Sensors and Actuators A 145–146 (2008) 283–290
Contents lists available at ScienceDirect
Sensors and Actuators A: Physical
journal homepage: www.elsevier.com/locate/sna
In-plane MEMS-based nano-g accelerometer with sub-wavelength optical
resonant sensor
U. Krishnamoorthy a,∗ , R.H. Olsson III a , G.R. Bogart b , M.S. Baker a ,
D.W. Carr b , T.P. Swiler a , P.J. Clews a
a
b
Advanced MEMS Technologies, Sandia National Laboratories, Albuquerque, NM 87185, USA
Symphony Acoustics, 103 Rio Rancho Drive, Suite B-4, Rio Rancho, NM 87124, USA
a r t i c l e
i n f o
Article history:
Received 30 June 2007
Received in revised form 11 March 2008
Accepted 11 March 2008
Available online 28 March 2008
Keywords:
Nano-grating
MEMS optical accelerometer
Optical resonant detection
In-plane accelerometer
Optical sensing
Low-g accelerations
a b s t r a c t
We have successfully demonstrated a series of results that push the limits of optical sensing, acceleration
sensing and lithography. Previously, we √
built some of the most sensitive displacement sensors with displacement sensitivities as low as 12 fm/ Hz at 1 kHz. Using reference detection circuitry in conjunction
with correlated double sampling methods, we lowered the 1/f noise
√ floor to 10 mHz, hence improving
the detection limit at low frequencies (10 mHz) by 77 dB to 50 fm/ Hz. We converted these highly sensitive displacement sensors to highly sensitive acceleration sensors through a direct mass integration
processes.
√ Our accelerometers have resonant frequencies as low as 36 Hz and thermal noise floors as low
as 8 nG/ Hz (where 1 G = 9.8 m/s2 ). We have pushed the limits
of shaker table experiments to indepen√
dently verify acceleration measurements as low as 10 ␮G/ Hz. Direct measurements with our integrated
sub-wavelength optical nano-grating
accelerometers have shown device sensitivities of 590 V/G and noise
√
floors corresponding to 17 nG/ Hz (at 1 Hz).
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
There is growing interest in extremely high-sensitivity,
compact accelerometers (nano-g sensitivity where 1 nanog = 9.81 × 10−9 m/s2 ) for a growing list of applications including
high-precision inertial navigation and seismic sensing for geophysical and oil-field applications. Such applications √
require
measurement of extremely small acceleration signals (nG/ Hz) at
very low frequencies (<100 Hz). Currently there are no chip-scale,
compact accelerometers that meet these high-sensitivity and lowbandwidth requirements.
The fundamental challenges in building such a device include
designing very high sensitivity displacement sensors with low
thermo-mechanical noise and minimizing 1/f noise. A practical
challenge in building these accelerometers on a micro-scale is in
integrating a large enough proof mass to the displacement sensor
and meet the sensitivity requirements for the device (since larger
mass relates to better sensitivity).
Our first step in building this accelerometer was to identify and
demonstrate a high-sensitivity optical displacement/motion sensing mechanism. Previously, optical sensing mechanisms based-on
∗ Corresponding author. Tel.: +1 505 844 6254.
E-mail address: ukrishn@sandia.gov (U. Krishnamoorthy).
0924-4247/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.sna.2008.03.017
diffraction-gratings have been used to demonstrate sub-angstrom
scale sensitivity [1]. Advantages of optical detection techniques
compared to capacitive or piezoresistive methods include high sensitivity and performance close to the Brownian noise limits of the
mechanical structure [2,3].
An in-plane nanophotonic resonant sensor based on multilayer sub-wavelength optical gratings was developed [2–4]. This
nano-optic sensor comprised of two sets of nano-gratings that
modulate the near-field intensity and polarization of an incident
light source based on the relative lateral motion of the two gratings. The in-plane sensing mechanism offers the advantage of
single-chip solutions to multiple axis motion/acceleration detection. We have previously published results from these sensors
√
demonstrating displacement sensitivities as low as 12 fm/ Hz at
1 kHz [2,3].
Further, we improved sensor performance at lower frequencies
to suit our target applications such as precision inertial navigation and seismic applications. We dramatically improved sensor
resolution by reducing 1/f noise limits in our detection circuitry
through design and implementation of specialized reference detection circuitry and correlated double sampling methods [5]. Next,
we added masses to these delicate sub-wavelength nano-gratingbased displacement sensors to build extremely high-resolution
accelerometers and compared them to similar accelerometers
reported in the past [7,8].
284
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
Fig. 2. Readout circuit used to cancel laser relative intensity noise for nano-grating
optical sensors [5].
Fig. 1. MEMS optical near-field resonant displacement sensor based on vertically
stacked sub-wavelength nano-gratings. The nano-gratings are attached to electrostatic actuators to control their motion and characterize their displacement
sensitivity [3].
2. MEMS optical displacement sensor
Previously, we demonstrated a new class of nano-grating based
optical MEMS resonant sensors√with extremely high lateral displacement sensitivities (12 fm/ Hz at 1 kHz) and greater than
120 dB of open loop dynamic range [3]. As mentioned earlier, these
sensors consist of two vertically offset layers of sub-wavelength
polysilicon nano-gratings separated by an air gap. They modulate the near-field intensity and polarization of an incident light
source in response to relative motion of the nano-gratings [2,3].
The reflected/transmitted optical beam intensity from the nanogratings is measured as a function of the relative lateral positions
of the nano-gratings. Electrostatic actuators were integrated into
the sensor to control nano-grating motion and measure the displacement sensitivity. Typical devices showed 10%/nm change in
reflectance vs. lateral position. [2]. A detailed description and principle of operation of this near-field MEMS optical resonant sensor
can be found in [4]. An SEM of this device is shown in Fig. 1. The
high sensitivity of this sensor made it a natural choice to be the
basis for our accelerometer design.
3. 1/f noise
Above several kHz, we can achieve the shot noise limit with commercially available reference detection circuitry. However, at the
lower frequencies needed for our applications outlined earlier, 1/f
components of both laser relative intensity noise (RIN) and amplifier noise dominate the response. This degrades the sensitivity of
the nano-grating sensors in applications that require excellent sensitivity at frequencies below 100 Hz. Reducing the 1/f noise in these
optical MEMS sensors is critical in extending their range to these
ultra-low frequencies.
In order to reduce 1/f noise, we designed and built reference
detection low noise readout circuitry that cancels laser relative
intensity noise (RIN) to frequencies as low as 0.7 Hz [5]. The low
frequency bandwidth of the sensor system was further reduced
to <10 mHz using a new MEMS correlated double sampling technique that canceled low frequency RIN, drift, and circuit 1/f noise
by greater than 77 dB [5].
The reference detection circuit (Fig. 2) implements the architecture presented in [6] but optimized for low frequency performance.
An additional reference diode is included in the circuit that can-
cels laser RIN. Details of this laser noise cancellation circuitry and
related analysis are published in [5].
While the reference detection circuitry rejects over 60 dB of
laser RIN, more rejection is needed to achieve the low frequency
performance requirements for our target applications. To further
reduce the 1/f corner and allow ultra-precise position measurement at very low frequencies a micromechanical correlated double
sampling (CDS) scheme was implemented [5].
The correlated double sampling technique uses the MEMS
grating element to perform the sampling. First, the electrostatic
actuators drive the nano-grating (shown in Fig. 1) to a known
position (Ex: zero displacement). This first sample position measurement is quantized. This measurement is used as a reference
for the actual displacement measurement. The two signals are subtracted digitally to further reduce the 1/f laser noise at the output.
This method requires the bandwidth of the electrostatically driven
MEMS device and the associated readout circuit to be at least three
times the sampling frequency to implement this scheme. Because of
aliased RIN, shot, and circuit noise the white noise floor is increased
√
by 6. However, this technique does cancel any low frequency noise
in the laser, position sensor, photodiodes, or reference detection
circuitry due to 1/f noise, drift, or thermal effects.
Hence, combining custom reference detection circuitry and correlated double sampling methodology, we reduced the 1/f noise
corner to 10 mHz and improved
√ the detection limit at low frequency
(10 mHz) by 77 dB to 50 fm/ Hz (Fig. 3).
4. Nano-g accelerometer
4.1. Operating principle
Building accelerometers from these high sensitivity displacement sensors involves the integration of an appropriate proof mass.
The acceleration experienced by of this proof mass is directly proportional to its displacement as shown in the following equation:
a = ωo2 x
(1)
where a is acceleration experienced by the proof mass, x is its displacement and ωo is resonant frequency of the proof mass given
by
ωo2 =
k
m
where k is its mechanical spring constant and m is its mass.
(2)
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
285
Fig. 5. Integration concept: nano-g accelerometer showing accelerometer chip with
integrated nano-grating displacement sensor and proof mass integrated with an
optical source (VCSEL) and detector (photodetector) chips.
Fig. 3. Displacement power spectral density of the reference detection circuit with
correlated double sampling scheme when
interfaced with a nano-grating MEMS
√
device [5]. The detection limit, 50 fm/ Hz extends down below 10 mHz. The reference detection and correlated double sampling techniques described above have
improved the displacement detection limit at low frequencies (10 mHz) by 77 dB.
4.2. Accelerometer design
For initial design simplicity, the accelerometer design we chose
was an open loop system without any feedback controls. To meet
our compact size requirements, our ideal choice was to build a
MEMS accelerometer using standard IC fabrication methodologies.
However, measuring nano-G accelerations requires displacement sensitivities in the femto-meter scale (where 1 fm = 10−15 m)
and low resonant frequencies (∼100 Hz) based on Eq. (1). This translates to a large mass, m, and a weak spring constant, k, from Eq. (2).
For example: An acceleration of 4 nG produces a displacement of
100 fm for a resonant frequency of 100 Hz.
For our accelerometer design (Fig. 4), we started with the optical
displacement sensor based on sub-wavelength in-plane√ nanogratings as mentioned earlier. The high resolution (12 fm/ Hz) of
this displacement sensor was essential to obtain the high sensitivities we needed for our compact low-frequency accelerometer
design.
To meet our low resonant frequency requirements (<100 Hz)
with our open-loop accelerometer design, we needed a large proofmass with very weak springs (see Eq. (2)). Here we had to push
our fabrication limits using a combination of surface micromachining and Deep Reactive Ion Etching (DRIE) to create through-wafer
structures that maximized the mass, m, while minimizing the
spring constant, k. In addition, the mass springs were designed
for large motion in one direction (y-axis) with very high crossaxis rejection. Our through-wafer etch process and lithographically
defined mass springs provided greater stiffness in the z-axis and
x-axis.
Considering the high sensitivity of our device, we needed to provide isolation of the mass and optical sensor from the thermally
induced stresses caused by the mismatch in coefficient of thermal
expansion between the silicon MEMS device and the attached package or substrate. We have designed a gimbal around the proof mass
whose springs help isolate higher frequency inputs and thermally
induced stresses.
We distributed multiple nano-grating sensors symmetrically
around the proof mass to enable differential sensing (Fig. 4). This
allows cancellation/rejection of both off-axis displacement errors
and common mode noise including (Ex: from thermal effects). Both
analytical and finite-element models were used to optimize the
gimbaled mass–spring system design.
Fig. 4. (Top-view) Accelerometer chip layout detail showing proof mass and device isolation platform (i.e. gimbal) with associated springs. Note that substrate section not
shown in layout. This design shows multiple nano-gratings located between the mass and gimbal for common-mode noise rejection through differential sensing. The insets
show details of the mass spring, gimbal spring and nano-grating models.
286
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
Finally, we adapted our design for simple and robust integration
of optical sources and detectors. We incorporated through-wafer
holes under the nano-grating-based displacement sensors to
accommodate transmission detection schemes with the facility
to package VCSELs and photodetectors above/below the device
(Fig. 5).
4.2.1. Spring design
A first order analytical model was generated for both the mass
and gimbal springs (Fig. 4). The mass springs were designed to be
very compliant in the y direction and relatively stiff in the x and
z directions. To simplify the analysis, we made a few assumptions
including a linear system approximation, small deflections and very
stiff spring trusses.
Using basic beam theory, the equations that define the springs
for the device are given by:Effective spring constant in y-direction,
ky
eff
=
Ehw3
2L3
(3)
20 ␮m. Cross-axis displacement coupling was negligible. The xdisplacement for 1 G acceleration in the x-direction (worst case)
was ∼0.1 nm. The z-displacement was not simulated but the analytical model predicted it to be ∼0.15 ␮m at 1 G and ∼0.15 fm at 1 nG
accelerations. ANSYS simulations of thermal effects predicted DC
shifts of ∼0.8 nm and ∼40 nm in the y-deflection for temperature
differentials of 1 ◦ C and 50 ◦ C, respectively.
4.2.2. Optical grating design
The two-layer optical gratings are defined in two vertically offset
polysilicon layers suspended in air and separated by an air gap. The
bottom grating was attached to the compliant mass and moved laterally with it. The upper grating layer was attached to the relatively
stiff gimbal.
A rigorous coupled wave analysis (RCWA) was used to simulate
and optimize the optical response of the grating for normal incidence [3,4]. The following design parameters were chosen for our
optical nano-gratings (Fig. 4):
Effective spring constant in x-direction,
kx
eff
=
2Ehw
L
(4)
Effective spring constant in z-direction,
kz
eff
=
Ewh3
2L3
(5)
where E = Young’s modulus (∼160 GPa for silicon) and h = spring
thickness.
Specifications for our lower mass device were:
Proof mass: 4500 ␮m × 4500 ␮m × 400 ␮m.
Mass springs: 4 sets of folded flexures with each spring dimension: 20 ␮m × 4300 ␮m × 400 ␮m.
Gimbal springs: 4 sets of folded flexures with each spring dimension: 30 ␮m × 1100 ␮m × 400 ␮m.
Mass, m, for the gimbal and proof mass were calculated to be
∼21.9 mg and 18.9 mg, respectively (based on device layout).
Deflections were calculated from the force balance equation:
F = ma = yk
(6)
a
where
k
Therefore, deflectiony = m ×
k is the appropriate spring
constant in the specified direction for the mass or gimbal element.
The ratio of spring constants for the mass are:
kx
ky
eff mass
kx
ky
eff mass
eff mass
eff mass
= 1.85 × 105
= 4.00 × 102
The ratio of the spring constants between the gimbal and the mass
in the y-direction is:
ky
ky
eff gimbal
eff mass
∼ 200
Hence, this design should provide adequate decoupling between
modes in the x, y and z direction for both the device proof mass and
gimbal.
We developed a 2D finite element model for this spring-mass
model using ANSYS® . Modal simulations predicted dominant resonant frequencies at 108 Hz (y translation of mass) and at 406 Hz (y
translation of gimbal).
The relative motion between the gimbal and mass is recorded
by the nano-grating. The simulation predicted a y-displacement
of ∼20 fm corresponding to 1 nG applied acceleration. The ydisplacement for a 1 G acceleration in the y-direction was ∼
incident wavelength, = 850 nm;
grating period, 0.84 ␮m;
grating width, w = 0.415 ␮m;
air gap, ga = 0.38–0.5 ␮m;
thickness of lower grating, t = 0.83 ␮m;
thickness of upper grating, t = 0.83 ␮m.
4.3. Fabrication
The accelerometers were fabricated at the Microsystems Development Laboratory (MDL) and Compound Semiconductor Research
Laboratory (CSRL) within Sandia National Labs in Albuquerque,
New Mexico. The fabrication process combined surface micromachining and deep reactive ion etching (DRIE) techniques. The
devices were fabricated in silicon. The two-layer polysilicon subwavelength nano-gratings for the optical displacement sensor were
fabricated in a surface micromachining process that has been published previously [2]. This involved fabrication of lithographically
defined sub-micron structures with tight tolerances [2]. Following the fabrication of the optical displacement sensors, the wafers
underwent a high-aspect ratio Deep Reactive Ion Etch (DRIE) of
the substrate silicon to structurally define the proof-mass and
the mechanical springs for both the proof-mass and its gimbal.
A layer of silicon dioxide was deposited over the optical gratings
to protect them from the silicon DRIE processing that followed. The
silicon wafers were patterned and etched from the backside. The
oxide layer under the nano-gratings provided an etch-stop for the
DRIE silicon etch. The DRIE process had to be optimized to ensure
uniform etching of the springs and through-wafer vias with target
aspect ratios >1:20. This was followed by a hydrofluoric acid (HF)
dip to remove buried oxide layers and release the optical sensors.
Finally, a CO2 supercritical drying step is used to prevent stiction of
the fragile nano-gratings to the substrate. Fig. 6 outlines the fabrication process for both the optical sensors and the integrated proof
mass.
Due to the high sensitivity of these devices, our yield from the
release and dicing steps was very low. We had to develop special
operating procedures for both steps to improve the yield of our
devices.
Two accelerometer designs with integrated proof-masses were
fabricated. Large proof-masses (18.9 mg and 33.6 mg, respectively) were chosen to lower the resonant frequency of the
device.
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
287
Fig. 6. Process flow for integrated nano-grating accelerometer fabrication. (a) Deposit oxide isolation layer (∼2 ␮m). (b) Deposit and pattern first polysilicon layer (0.82 ␮m).
(c) Deposit and pattern first sacrificial oxide layer (0.5 ␮m). (d) Deposit and pattern second polysilicon layer (0.82 ␮m). (e) Deposit second sacrificial oxide layer (2 ␮m) to
protect structures. (f) Grind and polish backside followed by patterning springs and mass.
Table 1
Nano-grating accelerometer dynamic characteristics measured in air
Device
1
2
Resonant frequency (Hz)
Q
Mass
Gimbal
Mass
Gimbal
44
43
1101
1131
22
18
83
113
4.4. Experimental results
4.4.1. Device set 1 (18.9 mg proof mass device)
Our first set of released devices were the lower mass (18.9 mg)
accelerometers. Although the large proof masses and weak
mechanical springs of this design survived the release process, the
nano-grating sensors did not survive.
Fig. 7. Phase/amplitude frequency response characteristics (in air) for nano-grating
accelerometer with lower-mass device (18.9 mg mass).
We successfully verified/tested the mechanical characteristics
of this device with a Laser Doppler Vibrometer (LDV) and a shaker
table setup. The device was mounted on a shaker table. Specific
accelerations were applied to the device and the corresponding relative motion of the mass that would be measured by the integrated
displacement sensor was externally measured with a LDV. We performed both dynamic and static mechanical tests on these smaller
mass accelerometers to verify our calculations.
We measured mass resonances as low as 43 Hz and a Q = 20 on
these lower proof-mass devices (Table 1) (Fig. 7).
We verified static displacement of the device for applied accelerations within the measurement system limits. A linear relationship
between the nano-grating displacements and applied acceleration
was observed (Fig. 8). The measured results were 8× better than
the simulation results. This can be attributed to the mechanical
Fig. 8. Verification of static characteristics of lower-mass nano-grating accelerometer using LDV setup on shaker table. Device performance is better than predicted
because fabricated mass springs were weaker than original design.
288
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
The results of these mechanical system tests for the lower mass
device gives us the mechanical operating limits for this accelerometer. They also give us performance verification for the accelerometer
independent of the nano-grating sensor.
Due to the high sensitivity of these devices to very low accelerations/forces, they were inherently sensitive to handling and
require specialized packaging. Our fragile optical gratings were very
susceptible to breakage during release/packaging. We designed
and developed a specialized package to provide necessary stops
to protect the device from high shock conditions. These packages were made by laser machining Quartz followed by careful
alignment and epoxy-based adhesion of the device chip to the
package. Care was taken to minimize the bond line to improve
robustness and prevent breakage of the gratings during shock conditions.
Fig. 9. Phase/amplitude frequency response characteristics (in air) for packaged
higher mass nano-grating accelerometer (33.6 mg mass)
springs being weaker than the original designs due to overetching
in the DRIE process.
We measured displacements of 1 nm for accelerations of 10 ␮G
and 1 ␮m displacements for accelerations
of 10 mG. This yields
√
an electronic noise floor of 0.5 nG/ Hz at frequencies as low as
10 mHz. The proof mass thermal
√ noise floor calculated from the
following equation is 11.7 nG/ Hz:
anoise,rms =
4kb Tωo
mQ
(7)
where kb = Boltzmann constant, and T = absolute temperature
(K).
We observed large mode separation between adjacent modes
with large spring constant ratios: ky /kz > 100 and ky /kx > 105 . Spring
constant ratio between gimbal and mass, ky gimbal /ky mass = 625.
Hence, this design provided significant decoupling between modes
in the x, y and z directions for the sensor.
4.4.2. Device set 2 (33.6 mg proof mass device)
With careful packaging and handling, we successfully released
and tested the larger mass (33.6 mg) accelerometers with integrated nano-grating sensors intact.
Similar to the previous device set, we used a Laser Doppler
Vibrometer (LDV) and shaker table to perform dynamic mechanical
tests on these larger mass devices. We measured resonant frequencies as low as 36 Hz (Fig. 9). To the best of our knowledge, these are
the lowest resonances for MEMS accelerometers measured to date.
Considering the complexity in fabrication and handling of these
large MEMS “micro-structures”, this is a significant achievement for
such devices that need to operate in open-loop without the benefit
of feedback systems.
To obtain direct measurements from our integrated sensor, we built an experimental setup to measure the optical
output from our nano-gratings sensor (Fig. 10). This setup
included an 850 nm laser diode source, several beam splitters
to direct/re-direct the optical path and custom reference detection circuitry [5] to lower the noise floor of the low-noise
photodiode. We used a high precision 3-axis piezo-controlled
flexure stage (NanomaxTM from Thorlabs Inc., with 5 nm resolution) to provide the extremely small displacements/accelerations
needed. The intensity of the optical beam reflected from the integrated nano-grating sensor was measured with the photodetector
Fig. 10. Experimental setup for optical testing of nano-grating accelerometers showing packaged device, optical setup, 5 nm resolution piezo-actuation stage and photodetector with reference detection electronics.
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
289
sensitive accelerometers. We have built integrated masses with resonant frequencies as low as 36 Hz. Due to the extreme sensitivity
of these devices, we developed custom packaging to compensate
for the inherent fragility of these devices. We measured sensitivity as high as 590 V/G
√ for the 33.6 mg mass device. With
−5
V/
Hz, corresponding to a resolution of
noise floors
of
10
√
17 nG/ Hz. This was about two
√times higher than the theoretical
thermal noise floor of 8 nG/ Hz. To the best of our knowledge these are the most sensitive MEMS accelerometers built to
date.
Acknowledgements
Fig. 11. Spectral noise characteristics of nano-grating accelerometer (33.6 mg mass)
with 10 Hz signal input.
circuitry described earlier and correlated with the applied acceleration.
We directly measured output signals of 3.80 mV for applied
accelerations as small as 6.45 ␮G. This corresponds to
√a sensitivity
of 589 V/G. The noise floor was measured at 10
√␮V/ Hz (Fig. 11).
This corresponds to a noise floor of 17 nG/ Hz. The theoretical thermal
noise floor was calculated, based on Eq. (7), to be
√
8 nG/ Hz. The measured dynamic range of our device spans 6
orders of magnitude (106 ).
To the best of our knowledge, these are the most sensitive MEMS
accelerometers built to date and are 40 dB more sensitive than the
best reported in-plane MEMS accelerometers [7,8].
5. Discussion
These nano-grating based optical accelerometers demonstrate
some of the highest sensitivities achievable to date. The strengths
of this device are the use of a high-sensitivity optical displacement sensing technique along with a low noise photodetection
circuitry to lower the 1/f noise floor. An inherent weakness of
this device is its high-sensitivity combined with an open-loop
scheme. This requires a large mass (from a microsystem perspective) and weak springs along with high cross-axis rejection
for successful operation. Our through wafer fabrication allowed
us to build such a device successfully. However, the device
is also inherently susceptible to random accelerations during
routine handling. Careful packaging of our device was critical
and necessary in improving its robustness and obtaining our
results. Alternatively, an integrated package or a closed loop
approach could be used in improving the robustness of such a
device.
6. Conclusions
In this work, we successfully demonstrated a highly sensitive nano-grating-based optical accelerometer.√
We have previously
shown displacement measurements of 12 fm/ Hz at 1 kHz based
on this optical transduction mechanism. We used previously
demonstrated circuit techniques√
to improve sensitivity at very low
frequencies by 77 dB to 50 fm/ Hz at 10 mHz. We successfully
developed capabilities to integrate large masses with fragile surface micromachined optical nano-gratings and create extremely
This work was supported by the Laboratory Directed Research
and Development (LDRD) program at Sandia National Laboratories.
Sandia National Laboratories is a multiprogram laboratory operated by the Sandia Corporation, Lockheed Martin Company, for the
United States Department of Energy’s National Nuclear Security
Administration under contract DE-AC04-94AL85000. We would
also like to acknowledge David Epp for shaker table measurements
of the nano-grating accelerometers. We would also like to acknowledge the staff of the Sandia Microsystems Development Laboratory
(MDL) and Compound Semiconductor Research Laboratory (CSRL)
who contributed to the fabrication of these devices.
References
[1] N.C. Loh, M.A. Schmidt, S.R. Manalis, Sub-10 cm3 interferometric accelerometer
with nano-g resolution, J. Microelectromech. Syst. 11 (June (3)) (2002) 182–187.
[2] B.E.N. Keeler, D.W. Carr, J.P. Sullivan, T.A. Friedmann, J.R. Wendt, Experimental
demonstration of a laterally deformable optical nanoelectromechanical system
grating transducer, Opt. Lett. 29 (11) (2004) 1182–1184.
[3] D.W. Carr, G.R. Bogart, B.E.N. Keeler, Femto-photonics: optical transducers utilizing novel sub-wavelength dual layer grating structure, in: Proc. of the Hilton
Head Solid-State Sensors, Actuators, and Microsystems Workshop, June, 2004,
pp. 91–92.
[4] D.W. Carr, J.P. Sullivan, T.A. Friedmann, Laterally deformable nanomechanical
zeroth-order gratings: anomalous diffraction studied by rigorous coupled-wave
analysis, Opt. Lett. 28 (September (18)) (2003) 1636–1638.
[5] R.H. Olsson III, B.E.N. Keeler, D.A. Czaplewski, D.W. Carr, Circuit techniques for
reducing low frequency noise in optical MEMS position and inertial sensors, in:
IEEE International Circuits and Systems Conf., May, 2007, pp. 2391–2394.
[6] P.C.D. Hobbs, Ultrasensitive laser measurements without tears, Appl. Opt. 36
(February (4)) (1997) 903–920.
[7] B.V. Amini, F. Ayazi, Micro-gravity capacitive silicon-on-insulator accelerometers, J. Micromech. Microeng. 15 (2005) 2113–2120.
[8] J. Chae, H. Kulah, K. Najafi, A monolithic three-axis micro-g micromachined silicon capacitive accelerometer, J. Microelectromech. Syst. 14 (2) (2005) 235–242.
Biographies
Uma Krishnamoorthy received the MS and PhD degree in
electrical engineering from University of California, Davis,
in 1999 and 2002. She was a Postdoctoral Researcher at E.
L. Ginzton Laboratory, Stanford University, CA, from 2002
to 2004. In the course of her research, she has developed
enabling technologies for design of reliable optical microscale components in optical switching, scanning and
spectroscopy applications. She has authored/co-authored
one book chapter and >25 journal and conference papers.
Currently, she is a Senior Member of Technical Staff
at Sandia National Laboratory in Albuquerque, NM,
where she works on the design of low-g MEMS optical
acceleration sensors and satellite components.
Roy H. Olsson III received the MS and PhD degrees in electrical engineering from
the University of Michigan, Ann Arbor, in 2001 and 2004. In July 2004, he joined
the Advanced MEMS Group at Sandia National Laboratories, Albuquerque, NM, USA,
where he is currently a Principal Member of the Technical Staff. His research interests
include: RF MEMS resonators, micromachined acoustic bandgap devices, microinertial sensors and MEMS neural interfaces, where he has authored in excess of
20 journal and conference papers.
290
U. Krishnamoorthy et al. / Sensors and Actuators A 145–146 (2008) 283–290
Greg R. Bogart received his BS degree in chemistry from
St. Cloud State University in St. Cloud, MN, and his PhD in
chemistry from Colorado State University in Fort Collins,
CO. His research interests involve development and transferring to manufacturing new technologies. He holds 12
US patents and is currently employed at Symphony Acoustics, Inc., in Rio Rancho, NM, where he is Vice President of
Engineering.
Michael S. Baker received his BS and MS degrees in
mechanical engineering from Brigham Young University,
Provo, UT, in 1999 and 2002. His research efforts were in
the area of on-chip actuation of MEMS bistable mechanisms. He is currently a Senior Member of the Technical
Staff in the Advanced MEMS Technologies department
at Sandia National Laboratories, Albuquerque, NM. His
research interests include compliant mechanism design
and methods of actuation in MEMS.
Dustin W. Carr received his PhD in physics from Cornell University in 2000, and
his BS in mathematics from Oklahoma State University in 1994. While a graduate
student his thesis research was on nanomechanical silicon structures. He was also a
staff member at the Cornell Nanofabrication Facility before joining Bell Laboratories,
Lucent technologies in Murray Hill, NJ, in 1999. While at Lucent, he worked on MEMS
fabrication for very large-scale optical cross-connects for use in telecom networks.
He managed a group that helped to transition a world class 200 mm CMOS research
fab into one of the world’s only 200 mm MEMS fabrication facilities. In 2003, he left
Lucent to join the technical staff at Sandia National Laboratories in Albuquerque.
While at Sandia, he studied ways to integrate nanophotonics and MEMS structures
for sensor applications. This work served as the motivation for the founding of Symphony Acoustics in 2006, where he is currently the chief technology officer, and is
working on optoelectronics and MEMS integration for acoustic and vibration sensing. Dr. Carr is widely published in peer-reviewed publications across a range of
topics in optics, MEMS, and nanofabrication. He is an inventor on 8 issued patents
with several more patents pending. In 2004 he was recognized by MIT’s Technology
Review Magazine as one of the TR100 top 100 young innovators.
Thomas P. Swiler is a Senior Member of the Technical Staff at Sandia National
Laboratories in the Thin Films, Vacuum, and Packaging Department. He received
a BS in ceramic science/BA in physics from Alfred University in 1986, and an MS
and PhD in materials science and engineering from University of Florida in 1988
and 1994, respectively. He is currently employed in microelectronics and MEMS
packaging, and his interests include devising novel package designs using lowtemperature cofired ceramics and modeling thermal and mechanical issues in
packages.
Peggy J. Clews is a Senior Member of the Technical Staff at Sandia National Laboratories in the MESAFab Operations Department. She received a BS in chemistry from
Nebraska Wesleyan University in 1980 and a BS in chemical engineering from Washington University in St. Louis in 1981. She is currently employed as a process engineer
in the MESA Fab complex which processes both CMOS and MEMS technologies. She
is interested in developing release and dry processes for unique MEMS devices and
applications.