IEEE SENSORS JOURNAL, VOL. XX, NO.

XX, MONTH X, XXXX 1

Active Temperature compensation for MEMS

capacitive sensor
Cuong Do, Member, IEEE, Ashwin A. Seshia, Senior Member, IEEE

Abstract — Temperature variation is one of the most crucial factors that

need to be cancelled in MEMS sensors. Many traditional methodologies
require an additional circuit to compensate for temperature. This work
describes a new active temperature compensation method for MEMS
capacitive strain sensor without any additional circuit. The proposed method
is based on a complement 2-D capacitive structure design. It consumes zero-
power, which is essential toward the realization of a low-power temperature-
compensated sensor in battery-powered or energy-harvesting applications.
The gauge factor of the developed MEMS capacitive strain sensor is 7. The best result showing a capacitance variation
of 1 ppm/oC compared with 13 ppm/oC on conventional design.

Index Terms—Active temperature compensation, Complement 2-D capacitive structure, Strain gauge.

I. Introduction

C APACITOR is a device that consists of two electrodes

separated by an insulator. For parallel plate capacitors, the
capacitance can be expressed as:

C= (Aε0εr)/d (1)

where C is the capacitance, A is the overlapping area of the two

plates, ɛ0 is the permittivity of free space, ɛr is the relative
permittivity of the dielectric material, and d is the distance
between the plates. Capacitive sensors are realized by varying
any of the three parameters of a capacitor: the distance between
two plates, overlapped area of plates, and the dielectric constant
of an insulator.
Based on those physical changes, a wide variety of capacitive Fig. 1. Thermal expansion of host structure under test
sensors have been developed, ranging from proximity sensing complex combination of the coefficient of thermal expansion
[1-5], position sensing [6, 7], humidity sensing [8, 9], tilt (CTE) of adhesive and host materials.
sensing [10, 11], strain sensing [12, 13, 14], etc. As the Most materials expand and contract when the temperature
capacitance is not dependent on the properties of the plate changes. Every material has its own coefficient of thermal
material, and thermal expansion is generally isotropic in three expansion (CTE), the change in length with temperature for a
dimensions, capacitive sensors are generally considered to solid material can be expressed as
having low-temperature dependence compared with other
sensor methods, such as resistive based [15] or resonant based 𝛼(T) = [𝑳T − 𝑳𝟎 ]/[𝑳𝟎 ∆T] (2)
[16]. However, in sensing applications where the capacitive
sensor adheres to package or other host materials, such as strain where L0 and LT represent, respectively, the initial and final
gauge as demonstrated in Fig. 1, the coefficient of thermal lengths with temperature change ΔT, the CTE, α(T), is usually
capacitance (CTC) will be affected and proportional to the found to be nonlinear with temperature [17]. For example,
silicon, which is widely used in MEMS structure, has the CTE
found to be varied from 2.57 ppm/K at 293 K to 4.33 ppm/K at
Manuscript received …. 1000 K [18].
Cuong Do was with the Engineering Department at the University In practice, MEMS silicon structure is usually glued onto a
of Cambridge. He is now with the College of Engineering and
Computer Science, VinUniversity, Hanoi, Vietnam (email: package carrier (e.g. polyimide, metals) using an adhesive
cuong.dd@vinuni.edu.vn) material (e.g. super glue, M-bond). Furthermore, the whole
Ashwin A. Seshia is with the Engineering Department at the strain gauge is then bonded onto a host infrastructure material
University of Cambridge, Cambridge, UK (email: (e.g. concrete, steel, aluminium, glass). Thus, the temperature
aas41@eng.cam.ac.uk)
effect on the capacitive sensor becomes extremely complex in

XXXX-XXXX © XXXX IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
8 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, MONTH X, XXXX

TABLE I
COEFFICIENT OF THERMAL EXPANSION (CTE) OF SOME MATERIALS RELATED
TO STRAIN SENSING

Materials CTE (ppm/K)

Polyimide ~20
Aluminium ~23
Super glue ~100
Steel ~11
Concrete ~12
Glass ~8.5

strain sensing applications. CTE of some of the above-

mentioned materials can be found in Table. I [17, 19]. The
CTEs of these materials are a few times higher than that of
silicon. Therefore, temperature effect will be exacerbated on to Fig. 3. MIDISTM MEMS platform process by CMC.
the sensor, including MEMS capacitive strain sensor.
Here, we introduce for the first time, an active compensation noted that, in a comb structure, the fringing effect is
method based on a complement 2-D capacitive structure design independent with the overlapping area. The substrate
to help to reduce the thermal expansion issue on MEMS contraction during cooling is following exactly similar
capacitive sensor without any additional complex behavior.
compensation circuit. If the substrate is isotropic, the expansion to x and y
directions should be equal and hence, Δx1 = Δx2. When B1 and
B2 are connected, the capacitance factor caused by thermal
II. DESIGN CONCEPT
expansion on two branches can be cancelled out and thus the
A concept of 2-D complement capacitive comb structure total capacitance is independent of temperature variation.
design is demonstrated in Fig. 2. The structures AB1 and AB2
are designed identically in term of distance, the number of 𝟒𝑵𝜺𝒓 𝒉(𝒙)
interdigitated fingers, thickness, overlapping area and gap. The 𝑪𝑻 = 𝑪𝑨𝑩𝟏 + 𝑪𝑨𝑩𝟐 = (5)
𝒈
only difference is the fingers orientation layout, as clearly seen
in Fig. 2. When the whole structure is heated, the substrate is
III. THE SENSOR
expanded to all directions [20]. By neglecting the fringing
effect, the capacitance between AB1 and AB2, CAB1, CAB2 can Based on the concept discussed in the previous section,
be found as: MEMS capacitive sensors have been developed and fabricated
on the MIDIS™ Platform, which is being offered as Multi-
𝑪𝑨𝑩𝟏 =
𝟐𝑵𝜺𝒓 𝒉(𝒙−∆𝒙𝟏)
(3) Project-Wafer (MPW) service through CMC Microsystems
𝒈 [21]. The process is 1.5 μm minimum feature size, 1.5 μm
minimum gap, 30 µm silicon thickness membrane and wafer-
𝟐𝑵𝜺𝒓 𝒉(𝒙+∆𝒙𝟐)
𝑪𝑨𝑩𝟐 = (4) level vacuum packaged. More details of the process are shown
𝒈
in Fig. 3. Fig. 4 shows the SEM picture of the fabricated Design
where N is the number of interdigitated fingers on each branch, 1 before it was encapsulated. The design follows the thermal
expansion compensation concept, as discussed above. Branches
h is the thickness of the fingers, x is the initial overlapping
AB3 and AB4 are similar to AB1 and AB2, respectively. B1,
length between fingers, g is the gap between two adjacent
B2, B3, B4 are connected by a top metal layer after
fingers, and Δx1, Δx2 are the changes in the overlapping encapsulation. In order to achieve a high-performance sensor
between fingers due to substrate expansion under heating. It is

Fig. 4. SEM picture of the developed MEMS capacitive sensor with

Fig. 2. An illustration of a temperature compensated capacitive sensor temperature compensation technique, Design 1.
based on complement 2-D capacitive comb structure design.
8 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, MONTH X, XXXX

Fig. 6. Cantilever beam bending test setup for MEMS strain sensors.

plate parasitic is relatively high due to the limited maximum

gap in this process. It will eventually affect the sensitivity of the
sensors as it travels in the opposite direction and inherits a
nonlinear capacitance versus distance relationship.
This is not an ideal comparison as Design 2 is not fully-
symmetrical as Design 1. However, the total number of finger,
NT, fingers gap, g, and the finger overlap, x, are mainly related
Fig. 5. SEM picture of the developed MEMS capacitive sensor without to the range and resolution of the sensors. The temperature
temperature compensation technique, Design 2.
sensitivity is mainly related to the relative finger overlapping,
TABLE II Δx. Hence, this relative comparison is still eligible to prove the
DIMENSIONS AND ESTIMATED PARAMETERS OF THE MEMS CAPACITIVE concept.
SENSORS

Description Design 1 Design 2 IV. EXPERIMENTAL RESULTS AND DISCUSSION

Chip size 1250 x 1250 μm 2
1150 x 820 μm 2
The fabricated sensors are attached by M-bond 200 on a
Device thickness, h 30 μm 30 μm cantilever beam bending test setup for strain sensing
measurement, as shown in Fig. 6. Both sensors are on the same
Total number of fingers, NT 2720 1428
die. The die encapsulation is grounded to remove parasitic
Capacitor finger gap, g 1.5 μm 1.5 μm capacitance of top metal layer, which is AlCu. The chip is wire-
Finger overlap, x 7.5 μm 10 μm bonded to a flexible PCB board for external connection. A
Estimated comb capacitance 7.22 pF 5.05 pF cantilever beam with a length of L, and thickness of t is
Estimated parallel parasitic 1.8 pF 1.3 pF
anchored at one end with the MEMS sensors located at a
position of p from the anchor. The beam material is steel with
with high capacitive and sensitivity, the interdigitated comb length, L, of 220 mm, thickness, t, of 3 mm, and the sensors are
fingers are designed in a compact form, as shown in the zoom- located at p = 35 mm. The displacement is measured by a
in sub-figure. The gap between fingers is designed at the micrometre, and the corresponding strain relationship can be
process minimum of g = 1.5 μm. To reduce the parallel plate found in the figure.
parasitic, the maximum gap is advised to be as big as possible.
However, the maximum distance between adjacent structures
of this process is only at 5 μm, hence some unwanted high
parasitic capacitance is predicted in this design. This will affect
the sensitivity and resolution of the sensors. However, the 2-D
complement capacitive concept described above is still
applicable to the parallel plate capacitor. Therefore, the
parasitic variations will also be cancelled out under temperature
variation.
In order to evaluate the effectiveness of the concept, a normal
design without temperature compensation technique has also
been developed to compare and contrast with the one in Fig. 4.
The SEM picture is shown in Fig. 5. The general design
structure concept of the interdigitated comb fingers layout is
annotated on the top of the SEM figure. When temperature
varies, the MEMS structure is expanded or contracted. Both
AB1 and AB2 capacitance will either increase or decrease at the
same time with temperature.
Table II shows designed dimensions and some estimated Fig. 7. Real-time monitoring and data logging strain sensing setup.
parameters of the two MEMS capacitive designs. The parallel
8 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, MONTH X, XXXX

TABLE III
EVAL-AD7150EBZ BOARD SETTING

Parameters Value Unit

Excitation frequency 32 kHz
Conversion time (sampling
5 (200) ms (Hz)
rate)
Input range 500 fF
Δ-Σ CDC Resolution 12 bit
1-bit resolution 0.012 fF

Fig. 7 presents the real-time measurement setup using a

commercial capacitive interface data acquisition board: EVAL-
AD7150EBZ for capacitance-to-digital converter. It has two
independent input channels. Hence two developed designs
could be tested at the same time. Real-time graphic monitoring
and data logging software is also provided. Fig. 9. Capacitance changes versus temperature.
Parameters setup for EVAL-AD7150EBZ board is shown in
Table III. The input range is set to the lowest possible range are around 0.03 to 0.05 fF. The CTC of Design 2 (conventional
setting (500 fF). At this setting and with 12-bit CDC, the 1 bit design) is found at ~13 ppm/oC, which is similar to the CTE of
resolution is 0.012 fF. The Σ-Δ modulator helps to significantly steel of ~11 ppm/oC as seen in Table. I. Whereas, the CTC of
minimize the noise and therefore helps to enhance the Design 1 (where compliment capacitive structure is utilized) is
measurement accuracy. The conversion time is set by default at found at only 1 ppm/oC. It is clearly shown the effectivity of the
5 ms per sample. proposed method to reduce the temperature effect on MEMS
The measured capacitance of the two designs are 10 pF and capacitive strain sensor.
7.1 pF for Design 1 and 2, respectively. They are ~10% higher
than the estimated values (including comb and parallel parasitic
capacitance). The differences could be contributed by the wire V. CONCLUSION
bonds, PCB and coaxial cable connections parasitic. This paper presents the design concept and characterization
Fig. 8 presents the measured output versus an applied input results for a new method to compensate for the temperature
strain for both designs up to 800 με with a nonlinearity of less effect of capacitance strain gauge. The sensors were fabricated
than 2% of full scale. Design 1 has a slightly higher sensitivity on silicon MEMS technology. The gauge factor is measured at
at ~0.07 fF/με than Design 2, which is measured at 0.05 fF/με. 7. The thermal effect has been assessed at a temperature up to
The equivalent Gauge Factors are found at 7 for both designs. 130oC. The technique requires no addition circuit or special
The temperature dependence characterization is carried out fabrication process.
with an environmental chamber to evaluate the temperature The proposed method improves the CTC of the sensor by 13
compensation effectivity of the concept. Both designs are on the times compared with the conventional design. Future work is
same dice, which adheres to a steel bar with M-bond 200 glue. required for the development of low-power front-end capacitive
Fig. 9 shows the capacitance changes of the two designs for 10
readout to improve resolution and sensitivity. Characterizations
samples each, with temperature from 25 to 130 oC. Error bars
of capacitive strain gauges for higher temperature operation (up
to 1000oC), long-term stability, and degradation mechanisms
are also required in future studies.
This method is also applicable to other types of capacitive
sensor, including proximity sensing, position sensing, humidity
sensing, and tilt sensing. Further research on this topic on how
to effectively employ this technique in those applications is
being carried out.
If we were to design this study again, we would make Design
2 as fully-symmetrical as Design 1 to have a better comparison.

REFERENCES
[1] Ojuroye, Olivia O., et al. "Embedded capacitive proximity and touch
sensing flexible circuit system for electronic textile and wearable
systems." IEEE Sensors Journal, vol. 19, no. 16, 2019.
[2] J. Wang, Jiaqi, Z. Zheng, J. Chan, and J.T. Yeow, "Capacitive
micromachined ultrasound transducers for intravascular ultrasound
imaging." Microsystems & Nanoengineering, vol. 6, no. 1, pp. 1-13,
Fig. 8. Capacitance changes versus strain characterization of two
2020.
developed MEMS sensors.
[3] Anzinger, Sebastian, et al. "Low Power Capacitive Ultrasonic
Transceiver Array for Airborne Object Detection." in IEEE 33rd
8 IEEE SENSORS JOURNAL, VOL. XX, NO. XX, MONTH X, XXXX

International Conference on Micro Electro Mechanical Systems Cuong Do received the B. B.Sc. degree in
(MEMS), 2020. electronics and telecommunication from
[4] T. Emadi, and A.B. Douglas, "A novel 6× 6 element MEMS capacitive Vietnam National University, Hanoi, in 2004,
ultrasonic transducer with multiple moving membranes for high the M.Eng. degree in electronics from
performance imaging applications." Sensors and Actuators A: Physical, Chungbuk National University, South Korea, in
vol. 222, pp. 309-313, 2015. 2007, and the Ph.D. degree in electronics from
[5] R.N. Miles, W. Cui, Q.T. Su, and D. Homentcovschi, “A MEMS low- the Cork Institute of Technology, Ireland, in
noise sound pressure gradient microphone with capacitive sensing”, 2012.
Journal of Microelectromechanical Systems, vol. 24, no. 1, pp. 241- From 2013 to 2017, he was a Research
248, 2015. Associate with the University of Cambridge. Since 2019, he has been
[6] Dong, Xianshan, et al. "Research on temperature characteristic of an Assistant Professor with the College of Engineering and Computer
parasitic capacitance in MEMS capacitive accelerometer." Sensors and Science, VinUniversity, Hanoi. His research interests include fields of
Actuators A: Physical, vol. 285, pp. 581-587, 2019. sensors and sensors fusion for healthcare applications.
[7] E.E. Aktakka, and K. Najafi, “A six-axis micro platform for in situ
calibration of MEMS inertial sensors”, in Proc. IEEE MEMS 29th Ashwin A. Seshia received his BTech in
International Conference, 2016. Engineering Physics in 1996 from IIT Bombay,
[8] Mutharpavalar, Ashaashvini, A. Y. Ahmed, and Nursyarizal Mohd Nor. MS and PhD degrees in Electrical Engineering
"Design, Modeling and Simulation of MEMS Resonator for Humidity and Computer Sciences from the University of
Sensor Application." in Prof. IEEE Sensors Applications Symposium California, Berkeley in 1999 and 2002
(SAS), 2020. respectively, and the MA from the University
[9] N. Lazarus, S.S. Bedair, C.C. Lo, and G.K. Fedder, “CMOS-MEMS of Cambridge in 2008. He joined the faculty of
capacitive humidity sensor”, Journal of Microelectromechanical the Engineering Department at the University
Systems, vol. 19, no. 1, pp. 183-191, 2010. of Cambridge in October 2002 where he is
[10] Rao, Kang, et al. "A high-resolution area-change-based capacitive presently a Professor of Microsystems Technology and a Fellow of
MEMS tilt sensor." Sensors and Actuators A: Physical, vol. 313, 2020. Queens' College. He is a Fellow of the Institute of Physics, a Fellow of
[11] L. Zhao, and E.M. Yeatman, “Micro capacitive tilt sensor for human the Institution for Engineering and Technology and a Fellow of the
body movement detection”, in Proc. Body Sensor Networks, pp. 195- Institute of Electrical and Electronics Engineers. Prof Seshia currently
200, 2007. serves on the editorial boards of the IEEE Journal of
[12] Shirhatti, Vijay, et al. "High-range noise immune supersensitive Microelectromechanical Systems and the IEEE Transactions on
graphene-electrolyte capacitive strain sensor for biomedical Ultrasonics, Ferroelectrics and Frequency Control, and the Executive
applications." Nanotechnology, vol. 30, no. 47, 2019. Committee of the European Frequency and Time Forum. He received
[13] M. Suster, J. Guo, N. Chaimanonart, W. Ko, and D. J. Young, “A High- the 2018 IEEE Sensors Technical Achievement Award (Advanced
Performance MEMS Capacitive Strain Sensing System”, Journal of Career - Sensor Systems) "for pioneering contributions to resonant
Microelectromechanical Systems, vol. 15, no. 5, Oct. 2006. microsystems with application to sub-surface density contrast imaging
[14] J. Li, L.P. Jon, T. Szymon, G. Andrew, and S. Sanjay, "Interdigital and energy harvesting systems".
capacitive strain gauges fabricated by direct-write thermal spray and
ultrafast laser micromachining." Sensors and Actuators A: Physical,
vol. 133, no. 1, pp. 1-8, 2007.
[15] Yaghootkar, B., Azimi, S., & Bahreyni, B. (2017). A high-performance
piezoelectric vibration sensor. IEEE Sensors Journal, vol. 17, no. 13,
pp. 4005-4012, 2017.
[16] C.D. Do, A. Erbes, J. Yan, K. Soga, and A.A. Seshia, “Vacuum
packaged low-power resonant MEMS strain sensor.”, Journal of
Microelectromechanical Systems, vol. 25, no. 5, pp. 851-858, 2016.
[17] F. Cverna, “ASM Ready Reference: Thermal properties of metals”,
ASM International, 2002.
[18] H. Watanabe, N. Yamada, and M. Okaji, “Linear Thermal Expansion
Coefficient of Silicon from 293 to 1000 K”, International Journal of
Thermophysics, vol. 25, No. 1, Jan. 2004.
[19] G. Murray, “Handbook of Materials Selection for Engineering
Applications”, CRC Press, 1997.
[20] Watanabe, H., Yamada, N., & Okaji, M. (2004). “Linear thermal
expansion coefficient of silicon from 293 to 1000 K”, International
journal of thermophysics, vol. 25, no. 1, pp. 221-236, 2014.
[21] https://www.cmc.ca/en/WhatWeOffer/Products/CMC-00200-
02990.aspx.