Abstract
A complete analysis of the morphology, crystallographic orientation, and resulting electrical properties of Pb(Zr0.53,Ti0.47) Pb(Nb1/3, Zn2/3)O3 (PZT-PZN) thin films, as well as the electrical behavior when integrated in a cantilever for energy harvesting applications, is presented. The PZT-PZN films were deposited using sol-gel methods. We report that using 20% excess Pb, a nucleation layer of PbTiO3 (PT), and a fast ramp rate provides large grains, as well as denser films. The PZT-PZN is deposited on a stack of TiO2/PECVD SiO2/Si3N4/thermal SiO2/Poly-Si/Si. This stack is designed to allow wet-etching the poly-Si layer to release the cantilever structures. It was also found that the introduction of the poly-Si layer results in larger grains in the PZT-PZN film. PZT-PZN films with a dielectric constant of 3200 and maximum polarization of 30 μC/cm2 were obtained. The fabricated cantilever devices produced ~300–400 mV peak-to-peak depending on the cantilever design. Experimental results are compared with simulations.
1. Introduction
The ability to retrofit systems with power consuming electronics without having to consider issues associated with providing an independent power source offers a significant advantage for devices in hard to reach locations [1].
In the past few years there has been an increase in research on small wireless electronic devices [1–3]. Mechanical vibrations have received attention as a potential source of power for sensors and wireless electronics in a wide variety of applications [4]. To accomplish this, improving the piezoelectric material used in the cantilever is of paramount importance [5].
Piezoelectric materials are widely used for various devices, including multilayer capacitors, sensors, and actuators [2, 3, 6]. By the 1950s, the ferroelectric solid solution Pb(Zr1−x Tix)O3 (PZT) was found to have exceptionally high dielectric and piezoelectric properties for compositions close to the morphotrophic phase boundary (MPB) [7, 8].
Research to improve the PZT properties, from the material and electrical standpoint, has been mainly accomplished by doping conventional PZT with different elements to “relax” the material [5, 9]. Previous investigations on the dielectric and electrical properties of many ceramic systems, such as barium titanate (BT), lead zirconate titanate (PZT), lead magnesium niobate (PMN), lead titanate (PT), PMN-PT, PZT-BT, and PMN-PZT have demonstrated the importance of the subject [10–13]. Recently, there has been a great deal of interest in the lead zirconate titanate-lead zinc niobate Pb(Zr0.53Ti0.47)O3-Pb(Zn1/3Nb2/3)O3 or (PZT-PZN) system because of its high dielectric, piezoelectric, and ferroelectric properties [14–16]. It is important to have not only a good material, but also a device design that matches the material properties. For this reason, the search for different device designs is of great interest. The cantilever structure is the most efficient structure that has been reported to date [2, 4].
In this paper, the synthesis and characterization of the PZT-PZN thin films for energy harvesting applications are presented. The films were subsequently integrated using silicon-compatible microfabrication techniques to fabricate cantilevers using a wet-etch release process.
Additionally, packaging and electrical characterization such as capacitance, polarization, and dielectric constant versus voltage were performed. Experimental voltage output at the resonance frequencies of the cantilevers is compared with simulations.
2. Experimental
The solutions used for the fabrication of PZT-PZN films were prepared using a sol-gel process that we previously described for Pb(Zr1−x,Tix)O3 [17, 18]. The starting precursors, zirconium (IV) butoxide, titanium (IV) isopropoxide, and niobium (V) ethoxide, were reacted in acetic acid (95%) and methanol. Lead (IV) acetate and zinc acetate dihydrate were added, and the solution was heated to 85°C to dissolve the acetates. The final PZT-PZN solution has a concentration of 0.4 M. A nucleation PbTiO3 (PT) layer was also prepared in a similar manner, but with using titanium (IV) isopropoxide and lead (IV) acetate. The Pb1.2(Zr0.53,Ti0.47)O3-Pb(Zn1/3,Nb2/3)O3 films were deposited from the solution using a spin-coating technique at 3000 rpm for 30 s, followed by a soft bake on a hot plate at 300°C for 10 min to evaporate the solvents. Finally, the crystallization step is done at 675°C for 30 min in a tube furnace in an air atmosphere, as previously described [9].
The PZT-PZN polycrystalline thin film was grown on a stack consisting of 3 μm of poly-Si, 100 nm of thermal SiO2 and 350 nm of Si3N4 deposited using LPCVD, 500 nm of PECVD SiO2 and 50 nm of sputtered Ti (Figure 1). The entire stack is annealed at 1000°C for 30 min in O2 to completely oxidize the Ti to TiO2. Next, the stack is subjected to a 4-step photolithography process to fabricate the cantilevers. The first step is the metallization shown in Figure 2(a). The second step is the encapsulation of the device with 100 nm of PECVD SiO2, Figure 2(b); the third step is the deposition of 50 nm Cr/500 nm Au for packaging-bonding purposes, shown schematically in Figure 2(c). Finally the fourth step is the release of the cantilever as shown in Figure 2(d). The release is carried out using 20% HF to etch the PZT-PZN, TiO2 and SiO2 layers. Next, a RIE plasma-etch step is used to remove the Si3N4 and the poly-Si layer is etched using 20% KOH at 50°C. This last step releases the cantilever. Images of the resulting cantilevers are shown in Figure 3, where (a) and (c) are optical microscopic images of two different cantilever designs, and Figures 3(b) and 3(d) are SEM micrographs of the same cantilever designs. The released cantilevers were mounted and wire-bonded to a chip carrier for poling and output voltage measurements, as show in Figure 4.
(a)
(b)
(c)
(d)
(a)
(b)
(c)
(d)
(a)
(b)
3. Results and Discussion
3.1. Material Characterization
The deposition of PZT-PZN films on the TiO2 buffer layer was optimized by varying the lead content, ramp rate, and insertion of a nucleation layer between the film and the underlying TiO2, as shown in Figure 1. Optimum conditions were 20% excess of Pb, direct sample insertion into the furnace previously heated to 675°C using what we call “a fast ramp rate” (a few seconds to insert the samples manually into the preheated furnace), and the use of a PbTiO3 (PT) nucleation layer between TiO2 and the PZT-PZN [9, 19, 20].
SEM results shown in Figure 5(a) indicate a dense film and large grains size (150 nm), which is a difficult characteristic to achieve in PZT-PZN thin films, but is essential for optimal electrical performance. A typical XRD diffraction pattern for the PZT-PZN is shown in Figure 5(b). XRD results demonstrate the absence of pyrochlore, but show a preferred perovskite orientation in the (110) plane.
(a)
(b)
(c)
AFM results (Figure 5(c)) show that the average roughness of the PZT-PZN films is 12 nm. The increased roughness is due to the poly-Si layer. However, the poly-Si layer is required for the wet-etch release of the cantilever. The results using the optimum conditions could be understood as a competition between two effects: lead loss and perovskite phase crystallization [9].
3.2. Electrical Characterization
The electrical characterization of the PZT-PZN thin film was done using interdigitated electrodes (IDE) on the released cantilevers. Capacitance versus voltage (C-V) measurements were performed on the PZT-PZN films to calculate the dielectric constant. C-V results are shown in Figure 6(a), where the maximum capacitance (20 pF) is observed at 5 V. The calculated dielectric constant for PZT-PZN is shown in Figure 6(b). These calculations were based on studies previously reported [21, 22]. The maximum dielectric constant calculated was 3200 at 5 V. Polarization measurements were also performed on the relaxor thin film and the results are shown in Figure 6(c). The films show a maximum polarization value of 31.2 μC/cm2 and a remnant polarization of 28.5 μC/cm2.
(a)
(b)
(c)
It is well known that a larger grain size can improve the dielectric constant of perovskite materials [23]. Therefore, the high dielectric constant and polarization results are attributed to the larger grain size observed in the PZT-PZN thin films.
In order to measure the output voltage due to vibrational excitation, the films were exposed to a poling step. The poling is accomplished by applying an electric field across the material at elevated temperatures; for thin films, longer poling times are typically required compared to bulk ceramics [24, 25].
The poling was performed after mounting and wire-bonding the devices to a chip carrier, as shown in Figure 4. The poling conditions were 200 kV/cm at 200°C for 50 min. C-V measurements before and after poling are shown in Figure 7. Two device structures are shown in this figure, “linear-shaped” (Figure 7(a)) and “snake-shaped” (Figure 7(c)) cantilevers. C-V results from both cantilevers increased after wire bonding because of an increase in contact area. If we compare both C-V measurements, we find that the capacitance for the “snake-shaped” cantilever is higher than for the “linear-shaped” cantilever. This may be explained because the contact area is higher in the “snaked-shape” cantilever.
(a)
(b)
(c)
(d)
The C-V curve after poling shifted by an average of 5 V from the original position, as show in Figures 7(b) and 7(d).
The cantilevers were subsequently tested by mounting the packaged device to a mechanical shaker (Bruel and Kjaer Instruments Inc. Model Type 4810). The force generated by the shaker is proportional to the acceleration, which is controlled by the applied voltage to a power amplifier (Pyle PLA2200). The amplifier was driven by a 12-V power supply and signals from the PC frequency generator. The output voltage from the cantilever was monitored using a Tektronix digital oscilloscope (TDS 210), as shown in Figure 8.
Voltage-output results for the two cantilever designs shown in Figure 7 are presented in Figure 9, where (a) is for the “linear-shaped” 2000 μm long cantilever, and (b) is for the “snake-shaped” cantilever. The “snake-shaped” cantilever shows an output voltage value of 400 mV peak-to-peak while the “linear-shaped” cantilever shows a 300 mV p-p output voltage at the resonant frequencies of 875 Hz and 960 Hz, respectively. The differences are consistent with simulation-based results using the finite element method (FEM). The results shown in Figure 10 show that for the “linear-shaped” cantilever (a), the area of highest stress is near the tip area, while for the “snake-shaped” cantilever (b), there is more stress distributed across the structure.
(a)
(b)
The FEM (COMSOL) package was used also to model the structural deformation and the voltage output at the resonant frequency, as determined from experimental analysis of the tested cantilevers. Results are shown in Figure 11, where (a) and (c) are the results of the simulated deformation at the respective resonant frequencies; for the “linear-shaped” cantilever a deformation of 104.6 μm at 960 Hz and 47.1 μm at 875 Hz for the “snake-shaped” cantilever.
(a)
(b)
(c)
(d)
For the “linear-shaped” cantilever the experimental results show an output voltage of 300 mV over a range of frequencies of 983 to 1053 Hz. Comparing this result to the simulation shown in Figure 10(b), we find that there is a 2% of difference between the experimental and calculated results. For the “snake-shaped” cantilever the experimental result showed a maximum voltage of 400 mV from 858 to 890 Hz with a difference of 2% between the simulated and experimental results, as shown in Figure 11(d). We believe that the voltage output can be improved by adding a weight at the end of the cantilever to reduce the resonant frequency and improve the deflection of the cantilevers, and consequently increase the voltage output [26].
4. Conclusions
PZT-PZN thin film and microcantilever fabrication processes have been developed to harvest vibrational energy. Integrated cantilevers were successfully fabricated using a wet-etch process to release the cantilevers. The properties of the relaxor films were optimized to yield the desired phase on the cantilever devices. A dielectric constant of 3200 and polarization of 30 μC/cm2 were obtained. A 300 to 400 mV p-p output voltage, depending on the cantilever design tested, was obtained. Simulations are in good agreement with experimental data.
Acknowledgments
The authors would like to thank to NSF Phase I SBIR grant and the Texas Emerging Technology Fund for partial financial support.