Lithiation of Crystalline Silicon As Analyzed by Operando Neutron Reflectivity Beatrix-Kamelia Seidlhofer,*,† Bujar Jerliu,§ Marcus Trapp,† Erwin Hüger,§ Sebastian Risse,† Robert Cubitt,⊥ Harald Schmidt,§,∥ Roland Steitz,† and Matthias Ballauff*,†,‡ †
Soft Matter and Functional Materials, Helmholtz-Zentrum Berlin für Materialien und Energie, Hahn-Meitner-Platz 1, 14109 Berlin, Germany ‡ Institute of Physics, Humboldt-University Berlin, 10099 Berlin, Germany § Institut für Metallurgie, Technische Universität Clausthal, AG Mikrokinetik, Robert-Koch-Str. 42, 38678 Clausthal-Zellerfeld, Germany ⊥ Institute Laue-Langevin, 71 avenue des Martyrs - CS 20156, 38042 Cedex 9 Grenoble, France ∥ Clausthaler Zentrum für Materialtechnik, Leibnizstraße 9, 38678 Clausthal-Zellerfeld, Germany S Supporting Information *
ABSTRACT: We present an operando neutron reflectometry study on the electrochemical incorporation of lithium into crystalline silicon for battery applications. Neutron reflectivity is measured from the ⟨100⟩ surface of a silicon single crystal which is used as a negative electrode in an electrochemical cell. The strong scattering contrast between Si and Li due to the negative scattering length of Li leads to a precise depth profile of Li within the Si anode as a function of time. The operando cell can be used to study the uptake and the release of Li over several cycles. Lithiation starts with the formation of a lithium enrichment zone during the first charge step. The uptake of Li can be divided into a highly lithiated zone at the surface (skin region) (x ∼ 2.5 in LixSi) and a much less lithiated zone deep into the crystal (growth region) (x ∼ 0.1 in LixSi). The total depth of penetration was less than 100 nm in all experiments. The thickness of the highly lithiated zone is the same for the first and second cycle, whereas the thickness of the less lithiated zone is larger for the second lithiation. A surface layer of lithium (x ∼ 1.1) remains in the silicon electrode after delithiation. Moreover, a solid electrolyte interface is formed and dissolved during the entire cycling. The operando analysis presented here demonstrates that neutron reflectivity allows the tracking of the kinetics of lithiation and delithiation of silicon with high spatial and temporal resolution. KEYWORDS: energy storage, silicon anode, lithium-ion batteries, lithiation, operando, time-resolved, neutron reflectivity lithiated graphite (372 mAh/g).3 In addition to this, silicon is an abundant element and can be manufactured industrially with high purity. The major drawback of Si anodes, however, is the enormous volumetric change of up to 310% upon full lithiation.3,4 This marked expansion is usually followed by mechanical fragmentation and poor cycling stability. Nanometric Si wires,5,6 nanotubes,7,8 spheres,9,10 thin films,11,12 and composite materials13 offer a way around this problem.14,15 Si spheres up to a critical diameter of a few hundred nanometers can be employed as Si anodes in Li-ion systems with good retention of capacity.9,16 Until now, a number of systems
T
he lithium-ion battery has become the leading electrochemical storage system in recent years. In commercial Li-ion batteries, storage of electrochemical energy relies on the intercalation of lithium ions in suitable cathode materials such as LiCoO2 or graphitic anodes.1 This mechanism has led to systems with good capacity retention over many cycles and has thus met the demands of typical applications in electronic devices such as smart phones or tablet computers. The maximum capacity stored in such a system, however, is limited by the number of interstitial sites in which Li ions can be inserted. For applications in automotive transportation,2 the capacity of the Li-ion batteries thus obtained does not suffice. In principle, anode materials such as silicon or germanium offer a way around this problem.3 The storage capacity in the highest lithiated phase, Li22Si5, is with a specific capacity of 3579 mAh/g, much higher than that of fully © 2016 American Chemical Society
Received: March 24, 2016 Accepted: July 22, 2016 Published: July 22, 2016 7458
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
www.acsnano.org
Article
ACS Nano
originates from the reaction of the silicon surface with the electrolyte.32 Subsequent work showed the formation and socalled “breathing” of a solid electrolyte interphase grown on an a-Si electrode after lithiation and delithiation.33 A protective layer of aluminum oxide has been used to maintain the integrity of the silicon electrode, and its effectiveness as an artificial solid electrolyte interphase was proven with NR studies.34 Here, we present a quantitative, time-resolved, and quasistatic study of electrochemical alloying of bulk crystalline silicon with lithium under operando conditions on the relevant nanometer scale by neutron reflectometry. We aim to present a precise operando analysis of all intermediate stages during lithiation and delithiation; therefore, we chose a very low current density of 2 μA/cm2. The neutron reflectivity curves are the corresponding Fourier images of the scattering length density profiles of the sample under investigation. Thus, the operando NR measurements of the bulk c-Si electrode during cycling lead to time-resolved information on lithium transport, distribution, and concentration inside the working electrode on the nanometer scale. The main goal is a precise kinetic study of the first lithiation in which the crystalline Si is transformed into amorphous LixSi and the first delithiation in which the amorphous LixSi is transformed into amorphous Si. This, in turn, can be compared to the second lithiation in which amorphous Si is alloyed with Li at the beginning of the reaction until the nonreacted silicon is reached and further lithiated. Special emphasis is laid on the spatial and temporal evolution of the lithiated phases and the interfacial layers together with their influence on the kinetics of the alloy formation.
containing nanoscopic Si particles have been presented, and the results obtained so far indicate that Si nanosystems may lead to a viable technology for high-capacity anodes in Li-ion batteries.3,17,18 A detailed microscopic understanding of the alloying kinetics of Si by Li is the necessary prerequisite for future applications. Transmission electron microscopy (TEM)19−26 has recently brought tremendous progress. These TEM studies have clearly revealed that the lithiation of Si proceeds in a very narrow reaction front in which the crystalline structure of the Si host is broken up, leading to an amorphous alloy.20,22 The kinetics of lithiation is coupled to the mechanical stress, which may be followed by a strong retardation of the reaction front.24,26 The rate of lithiation depends on the crystallographic plane, and certain TEM studies have revealed the details of alloy formation directly at the reaction front.20,23,24 Delithiation leads to amorphous silicon, and the second lithiation is therefore characterized by different kinetics.19 In situ TEM could also give information about oxide layers on the Si substrate and their transformation during the process of alloying.21 A detailed kinetic study by TEM, however, is difficult and restricted to distances over 10−20 nm. In addition, a quantitative analysis of the amount of Li incorporated in Si is only possible with crystalline samples. The electron beam is also able to inhibit the lithiation and induce delithiation even though no current has been applied. Therefore, the experiments must be conducted with extreme caution to avoid a possible disturbance of the electron beam.25 X-ray diffraction (XRD) is another suitable method to study the lithiation and to give important insight into the processes taking place during cycling.27−29 It was found that the mechanism of alloying is an electrochemically driven amorphization.27 When the potential of the electrode falls below ∼50 mV vs Li/Li+ during lithiation, the highly lithiated amorphous lithium silicide crystallizes to c-Li15Si4.28,29 During subsequent delithiation and lithiation steps, the amorphous aLixSi phases are formed together with amorphous Si.28,29 Neutron reflectivity (NR) offers an alternative, nondestructive method for the quantitative and time-resolved investigation of alloy formation between Si and Li in stratified systems on the nanometer scale.30−34 NR is sensitive to small changes of the scattering length density (SLD) as a function of depth. It is ideally suited to monitor the intake of Li in a Si electrode and Si slabs with precise crystallographic orientation. As the neutron beam easily passes the solid silicon electrode, which has a very low absorption coefficient of ∼0.004 mm−1 (at a wavelength λ = 0.5 nm),35 it is optimally suited to probe structural and compositional evolutions at the electrode/ electrolyte interface under working conditions (i.e., operando conditions), that is, in an electrochemical cell under welldefined conditions with a constant current and well-known potential. Moreover, Li has a negative scattering length and therefore exhibits a high contrast toward Si, which has a positive scattering length.30−34 By monitoring the changes in the SLD during lithiation, one can obtain the amount of incorporated lithium in LixSi.30 NR has already been employed ex situ as well as in situ to study amorphous Si layers.30−34 The studies resulted in precise information about the concomitant volume change.30,31 The spontaneous non-electrochemically driven reaction between an a-Si electrode and a liquid electrolyte (LiPF6/ethylene carbonate/dimethyl carbonate) has also been investigated. A layer with a thickness of 3.5 nm was found, which probably
RESULTS The measured open-circuit potential of the c-Si electrode vs Li/ Li+ in virgin state was 2.9 V. Lithiation and delithiation were carried out at a current of +25 μA for 720 min and −25 μA for 85 min (first cycle) and +25 μA for 720 min and −25 μA for 167 min (second cycle), respectively. The current applied during lithiation induced a rapid decrease of the electrode potential from the open-circuit potential to ∼0.13 V, where the potential remained until the end of the respective measurement (see Figure 3). The subsequent delithiation was induced by the reversal of the applied current. NR patterns were recorded for 5 min each during the ongoing lithiation and delithiation within a Qz range of 0.07−0.63 nm−1. Figure 1 shows typical examples of the measured reflectivity curves (see Supplementary Note 1 and Figures S1−S6). Since the neutron beam impinges on the electrode/electrolyte interface from the crystalline silicon substrate with higher SLD, no total external reflection is observed at low Qz values. After lithiation and delithiation, the applied current was stopped to equilibrate the system. Equilibrium was indicated by a constant potential with time and taken as a precondition for recording a neutron reflectivity pattern within a Qz range from 0.07 to 2.50 nm−1 with a total time per run of about 2 h (Figure 1 and Figures S1 and S4). The applied currents of +25 μA and −25 μA lead to changes in the shape and intensity of the NR curves of the virgin state and after lithiation/delithiation (Figure 1 and Figure S1−S6), respectively. The curve taken at the end of the first lithiation shows Kiessig oscillations with a frequency of ΔQz = 0.121 nm−1 (Figure 1) corresponding to d ∼ 52 nm. The curves of the virgin state and the one taken after 40 min of lithiation display a monotonic increase of RQz4 with Qz (Figure 1b). After 265 min, the recorded curve shows a decrease in RQz4 up to Qz = 0.25 nm−1 followed by an increase (Figure 1b). The 7459
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
ACS Nano
Figure 2. Characteristic SLD profiles of the virgin state (black dots), taken during lithiation and at the end of first lithiation (dark red) corresponding to the fits to the NR data in Figure 1 and Figures S1 and S2. After ∼370 min, the native SiO2 vanishes and the SLD further decreases with time inside the silicon electrode (dark gray). SLD values between ∼1.80 × 10−4 and −0.13 × 10−4 nm−2 are reached, equivalent to the deep and surface lithiation, respectively.
silicon surfaces exposed to air39 (Figure 3, violet), a SL (Figure 3, light green), and the electrolyte (El) (Figure 3, yellow). The surface layer may be composed of products induced by a decomposition reaction between silicon oxide and the liquid electrolyte, as recently observed by Veith and co-workers on a comparable system.32 During the early stage of lithiation, that is, before a deep lithiation of the Si electrode sets in (t < 390 min), several changes occur at the c-Si/SiO2/SL/El interface (Figure 2 and Figure 3). At the beginning of lithiation, a zone with a SLD of 1.34 × 10−4 nm−2 is formed in front of the electrode surface at a distance from 4.5 to 10.0 nm (Figure 2, red line, Figure 3, orange zone). The SLD of this phase is lower than the SLDc‑Si and lower than the SLDEl. Evidently, Li ions are enriched in this region. The maximum Li enrichment amounts to 13% with respect to the electrolyte. To the best of our knowledge, such a lithium enrichment zone has not been reported in the literature until now. The enrichment phase disappears after ∼100 min and occurs only during the first lithiation. Shortly after this, the enrichment zone diminishes and the SLDc‑Si of the c-Si electrode begins to decrease between 0 and −3.0 nm (Figure 3, yellow-red domain). After 235 min, the SLDSL decreases until it reaches the value of SLDEl = 1.66 × 10−4 nm−2; 130 min later (t = 365 min), the SiO2 layer vanishes. Simultaneously, the deep lithiation of c-Si begins. The dissolution of the native SiO2 layer and its transformation to a Li-rich layer (most likely, Li2O or Li x SiO y ) during lithiation has been reported in the literature.20,21,40 The SLD of the new Li-rich compound (e.g., SLDLi2O = 0.8 × 10−4 nm−2) may approach values similar to those found for LixSi, making it difficult to distinguish between the two distinct layers with NR due to low contrast matching. Final Stage of the First Lithiation. Two main regions with different SLDs are formed, namely, a smaller skin region inside the Si electrode with a minimum SLDs of −0.08 × 10−4 nm−2 starting from z ∼ 0 nm moving inward (toward negative z) followed by a larger growth region of SLDg = 1.80 × 10−4 nm−2 between −26.0 and −58.0 nm (Figure 2). The thickness of the growth region is ∼32 nm, and the thickness of the skin region is ∼18 nm by the end of the first lithiation. The total penetration depth is ∼50 nm, which is in good agreement with the separation of the Kiessig fringes (Figure 1 and Figures S1
Figure 1. (a,b) Neutron reflectivity patterns measured in virgin state (black), during lithiation after 40 min (red), after 265 min (blue), and after the first lithiation (cyan). For a better visualization, the curves were plotted as reflectivity R (a, the curves are shifted for a better view) and R·Q4 (b) vs time. Solid lines are the best fits to the data, based on the SLD profiles shown in Figure 2. The curves for the virgin state and first lithiation were measured over about 2 h and the curves during lithiation over 5 min total data acquisition time.
interference patterns of the Kiessig fringes36 at the end of first 2π and second lithiation were analyzed according to Q z , n = d n, with n being the reflection order. The slope of the linear regression yields a total thickness of the lithiated phase of dtotal,1 ∼ 52.0 ± 1.0 nm and dtotal,2 ∼ 87.0 ± 1.2 nm. The data were analyzed by fitting a box model to the experimental reflectivity curves (for more information, see Supplementary Note 2 and Figure S7).37 The scattering length densities of the c-Si electrode, SLDc‑Si = 2.07 × 10−4 nm−2, the native SiO2 layer, SLDSiO2 = 3.47 × 10−4 nm−2, and the backing electrolyte (El), 1 M LiClO4 in propylene carbonate, SLDEl = 1.66 × 10−4 nm−2, were obtained from their respective literature values.38 The theoretical reflectivity curve calculated on the basis of the interfaces between the working electrode, the native SiO2 layer, and the electrolyte did not match the experimentally recorded pattern of the virgin state. One additional surface layer (SL) of ∼2.0 nm thickness with SLDSL = 1.87 × 10−4 nm−2 at the SiO2/electrolyte interface was necessary to describe the measured reflectivity curve (Figures 1 and 2, black dotted line; Figure S1a,b). In the following, we will discuss the different stages of lithiation. Early Stage of the First Lithiation. Figure 2 displays selected SLD profiles, whereas Figure 3 shows a color map of the SLD profiles plotted against the lithiation time and distance from interface for the first lithiation. At the beginning of lithiation, four regions are visible (Figure 2, black dotted line): c-Si (Figure 3, green), SiO2 which is a common feature of 7460
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
ACS Nano
Figure 3. SLD profiles of the first lithiation as a function of time and distance from the interface. The right-hand side of the diagram displays the potential of the working electrode vs Li/Li+. At the beginning of lithiation, four regions are visible: c-Si (green), the native SiO2 layer (violet), a surface layer (light green), and the electrolyte (yellow). The following processes can be divided into three phases: surface enrichment (orange), surface lithiation (red/dark-red), and deep lithiation (red-yellow). Black dotted lines depict the example SLD curves shown in Figure 2 at a potential of 0.13 V vs Li/Li+.
and S4). It is evident that the lithiation process has to be subdivided into two distinct parts: (i) strong lithiation in a skin region of the electrode and (ii) weak lithiation in an adjacent growth region toward c-Si. The maximum lithium concentration in the skin region is found at a distance between −3.0 and 0 nm from the electrode/electrolyte interface at the lowest SLD and amounts to about xs,max ≈ 2.5 (from eqs 4 and 5). This value is comparable to x = 2.33 reported by Key and coworkers for the amorphous LixSi phase in crystalline silicon powder by nuclear magnetic resonance (NMR) spectroscopy and x ∼ 2.5 reported by Wang et al. by estimating the volume increase and the corresponding Li intake.29,41 In contrast to these results and our findings, Li et al. proposed a lithium content of x = 3.5, which was derived from in situ XRD measurements.28 The amount of lithium in the deep lithiation region is about x ≈ 0.1 in a-LixSi. A similar behavior was observed by Trill et al., who used nano-Si and the commercial cathode material LiCoO2 as electrodes.42 Trill et al. found that cycling led to lithiation of the nanoparticles until a certain penetration depth (∼30−40 nm) was reached. Further lithiation did not progress toward the interior of the particle but rather leads to a higher lithium content of the already lithiated region until Li15Si4 developed.42 At the end of lithiation and at potentials φ < 0.07 V, the lithiated phase was reported to be transformed into crystalline Li15Si4.28,29 We never reached a potential of φ < 0.07 V and also found no evidence by NR for the formation of the crystalline Li15Si4.28,29 First Delithiation. Figure 4b shows a color map of the first delithiation analogous to Figure 3, and Figure 4a displays characteristic SLD profiles of this process corresponding to the NR data shown in Figure S3. The main features are the decrease of thickness and lithium concentration of the LixSi phase. There is a finite amount of Li that remains in the skin region of the electrode after the delithiation is finished. Jerliu et al. reported a similar behavior for amorphous Si.30 In addition to this, a single thin layer of higher SLD is formed on the electrode/electrolyte interface. The changes in thickness and SLD occur until final values of ∼19.0 nm, 1.95 × 10−4 nm−2 (growth region), and 3.0 nm and 0.56 × 10−4 nm−2 (skin region) are reached (Figure 4a,b). The residual amount of Li in LixSi is characterized by x ≈ 1.1. The thickness of the formerly lithiated region is ddelith,total ∼ 20 nm smaller than in the lithiated state (dlith,total ∼ 50 nm), and the associated SLD equals the SLD of amorphous silicon (SLDg,delith = SLDa‑Si = 1.95 × 10−4
Figure 4. (a) Characteristic SLD profiles at the end of the first lithiation (red), during delithiation (gray), and at the end of the first delithiation (blue) corresponding to the fits to the NR data in Figure S3. (b) SLD profiles of the first delithiation as a function of time, distance from interface, and potential of the working electrode vs Li/Li+. The depletion of Li is represented by a color change from yellow/red (LixSi) to green (Si). After ∼40 min, a surface layer at the electrode/electrolyte interface is re-formed (light green). A residual amount of lithium remains in the electrode (red) after complete delithiation at t = 85 min. Black dotted lines depict the exemplary SLD curves shown in panel a.
nm−2) or LixSi with x ≈ 0.05. This finding agrees with earlier reports on the irreversible transformation of c-Si into a-Si after the first lithiation/delithiation cycle.3,4,19,28,29 The single thin layer at the electrode/electrolyte interface forms after ∼45 min with an SLDSL of 1.99 × 10−4 nm−2 and a thickness of 3.6 nm. We conclude that decomposition processes between the electrode and the electrolyte lead to the formation of this new surface layer. Second Cycle. Figure 5 displays the results obtained during the second lithiation and delithiation. During the second 7461
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
ACS Nano
Figure 5. SLD profiles of the second lithiation (a) and second delithiation (b) as a function of time and distance from the interface. The lithiation/delithiation of the Si electrode by Li is visualized by a color change from green (c-Si) to yellow/red (LixSi) and vice versa. Decomposition of the surface layer during lithiation and formation during delithiation is depicted as a green area at the electrode/electrolyte interface. Black dotted lines depict the exemplary NR curves shown in Figures S5 and S6.
Figure 6. (a) Changes of the complete LixSi layer thickness ds+g with time for the first and second lithiation and delithiation. The surface and deep lithiation are displayed by different slope kinetics of electrode penetration by lithium. (b) Changes of the lithium content x in LixSi in the skin region of the first and second lithiation. A theoretical error of 5% was calculated from the average of 10 SLD values.
lithiation, no enrichment zone is visible at the electrolyte/ electrode interface, and the surface layer formed during the first delithiation decomposes after ∼400 min of lithiation. In general, the second lithiation and delithiation follow the path paved by the first cycle. Again, growth and skin regions form during lithiation (dg,2 ∼ 76.0 nm, SLDg,2 = 1.80 × 10−4 nm−2; ds,2 ∼ 18.0 nm, SLDs,min,2 = −0.08 × 10−4 nm−2). The maximum lithium concentration in the skin region amounts to about x ≈ 2.4 in LixSi and is similar to the value found for the first lithiation (x ≈ 2.5) (Figure 6b) as well as the lithium amount in the growth region (x ≈ 0.1 in LixSi). The residual amount of Li in LixSi after delithiation is about x ≈ 1.1, and the thickness of the surface layer is 5.4 nm with an SLDSL of 1.99 × 10−4 nm−2.
results suggest that strong lithiation of the bulk silicon electrode is limited to the skin region, and further growth of the highly enriched LixSi phase (x ≈ 2.5) is hindered. Beyond a depth of ∼18.0 nm, the electrode is further penetrated by a small amount of lithium in a lowly enriched LixSi phase (x ≈ 0.1). Additionally, the internal roughness at the interface between cSi and a-LixSi (x ≈ 0.1) propagates over ∼10 nm in the z direction (Figures 2 and 4), while a thickness of ∼1 nm was reported by TEM measurements for the c-Si/a-LixSi phase boundary.20 Internal stress induced by the large volume changes accompanying the formation of the LixSi phase (x ≈ 2.5) may be the main reason for the two different lithium concentration areas arising during lithiation and the increased transition zone at the c-Si/a-LixSi interface. In addition, the current density we used in our experiments to detect all intermediate stages of lithiation and delithiation is very low at 2 μA/cm2, leading to different reaction conditions and kinetics compared to related TEM measurements with a much higher current density (e.g., 54 μA/cm2).44 These findings confirm earlier reports that the internal stress originating from the lithiation processes is able to decrease the driving force for lithiation or even to slow it down to a point where it ceases.3,24,26,42,43 In contrast to most previous reports, we did not detect a phase with the composition of the final lithium silicide c-Li15Si4 crystallizing at potentials of φ < 0.05 V. Such a potential has never been reached in our experiments. Reasons for this may be the low current density used and the high thickness of the electrode at delectrode = 1 cm. Li28 and Iaboni45 et al. recommend
DISCUSSION At first glance, our results seem in-line with present literature that finds 2.33 ≤ x ≤ 3.5.28,29,41 The coexistence of the two phases, c-Si and a-LixSi, during lithiation that has been revealed first by TEM measurements is clearly visible (Figures 2−5).3,19,20,22,41 However, neutron reflectivity reveals a different spatial distribution for the planar system under consideration here. While XRD, NMR, and TEM studies by Li,28 Key,29 Cui,19 Huang,20 and Wang41 et al. suggest a constant lithium concentration during lithiation of crystalline silicon powder or nanoelectrodes, our experiments on a planar, 1 cm thick single-crystal silicon electrode demonstrate that the Li concentration varies with penetration depth and time. The width and SLD of the skin region is identical for all cycles and therefore independent of the cycle number. The width of the growth region increases from 32.0 to 76.0 nm from the first to the second lithiation, whereas its SLD remains the same. These 7462
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
ACS Nano
m/s for the second lithiation, −7.6 × 10−12 m/s for the first delithiation and −7.1 × 10−12 m/s for the second delithiation, cf. Figure 6a). These values are smaller than the value of 2.5 × 10−11 m/s measured by Pharr et al. for Si nanopillars in the ⟨100⟩ direction.44 The main reason for this difference is located in the different currents used for both lithiation experiments.44 We used a current density of 2 μA/cm2 during the first and second cycle, whereas Pharr et al. used 45 μA/cm2. In addition, the lithiation kinetics of bulk c-Si and Si nanoparticles is expected to be different. The internal stress originating from volume expansion during lithiation hinders or slows the lithiation of bulk c-Si electrodes, as already mentioned before.3,24,26,43 Nanowires up to a size of 110 nm are able to accommodate the stress much easier than a planar slab.43 Figure 6b displays the lithium content in the skin region during cycling. After an induction time of tfirst lith ∼ 110 min and tsecond lith ∼ 20 min, it shows a nonlinear increase. A plateau is visible during relaxation. Another plateau is observed for the second lithiation after t ∼ 80 min, where the lithium content seems to stagnate while the thickness of the LixSi layer still increases (Figure 6a). At 195 min later (t ∼ 275 min), the lithium content shows a second increase and a second plateau at t ∼ 450 min when it reaches the final value of about x ∼ 2.4 at the end of lithiation and relaxation. A possible explanation for the stagnation of the lithium content is the breakthrough into the nonlithiated electrode material represented by an energy barrier which needs to be overcome. The delithiation shows a linear decrease of the lithium content until a value of x ∼ 1.1 in LixSi is reached.
the avoidance of the formation of c-Li15Si4 by controlling the potential of the electrode at φ > 0.05 V to prevent capacity fading and electrode damage due to inhomogeneous volume change in the c-Li15Si4/a-LixSi coexistence regions. The potential at which the breakthrough from surface to bulk lithiation occurs during the second lithiation is higher than that for the first lithiation (φbt,1 = 0.13 V < φbt,2 = 0.3 V). Hence, the second lithiation should occur faster and easier than the first. This observation confirms earlier reports on a difference in behavior of crystalline silicon electrodes on first and subsequent cycles due to the transformation of crystalline to amorphous Si after the first delithiation.19,29 Hence, the second lithiation occurs via a single-phase reaction.3,19,20,22 In contrast to the investigations of the lithiation of c-Si nanoparticles with TEM, the c-Si electrode in our experiments was lithiated up to a thickness of d ∼ 50 nm in the first lithiation. During the second lithiation, previously unlithiated material was penetrated as well by lithium so that the two-phase lithiation behavior was also observed for the second lithiation.3,19,20,22 A small amount of x ∼ 0.1 in LixSi, which was found in the growth region after lithiation, probably does not destroy the crystalline structure of silicon. Therefore, the growth region may be composed of crystalline silicon after delithiation, explaining the two-phase lithiation behavior also occurring during the second lithiation. Volume expansion and anisotropic swelling may lead to cracks at the electrode surface. An ex situ examination of the electrode with a light microscope showed no visible deterioration. It has been reported in the literature that the volume expansion normal to the ⟨100⟩ surface is less than that for the ⟨110⟩ or ⟨111⟩ surface.44 This effect together with the relaxation of the stress inside the electrode due to the lowly lithiated phase and the large electrode area of 12.9 cm2 explains the absence of cracks in the electrode. A feature that can be studied particularly well is the formation of surface layers on the Si electrode. During the second delithiation, this layer starts to form after ca. 35 min and grows to a final thickness of 5.4 nm with an SLDSL of 1.99 × 10−4 nm−2. For both cycles, the surface layer completely vanishes on lithiation for φ < 0.30 V and reappears on delithiation for φ > 0.39 V. The surface layer has the same SLD value for both delithiation steps and a thickness of 3.6 nm after the first delithiation and 5.4 nm after the second delithiation. We tentatively identify this layer as an solid electrolyte interface (SEI).40,42,46−49 Veith et al. reported the formation of a solid electrolyte interphase during lithiation and delithiation of a-Si.33 In contrast to our findings, the thickness of the SEI layer increases during lithiation and thins during delithiation. A possible reason for this difference may be the much smaller current density in our experiments and differing reaction conditions (electrolyte, silicon electrode, electrochemical experiments). We now turn to a quantitative evaluation of the kinetics of lithiation and delithiation. Figure 6a demonstrates that the thickness of the complete LixSi layer ds+g during lithiation and delithiation exhibits a linear growth with time. This indicates a reaction-controlled growth process in which the back reaction is characterized by a higher rate constant. The lithium atoms, penetrating the surface, quickly diffuse through the already lithiated layers to react with c-Si to a-LixSi at the reaction front. Since the latter process is slower, the reaction-controlled process is the rate-limiting step. The growth rate of ds+g of both lithiation and delithiation steps is nearly the same for both cycles (2.1 × 10−12 m/s for the first lithiation and 1.8 × 10−12
CONCLUSION This work presents a quantitative investigation of the lithiation of crystalline silicon by operando neutron reflectometry. From the analysis of the measured data, we derive precise concentration profiles of lithium in the c-Si electrode as a function of time. We find the following determinations: (1) The first lithiation proceeds via two stages: an early stage in which different reactions occur at the electrode surface and the final stage where the c-Si electrode is deeply lithiated. (2) The reactions that occur at the electrode/electrolyte interface are (a) the formation of a surface layer composed of decomposition products between silicon oxide and the liquid electrolyte without applying any current, (b) the formation and depletion of a lithium enrichment zone, (c) the decomposition of the native SiO2 and surface layer in the early stage of the first lithiation, and (d) the formation of a solid electrolyte interface during delithiation. (3) The lithiated zone is subdivided into two distinct parts: (a) strong lithiation in the skin region of the electrode with x ≈ 2.5 in LixSi and (b) weak lithiation in an adjacent growth region toward c-Si with x ≈ 0.1 in LixSi. (4) After complete delithiation, a well-defined amount of lithium remains present in the skin region of the electrode with x ∼ 1.1 in LixSi. (5) The velocity of lithiation of Si is ∼2 × 10−12 m/s for the first and second lithiation and ∼7 × 10−12 m/s for the first and second delithiation for a current density of 2 μA/cm2. (6) The thickness and the composition of the highly lithiated skin layer are independent of the cycle number. In contrast, the thickness of the growth region increases from the first to the second lithiation, whereas its composition is independent of cycle number. Thus, our operando neutron reflectometry studies demonstrate that crystalline Si slabs with ⟨100⟩ orientation can be successfully used as an anode material if the lithiation is limited 7463
DOI: 10.1021/acsnano.6b02032 ACS Nano 2016, 10, 7458−7466
Article
ACS Nano to ∼20 nm, for example, by a respective charge/discharge program.
electrode and the liquid electrolyte at a defined grazing angle. For the applied specular conditions, the incident angle θ equals the exit angle. The momentum transfer
EXPERIMENTAL SECTION Qz =
Electrochemical Cell. The detailed design of the electrochemical cell used in this work is described elsewhere.30,31 A silicon single crystal (Sil’tronix ST, Archamps, France, n-doped (phosphorus), resistivity < 0.005 Ω·cm, orientation ⟨100⟩, dimensions = 50 mm × 50 mm × 10 mm) was used. The crystalline silicon substrate further served as a working electrode (Figure 7a). A current collector was not
4π sin θ λ
(1)
between an incoming (primary) and an outgoing (reflected) beam is thus perpendicular to the interface. The ratio R = Ir/I0 of the intensities of the reflected and the primary beam defines the reflectivity of the interface and is recorded as a function of Qz. For sufficiently high angles, it is linked to the scattering length density profile across the interface by52 R(Q z) = RF(Q z)
1 SLD∞
∫
dSLD(z) exp(iQ zz)dz dz
2
(2)
where RF is the Fresnel reflectivity of the ideal interface, and it scales with Qz−4, and SLD∞ is the SLD of the bulk electrolyte solution. The SLD = ∑ibi/V is defined by the atomic composition of the material of interest, where bi is the neutron scattering length of component i and V is the molecular volume.37,53 In the case of a stratified system of layers of different SLD, the corresponding reflectivity curve deviates from a simple Qz−4 decay and exhibits pronounced fringes (Kiessig oscillations).36 The amplitude of these oscillations is a function of the SLD difference of the individual strata, and the spacing ΔQz of the oscillations is a direct measure of the thickness, d, where
Figure 7. (a) Scheme of the electrochemical cell used for neutron reflectometry studies. A crystalline silicon working electrode is in contact with the electrolyte consisting of 1 M LiClO4 in propylene carbonate. The counter electrode consists of metallic lithium. The layer formation of the Li−Si alloy indicated in dark blue is monitored as a function of time by the evaluation of the reflectivity profiles. (b) Scheme of the surface normal, z, vs time, t, at the electrode/electrolyte interface during lithiation.
d= necessary in the present setup because the highly doped crystalline silicon block was sufficiently conductive. Metallic lithium (Alfa Aesar, 99.9%) was used as the counter and reference electrodes. The electrolyte was a 1 M LiClO4 (Sigma-Aldrich, 99.99%, battery grade, dry) solution in propylene carbonate (Sigma-Aldrich, 99.7%, anhydrous). The solvent was dried for 1 week under argon atmosphere using a molecular sieve (Carl Roth, 0.3 nm) before the lithium salt was added. Additionally, a microporous polyethylene separator with a thickness of 20 μm (Brückner Maschinenbau, Germany) was placed between the two electrodes. The electrochemical cell was assembled in an argon-filled glovebox (water content