Introduction

Due to their various properties, zinc oxide nanoparticles in general and nanowires (ZnO NWs: linear particles with high aspect ratio) in particular are promising materials for a broad range of applications, e.g., piezoelectric sensors (Jeong et al. 2015) and piezoelectric energy harvesting (Lu et al. 2012; Wang 2008; Wang et al. 2015; Wang and Song 2006; Yang et al. 2009; Zhu et al. 2010), gas sensors (e.g., NOx, CO, O2) (Zhang et al. 2012), superhydrophobic surfaces (Mardosaitè et al. 2021), photocatalysis (Rackauskas et al. 2017), antibacterial activity (Sirelkhatim et al. 2015), spintronic devices, biosensors, etc.

ZnO is characterized by a wide band gap (3.37 eV) and a large exciton-binding energy (60 meV). Due to the wurtzite-type crystal structure (zincite), the asymmetric displacement of the ions Zn2+ and O2− upon pressure creates a piezoelectric current. Additionally, ZnO is photostable, temperature resistant and nontoxic. Further, it is a semiconductor of high potential for applications in optoelectronic devices, such as light-emitting diodes (LEDs), electrically pumped lasers, electron collection in photovoltaic cells and as transparent conducting oxide.

In order to explore the potential of zinc oxide nanoparticle-based technologies and to make them industrially viable, cost-efficient and scalable processes have to be developed, which allow tailoring the properties of the particles, such as their shape and size. Typical state-of-the-art processes to synthesize zinc oxide nanowires are wet-chemical methods (sol–gel processes, hydrothermal growth) (Arslan et al. 2022; Broitman et al. 2012; Hu et al. 2007; Maity et al. 2005; Wu et al. 2005; Xu et al. 2020; Zhang et al. 2003), laser ablation and pulsed laser deposition (Ghosh et al. 2013; Lynam et al. 2019; Rahm et al. 2006; Sun et al. 2004), electrochemical methods (e.g., potentiostatic, galvanostatic anodization) (Chang et al. 2002; Elias et al. 2011; Tello et al. 2021; Wang et al. 2002), physical and chemical vapour deposition (Falyouni et al. 2009; Huang et al. 2001; Khosravi-Nejad et al. 2019; Kong et al. 2001; Navarrete et al. 2022; Sharma et al. 2010), molecular beam epitaxy (Heo et al. 2003) and flame spray pyrolysis (Height et al. 2006).

Their main drawbacks are that they require a lot of energy, reaction time, seed layers, vacuum chambers, autoclaves, or catalysts. The hydrothermal growth, for example, allows oriented ZnO NW growth with a quantum dot seed layer at 65–95 °C for 3–7 h per growth step (Hu et al. 2007) (resulting in 2–8 μm long, vertically aligned ZnO NWs) or chaotic ZnO nanowire growth, which requires 12 h of autoclaving at 140 °C for 12 h, a cleaning procedure and several hours of drying (Wu et al. 2005) (resulting in up to 20 µm long nanowires). Treating large areas and generating large quantities of ZnO nanoparticles, especially nanowires, remain challenging tasks.

Plasma techniques, on the other hand, offer a broad bandwidth of different approaches to synthesize ZnO nanoparticles. E.g., it is possible to induce arcing in zinc swarf in oxygen-rich atmosphere with a magnetron (microwave air-plasma sputtering) to create ZnO nanoparticles of different length (Subannajui 2016). With electrodes emerged in an electrolyte with a zinc salt (e.g., ZnCl2), a solution plasma process allows forming ZnO nanoparticles with dopants when a salt containing the desired dopant (e.g. Fe) is added (Saqib et al. 2021). Similarly, an electrically conducting substrate, such as carbon fibers, may be coated with ZnO in a solution plasma process to form ZnO composite materials (Zhong et al. 2023). Even core–shell nanoparticles with ZnO shell can be synthesized from solid zinc sheets when they are emerged into a solution containing nanoparticles of a different kind and are treated with a plasma jet (Mohammed et al. 2022).

However, fast and cost-efficient process routes with high particle yield can be realized with atmospheric pressure plasma spraying (Primc et al. 2021). Generally, zinc-containing precursors (zinc nitrate, chloride, sulfate, acetate, …) or zinc powders are injected into a plasma jet, where they are converted into zinc oxide nanoparticles (flight-thru process). Primarily processes with zinc powder as starting material seem to be promising for scale-up, as they enable excellent ZnO nanoparticle production rates (24 g/min achieved by Peng et al. (2007), 20 g/min achieved by Ko et al. (2006)), limited mainly by the size of the plasma jet reactor, the zinc powder flow and the conversion rate. Depending on the reaction conditions, these nanoparticles may grow in different forms like spheres, rods, wires and tetrapods (Kim et al. 2008; Kim and Park 2009; Ko et al. 2006; Kumar et al. 2008; Lee et al. 2019; Lee et al. 2021; Liao 2006; Lin et al. 2005, 2009; Peng et al. 2007; Petzold et al. 2012). However, it is still challenging to control the growth of the ZnO nanoparticles and their morphology.

In this work, we develop a zinc oxide nanoparticle synthesis with a thermal atmospheric pressure plasma jet (APPJ) based on a direct current arc discharge. The water-cooled reactor wall allows collecting high yields of synthesized zinc oxide nanoparticles. Adopting the process parameters allows tailoring particle size and shape to form nanowires and nanorods as well as spherical particles and tetrapods. As proof of principle, we show that the synthesized ZnO nanowires have the potential to work as a piezoelectric material in flexible vibration sensors after poling when they are applied in a non-conductive matrix over electrodes.

Experimental

Zinc oxide nanoparticles (ZnO NPs) were synthesized from zinc powder (D50 = 5 µm, ECKART GmbH, Germany) and oxygen with an atmospheric pressure plasma jet reactor with water-cooled (~ 20 °C) outer wall (Fig. 1, scheme in a, picture in b). The commercial hot gas plasma jet (IC3 from INOCON Technologie GmbH, Austria) was operated with argon (5.0) as plasma gas (10 L/min) and mixtures of argon (5.0) and oxygen (5.0) as powder carrier gas (in sum always 10 L/min with 0–100 vol.-% O2). The powder was supplied from a powder feeder with an oscillating conveying channel (Medicoat AG, Switzerland) set to a flow rate of ~ 2 g/min.

Fig. 1
figure 1

Scheme (a) and picture (b) of the plasma jet reactor with the hot gas plasma nozzle IC3 (kindly provided by INOCON Technologie GmbH)

Full size image

A typical experiment is described as follows: with active cooling water, plasma gas and carrier gas flow, a direct current arc was ignited and maintained between the tip of the cathode and the plasma nozzle (hollow anode). Then, the operating discharge current was adjusted gradually (200–400 A). After 20 s of warm-up, the powder feeder was activated for 60 s. 10 s later, the plasma was deactivated, the temperature recorded (two positions, cf. Figure 1b), and the synthesized ZnO powder was collected from the main reactor wall in the area of the cooling coil.

The shape, length and thickness of the ZnO NPs were characterized using scanning electron microscopy (SEM). Due to the tiny diameters of the nanowires (reaching down to 10 nm), a TESCAN MIRA-3 microscope with a field-emission cathode was used. The Schottky field-emission cathode provides a narrow electron beam, which guarantees a superior image resolution at higher magnitudes (theoretically 1 nm) than a tungsten cathode SEM (Jiruše et al. 2014). For the analyses of the chemical elements, an OXFORD AZtecEnergy XT energy-dispersive X-ray spectroscopy (EDS) detector was used.

For the characterization by X-ray diffraction (XRD), the zincite/zinc nanoparticle powder samples were fixed on a glass specimen slide using a mixture of ethanol and glycerol (< 5%) for the dispersion of the particles. The height adjustment of the sample was carried out carefully to obtain a precision of about ± 0.25 mm. The characterization experiments by X-ray diffraction (XRD) were conducted on a D8 Discover diffractometer (Bruker, Germany) in parallel beam geometry (37 kV, 32 mA, Cu Kα radiation) to obtain phase composition, crystallite size and the lattice constants of the hexagonal zincite phase (ZnO). This device has a Sol-X energy-dispersive detector, an open Eulerian cradle and a polycapillary collimator. For each measurement, diffraction angles were scanned in the range of 20–100° 2θ with a corresponding step size of 0.05° and a counting time of 3.5 s/step. All experiments were conducted under air and laboratory conditions at 28 ± 1 °C.

The lattice cell parameters, the crystallite size and the phase contents were refined through the Rietveld technique; the TOPAS 5 Software by Bruker AXS, TOPAS V5 (1999–2014) was used. The crystallite size is here calculated assuming that each crystallite is composed of a set of columns along the scattering vector. The line broadening can be treated easily as a volume-weighted mean column length. No further assumptions in respect of shape and size are necessary. Furthermore, using the integral breadth (i. e., the width of a rectangle with the same height and area as the diffraction peak) and not the full width at half maximum of the diffraction peak, a Scherrer constant of 1 can be assumed (Laue 1926). The crystal structure data, taken from the Inorganic Crystal Structure Database (ICSD 2021) entries ICSD 26170 (hexagonal ZnO) and ICSD 247147 (elemental Zn), were used as starting values for the Rietveld refinement. Other crystal structures, e.g., atom position parameters or occupation factors, were not refined. A polycrystalline powder reference sample (LaB6 NIST 660a) and a solid polycrystalline corundum sample (Bruker) was measured using the identical diffractometer settings and taken as a reference for the correction of the instrumental influence regarding sample height error, peak width and shape.

For the piezoelectric sensors, 15 mg (labelled “ZnO NWs”) or 3.75 mg (labelled “¼ ZnO NWs”) of experimentally synthesized ZnO NWs were suspended into an acrylic resin (2.4 g Acrifix 2R 0190 (Röhm GmbH, Germany), 0.9 g methylmethacrylate (Sigma Aldrich, Merck KGaA, Germany) and 90 mg TPO-L (UV initiator from Sigma Aldrich, Merck KGaA, Germany)). To homogenize the ZnO nanowires, they were milled in a ball mill (Acrifix 2 R0190 with zirconium oxide beaker and milling balls: 500 rpm for 10 min in sequences of 10 s milling followed by 10 s break) before the other components were added. Further, a reference matrix was prepared without ZnO NWs.

For the piezoelectric experiments, the formulations were administered with a spatula onto dielectric substrates (self-adhesive Kapton tape (polyimide) on FR4 prepregs) with interdigit electrode structures (created with an OPTOMEC aerosol jet printer and PARU nano-silver ink in 350 µm line pitch and with 50 µm height). A poling potential (600 V and 700 V) was applied for 30 s or 60 s; after 15 s, a UV lamp was activated for 2 min to harden the acrylic resin. After curing, the piezoelectric currents of the sensors were recorded with a data acquisition system (DeweSoft Sirius, Slovenia).

Results and discussion

Zinc oxide nanoparticle synthesis and characterization

For the zinc oxide nanoparticle synthesis development, a commercially available plasma spray system (IC3 from INOCON Technologie GmbH, Austria) was combined with the water-cooled chamber of an oil diffusion pump as a reactor tube (cf. Figure 1, b). The upper part of the tube surrounding the plasma was hermetically sealed with a flange adapter, which allowed excluding the hardly controllable oxidative effect of ambient air. Preliminary experiments with zinc powder and oxygen-free carrier gas (argon) confirmed that the reactor was airtight. Figure 2 shows the different powders obtained on the flange adapter of the coating nozzle when zinc powder was injected into the plasma jet reactor without (a) and with oxygen (b). Without oxygen addition, the blue-greyish starting material (5 µm grain size) turns into a black powder of spherical zinc nanoparticles with < 200 nm diameter. With oxygen, on the other hand, zinc converts into its white oxide.

Fig. 2
figure 2

Pictures of the coating nozzle with flange adapter after the injection of zinc powder into the active plasma jet reactor without oxygen (a) and with oxygen (b)

Full size image

The influence of the plasma discharge current (200–400 A) and the oxygen rate (0–100% in the carrier gas) on the zinc oxide particle shape and size were studied in detail. The aim was to find reaction conditions that favour the growth of zinc oxide nanowires (particles with ≤ 200 nm diameter and ≥ 0.5 µm length).

When solely oxygen was used as powder carrier gas, small, primarily spherical nanoparticles (< 10 nm) were grown with ≤ 300 A (Fig. 3). At higher currents, the particles turned into nanorods with similar diameters and a length of 20–40 nm.

Fig. 3
figure 3

SEM micrographs (two magnifications) of the ZnO nanoparticles synthesized with fixed oxygen (100% O2 as carrier gas) but varying discharge current (200–400 A)

Full size image

When the current was fixed at 250 A and the oxygen rate of the carrier gas was reduced from 100 to 20% in argon, the spherical particles at 100% oxygen turned into nanoneedles, nanowires and particles with two or more branches (bipods, tripods, tetrapods) (Fig. 4). From 20 to 30% O2 the nanowire-like zinc oxide particles became nanorods, with 40% O2 they were already almost spherical nanoparticles. Overall, 20% oxygen in the carrier gas (10% oxygen in the plasma) enabled the best ZnO particle growth. Increasing the oxygen rate to 30% in the carrier gas (15% in the plasma) reduced the mean particle size from ~ 600 to ~ 300 nm (cf. Figure 4 and Fig. 9). Peng et al. (2007) described a similar observation for their plasma flight-thru process. They found that an increase in the oxygen rate supports the formation of ZnO nanowires up to a certain point, after which the particle size decreases. They argue that a higher number of ZnO nuclei is formed simultaneously with higher O2 flow rates, which increases the number of particles, but limits their growth. In their study, an oxygen rate of < 2.4% in the plasma resulted in decreased ZnO nanoparticles, when they investigated the effect of 0.5–30% O2 in the plasma.

Fig. 4
figure 4

ZnO nanoparticles synthesized with a fixed current (250 A) but varying oxygen rate (20–100%)

Full size image

To harden these findings, different currents were tested with a fixed oxygen rate of 20% in the carrier gas (Fig. 5). Nanoneedle- and nanowire-like particles were found from 250 to 350 A. Some of these particles were linear; others were partially branched (up to four branches). However, this confirms that, in our setup, 20% O2 in the carrier gas (10% in the plasma, respectively) is a sweet spot for the ZnO nanoneedle and nanowire synthesis.

Fig. 5
figure 5

SEM micrographs (two magnifications) of the ZnO nanoparticles synthesized with a fixed oxygen rate (20%) but varying current (200–400 A)

Full size image

The zinc and oxygen content in the nanoparticles were analysed with SEM/EDS. The atomic ratio of Zn:O in all samples was close to 1 when oxygen (or a mixture with argon) was injected as carrier gas (cf. Figure 6).

Fig. 6
figure 6

SEM/EDS images (a, b) and spectra table (c) from ZnO nanowires synthesized with a fixed oxygen rate (20%) but varying current (a: 250 A, b: 350 A)

Full size image

With XRD, the crystal structure of the ZnO nanoparticles was investigated. Typical XRD diffraction patterns of a Zn-rich sample (synthesized with 100% argon, 250 A: comparable to Fig. 2, a) and a ZnO-rich sample (synthesized with 30% O2, 250 A: comparable to Fig. 2, b) are shown in Fig. 7. There is a significant variation in the relative diffraction intensities corresponding to the phase composition or ZnO/Zn ratio. However, the changes in the diffraction angles indicating the lattice spacing of a specific maximum are only minor.

Fig. 7
figure 7

Typical XRD diffraction patterns of a zinc-rich ( +) and a zincite-rich (*) powder; the corresponding diffraction peaks according to the given Inorganic Crystal Structure Database (ICSD 2021) identification are indicated

Full size image

With the Rietveld refinement, lattice constants could be determined in the approximate ranges of a = 0.324 to 0.326 nm to and c = 0.520 to 0.524 nm. The observed ratios a/c are close to 0.62 (c/a ~ 1.6), which fits well with the reference data for the hexagonal zinc oxide phase (wurtzite-type structure: zincite, chemical composition ZnO) (Klingshirn 2007). Processes with oxygen (≥ 20%) in the carrier gas resulted in greyish-white (occasionally yellowish) powders consisting of nearly 100% zincite (> 95%). Without oxygen, black powders with zincite fractions below 43% were obtained from the bluish-grey starting material (as in Fig. 2a). Thus, the zinc powder was partially oxidized, although the plasma was protected from the surrounding air (with the remaining oxygen in the reactor and the starting material).

Additionally, the zincite fractions in the oxygen-deficient powders slightly differed in the a/c ratios from the oxygen-rich samples, which can be seen in Fig. 8. In the diagrams, the lattice constants a and c of the APPJ samples (crosses) are plotted with the corresponding data of ~ 70 entries of that phase in the Inorganic Crystal Structure Database (circles) (ICSD 2021).

Fig. 8
figure 8

Comparison of the ZnO lattice constants c and a in two magnifications (a, b); XRD data of the APPJ samples (crosses) and references (cycles). The samples processed without oxygen are found in region III, and those processed with oxygen in region IV

Full size image

The plotted phases can be arranged in four discrete regions:

Region I comprehends the samples investigated by Sowa and collaborators under high-pressure conditions (ICSD 2021). These samples display a strongly compressed cell.

Region II corresponds to the results of the work of Farooqi and Srivastava (Sowa and Ahsbahs 2006). Their work describes the synthesis of ZnO nanoparticles from ZnS nanoparticles by solid-state reaction. These samples display a larger a/c ratio in the range of 0.629.

Region III refers to APPJ samples in the actual work prepared without oxygen addition. The ZnO sample described in Ekicibil (Hasan Farooqi and Srivastava 2017), obtained by solid-state reaction, displays similar lattice constants as our samples in III.

Region IV contains actual samples prepared under oxygen and many published structures found in the database with stoichiometric compositions. Albertsson’s work (1989) includes high-temperature investigations shifted towards higher lattice constants due to thermal expansion (Ekicibil 2012)

It is evident from Fig. 8 that the oxygen deficiency during the process shifts the lattice constants a and c in a region not typical for zincite. This shift is caused most likely by oxygen defects in the crystal lattice. Thus, a deviation from the stoichiometric composition (ZnO1-x) may be induced in the zincite. The data of the ZnO particles synthesized with oxygen, on the other hand, correspond well with the literature values. These results confirm that the ZnO nanoparticles synthesized within the plasma jet with oxygen are composed almost entirely of a ZnO phase with a zincite structure as intended.

Comparing the crystallite size (determined with XRD) and the particle size (determined with SEM) reveals that both correlate similarly with the varied process parameters (discharge current and oxygen rate in the carrier gas) (cf. Figure 9). With a constant discharge current of 250 A, the crystallite size increases, when the oxygen rate is reduced from 100 to 20% in the carrier gas (50–10% in the plasma). The particle size reaches its maximum at 20% oxygen (10% in the plasma) with 250 A. As discussed above, with increasing oxygen flow rate the number of zinc oxide nuclei formed increases which is likely to act as an antagonistic mechanism to the particle growth. Kim et al. (2009) studied the crystallite size of ZnO nanoparticles synthesized with an APPJ reactor at 750 Torr (Kim and Park 2009). During their experiments, they found that higher O2 flow rates favour the crystallite growth, which was beneficial for the photocatalytic activity of their ZnO powders. However, the oxygen rates they tested were 10–25% in the plasma, so it cannot be excluded that they found the optimum for the crystallite growth with their setup, meaning that a further increase of the O2 rate might have limited or reduced the growth. A direct comparison is hard because of the different plasma setup used.

Fig. 9
figure 9

Particle size (solid bars) determined by SEM and crystallite size (dotted bars) determined by XRD concerning the process parameters discharge current (a with 20% O2 in the carrier gas) and oxygen rate in the carrier gas (b with a discharge current of 250 A)

Full size image

When the oxygen rate, on the other hand, is kept constant at 20% in the carrier gas, the particle and crystallite size maxima are found around 250 A. This means that the oxygen rate of 20% and the discharge current of 250 A allow optimal growth. In the reactor, a specific amount of energy is necessary to melt and evaporate the zinc particles (which also depends on the zinc flow rate) and to atomize/ionize the oxygen molecules. The small and curved exhaust tube causes turbulences in the reactor which keep the zinc and oxygen atoms and ions ready for reaction long enough inside the hot zone around the plasma so that zinc oxide nanoparticles are formed. The cooled walls, on the other hand, work as cold traps. There, zinc oxide particles grown in the gas phase are caught. The harsh cooling most probably provides as a quenching effect, stopping the particle growth. This explains why the particle size increases with increasing current first, as the hot plasma zone is extended and an optimum amount of energetic reactive species is formed. However, not only the temperature and kinetic energy of the gases and particles inside the reactor increase with the current when they pass the plasma jet, the gas and particle velocities increase, too (which is in accordance with numerical plasma simulations supported by the plasma jet manufacturer INOCON). The increasing velocities may be responsible for the particle size reduction with increasing current, because the particles are trapped faster on the cooled walls and their residence time in the gas phase is reduced significantly. Regarding the influence of oxygen, as discussed above, with increasing oxygen flow rate increases the number of zinc oxide nuclei formed which is likely to act as an antagonistic mechanism to the particle growth. Controlling the zinc oxide nanoparticle growth with our setup means, therefore, to set the favoured current and oxygen rate for the particles with the shapes desired.

Follow-up experiments revealed good repeatability for receiving nanowire-like ZnO particles with 250–350 A and 20% oxygen in the carrier gas. Similar results were obtained when compressed air was used instead of a mixture of oxygen in argon. This is of great interest for a high-throughput process and scale-up, as replacing the argon/oxygen mixture with compressed air reduces the process costs significantly.

In order to investigate the process yield, batch processes with different powder-feeding intervals were tested (1–5 min, 250 A, 20% oxygen in argon as carrier gas). After each experiment, the ZnO powder was sampled from the inner reactor wall of the main tube, where the majority of the product accumulates. Powder from the coating nozzle, the bottom of the tube and the side tube was discarded, meaning that the effectively collectible sample amount is slightly higher than that taken for calculating the process yield.

No attempts were made to filter the exhaust gas during these experiments. Previous tests with different filter procedures and materials like water bubbling and filter tissue did not increase the yield significantly. The best filtering effect was achieved with water, which allowed accumulating appr. 10–20% more sample powder. One crucial factor limiting the filter capacity was the back pressure, which, when too high, either triggered the safety relief valve of the powder feeder or shifted the reaction conditions in a manner, in which indeed more powder was trapped inside the reactor, but in the form of unreacted black zinc.

The process yield for ZnO was calculated as the ratio of the ZnO powder collected from the main tube and the transported zinc powder under the premise that the oxidation rate was 100% (according to XRD analysis, the real oxidation rate was > 95%). With increasing powder-feeding period, the absolute ZnO amount increased almost linearly until 4 min. The typical yield was ~ 40%. A plateau seemed to be reached with more extended powder-feeding periods and the yield began to decrease slightly. Most probably, the capacity of the walls to catch and keep powder despite the hot gas flows was reached: they were saturated with product. In a series of more than 20 batch processes with 4 min powder-feeding time, the process yield could be specified at 47 ± 9%. Thus, approximately half of the powder left the reactor with the gas flow. As mentioned above, our attempts to filter the exhaust gas hardly increased the total yield. However, this shows there is enough potential for further development, regarding industrial applications maybe in combination with a cyclonic separator.

Evaluation of the potential for piezoelectric sensors

Test sensors were developed based on so-called interdigit electrode structures (Fig. 10) to evaluate the potential of the plasma-synthesized ZnO nanowires for piezoelectric sensors. The electrodes were printed with an OPTOMEC aerosol jet using PARU nano-silver ink on dielectric substrates (self-adhesive Kapton tape on FR4).

Fig. 10
figure 10

From left: Design (a: dimensions in mm), photo (b) and microscope image (c) of interdigit electrodes made from silver ink printed onto dielectric substrates

Full size image

ZnO nanowires suspended in an acrylic resin were milled, mixed with methyl methacrylate and a UV initiator and applicated as a thick film (~ 200 µm) onto the electrodes, so that the space between the electrodes was filled with the nanowires containing matrix. To increase the piezoelectric effect of the nanowires, a poling voltage was applied (hysteresis poling) while the matrix was hardened under UV irradiation.

In most samples, the poling led to short circuits and arcing after some seconds, which irreversibly damaged the devices. The observed delay between poling start and short/arcing indicates that the nanoparticles started to align within the electric field until a conducting bridge between the electrodes was built. In a further test with acrylic resin without nanowires, a significantly increased dielectric strength and a significantly increased poling voltage up to arcing were observed. The alignment of the ZnO nanowires as chains during poling is thus likely to play a significant role in arcing. To avoid chain forming the concentration of the ZnO NWs in the matrix was reduced, and the poling voltage and time were varied.

Eventually, the successfully poled devices were tested.

The samples tested are listed in Table 1. ZnO nanopowders with nanowire-like structures were chosen for the piezoelectric experiments comparable to those in Fig. 4 and Fig. 5 that where synthesized with 250 A and 20% O2. Figure 11 shows the signals of the samples and the references (a) during bending, and a picture of the experimental setup (b) with a sample in front (yellow device). For these proof-of-concept experiments, the electrodes of the devices were contacted, and the electric charge was recorded as a function of time while the devices were stimulated mechanically. The films were bent by an operator using a ruler made of non-conductive material. Two thirds of the samples were fixed on a table, one third was free to be bent on the edge downwards. The reason for the different responses in Fig. 11 is that the sensors were not stimulated continuously. The pressure and pauses were varied by the operator on purpose, but were not quantified.

Table 1 Samples prepared for the piezoelectric experiments
Full size table
Fig. 11
figure 11

a Recorded charge upon bending of the samples and b setup of the piezoelectric experiments (the yellow sample is bent with a pen)

Full size image

The "Dummy" and the "Matrix" sample show some responses to the mechanical stimulation, which have to be attributed to the electrostatic charging of the polymer stick with which the samples were bent. Similarly, the unpoled ZnO-loaded sample (“ZnO, 0 V”) did not respond to the mechanical stimulation. Only with poling, the ZnO-loaded samples react with signal peaks to the stimulation, which can be attributed to the piezoelectric effect. Because of the low particle loading and, thus, low charge displacements the piezoelectric effect ranges in the order of magnitude of the electrostatic charging. The values are, therefore, to be understood as semi-quantitative. From the poled ZnO-loaded samples, the sample “ZnO, 600 V, 30 s” shows the most significant charge peaks, which may be attributed to the four times higher ZnO concentration in the matrix. In terms of amplitude with ~ 100 pC, however, it is still orders of magnitude below the piezoelectric effect of polymer-based sensors (e.g., PyzoFlex® from JOANNEUM RESEARCH). One approach to improve the sensors would be to use a piezoelectric matrix material such as poly(vinylidene fluoride) (PVDF) or a PVDF co-polymer. Dodds and co-workers showed that the voltages generated can be increased with the ZnO content (Dodds et al. 2013), in their study up to a factor of 1.7 with 10–20% ZnO NPs. In a comparative study of PVDF composites with different nanoparticles (SiO2, TiO2 and ZnO), Arjun Hari and co-workers found the most improvement of the output voltage when PVDF composites with 3% ZnO nanoparticles were tested.

Furthermore, the inner polarizability of the ZnO nanowires was investigated. For this purpose, sample “ZnO, 0 V”, in which the matrix curing took place without applying a polarization voltage and in which the nanowires were thus immobilized without a preferred direction, was subjected to hysteresis polarization. In the case of polarizability, charge shifts and, thus, currents would have to be measurable in the course of repeated repolarization. Such charge shifts and currents could not be observed, not even with a gradual increase in the voltage amplitude up to the arcing at 1500 V.

To improve the polarizability of the ZnO nanoparticles and, thus, the piezoelectric responses, it could be beneficial to dope the nanoparticles during the plasma synthesis with shallow donors (group III/VII) or shallow acceptors (group I, V) (Albertsson et al. 1989). Lithium, for example, has proven to enhance the output voltage of piezoelectric devices based on ZnO nanowires after poling (Consonni and Lord 2021; Shin et al. 2014).

However, the fact that in the case of previously immobilized nanowires breakdown only occurs at ~ 1500 V, while in the case of an acrylate matrix that has not yet been cured arcing already occurs at 600–700 V, may be seen as a further strong indication that the ZnO nanowires themselves strongly favour the formation of short circuits as they orient themselves to chains.

In conclusion, although minor piezoelectric signals are discernible, the ZnO nanowires are not yet qualitatively sufficient to fabricate usable sensors. The signals are too low to allow quantitative measurements of the piezoelectric coefficients. Therefore, future research should focus on implementing dopants like Li in the plasma flight-thru synthesis (e.g., as aerosol) in combination with piezoelectric matrices based on PVDF.

Conclusion

In the present work, we demonstrated the successful development of an atmospheric pressure plasma jet reactor for the morphology-controlled zinc oxide nanoparticle synthesis with zinc powder and oxygen. The particle morphology was found to depend on the oxygen rate in the carrier gas and the plasma, the discharge current and the energy inside the reactor, respectively. Zinc oxide nanoparticles with a high aspect ratio (nanorods and nanowires) grew, when 20% O2 was present in the carrier gas (10% in the plasma) and discharge currents of 250–350 A were applied. Piezoelectric signals could be detected upon mechanical stimulation when these nanoparticles were dispersed in a matrix of acrylic resin and fixed between finger-electrodes while poling. The signals were significantly greater than the responses assigned to electrostatic charging of the untreated reference, a ZnO-free and a non-poled sample. Hence, the signals of the poled ZnO sample prove that the synthesized ZnO nanoparticles (grown in the wurtzite structure) have the potential to animate piezoelectric sensors and devices. However, a further improvement of the particle morphology control in combination with high-yield harvesting and homogenization techniques will be necessary to advance towards scale-up for industrial purposes. To improve the polarizability of the ZnO NPs and, thus, the piezoelectric responses, on the other hand, the plasma flight-thru process should be adopted to enable the incorporation of dopants like Li into the ZnO NPs.