Enhanced reversibility and electrochemical performances of mechanically alloyed Cu₃P achieved by Fe addition

Abstract

Cu₃P is a potential anode material for lithium-ion batteries with its comparable gravimetric capacity, but several times higher volumetric capacity (4732 mA h cm⁻³) than graphite. However, the cycling stability of Cu₃P is poor at low discharge potentials and high current densities. In this work, Fe addition is employed as a simple strategy to modulate the composition and phase constitution of Cu₃P nanopowders synthesized by wet mechanical alloying, and thereby to tune the electrochemical performance of the anode. The addition of Fe results in a composite constitute containing Cu₃P as the major phase and some other minor phases including Cu, α-Fe and FeP, which are combinationally determined by X-ray diffraction, energy dispersive X-ray spectroscopy and Mössbauer spectroscopy. Electrochemical tests reveal that both the cycling stability and the rate capability of the electrodes are improved by Fe addition. The Cu₃P electrode with 10% Fe addition shows the best cell performance, with the capacity being remarkably improved by over 100%, from 82 mA h g⁻¹ to 178 mA h g⁻¹ after 50 cycles at 0.75C between 2.0 V and 0.5 V vs. Li/Li⁺. The improvement of the electrochemical performance is engendered by a synergetic effect of the microstructure change of the powders and the presence of Fe-related minor phases, leading to increased electronic conductivity as well as enhanced electrochemical reversibility of the electrode.

---

Introduction

In the last two decades, Li-ion batteries (LIBs) have been widely used as power units in portable electronic devices such as laptops, mobile phones and tablet intelligent kits.¹ However, people are putting forward even higher requirements for LIBs in terms of energy density, durability and safety for future applications, e.g. in electric vehicles and high-speed electronics. While the low specific capacity of current cathode materials (e.g. LiCoO₂ ^2–4 and LiFePO_4,^5,6 140–170 mA h g⁻¹) is the major hurdle that limits the energy density of most commercial cells, the currently used anode materials also have deficiencies which need to be overcome to meet the ever-growing demand for future LIBs.⁷ For example, commercial graphite has a too low intercalation potential close to 0 V vs. Li/Li⁺, which can bring safety problems due to easy formation of undesired Li dendrites;⁸ Li₄Ti₅O₁₂, in spite of its excellent safety and cycling stability, exhibit unfortunately low cell voltage, low capacity (∼160 mA h g⁻¹) and poor rate performance due to its extremely low electronic conductivity.^7,9,10

In recent years, metal phosphides have drawn increasing attention as possible anode materials for LIBs due to their unique advantages such as large reserve and low cost of raw materials, appropriate discharge potentials and high gravimetric/volumetric capacities. Focus has been made mainly on transition metal phosphides such as Mn–P,¹¹ Fe–P,^12,13 Co–P,^14,15 Ni–P,^16,17 Cu–P^17–22 and Zn–P,²³ due to miscellaneous crystal structures and variable valence of the transition metal atoms, which can provide a number of possibilities for Li reactions. In this category, Cu₃P stands out as a promising candidate with a close gravimetric capacity (363 vs. 372 mA h g⁻¹) and over five times higher volumetric capacity (4732 vs. 830 mA h cm⁻³) than graphite.¹⁹ However, microstructured Cu₃P prepared by conventional melting^19,21,22 or solid-state reactions^24–26 were unsatisfactory in electrochemical performance due to a lower conductivity than graphite and inferior Li⁺ transport in the material. To address these issues, most work has been focused on nanostructuring and morphology tuning of Cu₃P through various synthesis approaches such as mechanical alloying,^18,21,22 solvothermal synthesis^21,27–30 and template-based chemical methods,^31–33 with the aim to obtain a number of different nanostructures (nanoparticles, nanocrystals, nanorods or nanoarrays) of Cu₃P. For example, Villevieille et al.³¹ improved the cyclability of Cu₃P/Cu anode by fabricating Cu₃P nanoparticles on templated Cu nanorods, which also worked as current collectors, by a phosphorus vaporization method. Similarly, Ni et al.³² synthesized three dimensional (3D) and porous Cu₃P nanostructures on a Cu foam by a chemical corrosion method, and obtained excellent electrochemical performance due to the large surface area of the 3D nanoarchitecture. Recently, Stan et al.¹⁸ prepared Cu₃P nanopowders by mechanical alloying, which is an efficient and scalable approach for both material synthesis and nanostructuring, and reported the performance of Cu₃P/Li half cells as well as LiNi_0.5Mn_1.5O₄/Cu₃P and Li_1.2(Ni_0.13Mn_0.54Co_0.13)O₂/Cu₃P full cells. The capacity and cyclability of the mechanically alloyed Cu₃P were found closely related with the lower cut-off potential at discharge, which was best limited to 0.5 V vs. Li/Li⁺, with formation of Li_yCu_3−yP (0 ≤ y ≤ 2).

In spite of intensive efforts on nanostructuring and morphology tuning of Cu₃P, there has been no relevant report published to study the effect of compositional or phase constitute tuning on the electrochemical performances of Cu₃P, which is however an important aspect when designing high-performance electrode materials. In this work, we report the influence of Fe-addition on the composition, phase constitute, microstructure and electrochemical performance of Cu₃P nanopowders synthesized by wet mechanical alloying. Both the cycling and rate performance of the Fe-added Cu₃P electrodes are improved, which are maximized by adding 10 at% Fe (relative to Cu) in the ball milling system. The existing form of the added Fe in the ball-milled powders is analyzed by a series of combinational characterizations, and the role of each component on the performance improvement is discussed.

Experimental

Wet mechanical alloying was employed to synthesize the materials with nominal compositions of (Cu_1−xFe_x)₃P (x = 0, 0.05, 0.1 and 0.15). The original intention was to dope partial Cu atoms with Fe to modulate the lattice and phase composition, but final analysis reveals another situation of the constitute as presented later. For the mechanical alloying, stoichiometric amount of Cu, Fe (both 99.9%, Aladdin) and red phosphorus (AR, Aladdin) powders were loaded in stainless steel jars (filled with balls of the same material) in a glove box protected by Ar. Proper amount of hexane was added as a liquid dispersant to avoid powder agglomeration. All ball milling experiments were conducted for 10 h at a speed of 350 rpm, using a ball-to-powder weight ratio of 30 : 1. After ball milling, the slurry was collected and dried in vacuum at 60 °C for 12 h to remove the liquid. The resulted products were then ground with an agate mortar and pestle to obtain homogeneous powders for use.

The phase structure of the powders was studied by using X-ray diffraction (XRD) on a PANalytical Empyrean diffractometer using Cu K_α radiation (λ = 1.54056 Å). The morphology of the powders was observed using a Hitachi S3400N field emission scanning electron microscope (FESEM). Chemical compositions of the powders were analyzed by using energy dispersive X-ray spectroscopy (EDS) with an Oxford INCA PentaFET-x3 SDD attached to the FESEM. Mössbauer spectroscopy experiments with the transmission geometry were conducted for the samples. The radiation source was a ⁵⁷Co radiation source in the rhodium (Rh) matrix. The rate of the radiation source is calibrated with a standard α-Fe foil. The Mössbauer spectra were fitted using the software called WinNormos-for-Igors®. The ball-milled powders were cold-pressed to circle pellets under 80 MPa for evaluation of their electronic conductivity, which was measured with a conventional four-probe method.

The ball-milled powders (active materials, 80 wt%) were thoroughly mixed with acetylene black (conducting additive, 10 wt%) and polyvinylidene fluoride (PVDF, binder, 10 wt%) with a few drops of N-methyl pyrrolidone (NMP). The obtained slurry was then homogeneously spread on a piece of Cu foil to prepare the electrodes, which were then dried at 110 °C in vacuum for 12 h. Polypropylene films (Celgard 2400) were used as separators and were dried at 50 °C in vacuum for 12 h. Cu₃P/Li half-cells were assembled in a glove-box filled with Ar using 1 M LiPF₆ dissolved in a mixture of ethylene carbonate (EC), diethyl carbonate (DEC) and dimethyl carbonate (DMC) (1 : 1 : 1, v/v) as the liquid electrolyte. The galvanostatic charge/discharge tests were performed between 2.0 V and 0.5 V vs. Li/Li⁺ at different C rates (1C = 242 mA g⁻¹ for two mole Li⁺ extraction from one mole Cu₃P when discharged to 0.5 V) at room temperature using a BTS-5V2mA cell test instrument (NEWARE Electronic Co.). Cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS) tests were performed using a CHI660B electrochemical workstation (CH Instruments Co.). The frequency range and the perturbation voltage for the EIS tests were 100 000–0.1 Hz and 5 mV, respectively.

Results and discussion

The phase structures of the ball-milled powders are determined from the XRD patterns in Fig. 1. All the samples can be well indexed to the hexagonal Cu₃P phase (ICDD No. 71-2261) with a space group of P6₃cm. Obvious broadening of the diffraction peaks is observed, indicating extremely small grains of the powders, which can be verified by the FESEM results discussed later. In addition, an impurity minor phase can be judged from the small diffraction peak at ∼43.3°, which can be indexed to elemental Cu (ICDD No. 04-0836). No noticeable elemental P is found in the ball-milled powders from the XRD patterns, which is also verified by thermogravimetric (TG) analysis as shown in Fig. S1, ESI,† where the weight loss of the ball-milled powders is negligible (<0.5%) after the samples are heated to 400 °C. From the EDS results (Table 1). We can see clearly that Fe is successfully introduced into the material after ball milling. The measured Fe : Cu atomic ratios are in good agreement with the nominal ones. However, all the samples show evident non-stoichiometric compositions with deficiency in P. In other words, the obtained powders after wet ball milling are compositionally rich in metal element, which matches with the fact of the presence of secondary Cu phase in the powders, particularly with lower Fe-additions. The P loss during ball milling is proven by repeated experiments, but the reason is not very clear due to complexity of the side reactions that may take place during high-energy ball milling. The mechanism of P loss in this work is very similar with the reported Si loss when Si was subjected to ball milling in hexane,³⁴ because P and Si have much similarity in chemical properties and may undergo sublimation or chemical reactions with the organic media forming gaseous or liquid products that cannot be collected. It is noteworthy that an increased Fe-addition is able to mitigate the P loss during the milling process. This can be either due to a lowered energy barrier of the Cu–P reaction by Fe-addition, or associated with the occurrence of additional Fe–P reactions that can consume more P per mole metal than the case of Cu₃P, forming possible phases like Fe₂P, FeP, FeP₂ and FeP₄. Unfortunately, none of these phases can be easily detected solely by XRD, because their amount can be very tiny, and the obtained XRD patterns don't have sufficient resolution due to severe peak broadening and overlapping of the peaks. Mössbauer spectroscopy will be used to further identify the Fe-related phases which will be discussed later.

Fig. 1 XRD patterns of the ball-milled powders.

Table 1 Nominal and actual compositions of the ball-milled powders measure by EDS

Fe addition	Nominal composition		Real composition
	Cu/Fe/P (at%)	Formula	Cu/Fe/P (at%)	Formula
0	75/0/25	Cu3P	77.26/0/22.74	Cu3P0.88
5%	71.25/3.75/25	Cu2.85Fe0.15P	72.65/4.57/22.79	Cu2.85Fe0.17P0.89
10%	67.5/7.5/25	Cu2.7Fe0.3P	69.11/8.35/22.54	Cu2.7Fe0.32P0.88
15%	63.75/11.25/25	Cu2.55Fe0.45P	64.71/11.45/23.84	Cu2.55Fe0.45P0.94

---

The electrochemical performances of the Fe-added Cu₃P electrodes are displayed in Fig. 2. Initially, the cells were charged to 3.0 V after being discharged from the open-circuit potential (OCP) to 0.5 V. The short plateaus at the 1^st charge around 2.5 V is resulted by the oxidation of the remaining Cu, which can be understood as stepwise oxidation reaction: Cu − e → Cu⁺, Cu⁺ − e → Cu²⁺.³⁵ Such an oxidation reaction consumes reversible Cu and is undesired for the Cu₃P electrode. Therefore, from the 2^nd cycle on, the electrode is charged to only 2.0 V in order to ensure no Cu oxidation. It is observed that the oxidation plateaus is weakened with increasing Fe addition. This will be explained later from the view of the phase change after Fe-addition. The initial discharge capacities of the electrodes with different Fe-additions are very close to each other, reaching ∼250 mA h g⁻¹ at 0.75C (Fig. 2c). For the 1^st cycle, the pristine electrode shows the largest coulombic efficiency, which is found to decrease as the Fe-addition increases. After a few cycles of activation, the columbic efficiency of the Fe-added electrodes is gradually increased, being higher than that of the pristine one after 50 cycles. Also, the electrochemical polarization of all Fe-added electrodes is markedly reduced after 50 cycles (Fig. 2b), while the cycling stability is generally improved (Fig. 2c). For the electrodes with 0% and 5% Fe-addition, the capacity at the 50^th cycle fades quickly to only 31.6% (82 mA h g⁻¹) and 45.9% (118 mA h g⁻¹) of their initial capacities, respectively. In contrast, the other two electrodes with greater Fe-additions exhibit more stable capacities than the pristine electrode once they have experienced the decay of the first few cycles, although the capacity value of the 15% Fe-added electrode is lower. After 50 cycles, the electrode with 10% Fe-addition delivers the highest capacity of 178.5 mA h g⁻¹ (70% retention), which is improved by over 100% as compared to the pristine one (82 mA h g⁻¹, 32% retention). Beside the cycling performance, the rate capability of the electrodes is also improved by Fe-addition (Fig. 2d). At the largest current density of 360 mA g⁻¹ (1.5C), the capacity of the Cu₃P electrode with 10% Fe-addition achieves a discharge capacity of 155 mA h g⁻¹, which is twice that of the pristine electrode (71 mA h g⁻¹). However, further addition of Fe results in a reversed attenuation of the rate capability under large current densities.

Fig. 2 Electrochemical performances of the Fe-added Cu₃P electrodes: charge/discharge profiles of (a) the 1^st cycle between 3.0 V and 0.5 V and (b) the 50^th cycle between 2.0 V and 0.5 V at the same current density of 180 mA g⁻¹ (0.75C), (c) cycling performance and (d) rata capability of the electrodes between 2.0 V and 0.5 V.

The CV profiles of the electrodes are presented in Fig. 3. For the pristine electrode (Fig. 3a), three reduction peaks at 1.25 V (S), 0.86 V (A) and 0.68 V (B) are observed in the 1^st cycle, while only two oxidation peaks are present, at 1.08 V (B*) and 1.17 V (A*). The broad reduction peak S is attributed to irreversible reactions taking place on the surface of the active materials such as the formation of solid electrolyte interface (SEI) or reduction of surface oxidation layers. The lithiation/delithiation mechanism of Cu₃P has been well studied by Bichat and coworkers^19,36 by in situ XRD. A multi-step conversion mechanism involving sequential formation of Li_yCu₃P (y ≈ 0.33), LiCu₂P, Li₂CuP and Li₃P phases and simultaneous extrusion of Cu was proposed when discharged to 0.02 V vs. Li/Li⁺. With a higher cut-off potential (0.5 V) in this work, the redox reactions A/A* and B/B* can be ascribed to the two-step processes with maximum 2 mole Li reaction with 1 mole Cu₃P:^19,36

A/A* Cu₃P + yLi ↔ Li_yCu₃P (y ≤ 1/3)

B/B* Li_yCu₃P + zLi ↔ Li_y+zCu_3−y−zP + (y + z)Cu⁰ (0 ≤ z ≤ 2 − y)

Fig. 3 CV profiles of the Cu₃P electrodes with (a) 0%, (b) 5%, (c) 10% and (d) 15% addition of Fe tested at a sweeping rate of 0.05 mV s⁻¹ between 2.0 V and 0.5 V.

It is to note that A/A* is an alloying/dealloying reaction and leads to no change of the basic crystal structure of the material, while B/B* is an intercalation/deintercalation reaction involving formation of two different phases. The latter reaction is believed to be less reversible than the former one because the lattice change from Li_yCu₃P (a = b = 7.02 Å, c = 7.26 Å) to Li₂CuP (a = b = 4.03 Å, c = 7.79 Å) is more dramatic than the change from Cu₃P (a = b = 6.98 Å, c = 7.23 Å) to Li_yCu₃P.¹⁹ Such a big change may induce stresses in the material and can lead to pulverization and structural damage of the active materials. Furthermore, the extruded fine Cu particles at the 1^st discharge are easy to aggregate, which can somehow impede the oxidation reactions at the following charge and therefore reduce the overall reversibility of the electrode. The addition of Fe in Cu₃P is found to have several effects as can be seen from the CV results. First, the irreversible peak S is no longer noticeable after Fe-addition over 10%, which indicates less formation of SEI or side reactions before the 1^st lithiation (A/A*). Although the initial columbic efficiency of the Fe-added electrode is found less than the pristine electrode (Fig. 2c), it is to note that the electrodes were cycled under different conditions (charging potential and current density) with the CV test in the initial cycle. When a lower current density (18 mA g⁻¹) and the same potential was applied (Fig. S2, ESI†), we obtain an improvement of the initial columbic efficiency (72.2%, 74.8%, 78.1% and 75.8% for 0, 5%, 10% and 15% Fe-addition, respectively) after Fe-addition, which is then consistent with the less irreversible peak in the CV results. Second, the peak currents of the Fe-added electrodes (10% and 15%) after 50 cycles are much less reduced than those of the pristine and 5% Fe-added Cu₃P electrodes, which is further evidence of improved reversibility of the active materials after Fe-addition. Third, Fe-addition leads to noticeable shift of the redox potentials which are illustrated in Fig. 4. With the increase of Fe-addition, both A and B peaks of the 1^st cycle are shifted to the cathodic direction, while their oxidation potentials become more anodic. Interestingly, the 2^nd cycle redox potentials for all Fe-added electrodes are generally recovered to the initial level of the pristine electrode. From then on, the redox potentials are quite stable for higher Fe-additions (10% and 15%) upon further cycling (Fig. 3c and d), while the pristine electrode exhibit an unexpected shift of the reduction potential at the 50^th cycle (Fig. 3a).

Fig. 4 Redox potentials of the Fe-added Cu₃P electrodes obtained from the CV profiles.

In order to elucidate the role of Fe-addition on the electrochemical performance of Cu₃P electrode, the chemical form of the added Fe must be known. As the present XRD and EDS results cannot corroborate the original idea of Fe doping, the added Fe may exist as other forms in the powders. Mössbauer spectroscopy, which is able to get key information about the form of Fe, was therefore employed for further study. The recorded spectra of the powders with 10% and 15% Fe-additions are displayed in Fig. 5. Through detailed analysis, two sub-spectra can be resolved from the whole spectrum, including a six-peak spectrum and a doublet, whose hyperfine parameters are listed in Table 2. The six-peak spectrum (green line in Fig. 5a) can be indexed to the α-Fe phase, whose hyperfine magnetic field (H) is 32.81 T, being very close to the standard value of 33.0 T; the slight decrease of H may originate from the ultrafine grains of α-Fe caused by ball milling. The position of the red doublet in Fig. 5a is an indication of a Fe–P phase. As mentioned above and according to the binary phase diagram of Fe–P, there are several Fe–P alloys (Fe₂P, FeP, FeP₂ and FeP₄) that are possible to form during the ball milling process. By comparing the obtained isomer shift (δ) and quadrupolar splitting (ΔE_q) values with literature,^13,37 the doublet can only be indexed to FeP rather than any other Fe–P alloy. Quantitative analysis reveals relatively more α-Fe (67–68% in molar ratio) than FeP (32–33%) in the two samples. The sample with 10% Fe-addition shows a relatively smaller ΔE_q than the one with 15%, indicating a different electric field distribution around the Fe atoms, which may be caused by the existence of more Cu surrounding the FeP phase as inferred from their different compositions (Table 1). The Mössbauer spectroscopy results are able to reveal that one third of the added Fe is alloyed with P forming a new FeP phase while the rest remains as unreacted α-Fe in the powders. Although the XRD result in Fig. 1 is not able to directly evidence the α-Fe phase due to severe peak broadening, one can still realize an ambiguous shift of the major peak of Cu₃P with increasing amount of Fe-addition after a closer observation. By fitting this peak using the double peaks of α-Fe (ICDD No. 87-0722, 2θ = 44.76°) and Cu₃P (2θ = 45.09°), it is more easier to notice the presence of α-Fe in the powders (Fig. S3, ESI†). The above results manifest that the intended Fe doping in Cu₃P has failed under the present ball milling conditions, possibly due to the higher activity of Fe–P reactions and the non-stoichiometry of the resulted powders. Nevertheless, the electrochemical performances of the Cu₃P electrode are evidently improved by the as-formed secondary phases after Fe-addition.

Fig. 5 Mössbauer spectra of the ball-milled powders added with 10% and 15% Fe. (a) The whole spectra fitted by a doublet and six-line peak and (b) details of the doublet fittings.

Table 2 Hyperfine parameters of the phases observed in the ball-milled powders from the Mössbauer spectra. δ: isomer shift, ΔE_q: quadrupolar splitting, H: hyperfine magnetic field, 2Γ: line width, R: proportion

Fe addition	Peaks	δ (mm s^−1)	ΔEq (mm s^−1)	H (T)	2Γ (mm s^−1)	R
10%	Six-line peak	0.0016	−0.0083	32.81	0.24	68.6%
	Doublet	0.35	0.36	—	0.64	31.4%
15%	Six-line peak	0.0180	0.0070	32.85	0.25	67.7%
	Doublet	0.35	0.55	—	0.61	32.3%

---

Based on the above analysis, the phase constitute of the ball-milled powders can be presented as a mixture of Cu₃P, Cu, Fe and FeP phases. The ratio of each phase can be quantitatively evaluated according to the overall compositions (Table 1) and the known ratio of Fe to FeP (2 : 1). As the Fe/FeP ratio is almost the same for the 10% and 15% Fe-added samples, it is also employed for the calculation of the 5% Fe-added one. The chemical formula of the samples are therefore adjusted as shown in Table 3, which can more explicitly reflect the actual phase species and their content. It is to see that with increasing addition of Fe, both the FeP and α-Fe phases are monotonically increased, while the proportion of Cu₃P and Cu phase is reduced. The decreasing amount of the remaining Cu phase with increasing Fe-addition can perfectly explain the weakening of the small oxidation plateaus above 2.0 V in Fig. 2a. As the formation of FeP per mole of metal will consume more P than the formation of Cu₃P, P element can be more completely alloyed rather than lost during ball milling when Fe is added to a critical limit, which can explain the improved stoichiometry (less P loss) of the 15% Fe-added sample (Table 1). FeP is a well-studied anode material whose lithiation reactions are more cathodic than Cu₃P. For example, the potential of FeP + yLi → Li_yFeP (y ≤ 1) at the 1^st discharge was reported to be 0.42 V,¹² which is slightly lower than the potential of B/B* reaction for Cu₃P (0.68 V). The displacement reaction involved in FeP (LiFeP + 2 Li → Li₃P + Fe⁰) took place at ∼0.25 V and can be ignored in this study as the discharge is stopped at 0.5 V. It was also found that from the 2^nd cycle on, the potential of the intercalation reaction for FeP was increase to ∼0.6 V and then kept stable. In this regard, it is no wonder that the reduction potentials of the Cu₃P/FeP mixture electrode are generally shifted to the cathodic direction at the 1^st discharge and then return again in the following cycles, as shown in Fig. 3 and 4. Furthermore, the theoretical capacity of FeP phase (308 mA h g⁻¹ for one-electron reaction) is no less than that of Cu₃P (242 mA h g⁻¹ for two-electron reaction). As a result, the presence of minor FeP phase in Cu₃P has no significant influence on the initial discharge capacity of the electrode.

Table 3 Reformulated compositions of the ball-milled powders and the evaluated electronic conductivity of the cold-pressed pellets

Fe addition	Composition	Conductivity (S m^−1)
0	(Cu3P)0.88Cu0.36	0.022
5%	(Cu3P)0.84(FeP)0.05Cu0.33Fe0.12	0.098
10%	(Cu3P)0.78(FeP)0.1Cu0.26Fe0.22	0.147
15%	(Cu3P)0.79(FeP)0.15Cu0.08Fe0.3	0.054

---

The microstructure and electronic conductivity of the mechanically alloyed materials are examined to obtain more understanding of the improved electrochemical performances. The FESEM images of the as-milled powders are shown in Fig. 6. The pristine powders without Fe-addition are easy to aggregate, showing secondary particles of 3–5 μm in diameter while the primary grains are of the size from tens of nanometers to a few hundred nanometers. The addition of Fe is advantageous to reduce the aggregation of the primary grains, forming a refined microstructure of the powders containing a number of pores and gaps between the primary grains. However, the refined microstructure disappears when the Fe-addition is increased to 15%, and the particle aggregation becomes again noticeable being similar to the pristine powders. Fig. 7 shows the molar ratio of the total metal phases (Cu and α-Fe) calculated using the suggested compositions in Table 3 and also depicts the electronic conductivity of the materials as a function of Fe-addition. It is found that the ratio of the metal phases first increases and then decreases with the increase of Fe-addition. A similar relationship is observed on the electronic conductivity result. The sample with 10% Fe-addition exhibits the largest ratio of metal phases as well as the highest electronic conductivity among all electrodes. The residual Cu and Fe phases in the powders are believed to enable tuning of the electronic conductivity of the materials because they are intrinsically more conductive than FeP and Cu₃P, which can be one important reason for the enhanced rate capability of the Fe-added electrodes. Meanwhile, the refined microstructure of the powders is believed to be another important reason for the enhanced electrochemical performances as that can improve electrolyte infiltration and shorten the Li⁺ diffusion length during cycling.

Fig. 6 FESEM images of the ball-milled powders with (a and e) 0%, (b and f) 5%, (c and g) 10% and (d and h) 15% addition of Fe.

Fig. 7 Molar ratio of elemental Cu/Fe phases and the electronic conductivity of cold-pressed pellets from the ball-milled powders.

The charge transfer and diffusion kinetics of the electrodes are interpreted by EIS measurements. Fig. 8a shows the Nyquist plots of the EIS data for the electrodes after 3 cycles. Employing the equivalent circuit in the inset, the ionic resistance of the electrolyte (R_s), the surface layer resistance (R_f) and the charge transfer resistance (R_ct) of the electrodes are fitted (Table 4). It is found that the addition of Fe generally leads to a monotonic reduction of R_ct, indicating less irreversible SEI formation on the surface of the active materials, which is consonant with the weakening of the broad peak S in the CV curves (Fig. 3). The charge transfer resistance (R_ct) is first reduced and then increased with increasing Fe-addition, reaching the minimum when Fe-addition is 10%. This should be associated with its highest electronic conductivity, enabling more efficient transfer of electrons, and is also consistent with the best electrochemical performance of this electrode. The Li⁺ diffusion coefficient (D_Li⁺) in the electrode can be determined by plotting the linear range of Z′ as a function of ω^−1/2, which is displayed in Fig. 8b. In the low frequency range, Z′ and D_Li⁺ can be expressed by^38,39

Z′ = R_s + R_f + R_ct + δω^−1/2 (1)

where δ is the Warburg coefficient, V_m is the molar volume of the active material, A is the contact area of the electrode, F is the Faraday constant, and –dE/dX is the slope of OCP vs. mobile ion concentration X. Obviously, suppose other parameters are constant, D_Li⁺ is inversely proportional to δ, which is the slope of the Z′ − ω^−1/2 lines. The observed slopes in Fig. 8b can clearly demonstrate that the Li⁺ diffusion coefficient for the Cu₃P electrode is evidently enhanced by Fe-addition from 5% to 15%.

Fig. 8 (a) Nyquist plots of the EIS data after 50 cycles. The inset shows the equivalent circuit used for fitting. (b) The linear range of the real part of impedance Z′ as a function of ω^−1/2.

Table 4 Resistance parameters fitted from EIS spectra. R_s: ionic resistance of the electrolyte, R_f: surface layer resistance, R_ct: charge transfer resistance

Fe addition	Rs (Ω)	Rf (Ω)	Rct (Ω)
0	2.3	25.3	249
5%	1.9	17.0	160
10%	2.7	4.6	80
15%	4.7	1.4	120

---

Based on the above analysis, the enhanced electrochemical performances of the mechanically alloyed Cu₃P after Fe-addition can be understood as synergetic effect of the changes in both phase constitute and microstructure. Although the synthesis via wet mechanical alloying remains to be problematic due to undesired loss of P, the present study demonstrates a positive effect of Fe-addition, which leads to a mixture of Cu₃P and FeP phases with some remaining Cu and Fe nanoparticles dispersed in the matrix. The excessive Cu and Fe particles can act as lubricant for the ball milling and modify surface energies of the primary nanoparticles, being able to inhibit the coarsening of active Cu₃P particles (Fig. 6f and g). However, if the metal phases are too much (15% Fe-addition), they can agglomerate themselves and also cause coarsening of Cu₃P, as shown in Fig. 6h. As we know, the displacement reaction (B/B*) for Cu₃P is less reversible than the alloying reaction (A/A*) due to agglomeration of the extruded Cu particles, which can produce some “dead Cu” in the electrode causing incomplete recovery of Cu₃P during the following charge. This problem is the major factor accounting for the inferior cyclability of the pristine Cu₃P electrode, which can be reflected by its CV profile (Fig. 3a) where a dramatic reduction of the oxidation peaks as well as a cathodic shift of the reduction peak is observed at the 50^th cycle. The addition of Fe is able to enhance the overall reversibility of the redox reactions from two aspects. First, the coexistence of multiple phases leads to a favorable microstructure and mitigated aggregation of the active materials (Fig. 6), being able to better disperse the extruded Cu particles at discharge and thus make the oxidation reaction at charge more complete. Second, the presence of FeP phase is also beneficial to the electrochemical reversibility of the mixture electrode. As mentioned earlier, the FeP undergoes an intercalation/deintercalation reaction above 0.5 V (FeP + yLi → Li_yFeP) which is similar to the A/A* reaction for Cu₃P. Although this reaction requires more cathodic potential at the 1^st discharge, it is kinetically more reversible than the displacement reaction of Cu₃P (B/B*) in the same potential range and therefore can improve the cyclability and rate capability of the electrode.

As to the residual Cu phase, we think it is unfavorable for the overall performance of Cu₃P despite of its contribution to the enhancement of the material's electronic conductivity, because the extruded Cu can be grown more easily on the pre-existing Cu sites and then aggregate in there, therefore worsening the reversibility of the displacement reaction. This is confirmed by an additional study of a stoichiometric Cu₃P electrode obtained by properly increasing the P amount in the ball milling experiment, whose electrochemical performances are shown in Fig. S4, ESI.† The stoichiometric Cu₃P electrode displays better stability and rate capability than the non-stoichiometric electrode without Fe-addition. On the other hand, the addition of Fe (10%) overwhelms the disadvantage of the non-stoichiometry and shows the best electrochemical performances, which further highlights the role of the Fe-related phases in enhancing the electrochemical reversibility of the mixture electrode. In this regard, it is to foresee that if the stoichiometry of Cu₃P and the mixture phase are carefully controlled at the same time, the electrochemical performances of the electrode can be further improved.

Conclusions

Improved cycling performance and rate capability of mechanically alloyed nanostructure Cu₃P are achieved simply by addition of Fe. Detailed analysis reveals existence of elemental Cu and α-Fe as well as FeP minor phases in the Cu₃P powders instead of forming doped copper phosphide due to significant loss of P and the high activity of Fe with P in the wet ball milling process. The role of Cu/Fe phases can be interpreted as an inhibiter against aggregation of nanoparticles and a reinforcer for electrochemical reactions by enhancing the electronic conductivity of the materials. FeP, which is more reversible than Cu₃P in the applied potential range of 2.0–0.5 V vs. Li/Li⁺, is found to have influenced the redox potentials of the mixture electrode and generally enhanced the electrochemical reversibility. Under the synergetic effect of these minor phases, the capacity of the mixture electrode is remarkably improved from 82 mA h g⁻¹ to 178 mA h g⁻¹ after 50 cycles at 0.75C.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (21073029, 11234013, 21473022 and 51102039), the Science and Technology Bureau of Sichuan Province of China (2015HH0033), and Fundamental Research Funds for the Central Universities of China (ZYGX2012Z003 and ZYGX2015J027).

Notes and references

1. M. S. Whittingham, Chem. Rev., 2004, 104, 4271–4302.
2. X. Y. Dai, A. J. Zhou, J. Xu, B. Yang, L. P. Wang and J. Z. Li, J. Power Sources, 2015, 298, 114–122.
3. X. Y. Dai, L. P. Wang, J. Xu, Y. Wang, A. J. Zhou and J. Z. Li, ACS Appl. Mater. Interfaces, 2014, 6, 15853–15859.
4. X. Dai, A. Zhou, J. Xu, Y. Lu, L. Wang, C. Fan and J. Li, J. Phys. Chem. C, 2016, 120, 422–430.
5. S. Y. Chung, J. T. Bloking and Y. M. Chiang, Nat. Mater., 2002, 1, 123–128.
6. A. K. Padhi, K. S. Nanjundaswamy and J. B. Goodenough, J. Electrochem. Soc., 1997, 144, 1188–1194.
7. S. Goriparti, E. Miele, F. De Angelis, E. Di Fabrizio, R. P. Zaccaria and C. Capiglia, J. Power Sources, 2014, 257, 421–443.
8. H. Honbo, K. Takei, Y. Ishii and T. Nishida, J. Power Sources, 2009, 189, 337–343.
9. T. F. Yi, S. Y. Yang and Y. Xie, J. Mater. Chem. A, 2015, 3, 5750–5777.
10. Y. Wang, A. J. Zhou, X. Y. Dai, L. D. Feng, J. W. Li and J. Z. Li, J. Power Sources, 2014, 266, 114–120.
11. D. C. S. Souza, V. Pralong, A. J. Jacobson and L. F. Nazar, Science, 2002, 296, 2012–2015.
12. S. Boyanov, J. Bernardi, F. Gillot, L. Dupont, M. Womes, J. M. Tarascon, L. Monconduit and M. L. Doublet, Chem. Mater., 2006, 18, 3531–3538.
13. S. Boyanov, D. Zitoun, M. Menetrier, J. C. Jumas, M. Womes and L. Monconduit, J. Phys. Chem. C, 2009, 113, 21441–21452.
14. R. Alcántara, J. L. Tirado, J. C. Jumas, L. Monconduit and J. Olivier-Fourcade, J. Power Sources, 2002, 109, 308–312.
15. M. C. Lopez, G. F. Ortiz and J. L. Tirado, J. Electrochem. Soc., 2012, 159, A1253–A1261.
16. S. Boyanov, J. Bernardi, E. Bekaert, M. Ménétrier, M. L. Doublet and L. Monconduit, Chem. Mater., 2009, 21, 298–308.
17. L. Monconduit, J. Phys. Chem. C, 2014, 118, 10531–10544.
18. M. C. Stan, R. Klopsch, A. Bhaskar, J. Li, S. Passerini and M. Winter, Adv. Energy Mater., 2013, 3, 231–238.
19. B. Mauvernay, M. P. Bichat, F. Favier, L. Monconduit, M. Morcrette and M. L. Doublet, Ionics, 2005, 11, 36–45.
20. K. Wang, J. Yang, J. Y. Xie, B. F. Wang and Z. S. Wen, Electrochem. Commun., 2003, 5, 480–483.
21. M. P. Bichat, T. Politova, H. Pfeiffer, F. Tancret, L. Monconduit, J. L. Pascal, T. Brousse and F. Favier, J. Power Sources, 2004, 136, 80–87.
22. O. Crosnier and L. F. Nazar, Electrochem. Solid-State Lett., 2004, 7, A187–A189.
23. M. P. Bichat, L. Monconduit, J. L. Pascal and F. Favier, Ionics, 2005, 11, 66–75.
24. H. Pfeiffer, F. Tancret, M. P. Bichat, L. Monconduit, F. Favier and T. Brousse, Electrochem. Commun., 2004, 6, 263–267.
25. H. Pfeiffer, F. Tancret and T. Brousse, Electrochim. Acta, 2005, 50, 4763–4770.
26. M. S. Chandrasekar and S. Mitra, Electrochim. Acta, 2013, 92, 47–54.
27. S. L. Liu, Y. T. Qian and L. Q. Xu, Solid State Commun., 2009, 149, 438–440.
28. L. De Trizio, A. Figuerola, L. Manna, A. Genovese, C. George, R. Brescia, Z. Saghi, R. Simonutti, M. Van Huis and A. Falqui, ACS Nano, 2012, 6, 32–41.
29. S. L. Liu, S. Li, J. P. Wang, Q. Q. Shi and M. M. Li, Mater. Res. Bull., 2012, 47, 3352–3356.
30. L. De Trizio, F. De Donato, A. Casu, A. Genovese, A. Falqui, M. Povia and L. Manna, ACS Nano, 2013, 7, 3997–4005.
31. C. Villevieille, F. Robert, P. L. Taberna, L. Bazin, P. Simon and L. Monconduit, J. Mater. Chem., 2008, 18, 5956–5960.
32. S. B. Ni, J. J. Ma, X. H. Lv, X. L. Yang and L. L. Zhang, J. Mater. Chem. A, 2014, 2, 20506–20509.
33. J. Q. Tian, Q. Liu, N. Y. Cheng, A. M. Asiri and X. P. Sun, Angew. Chem., Int. Ed., 2014, 53, 9577–9581.
34. A. J. Zhou, X. B. Zhao, T. J. Zhu, T. Dasgupta, C. Stiewe, R. Hassdorf and E. Mueller, Intermetallics, 2010, 18, 2051–2056.
35. N. Pereira, L. Dupont, J. M. Tarascon, L. C. Klein and G. G. Amatucci, J. Electrochem. Soc., 2003, 150, A1273–A1280.
36. M. P. Bichat, T. Politova, J. L. Pascal, F. Favier and L. Monconduit, J. Electrochem. Soc., 2004, 151, A2074–A2081.
37. M. Womens, S. Boyanov, J. C. Jumas and L. Monconduit, Hyperfine Interact., 2008, 187, 57–69.
38. D. Zhang, B. N. Popov and R. E. White, J. Power Sources, 1998, 76, 81–90.
39. Y. S. Bai, X. Y. Wang, X. Y. Zhang, H. B. Shu, X. K. Yang, B. A. Hu, Q. L. Wei, H. Wu and Y. F. Song, Electrochim. Acta, 2013, 109, 355–364.

---

Footnote

† Electronic supplementary information (ESI) available: TG curve of the ball-milled powders and electrochemical performances of stoichiometric Cu₃P electrode. See DOI: 10.1039/c6ra01637e

---