Kinked silicon nanowires-enabled interweaving electrode configuration for lithium-ion batteries

A tri-dimensional interweaving kinked silicon nanowires (k-SiNWs) assembly, with a Ni current collector co-integrated, is evaluated as electrode configuration for lithium ion batteries. The large-scale fabrication of k-SiNWs is based on a procedure for continuous metal assisted chemical etching of Si, supported by a chemical peeling step that enables the reuse of the Si substrate. The kinks are triggered by a simple, repetitive etch-quench sequence in a HF and H2O2-based etchant. We find that the inter-locking frameworks of k-SiNWs and multi-walled carbon nanotubes exhibit beneficial mechanical properties with a foam-like behavior amplified by the kinks and a suitable porosity for a minimal electrode deformation upon Li insertion. In addition, ionic liquid electrolyte systems associated with the integrated Ni current collector repress the detrimental effects related to the Si-Li alloying reaction, enabling high cycling stability with 80% capacity retention (1695 mAh/gSi) after 100 cycles. Areal capacities of 2.42 mAh/cm2 (1276 mAh/gelectrode) can be achieved at the maximum evaluated thickness (corresponding to 1.3 mgSi/cm2). This work emphasizes the versatility of the metal assisted chemical etching for the synthesis of advanced Si nanostructures for high performance lithium ion battery electrodes.

Colloidal lithography: Silicon substrates (p-type, <100>, resistivity 10-25 Ωcm) have been used after 10 minutes ultrasonic treatment in methanol. Next, a short oxygen plasma treatment was performed to render their surface hydrophilic for the colloidal processing. A monolayer of 260 nm nominal diameter polystyrene (PS) spheres (Microparticles Gmbh) was assembled on the pre-treated Si substrate. In the next step, reactive ion etching was used to reduce the PS sphere diameter to 120 nm, while retaining the initial close-packed hexagonal ordering. A 15 nm Au layer is subsequently deposited on the PS-decorated Si substrate by physical vapor deposition. Adhesive tape followed by an overnight stay in dichloromethane was employed for the lift-off of the PS spheres. Chemical peeling: The sequential, top-down etching approach used in the fabrication of the k-SiNWs allows for the creation of a final porous nanowire segment. For this purpose, the etching was performed in a solution containing 4.6M HF and 8.3M H2O2 for 3 minutes ( Figure S1). The k-SiNWs were subsequently collected in methanol with a 10 seconds ultrasonic treatment.
k-SiNWs-based assembly with multi-walled carbon nanotubes (MWCNTs). The collected k-SiNWs are mixed with pre-dispersed MWCNTs (Nanocyl 7000) in methanol. The solution was further filtered through a Millipore system using a 0.45 μm nylon filter. The resulting composite was left to dry overnight in a vacuum oven at 50°C.
Electroless Ni deposition. The electroless deposition procedure was adopted from literature. 1 Briefly, the composite is immersed in an aqueous plating bath containing 1 g of nickel sulfate hexahydrate (Sigma Aldrich) as the source of Ni ions, 0.05 g dimethylamine borane (Acros Organic) as reducing agent, 0.5 g of sodium citrate (Sigma Aldrich) and 0.25 g of lactic acid (Sigma Aldrich) as complexant and, respectively, buffer in 50 ml deionized water. The pH was adjusted to 7 with ammonium hydroxide. The plating was performed at room temperature under continuous agitation. The onset of deposition was marked by the emergence of gas bubbles. After this point, the immersion time varied from 10 to 25 minutes depending on the desired thickness ( Figure S15). After the plating, the assembly was intensively rinsed in deionized water and left to dry in air. Morphology studies of the Ni coatings were mainly carried by scanning electron microscopy (SEM).
Electrochemical measurements. For the pristine assemblies, 500 nm of Cu was deposited on one side of the assembly as current collector by physical vapor deposition.
The electrochemical performances were recorded using a battery analyzer (Arbin Instruments). Coincells were assembled with either pristine k-SiNWs-based assembly or Ni-coated k-SiNWs-based assembly as the working electrode and Li metal foils as counter and reference electrodes. The electrolyte systems used were 1M LiTFSi in PYR13FSi (Solvionic) and 1M LiPF6 in EC:DEC (2% VC) (Solvionic). The separator was a Celgrad membrane for the organic electrolyte and a glass microfiber disk (Whatman GFD) for the ionic liquid electrolyte. The galvanostatic charge-discharge curves were recorded between 0.02 to 2 V vs Li/Li + . The half-cells were discharged (lithiated) and charged (delithiated) at different current densities as indicated throughout the main text.
Atomic force microscopy (AFM) analysis. The morphology of the assemblies of Si nanowires and carbon nanotubes was inspected by a Dimension Icon AFM (Bruker) in tapping mode using silicon cantilevers (Bruker) with a force constant of ∼42 N/m and an apex with a radius of curvature ∼10

nm.
Intermodulation atomic force microscopy (ImAFM). ImAFM uses an external multi-frequency lockin amplifier to excite the cantilever at two frequencies close to the resonance frequency peak and then it records the photo-detector response at many intermodulation products of the two drives. 2 Like any other AFM technique, the surface scanning is normally controlled by the host AFM.
Essentially, the ImAFM measures the amplitude dependence of the tip-surface force at each pixel of an image while scanning; thus, both amplitude and phase images are collected. 3,4 ImAFM uses a built-in method to determine the tip-surface force from the measured tip motion.
Normally, this procedure requires certain assumptions for the tip-surface force to reconstruct the measured data. Specifically, it reconstructs the tip-surface force F(d) as a function of cantilever deflection d = z -h, where h (the equilibrium tip height) and z (the tip position) are measured from a fixed position in the reference frame where the sample is at rest. Following Platz et al., we have approximated the nonlinear tip-surface force by a polynomial of finite degree N in the cantilever deflection as ( ) = . 5 The polynomial coefficients (in our case N = 19) gj can be obtained from the measured spectrum of odd-order intermodulation products. Once the polynomial approximation is selected to reconstruct the force curves as a function of cantilever deflection, we can use a specific force model to fit the data. For our samples, k-SiNWs deposited on oxidized Si (100) substrates, we have selected the Derjaguin -Muller -Toporov (DMT) model 6 considering a spherical tip and a flat surface to fit the following polynomial function: Here, R is the tip radius, ao is the interatomic distance, Fmin is the depth of the force minimum, and E* represents the effective elastic modulus.
Transmission electron microscopy (TEM) analysis. TEM analysis and electron tomography were performed on a Jeol JEM2100 transmission electron microscope microscope operated at 200 kV.
TEM images for electron tomography were collected over a tilt range of ±65° with tilt increments of 5°, at a nominal magnification of 35000 -60000 using the SerialEM software. The 3D reconstructions of the aligned tilt series were obtained using the simultaneous iterative reconstruction technique. Image filtering and alignment was done using Tomviz and the reconstruction performed using BrightfieldGui. Final image rendering was then done in Tomviz.
Least squares fitting of the measured α distribution. To understand the observed broad distribution of , we use the least squares method to fit the recorded experimental data.
Model. Let's consider a kinked nanowire with the kink at the location as shown in Scheme 1.
The angle formed by the arms and is , such that when = 180 ∘ , the nanowire is straight and when = 0 ∘ , the nanowire folds backward. The kink is viewed in a direction that might not be perpendicular to the plane in which the whole nanowire is situated. Therefore, the observed angle will be the projection ( ) of the angle that depends on the alignment of viewing plane with respect to the plane of the nanowire (and it varies from to 2 -). In this model, we have assumed that the viewing plane is parallel to the axis as shown in Scheme 1 with the following conditions:  Arm is in the direction;  Viewing direction is the z direction;  The arm is rotated, while keeping unchanged, and the viewing direction unchanged;  Point is at the origin .
The angle denotes the rotation of the arm around the axis, and is set to 0 when the arm is in the -plane. The coordinates of a point P on the arm will be: The projected angle ( ) will be such that: where ( ) is the viewed length of the segment .
Moreover, at = 0, = − ⋅ , This Equation 6 relates the apparent angle (i.e. ) with the real kink angle , and the rotation angle .
Distribution of viewing angles. Assuming that the distribution of angles is homogeneous, from 0 ∘ to 180 ∘ (0 to ), meaning that the nanowire do not penetrate the < 0 region, the distribution function is ( ) and it integrates to 1 between 0 and .
We must find ( ) corresponding to g( ) with the normalization condition: From Equation 6, we can write: Differentiating both sides, For > /2, substituting sin from Equation 9 as sin = 1 − , we obtain Using the rotational symmetry of angle from 0 to /2 and /2 to , the distribution function for between 0 and /2 can be written as: if < /2, ( ) = / ⋅ ; ∈ (0, ).
Assuming the overall distribution function is a linear sum of all the individual distributions associated with , the expected distribution function can be written as: where are the weights assigned to each .
Least squares fit. The angles formed between pairs of low indices crystallographic directions were matched with the position of peaks observed in the experimental data. These set of angles were then used as and fitted with least squares method, which gave their respective weights . Consequently, five prominent peaks were found sufficient to match the experimental data with an standard error of 0.0025. The authors have assumed a positive constant normalized weight fraction (between 0 and 1) for each angle ( ) such that the total weight sums up to 1.
Elemental mapping. Energy-dispersive X-ray spectroscopy (EDX) was performed with a JEOL 7600G SEM FEG.
Tensile tests. The mechanical properties of the assemblies were evaluated using a Universal Mechanical Tester (UMT-3, Bruker). The load sensor has a resolution of 0.5 mN and is able to accurately measure forces between 0.1 and 10 N. The specimens were cut in 20 mm long and 5 mm wide rectangles prior to the mechanical evaluation. The loading rate was 5 µm/s. An optical image with the specimen after tensile testing is shown in Figure S8b. An average of two measurements per assembly was performed. The consistency of the results is confirmed by the low error rates, as listed in Table S1. Young's modulus was extracted from the linear region of the stress-strain curves corresponding to the 0.16-0.40% strain range.
Experimental details Figure         The porosity is calculated as follows: where ρSi bulk= 2.33 g/cm 3 and ρC bulk= 2.26 g/cm 3 . Table S1. Physical parameters of the evaluated assemblies.