The TG-DSC curve of the mixture after ball milling is shown in Fig. 1. It can be seen from the diagram that the pyrolysis of the precursor mixture is mainly divided into three stages, which are mainly corresponding to the reduction of ferric iron in Fe2O3 and the formation of lithium iron phosphate crystals in the temperature range of 442.7 ℃~700.1 ℃. The weight loss in this stage is about 6.7 wt.%. When the temperature exceeds 700 ℃, the mass no longer changes obviously and tends to be stable,and the final weight is 69.1 wt.%. According to the thermogravimetric analysis of the precursor mixture, the precursor mixture will be calcined in the temperature range of 650 ℃~750 ℃.
Figure S1 shows the SEM diagram of LiFePO4/C synthesized under different conditions of orthogonal experiment, It is found that all the samples have no special morphology. The irregular block particles can be observed in the scanning diagram of S2, S3, S4, and S7, and the size is different. On the other hand, the scanning images of other samples show that the calcined products have different degrees of agglomeration, and only one plane with holes can be observed under the same magnification, which may be caused by the alkaline LiOH in the raw materials, such as S1, S6, S8 and S9 samples. Even only one larger particle was observed in S5 sample. The S2 sample has uniform particles and small particle size, which maybe correspond to better electrochemical properties. In order to verify this conjecture, we assembled all the samples into CR2032 button semi-cells for electrochemical performance tests.
Figure 2a shows the electrochemical impedance spectra of LiFePO4/C synthetic materials. The illustration in the figure is a fitted equivalent circuit diagram. It can be seen that the curve is composed of semicircle and slope. The intercept of semicircle in high frequency region represents the resistance RE of electrolyte, RC is the contact resistance at the collector/cathode interface, The intersection of the semicircle and the X axis represents the charge transfer resistance RCT. Among them, the intersection point between the curve of S2 and S7 and the X axis is smaller, which means that the charge transfer polarization resistance of the two samples is smaller.
Figure 2b shows the first circle charge and discharge curve of all samples under 0.5 C. All samples have a discharge plateau at about 3.4 V. The longer the platform length, the smaller the voltage difference between the charging platform and the discharging platform, the less obvious the polarization of the sample during the charging and discharging process, and the better the electrochemical performance. The specific capacity of S2 and S7 is obviously higher than that of other samples, and above 130 mAh·g− 1, they reach more than 76.5% of the standard specific capacity.
Figure 2(c, d) is a comparison of all samples cycled for 50 cycles at 0.5 C charge-discharge rate. We notice that the discharge specific capacity of S2 sample increases after 50 cycles, reaching 136 mAh·g− 1, indicating that the battery will have a slight polarization phenomenon in the initial charge-discharge process, and the subsequent discharge specific capacity will increase after the completion of polarization.
To select the optimal preparation conditions, we drew the horizontal factor diagram of the orthogonal experiment, as shown in Fig. 3. The results show that the influence of each factor on LiFePO4 is different, as follows: ball milling time > heating rate > calcination temperature > sintering time. To sum up, the optimum ball milling time for preparing LiFePO4 was 3 h, the heating rate was 5 ℃·min− 1, calcination temperature was 750 ℃, and the holding time was 4 h.
We try to improve the cyclic performance of the material by doping Na+ into LiFePO4. Figure 5a shows the XRD patterns of Na-0, Na-0.25, Na-0.5, and Na-0.75 samples. The peak shape of each sample is complete, and the position and intensity of each diffraction peak are consistent with the XRD pattern of LiFePO4 (PDF#40-1499). The sharp peak indicates that all the samples have good crystallinity and Na doping doesn’t change the structure. At the same time, the diffraction peaks of Na2CO3, LiOH, Fe2O3 and impurities were not observed in the XRD spectrum, indicating that the raw materials reacted completely and no impurities were introduced under these conditions. For to explore the effect of different Na+ doping content on cell parameters and cell volume, all samples were refined by Rietveld. The lattice parameters after refinement are shown in Table 3. As the amount of doping increases, the a, c and v values of the samples decrease, which has also been confirmed by Liu et al [13]. The values of a, c and v of Na-0.25 are the largest of the three samples. And the lithium ion channel of this sample is the widest. The decrease of lattice constant b is beneficial to the intercalation / delamination of Li+ and shortens the diffusion distance of Li+. Generally speaking, the lattice constant changes little with the increase of Na content. When the doping amount is too high, LiFePO4 with different structure may be formed.
Table 3
Lattice parameters of Li1 − xNaxFePO4/C
Sample | a/Å | b/Å | c/Å | V/Å3 |
Na-0 | 10.3470 | 6.0189 | 4.7039 | 292.947 |
Na-0.25 | 10.3646 | 6.0068 | 4.7145 | 293.516 |
Na-0.5 | 10.3499 | 6.0176 | 4.7124 | 293.496 |
Na-0.75 | 10.3354 | 6.0198 | 4.7082 | 292.930 |
Figure 4(a, b, c) shows the SEM images of three samples. There is no special appearance. The morphology and particle size of LiFePO4/C with different Na+ doping amount are basically the same. For to further explore the microstructure of the materials, Na-0.25 samples were scanned by TEM. The result is shown in Fig. 4d. Figure 4e shows the Bragg lattice of the sample, and both of them can see the ordered bright spots corresponding to the (121) crystal plane of lithium iron phosphate crystal respectively. Figure 4(f, g) is a high resolution transmission image of the sample. Na+ doping LiFePO4/C particles have high crystallinity, the particle surface is covered with a uniform carbon layer, and carbon exists in amorphous form. Continuous and complete carbon coating can not only effectively restrain the growth of grains, but also ensure the full contact between electrons to realize the rapid transfer of electrons [7, 32, 33].
Figure 5(b, c, d) and S2 show the XPS spectra of three samples, and further analyzes the surface element composition and valence information of LiFePO4/C cathode materials. The peaks of Na 1s, Fe 2p, O 1s, C 1s and P 2p can be observed from Figure S2. The XPS spectrum of Na is shown in Fig. 5b. The main peak of the three samples is approximately at 1071 eV, which is attributed to Na 1s. For to investigate the influence of doping on the oxidation state of Fe, the XPS spectrum of Fe 2p was studied, as shown in Fig. 5c. All samples have two peaks with binding energies around 710.2 and 723.8 eV, corresponding to Fe 2p1/2 and 2p3/2, respectively. This corresponds to Fe 2p in LiFePO4 [34]. The binding energy of the main peak and subsidiary peak of each sample has no obvious change with the different doping amount, indicating that the doping of Na+ has no obvious effect on the chemical valence of Fe (II). The lattice distortion is usually caused by the doping of other ions. This analysis shows that the charge difference caused by Na+ doping may be balanced by cation vacancies, which is helpful to improve the electronic conductivity. Figure 5d is the XPS spectrum of C1s. Three peaks approximately located at 283.8 eV, 284.5 eV, 287.2 eV were detected, corresponding to C-C, C-O, O-C = O [35].
Figure 6 shows the electrochemical performance diagram of four samples. We carried out electrochemical impedance spectroscopies (EIS) tests to explore the effect of Na doping on the kinetic behavior on the samples. The optimal equivalent circuit model is given in the figure, as shown in Fig. 6a. The RCT of the three doped samples are 355, 407 and 482 Ω, respectively. It is generally believed that the charge transfer resistance is closely related to the electrode reaction kinetics. The smaller the charge transfer resistance, the better the kinetic performance of the electrode. It can be easily seen that Na-0.25 samples can provide better kinetic behavior, which is anastomose with the above electrochemical tests.
For to further understand the structure of carbon, the Raman spectra of three samples were analyzed as shown in Fig. 6b. Two prominent peaks can be seen from the diagram, one is the G band related to graphite (sp2), which is located in ~ 1600 cm− 1, and the other is that the D band related to disordered carbon (sp3) is located in ~ 1360 cm− 1 [36]. In addition, according to the Raman analysis, the ID/IG values of the three samples are 0.480, 0.413 and 0.403 respectively. It shows that the LiFePO4/C cathode material with 0.25 doping content has higher degree of graphitization and better electrical conductivity.
Figure 6c shows the first charge-discharge curves of four samples at 0.5 C. The initial discharge specific capacities of the four samples are 135.9, 142.1, 130.8 and 123 mAh g− 1. The initial discharge specific capacity of Na-0.25 is the highest. With the increase of the doping amount, the decrease of the first charge/discharge specific capacity may be due to the different insertion/deintercalation modes of Na+ and Li+ during the charge-discharge process. It will stay in position 4a [37, 38]. With the decrease of Li+ intercalation/deintercalation, the first discharge capacity of the sample will decrease.
Figure 6d shows the cycle performance curve of four samples after 50 cycles at 0.5 C. It can be seen that the first discharge specific capacity of the sample will change with the increase of Na+ doping amount, which is 135.25, 139.35, 130.19 and 122.8 mAh·g− 1. The discharge specific capacity after 50 cycles is 131.45, 137.65, 129.2and 118.1 mAh·g− 1. This may be due to the fact that the activity of Na+ doped Na+ is lower than that of Li+. The Na+ maintains the 4a position and supports the one-dimensional channel, which makes the crystal structure of the material stronger and the cycling performance improved [39–41].
We use the first principle calculation method to verify the correctness of the experiment. The lattice constants and Fermi energy obtained from the optimized Li1 − xNaxFePO4 system are shown in Table 4. We can see from the table that with the increase of Na+ content, the lattice constant increases and the Fermi energy decreases. This is because the radius of Na+ is larger than that of Li+. After Na+ occupies the Li site, the lattice expands slightly and the volume increases. The band structures of Li1 − xNaxFePO4(x = 0, 0.25, 0.5, 0.75) are shown in Fig. 7(a-d). It can be seen that the position of the conduction band gradually moves down with the increase of the doping amount x, the width of the band gap becomes narrower. It means that the shorter the path from the valence band to the conduction band is, the less energy is needed, which improves the electronic conductivity of the system to a certain extent.
Table 4
Lattice constants and Fermi energies obtained after optimization of Li1 − xNaxFePO4 system
Doping amout | a/Å | b/Å | c/Å | V/Å3 | Fermi energy/V |
0 | 9.8527 | 5.7893 | 4.6635 | 266.010 | 4.40 |
0.25 | 9.8955 | 5.8430 | 4.7120 | 272.391 | 4.35 |
0.5 | 9.9116 | 5.9116 | 4.7701 | 279.397 | 4.26 |
0.75 | 9.9782 | 5.9528 | 4.8308 | 286.895 | 4.18 |
Figure 7(e-h) shows the density of states (DOS) for all samples. By comparison, it is found that the peak near the Fermi energy of the system becomes slightly sharp after doping. This means that the number of energy levels near the Fermi energy increases, which may increase the electronic conductivity of the doped LiFePO4. The energy band near the − 45 eV range should be contributed by Li-2s orbital electrons. With the increase of the amount of Na+ doping, the number of Li decreases, and the intensity of this peak decreases. At the same time, there is an energy band formed by the participation of Na-3s orbital electrons in -51 eV. Therefore, the gradual increase of the amount of doping will gradually reduce the band gap. But with too much doping, too many Na atoms will occupy more Li sites. In addition, its radius is larger than that of Li+, which leads to greater distortion of the lattice and decrease the electrochemical performance. It can be inferred that the Li+ diffusion of Li0.75Na0.25PO4 is least affected by the hindrance of Na, and the electrochemical performance should be relatively good. This is consistent with the experimental results.
Figure 8 shows the total density of states of Li0.75Na0.25FePO4 and the partial density of states of each element. By comparing the PDOS diagram of LFP, it can be found that the energy levels near the Fermi energy before and after doping are still mainly contributed by Fe-3d electrons. However, the band gap decreases after doping, indicating that the doping of Na is not directly involved in the formation of the energy level near the Fermi level, but indirectly changes the band gap of the system by affecting the electrons of the Fe-3d orbital. We compared the average length of Li-O bond before and after doping. It is found that before doping, the length is 2.10 Å, and after doping, the length is slightly increased to 2.14 Å. The deformation of the atomic position of Fe may lead to the increase of the length of Li-O bond, which slightly broadens the channel of Li ion migration. The conduction band in the range of 0 ~ 10 eV is mainly contributed by electrons on the Fe-3d and P-3p. The valence band in the range of -25 ~ 0 eV is also mainly contributed by the electron contribution of 2s, 2p of O and 3s, 2p of P. The peaks of the four are wide and the bonding is strong, so it is easy to form a [PO4] tetrahedron. It is found from the diagram that the electrons of Li atoms still have strong delocalization after doping. The interaction force with other atoms is weak, mainly electrostatic Coulomb interaction, and Li+ are relatively free, indicating that Li0.75Na0.25FePO4 can be used as cathode materials.