Integrated rocksalt-polyanion cathodes with excess lithium and stabilized cycling

Received: 8 January 2024 收到：2024 年 1 月 8 日
Accepted: 24 July 2024 接受：2024 年 7 月 24 日
Published online: 23 August 2024
Check for updates 检查更新

Moonsu Yoon (

0^{3, 6}

,Sipei Li

^{3}

, Baoming Wang

0^{3}

, Ethan Yupeng Zheng

0^{1}^{1,}

Jinhyuk Lee

\oplus^{7}

, Yongwen Sun

0^{4}

, Ying Han

^{4}

, Jim Ciston

0^{8}

, Colin Ophus

0^{8}

, Chengyu Song

^{8}

, Aubrey Penn

^{9}

, Yaqi Liao

^{10}

, Haijin Ji

^{10}

, Ting Shi

^{10}

, Mengyi Liao

^{10}

, Zexiao Cheng

^{10}

, Jingwei Xiang

^{10}

, Yu Peng

^{11}

, Lu Ma

^{12}

, Xianghui Xiao(

^{12}

, Wang Hay Kan (

^{13, 14}

, Huaican Chen

^{13, 14}

, Wen Yin

^{13, 14}

, Lingling Guo

^{15}

, Wei-Ren Liu

^{16}

, Rasu Muruganantham (

^{16}

, Chun-Chuen Yang

0^{17}

, Yuntong Zhu

^{1}

， Qingjie Li

^{3}

& Ju Li(

^{1, 3}

Abstract

Co- and Ni-free disordered rocksalt cathodes utilize oxygen redox to increase the energy density of lithium-ion batteries, but it is challenging to achieve good cycle life at high voltages $> 4.5 V ($ versus ${Li}^{-} {Li}^{+})$ . Here we report a family of Li-excess Mn-rich cathodes that integrates rocksalt- and polyanion-type structures. Following design rules for cation filling and ordering, we demonstrate the bulk incorporation of polyanion groups into the rocksalt lattice. This integration bridges the two primary families of lithium-ion battery cathodes-layered/spinel and phosphate oxidesdramatically enhancing the cycling stability of disordered rocksalt cathodes with 4.8 V upper cut-off voltage. The cathode exhibits high gravimetric energy densities above $1, 100 Wh {kg}^{- 1}$ and $> 70 %$ retention over 100 cycles. This study opens up a broad compositional space for developing battery cathodes using earth-abundant elements such as Mn and Fe.

Rapid growth of electricity storage capabilities with lithium-ion batteries (LIBs) is required to realize a sustainable energy infrastructure

^{1}

. In terms of resources, Co is

\sim 5 \times

the price of Li on a molar basis

^{2, 3}

, and Ni is

\sim 2 \times

(ref. 3 ); thus, we would run into Co or Ni crises before Li. For advanced LIB cathodes, eliminating Co and Ni usage would greatly improve the scalability of electricity storage

^{4}

. Disordered rocksalt (DRX) cathodes

^{5, 6}

are attractive for being potentially Co - and Ni -free, while having high energy densities (approaching

1, 100 Wh {kg}^{- 1}

(ref.7)). On the other hand, to reach high energy densities (

> 900 Wh {kg}^{- 1}

), high upper cut-off voltages (for example, 4.8 V versus

{Li}^{-} {Li}^{+}

for

{DRX}^{7 - 11}

) are required for cathodes, which means highly delithiated states with most of the

{Li}^{+}

-hosting sites vacant. This often triggers the participation of oxygen anion redox and eventually irreversible oxygen loss, as delithiation lowers the Fermi level towards or dropping below the top of the oxygen

2 p

band, especially at the surface and interfaces

^{6, 12 - 14}

. A heavy usage of hybrid anion- and cation-redox with more exotic oxygen valence

O^{α -} (0 < α < 2)

challenges the cycling stability of the
cathode since

O^{α -}

tends to be more mobile, leading to percolating lattice oxygen diffusion to the reactive surface, extensive side reactions with the electrolyte and finally structural and chemical instability at the surface and in the bulk

^{15 - 19}

. These are critical issues for

{DRX}^{6, 20}

and other high-energy-density cathodes

^{16, 21}

LIB cathodes are mainly constructed on face-centred cubic (FCC) oxygen or lower-symmetry polyanion framework (hexagonal close-packed, HCP , oxygen for

{LiFePO}_{4}

, the most useful polyanion cathode). The former has cation ordering in the parent rocksalt structure, which includes high-energy-density cathodes of

{LiCoO}_{2}

, Ni-rich layered cathodes and Li -/Mn-rich layered cathodes

^{22, 23}

(spinel and DRX cathodes are also rocksalt structure derivatives with FCC oxygen sublattice). They have high theoretical capacities

> 270 mAh g^{- 1}

, and extensive research has been conducted to improve their high-voltage stability. The latter is exemplified by

{LiFePO}_{4}

, with exceptional structural, electrochemical and thermal stability, yet limited by the low theoretical capacity (

170 mAh g^{- 1}

) and low energy density at the cathode level

^{24 - 26}

A full list of affiliations appears at the end of the paper. e-mail: dongyanhao@tsinghua.edu.cn; liju@mit.edu

Fig.1|Design of DRXPS cathodes. a, The structure of

M_{2} O_{4} \cdot b

, The structure of

M_{2 - u} O_{4}

. c, The structure of

M_{2 - u} {[{XO}_{4}]}_{x} O_{4 (1 - x)}

. d, Comparison of crystallographic tetrahedra size for polyanion olivine and rocksalt-type cathodes.

LMO, {LiMn}_{2} O_{4}; LNMO, {LiNi}_{0.5} {Mn}_{1.5} O_{4}; LCO, {LiCoO}_{2}; {NCM 111}^{-} {LiNi}_{1 / 3} {Co}_{1 / 3} {Mn}_{1 / 3} O_{2}; {LNO}_{2}, {LiNiO}_{2}

Marriage between the two families may offer synergistically improved energy density and stability. However, few reports

^{27}

of their integration testify to the incompatibility between rocksalt and polyanion structures.

This work seeks to resolve the above conundrum with the invention of integrated rocksalt-polyanion cathodes. These compositions originate fromDRX chemistries, and a major effort here is to improve the cycling stability under high upper cut-off voltages (required to deliver high capacity and energy density). We successfully produced a family of Li-excessCo-andNi-freedisordered rocksalt-polyanionic spinel (DRXPS) cathodes, with a general chemical formula of

{Li}_{2 + u - v} M_{2 - u} {[{XO}_{4}]}_{x} O_{4 (1 - x)}

. Here,

M

denotes transition metals such as Mn and

Fe, {XO}_{4}

denotes polyanion groups such as

{PO}_{4}, {SiO}_{4}

and

{SO}_{4}

, and

u, v

and

x

describe the designed stoichiometries. This family of compounds is called DRXPS because they are designed on a parent DRX structure and have bulk polyanion incorporation and spinel-type cation ordering (that gives a spinel diffraction pattern). Remarkable improvements of the cycling stability over reported DRX cathodes have been achieved in

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

{Li}_{1.67} {Mn}_{1.5} B_{0.17} O_{4}, {Li}_{1.67} {Mn}_{1.25} {Fe}_{0.25} P_{0.17} O_{4}

and four more compositions, all belonging to the DRXPS family. The DRXPS cathodes have high capacities (

> 350 mAh g^{- 1}

), high energy densities (

> 1, 100 Wh {kg}^{- 1}

), stable cycling (>70% energy density retention over 100 cycles), good rate performance and a highly tunable compositional space. The general design principles and experimental efforts presented here offer avenues for the future development of Co - and Ni -free cathodes.

Materials design 材料设计

Our task is to design high-capacity oxide cathodes with excess Li, anion redox activity, bulk polyanion incorporation and good electrochemical stability. Starting from the high-capacity FCC oxygen framework, a three-dimensionally connected spinel structure

M_{2} O_{4}

(Fig.1a, Li is not shown for simplicity) provides the best hybridization between transition metal (M)

d

and oxygen

2 p

orbitals under the constrained Li/M molar ratio of 1 (each oxygen is coordinated with one tetrahedral Li and three octahedral M). Further raising the Li/M molar ratio above (that is, replacing some M in Fig. 1a by Li ) increases the theoretical capacity, and anion redox is simultaneously activated with underbonded
oxygen (Fig. 1b). These underbonded oxygen can be oxidized upon charging to high voltages and may eventually leave the lattice in the form of outgassing if a percolative kinetic pathway exists from the bulk to the surface

^{28}

. We aim to shut down the labile oxygen percolation by incorporating some polyanion groups into the Li-excess lattice (Fig.1c), utilizing the strong

X - O

covalent bonds to mitigate oxygen instability.

Practical realization of the above is challenging and comes to the same incompatibility issue between rocksalt and polyanion structures discussed above. The main reasons are twofold. First, the cations in polyanion cathodes are not close-packed. The octahedral sites face-shared with

{XO}_{4}

tetrahedra need to be empty

^{24}

. This conflicts with cation-filling rules in layered and DRX cathodes (the octahedral sites are fully occupied). Second, X-O covalent bonds are short and strong, which results in much shorter

O - O

distances (characterizing the tetrahedral size) than the ones in rocksalt-structure cathodes. For example, the true tetrahedra size calculated from the P -O bond length in the polyanion olivine cathode

{LiFePO}_{4}

(refs. 29,30 ) is

12 - 15 %

smaller than that in rocksalt-structure cathodes

^{31 - 34}

(Fig.1d). This would result in large lattice distortion and, thus, difficulty in making a solid-solution phase between

{XO}_{4}

polyanions and 'normal O ’ anions.

We propose the following solution to the two problems mentioned above. For the first one, cation deficiency is an effective approach. Specifically, we were inspired by the polyhedral occupation rules in spinel cathodes: octahedra at 16 d sites are fully occupied, and octahedra at 16c sites (face-shared with tetrahedra at 8a sites) are empty. So spinel-like cation ordering is preferred. For the second one, typical high-temperature solid-state synthesis would not work, and we resort to lower-temperature mechanochemical synthesis. Without going into the detailed derivations of the optimal values of stoichiometry (

u, v

and

x

) in Supplementary Note 1, we show in the following sections that the above simple design rules are powerful enough to guide the synthesis of the DRXPS cathode series.

Structure and morphology of prototype

${L i}_{1.67} {M n}_{1.5} P_{0.17} O_{4}$

A prototype DRXPS cathode

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

was synthesized by a one-pot mechanochemical method. The obtained sample has a

Fig.

2 ∣

Structural characterization of

{Li}_{1.67} {M n}_{1.5} P_{0.17} O_{4} . a, X R D

patterns of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

. Open black circles are experimental, and solid black line is calculated. b, PDF of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4} . G (r) = 4 π r (ρ (r) - ρ_{0})

, where

ρ (r)

is the local atomic number density at distance

r

from a reference atom, and

ρ_{0}

is the average atomic density of the material. c, SEM image of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

. Scale bar, 200 nm . d,STEM-EDS mapping of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

. Scale bar,

100 nm . e

, TEM image of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

. Scale bar, 5 nm . Inset:SAED pattern. Scale bar,

2 {nm}^{- 1}

. f, STEM-EELS mapping of Mn, P and O, performed on a single-crystal grain close
to a zone axis, as indicated by the yellow dashed box. Note that the EELS signal of

P

is very weak due to the small atomic ratio of

P

in the composition; thus, the data supports but does not prove the uniform spatial distribution of P. Scale bars, 2 nm .g, Structural model of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4} \cdot

h, HAADF-STEM image of

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

. Inset: Atomic positions with alternating intensities at 16 d sites, characteristic of a spinel structure. Scale bar, 1 nm . i, Filtered image of

h

. Insets: 16d and 8a site signals. Scale bar, 1 nm .
composition close to the designed stoichiometry (shown by the inductively coupled plasma mass spectrometry (ICP-MS) data in Supplementary Table1). Its X-ray diffraction (XRD) pattern (Fig. 2a) matches with a single-phase cubic spinel structure (

a = b = c, α = β = γ = 90^{\circ}

;Supplementary Fig.1). Rietveld refinement yields a lattice constant

a = 8.1527 Å

(Supplementary Fig. 1 and Supplementary Table 2), which is slightly smaller than those of spinel cathodes (

8.246 Å

for

{LiMn}_{2} O_{4}

and

8.172 Å

for

{LiNi}_{0.5} {Mn}_{1.5} O_{4}

). Neutron powder diffraction measurement (Supplementary Fig. 2) and refinement (Supplementary Table3) were further conducted on

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

for better sensitivity on Li sites, which shows consistent results with XRD measurements. For more structural information, we conducted pair distribution function (PDF) analysis on

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

(Fig. 2b) and compared with references of

{LiMn}_{2} O_{4}

and

{LiFePO}_{4}

(Supplementary Fig. 3). First-nearest-neighbour

P - O

pair at

1.549 Å

was observed, which is slightly longer than the

P - O

pair in the

{PO}_{4}

group of

{LiFePO}_{4} (1.520 Å

). First-nearest-neighbour Mn-O pair at

1.893 Å

and

Mn - Mn

pair

2.858 Å

were observed, which are slightly shorter than the corresponding ones in

{LiMn}_{2} O_{4} (1.903 Å

for

Mn - O

and

2.887 Å

for

Mn - Mn)

. These elastic straining effects are consistent with our materials design (tensile strained for

{XO}_{4}

compared with

{LiFePO}_{4}

and compressive strained for

{MO}_{6}

compared with

{LiMn}_{2} O_{4}

). The effect smears at longer distances, for example, second-nearest-neighbour

Mn - O

distances (

3.418 Å \circ

) are similar in

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

and

{LiMn}_{2} O_{4}

. Raman spectroscopy measurement (Supplementary Fig. 4) was conducted for local structure analysis. The Raman peak at

\sim 940 {cm}^{- 1}

can be assigned to the

A_{\lg}

mode of

{PO}_{4}

(ref. 35), the peak at

\sim 600 {cm}^{- 1}

can be assigned to the symmetric stretching mode of

{MnO}_{6}

(refs. 36,37 ) and the peaks at

420 - 490 {cm}^{- 1}

can be assigned to the symmetric stretching modes of

{LiO}_{4}

and

{LiO}_{6}

(ref. 36). These Raman features support tetrahedral occupation of P , octahedral occupation of Mn and mixed tetrahedral/octahedral occupations of Li.

The scanning electron microscopy (SEM) image in Fig. 2c shows that

{Li}_{1.67} {Mn}_{1.5} P_{0.17} O_{4}

has an average size of

\sim 150 nm

for the agglomerates (see size distribution in Supplementary Fig. 5). Energy dispersive spectroscopy mapping in scanning transmission electron microscopy (STEM-EDS) (Fig. 2d) shows a uniform distribution of Mn, P and O . The transmission electron microscopy (TEM) image in Fig. 2e shows that the particles in Fig. 2c are polycrystalline, consisting of 5-10 nm ‘primary’ particles that are crystalline. A characteristic lattice spacing

d = 4.70 Å

can be identified, corresponding to the (111) plane of the spinel structure. The selected area electron diffraction (SAED) pattern (Fig. 2e, inset) further confirms the polycrystallinity, with diffraction rings corresponding to the (111), (311), (400), (511) and (440) peaks. Figure 2 f shows the electron energy loss spectroscopy (EELS) mapping

Integrated rocksalt-polyanion cathodes with excess lithium and stabilized cycling

Abstract

Materials design 材料设计

Structure and morphology of prototype

L i 1.67 M n 1.5 P 0 . 1 7 O 4 L i 1.67 M n 1.5 P 0 . 1 7 O 4 Li_(1.67)Mn_(1.5)P_(0.17)O_(4)\mathbf{L i}_{1.67} \mathbf{M n}_{1.5} \mathbf{P}_{\mathbf{0 . 1 7}} \mathbf{O}_{\mathbf{4}}

${L i}_{1.67} {M n}_{1.5} P_{0.17} O_{4}$