Reversible Movement of Zn2+ Ions with Zero-Strain Characteristic

Article pubs.acs.org/IC

Cite This: Inorg. Chem. XXXX, XXX, XXX−XXX

Reversible Movement of Zn2+ Ions with Zero-Strain Characteristic: Clarifying the Reaction Mechanism of Li2ZnTi3O8 Kazuhiko Mukai* Toyota Central Research and Development Laboratories, Inc., 41-1 Yokomichi, Nagakute, Aichi 480-1192, Japan

Downloaded via GUILFORD COLG on July 24, 2019 at 13:31:19 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Lithium zinc titanate spinel, Li2ZnTi3O8, has received significant attention as a negative electrode material for lithium-ion batteries (LIBs). However, its reaction mechanism has not been fully clarified yet, particularly for the large voltage hysteresis between discharge and charge c ur v e s . W e h en c e c lo se l y e xa mi n e d (L i 1 − x Zn x ) [Li1/3+x/3Ti5/3−x/3]O4 (LZTO) with 0 < x ≤ 0.5 by measuring its open-circuit voltage (OCV) and recording synchrotron radiation X-ray diffraction (XRD) patterns. Here, LZTO is a solid solution of Li[Li1/3Ti5/3]O4 (x = 0) and Li2ZnTi3O8 (x = 0.5), both of which have a spinel-framework structure. For the x = 0.5 sample, the OCV of the discharge reaction differed from that of the charge reaction, particularly at a capacity above 50 mAh·g−1. This difference was due to the migration of Zn2+ ions from tetrahedral sites to octahedral sites, and the Zn2+ ions moved back to tetrahedral sites during the charge reaction. Despite these drastic movements of Zn2+ ions, the cubic lattice parameter of the spinel was maintained during the whole reaction, i.e., zero strain. Perfect zero strain, which has never been reported for any LIB materials, was achieved with the x = 0.25 sample. The reaction mechanism with x = 0.5 and the contributions of the amount of Zn ions are discussed in detail.

■

INTRODUCTION The term “reversibility” in lithium-ion batteries (LIBs) has two meanings.1 One is the cycling performance, i.e., how long a charge capacity (Qcha) or discharge capacity (Qdis) maintains an initial value. The other is voltage hysteresis, namely, the difference between voltages in the charge and discharge curves of a certain cycle. The latter reversibility is sometimes overlooked, but the voltage hysteresis should be minimized as much as possible to avoid undesirable side reactions and/or generating heat in LIBs. Lithium zinc titanium oxide Li2ZnTi3O8 has attracted a great deal of attention as a negative electrode material for LIBs because of its large rechargeable capacity (Qrecha) of more than 170 mAh·g−1.2−14 Figure 1a schematically shows the crystal structure of Li2ZnTi3O8, which crystallizes into cubic spinel in which Li+ and Zn2+ ions are randomly distributed in the tetrahedral (tet) sites, while Li+ and Ti4+ ions order in a 1:3 ratio in the octahedral (oct) sites, resulting in a cation distribution of (Li1/2Zn1/2)tet[Li1/2Ti3/2]oct.15−17 The crystal structure of Li2ZnTi3O8 is assigned to the P4332 (or P4132) space group.15−17 Common electrochemical features have been observed in Li2ZnTi3O8:2−14 the voltage vs Li+/Li gradually decreases to ∼0.7 V up to a Qdis value of ∼50 mAh·g−1, then maintains a constant value at ∼0.6 V until ∼150 mAh·g−1, and finally decreases gradually again to ∼0.2 V. The subsequent charge (oxidation) curve is quite different from the previous discharge curve; i.e., a plateau appears at ∼1.5 V in the Qcha range between 50 and ∼150 mAh·g−1. The difference between the voltages of © XXXX American Chemical Society

the discharge and charge curves appears in further extended cycles without any significant capacity fading.2−14 The origins of the voltage hysteresis are not yet fully understood, although many studies have been devoted to improving the electrochemical properties of Li2ZnTi3O8, such as research on new synthetic routes,2−6 surface modifications,7,8 and doping/ substitution with other elements.9−13 To rationalize the voltage hysteresis in Li2ZnTi3O8, we measured the open-circuit voltage (OCV) and recorded ex situ X-ray diffraction (XRD) patterns in both the discharge and charge directions. The OCVs could reveal whether the voltage hysteresis arises from thermodynamic or kinetic factors, while the ex situ XRD results could reveal changes in the crystal structure of Li2ZnTi3O8, such as the cubic lattice parameter (ac) and cation distribution. This information is essential not only for elucidating the reaction mechanism of Li2ZnTi3O8 but also for obtaining insights into how to suppress the voltage hysteresis. We also examined the electrochemical and structural properties of (Li1−xZnx)[Li1/3+x/3Ti5/3−x/3]O4 (LZTO) with 0 < x < 0.5, where LZTO is a solid solution of Li2ZnTi3O8 and Li[Li1/3Ti5/3]O4. As shown in Figure 1b, Li[Li1/3Ti5/3]O4 adopts a spinel structure with the Fd3̅m space group, in which Li+ ions occupy both tetrahedral and octahedral sites, while Ti4+ ions occupy only the octahedral sites.18−20 Li[Li1/3Ti5/3]O4, i.e., the x = 0 compound, maintains its ac during the discharge and charge reactions, providing a zero-strain characteristic.18 Moreover, the Received: May 27, 2019

A

DOI: 10.1021/acs.inorgchem.9b01565 Inorg. Chem. XXXX, XXX, XXX−XXX

Article

Inorganic Chemistry

Figure 1. Schematic crystal structures of (a) Li2ZnTi3O8 and (b) Li[Li1/3Ti5/3]O4. Both materials have a spinel structure and a cation distribution of (Li1/2Zn1/2)tet[Li1/2Ti3/2]oct and (Li)tet[Li1/3Ti5/3]oct, respectively. The subscripts tet and oct represent tetrahedral and octahedral sites, respectively. Note that Li+ and Ti4+ ions in Li2ZnTi3O8 order in a ratio of 1:3 in the octahedral sites, resulting in a decline in the crystal symmetry from Fd3̅m to P4332 (or P4132). A solid solution of (Li1−xZnx)tet[Li1/3+x/3Ti5/3−x/3]octO4 can form in the x range between 0 and 0.5, where x is the amount of Zn2+ ions at the tetrahedral site. The amounts of Li+ and Ti4+ ions at the octahedral site also alter with x.

x = 0 compound shows a quite flat operating voltage at ∼1.55 V without voltage hysteresis.18−20 Therefore, the x dependence of the electrochemical properties and changes in the crystal structure could reveal the roles of Zn ions in the voltage hysteresis of LZTO. Consequently, the reversible movement of Zn2+ ions appeared in all of the LZTO samples investigated, although these samples demonstrated a zero-strain characteristic as for x = 0.

■

whether the OCVs of the discharge and charge reactions coincide. The voltage was monitored for 1 month after the cell was discharged up to a capacity (Q) of 125 mAh·g−1. Ex situ XRD patterns were recorded at a synchrotron radiation facility at the Aichi Synchrotron Radiation Center. Discharged and/or charged samples were prepared by electrochemical reactions at a current of 0.3 mA. When the current was switched off for at least 5 h, each electrode was removed from the lithium cell in the argon-filled glovebox. Then, the samples including AB and PVdF were packed into borosilicate glass capillaries with a diameter of 0.3 mm (W. Müller Glas Technik). The XRD data were recorded at the BL5S2 beamline using a two-dimensional (2D) detector, PILATUS 100 K (Dectris Ltd.). The wavelength of the X-rays was determined to be 0.799436(2) Å using standard silicon powder (NIST 640d). Rietveld analyses were performed using RIETAN-FP software,21 and crystal structures were drawn with VESTA software.22 Although RIETAN-FP software provided five significant figures for structural parameters, we rounded off the number to one to four decimal places considering observational error.

EXPERIMENTAL SECTION

Powder samples of LZTO with x = 0, 0.125, 0.25, 0.375, and 0.5 were synthesized using a conventional solid-state reaction technique. Reagent-grade LiOH·H2O, TiO2, and ZnO (Wako Pure Chemical Industries, Ltd.) powders were well mixed with a stoichiometric composition and then pressed into a pellet with a diameter of 23 mm and a thickness of ∼5 mm. Each pellet was heated at 750 °C for 12 h in air, followed by preheating at 400 °C for 12 h in air. The obtained pellet was crushed into a powder and then characterized via scanning electron microscopy (SEM; S-5500, Hitachi High-Technologies Co., Ltd.) and XRD (D8 ADVANCE, Bruker AXS, Inc.) measurements equipped with Fe Kα radiation. An osmium plasma coater (HPC-1S, Vacuum Device Inc.) was employed to prepare the SEM specimens. Electrochemical measurements were performed in a nonaqueous lithium cell using 1 M LiPF6 dissolved in ethylene carbonate (EC)/ diethyl carbonate (DEC) electrolyte (EC/DEC = 1/1 by volume, Kishida Chemical Co., Ltd.). A black slurry consisting of 88 wt % LZTO powder, 6 wt % acetylene black (AB, Denka Company Ltd.), and 6 wt % polyvinylidene fluoride (PVdF, Kureha Corp.) in N-methyl-2pyrrolidone (NMP, Nacalai Tesque, Inc.) was cast onto a copper foil. Then, NMP was evaporated under a vacuum at 120 °C for 12 h. A lithium metal sheet pressed onto a stainless steel plate (diameter = 19 mm) was used as the counter electrode, while two sheets of a polypropylene membrane with a thickness of 25 μm (TonenGeneral Sekiyu K. K.) were used as the separator. The lithium cells were fabricated in an argon-filled glovebox. Electrochemical measurements were conducted at 25 °C. Discharge and charge profiles were recorded at a current of 0.3 mA (≃0.2 mA· cm−2) in two voltage ranges: 1.0−3.0 and 0.2−3.0 V. OCVs were measured with an intermittent current of 0.3 mA for 20 min and a relaxation time of 5 h. For only the x = 0.5 sample, the OCV was additionally measured with a long period of relaxation time to clarify

■

RESULTS Particle Morphology. Figure 2 shows SEM images of the (a) x = 0, (b) x = 0.25, and (c) x = 0.5 samples at the 5 μm scale, and the corresponding enlarged SEM images at the 1 μm scale are shown in parts d, e, and f. The average size of primary particles ranges from 200 to 300 nm, regardless of x, but the particle morphologies seem to be different from each other. Particles of the x = 0 sample consist of several flat plates, while some particles of the x = 0.25 sample exhibit an octahedral shape with a side length of ∼300 nm. This octahedral shape is an indicator of well-developed crystal facets and has also been observed in other spinel oxides such as Li[Ni1/2Mn3/2]O423 and Li[CoxMn2−x]O4.24 Particles of the x = 0.5 sample show a nonuniform shape, which is a characteristic of Li2ZnTi3O8 compounds synthesized using a conventional solid-state reaction technique.5,10,12,13 As shown in Figure S1, the x = 0.125 and x = 0.25 samples exhibited similar particle shapes, while particles of the x = 0.375 sample resembled those of the x = 0.5 sample. B


Article

Inorganic Chemistry

311 diffraction line, indicating an increase in the occupancy (g) of Zn2+ ions in the tetrahedral sites (gtet). Furthermore, diffraction lines indicated by a ★ are clearly observed at diffraction angles below ∼50°. These diffraction lines were attributed to superlattice diffraction lines due to the 1:3 cation ordering between Li+ and Ti4+ ions in the octahedral sites. Hence, the XRD pattern of the x = 0.5 sample was assigned to a single phase of the P4332 space group, in which Li+ and Ti4+ ions occupy octahedral 4b and 12d sites, respectively.15−17 The primitive cell of the P4332 space group is still cubic, resulting in the same ac with the Fd3̅m space group.25 The 1:3 cation order in the octahedral sites is also reported for Li[Ni1/2Mn3/2]O423 and Fe[Li1/2Fe3/2]O4.26 Note that the x = 0.25 and 0.375 samples also exhibit superlattice diffraction lines, even though the ratio of Li/Ti is not exactly 1:3. This is probably due to a partial cation ordering between Li+ and Ti4+ ions in the octahedral sites. The mixed nature of a solid solution and superlattice has also been observed in a series of Li1/2+x/2Fe5/2−3x/2TixO4 spinels.26,27 We then performed Rietveld analyses using XRD data recorded with synchrotron radiation. Figure 3f shows the Rietveld analysis results of the x = 0.5 sample, and Table 1 lists its structural parameters. Owing to the 1:3 cation ordering, Li+ and Ti4+ ions occupy octahedral 4b and 12d sites, respectively. The ac value was determined to be 8.3688(1) Å, which is consistent with the reported values.15−17 The Rietveld analysis results for the x = 0.125, 0.25, and 0.375 samples are summarized in Figure S2 and Table S2. To balance the cation ordering and solid solution, we placed Li+ and Ti4+ ions at both 4a and 12d sites with Li+/Ti4+ ratios of 0.21:0.79 for x = 0.25 and 0.23:0.77 for x = 0.375 (Table S2). Figure 3g shows the change in ac as a function of x, as determined by the Rietveld analysis. The x dependence of ac is not monotonic but indicates a convex upward behavior with x. That is, the ac values of the x = 0.125, 0.25, and 0.375 samples were slightly greater than the estimated ac using the linear relationship between x = 0 [=8.3568(1) Å] and x = 0.5. Note

Figure 2. SEM images of the (a) x = 0, (b) x = 0.25, and (c) x = 0.5 samples at the 5 μm scale. Corresponding enlarged SEM images at the 1 μm scale are shown in parts d, e, and f.

Crystal Structure. Figure 3 shows the XRD patterns of the (a) x = 0, (b) x = 0.125, (c) x = 0.25, (d) x = 0.375, and (e) x = 0.5 powders recorded with conventional Fe Kα radiation. The x = 0 and 0.125 samples were assigned to a single-phase normal spinel structure with the Fd3̅m space group, in which the tetrahedral 8a and octahedral 16d sites were mainly occupied by Li+ and Ti4+ ions, respectively. As x increases from 0.125 to 0.5, the main diffraction line alters from the 111 diffraction line to the

Figure 3. XRD patterns of the (a) x = 0, (b) x = 0.125, (c) x = 0.25, (d) x = 0.375, and (e) x = 0.5 samples using the Fe Kα radiation. (f) Rietveld analysis results of the x = 0.5 sample using the synchrotron radiation and (g) ac as a function of x in Li4/3−2x/3ZnxTi5/3−x/3O4. The ★ symbol indicates the superlattice diffraction line due to the cation ordering between Li+ and Ti4+ ions. C


Article

Inorganic Chemistry Table 1. Structural Parameters of the x = 0.5 Sample Determined by Rietveld Analysis atom

Wyckoff position

x

y

z

Biso (Å2)

−0.002(1) −0.002(1) 0.625 0.125 0.390(1) 0.124(1)

−0.002(1) −0.002(1) 0.625 0.368(1) 0.390(1) 0.144(1)

−0.002(1) −0.002(1) 0.625 0.882(1) 0.390(1) 0.859(1)

0.4(1) 0.4(1) 0.2(1) 0.5(1) 0.2(1) 0.2(1)

occupancy (g)

Li1 8c 0.5 Zn 8c 0.5 Li2 4b 1.0 Ti 12d 1.0 O1 8c 1.0 O2 24e 1.0 Space group: P4332, ac = 8.3688(1) Å, Rwp = 3.08%, and S = 0.78.

Figure 4. Discharge and charge curves of the lithium cells with (a) x = 0.125, (b) x = 0.25, (c) x = 0.375, and (d) x = 0.5, operated at a current of 0.3 mA in the ranges of 1.0−3.0 V (black) and 0.2−3.0 V (red). dQ/dV curves for the (e) x = 0.125 and (f) x = 0.5 samples. (g) Cycling performance of the x = 0.125, 0.25, 0.375, and 0.5 samples.

that the ionic radius of Zn2+ ions [rZn2+ = 0.60 Å (C.N. = 4)] is close to that of Li+ ions [rLi+ = 0.59 Å (C.N. = 4)], while rLi+ = 0.76 Å (C.N. = 6) and rTi4+ = 0.61 Å (C.N. = 6), where C.N. is the coordination number.28 The increase in ac with x is thought to be a consequence of the expansion of the Li/TiO6 octahedra. A similar ac value (=8.37 Å) was obtained in (Li1/2Mg1/2)[Li1/2Ti3/2]O4.15 Discharge and Charge Profiles. Figure 4 shows the discharge and charge curves of lithium cells with (a) x = 0.125, (b) x = 0.25, (c) x = 0.375, and (d) x = 0.5, operated at a current of 0.3 mA. The black lines indicate discharge and charge curves in the voltage range between 1.0 and 3.0 V, while the red lines indicate those in the voltage range between 0.2 and 3.0 V. The x = 0.125 sample exhibits a flat operating voltage at ∼1.5 V without any voltage hysteresis, except at the discharge cutoff voltage of 1.0 V. The stable Qdis (or Qcha) value, namely, the Qrecha, reaches ∼130 mAh·g−1. When the discharge cutoff voltage is decreased to 0.2 V, voltage hysteresis clearly arises between the discharge and charge curves (≃0.7 V), although Qrecha increases to ∼180 mAh·g−1. Assuming a one-electron transfer reaction, the electrochemical reaction of x = 0.125 can be represented as

amount based on eq 1. A possible origin for the extra Qrecha value is discussed in a subsequent section. The voltage hysteresis region expands with increasing x. For instance, in the x = 0.5 sample, the discharge voltage remains at ∼0.7 V in the Qdis range between ∼60 and 150 mAh·g−1, whereas the charge voltage remains at ∼1.4 V in the corresponding Qcha range. The electrochemical characteristics of the x = 0.5 sample are similar to those of previous Li2ZnTi3O8.2−14 Note that the Qrecha value significantly decreases with x when the discharge cutoff voltage is 1.0 V; however, it maintains a constant value (≃180 mAh·g−1) when the discharge cutoff voltage is 0.2 V. This indicates that the substitution with Zn2+ ions strongly influences the electrochemical reaction below 1.0 V. To clarify the changes in the voltage hysteresis with x, parts e and f of Figure 4 show the dQ/dV curves of the x = 0.125 and x = 0.5, respectively. For the x = 0.125 sample, two peaks at 1.46 and 0.82 V appear in the cathodic (reduction) direction, while two peaks at 1.59 and 1.47 V appear in the anodic (oxidation) direction. This indicates a voltage hysteresis of 0.65 V in the x = 0.125 sample. For the x = 0.5 sample, the voltage hysteresis is estimated to be 0.63 V, although the cathodic and anodic peaks shift to lower voltages than those of the x = 0.125 sample. Hence, the magnitude of the voltage hysteresis is almost independent of x. We then examined the effects of the voltage hysteresis on cycling performance, and the results are shown in Figure 4g. The Qcha value of x = 0.125 slightly decreases from 179.5 to 159.1 mAh·g−1 over 30 cycles, providing a capacity retention of 88.6%.

Li5/4Zn1/8Ti13/8O4 + Li+ + e− ↔ Li 9/4Zn1/8Ti13/8O4 (1)

where the theoretical capacity (Qtheo) is calculated to be 168.95 mAh·g−1. The Qrecha value of ∼180 mAh·g−1 indicates that more Li+ ions and electrons are stored in x = 0.125 than the expected D


Article

Inorganic Chemistry

Figure 5. OCV measurements of the (a) x = 0.125, (b) x = 0.375, and (c) x = 0.5 samples as a function of capacity. The OCV values are indicated by red circles. Voltage hysteresis is the difference between voltages (or OCVs) in the discharge and charge curves, while polarization (ΔV) is the difference between the OCVs and voltages just after terminating the current. ΔV for x = 0.5 in the (d) discharge and (e) charge reactions. (f) OCV measurements after a long relaxation time (1 month) for x = 0.5. This discharge curve was also indicated by the blue solid line in part c.

However, the Qcha values of the other samples maintain almost initial Qcha values; for instance, the Qcha values of x = 0.5 are 162.1 mAh·g−1 for the 1st cycle and 156.9 mAh·g−1 for the 30th cycle, indicating a capacity retention of 96.7%. Therefore, the voltage hysteresis in LZTO does not degrade, and its cycling performance is stabilized in the lower voltage range. OCV Measurements. Figure 5 shows the OCV measurement results of the (a) x = 0.125, (b) x = 0.375, and (c) x = 0.5 samples, and Figure S3 shows those of the x = 0.125 sample. Red circles indicate the OCV values, which were the voltages 5 h after switching off the current. For the x = 0.125 sample, the OCV values of the discharge reaction overlap those of the charge reaction in the Q range between ∼30 and ∼110 mAh·g−1. The OCV values in this region exhibit a quite flat voltage at 1.55 V, which is similar to that of x = 0.18−20 The OCV gradually decreases as the discharge reaction proceeds and finally reaches ∼0.3 V at Q = 200 mAh·g−1. In contrast, the OCV values of the charge reaction are larger than 1.0 V, except at the beginning of the charge reaction. Therefore, differences between the OCV of the discharge and charge reactions are clearly observed at Q ≥ 130 mAh·g−1. As x increases from x = 0.125, the differences between the OCV of the discharge and charge reactions appear in the lower Q range. For the x = 0.25 sample, these differences appear at Q ≥ 90 mAh·g−1 (Figure S3), whereas, for the x = 0.375 sample, the differences were observed at Q ≥ 70 mA·g−1. For the x = 0.5 sample, the OCV values of the discharge reaction differed from those of the charge reaction over the entire Q range. Considering the polarizations (ΔV), namely, differences in the voltages between OCV and voltage just after terminating the current, ΔV as a function of Q is divided into three distinct regions. Figure 5 shows ΔV as a function of Q in the (d) discharge and (e) charge reactions. The ΔV value of the discharge reaction is less than 0.05 V at Q < 50 mAh·g−1, but it rapidly increases to 0.42 V at Q = 65 mAh·g−1. Then, the ΔV value ranges from 0.24 to 0.29 V up to Q = 146 mAh·g−1 and finally decreases to 0.1 V at Q ≥ 150 mAh·g−1. Three distinct

regions labeled (I), (II), and (III) also appear in the charge reaction. As Q decreases from ∼170 mAh·g−1, ΔV reaches a peak at Q ≥ 150 mAh·g−1, then maintains a constant value (∼0.1 V) at 90 mAh·g−1 ≤ Q < 150 mAh·g−1, and finally decreases to less than 0.03 V at 40 mAh·g−1 ≤ Q ≤ 80 mAh·g−1. The increase in ΔV at Q < 40 mAh·g−1 is typical at the end of the charge or discharge reaction, so the slight decrease in ΔV at the end of the discharge reaction is surprising. We fixed the relaxation time to 5 h for all of the OCV measurements, because such relaxation time was enough for Li1/2+x/2Fe5/2−3x/2TixO4 with x = 1.4 (Li1.2Fe0.4Ti1.4O4)27 and graphite.29 Note that the former compound exhibited almost the same OCV values during the discharge and charge reactions at Q ≥ ∼120 mAh·g−1, although the voltage hysteresis of ∼0.7 V was observed in the same Q range.27 However, OCVs were measured after a long relaxation time for the x = 0.5 sample, and the results are shown in Figure 5f. After discharging the cell up to Q = 125 mAh·g−1, the cell voltage gradually increases from 0.57 to 0.90 V during the initial 5 h. Subsequently, the voltage continues to increase and does not reach a constant value even after 1 month. The corresponding OCV value of the charge reaction was estimated to be 1.27 V. A relaxation time of more than 105 h, i.e., more than 10 years, would be required when the OCV value of the discharge reaction is equal to 1.27 V. On the basis of the OCV measurement results, the voltage hysteresis in LZTO is likely to originate from a kinetic factor, not a thermodynamic factor. Ex Situ XRD Measurements. Figure 6a shows the discharge and charge curves of the x = 0.5 sample for ex situ XRD measurements. The 10 different discharge and charge curves overlap, indicating the good reproducibility of the electrochemical reaction. The Q values of the discharge points at D1, D2, D3, D4, and D5 are 50, 100, 125, 150, and 167 mAh·g−1, respectively, while those of the charge points of C1, C2, C3, C4, and C5 are 116, 70, 44, 21, and 7.8 mAh·g−1, respectively. Figure 6b shows the XRD patterns during the discharge reaction at points D1−D5. As the discharge reaction proceeds, E


Article

Inorganic Chemistry

Figure 6. (a) Discharge and charge curves of the x = 0.5 sample for ex situ XRD measurements. (b) XRD patterns during the discharge reaction at points D1, D2, D3, D4, and D5 in part a. (c) XRD patterns during the charge reaction at points C1, C2, C3, C4, and C5 in part a. All of the samples contained AB and PVdF.

value is similar to 4 × d400, where d400 is the d value (≃2.09 Å) of the 400 diffraction line. Q dependence of the 400 diffraction line and 4 × d400 are shown in Figure S4. Figure 6c shows the XRD patterns during the charge reaction at points C1−C5 to clarify whether these changes are reversible. The XRD patterns during the charge reaction reveal the opposite trend to that of the discharge reaction. For instance, I(220) and I(311) increased during the charge reaction. Nevertheless, the diffraction angle of the 400 diffraction line also remained constant at 2θ ≃ 22° during the charge reaction (see inset). To construct a structural model for Rietveld analysis, we depict the Q dependence of I(220), I(311), and I(400) in Figure S5. The I(220) value was almost constant (∼28%) up to Q of 100 mAh·g−1 and then dropped to 6.8% at 125 mAh·g−1. The I(311) and I(220) values had a similar Q dependence, whereas the I(400) value exhibits the opposite trend. Specifically, it rapidly increased to 100% at around 100 mAh·g−1. These changes are unambiguously due to the reversible movement of

four striking features appear: (i) the intensities of the 220 and 311 diffraction lines decrease [i.e., I(220) and I(311), indicated by red and blue lines, respectively], (ii) the intensity of the 400 diffraction line [I(400) indicated by green lines] increases, (iii) the 400 diffraction line maintains its diffraction angle at 2θ ≃ 22° (see inset), and (iv) the cation ordering between Li+ and Ti4+ ions is preserved during the whole discharge reaction. Note that I(220) only reflects the diffraction line from the tetrahedral site. Since the atomic scattering factor of Li+ ions is very weak compared to that of Zn2+ ions, the decreases in I(220) and I(311) correspond to the decrease in gtet of Zn2+ ions, while the increase in I(400) corresponds to the increase in g of Zn2+ ions in the octahedral sites (goct). Similar changes in I(220), I(311), and I(400) are also reported in a series of spinel oxides (Fe) tet[Li1/2Fe3/2 ]octO4−(Li1/2Fe1/2 )tet[Li1/2 Fe1/2Ti]octO 4− (Li)tet[Li1/3Ti5/3]octO4.26 The constant angle of the 400 diffraction line means that no strain occurs in the lattice during the discharge reaction, i.e., zero strain. This is because the ac F


Article

Inorganic Chemistry

Figure 7. Rietveld analysis results of discharge/charge points (a) D1, (b) D5, (c) C1, and (d) C5. Variations in (e) gtet/(gtet + goct), (f) ac, and (g) Δac as a function of capacity. The right axes of parts e, f, and g indicate goct/(gtet + goct), Vc, and ΔVc, respectively. Circles and triangles indicate values for the discharge and charge reactions, respectively.

increasing Q. In contrast, the goct/(gtet + goct) value levels off ∼0% up to Q of ∼50 mAh·g−1 and then monotonically increases with Q. Note that differences rarely appear between gtet/(gtet + goct) and goct/(gtet + goct) of the discharge and charge reactions. This similarity indicates that the structural changes for the x = 0.5 sample are reversible, although a large voltage hysteresis appeared throughout the electrochemical reaction. Parts f and g of Figure 7 show the ac and change in ac (Δac) as a function of Q, respectively. Furthermore, the cubic lattice volume (Vc) and change in Vc (ΔVc) are shown by the right axes of parts f and g of Figure 7, respectively. The ac value increases from 8.3688(1) Å at Q = 0 mAh·g−1 to 8.3814(1) Å at Q = 100 mAh·g−1 and then remains almost constant at ∼8.382 Å with further increasing Q. No difference is observed between ac for the discharge and charge reactions. Vc, Δac, and ΔVc show Q dependences similar to that of ac. The Δac and ΔVc values are +0.15 and +0.46%, respectively, in the fully discharged state, which are slightly greater than those for x = 0 (Δac = −0.07% and ΔVc = −0.19%)18,30,31 but comparable with those for Li1.2Fe0.4Ti1.4O427 and LiCrTiO4.32 To the best of our knowledge, the zero-strain character of the x = 0.5 sample has been demonstrated for the first time, although a previous review paper on Li2ZnTi3O814 reported such zero-strain character without providing any evidence. Nonetheless, the ac value remained remarkably constant during the whole reaction, especially as the Zn2+ ions drastically move between the tetrahedral and octahedral sites. Ex situ XRD measurements were also performed for the x = 0.125, 0.25, and 0.375 samples in the fully discharged state (0.2 V), and their XRD patterns are shown in Figure 8a−c. Compared to the initial XRD patterns shown in Figure 3b−d and Figure S2, the 400 diffraction line became the most intense line with x = 0.5. However, the intensities of the 111 diffraction line were relatively strong, suggesting that some Zn2+ ions still occupy tetrahedral sites, even in the fully discharged state. We

Li+ and Zn2+ ions between the tetrahedral 8c sites and initially vacant octahedral sites. If the gtet value linearly decreases with Q, the Q dependence of I(220) and I(311) should be monotonic, as predicted by the simulation with the RIETAN-FP software21 (see the solid line in Figure S5). Hence, the rapid decrease in I(220) and I(311) at around 100 mAh·g−1 supports that the Zn2+ ions in the tetrahedral sites start to move to the octahedral sites at this composition. On the basis of the experimental results described above, we employed the following four conditions for our Rietveld analysis. As the discharge reaction proceeds, (i) the Li+ ions are inserted into the initially vacant octahedral 4b and 12d sites up to point D1 (50 mAh·g−1); (ii) then, Li+ and Zn2+ ions in the tetrahedral 8c sites move to the octahedral 4b and 12d sites, together with the insertion of Li+ ions from the outer area of the lattice; (iii) the moved Li+ and Zn2+ ions in the octahedral sites do not influence the octahedral sites already occupied with Li+ (4a) and Ti4+ (12d); and (iv) the Li+/Zn2+ ratio satisfies the total Li+/ Zn2+ ratio, including the amount of the inserted Li+ ions. Figure 7 shows the Rietveld analysis results of the discharge/ charge points (a) D1, (b) D5, (c) C1, and (d) C5, and Table 2 lists their structural parameters. The Rietveld analysis results of other discharge/charge points are shown in Figures S6 and S7, and their structural parameters are summarized in Tables S2 and S3. As evident from the reliability indices of Rwp and S, the above structural model well reproduced the results of the XRD analysis. Figure 7e shows the Q dependence of gtet/(gtet + goct) for both discharge (represented by circles) and charge (represented by triangles) reactions. Here, gtet is the g of Zn2+ in the 8c sites, goct is the sum of the g values of Zn2+ ions in the 4b and 12d sites, and gtet + goct should be equal to the initial amount of Zn2+ ions (=x). The Q dependence of goct/(gtet + goct) is shown in the right axis of Figure 7e. The gtet/(gtet + goct) value remains at ∼100% up to a Q of ∼50 mAh·g−1 and then decreases almost linearly with further G


Article

Inorganic Chemistry Table 2. Structural Parameters of the x = 0.5 Sample at the Discharge/Charge Points of D1, D5, C1, and C5 point

atom

Wyckoff position

x

y

z

Biso (Å2)

−0.001(1) −0.001(1) 0.625 0.125 0.125 0.125 0.387(1) 0.121(1)

−0.001(1) −0.001(1) 0.625 0.370(1) 0.125 0.790(1) 0.387(1) 0.144(1)

−0.001(1) −0.001(1) 0.625 0.880(1) 0.125 0.460(1) 0.387(1) 0.859(1)

0.9(1) 0.9(1) 0.5(1) 0.5(1) 0.5(1) 0.5(1) 0.2(1) 0.2(1)

0.105(1) 0.105(1) 0.625 0.125 0.125 0.125 0.125 0.125 0.384(1) 0.125(1)

0.105(1) 0.105(1) 0.625 0.375(1) 0.125 0.125 0.633(1) 0.633(1) 0.384(1) 0.139(1)

0.105(1) 0.105(1) 0.625 0.875(1) 0.125 0.125 0.617(1) 0.617(1) 0.384(1) 0.865(1)

1.4(1) 1.4(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.6(1) 0.6(1)

0.005(1) 0.005(1) 0.625 0.125 0.125 0.125 0.125 0.125 0.391(1) 0.127(1)

0.005(1) 0.005(1) 0.625 0.375(1) 0.125 0.125 0.640(1) 0.640(1) 0.391(1) 0.138(1)

0.005(1) 0.005(1) 0.625 0.875(1) 0.125 0.125 0.611(1) 0.611(1) 0.391(1) 0.871(1)

1.6(1) 1.6(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 1.3(1) 1.3(1)

−0.001(1) −0.001(1) 0.625 0.125 0.125 0.125 0.125 0.125 0.388(1) 0.125(1)

−0.001(1) −0.001(1) 0.625 0.368(1) 0.125 0.125 0.645(1) 0.645(1) 0.388(1) 0.145(1)

−0.001(1) −0.001(1) 0.625 0.882(1) 0.125 0.125 0.601(1) 0.601(1) 0.388(1) 0.859(1)

0.6(1) 0.6(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.8(1) 0.5(1) 0.5(1)

occupancy (g)

D1

Li1 8c 0.5 Zn1 8c 0.5 Li2 4b 1.0 Ti 12d 1.0 Li3 4a 0.120(1) Li4 12d 0.120(1) O1 8c 1.0 O2 24e 1.0 Space group: P4332, ac = 8.3767(1) Å, Rwp = 2.06%, and S = 0.36. D5 Li1 8c 0.074(1) Zn1 8c 0.074(1) Li2 4b 1.0 Ti 12d 1.0 Li3 4a 0.713(1) Zn2 4a 0.213(1) Li4 12d 0.713(1) Zn3 12d 0.213(1) O1 8c 1.0 O2 24e 1.0 Space group: P4332, ac = 8.3816(1) Å, Rwp = 2.24%, and S = 0.38. C1 Li1 8c 0.232(1) Zn1 8c 0.232(1) Li2 4b 1.0 Ti 12d 1.0 Li3 4a 0.514(1) Zn2 4a 0.134(1) Li4 12d 0.514(1) Zn3 12d 0.134(1) O1 8c 1.0 O2 24e 1.0 Space group: P4332, ac = 8.3791(1) Å, Rwp = 2.81%, and S = 0.52. C5 Li1 8c 0.484(1) Zn1 8c 0.484(1) Li2 4b 1.0 Ti 12d 1.0 Li3 4a 0.038(1) Zn2 4a 0.008(1) Li4 12d 0.038(1) Zn3 12d 0.008(1) O1 8c 1.0 O2 24e 1.0 Space group: P4332, ac = 8.3695(1) Å, Rwp = 2.43%, and S = 0.54.

discharged state, providing Δac = −0.13(1)% and ΔVc = −0.39(1)%. The negative Δac and ΔVc values are similar to those in the case of x = 0 (see the Δac value in ref 30), in which the oxygen positional parameter affected the shrinkage of the crystal lattice.19,30,31 As x increases from x = 0.125, the ac value at the fully discharged state tends to be larger than the initial ac value. Among the LZTO samples investigated, the x = 0.25 sample exhibited the same ac value (≃8.3645 Å), yielding Δac = 0% and ΔVc = 0%. Such “perfect” zero strain has never been observed among previously reported LIB materials.

conducted a Rietveld analysis using the Fd3̅m space group for x = 0.125 and the P4332 space group for x = 0.25 and 0.375. Indeed, as shown in Figure 8d, the gtet/(gtet + goct) value of x = 0.125 was determined to be 33.6(3)%, which is approximately twice that of x = 0.5 [14.8(3)%]. Since the gtet/(gtet + goct) value monotonically decreases with x, Zn2+ ions easily move to the octahedral sites when the initial amount of Zn2+ ions increases. In addition, the values of goct/(gtet + goct) and gtet/(gtet + goct) show opposite trends (see the right axis of Figure 8d). Table S4 lists the structural parameters of the fully discharged x = 0.125, 0.25, and 0.375 samples. Parts e and f of Figure 8 show ac and Δac in the fully discharged state as a function of x in Li4/3−2x/3ZnxTi5/3−x/3O4, respectively. The initial ac values are also shown by triangles in Figure 8e for comparison. Moreover, the right axes of parts e and f of Figure 8 show their Vc and ΔVc values, respectively. The ac value of x = 0.125 decreases to 8.3514(1) Å in the fully

■

DISCUSSION Reaction Mechanism. The reaction scheme of x = 0.5 was divided into three distinct regions considering the results of OCV and ex situ XRD measurements. In region (I) up to a Qdis of 50 mAh·g−1, Li+ ions were inserted into the initial vacant H


Article

Inorganic Chemistry

Figure 8. Rietveld analysis results of the (a) x = 0.125, (b) x = 0.25, and (c) x = 0.375 samples in the fully discharged state. Variations in (d) gtet/(gtet + goct), (e) ac, and (f) Δac as a function of x in Li4/3−2x/3ZnxTi5/3−x/3O4. The right axes of parts d, e, and f indicate goct/(gtet + goct), Vc, and ΔVc, respectively. These were the values at the fully discharged state. The triangles in part e indicate the initial ac values of the LZTO samples for comparison. The Δac value at x = 0 was obtained from data in ref 30.

Figure 9. (a) Comparison between observed Qdis and Qtheo as a function of x in Li4/3−2x/3ZnxTi5/3−x/3O4. Qdis(1.0−3.0 V) and Qdis(0.2−3.0 V) are the Q values measured in the ranges of 1.0−3.0 and 0.2−3.0 V, respectively. Qtheo is the Q based on the one-electron transfer reaction, and the value of (1 − x) corresponds to the initial amount of Li+ ions in the tetrahedral 8b sites. (b) Schematic of a possible conduction pathway for the electrochemical reaction of LZTO. (c) Oxygen sublattice of the x = 0.25 sample in the initial (left) and fully discharged state (right) along the [110] direction. Li+/Zn2+ ions in the tetrahedral 8b sites were discarded for the clarity of the display.

in which the ΔV value is relatively low compared to that of

octahedral 4a and 12d sites, while Li+ and Zn2+ ions in the tetrahedral 8b sites remained in their positions. The electrochemical reaction in this region is represented by

region (II) (see Figure 5d). In region (II) at 50 mAh·g−1 < Qdis < ∼160 mAh·g−1, Li+ ions were further inserted into the octahedral sites, together with the migration of Li+ and Zn2+

[Li 0.5Zn 0.5]tet [Li 0.5Ti1.5]oct O4 + y1Li+ + y1e− → [Li 0.5Zn 0.5]tet [Li y1]oct [Li 0.5Ti1.5]oct O4

ions from the tetrahedral 8b sites into the octahedral 4a and 12d (2)

sites. The electrochemical reaction in this region is described by I


Article

Inorganic Chemistry

Table 3. Selected Bond Distances, Bond Angles, and BVS of the x = 0.25 Sample in the Initial and Fully Discharged States sample

atom

initial

fully discharged state

Wyckoff position

distance (Å)

atoms

angle (deg)

BVS

M1 M1 M2

O1 O2 × 3 O2 × 6

8b 8b 4b

1.999(7) 2.102(6) 1.963(5)

O1 × 2 O2 × 2 O2 × 2

12d 12d 12d

1.986(3) 1.962(4) 1.993(5)

M1 M1 M2

O1 O2 × 3 O2 × 6

8b 8b 4b

1.444(9) 2.141(9) 1.811(7)

M3 M3 M3

O1 × 2 O2 × 2 O2 × 2

12d 12d 12d

2.109(3) 2.147(8) 2.047(3)

M4

O2 × 6

4a

2.175(4)

M5 M5 M5

O1 × 2 O2 × 2 O2 × 2

12d 12d 12d

2.377(4) 2.120(9) 2.022(3)

109.0(3) 109.9(3) 96.0(1) 96.3(1) 83.8(1) 96.9(2) 97.4(1) 82.2(1) 81.5(1) 83.1(1) 98.1(2) 118.8(2) 98.8(4) 94.1(1) 88.8(1) 88.4(2) 89.0(1) 97.4(2) 91.9(2) 89.2(1) 74.3(1) 93.9(2) 96.8(1) 92.3(1) 75.2(2) 88.5(2) 87.2(1) 87.8(2) 97.2(1) 77.3(1) 90.5(1)

1.09

M3 M3 M3

O1−M1−O2 O2−M1−O2 O2−M2−O2 O2−M2−O2 O2−M2−O2 O1−M3−O1 O1−M3−O2 O1−M3−O2 O1−M3−O2 O2−M3−O2 O2−M3−O2 O1−M1−O2 O2−M1−O2 O2−M2−O2 O2−M2−O2 O2−M2−O2 O1−M3−O1 O1−M3−O2 O1−M3−O2 O1−M3−O2 O2−M3−O2 O2−M3−O2 O2−M4−O2 O2−M4−O2 O2−M4−O2 O1−M5−O1 O1−M5−O2 O1−M5−O2 O1−M5−O2 O2−M5−O2 O2−M5−O2

3.51

3.35

0.48 5.29

2.43

0.95

1.03

for region (I),

[Li 0.5Zn 0.5]tet [Li y ]oct [Li 0.5Ti1.5]oct O4 + y2 Li+ + y2 e− 1

[Li1 − xZnx]tet [Li y ]oct [Li1/3 + x /3Ti5/3 − x /3]oct O4 + y2 Li+ + y2 e−

→ [Li 0.5 − pZn 0.5 − p]tet [Li y + y + pZn p]oct [Li 0.5Ti1.5]oct O4 1

1

2

(3)

→ [Li1 − x − pZnx − p]tet [Li y + y + pZn p]oct [Li1/3 + x /3Ti5/3 − x /3]oct O4

where p is the amount of Zn ions in the octahedral sites (=goct). Since this reaction accompanies the drastic movement of Zn2+ ions, a large value of ΔV (>0.2 V) was observed in this region. In region (III) at Qdis > 160 mAh·g−1, the Li+ ions were inserted into the tetrahedral 8b sites. The electrochemical reaction is given by

(6)

1

2

2+

+

[Li 0.5 − pZn 0.5 − p]tet [Li y + y + pZn p]oct [Li 0.5Ti1.5]oct O4 + y3 Li + y3 e 1

for region (II), and [Li1 − x − pZnx − p]tet [Li y + y 1

→ [Li1 − x − p + y Znx − p]tet [Li y + y 3

3

1

2

(4)

[Li1 − xZnx]tet [Li1/3 + x /3Ti5/3 − x /3]oct O4 + y1Li+ + y1e− 1

Zn p]oct [Li1/3 + x /3Ti5/3 − x /3]oct O4

2 +p

for region (III). Figure 9a shows the Qdis values in two different voltage ranges and Qtheo as a function of x. The Qdis value measured at 1.0−3.0 V [Qdis(1.0−3.0 V)] monotonically decreases from 133.8 mAh· g−1 at x = 0.125 to 39.6 mAh·g−1 at x = 0.5. These Qdis values are likely to correspond to y1, because, as shown in Figure 4a−d, the voltage hysteresis hardly appeared in these discharge and charge curves. Furthermore, these Qdis values are slightly lower than the calculated Qtheo × (1 − x) values, where (1 − x) is the initial amount of Li+ ions in the tetrahedral 8b sites. Obviously, Qdis is greater than Qtheo; i.e., the Qdis value measured at 0.2−3.0 V [Qdis(0.2−3.0 V)] corresponds to y3. Therefore, the Qdis value between Qdis(1.0−3.0 V) and Qdis(0.2−3.0 V) is attributed to y2, which produced the large voltage hysteresis between the discharge and charge curves. Comparing Qdis(0.2−3.0 V), Qdis(1.0−3.0 V), and Qtheo clarifies the contribution of y1 in region (I), y2 in region (II),

Hence, the extra insertion of Li+ ions into the 8b sites is the reason the actual Qrecha exceeds the Qtheo value on the basis of the one-electron transfer (=152.77 mAh·g−1). In addition, because this reaction did not accompany the movement of Zn2+ ions, the ΔV value decreased slightly to ∼0.1 V (see Figure 5d). On the other hand, during the charge reaction, the reaction proceeded in the order of region (III) → region (II) → region (I). As in the case of x < 0.5, the electrochemical reaction is generally represented by

→ [Li1 − xZnx]tet [Li y ]oct [Li1/3 + x /3Ti5/3 − x /3]oct O4

1

(7)

−

2

→ [Li 0.5 − p + y Zn 0.5 − p]tet [Li y + y + pZn p]oct [Li 0.5Ti1.5]oct O4

Zn p]oct [Li1/3 + x /3Ti5/3 − x /3]oct O4 + y3 Li+ + y3 e−

2 +p

(5) J


Article

Inorganic Chemistry and y3 in region (III). Note that a Qdis value greater than Qtheo based on the one-electron transfer was also achieved for x = 0 when the discharge cutoff voltage was set to 0.02 V.33 Origin of the Voltage Hysteresis. The voltage hysteresis in LZTO was undoubtedly caused by the movement of Zn2+ ions. As shown in Figure 9b, the Zn2+ ions should migrate from the tetrahedral 8b sites into the adjacent octahedral 4a or 12d sites during the discharge reaction. In contrast, the Zn2+ ions in the octahedral 4a or 12d sites move back to the tetrahedral sites during the charge reaction. The movement of Zn2+ ions correlates with the diffusion of Li+ ions from the tetrahedral 8b sites into adjacent octahedral 4a or 12d sites. In other words, for providing the conduction pathway of Li+ ions, Zn2+ ions should move back and forth. The conduction pathway of Zn2+ ions is essentially the same for the discharge and charge reactions, although g for the destination site is different for each reaction. During the discharge reaction, the octahedral 4a or 12d sites are already occupied by the initially inserted Li+ ions, and therefore, the Zn2+ ions should push aside the existing Li+ ions. However, during the charge reaction, the Zn2+ ions could easily move back to the almost empty tetrahedral 8b sites. This difference in situation produces a difference in the diffusivity of Zn2+ ions for the tetrahedral and octahedral sites, resulting in the extremely long relaxation time for the discharge reaction. Moreover, the low operating voltages below 1.0 V of the discharge reaction mean that Zn2+ ions are more stable in the tetrahedral sites, requiring a larger energy to migrate into the octahedral sites. The difference between LZTO and Li 1.2 Fe 0.4 Ti 1.4 O 4 spinels26,27 is interesting to note. The general formula of Li1.2Fe0.4Ti1.4O4 is represented by (Li0.7Fe0.3)tet[Li0.5Fe0.1Ti1.4]octO4 due to cation ordering in the octahedral sites.26 This cation ordering partially disappeared as the discharge reaction proceeded and appeared again with the charge reaction, indicating that the Li+, Fe3+, and Ti4+ ions in the octahedral sites participated in the electrochemical reaction.27 Li1.2Fe0.4Ti1.4O4 exhibited a large voltage hysteresis (∼0.7 V) at the end of the discharge reaction, but differences were rarely observed between the OCVs of the discharge and charge reactions. In the case of LZTO, the Li+ and Ti4+ ions, which are located in the octahedral 4b and 12d sites, respectively, did not participate in the electrochemical reaction. The utilization of the initially occupied octahedral sites produces another conduction pathway such as 8b → (24e) → 4b → (24e) → 12d, possibly increasing the diffusivity of Zn2+ ions during the discharge reaction. Zero-Strain Reaction Scheme. To clarify the reaction mechanism of the perfect zero strain in x = 0.25, Figure 9c shows the oxygen sublattice in the initial and fully discharged states along the [110] direction. Furthermore, Table 3 lists selected bond distances, bond angles, and bond valence sums (BVSs) at the initial and fully discharged states. The oxygen sublattice was initially deformed from the regular position (=0.375 for O1) due to a slight deviation in an oxygen positional parameter. However, in the fully discharged state, such deformation was relaxed, and the O1 atom was almost located at its regular position (=0.372, see Table S4). In addition, some bond distances and bond angles were drastically different in the fully discharged state; for instance, the bond distance between M2 and O2 decreased from 1.963(5) to 1.811(7) Å, where M1, M2, M3, M4, M5, and M6 in Table 2 correspond to mixed cations in the 8b, 4b, 12d, 4a, and 12d sites, respectively. Hence, the perfect zero strain in x = 0.25

is achieved by changes in bond distances and bond angles in the lattice. In the fully discharged state, the bond between M2 and O2 (≃1.81 Å) is excessively short, considering the values of rLi+ (=0.76 Å for C.N. = 6) and rTi3+ (=0.61 Å for C.N. = 6). Moreover, the bond distance between M1 and O1 exhibited the shortest value of 1.444(9) Å, indicating that the M1 atom is closest to the M4 atom in the octahedral 4a site. These very short bond distances provide extraordinarily small or large BVS values, although the BVS analysis was not applied to discharged or charged LIB materials. The origins of these short bonds are currently unclear, but as shown in Figure S8, the bond distance between M1 and O1 in x = 0.5 indicated a systematic change during the discharge and charge reactions. Further neutron diffraction measurements and extended X-ray absorption fine structure analyses could further elucidate changes in bond distances and/or bond angles for the LZTO samples. Finally, we wish to comment on an advantage of the x = 0.25 sample. One of the biggest issues in all-solid-state LIBs is constructing robust solid−solid interfaces between electrodes and solid electrolytes.34,35 The perfect zero strain is ideal for allsolid-state LIBs, because electrical and ionic conductions are maintained during the whole reaction. As shown in Figure 2e, the maximum size of the x = 0.25 particles was ∼300 nm. Furthermore, as listed in Tables S1 and S4, Δac of x = 0.25 was within 0.001 Å (0.1 × 10−3 nm). These results indicate that the clearance between the electrode and solid electrolyte can be suppressed within ∼0.36 Å (0.036 nm), which is significantly small compared with the size of the parent particle.

■

CONCLUSION We have clarified that the voltage hysteresis in the x = 0.5 sample is caused by the reversible movement of Zn2+ ions between the tetrahedral and octahedral sites. The Qrecha values of the x = 0.125, 0.25, 0.375, and 0.5 reached more than ∼170 mA·g−1, which was slightly greater than the Qtheo based on the oneelectron transfer. Capacity fading during the cycling at 25 °C was rarely observed, although the voltage hysteresis between the discharge and charge curves increased with x. OCV measurements revealed that such voltage hysteresis is due to a kinetic factor and that the relaxation in the discharge reaction is extremely slow compared to that in the charge reaction. Furthermore, a relaxation time of more than 10 years was estimated to be required for the OCV values of the discharge and charge reactions to coincide. According to ex situ XRD measurements on the x = 0.5 sample, the Zn2+ ions in the tetrahedral 8b sites migrated into the octahedral 4a and 12d sites during the discharge reaction. During the charge reaction, in contrast, the Zn2+ ions moved back to the tetrahedral 8b sites, and no differences in gtet/(gtet + goct) appeared between the discharge and charge reactions. Despite such a drastic movement of the Zn2+ ions, ac remained almost constant during the whole electrochemical reaction, providing the zero-strain character. The perfect zero strain, which has never been reported for LIB materials, was achieved with the x = 0.25 sample. We elucidated the contributions of the amount of Zn2+ ions in LZTO on the voltage hysteresis, OCVs, and changes in the crystal structures. However, the voltage hysteresis remained for all of the LZTO samples. Decreasing and/or eliminating such hysteresis will be required for practical applications to LIBs. Utilizing the other octahedral sites could be a solution for such a purpose. K


Article

Inorganic Chemistry

■

Li2ZnTi3O8 by Liquated Na2MoO4 to Boost Electrochemical Performance. ACS Appl. Mater. Interfaces 2017, 9, 43603−43613. (9) Tang, H.; Tang, Z.; Du, C.; Qie, F.; Zhu, J. Ag-Doped Li2ZnTi3O8 as a High Rate Anode Material for Rechargeable Lithium-Ion Batteries. Electrochim. Acta 2014, 120, 187−192. (10) Yi, T.-F.; Wu, J.-Z.; Yuan, J.; Zhu, Y.-R.; Wang, P.-F. Rapid Lithiation and Delithiation Property of V-Doped Li2ZnTi3O8 as Anode Material for Lithium-Ion Battery. ACS Sustainable Chem. Eng. 2015, 3, 3062−3069. (11) Chen, W.; Zhou, Z.; Wang, R.; Wu, Z.; Liang, H.; Shao, L.; Shu, J.; Wang, Z. High Performance Na-Doped Lithium Zinc Titanate as Anode Material for Li-Ion Batteries. RSC Adv. 2015, 5, 49890−49898. (12) Chen, W.; Liang, H.; Qi, Z.; Shao, L.; Shu, J.; Wang, Z. Enhanced Electrochemical Properties of Lithium Cobalt Titanate via Lithium-Site Substitution with Sodium. Electrochim. Acta 2015, 174, 1202−1215. (13) Meng, Z.; Wang, S.; Wang, L.; Hou, H. Synthesis of High Performance N-Doped Carbon Coated Li2ZnTi3O8 via a NTA-Assisted Solid-State Route. Dalton Trans. 2018, 47, 2711−2718. (14) Wu, Y.-R.; Pan, J.; Ren, S.; Xie, Y.; Yue, C.; Yi, T.-F. Review and Prospect of Li2ZnTi3O8-Based Anode Materials for Li-Ion Battery. Ionics 2019, 25, 373−397. (15) Blasse, G. Crystal Chemistry and Some Magnetic Properties of Mixed Metal Oxides with Spinel Structure; Philips Research Reports; Philips Research Laboratories: 1964; Vol. 3, pp 1−139. (16) Hernandez, V. S.; Torres Martinez, L. M.; Mather, G. C.; West, A. R. Structures and Polymorphism of Spinel-Like Phases, Li1.33xZn2−2xTi1+0.67xO. J. Mater. Chem. 1996, 6, 1533−1536. (17) He, L.; Mi, S.-B.; Jin, X.; Zhang, H.; Zhou, D.; Xiang, F.; Yang, H. Order−Disorder Phase Transition and Magneto- Dielectric Properties of (1 − x)LiFe5O8−xLi2ZnTi3O8 Spinel-Structured Solid Solution Ceramics. J. Am. Ceram. Soc. 2015, 98, 2122−2129. (18) Ohzuku, T.; Ueda, A.; Yamamoto, N. Zero-Strain Insertion Material of Li[Li1/3Ti5/3]O4 for Rechargeable Lithium Cells. J. Electrochem. Soc. 1995, 142, 1431−1435. (19) Mukai, K.; Kato, Y.; Nakano, H. Understanding the Zero-Strain Lithium Insertion Scheme of Li[Li1/3Ti5/3]O4: Structural Changes at Atomic Scale Clarified by Raman Spectroscopy. J. Phys. Chem. C 2014, 118, 2992−2999. (20) Mukai, K.; Kato, Y. Role of Oxide Ions in Thermally Activated Lithium Diffusion of Li[Li1/3Ti5/3]O4: X-ray Diffraction Measurements and Raman Spectroscopy. J. Phys. Chem. C 2015, 119, 10273−10281. (21) Izumi, F.; Momma, K. Three-Dimensional Visualization in Powder Diffraction. Solid State Phenom. 2007, 130, 15−20. (22) Momma, K.; Izumi, F. VESTA 3 for Three-Dimensional Visualization of Crystal, Volumetric and Morphology Data. J. Appl. Crystallogr. 2011, 44, 1272−1276. (23) Mukai, K.; Ikedo, Y.; Kamazawa, K.; Brewer, J. H.; Ansaldo, E. J.; Chow, K. H.; Månsson, M.; Sugiyama, J. The Gradient Distribution of Ni Ions in Cation-Disordered Li[Ni1/2Mn3/2]O4 Clarified by MuonSpin Rotation and Relaxation (μSR). RSC Adv. 2013, 3, 11634−11639. (24) Mukai, K.; Uyama, T. Toward Positive Electrode Materials with High-Energy Density: Electrochemical and Structural Studies on LiCoxMn2−xO4 with 0 ≤ x ≤ 1. ACS Omega 2017, 2, 5142−5149. (25) Haas, C. Phase Transitions in Crystals with The Spinel Structure. J. Phys. Chem. Solids 1965, 26, 1225−1232. (26) Mukai, K.; Kato, Y.; Nakano, H. Determination of Cation Distribution in the Fe[Li1/2Fe3/2]O4- LiFeTiO4-Li[Li1/3Ti5/3]O4 System: Mixed Nature of Solid Solution and Superlattice. J. Solid State Chem. 2017, 247, 67−76. (27) Mukai, K. A Series of Zero-Strain Lithium Insertion Materials that Undergo a Non-Topotactic Reaction. Electrochim. Acta 2018, 263, 508−514. (28) Shannon, R. D. Revised Effective Ionic Radii and Systematic Studies of Interatomic Distances in Halides and Chalcogenides. Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr. 1976, 32, 751−767. (29) Mukai, K.; Inoue, T.; Hasegawa, M. Rationalizing Thermal Reactions of C6Lix Negative Electrode with Nonaqueous Electrolyte. J. Power Sources 2017, 366, 185−192.

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.inorgchem.9b01565.

■

SEM images of the x = 0.125 and 0.375 samples; Rietveld analysis results for the x = 0.125, 0.25, and 0.375 samples and their structural parameters; the OCV measurements of the x = 0.125 sample; Q dependence of the 400 diffraction line and 4 × d400, I(220), I(311), and I(400) as a function of Q; Rietveld analysis results at discharged points D2, D3, and D4; Rietveld analysis results at charged points C2, C3, and C4l; structural parameters at discharged points D2, D3, and D4; structural parameters at charged points C2, C3, and C4; and change in bond distances for the x = 0.5 sample (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +81-561-71-7698. Fax: +81-561-63-6119. ORCID

Kazuhiko Mukai: 0000-0002-6154-6539 Notes

The author declares no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported in part by a Grant-in-Aid for Scientific Research (C), 25410207, from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The author appreciates Mr. Hiroaki Kadoura of TCRDL for SEM observations and Mr. Takeshi Uyama of TCRDL for help with the synchrotron XRD measurements. The synchrotron XRD was measured at the Aichi Synchrotron Radiation Center, Aichi Science & Technology Foundation, Japan (Proposal Nos. 201806019 and 201806066).

■

REFERENCES

(1) Pistoia, G. Lithium-Ion Batteries: Advances and Applications; Elsevier: The Netherlands, 2014. (2) Hong, Z.; Wei, M.; Ding, X.; Jiang, L.; Wei, K. Li2ZnTi3O8 Nanorods: A New Anode Material for Lithium-Ion Battery. Electrochem. Commun. 2010, 12, 720−723. (3) Wang, L.; Wu, L.; Li, Z.; Lei, G.; Xiao, Q.; Zhang, P. Synthesis and Electrochemical Properties of Li2ZnTi3O8 Fibers as An Anode Material for Lithium-Ion Batteries. Electrochim. Acta 2011, 56, 5343−5346. (4) Chen, W.; Liang, H.; Ren, W.; Shao, L.; Shao, L.; Shu, J.; Wang, Z. Complex Spinel Titanate as An Advanced Anode Material for Rechargeable Lithium-Ion Batteries. J. Alloys Compd. 2014, 611, 65− 73. (5) Li, Z.-F.; Cui, Y.-H.; Wu, J.-W.; Du, C.-Q.; Zhang, X.-H.; Tang, Z.Y. Synthesis and Electrochemical Properties of Lithium Zinc Titanate as An Anode Material for Lithium Ion Batteries via Microwave Method. RSC Adv. 2016, 6, 39209−39215. (6) Li, H.; Li, Z.; Cui, Y.; Ma, C.; Tang, Z. Long-Cycled Li2ZnTi3O8/ TiO2 Composite Anode Material Synthesized via a One-Pot CoPrecipitation Method for Lithium Ion Batteries. New J. Chem. 2017, 41, 975−981. (7) Yang, H.; Wang, X.-H.; Qi, Y.-X.; Lun, N.; Cao, Y.-M.; Bai, Y.-J. Improving the Electrochemical Performance of Li2ZnTi3O8 by Surface KCl Modification. ACS Sustainable Chem. Eng. 2017, 5, 6099−6106. (8) Yang, H.; Park, J.; Kim, C.-S.; Xu, Y.-H.; Zhu, H.-L.; Qi, Y.-X.; Yin, L.; Li, H.; Lun, N.; Bai, Y.-J. Uniform Surface Modification of L


Article

Inorganic Chemistry (30) Scharner, S.; Weppner, W.; Schmid-Beurmann, P. Evidence of Two-Phase Formation upon Lithium Insertion into the Li1.33Ti1.67O4 Spinel. J. Electrochem. Soc. 1999, 146, 857−861. (31) Ronci, F.; Reale, P.; Scosati, B.; Panero, S.; Rossi Albertini, V.; Perfetti, P.; di Michiel, M.; Merino, J. M. High-Resolution In-Situ Structural Measurements of the Li4/3Ti5/3O4 “Zero-Strain” Insertion Material. J. Phys. Chem. B 2002, 106, 3082−3086. (32) Ohzuku, T.; Tatsumi, K.; Matoba, N.; Sawai, K. Electrochemistry and Structural Chemistry of Li[CrTi]O4 (Fd3̅m) in Nonaqueous Lithium Cells. J. Electrochem. Soc. 2000, 147, 3592−3597. (33) Mukai, K.; Yamada, I. High-Pressure Study of Li[Li1/3Ti5/3]O4. Inorg. Chem. Front. 2018, 5, 1941−1949. (34) Sun, C. W.; Liu, J.; Gong, Y. D.; Wilkinson, D. P.; Zhang, J. J. Recent Advances in All-Solid-State Rechargeable Lithium Batteries. Nano Energy 2017, 33, 363−386. (35) Manthiram, A.; Yu, X. W.; Wang, S. F. Lithium Battery Chemistries Enabled by Solid-State Electrolytes. Nat. Rev. Mater. 2017, 2, 16103.

M


Reversible Movement of Zn2+ Ions with Zero-Strain Characteristic

Recommend Documents