Lattice pinning in MoO3 via coherent interface with stabilized Li+ intercalation

Large lattice expansion/contraction with Li+ intercalation/deintercalation of electrode active materials results in severe structural degradation to electrodes and can negatively impact the cycle life of solid-state lithium-based batteries. In case of the layered orthorhombic MoO3 (α-MoO3), its large lattice variation along the b axis during Li+ insertion/extraction induces irreversible phase transition and structural degradation, leading to undesirable cycle life. Herein, we propose a lattice pinning strategy to construct a coherent interface between α-MoO3 and η-Mo4O11 with epitaxial intergrowth structure. Owing to the minimal lattice change of η-Mo4O11 during Li+ insertion/extraction, η-Mo4O11 domains serve as pin centers that can effectively suppress the lattice expansion of α-MoO3, evidenced by the noticeably decreased lattice expansion from about 16% to 2% along the b direction. The designed α-MoO3/η-Mo4O11 intergrown heterostructure enables robust structural stability during cycling (about 81% capacity retention after 3000 cycles at a specific current of 2 A g−1 and 298 ± 2 K) by harnessing the merits of epitaxial stabilization and the pinning effect. Finally, benefiting from the stable positive electrode–solid electrolyte interface, a highly durable and flexible all-solid-state thin-film lithium microbattery is further demonstrated. This work advances the fundamental understanding of the unstable structure evolution for α-MoO3, and may offer a rational strategy to develop highly stable electrode materials for advanced batteries.

interaction 9 .A high theoretical specific capacity of about 372 mAh g −1 is expected, corresponding to the Mo 6+ /Mo 4+ redox couple with two lithium storage per Mo, which makes α-MoO 3 a promising positive electrode for Li-ion storage [10][11][12] .However, the insertion of Li + ions into the layered α-MoO 3 breaks the thermodynamically stable state and induces large lattice expansion (~16%), resulting in the irreversible phase transformation to acicular Li x MoO 3 (x ~0.25) and fast capacity fading (Fig. 1a) 13,14 .To prolong the cycle life of α-MoO 3 , various strategies including constructing nanostructures 15 , nitrogen doping 16 , and widening the vdW interlayer 10,13 of α-MoO 3 have been extensively explored.Nevertheless, there is still notable capacity loss, especially during the initial cycles for the modified α-MoO 3 10,13   , suggesting more effective approaches need to be developed to further stabilize the layered structure without compromising its large specific capacity.
Recent studies on constructing spinel-layered heterostructures for LiMnO 2 and Li-rich (Li 2 MnO 3 ) electrode materials suggest that interface engineering could be an effective way to improve the structural stability of Li + intercalation host 17,18 .To reduce the large lattice expansion of α-MoO 3 during the lithiation process, we propose a coherent epitaxial interface pinning design, which can inhibit the mobility of Mo atoms via strong ionic bonds at coherent interfaces.In this study, we constructed α-MoO 3 /η-Mo 4 O 11 intergrown heterostructure with coherent epitaxial interface by magnetron sputtering.η-Mo 4 O 11 is a nearly "zero-strain" Li + intercalation host (Fig. 1b), which could be the perfect structural stabilizer for the coherent interface design (Fig. 1c).When a coherent interface can be established between α-MoO 3 and η-Mo 4 O 11 , the high structural stability of η-Mo 4 O 11 can be transferred to α-MoO 3 , which could suppress lattice variation of α-MoO 3 during lithiation and delithiation.Therefore, the coherent interface pinning design could provide an effective structural regulation strategy to stabilize electrode materials with intercalation chemistry.
Herein, pure α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and pure η-Mo 4 O 11 thin films were prepared respectively by manipulating the O 2 /Ar flow ratio during the sputtering process.With the coherent interface, the large lattice expansion and phase transition in α-MoO 3 have been viably suppressed during the lithiation, which was evidenced by the ex situ XRD.The density functional theory (DFT) calculations indicate the interface pinning can effectively improve the structural stability of α-MoO 3 and suppress Mo migration during charge/discharge.Consequently, the α-MoO 3 /η-Mo 4 O 11 heterostructured electrode achieved a reversible specific capacity as high as ~350 mAh g −1 (0.1 A g −1 ; 1.0−3.5 V vs. Li/Li + ) and a high capacity retention of about 81% after 3000 cycles (at a specific current of 2 A g −1 and 298 ± 2 K) in the organic electrolyte, demonstrating greatly enhanced structural stability of α-MoO 3 within the heterostructure.Finally, we successfully fabricated an all-solid-state thin-film lithium microbattery based on the α-MoO 3 /η-Mo 4 O 11 electrode, lithium phosphorous oxynitride (LIPON) solid electrolyte, and Li metal electrode.Benefiting from the stable positive electrode-solid electrolyte interface during Li + intercalation/deintercalation processes, the microbattery showed a long-term cycling stability with about 74% capacity retention after 4000 cycles (at a specific current of 2 A g −1 and 298 ± 2 K).

Results
Construction of α-MoO 3 /η-Mo 4 O 11 heterostructure and structural characterizations The fabrication of α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films on Pt/Ti/glass and Pt/Ti/stainless steel substrates through direct current (DC) magnetron sputtering under O 2 /Ar atmosphere is schematically illustrated in Fig. 2a.During the sputtering process, the asformed ionized Ar (Ar + ) are driven towards the Mo target under the electric field.After Ar + bombardment, the target Mo atoms are released to react with the O -at ~400-800 K and deposit on the substrate in the form of Mo oxides 19 .X-ray diffraction (XRD) and corresponding Rietveld refinement were carried out to identify phase evolution and crystal structure of the films under different deposition conditions.The O 2 /Ar flow ratios of 8%, 13%, and 25% at a deposition temperature of 573 K were selected to prepare α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films, respectively, which are demonstrated by the powder XRD and glancing incidence XRD (GIXRD) patterns (Supplementary Figs. 1 and 2).Specifically, via Rietveld refinement, Fig. 2b reveals that the heterostructure possesses approximately 63 wt % α-MoO 3 and 37 wt% η-Mo 4 O 11 , respectively, indicating α-MoO 3 is the major phase for the heterostructure.The refined phase fraction error for η-Mo 4 O 11 in the heterostructure is 1.3% as provided by the MAUD.The Rietveld-refined XRD patterns demonstrate a monoclinic structure for η-Mo 4 O 11 (Supplementary Fig. 3, JCPDS No. 13-0142) and an orthorhombic structure for α-MoO 3 phase (Fig. 2c and Supplementary Table 1, JCPDS No. 05-0508) 13,20 .Moreover, the chemical composition and Mo valence state of the film can be well controlled by tuning the oxygen partial pressure and substrate temperature during deposition.As shown in Fig. 2d   the bracket shows the standard deviation 21 .The phase evolution in thin films deposited under different O 2 /Ar flow ratios is further confirmed by Raman measurements (Supplementary Fig. 4).The Raman spectrum of the MoO 3 /η-Mo 4 O 11 film was found to contain all the characteristic Raman bands of both α-MoO 3 and η-Mo 4 O 11 , demonstrating the successful fabrication of the heterostructure in the film by magnetron sputtering 22,23 .
The surface morphologies of various samples were investigated by field-emission scanning electron microscopy (FESEM).As shown in Supplementary Fig. 5a-c 3c).Supplementary Fig. 6 shows a low-magnification HAADF-STEM image of the α-MoO 3 /η-Mo 4 O 11 sample, revealing that the thin film is composed of nanograins with irregular shapes.An interfacial region in the α-MoO 3 /η-Mo 4 O 11 heterostructure is shown in Fig. 3d, in which η-Mo 4 O 11 nano-domains with diameters of ~3-5 nm are found to be embedded in the layered α-MoO 3 matrix (Fig. 3d and Supplementary Fig. 7).The atomic arrangement of Mo within the blue square agrees well with the crystal model of monoclinic η-Mo 4 O 11 along the [100] zone axis, whereas the atomic arrangement of Mo within the red square conforms to the crystal model of the layered α-MoO 3 along the [101] zone axis (Fig. 3d).Typically, the α-MoO 3 /η-Mo 4 O 11 heterostructure demonstrates a specific orientation relationship of (020) α //(020) η and [10-1] α //[001] η with a coherent interface between α-MoO 3 and η-Mo 4 O 11 (Fig. 3d).Generally, the direct joining between α-MoO 3 (020) and η-Mo 4 O 11 (020) planes (or a sharp interface) will generate a large lattice mismatch (f) of ~18.8%, defined by f = (d α −2d η )/d α , which is energetically unfavorable to form a coherent interface 24 .However, the HAADF-STEM image at the interfacial region demonstrates that a gradual transition zone with a thickness of ~4-5 nm instead of a sharp interface separates the two phases, thereby enabling the gradual release of misfit strain.In contrast to the sharp interface, the gradual transition interface enables the gradual release of misfit strain between the two lattices, leading to a continuous lattice mismatch less than 5% at the interfacial region.Therefore, the gradual transition interface is critical to reducing lattice mismatch between α-MoO 3 and η-Mo 4 O 11 , resulting in the construction of coherent interface between these two phases.Simultaneously, the large stress and strain issues associated with sharp interface can be effectively mitigated by the gradual transition interface design, further improving the structural stability of the heterostructure.Electron energy-loss spectroscopy (EELS) as a sensitive tool was employed to probe the oxidation state of Mo in the heterostructure.As shown in Fig. 3e, along the yellow line in the HAADF-STEM image, a series of energy-loss near-edge structures (ELNES) spectra of the O Kedge were recorded.The ELNES of O K-edge represents the transition of oxygen s core states to oxygen p states hybridized with Mo 4d orbitals [25][26][27] .In Mo-O octahedral configuration, the crystal field splitting leads to the generation of t 2g and e g symmetry bands, corresponding to peak A at about 532 eV and feature B at about 536 eV, respectively.For α-MoO 3 , both the t 2g and e g orbitals are empty, and thus O 1 s electrons can be excited into both orbitals.Consequently, the corresponding two peaks can be clearly detected at the O K-edge ELNES spectrum of MoO 3 .When the t 2g orbitals in η-Mo 4 O 11 are filled with electrons, an intensity decrease in peak A can be observed in O K-edge spectrum 25 .From the series of O K-edge ELNES spectra on the marked sites of α-MoO 3 /η-Mo 4 O 11 in Fig. 3e, it can be observed that the spectra of 3 and 6 show obviously decreased intensity at peak A, suggesting embedded nano-domains of η-Mo 4 O 11 in α-MoO 3 matrix.The evolution of Mo chemical state in different samples was also investigated by X-ray photoelectron spectroscopy (XPS), as shown in Supplementary Fig. 8. Based on the Mo 3d XPS spectra, the α-MoO 3 sample mainly contains Mo 6+ , and the trace of Mo 5+ suggests the existence of oxygen vacancy.In comparison, the Mo 5+ /Mo 6+ ratio notably increases in the α-MoO 3 /η-Mo 4 O 11 sample as half Mo are 5+ in η-Mo 4 O 11 , further confirming the formation of heterostructure.

Electrochemical performance
To compare the electrochemical behaviors, the α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films were investigated in Li metal coin cells with excess Li and an electrolyte of lithium perchlorate (LiClO 4 ) in a mixture of ethylene carbonate and dimethyl carbonate.All the cells were charged and discharged between 1.0 and 3.5 V (vs.Li/Li + ) at a specific current of 0.1 A g −1 .As shown in Fig. 4a, the initial charge/ discharge curves of the Li//α-MoO 3 half-cell show one pair of major potential plateaus at about 2.4 V (vs.Li/Li + ) with a large reversible capacity of ~300 mAh g −1 .Notably, the first discharge curve of the α-MoO 3 electrode displays a minor potential plateau at ~2.8 V vs. Li/Li + , corresponding to the small anodic peak in cyclic voltammetry (CV) curve in Supplementary Fig. 9, which can be attributed to the irreversible phase transition to the new phase of Li x MoO 3 with Li + entering the intralayer 10,13,14,28 .With such irreversible phase transition, the α-MoO 3 electrode exhibits fast capacity decay even within five cycles.As shown in Fig. 4b, the charge/discharge curves of the Li//η-Mo 4 O 11 halfcell present two pairs of potential plateaus between 1.5 and 3.5 V vs. Li/ Li + (corresponding to the two pairs of redox peaks in CV curves in Supplementary Fig. 9), delivering a large reversible capacity of about 350 mAh g −1 .With α-MoO 3 as the major phase in the heterostructure, Capacity (mAh g −1 ) Capacity (mAh g −1

)
Cycle number Cycle number 5.0 A g -1 Voltage (V vs. the charge/discharge curves of the α-MoO 3 /η-Mo 4 O 11 electrode display similar potential plateaus as the α-MoO 3 electrode but with a more sloping profile (Fig. 4c).A high reversible capacity of ~350 mAh g −1 was achieved and well maintained by the α-MoO 3 /η-Mo 4 O 11 electrode for the initial cycles.Surprisingly, the first discharge curve of the α-MoO 3 / η-Mo 4 O 11 electrode maintains a steep slope at ~2.8 V vs. Li/Li + without the appearance of potential plateau, suggesting successful suppression of the irreversible phase transition in the layered α-MoO 3 within heterostructure during lithiation.To further identify the phase transition during the lithiation/delithiation of these Mo-based electrodes, we carried out differential capacity analysis (dQ/dV) by numerical differentiation of data in Fig. 4a-c 29,30 .As shown in Supplementary Fig. 10, the features at ~2.8 V vs. Li/Li + in dQ/dV plot of α-MoO 3 are absent in its 5th cycle, suggesting the irreversible phase transition of the α-MoO 3 electrode.In sharp contrast, no irreversible phase transition is found in dQ/dV plot of α-MoO 3 /η-Mo 4 O 11 electrode, which maintains the pristine redox peaks well in its 5th cycle.Typical charge/discharge curves of the α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 electrodes at specific current from 0.1 A g −1 to 5 A g −1 are shown in Supplementary Fig. 11, and the corresponding rate capabilities of the three electrodes are compared in Fig. 4d.Layered α-MoO 3 , possessing a large interlayer distance, could favor fast Li + diffusion through the two-dimensional diffusion channels.The rate performance of α-MoO 3 , however, is compromised by its irreversible phase transition and fast structural degradation, which slow down its electrode kinetics.When α-MoO 3 is stabilized in the heterostructure, the α-MoO 3 /η-Mo 4 O 11 electrode exhibits both large specific capacity and high rate performance.At a high specific current of 5 A g −1 , the corresponding capacity retention of the α-MoO 3 /η-Mo 4 O 11 electrode (with reference to the capacity at 0.1 A g −1 ) is ~41%, which is larger than ~31% of α-MoO 3 and ~16% of η-Mo 4 O 11 .
The electrochemical impedance spectroscopy (EIS) spectra of the three electrodes were measured at open-circuit potentials and are shown in Supplementary Fig. 12.It can be observed that the charge transfer resistance of the α-MoO 3 /η-Mo 4 O 11 electrode calculated by fitted EIS is the smallest among the three electrodes, as shown in Supplementary Fig. 12 and Supplementary Table 3, further demonstrating the improved charge transfer efficiency for the heterostructure.Meanwhile, the chemical diffusion coefficient of lithium (D Li + ) of α-MoO 3 , η-Mo 4 O 11 , and α-MoO 3 /η-Mo 4 O 11 electrodes were calculated (Supplementary Fig. 13) by conducting the galvanostatic intermittent titration (GITT) technique during the lithiation process.The D Li + values in α-MoO 3 electrode are substantially higher than those in η-Mo 4 O 11 electrode owing to the large ionic diffusion channels.However, the irreversible phase transition results in the dramatic decrease of the diffusion coefficient of Li + in α-MoO 3 electrode.Compared with both α-MoO 3 and η-Mo 4 O 11 electrodes, α-MoO 3 /η-Mo 4 O 11 electrode possesses fast and stable Li + diffusion, which could be assigned to the solid-state interfacial chemistry and improved structural stability.
Based on the ultraviolet photoelectron spectroscopy (UPS) and ultraviolet-visible absorption spectroscopy measurements (Supplementary Fig. 14a), the schematic energy-level diagrams of α-MoO heterojunction.Given that η-Mo 4 O 11 domains are randomly embedded within the α-MoO 3 matrix, Li + ions need to be transferred from α-MoO 3 to η-Mo 4 O 11 during the discharge process.During the discharge process, the as-formed built-in electric field can accelerate Li + diffusion from α-MoO 3 to η-Mo 4 O 11 , thus improving the electrode kinetics for the discharge process.Under the built-in electric field, Li + could accumulate at η-Mo 4 O 11 side to neutralize the negative charges, and the electric field around the hetero-interface may finally vanish after the charge balance 32 .Upon the charge process, Li + ions are first extracted from α-MoO 3 owing to its direct contact with electrolyte.As α-MoO 3 possesses a more open structure for Li + diffusion, the fast leaching of Li + in α-MoO 3 will lead to a Li + concentration gradient across the α-MoO 3 /η-Mo 4 O 11 interface, thus generating a reversed built-in electric field during the charge process.Under such an electric field, the transfer of Li + from η-Mo 4 O 11 to α-MoO 3 can be accelerated, improving the electrode kinetics for charge process 33 .Overall, the constructed α-MoO 3 /η-Mo 4 O 11 hetero-interface is beneficial to improve Li + intercalation and deintercalation in this heterostructured electrode during both charge and discharge processes.
The structural stabilities of different electrodes were further investigated by cycling tests.Figure 4e compares the cycle performances of the three electrodes between 1.0 and 3.5 V (vs.Li/Li + ) at a specific current of 0.5 A g −1 .It is clear that the α-MoO 3 electrode exhibits fast capacity fading with only approximately 27% capacity retained after 300 cycles, suggesting continuous structural degradation during cycling.In comparison, the η-Mo 4 O 11 electrode presents superior structural stability with about 82% capacity retained after 300 cycles, demonstrating that it can be used as good structural stabilizer for the heterostructure design.Nevertheless, the specific capacity of the η-Mo 4 O 11 electrode at 0.5 A g −1 is the lowest among the three electrodes due to its poor rate capability.Remarkably, the α-MoO 3 /η-Mo 4 O 11 electrode displays both the highest specific capacity and the best cycle performance with a negligible capacity loss after 300 cycles.The superior structural stability of the heterostructure can be further demonstrated by a long-term cycling test.When cycled at a low specific current of 0.5 A g −1 , the α-MoO 3 /η-Mo 4 O 11 electrode can retain over 90% of the initial capacity after 1000 cycles.In addition, as depicted in Supplementary Fig. 15a, b, the redox peaks of α-MoO 3 /η-Mo 4 O 11 electrode are well retained after 1000 cycles and the corresponding dQ/dV plots at the 1st and 1000th cycles are nearly overlapped with minimal capacity loss, further demonstrating the high stability of the heterostructure with coherent interface.In sharp contrast, the α-MoO 3 displays an irreversible capacity loss of ~73% after 300 cycles at a specific current of 0.5 A g -1 .Moreover, the redox peaks in dQ/dV plot of α-MoO 3 are remarkably reduced and new peaks emerge even after 300 cycles, suggesting severe structural change.
As the α-MoO 3 /η-Mo 4 O 11 electrode was subjected to prolonged cycling at 2 A g −1 (Fig. 4f), the initial reversible capacity is ~170 mAh g −1 and about 81% of the initial capacity can be retained after 3000 cycles, substantially outperforming previously reported α-MoO 3 -based positive electrodes (Supplementary Table 4), including H 2 O molecule modified α-MoO 3 , oxygen-deficient α-MoO 3−x , α-MoO 3−x /carbon nanotube, and pre-intercalated α-MoO 3 (Supplementary Fig. 16) 10,13,15,23,34,35 .Importantly, the common phenomenon of fast capacity decay during initial cycles for the α-MoO 3 -based electrode as reported in literature is not observed in the present study for the α-MoO 3 /η-Mo 4 O 11 heterostructured electrode, suggesting greatly enhanced structural stability of α-MoO 3 within the heterostructure.Supplementary Fig. 17a-d presents the FESEM images of the α-MoO 3 and α-MoO 3 /η-Mo 4 O 11 thin films collected after different cycle numbers at 0.5 A g −1 , respectively.It is noted that the morphology of the α-MoO 3 /η-Mo 4 O 11 thin film was well maintained after 1000 cycles; however, severe morphology changes along with nanoparticles formation and aggregation, and cavities formation were observed in the α-MoO 3 thin film after only 5 cycles (Supplementary Fig. 18).The photo images of different electrodes after 300 cycles in Supplementary Fig. 19 visually confirm the excellent stability of the α-MoO 3 /η-Mo 4 O 11 thin film during the cycling process.In strong contrast, drastic exfoliation of the active material from the α-MoO 3 thin film electrode was detected after 300 cycles, indicating large volume change and electrode pulverization occurred during the cycling.

Lattice pinning via coherent interface
To grasp insight into the mechanism of the excellent structural stability of the α-MoO 3 /η-Mo 4 O 11 heterostructure, we have conducted ex situ XRD measurements on different electrodes at various charge and discharge states to elucidate the potential-dependent structure evolution and lattice variation (Fig. 5a, b).For α-MoO 3 , the gradual disappearance of the (020) peak belonging to the pristine α-MoO 3 along with the emerging (030) peak of the lithiated α-MoO 3 (Li x MoO 3 , x~0.25), at ~2.8 V vs. Li/Li + , confirms the irreversible phase transformation during the initial discharge process 10,13,14 .Simultaneously, the intensive (110) and (040) peaks ascribed to α-MoO 3 vanish and a new broad peak appears in the following lithiation process, indicating the original stacking order of Mo-O octahedron layers becomes strongly disordered 28 .In contrast to α-MoO 3 , all of the characteristic peaks for the α-MoO 3 /η-Mo 4 O 11 heterostructure are well retained with only reversible peak shift during charge/discharge, demonstrating α-MoO 3 layered structure can be stabilized without phase transition in the heterostructure.Meanwhile, the corresponding interlayered spacing of (020) planes of α-MoO 3 in heterostructure increases from 6.80 Å at 3.5 V to 6.96 Å at 1.0 V (expansion rate ~2%), whereas the initial interlayer spacing of the pure α-MoO 3 exhibits a considerably increased expansion rate of ~16% (from 6.80 to 7.90 Å), demonstrating the coherent interface can effectively pin lattice expansion of α-MoO 3 in heterostructure.Unlike α-MoO 3 , the monoclinic η-Mo 4 O 11 presents minimal and reversible lattice change during both lithiation and delithiation processes (Supplementary Fig. 20), suggesting that it could be the perfect structural stabilizer to constrain lattice variation of α-MoO 3 .It should be noted that the coherent interface not only suppresses the lattice variation of α-MoO 3 but also further reduces the lattice variation of η-Mo 4 O 11 in the heterostructure.As compared between Fig. 5b and Supplementary Fig. 20, the diffraction peak shift of η-Mo 4 O 11 in heterostructure during charge and discharge is even smaller than that of the pure η-Mo 4 O 11 .
The ex situ XPS measurements were performed to investigate the electrochemical reversibility of the α-MoO 3 /η-Mo 4 O 11 heterostructure (Fig. 5c).When fully delithiated at 3.5 V, the α-MoO 3 /η-Mo 4 O 11 heterostructure contains only Mo 6+ and Mo 5+ , whereas the pure α-MoO 3 presents ~18% Mo 4+ in addition to Mo 6+ and Mo 5+ , which suggests that more inactive Li + are trapped in the pure α-MoO 3 associated with the initial irreversible phase transition.Another interesting feature is that the fraction of Mo 4+ at the full lithiation state in α-MoO 3 /η-Mo 4 O 11 (~52%) is obviously higher than that in α-MoO 3 (~41%; Supplementary Fig. 21), suggesting that the enhancement in specific capacity of the heterostructure originates from the enhanced redox activity of Mo 6+ to Mo 4+ .The atomic structure of the cycled α-MoO 3 /η-Mo 4 O 11 sample, which underwent 100 cycles at a specific current of 0.5 A g −1 , was further analyzed by HADDF-STEM.As seen in Supplementary Fig. 22,  2 Theta (degree) is well constrained by the coherent interface due to the strong ionic bonds between α-MoO 3 and η-Mo 4 O 11 .With the strong lattice pinning effect, the layered structure of α-MoO 3 is perfectly stabilized in the heterostructure, allowing efficient Li + intercalation and deintercalation with minimal lattice variation.An increased interlayer spacing of 7.9 Å along the b axis in f-Li 0.25 MoO 3 was observed as compared to 6.8 Å in α-MoO 3 , which is in good agreement with the experimental result.Simultaneously, the distorted Mo 6+ ions in MoO 6 octahedra of α-MoO 3 was found to move to the octahedral center after lithiation (Fig. 6a-c).It is well known that the off-center displacement of d 0 Mo in MoO 6 octahedra of α-MoO 3 (Mo 6+ ) was ascribed to the secondorder Jahn-Teller (SOJT) effect through the interaction between the empty d-orbitals (d 0 ) of Mo cations and filled p-orbitals of oxygen ions [36][37][38][39] .The formation of Mo 5+ /Mo 4+ associated with Li + intercalation into the layered α-MoO 3 , however, breaks this distortion state, resulting in the migration of Mo ions back to the center of MoO 6 octahedra 40 .Therefore, lithiation in pristine α-MoO 3 induces large lattice expansion together with Mo migration, which could promote the irreversible phase transition with poor structural stability (Supplementary Fig. 24).Interestingly, the Mo ions in MoO 6 octahedra of c-Li 0.25 MoO 3 retain the same position as the pristine α-MoO 3 (Fig. 6d), suggesting the migration of Mo ions is driven by lattice expansion during lithiation.By suppressing the interlayer expansion along the b direction, the migration barrier of Mo ions is greatly enhanced during the discharge process, which effectively inhibits the displacement of Mo ions to the center of MoO 6 octahedra.In addition, partial density of states (PDOS) analysis indicates enhanced energy for the occupied Mo/O states in f-Li 0.25 MoO 3 as compared to that in c-Li 0.25 MoO 3 , demonstrating weakened Mo-O interaction stemming from increased interlayered spacing in f-Li 0.25 MoO 3 (Supplementary Fig. 25).As indicated in Supplementary Fig. 26, there are three types of oxygen atoms, i.e., O t , O a and O s , in the pristine α-MoO 3 .It can be seen that the Mo-O t bonds in f-Li 0.25 MoO 3 elongate from 2.372 to 2.461 Å, while the length of Mo-O t bond in c-Li 0.25 MoO 3 only changes from 2.372 to 2.371 Å (Supplementary Table 5) after lithiation.The notably increased Mo-O t bonds greatly increase the probability for Mo migration in α-MoO 3 , thus enabling the irreversible phase transition during the initial lithiation.Based on the DFT calculations, it is speculated that lattice pinning via the coherent interface can significantly increase the structural stability of α-MoO 3 in the heterostructure, successfully inhibiting the adverse phase transition (Fig. 6e).

All-solid-state thin-film microbattery
Owing to the solid−solid electrode /electrolyte interface in all-solidstate lithium batteries, the positive electrode materials with large volume change during charge/discharge are detrimental to interface stability, which could result in fast capacity fading and even battery failure 41 .Herein, the heterostructured α-MoO 3 /η-Mo 4 O 11 electrode with minimal lattice variation during cycling is anticipated to find desirable application in solid-state lithium batteries 42,43 .To identify the compatibility of the α-MoO 3 /η-Mo 4 O 11 electrode with solid electrolyte, we fabricated an all-solid-state thin-film lithium microbattery using α-MoO 3 /η-Mo 4 O 11 as the positive electrode, LiPON as the solid electrolyte, and metal lithium as the negative electrode (Fig. 7a and Supplementary Fig. 27).The suitable ionic conductivity (3 × 10 −6 S cm −1 ; 298 ± 2 K) and low electric conductivity (1 × 10 −10 S cm −1 ; 298 ± 2 K) of LiPON with high stability make it an appropriate solid electrolyte for solid-state thin-film lithium batteries 19,35,44 .The LiPON was sputtered using a Li 3 PO 4 target under N 2 atmosphere at 298 K.During the RF magnetron sputtering process, the introduced N 2 gas was split into nitrogen ions and electrons under the high electric field, resulting in a reactive nitrogen cation species.The ionized N species can react with the Li-P-O species to form LiPON 45,46 .As shown in the cross-section FESEM image of the solid-state battery (Fig. 7b), the LiPON affords close contact with both the α-MoO 3 /η-Mo 4 O 11 electrode and lithium metal electrode, guaranteeing efficient charge transfer at the interface with low interfacial impedance.In addition, the STEM image and corresponding energy-dispersive X-ray spectroscopy (EDS) elemental mappings are shown in Supplementary Fig. 28 and Fig. 7c, respectively.Well-separated and uniform distributions of Mo and P can be clearly observed in the sample, further demonstrating the intimate interface contact.
The charge/discharge curves of the solid-state thin-film lithium microbattery based on the α-MoO 3 /η-Mo 4 O 11 positive electrode at different specific currents are shown in Supplementary Fig. 29.At a low specific current of 0.05 A g −1 , the microbattery delivers a large capacity of ~300 mAh g −1 (~0.075 mAh cm −2 ) at the first cycle (Fig. 7d).Even at a high specific current of 2 A g −1 , it can still exhibit ~90 mAh g −1 , suggesting good rate performance.The specific capacity and rate performance of the α-MoO 3 /η-Mo 4 O 11 electrode in the solid-state thinfilm battery are relatively lower than those of the α-MoO 3 /η-Mo 4 O 11 electrode tested in organic electrolyte, suggesting limited charge transport kinetics in solid-state battery configuration.Nevertheless, the as-fabricated solid-state microbattery device is highly flexible and robust (Fig. 7e).The cycling test of the device showed negligible capacity change under various bending angles from 0 to 180°.Even after cycling at different bending states, the interface between α-MoO 3 /η-Mo 4 O 11 electrode and LiPON electrolyte maintains the intimate contact, suggesting the good flexibility of the solid-state microbattery and potential application as power sources for various flexible electronics (Supplementary Figs. 30 and 31).Importantly, Fig. 7f shows the long-term cycling performance of the solid-state microbattery, presenting ~74% capacity retention even after 4000 cycles at 2 A g −1 .The long-term cycling performance of the α-MoO 3 /η-Mo 4 O 11 electrode within a solid-state microbattery is well-placed among previously reported solid-state microbatteries (Supplementary Fig. 32) 19,35,45 , demonstrating the substantially enhanced battery cycle life in solidstate configuration when the α-MoO 3 /η-Mo 4 O 11 heterostructure is employed.
Based on the Galvanostatic charge/discharge profile in Supplementary Fig. 29, the α-MoO 3 /η-Mo 4 O 11 -based thin film micorbattery delivers a specific energy density of ~0.48 mWh with an areal energy density of 0.16 mWh cm -2 .Moreover, the α-MoO 3 /η-Mo 4 O 11 based solid-state microbattery can achieve superior electrochemical performance in comparison with the LiCoO 2 and LiMn 2 O 4 -based solid-state microbatteries, especially in specific capacity, max energy density, and cycle life 19,44 .Importantly, for the α-MoO 3 /η-Mo 4 O 11 and α-MoO 3 thin film deposition, a cheap Mo metal target can be directly used in sputtering, making the fabrication cost much lower than LiCoO 2 and LiMn 2 O 4 thin films.In addition, in contrast to traditional Li-contained positive electrode thin films (LiCoO 2 and LiMn 2 O 4 ) that necessitate a high annealing temperature around 973 K, the α-MoO 3 /η-Mo 4 O 11 as well as α-MoO 3 thin films can be well crystallized at a much lower annealing temperature of 673 K, making battery on-chip integration compatible without damaging the chip.

Discussion
In summary, we demonstrate a lattice pinning strategy via constructing coherent interface to stabilize α-MoO 3 in the intergrown α-MoO 3 / η-Mo 4 O 11 heterostructure.This strategy boosts Li + storage with substantially prolonged cycle life and improved rate capability.With the rational heterostructure design, the epitaxial lattice of α-MoO 3 is effectively constrained by η-Mo 4 O 11 at the coherent interface via the strong ionic bonds, which successfully suppresses lattice expansion along b direction and eliminates the adverse phase transition of α-MoO 3 during the lithiation process.The DFT calculations indicate that lithiation-induced interlayer expansion in α-MoO 3 weakens the Mo-O bonds and allows Mo migration in MoO 6 octahedra, which could drive the irreversible phase transition to form the unstable Li x MoO 3 phase.With minimal lattice variation during lithiation/delithiation, the α-MoO 3 /η-Mo 4 O 11 heterostructure affords substantially extended cycle life without the fast capacity loss within initial cycles as compared to individual α-MoO 3 .The high structural stability associated with minimal lattice variation of the α-MoO 3 /η-Mo 4 O 11 heterostructure is further demonstrated in all-solid-state thin-film lithium microbattery, which presents a long-term cycling stability with about 74% capacity retention after 4000 cycles at 2 A g −1 .This study demonstrates effective lattice pinning in layered α-MoO 3 via coherent interface design, and provides new insights to design highly stabilized intercalation hosts for sustainable batteries.

Materials synthesis
The Mo target is industrially available material, which was purchased from Wuxi Kaistar Electro-optic Material Co., LTD.The α-MoO

Materials characterizations
The crystallographic information of the samples was characterized by XRD (Bruker) with Cu Kα radiation, Raman spectroscopy (Jobin-Yvon T6400 Micro-Raman system), XPS (Phi Quantera SXM spectrometer equipped with an Al Kα X-ray excitation source).For the Rietveld refinement process with the materials analysis using diffraction (MAUD) software, data is first refined according to the background and instrument parameters.Then, we used the structural model of α-MoO 3 (a = 3.95 Å, b = 13.83Å, c = 3.69 Å, Z = 4, space group Pbnm, ICSD no.36167) as the starting model in the refinement.Moreover, "excess" peaks that cannot be indexed to α-MoO 3 (ICSD no.36167) were further refined using the structural model of η-Mo 4 O 11 (a = 24.54Å, b = 5.44 Å, c = 6.70 Å, Z = 4, space group P2 1 /a, ICSD no.15814), generating a lowreliability factor R wp .Specifically, the quantitative value (64/36) of weight fraction between α-MoO 3 and η-Mo 4 O 11 is extracted from the MAUD based on the following equation (Eq.( 1)) 47,48 : where V, M, Z, and S are the unit-cell volume of phase i, the molecular weight, the number of molecules per unit cell, and the scale factor in the α-MoO 3 /η-Mo 4 O 11 sample, respectively.Considering the thin film texture effect, Rietveld refinement on thin film XRD patterns was carried out by using the harmonic texture model implemented into the MAUD software [49][50][51] .For the electronic structure, we used UPS (Thermo ESCALAB 250XI), which provides valence band and work function position.The morphologies and microstructures of the samples were investigated by FESEM (FEI Quanta 250 F) and HRTEM (FEI Tecnai G2).
A JEM-ARM200F STEM fitted with a double-aberration corrector for both probe-forming and imaging lenses was utilized to perform the HAADF/ABF imaging and EELS; the electron microscope was operated at 200 kV.The attainable resolution of the probe defined by the objective pre-field was 78 pm.

Electrochemical measurements
Li coin cells using the α-MoO .The porosity and average pore size of the separator are 41%, and 0.043 μm, respectively.Moreover, the lithium foil has a diameter of 14.5 mm and a thickness of 500 μm.The D Li + is calculated by GITT according to the following equation (Eq.( 2)): where n m represents the mole value, τ is relaxation time, and S represents the area of the electrode, V m is the mole volume, ΔEt represents the changed voltage during the discharge, and ΔEs is the voltage response under the pulse current.The all-solid-state thin-film battery was fabricated through a layerby-layer deposition procedure.Stainless steel foil with a Ti/Pt transition layer was used as the substrate, and the α-MoO 3 /η-Mo 4 O 11 thin film positive electrode was first deposited by DC sputtering.The LiPON solid electrolyte film and lithium metal film were then sequentially deposited onto the α-MoO 3 /η-Mo 4 O 11 thin film (Supplementary Fig. 27).Typically, the LiPON electrolyte film with a thickness of 2.5 μm was deposited onto the positive electrode by RF reactive magnetron sputtering (SKY Technology Development Co., Ltd, China) using a Li 3 PO 4 target (75 mm in diameter and 5 mm in thickness) under N 2 atmosphere at 298 K for 15 h.Then, the lithium metal layer of approximately 2.5 μm was deposited on the LiPON layer by thermal evaporation (SKY Technology Development Co., Ltd, China).It should be noted that the area of LiPON is larger than those of the positive electrode and negative electrode to avoid a short circuit at the edge.After the deposition of 500 nm Cu current collector by RF reactive magnetron sputtering (SKY Technology Development Co., Ltd, China) using a Cu target, the all-solid-state thin-film microbattery was finally sealed with aluminum-plastic film.Galvanostatic charge-discharge measurements were operated in the voltage range between 1.0 and 3.5 V (vs.Li/Li + ) at different specific currents using a Neware BTS4000 battery test system.CV measurements were conducted on an electrochemical workstation (Biologic VSP).EIS measurements were carried out by applying a potentiostatic amplitude of 10 mV over the frequency range of 100.0 kHz to 0.1 Hz on the Biologic VSP.Six data points are collected per decade of frequency.The open-circuit voltage time applied before carrying out the EIS measurement is ~10 h.Zview software is used to fit the EIS and calculate the errors between the raw and fitted data.The electrochemical energy storage tests were carried out at 298 ± 2 K.

DFT calculation
First-principles calculations based on DFT were performed to understand the effect of Li-ion intercalation on the crystalline structure of α-MoO 3 using the Vienna Ab initio Simulation Package (VASP).The projector-augmented wave (PAW) method was used to describe the ionelectron interactions.Perdew-Burke-Ernzerhof (PBE) version of generalized gradient approximation (GGA) was adopted for the exchangecorrelation energy.A kinetic energy cutoff of 520 eV was used for the plane wave expansion of the valence electron wave functions.A dense Γcentered Monkhorst-Pack k-point mesh with a sampling density of 0.04 Å −1 , 10 −6 eV/cell in total energy, and 10 −2 eV/Å in force were adopted for the convergence criterion during structural optimization.Because of the layer structure of MoO 3 , van der Waals density functional (vdWDF) of optB86b-vdW functional was performed during structural optimization.Differential charge density was calculated by the difference between the total valence charge density of the structure and the superposition of the valence charge densities of neutral atoms.

Fig. 1 |
Fig. 1 | Coherent interface design to suppress lattice change of layered MoO 3 during lithiation.Schematic illustrations of a large lattice expansion of the individual α-MoO 3 , b minimal lattice expansion of the individual η-Mo 4 O 11 , and c suppressed lattice expansion of α-MoO 3 via coherent interface in the α-MoO 3 /η-Mo 4 O 11 heterostructure during lithiation.
, when the O 2 /Ar flow ratio was larger than 20%, Mo atoms can be completely oxidized to form α-MoO 3 .With the decrease of O 2 /Ar flow ratio to less than 20%, low valence molybdenum oxides such as η-Mo 4 O 11 were gradually generated together with α-MoO 3 , resulting in the formation of α-MoO 3 /η-Mo 4 O 11 heterojunction.Such heterojunction design can be easily achieved by controlling the O 2 /Ar flow ratio (10-20%) and substrate temperature (573-673 K).The standard deviation obtained during refinement for α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 is shown in Supplementary Table 2
, the increment of oxygen partial pressure not only induces the phase transition from η-Mo 4 O 11 to α-MoO 3 but also increases the surface roughness of the film.At a low O 2 /Ar flow ratio, the film exhibits a highly dense and smooth surface morphology, whereas at a high O 2 /Ar flow ratio, the film is composed of nanoflakes, resulting in a rough surface morphology.Therefore, increasing the O 2 / Ar flow ratio can enlarge particle size, change phase composition, and enhance the valance state of Mo ions of the films.The cross-section FESEM images of α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films are presented in Fig. 2e-g, which reveal that the corresponding thicknesses of different films are about 1.1, 1.0, and 1.3 µm, respectively.To investigate the heterostructure and interface between α-MoO 3 and η-Mo 4 O 11 , high-resolution transmission electron microscopy (HRTEM) and advanced spherical aberration-corrected scanning transmission electron microscopy (STEM) were employed for structural analysis.Ideal structure models of α-MoO 3 and η-Mo 4 O 11 superimposed on the high-angle annular dark field (HAADF)-STEM images were plotted to better illustrate the intergrowth of α-MoO 3 and η-Mo 4 O 11 phases and the formation of heterojunction.The atomicresolution HAADF-STEM images of α-MoO 3 (Fig. 3a) and η-Mo 4 O 11 (Fig. 3b) samples reveal lattice spacing of ~0.69 nm and 0.28 nm, respectively, which correspond to (020) plane of layered α-MoO 3 and (020) plane of η-Mo 4 O 11 .Noted that the distances between Mo atoms along [010] direction are not all the same due to the bilayer structure of α-MoO 3 .The corresponding lattice spacing of (020) plane of α-MoO 3 thus can be classified as the intralayer and interlayer spacing.The structures shown in Fig. 3a, b agree well with Mo atomic arrangements of the orthorhombic MoO 3 along [101] zone axis and monoclinic Mo 4 O 11 along [100] zone axis (Fig.

Fig. 3 |
Fig. 3 | Interface for the α-MoO 3 /η-Mo 4 O 11 heterostructure.a HAADF-STEM image for layered α-MoO 3 along [101] zone axis.The purple spheres represent Mo atoms and the yellow spheres represent oxygen atoms.b HAADF-STEM image for monoclinic η-Mo 4 O 11 along [100] zone axis.c Crystal structures of orthorhombic MoO 3 along [101] zone axis and monoclinic Mo 4 O 11 along [100] zone axis.d HAADF-STEM image for the α-MoO 3 /η-Mo 4 O 11 heterostructure.The blue square corresponds to the monoclinic Mo 4 O 11 viewed along the [100] zone axis while the red square corresponds to the orthorhombic MoO 3 viewed along the [101] zone axis.e The O K-edge ELNES spectra of the α-MoO 3 /η-Mo 4 O 11 heterostructure obtained along the yellow arrow in the left STEM image.
3 , η-Mo 4 O 11 , and heterostructured α-MoO 3 /η-Mo 4 O 11 were proposed as shown in Supplementary Fig. 14b.The Fermi levels of both α-MoO 3 and η-Mo 4 O 11 are all located near their Conduction Bands, suggesting their n-type semiconductor feature.As α-MoO 3 exhibits smaller work function (4.28 eV) as compared to that of η-Mo 4 O 11 (4.58 eV), electrons tend to transfer from α-MoO 3 to η-Mo 4 O 11 across the interface.The electron flow leads to the accumulation of positive charge on the α-MoO 3 side and negative charge on the η-Mo 4 O 11 side near the interface.Simultaneously, the energy levels of α-MoO 3 shift upward whereas those of η-Mo 4 O 11 bend downward near the interface until their Fermi levels reach equilibrium 31 .Thus, a built-in electric field with a direction pointing from α-MoO 3 to η-Mo 4 O 11 is formed at the α-MoO 3 /η-Mo 4 O 11 both α-MoO 3 and η-Mo 4 O 11 phases connected by the coherent interface can be clearly observed.The Mo atomic arrangements in the red square agree well with the orthorhombic α-MoO 3 viewed along [100] zone axis, while the Mo atomic arrangements in the yellow square agree well with η-Mo 4 O 11 viewed along [001] zone axis, demonstrating that the coherent interface between α-MoO 3 and η-Mo 4 O 11 is highly stable during the cycling process.Schematic Fig.5dillustrates that the origin of the irreversible phase transition and structure degradation in α-MoO 3 lie in the large lattice expansion during lithiation process.In sharp contrast, as illustrated in Fig.5e, the epitaxial lattice of α-MoO 3
DFT calculations were performed to gain in-depth insights into the improved structure stability of layered α-MoO 3 induced by the lattice pinning effect.f-Li 0.25 MoO 3 was used as a prototype for the lithiated state to compare with the pristine α-MoO 3 because the irreversible phase transition occurs at this lithiation state of Li x MoO 3 (x = 0.25).A hypothetical c-Li 0.25 MoO 3 structure with fixed interlayered spacing along the b axis as the pristine α-MoO 3 was used to simulate the pinning effect on the suppression of lattice expansion.Supplementary Fig.23a-cshows the relaxed lattice structures of α-MoO 3 , f-Li 0.25 MoO 3 , and c-Li 0.25 MoO 3 , respectively.

Fig. 6 |
Fig. 6 | Insights from theoretical calculations.a, b Differential charge density of MoO 3 and f-Li 0.25 MoO 3 .The blue and red regions represent the depletion and accumulation of charge, respectively.c Layered MoO 3 with Mo off-centering and the displacement of Mo ion being shifted toward the orbital center during lithiation.d Differential charge density of c-Li 0.25 MoO 3 .e α-MoO 3 with constrained lattice eliminates the migration of Mo ions and phase transition during lithiation.
3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films were deposited on Pt/Ti/glass or Pt/Ti/stainless steel substrates using a pure Mo target (75 mm in diameter and 5 mm in thickness) through DC magnetron sputtering (SKY Technology Development Co., Ltd, China) under O 2 and Ar atmosphere.Before deposition, a background pressure less than 1 × 10 −5 Pa was first reached in the chamber.The distance between target and substrate was kept at 10 cm and the substrate rotation is 20 rpm.The DC power was fixed at 60 W. The α-MoO 3 thin film was deposited at 573 K, in a working pressure of ~1 Pa introduced by mass flow controllers with 40.0 sccm Ar and 10.0 sccm O 2 (25% O 2 /Ar ratio).α-MoO 3 /η-Mo 4 O 11 thin film was deposited at 573 K, in a working pressure of ~1 Pa introduced by mass flow controllers with 40.0 sccm Ar and 5.0 sccm O 2 (13% O 2 /Ar ratio).η-Mo 4 O 11 thin film was deposited at 573 K, in a working pressure of ~1 Pa introduced by mass flow controllers with 40.0 sccm Ar and 3.0 sccm O 2 (8% O 2 /Ar ratio).All the films were deposited for 1.5 h and then annealed in the chamber at 673 K for 1 h to obtain the final α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films.The average thickness of α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films are about 1.3, 1.0, and 1.1 µm, respectively.The mass loadings of the α-MoO 3 , α-MoO 3 /η-Mo 4 O 11 , and η-Mo 4 O 11 thin films determined by a Sartorius Analytical Balance (CPA225D, with resolution of 10 µg) were 0.21, 0.25, and 0.30 mg cm −2 , respectively.

Fig. 7 |
Fig. 7 | All-solid-state thin-film lithium microbattery.a Schematic diagram illustrating the configuration of all-solid-state thin-film lithium microbattery.b The cross-section FESEM image of the thin-film microbattery.c The cross-section STEM image and corresponding EDS elemental mappings of the thin-film microbattery.d Rate performance of the thin-film microbattery.e Cycle performance of the flexible thin-film microbattery at different bending states at 0.05 A g −1 .f Long-term cycle performance of thin-film microbattery at 2 A g −1 .The solid electrochemical energy storage tests were carried out at 298 ± 2 K.