Introduction

Supercapacitor (SC) has been demonstrated to be a beneficial gadget for advance energy storage device for present and forthcoming technology, because it can work superior than conventional capacitors in criteria of stability, capacity, and energy density. Indeed, the performance of SC device is significantly governed by the electrochemical properties of electrodes, which depend strongly on the electrode materials. In general, as suggested and reported by numerous research articles, the efficient electrode should have a huge surface area, high porosity with appropriate pore size distribution and coated on a high conducting substrate1,2,3,4. In some kind of SC, the chemical surface of electrode, directly correlate to the oxidation state of the materials, is considered to be an important factor that influence the electrochemical properties, as well. In addition, distinct electrodes morphology obtained by different synthesis conditions and processes can also affect the electrochemical properties of SC. However, when benchmark to battery, SC has some disadvantages of low energy density and output voltage instability. Therefore, much efforts have been driven to overcome these problems to push SC to be properly applied as a novel energy storage device in a wide scale1,2,3,4.

Currently, TiO2 nanoparticles (NPs) was deliberated as an excellently potential metal oxide to be applied for SCs electrode owing to its proper pseudocapacitive behavior, easy to synthesis, nontoxic, low cost and ecological friendliness4,5,6,7. However, TiO2 NPs generally show low electrical conductivity, which can block their performance and needed to be improved. So far, applications of TiO2 NPs doped with metal ions (Cu, Co, Ni, Li, Ag) have been widely investigated4,8, because lithium (Li) is considered to be one of the most promising elements for use as a dopant. It has been applied in a number of disciplines, including passivation layers in perovskite solar cells, lithium-ion batteries, and nanosensors9. Li is also in charge of transporting to TiO2 NPs and ejecting electrons. Furthermore, lithium ions are very mobile, and can be doped with TiO2 to improve the conductivity and to improve electron transmission when oxygen vacancies are passivated8,9,10. For example, Teimouri et al.10 synthesized Li-doped TiO2 films that could show significantly improved conductivity with faster charge transfer in planar perovskite solar cells. Golvari et al.11 prepared dye-sensitized solar cells on the mesoporous beads of Li-doped TiO2 with 7.48% improvement of the device performance. The work of Lakra et al.12 confirmed a good capacitive behavior of synthesized TiO2 NPs and suggested the materials for SC electrodes application. Moreover, Wang et al.13 suggested that the faradaic storage behavior of nanocrystalline anatase TiO2 in aqueous electrolyte might be contributed to the conversion between Ti4+ and Ti3+ in the redox reaction. Meanwhile, the electrochemical properties of a symmetric hybrid SC with electrode of hydrothermally obtained SWCNTs/TiO2 had been studied by Lal et al.14, and the device delivered a high capacitance of 144 F g−1 with 20 Wh kg−1 of energy density and outstanding capacity retention of 95% after 50,000 cycles test. He et al.15 fabricated a current collector of SC based on TiO2 nanotube arrays (NTA) with a cathode of composite MnO2/TiO2 NTA, and attended a high capacitance of 1051 F cm−2. Moreover, a value of 608.2 F cm−2 was accomplished in case of using Fe2O3 modified TiO2 NTA as an anode, and the assembled asymmetric SC could retain about 91.7% capacitance after 5000 cycling tests. Another interesting material that had been considered as a promising applicant for efficient SCs electrode was heterostructure Co3O4/m-NTAs. In this study, Yu et al.16 reported a maximum value of 662.7 F g−1 for specific capacitance with retain 86.0% of the value after 4000 cycles test at 10 A g−1. Similarly, in the report of Li et al.17, an assembled symmetric solid-state SC based on TiO2-CNT electrodes could show a high value of 82.5 Wh kg−1 for energy density and 345.7 F g−1 at 1.0 A g−1 for specific capacitance. In addition, this TiO2-CNT SC could demonstrate a good cycling stability of 93.3% for 10,000 cycling test, which might be due to the fast ion diffusion on surface of the anatase structure. Furthermore, Kumar et al.18 found that a SC with carbon-supported TiO2 electrode could exhibit a specific capacitance 277.72 F g−1 at 25 mV s−1 in 1 M Na2SO4 aqueous electrolyte. In addition, Elshahawy et al.19 reported a value of 57.62 mF/cm for specific capacitance in 2 M KOH electrolyte of TiO2 nanorod arrays based SCs, and could retain a capacity of 91% at the 10000th cycle of test. Ojha et al.20 obtained the mesoporous Mn-doped TiO2 by a simple sol–gel and a solvothermal method. According to this work, they suggested that the enhanced electrochemical properties could be mainly ascribed by many factors such as a larger surface area, a mesoporous structure and an appropriate concentration of Mn doping that could lead to the improved conductivity of a wide band gap TiO2 NPs. Additionally, in the work of Hodaei et al.21, nitrogen-doped TiO2 NPs was obtained by a sol–gel process. The electrochemical study showed the enhanced capacitive performance, including a value of 311 F g−1 at 1 A g−1 for specific capacitance with remained 98.9% of the value after 4000 cycling tests. Regarding to the work of Hodaei et al.21, the authors reported that a sol–gel method was an efficient process for preparing pure and metal-doped TiO2 NPs to be applied for SCs electrode due to its unique chemical reaction that could yield high quality NPs. In addition, NPs of highly homogeneous size distribution with high surface area and very high purity (99.99%) could be obtained at a low temperature. Thus, a sol–gel synthesized metal-doped TiO2 NPs is expected to play an important role for increasing electrical conductivity that can further improve the whole capacitive performance of supercapacitor22,23.

Therefore, in this work, we focus on the electrochemical performance investigation of the as synthesized Li-doped anatase TiO2 NPs (LixTi1-xO2 NPs, x = 0, 0.05, 0.10, 0.15 and 0.20) obtained by a sol–gel process. Interestingly, all LixTi1-xO2 NPs (x = 0.05–0.20) electrodes could demonstrate a pseudocapacitive behavior with a high specific capacitance and good cycling stability compared to undoped sample. To the best of our knowledge, it is the first time to fabricate SC with electrodes based on Li-doped TiO2 NPs.

Experimental

Chemicals

Sigma Aldrich supplied titanium (IV) isopropoxide (C12H28O4Ti, 99.95%), lithium hydroxide (LiOH, 99.50%), polyethylene glycol, ammonium hydroxide (NH4OH, 99.50%), acetylene black (99.99%), polyvinylidene fluoride (PVDF) and ethanol (C2H5OH, 99%). Lithium sulfate (Li2SO4, 95%) is a product of Ajax Fine Chem Laboratory Chemicals. N-methyl-2-pyrrolidone (NMP, 99.5%) was obtained from RCI Labscan.

Synthesis of pure and Li-doped anatase TiO2 NPs

In the process of synthesis pure anatase TiO2 NPs, 5 ml polyethylene glycol was added to a well stirred deionized (DI) water:ethanol solution with 4:1 ratio by volume (40:10 ml). Then 10 ml of C12H28O4Ti was gradually dropped to this solution, while vigorously stirred at room temperature on a magnetic stirrer hot plate for further 20 min. Then, 2.5 wt.% aqueous ammonia (NH4OH) was added dropwise to carefully controlled the pH at 7, and further stirring for 30 min. After that, increased the temperature of a solution to 60 °C and kept on stirring until a wet gel was formed, and allowed to dry at 75 °C. The final product was achieved by crushing the dried gel, ground to fine powder and pyrolized at 500 °C in a furnace for 2 h, using a heating rate of 2 °C/min. LixTi1-xO2 NPs (x = 0.05, 0.10, 0.15 and 0.20) was synthesized by a similar way, only that lithium hydroxide (LiOH) of 0.05, 0.10, 0.15 and 0.20 by wt % was added in the mixture solution before adding NH4OH.

Electrodes fabrication for electrochemical properties study

The electrode slurries of LixTi1-xO2 NPs (x = 0, 0.05, 0.10, 0.15 and 0.20) were prepared by ball milling each product with PVDF and acetylene black at 80: 10: 10 wt% ratio in 500 µL NMP solvent at RT for 24 h. Each electrode was fabricated by dripping an active mass slurry of approximately 200 µL to coat on an area 1 cm2 at one end of ultrasonically cleaned nickel foam sheet of size 1 × 2 cm2, and dried for 2 h at 80 °C. After that, all electrodes were pressed at 1.5 tons for 1 min, and immersed in 0.5 M Li2SO4 aqueous electrolyte prior to electrochemical properties testing. The CV study was performed in an applied voltage of 0.0 to + 0.5 V at scan rate 10, 20, 30, 50, 100 and 200 mV s−1. The GCD study was performed at applied current density 1.5, 2, 4, 6, 8, 10 and 15 A g−1. The capacity retention was evaluated at the 2000th cycle of GCD test at 10 A g−1. The GCD results were used for the calculation of specific capacitance (Cs) using Eq. (1)24,

$${C}_{s}=\frac{I\Delta t}{m\Delta V}$$
(1)

where I, Δt, m, and ΔV stand for the constant discharge current (A), discharge time (s), mass of active material in electrode (g) and potential window (V), respectively.

Additionally, the energy density (Esd) and power density (Psd) of electrodes were determined from the GCD results, using Eqs. (2) and (3)24, respectively.

$${E}_{sd}=\frac{{C}_{s}\times \Delta {V}^{2}}{7.2}$$
(2)
$${P}_{sd}=\frac{{E}_{sd}\times 3600}{\Delta t}$$
(3)

Characterizations

X-ray source with CuKα (λ = 1.5406 Å) generated by X- ray diffractometer (Philips X’Pert) was used for crystal structure and phase identification of the products. Raman study using a laser of 532 nm excitation (DXR Smart, Thermo Scientific) was employed for TiO2 phase verification. Moreover, in order to confirm the existence of various modes of vibration between Ti and O bonding in the TiO2 crystalline structure, Fourier transform infrared spectroscopy (FTIR, Bruker, Senterra) was performed. The surface morphology inspection of products and particles size determination were accomplished by field emission scanning electron microscope (FE-SEM, FEI, Helios NanoLab G3 CX). In addition, the quantitative estimation for major elements in wt% of the products could be achieved using energy dispersive X-ray spectroscopy (EDS) with elemental mapping to display the distribution of elements. Furthermore, high magnified bright field images with selected area electron diffraction (SAED) patterns by transmission electron microscope (TEM, FEI, TECNAI G2 20) was performed for clearer observed products morphology and more accurate particles size determination, including phase and structure confirmation. An instrument of Autosorb1-Quantachrome was employed for the study of specific surface area and a type of pore distributed in samples through the Bruanauer Emmett-Teller (BET) and Barrett-Joyner-Halenda (BJH) techniques, respectively. Finally, an equipment of Wuhan Corrtest Instruments Corp Ltd. (Model CS350 Potentiostat/Galvanostat) was used for electrochemical properties studies of all LixTi1-xO2 NPs electrodes to obtain the CV, GCD and EIS results.

Results and discussion

The XRD patterns with Rietveld refinement fitting of LixTi1−xO2 NPs are displayed in Fig. 1a–e. In Fig. 1a–e, the most dominant XRD peaks at 25.26°, 36.92°, 47.96°, 53.92°, 55.01°, 62.83°, 70.24° and 75.04° correspond to the crystalline diffraction plane (101), (004), (200), (105), (211), (204), (116), and (215), respectively. The XRD results matched with the standard data of JCPDS: 21–1272 for the tetragonal anatase TiO2 crystalline phase of space group: I41/amd9,18,25. However, in a sample of x = 0.15 and 0.20 (Fig. 1d, e), many peaks of monoclinic Li4Ti5O12 phase with space group: C2/c observed at 17.32°, 30.95° and 44.49° correspond with the diffraction plane (111), (311) and (400), respectively, and matching to the standard data of JCPDS: 49-0207. The formation of Li4Ti5O12 phase might be due to the direct interaction of excess Li with pure anatase crystalline TiO2 phase during the growth process. It was suggested that this phase could provide a nonsymmetric stretching vibration of O–Ti–O that could result in reduced conductivity of the samples. Moreover, the cell parameter (a, b and c) with cell volume and various parameters (Rwp, Rp, Rex and GOF, definition for these parameters was given elsewhere) were evaluated by Rietveld refinement method using a standard data of JCPDS: 21-1272 (tetragonal phase with space group: I41/amd) and JCPDS: 49-0207 (monoclinic phase with space group: C2/c), as displayed in Fig. 1a–e, and summarize of the results was listed in Table 1. As seen in Table 1 and the excellent fitting of the XRD patterns in Fig. 1a–e, it can be concluded that Li loading significantly affect the cell parameters of anatase TiO2 phase. Obviously, the cell parameters and cell volume of samples decrease with increasing Li loading, leading to the deceased crystallite size of LixTi1−xO2 NPs. Generally, Ti4+ in a unit cell of TiO2 crystal system is bonded to six equivalent O2− atoms, leading to the formation of mixture distorted edge and corner-sharing TiO6 octahedra. Furthermore, in a unit cell of Li4Ti5O12 phase, a complicated structure is formed owing to the formation of LiO4 tetrahedra by the bonding of Li1+ with four O2− atoms at the cell corners that could be shared with others two equivalent LiO6 octahedra and ten TiO6 octahedra. Moreover, the percentage of anatase TiO2 phase and Li4Ti5O12 phase were determined in samples of x = 0.15 and 0.20, and found to be (95.12 and 4.88%) and (92.23 and 7.77%), respectively. Additionally, the X-ray line of the diffraction planes (101), (004), (200), (105), (211), (204), (116), and (215) were used for the evaluation of average crystallite sizes (DSh) of all samples, using the Scherrer’s equation (4).

Figure 1
figure 1

Rietveld refinement fitted XRD patterns of samples with different Li concentrations and plot of average crystallite size vs. Li concentration.

Table 1 Cell parameters and phase analysis investigated by XRD and Rietveld refinement fitting with the results of surface and pore analysis of LixTi1−xO2 NPs.
$$DSh =k\lambda/\beta \cos\theta ,$$
(4)

In Eq. (4), the parameters θ, λ and β are defined for Bragg angle, wavelength of X-ray and full width at half maximum, respectively. k is the constant and was taken as 0.9. The evaluated DSh values are 19.61 ± 0.59, 19.21 ± 0.56, 18.31 ± 0.52, 18.23 ± 0.64 and 18.12 ± 0.76 nm for LixTi1-xO2 NPs of x = 0, 0.05 0.10, 0.15 and 0.20, respectively. All the DSh values were summarized in Table 1, and the plot of DSh versus Li concentration is illustrated in Fig. 1f. Obviously seen in Fig. 1f, DSh decreases with increasing Li concentration. The decreased DSh was suggested to originate from the replacement of a large ionic radius of Ti4+ (0.745 Å) and Ti3+ (0.670 Å) by a small ionic radius of Li+ (0.60 Å) in the anatase TiO2 crystal structure. By comparing the ionic radius of Li+ (0.60 Å) and Ti4+ (0.745 Å), it is clear that a possible substitution of a small amount of Ti4+ by Li+ would be accompanied by a weak lattice expansion, due to the relatively small difference between their respective ionic radius, so that the Li+ ions can be dissolve into anatase TiO2 phase and Li4Ti5O12 phase26. However, the substitution might induce lattice expansion, resulting in a shift of the anatase peak to the lower angles. Although for the replacement of Ti4+ by Li+ ions, some Ti–O bonds are broken, which leads to the formation of oxygen vacancies, the contraction of lattice caused by oxygen deficiency is eliminated through lattice expansion induced by the presence of the slightly smaller lithium ions11. As a result, Li+ ions appear to be an appropriate option for modifying the local crystal structure at Ti4+ sites in TiO2, because they could operate as charge compensators and could additionally enhance capacitive properties due to the availability of more active sites27.

FTIR spectra of samples are displayed in Fig. 2a–e. The broaden vibration peaks around 3250–3350 cm−1 are assigned for O–H stretching modes, relating to the stretching vibration of the hydroxyl (O–H) group due to the formation of H2O molecules on surface11. Moreover, the appeared vibration peaks around 1640–1644 cm−1 are designated to the symmetric stretching of Ti–OH on surface18. Additionally, the observed peaks around 1110–1113 cm−1 and 600–625 cm−1 indicated the bonding of Ti–O in an anatase TiO2 structure11. In the samples with x = 0.10, 0.15 and 0.20, the observed peaks in a range 790–900 cm−1 are attributed to the symmetric C–H and asymmetric CH2 vibrations of an organic polyethylene glycol that could not be completely removed after calcination11. The strong vibration peaks in a range 410–625 cm−1 are associated with the vibration modes of O–Ti–O bonding in an anatase TiO2 structure.

Figure 2
figure 2

FTIR results of LixTi1−xO2 NPs with different Li concentrations.

Further structural analysis of LixTi1−xO2 NPs (x = 0, 0.05, 0.10, 0.15 and 0.20) was performed by Raman technique, as shown in Fig. 3a–e. In these figures, the major sharp Raman shift observed at ~ 144 cm−1 corresponds to the Eg(1) mode of anatase TiO218,28. The peak at 396 cm−1 corresponds to the B1g(1) mode, while another at 639 cm−1 corresponds to the Eg(2) mode, arising from the symmetric stretching mode of O–Ti–O bonding in crystallite anatase TiO228. The other one observed at 515 cm−1 was assigned for A1g + B1g(2) mode, corresponding to the antisymmetric bending vibration of O–Ti–O bonding in TiO2 structure18,28. Therefore, Raman results confirmed the anatase phase of all samples.

Figure 3
figure 3

Raman results of LixTi1−xO2 NPs with different Li concentrations.

Morphology and average particles size (Dps) of LixTi1−xO2 NPs are demonstrated by FE-SEM images with corresponding histograms in Fig. 4a–e. All images show the homogeneous distribution of agglomerated NPs with intercalated space between them, illustrating the micrographs of porous surface similar to spongy materials. However, a secondary monoclinic phase of Li4Ti5O12 NPs that existed in the samples with x = 0.15 and 0.20 could not be observed or identified by SEM micrographs in Fig. d-1, d-2, e-1, e-2 was suggested to be owing to the small amount of them compared to a major TiO2 phase, as estimated and listed in Table 1. The Dps values are 30.04 ± 4.92, 27.97 ± 6.56, 25.12 ± 2.64, 25.03 ± 5.53 and 24.66 ± 6.13 nm for samples of x = 0, 0.05 0.10, 0.15 and 0.20, respectively. The Dps values were listed in Table 1. As seen in Table 1, Dps decreases with increased Li loading. The uniform dispersion of agglomerated TiO2 NPs in electrodes could lead to enhance the porosity and form the conducting networks for charge transfer, as suggested by Prashad et al.20. Moreover, the observed porous structure of Li-doped anatase TiO2 NPs could increase the surface area of the electrode materials for the adsorption of electrolyte ions, resulting in more charges collection and could finally enhance the specific capacitance. Figure 5a–e display the EDS results and mapping of elements for LixTi1−xO2 NPs. The EDS results clearly show the uniform distribution of major elements Ti and O in the samples. However, Li element could not be detected due to its light-weight and a limitation of the instrument. The atomic percentages for Ti element were estimated to be about 55.0%, 53.7%, 52.9%, 50.9% and 50.5% for LixTi1−xO2 NPs with x = 0, 0.05 0.10, 0.15 and 0.20, respectively. The decreased amount of Ti was due to the Li replacement in anatase crystal structure of TiO2.

Figure 4
figure 4

FE-SEM images (×5000  and ×10,000  magnification) and histograms for particles size distribution of LixTi1−xO2 NPs.

Figure 5
figure 5

EDS results and mapping images of LixTi1−xO2 NPs.

In fact, more accurate particles size of LixTi1-xO2 NPs can be evaluated from TEM bright field images, including a better clear morphology of particles, as illustrated in Fig. 6a–e. As obviously seen, all images display NPs of very fine cuboidal shape with irregular size and agglomerated to form a spongy like-structure with roughly estimated individual particle size in an interval of 5–10 nm. Notably, the particle sizes estimated by TEM are smaller than those evaluated by SEM, which might be due to the dispersion of NPs during the sonication process of samples preparation prior to TEM performance. Moreover, the median size of NPs could be slightly reduced with the inclusion of Li+ ions on the Ti4+ and Ti3+ sites in the anatase TiO2 crystalline structure18,20. Furthermore, the unique morphology and homogenous size distribution in a narrow range of TiO2 NPs was suggested to result in the increase of materials porosity and surface area. Additionally, the halo ring shape with arranged spots on the circumferences of SAED patterns in all samples indicate a polycrystalline nature of the materials18. Moreover, all SAED patterns had been indexed to be composed of different crystalline planes that correspond to those of anatase phase TiO2, agreeing well with the XRD results shown in Fig. 1.

Figure 6
figure 6

TEM bright field images with insets showing indexed SAED patterns of LixTi1−xO2 NPs.

Figure 7a–e display the N2 adsorption/desorption isotherms results of LixTi1-xO2 NPs. Regarding to these results, the observed hysteresis loops of all samples exhibit the BET curve of type IV, corresponding to that of mesoporous materials11,20. The evaluated specific surface area of anatase TiO2 NPs (x = 0) was 163.01 m2 g−1, whereas those of LixTi1-xO2 NPs (x = 0.05, 0.10, 0.15 and 0.20) displayed the larger values of 180.23, 246.94, 239.92 and 241.93 m2 g−1, respectively. Moreover, the average pore size and total specific pore volume of LixTi1−xO2 NPs (x = 0, 0.05, 0.10, 0.15 and 0.20) by the BJH technique were found to be 6.54 ± 0.93, 6.31 ± 0.76, 5.91 ± 0.66, 5.71 ± 0.58 and 5.80 ± 0.61 nm, and 0.30, 0.31, 0.35, 0.37 and 0.38 cm3 g−1, respectively. All of these values were listed in Table 1, and their plots as a function of Li concentration are illustrated in Fig. 7f. It was suggested that the great increase of specific surface area with small pore size of Li-doped anatase TiO2 NPs could create the appropriate pathways for ions to diffuse into the surface of electrodes, resulting in the increased charge collection and improvement of the capacitive performance11,20. The morphology and pore size of the produced compounds indicated nanoscale particles that could provide high electrolyte–electrode interfacial surface area, resulting in the comprehensive permeation of electrolyte and minimizing the path of transport to accelerate the fast transfer of Li+ and e in Li-doped anatase TiO2 NPs cathode27. Furthermore, the mesoporous nature of electrode could improve the access of electrolyte into the bulk of the materials, while also providing high power tapping densities and robust structural and electrical interconnectivity across the electrode29.

Figure 7
figure 7

(a–e) Nitrogen sorption isotherms results with inset showing average pore diameter of LixTi1−xO2 NPs. (f) Plots of BET surface area, average pore size and total specific pore volume (inset) as a function of Li concentration.

Cyclic voltammetry (CV) measurements of LixTi1−xO2 NPs (x = 0, 0.05, 0.10, 0.15 and 0.20) electrodes were performed at a scan rate 50 mV s−1 in 0.5 M Li2SO4 electrolyte within a potential window 0.0–0.5 V. The capacitive performance of each electrode is shown in Fig. 8a. The electrochemical performance of all electrodes performed at different scan rates are displayed in Fig. 8b–f. As seen in Fig. 8a, the CV curve of Li0.10Ti0.90O2 NPs electrode reveals the largest size, indicating a superior electrochemical performance compared to other electrodes. According to Fig. 8a–f, the distorted rectangular shape with apparent redox peaks of CV curves are observed, suggesting a typical pseudocapacitive behavior of LixTi1−xO2 NPs electrodes4,16,20. Moreover, all curves exhibit the stability with increasing scan rate through the whole applied voltage range4,18,20. Additionally, it is obviously seen that with enhanced potential sweep rate from 10 to 200 mV s−1, the anodic and cathodic peaks are shifted to the negative and positive values, respectively. In addition, the appeared anodic and cathodic peaks in a voltage range 0.18–0.35 V were due to the faradaic redox reaction of TiO2 NPs. Generally, the appearance of redox peaks is correlated to the cation interaction on the TiO2 surface, which can be expressed as4,30:

$$\left({\text{TiO}}_{2}\right)surface +\mathrm{ Li}+ +\mathrm{ e}- \leftrightarrow ({\text{TiO}}_{2}-\mathrm{ Li}+) surface$$
(5)

where Li+ could be the Li2SO4 electrolyte. For CV curves of LixTi1−xO2 NPs electrodes displayed associated with Li+ intercalation and de-intercalation into LixTi1−xO2 NPs electrodes. The overall cell reaction for the Li+ insertion/extraction into LixTi1−xO2 NPs electrodes can be written as4,30:

Figure 8
figure 8

CV curves of LixTi1−xO2 NPs electrodes.

$$({\text{LixTi}}1-{\text{xO}}_{2}) surface +\mathrm{ Li}+ +\mathrm{ e}- \leftrightarrow ({\text{Lix}}({\text{Ti}}4+/{\text{Ti}}3+)1-{\text{xO}}_{2} -\mathrm{ Li}+) surface$$
(6)

Owing to the excellent ability of alteration between different oxidation states of Ti ions during the redox reaction, TiO2 was suggested to be a potential material for SCs electrode. Regarding to the redox reaction, Ti4+ was transferred to Ti3+ while charging, and converted to the initial state during discharging18,19, showing a pseudocapacitive behavior of n-type semiconductor for all LixTi1-xO2 NPs electrodes. In addition, it was reported that during discharge, a number of Li+ ions was inserted into the interstitial sites of the Li-doped TiO2 NPs framework, which implied a partial reduction of Ti4+ to Ti3+ state29,31,32,33,34. Moreover, it was also suggested that the redox reaction could be possibly affected by the thickness variation of diffusion layers in electrodes due to using different scan rates in the measurements. The obviously enhanced current with scan rates indicates a good rate capability of Li0.10Ti0.90O2 NPs electrode. While the conductivity of electrodes with higher Li concentration (x = 0.15 and 0.20) was observed to decrease, suggesting to be affected by the defect sites generated by the presence of Li4Ti5O12 phase in the materials.

For further investigation of electrochemical properties of LixTi1−xO2 NPs electrodes, the galvanostatic charge discharge (GCD) measurements were performed at 1.5 A g−1, and the results are displayed in Fig. 9a. In addition, more GCD results performed with variation of current density from 1.5 to 15 A g−1 for each electrode are displayed in Fig. 9b–f. In these figures, GCD curves of all electrodes illustrate a non-symmetrical voltage–time profile at each constant current density, indicating a pseudocapacitive behavior with superior electrochemical reversibility through the whole charge/discharge process4,18,20. The maximum specific capacitance (Cs) at 1.5 A g−1 of 822 F g−1 was achieved in LixTi1−xO2 NPs electrode with x = 0.10, while the Cs values of others electrode with x = 0, 0.05, 0.15 and 0.20 were evaluated to be 513, 666, 747 and 579 F g−1, respectively, as displayed in Fig. 10a, and summarized in Table 2. The longer discharge time than other electrodes at 1.5 A g−1 (see Fig. 9b–f) suggests a good electrochemical performance of Li-doped TiO2 Nps electrode with 0.10% Li loading. The GCD profile of Li0.10Ti0.90O2 NPs electrode, displaying a good performance with the highest Cs value, is in good agreement with its CV results shown in Fig. 8a, d. Generally, Psd and Esd obtained by the Ragone plot are two other important parameters used for evaluating the supercapacitor performance. Accordingly, Fig. 10b displays the Ragone plots of all LixTi1-xO2 NPs electrodes, illustrating the decreased Psd and Esd values with increasing current density, which might be correlated to the limit of ions diffusion in electrodes and electrolyte21. The energy density (Esd) and power density (Psd) at 1.5 A g−1 of LixTi1−xO2 NPs electrodes (x = 0, 0.05, 0.10, 0.15 and 0.20) are (64.12, 83.25, 102.75, 93.38 and 72.37 W h kg−1), and (51.04, 66.67, 176.04, 104.16 and 51.04 W kg−1), respectively. As seen, all Li-doped anatase TiO2 NPs electrodes revealed the improved Psd and Esd as compared to that of undoped anatase TiO2 NPs electrode. The increased electrical conductivity and surface area of Li-doped anatase TiO2 NPs with increasing Li content were suggested for the improved performance. Furthermore, the GCD results for electrochemical stability study at 10 A g−1 with 5000 cycles test are illustrated in Fig. 10c. According to these results, a high capacitance retention of 87.1%, 89.0%, 92.6%, 91.7% and 90.4% at the 5000th cycles were attained for LixTi1−xO2 NPs electrodes with x = 0, 0.05, 0.10, 0.15 and 0.20, respectively, and the values were listed in Table 2. Obviously, all electrodes of Li-doped anatase TiO2 NPs illustrated a superior capacitance retention as compared to undoped anatase TiO2 NPs18. Actually, it has been observed that good stability after testing for 5000th cycles of all LixTi1−xO2 NPs electrodes in aqueous electrolyte of 0.5 M Li2SO4 is the most interesting one, since three main effects are evident; (i) decrease in the aggregation and overlapping of LixTi1−xO2 NPs during a long time required by the charge–discharge process, (ii) the fast Li+ ions diffusion and increased electronic conductivity on surface of LixTi1−xO2 NPs, and (iii) a slight shift of the CV and GCD curves is a very promising strategy to produce an environment friendly supercapacitor, which is able to reach in the future for the targeted energy and power density of organic electrolyte-based systems with acceptable good electrochemical performance35,36,37,38. Additionally, the EIS results of all LixTi1−xO2 NPs electrodes obtained in a range 0.01–100 kHz frequency at 5 mV are shown in Fig. 10d, and all the obtained data were provided in Table 2. Next, the series resistance (Rs) of LixTi1−xO2, NPs electrodes with x = 0, 0.05, 0.10, 0.15 and 0.20 were determined to be 4.05, 4.03, 4.11, 4.10 and 4.03 Ω, respectively. Moreover, Li-doped anatase TiO2 NPs electrodes showed the low Rct value evaluated from a semicircle part in a high frequency region near the origin of the plots, meanwhile the plots connected to a low frequency region show almost the incline straight lines parallel to Z'' (Ω)4,18,20,21. As results, this might cause the formation of more vacancy sites in TiO2 lattice as well as the faster charge transferability in the Li-doped anatase TiO2 NPs electrodes, ascribed primarily due to the change of the intrinsic conductive properties of TiO2 NPs due to Li doping18,20. For Warburg resistance (WR), which describes the ion diffusion process of redox materials, the WR values for all electrodes in a low frequency range were generally evaluated from the slopes of Z'' plots that abruptly increase in a straight line next to a high frequency region. As seen in Table 2, the WR value of LixTi1−xO2 NPs electrode with x = 0.10 is lower than those of other electrodes. The lower value of WR implies the faster ion transferability from electrolyte to electrode, resulting in a greater Cs value18,20. The EIS results clearly indicate the decreased series resistance of Li-doped anatase TiO2 NPs electrodes as compared with the bare anatase TiO2 NPs electrode. Regarding to the ascribed results, it was suggested that LixTi1−xO2 NPs with x = 0.10 was an appropriate material to be applied for electrode of high-performance supercapacitors. To express the excellent and attractive capacitive value of Li0.10Ti0.90O2 NPs electrode, the comparison of its electrochemical performance with others TiO2-based electrode was illustrated in Table 3. Therefore, in this study, we thoroughly explored the influence of Li-doped anatase-TiO2 NPs on high-performance supercapacitors. Generally, pentavalent donor-type doping is required to increase the material's electrical conductivity, and Li+ ions can easily substitute Ti4+ and Ti3+ ions in the anatase-TiO2 lattice at a wide range of concentrations. According to the experimental results, adding a modest amount of Li to the material could potentially affect many factors such as crystal size, phase purity, morphology, surface area, and pore size distribution. Consequently, LixTi1−xO2 NPs electrode with x = 0.10 exhibited superior ion diffusivity, high conductivity, and small particle size compared to the other samples. As a result, Li0.10Ti0.90O2 NPs electrode exhibited a higher electrochemical activity compared to the others electrode. The increased lattice parameters could improve its overall performance by enhancing its metallic-like character, while having minor effects on its electrochemical properties. The CV and GCD tests showed that the Li0.10Ti0.90O2 NPs electrode exhibited a pseudocapacitive storing mechanism, which occurred on the electrode surfaces at Li-doped Ti sites. As a result, Li+ ions played an important role in the improvement of electrical conductivity, charge storage capacity and stability with capacitance retention reaching as high as 92.6% after 5,000 cycles GCD test.

Figure 9
figure 9

(a) GCD results of all LixTi1−xO2 NPs electrodes at 1.5 A g−1. (b–f) GCD results of all LixTi1−xO2 NPs (x = 0.0–0.20) electrodes performed at different current densities.

Figure 10
figure 10

Electrochemical performance of all electrodes, (a) specific capacitance vs. current density, (b) Ragone plots, (c) cycling stability at 10 A g−1 and (d) Nyquist plots of all electrodes in aqueous electrolyte of 0.5 M Li2SO4.

Table 2 Cs values and capacity retention with energy density, power density and EIS analysis of all LixTi1−xO2 NPs (x = 0.0–0.20) electrodes.
Table 3 Electrochemical performance comparison of TiO2-based electrodes synthesized by different methods, performed in different electrolytes and current densities or scan rates.

Conclusions

In summary, anatase LixTi1−xO2 NPs (x = 0, 0.05 0.10, 0.15 and 0.20) could be synthesized by the sol–gel process. The anatase phase with space group I41/amd of tetragonal LixTi1-xO2 NPs was confirmed by XRD results. Additionally, the monoclinic phase of Li4Ti5O12 was detected in samples with x higher than 0.10, suggesting for the reduced conductivity and electrochemical performance of samples. Raman spectra demonstrated characteristic peaks that represented the anatase phase of TiO2 in all samples. Moreover, FT-IR spectra also confirmed the existence of different modes of vibration between Ti and O atoms in the TiO2 structure. TEM and FE-SEM images revealed the homogeneously dispersed Li-doped TiO2 NPs in a spongy like morphology with estimated particles size in a range 5–10 nm by TEM. Such a morphology was suggested to enhance the porosity, and to form an excellent conducting network of NPs for charges transfer. The BET results illustrated the increased surface area of LixTi1−xO2 NPs with increasing Li loading. Electrochemical studies showed the pseudocapacitive behavior of all samples with high-quality performance achieved in a sample of x = 0.10 that revealed the highest value of 822 F g−1 for specific capacitance at 1.5 A g−1, and could retain 92.6% of its original value after 5000 cycles test. Therefore, Li0.10Ti0.90O2 NPs with excellent performance in terms of high capacitance, high power density (176.04 W kg−1) and high energy density (102.75 W h kg−1) was suggested to be an appropriate material for supercapacitors electrodes application.