A pomegranate-inspired nanoscale design for large-volume-change lithium battery anodes

Transcription

1 SUPPLEMENTARY INFORMATION DOI: /NNANO A pomegranate-inspired nanoscale design for large-volume-change lithium battery anodes Nian Liu, Zhenda Lu, Jie Zhao, Matthew T. McDowell, Hyun-Wook Lee, Wenting Zhao, and Yi Cui* NATURE NANOTECHNOLOGY 1

2 Materials and Methods Synthesis of 2 nanoparticles. In a typical synthesis, 400 mg Si nanoparticles (SiNPs, ~80 nm, MTI, Inc.) were first dispersed in a mixture of 320 ml ethanol and 80 ml water by ultrasonication, followed by the addition of concentrated ammonium hydroxide (4.0 ml). Under vigorous stirring, 1.6 ml tetraethoxysilane (TEOS, Aldrich) was added into the dispersion and the reaction was left at room temperature under stirring overnight. SiNPs coated by SiO 2 (Si@SiO 2 ) were collected by centrifugation and washed three times using water. The resulting Si@SiO 2 powder was re-dispersed in 20 ml water (~2 wt% of Si in water). The cost of fabrication could potentially be further reduced by using rice husks 1 as the source of nano-si and replacing tetraethoxysilane with sodium silicate 2,3. It can be even simpler by using partial thermal oxidation to introduce the SiO 2 sacrificial layer, thus avoiding tetraethoxysilane. Microemulsion-based assembly of Si@SiO 2 nanoparticles into clusters. The obtained water dispersion of Si@SiO 2 (4 ml) was mixed with 16 ml 1-octadecene (ODE, Aldrich) solution containing 0.3 wt% of emulsion stabilizer (amphiphilic block copolymer, Hypermer 2524, Croda USA) and homogenized at 7000 rpm for 1 min. The mixture was then heated at 95~98 o C for 2 h. After evaporation of water, the Si@SiO 2 nanoparticle clusters were collected by centrifugation, and washed with cyclohexane once. The final powder was calcined at 550 o C for 1 h in air to remove the organics and condense the SiO 2. Carbon coating on Clusters. 100 mg Si@SiO 2 clusters were dispersed in 30 ml water, and they were then mixed with 1 ml cetyl trimethylammonium bromide (CTAB, Aldrich, 10 mm) and 0.1 ml ammonia (NH 3 H 2 O, Aldrich, 28%) and vigorously stirred for 20 min to ensure the adsorption of CTAB on the cluster surface. Next, 40 mg resorcinol (Aldrich) and 56 μl formaldehyde solution (Aldrich, 37% wt% in H 2 O) were added and stirred overnight 4,5. To control the thikness of the resorcinolformaldehyde resin (RF) layer, different amount of resorcinol from 10 mg to 100 mg were added. X in RF-X in the paper denotes X mg resorcinol was used in the coating process. The final RF coated cluster were collected by centrifugation and washed with ethanol three times. The RF shell was carbonized under Ar at 800 o C for 2 h with a heating rate of 5 o C min -1. Silica layer etching. SiO 2 sacrificial layer was removed with HF solution to form void space for accommodating the large volume change of Si material during the charge/discharge process. Cluster@RF powders were immersed in 5 wt% HF aqueous solution for 30 minutes, followed by centrifugation and ethanol washing three times. The final Si pomegranate powders were obtained after drying in a vacuum oven. Characterization. The weight percentage of Si and C in the pomegranate secondary particles was determined from the weight loss curves measured under simulated air atmosphere (20% O % Ar, both are ultra purity grade gases from Airgas) on a TG/DTA instrument (Netzsch STA 449) with a heating rate of 5 o C/min. Si is slightly oxidized and causes mass increase, while carbon is oxidized to S2

3 gaseous species and causes mass decrease. To accurately determine the Si content, a pure Si control sample was measured at the same heating conditions and was subtracted from all the curves. Then Si content in the composite was determined by the lowest point of the subtracted curve. Other characterization was carried out using scanning electron microscopy (FEI Sirion), transmission electron microscopy (FEI Tecnai G2 F20 X-twin), X-ray photoelectron spectroscopy (PHI Versa Probe 5000, Physical Electronics, USA), and X-ray diffraction (PANalytical X Pert, Ni-filtered Cu Kα radiation). The cross-sectioned images of Si pomegranate structure after cycling were obtained using a FEI Strata 235B dual-beam SEM/FIB system that combined high-resolution SEM imaging and focused ion beam (FIB) milling. To expose the interior of a secondary particle, a focused gallium ion beam was used to carry out vertical dissection at the desired locations. In situ TEM characterization. In-situ TEM experiments were carried out using a specialized dualprobe electrical biasing holder (Nanofactory Instruments, AB). The movable metallic probes act as current collectors for a nanoscale electrochemical cell that is operated inside the TEM. One probe was a sharpened copper wire (the working electrode) that had the Si pomegranate structures drop-cast onto it. In some cases a thin carbon coating was added to the copper probe for better adhesion of the pomegranates. The second probe was a tungsten wire (the counter electrode) that had a piece of Li metal with a thin (~100 nm) oxide/nitride layer on the surface. The Li metal is the Li source, and the oxide/nitride layer acts as a solid electrolyte layer. Inside the TEM, the probes are positioned so that a single pomegranate structure is in contact with the Li electrode. By biasing the working electrode between -2.5 and -3 V versus the counter electrode, Li + ions flow through the oxide/nitride layer and are reduced at the working electrode, where they react with carbon and alloy with the Si in the pomegranate structure. Li diffuses through the carbon network to react with the Si, as seen in previous studies 6. Maintaining this voltage for about 30 min allows for the full lithiation of the pomegranate structure. Electrochemical characterization. Si pomegranate powders, carbon black (Super P, TIMCAL, Switzerland), and polyvinylidene difluoride binder (PVDF, Kynar HSV 900) with a mass ratio of 80:10:10 were mixed and stirred in N-methylpyrrolidone (NMP) overnight. The slurry was then cast onto a thin copper foil and dried. Prior to cell fabrication, the electrodes were degassed in a vacuum oven at 100 o C for at least 4 hr. Coin-type cells (2032) were fabricated inside an Ar-filled glovebox using Li metal foil as counter/reference electrode, along with a Celgard 2250 separator. The electrolyte employed was 1.0 M LiPF 6 in 1:1 w/w ethylene carbonate/diethyl carbonate (BASF) with 1 vol % vinylene carbonate (Novolyte Technologies) added to improve the cycling stability. Galvanostatic cycling was performed using a BioLogic VMP3 system or an MTI 8 Channels battery tester ( ma). If not mentioned otherwise, the galvanostatic voltage cutoffs were 0.01 and 1 V vs Li/Li +. The specific capacity was calculated based on the total mass of the Si pomegranate composite. The charge/discharge rate was calculated with respect to the theoretical capacity of Si (4200 mah g -1, 1C = 4200 ma g -1 ). The Coulombic efficiency was calculated as C dealloy /C alloy, where C dealloy and C alloy are the capacity of the anodes during Li extraction and insertion. To characterize the electrode after cycling, cells were charged to 1 V and opened. The Si pomegranate electrodes were washed in acetonitrile to remove the electrolyte while keeping the SEI. If S3

4 needed, the SEI layer was removed by soaking in 0.5 M HCl solution, and followed by rinsing with DI water 3 times. High mass loading cells shown in Fig. 4c in main text. To prepare high mass loading electrodes, Si pomegranate microbeads and CNTs (mass ratio 7:3) were first dispersed in NMP and vacuum filtrated with a nylon filter paper (pore size 0.45 μm). The obtained film was then peeled off from the filter paper and sintered at 500 o C for 1h under Ar at a ramp rate of 5 o C min -1. The film was cut into desired size for electrochemical test. The density of such electrodes is ~0.4 g cm -3. The film thickness and mass loading can be easily tuned by changing the starting dry mass of Si pomegranate and CNTs. The thickest electrode we have prepared is ~120 μm, with Si pomegranate areal loading of 3.12 mg cm -2. Coin-type cells (2032) were fabricated inside an Ar-filled glovebox using Li metal foil as counter/reference electrode and a Celgard 2250 separator. The electrolyte employed was 1.0 M LiPF 6 in 1:1 w/w ethylene carbonate/diethyl carbonate with 1 vol % vinylene carbonate and 10 vol % fluoroethylene carbonate added to improve the cycling stability. S4

5 Supporting Figures and Discussions Supplementary Fig. S1. Schematics (a) and pictures (b) of free Si nanoparticles (~80 nm), small secondary particles (1~2 μm), and large secondary particles (~5 μm) of the same Si nanoparticles. Each vial contains 0.4 g of tightly packed powders. The secondary particles were made with similar microemulsion methods described in the Methods section. The size of the secondary particles was tuned by changing the Si concentration in the water dispersion. S5

6 Supplementary Fig. S2. Optical micrographs of (a) 2 nanoparticles in water droplets in ODE, and (b) 2 clusters after water evaporation. The yellow color is due to the nano-sized Si primary particles. Microemulsion is the key step to produce spherical secondary particles. The surface hydrophilicity of 2 nanoparticles keeps it inside the water droplet, so that solid 2 clusters were obtained after water evaporation. If the nanoparticles stayed only at the water-oil interface, hollow microspheres would have been obtained instead. Supplementary Fig. S3. Typical SEM images of microemulsion-derived clusters of bare SiNPs. The diameters of the clusters range from 500 nm to 10 μm. All the clusters show highly spherical morphology. S6

7 Supplementary Fig. S4. (a-h) Centrifugation-separated clusters of bare SiNPs, with average diameters of 0.71 μm (a,e), 1.11 μm (b,f), 1.51 μm (c,g), and 2.53 μm (d,h). The highly spherical morphology remains after centrifugation. (i) Statistical analysis of the cluster diameter. S7

8 Supplementary Fig. S5. Optical micrograph of monodisperse Si pomegranate microbeads in NMP. The conformal carbon coating makes the microbeads highly dispersible in NMP. The yellow color is characteristic for the nano-sized Si primary particles inside the microbeads. S8

9 Supplementary Fig. S6. TEM images of the carbon framework after etching Si with NaOH. (a,b) RF100; (c,d) RF-40; (e,f) RF-20. At the top is the schematic showing the etching process. The thickness of the carbon layer is represented by t. The carbon forms a conformal coating at the surface and a continuous framework in the interior. Interestingly, carbon was present at the interior of the microbeads as well (Fig. 2e,f in main text), acting as a mechanically robust conductive framework. This is due to the progressive RF polymerization in solution5 ; resorcinol and formaldehyde diffuse into the inner space of the microbeads through packing gaps and form RF polymer coating both outside and inside. As a result of this mechanism, the carbon coating is thicker at the surface than in the middle. This is beneficial to its functions because the thick carbon at the surface fills the packing gap and blocks the electrolyte, while the interior only requires thin carbon to conduct Li and e- while keeping the Si content in the composite as high as possible. When the carbon thickness is less than 2.5 nm (e,f), the framework becomes less continuous and is not sturdy enough to support the secondary particles. S9

10 Supplementary Fig. S7. Schematic and SEM images of the Si pomegranates after (a) NaOH etching or, (b) 5 min intense sonication in ethanol. In the magnified SEM image at the bottom, it can be seen that the carbon shell is intact and the same as before etching or sonication. The robust carbon framework is important in stabilizing the secondary particle morphology while the primary nanoparticles inside go through drastic volume changes. S10

11 Supplementary Fig. S8. Zoomed-in CE plot for three cells shown in Fig. 4a in main text. The fact that Si pomegranate anode has higher CE than SiNP anode indicates that the electrolyte is not in direct contact with Si surface in Si pomegranate, otherwise it will behave like SiNPs. Another fact that Si pomegranate anode has higher CE than Si anode indicates that carbon is not only functioning as a conducting framework, but also an electrolyte-blocking layer (with the help of SEI outside). S11

12 Supplementary Fig. S9. Cycle performance of slurry-coated, calendered Si pomegranate anodes, with (a) 5% CNT or (b) no CNT. S12

13 Discussion on the infiltration of electrolyte The carbon shell in Si pomegranate may contain some micropores (smaller than could be identified under TEM) allowing small molecules to pass through. That allows SiO 2 etching to happen in the last synthesis step. We also notice that SiO 2 etching and SEI formation (electrolyte decomposition) has different mechanism as summarized in this table: Reactant Product SiO 2 etching Solid phase Liquid phase SEI formation (Electrolyte decomposition) Liquid phase Solid phase The products in SiO 2 etching are all soluble species and etching will proceed until all the solid SiO 2 is dissolved. During SEI formation, however, liquid phase reactants become solid. Moreover, the micropores are smaller than 2 nm while the thickness of the SEI is ~150 nm (Fig. 4d in main text). We therefore expect the micropores to be sealed by SEI after initial cycling. Also, with the expansion of Si, even if the electrolyte might initially infiltrate inside, it should be squeezed out and stay outside the particle after the SEI formation outside the particle. We carried out two additional experiments to further prove this point (results shown below). Supplementary Fig. S10. Dual-beam Focused Ion Beam (FIB) analysis and cross-sectioned SEM images of a Si pomegranate after 1000 deep cycles (delithiated). The FIB cutting is vertical, and the SEM images were taken from a 52 o tilted position. The yellow arrows in (b) indicate the void spaces inside the Si pomegranate. Most of the SEI forms outside the secondary particle. S13

14 Supplementary Fig. S11. Auger electron spectroscopy (AES) elemental mapping of the interior of cycled Si pomegranate. The top surface of the sample was removed by Ar ion beam sputtering (5 kv, 5 μa) for 4 minutes. The ion beam is not vertical so directional scoop channels were left on the particle as seen in the SEM image (lower left). Fluorine signal is used to identify the SEI, because the electrolyte contains LiPF 6 and fluoroethylene carbonate. The majority of F is outside the secondary particle. The closer it is to the middle, the less F there is. Pixel intensity along the vertical line is integrated and plotted as the yellow curve. It can be seen in the cross section images in Fig. S10 that the void spaces inside Si pomegranate still remain after 1000 deep cycles, which indicates most electrolyte is effectively blocked outside the secondary particle. The fluorine element mapping in Fig. S11 also indicates that the infiltration of electrolyte is limited. Moreover, the superior cycling performance over bare SiNPs and Si cluster@c (no void space) structures is another strong evidence for this mechanism (Fig. 4a in main text). If the electrolyte can freely infiltrate into carbon shell during cycling, the performance should be the same as bare SiNPs. And if one argues that carbon shell only improves the conductivity, Si cluster@c (no void space) should have better performance than Si pomegranate because of more intimate contact between Si and carbon. Si pomegranate's performance of 1000 cycles with 97% capacity retention (0.003% capacity decay per cycle) is even comparable to that of double-walled silicon nanotube (DWSiNT) (ref. 20 in manuscript, 6000 cycles with 85% capacity retention, % capacity decay per cycle), which is the most stable Si anode to the best of our knowledge. We believe the electrolyte blocking effect (spatially confined SEI formation) is one of the keys that enables the superior performance. S14

15 Supplementary Fig. S12. (a) First cycle voltage profiles of Si pomegranate with different carbon coating thicknesses. RF-0 denotes bare SiNPs. For ease of comparison, all the curves are normalized by delithiation capacity. The delithiation profiles almost overlap for all the curves. And the delithiation capacities are close to the lithiation capacities at the plateau below 0.1 V, corresponding to the reversible capacity contributed by Si. Therefore, the majority of the irreversible capacity is from the initial sloping part of the lithiation profile, corresponding to the lithiation of amorphous carbon and formation of SEI. (b) First cycle Coulombic efficiency (CE) comparison. Delithiation to a higher S15

16 potential increases the CE. (c) Relationship between first cycle CE and carbon content in the Si pomegranate. The dotted lines are merely for guidance of the eye. The carbon content was determined by TGA as shown in Fig. 2h in the main text. (d,e) First cycle delithiation capacity comparison and its relationship with carbon content. All the capacities are with respect to the total mass of Si and C in the pomegranate structure. The Si fraction in the composite is determined to be 60~91% by TGA (Fig. 2h in the main text). Thinner carbon gives lower carbon content in the composite, and increases the specific capacity. Thinner carbon also increases the first cycle CE, because amorphous carbon has a certain percentage of dangling bonds that react irreversibly with Li at low potential. RF-20 (The number 20 denotes the amount of added resorcinol in the polymer coating step, see Materials and Methods section above) has the minimum amount of carbon (9%) needed to support the secondary particle morphology. Its reversible capacity reaches 2465 mah g -1 with respect to the total mass of Si and C in pomegranate structure, while the first cycle CE reaches 82% when delithiated to 2 V. RF-40, containing 23% carbon, has better coverage of carbon. It gives 2285 mah g -1 capacity and 75% initial CE. Supplementary Fig. S13. XRD pattern of the Si pomegranates. All the peaks are attributed to crystalline Si. The broad peak between 20 and 25 o can be attributed to the amorphous carbon framework in the Si pomegranate structures. S16

17 Supplementary Fig. S14. Specific capacity (a) and voltage profiles (b) of the Si pomegranate anode cycled at various rates from C/50 to 2C in the potential window of 0.01 to 1 V versus Li/Li +. 1C = 4.2 A g -1 Si. S17

18 Supporting Movies Supplementary Movie S1. Lithiation of a Si pomegranate secondary particle with insufficient gap size (~15 nm) between Si and carbon. The expansion of Si upon lithiation ruptures the outer carbon shell. The movie is played at 5x speed. Supplementary Movie S2. Lithiation of a Si pomegranate secondary particle with sufficient gap size (~40 nm) between Si and carbon. The SiNPs expand towards the void spaces and the secondary particle morphology is stable upon lithiation. The movie is played at 25x speed. References 1. Liu, N., Huo, K., McDowell, M. T., Zhao, J. & Cui, Y. Rice husks as a sustainable source of nanostructured silicon for high performance Li-ion battery anodes. Sci. Rep. 3, 1919 (2013). 2. Kim, J. M. & Stucky, G. D. Synthesis of highly ordered mesoporous silica materials using sodium silicate and amphiphilic block copolymers. Chem. Commun (2000). doi: /b002362k 3. Corriu, R. J. P., Mehdi, A., Reye, C. & Thieuleux, C. Direct syntheses of functionalized mesostructured silica by using an inexpensive silica source. Chem. Commun (2004). doi: /b314415a 4. Liu, J. et al. Extension of The Stöber Method to the Preparation of Monodisperse Resorcinol Formaldehyde Resin Polymer and Carbon Spheres. Angew. Chemie Int. Ed. 50, (2011). 5. Li, N. et al. Sol-gel coating of inorganic nanostructures with resorcinol-formaldehyde resin. Chem. Commun. 49, (2013). 6. Liu, N. et al. A yolk-shell design for stabilized and scalable Li-ion battery alloy anodes. Nano Lett. 12, (2012). S18