Morphology evolution and impurity analysis of LiFePO₄ nanoparticles via a solvothermal synthesis process

Abstract

A solvothermal method is applied for synthesizing LiFePO₄ nanoparticles using ethylene glycol as solvent. Crystals are obtained with quite different morphologies at solutions of various acidity prepared via changing the primary LiOH/H₃PO₄ mole ratios. SEM, TEM, and HRTEM are used to analyze the samples. Element distribution in solid LiFePO₄ particles, mother solutions and washing solutions are tracked by ICP-OES and pH tests. Morphological test results show that the main exposed faces of samples transform from (100) as a rectangular shape to (010) as a spindle shape with the pH of the mother solutions increasing. Samples with predominant (010) faces are formed at less acidic solvothermal solutions. At the intermediate pH from 3.11 to 3.73, powders like long hexagon nanorods are synthesized with (100) and (010) faces exposed. XRD results show that the long hexagon nanorods have better crystal structures when synthesized at LiOH/H₃PO₄ = 2.7–3.0. Impurities like Fe₃O₄, Li₃PO₄, etc. are detected in the spindle shape LiFePO₄ powders. The amount of impurities is related to the synthesis process and increases with the pH of solvothermal solution increasing. High temperature treatment is useful for impurities transforming to LiFePO₄ and thus reduces the impurities. The long hexagon nanorods show better electrochemical performances: 169.9 mA h g⁻¹ at 0.1 C, and 129.8 mA h g⁻¹ at 10 C.

---

1. Introduction

Lithium iron phosphate (LiFePO₄) is one of the most promising cathode materials for lithium ion batteries for its high electrochemical performances, low cost and good stability.^1–6 However, compared with layered lithium transition metal oxides (LiMO₂ (M = Co, Mn, Ni)) and spinel lithium manganese oxide (LiMn₂O₄), LiFePO₄ has a rather low electron conductivity and lithium ion diffusion constant because of its one-dimensional lithium ion transfer channel.^7–9 Therefore, efforts like carbon coating,^10–12 supervalent doping,^6,13–15 nano-size and morphology tailoring^16–20 have been applied for solving this problem. Solvothermal (including hydrothermal) process has been researched a lot as an effective method to produce LiFePO₄ crystals which occurs a lower temperature, cheaper and softer chemistry environment compared with solid state synthesis.²¹ Synthesis procedure,²² solvent,^23,24 reaction temperature,²⁵ reaction time,²⁶ raw material ratio,^27–30 pH of the reaction solution,^31–34 reactant concentration^35,36 and source³⁷ are important in determining crystal habit, crystal composition, crystal nucleation and growth rate during solvothermal process. Ethylene glycol (EG) has been proved as an optimal solvent for solvothermal synthesizing nano size LiFePO₄ particles with lower defect concentration.^28,38

Nevertheless, during solvothermal process elements like Li, Fe and P et al. distributes in mother solutions, washing solutions and solid products which is different from that by solid state synthesis method. It is difficult for researchers to control element compositions of lithium iron phosphate by alternating Li, Fe and P ratios in precursors.^39–41 Composition range is small for nonstoichiometric lithium iron phosphate to stay as a homogeneous solid solution at ambient temperature. Phase separation and impurities' formation usually occur as Li : Fe : P deviates from 1 : 1 : 1.^20,42–44 Till now, there is no report on element distributions in solid phase products, mother solutions and washing solutions during solvothermal synthesis LiFePO₄ process. Reports on element distributions are mostly about element ratios of LiFePO₄ solid particles for those via hydrothermal methods^32,45–49 and solvothermal process.²⁹ Systemically investigations of element behaviour are essential for better understanding of reaction mechanisms and influence parameters of solvothermal synthesis LiFePO₄.

This paper reported several LiFePO₄ particles with different morphologies synthesized at different acidity (pH) solutions using EG as solvent. Element distributions were analyzed in solid products, mother solutions and washing solutions during the whole process. It's worth to mention that the orientation of LiFePO₄ particles changes from (100) to (010) as pH of mother solution increases from 2.56 to 5.80. The impurities also increased as pH increases. The impurity formation process was illustrated in details. High temperature treatment is useful not only for defect elimination but for impurity abatement.

2 Experimental

2.1 Synthesis

All the chemicals (AR grade) were purchased from Xilong Chemical Co., Ltd. Several samples were prepared by a solvothermal method using EG as solvent. The synthesis of LiFePO₄ was carried out in a 50 mL Teflon vessel, which was sealed in a stainless-steel autoclave. LiOH·H₂O, FeSO₄·7H₂O and H₃PO₄ (85 wt%) were chosen as Li, Fe, P sources. 7 mmol FeSO₄·7H₂O was used in this process (which was set as 1 in the following Table 1). The amount of LiOH·H₂O and H₃PO₄ was added just as calculated by Li : Fe : P mole ratio. The typical feeding sequence was chosen: (P → Fe) → Li. H₃PO₄ was mixed with FeSO₄'s EG solution firstly and then this mixture was dripped into LiOH's EG solution to make green black slurry. Then the slurry was transferred into a Teflon vessel. The prepared Teflon vessel was then heated at 180 °C for 10 h. After the solvothermal reactions finished, the supernatants were collected and diluted to one fifth with deionized water and the acidity (pH) of the solutions were measured. pH results in our work were all tested by this method. The obtained precipitates were washed with deionized water and ethanol for several times and then dried at 60 °C for over 6 h. In order to obtain carbon-coated LiFePO₄/C powder with good electronic conductivity, the samples were mixed with 10 wt% sucrose and sintered at 650 °C for 5 h under argon atmosphere. The product LiFePO₄/C powder has a ∼4% carbon content.

Table 1 Samples synthesized at different LiOH/H₃PO₄ mole ratios

Sample	Li : Fe : P (mole ratio)	LiOH/H3PO4	Shape
S1	2.7 : 1 : 1.5	1.8	Rectangular nanoplate
S2	2.4 : 1 : 1	2.4	Rectangular nanoplate
S3	2.7 : 1 : 1	2.7	Long hexagon nanorod
S4	3.0 : 1 : 1	3.0	Long hexagon nanorod
S5	3.1 : 1 : 1	3.1	Spindle plate
S6	3.15 : 1 : 1	3.15	Spindle plate

---

2.2 Characterization

The morphologies of the samples were observed using scanning electronic spectroscopy (SEM, JSM-5600LV, JEOL, Japan) at 20.0 kV. Transmission electron microscope (TEM) imaging was performed using a JEM-2011 electron microscope operated at 200 kV. The X-ray diffraction (XRD) patterns of samples were recorded on a Rigaku D/Max 2500 diffractometer operated at 40 kV voltage and a 200 mA current with Cu Kα radiation (λ = 1.5418 Å). The inductively coupled plasma optical emission spectrometry (ICP-OES) of samples was performed on n IRIS Intrepid II XSP (Thermo Fisher). Working parameters: RF Power: 1150 W, Nebulizer Flow: 26.0 PSI, Auxiliary gas: 1.0 LPM. The pH values of the supernatants were measured by a microprocessor pH meter (PHS-25, Shanghai).

2.3 Electrochemical measurements

The electrochemical performances were measured by using a 2032-type coin cell. The electrode was prepared by mixing a mixture of active materials, conductive graphite, acetylene black, and polytetrafluoroethene (PTFE) binder in a mass ratio of 6 : 2 : 1 : 1. Pure metallic lithium was used as anode. The electrolyte was 1 M LiPF6 dissolved in volume ratio of 1 : 1 : 1 with ethylene carbonate–dimethyl carbonate–ethylmethyl carbonate. The cells were assembled in an argon-filled glove box. Galvanostatical charging–discharging tests were carried out on a Land CT2001A cycler (Wuhan Kingnuo Electronic Co.) in a voltage range 2.5–4.2 V.

3 Results and discussion

3.1 Morphology evolution and crystal structure analysis

The LiFePO₄ powders were prepared via a solvothermal synthesis process by changing primary LiOH/H₃PO₄ mole ratio (i.e., changing reaction acidity). Fig. 1 shows the shape and size of the LiFePO₄ samples obtained by SEM. It shows well dispersed and uniform particles in Fig. 1. From Fig. 1(a) and (b), we can get that S1 and S2 particles are rectangular nanoplates with a large surface of 150 nm × 250 nm and thickness of 25 nm. There are few differences between S1 and S2. The samples S3 and S4 show quite different morphologies, which are about 40 nm × 50 nm × 90 nm and 40 nm × 60 nm × 110 nm nanorods with a large surface like long hexagon or rectangular. S4 nanorods have almost the same morphology with S3. While compared with S1–S4 nanoparticles, S5 and S6 crystals show larger sizes and a spindle like plate shape. The spindle plate grows larger when we increased the LiOH/H₃PO₄ mole ratio from 3.1 to 3.15. The width increased from 150 nm to 190 nm; the length increased from 300 nm to 350 nm; and the thickness increased from 60 nm to 80 nm. Orientation is an important parameter for LiFePO₄ as we discussed for its one dimensional lithium ion channel. We have done TEM examinations to determine the main exposed facets of these crystals with different morphologies shown in Fig. 2. Fig. 2(a) shows sample S1's rectangular surface details. According to its corresponding fast Fourier transformation (FFT) image and lattice distances, the S1 sample has predominantly (100) face exposed. In Fig. 1(c) and (d) there are two kinds of facets exposed of S3: rectangular and long hexagon. Fig. 2(b-1) shows the details of long hexagon faces. As marked in Fig. 2(b-1) (middle and right), this face is determined as (010) face. While the rectangular face in Fig. 2(b-2) is shown the same (100) face. It can be determined that there is a thin amorphous layer on the surface of the sample. For sample S5, the exposed main surface is the spindle (010) face as shown in Fig. 2(c). There are small particles on the surface of the big spindle crystal, which can also be observed in Fig. 1(e) and (f). Results in Fig. 1 and 2 show that the predominantly exposed faces of samples S1–S6 transform (100) faces to (010) faces as LiOH/H₃PO₄ increases from 1.8 to 3.15.

Fig. 1 SEM images of (a) S1, rectangular nanoplates; (b) S2, rectangular nanoplates; (c) S3, long hexagon nanorods; (d) S4, long hexagon nanorods; (e) S5, spindle plates; (f) S6, spindle plates.

Fig. 2 TEM (left), HRTEM (middle) and their corresponding FFT (right) images of (a) S1, rectangular nanoplates (exposed (100) face); (b) S3, long hexagon nanorods: (b-1) long hexagon shape shows the exposed (010) face, (b-2) rectangular shape shows the exposed (100) face; (c) S5, spindle plates shown the predominant face (010).

The thickness of crystals along [010] direction is detected using X-ray powder line-profile fitting based on the Scherrer function by scanning the (211/020) peak of 2θ range 28.6–31.0° (details are shown in ESI† part 2).^50–53 The results data for the six samples are 62.5 nm, 51.1 nm, 29.0 nm, 36.9 nm, 30.0 nm, 34.3 nm for S1–S6 individually which is different from those observed by SEM and TEM (150 nm, 150 nm, 40 nm, 50 nm, 80 nm, 80 nm for S1–S6). According to the measurement principles of XRD, the thickness of [010] direction is in fact the coherence length along [010] direction. The reasons for the differences might be defects or crystal distortions in LiFePO₄ crystals.⁵⁴ Fig. 3 shows the X-ray diffraction pattern of samples S1–S6 observed from solvothermal synthesis. All of them can be indexed to the standard pattern of orthorhombic olivine type LiFePO₄ (space group Pnma, JCPDS Card no. 83-2092). While for samples S5 and S6 with LiOH/H₃PO₄ = 3.1 and 3.15 (mole ratio) individually, the impurity iron diiorn(III) oxide Fe₃O₄ (space group Fd3̄m, JCPDS Card no. 227) is also indexed beside LiFePO₄ phase over 2θ range 60.9° to 63°. This impurity might refer to the small particles on the surface of the spindle plate shape crystals S5 and S6. For samples S1–S4 there are no evident impurities detected before the high temperature carbon coating. The detailed discussions about impurities are shown later.

Fig. 3 XRD patterns of the six LiFePO₄ crystals (S1–S6) obtained via solvothermal treatment. S1–S4 are well indexed to olivine LiFePO₄ patterns, while S5 and S6 samples show obvious impurity Fe₃O₄ patterns.

The crystal parameter changes with LiOH/H₃PO₄ mole ratio are also tracked by doing Rietveld refinements on XRD patterns of S1–S6 LiFePO₄ samples, shown in Table 2. Before carbon coated, the cell parameters a, b and cell volume V climb up and then decline with LiOH/H₃PO₄ increasing from 1.8 to 3.15, while c increases first then decreases. S3 and S4 with LiOH/H₃PO₄ mole ratio 2.7 and 3.0 crystallized the best before high temperature treatment. When samples are carbon coated at high temperature in Ar atmosphere, cell parameters a, b and V increase a lot and c decreases. The cell volumes of all samples have increased after high temperature carbon coated for better crystallization. The cell volume differences between samples before and after high temperature carbon coated are quite different. The value of that is smaller for samples S3 and S4 with LiOH/H₃PO₄ mole ratio 2.7 and 3.0.

Table 2 Lattice parameters a, b, c and V (cell volume) of LiFePO₄ particles with and without carbon coated prepared at different LiOH/H₃PO₄ mole ratios. The ΔV shows the smallest value of samples S3 and S4

Samples	Without carbon coated				With carbon coated				ΔV^a
	a	b	c	V	a	b	c	V
a ΔV is the cell volume difference between cells with and without carbon coated.
S1	10.3000	5.9859	4.6963	289.5498	10.3155	5.9986	4.6916	290.3095	0.76
S2	10.3016	5.9865	4.6973	289.6849	10.3108	5.9964	4.6943	290.2391	0.55
S3	10.3071	5.9918	4.6982	290.1524	10.3121	5.9999	4.6929	290.3574	0.20
S4	10.3064	5.9953	4.6949	290.1012	10.3131	6.0032	4.6909	290.4234	0.32
S5	10.3029	5.9848	4.7014	289.8946	10.3211	6.0014	4.6932	290.7008	0.81
S6	10.3011	5.9812	4.7012	289.6579	10.3243	6.0029	4.6966	291.0757	1.42

---

There are free ions like H⁺, H_xPO₄^x−3, FeH_yPO₄^y−1, Fe²⁺, Fe(OH)_z^2−z, etc. and complexes like Fe²⁺–EG, Fe²⁺–SO₄²⁻, LiFePO₄ crystal surface–EG, LiFePO₄ surface–SO₄²⁻, etc in the solvothermal reaction system. It has been approved the LiFePO₄ solvothermal (including hydrothermal) process is a dissolution-recrystallization process.²⁶ The upper free ions and complexes' concentration will have a big influence on LiFePO₄ crystal growth. However, the free ions' concentration and complexes' concentration are decided by solution acidity (pH) of the whole solvothermal process. Meanwhile, the precipitates like Li₃PO₄, Fe₃(PO₄)₂, and Fe(OH)₂'s dissolution process are also influenced by solution acidity (pH). Therefore, solution acidity (pH) is an important parameter that influences the whole LiFePO₄ solvothermal process. Here we can find the influences on LiFePO₄ crystal habit on Fig. 1 and 2. We tested the acidity (pH) of the supernatant of samples S1–S6 via solvothermal process by changing with LiOH/H₃PO₄ mole ratio. The results are 2.59, 2.78, 3.11, 3.73, 5.80, 6.55 for S1–S6 individually. The predominantly exposed faces of LiFePO₄ samples change from (100) face to (010) face with the solution acidity decreased (pH increased) shown in Fig. 4. To confirm the acidity (pH) influences on the morphology revolution, further experiments are done as that Li₂SO₄ is used as part of Li resource instead of LiOH for preparing LiFePO₄ particles. We use 2.7 : 0.25 : 1 : 1 (2.7-LiOH, 0.25-Li₂SO₄, 1-FeSO₄, and 1-H₃PO₄) as reactants ratio and get almost the same long hexagon nanorods as that by 2.7 : 1 : 1 (2.7-LiOH) instead of spindle plates made by 3.2 : 1 : 1 (3.2-LiOH). And we also checked the influence of feeding sequences of the Li, Fe and P resources on LiFePO₄ crystal morphology. Feeding sequence Fe → (P → Li) was applied to produce samples. H₃PO₄ was added to LiOH's solution first to form a white suspension, and then FeSO₄'s EG solution was dripped inside. Three samples with LiOH/H₃PO₄ = 1.8, 3.0 and 3.15 individually were produced. The morphology variation of samples (as shown in Fig. S1 and S2†) was almost the same as that of samples made by (P → Fe) → Li. As reported by Wang's work,⁵⁵ surface (100), (010) and (101) are lowest energy faces which prefer to be kept during LiFePO₄ crystal process. And naked faces like (100) and (010) faces include a lot of unsaturated coordinated Li and Fe and naked O. Then these naked faces will absorb H⁺, SO₄²⁻, Fe²⁺, OH⁻, EG, HO–CH₂–CH₂–O⁻ or ⁻O–CH₂–CH₂–O⁻, etc. to stabilize themselves. For the surface energy is changed by the absorption behaviour. Therefore the crystal habit is changed. There is an isoelectric point of LiFePO₄ particle in EG solvent which is always described as a pH value. Here we set the isoelectric point as i. When the solution pH < i, the net residual electric charge on the surface of the LiFePO₄ particles is expected to be positive, thus the absorption of cations is weakened. Faces are covered by different kinds of OH⁻, SO₄²⁻, HO–CH₂–CH₂–O⁻ or ⁻O–CH₂–CH₂–O⁻ ions and complexes selectively. While when pH > i, the particle surface is negatively charged. Particle surfaces will be capped by EG molecules. As reported by first principle calculations and hydrothermal experiments,^56,57 water-capped LiFePO₄ particles favour to form exposed (010) face. It is reasonable to hypothesis that EG-covered LiFePO₄ crystals have the (010) predominantly exposed face. High pH value creates a high supersaturating level which results in a fast nucleation process generating precursor precipitates. The saturated precursor is difficult to dissolution at higher pH value. The LiFePO₄ crystal growth kinetics is controlled by diffusion and surface reaction for this dissolution-recrystallization process. Thus the growth kinetics of LiFePO₄ crystals are controlled by dissolution process of precursors which resulted in larger crystal size. The whole process is schemed in Fig. 5.

Fig. 4 The particle morphology evolution with the acidity (pH) of solvothermal synthesis environment. Rectangular nanoplates with main exposed (100) face transfer to spindle plates predominantly exposed (010) face gradually as pH of mother solutions increases from 2.56 to 5.80.

Fig. 5 Schematic illustration of pH influences on LiFePO₄ solvothermal process. i is isoelectric point which refers to a certain pH value during the solvothermal process. when the solution pH < i, the net residual electric charge on the surface of the LiFePO₄ particles is expected to be positive, and faces attract different kinds of OH⁻, SO₄²⁻, HO–CH₂–CH₂–O⁻ or ⁻O–CH₂–CH₂–O⁻ negative ions and complexes; when pH > i, the particle surface is negatively charged. Particle surfaces will be capped by EG molecules and cations.

3.2 Electrochemical performances

Because of the LiFePO₄ growth in a preferred orientation for lithium ion transferring, it is reasonable for us to hypothesis the samples S3, S4, S5 and S6 will get excellent electrochemical performances. However, this is not the case. The electrochemical performances of the six carbon coated samples are shown in Fig. 6. S3 and S4 show the best power performances: with S3, the battery has a 131.3 mA h g⁻¹ specific discharge capacity and that with S4 has a 129.8 mA h g⁻¹ specific discharge capacity at 10 C. The S5 and S6 samples which should have got better performances, have got as worst as about 70 mA h g⁻¹ discharge capacity at 0.1 C and 15 mA h g⁻¹ at 10 C. S1 and S2 has a less discharge capacity of 84 and 118.1 mA h g⁻¹ at 10 C compared with S3 and S4. This is due to the larger crystal sizes along b axis for about 60 and 50 nm of S1 and S2 compared with about 40 nm for S3 and S4. The discharge performance of S4 varies with discharge current increasing from 0.1 C to 10 C as shown in Fig. 7. It has got a specific discharge capacity of 169.9, 155.8, 147.0, 145.5, 143.1, 138.5 and 129.8 mA h g⁻¹ individually. The bad performances of S5 and S6 are might because of the bigger crystal particle sizes compared with S1–S4 and more impurities as detected by SEM, TEM and XRD. Although the calculated results of the length along b axis of S5 and S6 are small (about 30 nm), while in Fig. 1 the SEM images shows about 60–80 nm length along [010] direction. The big difference between these two data might refer to a large number of defects or crystal distortions.⁵⁴ In addition, impurity content is also an important parameter that influences the LiFePO₄ particles' performances.

Fig. 6 Comparison of the specific discharge capacities of the carbon coated LiFePO₄ particles prepared via solvothermal method at different primary LiOH/H₃PO₄ mole ratios. S3 and S4 show the best rate performances with about 130 mA h g⁻¹ specific discharge capacity at 10 C at LiOH/H₃PO₄ = 2.7–3.0. S5 and S6 show the worst performances for about 70 mA h g⁻¹ at 0.1 C.

Fig. 7 The first cycle discharge curves of carbon coated LiFePO₄ synthesized at LiOH/H₃PO₄ = 3.0 at different C rates: 169.9, 155.8, 147.0, 145.5, 143.1, 138.5 and 129.8 mA h g⁻¹ at 0.1 C, 0.2 C, 0.5 C, 1 C, 2 C, 5 C and 10 C individually.

3.3 Impurity analysis

As mentioned in Section 1, compared with traditional solid-state synthesis LiFePO₄ method, liquid-phase method like hydrothermal or solvothermal way has a more complicated element distribution during the producing process. There are Li, Fe, P, S elements not only in solid products, the supernatant mother solutions, but in the washing solutions which are produced before clean LiFePO₄ samples are obtained. Thus it is very difficult to get stoichiometric LiFePO₄ samples which are critical for LiFePO₄ crystal structure and performances. To illustrate the element distribution map of solvothermal synthesis LiFePO₄ process, we tested the amount of Li, Fe, P, S elements in solid LiFePO₄ products, mother solutions and washing solutions as LiOH/H₃PO₄ mole ratio changes from 1.8 to 3.15. Table 3 shows the ICP-OES test results of Li, Fe, P, S amount in solid products and mother solutions. As it can be observed, (1) the amount of element Li, Fe, P, S in mother solutions decreases with LiOH/H₃PO₄ mole ratio increases. Almost all the Fe has been precipitated as solid phase and there's little left in mother solutions. 74.5% Li₂SO₄ solute in mother solution at LiOH/H₃PO₄ = 1.8 of S1, while this value decreased to 35.8% at LiOH/H₃PO₄ = 2.7 of S3, and when LiOH/H₃PO₄ ≥ 3, less than 20% Li₂SO₄ solute in mother solutions. (2) In solid phase products, Li/Fe > 1, and P/Fe < 1, compared with pure LiFePO₄ where Li : Fe : P = 1 : 1 : 1, which means that there are impurities in the solid phase products. There is S element detected in solid products and the amount increases with LiOH/H₃PO₄ increases. Li₂SO₄, FeSO₄ or other insoluble impurities stay in solid products. P/Fe is less than 1 in solid phase. So where is the other part of P? Take sample S3 as an example: P in solid phase is 93.4%, and 3.8% is in mother solution, the sum of P is 97.2% < 1 (as Fe/Fe₀ ≈ 1, here we neglect the difference between Fe and Fe₀). The left 2.8% should be washed away. To get this part of P, we tested the amount of elements in washing solution. The data are shown in Table 4. Neglect the loss of Fe during the washing process (0.13% is washed away). Combined Table 3 and 4, the material balance results of P is 1.003 with the error 0.3%.

Table 3 pH of mother solutions produced with various LiOH/H₃PO₄ and Li, Fe, P, S element amount in both solid phase products and mother solutions tested using ICP^b

Sample	LiOH/H3PO4	pH^a	Solid phase (mole ratio)			Mother solution (mole ratio)
			Li/Fe	P/Fe	S/Fe	Fe/Fe0	Li/Fe0	P/Fe0	S/Fe0
a pH of Mother solutions.b Fe0: initial FeSO4 content (mol).
S1	1.8	2.59	1.115	0.964	0.044	0.004	1.276	0.496	0.745
S2	2.4	2.78	1.094	0.911	0.073	0.0009	0.791	0.075	0.566
S3	2.7	3.11	1.161	0.934	0.071	0.0003	0.554	0.038	0.358
S4	3.0	3.73	1.264	0.963	0.082	0.0004	0.371	0.026	0.169
S5	3.1	5.80	1.303	0.887	0.170	0.0003	0.411	0.001	0.185
S6	3.15	6.55	1.350	0.887	0.171	0.0007	0.418	0.001	0.192

---

Table 4 Li, Fe, P, S amount in first 3 times' washing solutions during producing S3 process

Washing solution (mole ratio)	Fe/Fe0	Li/Fe0	P/Fe0	S/Fe0
First 3 times	0.0013	1.134	0.031	0.554

---

In summary, there are a lot of impurities in samples made from solvothermal process at various LiOH/H₃PO₄ mole ratios, even there is no evident extra peaks in S1–S4 samples. The impurities is amorphous phase existed on the surface of LiFePO₄ crystal as shown in Fig. 2(b). High temperature treatment enables the amorphous impurities to crystallize. Fig. 8 shows the comparisons of S1, S3 and S5 treated at 650 °C for 5 hours with and without carbon coated under Argon atmosphere. There are crystallized Li₃PO₄ and Fe₃O₄ impurities in both S3-a and S5-a which are annealed without carbon coated and the later have stronger impurity peaks as shown in Fig. 8(a). However after carbon coating, there is no evident impurities on the XRD patterns of S3-c. This is might because of surface reaction of Li₃PO₄ and Fe₃O₄ which existed on the surface of S3 and S5. Sucrose and the subsequent pyrolytic carbon also hinder the nucleation and growth of impurities. The Fe₃O₄ peaks intensity of S5-c have an obvious decrease compared with that of S5-a. A new Fe₃Fe₄(PO₄)₆ peak appears on the XRD pattern of S5-c.

Fig. 8 Comparison of XRD patterns of different LiFePO₄ samples annealed at 650 °C for 5 hours: (a) without carbon coated; (b) with carbon coated. After high temperature treatment, Li₃PO₄ and Fe₃O₄ impurities are indexed for samples S3-a and S5-a without carbon coating. Impurities disappear when carbon is coated on sample S3 surface. The impurities' relative intensity decreased for all samples coated by carbon. Fe₃Fe₄(PO₄)₆ is also indexed in XRD patterns of carbon coated sample S5-c.

As we discussed in our previous paper, reactions occurred as follows during the whole solvothermal process:

3Fe²⁺ + 2PO₄³⁻ ↔ Fe₃(PO₄)₂(s) (1)

3Li⁺ + PO₄³⁻ ↔ Li₃PO₄(s) (2)

Fe²⁺ + H₂O ↔ Fe(OH)₂(s) + 2H⁺ (3)

Li⁺ + Fe²⁺ + PO₄³⁻ ↔ LiFePO₄(s) (4)

when H₃PO₄ and FeSO₄'s mixed EG solution dripped into LiOH's EG solution, phosphate ions react with ferrous ions or lithium ions to form precipitates Fe₃(PO₄)₂ or Li₃PO₄. Ferrous ions also participate in a hydrolysis reaction. Precipitates like Fe₃(PO₄)₂, Li₃PO₄, Fe(OH)₂ constitute the precursors which solutes gradually in the solvothermal reaction process and form LiFePO₄ crystal. The left Fe₃(PO₄)₂, Li₃PO₄, Fe(OH)₂ precursors and Li₂SO₄ or FeSO₄ which was not washed away are the main source of impurities of solid products produced via solvothermal method.

(1) Fe(OH)₂. If we set Li/Fe = a, P/Fe = b, and S/Fe = c, using element conservation method we can get that the content of Fe(OH)₂ in the solid products is only related to a, b and c. (Set the mole ratio of LiFePO₄, Fe₃(PO₄)₂, Li₃PO₄, Li₂SO₄, FeSO₄, and Fe(OH)₂ in solid products as x₁–x₆ respectively. From element conservation, functions x₁ + 3x₃ + 2x₄ = a; x₁ + 2x₂ + x₃ = b; x₁ + 3x₂ + x₅ + x₆ = 1; x₄ + x₅ = c are built. We can get x₆ = 1 + a/2 − 3b/2 − c.) Table 5 gives the results of contents of Fe(OH)₂ in the six samples. The content of Fe(OH)₂ increases as the original LiOH/H₃PO₄ increases. This is because that when LiOH/H₃PO₄ increases the acidity of the reaction solution decreases which enhances reaction (3) and decreases the solubility of Fe(OH)₂. There are more Fe(OH)₂ left in the result powders which converts to iron oxide (Fe₃O₄) after drying or annealing. The content of Fe(OH)₂ might be decreased by reacting with Li₃PO₄ during high temperature annealing process. The reaction mechanism is shown as reaction (5).

Fe(OH)₂ + Li₃PO₄ ↔ LiFePO₄ + Li₂O + H₂O (5)

Table 5 Fe(OH)₂ contents of solid samples synthesized at different LiOH/H₃PO₄ mole ratios

Sample	S1	S2	S3	S4	S5	S6
LiOH/H3PO4 (mole ratio)	1.8	2.4	2.7	3.0	3.1	3.15
Fe(OH)2/Fe0 (mole ratio)	0.069	0.108	0.109	0.106	0.151	0.174

---

(2) Li₃PO₄. Reaction (1), (3) and (4) are competitive reactions which all consume ferrous ions. The reaction degree of reaction (3) increases with LiOH/H₃PO₄ increasing, which inhibited the generation of Fe₃(PO₄)₂ in the beginning of solvothermal process so that more phosphate ions can involve into reaction (2) and generate more Li₃PO₄. Fe(OH)₂ will release free Fe²⁺ after Fe₃(PO₄)₂ exhausted during solvothermal process. As it's more difficult for Fe(OH)₂ dissociation especially at high pH, the left Li₃PO₄ content increases as LiOH/H₃PO₄ increases when the solvothermal process ends. This is in constant with that the solid phase Li/Fe ratio increases when LiOH/H₃PO₄ increases as shown in Table 3. The left Li₃PO₄ content will be decreased through reaction with Fe(OH)₂ and Fe₃(PO₄)₂ on the LiFePO₄ particle surfaces during subsequent annealing process. The reaction of Li₃PO₄ and Fe₃(PO₄)₂ is shown as follows:

Fe₃(PO₄)₂ + Li₃PO₄ ↔ 3LiFePO₄ (6)

(3) SO₄²⁻. There is residual SO₄²⁻ in every LiFePO₄ samples which is usually washed for 3 times by water and once by ethanol. To check if this was due to inadequate wash, we have done a washing experiment by washing sample S3 for 20 times. Tables 6 and 7 list out the content of Li, Fe, P, S elements in solid products and washing solutions tested by ICP-OES. From Table 6 we can find that after 3 times wash, there are still 6–7% SO₄²⁻ in the samples. During the following 4–20 times' wash, the residual S changes little in solid products. Thus the S element might exist as insoluble impurities and it does not exclude the possibility that S enters into LiFePO₄ olivine lattice as a substitution at P site (partially replace of PO₄³⁻ by polyanion SO₄²⁻).

Table 6 Element contents of solid samples washed for different times

Solid phase(mole ratio)	Fe/Fe0	Li/Fe0	P/Fe0	S/Fe0
a Times washed by water.b Times washed by ethanol.
After 3^a + 1^b	0.999	1.041	0.882	0.072
After 10^a + 1^b	0.963	0.981	0.840	0.064
After 20^a + 1^b	0.926	0.922	0.799	0.062

---

Table 7 Element contents of washing solutions at different washing conditions

Washing solution (mole ratio)	Fe/Fe0	Li/Fe0	P/Fe0	S/Fe0
Within first 3 times wash	0.0013	1.134	0.031	0.554
4^th to 10^th time wash	0.035	0.057	0.042	0.0034
11^th to 20^th time wash	0.037	0.060	0.044	0.0033

---

Table 7 shows a lot of Li₂SO₄ was washed during the first 3 times by water. The content of Fe is as low as 0.13%, and it is reasonable to assume the washed 3.1% P comes from impurities like Li₃PO₄ instead of product LiFePO₄. During the 4^th to 20^th washing process, the total amount of Li, Fe and P are 11.7%, 7.2% and 8.6% individually, which is further more than that of S (0.67%). The main phase LiFePO₄ might be washed out during the 4^th to 20^th washing process. It needs further research about the existing form of S.

4 Conclusion

Six LiFePO₄ particles with different sizes and shapes are synthesized at various LiOH/H₃PO₄ mole ratios prepared by solvothermal synthesis method using ethylene glycol as solvent. Obvious morphology evolution is obtained during LiOH/H₃PO₄ mole ratio increasing from 1.8 to 3.15. Rectangular nanoplate with the main face of (100) transform to spindle plate with the predominant face of (010), and the intermediate state is a long hexagon nanorod with large (010) face and (100) face. Electrochemical tests of these samples show that the LiFePO₄ performances are related to the crystal sizes, crystal thickness along b axis and impurities' contents.

By analyzing the ICP-OES results of LiFePO₄ solid products, mother solutions, and washing solutions, element distributions are described in detail. Results indicate that the formation of impurities of Fe(OH)₂ (Fe₃O₄) and Li₃PO₄ is promoted as LiOH/H₃PO₄ mole ratio increases. The impurities on the surface of the product LiFePO₄ particles can transform to LiFePO₄ during high temperature annealing process to a certain extent. Almost all LiFePO₄ samples synthesized by solvothermal process have S element in it which cannot be washed away thoroughly even for 20 times. Further researches are needed to identify the form of S element existing in LiFePO₄ samples. It might exist as an insoluble impurity on the surface of LiFePO₄ or enter into the LiFePO₄ olivine lattice in solid solution.

Acknowledgements

This work is supported by the MOST (Grant no. 2013CB934000, no. 2011CB935902, no. 2014DFG71590, no. 2010DFA72760, no. 2011CB711202, no. 2013AA050903, no. 2011AA11A257 and no. 2011AA11A254), the Tsinghua University Initiative Scientific Research Program (Grant no. 2010THZ08116, no. 2011THZ08139, no. 2011THZ01004 and no. 2012THZ08129) and State Key Laboratory of Automotive Safety and Energy (no. ZZ2012-011), Suzhou (Wujiang) Automotive Research Institute (Project no. 2012WJ-A-01).

Notes and references

1. A. K. Padhi, K. S. Nanjundaswamy and J. B. Goodenough, J. Electrochem. Soc., 1997, 144, 1188–1194.
2. Z. Y. Bi, X. D. Zhang, W. He, D. D. Min and W. S. Zhang, RSC Adv., 2013, 3, 19744–19751.
3. L. Dimesso, C. Forster, W. Jaegermann, J. P. Khanderi, H. Tempel, A. Popp, J. Engstler, J. J. Schneider, A. Sarapulova, D. Mikhailova, L. A. Schmitt, S. Oswald and H. Ehrenberg, Chem. Soc. Rev., 2012, 41, 5068–5080.
4. Y. Gu, X. Zhang, S. Lu, T. Zhao and D. Jiang, Chin. J. Power Sources, 2013, 37, 147–151.
5. L. X. Yuan, Z. H. Wang, W. X. Zhang, X. L. Hu, J. T. Chen, Y. H. Huang and J. B. Goodenough, Energy Environ. Sci., 2011, 4, 269–284.
6. Y. Zhang, Q. Y. Huo, P. P. Du, L. Z. Wang, A. Q. Zhang, Y. H. Song, Y. Lv and G. Y. Li, Synth. Met., 2012, 162, 1315–1326.
7. C. Ouyang, S. Shi, Z. Wang, X. Huang and L. Chen, Phys. Rev. B: Condens. Matter Mater. Phys., 2004, 69, 104303.
8. G. Wang, X. Shen and J. Yao, J. Power Sources, 2009, 189, 543–546.
9. M. S. Islam, D. J. Driscoll, C. A. J. Fisher and P. R. Slater, Chem. Mater., 2005, 17, 5085–5092.
10. S. L. Bewlay, K. Konstantinov, G. X. Wang, S. X. Dou and H. K. Liu, Mater. Lett., 2004, 58, 1788–1791.
11. F. Y. Kang, J. Ma and B. H. Li, New Carbon Mater., 2011, 26, 161–170.
12. R. Orinakova, A. Fedorkova and A. Orinak, Chem. Pap., 2013, 67, 860–875.
13. K. L. Harrison, C. A. Bridges, M. P. Paranthaman, C. U. Segre, J. Katsoudas, V. A. Maroni, J. C. Idrobo, J. B. Goodenough and A. Manthiram, Chem. Mater., 2013, 25, 768–781.
14. S. B. P. Chang Kyoo Park, Ho. Chul Shin, W. Cho Il and Ho. Jang, Bull. Korean Chem. Soc., 2011, 32, 5.
15. M. Wagemaker, B. L. Ellis, D. Lützenkirchen-Hecht, F. M. Mulder and L. F. Nazar, Chem. Mater., 2008, 20, 6313–6315.
16. Z. Ma, G. Shao, Y. Fan, G. Wang, J. Song and T. Liu, ACS Appl. Mater. Interfaces, 2014, 6, 9236–9244.
17. C. Nan, J. Lu, L. Li, L. Li, Q. Peng and Y. Li, Nano Res., 2013, 6, 469–477.
18. B. J. Paul, S.-W. Kang, J. Gim, J. Song, S. Kim, V. Mathew and J. Kim, J. Electrochem. Soc., 2014, 161, A1468–A1473.
19. A. V. Murugan, T. Muraliganth, P. J. Ferreira and A. Manthiram, Inorg. Chem., 2009, 48, 946–952.
20. C. Sun, S. Rajasekhara, J. B. Goodenough and F. Zhou, J. Am. Chem. Soc., 2011, 133, 2132–2135.
21. B. Ellis, W. H. Kan, W. R. M. Makahnouk and L. F. Nazar, J. Mater. Chem., 2007, 17, 3248–3254.
22. L. Wang, X. He, W. Sun, J. Wang, Y. Li and S. Fan, Nano Lett., 2012, 12, 5632–5636.
23. S. Lemmer and F. Ruether, Chem. Eng. Sci., 2012, 77, 143–149.
24. W. Kang, C. Zhao, R. Liu, F. Xu and Q. Shen, CrystEngComm, 2012, 14, 2245–2250.
25. J. Chen and M. S. Whittingham, Electrochem. Commun., 2006, 8, 855–858.
26. X. Qin, X. Wang, H. Xiang, J. Xie, J. Li and Y. Zhou, J. Phys. Chem. C, 2010, 114, 16806–16812.
27. J. Ni, M. Morishita, Y. Kawabe, M. Watada, N. Takeichi and T. Sakai, J. Power Sources, 2010, 195, 2877–2882.
28. C. Nan, J. Lu, C. Chen, Q. Peng and Y. Li, J. Mater. Chem., 2011, 21, 9994–9996.
29. J. Lim, J. Gim, S.-W. Kang, S. Baek, H. Jeong and J. Kim, J. Electrochem. Soc., 2012, 159, A479–A484.
30. M. K. Devaraju and I. Honma, Adv. Energy Mater., 2012, 2, 284–297.
31. K. Kanamura, S. Koizumi and K. Dokko, J. Mater. Sci., 2008, 43, 2138–2142.
32. J. Liu, R. Jiang, X. Wang, T. Huang and A. Yu, J. Power Sources, 2009, 194, 536–540.
33. K. Dokko, S. Koizumi, H. Nakano and K. Kanamura, J. Mater. Chem., 2007, 17, 4803–4810.
34. J. Lee and A. S. Teja, J. Supercrit. Fluids, 2005, 35, 83–90.
35. F. Iskandar, A. S. Nisa and M. M. Munir, 2013 Joint International Conference on Rural Information & Communication Technology and Electric-Vehicle Technology (rICT & ICeV-T), 2013.
36. T. Muraliganth, A. V. Murugan and A. Manthiram, J. Mater. Chem., 2008, 18, 5661–5668.
37. Z. Dai, L. Wang, F. Ye, C. Huang, J. Wang, X. Huang, J. Wang, G. Tian, X. He and M. Ouyang, Electrochim. Acta, 2014, 134, 13–17.
38. K. Saravanan, P. Balaya, M. V. Reddy, B. V. R. Chowdari and J. J. Vittal, Energy Environ. Sci., 2010, 3, 457–463.
39. Y. Wang, J. Wang, J. Yang and Y. Nuli, Adv. Funct. Mater., 2006, 16, 2135–2140.
40. A. Yamada, S. C. Chung and K. Hinokuma, J. Electrochem. Soc., 2001, 148, A224–A229.
41. S. Lim, C. S. Yoon and J. Cho, Chem. Mater., 2008, 20, 4560–4564.
42. P. Axmann, C. Stinner, M. Wohlfahrt-Mehrens, A. Mauger, F. Gendron and C. M. Julien, Chem. Mater., 2009, 21, 1636–1644.
43. C. Delacourt, P. Poizot, J.-M. Tarascon and C. Masquelier, Nat. Mater., 2005, 4, 254–260.
44. Y. Wang, Z.-s. Feng, J. Hu, L. Yu, J.-j. Chen, L.-l. Wang, X.-j. Wang and H.-l. Tang, Electrochim. Acta, 2014, 117, 431–437.
45. J. Chen and J. Graetz, ACS Appl. Mater. Interfaces, 2011, 3, 1380–1384.
46. M.-H. Lee, T.-H. Kim, Y. S. Kim and H.-K. Song, J. Phys. Chem. C, 2011, 115, 12255–12259.
47. X. Sun and Y. Xu, Mater. Lett., 2012, 84, 139–142.
48. G. Meligrana, C. Gerbaldi, A. Tuel, S. Bodoardo and N. Penazzi, J. Power Sources, 2006, 160, 516–522.
49. D. Y. W. Yu, K. Donoue, T. Kadohata, T. Murata, S. Matsuta and S. Fujitani, J. Electrochem. Soc., 2008, 155, A526–A530.
50. U. Holzwarth and N. Gibson, Nat. Nanotechnol., 2011, 6, 534.
51. H. Borchert, E. V. Shevchenko, A. Robert, I. Mekis, A. Kornowski, G. Grübel and H. Weller, Langmuir, 2005, 21, 1931–1936.
52. D.-H. Kim and J. Kim, Electrochem. Solid-State Lett., 2006, 9, A439–A442.
53. L. Wang, G. Liang, X. Ou, X. Zhi, J. Zhang and J. Cui, J. Power Sources, 2009, 189, 423–428.
54. T. Azib, S. Ammar, S. Nowak, S. Lau-Truing, H. Groult, K. Zaghib, A. Mauger and C. Julien, J. Power Sources, 2012, 217, 220–228.
55. L. Wang, F. Zhou, Y. S. Meng and G. Ceder, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 76, 165435.
56. K. A. Persson, B. Waldwick, P. Lazic and G. Ceder, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85, 235438.
57. K. Dokko, S. Koizumi and K. Kanamura, Chem. Lett., 2006, 35, 338–339.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c4ra09484k

---