In situ X-ray diffraction studies on nominal composition of C₂Li under high pressure and temperature

Abstract

Superdense phase between graphite and Li metal, $\hbox {C}_{2}$Li, has been significant in the research on both lithium-ion batteries (LIBs) and graphite intercalated compounds (GICs). However, a detailed method for synthesizing $\hbox {C}_{2}$Li remains unknown owing to the limited information regarding $\hbox {C}_{2}$Li and difficulties in distinguishing $\hbox {C}_{2}$Li from $\hbox {C}_{6}$Li. Thus, we performed in situ X-ray diffraction measurements on samples with the nominal composition of $\hbox {C}_{2}$Li under high pressures and temperatures of up to 10 GPa and 400 $^{\circ }\hbox {C}$, respectively. We employed two types of $\hbox {C}_{2}$Li samples; one was a mixture of $\hbox {C}_{6}$ graphite powder and Li metal (C$_{6}+3$Li), and the other was a mixture of $\hbox {C}_{6}$Li and Li metal in which the $\hbox {C}_{6}$Li was prepared by the electrochemical discharge (reduction) reaction that occurs in LIBs. Considering changes in the d-value based on the 001 diffraction peak from $\hbox {C}_{6}$Li or $\hbox {C}_{2}$Li, C$_{6}$Li + 2Li is suitable for synthesizing $\hbox {C}_{2}$Li, although the nonaqueous electrolyte used for the electrochemical reaction should be removed to avoid structural transformations to lower-stage compounds such as $\hbox {C}_{12}$Li and $\hbox {C}_{18}$Li during the heating. These findings pave the way toward a method for synthesizing C$_{2}$Li, which could increase the energy density of LIBs and establish GICs with novel physical and electronic properties.

Introduction

Lithium-ion batteries (LIBs), in which Li$^{+}$ ions move back and forth between positive and negative electrode materials without destroying the core structures, are behind the scenes, powering today’s information technology and sustainability initiatives^1,2,3. Since the commercialization of LIBs in 1991, positive electrode materials with various structures and compositions have been introduced, such as LiCoO$_{2}$, Li[Ni$_{1/2}$Mn$_{3/2}$]O$_{4}$, and LiFePO$_{4}$^4,5,6. In contrast, graphite (C$_{6}$) has remained at the forefront negative electrode material for LIBs, owing to its extremely low operating voltage of $\sim$0.2 V vs. Li$^{+}$/Li and large rechargeable capacity of more than 300 mAh g$^{-1}$^7,8,9. The electrochemical discharge (reduction) and charge (oxidation) reactions of C$_{6}$ is represented by

$$\begin{aligned} \textrm{C}_{6} + \textrm{Li}^{+} + e^{-} \overset{\textrm{charge}}{\underset{\textrm{discharge}}{\leftrightarrows }} \textrm{C}_{6}\textrm{Li}, \end{aligned}$$

(1)

where the theoretical capacity ($Q_\textrm{theo}$) is calculated to be 372 mAh g$^{-1}$. As shown in Fig. 1a, Li$^{+}$ ions are intercalated between each graphite layer, forming a first-stage compound $\hbox {C}_{6}$Li with the P6/mmm space group^7,8. The repeated distance between the Li$^{+}$ layers along the $c_\textrm{h}$-axis, $I_\textrm{c}$, is approximately 3.74 Å, while the distance between Li$^{+}$ ions within the layer, which corresponds to the lattice parameter of the a-axis (= $a_\textrm{h}$), is approximately 4.30 Å^7,8.

Besides their application to LIBs, $\hbox {C}_{6}$Li and its related compounds have received a great deal of attention as graphite intercalated compounds (GICs) from the viewpoints of their physical and electronic properties¹⁰. Particularly under high pressure (HP) at 3–8 GPa and at high temperature (HT) above $\sim$300 $^{\circ }\hbox {C}$, C$_{6}$Li can reportedly accommodate more Li$^{+}$ ions until reaching the $\hbox {C}_{2}$Li composition^{11,12,13,14,15,16,17}. Meanwhile, in contrast to these moderate HP and HT conditions, the formation of $\hbox {C}_{2}$Li has been proposed in a diamond anvil cell at 27.5 GPa and $\sim$1500 $^{\circ }\hbox {C}$ together with several impurities such as Li$_{2}$O¹⁸. Table 1 summarizes HP-HT synthetic conditions and characterization methods on previous $\hbox {C}_{2}$Li studies. As shown in Fig. 1b, the space group and $I_\textrm{c}$ of $\hbox {C}_{2}$Li are identical to those of $\hbox {C}_{6}$Li, but $a_\textrm{h}$ of $\hbox {C}_{2}$Li is 1/$\sqrt{3}$ that of $\hbox {C}_{6}$Li because of the superdense arrangement of Li$^{+}$ ions between the graphite layers^13,14. Figure 1c shows the simulated X-ray diffraction (XRD) patterns for $\hbox {C}_{6}$, $\hbox {C}_{6}$Li, and $\hbox {C}_{2}$Li with respect to the distance between lattice planes (d). The XRD patterns of $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li resemble each other because of the small atomic scattering factor of Li atoms. Similarities between $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li are also observed in their physical and electronic properties^{12,13,14,15,16}. For instance, Belash et al.¹² reported a “smeared” superconducting transition temperature ($T_\textrm{c}$) of $\hbox {C}_{2}$Li at $\sim$1.9 K, meaning that its magnetic susceptibility gradually decreased from $\sim$2 K to 0.35 K, but it never rapidly dropped below $T_\textrm{c}$ as it does in the case of C$_{2}$Na ($T_\textrm{c}$ $\simeq$ 5 K)¹⁹ and $\hbox {C}_{8}$K ($T_\textrm{c}$ $\simeq$ 2.4 K)²⁰. Thus, the magnetic property of $\hbox {C}_{2}$Li is similar to that of $\hbox {C}_{6}$Li, which has demonstrated temperature-independent Pauli paramagnetic behavior down to 5 K^21,22.

Fig. 1

Characteristics and preparation of $\hbox {C}_{2}$Li. (a, b) Schematic crystal structures of $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li along the $c_\textrm{h}$-axis (left) and in the graphite layer (right). $I_\textrm{c}$ is the repeated distance between the Li$^{+}$ ions layers along the $c_\textrm{h}$-axis. The $a_\textrm{h}$ values of C$_{6}$Li and $\hbox {C}_{2}$Li are approximately 4.30 and 2.49 Å, respectively. (c) Simulated XRD patterns for $\hbox {C}_{6}$ ($P6_{3}/mmc$), $\hbox {C}_{6}$Li (P6/mmm), and $\hbox {C}_{2}$Li (P6/mmm) with respect to the d value. (d) Schematic of the procedure for synthesizing $\hbox {C}_{2}$Li in this study. Two samples of $\hbox {C}_{2}$Li were initially prepared in the nominal composition through two different methods, and the resulting samples are referred to as $\hbox {C}_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li. Then, each sample was installed in a HP capsule (Ta), pressurized at 10 GPa, and heated at 400 $^{\circ }\hbox {C}$ for 30 min. Energy-dispersive XRD measurements were conducted during the HP-HT synthesis. (e) Discharge curve of the C$_{6}$/Li cell for preparing $\hbox {C}_{6}$Li$_{x}$ with x = 0.98. Inset: $\hbox {C}_{6}$Li$_{0.98}$ pellet just after the discharge reaction.

Although $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li are structurally and physically similar, $\hbox {C}_{2}$Li is more attractive as a long-awaited negative electrode material for high-energy-density LIBs thanks to its larger $Q_\textrm{theo}$ (= 1116 mAh g$^{-1}$). Indeed, Bindra et al.¹⁷ reported a charge capacity of 910 mAh g$^{-1}$ using C$_{2}$Li prepared at 5 GPa and 300 $^{\circ }\hbox {C}$. However, the electrochemical charge reaction of this $\hbox {C}_{2}$Li compound is controversial, because the three plateaus observed at 91, 132, and 215 mV in $\hbox {C}_{2}$Li resembled the typical three plateaus observed with $\hbox {C}_{6}$Li at 89, 126, and 219 mV, respectively^7,17. These plateaus indicate the possibility that the $\hbox {C}_{2}$Li compound was not a single-phase of C$_{2}$Li but a mixture of $\hbox {C}_{6}$Li and Li phases or mixture of $\hbox {C}_{2}$Li, $\hbox {C}_{6}$Li, and Li phases. According to all of the papers on $\hbox {C}_{2}$Li published thus far^{11,12,13,14,15,16,17,18}, only three research groups have synthesized $\hbox {C}_{2}$Li using the HP method. In addition, methods for reliably preparing $\hbox {C}_{2}$Li and the reaction mechanisms that occur during its synthesis are still unknown, although Semenenko et al.¹¹ illustrated a HP assembly for preparing $\hbox {C}_{2}$Li in which a Li metal sheet was sandwiched between two graphite sheets. Meanwhile, $\hbox {C}_{2}$Li compounds have been prepared by ball-milling, but they contained defects and impurities such as $\hbox {C}_{12}$Li and $\alpha$-Li$_{3}$N^23,24.

In this study, we conducted in situ XRD studies on $\hbox {C}_{2}$Li samples with the nominal composition under HP (10 GPa) and HT (400 $^{\circ }$C) at the synchrotron radiation facility SPring-8. Understanding the structural changes in these samples will help optimize the synthetic conditions for $\hbox {C}_{2}$Li, thus enabling materials innovation in the development of both LIBs and GICs. As shown in Fig. 1d, we employed two $\hbox {C}_{2}$Li samples with the nominal composition. In one sample, $\hbox {C}_{6}$ graphite powder and Li metal were mixed in the appropriate proportions to reach the C$_{2}$Li composition (denoted as $\hbox {C}_{6}+3$Li). The other was a mixture of $\hbox {C}_{6}$Li$_{x}$ (x $\simeq$ 1) and Li metal, where C$_{6}$Li$_{x}$ was initially prepared by the electrochemical reaction that occurs in LIBs (denoted as $\hbox {C}_{6}$Li + 2Li). To advance the formation of $\hbox {C}_{6}$Li$_{x}$, the homogeneous reaction between the graphite layers and Li metal is expected to outperform simple mixture of $\hbox {C}_{6}$ and Li metal.

Results

Before discussing the results of the in situ XRD measurements, we first briefly describe the procedure of sample preparation. As shown in Fig. 1d, $\hbox {C}_{6}+3$Li was prepared by mixing graphite powder with Li metal in the C/Li molar ratio of 2:1 in an argon-filled glove-box. On the other hand, $\hbox {C}_{6}$Li was prepared by an electrochemical reaction using a nonaqueous electrolyte of 1 M LiPF$_{6}$ dissolved in a 1:1 (v/v) mixture of ethylene carbonate (EC) and diethylene carbonate (DEC), denoted as LiPF$_{6}$(EC+DEC). Fig. 1e shows the discharge curve of the C$_{6}$/Li cell operated at a current of 0.3 mA and 25 $^{\circ }\hbox {C}$. The cell voltage rapidly drops from $\sim$3.2 V to $\sim$0.8 V at the beginning of the discharge reaction and then remains at consecutive plateaus at $\sim$170, 99, and 68 mV as discharging proceeds. An enlarged discharge curve is shown in the inset of Fig. 1e. The plateau at $\sim$99 mV corresponds to the two-phase reaction between $\hbox {C}_{18}$Li (third-stage) and $\hbox {C}_{12}$Li (second-stage) represented by

$$\begin{aligned} 2\textrm{C}_{18}\textrm{Li} + \textrm{Li}^{+} + e^{-} \rightarrow 3\textrm{C}_{12}\textrm{Li}, \end{aligned}$$

(2)

whereas the plateau at $\sim$68 mV reflects the two-phase reaction between $\hbox {C}_{12}$Li (second-stage) and $\hbox {C}_{6}$Li (first-stage) described by^7,22

$$\begin{aligned} \textrm{C}_{12}\textrm{Li} + \textrm{Li}^{+} + e^{-} \rightarrow 2\textrm{C}_{6}\textrm{Li}. \end{aligned}$$

(3)

After discharging down to 0.02 V, the pressed graphite pellet appeared golden yellow, as shown in the inset of Fig. 1e. This indicates the homogeneous reaction between the graphite layers and Li metal. The x value in $\hbox {C}_{6}$Li$_{x}$ was calculated to be 0.98 based on the observed discharge capacity (= 364.1 mAh g$^{-1}$) and $Q_\textrm{theo}$. The $\hbox {C}_{6}$Li$_{0.98}$ pellet was rinsed with DEC twice, dried under vacuum for 1 h, and then mixed with Li metal in the argon-filled glove-box to adjust the C/Li molar ratio to 2:1

Table 1 HP-HT synthetic conditions and chactarization methods for $\hbox {C}_{2}$Li.

Powder samples of $\hbox {C}_{6}$, $\hbox {C}_{6}+3$Li, and $\hbox {C}_{6}$Li + 2Li were packed into a Ta capsule, which was installed in a pyrophyllite pressure medium in a cubic shape with a side length of 6 mm (Figure S1a). In situ XRD data were recorded at the BL14B1 beamline at SPring-8, Japan (Figure S1b). The pyrophyllite pressure medium was set in a cubic-type multi-anvil apparatus (Fig. 1d) and then immediately compressed up to $\sim$1.5 GPa at room temperature (RT) for $\hbox {C}_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li to avoid undesired reactions with moisture and oxygen in the air. Each sample was then pressurized up to 10 GPa at RT over a period of $\sim$2 h, then heated up to 400 $^{\circ }$C at 25 $^{\circ }\hbox {C}$ min$^{-1}$, where it was held for 30 min. Finally, it was de-pressurized down to ambient pressure over $\sim$2 h after quenching to RT from 400 $^{\circ }\hbox {C}$. Present HP-HT condition of 10 GPa and 400 $^{\circ }\hbox {C}$ is similar to those for previous studies on $\hbox {C}_{2}$Li^{11,12,13,15,17} except for one using diamond anvil apparatus (Table 1)¹⁸. Slightly higher pressure and temperature conditions were selected for homogeneous reactions between $\hbox {C}_{6}$Li (or $\hbox {C}_{6}$) and Li samples. Energy-dispersive-type in situ XRD measurements were performed during pressurizing, heating, and de-pressurizing. The experimental setup is described in detail in the materials and methods section and elsewhere^25,26,27.

The results of the in situ XRD measurements of $\hbox {C}_{6}$ graphite are illustrated in Figures S2a–S2e to gain fundamental data for $\hbox {C}_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li and confirm the consistency of our results with those of a previous HP study on graphite²⁹. The 002 diffraction peak shifts to a lower d value during pressurizing and returns to a higher d value during de-pressurizing (Figure S2a). Specifically, the $d_{002}$ value decreases from 3.246(1) Å at P = 1.5 GPa, then remains almost constant ($\sim$2.96 Å) at P = 10 GPa, and finally increases to 3.158(1) Å at P = 2 GPa (Figures S2b–S2d). This pressure dependence of $d_{002}$ is similar to that observed previously using a Drickamer-type apparatus²⁹.

Fig. 2

Results of the in situ XRD measurements of $\hbox {C}_{6} +3$Li. (a) Selected XRD patterns during pressurization and heating: initial state at P = 1.5 GPa and RT, before heating at P = 10 GPa, after heating at P = 10 GPa, and after heating at P = 1.5 GPa. The XRD patterns include the 001 and 002 diffraction peaks from $\hbox {C}_{6}$Li, the 002 diffraction peak from $\hbox {C}_{6}$, and the 110 diffraction peak and corresponding characteristic radiation K$\alpha$ and K$\beta$ peaks from the Ta capsule. (b) Time dependence of $d_{001}$. (c) Time dependence of temperature (left) and pressure (right) during the in situ XRD measurements. (d) The enlarged time dependence of $d_{001}$ together with the time dependence of temperature (right) are taken from the blue bands in (b) and (c). (e) Pressure dependence of $d_{001}$. The right and left arrows indicate the directions of pressurization and de-pressurization, respectively.

Figure 2a shows selected in situ XRD patterns of $\hbox {C}_{6}+$3Li, i.e., the initial state at P = 1.5 GPa and RT, before heating at P = 10 GPa, after heating at P = 10 GPa, and after heating at P = 1.5 GPa. Figure S3 shows the corresponding enlarged XRD patterns at $d \le$ 2 Å. According to the XRD pattern of the initial state, $\hbox {C}_{6}+3$Li is a mixture of $\hbox {C}_{6}$Li and $\hbox {C}_{6}$ (graphite), where Li$^{+}$ ions are intercalated between the graphite layers during mixing with the Li metal. Mechanical intercalation of Li$^{+}$ ions between the graphite layers originates from the soft and lubricating characteristics of Li metal and graphite, respectively^10,23,24,28. All the diffraction peaks, such as the 001 and 002 diffraction peaks from C$_{6}$Li, shift to lower d values at P = 10 GPa, but they almost return to their original d values after heating at P = 1.5 GPa. The characteristic radiation K$\alpha$ and K$\beta$ peaks and 110, 200, 211, and 220 diffraction peaks from the Ta capsule are also observed, in addition to the diffraction peaks from $\hbox {C}_{6}$Li and $\hbox {C}_{6}$ (see Figs. 2a and S3).

Figure 2b shows the time dependence of $d_{001}$, and Fig. 2c shows the time dependence of temperature (left) and pressure (right). The $d_{001}$ value gradually decreases from 3.633(1) Å at P = 1.5 GPa to 3.426(1) Å at 10 GPa, then ranges between $\sim$3.425 and 3.415 Å during heating at this pressure, and finally increases to 3.639(1) Å after heating at P = 1.5 GPa. The $d_{001}$ value corresponds to $I_\textrm{c}$ for both $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li (Figs. 1a and b). The decreasing trend in $d_{001}$ slightly reverses at $\sim$50 min, which corresponds to P $\simeq$ 4 GPa. As shown in Figures S2b and S2e, the $d_{002}$ value of C$_{6}$ graphite simply decreases during pressurization. Although the origins of the slight reverse are currently unclear, the rearrangement of $\hbox {C}_{6}$Li during pressurization is likely to contribute to such changes in $d_{001}$. A similar change in $d_{001}$ is also observed at $\sim$280 min ($\sim$3 GPa) during de-pressurization (Fig. 2b).

Figure 2d shows a close-up of the time dependence of $d_{001}$, which is an enlargement of the blue areas in Fig. 2b and c. The change in temperature with time is also shown on the right axis. During the temperature increase from RT to 400 $^{\circ }\hbox {C}$, $d_{001}$ reaches a broad maximum at $\sim$3.437 Å at $\sim$180 $^{\circ }$C, and then slightly decreases to $\sim$3.434 Å up to 400 $^{\circ }\hbox {C}$. The broad maximum in $d_{001}$ is related with the melting temperature of the Li metal ($\simeq$ 180 $^{\circ }\hbox {C}$ at ambient pressure). During constant heating at 400 $^{\circ }\hbox {C}$, $d_{001}$ gradually decreases to $\sim$3.415 Å with a changing slope. The decrease in $d_{001}$ suggests a de-intercalation of Li$^{+}$ ions from $\hbox {C}_{6}$Li and/or rearrangement of Li$^{+}$ ions between first-stage (C$_{6}$Li) and higher stages (C$_{12}$Li/C$_{18}$Li), as discussed later. Figure 2e shows the pressure dependence of $d_{001}$, where the arrows indicate the directions for pressurization and de-pressurization. Owing to the decrease in $d_{001}$ during heating at 400 $^{\circ }\hbox {C}$, $d_{001}$ during de-pressurization is smaller than that during pressurization down to $\sim$3 GPa.

Fig. 3

Results of the in situ XRD measurements of $\hbox {C}_{6}$ Li + 2Li. (a) Selected XRD patterns during pressurization and heating: before heating at P = 10 GPa at RT, during heating at P = 10 GPa at $\sim$200 $^{\circ }\hbox {C}$, after heating at P = 10 GPa, and after heating at P = 2 GPa. The XRD patterns include the 001 and 002 diffraction peaks from $\hbox {C}_{6}$Li, the 002 diffraction peak from $\hbox {C}_{6}$, and the 110 diffraction peak and corresponding characteristic radiation K$\alpha$ and K$\beta$ peaks from the Ta capsule. (b) Time dependence of d-spacing from $\hbox {C}_{6}$Li, $\hbox {C}_{12}$Li, $\hbox {C}_{18}$Li/C$_{24}$Li, and $\hbox {C}_{6}$ phases. (c) Time dependence of temperature (left) and pressure (right) during the in situ XRD measurements. (d) The enlarged time dependence of the d-spacing together with the time dependence of temperature (right) are taken from the blue bands in (b) and (c). (e) Contour plot of the d-spacing as a function of time.

Figure 3a shows selected in situ XRD patterns of $\hbox {C}_{6}$ + 2Li before heating at P = 10 GPa, during heating at P = 10 GPa and $\sim$200 $^{\circ }\hbox {C}$, after heating at P = 10 GPa, and after heating at P = 2 GPa. Similar to the $\hbox {C}_{6}+3$Li sample, C$_{6}$Li + 2Li consists of $\hbox {C}_{6}$Li and $\hbox {C}_{6}$ phases at the initial state, as understood by the 002 diffraction peak from C$_{6}$ together with the 001 and 002 diffraction peaks from C$_{6}$Li. The 001 diffraction peak from $\hbox {C}_{6}$Li splits into two peaks during heating at $\sim$200 $^{\circ }\hbox {C}$, and another weak diffraction peak appears at d $\simeq$ 3.2 Å. After heating, these diffraction peaks disappear, and in turn, an intense 002 diffraction peak from $\hbox {C}_{6}$ is observed at d $\simeq$ 3.0 Å.

To clarify the structural changes that occur during heating, Fig. 3b shows the time dependence of the d-spacing of several phases, and Fig. 3c shows the time dependence of the temperature (left) and pressure (right) during the in situ XRD measurements. The $d_{001}$ value before heating at P = 10 GPa is 3.443(1) Å, which is similar to that for $\hbox {C}_{6}+3$Li at P = 10 GPa [= 3.426(1) Å] (Fig. 2b). During heating, the d-spacing indicates a step-like decrease down to $\sim$3.0 Å and then gradually increases to $\sim$3.15 Å at P = 2 GPa.

As observed in the enlarged time dependence shown in Fig. 3d, the d-spacing decreases to $\sim$3.0 Å before reaching 400 $^{\circ }$C, stopping at two intermediate d-spacings of $\sim$3.35 and 3.20 Å. According to our in situ HT-XRD measurements on $\hbox {C}_{6}$Li$_{x}$ (x = 0.77) with the LiPF$_{6}$(EC+DEC) electrolyte under ambient pressure³⁰, the $\hbox {C}_{6}$Li phase changed into lower-stage compounds such as $\hbox {C}_{12}$Li and C$_{18}$Li, accompanied by the release of Li$^{+}$ ions from the graphite layers. Figure S4 compares the temperature dependence of changes in d-spacing between present and previous³⁰ studies. The transformation to $\hbox {C}_{6}$ graphite was accomplished at $\sim$320 $^{\circ }\hbox {C}$³⁰, similar to the present case. Hence, the step-like decrease in the d-spacing can most likely be attributed to structural changes into $\hbox {C}_{12}$Li and $\hbox {C}_{18}$Li due to reactions with the residual LiPF$_{6}$(EC+DEC) electrolyte, although the d-spacings for such phases under HP are currently unknown. Specifically, typical reactions with the LiPF$_{6}$(EC+DEC) electrolyte are represented by^30,31:

$$\begin{aligned} & 10\textrm{C}_{6}\textrm{Li}_{x} + x\textrm{PF}_{5} \rightarrow 5\textrm{C}_{12}\textrm{Li}_{x} + 5x\textrm{LiF} + x\textrm{P}, \end{aligned}$$

(4)

$$\begin{aligned} & 4\textrm{C}_{6}\textrm{Li}_{x} + 2x\mathrm{C_{3}H_{4}O_{3}} \rightarrow 2\textrm{C}_{12}\textrm{Li}_{x} + x\mathrm{(CH_{2}OCO_{2}Li)_{2}} + x\mathrm{C_{2}H_{4}}, \end{aligned}$$

(5)

and

$$\begin{aligned} 5\textrm{C}_{12}\textrm{Li}_{x} + x\textrm{PF}_{5} \rightarrow 10\textrm{C}_{6} + 5x\textrm{LiF} + x\textrm{P}. \end{aligned}$$

(6)

Figure 3e shows a contour plot of the d-spacing as a function of time to illustrate changes in the XRD peak intensities during heating. The intensity of the $\hbox {C}_{6}$ graphite peak becomes strong at $\sim$27 min, which corresponds to $\sim$320 $^{\circ }\hbox {C}$.

We then performed differential scanning calorimetry (DSC) to confirm the existence of the residual LiPF$_{6}$(EC+DEC) electrolyte. Figure 4a and b show the DSC profiles of the $\hbox {C}_{6}+3$Li and C$_{6}$Li + 2Li samples, respectively, which were identical to those subjected to the in situ XRD measurements illustrated in Figs. 2 and 3. No distinct exothermic reaction peak appears in the DSC profile of $\hbox {C}_{6}+3$Li. The endothermic reaction peak at $\sim$180 $^{\circ }\hbox {C}$ is attributed to the melting of the Li metal. In contrast, at least two broad exothermic reaction peaks are observed at $\sim$220 and 300 $^{\circ }\hbox {C}$ in the DSC profile of C$_{6}$Li + 2Li. The endothermic reaction peak related to Li melting is also observed at $\sim$180 $^{\circ }\hbox {C}$. As shown in Fig. 4c, the $\hbox {C}_{6}$Li$_{x}$ (x = 0.88) electrode with the LiPF$_{6}$(EC+DEC) electrolyte indicates at least three exothermic reaction peaks at $\sim$130, 260, and 300 $^{\circ }\hbox {C}$, which are attributed to decomposition of solid electrolyte interphase, formation of LiF, and decomposition of EC and DEC solvents³¹. Being an electrode, it contains a polyvinylidene fluoride (PVdF) binder (5 wt%) in addition to $\hbox {C}_{6}$Li$_{0.88}$. Comparing these DSC profiles reveals that the exothermic reaction peaks for $\hbox {C}_{6}$Li + 2Li were caused by reactions between $\hbox {C}_{6}$Li and residual LiPF$_{6}$(EC+DEC) electrolyte. Note that the amount of residual LiPF$_{6}$(EC+DEC) electrolyte is significantly small because the values of heat flow for $\hbox {C}_{6}$Li + 2Li are $\sim$1/20 compared to those for $\hbox {C}_{6}$Li$_{x}$ electrode.

Fig. 4

Comparison of DSC profiles. (a) C$_{6}+3$Li. (b) $\hbox {C}_{6}$Li + 2Li. The samples of $\hbox {C}_{6}+3$Li and C$_{6}$Li + 2Li are identical to those characterized by XRD (Figs. 2 and 3, respectively). The endothermic peak at $\sim$185 $^{\circ }\hbox {C}$ is caused by the melting of the Li metal. The DSC profile of $\hbox {C}_{6}$Li used for $\hbox {C}_{6}$Li + 2Li(5x) (not containing Li metal) is also shown in (b). (c) DSC profile of the C$_{6}$Li$_{x}$ (x = 0.88) electrode together with those for C$_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li shown in (a) and (b), respectively. The $\hbox {C}_{6}$Li$_{0.88}$ electrode contains the PVdF binder and LiPF$_{6}$(EC+DEC) electrolyte. The data for the C$_{6}$Li$_{0.88}$ electrode were obtained from Ref.³¹.

Fig. 5

Results of the in situ XRD measurements of $\hbox {C}_{6}$Li + 2Li(5x). (a) Selected XRD patterns during pressurization and heating: initial state at P = 1.5 GPa and RT, before heating at P = 10 GPa, after heating at P = 10 GPa, and after heating at P = 1.5 GPa. The XRD patterns include the 001 and 002 diffraction peaks from $\hbox {C}_{6}$Li, the 002 diffraction peak from C$_{6}$, and the 110 diffraction peak and corresponding characteristic radiation K$\alpha$ and K$\beta$ peaks from the Ta capsule. (b) Time dependence of $d_{001}$. c Time dependence of temperature (left) and pressure (right). (d) Enlarged time dependence of $d_{001}$ together with the time dependence of temperature (right). The enlarged time area is taken from the blue bands in in (b) and (c). (e) Pressure dependence of $d_{001}$. The right and left arrows indicate the directions of pressurization and de-pressurization, respectively.

Based on the results of DSC measurements, we prepared a new C$_{6}$Li + 2Li sample by rinsing it with DEC five times instead of twice and drying under vacuum for 24 h [denoted as C$_{6}$Li + 2Li(5x)]. Figure S5 shows the discharge curve of the C$_{6}$/Li cell for preparing $\hbox {C}_{6}$Li + 2Li(5x). The discharge curve and observed discharge capacity for $\hbox {C}_{6}$Li + 2Li(5x) are similar to those for $\hbox {C}_{6}$Li + 2Li. We also confirmed the DSC profile of the C$_{6}$Li used for $\hbox {C}_{6}$Li + 2Li(5x). As shown in Fig. 4b, exothermic reaction peaks at $\sim$220 and 300 $^{\circ }\hbox {C}$ are negligible, confirming that the LiPF$_{6}$(EC+DEC) electrolyte was removed from the sample.

Figure 5a shows selected in situ XRD patterns of C$_{6}$Li + 2Li(5x) in the initial state at P = 1.5 GPa and RT, before heating at P = 10 GPa, after heating at P = 10 GPa, and after heating at P = 1.5 GPa. $\hbox {C}_{6}$Li + 2Li(5x) is also in a mixture of $\hbox {C}_{6}$Li and $\hbox {C}_{6}$, as indicated by the 001 and 002 diffraction peaks from $\hbox {C}_{6}$Li and the 002 diffraction peak from C$_{6}$. The 001 and 002 diffraction peaks are observed even after heating at P = 10 and 1.5 GPa, confirming the absence of the LiPF$_{6}$(EC+DEC) electrolyte in $\hbox {C}_{6}$Li + 2Li(5x). The weak diffraction peaks during pressurizing are probably caused by changes in the preferred orientation of $\hbox {C}_{6}$Li layers or decrease in crystallinity. Figure 5b shows the time dependence of $d_{001}$, and Fig. 5c shows the time dependence of temperature (left) and pressure (right). The $d_{001}$ value decreases from 3.632(1) Å from P = 1.5 GPa to 3.550(1) Å at P = 10 GPa, then slightly increases to $\sim$3.60 Å during heating at 400 $^{\circ }\hbox {C}$, and finally increases to 3.703(1) Å at P = 1.5 GPa. The pressure dependence of $d_{001}$ is not monotonic, and several inflection points appear during pressurization or de-pressurization, probably due to the re-arrangement of $\hbox {C}_{6}$Li and insertion of Li$^{+}$ ions into the $\hbox {C}_{6}$Li lattice.

Figure 5d shows the enlarged time dependence of $d_{001}$ together with the time dependence of temperature (right). When the temperature increases from RT, $d_{001}$ increases from $\sim$3.545 Å to $\sim$3.60 Å at 400 $^{\circ }\hbox {C}$. The slight decrease in $d_{001}$ at $\sim$200 $^{\circ }\hbox {C}$ corresponds to the melting temperature of Li metal (Fig. 4a,b). The $d_{001}$ value remains constant at ($\simeq$ 3.60 Å) during heating at 400 $^{\circ }$C, which differs from the case for $\hbox {C}_{6}+3$Li (Fig. 2d). Furthermore, the $d_{001}$value of $\hbox {C}_{6}$Li + 2Li(5x) is $\sim$0.2 Å larger than that of $\hbox {C}_{6}+3$Li, suggesting that C$_{6}$Li + 2Li(5x) accommodates more Li$^{+}$ ions between the graphite layers. Note that the 110 diffraction peak from the Ta capsule (= $d_{110}$) is located at 2.30 Å at 10 GPa for both C$_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li(5x), confirming the significant difference in $d_{001}$ between $\hbox {C}_{6}+3$Li and $\hbox {C}_{6}$Li + 2Li(5x). As shown in Figure S6, full-width at half maximum of $d_{001}$ is almost constant ($\simeq$ 0.22$^{\circ }$) during the heating at 400 $^{\circ }$C. The $d_{001}$ value rapidly drops to $\sim$3.56 Å from $\sim$3.60 Å after the heating at 400 $^{\circ }\hbox {C}$, which is approximately four times larger than the change in $d_{002}$ for C$_{6}$ after the heating (Figure S2d). This means that the thermal expansion coefficient between the graphite layers is enlarged by the intercalation of Li$^{+}$ ions, enabling further intercalation of Li$^{+}$ ions during the heating up to the $\hbox {C}_{2}$Li composition. As shown in Fig. 5e, the pressure dependence of $d_{001}$ is more moderate than that of $\hbox {C}_{6}+3$Li; specifically, the change in $d_{001}$ for the entire measurement sequence is smaller than $\sim$0.15 Å for $\hbox {C}_{6}$Li + 2Li(5x) but larger than $\sim$0.22 Å for $\hbox {C}_{6}+3$Li.

Comparing changes in $d_{001}$ between $\hbox {C}_{6}+3$Li and C$_{6}$Li + 2Li(5x), $\hbox {C}_{6}$Li + 2Li(5x) is more suitable for synthesizing $\hbox {C}_{2}$Li. Because $I_\textrm{c}$ of $\hbox {C}_{2}$Li is identical to that of $\hbox {C}_{6}$Li (Fig. 1a,b), $d_{001}$ should be maintained or slightly enlarged during further insertion of Li$^{+}$ ions from the $\hbox {C}_{6}$Li composition. However, as shown in Fig. 2d, $d_{001}$ of $\hbox {C}_{6}+3$Li approached that for $\hbox {C}_{12}$Li ($\sim$3.41 Å), suggesting a de-intercalation of Li$^{+}$ ions from $\hbox {C}_{6}$Li. Meanwhile, $\hbox {C}_{6}+3$Li is likely to contain several higher stages such as $\hbox {C}_{12}$Li and $\hbox {C}_{18}$Li in a microscopic scale, because it was prepared by a simple mechanical mixing of C$_{6}$ and Li metal. Thus, the decrease in $d_{001}$ can also be explained by the rearrangement between the stages of $\hbox {C}_{6}$Li and C$_{12}$Li/C$_{18}$Li. In the case of $\hbox {C}_{6}$Li + 2Li(5x), $d_{001}$ slightly increased from $\sim$3.545 Å to $\sim$3.60 Å during the heating at 400 $^{\circ }\hbox {C}$ under HP. Moreover, $d_{001}$ after the heating at P = 1.5 GPa was $\sim$3.70 Å, which is larger than that of initial value ($\sim$3.64 Å). These results are related with the formation of $\hbox {C}_{2}$Li, which owes to the nearly stoichiometric and homogeneous $\hbox {C}_{6}$Li (specifically C$_{6}$Li$_{0.98}$) sample prepared by the electrochemical reaction.

Both crystal structure and composition should be examined to verify the formation of $\hbox {C}_{2}$Li, which has never been confirmed thus far, as understood by Table 1. Due to the limited amount of sample in the Ta capsule ($\sim$1.5 mg), further analyses such as XRD, DSC, and electrochemical measurements were not possible in this sutdy. Among them, we expect that open-circuit voltage measurements during the electrochemical charge process are effective because operating voltages correspond to the reactions between stage transformations of $\hbox {C}_{2}$Li $\rightarrow$ $\hbox {C}_{6}$Li, $\hbox {C}_{6}$Li $\rightarrow$ $\hbox {C}_{12}$Li, etc. The HP apparatus using Walker-type equipment provides a large amount of sample (> 30 mg). Further characterization including the optimization for HP-HT conditions would be required identify synthesized $\hbox {C}_{2}$Li samples.

Summary

We examined structural changes in samples of $\hbox {C}_{2}$Li at the nominal composition under HP and HT, to clarify the optimum conditions for synthesizing $\hbox {C}_{2}$Li. Although Semenenko et al.¹¹ employed a combination of $\hbox {C}_{6}$ and Li metal when they synthesized $\hbox {C}_{2}$Li for the first time, we utilized two types of $\hbox {C}_{6}$Li as a starting material, that is, $\hbox {C}_{6}+3$Li and C$_{6}$Li + 2Li. The former was prepared by simply mixing graphite powder with Li metal, whereas the latter was prepared by the electrochemical discharge reaction that occurs in LIBs. In the case of $\hbox {C}_{6}+3$Li, the value of $d_{001}$, which corresponds to $I_\textrm{c}$ both for $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li, gradually decreases from $\sim$3.434 Å to $\sim$3.414 Å during constant heating at 400 $^{\circ }\hbox {C}$, indicating the de-intercalation of Li$^{+}$ ions from the graphite layers. By contrast, in the case of C$_{6}$Li + 2Li(5x), $d_{001}$ remained constant at ($\simeq$3.60 Å) during heating at 400 $^{\circ }\hbox {C}$. Therefore, based on the changes in $d_{001}$ during this heating process, $\hbox {C}_{6}$Li + 2Li(5x) is more suitable for synthesizing $\hbox {C}_{2}$Li than $\hbox {C}_{6}+3$Li. The LiPF$_{6}$(EC+DEC) electrolyte should therefore be removed before mixing with Li metal, because the residual electrolyte induced the structural transformations to lower-stage compounds such as C$_{12}$Li and $\hbox {C}_{18}$Li.

The $\hbox {C}_{6}$Li + 2Li(5x) sample still contained a small amount of C$_{6}$ phase after the HP/HT synthesis, indicating the difficulties in synthesizing a single-phase of $\hbox {C}_{2}$Li. Difficulties in distinguishing $\hbox {C}_{2}$Li from $\hbox {C}_{6}$Li are also originated from their similarities in crystal structure and physical properties and small atomic scattering factor of Li$^{+}$ ions. Further HP studies and characterization are underway in our laboratory.

Methods

Sample preparation

A sample of $\hbox {C}_{6}$ graphite powder and a Li metal rod were provided by Nippon Graphite Industries and Honjo Metal, respectively. For C$_{6}$ + 3Li, approximately 0.1 g of graphite powder was mixed with Li metal using a mortar and pestle to reach a C/Li molar ratio of 2:1 in an argon-filled glove-box (DBO-2BLKP, Miwa Mfg). We cut the Li metal into fragments with dimensions of $\sim$1 mm $\times$ 1 mm to facilitate homogeneous mixing. The graphite powder turned from black to golden yellow during mixing, indicating that Li$^{+}$ ions intercalated between the graphite layers.

For $\hbox {C}_{6}$Li + 2Li and $\hbox {C}_{6}$Li + 2Li(5x), $\hbox {C}_{6}$Li$_{x}$ was initially prepared by an electrochemical discharge reaction. Approximately 0.2 g of the graphite powder was pressed into a pellet with a diameter of 20 mm and thickness of $\sim$1 mm. The pressed pellet was used as the working electrode, whereas the Li metal sheet pressed onto a stainless steel plate (diameter = 35 mm) was used as the counter electrode. The LiPF$_{6}$(EC+DEC) electrolyte was provided by Kishida Chemical. Two sheets of porous polyethylene membrane (TonenGeneral Sekiyu) were used as the separator. The C$_{6}$/Li cell was fabricated in the argon-filled glove-box and then discharged to 0.02 V at a current of 0.3 mA ($\simeq$ 0.096 mA cm$^{-2}$) and 25 $^{\circ }\hbox {C}$. The x values of $\hbox {C}_{6}$Li + 2Li and C$_{6}$Li + 2Li(5x) were 0.98 and 1, respectively, based on each discharge capacity and $Q_\textrm{theo}$ (= 372 mAh g$^{-1}$). The open-circuit voltages just before disassembling the cell were 69 mV for $\hbox {C}_{6}$Li + 2Li and 58 mV for $\hbox {C}_{6}$Li + 2Li(5x). The discharge curves for both cells are shown in Figure S5. The golden yellow C$_{6}$Li$_{x}$ pellet was removed from the cell in the argon -filled glove-box and then mixed with a piece of Li metal in the C/Li molar ratio of 2:1. Before mixing, the $\hbox {C}_{6}$Li$_{x}$ pellet for $\hbox {C}_{6}$Li + 2Li was rinsed with DEC twice and dried under vacuum for 1 h, whereas the $\hbox {C}_{6}$Li$_{x}$ pellet for $\hbox {C}_{6}$Li + 2Li(5x) was rinsed with DEC five times and dried under vacuum for 24 h.

In situ XRD measurements

In situ XRD patterns were recorded at the BL14B1 beamline at SPring-8, Japan. Approximately 1.5 mg of the $\hbox {C}_{6}$, $\hbox {C}_{6}+3$Li, C$_{6}$Li + 2Li, or $\hbox {C}_{6}$Li + 2Li(5x) sample was installed into a Ta capsule in the argon-filled glove-box. Then, the Ta capsule was packed into a cubic pyrophyllite pressure medium with a length of 6 mm, together with a BN sleeve, a graphite heater, and Mo rod (Figure S1a). Highly reduced materials such as $\hbox {C}_{6}$Li and Li metal are easily reacted with the moist air, producing Li$_{2}$O and Li$_{2}$CO$_{3}$ phases³⁰. The Ta capsule was sealed with a polytetrafluoroethylene tape and stored in an argon-filled vessel just before the in situ XRD measurements. Next, the Ta capsule was set in the center of six anvils (Fig. 1d) and quickly started to pressurize to $\sim$1.5 GPa. The working time for this procedure was within 5 min. As shown in Figs. 2a, 3a, and 5a, the diffraction lines from $\hbox {C}_{6}$Li were clearly observed. Thus, the C$_{6}+3$Li, $\hbox {C}_{6}$Li + 2Li, or $\hbox {C}_{6}$Li + 2Li(5x) sample was thought to maintain its initial reduced state in the Ta capsule. HP was generated using the cubic-type multi-anvil apparatus, where a top die with a piston or anvil was moved toward the lower side of the apparatus by a hydraulic cylinder (Figure S1b)^25,26,27. Each sample was pressurized to 10 GPa over a period of $\sim$2 h, then heated at 25 $^{\circ }\hbox {C}$ min$^{-1}$ up to 400 $^{\circ }\hbox {C}$, where it was held for 30 min (see Fig. 2c), and finally de-pressurized to ambient pressure for $\sim$2 h after quenching to RT from 400 $^{\circ }\hbox {C}$ (see Fig. 2c). Energy-dispersive XRD data were obtained during pressurization, heating, and de-pressurization with an acquisition time of 120 or 15 s. The energy of diffracted X-rays (E) was converted into d using the relation 2$d \sin \theta$ $= 12.4/E$ where 2$\theta$ = 4.5 $^{\circ }$ for all measurements. The peak positions of diffraction peaks were determined by fitting the diffraction patterns with a Gaussian function. The crystal structures of $\hbox {C}_{6}$Li and $\hbox {C}_{2}$Li were drawn using VESTA software³².

DSC measurements

DSC profiles were acquired at a scan rate of 5 $^{\circ }\hbox {C}$ min$^{-1}$ up to 450 $^{\circ }\hbox {C}$ (Thermo plus EVO2, DSC8230L, Rigaku). Approximately 20 mg of $\hbox {C}_{6}+3$Li, $\hbox {C}_{6}$Li + 2Li, or the C$_{6}$Li used for $\hbox {C}_{6}$Li + 2Li(5x) was installed into a stainless steel (SUS) pan with a diameter of 6 mm and a height of 5 mm. The temperature and change in enthalpy ($\Delta H$) were calibrated using the fusion temperatures and specific $\Delta H$s for several metals, as reported previously³¹. In this study, a positive $\Delta H$ value means that the reaction proceeds in exothermic, whereas a negative $\Delta H$ value means that the reaction proceeds in endothermic. The SUS pan was monitored before and after the measurements to ensure that no leakage occurred during the measurement.