A review of energy harvesting using piezoelectric materials: state-of-the-art a decade later (2008–2018)

Piezoelectric single crystals were developed to achieve superior coupling through uniform dipole alignment and outperform polycrystalline piezoceramics in many applications. The electromechanical coupling coefficient of single crystal piezoelectric materials can be significantly greater than monolithic materials, and in the highest performance materials can be several times greater than PZT [69–72]. The drawback to these materials is their higher cost, reduced toughness, and high damping [73]. However, despite these drawbacks, researchers have begun to incorporate piezoelectric single crystals in vibration-based energy harvesting systems to leverage their high electromechanical coupling.

Ren et al in 2010 [74] fabricated and tested a shear-based PMN-PT unimorph cantilever that was subjected to sinusoidal base excitation. The unimorph consisted of a 50.0 × 6.0 × 0.3 mm³ brass shim with a bonded 13.0 × 6.0 × 1.0 mm³ PMN-PT wafer and a tip mass of 0.5 g, and was able to generate 4.16 mW of power under a cyclic excitation force of 0.05 N at 60 Hz with a peak voltage output measured at 91.2 V. When compared to a similar PMN-PT cantilever operating in the d₃₁ mode, it was found that the shear mode device could generate considerably more power (approximately eight times more power). In 2009, Mathers et al [75] studied the application of interdigitated electrodes applied to a micro-scale PMN-PT cantilever-based energy harvester. The device had dimensions of 7.4 mm × 2.0 mm × 110 μm with a PMN-PT beam, a polydimethylsiloxane (PDMS) polymer coating, a PDMS tip mass, and interdigitated electrodes. Under excitation at its natural frequency of 1340 Hz and a displacement of 1 mm at the clamp, the device produced approximately 0.3 mW of power with a peak of 10 V.

In addition to single crystal PMN-PT, research has been performed on other material for energy harvesting, including lead magnesium niobate-lead zirconate titanate (PMN-PZT) single crystals. In 2008, Erturk et al [69] reported results using a PMN-PZT unimorph cantilever to harvest vibrational energy. This work used a small cantilever with a 20.0 × 5.0 × 0.5 mm³ piezoceramic applied to a 0.79 mm thick aluminum substrate that was excited at resonance (1744 Hz). The authors found that the device could generate a maximum power per base acceleration of 14.7 μW g⁻². In 2009, Moon et al [76] investigated a similar PMN-PZT cantilever and showed the device could generate 0.28 mW when excited under 1 g acceleration at resonance (630 Hz).

Lead zinc niobate-lead zirconate titanate PZN-PZT is another single crystal piezoelectric, first introduced in 2004 [77], that has been shown to be a high performance piezoelectric material for energy harvesting applications [78, 79]. In a work performed by Yue et al in 2017 [80], MnO₂ doped PZN-PZT nanopowders were synthesized and used to fabricate ceramics in a wide sintering temperature window. The material exhibits a d₃₃ coefficient of 314 pC N⁻¹, and a cantilever beam equipped with the piezoelectric material generated 98 μW of power under 10 m s⁻² and 90 Hz excitation when attached to a 1330 kΩ resistor. The authors report a high energy density of 29.2 μW mm^−3.

In a comparison study, Shahab et al in 2018 [81] investigated the performance of various soft and hard piezoelectric ceramics as well as single crystal piezoelectrics. The piezoelectric materials studied in this work include soft PZT-5H and PZT-5A ceramics, hard PZT-4 and PZT-8 ceramics, soft piezoelectric single crystals PMN-PT and PMN-PZT, and hard manganese doped PMN-PZT (PMN-PZT-Mn) materials. It was shown that in off-resonance harvesting, the soft materials outperform the hard ones; in contrary, in wideband random excitation including resonance harvesting, the hard materials outperform the soft materials. For off-resonance application, PMN-PT presented the highest electromechanical conversion efficiency. Yang et al in 2016 [82] reported similar conclusions on the energy harvesting performance of single crystal piezoelectric devices compared to PZT-based ceramics. The results showed that PZN-PT and PMN-PT single crystalline generators always outperformed PZT-based harvesters. Hwang et al in 2015 [83] also developed a flexible single crystalline PMN-PZT thin film energy harvester and installed it on the heel of a combat boot. The generated power from the piezoelectric was able to power 104 LEDs during normal walking.

While PZT offers superior piezoelectric properties to many alternatives, the toxicity of lead introduces inherent health risks in the use of PZT and other lead-based piezoelectrics. Ecological restrictions on the use of lead-based materials as well as the desire to use piezoelectric materials in medical devices has motivated the development of numerus lead-free piezoelectric ceramics [56, 84–86]. Lead-free piezoelectric materials consist of three main compositional families including titanate-based, alkaline niobate perovskite-based, and bismuth perovskite-based materials [87]. Performance evaluation of lead-free piezoelectric materials compared to lead-based materials has illustrated that some of these materials offer electromechanical, thermal, and mechanical properties superior to PZTs [88]. Much of the recent research on the topic of lead-free piezoelectric material is summarized in the review articles by Panda [55], Rodel [89], and Maurya [90], as well as a book by Priya and Nahm [56].

Recently, Wu et al in 2018 [91] demonstrated a lead-free flexible and high-performance piezoelectric material based on KNN-BNZ-AS-Fe (0.91K_0.48Na_0.52NbO₃-0.04Bi_0.5Na_0.5ZrO₃-0.05AgSbO₃-0.2%Fe₂O₃) with a coupling coefficient of d₃₃ = 500 pC N⁻¹. A 20 mm × 20 mm × 260 μm cantilever harvester made with this material generated 52 V and 4.8 μA under a compression force of 25 N at 2 Hz, which is sufficient to power 10 LEDs. Amongst the lead-free piezoelectric materials, BZT-BCT (Ba(Zr_0.2Ti_0.8)O₃-(Ba_0.7Ca_0.3)TiO₃) is one of the most widely studied materials due to its surprisingly high piezoelectric properties [92]. Yan et al in 2018 [93] proposed a high efficiency lead-free piezoelectric ceramic by adding Mn ions to a BZT-BCT ceramic with a d₃₁ coefficient of 130 pC N⁻¹. Experimental results on a 120 × 12 × 0.9 mm³ beam with a 10.5 × 10.5 × 0.5 mm³ ceramic patch showed that 1.198 mW of power was achieved under 5 g of acceleration at 64.5 Hz. In another work, this group developed a Mn-KNN (Mn-modified (K_0.5Na_0.5)NbO₃) lead-free piezoelectric material, and a 120 × 12 × 0.9 mm³ cantilever harvester with a 10 × 10 × 0.5 mm³ piezoelectric patch of this material generated 16 μW of power under 10 m s⁻² of acceleration at 90 Hz [94].

Another limitation in the application of PZTs is the limited working temperature of these materials due to phase instability and depolarization in high temperature applications, such as advanced energy generation systems and turbine engines [95]. Despite tremendous development in high temperature piezoelectric materials capable of working at temperatures up to 1000 °C in different bulk and thin film formations, the electromechanical coupling properties of the majority of these materials are relatively lower than conventional PZT ceramics [96, 97].

A 0–3-type composite was developed by Qaiser et al in 2018 [98] with embedded BiFeO₃ (BFO) grains in the Bi₃TaTiO₉ (BTTO) matrix to combine the acceptable piezoelectric coefficient of BFO and the high temperature resistance of BTTO. The composite shows a d₃₃ of 21 pC N⁻¹ and performs in temperatures as high as 500 °C. Li et al in 2017 [99] demonstrated an Mn-modified BiFeO₃-BiTiO₃ (BFO-BTO) lead-free ceramic with a high Curie temperature of 506 °C and d₃₃ of 169 pC N⁻¹. In a similar study, Tong et al in 2018 [100, 101] investigated the effect of Zn doping on the performance of BFO-BTO ceramics, which resulted in a piezoelectric with a d₃₃ of 192 pC N⁻¹ and depolarization temperature of 450 °C. Davis et al in 2018 [102] developed a novel non-ferroelectric piezoelectric material with a glass-ceramic composition for high temperature applications. The Sr-fresnoite with added SiO₂ material shows a d₃₃ of 10 pC N⁻¹, d₃₁ of 1.5 pC N⁻¹, d₁₅ of 34 pC N⁻¹, and relative permittivity as low as ε_r = 11.5 at temperatures higher than 300 °C, which makes this material a candidate for energy harvesting applications.

While monolithic piezoceramics offer high coupling coefficients, they cannot be conformed to curved surfaces, are generally brittle in nature making them vulnerable to breakage, and are typically dense due to the use of lead-based ceramics. To resolve the limitations of monolithic piezoceramic materials, researchers have devised composite piezoelectric devices consisting of an active piezoceramic phase embedded in a polymeric matrix phase. The resulting composites have increased strength and flexibility, as well as improved robustness due to the polymer matrix protecting the fragile ceramic. Research in this area has led to a broad range of active piezoelectric devices utilizing both fibers (Active Fiber Composites (AFC) [103] and Macro-Fiber Composites (MFC) [104]) and particles (0–3 composites). Early developments of 0–3 composites focused on the development of material performance and often termed the materials 'piezoelectric paints'. These materials offer an unprecedented ease of application to the desired surface and can be quickly coated over large surface areas through spray application or as discrete patches using a doctor blade or brush. Initial studies on PZT paint sensors based on 0–3 piezoelectric composites were performed in the mid 1980's by Klein et al [105] and Hanner et al [106] using a water-based suspension of piezoceramic and polymer.

More recently, a newer field of nanogenerators based on the use of vertically aligned piezoelectric nanowires (NWs) has emerged. One of the original works on piezoelectric nanogenerators was presented by Wang and Song in 2006 [107]. Their study fabricated a vertically aligned array of zinc oxide (ZnO) nanowires and experimentally tested the energy harvesting capacity using atomic force microscopy to deflect a single nanowire. Their study found the nanowires could generate significant energy when a Schottky barrier was formed, and created an entirely new field of energy harvesting. Through later developments of alternative architectures, the power density was increased to 2.7 mW cm⁻³ [108] which exceeds many MEMS-based devices.

The development of advanced materials for use in ZnO nanogenerators has led to several important design considerations due to the semiconductive nature of ZnO. Early reports realized that the electrical connection to the nanowire required the formation of a Schottky barrier to prevent screening of the piezoelectric potential formed in the material. It has been hypothesized that the rectifying contact was required to counteract the opposing polarizations in the symmetric rod under bending such that only one polarity could switch-on the diode at the metal-semiconductor junction and provide voltage to the external circuit [108, 109]. Liu et al in 2008 [109] performed a study using ultraviolet irradiation to tune the conductivity, and thus the carrier density, of the ZnO NW to evaluate the characteristics of the Schottky barrier at the interface between the metal electrode and the NW. The study demonstrated the critical role the semiconductive properties of the nanowires plays on the overall device performance. Briscoe et al [110] later demonstrated that it is possible to build a ZnO nanorod energy harvesting device using a semiconductor p-n junction, rather than a metal-semiconductor Schottky barrier.

The intrinsic properties of the n-type wurtzite-structured ZnO NWs influence the efficiency of the energy conversion in Schottky-type fabricated devices [110–112]. Therefore, in addition to the design of the Schottky barrier, intensive investigations have resulted in considerable improvement in the power generation of ZnO NWs through the reduction of the free-carrier concentration [112–117]. Passivation or doping with an acceptor are effective methods developed to engineer the carrier dynamics of ZnO NWs for energy harvesting [112, 114–119]. Lee et al in 2013 [115] utilized silver doping to increase the nanowire piezopotential through a reduction in free charge carriers that are typically formed in n-type ZnO and its corresponding hydrothermal synthesis methods. The Ag dopants act as shallow charge acceptors and can enable ZnO NWs to produce three times greater power in nanogenerators [115]. In addition to silver, Shih et al in 2014 [117] demonstrated the use of lithium in nanogenerators that produced open circuit voltages as high as 180 V. Lu et al in 2009 [120] demonstrated p-type ZnO nanowires that could switch the Schottky barrier to produce a positive rather than negative voltage.

While ZnO nanowires can be designed to achieve good energy harvesting performance, without doping or properly designed electrodes they exhibt high leakage current and limited overall performance. To overcome the limitations posed by the semiconductive propoerties of ZnO nanowires, other piezoelectric materials such as PZT and BaTiO₃ have been proposed for nanostructured harvesters. One of the initial works on piezoelectric nanocomposites using high aspect ratio fillers was presented by Feenstra and Sodano in 2008 [121] who developed a BaTiO₃ piezoelectric nanocomposite using electrospun fibers in an epoxy matrix. The BaTiO₃ nanowire paint developed in this work was compared experimentally to paint utilizing piezoelectric nanoparticles and, although it possessed a lower sensitivity than PVDF, it was found to provide as high as a three times increase in electromechanical coupling over the previous nanoparticle composite paint. This study was followed up with a theoretical analysis of the nanocomposites with high aspect ratio using micromechanics [122] and validation of the theory [123]. Zhou et al later in 2014 and 2016 demonstrated the process for PZT [124] and lead-free BZT-BCT (0.5Ba(Zr_0.2Ti_0.8)O₃–0.5(Ba_0.7Ca_0.3)TiO₃) [125] nanowires with the results showing more than nine times greater energy harvesting output when nanowires were used rather than equiaxial particles. Figure 1 shows an image of a nanocomposite cantilever with randomly oriented nanowires.

Zoom In Zoom Out Reset image size

Xu et al in 2010 [126] used the hydrothermal growth process developed by Lin et al [127] to grow vertically aligned PZT nanowires. The synthesis approach uses a seeded substrate to grow PZT nanowires via a competitive growth process that results in vertical alignment. Although fabrication of PZT thin films usually requires high temperatures (∼650 °C), the hydrothermal process enables the growth of vertically aligned single crystal PZT nanowire arrays at temperatures of only 230 °C, thus broadening their use by allowing integration with soft materials. Experimental testing of a nanowire array having an active area of 6 mm² showed the device to be able to generate 0.7 V peak output while having an average power density of 2.8 mW cm⁻³. Additionally, a seven-layer nanogenerator was found to be able to power a commercial laser diode when excited compressively by using a rectifying circuit with storage capacitors. A hydrothermal growth method was employed by Nafari et al in 2017 [128] to produce lead titanate (PbTiO₃) nanogenerators which could function in extreme environments. The study demonstrated that the high Curie temperature of PbTiO₃ allowed the energy harvester to function without loss of performance at temperatures as high as 375 °C.

In an effort to produce more environmentally friendly lead-free nanogenerators, Koka et al in 2013 [129, 130] developed a scalable hydrothermal growth process for vertically aligned arrays of BaTiO₃ and made comparisons to ZnO nanowires. The BaTiO₃ nanogenerator was designed using a proof mass mounted to the nanowire array such that the nanowires were compressed when subjected to base excitation, allowing the resonant frequency to be tuned to lower levels (<200 Hz) typically encountered by energy harvesting systems. The nanowire structure and device design are shown in figure 2. The results of testing showed that the BaTiO₃ nanowires could produce 20 times more energy that the ZnO nanowires although the Schottky barrier was not tuned and no doping was used. Koka and Sodano later in 2014 [131] demonstrated the growth of ultralong BaTiO₃ nanowires which reduced their stiffness such that a low resonant frequency of 155 Hz could be achieved.

Zoom In Zoom Out Reset image size

Many of the works performed on nanofiber piezoelectric energy harvesters are summarized in review articles by Chang et al in 2012 [132], Espinosa et al in 2012 [133], and Brisco and Dunn in 2015 [134]. Chang et al [132] reviews several types of piezoelectric nanofiber materials, but focuses on PVDF and PZT nanofiber generators. Espinosa et al [133] focuses on the characterization of nanomaterial properties and the performance of various nanomaterials, while Brisco and Dunn [134] focus their review on nanogenerator devices and material architecture.

In recent years, researchers have begun investigating the piezoelectric-like response of cellular polymer foam material for use in harvesting vibration energy [135]. The development of piezoelectret foam, also called ferroelectret foam, began in Finland in the 1980s [136]. This class of material is known as an electret; a dielectric material containing permanent electric charge (much like permanent magnets which contain permanent magnetic fields). While these materials are ferroelectret, as opposed to conventional piezoelectric materials which are ferroelectric, they exhibit piezoelectric-like behavior, therefore, are considered appropriate for inclusion in this review. Ferroelectret foam exhibits piezoelectric-like behavior thanks to the permanently charged internal voids of the structure. During fabrication of this material, a polarization process deposits the charge, which then becomes trapped in the voids. The application of mechanical or electrical stimuli causes the charged voids to act as macroscopic diploes, thus yielding piezoelectric-like properties. When compared to conventional piezoelectric polymer materials, piezoelectret foam has the advantage of a large piezoelectric coupling coefficient, up to d₃₃ = 250 pC N⁻¹ compared to d₃₃ = −33 pC N⁻¹ for PVDF (around seven times greater). Additionally, the cellular structure of ferroelectret foams provides a high level of compliance and low weight.

In 2014, Anton et al [137] investigated vibration energy harvesting using ferroelectret foam material. Their work utilized commercially available foams from Emfit, Corp. and created a pre-tensioned energy harvester with dimensions of 15.24 cm × 15.24 cm × 85 μm. When excited longitudinally (i.e. utilizing the '31' mode) with a peak-to-peak displacement of ±73 μm and at 60 Hz excitation (yielding an acceleration of ±10.38 g), the system generated 8 V peak. When configured to charge a 1 mF capacitor, experimental testing showed that the harvester could charge the capacitor to 4.67 V in 30 min while delivering an average power of 6.0 μW, an output power comparable to conventional piezoceramic and piezoelectric polymer materials.

Recent studies have also been performed to investigate the use of stacked piezoelectret foam harvesters that employ the '33' mode for improved energy harvesting performance. Pondrom et al in 2014 [138] formed 9- and 10-layer piezoelectret stacks. When harmonically excited in compression, the harvesters generated around 1.3 μW g⁻² of power for a load resistance of 100 MΩ. Later in 2016 [139], this group performed a systematic study to investigate the effect of seismic mass and number of layers on the energy harvesting performance of stacked and folded piezoelectret foams. An eight-layer irradiation cross-linked polypropylene (IXPP) stacked foam exhibited a power of 80 μW under an acceleration of 1 g, mass of 20 g, and resistive load of 93 MΩ. Ray and Anton in 2015 [140] extended the work of Pondrom et al [138] by increasing the stack layer count to 20 for improved performance. The 20-layer stack was excited harmonically in compression and configured to charge a capacitor. Experimental results showed an output of around 3.8 mW g⁻² for an optimal load resistance of 650 kΩ. Additionally, the stack was shown to charge a 1 mF capacitor to 1.2 V in 45 min when excited harmonically at resonance (124.4 Hz) with 0.5 g of acceleration. The work performed by Tefft in 2018 [141] expanded upon the study presented by Ray and Anton [140] by developing a more flexible 20-layer foam electret stack with composite graphene electrodes and without any adhesive. The results showed that a 1 mF capacitor could be charged to 1.025 V in 60 min under an acceleration of 0.5 g at resonance (93 Hz).

Another ferroelectret foam material developed by Mohebbi et al in 2017 [142] using nitrogen as the ionizing gas showed a d₃₃ coefficient as high as 550 pC N⁻¹, which is twice as high as the coefficients reported in the literature for previously fabricated foams using air as the ionizing gas. More recently, Zhang et al [143] in 2018 developed a ferroelectric nanogenerator using parallel tunnel films (laminated structure of fluorinated ethylene propylene (FEP) with large voids in between) with high transverse piezoelectric voltage coefficient of g₃₁ = 3 Vm N⁻¹ (corresponds to d₃₁ = 32 pC N⁻¹). The relatively high voltage coefficient (due to the small permittivity) of this material compared to PVDF with g₃₁ ≈ 0.2 makes it an ideal candidate for energy harvesting applications. A fabricated lightweight generator with dimensions of 3 × 5 × 8 mm³ showed an output power of 50 μW under an acceleration of 1 g and a seismic mass of 0.09 g. A thorough review of recent advances in ferroelectret foams and their piezoelectric properties was performed by Mohebbi et al in 2018 [144].

One issue correlated with typical piezoelectric energy harvesting systems which have linear resonance mechanisms is that these devices tend to exhibit weak power output when excited at frequencies away from the resonant frequency. This can happen as a result of fabrication imperfections and/or stochastic vibration [22]. Broadband and nonlinear energy harvesting techniques have been introduced and widely investigated during the past decade to improve the energy harvesting performance of linear systems by increasing the response bandwidth of the harvester. The frequency bandwidth of energy harvesting devices can be broadened using passive or active methods. Passive approaches manipulate the dynamics of the system to achieve reasonable response over a range of frequencies. Active approaches utilize an actuation mechanism to change the resonant frequency of the system to match the excitation frequency.

Various mechanical configurations have been utilized in passive broadband systems. Originally, the application of prestressed beams [192, 193] and attractive/repulsive magnetic forces [194] were suggested to adjust the resonant frequency of the harvester. While these passive systems were able to tune the resonant frequency, the frequency bandwidth of the system was not increased. In an attempt to broaden the harvester response bandwidth, Li et al in 2016 [195] presented a bi-resonant structure consisting of two PVDF bimorphs with proof masses and with different resonant frequencies. The cantilevers are placed carefully to ensure impact between the two beams as a result of base excitation at the resonant frequency of one of the beams (figure 5(a)). It was shown that the frequency bandwidth of the system was widened to a range of 14 Hz, which covers the resonant frequency of both of the beams, and the output power was improved by 80% compared to the power output of the two distinct beams.

Zoom In Zoom Out Reset image size

In active broadband systems, the resonant frequency of the harvester is altered by actuating the structure to match the excitation frequency. One of the original works in this area was presented by Roundy et al in 2005 [150], where a portion of the piezoelectric layers on a cantilever beam was utilized for actuating the structure in order to tune the resonant frequency to match the excitation frequency. However, it was found that the energy consumption of the actuating system was too large to be beneficial. In another work by Lallart et al in 2010 [204], an effective method for increasing the bandwidth of a beam harvester through actively tuning the resonant frequency of the system was suggested. A low-power control circuit was utilized to actuate the structure according to the displacement and acceleration feedback which resulted in insignificant power draw. More recently, a novel nonlinear approach to enhance the piezoelectric energy harvesting performance of beams and plates was proposed by Zhao et al in 2015 [196, 205] by utilizing acoustic black holes (ABHs) to design a dynamically tailored structure. Attaching piezoelectric elements on the ABHs, high energy density as well as broadband energy harvesting was presented due to a wavenumber sweep mechanism present in high frequency vibration (figure 5(b)).

More effective than passive and active broadband energy harvesting systems, nonlinear energy harvesting has been introduced and widely investigated in recent years. Introducing mechanical nonlinearities to an energy harvester, the frequency bandwidth of the system can be broadened while increasing the response amplitude and power output [206]. The majority of works performed on piezoelectric nonlinear energy harvesting have been based on piezoelectric-magnetic structures. Using multiple pairs of magnets, the resonator can be forced to oscillate around the dynamically stable positions in the system with one potential well (monostable [207–210]) or between multiple stable positions in the system with two, three, or four potential wells (bistable [211–213], tristable [214, 215], and quadstable [216]).

Initial works in this area were performed by Stanton et al [197], Cottone et al [217], and Erturk et al [218] in 2009. In the work of Stanton et al [197], a piezoelectric cantilever beam energy harvester with a tip magnet and two fixed magnets was proposed, which provides one stable position (monostable harvester) (figure 5(c)). Nonlinear oscillation of the cantilever beam around the stable position results in a wider frequency bandwidth and higher power output compared to a similar linear system. Changing the location of the fixed magnets in front of and behind the tip magnet, hardening and softening mechanisms in the beam dynamics are observed, respectively. Another nonlinear energy harvester design was presented by Cottone et al [217] using a bistable piezoelectric inverted pendulum device. The harvester consisted of a vertical piezoelectric bimorph equipped with a tip magnet and a fixed magnet with opposite polarity. Various magnet separation distances were investigated in order to achieve nonlinear oscillation between the two potential wells of the system. The nonlinear design was able to improve the output power of the device up to 600% compared to a similar linear system, while also achieving a wider range of operation frequencies.

A bi-stable Duffing oscillator with two stability positions was presented by Erturk et al [218] in 2009. The device consisted of a piezoelectric harvester beam with a tip magnet and two fixed magnets near the tip of the cantilever. Frequency response results showed that the effective bandwidth of the system was increased compared to a similar linear harvester with significantly improved power output. Another bi-stable system with similar configuration was suggested by Stanton et al in 2010 [198], and different separation distances between the magnets (as a bifurcation parameter) were analytically investigated (figure 5(d)). In 2013, Zhou et al [219] considered magnet rotation in addition to separation distance as another effective parameter on the performance of nonlinear multistable piezoelectric energy harvesters. The system was modeled analytically with the help of a discretization method (Rayleigh-Ritz) and Euler-Bernoulli beam theory. A vertical piezoelectric bimorph cantilever with a tip magnet and two rotatable external magnets was able to provide similar performance as the device reported by Stanton et al [197] in a smaller space. It was shown that the upsweep and downsweep test results are different, and for special magnet angles of 30 to 90 degrees, monostable duffing oscillation occurs within an 18 Hz bandwidth with noticeable superharmonic resonance. Later, in 2015, Jung et al [220] improved the analytical model of the system proposed by Zhou et al [219] by considering the linear terms of the magnetic force. The nonlinear dynamic and magnetic characteristics of the mono-, bi-, and tristable energy harvesters were modeled and discussed in detail. Huguet et al in 2018 [221] showed that by exploiting subharmonics found in bistable systems, the frequency bandwidth of a bistable piezoelectric harvester can be increased by 180%.

In 2015, Cao et al [222] presented a nonlinear piezoelectric device for energy harvesting from human gait. An analytical model was suggested considering nonlinear time-varying potential functions, as opposed to the constant potential functions used in previous works. The harvester consisted of a piezoelectric bimorph cantilever beam with a tip magnet and two rotatable magnets placed on an external frame. Experiments were performed on the human leg using different harvesters and it was shown that 3.21 μW, 18.73 μW, and 23.2 μW of maximum average output power could be generated using linear, monostable, and bistable harvesters, respectively. More recently, Harris et al in 2017 [223] demonstrated that, with the help of continuous wavelet transformation, phase portraits, and multiscale entropy analysis, a comprehensive platform to characterize the dynamic and electromechanical response of bistable harvesters could be created. The necessity of using a combination of solutions for characterization and optimization of nonlinear multistable harvesters with complex dynamics (such as random and chaotic vibration) was also shown by Wang et al in 2017 [224].

One of the main issues associated with bistable piezomagnetic systems is the relatively high potential well which limits the application of such harvesters to high oscillation amplitude applications to provide sufficient energy to cross the potential well [225]. In 2014, Zhou et al [199] presented a triple-well nonlinear piezoelectric energy harvester consisting of a piezoelectric beam with a tip magnet and two external magnets with a unique arrangement compared to traditional bistable systems (figure 5(e)). Theoretical model predictions and experimental test results over the frequency range between 10 to 35 Hz showed that a significant amount of energy can be harvested over the frequency range of 15.1 to 32.5 Hz under smaller vibration amplitudes compared to traditional bistable systems.

In addition to multistable systems based on cantilever beams and magnets, several researchers have proposed alternative designs of nonlinear energy harvesters. Tehrani and Elliott in 2014 [226] showed that using nonlinear Coulomb damping, a higher level of power can be harvested when the system is excited below the resonant frequency. Arrieta et al in 2010 [200] suggested the application of a bistable piezoelectric composite plate for nonlinear energy harvesting. Experiments under intermittent, limit cycle, and chaotic vibration showed that large power quantities in two distinct broadband frequency ranges can be obtained when the plate oscillates between two stable states (figure 5(f)). The bistable plate has several advantages over bistable cantilever beams, such as a non-magnetic structure which can mitigate the negative effects of magnet use on electronics, a more compact structure due to removing the magnets, and more adjustability because of the two lateral dimensions of the plate. In another study performed by Zhou et al in 2015 [201], a double magnet system consisting of two piezoelectric beam energy harvesters with interacting tip magnets was suggested (figure 5(g)). The system exhibits a multistable nonlinear behavior with a tunable frequency bandwidth through adjustment of the horizontal distance between the endmost magnets. Using a verified analytical model, it was shown that the system exhibits a more complex pattern of magnetic force due to the motion of the two oscillators. The frequency bandwidth of the device was also shown to be dependent on the linear parameters of each harvester without magnetic coupling, such as the horizontal distance between the harvesters and the size and properties of the tip magnets.

An architecture for a wideband piezoelectric energy harvester based on the bistable function of a simple buckled spring-mass system was proposed by Liu et al in 2013 [202]. The harvester consisted of two piezoelectric transducers with a displacement amplification mechanism and a central inertial mass all connected with flexible hinges (figure 5(h)). Analytical modeling and experimentation under chirp and band-limited white noise excitation showed that an output power density of 0.33 mW cm⁻³ could be harvested, which was significantly higher than the power density of the bistable systems introduced by Mann and Owens [227], Stanton et al [198], Erturk et al [228], Tang et al [229], and Arrieta et al [200].

Leadenham and Erturk in 2015 [230, 231] introduced a modified bistable beam harvester design with the potential of employing piezoelectric and electromagnetic generators. The beam was an M-shaped bent steel spring with 4 piezoelectrics on the roots of the beam and a proof mass in the middle of the beam. The nonconventional beam design was suggested to obviate the relatively high excitation level required to cross the potential wells of magnetic bistable beams and preloaded beams under compression. The system exhibited a nonlinear behavior at very low base excitation of 0.005 g, which was much lower than previous systems, and showed a 660% increase in the bandwidth of vibration at 0.04 g excitation compared to an equivalent linear system.

More recently, application of the internal resonance mechanism for increasing the frequency bandwidth of piezoelectric energy harvesters has been exploited by various researchers. Using an L-shaped beam-mass structure, Cao et al in 2015 [203] achieved a two-to-one internal resonance mechanism for nonlinear energy harvesting (figure 5(i)). It was shown that the frequency bandwidth of the proposed two degree of freedom system was significantly enhanced compared to two equivalent linear systems. An L-shaped beam-mass structure was also utilized by Chen et al in 2016 [232] to explore the feasibility of using the nonlinear modal interaction found in the internal resonance to improve the frequency bandwidth of energy harvesters as well as to study the effect of system parameters on the bandwidth. It was concluded that due to the presence of the peak associated with the two-mode component of the response, the bandwidth of the system increases compared to a linear system. Numerical analysis also showed that the frequency bandwidth is inversely proportional to the external load and piezoelectric electromechanical coupling factor. In a similar work, Liu et al in 2018 [233] also assessed the energy harvesting performance of an L-shaped structure. Experimental results showed that the frequency bandwidth and generated power is highly tunable by changing the length of the piezoelectric beams and the corner mass. A comparison between the investigated nonlinear harvester and a cantilever beam harvester showed that, by optimizing the device geometry, more power could be harvested from the L-shaped beam, while the power density of a cantilever beam is much higher than the L-shaped harvester. Another two-to-one internal resonance mechanism was introduced by Xiong et al in 2017 [234] using a tuned auxiliary resonator added to a primary oscillating structure equipped with a permanent magnet to provide nonlinear forces. Using a validated analytical model, the authors investigated the nonlinear dynamics and energy harvesting performance of the proposed system. Similar works on piezoelectric energy harvesting enhancement using internal resonance phenomenon have been presented by Harne et al in 2016 [235], Wu et al in 2018 [236], and Yang and Towfighian in 2019 [237].

More recently, Sun and Tse in 2018 [238] investigated the linear and nonlinear behavior of a horizontal asymmetric U-shaped piezoelectric energy harvester through finite element analysis, analytical modeling, and a series of experimental tests. The device consists of a primary and an auxiliary piezoelectric unimorph with two proof masses and two fixed permanent magnets. The magnets were deactivated for the linear analysis, while the nonlinear analyses take the magnetic interaction between the magnets and one of the tip masses into account. The results indicated that the nonlinear configuration outperformed the linear device with a maximum generated voltage of 14.18 V under a base excitation of 0.1 g at 15.41 Hz and a load resistor of 10 MΩ. The linear system exhibited 6.7 V of output under a base excitation of 0.1 g at 16.52 Hz. In 2018, Ahn et al [239] introduced a creative way to generate nonlinearity in the vibration response of a cantilever piezoelectric beam with the help of the nonlinear contact mechanism of an oscillating steel ball embedded inside the beam tip mass. Experimental testing was performed on the beam harvester by applying a base excitation with 3 m s⁻² amplitude. The proposed design showed an output power of 1.8 mW at 5 Hz and 13.5 mW at 15 Hz, while the conventional harvester presented an output power of 0.03 mW at 5 Hz and 14.8 mW at 15 Hz. At a fixed power level of 100 μW (required by a specific wireless sensor), the nonlinear harvester provided a 133% wider frequency bandwidth compared to the linear harvester.

Zhang et al in 2017 [240] investigated nonlinear energy harvesting from a primary structure based on the nonlinear energy sink (NES) principle. The developed device consisted of a base, a primary structure, and an energy harvesting NES. The primary structure is a platform supported by two steel plates and the NES structure is a fixed-fixed beam composed of two bimorphs, a steel beam, and two magnets attached to the beam. Later in 2018, Xiong et al [241] exploited the NES mechanism for simultaneous damping and energy harvesting. A theoretical model of a two degree of freedom system including a primary structure and an added NES structure was provided for analyzing the steady-state response, symmetry breaking and non-periodic responses, and frequency bandwidth with AC and DC electrical interfaces. Numerical simulations showed that 0.8 mW of power could be generated under 2 m s⁻² of base excitation at 12 Hz and with a 200 kΩ load resistor. A maximum frequency bandwidth of 9.44 Hz was reported for a 0.2 mW power threshold.

Lu et al in 2018 [242] introduced a nonlinear piezoelectric energy harvester consisting of three piezoelectric beams equipped with three permanent magnets, which forms an E-shaped energy harvester. Through an analytical model developed based on Hamilton's principle and a series of experiments, the effects of beam distancing on magnetic interaction, energy conversion efficiency, and jump phenomenon in frequency sweep tests were studied. The half-power bandwidth of the device is improved to 2.67 Hz compared to the 0.8 Hz of the corresponding linear system with a tip mass (i.e. 2.34 times improvement). Research on nonlinear piezoelectric energy harvesting systems has been very attractive and many other configurations of nonlinear harvesters have been investigated by researchers. The reader is referred to the reviews by Harne and Wang in 2013 [20] and Daqaq et al in 2014 [243] for more examples of research in nonlinear piezoelectric energy harvesting.

Low-power electronics based on very large scale integration (VLSI) circuitry have offered to significantly decrease the power consumption of many electronic components [244]. Progress has enabled chips to consume smaller levels of power, with many systems requiring only tens to hundreds of microwatts. Through decreased power consumption and the development of small-scale microelectronics, the use of microelectromechanical systems (MEMS) vibration-based energy harvesters has become feasible. Piezoelectric MEMS harvesters could enable fully self-powered microelectronics and sensors, and have use in many applications including biomedical devices where size and power usage must be optimized.

Jeon et al in 2005 [244] presented the first MEMS scale piezoelectric harvester to be fabricated and experimentally characterized. The author's device consisted of a micro-scale thin-film PZT cantilever harvester, as shown in figure 6(a), and utilized interdigitated electrodes to capitalize upon the higher d₃₃ mode of operation. The cantilever had dimensions of 100 × 60 × 0.48 μm³ and had a 13.9 kHz resonant frequency. Under resonant excitation with a tip displacement of 2.56 μm, the device produced 1.01 μW of power at 2.4 V. One of the primary drawbacks to micro harvesters is the high natural frequency which greatly exceeds the frequency of ambient vibration sources. Most mechanical systems exhibit operating frequencies in the range of tens or perhaps hundreds of Hz, therefore, devices with a fundamental resonant frequency in the kHz range are impractical.

Zoom In Zoom Out Reset image size

To address the high fundamental resonant frequency of MEMS energy harvesters, Fang et al in 2006 [245] developed a cantilever harvester with very high aspect ratio and a tip mass. The device, shown in figure 6(b), had dimensions of 2000 × 600 × 13.64 μm³ (the substrate thickness was 12 μm while the piezoceramic thickness was 1.64 μm) with a nickel tip mass applied, which resulted in a resonant frequency of 609 Hz. The authors found that a voltage output of 0.898 V and power of 2.16 μW were generated when excited with 1 g at resonance. The significantly reduced resonance frequency from Jeon et al's device [244] marked a drastic improvement to the design of MEMS vibration harvesters which operate under practical excitation frequencies. However, the operational frequency is still too high for the majority of mechanical systems. Liu et al in 2008 [246] continued the work of Fang et al with the development of an array of MEMS piezoelectric cantilevers, shown in figure 6(c), that allowed a larger operational bandwidth from 200–400 Hz.

In 2009, Shen et al [247] reduced the resonant frequency of a PZT-based silicon oxide wafer cantilever harvester to below 200 Hz. The cantilever is shown in figure 6(d) and had dimensions of 4800 × 400 × 22 μm³ with a 1 μm thick layer of PZT. Experimental testing demonstrated that the harvester had a fundamental frequency of 183.8 Hz and generated 0.32 μW under 0.75 g excitation. Lee et al in 2009 [248] compared MEMS energy harvesters functioning under the d₃₁ and d₃₃ piezoelectric modes. Through fabrication of both d₃₁ and d₃₃ PZT cantilevers using an aerosol deposition process, shown in figure 6(e) and (f), respectively, a direct comparison of the two harvester operational modes was performed. The d₃₁ harvester had dimensions of 3000 × 1500 × 11 μm³ and a proof mass, provided a resonant frequency of 255.9 Hz, and generated 2.765 μW at 2.675 V when excited at 2.5 g, whereas the d₃₃ harvester with interdigitated electrodes and the same dimensions and tip mass had a resonant frequency of 214 Hz, and generated 1.288 μW at 2.292 V under 2.0 g excitation. Through these results and further comparisons, the authors found that while the d₃₃ device produced a larger voltage, the d₃₁ device harvests more power.

Karami and Inman in 2011 [249] studied the use of novel geometries as a way to reduce the resonant frequency of MEMS piezoelectric energy harvesters. The authors proposed a zigzag structure, shown in figure 6(g), as a means of increasing the effective length of the beam without increasing the overall device dimension. While the geometry was proposed for MEMS devices, several macro-scale harvesters that contained varying number of elements in the zigzag (figure 6(g) shows an 8 member structure) were tested experimentally. The study's results demonstrated that 17 times reduction in resonant frequency could be achieved with an 11-member structure compared to a cantilever with the same beam length and thickness.

Berdy et al in 2012 [250] introduced a low frequency piezoelectric MEMS energy harvester with a fixed-fixed design (figure 6(h)) to reduce the torsion in the base of the beams when compared to the zigzag design presented by Karami and Inman [249]. The harvester was tested under an acceleration of 0.2 g at 49.7 Hz and a power of 118 μW (a power density of 5.02 μW mm⁻³ g⁻²) was generated. Another MEMS piezoelectric energy harvester was suggested by Yu et al in 2014 [256]. Initially, the device was designed and optimized using finite element simulation considering the resonant frequencies and output voltage of the harvester as the design parameters. A fabricated prototype of the harvester including a PZT cantilever array of five beams and an integrated large silicon proof mass was tested under a vibration excitation of 5 m s⁻² at 234.5 Hz. Using a power conditioning circuit consisting of impedance matching, AC-DC rectifying, instantaneous bleed-off, and voltage regulator circuits, a power output of 66.75 μW was achieved.

Integration of piezoelectric materials into MEMS devices has usually been conducted using complicated and expensive material processing techniques such as thin film deposition with laser deposition, sol-gel spin coating, screen printing, direct ink writing, and epitaxial growth [257]. Aktakka et al in 2010 [258] introduced a method to integrate a bulk piezoelectric ceramic, such as PZT and PMN-PT, on silicon substrates with a precise final film thickness of 5 to 100 μm. The method was to bond a bulk piezoelectric material to the silicon substrate and to thin the piezoelectric layer using an enhanced fixed-abrasive lapping/publishing process with a precise uniformity of ±0.5 μm. The advantages of the method include using commercially available piezoelectric materials, a wide range of achievable thicknesses, no polarization requirement after the process, and preserving the material properties of the original piezoelectric material. Energy harvesting test results on a 5 × 5 × 0.475 mm³ fabricated plate harvester showed that 10.2 μW of power can be generated under an acceleration of 2 g at 252 Hz, which was remarkably higher than previous works. Tang et al in 2014 [259] developed a d₃₃-mode MEMS energy harvester that consisted of a PMN-PT thick film and a silicon substrate. The d₃₃ coupling of the piezoelectric was exploited using interdigitated electrodes. The fabrication method previously introduced by Aktakka et al [258] was used in this study to thin a bulk PMN-PT. A fabricated prototype of the MEMS harvester with an active volume of 0.418 mm³ showed an output power of 7.182 μW under 1.5 g at 406 Hz. Later in 2017, this group applied a PZT thick film on a beryllium-bronzed substrate using the bonding and thinning technology to fabricate a piezoelectric MEMS harvester [260]. The fabricated device with an active volume of 30.6 mm³ exhibited a power output of 0.979 mW under an acceleration of 3.5 g at 77.2 Hz.

The integration of the harvester and power electronics on a single MEMS platform was studied by Marzencki et al in 2008 [251] as a method to further miniaturize the entire harvesting system. A micro-scale cantilever was combined with a miniature voltage multiplier circuit as a single System on a Package (SoP), as shown in figure 6(i). Although, the cantilever beam had a high resonant frequency (1511 Hz), resonant excitation at an amplitude of 0.4 g resulted in approximately 30 nW of regulated power at 3.0 V, and demonstrated an integrated system that could simultaneously harvest and condition energy. Rezaeisaray et al in 2015 [252] designed and fabricated a micro-energy harvester for frequency applications below 100 Hz. The compact design of the harvester includes a silicon substrate covered with aluminum nitride (AlN) piezoelectric material, and the electronic chip employed as a proof mass to reduce the natural frequency of the device (shown in figure 6(j)). The micro-harvester exhibited a power output of 136 nW, an open-circuit voltage of 1 V, and a frequency bandwidth of 10 Hz when subjected to base excitation of 0.2 g at 84.5 Hz and under an external load resistance of 2 MΩ. A validated finite element model of the system showed that by using PZT as the generator element and an SSHI power conditioning circuit, the power output can be increased to 3.1 μW.

Deposition of PZT layers on both sides of a stainless steel substrate using a customized aerosol deposition machine to fabricate an energy harvesting device was suggested by Kuo et al in 2016 [261]. A prototype of the cantilever harvester with dimensions of 9 mm × 6 mm × 90 μm was tested under a base excitation of 1.5 g at 140.8 Hz and a power output of 413 μW was measured. Jackson et al in 2017 [262] presented a CMOS compatible silicon-based MEMS cantilever harvester design using aluminum nitride as the piezoelectric layer to be embedded inside a pacemaker capsule with a diameter of 6 mm and length of 40 mm. The harvester showed a power output density of 97 and 454 μW cm⁻³ g⁻² under a heart rate of 60 and 240 bpm, respectively. A micro-scale self-powered viscosity and pressure sensing device was developed by Wang et al in 2017 [253] for an application in microfluidic systems. The device consisted of a layer of PVDF nanofibers deposited on a PDMS substrate to form a microchip (figure 6(k)). The device was tested under a droplet of water representing the droplets or bubbles found in microfluidic systems, and an open-circuit voltage of 1.8 V was collected. Simultaneously, the voltage signal generated by the harvester was analyzed to measure the pressure and viscosity of the microfluid as it passes the microchip.

Tian et al in 2018 [254] presented a low-frequency MEMS piezoelectric energy harvester consisting of a PZT thin film on a flexible phosphor bronze substrate with a proof mass. A rectangular hole was created on the beam harvester to reduce the resonance frequency of the system (shown in figure 6(l)). Exciting the device under a base acceleration of 1.5 g at 34.3 Hz, a power output of 216.66 μW (power density of 1713.58 μW cm⁻³) was collected for an optimal matched load resistance of 60 kΩ. In the work by Won et al in 2019 [255], a flexible piezoelectric MEMS beam harvester was developed using PZT thin film with a LaNiO₃ (LNO) buffer layer deposited on an ultra-thin NiCr-based austenitic steel metal foil substrate (figure 6(m)). The metal foil substrate provides larger compressive stress in the PZT thin film compared to a silicon substrate due to the high thermal expansion coefficient of the foil. Thermal expansion mismatch between the film and substrate is utilized to impose large biaxial stresses and to make a stress-tuned thin film during the fabrication process. A prototype of the harvester beam with the dimension of 2 mm × 4 mm × 25 μm was fabricated using a simple punching process. Experimental results showed that the device was able to generate a power output of 5.6 μW under a 11 kΩ load resistance when subjected to an acceleration of 0.5 g at 127 Hz.

While this section has reviewed several studies in the field of MEMS-based piezoelectric energy harvesters with the goal of reducing their resonant frequency, numerous other studies have been reported in the literature. The reader is referred to the comprehensive reviews of MEMS piezoelectric harvesting presented by Saadon and Sidek in 2011 [21], Priya et al in 2017 [263], and Tian et al in 2018 [264] as well as the book by Kottapalli et al in 2019 [265] for a more complete treatment of the field.

Poor energy conversion efficiency has been a consistent issue with piezoelectric energy harvesting devices due to weak transduction of vibration energy to generator components. Metamaterials and metastructures, with non-traditional physical behaviors, such as negative mass, stiffness, permittivity, permeability, and refraction, have attracted the attention of researchers in recent years to develop new forms of energy harvesting mechanisms with enhanced energy conversion efficiency. Phononic crystals and acoustic metamaterials are the most reported metamaterials for piezoelectric energy harvesting [266]. Phononic crystals are made of periodic distributions of inclusions embedded in a matrix exhibiting absolute acoustic band gaps and negative refraction, which can be used for mechanical filtering as well as to focus traveling elastic waves in specific directions [267]. However, the size of a phononic-based harvester working at low frequencies will be impractically large due to the long acoustic wavelength and sensitivity of performance to the lattice parameters and incident direction. Acoustic metamaterials, on the other hand, utilize independent unit cells with local resonators independent of lattice parameters and direction, which makes them a more practical candidate for energy harvesting [268]. In reviewing the literature, it can be noted that research on acoustic metamaterials for energy harvesting is still in its infancy, and most of the works carried out on developing metamaterials and metastructures are focused on the fundamentals.

One of the initial works on metamaterial energy harvesting was introduced by Gonella et al in 2009 [269], where a honeycomb lattice structure with a microstructure consisting of periodically distributed stiff piezoelectric cantilevers was proposed, and the interplay between phononic bandgaps and piezoelectric microstructure for energy harvesting was discussed (figure 7(a)). In another study by Wu et al in 2009 [270], a PVDF piezoelectric film was placed in the cavity of a sonic crystal (local defect created by removing a rod from a perfect sonic crystal) to convert acoustic energy to electric energy at the resonant frequency of the cavity (figure 7(b)). It was observed from experiments that a PVDF generator placed in the cavity generated 625 times higher voltage than that without the sonic crystal.

Zoom In Zoom Out Reset image size

In a work performed by Carrara et al in 2012 [271], an elliptical acoustic mirror was suggested in order to enhance structure-borne wave energy harvesting by focusing the propagating waves in a plate. The mirror consisted of several cylindrical stubs mounted on the surface of a plate, which focus the incoming wave energy at a specific point where a piezoelectric energy harvester is placed (figure 7(c)). Using stud spacing that is smaller or on the order of the wavelength of the propagated Lamb wave, the energy harvesting capacity in the frequency range between 25 and 150 kHz was investigated. A maximum power of 126 μW was generated with a 4.5 kΩ load resistance at 50 kHz. The generated power using the acoustic mirror was on average 3075% higher than the system without mirrors for the chosen resistance and frequency ranges. Later in 2013, this group introduced two other concepts of piezoelectric energy harvesters including localization using a 2D lattice structure with an imperfection, and guiding and channeling using an acoustic funnel (figures 7(d) and (e), respectively) [272]. The localization of the energy at the imperfection in the first design was exploited for tuned energy harvesting through matching the defect resonant frequency and excitation frequency. The second design was an acoustic funnel consisting of a periodic arrangement of several stubs on an aluminum plate featuring an open channel along which waves are guided. Using the acoustic funnel, the energy harvesting performance of a piezoelectric disc was increased by 84.5%, thus showing the effectiveness of the concept. The elliptical acoustic mirror introduced by Carrara et al in 2012 [271] utilized bulky cylindrical attachments, which drastically altered the thin host structure. In 2017, this research group investigated the feasibility of using structurally embedded mirrors using simulation and experimentation [273]. The structure used periodic metallic spheres of tungsten inserted into blind holes in a flat aluminum plate (figure 7(f)). Experimental results showed that the harvester was able to generate 11 times higher power using the embedded acoustic mirror structure as compared to the harvester without the mirror. Another concept studied by this group was omnidirectional elastic wave focusing and energy harvesting using a phononic crystal Luneburg lens [274]. A Luneburg lens was formed by hexagonal unit cells with blind holes of different diameters, which were determined according to the Luneberg lens refractive index distribution obtained by a finite element simulation (figure 7(g)). Two line acoustic wave sources were used to excite the plate and two piezoelectric harvesters were placed at the boundary of the lens. Experimental results demonstrated that the harvesters with the lens could generate 13 times higher power than the harvester without focusing the elastic waves. More recently, Sugino and Erturk in 2018 [277] introduced an analytical modeling framework for an energy harvesting metastructure based on local resonance. The structure was a periodic arrangement of piezoelectric cantilevers with tip masses attached to a primary beam structure. It was analytically shown that useful energy can be harvested from locally resonant metastructures without significantly diminishing their dramatic vibration attenuation in the locally resonant bandgap.

In order to enhance the stub-plate structure with defect proposed by Carrara et al [272], Qi et al in 2016 [275] proposed a concept to scavenge airborne acoustic waves in the direction perpendicular to the plate using a planar acoustic metamaterial structure. The system consisted of an array of silicon rubber stubs periodically deposited on a thin homogenous aluminum plate (figure 7(h)). A defect was created by removing four stubs, and a PZT-5H patch was attached to the defect. Simulation results demonstrated that the device was able to generate 8.8 μW of power when subjected to 2 Pa of acoustic incidence at a frequency of 2.25 kHz. A membrane-type acoustic metamaterial was designed and fabricated by Li et al in 2016 [276] including a pre-stretched thermoplastic circular membrane attached to a center mass and a PVDF patch (figure 7(i)). The membrane was originally designed for sound insulation, and the PVDF generator is placed at the point with maximum strain energy to harvest the sound energy. Experimental results showed that the proposed metamaterial device was not only able to block sound waves with over 20 dB sound transmission loss, but convert the absorbed energy to electric power with an efficiency of 15.3%. In another study by Hu et al in 2017 [278], the performance of a double-mass, acoustic metamaterial unit-cell with a piezoelectric transducer was investigated through analytical modeling. A mass-spring-damper model was initially used to simulate the performance of the unit-cell. The model was then extended to a multicell system, and the effect of various parameters including the piezoelectric coupling coefficient, mass ratio, stiffness ratio, and damping on the vibration suppression and energy harvesting performance was investigated.

Harvesting of wind energy is typically investigated using windmill-style harvesters and flutter-style harvesters. Energy harvesting from wind power has been far more attractive than many other sources because of wide availability and its perpetual nature, providing continuous mechanical energy [314]. One limitation in wind energy harvesting is the relatively low speed of wind near the ground due to boundary layer effects and physical obstructions such as trees and structures [315].

The original work on piezoelectric energy harvesting using a windmill-style device was presented by Priya et al in 2005 [316, 317] (shown in figure 8(a)). Following the work of Priya et al [316, 317], an optimized windmill harvester with a much simpler design was proposed by Myers et al in 2007 [318] (figure 8(b)). Since 2007, several researchers have tried to improve the design of windmill harvesters in order to reduce harvester dimensions and cut-in speed, and to increase power efficiency. A windmill harvester design was suggested by Tien and Goo in 2010 [319] consisting of a single PZT composite cantilever. A conventional fan with exciter teeth installed on the hub of the turbine was utilized to excite the harvester, as shown in figure 8(c). A prototype harvester equipped with a bimorph with dimensions of 72 × 12 × 0.5 mm³ was tested in a wind tunnel and a maximum power of 8.5 mW at 26 V was obtained, however, the corresponding wind speed was not described. A novel windmill harvester design using impact-induced resonance of piezoelectric beams was presented by Yang et al in 2014 [314]. The device, shown in figure 8(d), consists of a rotating fan in a polygon arrangement, three steel balls placed inside the polygon, and 12 piezoelectric bimorphs. The piezoelectric harvesters are struck with the steel balls when wind rotates the windmill and, consequently, the bimorphs oscillate at their natural frequency. A small-scale windmill with an overall diameter of 31 mm generated a peak power of 613 μW with an optimum load resistor of 20 kΩ at a rotational speed of 200 RPM (note, the wind speed was not reported).

Zoom In Zoom Out Reset image size

Considering the relatively low available energy level in small-scale windmill generators, several researchers have suggested different designs of contact-less windmill energy harvesters to minimize the effects of friction on power generation. In 2011, Bressers et al [320] introduced a contact-less wind turbine utilizing a nonlinear piezomagnetic configuration. The harvester consists of a vertical axis Savonius wind turbine rotor with permanent magnets and multiple piezoelectric bimorphs with tip magnets (figure 8(e)). Their optimization study resulted in a windmill with 2 blades and 4 magnets along with 6 piezoelectric beams (60 × 20 mm²) and with overall dimension of 16.51 × 16.51 × 22.86 cm³, which generated 1.2 mW of power at 9 mph (4 m s⁻¹) wind speed. A similar design of a windmill utilizing a contact-less mechanism was developed by Karami et al in 2013 [321]. Four vertical PZT bimorphs with dimensions of 58 × 12.7 ×0.38 mm³ were mounted on the base of a miniature Savonius vertical axis wind turbine. A permanent magnet was attached to the tip of each beam and five magnets were placed in the lower disc of the turbine blades such that when rotated, the magnetic interaction excites the cantilevers. Two versions of the piezoelectric windmill were tested in a wind tunnel and it was found that for the optimal design, as shown in figure 8(f), a power of 4 mW could be generated by a single bimorph at a wind speed of 10 mph (4.47 m s⁻¹). A startup speed as low as 2 m s⁻¹ was reported for the contact-less windmill. Another windmill-style piezoelectric energy harvester design was proposed by Zhang et al in 2017 [323] in which a rotating fan blade is attached to a turntable that strikes several piezoelectric beams. Wind tunnel tests showed that the device generated 2.57 mW of power at 14 m s⁻¹ wind speed and with a 10 MΩ load resistor.

A problem with the windmill designs proposed in this section is that they work at relatively high wind speeds (around 10 mph or 4.5 m s⁻¹), where even conventional electric generators operate efficiently. In order to reduce the cut-in speed, Kishore et al in 2014 [315] introduced a new wind turbine harvester design. The design is different from conventional wind turbines in many ways; the rotor has 8 fan-type blades with a flat-type profile, blade solidity of 50 %, and it employed a single piezoelectric bimorph generator (see figure 8(g)). Conventional turbines usually have 2–3 tapered blades with airfoil profiles, solidity of 5–7%, and employ electrical generators. The 72 mm diameter horizontal axis wind turbine rotor included 12 equally spaced permanent magnets attached around its periphery and a piezoelectric bimorph with a tip magnet clamped near the base of the turbine. In order to reduce the startup speed, the piezoelectric beam is used as an actuator for 2 s to start the energy harvesting process at 1.8 m s⁻¹ wind speed. Once started, the device switches to energy harvesting mode and generates 450 μW of power at the wind speed of 1.8 m s⁻¹. The initial actuation period consumes 7.2 s worth of harvested energy by the fan at this speed. The wind turbine was successfully used to charge a motion detector sensor. In another work, Rezaei-Hosseinabadi et al in 2015 [322] improved the power efficiency of contact-less wind turbines with a new arrangement of magnets and piezoelectric beams, as shown in figure 8(h). The 31 mm diameter turbine generated a maximum power of 363 μW (2 mW cm⁻³) and could operate at a cut-in speed as low as 0.9 m s⁻¹. In order to enhance the power harvesting efficiency of wind energy harvesters, Biccario et al in 2017 [324] suggested a new architecture of harvesting circuit that employs two storage capacitors: a small and fast-to-charge capacitor to power the active start-up circuit, and a large capacitor for storing the power. It was shown that voltage signals as low as 0.05 V can be harvested with only 200 ms start-up time with no amplitude limitation afterwards.

Table 1 summarizes the transducer type, generator material, dimensions, wind speed, cut-in speed, and output power of the different windmill-style piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.

Various flutter-style piezoelectric wind harvesting systems, as the second most reported configuration for energy harvesting from fluid flow, have been developed in the literature including fluttering beams with attached airfoils, wake galloping (a fluttering beam in the wake of a fixed bluff body), and fluttering beams with attached bluff bodies (galloping body). Using flutter-style harvesters, some issues associated with windmill-style harvesters, such as complexity, high fabrication and maintenance cost, and unfavorable scalability at small-scale due to high viscous drag and friction, can be alleviated [325].

In an initial investigation into flutter-style wind harvesting with an attached airfoil, Tan and Panda in 2007 [326] subjected a piezoelectric beam mounted at an angle to transverse airflow. In order to achieve improved coupling between airflow and the beam, a compliant plastic flapper was installed on the free end of the cantilever. A prototype of the harvester with overall dimensions of 76.7 × 12.7 × 2.2 mm³ showed a power output of around 155 μW when subjected to an optimal wind speed of 6.7 m s⁻¹. Similar designs of wind harvesters based on the stalk-leaf architecture were proposed by Li and Lipson in 2009 [327] and optimized in 2011 [328] (figure 9(a)). A plastic flapper was attached to the flexible PVDF stalk in parallel-flow stalk and cross-flow stalk configurations. Experimental test results showed that the cross-flow design could provide significantly more power than the parallel-flow design. A double layered PVDF cross-flow long stalk with dimensions of 72 × 16 × 0.41 mm³ harvested 615 μW power when subjected to 8 m s⁻¹ wind speed. The power density of this device was 17 times higher than the device presented by Tan and Panda [326]. Bryant and Garcia in 2011 [329, 330] expanded upon the work presented by Tan and Panda [326] by developing an aeroelastic flutter energy harvester containing a rigid flap connected to a piezoelectric cantilever harvester through a ball bearing revolute joint (figure 9(b)). When subject to airflow, oscillation of the flap causes flutter of the system. The beam had a length of 25.4 cm and width of 2.54 cm, while the flap had a width (span) of 13.6 cm and a length (semichord) of 2.97 cm. Two PZT harvesters with dimensions of 4.6 × 2.06 × 0.254 cm³ were placed on opposite sides at the beam's root. Experimental testing in a wind tunnel showed that the harvester could provide 2.2 mW of power at a wind speed of 7.9 m s⁻¹, with a cut-in speed of 2.6 m s⁻¹. Stamatellou and Kalfas in 2018 [331] also studied the energy harvesting performance of a PVDF cantilever with a plastic extension subjected to swirling airflow. The transducer generated 3 μW of power when placed in a swirling flow at 2 m s⁻¹ and with a load resistor of 150 Ω.

Zoom In Zoom Out Reset image size

In addition to flutter-style harvesters with attached airfoils, researchers have also investigated wake galloping-based flutter harvesters. Akaydin et al in 2010 [332] investigated the energy harvesting performance of a PVDF bimorph located in the wake of a cylinder at high Reynolds numbers subjected to turbulent flow (figure 9(c)). The three-way interactions of flow, electromechanical piezoelectric behavior, and electronics were numerically modeled using a multiphysics finite element model, and experimental validation was performed. The piezoelectric harvester had dimensions of 30 × 16 × 0.2 mm³ and was able to generate 4 μW of power when subjected to 7.23 m s⁻¹ wind speed. It is necessary to note that due to weak conversion of flow energy to mechanical vibration energy, the total efficiency of the device was very low. Experimental test results obtained from laminar airflow (provided by a fan) and turbulent airflow (provided by a wind tunnel) showed that 35 μW of power at 5 m s⁻¹ turbulent wind speed could be harvested, which is sufficient to power a commercial thermometer. Later in 2012 [336], this group presented a self-excited fluidic energy harvester consisting of a cylinder attached to the tip of a piezoelectric cantilever. Wind tunnel experiments showed that the device generated 0.1 mW of power at a wind speed of 1.192 m s⁻¹. In another study by Zhang and Wang in 2016 [337], a harvester consisting of a rigid cylinder attached to two piezoelectric bimorphs was placed against fluid flow to harvest energy from vortex induced vibrations and wake induced vibrations of an additional large cylinder. It was shown that the wake induced vibrations of the large cylinder led to more than 400 times higher power than a harvester with only vortex induced vibration (without additional cylinder). More recently, Ravi and Zilian in 2019 [338] developed a three-dimensional multiphysics finite element model for flow driven piezoelectric energy harvesting that involves the three-way interaction of fluid flow, piezoelectric material, and the controlling electrical circuit. A system of integral equations describing incompressible Newtonian flow, electromechanical behavior of a piezoelectric patch on an elastic substrate with equipotential electrodes, and the attached circuit was derived and solved using space-time finite element discretization with static condensation of the auxiliary fields and Galerkin's method. Usman et al in 2018 [339] presented a novel piezoelectric energy harvesting configuration based on wake galloping using a unimorph piezoelectric beam. Two cylinders with circular cross-sections of similar diameter are considered in the design while the upstream cylinder is fixed and the downstream cylinder is placed on top of a piezoelectric cantilever consisting of an MFC film. The lift component of force was considered the main cause of vibration. Test results showed that the system provides good performance for wind speeds higher than 4 m s⁻¹, and an optimum spacing distance equal to three times the diameter of the cylinders was achieved. Using an MFC beam with dimensions of 85 × 30 × 0.3 mm³, a peak voltage of 27 mV was achieved under a wind speed of 7 m s⁻¹.

A final method of flutter-based energy harvesting involves attaching a bluff body to a piezoelectric beam. Kwon in 2010 [333] developed a galloping body using a T-shaped cantilever beam that facilitates aeroelastic flutter under wind excitation (figure 9(d)). The occurrence of vortex shedding was observed around the T-shaped end of the device when placed in airflow causing flutter in the beam. A T-shaped harvester with overall dimensions of 100 × 60 × 30 mm³, and six MFC transducers (28 × 14 × 0.3 mm³ each) attached to the root of the beam, was placed inside a wind tunnel. Experimental results showed that the device could generate a maximum power output of 4.0 mw at a cut-in speed corresponding to flutter of 4 m s−1. Another flow energy harvester design consisting of a cross-flow PZT cantilever beam with a cylindrical extension was presented by Gao et al in 2013 [325] (figure 9(e)). The piezoelectric beam was equipped with a 31 × 10 × 0.127 mm³ PZT layer and the cylinder was a lightweight 36 mm long hollow cylinder. As a result of turbulence created around and in the wake of the cylinder, the piezoelectric beam could vibrate in the direction normal to the flow, and it was shown that 0.035 mW of average power could be generated at a wind speed of 5 m s⁻¹. Similar results were found by Amini et al in 2017 [340] for a comparable piezoelectric device.

The concept of piezoelectric energy harvesting under combined base excitation and vortex induced vibration (galloping in the beam with attached bluff body) was proposed by Dai et al in 2014 [341]. The harvester was a multilayer piezoelectric cantilever beam with a cylindrical tip mass similar to the design proposed by Akaydin et al [336]. A nonlinear distributed parameter model of the generator was developed and the results were compared to the experimental results of Akaydin et al [336] for the case of energy harvesting from beam galloping for the sake of verification. Considering the combined vibration sources, it was observed that the behavior of the system changed from periodic to period-n and quasi-periodic due to the co-presence of base excitation frequency and shedding frequency, and the output power of the system was significantly higher than the output power of the two separate vibration sources. A piezoelectric beam with dimensions of 267 × 32.5 × 0.635 mm³ generated a power output of 1.6 mW under a wind speed of 1.2 m s⁻¹ and base excitation of 0.05 g. In similar works, Yan et al in 2014 [342] and Bibo et al in 2015 [343] presented and verified analytical models for a piezoelectric bimorph cantilever beam with a tip mass (bluff body) under combined galloping and base excitation. For wind speeds below the galloping speed, a periodic response corresponding to the base excitation frequency with a corresponding high amplification in power generation was observed, while, for speeds higher than the galloping speed, two peak frequencies were observed in the response of the system.

Erturk et al in 2010 first introduced the concept of piezoaeroelastic systems (those that couple piezoelectricity and aeroelasticity) [344]. In this research, piezoelectric coupling was introduced to the plunge degree of freedom of a typical wing, and the developed linear lumped-parameter model was experimentally validated. Experiments showed that an electrical power of 10.7 mW could be generated at a linear flutter speed of 9.3 m s⁻¹ and a 100 kΩ load resistance. In addition, the effect of piezoelectric power generation on the linear flutter speed and generated nonlinearities in the system were discussed. Later in 2011 [345], this group enhanced the aeroelastic energy harvesting of a similar system by introducing combined piezoelectric-aeroelastic nonlinearities to the system. Piezoelectric devices were installed on the support beams of the airfoil in the plunge degree of freedom. Initially, the linear response of the device near the flutter speed was evaluated. Then, power output was improved by adding nonlinearities to the pitch degree of freedom leading to chaotic oscillations and twice as much harvested energy. Wind tunnel testing showed that the harvester could generate 27 mW of power at a wind speed of 10 m s⁻¹. Although the system tested was quite large and impractical, it demonstrated the benefits of nonlinearity in piezoaeroelastic energy harvesting. Later in 2013 [346], the same group proposed another airfoil-based harvester design that utilized piezoelectric transduction and electromagnetic induction for power generation. A 2D model with focus on linear system parameters was proposed to investigate various parameters including radius of gyration, chordwise offset of elastic axis from the centroid, frequency ratio, load resistance, and internal coil resistance. Adding a control surface to the airfoil and introducing a 3D model in 2014 [334], they showed that the power output of the system was improved. In addition, the 3D model offered broader design space and parameters to reduce the cut-in speed and to maximize the power output of the harvester (figure 9(f)).

The concept of piezoelectric grass, an array of vertical piezoelectric cantilever beams to harvest energy from turbulence induced vibration (figure 9(g)), was introduced by Hobeck and Inman in 2012 [335]. This concept involves arrays of robust piezoelectric beams designed for energy harvesting from low-velocity, turbulent flows. PZT cantilevers and PVDF cantilevers were employed to fabricate two prototypes of the presented concept. Wind tunnel testing demonstrated that a power of 1.0 mW could be generated by each beam (with substrate dimensions of 101.6 × 25.4 × 0.1016 mm³ and piezoelectric dimensions 45.97 × 20.57 × 0.1524 mm³) at a wind speed of 11.5 m s⁻¹.

More recently, in 2017, Silva and De Marqui [347] presented the concept of self-powered active control of base excitation and aeroeleastic oscillation of wings using a sensor-actuator piezoelectric system. A plate-like wing with two piezoelectric layers attached to the root of the wing on the top and bottom surfaces was analytically, numerically, and experimentally investigated. The top piezoelectric element was employed as an actuator and the bottom one was utilized as a sensor and energy harvester. Results showed that the device was able to fully damp the flutter vibration of the wing for a certain level of oscillation amplitudes. For small amplitude vibration, the system could not harvest enough energy to power the actuator, but the oscillation of the wing was damped to some extent due to the presence of the passive piezoelectric patches. In another study, the application of piezoelectric laminated plates in energy harvesting from yawed flow was suggested by Tang and Dowell in 2018 [348], and a computational model of the piezoelectric-aeroelastic coupled system was developed. Orrego et al in 2017 [349] proposed a flexible piezoelectric membrane placed in an inverted flag orientation where the trailing edge of the flag is fixed and the leading edge is free to move [350]. A flag with five 60 mm long piezoelectric beams generated a peak power of 1–5 mW cm⁻³ for wind speeds between 5 and 9 m s⁻¹.

Table 2 summarizes the transducer type, generator material, dimensions, wind speed, cut-in speed, and output power of the different flutter-style piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.

Research on energy harvesting from liquid flow has been more limited compared to energy harvesting from airflow due to the availability of the source. On the other hand, liquid flow sources, in particular water, can provide continuous energy accompanied with a higher energy density compared to airflow. Given the rich kinetic energy content of water flow along with a small dependence on environmental conditions, there is promising potential for energy harvesting from water sources.

Taylor et al in 2001 [351] introduced one of the earliest works on harvesting liquid flow using piezoelectric materials. They presented an energy harvesting eel consisting of a PVDF bimorph emerged in water flow (figure 10(a)). A prototype eel with dimensions of 24 cm × 7.6 cm × 150 μm was fabricated and tested in a flow tank. A peak voltage of around 3.0 V was measured at a water velocity of 0.5 m s⁻¹. In another work performed by Pobering and Schwesinger in 2004 [352], two types of hydropower piezoelectric harvesters including a PZT bimorph, and a PVDF fluttering flag, were proposed (figure 10(b)). Simulation results showed that the PVDF flag harvests up to 32 W m⁻² of power, while a PZT bimorph with dimensions of 5 × 3 × 0.060 mm³ could generate around 7 μW of power. In 2009, this group performed a comprehensive analytical, numerical, and experimental study on the power generation using a short piezoelectric cantilever beam subjected to liquid and gas flow [353]. Using a bluff body attached to the beam, a Von Karman vortex street (turbulence) was created in order to achieve a pressure differential on the beam surfaces and to apply upward and downward forces periodically. The system was analyzed analytically to investigate oscillation in the first resonant mode of the beam and to avoid neutralization of charge. The system was comparatively tested in wind and water channels, and a higher efficiency was observed when the system was subjected to water flow due to the higher energy density of water compared to air. For a geometrically non-optimized system, 0.8 V and 0.1 mW of power output was generated in the water channel for a flow speed of 45 m s⁻¹.

Zoom In Zoom Out Reset image size

In the study performed by Wang and Ko in 2010 [354], this group investigated the ability of a PVDF film to harvest energy from the fluctuations of pressure created by a pump in a fluid system. The harvester consists of an in-line flow channel, a flexible diaphragm, and a PVDF film with dimensions of 25 × 13 × 0.15 mm³ (PVDF thickness of 28 μm, the remainder is a polymer coating) attached to the diaphragm (figure 10(c)). The diaphragm transfers the pressure ripples of the liquid flow to the piezoelectric film. Placing the flow-based harvester in a pressure line with oscillations of 1.196 kPa at 26 Hz, 0.2 μW of power was experimentally obtained. Following this work, Wang and Liu in 2011 [356] proposed a similar design of the harvester with a PZT film as the generator component. Experimental results showed that using a PZT film with dimensions of 8 × 3 × 0.2 mm³, a power output of 0.45 nW could be generated for a fluctuation pressure of 20.8 kPa at 45 Hz.

In a more recent study conducted in 2015, Pineirua et al [355] investigated the effect of piezoelectric electrode arrangement along a fluttering flag through a series of analytical and experimental studies (figure 10(d)). The verified analytical model showed that the number and position of piezoelectric elements have a critical role in energy harvesting performance. It was shown that a larger number of piezoelectric segments can improve the efficiency of the fluttering harvester, in particular, for systems with higher fluid to solid inertial ratio (mass ratio) where the modal structure responds to shorter wavelengths and shorter electrodes can capture the flag deformation.

Table 3 summarizes the transducer type, generator material, dimensions, flow characteristic, and output power of the different fluid flow-style piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.

While the availability and relatively high kinetic energy content of air and water flow has attracted the attention of researchers to these sources, higher energy densities can be found in some other fluid sources including ocean wave and hydraulic systems. Ocean waves, with higher energy density of about 2–3 kW m⁻² compared to 0.1–0.5 kW m⁻² of wind near the surface (4–30 times higher energy density), have a promising potential for energy harvesting [357]. High pressure hydraulic lines with pressure ripples induced by pumps are another fluid source for energy harvesting reported in the literature.

One of the first works on energy harvesting from ocean waves was presented by Zurkinden et al in 2007 [358]. A PVDF cantilever beam with dimension of 30 × 3.75 ×1.25 mm³ was modeled on the sea bed and subjected to ocean waves through analytical and numerical analysis. A piezoelectric-buoy harvester design was introduced by Murray and Rastegar in 2009 [359] based on a two-stage electric generator concept (figure 11(a)). Three designs of the harvester including heaving-based, pendulum design pitching, and four-bar linkage pitching harvesters were presented. The output of a 76 mm diameter, and 914 mm long heaving-based harvester was 60–180 mW.

Zoom In Zoom Out Reset image size

In 2014, Xie et al [360, 361] proposed two designs of piezoelectric energy harvesters to generate power from transverse and longitudinal wave motion. The former harvester was a horizontal piezoelectric cantilever beam attached to a square column on a fixed support under a floating structure (figure 11(b)) and the later device was a piezoelectric cantilever column equipped with a tip proof mass (figure 11(c)). Analytical modeling results showed that using an optimized design of both devices, an RMS power of 30 W could be generated from the horizontal device with 2.4 × 1 m² cantilever dimension, and 55 W could be generated by the vertical harvester with 3 m cantilever length. Another piezoelectric-buoy energy harvester was proposed by the same research group in 2015 [362] for power harvesting from deep and intermediate ocean waves with relatively higher available power than sea bed and shallow waters. The buoy consisted of a slender floater attached to a larger cylindrical sinker to compensate the vertical oscillation of the buoy body (figure 11(d)). Several PZT beams were attached horizontally to the floater close to the ocean surface in order to generate electricity from the relative motion between the ocean wave and the buoy. Using a novel analytical model, the length and width of the floater, the diameter of the sinker, and the ratio of wave length to the length of the cantilever were optimized. The device with 1 m long and 0.2 m wide piezoelectric beams was able to generate 24 W of power from ocean waves. More recently, in 2017, the concept of energy harvesting from ocean waves using multistable mechanisms was introduced by Younesian and Alam [364], and the stability characteristics of a simple buoy-generator-restoring spring system were well discussed.

In addition to oceans and streams, the application of energy harvesting from rain drops has been suggested by several researchers [365–369]. In the work proposed by Ilyas and Swingler in 2015 [363], a PVDF-based energy harvester was presented. They utilized a commercial sensor as the active piezoelectric film under the impact of rain drops (figure 11(e)). Different stages of drop impact were discussed, modeled, and experimentally investigated, and results showed 2.5 nW of power generation from a single harvester with dimension of 25 × 13 × 3 mm³ and a load resistance of 2.2 MΩ for a single impact. Later, in 2017, they developed a multipiezoelectric generator to enhance the energy conversion efficiency of the harvester [370]. The device is a unimorph with 3 separate impact regions covered by PVDF patches and with overall dimension of 25 × 13 × 3 mm³ and connected in parallel to a 1 MΩ load resistor. Optimizing the surface angle, surface condition, and impact region, the efficiency of the device was improved to 0.671%. A review of energy harvesting from rain drops has been published by Wong et al in 2015 [371], and different harvester designs including piezoelectric bridge, solid film, and cantilevers were compared. It was found that the most efficient configuration for rain energy harvesting is a PVDF bridge. Outdoor environmental conditions (sunlight, wind, water, etc), application with large and typical raindrops, and non-continuous energy were listed as the main challenges of energy harvesting from raindrops.

In the work of Cunefare et al in 2013 [372], a hydraulic pressure energy harvester for electricity generation from the pressure ripple in closed hydraulic systems with a piezoelectric stack was presented. The high power intensity of hydraulic systems provides a suitable ground for piezoelectric energy harvesting. The energy harvester included a piezoelectric stack within a housing connected to the hydraulic line with an interface separating the fluid from the harvester. Two prototypes were fabricated for exploring the electromechanical performance and interface effects. The stack was a soft PZT with dimensions of 6.8 × 6.8 × 30 mm³. The interface was a 0.0762 mm thick aluminum diaphragm. The hydraulic system was a nine-piston pump operating at 1500 RPM with a fundamental ripple frequency of 225 Hz. The test results showed an output power of 1.2 mW from a dynamic pressure ripple of 400 kPa and with a 120 Ω resistor. The experiments also showed that the power efficiency was improved using area ratios more than unity at the interface. The output voltage was also calculated via an analytical model and compared to the test results successfully. The study showed promising potential of off-resonance energy harvesting from hydraulic systems. Zhou et al in 2018 [373] introduced a novel piezoelectric tubular energy harvester configuration in order to generate power from fluctuating fluid pressure inside tubes. The proposed system consists of a PZT tube with inner radius of 8 mm, outer radius of 10 mm, and length of 20 mm subjected to 0.2 MPa internal fluctuating pressure at 10 kHz attached to a resistive load of 2.62 kΩ. An analytical model of the device was developed and an exact solution was derived. Simulation results showed that 0.1 W of power could be generated using the proposed device.

Table 4 summarizes the energy source, transducer type, generator material, dimensions, input excitation, and output power of the different alternative fluid-based piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.

The ubiquitous presence of low power portable electronic devices and the lifespan limitations of batteries have encouraged the research community to develop wearable energy harvesting devices with the ability to generate micro- to milliwatts of energy. This amount of energy can be sufficient to power small electronics such as heart rate monitors and respiratory rate monitors, or, possibly, mobile phones.

One of the early works on wearable energy harvesting devices was presented by Paradiso's research group at MIT Media Laboratories in 1998 [377, 378], in which piezoelectrics were integrated into shoes. Similarly, in 2010, Rocha et al [379] presented a shoe harvester design consisting of PVDF polymers and an electrostatic harvester integrated into the sole of a shoe. Due to the relatively high available energy of human gait, other researchers have tried to improve the energy harvesting efficiency from shoes using different harvester designs. In the work performed by Xie and Cai in 2014 [380], an amplification mechanism including several piezoelectric bimorphs and sliders was proposed, and 0.41 mW cm⁻³ was experimentally obtained, which showed higher harvested power than previous works (figure 12(a)). In order to increase the power output of wearable harvesters in low frequencies, Jung et al in 2015 [381] suggested a flexible energy harvester with curved piezoelectrics located in the shoe insole. The curved generator was also utilized in a watch strap. Simulation results showed that 3.9 mW cm⁻² of power was available from the curved generator, while, the shoe experiment showed 25 V and 20 μA of electrical output for normal walking of a 68 kg person. Zhao and You in 2015 [382] suggested two shoe harvester designs to enhance power efficiency and comfortability (figure 12(b)). The harvesters consist of multilayer PVDF films embedded inside a plastic host and a flexible silicon rubber host, and sandwiched between two wavy surfaces. Test results showed that a higher power of 1 mW was generated by the harvester with plastic host, while the design with silicon host showed to be more comfortable. More recently, Ma et al in 2017 [383] introduced an interesting application of an insole-based piezoelectric energy harvesting device for unobtrusive user identification/verification. The idea is based on the uniqueness of gait pattern of each individual, which can be reflected in the output voltage signal of the energy harvester [384]. Experiments on 20 individuals showed an accuracy of identification of around 96% in addition to 160 μW of stored power in a storage unit.

Zoom In Zoom Out Reset image size

Work presented by Granstrom et al in 2007 [392] investigated the use of piezoelectric energy harvesting from the cyclic loading of the straps of a backpack. The work sought to developed a backpack with the capacity to generate power during walking as a means to reduce the amount of batteries carried by soldiers, emergency personnel, field workers, etc. Energy harvesting straps consisting of PVDF generators were tested and results were applied to verify a theoretical model of the piezoelectric straps under simulated tension measured from an instrumented backpack. The model predicted an average power of 45.6 mW could be generated by the backpack carrying a 444 N load. Expanding upon the work of Granstrom et al Feenstra et al in 2008 [385] investigated a backpack that used a mechanically amplified piezoelectric stack installed in the strap to harvest electrical power during walking (figure 12(c)). Results of experiments and simulations showed an average power of 0.4 mW could be generated by the piezoelectric stack with the backpack carrying a 220 N load.

Developing woven piezoelectric energy harvesters has been an interesting area of research that has emerged in the past decade. In an attempt to generate power from bending motion of the human body, Yang and Yun in 2011 [393] presented a flexible wearable piezoelectric device consisting of a PVDF layer attached on a curved polymer film. An initially curved structure provided a fast transition from the initial state to a bending state as a result of applied bending force. Tests performed on fabric with an embedded curved harvester, which was worn on an elbow joint, showed remarkable improvement in electrical output compared to flat piezoelectric structures. In addition to showing numerous advantages, such as a simple fabrication process, low cost, lightweight construction, and good wearability, the device was able to generate a high voltage output of 25 V for a very small amount of elbow motion. In another work, Zhang et al in 2015 [386] presented a fabric-like piezoelectric nanogenerator consisting of BaTiO₃ nanowires-polyvinyl chloride (PVC) hybrid fibers (a composite fiber fabricated by assembling high aspect ratio BaTiO₃ nanowires into PVC matrix), conventional cotton threads, and copper wire as electrodes for the piezoelectrics (figure 12(d)). Unlike brittle inorganic piezoelectric materials, the developed fabric was highly flexible, comfortable, and light, which made it promising for wearable application. Experiments on the fabric attached to a human elbow showed 10.02 nW of generated power with a 80 MΩ load resistor. The electrical output of this device was relatively low due to the existence of a gap between the copper electrodes and the piezoelectric fibers. An improved fabric-like piezoelectric energy harvester was proposed and tested by Song and Yun in 2015 [387]. The harvester had a fabric textile structure with polymer strains as warp threads and PVDF films with metal electrodes as the weft threads (figure 12(e)). Experimental results on an analytically optimized fabric showed that the device was able to generate a maximum power density of 125 μW cm⁻² with a 6.6 MΩ load resistance from body motion.

Several studies have proposed various mechanisms to convert the kinetic energy of limb motion to electrical energy. One major issue in harvesting the kinetic energy of the human body is the low frequency of body motion (less than 25 Hz [390]). Piezoelectric harvesters provide the highest energy conversion efficiency in resonant mode, which is typically higher than the frequency of motion in the human body. In order to obviate this limitation, different frequency upconversion mechanisms have been suggested. Pillatsch et al in 2012 [388] suggested a piezoelectric energy harvester with multiple piezoelectric bimorphs with a tip magnet and a rolling element that passes over the beams (figure 12(f)). As a result of external motion, the rolling mass rolls over the piezoelectric beams and, because of the snap-release mechanism, the beams resonate at their natural frequency. Experimental tests showed that the device could generate 2.1 mW of power at an excitation frequency of 2 Hz. This group presented another frequency upconversion mechanism in 2014 based on a rotating mass system to supply power for a wrist watch, similar to the system commercially used in Seiko Kinetic wrist watches, but with lower friction [389]. Using an eccentric rotating proof mass with a permanent magnet and a fixed piezoelectric bimorph with a tip magnet mass, this system generated 43 μW of power at a frequency of 2 Hz (figure 12(g)). This system is compatible with linear and rotational excitation modes. Similarly, the application of a piezoelectric system as a self-powered human activity recognition system was introduced and tested by Khalifa et al in 2015 [394] to obviate the power limitation of current accelerometers. Wahbah et al in 2014 [395] also showed that a combination of thermoelectric and piezoelectric resonant harvesters can be utilized on the wrist to generate power. The impact-driven piezoelectric energy harvester proposed by Wei et al in 2013 [396] is another example of a frequency upconversion system suggested for power harvesting from human body motion. A novel design of a low frequency piezoelectric energy harvester was proposed by Shukla and Bell in 2015 [390]. A frequency upconversion system including a rotor pendulum with multiple strikers and multiple PVDF bimorphs, called PENDEXE, was designed to generate power from waist motion (figure 12(h)). Tests showed that the device could generate 290 μW of power from the very low frequency motion of the waist (2 Hz) for normal walking.

In addition to the previously mentioned applications, energy harvesting systems using wearable piezoelectric generators from knee motion, jaw movement, breathing, and the ear canal have been presented. Pozzi et al in 2012 [391] suggested a knee joint wearable piezoelectric energy harvester with four fixed piezoelectric bimorphs and 74 plectra embedded in a rotating ring (figure 12(i)). The device attached to a human knee generated 2.06 mW of power from normal walking with the frequency upconversion mechanism. Later in 2016 [397], this group improved the generated power to 5.8 mW by altering the mechanical buckling mechanism with a magnetic buckling mechanism. Delnavaz and Voix in 2014 [398] showed that a hybrid generator consisting of a very simple piezoelectric ring and an electromagnetic generator located inside the ear canal could generate 0.5 μW of energy out of 5 mW of potentially available power. This group has also presented a piezoelectric energy harvester for power harvesting from jaw movement [399]. The system consists of a piezoelectric fiber composite plate placed under the chin and attached to a head-mounted device using two elastic rubber straps, and was found to generate 7 μW of power during chewing with slight discomfort for the user. Using an integrated array of piezoelectric films and a harvesting circuit placed within a pant belt, Abdi et al in 2014 [400] showed that 1.5 mW of power could be harvested from breathing. This power was generated without any active human engagement in the process and from a very low frequency of motion of breathing (between 0.25 Hz to 0.35 Hz).

Table 5 summarizes the energy source, transducer type, generator material, dimensions, input excitation, and output power of the different wearable piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.

The emersion of smart implantable medical devices in the past two decades has improved therapeutic and diagnostic methods to a large extent. One issue with implantable devices is the limited life time of batteries, which require intermittent replacement. In many cases, replacing the battery or device requires rigorous procedures associated with a high risk for patients. In order to tackle the current power limits of these devices, researchers have investigated implantable piezoelectric energy harvesters aimed at powering electronic devices embedded in the human body.

In 2011, Almouahed et al [401] suggested an instrumented knee replacement with four piezoelectric transducers placed in the tibial tray of the knee implant. An optimized design of the device with respect to the transducer location, material, and dimension, presented in 2016, showed that up to 4.276 mW of average power at an optimal resistive load of 50 kΩ could be obtained from a knee implant under a realistic knee load profile [402] (figure 13(a)). In a study performed by Holmberg et al in 2013 [403], a wireless self-powered sensory system was embedded inside the tibial tray of a knee replacement that utilized a piezoelectric stack which provides the required power for six capacitive force sensors. Experimental results for a 55 kg person showed that the piezoelectric harvester could harvest 1.051 mJ energy per step, which was sufficient to power the sensors, signal conditioning circuits, and wireless data transmitter with a low duty cycle. More recently, Safaei et al in 2018 [404] presented an instrumented knee replacement with four piezoelectric transducers located in the bearing of the implant for sensing knee forces and contact locations, and energy harvesting from human gait. Power harvesting test results showed that the energy generated from one hour of walking in normal conditions and stored in a capacitor was sufficient to power 9 min of sensing and data processing, and 5 s of wireless data transmission of a low-power biomedical sensing circuit published in the literature. An improved design of an instrumented knee bearing with six embedded piezoelectrics was proposed by the same group in order to measure compartmental forces and contact locations (figure 13(b)) [405]. The device was numerically optimized and experimentally validated. Result showed that the system was able to track the location of medial and lateral forces and location of contact points acting on the conforming surface of the bearing.

Zoom In Zoom Out Reset image size

Currently, implantable electronic devices, such as cardiac pacemakers and cardiac monitors, are powered with embedded batteries. Due to the limited lifetime of batteries, a surgical procedure is needed to replace the batteries, which exposes the patient to a variety of risks. Piezoelectric energy harvesting provides the potential to power the implanted electronic devices with energy collected from internal organs, such as the heart, lungs, diaphragm, etc. Karami and Inman in 2012 [34] investigated energy harvesting from heartbeat vibration using a nonlinear piezoelectric energy harvester to power pacemakers. Measured heartbeat data from the literature was used to perform simulations which demonstrated that a nonlinear cantilever harvester with dimensions of 27 × 27 × 0.080 mm³ could to generate over 3 μW of power over a broad range of heart rates from 7 beats per minute to 700 beats per minute. This group, later in 2016 [406], utilized several device configurations including fan-folded, elephant, and zigzag configuration for heartbeat energy harvesting, and showed that the fan-folded configuration allowed the device to generate the most power (figure 13(c)). Experimental results showed that this system generated 15.2 μW of average power, which is sufficient to power a pacemaker, and the output was robust to variation in heart rate [411]. Advantages of the system included MRI compatibility, small size, and a mechanism that was contact-less with heart tissue.

Several researchers have proposed various designs of flexible energy harvesters for power generation from cardiac motion. Hwang et al in 2014 [407] utilized a flexible single crystalline PMN-PT piezoelectric energy harvester to develop a self-powered cardiac pacemaker (figure 13(d)). Using PMN-PT with a piezoelectric strain constant of d33 = 2500 pC N⁻¹ (almost four times more than PZT and twenty times more than BaTiO₃), the implemented device placed in the cardiac muscle of a live rat showed a relatively high output current of 0.223 mA and output voltage of 8.2 V. Similarly, in 2015, Lu et al [408] developed an ultra-flexible PZT energy harvester to be integrated with the heart. The device was fabricated using transfer printing technology by applying PZT film between extremely soft substrate layers and implanted in a swine (pig) heart for in vivo testing (figure 13(e)). The effects of various suture fixation, mounting locations, and orientations were investigated in open/close chest and anesthesia/conscious conditions. A 3 V peak to peak generated voltage from this generator was found experimentally and is sufficient to power a cardiac pacemaker.

In an exceptional work, Dagdeviren et al in 2014 [409] applied several advanced material and engineering processes to develop a biocompatible and flexible piezoelectric energy harvester. A multilayer PZT structure was encapsulated in biocompatible material (polyimide) to minimize the risk of failure or immune system response (figure 13(f)). The device was evaluated in cell cultures, on large-scale, live animal models using various locations and orientations, and on different organs including the heart, lungs, and diaphragm. Adequate energy to power a pacemaker (1.2 μW cm⁻²) was obtained from the experiments, and the biocompatibility, mechanical properties, and electrical properties were tested for 20 million cycles under moist and hydrogel environment. More recently, Jeong et al in 2017 [412] fabricated a flexible harvester based on a lithium niobate (LiNbO₃)-doped KNaNbO₃ (KNN) lead-free piezoelectric thin film and implanted it into a porcine (pig) chest. In vivo tests showed that the harvester could generate up to 5 V and 700 nA from the heartbeat.

In a study presented by Deterre et al in 2014 [413], a very small piezoelectric energy harvester was developed for power generation from ordinary blood pressure variations to power a leadless pacemaker. The device was a microfabricated spiral-shaped piezoelectric beam within an ultra-flexible packaging equipped with a 10 μm diaphragm. The harvester had a diameter of 6 mm and volume of 21 mm³. Experimental results from an optimized design of the device showed a power density of 3 μJ cm⁻³ per heartbeat. The application of a sealed, flexible PVDF film for energy harvesting from pulsation of the ascending aorta of the heart was suggested by Zhang et al in 2015 [410] (figure 13(g)). Several experiments were performed to investigate the in vitro, in vivo, and sealing performance of the suggested flexible, implantable piezoelectric generator. The results showed that 681 nW and 30 nW were generated from in vitro and in vivo experiments, respectively, and the sealing package successfully prevented the device from being penetrated by saline water. The authors note that further optimization is needed on this design to improve the performance of the piezoelectric generator.

In the work of Jang et al in 2015 [414], a conceptual design of a piezoelectric artificial basilar membrane (ABM) was developed to be used as the front end of a cochlear implant. The fabricated prototype incorporated an array of eight piezoelectric cantilevers with a frequency range of 2.92 to 12.6 kHz. The results of their work demonstrated the feasibility of using the piezoelectric ABM for frequency separation. Following the work of Jang et al, the design of a fully-implantable cochlear implant using an array of thin film piezoelectric acoustic transducers was presented by Ilik et al in 2018 [415]. The system consisted of an array of piezoelectrics cantilevers designed to be placed on the eardrum or ossicles to generate the signal for neural stimulation. Eight piezoelectric cantilevers were placed facing each other with various lengths to result in a small size of 5 × 5 mm². Each cantilever was designed to cover a specific frequency range of acoustic sound pressure. Acoustic tests on a fabricated prototype of the harvester showed that a peak-to-peak voltage of 114 mV could be generated at a sound pressure level of 110 dB, which satisfies the power requirements of an implantable cochlear implant for stimulating the auditory nerves.

Table 6 summarizes the energy source, transducer type, generator material, input excitation, verification method, and output power of the different wearable piezoelectric harvesters described herein. This table attempts to provide the most relevant information for comparison purposes; additional details of each study can be found within the respective manuscripts.