Vertically aligned zinc oxide nanowires electrodeposited within porous polycarbonate templates for vibrational energy harvesting

A piezoelectric nanogenerator has been fabricated using a simple, fast and scalable template-assisted electrodeposition process, by which vertically aligned zinc oxide (ZnO) nanowires were directly grown within a nanoporous polycarbonate (PC) template. The nanowires, having average diameter 184 nm and length 12 μm, are polycrystalline and have a preferred orientation of the [100] axis parallel to the long axis. The output power density of a nanogenerator fabricated from the as-grown ZnO nanowires still embedded within the PC template was found to be 151 ± 25 mW m−3 at an impedance-matched load, when subjected to a low-level periodic (5 Hz) impacting force akin to gentle finger tapping. An energy conversion efficiency of ∼4.2% was evaluated for the electrodeposited ZnO nanowires, and the ZnO–PC composite nanogenerator was found to maintain good energy harvesting performance through 24 h of continuous fatigue testing. This is particularly significant given that ZnO-based nanostructures typically suffer from mechanical and/or environmental degradation that otherwise limits their applicability in vibrational energy harvesting. Our template-assisted synthesis of ZnO nanowires embedded within a protective polymer matrix through a single growth process is thus attractive for the fabrication of low-cost, robust and stable nanogenerators.


Introduction
Since 2006, when the concept of a zinc oxide (ZnO) piezoelectric nanogenerator (NG) first appeared in the literature [1], there has been a wide range of ZnO-based devices published with different geometries and architectures. These include single nanowires (NWs) [2], vertical NW arrays [3,4], lateral NW arrays [5], template-based nanotube (NT)/NW arrays [6,7] and other morphologies [8,9]. Alongside their use in NGs, ZnO NWs have also been studied for applications in solar cells [10,11] and gas sensing [12]. In the context of vibrational energy harvesting, NWs outperform bulk or thin film structures due to their high aspect ratio and high surfaceto-volume ratio, which provide improved sensitivity to lowamplitude ambient vibrations and reduced fragility [13]. Piezoelectric performance is typically quantified in the literature by the d 33 value, a normal component of the piezoelectric charge coefficient, defined as the rate of change of the dielectric displacement with respect to stress at constant electric field when the applied stress is parallel to the polarization [14]. The NW geometry is particularly attractive in this respect as theoretical [15][16][17][18][19] and experimental [20] studies have suggested that the bulk value of d 33 will vary from the equivalent NW values due to their surface-like nature, in particular that d 33 increases as the NW diameter decreases, and similarly for the Young's modulus, Y [20]. Piezo force microscopy measurements performed on ZnO NWs reported d 33 as 0.4-9.520. More recent studies have indicated a d 33 increase of ∼18% in ZnO NWs as compared to bulk [21]. For bulk ZnO, experimental measurements have shown the electromechanical coupling factor, k 33 2 which describes how effective a piezoelectric material is at converting mechanical energy to electrical energy [22], to be 0.17-0.23 [23,24] and for comparison, bulk barium titanate (BaTiO 3 ) to be 0.25 [24]. The piezoelectric performance of ZnO is therefore comparable to lead-free ceramics such as BaTiO 3 , with ZnO having the advantage of being relatively easy to synthesize into nanostructures [25].
The size, morphology and crystallinity of ZnO NWs are determined by the exact growth mechanism [26]. Wet chemical methods are typically relatively low-temperature, lowcost, scalable, and the least hazardous, as well as being effective at controlling size morphology and properties of the NWs. Within the wet chemical methods, electrodeposition is possibly the simplest, requiring the lowest temperature and pressure, and thus more cost-effective and scalable [26]. It is also possible to use a wide range of materials and control the growth through deposition parameters [27]. NWs grown by electrodeposition are in general found to exhibit better crystallinity and piezoelectric properties than those grown by other methods such as hydrothermal synthesis [28]. There are examples of polycrystalline ZnO NW arrays in the literature where the NWs consist of crystallites grown via templateassisted electrodeposition (TAED) in both anodized aluminum oxide (AAO) [29][30][31][32] and, more recently, polycarbonate (PC) templates [33] with various orientations. In comparison, there are limited examples of single crystal [002] axis orientated TAED NW arrays although it has been shown in AAO [34] and in PC [35], and also with a growth direction perpendicular to the (101) [36] plane or no preferred orientation [37].
Here we present vertically aligned, high aspect-ratio polycrystalline ZnO NWs, grown via TAED with preferred orientation of the [100] axis parallel to the long axis of the NWs hence with crystal plane growth parallel to the [002] caxis. With the NW array directly grown and embedded in a PC template, the resulting NG comprised a flexible ZnO-PC composite with an Ag sputtered electrode on one side and Al foil on the other. The NW electrodeposition was performed with a standard three-electrode cell directly within the PC The balanced reaction is therefore: Thus, under an applied potential, the nitrate ions reduce to nitrite ions in an aqueous solution which produces excess OH − ions in the region close to the working electrode. This increases the local pH and both the OHions and the Zn 2+ ions diffuse through the pores towards the base of the template. Depending on the reaction rate of the nitrate reduction, which is determined by the applied potential, there is either instantaneous nucleation directly onto the working electrode followed by growth for relatively high potentials or there is solution supersaturation at lower potentials where the Zn(OH) 2 , or ZnO if the temperature is above ∼50°C, precipitates in the electrolyte and then deposits onto the working electrode surface, again followed by growth [27,38,39]. This mechanism, which has previously been hypothesized, was recently experimentally confirmed for the first time by in situ transmission x-ray microscopy of electrodeposited ZnO by Tay et al [27]. The growth and the crystallinity of the NWs has been found to be affected by the choice of template, applied potential or current density, electrolyte bath composition, pH, temperature, stirring and sonication [40]. In the present study, PC rather than AAO templates were used because of the higher flexibility of the former which allows the final NG device to be more compliant and less prone to mechanical fracture. Additionally, the presence of the template reduces both the likelihood of electrical shorting when the electrodes are applied in the final NG device, as well as the likelihood of gases being adsorbed onto the surface of the NWs. This can help reduce the effect of a reduced polarization potential due to external screening of the polarization charges resulting in a reduced output voltage, without the need for surface passivation as previously reported [41,42].

Fabrication of ZnO NWs
A commercially available nanoporous PC track etched membrane (Whatman ® Cyclopore ® ) of diameter 25 mm with nominal pore diameter ∼200 nm, pore length ∼12 μm and porosity (4-20)% was first prepared by submerging it in deionized water and cleaned in an ultrasonic bath (U50, Ultrawave) for 3 min and subsequently dried at ∼100°C for 10 min. A ∼100 nm Ag electrode was sputtered on one side using a sputter coater (K575, EMITECH) at 40 mA for 4 min. A circular shadow mask was used to sputter a ∼3 cm 2 area on the ∼5 cm 2 template and a Cu wire was attached to the electrode using electrically conductive adhesive transfer tape (3M™), aluminum foil (Sphere Consumer Products plc) and silver conductive paint. The electrode was then covered in polyimide tape (3M™) to prevent deposition of the back surface. Zinc nitrate hexahydrate Zn(NO 3 ) 2 ·6H 2 O crystallized 99.0% (Sigma-Aldrich) was dissolved in deionized water to make a 0.1 M solution, pH≈4-5 (pH 510 series, Oakton instruments). The three electrodes, as described in the main text plus a magnetic follower (Fisher Scientific), were submerged in 100 ml of this solution and heated in a water bath to 75°C. A potentiostat (VersaSTAT 4, Princeton Applied Research) was used in chronoamperometry mode: -1 V was applied between the working electrode and the double junction Ag/AgCl (with saturated KCl solution) reference electrode (Sigma-Aldrich) for ∼15 min and the  current flow between the working and counter electrodes (platinum foil 20 mm×20 mm) was measured with the supplied software package (VersaStudio).

ZnO NW array characterization
Field-emission scanning electron microscopy (FE-SEM, FEI Nova NanoSEM) was used for imaging of the NWs. The NW diameters were measured by using the scaled scanning electron microscope (SEM) image (ImageJ). For the freed NWs, a small piece of the ZnO filled PC templates was placed on a SiO 2 wafer and then the PC template was dissolved with a few drops of ReagentPlus ® 99% chlorobenzene (Sigma-Aldrich). The sample was placed on a hotplate (C-MAG HS 4, IKA ® ) for ∼5 min to speed up the evaporation of the solvent and then placed in a furnace (Carbolite Furnaces) for 30 min at 500°C at a heating rate of 1°C min −1 to burn away the entire PC template. All samples were coated in palladium using a sputter coater at 40 mA for 20 s. X-ray diffraction (XRD) analysis was performed with a diffractometer (D8 Advance, Bruker) with a Cu Kα source and a non-rotating sample holder. For the high resolution TEM (HRTEM) (JEOL 4000EXII) imaging, the sample preparation was the same as for SEM except that the white-brown powder left after burning the template was dispersed in 0.5 ml of absolute ethanol (analytical reagent grade, Fisher Scientific). A droplet was then drop cast onto a Cu coated TEM grid and naturally dried in air for 20 min.

Fabrication and testing of the NG
From a ZnO NW filled PC template, a second electrode was added using firstly a layer of electrically conductive adhesive transfer tape with a piece of Al foil on top of the same area as the original electrode. A second Cu wire was attached with silver paint to this electrode. The energy harvesting set-up used to measure the voltage and current characteristics of the NG has been previously described [43] with a digital multimeter (Keithley 2002) for voltage measurements and a picoammeter (Keithley 6487) for current measurements. The whole setup was also enclosed in a grounded Al box which acts as a Faraday cage. Figure 1(b) demonstrates the typical current versus time plot for electrodepositions at -1.0 V. Immediately after the potential is applied (phase i), instantaneous nucleation occurs on the working electrode and this surface charge forms the first layer of the electric double layer, causing the current to suddenly increase. The sudden drop in current is then due to the formation of the second diffusion layer as the Zn ions flow towards the electrode and simultaneously reduce to form ZnO and precipitate on the surface. This causes a concentration gradient of Zn ions, decreasing towards the working electrode, thereby causing more of them to flow towards the electrode and continue to deposit. As the pores are filled in this way the current is approximately constant because the surface area exposed to the electrolyte is constant (phase ii). Once the pores are filled, the NWs emerge from the template and begin to form a film on the top and this steady increase in surface area causes a steady current increase (phase iii) [39]. The deposition was stopped at the onset of this stage in order to have an array of NWs without being connected by a film. Figure 2(a) shows a top-down SEM image of the NWfilled PC template showing a clear difference in contrast between the template and the ZnO. The presence of Zn is further confirmed by energy dispersive x-ray analysis in supplementary data S1. The black holes observed indicate regions where the NWs have not reached the surface or not filled the pores. This could be due to organic contaminants or trapped air in the pores. Slightly overgrown NWs are also visible in this image. The diameter distribution is displayed in figure 2(b), which gives an average NW diameter of (184±1) nm. The average pore size is ∼(227±92) nm [7] indicating that the NWs grow to less than the diameter of the pores themselves. The nominal thickness of the template is ∼12 μm and after freeing the NWs from the template through a controlled burning process, the approximate length is ∼10 μm as can be seen in figure 2(c). This difference is likely due to breakages of the NWs from the electrode during sample preparation. Figure 2(d) shows how the NWs break up into smaller pieces when released, and hence the template serves to protect the NWs while they are still embedded. Figure 3(a) shows an XRD plot of the ZnO NWs deposited at -1.0 V. The presence of other peaks besides the preferred orientation indicates that the ZnO NWs are either polycrystalline with an overall preferred orientation or they could be single crystals all with different orientations but most having the same preferred orientation. The texture coefficient (TC) of the (100) peak is 3.9 in this range and so a clear preferred orientation is visible (see supplementary data S2). A deposition at -1.2 V was also performed where [002] preferred orientation was found before cleaning the top Table 1. Electrical performance of the zinc oxide-polycarbonate nanogenerator presented here. I sc , V oc , P A , P d , P′, E, χ, η T , η s and χ′ are the average peak short-circuit current, open-circuit voltage, root-mean-square (rms) power density with respect to area, rms power density with respect to volume, specific rms power, average energy density per cycle, energy conversion efficiency, stress and strain figures of merit [22] and overall device efficiency. surface of the template with HCl, which switches to [100] axis orientation after cleaning. (The XRD results before and after cleaning at −1.2 V are presented in supplementary data figure S2.) At this higher voltage, the growth is faster which leads to the formation of a film on top of the template. This does not occur at the lower potential as long as the deposition is stopped at the correct time. Figure 3(b) shows the TEM image of a single freed ZnO NW. The NW is composed of large ZnO polygonal crystallites of sizes between 100 and 150 nm (one dimension). The inset shows the HRTEM image of one ZnO crystallite, which indicates the lattice fringe spacing of plane (100)=0.28 nm of phase-pure ZnO. The growth direction of the NW is along the [100] axis which is parallel to the long axis of the NW. Figure 3(c) shows the selected area electron diffraction pattern of the ZnO NWs that mostly shows a single crystalline pattern with a superimposed polycrystalline ring pattern from the adjacent polycrystalline crystallites of ZnO. Major diffraction planes such as (100), (101) and (102) could be indexed. The diffraction spots reveal a hexagonal lattice structure of ZnO NWs. From the XRD, SEM and TEM measurements it can be stated that the PC template can influence the growth rate of the different crystal planes. Without the template, NWs with the [002] axis orientated parallel to the long axis tend to grow because the high surface energy polar faces grow faster than the non-polar faces. This may be altered by the PC pore surface energy and in this case, the NWs are formed of single crystal nanoparticles with the [002] axis parallel to the plane of the template stacked on top of one another within each pore.

Results and discussion
A NG based on the ZnO NWs grown by TAED was fabricated and the electrical output was measured in response to periodic impacting by an oscillating mechanical arm at a set frequency in a bespoke energy harvesting measurement setup that has been previously described in [43]. An integrated schematic of the NG consisting of vertically aligned ZnO NWs embedded within a PC template is presented in supplementary data S3, which shows how the NG was subjected to a periodic impacting force parallel to the long axis of the NWs. The mechanical excitation was such that the NWs were being periodically compressed along their lengths and then released, resulting in the measured electrical output. The ZnO-PC nanocomposite NG is an example of a stress-driven NG [22,44], as the ZnO NWs have a relatively large areaweighted stiffness compared to the PC template in which they are embedded. Figure 4(a) displays the voltage measured across an impedance-matched resistance of 20 MΩ at 5 Hz, 10 Hz and 25 Hz and in (b) across 1 MΩ, 10 MΩ and 100 MΩ resistors at 5 Hz (see supplementary data S4 for the circuit diagram). The peak-to-peak voltage (V pp ) increases with frequency due to higher acceleration, and hence force, of the impacting arm. As the resistive load increases the voltage drop and hence V pp increases. The width of the peaks widens as the load increases due to the longer time constant (product of resistance and capacitance) resulting in slower charge and discharge. Figures 4(c)-(d) show the equivalent plots for current measured in series with the NG and the resistors. The output power was measured across ten different resistors and is plotted alongside root-mean-square voltage (V rms ) across each load in figure 4(e). The peak in output power occurred under impedance-matched conditions across a resistor ∼20 MΩ. Figure 4(f) shows the fatigue characteristics of the NG under continuous impacting through 24 h. After an initial ∼11% decrease in V pp within the first 2 h, the NG maintained a reasonably constant output voltage throughout the fatigue test, and the device did not show any visible physical damage. Table 1 outlines the full electrical characteristics with an estimated input energy of ∼3 μJ per cycle of the impacting arm (see supplementary data S4 and S5). Additionally, energy conversion efficiency and figures of merit, as defined in [22], were evaluated for the ZnO NWs grown by TAED (see supplementary data S4), and found to be comparable to previously reported values, while an overall device efficiency was also calculated as shown in supplementary data S5 and reported in table 1. The effect of crystallinity of the NWs on NG performance has previously been studied, and indicates that as the size of the crystallites within ZnO NWs [45] or nanotubes [6] increases, the voltage and current output increases. This is due to the decrease in the number of grain boundaries and defects for larger crystallites, which is seen in XRD patterns as the narrowing of the preferred orientation peak [6].

Conclusion
In summary, TAED was used to grow a vertically aligned ZnO NW array with [100] axis preferred orientation embedded in a PC template, which was then directly fabricated into a piezoelectric NG for applications in vibrational energy harvesting. This method is simple, cost-effective and scalable, with minimal steps required for the fabrication of a robust NG with good energy harvesting performance, as evidenced by voltage and current data at varying impacting frequencies and measured across varying loads, for up to 24 continuous hours. The aligned nature of the embedded nanowires ensures good piezoelectric performance across the entire device under impacting, while the PC template ensures mechanical stability and longevity of the device. The confined growth process achieved through TAED thus overcomes problems typically arising due to mechanical and/or environmental degradation in ZnO-based nanostructures that limits their applicability in vibrational energy harvesting. Importantly, the TAED method for growing ZnO nanowires and subsequent NG fabrication require only a single growth process and minimal postdeposition processing, and can thus be easily scaled up for low-cost and long-lasting energy harvesting applications.
ERC-2014-STG-639526, NANOGEN). SK-N, RAW and AD are grateful for financial support from this same grant. FLB and RAW thank the EPSRC Cambridge NanoDTC, EP/ G037221/1, for studentship funding. S-LS acknowledges support through the EPSRC grant EP/M010589/1. FLB fabricated the samples and FLB and TD performed the measurements reported. SK-N designed and guided the experimental work. AD and S-LS conducted the HRTEM imaging. RAW and FLB built the energy harvesting measurement set-up. FLB and SK-N co-wrote the paper. All authors discussed the results and commented on the paper. All authors have given approval to the final version of the manuscript. Supporting data for this are is available at the DSpace@Cambridge data repository (https://www. repository.cam.ac.uk/handle/1810/255931).