Project Brief

Please click on each title below to see content.

1. State of the Art on Mechanical Energy Harvesting and Piezoelectric Generators2. Project Objectives3. Manufacturing Process4. Characterization5. Modeling6. List of Equipment Used in This Project7. Bibliography

1. State of the Art on Mechanical Energy Harvesting and Piezoelectric Generators

Sustainable micro power supplies for wireless and mobile electronics is a strategic field in today’s energy research, which could offer a viable solution to provide the energy required by microsystems. Energy generation utilizing piezoelectric materials has been well-studied over the past two decades. These materials convert a physical pressure into the motion of electrons, and thus, the unused mechanical energy of our surroundings into electrical energy. Other types of mechanical energy harvesters (MEH) competing with piezoelectric generators (PGs) are listed in Table 1, with the benefits and drawbacks of each material or technology.

Please use the scroll bar at the bottom of the table or click on the table itself to scroll horizontally.

Table 1: comparative performance of mechanical energy harvesters [Gil08, Har11, Mat08, Fan12]
Electromechanical Transducer Benefits Drawbacks Power Density
  • Low frequency
  • Low voltage
  • Low efficiency at low power level
  • Complex manufacturing at micrometer scale
10 mW/(m/s2)2

1 mW/(cm3(m/s2)2)

  • Wide-frequency band
  • Low frequency
  • Easy manufacturing at micrometer scale
  • Needs electrical energy at starting point
10 µW/(m/s2)2

10 µW/(cm3(m/s2)2)

  • Wide-frequency band
  • Low frequency
  • High voltage
  • High efficiency
  • Easy manufacturing at micrometer scale
  • Low current
  • Ageing (not studied yet)
10 mW/cm3

(low frequency bending)

  • Wide-frequency band
  • High voltage
  • High efficiency
  • Easy manufacturing at micrometer scale
  • Low current
  • Limited temperature range
  • Ageing (not studied yet)
100 µW/(m/s2)2

200 µW/(cm3(m/s2)2)

Ambient mechanical sources are encountered on vehicles, industrial machines, traffic ways (roads, bridges, tunnels), passage ways, wind turbines, pipelines… or on human beings (during walk or hand movements). In order to estimate the interest to harvest such “free” energy, some important questions must be asked, in particular on the frequency range and the amplitude of the mechanical excitation, its truly free nature, its continuous or intermittent availability, the proximity of the electrical network, the interest of wireless electronic devices, etc.

Table 2 gives some examples of potential mechanical sources.

Please use the scroll bar at the bottom of the table or click on the table itself to scroll horizontally.

Table 2: examples of targeted mechanical energy sources
 Mechanical energy source  Frequency (Hz) Acceleration or velocity  Force (N)  Deformation  Power level
Hand (squeezing)  1  10N  1cm  100 mW
Rotating tyre of a car  60 to 80  up to 400 m/s2  300 W @ 105 km/h [Sin12]
 Rolling bicycle  10 to 30  10 m/s2  3.4 mW @ 21 km/h [Min08]
 Blood flow  1  0.93 W [Wan11]
 Exhalation  0.2  from 0 to 2 m/s  1W [Wan11, Lin13]

Various technologies of piezoelectric harvesters have been tested for energy harvesting since the early 2000s [Pri08, Def12]: bulk PZT (Lead Zirconate Titanate), PZT fiber composites, PZT thick or thin films, PVDF films. The densities of harvested power are comprised between 1 and 10 mW/cm3. Nevertheless, most of the prototypes are not integrated devices – including PG, power management block and storage element – and/or take up a volume that is too large to be compatible with embedded electronic devices.

In the meantime, thanks to bottom-up approaches, manufacturing process allow a better control of chemical composition and geometrical characteristics at nanometer scale. Today, literature presents an increasing number of prototype MEHs based on piezoelectric nanowires. The low maturity of the field results in numerous technological choices being investigated, with no clear material candidate being identified yet as the optimum one. The performance of each of these approaches depends on the material chosen (ZnO, CdS, GaN, or PZT), on the nanowires form (cylindrical, hexagonal, conical) as well as on the dimensions of the nanowires (length between 1μm and a few tens of microns, diameters between a few tens of nm and a few microns) [Cha12]. Typical values of power densities are a few µW/cm2 or mW/cm3, but it should be normalized by the applied mechanical excitation characteristics. There is no standard for the moment, for the mechanical operating conditions as well as for the electrical measurement protocol.

ZnO, with its direct semiconductor wide band gap of 3.37 eV at room temperature and its very large exciton binding energy of 60 meV, makes traditionally this material the candidate of choice for applications in ultra violet optoelectronics (LED, lasers or photodetectors). For us, the interest of the ZnO is elsewhere: in its exceptionally large piezoelectric constants. It means that a small deformation in the crystal can generate a significant applied voltage. Among the different ZnO nanostructures that can be obtained, nanowires and nanosheets are of particular interest.

2. Project Objectives

The heart of the project is to develop a prototype that integrates, on the same flexible chip, a microgenerator that converts this ambient mechanical energy into electrical energy that can recharge a storage element, for example a lithium battery, through a specific electrical converter.
In operation, application of a force induces in an apparent potential gradient across the microgenerator thickness, as a result of the piezoelectric effect. In turn, this potential can be used to drive charges in an external circuit.
The geometry of the final prototype is already fixed: 5 x 2.5 cm2 on flexible substrate (ISO 7816 smart card international standard), compatible with the dimensions of a solid state lithium thin film battery (LiB) [ST14]. The objective is to reach 150 µW supplied to the LiB, as it is the minimum power necessary to charge the LiB.
Three main structures of NW based piezoelectric generators are encountered: Vertically Oriented nanostructures of PG (VOPG), Laterally Oriented nanostructures of PG (LOPG) or Cylindrically Oriented nanostructures of PG around fibers (COPG) (cf. Table 3). As the natural resonance frequency of NWs is in the hundred of MHz range [Hin14], where no ambient source is available, the working mode of NW based piezoelectric generators is non resonant.

Please use the scroll bar at the bottom of the table or click on the table itself to scroll horizontally.

Table 3: principal structures of piezoelectric NW based generators
Type Vertically Oriented nanostructures of PG (VOPG) Laterally Oriented nanostructures of PG (LOPG) Cylindrically Oriented nanostructures of PG around fibers (COPG)
Schematics   Table3-Vertically-Oriented-Nanostructures-of-PG-VOPG Table3-Laterally-Oriented-Nanostructures-of-PG-LOPG Table3-Cylindrically-Oriented-Nanostructures-of-PG-Around-Fibers-COPG
Geometrical direction of NWs Orthogonal to the substrate Parallel to the substrate Orthogonal to the central fiber
Polymer matrix Possible Always At the bottom of the NWs only
Electrodes Plain or with particular shape (zigzag, with cones…) Plain or interdigitated Plain
Necessity to transfer the NWs No Depends on the growing process No
Operating principle Vertical compression or bending deformation Bending deformation Bending deformation

3. Manufacturing Process

Arrow Simplified Hydrothermal Synthesis

A possible route to reduce the price of these devices is using low cost manufacturing: hydrothermal synthesis appears as an interesting candidate, compared to MEMS technologies, that are often proposed in academic research. Energy harvesters can be practically assembled using low-cost solution assembly methods and over large-area substrates. Due to the relative ease of fabrication, the VOPG (Vertically Oriented nanostructures of PG) is by far and away the most developed (cf. Fig. 1a).

Fig. 1: Schematic diagrams of the fabrication process for the VOPG (Vertically Oriented Piezo Generator) on rigid or flexible substrate (a) [J6]; for the SNG (Stretchable NanoGenerator) on PDMS substrate (b) [J14]

At GREMAN, the prototypes are manufactured thanks to CERTeM technological platform, a fully equipped clean room and an electrical / physical / mechanical characterization laboratory.

The ZnO nanowires are synthesized by the hydrothermal reaction between 85°C and 100°C on a Au/Ti or ZnO coated substrate which may be rigid (silicon [J6], glass…) or flexible (Kapton®, Polyethylene naphthalate PEN [C9]…) or even stretchable [J14]. Figure 2 shows the top-view SEM image (a) and a cross section SEM image (b) of the NWs, and a XRD pattern (c), all showing that well-oriented and highly crystalline ZnO nanostructures are achieved.


Fig.2 : Top-view SEM image (a) and cross section SEM image (b) of the ZnO NWs; XRD pattern of the ZnO NWs (c), cross section SEM image (b) of the ZnO NWs covered with Parylene C and top electrode

High density piezoelectric ZnO nanowire growth was also demonstrated on metal coated PET substrates by another hydrothermal process: a galvanic cell reaction-assisted hydrothermal process with an aluminium sacrificing anode (cf. Fig. 3). This growth method offers several advantages with regards to ZnO nanowire material for industrial scale application:  the potential gradient induced by the cell allows for much higher growth rates (0.5 µm/hr at 100°C) and high density ZnO nanowires can be grown on specific substrate sites in the absence of seed layers and/or substrate with specific lattice parameters.

Process-flow-steps-for-PG-AssemblyFig.3 : Process flow steps for PG assembly. a-b) ZnO NWs are encapsulated by PDMS. c) Active area defined by evaporating Al/Ti (~400/100nm) on top of ZnO-polymer surface. d) Bonding of copper wires to the top and bottom surfaces. e) Cross-sectional SEM image of a fabricated PG [C9]

Arrow Full device manufacturing process (cf. Fig. 1)

Then the vertical ZnO NWs are completely encapsulated in an insulating polymer (cf. Fig. 2d), for various reasons. First, the insulating polymer layer provides an infinite potential barrier, preventing the induced electrons in the electrodes from “leaking” through the ZnO/metal interface. Second, this design replaces the Schottky contact in early designs which was difficult to realize on such highly doped ZnO NWs. Third, the polymer fills the gap between NWs by capillary force and forms a capping at the very top. This greatly enhances the VOPG’s efficiency as the applied force can now be transferred to all NWs even though they vary in length. Last, it serves as a buffer layer protecting NWs from intimate interaction with the electrode, improving the VOPG’s robustness.

The full manufacturing process is described in [J6]. Furthermore, some flexible and stretchable devices, called SNG stretchable nanogenerators (SNG), with increased performance, were manufactured following a similar but slightly different process (cf. Fig. 1b) [J14].

Arrow Density tunable growth of ZnO nanowires [J13]

In the literature, both by simulations and experimental results, it has been shown that the performances of electronic devices, such as piezoelectric nanogenerators, rely strongly on surface density and electrical properties of grown ZnO NWs. However, perfect control of the ZnO NW’s surface density, aspect ratio and alignment with controlled electrical properties has rarely been reported.

By careful addition of NH4OH, as an additive, in the growth solution, we obtained more than two orders of NW density variation (cf. Fig. 4). Although the exact chemical reactions during the HTG of ZnO NWs is still unclear, we have hypothesized, based on the experimental observations, the growth mechanism for the density controlled growth of ZnO NWs on seedless growth substrates. It is hypothesized that the amount of ammonium hydroxide has a direct effect over the concentration of Zn (II) complexes which will largely affect the Zn solubility in the solution. Consequently, the supersaturation of the growth solution can be controlled and so, the number of nuclei over the substrate.

To follow the effect of NH4OH on the electrical properties of ZnO NWs, we fabricated single ZnO nanowire field-effect transistors (NW-FETs) on Si/SiO2 substrates. It was observed that the free charge carrier density increases from 4.3 × 1016 to 2 × 1017 cm-3 while field-effect mobility decreases from 3.8 to 0.35 cm2/Vs, as the NH4OH concentration increases from 0 to 40 mM, hinting the creation of extra point defects with the addition of NH4OH in the growth solution [J13].


Fig. 4: SEM images of NWs grown for different concentrations of ammonia: (a) 0mM, (b) 10 mM, (c) 20 mM, (d) 30 mM, (e) 40 mM. The inset in each panel a-e shows the top view SEM image acquired from the same sample. The scale bar in the inset is 500nm. (f) Panel shows the variation of density of NWs with the change in NH4OH concentration [J13]

4. Characterization

Arrow Functional characterization

The fabricated PGs are characterized using a custom built test bench in order to assess their energy harvesting performance. The test bench setup is shown in Figure 5 [J14]. The test-bench includes a mechanical shaker (LDS V406) equipped with a ridged aluminium actuator arm, a power amplifier (LDS PA100E) used to drive the shaker, and a function generator (Agilent 33 250) used to control the magnitude and frequency of the shaker. The magnitude and frequency of the force applied to the device under test are in the [0; 13] N and [0;10] Hz range, respectively. The electrical characterization is performed using a differential amplifier, called double buffer circuit, which permits us to reduce the bias voltage effect during the measurement of the voltage generated by the PG (cf. Fig. 6) [J15].

custom-built-test-bench-for-the-PG-devices-functional-characterizationFig. 5: Custom-built test bench for the PG devices functional characterization.

Schematic-double-buffer-circuitFig. 6: Schematic of the differential amplifier, called double buffer circuit, which permits us to reduce the bias voltage effect during the measurement of the voltage generated by the PG

It is well known that for PG devices, the voltage generation is directly proportional to the applied stress. Therefore, it is expected that output voltage will be pressure sensitive. As seen from the Figure 7a, the open-circuit voltage (VOC) amplitude increases from 3 to 9.1 V with the increase of compressive pressure amplitude from 50 to 110 kPa, respectively, measured across a 128 MΩ resistive load. The peak output voltage values are plotted at corresponding applied pressure to calculate the sensitivity of the present device (Figure 7b). The sensitivity (S) is calculated by taking the slope of the linear fitting curve of the experimental data shown in Figure 7b. The obtained sensitivity value (S = 0.09 V kPa−1) is higher than previously reported triboelectric pressure sensors.


Fig. 7: a) Output open-circuit voltage w.r.t. applied compressive pressure; b) Summarized variation of peak voltage across 128 MΩ resistance at 5 Hz of frequency.

The output power and voltage are measured at different operating frequencies. Energy harvesting devices generally present internal impedance to the load circuits and in general, the maximum output power (Pmax) is obtained when the load impedance is matched to that of the harvester. Therefore, the optimal power transfer conditions for various operating frequencies are assessed by measuring the output power across wide range of resistive loads (RL), given by:

Pmax =(VRL × IRL )max

where VRL and IRL are the instantaneous peak voltage and current at a given resistive load, respectively. Figure 8a shows a plot of the peak output power against a wide range of resistive load for various operating frequencies. The low frequency range (3–7 Hz) is chosen for the present experiments as it is the typical vibrational frequency range present in our day-today life environment (walking, machine vibrations, etc.). The optimal load resistance for the maximum power transfer for various frequencies is found in between 10 and 20 MΩ. The maximum output power achieved from a SNG device is ≈580 nW at an operating frequency of 7 Hz and compressive pressure of 50 kPa. Further, Figure 8b shows the waveform of output voltages for different frequencies across 30 MΩ resistance under similar excitation conditions.

As can be seen from Figure 8, the magnitude of output power and voltage, respectively, increase with operational frequency. This is expected because the current generated from the piezoelectric material is proportional to the rate of strain.


Fig. 8: a) Peak output power as a function of load resistance at various frequencies. b) Output open-circuit voltage as a function of frequency at constant pressure of 50 kPa.

To avoid any discrepancy, we have also characterized reference devices, fabricated without incorporating ZnO NWs, under similar loading conditions. It can be seen from Figure 9, that the SNG devices gave 70 times higher peak-to-peak voltages compared with reference devices.


Fig. 9: The open circuit voltage versus time graph at constant force in forward and reverse connections: (a) the obtained signal from SNG ; (b) from reference device without ZnO NWs.

The proposed fabrication process for the SNG devices is substrate independent. The proof-of-device concept and its scalability are demonstrated by fabricating SNG on typical bank cards (5 × 3 cm). The integration of SNGs in the back plane of a flexible bank cards can be seen from the optical micrograph shown in Figure 10a: the three rectangular-shaped top electrodes are connected in parallel configuration to enhance the output current across the load resistance. The applied force on the SNG device is 13 N at 5 Hz, and the total compressive area is 8 cm2. The peak output power reaches a maximum value of 3 μW at an optimal load of 10 MΩ (cf. Fig 10b). Being pressed by a human palm, the peak value of open-circuit voltage and short-circuit current exceeded 27 V and 11 μA, respectively (cf. Fig 10c), with the peak maximum power of ≈35 μW at 100 kPa of compressive pressure.


Fig. 10: a) Image of the bank card and fabricated device at the back side of it. SEM image showing the as-grown ZnO NWs. Peak output power as a function of load resistance for an applied pressure of 16kPa (b) or 100 kPa (c) corresponding to hand tapping.

For the potential utilization of the electric output power generated from the SNG, we first demonstrated the charging of commercial capacitors and then to power consumer electronic devices such as LCDs [J14]. The SNG device is excited using mechanical shaker by applying 13 N of compressive forces. As shown in Figure 11a, many digits of a LCD screen can be driven without any external circuits using the SNG, implying the device significance in the field of consumer electronics.

Finally, the high sensitivity (0.09 V kPa−1) and flexibility of our SNG devices are sufficient to measure the slight body movements, proving the potential applications for wearable electronic systems such as human–machine interaction, fatigue detection, e-skin, etc. To obtain enhanced body motion sensor responses, we measured the voltage response from the SNG, by attaching it to the skin of index finger using a double-sided carbon scotch. The ability of the sensor device to detect the cyclic bending and releasing motion of index finger, is monitored and shown in Figure 11b. It is important to note that the metal electrodes used in the SNG devices do not possess stretchable features. However, the main idea is to show that the proposed fabrication process can be carried out on flexible and stretchable PDMS substrates with high performances.

Ability-of-the-sensor-device-to-detect-the-cyclic-bending-and-releasing-motion-of-index-fingerFig. 11: a) Schematic of circuit for steady and continuous display of LCDs driven by SNG device. b) The wearability test of the stretchable SNG devices being fixed on the middle of an index finger: The open-circuit voltage w.r.t. time showing peaks in positive side during bending stage and in negative during releasing stage.

Arrow Electrical characterization

The resarchers community working on ZnO NWs based generators knows that the high doping level of ZnO NWs synthesized by hydrothermal method is suspected to limit the performance of these types of devices, and their perspectives of industrialization. Hydrothermal synthesis is a cost-effective synthesis, fully compatible with industrial manufacturing processes. Therefore, controlling the natural doping of ZnO NWs is a challenge that may promote them for the next generation of piezo generators, based on a lead-free and biocompatible material.

For the extraction of electrical properties of NWs, we have fabricated bottom-gate single ZnO NW-FET on highly doped Si substrates with 170 nm thick thermally grown SiO2. Fabrication of ZnO NW-FET devices is done using standard electron-beam lithography (EBL) process (cf. Fig 12).

Using the IDS-VGS transfer characteristics, important electrical data of the semiconductor NWs, such as mobility and charge density can be extracted [J6,J7,J9].

In collaboration with University of Catania, optical properties of ZnO are also investigated by photoluminescence (PL) spectroscopy, which supplies detailed information about the optical electronic transitions between defect levels and valance/conduction band. The defect structure of ZnO NWs will also be investigated by the aid of electron paramagnetic resonance (EPR) spectroscopy. EPR spectroscopy is not only able to detect, distinguish and characterize stable or meta-stable band gap energy states but also gives detailed information about their microscopic origin.


Fig. 12: (a) Schematic illustration of the single ZnO NW-FET structure. (b) SEM image of a representative single ZnO NW device. (c) Transfer scan in both semi-log (left) and lin–lin scale (right). (d) Family of output scans at VG = -60 to 20V (5V steps) for the same device [J6]

5. Modeling

At GREMAN, since 2008 has been developed a strategy to overcome the different pitfalls in the modeling of the electromechanical conversion in ZnO NWs. Starting from governing equations that are constituted of three sets (mechanical, electrical and piezoelectric domains), different solving strategies can be set up. Non linearity induced by piezoelectric semiconducting coupling has already been studied thanks to finite element method for a single NW under static deflection [C5] but it seems to be hardly applicable at micro scale for thousands or millions of NWs. So a novel approach has been chosen based on a NW non linear equivalent circuit [J1]. This model can be easily implemented whether at nano or micro scale and can compute different physical quantities of interest for electromechanical devices such as voltages, currents, coupling coefficients or energetic efficiencies in dynamic regime. Simulations have been conducted with a resistive load connected at the output of the piezoelectric microgenerator, and should be enlarged to other types of loads like a Lithium microbattery.

6. List of Equipment Used in This Project

All the equipment is currently available through the collaborative technological platform CERTEM 2020.



  • Optical microscopy
  • Electro/Photo lithography
  • Raman spectroscopy
  • Metal deposition (PVD / e-beam evaporation)
  • Plasma-enhanced chemical vapor deposition (PECVD)
  • Metallic inkjet printer
  • Atomic Force Microscopy with SCM, SSRM, c-AFM modes
  • X-ray diffraction microscopy
  • SEM/STEM/TEM microscopy
  • Critical point dryer
  • Electrical probe system
  • kleindiek MM3A-EM micromanipulator
  • Tube furnace
  • Wet chemical bench

7. Bibliography

[Cha12] Chang et al., Nano Energy 1 (2012) 356-371
[Def12] M. Defosseux et al., Sensors and Actuators A 188 (2012) 489– 494
[Fan12] F.-R. Fan et al., Nano Energy 1 (2012) 328-334
[Gil08] J. M. Gilbert and F. Balouchi, Int J. Autom Comput, 05 (2008) 334
[Har11] Harb, Renewable Energy, 36/10 (2011) 2641
[Hin14] R. Hinchet, PhD thesis, Université de Grenoble, France, 4 April 2014.
[Lin13] H. I. Lin et al., ECS J. of Solid State Sci. and Technology 2(9) (2013) 400-404
[Mat08] Mathúna et al., Talanta 75 (2008) 613
[Min08] E. Minazara, D. Vasic, F. Costa, In proc. Int. Conf. on Renewable Energies and Power Quality (ICREPQ 2008), March 12-14, 2008, Santander, Spain.
[Pri08] S. Priya et al., in “Piezoelectric and Acoustic Materials for Transducer Applications”, eds. A. Safari, E. K. Akdogan, Springer (2008) 373-388
[Sin12] K. B. Singh, V. Bedekar, S. Taheri, S. Priya, Mechatronics 22 (2012) 970-988
[Wan11] Z. L. Wang, SMARTech digital repository, 2011, 139 pp.