Achieving over 90 % initial Coulombic efficiency and highly stable Li storage in 1 SnO 2 by constructing interfacial oxygen redistribution in multilayers 2

: 1 Among the promising high capacity anode materials, tin dioxide (SnO 2 ) represents 2 a classic and important candidate that involves both conversion and alloying reactions 3 toward Li storage. However, the inferior reversibility of conversion reactions usually 4 results in low initial Coulombic efficiency (ICE, ~ 60%), small reversible capacity 5 and poor cycling stability of electrodes. Here, we demonstrate that by carefully 6 designing the interface structure of SnO 2 -Mo, a breakthrough comprehensive 7 performance with ultrahigh average ICE up to 92.6 %, large capacity of 1067 mA h 8 g -1 and 100 % capacity retention after 200 cycles can be realized in a multilayer 9 Mo/SnO 2 /Mo electrode. The amorphous SnO 2 /Mo interfaces, which are induced by 10 redistribution of oxygen atoms between SnO 2 and Mo, can precisely adjust the 11 reversible capacity and cycling stability of the multilayers, while the stable capacities 12 of electrodes are parabolic with the interfacial density. Theoretical calculations and 13 in/ex-situ experimental investigation clearly reveal that oxygen redistribution in the 14 SnO 2 /Mo hetero-interfaces boosts the Li ions transport kinetics by inducing a built-in 15 electric field and improves the reaction reversibility of SnO 2 . This work provides a 16 new understanding of the interface-performance relationship of metal-oxide hybrid 17 electrodes and pivotal guidance for creating high performance Li-ion batteries.


Introduction 1
The increasing demand on higher energy density and safety for Li ion batteries 2 (LIBs) make it extremely important to exploit new anode materials with large capacity, 3 safe operating potentials and fast Li + diffusion kinetics in substituting the 4 conventional graphite anodes. 1-3 The past considerable efforts have verified that many 5 alloying-type metals/semiconductors (Sn, Al, Si, Ge, etc.) and conversion-type metal 6 oxides (Co3O4, Fe2O3, MnO2, etc.) meet well the criteria of higher capacity and safer 7 potentials toward Li storage but still suffer from unsatisfactory reversibility and 8 stability. 4, 5 Among those, tin dioxide (SnO2) anode, storing Li + through a combination 9 of conversion reaction (SnO2 + 4Li + + 4e -→Sn + 2Li2O) and then alloying reaction 10 (Sn + 4.4Li + →Li4.4Sn), with a theoretical capacity of 1494 mA h g -1 and moderate 11 lithiation potential range of 0.4-1.0 V vs. Li/Li + , 6,7 has been regarded as one of the 12 typical and important anode materials for both mechanism exploring and performance 13 tuning toward Li storage, which could ensure larger energy density and a higher level 14 of safety for the cells. 8,9 However, the following mentioned two major drawbacks of 15 metal and metal oxide anodes have seriously impeded the wide utilization of it: (1) the 16 large initial irreversible capacity loss characterized by low initial Coulombic 17 efficiency (ICE) mainly due to inferior reversibility of conversion reaction, 10,11 (2) 18 severe capacity fading and unstable electrode/electrolyte interfaces resulted from the 19 large volume effect of active phases during continuous cycling. 12 20 With respect to the reverse conversion reactions, Sn is oxidized to SnO2 by 21 obtaining O from the decomposed Li2O. As the Li2O and SnO2 have very similar 22 Fig. 2b also reveals that the initial reversible capacities of MSM and MSM-B can 17 still be maintained or even slightly increased after 50 cycles, which could be 18 attributed to the Li storage reactions at the SnO2/Mo interfaces and gradually 19 activation of active materials, as well as the decomposition of solid electrolyte 20 interphase (SEI) at high potential range above 2.0V. In contrast, the capacity of SnO2 21 has drastically decreased. Generally, most of the capacity fading in SnO2 electrode 22 results from the declining reversibility of the conversion reaction (Li2O + Sn → 1 SnO2), which can be reflected by the variation of differential charge capacity plots 2 (DCPs) at different cycles. 14, 18 As shown in Fig. 2c, in the SnO2 electrode，the 3 dealloying peaks of LixSn around 0.5 V gradually shift to higher potentials due to 4 increasing polarization resulted from the coarsening of Sn and LixSn phases. 5 Furthermore, the complete disappearance of DCP peaks within 1.0-2.0 V after 50 6 cycles suggests dramatically declining reversibility of conversion reactions in the 7 SnO2 electrode. In contrast, for the MSM, as shown in Fig. 2d, both the potential 8 positions and integral intensities of these DCP peaks remain stable even after 200 9 cycles, demonstrating the outstanding reversibility and stability of the alloying and 10 conversion reactions. SnO2/Mo interfacial density increased, implying the increased amorphous components 20 and defects such as oxygen vacancies, 48, 49 and another distinguishable diffraction 21 11 peak at 44°could be MoOx. 1 The cross-sectional morphology of different SnO2-Mo multilayers with obvious 2 columnar crystal feature was visually confirmed by scanning electron microscope 3 (SEM) and TEM images in Supplementary Figs. 6-9. Specifically, Fig. 3c shows the 4 cross-sectional structure of the MSM sample prepared by focused ion beam (FIB), 5 which consists of two layers of Mo and one layer of SnO2. The high-resolution TEM 6 (HRTEM) images and selected area electron diffraction (SAED) patterns demonstrate 7 the detailed structural information on each layer and the interface ( Fig. 3d-f). The 8 lattice fringes and diffraction rings reveal that most Mo and SnO2 are polycrystalline. 9 Nevertheless, there are some amorphous components around the columnar Mo grains, 10 as depicted in Fig. 3d. Meanwhile, from the HRTEM and fast Fourier transform (FFT) 11 images for the interface between SnO2 and Mo layers (Fig. 3f), it is revealed that there 12 is an amorphous region along the SnO2/Mo interface with a width of about 14 nm. 13 Moreover, as shown in Supplementary Fig. 9g, amorphous products are more 14 prominent in the MSM-B with greater interfacial density, which is consistent with the 15 results in Fig. 3b. By comparing energy-dispersive X-ray spectroscopy (EDS) 16 mapping and line scan of Sn, Mo, and O elements for the MSM electrode, as shown in 17  Table 2 and 4 Supplementary Fig. 11). As expected, there is no amorphous interface between MoO3 5 and SnO2 layers. And thus, the MoO3/SnO2/MoO3 sandwiched electrode has a lower 6 ICE of 77% and a charge capacity retention of 61.4 % after 50 cycles, which is close 7 to the pure SnO2 and much inferior to those of the MSM electrode. 8 and this effect is strongest near the interface. 54, 55 As clearly revealed by in-situ TEM 10 observation shown in Supplementary Video 1, the process of Li + quickly passing 11 through the outer Mo layer to the SnO2 layer has been found. After full lithiation, the 12 LixSn, Mo and Li2O products from SnOx and MoOx lead to a Li + rich region in the 13 micro-domain of SnO2 layer and a Li + poor region at the interface. Thus, a new E 14 builds with a direction from SnO2 layer to the interface, promoting the following Li + 15 extraction process. These theoretical calculations offer a thorough insight into the 16 interfacial effects of electrode materials for LIBs. Due to the charge transfer driving 17 force originating from the hetero-interface, MSM electrode exhibits high-rate 18 capability and low resistance toward Li + insertion and extraction, as displayed in Fig.  19 2a. 20

In/ex-situ characterizations to reveal the interfacial effect on highly 1 reversible and stable conversion reactions 2
Another important aspect that needs to be understood is how the SnO2/Mo 3 interfaces contributed to the additional reversible capacity and highly reversible and 4 stable conversion reactions in MSM electrode. From the in-situ XRD analysis for the 5 SnO2 electrode, as shown in Fig. 6a, the gradually disappearing of SnO2 and 6 appearing of Sn in conversion reaction and then appearing of LixSn phases during 7 alloying reactions happened along with the discharge from 3.0 V to 0.01 V. However, 8 the clearly observed diffraction peak of Sn as recharged to 3 V indicates poor 9 reversibility of the conversion reaction. For the MSM multilayers in the initial cycle, 10 as shown in Fig. 6b, the Sn and LixSn phases which generated in the conversion and 11 alloying reactions are basically undetectable, suggesting their ultrafine grain size 12 which homogenously dispersed within the Li2O, which leads to the largely reversible 13 conversion between SnO2 and Sn/Li2O in the repeated cycles in MSM. Besides, the 14 diffraction of Mo weakens and shifts to lower 2θ values when discharging, and 15 increases and shifts to the original 2θ values when charging. Obviously, the variation 16 of Mo peak throughout the cycle in MSM is much more obvious than that in the pure 17 Mo ( Supplementary Fig. 13), and agrees with the GI-XRD results as the interfacial 18 density increases (Fig. 3b). Therefore, the variation of Mo peak can be used to track 19 the evolution of the interfaces, and the interfaces should expand during the Li + 20 insertion process and partially recover during the Li + extraction process, indicating the 21 interfacial Li storage characteristic. Further tracking of the Mo peak in the subsequent 22 cycles (Fig. 6c, d) demonstrates that the interface remains stable after the 10th cycle. 1 The variation of the SnO2/Mo interfaces in MSM has also been directly revealed 2 by in-situ TEM observation (Supplementary Video 1). As shown in Fig. 6e, f, and  3 Supplementary Figs. 14 and 15, the interface zone becomes wider and more distinct 4 during the initial 10 cycles and then reaches its steady state. Therefore, the variation 5 of the SnO2/Mo interfaces observed by in-situ XRD and in-situ TEM demonstrates the 6 interfacial Li storage characteristic, which could be the absorption of Li at the 7 interfaces that contributes to the extra storage capacity in the MSM electrodes. 56, 57 8 The contribution of interfacial storage can be further enhanced by increasing the 9 interfacial density in the MSM-A and MSM-B multilayers. However, as the Mo and 10 SnO2 layers become much thinner (such as MSM-C, D) and the interfacial density is 11 too high, the Li adsorption energy should be enhanced due to the ultra-high interfacial 12 effect. 58, 59 It has been previously revealed that the interfacial Li adsorption helps to 13 increase the additional storage capacity in Li2S/graphene composites before the 14 amount of adsorbed Li atoms reaches 2. 58 The high Li adsorption energy should 15 hinder the Li + diffusion process, and the absorbed Li at the interfaces is difficult to 16 detach during the delithiation process, leading to the increased irreversible capacity 17 and decreased ICE (Supplementary Fig. 16). Besides, more oxygen in SnO2 should 18 redistribute to Mo in the multilayers with higher interfacial density, finally forming 19 MoOx and SnOx multilayers rather than original metal/semiconductor heterogeneous 20 interfaces. In this case, the driving force from concentration difference of O for fast 21 ion diffusion and the internal interfacial charge redistribution should be weakened, 22 18 resulting in inferior reversibility and stability of lithiation and delithiation reactions. 1 In contrast, in the SnO2 or MS electrodes without or just with limited SnO2/Mo 2 interface, poor structural stability and instantaneous collapse of active layer during 3 lithiation can be also found by in the in-situ TEM observation (Supplementary Video 4 2). These further consolidate that the Li storage behaviors at the SnO2/Mo interfaces 5 are responsible for the relationship between the capacity increase and interfacial 6 density, as displayed in Fig. 1d. 7 To investigate the interfacial effect on the promotion of highly reversible Li 8 storage, the composition distribution of Li in the depth direction (perpendicularly to 9 the current collector) in the MSM during the 1st cycle was investigated by 10 time-of-flight secondary ion mass spectrometry (TOF-SIMS), 60 as displayed in Fig.  11 6g and Supplementary Fig. 17. The depth profiles of Li as discharged to 0.01 V, 12 charged to 1V and 3V (Fig. 6g), indicate that a large and almost equal amount of Li 13 can be released during the processes of dealloying and reverse conversion reactions, 14 meeting well with the theoretical capacity proportion of these two reactions (771  15 vs.784 mA h g -1 ), demonstrating the fully reversible reactions in the MSM. 16 It is noted that XRD and TEM are hard to clarify the existence of Li2O and the 17 regenerated SnO2 which are usually amorphous. Thus, spectrum methods were 18 conducted to characterize the reversible conversion between Li2O and SnO2. Since 19 Li2O turns into Li2CO3 when exposed to air, the variation of Li2CO3 could evaluate 20 the variation of Li2O in the electrodes. As shown in Fig. 6h, the Fourier transform 21 infrared spectrometer (FTIR) spectra of the MSM electrode clearly reveal the 22 generation and disappearance of Li2O during the discharge/charge reactions. Li2O 1 starts to generate when discharged to 0.6 V and its content reaches the maximum at 2 0.01 V, indicating the completed conversion reaction. On the contrary, during the 3 charging process, Li2O gradually diminishes and then disappears at 2 V, indicating 4 that Li2O is completely reacted with Sn again and converted to SnO2. The regenerated 5 SnO2 is further proven by surface-enhanced Raman spectrum (SERS) collected from 6 the MSM electrode as shown in Supplementary Fig. 18. These fully confirm that the 7 reversibility and stability of the conversion reaction in SnO2 is greatly promoted in the 8 MSM electrode with amorphous interfaces caused by O redistribution, which is 9 responsible for the demonstrated high round-trip efficiency, increased capacity and 10 superior cycling stability in the SnO2-Mo multilayers. 11

Conclusions 12
To summarize, we assembled the SnO2-Mo multilayers with adjustable interfaces 13 and tunable oxygen distribution, and established a model to quantify the regulating 14 effect of SnO2/Mo interfaces on the stability and reversibility of conversion between 15 SnO2 and Sn/Li2O. Benefiting from the multifaceted interfacial effects on the 16 structural stability, Li storage capacity and reaction kinetics, Mo/SnO2/Mo exhibits 17 good cycling stability and accelerated charge-transfer kinetics, realizing a 18 breakthrough performance with high average ICE of 92.6 % and large capacity of 19 1067 mA h g -1 remained 100 % after 200 cycles. Furthermore, we clarified that the 20 redistribution of oxygen between SnO2 and Mo layers helps to form SnO2/Mo 21 20 amorphous interfaces, thereby providing additional reversible capacity and promoting 1 the highly reversible and rapid conversion reaction in lithiated SnO2. This work based 2 on interface engineering and modulation of oxygen distribution provides a novel 3 fundamental strategy to design highly reversible and stable conversion-type electrode 4 materials for large capacity Li storage. 5 4 Experimental 6

Preparation of multilayer thin film electrodes 7
All multilayer thin film electrodes were deposited on Cu foil substrates (6*6 8 cm 2 ) using a KYKY JGB-560 magnetron sputtering system. The targets of SnO2, Mo 9 and MoO3 have a diameter of 60 mm and a thickness of 5 mm. Under optimized 10 conditions, in a typical deposition, a radio frequency magnetron (power 120 W, 2.0 Pa 11 Ar as the working gas) was used during deposition after a base pressure of 1.0 *10 -4 12 Pa was achieved. The various multilayer thin films, namely, SnO2, SnO2-Mo and 13 SnO2-MoO3, were comparatively investigated as anodes for LIBs. 14 To prepare the sample for surface enhanced Raman spectroscopy (SERS) 15 measurement, colloid of Au nanoparticles was deposited on the surface of the SnO2 16 and Mo/SnO2/Mo electrodes. 17

Materials characterization 18
X-ray diffraction (XRD) was performed with a PANalytical X'Pert Pro Alpha-1 19 diffractometer using Cu Ka radiation. The microstructure was characterized using a 20 21 Carl Zeiss Supra 40 field emission scanning electron microscope (SEM) and a 1 high-resolution transmission electron microscope (TEM, JEOL JEM-2100F) 2 operating at 200 kV. The compositions of the layers were determined with an 3 energy-dispersive spectrometer (EDS) attached to the TEM. X-ray photoelectron 4 spectroscopy (XPS) was performed with a PerkinElmer PHI 5000c XPS system using 5 the C 1s peak at 284.8 eV as a reference. TOF-SIMS measurements were conducted 6 on a TOF-SIMS spectrometer (TESCAN GAIA3 model 2016 UHR SEM). The 7 TOF-SIMS measurements were conducted in positive mode. A pulsed 30 keV Ga + ion 8 beam was used in the high current mode for depth profiling. The SERS measurement 9 was conducted with a laser Raman spectrometer (Raman, Horiba) at an excitation 10 wavelength of 532.0 nm. The accurate component analysis was conducted by an 11 Escable 250 X-ray photoelectron spectrometer (Thermo Fisher Scientific, USA). 12 FTIR spectra were obtained by a Nicolet iS50FTIR spectroscopy, equipped with 13 attenuated total reflectance technique. The measurement worked in the range of 14 4000-700 cm -1 . 15 For the ex-situ measurement of electrochemically tested electrodes, the samples 16 were carefully stored and transferred to minimize air exposure. The reacted film 17 electrodes at different states were prepared by discharging/charging to a controlled 18 cutoff voltage. These electrodes were removed from the electrochemical cells, rinsed 19 with DMC and dried under vacuum in the ante-chamber of an argon filled glove box. 20

In-Situ Examinations 1
In-situ XRD analysis was conducted with a PANalytical X'Pert Pro Alpha-1 2 diffractometer using Cu Ka radiation and investigated by a homemade cell that was 3 sealed by Be foil as the X-ray penetrator, accompanied with the lithiation/delithiation 4 process. The 2θ range of each scan started from 25° to 45° with step increment of 0.02° 5 at step size of 0.26°. In-situ TEM measurement was performed on a JEOL 1400 TEM. 6 The MSM electrode was displayed on an Au rod as the working electrode, while a 7 small film of Li metal was scratched on a W wire as the counter electrode. A thin layer 8 of Li2O was expected to form as the solid electrolyte during the charge transfer 9 process. When physical contact between the two electrodes was encountered, a 10 potential bias of ± 3 V versus Li/Li + was applied to drive the diffusion of Li + .