1. Introduction
Al electrolytic capacitors are indispensable, essential, and key components in electric and electronic devices including computers, mobile phones, and automobiles because they can provide filtering, energy storing and transformation functions, rectifying, blocking, and coupling due their small size and self-healing property, which are associated with high specific capacitance. The aluminum electrolytic capacitor primarily consists of anode and cathode foil electrodes separated by a thin barrier-type aluminum anodic oxide film (Al2O3) and acting as dielectric layers on anode Al foil for accumulating electrical charges. The capacitance and property of an anodic oxide film determines the size and performance of capacitors. The developing trend of the miniaturization and integration of electronic and electron products necessitates increasing the specific capacitance of capacitors.
The specific capacitance of capacitors may be expressed with basic Formula (1)
    
where 
C is the capacitance, 
εr the relative dielectric constant of dielectric layer, 
εo the vacuum permittivity, 
S the surface area of the dielectric (electrode), and 
d the thickness of the dielectric layers [
1]. The 
d value depends on the formation potential (
Ea) and formation constant (
K). Equation (1) shows that increasing the 
C value can be achieved by increases in 
S, decreases in 
d, and increases in ε
r. Increases in S may usually be realized by the D.C or A.C electrochemical etching of Al foil before anodization to prepare high-density holes and tunnels on the Al foil, but now, the development of these methods has come to a standstill and a limit has been encountered. It is also very difficult to decrease 
d because 
d is related to the working voltage (
Ea) of the aluminum electrolytic capacitor. While 
Ea is decided, the 
d value is largely decided because 
K is almost a fixed value when other anodizing conditions remain unchanged. Increases in 
εr may be possible by incorporating relatively large 
εr value compounds into the aluminum anodic oxide film to form composite dielectric layers. The 
εr value of Al
2O
3 is about 8~10, below other valve metal oxides, such as Ta
2O
5, ZrO
2 [
1,
2,
3], Nb
2O
5 [
4], and TiO
2 [
5,
6,
7] and far less than the ferroelectric materials such as BaTiO
3 [
8,
9], Bi
4Ti
3O
12 [
10], and (Ba
0.5Sr
0.5)TiO
3 [
11]. The 
εr values of these oxides are summarized in 
Table 1. Introducing these high-
εr-value oxides into a single Al
2O
3 film is becoming a prospective and effective way to enhance the specific capacitance. 
By virtue of electrodeposition, the sol–gel method, the pore-filling method, and hydrolysis/precipitation, many groups have successfully prepared Ta
2O
5-Al
2O
3, Nb
2O
5-Al
2O
3, TiO
2-Al
2O
3, ZrO
2-Al
2O
3, BaTiO
3-Al
2O
3, and (Ba
0.5Sr
0.5) TiO
3-Al
2O
3 composite anodic oxide films [
12,
13,
14,
15,
16,
17,
18].
These studies focus on specific capacitance and flat Al foil. However, the other dielectric properties, apart from the capacitance of composite oxide film on DC-etched Al foil with high-density tunnels, have not been given sufficient attention in the past. Meanwhile, compared with electrodeposition, the pore-filling method, and hydrolysis/precipitation, the sol–gel process possesses noticeable advantages such as homogeneity, controllability, and low cost, promising industry application. In our previous work, a TiO
2-Al
2O
3 composite oxide film was successfully fabricated on an AC-etched Al foil via the sol–gel process [
19].
Therefore, in this study, a 60 V formation voltage TiO2-Al2O3 composite oxide film was produced on DC-etched Al foil with high-density tunnels by sol–gel, calcination, and anodization. Its composition and microstructure are studied by SEM and EDS. Its properties and performance are evaluated by EIS, specific capacitance measurement, loss tangent, and anode voltage–time curves.
  2. Experimental 
  2.1. Samples
The 120 μm thick commercial DC-etched Al foil with multitudinous tiny tunnels, 99.99% high purity, and >98% cubic texture was tailored into samples in a size of 1 cm × 5 cm with a handle.
  2.2. Sol–Gel Preparation and Coating
First, 10 mL tetrabutyl titanate was slowly added to the 35 mL dehydrated ethanol under intense magnetic stirring, which lasted for 10 min to obtain yellow clear solution A. Second, 2 mL glacial acetic acid and 10 mL distilled water were slowly dripped to the 35 mL dehydrated ethanol under intense magnetic stirring to obtain solution B, the pH of which was adjusted to about 3 by trace amount of hydrochloric acid. Third, solution A was slowly dripped into solution B under room-temperature water bath and intense stirring to obtain light-yellow TiO2 precursor sol.
The specimen was immersed in the sol for 2min and pulled out from the sol at 1 mm/s. After being taken out from the sol, the specimen was heated in drying oven for 5 min at 80 °C. The dried specimens were calcinated in a furnace at 500 °C for 5 min. This procedure was repeated two times.
  2.3. XRD Characterizing 
TiO2 gel was prepared by stirring and heating the sol at 40 °C in a water bath for 1 h and then were calcined at 500 °C for 5 min. The obtained powders were characterized by X-ray diffraction (XRD, D8ADVANCE). During characterization process, the scan rate was set as 10°/min and step size was set as 0.02° in the 2 θ range of 10–80° for phase identification.
  2.4. Anodizing
According to the EIAJ criterion, the samples were first anodized to 60 V in 13% C6H13NO4 electrolyte at 85 °C with a constant current of 25 mA·cm−2. Once the selected voltage was obtained, the anodizing continued for 10 min, while the voltage was held constant and the current was permitted to drop. The time variation in the anode voltage were monitored and recorded by a multi-meter combined with PC system during anodizing process. After anodization, the samples were taken out from the solution, rinsed with deionized water, blown dry, and annealed at 500 °C for 2 min. After annealing, the samples were given re-anodization at 60 V for 2 min. The blank specimen without any pretreatment suffered the same anodizing process for comparison.
  2.5. Micromorphology and Composition of Anodized Samples 
After electro-polishing in acetic acid and perchloric acid solutions, scanning electron microscope (SEM, ZESSEVo18, Carl Zeiss GmbH, Oberkochen, Germany) and energy-dispersed X-ray analyzer (EDX, JEOL JEM-2000ES, Tokyo, Japan) were adopted to examine and analyze the micromorphology and composition of anodized samples.
  2.6. Property of Anodized Specimens 
The specific capacitance (C60V) and loss tangent (tan δ) of the anodized samples was measured in 13% C6H13NO4 electrolyte at 30 °C by a digital LCR meter (YY2810A, Tianjin Anfutai Electronic Technology Co., Ltd., Tianjin, China). A pure Al sheet with a very large area was adopted as the counter electrode. The testing frequency was adjusted to 100 Hz. Three parallel samples are used for mean value and reliability as one group.
Dielectric properties of the anodic oxide film were further examined by EIS measurement (Chenhua, CHI660, Shanghai Chenhua Instrument Co., Ltd., Shanghai, China). In EIS measurement, after C60V and tan δ measurements for each group, only one sample was immersed in 13% C6H13NO4 electrolyte at 30 °C, and 100 mV of sinusoidal voltage in 10−2–105 Hz range was exerted. The software ZSimpWin 3.20 was used to analyze the EIS results. 
  3. Results and Discussion
  3.1. XRD Patterns
Figure 1 demonstrates the XRD pattern of TiO
2 precursor gel calcined at 500 °C for 5 min. The diffraction peaks correspond to an anatase structure, a manner well consistent with JCPDS card No. 21-1272. It is conspicuous that calcination at 500 °C for 5 min is enough for the crystallization of TiO
2 gel.
 During the sol–gel process, the following hydrolysis and polycondensation reactions can occur.
        
In the following calcination step, the above -Ti-O-Ti gel structure was further transformed into an anatase structure. It has been reported that the calcination temperature dominated over calcination time during the process of crystallization of TiO2 films, because when the annealing temperature is increased, the TiO2 films show a higher crystallinity with anatase structures. However, at 580 °C and higher temperatures, the highly pure Al foil substrate is estimated to recrystallize among the tunnel and even melt, which will probably reduce the specific capacitance and effective surface area of the Al foils. Therefore, it is reasonable to select 500 °C as the annealing temperature for crystalizing TiO2 precursor gel into an anatase structure.
  3.2. Anodizing Behavior of Samples 
Figure 2 shows the variations in the anode voltage with anodization time for the TiO
2-coated sample and a blank one. The voltage oscillation caused by the breakdown and reformation of an anodic film can be observed in the two curves for specimens, in agreement with others reports [
18]. However, the initial anode voltage for the TiO
2-coated sample is about 3V higher than the blank one. At the same time, the curve slope for the TiO
2-coated sample is steeper than the blank one. The thermal aluminum oxide film and TiO
2 layer resulting from calcination may account for differences in initial anode voltage and V-t slope. It is generally believed that amorphous or crystalline γ-Al
2O
3 (for anodizing pure Al foil) can form depending on electrolyte temperature and anodizing voltage, and a high electrolyte temperature and high anodizing voltage tend to form a crystalline one. The reports form H. Uchi, T. Kanno and R. S. Alwitt suggested that the formation constant, 
K value, for amorphous e Al
2O
3 anodic films is somewhat larger than that for crystalline ones. The thermal Al
2O
3 film is beneficial for the growth of an additional crystalline γ-Al
2O
3 film, and a TiO
2 layer can promote the formation of a composite dielectric oxide film with a low 
K, during anodization, contributing to power conservation for anodizing. 
   3.3. Surface Microstructure and EDS of Specimens
Figure 3 shows the surface SEM images of anodized samples (a,b) with a pure Al
2O
3 aluminum anodic oxide film and (c,d) with an Al
2O
3-TiO
2 composite anodic oxide film. Cubic tunnels characteristic of DC-etching can be easily discerned for both samples. After anodization, barrier-type anodic oxide films have formed on the walls of these tunnels. Some films were stripped off the tunnel walls and even fall off due to electropolishing in CH
3COOH, HClO
4, and C
2H
6O mixed solution. 
 The thickness of the pure Al2O3 barrier-type film is about 177.65 nm, much larger than the Al2O3-TiO2 composite one at only 87.26 nm. Therefore, the composite anodic film will maintain the effective surface area from DC-etching, beneficial for increasing the specific capacitance of the anode Al foils.
Figure 4 illustrates the area EDS of anodized specimens with a single Al
2O
3 aluminum anodic oxide film and an Al
2O
3-TiO
2 composite one. For the blank sample, Al, O, and C elements can be detected, characteristic of a pure Al
2O
3 aluminum anodic oxide film. Trace amounts of C elements can be attributed to the pollution of samples, impurities in the vacuum chamber, and ammonium adipate solution during anodization. For samples pretreated by TiO
2 precursor sol immersion and calcination, Al, O, Ti, and a trace of C elements can be discerned, corresponding to the Al
2O
3-TiO
2 composite anodic oxide film.
 Figure 5 illustrates a schematic model about the growing of the Al
2O
3-TiO
2 composite anodic oxide film. 
 After immersion in the TiO2 precursor sol and then drying, a lay of sol–gel was coated on the surface of DC-etched Al foil. After calcination, the above sol–gel was transformed into a TiO2 coating and a thermal layer was also formed. During the following anodization, the thermal layer was transformed into barrier-type Al2O3 anodic films under the electric field. At the same time, O2− anions were separated from the water at the base of the TiO2 layer, while TiO44− or TiO32− ions were transported inward and Al3+ ions transported outward. The movement of these ions led to the formation of an Al2O3-TiO2 composite anodic oxide film.
  3.4. Property of Anodized Specimens
The specific capacitance (C
60V) and loss tangent (tan δ) of blank samples and ones with an Al-Ti composite oxide film are listed in 
Table 2 and further illustrated in 
Figure 6. It can be found that pretreatment via TiO
2 precursor sol immersion followed by calcination can increase C
60V from 2.435 μF·cm
−2 to 3.013 μF·cm
−2, about a 23.7% growth rate. This can be attributed to the high relative dielectric constant, 
εr, and the low formation constant, 
K, of an Al-Ti composite oxide film, compared to a single Al
2O
3 film, leading to improvements in the effective surface area of a DC-etched Al foil electrode and its specific capacitance. The change in C
60V for both specimens is consistent with SEM observations and EDS and XRD analysis.
However, the tan δ for the Ti-Al2O3 composite oxide film is 0.041, 20.6% higher than that of the blank one. During the anodization process after TiO2 deposition, O2− ions detached from the water and TiO44− ions in the coating were transported inward and Al3+ ions in the substrate transported outward to form Al-Ti composite anodic oxide film under an electric field. More structural defects due to the doping effect by Ti4+ anions in the anodic oxide film may account for the increase in the tan δ value.
  3.5. EIS of the Anodized Specimens
The EIS of a pure aluminum anodic oxide film and the Ti-Al
2O
3 composite one is shown in 
Figure 7. The Nyquist plots in 
Figure 7a indicate that a single time-constant capacitance behavior was predominant throughout the 10
−2–10
5 Hz range in this study.
In Bode plots in 
Figure 7b, the |Z| values with 
f curves decline sharply with a −∞ slope around 10
−2~10
0 Hz region and suddenly change to a horizontal line with 
f enhancement. The phase angle θ with 
f displays a similar variation tendency to that of the |Z|-
f curves. These patterns are characteristic of dielectric materials. Both samples display the similar performance of |Z|-
f and θ-
f, suggesting that they possess the similar dielectric properties. However, the |Z| value in the Bode plot and the arc radius in the Nyquist plot for pure aluminum anodic oxide film are larger than those the for composite one.
As inserted in the diagram in 
Figure 7a, the EIS can be well simulated and interpreted by an equivalent electric circuit, consisting of a parallel combination 
Cox and 
Rox, connected in series to 
Rs, where 
Cox and 
Rox represent the specific capacitance and resistance of the film, and R
s is the solution resistance. The measured data and calculated data show good coincidence. According to the equivalent circuit and spectra data, the 
Cox and 
Rox values of the film can be calculated and listed in 
Table 3, which also includes film’s other electrical properties, such as C
60V and tan δ.
As shown by 
Table 3, the C
ox and C
60V values for the Al-Ti composite anodic film are higher than those for the blank one. According to the equation 
C = 
εrεoS/
d = 
εrεoS/(
Ea·
K), the C value depends on 
εr/
K under the same formation voltage, 
Ea. It has been reported that the 
εr/
K values of TiO
2 and are Al
2O
3, 30.0 and 7.7, respectively. Therefore, the 
Cox and 
C values of the specimen with the Al-Ti composite oxide film was higher than the blank Al
2O
3 film in this study.
However, the Rox of the Al-Ti composite anodic film is lower than that for the blank one. The decreases are possibly due to the higher number of defects in Al-Ti composite oxide film due to doping, which simultaneously accounts for the increase in tan δ.