Enhanced piezoelectric performance of the ZnO/AlN stacked nanofilm nanogenerator grown by atomic layer deposition

Recently, zinc oxide (ZnO) nanostructures have attracted great research interest to prepare the self-powered nanogenerators (NGs) due to the excellent piezoelectric properties.^1–3 In the past several years, significant efforts have been made to fabricate piezoelectric NGs (PENGs) based on ZnO nanowires (NWs).^4,5 However, the nanowire structures still suffer from a crucial issue that the synthesis conditions of nanowires are relatively harsh and only a limited number of piezoelectric materials like ZnO and gallium nitride (GaN) can be prepared as nanowires.⁶ It is difficult to carry out universal and large-scale factory preparation. Luckily, the structure of uniform nanofilms can overcome the shortcoming mentioned above, which is confirmed to have a simple fabrication process. For instance, Qin et al. fabricated a highly efficient and large area piezoelectric ZnO thin film NG on a Si substrate by pulsed laser ablation, which exhibited an output electrical voltage of ∼95 mV and a relatively high power density of 5.1 mW cm⁻².⁷ 

However, the piezoelectric performance of a pristine ZnO film is relatively weak,⁸ which can be summed up as leakage current, short-circuit current, and screening effect in ZnO thin films. Fabrication of stacked structures is widely regarded as an effective strategy to improve the performance of NGs due to the effective weakening of the piezoelectric potential screening effect caused by free electrons in a pristine piezoelectric semiconducting material. For example, Yin et al. fabricated a NiO/ZnO heterojunction by room temperature radio frequency (RF) magnetron sputtering, whose direct current (DC) output voltage and current density were approximately 21 times higher and 13 times larger compared with the pristine ZnO NGs.⁹ Moreover, the formation of CuI/ZnO heterojunction also deposited by RF magnetron sputtering was reported by Liu et al. They demonstrated that such a stacked structure poses a significant enhancement on the piezoelectric performance.¹⁰ 

The aluminum nitride (AlN) is also an attractive candidate as an inserted dielectric layer in the ZnO-based NG due to its relatively wide bandgap (6.20 eV).¹¹ The AlN interlayer can provide a relatively high contact potential barrier which contributes to reducing the internal loss of leakage currents and then improve the energy conversion efficiency of the stacked device.¹² Moreover, AlN is also a kind of piezoelectric material,¹³ which is beneficial to a further improvement of the piezoelectric performance of the entire stacked structure of ZnO/AlN nanofilms. In addition, ZnO and AlN have similar lattice structures and thermal expansion coefficients.¹⁴ Various film growth techniques have been developed to prepare ZnO and AlN films, including chemical vapor deposition, pulsed laser deposition, molecular-beam epitaxy, magnetron sputtering, sol-gel, and so on. Compared with the growth approaches mentioned above, atomic layer deposition (ALD) has shown great advantages in precise control of nanofilm thickness at the atomic level, uniform conformal coverage ability, and low growth temperature.¹⁵ Herein, the ZnO/AlN stacked nanofilms based on ALD technique was proposed to prepare a high performance ZnO-based PENGs in this work. The microstructure, composition, and piezoelectric performance of the designed PENGs are well characterized.

The substrates used in this work were 25 × 25 mm² standardly cleaned titanium (Ti) foils whose thickness is 0.1 mm. The smooth Ti substrates were ultrasonic cleaned properly with acetone, ethanol, and DI water and then dried with nitrogen gas (N₂) thoroughly. In fact, the Ti foil acts not only the substrate for ZnO/AlN nanofilms but also the conductive electrode. Then, the preparation procedure of the ZnO/AlN stacked nanofilm structure included two steps. First, AlN films with different thickness, namely, 5, 25, 50, 75, and 100 cycles, were deposited at 360 °C on the pre-cleaned Ti substrates by the TFS-200 ALD system (Beneq, Finland). The vapor of the trimethylaluminum [Al(CH₃)₃, TMA] precursor was pulsed into the reaction chamber by the high purity N₂ as the carrier gas, and the high purity NH₃ was pulsed into the reaction chamber through a separate gas line. Meanwhile, the high purity N₂ as the purge gas purging the gaseous by-products and residual gas out of the chamber between two valid pulses is of vital importance for avoiding unexpected gas reactions. Second, ZnO films were deposited on the AlN interlayers by ALD at 200 °C. The vapor of the diethylzinc [(C₂H₅)₂Zn, DEZ] precursor and DI water was pulsed into the reaction chamber by the high purity N₂, respectively. Finally, the Cr/Au film as the top electrode was deposited using physical vapor deposition (PVD) technique by PVD 75 (Kurt J. Lesker Company, USA) at the power of 80 W. In particular, each side of the samples needs to be pasted with tape to prevent leakage.

The thicknesses of the AlN interlayer and ZnO layer were measured by a SOPRA GES-5E spectroscopic ellipsometry (SE) system (Semilab, Hungary). The X-ray diffraction (XRD) experiments were conducted using a Bruker-D8 Advance powder X-ray diffractometer (Bruker, Germany) with Cu Kα radiation (wavelength λ = 1.5406 Å) working at 40 kV and 40 mA. Scanning electron microscopy (SEM) and atomic force microscope (AFM) were performed to identify the morphology of the ZnO/AlN stacked nanofilm samples. The cross-sectional-view SEM images were recorded on a SIGAMA HD field-emission SEM (Zeiss, Germany). The three-dimensional (3D) AFM images were measured by the Dimension Icon AFM (Bruker, Germany). The depth-profiling characterization of the ZnO/AlN/Ti samples was obtained by applying the glow discharge optical emission spectroscopy (GDOES) technique, using GD-Profiler 2 (HORIBA Scientific, France).

The piezoelectric properties of the PENG were studied by measuring the magnitude of the piezoelectric open circuit (OC) voltage as well as the piezoelectric output current across a 2 MΩ load resistance under a constantly applied compressive force. Figure 1(a) shows the schematic diagram of the self-made equipment which provides the compressive force. This equipment is composed of three major components: a 60-cm-long wood block with a fixed pulley as the base of the equipment, a stepper motor controlled by a DC power supply, and an adjustable weight connected with the motor by a string. When the stepper motor is running at a certain frequency under the DC power supply, the weight at the end of the string is driven to move up and down at the same frequency, providing compressive force to the ZnO/AlN stacked PENG. Besides, when measuring the electrical signals, the ZnO/AlN stacked PENG was connected to a Model SR560 low-noise preamplifier (Stanford Research Systems, USA) for noise reduction. The final results of the piezoelectric output signal were measured by a MDO3022 high-precision oscilloscope (Tektronix, USA). At this point, the establishment of the whole measurement system was completed. Here is the specific experimental scheme. For each individual PENG sample, set different compressive forces like 0.05, 0.10, 0.20, and 0.50 kgf and set different tapping frequencies like 0.67, 1.00, and 1.33 Hz for measurements. Particularly, the change of the tapping frequency can be achieved by adjusting the output voltage of the DC power supply. Finally, it is worth noting that all the measurements were conducted under the pressure of 1.01 × 10⁵ Pa.

Figure 1(b) illustrates the stack structure of ZnO/AlN nanofilm based PENGs. First, the AlN films with different thickness were deposited on the cleaned Ti substrates by ALD technique at 360 °C. Light yellow could be seen from the sample surface after deposition. The sample thickness of the AlN films was to be 1.9, 2.3, 3.6, 7.8, and 11.4 nm corresponding to different ALD cycle numbers of 5, 25, 50, 75, and 100. The average growth rate of the AlN layer could be estimated to be ∼0.1 nm per cycle. Subsequently, a layer of ZnO film with the thickness of 220 nm was deposited on the AlN films by another ALD reaction at 200 °C. After this step of deposition was completed, a uniform and bluish violet thin film was formed. Then, the Cr/Au film as the top electrode was obtained on the ZnO/AlN stacked structure through PVD technique. Finally, two terminal copper leads were glued on the two electrodes with silver paste for electrical measurements, respectively. At this point, the entire PENG was completed. The corresponding samples are named as Z/AX in this paper, where X = 0, 1.9, 2.3, 3.6, 7.8, and 11.4, representing the thickness of the AlN interlayer. In order to ensure the stability of the prepared PENGs as well as protect the top and the bottom electrodes, we employed two sheets of Kapton board to cover the whole PENG. The optical image of the as-made PENG is shown in Fig. 1(c). It is obvious that the prepared PENG can be easily bent by the human fingers, indicating it has a good flexibility.

Figure 2(a) shows the XRD patterns of the prepared samples with AlN interlayers with different thicknesses. Except for the peaks of Au electrodes and Ti substrates, all the samples show four characteristic diffraction peaks at 31.8°, 34.5°, 36.4°, and 56.7°, which is corresponding to the (100), (002), (101), and (110) planes of the hexagonal wurtzite structure of ZnO (JCPDS No. 75-0576), respectively. It suggests that the obtained ZnO films have a hexagonal wurtzite structure. With the increasing thickness of the AlN interlayer, the position of the ZnO (100) peak shifts to small angle. It implies that the lattice distortion exists in the ZnO films despite the lattice mismatch between the ZnO and AlN layers is relatively small. The thicker the AlN films, the higher the lattice distortion rate. However, no distinct diffraction peak of AlN was detected in the XRD pattern of the stacked samples, which is mainly due to the ultrathin thickness of the AlN layers obtained after no more than 100 cycles of the ALD-AlN reaction.¹⁵ In addition, no other impurity peak can be found in the XRD patterns. Moreover, the well-defined stacked nanofilm structure is further revealed by the cross-sectional-view SEM characterization result obtained on a typical sample Z/A11.4 on a Si substrate [Fig. 2(a), inset]. Additionally, the thickness of the ZnO and AlN films is estimated to be 220.0 and 11.4 nm from the SEM image, respectively. Figure S1 of the supplementary material displays the 3D AFM images of the sample Z/A2.3 on a Ti substrate, whose surface is not completely flat caused by the relatively rough Ti substrate. The AFM and SEM images together show a textured structure of the ZnO film, which will exhibit a better piezoelectric performance compared with the thoroughly flat film because of the larger mechanical deformation.¹⁶ 

To obtain the depth profile of ZnO/AlN/Ti stacked samples, the GDOES measurements were conducted, as shown in Fig. 2(b). It can be seen that the top layer of the prepared sample is a material consisted of Zn and O elements, that is, ZnO, and the interlayer is a material consisted of Al and N elements, that is, AlN. And the ZnO/AlN interface is marked by an increase in the intensities of Al and N depth profile curves and a decrease in the intensities of Zn and O depth profile curves. Similarly, the AlN/Ti interface is marked by an increase in the intensity of Ti depth profile curves and a decrease in the intensities of all the other chemical elements. Moreover, our previous work has conducted the X-ray photoelectron spectroscopy (XPS) measurement of the AlN film deposited on a Si substrate, whose preparation process is completely the same as this work, confirming the real existence of Al–N bonds.¹⁷ 

Figure 3 show the open circuit (OC) piezoelectric output voltage of the ZnO/AlN stacked nanofilm PENGs with AlN interlayers at room temperature under the same mechanical deformation, namely, using the same compressive force of 0.20 kgf and the same tapping frequency of 1.00 Hz. It can be seen that the samples Z/A0, Z/A1.9, Z/A2.3, Z/A3.6, Z/A7.8, and Z/A11.4 show a piezoelectric OC voltage of about 1.3, 2.4, 4.0, 3.2, 2.4, and 2.4 V, respectively. Obviously, the value of the OC voltage is completely dependent on the existence of the AlN interlayer. Compared with a pristine ZnO PENG (i.e., sample Z/A0), the OC voltage of the ZnO/AlN nanofilm stacked PENG increases at least 84.6% (i.e., sample Z/A1.9), achieving a really effective improvement. This phenomenon is in good agreement with other works about the ZnO-based PENGs inserted with dielectric layers.¹⁸ 

The detailed piezoelectric mechanism of the stacked structure is discussed here. As shown in Fig. 4(a), the work function of Au is 5.10 eV¹⁹ and the electron affinity of ZnO is approximately 4.35 eV,²⁰ thus forming a Schottky contact between the Au film and the n-type ZnO film in this work. However, the work function of Ti is 4.33 eV,¹⁹ which is smaller than the electron affinity of ZnO, leading to an ohmic contact between them.²¹ The introduction of an AlN interlayer leads to a change of the contact state between ZnO and Ti substrate owing to a small electron affinity (∼1 eV) and a relatively high bandgap (6.20 eV) of AlN.¹¹ As a result, the leakage or short circuit current caused by the ZnO internal structure can be effectively reduced to improve the piezoelectric OC voltage significantly. Besides, AlN is also a kind of piezoelectric material¹³ and will further improve the piezoelectric performance of the entire stacked structure. The schematic energy band diagram for the Au/ZnO/AlN/Ti stacked nanofilm structure under stress-free condition is shown in Fig. 4(b). When an external force is applied to the PENG, a corresponding piezoelectric potential will be generated in the piezoelectric material. As can be seen from Fig. 4(c), when the stacked structure is applied with a compressive force, the negative polarization charges generated from the ZnO internal structure will accumulate in the inner interface region of ZnO/Au. On the other hand, the positive polarization charges will accumulate in the inner interface region of ZnO/AlN. A piezoelectric potential is then established and electrons will flow from the Ti electrode through the external circuit to the Au electrode, forming a positive current. Thanks to the existence of the AlN interlayer, the high barrier will block the leakage of the electrons effectively. When the compressive force is released, the originally accumulated electrons will flow through the external circuit back to the Ti electrode with a sudden drop of the internal piezoelectric potential. Such a process is faster than the charge accumulation process. Similarly, if a tensile force is applied to the stacked structure, the positive and negative polarization charges will, respectively, accumulate in the inner interface of ZnO/Au and the inner interface of ZnO/AlN, as shown in Fig. 4(d). As a result, the electrons will flow through the external circuit in the opposite direction. To further verify the piezoelectricity mechanism discussed above, we subsequently conducted a switching-polarity test. Typically, when the sample Z/A2.3 was connected reversely with the oscilloscope, the pulses were also reversed, as shown in Fig. S2.

Additionally, the piezoelectric OC voltage is closely related to the thickness of the AlN interlayer, which can be seen more clearly in Fig. S3. Specifically, the OC voltage of the stacked samples under a constantly applied compressive force of 0.20 kgf at 1.00 Hz kept increasing as the thickness of AlN increased from 1.9 nm of Z/A1.9 to 2.3 nm of Z/A2.3. The sample Z/A2.3 yielded a maximum OC voltage of 4.0 V, which is around 3 times higher than the value of the sample Z/A0. However, the OC voltage decreased as the thickness of AlN further increased above 2.3 nm (i.e., sample Z/A3.6, Z/A7.8, and Z/A11.4). The variation law of the AlN thickness and the piezoelectric OC voltage mentioned above may be caused by the following two aspects. First, when the thickness of the AlN dielectric layer increases, the resistance also increases, which helps to weaken the screening effect and reduce the leakage current. Therefore, the piezoelectric performance of the stacked devices will be improved to some extent. On the other hand, with the thickness of the AlN dielectric layer gradually increasing, the increase in the movable charges and trapped charges in the dielectric layer will weaken the piezoelectric performance of the stacked devices. And it is obvious that if the whole stacked device is thicker, the loss in the whole energy conversion process will be higher. The yield of mechanical energy transfer (η_M) and the yield of electrical energy transfer (η_E) can be calculated as¹² 

where d₁, d₂, E₁, E₂, and ε₁, ε₂ are the thickness, Young’s modulus, and dielectric constant of the insulating and piezoelectric layers, respectively. Since Young’s modulus ratios and dielectric constant ratios have already been determined by the selected two materials, when the thickness of the AlN interlayer d₁ is smaller as the thickness of ZnO d₂ is fixed, the energy transfer efficiency η_M and η_E will be larger, leading to the better piezoelectric performance of the stacked structure. Taking into account these two aspects, the amplitude of the piezoelectric OC voltage is supposed to have a maximum point corresponding to a certain thickness of the AlN dielectric layer.

It is well known that only piezoelectric OC voltage is not enough to characterize the piezoelectric performance of a PENG. Therefore, the piezoelectric output current of the stacked PENGs were measured as well.^22,23 The output power measurement of the stacked PENG Z/A2.3 was performed across different load resistance from 0.5 MΩ to 5 MΩ as the pre-experiment (Fig. S4). The results demonstrated that when across a 2 MΩ resistor, the sample Z/A2.3 exhibited the largest piezoelectric output power. So in the follow-up experiments, the piezoelectric output current of the stacked PENGs with AlN interlayers was measured across a 2 MΩ resistor under the same constantly applied force (i.e., 0.20 kgf, 1.00 Hz). As shown in Fig. 5(a) and more clearly in Fig. S5, the corresponding output current of PENGs Z/A0, Z/A1.9, Z/A2.3, Z/A3.6, Z/A7.8, and Z/A11.4 is 0.54, 0.68, 1.10, 0.95, 1.05, and 0.96 µA, respectively. It can be easily found that the stacked PENG Z/A2.3 achieves the largest piezoelectric output current of 1.10 µA and the largest output power of 2.42 µW with a load resistance of 2 MΩ. Additionally, the enlarged plot of the piezoelectric output current of the Z/A2.3 PENG across a 2 MΩ load resistance is shown in Fig. S6.

To figure out the most optimized external working conditions of the ZnO/AlN stacked nanofilm PENGs, the effects of the value and the frequency of the compressive force on the piezoelectric output current of the stacked PENGs across a 2 MΩ load resistance were investigated as well. According to the experimental results mentioned above, the sample Z/A2.3 with the best piezoelectric performance was selected. Figure 5(b) shows the piezoelectric output current of the sample under different compressive forces with a tapping frequency of 1.00 Hz. As the compressive force is 0.05, 0.10, 0.20, and 0.50 kgf, the piezoelectric output current of the device is 0.82, 1.10, 1.10, and 0.92 µA. The corresponding piezoelectric output power is 1.34, 2.42, 2.42, and 1.69 µW. It is clear the piezoelectric output power of the sample reaches the maximum when the compressive force is 0.10 or 0.20 kgf. Such a result may be caused by the following two aspects. First, the amount of mechanical deformation increases with the increase in the compressive force, thus improving the piezoelectric output power of the sample device. Second, the stability of the applied compressive forces are limited by our measurement equipment, that is, if the compressive force is too large, the stability of the equipment will decrease in some extent, affecting the measurement results of the piezoelectric output signals. The effects of these two aspects are completely opposite, leading to an optimal piezoelectric output power between the compressive force of 0.10 and 0.20 kgf. However, due to the restriction of the given weight, it is unable to apply continuously variable force. Anyway, our results demonstrate that such stacked PENGs are suitable for the case when the compressive force is not very large, for instance biomedical devices.

Besides, the effect of the frequency of the compressive force was also investigated with the same stacked PENG (i.e., sample Z/A2.3). Remarkably, we set the compressive force at 0.20 kgf determined according to the previous results. The measurement results under different tapping frequencies from 0.67 Hz to 1.33 Hz are shown in Fig. 5(c). As the tapping frequency is 0.67, 1.00, and 1.33 Hz, the piezoelectric output current is 1.08, 1.10, and 1.12 µA, and the corresponding piezoelectric output power is 2.33, 2.42, and 2.51 µW, respectively. As it can be seen, when under relatively low tapping frequencies (<10 Hz), the piezoelectric output power of the sample device increases with the increase in the frequency. This is mainly caused by the increase in the impulse of the weight as well as the amount of mechanical deformation.

In summary, highly efficient and stable PENGs based on the ZnO/AlN stacked nanofilm structure with precisely controlled AlN thickness prepared by ALD technique have been fabricated in this work. The stacked PENG with an AlN interlayer of 2.3 nm achieved the maximum piezoelectric OC voltage of 4.0 V and an output power of 2.42 µW across a 2 MΩ load resistance under the constantly applied compressive force of 0.20 kgf at 1.00 Hz. In addition, the studies on the external working factors indicate that the as-made PENGs will be suitable to be used in some micro-electromechanical systems, such as wearable devices and biomedical applications. Based on these results, the findings are expected to stimulate a new feasible direction for the development of highly energy-efficient ZnO-based NGs as well as further expand the application scope for nanosystems.

See supplementary material for the additional figures and data.

This work is supported by the National Key R&D Program of China (No. 2016YFE0110700), the National Natural Science Foundation of China (Nos. U1632121 and 61376008), and the Natural Science Foundation of Shanghai (No. 18ZR1405000).