Cyclability evaluation on Si based Negative Electrode in Lithium ion Battery by Graphite Phase Evolution: an operando X-ray diffraction study

Artificial graphite (FSN) additive is employed as internal structural label for projecting cyclability of Si material native electrode in a mass ratio of Si/FSN = 1.0 in Li ion battery (LIB). Results of operando X-ray diffraction analysis on Si-FSN negative electrode in LIB demonstrate that one can evaluate the lithiation and delithiation affinity of active material by referring phase transition delay of graphite as affected by experimental splits in a formation process of LIB. We prove that a thin layer of surface amorphous structure and residual lattice strain are formed in Si by high energy ball-milling treatment. Those manipulations improve Li intercalation kinetics and thus enabling a capacity fading of less than 10% (from 1860 to 1650 mAhg−1) for Si negative electrode in 50 cycles. Of utmost importance, this study discloses a robust assessment for revealing mechanism on amorphous and strain related silicide formation and predicting cyclability of negative electrode by quantitative phase evolution rate of FSN additive in LIB.


Results and Discussion
Physical Structure Characterizations on Si based materials. Crystal structure of Si P and Si H+W is revealed by using XRD analysis. Structure parameters are quantitatively determined by Rietveld refinement on their diffraction patterns (Fig. S1) and corresponding lattice parameters of Si P + FSN and Si H+W + FSN are listed in Table S1. Accordingly, lattice parameter is 5.435(6) Å and unit cell volume (V) is 160.598(9) Å 3 for Si H+W . For Si P , lattice constant is 5.431(3) Å and unit cell volume is 160.218(0) Å 3 . As a result, one can notice a slight distortion and lattice expansion by 0.24% on Si crystal (Si H+W ) by high energy ball-milling treatment. Diffraction patterns of the two electrode are compared in Fig. 1 and intensity is normalized by that of FSN (002) peak. Accordingly, all diffraction peaks of Si H+W + FSN possess lower intensity, broader peak width accompanied with stronger diffusion scattering background (denoted by Q and Q" in inset) as compared to those of Si P + FSN. As consistently proved by high resolution transmission electron microscopy (HRTEM) images in Fig. S2 and XRD analysis (Table 1), those characteristics can be ascribed for formation of amorphous Si species, certain short range disorder structure (sub-nanometer domains), and increased preferential (111) facets in Si H+W by HEMM. Quantitative structural parameters are summarized in Table 1. Accordingly, crystal structure of Si phase remaining unchanged by HEMM treatment. For Si P + FSN, average coherent length (D avg ) is 834.9 Å for (111), 790.5 Å for (220), and 757.1 Å for (311) facets in Si phase. After HEMM treatments, D avg of Si H+W + FSN is respectively reduced by 20% in (111) facet (624.2 Å), ~43% in (220) facet (451.7 Å), and 46% in (311) facet (455.4 Å) as compared to those of Si P + FSN. In a meantime, H (111) /H (220) and H (111) /H (311) is respectively increased to 2.215 and 4.493 revealing its significant preference along Si (111) facet as compared to that of ideal Si crystal and Si P . Changes of those parameters resembles the "damage" of Si phase by high energy treatment and thus increasing the asymmetric crystal ratio in (111) facet.

Interface Properties of Experimental Si based Negative Electrode Materials by DFT calculation and EIS characterization.
Affinity of Li silicide formation is a crucial factor in performance of Si materials as negative electrode in LIB. This factor is a combination results including atomic packing density of facets, size of Si crystal, and anchoring ligands as dock for Li intercalation. Effects of proposed high energy milling treatment on Li silicide formation affinity are evidenced by cross-referencing DFT (with structure models determined by XRD fitting) and XRD determined crystal structure evolution of experimental materials in operando LIB. Figure 2 compares top view of DFT calculated atomic packing structure for models of Li 15 Si 4 clusters in Si Figure 1. XRD patterns of as-prepared Si P + FSN and Si H+W + FSN electrodes. Inset presents room in for the two diffraction patterns in a range from 11.4° to 13.2°. Wavelength of incident X-ray is 0.689 Å (18 keV slabs and corresponding formation energy (E form (eV/atom)) with facets indexed by the first three diffraction peaks in Fig. 1. Accordingly, the two Li 15 220) and (311)), silicide clusters tend to accumulate in slab corners. Given that accumulation of silicide occurs both in slabs with a large (253.7 Å 2 ) and small (207.1 Å 2 ) areas, steric effects should be a minor factor in Li silicide formation as compared to facet selectivity (i.e., selectivity of facet) in crystal surface once improper atomic packing configuration is adopted. With proper atomic packing configuration (i.e., (111) facet), E form is substantially reduced by 0.038 (eV/atom) with slab size by ~70 Å 2 (from 207.1 to 134.5 Å 2 ). Above interpretations could be explained by effects of interface lattice mismatch and atomic arrangements on heterogeneous crystal growth of silicide in a Si crystal surface. For growing polycrystalline thin films on a crystal with completely different structure symmetry, increasing surface atomic packing density reduces local steric barrier for the subsequent position of film atoms. On the other hand, with a large local steric barrier in opened facets (i.e., (220) and (311)) reduction of E form with slab size is insignificant and is because of the higher energy for silicide to packing in surface defects. It is surely that presence of surface defects facilitates Li intercalation to growth Li silicide, however, such assessment and effect might lead to side effects (such as electrolyte decomposition and Si oxidation, etc.) and are not the topic to be discussed in this study. Effects of preferential facet and particle size (predicted by DFT calculation and XRD analysis) on Li intercalation and Li silicide formation are complementary revealed combining electrochemical and operando XRD analysis on LIB with experimental Si materials (Si P + FSN and Si H+W + FSN) as negative electrode. Figure 3 displays the Nyquist plots of electrochemical impedance spectra (EIS) obtained from experimental Si based negative electrode at 0% lithiated state. These cells exhibited a semicircle in high frequency and a straight line in low frequency ranges, which are attributed to kinetics of charge transfer in interface and Li ions diffusion in bulk electrodes, respectively. As can be seen, a lower electrode impedance of Si H+W indicates its smaller interfacial charge-transfer resistance (Rct) as compared to that of Si P . Such a phenomenon reveals the facilitation of Li intercalation in Si H+W surface due to the surface amorphization (probed by HRTEM images), preferential (111) facets, and reduced D avg (at (220) and (311) facets) in Si H+W . In a meantime, a uniform and local disordered amorphous Si thin layer (as consistently proved by HRTEM and XRD analyses) is formed in fresh silicon surfaces by wet milling which provides an easy access for Li intercalation. In addition, physical impacts of high energy mechanical milling and wet milling process reduce the primary and agglomerate size which shorten the Li ion diffusion length in a silicon particle. Presence of FSN provide additional Li storage sites. Those Li storage sites share the redox loading and thus reduce chemical stress in Si surface. As denoted by reducing of semicircle diameters, presence of FSN modulates the redox kinetics and thus reduces impedance of Li intercalation in silicon-based electrode.
Internal standard structure labelling on Li affinity of Si based materials. Synchrotron based operando XRD analysis further confirms the effects of crystal size, facet selectivity on silicide formation affinity, therefore, projects cyclability of Si materials in LIB. XRD patterns of LIBs containing negative electrodes of Si P + FSN and Si H+W + FSN in the first lithiation/delithation cycle are compared in Fig. 4a,e, respectively. As can been seen, the three peaks at 11.3° (P), 11.8° (Q), and 12.6° (R) are diffraction lines from LiC 12 (002), graphite (002) (C (002)), and Si (111) facets. Rest of peaks at 9° and 10.48° are diffraction lines for (211) (peak S) and (220) facets (peak T) of Li 15 Si 4 . For Si P + FSN (Fig. 4a), intensities of diffraction peaks for silicon phase gradually decrease with increasing lithiation ratio to 50%. After that, Si phase is dramatically vanished and Li 15 Si 4 crystal phase by the end of lithiation. In delithiation process, intensity of peak S and T is decreased indicating a gradual extraction of Li-ions from Li 15 Si 4 . As consistent revealed in literature, no silicon peaks are found at the end of delithiation process. This phenomenon can be rationalized by formation of amorphous Si due to a severe Li retention. As compared to those of Si P + FSN, structure evolutions of Si and silicide phases go even faster in Si H+W + FSN (Fig. 4e). For clarifying lithiation affinity of Si P + FSN (Si H+W + FSN), quantitative results for crystal structure evolution of graphite (C), Si, and Li 15 Si 4 phases in the first lithiation and delithiation cycle of LIB are cross-referenced in Fig. 4b-d (Fig. 4f-h). In these figures, unit of x-axis is number of diffraction patterns, that of y-axis is d-spacing. Shown in Fig. 4b and Fig. 4f, insertion of Li ions induces a graphite phase transformations in several stages including stage 4L (whose composition is not well defined), stage 3L (LiC 24 , d (002)G ~3.47 Å), dilute lattice-gas disordered stage 2 (LiC 18 ), stage 2 (LiC 12 , d (002)G ~3.53 Å), and stage 1 (LiC 6 , d (001)G ~3.70 Å) both in Si P + FSN and Si H+W + FSN 16 . In those stages, numerals (1, 2, 3, and 4) refer to number of empty layers between each Li-filled layer, and different stage reflects different d-spacing 16 . By cross-referencing changes of d-spacing between crystal structures of graphite and active material, an interesting correlation to the affinity of Si with adopted treatments are revealed in Fig. 4b-d. Figure 4b shows 2D contour for changes of inter-planar spacing of graphite (002) facet (C (002)) in Si P + FSN. As shown, increasing d-spacing at graphite (002) facet (d (002)G ) from 3.36 Å to 3.52 Å indicates a successive evolution of Li-intercalated LiC x to stage 2 phase with capacity from 0 to 150 mAh g −1 in lithiation process. By increasing capacity to 150 mAh g −1 , d (002)G is linearly expanded to that of Li enriched stage (stage 2, LiC 12 ). Given that mass ratio of FSN is 50 wt% (equivalent to an ideal capacity of ~170 mAh g −1 ) in active material, transformation of graphite to stage 2 (LiC 12 ) suggests that Li ions are mostly intercalate in FSN. In this event, FSN performs a substantial higher affinity for Li intercalation as compared to that of Si P in active material. After capacity higher than 150 mAh g −1 , d (002)G is slightly increased by 0.01 Å (from 3.52 to 3.53 Å) by a subsequent lithiation to 1720 mAh g −1 . For structural interpretation in details, changes of peak intensity for Si phases in Si P + FSN and Si H+W + FSN with lithiation ratio are compared in Figs S3a and S3b. Accordingly, with an absence of LiC 6 phase and vibration of Si (111) peak intensity (Fig. 4c), one can notice that most of Li is intercalated in Si phase with lithiation ratios With a mass ratio of 50 wt% in active materials, this value is decreased by ~18% as compared to theoretical capacity of Si (1850 mAh g −1 ) and is possibly due to a crack of Li x Si phases from Si surface in a Si P + FSN electrode. In a Si H+W + FSN electrode (Fig. 4e), four stages including stage 4L (100-590 mAh g −1 ), 3L (590-1200 mAh g −1 ), stage 2 (1200-1890 mAh g −1 ), and stage 1 (1900-2030 mAh g −1 ) are found in graphite phase evolution by lithiation from 0 to 100% (2030 mAh g −1 ). Among them, stage 1 is fully lithiated graphite phase and can only be formed by a kinetics balance between intercalation and diffusion rates of Li ions in graphite surface in a negative electrode lithiated higher than 94%. It is important to note that active materials possess a higher than 98% of ideal capacity of graphite and Si phases in Si H+W + FSN; where capacity contribution is ~170-180 mAh g −1 for graphite phase and ~1800-1900 mAh g −1 for Si phase. Compared to that of Si P + FSN, graphite phase delay with lithiation ratio indicates that affinity of Li ion to surface modified Si phase is substantial higher than that of graphite phase in Si H+W + FSN. Those scenarios are direct evidences rationalizing the strong lithiation preferential of Si (111), amorphous Si, and defect regions in Si H+W + FSN electrode. Facile delithiation from Si H+W + FSN is consistently revealed by phase transition of graphite. Shown in Fig. 4f, d (002)G hold in stage 1 by delithiation from 0 to 125 mAh g −1 and then move to stage 2 until 235 mAh g −1 . Further delithiation from 235 to 375 mAh g −1 results in a transition from stage 3L to stage 4L in graphite. In this region, this delithiation value is doubled to that can be offered by graphite meaning that Li extraction is mainly from Si phase in a Si H+W + FSN electrode. On the other hand, in delithiation process of a Si P + FSN electrode (Fig. 4b), stage 2 to stage 4L transition is found by delithiation from 0 to 160 mAh g −1 meaning that most of capacity is contributed from graphite phase (i.e., activation energy for Li extraction from graphite is lower than that from Si phase). Preference of lithiation/delithiation remaining hold even at the 50 th cycles and is consistently revealed by comparing changes of d (002)G between Si P + FSN and Si H+W + FSN with respect to lithiation ratios of LIBs (Fig. S3).
Structure evolutions of Si phases provide complimentary information to the preferential lithiation of active materials in negative electrode of LIB. Shown in Fig. 4g, position of Si(111) peak for Si P + FSN remaining unchanged in lithiation process. In this region, intensity of Si(111) peak is vibrating between 200 to 1100 a.u. with increasing capacity to 1490 mAh g −1 and then dramatically decreased to 0 by a subsequent lithiation till 1570 mAh g −1 (Fig. S3a). A dramatic vibration of peak intensity implies a crack of Si powder due to a strong lattice mismatch between Li silicide. This hypothesis is proved by presence of wide range scattering signals (denoted by yellow arrows) and diffraction line of Li 15 Si 4 (211) by increasing capacity higher than 500 mAh g −1 in Fig. 4d. Schematic representation for silicide formation induced interface crack in Si P is shown in Fig. 5. For Si H+W + FSN, Li 15 Si 4 (211) peak intensity is increased from 0 to 40 a.u. by delithiation from 0 to 500 mAh g −1 and then progressively decreased to 0 in a subsequent delithiation till 1200 mAh g −1 (Figs 4h and S3b). As compared to those of in Si P + FSN, substantially weakened intensity with a broad width and delayed response of Li 15 Si 4 (211) peak reveal a suppression of Li silicide. Such a characteristic can be attributed to formation of local disordered Si/SiO x and increased ratio of (111) facets dimension with proper interface to facilitate Li intercalation and formation of amorphous Li silicide in Si H+W + FSN surface ( Fig. 1 and Table 1). Effects of HEMM treatments on facilitating Li accommodation in Si surface remaining hold in long-term cycling test till the 50 th cycle which again consistently proved the facilitations of silicide formation in (111) facets and reduced D avg as predicted by DFT calculation and EIS analysis. Meanwhile, the same scenario on graphite phase evolutions delay proves the substantial improvement of Li intercalation/extraction performances of Si materials even hold after 50 cycles of LIB. Details of graphite evolutions in 50 th cycles are given in ESI (Fig. S2) and latter sessions. Rates of d (002)G to lithiation ratio (Δd (002)G /ΔL) of negative electrodes are compared Fig. S4 and corresponding peak area (which can be serve as a qualitative index for extent of graphite phase transition) are compared in Table S2 for further revealing the graphite evolution as affected by affinity of Si to Li + ions. Shown in Fig. S4a, the four Δd (002)G /ΔL peaks in (1) 3.9%, (2) 6.8%, (3) 8.1%, and (4) 94.8% of lithiation ratios correspond to the maximum rate for graphite phase transition (i.e., stage 4L, stage 3L, stage 2, and stage 1) in Si P + FSN. The first three peaks incur 97.8% of area indicate that most of active sites in graphite phase are lithiated in a lithiation ratio of 8.1% (i.e., ~260 mAh g −1 ) for negative electrode. For the case of Si H+W + FSN, the Δd (002)G /ΔL peaks at lithiation ratios of (1) 4.6%, (2) 28.6%, (3) 30.2%, (4) 36.1, and (5) 51.8% suggest the presence of five transient states in graphite phase of Si H+W + FSN. In this event, as revealed by area of all Δd (002)G /ΔL peaks, most active sites (~96.1%) in graphite phase are lithiated in 36.1% of lithiation (~1240 mAh g −1 ) for Si H+W + FSN. In a subsequent lithiation, a broad peak across 36.1 to 58.1% can be attributed to formation of LiC 6 transient state. As compared to that of Si P + FSN, transition of graphite phases is delayed by 20-25% of lithiation ratio in Si H+W + FSN. Given that phase transition rate of active materials is dominated by their affinity to Li + ion, such a phase delay again consistently revealed the improvement of Li affinity on Si phase in Si H+W + FSN. Taking results of operando XRD analyses together, changes of graphite and Si phases with lithiation ratios in the first lithiation step of Si P + FSN and Si H+W + FSN can be respectively summarized in S4a and S4B. As compared to that of the first cycle, transition of graphite phase is further delayed by ~14 in Si P + FSN and ~10% in Si H+W + FSN. Such a scenario can be attributed to a reduced lithiation/delithiation barrier by LiSi x formation again proving the concept of graphite phase evolution as internal structural label to Li + ion affinity of active materials in LIB.

Crystal structure affinity to formation of Li silicide in long-cycle LIB cells. Quantitative structural
parameters on influences of surface modification to cyclability of silicon are determined by fitting the experimental diffraction patterns by the LAMP program. Results of graphite phase evolution (changes of d (002)G ) with lithiation/delithiation ratios of Si P + FSN in the 1 st and 50 th cycles are shown Fig. S5a. In lithiation process, Li + ions move from Li metal to the negative electrodes. In Si P + FSN, d (002)G is significantly increased to a value of ~3.53 Å by increasing lithiation ratios from 5 to ~15%. Such a value is commonly known as a d-space of fully lithiated graphite (LiC 12 ). Without significant differences of Si phase evolution in LIB, such a result suggests a lower diffusion barrier for Li + ions in solid electrolyte interface (SEI) and graphite surface as compared to that in Si surface. For the case of Si H+W , situation goes to the opposite. As shown in Fig. S5b, d (002)G is increased to that of stage 4L (3.37 Å) by increasing lithiation ratios from 5% to 30%. After that, d (002)G is then dramatically increased to that of stage 3L (3.47 Å) when ratio of lithiation is 40% and then progressively increased to 3.53 Å (stage 2) till 100%. Hereafter, one can notice a delay of d (002)G in Si H+W + FSN as compared to that in Si P + FSN in lithiation/ delithiation processes. Such a delay on d (002)G with lithiation extent coincide to an inverse proportional of formation energy of Li 15 Si 4 in Si surface as consistently proved by DFT calculations. Changes of d (002)G with capacity in lithiation and delithiation processes of Si P + FSN and Si H+W + FSN in the 50 th cycle are respectively shown in Fig. S6c and S6d. Accordingly, trends of d (002)G evolutions in Si P + FSN in the 50 th cycle is differed from that in the 1 st cycle. Shown in Fig. S6c, changes of d (002)G in Si P + FSN is significantly retarded in the 50 th delithiation process as compared to that in the 1 st delithiation process. Such a delayed response on graphite phase and be explained by formation of significant amount of amorphous Si and retained Li silicide (i.e., irreversible capacity) in Si P + FSN in long-cycle test. Meanwhile, as consistently explained by vibration of Si(111) peak intensity, loss of capacity reveals a crack (pulverization) of Si particle in Si P + FSN in long cycle test which increases internal resistance between Si phases and back contact electrode. Changes of d (002) G for Si P + FSN and Si H+W + FSN in delithiation process are compared in Fig. S6d. As depicted, Si P + FSN and Si H+W + FSN perform a similar decay trend on d (002)G . It means that graphite phase possesses a similar energy barrier for Li intercalation in both the two electrodes in the 50 th delithiation process. It is worth to note that, as compared to that of Si H+W + FSN in the 1 st cycle, d (002)G is suspended in stage IV by further increasing delithiation by 8-9%. It rationalizes the Li intercalation in Si phase is further facilitated in the 50 th cycle for Si H+W + FSN in a LIB. Those results prove that formation of surface amorphous Si layer and increase ratio of (111) facet domain size with local distortion and surface modification by HEMM treatments improve the affinity of Li intercalation and extraction in Si surface (Fig. 5). Such a method is easy assessable therefore promising the development of Si based materials in LIB applications.
Cycle performance test further confirms the prediction on cyclability of Si materials by operando XRD analysis. Figure 6 shows the (a) specific capacity and (b) charging capacity retention (CR) of Si P , Si P + FSN, Si H+W and Si H+W + FSN negative electrodes in LIBs till the 50 th lithiation/delithiation cycle. In the cycle test, rate is C/6 and potential range is 2 mV to 1.5 V. Accordingly, Si P possesses a capacity of 3236 mAh g −1 in lithiation process and 2655 mAh g −1 in delithiation process in the 1 st cycle (formation stage). For Si H+W , capacity is 3470 mAh g −1 in lithiation and 2845 mAh g −1 in delithiation processes. It is important to note that both the Si P and Si H+W perform a columbic efficiency of ~82% in formation stage, however, with completely different fading manners of CR fading in a cycle test. For Si P , CR is exponentially decreased to 10% till the 50 th cycle. On the other hand, CR is decreases by ~15% in the first five cycles (region A) and then slightly decreased by ~4% in the subsequent cycles (region B) for Si H+W . As compared to that of Si P , a substantially reduced CR decay in region A implies a reduction of energy barrier for silicide formation therefore improving cycle stability of Si H+W . Addition of conducting graphite (FSN) further enhances the difference of capacity fading mode between Si P and Si H+W . Show in Fig. 6b, fading rate of Si P is substantially reduced by mixing FSN in a weight ratio of FSN/ Si P = 1.0. Although capacity fading is improved by 19.6%, its CR fading remaining reaction control. On the other hand, CR fading of Si H+W shows the opposite way to that of Si P when mixing with FSN in the same weight ratio. In the 2 nd cycle, a slight increase of capacity might be attributed to the excess Li storage in FSN as compared to that of the 1 st cycle. Such a hypothesis is further revealed by results of operando XRD analysis and impedance test which proving the insufficient/uncompleted Li intercalation in FSN in the first lithiation process. Similar to that of Si P , CR fading of Si H+W is improved by 12% by FSN additive with cycle number to 50.

Conclusions
Graphite phase evolution is employed to evaluate cyclability of Si based materials as negative electrode in an operando LIB cell. At the first cycle, delay of d (002)G expansion with increasing capacity reveals a strong preference of Li ion intercalation in silicon phase in negative electrode of Si H+W + FSN comprising modified Si powder in lithiation process. As compared to that of Si H+W + FSN, d (002)G is linearly increased and stabled to that of fully lithiated graphite (3.53 Å) with respect to a capacity of ~170-180 mAh g −1 revealing that Li + ions are mostly intercalated in graphite phase by lithiation to 5% in negative electrode of Si P + FSN comprising pristine Si powder. Those scenarios are further confirmed by cross-referencing results of XRD, HRTEM, and DFT calculation indicating that performance of Si materials are improved by formation of preferential (111) facet accompanied with certain amorphous structure by HEMM treatment. A most important finding is that d (002)G delay remaining hold in an operando LIB cell even in the 50 th lithiation/delithiation cycle. Such a scenario, in a fact of phase evolution dominated by Li interaction, proves that changes of d (002)G with lithiation ratios is a robust qualitative index for predicting the cyclability of active materials in the formation stage of a LIB.

Methods
Experimental details -Sample preparation, LIB assembly, and Physical Structure Characterizations. Commercial silicon powder was purchased from Fuzhou Hokin Chemical Technology, China (Si P , ~10 um). To modify surface conformation of Si P , high energy mechanical milling (HEMM) and wet ball-milling in a planetary miller for 20 hours at room temperature were employed and resulting product is named as Si H+W 15 . Artificial graphite (FSN, Shanshan Technology, China, ~15 um) was employed as an internal structure standard for labelling lithiation and delithiation affinity of active materials in negative electrode of LIB. The Figure 6. Changes of (a) capacity and (b) capacity retention with lithiation/delithiation cycles of LIBs equipped with Si P , Si P + FSN, Si H+W , and Si H+W + FSN negative electrodes. In these tests, rate is C/6 and potential is ranged between 0.02 V and 1.5 V vs. Li/Li + .