T E C H N I C A L P A P E R
MEMS design and fabrication of an electrostatic
vibration-to-electricity energy converter
Yi Chiu Æ Chiung-Ting Kuo Æ Yu-Shan Chu
Received: 28 June 2006 / Accepted: 20 November 2006 / Published online: 4 January 2007
Springer-Verlag 2007
Abstract This paper presents a micro electrostatic vibration-to-electricity energy converter based on the micro-electromechanical system. For the 3.3 V supply voltage and 1 cm2chip area constraints, optimal design parameters were found from theoretical calculation and Simulink simulation. In the current design, the output power is 200 lW/cm2 for the optimal load of 8 MW. The device was fabricated in a silicon-on-insu-lator wafer. Mechanical and electrical measurements were conducted. Residual particles caused shortage of the variable capacitor and the output power could not be measured. Fabrication processes are being refined to remove the back silicon substrate to eliminate residual particles and parasitic capacitance.
1 Introduction
Due to the advance of CMOS VLSI technology, the power consumption of electronic devices has been re-duced considerably. The low-power technology en-ables the development of such applications as wireless sensor networks (Rabaey et al. 2000) or personal health monitoring (Tashiro et al.2000), where remote or independent power supply is critical for building more compact or longer-lifetime systems. In particular,
energy scavenging from ambient natural sources, such as vibration (Roundy et al. 2002), radioisotope (Lal et al.2005) and ambient heat (Douseki et al.2003), is attracting much recent interest as a self-sustainable power source for these applications. Among various approaches, electrostatic vibration-to-electricity con-version using the micro-electromechanical systems (MEMS) technology is chosen in this study due to its compatibility to IC processes and the ubiquity of the energy source in nature.
The output power of a vibration driven energy converter is closely related to the nature of the vibration source, which must be known in order to design the converter and estimate the generated power. The vibration spectra of several household appliances were measured. A typical vibration source has a peak acceleration of about 2.25 m/s2 at about 120 Hz, as shown in Fig.1. These values are used in the following static and dynamic analysis for the design of the converter.
2 Design
A variable capacitor Cv formed by an in-plane
gap-closing comb structure is the main component in the energy converter (Roundy et al. 2002; Williams and Yates1996), as shown in Fig.2. Figure3 shows a schematic circuit that can be used to extract the con-verted energy. The variable capacitor Cvis charged by
an external voltage source Vinthrough the switch SW1
when Cvis at its maximum Cmax. When Cvis charged
to Vin, SW1 is opened and then the capacitance is
changed from Cmax to Cmin due to the electrode
dis-placement caused by vibration. In this process, the
DTIP2006
Y. Chiu (&) C.-T. Kuo Y.-S. Chu
Department of Electrical and Control Engineering, National Chiao Tung University, 1001 Ta Hsueh Road, Hsinchu 300, Taiwan, ROC
e-mail: [email protected] DOI 10.1007/s00542-006-0348-z
charge Q on the capacitor remains constant (SW1 and SW2 both open). Therefore, the terminal voltage on the capacitor is increased and the vibration energy is converted to the electrostatic energy stored in the capacitor. When the capacitance reaches Cmin(Vmax),
SW2 is closed and the charge on Cvis transferred to a
storage capacitor Cstor. SW2 is then opened and Cv
goes back to Cmax, completing one conversion cycle.
During the period when SW2 is open, the charge on Cstor is discharged by the load resistance RL with a
time constant s = RLCstorbefore it is charged again by
Cv. In the steady state, the initial and final terminal
voltages VLof the discharge process become constant,
as shown in Fig.4.
It can be shown that the steady-state final terminal voltage Vsat in the charge-discharge cycle can be
ex-pressed as Vsat¼ Cmax CstorVin 1þCmin Cstor expðDt=RLCstorÞ 1 , ð1Þ
where Dt = conversion cycle time = 1/2f and f is the vibration frequency. When the voltage ripple of the charge-discharge cycle is small, as will be shown subsequently, the output power can be estimated by
Pout¼ V2
sat
RL , ð2Þ
which is in general proportional to C2 max.
In the comb structure, Cmax is determined by the
minimum finger spacing. In a previous design (Chu et al. 2005), the minimum finger spacing was kept at 0.5 lm to prevent shortage of the uninsulated fingers (Fig. 5a). If a dielectric coating is applied to the side walls of the fingers (Fig.5b), they become insulated and the minimum spacing can be further reduced to increase Cmax and Pout. In this design, the total
capaci-tance between fingers becomes Cdielectric|| Cair|| Cdielectric.
If a layer of 500-A˚ -thick silicon nitride is used as the dielectric material due to its process compatibil-ity and high dielectric constant (er ~ 7), Cmax can be
y z Cmax y z Cmin Displacement due to vibration Displacement due to vibration
Fig. 2 Variable capacitor schematic
Vin Cv Cstor RL
VL
SW1 SW2
Fig. 3 Operation of the electrostatic energy converter
VL SW2 open (discharging) SW2 close (charging) Time Vsat . . . .
Fig. 4 Output terminal voltage VL in the steady-state charge-discharge cycle 0 100 200 300 400 500 0.01 0.1 1 10 Frequency (Hz) Acceleration (m/s 2)
increased by a factor of four compared to the previous design. It is also noted that the dielectric coating barely increases Cmin. Therefore, the expected increase of
output power will not be affected by the change of Cmin.
2.1 Static analysis
In Eq. 1, RL and Cstor can be chosen so that the
dis-charge time constant s = RLCstor is much larger than
the conversion cycle time Dt. The output voltage ripple in the steady state can therefore be neglected. In this case, Vsatcan be approximated as
Vsat¼ CmaxVin Cmin 1þRLDtCminþ
Dt RLCstor
ð3Þ
Usually Cminis a small value (in the order of 100 pF).
The other circuit components in Eq. 3 can then be chosen so that Cstor>> Cmin and RLCmin<< Dt. The
equation can then be simplified as
Vsat
CmaxVin CminRLDtCmin
ð4Þ
The power output becomes
PoutV 2 sat RL CmaxVin Dt 2 RL ð5Þ
For a typical low-power sensor node or module, the minimum output power requirement is about 200 lW. In such a module, a power management circuit is needed to convert the high output voltage to lower ones for various sensing and signal processing units. To be compatible with the power management circuit, the maximum output voltage is limited to about 40 V. With these constraints in Eq. 2, the range of RLcan be
found to be RL £ 8 MW. Even though a smaller RLcan
be used, Cmaxmust then be increased in order to satisfy
the voltage and power requirement (Eqs. 4, 5), which in turn have adverse effects in the dynamic behavior of the converter. Therefore, RL= 8 MW and hence
Cmax= 7 nF are used in the following calculation.
The output power Pout for various Cstorand RL is
shown in Fig.6 for Cstor>> Cmin. It can be seen that
the output power does not depend strongly on the storage capacitor Cstor when Cstor relatively large.
Nevertheless, a large Cstor will result in long initial
charge time when the converter starts to work from a static initial status. Hence, a reasonable Cstorof 20 nF
is used.
From Eq. 1 and with the values of Cstor and RL
obtained from above, input voltage Vin of 3.3 V,
vibration frequency of 120 Hz, and chip area size of 1 cm2, Fig.7 shows the calculated output saturation voltage and power as a function of the initial finger gap distance for various silicon nitride thickness. The finger thickness, length, and width are 200, 1,200 and 10 lm, respectively (Chu et al. 2005). The dimensions of the fingers are based on the available deep etching process capability. The minimum gap distance is assumed to be 0.1 lm, which is controlled by mechanical stops. It can be seen that with a 500-A˚ -thick nitride, the initial
fin-dielectric coating (a)
(b)
Fig. 5 Variable capacitor at Cmaxposition: a without coating, b with dielectric coating
1 2 3 4 5 6 7 8 9 10 5 10 15 20 25 30 35 40 45 50 RL (MOhm) Cstor (nF) 22 0 2 2 0 22 .0 4 4 .0 4 4 .0 44.0 66 .1 66 .1 66 1 8 8 .1 8 8 .1 88.1 1 1 0 .1 1 1 0 .1 11 0.1 1 3 2 .1 1 3 2 .1 132.1 1 5 4 .2 15 4 .2 154.2 1 7 6 .2 17 6 .2 176.2 1 9 8 .2 1 9 8 .2 198.2 2 2 0 .2 22 0 .2
ger gap has an optimal value of 35 lm for a power output of 200 lW and output voltage of 40 V.
2.2 Dynamic analysis
After the dimensions of the variable capacitor are determined from the static analysis, the dynamics of the micro structure is analyzed so that the desired maximum displacement, and hence Cmax, can be
achieved by the target vibration source. The electro-mechanical dynamics of the converter can be modeled as a spring-damper-mass system with the equation of motion,
m€zþ beðzÞ þ bmðz; _zÞ þ kz ¼ m€y; ð6Þ where y is the displacement of the device frame caused by vibration, z is the displacement of the shuttle mass m with respect to the device frame, bmðz; _zÞ is the mechanical damping force representing energy loss caused by the squeezed film effect, and be(z) is the
electrostatic force acting on the MEMS structure. Notice that the mechanical damping bmis a function of
both the displacement z of the shuttle mass and its velocity _z (Roundy et al.2002).
A Simulink model was built to simulate the system dynamic behavior based on Fig.3and Eq. 6, as shown in Fig.8. The charge redistribution box calculates the charging and discharging events when Cvreaches Cmax
or Cmin. This process represents the power output. Due
to the limited shuttle mass that can be achieved in a MEMS process using only silicon, an external attached mass m is considered in order to increase the dis-placement of the variable capacitor and the energy conversion efficiency.
For various attached mass, Fig.9 shows the maxi-mum achievable displacement and required spring constant. It can be seen that a mass of 7.2 g is required to achieve the maximum of 34.8 lm according to the static design. The corresponding spring constant, 4.3 kN/m, is used to design the spring structures. With these values, the output voltage simulated by the Simulink model as a function of time is plotted in Fig.10. The charge-discharge cycles are evident and the saturation voltage Vsat is close to the expected
value of 40 V. Table1 summarizes the important de-vice design parameters according to both the static and dynamic analyses.
3 Fabrication
A schematic device layout is shown in Fig.11. The center hole is used to fix the position of the attached mass. A silicon-on-insulator (SOI) wafer with a 200-lm-thick device layer was used for large capacitance. The oxide layer and the handle wafer are 2 and 500 lm 5 10 15 20 25 30 35 40 45 50 0 100 200 300 400 500
Output power Pout ( uW) 0
20 30 40 50 60
Finger initial gap (um)
O u tp u t sa tu ra tio n vo lta g e V sa t (V o lts) 0.05um-thick nitride 0.1um-thick nitride 0.2um-thick nitride Pout Vsat
Fig. 7 Output saturation voltage and power versus initial finger gap (RL= 8 MW, Cstor= 20 nF) acceleration charge redistribution pull-in detection 1/s velocity 1/s vibration source displacement contact detection
Fig. 8 Dynamic simulation model
2 3 4 5 6 7 8 9 20.0 22.5 25.0 27.5 30.0 32.5 35.0 M a x di s pl a cement ( u m) 2 3 4 5 6 7 8 91 2 3 4 5 6 7 Mass (g) Spring con s ta nt (kN/m)
Fig. 9 Maximum displacement and spring constant versus attached mass
thick, respectively. Figure12shows an earlier fabrica-tion process without the dielectric coating. The vari-able capacitor structure is first defined by deep reactive ion etching (DRIE) (Fig.12a). After the sacrificial oxide layer is removed using HF solution (Fig. 12b), aluminum is evaporated for electrical contact (Fig. 12c). A steel ball is then attached to the central plate to adjust the resonant frequency to match the vibration source and improve the conversion efficiency (Fig. 12d).
The fabricated first-generation device is shown in Fig.13(Chu et al. 2005). The width of the finger was reduced to 6.8 lm due to the variation in photoli-thography and RIE processes. The deviation will affect the characteristics of the converter such as the resonant frequency, output power, and output voltage.
4 Measurement
4.1 Mechanical measurement
The displacement of the device without the attached mass was measured using a PROWAVE JZK-1 shaker. The measured response is shown in Fig.14. Since the mass was not attached, the vibration acceleration was increased to 40 m/s2 for easy observation. The maxi-mum displacement was about 10 lm at 800 Hz, and the quality factor Q=x0/Dx was about 10, where x0 was
the resonant frequency and Dx was the resonant
0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 30 35 40 45 Time (s) O u tp u t vol ta ge ( Vol ts )
Fig. 10 Output voltage versus time
Table 1 Design parameters of the energy converter
Parameter Description
W Width of shuttle mass 10 mm
L Length of shuttle mass 8 mm
Lf Length of finger 1,200 lm
Wf Width of finger 10 lm
m Shuttle mass 7.2 g
d Initial finger gap 35 lm
dmin Minimum finger gap 0.1 lm
Cstor Storage capacitance 20 nF
K Spring constant 4.3 kN/m
T Dielectric layer thickness 500 A˚
er Dielectric constant 7 (SiN)
RL Load resistance 8 MW
Vsat Output voltage ~40 V
Pout Output power ~200 lW
Fig. 11 Device layout schematic
Si Al
a) c)
b) d)
Fig. 12 Fabrication process: a define structure by DRIE, b etch oxide by HF solution, c apply Al by thermal evaporation, and d attach external mass
bandwidth shown in Fig.14. The mass of the center plate was approximately 0.038 g, thus the spring con-stant can be calculated as k¼ x2
0m¼ 960 N/m. The measured spring constant is different from the design mainly due to the reduced spring width as shown in Fig.13.
4.2 Electrical measurement
The electrical measurement was conducted using an INSTEK-LCR-816 LCR meter and a HP-4192A impedance analyzer. The measured capacitance with-out vibration was abwith-out 500 ~ 600 pF, while the cal-culated capacitance Cmin is about 50 pf. The major
contribution of the large capacitance is the parasitic capacitance Cpar between the center plate and the
underneath substrate.
In addition to the parasitic capacitance, there was also a parallel parasitic conductance. The measured conductance varied from die to die with an average resistance of 2.5 kW. It is suspected to be caused by the residual particles left in the device after the release process. The presence of the parasitic capacitance and conductance had hindered the measurement of output power. New devices are being fabricated with the underneath substrate removed to prevent residual particles and reduce parasitic capacitance.
5 Conclusion
The design and analysis of a micro vibration-to-elec-tricity converter have been presented. The device was fabricated in a SOI wafer. The reduced feature size of the fabricated device resulted in the decrease of spring constants. Mechanical and electrical measurements of the fabricated device were conducted. Impedance measurements showed an unwanted parasitic conduc-tance which resulted in the failure of output power measurement. Improvement of the fabrication pro-cesses is being conducted.
Acknowledgment This project was supported in part by the National Science Council, Taiwan, R.O.C, under the grant No. NSC 93-2215-E-009-066. The authors are grateful to the National Center for High-performance Computing, Taiwan, ROC, for computer time and facilities.
References
Chu YS, Kuo CT, Chiu Y (2005) A MEMS electrostatic vibration-to-electricity energy converter. In: Proceedings of PowerMEMS, pp 49–52
Douseki T, Yoshida Y, Utsunomiya F, Ito N, Hama N (2003) A batteryless wireless system uses ambient heat with a reversible-power-source compatible CMOS/SOI DC-DC converter. In: Proceedings of the IEEE international solid-state circuits conference, pp 388–389
Lal A, Duggirala R, Li H (2005) Pervasive power: a radioiso-tope-powered piezoelectric generator. IEEE Pervasive Comput 04(1):53–61 width ~ 6.8 µm center plate mecha-nical stop variable capacitor depth ~ 200 µm spring (a) (b) (c)
Fig. 13 Fabricated device: a top view, b cross section of comb fingers, c overview of the converter with attached mass
500 600 700 800 900 1000 0 1 2 3 4 5 6 7 8 Frequency(Hz) Re s pon s e (|z/y|) measured data fit
∆
ω
ω
0Rabaey JM, Ammer MJ, da Silva JL Jr, Patel D, Roundy S (2000) Picoradio supports ad hoc ultra low-power wireless networking. IEEE Comput Mag 33:42–48
Roundy S, Wright PK, Pister KS (2002) Micro-electrostatic vibration-to-electricity converters. In: Proceedings of 2002 ASME International Mechanical Engineering Congress and Exposition, pp 393–409
Tashiro R, Kabei N, Katayama K, Ishizuka Y, Tsuboi F, Tsuchiya K (2000) Development of an electrostatic gener-ator that harnesses the motion of a living body: (use of a resonant phenomenon). JSME Int J ser C 43(4):916–922 Williams CB, Yates RB (1996) Analysis of micro-electric
