second formula sequence (A.2), breaking of the shock-protection structures is judged with if-then clauses. Here, we define the direction from the proof mass to the bottom fixed electrode as positive, thus the syntax if( — 1 * Ftotal > Fs) shows the comparison of Fs with Ftotal that is an input force in the direction from the proof mass to the shock-protection structure. When the shock-protection structure is supposed to be broken, the condition is given by

F total - Fs,

where Ftotal is equal to or larger than Fs. If the condition (2) is true, then the flag parameter sb is set to be one.

The third formula sequence (A.3) resets the integration of the acceleration to calculate the velocity by using the Verilog-A function idt at the shock-protection broken condition. A logical flag viOinteg is set to be one when the displacement of the proof mass V(xs) comes to the mechanical value limit in the positive direction or limit_m in the negative direction. Here, the distance between the proof mass and the shock-protection structure is defined as the parameter stopperpos, and the parameter limit_m means the distance between the proof mass and the upper virtual shock-protection structure to set the boundary condition in the calculation of the capacitance. The parameter limit shows the distance between the proof mass and the substrate. In a similar manner, integration-reset procedure is performed at the workable condition of the MEMS structure in the fourth formula sequence (A.4). Different from the formula sequence (A.3), parameter stopperpos is used instead of limit_m at the judgment on the negative direction. Thus, when the MEMS structure is workable, the condition can be represented by

stopperpos < V(xs) < limit.

In next formula sequence (A.5), the velocity of the proof mass vi1 is calculated by the integration of the acceleration at, where integration is reset according to the value of the flag viOinteg. Subsequently, the value of the velocity vi1 is output as voltage by the Verilog-A syntax V( vs) < + vil. Likewise, the displacement of the proof mass xi1 is calculated by the integration of the velocity in the sixth formula sequence (A.6).

ARTICLE IN PRESS

The seventh formula sequence (A.7) shows the procedure to set the displacement into the stop position. Here, if the calculated displacement xi1 is equal to or larger than the parameter limit in the positive direction, the displacement value is set to be the value of limit. On the other hand, according to the formula sequence (A.3) or (A.4), displacement xi1 is equal to or less than the parameter stopperpos or limit_m in the negative direction, the displacement value is set to be either value. Then, the value of the displacement xi1 is output as voltage.

2.4. EOM module

A program code for the EOM module is described in Appendix B. that uses four sets of formula sequences. Coding lines are simplified as compared to our previous work, because calculation procedure of the velocity and displacement are included into the shock-protection module. The first formula sequence (B.1) shows the initializing step of the parameters, and the second formula sequence (B.2) shows the calculation of the sum of input forces and the acceleration by the equation of

motion. Then, the calculated values of the input force and the acceleration are output to the shock-protection module as voltages (B.3). After the shock-protection module calculation, the electrical port v1 and v2 output the velocity V(vs) as a voltage, and the port x1 and x2 output the displacement V(xs) as a voltage (B.4).

2.5. Tensile test for adhesion strength evaluation

Tensile tests [28-30] were performed to experimentally obtain the adhesion strength of multi-layer metal structures. Fig. 6 shows schematic images of the test samples used in the experiments. In multilayer metal structures, Ti layers are used for adhesion layers that form material interfaces such as SiO2/Ti, Ti/seed-Au, and electroplated-Au/ Ti. Thus, three types of samples were employed to investigate the adhesion strength of the Ti adhesion layers. Fig. 6(a) shows sample (i) with SiO2 and Ti layers. The SiO2 layer was thermally formed on a Si substrate. Fig. 6(b) represents sample (ii) with SiO2, Ti, seed-Au, and electroplated-Au layers. Sample (iii) in Fig. 6(c) has two sets of

Fig. 4. Equivalent circuit model of MEMS capacitive accelerometer with a shock-protection module.

ARTICLE IN PRESS

D. Yamane et al. / Microelectronics Reliability xxx (2016) xxx-xxx

Fixed electrode Spring

Fig. 5. Device schematic image with shock-protection module parameters.

electroplated-Au layers with each seed-Au and Ti layers. The thickness of each layer is as follows: Si substrate 625 |am, SiO2 1 |Jm, Ti 100 nm, seed-Au 70 nm, electroplated-Au 10 |jm. All the test samples were developed by the same MEMS fabrication process as used for the multilayer metal shock-protection structures. Each test sample is diced into a square of 12 mm x 12 mm in area. Fig. 7(a) shows a photograph of a test sample with a stud pin, which was attached to the sample with an adhesive material, as schematically shown in Fig. 7(b); tensile tests were performed by pulling up the pin. In this work, adhesion strength (Ps), which is the force of adhesion per unit area, is defined by [30]

Ps = Fp/A,

where FP is the force required to pull the metal layers off the substrate, and A is the area of the head surface of the stud pin. Measured adhesion strength as a function of the stroke length in the tensile tests is shown in Fig. 8. The experiment was performed by a tensile testing machine (AG-X, Shimadzu Corp.) with a stroke speed of 0.5 mm/min. From the test results, the Ps was experimentally evaluated to be >4.7 MPa. Thus, in this work, we use 4.7 MPa for the simulation parameter of Ps.

Table 1

Parameters of the multi-physics simulation.

Symbol Parameter Value Unit

£o Dielectric constant 8.854 x F/m

Relative permittivity of SiO2 4.2 -

S Proof mass area 2.776 x m2

m Proof mass 7.582 x kg

gini Initial gap between the proof mass and the 17.6 |im

substrate

gst Initial gap between the proof mass and the 1.46 |im

shock-protection

k Spring constant 1.5 N/m

c Damping coefficient 1.85 x 10 -5 Pa-s

t Thickness of the SiO2 layer 1 |m

Ss Shock-protection anchor area 18.1 x m2

Ps Adhesion strength 4.7 x 106 Pa

Fig. 6. Multi-layer metal samples. (a) Sample (i): Si/SiO2/Ti, (b) sample (ii): Si/SiO2/Ti/ seed-Au/electroplated-Au, and (c) sample (iii): Si/SiO2/Ti/seed-Au/electroplated-Au/Ti/ seed-Au/electroplated-Au.

3. Results and discussions

3.1. Experimental evaluation

To compare simulation and experiment results, we carried out C-G (electrostatic capacitance as a function of input acceleration) measurement. The MEMS device was implemented in a ceramic package and set on a vibration exciter (WaveMaker05, Asahi Seisakusho) to apply vertical acceleration. The capacitance was measured with a semiconductor device analyzer (B1500A, Agilent Tech., Inc.). Simulation parameters used in the shock-protection module were listed in Table 1. Other simulation parameters of the MEMS accelerometer were extracted from the device with the sensing range of ± 3 G as described in reference [3]. Fig. 9 shows the measurement and simulation results of C-G characteristics. When the input acceleration was within the sensing range, which was from 0 G to 3 G, measured capacitance changed with increased input acceleration. As the input acceleration was 3 G or larger, measured capacitance was saturated due to the proof mass ramming into shock-protection structures. In the proposed multi-physics simulation, the shock-protection module added the force Fs from the shock-protection structures when input acceleration was over 3 G. Thus, the simulation results are consistent with the measurement results. From this evaluation, we confirmed that the shock-protection module for the multi-physics simulation could handle the analysis of the shock-protection structures even when input acceleration was out of sensing range.

3.2. Transient analysis

Fig. 10 shows the transient analysis results of the MEMS accelerom-eter integrated with the shock-protection module by the multi-physics simulation along with the simulation parameters as summarized in Table 1; (a) at the acceleration level of 1 G, the capacitance change showed the same sinusoidal curve as that of the input acceleration.

Fig. 7. Tensile test sample. (a) Photo of a test sample and (b) schematic of the cross-section view.

ARTICLE IN PRESS

D. Yamane et al. / Microelectronics Reliability xxx (2016) xxx-xxx

Fig. 8. Measured adhesion strength as a function of stroke length.

(b) When the acceleration level was 5 G, the proof mass made physical contact with the shock-protection structure, and thus the capacitance change became non-linear characteristics. (c) At the acceleration level of 1 kG, the movable structures touched either the shock-protection structure or the substrate as shown in Fig. 5. When the input acceleration was higher than 11 kG, the shock-protection structures were broken, and the capacitance came close to zero level. Those analysis methods can be useful to predict the transient behavior and the mechanical failure of the shock-protection structures at the input of excess acceleration.

3.3. Capacitance-acceleration analysis

Fig. 11 shows the non-linear analysis result of multi-physics C-G simulation at the input of excess acceleration. In the simulation, we used the parameters presented in Table 1. The capacitance changed

Fig. 9. Measured and multi-physics C-G simulation results of the MEMS accelerometer.

Fig. 10. Transient analyses of the MEMS accelerometer using multi-physics simulation with the shock-protection module. (a) Input acceleration of the sine wave of 1 G at 50 Hz, (b) input acceleration of the sine wave of 5 G at 50 Hz, and (c) (i) 0-100 ms: input acceleration of the sine wave of 1 x 103 G at 50 Hz, (ii) after 100 ms: input acceleration of 15 x 103 G at 50 Hz.

until input acceleration reached the sensing limit of3 G. When input acceleration was > 3 G, the capacitance showed a constant value due to the physical contact between the proof mass and the shock-protection structure. Then, at the input acceleration of 11 kG, the shock-protections were broken and the proof mass moved away from the substrate. The anchor area (Ss) of the shock-protection structure was determined by multiplying the anchor area (As) by the total number (N) of the anchors of the shock-protection structures in the device, as shown in Fig. 11.The simulation results indicates that the adhesion strength of the shock-protection structure has the potential shock tolerance of up to around 1000 G-10,000 G for consumer electronics applications. This analysis can be employed to estimate the mechanical characteristics of the shock-protection structures.

4. Conclusions

We showed a novel approach for the evaluation and modeling of adhesion layer in shock-protection structures for the MEMS accelerometer. The adhesion force in the metal adhesion layers was examined by the tensile tests, and the equivalent circuit model for the shock-protection module in the multi-physics simulation was proposed to evaluate the mechanical characteristics of the shock-protection structures. The simulation results with the shock-protection module were consistent with the measured C-G characteristics. The transient and C-G simulations at the input of excess acceleration were also demonstrated to show the non-linear behavior analyses of the shock-protection structures. Those results show that the proposed module is useful for the multi-physics simulation to analyze the mechanical characteristics of the MEMS accelerometer.

Conflicts of interests

No conflict of interest is known to exist in this work.

Acknowledgements

The authors would like to thank Dr. T. Maruno, Dr. Y. Akatsu, M. Yano, K. Kudo and M. Fujinuma with NTT-AT Corp. for technical discussions. This work was supported by CREST, JST, and JSPS KAKENHI Grant Number 15K17453.

ARTICLE IN PRESS

D. Yamane et al. / Microelectronics Reliability xxx (2QÍ6) xxx-xxx

0.1 1 10 100 1k 10k 100k Input Acceleration (-Z direction) [G]

Fig. 11. Multi-physics C-G simulation for the non-linear analysis of the shock-protection structures.

Appendix A. Verilog-A expression of shock-protection module

Verilog-A expression of shock-protection module begins with the definition of initial conditions as follows:

analog begin

@(initial_step>) begin

vi0=vini;

vil=vini;

xi0=xini;

V(vs)<+vini;

V(xs)<+xini;

Fs=Ps*Ss;

vi0integ=0;

(A. 1)

where, vini and xini are the initial proof-mass velocity and the initial proof-mass position, respectively. In this work, vini is set to be 0, and xini is set to be 0.

After the initializing step, breaking of shock-protection structures is judged as follows:

at = V (a); Ftotal = V (Ft); if (-1 * Ftotal> = Fs) sb= 1;

where a and Ft are the acceleration of proof mass and the total force on poof mass, respectively.

Then, at stopper broken condition, the integration of acceleration is reset as follows:

if(sb==l) begin if(V(xs)> =ttmit) begin if(at>0)

vi0integ=l;

vi0integ=0;

else if(V(xs)<=limitjn) begin if(at<0)

vi0integ=l;

vi0integ=0;

vi0integ=0;

At stopper workable condition, the integration of acceleration is reset as follows:

else begin

if(V(xs)>-limit) begin if(at>0)

vi0integ=l;

vi0integ=0;

else if(V(xs)<=stopperpos) begin if(at<0)

vi0integ=l;

end else

vi0integ=0;

vi0integ=0;

The velocity (vii) and the displacement (xii) of proof mass is calculated as follows:

vil = idt(at, viO, viOinteg); V(vs)< + vil ; xil = idt(vi1, xiO) ;

Finally, the proof-mass displacement (xil) is set to be stop position as follows:

ARTICLE IN PRESS

D. Yamane et al. / Microelectronics Reliability xxx (2016) xxx-xxx

analog begin

@(initial_step) begin

viO=vini;

vil=vini;

xiO=xini; (A.7)

V(vl)<+vini;

V(v2)<+vini;

V(xl)<+xini;

V(x2)<+xini;

Appendix B. Verilog-A expression of EOM module

Verilog-A expression of EOM module begins with the definition of initial conditions as follows:

analog begin

@(initial_step) begin

viO=vini;

vil-vini;

xiO=xini; (B-l)

V(vl)<+vini;

V(v2)<+vini;

V(xl)<+xini;

V(x2)<+xini;

where v1, v2, x1 and x2 are the proof-mass velocity transferred to suspension module, the proof-mass velocity for monitoring, proof-mass position transferred to suspension module, and proof-mass position transferred to actuator module, respectively.

Then, the sum of input forces and the acceleration are calculated by the equation of motion as follows:

vd1 = i(Fe1)+ I (Fe2); vd2 = I(Fm1) + I(Fm2) ; Ftotal = vd1 + vd2; at = Ftotal/m ;

where Fe1, Fe2, Fm1 and Fm2 are electrostatic force that is not used in this module, electrostatic attractive force generated by actuator module, elastic restoring force generated by suspension module, and inertial force generated by acceleration module, respectively.

The calculated values of the input force and the acceleration are output to the shock-protection module as follows:

V (a)< + at;

V (Ft)< + Ftotal;

After the calculation in the shock-protection module, the velocity and the displacement of proof mass are output as follows:

V (v1)< + V (vs)

V (v2)< + V (vs)

V (x1 )< + V (xs)

V (x2)< + V (xs) end

endmodule

G. Krishnan, C.U. Kshirsagar, G.K. Ananthasuresh, N. Bhat, Micromachined high-resolution accelerometer, J. Indian Inst. Sci. 87 (3) (Sept. 2007) 333-361.

D. Yamane, T. Konishi, T. Matsushima, H. Toshiyoshi, K Masu, K. Machida, A 0.1G-to-20G integrated MEMS inertial sensor, Jpn. J. Appl. Phys. 54 (8) (July 2015) (pp. 04DE06.1-04DE06.6).

D. Yamane, T. Konishi, T. Matsushima, K. Machida, H. Toshiyoshi, K. Masu, Design of sub-1 g microelectromechanical systems accelerometers, Appl. Phys. Lett. 104 (2014) 074102.

K. Machida, S. Shigematsu, H. Morimura, Y. Tanabe, N. Sato, N. Shimoyama, T. Kumazaki, K. Kudou, M. Yano, H. Kyuragi, A novel semiconductor capacitive sensor for a single-chip fingerprint sensor/identifier LSI, IEEE Trans. Electron Devices 48 (10) (Oct. 2001) 2273-2278.

K. Machida, T. Konishi, D. Yamane, H. Toshiyoshi, K. Masu, Integrated CMOS-MEMS technology and its applications, ECS Trans. 61 (6) (May 2014) 21-39. M.-H. Tsai, C.-M. Sun, W.-J. Mao, W. Fang, The CMOS-MEMS 3-axis capacitive accelerometer to meet the commercial specifications, Proc. IEEE MEMS 2016, pp. 1002-1005.

V.T. Srikar, S.D. Senturia, The reliability of microelectromechanical systems (MEMS) in shock environments, J. Microelectromech. Syst. 11 (3) (June 2002) 206-214. S.W. Yoon, S. Lee, N.C. Perkins, K. Najafi, Shock-protection improvement using integrated novel shock-protection technologies, J. Microelectromech. Syst. 20 (4) (Aug. 2011) 1016-1031.

J.W. Kim, H.W. Park M.K. Choi, W.K Jeung, J.W. Lee, Drop impact reliability of MEMS inertial sensors with membrane suspension for mobile phones, Proc. IEEE ECTC 2012, pp. 344-349.

D.S. Eddy, D.R. Sparks, Application of MEMS technology in automotive sensors and actuators, Proc. IEEE 86 (8) (Aug. 1998) 1747-1755.

J.D. Turner, A.L. Austin, Sensors for automotive telematics, Meas. Sci. Technol. 11 (2) (Feb. 2000) R58-R79.

T.G. Brown, Harsh military environments and microelectromechanical (MEMS) devices, Proc. IEEE Sensors 2003, pp. 753-760.

J.-M. Stauffer, Current capabilities of MEMS capacitive accelerometers in a harsh environment, IEEE Aerosp. Electron. Syst. Mag. 21 (11) (Nov. 2006) 29-32.

H.R. Shea, Reliability of MEMS for space applications, Proc. SPIE, vol. SPIE-6111,2006 (pp. 61110 A1-61110 A10).

U. Wagner, J. Franz, M. Schweiker, W. Bernhard, R. Muller-Fiedler, B. Michel, O. Paul, Mechanical reliability of MEMS—structures under shock load, Microelectron. Reliab. 41 (2001) 1657-1662.

S.W. Yoon, N. Yazdi, N.C. Perkins, K. Najafi, Micromachined integrated shock protection for MEMS, Sensors Actuators A Phys. 130-131 (2006) 166-175. S. Mariani, A. Ghisi, F. Fachin, F. Cacchione, A. Corigliano, S. Zerbini, A three-scale FE approach to reliability analysis of MEMS sensors subject to impacts, Meccanica 43 (2008) 469-483.

M. Naumann, J. Mehner, D. Lin, T.F. Miller, Design and application of flexible stops for MEMS devices, Proc. IEEE Sensors 2010, pp. 168-173.

Y.-K. Tao, Y.-F. Liu, J.-X. Dong, Flexible stop and double-cascaded stop to improve shock reliability of MEMS accelerometer, Microelectron. Reliab. 54 (2014) 1328-1337.

P. Lall, N. Kothari, J. Glover, Mechanical shock reliability analysis and multiphysics modeling of MEMS accelerometers in harsh environments, Proc. ASME InterPACK2015, 2015 (pp. V003T07A007).

A. Corigliano, F. Cacchione, A. Frangi, S. Zerbini, Numerical modelling of impact rupture in polysilicon microsystems, Comput. Mech. 42 (2) (July 2008) 251 -259. A. Ghisi, S. Kalicinski, S. Marini, I.D. Wolf, A. Corigliano, Polysilicon MEMS Acceler-ometers Exposed to Shocks: Numerical-Experimental Investigation, J. Micromech. Microeng. 19 (3) (Feb. 2009) 035023 (12 pp).

T. Konishi, K. Machida, S. Maruyama, M. Mita, K. Masu, H. Toshiyoshi, A single-platform simulation and design technique for CMOS-MEMS based on a circuit simulator with hardware description language, J. Microelectromech. Syst. 22 (3) (June 2013) 755-767.

W.M. v. Spengen, MEMS reliability from a failure mechanisms perspective, Microelectron. Reliab. 43 (2003) 1049-1060.

R. Ardito, A. Corigliano, A. Frangi, Modelling ofspontaneous adhesion phenomena in micro-electro-mechanical systems, Eur. J. Mech. A Solids 39 (2013) 144-152. T. Konishi, D. Yamane, T. Matsushima, K. Masu, K. Machida, H. Toshiyoshi, A capacitive CMOS-MEMS sensor designed by multi-physics simulation for integrated CMOS-MEMS technology, Jpn. J. Appl. Phys. 53 (2014) (pp. 04EE15.1-7). K.L. Mittal, Adhesion measurement of thin films, Electrocompon. Sci. Technol. 3 (1976) 21-42.

T. Horiuchi, T. Yamaguchi, A. Kinbara, Adhesive strength ofevaporated gold films on glass substrate, Shinku 11 (8) (1968) 285-289.

R. Jacobsson, B. Kruse, Measurement of adhesion of thin evapolated films on glass substrates by means of the direct pull method, Thin Solid Films 15 (1973) 71-77.

References

[1] N. Yazdi, F. Ayazi, K. Najafi, Micromachined inertial sensor, Proc. IEEE 86 (8) (Aug. 1998) 1640-1659.