A Step toward High-Energy Silicon-Based Thin ... - ACS Publications

Apr 24, 2017 - Institute of Physical Chemistry, MEET Battery Research Center, University of Münster, Corrensstr. 46, 48149 Münster, Germany. ‡. Helmho...
0 downloads 0 Views 11MB Size
A Step toward High-Energy Silicon-Based Thin Film Lithium Ion Batteries Antonia Reyes Jiménez,† Richard Klöpsch,† Ralf Wagner,† Uta C. Rodehorst,† Martin Kolek,† Roman Nölle,† Martin Winter,*,†,‡ and Tobias Placke*,† †

Institute of Physical Chemistry, MEET Battery Research Center, University of Münster, Corrensstr. 46, 48149 Münster, Germany Helmholtz Institute Münster, IEK-12, Forschungszentrum Jülich GmbH, Corrensstr. 46, 48149 Münster, Germany



S Supporting Information *

ABSTRACT: The next generation of lithium ion batteries (LIBs) with increased energy density for large-scale applications, such as electric mobility, and also for small electronic devices, such as microbatteries and on-chip batteries, requires advanced electrode active materials with enhanced specific and volumetric capacities. In this regard, silicon as anode material has attracted much attention due to its high specific capacity. However, the enormous volume changes during lithiation/delithiation are still a main obstacle avoiding the broad commercial use of Si-based electrodes. In this work, Si-based thin film electrodes, prepared by magnetron sputtering, are studied. Herein, we present a sophisticated surface design and electrode structure modification by amorphous carbon layers to increase the mechanical integrity and, thus, the electrochemical performance. Therefore, the influence of amorphous C thin film layers, either deposited on top (C/Si) or incorporated between the amorphous Si thin film layers (Si/C/Si), was characterized according to their physical and electrochemical properties. The thin film electrodes were thoroughly studied by means of electrochemical impedance spectroscopy, Raman spectroscopy, X-ray photoelectron spectroscopy, and atomic force microscopy. We can show that the silicon thin film electrodes with an amorphous C layer showed a remarkably improved electrochemical performance in terms of capacity retention and Coulombic efficiency. The C layer is able to mitigate the mechanical stress during lithiation of the Si thin film by buffering the volume changes and to reduce the loss of active lithium during solid electrolyte interphase formation and cycling. KEYWORDS: lithium ion batteries, magnetron sputtering, anode, carbon/silicon, multilayer structure, thin films mAh g−1 corresponding to crystalline Li15Si4.11 Within the highcapacity “lithium alloying” materials,12 including silicon (Si), tin, and germanium, Si is currently considered as the most promising material to replace graphite as anode material. This is not only due to its high gravimetric and volumetric capacity but also due to the high abundance and low costs.11,13,14 In addition, Si displays a relatively low operating potential of ≈0.4 V vs Li/Li+, which is important to enhance the cell voltage, as well as a relatively high energy efficiency,8 which is displayed by the low voltage hysteresis between lithiation and delithiation. However, Si still suffers from severe drawbacks hindering its commercial breakthrough so far, which are most likely related to the large volume changes (up to 300%) during lithiation/ delithiation and its low intrinsic electronic conductivity.11,15 As a result, the large stress and strain leads to particle cracking and pulverization, mechanical cracking of the composite electrode,

T

he next generation of lithium ion batteries (LIBs) applied in energy storage either for large-scale applications, such as automotive batteries or stationary storage, and for small electric devices, such as medical implants or microsensors, needs further development of new active anode and cathode materials with higher specific and volumetric capacities. In turn, it will be possible to increase the gravimetric and volumetric energy and power densities of LIBs. Besides the need for improved positive electrode materials, the required energy target value at cell level can only be achieved with novel anode materials delivering a high specific capacity as well as a sufficiently high cycling and rate performance.1−4 Currently, graphite is still the state-of-the-art anode material in commercial LIBs due to its outstanding dimensional stability, low operating potential of 0.2 V vs Li/Li+,5 high but surfacearea-dependent Coulombic efficiency,6,7 high energy efficiency,8 low costs, and nontoxicity.9 However, its theoretical specific capacity of only 372 mAh g−110 is low compared to the one of silicon, which has a theoretical specific capacity of 3579 © 2017 American Chemical Society

Received: February 9, 2017 Accepted: April 24, 2017 Published: April 24, 2017 4731

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

www.acsnano.org

Article

ACS Nano

Figure 1. Schematic illustration of (a) all-solid-state (micro-) battery and (b) magnetron sputtering process.

the volume expansion of silicon.29,30 In addition, the design of multiphase composite materials is a promising approach, in which a second or more components are introduced, which exhibit no or minor volume changes and preferentially a high electronic conductivity.24 Examples are C/Si composite materials,31 Si/SiOx/C composite materials,32,33 or intermetallic compounds such as Si/TiSi234 or NiSi2/Si,35−37 as well as the implementation of conductive polymers.38 Besides the modification of Si nanoparticles, another approach to enhance the cycle life is the preparation of Sibased thin film layers (typically 50−500 nm).39−48 Thin film electrodes are of special interest for usage in microbatteries powering miniaturized electronic devices, for example, for medical applications.2,49 A schematic illustration of a possible setup of an all-solid-state (micro-) battery is depicted in Figure 1a, in which a solid electrolyte can be used between the positive and negative thin film electrodes. As these thin film electrodes also suffer from the large volume changes, there are several strategies for improving the performance including doping of Si thin films,50 usage of binary Si thin film systems such as Mg−Si,46 Ni−Si,51 or Si−Al,52

and disconnection of the active Si particles from the electronically conductive network.15,16 Furthermore, in contrast to graphitic carbon anodes, where the electrolyte decomposition and corresponding solid electrolyte interphase (SEI) formation are mainly limited to the initial cycles,17−19 for lithium alloying (e.g., Si-based anodes), it is a dynamic process of breaking off and re-forming, due to the constant structural changes as well as changes in the electrode thickness and porosity during cycling.15,20−24 This leads to a huge irreversible capacity as well as high active lithium loss and continuous increase in interfacial resistance upon charge/discharge cycling, which will result in rapid capacity fading.15,25 In the past years, several concepts and strategies to reduce the detrimental effects of the volume expansion/shrinkage and the ongoing side reactions with respect to electrolyte degradation have been reported. One major strategy is the reduction of the Si particle size to the nanometer range, whereas various structures such as nanowires, nanospheres or nanotubes have been suggested.26−28 A further main strategy related to the performance improvement is the design of porous structures that can offer sufficient void space to absorb 4732

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

ACS Nano

Figure 2. Schematic illustration of (a) carbon/silicon (C/Si) and (b) silicon/carbon/silicon (Si/C/Si) thin film electrodes.

structuring of electrodes,45,49,53,54 surface modifications or coatings,42,47,55 as well as multilayer systems.56−58 In this work, the focus is set on the manufacturing of Sibased thin film electrodes as well as their designed surface and structure modification by amorphous carbon layers to increase the mechanical stability and, thus, the electrochemical performance. The Si thin film electrodes are prepared by magnetron sputtering, as illustrated in Figure 1b. In the process of sputtering, a target is bombarded by highly energetic ions, which are created in a glow plasma. This process causes the removal of target atoms, which then are deposited on a substrate as a thin film layer.59 Compared to bulk silicon anodes, which contain at least 10 wt % of inactive materials, such as binders and conductive additives, Si thin film anodes contain no inactive materials, thus delivering an increased specific capacity.60 In addition, binder-free electrodes can be considered as good model electrodes for basic active material studies, as binders have a well-known influence on electrochemical performance.61−63 In this work, the influence of amorphous carbon thin film layers, either deposited on top (C/Si) or incorporated between the amorphous Si thin film layers (Si/C/Si), on the electrochemical characteristics, including the specific capacity, Coulombic efficiency, and cycling stability, is systematically investigated. In particular, the impact of the thickness of the carbon layer on these characteristics is studied. Furthermore, a major part of this work is attributed to the characterization of the carbon/silicon multilayer thin film electrodes in terms of electronic conductivity as well as mechanical stability after charge/discharge cycling.

RESULTS AND DISCUSSION Structural Characterization of Si, C, C/Si, and Si/C/Si Thin Films. Next to pure Si and pure C thin film electrodes, two different types of electrodes were prepared in order to study the influence of an amorphous carbon layer on the electrochemical performance, as schematically depicted in Figure 2. The first approach is motivated by the protection of the susceptible Si electrode/electrolyte interface by deposition of a protective, because less electrolyte-reactive, amorphous C layer on top of the Si layer, as shown in Figure 2a. The thickness of the C layers deposited on the Si electrode was set to 5, 10, and 50 nm, which corresponds to an amount of 3.5, 6.7, and 26.3 wt % of C, respectively, related to the active electrode mass. The second approach is motivated by the reduction of electrode degradation processes due to the mechanical stress induced by the large volume expansion/ contraction of Si during lithiation/delithiation, respectively. Furthermore, pure Si electrodes only have a low electronic conductivity, which limits their performance at increased charge/discharge current rates. Therefore, a multilayer Si/C/ Si system is prepared to buffer the volume expansion and to improve the electronic conductivity throughout this multilayer electrode stack. The multilayer system is composed of 70 nm Si/x nm C/70 nm Si, where the carbon thickness was varied (x = 5, 10 and 50 nm) (Figure 2b). In order to prove the successful preparation of the Si, C, and C/Si thin film electrodes, the pristine electrodes were investigated by means of Raman spectroscopy. Figure 3 shows the Raman spectrum of pure Si (140 nm) and pure C (200 nm) thin film electrodes as well as a C/Si (50/140 nm) electrode. The pure Si electrode shows four bands, correspond4733

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

ACS Nano

Figure 4. XPS depth profiling of a pristine multilayer Si/C/Si (70 nm/50 nm/70 nm) thin film electrode.

Figure 3. Raman spectra of the pristine Si (140 nm), C (200 nm), and C/Si (50/140 nm) electrodes.

lithiation/delithiation processes, the irreversible electrolyte reduction and SEI formation, as well as the Li+ ion transport properties. In Figure 5, the first three cycles of a CV measurement at a scan rate of 0.025 mV s−1 of a pure Si thin film (a), C/Si (b), and Si/C/Si (c) electrodes are depicted. The observed broad cathodic peaks in the potential region from

ing to four different phonon bands. The transverse acoustic (TA) at 150 cm−1, the longitudinal acoustic (LA) at 300 cm−1, the longitudinal optic (LO) at 400 cm−1, and transverse optic (TO) at 480 cm−1 are the typical vibrational modes of amorphous Si. Furthermore, the spectrum does not show any feature at 520 cm−1 arising from crystalline Si, thus indicating a solely amorphous phase.46,64 In the Raman spectrum of the C/Si electrode, the intensity of the aforementioned silicon-typical bands is diminished and two new bands arise, thus proving the existence of an additional layer of carbon on the Si surface. The typical D band at 1360 cm−1 and the typical G band at 1580 cm−1 show that this surface layer is composed of amorphous carbon.65 In addition, the presence of the D band and strong overlapping of these two bands for both the pure C electrode and the C/Si electrode verify the amorphous structure of carbon.66 It should be noted that the amorphous carbon in the layer is different in structure than carbon black, with the usually used conductive additive in lithium ion battery electrodes.67,68 To prove the successful preparation of the Si/C/Si electrode (see Figure 2b), the pristine electrode (70 nm/50 nm/70 nm) was investigated by means of X-ray photoelectron spectroscopy (XPS) depth profiling. Figure 4 shows the distribution of the atomic concentration of Cu 2p, O 1s, C 1s, and Si 2p over a sputtering time of 2000 s. As the elements within the electrode have different etching rates and are due to the fact that in each sputtering step the electrode material is not removed uniformly, the depth profile exhibits broad atomic distributions rather than sharp interfaces. The surface of the electrode is covered by SiO2 as seen in the high atomic concentrations of Si 2p and O 1s, that is ∼40 atom % for Si and ∼30 atom % for O. With further sputtering time, the O 1s signal strongly decreases to below 10 aton % after ca. 100 s. The course of the C 1s concentration as a function of the sputtering time evidences the successful existence of the multilayer structure of the Si/C/Si electrode. After ≈1400 s of sputtering, the copper current collector can be observed as the Cu 2p signal increases to almost 100 atom %. Determination of the Li+ Ion Transport Properties of C/Si and Si/C/Si Thin Film Electrodes. Cyclic voltammetry (CV) was used to study the influence of the additional carbon layer of the Si thin film anodes on the reversibility of the

Figure 5. Cyclic voltammograms of the first three cycles of (a) Si (140 nm), (b) C/Si (50 nm/140 nm), and (c) Si/C/Si (70 nm/50 nm/70 nm) thin film electrodes, measured in a half cell setup vs metallic lithium. Scan rate = 0.025 mV s−1; potential range = 1.50 to 0.04 V vs Li/Li+. For better comparability, horizontal lines are added at 0.025 and −0.030 mA cm−2. 4734

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

ACS Nano Table 1. Overview of the Resistance and Electronic Conductivity of the Different Thin Film Electrodes thickness/nm resistance/Ω electronic conductivity/mS cm−1

Si

C

C/Si

Si/C/Si

140 0.3078 ± 5.0 × 10−5 6.60 ± 0.01

200 0.2586 ± 1.4 × 10−3 7.86 ± 0.04

50/140 0.2563 ± 1.1 × 10−3 7.93 ± 0.03

70/50/70 0.2578 ± 2.1 × 10−3 7.89 ± 0.06

0.40 to 0.04 V vs Li/Li+ in all CV measurements are attributed to the formation of Si−Li alloys with different composition.69 The first cathodic peak is attributed to the formation of the P-I phase (Li−50 atom % Si; LiSi) from amorphous Si, which is superimposed with the phase transition between P-I and P-II phases (Li−30 atom % Si; Li7Si3). The second cathodic peak is related to the formation of the P-III phase (Li−24 atom % Si; Li3.16Si) and most likely to the formation of crystalline silicon (for pure Si).70 In the anodic scan, Li+ ion extraction takes place from 0.1 to 0.6 V vs Li/Li+. The observed peaks are attributed to the phase transition between the Si−Li alloys and amorphous Si, while the first peak is attributed to the phase transition between the P-III and P-II phases, and the second peak corresponds to the transition from P-II to P-I phases.16,70 In the case of the pure Si electrode (Figure 5a), the first cycle displays an increased current in the potential region from 1.3 to 0.4 V vs Li/Li+, which can be attributed to the reductive electrolyte decomposition and concomitant formation of the SEI.17,69,71 In the CV measurement of the C/Si thin film electrode (Figure 5b), the observed broad peaks in the potential region from 0.40 to 0.04 V vs Li/Li+ are also attributed to the formation of Si−Li alloys with different compositions as well as to the lithiation of the amorphous carbon. In the anodic scan, Li+ ion extraction from Si and C takes place from 0.2−0.6 V vs Li/Li+. Figure S1 (Supporting Information) shows the CV measurement of the amorphous carbon thin film (thickness: 200 nm) at a scan rate of 0.5 mV s−1. Only a low reversible capacity storage is seen (≈230 mAh g−1). Therefore, the additional capacity of the amorphous carbon was not considered for the calculation of the capacity of amorphous carbon containing Si thin film electrodes. Figure 5c shows the measurement of Si/C/Si multilayer electrode with reduction peaks at 0.26 and 0.08 V vs Li/Li+ as well as oxidation peaks at 0.3 to 0.6 V vs Li/Li+. Overall, it has to be noted that the carbon phase (either on top or between two Si layers) suppresses the crystallinity of Si and the possible formation of the Li15Si4 phase during lithiation. In the case of the pure Si thin film electrode, the peak at ≈40 mV vs Li/Li+ during the cathodic scan (Figure 5a and Figure 8) can be assigned to the formation of the crystalline Li15Si4 phase. In general, this phase has been shown to form during the lithiation of Si when the potential is less than 50 mV.72 However, Hatchard et al. reported that the Li15Si4 phase is not formed in Si thin films with a thickness of less than 2.5 μm, which is not in agreement with our results.73 In the case of the Si thin film electrodes including the additional amorphous C layer, we assume that due to the higher overpotential during lithiation (see Figure 8), this threshold potential value is not reached, and thus, the Li15Si4 phase is not formed. The Coulombic efficiency (CE) of pure Si amounts to 37, 70, and 76% for the first, second, and third cycle, respectively. Furthermore, the CE for the C/Si electrode in the first, second, and third cycle is 47, 76, and 80%, respectively, whereas the Si/ C/Si multilayer system shows a CE of 60, 89, and 91%. By this, a clear trend (i.e., an increase of the initial CE when going from

the pure Si electrode to the multilayer Si/C/Si system) can be observed. CV measurements were also performed at different scan rates ranging from 0.025 to 1.000 mV s−1 to investigate the Li+ ion diffusion coefficient of the different Si-based electrodes with and without the amorphous C thin film layer, as described in detail in the literature.39 The measurements for the three different electrodes were each performed in a single cell at different scan rates. At each scan rate, three cycles were conducted. The second cycle was chosen for the assessment of the Li+ ion diffusion coefficient. The CV measurements of the second cycle at selected scan rates are depicted in Figure S2 (Supporting Information). In all cells, the specific peak current increases with increasing scan rate, and the peak maxima shift to lower potential values in the cathodic scan and to higher potentials in the anodic scan due to the presence of increased overpotentials. For semi-infinite diffusion, the peak current (ip) is proportional to the applied scan rate (v) which is expressed by the Randles-Sevcik equation:74,75 i p = (2.69 × 105)n3/2AD01/2ν1/2C0

In this equation, n is the number of electrons (=1e−) transferred in the Faradaic reaction, A is the surface area of the electrode (≈1.13 cm2), D0 is the Li+ ion diffusion coefficient, v is the scan rate, and C0 is the bulk concentration of Li+ ions in the electrode (≈0.15 mol/cm3). From the slope of ip versus V1/2, it is possible to calculate the D0 of investigated electrodes, as shown in Figure S3 (Supporting Information). The calculated D0 for pure Si, C/ Si, and the multilayer Si/C/Si system, based on the data obtained from the CV measurement at the scan rate of 0.025 mV s−1, amounts to 1.41 × 10−10, 6.73 × 10−11, and 9.74 · 10−11 cm2 s−1, respectively. Therefore, the Li+ ion diffusion is the highest in the pure silicon electrode, whereas lower Li+ ion diffusion coefficients are obtained for the silicon−carbon electrodes. This trend can also be observed at various scan rates (see Table S1, Supporting Information). On one hand, the presence of the additional carbon layer slows down the Li+ ion diffusion. On the other hand, the amorphous carbon layer leads to better structural integrity of the Si electrode. The existence of cracks within the Si electrode leads to additional surface area accessible to the electrolyte, thus resulting in faster Li+ ion diffusion due to shorter Li+ ion diffusion distances. It is assumed that both effects contribute to the higher diffusion coefficient of the pure Si electrode. Determination of the Surface Topography and Electron Transport Properties of C/Si and Si/C/Si Thin Film Electrodes. The electronic conductivity of the different electrodes was investigated by means of electrochemical impedance spectroscopy (EIS). The electronic conductivity values were determined by fitting from the impedance data. Table 1 shows the obtained values of the electronic conductivity of investigated electrodes with their corresponding film thicknesses. The results prove that all electrodes containing carbon, such as C/Si and Si/C/Si (both ≈7.9 mS cm−1), display an increased electronic conductivity compared to pure 4735

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

ACS Nano

Figure 6. AFM topographic (left side) and current images (right side) of a 20 μm × 20 μm area of pristine (a,b) Si (140 nm), (c,d) C (200 nm), (e,f) C/Si (50 nm/140 nm), and (g,h) Si/C/Si (70 nm/50 nm/70 nm) thin film electrodes.

silicon (≈6.6 mS cm−1). Atomic force microscopy (AFM) was further used to study the surface topography of the different

electrodes and to investigate their local electronic conductivity (current sensing AFM). Figure 6 shows the topography (left 4736

DOI: 10.1021/acsnano.7b00922 ACS Nano 2017, 11, 4731−4744

Article

ACS Nano side) and the current distribution (right side) of pure Si (140 nm), pure C (200 nm), C/Si (50 nm/140 nm), and multilayer Si/C/Si (70/50/70 nm). The observed surface inhomogeneity (±150 nm) arises from the used copper foil (see Figures S4−S6 in the Supporting Information). In theory, the local electronic conductivity (σ) is described by the following equation: σ=

I h U A

where σ is proportional to the current (I) and the electrode thickness (h) as well as reverse proportional to the applied voltage (U) and locally investigated electrode’s surface area (A). For the measurement of the current distribution, the samples were electronically connected to the AFM system, and a voltage bias was applied to the sample. The voltage bias was adjusted to 1.5 V for the pristine silicon and the Si/C/Si electrodes as well as to 0.5 V for the pure C and the C/Si electrodes, to stay inside the measurement range of the current distribution. The circuit was closed through a full platinum cantilever and a current sensing nose cone assembly. The topography of the pure Si electrode correlates with the topography of pristine copper foil (see Figures S4 and S5, Supporting Information), showing a maximum peak height of 350 nm (Figure 6a). As seen in Figure 6b, the pure silicon electrode displays an inhomogeneous current distribution and low average current value lower than the limit of detection of