Phosphorus-doped TiO₂ mesoporous nanocrystals for anodes in high-current-rate lithium ion batteries

TiO₂ mesoporous nanocrystals (TMC) were firstly prepared by a common hydrothermal process, which described and characterized detailly in supplementary information (figure S1). For d-TMC production, the mixture of prepared TMC sample and NaBH₄ with the weight ratio of [TMC]:[NaBH₄] = 2:1 was collected and thoroughly mixed using an Agate pestle and mortar, followed by heating in the furnace at 350 °C under argon atmosphere for 4 h. The sample then cooled down, washed with DI water for several times, and dried at 80 °C to get the light brown powder.

Before P-doping process, the d-TMC sample was firstly manually ground for 10 min in an agate mortar to decrease the particle size and prevent sample aggregation. Afterwards, P-TMC was fabricated by a phosphorylation process in a low-cost reactor made from two-neck round-bottom flask connected with a glass tube (figure S2), in which the gas-solid P-doping reaction would perform at 300 °C for 2 h with the assistance of sodium hypophosphite (NaH₂PO₂, Acros Organics, >99%) based on d-TMC. In detail, NaH₂PO₂ was placed on the one of a flask and the d-TMC was placed on the other one. As the temperature increases, phosphine gas slowly evolved and passed through the glass tube along with the argon carrier gas for accelerating the doping of the phosphine gas on the surface of the TiO_2–x. Finally, the brown powder of P-TMC was obtained and the P doping amount was about 2.8 at%.

X-ray diffractometry (XRD, D8-SSS with Cu kα radiation (λ = 0.1506 nm)) and Raman spectroscopy (Nanofinder 30 from Tokyo Instruments, Inc., excitation line at 632.8 nm) were used to characterize the crystal form and structure of the samples. Field emission scanning electron microscope (FESEM, Zeiss Ultra Plus, accelerating voltage of 3 kV) and high-resolution transmission electron microscope (HRTEM, JEM 2010, 200 kV) were used to characterize the morphology and elemental distribution of the prepared samples. The electronic states of as-synthesized samples were determined by x-ray photoelectron spectroscopy (XPS, ULVAC-PHI PHI 5000 Versa P-robe x-ray photoelectron spectrometer, Al Kα x-ray). The surface area was measured from the N₂ adsorption–desorption isotherms by using an automated gas sorption analyzer (Quantachrome Instruments autoSorb iQ-TPX) with the Brunauer–Emmett–Teller (BET) method. Electron paramagnetic resonance (EPR) analysis was carried out using a Bruker EMX-plus (PREMIUM X)-10/12 spectrometer.

The anodes were made by dispersing active material, carbon black (Alfa-Aesar, >99%), and poly(vinylidene fluoride) (PVDF) (Sigma-Aldrich, >99%) binder in N-methylpyrrolidone (NMP) (Alfa-Aesar, >99%) solvent, at a weight ratio of 7:2:1. The active material loading was about 1.0 mg cm⁻². The lithium-ion battery test was carried out with 2032-type coin cells, which were assembled in an Ar-filled dry glove box with the samples as the positive electrodes and the lithium foil as the negative electrode. The electrolyte is the 1

M LiPF₆ dissolved in a mixture of ethylene carbonate (EC)/diethyl carbonate (DEC) (1:1, v/v) (Guangzhou Tinci Materials Technology Co., Ltd China). Celgard 2325 membrane was used as the separator. Electrochemical evaluations were carried out in the voltage window of 1.0–3.0 V versus Li/Li⁺ by using a Wonatech WBCS 3000 battery tester (Korea) and Landt CT3002A battery test system (China). The cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) tests were operated on a CHI 611E electrochemical workstation.

Figure 1 schematically illustrates the formation mechanism of the P-doped anatase TiO₂ mesoporous nanocrystals. Firstly, the TiO₂ mesoporous nanocrystals (TMCs) were synthesized through a hydrothermal process. Subsequently, the TMCs were mixed with NaBH₄ and then heated under Ar atmosphere to create oxygen vacancy defects on TiO₂ nanocrystals and metastable surface layer, which was beneficial to facilitate efficient phosphorus doping due to the increased electronic conductivity and enhanced diffusion and mobility of heteroatoms [21]. Afterwards, during P-doping process of phosphorylation, P (V) dopants substituted Ti sites in TiO₂ lattice happened, in which the charge compensation was accomplished by the reduction of Ti⁴⁺ to Ti³⁺ per P⁵⁺ introduced and/or the formation of oxygen interstitials, resulting in an increase in electron mobility. Finally, P-doped TiO₂ mesoporous nanocrystals with enhanced electronic conductivity and fast diffusion path for Li⁺ ions were carried out. Figure 2(a) gives the XRD pattern of the d-TMC and P-TMC materials. Both results show the characteristic anatase TiO₂ pattern with no other impurity peaks and no obvious difference between the two, implying a lack of bulk structural change to the as-synthesized TiO₂ samples. Nevertheless, the ionic radius of P⁵⁺ (38 pm) is smaller that of Ti⁴⁺ (60 pm); thus, substituting P⁵⁺ for Ti⁴⁺ may lead to the TiO₂ lattice contraction. Figure 2(b) and figure S3 display that the peak positions (2-theta) of the two anatase planes in TiO₂, (101) and (200), have a slight shift toward higher diffraction values with P-doping. This phenomenon is caused by the contraction of the TiO₂ lattices. The fringe spacings of 3.50 (d101) and 1.87 Å (d200) for d-TMC is decreased to 3.47 and 1.86 Å for P-TMC, according to the calculation result from Bragg's law of nλ = 2d sinθ where θ is the incident angle, n is the reflection order, λ is the wavelength of the incident x-rays, and d is the interplanar spacing between planes in the atomic lattice. Therefore, the changes of XRD peaks demonstrate that the P is successfully doped into TiO₂ crystal framework through the gas phosphorylation process. Figures 2(c) and (d) display the Raman spectroscopy of the d-TMC and P-TMC samples. For the d-TMCs, a strong band at 153 cm^–1 (vibration mode E_g), three bands at 398, 514 and 636 cm^–1 (modes B_1g, A_1g and E_g), and a weak band at 201 cm^–1 (mode E_g) are displayed, confirming that our samples belong to anatase TiO₂ [34, 35]. This result is consistent with the results obtained by XRD. From the Raman analysis, we can observe that all the bands of the P-TiO₂ sample become broadening and weakening of intensity as compared to the undoped one due to the formation of Ti–O–P with asymmetric vibration and the existence of oxygen vacancies [34, 36].

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Morphologies of d-TMC and P-TMC are revealed by SEM micrographs and presented in figures 3(a) and (b). Both samples appear smaller spherical-like nanostructures. Interesting, reduction of particle size is observed as doping of phosphorus. To further elucidate the size and structure of these particles for both samples, TEM measurements are carried out. Figures 3(c)–(f) shows the low magnification and high-resolution TEM images of d-TMC samples with and without P-doping. It reveals the almost spherical shaped particles for both cases, but with the particle size distribution of 5−30 nm for d-TMC and about 2−8 nm for P-TMC, respectively. In addition, according to figure 3(e), the lattice spacing of 0.358 nm indexed to (101) facets of anatase TiO₂ is found out for d-TMC sample. After P-doping, we can observe that part of P-TMCs interplanar spacing becomes narrower due to the smaller ionic radius of the doping element, which is consistent with the XRD result. In addition, the EDX element mapping of the P-TMC sample demonstrate the uniform distribution of Ti, O, and P elements in the sample (figure S4). Therefore, the above observations confirm that the P elements can serve as cation doping into d-TMCs with our designed efficient synthesis methodology.

Zoom In Zoom Out Reset image size

To further check the chemical structure and oxidation states of elements that exist in P-TMC and d-TMC, XPS analysis are carried out. The survey scanned spectra (figure 4(a)) demonstrate that Ti, C, and and O elements exist in undoped d-TMC. One extra peak of P in addition to these peaks appears in P-TMC, where the P content is about 2.8 atom% from the measurement. In figure 4(b) of the high resolution XPS spectra related to O 1s, there are three deconvolution peaks located at around 532.3, 531.4, and 529.9 eV observed. The peak located at approximately 532.3 eV is corresponding to the surface oxygen about the –OH adsorbed on the material surface (O_I). The peak at 531.4 eV is assigned to the surface-adsorbed oxygen and/or the surface oxygen defects (O_II). The peak observed at about 529.9 eV represents the lattice oxygen (O_III). And the relative content of oxygen vacancies could be calculated by the ration of (O_I + O_II)/O_III [37]. It could be obtained that the peak area ratio of (O_I + O_II)/(O_III) in P-TMC is 1.3, which is higher than that of d-TMC (0.7). This result demonstrates that the oxygen vacancy content could be increased by P-doping. For the P 2p spectrum illustrated in figure 4(c), there is only one peak at approximately 133.40 eV corresponding to phosphorus in the form of pentavalent-oxidation state (P⁵⁺) observed. Also, no peak at around 129 eV corresponding to characteristic peak of P in Ti–P bonds formed as P atoms replacing O atoms in the lattice is obtained, implying the substitution of Ti sites by P dopants in the crystal lattice of TiO₂ [38]. Further seen from figure 4(d) of Ti 2p spectrum, it shows that there are two characteristic peaks in the Ti 2p region. The peaks at binding energies 458.83 and 464.66 eV for d-TMC, and 458.61 and 464.55 for P-TMC, refer to Ti 2p_3/2 and Ti 2p_1/2, respectively, of TiO₂. It is found that the Ti 2p binding energy in P-TMC is slightly shifted to lower energy states because of the existence of lower binding energy in comparison with that in d-TMC sample. It could be related to the introduction of oxygen vacancies and a slight increase of Ti³⁺ states with P dopants, as demonstrated by other reported works about TiO₂ with defects [21, 39, 40]. For the Ti 2p_3/2 and Ti 2p_1/2 peaks, the signals of Ti⁴⁺ and Ti³⁺ are observed for both samples, and the Ti³⁺ species are created to maintain the charge balance of TiO₂ as oxygen vacancies formation [41]. In addition, the Ti³⁺ signals slightly increased and the Ti⁴⁺ signals decreased after P-doping, which could be ascribed to the replacement Ti⁺⁴ sites by P and the formation of Ti³⁺ for charge balance [42].

Zoom In Zoom Out Reset image size

In addition, the existence of oxygen vacancies is confirmed by EPR analysis, which is used to verify the electron structure of the as-synthesized samples for providing valuable information about the existence of oxygen vacancies, trapped electrons, and surface defects (figure 5 and figure S5). No obvious EPR signal is observed for the non-defective TiO₂. However, a strong EPR signal at g = 2.01 (g-factor) can be seen in d-TMC sample, which is characteristic of unpaired electrons trapped by oxygen vacancies, indicating the presence of oxygen vacancies in TiO₂ lattice [41, 43]. In contrast, the P-TMC sample shows a stronger EPR signal at 2.01, revealing that a higher oxygen vacancy content generated after P-doping process.

Zoom In Zoom Out Reset image size

The isothermal adsorption/desorption curves and pore size distribution curves of d-TMC and P-TMC samples are presented in figure 6 and indicate that the P-TMC sample possesses a large specific surface area with a mesoporous structure. Firstly, both isotherm curves depict the typical type IV behavior with H3-type hysteresis loops due to capillary condensation in the mesoporous channels and/or cages [44]. Using the multipoint BET method, the specific surface area of d-TMC is measured to be 159.757 m² g^–1, which is much smaller than that of P-TMC one with surface area of 234.164 m² g^–1. Additionally, it is observed from the plot that the pore size distribution for both samples is with the average pore size ∼2.0 nm determined by Barrett–Joyner–Halenda (BJH) method. These results declare that the P-TMC sample can create more active site for the contact between the electrolyte and the electrode material, which can promote electrochemical redox reaction and facilitate the ion transport in the electrolyte during charge/discharge process, and then improve the electrochemical performances.

Zoom In Zoom Out Reset image size

To elucidate the positive P-doping effect, both CV and galvanostatic charge–discharge (GCD) measurements were firstly evaluated for d-TMC and P-TMC cells. CV analysis is first performed on a half-cell in a potential range between 1.0 and 3.0 V versus Li/Li⁺ at a scanning rate of 0.5 mV s⁻¹. As illustrated in figure 7(a), it is observed that both cells exhibit similar redox curves with two pairs of redox peaks, suggesting these cells undergo a two-step electrochemical process. The pair of peaks located at near 1.71/2.03 V and 1.72/1.99;V for P-TMC and d-TMC, respectively, can be assigned to Li⁺ insertion/extraction into anatase TiO₂, resulting in reversible phase transition between tetragonal TiO₂ to orthorhombic Li_x TiO₂ [45]. The other pair of peaks observed at approximately 1.50/1.65 V and 1.49/1.60 V for P-TMC and d-TMC are characteristic for the pseudocapacitive Li⁺ storage at the surface due to the formation of oxygen vacancy trivalent titanium (Ti³⁺) [46]. Importantly, the P-TMC cell not only displays a higher current density but also presents enlarged area of CV curve than the d-TMC one (the area of CV is 0.522 mW mg⁻¹ for P-TMC and 0.302 mW mg⁻¹ for d-TMC, respectively), indicating the improvement of energy storage capacities by P-doping. Figure 7(b) illustrate the cycling performance of these anodes at 1 C (170 mA g⁻¹) for 100 cycles. Both anodes exhibit gradual capacity fading but bear the specific capacity of ∼149.1 and 107.6 mAh g⁻¹ after 100 cycles for P-TMC and d-TMC, respectively. We believe that the improved specific capacity of the P-TMC stems from phosphorus-doping in TiO₂ for increasing the electric conductivity of the electrode and delivering faster charge transfer during the Li⁺ intercalation/deintercalation processes. This enhancement can be supported by the EIS data. As revealed by the EIS spectrum (figure 7(c) and figure S6(a)), the P-TMC cell displays a smaller semi-circle in the high-frequency region corresponding to lower charge transfer resistance (R_ct) and a higher Warburg slope in the low-frequency region referring to faster ion diffusion speed and ion storage kinetics, as compared to those of the d-TMC and TMC cells [47, 48]. To further discuss the ion diffusion of these cells (figure 7(d) and figure S6(b)), the corresponding ion diffusion coefficient D, which is related to Warburg impedance coefficient (σ_w) from the linear fitting of Z' and ω^−1/2 in the low-frequency range, is investigated according to the following equations [49, 50]:

$$\begin{eqnarray}&&{Z}_{\mathrm{re}\,}={R}_{{\rm{s}}}+{R}_{\mathrm{ct}}+{\sigma }_{w}{\omega }^{-1/2}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&D=\frac{1}{2}{\left(\frac{{RT}}{A{F}^{2}{\sigma }_{w}C}\right)}^{2}\end{eqnarray} \tag{ 2 }$$

Here R is the gas constant, T is the absolute temperature, A is the contact area between electrode and electrolyte, F is the Faraday constant, and C is the molar concentration of ions. The Li⁺ ion diffusion coefficient D is in direct proportion to (1/σ_w)² from the equation. By calculating the slope of the line after fitting, the σ_w of P-TMC cell is obtained to be 19.50 Ω cm²⋅S^−1/2, which is lower than the values for d-TMC (51.47 Ω cm²⋅S^−1/2) and TMC (141.91 Ω cm²⋅S^−1/2). It indicated the P-TMC sample can provide better lithium-ion diffusion kinetics. The above results confirm that phosphorus doping in TiO₂ could boost the mobility of ions/electrons, beneficial to improve the reversible capacity and lead to improved electrochemical performances for P-TMC sample.

Zoom In Zoom Out Reset image size

Figure 8(a) shows the CV curves of the P-TMC sample in a half-cell configuration scanned at a scanning rate of 0.2 mV s⁻¹, which can be implemented as the primary tool to understand the electrochemical redox reactions during charge/discharge process. The first cathodic CV sweep of the P-TMC displays three peaks, locating at ∼2.32, 1.57, and 1.28 V. The broad peak centered at 2.32 V in the first cathodic scan without an anodic pair disappears in the subsequent cycles, revealing the irreversible formation of the solid electrolyte interphase (SEI) layers, and this may be due to the extension of the voltage window of electrolyte [51, 52]. The cathodic peak at 1.57 V in the first cathodic sweep is assigned to the insertion of Li⁺ into the TiO₂ structure and another one at 1.28 V is corresponding to the pseudocapacitive Li⁺ intercalation at surface. An obvious anodic peak located at 1.97 V associated with the lithium extraction from anatase lattice is revealed. Importantly, the CV curves turn gradually steady when the number of cycles is increased. The basically overlapped CV curves after the fourth scans indicates the cell's good reversibility which can uptake and release Li⁺ ions with good reversibility. In addition, figure 9(a) shows the CV curves recorded at various sweep rates, which can be applied to elucidate the pseudo-capacitive electrochemical behaviors involved in LIB. To be specific, the effects of diffusion-controlled and surface pseudocapacitive reactions could be distinguished from the power-law equation of i = av^b, in which a and b are the constants for the power-law exponent, and v is the sweep rate [47, 53]. In general, the b-value, determined by the relationship between peak current and the sweep rate and derived from the slope of log(i) against log(v), is applied to qualitatively classify the electrochemical behaviors for diffusion-control process as the b-value is close to 0.5 and pseudo-capacitive behavior as the b-value is approaching 1, respectively. From figure 9(b), we could observe that b-values for peak 1 and 2 are 0.61 and 0.64, respectively, indicating its dominance of the diffusion-limited redox reaction. For the peak 3 and 4, the b-values are 0.81 and 0.89, demonstrating the corresponding redox processes are dominated by pseudocapacitive behavior from storage of Li ions occurring at or near the surface of the electrode material. In addition, figure 9(c) indicates that the contribution ratio of the pseudocapacitive capacity reaches 76.07% at the sweep rate of 0.6 mV s^–1. Obviously, the pseudocapacitive contribution increases with the increasing scan rate (figure 9(d)), which are beneficial to the rapid transport kinetics of lithium ions and high rate performance.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 8(b) displays the charge–discharge profiles of the P-TMC anode at 0.2 C. During the first discharge process, the P-TMC anode delivers a high specific capacity of 284.7 mAh g⁻¹ due to the Li-ion insertion into the active material and the SEI formation. Afterwards, the anode exhibits a reversible capacity of about 185 mAh g⁻¹, with the initial coulombic efficiency (CE) of ∼65.0% during the following initial charge process, meaning lithiate-irreversible process occurs to form an irreversible and ionically conductive Li_x TiO₂ surface layer in the first charge/discharge cycle [54]. Nevertheless, the CE values rapidly increase and achieve to ∼99% in subsequent cycles. It is noted that there is still large room for increasing the initial CE of the P-TMC anode for practical application. For example, controlling the structure/morphology design of the electrode materials and the cell configuration assembly further [55]. Therefore, the development of a nanostructured P-doped TiO₂-based material with optimal architectural design and microstructural engineering will be the focus of our future work in this system.

The improved electrochemical performance for P-TMC sample for LIB application compared to the d-TMC one and TMC sample after annealing was additionally detected using the rate capability measurements at various current rates ranging from 0.2 to 20 C for several consecutive discharge/charge cycles. (figure 8(c) and figure S7). The discharge capacities of 187, 174, 162, 151, 136, and 123 mAh g^–1 are recorded at the increasing current rates of 0.2, 0.5, 1, 2, 5, and 10 C, respectively. At the highest current rate of 20 C (2 min per charge or discharge), the specific capacity for P-TMC is 104 mAh g^–1, much higher than that of d-TMC (65 mAh g^–1) and TMC based one (20 mAh g^–1). Moreover, the discharge capacity could be recovered to 176 mAh g^–1 with approximately ∼94% capacity retention when the current rate returns to 0.2 C, which represents an excellent rate capability. Compared to the lithium storage performance of other previously reported high efficiency doped titanium oxide-based materials (figure 8(d)), the P-TMC sample exhibits competitive rate capability, owing to improved electronic conductivity induced by P-doping [56–66]. To further verify the improved rate capability of the P-TMC sample, the longer cycling at a high rate of 10 C was also carried out. As it is shown in figure 8(e), even after 2500 cycles at 10 C-rate, the P-TMC electrode exhibits approximately 101 mAh g^–1 with a capacity retention of 82.6%. It indicates that the average capacity decay for P-TMC is merely around 0.007 mAh g⁻¹ per cycle, indicating its good cycle life. The good cycling performance of P-TMC may be due to the stable structure, in which the existence of mesopores among TiO₂ nanocrystals can buffer the high rate insertion–extraction of Li⁺ ions during cycling test. As can be seen in the inset of figure 8(e), both spherical-like and mesoporous structure is still preserved after 2500 cycles at 10 C, even the aggregation of nanoparticles occurs. It indicates that the good structural integrity of the P-TMC anode. Additionally, it is seen from table S1 that the as-prepared P-TMC sample composite here not only exhibits competitive rate capability but also better cycling stability in comparison to other TiO₂-based nanomaterials reported recently for LIB anodes. These results demonstrated that our designed P-TMC is an ideal anode material for LIBs as the mesoporous structure can increase the contact area with electrolyte by high surface area and the existence of oxygen vacancy as well as P-doping can improve electronic conductivity and shorten the Li⁺ diffusion path, finally greatly enhancing the lithium ion storage capability.