Silicon Nanowire Degradation and Stabilization during Lithium

Apr 28, 2014 - ... on reinforcing the nanowire base, applying strain-graded structures to ...... Chunzeng Li , Stephen Minne , Xingcheng Xiao , Brian ...
1 downloads 0 Views 5MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Silicon Nanowire Degradation and Stabilization during Lithium Cycling by SEI Layer Formation Jeong-Hyun Cho*,†,‡ and S. Tom Picraux*,‡ †

Department of Electrical and Computer Engineering, University of Minnesota, Twin Cities, Minnesota 55455, United States The Center for Integrated Nanotechnologies, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, United States

‡

S Supporting Information *

ABSTRACT: Silicon anodes are of great interest for advanced lithium-ion battery applications due to their order of magnitude higher energy capacity than graphite. Below a critical diameter, silicon nanowires enable the ∼300% volume expansion during lithiation without pulverization. However, their high surface-tovolume ratio is believed to contribute to fading of their capacity retention during cycling due to solid-electrolyte-interphase (SEI) growth on surfaces. To better understand this issue, previous studies have examined the composition and morphology of the SEI layers. Here we report direct measurements of the reduction in silicon nanowire diameter with number of cycles due to SEI formation. The results reveal significantly greater Si loss near the nanowire base. From the change in silicon volume we can accurately predict the measured specific capacity reduction for silicon nanowire half cells. The enhanced Si loss near the nanowire/metal current collector interface suggests new strategies for stabilizing nanowires for long cycle life performance. KEYWORDS: Lithium-ion battery, solid electrolyte interphase (SEI), silicon nanowire, fracture, surface morphology, current density

A

nanowires to fracturing over longer periods of cycling (many hundreds to thousands of cycles) have been achieved by designing more complex shapes for the nanowires.8,9 In addition, using nanowire growth through templates, we have recently shown that capacity retention over extended cycling can be significantly enhanced by improved design of the nanowire contact with the current collector electrode.10 This approach avoids the formation and stress build-up in Si islands around the nanowire−electrode interface which can lead to progressive nanowire delamination. While these advances have improved the viability of Si nanowire anodes, there remains an additional fundamental issue for high cyclability performance for all high capacity nanoscale structures due to their high surface-to-volume ratio. During cycling, the electrolyte undergoes a reduction reaction, and a thin passivation layer, called the solid-electrolyteinterphase (SEI) layer, forms on the anode surface in Li-ion batteries. Upon initial cycling this SEI layer forms and grows as a result of inorganic and organic electrolyte decomposition because of the low potential of the anode as Li-ions are inserted into the anode.11 A stable and continuous SEI layer can protect the anode from subsequent irreversible reactions with the electrolyte. However, given the reactive nature of Si, if active Si surface sites are re-exposed to the electrolyte during continued cycling, for example due to stress-induced fracture of the SEI layer or the Si surface region, a loss of active Si material, electrolyte, and Li will occur. If present, this lack of SEI layer

mong many different types of rechargeable batteries, lithium-ion batteries are progressively being used not only in portable electronic systems, but also in electrical vehicles and aerospace systems. Advantages for these applications include the lack of a memory effect, low self-discharge, and relatively high energy and power densities compared to those of other battery types, such as lead-acid, nickel cadmium, and nickel metal hydride. However, the power and energy densities of commercialized lithium-ion batteries for portable electronics and transportation systems must continue to be improved to meet consumer requirements. The drive to enhance energy and power capacity without sacrificing cyclability for extended rechargeable battery life has led to great interest in silicon nanowire-based anode materials, due to their order of magnitude greater energy capacity compared to today’s commercially used graphite anode materials.1 These onedimensional structures, contacted at their ends to the current collector, allow enough space between the nanowires to accommodate the large volume changes for Si (∼300%) during charging (Li-ion insertion: lithiation) and discharging (Li-ion extraction: delithiation), as well as allowing axial and radial directional stress relaxation of the wires. This stress relaxation alleviates the progressive pulverization of Si which occurs in bulk and thick film structures due to the large volume change during cycling.2−5 To avoid stress-induced fracturing, the Si-based anode materials must be below a maximum critical size. Critical values of ∼150 nm for silicon nanoparticles6 and ∼300 nm in diameter for silicon nanowires have been reported,7 below which Si can be fully lithiated and delithiated without apparent fractures and cracks. Further improvements in the stability of Si © 2014 American Chemical Society

Received: January 12, 2014 Revised: April 13, 2014 Published: April 28, 2014 3088

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

Figure 1. Conceptual (a−c) schematic and (d−f) corresponding SEM images showing morphological changes of silicon nanowire anodes as a result of SEI layer growth. (a,d) Pristine silicon nanowires on PtSix pedestals on Pt thin film electrodes synthesized in porous anodic aluminum oxide templates. (b,e) Propagation of SEI layers into a SiNW by consuming the Si anode materials while Li cycling, (b) cross sectional and (e) side view images before removing SEI layers. (c,f) Nanowires after Li cycling followed by the removal of the SEI layers. (d,f) Insets: SEM images captured with 45° tilt angles clearly show the SiNW diameter decreased relative to the reference PtSix pedestal diameter.

needed which directly examine the change in the SiNWs due to SEI layer evolution during cycling. Furthermore, the morphology changes in the NWs should not be assumed to be uniform along the length of the NWs, since a high stress will be generated near the interface at the base where the NWs are connected to the metal electrode which acts as the current collector, and also since the electron current density in the NWs during cycling will increase with increasing distance down the wire toward the electrode. In this work we have investigated SEI layer formation on SiNWs with cycling and directly determine, for the first time to our best knowledge, the loss of Si from SiNW anodes due to SEI layer growth as a function of Li ion cycling. By using template growth to synthesize the SiNWs on a platinum silicide (PtSix) pedestal, which does not actively undergo SEI formation or measurable dimensional changes during lithium cycling, we create a built-in reference diameter for every nanowire (NW). We hypothesized that morphological change in the SiNW diameter would not necessarily be uniform along the length of the NWs, and using our diameter reference we have shown that a more rapid loss of Si occurs near the bottom of the SiNWs close to the current collectors where the stress and free electron density (current density) are the highest during battery cycling. Since connectivity between the SiNW and electrode must be maintained, the physical stability near the bottom of the SiNWs is one of the critical factors determining the life of the SiNW anodes for Li ion cycling. We find the SEI layers can initially result in loss of Si at rates as high as 0.1−0.5 nm/cycle near the NW base. This rapid initial decrease in SiNW diameter is observed within the first ∼150

and Si stability will result in a continuing irreversible capacity decrease with cycling.12−17 To prevent such capacity fade, the SEI layer and Si nanowire must be physically stable after some limited number of cycles. Because of the large surface-tovolume ratio of nanoscale materials, such SEI layer stability is particularly critical for maintaining long-term cyclability.9,14 Previously, many literature reports have shown physical conditions of the Si anode materials by in situ observations.6,18,19 However, there have been relatively few investigations focused on the formation and stability of SEI layers on Si nanowires (SiNWs).20−22 Those studies have primarily examined the morphology and composition of the SEI layers after a few cycles based on scanning electron microscopy (SEM), X-ray photoelectron spectroscopy (XPS), time-of-flight secondary ion mass spectrometry (ToF-SIMS), and Fourier transform infrared spectroscopy (FT-IR) analysis.20−22 They have found that the nanowires themselves are not fractured when below the critical diameter, and cracks appear only in the SEI layer around SiNWs.21 While these studies have helped to understand the surface chemistry of the SEI layers on SiNWs, as the investigators have pointed out, it is not possible to distinguish the SiNWs from the rough SEI layers formed on the surface of the SiNWs.21 Furthermore, the irregular surfaces of the SEI layer on individual SiNWs22 (which can even lead to a dense array of NWs being buried by SEI material21−24) makes it extremely challenging to draw conclusions about the morphological changes of the SiNWs with the cycle number. It is, however, the retention of the active Si anode material upon repeated cycling and SEI growth that is essential for Li ion storage and anode capacity retention. Thus, studies are 3089

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

nm of the NW base with a decreased loss further up the length of the NW. By modeling the distant-dependent SiNW diameter change with cycling we are able to accurately predict the measured decrease in specific capacity with SEI layer growth over the first 200 cycles. The results imply that the size and evolution of the individual SiNWs prior to stabilization of the SEI layer play a critical role in the long term capacity retention during cycling for SiNW-based anode materials. To investigate the morphological changes of SiNWs during Li-ion cycling, phosphorus-doped n-type SiNWs 46 nm ± 3 nm in diameter and about 1 μm long were synthesized using anodic aluminum oxide (AAO) templates.10 During initial SiNW growth, short segments (∼100 nm long) of PtSix nanowires at the base of the SiNWs form inside the templates and serve as current collectors attached to the Pt thin film electrode. After SiNW synthesis the AAO template is removed. The detailed experimental procedure is described in the Supporting Information (Experimental methods). As illustrated in Figure 1a and shown in the SEM images of Figure 1d, the diameters of the PtSix pedestals and SiNWs are the same before Li-ion cycling because they are all synthesized within the same uniform nanopores of the AAO template. Since we observed that the delithiated reference PtSix pedestal has negligible volume change due to Li-ion cycling, this pedestal’s diameter provides a precise built-in reference to observe any change in the SiNW diameter upon removal of the SEI layer. As Li cycling proceeds, the electrolyte decomposes and forms a SEI layer on the surface of the SiNWs. The SEI layer increases in thickness9 until a stable thickness is reached. Due to the reactivity of Si, however, the SEI layer can continue to grow if the Si surface is newly re-exposed to the electrolyte, for example due to SEI layer or Si cracking. Since SEI layer formation consumes some of the Si as it forms or reforms, the Si diameter can be reduced by the SEI formation with increasing number of cycles (Figure 1b and e). By removing the SEI layer and comparing each SiNW to its PtSix reference pedestal, we can directly determine the diameter changes and the volume loss of Si from the SiNWs with Li-ion cycling (Figure 1c and f). To investigate SEI layer formation with cycling, four half cells were assembled and the mass gain of the grown SEI layer was obtained by measuring the mass changes of the four samples before and after 40, 80, 120, and 160 cycles (charging and discharging) using a microbalance (Sartorius SE), which measures down to a 0.1 μg change (Figure 2). The measured mass was normalized by the mass of the SiNWs before cycling to correct for small differences between samples in the total number and length of SiNWs (see Supporting Information, Normalized mass of SEI layer growth). As shown by the normalized mass increase in Figure 2, the SEI layer initially grows rapidly for the first 40 cycles, reaching a value of 5.0 which is 75% of the total SEI growth for 160 cycles. The rate of growth then slows between 40 and 80 cycles and upon continued cycling rapidly saturates to a value of ∼6.6 for the normalized total SEI mass between 80 and 120 cycles, and showing less than a 2% increase between 120 and 160 cycles. This result implies that SiNW-based anode materials may suffer from significant morphological changes up to about 80 cycles due to SEI growth, after which the materials are rapidly stabilized with additional Li cycling. To check the consistency of the mass gain measurements with the SEM images obtained, we consider the measured absolute mass (110.3 μg) due to SEI layer growth after 40 cycles, which is an increase of five times the mass of the pristine SiNWs (22 μg). If we assume a density

Figure 2. Relationship between the mass gain of SEI layer and electrochemical performance with Li ion cycling. The mass of SEI layers is given for four samples and determined by measuring the mass of SiNW based anode materials before and after 40, 80, 120, and 160 cycles. The normalized mass of SEI was obtained by dividing the mass of SEI layers by pristine SiNW mass. Specific discharge capacity measurements were collected between 20 mV and 1.5 V at 0.5C in a SiNW-based half cell. Square symbols are the calculated results of a model of the specific capacity change with cycle number based on the loss of Si: solid yellow line is a curve fit to the model, C = 2987−1353 {1 − exp[−0.0369(cycle no. −1)]}. The 2987 mA·h/g in the curve fit equation is the measured value of discharge specific capacity in the first cycle.

for the SEI layer comparable to the SEI layer density (2.1 g/ cm3)25 for carbon based anode materials, the observed weight gain corresponds to a SEI shell thickness of ∼35 nm around each SiNW. This implies an increase in diameter from 46 nm for the pristine SiNWs to ∼120 nm after 40 cycles with the surrounding SEI layer present. These results are qualitatively consistent with SEM observations for various Li cycling conditions and serve to emphasize the importance of SEI layer formation in the morphological evolution of nanoscale Si anodes. We have found that the observed rapid initial SEI growth followed by saturation to a constant value correlates closely with the observed change in specific capacity of the SiNWs with cycling. As shown in Figure 2 for similarly prepared SiNW anode half cells, the initial specific discharge capacity (∼2986 mA·h/g) of the SiNW-based half cell at first rapidly decreases, reaching 1947 mA·h/g at 40 cycles, after which the specific capacity decreases more gradually up to 80 cycles and then stabilizes at a constant value ∼1900 mA·h/g between 80 and 120 cycles. As seen in Figure 2, the initial decay in the specific capacity with the cycle number followed by reaching a stabilized level is seen to correlate well with the initial growth in the SEI layer followed by reaching a stabilized value as measured by the SEI normalized mass gain. This result suggests the loss of capacity may be the result of the loss of active Si from the nanowires due to SEI formation. A model to explore this interpretation, which quantitatively correlates the measured loss of Si with the specific capacity decrease, will be presented later in this paper. To examine the influence of the SEI layer growth on the SiNW morphology and the possible loss of Si material from the NWs as a result of SEI layer growth, we measured the change in diameter of the SiNWs after removing the SEI layer for a fixed number of lithiation cycles (Figure 3). We were particularly interested in observing the changes of the SiNWs near their bottom, the base region immediately above where the SiNWs connect to the current collectors (PtSix pedestal). This area 3090

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

Figure 3. Morphological (diameter) change of SiNWs, resulting from Si consumption, as the number of cycles increases. Diameters were measured at a point located at 30 nm away from the interface between SiNWs and PtSix reference pedestals. (a) Cross-sectional area and diameter changes of SiNWs. (b) Average Si consumption rate versus the number of battery cycles and (inset) the diameter of SiNWs remaining after cycling. The Si consumption rates per cycle for each datum point were obtained by the measurement of the diameter changes of two samples corresponding to the extent of the shaded areas. A mathematical expression of these average Si consumption rates is (Dn − Dn+40)/40, where n is 0, 40, 80, and 120 and D# is the diameter of a SiNW after # number of cycles. The two circles in the inset schematically represent the diameter of SiNWs before and after cycling. The diameter (D) decreased with each charge/discharge cycle due to SEI growth consuming Si on the surface of SiNWs. (c−f) SEM images of SiNWs showing the morphological changes of SiNWs near their base in comparison to the PtSix pedestal reference after (c) 40, (d) 80, (e) 120, and (f) 160 cycles.

hypothesis, further investigations of the electron mobility and Li atom diffusivity in SiNWs as well as Li-ion mobility in the electrolyte within NW forests are required. The initial diameter of the SiNWs was 46 nm ± 3 nm as obtained by measuring the diameter of the PtSix pedestal as discussed previously (Figure 3). Solid arrows in the SEM images (Figure 3c−f), captured after cycling and SEI layer removal, indicate the places where the diameter of the PtSix pedestals was determined, which represent the diameter of the pristine SiNW. The dashed arrows indicate the diameter of the SiNWs after cycling to the indicated number of cycles, all at a distance of 30 nm along the NW from the interface between SiNWs and PtSix (Figure 3c−f). There is no change in the diameter (volume) of the PtSix pedestal with cycling because metal silicides typically have no significant reaction with Liions.3,28,29 However, the morphological changes of the SiNWs near the base are significant (Figure 3c−f), with the diameter decreasing with an increasing number of cycles. Before cycling, the diameter and cross-sectional area of the SiNWs were 46 ± 3 nm and 1685 nm2 ± 226 nm2, respectively (Figure 3a). After 40 cycles only ∼60% of the diameter (35% of cross-sectional area) of the SiNW near the base of the NW remained (Figure S2 in Supporting Information). The SEI layer kept increasing with each cycle, and the diameters of the SiNWs decreased to 11 ± 2 nm in 120 cycles and further decreased at a rate of 0.1 ± 0.05 nm/cycle after 120 cycles to 7 ± 2 nm in diameter at 160 cycles (Figure 3a,b). As shown in Figure 3b, the Si consumption rate decreased as cycle number increased and reached down to 0.1 nm/cycle in 121−160 cycles. This result also shows that, when the diameters of SiNWs become smaller with each charge/ discharge cycle, the Si consumption rate also decreased [about

adjacent to the metal current collector is a transition region with high stress23 (limited radial strain/stress relaxation in SiNWs near the interface) due to the large expansion of the Si compared to the Pt silicide upon Li cycling. Because of the abrupt stress and limited stress relaxation, the SiNWs near the interface between SiNWs and the metal current collectors (PtSix pedestal) are physically the most vulnerable point of the SiNWs to potential degradation mechanisms. Moreover, the SiNW base region is the place at which the rapid lithiation reaction with Si would be expected to occur first during charging (lithiation), because electrons supplied from the current collector are immediately available to combine with the lithium ions surrounding the SiNW base with minimal impedance compared to the other end of the nanowires. We also note that the region at the base of the SiNWs is the area of highest current density during lithiation as well as delithiation, since the free electron transport controls Li cycling. In general, it is known that the flow of the free electrons is limited by Li atom diffusion and mobility.26 This condition is because the mobility of electrons in most active materials is much higher than that of ions27 and implies that it is not free electron flow, but Li-ion migration that controls the Li alloying reaction. However, the high mobility of electrons is invalid at low Li concentrations in Si anode material during Li cycling.27 Due to this aspect, we should consider not only radial diffusion of Liions from the electrolyte (the case controlled by Li-ions), but also axial diffusion of electrons from the metal current collector (the case controlled by free electrons),27 where the process of Li-ion insertion might be governed by free electrons under certain conditions, such as the low Li concentrations and high impedance conditions of SiNWs. To clearly verify this 3091

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

Figure 4. Morphological changes of SiNWs along their axial distance Z from the interface between current collectors and SiNWs after cycling for 160 times at 0.5C and SEI layer removal. (a) Nanowire width measured at different axial distances Z for a SiNW shown by the SEM image in (b). The left inset in (a) schematically illustrates the distribution of free electron density (current density) in a SiNW during a lithiation process as the electrons combine with Li-ions. The right inset is an approximate stress (σ)−strain (ε) distribution along the distance Z for a SiNW. (b) A SEM image showing a SiNW grown on a PtSix pedestal. Each dashed arrow in the figure indicates points at which the widths were measured and plotted in (a). (c) A SEM image (top view) of about 1 μm long SiNW. Solid arrows indicate the narrow SiNWs close to the current collectors, base, and the dashed arrows indicate the thicker SiNWs away from the current collectors. (d) Top view of a SEM image of ∼20 μm long SiNWs after 160 cycles. SiNWs shown in the image are far away from the current collectors. Bright dots (see white arrows) indicate the Au growth catalyst at the end of the SiNWs. (e) Side view of a SEM image of ∼1 μm long SiNW bundles after 160 cycles.

region at between Z = ∼152 nm and ∼267 nm, and (c) wide nanowire diameters above Z = ∼267 nm. The width of the SiNW at Z = ∼11 nm was ∼12 nm, and the width gradually decreased to ∼10 nm at Z = ∼152 nm. At that point (Z = ∼152 nm) the width rapidly increases and slowly levels off by Z = ∼350 nm, reaching ∼36 nm in diameter. We hypothesize that the increased diameter change near the NW base is due to two factors: limited radial strain/stress relaxation in SiNWs near the interface and different current densities along the nanowires during lithiation. In the first factor the strain/stress changes near the interface might accelerate cracking of SiNWs and/or SEI layers and the resulting SEI growth. At the interface, minimum radial strain is allowed in SiNWs during lithiation/delithiation processes owing to the rigid current collectors under the SiNWs not allowing volume changes; however, maximum radial stress arises because of the restricted radial strain relaxation (restricted plastic/elastic deformation along the radial direction), increasing the possibility of crack development (Figure 4a, right inset). The second possible reason for the diameter change as the axial distance (Z) increases is that the electrons in this one-dimensional structure are supplied from the bottom (metal current collector at the NW base) and disperse with distance Z along the nanowires during charging (lithiation) to

0.47, 0.25, 0.16, and 0.10 nm/cycle at 36, 23, 14, and 9 nm diameter (D), respectively (Figure 3b, inset)], indicating less degradation of Si. We believe this reduction in Si loss is mainly caused by two reasons. First, the thicker SEI layers may have increased stability. Second, the physical robustness of the nanoscale SiNWs and overall SEI shell/NW core structure may be gradually enhanced as the nanowire diameter decreases.7 For example, the volume changes of the smaller diameter SiNWs may not be large enough to physically affect (crack) the thick SEI layer which is 2 orders of magnitude larger than that of the thin SiNWs in volume. Also, the dimensional change in radius upon Li cycling for the SiNW is less for a given % change as the diameter decreases. To better understand this unexpected large decrease in SiNW diameter near the NW base, we investigated the morphological and diameter changes along the length of the SiNWs. In Figure 4a we show the SiNW diameter vs distance (Z) along the NW axis, as measured from the interface between the SiNW and PtSix current collector pedestal, before and after cycling 160 times. The diameters after 160 cycles were obtained from a delithiated SiNW shown in Figure 4b, with the dashed arrows in the figure indicating points at which the widths were measured. After cycling we observe three regions: (a) narrow diameters close to the current collector, (b) a transitional 3092

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

The specific capacity decrease shown in Figure 2 was modeled using the measured loss of Si material with the cycle number (Table S2 in the Supporting Information). The Si loss is determined by the distance-dependent reduction in diameter along the SiNW length after removal of the SEI layer (Figures 3 and 4). The amount of Si remaining for the nanowires was multiplied by a constant of 0.8 to take into account an expected porosity of ∼20% induced in the SiNWs by the lithium cycling. We took the porosity to be constant with cycling since the initial change in porosity with cycling from full density Si nanowires is known to occur rapidly in the first few (∼5) cycles after which the volume change of the porosity is saturated and no significant change is shown.31 This constant is the only free parameter in the model which is seen to be in good agreement with the observed absolute value of the specific capacity without requiring an assumption of additional loss of SiNWs due to detachment (see Figure S5). Here, the amount of Si remaining after a fixed number of cycles determines the remaining specific discharge (delithiation) capacity, where the measured value of 2987 mA·h/g at the start of cycling (1 cycle) was used for the initial specific capacity [detailed information regarding the specific capacity model is given in the Supporting Information (Model for loss of specific capacity using the loss of Si material from the NWs)]. The resultant prediction of specific capacity vs cycle number for the four cycling conditions used is shown by the yellow squares in Figure 2 with the solid yellow line as a fit to the data. As seen in Figure 2, the model for the specific capacity, based on the SEI layer-induced change in SiNW volume, shows very good agreement with the discharge capacity measurement over the entire range of cycles (from 1 to 200 cycles). This result shows a direct correlation between the SEI-induced degradation and loss of Si anode material, and the early capacity fading as the number of battery cycles increases, suggesting that stabilization of the SEI layer plays a critical role in maintaining high specific capacity with long-term capacity retention for SiNWs. We note that the observed SiNW degradation leading to progressive SEI layer formation with cycle number does not directly correlate with the Coulombic efficiency. The average Coulombic efficiency of the SiNW based half-cell battery was about 93% until 80 cycles, and from 81 to 200 cycles the average Coulombic efficiency was 91%. We suggest in our case a parasitic reaction such as electrolyte oxidation32 may be one of the major causes of the Coulombic inefficiency.10 While little is known about the microscopic nature of SEI layer formation and dissolution for Si during Li ion cycling,20 it is of interest to consider what logical conclusions might be reached from the present results, especially regarding the connection between SEI layer evolution and the loss of Si during cycling. It is not known if a stable SEI layer on Si forms after some period of cycling (as for carbon anodes). The relatively small change in SiNW diameter far from the NW base (>300 nm in Figure 4a) along with the saturation of specific capacity fading (Figure 2) suggests that a stabilized SEI layer may be formed, perhaps after some initial cracking of the SEI layer due to the stresses involved. While the SEI layer results primarily from the decomposition of the electrolyte and recent XPS and ToF-SIMS measurements of SEI layers on Si did not observe measurable amounts of Si in the SEI,4,13,20,22 some reaction of the Si surface during SEI formation could occur, for example, giving rise to lithium silicates (LixSiOy: mostly Li4SiO4) and lithium oxide (Li2O) during a lithiation process under the SEI layer.33 We believe this effect could account for

neutralize the positively charged Li. This leads to full lithiation in a shorter time (faster lithiation) at the nanowire bottom where there is lower impedance for electron transport. However, at the top of the nanowires, the lithiation process will be slower than at the bottom due to lower availability of free electrons since they must pass through the length of the nanowires while Li-ions are being actively incorporated all along the nanowire and combining with free electrons supplied from the current collector (Figure 4a left insect). The faster Li cycling might actively degrade the anode material (Si) and/or induce the SEI layer to be physically less stable near the NW base, leading to more SEI layer formation on the surface of newly exposed Si and consuming more Si anode material. Although there is some variation in the values of nanowire width along the distance Z for different nanowires, we observed consistent results from all of the observed nanowires as shown in top (Figure 4c) and side (Figure 4e) view of a SEM image of SiNW bundles captured after 160 cycles. It should be noted that if the free electron density is the only parameter affecting degradation of Si at the base of NWs, the degradation of the SiNWs should be more severe with cycling because the free electron density increases as the SiNW diameter decreases near the NW base with the cycle number. However, the degradation rates presented in Figure 3b show the opposite result. We believe this reduction in degradation is due to the effect of a critical size,6,7 below which Si can be fully charged and discharged without apparent fractures. As we mention earlier, as the diameter of SiNW decreases, the physical stability of SiNWs to fracturing is enhanced because of two factors. First, the thicker SEI layers may have induced stability. Second, the physical robustness of the nanoscale SiNWs7 and overall SEI shell/NW core structure may be gradually enhanced as the nanowire diameter decreases. We believe that these two factors may dominate the physical stability of the NWs as the diameter decreases and overcomes the defect (physical instability) caused by fast alloying rates. We also examined the morphology of much longer (∼20 μm long) SiNWs after cycling to 40, 80, 120, and 160 cycles. Figure 4d shows a top view of these SiNWs after SEI layer removal, where the bright dots indicate the gold (Au) catalyst used for the vapor−liquid−solid growth which serve to mark the end of each NW. Consistent with the results for the 1 μm long SiNWs, the width (∼90 nm in this case) of the SiNWs close to the tip (∼20 μm away from the current collector) did not change significantly after initial cycling (between 40, 80, 120, and 160 cycles) (see Figure S3 in Supporting Information), whereas the width near the base of the SiNW (close to the current collectors) was significantly decreased from that of the tip with cycling. Since shorter NWs will have a lower free electron current density near the nanowire base at a given cycle rate, and a smaller impedance between NW tip and base (see Supporting Information, Figure S4), it is possible that shorter SiNWs may allow for greater physical stability of the NW near the base with less Si loss with cycling. However, additional studies would be needed to confirm this possibility. The present result of different morphological (diameter) changes along the length of the SiNWs, indicating that the SEI layer growth by the diffusion of electrolyte through SEI layers30 is relatively small near the base of the NWs, compared to the SEI formed by fracture of Si anode materials during Li ion cycling. In contrast, if the electrolyte diffusion dominated the formation of the SEI layer, the diameter after many cycles should be relatively uniform along the entire length of the SiNWs. 3093

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

current collectors) may be also present in other onedimensional (nanowire-based) anode materials. These results highlight the importance of the design of robust material transitions in the region near the attachment of nanowires to the current collector and suggest strategies based on reinforcing the nanowire base, applying strain-graded structures to minimize the abrupt stress change,34 and/or reducing current density during lithiation for enhancing the performance of Si nanowire anodes in Li ion batteries.

the observed Si loss along the NW far from the NW−electrode interface (e.g., where the Si loss corresponds to ∼10 nm over 160 cycles in Figure 4a). However, near the base of the SiNW there is a substantially greater amount of Si loss on average during early cycling, with the Si consumption rate as large as 0.47 nm/cycle initially and remaining as high as 0.25 nm/cycle at between 40 and 80 cycles. Such a high rate of Si loss and necking of the SiNW near the base in the present case suggests that microscopic bits of Si may be lost during this early cycling phase. The strain present due to the expansion and contraction of the Si relative to the PtSix electrode base as well as the somewhat more rapid lithiation near the base as discussed earlier may contribute to nanoscale cracks and surface flaking of the SiNW initially. Such loss of Si could lead to the rapid initial decrease in SiNW diameter near the base, with the SEI layer and SiNW stabilizing after longer times at smaller diameters. Thus, reinforcing the stability of the SiNWs near the base, possibly through stress reduction and dimensional changes, may enable greater capacity retention and longer term stability with lithium cycling. In general we find the saturation of SEI growth and the formation of a stabilized SEI layer indicates that the propagation of cracks (fracture) in the SEI layer and/or Si steadily diminishes, finally resulting in the formation of a stabilized SEI layer along the entire SiNW. The SEI layer becoming physically stabilized thus can lead to the relatively steady specific capacities which were achieved after 80 cycles (Figure 2). Consistent with previous literature,9,14 these results indicate that building a stabilized SEI passivation layer is one of the critical strategies in order to enhance the cyclability of Liion batteries. We note that, although this concept has been addressed by many researchers and the cycling stability has been improved by various diverse methods of realizing more stable SEI layers, to date no literature reports have demonstrated, to our knowledge, the direct correlation between the propagation of SEI growth and Si loss, which is shown here to closely correlate with specific capacity fading as the number of battery cycles increases. We have previously shown that template growth of Si nanowire anodes can lead to significant enhancements in specific capacity retention over long Li-ion cycling periods (>1000 cycles) due to stress reduction near the base of the nanowire.10 Here, we have investigated the influence of SEI layer formation on silicon nanowire morphological changes during the first few 100 cycles and have shown that the growth of the SEI layer results in the loss of Si nanowire volume which directly correlates with the reduction in specific capacity retention over the first ∼100 Li ion cycles. The reduction in Si nanowire diameter with cycling due to SEI layer growth is found to be much greater near the base of the nanowire where it attaches to the current collector. Continued formation and dissolution and/or cracking of the SEI layer, possibly associated with small fractures in the Si nanowire surface, may expose fresh Si to the electrolyte near the Si nanowire base, leading to the loss of microscopic bits of Si, and resulting in additional Si material being consumed. We hypothesize that this narrowing in the nanowire width with SEI formation near the current collector may be a result of a more rapid lithiation near the nanowire base, together with the abrupt stress change in this region. In contrast, the SiNW diameters far away from the current collectors show relatively little loss of Si and degradation. We suggest this phenomenon (nonuniform morphological changes of SiNWs with distance from the

■

ASSOCIATED CONTENT

S Supporting Information *

Experimental methods, normalized mass of SEI layer growth, changes of diameter and cross-sectional area before and after cycles, morphology of ∼20 μm long SiNWs (top view) far away from the current collectors, free electron current density and NW impedance, and model for the loss of specific capacity using the loss of Si material from the NWs. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This material is based upon work supported as part of the Nanostructures for Electrical Energy Storage, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DESC0001160. The work was performed, in part, at the Center for Integrated Nanotechnologies, a U.S. Department of Energy, Office of Science user facility. Los Alamos National Laboratory, an affirmative action equal opportunity employer, is operated by Los Alamos National Security, LLC, for the National Nuclear Security Administration of the U.S. Department of Energy under contract DE-AC52-06NA25396. J.H.C. acknowledges support from a start-up fund at the University of Minnesota, Twin Cities.

■

REFERENCES

(1) Boukamp, B. A.; Lesh, G. C.; Huggins, R. A. J. Electrochem. Soc. 1981, 128, 725−729. (2) Chan, C. K.; Peng, H.; Liu, G.; McIlwrath, K.; Zhang, X. F.; Huggins, R. A.; Cui, Y. Nat. Nanotechnol. 2008, 3, 31−35. (3) Cho, J. H.; Li, X.; Picraux, S. T. J. Power Sources 2012, 205, 467− 473. (4) Chan, C. K.; Ruffo, R.; Hong, S. S.; Huggins, R. A.; Cui, Y. J. Power Sources 2009, 189, 34−39. (5) Zhao, K.; Pharr, M.; Wan, Q.; Wang, W. L.; Kaxiras, E.; Vlassak, J. J.; Suoa, Z. J. Electrochem. Soc. 2012, 159, A238−A243. (6) Liu, X. H.; Zhong, L.; Huang, S.; Mao, S. X.; Zhu, T.; Huang, J. Y. ACS Nano 2012, 6, 1522−1531. (7) Ryu, I.; Choi, J. W.; Cui, Y.; Nix, W. D. J. Mech. Phys. Solids 2011, 59, 1717−1730. (8) Hertzberg, B.; Alexeev, A.; Yushin, G. J. Am. Chem. Soc. 2010, 132, 8548−8549. (9) Wu, H.; Chan, G.; Choi, J. W.; Ryu, I.; Yao, Y.; McDowell, M. T.; Lee, S. W.; Jackson, A.; Yang, Y.; Hu, L.; Cui, Y. Nat. Nanotechnol. 2012, 7, 310−315. (10) Cho, J. H.; Picraux, S. T. Nano Lett. 2013, 13, 5740−5747. 3094

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095

Nano Letters

Letter

(11) Verma, P.; Maire, P.; Novak, P. Electrochim. Acta 2010, 55, 6332−6341. (12) Aurbach, D. J. Power Sources 2000, 89, 206−218. (13) Lee, Y. M.; Lee, J. Y.; Shim, H. T.; Lee, J. K.; Park, J. K. J. Electrochem. Soc. 2007, 154, A515−A519. (14) Wu, H.; Zheng, G.; Liu, N.; Carney, T. J.; Yang, Y.; Cui, Y. Nano Lett. 2012, 12, 904−909. (15) Pinson, M. B.; Bazant, M. Z. J. Electrochem. Soc. 2013, 160, A243−A250. (16) Smith, A. J.; Burns, J. C.; Zhao, X.; Xiong, D.; Dahn, J. R. J. Electrochem. Soc. 2011, 158, A447−A452. (17) Yoshida, T.; Takahashi, M.; Morikawa, S.; Ihara, C.; Katsukawa, H.; Shiratsuchi, T.; Yamaki, J. J. Electrochem. Soc. 2006, 153, A576− A582. (18) Liu, X. H.; Wang, J. W.; Huang, S.; Fan, F.; Huang, X.; Liu, Y.; Krylyuk, S.; Yoo, J.; Dayeh, S. A.; Davydov, A. V.; Mao, S. X.; Picraux, S. T.; Zhang, S.; Li, J.; Zhu, T.; Huang, J. Y. Nat. Nanotechnol. 2012, 7, 749−756. (19) Liu, X. H.; Zhang, L. Q.; Zhong, L.; Liu, Y.; Zheng, H.; Wang, J. W.; Cho, J. H.; Dayeh, S. A.; Picraux, S. T.; Sullivan, J. P.; Mao, S. X.; Ye, Z. Z.; Huang, J. Y. Nano Lett. 2011, 11, 2251−2258. (20) Xu, W.; Vegunta, S. S.; Flake, J. C. J. Power Sources 2011, 196, 8583−8589. (21) Chan, C. K.; Ruffo, R.; Hong, S. S.; Cui, Y. J. Power Sources 2009, 189, 1132−1140. ́ J.; Chagnes, A.; Ozanam, F.; (22) Pereira-Nabais, C.; Swiatowska, Gohier, A.; Tran-Van, P.; Cojocaru, C.; Cassir, M.; Marcus, P. Appl. Surf. Sci. 2013, 266, 5−16. (23) Qu, J.; Li, H.; Henry, J. J., Jr.; Martha, S. K.; Dudney, N. J.; Xu, H.; Chi, M.; Lance, M. J.; Mahurin, S. M.; Besmann, T. M.; Dai, S. J. Power Sources 2013, 198, 312−317. (24) Yao, Y.; Liu, N.; McDowell, M. T.; Pasta, M.; Cui, Y. Energy Environ. Sci. 2012, 5, 7927−7930. (25) Deshpande, R.; Verbrugge, M.; Cheng, Y.; Wang, J.; Liuc, P. J. Electrochem. Soc. 2012, 159, A1730−A1738. (26) Wan, W.; Zhang, Q.; Cui, Y.; Wang, E. J. Phys.: Condens. Matter 2010, 22, 415501. (27) Song, J.; Bazant, M. Z. J. Electrochem. Soc. 2013, 160, A15−A24. (28) Netz, A.; Huggins, R. A.; Weppner, W. Ionics 2001, 7, 433−439. (29) Netz, A.; Huggins, R. A. Solid State Ionics 2004, 175, 215−219. (30) Ploehn, H. J.; Ramadass, P.; White, R. E. J. Electrochem. Soc. 2004, 151, A456−A462. (31) Choi, J. W.; McDonough, J.; Jeong, S.; Yoo, J. S.; Chan, C. K.; Cui, Y. Nano Lett. 2010, 10, 1409−1413. (32) Smith, A. J.; Burns, J. C.; Dahn, J. R. Electrochem. Solid-State Lett. 2010, 13, A177−A179. (33) Philippe, B.; Dedryvere, R.; Allouche, J.; Lindgren, F.; Gorgoi, M.; Rensmo, H.; Gonbeau, D.; Edstrom, K. Chem. Mater. 2012, 24, 1107−1115. (34) Krishnan, R.; Lu, T. M.; Koratkar, N. Nano Lett. 2011, 11, 377− 384.

3095

dx.doi.org/10.1021/nl500130e | Nano Lett. 2014, 14, 3088−3095