MOEMS Vibration Sensor for Advanced Low-frequency … · 838 W. Hortschitz et al. / Procedia Engineering 87 ( 2014 ) 835 – 838 Figure 4: Transfer characteristics of the MOEMS with

Procedia Engineering 87 ( 2014 ) 835 – 838

Available online at www.sciencedirect.com

1877-7058 © 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).Peer-review under responsibility of the scientific committee of Eurosensors 2014doi: 10.1016/j.proeng.2014.11.282

ScienceDirect

EUROSENSORS 2014, the XXVIII edition of the conference series

MOEMS Vibration Sensor for Advanced Low-Frequency

Applications with pm Resolution

W. Hortschitza,∗, A. Kainzb, H. Steinera, M. Stiftera, F. Kohla, J. Schalkoa,b, T. Sautera,

F. Keplingerb

aCenter for Integrated Sensor Systems, Danube University Krems, Wiener Neustadt, AustriabInstitute of Sensor and Actuator Systems, Vienna University of Technology, Vienna, Austria

Abstract

The majority of MEMS vibration sensors requires relatively high resonance frequencies of several kilohertz toavoid mechanical contact of moving parts such as electrode plates. In contrast to that, our hybrid micro-opto-electro-mechanical system (MOEMS) enables sensor applications at low frequencies. This work describes a distinct MOEMSfeaturing extremely low resonance frequencies of below 200 Hz. It is operated ambient air without closed loop feed-back, extensive electronics and cooling. Due to the soft suspension of the micro-mechanical sub-system the funda-mental limit, the Brownian noise floor, is reached. The resulting noise equivalent displacement for frequencies above

the resonance is 1.9 pm/√

Hz, which is equivalent to 0.29 μg/√

Hz below the resonance. We describe the design spacefor sensors with further enhanced sensitivity and resonance frequencies far below 100 Hz.

© 2014 The Authors. Published by Elsevier Ltd.

Peer-review under responsibility of the scientific commitee of Eurosensors 2014.

Keywords: MOEMS, vibration, sensor, low frequency, high resolution

1. Introduction

Micro-mechanical vibration sensors or accelerometers were used for decades in automotive industry, to monitor

bearings in fabrication lines, and more recently, in consumer electronics like cameras for image stabilization or game

consoles [5]. Nowadays, capacitive MEMS sensors are applied in almost all of these cases. This implies, however,

relatively stiff systems and, therefore, high resonance frequencies to avoid damage of the sensor components in case

of a mechanical shock [1]. Furthermore, most of these sensors are operated in a closed loop mode to extend the

dynamic range at the expense of a high power consumption. The presented competitive sensor principle depicted

in Fig. 1 features laterally deflectable openings which modulate a perpendicularly introduced light flux. The light

flux is generated by an LED (Osram SFH 4680) and received by a photo-transistor (Osram SFH 3600). Therefore,

the seismic mass of the MEMS can be deflected in-plane almost freely without any risk of damage or destruction of

electrodes or readout elements.

∗Corresponding author. Tel.: +43 2622 2342 0; fax: +43 2622 23 42 0 99

E-mail address:[email protected]

© 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).Peer-review under responsibility of the scientific committee of Eurosensors 2014

836 W. Hortschitz et al. / Procedia Engineering 87 ( 2014 ) 835 – 838

Figure 1: Schematic cross-section of the sensor for a fully encapsulated

device. The emitted light flux is modulated by two micro-structured grat-

ings. The first is fabricated as chromium layer fixed to the MEMS frame

while the second one is suspended via springs inside device layer of an

SOI wafer. The output signal of the phototransistor is proportional to the

deflection of the movable mass.

Figure 2: Micrographs of the fabricated MEMS components of the

MOEMS sensor. The perforated seismic mass is ~2.5x2.5 mm² in

size and was fabricated into a 20 μm thick device layer of an SOI

wafer. The inset shows the fixed Cr layer on glass, bonded to the

SOI chip, forming a light modulation aperture.

2. Constraints for the sensitivity

The MOEMS incorporates two main subsystems, i. e. the mechanical and the opto-electrical one. Both subsys-

tems exhibit constraints regarding the resolution limits of the combined sensor system. The main limitation for the

mechanical subsystem is the Brownian motion noise of the seismic mass while the optoelectronics is constrained by

the dark current of the receiver element, e.g., the photo-transistor. For the presented sensor, the Brownian noise is the

dominating restriction for the sensitivity. Hence, it will be discusse subsequently in greater detail.

The microstructure depicted in Fig. 2 was fabricated into the 20 μm thick device layer of an SOI wafer. The

perforated, oscillating mass of 2.5x2.5 mm² is suspended by four folded beams. The low total stiffness of the structure

of c = 0.23 N/m results in a relatively large Brownian noise equivalent deflection NEDBr where the underlying

Brownian noise force exhibits a white noise characteristic. The transfer characteristic Hn between an input force

and the differential deflection xd can be approximated by Hn(ω � ω0) ≈ 1/c for frequencies far below the angular

resonance frequency ω0 of the MEMS. The noise equivalent deflection induced by the Brownian force is, therefore,

given by

NEDBr = Fn · Hn = Fn/c =√

4kBTd/c , (1)

where kB is the Boltzmann constant, T is the absolute temperature and d is the damping coefficient. Hence, the noise

equivalent deflection does not depend on the seismic mass m but on the damping coefficient d and the stiffness c,

for frequencies far below the angular resonance frequency. The damping can be minimized by reducing the ambient

pressure, requiring a hermetical package of the MEMS component. Such sealings are challenging to fabricate and

hence not considered in this paper. The work focuses on the impact of the stiffness instead. The sensor design allows

to increase the stiffness c in order to reduce the influence of the Brownian noise. This leads to a higher natural angular

frequency, which can be compensated by increasing the seismic mass.

837 W. Hortschitz et al. / Procedia Engineering 87 ( 2014 ) 835 – 838

Metal housing

with shaker unit

Lock-in for

circuit output

Lock-in for output

of laser-vibrometer

Shaker

ampli er

Waveform-

generator

Laser-vibrometer

sensing head

Laser-vibrometer-

controller and PC

(a) Overall measurement set-up for characterizing the MOEMS vibration

sensors. The laser-Doppler-vibrometer is part of a Polytec MSA-400 mi-

cro system analyzer. The sensor is mounted inside the metal housing in

the middle. A detailed view of the sensor mounting is depicted in Figure

(b). The shaker unit is excited via a waveform generator and an appropi-

ate piezo amplifier.

Micro system analyzer

laservibrometer

Excitation via

shaker unit

Sensor

mounting

Am

pli

er

Supply

Exci

tation v

ia

shaker

unit

MEMS

(b) Detailed vibration measurement set-up for transducer character-

ization. In the depicted set-up the optical transmission axis of the

transducer is oriented horizontally, whereas the vertically aligned

laser beam of the Doppler-vibrometer probes the motion of the foot-

plate through a hole in the sample holder.

Figure 3: Measurement set-up with detailed view on the sensor mounting.

3. Setup and Results

The devices were characterized with the set-up depicted in Fig. 3. The MEMS is operated in ambient air without

closed loop feedback, extensive electronics or cooling. The results were obtained by exciting the sensor with a custom

made piezo-electical shaker unit (Fig. 3b). Excitation amplitude and phase were determined with a laser Doppler-

vibrometer (Polytec MSA-400) while simultaneously recording the output signal of the sensor’s transimpedance am-

plifier with a Stanford Research Lock-In amplifier (SR830). The Lock-In amplifier was also used to estimate the

electrical noise floor of the opto-electronics while the LED was switched off. Further details of the measurement

set-up and the data evaluation can be found in [4] and [2]. The measurement results depicted in Fig. 4 were fitted

with an iterative least squares algorithm to gain the mechanical parameter of the sensor. The resolution was computed

with the mechanical parameters listed in Fig. 4 and Eq. 1 to be NEDBr = 1.9 pm/√

Hz. The dark current noise of the

phototransistor and the output noise of the transimpedance amplifier were determined and recalculated with the sensi-

tivity from Fig. 4 to be equivalent to NEDLED,off = 0.0146 pm/√

Hz. Hence, Brownian noise is currently limiting the

sensitivity of the sensor. This limitation can easily be circumvented for a constant resonance frequency by increasing

the stiffness via a thicker device layer [3].

4. Conclusion

The discussed sensor is operated in ambient air without closed loop feedback, extensive electronics, or cooling.

Due to the soft suspension of the seismic mass in the micro-mechanical sub-system of the sensor of 0.23 N/m, one

fundamental limit, the Brownian noise floor, is reached. The resulting noise equivalent displacement for frequencies

above the resonance frequency is 1.9 pm/√

Hz, which is equivalent to 0.29 μg/√

Hz below the resonance. A future

work will concentrate on further increasing the sensitivity by adjusting the mechanical parameters stiffness and mass

as described in this paper while simultaneously decreasing the resonance frequency to gain more bandwidth for

displacement sensing.

The presented extremely sensitive low frequency (MOEMS) vibration sensor with low resonance frequency paves

the way for micro-mechanical systems to be used in new fields of applications such as seismology or novel medical

applications.

838 W. Hortschitz et al. / Procedia Engineering 87 ( 2014 ) 835 – 838

Figure 4: Transfer characteristics of the MOEMS with a resonance frequency of 196 Hz; The excitation amplitude x̃f is 6 nm. The mechanical

parameters for stiffness, mass and damping which were fitted with a least squares fit are listed inside the plot. The phase of the sensor system is

ϕS = ϕOut − ϕf , the phase of the sensor foundation is ϕf while ϕOut is the phase of sensor output.

References

[1] Acar, C., Shkel, A. M., 2003. Experimental evaluation and comparative analysis of commercial variable-capacitance MEMS accelerometers.

Journal of Micromechanics and Microengineering 13 (5), 634.

[2] Hortschitz, W., 2013. Hybrid MOEMS Displacement Sensor and Accelerometer. Ph.D. thesis, Vienna University of Technology.

[3] Hortschitz, W., Encke, J., Kohl, F., Sauter, T., Steiner, H., Stifter, M., Keplinger, F., oct. 2012. Optimized hybrid MOEMS sensors based on

noise considerations. In: Sensors, 2012 IEEE. pp. 1 –4.

[4] Hortschitz, W., Steiner, H., Sachse, M., Stifter, M., Kohl, F., Schalko, J., Jachimowicz, A., Keplinger, F., Sauter, T., nov. 2011. An Optical

In-Plane MEMS Vibration Sensor. Sensors Journal, IEEE 11 (11), 2805 –2812.

[5] Tanaka, M., 2007. An industrial and applied review of new MEMS devices features. Microelectronic Engineering 84 (58), 1341 – 1344,

proceedings of the 32nd International Conference on Micro- and Nano-Engineering.