measurements. The spectra show electrolyte in contact with the carbon fibre based gas diffusion layer

Transcription

1 Supplementary Figure 1. 7 Li NMR spectra of electrolyte in contact with potential substrates for in-situ NMR measurements. The spectra show electrolyte in contact with the carbon fibre based gas diffusion layer support (CFGDL), SUS304 stainless steel (SS) mesh, SUS304 SS sheet, and SUS316 SS sheet as described in the Methods. 7 Li spectra were acquired with 1500 scans on a Techmag LapNMR (300 MHz) spectrometer at room temperature, applying π/2 pulses of 3 μs. All 7 Li shifts reported in this publication are referenced to a 1 M LiCl solution at 0 ppm. 1

2 20 15 Mean:~60 nm Standard deviation:~40 nm ~60NWs are sampled Count SiNW diameter [nm] Supplementary Figure 2. Histogram of diameter of the as-grown SiNWs on CFGDL. Sixty NWs are analysed by scanning electron microscopy (SEM) images, the average of which is centred at ~60 nm with a standard deviation of 40 nm. 2

3 Supplementary Figure 3. Electrochemistry of SiNWs on stainless steel (SS, SUS304). The batteries in 2032 coin cells are galvanostatically cycled with 2 0 V potential limit at C/25 for 10 cycles, (a) specific capacity for discharge and charge and (b) overlaid voltage profiles as a function of specific capacity. 3

4 Supplementary Figure 4. dq/dv profiles as a function of voltage for SiNWs on SS galvanostatically cycled at C/75. (a) Discharge and (b) charge in 2032 coin cells with potential limits of 0-2 V over 8 cycles.. 4

5 Supplementary Figure 5. dq/dv profiles of reference materials obtained in coin cells as a function of voltage. (a,b) SiNW-CFGDL composite, (c,d) Bare CFGDL, (e.f) Au colloids on SS, (g.h) Si thin-film on SS. All the cells are galvanostatically cycled at C/25 with potential limits of 2 0 V. The left column (a, c, e, g) shows dq/dv profiles from V on discharge. The right column (b,d,f,h) shows dq/dv profiles for V on charge. Note that no visible peaks are observed in the dq/dv plots above 0.5 V on discharge and 0.8 V on charge. For the dq/dv profiles shown here and in all subsequent figures are offset in the y-axis so that the differences between the dq/dv profiles are more clearly seen. 5

6 Supplementary Figure 6. dq/dv profiles of SiNWs on stainless steel (SS) as a function of cut-off voltage. The cells in 2032 type coin cells are galvanostatically cycled at C/25 with gradually decreasing cut-off voltages on discharge from 80 to 40 mv (decreasing by 10 mv steps), followed by holding at the cut-off voltage for 12 hours and charging up to 2 V, (a) discharge (b) charge. 6

7 Supplementary Figure 7. Ex-situ XRD patterns of lithiated-sinws on SS after galvanostatic cycling at C/25. The cells are discharged/charged to different target potentials in the 1 st cycle. The mass of SiNWs used is given in the figure. 7

8 Supplementary Figure 8. Ex-situ XRD patterns of lithiated-sinws on SS after galvanostatic cycling at C/75. The cells are discharged/charged to different target potentials in the 1 st cycle. The mass of SiNWs used is given in the figure. 8

9 Supplementary Figure 9. Ex-situ XRD patterns of lithiated-au on SS after galvanostatic cycling at C/30. The cells are discharged/charged to different target potentials in the 1 st cycle. The mass of Au film and colloid on SS is ~50 µg. 9

10 Supplementary Figure 10. TEM images of cycled SiNWs. Various diameters after (a-d) one cycle and (e-h) two cycles. Prior to the observation, a coin cell battery made of the SiNW-CFGDL composite is cycled exsitu galvanostatically at C/25 with potential limits of 2-0 V. Scale bars are (a) 50, (b) 10, (c,d) 200 nm, (e) 30, (f) 60, (g,h) 100 nm, respectively. 10

11 Supplementary Figure 11. In-situ NMR spectroscopy for bare carbon fibre based gas diffusion support (CFGDL). (a) In-situ 7 Li NMR spectra of bare CFGDL as a function of accumulated specific capacity for galvanostatic cycles at C/25 with potential limits of 2-0 V as described in the Methods. (b) Contour figure associated and associated voltage profile. (c) Stacked 7 Li NMR spectra for the 1 st cycle from 10 to 10 ppm. The spectra in (c) are offset in the y-axis with a constant pitch. 11

12 Supplementary Figure 12. Stacked ex-situ MAS 7 Li NMR spectra for the SiNWs on stainless steel (SS). The cells are galvanostatically discharged/charged at C/75 and held at target voltages. The spectra were acquired with a Hahn echo sequence. The absence of any significant intensity in the spectra obtained at 300 mv on the 1 st discharge and at 2 V at the top of the 2 nd discharge indicates that the electrolyte and the solid electrolyte interphase (SEI) on the SiNWs has been removed by the DMC-wash step to a concentration that is below the NMR detection limit. The spectra show that the Li resonances originating from c-li 3.75 Si on charge (at 200 and 380 mv on the 1 st charge) are centred at 10 0 ppm, and while the peaks originating from the large Si clusters/network (seen at 300 mv on the 2 nd discharge) are centred at 5 0 ppm. 12

13 Supplementary Figure 13. In-situ 7 Li NMR spectra obtained from the SiNW- CFGDL composite battery following galvanostatic cycling (at C/30) between 2V and 0 V (1 st two cycles) and 2V and 50 mv (3 rd cycle), plotted so as to show the higher voltage regions obtained on discharge down to approximately 140 mv. In this Figure and the subsequent three sets of figures, the upper row (a,b,c) focuses on the lowintensity peaks of the 7 Li in-situ NMR spectra in the 30 to 30 ppm region and the lower row (d, e, f) shows the most intense/central peaks in the 5 to 5 ppm region. Each spectrum in this and the subsequent three sets of figures are offset in the y-axis with a constant pitch so that the difference between each spectrum is clearly seen. 13

14 Supplementary Figure 14. Stacked in-situ 7 Li NMR spectra obtained from the same experiment as that shown in supplementary Figure 13, but plotted so as to show the region between 150 mv and 0V on discharge and from 0 to 450 mv on charge. The 7 Li resonances at around ppm, 0 10 ppm and 10 ppm are labelled as P1 (Li nearby small Si clusters), P2 (isolated Si anions, c-li 3.75 Si, and extended Si networks), and P3 (the over-lithiated crystalline phase, c-li 3.75+δ Si), respectively. 14

15 Supplementary Figure 15. Stacked in-situ 7 Li NMR spectra obtained from the same experiment as that shown in supplementary Figure 13, but plotted so as to show the region above 400 mv on charge. 15

16 Supplementary Figure 16. Stacked in-situ 7 Li NMR spectra for potentiostatic cycling from 2 V to 150 mv on discharge. The cell made from the SiNW- CFGDL composite is cycled with a potentiostatic schedule with potential limits of 2-0 V over two cycles. In the 3 rd cycle the potentiostatic discharge was stopped at 30 mv (see Methods). 16

17 Supplementary Figure 17. Stacked in-situ 7 Li NMR spectra obtained from the same experiment as that shown in supplementary Figure 16, but plotted so as to show the spectra obtained for potentiostatic cycling to voltages higher than 400 mv on charge. The upper row (a,b) focuses on the low-intensity peaks of the 7 Li insitu NMR spectra in the 30 to 30 ppm region and the lower row (c, d) shows the central peaks in the 5 to 5 ppm region. 17

18 Supplementary Figure 18. Deconvolved in-situ 7 Li NMR spectra with galvanostatic cycling (C/30). These spectra were obtained from the same experiment as that shown in supplementary Figure 13. The spectra (enlarged to show the weaker, broader peaks) from 100 mv to 0 mv on discharge and from 0 mv to 450 mv on charge are shown. The deconvolution is conducted with a Voigt function fitting with the software Fityk ( 45 The deconvolution parameters are summarised in Supplementary Table 1. P1 (small Si clusters, ppm) and P3 (over-lithiated Si, 10 ppm) are indicated in the spectra. 18

19 Supplementary Figure 19. Deconvolved in-situ 7 Li NMR spectra with potentiostatic cycling. These spectra were obtained from the same experiment as that shown in supplementary Figure 16. The spectra (enlarged to show the weaker, broader peaks) from 100 mv to 0 mv on discharge and from 0 mv to mv on charge are shown. The deconvolution parameters are summarised in Supplementary Table 2. P1 and P3 are labelled in the spectra. 19

20 Supplementary Figure 20. Stacked in-situ 7 Li NMR difference spectra. The difference spectra are shown from 70 mv to 0 V in the 2 nd discharge with (a) galvanostatic at C/30 and (b) potentiostatic cycling. For both galvanostatic and potentiostatic experiments, the spectrum at ~70 mv is subtracted from the series of spectra so as to clearly see changes in the following spectra. The 7 Li resonances at around ppm, 0 10 ppm and 10 ppm are labelled as P1, P2 (isolated Si anions, c-li 3.75 Si, and extended Si networks), and P3, respectively. The formation of c-li 3.75 Si (P2; red arrow) and its conversion to c-li 3.75+x Si (P3) is accompanied by a sudden broadening, shift to lower frequency and decrease in intensity of the P1 resonance as shown by the blue arrow in b. The changes seen in the in-situ NMR spectra in the 0 5 ppm region are much less for bare CFGDL around 70 mv 0V as compared to SiNW-CFGDL, in agreement with the hypothesis that the 50 mv Li-Si process is responsible for the P2 resonance (c-li 3.75 Si). 20

21 Supplementary Figure 21. Schematics of in-situ 7 Li NMR spectrum intensity increase/decrease in the 2 nd cycle. (a) Galvanostatic (C/30) and (b) potentiostatic (stepped potential electrochemical spectroscopy, SPECS) cycling. (c) Bare CFGDL with galvanostatic cycling at C/25, showing the increase and decrease of peaks by blue and red arrows, respectively. The 7 Li resonances at around ppm, 0 10 ppm and 10 ppm are labelled as P1 (Li nearby small Si clusters), P2 (isolated Si anions, c-li 3.75 Si, and extended Si networks), and P3 (the over-lithiated crystalline phase, c-li 3.75+δ Si), respectively. 21

22 Supplementary Figure 22. Fractional lithium concentration, C, versus formation enthalpy per atom. The formation enthalpy is calculated with density functional theory (see Methods), showing a convex hull of lithium-silicon structures near c-li 3.75 Si. 22

23 P1 P3 1st discharge 1st charge 2nd discharge 2nd charge 3rd discharge 3rd discharge mv centre (ppm) standard errors (+/-) intensity standard errors (+/-) centre (ppm) standard errors (+/-) intensity standard errors (+/-) Supplementary Table 1 Summary of deconvolution parameters for P1 and P3 from Supplementary Figure 18 (galvanostatic cycling at C/30). 23

24 P1 P3 1st discharge 1st charge 2nd discharge 2nd charge 3rd discharge 3rd discharge mv centre (ppm) standard errors (+/-) intensity standard errors (+/-) centre (ppm) standard errors (+/-) intensity standard errors (+/-) Supplementary Table 2 Summary of deconvolution parameters for P1 and P3 from Supplementary Figure 19 (potentiostatic cycling). 24