Electrochemical Investigations of Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg²⁺, Zn²⁺) Cathodes for Lithium Ion Batteries

Abstract

The Mn²⁺ and Mg²⁺ (or Zn²⁺) co-doped Li₂Fe_0.8-xMn_0.2M_xSiO₄ (x = 0.05 and 0.1) are synthesized by a solid-state reaction route. Compared with the single doped Li₂Fe_0.8Mn_0.2SiO₄, the co-doped samples show improved cycling performance. The capacity retention can stay above 50% after 50 cycles, which is significantly higher than 30.4% for Li₂Fe_0.8Mn_0.2SiO₄. This phenomenon could be attributed to the increased structural stability caused by the incorporation of the electrochemically inactive M²⁺ ions. However, except for Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄, the other samples show decreased capacities, especially in the case of the Mn/Zn co-doping. Further tests indicate that the promotion of Li⁺ diffusivity may be a key reason for the improved rate and cycling performances. By contrast, the incorporation of Zn²⁺ impaired the cell performances such as increased internal polarization, hindered charge transfer, decreased Li⁺ diffusivity. In this work, the Mg²⁺ with smaller radius seems to be a better choice as the co-doping element at Fe sites than Zn²⁺.

Keywords:
chemically modified electrode; lithium battery; applied electrochemistry; composite and nanocomposite materials

Introduction

Compared with the other cathodes from the silicate family, Li₂FeSiO₄ has its own characteristics, such as: structural rearrangement, stable interface film, thermal regeneration. However, low electronic conductivity and slow lithium-ion diffusion are the biggest obstacles for the commercial application of Li₂FeSiO₄. To date, numerous approaches have been investigated to circumvent these main issues by improving the intrinsic and extrinsic properties of bulk Li₂FeSiO₄, by e.g., cation doping, carbon coating and optimized preparation technology, etc. Some nano-Li₂FeSiO₄/C composites with novel shapes, such as Li₂FeSiO₄/C/CNS,¹Hierarchical shuttle-like Li₂FeSiO₄/C,² rod like Li₂FeSiO₄³ and nano-spherical Li₂FeSiO₄/C,⁴ can approach or even exceed the theoretical capacity of 166 mAh g^-1. Especially, the Li₂FeSiO₄/C/MWCNTs (MWCNTs: multi walled carbon nanotubes) composites reported by Peng et al.⁵ had a high discharge capacity of 206.8 mA h g^-1 in the second cycle at 0.1 C, even at 20 C, the electrode could maintain 82 mAh g^-1 after 500 cycles. Moreover, the porous Li₂FeSiO₄/C nano-composites prepared by a simple P123 (EO₂₀PO₇₀EO₂₀) assisted sol-gel method, also delivered a first discharge capacity of 230 mAh g^-1 at 0.1 C corresponding to 1.39 mol Li⁺ inserted per molecular formula.⁶ These works indicate that Li₂FeSiO₄ has a lot of potential as a promising cathode material for Li-ion battery.

In the past few years, cation doping was considered as one of the most effective methods to improve electronic conductivity and ion mobility. Various doping elements have been tried to improve the performance of Li₂FeSiO₄ materials. For example, Zhang et al.⁷ studied the impact of different doping elements including Mg²⁺,⁷ Zn²⁺/Cu²⁺/Ni²⁺,⁸ Cr³⁺,⁹ on the electrochemical performance of Li₂FeSiO₄, and some improved results were observed. Among the related reports, Mn²⁺ doping is of special concern, because Mn²⁺ ion can solve into the crystal structure of Li₂FeSiO₄, and make up for the intrinsic disadvantage of low theoretical capacity (166 mAh g^-1). For example, Shao et al.¹⁰ prepared carbon-coated nanostructured Li₂Fe_0.5Mn_0.5SiO₄ via a combination of spray pyrolysis and wet ball milling method, which displayed a discharge capacity as high as 149 mAh g^-1. Different from the above optimum Fe/Mn ratio, Guo et al.¹¹ pointed out that Li₂Fe_0.9Mn_0.1SiO₄/C synthesized by mechanical activation-solid-state reaction had a distinguished discharge capacity of 158.1 mAh g^-1 in the first cycle.¹⁰ Deng et al.¹² also reported that the Li₂Fe_0.5Mn_0.5SiO₄ prepared via citric acid assisted sol-gel method showed an initial discharge capacity above 170 mAh g^-1. Especially, Li₂Fe_0.75Mn_0.25SiO₄/C reported by Wang and co-workers¹³ using a modified sol-gel methods, inserted more than 1.0 mol Li⁺per molecular formula, and showed an initial discharge capacity of 201 mAh g^-1. However, all the Mn-doped samples presented worse cycling performance than pristine materials. About this, Bini et al.¹⁴ pointed out that a significant cation disorder could top up to 80%, based on the Li/Fe(Mn) anti-site defect, which should seriously influence the structural stability of Li₂Fe_1-xMn_xSiO₄/C composites. Moreover, Chen et al.¹⁵ also found out that the amorphization and structural rearrangements of Li₂Fe_1-xMn_xSiO₄ during the initial cycle and the redox processes of Mn species could be only involved in the first few cycles. Therefore, the structural stability and the efficiency of Mn species are the main problem facing the current ferromanganese lithium silicate materials.

Recently, the introduction of a second dopant has been proposed and widely studied in order to improve the electrochemical response of cathode material. Hu et al.¹⁶ compared the properties of single doped LiMn_0.9Fe_0.1PO₄ and co-doped LiMn_0.9Fe_0.05Mg_0.05PO₄, and found that the LiMn_0.9Fe_0.05Mg_0.05PO₄ sample has much higher reversible capacity and rate capability as compared to LiMn_0.9Fe_0.1PO₄. Cui et al.¹⁶ synthesized zinc- and PO₄-doped LiMn₂O₄ via traditional solid state reaction. The discharge capacity and cycling performance of the co-doped LiZn_0.05Mn_1.95O_3.96(PO₄)_0.025 were found to be superior to single doped LiZn_0.05Mn_1.95O_3.99.

With the expectation to improve the cycling capability of Li₂Fe_1-xMn_xSiO₄ system, we choose common divalent cations (Mg²⁺ and Zn²⁺) combined with Mn²⁺ as the co-doping elements at Fe sites and prepared Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg or Zn, x = 0.05, 0.1) cathodes by a solid state reaction. Their electrochemical properties were investigated and the possible synergistic mechanisms were discussed.

Experimental

The used metal salts: LiCH₃COO.2H₂O, FeC₂O₄.2H₂O, Mn(CH₃COO)₂.4H₂O, ZnO and Mg(CH₃COO)₂.4H₂O, were purchased from Tianjin Guangfu Fine Chemical Research Institute, and nano-SiO₂ was provided from Henan Huanyu Power Source Co., Ltd. To prepare cathode materials, the stoichiometric amounts of SiO₂ and metal salts were first mixed together. To avoid the excessive impurities in the final products, only 5 wt.% mass of glucose (C₆H₁₂O₆.H₂O) as carbon source was introduced to the above mixture, and the mixture were ground to fine powder together with an agate mortar and pestle. The resulting intimate mixtures were first dehydrated at 200 °C for 1 h, and then calcined at 700 °C for 10 h in a vacuum tube furnace keeping -0.1 MPa vs. normal atmosphere. According to the ratio of metal salts in the initial precursor, the obtained final products with about 1.6 wt.% carbon (determined by high frequency infrared carbon-sulfur determination, CS 600, Germany Eltar), were named as Li₂Fe_0.8-xMn_0.2M_0.05SiO₄ or Li₂Fe_0.8-xMn_0.2M_0.1SiO₄ (M = Mg or Zn).

The crystal structures were analyzed by powder X-ray diffraction (Ultima IV, Rigaku) employing Cu K_a1 radiation (λ = 0.154056 nm) in the two-theta range of 10-80°. The morphologies and chemical distributions were observed using a field-emission scanning electron microscope (FESEM, JSM-7500F, Japan JEOL) equipped with an energy dispersive X-ray analyzer (EDX, Oxford Inca PentaFET ×3).

The mixture of Li₂Fe_0.8-xMn_0.2M_xSiO₄ powder, acetylene black and polyvinylidene fluoride with a weight ratio of 80:10:10 was used as the cathode. The dry strip cathodes were cut into the disks (16 mm in diameter) containing 6-8 mg active materials. Then, the Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li coin cells were fabricated in an Ar-filled glove box. The cycling and rate performances were characterized on a 5 V per 2 mA battery testing system (Land 2005A, Wuhan Jinnuo Electronics Co., Ltd.) with cut-off voltage of 1.5-4.5 V vs. Li⁺ /Li at a constant current from 0.06 to 2 C. The electrochemical impedance spectra (EIS) of the cycled cells were obtained using an electrochemical workstation (CHI 604D, Shanghai CH Instruments, Inc.) in a range of 0.1-100 kHz at the amplitude of 5 mV. The cyclic voltammetry (CV) was also tested with CHI 604D at a scanning rate of 1.0 mV s^-1 at room temperature. Capacity intermittent titration technique (CITT) proposed by Tang et al.¹⁸ was used to investigate the Li-ion diffusion coefficient (D_Li+). The cell was first galvanostatically-charged (5 mA g^-1) to a certain cutoff voltage, then potentiostatically-charged at this cutoff voltage until the current tends to zero. Here, the set current is 0.05 mA g^-1. The whole charge voltage ranges (1.5-4.5 V) were divided into many small voltage regions and separated from one another by 0.1 V. When the charging was completed, the tested cell was discharged with 10 mA g^-1 current to 2.0 V. The above procedures were continuously repeated for five cycles.

Results and Discussion

Figure 1 compares the X-ray diffraction (XRD) patterns of the synthesized Li₂Fe_0.8-xMn_0.2M_xSiO₄ samples. The Li₂Fe₃O₄ impurity can be easily detected, small Li₂SiO₃ also emerged in the Mn/Mg co-doped samples. The remaining diffraction peaks are very similar to each other, indicating that small amounts of co-doping elements have not changed Li₂FeSiO₄ architecture. However, the broad peaks indicate that in neither case the crystallinity is good. The quality of XRD patterns does not allow a good Rietveld refinement, but the matching results of XRD patterns, especially in low angle region, suggest that the obtained samples should be a mixed phases of P2₁/n and Pmn2₁ co-existance.¹⁹ Furthermore, the co-doping of Mg²⁺ or Zn²⁺ caused some observable variations in the diffraction peak positions. For example, the strongest diffraction peak around 33° gradually shifted toward higher 2θ values with the increase of Mg content from 0 to 0.1, which could be easily seen from the enlarged patterns located at the right side of Figure 1. Recently, Dimesso et al.²⁰ reported the influence of different Mg-ion additions on the structure of the olivine-structured LiCo_1-xMg_xPO₄. Their results showed a shift in the positions of the (311) and (121) reflections towards higher diffraction angles. By contrast, due to the similar radius of Zn²⁺ (0.74 Å) and Fe²⁺ (0.76 Å), the Mn/Zn co-doping did not cause changing in the position of the reflection at 33°/2θ, and a shoulder peak always well evident. The XRD results suggest that Mg²⁺ and Zn²⁺ have been successfully introduced into Li₂FeSiO₄ matrix structure.^10,21-23 Moreover, the XRD patterns do not show any peak related to crystal carbon, which indicates the introduced carbon should be in amorphous form in the final samples.

Figure 1
XRD patterns of Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg and Zn, x = 0, 0.05 and 0.1).

To clarify the surfacial characteristics of Li₂Fe_0.8-xMn_0.2M_xSiO₄ samples, SEM (scanning electron microscopy) observations were carried out in combination with EDX tests. As shown in Figure 2, there were no significant morphological differences between these SEM images. All samples exhibited agglomerated particles mainly arising from the employed solid-phase method. However, based on the local view, the Mn/Zn co-doping seems to produce more serious agglomerates and less uniform particle size. The further EDX (energy dispersive X-ray) spot data in Table 1 show that the atomic ratio of Fe/Mn/M/Si/O in samples was very close to the designed ratio of the Li₂Fe_0.8-xMn_0.2M_xSiO₄ formula. It demonstrates that the co-doped elements incorporate very well with Li₂Fe_0.8-xMn_0.2M_xSiO₄ host materials.

Figure 2
SEM-EDX spot images of original magnification 10,000× of Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg and Zn, x = 0.05 and 0.1).

Table 1
EDX spot data of Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg and Zn, x = 0.05 and 0.1)

Sample	Atomic / %
	Fe K	Mn K	M2+ K	Si K	O K
Li2Fe0.75Mn0.2Mg0.05SiO4	12.22	3.39	0.88	16.69	66.82
Li2Fe0.7Mn0.2Mg0.1SiO4	11.74	3.25	1.77	16.22	67.02
Li2Fe0.75Mn0.2Zn0.05SiO4	12.62	3.32	0.83	16.71	66.52
Li2Fe0.7Mn0.2Zn0.1SiO4	12.64	3.56	1.95	17.23	64.62

Although Li₂FeSiO₄ capacity could be greatly improved by Mn²⁺ doping, Li₂Fe_0.8Mn_0.2SiO₄ has been demonstrated experimentally with the 30-40% capacity loss after 50 charge-discharge cycles in many reports.^10,12,24 A similar observation can be seen in our experiments. As shown in Figure 3b, the Li₂Fe_0.8Mn_0.2SiO₄/Li cell exhibited a serious capacity loss and only maintained about 30.4% of discharge capacity at 50^th cycle (38.6 mAh g^-1), as compared to that at 1^st cycle (127 mAh g^-1). Moreover, it is necessary to point out that the present Li₂Fe_0.8Mn_0.2SiO₄ performance is lower than that in our previous report.²⁵ This is because that smaller carbon was introduced in the present series experiments in order to avoid more impurities resulting from excessive carbon, and highlight the effect of co-doping.

Figure 3
Typical initial dQ/dV and charge-discharge curves of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells (M = Mg and Zn, x = 0, 0.05 and 0.1).

Based on the charge-discharge data in Figure 3, we can summarize some key things as follow: (i) the initial charge curves of all samples are obviously higher than those in the subsequent cycles, which means that the single-doping of Mn²⁺ or the further co-doping of Mg²⁺ or Zn²⁺ do not change the inherent characteristic of structural rearrangement of Li₂FeSiO₄ material;^26,27 (ii) the initial dQ-dV curves in Figure 3a exhibit two clear voltage plateaus at about 3.5 and 4.4 V. After Mg²⁺ or Zn²⁺ co-doping, the area of oxidation peak located at about 3.5 V increases. At the same time, a shift of reduction peak located at about 2.5 V towards the right means less internal polarization. Novikova et al.²⁸ also observed similar phenomenon that the slope of the voltage plateau for the co-doped samples is lower than that of single-doped sample, and they attribute it to the improved ionic conductivity; (iii) the initial discharge capacities of the co-doped samples are lower than that of 127 mAh g^-1 for the single-doped sample. For example, in the case of the Mn/Mg co-doped samples, when x is 0.05 and 0.1, the initial discharge capacities are 118 and 108.5 mAh g^-1, respectively. As for the Mn/Zn co-doped samples, lower discharge capacities of 96.8 and 79.6 mAh g^-1 are observed. This result can be interpreted as being due to the fact that the doped Mg²⁺ or Zn²⁺ in the crystal lattice cannot be oxidized to M³⁺ or M⁴⁺, they usually stay electrochemically inactive.²⁹ The fact that the more M²⁺ co-doping causes the lower discharge capacities is in agreement with the above inference. (iv) Cycling performances have been greatly improved by co-doping. For example, Li₂Fe_0.8-xMn_0.2Mg_xSiO₄ samples show the discharge capacities of 74.9 and 63.2 mAh g^-1 after 50 cycles, when x is 0.05 and 0.1, and the corresponding capacity retentions are 63.5 and 58.2%, as compared to their initial discharge capacities, respectively. Similarly, capacity retention is 53.6 and 69.5% for Li₂Fe_0.8-xMn_0.2Zn_xSiO₄ samples, in spite of their lower discharge capacities of 51.9 and 55.3 mAh g^-1 at 50^th cycle. We think that the inactive M²⁺ could have the buttressing effect on crystal lattice when Li-ions are extracted from host cathodes.³⁰ Therefore, the co-doped materials have more resistance to amorphisation than Li₂Fe_0.8Mn_0.2SiO₄. However, along this inference, Li₂Fe_0.8-xMn_0.2Mg_xSiO₄ samples show the opposite result. The capacity retention of x = 0.1 is only 58.2%, less than 63.5% of x = 0.05. A similar phenomenon was explained by Huang et al.³¹ that the existence of too many doping ions in the crystal lattice may induce large changes to the structure and cause the phase instability. (v) Among all samples, Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄ shows the most excellent performance in charge-discharge capacity and cycling property.

Due to low Li⁺ diffusivity and poor electronic conductivity of silicate cathodes, Li₂MSiO₄ materials usually have a major drawback to their capacity performances, especially at high rates. As shown in Figure 4, all Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells present poor rate performances in the current range from 0.25 to 2 C. Among these samples, only Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄/Li cell has an improved rate performance, whose discharge capacity is 78.1, 69.7, 58.2, 49.8, 40.7 and 35.6 mAh g^-1 at the current rates of 0.25, 0.375, 0.5, 0.75, 1.0, 2.0 C, respectively. However, when Mg²⁺ content increases to 0.1, the sample has a similar performance to that of Li₂Fe_0.8Mn_0.2SiO₄. By contrast, both the Mn/Zn co-doped samples presented worse rate performances, especially Li₂Fe_0.7Mn_0.2Zn_0.1SiO₄ even does not work at 2.0 C.

Figure 4
Rate capabilities of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells (M = Mg and Zn, x = 0, 0.05 and 0.1).

Figure 5 depicts the CV curves of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells, conducted at a scan rate of 1.0 mV s^-1 to elucidate the redox chemistry in a better way. All the CV profiles are similar. A strong reduction peak with good repeatability appears at about 2.4 V in the cathodic scan, while the anodic scan exhibits two broad oxidation peaks, in which the lower voltage oxidation peak (LOP) at about 3.4 V can be attributed to the dominating Fe²⁺/Fe³⁺ redox process on cycling,^6,32 it is located at a slightly higher potential (3.6-3.8 V) in the first cycle. In view of Wang et al.³³ the CV curves of Li₂MnSiO₄/Li cells showed two oxidation peaks located at 4.4 and 4.8 V corresponding to the Mn²⁺/Mn³⁺ and Mn³⁺/Mn⁴⁺ redox couples during the initial charge, respectively. Then they moved obviously to the lower voltages in the subsequent cycles. Therefore, the higher voltage oxidation peak (HOP, > 4.2 V) in our experiment can be mainly associated with the Mn²⁺/Mn³⁺ redox couple except for the initial charge. Possibly, there may be a few Fe³⁺/Fe⁴⁺ reactions occurred at local crystal structure.

Figure 5
The first five CV profiles (a-e) and mean peak separation; (f) of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells (M = Mg and Zn, x = 0, 0.05 and 0.1).

In the case of Li₂Fe_0.8Mn_0.2SiO₄, the oxidation peak (> 4.2 V) of the second scan is significantly lower than that of the first cycle, as shown in Figure 5a. This observation just matches up with the conclusion suggested by Chen et al.¹⁵ that the redox processes of Mn species only can be involved in the first few charge-discharge cycles. As for the Mn/Mg co-doped samples (Figures 5b and 5c), we can notice three changes: (i) during the first lithium de-intercalation process, the Mn/Mg co-doped samples have lower first LOP than the other samples. When x = 0.05, LOP presents a significant peak at 3.3 V accompanied with a weak transition peak at 3.7 V. However, when x = 0.1, the LOP always stays at about 3.35 V. By contrast, the corresponding LOP locate at 3.75 V for Li₂Fe_0.8Mn_0.2SiO₄, 3.8 and 3.65 V for the Mn/Zn co-doped samples, respectively. This fact indicates that, the Mg-substituted system has a smaller inner polarization, which usually means an easier lithium de-intercalation from host cathode. Here, the lower HOP suggests that Mg²⁺ doping could reduce the energy barrier for lithium migration; (ii) the intensity of HOP (> 4.2 V) only showed a slight decline with increasing cycles, suggesting that the incorporation of Mg²⁺ may be better to improve the oxidation from M³⁺ to M⁴⁺; (iii) CV profiles of Li₂Fe_0.8-xMn_0.2Mg_xSiO₄ samples have a better reproducibility in the test range of 1^st-5^th cycle, implying the better electrode stability. By contrast, the Mn/Zn co-doped samples have extremely similar CV characteristics to Li₂Fe_0.8Mn_0.2SiO₄ sample except for some slight differences in redox peak position and response current intensity. Notice that the HOP intensity (> 4.2 V) is smaller and decreases rapidly; this suggests that the incorporation of Zn²⁺ could impair the oxidation from M³⁺ to M⁴⁺.

Then, we calculated the intervals between the oxidation peak and the reduction peak during the first five CV cycles, which have been compared in Figure 5f. Because the oxidation peaks of HOP (> 4.2 V) is incomplete, limited by the voltage upper limit of 4.5 V, all the data come from the oxidation peak around 3.4 V and the reduction peak around 2.4 V. Moreover, the mean peak separation were used in the 2^nd-5^th CV cycles, the single difference was used at the first CV when considering structural rearrangement. This indicates that the introduction of Mg²⁺ can help to lessen the inner polarization during the initial structural rearrangement process. In the subsequent CV (2^nd-5^th), the single doped sample (Li₂Fe_0.8Mn_0.2SiO₄) presents the smallest value (0.7748 V), while the peak intervals after M²⁺ co-doping have different growth. Especially, in either case, the Mn/Zn co-doped samples show noticeable increases in polarization. The fact that more M²⁺ doping leads to the greater polarization suggests that the incorporation of electrochemically inactive M²⁺ ions could impede the extraction of Li-ions at LOP to some extent.²⁹ This result seems to be contradictory with cycling performances, because smaller resistance usually means better cell performance. However, note that CV data only present the mean peak intervals of 2^nd-5^th scan, thus taking them as evidence to explain long-cycling performance seems to be insufficient. In addition, the absence of HOP (> 4.2 V) might also bring certain errors in CV results.

Figure 6 shows XRD patterns and EIS spectra of the cycled Li₂Fe_0.8-xMn_0.2M_xSiO₄ samples. In Figure 6a, all the XRD patterns are similar and appear no sharp peaks. This proved a transformation from crystal structure into amorphous phase occurred after 50 cycles. Different from prior CV results presented in the first few cycles, the co-doped samples did not show a long-life advantage in structural stability. The Nyquist plots shown in Figure 6b are composed of a semicircle at middle frequency and a slope line at low frequency. The sloped line is related to the Li⁺ diffusion in the electrode material, and the semicircle is connected with the interfacial impedance (R_i) including charge-transfer impedance (R_ct) and solid electrolyte interphase impedance (R_SEI).³⁴ The spectra can be fitted by the equivalent circuit in our previous work.²⁵ The obtained R_i values were 278 Ω for Li₂Fe_0.8Mn_0.2SiO₄, 289 and 404 Ω for the Mg co-doped samples, 626 and 667 Ω for the Zn co-doped samples, respectively. This result indicates that the co-dopping lead to bigger R_i values, especially for the incorporation of Zn²⁺. Therefore, the above XRD and EIS results cannot give sufficient reasons for the improved performance of Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄. To further reveal this issue, we carried out CITT measurements of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells.

Figure 6
XRD patterns (a) of Li₂Fe_0.8-xMn_0.2M_xSiO₄ after 50 cycles; (b) EIS spectra of the cycled cells.

The CITT technique was used to determine D_Li+ values. Figure 7 compares the obtained CITT curves with various stair-step shapes. It can be seen that the voltage increases gradually in galvano-charge step and the current decreases gradually in the potentio-charge step. According to the reports,³⁵ the following equation can be derived from the spherical diffusion model.

(1)

Figure 7
The first two CITT curves of Li₂Fe_0.8-xMn_0.2M_xSiO₄/Li cells (M = Mg and Zn, x = 0, 0.05 and 0.1)

where q is the ratio of Q_p and Q_g, t_G is the galvano-charge time (s), r is the mean crystallite size, α_j is constant. Therefore, the equation gives the analytical solution associating q value with the D_Li+. We can get the desired data from the CITT curves in Figure 7, and then select the equations fitted by linear least-square fits according to for different ranges of q value to calculate D_Li+ values.³⁶

The obtained D_Li+ values are compared in Figure 8. It can be seen that the D_Li+ values varied non-linearly with the charge voltage as a wavy shape, all the D_Li+ values are almost in the same order of magnitude (10^-16). Although the curve shapes are similar, the D_Li+ values of Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄ in the first cycle are higher than that of Li₂Fe_0.8Mn_0.2SiO₄ sample, especially when the voltage is higher than 4.2 V. This superiority is still retained during the second charging, in spite of the reversed results in the range of 2.8-4.2 V. The higher D_Li+ values mean better Li diffusivity, so it proves that the incorporation of Mg²⁺ can promote the extraction/insertion of Li⁺ in the corresponding range. Also, note that the most D_Li+ values at the same voltage in the second charging are higher than that in the first charging. However, for the Li₂Fe_0.75Mn_0.2Mg_0.05SiO₄ sample, the obtained D_Li+ data in the first and second cycle are comparable. Recently, Araujo et al.³⁷ pointed out that the D_Li+ values of Li₂FeSiO₄ at room temperature was only about 10^-20-10^-17 cm² s^-1, less than other with cathode materials (10^-13-10^-7 cm² s^-1), therefore the kinetics of diffusion performance need to be improved. In our experiment, the Mg²⁺ co-doping showed some positive effects on Li⁺ mobility, especially above 4.2 V range, this may be one reason for its improved cell performances. It is need to point out that, because this study only compares the D_Li+ values of the first two charging, so the corresponding data might not reflect the real potential of electrodes for their long cycling.

Figure 8
The D_Li+-voltage curves determined from CITT data.

Conclusions

In this study, Li₂Fe_0.8-xMn_0.2M_xSiO₄ (M = Mg and Zn, x = 0, 0.05 and 0.1) show very distinct cell performances. Firstly, the incorporation of the electrochemically inactive M²⁺ ions can serve to strengthen crystal structure and improve the cycling performance significantly. However, note that the capacities show a serious loss with cycling for the Mn/Zn co-doped samples. Secondly, the Mg²⁺ co-doping can bring more benefits, such as the increased capacity (x = 0.05) and the improved rate performance. Thirdly, the XRD and EIS results showed that co-doping leads to the enlarged interfacial resistance and amorphous crystal structure after 50 cycles, but the CV and CITT results suggest that the Mg²⁺ co-doping should improve Li-ion diffusion behavior at least in the first several cycles, especially above 4.2 V range. In conclusion, by comparing the Mg²⁺ and Zn²⁺ co-doped samples, the Mg²⁺ with smaller radius seems to be a better choice in this work. The study on the Fe-site co-doping of Li₂FeSiO₄ including more elements will be performed in the future.

Acknowledgments

This work was financially supported by National Natural Science Foundation of China (NSFC) (No. 21003087) and Natural Science Foundation of Shanxi Province (No. 2014011017-4) and National Student Innovation Training Program of China (No. 201510118002) and Students Innovation Training Program of Shanxi Province (No. 2015152).

References

• ¹
  Yang, J. L.; Kang, X. C.; Hu, L.; Gong, X.; He, D. P.; Peng, T.; Mu, S. C.; J. Alloys Compd. 2013, 572, 158.
• ²
  Yang, J. L.; Kang, X. C.; He, D. P.; Peng, T.; Hu, L.; Mu, S. C.; J. Alloys Compd. 2013, 242, 171.
• ³
  Devaraju, M. K.; Tomai, T.; Honma, I.; Electrochim. Acta 2013, 109, 75.
• ⁴
  Qu, L.; Fang, S. H.; Zhang, Z. X.; Yang, L.; Hirano, S. I.; Mater. Lett. 2013, 108, 1.
• ⁵
  Peng, G.; Zhang, L. L.; Yang, X. L.; Duan, S.; Liang, G.; Huang, Y. H.; J. Alloys Compd. 2013, 570, 1.
• ⁶
  Wu, X. Z.; Jiang, X.; Huo, Q. S.; Zhang, Y. X.; Electrochim. Acta 2012, 80, 50.
• ⁷
  Zhang, S.; Deng, C.; Fu, B. L.; Yang, S. Y.; Ma, L.; J. Electroanal. Chem. 2010, 644, 150.
• ⁸
  Deng, C.; Zhang, S.; Yang, S. Y.; Fu, B. L.; Ma, L.; J. Power Sources 2011, 196, 386.
• ⁹
  Zhang, S.; Deng, C.; Fu, B. L.; Yang, S. Y.; Ma, L.; Electrochim. Acta 2010, 55, 8482.
• ¹⁰
  Shao, B.; Abe, Y.; Taniguchi, I.; Powder Technol. 2013, 235, 1.
• ¹¹
  Guo, H. J.; Cao, X.; Li, X. Q.; Li, L. M.; Li, X. H.; Wang, Z. X.; Peng, W. J.; Li, Q. H.; Electrochim. Acta 2010, 55, 8036.
• ¹²
  Deng, C.; Zhang, S.; Yang, S. Y.; J. Alloys Compd. 2009, 487, L18.
• ¹³
  He, X. G.; Yang, G. L.; Sun, L. Q.; Xie, H. M.; Wang, R. S.; Chem. J. Chinese U. 2010, 31, 2148.
• ¹⁴
  Bini, M.; Ferrarin, S.; Capsoni, D.; Spreafico, C.; Tealdi, C.; Mustarelli, P.; J. Solid State Chem. 2013, 200, 70.
• ¹⁵
  Chen, R. C.; Heinzmann, R.; Mangold, S.; Chakravadhanula, V. S. K.; Hahn, H.; Indris, S.; J. Phys. Chem. C 2013, 117, 884.
• ¹⁶
  Hu, C.; Yi, H.; Fang, H.; Yang, B.; Yao, Y.; Ma, W.; Dai, Y.; Electrochem. Commun. 2010, 12, 1784.
• ¹⁷
  Cui, P.; Liang, Y. S.; Solid State Ionics 2013, 249-250, 129.
• ¹⁸
  Tang, X. C.; Li, L. Q.; Huang, B. Y.; He, Y. H.; Solid State Ionics 2006, 177, 687.
• ¹⁹
  Sirisopanaporn, C.; Dominko, R.; Masquelier, C.; Armstrong, A. R.; Maliad, G.; Bruce, P. G.; J. Mater. Chem. 2011, 21, 17823.
• ²⁰
  Dimesso, L.; Spanheimer, C.; Jaegermann, W.; J. Alloys Compd. 2014, 582, 69.
• ²¹
  Dong, Y. Z.; Zhao, Y. M.; Duan, H.; J. Electroanal. Chem. 2011, 660, 14.
• ²²
  Dimesso, L.; Spanheimer, C.; Jaegermann, W.; J. Power Sources 2013, 243, 668.
• ²³
  Muraliganth, T.; Manthiram, A.; J. Phys. Chem. C 2010, 114, 15530.
• ²⁴
  Gong, Z. L.; Li, Y. X.; Yang, Y.; Electrochem. Solid-State Lett. 2006, 9, A542.
• ²⁵
  Gao, K.; Ionics 2014, 20, 809.
• ²⁶
  Liivat, A.; Solid State Ionics 2012, 228, 19.
• ²⁷
  Eames, C.; Armstrong, A. R.; Bruce, P. G.; Islam, M. S.; Chem. Mater. 2012, 24, 2155.
• ²⁸
  Novikova, S.; Yaroslavtsev, S.; Rusakov, V.; Kulova, T.; Skundin, A.; Yaroslavtsev, A.; Electrochim. Acta 2014, 122, 180.
• ²⁹
  Reddy, M. V.; Jie, T. W.; Jafta, C. J.; Ozoemena, K. I.; Mathed, M. K.; Nairf, A. S.; Peng, S. S.; Idris, M. S.; Balakrishna, G.; Ezema, F. I.; Chowdari, B. V. R.; Electrochim. Acta 2014, 128, 192.
• ³⁰
  Singh, G.; Gupta, S. L.; Prasad, R.; Auluck, S.; Gupta, R.; Sil, A.; J. Phys. Chem. Solids 2009, 70, 1200.
• ³¹
  Huang, J. S.; Yang, L.; Liu, K. Y.; Tang, Y. F.; J. Power Sources 2010, 195, 5013.
• ³²
  Singh, S.; Mitra, S.; Electrochim. Acta 2014, 123, 378.
• ³³
  Wang, F.; Wang, Y. M.; Sun, D. M.; Wang, L.; Yang, J.; Jia, H. P.; Electrochim. Acta 2014, 119, 131.
• ³⁴
  Nytén, A.; Stjerndahl, M.; Rensmo, H.; Siegbahn, H.; Armand, M.; Gustafsson, T.; Edstrom, K.; Thomas, J. O.; J. Mater. Chem. 2006, 16, 3483.
• ³⁵
  Tang, X. C.; Pan, C. Y.; He, L. P.; Li, L. Q.; Chen, Z. Z.; Electrochim. Acta 2004, 49, 3113.
• ³⁶
  Tang, X. C.; Song, X. W.; Shen, P. Z.; Jia, D. Z.; Electrochim. Acta 2005, 50, 5581.
• ³⁷
  Araujo, R. B.; Scheicher, R. H.; de Almeida , J. S.; Ferreira da Silva, A.; Ahuja, R.; Solid State Commun. 2013, 173, 9.

Publication Dates

• Publication in this collection
  Nov 2016

History

• Received
  07 Jan 2016
• Accepted
  23 Mar 2016