Introduction

Rechargeable Li-ion batteries have become major power sources in a growing number of applications, including the automotive industry. This creates new requirements such as high rate performance during both charging and discharging. To describe the rate limiting process, all internal system resistances must be known. Among several processes taking place in Li-ion batteries, the Li+ charge transfer reaction at both the cathode and anode seems to be essential to gain insight into the internal cell impedance. The most studied anode materials are different carbons and LiMn2O4, while LiCoO2 and LiFePO4 are basic cathode materials. Kinetics of anodic [1–8] and cathodic processes [9–24] have been studied. Results of reported studies are not comparable. However, one of the most important factors is the resistance (impedance) associated with the charge transfer process, which has not been fully clarified for cathodes. The general aim of this study was to compare the charge transfer process taking place at LiMn2O4, LiCoO2, and LiFePO4 cathodes using the same methodology and electrolytes.

Experimental

Materials

LiFePO4 (carbon coated, battery grade, Aldrich), LiMn2O4 (Aldrich), LiCoO2 (Aldrich), carbon black (CB, Fluka), poly(vinylidene fluoride) (PVdF, M W = 180,000 Fluka), lithium foil (Aldrich, 0.75 mm thick), and N-methyl-2-pyrrolidinone (NMP, Fluka) were used as received. LiPF6 solution (1 M) in a mixture of ethylene carbonate and dimethyl carbonate (EC + DMC 1:1, Aldrich) was used as electrolyte.

Tested cathodes were prepared on golden current collectors by a casting technique from a slurry of electrode material, carbon black, and PVdF suspension in NMP. After solvent (NMP) evaporation at 120 °C in a vacuum, a layer of the electrode, containing the active material, the electronic conductor (CB), and a binder (PVdF) was formed.

Procedures and measurements

Electrochemical properties of the cells were characterized using electrochemical impedance spectroscopy (EIS) and galvanostatic charging/discharging tests. The cycling measurements were taken with the use of the ATLAS 0461 MBI multichannel electrochemical system (Atlas-Sollich, Poland). Impedance spectroscopy measurements were performed using the Gamry 1000 multichannel electrochemical system (USA) at different temperatures. Tested anodes were separated from metal-lithium counter and reference electrode by the glass microfiber GF/A separator (Whatmann, 0.4–0.6-mm thick), all placed in an adopted Swagelok® connecting tube. Typically, the mass of the lithium was ca. 31 mg (0.785 cm2), while that of cathodes was 3.0–4.0 mg. The cells were assembled in a glove box in the dry argon atmosphere. After electrochemical measurements, the cells were disassembled and the cathodes were washed with DMC and dried in vacuum at room temperature. The morphology of electrodes was observed with a scanning electron microscope (SEM, Tescan Vega 5153). The BET surface of pristine electrode materials was determined with an Autosorb iQ apparatus (Quantochrome Instruments, UK) and particle size distribution with a Zetasizer Nano ZS (Malvern Instruments Ltd., UK).

Results and discussion

BET surface, size distribution, and SEM images

Specific surface area of cathode materials from BET analysis was between 15.4 and 2.05 m2 g−1 (Table 1). Particle size distribution is shown in Fig. 1. LiFePO4 and LiCoO2 materials contained particles of diameters between 1 and 4 μm, with the maximum amount of ca. 2 μm. The size distribution of LiMn2O4 particles was broader (2 and 7 μm with a maximum at 4–6 μm). Figures 2, 3, and 4 show SEM images of cathodes (1) after electrode formation but before its cycling and (2) in the discharged state after the second cycle. In the case of all cathodes, their cycling results in the formation of small particles. In the case of LiFePO4 and LiMn2O4, the diameter of particles remained similar while LiCoO2 was converted into material of a smaller diameter.

Table 1 Parameters of tested electrodes (expressed versus mass of active material): mass m, specific BET surface area S BET, real surface area S calculated as m·S BET, passivation film resistance R f, resistance of charge transfer process R ct, exchange current density j o, activation energy of Li+ transport in passivation film (\( {E}_f^{\#} \)), and of charge transfer process (\( {E}_{ct}^{\#} \))
Full size table
Fig. 1
figure 1

Particle size distribution of electrode materials

Full size image
Fig. 2
figure 2

SEM images of LiMn2O4 cathode a before and b after 4 cycles (magnification ×5000)

Full size image
Fig. 3
figure 3

SEM images of LiCoO2 cathode a before and b after 4 cycles (magnification ×5000)

Full size image
Fig. 4
figure 4

SEM images of LiFePO4 cathode a before and b after 4 cycles (magnification ×5000)

Full size image

Impedance and kinetic parameters

Figure 5 shows impedance spectra of tested cathode materials at room temperature. All curves consist of two semicircles at the high frequency region and a straight line at low frequencies (in the case of LiFePO4, only a part of the first semicircle can be seen). The most striking difference is the impedance (resistance) magnitude: tens of ohm in the case of LiFePO4, hundreds of ohm for LiMn2O4, and thousands of ohm for LiCoO2. Impedance spectra were deconvoluted according to the equivalent circuit shown in Fig. 6. It was selected from a library of circuits based on two time constants due to best correlation of fits with experimental data. Passivation film (R f) and charge transfer (R ct) resistances obtained from the deconvolution procedure are shown in Table 1. Both R f and R ct given in ohms are expressed versus the geometrical surface area of electrodes (1.27 cm2). It can be seen from Table 1 that resistance of the passivation film differs by two orders of magnitude depending on the electrode material: it was 4.9 Ω for LiFePO4, ca. 20 times more for LiMn2O4 (102 Ω), and as high as 645 Ω for LiCoO2. Charge transfer resistances (R ct) estimated from the deconvolution procedure were between 6.0 and 88.5 Ω (Table 1). Those resistances can be expressed against the anode real surface area A (estimated from BET measurements S = m·S BET) as R ct·S (expressed in Ωcm2).

Fig. 5
figure 5

Impedance spectra of LiMn2O4, LiCoO2, and LiFePO4 cathodes taken at 298 K. Counter-electrode: lithium metal

Full size image
Fig. 6
figure 6

Equivalent circuit used for impedance spectra deconvolution

Full size image

Charge transfer resistances may be converted into surface area independent of exchange current densities:

$$ {j}_0=\frac{RT}{FS}\frac{1}{R_{ct}} $$
(1)

Both R ct·S and j o values are given in Table 1. It can be seen that while charge transfer resistances differ for one order of magnitude, the corresponding R ct·S values are comparable: from ca. 4.7 to ca. 2.5 kΩcm2. Consequently, exchange current density for all the three cathode materials is also comparable, amounting to 10−2 mAcm−2: from ca. 0.55 10−2 mAcm−2 (LiMn2O4) to ca. 0.94·10−2–1.01 10−2 mAcm−2 (LiCoO2 and LiFePO4). Exchange current densities can be found in the literature for the LiFePO4 material [16–18]. In general, j o values are reported in a broad range between 10−5 mAcm−2 (5.19 10−5 mAcm−2 [17], 2.12 10−5 mAcm−2 [18]), and 10−1 mAcm−2 (1.7·10−1 mAcm−2 [16]). However, a comparison of the present results with literature data is difficult because (i) different solvents or electrolytes were used and (ii) the real surface areas of electrodes were not reported or even not taken into account in calculations. For the other cathodes (LiCoO2 and LiMn2O4), R ct·S and j o values are not available.

Kinetic parameters of cathodes may be compared to the corresponding data for the metallic lithium [9, 25–30] anode. A typical reported value is of the order of 10–10−1 mA cm−2, depending on the solvent, electrolyte, and its concentration [30]. Exchange current density values, expressed versus the active material specific surface, suggest that the kinetics of the charge transfer taking place at the cathode is slower in comparison to metallic lithium, while the surface area is larger. To our knowledge, available literature does not present exchange current densities for LiC6 (graphite) anodes working together with classical LiPF6 electrolytes. Generally, R ct values determined from impedance spectroscopy for a lithiated graphite anode can be found in the literature; however, resistance depends on the electrode size (real surface area), which usually is not mentioned. Therefore, the j o value for LiC6 was measured in the present study in the same way as for cathodes (the impedance spectrum for the LiC6 anode is shown in Fig. 7) and data are listed in Table 1. It can be seen that the exchange current density is comparable to that characteristic of anodes (0.77·10−2 mA cm−2).

Fig. 7
figure 7

Impedance spectrum of LiC6 anode taken at 298 K. Counter-electrode: lithium metal

Full size image

Values of R f and R ct measured at different temperatures and plotted as −lnR = f(T −1) provide electrode size-independent charge transfer activation energy E #. Figure 8 shows the Arrhenius plot for a LiFePO4 cathode as an example. Both \( {E}_f^{\#} \) and \( {E}_{ct}^{\#} \) values for all cathodes are shown in Table 1. It can be seen that activation energies for the Li+ ion conduction in passivation film and charge transfer reaction are similar in the case of LiMn2O4 and LiCoO2 while in the case of the LiFePO4 cathode, both \( {E}_f^{\#} \) and \( {E}_{ct}^{\#} \) are considerably lower. Activation energies for the charge transfer process taking places at cathodes can be found in the literature [11–15]. Again, a comparison of data is difficult due to different solvents, electrolytes, and their concentrations. Some data are reported for polymer electrolytes. In addition, the state of cathodes intercalation is different, which is equivalent to a comparison of different compounds. For example, the \( {E}_{ct}^{\#} \) value found here for the LiCoO2/1 M LiPF6 in EC + DMC system is 48.9 kJmol−1, comparable to that reported for the LiCoO2/1 M LiClO4 or LiCF3SO3 in PC (46–48 kJmol−1 [14]). Both values were measured at a potential of ca. 3.9 V versus metallic lithium. The corresponding activation energy measured at a higher potential (4.2 V), equivalent to a lower degree of intercalation, was reported to be much lower (ca. 25 kJmol−1 [15]).

Fig. 8
figure 8

Arrhenius plot for the charge transfer (R ct) and passivation film (R f) resistances for the LiFePO4 electrode

Full size image

Compatibility of cathodes with the LiC6 anode

Li-ion batteries typically contain a carbon anode, which capacity (for graphite ca. 370 mAh g−1) is usually higher in comparison to that characteristic of cathodes (Table 2). However, the capacity of the CF x material is very high (ca. 900 mAh g−1), and hence, literature kinetic data for this cathode [31] are also shown in Table 2. The ratio of the cathode and graphite mass should be equal to the corresponding ratio of specific capacity (n c/n G = q c/q G) to maximize active material utilization. The ratio m c/m G indicates also the mass of the cathode compatible with 1 g of carbon material. It can be seen from Table 2 that the mass ratio n c/n G for LiMn2O4, LiCoO2, and LiFePO4 cathodes is between 3.08 and 2.18 in contrast to the CF x material (n CFx/n G = 0.41). On the other hand, from the point of view of power, the ratio of anodic and cathodic charge transfer resistance should be close to 1 (a more resistive electrode determines the operation rate). The ratio of cathodic and anodic resistances R c/R G can be calculated from the ratio of electrode surface S G/S c and exchange current densities \( {j}_o^G \)/\( {j}_o^c \):

$$ \frac{R_c}{R_G}=\frac{S_G{j}_o^G}{S_c{j}_o^c} $$
(2)
Table 2 Calculated compatibility parameters of graphite anode (G) and cathodes (c): theoretical capacity q, mass n of cathode material compatible with 1 g of graphite (n c = 1 g·(q G/q c)), real surface area of electrode material S (calculated as n·S BET), ratio of graphite and cathode real surface (S G/S c), exchange current density (\( {j}_o^G \)/\( {j}_o^c \)), and charge transfer resistance (R c/R G)
Full size table

The R c/R G ratio values shown in Table 2 fall within a broad range between 0.13 and 10.04. While LiMn2O4 and LiCoO2 cathodes (R c/R G ≈ 1) are kinetically compatible with the carbon anode, the situation in the case of other cathodes is different. The LiFePO4 material shows resistance by one order of magnitude lower than the equivalent amount of carbon. In contrast, the CF x cathode is characterized by resistance by one order of magnitude higher than the equivalent amount of the anode.

Conclusions

The most striking difference of EIS curves is the impedance magnitude: tens of ohms in the case of LiFePO4, hundreds of ohms for LiMn2O4, and thousands of ohms for LiCoO2. Charge transfer resistances (R ct) for the lithiation/delitiation process estimated from the deconvolution procedure were 6.0 Ω (LiFePO4), 55.4 Ω (LiCoO2), and 88.5 Ω (LiMn2O4). Exchange current density for all the three tested cathodes was found to be comparable (0.55·10−2–1·10−2 mAcm−2, T = 298 K).

Corresponding activation energies for the charge transfer process, \( {E}_{ct}^{\#} \), differed considerably: 66.3, 48.9, and 17.0 kJmol−1 for LiMn2O4, LiCoO2, and LiFePO4, respectively. Consequently, temperature variation may have a substantial effect on j o in the case of LiMn2O4 and LiCoO2 cathodes.