Facile growth of aluminum oxide thin film by chemical liquid deposition and its application in devices

Uniform and continuous Al2O3 thin films were prepared by the chemical liquid deposition (CLD) method. The breakdown field strength of the amorphous CLD-Al2O3 film is 1.74MV/cm, making it could be used as a candidate dielectric film for electronic devices. It was further proposed to use the CLD-Al2O3 film as an electron blocking layer in a triboelectric nanogenerator (TENG) for output performances enhancement. Output voltages and currents of about 200V and 9μA were obtained, respectively, which were 2.6 times and 3 times, respectively, higher than TENG device without an Al2O3. A colloidal condensation-based procedure controlled by adjusting the pH value of the solution was proposed to be the mechanism of CLD, which was confirmed by the Tyndall effect observed in the growth liquid. The results indicated that the CLD could serve as a low-cost, room temperature, nontoxic and facile new method for the growth of functional thin films for semiconductor device


Introduction
Al 2 O 3 is a functional material with good electrical insulation, large forbidden band width (E g ≈ 8.8 eV), high transparency, high refractive index, large mechanical strength, and good thermal conductivity [1][2][3][4][5][6][7][8]. It has stable chemical properties at high temperature. Al 2 O 3 has a dielectric constant of about 8-10, which is more than twice as high as SiO 2 . Al 2 O 3 is also superior to SiO 2 and silicon nitride in its ability to resist radiation and to block impurities such as sodium ions [9]. It is more resistant to various acids and base corrosion compared to SiO 2 . Therefore, Al 2 O 3 could be used as an excellent dielectric and passivation layer in semiconductor electronic devices.
Currently, standard Al 2 O 3 film preparation processes primarily comprise electron beam evaporation [10], magnetron sputtering [11], atomic layer deposition (ALD) [12], and plasma-enhanced chemical vapor deposition (PECVD) [13]. Even though Al 2 O 3 films prepared by these methods are typically of high quality, these methods are complicated in process and generally require growth of the film under high temperature and vacuum conditions relying on very expensive equipment.
In this article, we introduce a simple and low-cost method that uses nontoxic liquid precursors to grow the Al 2 O 3 film. It is named as chemical liquid deposition (CLD), in parallel with the commonly known chemical vapor deposition method. The thin film growth setup was simple, inexpensive, and easy to operate. Al 2 O 3 thin films were prepared on the surface of Si substrates and indium tin oxide (ITO)-coated glasses (i.e., ITO glass) under ambient pressure and room temperature conditions. The deposition had the characteristics of uniform film growth on the substrate. The surface morphology, structural composition, electrical properties, and passivation properties of Al 2 O 3 films prepared by the CLD method (CLD-Al 2 O 3 ) were characterized by comprehensive material analysis. Triboelectric nanogenerators (TENGs) are a new type of simple and low-cost energy harvesting devices that could effectively convert various daily wasted mechanical energy into electric energy by utilizing triboelectric charging and electrostatic induction effects [14][15][16][17]. As an example of the device application, we used CLD-Al 2 O 3 as an electron-blocking layer in TENG, which effectively prevented charge leakage during operation and enhanced the output performance of TENG. Experiments showed that the output performance of the TENG with the Al 2 O 3 blocking layer was significantly improved compared to the conventional TENG structure. In addition, the electron blocking layer located at the bottom of the friction layer provided a long-term performance enhancement for TENG. Finally, through the observation of the Tyndall effect, we had revealed that the growth mechanism of the CLD method was essentially a colloidal condensation process caused by adjusting the pH value. Note that this mechanism should not be confused with the traditional sol-gel method. Our results are important for the theoretical interpretation of CLD and could be used as the guideline for the design of CLD of other functional materials.

Preparation of Al 2 O 3 thin films by CLD
The chemical reagents used in this experiment were purchased from Sinopharm Chemical Reagent Co., Ltd. Figure 1a shows a schematic process for the Al 2 O 3 -CLD, and the specific preparation procedures are as follows: (1) Al 2 (SO 4 ) 3 ·18H 2 O (50 g) was dissolved in 50 mL of deionized (DI) water at room temperature. The solution was filtered (maximum pore size 15 µm) to remove few undissolved large particles and to obtain 73.25 mL colorless viscous liquid, whose nominal pH is about −0.26. (2) The prepared Al 2 (SO 4 ) 3 ·18H 2 O solution was poured into a beaker and stirred using a magnetic stirrer, and 16.115 g of NaHCO 3 powder was slowly added into the solution. Each time when the NaHCO 3 was added, it was necessary to wait for the CO 2 foam to substantially disappear before continuing to add more. A small amount of residual NaHCO 3 pellets were treated with an ultrasonic device to dissolve them thoroughly to obtain 74.72 mL of a colorless transparent viscous liquid, whose pH was about 2.89. This liquid was prone to produce aluminumsodium vanadium, so it should not be placed for a long time.
(3) The as-prepared liquid was immediately diluted with water by a volume ratio of 1:1, and the suspended residual matter in the liquid was filtered out, thereby obtaining 149.43 mL of colorless transparent liquid. Now the pH value is about 3.35. (4) This liquid was further diluted with water with a volume ratio of 1:5 to obtain 896.58 mL of the growth liquid, whose pH was about 3.74. The Si substrates or ITO glasses were ultrasonically washed in acetone, absolute ethanol, and DI water for 10 min each and then dried. After the cleaning was completed, the surface was treated with oxygen plasma for 15 min to obtain some degree of hydrophilicity, so that a large amount of surface hydroxyl groups (OH groups) were decorated on the surface, which were advantageous for enhancing the chemical adsorption of the film material to the substrates. The samples were then placed on the surface of a Teflon stent in a Teflon beaker and immersed in the growth liquid, as shown in Figure  1b. The liquid was agitated with a magnetic stirrer at 200 rpm. After few hours at room temperature, the growth liquid eventually turned little turbid. The samples were taken out, cleaned with DI water, and dried in an oven at 60°C. (5) Finally, the as-prepared samples were thermally annealed optional at 400°C for 1 h using a muffle furnace (heating rate 1.5°C/min) and were taken out after natural cooling.

Fabrication process of the TENG device with Al 2 O 3 electron blocking layer
The TENG device consisted of a polydimethylsiloxane (PDMS)/Al 2 O 3 /ITO complex film and an Al metal layer operating in a vertical contact-separation mode [18]. PDMS (Sylgard 184, Dow Corning) base and a curing agent were mixed at a mass ratio of 10:1 and were spin coated at 3,000 rpm for 30 s onto the Al 2 O 3 -coated ITO glass. The PDMS was then cured at 150°C for 10 min. PDMS and Al act as negative and positive friction layers, respectively. The surfaces (contact area: 2 × 2 cm 2 ) of the PDMS and Al film were periodically contacted and separated to generate a potential difference.

Measurement and characterization
The chemical compositions of the samples were analyzed by energy-dispersive spectroscopy (EDS, AMETEK) and X-ray photo-electron spectroscopy (XPS, Thermo Scientific, ESCALAB 250). The morphology of the Al 2 O 3 films was characterized by ultra-high resolution field emission scanning electron microscopy (SEM, Thermo Scientific, Verios G4 UC) and atomic force microscopy (AFM, Bruker, Multimode 8). The crystal structure analysis of the Al 2 O 3 films was performed using X-ray diffraction (XRD, Empyrean, DY1602). The thickness and the refractive index of the Al 2 O 3 films were measured using ellipsometry (J.A. Woollam, M2000). The breakdown voltage of the film was measured by the semiconductor characterization system (Keithley, 4200). A signal generator (RIGOL, DG4162), a power amplifier (SINOCERA, YE5872A), and a vibration exciter (SINOCERA, JKZ-10) were combined to form a controllable tapping system to drive the TENG to work with the same external force and frequency. An oscilloscope (RIGOL, DS1102E) was used to record the voltage signal of the TENG output. The output current and the transferred charge were measured by using an electrometer (Keithley, 6514).

Results and discussion
The double hydrolysis reaction of Al 2 (SO 4 ) 3 and NaHCO 3 in the experiment is given as follows: where Al(OH) 3 is actually xAl 2 O 3 ·yH 2 O and ΔH is the chemical reaction enthalpy (ΔH > 0). Al(OH) 3 is a typical amphoteric hydroxide, which is soluble not only in strong acids but also in strong bases to form hydroxyl complexes. The solubility of Al(OH) 3 has a certain relationship with the pH value. Al(OH) 3 can be dissolved in the acidic environment with pH < 3.4 or an alkaline environment with pH > 12.9, and it does not substantially dissolve in the pH rang of 4-11.
In this experiment, a very high concentration of Al 2 (SO 4 ) 3 solution was first prepared, and the solution exhibited strong acidity due to the hydrolysis of Al 2 (SO 4 ) 3 . After slowly adding an appropriate amount of NaHCO 3 powder, the Al(OH) 3 nominally formed by the double hydrolysis reaction will then all "dissolve" in the still strongly acidic liquid. However, Na 2 SO 4 , one of the reaction products, could form a double-salt aluminumsodium vanadium (NaAl(SO 4 ) 2 ·12H 2 O) with Al 2 (SO 4 ) 3 . As aluminum-sodium vanadium could easily crystallize, it was necessary to dilute the liquid with water rapidly after the reaction of Al 2 (SO 4 ) 3 with NaHCO 3 was completed. Then, the added water (volume ratio 1:1) had adjusted the pH to 3.35, very close to the critical value. This liquid was chemically relatively stable and could be stored if necessary (not more than a few weeks). Just before the film growth, DI water with five times the volume of the liquid was poured into the liquid to adjust the pH to 3.74, which was greater than the critical value of 3.4. At this point, the Al(OH) 3 in the liquid gradually precipitated onto the sample to initiate the nucleation and the subsequent film formation. The surface was therefore deposited with a layer of Al(OH) 3 film, which was then annealed to eliminate the possibly existing H element, although this step was usually optional, since the amount of H is negligible. respectively. This figure also shows that the surface morphology of the Al 2 O 3 film is continuous, relatively flat, and uniform, with a few bumps occasionally. Generally, an insulating layer with the smooth surface is favorable for forming good interface contact with electron devices, which can improve the electrical performance and the stability of the devices [19].

Characterization of Al
Because the Al 2 O 3 film was very thin, we did not detect any signal in the small incident angle XRD measurement [20]. Therefore, we performed XRD tests on the precipitate collected in the growth liquid instead, because the precipitate had exactly the same composition and structure as the film. The precipitate was filtered out from the growth liquid that had grown for 12 h, washed, dried, and annealed in a muffle furnace at 1,000°C for 1 h. The XRD curve was compared with the JCPDS card (46-1212 [21]) and confirmed as α-Al 2 O 3 , as shown in Figure 3b. We also characterized the powder annealed at 500°C but did not obtain conclusive results indicating a crystallization. Therefore, increasing the annealing temperature can indeed crystallize this material. However, for most applications in electronic devices, this crystallization is unnecessary. In general, films with an amorphous structure are isotropic, and their insulation and other properties are usually more uniform than the polycrystalline counterpart. In addition, the amorphous structure usually makes the surface of the dielectric layer flatter. If the film has a polycrystalline structure, the grain boundaries will become channels for impurity scattering and leakage current. The anisotropy will reduce the dielectric stability of the film. Furthermore, the grain boundary will increase the surface roughness of the film, increase the density of defect states at the interface with which the semiconductor layer is in contact, and directly affect the application of the dielectric film.
Several ITO glass substrates were simultaneously immersed in the growth liquid, and one sample was taken out every half an hour. After annealing at high temperature, the thickness and the refractive index of the grown Al 2 O 3 films were measured by ellipsometry. The relation between the thickness and the growth time is shown in Figure 3c. It can be seen from the figure that the deposition rate is faster in the range of 0-0.5 h, which is about 32 nm/h. As the growth time increases, the growth rate of the film decreases, and it remains nearly constant at about 10 nm/h at 0.5-1.5 h. When the growth time exceeds 1.5 h, the film thickness remains almost unchanged, indicating that the liquid has already substantially lost its growth ability. Figures 4a and b show the EDS spectrum of the Al 2 O 3 film grown on the ITO surface and Si substrate (with a scan range of 0 to 5 keV, 400°C annealing), respectively. This figure also shows that there are clear O and Al signals at 0.53 and 1.49 keV, respectively. Apart from the In peak, there are no additional peaks detected, which indicates that the chemical composition of the CLD-Al 2 O 3 film method is very pure. According to the peak values in Figure 4a, it can be quasi quantitatively inferred that the atomic ratio of the O element to the Al element is about 7:1, which deviates from the expected 3:2 ratio. The main reason is that the thickness of the Al 2 O 3 film is small. A large amount of O element contained in the thick ITO film on the base surface is inevitably detected, so that the content of the O element appears significantly larger. The chemical composition of the CLD-Al 2 O 3 film on the Si substrate (400°C annealing) was investigated by XPS measurements as well. In Figure 4c, the XPS spectrum shows a singlet Al 2p peak at 74.4 eV, which is consistent with the peak position of the oxidation state Al in the literature [22]. The XPS O 1s peak of the CLD-Al 2 O 3 film is shown in Figure 4d. The black solid line shows the measured data, and the green dotted line represents the Gaussian peak fitting result. The O 1s peak in spectrum is deconvoluted into two sub-peaks corresponding to O 2− bonded with Al 3+ ions in the Al 2 O 3 film and oxygen associated with the OH − in aluminum hydroxide, whose centers are located at around 531.05 and 532.35 eV, respectively [23]. It is worth noting that the associated OH groups are mainly due to water absorbed from the environment in the Al 2 O 3 film [22].

Output performance of the TENG with CLD-Al 2 O 3 layer
We applied the CLD-Al 2 O 3 thin film in TENG as an electron blocking layer to enhance the output performance. At the same time, we also fabricated TENG without Al 2 O 3 as a reference sample. The fabrication process is described in detail in the experimental section, and Figure   separation cycles, the friction generated electrons are accumulated on the surface of the PDMS friction layer. At the same time, the bottom electrode will be positively charged due to electrostatic induction, so an electric field will be generated between the contact surface and the bottom electrode (the direction of the field is upward), as shown in Figure 6a. At this moment, there are two electron transfer procedures: a drift process  caused by the electric field and a diffusion process caused by the concentration gradient of electrons [24]. Since the electric field intensity generated by the electrostatic induction is large, the electron diffusion transfer is comparably negligible. Therefore, in Figure 6b, only the drift motion is considered. Although the number of actually drifted electrons are not large due to the fact that the PDMS is an insulator, some electrons do have high enough energy to climb over the lowest unoccupied molecular orbital (LUMO) level of the PDMS (equivalent to the conduction band bottom of an insulator). These hot electrons eventually reach the bottom of the PDMS and escape from the bottom electrode, leading to the attenuation of the triboelectric charge stored in the friction layer. This reduces the output performance of TENG. However, the band gap of Al 2 O 3 is wider than the highest occupied molecular orbital (HOMO)-LUMO gap of PDMS [25][26][27]. Therefore, after the introduction of Al 2 O 3 , it blocks the hot electrons, suppressing the charge transport. That is, the Al 2 O 3 can effectively prevent the electrons from escaping at the bottom of PDMS, thereby improving the triboelectric charge storage capacity (see Figures 6c  and d). We have measured the breakdown field of the CLD-Al 2 O 3 thin film to be about 1.74 MV/cm, as shown in Figure 3d. In this measurement, a 23 nm Al 2 O 3 film was sandwiched between two parallel electrodes. This breakdown field value is in the same order as that of the standard Al 2 O 3 [28]. Today, in TENG community, to enhance the device output performance, most research has focused on the surface optimization of the surface friction layer [29][30][31][32]. However, these approaches yield little contributions toward the performance gain in longterm operation when surface modification/engineering is inevitably physically worn out by intimate contact and friction between the triboelectric materials [33]. Different from that strategy, the Al 2 O 3 electron blocking layer here is located far away from the friction layer surface. Therefore, the resulted enhancement would not diminish even after the friction layer surface was worn during long-term operation. In addition, instead of competing with other methods, our technique could actually complement them. It is generic and could be applied in combination with other methods.

Growth mechanism of the CLD-Al 2 O 3
Comparing with other more conventional Al 2 O 3 deposition methods, the CLD technique may achieve the similar material quality while using cheaper and more convenient setup that does not rely on vacuum systems. Especially, the deposition temperature of CLD is much lower than other methods (see Table 1). Therefore, unless the substrate is sensitive to the aqueous environment, CLD could be used as a facile method to grow aluminum oxide thin films on electron devices. Nevertheless, an interesting question remains to be solved. That is, the deposition mechanism is yet to be unveiled. We propose the CLD-Al 2 O 3 thin film growth mechanism as follows. First, owing to the abundance of OH groups in the growth liquid, OH are decorated onto the surface of substrate. Then, the OH groups and the Al(OH) 3 clusters undergo a continuous dehydration and polymerization, which result in the CLD-Al 2 O 3 particle and subsequently thin film deposition. However, what is the status of the precursor? Are the Al(OH) 3 in ionic form, i.e., Al 3+ and OH − , like in a solution, or are they already forming colloids? To clarify this, we have let a laser beam pass through a 1:1 diluted liquid (see step 3 in Section 2.1 and Figure 1a). In a dark room, observing from the side, a bright "light path" of the incident light travelling in the liquid can be vividly seen, i.e., the Tyndall effect, as shown in Figure 3e. Therefore, some Al (OH) 3 colloids are already formed in the liquid in step 3 (see Section 2.1) before the deposition. Nevertheless, we note that our CLD method is fundamentally different from the existing sol-gel method and should not be mixed up. CLD has the advantage that a solid film can be directly grown on the surface of the sample in the growth liquid, and postannealing is not mandatory in the process flow. If needed, however, postannealing could be used to increase the crystallinity of the film. On the contrary, the sol-gel method usually requires a plurality of heat treatments in the process of preparing the sol. After immersing the substrate in the sol, no film is formed in situ. Taking the sample out of the sol, a solid film is produced only after the high-temperature annealing, which is not an optional procedure. Therefore, the CLD method is intrinsically a room temperature process that accomplishes thin film deposition already in the growth liquid, which is not possible using the sol-gel method. The growth is in essence a pHcontrolled condensation process of colloidal nanoparticles. No matter we put in a substrate or not, the condensation will anyway take place and that is why the liquid gradually becomes turbid. When a substrate is immersed in the growth liquid, the nanoparticles will nucleate on its surface and growth into a thin film. The particles are so fine that the film is rather flat and pin hole free. Of course, it is suitable for conformal coating of ultrathin films. Eventually, when the nanoparticles are exhausted, the liquid losses its ability to grow films. Actually, the growth already essentially stops when the colloidal particles become too large to be adsorbed. Note that our experiments do not exclude the growth from single ions. Although the liquid is proved to contain Al (OH) 3 sol, since the chemical reaction (see equation (1)) is dynamically taking place, it does not exclude the possibility that single Al(OH) 3 particles coming from the hydrolysis of single Al 3+ ions can be directly incorporated into the thin film (without forming nanoparticles). Therefore, the CLD of Al 2 O 3 is a complicated procedure involving Al(OH) 3 formation at both nanometer and single-ion level. With the growth mechanism revealed, one could specifically select chemical reactions that generate colloids, which can be condensed in a controlled fashion to grow the needed thin films in a facile manner. It opens the gateway to the synthesis of a bunch of functional films.

Conclusions
In summary, Al 2 O 3 thin films were deposited on substrates by using the CLD method, which is a facile and low-cost method. The growth conditions are carefully optimized. The CLD method can grow large area thin films at room temperature with a reasonably good material quality suitable for electronic applications. The results show that the surface of the Al 2 O 3 films is very flat. The breakdown field strength can reach 1.74 MV/cm. We have applied the as-grown Al 2 O 3 films as the electron blocking layer of TENG, which provides a simple, lowcost, and effective method for improving the output performance of TENG. The open-circuit voltage and short-circuit current of the Al 2 O 3 -TENG is about 2.6 and 3 times, respectively, that of the Al 2 O 3 -free TENG. Finally, the growth mechanism is confirmed to include a pHcontrolled colloidal condensation procedure, which could be used as the guideline to design the CLD of other thin films.