Comparison of structure and electrochemical properties for 5 V LiNi_0.5Mn_1.5O₄ and LiNi_0.4Cr_0.2Mn_1.4O₄ cathode materials

Abstract

Spinel LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ cathode materials have been successfully synthesized by the sol–gel method using citric acid as a chelating agent. The structure and electrochemical performance of these as-prepared powders have been characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), and the galvanostatic charge–discharge test in detail. XRD results show that there is a small Li _y Ni_1-y O impurity peak placed close to the (4 0 0) line of the spinel LiNi_0.5Mn_1.5O₄, and LiMn_1.4Cr_0.2Ni_0.4O₄ has high phase purity, and the powders are well crystallized. SEM indicates that LiMn_1.4Cr_0.2Ni_0.4O₄ has a slightly smaller particle size and a more regular morphological structure with narrow size distribution than those of LiNi_0.5Mn_1.5O₄. Galvanostatic charge–discharge testing indicates that the initial discharge capacities of LiMn_1.4Cr_0.2Ni_0.4O₄ and LiNi_0.5Mn_1.5O₄ cycled at 0.15 C are 129.6 and 130.2 mAh g⁻¹, respectively, and the capacity losses compared to the initial value, after 50 cycles, are 2.09% and 5.68%, respectively. LiMn_1.4Cr_0.2Ni_0.4O₄ cathode has a higher electrode coulombic efficiency than that of the LiNi_0.5Mn_1.5O₄ cathode, implying that Ni and Cr dual substitution is beneficial to the reversible intercalation and de-intercalation of Li⁺.

Introduction

Lithium ion batteries are regarded as promising new power sources for hybrid electric vehicles as well as for portable electronic devices due to their long cycle life and high energy density. Their most promising cathodes are olivine-type LiFePO₄ and spinel LiMn_2-x M _x O₄ (M—transition metal or Mg, Al, etc.) because of their low material cost, high natural abundance, environmental harmlessness, and good safety compared with the LiCoO₂ used in current batteries [1–20]. While LiFePO₄ has been demonstrated to have high reversible capacity and intrinsic safety, thanks to its electrochemically stable phosphate group [1, 2], LiMn_2-x M _x O₄ is a better choice in high-power applications due to its much higher average discharge potential, especially when doped with transition metals. On the other hand, spinel LiMn₂O₄ has a higher potential (4 V vs. Li/Li⁺) and a higher material density (4.2 g cm⁻³) than those of LiFePO₄ (3.45 V vs. Li/Li⁺ and 3.6 g cm⁻¹³, respectively). In addition, LiFePO₄ also suffers from very low electronic conductivity. Transition metal oxide spinels possess rich crystal chemistries due to the ability of transition metals to adopt different valence states. This can be induced by changing the starting composition, annealing temperature, or atmosphere, i.e. oxidizing or reducing. Among LiMn_2-x M _x O₄ (M = transition metal) materials, LiMn_1.5Ni_0.5O₄ shows good cycling stability on repeated lithium ion extraction and insertion, making it the most attractive material for practical application [8, 9]. LiNi_0.5Mn_1.5O₄ shows not only high oxidation voltage but also higher stability than layered lithium cobalt nickel oxide [10]. However, it has been reported that LiNi_0.5Mn_1.5O₄ synthesized at above 650 °C loses oxygen and disproportionates to a spinel with a smaller Ni content and Li _y Ni_1-y O, which leads to a decrease in the nickel content in the spinel phase and a consequent introduction of some Mn³⁺. This is closely related to the oxygen deficiency during the processing[11]. The presence of Mn³⁺ in LiNi_0.5Mn_1.5O₄ causes the development of a 4-V plateau and a decrease in 5-V capacity. In addition, Lee et al. [12] have also reported that the solubility limit of Ni in the LiNi _x Mn_2-x O₄ system is 0.415. It has been reported that Cr³⁺ ions have high oxygen affinity, providing structural stability during cycling [13, 14]. So far, many chemical routes have been used to synthesize spinel LiNi_0.5Mn_1.5O₄ powders. These synthesis methods include solid-state reaction [15], sol–gel [16], emulsion drying [17], carbonate processing [18], and molten salt processes [19], etc. A sol–gel process has been used for the synthesis to obtain homogeneity and narrow particle size distribution of the final LiNi_0.5Mn_1.5O₄ product, and it has been widely used to prepare cathode materials for lithium ion batteries [16, 20]. Wu et al. [21] have reported that the substitution of chromium ions in LiMn_2-x Cr _x O₄ results in a conversion of Cr³⁺ to Cr⁶⁺ for x > 0.2 and in the formation of LiCrO₂ impurity for x > 0.8. In this article, the spinels LiMn_1.4Cr_0.2Ni_0.4O₄ and LiNi_0.5Mn_1.5O₄ were synthesized by the sol–gel method, and their structures and electrochemical behavior were investigated. It provides not only an effective route of synthesis but also a way to modify Li–Mn–O spinel as a promising cathode material.

Experimental

Preparation of materials

The samples of LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ spinel were prepared by a sol–gel method using citric acid as a chelating agent. Stoichiometric amounts of Li(CH₃COO)·2H₂O(AR, 99%), Mn(CH₃COO)₂·6H₂O(AR, 99%), Ni(NO₃)₂·6H₂O(AR, 99%) and Cr(NO₃)₃·9H₂O (AR, 99%) were separately dissolved in citric acid solution with continuous stirring, and the molar ratio of citric acid to total metal cations was fixed at 1:1. The pH of the mixed solution was adjusted to 7 by adding NH₃·H₂O (AR, 25%). The mixed solution was then heated with continuous stirring up to boiling point until a gel was obtained. The gel was dried at about 110 ^NC for 12 h in a vacuum drying oven, resulting in the formation of amorphous powders. The powder was again dried at 850 ^NC for 20 h in air to obtain the final spinel product.

Physical characterization of materials

Powder X-ray diffraction (XRD) was performed using a Rigaku D/MAX-RC X-ray diffractometer (Hitachi, Japan) with Cu Kα₁ (45 kV, 50 mA, step size = 0.02°, 10° < 2θ < 90°) monochromated radiation to identify the crystalline phase of the materials. The particle morphologies of the samples were examined using a scanning electron microscope (Hitachi, S-4000 and S-570).

Electrochemical performance of materials

Charge–discharge performance of the cell was characterized galvanostatically on Land 2000T (Wuhan, China) tester at 0.15 C charge–discharge rate between 3.5 and 4.98 V (vs. Li/Li⁺). The upper limit was set at 4.98 V to prevent increasing the charge capacity by effect of side reactions such as release of oxygen from the spinel lattice above 5.0 V.

Preparation of lithium ion batteries

The coin cell cathode paste was prepared by mixing 85 wt.% of active material with 5 wt.% of polyvinylidene fluoride and 10 wt.% of acetylene black in N-methyl-2-pyrrolidinone and, subsequently, a slurry was made. The mixed slurry was coated onto an aluminum current collector and vacuum-dried at 110 °C overnight before use. Coin cells were prepared in a dry argon atmosphere inside a glove box. The cathode and Li anode were separated inside the coin cell using Celgard 2300 membrane. Commercial LiPF₆ (1 M) in a 1:1 (by volume) mixture of ethylene carbonate and dimethyl carbonate was used as electrolyte.

Results and discussion

Micrographs and structure analysis

Figure 1 shows the XRD patterns of as-prepared LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ powders. All the peak signatures of the XRD pattern for LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ conform to JCPDS card No-35-782, suggesting that they are both single-phase spinel compounds. The sharp peaks in the patterns show good crystallinity of the cathodes. In addition to the spinel reflections, the pattern shows a few weak reflections corresponding to the impurity phase Li _x Ni_1-x O having the rock salt structure, as is observed in Fig. 1a, which is in agreement with previous reports [11], indicating that Ni substitution alone did not help to form a pure spinel structure. The reflections marked with (h, k, l) values refer to the spinel phase, and those marked with (*) refer to the impurity phase Li _y Ni_1-y O. The formation of the Li _y Ni_1-y O impurity phase results in a decrease in Ni content and Mn valence in the spinel phase as shown by the generalized reaction:

$${\text{LiMn}}_{{\text{1}}{\text{.5}}} {\text{Ni}}_{{\text{0}}{\text{.5}}} {\text{O}}_{\text{4}} \to \alpha {\text{Li}}_y {\text{Ni}}_{{\text{1 - y}}} {\text{O + }}\beta {\text{LiMn}}_{{\text{1}}{\text{.5 + }}x} {\text{Ni}}_{{\text{0}}{\text{.5 - }}x} {\text{O}}_{\text{4}} {\text{ + }}\gamma {\text{O}}_{\text{2}} $$

(1)

where α, β, and γ define, respectively, the relative amounts of the Li _y Ni_1-y O, LiNi_0.5-x Mn_1.5+x O₄, and O₂ phases. This observation suggests that the solubility limit of Ni in the LiMn_2-x Ni _x O₄ spinel phase may be x < 0.5, which is in agreement with the previous reports [12]. The integration of the XRD pattern shows that the content of Li _y Ni_1-y O is approximately 1% of the total spinel phase, from the peak area comparison in the diffractograms. LiMn_1.4Cr_0.2Ni_0.4O₄ materials are confirmed as a cubic spinel structure with a space group of Fd3m in which lithium ions occupy the tetrahedral (8a) sites, transition metals (Ni, Cr and Mn) are located at the octahedral (16d) sites, and oxygen atoms reside in the 32e sites, as shown in Fig. 1b. This indicates that the Mn site in LiMn₂O₄ can be fully substituted by Ni and Cr. The results show that the Cr doped LiMn_1.4Cr_0.2Ni_0.4O₄ contains more oxygen than that of undoped LiNi_0.5Mn_1.5O₄, implying that Cr plays a crucial role in the maintenance of the oxygen in the cathode. The lattice parameters and unit cell volumes of these compounds are listed in Table 1. The lattice parameters were calculated through the least square program method from the diffraction data of LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ and were found to be about 8.174 and 8.169 Å, respectively, which is slightly less than that of a pure LiMn₂O₄ cathode material [8.24762(16) Å (35-0782 JCPDS file)]. This prevents a phase transition of the material. However, LiMn_1.4Cr_0.2Ni_0.4O₄ has a smaller lattice parameter than that of LiNi_0.5Mn_1.5O₄. This may be explained by the fact that: (1) the ionic radius of Cr³⁺ ion (0.615 Å) is smaller than those of Mn³⁺(0.66 Å) and Ni (0.69 Å)[22], leading to the shrinkage of the spinel framework; (2) the bonding energy of Cr–O (427 kJ mol⁻¹) is stronger than that of Mn–O (402 kJ mol⁻¹) and Ni–O (391.6 kJ mol⁻¹), revealing that the former has a higher octahedral site preference energies [22, 23], which results in the diminution of the bond length. The stronger Cr–O bond can also stabilize the spinel structure by assisting retention of the local symmetry during cycling.

Fig. 1

Typical XRD patterns for the samples: a LiNi_0.5Mn_1.5O₄; b LiMn_1.4Cr_0.2Ni_0.4O₄

Table 1 Structural parameters of LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄

In a spinel-framework structure having space group symmetry of Fd3m, the mean bond length of cations (R) is given by the following equations:

$$R_{{\text{Mn - Mn}}\left( {{\text{Ni, Cr}}} \right)} = \frac{{\sqrt 2 }}{4}a$$

(2)

$$R_{{\text{Mn}}\left( {{\text{Ni, Cr}}} \right){\text{ - O}}} = a\sqrt {3u^2 - 2u + 0.375} $$

(3)

where a is the lattice parameter of the spinel phase, and u is the oxygen positional parameter (u, 0.267) [24]. The calculated results are recorded in Table 1. The results show that the substitution of Cr shortens the mean bond length of cations, increases the mean bond energy, and improves structural stability. This means that the Cr-doped LiMn_1.4Cr_0.2Ni_0.4O₄ may have excellent electrochemical cycle stability. Ohzuku et al. [25] have reported that any occupancy of the substituent ions in the 8a tetrahedral lithium sites will lead to unfavorable electrochemical characteristics. According to Ohzuku et al. [25], the integrated intensity ratios of the (4 0 0)/(3 1 1) and (2 2 0)/(3 1 1) peaks are indices of the extent of occupancy of the substituent ions in the 8a lithium sites. There is an obscure (2 2 0) peak (at 2θ≈31°) in LiMn_1.4Cr_0.2Ni_0.4O₄ as shown in Fig. 1, which is associated with the presence of heavy cations in tetrahedral 8a sites of the spinel-type structure. However, in the LiMn_1.4Cr_0.2Ni_0.4O₄ samples, the integrated intensity ratio of the (4 0 0)/(3 1 1) increases compared to LiNi_0.5Mn_1.5O₄. This suggests that nickel shows a propensity to occupy the 8a lithium sites in LiNi_0.5Mn_1.5O₄ due to the Li _y Ni_1-y O impurity.

Figure 2 shows the scanning electron microscopy (SEM) image of these materials. LiMn_1.4Cr_0.2Ni_0.4O₄ powders have a uniform, nearly cubic structural morphology with narrow size distribution under 1 μm. These can be attributed to the action of the chelating agent citric acid, which forms a complex network where the metal ions are uniformly distributed in the matrix. Thus, it prevents phase separation and leads to the formation of homogeneous sized particles during thermal decomposition. However, LiNi_0.5Mn_1.5O₄ powders have a wide particle size distribution ranging from 1 to 5 μm. The result suggests that Cr ions are successively substituted for Mn in the LiMn_1.4Cr_0.2Ni_0.4O₄ host structure, resulting in sufficient contact between active materials and electrolyte and resulting in favorable diffusion and transmission of Li⁺ in the electrode.

Fig. 2

SEM pictures of LiNi_0.5Mn_1.5O₄ (a) and LiMn_1.4Cr_0.2Ni_0.4O₄ (b)

Electrochemical performance analysis

Figure 3 shows the room temperature charge–discharge characteristics of coin cells carried out galvanostatically between 3.5- and 4.98-V cutoff limits, and Table 2 lists the first discharge capacities obtained in different voltage regions. From the charge–discharge curves, it can be seen that LiMn_1.4Cr_0.2Ni_0.4O₄ exhibits three potential plateaus, but LiNi_0.5Mn_1.5O₄ shows only two potential plateaus. One is around the 4.1-V range, which corresponds to the redox reaction involving Mn³⁺/Mn⁴⁺. The other plateaus are observed in the 4.7- and 4.8-V potential regions. These two plateaus correspond to the redox reactions involving Ni²⁺/Ni⁴⁺ and Cr³⁺/Cr⁴⁺, respectively. The delithiation (charge)/lithiation (discharge) reactions into/from spinel LiMn₂O₄, LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ proceed reversibly according to the following equations:

$${\text{Li}}^{\text{ + }} {\text{Mn}}^{{\text{3 + }}} {\text{Mn}}^{{\text{4 + }}} {\text{O}}_{\text{4}} \rightleftharpoons {\text{Mn}}_{\text{2}}^{{\text{4 + }}} {\text{O}}_{\text{4}} {\text{ + Li}}^{\text{ + }} {\text{ + e}}^ - $$

(4)

$${\text{Li}}^{\text{ + }} {\text{Mn}}_{{\text{1}}{\text{.5}}}^{{\text{4 + }}} {\text{Ni}}_{{\text{0}}{\text{.5}}}^{{\text{2 + }}} {\text{O}}_{\text{4}} \rightleftharpoons {\text{Mn}}_{{\text{1}}{\text{.5}}}^{{\text{4 + }}} {\text{Ni}}_{{\text{0}}{\text{.5}}}^{{\text{4 + }}} {\text{O}}_{\text{4}} {\text{ + Li}}^{\text{ + }} {\text{ + e}}^ - $$

(5)

$${\text{LiMn}}_{{\text{1}}{\text{.4}}}^{{\text{4 + }}} {\text{Cr}}_{0.2}^{{\text{3 + }}} {\text{Ni}}_{{\text{0}}.4}^{{\text{2 + }}} {\text{O}}_{\text{4}} \rightleftharpoons {\text{Mn}}_{{\text{1}}{\text{.4}}}^{{\text{4 + }}} {\text{Cr}}_{0.2}^{{\text{4 + }}} {\text{Ni}}_{{\text{0}}.4}^{{\text{4 + }}} {\text{O}}_{\text{4}} {\text{ + Li}}^{\text{ + }} {\text{ + e}}^{\text{ - }} $$

(6)

Fig. 3

First discharge profiles of LiNi_0.5Mn_1.5O₄ (a) and LiMn_1.4Cr_0.2Ni_0.4O₄ (b) series of spinel oxides, recorded at 0.15 C discharge rate between 4.98 and 3.5 V

Table 2 Cycle performance data of the prepared powders for the first cycle and the 50th cycle

As shown by these equations, the charge–discharge reactions may be classified into three processes: Eq. 4 corresponding to Mn³⁺/Mn⁴⁺, Eq. 5 corresponding to Ni²⁺/Ni⁴⁺, and Eq. 6 corresponding to Ni²⁺/Ni⁴⁺ and Cr³⁺/Cr⁴⁺, respectively. The initial capacities of LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ are 130.2 and 129.6 mAh g⁻¹, respectively. It is clear that both samples have nearly equal initial capacities, but LiMn_1.4Cr_0.2Ni_0.4O₄ has a higher discharge potential plateau than that of LiNi_0.5Mn_1.5O₄. In addition, LiMn_1.4Cr_0.2Ni_0.4O₄ has higher discharge capacity (at 5–4.1 V) than that of LiNi_0.5Mn_1.5O₄, so that the former has a better electrochemical performance than the latter, and it also indicates that the latter has a higher content of Mn³⁺ ions. The atomic radius of Mn⁴⁺ ion is smaller than that of Mn³⁺ ion, which means that LiMn_1.4Cr_0.2Ni_0.4O₄ has a smaller lattice parameter than that of LiNi_0.5Mn_1.5O₄ because the latter contains more Mn³⁺ ion than the former, which is consistent with the XRD results in Table 1. This prevents the phase transition of the spinel material. The differences in Mn³⁺ ion content between LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ materials may be due to the high chemical affinity of Cr for oxygen [26], which provides excess capacity in the 4.8-V region, stable intercalation in the high voltage range.

Figure 4 shows typical galvanostatic charge–discharge cycling performance curves for electrodes cycled at 0.15 C made from prepared powders. Table 2 summarizes their cycling performance. The capacity loss was calculated after the first 50 cycles, which is defined as (C ₁ – C _n )/C ₁ × 100%, where C ₁ and C _n are the discharge capacities of the first and nth cycle, and n is the number of cycles. It is found that both samples have better cycling performance, as shown in Fig. 4, which is related to the presence of the dopant ions. This may be explained that the Jahn–Teller distortion is effectively constrained by the substitution of Ni²⁺ and Cr³⁺ for Mn³⁺ ions. LiMn_1.4Cr_0.2Ni_0.4O₄ has a larger discharge capacity after two cycles and much lower capacity loss rate after 50 cycles than those of LiNi_0.5Mn_1.5O₄, as is shown in Table 2. The reason for this may be as follows. It has been reported that HF generated during cycling when using LiPF₆-based electrolyte was responsible for the dissolution of manganese [27]. In addition, Mn³⁺ is considered to be the main source of the dissolution of manganese via Hunter’s reaction [28]:

$${\text{4H}}^{\text{ + }} + 2{\text{LiMn}}^{{\text{3 + }}} {\text{Mn}}^{{\text{4 + }}} {\text{O}}_{\text{4}} \to 3\left( {\lambda {\text{ - MnO}}_{\text{2}} } \right){\text{ + Mn}}^{{\text{2 + }}} {\text{ + 2Li}}^{\text{ + }} + 2{\text{H}}_{\text{2}} {\text{O}}$$

(7)

Fig. 4

Cycling performance of different cathode materials: a LiMn_1.4Cr_0.2Ni_0.4O₄ and b LiNi_0.5Mn_1.5O₄

The protonated λ-MnO₂ cannot act as a host for lithium ions during cycling because of the strong binding energy of the protons to the oxygen sites around the 16d Mn [29]. Furthermore, F⁻ can also trap soluble Mn²⁺ and lead to the precipitation of MnF₂. At the same time, the deliquescent manganese fluoride deposits on the surface of negative electrode on which the passivation layer forms. This reduces conductance and blocks the transport of electrons or ions [30], reducing the electrochemical performance of LiNi_0.5Mn_1.5O₄. According to the analysis in Fig. 3, doping with Ni and Cr ions reduces the Mn³⁺ content in LiMn_1.4Cr_0.2Ni_0.4O₄, slowing the dissolution of Mn³⁺ ions and thereby increasing cycling performance. Furthermore, better crystallinity and more regular morphology of LiMn_1.4Cr_0.2Ni_0.4O₄ powders would help to release the stresses generated by the repetitive Li⁺ intercalation; smaller particles (see Fig. 2b) can provide more interfacial area for contact within the liquid electrolyte and hence can increase the opportunity for lithium ions to intercalate into the host structure [31]. These are consistent with the XRD and SEM results discussed above; moreover, LiMn_1.4Cr_0.2Ni_0.4O₄ has a higher theoretical capacity than that of LiMn_1.5Ni_0.5O₄ due to the low molecular weight of Cr compared with that of Ni [32].

To study the influence of the dopant ions on the electrochemical behavior of spinel lithium manganese oxide, the result of the coulomb efficiency of both electrodes is plotted in Fig. 5. This efficiency is defined as the discharge capacity divided by the charge capacity in one charge/discharge cycle. It is interesting that the coulombic efficiencies of both electrodes in the first charge–discharge cycle are less than 85%, indicating that a small fraction of lithium ions is incapable of intercalating back into the host structure due to electrolyte decomposition at high voltage. However, after about two cycles, the coulomb efficiencies of both electrodes are increased to nearly a constant value (more than 90%). The improved cycling efficiency of the samples after the first charge–discharge process may be attributed to the formation of protective layer on the electrode surface. It is obvious that the LiMn_1.4Cr_0.2Ni_0.4O₄ electrode has a higher mean coulombic efficiency than that of the LiNi_0.5Mn_1.5O₄ electrode. This result implies that Ni and Cr dual substitution is beneficial to the reversible intercalation and de-intercalation of Li⁺. It also shows that the LiMn_1.4Cr_0.2Ni_0.4O₄ material has a better electrochemical performance than LiNi_0.5Mn_1.5O₄, which may be related to its better crystallinity and more regular morphology. The features of LiMn_1.4Cr_0.2Ni_0.4O₄ examined here are very desirable for its use as a cathode to improve the electrochemical properties of lithium ion batteries.

Fig. 5

Coulombic efficiency of LiNi_0.5Mn_1.5O₄ (a) and LiMn_1.4Cr_0.2Ni_0.4O₄ (b) in 50 cycles

Conclusions

LiNi_0.5Mn_1.5O₄ and LiMn_1.4Cr_0.2Ni_0.4O₄ cathode materials were successfully synthesized by sol–gel method. The I(400)/I(311)ratio as well as the lattice parameter and smaller particle size of LiMn_1.4Cr_0.2Ni_0.4O₄ are expected to show better electrochemical performance than that of LiMn_1.5Ni_0.5O₄. LiMn_1.4Cr_0.2Ni_0.4O₄ has a uniform and narrow size distribution under 1 μm. The substitutions of a small amount of Cr³⁺ simultaneously for both Mn⁴⁺ and Ni²⁺ to give LiMn_1.4Cr_0.2Ni_0.4O₄ are found to suppress the 4-V plateau and increase the 5-V capacity. LiMn_1.4Cr_0.2Ni_0.4O₄ shows a better combination of high 5-V capacity (120.5 mAh g⁻¹ at 5–4.1 V), excellent capacity retention (97.91% in 50 cycles), and coulombic efficiency. The excellent performance of LiMn_1.4Cr_0.2Ni_0.4O₄ coupled with its low cost may make it attractive for high-power applications such as electric and hybrid vehicles.